[1] 11w The Insulin-Like Growth Factor I Receptor Regulates Glucose Transport by Astrocytes

[1] 176w Previous findings indicate that reducing brain insulin-like growth factor I receptor (IGF-IR) activity promotes ample neuroprotection. We now examined a possible action of IGF-IR on brain glucose transport to explain its wide protective activity, as energy availability is crucial for healthy tissue function. Using 18 FGlucose PET we found that shRNA interference of IGF-IR in mouse somatosensory cortex significantly increased glucose uptake upon sensory stimulation. In vivo microscopy using astrocyte specific staining showed that after IGF-IR shRNA injection in somatosensory cortex, astrocytes displayed greater increases in glucose uptake as compared to astrocytes in the scramble-injected side. Further, mice with the IGF-IR knock down in astrocytes showed increased glucose uptake in somatosensory cortex upon sensory stimulation. Analysis of underlying mechanisms indicated that IGF-IR interacts with glucose transporter 1 (GLUT1), the main facilitative glucose transporter in astrocytes, through a mechanism involving interactions with the scaffolding protein GIPC and the multicargo transporter LRP1 to retain GLUT1 inside the cell. These findings identify IGF-IR as a key modulator of brain glucose metabolism through its inhibitory action on astrocytic GLUT1 activity.

[1] 9w R eduction of brain IGF-IR activity provides salutary effects.

[2] 104w Intriguingly, neuroprotective actions of lowered brain IGF-IR encompass a wide variety of insults of different etiology (Biondi et al., 2015;Cohen et al., 2009;De Magalhaes Filho et al., 2016;Gontier et al., 2015), and even prolong lifespan (Kappeler et al., 2008). While diverse mechanisms have been invoked to explain the ample spectrum of beneficial actions of reduced IGF-IR activity (Gazit et al., 2016;Kenyon, 2010;Lopez-Otin et al., 2013;Vilchez et al., 2014), it is possible that IGF-IR targets additional mechanisms of wide functional impact such as energy balance, as insulin-like receptors and their ligands are well known modulators of glucose and lipid handling in different tissues and species.

[3] 95w Conversely, IGF-I, the preferred ligand of IGF-IR, is generally considered a neuroprotective factor and has been proposed as a treatment for various neurodegenerative diseases (Fernandez and Torres-Aleman, 2012). This poses the paradox that either increasing IGF-I or reducing its receptor appears to lead to beneficial actions in the brain (Cohen and Dillin, 2008;Fernandez and Torres-Aleman, 2012). While these apparently contradictory observations remain largely unexplained (but see (O'Neill et al., 2012), one possibility is that IGF-IR has ligand independent actions, as recently reported for apoptotic signaling through insulin receptor (IR) and IGF-IR (Boucher et al., 2010).

[4] 74w In the present work we analyzed a possible involvement of the IGF-I receptor on glucose handling by the brain, a key aspect in tissue homeostasis that could in theory form part of neuroprotection by insulin-like receptors, but remains little explored. We now describe that the IGF-I receptor inhibits in a ligand-independent manner the activity of glucose transporter 1 (GLUT1) in astrocytes, adding further support for a broad beneficial effect of reducing brain IGF-IR levels.

[1] 200w We first examined whether IGF-IR affects brain glucose handling by reducing its levels with shRNA interference and measuring brain glucose uptake with 18 F fluoro-2 deoxyglucose micro-PET imaging. A lentiviral vector expressing IGF-IR shRNA (Supporting Information Fig. 1A) was stereotaxically injected into one side of the somatosensory cortex of wild type mice, whereas the contralateral side was injected with scrambled shRNA (Supporting Information Fig. 1B), using procedures already described in detail elsewhere (Carro et al., 2005). After measuring baseline levels, we stimulated the whiskers bilaterally to increase glucose flux into the somatosensory cortex. Whereas glucose uptake was increased in both sides after whisker stimulation (Supporting Information Fig. 1C), normalization of glucose levels in the side injected with IGF-IR shRNA with those in the scrambleinjected side, both under basal conditions and after stimulation of the whiskers, revealed a significantly greater increase in glucose uptake after stimulation (Fig. 1). No changes in glucose uptake were seen in a brain area such as the caudateputamen that is not activated after whisker stimulation. Despite the limitations of PET analysis in small animals such as mice (Kuntner et al., 2009), these results indicate that reducing brain IGF-IR enhances brain glucose uptake in active brain regions.

[2] 102w Because it has been shown that astrocytes are the predominant type of cell involved in glucose uptake upon activation of the somatosensory cortex (Chuquet et al., 2010), we used in vivo fluorescence microscopy (Supporting Information Fig. 2) to visualize astrocytes in mice injected with IGF-IR shRNA in the somatosensory cortex, using procedures already described (Perez-Alvarez et al., 2013). We found significantly larger increases after stimulation of the whiskers in glucose transport (detected with 6-NBDG) in somatosensory cortex astrocytes (identified with the astrocyte-specific fluorescent red marker SR101) injected with IGF-IR shRNA, as compared to scramble RNA-injected astrocytes (Fig. 2B-D and Supporting Information video).

[3] 144w As injection of the lentiviral particles transducing IGF-IR shRNA in the brain affected all types of brain cells, we needed to confirm whether IGF-IR in astrocytes was involved in glucose uptake, Thus, we performed PET analysis in mice with reduced IGF-IR specifically in astrocytes (As IGF-IR 1/2 ). Compared to littermates, a significant increase in glucose uptake in somatosensory cortex after sensory stimulation was seen in As IGF-IR 1/2 mice (Fig. 2D). PET analysis was performed in heterozygous mice (As IGF-IR 1/2 ) because homozygous mice (As IGF-IR 2/2 ) have reduced brain size (not shown). Astrocytes from As IGF-IR 1/2 mice showed reduced responses to IGF-I (1 nM), as determined by phosphorylation of Akt (Supporting Information Fig. 3A), even though total brain levels of IGF-IR were not significantly reduced in these mice, as many other brain cells express IGF-IR (Supporting Information Fig. 3B).

[4] 156w We then examined mechanisms whereby IGF-IR reduces glucose transport using astrocyte cultures. We first confirmed that reduction of IGF-IR in shRNA IGF-IR-transfected astrocytes significantly increased glucose transport (Fig. 3A). Because astrocytes also express IR (Supporting Information 1B). After allowing 2 weeks of recovery, animals were submitted to PET scans, and after basal (white bars) measurements of 18 F-FDG uptake they were bilaterally stimulated (black bars) in their whiskers. Basal and stimulated responses in the shRNA IGF-IR injected site were normalized to responses in the scramble-injected side. Significantly enhanced uptake in the somatosensory cortex injected with shRNA IGF-IR was seen in response to whisker stimulation. No changes were appreciated in a unstimulated area such as the caudate putamen, that was analyzed to determine region specificity in glucose responses to sensory stimulation (*P < 0.05 vs. basal). Lower panel: representative PET images under basal conditions and after whisker stimulation are shown. Greater signal (red) is seen after stimulation.

[5] 110w Fig. 4A), we also tested its role on glucose transport. Intriguingly, reduction of IR levels (Supporting information Fig. 4B) decreased glucose transport in astrocytes (Fig. 3A). Opposing actions of IR and IGF-IR on glucose transport were confirmed by the observation that when both receptors were reduced, the effects cancelled each other (Fig. 3A). Modulation of basal glucose transport in astrocytes after reduction of either IR or IGF-IR reflects ligand-independent actions of these receptors (Boucher et al., 2010), as experiments were conducted under serum-free conditions, and addition of either IGF-I or insulin in the presence or absence of the respective shRNAs did not alter glucose transport in any case (not shown).

[6] 164w Because GLUT1 is the main facilitative glucose transporter in astrocytes (Morgello et al., 1995), we determined its role in IGF-IR and IR actions. We found that decreasing IR levels reduced GLUT1 mRNA (Fig. 3B), resulting in decreased GLUT1 levels at the cell membrane (Fig. 3C). Confirming these observations in vitro, we recently found reduced levels of brain GLUT1 mRNA in mice lacking IR in astrocytes (Garcia-Caceres et al, in press). Indeed, these mice show wide disturbances in brain glucose handling. Thus, astrocytic IR modulates GLUT1 expression, and consequently GLUT1 protein levels at the cell membrane are also affected. On the other hand, reduction of IGF-IR by shRNA did not affect GLUT1 mRNA levels (Fig. 3B), but resulted in increased amounts of GLUT1 at the cell membrane of astrocytes, as determined by flow cytometry and membrane protein biotinylation (Fig. 3C,D). The latter suggests that IGF-IR retains GLUT1 inside the cell and agrees with the observation that reduction of IGF-IR levels increases glucose transport in astrocytes.

[7] 167w We then started to determine possible underlying mechanisms. IGF-IR associates with GIPC (GAIP-interacting protein, C terminus) (Booth et al., 2002;Bunn et al., 1999), a scaffolding protein that participates in protein trafficking and binds to many partners, including GLUT1 (Bunn et al., 1999). We hypothesized that GIPC may simultaneously interact with IGF-IR and GLUT1 through its PDZ domain because GIPC can dimerize (Katoh, 2013). In support of this possibility we found that IGF-IR co-immunoprecipitates not only with GIPC, as already reported, but also with GLUT1, whereas GIPC, as expected, co-immunoprecipitates also with GLUT1 (Fig. 4A). Proximity ligation assays (PLA) confirmed an interaction between IGF-IR and GLUT1 (Fig. 4B). Significantly, neither GLUT1, or as previously shown (Ligensa et al., 2001), nor GIPC, co-immunoprecipitate with the IR (Fig. 4A). To further establish a role of GIPC in the interaction between IGF-IR and GLUT1 we reduced GIPC levels in astrocytes using GIPC shRNA (Supporting Information Fig. 4C) and found that the interaction between IGF-IR and GLUT1 was significantly reduced (Fig. 4C).

[8] 112w We examined a potential intermediary role of lipoprotein-receptor associated protein 1 (LRP1) in this mechanism as LRP1 is a multicargo membrane protein that associates to the IGF-I receptor in other type of brain cells (Nishijima et al., 2010) and is involved in GLUT translocation in neurons (Liu et al., 2015). We confirmed that LRP1 and IGF-IR interact also in astrocytes (Supporting Information Fig. 4D), and that LRP1 is required for IGF-IR to interact with GLUT1 and GIPC as its reduction with shRNA (Supporting Information Fig. 4E) reduces the amount of IGF-IR that immunoprecipitates with GLUT1 and GIPC (Fig. 4D). No direct interaction of LRP1 with GLUT1 was observed (Supporting Information Fig. 4F).

[1] 66w The present observations indicate that IGF-IR regulates glucose transporter 1 activity in astrocytes, impacting in this way on overall brain glucose transport. When IGF-IR levels are reduced, GLUT1 locates at the cell membrane and transports glucose inside the cell. Thus, unbound GLUT1 may contribute to basal glucose transport in astrocytes. Conversely, when Hernandez-Garz on et al.: Insulin-like Growth Factor I Receptor and Brain Glucose Month 2016

[2] 141w GLUT1 is bound to IGF-IR, it remains inside the cell and associates to IGF-IR through protein-protein interactions involving GIPC and LRP1 (see Summary Graphic). As a facilitative glucose transporter, GLUT1 takes up extracellular glucose in a concentration-dependent manner, which explains increased glucose transport in shRNA IGF-IR-injected mice receiving sensory stimulation. Indeed, local blood flow is increased in response to enhanced neuronal activity (Roy and Sherrington, 1890), and astrocytes, that fully wrap blood vessels (Mathiisen et al., 2010), will take up more glucose through GLUT1. In turn, previous observations indicate that activity-dependent modulation of astrocytic glucose transport relies on stimulation of GLUT1 by glutamate via engagement of the Na 1 -glutamate co-transporter and intracellular Na 1 -Ca 11 co-signaling (Loaiza et al., 2003;Porras et al., 2008). Collectively, these observations provide additional information about the role of astrocytic GLUT1 in brain glucose metabolism.

[3] 123w In its IGF-IR-free state, GLUT1 is in the membrane and may be active. In this regard, and not withstanding different sensitivities of the methods used, the increase in glucose transport seen in astrocytes with reduced levels of IGF-IR is more robust that the increase seen in membrane GLUT1 levels. This could mean that the intrinsic activity of GLUT1, that can be regulated (Barnes et al., 2002), is also affected by IGF-IR. Thus, when associated to IGF-IR, GLUT1 is retained inside the cell and its activity is downregulated. Bound and unbound GLUT1 may constitute two separate pools in astrocytes. We also speculate that when bound to IGF-IR, GLUT1 may become subject to regulation by extracellular signals (glutamate. . .etc), but further work is needed.

[4] 234w This study confirms the utility of 6NBDG as a probe for glucose uptake, further supporting the notion that astrocytes are major contributors in glucose metabolism, as seen previously both in vivo (Chuquet et al., 2010), and in different in vitro preparations (Barros et al., 2009b;Jakoby et al., 2014). Importantly, both in anesthetized animals and in slices without anesthetics, 6NBDG showed preferential astrocytic uptake. Another important aspect confirmed by our findings is that IR and IGF-IR display ligand-independent activities that may, or may not, be related to the actions of their ligands (Boucher et al., 2010). In this regard, different authors have shown that IGF-I and insulin also affect glucose handling by astrocytes at different levels, including enhanced glucose uptake (Kum et al., 1992;Masters et al., 1991) and/ or enhanced glycogen production (Dringen and Hamprecht, 1992;Hamai et al., 1999;Muhic et al., 2015). These observations suggest that insulin peptides and their receptors form an intricate glucose regulatory network in astrocytes that may even act in apparently opposing manners. Indeed, and through entirely different mechanisms, IGF-IR exerts an intrinsic inhibitory action on astrocytic glucose transport, while IR displays an intrinsic stimulatory activity. In this way, glucose uptake in astrocytes may in part be determined by a balance between IGF-IR and IR levels. This suggests that physiological and pathological processes impacting on astrocytic insulin and IGF-I receptor levels will influence glucose transport by the brain in opposite directions.

[5] 153w Collectively, these observations would help reconcile the apparent controversy on the role of these receptors and their ligands in the brain (Cohen and Dillin, 2008), which largely arises from studies in invertebrates harboring a single insulinlike receptor (Kenyon, 2010). Thus, while in C elegans a single insulin-like receptor is modulated by many different ligands, even in an antagonistic fashion (Matsunaga et al., 2012), the acquisition of new insulin-like receptors in vertebrates has allowed the appearance of interactions among them. Based on present findings we consider that reported actions of insulin-like receptors in invertebrates should not be immediately inferred to be similar in vertebrates. The corollary of these observations is that invertebrate models of insulin-like receptor physiology in mammals should take into account the existence of two tyrosine kinase receptors, IGF-IR and IR, not present in invertebrates that may display cooperative (Boucher et al., 2010) or opposing activities (present observations), depending on biological context.

[6] 169w Finally, since glucose transport by the brain deteriorates during aging and its associated pathologies (Mergenthaler et al., 2013), these observations provide further support for the use of strategies regulating brain IGF-IR levels to support healthy as well as pathological aging. Indeed, lowering IGF-IR in Alzheimer's disease (AD) brains will not only diminish amyloidosis (Cohen et al., 2009), and related pathology such as hippocampal hyperactivity (Gazit et al., 2016), but will also likely contribute to normalize glucose dysregulation present as a characteristic alteration of this disease (La Joie et al., 2012). Future work should examine major components of the IGF-IR pathway in astrocytes described herein for brain glucose regulation in experimental models of normal and pathological aging, as for example the recently described role of GLUT1 in AD (Winkler et al., 2015). Altogether, these set of observations support a role of an interaction between insulin and IGF-I receptors in modulating glucose handling by astrocytes, adding a new layer of regulation by astrocytes of brain energy economy (Allaman et al., 2011).

[1] 412w Adult (3-5 months old) and neonatal C57BL6/J mice were used. Mutant mice with reduced levels of IGF-IR in astrocytes (AsIGF-IR 1/2) were obtained by crossing IGF-IR floxP/floxP mice (the IGF-IR gene flanked by LoxP sites; Jackson Labs) with GFAP-Cre mice (Cre expression under the human GFAP promoter; Jackson Labs), both in a C57BlJ background. Littermates with IGF-IR flox/flox , IGF-IR flox/2 , and IGF-IR 1/1 Cre 1/? were pooled and used as controls. GFAP-Cre mice crossed with Rosa26 tomato-eGFP (a reporter mouse line from Jackson Labs) mice express GFP in astrocytes, whereas AsIGF-IR 2/2 mice expressed Cre in astrocytes (not shown). While levels of IGF-IR in the brain of AsIGF-IR 2/2 mice were significantly reduced (see below), levels of IR were normal (not shown). Animals were genotyped by PCR using primers for GFAP-Cre forward: ACT CCT TCA TAA AGC CCT and reverse: ATC ACT CGT TGC ATC GAC CG, and for IGF-IR forward: CTT CCC AGC TTG CTA CTC TAG G and reverse: CAG GCT TGC AAT GAG ACA TGG G. Two other transgenic mice lines (hGFAP-CreERT2 and IR loxP/loxP mice) were crossed to obtain GFAPIR-KO mice lacking IR in astrocytes when injected with tamoxifen, as explained in detail elsewhere (Garcia-Caceres et al., in press). hGFAP-CreERT2 mice, an inducible transgenic mouse line under the control of a GFAP promoter and estrogen (C57BL/6J background, FM Vaccarino, Yale University School of Medicine) were mated with IR loxP/loxP mice (generated by R Kahn, Joslin Diabetes Center), and breeding cages were maintained by mating IR loxP/loxP and IR loxP/loxP ;hGFAP-CreERT2 mice. To excise loxP sites by Cre recombination, 6 weeks-old male mice were administrated a daily tamoxifen injection (10 mg/kg, intraperitoneal) for 5 days. Tamoxifen (Sigma) was dissolved in sunflower oil at a final concentration of 10 mg/mL at 37 , and then filter sterilized and stored for up to 7 days at 4 C in the dark. IR loxP/loxP mice were used as controls and also were injected with tamoxifen. PCR genotyping was carried out using primer sets binding to Cre (Cre-1084, 5 0 -GCG GTC TGG CAG TAA AAA CTA TC-3'; Cre-1085, 5 0 -GTG AAA CAG CAT TGC TGT CAC TT-3'; Cre-42, 5 0 -CTA GGC CAC AGA ATT GAA AGA TCT-3'; Cre-43, 5 0 -GTA GGT GGA AAT TCT AGC ATC ATC C-3 and crossing the loxP site (oKAHN03: 5-GAT GTG CAC CCC ATG TCT G-3'; oKAHN04: 5-TCT ATC AAC CGT GCC TAG AG-3'; oKAHN05: 5-CTG AAT AGC TGA GAC CAC AG-3').

[2] 28w Animal procedures followed European (86/609/EEC & 2003/65/EC, European Council Directives) guidelines and studies were approved by the respective local Bioethics Committees. All in vivo experiments were done blinded.

[3] 79w Astroglial cultures with >95% GFAP-positive cells were prepared as described (Fernandez et al., 2007). Postnatal (day 3-4) brains from wild type, mutant As IGF-IR , and littermate mice were dissected and immersed in ice-cold Hank's balance salt solution (HBSS, Life Technologies, Spain). Cortex and hippocampus were removed and mechanically dissociated. The resulting cell suspension was centrifuged and plated in DMEM/F-12 (Life Technologies) with 10% fetal bovine serum (Life Technologies) and 100 mg/mL of antibiotic-antimycotic solution (Sigma-Aldrich). After 15-20 days,

[4] 96w We used 6-NBDG to measure glucose transport in astrocytes as shown by others (Barros et al., 2009a). Briefly, cells well starved in serum-free media for 3 hours. Then IGF-I (PreProTec, UK), insulin (Sigma, USA), or vehicle were added to a final concentration of 1nM. We then added 6-NBDG (Setareh biotech, USA) to a final concentration of 30 mM. Cultures were kept for 3 hours at 378C and then ice cold PBS was added and cells trypsinized. Cells were collected and FBS and PBS added. Fluorescence intensity was measured by flow cytometry (FACSAria cytometer, BD Biosciences, USA).

[5] 6w In Situ Proximity Ligation Assays (PLA)

[6] 204w GLUT1 -IGF-IR interactions were detected in astrocytes grown on glass coverslips using the Duolink II in situ PLA detection Kit (OLink; Bioscience, Sweden) as previously described (Gonzalez et al., 2012). Astrocytes were fixed in 4% paraformaldehyde for 10 min, washed with PBS containing 20 mM glycine to quench the aldehyde groups, permeabilized with the same buffer containing 0.05% Triton X-100 for 5 min and successively washed with PBS. After 1 h/378C with the blocking solution in a pre-heated humidity chamber, astrocytes were incubated overnight with primary antibodies: rabbit polyclonal anti-GLUT1 antibody (1:100, ref. sc-7903; Santa Cruz Biotechnology) and monoclonal mouse anti-IGF-I receptor antibody (1:100, ref. sc-463; Santra Cruz Biotechnology) and were processed following the instructions of the supplier using the PLA probes detecting rabbit or mouse antibodies (Duolink II PLA probe anti-Rabbit plus and Duolink II PLA probe anti-Mouse minus diluted in antibody diluent to a concentration of 1:5) and a DAPI-containing mounting medium. Samples were observed in a Leica SP2 confocal microscope (Leica Microsystems, Germany) equipped with an apochromatic 63X oil-immersion objective. For images of each field a maximum projection (superimposed sections) in two channels (one per staining) of 6 to 12 Z stacks with a step size of 1 mm were acquired.

[7] 40w Translocation of GLUT1-Flag and IGF-IR to the cell membrane was evaluated following previously published procedures (Koshy et al., 2010). In brief, cultured astrocytes were labeled with anti IGF-IRa (SC-463, 1:50, Santa Cruz Biotechnology) or anti Flag M2 (F1804, 1:1000, Sigma-Aldrich)

[8] 22w and secondary antibody Alexa Fluor 488 (A-11008, 1:1000, Life Technologies), and fixed before assessing fluorescence intensity by flow cytometry (FACS Aria, BD).

[9] 80w Mice were anesthetized by inhalation of a mixture of isofluorane/oxygen (5% induction, 2% maintenance). After removing the duramater, the tip of the glass pipette was placed onto the surface of the brain. Two microliter of lentiviral shRNA against IGF-I receptor (4 3 10 10 pfu/mL) were administered per mouse. Administration was made through a glass pipette connected to a Hamilton syringe. Rate of infusion was 1 mL per 10 min. Stereotaxical coordinates were 21.06 mm from bregma and 21mm lateral.

[10] 202w In Vivo Astrocyte Glucose Uptake Glucose uptake by astrocytes was evaluated as previously described by others (Chuquet et al., 2010) according to the experimental set-up shown in Supporting Information Fig. 1A, following procedures described in detail elsewhere (Perez-Alvarez et al., 2013). Astrocytes were labeled with sulforhodamine 101 (100 mg/kg, i.p. SR101; Sigma-Aldrich). Mice were anesthetized with urethane (1.7g/kg, i.p. Sigma-Aldrich) and their femoral artery cannulated. A 4 mm craniotomy around the area of interest was made. After 5 min, the cortex was washed and a drop of low melting point agarose (1% in HEPES-buffered solution. Sigma-Aldrich) and a 5 mm glass coverslip were placed carefully over the exposed cortex. Dental cement (Fortex, Facident, Spain) was applied to fix the coverslip. A light aluminum frame (2 3 3.5cm) with a central circular hole (10mm diameter) was attached to the skull centered on the craniotomy area and fixed with dental cement. The cranial frame was fixed to a heavy aluminum base. The base was moved to the imaging stage. The animals' body temperature was monitored during the procedure using a rectal probe (Technomed Europe, The Netherlands) and regulated using a heating pad (RS Amidata, Spain) controlled by a thermostat (Cibertec, Spain) set at 378C.

[11] 66w 6-NBDG administration. 6-NBDG (Setareh biotech, USA) was dissolved in a solution of 55% of HEPES-ringed buffer and 45% of DMSO, pH 7.42, to a concentration of 5 mg/mL. A 300 mL Hamilton syringe (Hamilton, USA) filled with the 6-NBDG solution was connected to the femoral vein cannula and placed in a micro-injector pump (Harvard Apparatus, USA). 6-NBDG was pumped at a rate of 20 lL/min/100 g.

[12] 91w The whiskers of the animal's snout were stimulated with 100 ms puffs of air produced at 5 Hz by a pressure injector (Dagan, USA) for 30 sec controlled by an Axon Digidata 1322A and pClamp software (Molecular Devices, USA). Air was ejected at 1 bar pressure via capillary glass, attached to plastic tubing, positioned 1 cm lateral and anterior to the animal's nose to stimulate the whole left whisker pad. At the same time, tail pinching was performed at 2 Hz with steel forceps, providing a pairing protocol for astrocyte stimulation.

[13] 98w Astrocytes were distinguished by using the red signal emitted by SR-101. Astrocytes concentrate SR-101, and their soma appear intensely bright (Nimmerjahn et al., 2004). Each image of the sequence was aligned over the previous image with Align Slice (ImageJ, National Institutes of Health, USA) to correct x-y deviation caused by possible drift of the tissue. Fluorescence intensity was measured in a region of interest (ROI) strictly limited to the somatic area. Signals were expressed as relative fluorescence changes (DF/F0), where F0 was the mean of the baseline period. Astrocytes showing variations greater than 1 were considered as responders.

[14] 260w Briefly, fasted mice were injected i.p. with the positron emitting radiotracerfoot_1 F-FDG (18.5 MBq in 0.2 mL of 0.9% NaCl, Instituto Tecnol ogico PET, Spain). During an uptake period of 45 min animals were anesthetized by inhalation of a mixture of isoflurane/oxygen (5% for induction and 2% for maintenance) and then placed on the bed of the tomograph. The duration of the PET acquisition was 20 min, immediately followed by a CT (computed tomography) scan. The scanner used was a specific small animal PET-CT hybrid tomograph (Albira ARS, Oncovision, Spain). After acquisition, PET images were reconstructed with an ordered subset expectation maximization (OSEM) algorithm, and with applied corrections for randoms, scatter, attenuation, dead time and radio element decay, whereas for the CT images a filtered back projection algorithm was used. For metabolic activity quantification, the procedure used was as follows: first, the CT image of the skull from each animal was co-registered to a magnetic resonance image (MRI) template of mouse brain in which the regions of interest (ROIs) were previously delineated. After the CT image was co-registered, the spatial mathematic transformation was saved and then applied to its own fused PET image, allowing the correct matching between the PET image and the MRI template. Once the 18 F-FDG uptake in the different brain regions was calculated (in kBq/cc units), the activity of each left hemisphere region was normalized to its homologous region in the right hemisphere and expressed as proportional uptake (left/right). All processes of visualization, co-registration and quantification were performed using PMOD 3.0 software (PMOD Technologies Ltd., Switzerland).

[15] 87w Normal distribution tests were carried out in all initial set of experiments and a non-parametric Wilcoxon test was applied accordingly. For samples with normal distribution, parametric tests include one-way ANOVA followed by a Tukey HSD or t-test. A P < 0.05 was considered significant. Results are shown as mean 6 s.e.m. No statistical methods were used to predetermine sample sizes. Data collection and analysis were performed blinded to the conditions of the experiments only in in vivo experiments. There was no randomization of data collection or processing.

[1] 204w For viral transduction we used a three-plasmid system previously described (Dull et al., 1998). The co-transfection system consisted of an shRNA plasmid against either IGF-IR or IR (also used for transfecting primary astrocyte cultures), a packaging construct (pCMV-dR8.2 Dvpr) and the vesicular stomatitis virus G-protein envelope (pMD2.G, Addgene, USA). shRNAs against Glut-1, IGF-IR, IR, scramble sequence and EGFP were from Origene (HuSH-29, Origene, USA): shRNA against GIPC was constructed as described in www.addgene.org/tools/protocols/plko/using the primers: 5 0 CCGGACTCACCGAACCTCGGAAGGCCTCGAGGCCTTCC GAGGTTCGGTGAGT TTTTTG3 0 and 5 0 AATTCAAAAAACTC ACCGAACCTCGGAAGGCCTCGAGGCCTTCCGAGGTTCGGT GAGT3 0 directed against the 684-704 fragment of GIPC mRNA. The transfer vector (5 lg), the envelope (2 lg), and the packaging plasmids (5 lg) were co-transfected using calcium phosphate in human embryonic kidney 293 T cells (6 3 106 cells per dish) cultured in DMEM with 10% FCS and 1% penicillin/streptomycin. Lysosomal function was inhibited with cloroquine prior to transfection. The supernatant containing the viral particles was collected, filtered and stored at 2808C until use. Viral concentration was titrated as described (Munive et al., 2016). Infection efficiency was 80% as determined using GFP-expressing viral particles. shLRP-1 was obtained as described (Nishijima et al., 2010). GLUT1-Exo Flag was a kind gift of JC Rathmell (Wieman et al., 2007).

[2] 13w Glucose transporter 1 IGF-IR Insulin-like growth factor I receptor IR Insulin receptor LRP1

[3] 84w Lipoprotein-receptor associated protein 1 LSCM Laser scanning confocal microscopy PLA Proximity ligation assays astrocytes were re-plated at 1.2 3 10 5 cells/well. For transfection, astrocytes were electroporated (2 3 10 6 astrocytes with 2 mg of plasmid DNA) before seeding using an astrocyte Nucleofector Kit (Amaxa, Lonza, Switzerland). After electroporation, cells were plated to obtain a final cell density on the day of the experiment similar to that obtained with the transfection method. The transfection efficiency was 60-80%, as assessed with a GFP vector.

[4] 45w Cell surface proteins were biotinylated following the manufacturer's instructions (EZ-Link TM Sulfo-NHS-SS-Biotin, Thermo Scientific). Biotinylated proteins were purified by affinity chromatography using NeutroAvidinAgarose Resin (Thermo Scientific) and resolved by Western blot. The membrane protein Na 1 /K 1 ATPase was used as a loading control.

[5] 37w Quantitative PCR Total RNA isolation from cell lysates or brain tissue was carried out with Trizol. One mg of RNA was reverse transcribed using High Capacity cDNA Reverse Transcription Kit (Life Technologies) according to the manufacturer's instructions.

[6] 113w For quantification of specific genes, total RNA was isolated and transcribed as above and 62.5 ng of cDNA was amplified using TaqMan probes for GLUT1, GluT4, IGF-IR or IR, and 18S as endogenous control (Life Technologies). Each sample was run in triplicate in 20 lL of reaction volume using TaqMan Universal PCR Master Mix according to the manufacturer's instructions (Life Technologies). All reactions were performed in a 7500 Real Time PCR system (Life Technologies). Quantitative real time PCR analysis was carried out as described (Pfaffl, 2001). Results were expressed as relative expression ratios on the basis of group means for target transcripts versus reference 18S transcript. At least three independent experiments were done.

[7] 42w Imaging was performed with a custom-built confocal laser (CVI Melles Griot, UK) scanning microscope consisting of an Olympus FV300 laser scanning confocal system coupled to an Olympus BX61WI upright microscope (Olympus, Japan) and a Olympus LUMPLFL 60XW/IR water immersion objective (0.9NA; Olympus)