Dissecting the contribution of host genetics and the microbiome in complex behaviors.

The core symptoms of many neurological disorders have traditionally been thought to be caused by genetic variants affecting brain development and function. However, the gut microbiome, another important source of variation, can also influence specific behaviors. Thus, it is critical to unravel the contributions of host genetic variation, the microbiome, and their interactions to complex behaviors. Unexpectedly, we discovered that different maladaptive behaviors are interdependently regulated by the microbiome and host genes in the Cntnap2-/- model for neurodevelopmental disorders. The hyperactivity phenotype of Cntnap2-/- mice is caused by host genetics, whereas the social-behavior phenotype is mediated by the gut microbiome. Interestingly, specific microbial intervention selectively rescued the social deficits in Cntnap2-/- mice through upregulation of metabolites in the tetrahydrobiopterin synthesis pathway. Our findings that behavioral abnormalities could have distinct origins (host genetic versus microbial) may change the way we think about neurological disorders and how to treat them.

TMEFF1 is a neuron-specific restriction factor for herpes simplex virus.

The brain is highly sensitive to damage caused by infection and inflammation1,2. Herpes simplex virus 1 (HSV-1) is a neurotropic virus and the cause of herpes simplex encephalitis3. It is unknown whether neuron-specific antiviral factors control virus replication to prevent infection and excessive inflammatory responses, hence protecting the brain. Here we identify TMEFF1 as an HSV-1 restriction factor using genome-wide CRISPR screening. TMEFF1 is expressed specifically in neurons of the central nervous system and is not regulated by type I interferon, the best-known innate antiviral system controlling virus infections. Depletion of TMEFF1 in stem-cell-derived human neurons led to elevated viral replication and neuronal death following HSV-1 infection. TMEFF1 blocked the HSV-1 replication cycle at the level of viral entry through interactions with nectin-1 and non-muscle myosin heavy chains IIA and IIB, which are core proteins in virus-cell binding and virus-cell fusion, respectively4-6. Notably, Tmeff1-/- mice exhibited increased susceptibility to HSV-1 infection in the brain but not in the periphery. Within the brain, elevated viral load was observed specifically in neurons. Our study identifies TMEFF1 as a neuron-specific restriction factor essential for prevention of HSV-1 replication in the central nervous system.

Human TMEFF1 is a restriction factor for herpes simplex virus in the brain.

Most cases of herpes simplex virus 1 (HSV-1) encephalitis (HSE) remain unexplained1,2. Here, we report on two unrelated people who had HSE as children and are homozygous for rare deleterious variants of TMEFF1, which encodes a cell membrane protein that is preferentially expressed by brain cortical neurons. TMEFF1 interacts with the cell-surface HSV-1 receptor NECTIN-1, impairing HSV-1 glycoprotein D- and NECTIN-1-mediated fusion of the virus and the cell membrane, blocking viral entry. Genetic TMEFF1 deficiency allows HSV-1 to rapidly enter cortical neurons that are either patient specific or derived from CRISPR-Cas9-engineered human pluripotent stem cells, thereby enhancing HSV-1 translocation to the nucleus and subsequent replication. This cellular phenotype can be rescued by pretreatment with type I interferon (IFN) or the expression of exogenous wild-type TMEFF1. Moreover, ectopic expression of full-length TMEFF1 or its amino-terminal extracellular domain, but not its carboxy-terminal intracellular domain, impairs HSV-1 entry into NECTIN-1-expressing cells other than neurons, increasing their resistance to HSV-1 infection. Human TMEFF1 is therefore a host restriction factor for HSV-1 entry into cortical neurons. Its constitutively high abundance in cortical neurons protects these cells from HSV-1 infection, whereas inherited TMEFF1 deficiency renders them susceptible to this virus and can therefore underlie HSE.

CLSTN3ß enforces adipocyte multilocularity to facilitate lipid utilization.

Multilocular adipocytes are a hallmark of thermogenic adipose tissue1,2, but the factors that enforce this cellular phenotype are largely unknown. Here, we show that an adipocyte-selective product of the Clstn3 locus (CLSTN3ß) present in only placental mammals facilitates the efficient use of stored triglyceride by limiting lipid droplet (LD) expansion. CLSTN3ß is an integral endoplasmic reticulum (ER) membrane protein that localizes to ER-LD contact sites through a conserved hairpin-like domain. Mice lacking CLSTN3ß have abnormal LD morphology and altered substrate use in brown adipose tissue, and are more susceptible to cold-induced hypothermia despite having no defect in adrenergic signalling. Conversely, forced expression of CLSTN3ß is sufficient to enforce a multilocular LD phenotype in cultured cells and adipose tissue. CLSTN3ß associates with cell death-inducing DFFA-like effector proteins and impairs their ability to transfer lipid between LDs, thereby restricting LD fusion and expansion. Functionally, increased LD surface area in CLSTN3ß-expressing adipocytes promotes engagement of the lipolytic machinery and facilitates fatty acid oxidation. In human fat, CLSTN3B is a selective marker of multilocular adipocytes. These findings define a molecular mechanism that regulates LD form and function to facilitate lipid utilization in thermogenic adipocytes.

Heteromeric clusters of ubiquitinated ER-shaping proteins drive ER-phagy.

Membrane-shaping proteins characterized by reticulon homology domains play an important part in the dynamic remodelling of the endoplasmic reticulum (ER). An example of such a protein is FAM134B, which can bind LC3 proteins and mediate the degradation of ER sheets through selective autophagy (ER-phagy)1. Mutations in FAM134B result in a neurodegenerative disorder in humans that mainly affects sensory and autonomic neurons2. Here we report that ARL6IP1, another ER-shaping protein that contains a reticulon homology domain and is associated with sensory loss3, interacts with FAM134B and participates in the formation of heteromeric multi-protein clusters required for ER-phagy. Moreover, ubiquitination of ARL6IP1 promotes this process. Accordingly, disruption of Arl6ip1 in mice causes an expansion of ER sheets in sensory neurons that degenerate over time. Primary cells obtained from Arl6ip1-deficient mice or from patients display incomplete budding of ER membranes and severe impairment of ER-phagy flux. Therefore, we propose that the clustering of ubiquitinated ER-shaping proteins facilitates the dynamic remodelling of the ER during ER-phagy and is important for neuronal maintenance.

The endothelial protein PLVAP in lymphatics controls the entry of lymphocytes and antigens into lymph nodes.

In the lymphatic sinuses of draining lymph nodes, soluble lymph-borne antigens enter the reticular conduits in a size-selective manner and lymphocytes transmigrate to the parenchyma. The molecular mechanisms that control these processes are unknown. Here we unexpectedly found that PLVAP, a prototypic endothelial protein of blood vessels, was synthesized in the sinus-lining lymphatic endothelial cells covering the distal conduits. In PLVAP-deficient mice, both small antigens and large antigens entered the conduit system, and the transmigration of lymphocytes through the sinus floor was augmented. Mechanistically, the filtering function of the lymphatic sinus endothelium was dependent on diaphragms formed by PLVAP fibrils in transendothelial channels. Thus, in the lymphatic sinus, PLVAP forms a physical sieve that regulates the parenchymal entry of lymphocytes and soluble antigens.

TRPV4 Channel Signaling in Macrophages Promotes Gastrointestinal Motility via Direct Effects on Smooth Muscle Cells.

Intestinal macrophages are critical for gastrointestinal (GI) homeostasis, but our understanding of their role in regulating intestinal motility is incomplete. Here, we report that CX3C chemokine receptor 1-expressing muscularis macrophages (MMs) were required to maintain normal GI motility. MMs expressed the transient receptor potential vanilloid 4 (TRPV4) channel, which senses thermal, mechanical, and chemical cues. Selective pharmacologic inhibition of TRPV4 or conditional deletion of TRPV4 from macrophages decreased intestinal motility and was sufficient to reverse the GI hypermotility that is associated with chemotherapy treatment. Mechanistically, stimulation of MMs via TRPV4 promoted the release of prostaglandin E2 and elicited colon contraction in a paracrine manner via prostaglandin E receptor signaling in intestinal smooth muscle cells without input from the enteric nervous system. Collectively, our data identify TRPV4-expressing MMs as an essential component required for maintaining normal GI motility and provide potential drug targets for GI motility disorders.

STING controls nociception via type I interferon signalling in sensory neurons.

The innate immune regulator STING is a critical sensor of self- and pathogen-derived DNA. DNA sensing by STING leads to the induction of type-I interferons (IFN-I) and other cytokines, which promote immune-cell-mediated eradication of pathogens and neoplastic cells1,2. STING is also a robust driver of antitumour immunity, which has led to the development of STING activators and small-molecule agonists as adjuvants for cancer immunotherapy3. Pain, transmitted by peripheral nociceptive sensory neurons (nociceptors), also aids in host defence by alerting organisms to the presence of potentially damaging stimuli, including pathogens and cancer cells4,5. Here we demonstrate that STING is a critical regulator of nociception through IFN-I signalling in peripheral nociceptors. We show that mice lacking STING or IFN-I signalling exhibit hypersensitivity to nociceptive stimuli and heightened nociceptor excitability. Conversely, intrathecal activation of STING produces robust antinociception in mice and non-human primates. STING-mediated antinociception is governed by IFN-Is, which rapidly suppress excitability of mouse, monkey and human nociceptors. Our findings establish the STING-IFN-I signalling axis as a critical regulator of physiological nociception and a promising new target for treating chronic pain.

FAM72A antagonizes UNG2 to promote mutagenic repair during antibody maturation.

Activation-induced cytidine deaminase (AID) catalyses the deamination of deoxycytidines to deoxyuracils within immunoglobulin genes to induce somatic hypermutation and class-switch recombination1,2. AID-generated deoxyuracils are recognized and processed by subverted base-excision and mismatch repair pathways that ensure a mutagenic outcome in B cells3-6. However, why these DNA repair pathways do not accurately repair AID-induced lesions remains unknown. Here, using a genome-wide CRISPR screen, we show that FAM72A is a major determinant for the error-prone processing of deoxyuracils. Fam72a-deficient CH12F3-2 B cells and primary B cells from Fam72a-/- mice exhibit reduced class-switch recombination and somatic hypermutation frequencies at immunoglobulin and Bcl6 genes, and reduced genome-wide deoxyuracils. The somatic hypermutation spectrum in B cells from Fam72a-/- mice is opposite to that observed in mice deficient in uracil DNA glycosylase 2 (UNG2)7, which suggests that UNG2 is hyperactive in FAM72A-deficient cells. Indeed, FAM72A binds to UNG2, resulting in reduced levels of UNG2 protein in the G1 phase of the cell cycle, coinciding with peak AID activity. FAM72A therefore causes U·G mispairs to persist into S phase, leading to error-prone processing by mismatch repair. By disabling the DNA repair pathways that normally efficiently remove deoxyuracils from DNA, FAM72A enables AID to exert its full effects on antibody maturation. This work has implications in cancer, as the overexpression of FAM72A that is observed in many cancers8 could promote mutagenesis.

NF1 mutation drives neuronal activity-dependent initiation of optic glioma.

Neurons have recently emerged as essential cellular constituents of the tumour microenvironment, and their activity has been shown to increase the growth of a diverse number of solid tumours1. Although the role of neurons in tumour progression has previously been demonstrated2, the importance of neuronal activity to tumour initiation is less clear-particularly in the setting of cancer predisposition syndromes. Fifteen per cent of individuals with the neurofibromatosis 1 (NF1) cancer predisposition syndrome (in which tumours arise in close association with nerves) develop low-grade neoplasms of the optic pathway (known as optic pathway gliomas (OPGs)) during early childhood3,4, raising  the possibility that postnatal light-induced activity of the optic nerve drives tumour initiation. Here we use an authenticated mouse model of OPG driven by mutations in the neurofibromatosis 1 tumour suppressor gene (Nf1)5 to demonstrate that stimulation of optic nerve activity increases optic glioma growth, and that decreasing visual experience via light deprivation prevents tumour formation and maintenance. We show that the initiation of Nf1-driven OPGs (Nf1-OPGs) depends on visual experience during a developmental period in which Nf1-mutant mice are susceptible to tumorigenesis. Germline Nf1 mutation in retinal neurons results in aberrantly increased shedding of neuroligin 3 (NLGN3) within the optic nerve in response to retinal neuronal activity. Moreover, genetic Nlgn3 loss or pharmacological inhibition of NLGN3 shedding blocks the formation and progression of Nf1-OPGs. Collectively, our studies establish an obligate role for neuronal activity in the development of some types of brain tumours, elucidate a therapeutic strategy to reduce OPG incidence or mitigate tumour progression, and underscore the role of Nf1mutation-mediated dysregulation of neuronal signalling pathways in mouse models of the NF1 cancer predisposition syndrome.

Tonic prime-boost of STING signalling mediates Niemann-Pick disease type C.

The classic mode of STING activation is through binding the cyclic dinucleotide 2'3'-cyclic GMP-AMP (cGAMP), produced by the DNA sensor cyclic GMP-AMP synthase (cGAS), which is important for the innate immune response to microbial infection and autoimmune disease. Modes of STING activation that are independent of cGAS are much less well understood. Here, through a spatiotemporally resolved proximity labelling screen followed by quantitative proteomics, we identify the lysosomal membrane protein Niemann-Pick type C1 (NPC1) as a cofactor in the trafficking of STING. NPC1 interacts with STING and recruits it to the lysosome for degradation in both human and mouse cells. Notably, we find that knockout of Npc1 'primes' STING signalling by physically linking or 'tethering' STING to SREBP2 trafficking. Loss of NPC1 protein also 'boosts' STING signalling by blocking lysosomal degradation. Both priming and boosting of STING signalling are required for severe neurological disease in the Npc1-/- mouse. Genetic deletion of Sting1 (the gene that encodes STING) or Irf3, but not that of Cgas, significantly reduced the activation of microglia and relieved the loss of Purkinje neurons in the cerebellum of Npc1-/- mice, leading to improved motor function. Our study identifies a cGAS- and cGAMP-independent mode of STING activation that affects neuropathology and provides a therapeutic target for the treatment of Niemann-Pick disease type C.

mTORC1 Activation Requires DRAM-1 by Facilitating Lysosomal Amino Acid Efflux.

Sensing nutrient availability is essential for appropriate cellular growth, and mTORC1 is a major regulator of this process. Mechanisms causing mTORC1 activation are, however, complex and diverse. We report here an additional important step in the activation of mTORC1, which regulates the efflux of amino acids from lysosomes into the cytoplasm. This process requires DRAM-1, which binds the membrane carrier protein SCAMP3 and the amino acid transporters SLC1A5 and LAT1, directing them to lysosomes and permitting efficient mTORC1 activation. Consequently, we show that loss of DRAM-1 also impacts pathways regulated by mTORC1, including insulin signaling, glycemic balance, and adipocyte differentiation. Interestingly, although DRAM-1 can promote autophagy, this effect on mTORC1 is autophagy independent, and autophagy only becomes important for mTORC1 activation when DRAM-1 is deleted. These findings provide important insights into mTORC1 activation and highlight the importance of DRAM-1 in growth control, metabolic homeostasis, and differentiation.

MFSD12 mediates the import of cysteine into melanosomes and lysosomes.

Dozens of genes contribute to the wide variation in human pigmentation. Many of these genes encode proteins that localize to the melanosome-the organelle, related to the lysosome, that synthesizes pigment-but have unclear functions1,2. Here we describe MelanoIP, a method for rapidly isolating melanosomes and profiling their labile metabolite contents. We use this method to study MFSD12, a transmembrane protein of unknown molecular function that, when suppressed, causes darker pigmentation in mice and humans3,4. We find that MFSD12 is required to maintain normal levels of cystine-the oxidized dimer of cysteine-in melanosomes, and to produce cysteinyldopas, the precursors of pheomelanin synthesis made in melanosomes via cysteine oxidation5,6. Tracing and biochemical analyses show that MFSD12 is necessary for the import of cysteine into melanosomes and, in non-pigmented cells, lysosomes. Indeed, loss of MFSD12 reduced the accumulation of cystine in lysosomes of fibroblasts from patients with cystinosis, a lysosomal-storage disease caused by inactivation of the lysosomal cystine exporter cystinosin7-9. Thus, MFSD12 is an essential component of the cysteine importer for melanosomes and lysosomes.

Violet-light suppression of thermogenesis by opsin 5 hypothalamic neurons.

The opsin family of G-protein-coupled receptors are used as light detectors in animals. Opsin 5 (also known as neuropsin or OPN5) is a highly conserved opsin that is sensitive to visible violet light1,2. In mice, OPN5 is a known photoreceptor in the retina3 and skin4 but is also expressed in the hypothalamic preoptic area (POA)5. Here we describe a light-sensing pathway in which POA neurons that express Opn5 regulate thermogenesis in brown adipose tissue (BAT). We show that Opn5 is expressed in glutamatergic warm-sensing POA neurons that receive synaptic input from several thermoregulatory nuclei. We further show that Opn5 POA neurons project to BAT and decrease its activity under chemogenetic stimulation. Opn5-null mice show overactive BAT, increased body temperature, and exaggerated thermogenesis when cold-challenged. Moreover, violet photostimulation during cold exposure acutely suppresses BAT temperature in wild-type mice but not in Opn5-null mice. Direct measurements of intracellular cAMP ex vivo show that Opn5 POA neurons increase cAMP when stimulated with violet light. This analysis thus identifies a violet light-sensitive deep brain photoreceptor that normally suppresses BAT thermogenesis.

Structure and mechanism of the ER-based glucosyltransferase ALG6.

In eukaryotic protein N-glycosylation, a series of glycosyltransferases catalyse the biosynthesis of a dolichylpyrophosphate-linked oligosaccharide before its transfer onto acceptor proteins1. The final seven steps occur in the lumen of the endoplasmic reticulum (ER) and require dolichylphosphate-activated mannose and glucose as donor substrates2. The responsible enzymes-ALG3, ALG9, ALG12, ALG6, ALG8 and ALG10-are glycosyltransferases of the C-superfamily (GT-Cs), which are loosely defined as containing membrane-spanning helices and processing an isoprenoid-linked carbohydrate donor substrate3,4. Here we present the cryo-electron microscopy structure of yeast ALG6 at 3.0 Å resolution, which reveals a previously undescribed transmembrane protein fold. Comparison with reported GT-C structures suggests that GT-C enzymes contain a modular architecture with a conserved module and a variable module, each with distinct functional roles. We used synthetic analogues of dolichylphosphate-linked and dolichylpyrophosphate-linked sugars and enzymatic glycan extension to generate donor and acceptor substrates using purified enzymes of the ALG pathway to recapitulate the activity of ALG6 in vitro. A second cryo-electron microscopy structure of ALG6 bound to an analogue of dolichylphosphate-glucose at 3.9 Å resolution revealed the active site of the enzyme. Functional analysis of ALG6 variants identified a catalytic aspartate residue that probably acts as a general base. This residue is conserved in the GT-C superfamily. Our results define the architecture of ER-luminal GT-C enzymes and provide a structural basis for understanding their catalytic mechanisms.

Recapitulating the human segmentation clock with pluripotent stem cells.

Pluripotent stem cells are increasingly used to model different aspects of embryogenesis and organ formation1. Despite recent advances in in vitro induction of major mesodermal lineages and cell types2,3, experimental model systems that can recapitulate more complex features of human mesoderm development and patterning are largely missing. Here we used induced pluripotent stem cells for the stepwise in vitro induction of presomitic mesoderm and its derivatives to model distinct aspects of human somitogenesis. We focused initially on modelling the human segmentation clock, a major biological concept believed to underlie the rhythmic and controlled emergence of somites, which give rise to the segmental pattern of the vertebrate axial skeleton. We observed oscillatory expression of core segmentation clock genes, including HES7 and DKK1, determined the period of the human segmentation clock to be around five hours, and demonstrated the presence of dynamic travelling-wave-like gene expression in in vitro-induced human presomitic mesoderm. Furthermore, we identified and compared oscillatory genes in human and mouse presomitic mesoderm derived from pluripotent stem cells, which revealed species-specific and shared molecular components and pathways associated with the putative mouse and human segmentation clocks. Using CRISPR-Cas9-based genome editing technology, we then targeted genes for which mutations in patients with segmentation defects of the vertebrae, such as spondylocostal dysostosis, have been reported (HES7, LFNG, DLL3 and MESP2). Subsequent analysis of patient-like and patient-derived induced pluripotent stem cells revealed gene-specific alterations in oscillation, synchronization or differentiation properties. Our findings provide insights into the human segmentation clock as well as diseases associated with human axial skeletogenesis.

The innate immunity protein IFITM3 modulates γ-secretase in Alzheimer's disease.

Innate immunity is associated with Alzheimer's disease1, but the influence of immune activation on the production of amyloid-ß is unknown2,3. Here we identify interferon-induced transmembrane protein 3 (IFITM3) as a γ-secretase modulatory protein, and establish a mechanism by which inflammation affects the generation of amyloid-ß. Inflammatory cytokines induce the expression of IFITM3 in neurons and astrocytes, which binds to γ-secretase and upregulates its activity, thereby increasing the production of amyloid-ß. The expression of IFITM3 is increased with ageing and in mouse models that express familial Alzheimer's disease genes. Furthermore, knockout of IFITM3 reduces γ-secretase activity and the formation of amyloid plaques in a transgenic mouse model (5xFAD) of early amyloid deposition. IFITM3 protein is upregulated in tissue samples from a subset of patients with late-onset Alzheimer's disease that exhibit higher γ-secretase activity. The amount of IFITM3 in the γ-secretase complex has a strong and positive correlation with γ-secretase activity in samples from patients with late-onset Alzheimer's disease. These findings reveal a mechanism in which γ-secretase is modulated by neuroinflammation via IFITM3 and the risk of Alzheimer's disease is thereby increased.

Rescue of oxytocin response and social behaviour in a mouse model of autism.

A fundamental challenge in developing treatments for autism spectrum disorders is the heterogeneity of the condition. More than one hundred genetic mutations confer high risk for autism, with each individual mutation accounting for only a small fraction of cases1-3. Subsets of risk genes can be grouped into functionally related pathways, most prominently those involving synaptic proteins, translational regulation, and chromatin modifications. To attempt to minimize this genetic complexity, recent therapeutic strategies have focused on the neuropeptides oxytocin and vasopressin4-6, which regulate aspects of social behaviour in mammals7. However, it is unclear whether genetic risk factors predispose individuals to autism as a result of modifications to oxytocinergic signalling. Here we report that an autism-associated mutation in the synaptic adhesion molecule Nlgn3 results in impaired oxytocin signalling in dopaminergic neurons and in altered behavioural responses to social novelty tests in mice. Notably, loss of Nlgn3 is accompanied by a disruption of translation homeostasis in the ventral tegmental area. Treatment of Nlgn3-knockout mice with a new, highly specific, brain-penetrant inhibitor of MAP kinase-interacting kinases resets the translation of mRNA and restores oxytocin signalling and social novelty responses. Thus, this work identifies a convergence between the genetic autism risk factor Nlgn3, regulation of translation, and oxytocinergic signalling. Focusing on such common core plasticity elements might provide a pragmatic approach to overcoming the heterogeneity of autism. Ultimately, this would enable mechanism-based stratification of patient populations to increase the success of therapeutic interventions.

PGRMC2 is an intracellular haem chaperone critical for adipocyte function.

Haem is an essential prosthetic group of numerous proteins and a central signalling molecule in many physiologic processes1,2. The chemical reactivity of haem means that a network of intracellular chaperone proteins is required to avert the cytotoxic effects of free haem, but the constituents of such trafficking pathways are unknown3,4. Haem synthesis is completed in mitochondria, with ferrochelatase adding iron to protoporphyrin IX. How this vital but highly reactive metabolite is delivered from mitochondria to haemoproteins throughout the cell remains poorly defined3,4. Here we show that progesterone receptor membrane component 2 (PGRMC2) is required for delivery of labile, or signalling haem, to the nucleus. Deletion of PGMRC2 in brown fat, which has a high demand for haem, reduced labile haem in the nucleus and increased stability of the haem-responsive transcriptional repressors Rev-Erbα and BACH1. Ensuing alterations in gene expression caused severe mitochondrial defects that rendered adipose-specific PGRMC2-null mice unable to activate adaptive thermogenesis and prone to greater metabolic deterioration when fed a high-fat diet. By contrast, obese-diabetic mice treated with a small-molecule PGRMC2 activator showed substantial improvement of diabetic features. These studies uncover a role for PGRMC2 in intracellular haem transport, reveal the influence of adipose tissue haem dynamics on physiology and suggest that modulation of PGRMC2 may revert obesity-linked defects in adipocytes.

Innervation of thermogenic adipose tissue via a calsyntenin 3ß-S100b axis.

The sympathetic nervous system drives brown and beige adipocyte thermogenesis through the release of noradrenaline from local axons. However, the molecular basis of higher levels of sympathetic innervation of thermogenic fat, compared to white fat, has remained unknown. Here we show that thermogenic adipocytes express a previously unknown, mammal-specific protein of the endoplasmic reticulum that we term calsyntenin 3ß. Genetic loss or gain of expression of calsyntenin 3ß in adipocytes reduces or enhances functional sympathetic innervation, respectively, in adipose tissue. Ablation of calsyntenin 3ß predisposes mice on a high-fat diet to obesity. Mechanistically, calsyntenin 3ß promotes endoplasmic-reticulum localization and secretion of S100b-a protein that lacks a signal peptide-from brown adipocytes. S100b stimulates neurite outgrowth from sympathetic neurons in vitro. A deficiency of S100b phenocopies deficiency of calsyntenin 3ß, and forced expression of S100b in brown adipocytes rescues the defective sympathetic innervation that is caused by ablation of calsyntenin 3ß. Our data reveal a mammal-specific mechanism of communication between thermogenic adipocytes and sympathetic neurons.