The Clock Mechanism Influences Neurobiology and Adaptations to Heart Failure in Clock∆19/∆19 Mice With Implications for Circadian Medicine

In this study we investigated the role of the circadian mechanism on cognition-relevant brain regions and neurobiological impairments associated with heart failure (HF), using murine models. We found that the circadian mechanism is an important regulator of healthy cognitive system neurobiology. Normal Clock∆19/∆19 mice had neurons with smaller apical dendrite trees in the medial prefrontal cortex (mPFC), and hippocampus, showed impaired visual-spatial memory, and exhibited lower cerebrovascular myogenic tone, versus wild types (WT). We then used the left anterior descending coronary artery ligation model to investigate adaptations in response to HF. Intriguingly, adaptations to neuron morphology, memory, and cerebrovascular tone occurred in differing magnitude and direction between Clock∆19/∆19 and WT mice, ultimately converging in HF. To investigate this dichotomous response, we performed microarrays and found genes crucial for growth and stress pathways that were altered in Clock∆19/∆19 mPFC and hippocampus. Thus these data demonstrate for the first time that (i) the circadian mechanism plays a role in neuron morphology and function; (ii) there are changes in neuron morphology and function in HF; (iii) CLOCK influences neurobiological gene adaptations to HF at a cellular level. These findings have clinical relevance as patients with HF often present with concurrent neurocognitive impairments. There is no cure for HF, and new understanding is needed to reduce morbidity and improve the quality of life for HF patients.

Ischemic heart disease leading to myocardial infarction (MI, heart attack) and heart failure (HF) is a leading cause of morbidity and mortality worldwide 1 . Cognitive impairment, depression or brain changes are frequently observed in patients with HF [2][3][4] . Here we investigated a role for the circadian mechanism in neurocognitive impairments in HF, using murine models. In support of this approach, recent experimental and clinical studies show that the circadian mechanism is important for cardiovascular physiology. It underlies daily rhythms in heart rate 5 , blood pressure 6 , and organizing of the cardiac genome [7][8][9][10][11][12][13][14] , proteome [15][16][17] , contractility and metabolism (reviewed in 11,[18][19][20]. Moreover, we and others have shown that the circadian mechanism contributes to remodeling in ischemic heart disease and HF 9,12,[21][22][23][24][25][26][27][28][29][30][31][32] . However, how the circadian mechanism affects the neurobiological adaptations in HF is not known, in large part because studies have focused only on the heart and have not investigated what happens concurrently in the brain. The circadian system coordinates our physiology with the diurnal environment -mammals are awake in the day or at night (reviewed in 33,34 ). Briefly, the system is hierarchically orchestrated; time-setting light cues interaction between effects of HF and distance from the soma (two-way ANOVA, F(14,105) = 1.6, P = 0.1). In contrast, apical dendrite length is not different in HF than at baseline for CA1 neurons (two-way ANOVA, F(1, 102) = 0.2, P = 0.7) (Fig. 2b, right). Thus these data show that in WT mice, HF is associated with smaller apical dendrite trees in the mPFC layer 2/3 neurons but not in hippocampus CA1 neurons.
Next, we determined what happens in Clock ∆19/∆19 mice with HF. Intriguingly, apical dendrite length is greater in HF than at baseline in mPFC neurons (two-way ANOVA, F(1, 105) = 4.6, P = 0.03) (Fig. 2c, left), with no interaction between the effects of HF and distance from the soma (two-way ANOVA, F(14, 105) = 1.0, P = 0.4). Apical dendrite length is also greater in HF than at baseline in hippocampus CA1 neurons (two-way ANOVA, F(1, 102) = 6.5, P = 0.01) (Fig. 2c, right), with no interaction between the effects of HF and distance from soma (two-way ANOVA, F(16, 102) = 0.2, P = 1.0). Thus these data show that in Clock ∆19/∆19 mice, HF leads to larger apical dendrite trees in both the mPFC layer 2/3 neurons and in hippocampus CA1 neurons, and that this response is different than what happens in HF WT mice.
Finally, we compared the direction of change in apical dendrite morphology in Clock ∆19/∆19 versus WT mice in HF. We found that the Clock ∆19/∆19 HF mice exhibit an increase in mPFC apical dendrite length, especially in the first 100 μm from the soma, whereas the WT HF mice exhibit a large decrease in mPFC apical dendrite length near the soma (Fig. 2d, left). Intriguingly, despite the significant difference in tree size between the genotypes at baseline (Fig. 2a), there is convergence to similar end lengths across genotypes in HF (two-way ANOVA, F(1, 120) = 0.4, P = 0.5). These effects were also observed in the CA1 neurons of the hippocampus in the HF mice, as Clock ∆19/∆19 HF mice exhibit an increase in apical dendrite length at all distances from the soma, whereas WT mice exhibit a decrease in apical dendrite length within 125 μm of the soma, and increases at further distances www.nature.com/scientificreports www.nature.com/scientificreports/ Figure 2. Apical dendrite morphology differs in Clock ∆19/∆19 vs. WT mice, baseline and HF. (a) Apical dendrite morphology was analyzed for mPFC layer 2/3 and hippocampus CA1 pyramidal neurons using a modified three-dimensional Sholl analysis. This analysis measured the length of apical dendrite between concentric spheres radiating outward from the soma. (a) Normal Clock ∆19/∆19 mice have less apical dendrite length compared to WT mice in the mPFC (left, P = 0.0001) and the hippocampus (right, P = 0.001). (b) WT mice with HF exhibit decreased apical dendrite length in the mPFC (left, P = 0.003), but no change in apical dendrite length in hippocampus (right), as compared to non-HF WT controls. In contrast, (c) Clock ∆19/∆19 mice with HF exhibit increased apical dendrite length in the mPFC (left, P = 0.03), and increased apical dendrite length in the hippocampus (right, P = 0.01), versus Clock ∆19/∆19 controls. (d) Thus HF is associated with changes in apical dendrite length, and the direction and magnitude of change is different in Clock ∆19/∆19 HF versus WT HF mice, in mPFC (left) and hippocampus (right) neurons. For mPFC: n = 4 mice per baseline group, n = 5 mice per HF group. For hippocampus: n = 4 mice per baseline group, n = 4 mice per HF group. Four neurons were traced and averaged for each mouse, and data are shown as mean ± SEM. *Indicates P < 0.05 by Bonferroni post-hoc analysis.
www.nature.com/scientificreports www.nature.com/scientificreports/ (Fig. 2d, right). As with the mPFC neurons, these differing directions of change in dendrite length in the CA1 hippocampus neurons leads to a convergence of similar-sized apical dendrite trees in each genotype with HF (two-way ANOVA, F(1, 102) = 0.1, P = 0.8). Thus collectively these data show that not only does neuron morphology exhibit changes with HF, but also the circadian mechanism plays a role as Clock ∆19/∆19 HF mice exhibit differently evolving neuron morphology as compared to WT HF mice.
We then compared whether HF is associated with changes in basal dendrite length in WT mice. We found that basal dendrite length increases in HF in both mPFC neurons (two-way ANOVA, F(1, 77) = 12.4, P = 0.0007) (Fig. 3b, left) and hippocampus neurons (two-way ANOVA, F(1, 48) = 4.0, P = 0.049, Fig. 3b, right). In the mPFC neurons, this effect of HF is most pronounced at 75 µm and 100 µm away from the soma of the neuron (Bonferroni's post-hoc analysis, each P < 0.008), but there is no interaction between the effects of HF and distance from soma in either neuron cell type (two-way ANOVA, mPFC: F(10, 77) = 2.0, P = 0.05; hippocampus: F(7, 48) = 0.6, P = 0.8). Thus these data reveal that in WT mice, there are changes in basal dendrite length associated with HF.
Next, we investigated basal dendrite length in Clock ∆19/∆19 mice with HF. We found no change in basal dendrite length in Clock ∆19/∆19 mice in the mPFC (two-way ANOVA, F(1, 77) = 0.4, P = 0.5) (Fig. 3c, left). Basal dendrite length in the hippocampus, however, increases (two-way ANOVA, F(1, 48) = 13.5, P = 0.001) (Fig. 3c, right). The effect of HF is most pronounced at 75 µm and 100 µm away from the soma of the neuron (Bonferroni's post-hoc analysis, each P < 0.05), but there is no interaction between the effects of HF and distance from soma (two-way ANOVA, F(7, 48) = 1.8, P = 0.12). Thus collectively these data show that HF increases the size of basal dendrite trees in mPFC and hippocampus for WT mice, but only in the hippocampus for Clock ∆19/∆19 HF mice.
Finally, we compared the direction of change in basal dendrite morphology in Clock ∆19/∆19 HF versus WT HF mice. As with the apical dendrites, the basal dendrites also exhibit differences by genotype. In contrast to the Clock ∆19/∆19 HF mice, the WT HF mice exhibit an increase in mPFC basal dendrite complexity (Fig. 3d, left). No difference in direction was observed between genotypes in response to HF in basal dendrites of hippocampal neurons (Fig. 3d, right). Thus, these data further support the evolution of neuron morphology in HF, and that the circadian mechanism plays a role, as mutation of CLOCK alters the morphological response in basal dendrites differently as compared to the WT HF response.
We then compared what happens in Clock ∆19/∆19 and WT mice with HF. The first time point of 1 week was first investigated as it is a robust period of cardiac remodeling including inflammatory processes such as elaboration of cytokines which can influence cognitive function. We found that OiP memory is impaired in both WT and Clock ∆19/∆19 mice, at both retention delays (two-way split-plot ANOVA: genotype x delay F(1, 31) = 4.60, P < 0.05, WT n = 17, Clock ∆19/∆19 ) (Fig. 4b). One-sample t-tests suggest impaired memory in all conditions (WT 45 seconds: t(16) = 0.81, P = ns; WT 5 minutes: t(16) = 0.95, P = ns; Clock ∆19/∆19 45 seconds: t(15) = 0.16, P = ns; Clock ∆19/∆19 5 minutes: t(15) = 2.08, P = ns, as the discrimination ratios did not differ from zero. Thus, myocardial infarction is associated with impaired OiP memory tests for both genotypes during the early remodeling period.
Circadian regulation of cerebrovasculature. Neurological deficits in HF may correlate with altered blood flow regulation in small vessels, thus we next performed pressure myography to characterize the tone in the posterior cerebral arteries (PCA) of Clock ∆19/∆19 and WT mice. At baseline, the myogenic responsiveness in Clock ∆19/∆19 PCA is significantly lower than in WTs (P < 0.05 at pressures>60 mm Hg) (Fig. 5a). We then compared whether HF leads to changes in myogenic tone in WT mice. We found that the myogenic responsiveness is significantly increased in the HF WT PCA at 8 weeks post-myocardial infarction, as compared to controls week post-myocardial infarction, OiP performance is impaired in both Clock ∆19/∆19 mice and WT mice at both immediate (45 seconds) and 5-minute retention delays (n = 16 Clock ∆19/∆19 , n = 17 WT). (c) In the 8-week HF mice, OiP performance is impaired in Clock ∆19/∆19 vs. WT mice at immediate (45 second) delays, and is for both Clock ∆19/∆19 and WT mice at 5-minute retention delays (n = 16 Clock ∆19/∆19 , n = 17 WT). Object oddity discrimination is similar at (d) baseline (n = 5/group), (e) 1 week post-myocardial infarction (n = 6/group), and in the (f) 8 week HF mice (n = 16 WT, n = 17 Clock ∆19/∆19 ), as the oddity preference significantly differed from 0.33 (chance performance). However, Clock ∆19/∆19 mice performed worse than WT at 1 week post-myocardial infarction *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. www.nature.com/scientificreports www.nature.com/scientificreports/ (P < 0.05 at pressures of 40 mmHg and 60 mmHg) (Fig. 5b). Next, we investigated myogenic tone in Clock ∆19/∆19 PCA in HF. We found that the myogenic responsiveness is also significantly increased in the HF Clock ∆19/∆19 PCA by 8 weeks post-myocardial infarction versus controls, but moreover this increase in tone occurs over broader range of physiologic pressures and is greater in magnitude than WT PCA (P < 0.05 at pressures>20 mmHg to 100 mmHg) (Fig. 5c). Intriguingly, PCA of all 4 groups show similar vasoconstriction to phenylephrine and thus similar capacity (Fig. 5d), however, Clock ∆19/∆19 PCA exhibit lower myogenic tone at baseline, and a greater rise in myogenic tone in HF, as compared to WT PCA (Fig. 5e). These differences in cerebrovascular tone in Clock ∆19/∆19 versus WT PCA, and in HF, are illustrated in Fig. 5f. Circadian regulation of healthy cognitive systems neurobiology, and adaptations to HF are summarized in Fig. 5g.
Neural gene expression differs in response to heart disease. Finally, we investigated mechanisms underlying the dichotomous neurobiological responses between Clock ∆19/∆19 and WT mice. We found no obvious differences in early cardiac remodeling -inflammasome genes and serum cytokines post-MI ( Supplementary  Fig. S1). These findings are consistent with the notion that rather than being systemically driven, CLOCK regulated neurobiological responses occur at a cellular level in the brain. Thus we next examined gene expression in the mPFC and hippocampus, using microarrays and bioinformatics analyses. Global gene profiles were determined at baseline (BL), MI, and HF, in the mPFC and hippocampus of Clock ∆19/∆19 and WT mice (Fig. 6a). We found significant differences in genes involved in neural growth, cytoskeleton, signalling, and metabolism in the Clock ∆19/∆19 mice (Fig. 6b), consistent with the notion that CLOCK acts at a cellular level. The genes plotted are defined in Table 2, and these and additional genes are further detailed in Supplementary Table S1.
We investigated three paradigms. First, we determined which genes differed at baseline in the mPFC and/ or hippocampus of healthy Clock ∆19/∆19 mice. We found that the core circadian mechanism and its output genes were remarkably dysregulated in the Clock ∆19/∆19 mPFC, and hippocampus, consistent with the notion that the circadian mechanism contributes to neurobiological adaptations within cognition-relevant brain regions (Fig. 6c). Second, we investigated neurobiological gene adaptations to MI. In the mPFC, we found significant differences in genes involved in neuron signalling, and in stress (addiction) pathways in the Clock ∆19/∆19 mice (Fig. 6d). In the hippocampus, differences predominantly mapped to metabolic pathways (data not shown, lists are in Supplementary Table S1). Finally, we investigated why Clock ∆19/∆19 mice respond differently to HF, versus WT mice. To do this, we profiled stress response pathways in the brain, with a focus on altered stress response signalling at baseline in Clock ∆19/∆19 mice, and how these same genes responded during HF where there is phenotypic convergence (Fig. 6e). We identified unique cassettes of 64 genes in the mPFC (top) and 122 genes in the hippocampus (bottom) in which altered expression at baseline converged in response to HF. Taken together, these findings support the notion that the circadian mechanism influences normal neurobiology, and provide insights into the role of CLOCK in neurobiological adaptations within cognition-relevant brain regions in HF.

Discussion
The circadian mechanism plays a critical role in cardiac remodeling in HF. Since neurological conditions (e.g. cognitive impairment, depression) frequently coincide in patients with HF, we hypothesized that the circadian mechanism also influences neuropathology in HF. In this study, we found that CLOCK is pivotal to maintain normal neuron morphology; Clock ∆19/∆19 mice have smaller apical dendrite trees in mPFC layer 2/3 and hippocampus CA1 neurons, versus WT mice. We also observed functional consequences, as visual-spatial memory differs in Clock ∆19/∆19 versus WT mice on the OiP memory tasks. Moreover, normal Clock ∆19/∆19 mice exhibit lower cerebrovascular myogenic tone, versus WT mice. We then investigated what happens in HF, by using the left anterior descending coronary artery ligation model. We observed significant differences in the magnitude and direction of adaptations to HF, including those to neuron morphology in the mPFC and hippocampus, visual-spatial memory, and PCA myogenic responsiveness in Clock ∆19/∆19 versus WT mice. Intriguingly, despite the differences in adaptations to HF, they converged towards similar end profiles in neuron morphology, memory, and cerebrovascular tone in Clock ∆19/∆19 and WT mice. This dichotomous response was mediated in part by differences in baseline gene expression in growth and stress response pathways in neurocognitive regions in the brains of Clock ∆19/∆19 mice, followed by differential activation of neural mRNA pathways in response to MI, with convergence in HF. Collectively these studies establish a new connection between the circadian mechanism and cognitive system neurobiology, and mechanisms underlying adaptations to this system in HF.
One of the novel findings of this study is that the circadian mechanism, and specifically CLOCK, influences cognitive system neurobiology. We used a genetic approach to elucidate a role for CLOCK in neuron morphology/function. We found that Clock ∆19/∆19 mice have pyramidal neurons with altered dendrite trees within layer 2/3 of the mPFC and in CA1 of the hippocampus, and impairments in visual-spatial memory, as compared to WT mice. This is the first study showing that CLOCK plays a role in maintaining neuronal health, and adds to the growing notion that core clock genes regulate aspects of neuron cell biology. Indeed, it was previously shown in mice that loss of another circadian mechanism gene, namely Bmal1, induces astrogliosis in the cerebral cortex and hippocampus 38 , and is associated with impairments in learning and memory 39 . Moreover, several recent environmental studies provide indirect support for the notion that the circadian mechanism mediates neuron morphology and cognition. For example, circadian desynchrony in hamsters, induced by experimental "jet lag", decreases hippocampal cell proliferation and neurogenesis, and impairs hippocampus-dependent learning and memory 40 . Another study showed that inducing circadian desynchrony by altering the external light and dark environment reduces the size of apical dendrite trees for pyramidal neurons in the mPFC, and impairs ability to shift learned behaviors 41 . Thus, CLOCK contributes to the neuron morphology and dependent function of neurons within cognitive circuits. These findings also provide a foundation for investigating a novel role for the circadian mechanism in the neuropathology of HF. www.nature.com/scientificreports www.nature.com/scientificreports/ Another key finding is that there are changes in neuron morphology in the mPFC and hippocampus associated with HF, and that this correlates temporally with impaired cognitive responses. Recent clinical studies support our findings that HF affects the brain and contributes to cognitive and emotional abnormalities. For example, similarly to phenylephrine, suggesting similar capability to respond. However, as compared to WT PCA, the Clock ∆19/∆19 PCA exhibit reduced myogenic tone at baseline, and (e) a greater delta change in myogenic tone in response to HF (e), supporting the notion that the circadian mechanism can influence responses in cerebrovasculature. (f) Summary of the different responses of WT PCA versus Clock ∆19/∆19 PCA. All PCA were collected during the middle of the animals' wake period (Zeitgeber time (ZT19)). *P < 0.05, n = 6 WT PCA, n = 8 WT HF PCA, n = 7 Clock ∆19/∆19 PCA, n = 7 Clock ∆19/∆19 HF PCA. (g) The circadian mechanism is an important regulator of healthy cognitive system neurobiology. Neurobiological adaptations to HF differ in magnitude and direction in Clock ∆19/∆19 versus WT mice, including neuron morphology, visual-spatial memory and cerebrovascular myogenic tone, leading to convergence of end stage measures. These findings highlight the need to better understand how the circadian mechanism affects neurobiological adaptations to HF, a leading cause of morbidity and mortality worldwide. Large black doublesided arrow denotes comparison of normal Clock ∆19/∆19 and WT mice. Open white single sided arrows denote comparison of each genotype at baseline and in HF. ( www.nature.com/scientificreports www.nature.com/scientificreports/  Table 2, and all gene shown as well as additional identified genes are further detailed in Supplementary Table S1. ( www.nature.com/scientificreports www.nature.com/scientificreports/ patients with HF have damage in cognitive regulatory areas including the cerebral cortex and hippocampus 42 , and emotional deficits are commonly found among patients with ischemic heart disease and HF [2][3][4][43][44][45][46] . Experimentally, De Sliva and colleagues recently showed that mice subjected to myocardial ischemia/reperfusion (mI/R) exhibit reactive gliosis throughout the hippocampus, and impaired performance on fear-conditioning and on object location memory tests 47 , thus providing further support for our findings. It is worth noting that here we used a model of HF induced by permanent ligation of the coronary artery, which produces larger infarcts and more rapid progression to HF than mI/R. This bears clinical relevance, as many human patients post-MI do not reach hospitals in a timely fashion and undergo reperfusion, or the procedure is unsuccessful with incomplete revascularization or no reflow.
We also investigated whether the circadian mechanism directly acted on cerebrovascular blood flow, as might be expected to regulate perfusion in HF. In support of this approach the circadian mechanism is implicated in a regulation of diverse vascular beds [48][49][50] , and in the regulation of daily patterns of blood pressure in health and disease (reviewed in 51 ). This is regulated in large part by myogenic response, which is a mechanism employed by resistance arteries to match resistance transmural pressure, and an important regulator of blood flow 52 . We found that the Clock ∆19/∆19 mice significantly increased their PCA in HF, as compared to normal Clock ∆19/∆19 mice. Moreover, this response was greater in magnitude and over a wider range of physiological pressures, as compared with the WTs in HF. However, we also observed that there was no difference when we compared the overall myogenic responsiveness of Clock ∆19/∆19 HF and WT-HF mice, suggesting that while they exhibited a different magnitude of response, the end outcomes were similar. This suggests that the mechanism may not be at the level of cerebrovascular blood flow in Clock ∆19/∆19 HF mice, as regulation is similar in that vascular bed. That is, it seems likely that CLOCK influences adaptations at a cellular level, whether neuron or vascular cell, and not at the systems level.

Gene
Gene Name

Phactr2
Phosphatase And Actin Regulator 2

Cspg5
Chondroitin Sulfate Proteoglycan 5 Table 2. Key genes underlying dichotomous neurobiological responses in Clock ∆19/∆19 mice. Genes listed are as profiled in Fig. 6, and were identified as significantly different in the mPFC, or hippocampus, or both. Further details about these genes are provided in Supplementary Table S1, including Affymetrix identifiers, fold change values, gene expression values under all conditions tested (baseline, MI, HF) in Clock ∆19/∆19 and WT mice, chromosome location, and tags to esemble, entrezgene, genebank and GO databases. Additional genes identified but not mapped in Fig. 6 are also provided in Supplementary Table S1. www.nature.com/scientificreports www.nature.com/scientificreports/ To investigate how CLOCK influences neuronal adaptations at a cellular level, we asked two important questions. First, we investigated which genes differed in the cognition relevant brain regions of Clock ∆19/∆19 versus WT mice, which could help to explain the neurobiological impairments in the normal Clock ∆19/∆19 mice. Microarray and bioinformatics analyses revealed that Clock ∆19/∆19 mice exhibit altered expression of genes involved in neural morphology, signaling, and metabolism. For example, one of the genes identified was gamma-aminobutryic acid type A receptor alpha2 subunit (Gabra2) gene, which plays a critical role in stress responses in the brain 53 . We also found differences in the fatty acid binding protein 7 (Fabp7) gene, which is involved in neural structure in the developing brain 54,55 . Differences were also found in neuroligin 3 (Nlgn3) and in myelin associated glycoprotein (Mag), which are involved in neural cell interactions 56 , and in mitofusin 2 (Mfn2) which is involved in mitochondrial activities 57 . These genes are highlighted in Fig. 6c, Table 2, and additional details and genes are in Supplementary Table S1. Intriguingly, gene expression appeared to be differentially expressed brains of healthy Clock ∆19/∆19 mice, as well as under the disease conditions of MI and HF. The holding expression pattern greatly resembles what is also observed with the circadian mechanisms genes that are a part of CLOCK transcriptional control. Thus these observations suggest that these new genes may also be under CLOCK regulatory control; that is, novel output genes regulated by the circadian mechanism in the brain.
Second, we investigated how gene adaptations differed in response to heart disease in the Clock ∆19/∆19 versus WT mice. We especially focused on genes that showed altered stress responses in signaling at baseline, but then converged and were similar in HF; these could help explain the phenotypic convergence observed at HF. The genes are highlighted in Fig. 6e, Table 2, and additional details and genes are in Supplementary Table S1. Some of the genes identified may directly trigger neuronal adaptations at the cellular level because they regulate key physiological functions such as neuronal synaptic plasticity (e.g. calmodulin dependent protein kinase II inhibitor 2, Camk2n2), neural growth and differentiation (e.g. Necdin, Ndn; chondroitin sulfate proteoglycan 5, Cspg5), and neurotransmission (e.g. glutamate metabotropic receptor 3, Grm3). However, these genes may also indirectly trigger neuronal adaptations at the cellular level; that is, they can make the cells more susceptible to secondary consequences. For example, changes in the neural cell membranes may increase susceptibility to damage from oxidative stress in both Clock ∆19/∆19 and WT mice; indeed, oxidative stress mediators are well known to play an important role in the pathophysiology of HF 58 . Thus taken together, differential activation of the genes in these pathways can help to explain how CLOCK drives dichotomous neurobiology that converges in HF.
One additional point worth noting is that these studies were done in male mice. However, there has been a flurry of recent studies describing how heart disease manifests differently by biological sex and gender, and importantly, that the circadian clock mechanism plays a role 21,24,[59][60][61] . Future studies examining differences in genes expression in male versus female brain regions are warranted, especially in the context of designing circadian medicine based therapies to reduce neurocognitive impairments in HF patients of both biological sexes.
The results of this study have important clinical implications for patients with circadian rhythm disturbances. In humans recovering from MI, circadian desynchrony can occur as a consequence of environmental conditions in our intensive and coronary care units. That is, frequent patient-staff interactions at night, light and noise conspire to disturb sleep and circadian rhythms 62,63 . Experimentally, circadian rhythm desynchrony, even short-term for just the first few days after MI, impairs healing and exacerbates maladaptive cardiac remodeling in the murine model 23 . We would postulate, based on this study, that circadian rhythm disturbances will exacerbate the neurobiological impairments that develop in MI patients. Moreover, subsequent circadian disturbances could cause further neurobiological impairments as patients progress to HF. For example, individuals are subjected to a wide variety of circadian rhythm disturbances in contemporary society such as through night shift work, sleep disorders, and social jet lag (e.g. reviewed in 18 ). Studies investigating the prevalence and severity of neurobiological changes in shift workers who develop HF, or patients with sleep disorders and HF, are clearly indicated. Understanding how the circadian mechanism contributes to neuropathology can lead to new strategies to reduce neurocognitive impairment and improve the quality of life for patients with HF.
In summary, we show that the circadian mechanism influences neurobiology in the brain's cognitive systems. The circadian mechanism is an important regulator of healthy cognitive system neurobiology, and loss of CLOCK leads to adverse changes in the neurobiology of the cognitive system. These data also elucidate a role for the circadian mechanism in neurocognitive adaptations in HF. This is important because clinically many patients with HF often present with concurrent cognitive impairments, and there is no cure for HF. New understanding of the circadian mechanism, and it's role in brain pathophysiology, can lead to new approaches to reduce morbidity and improve the quality of life for HF patients.

Methods
Animals. All studies were approved by the University of Guelph Institutional Animal Care and Use Committee and in accordance with the guidelines of the Canadian Council on Animal Care. Male C57Bl/6 mice were obtained from Charles River, Quebec, Canada. Male Clock ∆19/∆19 mice 35 were obtained from our breeding colony at the University of Guelph, and genotyped as described previously [21][22][23][24] . All mice were housed in the Central Animal Facility at the University of Guelph under a 12-hour light (L) and 12-hour dark (D) cycle with lights on at 8:00am (Zeitgeber Time 0 (ZT0)) and lights off at 8:00 pm (ZT12), and were provided with food and water ad libitum. Activity was recorded from individual cages equipped with running wheels and analyzed using ClockLab (Actimetrics) 21-23 . Left anterior descending coronary artery ligation model and pathophysiology. Mice (8 weeks of age) were subjected to left anterior descending coronary artery ligation for 8 weeks (HF model) and pathophysiologic assessments, as described previously 23,24 . Briefly, mice were anesthetized with isoflurane, intubated, and ventilated (Harvard Apparatus Model 687) throughout the procedure. A local anesthetic of 50:50 bupivacaine and lidocaine solution was administered prior to incision. An incision was made at the 3 rd intercostal space on the left www.nature.com/scientificreports www.nature.com/scientificreports/ side. A prolene 7-0 suture (Ethicon) was passed underneath the LAD at 1mm below the edge of the left auricle, and ligated. The chest and skin were closed with silk 6-0 sutures (Ethicon). All surgeries were performed in a very short window of time, between ZT01 and ZT03. Shams were subjected to the same procedures, but without LAD coronary artery ligation. Mice were administered buprenorphine (0.1 mg/kg) upon awakening.
Pathophysiologic assessments were made prior to procedures, and at 1 week post-myocardial infarction, and at 8 weeks post-infarction (HF model), by echocardiography and in-vivo hemodynamics analyses, as described previously [21][22][23][24]26,60 . Briefly, cardiac function and morphometry were assessed in a blinded manner under light anesthesia (1.5% isoflurane), on a GE Vivid7 Dimension ultrasound equipped with a 14 MHz linear-array transducer. All measurements were acquired at the mid-papillary level, to determine LV internal dimensions at end-diastole (LVIDd), LV internal dimensions at end-systole (LVIDs), % ejection fraction (EF), % fractional shortening (FS) and heart rate (HR). At least 5 different images were analyzed per heart, with n = 8 hearts per group. For in vivo hemodynamics, mice were placed under 3.5% isoflurane, intubated, and body temperature was continuously monitored and maintained at 37 °C. A 1.2-Fr pressure-volume catheter (Transonic) was advanced into the LV, and measurements were recorded using an ADInstrument PowerLab. LV end systolic pressure (LVESP) and end diastolic pressure (LVEDP), and volumes (LVESV, LVEDV), stroke volume (SV), cardiac output (CO), maximum and minimum first derivative of LV pressure (dP/dtmax, dP/dtmin), and systolic and diastolic blood pressure (SBP, DBP) were recorded. Mean arterial blood pressure (MAP) was calculated as DBP+[(SBP−DBP)/3]. Following collection of hemodynamics data, mice were sacrificed by isoflurane overdose and cervical dislocation. Body weight (BW), heart weight (HW) and tibia length (TL) measurements were collected.
Brain collection and Golgi-Cox staining. HF mice were sacrificed at 8 weeks post-myocardial infarction (and age-matched controls) by isoflurane and cervical dislocation. Mice were decapitated and the brain quickly removed. Golgi-Cox staining was performed 64,65 . Briefly, whole brains were immediately placed into Golgi-Cox impregnation solution (1% potassium dichromate, 0.8% potassium chromate, 1% mercuric chloride) and incubated in this solution for 25 days in the dark at room temperature. Brain sections containing the mPFC or dorsal hippocampus were made at 500 µm thickness using a vibratome, developed using ammonium hydroxide, fixed with Kodak Fixative A, mounted onto microscope slides, dehydrated in ethanol gradients, then cover slipped using anhydrous mounting medium.
Neuron imaging, tracing, and morphology analysis. Neuron imaging and tracing were performed 64 .
Briefly, neurons were imaged in bright field using an Olympus BX53 upright microscope (Olympus, Richmond Hill, ON, Canada) controlled using Neurolucida software (version 10, MBF Bioscience, Williston, VT). Overlapping image stacks containing neurons to be traced were captured using an Olympus UPlanSApo 30X, 1.05 NA silicon oil immersion objective. Pyramidal neurons in layer 2/3 of the mPFC and the CA1 region of the hippocampus were imaged and manually traced in three dimensions using the Neurolucida software. All experimenters were blinded to the treatment group prior to neuron imaging and tracing. Four neurons were sampled randomly from each brain region, from n = 4-5 mice per group, based on the criteria of: (i) being fully contained within the thickness of the slice, (ii) not being occluded by other stained neurons, and (iii) having stained dendrites that were fully intact. Quantitative analyses of apical and basal dendrites for each neuron were performed using Neurolucida Explorer (MBF Bioscience). Data for the neurons within each brain region were averaged per mouse, and statistical analyses were performed with the mouse serving as the level of sample.
Object-in-place (OiP) memory task. Four objects varying in size (7-20 cm), color, material (glass, aluminum, ceramic and plastic) and texture were placed in each corner of an open field, 5 cm away from the walls as described previously 66 . The objects were washed with 50% ethanol between trials to eliminate exploration bias due to olfactory cues. The open field (45 × 45 × 30 cm 3 ) contains no spatial cues on the apparatus walls and a bare floor. It was constructed of white, plastic-coated corrugated cardboard. Spatial cues were present in the testing room (i.e. television, shelving, camera and colored door), and a ceiling-mounted white light illuminated the room. All mice experienced two habituation sessions on successive days, where they explored an empty open field for 10 min. Learning occurred 1 day following habituation, during a 10-minute sample phase where mice were allowed to freely explore four different novel objects. To assess immediate and short term memory, mice experienced a 2-minute choice phase 45 seconds and 5 minutes later where they viewed the same four objects, but two objects had switched locations (right or left, counterbalanced) creating a 'novel side' . The discrimination ratio (DR) was calculated as [(novel object exploration − familiar object exploration)/(total object exploration)]. Preference for the objects on the novel side gave a DR value significantly greater than zero, and was interpreted as being indicative of memory. MI surgeries on all mice occurred 1 day following the last choice phase. At 1 week and 8 weeks following MI surgeries, the mice repeated the same sample and choice phases as previously described with an additional set of 4 novel objects for each time point. Mouse exploration was scored by two researchers. 95% interrater reliability was maintained to ensure consistency in the recording of data.
Object oddity perceptual task. To assess whether there was impairment of basic object perceptual discrimination following MI, an object oddity task was used after OiP testing, as described previously 67,68 . The same open field used for the OiP test was used for this task. Three objects, two identical, one unique, were place on one side of the open field, 5 cm away from the wall. Mice experienced a single 10-minute sample phase in which they were allowed to freely explore the three objects. To ensure object preference was due to one object being unique rather than a bias for an object, the order of the three objects and the selection of the unique object were switched after each mouse. Oddity preference was calculated as [unique object exploration]/[total exploration]. An oddity preference ≥0.33 indicated a greater preference for the unique object and intact perception. As with the OiP task, mouse exploration was scored by two researchers. 95% interrater reliability was maintained to ensure consistency in the recording of data.
www.nature.com/scientificreports www.nature.com/scientificreports/ Statistical analysis. All values are expressed as mean ± SEM. Echocardiography, hemodynamics and histology data were analyzed using two-way analysis of variance (ANOVA) followed by Tukey's post-hoc analysis. Neuron morphology data were analyzed by two-way ANOVA followed by Bonferroni's multiple comparison correction. Behavioural data were analyzed using a split-plot two-way ANOVA followed by Bonferroni's multiple comparison correction. Myography data were analyzed using two-way ANOVA followed by Bonferroni's multiple-comparison correction. Microarray gene expression data was analyzed by GeneSpring GX v14.9 software (Agilent Technologies Inc). P-values < 0.05 were considered statistically significant. All values were analyzed in GraphPad Prism 6 statistical software and plotted in Prism 6 or Microsoft Excel.