Temporal variations in presynaptic release probability in the lateral habenula

Rhythmicity plays an important role in a number of biological systems. The habenular complex is reported to contain an intrinsic molecular clock and to show rhythmic expression of circadian clock genes and proteins including per2/PER2. In this study, we observed that there is a temporal rhythmicity in the presynaptic efficacy of the lateral habenula (LHb) neurons. We collected a substantial number of recordings at different time points of the day during the light phase. The frequency and amplitude of spontaneous excitatory transmission were increased in the afternoon compared to recordings performed in the morning. In addition, the paired-pulse ratio and the success rate of minimal stimulation were also significantly different depending on the time of the recording. We did not see any significant differences in recordings obtained from pyramidal neurons of the hippocampus in the same brain slices. Taken together, our data indicates that the LHb exhibits intrinsic temporal oscillation in basal neurotransmission and in presynaptic release probability. Given the rapidly growing interest on the function of the LHb, more careful examination of synaptic transmission in the LHb is thus required.

control. We collected a substantial number of recordings of synaptic transmission in the LHb during restricted time windows of the day during the light phase. The heterogeneity of synaptic transmission maintained within the recording time; however we observed a very significant shift in the efficacy of neurotransmission in the LHb depending on the time of the recording. Our observations suggest that more careful examination is required to dissect the function of the LHb in future studies.
To minimize any technical interference due to the long delay (> 6 hours) between slice preparation and recording for the ZT8(ZT1) group, and to match the qualities of the recordings obtained during between different time windows, we prepared acute brain slices at ZT1 and ZT7, and then recorded during ZT2-6 and ZT8-12, respectively. Consistent with previous observations, the frequency of mEPSCs was higher during ZT8-12 (ZT8(ZT7) group) compared to the frequency during ZT2-6 (ZT2(ZT1) group) ( Fig. 2A, p < 0.01 by t-test, p < 0.01 by bootstrap). When the same dataset was re-analyzed for narrower time windows (2-hr time blocks), there was a significant difference in mEPSC frequencies between recording time windows (Fig. 2C, F(3, 86) = 3.29, p < 0.05, one-way ANOVA), however there was no difference within defined groups (p > 0.2 between ZT2-4 and ZT4-6; p > 0.9 between ZT8-10 and ZT10-12, Tukey's post hoc test). Notably, we observed a bimodal distribution of mEPSC frequencies in randomly chosen LHb neurons in both groups (Fig. 2B, p < 0.001, Shapiro-Wilk test), which is in agreement with our previous observations in VTA-projecting LHb neurons 21 . This observation suggests that the heterogeneity of synaptic transmission observed in the LHb is the physiological characteristics of this particular brain area which may not be driven or affected by circadian changes. The threshold for high-frequency mEPSCs was about 7-8 Hz, as reported previously 21 . However, the overall distribution of mEPSC frequencies was shifted toward the right in the ZT8 groups compared to the ZT2(ZT1) group, favoring the appearance of high-frequency cells in the afternoon. In addition, the frequency of mEPSCs in one out of the total of 41 recorded neurons in the ZT2(ZT1) group was higher than 7 Hz, while 8 out of the total of 45 LHb neurons showed high-frequency (> 7 Hz) in the ZT8(ZT7) group. The dominance of neurons with high-frequency mEPSCs in the ZT8(ZT7) group was statistically significant, supporting that mEPSC frequencies are different depending on the time of the day (p < 0.05, χ 2 test, Fig. 2B, shaded area). Interestingly, similar observations were made with our bulk recording reanalysis for each hour. The dominance of high-frequency mEPSCs was temporarily varied during the light phase of daily cycle (Fig. 1A, p < 0.05, χ 2 test). The mEPSC amplitudes in the ZT8(ZT7) group were increased compared to those of the ZT2(ZT1) group (Fig. 2D, p < 0.05 by t-test, p < 0.05 by bootstrap), although two-hr time block analysis revealed no significant difference in mEPSC amplitudes between time windows (Fig. 2F, F(3, 86) = 0.94, p > 0.4). In addition, the mEPSC amplitudes did not follow a normal distribution in either group (Fig. 2E, p < 0.05, Shapiro-Wilk test). Neurons with large-amplitude mEPSCs (> 30 pA) were significantly more abundant in the ZT8(ZT7) group (7 out of 56 cells in the ZT2(ZT1) group vs. 12 out of 34 cells in the ZT8(ZT7) group, p < 0.05, χ 2 test, Fig. 2E, shaded area). However, bulk recording reanalysis of the large-amplitude dominance in every hour did not reach statistical significance (Fig. 1B, p > 0.5, χ 2 test). We found no statistical differences in recordings obtained during ZT8-12 in brain slices prepared at ZT1 and ZT7 (frequency: p > 0.2, amplitude: p > 0.3). This observation suggests that the time of sacrifice did not entrain the intrinsic clock of the LHb. Therefore, intrinsic temporal variations in mEPSC frequencies and amplitudes are maintained in vitro for at least 12 hours.
Next, to determine whether temporal variations in spontaneous transmission commonly occurs in other brain areas known to be circadianly controlled, we performed timely controlled recordings of spontaneous excitatory transmission from the hippocampal neurons as done in the LHb. There was no significant difference in both frequencies and amplitudes of mEPSCs between ZT2(ZT1) group and ZT8(ZT7) group (Fig. 3A,C, frequency: p > 0.4; amplitude: p > 0.8). When analyzed for 2-hr time blocks, we failed to observe any temporal differences between different time windows of recording ( Fig. 3B,D, frequency: F(3, 62) = 1.28, p > 0.2; amplitude: F(3, 62) = 0.16, p > 0.9). These observations strongly suggest that temporal variations in the spontaneous transmission during the light phase are unique characteristics for LHb neurons.
The probability that an action potential invading the presynaptic terminal leads to successful neurotransmitter release varies depending on the specific synapses 22 and can be modified under a variety of circumstances 23 . Changes in mEPSC frequencies often suggest changes in release probability 24,25 . Thus, we measured the presynaptic release probability of LHb neurons or hippocampal neurons by giving a pair of consecutive stimulations (50-ms apart) at different times of day. The reduced amplitude of evoked excitatory postsynaptic currents (eEPSCs) upon the second stimulation compared to the first stimulation often suggests the shortage of available synaptic vesicles due to insufficient time for vesicle recycling 23 . The reduction in the amplitude in response to the second stimulation is considered to be greater in synapses with higher release probability 23 . The paired pulse ratios (PPRs) of eEPSCs in the ZT8(ZT7) group were significantly lower than those of the ZT2(ZT1) group ( Fig. 4A,B, p < 0.001). This observation, together with our observations regarding mEPSC frequencies, suggests that the presynaptic release probability is increased in the LHb in the ZT8(ZT7) group compared to the ZT2(ZT1) group. We failed to observe any significant differences in the PPRs measured in the hippocampus between the two groups ( Fig. 4A,B, p > 0.4). We also examined the failure rate upon minimal stimulation intended to activate only few synapses to address changes in presynaptic efficacy. The failure rate was ~44% in the ZT8(ZT7) group and ~64% in the ZT2(ZT1) group, providing further support for our observation of increased release probability in the afternoon compared to the morning (Fig. 4C, p < 0.01). The average amplitude of the successful evoked events was comparable in both groups (Fig. 4D, p > 0.7).
To examine whether there are any alterations in the relative expressions of postsynaptic AMPA receptors with different permeabilities, we measured the rectification index in the presence of spermine in the pipette solution. We found no significant differences between the two groups ( Fig. 4D, p > 0.8), suggesting that the oscillations in synaptic transmission may be primarily encoded presynaptically. Taken together, our observations strongly suggest that there are temporal variations in presynaptic release probability in the LHb during the light phase.

Discussion
In this study, we showed that there are temporal variations in the efficacy of synaptic transmission occurring in the LHb. The synaptic efficacy in the LHb seems to be heterogeneous, resulting in bimodal distribution of mEPSC frequencies. Two populations of neurons are observed in both recording time windows however, the relative size of two populations seems to differ depending on the time windows of recording across the light phase. Interestingly, the rhythmicity seems mainly presynaptic: the frequency of mEPSCs and the release probability were increased in the afternoon compared to the morning while the relative expression of Ca 2+ -permeable AMPA receptors remained comparable. This oscillation was maintained ex vivo for up to 12 hours.
The LHb is a highly heterogeneous structure, in which physiological properties of nearby neurons are not always comparable 21,[26][27][28][29] . Our current observations add further complexity to the examination of the LHb. Randomized collection of data of substantially large size is not likely to skew the observation. However, if synaptic transmissions are recorded under two different conditions during separate time windows for the majority of each of the experiments, the observed differences could be due to intrinsic temporal variations and not the conditions or treatments of interest. Thus, we urge others to be more careful when analyzing synaptic transmission in the LHb in future studies.
Even though the heterogeneity per se in synaptic transmission occurring in the LHb is not likely under the temporal control, the relative occurrence of high activity neurons seems to be temporally controlled. Current observations cannot pinpoint the origin of temporal control whether it is governed by circadian changes in direct or indirect inputs including from the SCN or by oscillations in expression of clock proteins within the LHb. Given that we observed presynaptic variations as shown in changes in mEPSC frequencies and PPR, it is possible that circadian oscillations in input areas of the LHb may mediate temporal changes in neurotransmission onto the LHb. However, we failed to observe any temporal variations in synaptic transmission occurring in the hippocampus (Fig. 3), another area which is also known to be regulated by SCN 16,24,[30][31][32] and the excitability and synaptic plasticity in which are reported to be under the circadian control 33 . Some studies reported that temporal variations in the habenula complex was limited to the LHb 13,20 however, recently, the medial habenular neurons are also shown to exhibit daily variations similar to our observations. In this study, daily variations were no longer Scientific RepoRts | 7:40866 | DOI: 10.1038/srep40866 observed in clock gene deficient animals, suggesting that the temporal dynamics could be directly regulated by the expression of clock-related genes 34 . Thus, the excitability of the entire habenula complex is likely under the circadian control.
Previous studies using organotypic LHb slice cultures or in vivo preparations also showed that there is a circadian oscillation in spontaneous firing rates in the LHb. Our current study in ex vivo acute brain slice preparation showed that neurotransmission onto LHb neurons is also temporally variable. More importantly, the variations include a significant shift in the dominance of highly active population of neurons and primarily rise from presynaptic efficacy of neurotransmitter release. This is of particular interest given that the LHb activity in animal models of depression were presynaptically but not postsynaptically potentiated and the high-frequency neurons are likely to make greater contribution to the observed abnormal potentiation in helpless rodent models 34 .
Currently available literatures suggest that the temporal variations in synaptic transmission and excitability are likely originated from the intrinsic circadian drive onto LHb neurons. Blockade of neuronal activity by tetrodotoxin (TTX) treatment failed to alter the circadian variations observed in the LHb 12 . The temporal variations in firing rates in the LHb 12 and the mHb 34 were not observed in animals lacking core circadian clock genes. What then may be the underlying cellular mechanisms mediating temporal variations in release probability in LHb synapses? One possible candidate is prokineticin 2, one of important circadian messenger which is released from the SCN and govern behavioral circadian rhythm 35 . Both mRNA and proteins for prokineticin 2 receptros are highly expressed in the LHb 35-37 and prokineticin 2 is shown to alter the presynaptic release probability of GABA in the LHb 12 . In the hypothalamic paraventricular nucleus 38 , prokineticin 2 application increases the frequency of EPSCs in subpopulation of neurons. Therefore, the temporal oscillations in prokineticin release onto LHb neurons may mediate the observed variations in synaptic transmission. However, in SCN brain slices 39 and in trigeminal ganglion primary cultures 40 , prokineticin 2 is reported to suppress GABA-mediated currents postsynaptically. Therefore future studies are required to examine the effect of prokineticin on the temporal variations in the LHb.
Alternatively, the voltage-gated Ca 2+ channel (VGCC), which has been shown to regulate presynaptic vesicle release may play a role in temporal variations observed in the LHb. Several types of VGCCs such as the P/Q-and T-types are known to show clear circadian patterns of expression level changes in SCN neurons 1,41 . In addition, rhythmicity of Ca 2+ influx in SCN neurons was maintained even in the presence of TTX 42,43 . The circadian oscillations of the Per1 and Per2 proteins are blocked by the inhibition of VGCCs 44 , suggesting that oscillatory Ca 2+ flux, in combination with the rhythmic expression of VGCCs, are necessary for generating endogenous circadian oscillations and mediating rhythmic synaptic transmission. Therefore, Ca 2+ oscillation mediated by the VGCCs in LHb neurons may contribute to the generation of diurnal differences in synaptic transmission in the LHb.
A number of studies suggest that circadian rhythmicity is important, primarily for physiologically anticipatory activities [45][46][47][48] . Disruption in circadian rhythms often accompanies other clinical conditions. One shared symptom observed in patients with schizophrenia and depression is the presence of aberrant circadian rhythms 49,50 , including insomnia 3,50 . Several studies employing different stressors observed impaired circadian rhythms of molecular clock proteins including Per1, Clock, and Cry2 in the amygdala, the limbic forebrain and the hypothalamus 51,52 . These studies suggest that stress can disrupt circadian rhythms of brain areas related to emotional processing. Recently, circadian disruption with shorter lengths of the daily cycle has been shown to directly lead to depressive behaviours in rodents in the absence of other stressors through the action of melanopsin-expressing neurons 53 . The LHb is known to project to major monoamine centres such as the VTA, the dorsal raphe and the locus coeruleus. Previously, we have shown that mEPSCs from VTA-projecting LHb neurons are presynaptically potentiated in animal models of depression 21 . Temporal variations in the release probabilities of LHb neurons may thus determine the strengths of the synaptic inputs to monoamine centres, thereby differentially regulating the release of monoamines. Therefore, it is likely that presynaptic temporal variations in neurotransmission of the LHb are altered in different emotional contexts, such as in addicted or stressed conditions. Future studies are anticipated to elucidate molecular mechanisms underlying the rhythmicity of neurotransmission in the LHb and possible alterations in response to different emotional stimuli.

Materials and Methods
Animals. All procedures were carried out in accordance with the guidelines of the National Institutes of Health for animal care and use (http://oacu.od.nih.gov/regs/index.htm) and approved by the Institutional Animal Care and Use Committee of the Konkuk University (KU12063 and KU14155, Seoul, Korea). Male Sprague-Dawley rats were purchased from the Orient Bio company (a branch of Charles River, Gapyung, Korea) and group-housed with free access to food and water under standard conditions. Animals remained in a climate-and light-controlled environment (22 ± 1 °C, 45% humidity, 12:12-hour light/dark cycle with lights on at 7 am) for at least a week before the experiments. Zeitgeber time (ZT) 0 was defined as lights-on (07:00) and ZT12 was defined as lights-off (19:00). Slice preparation. Animals (6-8 weeks old) were anaesthetized using isoflurane. After immediate decapitation, brains were stored in ice-cold dissection buffer (in mM: 212 sucrose, 3 KCl, 26 NaHCO 3 , 1.25 NaH 2 PO 4 , 7 MgCl 2 , and 10 glucose, gassed with 95% O 2 and 5% CO 2 ). Transverse slices (350-μ m thick) containing the hippocampus and the LHb were prepared using a Leica VT 1000 S Vibratome. Brain slices were transferred to a recovery chamber containing artificial cerebrospinal fluid (aCSF) (in mM, 118 NaCl, 2.5 KCl, 11 glucose, 1 NaH 2 PO 4 , and 26.2 NaHCO 3 , gassed with 95% O 2 and 5% CO 2 ) at 35 °C for one hour and then stored at room temperature. Slice preparation was performed at ZT1 (08:00) and ZT7 (14:00). All experiments were performed at 27-30 °C.
Electrophysiology. CA1 pyramidal neurons or LHb neurons (mostly in the medial part of the LHb) were voltage-clamped to − 60 mV in warmed aCSF using Axopatch 200B and Clampex 10.3 (Molecular Devices), or HEKA EPC8 and pulse v8.8 (HEKA Electronik), filtered at 5 kHz and sampled at 10 kHz. Glass pipettes with a resistance of 2-6 MΩ were filled with an internal solution containing the following (in mM): 115 Cs methanesulphonate, 20 CsCl, 10 HEPES, 2.5 MgCl 2 , 0.6 EGTA, 5 QX314, 4 Na 2 -ATP, 0.4 Na 2 -GTP, and 10 Na-phosphocreatine (pH 7.3). For recordings of evoked transmission, QX314 (5 mM) was added to the internal solution. The Schaffer collateral pathway (in the hippocampus) or the stria medullaris (in the LHb) was stimulated to elicit eEPSCs using platinum/iridium cluster electrodes. Each response was recorded for 15 sweeps with 20 sec intervals to average the amplitude of eEPSCs. The ionotropic GABAergic receptor antagonist picrotoxin (PTX, 50 μ M; Sigma, in DMSO) was added to the aCSF to exclude GABAR-mediated inhibitory synaptic transmission. mEPSCs were recorded in the presence of 1 μ M TTX and 50 μ M PTX, and analysed manually to avoid false-positive and false-negative events using Mini Analysis software (Synaptosoft). Data analysis. Data were analyzed using Clampfit 10.3 (Molecular Devices) or Mini Analysis software (Synaptosoft). Values are presented as means ± standard error of the mean (SEM). n indicates the number of cells studied. Shapiro-Wilk test was used for testing normality for data sets. Two-tailed unpaired t-tests, bootstraps for mean and, one-way ANOVA were used for statistical comparisons between groups unless stated otherwise. For bootstrapping which relies on random sampling with replacement to improve the comparison between not-normally distributed data sets, means (N i and M i ) of two data sets (N and M of size n and m) which were randomly sampled n and m times, respectively were calculated. The generation of means was repeated 10,000 times. If N j was more than M j fewer than 1 or 5% of the times, then the probability that N is more than M was considered to be less than 0.01 or 0.05, respectively.