Fatty acid oxidation organizes mitochondrial supercomplexes to sustain astrocytic ROS and cognition

Na u e Me abolism | Volume 5 | Augus 2023 | 1290–1302 1290 na u e me abolism Le e h ps://doi.o g/10.1038/s42255-023-00835-6 Fa y acid oxida ion o ganizes mi ochond ial supe complexes o sus ain as ocy ic ROS and cogni ion B enda Mo an -Fe ando   1,2,5, Daniel Jimenez-Blasco   1,2,3,5, Paula Alonso-Ba an   1,2, Jesús Agulla   1,2, Rebeca Lap esa   1,2, Da io Ga cia-Rod iguez   1,2, Sa a Yun a-Sanchez   1,2, I ene Lopez-Fabuel   1,2, Emilio Fe nandez1,2,3, Pe e Ca melie    4, Angeles Almeida   1,2 , Ma ina Ga cia-Macia   1,2,3 & Juan P. Bolaños   1,2,3 Ha ing di ec access o b ain ascula u e, as ocy es can ake up a ailable blood nu ien s and me abolize hem o ul il hei own ene gy needs and deli e me abolic in e media es o local synapses1,2. These glial cells should be, he e o e, me abolically adap able o swap di e en subs a es. Howe e , in i o and in i o s udies consis en ly show ha as ocy es a e p ima ily glycoly ic3–7, sugges ing glucose is hei main me abolic p ecu so . No ably, ansc ip omic da a8,9 and in i o10 s udies e eal ha mouse as ocy es a e capable o mi ochond ially oxidizing a y acids and ha hey can de oxi y excess neu onal-de i ed a y acids in disease models11,12. S ill, he ac ual me abolic ad an age o a y acid use by as ocy es and i s physiological impac on highe -o de ce eb al unc ions emain unknown. He e, we show ha knockou o ca ni ine-palmi oyl ans e ase-1A (CPT1A)—a key enzyme o mi ochond ial a y acid oxida ion—in adul mouse as ocy es causes cogni i e impai men . Mechanis ically, dec eased a y acid oxida ion ewi ed as ocy ic py u a e me abolism o acili a e elec on lux h ough a supe -assembled mi ochond ial espi a o y chain, esul ing in a enua ion o eac i e oxygen species o ma ion. Thus, as ocy es na u ally me abolize a y acids o p ese e he mi ochond ial espi a o y chain in an ene ge ically ine icien disassembled con o ma ion ha secu es signalling eac i e oxygen species and sus ains cogni i e pe o mance. To asce ain he exp ession le els o genes coding o a y acids use in as ocy es and neu ons, we pe o med quan i a i e PCR wi h e e se ansc ip ion (RT–qPCR) analyses, which e ealed inc eased messenge RNA abundances in ca ni ine-palmi oyl ans e ase-1A (Cp 1a)— espon- sible o long-chain acyl-CoA en y in o mi ochond ia13—and dec eased ace yl-CoA ca boxylase-1 (Acc1)— esponsible o he biosyn hesis o CPT1A − inhibi o malonyl-CoA 14 -mRNA abundances in mouse p ima y as ocy es when compa ed wi h neu ons (Supplemen a y Fig. 1a). In addi ion, he mRNA abundances o mi ochond ial Acc2 iso o m, which is ound highly en iched in oxida i e issues such as skele al Recei ed: 29 No embe 2022 Accep ed: 2 June 2023 Published online: 17 July 2023 Check o upda es 1Ins i u e o Func ional Biology and Genomics (IBFG), Uni e si y o Salamanca, CSIC, Salamanca, Spain. 2Ins i u e o Biomedical Resea ch o Salamanca (IBSAL), Uni e si y Hospi al o Salamanca, Salamanca, Spain. 3Cen e o Biomedical In es iga ions Ne wo k on F ail y and Ageing (CIBERFES), Mad id, Spain. 4Labo a o y o Angiogenesis and Vascula Me abolism, Vesalius Resea ch Cen e , Leu en, Belgium. 5These au ho s con ibu ed equally: B enda Mo an -Fe ando, Daniel Jimenez-Blasco.  e-mail: [email p o ec ed]; ma inaga [email protected]; [email p o ec ed] Na u e Me abolism | Volume 5 | Augus 2023 | 1290–1302 1291 Le e h ps://doi.o g/10.1038/s42255-023-00835-6 i us (AAV) pa icles exp essing C e ecombinase go e ned by he as ocy ic-speci ic glial- ib illa y acidic p o ein (GFAP) sho -p omo e (PHP.eB-AAV-g aABC 1 D-C e-GFP) (Fig. 1a). This ea men was e icien , as judged by he wide exp ession o g een luo escen p o ein (GFP) ac oss he b ain (Supplemen a y Fig. 1c). Con ols (wild- ype, WT) we e Cp 1alox/lox mice ha ecei ed equi alen doses o he same i us pa icles, excep ha hey lacked C e ecombinase, and all mice we e analysed a e 1–9 mon hs (Fig. 1a). As shown in Fig. 1b (Supplemen a y Fig. 1d), PHP.eB-AAV-g aABC 1 D-C e-GFP ea men caused a signi ican educ ion in b ain CPT1A p o ein abundance. Gi en ha he b ain con ains o he cell ypes besides as ocy es, we also analysed CPT1A abundance in ex i o as ocy es immunomagne ically isola ed om he b ain o he adul CPT1A KO mice (Fig. 1a), which e ealed CPT1A abolishmen speci ically in he as ocy e-posi i e (ACSA+) ac ion, bu no in he as ocy e-nega i e (ACSA − ) ac ion (Fig. 1c). To asce - ain he unc ional e icacy o CPT1A KO, ex i o eshly isola ed b ain slices om adul mice we e incuba ed wi h [U-14C]palmi ic acid o assess he a e o 14CO2 p oduc ion as an index o a y acid oxida ion lux. As shown in Fig. 1d, oxida ion lux in he b ain was signi ican ly educed by oughly 75% in CPT1A KO when compa ed wi h WT mice. To explain whe he loss o as ocy ic CPT1A al e ed o he pa hways o b ain me abolism, we pe o med un a ge ed me abolomics in b ain samples. As depic ed in he olcano plo (Supplemen a y Fig. 1e) and in he hea map (Supplemen a y Fig. 1 ), we ound 17 me aboli es signi i- can ly dec eased and 43 me aboli es signi ican ly inc eased in he b ain o he as ocy e-speci ic CPT1A KO mice. The esul s e ealed inc eased abundance in long-chain a y acids and long-chain acyl-ca ni ine de i a i es, and dec eased abundance in sho -chain a y acids muscle and hea 15, o long-chain ace yl-CoA dehyd ogenase (Acadl), which ca alyses he ini ial s ep o mi ochond ial a y acid oxida ion, o he endoplasmic e iculum-loca ed Cp 1c and o mi ochond ial i unc ional p o ein α (M pα) ha is unc ionally esponsible o elec on ans e om mi ochond ial long-chain a y acids o mi o- chond ial espi a o y complexes I (CI) and III (CIII)16, we e ound o be highe in as ocy es e sus neu ons (Supplemen a y Fig. 1a). Al hough hese a e ela i e alues, hey a e in cohe ence wi h p e ious obse a- ions8–10 sugges ing ha as ocy es a e be e equipped han neu ons o mi ochond ially oxidize long-chain a y acids. To unc ionally sus ain his s a emen , he oxygen consump ion a e (OCR) was analysed in as ocy es and neu ons using he Seaho se echnology in glucose-based medium in ei he he absence o p esence o e omoxi : a po en and i e e sible inhibi o o CPT1 ( e . 17). As shown in Supplemen a y Fig. 1b, basal mi ochond ial espi a ion was ound o be oughly 1.7- old highe in neu ons when compa ed wi h as ocy es, con i ming p e ious indings18. No ably, he p opo ion o mi ochond ial OCR inhibi ion by e omoxi was oughly 20% in neu ons and oughly 35% in as ocy es (Supplemen a y Fig. 1b), indica ing ha a y acids a e p e e ed mi o- chond ial espi a o y subs a es o as ocy es han neu ons. Roughly 62% o as ocy ic ATP-linked mi ochond ial espi a ion was sus ained by a y acids (Supplemen a y Fig. 1b). To in es iga e he me abolic ad an age o mi ochond ial a y acid oxida ion in as ocy es in i o, o he wise o e coming he po en- ial d awbacks o pha macological inhibi o s, we gene ically engi- nee ed an as ocy e-speci ic Cp 1a knockou (KO) mouse model. To do so, 2-mon h-old Cp 1alox/lox mice19 we e in a enously injec ed, ia he e o-o bi al sinus 20 , wi h PHP.eB se o ype adeno-associa ed MW (kDa) 55 88 CPT1A β-Tubulin b CPT1A KO WT B ain Exons 11–12 loxP loxP Cp 1alox/lox mouse Cp 1a gene PHP.eB-AAV-g aABC1D-C e-GFP PHP.eB-AAV-g aABC1D-GFP WT mouse As ocy e-speci ic CPT1A KO mouse a WT as ocy es CPT1A KO as ocy es In i o Ex i o 1–9 mon hs c d e 0 0.6 1.2 1.8 0.0199 0 0.5 1.0 1.5 0.0009 Linolena e (18:3) (a.u.) 0 0.5 1.0 1.5 0.0033 0 0.6 1.2 1.8 0.0307 0 0.7 1.4 2.1 0.0648 0 0.5 1.0 1.5 0.0282 0 0.6 1.2 1.8 0.0089 0 0.4 0.8 1.2 0.1028 Py u a e (a.u.) Palmi olea e (16:1) (a.u.) Palmi oleylca ni ine (16:1) (a.u.) Cap oa e (6:0) (a.u.) Cap yla e (8:0) (a.u.) Palmi oylca ni ine (16:0) (a.u.) A achidonoylca ni ine (20:4) (a.u.) WT CPT1A KO 0.0184 [U-14C]Palmi ic acid oxida ion (a.u.) WT CPT1A KO 14CO2 β-Oxida ion +TCA [U-14C]Palmi ic acid B ain slices 0 0.5 1.0 1.5 0 0.6 1.2 1.8 0.0009 Hexadecadienoa e (16:2) (a.u.) MW (kDa) 55 88 CPT1A GFAP CPT1A KO WT CPT1A KO WT Immunomagne ically sepa a ed ex i o cells ACSA+ (As ocy es) ACSA– (As ocy es–) Fig. 1 | In i o as ocy e-speci ic Cp 1a KO inhibi s a y acid oxida ion and al e s he me abolomics pa e n in he b ain. a, S a egy used o gene a e as ocy e-speci ic Cp 1a KO mice and o immunomagne ically pu i y CPT1A KO as ocy es om adul b ain. C ea ed wi h BioRende .com. b, Wes e n blo agains CPT1A p o ein in as ocy e-speci ic Cp 1a KO b ain. β-Tubulin was used as a loading con ol; n = 2 mice pe condi ion (Supplemen a y Fig. 1d). c, Wes e n blo ing agains CPT1A p o ein in ACSA+ (as ocy es) and ACSA− (no as ocy es) cells, immunomagne ically isola ed om as ocy e-speci ic Cp 1a KO mouse b ain; n = 2 mice pe condi ion. GFAP was used as as ocy e en ichmen and loading con ols. d, Ra e o 14CO2 p oduc ion om [U-14C]palmi ic acid in b ain slices o WT and as ocy e-speci ic Cp 1a KO mice. Da a a e mean ± s.e.m. P alue is indica ed (n = 3 biologically independen samples; unpai ed S uden ’s - es , wo-sided). e, Concen a ions o a selec ion o me aboli es al e ed in he me abolomics s udy o he b ain samples om as ocy e-speci ic Cp 1a KO when compa ed wi h WT mice. Da a a e mean ± s.e.m. P alues a e indica ed (n = 6 mice pe condi ion; unpai ed S uden ’s - es , wo-sided). a.u., a bi a y uni s. Na u e Me abolism | Volume 5 | Augus 2023 | 1290–1302 1292 Le e h ps://doi.o g/10.1038/s42255-023-00835-6 (Fig. 1e), sugges ing dec eased long-chain and inc eased sho -chain a y acid use. No ably, py u a e concen a ion was signi ican ly dec eased by oughly 26% in he b ain o as ocy e-speci ic CPT1A KO mice (Fig. 1e), sugges ing an al e a ion in he me abolism o his glycoly ic-end p oduc in e media e. We nex aimed o u he cha ac e ize me abolically CPT1A KO as ocy es. To do so, as ocy es in p ima y cul u e om Cp 1alox/lox mice we e ansduced wi h adeno i uses exp essing C e ecombinase unde he po en cy omegalo i us (CMV) p omo e (AdV-CMV-C e-GFP) (Fig. 2a). Cp 1alox/lox as ocy es ansduced wi h he AdV lacking C e ecombinase (AdV-CMV-GFP) we e used as con ols (WT). RT–qPCR analysis e ealed Cp 1a, no Cp 1b o Cp 1c, mRNAs dec ease in AdV-CMV-C e-GFP ansduced as ocy es (Supplemen a y Fig. 2a). As shown in Fig. 2b (Supplemen a y Fig. 2a), CPT1A p o ein was e i- cien ly knocked ou in AdV-CMV-C e-GFP ansduced as ocy es when compa ed wi h WT cells. CPT1B, CPT1C and CPT2 p o ein abundances we e una ec ed (Supplemen a y Fig. 2c). The elease o 3-hyd oxybu y a e om AdV-CMV-C e-GFP ansduced as ocy es was signi ican ly educed, sugges ing impai ed ke ogenesis in CPT1A KO as ocy es (Supplemen a y Fig. 2d). We u he assessed he a e o [1- 14 C]palmi ic acid con e sion o 14 CO 2 and o 14 C-ke ones, which we e educed by oughly 50% in CPT1A KO when compa ed wi h WT as ocy es (Fig. 2c,d and Supplemen a y Fig. 2e). E omoxi , which inhibi s bo h CPT1 (mi ochond ial) and ca ni ine oc anoyl ans e ase (pe oxisomal)-media ed a y acid up ake21, i ually abolished [U-14C] palmi ic acid oxida ion (Supplemen a y Fig. 2 ). Thus, he coope a ion o a cellula compa men , p obably he pe oxisome 22 , no depend- ing on CPT1A, by con e ing long-chain in o sho -chain a y acids is likely o con ibu e o he mi ochond ial oxida ion o a y acids. Since as ocy e ene gy me abolism is hough o be la gely sus- ained by glycolysis3–5, we nex aimed o assess he impac o a y acid oxida ion on glucose me abolism. To do his, we i s analysed he a e o [6- 14 C]glucose oxida ion o 14 CO 2 , a p ocess ha akes place in he ica boxylic acid (TCA) cycle. As shown in Fig. 2e, [6- 14 C]glucose deca - boxyla ion inc eased in CPT1A KO as ocy es. 14C6-Glucose ia glycolysis labels 14 C 3 -py u a e, which deca boxyla es exclusi ely a he TCA cycle depending on he a es o glycolysis, mi ochond ial py u a e impo and py u a e dehyd ogenase (PDH) ac i i y. To iden i y which o hese s eps accoun s o he inc eased a e o [6-14C]glucose deca boxy la ion, we assessed [1- 14 C]py u a e deca boxyla ion, which exclusi ely akes place a PDH, a e mi ochond ial py u a e impo . We ound a signi ican inc ease in 14 CO 2 o ma ion om [1- 14 C]py u a e in CPT1 KO as ocy es (Fig. 2 ), indica ing enhanced PDH deca boxyla ion a e. This was ali- da ed by he assessmen o 293 Se phospho yla ion s a us o he PDH A1 (PDHA1) subuni , which was educed (Supplemen a y Fig. 2g) indica ing PDH ac i a ion 23 , an end ha was con i med by he PDH-speci ic ac i i y (Supplemen a y Fig. 2h). This esul , which explains he obse ed educ- ion in py u a e concen a ion in he me abolomics analysis (Fig. 1e), indica es ha , on Cp 1a loss, as ocy es unde go a me abolic ewi ing consis ing in enhanced py u a e deca boxyla ion o ace yl-coenzyme A. Lac a e is he main me abolic a e o glycoly ically de i ed py u- a e in as ocy es3–5. We he e o e assessed whe he he enhanced mi ochond ial py u a e deca boxyla ion al e ed as ocy ic- eleased lac a e. As shown in Fig. 2g, lac a e o ma ion was educed by a p opo - ion ( oughly 118 nmol h−1 mg p o ein−1) consis en wi h an inc eased py u a e deca boxyla ion ( oughly 65 nmol h −1 mg p o ein −1 ) (Fig. 2 ) in CPT1A KO as ocy es, an e ec ha could no be accoun ed o by changes in he lux o glycolysis, as speci ically measu ed by he a e o [3- 3 H]glucose con e sion in o 3 H 2 O (Fig. 2h). These esul s indica e ha , on Cp 1a KO, as ocy es ewi e he me abolic a e o py u a e o inc ease i s mi ochond ial oxida ion wi hou a ec ing glycolysis. Glycolysis and β-oxida ion hus appea o be independen ly egula ed pa hways aimed o sus ain di e en ace s o as ocy e me abolism. Gi en he con ibu ion o a y acid oxida ion in sus aining as o- cy ic mi ochond ial espi a ion (Supplemen a y Fig. 1b), we nex analysed he OCR in CPT1A KO as ocy es. As shown in Fig. 2i, loss o CPT1A inc eased by oughly 1.5- old he mi ochond ial basal and ATP-linked espi a ion, wi hou s a is ically signi ican ly a ec ing maxi- mal and non-mi ochond ial espi a ion. These esul s appa en ly con- as wi h hose showing dec eased basal mi ochond ial espi a ion by e omoxi (Supplemen a y Fig. 1b). Howe e , e omoxi acu ely inhibi s CPT1, whe eas he gene ic de iciency o Cp 1a allows as ocy es o adap o CPT1A loss. Thus, lack o CPT1A seems o ep og amme as ocy ic me abolism o a con o ma ion whe eby he mi ochond ial espi a ion is imp o ed. Such an imp o emen is no he consequence o inc eased mi ochond ial mass, acco ding o p o ein abundance pa ame e s (Supplemen a y Fig. 2i). Gi en ha elec on lux h ough mi ochon- d ial CI is he mos ene ge ically e icien cou se o conse e mi ochon- d ial ene gy, we analysed he p opo ion o mi ochond ial espi a ion ha is con ibu ed by his complex. To do his, we inhibi ed mi o- chond ial espi a ion wi h CI-speci ic inhibi o o enone, bo h in in ac (Supplemen a y Fig. 2j) and in digi onin-pe meabilized cells in he p esence o CI-subs a es and ADP (Supplemen a y Fig. 2k), which e ealed ha he con ibu ion o CI o sus ain mi ochond ial espi- a ion was signi ican ly enhanced in CPT1A KO e sus WT as ocy es. Al oge he , hese da a indica e ha mi ochond ial oxida ion o endo- genous a y acids in as ocy es p ese es he mi ochond ial espi a- o y chain con o ma ion in an ene ge ically less ac i e mode. To unde s and he molecula mechanism whe eby a y acid oxida ion keeps less ac i e mi ochond ial espi a ion in as ocy es, we sough o in es iga e he supe -assembly o he espi a o y complexes, hough o egula e mi ochond ial espi a ion24–26. To do so, mi ochon- d ia we e isola ed om CPT1A KO and WT as ocy es and hei p o eins subjec ed o blue na i e gel elec opho esis (BNGE) ollowed by wes - e n blo ing agains CI, CIII and CIV subuni s. The analysis e ealed ha loss o CPT1A p omo ed a signi ican inc ease in mi ochond ial supe complexes (SC) o ma ion (Fig. 3a,b and Supplemen a y Fig. 3a), p obably explaining he obse ed inc ease in CI-sus ained espi a ion (Supplemen a y Fig. 2j,k). The inc eased SC o ma ion and CI-sus ained espi a ion could no be asc ibed o a pu a i e enhancemen in he p o ein abundances o he mi ochond ial espi a o y chain complexes (Supplemen a y Fig. 3b) no in an enhancemen in CI-speci ic ac i i y (Supplemen a y Fig. 3c). The CI–III- and CIV-speci ic ac i i ies sig- ni ican ly inc eased in CPT1A KO as ocy es (Supplemen a y Fig. 3c), in cohe ence wi h enhanced mi ochond ial espi a ion (Fig. 2i) and SC o ma ion (Fig. 3a,b and Supplemen a y Fig. 3a). E en hough i is he subjec ma e o deba e 25,26 , besides he egula ion o espi a- ion, CI supe -assembly in SC has been sugges ed o egula e eac i e oxygen species (ROS) p oduc ion in hea 27 and b ain cells including neu ons and as ocy es18. Acco dingly, we nex in es iga ed whe he he obse ed supe -assembly o he mi ochond ial espi a o y chain in CPT1A KO as ocy es had an impac on ROS abundance. As shown in Fig. 3c, hyd ogen pe oxide (H2O2) was signi ican ly lowe in CPT1A KO as ocy es. To es ablish a causal link be ween inc eased SC o ma ion and dec eased H 2 O 2 gene a ion in CPT1A KO as ocy es, we knocked down CI subuni NDUFS1 (Supplemen a y Fig. 3d), a s a egy p e iously used o educe CI le els18. This ea men caused disas- sembly o CI-con aining SC in as ocy es (Supplemen a y Fig. 3e), p e en ed he inc ease in mi ochond ial basal espi a ion (Fig. 3e) wi hou s a is ically signi ican ly a ec ing maximal, ATP-linked and non-mi ochond ial espi a ion (Supplemen a y Fig. 3 ), and inc eased H2O2 in he CPT1A KO as ocy es (Fig. 3 ). Whe eas highe - esolu ion s uc u al wo k would be equi ed o con i m his mechanism25, al o- ge he , hese da a indica e ha a y acid oxida ion keeps he as ocy ic mi ochond ial espi a o y chain unde less ene ge ically a ou able s uc u al con o ma ion ha is able o sus ain ROS gene a ion. As ocy e ROS cons i u e edox signals ha modula e b ain me abolism o sus ain mouse beha iou 28. In addi ion, as ocy es a e able o ans o m a y acids in o ke one bodies10,29,30 (Fig. 2d and Supplemen a y Fig. 2d) ha , acco ding o s udies pe o med in Na u e Me abolism | Volume 5 | Augus 2023 | 1290–1302 1293 Le e h ps://doi.o g/10.1038/s42255-023-00835-6 D osophila, may be shu led o neu ons o oxida i e use 31,32 . Mo e- o e , he e we show ha , as long as as ocy es oxidize a y acids, py u- a e is con e ed o lac a e, a me aboli e ha also may be shu led o neu ons33,34. Gi en ha knocking ou Cp 1a in as ocy es impai ed ROS gene a ion, which physiologically main ains neu onal in eg i y and mouse cogni ion 28 , we nex in es iga ed i s impac on neu onal unc ion and beha iou . Neu ons cocul u ed wi h CPT1A KO as o - cy es (Fig. 3g) unde wen loss in an ioxidan glu a hione (Fig. 3h), in consonance wi h p e ious obse a ions 28 , sugges ing neu onal edox s ess. In ac , hese neu ons showed inc eased mi ochond ial ROS (Fig. 3i), mi ochond ial memb ane po en ial (∆ψ m ) dis up ion (Fig. 3j) and apop o ic dea h (Fig. 3k). Neu ons exp essing a mi ochond ial iso o m o he an ioxidan enzyme ca alase (mi ochond ial ca alase (mCAT) neu ons), which e icien ly p e en ed he inc eased mi o- chond ial ROS caused by coincuba ion wi h CPT1A as ocy es (Fig. 3i), abolished ∆ψm loss and apop osis (Fig. 3j,k). These bioene ge ic al e a- ions caused neu onal dys unc ion, as judged by he educed mRNA abundances o he neu onal unc ional ma ke s35,36 c-Fos and A c in a mi ochond ial ROS-dependen manne (Fig. 3l). To asce ain he in i o impac o ou indings, we immunomagne ically isola ed as o- cy es om bo h WT and as ocy e-speci ic Cp 1a KO mice (Fig. 4a). Cha ac e iza ion o neu al cell ma ke s con i med he pu i y o he ACSA + (as ocy es) ac ion in bo h geno ypes (Supplemen a y Fig. 4a). Analysis o he mi ochond ial espi a o y chain SC in he mi ochond ial ac ions o hese cells e ealed ha loss o CPT1A inc eased SC o ma ion, an e ec ha was obse ed bo h in male and emale mice (Supplemen a y Fig. 4b). Immunomagne ic isola ion o neu ons om as ocy e-speci ic Cp 1a KO mice (Fig. 4a), which esul ed in an en iched ac ion (Neu on + ) acco ding o he neu al cell ma ke s (Supplemen a y Fig. 4a), ollowed by SC analysis o hei mi ochon- d ial ac ions, e ealed CI disassembly om SC bo h in male and emale mice (Supplemen a y Fig. 4c). These esul s indica e ha loss o CPT1A in as ocy es causes mi ochond ial dys unc ion in neighbou ing neu ons. In good ag eemen wi h he da a ob ained in p ima y cul- u ed as ocy es, as ocy es isola ed om as ocy e-speci ic Cp 1a KO adul mice showed inc eased basal and ATP-linked espi a ion (Fig. 4b and Supplemen a y Fig. 4d), dec eased H2O2 and mi ochond ial ROS (Fig. 4c) wi h unchanged ∆ψm (Supplemen a y Fig. 4e) when g Ra e o lac a e elease (µmol h–1 mg p o ein–1) e c 0 1 2 3 4 0.0105 [1-14C]Palmi ic acid con e sion o 14CO2 (nmol h–1 mg p o ein–1) WT CPT1A KO h [3-3H]Glucose con e sion o 3H2O (µmol h–1 mg p o ein.1) i MW (kDa) 42 88 CPT1A β-ACTIN b CPT1A KO WT P ima y as ocy es 0 125 250 375 500 WT CPT1A KO 0 50 100 150 0.0343 Basal espi a ion OCR (pmol min–1 × 10–4 nuclei) OCR (pmol min–1 × 10–4 nuclei) OCR (pmol min–1 × 10–4 nuclei) OCR (pmol min–1 × 10–4 nuclei) OCR (pmol min–1 × 10–4 nuclei) 0 200 400 600 0.7274 Maximal espi a o y capaci y 0 20 40 60 Non-mi ochond ial espi a ion 0 30 60 90 0.0105 ATP-linked espi a ion Exons 11–12 loxP loxP Cp 1alox/lox mouse Cp 1a gene AdV-CMV-C e-GFP AdV-CMV-GFP a WT as ocy es CPT1A KO as ocy es 5 days B ain co ex dissocia ion (0–24 h neona es) P ima y cul u es 14CO2 TCA [6-14C]Glucose [1-14C]Py u a e Lac a e Py u a e [3-3H]Glucose Glycolysis 3H2O PDH WT CPT1A KO WT CPT1A KO WT CPT1A KO WT CPT1A KO TPI Time (min) 1 18 35 52 69 86 103 120 Olig FCCP Ro An WT CPT1A KO 0.3403 0 2.5 5.0 7.5 10.0 0.0119 d WT CPT1A KO 0 1 2 3 40.0316 0 200 400 600 0.0046 0 1 2 3 0.1224 [6-14C]Glucose con e sion o 14CO2 (nmol h–1 mg p o ein–1) [1-14C]Py u ic acid con e sion o 14CO2 (nmol h–1 mg p o ein–1) TCA [1-14C]Palmi a e 14CO2 Ke ogenesis [1-14C]Palmi a e 14Ke ones 0 0.8 1.6 2.4 0.0374 [1-14C]Palmi ic acid con e sion o 14C-ke ones (nmol h–1 mg p o ein–1) 14CO2 Fig. 2 | KO o Cp 1a in as ocy es inhibi s a y acid oxida ion and me abolic ewi ing enhancing mi ochond ial oxygen consump ion. a, S a egy used o ob ain Cp 1a KO as ocy es in as ocy es in p ima y cul u e. C ea ed wi h BioRende .com. b, Wes e n blo agains CPT1A p o ein in Cp 1a KO as ocy es in p ima y cul u e 5 days a e AdV-CMV-C e-GFP ansduc ion; n = 3 biologically independen cell cul u e p epa a ions; unpai ed S uden ’s - es , wo- ailed. β-Ac in was used as a loading con ol (Supplemen a y Fig. 2b). c, 14CO2 p oduc ion om [1-14C]palmi ic acid in WT and Cp 1a KO as ocy es in p ima y cul u e. Da a a e mean ± s.e.m. P alue is indica ed (n = 4 biologically independen samples; unpai ed S uden ’s - es , wo-sided). d, 14Ke ones p oduc ion om [1-14C]palmi ic acid in WT and Cp 1a KO as ocy es in p ima y cul u e. Da a a e mean ± s.e.m. P alue is indica ed (n = 4 biologically independen samples; unpai ed S uden ’s - es , wo-sided). e–h, 14CO2 p oduc ion om [6-14C]glucose (e) o [1-14C]py u ic acid ( ), a e o lac a e eleased (g) and glycoly ic lux as measu ed by he a e o [3-3H]glucose con e sion in o 3H2O (h), in WT and Cp 1a KO as ocy es in p ima y cul u e. TPI, iosephospha e isome ase. Da a a e mean ± s.e.m. P alues a e indica ed; n = 6 (e), 6 ( ), 8 (g) and 6 (h) biologically independen cell cul u e p epa a ions; pai ed S uden ’s - es , wo-sided. i, OCR analysis and calcula ed pa ame e s in WT and Cp 1a KO as ocy es in p ima y cul u e. Da a a e mean ± s.e.m. P alues a e indica ed (n = 5 biologically independen cell cul u e p epa a ions; unpai ed S uden ’s - es , wo-sided) (Supplemen a y Fig. 2j,k). Na u e Me abolism | Volume 5 | Augus 2023 | 1290–1302 1294 Le e h ps://doi.o g/10.1038/s42255-023-00835-6 compa ed wi h hose isola ed om WT mice. Neu ons isola ed om as ocy e-speci ic Cp 1a KO adul mice showed dec eased basal and ATP-linked espi a ion (Fig. 4b and Supplemen a y Fig. 4d), inc eased H2O2 and mi ochond ial ROS (Fig. 4c) wi h educed ∆ψm (Supplemen- a y Fig. 4e) when compa ed wi h hose isola ed om WT mice. Thus, neu ons adjacen o CPT1A KO as ocy es de elop adap i e changes d 0 200 400 600 WT siCon ol CPT1A KO siCon ol WT siNDUFS1 CPT1A KO siNDUFS1 Time (min) 1 18 35 52 69 86 103 120 OCR (pmol min–1 × 10–4 nuclei) Basal espi a ion (pmol min–1 × 10–4 nuclei) Olig FCCP Ro An H2O2 abundance (µmol µg–1 h–1) WT CPT1A KO siCon ol siNDUFS1 + + – – – – + + 0 2 4 60.508 0.033 0 80 160 240 0.0315 0.9917 0.0218 0.9999 WT CPT1A KO siCon ol siNDUFS1 e c WT CPT1A KO a I2+III2 I+III2 I NDUFSA9 (CI) WT CPT1A KO SDHA II CI supe -assembly (SC-CI/CI) WT CPT1A KO 0 0.5 1.0 1.5 2.0 0.0484 b CIII supe -assembly (SC-CIII/CIII) 0 1 2 30.0360 UQCRC2 (CIII) SDHA WT CPT1A KO WT CPT1A KO 0 1 2 3 4 0.0300 0 60 120 180 Mi ochond ial ROS in cocul u ed neu ons (Mi oSox luo escence) (a.u.) 0.0005 0.0093 WT mCAT Neu ons 0 30 60 90 0.0428 Glu a hione in cocul u ed neu ons (nmol mg p o ein–1) 0 50 100 150 0.0385 0.0150 Mi ochond ial memb ane po en ial in cocul u ed neu ons (a.u.) WT mCAT Neu ons 0 5 10 15 0.0027 0.0013 Apop osis in cocul u ed neu ons (%) WT mCAT Neu ons 0 0.5 1.0 1.5 0.0088 0 0.5 1.0 1.5 2.0 0.0278 A c 0 0.5 1.0 1.5 2.0 0.3556 0 0.5 1.0 1.5 2.0 0.5572 mRNA abundance in cocul u ed neu ons ( old) cFos A ccFos WT neu ons mCAT neu ons l 2.0 As WT As CPT1A KO k As WT As CPT1A KO g h i j Analysis o neu ons 3 days Cocul u e WT o mCAT neu ons (pla es) WT o CPT1A KO as ocy es (inse s) H2O2 abundance (µmol µg–1 h–1) I+III2+IV 2.5 1.5 1.0 0.1 MW (MDa) 1.7 2.5 1.5 0.5 0.1 1.7 MW (MDa) I2+III2 I+III2 III2 II I+III2+IV Fig. 3 | KO o Cp 1a in as ocy es induces mi ochond ial SCs leading o inc eased espi a ion and dec eased ROS a ec ing bioene ge ics and unc ion o cocul u ed neu ons. a, F ee CI and CI-con aining SCs (SC-CI) in WT and Cp 1a KO p ima y as ocy es, analysed by BNGE ollowed by immunoblo ing agains CI subuni NDUFA9. Da a a e mean ± s.e.m. P alues a e indica ed (n = 3 biologically independen cell cul u e p epa a ions; unpai ed S uden ’s - es , wo-sided). b, F ee CIII and CIII-con aining SCs (SC-CIII) in WT and Cp 1a KO p ima y as ocy es, analysed by BNGE ollowed by immunoblo ing agains CIII subuni UQCRC2. Da a a e mean ± s.e.m. P alues a e indica ed (n = 3 biologically independen cell cul u e p epa a ions; unpai ed S uden ’s - es , wo-sided). c, H2O2 p oduc ion in WT and CPT1A KO as ocy es in p ima y cul u e. Da a a e mean ± s.e.m. P alues a e indica ed (n = 6 independen cell cul u e p epa a ions; unpai ed S uden ’s - es , wo-sided). d, OCR analysis in WT and Cp 1a KO as ocy es in p ima y cul u e, ei he ans ec ed wi h sc ambled (con ol) o NDUFS1 siRNAs. Da a a e mean ± s.e.m. P alues a e indica ed (n = 4 biologically independen cell cul u e p epa a ions) (Supplemen a y Fig. 3 ). e, Basal espi a ion in WT and CPT1A KO as ocy es in p ima y cul u e, ei he ans ec ed wi h sc ambled (con ol) o NDUFS1 siRNAs. Da a a e mean ± s.e.m. P alues a e indica ed (n = 4 biologically independen cell cul u e p epa a ions; wo-way ANOVA ollowed by Tukey). , H2O2 p oduc ion by WT and CPTA1A KO as ocy es in p ima y cul u e, ei he ans ec ed wi h sc ambled (con ol) o NDUFS1 siRNAs. Da a a e mean ± s.e.m. P alues a e indica ed (n = 3 biologically independen cell cul u e p epa a ions; mul iple unpai ed S uden ’s - es ). g, S a egy used o assess he e ec o Cp 1a KO as ocy es on WT o mCAT neu ons in p ima y cul u e. C ea ed wi h BioRende .com. h–l, Glu a hione concen a ion (h), mi ochond ial ROS (i), ∆ψm (j), apop osis (k) and c-Fos and A c mRNA abundances (l) in WT o mCAT-exp essing ansgenic neu ons a e cocul u e wi h WT o Cp 1a KO as ocy es; n = 3 (h), 4 (i), 4 (j), 4 (k, WT), 4 (k, mi oCAT) and 4 (l) biologically independen cell cul u e p epa a ions; pai ed S uden ’s - es , wo-sided o simple compa isons and wo-way ANOVA ollowed by Tukey o mul iple compa isons. Na u e Me abolism | Volume 5 | Augus 2023 | 1290–1302 1295 Le e h ps://doi.o g/10.1038/s42255-023-00835-6 ha esul in mi ochond ial dys unc ion and edox s ess. To assess whe he he obse ed e ec s on neu ons ha e beha iou al implica- ions, mice we e subjec ed o a ba e y o pe o mance es s. The esul s e ealed ha as ocy e-speci ic Cp 1a KO mice did no de elop a signi ican impai men in he open ield pe o mance (Supplemen a y Fig. 5a) o in he o a od es (Supplemen a y Fig. 5b), indica ing lack o anxie y and mo o coo dina ion. Howe e , as ocy e-speci ic Cp 1a KO mice showed an impai men in he wo king memo y, as judged by he obse ed dec eased disc imina ion index in he no el objec ecogni- ion es (Fig. 4d and Supplemen a y Fig. 5c), as well as an impai men in he long- e m spa ial memo y acco ding o he Ba nes maze es (Fig. 4e and Supplemen a y Fig. 5d). Albei h ough a di e en mechanism, a simila ou pu akes place in neu on-speci ic Cp 1c-iso o m gene ic abla ion37. Al oge he , hese esul s indica e ha a y acid oxida ion in as ocy es is essen ial o main ain mi ochond ial ROS o ma ion, neu onal ene gy i ness and cogni i e pe o mance in mouse. In conclusion, he e we desc ibe ha mi ochond ial oxida ion o a y acids in as ocy es exhibi s signalling and highe -o de ce eb al ad an ages. This is mainly suppo ed by he indings ha as ocy ic-speci ic gene ic dele ion o a key s ep in a y acid use, CPT1A, impai s he physiological p oduc ion o signalling mi ochon- d ial ROS. In con as o glucose ca abolism ha , ia py u a e oxi- da ion, mainly conse es educing equi alen s as NADH(H+), a y acid β-oxida ion conse es bo h NADH(H+) and FADH2. No ably, in as ocy es, mos a y acid-de i ed ace yl-coenzyme A is con e ed in o ke one bodies10 ins ead o he NADH(H+)-gene a ing TCA cycle, WT CPT1A KO 0 s 4 s eBa nes maze Basal Tes 0 60 120 180 0.5979 0.0344 Basal Tes 0 5 10 15 0.0388 Basal Tes 0 50 100 150 0.0170 Escape quad an Escape hole WT CPT1A KO Time spen in escape quad an (s) Time explo ing he escape quad an (s) Dis ance a elled in escape quad an (m) WT CPT1A KO 0 s 1.7 s Familia New –1.0 –0.5 0 0.5 1.0 0.0182 WT CPT1A KO dNo el objec ecogni ion WT CPT1A KO Disc imina ion index (a.u.) cAs ocy es (ex i o) H2O2 abundance (µmol µg–1 h–1) Mi ochond ial ROS (Mi oSox luo escence) (a.u.) H2O2 abundance (µmol µg–1 h–1) Neu ons (ex i o) Mi ochond ial ROS (Mi oSox luo escence) (a.u.) 0 150 300 450 WT CPT1A KO 0 50 100 150 0.0376 WT CPT1A KO 1 50 70 90 110 30 10 Time (min) Olig FCCP Ro /An Neu ons (ex i o) WT CPT1A KO 0 1 2 3 40.0076 0 60 120 180 0.0011 0 150 300 450 WT CPT1A KO Time (min) 1 50 70 90 110 30 10 Olig FCCP Ro /An As ocy es (ex i o) OCR (pmol min–1 × 10–4 nuclei) OCR (pmol min–1 × 10–4 nuclei) 0 50 100 150 0.0204 WT CPT1A KO Basal espi a ion (pmol min–1 × 10–4 nuclei) Basal espi a ion (pmol min–1 × 10–4 nuclei) WT CPT1A KO 0 60 120 180 0.0007 0 1 2 3 40.0429 WT mouse As ocy e-speci ic CPT1A KO mouse a WT as ocy es CPT1A KO as ocy es WT neu ons WT neu ons om CPT1A KO mice Seaho se Fluo ime ySC analysis Flow cy ome y b Fig. 4 | As ocy e-speci ic Cp 1a KO mice enhance in as ocy es bu dec ease in neu ons mi ochond ial SCs and espi a ion causing cogni i e impai men . a, S a egy used o immunomagne ically isola e as ocy es and neu ons om WT o as ocy e-speci ic CPT1A KO adul mice. C ea ed wi h BioRende .com. b, OCR analysis and calcula ed basal espi a ion in immunomagne ically isola ed as ocy es ( op) and neu ons (bo om) om WT and as ocy e-speci ic Cp 1a KO mice. Da a a e mean ± s.e.m. P alues a e indica ed (n = 4 mice pe geno ype; unpai ed S uden ’s - es , wo-sided). c, H2O2 and mi ochond ial ROS analyses in immunomagne ically isola ed as ocy es ( op) and neu ons (bo om) om WT and as ocy e-speci ic Cp 1a KO mice. Da a a e mean ± s.e.m. P alues a e indica ed (n = 5 mice pe geno ype; unpai ed S uden ’s - es , wo-sided). d, No el objec ecogni ion es in WT and as ocy e-speci ic Cp 1a KO mice. Rep esen a i e pa hs and spa io empo al quan i a i e hea maps a e shown. Da a a e mean ± s.e.m. P alues a e indica ed (n = 9 (WT) o 7 (CPT1A KO) mice; unpai ed S uden ’s - es , wo-sided). e, Ba nes maze es in WT and as ocy e- speci ic Cp 1a KO mice 8 days a e aining. Spa io empo al quan i a i e hea maps a e shown. Da a a e mean ± s.e.m. P alues a e indica ed (n = 9 (WT) o 7 (CPT1A KO) mice; wo-way ANOVA ollowed by Tukey). P alues in he igu e. Na u e Me abolism | Volume 5 | Augus 2023 | 1290–1302 1296 Le e h ps://doi.o g/10.1038/s42255-023-00835-6 hus s eng hening he ela i e con ibu ion o a y acid-de i ed FADH 2 o elec on lux o CIII h ough elec on ans e ing-ubiquinone oxido educ ase. Consis en wi h his, he ela i e mRNA exp ession o Acadl and M pα, esponsible o long-chain a y acid elec on ans e o CIII ( e . 16), is high in as ocy es. By con as , impai men o a y acid oxida ion by CPT1A loss adap s as ocy e me abolism owa ds inc eased py u a e mi ochond ial oxida ion, CI assembly in o SC and mi ochond ial espi a ion. This mechanism is cohe en wi h he ecen ly p oposed model23 ha PDH ac i i y con e ges wi h he unc ional o ganiza ion o he mi ochond ial espi a o y chain o wa an op imal me abolic adap a ions. Ou da a also con i m 18 ha mi ochond ial espi a o y chain is o ganized in as ocy es unde a con o ma ion in which CI is no ully assembled in SC, pe mi ing a y acids o con ibu e o mi ochond ial espi a ion ia elec on ans e ing-ubiquinone oxido educ ase. Al hough his pa hway is known o be less ene ge ically e icien han CI-d i en espi a ion, i allows a ela i ely high ROS gene a ion18 wi h signalling pu poses28. By g an ing p e e en ial use o a y acids, ou esul s hus indica e ha as ocy es p io i ize he gene a ion o mi ochond ial ROS, essen ial o sus aining cogni i e pe o mance28, o e a bioene ge ic bene- i (Supplemen a y Fig. 5e). As ocy es a e hough o la gely mee hei ene gy needs om glycolysis 3,7,38–41 , a pa hway ha he e we show coexis s wi h a y acid oxida ion. Howe e , in ligh o ou da a, he con ibu ion o hese wo pa hways o bioene ge ics and signalling unc ions a e no analogous. Thus, al hough a y acids a e oxidized ia he TCA cycle a a highe a e han glucose in as ocy es, a y acids show highe espi a ion linked wi h ATP p oduc ion, indica ing mo e uel is equi ed o ene gy homeos asis. Glycolysis and a y acid oxida- ion hus appea o be wo pa hways ha each sus ain essen ial ace s o as ocy e me abolism and ROS signalling. Me hods Cp 1alox/lox mice All p o ocols we e pe o med acco ding o he Eu opean Union Di ec i e 86/609/EEC and Recommenda ion 2007/526/EC, ega ding he p o ec ion o animals used o expe imen al and o he scien- i ic pu poses, en o ced in Spanish legisla ion unde he law 6/2013. P o ocols we e app o ed by he Bioe hics Commi ee o he Uni e si y o Salamanca o CIC bioGUNE (posi on emission omog aphy and magne ic esonance spec oscopy) in acco dance wi h he Spanish legisla ion (RD53/2013). Cp 1alox/lox mice we e gene a ed by in oducing wo loxP si es lanking a segmen comp ising exons 11 and 12 o Cp 1a gene ia homologous ecombina ion in emb yonic s em cells unde a C57BL/6J backg ound 19 . Animals we e b ed a he Animal Expe imen a ion Facili y o he Uni e si y o Salamanca in cages (maximum o i e animals pe cage) wi h a 12 h ligh and da k cycle (ligh om 08:00). The humidi y was 45–65% and he empe a- u e was 20–25 °C. Animals we e ed ad libi um wi h a solid die (20% p o eins, 45% lipids and 35% ca bohyd a es, plus mine als and i amins) and wa e . In i o gene a ion o as ocy e-speci ic Cp 1a KO mice This was ca ied ou using a alida ed AAV s a egy20. Essen ially, AAV pa icles o he PHP.eB capsid (se o ype), known o e icien ly ans- duce he cen al ne ous sys em ia in a enous injec ion 42 , exp essing C e ecombinase d i en by he as ocy e-speci ic sho GFAP p omo e (PHP.eB-AAV-g aABC 1 D-C e-GFP) we e adminis e ed in a enously (50 µl aliquo s o a phospha e-bu e ed saline (PBS) solu ion con ain- ing 0.001% Plu onic F-68, Sigma-Ald ich and 1 × 1011 i al genomes, VG) h ough he e o-o bi al sinus o 2-mon h-old Cp 1a lox/lox male mice unde a b ie se o lu ane anaes hesia (Se o ane, a 6% o ini ia ion ollowed by oughly 3% o main enance in ai wi h supplemen s o O 2 and NO 2 0.4 and 0.8 l min −1 , espec i ely, using a gas dis ibu ion column, He sill H-3, and a apo ize , In e Med Penlons Sigma Del a). We used he e o-o bi al sinus in a enous ou e because o he highe success a e obse ed when compa ed wi h he ail o empo al ones 43 . Siblings o WT mice ecei ed equi alen amoun s o he same AAV pa icles ha did no ha bou C e ecombinase. Mice we e used om 4 weeks a e AAV injec ions. P ima y cul u es o as ocy es As ocy es in p ima y cul u e we e ob ained om he co ex o 0–24 h old Cp 1alox/lox mouse neona es28. Cell suspensions we e seeded in 175 cm 2 plas ic lasks in low glucose (5.5 mM) Dulbecco’s Modi ied Eagle’s Medium (DMEM) supplemen ed wi h 10% e al bo ine se um and 4 mM glu amine, and incuba ed a 37 °C in a humidi ied 5% CO2-con aining a mosphe e. To de ach non-as ocy ic cells, a e 7 days in i o (DIV), he lasks we e shaken a 150  .p.m. o e nigh . The supe na an was disca ded, and he a ached, as ocy e-en iched cells we e eseeded a 0.8–1 × 10 5 cells pe cm 2 in he app op ia e pla es. Cells we e used a 9 DIV. Indi idual p ima y cul u es o mouse co ical neu ons we e p epa ed om E14.5 day-old mCAT28 o WT mice, seeded a 2.0 × 105 cells pe cm2 in six-well o Seaho se pla es coa ed wi h poly-d-lysine (10 µg ml−1) and incuba ed in Neu obasal A sup- plemen ed wi h 2 mM glu amine, 5.5 mM glucose, 0.22 mM py u a e and 2% an ioxidan B27 supplemen . Cells we e incuba ed a 37 °C in a humidi ied 5% CO2-con aining a mosphe e. A 72 h a e pla ing, he medium was eplaced by 2% o he minus an ioxidan ( ha is, lack- ing i amin E, i amin E ace a e, supe oxide dismu ase, ca alase and glu a hione) B27 supplemen . Neu ons we e used on day 6. To ob ain as ocy e-neu onal cocul u es, as ocy es a 8 DIV we e eseeded on semipe meable polyes e T answell memb ane inse s (4.5 cm2, 0.4 µm po e size; Co ning) and allowed o a ach o 24 h. A e his ime, as ocy es we e ansduced wi h he adeno i al pa icles and, a e 4 days, as ocy e-con aining inse s we e placed o e 3 DIV neu ons and cocul u ed in Neu obasal A supplemen ed wi h 2 mM glu amine, 5.5 mM glucose, 0.22 mM py u a e and 2% B27 minus an ioxidan supplemen o 3 days. Immunocy ochemis y agains a neu onal (β-Tubulin III: 1/300; T2200; Sigma), as ocy ic (GFAP: 1/800; AB5541; Millipo e), oligodend ocy es (O4; 1/300; om mouse hyb idoma kindly dona ed by I. Fa iñas’ labo a o y) and mic oglial ma ke (CD45; 1/200; 553076; BD) was pe o med o de e mine he pu i y o he cul u es, which was oughly 100% as ocy es ( o as ocy e-en iched cul u es) and 99.02% neu ons, 0.43% as ocy es, 0.11% oligodend ocy es, 0.13% mic oglia and 0.31% o he cells ( o neu on-en iched cul u es). Gene a ion o Cp 1a KO as ocy es in p ima y cul u e This was ca ied ou by ansducing 9 DIV p ima y as ocy es, ob ained om Cp 1alox/lox mice, wi h adeno i al pa icles ha bou ing C e ecom- binase d i en by he ubiqui ous ci omegalo i us (CMV) p omo e (AdV-CMV-C e-GFP). As ocy es om he same cul u es ansduced wi h equi alen amoun s o he same AdV lacking C e ecombinase (Ad V-CMV-GFP) we e used as WT as ocy es. Cells we e used 5 days a e ansduc ion. Geno yping by PCR Fo Cp 1alox/lox geno yping, a PCR wi h he ollowing p ime s was pe o med 5′-CAGCTGCTCCACACCAAGGCT-3′ ( o wa d) and 5′-TGCCCTTCTACTGTCACATGG-3′ ( e e se), esul ing in a 403 base- pai (bp) band o Cp 1lox/lox mice and 209 bp o WT19. PCR condi- ions we e 30 s a 98 °C, 30 cycles o 5 s a 98 °C, 5 s a 60 °C, 10 s a 72 °C and a inal ex ension o 2 min a 72 °C. P ime s o geno yping he mCAT allele we e 5′-CTCCCAAAGTCGCTCTGAGTTGTTATCA-3′, 5′-CGATTTGTGGTGTATGTAACTAATCTGTCTGG-3′ and 5′-GCAGTG AGAAGAGTACCACCATGAGTCC-3′, which yielded a 778-bp band o he WT allele and a 245-bp band o he mCAT allele. PCR condi ions o mCAT geno yping we e 5 min a 94 °C, 35 cycles o 30 s a 94 °C, 30 s a 65 °C, 3 min a 68 °C and 8 min a 68 °C. PCR p oduc s we e esol ed in 3% aga ose gel using he 1 kilobase DNA ladde plus (The mo Fishe Scien i ic). Na u e Me abolism | Volume 5 | Augus 2023 | 1290–1302 1297 Le e h ps://doi.o g/10.1038/s42255-023-00835-6 qPCR wi h e e se ansc ip ion This was pe o med in o al RNA samples, pu i ied om p ima y cul- u es o as ocy es and neu ons using he GenElu e Mammalian To al RNA Minip ep Ki (Sigma), ollowing he manu ac u e ’s p o ocol. Ampli ica ions we e pe o med in 100 ng o RNA, using Powe SYBR G een RNA- o-CT 1-S ep ki (Applied Biosys ems). The p ime s we e ( o wa d and e e se, espec i ely) 5′-GGATGGCTATGGTCAAGGTC-3′ and 5′-GGCCTCACAGACTCCAGGTA-3′ o Cp 1a; 5′-TGCTCCATGG CAACTGCTAT-3′ and 5′-ACTCCCAGAGGTGCCCAAT-3′ o Cp 1b; 5′-CGCCCAGTATGAGAGGATGT-3′ and 5′-CCCTACACGGAAGAA TCTGC-3′ o Cp 1c; 5′-GCAGTGGTCTTCGAGTGGAT-3′ and 5′-CAG CTGCCTTCAGACCATCA-3′ o Acc1; 5′-GAGTGGAAGCGGTCTC ACAG-3′ and 5′-GCAAGCCTTCGTCCACATCC-3′ o Acc2; 5′-TCATTG CCAAGGCGGTTGAT-3′ and 5′-GCCATGGACTCAGTCACATAC-3′ o Acadl; 5′-CATGCGAATCCTCCAGGAAG-3′ and 5′-GCTACATCCAC ACCCACTTC-3′ o M pα; 5′-GGGAATGGTGAAGACCGTGT-3′ and 5′-CCGTTCCCTTCGGATTCTCC-3′ o c-Fos; 5′-CACTCTCCCGTGA AGCCATT-3′ and 5-TCCTCCTCAGCGTCCACATA-3′ o A c and 5′-AG AGTCATGAGCTGCCTGAC-3′ and 5′-CAACGTCACACTTCATGATG-3′ o β-ac in. The mRNA abundance o each ansc ip was no malized o ha o β-ac in ob ained in he same sample. The esul ing no ma- lized alues in as ocy es we e exp essed as he old change e sus he co esponding no malized alues in neu ons. When compa ed CPT1A KO wi h WT as ocy es, he esul ing no malized alues in CPT1A KO we e exp essed as he old change e sus he co esponding no ma- lized alues in WT as ocy es. Immunomagne ic pu i ica ion o as ocy es and neu ons om adul b ain Mouse adul b ain (minus ce ebellum and ol ac o y bulb) was disso- cia ed using he adul mouse b ain dissocia ion ki (Mil enyi Bio ec). The issue, once clean, was agmen ed wi h a s e ile scalpel in 2 ml pe hemisphe e o a disin eg a ion solu ion (Ea le’s Balanced Sal Solu ion, EBSS, 116 mM NaCl, 5.4 mM KCl, 1.5 mM MgSO4, NaHCO3 26 mM, NaH2PO4·2H2O 1.01 mM, glucose 4 mM, phenol ed 10 mg l−1, supple- men ed wi h albumin 14.4 µl ml −1 and DNase ype I 26 µl ml −1 , pH 7.2, ypsin 10.8 µl ml−1), and i was ypsinized a 37 °C in a he mos a ed ba h o 5 min, shaking equen ly o a oid decan a ion o he issue. I was u he mechanically disin eg a ed by i u a ion using a 5 ml se ological pipe e i e imes. Then, he suspension was e u ned o he he mos a ed ba h o 10 min, shaking equen ly. T ypsin ac i i y was s opped by adding 10% e al se um, be o e cen i uging he issue a 700g o 5 min in a mic o uge a 4 °C. Once he enzyma ically disin- eg a ed issue had been decan ed, he pelle was esuspended in a ypsin- ee disin eg a ion solu ion (EBSS + 13 µl ml−1 DNase + 20 µl ml−1 albumin) o mechanical i u a ion using a Pas eu pipe e. App oxi- ma ely i e passages we e pe o med pe a olume o 4 ml and pe hemisphe e. The supe na an was cen i uged o 3 min a 700g and he numbe o cells in he pelle was coun ed. Once a homogeneous suspen- sion o indi idualized adul neu al cells was achie ed, cell popula ion sepa a ions we e pe o med using MACS Technology using ei he he as ocy e-speci ic an i-ACSA-2 Mic obead Ki o he neu on-speci ic Neu on Isola ion Ki , acco ding o he manu ac u e ’s p o ocol (MACS Technology). We con i med he iden i y o he isola ed ac ions by wes e n blo ing agains as ocy ic (GFAP), neu onal (MAP2)-speci ic ma ke s and he pu i y wi h mic oglial (Iba1) and oligodend oglial (OLIG2)-speci ic ma ke s. Cell ans ec ions Fo NDUFS1 knockdown expe imen s, we used small in e e ing RNAs (siRNAs) agains NDUFS1 (siNDUFS1; s105592; Li e Technologies) and a siRNA con ol (siCon ol; 4390843; Li e Technologies). T ans ec ions wi h siRNAs we e pe o med wi h Lipo ec amine RNAiMAX eagen (Li e Technologies) acco ding o he manu ac u e ’s p o ocol using a siRNA inal concen a ion o 9 nM. Cells we e used a e 3 days. De e mina ion o me abolic luxes To assess a y acid, glucose and py u a e oxida i e luxes, we used adiome ic app oaches. To do his, as ocy es we e seeded in 8 cm 2 lasks hanging a mic ocen i uge ube con aining ei he 1 ml o benze- honium hyd oxide (Sigma) ( o 14 CO 2 equilib a ion) o 1 ml o H 2 O ( o 3 H 2 O equilib a ion). Fo b ain slices, hese we e placed in 25-ml glass lasks ha bou ing a cen al well wi h a ube con aining 0.8 ml o benze honium hyd oxide. All incuba ions we e ca ied ou in KRPG (NaCl 145 mM; Na2HPO4 5.7 mM; KCl 4.86 mM; CaCl2 0.54 mM; MgSO 4 1.22 mM; pH 7.35) con aining 5 mM d-glucose a 37 °C in he ai - he mos a ized chambe o an o bi al shake . To ensu e adequa e oxygen supply o oxida i e me abolism h oughou he incuba ion pe iod, he lasks’ a mosphe es we e gassed wi h ca bogen (5% CO 2 /95% O 2 ) be o e sealing hem wi h ubbe caps. To measu e he ca bon lux om a y acids o CO2, cells (o b ain slices) we e incuba ed in KRPG (5 mM glucose) bu e wi h 0.25 µCi ml −1 o ei he [U- 14 C]- o [1- 14 C] palmi ic acid (plus 10 µM palmi ic acid)44, as indica ed in he igu es. To measu e he ca bon lux om glucose o CO2 h ough he TCA cycle, cells we e incuba ed in KRPG (5 mM d-glucose) wi h 0.25 µCi ml −1 D-[6-14C]glucose45. To measu e he ca bon lux om py u a e o CO2 h ough mi ochond ial py u a e up ake ollowed by PDH ac i i y, cells we e incuba ed in KRPG (5 mM d-glucose) wi h 0.25 µCi ml−1 [1-14C]py u- a e (plus 1 mM py u a e). Incuba ions we e e mina ed a e 90 min by he addi ion o 0.2 ml 20% pe chlo ic acid (Me ck Millipo e) and, a e a u he 60 min, he ube con aining benze honium hyd oxide (wi h he apped 14 CO 2 ) was used o de e mine he adioac i i y using a liquid scin illa ion analyse (T i-Ca b 4810 TR, Pe kinElme ). The glycoly ic lux was measu ed by assaying he a e o 3H2O p oduc ion om [3- 3 H]glucose using a simila s a egy using 3 µCi ml −1 o d-[3- 3 H] glucose in KRPG bu e (5 mM d-glucose) o 120 min ( e . 45). A e incuba ions we e e mina ed wi h 0.2 ml 20% pe chlo ic acid, he cells we e u he incuba ed o 96 h o allow o 3 H 2 O equilib a ion wi h H 2 O p esen in he cen al mic ocen i uge ube. The 3H2O was hen meas- u ed by liquid scin illa ion coun ing (T i-Ca b 4810 TR, Pe kinElme ). The speci ic adioac i i y was used o he calcula ions. Unde hese expe imen al condi ions, 75% o he p oduced 14 CO 2 and 28% o he p oduced 3 H 2 O we e eco e ed and we e aken in o accoun o he calcula ions45. To assess he con e sion o a y acids o ke ones, as o- cy es we e seeded in 8 cm2 lasks wi h KRPG (5 mM glucose). The a e o ke one body o ma ion was de e mined by adding 0.25 µCi ml−1 o [1- 14 C]palmi ic acid bound o delipida ed bo ine se um albumin (BSA) (plus 10 µM palmi ic acid) o 2 h. A e incuba ions we e e mina ed wi h 0.2 ml 20% ( / ) pe chlo ic acid. Ke one bodies we e ex ac ed as a non- ola ile, acid-soluble p oduc 10. To do his, 1 ml o he medium was aken and added o a delipida ed 50 ml cen i uge ube (352070; FalconTM) con aining 8  ols o a chlo o o m:me hanol mix u e (2:1, / ) and 2  ols o KCl (0.1 M). A e shaking, i was cen i uged o 5 min a 3,000g. The uppe aqueous phase was aken and ans e ed o ano he ube con aining 8  ols o chlo o o m:me hanol mix u e (2:1  / ). A e shaking and cen i uging unde he same condi ions, he uppe aqueous phase was aken and ans e ed o a liquid scin il- la ion ial o coun ing. Lac a e de e mina ion Lac a e concen a ions we e measu ed in he cul u e medium spec o- pho ome ically45 by he de e mina ion o he inc emen s in he abso b- ance o he samples a 340 nm in a mix u e con aining 1 mM NAD + , 8.25 U o lac a e dehyd ogenase in 0.25 M glycine, 0.5 M hyd azine and 1 mM e hylenediamine e aace ic acid (EDTA) bu e , pH 9.5. β-Hyd oxybu y a e de e mina ion As ocy es we e incuba ed o 48 h in esh medium, which was col- lec ed and snap ozen a −80 °C. β-Hyd oxybu y a e was de e mined using a spec opho ome ic-based de ec ion ki (MAK134, Sigma) in 40 µl o samples ollowing he manu ac u e ’s ins uc ions. Na u e Me abolism | Volume 5 | Augus 2023 | 1290–1302 1298 Le e h ps://doi.o g/10.1038/s42255-023-00835-6 OCR assessmen OCRs o cell in p ima y cul u e o eshly immunomagne ically isola ed we e measu ed in eal- ime in an XFe24 Ex acellula Flux Analyse (Seaho se Bioscience; Seaho se Wa e Desk op so wa e .2.6.1.56). This equipmen measu es he ex acellula medium O 2 lux changes o cells seeded in XFe24-well pla es. Regula cell medium was emo ed and washed wice wi h DMEM unning medium (XF assay modi ied supplemen ed wi h 5 mM glucose, 2 mM l-glu amine, 1 mM sodium py u a e, 5 mM HEPES, pH 7.4) and incuba ed a 37 °C wi hou CO2 o 30 min o allow cells o p e-equilib a e wi h he assay medium. Oligo- mycin, FCCP (ca bonyl cyanide-p- i luo ome hoxyphenylhyd azone) and a mix u e o o enone and an imycin, dilu ed in DMEM unning medium, we e loaded in o po -A, po -B and po -C, espec i ely. Final concen a ions in XFe24 cell cul u e mic opla es we e 1 µM oligomycin, 2 µM FCCP, 1 µM o enone and 2.5 µM an imycin. The sequence o measu emen s was as ollows unless o he wise desc ibed. The basal le el o OCR was measu ed h ee imes, and hen po -A was injec ed and mixed o 3 min, a e OCR was measu ed h ee imes o 3 min. Same p o ocol wi h po -B and po -C. OCR was measu ed a e each injec ion o de e mine mi ochond ial o non-mi ochond ial con ibu ion o OCR. All measu emen s we e no malized o a e age h ee measu emen s o he basal (s a ing) le el o cellula OCR o each well sub ac ing he non-mi ochond ial OCR. Each sample was measu ed in 3–5 eplicas. Expe imen s we e epea ed 3–5 imes in biologically independen cul u e p epa a ions. Non-mi ochond ial OCR was de e mined by OCR a e injec ion o an imycin plus o e- none oge he o sepa a ely. Maximal espi a ion was de e mined by maximum OCR a e a e FCCP injec ion minus non-mi ochond ial OCR. ATP p oduc ion was de e mined by he las OCR measu emen be o e oligomycin injec ion minus he minimum OCR measu emen a e oligomycin injec ion. When indica ed, e omoxi (100 µM) was injec ed in po -A o de e mine a y acids-dependen espi a ion, which was ob ained by sub ac ing he minimum OCR alue a e e omoxi o ha be o e e omoxi injec ion. To es ima e CI-sus ained mi ochond ial espi a ion, OCR be o e o enone injec ion was sub- ac ed he minimum OCR measu emen a e o enone injec ion, and hen om his alue he non-mi ochond ial OCR was sub ac ed. We also de e mined CI-sus ained espi a ion in pe meabilized cells. To do so, XF DMEM bu e was swi ched o manni ol and suc ose (con ain- ing 70 mM suc ose, 220 mM manni ol, 10 mM KH 2 PO 4 , 5 mM MgCl 2 , 2 mM HEPES and 1 mM EGTA, pH 7.2) bu e and he basal OCR le el moni o ed. To assess OCR in pe meabilized cells, digi onin (25 µg ml −1 ), l-glu amine:mala e (Gln:Mal; 4 mM:0.5 mM) and ADP (1 mM) we e added o s imula e NAD + educ ion and espi a ion. Ro enone and an i- mycin we e added sequen ially o independen ly calcula e CI-sus ained espi a ion and non-mi ochond ial OCR. CI-sus ained espi a ion was calcula ed as he di e ence be ween OCR a e digi onin/Gln/Mal/ADP and he OCR a e an imycin (non-mi ochond ial OCR). Speci ic ac i i y o he mi ochond ial espi a o y complexes Cells we e collec ed and suspended in 10 mM phospha e bu e (KH2PO4; pH 7.0). A e h ee cycles o eezing and hawing o ensu e cellula dis up ion, he speci ic ac i i ies o CI, CI–III, CII–III, CIV and ci a e syn hase we e de e mined. Ro enone-sensi i e CI (NADH-ubiquinone oxido educ ase) ac i i y46 was measu ed in KH 2 PO 4 (25 mM, pH 7.2) in he p esence o 10 mM MgCl 2 , 2.5 mg ml −1 BSA, 0.15 mM NADH and 1 mM KCN. Changes in abso bance a 340 nm (30 °C) (ε = 6.81 mM−1 cm−1) we e eco ded a e he addi ion o 50 µM ubiquinone and 10 µM o enone. CI–III (NADH-cy och ome c oxido e- duc ase) ac i i y was de e mined in KH2PO4 (25 mM; pH 7.2) in he p esence o 10 mM MgCl 2 , 50 mg ml −1 BSA, 300 mM KCN and 330 mM o oxidized cy och ome c. Changes in abso bance we e eco ded (550 nm; 30 °C) (ε = 19.2 mM −1 cm −1 ) a e he addi ion o 10 mM NADH and 10 µM an imycin A plus 25 µM o enone. CII–III (succina e-cy och ome c oxido educ ase) ac i i y47 was de e mined in he p esence o 100 mM phospha e bu e , plus 0.6 mM EDTA(K+), 2 mM KCN and 200 µM cy och ome c. Changes in abso bance we e eco ded (550 nm; 30 °C) (ε = 19.2 mM −1 cm −1 ) a e he addi ion o 20 mM succina e and 10 µM an imycin A. Fo CIV (cy och ome c oxidase) ac i i y, he i s - a e cons an , k (min−1 mg p o ein−1) o cy och ome c oxida ion was de e - mined48 in he p esence o 10 mM phospha e bu e (KH2PO4; pH 7.0) and 50 µM educed cy och ome c; abso bance was eco ded e e y minu e a 550 nm, 30 °C (ε = 19.2 mM −1 cm −1 ). Ci a e syn hase ac i i y 49 was measu ed in he p esence o 93 mM o T is-HCl, 0.1% ( / ) T i on X-100, 0.2 mM ace yl-CoA and 0.2 mM 5,5-di hio-bis-(2-ni obenzoic acid) (DTNB); he eac ion was s a ed wi h 0.2 mM o oxaloace a e and he abso bance was eco ded a 412 nm (30 °C) (ε = 13.6 mM−1 cm−1). Da a we e exp essed as he a io o he ac i i ies o each complex agains he ci a e syn hase ac i i y. PDH ac i i y PDH ac i i y was de e mined by he educ ion o NAD+ o NADH, cou- pled o he educ ion o a epo e dye o yield a colou ed eac ion p oduc wi h an inc ease in abso bance a 450 nm a oom empe a u e, using he PDH Enzyme Ac i i y Mic opla e Assay Ki (Abcam, ca alogue no. ab109902) ollowing he manu ac u e ’s ins uc ions. Cell homoge- na es (300 µg o p o ein) we e added o each well and he solubilized PDH enzyme was immunocap u ed o 3 h. A e washing wice wi h s abilize , esh assay solu ion was added and he abso bance o each well measu ed a 37 °C by a kine ic p og am a 450 nm o 30 min wi h a 60 s eading in e al in a Va ioskan Flash (The mo Scien i ic). PDH ac i i y (µOD × min−1) was exp essed as he ini ial eac ion a e de e - mined om he slopes o he cu es gene a ed. De e mina ion o glu a hione concen a ions Cells we e lysed wi h 1% (w/ ) o sul osalicylic acid, cen i uged a 13,000g o 5 min a 4 °C, and he supe na an s we e used o he de e mina ion o o al glu a hione ( ha is, educed glu a hione plus wice he concen a ion o oxidized glu a hione), using oxidized glu a hione (0–50 µM) as s anda d as desc ibed p e iously 50 . To al glu a hione was measu ed in eac ion bu e (0.1 mM NaHPO4, 1 mM EDTA, 0.3 mM DTNB, 0.4 mM NADPH, glu a hione educ ase 1 U ml−1, pH 7.5) by eco ding he inc ease in he abso bance a e he eac ion o educed glu a hione wi h DTNB o 2.5 min a 15 s in e als using a Va ioskan Flash (The mo Fishe ) spec opho ome e (λ = 405 nm). Glu a hione concen a ion (nmol mg−1 p o ein) was calcula ed om he slopes ob ained in he samples, ex apola ing hem o hose ob ained in he s anda d. Flow cy ome ic de ec ion o CPT1A To assess he adeno i al pa icles-media ed C e ecombinase ans- duc ion e iciency in p ima y as ocy ic cul u es, cells we e ixed, pe meabilized using he Fix&Pe m ki (Bec on Dickinson Biosciences) and incuba ed wi h an i-CPT1A an ibody (1/500) o 1 h a oom empe - a u e. Then, cells we e incuba ed wi h he seconda y Cy5-conjuga ed an ibody o 30 min a oom empe a u e and analysed in he FACScali- bu low cy ome e (15 mW a gon ion lase ; CellQues so wa e, Bec on Dickinson Biosciences) using FL1 and FL4 channels o GFP and CPT1A labelling, espec i ely. Flow cy ome ic analysis o apop o ic cell dea h Cells we e ca e ully de ached om he pla es using 1 mM EDTA ( e asodium sal ) in PBS (pH 7.4). APC-conjuga ed annexin-V and 7-amino-ac inomycin D (7-AAD) (Bec on Dickinson Biosciences) we e used o de e mine quan i a i ely he pe cen age o apop o ic neu ons by low cy ome y. Cells we e s ained wi h annexin-V-APC and 7-AAD in binding bu e (100 mM HEPES, 140 mM NaCl, 2.5 mM CaCl2), acco ding o he manu ac u e ’s ins uc ions, and 5 × 104 cells we e analysed, in h ee eplica es pe condi ion, on a FACScalibu low cy ome e (15 mW a gon ion lase ; CellQues so wa e, Bec on Dickinson Biosciences),