Leptin coordinates efferent sympathetic outflow to the white adipose tissue through the midbrain centrally-projecting Edinger-Westphal nucleus in male rats

The centrally-projecting Edinger-Westphal nucleus (EWcp) hosts a large population of neurons expressing urocortin 1 (Ucn1) and about half of these neurons also express the leptin receptor (LepRb). Previously, we have shown that the peripheral adiposity hormone leptin signaling energy surfeit modulates EWcp neurons' activity. Here, we hypothesized that Ucn1/LepRb neurons in the EWcp would act as a crucial neuronal node in the brain-white adipose tissue (WAT) axis modulating efferent sympathetic outflow to the WAT. We showed that leptin bound to neurons of the EWcp stimulated STAT3 phosphorylation, and increased Ucn1-production in a time-dependent manner. Besides, retrograde transneuronal tract-tracing using pseudorabies virus (PRV) identified EWcp Ucn1 neurons connected to WAT. Interestingly, reducing EWcp Ucn1 contents by ablating EWcp LepRb-positive neurons with leptin-saporin, did not affect food intake and body weight gain, but substantially (+26%) increased WAT weight accompanied by a higher plasma leptin level and changed plasma lipid profile. We also found that ablation of EWcp Ucn1/LepRb neurons resulted in lower respiratory quotient and oxygen consumption one week after surgery, but was comparable to sham values after 3 and 5 weeks of surgery. Taken together, we report that EWcp/LepRb/Ucn1 neurons not only respond to leptin signaling but also control WAT size and fat metabolism without altering food intake. These data suggest the existence of a EWcp-WAT circuitry allowing an organism to recruit fuels without being able to eat in situations such as the fight-or-flight response.

The EWcp is the major seat of urocortin 1 (Ucn1) neurons in the brain (Kozicz et al., 1998;Bittencourt et al., 1999). Ucn1 is an important modulator of metabolic functions and feeding behavior (Kuperman and Chen, 2008). Intracerebroventricular administration of Ucn1 strongly suppresses food intake (Benoit et al., 2000;Wang et al., 2001), and increases body temperature and oxygen consumption (De Fanti and Martínez, 2002). In the mouse, EWcp/Ucn1 neurons respond to in vitro leptin administration with a slow activation of the JAK2-STAT3 signaling pathway (Xu et al., 2011). Systemic injection of leptin increases Ucn1-content of the EWcp, whereas mice lacking LepRb exhibit decreased Ucn1 expression (Xu et al., 2011).
While most studies have been focused on leptin's action on the brain, the brain also modulates the efferent sympathetic outflow to the adipose tissue (Zhang et al., 2014). As the primary energy store, the adipose tissue plays an important role in energy balance. Histologically and functionally two distinct types of fat tissues can be distinguished: white adipose tissue (WAT) and brown adipose tissue (BAT). The WAT stores chemical energy in the form of triglycerides whereas the BAT dissipates chemical energy in the form of heat through thermogenesis (Zhang et al., 2014). Because BAT has a relatively small fat storage capacity and a high metabolic demand to support a continuous thermogenic response, it relies on continuous free fatty acids from WAT. The sympathetic nervous system (SNS) is the main driver for adipose tissue lipolysis. In the WAT, SNS stimulation leads to the breakdown of lipids, causing smaller fat cell size and reducing leptin release. In the BAT, SNS stimulation induces lipolysis and thus Uncoupling Protein 1 (UCP1) activation (a marker for thermogenesis). Therefore, it has been hypothesized that coordinated sympathetic action of BAT and WAT would maximize leptin's effect on energy expenditure and fat metabolism (Zeng et al., 2015). However, an important research gap exists in understanding how leptin actions would be connected to the sympathetic activation of WAT and BAT.
Although the Edinger-Westphal nucleus has been traditionally considered as the parasympathetic component of the oculomotor complex (e.g. Warwick, 1954;Klooster et al., 1993), retrograde infection of EWcp neurons has been reported after pseudorabies virus (PRV) injection into the sympathetic stellate ganglion, BAT, pancreas, spleen and adrenal gland (Jansen et al., 1997;Farkas et al., 1998;Cano et al., 2001;Zhang et al., 2011;Shah et al., 2013). Besides, Ucn1 fibers occur in the lamina VII of the mouse spinal cord (Korosi et al., 2007). Importantly, the two other major Ucn1-containing brain areas, the supraoptic nucleus and lateral superior olive (Bittencourt et al., 1999) do not project to the lateral horn of the spinal cord (Tucker and Saper, 1985). These facts strongly implicate EWcp/Ucn1 neurons in controlling the sympathetic nervous system's impact on adipose tissue function (for review see: Cano et al., 2021).
Based on the above data, we hypothesized that LepRb neurons in the EWcp would form an important component of a midbrain-SNS-WAT axis. More specifically, EWcp/Ucn1 neurons would 1) receive a message via leptin about the abundance of lipids in the WAT; 2) integrate this message via second messenger systems such as JAK2-STAT3, MAPK/ ERK, and PI3K/AKT to mediate EWcp neuron's activity, and modulate sympathetic tone that, in turn, would control WAT function. To test this hypothesis, we first assessed whether leptin directly bound to EWcp/ Ucn1 neurons. Next, we comprehensively evaluated the effect of peripheral leptin administration on second messenger signaling (pSTAT3, pAKT, and pERK) and Ucn1 peptide and mRNA dynamics in the EWcp. To test that EWcp-Ucn1 neurons indeed innervate the WAT, we applied retrograde viral tracing. Finally, we investigated whether ablation of leptin receptor-expressing EWcp-Ucn1 neurons would influence WAT size, leptin dynamics, and metabolic parameters such as oxygen consumption, carbon dioxide production, and respiratory quotient (calorimetry).

Animals
Studies were performed with male Wistar rats (225-390 g). Animals from our Animal Facility of the Department of Anatomy in Pécs (original breeding pairs purchased from Charles River), were used for leptin administration (28 rats), LepRb binding (8 rats), LepRb neuron ablation (32 rats) and calorimetry (8 rats). Retrograde tracing was done in 8 rats from Toxi-coop, Budapest, Hungary. Rats were housed in standard plastic cages (40 × 25 × 20 cm; 6 rats/cage) in a temperature-and humidity-controlled environment, on a 12-hrs light/12-hrs dark cycle (lights on at 6:00 AM), with free access to tap water and rat chow. They were allowed 1 week of acclimation before an experiment. The use of male rats was preferred in this study because the EWcp/Ucn1 neurons are estrogen sensitive and the effect of the estrus cycle could have influenced the results (Derks et al., 2010).
Animals were sacrificed by an overdose of anesthetic injection (sodium-pentobarbital, Nembutal, Sanofi, Budapest, Hungary; 100 mg/kg body weight). All efforts were made to minimize the number of animals used and their suffering. In vivo experimental procedures were permitted by the National Food Chain Safety Office in Hungary (license numbers: BA02/2000-26/2011and BA02/2000-22/2017. The licenses were given based on the scientific approvals of the Animal Welfare Committee at Pécs University and the National Scientific Ethical Committee on Animal Experimentation in Hungary.

Cy3-labeled leptin binding
To test if leptin binds specifically to EWcp-neurons, rats were transcardially perfused with 50 ml of 0.1 M sodium phosphate-buffered saline (PBS; pH 7.4), decapitated, and their brains rapidly dissected.
Then, 100 μm-thick brain sections containing the EWcp were incubated with 1 μМ (n = 4 rats) or 50 nM (n = 4 rats) Cy3-labeled leptin (Phoenix Pharmaceuticals, Burlingame, CA, USA) for 20 min at 4 • C. As a negative control EWcp sections from the same rats were incubated with inactivated, Cy3-labeled leptin that was boiled for 10 min before use. Subsequently, the sections were fixed with 4% paraformaldehyde (PFA), mounted on glass slides, and analyzed with a confocal laser scanning microscope (Leica Microsystems TCS SP2 AOBS system, Wetzlar, Germany). For further details of this method, see Kieffer et al. (1997).

Leptin injection
Rats (n = 4 per group) were injected intraperitoneally (i.p.) with either leptin (3 mg/kg in sterile PBS, pH 7.4), or an equal volume vehicle, and sacrificed 0, 1, 2 or 4 h later. The dose of leptin injection was determined based on an earlier work by Rizk et al. (2001). Recombinant mouse leptin was obtained from the National Hormone and Peptide Program (Dr. A.F. Parlow, Los Angeles, CA, USA). After deep anesthetization with Nembutal (100 mg/kg body weight), they were transcardially perfused with 50 ml of 0.1 M PBS (pH 7.4) followed by 250 ml 4% ice-cold PFA in 0.2 M Millonig sodium phosphate buffer (pH 7.4). After decapitation, brains were dissected and stored in the PFA for 2 days. Then, brains were transferred into 30% sucrose in PBS, and when completely submerged, 6 series of 25 μm-thick coronal sections were cut at the midlevel of the EWcp (5.0-7.0 mm caudal to the level of Bregma; see Paxinos and Franklin, 2003) on a freezing microtome (Microm, Walldorf, Germany), saved in sterile antifreeze solution (0.1 M PBS, 30% ethylene glycol, 20% glycerol), and stored at − 20 • C until immunohistochemistry for pSTAT3, pAKT, pERK and Ucn1.

Retrograde transneuronal tract-tracing
For retrograde tracing, Ba-DupGreen (BDG), a recombinant, Barthaderived pseudorabies virus expressing green fluorescent protein (GFP; Zs. Boldogköi, University of Szeged, Szeged, Hungary) was used (Boldogköi et al., 2002). This virus strain shows highly specific retrograde, transneuronal spread, which has been extensively characterized in our previous studies (e.g. Dénes et al., 2005;Dénes et al., 2006). Considering the spread of the virus in sympathetic circuits, upon uptake by nerve terminals in the target organ (i.e. WAT) viral capsids travel in a retrograde direction to ganglia, replicate and infect neurons that reside in the intermediolateral cell column of the spinal cord (IML). IML neurons are innervated by various autonomic nuclei in the brain stem and the hypothalamus as it has also been demonstrated in various track tracing studies using different PRV strains. BDG has special features compared to earlier PRV derivatives, as it is shows highly specific retrograde, transneuronal spread and expresses GFP with immediate-early kinetics under the regulation of the human CMV promoter, allowing early identification of infected neurons in different brain areas.
BDG virus was grown in porcine kidney (PK-15) cells to a titer of 6 × 10 8 plaque-forming units (PFU)/ml. BDG was concentrated to 1.5 × 10 10 PFU/ml by ultracentrifugation at 70,000 g (for details see Dénes et al., 2006). The virus stocks were aliquoted and stored at − 80 • C. Aliquots were thawed immediately before injection and stored at 4 • C during surgery.
Finally, the peritoneum and abdominal wall was sutured.
Rats were housed individually and monitored twice a day. Five days later, animals were sacrificed by deep Nembutal (100 mg/kg) anesthetization and transcardially perfused with 50 ml saline followed by 400 ml ice-cold Zamboni's solution (4% PFA and 0.2% picric acid in 0.1 M PBS; pH 7.4). Brains were postfixed in the same fixative for 16 h, and cryoprotected in 10% sucrose. Four series of 30 μm thick sections were cut and were stored at − 20 • C in sterile antifreeze solution as described above. Productive infection was observed in 6 rats, by identification of GFP-containing neurons along the sympathetic neuraxis (Dénes et al., 2005).

Neuron ablation
Rats were anesthetized by i.p. injection of a mixture of 78 mg/kg ketamine (Calypsol, Richter Gedeon, Budapest, Hungary) and 13 mg/kg xylazine (Sedaxylan, Eurovet Animal Health BV, Bladel, The Netherlands). Then, the scull was drilled on the left side, an incision made on the dura mater, and an injection needle stereotactically inserted at an angle of 19 • from dorsolateral to ventromedial, 4.8 mm posterior and 1.5 mm lateral to Bregma. Either unconjugated saporin (53 ng in 80 nl MQ; Kit-47; Advanced Targeting Systems, San Diego, CA, USA) or leptin-conjugated saporin (90 ng in 80 nl MQ; Kit-47; Advanced Targeting Systems) was injected into the midbrain, to a depth of 5.8 mm from the dura mater, directly dorsal to the EWcp. Fourteen days later, rats were weighed, sacrificed, and their brains processed and stored in the same way as described above for the leptin injection study. Left and right epididymal WAT and BAT were collected. In case of BAT, a midline skin incision was made along the spinous processes of lower cervical and thoracic vertebrae to expose the interscapular area. Then, by careful dissection, the tissues were separated in the midline till the spinous processes of vertebrae got visible. This was conducted in order to accurately separate the left and right BAT. After dissection and removal of the right and left WAT and BAT tissues, they were weighed and cut into two pieces. One part was put in containers snap-frozen and stored at − 20 • C, the other piece was immersion-fixed and embedded in paraffin.
Sections of the midbrain (25 μm) were made using a vibratome as described above. Three series of 4 μm-thick sections of WAT and BAT were mounted on SuperFrost® microscope slides (Menzel-Gläser, Braunschweig, Germany) for histological stainings (see below).

Calorimetry
Rats were housed individually in metabolic chambers. Four chambers were used at the same time. O 2 consumption and CO 2 production were recorded by indirect calorimetry with an Oxymax calorimeter (Equal Flow, Columbus, OH, USA; for details of its working, see ref Füredi et al., 2017), starting with a 5-day control period before the surgery (on day 6) and ending with follow up tests in weeks 1, 3 and 5. O 2 consumption (ml/kg/min) and CO 2 production (ml/kg/min) were measured in air that had been perfused through the metabolic chamber, with 10 min intervals, except during daily chamber opening for cleaning (from 09.00 till 09.30 a.m.). O 2 and CO 2 data were averaged over 12-h periods, representing the light and the dark period, and the respiratory quotient (RQ) calculated. Bodyweight and 24-h food consumption were measured daily between 9.00 and 10.00 a.m.

Leptin titer and lipid profile
Blood samples were taken 2 weeks after surgery and the plasma tested for leptin titer and lipid profile: plasma triglycerides, cholesterol, high-density lipoproteins (HDL), low-density lipoproteins (LDL), and non-high-density lipoproteins (NHDL). Leptin was determined with a leptin ELISA kit (EZML-82K; Merck, Darmstadt, Germany). Plasma lipids were measured following standard protocols in the Department of Laboratory Medicine at our Radboudumc.

Hematoxylin-eosin staining
Standard HE staining was performed by immersing xylenedeparaffinated and alcohol-hydrated sections in hematoxylin and then in 0.1% eosin, both for 10 min. The sections were embedded in Entellan (Merck, Darmstadt, Germany).
To reveal WAT innervating Ucn1 neurons, double immunolabeling of GFP and Ucn1 was performed. EWcp sections were processed as described for single immunolabeling, but with incubation in a mixture of primary chicken anti-GFP serum (1:2000) and primary rabbit anti-Ucn1 serum (1:30,000), for 16 h, and then in a mixture of Cy2-conjugated anti-chicken and Cy3-conjugated anti-rabbit sera (1:100) for 2 h. Finally, sections were mounted on gelatin-covered slides, air dried and coverslipped with a few drops of FluorSave™ (Merck).

RNA extraction from adipose tissue and cDNA synthesis
Small pieces of WAT or BAT (~1-10 mm 3 ) were used for RNAse-free RNA isolation. Tissue was incubated with 500 μl TRIzol® (Invitrogen) on ice, for 30 min, crushed using small glass cylinders, and sonificated to maximize RNA yield in an Ultrasonic processor (Sonics & Materials, Newton, USA), for 30 s. After adding 200 μl chloroform, the sample was centrifuged for 15 min at 4 • C, at 14,000 rpm. The aqueous phase was incubated for 10 min at 20 • C with 1 μl (20 μg) of glycogen (Thermo Fisher Scientific, Waltham, MA, USA) and 500 μl iso-propanol, and centrifuged again at 4 • C and 14,000 rpm, for 10 min. The cell pellet was rinsed for 1 min with 1 ml 75% ethanol, centrifuged at 4 • C and 14,000 rpm, for 5 min, air-dried, dissolved in 50 μl diethyl pyrocarbonate (DEPC)-treated MQ water, and its RNA concentration determined with a ND-1000 spectrophotometer (NanoDrop, Wilmington, USA).

Quantitative RT-PCR
Q-PCR was performed with a 7500 Real-Time PCR machine (Applied Biosystems, Foster City, USA), in FrameStar 96 well plates (4titude, Surrey, UK) filled with a mixture of 1.5 μl DEPC-treated MQ, 12.5 μl Power SYBR® Green mixture (2x; Applied Biosystems) and 3 μl (15 pmol) UCP1 (forward CAGAACTCACCGCTCAACAG, reverse ATC-CATGGACGTTGCTTGTC) or ADRB3 (forward CCACA-GAAAGCTTGTCAACACT, reverse AATGGCAGGGGACGTCAT) primers (both from Biolegio, Nijmegen, The Netherlands) to which 5 μl cDNA had been added. As a reference gene, ribosomal subunit 18S was used (forward: TTGCTGATCCACATCTGCTGG, reverse ATTGCCGA-CAGGATGCAGAA) as this housekeeping gene appeared to be reliable (Derks et al., 2008). The reaction cycle was as follows: a single denaturation step of 10 min at 95 • C, followed by 40 repetitions of 15 s of denaturation at 95 • C and 1 min of elongation at 60 • C. Next, dissociation was performed during 15 s at 95 • C, 1 min at 60 • C, and finally 15 s at 95 • C. qPCR data were analyzed using the Applied Biosystems 7500 PCR machine SDS 10 software. The 2 − ΔΔCt method has been applied to further evaluate the gene expression after assessment of melting curves.

Image analysis
Images were made with either an Orthoplan light microscope (Leitz, Wetzlar, Germany) equipped with a DFC420C digital camera (Leica Microsystems, Wetzlar, Germany), or a DM IRE2 inverted fluorescence microscope (Leica). For quantitation of the amount of immunoreactive material in the EWcp, two parameters were studied: 1) the total number of immunoreactive neuronal profiles present in 4 coronal sections of the EWcp between − 5.5 mm and − 6.5 mm to the Bregma, and 2) per neuron, the specific immunoreactivity (ir) signal density (SSD) averaged over all neurons present in these sections, using Image J software (version 1.37, NIH, Bethesda, MD, USA). The SSD was corrected for background density outside the EWcp and expressed in arbitrary units per neuron. Measuring the size of adipocytes was done using the ImageJ software. Sizes of adipocyte fat vacuoles were expressed as the average of measurements performed by two independent observers.

Statistics
Data are presented as means with the standard error of the mean (SEM). Data shown in Figs. 3-5 were analyzed by two-way analysis of variance (ANOVA), with 'leptin' and 'time' as independent variables, preceded by tests for normality (Shapiro-Wilk test;Shapiro and Wilk, 1965) and homogeneity of variance (Bartlett's Chi-square test; Snedecor and Cochran, 1989), followed by Fisher's post hoc analysis (Stat Soft, Tulsa, OK, USA) (α = 5%). Datasets in Figs. 7-10 were evaluated using two-tailed type 2 Student's t-test after testing for normality and variance.

The sensor: leptin binds to EWcp neurons
Previously we showed that approximately ~45% of EWcp-Ucn1 neurons expressed the short form of leptin receptor and responded to i.p. leptin injection (Xu et al., 2011). However, to study that leptin can physically bind to leptin receptor on EWcp-Ucn1 neurons and its effect on Ucn1 neuronal structure and function is not mediated by indirect neuronal pathways, we incubated acute midbrain slices containing the EWcp with Cy3-labeled leptin. Cy3-labeled leptin binding was present in the EWcp. Cy3-labeled leptin in 1 μM concentration caused a clear strong labeling of EWcp neurons. The 50 nM Cy3-labeled leptin provided a punctate cell surface signal in the EWcp (Fig. 1). This provides evidence that leptin binds to leptin receptor on EWcp-Ucn1 neurons.

The integrator: effect of leptin administration on EWcp-Ucn1 neuronal signaling
Previously, we reported that leptin activated STAT3 signaling in the EWcp-Ucn1 neurons within 2-4 h of i.p. injection (Xu et al., 2011). However, previous research has shown that leptin also recruits intracellular pathways other than STAT3 (for review see Liu et al., 2018). To obtain a comprehensive understanding of intracellular signaling pathway(s) recruited by leptin, we assessed the presence and dynamics of pSTAT3, pAKT and pERK, in a time-dependent manner. Leptin crosses the blood-brain barrier with differential dynamics (Banks et al., 2000;Faouzi et al., 2007). Therefore, we examined brain samples 0, 1, 2 and 4 h after leptin injection. To validate our approach, we first investigated the widely characterized and validated effect of i.p. leptin on pSTAT3 activation in the mediobasal hypothalamus (MBH), with saline-injected rats as controls. In line with previous literature (Xu et al., 2011), the strongest pSTAT3-ir in neuronal cell nuclei was observed 2 h after Fig. 1. Leptin binds to neurons in the centrally-projecting Edinger-Westphal nucleus (EWcp). Representative confocal images of acute slices of the rat brain treated with 1 μM (A) and 50 nM (B) Cy3-conjugated leptin. Note that the red immunofluorescence marks neuronal perikarya in the EWcp area suggesting that leptin binds to its receptors (see also higher magnification inserts). As a control, inactivated (boiled) Cy3-conjugated leptin treatment was performed that did not give any detectable red signal (C). Bars 30 μm.
injection ( Supplementary Fig. 1). Similarly, the most robust activation of pSTAT3 in the EWcp was observed at 2 h (Fig. 2), whereas no immunopositive cell nuclei were seen at 1 h and 4 h (Fig. 2). Saline-injected rats did not show pSTAT3-ir either in MBH or EWcp at any time points.
In line with literature data, in the MBH, leptin administration induced pAKT-ir relative to saline-treated animals, with a maximal effect at 4 h ( Supplementary Fig. 1). Interestingly, no pAKT-ir was observed in the EWcp at any time point after leptin administration (Fig. 2) suggesting that leptin does not recruit the AKT signaling pathway in the EWcp.
These results provide additional evidence that leptin directly recruits EWcp-Ucn1 neurons and primarily recruits STAT3 and pERK signaling pathways.

The integrator: effect of leptin administration on Ucn1 mRNA and peptide abundance
Previously we reported that leptin increased Ucn1 protein expression 4 h after i.p. leptin injection (Xu et al., 2011). To show that such a leptin-mediated increase in Ucn1 peptide abundance is accompanied by increased mRNA expression, we quantitatively assessed the Ucn1 mRNA content of the EWcp neurons using in situ hybridization. We found a clear hybridization signal in EWcp neurons (Fig. 4). The strength of hybridization was quantified by counting the number of mRNA-expressing neurons and determining the neuronal SSD of the hybridization signal. We found no effects of leptin or time or a leptin × time interaction on the number of neurons (Fig. 4I). However, the SSD of the neurons showed effects of leptin (F1,24 = 11.46; P < 0.005) and time (F3,24 = 3.71; P < 0.05) as well as a leptin × time interaction (F3,24 = 3.5; P < 0.05; Fig. 4E-H). In saline-injected rats, the SSD revealed a tendency to increase at 1 h after injection (+18%; P = 0.09; Fig. 4J), but at 4 h it was significantly lower than at 1 h (− 32%; P < 0.005) and 2 h (− 22%; P < 0.05). In the leptin group, at 1 h, the SSD was higher than at time 0 (+24%; P < 0.05) and stayed at that level at 2 and 4 h. As a result of these time-dependent responses in both groups, 4 h after injection the SSD was markedly higher (+54%) in the leptin group than in the saline group (P < 0.0005).
To determine the amount of Ucn1 peptide in the EWcp in response to leptin injection, we measured the numbers and SSD of the respective Ucn1-ir neurons. For the number, no effect of leptin and time or a leptin × time interaction was observed at any time point after injection (Fig. 5). However, the SSD revealed an almost significant effect of leptin (F1,24 = 3.74, P = 0.06) and post hoc analysis showed that after 4 h (P < 0.05), the SSD of Ucn1-ir neurons was markedly higher (+48%) in the leptin-injected rats than in the controls (Fig. 5).
These results confirm our previous observation, and reveal that leptin first increased Ucn1 mRNA expression followed by increased Ucn1 peptide abundance.

The effector -retrograde tracing from WAT
Pre-sympathetic outflow from EWcp has been previously established towards the stellate ganglion, BAT, pancreas, and spleen (Jansen et al., 1997;Farkas et al., 1998;Cano et al., 2001;Zhang et al., 2011). To explore whether EWcp/Ucn1 neurons are a source of pre-sympathetic innervation to the WAT too, we injected PRV coupled with BDG into the epididymal fat pad. Five days later, we found strong GFP-ir in EWcp, indicating productive infection and the presence of BDG. Each mid-level section of the EWcp showed about 10 immunopositive neurons (Fig. 6A,  D). Double immunolabeling demonstrated that about 80% of the labeled neurons were Ucn1-positive (Fig. 6C, F).
These results strongly suggest that EWcp-Ucn1 neurons provide a higher-order pre-sympathetic innervation of WAT. After a transient strong tendency of increase at 1 h, the specific signal density (SSD) of pERK remained stable at any time points (closed circles in J) in PBS-injected animals. In contrast, 4 h after leptin injection (F) the number of pERK-immunoreactive cells decreased compared to all time points (closed circles in graph I; #P < 0.05). The pERK-SSD transiently increased 1 h after leptin injection (F) followed by a strong decrease at 4 hs (see closed circles in graph J; #P < 0.05 compared to all other time points). Leptin-treated rats showed significantly lower pERK cell count and SSD than PBS injected animals at 4 h (*P < 0.01). Bars: 50 μm.

The effectorthe impact of EWcp/LepRb/Ucn1 neurons on white adipose tissue
To assess EWcp/LepRb/Ucn1 neurons' functional effects on WAT, we injected leptin-conjugated saporin toxin into the EWcp. Successful ablation of LepRb-positive EWcp neurons was assessed by Ucn1immunohistochemistry. As it is rather difficult to perform a midline injection into the EWcp without damaging the superior sagittal sinus, we recently developed an oblique approach (for details see ref. Füredi et al., 2017). With this, our success rate was about 60% with rats revealing a good injection site targeting the EWcp.
EWcp/Ucn1 neuronal ablation resulted in increased weight of the right WAT (+26%; t(11) = -2.162; P < 0.05; control: 0.125 ± 0.012; ablation: 0.158 ± 0.01; the weights of left and right of WAT were corrected for body weight) compared to controls (Fig. 7B). We found no significant effect in the left WAT. Next, we tested whether LepRb/Ucn1 neuron ablation influences WAT size. We counted the number of WAT cells in HE-stained WAT sections. WAT adipocytes showed a relatively large variation in their size both in control and in leptin-saporin-treated rats. In rats with EWcp/ LepRb/Ucn1 neuron ablation, we observed a larger number of large adipocytes and fewer small adipocytes than controls on both left and right WAT (P < 0.05). This change in WAT size-frequency distribution was also mirrored by the mean adipocyte surface area, which showed 25% increase for the left WAT and a 20% increase for the right WAT, compared to the controls (P < 0.05; Fig. 7D and E).
Finally, we also assessed lipolysis in WAT by measuring p-HSL-ir in the right WAT. HSL is the main player involved in the lipolytic pathways. No effect of leptin-saporin injection on the density of p-HSL-ir was observed (P = 0.42; control: 35.33 ± 4.03; ablation: 37.29 ± 5.61; Fig. 7C) suggesting that the ablation of EWcp-Ucn1 neurons does not   These results show that EWcp/Ucn1 neuronal ablation increases right WAT weight and the number of large adipocytes and the mean adipocyte surface area in both left and right WAT but does not directly affect lipolysis.
These results confirm previous observations and show that EWcp/ LepRb/Ucn1 neuron ablation influences BAT weight and increases UCP1 as well as ADRB3 mRNA expression in the BAT.
These results reveal that with EWcp/LepRb/Ucn1 ablation alters plasma lipid profiles.
These results demonstrate that EWcp Ucn1 neuron loss reduces respiratory quotient and oxygen consumption during the dark period.

Discussion
We provide evidence for a midbrain-WAT regulatory circuit via midbrain LepRb/Ucn1 neurons, supported by the following findings: 1) leptin can bind to neurons in the EWcp, 2) peripheral administration of leptin recruits pSTAT3 signaling at 2 h, and decreases pERK signaling at 4 h, 3) peripheral leptin differentially modulates the expression of Ucn1, 4) EWcp-Ucn1 neurons provide pre-sympathetic innervation to the WAT and 5) LepRb-Ucn1 neurons in the EWcp can regulate WAT and BAT structure and function in a different manner. Below these findings will be discussed in detail.

Leptin recruits different signaling pathways in EWcp neurons
We found that leptin indeed binds to its receptor in the EWcp-Ucn1 neurons, suggesting that leptin can directly regulate Ucn1 neurons' activity in the EWcp. This is corroborated by the finding that leptin treatment induced pSTAT3 at 2 h after leptin administration. However, we saw no pSTAT3 at 4 h after the injection. This activation pattern of pSTAT3 in the EWcp is in line with the temporal dynamics of leptin action in the dorsomedial and ventromedial hypothalamus (DMH, VMH; Faouzi et al., 2007), but it is in contrast to leptin-induced STAT3 phosphorylation in the ARC, which peaks already at 30 min after leptin injection and persists for at least 6 h (Guo et al., 2004). The relatively transient action of leptin on the EWcp is in line with our observation of an immediate, short-term effect of leptin on the electrical activity of the EWcp (Xu et al., 2011).
Leptin activates other signaling molecules downstream of LepRb, such as those constituting the PI3K-AKT (Niswender et al. 2001(Niswender et al. , 2004 and MAPK-ERK pathways (Rahmouni et al., 2009). In the present study, no phosphorylation of AKT was detected in the EWcp, suggesting that this pathway is not recruited by leptin action in the EWcp. This idea should not be surprising, because no changes in pAKT were observed upon leptin infusion in the hypothalamus either (Burgos-Ramos et al., 2010) and the effect of leptin on food intake, does not seem to involve AKT activation (Carvalheira et al., 2005). Concerning pERK, the picture is different. We found that 4 h after leptin injection, the amount of pERK-ir was markedly reduced compared to rats injected with saline. This effect differs from the effect of leptin on pERK in other parts of the brain. For instance, leptin rapidly (within 30 min) activates pERK in the hypothalamus, but no leptin effects are seen in extrahypothalamic nuclei like the nucleus of the solitary tract (Benomar et al., 2005;Rahmouni et al., 2009).
Nonetheless, leptin's ability to activate STAT3 indicates that it may be involved in controlling long-term transcriptional changes, such as Ucn1 gene expression. Indeed, we show that Ucn1 gene expression is induced in a time-dependent fashion by peripheral leptin. Specifically, as early as 2 h after leptin injection, Ucn1 mRNA content was elevated, and this elevation continued through 4 h after leptin injection. An increased abundance of Ucn1-ir was only observed 4 h after leptin injection. We conclude that leptin directly binds to LepRb on the EWcp/ Ucn1 neurons, and induces STAT3 activation, which in turn increases Ucn1 biosynthesis evidenced by an initial increase in Ucn1 mRNA expression followed by an increase in Ucn1 peptide abundance.
Here a possible limitation of these results must be noted. The potentially painful i.p. injection may have caused some acute stress that could have affected Ucn1 mRNA and peptide abundance in EWcp cells. However, in our earlier studies we did not detect remarkable Ucn1 neuron activation upon intraperitoneal saline injection (Gaszner et al., 2004) and the acute pain-induced Ucn1 neuron activation results in a faster upregulation of Ucn1 (Rouwette et al., 2012) than observed here. Furthermore, there was a non-significant increase of Ucn1 mRNA expression 1 h after PBS injection that then gradually decreased to baseline level.

EWcp/LepRb/Ucn1 neurons regulate WAT via the autonomic nervous system
Earlier studies showed that when PRV was injected into WAT specifically, neurons in the paraventricular nucleus (PVN) and medial preoptic area (MPA) within the hypothalamus, and many nuclei in the brainstem and the spinal cord were strongly labeled (Bamshad et al., 1998). Sympathetic nerve fibers arise from the intermediolateral (IML) column of the thoracic spinal cord and project to stellate ganglia located just outside of the spinal cord. In turn, stellate ganglia give rise to postsynaptic sympathetic nerve fibers, which subsequently innervate the WAT (Geerling et al., 2014). Here we provide evidence that EWcp Ucn1 neurons were retrogradely labeled after PRV-GFP injection into WAT as well. BDG has special features compared to earlier PRV derivatives, as it shows highly specific retrograde, transneuronal spread and expresses GFP with immediate-early kinetics under the regulation of the human CMV promoter, allowing early identification of infected neurons in different brain areas. These properties are due to specific modifications of the viral genome, which have been described in detail earlier (Boldogköi et al. 2002(Boldogköi et al. , 2004. The route of virus infection is closely identical in the case of different target organs until the infection reaches the brain, which has been comprehensively characterized in previous studies. Thus, we have only enclosed pictures of the EWcp nucleus and have not investigated ganglia outside the spinal cord and neurons in the IML in the present paper. Viral labelling in the EWcp indicates specific and indirect neuronal control of the WAT by the EWcp. With this approach it is not possible to tell whether the connections between the EWcp and the IML are direct or indirect and this is a true limitation of the study, because they may take place partially via other autonomic nuclei in the medulla. Another limitation we have to state is that we do not provide direct experimental evidence if the leptin-sensing EWcp/LepRb/Ucn1 are the same cells that EWcp/Ucn1 neurons project to the IML, although we assume, this is the case. Related to this, one may also argue that besides the EWcp other major Ucn1-containing brain areas (e.g. supraoptic nucleus, lateral superior olive; see in ref. Bittencourt et al., 1999) areas may also project to the IML. However, we could not find any published evidence that direct nerve fiber connections would exist between the supraoptic nucleus, the lateral superior olive and the IML (Tucker and Saper, 1985). Nevertheless, together with the fact that Ucn1 fibers innervate the sympathetic outflow area of the thoracic spinal cord (Korosi et al., 2007), support the view that EWcp is one of the brain regions that regulate SNS components (Cano et al., 2021).
In general, increased sympathetic output from the SNS toward WAT increases plasma TG levels by stimulating hepatic VLDL production, and lipolysis in WAT. The latter results in the release of free fatty acids (FFA), which are transported to the liver to fuel the increased synthesis of VLDL. Ablation of LepRb/Ucn1 neurons resulted in significant WAT weight increase due to an increase in adipocyte size. Also, compared to control rats, rats with EWcp/LepRb/Ucn1 neuron ablation had lower plasma TG, lower LDL and HDL, which is probably due to decreased lipolytic activity. Decreased lipolysis leads to a decrease in circulating FFA, which results in a substrate shortage for hepatic very-low-density lipoprotein (VLDL) production, consequently reducing plasma levels of LDL and TG and increased plasma HDL. Taken together, the fat mass and plasma lipid analysis suggest that the ablation of LepRb/Ucn1 neurons in the EWcp leads to an overall decrease in lipolysis, indicating a decrease in the SNS outflow to WAT.
To gain more insights into the regulation of EWcp on adipose metabolism, we further investigated the BAT in the rats with EWcp/ LepRb/Ucn1 ablation. The SNS densely innervates the BAT, and SNSmediated BAT thermogenesis is controlled by various brain areas (Boulant, 2000;Watanabe et al., 1994). Norepinephrine (NE) from SNS terminals stimulates ADRB3-activated BAT thermogenesis ( Bartness and Song, 2007), enhances expression and synthesis of uncoupling protein-1 (UCP1). We found that the ablation of LepRb/Ucn1 neurons resulted in significant BAT weight increase. In addition, rats with EWcp/LepRb/Ucn1 neuron ablation showed an increased expression of ADRB3 mRNA and UCP1 mRNA in the BAT, suggesting BAT activation happens indeed through an increased SNS activity. These findings corroborate that the EWcp/LepRB/Ucn1 cells contribute to the higher-order control of SNS and BAT thermogenesis that is in agreement with our recent finding that local injection of melanocortin 4 receptor agonists activates EWcp/Ucn1 neurons, leading to the rise of energy metabolism and heat loss via increased peripheral vasodilation (Füredi et al., 2017).
When compiling collected data from WAT and BAT, a paradoxical picture emerges at first sight. On one hand, the EWcp seems to be a brain site involved in adipose metabolism regulation through controlling SNS activity. On the other hand, data of ADRB3 and UCP1 induction would suggest increased SNS innervations to the BAT. Although these data may seem paradoxical, when we consider that animals were all sacrificed two weeks after the surgery, we could propose a potential mechanism to resolve this paradox. Specifically, we suggest that the ablation of EWcp/ LepRb/Ucn1 neurons leads to a decreased Ucn1 output towards the SNS. Decreased SNS activity in WAT leads to decreased TG levels and decreased lipolysis resulting in increased WAT size. The concomitant increase in leptin, however, is unable to upregulate Ucn1 production because of the ablation of EWcp/LepRb/Ucn1 neurons. However, leptin would still be able to reach other hypothalamic nuclei to activate the BAT. These results indicate that BAT activation is not primarily regulated by the EWcp/LepRb/Ucn1 neurons, but by the effect of leptin outside the EWcp.

Involvement of EWcp/LepRb/Ucn1 EWcp in energy balance
Maintenance of energy homeostasis and body weight is achieved by an intricate balance between energy intake (food intake) and expenditure. The ablation of LepRb/Ucn1 neurons in the EWcp has no apparent impact on food intake and body weight. Although not significant, these observations suggest an important notion that the effect of EWcp/ LepRb/Ucn1 ablation is independent of leptin/Ucn1 anorexic effects. Earlier studies have found that central administration of Ucn1 results in a potent suppression of food intake (Spina et al., 1996) and increases whole-body oxygen consumption (De Fanti and Martínez, 2002). Interestingly, the fact that within the first week after the surgery, reduced LepRb/Ucn1 expression in the EWcp has led to a transient decrease in respiratory quotient (RQ, indicative of increased fat usage) in the light period; effects that were not seen for the remaining of the experiment, suggests that EWcp/LepRb/Ucn1 neurons modulate specifically WAT lipid mobilization and fat oxidation. In addition to this role of EWcp/LepRb/Ucn1 neurons, recently we have shown that they also contribute to energy intake/expenditure (Füredi et al., 2017).

Conclusions and future perspectives
Internal states, such as e.g. fight-or-flight response, profoundly influence physiology and behavior. A study by Lovett-Barron et al. (2017) identified a set of diverse neuromodulatory systems, including the EWcp governing internal states. The data summarized in this study indicate that WAT-derived leptin binds to the LepRb in the EWcp, increases Ucn1 production in the EWcp, and leads to enhanced sympathetic innervation of WAT. This leads to increased plasma TG levels as well as lipolysis in WAT without altering food intake. Collectively, we propose that the EWcp mediated activation of SNS is required to maintain energy homeostasis under different internal states (e.g. fight-or-flight response; Kozicz et al., 2011 andXu, 2014) in which an organism needs to recruit fuels without being able to eat.

Author contributions
LX and B Geenen performed the histological techniques, PCR experiments, evaluated the results and performed the statistics. LX wrote the manuscript draft. NF, CL performed the animal surgery, metabolic measurements and immunocytochemistry. EP and MB supervised the metabolic experiments, evaluated the results and prepared the corresponding figures. Á D performed the virus injections and evaluated the related results. KJK supervised virus injections and approved the draft manuscript. BG performed and supervised the animal experiments, collected tissue samples, performed the histological techniques and morphometry. TK designed the experiments, supervised data assessment, edited the final version of manuscript which was approved by all authors.