Multisensory integration in early vestibular processing in mice : 3 The encoding of passive versus active motion

41 42 The mouse has become an important model system for studying the cellular basis of 43 learning and coding of heading by the vestibular system. Here we recorded from single 44 neurons in the vestibular nuclei (VN) to understand how vestibular pathways encode 45 self-motion under natural conditions, during which proprioceptive and motor-related 46 signals as well as vestibular inputs provide feedback about an animal’s movement 47 through the world. We recorded neuronal responses in alert behaving mice focusing on 48 a group of neurons, termed vestibular only cells, that are known to control posture and 49 project to higher-order centers. We found that the majority (70%, n=21/30) of neurons 50 were bimodal, in that they responded robustly to passive stimulation of proprioceptors 51 as well as passive stimulation of the vestibular system. Additionally, the linear 52 summation of a given neuron’s vestibular and neck sensitivities predicted well its 53 responses when both stimuli were applied simultaneously. In contrast, neuronal 54 responses were suppressed when the same motion was actively generated, with the 55 one striking exception that the activity of bimodal neurons similarly and robustly 56 encoded head on body position in all conditions. Our results show that proprioceptive 57 and motor-related signals are combined with vestibular information at the first central 58 stage of vestibular processing in mice. We suggest that these results have important 59 implications for understanding the multisensory integration underlying accurate postural 60 control and the neural representation of directional heading in the head direction cell 61 network of mice. 62 63 64 65 66 67 68 Introduction 69 70 The advent of genetic engineering has resulted in numerous mutant mouse strains with 71 behavioural phenotypes suggestive of vestibular dysfunction including head tilting and 72 bobbing, locomotive circling, and ataxia. Behavioural studies have characterized 73 vestibularly driven eye movements in such mutants to gain insight into the mechanisms 74 underlying motor learning and compensation (e.g., Beraneck et al. 2008; Schonewille et 75 al. 2011; Katoh et al. 2008; Faulstich et al. 2006; Stahl et al. 2004). In addition, in vitro 76 studies (Camp et al, 2006; Nelson et al, 2003; Sekirnjak and du Lac 2002, 2006; 77 Bagnall et al, 2007, 2008) have characterized the intrinsic membrane dynamics of 78 neurons at the first central stage of sensory processing (i.e., vestibular nuclei (VN) 79 neurons) in mice. However, the gap between in-vitro recordings and behaviour has not 80 been bridged, since the information encoded by VN neurons in alert mice remains 81 poorly understood. 82 83 The ability to sense vestibular information is integral to the generation of reflexes that 84 stabilize gaze and posture as well as the perception of self-motion. As an animal moves 85 through its environment, however, self-motion cues are also available from the muscles 86 and joints (proprioception) and motor commands producing movement (e.g. motor 87 efference copy). Thus to understand the neural encoding of self-motion, it is vital to 88 establish whether and how these cues are combined with vestibular signals at the level 89 of individual neurons. To date, however, in the only in-vivo characterization of the 90 vestibular nuclei undertaken in awake behaving mice (Beraneck and Cullen, 2007), 91 animals were restrained such that neural sensitivities were quantified in response to 92 passive vestibular stimulation alone. 93 94 Accordingly, to understand the encoding of self-motion under natural conditions we 95 recorded from single VN neurons in head-unrestrained wild type mice during passive 96 and active movement. We addressed three important questions. First, do murine VN 97 neurons show robust responses to passive stimulation of proprioceptors? Second, can 98 responses to combined proprioceptive and vestibular stimulation be predicted by the 99 weighted linear sum of responses to the individual cues? Third, are these inputs 100 combined differently for passive versus active movements? Our findings show that 101 vestibular and neck proprioceptive information is integrated by single vestibular nuclei 102 neurons in alert, behaving mice. However, while linear summation of the vestibular and 103 neck sensitivities predicts well neuronal responses in passive conditions, neuronal 104 responses are suppressed when the same motion was actively generated with one 105 exception. Notably, the majority of murine VN neurons display a constant and robust 106 sensitivity to static head orientation produced by both passive and active movements. 107 Because these neurons likely send descending projections to spinal pathways, our 108 results suggest that the efficacy of vestibulo-spinal reflexes is reduced during active 109 motion. Furthermore, because work in primate and cat further suggest these neurons 110 send ascending projections to the vestibular cerebellum (Cheron et al. 1996; Reisine 111 and Raphan 1992) and thalamus (reviewed in Cullen 2012), our results have significant 112 implications for understanding the contribution of vestibular pathways to spatial 113 perception in mice. Importantly, vestibular inputs are commonly thought to be vital cue 114 for the generation of the head direction signal coded by the head direction cell circuit, 115 yet our results show that the motion information encoded at the first central stage of 116 vestibular processing reflects the integration of allocentric (i.e., vestibular) and 117 egocentric (i.e., proprioceptive, motor efference copy) signals. Accordingly, we 118 speculate that this multisensory integration contributes to the genesis of an internal 119 representation of directional heading in mice. 120 121 122 123 124 125 Methods 126 Eighteen male C-57bl6 (30–35 g; Charles River Laboratories) adult mice were 127 included in this study. The procedures were approved by the McGill University Animal 128 Care Committee and were in strict compliance with the guidelines of the Canadian 129 Council on Animal Care. 130 131 Head-post implantation and craniotomy 132 Surgical techniques and anesthesia protocols used were adapted from those 133 previously described by Beraneck and Cullen (2007). Briefly, mice were anesthetized 134 with an intraperitoneal injection of a mixture of atropine (5 X 10 mg/g), ketamine (10 135 mg/g), acepromazine maleate (2.5 X 10 mg/g), xylazine (10mg/g), and sterile saline. 136 Anesthetized mice were then secured in a stereotaxic frame. A 1 mm diameter 137 craniotomy was performed to allow access to the vestibular nuclei. A dental cement 138 chamber (C&B Metabond) was then constructed around the craniotomy, and a custom 139 built head holder was cemented to the implant anterior to the chamber. Following 140 surgery, Carbapen (0.05 mg/kg) was utilized for postoperative analgesia, and 141 antibacterial cream (2%, Astra Pharma, Ontario) was applied to the incision site to 142 prevent infection. Animals were kept in isolated cages and closely monitored during the 143 first 72 hours. 144 145 Recording sessions 146 147 During experiments, mice were placed in a custom built Plexiglas tube at the center of a 148 turntable. Their heads were fixed with a stainless steel post to align the horizontal 149 semicircular canals with the horizontal plane (i.e., 35° nose down; Calabrese and Hullar 15