Sympathetic neuropeptide Y protects from obesity by sustaining thermogenic fat

Abstract

Human mutations in neuropeptide Y (NPY) have been linked to high body mass index but not altered dietary patterns¹. Here we uncover the mechanism by which NPY in sympathetic neurons^2,3 protects from obesity. Imaging of cleared mouse brown and white adipose tissue (BAT and WAT, respectively) established that NPY⁺ sympathetic axons are a smaller subset that mostly maps to the perivasculature; analysis of single-cell RNA sequencing datasets identified mural cells as the main NPY-responsive cells in adipose tissues. We show that NPY sustains the proliferation of mural cells, which are a source of thermogenic adipocytes in both BAT and WAT^4-6. We found that diet-induced obesity leads to neuropathy of NPY⁺ axons and concomitant depletion of mural cells. This defect was replicated in mice with NPY abrogated from sympathetic neurons. The loss of NPY in sympathetic neurons whitened interscapular BAT, reducing its thermogenic ability and decreasing energy expenditure before the onset of obesity. It also caused adult-onset obesity of mice fed on a regular chow diet and rendered them more susceptible to diet-induced obesity without increasing food consumption. Our results indicate that, relative to central NPY, peripheral NPY produced by sympathetic nerves has the opposite effect on body weight by sustaining energy expenditure independently of food intake.

Fig. 1. One-third of sympathetic axons in iWAT and iBAT are NPY⁺ and preferentially innervate the vasculature.

a, Confocal images of an axonal bundle dissected from iWAT of a male lean adult mouse, stained with anti-TH (red) and anti-NPY (cyan). b, Confocal images of a capillary within cleared iWAT from a lean adult mouse, stained with anti-CD31 (red) and anti-NPY (cyan). c, Confocal images of cleared iWAT from lean adult male WT mice, stained with anti-TH (green), anti-NPY (white) and anti-CD31 (red). d, Confocal images of cleared iWAT in c from lean adult male mice, stained with anti-TH (green), anti-NPY (white) and anti-CD31 (red). e,f, Quantification of images in d for TH/NPY overlap: TH⁺NPY⁻ (dark green), TH⁺NPY⁺ (light green) and TH⁻NPY⁺ (white); percentages shown on a histogram (e) and Venn diagram (f); n = 5 views from three mice. g, Quantification of images in d showing the proportions of TH⁺ and NPY⁺ axons overlapping with CD31⁺ endothelial cells (n = 4 views from three mice; P = 0.0035). h, Confocal images of cleared iBAT from lean adult male mice stained with anti-TH (green), anti-NPY (white) and anti-CD31 (red). Arrowheads indicate TH⁺NPY⁻ axons. a–d,h, Representative images are shown from three experiments. All values shown are mean ± s.e.m. Statistical comparisons were made using two-tailed Student’s t-tests, **P < 0.01. Scale bars, 50 μm (a,c), 100 μm (b,d,h).

Fig. 2. Npy1r is mainly expressed in mural cells that are a source of thermogenic adipocytes.

a,c, Dot plots showing the expression of Npy1r, Des, Myh11, Acta2 (αSMA), Tagln, Cspg4 (NG2), Pdgfrb, Pdgfra and Pecam1 (CD31) in the SVF of iWAT (a) and iBAT (c). b,d, Embedding plots showing the expression of Npy1r, Des, Myh11 and Acta2 (αSMA) in the SVF of iWAT (b) and iBAT (d). Mural cell clusters are highlighted by red dashed lines. e, Expression of Npy1r (P = 0.0002) and Des (P = 0.0024) in sorted CD31⁻CD45⁻PDGFRα⁻NG2⁺ mural cells and CD45⁺ immune cells (n = 5 and n = 3 biologically independent samples, respectively). Gapdh was used as the reference gene. f, Confocal images of capillary dissected from iWAT of WT mice and stained with anti-NPY1R (cyan) and anti-CD31 (red). g, Confocal images of vessels dissected from Npy1r^Cre;Rosa26^tdTomato mice stained with anti-DES (green). h, Confocal images of vessels dissected from iBAT of ND-treated, room temperature-housed, 12-week-old Npy1r^Cre;Rosa26^tdTomato mice stained with anti-UCP1 (cyan). i, Zoomed-in images of h. f–i, Representative images from two experiments. All values mean ± s.e.m. Statistical comparisons were made using two-tailed Student’s t-tests, **P < 0.01, ***P < 0.001. NK, natural killer cells; NKT, nature killer T cells; MonoMF, monocytes and macrophages. UMAP, uniform manifold approximation and projection. Scale bars, 10 μm (f), 20 μm (g), 100 μm (h), 50 μm (i).

Fig. 3. NPY promotes the proliferation of both iWAT and iBAT mural cells.

a, Schematic illustrating isolation of mural cells from iWAT. b, Confocal images showing iWAT mural cells treated with or without 1 μM NPY and stained with DES (cyan) or DAPI; 5-ethynyl-2′-deoxyuridine (EdU) (red) indicates proliferating cells. c, Percentage of EdU⁺ mural cells quantified based on images in b (n = 10 biologically independent samples; P = 0.0008). d, Schematic showing isolation of mural cells from iBAT. e, Confocal images showing iBAT mural cells treated with or without 1 μM NPY and stained with DES (cyan) and DAPI. EdU (red) indicates proliferating cells. f, Percentage of EdU⁺ mural cells quantified based on images in d (n = 3 biologically independent samples; P = 0.0370). b,e, Representative images are shown from three and two experiments, respectively. All values mean ± s.e.m. Statistical comparisons were made using two-tailed Student’s t-tests, *P < 0.05, ***P < 0.001. NC, negative control (without NPY). Scale bars, 80 μm.

Fig. 4. Loss of NPY from sympathetic neurons depletes mural cells and increases vascular leakiness in both iBAT and iWAT.

a,b, Light-sheet images of cleared iBAT from 15-week-old ND-treated mice (a) and of iWAT from 30-week-old ND-treated mice (b) stained with anti-PV1 (red) and anti-TAGLN (cyan). c,d, Expression levels of mural cell markers Des (n = 8 mice; P = 0.0007) and Rgs5 (n = 7 mice; P = 0.0374) (c) and proinflammatory genes Tnf (n = 7 mice, P = 0.0284) and Il6 (n = 7 mice, P = 0.0004) (d) in BAT of ND-treated, 12-week-old, male Th^Cre;Npy^flox/flox and Npy^flox/flox mice. e, Percentage of DES⁺ mural cells in iBAT (Npy^flox/flox versus Th^Cre;Npy^flox/flox, P = 0.0084) and iWAT (Npy^flox/flox versus Th^Cre;Npy^flox/flox, P = 0.0038) of 17-week-old, HFD-treated mice, measured by flow cytometry (n = 4 mice). a,b, Representative images from three experiments. All values mean ± s.e.m. Statistical comparisons were made using two-tailed Student’s t-tests, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Scale bars, 200 μm (a), 500 μm (b). Hypoxanthine–guanine phosphoribosyltransferase (Hprt) was used as the reference gene for qPCR.

Fig. 5. Loss of NPY from sympathetic neurons whitens BAT, decreases energy expenditure and thermogenesis and renders mice more susceptible to DIO without increasing food intake.

a, iBAT dissected from ND-treated, 12-week-old, male Npy^flox/flox and Th^Cre;Npy^flox/flox mice. Dashed lines delineate iBAT lobes. b, Expression of thermogenic genes Ucp1, Cidea (P = 0.0024), Pgc1a (P = 0.0018) and Prdm16 (P = 0.0048), and of adipogenic gene Pparg (P = 0.0012), in iBAT from ND-treated, 12-week-old, male Th^Cre;Npy^flox/flox and Npy^flox/flox mice (n = 6 mice for Ucp1, n = 7 mice for other genes). c, Daily energy expenditure of 7-week-old male Th^Cre;Npy^flox/flox and Npy^flox/flox mice, measured using an indirect calorimetry system (n = 9). d,e, Changes in iBAT temperature (Th^Cre versus Th^Cre;Npy^flox/flox mice, P = 0.0198; Npy^flox/flox versus Th^Cre;Npy^flox/flox mice, P = 0.0440) (d) and body weight (Th^Cre versus Th^Cre;Npy^flox/flox mice, P = 0.0008; Npy^flox/flox versus Th^Cre;Npy^flox/flox mice, P = 0.0225) (e) of 12-week-old mice following a 14-h fast (n = 6 for Th^Cre;Npy^flox/flox mice and n = 3 for Th^Cre and Npy^flox/flox mice). f–i, Body weight (week 30, P = 0.0003) (f) and weights of iBAT (week 12, P = 0.0006; week 24, P = 0.0387; week 30, P = 0.0058) (g), iWAT (week 30, P = 0.0004) (h) and visceral WAT (vWAT) (week 24, P = 0.0338; week 30, P = 0.0030) (i) from ND-treated, male Th^Cre;Npy^flox/flox and Npy^flox/flox mice (n = 4 mice). j,k, Weekly body weight (n = 8; week 13, P = 0.0007; weeks 14–16, P < 0.0001) (j) and averaged daily food consumption (n = 4 and n = 6 cages) (k) from male, HFD-treated, Th^Cre;Npy^flox/flox and Npy^flox/flox mice. l,m, Weights of iBAT (n = 5 and n = 6 mice, P = 0.0023) (l), and iWATs (n = 6 and n = 7 mice, P = 0.0002) and vWAT (n = 6 and n = 7 mice) (m), from HFD-treated, 17-week-old, male Th^Cre;Npy^flox/flox and Npy^flox/flox mice. k–m, Colour coding as in j. n, Average adipocyte size in iBAT from ND- and HFD-treated male Th^Cre;Npy^flox/flox and Npy^flox/flox mice (n = 2 for ND-treated mice, n = 4 for HFD-treated mice, P = 0.0436). a, Representative images from three experiments. All values mean ± s.e.m. Statistical comparisons were made using two-tailed Student’s t-tests, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Hprt was used as the reference gene for qPCR.

Extended Data Fig. 1. Npy is expressed by a subpopulation of sympathetic neurons, and NPY⁺ axons preferentially innervate the 5^th order arteriole branches; NPY variants are significantly associated with human body mass index (BMI).

(a) Schematic showing the tissues innervated by each sympathetic ganglia. (b) Single-cell RNA-Seq dataset of sympathetic ganglia chain published by Linnarsson Lab to show the clustering of neurons. (c) Violin plot showing the expression of Npy. (d-e) Confocal image of cryo-sectioned (d) superior-cervical ganglion (SCG) and (e) stellate ganglia (SG) of ND-treated WT male 8-week-old mice stained with anti-TH (red) and anti-NPY (cyan). Scale bar = 100 μm. (f) The percentage of NPY⁺ sympathetic neurons quantified based on images as in Extended Data Fig. 5d,e (n = 5 sections from 3 mice). (g) Confocal images of cleared adipose tissue stained with anti-NPY (cyan) and anti-CD31 (red). The numbers indicate the branch orders. Scale bar = 150 μm. (h) Zoom-in of Extended Data Fig. 1g, scale bar = 50 μm. (i-j) (i) The diameters of vessels and (j) the density of NPY⁺ axons at each branch level quantified based on images as in Extended Data Fig. 1g (n = 6 views from 2 mice); NPY⁺ axon density on Branch #1 vs. #5, P < 0.0001. (k) Light-sheet images of cleared adipose tissue stained with anti-NPY (red) and anti-EPHB4 (cyan). Scale bar = 500 μm. (l) Confocal images of cleared adipose tissue stained with anti-NPY (red) and anti-SOX17 (cyan). Scale bar = 100 μm. (m) Plot showing associations of human NPY gene variants with different traits (https://hugeamp.org/gene.html?gene=NPY). For d-h, representative images were shown from 3 experiments; for k,l, representative images were shown from 2 experiments. All values are expressed as mean ± SEM. Statistical comparisons were made using 2-tailed Student T-tests, ****P < 0.0001.

Extended Data Fig. 2. NPY1R-tdTomato is colocalised with NG2 and αSMA in mural cells.

(a) Embedding plot showing the co-expression of mural markers Acta2 (αSMA), Myh11, and Tagln (red) with Npy1r (green). Co-expression is in yellow. (b) Confocal images of vessels dissected from adipose tissue of Npy1r^Cre; Rosa26^tdTomato mice stained with anti-αSMA (green), scale bar = 100 μm. (c) Zoomed-in images of vessels as in Extended Data Fig. 2b, scale bar = 20 μm. (d) Confocal images of vessels dissected from Npy1r^Cre; Rosa26^tdTomato mice stained with anti-αSMA (green) and anti-NG2 (cyan), scale bar = 100 μm. (e) Zoomed-in images of vessels as in Extended Data Fig. 2b, scale bar = 20 μm. For b-e, representative images were shown from 2 experiments.

Extended Data Fig. 3. Previously identified progenitors of thermogenic adipocytes (Lin⁻/CD81⁺) in iWAT express Npy1r and mural cell markers, and mural cells isolated from both iWAT and iBAT can differentiate into thermogenic adipocytes.

(a) Schematic of isolating previously identified progenitors of thermogenic adipocytes (Lin⁻/CD81⁺) in iWAT. (b-c) The expression level of (b) Npy1r (P = 0.0027), and (c) mural cell markers Des (P = 0.0003) and Acta2 (P = 0.0009) in CD81⁻ vs. CD81⁺ progenitors (n = 4 biologically independent samples). (d) Dot plot showing the expression of Npy1r, Des, Myh11, Acta2, Tagln, Pdgfrb, Cspg4, Pdgfra, Pecam1, and Cd81 in different cells in the SVF of mice iWAT. The size of the dot represents the percentage of cells expressing a certain gene, and the darkness of the dot represents the expression level. NPY1R⁺ mural cell clusters are highlighted with red rectangles. (e) Schematic of isolating mural cells from adipose tissue. (f) The expression level of mural cell markers Des, Pdgfrb, and Acta2 in isolated mural cells. The red dashed line indicates the expression level of reference gene Hprt (n = 4 biologically independent samples). (g,i) Schematic of isolating and differentiating mural cells from (g) iWAT and (i) iBAT. (h,j) The expression of thermogenic and adipogenic genes in adipocytes differentiated from the stromal-vascular fraction (SVF) and mural cells isolated from (h) iWAT (n = 5&6 biologically independent samples; Cidea P = 0.0006, Pgc1a P = 0.0037, Pparg P = 0.0.0364) and (j) iBAT (n = 5&5 biologically independent samples for Ucp1 and Cidea P = 0.0071; n = 5&3 biologically independent samples for Pgc1a P = 0.0001 and Pparg P = 0.0003). All values are expressed as mean ± SEM. Statistical comparisons were made using 2-tailed Student T-tests, *p < 0.05, **p < 0.01, ***p < 0.001. The icon of scissors in c,g,i is from Flaticon (https://flaticon.com). Hprt was used as the reference gene for qPCR.

Extended Data Fig. 4. NPY promotes the proliferation of mural cells via ERK1/2 and increases the proportion of mural cells ex vivo.

(a) Confocal images of isolated mural cells treated with or without 1 μM NPY and 2 μM PD98059, an ERK inhibitor, stained with anti-DES (cyan) and DAPI (blue). EdU (red) was used to indicate proliferation. Scale bar = 80 μm. (b) The percentage of EdU⁺ cells quantified based on images as in Extended Data Fig. 4a (n = 6 biologically independent samples). (c) Schematic of isolating iWAT SVF and CD81⁺ cells for NPY treatment experiments and flow cytometry analysis. (d) Gating strategy for measuring the percentage of αSMA⁺ mural cells in cultured SVF and CD81⁺ cells by flow cytometry. (e-f) The percentage of αSMA⁺ mural cells in (e) SVF (n = 6 biologically independent samples, P = 0.0191) and (e) CD81⁺ cells (n = 8 biologically independent samples, P = 0.0141) after incubating with or without 1 μM NPY. (g) Schematic showing NPY treatment increases the percentage of mural cells in cultured SVF or CD81⁺ cells, and mural cells are the progenitor of thermogenic adipocytes. For a, representative images were shown from 3 experiments. All values are expressed as mean ± SEM. Statistical comparisons were made using 2-tailed Student T-tests, *p < 0.05. The icon of scissors in c is from Flaticon (https://flaticon.com).

Extended Data Fig. 5. NPY facilitates the neogenesis of thermogenic adipocytes from iWAT SVF and mural progenitors isolated from both iWAT and iBAT.

(a) Schematic of differentiating primary stromal vascular fraction (SVF) of iWATs to thermogenic adipocytes. (b) The expression Ucp1, Cidea, Pgc1a, Serca2b, Pparg, and Plin1 (n = 4&3&3 biologically independent samples for Plin1 and n = 3&5&5 biologically independent samples for other genes) in thermogenic adipocytes differentiated from primary SVF with or without 1 μΜ NPY. No Induction vs. Induction: Ucp1, P = 0.0052; Pgc1a, P = 0.0006; Pparg, P = 0.0001. Induction vs. Induction+NPY: Ucp1, P = 0.0350; Pgc1a, P = 0.0113; Serca2b, P = 0.0012; Pparg, P = 0.0190. (c) Schematic of differentiating beige adipocyte progenitors (Lin⁻/CD81⁺) isolated from iWAT. (d) The expression of thermogenic and adipogenic genes in adipocytes differentiated from iWAT beige adipocyte progenitors (Lin⁻/CD81⁺) with or without 1 μM NPY (n = 4 biologically independent samples). No Induction vs. Induction: Ucp1, P = 0.0027; Cidea, P = 0.0043; Pgc1a, P = 0.0001; Prdm16, P = 0.0261; Dio2, P = 0.0052; Elvol3, P = 0.0246; Pparg&Adipoq, P < 0.0001. Induction vs. Induction+NPY: Ucp1&Cidea&Dio2, P < 0.0001; Prdm16, P = 0.0071; Elvol3, P = 0.0400. (e) Oxygen consumption in adipocytes differentiated from beige adipocyte progenitors (Lin⁻/CD81⁺) with or without 1 μM NPY (n = 5 biologically independent samples per group). At 60 min, P = 0.0077; at 70 min, P = 0.0056. (f) Schematic of differentiating iBAT mural cells (PDGFRβ⁺) to thermogenic adipocytes. (g) The expression of thermogenic and adipogenic genes in adipocytes differentiated from iBAT mural cells or SVF (n = 4&5&5). No Induction vs. Induction: Ucp1, P = 0.0002; Pgc1a, P < 0.0001; Prdm16, P = 0.0261; Pparg, P = 0.0040. Induction vs. Induction+NPY: Ucp1, P = 0.0259; Pgc1a, P = 0.0072. All values are expressed as mean ± SEM. Statistical comparisons were made using 2-tailed Student T-tests (b,d,g) or 2-way ANOVA (e), *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Hprt was used as the reference gene for qPCR.

Extended Data Fig. 6. NPY sustains mural cells and their ability to differentiate into thermogenic adipocytes against PDGF-BB.

(a) Confocal images of primary mural cells stained with anti-DES (cyan) and anti-PDGFRα (red), cultured under different conditions. Scale bar = 100 μm. (b) The percentage of PDGFRα⁺ cells under different stimulating conditions (n = 6 biologically independent samples). No drug vs. PDGF-BB, P = 0.0138; PDGF-BB vs. PDGF-BB + NPY, P < 0.0001. (c-e) The expression of (c) Npy1r (n = 8 biologically independent samples), (d) mural cell markers Pdgfrb (n = 8 biologically independent samples), and Rgs5 (n = 5&8&8 biologically independent samples) and (e) secretory factors in maintaining vascular integrity includingTgfb1 (n = 8) and Angpt1 (n = 4&8&8 biologically independent samples) in primary mural cells cultured with or without 1 μΜ NPY or 250 ng/mL PDGF-BB. No drug vs. PDGF-BB: Npy1r, P = 0.0005, Pdgfrb&Rgs5, P < 0.0001; Angpt1, P = 0.0032. PDGF-BB vs. PDGF-BB + NPY: Npy1r, P = 0.0004; Rgs5, P = 0.0051; Tgfb1, P = 0.0043; Angpt1, P = 0.0001. (f) Schematic of differentiating mural cells with or without 1 μΜ NPY or 250 ng/mL PDGF-BB. (g-h) The expression of (g) thermogenic and adipogenic genes in adipocytes differentiated from mural cells with or without 1μΜ NPY or 250 ng/mL PDGF-BB (n = 5&6 biologically independent samples). No drug vs. PDGF-BB: Ucp1, P = 0.0001; Cidea, P = 0.0019; Prdm16, P = 0.0260; Pparg, P = 0.0045. PDGF-BB vs. PDGF-BB + NPY: Ucp1, P = 0.0154; Prdm16, P = 0.0240; Pparg, P = 0.0134. For a, representative images were shown from 3 experiments. All values are expressed as mean ± SEM. Statistical comparisons were made using 2-tailed Student T-tests, *p < 0.05. Hprt was used as the reference gene for qPCR.

Extended Data Fig. 7. High-fat diet-induced obesity (DIO) depletes NPY⁺ innervation and mural cells and leads to vascular leakiness.

(a) Light-sheet images of cleared iWAT of normal diet (ND)-treated and high-fat diet (HFD)-treated 17-week-old male WT mice stained with anti-NPY (cyan) and anti-CD31 (red). Scale bar = 500 μm. (b) NPY⁺ innervation calculated as the ratio of NPY to CD31 quantified based on images as in Extended Data Fig. 7a (n = 3 mice, P = 0.0112). (c,d) The concentration of NPY in (c) iWAT (n = 5&4&4 mice) and (d) blood plasma of ND and HFD-treated 17-week-old male WT mice, and ND-treated Ob/Ob 17-week-old male mice (n = 6&7&2 mice). P = 0.0002 for both comparisons. (e) Light-sheet images of cleared iWAT of ND- and HFD-treated 17-week-old male WT stained with anti-DES (cyan) and anti-CD31 (red). Scale bar = 500 μm. (f) Mural cell coverage calculated as the ratio of DES⁺ cells to CD31⁺ cells quantified based on images as in Extended Data Fig. 7e (n = 3 mice, P = 0.0021). (g) Light-sheet images of iWATs of ND- and HFD-treated 17-week-old male mice stained with anti-DES (cyan) and anti-PV1 (red). Scale bar = 500 μm. (h) PV-1 density quantified based on images as in Extended Data Fig. 7g (n = 3&4 mice, P = 0.0053). For a,e,g, representative images are shown from 3 experiments. All values are expressed as mean ± SEM. Statistical comparisons were made using 2-tailed Student T-tests, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Hprt was used as the reference gene for qPCR.

Extended Data Fig. 8. Th^Cre;Npy^flox/flox mice have depleted NPY from the sympathetic nervous system and unchanged sympathetic innervation, central or plasma NPY level, causing mural cell depletion.

(a,b) The expression of Npy and Th in (a) SCG (n = 3&4 mice; Npy, P = 0.0190) and (b) SG (n = 5 mice; Npy, P = 0.0003) of 12-week-old Th^Cre;Npy^flox/flox mice and Npy^flox/flox male mice. (c,d) The percentage of (c) PDGFRα⁻ /NG2⁺ mural cells (n = 4&2 mice) and (d) CD45⁺ immune cells (n = 2&3 mice) in iBATs and iWATs of 30-week-old, ND-treated mice measured by flow cytometry. (e,f) The image of cleared iWAT of 12-week-old Th^Cre;Npy^flox/flox mice and Npy^flox/flox male mice stained with (e) anti-NPY (cyan) or (f) anti-TH (green) and anti-CD31 (red). Scale bar = 500 μm. (g,h) Light-sheet image of cleared iBAT of 12-week-old Th^Cre;Npy^flox/flox mice and Npy^flox/flox male mice stained with (g) anti-NPY (cyan), scale bar = 500 μm, or (h) anti-TH (green), scale bar = 200 μm. (i) The gating strategy of flow cytometry to quantify the percentage of CD45⁺ immune cells DES⁺ mural cells, used for Fig. 4e. (j) The representative flow cytometry analysis of the percentage of mural cells (DES⁺) in the iBAT (top) and iWAT (bottom) of HFD-treated male Th^Cre;Npy^flox/flox mice and Npy^flox/flox mice. For e-h, representative images are shown from 2 experiments. All values are expressed as mean ± SEM. Statistical comparisons were made using 2-tailed Student T-tests, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Gapdh was used as the reference gene for qPCR.

Extended Data Fig. 9. Loss of sympathetic NPY increases vascular leakiness, decreases energy expenditure (EE) and cold tolerance, inhibits thermogenesis and beiging, and increases respiratory exchange ratio (RER) and lipid droplet size.

(a) Intravital microscopy of epididymal WATs of 23-week-old male Npy^flox/flox and Th^Cre;Npy^flox/flox mice 12 min after i.v. injection of 70 kDa Dextran at the orbital plexus. (b) Fluorescence intensity ratio between tissue parenchyma and vessel calculated based on images as in Extended Data Fig. 9a (n = 5&4 mice). P = 0.0390. (c) Dextran intensity in the vessels of eWAT right after i.v. injection (n = 5&4 mice). (d) Dextran intensity in vessels over time measured with a 2-photon microscope. Arrow indicates the time when Dextran was injected. (e) Blood flow calculated using the average of the raw intensity of Dextran inside the artery over 100 s right after the stabilization of the Dextran signal following administration (n = 5&4 mice). (f) The quantification of daily energy expenditure (EE) recorded and quantified using an indirect calorimetry system (n = 9 mice). Full day, P = 0.0416; dark, P = 0.0288. (g) The percentage of lean body weight (LBM) of 8-week-old male mice measured by MiniSpec MRI (n = 9). (h) The locomotor activity measured using Multitake cage (n = 6 mice). (i) Respiratory exchange ratio (RER) of 8-week-old mice recorded using a metabolic cage. (j) The quantification of (i) (n = 9 mice). Full day, P = 0.0173; light, P = 0.0136; dark, P = 0.0243. (k) Rectal temperature of 20-week-old Npy^flox/flox and Th^Cre;Npy^flox/flox mice under cold exposure recorded using a rectal probe every hour (n = 3&4 mice). 8 h, P = 0.0158. (l,m) The expression of thermogenic and adipogenic genes in (l) iBATs (n = 3&4 mice) and (m) iWATs (n = 3 mice) of cold-challenged Npy^flox/flox and Th^Cre;Npy^flox/flox mice. iBAT: Ucp1, P = 0.0106; Dio2, P = 0.0012; Prdm16, P = 0.0385; Pparg, P = 0.0412. iWAT: Prdm16, P = 0.0024. (n,o) Histological staining of BATs of (n) ND- and (o) HFD-treated 17-week-old male Th^Cre;Npy^flox/flox and Npy^flox/flox mice. Scale bar = 50 μm. For a, representative images are shown from 4 experiments. All values are expressed as mean ± SEM. Statistic comparisons were made by 2-tailed Student T-tests (b,l,m) or 2-way ANOVA (f,j,k), *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Hprt was used as the reference gene for qPCR.

Extended Data Fig. 10. Ablating NPY from sympathetic neurons using Th^Cre;Npy^flox/flox mice does not affect NPY levels in VTA, hypothalamus or plasma, and ablating NPY from the hindbrain does not affect body weight, BAT weight or thermogenesis.

(a,c) Confocal image showing a slice of (a) ventral tegmental area (VTA, Scale bar = 500 μm) and (c) arcuate nucleus (Arc, Scale bar = 200 μm) of the hypothalamus of an 8-week-old NPY-GFP male mouse stained with anti-TH (red), anti-GFP (green), and DAPI (blue). (b,d) Quantification of TH/NPY-GFP overlap in (b) VTA and (d) hypothalamus (n = 3 mice). (e) The expression of Npy, Th, and Cre in the hypothalamus of 12-week-old Th^Cre;Npy^flox/flox mice and Npy^flox/flox male mice (n = 3). (f) NPY concentration in the blood plasma of 12-week-old Th^Cre;Npy^flox/flox and Npy^flox/flox male mice (n = 6&5). (g) Schematic of knocking out NPY in the hindbrain by AAV injection. (h) Schematic of metabolic phenotyping of mice with NPY knocked out from the hindbrain (Hindbrain: Npy^flox/flox; AAV-CRE-GFP) and WT (Npy^flox/flox; AAV-GFP) mice. (i) Confocal images of the sliced hindbrain of Hindbrain: Npy^flox/flox; AAV-CRE-GFP and WT mice stained with anti-NPY (magenta). Scale bar = 100 μm. (j) Weekly body weight of ND-treated male Hindbrain: Npy^flox/flox; AAV-CRE-GFP and WT mice (n = 7). (k) iBAT weight of 14-week-old ND-treated male Hindbrain: Npy^flox/flox; AAV-CRE-GFP and WT mice (n = 7). (l) iBAT temperature change after a 14-hour fast (n = 7). (m) The expression levels of thermogenic and adipogenic genes in the iBAT of RT-housed ND-treated male Hindbrain: Npy^flox/flox; AAV-CRE-GFP and WT mice (n = 5). For a,c,i, representative images are shown from 3 experiments. All values are expressed as mean ± SEM. Hprt was used as the reference gene for qPCR.