Catecholaminergic innervation and D2-like dopamine receptor-mediated modulation of brainstem nucleus incertus neurons in the rat

Nucleus incertus (NI) is a brainstem structure involved in the control of arousal, stress responses and locomotor activity. It was reported recently that NI neurons express the dopamine type 2 (D2) receptor that belongs to the D2-like receptor (D2R) family, and that D2R activation in the NI decreased locomotor activity. In this study, using multiplex in situ hybridization, we observed that GABAergic and glutamatergic NI neurons express D2 receptor mRNA, and that D2 receptor mRNA-positive neurons belong to partially overlapping relaxin-3- and cholecystokinin-positive NI neuronal populations. Our immunohistochemical and viral-based retrograde tract-tracing studies revealed a dense innervation of the NI area by fibers containing the catecholaminergic biosynthesis enzymes, tyrosine hydroxylase (TH) and dopamine β-hydroxylase (DBH), and indicated the major sources of the catecholaminergic innervation of the NI as the Darkschewitsch, raphe and hypothalamic A13 nuclei. Furthermore, using whole-cell patch clamp recordings, we demonstrated that D2R activation by quinpirole produced excitatory and inhibitory influences on neuronal activity in the NI, and that both effects were postsynaptic in nature. Moreover, the observed effects were cell-type specific, as type I NI neurons were either excited or inhibited, whereas type II NI neurons were mainly excited by D2R activation. Our results reveal that rat NI receives a strong catecholaminergic innervation and suggest that catecholamines acting within the NI are involved in the control of diverse processes, including locomotor activity, social interaction and nociceptive signaling. Our data also strengthen the hypothesis that the NI acts as a hub integrating arousal-related neuronal information.


Introduction
Maintaining an appropriate level of arousal and locomotor activity is essential for successful functioning within a complex and changing environment, and adequate responding to external stimuli. One of the brain areas involved in the control of these processes is the nucleus incertus (NI), a brainstem structure localized bilaterally beneath the fourth cerebral ventricle in the human (Streeter, 1903), macaque (Ma et al., 2009), rat (Goto et al., 2001;Olucha-Bordonau et al., 2003) and mouse (Smith et al., 2010); and also described in fish (Donizetti et al., 2008).
Indeed, functional studies indicate that NI neuronal populations control arousal, locomotion, anxiety, the stress response and food intake (Blasiak et al., 2017;Lu et al., 2020;Ma et al., 2013Ma et al., , 2017aSzőnyi et al., 2019). Specifically, chemo-and opto-genetic activation of NI neurons promotes locomotor activity, arousal levels, and hippocampal theta power (Lu et al., 2020;Ma et al., 2017a;Szőnyi et al., 2019), whereas optogenetic inhibition of NI NMB neurons lead to deceleration of locomotion speed and a decrease in arousal (Lu et al., 2020). In female rats, depletion of RLN3 levels in the NI caused increased anxiety-like behavior in the open field, along with a decrease in body weight and an imbalance in food intake (de Á vila et al., 2020), and infusion of an antagonist of the cognate relaxin-3 receptor, RXFP3, into the paraventricular hypothalamic nucleus prevented stress-induced, binge-eating behavior (Kania et al., 2020).
NI-controlled functions are closely related to dopaminergic signaling, which is critically involved in the control of locomotor activity, anxiety, motivation and arousal levels (Klein et al., 2019). Dopamine is known to exert its actions through two G-protein-coupled receptor families: D1-like receptors (D1R), comprising D1 and D5 receptors, and D2-like receptors (D2R), comprising D2, D3 and D4 receptors (Beaulieu et al., 2015). Notably, dopamine is not the only endogenous ligand for these receptors, as the catecholamines, noradrenaline and adrenaline, act as D2R agonists (Lanau et al., 1997;Sánchez-Soto et al., 2016). A recent study reported that D2, but not D3, receptors are present in the NI in RLN3-and CRHR1-expressing neurons, and D2 receptor activation in the NI leads to hypo-locomotion (Kumar et al., 2015). However, the source of the catecholaminergic innervation of NI remains unknown.
The most common effect of D2R stimulation observed is inhibition of neuronal activity resulting from an increase in potassium conductance and subsequent membrane hyperpolarization (Missale et al., 1998), but D2R stimulation can also exert an excitatory effect, via activation of nonselective cation currents (Aman et al., 2007;Haj-Dahmane, 2001). However, it remains unclear, what influence D2R activation has on NI neuronal activity.
In these studies, we performed multiplex fluorescent in situ hybridization (RNAscope) to verify the neurochemical profile of NI neurons expressing D2 receptor mRNA. With the use of viral vector-based, retrograde neural tract-tracing combined with immunofluorescent staining, we identified the source of the catecholaminergic innervation of the NI in multiple regions, including hypothalamic, midbrain and brainstem structures. Finally, we performed single-cell, patch-clamp recordings of NI neurons to determine the physiological effect of D2R activation on their activity.

Ethical approval and animals
Experiments were conducted in accordance with the EU Directive 2010/63/EU on the protection of animals used for scientific purposes, the Polish Act on the Protection of Animals Used for Scientific or Educational Purposes of January 15, 2015 and approved by the 2nd Local Institutional Animal Care and Use Committee (Krakow, Poland). All efforts were made to minimize stress prior to experimentation and the number of rats used.
Experiments were conducted using male, Sprague-Dawley rats (Institute of Zoology and Biomedical Research, Jagiellonian University, Krakow, Poland), which were kept in plastic cages lined with woodshaving bedding, under constant temperature conditions (21 ± 2 • C), maintained on a 12-12 light-dark cycle with ad libitum access to fresh water and standard laboratory rodent chow. Rats were separated from dams at four weeks old and kept in same-sex cages (up to 7 rats per cage), until use in experiments. For in situ hybridization, immunostaining and neural tract-tracing studies, 10-12 week-old rats were used, and for patch-clamp experiments, 4-6 week-old rats were used. Overall, 124 rats were used in these studies.
Images for rounds 1 and 2 were acquired and processed using an Axio Imager M2 fluorescence microscope (Zeiss, Gottingen, Germany) with an automatic z-stage and Axiocam 503 mono camera (Zeiss), and subsequently with the following software: Zen (3.1 blue edition and 3.0 SR black edition, Zeiss), CorelDraw 2020 (Corel Corporation, Ottawa, Canada), ImageJ (Schneider et al., 2012) and HiPlex Image Registration Software v1.0 (ACD). The borders of the NI were delineated on the basis of the presence of RLN3 mRNA-expressing neurons and a rat brain atlas (Paxinos and Watson, 2007). All T1-T6 mRNA-expressing cells in the NI area from one brain hemisphere per slice (right and left interchangeably), were counted semi-automatically with an ImageJ Cell Counter plugin. Neurons were identified by the presence of an explicit cell-like distribution of fluorescent mRNA dots and/or a nucleus stained with DAPI. A cell was considered as expressing a specific mRNA when at least two unambiguous dots of specific fluorescence were present within its boundary. All counted neurons were categorized by type, depending on mRNA species co-expression (juxtaposition within a cell).

Immunostaining of TH and DBH fibers in the nucleus incertus
Rats (n = 3) were anesthetized with an intraperitoneal injection of pentobarbital (240 mg/kg) and sacrificed by transcardial perfusion with 250 ml of PBS followed by 250 ml of 4% formaldehyde solution (freshly made from paraformaldehyde powder). Brains were then fixed in 4% formaldehyde at 4 • C overnight, and cut into 50 μm coronal sections, using a Leica VT 1000S vibrating microtome (Leica Instruments, Heidelberg, Germany). Every fourth section containing NI underwent an immunostaining procedure: blocking and permeabilization for 1 h at room temperature (10% normal donkey serum (NDS), 0.3% Triton X-100 in PBS), incubation with primary antibodies for 72 h at 4 • C (one set of sections using rabbit anti-TH (1:100) and another set using rabbit anti-dopamine β-hydroxylase (DBH) (1:4000), 2% NDS, 0.3% Triton X-100 in PBS), followed by incubation with secondary antibodies overnight at 4 • C (donkey anti-rabbit Cy3 (1:400), 2% NDS in PBS). Incubation with primary and secondary antibodies was followed by several washes in PBS. Fluoroshield™ with DAPI was used for section mounting. TH-and DBH-immunoreactive fibers in the NI were imaged using a confocal microscope (LSM710 on Axio Observer Z1, Zeiss) with an EC Plan-Neofluar 20 × /0.50 M27 objective.

Retrograde neural-tract tracing with viral vectors
Rats (n = 4) were anesthetized with an intraperitoneal injection of ketamine and xylazine (100 mg/kg ketamine + 10 mg/kg xylazine) and with additional doses of ketamine as required (33 mg/kg), and placed in a stereotaxic apparatus (SAS-4100; ASI Instruments, Warren, USA). Retro-AAV2-hSyn-mCherry viral vector was injected into the NI bilaterally (stereotaxic coordinates [mm]: AP 3.7 from lambda, ML ± 0.15, DV 7.2, with the rat's head angled forward by 15 • , 100 nl into each hemisphere), using glass microinjection pipettes (50 μm tip diameter), pulled from borosilicate glass capillaries on a vertical puller (Narishige, Tokyo, Japan) and connected to a 1 μl Hamilton syringe (Hamilton, Bonaduz, Switzerland). In order to improve subsequent immunostaining, after 7 days of recovery, rats were anesthetized as described and a colchicine injection was made into the lateral ventricle ([mm]: AP -0.7 from bregma, ML 1.8, DV -4; 5 μl of 20 mg/ml solution into one hemisphere). After 24 h, rats were anesthetized with an intraperitoneal injection of pentobarbital (240 mg/kg) and sacrificed by transcardial perfusion as described. After post-fixation, brains were cut into 50 μm coronal sections using a Leica VT 1000S vibrating microtome (Leica Instruments). Sections containing NI were mounted onto glass slides, coverslipped with Fluoroshield™ and injection sites were imaged using an Axio Imager M2 fluorescence microscope with an A-Plan 10 × /0.25 objective (Zeiss). Reconstructions of the injection sites were made using CorelDRAW software (Corel Corporation), according to the rat brain atlas (Paxinos and Watson, 2007). Every fourth section, beginning at the coronal level of the medial septum through to the NI (i.e., from bregma +1.3 mm to − 10.3 mm) underwent immunostaining. Free-floating sections were blocked and permeabilized for 1 h at room temperature (10% NDS, 0.3% Triton X-100 in PBS, respectively), incubated with primary antibodies solution for 72 h at 4 • C (mouse anti-TH (1:250), rabbit anti-mCherry (1:1000), 2% NDS, 0.3% Triton X-100 in PBS) and subsequently with secondary antibody solutions overnight at 4 • C (donkey anti-mouse Alexa 647-conjugated antibody (1:400), donkey anti-rabbit Cy3-conjugated antibody (1:400), 2% NDS in PBS). Incubations with primary and secondary antibodies were followed by several washes in PBS. After mounting with Fluoroshield™, sections were examined and imaged using an Axio Imager M2 fluorescence microscope with an A-Plan 10 × /0.25 objective (Zeiss). Tyrosine hydroxylase (TH) and mCherry immunoreactive (-ir) neurons were counted using ZEN 2.1 software (Zeiss) and multiplied by 4 to estimate the number of TH-ir neurons innervating the NI.
NI neurons were localized and approached using an Examiner D1 microscope (Zeiss) equipped with video-enhanced infrared differential interference contrast. Cell-attached and whole-cell configurations were obtained using a negative pressure delivered by mouth suction. SEC 05LX amplifiers (NPI, Tamm, Germany), Micro 1401 mk II converters (CED, Cambridge Electronic Design, Cambridge, UK) and Signal and Spike2 software (CED) were used for signal recording and data acquisition. Recorded signal was lowpass filtered at 3 kHz and digitized at 20 kHz. All drugs were applied via a perfusion system. The activity of one neuron per slice was recorded.

Electrophysiological data analysis
Only neurons with stable input resistance throughout the recording, assessed on the basis of the voltage or current responses to hyperpolarizing current or voltage steps (applied every 60 s), were included in the final analysis. The change in the recorded whole-cell current or voltage in response to the drug application was considered significant if it differed from the baseline by more than three standard deviations (SD). Electrophysiological data were analyzed using custom Spike2 and MATLAB (MathWorks Inc., Natick, MA, USA) scripts. To verify the effect of quinpirole on recorded synaptic activity, 200 s epochs of the baseline recoding and 200 s epochs after quinpirole application, were analyzed using Mini Analysis software (Synaptosoft Inc., Fort Lee, NJ, USA). Events were manually detected to measure frequency, amplitude and rise time of postsynaptic currents and the decay time constant of averaged current trace. Analysis of the influence of quinpirole on synaptic activity was performed on all recorded neurons, regardless of whole-cell current change in response to D2R agonist. Statistical analysis was performed using GraphPad Prism v6.00 for Windows (GraphPad Software Inc., La Jolla, CA, USA). All data underwent a test for normality of distribution (Shapiro-Wilk normality test) and outlier detection (ROUT method, Q = 1%) and outliers were eliminated from the analysis. Differences were considered statistically significant at p < 0.05. All tests were two-tailed, and tests used (paired and unpaired t-tests, Mann-Whitney test and Wilcoxon test) are stated in the Results and Figures. Values are provided as mean ± SD, where data was normally distributed or as median ± interquartile ranges, if otherwise.

Post-recording immunostaining
After recordings, slices underwent immunofluorescent staining in order to verify the neurochemical content of examined neurons. Slices were fixed overnight with 4% formaldehyde at 4 • C. Fixed, free-floating sections were blocked and permeabilized with 10% NDS and 0.6% Triton X-100 in PBS, respectively, at 4 • C overnight or for 3 h at room temperature. Subsequently, after washing in PBS, slices were incubated with mouse anti-RLN3 (1:15), ExtrAvidin-Cy3 (1:200), 2% NDS and 0.3% Triton X-100 in PBS for 48-72 h at 4 • C and, after several washing steps (in PBS), with secondary antibody solution: anti-mouse Alexa 647 (1:400) and 2% NDS in PBS at 4 • C overnight. Slices were mounted onto glass slides, coverslipped with Fluoroshield™ and imaged using a fluorescence microscope (Axio Imager M2, Zeiss, with an A-Plan 10 × /0.25 objective or EC-Plan-Neofluar 20 × /0.25 objective) to assess the presence of RLN3 immunoreactivity within recorded cells. However, a lack of RLN3 immunoreactivity was not used as a prerequisite for assigning an NI neuron as non-RLN3, due to the possible dilution of antigen by the intrapipette solution during patch-clamp recording.

Different neurochemical populations of nucleus incertus neurons express D2 receptor mRNA
Brain sections containing NI were subjected to RNAscope™ HiPlex in situ hybridization with probes for RLN3, CCK, and vesicular GABA (vGAT1) and glutamate (vGlut2) transporter mRNA, as well as D2 receptor mRNA, to investigate the expression and distribution of different mRNA species in NI neurons, with a focus on D2 receptor mRNAcontaining populations.
The mean number of NI cells per section, expressing at least one of the mRNA species tested was 1178, of which 422 (36%) were D2 receptor mRNA-positive. Cell counting established that 59% of counted NI neurons expressed vGAT1 mRNA and 48% vGlut2 mRNA. Among them, a small group exhibited a 'vGAT1 + vGlut2 mRNA' phenotype (7% of all cells). Similar numbers of vGAT1 and vGlut2 mRNA-expressing cells, as well as vGAT1/vGlut2 mRNA-expressing cells, were also D2 receptor mRNA-positive (52%, 56% and 8% of total D2 receptor mRNAexpressing cells, respectively).
vGAT1-and vGlut2 mRNA-expressing neurons were found to express RLN3 and CCK mRNA, with 21% of all counted cells vGAT1/RLN3 mRNA-positive; including a small group of vGAT1/vGlut2/RLN3 mRNA-expressing cells. RLN3 mRNA-expressing neurons always coexpressed vGAT1 mRNA, and vGlut2/RLN3 only mRNA-positive neurons were not identified. CCK mRNA-positive cells accounted for 11% of total counted cells and comprised: vGAT1/CCK, vGAT1/vGlut2/CCK and very rarely CCK only mRNA-expressing neurons. Interestingly, a small group (4%) of the total NI neurons counted expressed RLN3 and CCK mRNA, with vGAT1/RLN3/CCK mRNA and rare vGAT1/vGlut2/ RLN3/CCK mRNA-positive cells observed. Importantly, all identified cell types included subpopulations that expressed D2 receptor mRNA, except the CCK mRNA only and vGAT1/vGlut2/RLN3/CCK mRNApositive cells, in frequencies roughly proportional to the general abundance of each cell type within the NI. Scarce D2 receptor mRNA only expressing neurons were also detected. For a detailed summary see Tables 1 and 2, and Fig. 1.
Notably, different neurochemical NI cell types were differentially distributed throughout the structure, as reported . vGlut2 mRNA-expressing neurons were mostly located laterally, while RLN3 and CCK mRNA-positive cells were more medial, and vGAT1 and D2 receptor mRNA-expressing neurons were evenly distributed.

Catecholaminergic fibers densely innervate the nucleus incertus
Immunostaining for TH, the rate-limiting enzyme in catecholamine biosynthesis, and DBH, the synthetic enzyme for noradrenaline, revealed the presence of dense TH-ir and DBH-ir fibers within the NI region ( Fig. 2A). Both TH-ir and DBH-ir fibers were observed in close apposition to RLN3-ir neurons, suggesting the presence of synaptic connections between them (Fig. 2B).

Multiple sources of the descending catecholaminergic innervation of the nucleus incertus
In studies to determine which brain structures provide the catecholaminergic innervation of the NI, the brains from four rats that received an intra-NI injection of a retrograde viral vector (retro-AAV2-hSyn-mCherry) were examined for colocalization of mCherry and TH immunoreactivity throughout the brain (+1.3 mm to − 10.3 mm from bregma). In all four brains (cases A-D, Fig. 3A and B), TH-ir and mCherry-ir cells were consistently observed in the periaqueductal gray Darkschewitsch nucleus (Fig. 3C), raphe nuclei (including the paramedian raphe nucleus, dorsal raphe nucleus (dorsal, lateral and caudal parts), caudal linear nucleus of the raphe, rostral linear nucleus of the raphe) (Fig. 3D) and in the A13 dopaminergic group (Fig. 3E). In three  (23) 194 ± 78 (16) vGlut2/CCK 7 ± 3 (1) 9 ± 2 (1) D2 only 2 ± 2 (0.1) -CCK only 1 ± 1 (0.1) -brains, colocalization of mCherry and TH immunoreactivity was observed in the lateral and pleoglial periaqueductal gray and the A11 dopaminergic cell group. Single catecholaminergic neurons immunoreactive for mCherry were also present in other brain structures, but were not observed in more than two brains studied (Table 3).

D2-like receptor activation in nucleus incertus has excitatory and inhibitory effects
Recorded NI neurons were divided into two groups based on their unique electrophysiological properties. Type I neurons were identified by the presence of a delay before the first action potential after a hyperpolarizing current pulse, underlined by a robust A-type potassium current; and type II neurons were characterized by a lack of a delay and/ or the presence of rebound depolarization after the hyperpolarizing current pulse, underlined by calcium current passing through voltagedependent channels (Blasiak et al., 2015;Szlaga et al., 2022).
Whole-cell, zero current clamp recordings revealed that 64.7% (11 of 17) of the recorded neurons were responsive to bath application of the D2R agonist, quinpirole (20 μM) in sACSF. Within this group, six neurons (56%) were excited, and the remaining five neurons were inhibited by D2R activation. Notably, type I neurons were inhibited (n = 5) and excited (n = 2) by quinpirole application, whereas type II neurons were exclusively excited (n = 4; Fig. 4A and B).
In studies to test for the possible pre-vs post-synaptic localization of D2R in the NI, five neurons that were responsive to quinpirole in sACSF, were re-treated with the agonist under conditions of pharmacological isolation, i.e., in the presence of TTX, and ionotropic GABAergic and glutamatergic receptors antagonists. All tested neurons remained responsive to quinpirole under pharmacological isolation; four neurons (type I) were hyperpolarized (− 2.44 ± 1.07 mV, mean voltage change) and one neuron (type II) was depolarized by 4.31 mV (Fig. 4A and B).
In voltage-clamp recordings (holding potential − 50 mV), 52% (34 of 66) of recorded cells were sensitive to quinpirole (20 μM) in sACSF, and within this group, in 74% of neurons (n = 25, 13 type I, 12 type II) an increase in inward current in response to quinpirole was observed (14.67 ± 9.42 pA, mean current change). In the remaining 26% (n = 9, 7 type I, 2 type II), an increase in outward current was recorded (13.6 ± 6.53 pA, mean current change; Fig. 4C and D). The mean change in quinpirole-induced inward current was significantly different between NI neuron types (Mann-Whitney unpaired test p = 0.0001, Fig. 4E), with type I neurons displaying significantly greater excitatory responses to quinpirole application. Similar to current-clamp recordings, in voltage-clamp experiments, the majority (n = 17, 80.95%) of neurons responsive to quinpirole application in sACSF, exhibited a change in whole-cell current after quinpirole application in TTX-enriched ACSF; 82.35% (n = 14) of responsive neurons exhibited an increase in inward current (8 type I, 6 type II, mean amplitude of the increase of inward current: 14.29 ± 8.19 pA, Fig. 4F, H), and in 17.65% (n = 3), an increase in outward current was observed (2 type I, 1 type II, mean amplitude of increase of outward current: 8.07 ± 1.86 pA, Fig. 4H). In TTX-enriched ACSF, type I neurons displayed a tendency for a higher amplitude of D2R activation-induced inward current (Mann-Whitney test, p = 0.081), suggesting the larger increase in inward current is due to postsynaptic actions of D2R (Fig. 4F).
Both RLN3-immunopositive and RLN3-immunonegative neurons were responsive to D2R activation (Fig. 4I). Of nine RLN3-ir neurons (7 type I, 2 type II), six were inhibited and the remaining three were excited by D2R activation in sACSF.

D2R activation influences synaptic currents in the nucleus incertus in a neuron type-specific manner
In order to assess possible effects of D2R activation on spontaneous postsynaptic currents in type I (n = 21) and type II (n = 15) NI neurons, quinpirole (20 μM) was applied during voltage-clamp recordings performed in sACSF. Subsequent analysis revealed that D2R activation caused a drop in frequency of iPSCs by 0.12 ± 0.18 Hz (paired t-test, p = 0.024) in type II neurons (Fig. 5F), and a decrease in ePSC rise time by 0.19 ± 0.32 ms (paired t-test, p = 0.011) in type I neurons (Fig. 5L). The remaining parameters associated with the recorded synaptic currents were not affected by quinpirole application (Table 4).

Discussion
In these studies, we characterized the catecholaminergic innervation of the NI at the molecular, circuit and electrophysiological level. Using multiplex in situ hybridization histochemistry, we demonstrated that both GABAergic and glutamatergic NI neurons express D2 receptor mRNA, and that transcripts for these dopaminergic receptors were present in one-third (36%) of all NI cells counted. Our immunohistochemical data and viral-based, retrograde tract-tracing studies revealed an abundance of TH-and DBH-ir fibers within the NI area, and identified the major descending sources of the NI catecholaminergic innervation in the periaqueductal gray, raphe nuclei and hypothalamic A13 nuclei. Finally, using whole-cell, patch-clamp recordings, we demonstrated that a majority of NI neurons is sensitive to D2R activation by quinpirole, in a cell type-dependent manner; and that D2R activation had a direct excitatory or inhibitory influence on NI neuronal activity. Our results indicate that the rat NI is under a strong catecholaminergic influence, that can be both dopaminergic and noradrenergic in nature, and provide insights into the possible involvement of this innervation in the regulation of diverse functions including locomotor activity, social interactions and nociceptive signaling. Moreover, our findings support the hypothesis that the NI acts as an integration hub for arousal-related neural information.
The NI has attracted considerable research recently, and has been shown to play a role in theta rhythm control, locomotor activity, arousal, and stress responses in rats and mice (Lu et al., 2020;Szlaga et al., 2022;Szőnyi et al., 2019;Trenk et al., 2022). Chemo-and opto-genetic activation of NI neurons increased theta power and locomotor speed and induced arousal (Lu et al., 2020;Ma et al., 2017a). It was also shown that D2R activation in the rat NI by intra-NI infusion of the D2R agonist, quinpirole, induced home cage hypolocomotion and suppression of the velocity and distance travelled in a novel environment (Kumar et al., 2015), suggesting the main effect of D2R activation in the NI is inhibition of neuronal activity. However, the NI consists of heterogeneous neuronal populations that differ in their in vitro electrophysiological properties and their neurochemical phenotype, and distinct NI neuronal populations belong to separate neuronal circuits (Cervera-Ferri et al., 2012;Sutin and Jacobowitz, 1988;Szlaga et al., 2022;Trenk et al., 2022). Our present is situ hybridization data revealed that D2 receptor mRNA is present in both putative GABAergic and glutamatergic NI neurons, and that ~50% of RLN3 and ~25% of CCK NI neurons express D2 receptor transcripts. These data indicate that different NI neuronal populations, possibly involved in distinct neuronal  (caption on next page) A. Szlaga et al. circuits, are sensitive to D2R agonists. The potential physiological importance of D2R signaling in the NI is reflected by the fact that D2 receptor mRNA is present in one third of all NI cells, which is consistent with the relatively dense innervation of the NI by catecholaminergic fibers, as shown by the levels of TH and DBH staining. These data further establish the anatomical and molecular basis of the observed behavioral effects of D2R activation in the NI (Kumar et al., 2015), and indicate the need for additional research to investigate the involvement of specific, D2R-expressing NI neuronal populations in defined neuronal circuits. Our immunohistochemical and neural tract-tracing data indicate that the catecholaminergic innervation of the NI may be both dopaminergic and noradrenergic in nature, as TH-(the rate-limiting enzyme of catecholamine synthesis) and DBH-(the enzyme that catalyzes the synthesis of noradrenaline from dopamine) positive fibers were detected within the NI area, and TH-positive neurons from both dopaminergic and noradrenergic brain structures were shown to directly innervate the NI. Brain areas identified as major sources of the catecholaminergic input to the NI were the Darkschewitsch nucleus (Dk), raphe nuclei and A13 dopaminergic cell group. Input from the Dk to the NI in the mouse has been described (Lu et al., 2020), but the neurochemical nature of this projection was unknown until now. We demonstrated that the Dk-NI projection is, at least in part, catecholaminergic, but to the best of our knowledge, it is not established what kind of catecholamine(s) is/are synthetized in the Dk. An early study of intracerebroventricular injections of radioactive noradrenaline, suggested the possible noradrenergic nature of Dk neurons in the rat (Reivich and Glowinski, 1967). However, in situ hybridization data does not detect the presence of DBH mRNA, while confirming the presence of TH mRNA in this brain area in mice (Allen Brain Atlas, 2004).
Surprisingly little is known about the neurophysiological role of Dk neurons, and most information come from studies in cats. Early anatomical data suggested the Dk is involved in oculomotor control (Bianchi and Gioia, 1986), while some later anatomical studies implicated Dk in the control of locomotor activity (Onodera and Hicks, 1995;Rutherford et al., 1989). More recent studies in mice have revealed that neurons in Dk and the neighboring Edinger-Westphal nucleus (which also innervates the NI) express mRNA for urocortin, a peptide closely related to corticotropin-releasing hormone (CRH), which indicates a possible role of Dk in stress responses (Weninger et al., 2000). Notably, urocortin-positive fibers have been identified in the rat NI (Bittencourt et al., 1999). Taken together, these data on the anatomical relationship between the Dk and NI, suggest this pathway is involved in stress and locomotor activity control.
Among the structures identified in the current studies as sources of a catecholaminergic innervation of the NI were the raphe nuclei, which together with the dopaminergic neurons of the ventrolateral periaqueductal gray (vlPAG), can be considered as a dorso-caudal extension of the A10 group (with the majority of dopaminergic cells localized in the ventral tegmental area) (Léger et al., 2010;Li et al., 2014;Stratford and Wirtshafter, 1990). Bidirectional connections between the raphe nuclei/PAG and the NI have been described in rat and mouse (Goto et al., 2001;Lu et al., 2020), but the role of these pathways is currently unclear. It has been reported that dorsal raphe nucleus (DRN) and PAG dopamine neurons are involved in arousal control, and their optogenetic activation promoted wakefulness, whereas their chemogenetic inhibition strongly opposed wakefulness, even in the presence of salient stimuli (Cho et al., 2017;Lu et al., 2006). Therefore, the dorsal raphe/vlPAG dopaminergic innervation of the NI, may be an important element of the arousal control circuit, as it has been shown that chemoand opto-genetic activation of NI neurons promotes arousal, cortical desynchronization, hippocampal theta rhythm and locomotor speed (Lu et al., 2020;Ma et al., 2017a).
shaping social interactions, oxytocin is a key modulator in pain and analgesia control (Rash et al., 2014). Other key players in nociceptive processing are the A13 and A11 dopaminergic neurons located in the medial portion of the zona incerta (Moriya and Kuwaki, 2021;Puopolo, 2019). Projections from this brain area to the NI have been described in the rat (Goto et al., 2001), and we have identified the catecholaminergic nature of these connections. Thus, it will be of interest to investigate the involvement of the NI and/or its associated peptides systems, such as the relaxin-3/RXFP3 network, with nociceptive processing, especially as pain, including chronic pain, is a significant and growing public health problem (Mills et al., 2019). A11 and A13 dopaminergic neurons are also implicated in the modulation of locomotor activity, as they innervate locomotor centers in the brainstem and spinal cord, and optogenetic activation of A11 neurons increases locomotion (Koblinger et al., 2018;Sharma et al., 2018). In view of the aforementioned close relationship of NI neuronal activity with locomotor activity, the demonstrated catecholaminergic connections from the A13 and A11 areas should also be verified in this context, in future studies.
The possible association of catecholaminergic signaling within the NI with diverse neuronal populations belonging to different neuronal circuits, is supported not only by the current in situ hybridization and neural tract-tracing data, but also by the different responses of NI neurons to D2R activation. We observed that quinpirole-induced D2R activation could result in both excitation and inhibition of recorded neurons, and that the type of response was dependent on the NI cell type. Type II neurons were almost exclusively excited, while type I NI neurons were equally excited or inhibited upon D2R activation. This effect also occurred under conditions of pharmacological blockade of spiking and synaptic activity (in the presence of TTX and ionotropic glutamate and GABA receptor antagonists), which indicates that the D2R activated were situated postsynaptically on recorded neurons. However, this does not exclude the possibility of presynaptic expression of D2R within the NI.
In accordance with the excitatory effect of D2R activation on type II NI neurons, manifested either as depolarization in the current clamp or an increase in the whole cell inward current in voltage clamp mode, were our observations of the influence of quinpirole on postsynaptic currents (PSCs) in these neurons. The recorded decrease in frequency of inhibitory PSCs and concomitant reduction in the rise time of the excitatory PSC, very likely contributed to the observed excitatory action of D2R activation. Importantly, we recently demonstrated that the electrophysiological characteristics of a given NI cell population recorded ex vivo, is closely associated with the neuronal circuit to which it belongs, and that the NI-medial septum axis is composed exclusively of type I NI neurons . Also, our in vivo recordings revealed that the type of electrophysiological activity differentiates neurons that belong to different neuronal circuits, and that only regular, fast-firing NI neurons (putative type I ex vivo) innervate the medial septum, whereas irregular and bursting NI neurons (putative type II ex vivo) receive feedback input from the medial septum . Therefore, diverse and cell-type, specific effects of D2R activation may be related to the neuronal circuit to which a given catecholamine-sensitive neuron belongs, and depend on the Amplitude of quinpirole-induced inward current in type I and type II NI neurons. Note the significantly smaller quinpirole-induced current in type II than in type I neurons (Mann-Whitney unpaired test, p = 0.001, median ± interquartile ranges presented). (F) Amplitude of quinpirole-induced inward current in type I and type II NI neurons under conditions of pharmacological isolation (Mann-Whitney unpaired test, p = 0.081, median ± interquartile ranges presented). (G) Number of NI neurons excited and inhibited by quinpirole and non-responsive (numbers summed from current-and voltage-clamp recordings) in sACSF. (H) Number of NI neurons excited and inhibited by quinpirole and non-responsive (numbers summed from current-and voltage-clamp recordings) under conditions of pharmacological isolation. (I) Fluorescence microscope image of an exemplary NI neuron (indicated by the arrow) stained after recording for RLN3 (left panel) and biocytin (middle).
A. Szlaga et al. intracellular signaling cascades within a given type of neuron.
The heterogeneous nature of neuronal responses to D2R activation has been described earlier, although an inhibitory effect is more commonly observed (Bonci and Hopf, 2005). The inhibitory D2R action is mediated by coupling to G αi/o -protein and inhibition of adenylyl cyclase, and a subsequent drop in cAMP production (Missale et al., 1998). This may result in the reduction of Ca 2+ flow through L-and N-type channels (Hernádez-López et al., 2000;Yan et al., 1997), activation of inwardly-rectifying potassium channels (GIRKs) (McCall et al., 2019), and interaction with Kir3 potassium channels (Lavine et al., 2002). The excitatory action of D2R activation has been shown to be associated with Gβγ protein coupling and activation of phospholipase C (PLC) (Valler et al., 1990). PLC can directly modulate inositol 1,4,5-trisphosphate (IP3), diacylglycerol (DAG) and phosphatidylinositol 4, 5-bisphosphate (PIP2) signals, and lead to an increase in calcium levels in the cytoplasm (Putney and Tomita, 2012). For example, a direct excitatory action of D2R activation was observed in parvalbumin interneurons in primary motor cortex (Cousineau et al., 2020), supraoptic nucleus neurons (Yang et al., 1991) and dorsolateral geniculate nucleus interneurons (Munsch et al., 2005). D2R activation was also shown to have an excitatory action on dorsal raphe serotonin neurons, via activation of a nonselective cation current (Aman et al., 2007;Fig. 5. Influence of D2R activation on spontaneous synaptic currents in NI neurons. (A) Exemplary trace of voltage-clamp recording (holding potential − 50 mV). Inward currents represent excitatory postsynaptic currents (ePSC) that were blocked by the glutamate receptor antagonists, CNQX and DL-AP5, and outward currents represent inhibitory postsynaptic currents (iPSC), that were blocked by the GABA A R antagonist bicuculline (Bic). (B-E) Frequency and amplitude of iPSC and ePSC in type I NI neurons at baseline (base; in black) and after D2R agonist, quinpirole (20 μM) application (quin; in orange). (F-I) Frequency and amplitude of iPSC and ePSC in type II NI neurons at baseline (base; in black) and after D2R agonist quinpirole (20 μM) application (quin; in orange). A drop in iPSC frequency upon D2R activation was observed in type II neurons (F). (J-M) Rise time (left inset panels) and decay time (right inset panels) of iPSC (left panel) and ePSC (right panel) type I and type II NI neurons. A decrease in ePSC rise time was observed after D2R activation in type I NI neurons (L). Other parameters of iPSC were not influenced by D2R agonist application. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Haj-Dahmane, 2001), while we have demonstrated that NI is another brainstem structure in which D2R activation can have a direct excitatory effect.
The diversity in the origin of the NI catecholaminergic innervation, the opposite effects of D2R activation on neuronal activity, as well as the variety of neurochemical profiles of NI neurons expressing D2 receptor mRNA, strongly suggest that catecholamines acting within the NI are involved with distinct neuronal circuits. This in turn strengthens the hypothesis that the NI is a hub integrating arousal-related information (Gil-Miravet et al., 2021;Ma et al., 2017b;Olucha-Bordonau et al., 2018;Ryan et al., 2011). Finally, the current data indicate a need for further studies of the effects of catecholamines on NI neuronal activity in which the catecholaminergic origin to the NI is specified.

Technical considerations regarding retrograde tracing studies
It is important to note that in the current study, some structures were retrogradely labeled in some but not all examined brains, which may be due to differences in the area of NI that was transfected with viral vectors in the different cases. This finding suggests that the described retrogradely-labeled structures innervate defined areas of the NI. Therefore, further studies are needed to map the distribution of THpositive fibers, originating from defined dopaminergic and noradrenergic brain structures, across the NI area. It is also worth noting that in tracing studies there is the possibility that some of the retrogradely labeled neurons, accumulate viral vectors (or other molecules in the case of conventional tracers) from the near vicinity of the structure targeted for injection. Therefore, the relative abundance of retrogradely-labeled neurons is an important aspect of interpreting such labelling results.

Funding
This study was supported by research grants from The National Science Centre Poland (UMO-2018/30/E/NZ4/00687 to A.B.) and the Institute of Zoology and Biomedical Research of the Jagiellonian University in Krakow (DS/D/WB/IZiBB/17/2019 to A.S.).

Declarations of interest
None.

Data availability
Data will be made available on request.