Altered allostatic regulation of wakefulness and slow-wave-sleep spectral quality in mice with hypocretin (orexin) receptor 1 inactivation in noradrenergic cells

The neuromodulators noradrenaline (NA) and hypocretins (Hcrt) both have major, but distinct, roles in implementing the brain circuit processing modes that define wakefulness and waking behaviors. The A6 NA cell group in the locus coeruleus (LC-NA) receives dense projections from the Hcrt cells of the hypothalamus, and acts as critical mediator of Hcrt-induced arousal. The range of behaviors relying on the Hcrt-to-LC-NA neuronal connection however remains undefined. We generated a c onditionally inactivated allele of Hcrtr1 , the Hcrt receptor gene expressed by LC-NA cells, and used a Dbh-Cre transgene to create mice with NA-specific Hcrtr1 inactivation ( Hcrtr1 Dbh-CKO mice). Electrocortical activity (ECoG) of these mice was analyzed in distinct behavioral contexts, and contrasted to the one of Cre -less control littermates. While baseline waking was grossly normal albeit for enhanced activity in the slow-delta ( δ ) range, upon exposure to a novel environment, Hcrtr1 Dbh-CKO mice displayed (i) a slowing of the waking ECoG, with enhanced δ and inter- δ/θ band activity, (ii) blunting of the θ rhythm and fast- γ activity, while (iii) β and slow- γ activities were enhanced. Moreover, locomotor and ECoG responses to sleep deprivation (SD) were markedly reduced. Surprisingly, while deficits in manually-enforced wakefulness resembled those following cage change (CC), spontaneous dark phase waking, and after providing nest material, were enhanced in θ and fast- γ oscillatory activities. Our data suggest that Hcrtr1 Dbh-CKO mice are deficient in adaptive regulation of wakefulness in stress-associated conditions with electrocortical signs of decreased alertness and hippocampal θ activation. Furthermore, their waking θ and fast- γ ECoG power show concerted, but opposite, trends in stressful versus rewarding contexts. Finally, slow-wave-sleep (SWS) following SD and CC both showed a profound deficit in slow- δ activity, suggesting that Hcrt-to-NA signalling is critical in active waking to induce the homeostatic slow- δ rebound that characterizes ensuing SWS. loss of noradrenergic Hcrtr1 signalling in our mutant mice. Deciphering the underlying mechanisms, and potential therapeutic use of Hcrt receptor and noradrenergic pharmacological agents, will particularly benefit from a better understanding of how Hcrt neuronal firing patterns affect LC-NA cell firing patterns, and the identification of the sites at which Hcrtr1 signalling mediates its behavioral affects. show a selective deficit in slow- δ , relative to Hcrtr1 Dbh-CTR (CTR) littermates, with genotype differences becoming more pronounced under high sleep pressure, i.e. in dark phase, post-SD and post-CC SWS. ( e ) To emphasize genotype effects, SWS spectra of Hcrtr1 Dbh-CKO mice in distinct experimental contexts are referenced to spectra of Hcrtr1 Dbh-CTR mice littermates in the same contexts. The resulting CKO/CTR ECoG power ratio spectral distribution shows that CKO mice’ spectra differ most from CTR mice’ spectra in the slow- δ band, and post-SD and post-CC SWS resemble each other the most in their response to Hcrt-to-NA lesioning.


Introduction
Wakefulness is on one hand under circadian control, and on the other associated with adaptive responses to external stimuli. All these processes correlate with activity changes in cortical and subcortical neural networks that can be assessed in the electrocorticogram (ECoG). By neuromodulation of synaptic transmission throughout the brain, noradrenaline (NA) plays a major role in implementing the functional circuitry and processing modes that define the waking state.
A role of NA in regulation of cross-regional brain synchrony across cortex, striatum, thalamus and midbrain dopaminergic nuclei was suggested by using a mouse model of NA-depletion (Dzirasa et al. 2010). In these mice, NA depletion led to a profound disruption of coherence across these circuits, and to an altered behavior of the animal featuring transient hyperactivity, stereotypies and excessive grooming (Dzirasa et al. 2010). Modulation of NA signalling by pharmacological agents has also been shown to profoundly alter the waking ECoG spectral profile in rats (Sebban et al. 1999); (Berridge and Espana 2000).
The NA-producing cell group residing in the locus coeruleus (LC, or A6) produces approximately half of the brain NA content (Berridge 2008). Spontaneous tonic activity of LC-NA cells varies with behavioral states, and in wakefulness with arousal (Foote et al. 1980). Superimposed on the tonic discharge rate, exposure to novel and/or salient stimuli induces phasic bursts of activity in LC-NA cells, which differentially modulate NA release and processing in target networks (Devilbiss and Waterhouse 2011). Thus NA tone in LC-NA targets depends on LC-NA neuronal firing patterns, which in wakefulness are strongly modulated by hypocretin (Hcrt) (Walling et al. 2004). Among the wide range of afferents that project onto LC-NA cells, a prominent input indeed originates L i , F r a n k e n & V a s s a l l i | 4 effector in Hcrt-mediated arousal. Hcrt input to LC-NA has also been shown to be important in regulating wakefulness in the zebrafish (Singh et al. 2015).
Specific inactivation of Hcrtr1 signalling in NA targets remains unexplored. Here we created a mouse model with Hcrtr1 gene inactivation in NA cells, by designing a conditional, Cre-dependent Hcrtr1 allele (Vassalli et al. 2015), and crossing it with a Cre-expressing Dopamine β-hydroxylase (Dbh) PAC transgene (Parlato et al. 2007). The waking electrocorticogram of the resulting Hcrtr1 Dbh-cko mice was examined in distinct behavioral paradigms. Relative to control littermates, spontaneous baseline wakefulness of Hcrtr1 Dbh-cko mice demonstrated enhanced δ activity, suggestive of reduced alertness, while challenging behavioral contexts induced more profound spectral changes. Our study evidences the role of an intact Hcrt-to-NA signaling pathway to build an appropriate θ/fast−γ response in stress-associated environments, but also evidences circumstances in which the Hcrt-to-NA signaling pathway may serve to curb hyperarousal.

Results
Generating mice with NA cell-specific Hcrtr1 deficiency. The Hcrtr1 gene was engineered in embryonic stem cells to create a CRE-conditional Hcrtr1 knockout (KO) (floxed, Hcrtr1 flox ) allele (Fig. 1a). This allele is designed so as to be functional in absence of the CRE recombinase, while CRE mediates loxP site-recombination resulting in (i) deletion of the DNA encoding the signal peptide and N-terminal polypeptide up to within HCRTR1 transmembrane domain 3 (TM3) (126 aa in total), (ii) insertion of a promoter-less Gfp to report Hcrtr1 promoter expression, and (iii) termination of transcription downstream of Gfp by insertion of a polyadenylation signal insuring effective Hcrtr1 KO. To generate NA cell-specific HCRTR1 deficient mice, Hcrtr1 flox mice were crossed to transgenic mice expressing CRE from a Dopamine-β-Hydroxylase BAC transgene (Tg(Dbh-Cre); Parlato et al., 2007). Breedings between Hcrtr1 flox/flox mice where one parent is hemizygous for Tg(Dbh-Cre) produces two offspring groups, differing only in presence or absence of Cre: Hcrtr1 flox/flox Tg(Dbh-Cre) + mice, in which NA-cells express Cre (hereafter referred as Hcrtr1 Dbh-CKO ), and Hcrtr1 flox/flox littermates, in which NA-cells do not express Cre (Hcrtr1 Dbh-CTR , or control mice). All analyses rely on pair-wise phenotype comparison between these two groups.
Moreover, a constitutive Hcrtr1 KO allele (Hcrtr1 ∆-GFP , Fig. 1a) spontaneously emerged in the colony through expression of Tg (Dbh-Cre) in the germ-line of a minority of Hcrtr1 Dbh-CKO females, allowing us to generate a strain carrying the Hcrtr1 loxP-recombined (KO) allele in all cells. This line expresses Gfp under control of the endogenous Hcrtr1 promoter, permitting to map Hcrtr1 expression in Hcrtr1 ∆-GFP heterozygote animals (Vassalli et al. 2015); Fig. 1d).
Impaired electrocortical response to sleep deprivation (SD). A 6h period of enforced wakefulness through 'gentle handling' was initiated at light onset (ZT0) of the 3rd day, i.e. at a circadian time of maximal sleep propensity (Fig. 8a). In these 6h, relative to controls, Hcrtr1 Dbh-CKO mice displayed increased activity across an extended range of low frequencies, compared to what had been observed in baseline (Fig. 3e, compared to d). Thus, EcoG power was enhanced across a wider δ range as well as in θ (0.75-8.0 Hz), and β (13.5-20.0 Hz) frequencies (independent two-tailed student's t-test ， P <0.05). A shift of ECoG power toward lower frequencies (over the δ, θ and α ranges) is known to correlate with decreased arousal (Cajochen et al. 1995;Aeschbach et al. 1997;Phipps-Nelson et al. 2011), while γ power wanes (Lal and Craig 2001;Lal and Craig 2002). When normalized to the baseline waking ECoG, SD was moreover found to induce a lesser increase in 8.75-9.5 Hz power in Hcrtr1 Dbh-CKO mice than it did in controls, while a lesser decrease in inter-δ/θ band (4.5-7.0 Hz) activity is observed (Post-hoc independent two-tailed student's t-test，P <0.05). This indicates that the θ rhythm acceleration, and θ peak sharpening, that normally occur in active waking in rodents, are blunted in Hcrtr1 Dbh-CKO mice relative to their Cre-less littermates during SD.
Voluntary behaviors, both in threat-avoidance and appetitive arousal, correlate at the ECoG level, with the emergence of a high-power, narrow-bandwidth θ (~8-10 Hz) rhythm of hippocampal origin, as a stereotypical θ pattern becomes visible in the ECoG signal (Buzsaki, 2006). We have formulated an EEG-based algorithm to automatically score and quantify the behavioral state in which this pattern dominates (TDW, Vassalli & Franken, 2017). During SD, mice spend approximately 50% of their time in TDW (Vassalli & Franken, 2017). Hence we calculated the frequency of the θ rhythm in our mice during SD. Hcrtr1 Dbh-CKO showed a significantly slower θ peak frequency (TPF) than controls ( Fig. 3e;  Thus while SD induces the emergence of an ECoG θ synchrony in both genotypes compared to baseline waking ( Fig. 1e compared to 1d), the frequency, and the synchrony, of this rhythm appear reduced in Hcrtr1 Dbh-CKO mice relative to controls, suggesting that the hippocampal response to SD-associated arousal is blunted in Hcrtr1 Dbh-CKO mice Because TDW tends to correlate with enhanced locomotor activity (LMA), but can also uncouple not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/123075 doi: bioRxiv preprint first posted online Mar. 31, 2017; from it (Vassalli & Franken, 2017), as enhanced hippocampal θ can also be seen in immobile rats following running bouts (McFarland et al., 1975), or under motivational activation (Slawinska et al., 1998), we investigate LMA in our mice. LMA was found to be unchanged in the two baseline days, but to be markedly reduced during the 6h SD in Hcrtr1 Dbh-CKO mice relative to control littermates ( Fig. 3c; 929±120 vs. 1631±190 counts, in CKO vs. CTR; Fi; Mann-Whitney U test,

P=0.01).
We next investigated the spectral dynamics of the waking ECoG in the course of the 3-day recordings (baseline BL1+BL2 days, 6h SD and recovery, see Fig. 2) in Hcrtr1 Dbh-CKO and Hcrtr1 Dbh-CTR littermates. Because waking spectra are strongly influenced not only by environmental contingencies, but also by homeostatic sleep pressure, waking ECoG power density values in the following analyses were referenced to the waking ECoG power density values in the last 4h of baseline light phase, i.e. when homeostatic sleep pressure is minimal, following the major sleep period.
Waking ECoG power dynamics during the 3-day recordings are depicted for each genotype in

c). This analysis reveals a prominent deficit in
Hcrtr1 Dbh-CKO mice' θ rhythm regulation during the 6h SD (blue color-code), while δ and inter-δ/θ activities are markedly enhanced (orange/red color-codes), the latter in particular during the 2 nd half of the SD, as sleep pressure builds up (Vyazovskiy and Tobler 2005). Another frequency range showing a deficit in power specifically during SD, in particular its 1 st half, paralleling θ deficit, is the fast-γ (70-90 Hz) range.
Surprisingly, periods of high spontaneous activity (baseline dark phase waking maxima, Fig. 4b) were found to correlate, conversely, with enhanced θ/fast-γ activity in Hcrtr1 Dbh-CKO mice relative to controls (Fig. 4c), an observation that will be confirmed in later recording days (see Fig. 6h), and be replicated during the nesting-associated wakefulness experimental paradigm (Fig. 5g).
Of note, maxima in ECoG power ratio time-courses (in heatmap spectral representations as depicted in Fig. 4a, or in dynamic analyses of ECoG power in specific frequency ranges normalized to their power in reference spectra, as depicted in Fig. 4d), do not represent maxima in brain oscillatory power, but the frequencies at which power densities maximally differ (see SFig. 1 for an illustration).
In sum, the full adaptive response of waking electrocortical parameters normally induced by the experimenter's 'gentle handling' manipulations appears impaired in Hcrtr1 Dbh-CKO mice.
Altered electrocortical response to nesting-associated environmental enrichment, with θ/fast-γ responses inverse of those observed during SD. To explore the response of Hcrtr1 Dbh-CKO mice to rewarding stimuli, rather than stress-associated stimuli such as SD, mice were exposed to a nest building experimental paradigm. Nest construction is an extremely robust, highly motivated behavior in mice ((Gaskill et al. 2013); (Eban-Rothschild et al. 2016)). Nest not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/123075 doi: bioRxiv preprint first posted online Mar. 31, 2017; material (a Nestlet TM , see Methods) was introduced in the cage at dark onset (ZT12), and mice were left undisturbed across the night, up to light onset of the next day, at which time nest construction was assessed. All animals (CKO n=5; CTR n=5) had converted the Nestlet into an elaborated nest. Preliminary video-based behavioral analysis (in n=4/10 mice), indicated that most of nest-building activity took place in the end of the night phase, preceding the major sleep period. ECoG waking analysis thus focused on the last 6h (ZT18-24, Figs 5a, c), or last 2h (ZT22-24, Figs 5b, d), of the night. Providing nest material at the beginning of the night was found to markedly alter night phase wakefulness relative to SD-induced wakefulness, or to baseline waking. Surprisingly, wakefulness of Hcrtr1 Dbh-CKO mutant mice was found to display enhanced fast-γ (50-80 Hz) power relative to controls in those time windows , in particular in the last 2h of the night, suggesting that Hcrt-to-NA signalling normally has a negative effect on fast-γ oscillations in this behavioral context. The spectral dynamics of wakefulness across the nesting night and the following 12h light phase were next examined. This confirmed the occurrence of prominent fast-γ power maxima (50-90 Hz) in Hcrtr1 Dbh-CKO mice, which could be seen to occur mostly in the beginning and the end of the night in which nest material was available (Fig. 5g). Enhanced fast-γ activity was furthermore seen to correlate in its time-course with a θ rhythm enhancement. Enhanced θ rhythm and fast-γ power in Hcrtr1 Dbh-CKO mice in the last 2h of the night (ZT22-24) furthermore correlated with a higher waking content (min/h) relative to control mice ( Fig. 5f left), that might correlate with an increased nest-building activity. The following light phase also displayed three, shorter-lived, peaks of enhanced fast-γ activity, also temporally matched with peaks in θ rhythm.
The differential spectral dynamic analysis of wakefulness also confirmed Hcrtr1 Dbh-CKO mice' enhanced slow-γ (30-50 Hz) activity compared to CTR in the nesting paradigm. Periods of higher slow-γ (30-50 Hz) power in CKO mice were seen to occur prominently in the first half of the night, and the first half of the following light phase.

Impaired electrocortical response to transfer from nesting cage to a new environment.
After two intervening undisturbed days, mice were removed from their home cage where they had built their nest, and transferred to a fresh cage at ZT3, a time at which mice spontaneously mostly sleep (Fig. 8a). Cage change (CC) induces a powerful arising effect, and, although Hcrtr1 Dbh-CKO mice tended to show reduced sleep latency following CC relative to control littermates (Fig. 6b), the two groups responded with a similar increase in time-spent-awake compared to their baseline values. The spectral quality of the waking ECoG was analyzed from CC not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The characteristic θ peak that defines active waking was apparent in both genotypes, but markedly blunted in Hcrtr1 Dbh-CKO mice, which, similarly as observed during SD, displayed enhanced 3.25-6.25 Hz range activity, suggestive of reduced alertness. When CC-induced wakefulness was contrasted to wakefulness of the preceding 24h in the home cage ( Fig. 6a bottom), and compared in the two genotypes (two way ANOVA: genotype: F 1, 3276 =180.562; P <0.001; interaction: F 272, 3276 =1.941; P <0.001), Hcrtr1 Dbh-CKO mice displayed a lesser decrease in inter-δ/θ band activity than controls, again consistent with a partially failed narrowing, and acceleration, of θ oscillatory frequencies (see also Suppl. Table 2 for θ peak frequency values).
Furthermore, the fast-γ response to CC across a wide frequency range (53-80 Hz) was profoundly diminished in Hcrtr1 Dbh-CKO mice (Fig. 6a bottom; post-hoc independent two-tailed student t-test, P <0.05). This effect is clearly apparent in heatmap representations of the waking ECoG spectral dynamics from CC to SWS-onset (Figs 6c-d). Thus, the fast-γ power response of Hcrtr1 Dbh-CKO mice following CC is inversed relative to the one observed in these same mice, two days beforehand during the nesting night. This suggests that perception of the two behavioral contexts that are nesting in the dark phase, and forced exploration of a novel environment in the light phase, profoundly differ in their processing by the Hcrt-to-NA pathway, as further discussed below.
The heatmap representations of the spectral dynamics of CC-to-SWS wakefulness (Fig. 6c), and differential CKO-CTR dynamics thereof (Fig. 6d), also powerfully illustrate the blunted θ rhythm response of Hcrtr1 Dbh-CKO mice to CC, compared to the Hcrtr1 Dbh-CTR mice' response.
Consistently with reduced θ synchrony in Hcrtr1 Dbh-CKO mice, the decrease in inter-δ/θ band activity relative to its value at ZT8-12 of the preceding light phase, that is evident in controls ( Fig. 6c right), is blunted in Hcrtr1 mutant mice (Figs 6c left, and 6d).
We then performed a more specific dynamic analysis of ECoG power within specific frequency ranges. Based on their evident modulation in the course of the CC-to-SWS-onset waking paradigm as assessed in the heatmaps, three frequency ranges were analyzed (3.5-6.0 Hz (interδ/θ band), 7.5-11.5 Hz (θ), and 65.0-75.0 Hz(fast-γ); Fig. 6e). As time progresses after CC, novelty decreases and sleep pressure increases, the 3.6-6.0 Hz activity increased in both genotypes, while the θ and fastγ activities decreased in both genotypes. At all time-points, however, and increasingly so at time progressed following CC, the 3.6-6.0 activity was higher in CKO than in CTR mice, while the θ and fast-γ activities were lower in CKO than in CTR littermates (independent two-tailed student's t-test ， P <0.05). In sum, although both Hcrtr1 Dbh-CKO and Hcrtr1 Dbh-CTR mice display a dramatic induction of θ synchrony, with a narrowing and acceleration of the θ rhythm, and of fast-γ oscillatory power, following CC ( Fig. 6c), both these processes appeared significantly impaired in Hcrtr1 Dbh-CKO mice (Fig. 6d).
To confirm that our unexpected observations of an opposite regulation of fast-γ oscillations in not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/123075 doi: bioRxiv preprint first posted online Mar. 31, 2017; waking of Hcrtr1 Dbh-CKO mice in SD and after CC, on one hand (stress-associated stimuli), and in spontaneous and nesting-induced waking, on the other hand, was stable through time, we extended our spectral dynamic analysis of wakefulness throughout the day preceding CC, and the entire CC day. This revealed that spontaneous dark phase waking (in particular in ZT12-18) in the night preceding CC, and the night following CC, as had been observed for baseline recording days 1 and 2, witness a concerted surge in θ and fast-γ power, present in both genotypes, but enhanced in Hcrtr1 Dbh-CKO animals ( Fig. 6f-h).
Slow-wave-sleep of Hcrtr1 Dbh-CKO mice, in particular following active waking, is markedly deficient in the slow-δ oscillatory component. Because SWS quality critically depends on preceding wakefulness quality, in particular on prior time animals spent in the TDW waking substate (Vassalli & Franken, in revision), we next examined SWS subsequent to the distinct behavioral contexts the mice were exposed to. We thus analyzed the mice' global baseline (48h) SWS spectral profiles (Fig. 7a), the profiles of the first 20 min of SWS following baseline light-onset ( Fig. 7c), i.e. when sleep homeostatic drive is maximal, after the mice' major waking period ( Fig. 3a), the first 20 min of SWS following SD (Fig. 7b), and following CC-induced waking ( Fig. 7d). A slow (~0.75-2.00 Hz) and fast (~2.50-4.25 Hz) ranges of δ power were seen to respond differently in Hcrtr1 Dbh-CKO mice, with a more pronounced deficiency in slow-δ than in fastδ power, respective to controls. While this deficit was visible but below significance in baseline, it became more pronounced in post-SD and post-CC SWS, suggesting that SWS subsequent to stress-associated active waking is more dependent on Hcrt-to-NA signalling than baseline waking is. In SWS following nesting-induced waking we did not observe a δ, or selective slow-δ, power deficit (not shown). Post-SD SWS of Hcrtr1 Dbh-CKO mice moreover displayed enriched β and slow-γ activities (Fig. 7b). Increased β power during non-REM sleep is reported in insomnia in humans (Spiegelhalder et al., 2012).
SWS spectra of Hcrtr1 Dbh-CKO mice in these distinct experimental contexts were next plotted in reference to the spectra of control littermates in the same contexts (Fig. 7e). The resulting CKO/CTR ECoG power ratio spectral distribution confirms that the major genotype differences in SWS quality reside in the slow-δ band. Moreover, the altered spectral responses of CKO mice' SWS relative to CTR mice' SWS were seen to closely overlap, and be more pronounced in post-SD and post-CC SWS, than they were in baseline SWS, emphasizing again that SD and CC similarly affect electrocortical activity across behavioral states, and similarly rely on Hcrt-to-NA signalling.
Time-course analyses of SWS slow-δ and fast-δ power (Fig. 8b) across the 3-day recording revealed that at all time points slow-δ power is dimished in Hcrtr1 Dbh-CKO mice relative to controls, however this deficit is enhanced following major waking periods (at the end of the night, and after SD, Fig. 8b; and after CC-induced waking (Fig. 8d).
The significance of different δ oscillatory frequency ranges in SWS, and their potential impact on homeostatic recovery, remain unclear. However, a correlation of the slow-δ component of SWS not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/123075 doi: bioRxiv preprint first posted online Mar. 31, 2017; with prior active wakefulness, TDW and potentiation events was reported in several studies, including by ourselves in Hcrt-KO mice (Vassalli & Franken), and is further discussed below. Our present data suggest that events associated with active waking critically require Hcrt-to-NA signalling for subsequent SWS expression of a slow-δ oscillatory component.

Discussion
Here we record brain activity in a novel mouse line engineered to carry a non-functional Hcrtr1 gene in NA cells, in four distinct behavioral contexts (spontaneous baseline waking, gentle handling-enforced waking, nesting-associated, and CC-induced waking), and follow their electrocortical and behavioral state response. Relative to control littermates, the spectral oscillatory profile of Hcrtr1 Dbh-CKO mice' wakefulness is more extensively altered as mice are placed under behavioral challenge. Moreover the electrocortical alterations are found to be highly context-dependent, with, in particular, an opposite regulation of fast-γ oscillations in distinct contexts. Because these changes were robust and reproducible, we suggest they may eventually inform on novel aspects of noradrenergic regulation of behaviors and affects.
The TDW waking substate (Vassalli & Franken) (or 'active waking') is characterized at the ECoG level by a narrowing and frequency increase of the θ oscillatory peak, and enhanced fast-γ activity, as the mouse gets aroused. Our most salient finding is that the induction of these two read-out's of arousal is markedly diminished in Hcrtr1 Dbh-CKO mice relative to control littermates in SD and following CC, whereas in spontaneous dark phase and nesting-associated wakefulness, the θ and fast-γ oscillatory responses now show an inverse relation to controls, i.e. show marked enhancement in Hcrtr1 Dbh-CKO mice.
Terminology. Waking in rodents is often partitioned in two broad substates, quiet and active waking (McGinley et al. 2015). In quiet waking and automatic behaviors, activity distributed across the 4.0-7.0 Hz range is known to increase with sleep drive, and perception of sleepiness ((Franken et al. 1991a;Vyazovskiy and Tobler 2005)and see below. Although it is commonly referred to as 'θ', we here refer to it as the 'inter-δ/θ' frequency range, to distinguish it from the hippocampally-driven θ rhythm of active wakefulness (Buzsaki 2002). In voluntary behaviors, such as in explorative locomotion, foraging, but also in alert immobility, a higher-frequency (7-10 Hz), narrow-bandwidth θ rhythm of hippocampal origin appears, as the raw EEG signal displays a regular, stereotypical rhythmic pattern (McFarland et al. 1975;Slawinska and Kasicki 1998), that we designed an algorithm to recognize, and define as the TDW substate (Vassalli & Franken).
Altogether, relative to Hcrtr1 Dbh-CTR littermates, Hcrtr1 Dbh-CKO mice' wakefulness manifested: (i) a slowing of the electrocorticogram, with enhanced power in the δ and inter-δ/θ ranges, (ii) enhanced β and slow-γ activity, particularly pronounced in nesting-associated-waking, (iii) impaired θ rhythm induction under SD and following CC. not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
(v) Conversely, fast-γ power is enhanced in spontaneous dark phase and nesting behaviors.
Hcrt-to-NA signalling deficiency causes a slowing of the ECoG in stress-associated contexts (SD and CC). Baseline waking of Hcrtr1 Dbh-CKO mice shows enhanced power in a restricted, slow-δ (1.25-2.75 Hz), range, while SD, nest material provision and CC induce ECoG alterations over the entire spectral range, encompassing the δ, inter−δ/γ , β, slow-and fast-γ ranges. SD and CC induce enhanced δ and inter-δ/θ activities, both of which are reported to be markers of homeostatic sleep pressure in the waking ECoG of rodents and humans (Franken et al. 1991b;Cajochen et al. 1995;Finelli et al. 2000;Huber et al. 2000;Vyazovskiy & Tobler 2005),

2015)
Whether enhanced β and slow-γ activity in Hcrtr1 Dbh-CKO mice reflects dopaminergic deregulation is unknown, but its association with a motor impairment can be tested in our mice, and is supported by the fact that Hcrt-KO mice also show enhanced slow-γ (30-47.5 Hz) activity in wakefulness of the minute that precedes cataplexy attacks (Vassalli et al. 2013). The latter are most often triggered during motivated behaviors including nest-building, and the postural collapse that characterizes them is sometimes preceded by evident impaired motor control, such as leg stumbling, thought to represent partial cataplexies. Our study suggests that some of these effects critically rely on Hcrt input.
Our observation of impaired θ rhythm induction, but enhanced β/slow-γ activity, in Hcrtr1 Dbh-CKO mice during SD corroborate a rat study in which stimulation of the LC was found to enhance θ activity, but suppress 12 to 40 Hz frequencies, through a pathway involving inhibition of a type of not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
Inversed θ/fast-γ responses of Hcrtr1 Dbh-CKO mice in SD and CC vs. spontaneous dark phase and nesting-associated waking. Relative to controls, Hcrtr1 Dbh-CKO mice display lower fast-γ activity during SD and post-CC wakefulness, while spontaneous dark phase waking, and increasingly so dark phase waking in presence of nest material, display higher fast-γ power. What could explain these opposite θ/fast-γ responses in distinct behavioral contexts? Hcrtr1 signalling has been implicated in fear and fear memory, in particular for memories with a high emotional component (Soya et al. 2013 One working hypothesis therefore would be that loss of Hcrtr1 expression in LC-NA cells causes blunting of the fear response, and of the θ/fast-γ oscillatory activation associated with frightened arousal, upon human manipulation (during SD), or exposure to a foreign environment (after CC), reflected in the electrocorticogram with failed θ and fast-γ responses.
In contrast, in spontaneous dark phase and nesting behaviors, the θ rhythm and fast-γ responses may be mediated by other pathways, that are preserved in Hcrtr1 Dbh-CKO mice. Because nesting and spontaneous dark phase behaviors are rewarding (Gaskill et al. 2013), induce arousal, the TDW state, and associated θ/fast-γ oscillatory activation, a reduced fear response may permit expression of enhanced behavioral activation during nest-building, reflected in the enhanced θ and fast-γ response of Hcrtr1 Dbh-CKO mice in the two latter contexts. Supporting this argument, hippocampal θ is induced not only by locomotion, but also by motivational activation (Slawinska and Kasicki 1998). This raises the interesting possibility that an enhanced level of θ/fast-γ oscillatory activation in Hcrtr1 Dbh-CKO mice may, in some circumstances, lead to hyperarousal.
Hcrt-to-NA signalling deficiency did not cause overt cataplexy. Hcrtr1 Dbh-CKO mice did not demonstrate obvious electrocortical, or behavioral, evidence of cataplexy (Chemelli et al. 1999); (Vassalli et al. 2013). A strong body of evidence in the canine model of narcolepsy implicated the NA system in the control of cataplexy (Nishino and Mignot 1997), leading to the noradrenergic hypoactivity hypothesis. Although a more thorough, video-based, behavioral assessment of our mice is warranted, our preliminary observation is in concordance with the finding that restoration of Hcrtr1 in LC of mice lacking Hcrt receptors did not suppress cataplexy (Hasegawa et al. 2014).
Also, genetic canine models of narcolepsy are caused by loss-of-function mutations in Hcrtr2 and not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/123075 doi: bioRxiv preprint first posted online Mar. 31, 2017; not Hcrtr1 (Lin et al. 1999). One explanation may be that, an yet unknown Hcrtr2-expressing cell population critically controls LC-NA neurons and cataplexy.
A limitation of the present work is that Hcrtr1-mediated signalling is inactivated in our mice as soon as Dbh is expressed, i.e. around mid-embryonic development (Cre-mediated activity was first evidenced in Tg(Dbh-Cre) mice at E11.5; (Parlato et al. 2007). Thus the effects we observe may differ, and may be milder, from those of acute NA cell-specific Hcrtr1 inactivation. Exploring the latter is feasible in our mice through local viral transduction of a NA cell-specific Cre-expressing vector in the LC, or another NA cell group that express Hcrtr1, in adult Hcrtr1 flox/flox mice.
Hcrt-to-NA signalling deficiency affects the spectral quality of SWS. How does Hcrt-to-NA signalling relate to wake/sleep homeostatic processes? Our work revealed that this pathway is essential during waking in stressful conditions for normal expression of slow-δ oscillations in ensuing sleep.
A selective deficit in slow-δ is also seen in SWS of Hcrt ko/ko mice following spontaneous baseline waking (Vassalli & Franken), as well as in SWS of rats following SD if they were depleted in cortical NA (Cirelli et al. 2005). These rats furthermore display reduced cortical expression of synaptic activity-associated genes during SD waking (Cirelli et al. 2004), as do Hcrt-KO mice in spontaneous baseline dark phase (Vassalli & Franken). The significance for sleep recovery function of a specific slow-δ deficit in SWS, is unclear, but the above studies suggest that slow-δ power in SWS is dependent upon NA-dependent, and possibly Hcrt-dependent, potentiation events in preceding wakefulness. Because θ activity is known to be associated with heightened synaptic plasticity and memory-related potentiation events, and long-term potentiation (LTP) in the hippocampus is optimally induced by θ frequency stimulation (Larson et al. 1986), reduced θ promotion under SD and CC in Hcrtr1 Dbh-CKO mice may also cause a deficit in hippocampal potentiation events, and thus affect learning and memory.
We previously found that time spent in TDW state is a predictor of sleep homeostatic drive, as δ power in ensuing SWS is correlated with prior time spent in TDW, rather than in all waking (Vassalli & Franken). Hcrt-deficient mice, which are profoundly impaired in maintaining TDW in baseline conditions, express a subsequent SWS markedly diminished in δ power, in particular in the slow-δ range (Vassalli & Franken, 2017). Thus impaired slow-δ in SWS of Hcrtr1 Dbh-CKO mice following SD, may be related to their deficit in θ rhythm sharpening and frequency during SD.
Supporting this, TDW state expression tends to be reduced in Hcrtr1 Dbh-CKO mice during SD, although this variable failed statistical testing in our mouse sample.
Using simple electrocortical indices of the waking state and SWS quality, we here provide suggestive evidence of the functional relevance of the Hcrt-to-NA connection in adaptive regulation of behaviors amid changing contexts. We expect many behaviors to be affected by the not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/123075 doi: bioRxiv preprint first posted online Mar. 31, 2017; loss of noradrenergic Hcrtr1 signalling in our mutant mice. Deciphering the underlying mechanisms, and potential therapeutic use of Hcrt receptor and noradrenergic pharmacological agents, will particularly benefit from a better understanding of how Hcrt neuronal firing patterns affect LC-NA cell firing patterns, and the identification of the sites at which Hcrtr1 signalling mediates its behavioral affects. not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.

Methods
Evidence of tissue-specific, Cre-dependent Hcrtr1 flox allele recombination. We designed 2 PCR assays flanking Hcrtr1 flox loxP sites, allowing amplification of a diagnostic fragment from the recombined allele, and a 1.1 kb larger fragment (inter-loxP distance) from the non-recombined allele. The corresponding recombined fragments were produced from DNA prepared from tissue slices containing the LC (250 um X 1mm X 1mm tissue piece) from Hcrtr1 Dbh-CKO animals, but not from Hcrtr1 Dbh-CTR littermates, or not from their ear DNA. PCR products were gel purified and the not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/123075 doi: bioRxiv preprint first posted online Mar. 31, 2017; loxP recombined junction was sequenced in both orientations, allowing to confirm accurate recombination at the nucleotide level in 2 Hcrtr1 Dbh-CKO animals.

Generation of the Hcrtr1 KO/GFP reporter (Hcrtr1 ∆-Gfp ) allele. Spontaneous expression of
Tg(Dbh-Cre) in the germ-line was found to occur in a minority of Hcrtr1 flox/flox +/Tg (Dbh-Cre) females, allowing us to generate a line carrying the recombined allele derived from Hcrtr1 tm1.1Ava in the germ-line (Hcrtr1 ∆-GFP , or Hcrtr1 tm1.2Ava MGI: 5637400). Accuracy of the loxP site recombination at the Hcrtr1 ∆-Gfp allele was verified by sequencing of genomic DNA from ear biopsies of 2 male mice homozygous for the mutation.
Mice. All ECoG data are derived from 10-13 week-old males (weight 27-31 g). Mice were individually housed with food and water ad libitum under a 12h light/12h dark cycle (lights on at 8:00 am, defined as Zeitgeber Zero, ZT0). All experiments were performed in accordance with Swiss federal legislation, using protocols approved by the State of Vaud Veterinary Office,

Switzerland.
ECoG/EMG, video and locomotor activity (LMA) recording. Electrodes for differential frontoparietal EEG and EMG recording were implanted as described (Mang and Franken 2012).
Hardware (EMBLA A-10) and software (Somnologica-3) were from Medcare Flaga (EMBLA, Thornton, USA). EEG and EMG signals were amplified, filtered, analog-to-digital converted, and stored at 200 Hz. Each 4-s epoch was assigned a behavioral state score using published criteria (Franken et al. 1998). Behavioral monitoring was performed using infrared video-cameras allowing close frontal viewing. Locomotor activity was assessed at 1-min resolution using topmounted passive infrared motion detectors (Visonic Ltd, Tel Aviv, Israel) and analyzed with ClockLab software (ActiMetrics, IL, USA).
ECoG power spectral density analysis. The ECoG signal was subjected to discrete Fourier transform to yield power density spectra (0.75-90 Hz) in 4s consecutive epochs (0.25 Hz frequency resolution; Hamming window function). Values between 47.5 to 52.5 Hz were excluded because of the 50 Hz AC power line noise. Higher-frequency artefacts (e.g. 71.25-77.25 Hz) of unknown origin were present in some mice (SD: 6/16; 3 CKO, 3 CTR; CC: 4/14; 1 CKO, 3 CTR; Nest experiment: 4/10; 1 CKO; 3 CTR), and values in these frequency ranges were removed from spectral analysis. Artefacted epochs were tagged and removed from spectral analysis but not from time-spent-awake/-in-SWS, when behavioral state could be assessed. For each behavioral state and animal, a mean ECoG spectrum was obtained by averaging the spectra of all epochs that were artefact-free and flanked by the same state (Franken et al. 1998). To account for interindividual differences in overall ECoG power, ECoG spectra of were expressed as a percentage of an individual reference value calculated as the total ECoG power across frequencies (0.75-40 Hz) not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/123075 doi: bioRxiv preprint first posted online Mar. 31, 2017; and behavioral states. This reference was weighted so the relative contribution of each state is identical in all animals (Franken et al. 1998). Reference values for the two genotype did not differ (Hcrtr1 Dbh-CKO vs. Hcrtr1 Dbh-CTR 1181± 127 uV 2 vs. 1181± 120 uV 2 , t 14 = 0.00105, P=0.99). The dominant frequency of θ oscillations (θ peak frequency, TPF) was determined for each animal as the frequency at which ECoG power density of the mean spectral profile was highest in the 5-10 Hz range during waking (Hasan et al. 2009;Mang et al. 2016).
Baseline and sleep deprivation (SD)-induced wakefulness: '3-day recordings' consisted of two undisturbed days (BL1+BL2), followed by a 6h SD initiated at light onset (ZT0) of the 3 rd day. SD was achieved by 'gentle handling' (Mang and Franken 2012). To further stimulate behavioral activity during SD, tissue paper was introduced in homecage at ZT2 and ZT5, and midway through SD (ZT3), mice were transferred into a fresh cage. Following SD, mice were left undisturbed for the rest of the day, and for at least another 2 intervening days.
Nest material interaction experiment: Nest material previously shown to trigger cataplexy in Hcrtdeficient mice (Vassalli et al. 2013) was used, i.e. Nestlets TM (Ancare, Ref. Nr. 14010), made of highly shreddable cotton. One Nestlet was introduced in the cage at dark onset (ZT12) of day 6, and mice were left undisturbed until next light-onset and nest-building assessment. Mice were then left undisturbed for another ≥2 more days before the next experiment.
Cage change (CC): On ZT3 of day 9, mice were transferred from their homecage where the nest had been built to a fresh cage. Latency to SWS onset was defined as the time until the first SWS episode lasting ≥2 min and interrupted by ≤3 epochs scored as wakefulness (Hunsley et al.
Secondary antibodies were donkey IgGs coupled to Alexa dyes used at 1:500 for 1h at r.t.
Images were acquired on an inverted Zeiss LSM710 confocal laser-scanning microscope using ZEN, and analyzed using ImageJ.
Statistics. Data were tested for normal distribution using Shapiro-Wilk normality test. If the test passed, genotype differences in TPF and accumulated LMA during SD, or ECoG power within specific frequency bands were evaluated by independent two-tailed student's t-test. Otherwise, nonparametric Mann-Whitney test was used. Time-course of hourly values of W, SWS, PS, LMA, ECoG power spectra, were analyzed by two-way ANOVA with factors "genotype" and "time-ofday", or "genotype" and "frequency bin", and their interaction. When significance was reached for the main factors and/or their interactions, genotype contrasts were assessed by unpaired, twotailed Student's t tests. Within-group effects, such as response to SD or nest material, were assessed using paired, two-tailed student's t-tests. not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/123075 doi: bioRxiv preprint first posted online Mar. 31, 2017;  n=7) were sequentially exposed to distinct behavioral paradigms. Spontaneous waking ECoG is analyzed in two baseline (undisturbed) days, followed by a 6h 'gentle handling' sleep deprivation (SD), initiated at light onset (ZT0) of the 3 rd day. After 2 intervening days, nest material (+Nestlet) is introduced at dark onset (ZT12), and the waking ECoG analyzed in the following 24h. After an additional 2 intervening days, the mouse is removed from its nesting cage, and transferred to a fresh cage 3h after light onset (ZT3). Post-CC waking is analyzed until the next SWS onset (see Methods). Shaded areas represent the 12h dark phases. not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure Legends
The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/123075 doi: bioRxiv preprint first posted online Mar. 31, 2017; are referenced to each mouse' total ECoG power across all states and frequencies in baseline (Franken et al. 1998). Note non-linear axes. Blown-up views across 0.75 to 15 Hz (d), and 6.5 to 10 Hz (e), with linear axes, are shown in insets. Baseline dark phase waking shows enhanced δrange activity in CKO mice relative to CTR, while SD-waking shows an extended range of power increase, over δ, inter-δ/θ band and β-range activities. SD-waking displays a slowing of the θ rhythm in CKO relative to CTR mice (e inset). θ peak frequency (TPF) value (mean±SEM) is shown for CKO (black symbol) and CTR (grey) mice. Red bars and asterisk indicate significant genotype differences (two-way ANOVA, followed by post-hoc independent two-tailed student ttest, P<0.05). recording. To emphasize genotype differences, the two spectral dynamics shown in (a) were substracted from each other. White lines mark significant genotype differences (P=0.05). Note that activity in the inter-δ/θ band is increased more in CKO mice than in CTR, in particular in the 2 nd half of the SD. θ frequency narrowing and acceleration, as well as activity in the 70-90 Hz range, are reduced in CKO relative to CTR, in particular in the 1st half of the SD. Importantly, note that the color codes for (c) and (d) differ in the code indicating 'no change': in (c), a 100% reference value is coded as dark-blue. In (d), a zero value (CKO=CTR) is coded as light-green. (d) Time-course of ECoG power within specific frequency ranges are plotted across time relative to its mean value in baseline ZT8-12 wakefulness [%]. indicates a 2-fold change. Red bars indicate significant genotype differences between genotypes (two-way ANOVA, followed by post-hoc independent two-tailed student t-test, p<0.05). (e) Heatmaps representing the waking spectral dynamics across the nesting night and the following not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/123075 doi: bioRxiv preprint first posted online Mar. 31, 2017; light phase were constructed as for the B1-B2-SD-day recording shown in Fig. 4a. Plotted are ECoG power densities for each 0.25 Hz frequency bin relative to their values in wakefulness of baseline (B1+B2) ZT8-12 periods [%]. (g) A differential CKO-CTR heatmap of the nesting 24h period was constructed similarly to Fig. 4g for the B1-B2-SD recording. Note that the color code for (e) and (g) differ. In (e), 100% reference value is coded as dark-blue. In (g), a zero value (CKO=CTR) is coded as light-green. (f) Time-courses of time-spent-awake [min/h]. Red bar indicates significant genotype differences between genotypes (two-way ANOVA, followed by posthoc independent two-tailed student t-test, p<0.05). Red bars indicate significant genotype differences (two-way ANOVA, followed by post-hoc independent two-tailed student t-test, P<0.05). (b) Latencies to SWS-onset tended to be shorter in CKO than CTR mice.
(c). Spectral dynamics of the waking ECoG after cage change, across time (ZT3 to SWS-onset), frequencies (0.75 -90.0 Hz), and power (as % of each mouse' mean waking ECoG power in that frequency bin in the preceding day's ZT8-12 time interval; plotted as log 2 value and color-coded as shown on the right). x-axis shows ZT time, adjusted to individual waking latencies to SWSonset. (d) Differential CKO-CTR waking power dynamics are shown, similarly as in Fig. 4c and Fig.   5g. Note that the color code for (c) and (d) differ. In (c), 100% reference value is coded as darkblue. In (d), a zero value (CKO=CTR) is coded as light-green. (e) Time-courses of waking ECoG power over the CC-to-SWS-onset waking period within specific frequency ranges that showed modulation in c-d (3.5-6.0 Hz; 7.5-11.5 Hz; 65-75 Hz). Red stars, significant genotype differences (independent two-tailed student t-test, p<0.05). Black star, marginally significant genotype difference (independent two-tailed student t-test, P=0.051). The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/123075 doi: bioRxiv preprint first posted online Mar. 31, 2017; show a selective deficit in slow-δ, relative to Hcrtr1 Dbh-CTR (CTR) littermates, with genotype differences becoming more pronounced under high sleep pressure, i.e. in dark phase, post-SD and post-CC SWS. (e) To emphasize genotype effects, SWS spectra of Hcrtr1 Dbh-CKO mice in distinct experimental contexts are referenced to spectra of Hcrtr1 Dbh-CTR mice littermates in the same contexts. The resulting CKO/CTR ECoG power ratio spectral distribution shows that CKO mice' spectra differ most from CTR mice' spectra in the slow-δ band, and post-SD and post-CC SWS resemble each other the most in their response to Hcrt-to-NA lesioning.  do not represent peaks of ECoG oscillatory activity. To contrast two power spectra, or to depict a spectral profile relative to another (reference) spectral profile (e.g. ECoG waking spectrum during SD vs. waking spectrum in baseline), the ratio of the two spectra can be plotted as a spectral distribution which should be interpreted with caution: a peak in this spectral distribution represents a frequency at which the two spectra differ maximally. This peak should not be confused with the peak of the oscillations on which the two spectra being compared are based. Illustrating this, the mean ECoG power spectrum during SD-wakefulness is shown in color, and the reference waking spectrum to which it is contrasted (reference), i.e. waking in the last 4h of baseline light phase (BL ZT8-12) is shown in grey. (Here, mean spectra of n=7 Hcrtr1 Dbh-CTR mice are depicted as an example). SD induces the emergence of a sharp θ rhythm, with a peak frequency at 8.5 Hz. (b) depicts the ECoG power ratio between these two spectra, revealing a peak at 9.25 Hz, the frequency at which the two spectra differ the most. This peak does not evidence any 9.25 Hz brain oscillation. not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.