Altered resting-state connectivity within default mode network associated with late chronotype Charlotte Mary Horne1,2 and Ray Norbury1 1 Department of Psychology University of Roehampton 2 Address for correspondence: Charlotte Mary Horne Department of Psychology University of Roehampton Whitelands College, London, UK SW15 4JD E:[email protected] T: +44(0) 20 8932 5788 Abstract 1 Current evidence suggests late chronotype individuals have an increased risk of developing depression. However, the underlying neural mechanisms of this association are not fully understood. Forty-six healthy, right-handed individuals free of current or previous diagnosis of depression, family history of depression or sleep disorder underwent resting- state functional Magnetic Resonance Imaging (rsFMRI). Using an Independent Component Analysis (ICA) approach, the Default Mode Network (DMN) was identified based on a well validated template. Linear effects of chronotype on DMN connectivity were tested for significance using non-parametric permutation tests (applying 5000 permutations). Sleep quality, age, gender, measures of mood and anxiety, time of scan and cortical grey matter volume were included as covariates in the regression model. A significant positive correlation between chronotype and functional connectivity within nodes of the DMN was observed, including; bilateral PCC and precuneus, such that later chronotype (participants with lower rMEQ scores) was associated with decreased connectivity within these regions. The current results appear consistent with altered DMN connectivity in depressed patients and weighted evidence towards reduced DMN connectivity in other at-risk populations which may, in part, explain the increased vulnerability for depression in late chronotype individuals. The effect may be driven by self-critical thoughts associated with late chronotype although future studies are needed to directly investigate this. Keywords: Chronotype, fMRI, Resting state, Default Mode Network, Precuneus, Major Depression Introduction 2 An individual’s diurnal preference is referred to as their chronotype (J. A. Horne & Ö stberg, 1976). Early chronotypes (‘larks’) rise early in the morning and reach their peak early in the day (Schmidt, Collette, Cajochen, & Peigneux, 2007; Schmidt et al., 2012), whereas late chronotypes (‘night owls’) prefer to rise and sleep late. This allows the individual to synchronise their optimal time of day to their circadian profile. Although chronotype is considered a relatively stable trait, it is also known to shift across the lifespan. Children tend to be early chronotypes, shifting towards late chronotype during adolescence before progressively moving back to early chronotype during later life (Randler, Freyth-Weber, Rahafar, Florez Jurado, & Kriegs, 2016). A growing body of research suggests an association between late chronotype and psychological health – particularly major depression. For example, in a longitudinal study Haraden et al. (Haraden, Mullin, & Hankin, 2017) reported that late chronotype predicted depressive symptoms and the onset of depressive episodes. Late chronotype is also associated with increased likelihood of reporting depressive symptoms (Hidalgo et al., 2009; Levandovski et al., 2011), diagnosis of depression (Antypa, Vogelzangs, Meesters, Schoevers, & Penninx, 2016) and use of antidepressant medication (Merikanto et al., 2015; Merikanto et al., 2013). Together, these findings suggest an important link between late chronotype and depression. Converging evidence suggests that depression is associated with disruption to the Default Mode Network (DMN). This network is preferentially activated when the participant is internally focused; for example, during daydreaming, future planning, memory retrieval and thinking from different perspectives (Raichle et al., 2001). The DMN can be studied using resting-state Functional Magnetic Resonance Imaging (rsFMRI) (Broyd et al., 2009; Dutta, McKie, & Deakin, 2015; K. Zhang et al., 2016) where Blood Oxygenation-Level Dependent (BOLD) data is acquired while participants are at rest (not performing any 3 specific task) and are given minimal instruction (e.g. keep their eyes open; let their mind wander). A number of techniques have been developed to probe rsFMRI data. In model-free approaches such as Independent Components Analysis (ICA), the spatio-temporal structure of the data is characterised into a number of independent components reflecting a functional network, physiological noise or image/acquisition artifact. Further processing is then required to either manually select relevant networks and discard noise components or use a well validated resting-state network map (Smith et al., 2009), to generate subject-wise networks for subsequent analysis (Filippini et al., 2009). Other analytical approaches include seed- based correlation analyses - where the time-course from a ‘seed’ voxel is extracted and correlated with activity at each voxel. Regional and network homogeneity assess the correlation between a given voxel and its nearest neighbours or all voxels in the entire network, respectively. Whereas Amplitude of Low Frequency Fluctuations (ALFF) and fractional ALFF (fALFF) quantify the amplitude of low frequency oscillations generated in the brain at rest. Irrespective of the method employed, a number of brain regions have been consistently associated with the DMN. These include the posterior cingulate cortex (PCC)/precuneus, medial prefrontal cortex (mPFC), ventral anterior cingulate cortex (ACC), and inferolateral parietal cortex (Dutta et al., 2015). The literature describes a dissociation between the anterior (centred on the mPFC) and posterior regions (centred on the PCC and inferolateral parietal lobes) of the DMN (Zhou et al., 2010). The anterior DMN is implicated in self-referential processing and emotion regulation whereas the posterior DMN is implicated in consciousness and memory processing (X. L. Zhu et al., 2012). A pattern of altered connectivity within the DMN has been identified in depressed individuals. However, the directionality and extent of this altered connectivity is unclear as 4 both increases and decreases have been reported across a broad range of brain regions (for reviews please see (Brakowski et al., 2017; Broyd et al., 2009; Dutta et al., 2015; M. Greicius, 2008; K. Zhang et al., 2016). These inconsistencies might be related to differences in the depressed population studied; for example, age, medication status, pharmacotherapy and depression severity. By studying patients with depression, it is also difficult to determine whether altered connectivity arises as a result of the disorder or pre-exists development of the disorder. To address the above problem, it is necessary to examine DMN connectivity in healthy, never-depressed individuals at increased risk for depression. Servaas et al., reported high neuroticism (a recognised risk factor for depression) was associated with reduced functional connectivity between an occipito-parietal seed region (including the PCC, precuneus and cuneus) and middle cingulate gyrus, insula, and postcentral gyrus, and increased connectivity between the same region and the calcarine sulcus, lingual gyrus and inferior frontal gyrus following exposure to self-referential criticism (Servaas et al., 2013). Reduced connectivity within the DMN has also been reported in children at familial risk of depression (i.e. children that have a 1st degree relative with depression) (Bellgowan et al., 2015) although hyperconnectivity of the DMN has also been reported in this at-risk group (Chai et al., 2016; Norbury, Mannie, & Cowen, 2011; Posner et al., 2016). Two of these studies (Chai et al., 2016; Posner et al., 2016) also reported a reduced anticorrelation between the DMN and the central executive network (CEN) including the dorso-lateral prefrontal cortex (DLPFC) suggesting an impairment in attentional control or impulsivity within these groups (Chai et al., 2016; Posner et al., 2016). Healthy participants that have experienced childhood trauma, another recognised risk factor for depression, also show altered DMN connectivity (Lu et al., 2017). 5 Additionally, a number of studies have identified altered DMN connectivity in healthy individuals with subclinical symptoms of depression. Rzepa and McCabe reported decreased functional connectivity between the amygdala and PCC/precuneus as well as between the DMPFC and precuneus in adolescents with high scores on the mood and feelings questionnaire (MFQ) compared to age-matched participants scoring low on the same instrument (Rzepa & McCabe, 2016). Zhang et al., reported that individuals with a cognitive vulnerability for depression (a negative cognitive style where the individual attributes negative events to external causes that are beyond their control) had reduced ALFFs in bilateral orbitofrontal cortex and increased ALFF in insula cortex and left fusiform gyrus, similar to differences observed in the depressed patient group (X. Zhang et al., 2016). Similarly, reduced fractional ALFF (fALFF) in bilateral precuneus and left posterior cerebellum was associated with higher nonclinical depressive symptomology in young adults (Wei et al., 2015). Taken together, the data indicate that healthy participants with sub- threshold symptoms of depression show altered functional connectivity within the DMN, similar to other at-risk groups (Bellgowan et al., 2015; Chai et al., 2016; Lu et al., 2017; Posner et al., 2016; Servaas et al., 2013), which may represent