A brainstem–hypothalamus neuronal circuit reduces feeding upon heat exposure – Nature

    A brainstem–hypothalamus neuronal circuit reduces feeding upon heat exposure – Nature


    Experimental procedures on mice conformed to the 2010/63/EU directive and were approved by the Austrian Ministry of Education, Science and Research (66.009/0145-WF/II/3b/2014 and 66.009/0277-WF/V3b/2017). All procedures were planned to reduce suffering, as well as mouse numbers. Mice were kept under standard housing conditions (12 h:12 h reverse light:dark cycle with light on at 22:00 and off at 10:00, 25 °C), with food and water available ad libitum. For acute thermal manipulations, ex vivo electrophysiology, neuroanatomy, and behavioural tests, C57Bl6/J mice were used. Raxtm1.1(cre/ERT2)Sbls/J mice (Rax-CreERT2; JAX 025521) were crossed with B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J (referred to as Ai14; JAX 007914), B6;129S6-Polr2aTn(pb-CAG-GCaMP5g,-tdTomato)Tvrd/J (PC-G5-tdT; JAX 024477) or B6;129S-Slc17a6tm1.1(flpo)Hze/J (Slc17a6-IRES2-FlpO-D; JAX 030212) at ages between postnatal days (P)60-90. B6;129P2-Mapttm2Arbr/J (referred to as TaumGFP-loxP; JAX 021162) were used for transsynaptic labelling. Agrp+ neurons were visualized by crossing Agrptm1(cre)Lowl/J (JAX 012899) and Ai14 reporter mice. Mice of both sexes were used for experiments, as indicated. For primary cultures of tanycytes, both male and female Wistar rats were used. To sample the CSF, male Wistar rats were cannulated, as approved by the Ethical Review Board of Semmelweis University (PE/EA/1234-3/2017, Hungary).

    Acute heat exposure

    C57Bl6/J mice of both sexes aged P60–P100 were housed individually and habituated in an Aria BIO-C36 EVO incubator (Tecniplast) at 25 °C with a reverse 12 h:12 h light:dark cycle (light on at 22:00) with 42% humidity for 3 days. One day before acute thermal challenge, the temperature of an MIR-254 incubator (Sanyo) was set to the relevant target temperature. To maintain humidity in the incubator, a Becher glass filled with 1 l water was placed in the incubator. Humidity (~42%) and CO2 levels (~396 ppm) were continuously measured with a CO100 CO2 monitor (EXTECH Instruments). At 09:00 on day 4 (that is, 1 h before the beginning of the dark (off) phase of the light cycle), the mice were placed in new experimental cages without food and water, exposed to 25 °C for 1 h, and then returned to their home cages. On day 5 (09:00), mice were again placed in experimental cages without food and water, and then exposed to either 4 °C or 40 °C for 1 h. Subsequently, the mice were returned to their cages in an Aria BIO-C36 EVO incubator (Tecniplast) set at 25 °C.

    To record skin temperature, C57Bl6/J male mice were singly housed, with their interscapular area above the main brown fat depot shaved 2–3 days prior to the experiments. Mice were then exposed to 40 °C for 1 h. Control mice were kept at 25 °C. Body temperature was recorded at both the interscapular area and the perianal region of each mouse using an infrared thermometer60,61 (DET-306, Femometer). Baseline temperature was acquired 15 min prior to the thermal challenge, followed by switching them to a thermo-controlled chamber (Memmert, MEMM-OT3007S) set to 40 °C, and left undisturbed for 1 h. Temperature recordings resumed at intervals of 15 min for another 180 min after heat exposure ended, with the mice returned to their home cages.

    Measurement of food intake and body weight

    Food pellets and mice were measured on an Entris II Essential line scale with 0.01 g accuracy (Sartorius, 1000059011) to determine food intake and body weight, respectively. Baseline parameters were determined 1 h prior to thermal manipulation. In select experiments, food pellets were weighed 2 h (12:00), 4 h (14:00) and 24 h (09:00) after thermal challenge.

    In multiparametric experiments (Fig. 5h,i and Extended Data Fig. 10i–k), food and fluid intake, as well as horizontal movement were simultaneously recorded by using PhenoTyper cages (Noldus). Herein, food intake was approximated by recording the time spent to consume food when the infrared beam within the pellet dispenser was interrupted by the nose-pokes of the mice (Δt). The same technical setup was used to measure the time spent to drink. Data were analysed by Ethovision XT15 (Noldus).


    For immunofluorescence labelling, mice were anaesthetized with isoflurane and transcardially perfused with ice-cold phosphate buffer (PB) (0.1 M, pH 7.4) followed by ice-cold paraformaldehyde (4% in 0.1 M PB). Subsequently, the brains were removed and kept in the same fixative at 4 °C overnight. Next, the brains were washed with 0.1 M PB and stored with 0.025% NaN3 as antifungal agent at 4 °C until processing. Fifty-micrometre-thick coronal sections spanning the ARC and PBN were cut on a vibratome (V1000S; Leica) in 0.02 M tris-buffered saline (TBS). Free-floating sections were stored in 0.02 M TBS supplemented with 0.025% NaN3 at 4 °C. To produce 30-µm glass-mounted sections, brains were cryoprotected in 0.1 M PB containing 30% sucrose and 0.025% NaN3. Then, brains were flash-frozen in liquid N2, and embedded in optimal cutting temperature embedding matrix (OCT, Tissue-Tek). Coronal sections were cut on a cryostat microtome (CryoStar NX70; Thermo Scientific). Brain sections were washed in 0.02 M TBS, then blocked with a solution containing 5% normal donkey serum, 2% bovine serum albumin (BSA, Sigma Aldrich), 0.3% Triton X-100 in 0.02 M TBS at 22–24 °C for 2 h. Select combinations of primary antibodies were used as follows: guinea pig anti-cFOS (1:1,000; Synaptic Systems, 226005), rabbit anti-cFOS (1:2,000; Synaptic Systems, 226003), rabbit anti-DsRed (1:200; Clontech/Takara, 632496), rabbit anti-RFP (biotinylated, 1:1,000; Rockland, 600-406-379), chicken anti-RFP (1:500; Rockland, 600-901-379), goat anti-GFP (1:200; Abcam, ab6662), goat anti-mCherry (1:500; Antibodies Online, ABIN1440058), guinea pig anti-GluA1 (1:100; Alomone Labs, AGP-009), rabbit anti-GluA2 (1:100; Alomone Labs, AGC-005), chicken anti-NeuN (1:500; Millipore, ABN91), rabbit anti-p44/42 MAPK (pERK1/2Thr202/Tyr204; 1:200; Cell Signaling Technology, 9101S), rabbit anti-TH (1:500; Millipore, AB152), goat anti-VEGFA (1:100; R&D Systems, AF-493-NA), guinea pig anti-VGLUT2 (1:200; Synaptic Systems, 135404), rabbit anti-VGLUT2 (1:500; Synaptic Systems, 135403), and chicken anti-vimentin (1:500; Synaptic Systems, 172006). Cocktails of the antibodies were incubated on an orbital shaker in 0.02 M TBS to which 2% normal donkey serum, 0.1% BSA, 0.3% Triton X-100 and 0.025% NaN3 had been added at 4 °C for 3–4 days. Secondary antibodies included: Alexa Fluor 488 donkey anti-rabbit IgG (1:2,000; Invitrogen, AB21206), Alexa Fluor 488-conjugated AffiniPure donkey anti-guinea pig IgG (1:300; Jackson ImmunoResearch, 706-545-148), Alexa Fluor 488-conjugated AffiniPure donkey anti-mouse IgG (1:300; Jackson ImmunoResearch, 715-545-151), Alexa Fluor 647-conjugated AffiniPure donkey anti-rabbit IgG (1:300; Jackson ImmunoResearch, 711-605-152), Cy2-conjugated AffiniPure donkey anti-goat IgG (1:300; Jackson ImmunoResearch, 705-225-147), Cy2-conjugated AffiniPure donkey anti-rabbit IgG (1:300; Jackson ImmunoResearch, 711-225-152), Cy3-conjugated AffiniPure donkey anti-chicken IgG (1:300; Jackson ImmunoResearch, 703-165-155), Cy3-conjugated AffiniPure donkey anti-guinea pig IgG (1:300; Jackson ImmunoResearch, 706-165-148), Cy3-conjugated AffiniPure donkey anti-rabbit IgG (1:300; Jackson ImmunoResearch, 711-165-152), Cy5-conjugated AffiniPure donkey anti-chicken IgG (1:300; Jackson ImmunoResearch, 703-175-155), Cy5-conjugated AffiniPure donkey anti-guinea pig IgG (1:300; Jackson ImmunoResearch, 706-175-148) and Cy5-conjugated streptavidin (1:200; Jackson ImmunoResearch, 016-170-084). Secondary antibodies were applied in 0.02 M TBS containing 2% BSA, 0.3% Triton X-100, and Hoechst 33,342 (1:10,000; Sigma Aldrich, used as nuclear counterstain, B2261) on an orbital shaker at 22–24 °C for 2 h. After washing in 0.02 M TBS, sections were glass-mounted and coverslipped with an antifade solution consisting of 10% Mowiol (Sigma, 81381), 26% glycerol (Sigma, G7757), 0.2M Tris buffer (pH 8.0) and 2.5% Dabco (Sigma, D27802). Ex vivo brain slices (250–300 µm) after patch-clamp recordings were cleared in an ascending series of glycerol (25%, 50%, 80% and 100% for 1 h each, and 100% overnight), and mounted with the same antifading solution as above.

    Chromogenic histochemistry and electron microscopy for VGLUT2

    To localize VGLUT2 in the periventricular area, samples were prepared as previously published38. In brief, mice (n = 4) were transcardially perfused with ice-cold 0.1 M PB (20 ml), followed by 4% PFA and 0.1% glutaraldehyde (GA) in 0.1 M PB. Sections were washed three times in 0.1 M PB. Endogenous peroxidase activity was blocked by treating the sections with 1% H2O2 for 10 min. Next, sections were blocked (see ‘Immunohistochemistry’) and immunolabelled with a rabbit anti-VGLUT2 antibody (1:1,000; a gift from M. Watanabe)62 and incubated at 4 °C for 2 days to reveal presynaptic terminals in apposition to tanycytes. Following repeated washes in 0.1 M PB, sections were exposed to biotinylated anti-rabbit secondary antibody (Vector Labs BA-1000) at 22–24 °C for 2 h. Next, sections were washed in 0.1 M PB and incubated with pre-formed avidin–biotin–peroxidase complexes (ABC Elite; Vector Laboratories) at 4 °C overnight. Thereafter, sections were osmificated, dehydrated, embedded in durcupan (Fluka, ACM), and cut at 60 nm on an Ultracut UCT microtome (Leica). Imaging was performed on a Transmission Electron Microscope FEI Tecnai 10 (100kV) equipped with a TEM side-mounted camera (EMSIS MegaView III G3).

    Electron microscopy for vimentin, TH and tdTomato

    Male C57Bl6/N mice (n = 3) were used for vimentin plus TH immunostaining. Mice were perfused with a fixative containing 4% PFA, 15% picric acid (by volume) and 0.08% GA in 0.1 M PB. Tissue was post-fixed overnight in GA-free fixative, then washed in PB. Sections containing intact ARC were kept in 10% sucrose in 0.1 M PB for 30 min and 20% sucrose in 0.1 M PB for 1 h. The sections were rapidly freeze/thawed (3×), washed (3×) with 0.1 M PB, and double-stained with chicken anti-vimentin antibody (1:1,000; Sigma in goat blocking serum) and mouse anti-TH antibody (1:3,500 Sigma) on a shaker at 4 °C for 48 h. After repeated washes in PB, sections were incubated for 1.5 h in biotinylated goat anti-mouse and biotinylated goat anti-chicken IgG (1:200 each in goat blocking serum; Vector Labs) at 22–24 °C. Sections were then washed (3×) and incubated in ABC complex (1:100 in PB; ABC Elite kit, Vector Labs) at 22–24 °C for 1.5 h. The immunoreaction was visualized with 3,3-diaminobenzidine (DAB), then extensively washed. Agrp-Cre::Ai14 mice were perfused as above, and carried through the same procedures as above but the sections were incubated in chicken anti-RFP antibody (1:2,000; Rockland) at 4 °C for 48 h. This was followed by biotinylated goat anti-chicken IgG, then ABC (both for 1.5 h) to visualize tdTomato+ (Agrp-Cre) neurons. Following the DAB reaction, sections were osmificated (1% OsO4 in 0.1 M PB) for 30 min, washed in PB followed by double-distilled H2O, and 50% ethanol. Sections were kept in 1% uranyl acetate in 70% ethanol for 1 h, washed in 95% and 100% ethanol, washed (2×) in propylene oxide, and left in a solution of 50% propylene oxide and 50% durcupan for 3 h. Sections were left in pure durcupan overnight, flat-embedded on liquid release-coated slides, coverslipped with Aclar (Electron Microscopy Sciences), glued and trimmed. Sections were collected on Formvar-coated single slot copper grids and imaged using a Philips Tecnai T-12 Biotwin electron microscope.

    Fluorescence in situ hybridization

    PFA-fixed 30-µm glass-mounted sections were used for FISH. We followed the HCR 3.0 protocol for ‘generic sample on slide’ per the manufacturer’s recommendations (Molecular Instruments; https://files.molecularinstruments.com/MI-Protocol-RNAFISH-FrozenTissue-Rev2.pdf) with Agrp, Flt1, Pomc, Th and Vegfa probes. In brief, slides were defrosted and gradually dehydrated in an ascending ethanol gradient (50%, 70%, 100%) for 5 min each at 22–24 °C. Tissue samples were then hybridized by incubation with 1.2 µl of 1 µM stock of each probe (1.2 pmol) in a humid chamber at 37 °C overnight. Excess probe was washed with warm washing buffer (37 °C) mixed with 5× SSCT buffer (that is, sodium chloride/sodium citrate (5× SCC) and 0.1% Tween 20; Sigma Aldrich, 9005-64-5) at scaled composition (75% washing buffer/25% 5× SSCT; 50% washing buffer/50% 5× SSCT; 25% washing buffer/75% 5× SSCT; 100% 5× SSCT) for 15 min each at 37 °C. Next, 2 µl of amplifiers (hairpins) were diluted (from 3 µM stock) in 100 µl amplification buffer and applied to the samples in a humid chamber at 22–24 °C for 12 h. Thereafter, slides were washed in 5× SSCT buffer. Nuclei were counterstained with Hoechst 33,342 (1:10,000; Sigma Aldrich, B2261) diluted in 5× SSCT at 22–24 °C for 15 min. After another wash with 5× SSCT, the samples were coverslipped with an antifade solution made up of 10% Mowiol (Sigma, 81381), 26% glycerol (Sigma, G7757), 0.2 M Tris buffer (pH 8.0), and 2.5% Dabco (Sigma, D27802).

    Confocal and epifluorescence imaging

    Confocal micrographs were acquired on Zeiss LSM710, LSM880/Airyscan or Zeiss LSM900/Airyscan 2 setups. We used a Zeiss AXIO Observer ApoTome.2 platform for epifluorescence microscopy. The number of VGLUT2+ presynapses contacting vimentin+ tanycytes were determined by using a Zeiss LSM880/Airyscan microscope equipped with a Plan-Apochromat 63×/1.4 NA oil objective (Zeiss). We separately acquired 2 × 2 tile scans covering each tanycyte subcategory in coronal brain sections at both −1.94 mm and −2.30 mm relative to bregma. Orthogonal z-stacks were acquired at a depth of 25 µm. Images to quantify the intensity of pERK1/2 were captured on an LSM880 microscope equipped with a Plan-Apochromat 25×/0.8 Imm Korr DIC M27 objective (Zeiss). Images showing complementary GluA2 and VGLUT2 signals within individual synapses were captured on a Zeiss LSM900/Airyscan 2 microscope equipped with a Plan-Apochromat 40×/1.4 NA oil objective.

    Image analysis

    Confocal images were loaded in either Imaris 9.0.2 (Biplane) or Fiji 1.52e (https://imagej.net/Fiji).

    Mapping of VGLUT2+ presynaptic terminals in apposition to tanycytes

    α1-, α2-, β1- and β2-tanycytes (all vimentin+) were separately captured at −1.94 mm and −2.30 mm relative to bregma, and at a tissue depth of 25 µm (z-scan) on a Zeiss LSM880 microscope with their images loaded in Imaris x64 9.0.2 later (Bitplane). Tanycyte filaments were reconstructed along their vimentin signal using the built-in extension ‘Filament tracer’. First, we determined the thickness of the basal process on x, y and z axes (~1 µm). Subsequently, we traced these basal processes by using the ‘Autopath’ method, and by setting the seeding point on the soma of each tanycyte separately. Next, tracing was centred, smoothed, and adjusted to a diameter of 1 µm. To quantify and to reconstruct the VGLUT2 signal in putative presynapses, we first set their diameter to <0.5 µm. Subsequently, we isolated any such VGLUT2 signal with the built-in ‘Spots’ extension to reconstruct spheres. We then used a ‘find spots close to filaments’ Imaris XTension to quantify the density and distribution of those VGLUT2+ presynapses (spots) that apposed vimentin+ tanycyte processes (filaments). The maximal accepted distance from the spot centre (VGLUT2+) to the filament edge (vimentin+) was set to <0.5 µm. Thus, the total number of spots within 0.5 µm was used for statistical analysis.

    cFOS in tanycytes and neurons

    To quantify the number of tanycytes activated by acute thermal manipulation in C57Bl6/J mice of both sexes or after chemogenetically activating glutamate inputs in B6;129S-Slc17a6tm1.1(flpo)Hze/J mice bilaterally injected with either AAV-EF1a-FRT-hM3D(Gq)-mCherry or AAV2/1-Syn-FRT-hM3D(Gq)-mCherry virus particles, we counted the absolute number of cFOS+ nuclei both in vimentin+ tanycytes along the wall of the third ventricle, and in mCherry+ neurons in the PBN per section from confocal micrographs at a tissue depth of 25 µm (z-scans).

    Intensity analysis for pERK1/2 and VEGFA

    Five-by-three tiled confocal images over the cross-section of the third ventricle were acquired on a Zeiss LSM880 microscope at an image depth of 8 bit. Confocal micrographs were loaded in Fiji 1.52e, and their signal intensity for either pERK1/2 or VEGFA was quantified in pre-defined tanycyte subgroups in male mice kept at either 25 °C or 40 °C. Images were acquired at identical settings (including laser power output, digital gain/offset) to allow for comparisons be made on signal intensities between the experimental groups.

    Vegfa expression and localization

    Confocal images of Vegfa mRNA (FISH) from brains of both control and heat-exposed C57Bl6/J male mice that had received scrambled RNAi or Vegfa-targeting RNAi cocktails in the third ventricle were acquired on a Zeiss LSM710 microscope as 2 × 5 image tiles. Thus, the entire length of the ventricular wall was imaged as a z-stack of 25 µm. We reconstructed the wall of the third ventricle with the ‘Surface’ method (over a nuclear signal), thus limiting data collection to only the perikarya of tanycytes. To quantify the number of Vegfa mRNA precipitates in tanycytes, images were loaded in Imaris (Bitplane) with the Vegfa signal in the somata of tanycytes transformed into spots with a maximal diameter of <0.5 µm. Then, the number of spots (Vegfa) that had been in close apposition to the surface was determined by the ‘Find spots close to surface’ Imaris XTension (threshold set to 1 unit) and used for statistical analysis.

    Chemogenetic induction of PBN projections onto tanycytes

    To test whether tanycytes are directly activated by long-range glutamatergic projections, the PBN of B6;129S-Slc17a6tm1.1(flpo)Hze/J was bilaterally injected with AAV-EF1a-FRT-hM3D(Gq)-mCherry or AAV2-Syn1-FRT-hM3D(Gq)-mCherry particles. Twenty-one days after virus delivery, mice were moved to an incubator (Tecniplast, Aria BIO-C36 EVO) set at 25 °C with a reverse 12 h:12 h light:dark cycle for 24 h. The following day, mice were injected intraperitoneally with either sterile physiological saline or CNO (5 mg kg−1; Tocris, 6329) dissolved in saline. After 1.5 h, mice were transcardially perfused with 0.1 M PB followed by ice-cold 4% PFA for histochemistry.

    RNA isolation from the wall of the third ventricle wall and quantitative PCR

    Two groups of P60-P90 C57Bl6/J male mice (n = 4 per group) were acutely exposed to 40 °C for 1 h and compared to mice kept at 25 °C. Their brains were rapidly removed, and 1-mm coronal brain slices were cut by using a steel brain matrix (Stoelting, 51386). The wall of the third ventricle was manually dissected, flash-frozen in liquid N2, and stored at −80 °C until processing. RNA was extracted with the RNeasy mini kit (Qiagen, 74536). To eliminate genomic DNA, samples were treated with DNase I. Thereafter, RNA was reverse transcribed to cDNA with the high-capacity cDNA reverse transcription kit (Applied Biosystems, 4368814). Quantitative real-time PCR was performed (CFX-connect, Bio-Rad) with primer pairs as follows: mouse Vegfa (forward: 5′-gaggggaggaagagaaggaa-3′, reverse: 5′-ctcctctcccttctggaacc-3′) and mouse glyceraldehyde-3-phosphate dehydrogenase (Gapdh; forward: 5′-aactttggcattgtggaagg-3′, reverse: 5′-acacattgggggtaggaaca-3′), which were designed with the NCBI Primer Blast software. Quantitative analysis of gene expression was performed with SYBR Green master mix (Life Technologies, 4364344). Expression levels were normalized to Gapdh, used as a housekeeping standard. Fold changes were determined with the Livak method63.

    Primary cultures of tanycytes

    Primary cultures of tanycytes were generated as described64. P10 Wistar rats (local breeding) were decapitated, and their brains were extracted and immersed in ice-cold sterile Hank’s balanced salt solution (HBSS; Thermo Fisher). The median eminence was dissected under a stereomicroscope (Leica, M205) and crushed on 80-µm nylon meshes. Dissociated cells were cultured in DMEM/F12-phenol red free medium (Thermo Fisher) supplemented with 10% fetal calf serum (Invitrogen). Primary cultures of tanycytes were kept in 5% CO2 atmosphere at 37 °C. Media were half-refreshed every three days. Two days before protein extraction, primary cultures of tanycytes were split in 6-well plates and cultured in DMEM/F12-phenol red free medium supplemented with 5 µg ml−1 insulin from bovine pancreas (Sigma) and 100 µM putrescine dihydrochloride (Sigma).

    Protein extraction from cultured tanycytes

    Primary cultures of tanycytes were washed with ice-cold HBSS (Thermo Fisher), harvested, and pelleted at 1,000 rpm for 60 s. The supernatant was discarded. Pellets were resuspended in 300 mM NaCl, 50 mM HEPES (pH 8.0), 1% IGEPAL CA-630, 0.1% sodium deoxycholate, 1 mM DTT, 1 mM protease inhibitors (EDTA-free, Roche) and incubated on ice for 10 min. Cell lysates were flash-frozen in liquid N2 and stored at −80 °C.

    Mass spectrometry

    Bands on SDS gels (n = 3 biological replicates) were cut into three pieces each and the corresponding proteins were extracted. The proteins of each band were collected as fractions (three for each sample) and subjected to tryptic digest and post-digest purification.

    Approximately 1 µg of tryptic peptides (4.5 µl injection volume) from each fraction (in total three) were separated by an online reversed-phase (RP) HPLC (Dionex Ultimate 3000 RSLCnano LC system, Thermo Scientific) connected to a benchtop Quadrupole Orbitrap (Q-Exactive Plus) mass spectrometer (Thermo Fisher Scientific). Online separation was performed on analytical (nanoViper Acclaim PepMap RSLC C18, 2 μm, 100 Å, 75 μm internal diameter × 50 cm, Thermo Fisher Scientific) and trap (Acclaim PepMap100 C18, 3 μm, 100 Å, 75 μm internal diameter × 2 cm, Thermo Fisher Scientific) columns. The flow rate for the gradient was set to 300 µl min−1, with an applied maximum pressure at 750 mbar. The liquid chromatography method was a 175-min run and the exponential gradient was set at 5–32% buffer B (v/v%; 80% acetonitrile, 0.1% formic acid, 19.9% ultra-high purity LC-MS water) over ~118 min (7 curves). This was followed by a 30-min gradient of 50% buffer B (6 curves) and then increased to 90% of buffer B for another 5 min (5 curves). The liquid chromatography eluent was introduced into the mass spectrometer through an integrated electrospray metal emitter (Thermo Electron). The emitter was operated at 2.1 kV and coupled with a nano-ESI source. Mass spectra were measured in positive ion mode applying top ten data-dependent acquisition (DDA). A full mass spectrum was set to 70,000 resolution at m/z 200 (Automatic Gain Control (AGC) target at 3 × 106, maximum injection time of 30 ms and a scan range of 350–1,800 (m/z)). The MS scan was followed by a MS/MS scan at 17,500 resolution at m/z 200 (AGC target at 1 × 105, 1.8 m/z isolation window and maximum injection time of 70 ms). For MS/MS fragmentation, normalized collision energy for higher energy collisional dissociation was set to 30%. Dynamic exclusion was at 30 s. Unassigned and +1, +8 and > +8 charged precursors were excluded. The minimum AGC target was set to 1.00e3 with an intensity threshold of 1.4e4. Isotopes were excluded. Targets were accepted if more than two peptide fragments covered each and listed in Extended Data Table 1.

    CSF extraction and VEGF ELISA

    Wistar rats of ~P60 of age (all male, n = 3 for 25 °C and n = 4 for 40 °C) were allowed to habituate to the experimental setting in an incubator (Tecniplast, Aria BIO-C36 EVO) at 25 °C with a reverse 12 h 12 h light:dark cycle for 3 days. Next, rats were acutely exposed to either 25 °C or 40 °C for 1 h, anaesthetized intramuscularly with a mixture of ketamine (50 mg kg−1) and xylazine (4 mg kg−1), and their heads were mounted in a stereotaxic frame (RWD). For CSF sampling, the fourth ventricle was approached. For this, the skin was incised, nuchal muscles were retracted to the sides, and partially removed. The dorsal wall of the ventricle formed by the lamina epithelialis was identified as a silvery membrane caudal to the cerebellum between the rim of the foramen magnum and first cervical vertebra. The membrane was pierced with a 26G syringe and 15 µl CSF was removed from the fourth ventricle using a standard 20-μl laboratory pipette (Eppendorf). Samples were flash-frozen in liquid N2 and stored at −80 °C. To test the VEGF content of the CSF, we used a rat VEGF ELISA Kit (Sigma Aldrich; RAB0511) as per the manufacturer’s instructions. An ELISA plate reader set at 450 nm (Glomax Multi+, Promega) was used to read out VEGF levels in 20-μl sample volumes. VEGF concentrations were expressed in pg ml−1.

    Electrophysiology, Ca2+ imaging, optogenetics and analysis

    Acute coronal slices comprising, in the rostrocaudal axis, the medial-caudal portion of the third ventricle were obtained from P60-P90 male C57Bl6/J, Raxtm1.1(cre/ERT2)Sbls/J::B6;129S6-Polr2aTn(pb-CAG-GCaMP5g,-tdTomato)Tvrd/J and B6;129S-Slc17a6tm1.1(flpo)Hze/J mice. Mice were anaesthetized with isoflurane (5%, 1 l min−1 flow rate) prior to decapitation, and their brains were rapidly dissected out. Two hundred fifty-µm-thick coronal slices were cut on a vibratome (VT1200S, Leica) in ice-cold cutting solution (pH 7.3) containing (in mM): 135 N-methyl-d-glucamine, 1 KCl, 1.2 KH2PO4, 10 glucose, 20 choline bicarbonate, 1.5 MgCl2, and 0.5 CaCl2 and continuously oxygenated with 95% O2/5% CO2. Acute slices of the caudal portion of the hypothalamic third ventricle/ARC were incubated at 32 °C for 1 h and allowed to cool to 25 °C in oxygenated ACSF (pH 7.3) containing (in mM): 124 NaCl, 3 KCl, 1.25 KH2PO4, 2 MgCl2, 2 CaCl2, 26 NaHCO3, and 10 mM glucose. For recordings, brain slices were transferred to a recording chamber (Examiner.D1, Zeiss) and superfused with ACSF (25 °C) at a rate of 3 ml min−1 with a peristaltic pump (PPS5, Multichannel Systems). Tanycytes and neurons were recorded through patch pipettes (3–5 MΩ) made from borosilicate glass capillaries pulled on a P100 glass puller (Sutter Instruments). Patch pipettes were filled with an intracellular solution containing (pH 7.3, 300 mOsm; in mM): 125 K-gluconate, 20 KCl, 0.1 EGTA, 2 MgCl2, 10 HEPES, 2 Na-ATP, 0.4 Na-GTP, 10 phosphocreatine and 0.5% biocytin (Tocris, 3349).


    To record glutamatergic inputs onto tanycytes, both sEPSCs and tonic currents were recorded at −70 mV using a Multiclamp 700B amplifier (Molecular Devices), sampled at 10 KHz, and filtered at 2 KHz. EPSCs were analysed using the Mini Analysis Program (Synaptosoft). Both the amplitude and frequency of sEPSCs were statistically tested in both α- and β-tanycytes. s-AMPA (100 µM; Tocris, 0254) was superfused to test for tonic currents. To define voltage responses to currents ramps, tanycytes were recorded in current-clamp mode with the holding current set at 0 pA. Current injections were applied for 1 s with consecutive steps of current of 5 pA for 20 sweeps. To determine the effect of the threshold for neuronal spiking on VEGFA release, acute slices were either superfused with ACSF (control) or with axitinib (40 µM; LC Laboratories A-1107), a selective inhibitor of VEGF receptors. To define their action potential thresholds, patch-clamped neurons were recorded in current-clamp mode with the holding current set at 0 pA. Patch-clamped neurons in the ARC and apposing α-tanycytes were recorded in repeated measures first at 25 °C and after increasing the temperature of the recording chamber to 38 °C by using a temperature controller (Warner Instruments, TC-324C). The voltage value corresponding to the exponential rise of the action potential was used for statistical analysis (Clampfit, Molecular Devices).

    Ca2+ imaging

    We recorded neuronal input-dependent Ca2+ transients in tanycytes from acute slices from Raxtm1.1(cre/ERT2)Sbls/J crossed with B6;129S6-Polr2aTn(pb-CAG-GCaMP5g,-tdTomato)Tvrd/J mice (n = 8, males). We used an AxioExaminer.D1 microscope (Zeiss) and visualized Ca2+ transients with a water-immersion W40×/1.0 DIC VIS-IR Plan-Apochromat objective (Zeiss) and a CoolSnap HQ2 camera (Photometrics). We first proceeded to patch-clamp neurons proximal to the wall of the third ventricle. To induce action potentials in patch-clamped neurons, we injected steps of currents ranging between 10 pA and 30 pA for 500 ms. Simultaneously, a VisiChrome monochromator (Visitron Systems) was used to visualize GCaMP5g in tanycytes. To demonstrate the AMPA receptor (AMPAR) dependence of Ca2+ transients, tanycytes were imaged while ACSF was supplemented with 2,3-dioxo-6-nitro-7-sulfamoyl-benzo[f]quinoxaline (NBQX, 20 µM, Tocris, 1044). In recordings where neuronal activity was pharmacologically manipulated, acute slices from Raxtm1.1(cre/ERT2)Sbls/J crossed with B6;129S6-Polr2aTn(pb-CAG-GCaMP5g,-tdTomato)Tvrd/J mice were placed on µ-Dish 35 mm high chamber for cell culture imaging (Ibidi) mounted on an inverted LSM880 confocal microscope (Zeiss), and visualized with a Plan-Apochromat 20×/0.8 M27 objective (Zeiss). ACSF, 100 µM picrotoxin (Tocris, 1128), 5 µM TTX (Tocris, 1069), 100 µM s-AMPA (Tocris, 0254), 20 µM NBQX (Tocris, 1044) and KCl 50 mM were superfused at a rate of 1.5 ml min−1 with a peristaltic pump (PPS5, Multichannel Systems). Single-plane images of the GCaMP5g signal were captured upon excitation with a 488-nm laser at 5.5% of total efficient power output to avoid phototoxicity. A frame dimension of 512 × 512 pixels at 8 bit with a rate of 600 ms was used with the pinhole set at 447 µm. To analyse Ca2+ transients, image series were loaded in Fiji and the intensity of GCaMP5g transients was calculated from manually drawn regions of interest over tanycyte somata and basal processes proximal to the third ventricle. The GCaMP5g signal was normalized to the difference between the signal intensity in tanycytes during their period of inactivity and background.

    ChR2-assisted circuit mapping

    Ex vivo coronal brain slices (300 μm) encompassing the medial-caudal portion of the third ventricle were cut from B6;129S-Slc17a6tm1.1(flpo)Hze/J mice bilaterally injected with AAV1-CAG-FLEXFRT-ChR2(H134R)-mCherry in the PBN to test possible monosynaptic inputs onto tanycytes. Brain slices were superfused with oxygenated ACSF containing 1 µM TTX (Tocris, 1069) and 100 µM 4-aminopyridine (Sigma Aldrich, 275875) with a peristaltic pump (Multichannel systems, PPS2) at a flow rate of 3 ml min−1 at 25 °C throughout. A BX51WI microscope (Olympus) equipped with a DIC prism (Olympus, WI-DICHTRA2), and LUMPlanFI/IR 60X/0.90W and Plan N4×/0.10 objectives (Olympus) was used. channelrhodopsin-2(ChR2)-mCherry+ axons in close apposition to the third ventricle were excited with a CoolLED (pE-100) light source at 535 nm and imaged on an ORCA-Fusion digital camera (Hamamatsu, C14440). Tanycytes were clamped at a holding potential of −70 mV, and data acquired on an EPC10 USB Quadro patch-clamp amplifier (HEKA) were sampled at 20 KHz, and filtered at 2 KHz. ChR2–mCherry+ terminals were excited with 50-ms light pulses at 470 nm (CoolLED, pE-100) synchronized to the recording of possible optically induced EPSCs in tanycytes. The time response (in ms) and amplitude (in pA) of EPSCs were analysed in PatchMaster Next (HEKA).

    Effects of TRPV2 inhibition and 38 °C on food intake

    We injected tranilast (20 mg kg−1, intraperitoneally T0318-10MG; Sigma Aldrich) in C57Bl6/N mice (n = 4) and compared its effect with naive controls (n = 4) and mice injected with DMSO used as a vehicle (D2650; Sigma Aldrich). Mice were injected with either tranilast or DMSO 10 min before being exposed to 25 °C and then to 38 °C for 1 h on consecutive days. The tranilast concentration was chosen based on dose conversion from human to mouse (considering the body surface area according to US Food and Drug Administration guidelines: http://www.fda.gov/downloads/Drugs/Guidances/UCM078932.pdf). An equivalent mg kg−1 dose for tranilast in mice was calculated by multiplying its human dose (100 mg per 60 kg, equivalent to 1.6 mg kg−1 for human) by the body surface area conversion factor in mice (12.3), resulting in a dose of 19.68 mg kg−1 in mouse.

    Stereotaxic surgery for viral injections

    All mice undergoing stereotaxic delivery of AAV viral particles were processed 21 days after virus delivery. Anaesthesia was induced with isoflurane (5%; 0.6 l min−1 flow rate). The mice were then mounted in a stereotaxic frame (RWD) with anaesthesia maintained with isoflurane (1.5%; 0.6 l min−1 flow rate) through a snout mask. Viral particles were delivered with a micropipette (Drummond) mounted on either a Quintessential Stereotaxic Injector (Stoelting) or an R-480 nanolitre microinjection pump (RWD) at a speed of 100 nl min−1. The pipette was slowly withdrawn 10 min after AAV delivery.

    B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J and B6;129P2-Mapttm2Arbr/J mice used for the transsynaptic mapping of neuronal afferents to tanycytes were unilaterally injected (lateral ventricle) with rAAV8-EF1a-mCherry-IRES-WGA-Cre particles (UNC Vector Core; 1.0 µl) at the following coordinates (all relative to bregma): anterior–posterior (AP): −0.1 mm, lateral (L): 0.9 mm, dorsoventral (DV): −2.3 mm.

    To perform long-range axonal tracing to the third ventricle, C57Bl6/J mice were unilaterally injected (in the ARC) with AAVrg-CAG-GFP particles (70 nl, Addgene, 37825) as above at the following coordinates (all relative to bregma): AP: −1.94 mm; L: 0.25 mm; DV: −5.86 mm.

    For ChR2-assisted-circuit mapping to assess monosynaptic inputs from the PBN to tanycytes, B6;129S-Slc17a6tm1.1(flpo)Hze/J mice were bilaterally injected with AAV1-CAG-FLEXFRT-ChR2(H134R)-mCherry particles (250 nl, Addgene, 75470-AAV1) as above at the following coordinates (all relative to bregma): AP: −5.2 mm; L: ±1.25 mm; DV: −2.8 mm.

    To test tanycyte activation following chemogenetic manipulation of PBN projections and also in behavioural tests, B6;129S-Slc17a6tm1.1(flpo)Hze/J and Raxtm1.1(cre/ERT2)Sbls/J mice were crossed to obtain Slc17a6-FlpO::Rax-CreERT2 mice that were bilaterally injected with either AAV2/1-Syn-FRT-hM3D(Gq)-mCherry particles (Viral Vector Core Facility, Canadian Neurophotonics Platform; RRID:SCR_016477) or AAV-EF1a-FRT-hM3D(Gq)-mCherry particles (Molecular Biology Services, Institute of Science and Technology Austria) at volumes of 250 nl each at the following coordinates (all relative to bregma): AP: −5.2 mm; L: ±1.25 mm; DV: −2.8 mm.

    To block tanycyte-dependent VAMP2-mediated exocytosis in behavioural experiments, Rax-CreERT2 mice or Slc17a6-FlpO::Rax-CreERT2 mice were medially injected in the third ventricle with AAV-TeLC-FLEX-GFP56 or AAV2-FLEX-GFP (control) viruses (1.0 µl; coordinates relative to bregma: AP: −1.70 mm; L: ±0.0 mm; DV: −5.85 mm).

    To block VAMP2-mediated exocytosis in PBN neurons, Flp-dependent AAV2-FlpON-TeLC-GFP or AAV2-FlpON-GFP (control) viruses were injected in the PBN of Slc17a6-IRES2-FlpO-D-mice (250 nl) at the coordinates: AP: −5.2 mm; L: ±1.25 mm; DV: −2.8 mm (all relative to bregma).

    Tamoxifen injection

    Raxtm1.1(cre/ERT2)Sbls mice used for histochemical analysis, as well as Raxtm1.1(cre/ERT2)Sbls/J mice crossed with B6;129S6-Polr2aTn(pb-CAG-GCaMP5g,-tdTomato)Tvrd/J mice for Ca2+ imaging were injected intraperitoneally for 3 consecutive days with 150 mg kg−1 tamoxifen (Sigma, T5648), and processed 3 days following the last injection. For behavioural tests, Raxtm1.1(cre/ERT2)Sbls mice crossed with B6;129S-Slc17a6tm1.1(flpo)Hze/J mice were injected intraperitoneally for 3 consecutive days with 50 mg kg−1 4-hydroxytamoxifen (Sigma, H6278) to ensure maximal recombination of the AAV-TeLC-FLEX-GFP construct in tanycytes.

    Behavioural tests and controls

    To test the effect of acute heat exposure on food intake, P60 C57Bl6/J mice were habituated to the experimental room set to 25 °C for 24 h. Next, mice were transferred to thermo-controlled cabinets (Sanyo Incubator, MIR-254) preset to either 25 °C (control) or 40 °C for 1 h. Following heat exposure, mice were single housed in PhenoTypers (Noldus) placed into incubators (Memmert, MEMM-OT3007S and Tecniplast, Aria BIO-C36 EVO) set to 25 °C with a reversed 12 h:12 h light:dark cycle for another 24 h. Food and fluid intake, as well as mobility were monitored over 24 h after acute thermal manipulation by weighing the food pellet, measuring the volume of water consumed, or scoring the frequency of eating bouts and general mobility (both in EthoVision XT15; Noldus). Behavioural tests were designed such that each mouse served as its own control (baseline versus post-heat exposure data), allowing statistical analysis through repeated-measures analysis of variance (ANOVA).

    To test if neuronal activity-induced VEGFA release from tanycytes affected food intake, P60–P70 male C57Bl6/J mice were intracerebroventricularly infused with 1 nmol/1.5 µl of either Accell mouse Vegfa siRNA (Vegfa-RNAi; Dharmacon, E-040812-00-0020) or Accell non-targeting siRNA (control; Dharmacon, D-001950-01-20) in the third ventricle (AP: −1.70 mm; L: ±0.0 mm; DV: −5.85 mm relative to bregma). First, we tested the knockdown efficiency of Vegfa-RNAi by infusing P60–P70 male C57Bl6/J mice (n = 4 per group) with either scrambled RNAi or Vegfa-RNAi (stereotaxic surgery was identical as described above). Eight days after RNAi infusion, mice were perfused with ice-cold 4% PFA and the brains processed for FISH. Next, to test the impact of reduced VEGFA release on food intake upon heat exposure, P60–P70 male C57Bl6/J mice were intracerebroventricularly infused with either scrambled RNAi (control) or Vegfa-RNAi (n = 8 per group). Mice were single housed in PhenoTypers (Noldus) placed into incubators (Tecniplast, Aria BIO-C36 EVO) at 25 °C with a reversed 12 h:12 h light:dark cycle, and allowed to recover for 8 days. From day 3 to 8 post-surgery, we monitored both food intake and body mass by weighting the food pellets and mice, respectively. On days 9 and 10, mice were subjected to thermal challenge (40 °C, 1 h) in incubators (Sanyo, MIR-254). This was followed by measuring foor intake and body mass for 24 h as above (PhenoTypers, Noldus).

    To test the effect of the chemogenetic activation of PBN projections onto tanycytes, male Raxtm1.1(cre/ERT2)Sbls::B6;129S-Slc17a6tm1.1(flpo)Hze/J mice were stereotaxically injected with AAV-TeLC-FLEX-GFP (third ventricle) and AAV-FRT-hM3D(Gq)-mCherry (PBN, bilaterally) to simultaneously manipulate tanycytes and glutamatergic output form the PBN. All tests were performed in a self-controlled design to use the same mice before and after blocking VAMP2-mediated exocytosis from tanycytes, by the temporally controlled recombination of the AAV-TeLC-FLeX-GFP construct that encodes TeLC (Fig. 5g). Twenty-one days after virus delivery, mice were placed individually in PhenoTypers (Noldus) mounted in ventilated and temperature-controlled (29 °C) cabinets (Memmert, MEMM-OT3007S) with a reversed 12 h:12 h light:dark cycle. Food intake, locomotion, and drinking were monitored with EthoVision XT15 (Noldus). Mice were allowed to habituate for 2 days to the experimental setup (days 21,22). Next, baseline activity was recorded for 24 h (day 23). On day 24, mice were treated with 3 mg kg−1 CNO (Tocris, 6329) by both intraperitoneal delivery and in the drinking water, together with 5 mM saccharine (Sigma), to test the effect of chemogenetically activating PBN projections on feeding, drinking, and locomotor activity, whilst leaving VAMP2-mediated exocytosis from tanycytes unaffected. Thereafter, mice were placed individually in home cages for Cre-dependent recombination of the TeLC construct to take place into Rax-expressing tanycytes by injecting 50 mg kg−1 4-hydroxytamoxifen (Sigma) for 3 days (days 25–27). Mice were then allowed to recover for another 3 days (days 28–30). On day 31, we returned the mice to the PhenoTypers and allowed them to habituate for another 48 h (days 31 and 32). Thereafter, we recorded (for 24 h, day 33) their baseline activity following the TeLC-dependent block of VAMP2 in tanycytes. The next day (day 34), we triggered neuronal activity in the PBN by injecting CNO (3 mg kg−1) and using it as an additive to the drinking water together with saccharine (5 mM), and tested feeding, drinking, and locomotor activity again. On the last day (day 35), mice were transcardially perfused with ice-cold 4% PFA. Their brains were routinely processed to verify the accuracy of virus delivery. No mouse was excluded from the analysis.

    To test if blocking VAMP2-mediated exocytosis in PBN neurons projecting to tanycytes affected food intake following acute heat exposure, Slc17a6tm1.1(flpo)Hze/J mice were bilaterally injected in the PBN with either AAV2-FlpON-GFP (control) or AAV2-FlpON-TeLC-GFP. Twenty-one days after virus delivery, mice were sequentially exposed to either 25 °C (control) or 40 °C for 1 h (on consecutive days). Food intake was determined by measuring the weight of food pellets. To test if blocking VAMP2-mediated exocytosis in tanycytes could modify food intake following acute heat exposure, Raxtm1.1(cre/ERT2)Sbls/J mice were medially injected in the third ventricle with either AAV2-FLeX-GFP (control) or AAV2-FLeX-TeLC-GFP. To induce Cre-dependent recombination, mice were injected with tamoxifen (150 mg kg−1) for 3 consecutive days, starting 2 days after surgery. Twenty-one days after virus delivery, mice were sequentially exposed to either 25 °C (control) or 40 °C for 1 h (on consecutive days). In both experiments, food intake was determined by measuring the weight of food pellets.

    Statistics and reproducibility

    Data were analysed using GraphPad Prism 8.0.2 (GraphPad). Two sets of independent samples were compared using two-tailed Student’s t-test. Repeated measures of pair-wise comparisons were analysed by paired two-tailed Student’s t-test. Multiple sets of measurements involving one independent variable were analysed by one-way ANOVA and further justified by Bonferroni’s post hoc comparison. Repeated-measures two-way ANOVA and three-way ANOVA were used to evaluate between and within factors, with Bonferroni’s post hoc test applied throughout. The Kolmogorov–Smirnov test was used to analyse cumulative distribution. Data were expressed as means ± s.e.m. throughout, except in box-and-whisker plots that show median ± interquartile ranges, and minimum and maximum values. Statistical significance was indicated as *P < 0.05, **P < 0.01 or ***P < 0.001. For neuroanatomy, a minimal desired cohort size of n = 3 mice was chosen, with higher mouse numbers specified in the relevant figure legends.

    Statistical output for main figures 

    Figure 1b: two-way repeated-measures ANOVA: interaction (sex versus temperature): F = 0.005, P = 0.942; sex: F = 7.969, P = 0.013; temperature: F = 32.240, P < 0.001. Bonferroni’s multiple comparison: t = 4.067, **P = 0.002 (males at 25 °C versus 40 °C); t = 3.963, **P = 0.003 (females at 25 °C versus 40 °C).

    Figure 1c: repeated-measures ANOVA: F = 18.030, p < 0.001.

    Figure 1f: two-way ANOVA: interaction (sex versus temperature): F = 1.497, P = 0.249; sex: F = 3.589, P = 0.087; temperature: F = 81.700, P < 0.0001. Bonferroni’s multiple comparison: t = 6.788, ***P < 0.001 (males 25 °C versus 40 °C); t = 5.969; ***P < 0.001 (females 25 °C versus 40 °C).

    Figure 3b, middle: frequency: Student’s t-test (two-sided), t = 0.476, P = 0.639; α- versus β-tanycytes.

    Figure 3b, right: amplitude: Student’s t-test (two-sided), t = 3.006, **P = 0.007; α- versus β-tanycytes.

    Figure 4b: Student’s t-test (two-sided), t = 7.120, ***P < 0.001.

    Figure 4e: repeated-measures ANOVA: interaction: F = 3.974, P = 0.066; treatment (ACSF versus axitinib): F = 1.947, P = 0.185; temperature: F = 23.880, P < 0.001; subject, F = 6.723; P < 0.001. Bonferroni’s multiple comparison: ACSF (25 °C versus 38 °C), t = 4.865; ***P < 0.001; axitinib (25 °C versus 38 °C) t = 2.046; P = 0.1201.

    Figure 4g: Student’s t-test (two-sided), t = 3.143, *P = 0.020.

    Figure 4h: repeated-measures ANOVA: interaction (treatment versus temperature), F = 1.081, P = 0.316; temperature: F = 17.310, P = 0.001; treatment: F = 3.089, P = 0.094. Bonferroni’s multiple comparison: temperature: control, t = 3.677, **p = 0.005; Vegfa-RNAi, t = 2.207, P = 0.089 (not significant).

    Figure 4k: Student’s t-test (two-sided), **P < 0.01.

    Figure 5b, right: Student’s t-test (two-sided), **P < 0.01; n = 3 mice per group.

    Figure 5c: two-way ANOVA: interaction (TeLC versus temperature): F = 8.682, P = 0.042; GFP versus TeLC: F = 0.683, P = 0.455; temperature: F = 16.34, P = 0.0156. Bonferroni’s multiple comparison: P = 0.016 (GFP; 25 °C versus 40 °C); P = 0.964 (TeLC; 25 °C versus 40 °C).

    Figure 5i: two-way repeated-measures ANOVA: time: F = 6.202, P = 0.026; treatment: F = 6.839, P = 0.048; interaction (time versus treatment): F = 1.944, P = 0.208.

    Reporting summary

    Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

    Source link