An early endothelial cell–specific requirement for Glut1 is revealed in Glut1 deficiency syndrome model mice

Paucity of the glucose transporter-1 (Glut1) protein resulting from haploinsufficiency of the SLC2A1 gene arrests cerebral angiogenesis and disrupts brain function to cause Glut1 deficiency syndrome (Glut1 DS). Restoring Glut1 to Glut1 DS model mice prevents disease, but the precise cellular sites of action of the transporter, its temporal requirements, and the mechanisms linking scarcity of the protein to brain cell dysfunction remain poorly understood. Here, we show that Glut1 functions in a cell-autonomous manner in the cerebral microvasculature to affect endothelial tip cells and, thus, brain angiogenesis. Moreover, brain endothelial cell–specific Glut1 depletion not only triggers a severe neuroinflammatory response in the Glut1 DS brain, but also reduces levels of brain-derived neurotrophic factor (BDNF) and causes overt disease. Reduced BDNF correlated with fewer neurons in the Glut1 DS brain. Controlled depletion of the protein demonstrated that brain pathology and disease severity was greatest when Glut1 scarcity was induced neonatally, during brain angiogenesis. Reducing Glut1 at later stages had mild or little effect. Our results suggest that targeting brain endothelial cells during early development is important to ensure proper brain angiogenesis, prevent neuroinflammation, maintain BDNF levels, and preserve neuron numbers. This requirement will be essential for any disease-modifying therapeutic strategy for Glut1 DS.

Tip cell defects are evident in Glut1-stained cortical sections of PND14 Glut1 D/+ and Tie2-Cre;Glut1 FL/+ mutants but rescued following repletion of the protein; tip cells indicated by arrows. Also depicted are magnified images of representative tip cells from each of the mice. Note the especially intense Glut1 staining at termini of control tip cell (arrowheads). Tip cell numbers are (B) reduced and have (C) significantly fewer and (D) shorter lamellipodia in Glut1 D/+ and Tie2-Cre;Glut1 FL/+ mutants than they do in healthy controls or mutants restored for Glut1. Note: *,***, P < 0.05 and P < 0.001 respectively, one-way ANOVA, n≥9 regions from each of N=3 mice of each genotype examined for panels B -D. Thalamic brain sections of PND14 mice stained for Glut1 reveal a relative paucity and defects of mutant endothelial tip cells (arrows). Quantified results of (B) tip cell numbers, (E) lamellipodia counts and (F) lamellipodia size in mutants and controls. Note: **, ***, P < 0.01, P < 0.001 respectively, t test, n≥9 regions from each of N=3 mice of each genotype examined for panels B -D.

Supplementary Information
Rotarod test. To administer the rotarod test, mice were subjected to a training period of 5 minutes on an accelerating rotarod (Ugo Basile Inc., Italy) three times a day for four consecutive days. Measurements were recorded on the fifth day at a setting of 25rpm. Duration of time on the rotating rod was recorded and the experiment terminated if a mouse surpassed 1000s.
Brain parenchymal and vessel fractions. Mice were perfused with 1X PBS, whole brains extracted and the tissue gently homogenized by means of Dounce-type glass homogenizer in 1ml PBS. The extract was centrifuged at 1000g for 5min, the resulting pellet re-suspended once again in 1ml PBS and the centrifugation step repeated. The supernatant was then removed and the pellet re-suspended in 1ml of an 18% dextran solution in PBS. This suspension was centrifuged at 10,000g for 1min, the pellet saved and the supernatant containing the parenchymal fraction transferred to new tube. The pellet was once again re-suspended in 1ml of the 18% dextran and the centrifugation repeated. This process was repeated a third time and the pellets containing the vessel fractions were pooled and stored at -80°C until use. The supernatant fractionscontaining the neuropil -were similarly combined and stored until use.
Quantification of neurons and activated glia in thalamic brain. Thalamic brain in the region of the ventral posteromedial (VPM) nucleus was imaged at a magnification of 10X and the external medullary lamina used as an anatomical marker to demarcate identical regions of the nucleus in the different mice. Neurons within the demarcated (also see fig. S7) dorsal two-thirds of the nucleus (dorsal VPM) were enumerated. Counts were conducted by the ImageJ software suite (NIH, Bethesda, MD). Activated glial cells were enumerated manually in images of the dorsal VPM acquired at a magnification of 63X.
Live-imaging of brain microvasculature. Five month old mice were anesthetized (1500 mg/kg urethane and 500 mg/kg glycopyrrolate, administered I.P.). Next, a craniotomy on the right hemisphere between bregma and lamda was performed and then fluorescein-conjugated dextran (2000 kDa, 0.1 ml from 25 mg/ml) injected into the tail vein to enable visualization of the capillaries. Images were acquired using a home-built twophoton laser scanning microscope (41, 42) equipped with a 20X, 0.95 NA objective lens (XLUMPLanFl, Olympus). Stacks of angiograms (~ 510 x 510µm) were constructed beginning at the cortical surface down to a depth of ~ 500µm. Images were acquired every 2µm in the z-axis. Microvascular length was quantified by modifying an image processing pipeline previously described (42). Image analysis was performed using ImageJ and MATLAB. Briefly, three sub-regions (510 x 510 x 20µm) at depths of 200, 300 and 400µm were selected from the stack. The mean image of the sub-region was first pre-processed with a tubeness filter to enhance the features of the vessels. Then, an automatic intensity thresholding was applied to segment the vessels.
Capillary diameter was further determined by skeletonization and Euclidean distance map. Only blood vessel segments with diameters of < 6µm were included in the final result.    Marked reduction in Glut1 transcripts in brain tissue of 5-month old CreER;Glut1 FL/FL mutants administered tamoxifen at 8 weeks of age to inactivate the floxed alleles; ***, P < 0.001, t test, n=3 controls and n=7 mutant mice. (B) Western blot of Glut1 protein in brain tissue of CreER;Glut1 FL/FL mutants treated with tamoxifen at 8 weeks of age. (C) Quantified result of Glut1 protein in brain tissue of tamoxifen-treated CreER;Glut1 FL/FL mutants; ***, P < 0.001, t test, n=3 mice of each cohort. (D) Depleting Glut1 below 50% in adult mice severely impairs motor performance on the rotarod. Note precipitous decline in performance 4 weeks after tamoxifen administration; *, ***, P < 0.05, P < 0.001 respectively, t test, n≥6 mice of each cohort analyzed. (E) Kaplan-Meier survival curves depict a modest reduction in lifespan of CreER;Glut1 FL/FL mutants depleted of Glut1 below heterozygous levels during adult life; n=10 mice of each cohort.