MCT2 Expression and Lactate Influx in Anorexigenic and Orexigenic Neurons of the Arcuate Nucleus

Hypothalamic neurons of the arcuate nucleus control food intake, releasing orexigenic and anorexigenic neuropeptides in response to changes in glucose concentration. Several studies have suggested that the glucosensing mechanism is governed by a metabolic interaction between neurons and glial cells via lactate flux through monocarboxylate transporters (MCTs). Hypothalamic glial cells (tanycytes) release lactate through MCT1 and MCT4; however, similar analyses in neuroendocrine neurons have yet to be undertaken. Using primary rat hypothalamic cell cultures and fluorimetric assays, lactate incorporation was detected. Furthermore, the expression and function of MCT2 was demonstrated in the hypothalamic neuronal cell line, GT1-7, using kinetic and inhibition assays. Moreover, MCT2 expression and localization in the Sprague Dawley rat hypothalamus was analyzed using RT-PCR, in situ hybridization and Western blot analyses. Confocal immunohistochemistry analyses revealed MCT2 localization in neuronal but not glial cells. Moreover, MCT2 was localized to ∼90% of orexigenic and ∼60% of anorexigenic neurons as determined by immunolocalization analysis of AgRP and POMC with MCT2-positives neurons. Thus, MCT2 distribution coupled with lactate uptake by hypothalamic neurons suggests that hypothalamic neurons control food intake using lactate to reflect changes in glucose levels.


Introduction
Neurons within the arcuate nucleus (AN) of the hypothalamus are capable of responding to changes in glucose and lactate concentrations [1,2]. These neurons couple blood glucose fluctuations with a complex network of neurochemical and neurophysiologic responses that control energy expenditure and feeding behavior [3]. The hypothalamic melanocortin pathway plays a critical role, as it is directly regulated by hormones and nutrients [4]. The melanocortin pathway is formed from neuronal populations localized in the AN with antagonistic actions. Orexigenic neurons that synthesize neuropeptide Y (NPY) and agouti-related peptide (AgRP) increase food intake and reduce energy expenditure. Contrary effects were observed when anorexigenic neurons, including proopiomelanocortin (POMC) neurons, are stimulated [3]. On the other hand one population within the hypothalamus, glucose-exited neurons (GE), increases its firing rate in response to elevated glucose levels. Alternative, neurons that reduced their firing rate in response to increased glucose levels are known as glucose-inhibited neurons (GI) [5,6]. A high percentage of GI neurons are immunoreactive for NPY [7]. In contrast, it has been proposed that POMC neurons could be correspond to GE neurons [8,9].
AN neurons are not in direct contact with blood or cerebrospinal fluid, and therefore, cannot directly detect and respond to changes in glucose concentration [5,10,11,12,13].
Previous studies have proposed that the hypothalamic glial cells (i.e., tanycytes) release lactate to neighboring neurons through monocarboxylate transporters (MCT) in response to high glucose, resulting in neuronal activation [11,12,14,15,16,17]. Furthermore, dynamic bioluminescence imaging, which records cytosolic ATP ([ATP] c ) in real-time, revealed no significant changes in [ATP] c (,2%) in response to high glucose (15 mM) in hypothalamic neurons, contrary to expected results. However, these neurons responded to lactate with a significant increase in [ATP] c [14], suggesting that lactate but not glucose induces a neuronal response in high glucose conditions. The participation of lactate in the glucosensing mechanism and feeding behavior is supported by a study by Lam et al. [18] in which hypothalamic injection of lactate mimicked the effects of hypothalamic glucose administration. Moreover, GE neurons located in the VMH respond to lactate (15 mM) [2].
Tanycytes release lactate through MCTs, and b1 ventral tanycytes (b1v), which contact an orexigenic area of AN, express MCT1 [10]. MCT4 distribution in the processes of b1 dorsal tanycytes (b1d) contacting the anorexigenic zone of AN was also observed [11]. Moreover, the selective distribution of MCTs between neurons and glial cells suggests that lactate plays a key role in brain energy metabolism, and that MCT2 is essential for neuronal lactate uptake [19,20,21,22,23]. However, the expression of MCTs in hypothalamic AN neurons in normal animals has not been analyzed to date.
In this study, the expression and function of MCT2 in hypothalamic neurons was analyzed. Using primary hypothalamic cell cultures and the hypothalamic neuronal cell line, GT1-7, incorporation of lactate was observed through MCT2. Moreover, RT-PCR, in situ hybridization, Western blot, and immunohistochemistry analyses demonstrated MCT2 localization in orexigenic and anorexigenic neurons. The distribution of MCTs in the hypothalamus suggests that neuronal and glial cells interact through lactate to regulate the activity of neurons that control food intake.

Ethics Statement
All animals were handled in strict accordance with the Animal Welfare Assurance (permit number 2010101A), and all animal work was approved by the appropriate Ethics and Animal Care

Primary Cell Culture
A mixed primary culture of hypothalamic neurons and glia was prepared from Sprague-Dawley rats at 19 days of gestation and maintained in culture at 37uC and 5% CO 2 . Briefly, the hypothalamic region of the brain was microdissected at 4uC in buffer containing 10 mM HEPES (pH 5 7.4; 340 mOsm/L). Subsequently, the tissue was subjected to enzymatic desegregation for 10 min at 37uC in 0.25% trypsin (Invitrogen, Rockville, MD, USA) and EDTA 0.20% (Sigma-Aldrich, St. Louis, MO, USA). After the tissue was then transferred to MEM (Invitrogen) with 10% (v/v) fetal bovine serum (FBS) (Thermo Fisher Scientific Inc., Waltham, MA, USA) and 2 mg/mL DNAse I (Sigma-Aldrich), mechanical disaggregation was performed. The cells (,5610 5 cells/cm 2 ) were plated on coverslips 18 mm in diameter that were coated with 0.2 mg/mL poly-L-lysine (Sigma-Aldrich) in 12-well plates (Corning Costar, NY, USA). After 30 min, Neurobasal medium supplemented with B27, 2 mM glutamine, 100 U/mL penicillin, 100 mg/mL streptomycin, and 2.5 mg/mL Fungizone (all Invitrogen) was added. After 48 h, the cultures were treated with 15 mg Ara-C (Sigma-Aldrich) and maintained in culture for 7 days until subsequent immunocytochemistry and fluorimetric analyses.

Immortalized Cell Culture
The mouse hypothalamic tumor cell line, GT1-7 [24] (kindly provided by Dr. P. Mellow) was maintained at 37uC and 5% CO 2 on Petri dishes (Corning Costar) in high glucose DMEM supplemented with 10% FCS, 3.75 mM glutamine, and 2.5 mg/ mL Fungizone (all Invitrogen), according to [24]. The dishes with the highest density of confluent cells were expanded (1:5) and used for immunocytochemistry, lactate uptake, and immunoblot experiments.

Immunocytochemistry
Cells (5610 5 cells/cm 2 ) were grown on 0.2 mg/mL poly-Llysine-coated glass cover slides (Sigma-Aldrich) in 24-well plates and fixed with 4% paraformaldehyde in PBS for 30 min. The immunocytochemistry was performed as previously described [11], using the following primary antibodies: mouse anti-MAP2 The slides were analyzed using confocal laser microscopy (D-Eclipse C1 Nikon, Tokyo, Japan).

Fluorimetric L-Lactate Analyses
Changes in intracellular pH (pHi) in hypothalamic neurons were visualized using fluorescence measurements of intracellular 29,79-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein acetoxymethyl ester (BCECF-AM). Cells were incubated in cell culture medium with 5 mM BCECF-AM (Invitrogen) for 30 min at 37uC in a 95% O 2 -5% CO 2 atmosphere. After loading the probe, the   To inhibit the main H + transport system in neurons, dicyclohexylcarboxiimide, a nonspecific H + -ATPase inhibitor, was added to the cells 20 min before the pHi measurements [25]. The excitation fluorescence ratio of 495/440 nm was determined using an emission wavelength of 535 nm in a Ti microscope (Nikon) conditioned with search filters and D495/10X D440/10X excitation, emission D535/25m and dichroic 515dcxr (Chroma Technology Corp., Bellows Falls, VT, USA) and a CoolSNAP HQ2 camera (Photometrics, Tucson, AZ, USA). After a basal fluorescence ratio was obtained 10, 20 and 50 mM of L-lactate were added to the cell suspension, and the fluorescence ratio was recorded until it reached a new steady-state level. All experiments were performed in triplicate, and results represent an average of at least two separate experiments.

In situ Hybridization
Using sense 59-GTC TCA TCT CCG AAT CAG TGT TCA G-39 and antisense 5-CCC GTT ACT CAG TGT TTG CGT G-39 primers, a PCR product of 469 bp was obtained from the hypothalamus and subcloned into pCRH-4-Blunt-TOPOH (Invitrogen) to generate sense and antisense digoxigenin-labeled riboprobes. RNA probes were labeled with digoxigenin-UTP by in vitro transcription with T3 or T7 RNA polymerase following the manufacturer's procedure (Invitrogen). In situ hybridization was performed on rat frontal brain sections of 7 mm mounted on poly-L-lysine-coated glass slides [26]. The sections were baked at 60uC for 1 h, deparaffinized in xylene, and rehydrated in graded ethanol. Following incubation with 1 mg/ml proteinase K (Sigma) in PBS (pH 7.4) for 5 min at 37uC, the tissue sections were fixed with 4% paraformaldehyde at 4uC for 5 min, washed in cold PBS, and then acetylated with 0.1 M triethanolamine-HCl (pH 8.0) and 0.25% acetic anhydride at room temperature for 10 min. After a brief wash with PBS, the sections were incubated in prehybridization solution (Novagen) at 37uC for 30 min, and then 25 ml of hybridization mix (50% formamide, 0.6 M NaCl, 10 mM Tris-HCl [pH 7.5], 1 mM ethylenediaminetetraacetic acid [EDTA], 16 Denhart's solution, 10 mM DL-dithiothreitol [DTT], 500 mg/mL yeast tRNA, 50 mg/ml heparin, 500 mg/ml DNA carrier, and 1:25 dilution of riboprobe) were added to each slide. The slides were covered with glass coverslips and placed in a humidified chamber at 42uC overnight. After removal of the coverslips, the slides were rinsed in 46 saline-sodium citrate (SSC) and washed twice for 15 min at 42uC. The slides were washed twice at 37uC for 15 min each in 26SSC, 16SSC and 0.36SSC. Visualization of digoxigenin was performed by incubation with a monoclonal antibody coupled to alkaline phosphatase (antidigoxigenin alkaline phosphatase Fab fragments diluted 1:500; Roche Applied Science) at room temperature for 2 h. Nitroblue tetrazolium chloride and 5-bromo-4-chloro-3-indolyl-phosphate (Roche Applied Science) were used as substrates for the alkaline phosphatase. Controls included use of the sense riboprobe and omission of the probe.

Intracerebral Ventricular (i.c.v) Injection of Colchicine
Rats were anesthetized with an intraperitoneal injection mix of ketamine-xilazine (90 mg/Kg-10 mg/Kg). Using a cannula (28 gauge), 10 mL of 10 mg/mL colchicine (Sigma-Aldrich) was injected into the third ventricle (AP 23.14 mm, ML 0.0 mm, DV 9.2 mm). After 20 h, rat brains were fixed for vascular perfusion using 4% PFA and processed for cryosectioning into 30mm sections. The samples were used for subsequent immunohistochemical detection of neuropeptides.

Immunohistochemistry
Immunohistochemical analyses was performed as previously described [10]

Image Analyses
To analyze MCT2/vimentin, MCT2/GLUT1, MCT2/GFAP, vimentin/GLUT1 and vimentin/GFAP co-localization within various regions of interest in the hypothalamic region, NIS-Elements software (Nikon, Nikon Instruments INC), was used, and Pearson's coefficient (Rr) values were calculated. This coefficient measures the overlapping level between the pixels of two fluorescent channels, ranging from +1 to 21 (0-100% colocalization) [27]. The statistical analysis was performed compar-ing the Rr calculated between the glial markers (control) with the Rr calculated between the control and the protein of interest.
Quantification of POMC-and AgRP-positive neurons to MCT2 was performed using ImageJH software (National Institute of Health, Bethesda, Maryland, USA). Each image was analyzed as follows. A Gaussian Blur filter of 1 pixel size was applied to the POMC and AgRP channels. Using a threshold of 75, the significant intensity areas over 190 pixels were selected. Subsequently, only the areas with an average intensity greater than 130 were chosen and selected as POMC or AgRP cells. After that, only those areas whose average MCT2 channel intensity was over 1 and whose minimum MCT2 intensity was less than 78 were selected for analysis, excluding the cells out of the region of interest. The MCT2 intensity inside every POMC and AgRP cell was calculated and only those with a sum over 12000 were chosen as MCT2-positive cells. For cells that were within clusters, the  The brain of each rat and brain of mouse were removed, and the hypothalamic area or medulla renal was isolated and further dissected. Total RNA from hypothalamus, control tissues, or cell cultures were isolated using Trizol (Invitrogen) and treated with DNase I (Fermentas International INC

Statistical Analysis
All values are expressed as the means 6 SEM. Statistical comparisons between groups were performed using one-way ANOVA followed by Fisher's test among more than three groups or the unpaired Student's t-test for two groups. All analyses were performed using GraphPad Prism 4.0 Software (GraphPad Software Inc., San Diego CA).

Functional Characterization of MCT2 in Primary Hypothalamic Cultures and in the GT1-7 Cell Line
Tanycytes within the AN release lactate through MCT1 and MCT4 in response to glucose [11]. Therefore, the uptake of lactate by hypothalamic neurons was evaluated; MCT expression in these cells was also assessed. Mixed primary cultures enriched in hypothalamic neurons were assessed for neuronal markers, such as MAP-2 (Fig. 1a, arrows), neurofilaments (Fig. 1d, arrows), and N-  (Fig. 1g, arrows), and the tanycyte marker, vimentin (Fig. 1b,  e and h, arrows), by immunocytochemistry. There were two distinguishable cell populations that did not co-express these markers, neurons and tanycytes (Fig. 1c, f and i, arrows). The coexpression of the neuropeptides, AgRP, CART, POMC and NPY, with the neuronal marker, MAP-2, was also analyzed (Fig. 2a-h); in vitro hypothalamic neurons expressed the orexigenic neuropeptides, AgRP ( Fig. 2a and e, arrows), the anorexigenic neuropeptides CART (Fig. 2b and f, arrows), and POMC ( Fig. 2c and g,  arrows). However, NPY was not detected (Fig. 2d and h, arrows) in the cell culture.
Lactate incorporation via MCTs was also analyzed in these mixed hypothalamic primary cultures using fluorimetric assays. Since MCTs are monocarboxylate/H + symporters, the in vivo uptake of lactate was undertaken in the presence of H + -ATPase inhibitors and in the absence of bicarbonate. The response of the neurons to the addition of 20 mM L-lactate was detected as a decrease in the ratio of fluorescence that stabilized a few seconds after lactate was added (Fig. 3g), indicating that hypothalamic neurons incorporate lactate. However, using concentrations below 20 mM of L-lactate, reproducible changes in the fluorescence ratio were not observed (data not shown); therefore, this technique did not permit functional characterization of the MCT isoform present in this culture. To resolve this problem, MCT expression and lactate uptake was subsequently analyzed in the hypothalamic neuron cell line, GT1-7 [24,28].
The expression of MCT1, MCT2 and MCT4 in GT1-7 cells was analyzed using RT-PCR with mouse-specific primers and conditions optimized using RNA from mouse hypothalamus. The amplified cDNA bands were 400 and 476 and 369 bp, which are the expected sizes for MCT1, MCT2 and MCT4, respectively (Fig. 4a, lanes 1, 5 and 9). GT1-7 cells expressed MCT1 and MCT2 but not MCT4 (Fig. 4a, lanes 2, 6 and 10). No amplification product was observed in samples without reverse transcriptase (Fig. 4a, lane 3, 7 and 11) or water (Fig. 4a, lane 4, 8  and 12), indicating the absence of genomic DNA or contamination in the PCR reaction. MCT1 and MCT2 expression was also demonstrated using immunoblot analysis of proteins isolated from rat hypothalamus (positive control) (Fig. 4b, lanes 1 and 4) and GT1-7 cells (Fig. 4b, lanes 2 and 5). Analysis of hypothalamus protein extract using anti-MCT preabsorbed with inductor peptides showed an absence of bands (Fig. 4b, lanes 3 and 6). Therefore, RT-PCR and Western blots analyses confirmed the expression of MCT1 an MCT2 in GT1-7 cells.
Transport assays revealed a dependence on pH for lactate uptake in GT1-7 cells. At a substrate concentration of 0.1 mM, lactate uptake at pH 5.0 was more than five times faster than that observed at pH 8.5 when the assays were performed at 20uC (Fig. 4c). The proton effect was found to be of a non-cooperative nature, which was corroborated by a Hill plot that yielded a straight line with a slope (Hill coefficient) of 0.48 (Fig. 4d). To determine whether other monocarboxylates might also serve as substrates for the lactate transporter in GT1-7 cells, competition experiments were performed. At an extracellular concentration of 0.1 mM lactate (20uC, pH 7.0), transport was strongly inhibited by 10 mM pyruvate (Fig. 4e). Incubation with MCT inhibitors, including 1 mM phloretin, 5 mM 4-CIN and 1 mM DIDS, decreased lactate uptake up to 70%, 40% and 50% respectively. However, 3 mM pCMBS did not significantly inhibit lactate uptake (Fig. 4e). The uptake of 0.1 and 25 mM lactate at 4uC was almost linearly correlated up to 60 s ( Fig. 5a and b). Using these optimized conditions, the basic transport parameters for lactate uptake in GT1-7 cell culture were determined. Dose-response studies performed up to 30 s revealed that lactate transport was saturated at concentrations above 15 mM (Fig. 5c). Data analysis suggested the presence of two kinetic components for lactate transport. When the data was transformed and plotted according to Hanes-Woolf, this observation was confirmed (Fig. 5d). Transformation showed apparent Km values of 0.5 mM and 6 mM and V max values of 1.5 (nmol/min per 10 6 cells) and 15 (nmol/min per 10 6 cells), corresponding to MCT2 and MCT1 respectively.

MCT2 is Expressed in the Hypothalamus
The expression of MCT2 in rat hypothalamus was analyzed by RT-PCR with specific primers (Fig. 6a). The amplified DNA band was approximately 416 bp, which was the expected size for the MCT2 amplification product (Fig. 6a, lanes 2). The PCR conditions were optimized using RNA from the cerebral cortex as a positive control for MCT2 expression [28] (Fig. 6a, lane 1). No amplification product was observed in samples without reverse transcriptase (Fig. 6a, lane 3) or water (Fig. 6a, lane 4). Immunoblot analysis using protein extracts isolated from the cerebral cortex (positive control) and hypothalamus showed MCT2 expression (45 kDa band) in both tissues (Fig. 6b, lanes 1  and 2). Bands were absent in samples incubated with anti-MCT2 preabsorbed with inductor peptides (Fig. 6b, lanes 3). MCT2 mRNA expression was also analyzed using in situ hybridization with digoxigenin-labeled cRNA probes specific for MCT2 (Fig. 6cg). A positive hybridization signal was observed in neurons of the entorhinal cortex (Fig. 6d, arrow head), in the ependymal layer of lateral ventricle (Fig. 6e, arrow head) and in some choroid plexus cells (Fig. 6e, arrow). In the basal hypothalamus (Fig. 6c), the reaction was concentrated in the AN neurons, principally in the peri-ventricular (Fig. 6g, arrow head) and lateral regions of the AN (Fig. 6h, arrows head). Negative hybridization was observed using the sense probe ( Fig. 6 inset in c-g).

MCT2 is Localized in Hypothalamic Neurons
The high specificity of the antibodies used in this study was evaluated in control tissues and several cerebral areas (Fig. S1) [19,20,23,29,30,31,32]. Anti-MCT2 reactivity was localized in the kidney (Fig. S1a) and brain (Fig. S1b-h). Immunohistochemical studies revealed MCT2 reactivity in some collecting ducts of the kidney medulla (Fig. S1a, arrows) and in brain endothelial cells (Fig. S1b, arrows). Co-distribution analyses performed with anti-MCT2 and anti-GLUT1 in the brain cortex showed MCT2 immunoreaction in astrocytes that surrounded GLUT1-positive vessels in several regions of the brain cortex and in the marginal zone ( Fig. S1c-e, arrows). Co-distribution was only detected in microvessels (Fig. S1c-d, arrowheads). Intense immunoreaction was detected in ependymal cells of the lateral ventricle ( Fig. S1f-g,  arrows) and subependymal astrocytes of the dorsal third ventricle (Fig. S1h, arrows).
In the hypothalamus, MCT2 localization was analyzed using two tanycytes markers: vimentin and GLUT1 (Fig. 7b and d).
Intense MCT2 reactivity was detected in the peri-ventricular and laterals AN regions ( Fig. 7c and d, arrows), excluding the cellular bodies of tanycytes (schematic representation, Fig. 7a) that form the ventricular wall (Fig. 7c, arrowheads). A positive reaction was also detected in astrocytes of the median eminence and in the terminal processes of b1d tanycytes negative for vimentin ( Fig. 7cd, asterisks). Using high magnification images, MCT2 hypothalamic cellular localization was defined (Fig. 7e-p). In the dorsal area of the peri-ventricular AN, MCT2 was detected in the hypothalamic parenchyma surrounding neuronal bodies (Fig. 7e,  arrows), and co-localization with vimentin or GLUT1 was not observed (Fig. 7f-h, arrows). Co-distribution of vimentin and GLUT1 was detected only within the cellular bodies of a-tanycytes (Fig. 7h, triple arrow). In the AN ventral area, MCT2 immunoreactivity was concentrated in the peri-ventricular zone that surrounds neuronal bodies (Fig. 7i, arrows). As in the dorsal area, no co-localization between MCT2 and b1v tanycytes markers was detected (Fig. 7j-l, arrows). In the lateral AN, MCT2 formed a cellular net around the neuronal bodies (Fig. 7m,  arrows), which did not express vimentin or GLUT1 (b1d tanycytes markers) (Fig. 7n-p, arrows). To corroborate the absence of colocalization between MCT2 and the glial markers, Rr values were calculated. The Rr value estimated for MCT2/vimentin was 20.012 while that for MCT2/GLUT1 was 20.027, which were significant different from the Rr value obtained for vimentin/ GLUT1 control markers (0.564; p,0.0001), confirming the observed results. In order to corroborate MCT2 neuronal localization, the ventral peri-ventricular area of the AN was analyzed using multilabeling analysis of high magnification images (Fig. 8). b1v tanycyte projections were detected using antivimentin (Fig. 8a, arrows) and anti-GFAP (Fig. 8b, arrows). MCT2 was located in the hypothalamic parenchyma associated with cellular bodies and processes (Fig. 8c, arrows). No colocalization of MCT2, vimentin, and GFAP was observed, corroborating MCT2 neuronal localization in both neuronal soma (Fig. 8e, arrows) and processes (Fig. 8e, arrow heads). The Rr value estimated for MCT2/Vimentin co-localization was 20.076 and that for MCT2/GFAP co-localization was 20.040, which was significantly different from the value observed for the glial vimetin/GFAP control markers (0.192, p,0.0002).

MCT2 is Localized in Orexigenic and Anorexigenic Arcuate Nucleus Neurons
MCT2 distribution in orexigenic and anorexigenic AN neurons was evaluated in rostro-caudal rat brain slices, treated with colchicine, using anti-POMC and anti-AgRP antibodies (Fig. S2). POMC neurons are generally located distal to the third ventricle with AgRP neurons distributed proximal to ventricle, dividing AN into anorexigenic (lateral) and orexigenic (ventricular) areas ( Fig.  S2b-f). However, the rostral AN sub-population of POMC neurons are close to the third ventricle (Fig. S2b-c), generating a sub-compartment that contains both anorexigenic and orexigenic neurons. This neuronal distribution is reduced towards the caudal area; in the AN medial area, few POMC neurons are found within the same area as AgRP neurons (Fig. S2d). Consistently, in the caudal zone of the AN, these two populations are completely segregated (Fig. S2e-f). As shown in Figure 9, AgRP and POMC localization was analyzed in the rostral (Fig. 9a), medial (Fig. 9b) and caudal (Fig. 9c) areas. MCT2 was principally located proximal to the third ventricle surrounding primary AgRP neurons ( Fig. 9a1-c1, arrows) and to a lesser degree POMC neurons ( Fig. 9a1-c1, arrow heads). MCT2 was localized in neuronal membranes, and its localization was not affected by colchicine treatment (data not shown). In a more detailed analysis, changes in the localization of MCT2 from the rostral to caudal region of hypothalamus were detected. In the rostral area, MCT2-positive neurons were located in a similar proportion to AgRP and POMC neurons (95 v/s 85%) (Fig. 9a2-a3, arrows). On the contrary, in the medial AN, more AgRP than POMC neurons positive for MCT2 were observed (90 v/s 70%) (Fig. 9b2-b3, arrowheads). In the caudal zone of the AN, there were less MCT2-reactive POMC neurons, contrary to that observed for AgRP ( Fig. 9c2-c3, arrows) (50 v/s 81%). Therefore, these analyses confirm the localization of MCT2 in neurons that control feeding behavior.
Because NPY and CART neurons also participate in controlling feeding behavior and are also present in the AN, MCT2 distribution in these neurons was assessed using anti-NPY and anti-AgRP (Fig. 10) or anti-POMC and anti-CART (Fig. 11) antibodies in the AN medial area. Analyses of orexigenic markers showed that a high number of neurons located in the periventricular AN co-express NPY and AgRP neuropeptides (Fig. 10a-d). High MCT2 reactivity was found surrounding orexigenic neurons in both the dorsal (Fig. 10e-h, arrows and inset) and ventral area of the peri-ventricular AN (Fig. 10i-l, arrows and inset). MCT2 distribution using anorexigenic markers in the same area was also undertaken (Fig. 11a-d). High magnification of the AN dorsal area revealed a low number of neurons that co-express POMC and CART neuropeptides ( Fig. 11e-h, arrows), which is contrary to that observed in ventral area where a major number of neurons co-express both neuropeptides ( Fig. 11i-l, arrows). However, MCT2 was localized in dorsal ( Fig. 11e-h, arrows) and ventral ( Fig. 11i-l, arrows) neurons. Furthermore, intense MCT2 expression was observed in another neuronal type that was not reactive to POMC or CART antibodies ( Fig. 11h-l inset) or AgRP/NPY (Fig. 10, asterisk), suggesting that other neuronal type expressed this transporter.

Discussion
Several neuronal populations capable of responding to changes in glucose concentration are located within the AN. These neurons couple the fluctuations in blood glucose levels with a complex network of neurochemical and neurophysiologic responses that control food intake and feeding behavior [3]. However, the mechanism by which these neurons detect changes in glucose levels it not clear. Several studies have proposed that the glucose sensing mechanism is driven by metabolic interactions between glial and neural cells that are mediated by lactate and MCTs [11,12,14,15,18]. MCT2 expression was demonstrated in the neurons of the cortex, hippocampus and cerebellum [19,20,33], but its function has yet to be analyzed in these cells. Because the expression and function of MCTs in hypothalamic neurons has not been determined, this study analyzed uptake of lactate through MCTs in AN neurons, showing that MCT2 is localized in neurons that control food intake.
Expression of functional MCTs was analyzed in hypothalamic neuroendocrine neurons, a mixed primary cell culture, and an immortalized hypothalamic cell line (GT1-7) in the present study. In primary cultures, MCT2 was expressed in neuroendocrine neurons; however, limitations of the fluorimetric assays prevented further functional characterization of the MCTs expressed in these cells. Lactate uptake was observed in hypothalamic neurons in vivo. Because neurons cultured in vitro induced the expression of MCT1 and MCT4, transporters with high transport capacity (Km 6 and 55 mM, respectively) [34,35], their expression may mask MCT2 (Km 0.5 mM) [21,36]. Therefore, the kinetic characterization of MCT2 was performed in GT1-7 cells using radiometric methods. RT-PCR and Western blot analyses confirmed the expression of MCT1 and MCT2 in these cells, demonstrating that this immortalized cell culture is a suitable model of neuroendocrine neurons to evaluate MCT function. Transport and competition assays revealed that GT1-7 cells express functionally active MCT1 and MCT2. The lower affinity transport component had the expected properties of MCT1, with a transport Km of 6 mM [34,37]. In contrast, the higher affinity transporter revealed an apparent Km of approximately 0.5 mM, which fits the description of MCT2 [21,36]. Furthermore, increased lactate uptake was observed with lower pH values, which is similar to that previously reported for MCT1 and MCT2 [21,34,36,37]. Lactate transport was strongly inhibited by 10 mM pyruvate as both isoforms transport this substrate [21,34,37,38]. Additionally, 1 mM floretin decreased lactate uptake by 70%, according with the IC 50 reported for MCT1 (30 mM) and MCT2 (15 mM) [21,34,37]. 4-CIN also strongly inhibited lactate uptake, which is in accordance with previous findings [21,34,37]). In addition, a smaller but significant inhibitory effect on lactate uptake was observed with 1 mM DIDS, which was similar to that observed in MCTexpressing oocytes [21,37]. However, 3 mM pCMBS, an inhibitor of MCT1 but not MCT2 [21,34], did not significantly inhibit lactate uptake, indicating that MCT2 is the preferred transporter for lactate incorporation in GT1-7 neuronal cells.
The in vitro expression and functional analyses of the present study indicated that MCT2 is used to incorporate lactate in the hypothalamic neurons. However, only one study demonstrated in situ MCT2 expression in the hypothalamic neurons of rats fed a high fat diet, which induced MCT2 over-expression in several hypothalamic nuclei including the AN [39]. The present study is the first to demonstrate MCT2 expression in hypothalamic AN neuroendocrine neurons in rats, under normal dietary conditions, supporting the idea of a glucose sensing mechanism driven by glianeuron interactions through lactate.
RT-PCR and immunoblot analyses revealed that MCT2 is expressed in hypothalamic tissue, which was confirmed by in situ hybridization and immunohistochemical analyses. Both studies showed high MCT2 expression in hypothalamic neurons localized in the peri-ventricular and lateral regions of AN. However, several neuronal populations with antagonistic actions on the food intake, namely orexigenic and anorexigenic neurons, are localized within the hypothalamus [4]. In the present study, detailed immunohistochemical analyses showed that MCT2 is localized in periventricular and distal AN neurons in both cellular populations. However, this distribution depends on the hypothalamic area examined. For example, a similar number of orexigenic and anorexigenic neurons reactive to MCT2 were observed in the rostral area of the basal hypothalamus. In contrast, in the more caudal areas, MCT2 was primarily localized within NPY and AgRP neurons. This change in MCT2 distribution in orexigenic and anorexigenic AN neurons is correlated with the localization of glucose-inhibited (GI) and glucose-exited (GE) neurons [40]. NPY neurons are GI neurons, which reduce their electrical activity in response to high glucose [7], but the electrical-glucose identity of anorexigenic neurons (POMC and CART) has not been completely elucidated [8,9,41]. Electrophysiology studies performed in transgenic mice indicate that these neurons could be corresponding to GE neurons [8,9]. Thus, neuroendocrine neurons may couple the neuronal activity with changes in glucose concentration and feeding behavior. Therefore, AN may represent an anteroposterior compartment in which MCT2 localization generates metabolic niches for orexigenic (GI) and anorexigenic (GE) neurons, controlling food intake and feeding behavior (discussed later).
Several studies support the idea that lactate is a key signal in hypothalamic glucose sensing and food intake [18,42,43,44,45]. Furthermore, lactate perfusion of the VMH suppresses the hypoglycemic counter-regulatory response, with a strong diminution in glucagon and epinephrine release [42]. Moreover, central inhibition of lactate dehydrogenase by administration of oxamate abolishes the effects of ICV-glucose on blood glucose lowering. Thus, blocking lactate metabolism in the hypothalamus results in a 40% reduction in the inhibitory action of glucose on circulating glucose levels [18]. Besides, ICV injection of lactate reduces blood glucose levels and food intake, resulting in body weight loss in eight-week-old rats [45]. Interestingly, the inhibition of lactic dehydrogenase or K ATP channels in the hypothalamus increases glucose blood levels in the presence of systemic lactate [43]. These findings provide evidence for both a direct intracellular mechanism of K ATP channel regulation, which does not involve increased [ATP] c [46], as well as a separate intercellular signaling mechanism mediated by lactate released from neighboring glial cells.
Based on the results presented here and our previous studies, we postulate that neuron-glial interactions through lactate regulate the activity of neurons that control food intake and feeding behavior. The specific localization of tanycytes in direct contact with cerebrospinal fluid [10] coupled with GLUT2/GK expression strongly [12,15] supports the idea that these cells have a high capacity to uptake glucose. Furthermore, our recent work has shown that tanycytes have the capability to increase intracellular calcium levels in response to high glucose [16], demonstrating that these cells can respond to glucose. Furthermore, tanycytes can release lactate through both MCT1 and MCT4 [11]. The coincident expression of MCT2 in orexigenic neurons with the localization MCT1 in b1v-tanycytes distributed in the periventricular area [11] led us to propose that these glial cells regulate the activity of GI anorexigenic neurons. We postulate that lactate released by tanycytes, but not glucose itself; reduces the firing rate of these neurons in hyperglycemic conditions. Moreover, the localization of MCT4 in b1d-tanycytes suggests that these cells could couple with anorexigenic neurons located in the lateral AN through MCT2. In high glucose, these neurons uptake the lactate released by tanycytes, increasing their firing rate and generating satiety through two feasible pathways. The first pathway is mediated by anorexigenic neuropeptides, including a-MSH (POMC-derived), from GE-anorexigenic neurons [8,9]. The second alternative is an indirect pathway where GE and anorexigenic neurons are two different cells [41], and therefore the GE neuron regulates the secretion of anorexigenic neuropeptides.
In conclusion, we postulated that lactate has a dual role in the control of the feeding behavior, depending on the subtype of neuronal and glial cells activated in the process. This evidence suggests an important rol for tanycytes in the glucosensing mechanism driven by lactate for indirectly controlling the neuronal function. Figure S1 MCT2 is localized in kidney and brain tissue. (a) Immunohistochemical localization of MCT2 in the basolateral region of the collecting tubules of the renal medulla (arrows). (b) MCT2 is expressed in capillaries in the cerebral cortex (arrows). (c-h) Co-distribution of MCT2 (green) and GLUT1 (red) in the CNS. MCT2 is localized in astrocytes surrounding blood vessels positive for GLUT1 (c-d, arrows), which form the marginal glia (e, arrows); co-distribution was only detected in the microvasculature (c-d, arrow heads). An intense MCT2 reaction was detected in the lateral ventricle ependymal cells (f-g, arrows) and subependymal astrocytes (h, arrows). Absence of co-localization with GLUT1 was observed in the areas analyzed. CD: collecting ducts, CP: choroid plexus, LV: lateral ventricle, IIIV: third ventricle. Scale bar: 50 mm.