Alterations in circadian entrainment precede the onset of depression-like behavior that does not respond to fluoxetine

All experiments were performed in agreement with the European and Swedish national regulation following approval by the local Animal Ethics Committee (Stockholms Norra djurförsöksetiska nämnd).

Timed-pregnant C57Bl/6 mice (=34 per treatment) (Charles River, SCANBUR Research, Sollentuna, Sweden) were injected subcutaneously with 0.05 mg kgper day DEX (Sigma-Aldrich, Stockholm, Sweden) from gestational day (GD) 14 (the day the postcoital plug was noted was considered GD0) until delivery (recorded as postnatal day (PND) 0). This dose induces a moderate fetal growth retardation without affecting litter size, gestational length or maternal behavior.Control mice were delivered by sham-treated females, that is, they were handled the same way, at the same time of the day as the exposed dams, and were injected with an equivalent volume of vehicle. The litters were culled to four pups per litter at PND3 and the pups were weighed at PND 3, 7, 14 and 21. N −1 ²⁵

The exposure to MeHg has been described elsewhere.Briefly, pregnant C57Bl/6 dams (=6 per treatment) received MeHg (CHHgOH) at the dose of 0.5 mg kgper day via drinking water from GD7 till day 7 after delivery. Control dams received tap water. ¹⁶ −1 N 3

At weaning (PND21), the mice were implanted subcutaneously with sterile radio frequency identification tags (Trovan Unique 100A, Trovan, Douglas, Isle of Man, UK) under brief isoflurane anesthesia. The transponders allow unambiguous identification of animals and are also used for monitoring the activity in a homecage environment. After implantation, the pups were redistributed to new cages so that each cage would house a maximum of five mice originating from different litters, and the distribution was maintained throughout the study. The mice were kept in an animal facility under 12:12-h light–dark (LD) cycle (light intensity 50 lx; light on at 0600 hours) at constant temperature (22±1 °C) and humidity (50±5%).

All experiments involving circadian rhythms were performed in an isolated room (22±1 °C; 50±5% relative humidity), with a 12:12-h LD cycle (light intensity 200 lx; light on at 0600 hours). The circadian zeitgeber ('time giver') time (ZT) 0 corresponds to the onset of the light phase. The interaction with human experimenters was limited to changing the cage and replenishing food and water, which occurred at random times throughout the experiment.

The mice were first screened in a battery of behavior tests, including open field, and a social recognition test (see). Supplementary Materials and Methods

Adult male mice aged 7 weeks, 3 mo or 12 mo were tested for depression-like behavior in the forced swim tests, as described earlier.Briefly, the animals were individually placed in glass cylinders (24 cm height, 12 cm diameter) filled with water (23.5 °C) to a depth of 16 cm. The animals were exposed to a 15-min pre-test followed by a 6 min test 24 h later. Test sessions were videotaped and analyzed offline by one investigator who was blind to the treatment and exposure conditions. Immobility was defined as passive floating for at least 2 s. After documenting depression-like behavior at the age of 12 mo, the animals were treated with FLX (a selective serotonin reuptake inhibitor antidepressant) dissolved in drinking water (80 mg l) for 21 days before repeating the test. This dosage did not reduce water intake and resulted in FLX plasma levels within therapeutic range in humans.In addition to forced swim tests, we tested depression-like behavior in a tail-suspension test, as described in the. ¹⁶ −1 , ^{4 26} Supplementary Materials and Methods

The subgranular zone of the hippocampal dentate gyrus (DG) is one of the brain regions that retain the neurogenic potential in adult animals.To investigate neurogenesis, we estimated the progenitor proliferation and the maturation of newly generated neurons in the DG by immunohistochemical methods (see). Progenitor proliferation was assessed by counting EdU-positive cells in the subgranular zone (defined as a two-cell-diameter layer subjacent to the granule cell layer of the DG) after pulse-labeling by systemic administration of EdU (50 mg kgper day intraperitoneally at ZT12 for 7 days). Maturation of newly generated neurons was assessed by counting doublecortin (DCX)-positive neuroblasts throughout the granule cell layer of the DG. The cell counting was performed in a stereological design using vertical sections. Equally spaced series of sagittal sections (20 μm thick; 200 μm between consecutive slices) were collected starting from the first occurrence of the hippocampal structure, until the dorsal hippocampal commissure (lateral 3.5–0 mm in stereotaxic coordinates). The total number of cells was estimated by multiplying the number of counted cells with the inverse sampling fraction. ²⁷ −1 ²⁸ Supplementary Materials and Methods

The expression of the glucocorticoid receptor (GR) was assessed by measuring the fluorescence intensity in the DG. The intensity of the positive signal was estimated in the granule cell layer (manually delineated), and the background intensity was evaluated in the molecular and polymorph layers of the DG.

We investigated the diurnal rhythm of GC secretion at 12 mo by collecting spontaneous fecal boli between ZT12 and ZT12-14 (that is, immediately after the transition between the light and dark phases). The feces were collected in sterile Eppendorf tubes and stored at −80 °C until further processing. Samples from each mouse (=8–10 per group) were collected on two occasions (7-day interval between samplings). The concentration of corticosterone metabolites in dry fecal extracts was measured by enzyme immunoassay according to the manufacturer's instructions (DetectX, Arbor Assays, Ann Arbor, MA, USA). N

We investigated the expression of clock genes in hippocampi harvested from male mice 3 and 12 mo (=4 per group) maintained in a 12:12-h LD cycle for at least 7 days before sampling. The mice (two mice per cage) were killed at ZT 3 and ZT12 (corresponding to the peak and trough in messenger RNA (mRNA) expression of core clock genes involved in the main feedback loop, such asand; see also Harbour) by an overdose of anesthetic (sodium pentobarbital, 150 mg kg). The blood was removed by transcardial perfusion with ice-cold buffered saline. The hippocampus was quickly dissected on ice and stored at −80 °C until processing. The relative expression of clock genes was assessed by quantitative real-time PCR withas a housekeeping gene (see also). The clock genes (,and) were selected for analysis based on their involvement in the core feedback loops of molecular clock, as well as on their documented regulation of neurogenesis. N Bmal1 Per/Cry et al. Gapdh Bmal1 Per1 Rev-Erb α ²⁹ −1 ,, ^{30 31 32} Supplementary Materials and Methods

We recorded the spontaneous activity of mice aged 1, 3, 5 and 12 mo (=7–8 per group) using the TraffiCage system (NewBehavior, Zürich, Switzerland), as described elsewhere.Briefly, the system consists of an array of antennas placed under the cage with group-housed, freely moving mice. The antennas read the radio frequency identification tags and provide an approximate location of each animal with a time resolution of 20 ms. A 'visit' was defined as the time interval during which an animal is detected constantly by the same antenna, and was used as activity count. The time series of visits are exported as ASCII files and analyzed using custom algorithm implementations in Matlab R2013b (The MathWorks, Natick, MA, USA). N ³³

To minimize the effects of novelty and possible interference from circadian rhythm entrainment, we derived the baseline measurements based on three LD cycles after an acclimation period of at least three LD cycles. The mice were then exposed to constant darkness (DD; free-running period) for 2 weeks. Spontaneous activity in free-running conditions was analyzed over the last 72 h of recording in DD. Circadian re-entrainment was induced by resuming the LD cycle. The effects of forced resynchronization were analyzed during the first 3 days of circadian re-entrainment. Each recording session included two cages monitored simultaneously, one control and one test cage, placed randomly on either TraffiCage plates.

We analyzed spontaneous activity by detrended fluctuation analysis. The method is based on linear regression analysis of the residual variance of the time series against the timescale used for detrending on double-logarithmic plots. The correlation coefficient of a linear regression in double-logarithmic plot translates into a scaling exponent, and describes the long-term autocorrelation patterns embedded in the time series.Random fluctuations yield a scaling exponent around 0.5, and values close to 1 suggest strong underlying regularity; complex time series with fractal-like patterned irregularity yield values between 0.5 and 1.The scaling exponent in young, healthy rodents and humans is around 0.8,and the loss of patterned irregularity has been suggested to be a hallmark of disease. ^{34 34} , ^{35 36 37}

The analysis of circadian rhythmicity consisted of rhythmometry by means of cosinor analysis.The period of spontaneous activity was estimated as the highest peak in the-periodogrambetween 20 and 25 h with a 5-min resolution. , ^{38 39} 2 ⁴⁰ χ

The activity counts were binned in 5-min non-overlapping epochs, and then smoothed with a weighted average using a sliding Gauss window (4 h width). The epochs with activity above the individual's average were considered 'active epochs'. The active phase was defined as a sequence of active epochs contiguous or separated by gaps no larger than 1 h. The onset and the offset of the active phase were defined as the ZT corresponding to the beginning and the end of the active phase, respectively. The duration of the active phase was calculated as the time span between the onset and the offset of the active phase during one LD cycle. For steady entrainment conditions, the analysis of active phase is based on three consecutive LD cycles.

The Matlab routines developed for TraffiCage data analysis are available upon request.

Tissue samples (~0.25 cm) were harvested from the ear of adult (6 mo) control and Dex-exposed mice under terminal anesthesia. The tissue was rinsed in Hank's balanced salt solution (Life Technologies Europe, Stockholm, Sweden), then minced with sterile razor blade into Collagenase (Type XI-S) (Sigma-Aldrich) (30 min at 37 °C). After digestion, 3 ml of Dulbecco's modified Eagle's medium (DMEM; Life Technologies Europe) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin (Life Technologies Europe) was added to a 6-cm plate and the samples were incubated at 37 °C for at least 6 days. After passaging (0.05% Trypsin-EDTA; Invitrogen, Life Technologies Europe), the cells were plated in 12 multi-well plates in MEF medium (DMEM medium+10% fetal bovine serum+1% pen/strep) at a density of at least 50k cm. After 24 h, the expression of clock genes was synchronized by exposing the fibroblasts to 1μDex. The cells were collected between 6 and 36 h after synchronization. The relative expression ofwas assessed by quantitative PCR withas the housekeeping gene (see also). Circadian oscillations in clock gene expression were analyzed by means of cosinor rhythmometry. 2 − 2 , ^{38 39} m Bmal1 Gapdh Supplementary Materials and Methods

All statistical analyses were performed in Statistica version 12 (Statsoft Scandinavia, Uppsala, Sweden). Unless otherwise specified, we used simple, factorial or mixed (repeated measures between-group) design analysis of variance models followed by unequalHSDtest or contrast analysis. The results are shown as average and s.e.m. The number of independent samples in each group is indicated in the figure legend. The statistical power of all analyses is higher than 0.8. N post hoc

The exposure to DEX from GD14 until delivery induced a mild but consistent decrease in intrauterine growth rate (). After delivery, the body weight was lower in both male and female DEX-exposed mice, as previously shown in rats.The difference was consistent until weaning (PND21), but disappeared soon thereafter (no significant difference at PND28;). The behavioral outcomes of prenatal exposure to DEX, including hyperactivity in the open field and impaired social behavior in the social recognition test, were only present in the male offspring (). The sex differences were in agreement with previous studies,and therefore we used only the male offspring for the following experiments. Supplementary Figure 1A Supplementary Figure 1B Supplementary Figure 2 , ^{25 41} , ^{42 43}

We tested male offspring in the forced swim test at several ages, and found that DEX-exposed mice showed increased immobility time at 12 mo, but not earlier (). The depression-like phenotype was confirmed in the tail-suspension test (). We next asked whether antidepressant treatment could reverse the behavioral phenotype. We selected FLX based on our previous experienceand on the established effectiveness of selective serotonin re-uptake inhibitors in models of depression induced by prenatal stress.Treatment with FLX for 21 days before repeating the forced swim test did not affect the immobility time in DEX-exposed mice, but decreased it only in controls (). Figure 1 Supplementary Figure 3 Figure 1 ⁴ , ^{44 45}

We investigated neurogenesis at 12 mo of age and observed that DEX-exposed mice had a lower number of EdU-positive cells in the subgranular zone () and less DCX-positive cells in the granular layer of the DG (). FLX treatment did not have any significant effect on neurogenesis in DEX-exposed mice. However, in agreement with earlier reports,FLX decreased the number of DCX-positive cells in controls. Figure 2a Figure 2b ,, ^{46 47 48}

The antidepressant effect of FLX requires rhythmic GC secretion,therefore we investigated the diurnal rhythm of GC secretion by measuring the concentration of GC metabolites in feces. DEX-exposed mice exhibited lower levels of corticosterone metabolites that did not show significant diurnal oscillations (). ⁴⁹ Figure 2c

Hypothalamic–pituitary–adrenal axis activity regulation, including the circadian fluctuations in GC secretion, involves hippocampal GR in the feedback loop.The analysis of GR expression in the hippocampus revealed that 12 mo-old DEX-exposed mice had a significantly lower GR signal intensity in DG (; see also), in agreement with earlier reports. , ^{50 51} , ^{52 53} Figure 2d Supplementary Figure 4

Neurogenesis timing and progression are regulated by clock genes, particularly byand,and oscillations in circulating GC induce their expression in the hippocampus.The quantitative PCR analysis of clock genes in the hippocampi showed that in 12-mo-old depressed DEX-exposed mice there were no diurnal oscillations in the expression of,and(). We next asked whether similar alterations in clock gene expression may occur before the onset of depression-like behavior, and analyzed hippocampi harvested from 3-mo-old mice. Notably, the diurnal oscillations in,andwere abolished already at a young age in DEX-exposed mice (). Bmal1 Per1 Bmal1 Per1 Rev-Erb α Bmal1 Per1 Rev-Erb α , ^{30 31} , ^{54 55} Figure 2e Figure 2f

Diurnal fluctuations in GC are also involved in regulating circadian entrainment of spontaneous activity, particularly in response to changes in the LD cycle.Therefore, we analyzed the locomotor activity in the homecage in steady-state conditions (LD cycle or DD), as well as the response to circadian re-entrainment (forced synchronization;). Circadian entrainment allows the anticipation of regular events (for example, onset of dark/light phase). In spontaneous activity, this is illustrated by the acrophase occurring before ZT 18 (that is, the middle of the dark phase) in the context of steady entrainment with a 24-h circadian period. In contrast, forced synchronization transiently increases the amplitude of circadian oscillations and delays the acrophase to around ZT 18. DEX-exposed mice displayed larger amplitude than controls during steady entrainment already from 1 mo of age, and showed no significant difference between steady entrainment and forced synchronization (). The acrophase in DEX-exposed mice occurred close to ZT 18 at all ages tested, and was significantly delayed by forced synchronization only at 1 mo (). In contrast, the acrophase of spontaneous activity in control mice occurred consistently before ZT 18 under steady entrainment conditions, and was significantly delayed by forced resynchronization at 1, 3 and 5 mo (). This suggested that photic entrainment was stronger in DEX-exposed mice than in controls, particularly in steady entrainment conditions. Therefore, we next analyzed the onset and the offset of the active phase in relation to the LD cycle. We found that the onset of active phase virtually coincides with the beginning of the dark phase in DEX-exposed mice, but precedes the dark onset in controls (). In contrast, the offset of the active phase occurs very soon after the end of the dark period in DEX-exposed mice, whereas in controls it occurs consistently later than the end of the dark phase (). This led to DEX-exposed mice having a shorter active phase than controls at all ages tested (). ,,, ^{56 57 58 59} Figure 3a Figure 3b Figure 3c Figure 3c Figure 3d Figure 3d Figure 3e

To confirm this hypothesis, we analyzed the spontaneous activity in response to a 6-h phase advance at the age of 6 mo () and observed that the acrophase is advanced by 6 h immediately after the phase advance in the LD cycle (). When we analyzed the onset and the offset of the active phase in relation to the LD cycle, we found that the onset of active phase coincides with the beginning of the dark phase immediately after the phase advance in the LD cycle. In contrast, the onset of activity is lagging behind the onset of the dark phase in control mice (). In addition, the offset of activity in DEX-exposed mice occurs earlier during the first two LD cycles after the phase advance, but later coincides with the offset of the dark phase (). In control mice, the onset of the light phase suppresses spontaneous activity immediately after the phase advance in the LD cycle (). This pattern of changes is reflected in the duration of the active phase, which is shortened only during the first LD cycle after the phase advance in DEX-exposed mice, whereas in controls it was persistently shorter after the phase advance. Altogether, these data suggest that the diurnal rhythms in spontaneous activity in DEX-exposed mice are more rigid in relation to the LD cycle, and that DEX-exposed mice entrain a circadian rhythm considerably faster than controls. Figure 4a Figure 4b Figure 4c Figure 4c Figure 4c

Circadian entrainment also implies that the synchronized oscillation is self-sustained, and has a period similar to that of the zeitgeber. Forced synchronization induced circadian re-entraining in both controls and DEX-exposed mice, and the circadian period was close to 24 h at young ages (). However, in young DEX-exposed animals during steady entrainment the circadian period displayed a tendency to deviate from 24 h, and the difference became significant in steady entrainment at the age of 5 mo (). Moreover, at 12 mo, circadian entrainment failed in DEX-exposed mice even during forced synchronization, and the circadian period did not vary between free-running, forced synchronization and steady entrainment (). Figure 5a Figure 5a Figure 5a

To investigate whether the alterations in circadian rhythmicity are due to central clock dysfunction, or solely due to diurnal entrainment, we evaluated the effects of photic entrainment in comparison with free-running conditions. In free-running conditions, intrinsic rhythmicity in spontaneous activity is controlled by the central clock located in the suprachiasmatic nucleus.We found no difference between DEX-exposed mice and controls in either free-running period or scaling exponent (), which rules out a possible central clock dysfunction in the DEX-exposed mice. The transient increase in scaling exponent observed during forced synchronization by photic entrainment can be explained by a strong underlying rhythm, presumably diurnal. This effect was consistently detected in DEX-exposed mice at all ages, but in controls it reached significance only at the ages of 1 and 3 mo (). During steady entrainment, DEX-exposed mice displayed consistently higher scaling exponent than in free-running conditions, indicating that photic entrainment induces a more pronounced background rhythm in spontaneous activity in DEX-exposed mice. Taken together, these data confirm that the core alteration is the impairment in circadian entrainment. , ^{35 36} Figures 5a and b Figure 5b

Skin fibroblasts express functional molecular clock machinery,and the circadian oscillations in clock gene expression maintain to a large extent the features of circadian rhythms in the central clock.The molecular clock machinery in the fibroblasts acts as a peripheral oscillator and is subject to entrainment by the central oscillator (located in the suprachiasmatic nucleus),similar to spontaneous activity.The possibility to synchronize self-sustained oscillations, and to reset the phase is preserved in cultured fibroblasts.We therefore investigated the expression of clock genes in fibroblast isolated from controls and DEX-exposed mice aged 6 mo, the age when we documented the facilitated entrainment of spontaneous activity. After synchronization, the cells were harvested at different time points and the oscillations inmRNA expression were investigated by cosinor rhythmometry.Fibroblasts isolated from adult DEX-exposed mice displayed a smaller amplitude of oscillations inmRNA expression (). This indicates that while the expression ofcan be synchronized across the fibroblast cell population, the cross-synchronization dissipated faster in the fibroblasts derived from DEX-exposed mice, and is consistent with the facilitated circadian re-entrainment we observed in spontaneous activity. , ^{19 20} , ^{21 60 60 61} , ^{20 22} , ^{38 39} Bmal1 Bmal1 Bmal1 Figure 5c

We reported earlier persistent depression-like behavior associated with impaired neurogenesis induced by developmental exposure to low levels of the food contaminant MeHg.In contrast to the DEX-exposed mice, depression appeared at young age, and FLX treatment could restore neurogenesis and reverse the behavioral phenotype.To evaluate the possible occurrence of alterations in circadian entrainment associated with MeHg-induced depression, we reanalyzed the available recordings of spontaneous activity under steady entrainment conditions (see Onishchenko). The amplitude of diurnal fluctuations in spontaneous activity was lower than in controls (), however, MeHg-exposed mice showed robust anticipation of phase change (acrophase occurring consistently before ZT 18;), maintained a circadian period close to 24 h () and the scaling exponent was similar to controls (). Taken together, these findings indicate that, in contrast to DEX-exposed mice, circadian entrainment is not impaired in MeHg-exposed mice. , ^{4 16} , ^{4 15 16} et al. Figure 6a Figure 6b Figure 6c Figure 6d