Loss of cerebellar function selectively affects intrinsic rhythmicity of eupneic breathing

Respiration is controlled by central pattern generating circuits in the brain stem, whose activity can be modulated by inputs from other brain areas to adapt respiration to autonomic and behavioral demands. The cerebellum is known to be part of the neuronal circuitry activated during respiratory challenges, such as hunger for air, but has not been found to be involved in the control of unobstructed breathing at rest (eupnea). Here we applied a measure of intrinsic rhythmicity, the CV2, which evaluates the similarity of subsequent intervals and is thus sensitive to changes in rhythmicity at the temporal resolution of individual respiratory intervals. The variability of intrinsic respiratory rhythmicity was reduced in a mouse model of cerebellar ataxia compared to their healthy littermates. Irrespective of that difference, the average respiratory rate and the average coefficient of variation (CV) were comparable between healthy and ataxic mice. We argue that these findings are consistent with a proposed role of the cerebellum in the coordination of respiration with other rhythmic orofacial movements, such as fluid licking and swallowing.


INTRODUCTION 46
The cerebellum has extensive reciprocal connections with the brain stem and cerebellar neuropathology is known to affect brain stem controlled processes, such 48 as cardiovascular and respiratory function (Harper et al., 1998;Paton et al., 1991;49 Rector et al., 2006). The involvement of the cerebellum in respiration seems to be 50 central to respiratory challenges, such as hypoxia or hypercapnia (Macey et al., 2005;51 Parsons et al., 2001). Neurons in the medial cerebellar nuclei project to brain stem 52 areas containing the respiratory pattern generating circuits (Lu et al., 2013) and 53 neuronal activity in the medial cerebellar nuclei is entrained by the respiratory 54 rhythm (Lu et al., 2013;Xu and Frazier, 2002). Investigations into eupneic 55 respiration, however, failed to implicate the cerebellum in animal models (Moruzzi, 56 1940;Xu and Frazier, 2002;Xu et al., 1995) or humans (Ebert et al., 1995). Previously 57 published findings suggested that ablation of the fastigial nucleus altered respiratory 58 responses to hypercapnia but did not alter eupneic breathing (Xu and Frazier, 2002). 59 However, a recent study conducted in juvenile mice reported the possibility of only a 60 modest influence of cerebellar cortical function during respiration as determined by 61 a measure of breathing regularity called inter breath interval (van der Heijden and 62 Zoghbi, 2018). Collectively, these previous studies that investigated the involvement 63 of the cerebellum in respiration have been complicated by the use of anaesthetized 64 conditions or analyses conducted before postnatal day 30, an age prior to which 65 functional cerebellar circuits have yet to reach maturity (Arancillo et al., 2015). We 66 therefore sought to test the role of the mature cerebellum in eupneic breathing in 67 awake conditions using multiple physiological measures of system rhythmicity. 68 To quantitatively address this problem, we compared the average respiratory 69 rate, coefficient of variation (CV) of the respiratory rhythm and the intrinsic 70 rhythmicity of respiration (CV2) (Holt et al., 1996) of eupneic breathing in healthy 71 mice (controls, CT) and in adult mice with cerebellar ataxia. Cerebellar ataxia was 72 induced using the Cre/LoxP genetic approach to selectively block Purkinje cell 73 GABAergic neurotransmission (L7 Cre ;Vgat flox/flox ), thereby functionally disconnecting 74 the cerebellar cortex from the cerebellar nuclei (White et al., 2014). Spontaneous 75 respiratory behavior was measured in a plethysmograph. 76 Cerebellar ataxia did not affect the average respiratory rate or the CV of the 77 respiratory rhythm. However, compared to their CT littermates, L7 Cre ;Vgat flox/flox 78 mutant mice (MU) showed increased intrinsic rhythmicity, as measured by the CV2. 79 The CV2 evaluates the similarity of pairs of intervals, and is thus sensitive to brief 80 changes in rhythmicity at a temporal resolution of individual interval durations. The 81 CV2 provides a measure of intrinsic rhythmicity in the sense that this measure is 82 sensitive to short periods of highly regular breathing but less sensitive to slow rate 83 variations than the CV. Based on our results and findings from previously published 84 studies, we propose a role for the cerebellum in the coordination of multiple rhythmic 85 orofacial movements, such as coordinating respiration with swallowing. 86 87 88 89

90
To measure spontaneous respiratory behavior, mice were placed in a 91 plethysmograph chamber, where they could move freely (Fig. 1). Spontaneous 92 respiration was monitored for 30 min and compared between MU mice and their CT 93 litter mates (Table 1). Male and female mice of both phenotypes were divided into 94 two age groups, 2-3 or 5-7 months old (Table 1), corresponding approximately to the 95 transition from adolescence/young adults to mature adults in human development 96 (Flurkey et al., 2007). 97 Respiratory activity in the plethysmograph chamber causes rhythmic 98 pressure changes, which were measured using a pressure transducer (Fig. 1A). 99 Rhythmic decreases in voltage output from the transducer corresponded to 100 inhalation movements, and the times of voltage trough minima were marked as "end-101 of-inhalation" times ( Fig. 1B). All further analysis was based on these temporal 102 markers. 103 The average respiratory rate was evaluated by calculating the average 104 duration of inter-inhalation intervals. Comparison of mean interval durations 105 between MU mice and CT littermates within age groups revealed no difference 106 between groups (p=0.99, adjusted for sex and age; Fig. 2

-see Materials and 107
Methods/Statistical analysis for details). Older mice, however, had shorter average 108 interval durations than the younger mice (p<0.001; adjusted for sex), consistent with 109 a known age-related increase in respiratory rate (Rodriguez-Molinero et al., 2013). 110 Comparison of respiratory rate by sex showed a systematic, albeit not very 111 pronounced difference (p=0.048, adjusted for age), with female mice having higher 112 respiratory rates than males. 113 The overall variability of the respiratory rhythm, quantified as the coefficient 114 of variation of the inter-inspiration interval distribution (CV = standard deviation of 115 interval distribution/mean of interval duration) was, like the mean interval duration, 116 dependent on age (p<0.001, with lower values for younger mice), but not on sex 117 (p=0.70). Also in analogy to the mean interval duration above, we could not find a 118 difference between MU mice and their CT littermates with respect to CV (p=0.95, 119 adjusted for age) (Fig. 3). 120 The CV measures variability based on the distribution of intervals across the 121 entire observation time. Brief but reoccurring changes in rhythmicity cause no 122 change in the CV, as long as the overall interval distribution remains the same or 123 similar. In 1996, Holt et al. introduced a novel measure of variability, the CV2 (see 124 methods), specifically designed to detect the brief, reoccurring changes in 125 rhythmicity, which they called "intrinsic rhythmicity" (Holt et al., 1996). 126 Comparison of the CV2 of respiratory behavior in CT and MU mice revealed a 127 lower CV2 in MU mice compared to the CT littermates (p < 0.001), when adjusting for 128 age (p<0.001) and sex (p=0.048) (Fig. 4). While the CV2 increased with age in the 129 control mice (p<0.001), it did not change substantially with age in L7 Cre ;Vgat flox/flox 130 mice (p=0.16) (Fig. 4). Irrespectively, in the combined data, the interaction effect 131 between age and phenotype was not significant (p = 0.27); something that would be 132 expected if the age-CV2 relationship holds for only one of the two genotypes. 133 The reduced CV2 in MU mice suggests that the difference in the respiratory 134 sequences in MU and CT mice is based on the temporal modulation of a subset of 135 individual respiratory intervals (see also discussion). In mutant mice, those 136 modulations seem to be less pronounced, resulting in a smaller CV2 value, reflecting 137 reduced intrinsic variability. To address our hypothesis that the difference in CV2 138 The cerebellum is known to be part of the neuronal circuitry activated during 148 respiratory challenges, such as hypercapnia or hypoxia (Macey et al., 2005;Parsons 149 et al., 2001). There has been, however, no clear experimental evidence supporting a 150 role of the cerebellum in eupneic breathing (Ebert et al., 1995;Moruzzi, 1940;van der 151 Heijden and Zoghbi, 2018;Xu and Frazier, 2002;Xu et al., 1995). Consistent with 152 previous findings, our results show that loss of cerebellar function does not affect the 153 average respiratory rate or the coefficient of variation of eupneic respiration in mice. 154 Analysis of the CV2, however, showed a significantly reduced variability, i.e. a lower 155 CV2, in MU mice compared to their CT littermates (Fig. 4). The CV2 is uniquely 156 sensitive to brief changes in rhythmicity, which could affect only one or two intervals 157 at a time (Holt et al., 1996). Our findings suggest that loss of cerebellar function is 158 associated with a loss of modulation of select respiratory intervals. A well-known 159 biological purpose for an occasional modulation of respiratory intervals is the 160 coordination of breathing with swallowing movements, which is crucial to protect the 161 airways from aspirating fluid or food (Hardemark Cedborg et al., 2009;Yagi et al., 162 2017). The typical observation in healthy subjects is an extended pause of the 163 respiratory rhythm during swallowing (swallowing apnea) (Hardemark Cedborg et 164 al., 2009;Nishino, 2012). Inappropriate temporal coordination of swallowing and 165 respiratory movements can lead to dysphagia, a common observation in neurological 166 disorders such as stroke or Parkinson's disease (Yagi et al., 2017). Based on our 167 observations we propose that the cerebellum is involved in the temporal 168 coordination of breathing and swallowing movements, specifically that the extension 169 of the respiratory pause during swallowing requires an intact cerebellum. We tested 170 this idea in silico by artificially prolonging the durations of a subset of respiratory 171 intervals in the respiratory sequences of ataxic mice. Increasing the duration of every 172 10 th respiratory interval by 50% increased the CV2 of the MU sequence to values that 173 no longer differed from CV2 values of CT mice (Fig. 5). At the same time, the changes 174 in average rate and CV caused by this manipulation were insignificant. 175 Interestingly, it has been reported in humans that breathing-swallowing 176 coordination changes with age (Wang et al., 2015). Older individuals tend to show a 177 reduced frequency of the protective expiration-swallow-expiration pattern, which 178 would increase the risk of aspiration, but accompanying changes, such as longer 179 swallowing apnea, seem to compensate for deficient breathing-swallowing 180 coordination (Wang et al., 2015). While our data suggest an age-related difference in 181 the CV2 of the respiratory sequence, answers to the question of whether a CV2 182 increase with age is linked to cerebellar function will require a larger sample size size 183 and preferably longitudinal data from the same specimen as they age. 184 Besides the data we presented here, there are several other lines of evidence 185 supporting a role for the healthy cerebellum in the coordination of respiration with 186 rhythmic orofacial movements. Extensive reciprocal projections connect the 187 cerebellum and the brain stem (Asanuma et al., 1983;Teune et al., 2000;Whiteside 188 and Snider, 1953), providing the necessary anatomical substrate for a cerebellar 189 coordination of brain stem controlled breathing with other orofacial movements. A 190 detailed anatomical and electrophysiological investigation of the medial cerebellar 191 nuclei showed that neurons project broadly to the brain stem, including the area 192 containing the respiratory pattern generating circuits (Lu et al., 2013). 193 Electrophysiological recordings from medial nucleus neurons showed that their 194 spiking activity is correlated with one or two of three different orofacial movements: 195 respiration, fluid licking and mystacial whisker movements (Lu et al., 2013;Xu and 196 Frazier, 2002). In addition, a significant sub population of the medial cerebellar 197 neurons send axon collaterals to more than one brain stem site, a projection pattern 198 that could serve the coordination of brain stem activity at the two sites. Swallowing 199 was not measured in that study, but several lines of evidence support a role for the temporal coordination of sensorimotor activity during behavior (Braitenberg, 1961;229 Ivry et al., 2002;Mauk and Buonomano, 2004).

Measurements of respiratory behavior 247
Respiratory behavior was monitored for 30 min by placing mice in a plethysmograph 248 consisting of a glass container within which mice could move about freely (Fig. 1). 249 The air in the chamber was continuously replaced by a constant flow (1 l/min) of 250 fresh air, directed into the chamber through an opening in the lid with a second 251 opening in the lid serving as an outlet. While the mice were in the chamber, a box was 252 placed over the chamber, blocking direct light. Descriptive boxplots including all animals by phenotype as well as age at time of 276 testing and sex are shown in Fig. 2. Linear models were fitted relating the respective 277 outcome measure to phenotype while adjusting for age and sex. Model fit was verified 278 by residual plots (such as quantile-quantile plots) to verify approximate normality of 279 residuals (not shown). The most important conclusions for CV2 were verified by a 280 bootstrap resampling approach (not shown). Because respiratory function might be 281 affected by age and/or sex, it is important to adjust for these factors when evaluating 282 systematic differences with respect to phenotype. We do this by adding factors for 283 sex and age-group in our regression approach (Kutner et al., 2005). We follow the 284 common approach to exclude a factor if it is not statistically significant (p <= 0.05) 285 because each additional factor reduces the degrees of freedom in the testing for the 286 phenotype-difference, which is our main focus.

Figure 2) Comparison of mean respiratory interval durations. 345
Mean interval duration is lower in older mice (right panel; p<.001), but does not differ 346 between MU and CT mice within each age-group (left and right box within each panel; 347 p = 0.99, adjusted for age and sex). There seems to be a systematic albeit modest 348 difference by sex (p = 0.048, adjusted for age). The CV2 is lower in the MU mice compared to their CT littermates (left vs. right box 365 in each panel; p<.001) when adjusting for age (p<.001) and sex (p = 0.048). In 366 addition, the CV2 was significantly higher in older compared to younger CT mice 367 (p<.001 in the CT subgroup), but possibly remained unchanged across age in the MU 368 mice (p = 0.16 in the MU subgroup). However, in the combined data, the interaction 369 effect between age and phenotype is not statistically significant (p = 0.19); something 370 that would be expected if the age-CV2 relationship holds for only one of the two 371 phenotypes. Using the same analytical approach as in the original respiratory sequence measured 380 in MU mice, one would now conclude that the phenotype is not statistically significant 381 (p = 0.31; age remains significant with p <.001), The respiratory sequences of MU 382 mice was modified in silico by extending the duration of every 10 th interval by 50%, 383 as described in the methods section. 384 385 386 387