Diabetologia (2023) 66:1557–1575 https://doi.org/10.1007/s00125-023-05935-2 ARTICLE Disruption of cortical cell type composition and function underlies diabetes‑associated cognitive decline Karis Little1 · Aditi Singh1 · Angel Del Marco2,3 · María Llorián‑Salvador1,4,5 · Maria Vargas‑Soria2,3 · Mireia Turch‑Anguera5,6,7,8 · Montse Solé5,6,7,8 · Noëlle Bakker9 · Sarah Scullion1 · Joan X. Comella5,6,7,8 · Ingeborg Klaassen9 · Rafael Simó4,5,10 · Monica Garcia‑Alloza2,3 · Vijay K. Tiwari1,11,12,13 · Alan W. Stitt1 · on behalf of the RECOGNISED consortium Received: 12 December 2022 / Accepted: 28 March 2023 / Published online: 23 June 2023 © Crown 2023, corrected publication 2023 Abstract Aims/hypothesis Type 2 diabetes is associated with increased risk of cognitive decline although the pathogenic basis for this remains obscure. Deciphering diabetes-linked molecular mechanisms in cells of the cerebral cortex could uncover novel therapeutic targets. Methods Single-cell transcriptomic sequencing (scRNA-seq) was conducted on the cerebral cortex in a mouse model of type 2 diabetes (db/db mice) and in non-diabetic control mice in order to identify gene expression changes in distinct cell subpopulations and alterations in cell type composition. Immunohistochemistry and metabolic assessment were used to validate the findings from scRNA-seq and to investigate whether these cell-specific dysfunctions impact the neurovascular unit (NVU). Furthermore, the behavioural and cognitive alterations related to these dysfunctions in db/db mice were assessed via Morris water maze and novel object discrimination tests. Finally, results were validated in post-mortem sections and protein isolates from individuals with type 2 diabetes. Results Compared with non-diabetic control mice, the db/db mice demonstrated disrupted brain function as revealed by losses in episodic and spatial memory and this occurred concomitantly with dysfunctional NVU, neuronal circuitry and cerebral atrophy. scRNA-seq of db/db mouse cerebral cortex revealed cell population changes in neurons, glia and microglia linked to functional regulatory disruption including neuronal maturation and altered metabolism. These changes were validated through immunohistochemistry and protein expression analysis not just in the db/db mouse cerebral cortex but also in post- mortem sections and protein isolates from individuals with type 2 diabetes (74.3 ± 5.5 years) compared with non-diabetic control individuals (87.0 ± 8.5 years). Furthermore, metabolic and synaptic gene disruptions were evident in cortical NVU cell populations and associated with a decrease in vascular density. Conclusions/interpretation Taken together, our data reveal disruption in the cellular and molecular architecture of the cerebral cortex induced by diabetes, which can explain, at least in part, the basis for progressive cognitive decline in individuals with type 2 diabetes. Data availability The single-cell sequencing data that supports this study are available at GEO accession GSE217665 (https://​ www.​ncbi.​nlm.​nih.​gov/​geo/​query/​acc.​cgi?​acc=​GSE21​7665). Keywords Cognitive decline · Cortex · Diabetes · Metabolism · Neuroscience · Neurovascular unit Abbreviations GFAP Glial fibrillary acidic protein DCX Doublecortin GO Gene ontology DEG Differentially expressed gene IHC Immunohistochemistry MWM Morris water maze NeuN Neuronal nuclear protein Karis Little, Aditi Singh and Angel del Marco contributed equally; NVU Neurovascular unit authorship order was determined based on date of joining the project. ROI Region of interest scRNA-seq Single-cell RNA sequencing Monica Garcia-Alloza, Vijay K. Tiwari and Alan W. Stitt share SVZ Subventricular zone senior authorship. WGCN Weighted gene correlation network WT Wild-type Extended author information available on the last page of the article 13 Vol.:(0123456789) 1558 Diabetologia (2023) 66:1557–1575 transcriptomic signatures in the cerebrovasculature in Introduction Huntington’s disease [7] and another study revealed sev- Type 2 diabetes constitutes a huge burden on healthcare sys- eral Alzheimer’s disease risk genes mapping to endothe- tems worldwide due to multi-morbidities such as blindness, lial and microglial cell populations [8]. Furthermore, renal failure and cardiovascular-related death [1]. There is neurodegenerative disease-linked shifts in brain cell mounting evidence that type 2 diabetes is also associated populations have been reported [9]. Neuronal and glial with cognitive decline [2–4]. Individuals with type 2 diabe- population changes have been observed in Alzheimer’s tes have an approximate twofold increased risk of develop- disease [10] and may co-occur with reduced cortical ing Alzheimer’s disease compared with non-diabetic, age- thickness in individuals [11]. matched individuals [3]. Analysis from a single-cell RNA sequencing Clinical evidence and pre-clinical data suggest that com- (scRNA-seq) dataset revealed profound changes to the mon pathways may be involved in type 2 diabetes and cogni- transcriptome of vascular, glial and neuronal cells from tive decline, including glial dysfunction, neurodegeneration the retina of db/db mice, showing a useful linkage to the and inflammation [5, 6], arising from dysfunctional insulin pathogenesis of diabetic retinopathy [12]. The scRNA- signalling, chronic hyperglycaemia and oxidative stress. seq approach has also been used to characterise hip- These pathways may contribute to neurovascular unit (NVU) pocampal changes in db/db mice, indicating the role dysfunction in the brain and retina during type 2 diabetes of microglial populations in diabetes-induced cognitive [4]. However, the underlying pathways that drive neurode- decline [13]. The impact of type 2 diabetes on the brain generation and cognitive decline in type 2 diabetes are not NVU, particularly in the cortex, have not been fully fully understood. explored. Moreover, whether the cognitive decline in The NVU plays an important homeostatic role in the type 2 diabetes is associated with disruption in brain central nervous system and is formed by interdependency cell populations remains unknown. of vascular cells, glia, neurons and professional immune The aim of the present study was to provide novel cells. Besides ensuring integrity of the blood–brain bar- insights into type 2 diabetes-associated cognitive decline rier, the NVU is important for immune privilege and and the underlying associated pathology in the cerebral regulating blood f low to meet the oxygenation and cortex by combining state-of-the-art physiological and nutrient demands of the highly metabolically demand- molecular measurements in db/db mice as a model and ing neuropile. A recent study revealed human-specific validation in human samples. 13 Diabetologia (2023) 66:1557–1575 1559 Table 1  Metabolic characterisation of 14-week-old db/db and control analysed, following procedures shown in ESM Methods. mice ‘What’ was the difference in time exploring familiar and Characteristic Control mice db/db mice recent objects, ‘where’ was the difference in time exploring displaced and non-displaced objects and ‘when’ was the dif- Body weight (g) 25.20 ± 2.17 41.66 ± 2.04** ference between time exploring familiar non-displaced and Glucose (mmol/l) 6.92 ± 0.25 31.00 ± 2.30** recent non-displaced objects. All measurements were per- Insulin (pmol/l) 104.64 ± 20.93 517.96 ± 156.96* formed in triplicate and data were expressed as percentage Data are presented as mean ± SEM and are representative of five or of control values. Discrimination index was calculated as six mice/group (control [WT] three male and two female mice; db/db the percentage of time exploring recent objects/total explo- three male and three female mice) ration time. Body weight, glucose levels and insulin levels were significantly increased in db/db mice when compared with control mice at 14 weeks of age Morris water maze Spatial learning and memory abilities ** were analysed in the Morris water maze (MWM) (see ESM p<0.001 vs control; *p=0.033 vs control Methods). Time required to locate the platform in the acquisi- tion phase, number of entrances in quadrant 2 and swimming Methods velocity during the retention phase were analysed using Smart software version 3.0.04 (Panlab S.L.U., Barcelona, Spain). Animals db/db (C57BLKS/heterozygous db/+ mice [BKS.Cg- +Leprdb/+Leprdb/OlaHds]) (RRID: IMSR_ENV:HSD-173) Tissue dissociation For scRNA-seq, one male homozygous were purchased from Harlan Laboratories (Netherlands) or (db/db) mouse and one male heterozygous (db/+) mouse Janvier (France). Since heterozygous (db/+) mice do not show were killed at 16 weeks of age. Whole cortex was dis- a specific phenotype [14], wild-type (WT) and db/+ mice are sected and dissociated using the papain dissociation system commonly considered a single control group [15, 16]. For For (Worthington, Lakewood, NJ, USA). Cortex samples were immunohistochemistry (IHC) and behavioural studies, WT digested in papain (20 U/ml) for 45 min at 37°C. After treat- mice (BKS.Cg-/OlaHds) belonging to the same colony were ing with ovomucoid, cells were re-suspended in PBS with used as the control group. For transcriptional studies and corti- 0.04% wt/vol. BSA. Cell counts and viability was checked cal metabolic assessment, db/+ mice were used as the control using Trypan Blue (Sigma-Aldrich, Missouri, USA). group. Mice were housed in a standard pathogen-free experi- mental facility, exposed to a 12 h light–dark cycle and had free scRNA‑seq library preparation, sequencing and associated access to food and water. Five or six male/female mice per analysis scRNA-seq libraries were generated using the group aged 14–16 weeks (representing an established but not 10x Genomics Chromium Single Cell 3′ Reagent Kit v3 yet advanced stage of diabetes) were used. Metabolic determi- and sequenced on Illumina NovaSeq6000 S2 200 flow cell nations (Table 1) were carried out as described in electronic at the Genomics Core Technology Unit (GCTU), Queen’s supplementary material (ESM) Methods. University Belfast. Analysis was completed as discussed in ESM Methods using Seurat R package v.4.2.0 (https://​satij​ Actimetry and new object discrimination task Episodic alab.​org/​seurat/; developed by P. Hoffman, S. Lab and col- memory for ‘what,’ ‘where’ and ‘when’ paradigms was laborators, New York City, NY, USA). Table 2  Marker genes used for Cell type Marker gene Gene name identification of cell types Mature neurons Syt1 Synaptotagmin 1 Immature neurons Sox11 SRY-box transcription factor 11 Inhibitory neurons Gad1, Gad2 Glutamate decarboxylase 1, 2 Astrocytes Gja1 Gap junction protein α1 Oligodendrocytes Cldn11 Claudin 11 Oligodendrocyte precursor cells Pdgfrb Platelet-derived growth factor β Microglia Tmem119 Transmembrane protein 119 Endothelial Cldn5 Claudin 5 Pericytes Kcnj8 Potassium inwardly rectifying channel subfamily J member 8 Vascular smooth muscle cells Slc6a13 Solute carrier family 6 member 13 Choroid plexus cells Ttr Transthyretin 13 1560 Diabetologia (2023) 66:1557–1575 Control Acquisition phase 3 Day 0: 2 Day Day Day Day Day time point -8 -7 -6 -5 -1 14–16 weeks IHC and (retention) db/db memory 3 study Actimetry NOD test NOD test Metabolic MWM test 3 test Day 1 Day 2 assessment and brain collection 13 Diabetologia (2023) 66:1557–1575 1561 ◂Fig. 1  Cognitive impairment and brain atrophy in db/db mice. (a) Iba1 and GFAP staining Sections were pre-treated with Schematic detailing study design and time points. (b) The new object Tween-20 (BP337; Fisher Scientific, UK) (0.05%) (wt/vol.) discrimination test showed that episodic memory was not affected in db/db mice when ‘what’ (p=0.441) or ‘where’ (p=0.484) para- in citrate buffer (10 mmol/l) for 2 h at 65°C and blocked in digms were analysed. However, the ‘when’ paradigm was signifi- 3% (wt/vol.) BSA (A7906; Sigma-Aldrich, Madrid, Spain) cantly impaired in db/db mice (*p=0.016). Data are representative and 0.5% (wt/vol.) Triton X100 (BP151; Fisher Scientific) for of five mice. (c) Discrimination index was not affected in db/db mice 1 h. Sections were incubated with anti-Iba1 (ionised calcium (p=0.530). Data are representative of five or six mice. (d) Learning impairment was observed in db/db mice in the acquisition phase of binding adaptor molecule 1) (ESM Table 1) and anti-GFAP the MWM. Times required to locate the platform were significantly (glial fibrillary acidic protein) (ESM Table 1) antibodies longer in db/db mice than control mice in sessions 5–8 (session 1 overnight at 4°C in 3% (wt/vol.) BSA and 0.5% (wt/vol.) Tri- p=0.729, session 2 p=0.637, session 3 p=0.294, session 4 p=0.168, ton X100 followed by incubation with secondary antibodies session 5 ***p<0.001 vs control, session 6 ***p<0.001 vs control, session 7 **p=0.003 vs control, session 8 *p=0.025 vs control). Data (diluted in PBS) (ESM Table 1) for 1 h at room temperature. are representative of five or six mice. White squares, mean value for db/db mice; grey squares, mean value for control mice. (e) The Vascular staining Antigen retrieval was performed using cit- number of entries into quadrant 2 (where the platform was located rate buffer at 95°C for 30 min in a water bath. Slides were along the acquisition phase) was significantly lower in db/db mice (*p<0.05). (f) Brain weight was reduced in db/db mice when com- permeabilised in 2% (wt/vol.) Triton-X for 30 min. Slides pared with control mice (**p<0.001 vs control). (g) Hemisection size were blocked in 5% donkey serum in 0.2% (wt/vol.) Triton- was significantly reduced in db/db mice (*p=0.039 vs control). (h) X for 1 h at room temperature. Primary antibodies (ESM Cortical thickness was significantly reduced in db/db mice (*p=0.013 Table 1) were added and slides were incubated overnight at vs control). (i) Representative images showing reduced cortical thick- ness in db/db mice. Scale bar, 250 µm. Schematic in (a) was created 4°C in 0.5% donkey serum and 0.2% Triton-X followed by with BioRender.com incubation with secondary antibodies (ESM Table 1) for 2 h at room temperature, in PBS. Identification of cell types, marker genes and differential Image analysis Images were taken via Olympus DP71 cam- expression analysis Following procedures detailed in ESM era or Leica SP8 confocal. Counting of cell populations was Methods, cortical cell types were identified by expression performed by selecting 15–20 regions of interest (ROIs) of established markers (Table 2). Differential gene expres- (7960.215 µm2/ROI) from each cortical section (333–417 sion, single-cell regulatory network, weighted gene cor- ROIs/group) and counting using Image J Software version relation network (WGCN) [17] and pseudotime trajectory 1.49 (National Institutes of Health, MD, USA). [18] analyses were performed. Metabolic assessment of cortical brain punches The method Tissue preparation and staining Immediately after mice published by Qi et al (2021) [22] was adapted in order to were killed via overdose of pentobarbital, db/db and control assess glycolysis in acute coronal slices from the cortex of mouse brains were harvested and weighed. Left hemispheres five male db/db and five male db/+ mice, as described in were fixed in 4% wt/vol. paraformaldehyde for 1 week and ESM Methods. The ‘glycolysis stress test’ protocol from Agi- 30 µm coronal brain sections were obtained in a Microm lent (USA) was optimised for use with cortex slices (ESM HM450 microtome (ThermoFisher, Spain). Sections were Table 2). Protocol length and injection timings were as fol- stored in 50% wt/vol. glycerol at −20°C until used. For stain- lows: baseline, 4 cycles; glucose injection, 5 cycles; oligo- ing procedures, consecutive sections were selected at 1.5, mycin A injection, 5 cycles; and 2-deoxy-d-glucose (2-DG), 0.5, −0.5, −1.5, −2.5 and −3.5 mm from Bregma. Staining 10 cycles. Each cycle consisted of 3 min mix, 1 min wait, 2 was carried out for neuronal nuclear protein (NeuN) [19], min measure. Analysis of basal glycolysis, glycolytic capac- BrdU and doublecortin (DCX) [20] as previously described. ity and glycolytic reserve was performed in Wave Software version 2.6.3.5 (Agilent). Cresyl violet staining Sections were processed as described [21] (see ESM Methods). Images were acquired using an Human tissue For IHC staining, post-mortem brain samples optical Olympus Bx60 microscope with an attached Olympus of the frontal cortex of eight individuals were provided by DP71 camera (Olympus, Tokyo, Japan) and mmi CellTools the Netherlands Brain Bank, Amsterdam. The brain samples software version 4.3 (Molecular Machines & Industries, Ech- were from donors with or without diabetes mellitus. Clas- ing, Germany). Hemisection size and cortical thickness were sification of the donors was based on clinical diagnosis and measured using Adobe Photoshop Elements 2.0 version 6.2 confirmed by autopsy. The staining procedures were per- (Adobe, USA) and Image J version 1.49 (National Institutes formed as described in ESM Methods. Antibody information of Health, MD, USA) software. is included in ESM Table 3. 13 1562 Diabetologia (2023) 66:1557–1575 a b Ctrl db/db Ctrl Astrocytes Cortex Endothelial cells Microglia Mature neurons Oligodendrocyte progenitor cells Oligodendrocytes db/db Macrophages / dendritic cells Pericytes 30 Astrocytes / arachnoid barrier cells 20 10 Immature neurons t-SNE2 0 Olfactory ensheathing glia -10 -20 Ependymocytes -30 Choroid plexus epithelial cells Single-cell -30 -20 -10 0 10 20 30 t-SNE1 suspension UMAP Vascular smooth muscle cells c d Rps29 Tspan7 Mt3 5 3 100 6 Expression level 4 90 4 3 2 ASC 80 EC 2 2 MG 1 70 1 mNEUR Proportion of cells OPC 60 0 0 0 VLMC OLG Ctrl db/db Ctrl db/db Ctrl db/db 50 MAC/DC 40 PC ASC/ABC e 8 Apoe Ptgds 5 Rpsa ImmN 8 Expression level 30 OEG 6 6 4 EPC 20 CPC 3 4 4 10 2 2 2 1 0 Ctrl db/db 0 0 0 Ctrl db/db Ctrl db/db Ctrl db/db f g 5 Actb Mbp Gad2 Rps21 6 4 3 Expression level 4 Expression level 3 3 4 2 2 2 2 1 1 1 0 0 0 0 Ctrl db/db Ctrl db/db Ctrl db/db Ctrl db/db Gria1 Egr1 Hba-a2 Meg3 3 3 6 Expression level 3 Expression level 4 2 2 2 2 1 1 1 0 0 0 0 Ctrl db/db Ctrl db/db Ctrl db/db Ctrl db/db 13 Diabetologia (2023) 66:1557–1575 1563 ◂Fig. 2  Single-cell sequencing of diabetic mouse brain reveals changes following the regulations of local ethics committees in Uni- in glial and neuronal populations. (a) Schematic of scRNA-seq data versitat de Barcelona, and in accordance with the Helsinki derivation. db/db and db/+ (control) mouse cortex was dissociated in papain when mice were 16 weeks old. The resulting single-cell solu- Declaration and the Convention of the Council of Europe on tion was sequenced via 10× genomics and analysis was performed Human Rights and Biomedicine. in R. (b) Clustering of the cortex showed similar cell populations in diabetic and control mouse brain. (c) Proportion of each cell type in db/db and db/+ mouse brain cortex, indicating the difference in astro- cytes, microglia and neuronal populations. (d–g) Significant differen- Results tial expression of genes (all p<0.001) in astrocytes (d), microglia (e), mature neurons (f) and immature neurons (g). Expression levels for db/db mice exhibit altered memory function individual genes are presented as the normalised counts. Schematic and reduced cortical thickness The db/db mouse dis- in (a) was created with BioRender.com. ABC, arachnoid barrier cells; ASC, astrocytes; CPC, choroid plexus cells; Ctrl, control; DC, plays key characteristics of type 2 diabetes including obesity, dendritic cells; EC, endothelial cells; EPC, ependymocytes; ImmN, hyperglycaemia and hyperinsulinaemia (Table 1) [21, 23]. immature neurons; MAC, macrophages; MG, microglia; mNEUR, A schematic of the study is shown in Fig. 1a. mature neurons; OEG, olfactory ensheathing glia; OLG, oligodendro- cytes; OPC, oligodendrocyte progenitor cells; PC, pericytes; VLMC, vascular leptomeningeal cells Episodic memory was compromised in db/db mice (Fig. 1b). While no differences were detected for ‘what’ or ‘where’ paradigms, a significant impairment was observed For western blotting, post-mortem brain samples were for ‘when’ paradigm. A similar discrimination index was obtained as frozen pieces of cortical tissue from the neu- observed when comparing the groups of mice (Fig. 1c). rological tissue biobank of Hospital Clínic – IDIBAPS in Assessment of spatial memory using the MWM revealed no Barcelona, Spain. Control donors did not have signs of neu- significant group × session effect (F(1,7)=1.009, p=0.427), rological alterations. Samples from individuals with type 2 although individual session assessment revealed that db/db diabetes clinically diagnosed with dementia were excluded mice had great difficulty in learning the location of the hid- from the study. The most relevant associated comorbidities den platform. This was reflected from session 5 onwards are presented in ESM Table 4. Proteins were extracted and (Fig. 1d). The number of entries into the quadrant where the western blots were run as described in ESM Methods. Anti- platform used to be located was significantly lower in db/ body information is included in ESM Table 3. db mice (Fig. 1e). Swimming speed was reduced in db/db mice (ESM Table 5) and while this may impact the results Analysis and statistics Randomisation and blinding at the of the MWM, we have also shown that walking speed (and experimental phase was not possible due to the obese nature therefore motor activity) is not compromised in these mice. of db/db mice. Quantification and analysis of mouse data At the time of death, brain weight was significantly lower was done by two independent, blinded researchers. Student’s in db/db mice (Fig. 1f) [24]. Cresyl violet staining revealed t test for independent samples, or Mann–Whitney test was that hemisection sizes were also significantly smaller in db/ used to compare control and db/db mouse groups. Two-way db mice (Fig. 1g). db/db mice presented significantly thin- ANOVA (group × session) was used to analyse the acquisi- ner cortices when compared with control mice (Fig. 1h,i). tion phase in the MWM. IBM SPSS Statistics software ver- sion 29.0.0.0 (IBM Corporation, NY, USA) was used for all Single‑cell transcriptome of db/db mouse cortex reveals statistical analysis and results are expressed as mean ± SEM. massive changes in identity and composition of various key cell types Single-cell transcriptome analysis of cortex from Study approval The Animal Care and Use Committee of db/db mice revealed significant changes in neuronal and glial the University of Cadiz, in accordance with the Guidelines populations (Fig. 2a,b). There was a marked increase in the for Care and Use of experimental animals (European Com- microglia population of the db/db mouse cerebral cortex mission Directive 2010/63 /UE and Spanish Royal Decree accompanied by a loss of astrocytes and mature neurons 53/2013), approved all experimental procedures. The Ani- (Fig. 2c). An increased fraction of immature neurons was mal Welfare Ethical Review Body at Queen’s University observed, alongside an increase in oligodendrocyte progeni- Belfast approved this study, and all procedures were con- tors and concomitant decrease in mature oligodendrocytes ducted under the regulation of the UK Home Office Animals (Fig. 2c). Pseudotime trajectory on glia populations (astro- Scientific Research Act 1986. The collection of the human cytes and oligodendrocytes together) (ESM Fig. 1) showed brain material for immunostaining was done in accordance cellular heterogeneity and an altered lineage trajectory in with Dutch laws and the rules outlined by the local ethics db/db vs control mice. These results suggest altered neuronal committee. For western blotting, samples were obtained development, glial development and maturation trajectory in by the biobank after corresponding ethical approvals and the db/db mouse cortex. 13 1564 Diabetologia (2023) 66:1557–1575 a c 2.0 e Pax6 (bits) 0.5 Pax6(+) Nr2e1(+) Pax6 Nr2e1 Expression 3 3 1.0 Regulon specificity score Emx2 (+) level 2 2 Sox2 (+) 1 1 Sox21 (+) Pou3f2 (+) 0.0 0.4 5’ 1 2 3 4 5 6 7 8 9 10 12 14 16 17 18 3’ 0 0 2.0 Nr2e1 (bits) Ctrl db/db Ctrl db/db 1.0 Emx2 4 Sox2 Expression 3 3 level 0.2 0.0 2 2 5’ 1 2 3 4 5 6 7 8 9 10 3’ 2.0 1 1 Emx2 (bits) 0 0 Ctrl db/db Ctrl db/db 1.0 Expression 0.0 4 Lhx2 4 Hes5 0.0 level 3 3 0 100 200 300 400 5’ 1 2 3 4 5 6 7 8 9 3’ 2 2 b Rank d 2.0 1 1 Lhx2 (bits) Sox21(+) 0 0 0.5 Hes5(+) Ctrl db/db Ctrl db/db Lhx2(+) Emx2(+) 1.0 Regulon specificity score Sox6 (+) Pou3f3(+) Pou3f2 Pou3f3 0.4 3 3 Etv4(+) 0.0 Expression Sox2(+) 5’ 1 2 3 4 5 6 7 8 3’ 2.0 2 2 level 0.3 Hes5 (bits) 1 1 1.0 0 0 Ctrl db/db Ctrl db/db 0.2 0.0 5’ 1 2 3 4 5 6 7 8 9 10 3’ 2.0 0.1 Sox6 Sox21 Expression Sox21 (bits) 3 3 level 2 2 1.0 0.0 1 1 0 100 200 300 400 0.0 0 0 5’ 1 2 3 4 5 6 7 8 9 10 3’ Ctrl db/db Ctrl db/db Rank f h j 2.0 Mafb (bits) Mafb Lyl1 Expression Mafb(+) 3 1.0 4 0.5 2 level 3 Regulon specificity score 2 Lyl1(+) 1 1 0.0 0 0.4 5’ 1 2 3 4 5 6 7 8 9 10 11 12 13 3’ 0 Cebpa(+) Irf5(+) Ctrl db/db Ctrl db/db 2.0 Ikzf1(+) Lyl1 (bits) 0.3 Prdm1(+) Irf8(+) 1.0 Cebpa Irf1 3 Expression 3 0.2 Elf4(+) 2 level 0.0 2 5’ 1 2 3 4 5 6 7 8 9 10 11 12 3’ 2.0 1 1 Cebpa (bits) 0.1 0 0 Ctrl db/db Ctrl db/db 1.0 0.0 400 0.0 0 100 200 300 5’ 1 2 3 4 5 6 7 8 9 10 11 12 13 3’ Ikzf1 Prdm1 Expression Rank 2 3 2 i level 2 g 1 Ikzf1 (bits) Mafb(+) 0.6 1 Lyl1(+) 0 0 1 Stat6(+) Ctrl db/db Ctrl db/db Elf4(+) Regulon specificity score Maf(+) Ikzf1(+) 0 5’ 1 2 3 4 5 6 7 8 9 10 11 12 13 3’ Irf8 3 Elf4 2.0 Expression 0.4 Cebpa(+) 3 Irf5(+) 2 Elf4 (bits) level 2 1 1 1.0 Irf8(+) 0 0 Ctrl db/db Ctrl db/db 0.2 0.0 5’ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 3’ 2.0 Stat6 Stat6 (bits) 3 Maf Expression 4 2 level 1.0 3 2 0.0 1 1 0 0 0 100 200 300 400 0.0 5’ 1 2 3 4 5 6 7 8 9 10 3’ Ctrl db/db Ctrl db/db Rank Fig. 3  Regulon analysis. (a–d) Top regulons (a, b) and top unique motifs in db/+ (f, h) and db/db (g, i) mouse microglia. (j) Differential expres- (c, d) in db/+ (a, c) and db/db (b, d) mouse astrocytes. (e) Differential sion patterns of top regulon genes in db/+ and db/db mouse microglia expression patterns of top regulon genes in db/+ and db/db mouse astro- cell populations. Expression levels for individual genes are presented as cyte cell populations. (f–j) Top regulons (f, g) and unique motifs (h, i) the normalised counts. Neuronal populations are shown in ESM Fig. 2 13 Diabetologia (2023) 66:1557–1575 1565 Fig. 4  Neuron compromise and microglia burden are increased in db/db mice. (a) Schematic describing the staining process. Indirect staining with primary antibody and secondary fluo- rescently conjugated antibody was used to identify cell subtypes (immature neurons, mature neurons, astrocytes and microglia). (b) NeuN/DAPI ratio was significantly lower in the cortex from db/db mice, when compared with control mice (***p<0.001 vs control). Representative images of corti- cal sections from control and db/db mice, with NeuN (red) and DAPI (blue) staining, are shown. Scale bar, 50 µm. (c, d) No differences were observed when DCX (p=0.407) (c) or BrdU (p=0.267) (d) densities were compared in the cortex from db/db and control mice. (e, f) DCX burden (e) was sig- nificantly increased in the SVZ from db/db when compared with control mice (*p=0.012 vs control). Similarly, BrdU density (f) was significantly higher in the SVZ from db/db mice (**p=0.004 vs control). (g) Representative images of the SVZ from control and db/db mice (green, ­BrdU+; red, ­DCX+). Scale bar, 50 μm. (h) Microglia burden was increased in db/db mice when compared with control mice (*p=0.046). Representative images of microglia staining in the cortex from control and db/db mice (green, Iba-1) are shown. Scale bar, 25 μm. (i) Astrocyte burden was not affected in db/db mice when compared with control mice (p=0.142). Representative images of astrocyte staining in the cortex from control and db/db mice (red, GFAP) are shown. Scale bar, 25 μm. Schematic in (a) was created with BioRender.com. ASC, astrocytes; ImmN, immature neurons; MG, microglia; mNEUR, mature neurons 13 1566 Diabetologia (2023) 66:1557–1575 a b 1.0 1.0 0.9 0.9 0.8 0.8 Height Height 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 Module Module colours colours c Module cell - type relationships d ImmN Module cell - type relationships Grey Grey MG Salmon Pink Purple mNEUR Blue Correlated gene modules Green Black Correlated gene modules Turquoise Midnight-blue R Brown Brown 1.0 ImmN Black Turquoise 0.5 Tan Magenta 0.0 Yellow Green-yellow -0.5 Salmon Tan -1.0 Purple Midnight-blue ASC Green ASC Red Light cyan Yellow Cyan Cyan Green-yellow mNEUR Blue Red Pink Magenta ImmN MG OEG OLG ImmN ASC/ ABC ASC CPC EC EPC MAC/ DC OPC/ VLMC PC VSMC MG OLG mNEUR ASC ASC/ ABC EC CPC EPC MAC/ DC PC OPC/ VLMC VSMC mNEUR MG e Regulation of primary metabolic process ASC MG Fc-γ receptor signalling pathway GO terms Branched-chain amino acid metabolic GO terms Regulation of lipid metabolic process Fc receptor mediated stimulatory signalling pathway Glucan catabolic process Regulation of lamellipodium assembly Glycogen catabolic process Fc-γ receptor signalling pathway in phagocytosis 0 1 2 3 4 5 6 7 0 3 6 9 12 -log10(p value) -log10(p value) Regulation of mRNA splicing, via spliceosome ImmN mNEUR Protein transport Regulation of mRNA stability GO terms GO terms Synaptic vesicle exocytosis mRNA metabolic process Chromatin remodeling Signal release from synapse ATP-dependent chromatin remodelling Neurotransmittor secretion 0 3 6 9 12 15 18 0 2 4 6 8 10 -log10(p value) -log10(p value) 13 Diabetologia (2023) 66:1557–1575 1567 ◂Fig. 5  WGCN analysis and GO terms for correlated modules. (a, b) increased expression of marker genes for inhibitory neuronal Cluster dendrogram for gene modules showing modules of high inter- characteristics (Sp9 and Meis1) (ESM Fig. 2f–j). These out- connection in control (a) and db/db (b) mice. Height in cluster den- drograms indicates the inter-cluster correlation distance and modules comes suggest a shift in the gene regulatory networks and are marked by different colours in the horizontal bar (grey represents driver transcription factors underlying distinct cell fates in unassigned genes). (c, d) The relationship between gene sets and cell db/db mouse cortex. The corresponding scale independence types is shown for control (c) and db/db (d). Each column is a cell and mean connectivity are shown in ESM Fig. 3. type and each row represents module eigengenes. Each unit is col- oured in the correlation coefficient (R) on a scale of −1 to +1 (black, negative; red, positive; white, null). (e) Unique GO terms for corre- lated gene module in db/db mouse cell types. Cell images were cre- Ex vivo investigation validates neuronal and glial population ated with BioRender.com. ABC, arachnoid barrier cells; ASC, astro- alterations in db/db mice Immunohistochemistry (Fig. 4a) cytes; CPC, choroid plexus cells; DC, dendritic cells; EC, endothelial cells; EPC, ependymocytes; ImmN, immature neurons; MAC, mac- was used to quantify the numbers of immature and mature rophages; MG, microglia; mNeur, mature neurons; OLG, oligoden- neurons (Fig. 4b–g). ­NeuN+ cell/DAPI ratio was signifi- drocytes; OPC, oligodendrocyte progenitor cells; PC, pericytes; cantly reduced in db/db mice (Fig. 4b). A slight increase in VLMC, vascular leptomeningeal cells; VSMC, vascular smooth mus- the density of D­ CX+ and proliferating cells was observed cle cells in db/db mice (Fig. 4c,d). Since the cortex is not a classical neurogenic niche, proliferation and neurogenesis were also Molecular analysis of cortical cell populations reveals criti‑ assessed in the subventricular zone (SVZ). Both DCX bur- cal changes in diabetic mouse brains Focused molecular den and B ­ rdU+ cell density in the SVZ were significantly analysis on neuronal and glial specific transcripts (Fig. 2d–g) increased in db/db mice when compared with control mice demonstrated that genes encoding ribosomal protein S21 and (Fig. 4e–g). S29 (Rps21 and Rps29, respectively) were downregulated in Microglia burden was increased in the cortex from immature neurons as well as astrocytes, with upregulation db/db mice when compared with control mice (Fig. 4h), of ribosomal protein SA (encoded by Rpsa) in microglia. while no differences in astrocyte burden were observed Transcripts linked to inflammation and microglial activation (Fig. 4i). were altered in db/db mouse microglia: expression of genes encoding apolipoprotein E (Apoe) [25] and prostaglandin D2 Correlated gene modules reveal the altered cellular behav‑ synthase (Ptgds) [26] was upregulated. Furthermore, down- iour of metabolism, synapse and inflammatory pathways regulated genes in mature neurons were major mediators and in db/db mouse cortex WGCN analysis was performed regulators of synaptic neuronal functions such as the genes (Fig. 5a–d). Gene ontology (GO) analysis of correlated encoding glutamate ionotropic receptor AMPA type subunit unique gene modules in cells revealed altered mRNA 1 (Gria1) [27] and myelin basic protein (Mbp) [28]. Mature metabolism in immature neurons and changes in glycogen neurons showed downregulation of genes encoding mater- and lipid metabolism in astrocytes of diabetic mouse brains nally expressed gene 3 (Meg3) and β-actin (Actb). Astrocytes (Fig. 5e). Furthermore, pathways implicated in inflammation displayed downregulation of actin polymerisation regulat- and phagocytosis in microglia were uniquely altered in dia- ing genes tetraspanin 7 (Tspan7) [29] and metallothionein betic cortex, whereas mature neurons had altered pathways III (Mt3) [30]. Immature neurons showed downregulation relating to synaptic processes (Fig. 5e). of the gene encoding early growth response-1 (Egr1), an important transcription factor in memory formation [31]. Vascular abnormalities observed in db/db mouse cortex Pre- This occurred alongside upregulation of HbA-a2 (encoding vious reports have suggested that brain vasculature is affected a globin family protein) and Gad2, which are known to be in type 2 diabetes models [24]. We observed reduced linked to schizophrenia and bipolar disorder [32]. Analysis ­laminin+ vessels in db/db mice (Fig. 6a, d). ‘Hotspots’ of by the algorithm pySCENIC identified enhanced expres- leakage were observed in the db/db mouse cortex but not in sion of the transcription factor genes (Hes5, Lhx2, Sox21, the control mice cortex (Fig. 6e). Expression of endothelial Sox2) and neurogenesis regulatory factors in db/db mouse marker genes including Lgals1 (encoding galectin 1) and astrocytes (Fig. 3a–e), suggesting a potentially ambiguous Lamc3 (encoding laminin γ subunit 3) was reduced in db/db lineage trajectory between db/db and control mice. Top mice (ESM Fig. 4). The top differentially expressed genes microglial regulons were slightly discrete between control (DEGs) in db/db mouse endothelial cells included Malat and db/db mice (Fig. 3f–j). The mature neuronal population (encoding metastasis-associated lung adenocarcinoma tran- (db/db) showed top regulons that are known transcription script 1), which may be involved in angiogenesis [33], and factors (Elk1 and Egr4) regulating early immediate genes Cts3, which is dysregulated in cerebral amyloid angiopathy (ESM Fig. 2a–e), while immature neuronal cells (db/db) had (Fig. 6h) [34]. There was an increase in acellular capillaries 13 1568 Diabetologia (2023) 66:1557–1575 Fig. 6  Vascular alterations in db/db mouse cortex. (a) ­Laminin+ leakage in db/db mouse cortex (blue, laminin; red, IgG; green, vessel area was significantly reduced at various levels of Bregma albumin) (e); lectin (green) and Col-IV (red) in control and db/db in the cortex of db/db mice (*p<0.05, **p<0.01). (b) A significant mouse cortex for assessment of acellular capillaries (arrows) (f); increase in acellular capillaries was observed in db/db mouse cor- and lectin (green) and PDGFR-β (red) in the control and db/db tex (*p<0.05) at Bregma level +0.5 mm. (c) A significant loss of mouse cortex for measurement of pericyte coverage (g). Scale pericyte coverage (PDGFR-β) was observed in db/db mouse cor- bar, 50 µm. (h) DEGs in endothelial cells. (i) DEGs in pericytes. tex at Bregma +0.5 mm (**p<0.01). (d–g) Representative images: Expression levels for individual genes (all p<0.001) are presented laminin staining (purple) (d); ‘hot-spot’ (arrow) area of vessel as the normalised counts 13 Diabetologia (2023) 66:1557–1575 1569 in the db/db mouse cortex (Fig. 6b,f) alongside a loss of (Fig. 8b). Iba-1 was not significantly changed in any corti- pericyte coverage (Fig. 6c,g). The top DEGs in pericytes cal region in individuals with type 2 diabetes when com- included Cald1 (encoding downregulation of caldesmon 1), pared with healthy control individuals. A reduction in zonula which is involved in smooth muscle contraction [35]. Dys- occludens-1 (ZO-1) protein levels was observed in individu- regulation of metabolic genes was observed in db/db mouse als with type 2 diabetes in the parietal and temporal cortex pericytes, with a loss of mitochondrial complex-related genes (ESM Fig. 5b). Occludin was unexpectedly increased in the such as mt-Co1 and mt-Nd4 (encoding mitochondrial NADH parietal cortex of individuals with type 2 diabetes (ESM dehydrogenase 4) (Fig. 6i). Fig. 5b). A statistically significant loss of platelet-derived growth factor receptor (PDGFR) α/β was observed in the Cortical metabolism is significantly altered in db/db mice, parietal cortex of individuals with type 2 diabetes (ESM with dysregulated metabolic pathway genes observed in Fig. 5c). specific cortical populations Our results indicated that metabolic dysfunction in the type 2 diabetes brain may be a consequence of NVU impairment. To validate these Discussion findings in vivo, metabolic profiling of the cortex was performed using acute brain slices (Fig. 7a). Glycolytic Recent evidence has shown a relationship between type 2 function was increased in db/db mouse cortex com- diabetes and cognitive decline, dementia and Alzheimer’s pared with controls (Fig. 7b). Non-glycolytic acidifica- disease [2-4]. Hyperglycaemia, insulin resistance, hyperten- tion (Fig. 7c), basal glycolysis (Fig. 7d) and glycolytic sion and dyslipidaemia in type 2 diabetes have a profound capacity (Fig. 7e) were all significantly increased in the effect on the vasculature, leading to complications such as db/db mouse cortex; however, no significant difference stroke, cardiovascular disease and diabetic retinopathy. Type was observed in glycolytic reserve (Fig. 7f). A number 2 diabetes also has a damaging effect on neurons and glia, of metabolism-related genes were differentially expressed often relating to cell–cell communication defects across the in the db/db mouse cortex (Fig. 7g). For example, astro- retinal NVU [12]. The current study has detailed how db/db cytes and mature neurons showed increased expression of mice manifest cognitive deficiencies in unison with reduced electron transport chain subunit genes and NADH dehy- brain weight and cortex size. Here, for the first time, we have drogenase subunits. Together, these data indicate that the catalogued a map of transcriptional profiles of the cerebral metabolic processes of cortical NVU cell populations are cortex from db/db mice, with a focus on the cell populations altered in db/db mice. of the NVU. Our scRNA-seq analysis identified a decrease in mature neurons, alongside an increase in microglia and immature Cortical cell populations are also changed in humans with neurons in the db/db mouse cortex. These novel findings, type 2 diabetes To validate our db/db mouse model find- including disruption of several key neuronal regulatory ings, we studied samples from people with type 2 diabetes. genes, were complemented by the observed brain atrophy Neuronal and glial cell populations were assessed in a small and cognitive decline in the db/db mouse. Meg3 and Mbp population of cortex samples from non-diabetic individuals both showed reduced expression in db/db mice, indicating and from individuals with type 2 diabetes. Staining of NeuN functional disruption. Downregulation of Mbp in mature in frontal cortex sections indicated a qualitative reduction neurons may indicate that golli-Mbp [28] (a known pro- in neuronal cells in the frontal cortex of individuals with tein found outside the myelin sheath), which is involved type 2 diabetes (Fig. 8a), although semi-quantitative analysis in neuronal development [36] and ­Ca2+ homeostasis, may of NeuN western blots from parietal, temporal and frontal be affected in type 2 diabetes. We also observed some cortex samples showed no significant difference (Fig. 8c). A evidence of NeuN reduction in individuals with type 2 similar observation was seen after analysis of postsynaptic diabetes. Together, these data suggest that type 2 diabetes density protein 95 (PSD95) (ESM Fig. 5a). Noticeably, a may affect neuronal maturation, thereby having a profound reduction in GFAP protein was observed in samples from impact on cognition. This finding is reinforced by similar type 2 diabetes temporal cortex only (Fig. 8b,d). GFAP findings in Alzheimer’s disease [37] and other neurode- staining of human cortex showed an increased localisation generative diseases associated with progressive cognitive with blood vessels (stained with wheat germ agglutinin) decline [38]. 13 1570 Diabetologia (2023) 66:1557–1575 13 Diabetologia (2023) 66:1557–1575 1571 ◂Fig. 7  Metabolic profiling of db/db mouse cortex. (a) Schematic glucose metabolism and proinflammatory responses [44]. describing the process of obtaining acute coronal sections for meta- However, high levels of leptin can influence ApoE expres- bolic profiling by Seahorse X FE 96. Male db/db or control mice aged 16 weeks (n=5 mice per group) were used to generate acute coronal sion [45], therefore this finding should be explored in other slices (250 µm) using the Leica VT100s Vibratome. The Seahorse diabetic models before a conclusion is made. In diabetic analyser was used to assess glycolytic function, via an adapted glyco- retinopathy, neurodegeneration is known to be an impor- lytic stress test protocol. (b) Traces from n=5 db/db and n=5 control tant early event (prior to vascular complications [46]), mice (representative of seven or eight punches per mouse) following the adapted glycolysis stress test protocol. (c) Non-glycolytic acidifi- and is thought to be driven by chronic inflammation. A cation was significantly increased in db/db mice. (d) Basal glycoly- recent study showed that depleting microglia in a high-fat- sis was significantly increased in db/db mice. (e) Glycolytic capacity diet-induced model of diabetes was able to improve was significantly increased in db/db mice. (f) No significant changes cerebrovasculature deficits [47]. Further understanding of were observed between db/db and control mice when comparing gly- colytic reserve. n=5 mice per group; seven or eight cortical punches the neuroinflammatory pathways mediated by microglia per mouse. *p<0.05, **p<0.01. Unpaired Student’s t test. Data are in the type 2 diabetes cortex will be beneficial in under- presented as mean values ± SEM. (g) Metabolism-related DEGs (all standing the role of neuroinflammation in the progress of p<0.001) in db/+ (control) and db/db mouse cortex. Heatmap depicts diabetes-related cognitive decline. normalised expression counts scaled between +1 and −1. Schematic in (a) was created with BioRender.com. ECAR, extracellular acidifi- Diabetes is known to have a profound effect on vascular cation rate function. Our IHC data indicated reduced ­laminin+ vascular density in the db/db mouse cortex, mirrored by increased acellular capillaries and loss of pericytes. This is in line with Previous reports have suggested that glial ­GFAP+ area is findings in the db/db retina, where vascular alterations can increased in the db/db mouse cortex [39]. Our findings in db/ be seen from 10 weeks [48]. We also observed evidence of db mice and humans with type 2 diabetes remain ambiguous. vascular leakage in specific ‘hot spots’ in the db/db mouse Nevertheless, we observed a distinct transcriptional profile cortex, although this requires further characterisation. Top of cortical astrocytes in db/db mice. A notable example is DEGs in endothelial cells suggested alterations to angiogen- Mt3 downregulation in db/db mouse astrocytes. This change esis and vessel disease, while smooth muscle contraction in expression has been related to reduced uptake of Aβ and metabolism were dysregulated in pericytes. We saw a leading to extracellular accumulation in Alzheimer’s dis- statistically significant reduction of ZO-1 protein expres- ease [40], highlighting a link between Alzheimer’s disease sion in the parietal and temporal cortex of individuals with pathology and diabetes-induced cognitive decline. Mt1 and type 2 diabetes. We also observed region-specific increases Mt3 are also involved in protection against oxidative stress in occludin expression, although the cellular source of [30] and their downregulation in db/db mouse astrocytes this upregulation remains to be clarified. Loss of vascu- may contribute to oxidative damage and enhanced inflam- lar density and pericyte coverage have been reported in mation in the type 2 diabetes cortex. diabetic retinopathy [49] and also in the cortex in Alzheimer’s Cortical astrocyte metabolism was affected in db/db mice, disease [8], strengthening the assertion that progres- including upregulation of genes involved in the formation of sive NVU dysfunction in the brain during type 2 diabe- mitochondrial complexes (e.g. mt-Co2, mt-Nd1, mt-Nd2) and tes is linked to neuronal depletion and cognitive decline. dysregulated lipid metabolism. A previous study reported Together, these findings of altered neuronal maturation, that alterations in astrocyte lipid metabolism may be linked astrocyte metabolism and microglial inflammation in the db/ to degeneration of motor neurons [41]. Astrocytes play an db mouse cortex highlight the impact of complex neuroglial important role in NVU function, interacting with and sup- interactions in type 2 diabetes-induced cognitive decline. porting neurons and blood vessels [42] and these changes We report that db/db mouse cortex showed a significant in metabolic function in cortical astrocytes may contribute increase in glycolysis and alterations to a number of genes to NVU dysfunction in db/db mice. involved in oxidative phosphorylation (e.g. mt-Nd2 and Microglia are important modulators of homeostasis mt-Nd5 were reduced in immature neurons). This result in the cerebral cortex. We found an increased number of suggests that at least some cells in the type 2 diabetes microglia in the cortex of db/db mice, occurring in unison cortex may switch to anaerobic glycolysis as the main sup- with overexpression of inflammatory pathways and genes ply of energy (Warburg effect). Interestingly, this War- involved in myelin damage, indicating neuronal injury in burg effect metabolism shift has been shown to drive neu- the diabetic brain [43]. ApoE is an established genetic risk ronal degeneration in sporadic Alzheimer’s disease [50]. factor for Alzheimer’s disease with roles in Aβ clearance, Metabolism was also dysregulated in mature neurons and 13 1572 Diabetologia (2023) 66:1557–1575 Fig. 8  Cortex samples from individuals with type 2 diabetes reveal ting of human cortex samples (parietal, temporal and frontal) shows alterations in neuronal and glial cell populations. (a) Representative the expression of NeuN in each location, n=3 individuals per group. images show reduced NeuN staining (yellow) in frontal cortex from (d) Western blotting of human cortex samples (parietal, temporal and a donor with type 2 diabetes compared with cortex from a control frontal) shows the expression of GFAP and Iba-1 in each location. donor. Laminin staining is shown in purple, alongside DAPI in blue, GFAP expression was reduced in the temporal cortex (*p<0.05 vs n=4 or 5 individuals per group. (b) Representative images showing control samples), n=3 individuals per group. Western blot quantifica- GFAP staining of the frontal cortex (purple) alongside wheat germ tion was analysed by normalising the protein levels to the Ponceau agglutinin (yellow), n=4 or 5 patients per group. (c) Western blot- stain. C, control; T2D, type 2 diabetes; WGA, wheat germ agglutinin 13 Diabetologia (2023) 66:1557–1575 1573 astrocytes. Mitochondrial dysfunction in diabetes has been script. All authors approved the final version of the article. AWS is the widely described [51]. We propose that altered metabolic guarantor of this work. function of key component cells of the brain NVU may play a major role in the loss of vascular integrity and Open Access This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- homeostasis leading to increased risk of neurodegenera- tion, distribution and reproduction in any medium or format, as long tion and cognitive decline. as you give appropriate credit to the original author(s) and the source, Our study reveals distinct changes to the cells of the provide a link to the Creative Commons licence, and indicate if changes cerebral cortex in type 2 diabetes. One limitation of our were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated approach is the use of IHC quantification to validate otherwise in a credit line to the material. If material is not included in changes to cortical NVU populations, as IHC is primar- the article’s Creative Commons licence and your intended use is not ily a qualitative technique. Future analysis should focus permitted by statutory regulation or exceeds the permitted use, you will on characterisation and quantitative validation of corti- need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. cal cellular heterogeneity and subcellular vulnerability in type 2 diabetes. Our dataset facilitates opportunities for molecular analysis and their validation in cell–cell communication in diabetes-related cortical dysfunction. References Future studies in such diabetes models will provide further 1. Karuranga S, da Rocha Fernandes J, Huang Y, Malanda B (2017) insights into the therapeutic potential of targeting these IDF diabetes atlas eighth edition 2017. International Diabetes gene alterations in the NVU for treatment of diabetic Federation, Brussels 2. Koekkoek PS, Janssen J, Kooistra M et al (2016) Case-finding comorbidities of memory and cognition. for cognitive impairment among people with type 2 diabetes in primary care using the Test Your Memory and Self-Administered Supplementary Information The online version contains peer-reviewed Gerocognitive Examination questionnaires: the Cog-ID study. but unedited supplementary material available at https://​doi.​org/​10.​ Diabet Med 33(6):812–819. https://​doi.​org/​10.​1111/​dme.​12874 1007/​s00125-​023-​05935-2. 3. Biessels GJ, Staekenborg S, Brunner E, Brayne C, Scheltens P (2006) Risk of dementia in diabetes mellitus: a systematic Acknowledgements The authors wish to thank K. Harkin (Queen’s review. Lancet Neurol 5(1):64–74. https://d​ oi.o​ rg/1​ 0.1​ 016/S ​ 1474-​ University Belfast) for assistance with single-cell dissociation experi- 4422(05)​70284-2 ments. The authors would also like to acknowledge the support provided 4. Little K, Llorián-Salvador M, Scullion S et al (2022) Common by Genomics Core Technology Unit (GCTU), as well as resources and pathways in dementia and diabetic retinopathy: understanding the consultancy provided by High Performance Computational Services mechanisms of diabetes-related cognitive decline. Trends Endo- (Kelvin Cluster) team and the Advanced Imaging Core Technology Unit crinol Metab 33(1):50–71. https://​doi.​org/​10.​1016/j.​tem.​2021.​10.​ (AICTU), Queen’s University Belfast. The authors thank the Animal 008 Facility of the University of Cadiz and its personnel for their help and 5. Ma T, Li Z-Y, Yu Y et al (2022) Gray and white matter abnormal- support and thank Servicios Centrales de Investigacion en Biomedicina ity in patients with T2DM-related cognitive dysfunction: a sys- (SC-IBM) from Universidad de Cadiz for the resources and technical temic review and meta-analysis. Nutr Diabetes 12(1):39. https://​ support. Schematics were created with BioRender.com. doi.​org/​10.​1038/​s41387-​022-​00214-2 6. Dutta BJ, Singh S, Seksaria S, Das Gupta G, Singh A (2022) Data availability The single-cell sequencing data that support this Inside the diabetic brain: insulin resistance and molecular mecha- study are available at GEO accession GSE217665 (https://​www.​ncbi.​ nism associated with cognitive impairment and its possible thera- nlm.​nih.​gov/​geo/​query/​acc.​cgi?​acc=​GSE21​7665) peutic strategies. Pharmacol Res 182:106358. https://​doi.​org/​10.​ 1016/j.​phrs.​2022.​106358 Funding This work is supported by the RECOGNISED consortium grant 7. Garcia FJ, Sun N, Lee H et al (2022) Single-cell dissection of the (EU GA 847749) to MGA and AWS and by the Deutsche Forschungsge- human brain vasculature. Nature 603(7903):893–899. https://​doi.​ meinschaft TI 799/1-3 and Novo Nordisk Foundation P3110103 grants org/​10.​1038/​s41586-​022-​04521-7 to VKT. MLS is funded by Maria Zambrano fellowship from Spanish 8. Yang AC, Vest RT, Kern F et al (2022) A human brain vascu- Ministry of Science, Innovation and Universities, funded by European lar atlas reveals diverse mediators of Alzheimer’s risk. Nature Union ‘Next Generation EU’ (Universitat Autònoma de Barcelona). The 603(7903):885–892. https://d​ oi.o​ rg/1​ 0.1​ 038/s​ 41586-0​ 21-0​ 4369-3 study funders were not involved in the design of the study; the collec- 9. Velmeshev D, Schirmer L, Jung D et al (2019) Single-cell genom- tion, analysis, and interpretation of data; writing the report; and did not ics identifies cell type-specific molecular changes in autism. Sci- impose any restrictions regarding the publication of the report. ence 364(6441):685–689. https://​doi.​org/​10.​1126/​scien​ce.​aav81​ 30 Authors’ relationships and activities AWS is an associate editor for 10. Mathys H, Davila-Velderrain J, Peng Z et al (2019) Single- Diabetologia. The authors declare that there are no other relationships cell transcriptomic analysis of Alzheimer’s disease. Nature or activities that might bias, or be perceived to bias, their work 570(7761):332–337. https://​doi.​org/​10.​1038/​s41586-​019-​1195-2 11. Lerch JP, Pruessner JC, Zijdenbos A, Hampel H, Teipel SJ, Evans Contribution statement KL, AS and ADM contributed equally to AC (2005) Focal decline of cortical thickness in Alzheimer’s dis- experiments, analysis and writing of the manuscript. Authorship order ease identified by computational neuroanatomy. Cereb Cortex in the manuscript was determined based on date of joining the project. 15(7):995–1001. https://​doi.​org/​10.​1093/​cercor/​bhh200 MLS, MVS, MTA, MS, NB and SS contributed to data collection and 12. Niu T, Fang J, Shi X et al (2021) Pathogenesis study based on analysis, and revision of the article drafts. JXC, IK, RS, MGA, VKT high-throughput single-cell sequencing analysis reveals novel and AWS contributed to discussion, and reviewed/edited the manu- 13 1574 Diabetologia (2023) 66:1557–1575 transcriptional landscape and heterogeneity of retinal cells in 28. Landry CF, Ellison JA, Pribyl TM, Campagnoni C, Kampf K, type 2 diabetic mice. Diabetes 70(5):1185–1197. https://​doi.​org/​ Campagnoni AT (1996) Myelin basic protein gene expression in 10.​2337/​db20-​0839 neurons: developmental and regional changes in protein targeting 13. Ma S, Bi W, Liu X et al (2022) Single-cell sequencing analysis of within neuronal nuclei, cell bodies, and processes. J Neurosci Off the db/db mouse hippocampus reveals cell-type-specific insights J Soc Neurosci 16(8):2452–2462. https://d​ oi.o​ rg/1​ 0.1​ 523/J​ NEUR​ into the pathobiology of diabetes-associated cognitive dysfunc- OSCI.​16-​08-​02452.​1996 tion. Front Endocrinol 13:891039. https://​doi.​org/​10.​3389/​fendo.​ 29. Bassani S, Passafaro M (2012) TSPAN7: a new player in excita- 2022.​891039 tory synapse maturation and function. Bioarchitecture 2(3):95–97. 14. Jiménez-Palomares M, Ramos-Rodríguez JJ, López-Acosta JF https://​doi.​org/​10.​4161/​bioa.​20829 et al (2012) Increased Aβ production prompts the onset of glucose 30. Koh J-Y, Lee S-J (2020) Metallothionein-3 as a multifunctional intolerance and insulin resistance. Am J Physiol Endocrinol Metab player in the control of cellular processes and diseases. Mol Brain 302(11):1373–1380. https://​doi.​org/​10.​1152/​ajpen​do.​00500.​2011 13(1):116. https://​doi.​org/​10.​1186/​s13041-​020-​00654-w 15. Infante-Garcia C, Ramos-Rodriguez JJ, Galindo-Gonzalez L, 31. Veyrac A, Besnard A, Caboche J, Davis S, Laroche S (2014) The Garcia-Alloza M (2016) Long-term central pathology and cogni- transcription factor Zif268/Egr1, brain plasticity, and memory. tive impairment are exacerbated in a mixed model of Alzheimer’s Prog Mol Biol Transl Sci 122:89–129. https://​doi.​org/​10.​1016/​ disease and type 2 diabetes. Psychoneuroendocrinology 65:15–25. B978-0-​12-​420170-​5.​00004-0 https://​doi.​org/​10.​1016/j.​psyne​uen.​2015.​12.​001 32. Davis KN, Tao R, Li C et al (2016) GAD2 Alternative transcripts 16. Zheng H, Zheng Y, Zhao L et al (1863) (2017) Cognitive decline in the human prefrontal cortex, and in schizophrenia and affec- in type 2 diabetic db/db mice may be associated with brain region- tive disorders. PLoS One 11(2):e0148558. https://d​ oi.o​ rg/1​ 0.1​ 371/​ specific metabolic disorders. Biochim Biophys Acta Mol Basis journ​al.​pone.​01485​58 Dis 1:266–273. https://​doi.​org/​10.​1016/j.​bbadis.​2016.​11.​003 33. Michalik KM, You X, Manavski Y et al (2014) Long noncod- 17. Langfelder P, Horvath S (2008) WGCNA: an R package for ing RNA MALAT1 regulates endothelial cell function and ves- weighted correlation network analysis. BMC Bioinformatics sel growth. Circ Res 114(9):1389–1397. https://​doi.​org/​10.​1161/​ 9(1):559. https://​doi.​org/​10.​1186/​1471-​2105-9-​559 CIRCR​ESAHA.​114.​303265 18. Qiu X, Mao Q, Tang Y et al (2017) Reversed graph embed- 34. Levy E, Jaskolski M, Grubb A (2006) The role of cystatin C in ding resolves complex single-cell trajectories. Nat Methods cerebral amyloid angiopathy and stroke: cell biology and animal 14(10):979–982. https://​doi.​org/​10.​1038/​nmeth.​4402 models. Brain Pathol 16(1):60–70. https://d​ oi.o​ rg/1​ 0.1​ 111/j.1​ 750-​ 19. Carranza-Naval MJ, del Marco A, Hierro-Bujalance C et al (2021) 3639.​2006.​tb005​62.x Liraglutide reduces vascular damage, neuronal loss, and cogni- 35. Mayanagi T, Sobue K (2011) Diversification of caldesmon-linked tive impairment in a mixed murine model of Alzheimer’s disease actin cytoskeleton in cell motility. Cell Adh Migr 5(2):150–159. and type 2 diabetes. Front Aging Neurosci 13(December):1–17. https://​doi.​org/​10.​4161/​cam.5.​2.​14398 https://​doi.​org/​10.​3389/​fnagi.​2021.​741923 36. Reyes SD, Campagnoni AT (2002) Two separate domains in 20. Hierro-Bujalance C, del Marco A, José Ramos-Rodríguez J et al the golli myelin basic proteins are responsible for nuclear tar- (2020) Cell proliferation and neurogenesis alterations in Alzhei- geting and process extension in transfected cells. J Neurosci Res mer’s disease and diabetes mellitus mixed murine models. J Neu- 69(5):587–596. https://​doi.​org/​10.​1002/​jnr.​10319 rochem 154(6):673–692. https://​doi.​org/​10.​1111/​jnc.​14987 37. Li B, Yamamori H, Tatebayashi Y et al (2008) Failure of neu- 21. Ramos-Rodriguez JJ, Ortiz O, Jimenez-Palomares M et al ronal maturation in Alzheimer disease dentate gyrus. J Neuro- (2013) Differential central pathology and cognitive impairment pathol Exp Neurol 67(1):78–84. https://​doi.​org/​10.​1097/​nen.​ in pre-diabetic and diabetic mice. Psychoneuroendocrinology 0b013​e3181​60c5db 38(11):2462–2475. https://​doi.​org/​10.​1016/j.​psyne​uen.​2013.​05.​ 38. Orvoen S, Pla P, Gardier AM, Saudou F, David DJ (2012) Hun- 010 tington’s disease knock-in male mice show specific anxiety- 22. Qi G, Mi Y, Yin F (2021) Characterizing brain metabolic func- like behaviour and altered neuronal maturation. Neurosci Lett tion ex vivo with acute mouse slice punches. STAR Protoc 507(2):127–132. https://​doi.​org/​10.​1016/j.​neulet.​2011.​11.​063 2(2):100559. https://​doi.​org/​10.​1016/j.​xpro.​2021.​100559 39. Liu Y, Chen D, Smith A, Ye Q, Gao Y, Zhang W (2021) Three- 23. Bogdanov P, Corraliza L, Villena JA et al (2014) The db/db dimensional remodeling of functional cerebrovascular archi- mouse: a useful model for the study of diabetic retinal neurode- tecture and gliovascular unit in leptin receptor-deficient mice. J generation. PLoS One 9(5):e97302. https://​doi.​org/​10.​1371/​journ​ Cereb blood flow Metab Off J Int Soc Cereb Blood Flow Metab al.​pone.​00973​02 41(7):1547–1562. https://​doi.​org/​10.​1177/​02716​78X21​10065​96 24. Ramos-Rodriguez JJ, Jimenez-Palomares M, Murillo-Carretero 40. Lee S-J, Seo B-R, Koh J-Y (2015) Metallothionein-3 modulates MI et al (2015) Central vascular disease and exacerbated pathol- the amyloid β endocytosis of astrocytes through its effects on ogy in a mixed model of type 2 diabetes and Alzheimer’s disease. actin polymerization. Mol Brain 8(1):84. https://​doi.​org/​10.​1186/​ Psychoneuroendocrinology 62:69–79. https://​doi.​org/​10.​1016/j.​ s13041-​015-​0173-3 psyne​uen.​2015.​07.​606 41. Mou Y, Dong Y, Chen Z et al (2020) Impaired lipid metabolism 25. Krasemann S, Madore C, Cialic R et al (2017) The TREM2-APOE in astrocytes underlies degeneration of cortical projection neu- pathway drives the transcriptional phenotype of dysfunctional rons in hereditary spastic paraplegia. Acta Neuropathol Commun microglia in neurodegenerative diseases. Immunity 47(3):566– 8(1):214. https://​doi.​org/​10.​1186/​s40478-​020-​01088-0 581. https://​doi.​org/​10.​1016/j.​immuni.​2017.​08.​008. (e9) 42. Liu C-Y, Yang Y, Ju W-N, Wang X, Zhang H-L (2018) Emerging 26. Mohri I, Taniike M, Taniguchi H et al (2006) Prostaglandin roles of astrocytes in neuro-vascular unit and the tripartite synapse D2-mediated microglia/astrocyte interaction enhances astroglio- with emphasis on reactive gliosis in the context of Alzheimer’s sis and demyelination in twitcher. J Neurosci Off J Soc Neurosci disease. Front Cell Neurosci 12:1–12. https://​doi.​org/​10.​3389/​ 26(16):4383–4393. https://​doi.​org/​10.​1523/​JNEUR​OSCI.​4531-​ fncel.​2018.​00193 05.​2006 43. Zhang J, Sun X, Zheng S et al (2014) Myelin basic protein induces 27. Shen K, Limon A (2021) Transcriptomic expression of AMPA neuron-specific toxicity by directly damaging the neuronal plasma receptor subunits and their auxiliary proteins in the human brain. membrane. PLoS One 9(9):e108646. https://d​ oi.o​ rg/1​ 0.1​ 371/j​ ourn​ Neurosci Lett 755:135938. https://​doi.​org/​10.​1016/j.​neulet.​2021.​ al.​pone.​01086​46 135938 13 Diabetologia (2023) 66:1557–1575 1575 44. Fernandez CG, Hamby ME, McReynolds ML, Ray WJ (2019) on retinal neuroinflammation and vascular leakage in experimen- The role of APOE4 in disrupting the homeostatic functions of tal diabetes. Biomedicines 9(8):1–10. https://​doi.​org/​10.​3390/​ astrocytes and microglia in aging and Alzheimer’s disease. Front biome​dicin​es908​0926 Aging Neurosci 11:14. https://​doi.​org/​10.​3389/​fnagi.​2019.​00014 49. Hammes HP, Feng Y, Pfister F, Brownlee M (2011) Diabetic retin- 45. Shen L, Tso P, Wang DQ-H et al (2009) Up-regulation of apolipo- opathy: targeting vasoregression. Diabetes 60(1):9–16. https://d​ oi.​ protein E by leptin in the hypothalamus of mice and rats. Physiol org/​10.​2337/​db10-​0454 Behav 98(1–2):223–228. https://​doi.​org/​10.​1016/j.​physb​eh.​2009.​ 50. Traxler L, Herdy JR, Stefanoni D et al (2022) Warburg-like 05.​013 metabolic transformation underlies neuronal degeneration in 46. Simó R, Stitt AW, Gardner TW (2018) Neurodegeneration in dia- sporadic Alzheimer’s disease. Cell Metab 34(9):1248–1263. betic retinopathy: does it really matter? Diabetologia 61(9):1902– https://​d oi.​o rg/​1 0.​1 016/j.​c met.​2 022.​0 7.​0 14. (e6) 1912. https://​doi.​org/​10.​1007/​s00125-​018-​4692-1 51. Avram VF, Merce AP, Hâncu IM et al (2022) Impairment of mito- 47. Shen Q, Zhang G (2021) Depletion of microglia mitigates cerebrovas- chondrial respiration in metabolic diseases: an overview. Int J Mol cular dysfunction in diet-induced obesity mice. Am J Physiol Metab Sci 23(16):1–25. https://​doi.​org/​10.​3390/​ijms2​31688​52 321(3):E367–E375. https://​doi.​org/​10.​1152/​ajpen​do.​00086.​2021 48. Simó R, Bogdanov P, Ramos H, Huerta J, Simó-Servat O, Hernán- Publisher’s note Springer Nature remains neutral with regard to dez C (2021) Effects of the topical administration of semaglutide jurisdictional claims in published maps and institutional affiliations. Authors and Affiliations Karis Little1 · Aditi Singh1 · Angel Del Marco2,3 · María Llorián‑Salvador1,4,5 · Maria Vargas‑Soria2,3 · Mireia Turch‑Anguera5,6,7,8 · Montse Solé5,6,7,8 · Noëlle Bakker9 · Sarah Scullion1 · Joan X. Comella5,6,7,8 · Ingeborg Klaassen9 · Rafael Simó4,5,10 · Monica Garcia‑Alloza2,3 · Vijay K. Tiwari1,11,12,13 · Alan W. Stitt1 · on behalf of the RECOGNISED consortium 6 * Monica Garcia‑Alloza Cell Signaling and Apoptosis Group, Vall d’Hebron Institut monica.garcia@uca.es de Recerca (VHIR), Barcelona, Spain 7 * Vijay K. Tiwari Departament de Bioquímica i Biologia Molecular i Institut de tiwari@health.sdu.dk Neurociències, Universitat Autònoma de Barcelona (UAB), Bellaterra, Spain * Alan W. Stitt 8 a.stitt@qub.ac.uk Centro de Investigación en Red en Enfermedades Neurodegenerativas (CIBERNED - ISCII), Madrid, Spain 1 The Wellcome‑Wolfson Institute for Experimental Medicine, 9 Ocular Angiogenesis Group, Department of Ophthalmology, School of Medicine, Dentistry & Biomedical Science, Amsterdam UMC location University of Amsterdam, Queen’s University Belfast, Belfast, Northern Ireland, UK Amsterdam, the Netherlands 2 Division of Physiology, School of Medicine, University 10 Centro de Investigación Biomédica en Red de Diabetes y of Cadiz, Cadiz, Spain Enfermedades Metabólicas Asociadas (CIBERDEM-ISCIII), 3 Instituto de Investigacion e Innovacion en Ciencias Madrid, Spain Biomedicas de la Provincia de Cadiz (INIBICA), Cadiz, 11 Institute of Molecular Medicine, University of Southern Spain Denmark, Odense C, Denmark 4 Department of Medicine, Universitat Autònoma de Barcelona 12 Danish Institute for Advanced Study (DIAS), Odense M, (UAB), Barcelona, Spain Denmark 5 Diabetes and Metabolism Research Unit, Vall d’Hebron 13 Department of Clinical Genetics, Odense University Institut de Recerca (VHIR), Vall d’Hebron University Hospital, Odense C, Denmark Hospital, Barcelona, Spain 13