Fig 2: Blood-brain barrier permeability measurements in male Ins2AKITA and WT littermate controls. (a) Representative stereomicroscope fluorescence images of brains showing 1 kDa Alexa Fluor 555 cadaverine permeability in Ins2AKITA and WT after 2 h of dye circulation (n = 2). (b) Representative confocal images of coronal sections of 1 kDa Alexa Fluor 555 cadaverine injected mouse brains. ANPEP positive mural cells, green; PECAM1 positive vasculature, white. No Alexa Fluor 555 cadaverine leakage into brain parenchyma was observed either in Ins2AKITA or WT mice (n = 2, scale bar 30 µm). (c) Quantification of 1 kDa Alexa Fluor 555 cadaverine permeability in 26.5–32 week-old Ins2AKITA and WT mice after 2 h circulation (n = > 8, 3 independent experiments). y-axis shows the fold change in relative fluorescence units (RFU) per gram of brain tissue in relation to WT. (d) Quantification of 1 kDa Alexa Fluor 488 cadaverine permeability in 38 week-old Ins2AKITA and WT mice after 2 h circulation (n = 3). y-axis shows the fold change in relative fluorescence units (RFU) per gram of brain tissue in relation to WT. (e) Evans Blue dye permeability in 30 week-old Ins2AKITA and WT littermate control mice after overnight circulation (n = > 2). y-axis shows optical density (OD) at 620 nm per gram of tissue. PdgfbRet/Ret served as positive control for tracer leakage into the brain parenchyma. n.s., not significant, student’s t test. Data is presented as mean ± SEM.

Fig 3: Microglia contribute to hypercapnia-induced vasodilation. (a) In vivo two-photon resonant (32-Hz) imaging was performed in the somatosensory cortex of CX3CR1tdTomato mice during hypercapnia (by inhalation of 10% CO2 under normoxic conditions). The middle panel shows the maximal vasodilation provoked by hypercapnia. Scale bar, 20 µm. (b) Identical hypercapnic challenge and imaging protocol was performed in CX3CR1GFP/+ mice after intracortical injection of SR101 to visualize astrocytes. The number of phylopodia formed at the end of contacting microglial processes (arrowheads) increased in response to hypercapnia. n = 5 mice; *, P = 0.0316, Mann–Whitney U test. Scale bar, 20 µm. (c) Perivascular microglia respond rapidly to hypercapnia with [Ca2+]i pulses in small (arrowheads) and large processes as assessed in CX3CR1CGaMP5g–tdTomato mice. Individual processes were followed with in vivo two-photon resonant (31-Hz) imaging; see also Video 7. Scale bar, 10 µm. n = 4 mice; ***, P = 0.001, Mann–Whitney U test. (d) In vivo two-photon imaging reveals impaired vasodilation at the level of penetrating arteries in the absence of microglia. n = 22 and n = 18 vessels from eight control and six depleted mice; **, P = 0.0013, unpaired t test. The experimental protocol shown for hypercapnic (hyp.) challenge was identical for in vivo two-photon imaging (a–d and j–l) and LSCI (e–i). (e) Difference images show reduced CBF response in microglia-depleted mice to hypercapnic challenge (ROIs are labeled with arrowheads). Scale bar, 1 mm. (f) The average kinetics of hypercapnic responses show difference in depleted mice compared with controls. n = 14–12 ROIs from seven control and six depleted mice, two ROIs/mouse (f and g); ****, P < 0.0001, Mann–Whitney U test (f); *, P = 0.0472, unpaired t test (g). (g and h) Hypercapnia-evoked CBF response is markedly decreased in the absence of microglia under ketamine-medetomidine (Ket./med.; g) or Ket./med. (h) anesthesia after administration of atipamezole (Atip.). n = 12–10 ROIs from six control and five depleted mice, two ROIs/mouse; *, P = 0.0436, unpaired t test. (i) Hypercapnia-evoked CBF response is markedly decreased in P2Y12R KO mice as assessed by LSCI. n = 16 control, n = 13 P2Y12R KO; *, P = 0.0131, unpaired t test. (j) In vivo two-photon imaging reveals that elimination of P2Y12R impairs hypercapnia-induced vasodilation in double transgenic (CX3CR1GFP/+ × P2Y12R KO) mice compared with P2Y12R-competent CX3CR1GFP/+ mice. n = 8 and 8 vessels from n = 5 control and n = 5 P2Y12R KO mice; *, P = 0.0104, Mann–Whitney U test. Scale bar, 20 µm. (k) The number of phylopodia formed at the end of perivascular microglial processes in response to hypercapnia is significantly reduced in P2Y12R KO mice. n = 5 control and n = 5 P2Y12R KO mice; **, P = 0.003, Mann–Whitney U test. (l) During hypercapnic challenge, neuronal activity did not differ between control, microglia-depleted, and P2Y12R KO mice. n = 49 single units in control, n = 44 in depleted, and n = 61 in P2Y12R KO group; P = 0.4852, Kruskal–Wallis test with Dunn’s multiple comparison. (m) Single image planes for CLSM imaging show small blood vessel segments from the second to third layer of the neocortex in acute brain slices. Lectin (blue) outlines the vessels, CD13 labels contractile elements (pericytes and smooth muscle cells), microglial P2Y12R is orange, and cGMP signal can be seen in green. cGMP levels were measured within areas (outlined by white dashed line) masked based on CD13 staining. A low level of basal cGMP levels can be seen under control conditions, while hypercapnia induced a robust increase in vascular cGMP levels. Preincubation with the P2Y12R inhibitor PSB0739 abolished hypercapnia-induced cGMP elevation. As a control, application of the NO donor SNP also induced robust cGMP production. Scale bar is uniformly 15 µm. n = 3 mice; ***, P < 0.0001, Kruskal–Wallis test. Data are expressed as mean ± SEM (b–d and f–l) and median ± IQR (m). LSCI data have been pooled from two to three independent experiments.

Fig 4: Characterization of pericyte coverage and pericyte-specific gene and protein expression in 26.5–32 week-old Ins2AKITA and WT littermate controls. (a,b) Representative images of pericyte-specific protein expression in WT and Ins2AKITA mice, (a) Aminopeptidase N (ANPEP, green) and Vitronectin (VTN, red); (b) Laminin alpha 2 (LAMA2, green) and desmin (DES, red). Endothelium visualized with PECAM1, cyan. n = 2, scale bar 30 µm. (c) The skeletal length of PECAM1 positive capillaries and ANPEP positive pericytes in Ins2AKITA and WT was measured and plotted as the percentage of the pericyte length over vessel length (n = 6, *p = 0.0125, student’s t test). (d,e) qPCR analysis on isolated brain microvasculature fragments for pericyte-specific gene Pdgfrb (d) and Rgs5 (e). WT controls are set as 1 and Ins2AKITA results are presented as fold change over WT (n = 5). n.s. = not significant, student’s t test. Data is presented as mean ± SEM.

Fig 5: Representative image of immunostaining corresponding to the glucose transporter protein SLC2A1 (GLUT1) in 30 week-old Ins2AKITA and WT cerebral cortex. Mural cells (ANPEP, green), glucose transporter 1 (GLUT1, red), endothelium (PECAM1, cyan). n = 2, scale bar 50 µm.