A brainstem-hypothalamus neuronal circuit reduces feeding upon heat exposure

Abstract

Empirical evidence suggests that heat exposure reduces food intake. However, the neurocircuit architecture and the signalling mechanisms that form an associative interface between sensory and metabolic modalities remain unknown, despite primary thermoceptive neurons in the pontine parabrachial nucleus becoming well characterized¹. Tanycytes are a specialized cell type along the wall of the third ventricle² that bidirectionally transport hormones and signalling molecules between the brain's parenchyma and ventricular system^3-8. Here we show that tanycytes are activated upon acute thermal challenge and are necessary to reduce food intake afterwards. Virus-mediated gene manipulation and circuit mapping showed that thermosensing glutamatergic neurons of the parabrachial nucleus innervate tanycytes either directly or through second-order hypothalamic neurons. Heat-dependent Fos expression in tanycytes suggested their ability to produce signalling molecules, including vascular endothelial growth factor A (VEGFA). Instead of discharging VEGFA into the cerebrospinal fluid for a systemic effect, VEGFA was released along the parenchymal processes of tanycytes in the arcuate nucleus. VEGFA then increased the spike threshold of Flt1-expressing dopamine and agouti-related peptide (Agrp)-containing neurons, thus priming net anorexigenic output. Indeed, both acute heat and the chemogenetic activation of glutamatergic parabrachial neurons at thermoneutrality reduced food intake for hours, in a manner that is sensitive to both Vegfa loss-of-function and blockage of vesicle-associated membrane protein 2 (VAMP2)-dependent exocytosis from tanycytes. Overall, we define a multimodal neurocircuit in which tanycytes link parabrachial sensory relay to the long-term enforcement of a metabolic code.

Fig. 1. Acute heat reduces food intake and selectively activates α-tanycytes.

a, Schematic of the experimental paradigm to link acute thermal challenge (1 h) to reduced food intake in wild-type male and female mice. b, Food intake in male and female mice during a 24-h period after exposure to 40 °C (1 h) versus continuity at 25 °C (n = 8 per sex). c, Core body temperature during (pink shaded region) and after thermal challenge. Data are presented as a facet-wrap plot with measurements taken at 15-min intervals (n = 8 male mice). d, Acute heat-induced cFOS expression in the PBN (arrows). scp, superior cerebellar peduncle. Scale bar, 120 μm. e, cFOS immunoreactivity in vimentin⁺ α-tanycytes (at −2.30 mm relative to bregma) at the indicated temperatures (n = 3 male mice per group; data for female mice are in Extended Data Fig. 1e). Scale bars, 4 µm. f, The number of cFOS⁺ α-tanycytes per 50 µm of ventricular surface in control versus heat-exposed mice of both sexes. Data are mean ± s.e.m., with circles (b,f) denoting individual data points. b,c,f, Detailed statistics are provided in Methods. *P < 0.05, **P < 0.01, ***P < 0.001. Source Data

Fig. 2. Parabrachial neurons innervate α-tanycytes.

a, Left, VGLUT2⁺ presynapses (black) formed synapse-like contacts on tanycytes (T) along the third ventricle (3V). The outlined region is enlarged on the right. Right, VGLUT2⁺ presynapse (semi-transparent cyan) apposing the soma of a tanycyte (semi-transparent magenta). Scale bars: 2 μm (left), 200 nm (right). b, Schema of transsynaptic labelling. Tanycytes along the third ventricle served as the primary transduction site for WGA–Cre to mark synaptically connected neurons (cyan). LV, lateral ventricle. c, Top, transsynaptic labelling using rAAV8-EF1a-mCherry-IRES-WGA-Cre in Ai14-tdTomato-loxP mice. Vimentin⁺ tanycytes co-expressed mCherry (arrowheads) and were synaptically connected, among others, to local neurons (arrows). Bottom, putative contact between a vimentin⁺ tanycyte process (arrowheads) and an mCherry⁺ transsynaptically labelled neuron (arrow, cyan). Scale bars: 20 µm (top), 6 µm (bottom). d, Transsynaptic labelling by rAAV8-EF1a-mCherry-IRES-WGA-Cre in Tau^mGFP-loxP mice. mCherry⁺ tanycytes were innervated by local NeuN⁺ neurons (mGFP, colour-coded to cyan, arrows in both overview and inset). Arrowheads denote neuronal processes. Scale bars: 40 µm (main image), 2 µm (inset). e, GFP⁺ parabrachial neurons labelled with AAVrg-CAG-GFP injected at the level of ARC. Scale bar, 50 µm. f, Transsynaptic labelling by rAAV8-EF1a-mCherry-IRES-WGA-Cre in Ai14 mice. Parabrachial neurons (tdTomato⁺, cyan) synaptically connected to tanycytes (see also c) resided laterally to the scp. Scale bar, 50 µm. g, Cartoon of the chemogenetic manipulation of parabrachial glutamatergic neurons in Slc17a6-IRES2-FlpO-D mice. h, Transduced neurons expressed mCherry. cFOS⁺ neurons are shown 1.5 h after either saline or CNO (5 mg kg⁻¹, intraperitoneal) injection. Scale bars, 50 µm. i, cFOS activation (grey) in α-tanycytes (green) 1.5 h after administration of saline or CNO. Scale bars, 10 µm. j, Left, mCherry⁺ parabrachial efferents (magenta) from glutamatergic neurons apposed (arrows) vimentin⁺Hoechst 33,421⁺ tanycytes (composite; outlined region is enlarged on the right) co-expressing cFOS. Right, mCherry⁺ parabrachial efferents (arrowheads) apposed vimentin⁺ processes. Scale bars: 8 µm (left), 1 µm (right).

Fig. 3. Tanycyte responses to synaptic afferent modulation.

a, GluA1 and GluA2 in tanycytes. Scale bars, 100 µm. b, Top left, biocytin-filled, vimentin⁺ tanycyte (T) with GluA2 expression (main image; scale bar, 5 µm), with enlarged view showing GluA2 in a vimentin⁺ basal filament (arrowheads; scale bar, 800 nm). Frequency (middle) and amplitude (right) of sEPSCs recorded in α- and β-tanycytes (both n = 11 cells). Detailed statistics are presented in Methods. c, Schema of ex vivo experiments. Top, configuration at test (t₀). Bottom, action potentials in neurons evoked at t₁. d, Left, Ca²⁺ transients in tanycytes (arrowheads) in response to an evoked action potential (AP) of 30 pA per 100 ms (arrow) (Supplementary Video 1). Middle, relative fluorescence intensity for GCaMP5g (F/F₀) in tanycytes upon action potential induction in neurons (arrows). Right, time lag of GCaMP5g relative fluorescence after the last action potential (trains of 8 action potentials; 305.2 ± 35.25 ms, n = 30 tanycytes, n = 6 experiments). Scale bars, 20 µm. e, Left, a biocytin-filled neuron in the ARC of a Rax-CreER^T2::PC-G5-tdT mouse. Right, intersection between a biocytin⁺ neuronal process (cyan) and a tanycyte (tdTomato⁺, magenta) in the outlined region in the left image. Scale bars: 20 µm (left); 2 µm (right). f, Left, cartoon showing tanycytes (green) tested for optogenetically induced EPSCs by stimulating PBN efferents (ChR2–mCherry, red) with 50-ms pulses of 470-nm light. Middle, bright-field (BF) view of tanycytes along the third ventricle overlaid on an mCherry⁺ afferent (arrowheads; scale bar, 10 µm). Right, a putative intersection between a tanycyte and afferent (scale bar, 2 µm). g, Left, optically induced EPSCs (arrows) in tanycytes. Time lag (middle; 256.5 ± 34.66 ms) and amplitude (right; 6.759 ± 0.48 pA; n = 29 EPSCs from n = 7 tanycytes, n = 4 independent experiments). b,d,g, In box plots, the centre line is the median, box edges delineate top and bottom quartiles, whiskers extend to minimum and maximum values and circles depict individual data points. Source Data

Fig. 4. Tanycytes produce VEGFA upon heat exposure of mice.

a, Vegfa mRNA (green precipitate) in α1-tanycytes of mice at 25 °C and 40 °C. Scale bars, 5 µm. b, Quantification of Vegfa punctae in α1-tanycytes. c, Basal processes of tanycytes (green overlay) apposed the perikarya of both TH⁺ (left) and AgRP⁺ neurons (right). Scale bars, 2 µm. d, Co-localization (arrows) of Flt1 and either Th (left) or Agrp (right) in the ARC. Scale bars, 3 µm. e, Threshold of spontaneous action potentials in neurons sequentially recorded at 25 °C and 38 °C in artificial CSF (ACSF) alone (control) or ACSF supplemented with axitinib (40 µM; n = 8 pairs). In box plots, the centre line is the median, box edges indicate interquartile ranges, and whiskers extend to minimum and maximum values. f, Experimental design for infusion of scrambled RNAi (control) or Vegfa-RNAi in the third ventricle. g, Left, Vegfa mRNA (green precipitate) in α-tanycytes in control or after Vegfa-RNAi infusion. Right, quantification of Vegfa punctae in α-tanycytes (n = 4 mice per condition). Scale bars, 5 µm. h, Food intake during a 24-h period after acute thermal challenge (1 h) with pre-treatments as indicated (n = 8 mice per group). i, Schema of the experiment with AAV-FLEX-GFP and AAV-FLEX-TeLC-GFP virus particles infused in the third ventricle of Rax-CreER^T2 mice. j, Orthogonal image stack showing VAMP2 along a vimentin⁺ process. Scale bar, 500 nm. k, Temperature switching reduced food intake in control mice expressing only GFP in tanycytes, whereas TeLC–GFP expression in tanycytes abolished the temperature sensitivity of food intake (n = 4 mice per condition). b,g,h,k, Data are mean ± s.e.m. b,e,h, Individual data points are displayed as circles. Sections were routinely counterstained with Hoechst 33,421. e,g,h,k, Detailed statistics are presented in Methods. NS, not significant. Source Data

Fig. 5. Tanycytes link parabrachial activity to feeding.

a, Cartoon showing a TeLC-based strategy to inhibit glutamate release. b, Left, exposure to 40 °C (1 h) resulted in cFOS⁺ tanycytes in mice that had received control viruses (AAV-FLEX-GFP). By contrast, cFOS⁻ (inactive) tanycytes were mostly seen after thermal challenge of mice treated with AAV-FLEX-TeLC-GFP. Right, quantification of cFOS⁺ tanycytes. c, Inactivation of glutamatergic PBN neurons by TeLC attenuated the sensitivity of food intake to acute heat. d, Schema of the intersectional genetic approach used to simultaneously modulate VGLUT2⁺ PBN neurons and RAX⁺ tanycytes. e, TeLC–GFP-expressing hypothalamic tanycytes. f, hM3D(G_q)-mCherry-expressing Slc17a6⁺ neurons in the PBN. e,f, Scale bars, 50 µm. g, Experimental timeline. h, Facet-wrap timeline plot showing the cumulative eating time during baseline (24 h), CNO exposure (3 mg kg⁻¹; CNO 1), baseline to TeLC (recombined with 4-hydroxytamoxifen), and AAV2-FLEX-TeLC-GFP recombined in tanycytes and injected with CNO (3 mg kg⁻¹; CNO 2 TeLC). Data are means ± s.e.m., solid circles show individual data points. i, Cumulative eating time in 24 h. Group designations and colouring correspond to those in h. b,c,i, Data are means ± 95% confidence interval of the s.d. (n = 3–4 per group), with dots representing individual data points. Detailed statistics are provided in Methods. Source Data

Extended Data Fig. 1. Acute heat restricts food intake and activates α tanycytes.

a. Time-resolved measurement of food intake after exposure to 40 °C for 1h in male mice. Control measurements (25 °C, 1h) were performed in the same mice 24h earlier. Data are from n = 4 mice, **p < 0.01, *p < 0.05, Student’s t-test. b. Diurnal activity of male (♂, left) and female (♀, right) mice, expressed as the distance moved (cm) in a period of 20h following exposure to 25 °C (grey line and shading) and then to 40 °C (red line and shading) for 1h. Time intervals were expressed as zeitgeber time points (ZT), wherein ZT12 and ZT0 coincided with the onset of the dark and the light phases, respectively. Data were presented as facet-wraps with measurements at 1h intervals for 20 h; n = 8 mice/sex, repeated-measures ANOVA: males (♂), interaction (time vs. temperature): F = 2.223; p = 0.002; time: F = 5.047; p < 0.001; temperature: F = 0.017; p = 0.898 and females (♀): interaction (time vs. temperature): F = 2.700; p < 0.0001; time: F = 11.190; p < 0.0001; temperature: F = 6.536; p < 0.001. c. Changes in body weight (g) for each mouse 24h after exposure to 25 °C (grey) and then to 40 °C (pink bars) for 1h. Data from both males (♂) and females (♀) are shown. Data in bar graphs were expressed as means ± s.e.m., while individual changes are shown by interconnected circles; n = 8/sex; repeated-measures ANOVA: interaction (sex vs. temperature): F = 0.796; p = 0.3871; sex: F = 2.266; p = 0.1545; temperature: F = 8.713; p = 0.0105. Bonferroni’s multiple comparisons for temperature in males: t = 1.456; p = 0.334; females: t = 2.718; p = 0.033. d. Temperature changes in the intrascapular region (BAT, left) and the perianal region (right), expressed in Celsius (°C). Data represents facet-wraps with measurements at 15-min intervals before, during, and after exposing mice to acute heat for 1h (40 °C, pink shading); n = 8 mice/group, repeated measures ANOVA: BAT: F = 11.260; p < 0.001; perianal: F = 36.050; p < 0.001. e. Confocal micrographs showing cFOS immunoreactivity (in cyan) in vimentin⁺ α-tanycytes (in magenta; at −2.30 mm relative to bregma) at the temperatures indicated in female mice (♀; n = 4/group). Data in facet-wrap plots were expressed as means ± 95% confidence interval of the s.d. throughout. Abbreviation: 3V, 3^rd ventricle. Scale bar = 5 µm. Source Data

Extended Data Fig. 2. Acute heat exposure induces cFOS expression in tanycytes.

a. Left: Confocal micrographs showing cFOS immunoreactivity (in cyan) at −1.94 mm (relative to bregma) after 1h exposure to either 40 °C or 4 °C vs. 25 °C, used as control, in male (♂) mice. The dorsomedial hypothalamic nucleus (DMH), the ventromedial hypothalamic nucleus (VMH) and the arcuate nucleus (ARC) were indicated. Arrowheads point to cFOS⁺/vimentin⁺ (the latter marker in magenta) tanycytes. Scale bar = 70 µm. Right: Quantitative analysis of cFOS⁺ α-tanycytes. Data in bar graphs represent means ± s.e.m. with individual data points shown as circles; n = 3/condition; one-way ANOVA: F = 16.34; p = 0.004. Bonferroni’s multiple comparison: 25 °C vs. 40 °C: t = 5.102; p = 0.007; 25 °C vs. 4 °C: t = 0.3189; p ~ 1.0 (n.s., non-significant); 40 °C vs. 4 °C: t = 4.783; p = 0.009. b. Left: Confocal micrographs showing cFOS immunoreactivity (in cyan) at −2.30 mm (relative to bregma) after 1h exposure to 40 °C or 4 °C vs. 25 °C, used as control, in male (♂) mice. Arrowheads point to cFOS⁺/vimentin⁺ (the latter marker in magenta) α-tanycytes. Scale bar = 70 µm. Right: Quantitative analysis of cFOS⁺ α-tanycytes. Data in bar graphs show means ± s.e.m. with individual data points depicted as circles; n = 3/condition; one-way ANOVA: F = 30.460; p < 0.001. Bonferroni’s multiple comparison: 25 °C vs. 40 °C: t = 7.590; p < 0.001; 25 °C vs. 4 °C: t = 2.219; p = 0.205; 40 °C vs. 4 °C: t = 5.372; p = 0.005. c. Right: Confocal micrographs of cFOS⁺ α-tanycytes (in cyan) at −1.94 mm (relative to bregma) after 1h exposure to 40 °C vs. 25 °C in female (♀) mice. Arrowheads denote cFOS⁺/vimentin⁺ tanycytes (in magenta). Scale bar = 100 µm. Right: cFOS⁺ α-tanycytes at −1.94 mm (relative to bregma) at the temperatures indicated. Data in bar graphs represent means ± s.e.m. with individual data points shown as circles; n = 4/condition; t = 14.700; p < 0.0001; Student’s t-test. d. Left: Confocal micrographs with cFOS immunoreactivity (in cyan) at −2.30 mm (relative to bregma) following the temperature manipulations as indicated in females (♀). Arrowheads point to cFOS⁺/vimentin⁺ α-tanycytes (in magenta). Scale bar = 50 µm. Right: cFOS⁺ α-tanycytes at −2.30 mm. Data in bar graphs are means ± s.e.m. with individual data points shown as circles; n = 4/condition; t = 6.331; p < 0.001; Student’s t-test. Source Data

Extended Data Fig. 3. Acute heat induces pErk1/2 in tanycytes.

a. Left: Photomicrographs showing pErk1/2 immunoreactivity (in cyan) in tanycytes co-labeled with vimentin (in magenta) at −1.94 mm (relative to bregma) after 1h exposure to 40 °C vs. mice kept at 25 °C. The dorsomedial hypothalamic nucleus (DMH), the ventromedial hypothalamic nucleus (VMH) and the arcuate nucleus (ARC) were indicated. Scale bar = 70 µm. Right: pERK1/2 immunoreactivity in α-tanycytes in control (25 °C, in grey) and after exposure to 40 °C for 1 h (in pink). Data in bar graphs represent means ± s.e.m. with individual data points shown as circles; n = 3/condition; t = 3.002; p = 0.039; Student’s t-test. b. Left: Photomicrographs showing pErk1/2 immunoreactivity (in cyan) in vimentin⁺ α-tanycytes (in magenta) residing at −2.30 mm (relative to bregma) after 1 h thermal manipulation, relative to mice kept at 25 °C. Scale bar = 70 µm. Right: pErk1/2 immunoreactivity in α-tanycytes at −2.30 mm (relative to bregma) as indicated. Data in bar graphs are means ± s.e.m. Individual data points appear as circles; n = 3/condition; t = 3.565; p = 0.023; Student’s t-test. Source Data

Extended Data Fig. 4. Parabrachial neurons innervate tanycytes.

a. Confocal image of GFP⁺ neurons within the arcuate nucleus (ARC; in cyan) transduced with an AAVrg-CAG-GFP vector and positioned ventrolateral to the 3^rd ventricle (3V). Scale bar = 100 µm. ‘1’. Panoramic image from Fig. 2e, showing GFP⁺ neurons (in cyan) in the locus coeruleus (LC) and parabrachial nucleus (PBN) retrogradely labelled with an AAVrg-CAG-GFP construct. Scale bar = 100 µm. b. Panoramic confocal image of a GFP⁺ neuron (in cyan). Open rectangle denotes the position of ‘1’ in the PBN trans-synaptically labelled by rAAV8-EF1a-mCherry-IRES-WGA-Cre in Tau^mGFP-loxP mice. Scale bar = 50 µm (b), 8 µm (1). c. Tiled image survey of Fig. 2f showing GFP⁺ neurons (in cyan, arrows) in the PBN trans-synaptically labelled by a rAAV8-EF1a-mCherry-IRES-WGA-Cre virus in Ai14 mice. Scale bar = 50 µm. Abbreviations: 4V, 4^th ventricle; Aq, aqueduct; scp, superior cerebellar peduncle. d. Number of cFOS⁺ mCherry-transduced neurons in the PBN. Data in bar graphs represent means ± s.e.m. for saline (grey, n = 5) and CNO (5 mg/kg in red, n = 6) groups. Circles correspond to individual data points; t = 4.094; p = 0.003; Student’s t-test. e. Confocal images of the 3^rd ventricle (3V) showing periventricular PBN efferents (mCherry, in magenta) with vimentin⁺ tanycytes in green. Left and right images are from coordinates −1.94 mm and −2.30 mm (relative to bregma), respectively. Scale bars = 70 µm. Source Data

Extended Data Fig. 5. Chemogenetic induction of PBN neurons activates tanycytes.

a. Confocal images along the wall of the 3^rd ventricle (−1.94 mm, relative to bregma) showed cFOS activation (in grey) both in vimentin⁺ α-tanycytes (in green) and neurons following an i.p. injection of either saline or CNO (5 mg/kg; 1.5 h) in Slc17a6-IRES2-FlpO-D mice in which AAV2-Syn1-FRT-hM3D(Gq)-mCherry particles were stereotaxically delivered to the PBN 21 days earlier. Scale bar = 15 µm. a₁. The number of cFOS⁺ α-tanycytes (−2.30 mm, relative to bregma) was expressed along 50 μm ventricular wall segments. Data in bar graphs are means ± s.e.m.; for saline (grey, n = 5) and CNO-injected mice (5 mg/kg in red, n = 6). Open circles correspond to individual data points; t = 4.732; p = 0.001; Student’s t-test. Source Data

Extended Data Fig. 6. Synaptic activation of tanycytes.

a. Image survey of the third ventricle (3V) used to determine the density and distribution of synaptic inputs to vimentin⁺ tanycytes (in magenta). Dorsoventral subsetting of the ventricle wall (dashed lines) helped to identify the positions of α1, α2, β1, and β2 tanycytes. Scale bars = 100 µm. b. Image series to illustrate the image analysis pipeline used to quantify synaptic input onto tanycytes. 1,2: Confocal images captured at 63x primary magnification show vimentin⁺ α-tanycytes (in magenta) and VGLUT2⁺ putative presynapses (in green). Scale bar = 10 µm. 3: ‘Filament’ and ‘spots’ rendering algorithms were superimposed on vimentin (in magenta) and VGLUT2 (in green), respectively. Scale bar = 10 µm. The image shows how VGLUT2⁺ terminals were reconstructed as ‘spots’ and accepted as relevant only within a distance of < 0.5 µm from tanycyte filaments by using the built-in MatLab ‘find-spots-close-to-filaments’ function in Imaris. 4: Magnified view of a subset of VGLUT2⁺ spheres (in green) that were < 0.5 µm to the vimentin signal (in magenta). Scale bar = 2 µm. 5: VGLUT2⁺ inputs contacting β2-tanycytes. Scale bar = 20 µm. 6: Reconstruction of VGLUT2⁺ inputs contacting β2-tanycytes. Scale bar = 5 µm. c. Boxplots showing the probability of α1, α2, β1, and β2 tanycytes receiving VGLUT2⁺ inputs at -1.94 mm (relative to bregma, left) and −2.30 mm (relative to bregma, right). None of the tanycyte subtypes was significantly different; n = 7 mice, n = 50 processes reconstructed/subtype/animal, n = 200 processes/animal in total. d. Boxplots of the number of VGLUT2⁺ terminals on tanycytes at the rostro-caudal locations as above. None of the tanycyte subtypes was significantly different in n = 7 mice, n = 50 processes reconstructed/subtype/animal, n = 200 processes/animal in total. e. Quantitative analysis of VGLUT2⁺ punctae in a distance of < 0.5 µm to reconstructed filaments from both rostral and caudal regions. Data were shown as cumulative plots of VGLUT2⁺ terminals juxtaposing α- and β-tanycytes (pooled) at both rostral and caudal positions relative to bregma (b.) as above; n = 500 filaments were reconstructed/tanycyte subtype from n = 7 male mice (P60-90). Kruskal-Wallis test: KW = 96.040; p < 0.001 for α (caudal) vs. β (caudal); for α (rostral) vs. α (caudal); for β (caudal) vs. β (rostral). f. Left: Topographic view of tanycytes expressing GluA2 in apposition to VGLUT2⁺ terminals in Rax-CreER^T2::Ai14 mice. Open rectangle (‘1’) shows the location of the image series to the right. Scale bar = 5 µm. Right: Image series in left-to-tight order: high-resolution images of VGLUT2⁺ presynapses (‘pre’; in grey), tdTomato⁺ tanycytes (in magenta), GluA2 subunits (‘post’, in cyan), and their overlay. Scale bar = 500 nm. g. Current-clamp trace showing the membrane potential (mV) in tanycytes upon current steps. Vertical and horizontal scales are 20 mV and 50 ms, respectively. h. Boxplots define the resting membrane potential (mV) in α- (n = 23) and β-tanycytes (n = 20); t = 0.624; p = 0.536. i. Boxplots showing the membrane resistance (MΩ) in α (n = 23) and β tanycytes (n = 20); t = 1.980; p = 0.054. j. Boxplots depict the membrane capacitance (pF) in α- (n = 23) and β-tanycytes (n = 20); t = 0.960; p = 0.343. k. Boxplots for the time constant (τ) to dissipate current (ms) in α- (n = 23) and β-tanycytes (n = 20); t = 1.974; p = 0.0552. Open circles in h-k are individual data points; data were analyzed by Student’s t-test throughout. l. Representative trace showing the inward deflection of a tonic current in response to s-AMPA (100 µM) and its reversal upon wash-out. Vertical and horizontal scales are 50 pA and 100 s, respectively. m. top: Failure rate of optogenetically-induced EPSCs in tanycytes. Boxplot depicts means ± s.e.m., with individual data points shown as open circles. Bottom: Distribution of the lag-time of EPSCs (ms) relative to the onset of optical stimuli. Data were binned as indicated. Source Data

Extended Data Fig. 7. Neuronal activity triggers Ca²⁺ transients in tanycytes.

a. Top left: GCaMP5g transients in tanycytes after evoking a single action potential (AP) in a synaptically-connected neuron in brain slices superfused with ACSF. Images show Ca²⁺ transients in tanycytes (frame response, orange-fire, arrow) subsequent to injecting current in a patch-clamped neuron (frame AP, neuron is marked by arrow; see also Supplementary Videos 1, 2). Bottom left: Representative traces of GCaMP5g relative fluorescence intensity transients (F/F₀) in tanycytes following an evoked AP (arrows). Top right: NBQX (20 µM) blocked GCaMP5g-encoded Ca²⁺ transients in tanycytes (frame response, orange-fire, arrow) following AP induction in the same neuron (frame AP; see also Supplementary Video 3). Bottom right: Representative traces of GCaMP5g F/F₀ transients in tanycytes following evoked AP (arrows). Time was expressed as frames with inter-frame interval of 300 ms. b. AMPA (100 µM)-induced GCaMP5g-encoded Ca²⁺ transients in tanycytes. Confocal time series (20x) from 300-µm coronal hypothalamic slices from Rax-CreER^T2::PC-G5-tdT mice. Left: Representative frames of GCaMP5g basal activity as F/F₀ transients in tanycytes superfused with ACSF. Middle: AMPA (100 µM)-induced increase in GCaMP5g transients. Right: The effect of AMPA was abolished by NBQX (20 µM; see also Supplementary Videos 4–6). Time was expressed as frames with inter-frame interval of 600 ms. Scale bar = 20 µm. c. PTX (100 µM)-induced GCaMP5g-encoded Ca²⁺ transients in tanycytes. Confocal time series (20x) from Rax-CreER^T2::PC-G5-tdT mice as above. Left: Sequential frames of basal GCaMP5g activity (F/F₀) when tanycytes were superfused with ACSF. Middle: PTX (100 µM)-induced increase in GCaMP5g transients, which were substantially reduced when co-applying TTX (5 µM, right; see also Supplementary Videos 7–9). Scale bars = 20 µm. Time was expressed as frames with an inter-frame interval of 600 ms. d. Top: KCl (50 mM)-induced tanycyte activation. Confocal time series (20x) from 300-µm coronal hypothalamic slices from Rax-CreER^T2::PC-G5-tdT mice. Left: Representative frames of GCaMP5g basal activity (F/F₀,) in tanycytes superfused with ACSF. Right: KCl (50 mM)-induced increase in GCaMP5g-encoded Ca²⁺ events in tanycytes (F/F₀, trace; see also Supplementary Videos 10, 11). Scale bars = 20 µm. Time was expressed as frames with an inter-frame interval of 600 ms.

Extended Data Fig. 8. Neuropeptides, SNARE proteins, and AMPA receptor subunits in tanycytes.

a. Fold changes for differentially-expressed neuropeptides and signaling molecules in tanycytes (group ‘1’) relative to ependymocytes, astrocytes, tanycytes ‘2’, and two clusters of neurons after re-processing open-label single-cell RNA-seq data from ref. . Scaled fold changes with a false discovery rate of > 25% were included. Expression scale: 5 (red), 0 (white), −10 (blue). b. Fold changes for differentially-expressed SNARE complex-related genes and AMPA receptor subunits in tanycytes ‘1’ relative to ependymocytes, astrocytes, tanycytes ‘2’, and two neuronal clusters, with data re-processed from ref. . Scaled fold changes with a false discovery rate of > 25% were included. Expression scale: 2 (red), 0 (white), −2 and −4 (blue). Source Data

Extended Data Fig. 9. Tanycytes produce VEGFA upon heat exposure.

a. Image series showing the analysis pipeline used to quantify the number of Vegfa mRNA in tanycytes at both 25 °C and at 40 °C (1 h) in vivo. Images show the surface reconstruction of tanycytes along the wall of the third ventricle (counterstained by Hoechst 33,342, in grey) and the Vegfa mRNA probe (in cyan). To determine the number of Vegfa mRNAs in apposition to the surface of tanycytes, we used the ‘find-spots-close-to-surface’ Imaris function. 1: Confocal image of tanycytes labelled with Hoechst 33,342 (in grey). 2: Vegfa mRNA signal (in cyan). 3: Vegfa ‘spots’ (spheres, in cyan) rendered for reconstruction by assigning each sphere to 0.4 µm x 0.4 µm in diameter over the Vegfa mRNA signal along the Hoechst 33,342-labelled ventricular surface. 4: Raster shows the number of Vegfa ‘spots’ (in cyan) in α-tanycytes. Scale bar = 20 µm. ‘1’. Fold change of Vegfa mRNA expression in the wall of the 3^rd ventricle microdissected from mouse hypothalami at 25 °C vs. 40 °C. Data in bar graphs correspond to means ± s.e.m. with individual data points shown as circles; n = 4 mice/group; t = 2.751; p = 0.033; Student’s t-test. b. VEGFA immunoreactivity in tanycytes lining the third ventricle (arrows), and its change after thermal manipulation. Left: Grey-scaled VEGFA immunoreactivity. Right: Overlay of VEGFA (in cyan) and vimentin (in pink). Scale bars = 70 μm. ‘1’. Quantitative data on VEGFA immunoreactivity in tanycytes (as in panel ‘b’) from n = 4 mice/group; p < 0.05 (Student’s t-test). c. VEGFA immunoreactivity in tanycytes (arrows), and its change after chemogenetic activation by CNO. Saline was used as vehicle control. Left: Grey-scaled VEGFA immunoreactivity. Right: Overlay of VEGFA (in cyan) and vimentin (in pink). Scale bars = 70 μm. ‘1’. Quantitative data on VEGFA immunoreactivity in tanycytes (as in panel ‘c’) from n = 4 mice/group; p < 0.05 (Student’s t-test). d. VEGFA levels (pg/ml) in the CSF of rats exposed for 1h to 25 °C (in grey, n = 3) or 40 °C (in red, n = 4). Data in bar graphs show means ± s.e.m. with open circles being individual data points; t = 0.306, p = 0.772; Student’s t-test. e. Coincident detection of tdTomato (corresponding to Agrp⁺ neurons; in red) and vimentin⁺ tanycytes (in green) in the arcuate nucleus (ARC). Scale bar = 20 μm. f. Anatomical arrangement between vimentin⁺ tanycytes and Th⁺ neurons in the ARC. Scale bar = 5 µm. g-i. Confocal surveys of the ARC for the distribution of Th⁺, Agrp⁺, and Pomc⁺ neurons (all in magenta), and their co-expression of Flt1 (or not, see for Pomc⁺ neurons). Hoechst 33,342 was used as nuclear counterstain (grey). Scale bars = 50 µm (overviews), 2 µm (i, inset), 1 µm (g, inset). j. Left: Changes in food intake upon exposure to 38 °C vs. 25 °C. Middle: Food intake in mice acutely injected with DMSO (0.001%; used as vehicle) prior to exposure to 38 °C. Right: Pre-treatment with tranilast (20 mg/kg) prior to exposure to 40 °C. Note that tranilast did not alter food intake at the concentration tested. Individual data points are shown from n = 4 male mice/group. *p < 0.05, **p < 0.01 (Student’s t-test, pair-wise comparisons). Data in bar graphs (a‘1’,b‘1’,c‘1’,d) were expressed as means ± s.e.m. throughout. Results in box-and-whisker plots (j) show medians ± interquartile ranges. Source Data

Extended Data Fig. 10. Tanycytes link the parabrachial nucleus to feeding.

a. Left: Immunohistochemical detection of VEGFA (arrows) under control conditions and after Vegfa-specific RNAi infusion in the 3^rd ventricle, with fluorescence intensity quantified in n = 4 mice/group (right). *p < 0.05 (Student’s t-test). Scale bars = 20 μm. b,c. Timeline analysis (facet-wrap) of food intake (b; n = 4/group) and body weight (c, n = 8/group) expressed in grams (g) over 5 days prior acute thermal manipulation of mice injected with control or Vegfa-RNAi. Repeated-measures ANOVA for food intake: interaction (RNAi vs. days): F = 1.457; p = 0.246; days: F = 57.150; p < 0.001; RNAi: F = 1.033; p = 0.349, and for body weight: interaction (RNAi vs. days): F = 3.010; p = 0.004; days: F = 1.427; p = 0.244; RNAi: F = 0.505; p = 0.489. d. Difference in body weight expressed in g after 24h at 25 °C and after being exposed to 40 °C in control vs. Vegfa-RNAi. Data in bar graphs were expressed as means ± s.e.m. with interconnected circles showing changes per animal (n = 8/group). Repeated-measures ANOVA: interaction (RNAi vs. temperature), F = 0.278; p = 0.614; temperature: F = 3.783, p = 0.093; RNAi: F = 0.095; p = 0.767. e. Diurnal activity expressed as the mean distance moved (cm) 20h after exposure to 25 °C or 40 °C. Time intervals were expressed as zeitgeber (ZT), with ZT12 and ZT0 coinciding with the onset of the dark and light phases, respectively (n = 8/group). Repeated-measures ANOVA: interaction (RNAi vs. temperature): F = 0.049; p = 0.8249; interaction (RNAi vs. time): F = 6.927; p < 0.001; interaction (time vs. temperature): F = 2.510; p < 0.001; time: F = 15.210; p < 0.0001; RNAi: F = 12.620; p = 0.001; temperature: F = 3.908; p = 0.058. f. Benchmarking the specificity of GFP (control) and TeLC-GFP expression in tanycytes after injection of AAV particles in the third ventricle (3V). The vimentin immunosignal was color-coded in light grey to enhance visual clarity, and to highlight somatic GFP fluorescence in only the outermost cell layer of the ventricular wall. Scale bars = 20 μm. g. Vimentin⁺ tanycytes (in magenta) expressed VAMP2 (in cyan). Open rectangle denotes the location of the orthogonal image stack shown in Fig. 4i. Arrows point to tanycyte (T) processes with VAMP2 signal. Scale bar = 2 µm. h. Histochemical controls of the successful transduction of PBN neurons with AAVs expressing either GFP only or a TeLC-GFP fusion construct. Abbreviation: scp, superior cerebellar peduncle. Scale bar = 70 µm. i. Individual bodyweights across the phases of behavioral testing (n = 4; F = 3.504; p = 0.148; repeated-measures one-way ANOVA). j. Facet-wrap timeline plot of cumulative drinking time (s) during 24-h of baseline, in mice exposed to CNO (3 mg kg⁻¹; CNO 1), repeated baseline after TeLC delivery and recombination by 4-hydroxytamoxifen (baseline TeLC), and in mice with AAV2-FLEX-TeLC-GFP recombined in tanycytes and injected with CNO (3 mg/g; ‘CNO #2 TeLC’, in blue), n = 4; repeated-measures two-way ANOVA: interaction: F = 5.281; p < 0.0001; time (24 h): F = 58.510; p < 0.0001; treatment: F = 6.153; p = 0.009; subject: F = 4.936; p < 0.0001. k. Facet-wrap timeline plot of the cumulative distance moved (cm). Groups were identical to (i and j); n = 4; repeated-measures two-way ANOVA: interaction: F = 11.130; p < 0.0001; time (24 h), F = 48.930; p < 0.0001; treatment: F = 47.570; p < 0.0001; subject: F = 3.381; p = 0.0003. Data in facet-wrap plots were expressed as means ± 95% confidence interval of the s.d. throughout. Solid circles represent individual data points at time and treatment. Source Data