Alterations in circadian entrainment : translational value in major depression

Abstract

Adverse conditions of different origin occurring during development can have a negative impact on the nervous system and increase risk for the onset of mental disorders. In mice prenatal exposure to excess glucocorticoids (GC) leads to late onset depression-like behaviour in males which is preceded by alterations in circadian entrainment. The aims of this thesis were to (1) identify sex-related differences in behavioural outcomes with a special focus on circadian activity and mood-related behaviours; (2) investigate the mechanisms behind the behavioural alterations; (3) implement novel computational techniques for characterisation of behaviour in rodents; and (4) explore the translation potential of the analysis of activity in patients suffering from depression.

Pregnant C57Bl/6 were exposed to the synthetic GC dexamethasone (Dex) (0.05 mg/Kg/day, i.p.) from gestational day 14 until delivery. We analysed the behaviour of both males and female offspring at the age of 6 months focusing on circadian entrainment of spontaneous activity in the home cage environment. There were sparse differences between Dex and control (Ctrls) males at baseline, which were consistent with decreased social behaviour. In contrast, Dex-females were spontaneously hyperactive, a phenotype resembling ADHD models.

Next, we investigated the changes in spontaneous activity following an abrupt advance in dark phase onset (phase shift). Dex-males re-entrained activity faster, while in Dex-females re-entrainment was delayed as compared to Ctrls. We implemented a multivariate approach to analysing the spatial and temporal organisation of behaviour which included affinity propagation (AP) clustering and uniform manifold approximation and projection (UMAP) using a set of 129 features extracted form activity recording. AP-clustering allowed the identification of features with similar patterns of variations following the phase shift, while UMAP supported the visualisation and quantification of changes in organisation of behaviour. The organisation of behaviour changed in both males and females, particularly in the dark phase. Dex-males displayed fewer, but more persistent alterations, while in Dex-females most alterations were transient and returned to baseline values within 5 days after the phase shift. The behaviour alterations in response to the phase shift pointed at alterations in the function of the central clock, the suprachiasmatic nucleus (SCN). In Dex-males the coupling between SCN and peripheral oscillators was disrupted, but was restored by enhancing GC receptor-mediated signalling by the antidepressant DMI. Furthermore, DMI also prevented the onset of depression. In Dex-females instead, the coupling between SCN and peripheral oscillators was preserved, supporting the idea that altered coupling is an early sign specific for depression onset.

To investigate the connection between Dex-female behaviour and central clock alterations we performed gene expression analysis by RNA- sequencing followed by SPIA pathway analysis and identified significant alterations in glutamatergic, GABAergic, and dopaminergic signalling. We selected a set of relevant genes to validate the altered expression by cross-referencing differentially expressed genes in the SCN with the international mouse phenotype consortium (IMPC) database for associated phenotypes. Using this approach, we identified several differentially regulated genes supproting a decrease in dopamine signalling and the ADHD-like phenotype in Dex-females.

Spontaneous activity can be measured also in human subjects by means of actigraphy. We used actigraphy recordings from patients suffering from major depression to explore possible correlations between patterns of activity and symptom severity or response to internet- delivered cognitive behavioural therapy (iCBT) as antidepressant treatment. We implemented an independent homogenous training of model ensembles using systematic bootstrapping (with replacement) followed by pruning based on goodness-of-fit criteria. The aggregated output outperformed individual models by a factor of 4, indicating that ensemble training increases the accuracy of fitting.

We further applied an external validation procedure using an independent dataset to provide proof-of-concept evidence for the possibility to model symptom severity estimated by MADRS-S. Our data suggests that higher symptom severity is associated with less complex patterns of activity, stronger coupling with circadian entrainers and less robust circadian rhythms, while larger improvement following iCBT is associated with less fragmented activity, more robust circadian rhythms, and higher day-to- day variability before treatment. To summarize, these studies suggest a strong connection between circadian activity and neuropsychiatric disorders. Further experimental and clinical investigations are needed to dissect all relevant systems and molecules involved to provide new insight into mental disorders.