The Mitochondrial Pyruvate Carrier Coupling Glycolysis and the Tricarboxylic Acid Cycle Is Required for the Asexual Reproduction of Toxoplasma gondii

T. gondii is a zoonotic parasite capable of infecting many warm-blooded organisms, including humans. Among others, a feature that allows it to parasitize multiple hosts is its exceptional metabolic plasticity. ABSTRACT Toxoplasma gondii is an obligate intracellular parasite capable of infecting humans and animals. The organism has extraordinary metabolic resilience that allows it to establish parasitism in varied nutritional milieus of diverse host cells. Our earlier work has shown that, despite flexibility in the usage of glucose and glutamine as the major carbon precursors, the production of pyruvate by glycolytic enzymes is central to the parasite’s growth. Pyruvate is metabolized in a number of subcellular compartments, including the mitochondrion, apicoplast, and cytosol. With the objective of examining the mechanism and importance of the mitochondrial pool of pyruvate imported from the cytosol, we identified the conserved mitochondrial pyruvate carrier (MPC) complex, consisting of two subunits, MPC1 and MPC2, in T. gondii. The two parasite proteins could complement a yeast mutant deficient in growth on leucine and valine. Genetic ablation of either one or both subunits reduced the parasite’s growth, mimicking the deletion of branched-chain ketoacid dehydrogenase (BCKDH), which has been reported to convert pyruvate into acetyl-coenzyme A (CoA) in the mitochondrion. Metabolic labeling of the MPC mutants by isotopic glucose revealed impaired synthesis of acetyl-CoA, correlating with a global decrease in carbon flux through glycolysis and the tricarboxylic acid (TCA) cycle. Disruption of MPC proteins exerted only a modest effect on the parasite’s virulence in mice, further highlighting its metabolic flexibility. In brief, our work reveals the modus operandi of pyruvate transport from the cytosol to the mitochondrion in the parasite, providing the missing link between glycolysis and the TCA cycle in T. gondii. IMPORTANCE T. gondii is a zoonotic parasite capable of infecting many warm-blooded organisms, including humans. Among others, a feature that allows it to parasitize multiple hosts is its exceptional metabolic plasticity. Although T. gondii can utilize different carbon sources, pyruvate homeostasis is critical for parasite growth. Pyruvate is produced primarily in the cytosol but metabolized in other organelles, such as the mitochondrion and apicoplast. The mechanism of import and physiological significance of pyruvate in these organelles remains unclear. Here, we identified the transporter of cytosol-derived pyruvate into the mitochondrion and studied its constituent subunits and their relevance. Our results show that cytosolic pyruvate is a major source of acetyl-CoA in the mitochondrion and that the mitochondrial pyruvate transporter is needed for optimal parasite growth. The mutants lacking the transporter are viable and virulent in a mouse model, underscoring the metabolic plasticity in the parasite’s mitochondrion.

compromises the parasite's growth and virulence (18,20). In particular, mutagenesis of PYK1 and PYK2 suggests a critical role of pyruvate homeostasis in tachyzoite proliferation. PYK1 is needed for pyruvate production and parasite growth, irrespective of the carbon source (glucose or glutamine). Structurally similar compounds like lactate and alanine can be converted to pyruvate and partly restore the growth defect of the PYK1-depleted mutants (20).
Pyruvate is utilized by several metabolic pathways in multiple organelles. While the PYK1-derived cytosolic pool of pyruvate is crucial for tachyzoites, the underlying cause is unclear. It can be converted to lactate and amino acids in the cytosol and/or imported into the mitochondrion, producing acetyl-coenzyme A (CoA) for the TCA cycle. The latter reaction is catalyzed by a branched-chain ketoacid dehydrogenase (BCKDH) enzyme (21), which has been repurposed to serve the catalytic role of pyruvate dehydrogenase (PDH) in apicomplexan parasites. The cytosolic pyruvate may also enter the apicoplast to fuel the methylerythritol 4-phosphate (MEP) pathway for producing isoprenoids, as well as the FASII pathway for fatty acid synthesis. Because the apicoplast-localized PYK2 is dispensable even though the MEP and FASII pathways are required for tachyzoite growth, it is reasonable to assume that the organelle imports pyruvate from the cytosol. Likewise, we postulate the presence of a transport system in the parasite mitochondrion, which constituted the focus of this study of the tachyzoite stage of T. gondii.

RESULTS
T. gondii encodes two functional MPC proteins. To search for the mitochondrial pyruvate carrier (MPC) subunits in T. gondii, bona fide MPCs from humans and yeasts were used as the query sequences for BLAST analyses in ToxoDB (https://toxodb.org). The top hits were used as baits for reciprocal BLAST searches in human and yeast genomes. We found two orthologs in Toxoplasma, termed MPC1 (TGGT1_235880) and MPC2 (TGGT1_204370). The open reading frames of T. gondii MPC1 (TgMPC1) and TgMPC2 encode relatively small proteins of 160 and 135 amino acids, respectively. Their sequences are well conserved with respect to the corresponding MPC subunits from other eukaryotes, as suggested by multiple sequence alignment ( Fig. 1A and B). To assess the subcellular localization of TgMPCs, a spaghetti monster-hemagglutinin (smHA) epitope tag (22) was fused to their C-terminal ends using CRISPR/Cas9-mediated homologous recombination at the 39 end of the target gene ( Fig. 1C) (23). TgMPC1-smHA and TgMPC2-smHA colocalized with a mitochondrial marker, T. gondii heat shock protein 60 (TgHSP60), in transgenic parasites (Fig. 1D).
Next, we tested the function of TgMPC1 and TgMPC2 by expressing them in a Saccharomyces cerevisiae yeast mutant lacking its endogenous MPC1 (ScDmpc1) or MPC2 (ScDmpc2) (24). As shown by the results in Fig. 1E and F, the ScDmpc1 and ScDmpc2 strains were auxotrophic for leucine and valine, which serve as carbon sources in the absence of mitochondrial pyruvate transport. The growth of both mutants could be restored by ectopic expression of the corresponding yeast proteins (ScMPC1 and ScMPC2; positive controls), whereas the empty-vector negative control was unable to rescue the auxotrophic behavior of either strain for branched-chain amino acids (BCAAs). The expression of TgMPC1 partly complemented the growth of ScDmpc1 and ScDmpc2 mutants on plates without valine and leucine, while TgMPC2 conferred a complete restoration of the growth of ScDmpc1 and ScDmpc2 mutants. We also performed functional complementation assays in defined liquid cultures without or with BCAAs, confirming our results (Fig. S1 in the supplemental material).
MPC1 and/or MPC2 is dispensable for parasite growth but required for optimal parasite growth. To evaluate the physiological significance of the TgMPCs in T. gondii, we first deleted the MPC1 gene by replacing it with a pyrimethamine-resistant DHFR*, a mutant version of Toxoplasma dihydrofolate reductase ( Fig. 2A) (23). The resulting parasite clones were screened by diagnostic PCRs (PCR1, PCR2, and PCR3) (Fig. 2B). We also constructed a complemented strain expressing MPC1 (CoMPC1) in the deletion mutant to facilitate subsequent phenotypic analyses. To do this, an HA-tagged TgMPC1 expression cassette was inserted into the UPRT locus of the mutant (Fig. 2C). The protein localized in the mitochondrion, as expected (Fig. 2D). The Dmpc1 mutant proliferated more slowly than the parental strain, and complementation by ectopic MPC1 restored the replication defect (Fig. 2E). Consequently, the plaques formed by the knockout strain were much smaller than the plaques formed by the parental and CoMPC1 strains ( Fig. 2F to H). These results demonstrated a need for MPC1 for tachyzoite growth under standard culture conditions. The alignments were performed using Clustal X2. (C) TgMPC1 or TgMPC2 subunits were C-terminally tagged with the spaghetti monster tag containing 10 HA epitopes (smHA) at the endogenous gene locus through CRISPR/Cas9-mediated site-specific gene insertion in the RH Dku80 strain. The blue arrowhead indicates the CRISPR cut site. (D) Immunofluorescence assays of the 10 HA-tagged strains constructed as shown in panel C to check the localization of MPC1 and MPC2. TgHSP60 was used as a mitochondrion-specific marker for colocalization purposes. (E and F) Serial dilutions of the indicated yeast strains were spotted on plates with (1) or without (2) valine and leucine and grown at 30°C for 3 days.
Studies of other organisms suggest that functional MPC comprises two subunits in a heterodimeric complex. However, human MPC2 has also been reported to form a homodimeric complex (25). We, therefore, tested the impact of MPC2 inactivation on parasite growth. A Dmpc2 single mutant was generated, and its plaque growth and replication compared to those of the parental strain (Fig. S2). It displayed a defect similar to that of the Dmpc1 mutant. To further test whether the loss of both subunits would have additive effects, we constructed a Dmpc1-Dmpc2 double mutant (D1-D2) by deleting TgMPC1 from the Dmpc2 strain. The recombination-specific screening PCRs (similar in design to those depicted in Fig. 2A   in a D1-D2 clone (Fig. 3A). The growth assays suggested that the D1-D2 double mutant phenocopied each MPC single mutant (Dmpc1 or Dmpc2) (Fig. 3B to E and Fig. S2C to E). These results suggest that TgMPC1 and TgMPC2 are not functionally redundant but may instead function as a heterodimeric complex.
The MPC deletion strains mimic the BCKDH-deficient mutants. TgBCKDH converts pyruvate to acetyl-CoA in the parasite's mitochondrion (21). Therefore, we reasoned that if the glycolysis-derived pyruvate imported by MPC proteins was the only source for acetyl-CoA synthesis in the mitochondrion, a mutant with their deletion should mimic the growth of the BCKDH mutant. To test this notion, we generated a Dbckdh mutant by replacing the corresponding gene with DHFR*, using a strategy similar to the one depicted in Fig. 2A. Indeed, the Dbckdh and D1-D2 strains exhibited nearly identical growth patterns in the plaque and replication assays (Fig. 4). In yeast and mammalian cells, BCAAs like leucine and valine can serve as a source of acetyl-CoA and thereby render MPC proteins dispensable ( Fig. 1E and Fig. S1). To examine whether the degradation of BCAAs has any role in acetyl-CoA production in mutants lacking MPC, we targeted the branched-chain aminotransferase (BCAT) that was required for BCAA degradation and constructed a Dmpc2-Dbcat double mutant (Fig.  S2B). Surprisingly, the growth of the Dmpc2-Dbcat mutant was similar to that of the MPC mutants ( Fig. 3B to E). These results suggest that an interrupted supply of pyruvate in the mitochondrion cannot be compensated by the degradation of BCAAs and that pyruvate imported by MPC is a major source of mitochondrial acetyl-CoA.
Deletion of MPC impairs the central carbon metabolism of tachyzoites. To explore the mechanisms underlying the growth defect observed in the MPC mutants, we investigated metabolic changes in MPC1 knockout tachyzoites. We first examined the efficiency of acetyl-CoA production from glucose by labeling parasites with [ 13 C 6 ]glucose for 4 h, followed by liquid chromatography-mass spectrometry (LC-MS) (Fig. 5A). In the parental RH parasites, .60% of the total acetyl-CoA pool was labeled with 13 C (with one or more carbons being 13 C), which, however, was reduced to about 30% in the Dmpc1  mutant. The decrease of acetyl-CoA labeling was restored in the MPC1-complemented strain. We also tested the impact of MPC1 mutation on the carbon flux via glycolysis, the TCA cycle, and the pentose phosphate pathway (PPP). Tachyzoites were labeled with [ 13 C 6 ]glucose, and the inclusion of 13 C into key metabolites was quantified. The Dmpc1 strain displayed impaired flux of glucose-derived carbon into several metabolic intermediates, including PEP, lactate, citrate, malate, succinate, and ribose-5-phosphate. As expected, MPC1 complementation restored the observed metabolic phenotype (Fig. 5B).
The MPC and BCKDH mutants show attenuated virulence in mice. To investigate the in vivo importance of pyruvate metabolism, the Dmpc1, Dbckdh, and CoMPC1 strains were used to infect ICR mice by intraperitoneal (i.p.) injection (100 parasites/ mouse), and the mice were monitored for survival. All animals infected with the parental strain died within 9 days, whereas those infected by Dmpc1 or Dbckdh mutants survived 3 or 4 days longer, suggesting a mild attenuation of virulence. The CoMPC1 strain was nearly as virulent as the parental parasites, as anticipated (Fig. 6A). Our extended work determined the proliferation of these strains in mice infected with tachyzoites (5 days postinfection with 1 Â 10 5 parasites i.p.). The peritoneal fluids were collected, and the parasite burden was scored by quantitative PCR. The animals infected with the parental or CoMPC1 strain harbored about 1 Â 10 8 cells in the perito-   neal fluids (Fig. 6B). However, the parasite burdens in mice infected with the Dmpc1 or Dbckdh mutant were close to 1 Â 10 5 (0.1% of the burden in controls), suggesting a strong propagation defect in the indicated mutants. MPC1 is required for parasite differentiation and chronic infection. Finally, we determined the role of MPC in the onset of chronic infection. In this regard, we deleted MPC1 in a type II strain, ME49, that is competent for bradyzoite formation in vivo and in vitro. Although the ME49Dmpc1 mutant was viable and could be maintained in culture, it proliferated much more slowly under normal (tachyzoite) growth and alkaline (bradyzoite induction at pH 8.2 under ambient CO 2 ) conditions ( Fig. 7A to B). When parasites were cultured in an alkaline medium for 3 days and then stained with Dolichos biflorus agglutinin-fluorescein isothiocyanate (DBA-FITC) to monitor the formation of encysted bradyzoites, the ME49Dmpc1 mutant was less efficient in forming DBA-positive (DBA 1 ) bradyzoites (Fig. 7C). We also tested the impact of MPC1 deletion on cyst development in vivo. The ME49 and ME49Dmpc1 strains were used to infect ICR mice (100 tachyzoites/mouse), and the survival of animals was observed for 30 days. The mice infected with the mutant had a mortality rate of 66.7%, compared to 88% with the parental strain, suggesting decreased virulence upon the deletion of MPC1 (Fig. 7D). When the cyst burden in the brains of surviving animals was analyzed on day 30, the Dmpc1 mutant was found to have formed nearly 10 times fewer cysts than the ME49 strain (Fig. 7E). Also, the mutant produced much smaller cysts as determined by the diameters of cysts (Fig. 7F). These results demonstrate that MPC1 is required for bradyzoite development both in vitro and in vivo.

DISCUSSION
Pyruvate is produced in the cytosol and utilized by several metabolic pathways in different organelles of T. gondii. This study aimed to examine the role of pyruvate metabolism in the parasite's mitochondrion. We identified the mitochondrial pyruvate carrier in T. gondii, comprising MPC1 and MPC2 proteins highly conserved across eukaryotes. Our results from deleting the genes individually or in combination show that both subunits are needed for optimal parasite growth. Mutants lacking functional MPC experienced decreased flux through glycolysis and the TCA cycle, as well as reduced incorporation of glucose-derived carbons into acetyl-CoA (Fig. 8). These metabolic changes were similar to what was observed in the Dbckdh mutant. On the other hand, tachyzoites lacking MPC or BCKDH were still viable and virulent. These observations emphasize the extreme flexibility of tachyzoites to thrive and establish parasitism in various nutritional milieus. The dispensability of MPC also suggests alternative sources for pyruvate or acetyl-CoA in the mitochondrion of T. gondii (Fig. 8). Unlike yeast and mammalian cells, which can utilize BCAAs and fatty acids to make acetyl-CoA and thereby maintain normal growth in the absence of MPC (26), the tachyzoite mutants of T. gondii are impaired even under nutrient-rich conditions. Toxoplasma does require BCAAs (leucine, isoleucine, and valine) for growth; however, they are probably not a significant source for acetyl-CoA synthesis in the mitochondrion of tachyzoites, because deletion of the BCAT gene, which is involved in BCAA degradation, did not noticeably affect tachyzoite growth (27). In addition, the Dmpc2-Dbcat double mutant phenocopies the Dmpc2 mutant, suggesting that BCAAs are not used for acetyl-CoA production in the tachyzoite mitochondrion and are probably needed as essential amino acids for protein synthesis. In mammalian cells, alanine can also be transported into mitochondria and then converted to pyruvate by glutamic-pyruvic transaminase 2 (GPT2) in the mitochondrial matrix (28)(29)(30). GPT2 homologs can be found in the Toxoplasma genome, but their contribution to acetyl-CoA production and parasite growth deserves further exploration. Likewise, b oxidation of fatty acids as a possible source of mitochondrial acetyl-CoA has not been investigated in tachyzoites. However, the lack of canonical mitochondrial acyl-carnitine/carnitine carriers and low expression of b oxidation proteins in tachyzoites rule out a contribution of fatty acids to acetyl-CoA synthesis (31). The tolerance of MPC deletion in Toxoplasma brings into question the role of the TCA cycle. Acetyl-CoA not only fuels energy production but is also required for the acetylation of diverse biomolecules, including proteins, DNA, and RNA. It is currently unknown what underlies the poor growth of the Dmpc and Dbckdh mutants. T. gondii harbors additional enzymes in other organelles to produce acetyl-CoA for different reactions (32), which include acetyl-CoA synthetase (ACS) and ATP citrate lyase (ACL) in the cytosol (32) and apicoplast-localized PDH (33). Disruption of these enzymes is not lethal except for the simultaneous deletion of ACS and ACL. Loss of ACS and ACL causes hypoacetylation of nucleocytosolic and secretory proteins, whereas the Dbckdh strain displays hypoacetylation of mitochondrial proteins, suggesting acetylation of organelle proteins by a locally produced pool of acetyl-CoA. Whether the change in mitochondrial protein acetylation is responsible for the growth defect of the Dmpc and Dbckdh mutants merits further investigation.
Another major role of the mitochondrial acetyl-CoA is to drive energy production. T. gondii has a fully functional TCA cycle (12,13), but its physiological role is not well defined. Conditional depletion of the succinyl-CoA synthetase causes only a 30% reduction in the parasite's growth (34), and the lack of a more severe growth defect suggests that the TCA cycle is not essential in tachyzoites. Notably, the Dmpc and Dbckdh strains phenocopy the depletion of succinyl-CoA synthetase, endorsing a nonessential role of the TCA cycle. Likewise, six of the eight TCA cycle enzymes in Plasmodium falciparum can be deleted without impairing the erythrocytic development of the malaria parasites (35). MacRae et al. reported that sodium fluoroacetate (NaFAc), a potent inhibitor of aconitase, can abolish the plaque growth of T. gondii tachyzoites, suggesting a critical role of the TCA cycle (13). The authors also described a GABA shunt in T. gondii that could produce succinate from glutamine as a carbon source independently of succinyl-CoA synthetase. It is unclear whether inhibition of aconitase is the only cause of growth arrest by NaFAc or if it exerts a pleiotropic effect by inhibiting other pathways, including mammalian enzymes. The physiological roles of the TCA cycle in T. gondii, therefore, warrant systematic investigation using a chemogenetic approach.

MATERIALS AND METHODS
Parasite strains and cell cultures. The type I (RHDhxgprt and RH Dku80), type II (ME49), and subsequent derivative strains of T. gondii were propagated in human foreskin fibroblast (HFF) cells (ATCC, USA) as described elsewhere (36). HFF cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS; Life Technologies, USA), 100 mg/mL streptomycin, and 5 mM L-glutamine. The Dmpc1 and Dmpc2 mutants of S. cerevisiae and the pJR3455A vector for the ectopic expression of MPC proteins (37) were provided by Jared Rutter (University of Utah, USA).
Construction of plasmids. The primers used in this study (Table S1) were synthesized by Tsingke Biotechnology (China) and Life Technologies (Germany). Locus-specific CRISPR plasmids were generated by replacing the UPRT-targeting guide RNA (gRNA) in pSAG1::Cas9-U6::sgUPRT with the target-specific gRNAs. The donor plasmids used to delete MPC1, MPC2, or BCKDH genes were generated by cloning the corresponding 59 and 39 homology arms and the drug selection marker (DHFR*) into pUC19, using the ClonExpress II one-step cloning kit (Vazyme Biotech, China) (20). The pTub::MPC1-HA-CAT construct for MPC1 complementation (CoMPC1) was made by replacing TgLDH1 in pCom-LDH1 (38). The yeast expression constructs in the pJR3455A vector were generated by restriction cloning of the corresponding open reading frames. The PCR amplicons and vector were digested by XbaI and XhoI and ligated using T4 DNA ligase (TaKaRa, Japan), followed by transformation of E. coli strain XL1b and screening of positive clones. All constructs were verified by sequencing before downstream applications.
MPC complementation in S. cerevisiae. The empty vector (pJR3455A) and indicated MPC constructs (ScMPC1/2 and TgMPC1/2) were transformed into the ScDmpc1 or ScDmpc2 mutants following a protocol for yeast culture and transformation described previously (39). Briefly, the ScDmpc1 or ScDmpc2 mutant was grown in 2% yeast extract, 1% peptone, and 2% glucose. Following transformation with the plasmids described above, all yeast strains were cultured in synthetic dropout (uracil-free) minimal medium (0.67% yeast nitrogen base; Difco) supplemented with appropriate amino acids and 2% glucose. The transfectants were cloned on selective plates and tested for growth complementation in the presence or absence of leucine and valine (30°C for 3 days on plates or up to 32 h in liquid culture).
Construction of transgenic parasites. To insert an HA tag at the C terminus of MPC1 and MPC2, the cassette containing the smHA epitope and DHFR * selection marker was amplified using the pSL24m-Linker-smFP-DHFR-LoxP-T7 plasmid (22) and primers with arms (50 bp) homologous to the target genes. The amplicons were cotransfected into purified tachyzoites of the RH Dku80 strain along with the corresponding locus-specific CRISPR plasmid. Transfected parasites were selected with 1 mM pyrimethamine (Sigma-Aldrich, USA), and stable transfectants were examined by immunofluorescence assay. To construct the Dmpc1, Dmpc2, and Dbckdh mutants, corresponding homology templates (MPC1::DHFR, MPC2::DHFR, or BCKDH::DHFR) were transfected into tachyzoites of the RHDhxgprt strain together with gene-specific gRNA constructs. Transgenic parasites were selected with 1 mM pyrimethamine and cloned by limiting dilution, followed by diagnostic PCRs (PCR1, PCR2, and PCR3, as indicated in the figures). The Dmpc1-Dmpc2 double deletion mutant was constructed in two steps. First, the DHFR* selection marker (flanked by loxP sites) in the Dmpc2 strain was removed by transfecting a plasmid expressing the Cre recombinase (pmin-Cre-eGFP) (40). The Dmpc2 clones lacking the DHFR* marker were used to further delete the MPC1 gene, as described above. The complementation strain CoMPC1 was constructed by inserting an MPC1 expression cassette at the UPRT locus of the Dmpc1 strain. Briefly, the pTub::MPC1-HA-CAT amplicon was generated from the pTub::MPC1-HA-CAT plasmid and transfected into tachyzoites along with the pSAG1:: Cas9-U6::sgUPRT plasmid. Subsequently, the parasites were selected with 30 mM chloramphenicol (Sigma-Aldrich, USA) (38).
In vitro phenotyping of parasite mutants. The plaque and replication assays to determine the overall growth and proliferation fitness of tachyzoites were performed following the protocols reported previously (41). To evaluate the bradyzoite differentiation efficiencies of the ME49 and ME49Dmpc1 strains, parasites were cultured for 3 days in alkaline RPMI 1640 medium supplemented with 1% FBS and 50 mM HEPES (pH 8.2, ambient CO 2 ). Samples were fixed by 4% paraformaldehyde, permeabilized in 0.1% Triton X-100 Parasite proliferation and virulence assays in mice. To assess the parasite virulence, 8-week-old female ICR mice were infected with purified tachyzoites through intraperitoneal (i.p.) injection. Each strain was tested in 5 to 25 mice (depending on the strain tested) with a dose of 100 tachyzoites/animal (38). The mice were monitored daily for their survival over a period of 30 days. To determine the proliferation of parasites, 1 Â 10 5 tachyzoites of indicated strains were injected (i.p.) into mice. Five days postinfection, the peritoneal liquid was examined for the parasite burden by quantitative PCR (qPCR) of the T. gondii glyceraldehyde-3-phosphate dehydrogenase 1 (TgGAPDH1) (primers in Table S1). All animal experiments were approved by the ethical committee of Huazhong Agricultural University (permit number HZAUMO-2019-095).
Metabolic labeling and measurements. To monitor the carbon flux through glycolysis and the TCA cycle, 3 Â 10 7 tachyzoites were incubated in DMEM containing 8 mM [ 13 C 6 ]glucose (37°C for 4 h). Subsequently, the parasites were washed once with glucose-free DMEM and twice with phosphate-buffered saline (PBS) and then resuspended in 80% methanol. Samples were lysed by ultrasonication (5 cycles of 1 min each with 1-min intervals on an ice-water interface) in a bath sonicator, incubated for 30 min at 220°C, and centrifuged (15,000 Â g for 15 min at 4°C). The supernatant of each sample was collected for further analysis to determine the incorporation of 13 C into glycolytic and TCA cycle intermediates. The supernatant (1 mL) was evaporated to dryness and then reconstituted in 50 mL of 50% aqueous acetonitrile (1:1 [vol/vol]) prior to ultra-high-performance liquid chromatography-high-resolution mass spectrometry (UHPLC-HRMS) analysis.
Chromatographic separation was performed on a ThermoFisher UltiMate 3000 UHPLC system using a Waters BEH amide column (2.1 mm by 100 mm, 1.7 mm). The injection volume was 2 mL, and the flow rate was adjusted to 0.35 mL/min. The mobile phases consisted of water with 15 mM ammonium acetate (pH 8.5, phase A) and acetonitrile in water (90:10 [vol/vol], phase B). A linear-gradient elution with the following program was performed: 0 to 2 min, 90% B; 14 min, 75% B; 15 min, 65% B; 15.2 to 16.9 min, 50% B; 17 to 20 min, 90% B. Eluents were analyzed on a ThermoFisher Q Exactive hybrid quadrupole-Orbitrap mass spectrometer (QE) set in heated electrospray ionization-negative (HESI 2 ) mode. The spray voltage was set to 3,500 V, while the capillary and probe heater temperatures were kept at 250°C and 300°C. The sheath and auxiliary gas flow rates and the S-lens radiofrequency (RF) level were adjusted to 35 arbitrary units (AU), 10 AU, and 50 AU, respectively. The instrument was operated in a full-scan high-resolution mode (70,000 full width half maximum [FWHM]; m/z, 200) in a range of 70 to 1,050 m/z with the automatic gain control (AGC) target setting at 1 Â 10 6 .
Acetyl-CoA was analyzed by LC-MS, performed using the 6500 plus Qtrap mass spectrometer (AB Sciex, USA) coupled to the Acquity UPLC H class system (Waters, USA). An Acquity UPLC HSS T3 column (2.1 by 100 mm, 1.8-mm particle size; Waters) was employed, using water with 5 mM ammonium acetate as mobile phase A and methanol as mobile phase B. The linear gradient was set as follows: 0 min, 0% B; 1.5 min, 0% B; 6 min, 95% B; 7.4 min, 95% B; 7.5 min, 0% B; and 10 min, 0% B (flow rate, 0.3 mL/min). The column chamber and sample tray were kept at 35°C and 10°C, respectively. The mass data were acquired in multiple reaction monitor (MRM) mode for acetyl-CoA, acetyl-CoA (*M2) and acetyl-CoA (*M3) with transitions of 810/303, 812/305 and 813/306 in positive mode. The ion transitions were optimized using the chemical standards. The nebulizer gas (Gas1), heater gas (Gas2), and curtain gas were set at 55, 55, and 30 lb/in 2 , respectively. The ion spray voltage was 5,500 V, and the optimal probe temperature was set at 550°C. The Sciex OS1.6 software was applied for metabolite identification and peak integration.
Protein sequence analysis. Protein sequences were retrieved from NCBI (https://www.ncbi.nlm.nih .gov). The multiple-sequence alignment was performed using ClustalX2 and the following sequences: T. Data presentation and statistics. All assays were performed at least three times independently, unless stated otherwise. Statistical analyses were executed in GraphPad Prism 8.0.2 (GraphPad Software, Inc., USA) using two-way analysis of variance (ANOVA) with the Bonferroni posttest or Student's t test. The data in graphs depict the mean values 6 standard errors of the means (SEM) or standard deviations (SD), as indicated in figure legends (*, P # 0.05; **, P # 0.01; ***, P # 0.001). The plaque and immunofluorescence images shown are representative of multiple experiments.
Data availability. All data generated or used in this study are presented in the paper and the supplemental material.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 1.8 MB.
(grant number M0074) endowed to B.S. and N.G. by the Sino-German Center for Research Promotion (CDZ). The funders had no role in the design, data collection, analysis, preparation, or decision to publish this work. B. Shen and N. Gupta designed the research and acquired funding. C. Lyu, Y. Chen, Y. Meng, and I. EI-Debs performed the experiments. C. Lyu, J. Yang, S. Ye, and Z. Niu provided methods. C. Lyu, N. Gupta, and B. Shen analyzed the data. C. Lyu, N. Gupta, and B. Shen wrote the paper. All authors have read and approved the final version of the paper.
The authors have no conflict of interest to declare.