The NeuroD6 Subtype of VTA Neurons Contributes to Psychostimulant Sensitization and Behavioral Reinforcement

Abstract Reward-related behavior is complex and its dysfunction correlated with neuropsychiatric illness. Dopamine (DA) neurons of the ventral tegmental area (VTA) have long been associated with different aspects of reward function, but it remains to be disentangled how distinct VTA DA neurons contribute to the full range of behaviors ascribed to the VTA. Here, a recently identified subtype of VTA neurons molecularly defined by NeuroD6 (NEX1M) was addressed. Among all VTA DA neurons, less than 15% were identified as positive for NeuroD6. In addition to dopaminergic markers, sparse NeuroD6 neurons expressed the vesicular glutamate transporter 2 (Vglut2) gene. To achieve manipulation of NeuroD6 VTA neurons, NeuroD6(NEX)-Cre-driven mouse genetics and optogenetics were implemented. First, expression of vesicular monoamine transporter 2 (VMAT2) was ablated to disrupt dopaminergic function in NeuroD6 VTA neurons. Comparing Vmat2lox/lox;NEX-Cre conditional knock-out (cKO) mice with littermate controls, it was evident that baseline locomotion, preference for sugar and ethanol, and place preference upon amphetamine-induced and cocaine-induced conditioning were similar between genotypes. However, locomotion upon repeated psychostimulant administration was significantly elevated above control levels in cKO mice. Second, optogenetic activation of NEX-Cre VTA neurons was shown to induce DA release and glutamatergic postsynaptic currents within the nucleus accumbens. Third, optogenetic stimulation of NEX-Cre VTA neurons in vivo induced significant place preference behavior, while stimulation of VTA neurons defined by Calretinin failed to cause a similar response. The results show that NeuroD6 VTA neurons exert distinct regulation over specific aspects of reward-related behavior, findings that contribute to the current understanding of VTA neurocircuitry.


Introduction
The midbrain dopamine (mDA) system mediates a diverse spectrum of behaviors and their dysfunction is correlated with a range of severe behavioral disorders including substance use disorder, schizophrenia, ADHD and Parkinson's disease (PD). Consequently, therapies based on modulating the activity of the mDA system are commonly prescribed, however, due to their unselective nature, current treatments often fail to alleviate symptoms and instead cause adverse effects (Weintraub, 2008;Divac et al., 2014). One reason for the lack of successful treatment is incomplete understanding of the underlying neurobiology. Indeed, it is increasingly understood that the mDA system is highly heterogeneous (for review, see Pupe and Wallén-Mackenzie, 2015;Morales and Margolis, 2017). Beyond the classical separation into the ventral tegmental area (VTA) and substantia nigra pars compacta (SNc), with VTA projections to cortical and limbic target areas and SNc projections to the dorsal striatum subserving cognitive/affective and motor functions, respectively (Björklund and Dunnett, 2007), a higher level of complexity is now being unfolded: afferent and efferent projections, electrophysiological patterns, capacity for glutamate or GABA co-release, and responsiveness to appetitive or aversive stimuli are some of the properties that distinguish mDA neurons from each other (Lammel et al., 2011;Beier et al., 2015;Menegas et al., 2015;Faget et al., 2016).
Likely coupled to this functional diversity is a complex diversity in molecular identity. Microarray-based analyses have identified gene expression patterns enriched in VTA over SNc DA neurons (Chung et al., 2005;Greene et al., 2005;Viereckel et al., 2016) while single cell profiling has begun to identify combinatorial gene expression patterns that molecularly define subtypes of mDA neurons (Poulin et al., 2014;La Manno et al., 2016;Hook et al., 2018). Based on this new knowledge, intersectional genetic approaches were recently described in which the distinct projection pathways of several newly defined subtypes of mDA neurons were identified (Poulin et al., 2018). By forwarding the current knowledge toward molecularly defined, and thus targetable, subtypes of mDA neurons with distinct projection patterns, these recent advances enhance the possibility of improving selectivity in treatment of dopaminergic disorders. However, a key issue that remains to be resolved is how each molecularly defined subtype of DA neuron contributes to the complex range of behaviors ascribed to the mDA system.
The gene encoding the transcription factor NeuroD6 (also known as NEX1M) has recently gained attention due to its selective expression within subsets of VTA DA neurons while being excluded from the SNc (Viereckel et al., 2016;Khan et al., 2017;Kramer et al., 2018). VTA DA neurons are of particular interest for several reasons. First, the importance of VTA DA neurons in several aspects of behavioral reinforcement and conditioning has been established through classical studies (for review, see Di Chiara and Bassareo, 2007;Ikemoto, 2007), and more recently, by the use of optogenetics (Tsai et al., 2009;Kim et al., 2012;Ilango et al., 2014;Pascoli et al., 2015). However, detailed knowledge of the exact nature of those particular DA neurons that contribute to each of these complex behaviors remains elusive. Second, medial DA neurons mediate the most potent responsiveness to addictive drugs via their projection to the nucleus accumbens shell (NAcSh; Ikemoto and Bonci, 2014). The possibility to ascribe specific aspects of drug responses to a distinct subtype of VTA DA neurons would therefore enhance the understanding of addictive behavior. Third, certain VTA neurons show resistance to degeneration in PD (Brichta and Greengard, 2014); however, depending on their role in behavioral regulation, surviving VTA neurons might contribute to non-motor symptoms including behavioral addictions (Cenci et al., 2015).
While NeuroD6-expressing DA neurons were recently identified as neuroprotected in experimental PD (Kramer et al., 2018), the potential role of this newly described subtype of VTA neurons in behavioral regulation has remained unexplored. Here, we implemented NeuroD6-Cre mice (also known as NEX-Cre) to create opportunities for targeting and manipulation of the NeuroD6 subtype VTA neurons. We show that gene targeting of vesicular monoamine transporter 2 (VMAT2) within this particular DA neuron subtype elevated the locomotor response to psychostimulants while activation of NeuroD6-Cre neurons by optogenetic stimulation in the medial VTA induced DA release and glutamatergic postsynaptic responses in the NAcSh. In vivo optogenetic activation of the NeuroD6-Cre VTA subpopulation in a realtime place preference (RT-PP) failed to trigger a conditioned response (CR) but induced place preference upon direct
For double and triple ISH using riboprobes [fluorescent ISH (FISH)  Detection of Th, Dat, Vglut2, Viaat, Calb2, NeuroD6 mRNA, and Vmat2 probe 1 and probe 2 mRNA in brain tissue using ISH was performed following a previously published protocol (Viereckel et al., 2016). Briefly, mice were sacrificed and brains dissected. Coronal cryosections were prepared, air-dried, fixed in 4% paraformaldehyde and acetylated in 0.25% acetic anhydride/100 mM triethanolamine (pH 8) followed by hybridization for 18 h at 65°C in 100 l of formamide-buffer containing 1 g/ml digoxigenin (DIG)-labeled probe for colorimetric detection or 1 g/ml DIG-or 1 g/ml fluorescein-labeled probes for fluorescent detection. Sections were washed at 65°C with SSC buffers of decreasing strength, and blocked with 20% FBS and 1% blocking solution. For colorimetric detection, DIG epitopes were detected with alkaline phosphatase-coupled anti-DIG fab fragments at 1:500 and signal developed with NBT/BCIP. For fluorescent detection, sections were incubated with HRPconjugated anti-fluorescein antibody at 1:1000 concentration (Roche catalog #11426346910, RRID:AB_840257). Signals were revealed with the TSA kit (PerkinElmer catalog #NEL749A001KT) using biotin tyramide at 1:75 concentration followed by incubation with neutravidin Oregon Green conjugate at 1:750 (Invitrogen catalog #A-6374, RRID:AB_2315961). HRP-activity was stopped by incubation of sections in 0.1 M glycine and 3% H 2 O 2 . DIG epitopes were detected with HRP-conjugated anti-DIG antibody at 1:1000 (Roche catalog #11207733910, RRID:AB_514500) and revealed with TSA kit (PerkinElmer catalog #NEL744A001KT) using Cy3 tyramide at 1:200. For triple FISH, TH mRNA was detected with dinitrophenyl (DNP)-labeled probe; NeuroD6 mRNA with DIG-labeled probe and Vglut2 mRNA with fluorescein-labeled probe. The protocol was the same as described above until revelation: DIG epitopes were detected with HRP anti-DIG fab fragments at 1:3000 and revealed using Cy3 tyramide at 1:50 followed by glycine and H 2 O 2 treatment. Fluorescein epitopes were detected with HRP anti-fluorescein fab fragments at 1:5000 and revealed using Cy2 tyramide at 1:250 by glycine and H 2 O 2 treatment. DNP epitopes were detected with HRP anti-DNP fab fragments at 1:1000 and revealed using Cy5 tyramide at 1:50, followed by incubation with DAPI. Fluorophore tyramides were synthetized as previously described (Hopman et al., 1998). All slides were scanned and analyzed on NanoZoomer 2.0-HT Ndp2.view (Hamamatsu). Stereotaxic reference atlases (Franklin and Paxinos, 2008;Fu et al., 2012) were used to outline anatomic borders.

Validation of NEX-Cre-mediated recombination of floxed Vmat2 exon 2
Upon genotyping, PCR-validated Vmat2 lox/lox;NEX-Cre-tg (cKO) and Vmat2 lox/lox;NEX-Cre-wt (Ctrl) mice were sacrificed and brains analyzed by ISH to verify NEX-Cre-driven re-combination of the floxed exon 2 of the Vmat2 gene in cKO mice. Littermate Ctrl mice were used to validate wild-type Vmat2 mRNA. A Vmat2 mRNA two-probe approach was implemented to visualize cells positive for wild-type Vmat2 mRNA and cells positive for a truncated Vmat2 mRNA generated on NEX-Cre-driven recombination of the floxed Vmat2 exon 2. Probe 1 (green) was designed for detection of Vmat2 mRNA derived from exon 6 -15 and probe 2 (blue) for detection of mRNA from exon 2. In control mice, both probe 1 and probe 2 can bind their target mRNA (wild-type Vmat2 mRNA). Combination of probe 1 and probe 2 gives rise to combined blue and green labeling in wild-type DA neurons. In cKO mice, Vmat2 exon 2 will be deleted specifically in cells expressing the NEX-Cre transgene, leading to production of Vmat2 mRNA missing exon 2 but maintaining exons 6 -15. In Vmat2-expressing cells that do not express the NEX-Cre transgene in cKO mice, wild-type Vmat2 mRNA will be produced. Vmat2-targeted cells can thus be identified based on lack of blue color (probe 2) and presence of green color only (probe 1). Thus, using the Vmat2 mRNA two-probe-approach, the color shift from complete overlap of blue and green color in Ctrl mice to the presence of green-only cells in cKO mice is used to verify Cre-LoxPmediated cKO of the Vmat2 gene.

Baseline locomotion
Spontaneous locomotion and habituation in a novel environment were monitored for 30 min upon placing the mice in Makrolon polycarbonate boxes containing 1.5-cm bedding and a transparent Plexiglas lid. Locomotor behavior of the mice was recorded by the EthovisionXT software (Noldus, RRID:SCR_000441).

Sucrose preference test
Preference to sucrose was assessed in the home cage of the mice. The mice were housed individually in cages containing two drinking bottles. After 48 h of habituation to the experimental set up, they were presented to one bottle of tap water and one of sucrose solution (1%, 3%, and 10%) that were replaced and weighted every 24 h. Each concentration was tested twice and the position of the bottles was alternated to avoid side bias.

Ethanol preference test
Individually housed mice had access to one bottle of tap water and one of alcohol solution (3%, 6%, and 10%) that were replaced and weighted every 24 h. Each concentration of ethanol was tested four times.

Cocaine-induced locomotion
Mice were placed in Makrolon polycarbonate boxes containing 1.5-cm bedding and a transparent Plexiglas lid and their locomotor behavior was recorded 30 min before and 60 min after injection of saline or cocaine (5, 10, and 20 mg/kg, i.p.) on four consecutive days. Locomotor behavior of the mice was recorded by the EthovisionXT software (Noldus, RRID:SCR_000441).

Amphetamine sensitization
Upon habituation, mice received a saline injection (day 1) followed by 4 d of amphetamine injections (days 2-5, 3 mg/kg, i.p.) followed by a last injection on day 17. Locomotion was recorded 30 min before and 1.5 h after injection. Locomotor behavior of the mice was recorded by the EthovisionXT software (Noldus, RRID:SCR_000441).

Conditioned-placed preference (CPP)
An apparatus (Panlab, Harvard Apparatus) consisting of two-main compartments [20 cm (W) ϫ 18 cm (L) ϫ 25 cm (H)] with distinct wall and floor texture patterns and one connecting, transparent compartment (20 ϫ 7 ϫ 25 cm) was used. The CPP procedure was conducted throughout 6 d. Firstly, during the pre-test, the mice were placed in the apparatus and left to freely explore. This session was used to assess initial preferences and to calculate the preference score (see below). During the next four consecutive conditioning days, the mice were constrained in one of the two main compartments and received drug injections (cocaine, 20 mg/kg or amphetamine, 3 mg/kg; i.p.) in the least preferred compartment or saline injections in the opposite one. The conditioning sessions were repeated twice a day [morning (A.M.), afternoon (P.M.)] and the treatment was alternated between days. Thus, the mice received in total four injections of saline and four injections of the drug, counterbalanced between sessions and genotypes. On the test day, the mice were placed again in the apparatus and were let to freely explore. The preference score was calculated by subtracting the time in seconds the animal spent in the drug-paired compartment during pre-test from the time spent in the same compartment during the test (⌬Sec). All sessions lasted 30 min, and the locomotor behavior of the mice was recorded by the EthovisionXT software (Noldus, RRID: SCR_000441).

Imaging, cell counting, and analysis of projection target areas Quantification of FISH
Manual counting of cells expressing mRNAs of interest was performed in two to three mice per probe pair with Th mRNA as reference for outline of the VTA and Th, Dat, Viaat, or Vglut2 mRNA as reference for distinct cell soma. A signal for a particular mRNA was considered as specific for a particular cell when five contiguous fluorescent puncta were present within the outline of the cell soma.

Quantification of immunohistochemistry
Sections of Calb2-Cre and NEX-Cre mice injected with AAV5-EF1a-DIO-ChR2(H134)-eYFP containing the VTA (-3.28 to -3.80 mm from bregma according to Franklin and Paxinos, 2008) were immunostained for eYFP and TH as described above. Z-stacks in four different positions within the VTA, (VTA1-VTA4, of which VTA1 and VTA3 represented medial VTA and VTA2 and VTA4 lateral VTA on two different bregma levels), were acquired using a Zeiss Confocal microscope (LSM 700, 20ϫ magnification). Co-labeling of YFP and TH was identified for each fluorescent channel and counted manually using the Im-ageJ software (RRID:SCR_003070). A minimum of three mice of each genotype was processed and analyzed.

Analysis of projection areas
Fluorescent microscopy (Zeiss Confocal microscope) was used to detect eYFP-positive fibers in sections derived from the whole brain of NEX-Cre, Calb2-Cre, DAT-Cre, and NEX-Cre mice injected into the VTA with AAV5-EF1a-DIO-ChR2(H134)-eYFP. A minimum of two mice of each genotype was analyzed by two persons blind to the genotype of the mice.

Carbon fiber microelectrodes
Carbon fiber working electrodes were fabricated by aspirating 7-m diameter carbon fibers (Cytec Engineered Materials) into borosilicate glass capillaries (1.2 mm O.D., 0.69 mm I.D., Sutter Instrument Co). Capillaries were adjusted (Sutter Instrument, P-97) and sealed with epoxy (EpoTek 301, Epoxy Technology). Electrodes were tested on bath applications of known concentrations of DA. Only electrodes showing good reaction kinetics (current vs time plots, and current vs voltage plots) were used.

FSCV
A Dagan Chem-Clamp potentiostat (Dagan Corporation) and two data acquisition boards (PCI-6221, National Instruments) run by the TH 1.0 CV program (ESA) were used to collect all electrochemical data. Cyclic voltammograms were obtained by applying a triangular wave form potential (Ϫ0.4 to ϩ1.3 V vs Ag/AgCl) repeated every 100 ms at a scan rate of 200 V/s (low pass Bessel filter at 3 kHz). Each cyclic voltammogram was a backgroundsubtracted average of 10 successive cyclic voltammograms taken at the maximum oxidation peak current. All electrodes were allowed to cycle for at least 15 min before recording to stabilize the background current. The recorded current response was converted to DA concentration via in vitro electrode calibration with standard DA solution after each experiment. For optically evoked DA release, photostimulation during FSCV recordings was generated through a 3.4-W 447-nm LED mounted on the microscope oculars and delivered through the objective lens. Photostimulation was controlled via a DigiData 1440A, enabling control over duration and intensity. Illumination intensity typically scaled to 3 mW/mm 2 . Acquired data were analyzed and plotted using MATLAB (RRID:SCR_001622) routines and statistical analysis was performed using Prism 6.0 (GraphPad Software, RRID: SCR_002798)

Place preference upon optogenetic stimulation
The three-compartment apparatus (Panlab, Harvard Apparatus) used in the CPP experiments (above) was also implemented in the optogenetics-driven place preference experiments to address RT-PP upon photostimulation and CR, the association to compartment previously paired with photostimulation. Similar to protocols previously described by others (Root et al., 2014;Qi et al., 2016), the entry of the mouse into one of the two main compartments was paired with intracranial VTA photostimulation (10-ms pulse width, 20 Hz, 10 mW) while the interconnecting compartment was not coupled to light stimulation (neutral) at all. The EthovisionXT tracking software (Noldus, RRID:SCR_000441) was used to monitor behavior and trigger laser stimulation. Behavior was assessed over the course of eight experimental days subdivided into two recording phases with a minimum 3-d rest period in between ("phase 1," days 3-5, and "reversal phase," days 6 -8). On day 1 ("habituation"), the mouse was connected to the optic fiber cord and allowed to acclimatize. On day 2 ("pre-test"), the mouse was placed in the three-compartment apparatus for 15 min to freely explore, while attached to the optic fiber cord but without receiving any photostimulation; the preference for each compartment was evaluated. During 30 min-long recordings on days 3 and 4 ("RT-PP"), entry into the assigned light-paired compartment (non-preferred in pre-test) resulted in blue laser photostimulation delivered as continuous train of pulses (10-ms pulse width, 20 Hz, 10 mW). On day 5 ("CR"), the time spent in each compartment was measured for 15 min with no delivery of photostimulation. In the reversal phase, the protocol was repeated but with stimulation in the opposite compartment compared to phase 1. "High-power" experiments followed the same structure except that the mice received a stimulation of higher power (5-ms pulse width, 20 Hz, 20 mW).
For the Neutral Compartment Preference (NCP) test, a modified version of the test described above was used with the following changes: Entry into either one of the two main compartments was coupled to light stimulation, while only entry the interconnecting compartment had no consequence. The experiment took place on three consecutive days: During the first 2 d (Stim1 and Stim2), the mice received stimulation upon entry in any of the main compartments during 30 min long sessions while the third day was stimulation-free (15 min long session) and used to study the presence of any CRs induced by the experience with the stimulation.

Experimental design and statistical analysis
Regular and repeated measures (RM) two-way ANOVA and unpaired t tests were used to compare mean scores of Ctrl and cKO mice in behavioral tests. To analyze cocaine-induced locomotion during CPP, a mixed-effects model was used. Post hoc comparisons were performed by Sidak's multiple comparison test. Unpaired t test was used to compare mean DA release between ChR2-and eYFP (control)-injected DAT-Cre, Calb2-Cre and NEX-Cre mice for each region where the measurements were performed. Paired t tests were used to compare pre-DNQX and post-DNQX EPSP recordings. Two-way RM ANOVA with day and chamber were used as factors throughout the optogenetic experiments followed by Tukey's post hoc test. When the days of stimulation were averaged, one-way ANOVA was used to unravel the effect of compartment (paired, unpaired, neutral) on time spent and Tukey's multiple comparison test for post hoc analysis. Data are presented as mean Ϯ SEM unless stated otherwise. Data analysis was performed with Prism8 (RRID: SCR_002798). Detailed statistical information is shown in Table 1.

NeuroD6 mRNA is found in a modest population of the medial VTA where it co-localizes extensively with dopaminergic markers and with a glutamatergic marker to minor degree
To address the distribution pattern and neurotransmitter identity of NeuroD6-expressing neurons, doublelabeling FISH was first performed in which NeuroD6 mRNA ( Fig. 1A,C) was compared to tyrosine hydroxylase (Th) mRNA encoding the rate-limiting enzyme (TH) of DA synthesis (Fig. 1B,C). Using the distribution pattern of Th mRNA as reference, DA neurons of the SNc and VTA were identified, including the paranigral (PN), parainterfascicular (PIF), parabrachial pigmented nucleus (PBP), interfascicular nucleus (IF), and rostral linear nucleus (RLi) subareas of the VTA ( Fig. 1A-C). NeuroD6 mRNA was excluded from the SNc, but was detected in scattered VTA neurons. Most NeuroD6 neurons were found within the PN, PIF, and PBP subareas of the VTA, followed by fewer NeuroD6 neurons in the IF and RLi (Fig. 1A,C). Co-detection analysis showed that all neurons detected as positive for NeuroD6 mRNA within the PN, PIF, PBP, IF, and RLi were positive for Th mRNA (Fig. 1C). Quantification verified that 100% of NeuroD6 mRNA-positive cells in the PN/PIF, PBP, IF, and RLi were positive for Th mRNA, while 12% of all Th-expressing neurons within these VTA subareas contained NeuroD6 mRNA (Fig. 1D). To further address the dopaminergic identity of NeuroD6 . Quantification verified that 100% of NeuroD6 VTA neurons were positive for Th (NeuroD6ϩ/Thϩ), and showed that 12% of these NeuroD6ϩ/Thϩ VTA neurons were also positive for Vglut2 mRNA. 12% thus displayed a NeuroD6ϩ/Thϩ/ Vglut2ϩ triple-positive molecular phenotype, while the remaining 88% of NeuroD6/Th neurons were negative for Vglut2 (NeuroD6ϩ/Thϩ/Vglut2-; Fig. 1O). NeuroD6ϩ/ Thϩ/Vglut2ϩ and NeuroD6ϩ/Thϩ/Vglut2-VTA neurons were distributed throughout the VTA with highest density in PN, PIF, and PBP subareas ( Fig. 1M,P).

Conditional ablation of the Vmat2 gene in NeuroD6-Cre VTA neurons, a model for spatially restricted DA deficiency
To analyze the consequences of lost ability for vesicular packaging of DA in NeuroD6 VTA DA neurons, the Slc18a2/Vmat2 gene encoding VMAT2 was targeted using a NeuroD6-Cre (NEX-Cre) transgenic mouse line. By breeding NEX-Cre mice with Vmat2 lox/lox mice, Vmat2 lox/ lox;NEX-Cre-tg (cKO), and littermate control (Ctrl) mice were produced ( Fig. 2A). Upon PCR-based analysis of geno-type, brains from Ctrl and cKO mice were analyzed by ISH to verify loss of full-length Vmat2 mRNA in cKO mice. Due to the scarcity of NeuroD6-positive neurons in the VTA, a Vmat2 mRNA two-probe approach was used to allow detection of gene-targeted neurons. Vmat2 probe 1 was designed to detect all cells positive for Vmat2 mRNA, while Vmat2 probe 2 was designed to bind mRNA derived from exon 2, the exon targeted for recombination by Cre recombinase (Fig. 2B). In the ventral midbrain of control mice, probe 1 (green) and probe 2 (blue) were detected throughout the VTA and SNc areas with complete overlap (Fig. 2C, left panel). In the corresponding area of cKO mice, the majority of cells were positive for both probe 1 and probe 2 with complete overlap (Fig. 2C, right panel). However, throughout the PN, PIF, PBP, and IF VTA subareas, sparse cells showing green color only (probe 1) were detected, thus visualizing Vmat2-gene targeted cells among the mass of VTA DA neurons positive for both Vmat2 probes 1 and 2 ( Fig. 2C, right panel). Having confirmed NEX-Cre-mediated recombination of the floxed Vmat2 gene within scattered neurons of the VTA, other brain areas in which monoaminergic neurons reside were addressed by oligo ISH. Apart from the modest VTA population positive for NeuroD6 mRNA, NeuroD6 mRNA was not detected within any other monoaminergic cell group, identified by Th and Vmat2 mRNA (Extended Data Fig. 2-1). However, as previously reported (Goebbels et al., 2006), NeuroD6 was abundant in several nondopaminergic brain structures, primarily the cerebral cortex and hippocampus (Extended Data Fig. 2-1). In accordance with the lack of NeuroD6 in all monoaminergic cell groups apart from the VTA, Vmat2 probe 1 and probe 2 showed complete overlap in these areas, including locus coeruleus, ventromedial hypothalamus, and nucleus raphe obscurus, while none displayed labeling from probe 1 only (Fig. 2D). These experiments showed that in cKO mice, Vmat2 mRNA was selectively ablated within the VTA. To address whether the targeted deletion of Vmat2 in NeuroD6 neurons of the VTA affected the morphology of the midbrain DA system, distribution patterns of Th mRNA and TH protein were addressed, none of which revealed any gross anatomic difference in the dopaminergic system between Ctrl and cKO mice (  Heightened locomotor response to psychostimulants upon gene-targeting of Vmat2 in NEX-Cre VTA neurons To address whether it is possible to dissociate an explicit behavioral role of DA neurotransmission exerted by NeuroD6 VTA DA neurons from the range of behaviors ascribed to the mDA system, Vmat2 lox/lox;NEX-Cre-tg cKO mice were tested in a battery of tests relevant to the mDA system and compared to Vmat2 lox/lox;NEX-Cre-wt Ctrl mice. To assess body weight, mice were weighed every week from weaning to adulthood. cKO mice were similar to their Ctrl littermates weight-wise (effect of age: F (4158) ϭ 79.8, p Ͻ 0.001; genotype: F (1158) ϭ 4.67 p ϭ 0.032; no age ϫ genotype interaction, no post hoc differences between genotypes; Fig. 3A).

Baseline locomotion
The habituation response to a novel environment, a gross measure of stress and exploratory behavior, was addressed. Both Ctrl and cKO mice showed the same rate of reaching a stable plateau in baseline locomotion (effect of time: F (5160) ϭ 69.5, p Ͻ 0.001; effect of genotype: F (1,32) ϭ 0.00912, p ϭ 0.535; Fig. 3B).

Sucrose and ethanol preference
A sucrose bottle preference test was next performed. Both Ctrl and cKO mice preferred the ascending concentrations of sucrose solutions over water (effect of concentration: F (2,66) ϭ 151, p Ͻ 0.001), but no differences between the genotypes were observed (effect of genotype: F (1,33) ϭ 1.12, p ϭ 0.297; Fig. 3C). The rewarding effect of alcohol was subsequently measured by using increasing concentrations of ethanol (3%, 6%, 10%) presented in a bottle preference test. Again, both Ctrl and cKO mice preferred the presented reward over water (effect of concentration: F (2,52) ϭ 14.2, p Ͻ 0.001), but there was no difference between the genotypes (effect of genotype: F (1,26) ϭ 0.969, p ϭ 0.334). However, post hoc analysis showed that Ctrl mice significantly preferred the 6% and 10% concentrations over the 3% solution ( § § §p Ͻ 0.001 3% vs 6% and 10% ethanol in ctrl mice), while a trend toward significant differences in cKO mice was observed only between the 3% and 10% ethanol solutions (3% vs 10%: p Ͻ 0.072; Fig. 3D).

CPP
To study the reinforcing effects of psychostimulants, a CPP procedure was applied (Fig. 3G). Both Ctrl and cKO mice showed preference for the cocaine-paired or amphetamine-paired compartment over the saline-paired compartment with no significant difference between genotypes (ctrl vs cKO cocaine: p ϭ 0.860, amphetamine p ϭ 0.744; Fig. 3H,J). In addition to preference, locomotion was monitored during the conditioning sessions. cKO mice displayed increased locomotor responses after repeated administration of cocaine compared to Ctrl mice (effect of session; F (3,75) ϭ 4.4, p ϭ 0.006; effect of genotype F (1,25) ϭ 5.2, p ϭ 0.031, no differences in post hoc analysis; Fig. 3I). In contrast, in the CPP paradigm, repeated administration of amphetamine did not induce elevated locomotion in cKO over Ctrl mice (effect of session; F (3,85) ϭ 24.0, p Ͻ 0.001; effect of genotype F (1,30) ϭ 0.0631, p ϭ 0.803; Fig. 3K).

NeuroD6 mRNA co-localizes partly with Calb2 mRNA, but Calb2 mRNA is abundant throughout VTA and SNc
To further characterize the molecular identity of Neu-roD6 VTA neurons, FISH was next used to address the putative overlap between NeuroD6 and Calb2 mRNAs. Distribution patterns of NeuroD6 and Calb2 mRNAs within midbrain DA neurons were recently described without addressing their putative overlap (Viereckel et al., 2016). In contrast to the selective localization of NeuroD6 mRNA within the VTA and its exclusion from the SNc, Calb2 mRNA was abundant in both VTA and SNc (Fig. 4A). The restricted number of NeuroD6 neurons in the VTA showed partial overlap with Calb2 mRNA: 54% of all NeuroD6 VTA neurons were positive for Calb2 mRNA while 20% of Calb2 neurons expressed NeuroD6 mRNA (Fig. 4A). Fur-continued approach for detection of Vmat2 mRNA by ISH. Probe 1 detects exons 6 -15 and probe 2 detects exon 2 of the Vmat2 gene. Exon 2 is floxed in Vmat2 lox/lox mice leading to failure of probe 2-binding to Vmat2 mRNA in cKO neurons. C, Implementation of Vmat2 mRNA two-probe approach in Vmat2 lox/lox;NEX-Cre-wt (Ctrl, left panel) and Vmat2 lox/lox;NEX-Cre-tg (cKO, right panel) brains. Wild-type neurons are positive for both Vmat2 probes, while cKO neurons are only positive for probe 1 due to targeted deletion of exon 2 (detected by probe 2). Probe 1 detected in green and probe 2 detected in blue results in green-blue double-labeling in wild-type cells and green-only labeling in cKO cells. Green arrows point to green-only cells, i.e., VMAT2 cKO cells. D, Vmat2 mRNA two-probe ISH in additional monoaminergic areas. E, TH immunohistochemistry in Ctrl and cKO midbrain and striatum. LC, locus coeruleus; ROB, raphe nucleus obscurus; VMH, ventromedial hypothalamus; VTA, ventral tegmental area; SNc, substantia nigra pars compacta; DStr, dorsal striatum; NAc, nucleus accumbens; OT, olfactory tubercle. TH, Tyrosine hydroxylase; Vmat2/VMAT2, Vesicular monoamine transporter 2; Ctrl, control; cKO, conditional knockout.

Spatially restricted striatal innervation by NeuroD6-Cre and Calb2-Cre VTA neurons
Next, to allow analysis of projections, signaling properties and behavioral regulation of NEX-Cre and Calb2-Cre VTA neurons, optogenetics was implemented. Upon infusion of viral particles carrying a double-floxed DIO-ChR2-eYFP genetic construct encoding both Channelrhodopsin (ChR2) and the enhanced yellow fluorescent protein (eYFP) into the VTA, mice were analyzed in different parameters. DAT-Cre and Vglut2-Cre transgenic mice were used as controls based on their representation of VTA and SNc dopaminergic and glutamatergic neurons, respectively (Stuber et al., 2010;Hnasko et al., 2012;Pascoli et al., 2015;Qi et al., 2016;Yoo et al., 2016). First, Cre-driven expression of the DIO-ChR2-eYFP construct in DAT-Cre, Vglut2-Cre, Calb2-Cre and NEX-Cre mice was continued SEM. Ctrl (N ϭ 14) and cKO (N ϭ 21; ###p Ͻ 0.001 effect of sucrose concentration). D, Ethanol preference expressed as percentage of preference for ethanol solution over tap water Ϯ SEM. Ctrl (N ϭ 14) and cKO (N ϭ 14; ###p Ͻ 0.001 effect of ethanol concentration, § § §p Ͻ 0.001 3% vs 6% and 10% in ctrl mice). E, Cocaine-induced locomotion. Top, Administration schedule. Bottom, Average distance moved 1 h after injection of saline and 5, 10, 20 mg/kg of cocaine; Ctrl (N ϭ 14) and cKO (N ϭ 21) mice. Data expressed as total distance moved during the 1-h recording period Ϯ SEM (###p Ͻ 0.001 effect of session analyzed histologically by comparing YFP with TH immunolabeling (Fig. 5A). In DAT-Cre, Vglut2-Cre, Calb2-Cre, and NEX-Cre mice, YFP fluorescent labeling was identified in the VTA, verifying the activity of each Cre-driver to recombine the floxed optogenetic construct (Fig. 5B-F). YFP co-localized extensively with TH in the VTA. YFP was strongest and most abundant in the VTA of DAT-Cre mice, while Vglut2-Cre, Calb2-Cre, and NEX-Cre mice all showed lower amount of cells positive for YFP (Fig. 5B-F). Next, to reveal target areas, sections throughout the entire brain of all four Cre-driver mouse lines were analyzed and compared. Some target areas were the same for all four Cre-drivers, including the NAcSh and ventral pallidum, while others differed, such as the distribution within the medial and lateral habenula (Table 2). Overall, the density of YFP-positive fibers was substantially lower in NEX-Cre and Calb2-Cre mice than in DAT-Cre and Vglut2-Cre mice. Following analysis of sections throughout the brain, the VTA and striatum were analyzed in more detail. DAT-Cre mice showed strong cellular YFP labeling within all VTA subareas (sparse in RLi) and within the SNc, primarily on the injected side (   (Fig. 5G,H).

Optogenetic stimulation in striatal target areas of NeuroD6 and Calb2 VTA neurons verifies DA release
To address neurotransmitter release, extracellular DA concentration upon optogenetic stimulation was recorded using FSCV in slice preparations. DAT-Cre, NEX-Cre, and Calb2-Cre mice injected with the same DIO-ChR2-eYFP construct as above (Fig. 6A) were analyzed upon photostimulation and subsequent recording within the NAcSh and OT (Fig. 6B). Cre-mice injected with DIO-eYFP were used as controls. DA levels (ϳ1 M) were readily recorded upon photostimulation in both the NAcSh of DIO-ChR2 injected DAT-Cre (0.9699 Ϯ 0.1471 M) and NEX-Cre mice (0.4701 Ϯ 0.08043 M), while a lower signal was obtained in the NAcSh of Calb2-Cre/ChR2 mice (0.01509 Ϯ 0.002845 M; Fig. 6C,D). Upon photostimulation and recording in the OT, lower DA levels (ϳ200 nM) than those measured in the NAcSh were obtained in DAT-Cre/ChR2 mice (0.2129 Ϯ 0.01291 M) while even smaller levels were detected in both Calb2-Cre/ChR2 (0.02097 Ϯ 0.002712 M) and NEX-Cre/ChR2 mice (0.01362 Ϯ 0.002304 M; Fig. 6C

Optogenetic activation of NeuroD6 VTA neurons, but not Calb2 VTA neurons, induces place preference
Finally, in vivo optogenetic stimulation in the VTA of NEX-Cre and Calb2-Cre mice was applied to assess whether this would induce place preference behavior. Again, DAT-Cre and Vglut2-Cre mice were used as references for comparison to Calb2-Cre and NEX-Cre mice. Mice received DIO-ChR2-eYFP or DIO-eYFP (control) injection and implantation of optic fibers above  (Fig. 8A,G), and were analyzed for RT-PP and CR (Fig.  8A).

Discussion
It is well established that the VTA is involved in a range of functions, including behavioral reinforcement, reward, aversion, motivation and incentive salience (Morales and Margolis, 2017). However, an area of active investigation is how the VTA can possess the ability to contribute to all of these diverse functions, some even contrasting. It is now becoming increasingly clear that functional diversity within the mDA system might be matched by molecular and anatomic heterogeneity (Lammel et al., 2011(Lammel et al., , 2012Roeper, 2013;Morales and Margolis, 2017;Poulin et al., 2018). Why is this important? The possibility to determine the exact identity of neurons that contribute to a particular behavior opens up entirely new perspectives in the opportunity to to selectively target only those neurons that contribute to clinical symptoms without causing sideeffects by affecting adjacent neuronal populations. In this study, we used Cre-driven mouse genetics and optogenetics to begin to disentangle the contribution of the newly described NeuroD6 VTA subtype (Viereckel et al., 2016;Khan et al., 2017;Kramer et al., 2018) in rewardrelated behaviors commonly ascribed to the VTA DA system. The main finding of our study is that despite their restricted number, NeuroD6 VTA neurons contribute to psychostimulant-induced hyperlocomotion and that their activation induces place preference behavior.

NeuroD6 VTA neurons represent a modest neuronal population within the VTA with molecular capacity for dopaminergic and glutamatergic neurotransmission
In the current study, we showed that NeuroD6 VTA neurons constitute a modest proportion (circa 12%) of all VTA neurons expressing the gene encoding TH within the PN, PIF, PBP, IF, and RLi subareas. Within these VTA subareas, all NeuroD6-positive neurons were positive for both Th and Dat mRNAs, markers of dopaminergic neurons. In addition, while no or very few NeuroD6 neurons were positive for Viaat mRNA, a marker of GABAergic neurons, 12% of the NeuroD6/Th double-positive neurons within the VTA were positive for Vglut2 mRNA, suggesting a capacity for dual dopaminergic/glutamatergic neurotransmission. Indeed, DA/glutamate co-release has in several studies been shown as a property of certain mDA neurons where it has been proposed to play a role in  reward-related behavior reinforced by DA (for recent review, see Trudeau and El Mestikawy, 2018). The identification of co-labeling of NeuroD6 mRNA with Th, Dat and Vglut2 mRNAs within distinct neurons was partly in accordance with our analysis of a NEX-Cre transgenic mouse line, implemented here to achieve manipulation of the NeuroD6 VTA neurons, which identified substantial co-localization between NEX-Cre-driven reporter gene expression (YFP) and TH immunofluorescence. However, lack of TH/YFP co-localization was also identified. The findings showing that the majority of NeuroD6 VTA neurons expressed DA markers were in accordance with our electrophysiological data in which optogenetic VTA stimulation of NEX-Cre neurons enabled the identification of DA release, as further discussed below. Further, optogenetic stimulation also gave rise to EPSCs of glutamatergic nature, while no GABAergic currents were detected, in agreement with the co-localization of NeuroD6 mRNA with Vglut2 mRNA but lack of significant co-localization with Viaat mRNA.
In the context of transgenic mice, it is noteworthy that our result showing non-complete overlap between NEX-Cre-driven reporter gene expression and TH, which contrasts the parallel finding that all VTA neurons positive for endogenous NeuroD6 mRNA also label for Th mRNA, are in accordance with a recent study in which a substantial number of non-dopaminergic NEX-Cre VTA neurons were identified (Kramer et al., 2018). Collectively, these findings propose that interpretation of VTA-data originating from the current NEX-Cre mouse line should be considered with awareness of complex downstream neurocircuitry. Further, as extensively discussed in the literature, regulatory promoters implemented experimentally to drive Cre expression may give rise to transient and/or ectopic Cre activity that fails to mimic endogenous gene expression due to gene regulatory events, not least during developmental phases. Indeed, patterns of ectopic Cre activity have been described for other transgenic mouse lines, including DAT-Cre and TH-Cre transgenic mouse lines commonly implemented for the study of DA neurons (Lindeberg et al., 2004;Stamatakis et al., 2013;Lammel et al., 2015;Nordenankar et al., 2015;Pupe and Wallén-Mackenzie, 2015;Stuber et al., 2015;Morales and Margolis, 2017). While the current NEX-Cre transgenic line has been thoroughly validated recently for the study of VTA neurons (Khan et al., 2017;Kramer et al., 2018), to direct selectivity to VTA DA neurons, we here implemented a conditional genetic approach to specifically abrogate vesicular packaging of DA in NEX-Cre neurons. Further, we used optogenetically driven neuronal activation to study effects upon direct stimulation of NEX-Cre VTA neurons.

Targeting of the Vmat2 gene in NEX-Cre VTA DA neurons allowed identification of a role in psychostimulant-mediated response
To enable the study of how reward-related behaviors classically associated with the mDA system would be affected if the NEX-Cre DA neuron subtype lost its ability for dopaminergic function, a conditional gene-targeting approach was implemented in which VMAT2 was ablated specifically from NEX-Cre neurons. Since we could show that NeuroD6 and Vmat2 mRNAs only co-localized within the VTA, no other monoaminergic population should suffer from loss of VMAT2 by this approach. Indeed, the results confirmed that Vmat2 mRNA was selectively knocked out within the VTA, while all other monoaminergic neurons maintained normal Vmat2 mRNA. Thus, the Vmat2 lox/lox;NEX-Cre mouse line forms a new mouse model of DA-release deficiency from a restricted group of VTA DA neurons characterized by NeuroD6 promoter activity. Based on the importance of mDA system in processing natural and drug rewards (Kalivas et al., 1992;Di Chiara and Bassareo, 2007;Ikemoto, 2007;Baik, 2013;Robinson and Berridge, 1993), we addressed the behavioral responses of Vmat2 lox/lox;NEX-Cre-tg cKO mice and Vmat2 lox/lox;NEX-Cre-wt control mice to sugar, ethanol and the psychostimulants amphetamine and cocaine. cKO mice displayed higher locomotor activation on repeated administration of psychostimulants than control mice. In contrast, sugar preference and CPP to cocaine and amphetamine were similar between cKO and control mice, and both genotype groups showed a preference for increasing dose of ethanol, albeit in different patterns.
While acute administration of cocaine failed to cause differences in locomotor responses between cKO and control mice, repeated administration caused exaggerated locomotor behavior in cKO mice when measured in the CPP paradigm. In contrast, with repeated amphetamine injections, the locomotor response was elevated above control levels in the open field, but not in the CPP.
The tests implemented were designed to study different behavioral parameters, and results obtained in different setups and by different drugs are therefore not directly comparable. What may seem as apparent discrepancies might be related to several different properties. Firstly, the size and properties of the test environment were substantially different between setups. The open field test took place in an environment that resembled the home cage. Locomotion was recorded during the conditioning phase when the mice were confined to a much smaller compartment with specific patterns and no bedding. Secondly, the injection regime differed between tests. In the open field, mice received acute injections of cocaine or were sensitized to amphetamine by receiving daily injections continued percentage of time spent in each compartment during days 3, 4, 6, and 7 Ϯ SEM (bar graphs; right; ‫ء‬p Ͻ 0.05, ‫‪p‬ءءء‬ Ͻ 0.001 vs light-paired compartment; #p Ͻ 0.05, ##p Ͻ 0.01, ###p Ͻ 0.001 vs unpaired compartment). DAT-Cre N ϭ 10; Vglut2-Cre N ϭ 7; Calb2-Cre N ϭ 7; NEX-Cre N ϭ 5. F, High-power stimulation of bilaterally injected NEX-Cre mice (N ϭ 4). G, Schematic illustration of optical fiber placement in mice analyzed in RT-PP analysis. NS, non-significant. DAT, Dopamine transporter; Calb2, Calbindin 2 (Calretinin); NEX, NeuroD6; Vglut2; Vesicular glutamate transporter 2; ChR2; Channelrhodopsin 2; eYFP, enhanced Yellow fluorescent protein.
after a 30-min habituation period. In contrast, in the CPP experiment, the mice received in total four injections of the drug in two non-consecutive days without any previous habituation period. Finally, the recording period was shorter in the CPP compared to the open field (30 min vs 1.5 h), a parameter that could mask the long-lasting effects of amphetamine on locomotion. Further, the observation of heightened, rather than reduced, psychostimulant-induced locomotion might seem counter-intuitive: Loss of VMAT2 should lead to decreased packaging and release of DA which might be expected to cause reduced locomotion compared to control levels. However, the results obtained from our spatially selective cKO mice are similar to the heightened amphetamineinduced hyperlocomotion observed in a study of mice heterozygous for Vmat2 in all DAT-Cre neurons (Isingrini et al., 2016). Thus, lowering the level of VMAT2 throughout all DAT-Cre neurons or ablating it within the NEX-Cre VTA DA population give rise to similar behavioral consequences. Further analyses focused around VMAT2 in psychomotor behavior will be necessary to pin-point this matter, however, developmental adaptations, a common feature of KO strategies induced during embryonal development, may underlie the heightened locomotor response.

Striatal optogenetic stimulation in NEX-Cre mice induced DA release and glutamatergic EPSCs
Complementary to the cKO approach, we used optogenetics-based experiments in which the NEX-Cre VTA population could be directly stimulated. This type of manipulation provides high spatial and temporal resolution (Deisseroth, 2015) and thus has the advantage of enabling selective stimulation of Cre-driven neurons in real time with the benefit of directly pin-pointing the role of molecularly defined neurons in measurable behavior. By analysis of optogenetic reporter gene (eYFP) expression upon injection into the VTA of NEX-Cre mice, we showed that NeuroD6 VTA neurons projected mainly to the NAcSh of the striatal complex, with substantially lower density than observed upon similar injection in DAT-Cre and Vglut2-Cre mice used here as controls (Stuber et al., 2010;Hnasko et al., 2012;Pascoli et al., 2015;Qi et al., 2016;Yoo et al., 2016). NEX-Cre VTA projections also reached several additional areas, but with even lower density than seen in the NAcSh, including the OT, medial habenula and ventral pallidum. In accordance with the co-localization of eYFP with TH immunoreactivity, we could verify that NEX-Cre VTA neurons released DA in both the NAcSh and OT upon striatal optogenetic stimulation. Although the levels were lower than those observed upon similar stimulation of DAT-Cre-positive VTA neurons, they were significantly higher than those observed in control experiments, demonstrating that the NEX-Cre VTA population indeed releases measurable amounts of DA in their target areas. To investigate whether the TH-negative cellular population, present most profoundly in the medial VTA, was of glutamatergic or GABAergic nature, patch-clamp electrophysiology was performed which showed that optogenetic stimulation of NEX-Cre terminals induced EPSCs, but not IPSCs, in NAcSh, thus verifying glutamatergic neurotransmission. While glutamatergic postsynaptic currents were evidently a result of the optogenetic stimulation of NEX-Cre VTA neurons, it remains to be established if the rare endogenous NeuroD6ϩ/Thϩ/NeuroD6ϩ triple-positive neurons observed in our histologic analysis are sufficiently potent to drive a similar postsynaptic response in the natural situation, that is, upon excitation of the NeuroD6 VTA neurons in a non-transgenic context. Finally, the current setup did not allow us to conclude if the EPSCs were of monosynaptic or polysynaptic nature. The short onset of EPSCs was suggestive of monosynaptic transmission, however, electrophysiological approaches combined with pharmacological agents will be necessary to fully define the signaling properties.

Optogenetic stimulation of NEX-Cre VTA neurons reveals a role in place preference behavior
Optogenetic stimulation of the mDA system of TH-Cre and DAT-Cre mice has been demonstrated to potently induce DA release and real time place preference (Tsai et al., 2009;Stuber et al., 2010;Yoo et al., 2016). The same type of activation of VTA in Vglut2-Cre mice has been described to cause postsynaptic glutamatergic currents and to induce either place preference or place avoidance, depending on stimulation parameters (Hnasko et al., 2012;Wang et al., 2015;Qi et al., 2016;Yoo et al., 2016). Using DAT-Cre and Vglut2-Cre mice as references, we could show here that optogenetic stimulation within the VTA of NEX-Cre mice induced a significant preference for the light-paired compartment. The magnitude of the preference observed was, however, smaller in NEX-Cre than in DAT-Cre mice. This difference is likely related to the substantially smaller population of VTA neurons activated upon photostimulation in the NEX-Cre compared to DAT-Cre VTA and the different projection patterns of these neuronal populations. This is supported by the analysis of YFP-positive fibers, which differ substantially between DAT-Cre and NEX-Cre mice. VTA-injection of ChR2-YFP in DAT-Cre mice results in strong YFPfluorescence in all innervation areas ascribed to the mDA system. In contrast, the same injection into the VTA of NEX-Cre mice results in substantially lower YFP-derived fluorescence in the VTA and sparse fluorescence in target areas.
Despite smaller magnitude, the ability of NEX-Cre VTA neurons to induce real time place preference is an important finding as it demonstrates the possibility of identifying spatially restricted groups of VTA neurons that are sufficient to induce a measurable behavior. Further arguing for the importance of this result, the optogenetically induced preference behavior displayed by NEX-Cre mice was strengthened by viral injections in bilateral, rather than unilateral, manner as well as by increased laser power. The results of these experimental manipulations suggest that the enhanced recruitment of NEX-Cre neurons strengthened the behavioral output. While additional studies will be required to completely disentangle the behavioral role of NeuroD6 VTA neurons, the current optogenetics-based setup already enabled us to demonstrate that VTA activation in NEX-Cre mice could induce place preference in real-time, but that it failed to result in CR, defined as significant place preference even in absence of actual optogenetic stimulation. This contrasts the strong CR observed in the DAT-Cre mice, and hence, activation of VTA populations in NEX-Cre mice and DAT-Cre mice differ in more than one parameter: Magnitude in real time place preference and presence of a detectable CR. In contrast to the preference behavior displayed by NEX-Cre and DAT-Cre mice, optogenetic stimulation of VTA Vglut2-Cre neurons led to real time place avoidance defined here as reduced time spent in the stimulation-paired compartment. This result is consistent with a recent study which found that real time avoidance coincided with a frequencydependent increase in entries to the light-paired compartment and robust self-stimulation in an operant task . In contrast, another study found that photostimulation of Vglut2-Cre neurons in VTA induced modest real time place preference and self-stimulation (Wang et al., 2015). These data show that the behavioral effects of VTA glutamate neuron stimulation are sensitive to the task, including the design of the apparatus and stimulus parameters. In this context, it is noteworthy that VTA neurons of the NEX-Cre transgenic mouse line, with their mixture of dopaminergic and glutamatergic signaling properties, might have shown lower level of place preference than DAT-Cre mice not only due to the smaller number of neurons and sparser projections, but also as their activation might have caused a glutamate-mediated avoidance behavior that counterbalanced the behavioral preference for light stimulation.

NeuroD6 and Calb2 mRNAs show partial overlap, but NEX-Cre and Calb2-Cre VTA neurons have distinct projections and role in behavior
Parallel to the focus on NeuroD6 VTA neurons in neurocircuitry and behavioral regulation, our histological analysis enabled us to identify a degree of co-localization between NeuroD6 and Calb2 mRNAs. While NeuroD6 mRNA was uniquely found in the VTA and excluded from the SNc, Calb2 mRNA was found distributed throughout these dopaminergic areas. However, histological analysis showed a degree of co-localization between NeuroD6 and Calb2 mRNAs, a finding which adds to the recent molecular description of NeuroD6 as co-localized with gastrinreleasing peptide (GRP) and additional markers (Khan et al., 2017;Kramer et al., 2018;Poulin et al., 2018). Beyond the partial co-localization of NeuroD6 and Calb2 mRNAs, the results demonstrate that Calb2 VTA neurons constitute a substantially larger proportion within the mDA population, show considerable expression of the gene encoding VIAAT, and are present in the SNc, an area devoid of NeuroD6 neurons. Our neurophysiological circuitry analyses of Calb2-Cre mice showed that Calb2-Cre VTA neurons belong to the category of VTA/SNc neurons that projects to the OT where their stimulation resulted in DA release and glutamatergic postsynaptic currents. While it was recently described that activation of dopaminergic fibers from VTA to the medial OT can induce place preference in DAT-Cre mice (Zhang et al., 2017), a similar response was not observed here on Calb2-Cre VTA stimulation. These differences might be explained by the difference in density of the innervation patterns in the OT between the DAT-Cre and Calb2-Cre mice. The difference in preference behavior between NeuroD6-Cre and Calb2-Cre mice shows that distinct VTA neurocircuitry is crucial for the behavioral output.
Unraveling the behavioral roles of NeuroD6 VTA neurons stands to benefit current decoding of VTArelated disorders The behavioral complexity mediated by the VTA is implicated in a range of neuropsychiatric conditions including substance use disorder, schizophrenia, and ADHD for which clinical interventions based on increasing, decreasing, stabilizing or modulating the mDA system are commonly prescribed. In addition, since VTA DA neurons are less susceptible to degeneration in PD than SNc DA neurons, molecular differences are intensively searched for. GRP, in several studies identified as a marker for VTA DA neurons (Chung et al., 2005;Greene et al., 2005;La Manno et al., 2016;Viereckel et al., 2016) was recently shown to co-localize with NeuroD6 (Kramer et al., 2018). Several lines of evidence suggest that a discrete Neu-roD6/GRP VTA subtype should be of specific interest. Overexpression of the gene encoding GRP increased the survival rate of cultured DA neurons in a parkinsonian experimental model (Chung et al., 2005) and GRP-positive mDA neurons remain in biopsies from deceased PD patients (Viereckel et al., 2016). Further, NeuroD6 increases neuronal survival in a toxin model of PD (Kramer et al., 2018). The NeuroD6/GRP VTA subtype might thereby possess resistance to PD. Our current results show that, despite their modest representation within the VTA, NeuroD6-expressing VTA neurons are implicated in distinct aspects of reward-related behavior. Their resistance to PD may thus contribute to the cause of behavioral dysfunction observed in the non-motor symptom domain of PD, including treatment-induced complications that resemble aspects of neuropsychiatric diseases, such as behavioral addictions (Cenci et al., 2015).
Current molecular profiling of DA neuron subtypes should prove valuable for prospects of selective treatment in conditions related to VTA dysfunction. Of essence to achieve such selectivity is the systematic decoding of the explicit behavioral roles mediated by distinct VTA neurons. In this study, we initiated such analysis and now propose that NeuroD6 VTA neurons are of particular interest for further analysis of motivated and addictive behavior as they are here implicated in reward-related behavior measured as real time place preference and as their controlled dysregulation alters the responsiveness to psychostimulants. Our findings should prove useful for future investigations aimed at advancing the knowledge of VTA neurocircuitry in healthy conditions and in neuropsychiatric illness implicating the VTA.