Pathogenic hypothalamic extracellular matrix promotes metabolic disease

The study addresses how metabolic disease, particularly obesity and type 2 diabetes, induces insulin resistance within hypothalamic neurons of the arcuate nucleus (ARC), a key hub for energy and glucose homeostasis. Building on evidence that extracellular matrix (ECM) remodelling (fibrosis) in peripheral tissues contributes to insulin resistance, and that perineuronal nets (PNNs) form around ARC neurons (notably AgRP neurons) and can influence neuronal function, the authors hypothesize that obesity drives pathogenic remodelling of the ARC PNN. This neurofibrosis could impede insulin access and signalling in ARC neurons, promoting neuronal insulin resistance and systemic metabolic dysfunction. The purpose is to define whether ARC ECM remodelling occurs in metabolic disease, determine its turnover dynamics, establish causality for metabolic phenotypes, elucidate mechanisms (insulin access/signalling, neuronal excitability), identify inflammatory drivers, and test therapeutic strategies targeting neurofibrosis.

• Peripheral fibrosis and insulin resistance: Excess ECM deposition in adipose tissue, skeletal muscle, and liver impairs insulin action and signalling, with human and animal studies linking fibrosis severity to insulin resistance.
• Brain ECM and PNNs: Brain ECM remodelling has been noted after injury and in neurological disease. PNNs form around maturing neurons, including AgRP neurons in the ARC, and influence neuronal structure and function; microglia can regulate PNN plasticity. Nutritional status can dynamically remodel ECM in regions adjacent to the ARC (median eminence), suggesting ECM involvement in metabolic regulation.
• Hormonal signalling in ARC: Insulin and leptin modulate ARC neuronal activity and systemic metabolism; obesity is associated with altered neuronal excitability and hormone resistance. BBB transport of leptin remains intact in obesity, whereas central insulin action is impaired in humans and rodents. These observations frame the rationale that hypothalamic ECM remodelling could be a missing mechanistic link between obesity, neuronal insulin resistance, and systemic metabolic disease.

• Animal models: C57BL/6J mice subjected to high-fat, high-sugar (HFHS) diet to induce diet-induced obesity (DIO); additional models included genetically obese/diabetic db/db mice and late-stage type 2 diabetes induced by HFHS plus low-dose streptozotocin.
• Histology and ECM detection: PNNs visualized using Wisteria floribunda agglutinin (WFA) lectin staining in coronal brain sections. PNN area and intensity quantified in the ARC (and control regions: ventromedial hypothalamus, retrosplenial granular cortex). ECM components assayed included hyaluronic acid, link proteins (HAPLN1), tenascin C, and CSPGs (aggrecan, versican, neurocan, brevican). Co-localization analyses identified aggrecan as the predominant CSPG overlapping with WFA-labelled PNN in ARC.
• Glycosaminoglycan composition: Quantitative analysis of enzymatically released chondroitin sulfate glycosaminoglycan (CS-GAG) disaccharides from ARC homogenates via zwitterionic hydrophilic interaction liquid chromatography (ZIC-HILIC) with 2-aminobenzamide labelling.
• PNN turnover (PNN tracker): Developed a pulse-chase approach by stereotaxic intra-ARC injection of biotinylated WFA (WFA-biotin) to label existing PNNs, followed by ex vivo detection of residual pulsed WFA-biotin and total PNN (WFA-FITC) at 0, 1, 3, 5, and 10 weeks. Validation included unilateral pulse (vs saline) and enzymatic digestion with chondroitinase ABC (chABC) to confirm specific labelling of PNN present at pulse time.
• Enzymatic PNN disassembly: Bilateral intra-ARC injections of chABC to digest CS-GAG chains and disassemble PNNs in obese mice; vehicle controls included pair-feeding groups.
• Metabolic phenotyping: Body weight, adiposity, food intake, respiratory calorimetry (energy expenditure, VO2, RER), ambulatory activity. Thermogenesis assessed via gross/histological analysis of inguinal white adipose tissue (ingWAT) and brown adipose tissue (BAT), UCP1 immunohistochemistry, and dermal thermography.
• Glucose homeostasis: Glucose tolerance tests, fasting insulin, HOMA-IR. Hyperinsulinaemic–euglycaemic clamps quantified glucose infusion rate (GIR), endogenous glucose production, and tissue-specific glucose uptake (skeletal muscle, BAT, ingWAT).
• Insulin access and signalling: Peripheral injection of FITC-labelled insulin (insulin-FITC) to assess ARC entry, internalization, and downstream signalling (p-AKT Ser473 and insulin receptor phosphorylation). Cerebrospinal fluid infusion of insulin-FITC to bypass BBB. Controls included FITC alone and leptin.
• In vitro binding assays: Plate-binding of insulin-FITC to coated CSPG mixtures, aggrecan, or chondroitin 4-sulfate (C4S), with modulation by chABC or polyarginine to assess charge-dependent interactions.
• Neuronal electrophysiology: Whole-cell patch clamp in AgRP neurons from DIO mice with or without ARC PNN digestion to measure spontaneous firing, firing frequency, resting membrane potential, and K+ currents (with tetrodotoxin). Pharmacological blockade of insulin receptor (S961) tested insulin-dependence of effects.
• Cell-specific insulin receptor targeting: CRISPR/Cas9-based insulin receptor (Insr) deletion in AgRP neurons using AgRP-Cas9 mice (Agrp-IRES-Cre; Rosa26-LSL-Cas9-GFP) and AAV delivering two Insr sgRNAs (AAV-gIR) vs scrambled control (AAV-gScrambled). Validation by PCR amplicon size shift and impaired insulin signalling in AgRP neurons.
• Inflammation manipulation: Anti-inflammatory AAVs expressing soluble TNFR1α (sTNFR1α) and soluble TGFβ receptor (sTGFβR) delivered to ARC prior to DIO; pro-inflammatory AAVs overexpressing TNFα and TGFβ delivered to ARC of chow-fed mice. Outcomes: hypothalamic inflammatory gene expression, ECM protease/inhibitor gene profiles, PNN metrics (WFA, CS-GAGs), and systemic metabolic phenotypes; interaction with chABC to test causality of ECM remodelling.
• Pharmacology—neurofibrosis inhibition: Fluorosamine (per-O-acetylated-4-F-N-acetylglucosamine), an inhibitor of CS-GAG chain assembly/elongation, delivered intracerebroventricularly (ICV) or intranasally. Biodistribution assessed with biotin-conjugated fluorosamine. Metabolic and signalling endpoints as above; epistasis with AgRP insulin receptor deletion to test neuronal mechanism of action.
• Statistics: Two-tailed t-tests, one-way or two-way ANOVA with multiple comparisons (Dunnett/Tukey), repeated measures where appropriate; linear regression for turnover kinetics; n denotes biologically independent mice; results replicated across independent experiments.

• Obesity induces ARC neurofibrosis: HFHS feeding increased ARC PNN area and intensity (WFA staining) within 4 weeks, with further increases at 8–12 weeks (p<0.0001), while RSG and VMH did not show similar remodelling, indicating ARC specificity.
• CS-GAG augmentation: Quantitative ZIC-HILIC analysis showed significantly elevated CS-GAG disaccharide abundances in ARC of DIO mice, aligning with immunohistochemical PNN augmentation.
• Aggrecan predominates in ARC PNN: Although multiple CSPGs increased with obesity (versican, neurocan, brevican, aggrecan), aggrecan co-localized most with WFA-labelled PNN in ARC, suggesting a principal structural role.
• Neurofibrosis localizes to AgRP neurons: Under chow, ~45% of AgRP and ~24% of POMC neurons were PNN-ensheathed. During obesity, PNN coverage increased specifically around AgRP neurons with enhanced WFA/aggrecan signals, independent of neuron number changes.
• Rapid and region-specific PNN turnover: In chow-fed mice, ARC PNN exhibited a turnover period of ~5 weeks; RSG PNN persisted ≥5 weeks and blood vessel-associated PNN showed no degradation over this period.
• Obesity attenuates PNN degradation: Pulse-chase revealed significantly slower PNN degradation in obese vs lean mice, with pulsed WFA-biotin persisting up to 10 weeks in obese ARC (vs ~5 weeks in lean). Obesity reduced expression of ECM proteases and increased inhibitors in mediobasal hypothalamus, consistent with impaired degradation.
• Disassembling ARC neurofibrosis improves metabolism: Intra-ARC chABC in DIO mice caused progressive weight loss, reduced adiposity, decreased caloric intake, increased energy expenditure and thermogenesis (ingWAT, BAT), improved glucose tolerance and lower HOMA-IR. Hyperinsulinaemic–euglycaemic clamps showed higher GIR and improved tissue glucose uptake. Benefits occurred even before notable weight loss and were recapitulated in db/db mice.
• Neurofibrosis impedes insulin access/signalling: Obese mice showed reduced insulin-FITC entry and diminished insulin-induced p-AKT in ARC; chABC restored both. CSF infusion experiments indicated improved insulin access independent of BBB transport. Effects were insulin-specific (no effect for FITC alone; leptin entry unaffected, and leptin sensitivity not restored by PNN digestion).
• Insulin-PNN interactions: In vitro, insulin-FITC bound CSPG mix, aggrecan, and C4S; binding was abolished by chABC or polyarginine, implicating negatively charged GAGs in insulin sequestration.
• Neuronal physiology: Obesity increased spontaneous firing in AgRP neurons (>82%); PNN digestion reduced this to ~33%, decreased firing frequency and depolarization, and enhanced K+ currents. Upregulation of K+ channels after PNN digestion was blunted by insulin receptor antagonist S961, indicating insulin-dependent normalizing effects.
• AgRP insulin signalling is required: CRISPR-mediated Insr deletion in AgRP neurons attenuated chABC-induced benefits on body weight, adiposity, energy expenditure, food intake, and glucose tolerance, demonstrating that reinstating AgRP insulin signalling mediates key therapeutic effects.
• Inflammation drives neurofibrosis: Anti-inflammatory ARC AAVs (sTNFR1α, sTGFβR) reduced hypothalamic inflammation, normalized ECM remodelling gene profiles, decreased ARC PNN (WFA, CS-GAGs), and improved systemic metabolic parameters and ARC insulin responsiveness. Conversely, ARC overexpression of TNFα/TGFβ in lean mice increased PNN remodelling, induced metabolic dysfunction, and altered ARC circuitry; chABC partially reversed these effects, establishing PNN remodelling as a causal node in inflammation-induced metabolic disease.
• Small-molecule inhibition of neurofibrosis is therapeutic: Central fluorosamine reduced ARC PNN (with minimal effects in other regions), promoted weight loss, reduced adiposity, increased energy expenditure, enhanced satiety, and improved glycaemic control and insulin sensitivity, including in late-stage T2D mice. Benefits depended on intact insulin receptor signalling in AgRP neurons (lost in AAV-gIR mice). Intranasal fluorosamine achieved ARC accumulation, attenuated neurofibrosis, and reproduced metabolic improvements, demonstrating translational feasibility.

This study demonstrates that obesity induces pathological remodelling (neurofibrosis) of the ARC perineuronal net, particularly around AgRP neurons. Neurofibrosis reduces PNN turnover via downregulation of ECM proteases and concomitant upregulation of their inhibitors, resulting in augmented CS-GAG-rich matrices dominated by aggrecan. Mechanistically, negatively charged GAGs bind insulin and impede its penetrance and signalling within the ARC, producing neuronal insulin resistance. Enzymatic or pharmacologic attenuation of ARC neurofibrosis restores insulin access and signalling, normalizes AgRP neuronal excitability (via insulin-dependent enhancement of K+ currents and reduced firing), decreases inhibitory AgRP tone to PVH melanocortin circuits, and corrects whole-body energy balance and glucose homeostasis. Causality is established by the requirement for AgRP insulin receptor signalling for chABC- and fluorosamine-induced metabolic benefits. Furthermore, hypothalamic inflammation is identified as an upstream driver of neurofibrosis; anti-inflammatory interventions prevent PNN remodelling and metabolic dysfunction, whereas pro-inflammatory signalling is sufficient to induce both, with PNN disassembly mitigating these effects. Collectively, these findings resolve how hypothalamic ECM remodelling contributes to neuronal and systemic insulin resistance, positioning neurofibrosis as a central, targetable mechanism in metabolic disease.

The work identifies ARC perineuronal net remodelling as a pathogenic driver of obesity-related insulin resistance and metabolic dysfunction. Neurofibrosis impedes insulin access and signalling in AgRP neurons, elevating their excitability and disrupting metabolic neurocircuitry. Disassembling the ARC PNN—enzymatically or via the small-molecule fluorosamine—restores neuronal insulin responsiveness, rebalances energy intake and expenditure, and improves glycaemic control. Inflammation is a causal upstream regulator of neurofibrosis. Therapeutically, targeting hypothalamic ECM with neurofibrosis inhibitors (including intranasal delivery) represents a promising strategy for treating obesity and type 2 diabetes. Future directions include: defining molecular regulators of ARC PNN turnover and region specificity; evaluating long-term safety, specificity, and durability of neurofibrosis inhibition; optimizing brain-targeted delivery platforms; assessing effects across sexes, ages, and comorbidities; and translating to human studies to validate efficacy and mechanisms.