Diet suppresses glioblastoma initiation in mice by maintaining quiescence of mutation-bearing neural stem cells

Summary Glioblastoma is thought to originate from neural stem cells (NSCs) of the subventricular zone that acquire genetic alterations. In the adult brain, NSCs are largely quiescent, suggesting that deregulation of quiescence maintenance may be a prerequisite for tumor initiation. Although inactivation of the tumor suppressor p53 is a frequent event in gliomagenesis, whether or how it affects quiescent NSCs (qNSCs) remains unclear. Here, we show that p53 maintains quiescence by inducing fatty-acid oxidation (FAO) and that acute p53 deletion in qNSCs results in their premature activation to a proliferative state. Mechanistically, this occurs through direct transcriptional induction of PPARGC1a, which in turn activates PPARα to upregulate FAO genes. Dietary supplementation with fish oil containing omega-3 fatty acids, natural PPARα ligands, fully restores quiescence of p53-deficient NSCs and delays tumor initiation in a glioblastoma mouse model. Thus, diet can silence glioblastoma driver mutations, with important implications for cancer prevention.


INTRODUCTION
Increasing evidence indicates that glioblastoma (GBM) originates from neural stem cells (NSCs) of the subventricular zone (SVZ) neurogenic niche. 1,2SVZ NSCs, which are largely quiescent (qNSCs) in the normal adult brain, 1 were shown to bear mutations in cancer-driving genes, including in TERT, TP53 and EGFR in GBM. 2 This suggests that these genes may control transitions between quiescence and activation to a proliferative state at the onset of tumorigenesis.Therefore, deciphering the impact of GBM-relevant mutations on the biology of qNSCs should provide understanding of disease etiology and reveal approaches for cancer prevention.
The transcription factor p53 (TP53 in humans, Trp53 in mice) is the most frequently mutated gene in human cancer, and the p53 pathway is altered in 87% of GBM patients. 3Interestingly, p53 has also been linked to the regulation of adult murine SVZ neurogenesis. 4,5In rodents, neurogenesis arises from a subpopulation of NSCs termed ''type-B cells.'' 6Although largely quiescent, adult murine type-B cells can activate to a proliferative state (active NSCs, [aNSCs]) from which they give rise to transit amplifying progenitors (type-C cells), which in turn fuel the production of neurons and glia. 6Analysis of constitutive p53 knockout mouse models showed that p53 restrains type-C cell proliferation and neuronal differentiation. 4,5In contrast, its role in qNSCs is less clear, with reports of no changes to a mild increase, 4,5 likely due to compensatory effects.Here, we combined a conditional and inducible p53 knockout mouse model (p53 icKO ) with mechanistic assays and GBM mouse models to examine the function of p53 and its effectors in adult SVZ qNSCs and glioma initiation.

RESULTS p53 enforces NSC quiescence
To examine the acute effects of p53 loss in qNSCs and lineagetrace recombined cells and their progeny, p53 icKO (p53 LoxP/LoxP ; tdTom fl/+ ;GLAST::CreERT2) or control mice (p53 +/+ ;tdTom fl/+ ; GLAST::CreERT2) were treated with tamoxifen at 7 weeks of age. 4,5Following confirmation of recombination efficiency (>60%) (Figures S1A and S1B) and type-B cell specificity (Figures S1C and S1D) by immunocytochemistry and FACS analysis, respectively, we performed 5-ethynyl-2'-deoxyuridine (EdU)label retention experiments to investigate effects of p53 loss both in type-B cells that had previously activated and returned to quiescence (resting qNSCs) and in type-B cells that had not yet activated (dormant qNSCs). 7For resting qNSCs, EdU was administered for 7 days, followed by a 5-day tamoxifen administration and analysis of the SVZ in whole-mount preparations 3 days later (Figure 1A).The percentage of resting qNSCs (identified as radial EdU + /Ki67 À /tdTom + cells) was markedly reduced in p53 icKO mice relative to controls, indicative of premature re-activation upon p53 loss (Figures 1B and 1C).This was accompanied by an increase in pairs of EdU À /Ki67 + and Ascl1 + /td-Tom + type-C cells, which likely represent the immediate progeny of non-label-retaining recombined NSCs (Figures 1D, 1E, S1E, and S1F). 7In contrast, the numbers of 3-and 4-cell clusters were similar in control and p53 icKO animals (Figures S1G and S1H), confirming that effects were selective to type-B cells.To examine dormant qNSCs, we administered EdU for 14 days, a time-window during which most resting qNSCs incorporate EdU, while dormant qNSCs remain unlabeled. 7Recombination of p53 was induced during the last 5 days of the labeling period, and whole mounts analyzed 1 day later to identify dormant qNSCs undergoing activation based on Ki67 positivity and lack of EdU (Figure 1F).The loss of p53 induced premature activation of dormant qNSCs (Figures 1G and 1H), accompanied by a trend toward an increase in EdU À /Ki67 + type-C cell pairs, which did not reach significance (Figure 1I) and a significant increase in EdU + /Ki67 + type C cell pairs.The latter likely represented immediate progenitors of recombined qNSCs (resting and dormant) that activated during the tamoxifen/EdU administration period (Figure 1J).Thus, p53 maintains type-B cell quiescence and its deletion aberrantly activates resting and dormant qNSCs.neural stem/progenitor cells (NPCs) acquire phenotypic and molecular hallmarks of quiescence upon direct cell-cell contact with primary brain microvascular endothelial cells (bmvEC). 8To further validate this model, we transcriptionally profiled NPCs alone or in co-culture by RNA sequencing (RNA-seq) (Table S1).0][11][12][13] Luciferase reporter assays confirmed that p53 activity increased in co-cultured NPCs compared with alone NPCs (Figure 2A).Furthermore, acute p53 recombination in primary p53 LoxP/LoxP NPCs (p53 À/À NPCs) resulted in a less pronounced cell-cycle arrest in co-culture relative to p53 +/+ NPCs controls (Figures 2B, S2D, and S2E), 14 confirming that the co-culture system reflects in vivo phenotypes and can inform p53 effectors.We therefore examined p53-controlled transcriptional programs by subjecting p53 +/+ and p53 À/À NPCs alone and in coculture to bulk RNA-seq (Table S1).Hierarchical clustering of the genes shared between co-cultured NPCs and qNSCs in vivo (Figure S2A), identified three groups of genes in p53 À/À NPCs (Figures 2C and 2D; Table S1): genes that did not significantly change (cluster 2), increased (cluster 1) or decreased (cluster 3) in expression upon p53 loss.Cluster 2 comprised type-B cell identity genes, including glial markers, adhesion, and Notch signaling, suggesting that p53 does not control type-B fate. 9,10onsistently, immunostaining confirmed that GFAP expression and neurosphere-like morphology were similar between p53 +/+ and p53 À/À NPC co-cultures (Figure S2E).Cluster 1 included cell-cycle genes, as expected 15 (Figure 2B).Cluster 3 was enriched in lipid-metabolism signatures, suggesting that p53 may control qNSCs metabolic state.
In response to DNA damage, p53 mediates cell-cycle arrest predominantly through transcriptional activation of p21/ Cdkn1a. 16,17We therefore assessed p21 function in our system, by comparing the response of wild-type (p21 +/+ ) and p21 knockout (p21 À/À ) NPCs with co-cultured NPCs.Surprisingly, we found no differences (Figure S2F).Consistent with this, p53 binding to the p21 promoter and p21 mRNA levels were similar in alone and co-cultured wild-type NPCs (Figures 3E and S2G), indicating that p53-induced quiescence is p21 independent.
9][20][21] As p53 À/À NPC co-cultures displayed changes in lipid metabolism signatures (Figure 2D), we examined the role of FAO downstream of p53.Many enzymes involved in FA metabolism were deregulated in the transcriptomes of p53 À/À relative to p53 +/+ NPCs (Figure 2E).To determine whether this was accompanied by functional changes, we first measured FA-driven oxygen consumption in p53 +/+ and p53 À/À NPCs cultured alone and with bmvECs.While p53 +/+ NPCs increased FAO in co-culture (Figures 2F and2H), p53 À/À NPCs did not (Figures 2G and 2H).Interestingly, despite FAO signatures being downregulated in both proliferating and quiescent p53 À/À NPCs, a change in metabolic state only occurred in co-culture, confirming the quiescence specificity of FAO. 18To determine whether p53-regulated FAO is required for energy production, we measured ATP and total adenine nucleotide pool levels in p53 +/+ and p53 À/À co-cultures before and after treatment with the FAO inhibitor etomoxir (ETO).Both were significantly decreased in ETO-treated wild-type co-cultures, whereas no differences were detected in the absence of p53 (Figures 2I and S2H).This confirms that FAO is used for energy production 18 and that p53 is a key regulator of this metabolic program.
In turn, PPARGC1a enhances PPARa activity to induce transcription of lipid-catabolic enzymes resulting in increased FAO and quiescence (Figure 4N).To test this model functionally, we asked whether increasing PPARa activity through the administration of exogenous ligands would compensate for the PPARGC1a decrease in p53 À/À NSCs and rescue quiescence. 24reatment with the PPARa agonist WY-14643 increased FAO genes expression and restored the cell-cycle arrest of p53 À/À co-cultured NPCs, without affecting stemness (Figures 3F and S3B-S3D).The WY-14643 rescue was dependent on PPARa and not caused by off-target effects because it was lost in Ppara knockout cells (Figures S3E, S4F, and S4G).To confirm that the effects of WY-14643 were mediated by FAO downstream of PPARa, we performed two parallel experiments.We measured the FA-driven oxygen consumption rate in quiescent p53 +/+ and p53 À/À NPCs untreated or treated with WY-14643 and found a complete rescue of FAO in p53 À/À NPCs upon WY-14643 treatment (Figures 3G-3I).Next, we exposed quiescent p53 +/+ and p53 À/À NPCs to 13 C-palmitate in the presence or absence of WY-14643 and traced the incorporation of radiolabeled carbons into tricarboxylic acid (TCA) cycle intermediates and amino acids derived from TCA intermediates using gas chromatography-mass spectrometry (GC-MS).We found that while 13 C incorporation into both was significantly decreased in untreated p53 À/À co-cultures, as expected, WY-14643 treatment restored it to the levels of p53 +/+ co-cultures (Figures 3J and S3F).
WY-14643 treatment of p53 +/+ NPCs monocultures also phenocopied the cell-cycle arrest of p53 +/+ NPC co-cultures while treatment of p53 +/+ NPC co-cultures resulted in a more pronounced arrest, which was paralleled by an increase in FAO in both metabolic assays, further confirming the link between fatty-acid metabolism and quiescence (Figures 3F and  3G).These experiments indicate that p53 mediates quiescence at least in part through regulation of FAO via a PPARGC1a/ PPARa axis.

Dietary fish-oil supplementation delays tumor initiation
Our results suggest that premature activation of qNSCs through metabolic remodeling may be a mechanism by which p53 mutations drive GBM.We therefore hypothesized that restoring PPARa-dependent FAO downstream of p53 loss may suppress tumorigenesis.PPARa is a nutrient sensor and can be activated by dietary polyunsaturated FAs (PUFAs), providing a potential strategy for tumor prevention through diet. 25To test this, we first examined in vitro effects of docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA) omega-3 PUFAs, the main components of fish oil.Treatment of p53 +/+ and p53 À/À NPC co-cultures with EPA and DHA increased expression levels of FAO genes in both genotypes, restoring them to wild-type levels in p53 À/À NPCs (Figure S3G).This was accompanied by a PPARa-dependent rescue of quiescence in p53 À/À NPCs (Figures 3F and  S3E) and further suppression of proliferation in p53 +/+ cells both alone and in co-culture as observed with WY-14643 (Figure 3F).Next, we tested the in vivo effects of dietary supplementation with fish oil.Fish oil, or PBS control, were administered to p53 wt or p53 icKO mice for a total of 3 weeks, and during the last 5 days tamoxifen was added prior to analysis of the SVZ 24 h later (Figure 4A).Strikingly, fish-oil supplementation fully reversed the effect of acute p53 loss in qNSCs p53 icKO , suppressing the increase in activated type-B and early type-C progenitors (Figures 4B and 4C).7][28] The model is driven by the combined inactivation of the tumor suppressor genes Nf1, Pten, and Trp53 (alongside tdTomato overexpression, hereafter NPP model) in endogenous postnatal SVZ NSCs via electroporation (Figure S4B).As in genetically engineered mouse models carrying the same mutations, 29,30 the NPP model gives rise to brain tumors with histological and molecular features of GBM. 27,28Importantly, the introduction of the mutations dramatically increased qNSC activation, relative to control mice (legend continued on next page) electroporated with a control tdTomato plasmid (Figures 4D and  S4A), indicating that aberrant activation of qNSCs is an early event in tumor initiation.Furthermore, acute treatment of tdTomato-electroporated brains with an 8-day course of ETO during the same time period (p4-p11), phenocopied effects of p53 deletion in qNSC leading to their increased activation to a proliferative state compared with controls (Figures 4E and S4A).
To determine whether p53 phenotypes could be reversed by diet in this tumor model, females were fed a high-fish-oil or normal control diet from conception through lactation.Tumorigenesis was then induced in the pups and brain tissue analyzed 3.5 weeks later, a time at which early lesions can be detected in control animals (Figures 4F and 4G).Remarkably, resulting lesions were significantly smaller and less proliferative in the high-fish-oil diet-fed group relative to controls, as judged by tdTomato + area and EdU incorporation, respectively (Figures 4G-4I).To determine whether fish-oil administration acts through FAO, we repeated the protocol described above in the presence or absence of a 14-day course of ETO (Figure 4J) and found that FAO inhibition completely abrogated fish-oil effects, leading to the development of larger and more proliferative lesions, similar to early NPP control tumors (Figures 4K and 4L).Furthermore, we assessed whether FAO induction downstream of p53 and fish oil depends on the identified PPARGC1a/PPARa axis.The NPP construct was modified to incorporate the Ppargc1a gene (Figure S4B), resulting in its overexpression in the tumor cells (Figures S4C and S4E) and, importantly, upregulation of p53-regulated FAO genes (Figure S4D).Ppargc1a-overexpressing NPP tumors induced in animals fed a normal diet were significantly smaller and less proliferative than controls, phenocopying fish-oil effects (Figures 4K and 4L).Furthermore, Ppargc1a tumor-suppressive effects were fully reversed by ETO administration, confirming that they were FAO-dependent (Figures 4K and 4L).We next deleted the Ppara gene by introduction of a CRISPR sgRNA to Ppara into the NPP construct (Figure S4B, S4F and S4G) and used this construct to initiate tumors in animals fed a control or high-fish-oil diet.
Remarkably, Ppara deletion abrogated the effects of fish oil, leading to the development of tumors of similar size as control NPP tumors fed a normal diet.In normal diet, NPP tumor development was unaffected by knockout of Ppara, as expected for p53-deficient tumors with compromised Ppara activity (Figures 4K and 4L).Thus, fish oil suppresses initiation through PPARa-mediated FAO.
Finally, to confirm that these effects resulted from dysregulation of NSC quiescence, we assessed the proportions of qNSCs and aNSCs in all experimental groups described above (Figure 4M).Tumor size was directly related to NSC activation state, with all groups in which p53/PPARGC1a/PPARa/FAO was disrupted displaying aberrant activation; those in which the pathway was rescued, namely fish-oil administration and Ppargc1a overexpression, retaining a much greater proportion of qNSCs in the SVZ.These experiments further underscore the functional link between FAO and quiescence.We conclude that diet counteracts tumorinitiating mutations to delay gliomagenesis.

DISCUSSION
][33] How these mutations remain silent despite their tumor-initiating potential, is a fundamental and still unresolved question in cancer biology.It has been proposed that a normal tissue microenvironment acts as a tumor-suppressive mechanism by dominantly keeping mutations in check. 31,34Our study identifies diet as a key contributing factor in mutation silencing.
Mutations in p53 that have been found to be shared between tumor-free SVZ and matching tumors of GBM patients 2 pointing to NSCs as GBM cells of origin and to p53 as a key tumor-initiating mutation.][20][21] Unlike in other systems where p53 directly controls FAO genes, 40,41 in qNSCs p53 induces FAO via a PPARGC1a/ PPARa axis.3][44][45] It is tempting to speculate that the p53/ PPARGC1a/PPARa may represent a general homeostatic pathway in normal cells, which is highjacked in cancer to promote tumor growth by rewiring the cell metabolic state.
These findings have important therapeutic implications as we showed that, in the context of the NPP model, early GBM development was dramatically suppressed by fish oil.This suggests that dietary intervention may be an effective therapeutic strategy for suppressing tumor initiation.As quiescent stem cells often share a common metabolic profile and can act as cancer cellof-origin across many tissues, dietary intervention may provide a more general approach for cancer prevention. 46mitations of the study It is possible that additional effectors besides increased FAO mediate quiescence downstream of p53 and future studies should explore this possibility.It would be equally important to examine the long-term impact of fish-oil supplementation on long-term survival and the relevance of our mouse findings to human disease.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Lead Contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Simona Parrinello (s.parrinello@ucl.ac.uk).

Materials Availability
This study did not generate new unique reagents.were obtained from The Jackson Laboratory (Jax 003257).p53 icKO were generated by crossing GLAST::CreERT2 mice to animals carrying a loxP-flanked Trp53 gene (p53 LoxP/LoxP ) or p53 +/+ and to a Rosa26::tdTom inducible reporter strain (Jax 007914). 48,49,60Animals were in a mixed 129xC57BL/6J background and both males and females were analysed between 7-10 weeks of age.For tumour initiation experiments, C57Bl6/J mice were purchased from Charles River (Jax 000664).Approximately an equal number of male and female mice were used per experiment.Mice were group-housed (where possible) in individually ventilated cages and maintained with 12-hour light/dark cycles with water and chow available ad libitum.

Data and Code
Neural progenitor cell culture NPCs were isolated from the SVZ of postnatal day 9-12 (P9-P12) mouse brains as previously described. 8Briefly, following microdissection, the SVZ was digested by incubation in HBSS (Invitrogen, 14170-088) supplemented with 0.25% trypsin and 60 U ml À1 Dnase I (Sigma, D4263) for 2 minutes at 37

METHOD DETAILS
Tamoxifen, EdU, Fish oil and Etomoxir administration Tamoxifen (Sigma, T5648) was administered to 7-9 weeks-old mice by intraperitoneal injection (i.p.) at 100mg/kg/d.For assessment of resting type-B cells, EdU (Santa Cruz, sc-284628) was administered in the drinking water (0.2 mg/mL) ad libitum for 7 days, followed by tamoxifen administration for five consecutive days and mice were sacrificed 3 days later.This protocol enables identification of resting qNSCs by labelling type-B cells that incorporated EdU during the 7-day pulse and then re-entered quiescence during the 8-day chase period, thereby appearing as EdU + /Ki67 -/tdTom + cells with radial morphology.To examine dormant type-B cells, EdU was administered in the drinking water ad libitum for 14 days.Mice were injected with tamoxifen over the last 5 days of EdU labelling and sacrificed 24h later.This enables identification of dormant qNSCs that activate during the 1-day chase period as radial EdU -/Ki67 + /tdTom + cells.For Omega-3 FA administration experiments, 6-weeks old mice were given daily 100mL of fish oil (Sigma, F8020) corresponding to 0.2-0.43g/kg/d docosahexaenoic acid (DHA) and 0.4-0.67 g/kg/d eicosapentaenoic acid (EPA) by oral gavage for 3 weeks.Tamoxifen was administered over the last 5 days of fish oil administration and SVZs collected 24h later.To prevent oxidation of the fish oil, aliquots were protected from direct light and supplemented with 40mM EDTA and 0.5mg/ml Ascorbyl Palmitate (Sigma, PHR1455).Control 6-weeks old mice received 100mL of PBS containing 40mM EDTA and 0.5mg/ml Ascorbyl Palmitate.For the tumour initiation experiments, we used a de novo somatic GBM model based on CRISPR/Cas9-mediated deletion of Nf1, Pten and Trp53 tumour suppressors (NPP), as described. 27Briefly, sgRNAs (single guide RNA) were expressed from a PiggyBac vector alongside a tdTomato reporter to fluorescently label resulting tumours.PiggyBase transposase was co-expressed with Cas9 in a second non-integrating plasmid.NPP-Pparpgc1a-OE and NPP-Ppara-KO constructs were generated using InFusion Kit (Clontech, 638917) and T4 DNA Ligase (NEB, M0202S), following manufacturer's instructions.To generate the NPP-Ppara-KO, sgRNA to target Ppara (5'-GCCGGGGGACTCGTCCGTGC-3'), as reported, 62 was cloned in the NPP plasmid 3' of the Nf1, Pten, Trp53 sgRNAs.Mouse Ppargc1a CDS was cloned after the tdTomato sequence as a polycistronic construct with a T2A linker for the generation of the NPP-Pparpgc1a-OE plasmid. 27,28C57BL/6J female mice were randomly placed on an ad libitum fish oil supplemented diet (Teklad Global 2020X diet supplemented with 30-32g fish oil/kg containing 13.5% EPA and 10.5% DHA) or control diet (Teklad Global 2020X diet) following the observation of a copulation plug.Pups from these females were injected at P2 with the two plasmids mixed at a ratio of 0.8:1 using a Femtojet microinjector (Eppendorf) directly into the lateral ventricle.Plasmids were electroporated into the sub-ventricular zone using the Gemini X2 Generator set for 5 square pulses, 50 msec/pulse at 100 volts, with 950 msec intervals 63 .Mice were maintained on fish oil supplemented diet or control diet until collection.After 3.5 weeks animals were i.p. injected with EdU (50mg/kg) 2 hours prior to sacrifice by transcardial perfusion of paraformaldehyde (PFA, 4%) under terminal anaesthesia.The brains were collected, stored overnight in PFA at 4 C prior to Vibratome sectioning (50mm) and subsequent analysis of tumour development.For the inhibition of FAO during tumour initiation in vivo, pups were treated with etomoxir (Sigma-E1905) starting 10 days after electroporation and continuing for 2 weeks sacrifice.This was achieved through lactation via etomoxir i.p. injections of dams at 10mg/kg every other day during the first week and through i.p. etomoxir injection of the pups at 5mg/ kg every other day during the second week.For the analysis of acute NSC activation in response to tumour-initiating mutations, pups at P2 were electroporated with the EF1a-tdTomato (control) or NPP (Nf1, Pten, Trp53) plasmids and SVZ wholemount preparations were analysed 9 days after electroporation (P11).To assess effects of FAO inhibition of NSC activation, EF1a-tdTomato pups received either PBS (vehicle) or etomoxir at P4 through lactation via i.p. injection to the dams every other day (10 mg/kg) and collected at P11.

Co-culture experiments and cell treatments
Co-culture experiments were performed as previously described and analysed 48h later. 8In brief, bmvEC were plated at a density of 2.6x10 4 cells/cm 2 .The following day, NPCs were plated either alone or onto endothelial monolayers at a density of 34x10 3 cells/cm 2 .After 48h, the NPCs were removed by selective trypsinization before analysis.Treatments were as follows: PPARa agonist WY-14643 was purchased from Sigma (C7081) and added to NPCs alone or in co-culture with bmvECs at the concentration of 200mM for 24h.BSA-conjugated Eicosapentaenoic Acid (EPA, 20:5n-3) (Cloud-Clone Corp., #CPO122Ge11) and BSA Conjugated Docosahexaenoic Acid (DHA, 22:6n-3) (Cloud-Clone Corp., #CPO632Ge11) were used at the concentration of 50mM each for 6h or 24h.For the EdU incorporation assay, 10mM EdU was added to the culture media for 2h prior to fixation and detection using the Click-iTÔ EdU Alexa FluorÔ 647 Flow Cytometry Assay Kit (Thermo Fisher Scientific, C10424), according to manufacturer's instructions.DNA was stained with DAPI.The cells were analysed on Fortessa X20s flow cytometer (BD).For p53 loss-of-function experiments, p53 loxP/loxP NPCs were infected with adenoviruses expressing a codon-improved Cre (iCre) under a CMV promoter (Ad-CMV-iCre, Vector Biolabs, #1045).Adenoviruses carrying CMV promoter only were used as controls (Ad-CMV-Null, Vector Biolabs, #1300).NPCs were used within 2-3 passages post infection to minimise compensatory effects of p53 loss.For luciferase assays, NPCs were nucleofected with a reporter plasmid containing the p53 binding element (PG13-CAT) prior to seeding in alone and bmvEC co-culture for 48h. 64,65Luminescence was normalized to viable cell numbers in each condition.For the analysis of adenine nucleotides, NPCs were treated with etomoxir at the final concentration of 100mM starting on day 2 of co-culture for 24h.

FAO assay
Fatty acid oxidation was measured using the commercially available Fatty Acid Oxidation (Abcam, ab217602) and Extracellular Oxygen Consumption Assay kits (Abcam, ab197243) according to the manufacturer's instructions.Treatment with 25mM FCCP served as positive and samples without cells as negative controls, respectively.Extracellular O 2 consumption was measured every 90 seconds for 2h at Ex/Em=380/650 nm using the VarioSkan LUX Plate reader set at 37 C. (ThermoFisher Scientific).Oxygen Consumption Rate (OCR) was determined by calculating the slope of the linear proportion of the signal profiles.

Sample extraction for LC-HRMS analysis
Cell samples were extracted with 500 mL of extraction solvent (water/methanol, 20:80 v/v) and they were spiked with a known concentration of AMP 13C1015N5, ADP 15N5 and ATP 13C10 for their calibration curves (standard addition).Cell extracts were sonicated in an ultrasonic water bath (15 min) and centrifuged (13,000 g, 10 min, 5 C).Pooled quality controls (QCs) were created by pooling equal aliquots of each study sample, in order to assess technical reproducibility across the batch.

Nucleotides analysis by LC-HRMS
Liquid chromatographic analyses were performed on a Vanquish Flex Binary UHPLC system (Thermo Scientific Inc., MA, USA) coupled to a benchtop hybrid quadrupole-Orbitrap Q Exactive mass spectrometer (Thermo Scientific Inc., Bremen, Germany).Chromatographic separation of extracts was achieved using a SeQuantÒ ZICÒ-cHILIC column (3mm,100A ˚100 x 2.1 mm) held at a temperature of 45 C and a flow rate of 0.30 mL/min.Mobile phase was consisted of 90% acetonitrile with 10 mM ammonium acetate pH = 4.6 (solvent A) and 10 mM ammonium acetate pH = 4.6 (solvent B).The gradient elution started with 5% of solvent B increasing to 45% of B over 14 min.This condition was kept constant until 15.5 min followed by re-equilibration at the initial conditions, yielding a total run time of 20 min.Ionization was performed in the negative mode using a heated electrospray ionisation source, with the following parameters: spray voltage 3.0 KV, heater temperature 400 C, capillary temperature 320 C, S-lens RF level 50, sheath and auxiliary gas flow rate, 48 and 11 units, respectively.The mass accuracy was calibrated before sample analysis.Mass spectrometric data were acquired at high-resolution (70,000 at m/z 200) in profile mode using a Full MS scan method (m/z 70 to 700).Automatic gain control (AGC) was set to 1e6 and maximum injection time 250 ms.Xcalibur version 4.1 was used for data acquisition and processing.Peak areas of the Extracted Ion Chromatograms (EIC) were used in the data processing.

13
C-palmitate tracer experiments and GC-MS analysis A 100mM stock of [U- 13 C]-palmitic acid (Sigma, 605573) was prepared in ethanol.This was diluted 100-fold into DMEM/F12 media containing 2% w/v fatty acid-free BSA (Sigma A8806) and sonicated for 15 minutes in a water bath sonicator at room temperature to form soluble BSA/palmitate complexes.The resulting media contained 1mM [U- 13 C]-palmitic acid, 2% BSA and 1% ethanol (170 mM) with a palmitate:BSA molar ratio of 3.3:1.Following filtration through a 0.2 mm PES filter (Millipore, Burlington, MA) this was diluted 1:10 in SVZ culture medium, giving a final concentration of 100mM [U- 13 C]-palmitic acid.
Cells in co-culture, in the presence or absence of PPARa agonist WY-14643 (200mM), were incubated with 100mM U-13 C-palmitate SVZ medium for 6 hr or 24 hr.Cell were selectively dissociated from the endothelial monolayer and rapidly quenched on dry ice/ ethanol.Samples were then pelleted at 0 C, washed with ice-cold PBS pH 7.4, transferred to 2 mL tubes and metabolites extracted at 4 C for 1 hr (with 3 x 8 min sonications in a water bath sonicator) using 600 mL chloroform/methanol (2:1, v/v, containing 13 C lauric acid internal standard at 5nmol per sample).Extracts were transferred to 1.5 mL tubes and dried in a SpeedVac.The pellet was then re-extracted with 450 mL methanol/water (2:1, v/v, containing 1 nmol scyllo-Inositol internal standard; 4 C, 8 min sonication), the extract was combined with the first extract and then re-dried.Polar and apolar metabolites were separated by phase partitioning with chloroform/methanol/water (1:3:3, v/v/v) and then analyzed by GC-MS.
GC-MS data acquisition was performed largely as previously described, 67 using an Agilent 7890B-5977A GC-MSD in EI mode after derivatization of twice methanol-washed dried extracts by the addition (a) for polar metabolites of 20 mL of a 20 mg/mL solution of methoxyamine hydrochloride in pyridine (both Sigma) at room temperature for >16 hr and 20 mL BSTFA + 1% TMCS (Sigma) at RT for >1 hr, or (b) for fatty acids of 25 mL chloroform/methanol (2:1, v/v) and 5 mL MethPrepII (Grace Alltech) at room temperature with no incubation.GC-MS parameters were as follows: carrier gas, helium; flow rate, 0.9 mL/min; column, DB-5MS (Agilent); for polar analyses: inlet, 270 C; temperature gradient, 70 C (2 min), ramp to 295 C (12.5 C/min), ramp to 320 C (25 C/min, 3 min hold); for apolar analyses: inlet, 250 C; temperature gradient, 70 C (1 min), ramp to 230 C (15 C/min, 2 min hold), ramp to 325 C (25 C/min, 3 min hold).Scan range was m/z 50-550 (polar) and 50-565 (apolar).Data were acquired using MassHunter software (version B.07.02.1938).Data analysis was performed using MANIC software, an in house-developed adaptation of the GAVIN package. 68Metabolites were identified and quantified by comparison to authentic standards, and label incorporation estimated as the percentage of the metabolite pool containing one or more 13 C atoms after correction for natural abundance.
Quantitative RT-PCR For in vitro experiments, RNA was extracted using RNeasy mini kit (Qiagen, 74104) following the manufacturer's instructions.RNA was reverse transcribed using iScript gDNA clear cDNA synthesis kit (Bio-rad, 1725034) and quantitative PCR was performed using the qPCRBIO SyGreen Mix Lo-Rox (PCR Biosystems, PB20.11).For assessment of acutely FACS-purified qNSCs and aNSCs, RNA was extracted using RNeasy Plus MicroKit (Qiagen, 74034) according to the manufacturer's instructions and cDNA libraries were prepared using the Smart-seq2 protocol. 69Primers used are listed in Table S2.

RNA-Sequencing
For RNA sequencing, RNA was isolated using RNeasy mini kit (Qiagen, 74104) according to manufacturer's instructions.Libraries were prepared using the Truseq mRNA stranded kit and quality control checks were performed by Qubit and Bioanalyser analysis.Libraries were then pooled and run on a MiSeq Nano Flow Cell (V2 reagents) (Single Read 26 cycles) to check the balance of the libraries within the pool and adjusted where necessary.Clonal clusters of each library were then amplified onto an Illumina Flow Cell using the Illumina cBot system and sequenced on a HiSeq 2500 (v4 chemistry) as a Paired-End 100bp.

Figure 3 .
Figure 3. p53 mediates quiescence through PPARa (A) RNA-seq log 2 expression ratios of Ppara and Ppard targets in indicated cultures.(B) RNA-seq log 2 expression ratios of FAO regulators in indicated cultures.
Availability d The scRNA-seq and CUT&RUN data have been deposited at GEO. Accession numbers are listed in the key resources table.
d This paper does not report original code.dAnyadditionalinformation required to reanalyse the data reported in this paper is available from the lead contact upon request.EXPERIMENTAL MODEL AND SUBJECT DETAILSMiceAll procedures were performed in compliance with the Animal Scientific Procedures Act, 1986 and approved by the UCL Animal Welfare and Ethical Review Body (AWERB) in accordance with the International guidelines of the Home Office (UK).GFAP::GFP mice C. Single cell suspensions were plated onto poly-L-lysine (PLL)-coated plates in SVZ explant medium consisting of DMEM/F12 (Invitrogen, 11320074), 3% FBS (Invitrogen), 20 ng ml À1 EGF (Peprotech, 315-09-1000) for 48 h.NPCs were routinely grown, for up to 8-10 passages, in SVZ culture medium consisting of DMEM/F12 (Life technologies, 11320074), 0.25% FBS, N2 (Life technologies, 17502001), 20 ng ml À1 EGF, 10 ng ml À1 bFGF (Peprotech, 450-33A) and 35 mg ml À1 bovine pituitary extract.For the Cdkn1a loss-of-function experiments, NPCs were isolated from the SVZs of p21 +/+ or p21 -/- mice, a kind gift of Dr Owen J. Sansom 61 (Cancer Research UK Beatson Institute, Glasgow, UK).