An Inhibitory Medial Preoptic Circuit Mediates Innate Exploration

Animals have an innate motivation to explore objects and environments with unknown values. To this end, they need to activate neural pathways that enable exploration. Here, we reveal that photostimulation of a subset of medial preoptic area (MPA) neurons expressing the vesicular-GABA transporter gene (vgat+) and sending axonal projections to the ventrolateral periaqueductal gray (vPAG) increases exploration in a chamber but causes no place preference when tested there without photostimulation. Photoinhibition of MPAvgat–vPAG projections leads to no emotional changes as measured by normal activity in an open field assay. Electrophysiological recordings revealed that most GABAergic vPAG neurons are inhibited by MPAvgat neurons. In contrast to a previous report that suggested that MPAvgat–vPAG neurons may impart positive valence to induce place preference, our results suggest that these neurons can increase innate exploration.


INTRODUCTION
In nature, animals explore the environment for their survival. Such exploration is required for finding food, prey, monitoring for predators, and investigating males, females, and offspring. Various types of motivated behavior can lead to exploratory behavior, such as hunting, consuming, aggression, reproduction, parental care, and defensive behavior.
Exploration can be broadly divided into two types: extrinsic exploration is driven by the completion of a goal, such as food to satiate hunger or scanning for predators, whereas intrinsically driven exploration occurs due to non-homeostatic or reproductive drives and has been speculated to be curiosity and novelty-seeking driven, driving investigation of objects, stimuli, and environments for its own sake (Hughes, 1997).
Studies that have attempted to elucidate the mechanisms underlying intrinsic exploration have typically involved inducing altered states of arousal or anxiety and observing the effects that such manipulations have on exploratory behavior. Such research includes the use of drugs such as methamphetamines, amphetamines, and methylphenidate (Berlyne et al., 1966;Dyne and Hughes, 1970;Robbins and Iversen, 1973) to alter states of arousal and the use of benzodiazepines and chlordiazepoxide (Hughes, 1972;Russell, 1973;Crawley et al., 1981;Holmes and Rodgers, 1999) to alter anxiety. Although such studies help to identify possible relationships between brain states and intrinsic exploration, the brain regions and neural circuitry underlying such phenomena have received little attention.
The medial preoptic area (MPA) is part of the anterior hypothalamus and has been implicated in playing a role in many types of innately motivated behavior, such as hunting (Park et al., 2018), anxiety (Zhang et al., 2021), reproductive (Wei et al., 2018;McHenry et al., 2017), and parental behavior (Wu et al., 2014;Kohl et al., 2018;Li et al., 2019;Zhang et al., 2021). Although such studies have also shown to a degree the capability of the MPA to increase extrinsic exploration, due to alterations in underlying homeostatic drives or reproductive drives, the potential role of the MPA in modulating intrinsic exploration has received little attention. A recent study (Zhang et al., 2021) showed that optogenetic activation of vgat neurons in the MPA is capable of increasing time spent in the stimulated side of a real-time place preference (RTPP). However, this experiment by itself fails to distinguish between intrinsic and extrinsic exploration, since changes seen in RTPP could be driven either due to changes in exploratory behavior itself or due to a reinforcing effect. Since exploration induced by motivational drives of the MPA would be expected to be associated with either aversive or appetitive valences, we sought to distinguish between extrinsic and intrinsic exploration by validating whether any changes in exploratory activity were accompanied by driving reinforcement through the use of conditioned place preference (CPP) tests. Thus, to this end, we employed optogenetics to specifically modulate vgat neurons of the MPA and validate if such modulations alter intrinsic exploration.

Animals
The animal study was conducted according to the Korean Advanced Institute of Science and Technology (KAIST) Guidelines for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee (Protocol No. KA2020-63). Vgat-ires-Cre (mixed background, Jackson lab, 016962), aged 7-8 weeks, were maintained under a 12-h light/dark cycle with ad libitum access to food and water. Both sexes of mice for behavioral experiments were group housed (three to five mice per cage), except for when they were isolated before a home cage test. Behavioral experiments were conducted 3-4 weeks after virus injection and surgical procedures. For photostimulation experiments, we used six mice to photoactivate MPA vgat neurons and four mice as controls. Seven mice were used to photoactivate PAG-projecting MPA vgat neurons and three mice as controls. Seven mice were used for photoinhibition of PAG-projecting MPA vgat neurons and four mice as controls.

Behavioral Experiments
All behavioral tests were conducted in a sound attenuation booth during the dark cycle, and all mice were handled for 10 min each day for 5 days before performing any test, except RTPP/CPP tests. The mice had an interval of 72 h between experiments. All mice were evaluated for viral expression and excluded if they lacked proper expression or implantation of fiber optic cannula with reference to a brain atlas (Paxinos and Franklin, 2008). No sexdependent changes were observed; thus, male and female mice were pooled together.

Open Field Test
Mice were habituated in an open field test chamber (40 cm × 40 cm × 40 cm) for 10 min. The experiment was carried out for a total of 9 min. Sessions were conducted as light OFF-ON-OFF, each session lasting for 3 min. The locomotion of mice was recorded with a camera to measure distance movements (Kalueff et al., 2006).
We use the Gaussian mixture model to analyze locomotive patterns. To analyze locomotive patterns, we plotted the density of the log speed and used a Gaussian mixture model to analyze the distribution of these log speeds. The parameters of the model are estimated by using the Expectation-Maximization (EM) algorithm. The algorithm estimates the maximum likelihood parameters (mean, variance, and weights) of a mixture with a given number of clusters. We used three clusters: lingering, progressing, and rapid movement (fast progressing). The EM algorithm is an iterative algorithm that starts with user-given initial values and incrementally improves the likelihood function until further iterations yield only a negligible improvement (Drai et al., 2001).

Attached Object Test
Mice were habituated in a test chamber (40 cm × 40 cm × 40 cm) for 10 min. After habituation, an object (2 cm × 2 cm × 2 cm) was attached to the center of the chamber. The experiment was carried out for a total of 9 min. Sessions were conducted as light OFF-ON-OFF, each session lasting for 3 min. The locomotion of mice was recorded with a camera to measure frequency of approaching toward the object zone (20 cm × 20 cm). Linear regression was done by using Microsoft Excel.

Home Cage Test
Mice were habituated in their home cage for 10 min while connected to an optic cable. The experiment was carried out for a total of 9 min. Experiments were conducted as sessions of light OFF-ON-OFF, with each session lasting for 3 min. The locomotion of mice was recorded with a camera to measure rearing behavior.

Real-Time Place Preference and Conditioned Place Preference Test
The RTPP and CPP apparatus consisted of two rectangular chambers (20 × 18 cm 2 ) with distinct wall drawings and a corridor separating them. One rectangular chamber had a polka dot pattern, and the other rectangular chamber had a striped pattern. A video-tracking and analysis system (EthoVision XT 11.5 software, Noldus, Wageningen, Netherlands) recorded and analyzed all animal movements. The protocol for RTPP/CPP tests was taken from a reference (Tan et al., 2012). The paradigm consisted of three sessions over 5 days. On day 1, as a pretest session, mice freely explored the chamber for 15 min without light. We excluded mice showing a side preference higher than 35%. Days 2-4 were conditioning sessions; mice were trained for 30 min with photostimulation given in a light paired chamber. The chamber that was light paired was randomly assigned to each mouse in a counterbalanced manner. Photostimulation was triggered whenever mice entered the light-paired chamber, through a transistor-transistor logic (TTL) signal using a mini I/O box with EthoVision XT. To avoid overheating of the brain, lasers were turned off if mice stayed in the light-paired chamber for longer than 30 s. If mice continued to stay in the light-paired chamber, 1 min after the light pulse was turned off, the laser was turned on again. On day 5 as a posttest session, 24 h after the conditioning session, mice explored the chambers for 15 min without light.
Time in stim chamber (%) was calculated as (Time spent in the light-paired chamber)/(Time spent in either chamber) × 100.
Photostimulation was delivered through an OptoPatcher (A-M Systems, Sequim, WA, United States), connected to a laser source (473 nm; Shanghai Lasers, Shanghai, China), through a patch cord with an NA of 4.8. Light intensity at the end of the optic fiber was measured as 0.4 mW. CRACM experiments (Petreanu et al., 2007) were conducted in voltage-clamp mode at −60 and −10 mV to detect excitatory and inhibitory postsynaptic currents, respectively. Three single light pulses (10 ms) were delivered 1 s apart by triggering a pulse generator with pClamp software.

Statistics
No statistical analyses were performed to predetermine sample sizes. The sample sizes used were similar to those used in many previous studies (Kim et al., 2017;Park et al., 2018). All data analyses were performed using SigmaPlot (12.0; Systat Software). For parameters that followed a normal distribution (Shapiro-Wilk test, p > 0.05), differences between two groups were analyzed with the Student's t-test, and comparisons of three or more groups were performed with the analysis of variance (ANOVA). The Wilcoxon signed-rank test, Mann-Whitney U test, and the signed-rank test were used for data that were not normally distributed. All statistical test were two-sided, and p < 0.05 were considered statistically significant.

MPA vgat Neurons Induce Exploration Without Reinforcement
To investigate if GABAergic neurons in the MPA were associated with exploration behavior, we first performed optogenetic experiments (Zhang et al., 2006). We unilaterally injected adeno-associated virus (AAV) particles containing a Cre-dependent, channelrhodopsin (ChR2) virus (AAV.EF1α.DIO.hChR2.mCherry) into the MPA of vgatires-Cre mice and implanted fiber-optic cannulas over the vPAG. AAV.EF1α.DIO.mCherry virus was injected as a control ( Figure 1A). Histology confirmed expression of the viruses in MPA neurons ( Figure 1B). We conducted behavioral experiments 4 weeks after surgery. Blue illumination (473 nm, 20 Hz, 5 ms, 3 mW, pulse) was delivered through the implanted optic fiber to activate the soma of MPA vgat neurons. In an open field test (OFT), an exploratory test (Brown et al., 1999) Since exploration by mice is characterized by alternating bouts of rapid movement, progression, and lingering, we further analyzed locomotive patterns shown by each group of mice by log-transforming speeds of mouse trajectories and separating them into three clusters (slow, medium, fast) using the Expectation-Maximization algorithm (Drai et al., 2001) (Supplementary Figures 1A,B). Photostimulation resulted in a rightward shift in the mean of each cluster (ChR2 mice: slow, 0.22 ± 0.05; medium, 0.81 ± 0.02; fast, 1.40 ± 0.02; mCherry mice: slow, −0.12 ± 0.03; medium, 0.51 ± 0.05; fast, 1.20 ± 0.06) {Supplementary Figure 1C Extrapolating with linear regression, we found that the increase in frequency in the object zone of ChR2-expressing mice was higher compared the mCherry-injected mice, even when considering their increased speed (Supplementary Figure 2C).
To test whether artificial activation of MPA vgat neurons induced exploration and could be reinforcing, we conducted RTPP and CPP tests (Tan et al., 2012). An RTPP/CPP test box containing two chambers with distinct wall drawings and a corridor separating them was used. Mice freely explored the Frontiers in Neuroscience | www.frontiersin.org two chambers for 30 min/day, while receiving photostimulation of MPA vgat neurons in one of the chambers, over 3 days. During pre-and posttest sessions, mouse freely explored the test box without photostimulation for 15 min. We found that the average time in the stimulation chamber during the conditioning session increased significantly when compared to the pretest session in ChR2 mice (ChR2_pre, 50.87 ± 2.10; ChR2_con, 89.89 ± 2.28) [Figures 1E,F; 1F, t(5) = 11.607, p < 0.001, paired t-test]. However, mCherry control mice showed no differences (mCherry_pre, 48.98 ± 3.10; mCherry_con, 50.69 ± 2.09) [ Figure 1G; t(3) = 1.074, p = 0.361, paired t-test]. Interestingly, during the posttest session, 24 h postconditioning, ChR2expressing mice (53.00 ± 2.46) and control mice (51.73 ± 3.09) showed no difference in time spent in the stimulation chamber [ Figure 1H; t(8) = 0.325, p = 0.753, unpaired t-test]. To validate if activation of MPA vgat neurons has a reinforcement effect, we analyzed the RTPP data across 5-min time windows. On conditioning day 1, ChR2-expressing mice did not show any progressive increases in the time spent in the stim chamber {Supplementary Figures 2B, 3A Based on this CPP result, the observed preference in the stimulation chamber during the RTPP test can be explained due to an increase in exploration. Together, these results suggest that GABAergic neurons in the MPA can induce exploration behavior but fail to condition mice.

MPA vgat Neurons Send Inhibitory Output to vPAG vgat Neurons
To identify the functional connectivity of GABAergic neurons in the MPA, we performed whole-cell patch-clamp recordings.

MPA vgat -vPAG Projections Induce Exploration Behavior Without Reinforcement
To activate GABAergic neurons in the MPA projecting to the vPAG, we injected AAV.EF1α.DIO.hChR2.mCherry virus unilaterally into the MPA of vgat-ires-Cre mice and implanted fiber-optic cannulas over the vPAG. We injected AAV.EF1α.DIO.mCherry virus into vgat-ires-Cre mice as a control (Figure 3A) 16) = 1.267, p = 0.308]; two-way repeated measures ANOVA}, although we found the correlation between frequency in object zone and velocity to be steeper for ChR2-injected mice (Supplementary Figure 2E).

DISCUSSION
In the present study, we reveal that vPAG projecting MPA vgat neurons induce exploration and that these vgat neurons in the MPA send inhibitory input to vgat neurons of the vPAG. We found that activation of MPA vgat neurons increases locomotion and time spent in the stimulation chamber during RTPP but do not show any preference during CPP. We interpreted this increased time spent in the simulation chamber during realtime tests as not being reinforcing but as increasing exploration. We recapitulated this result when activating the MPA vgat -vPAG circuit. However, inhibition of the MPA vgat -vPAG circuit did not decrease exploration behavior.

Activation of MPA GABAergic Neurons Is Not Reinforcing
A previous study showed that vgat neurons, a subset of GABAergic neurons, in the MPA encode positive value (Zhang et al., 2021). Indeed, we recapitulated the results from the RTPP and open field tests in this study. However, our study showed that activation of MPA vgat neurons and MPA vgat -vPAG projections did not induce any difference in time spent in a stimulation chamber 24 h after a conditioning session (Figures 1H, 3F), suggesting that the increased preference seen during RTPP was not due to a reinforcing effect. We found that there was no progressive increase in preference during RTPP tests, consistent with the notion that activation of MPA vgat neurons does not have a reinforcing effect. One important caveat to note is that we cannot rule out the hypothesis that these effects may have been caused by contextual memory impairments induced by activation of MPA neurons. Taken together, we tentatively interpret these results to mean that MPA vgat neurons promote exploration in a non-reinforcing manner, although further experiments will be needed to conclusively rule out a change in memory.

Inhibition of MPA GABAergic Neurons Does Not Decrease Exploration
During photoinhibition of the MPA vgat -vPAG circuit, we found no decreases in either locomotion or time spent in the stimulation chamber during an RTPP test. Although these results may tentatively suggest that MPA vgat -vPAG is not necessary for exploration, it is important to note that our experiments did not include other stimuli during the test. Previous studies on the function of MPA neurons projecting to the vPAG have shown that these neurons respond to particular stimuli, such as CaMKIIα neurons responding to an object or prey, vglut2 neurons responding to stress, and galanin neurons responding to pups (Wu et al., 2014;Kohl et al., 2018;Park et al., 2018;Zhang et al., 2021). Thus, it is plausible that there will be certain stimuli or conditions that could cause activation in the MPA vgat -vPAG circuit that in turn facilitate exploration. The absence of such a stimulus would result in no difference in exploration when inhibiting this circuit, which may account for our results. Furthermore, the vPAG is likely to receive presynaptic input from other brain regions, which may have also contributed to the lack of any effect seen from inhibiting the MPA vgat -vPAG circuit. Further studies will be needed to reveal the detailed function of this circuit.

The MPA Is Associated With Broad Exploration
Exploratory behavior can consist of orienting responses, locomotor responses, and investigatory responses, which refer to orientation of sensory organs, displacement of the whole body, and manipulation of objects in the environment, respectively (Hughes, 1997). It is thus possible that the MPA is more broadly involved with all types of exploratory behaviors, whereas the MPA-vPAG connection is restricted to only mediating the locomotory responses of exploration, without affecting investigatory responses (Supplementary Figure 2). This would explain why photoactivation of MPA-vPAG resulted in preference during RTPP and increased locomotion but no changes in object exploration. However, this locomotory response appears to occur in a non-reinforcing manner, since photostimulation of the MPA-vPAG projection could not condition mice during CPP. Notably, this increase in locomotion occurred independently of changes in locomotor speed, suggesting that it was not a direct motor effect (Supplementary Figure 3). Further analysis of locomotor patterns showed that alternations between lingering, progressing, and rapid movement, which are hallmarks of exploratory behavior (Drai et al., 2001), were maintained during photostimulation (Supplementary Figure 1). Since the medial septum is known to project to the MPA, which has been shown to mediate speed increases during locomotion , it is plausible that direct activation of MPA neurons can result in changes in speed during photostimulation. On the other hand, such changes in speed may be dependent on specific projections of the MPA to downstream sites, rather than a generalized effect, which may account for why the MPA-vPAG projection did not show any changes in speed during photostimulation (Supplementary Figure 3). Overall, our results suggest that MPA vgat neurons are involved in mediating exploration in a non-reinforcing manner, which is essential for organisms in familiarizing themselves with new environments.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

ETHICS STATEMENT
The animal study was reviewed and approved by the Korean Advanced Institute of Science and Technology (KAIST) Guidelines for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee (Protocol No. KA2020-63).

AUTHOR CONTRIBUTIONS
DK designed the study and coordinated the experiments. JR performed the behavioral, optogenetic, and histological experiments and analyzed the data. SP performed the electrophysiology experiments. All authors participated in writing the manuscript.

ACKNOWLEDGMENTS
We thank A. Shin for assistance with experiments.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fnins.2021. 716147/full#supplementary-material Supplementary Figure 1 | The pattern of the density function for speeds was conserved. (A) The distributions were fitted with the EM algorithm. The density estimation for the distribution of the log speed for the ChR2 group (n = 6; n = 3 males; n = 3 females) during MPA vgat neural photoactivation in the OFT. (B) The density estimation for the distribution of the log speed for the mCherry group (n = 4; n = 1 male; n = 3 females). (C) The mean of each distribution for slow, medium, and fast clusters in the OFT; slow: * * * p < 0.001; medium: * * * p < 0.001; fast: * * p = 0.003. Error bars represent SD. (D) The variance of each distribution for slow, medium, and fast clusters in the OFT; slow: p = 0.964; medium: p = 0.330; fast: p = 0.541. Error bars represent SD. (E) The weight of each distribution for slow, medium, and fast clusters during the OFT; slow: p = 0.861; medium: p = 0.683; fast: p = 0.814. Error bars represent SD. (F) The density estimation for the distribution of the log speed for the MPA vgat -PAG ChR2-expressing group (n = 7; n = 4 males; n = 3 females) during photoactivation in OFT. (G) The density estimation for the distribution of the log speed for the mCherry control group (n = 3; n = 1 male; n = 2 female). (H) The mean of each distribution for slow, medium, and fast clusters during photostimulation of PAG-projecting MPA vgat neurons in the OFT; slow: * * * p < 0.001; medium: * p = 0.025; fast: p = 0.534. Error bars represent SD. (I) The variance of each distribution for slow, medium, and fast clusters during photostimulation of PAG-projecting MPA vgat neurons in the OFT; slow: p = 0.537; medium: p = 0.169; fast: p = 0.889. Error bars represent SD. (J) The weight of each distribution for slow, medium, and fast clusters during photostimulation of PAG-projecting MPA vgat neurons in the OFT.