Dynamic Change of Intracellular Metabolism of Microglia Evaluated by Transcriptomics in an Alzheimer’s Mouse Model

Abstract

Background:

Microglia play diverse roles in Alzheimer’s disease (AD). Intracellular metabolism has been indicated an important factor in modulating the function of microglia. However, it is not clear whether the intracellular metabolism of microglia changes dynamically in different stages of AD.

Objective:

To determine whether microglia intracellular metabolism changes dynamically in different stages of AD.

Methods:

Microglia were extracted from APP^Swe/PS1^dE9 (APP/PS1) mice and wild-type littermates at 2, 4, and 8 months old by fluorescence-activated cell sorting and used for RNA-sequencing analysis and quantitative PCR. Morphologies of amyloid plaques and microglia were detected by immunofluorescence staining.

Results:

Compared with control littermates, the microglia of APP/PS1 mice exhibited significant transcriptional changes at 2-month-old before microglia morphological alterations and the plaque formation. The changes continued drastically following age with defined morphological shift of microglia and amyloid plaque enhancement in brains. Further analysis of those genotype and age dependent transcriptomic changes revealed that differentially expressed genes were enriched in pathways related to energy metabolism. Compared with wild-type mice, there were changes of some vital genes related to glucose metabolism and lipid metabolism pathways in APP/PS1 mice at different ages. Glucose metabolism may play a major role in early activation of microglia, and lipid metabolism may be more important in later activation period.

Conclusion:

Our results showed that microglia actively participate in the pathological progress of AD. The intracellular metabolism of microglia changed significantly in different stages of AD, even preceding amyloid-β deposition.

Keywords

Alzheimer’s disease metabolism microglia transcriptomics

INTRODUCTION

Alzheimer’s disease (AD) is one of the most common neurodegenerative diseases threatening the lives of tens of millions of people [1, 2]. However, effective treatments to improve patients’ prognosis are still lacking to date [2].

Microglia, as immune cells of central nervous system (CNS), play important roles in recognizing exogenous and endogenous attacks, triggering immune response, defending pathogens and other injuries, and maintaining the homeostasis of CNS internal environment [3 –5]. The intensive involvement of microglia in AD pathogenesis has been confirmed by both clinical and animal studies [6 –9]. In the course of AD, microglia play complex functions. On one hand, microglia have been found to promote the clearance of amyloid-β (Aβ) and necrotic neurons [6, 10]. On the other hand, microglia can exacerbate neuron damage [11, 12] by directly or indirectly phagocytizing synapses [13], aggravate tau pathology [14], and secret a variety of inflammatory factors [15], resulting in further deterioration of neurocognitive function [16 –18]. However, it remains unclear how these diverse functions transform in microglia following the progression of AD pathology.

It is well known that the maintenance of cell function is closely related to its metabolic state. Research has found a novel subtype of microglia in AD by single-cell RNA sequencing (RNA-seq) named disease-associated-microglia (DAM). DAMs are phagocytic cells located near Aβ plaques; interestingly, they express high level of lipid metabolism pathways [18]. Perturbed energy metabolism of microglia in AD has also been demonstrated recently [19]. Therefore, a systematic understanding of the current cellular metabolic status of microglia in different morphological/functional changes can help understand the reasons behind the functional changes of microglia.

RNA-seq technology is used to quickly and com-prehensively compare the changes of cell transcription level. Many bulk RNA-seq technologies, and even single cell RNA-seq technologies, have been used to detect and observe the function and metabolism of microglia in AD [18 , 21]. And in these studies, RNA-seq analysis of microglia have been performed in geriatric and young mice to identify the transcriptomic alterations in AD mouse models. However, they did not pay attention to the dynamic changes of energy metabolism related pathways of microglia during the whole process of plaque formation.

Therefore, in the present study, we report micro-glia morphology and Aβ plaques distribution in APP^Swe/PS1^dE9 (APP/PS1) mice at different ages. More importantly, we demonstrate the transition of microglia gene expression profiles, with special attention on energy metabolism alongside the progression of Aβ pathology in APP/PS1 mice and wild-type littermates.

MATERIALS AND METHODS

Animals

APP/PS1 male mice (The Jackson Lab, NO. 004462) and their wild-type littermates were sacrificed at the age of 2, 4, and 8 months old. All animals were housed in standard specific pathogen free (SPF) conditions with access to sterilized water and food (Beijing Keao Xieli Feed Co., LTD.) ad libitum. Animal handling and experiment procedures were performed in accordance with the guide for the care and use of laboratory animals by Medical Experimental Animal Administrative Committee of Fudan University. All efforts were made to reduce the number of animals and minimize the animals’ suffering.

Isolation of microglia

Three mice in each group were sacrificed for mi-croglia isolation and RNA-seq detection. The micro-glia from different mice in the same group were not pooled. Therefore, the microglia from each mouse were used as an independent sample for cell enrichment, library construction, subsequent sequencing, and analysis.

Mice were anesthetized by intraperitoneal injection of 1%pentobarbital sodium and perfused with sterile ice-cold phosphate buffered solution (PBS). After perfusion, bilateral cerebral cortex and hippocampus were obtained for microglia extraction. Cell suspension were obtained by mechanical separation and magnetic bead demyelination [22]. That is, tissue was firstly transferred to a 3.5 cm dish filled with solution consisting of Liberase TL (Roche), DNase I (Invitrogen), and RNase inhibitor (Promega), and dissociated with a scalpel. The dish was then incubated in 37°C for 30 min. Then, enzymatic reaction was ended with 20μL 0.5 M Ethylene Diamine Tetraacetic Acid (EDTA). The solution was pipetted repeatedly and filtered with 70μm cell strainer (Falcon) to obtain a single-cell suspension, and spun at 300 g, 4°C for 10 min. After discarding the supernatant, magnetic-activated cell sorting (MACS) buffer of 1.8 mL was added to the precipitation. After a gentle blending, the liquid was transferred to a new 15 mL centrifuge tube. Myelin removal beads (Miltenyi Biotec) and RNase inhibitor was added. Then the mixture solution was incubated at 4°C for 15 min and centrifuged at 300 g, 4°C for 10 min. The pellet was collected, resuspended, and passed through LD column (Miltenyi Biotec). The negatively sorted acquisition was a mixture of cells.

Afterwards, microglia from the mixture cells were enriched via specific surface antibody expressed by microglia and fluorescence-activated cell sorting (FACS) technology [23] Mixed cell suspensions were centrifuged at 500 g, 4°C for 10 min. Pelleted cells were resuspended in FACS buffer and blocked with CD16/32 to reduce non-specific binding. Then, the cells were stained with FITC-CD11b (1:100, Biolegend), PE/CY7-CD45 (1:100, Biolegend) and APC-Ly6c (1:100, Biolegend). DAPI was added before loading.

Next, the BD FACS Aria II (BD Bioscience) was used for microglia sorting. After antibody incubation, the “total cells” were first separated from cell debris and remaining myelin sheath through a gate in side scatter (SSC) and forward scatter (FSC). In order to obtain high quality RNA samples, “living cells” were then selected as DAPI-negative cells from the “total cells”. Because of the high expression of CD11b and moderate expression of CD45 in microglia, we screened out the “corresponding microglia population” from “living cells” through these two antibodies. We then screened out “the corresponding microglia population with Ly6C negative” from “the corresponding microglia population” by using the Ly6C antibody which only expressed in macrophage whereas not in microglia. The cells labeled with “CD11b⁺CD45^{^lo}Ly6c^-” were enriched and named as “positive cells” for subsequent RNA-seq (83,000 cells/per mouse) and post-sorting cell validation (5,000 cells/per mouse). We also enriched “CD11b^-” cells and named them as “negative cells” for post-sorting cell type validation. The enriched “positive cells” for RNA-seq were collected into a 1.5 mL centrifuge tube containing 750μL TRIzol LS (Life). Cells for post-sorting validation were collected into a sterile PCR tube containing 10μL resuspension buffer and 1μL lysis enhancer (cells direct one step for QRT PCR, Invitrogen). All cell samples were frozen on dry ice and stored at –80°C until further processing.

RNA extraction, amplification, and library generation

The RNA in TRIzol LS was extracted by RNeasy Micro Kit (Qiagen). The integrity of the RNA was evaluated by Agilent 2100 Bioanalyzer. For the qualified RNA samples, the initial loading amount was adjusted to 1 ng, and the Single Cell Full length mRNA Amplication kit (Vazyme) was used to reverse transcription and amplification. Agilent 2100 bioanalyzer was then used to evaluate the quality and concentration of the amplified cDNA. The qualified cDNA after amplification should be a large single peak distributed in 400–10000 bp. After adjusting the amount of initial loading of cDNA to 1 ng, the library was generated by using TruePrep^TM DNA Library Prep Kit V2 (Vazyme) for Illumina.

RNA sequencing

RNA sequencing was completed in Beijing Novogene Technology Co., Ltd. Raw reads in FastQ format were first processed by internal Perl scripts. In this step, clean data was obtained by deleting reads containing adapter, reads containing ploy-n, and low-quality reads from the original data. The contents of Q20, Q30, and GC were also calculated. The data should meet the standards of Q20 > 85%, Q30 > 80%, and GC content in the range of 45%–55%at the same time. All downstream analyses were based on high quality clean data. Hisat2v2.0.5 was used to construct the reference genome index, and Hisat2v2.0.5 was used to align the paired clean reading with the reference genome. Featurecounts v1.5.0-p3 was used to calculate the number of reads for each gene. Then the Fragments Per Kilobase per Million (FPKM) of each gene was calculated according to the length of each gene, and the gene was read and counted. DESeq2 p value < 0.05 |log2 Fold Change |> 0.0 was used to calculate the difference of gene expression. To predict the cellular and metabolic functions associated with the observed changes in transcript levels, the differentially expressed genes (DEGs) were categorized according to predicted protein function using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (http://www.genome.jp/kegg/). A Gene Ontology (GO) enrichment analysis was also performed, and all DEGs were mapped to GO terms in the Gene Ontology database (http://www.geneontology.org/). Gene set enrichment analysis (GSEA) was performed with gene score enrichment analysis (http://www.broadinstitute.org/gsea/index.jsp).

Single cell qRT-PCR

Four tubes of negative cells and four tubes of positive cells were randomly selected for verification after sorting. Regarding the limited cell number and RNA contained, the standard procedure will not provide with enough RNA for subsequent PCR validation. Single-cell PCR technology was adapted from published literature [24]. In detail, the samples were incubated at 75°C for 10 min and then chilled to 4°C for 5 min. Then, 10μL 2×Reaction Buffer Mix and 2μL RT Enzyme Mix (Superscript III first-strand synthesis Supermix for qRT-PCR, Invitrogen) were added and mixed. To synthesize cDNA, the samples were incubated at 25°C for 10 min, 50°C for 30 min, 85°C for 5min, 4°C for 5 min sequentially. To remove RNA, 1μL RNase H was added and incubated at 37°C for 20 min. To validate the expression of microglia-specific markers (Cx3cr1, Iba1, Tmem119, and Trem2), quantitative Real-time PCR (qRT-PCR) was performed, and 2μL of the samples were used as template for 60–70 cycles of amplification [24]. Primers used and corresponding sequence were listed in Table 1. The changes in the gene expression of the microglia markers between positive and negative cells were normalized using the reference gene (Actin) expression results. The relative gene expression was calculated by the 2–ΔΔCT method.

Table 1

Reagents used and corresponding information

Gene	Forward primer	Reverse primer	Commercial
Tmem119	GTGTCTAACAGGCCCCAGAA	AGCCACGTGGTATCAAGGAG	BioSune Biotechnology (ShangHai) Co.
Trem2	CTGGAACCGTCACCATCACTC	CGAAACTCGATGACTCCTCGG	BioSune Biotechnology (ShangHai) Co.
Cx3cr1	CAGCATCGACCGGTACCTT	GCTGCACTGTCCGGTTGTT	BioSune Biotechnology (ShangHai) Co.
Iba1	ATCAACAAGCAATTCCTCGATGA	CAGCATTCGCTTCAAGGACATA	BioSune Biotechnology (ShangHai) Co.
Actin	AGAAGGACTCCTATGTGGGTGA	CATGATCTGGGTCATCTTTTCA	BioSune Biotechnology (ShangHai) Co.
Antibody	Full name	Isotype	Cat	Conc.	Commercial
6E10	Purified anti-β-amyloid, 1–16	mouse IgG1	83003	1mg/ml	Biolegend
Iba1	Anti Iba1, Goat	goat IgG	ab5076	0.5mg/ml	Abcam
DAPI	DAPI ready-made solution	N/A	28718-90-3	1mg/ml	Sigma
Alexa Fluor 488 Donkey-anti-Mouse IgG		donkey IgG	A11055	2mg/ml	Invitrogen
Alexa Fluor 555 Donkey-anti-Mouse IgG		donkey IgG	A32773	2mg/ml	Invitrogen

Immunofluorescent staining

Three mice in each group were sacrificed for immunofluorescent staining. Mice were anesthetized and perfused with sterile ice-cold PBS and 4%polyformaldehyde. Brains were removed, dehydrated by sucrose solutions and coated by Optimal Cutting Temperature (O.C.T) compound (SAKURA). A serial of 30μm slices were dissected coronally from olfactory bulb to cerebellum. Representative sections of slices (bregma –0.95 mm to bregma –3.79 mm) were selected and stained. Brain slices were incubated with 6E10 antibody (1:500, BioLegend) for amyloid staining and Iba1 (1:500, Abcam) for microglia staining at 4°C overnight after blocking with 3%donkey serum. Slices were rinsed three times with PBS the next day and incubated with fluorescent Alexa Fluor 555 Donkey-anti-Mouse IgG (1:2000, Invitrogen) and Alexa Flour 488 Donkey-anti-Goat IgG (1:2000, Invitrogen) secondary antibodies accordingly. Slices were then washed as described above and stained with DAPI. Pictures were captured by confocal microscope (Nikon AIR-MP). Three microscopic fields were randomly captured in each slice with the same reference position. The plaque area was calculated on 10 slices per animal in a blind manner. The plaque area was measured by Image-Pro plus software (Mediacybernetics, Acton, MA, USA) and recorded as percentage of area occupied by plaques per field. FIJI software was used to analyze microglia morphology, and the microglia process length/cell and number of endpoints/cell were quantified according to the protocol [25].

Statistical analysis

Results were presented as mean±standard errors of the mean (SEM). Statistical analysis was performed by the GraphPad Prism7. Comparisons be-tween two groups were performed using two-tailed unpaired t test. One-way analysis of variance (ANOVA) with Tukey’s post hoc test was used to compare three or more independent groups. p-values less than 0.05 were considered statistically significant.

RESULTS

Enhancement of microglia activation and amyloid plaques at different ages

The APP/PS1 and age-matched littermate mice at three time points (2, 4, and 8 months) were studied. Representative confocal images of cortex with enlarged picture of microglia were shown in Fig. 1A. No Aβ plaque deposition was detected in brains of the wild-type mice. There was an overall increase in total plaque area (or plaque burden) and activated microglia with age in the APP/PS1 mice, and no significant Aβ plaque formation or microglia activation was observed at the age of 2 months. At the age of 4 months, a few activated microglia were found around the plaques. At the age of 8 months, microglia were significantly activated and found largely around senile plaques, with shorter and thicker processes, fewer ramified processes and swollen or elongated cell bodies (Fig. 1A). Statistical analysis confirmed the significant increase of plaque formation (Fig. 1B). As to the statistics of microglia morphology, we quantified the number of endpoints/cell (Fig. 1C) and microglia process length/cell (Fig. 1D) and found that the number of microglia process endpoints/cell and process length/cell was significantly decreased with age.

Fig. 1

Aβ plaques formation and microglia activation following different age in APP/PS1 mice and wild-type littermates. A) Representative confocal images showing microglia (stained with Iba1) and Aβ plaques (stained with 6E10) in APP/PS1 and wild-type mice of 2, 4, and 8 months old. Scale bar 20μm. B) Quantification of area occupied by Aβ plaques in the APP/PS1 mice of 2, 4, and 8 months old. With age enhancement, the number of Aβ plaques in the APP/PS1 mice increased significantly. C-D) Quantification of microglia process endpoints/cell and process length in the APP/PS1 and wild-type mice of 2, 4, and 8 months old. Microglia process endpoints/cell and length/cell was significantly decreased in APP/PS1 mice with age. Data are presented as means±SEM, n = 3. ^**p < 0.01; ^***p < 0.001, ^****p < 0.0001.

Morphology is usually the downstream manifestation of molecular biological changes. In order to find out the underlying molecular biological changes in different pathological periods, microglia of APP/PS1 and wild-type mice at different ages were extracted, and gene expression was analyzed by RNA-seq.

Expression changes of genes in the APP/PS1 mice

In order to investigate the changes of microglia from molecular level in different ages and pathological status, we isolated microglia from APP/PS1 mice and wild-type mice at 2, 4, and 8 months old separately and analyzed their transcriptomic profiles. Those are key time points represents AD pathological stages of before and after plaque formation. The extraction process of microglia is showed in Fig. 2A. The expression of microglia-specific markers was significantly higher in the collected cells than that in negative cells (Fig. 2B).

Fig. 2

Microglia extraction and validation. A) Schematic diagram of the microglia dissociation process. Cell suspensions of mouse brain samples were got after mechanical disruption and enzyme treatment. After removal of debris by MACS, microglia were purified by FACS. B) The levels of microglia-specific markers were significantly high in the collected cells as compared with that in negatively sorted cells. The y-axis represents normalized expression level from qRT-PCR. Error bars show SEM, n = 4. ^*p < 0.05, ^**p < 0.01 Student’s t test. C) Confocal images showing Iba1 (microglia) and DAPI (nucleus) staining of post-sorting cells. Scale bar 10μm. FACS, Fluorescence-activated cell sorting; MACS, Magnetic-activated cell sorting; qRT-PCR, Quantitative Real-time PCR.

RNA sequencing of the collected cells of three mice at each timepoint per subgroup was performed, and the DEGs were identified by comparing the APP/PS1 mice with age-matched wild-type littermates. The results showed that there were 1,028 upregulated and 811 downregulated genes in 2-month-old APP/PS1 mice as compared to wild-type littermates (Fig. 3A), 667 upregulated and 776 downregulated at 4 months old (Fig. 3B), and 1,218 upregulated and 1,172 downregulated at 8 months old (Fig. 3C). The DEGs between different groups were further analyzed. Most DEGs were unique to certain age group: there were 634 DEGs only upregulated in 2 months groups, 555 DEGs only at 4 months, and 868 DEGs only at 8 months groups (Fig. 3D, left panel). Consistently, the number of non-overlapping downregulated DEGs were 416, 648, and 798 in 2-, 4-, and 8-months groups respectively (Fig. 3D, right panel). There were only 18 upregulated and 37 downregulated DEGs were shared by all three age groups. These data indicated that gene expression changes were uniquely genotype- and age-dependent.

Fig. 3

The identification of DEGs in microglia between the APP/PS1 mice and age-matched wild-type littermates. A-C) Volcano plot of DEGs of microglia in 2, 4, and 8 months old of the APP/PS1 mice and wild-type littermates respectively. Numbers in plot indicated the expressions of genes that were significantly upregulated, downregulated, or unchanged. Top five highly significantly genes are labelled. D) Venn diagram showing the overlap of upregulated (left panel) or downregulated (right panel) genes in microglia between 2-, 4-, and 8-month-old APP/PS1 mice and wild-type littermates. DEGs, Differentially expressed genes.

Involvement of biological pathways in microglia of the APP/PS1 mice

As shown in Fig. 4A, the enriched KEGG pathways were mainly associated with metabolism and proliferation-associated genes as compared APP/PS1 mice with wild-type mice at 2 months old, including PI3K-Akt signaling pathway, Rap1 signaling pathway, MAPK signaling pathway, EGFR tyrosine kinase inhibitor resistance, and mitophagy-animal. The GO enrichment analysis indicated that enriched biological processes (BPs) were mainly centered on angiogenesis and chemotaxis (Fig. 4B). At 4 months old, the enriched KEGG pathways were mainly related to immune reactions (Fig. 4C), and enriched BPs were centered on inflammatory response and cytokine (Fig. 4D). At 8 months old, KEGG pathways related to metabolism and proliferation including MAPK signaling pathway and various cancer pathways were also enriched, and the enriched BPs were mainly centered on protein modification and various metabolic regulations. These data imply that energy metabolism may be dynamically changed. As energy metabolism is important for immune cells to perform multiple functions, analysis was then focused on the genes related to energy metabolism at different ages.

Fig. 4

The enriched pathways of DEGs by KEGG and GO analysis in the APP/PS1 mice and wild-type littermates. A, C, E) The DEGs enriched pathways by KEGG analysis as compared APP/PS1 mice with wild-type littermates at 2 months old (A), 4 months old (C), and 8 months old (E). GeneRatio is the ratio of the number of DEGs to the total number of genes in a certain pathway. The color and size of the dots represent the range of the -log10 (p value) and the number of DEGs mapped to the indicated pathways, respectively. The figure shows the top 20 enriched pathways. B, D, F) The DEGs enriched pathways by GO analysis as compared the APP/PS1 mice with wild-type littermates at 2 months old (B), 4 months old (D), and 8 months old (F). The y-axis indicates the name of biological processes, and bar show -log10 (p value). The figure shows the top 20 enriched biological processes. DEGs, differentially expressed genes; KEGG, Kyoto Encyclopedia of Genes and Genomes; GO, GeneOntology.

Glucose metabolism related genes in microglia change before plaque formation in the APP/PS1 mice compared with wild-type mice

As reported, glycolysis and oxidative phosphorylation are crucial to maintain microglia function, especially in immune response [26]. So, we summarized DEGs related to oxidative and glycolytic energy metabolism (Table 2). As shown in Fig. 4, before Aβ plaque formation, the profile of DEGs has indicated changes in various metabolic pathways. Compared with wild-type mice, the expression of glucose transporter Slc2a3 was upregulated in the APP/PS1 mice at the age of 2 months, i.e., before plaque formation, as well as many other DEGs related to glucose metabolism. This indicates that glucose metabolism may be already changing before microglia activation and plays different roles in different stages of microglia activation, which needs further study.

Table 2

Differentially expressed genes (DEGs) related to glucose metabolism comparing APP/PS1 to wild-type littermates at 2, 4, and 8 months

Gene	log2FoldChange	p value	Function
2M
Slc2a3	1.476101808	3.7061E-06	Facilitative glucose transporter [38]
Txnip	–0.654617628	0.000926007	Mediate oxidative stress as a transcription repressor [39]
Nfkb1	–0.472510045	0.007127765	pleiotropic transcription factor [40]
Gapdh	–0.968477099	0.007830307	Catalyzing the first step of glycolysis [41]
Pck2	–0.508505136	0.008098957	Catalyzes the conversion of oxaloacetate to PEP
Pgam1	–0.492134292	0.019843655	Interconversion of 3- and 2-PHG with 2,3-BPG
Aldoc	0.513744297	0.024592733	Synthesizes D-glyceraldehyde 3-phosphate [42]
Pfkl	–0.441554623	0.024880208	Catalyzes the phosphorylation of D-fructose 6-phosphate [43]
Pgm2l1	0.915989244	0.028238599	Synthesizes glucose 1,6-biphosphate [44]
Ldhb	–0.363937958	0.031012731	Synthesizes (S)-lactate from pyruvate
Slc25a1	–0.413577867	0.035034595	Citrate transporter [45]
4M
Pfkfb3	–0.759367304	7.24143E-06	Synthesis and degradation of fructose 2,6-BP
Sod1	–0.759692865	0.000305798	Destroys cellular radicals
Eno3	–0.808061698	0.000445806	Synthesizes pyruvate from D-glyceraldehyde 3-phosphate
Gapdh	–2.426940983	0.002215664	See above
Hk1	–0.63298537	0.002405624	Catalyzes phosphorylation of hexoses. Involved in innate immunity and inflammation [46]
Nfkbia	–0.421814923	0.003838265	Inhibits the activity of dimeric NF-kappa-B/REL complexes [47]
Hk3	–0.442665876	0.031679998	Catalyzes the phosphorylation of hexose [48]
8M
Cox6a2	1.747698011	5.96278E-09	Component of the cytochrome c oxidase
Ndufs6	0.583049652	0.000707909	Accessory subunit of the mitochondrial Complex I
Slc2a1	0.801352211	0.000835367	Facilitative glucose transporter [49]
Nfkb1	–0.427440556	0.002033065	See above
Sdha	0.456369353	0.006563523	Flavoprotein subunit of succinate dehydrogenase that is involved in the mitochondrial complex II [50]
Gapdh	–0.834914611	0.013265652	See above
Ndufa1	0.485305323	0.014975494	Catalytical subunit of the mitochondrial Complex I [51]
Pgk1	0.392696975	0.023779092	Catalyzes conversion of 1,3-DPG to 3-PG [52]
Hk2	0.302020622	0.024132627	Catalyzes the phosphorylation of hexose [53]
Slc25a1	–0.375066679	0.028974393	See above
Aldoa	0.355080319	0.030292151	Synthesizes D-glyceraldehyde 3-phosphate and glycerone phosphate from D-glucose.
Txnip	–0.290196565	0.045018003	See above

PEP, phosphoenolpyruvate; BPG, bisphosphoglycerate; BP, bisphosphate; DPG, diphosphoglycerate; PG, phosphoglycerate. FoldChange: The ratio of average read count in APP/PS1 mise verses wild-type mice.

Dynamic regulation of lipid metabolism pathways of microglia in the APP/PS1 mice compared with wild-type mice

Lipid is another important energy source of microglia. In order to identify the lipid metabolism of microglia between APP/PS1 and wild-type mice with age, GSEA was carried out. No pathway relative to lipid metabolism was found at 2 months old. GSEA indicated that positive regulation of lipid transport, cholesterol transport, and sterol transport pathways were downregulated in APP/PS1 mice microglia compared with wild-type littermates at 4 months of age (Fig. 5A). As to 8 months old, acylglycerol homeostasis pathway and triglyceride homeostasis pathway was upregulated, and regulation of fatty acid β oxidation pathway was downregulated in APP/PS1 mice compared with wild-type mice (Fig. 5B).

Fig. 5

GSEA enrichment plots of representative gene sets that were significantly related to lipid metabolism. A) The positive regulation of lipid transport, cholesterol transport and sterol transport pathways gene sets were enriched in downregulated genes (loss of function in APP/PS1 mice compared with wild-type littermates) at 4 months old. B) Acylglycerol homeostasis pathway and triglyceride homeostasis pathway were enriched in upregulated genes, while regulation of fatty acid β oxidation pathway was enriched in downregulated genes in APP/PS1 mice compared with wild-type mice at 8 months old. GSEA, gene set enrichment analysis.

Changes of energy metabolism related pathways in microglia at different ages

Changes of glucose and lipid metabolism in the mi-croglia of APP/PS1 mice compared to wild-type mice at different ages were investigated and describe above. It is known that a multitude of functionally diverse microglia populations exists in a dynamic equilibrium in brain, which is constantly changing with age [27]. So, we then asked whether microglia of APP/PS1 mice at different ages have differential expression of genes related to energy meta-bolism. When comparing the 2-month-old APP/PS1 mice with 4-month-old APP/PS1 mice, pathways related to energy metabolism were centered on glyco-metabolism, such as glucose transmembrane transporter activity, monosaccharide transmembrane transporter, and hexose transmembrane transporter activity (Fig. 6A). Interestingly, when comparing the 4-month-old APP/PS1 mice with 8-month-old APP/PS1 mice, pathways related to energy metabolism were centered on lipid metabolism, such as acylglycerol homeostasis pathway, triglyceride homeostasis pathway, and cholesterol transporter activity pathway (Fig. 6B). That is, with increasing activation of microglia, the various glucose transmembrane transporter activities were strengthened, while with a large number of microglia activated, pathways related to lipid metabolism were enhanced.

Fig. 6

GSEA enrichment plots of representative gene sets that were significantly related with energy metabolism among APP/PS1 mice. A) The glucose transmembrane transporter activity, monosaccharide transmembrane transporter activity and hexose transmembrane transporter activity gene sets were enriched in downregulated genes in 2-month-old APP/PS1 mice compared with 4-month-old APP/PS1 mice. B) Acylglycerol homeostasis pathway, triglyceride homeostasis pathway and cholesterol transporter activity pathway were enriched in downregulated genes in 4-month-old APP/PS1 mice compared with 8-month-old APP/PS1 mice. GSEA, gene set enrichment analysis.

Fig. 7

GSEA enrichment plots of representative gene sets that were significantly related with energy metabolism among wild-type mice. The acetyl-CoA biosynthetic process, citrate metabolic process, and tricarboxylic acid cycle gene set were enriched in upregulated genes in 2-month-old wild-type mice compared with 8-month-old wild-type mice. GSEA, gene set enrichment analysis.

It is reported that microglia showed different functional status with age in wild-type mice, such as immune response and cytokine, etc. [20], while whether functional changes are accompanied by changes in energy metabolism pathways have not been studied. In order to explore the energy metabolism changes in wild-type mice, and to observe whether the changes were consistent with APP/PS1 mice, we further analyze the GSEA related with energy metabolism in wild-type mice. No significant difference in microglia energy metabolism was seen between 4-month-old wild-type mice and 2-month-old wild-type mice, and between 4-month-old wild mice and 8-month-old wild mice. Acetyl-CoA biosynthetic process, citrate metabolic process, and tricarboxylic acid cycle were upregulated in 2-month-old wild-type mice compared with 8-month-old wild-type mice (Fig. 7), However, no significant changes in lipid metabolism were observed between the two groups. As for mice, young individuals were represented by 2-month-olds and middle-age individuals were represented by 8-month-olds. Therefore, we speculate that aging may affect the energy metabolism of microglia in wild-type mouse, especially the decrease of the aerobic oxidation process of glucose metabolism. The metabolic changes of microglia in wild-type mice from 2 months to 8 months were significantly different from those in APP mice from 2 months to 8 months.

DISCUSSION

The role of microglia playing in AD is drawing more and more attention these years. Microglia are known to undergo not only morphological changes but more importantly redirection of physiological functions and induction of activating inflammatory pathways, which may contribute to disease progression [16]. It is not surprising that inflammatory response should be strictly controlled to prevent damage caused by excessive cascade reaction. However, how to regulate microglia effectively remains unclear. This is due to the complexity of microglia function and specificity at different ages. More importantly, the underlying mechanisms regulating the remodulation of its function are still unknown. Answers to these questions may help to find intervention points. In our study, we found that microglia displayed a defined morphological shift over time, and gene expression changes of microglia were uniquely genotype and age-dependent. Moreover, many pathways related with energy metabolism were enriched with age. Glucose metabolism and lipid metabolism may play distinct roles in different activation state of microglia, which provides important evidence for the modification of microglia function.

Our transcriptomic results demonstrated that microglia in APP/PS1 mice exhibit different mRNA profiles compared to their wild-type littermates at different stages of senile formation. To be specific, these changes can be observed at 2 months old, much antedate by Aβ deposition and even morphological changes of microglia. Despite some DEGs presented throughout the timeline, majority DEGs were specific to a certain age group. This indicates a constantly changing gene expression profile of microglia and implies downstream functional participation of AD pathogenesis. It is surprising that as early as 2 months old, when Aβ plaques were not formed in the brain, many pathways related to energy metabolism and cell function had already changed. Enriched BPs indicated that microglia is beginning to prepare for immune function changes through angiogenesis and regulation of chemotaxis. These may imply that microglia can detect the subtlest changes in the microenvironment and respond. In accordance, the clinical imaging technologies including positron emission tomography and single emission computed tomography have found microglia activation in patients with mild cognitive impairment (MCI) [28, 29]. Moreover, the majority of GWAS (genome wide association studies) identified AD risk genes are expressed selectively or preferentially in microglia relative to other cell types in the brain [9]. Our results provided another important evidence that microglia may serve as a marker for the early diagnosis of AD.

Addressing functional remodulation of microglia to specific intracellular pathways, energy metabolism drew most attention as expounded above. Results indicated that microglia had different preference of energy source according to Aβ formation. Genes incorporated in lipid, cholesterol, and sterol transportation pathways were downregulated in APP/PS1 mice compared to their wild-type littermates at 4 months old, while acylglycerol homeostasis and triglyceride homeostasis pathways were upregulated when it came to the 8-month-old mice. These changes in gene expression profile indicates active metabolic adjustment of microglia in the process of disease progression. Specifically, as the disease progresses, intracellular energy metabolism of microglia is constantly attuning to its functional demand.

Since energy metabolism is tightly regulated, we did some further analysis of metabolic mode of microglia isolated from APP/PS1 mice with different age. We found that glucose, monosaccharide, and hexose transmembrane transporter activity pathways were downregulated in 2-month-old APP/PS1 mice relative to 4-month-old ones. Comparing 4-month-old APP/PS1 mice with 8-month-old ones, acylglycerol, triglyceride homeostasis, and cholesterol transporter pathways were downregulated. Conclusively, the energy remodulation of microglia in APP/PS1 mice showed a bimodule pattern, in which intensive regulation was adjusted from glucose to lipids. As to wild-type mice, glucose metabolism may be the most critical metabolic change with age. Conclusively, the energy remodulation of microglia in APP/PS1 mice and wild-type mice was different. The mechanism of this change and its profound consequences still need further investigations.

In our study, we have only explored the metabolism changes in APP/PS1 mice. The genetic predisposition plays an important role in AD pathogenesis and mouse with different genetic background manifest quite distinct AD-related phenotypes. However, as mentioned in the introduction section, we found the expression levels of Apolipoprotein E (Apoe), lipoprotein lipase (Lpl) in 5xFAD mice were increased with the progression of AD [18]. Moreover, Marschallinger et al. found lipid-related genes were enriched in this lipid droplets (LDs)-high microglia, and ‘fatty acid β-oxidation’ was one of the top enriched pathways. These microglia were defective in phagocytosis, produced high levels of reactive oxygen species, and secreted proinflammatory cytokines [30]. Based on these, we speculated that metabolism changes of microglia in different models are general characteristics, which may play important roles in function modulation.

Further research is needed on the causal relationship between energy metabolism and functional/morphological changes of microglia. Many studies demonstrated that inflammation affects cell energy metabolism [31, 32]. Recent studies discovered how energy metabolism affects inflammatory status [10]. Ulland et al. found that in both TREM2 knockout mice and AD patients with TREM2 gene mutation, microglia showed decreased mitochondrial mass and increased apoptosis [33], and meanwhile, its phagocytic function is impaired, resulting in failure to complete clearance of Aβ plaque [12]. This functional impairment was rescued by supplementary of exogenous energy [19, 34]. Baik et al. reported that chronic exposure to Aβ could lead to metabolic defects in microglia, while treatment with IFN-γ can restore glycolytic metabolism and immunological function of microglia [35]. In contrast, IL-4 can increase oxygen consumption rate, basal respiratory rate, and ATP production, accompanied by the expression of anti-inflammatory markers [36]. In the future, specifically manipulating the energy metabolism genes in microglia to observe the function of microglia and its effect on AD may help us understand the effect of intracellular metabolism on microglia function, and even clarify the causal relationship between them.

Most of the previous studies focused on the functional changes of microglia, as microglia function is an important factor in the progress of AD. However, the underlying reason for the functional changes is still unclear. In this paper, we firstly systematically describe the differences of energy metabolism of microglia in different stages of AD, and associate them with the functional status of microglia, which provides a basis for the bioenergetic regulation of microglia playing important roles in AD. In a recent study, Minhas et al. found that remodeling the energy metabolism of macrophages and microglia, especially glucose metabolism, can reduce systemic and cerebral inflammation, restore hippocampal synaptic plasticity, and reverse cognitive function [37]. This article also proved the importance of energy metabolism dysregulation on the functional changes of microglia to a certain extent. We hope that more attention will be drawn to study the relationship between microglia energy metabolism and cell function in the future.

In addition, we still have some shortcomings. All metabolic changes were judged by transcriptomics. In the future, metabolomics or metabolic flow test can be used to accurately determine the metabolism of microglia in different periods.

In conclusion, this study proved energy meta-bolism remodulation of microglia in APP/PS1 mice brain with transcriptional evidence. The fine tuning of microglia transcription regulation and its constant adjustment are probably implying functional changes, by which incorporated into brain internal environment and affect disease progression in turn. For future studies, through comprehension of the interaction between microglia energy metabolism and brain microenvironment may provide promising diagnosis and treatment of AD.

Footnotes

ACKNOWLEDGMENTS

The study was financially supported by the Na-tional Key Research and Development Program Foundation of China (No.2016YFC1306403), the National Natural Science Foundation of China (No.81870122), the Medical and Health Science and Technology Program Foundation of Xiamen Science and Technology Bureau (No.3502Z20194028), Zhongshan Hospital (No.2020ZSQN23), and the Shanghai Municipal Science and Technology Major Project (No.2018SHZDZX01) and ZJLab.

Authors’ disclosures available online ()

References

Patterson

(2018) World Alzheimer Report 2018. The state of the art of dementia research: New frontiers. Alzheimer’s Disease International, London.

Alzheimer’s Disease International (2019) World Alzheimer Report 2019. Attitudes to dementia. Alzheimer’s Disease International, London.

Hickman

, Izzy

, Sen

, Morsett

, El Khoury

(2018) Microglia in neurodegeneration. Nat Neurosci 21, 1359–1369.

Heneka

, Carson

, El Khoury

, Landreth

, Brosseron

, Feinstein

, Jacobs

, Wyss-Coray

, Vitorica

, Ransohoff

, Herrup

, Frautschy

, Finsen

, Brown

, Verkhratsky

, Yamanaka

, Koistinaho

, Latz

, Halle

, Petzold

, Town

, Morgan

, Shinohara

, Perry

, Holmes

, Bazan

, Brooks

, Hunot

, Joseph

, Deigendesch

, Garaschuk

, Boddeke

, Dinarello

, Breitner

, Cole

, Golenbock

, Kummer

(2015) Neuroinflammation in Alzheimer’s disease. Lancet Neurol 14, 388–405.

Salter

, Stevens

(2017) Microglia emerge as central players in brain disease. Nat Med 23, 1018–1027.

Sarlus

, Heneka

(2017) Microglia in Alzheimer’s disease. J Clin Invest 127, 3240–3249.

Karch

, Goate

(2015) Alzheimer’s disease risk genes and mechanisms of disease pathogenesis. Biol Psychiatry 77, 43–51.

Brown

, St George-Hyslop

(2017) Deciphering microglial diversity in Alzheimer’s disease. Science 356, 1123–1124.

Hansen

, Hanson

, Sheng

(2018) Microglia in Alzheimer’s disease. J Cell Biol 217, 459–472.

10.

Unger

, Schernthaner

, Marschallinger

, Mrowetz

, Aigner

(2018) Microglia prevent peripheral immune cell invasion and promote an anti-inflammatory environment in the brain of APP-PS1 transgenic mice. J Neuroinflammation 15, 274.

11.

Ghosh

, Castillo

, Frias

, Swanson

(2018) Bioenergetic regulation of microglia. Glia 66, 1200–1212.

12.

Shi

, Holtzman

(2018) Interplay between innate immunity and Alzheimer disease: APOE and TREM2 in the spotlight. Nat Rev Immunol 18, 759–772.

13.

Rajendran

, Paolicelli

(2018) Microglia-mediated synapse loss in Alzheimer’s disease. J Neurosci 38, 2911–2919.

14.

Asai

, Ikezu

, Tsunoda

, Medalla

, Luebke

, Haydar

, Wolozin

, Butovsky

, Kügler

, Ikezu

(2015) Depletion of microglia and inhibition of exosome synthesis halt tau propagation. Nat Neurosci 18, 1584–1593.

15.

Calsolaro

, Edison

(2016) Neuroinflammation in Alzheimer’s disease: Current evidence and future & directions. Alzheimers Dement 12, 719–732.

16.

Shemer

, Erny

, Jung

, Prinz

(2015) Microglia plasticity during health and disease: An immunological perspective. Trends Immunol 36, 614–624.

17.

Block

, Zecca

, Hong

(2007) Microglia-mediated neurotoxicity: Uncovering the molecular mechanisms. Nat Rev Neurosci 8, 57–69.

18.

Keren-Shaul

, Spinrad

, Weiner

, Matcovitch-Natan

, Dvir-Szternfeld

, Ulland

, David

, Baruch

, Lara-Astaiso

, Toth

, Itzkovitz

, Colonna

, Schwartz

, Amit

(2017) A unique microglia type associated with restricting development of Alzheimer’s disease. Cell 169, 1276–1290.e1217.

19.

Aldana

(2019) Microglia-specific metabolic changes in neurodegeneration. J Mol Biol 431, 1830–1842.

20.

Pan

, Ma

, Yu

, Zhang

, Wan

(2020) Transcriptomic profiling of microglia and astrocytes throughout aging. J Neuroinflammation 17, 97.

21.

Kamphuis

, Kooijman

, Schetters

, Orre

, Hol

(2016) Transcriptional profiling of CD11c-positive microglia accumulating around amyloid plaques in a mouse model for Alzheimer’s disease. Biochim Biophys Acta 1862, 1847–1860.

22.

Bennett

, Bennett

, Liddelow

, Ajami

, Zamanian

, Fernhoff

, Mulinyawe

, Bohlen

, Adil

, Tucker

, Weissman

, Chang

, Li

, Grant

, Hayden Gephart

, Barres

(2016) New tools for studying microglia in the mouse and human CNS. Proc Natl Acad Sci U S A 113, E1738–1746.

23.

Galatro

, Vainchtein

, Brouwer

, Boddeke

, Eggen

BJL

(2017) Isolation of microglia and immune infiltrates from mouse and primate central nervous system. Methods Mol Biol 1559, 333–342.

24.

Yazawa

, Hsueh

, Jia

, Pasca

, Bernstein

, Hallmayer

, Dolmetsch

(2011) Using induced pluripotent stem cells to investigate cardiac phenotypes in Timothy syndrome. Nature 471, 230–234.

25.

Young

, Morrison

(2018) Quantifying microglia morphology from photomicrographs of immunohistochemistry prepared tissue using ImageJ. J Vis Exp 136, 57648.

26.

Orihuela

, McPherson

, Harry

(2016) Microglial M1/M2 polarization and metabolic states. Br J Pharmacol 173, 649–665.

27.

Song

, Colonna

(2018) The identity and function of microglia in neurodegeneration. Nat Immunol 19, 1048–1058.

28.

Fan

, Brooks

, Okello

, Edison

(2017) An early and late peak in microglial activation in Alzheimer’s disease trajectory. Brain 140, 792–803.

29.

Hamelin

, Lagarde

, Dorothée

, Leroy

, Labit

, Comley

, de Souza

, Corne

, Dauphinot

, Bertoux

, Dubois

, Gervais

, Colliot

, Potier

, Bottlaender

, Sarazin

(2016) Early and protective microglial activation in Alzheimer’s disease: A prospective study using 18F-DPA-714 PET imaging. Brain 139, 1252–1264.

30.

Marschallinger

, Iram

, Zardeneta

, Lee

, Lehallier

, Haney

, Pluvinage

, Mathur

, Hahn

, Morgens

, Kim

, Tevini

, Felder

, Wolinski

, Bertozzi

, Bassik

, Aigner

, Wyss-Coray

(2020) Lipid-droplet-accumulating microglia represent a dysfunctional and proinflammatory state in the aging brain. Nat Neurosci 23, 194–208.

31.

Wang

, Pavlou

, Du

, Bhuckory

, Xu

, Chen

(2019) Glucose transporter 1 critically controls microglial activation through facilitating glycolysis. Mol Neurodegener 14, 2.

32.

Lynch

(2020) Can the emerging field of immuno-metabolism provide insights into neuroinflammation? Prog Neurobiol 184, 101719.

33.

Ulland

, Song

, Huang

, Ulrich

, Sergushichev

, Beatty

, Loboda

, Zhou

, Cairns

, Kambal

, Loginicheva

, Gilfillan

, Cella

, Virgin

, Unanue

, Wang

, Artyomov

, Holtzman

, Colonna

(2017) TREM2 maintains microglial metabolic fitness in Alzheimer’s disease. Cell 170, 649–63.e613.

34.

Hong

, Stevens

(2017) TREM2: Keeping microglia fit during good times and bad. Cell Metab 26, 590–591.

35.

Baik

, Kang

, Lee

, Choi

, Chung

, Kim

, Mook-Jung

(2019) A breakdown in metabolic reprogramming causes microglia dysfunction in Alzheimer’s disease. Cell Metab 30, 493–507.e496.

36.

Holland

, McIntosh

, Finucane

, Mela

, Rubio-Araiz

, Timmons

, McCarthy

, Gun’ko

, Lynch

(2018) Inflammatory microglia are glycolytic and iron retentive and typify the microglia in APP/PS1 mice. Brain Behav Immun 68, 183–196.

37.

Minhas

, Latif-Hernandez

, McReynolds

, Durairaj

, Wang

, Rubin

, Joshi

, He

, Gauba

, Liu

, Wang

, Linde

, Sugiura

, Moon

, Majeti

, Suematsu

, Mochly-Rosen

, Weissman

, Longo

, Rabinowitz

, Andreasson

(2021) Restoring metabolism of myeloid cells reverses cognitive decline in ageing. Nature 590, 122–128.

38.

Deng

, Sun

, Yan

, Ke

, Jiang

, Xiong

, Ren

, Hirata

, Yamamoto

, Fan

, Yan

(2015) Molecular basis of ligand recognition and transport by glucose transporters. Nature 526, 391–396.

39.

Jin

, Seo

, Kim

, Woo

, Lee

, Yi

, Lee

, Choe

, Lee

, An

, Hong

, Park

(2011) TXNIP potentiates Redd1-induced mTOR suppression through stabilization of Redd1. Oncogene 30, 3792–3801.

40.

Fliegauf

, Bryant

, Frede

, Slade

, Woon

, Lehnert

, Winzer

, Bulashevska

, Scerri

, Leung

, Jordan

, Keller

, de Vries

, Cao

, Yang

, Schäffer

, Warnatz

, Browett

, Douglass

, Ameratunga

, van der Meer

, Grimbacher

(2015) Haploinsufficiency of the NF-κB1 subunit p50 in common variable immunodeficiency. Am J Hum Genet 97, 389–403.

41.

Lee

, Ma

, Sacharidou

, Mi

, Salato

, Nguyen

, Jiang

, Pascual

, North

, Shaul

, Mettlen

, Wang

(2015) A protein kinase C phosphorylation motif in GLUT1 affects glucose transport and is mutated in GLUT1 deficiency syndrome. Mol Cell 58, 845–853.

42.

Tisdale

(2002) Glyceraldehyde-3-phosphate dehydrogenase is phosphorylated by protein kinase Ciota /lambda and plays a role in microtubule dynamics in the early secretory pathway. J Biol Chem 277, 3334–3341.

43.

Magnani

, Bianchi

, Casabianca

, Stocchi

, Daniele

, Altruda

, Ferrone

, Silengo

(1992) A recombinant human ‘mini’-hexokinase is catalytically active and regulated by hexose 6-phosphates. Biochem J 285 (Pt 1), 193–199.

44.

Burnichon

, Brière

, Libé

, Vescovo

, Rivière

, Tissier

, Jouanno

, Jeunemaitre

, Bénit

, Tzagoloff

, Rustin

, Bertherat

, Favier

, Gimenez-Roqueplo

(2010) SDHA is a tumor suppressor gene causing paraganglioma. Hum Mol Genet 19, 3011–3020.

45.

Scherer

, Brockman

, Chen

, Maniatis

, Ballard

(1995) Signal-induced degradation of I kappa B alpha requires site-specific ubiquitination. Proc Natl Acad Sci U S A 92, 11259–11263.

46.

Arakaki

, Pezza

, Cronin

, Hopkins

, Zimmer

, Tolan

, Allen

(2004) Structure of human brain fructose 1,6-(bis)phosphate aldolase: Linking isozyme structure with function. Protein Sci 13, 3077–3084.

47.

Palma

, Agostini

, Mason

, Dachà

, Piccoli

, Biagiarelli

, Fiorani

, Stocchi

(1996) Purification and characterization of the carboxyl-domain of human hexokinase type III expressed as fusion protein. Mol Cell Biochem 155, 23–29.

48.

Stroud

, Surgenor

, Formosa

, Reljic

, Frazier

, Dibley

, Osellame

, Stait

, Beilharz

, Thorburn

, Salim

, Ryan

(2016) Accessory subunits are integral for assembly and function of human mitochondrial complex I. Nature 538, 123–126.

49.

, Clark

, Mason

, Keenan

, Hill

, Goddard

3rd , Peters

, Driggers

, Hsieh-Wilson

(2012) Phosphofructokinase 1 glycosylation regulates cell growth and metabolism. Science 337, 975–980.

50.

Bollong

, Lee

, Coukos

, Yun

, Zambaldo

, Chang

, Chin

, Ahmad

, Chatterjee

, Lairson

, Schultz

, Moellering

(2018) A metabolite-derived protein modification integrates glycolysis with KEAP1-NRF2 signalling. Nature 562, 600–604.

51.

Veiga-da-Cunha

, Vleugels

, Maliekal

, Matthijs

, Van Schaftingen

(2008) Mammalian phosphomannomutase PMM1 is the brain IMP-sensitive glucose-1,6-bisphosphatase. J Biol Chem 283, 33988–33993.

52.

Cheung

, Ludwig

, Vousden

(2012) Mitochondrial localization of TIGAR under hypoxia stimulates HK2 and lowers ROS and cell death. Proc Natl Acad Sci U S A 109, 20491–20496.

53.

Pop

, Williams

, Struys

, Monné

, Jansen

EEW

, De Grassi

, Kanhai

, Scarcia

, Ojeda

MRF

, Porcelli

, van Dooren

SJM

, Lennertz

, Nota

, Abdenur

, Coman

, Das

, El-Gharbawy

, Nuoffer

, Polic

, Santer

, Weinhold

, Zuccarelli

, Palmieri

, Salomons

(2018) An overview of combined D-2- and L-2-hydroxyglutaric aciduria: Functional analysis of CIC variants. J Inherit Metab Dis 41, 169–180.