Brain glucose sensing is critical for survival during hypoglycaemia and tunes the level of defended blood glucose, which goes up in diabetes. Neuronal glucose sensing neurons and mechanisms have been identified, but how these neurons access blood concentrations of glucose to adjust their output and maintain glucose homeostasis is unclear. Here, we demonstrate that adult oligodendrogenesis in the median eminence (ME) is modulated by changes in circulating glucose levels and rapidly upregulated by hypoglycaemia. We show that genetic blockade of new OL production in adult mice impairs the regulation of glucose homeostasis, the integrity of the ME blood-hypothalamus barrier, and neuronal glucose sensing in the arcuate nucleus of the hypothalamus (ARH). Unexpectedly, functional integrity of adult-formed myelin is not required for the maintenance of glucose homeostasis. Instead, newly formed OLs exert their glucoregulatory actions via the synthesis of A disintegrin and metallopeptidase with thrombospondin motifs 4 (ADAMTS4), a metallopeptidase expressed exclusively by OLs and dependent on adult OL genesis to maintain its expression in the ME. Both lack of Adamts4 and ADAMTS4 gain-of-function are associated with impaired glucose homeostasis and remodelling of the blood-hypothalamus barrier, indicating that optimal ADAMTS4 expression is required for the integrity of vascular permeability and normal glycaemic control. Finally, we show that ME ADAMTS4 expression is regulated by changes in peripheral glycaemia and is dysregulated in diabetes, providing a mechanism by which ME OLs contribute to the regulation of glucose homeostasis. ### Competing Interest Statement The authors have declared no competing interest.

Evidence from mouse models suggests that brain serotonergic pathways control blood glucose. We hypothesized that sumatriptan (5HT1B‐receptor agonist) would alter glucose homeostasis in humans.

Discrete brain areas, including hypothalamic nuclei, have been implicated in glucose homeostasis. Some contain molecular machinery capable of detecting changes in blood glucose such as glucokinase (GK) , a low affinity hexokinase important for pancreatic beta cell glucose-sensing. A subset of arcuate nucleus AGRP neurons express GK. We created GKAGRP knockout (KO) mice by crossing AGRP Cre with GK floxed mice, hypothesizing that they would display impaired gluco-regulation. Despite established sex differences in metabolic neurocircuitry, female mice are not always studied. In pilots, we found impaired glucose handling in i.p. glucose tolerance tests in older GKAGRP KO females, but not younger female, nor male mice. Here, we perform detailed phenotyping, using sequential euglycemic (with stable glucose isotope tracers) and hyperglycemic clamp studies in conscious, free moving, 19-23 week old female mice with implanted carotid artery and jugular vein catheters. As shown below, we found that GKAGRP KO females were insulin resistant with a lower dextrose infusion rate and reduced suppression of hepatic glucose production during euglycemic clamps. They also displayed reduced insulin secretory responses during hyperglycemic clamps. Our data support a sexually-dimorphic central gluco-regulatory role for AGRP neurons which requires the presence of GK to facilitate normal insulin release and action. Disclosure M.Josipovic: Research Support; Novo Nordisk A/S. E.O.Staricoff: Research Support; Novo Nordisk A/S. C.H.Riches: Research Support; Novo Nordisk. L.Sheppard: None. M.Evans: Advisory Panel; Pila Pharma, Zucara Therapeutics, Other Relationship; Abbott Diabetes, Dexcom, Inc., Medtronic, Novo Nordisk, Research Support; AstraZeneca, Sanofi, Speaker’s Bureau; Lilly Diabetes. Funding European Union’s Innovative Medicines Initiative 2 Joint Undertaking (grant agreement 777460) also supported by EFPIA, T1D Exchange, JDRF, International Diabetes Federation (IDF) , The Leona M. and Harry B. Helmsley Charitable Trust, Wellcome Trust, Medical Research Council UK and Gates Cambridge Trust.

A century after the life-transforming discovery and purification of insulin, many people living with type 1 diabetes (T1D) are not reaching glycaemic goals.1,2 Therapeutic approaches to help people with T1D achieve glucose targets and (equally importantly) reduce the burden of living with diabetes include structured education, new insulins and technology for delivering insulin, measuring glucose, decision support and closed loop technology to automate insulin delivery. The Freestyle Libre was first launched in Europe in 2014 with uptake in UK use having increased since it was made available on NHS prescription in 2017. There have been changes from the original device with the introduction of an algorithm to improve accuracy and the launch of the second generation Libre2 device allowing the optional use of alarms. Current UK T1D penetrance is around 50% in England (with a marked increase since April 2019 facilitated by NHS England as part of the NHS Long Term Plan), with higher rates in Scotland, Wales and Northern Ireland. A recent Health Technology Wales guidance has recommended broader use for all people with insulin-treated diabetes, not just T1D.3 Worldwide use is also increasing markedly. This global growth in the use of the Libre has occurred despite the lack of a supporting body of evidence from randomised controlled trials (RCTs) showing lowering of HbA1c, the traditional outcome metric for diabetes trials.4 For readers’ interest, an RCT (FLASH-UK) has been examining this in the UK, with participants with T1D randomised to Libre2 versus control finger prick testing.5 At the time of writing, FLASHUK had just completed follow-up and the results are eagerly anticipated. Despite the current absence of RCT data, there are, however, many real-world observations showing improved clinical outcomes with the Libre.6 In the UK, data show reductions in HbA1c and a striking reduction in severe hypoglycaemia and diabetic ketoacidosis with use of the Libre in Scotland.7 The Association of British Clinical Diabetologists (ABCD) has been running nationwide audits of medications introduced into real-world use in the UK since 2004. An ABCD audit of Libre outcomes has been running since 2017, reporting reduced HbA1c, improved hypoglycaemia awareness and reductions in hospital attendances for dysglycaemia.8 This edition of the journal contains an examination of data from the ABCD Libre audit, asking whether prior structured education affects the outcomes with flash glucose monitoring. In particular, clinical outcomes were compared between those who had undergone Dose Adjustment for Normal Eating (DAFNE) structured education, other structured education or neither. Structured education to support self-management of T1D includes a variety of programmes across the UK and elsewhere with variable approaches/ quality assurance, evidence and governance/structure.9 DAFNE is currently delivered in 99 centres and based on principles of therapeutic education with a written curriculum, multidisciplinary team working with defined accreditation, quality assurance and RCT and real-world evidence for efficacy.10,11 This includes (but is not limited to) equipping participants with the ability to appraise and utilise glucose information judiciously. A priori, it would have been possible to hypothesise that those undergoing structured education/DAFNE might be better placed to interpret and benefit from more comprehensive glucose data provided by the Libre. An alternative hypothesis would be that those who had undergone structured education had already partbenefited from the ability to interpret glucose information and would have less incremental gain from the Libre. Of note, there is a large repository of free online training (including the Diabetes Technology Network-UK resources cited in the paper) targeted specifically at how to use and interpret Libre data which would have been available to all regardless of previous structured education and, indeed, many services would have encouraged or even mandated evidence that people had undergone this more targeted training. The study included 14,880 patients, stratified into three groups based on prior structured education status: 4,215 DAFNE graduates, 3,964 other structured education graduates and 6,701 patients who had not received structured education. The main outcomes were the impact of previous education on glycaemic control assessed by HbA1c levels, and hypoglycaemia awareness measured by the standardised GOLD score. At follow-up, all three groups showed improvements from Libre initiation, with reduced HbA1c (by 8.10 mmol/mol, 6.61 mmol/ mol and 6.22 mmol/mol, respectively) and GOLD score (by 0.33, 0.30 and 0.34, respectively). There was no statistical difference between groups in terms of the magnitude of these changes (p=0.83 for HbA1c, p=0.42 for GOLD). Interestingly, on linear regression modelling, the authors show that a higher baseline 1 Gates Trust Cambridge Scholar/ PhD Student, Wellcome Trust/ MRC Institute of Metabolic Science, University of Cambridge, Cambridge UK 2 University Professor of Diabetic Medicine and Honorary Consultant Physician, Wellcome Trust/ MRC Institute of Metabolic Science & Department of Medicine, University of Cambridge, Cambridge UK