Into the deep: The subthalamic and para-subthalamic nuclei in behavioral avoidance

The subthalamic nucleus (STN) is a key component of the brain network for movement control. However, the STN is strikingly heterogeneous and also intricately engaged in limbic and cognitive functions. The STN shows aberrant firing activity in several neurological and neuropsychiatric disorders, including Parkinsońs disease (PD). Deep brain stimulation (DBS) in the STN alleviates motor impairment in PD, but patients have reported altered mood as adverse side-effect. Recent observations suggest that optogenetic STN activation in mice induces flight behavior. We hypothesized that STN activation stand at risk of causing an aversive response with behavioral avoidance as consequence. The STN is directly adjoined with the para-STN (pSTN), a hypothalamic area correlated with appetitive and aversive behavior. STN-DBS aiming to correct STN might thereby also modulate pSTN. To dissociate the impact of STN and pSTN, we took advantage of selective promoters in mice, identified in our recent RNA- sequencing of the subthalamic area, to selectively direct optogenetic excitation. Acute photostimulation resulted in aversion via both the STN and pSTN, but only STN- stimulation-paired cues resulted in conditioned avoidance. Viral-genetic tracing coupled with electrophysiological recordings identified a polysynaptic pathway from the STN to the lateral habenula, a critical hub for aversion and associated with clinical depression. This study demonstrates that STN activation is directly correlated with aversion, and thereby contributes neurobiological underpinnings to emotional affect upon STN manipulation with implications for STN-targeted treatment outcome.

Suppl. Fig. 1). 151 To first validate the stimulation protocol, a series of electrophysiological analyses were 152 performed in STN Pitx2 /ChR2 mice. In vivo single cell electrophysiological recordings 153 upon optogenetic stimulation of the STN were performed (Suppl. Fig. 2). An 154 optotagging protocol (Suppl. Fig. 2) was used to stimulate and record within the STN. 155 To observe the reaction of STN neurons to photostimulation, peri-stimulus time 156 histograms (PSTH protocol, 0.5 Hz, 5 ms bin width, 5-8 mW) were created by applying 157 a 0.5 Hz stimulation protocol for at least 100 seconds. Action potentials in ChR2-158 positive STN cells were successfully evoked by STN photostimulation (Suppl. Fig. 2). 159 Once neuronal activity of STN neurons returned to baseline, a photostimulation 160 protocol intended for behavioral experiments was validated (Behavior protocol, 20 Hz, 161 5 ms pulses, 5-8 mW; 10 seconds) which increased the frequency and firing rate of 162 STN neurons for the whole duration of the stimulation, after which they returned to 163 normal (Suppl. Fig. 2). 164 The excitability of STN neurons was next confirmed by patch-clamp recordings in the 165 STN of STN Pitx2 /ChR2 mice with the optic probe placed above the recording site (Suppl. 166 Fig. 2). ChR2-YFP expression was strong in the STN (Suppl. Fig. 2), and all of the STN 167 neurons tested responded to continuous (Suppl. Fig. 2) or 20 Hz trains (Suppl. Fig. 2) 168 of light stimulation by sustained ChR2-mediated currents. When light stimulation was 169 applied in brain slices from STN Pitx2 /CTRL control mice, no current was observed in 170 STN neurons.  stimulation, but they did perform repetitive face grooming. pSTN Tac1 /ChR2 mice 195 showed a different response curve than STN Pitx2 /ChR2 mice, as the grooming did not 196 activation caused conditioned preference, rather than avoidance, to the 262 photostimulation-paired compartment (Fig. 2O, S). Similar as STN mice, pSTN mice 263 showed reversal with the compartmental change in source of stimulation, but none of 264 the groups reached significance on the reversal test day. 265 Taken together, these analyses identified a robust correlation between STN activation 266 and place avoidance both upon direct stimulation (Opto-Avoidance-EPM and Opto-RT-267 CPP; real-time) and upon exposure to aversion-paired cues (Opto-RT-CPP; test days). Based on the findings above, we hypothesized that optogenetic stimulation causing 277 aversive behavioral avoidance should be sufficient to induce a negative reinforcement 278 behavior. To test this hypothesis, we next attempted to assess whether STN Pitx2 /ChR2 279 mice could learn to make active nose-pokes to terminate STN optogenetic activation. 280 In this opto-negative reinforcement paradigm (Opto-NR), all STN Pitx2 /ChR2 mice failed 281 to acquire an active nose-poke response to terminate optogenetic stimulation and the 282 experiment was aborted. 283 Instead, we decided to address the hypothesis that STN optogenetic activation, shown 284 to induce place avoidance, would disrupt positive reinforcement behavior. We also 285 hypothesized that STN and pSTN mice would respond differently, given their 286 differential response in the Opto-RT-CPP conditioning test. To test these hypotheses, 287 an instrumental sugar positive reinforcement paradigm (Sugar-PR) was implemented 288 using operant boxes equipped with two nose-poke (NP) apertures coupled to a reward 289 dispenser with one active setup (NP leads to sugar delivery; active NP) and one 290 inactive setup (no sugar delivery; inactive NP). The original schedule applied consisted 291 of 3 phases (Phase 1, training to nose-poke for sugar reward in a fixed ratio (FR) 292 paradigm; Phase 2, nose-poke to earn sucrose paired with subthalamic photostimulation, Phase 3, sugar reinstatement; photostimulation removed ( Fig. 3A,  294 and Suppl. Fig. 5). 295 In the absence of photostimulation (Phase 1), all STN Pitx2 /ChR2 (Fig. 3B, C) and control 296 (Suppl. Fig. 6) mice showed a significant higher number of active compared to inactive 297 NPs, earning sugar reinforcers. This was the expected response, given the positively 298 reinforcing properties of sugar. With the coupling to STN-photostimulation (Phase 2), 299 control mice continued the same behavior whereas STN Pitx2 /ChR2 mice responded 300 strongly with reduced nose-poke activity on the active NP, a direct consequence of 301 optogenetic STN activation. The difference between active and inactive NPs was no 302 longer significant as the active NPs dropped with the onset of photostimulation. With 303 subsequent removal of photostimulation (Phase 3), STN Pitx2 /ChR2 mice resumed 304 nose-poking (similar to Phase 1), with the number of active NPs significantly higher 305 than the inactive ones. No difference in the number of active NPs were observed 306 between STN Pitx2 /ChR2 and control mice during Phases 1 and 3, while there was a 307 significant decrease selectively for STN Pitx2 /ChR2 mice in the photostimulation phase 308 (Suppl. Fig. 6). Similar results were obtained with STN Vglut2 /ChR2 mice, which made 309 more active than inactive nose-pokes during phases 1 and 3, but with significant 310 decrease during phase 2, leading to significantly lower number of active NPs in phase 311 2 compared to both phase 1 and 3 (Fig. 3D, E). 312 When testing the STN PV /ChR2 mice, no effect of STN-photostimulation was observed. 313 administered as a consequence of the active NPs (Suppl. Fig. 5). This led to an initial 316 increase in the number of active NPs (seeking) due to the absence of the reinforcing 317 stimulus (sugar), followed by a progressive decrease in number of active NPs 318 (extinction). After this followed a Phase 4, in which sugar was again available (Fig. 3F, 319 G). As expected, this restoration of sugar only caused the number of active NPs to 320 again become significantly larger than the inactive ones. Thus, the mice responded to 321 sugar but not to STN photostimulation; no difference was observed between 322 STN PV /ChR2 and control mice in any of the 4 phases (Suppl. Fig. 6). This result 323 supports the outcome of the Opto-Avoidance-EPM test in that the PV + STN 324 subpopulation does not play a first order role in aversive response. 325 The effect of pSTN stimulation in the positive reinforcement test resulted in a response 326 that did not resemble aversion. pSTN Tac1 /ChR2 mice did not show a significant 327 aversive response upon optogenetic stimulation. Unexpectedly, pSTN mice continued 328 to nose-poke for sugar in the presence of photostimulation (Fig. 3H, I). Further, using 329 a Phase 3 in which active NPs were coupled with photostimulation alone (no sugar), 330 no decrease in the number of active NPs was achieved, but instead the number of 331 inactive NPs increased until reaching the same values as the active ones (Fig. 3H, I). 332 To assess this surprising response further, two subsequent phases were added. Now, 333 pSTN-photostimulation and sugar delivery were removed, leaving the acoustic cue as 334 consequence of the active NPs (Phase 4), and subsequently removing also the cue, 335 so that active NPs were equal to inactive NPs (Phase 5) (Suppl. Fig. 5). 336 pSTN Tac1 /ChR2 mice responded during the first sessions of Phase 4 with strongly 337 increased number of active NPs (seeking) and then proceeded towards extinction 338 during the following sessions (Fig. 3H). This result was different to what observed with 339 controls as these showed a strong increase in active NPs when sugar was removed 340 (Suppl. Fig. 6). In contrast to STN-mediated aversion, optogenetic activation of the 341 pSTN did not negatively affect sugar consumption, instead mice maintained their self-342 administration activity in the presence of pSTN stimulation. 343

Distinctly different projections for STN and pSTN 344
Projection targets of STN Pitx2 , STN Vglut2 , STN PV and pSTN Tac1 neurons were next 345 compared to identify circuitry components of the identified behavior (Fig. 4). To first explore any potential impact of STN-activation on LHb activity, in vivo 375 extracellular recordings were performed in STN Pitx2 /ChR2 mice (Fig. 5A). An optic fiber 376 was positioned above the STN and a stimulation protocol (PSTH 0.5 Hz, 5 ms pulses, 377 5-8 mW; at least 100 seconds) was implemented (Fig. 5B). Post-recording, pontamine 378 staining and neurobiotin-marked neurons confirmed the positioning of the recording 379 electrodes within the LHb (Fig. 5C, D). Analysis showed that STN stimulation evoked 380 an excitatory response in 50% of the recorded LHb neurons (onset latency =10.08 ms 381 +/-0.81 ms) (Fig. 5E, F). Thus, a functional STN -LHb connection could be observed. 382 With an apparent lack of direct connection between the STN and the LHb in STN Pitx2 383 mice (confirmed by the absence of eYFP fibers in LHb; Fig. 5D Fig. 7). 424 In the Opto-RT-CPP paradigm (Fig. 7B), an active avoidance behavior to STN-VP-

Discussion 452
The STN is a small thalamic brain structure described as critical to the central motor 453 programs that enable us to move our body purposefully, allowing only relevant movement to be activated upon command. In this study, we identified aversion and 455 avoidance-behavior as a direct consequence of optogenetic STN excitation in mice. 456 This is an important finding considering the strategical position of the STN at the 457 intersection between motor execution, and cognitive and affective function. Excitation 458 of the immediately adjacent pSTN caused a different pattern of avoidance than STN 459 excitation, in accordance with the different neurocircuitries identified for STN and 460 pSTN, respectively. The findings of this study provide a neurobiological framework for 461 emotional affect in the natural context and with implications for neurological and 462 neuropsychiatric disorders in which STN dysfunction is a critical brain pathology. 463 STN is best known for, or at least most studied for, its role in movement inhibition in 464 PD. However, PD contains both motor and non-motor symptom domains 85-87 . PD is 465 progressive in nature, and while no cure exists that restores movement control, 466 treatments alleviating motor incapacity have existed for some decades. Of these, 467 dopamine replacement therapy is the treatment of choice early in the disease, whereas 468 DBS within, or near, the STN is a preferred treatment in late-stage PD 9 . In contrast to Aversion processing is essential to health and well-being, but its disturbed function can 497 have deleterious effects for affected individuals. Aversive stimuli, such as the threat of 498 a predator or the perception of an imminent dangerous situation induce the activation 499 of a "survival state" designed to avoid or reduce the possible harmful outcome. In this 500 context, aversive learning allows the animals to detect the aversive stimulus and learn 501 to actively avoid it, while aversive Pavlovian conditioning serves to associate neutral 502 stimuli to the hostile situation and environment reducing the probability of a related 503 behavior being expressed 105 . Obtaining a reward or avoiding punishment shapes 504 decision-making and motivates learning with several brain regions playing a part is 505 such vital responses 105 . Patients with major depression disorder show mood-506 congruent biases in information-processing, implicating an association between 507 depression and enhanced aversive pavlovian control over instrumental behavior 106 . 508 In contrast to the recent interest in pallidal and habenular structures in aversion 509 processing, the STN has been far less explored in this context. Main interest for the 510 STN using the spatio-temporal specificity of optogenetics has focused around its role 511 in motor control, given its key role in PD and additional movement disorders. functionally. However, one might consider that stimulating electrodes placed in the 570 STN could affect immediately surrounding structures such as the pSTN, to which no 571 specific anatomical border is present to guide stereotaxic surgery. With recent 572 advances in transcriptomics, we recently identified Tac1 as a selective promoter for 573 pSTN 65 , a finding which has been successfully implemented in recent study of the 574 small and elusive pSTN 45,47,50 . Here, taking advantage of a range of selective 575 promoters to dissociate the STN and pSTN, we reasoned that a comparison between 576 these anatomically associated structures would give clues to their roles in affective 577 function, and hence provide information regarding their potential impact on DBS-578 induced side-effects. Indeed, by comparing projectivity and optogenetic excitation of 579 STN and pSTN, their similar but yet clearly distinct roles in aversion and avoidance 580 behavior was evident. 581 Further, not only adjacent structures, but also the functional heterogeneity of the STN 582 itself poses challenges when electrodes are placed in the area and allowed to take 583 control over the multitude of activities that are regulated here. Here, we used Vglut2 584 and Pitx2 promoters to direct opsin selectivity to the whole STN, and also tested PV, identified as promoter active in a subpopulation of STN neurons in both rodents and 586 primates 65,67-70 . Based on the finding that STN excitation causes avoidance behavior, 587 we speculated that the PV-expressing STN subpopulation might contribute to aversion-588 related function. Curiously, the STN PV subpopulation mimicked some, but not all, 589 behaviors assessed and was not concluded as a major player in behavioral avoidance. 590 More work is needed to reveal the exact molecular identity of STN neurons engaging 591 in the newly identified avoidance-type behavior which mice display in the current 592

analyses. 593
To advance precision in treatment and reach symptom-alleviation without causing 594 adverse side-effects, experimental strategies that enable revelation of the full 595 repertoire of behaviors mediated by the subthalamic area are necessary. Considering 596 that numerous adverse side-effects of STN-DBS that have been reported, including 597 low mood state, depression, personality changes and even suicide 13,127 , manipulating 598 the STN might come with a certain risk. While this issue has been challenging to 599 resolve, any direct causality between STN excitation and behavioral aversion in mice 600 is clearly crucial to take into consideration as putatively important not only 601 experimentally, but also clinically. The present results underscore the importance of shown that STN lesioning in rats leads to altered emotional state in response to various 615 rewarding and aversive stimuli 138 . Taken together, both experimental and clinical data 616 clearly highlight the STN structure as critical in affective processing.
Here, we describe the identification of the STN as a non-canonical source of negative 618 emotional value. Well embedded within the brain circuitry of negative reinforcing 619 properties, the presented results point towards a pivotal role of the clinically relevant 620 STN in aversive learning. Evidently, aberrant activity of STN circuitry, as well as its 621 manipulation, may cause both beneficial and detrimental effects. Further studies will 622 be needed to fully reveal the complete neurocircuitry engaging the STN in aversion 623 processing. 624      VP instead of STN. Same battery of behavior tests as detailed above (Fig. 2, 3).                                                                  In situ hybridization histochemistry

Brain preparation
To prepare brain sections for mRNA analysis by fluorescent in situ hybridization (FISH), brains were quickly dissected from mice euthanized by cervical dislocation, snap-frozen in cold (-30°C to -35°C) 2-methylbutane (≥99%, Honeywell) and kept at -80°C until usage. Coronal serial sections were prepared on a cryostat at 16µm thickness and placed onto Superfrost glass slides in series of 8 slides (Thermo Fischer). Slides prepared for FISH analysis were stored in -80°C until usage. Specificity of probes was verified using NCBI blast.
FISH experiments were performed following our previously described protocol 6 .
Cryosections were air-dried, fixed in 4% paraformaldehyde and acetylated in 0.25% Colocalization was determined by the presence of the signals for both probes in the soma of the same cell.

Stereotaxic virus injection and fiber optic probe implantation
Virus injection: Stereotaxic injections were performed in Pitx2 Cre , Vglut2 Cre , PV Cre , STN, EP and VP neurons were recorded in acute brain slices, prepared as previously described 8,9  For optogenetic stimulation, a LED laser source (Prizmatix, Israel) connected to optic fiber (∅: 500 µm) was placed above the brain slice. Light intensities ranged from 4mW

Criteria for exclusion from the analysis
Mice were excluded from individual test analysis if: 1) In the Opto-Open Field they jumped out of the arena during photostimulation so that the analysis of other parameters was made impossible.
2) In the Opto-RT-CPP if, during the pre-test (Day 1), there was a strong initial preference, such that the time spent in a compartment exceeded 80% of the total, or a strong avoidance, so that the time spent in a compartment was less than 20% of the total.
3) During Sugar-PR if they have not acquired a stable rate of self-sugaring, which consisted of at least 10 sugar lozenges received/session with <20% variation during three consecutive days or if they have done more than 30 active nose-pokes in a short time during the initial sessions of the paradigm.

4) In
Opto-Avoidance-EPM if the mice were immobilized due to entanglement of the optical fibers during the test.
Mice were excluded from all analyzes if post-hoc histological examination revealed insufficient viral expression or the cannulas were not close enough to the site of interest. Mice were excluded if histological analysis was not possible due to premature death of the animal or if the animal had to be sacrificed for health reasons during the study. In case of loss of the optical assembly during the study, only the data acquired up to that moment were included. In one case the animal was excluded because an incorrect genotype was revealed during post-hoc analysis.

Habituation
Three weeks after surgery and before the first behavioral test, all mice were handled and habituated to the experimental room and to the optic cables to reduce the stress during the day of the experiment. Before each behavioral test, mice were acclimatized for 30 minutes in the experimental room.

Opto-Open Field
Mice were individually placed in neutral cages for 3 minutes in order to recover after during the "Pre-Test" for either one of the two compartments (<25% or >75% of time spent) were excluded from the statistical analysis.

Opto-Avoidance-EPM
The re-designed version of the EPM test aims to assess the aversive effect induced by the photostimulation in comparison to the natural aversion experienced in the open arms of the apparatus. In the Opto-Avoidance-EPM test, photostimulation was activated upon entry into any of the closed arms, and disabled by leaving it. In contrast, visiting the open arms or occupying the center zone had no effect on photostimulation. The

Sugar Positive Reinforcement Paradigm (Sugar-PR)
Under this protocol mice learned to make active nosepokes to receive sugar pellets.
The experiment is carried out during several consecutive sessions (30 minutes; 1 session/day). Mice were food restricted by administering one daily feeding of 2.5-3g of standard food following each daily session.
Training and testing took place in operant boxes (MED-PC, Med Associates inc, Fairfax, USA) interfaced with lasers for optogenetic stimulation and equipped with nosepoke (NP) devices on each side of a food dispenser. One NP device was designated as active and marked by a light while the device on the other side of the dispenser was selected as inactive. Nose-poking to the active NP carried out when the light is on ("delivering phase"), names as "Active delivering NPs", resulted in a cue tone presentation (0.5s), a 20 mg sucrose pellet delivery (5TUT, TestDiet, St. Louis, USA) and/or laser activation for optogenetic stimulation (20Hz, 5ms, 5mW) according to the different phases of the task, while nose-poking to the inactive side resulted in no cues presentation, sugar delivery and/or laser activation. During 20s that follows a sugar pellet delivery and/or optogenetic stimulation the visual cue for the active NP goes off ("no delivering phase") and active nose-poking during this period did not result in auditory cue presentation, sugar delivery and/or laser activation. Active NPs carried out during this period were registered as "Active no delivering NPs" and summed to the "Active delivering NPs" to obtain the total Active NPs. Nose pokes in the inactive hole, did not activate the sugar dispenser, the laser and the acoustic cue at any time but they were registered as "Inactive delivering NPs" when carried out during the "delivering phase" and as "Inactive no delivering NPs" when performed during the "no delivering phase". "Inactive no delivering NPs" were summed to the "Inactive delivering NPs" to obtain the total Inactive NPs.
Phase 1 (Acquisition phase): Sugar was delivered in response to an Active delivering NPs. This phase lasts until the animal acquires a stable sugar SA rate, which consists of at least 10 received sugar pellet/session with a <20% variation during three consecutive days. Animal that did not acquired a stable sugar SA rate within the third week were excluded from the experiment.
Phase 2 (Sugar-Stimulation phase): Optogenetic stimulation (10 s stimulation 20Hz, 5ms, 5mW) was delivered as a consequence of an active delivering NPs simultaneously to the activation of the sugar dispenser. This phase lasts 5 consecutive sessions (1 session/day).
Phase 3: Depending on the response upon pairing sugar delivering and optogenetic stimulation mice were exposed either to a "only sugar" or to an "only stimulation" phase 3. In the case of a reduction in the number of active NPs, and in order to exclude that such response was not due to a loss of interest for sugar, the subsequent and last phase is characterized "sugar reinstatement" during the following phase 3, restoring the same conditions previously applied during phase 1. If NP activity was similar or higher during phase 2 compared to phase 1 mice were moved to the "only stimulation" phase 3 (5 sessions) in order to assess whether optogenetic stimulation alone was sufficient to maintain a comparable NP activity.
Phase 4: Depending on the response upon "only stimulation" phase 3 mice were exposed either to a "only sugar" phase 4 or to an "only cues" phase 4. In the case of extinction of the active response, the subsequent and last phase 4 (2 sessions) was characterized by "sugar reinstatement" restoring the same conditions previously applied during phase 1. If NP activity was not affected during phase 3 compared to phase 2 mice were moved to the "only cue" phase 4 (3 sessions) where mice were only exposed to visual and auditory cues previously accompanied with sugar and optogenetic stimulation.
Phase 5: Response extinction was finally asses in a last phase (3 sessions) in which active NP activity was followed by no sugar delivery, no optogenetic stimulation and no cues presentation.

Post-injection histological analysis
Following behavioral analyses, recombinase mice were deeply anesthetized and perfused trans-cardially with phosphate-buffer-saline (PBS) followed by ice-cold 4% formaldehyde. Brains were extracted and 60 μm sections were cut with a vibratome.
Fluorescent immunohistochemistry was performed to enhance the YFP signal. All mice were analysed. Mice that displayed strong cellular YFP labeling in the expected area (STN or pSTN) and in which optic cannulas could be confirmed as positioned immediately above the STN or pSTN, respectively were included in the statistical analyses of the electrophysiological and behavioral experiments.
For enzymatic immunohistochemistry mice were deeply anesthetized and perfused trans-cardially with phosphate-buffered-saline (PBS) followed by ice-cold 4% formaldehyde. Brains were extracted and 40 μm and 60 μm sections were cut with a cryostat and a vibratome, respectively. After rinsing in PBS, the endogenous peroxidase inhibition was performed in PBS containing 0.3% X-100 Triton and 1% H2O2 for 15 min. Sections were rinsed in PBS and incubated for 90 min in PBS containing 0.3% X-100 Triton and 5% blocking solution (normal goat serum) followed by incubation with primary antibody diluted in 5% normal goat serum in PBS, overnight