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.

Keywords: hypothalamus; motor control; predictive control; skilled movements.

© The Author(s) 2022. Published by Oxford University Press on behalf of the National Academy of Sciences.

Figures

Fig. 1.

Correlating premovement HON activity and anticipatory movements. (A) Setup diagram (see “Methods” for description). (B) Single trial structure. HON signals are analyzed between the go cue and movement initiation. (C) Session structure, with challenge-containing trials (left) and without (right). (D) Left, mean premovement HON-GCaMP6s activity ( n = 15 challenge sessions from four mice). Right, corresponding prechallenge acceleration vectors. (E) Same as (D), but without challenge (n = 9 sessions from four mice). (F) HON-GCaMP6s signal amplitude across challenge sessions (n = 15 sessions from four mice) and no-challenge sessions (n = 9 sessions from four mice). Red dashed line indicates the start of the challenge-containing trails (or not challenge control). 2-way RM ANOVA, challenge F(1, 22) = 14.99, P < 0.001; challenge x trials F(7,154) = 4.439, P < 0.001). (G) Same as (A), for anticipatory movements (challenge, n = 38 sessions from 12 mice; no challenge, n = 32 sessions from 12 mice). 2-way RM ANOVA, challenge F(1,102) = 13.99, P < 0.001; challenge x trials F(7,714) = 11.59, P < 0.0001). (H) Relationship between premovement HON-GCaMP6s signals and corresponding anticipatory counter movement (challenge-opposing acceleration component) with challenge (black) and without challenge (gray). Each point is the average trial epoch (block of 25) for each mouse during trials 50 to 200. Challenge n = 18 trial blocks from three mice, Pearson's correlation: r2 = 0.506, P = 0.0009. No challenge n = 24 trial blocks from four mice, Pearson's correlation: r2 = 0.016, P = 0.557, ns: P > 0.05. I. Same as in (H), but for challenge-perpendicular acceleration component. Challenge n = 18 trial blocks from three mice, Pearson's correlation: r2 = 0.106, P = 0.186, ns: P > 0.05. No challenge n = 24 trial blocks from four mice, Pearson's correlation: r2 = 0.078, P = 0.186, ns: P > 0.05. (J) Same as in (H), but for complete acceleration vector. Challenge n = 18 trial blocks from three mice, Pearson's correlation: r2 = 0.237, P = 0.0040. No challenge n = 24 trial blocks from four mice, Pearson's correlation: r2 = 0.014, P = 0.580.

Fig. 2.

Causal roles of premovement HON signals in subsequent movements. (A) Top, temporal targeting of HON optogenetic interference. Middle, averages of the acceleration vectors of HON optostimulation (orexin::C1V1) sessions in baseline (trials 26 to 50), early (trials 51 to 75), middle (trials 101 to 125), and late (trials 176 to 200) epochs ( n = 4 mice). Acceleration vectors are normalized to the magnitude of both acceleration components (x and y) of baseline trials 26 to 50. Bottom, same as (B), but for HON- optosilencing (orexin::ArchT, n = 4 mice). (B) Group data of anticipatory movements during HON optostimulation (magenta, orexin:: C1V1, n = 4 mice), HON optosilencing (cyan, orexin::ArchT, n = 4 mice), and control (orange, laser off, n = 11 mice). (C) Same as (B), but comparing prechallenge (trials 1 to 50) trial averages with during-challenge (trials 100 to 200) trial averages. Data are means ± SEM. Indicated statistics are from Holm–Sidak’s multiple comparison tests (ns = P > 0.05, *P < 0.05,**P < 0.01, ***P < 0.001) following 2-way RM ANOVA, group F(2, 15) = 5.662, P < 0.05. (D) Same as (B), but comparing slopes of linear fits to the relationship between anticipatory movement and trial number during the first 50 trials of challenge (a metric of learning to anticipate across trials). Data are means ± SEM. Indicated statistics are from Holm–Sidak’s multiple comparison tests (ns =   P > 0.05, *P < 0.05,**P < 0.01, ***P < 0.001) following 1-way ANOVA, F(2, 16) = 2.774, P < 0.01. (E) Control experiment for (B). Data are means ± SEM of n = 8/10/9 sessions from six mice for laser ON (excitation light pattern)/laser OFF/laser ON (silencing light pattern). 2-way RM ANOVA, F(92, 24) = 0.01039, P = 0.9897. (F) Same experiment as in (B), but repeated without the challenge (the whole acceleration vector is used here since no lateral force was present). Data are means ± SEM of n = 4/23/4 sessions from 4/8/4 mice (HON optostimulation/control/HON optosilencing = C1V1/control/ArchT).