Ventromedial medulla inhibitory neuron inactivation induces REM sleep without atonia and REM sleep behavior disorder

Abstract

Despite decades of research, there is a persistent debate regarding the localization of GABA/glycine neurons responsible for hyperpolarizing somatic motoneurons during paradoxical (or REM) sleep (PS), resulting in the loss of muscle tone during this sleep state. Combining complementary neuroanatomical approaches in rats, we first show that these inhibitory neurons are localized within the ventromedial medulla (vmM) rather than within the spinal cord. We then demonstrate their functional role in PS expression through local injections of adeno-associated virus carrying specific short-hairpin RNA in order to chronically impair inhibitory neurotransmission from vmM. After such selective genetic inactivation, rats display PS without atonia associated with abnormal and violent motor activity, concomitant with a small reduction of daily PS quantity. These symptoms closely mimic human REM sleep behavior disorder (RBD), a prodromal parasomnia of synucleinopathies. Our findings demonstrate the crucial role of GABA/glycine inhibitory vmM neurons in muscle atonia during PS and highlight a candidate brain region that can be susceptible to α-synuclein-dependent degeneration in RBD patients.

Fig. 1

GABA/glycine neurons of the ventromedial medulla (vmM) are specifically active during PS rebound. a Photomicrographs comparing the distribution in RMg (upper row), GiV (middle row) and 7–8 sp Rexed’s layer of lumbar spinal cord (SC, lower row) of GlyT2+ neurons expressing c-Fos in control (PSC, first column) and in rats submitted to a PS deprivation (PSD, second column), a PS rebound (PSR, third column) or a protocol of forced locomotion (STEP, fourth column). Neurons expressing c-Fos were colored in brown (nuclear staining, arrowheads) whereas GlyT2+ neurons were colored in blue (cytoplasmic staining, double arrowhead). Notice that the number of double-labeled neurons c-Fos+/GlyT2+ (arrows) in RMg (top line) and GiV (middle line) is higher during PSR compared to the 3 other experimental conditions. At spinal level, numerous c-Fos+/GlyT2+ neurons were found quite only in STEP rats. Scale bars: 50 µm. b Histograms (mean ± SEM; n = 4 PSC, n = 5 PSD, n = 6 PSR, n = 4 STEP) illustrating the percentage of c-Fos+ that were also GABA/glycine in nature (c-Fos+/GlyT2+) encountered within the vmM and lumbar SC for each experimental condition. c, d Drawings of sections showing the distribution of single c-Fos+ (black dots) and double-labeled c-Fos+/GlyT2+(red dots) neurons in the same representative PSR rat within the RMg and GiV (c, frontal sections) and lumbar SC at T13-L2 levels (d, horizontal section). Kruskal–Wallis tests followed by Mann–Whitney U tests *p < 0,05 compared to PSC, ^#p < 0,05 compared to PSD, ^§p < 0,05 compared to PSR. 7 facial nucleus, 7–8 sp 7–8 Rexed’s spinal cord layers, Amb ambiguous nucleus, DPGi dorsal paragigantocellular nucleus, Gi gigantocellular reticular nucleus, GiA gigantocellular reticular nucleus (pars alpha), GiV gigantocellular reticular nucleus (pars ventral), IO inferior olivary complex, LPGi lateral paragigantocellular nucleus, MTn motoneurons, MVe medial vestibular nucleus, PCRt parvicellular reticular nucleus, Pr praepositushypoglossi nucleus, py pyramidal tract, RMg raphe magnus nucleus, SLD pontine sublaterodorsal tegmental nucleus, sp5 trigeminal spinal tract, SpVe spinal vestibular nucleus

Fig. 2

Neurons of the vmM specifically activated during PS send direct projections to the lumbar motoneurons. a Drawing illustrating the experimental paradigm with a FG injection in lumbar motoneurons at spinal L1–L2 levels in rats allowed to sleep recover (PSR) after a 72-h deprivation of PS using the flower-pot method (n = 4). b–d Photomicrographs at high magnification of sections at the level of RMg (b), GiV (c), and 7–8 sp Rexed’s layers at lumbar SC (d), double immunostained for c-Fos and FG in a representative PSR rat. Neurons expressing c-Fos were colored in  black (nuclear staining, arrowheads) whereas FG+ neurons were colored in brown(cytoplasmic staining, double arrowhead). Notice the massive numbers of double-labeled neurons c-Fos+/FG+ (arrows) in RMg and GiV compared to lumbar SC where scattered, if any c-Fos+/FG+ cells are seen after PS recovery. e Percentage of c-Fos+ neurons in the vmM that are retrogradely labeled after a FG injection in the lumbar SC (PSR n = 4, values are mean ± SEM). f Drawings of two frontal medulla sections illustrating the bilateral distribution of single FG+ (violet circles), single c-Fos+ (black dots), and double-labeled c-Fos+/FG+ neurons (red dots) in the RMg and GiV in the same PSR rat. g Photomicrograph overview of one horizontal lumbar section at the level of 7–8 sp Rexed’s layers showing fibers emanating from transducted GABA/glycine vmM neurons with AAV as immunolabeled for mCherry (black fibers). Notice the very high density of mCherry immunolabeled fibers surrounding the soma of lumbar motoneurons (brown cytoplasmic staining), often in close apposition with them suggesting that vmM neurons have direct synaptic contacts with somatic motoneurons. Scale bars: 40 µm in b–d; 50 µm in g overview and 20 µm in enlarged square

Fig. 3

Verification of the genetic inactivation of endogenous vGAT mRNA and native vGAT protein in GABA/glycine vmM neurons. a Scheme of the in vivo experiments: AAV-shRNA injections were made in three points to genetically inactivate GABA/glycine neurons present bilaterally in the GiV, and GiA and the midline RMg and to trace anterogradely their efferent projections of transducted neurons until the lumbar spinal cord. b, c Drawings reporting the location and the extent of AAV injection sites at day 30 post-injection for each Ctrl-shRNA (b, n = 5) and vGAT-shRNA (c, n = 7) treated rat considered for the physiological study. The injection spreading was demarcated by the spontaneous mCherry fluorescence. Notice that for these rats, the AAV injections covered the largest part of the vmM and avoided the LPGi laterally and Gi dorsally. d, e Low power photomicrographs of the RMg/GiA (d) and the GiV (e) in a representative control rat showing the high number of transducted neurons after AAV injection. h, i Photomicrographs of adjacent sections treated for vGAT mRNA ISH. Notice the high number of strongly labeled neurons in RMg, GiA, and GiV indicating that the expression of vGAT mRNA is normal. f, g Photomicrographs of vmM sections taken from a representative vGAT-shRNA rat after the ISH labeling of vGAT mRNAs. Note the absence of labeled neurons in RMg, GiA, and GiV within the injection site. j Confocal photomicrographs comparing in representative Ctrl-shRNA (left) and vGAT-shRNA (right) rats the expression in lumbar motoneurons pool of the native vGAT protein (green, middle panels) in synaptic terminals emanating from transducted vmM neurons (mCherry, top panels). As evidenced on merged photomicrographs, the vGAT protein is expressed in axons and synaptic terminals (arrows) emanating from vmM neurons in Ctrl-shRNA rats (yellow, bottom left panel) but is virtually absent in vGAT-shRNA rats (bottom right panel). Scale bars: 100 µm in b, c; 40 µm in enlarged squares; 5 µm in f

Fig. 4

Loss of muscle atonia during PS after the genetic inactivation of GABA/glycine vmM neurons. a, b Histogram comparing the daily percentages of W, SWS, and PS in Ctrl-shRNA (open bars) vs. vGAT-shRNA (filled bars) rats at day 30 after AAV injections. Notice the significant reduction of the PS episode duration in vGAT-shRNA vs. Ctrl-shRNA rats (b). c Normalized mean power spectrum in a control (black) and experimental (red) rats showing no differences in EEG during PS between both groups. d, e Raw polysomnographic recordings during PS of two representative Ctrl-shRNA (d) and vGAT-shRNA rats (e) showing the loss, although irregular, of muscle atonia (arrows) concomitant to a strong increase of phasic twitches on nuchal EMG after the genetic inactivation of GABA/glycine vmM neurons. f Dot plots comparing mean EMG values during PS and SWS episodes (bars) in Ctrl-shRNA (open circles) vs. vGAT-shRNA (filled circles) rats. Dashed lines connect SWS with PS mean values for each animal. Control rats show an expected diminution of mean EMG values during PS compared to preceding SWS (i.e., PS atonia). In experimental rats, mean EMG values are comparable during PS and SWS unraveling the loss of atonia during PS. Interestingly, PS EMG is significantly increased in vGAT-shRNA rats compared to PS EMG in Ctrl-shRNA congeners. g Dot plots represented in decibels showing that PS/SWS ratio of mean EMG values is increased in experimental (filled circles) vs. control (open circles) rats. Mann–Whitney U tests, *p < 0,05 compared to Ctrl-shRNA; Wilcoxon signed rank test ^#p < 0,05 compared to mean EMG values during SWS

Fig. 5

RBD-like behaviors are displayed during PS after the genetic inactivation of GABA/glycine vmM neurons. a, b Examples of captured video images of representative Ctrl-shRNA (a) and vGAT-shRNA rats (b), both during a PS episode. Each red point corresponds to a gray color changed pixel between two successive images (see Materials and Methods sections). The density and location of red points respectively reflect intensity of rat’s movements and body territories where they occurred. c Histogram comparing in both rat groups the mean actimetry per second of SWS and PS calculated by counting the number of changed pixels between two successive video images due to animal movements. Oneiric movements appear only during PS in vGAT-shRNA rats. d, e Histograms comparing in bot rat groups the mean number of motor events per PS bout (d) and the mean duration of motor events (in seconds, e) per PS episode. f Histogram showing the percentage of PS time with movements in both rat groups. Note that all actimetry parameters calculated thanks to video recordings and that directly reflect the movements displayed by rats during PS are significantly higher in vGAT-shRNA vs. Ctrl-shRNA animals. Mann–Whitney U tests, *p < 0,05; **p < 0,005 compared to Ctrl-shRNA. g, h Activity maps of one representative Ctrl-shRNA (g) and vGAT-shRNA rats (h) illustrating EMG nuchal changes and actimetric phasic events during SWS, SWS preceding PS, and PS. SWS and PS episodes are sorted vertically from the highest (top, yellow values) to the lowest (bottom, blue values) mean muscle tone combined with actimetry events. The SWS episodes preceding PS are sorted in the same order as their related PS episodes. The x-axis represents the episode duration normalized between 0 and 2π. Notice the increase of muscle tone and actimetry motor events during PS in vGAT-shRNA animals and the higher probability of movement occurrence during the last third of PS bouts