Loss of succinyl-CoA synthetase in mouse forebrain results in hypersuccinylation with perturbed neuronal transcription and metabolism

SUMMARY

Lysine succinylation is a subtype of protein acylation associated with metabolic regulation of succinyl-CoA in the tricarboxylic acid cycle. Deficiency of succinyl-CoA synthetase (SCS), the tricarboxylic acid cycle enzyme catalyzing the interconversion of succinyl-CoA to succinate, results in mitochondrial encephalomyopathy in humans. This report presents a conditional forebrain-specific knockout (KO) mouse model of Sucla2, the gene encoding the ATP-specific beta isoform of SCS, resulting in postnatal deficiency of the entire SCS complex. Results demonstrate that accumulation of succinyl-CoA in the absence of SCS leads to hypersuccinylation within the murine cerebral cortex. Specifically, increased succinylation is associated with functionally significant reduced activity of respiratory chain complex I and widescale alterations in chromatin landscape and gene expression. Integrative analysis of the transcriptomic data also reveals perturbations in regulatory networks of neuronal transcription in the KO forebrain. Together, these findings provide evidence that protein succinylation plays a significant role in the pathogenesis of SCS deficiency.

In brief

Lancaster et al. present a mouse model of forebrain-specific succinyl-CoA synthetase deficiency, resulting in accumulation of TCA cycle intermediate succinyl-CoA and causing global protein hypersuccinylation. Hypersuccinylation within the forebrain is associated with changes in neuronal transcriptional networks and mitochondrial respiratory dysfunction, suggesting that hypersuccinylation contributes to pathogenesis of SCS deficiency.

Graphical Abstract

INTRODUCTION

Protein acylation is a post-translational modification emerging as a critical regulator of metabolic function and disease. Many types of protein acylation have been investigated, including acetylation, succinylation, propionylation, and malonylation.¹ Fatty acyl-coenzyme A (acyl-CoA) species are the predominant donors for protein acylation, and these modifications are tightly regulated by the metabolic generation of acyl-CoA intermediates within the mitochondrion.² Acylation is a reversible modification of lysine residues, affecting both the charge and size of the amino acid side chain.^1,3 These lysine modifications have been shown to have broad impacts on many cellular processes, including epigenetic regulation of transcription and protein localization, structure, and function, with mitochondrial metabolism being largely affected.⁴ While metabolic pathways within the mitochondria, such as the tricarboxylic acid (TCA) cycle, urea cycle, and electron transport chain (ETC) are largely influenced by lysine acylation, no mitochondrial acyltransferase has been identified. Therefore, acylation within the mitochondria is widely believed to occur nonenzymatically and much of the current research focuses on perturbations of the sirtuin family of mitochondrial deacylases.^5,6

Lysine acylation is principally dependent on cellular concentrations of the acyl-CoA species generated by intermediary metabolism. Succinyl-CoA is a metabolic intermediate generated within the TCA cycle of the mitochondrial matrix and is the donor for lysine succinylation. The biological significance of lysine succinylation has been predominantly studied in knockout (KO) models of SIRT5, the mitochondrial localized desuccinylase, and dysregulation of lysine succinylation has been associated with several disease states, including heart disease, diabetes, Alzheimer’s disease, and mitochondrial disorders.^7–10 Recently, hypersuccinylation was reported in cultured fibroblasts from patients with mutations in SUCLA2, encoding α subunit of the TCA cycle enzyme, succinyl-CoA synthetase (SCS).⁷

SCS is responsible for catalyzing the reversible conversion of succinyl-CoA to succinate coupled with substrate level phosphorylation in the TCA cycle. In eukaryotes, SCS is a heterodimeric complex with a single catalytic α subunit, SUCLG1, and two isoforms of the β subunit, ATP-generating SUCLA2, and GTP-forming SUCLG2. Biallelic mutations in both SUCLG1 and SUCLA2 have been associated with mitochondrial encephalomyopathy in human patients, with a typical onset in early childhood.¹¹ Common symptoms include intellectual disability, hypotonia, dystonia, and sensorineural deafness, with methylmalonic aciduria and mitochondrial DNA (mtDNA) depletion in muscle tissue, although pathogenic mechanisms remain to be elucidated.^11–13 Furthermore, hypersuccinylation was reported as a molecular phenotype in SUCLA2^−/− patient fibroblasts, which was recapitulated in sucla2^−/− zebrafish larvae by 7 days post-fertilization.⁷ However, the role of these molecular perturbations in the pathogenesis of SCS deficiency is poorly understood and, until recently, investigation into these mechanisms has largely relied on in vitro studies, as the generation of animal models has proven difficult, with homozygous loss-of-function variants in Sucla2 resulting in developmental lethality.^7,14

This study presents a mouse model that bypasses embryonic lethality via spatiotemporal control of Sucla2 mutation using the Cre-Lox system for mutagenesis.¹⁵ Through restriction of Sucla2 mutation to the postnatal mouse forebrain, KO mice survive, allowing investigation into adult phenotypes of Sucla2 deficiency in the mouse brain. This report presents the characterization of the metabolic consequences of forebrain-specific KO of Sucla2, including widescale changes in global succinylation. Using immunoprecipitation (IP) and concomitant mass spectrometry, a highly enriched succinylated proteome is identified in the Sucla2-deficient forebrain. The research presented herein combines affinity purification mass spectrometry (MS), global proteomics, and transcriptomics datasets to investigate the relationship between global changes in the succinylome and forebrain-specific pathogenic phenotypes of SCS deficiency. This model will provide a critical tool for understanding the role of lysine succinylation in SCS deficiency and the consideration of protein acylation as a potential therapeutic target for multiple diseases with metabolic dysfunction.

RESULTS

Forebrain-specific KO of Sucla2 causes downregulation of the entire SCS complex and metabolic perturbations in the absence of mtDNA depletion

Currently, the only reported mouse model of constitutive Sucla2 deficiency exhibits embryonic lethality, demonstrating a continued need for an adult in vivo model of SCS deficiency.¹⁴ While whole-body KO of Sucla2 is a lethal phenotype in mice, this report tested the hypothesis that tissue-specific restriction of Sucla2 mutagenesis would allow adult murine survival. Given that SUCLA2-deficient phenotypes in human patients predominantly affect high energy tissues, the CamKIIα-Cre transgene was used to drive forebrain-specific postnatal recombination of a CRISPR-Cas9-generated floxed allele of Sucla2 (Figure 1A).^16,17 The spatial restriction of Sucla2 deficiency resulted in complete survival of the KO mice (Table S1), and molecular validation via qPCR confirmed a 68% reduction in Sucla2 gene expression in the mutant cerebral cortex when compared with wild-type (WT) controls (Figure 1B). Reverse transcriptase qPCR (qRT-PCR) analysis also revealed a slight decrease in the mRNA expression of both Suclg1 and Suclg2 (Figure 1B). The reduction in SCS transcript levels is exacerbated at the protein level, where western blot analysis indicates that forebrain-specific KO of Sucla2 results in decreased protein expression of both SUCLG1 and SUCLG2 at levels comparable to SUCLA2 loss (Figures 1C and 1D). These data demonstrate significant post-transcriptional downregulation leading to 80%–90% reduction of the entire SCS complex in the Sucla2^−/− cerebral cortex. In addition, immunohistochemical (IHC) analysis co-staining for neuronal marker NeuN and SUCLA2 reveals marked (p < 0.0001) reduction of SUCLA2 within the neurons of the cerebral cortex (59% reduction) and hippocampus (69% reduction) of Sucla2 mutants compared with control mice (Figures 1E–1H). These data suggest that the knockdown of Sucla2 by neuron-specific cre-recombination¹⁶ leads to SCS loss within the neurons of the forebrain.

Figure 1. Forebrain-specific knockout of Sucla2 leads to post-transcriptional downregulation of the entire SCS complex.

(A) Graphical representation of the Sucla2 alleles generated, including a floxed allele (Sucla2^Flox) with loxP sites (orange triangles) flanking exons 3–4, and a constitutive deletion allele (Sucla2^Δ). Recombination of the floxed allele is achieved in the presence of Cre transgene expression driven by the CamKIIα promoter.¹⁶

(B) Gene expression of SCS subunits in the cerebral cortex measured by qRT-PCR (n = 16–17). Data shown are pooled from three separate experiments, and non-normalized results are provided in Table S2.

(C) Composite western blot depicting protein expression of all SCS subunits in the cerebral cortex and GAPDH as a loading control (n = 5).

(D) Combined densitometry analysis of three separate western blot experiments (n = 13). Bands of SCS subunits in the cortex were quantified in ImageJ and normalized to the loading control. Non-normalized results are provided in Table S2.

(E and G) Representative images of SUCLA2-positive (+) cells (green) and NeuN⁺ cells (neuronal marker, red) from cortex (E) and hippocampus (G) in brain slices from Sucla2 control and mutant mice. Scale bars, 100 μm.

(F and H) Quantification of the percentage of SUCLA2⁺ cells on the neuronal cells in the cortical area (F) and hippocampal area (H). All data presented in bar graphs are represented as means ± SD. WT = Sucla2^+/+, CamKIIα-Cre positive (blue); and KO are either Sucla2^Flox/Δ or Sucla2^Flox/Flox and CamKIIα-Cre positive (red). Equivalent levels of KO between the two mutant genotypes are shown in Figure S1. Statistically significant differences are depicted by asterisks, where *p < 0.05,

**p < 0.01, ****p < 0.0001 by standard unpaired t tests.

To further validate the reduction of the entire SCS complex, both ADP- and GDP-specific activities of SCS were measured in whole-cell lysates isolated from the cerebral cortex of mutant and control mice (Figure 2A). GDP-specific activity was between 30% and 50% that of the ADP-specific activity regardless of genotype, which is consistent with the literature reporting SUCLA2 as the predominant β-isoform in the mouse brain.¹⁸ Furthermore, the enzymatic data show a 50% reduction in ADP-specific activity and a 72% reduction in GDP-specific activity in the mutant cortex compared with WT controls, providing further evidence of significant reduction (58%) of total SCS activity. SCS deficiency is associated with hallmark phenotypes of mtDNA depletion and methylmalonic aciduria in human patients.^12,13,19,20 However, mtDNA depletion is not observed in the Sucla2^−/− cerebral cortex; rather, there is a slight elevation in mtDNA copy number (Figure 2B). Strikingly, there is a 14-fold increase in the concentration of methylmalonic acid (MMA) measured in whole-cell lysates of Sucla2-deficient cortex compared with WT controls (Figure 2C; Data S1). Methylmalonyl-CoA is the coenzyme A-inked derivative of MMA that feeds into the TCA cycle through reversible conversion to succinyl-CoA via methylmalonyl-CoA mutase. Therefore, accumulation of MMA in the Sucla2^−/− cortex provides additional evidence of SCS deficiency and a subsequent buildup of succinyl-CoA. Therefore, succinyl-CoA levels were measured directly in whole-cell lysates of cerebral cortex, and results indicate a 50% increase in mutant cellular concentrations of succinyl-CoA compared with controls (Figure 2D; Data S1).

Figure 2. Postnatal SCS deficiency causes metabolic perturbations in the cerebral cortex in the absence of mtDNA depletion.

(A) ADP- and GDP-specific SCS enzyme activities measured in the forward direction in whole-cell lysates of murine cortex (n = 4). Total activity is defined as the sum of nucleotide-specific activities.

(B) MtDNA copy number measured by qPCR. Results are pooled from four separate experiments, and non-normalized results are provided in Table S3 (n = 18–19).

(C–F) Relative concentrations of significantly altered metabolites in whole-cell lysates of murine cortex (n = 4–9), including levels of methylmalonic acid (MMA) (C), succinyl-CoA (D), amino acids (E), and acyl-carnitines (F). Raw data for all measured metabolites are provided in Data S1. WT = Sucla2^+/+, CamKIIα-Cre positive (blue); and KO are either Sucla2^Flox/Δ or Sucla2^Flox/Flox and CamKIIα-Cre positive (red). All data are represented as mean ± SD. Statistically significant differences are depicted by asterisks, where *p < 0.05, **p < 0.01, ****p < 0.0001 by standard unpaired t tests, with correction for multiple testing where applicable.

To further characterize the metabolic signature of the model, mass spectrometry was used to measure the cellular concentrations of amino acids and acylcarnitines (Data S1). The results demonstrate a mutant-specific increase in glutamate and taurine amino acids and a decrease in aspartate, glutamine, and asparagine (Figure 2E). Notably, most of the amino acids with altered cellular concentrations in the mutant cortex are directly involved in TCA cycle metabolism, suggesting potential perturbations beyond amino acid metabolism in Sucla2-deficient neurons. Moreover, marked elevations in methylmalonic and succinyl (C4-dicarboxylic of C4-DC) and propionyl (C3) acylcarnitines were observed (Figure 2F). The elevation of propionyl-, methylmalonyl-, and succinyl-carnitines provides evidence not only for SCS deficiency, but also for cellular accumulation of the respective acyl-CoA species. These data support the observation of accumulation of cerebral succinyl-CoA resulting from the loss of SCS (Figure 2D), while also demonstrating metabolic perturbations that mirror patient biochemical phenotypes.

Buildup of succinyl-CoA in the TCA cycle results in global increase in protein succinylation in mutant cortex

Given the accumulation of succinyl-CoA in the cortex of Sucla2^−/− mice, the possibility of a global increase in cerebral lysine-succinylation was investigated. This hypothesis was confirmed via Western blot analysis showing a 100-fold increase of succinylated lysine in Sucla2 mutant cortex (Figures 3A, 3B, and S2). In addition, IHC staining showed a nearly 3-fold increase in succinylated lysine immunoreactivity in the cortical area of Sucla2 mutant compared with control mice (Figures 3C and 3D). To identify and quantify the specific targets of lysine succinylation in both mutants and controls, IP using an antibody against succinyllysine was performed to enrich for succinylated peptides, followed by nano-liquid chromatography-tandem MS (LC-MS/MS) acquired with high-field asymmetric waveform ion mobility spectrometry (Data S2). A proportion of the same sample extracts prepared for IP were subjected to global proteomics (LC-MS/MS) (Data S3; Figure S3) without succinyl-enrichment, and succinylated peptide abundances were normalized to total protein levels. To accurately measure the differential abundance of succinylated peptides, a spike-in control of chicken lysozyme was added to each sample following IP relative to the amount of starting protein in each sample to allow for proper normalization. All proteomic data were deposited at massIVE: https://massive.ucsd.edu/ with login details provided in the STAR Methods. With 21,040 peptides identified, 4,747 peptides across 1,082 proteins had at least one succinyl modification, and 92% of the quantified succinylated peptides were significantly (p < 0.05) hypersuccinylated in the mutant cortex (Figure 3E, blue and red).

Figure 3. SCS deficiency results in global increase in cortex protein succinylation.

(A) Western blot analysis performed using an antibody against succinyllysine, with GAPDH expression used as a loading control (n = 5).

(B) Combined densitometry analysis of three separate western blot experiments (n = 13). Global intensity of WT (blue) and KO (red) bands were quantified in ImageJ and normalized to the loading control. Bar graph is shown as mean ± SD, where ****p < 0.0001 by a standard unpaired t test. Western blot images for the two additional blots are included in Figure S2.

(C) Representative images of succinyllysine (green) in brain slices from Sucla2 control and mutant mice. Scale bar, 100 μm.

(D) Quantification of succinyllysine immunoreactivity (IR) in the cortical area. Bar graphs are means ± SEM (n = 6), where ****p < 0.0001 by standard unpaired t test.

(E) Volcano plot depicting succinylated peptide abundance quantified via LC-MS/MS analysis of TMT-labeled samples. The log-plot shows fold change (FC) vs. p value of KO vs. WT peptide abundances. Peptides where p < 0.05 are shown in blue, and peptides where p < 0.05 and absolute FC ≥ 2 are shown in red. n = 6 mice per genotype. Results for the succinyl-enriched and global proteomic experiments are provided in Data S2 and Data S3, respectively.

(F) The distribution of total number of succinylated lysine sites per protein. Blue bars represent the proportions of all succinylated proteins identified (n = 1,189) and red bars represent the proportions of hypersuccinylated proteins identified (p < 0.05) (n = 661). The distributions were compared via chi-square analysis (p < 0.0001). Raw counts for each group are provided in Figure S2B.

(G) Subcellular localization of hypersuccinylated proteins (red, KO vs. WT p < 0.05 for at least one lysine) and succinylated proteins that are not enriched in the mutant cortex (blue, KO vs. WT p > 0.05). The distributions were not significantly different via chi-square analysis. Raw counts for each group are provided in Figure S2D.

(H) Subcellular localization of hypersuccinylated proteins (KO vs. WT p < 0.05 for at least one lysine) with one succinylated lysine (blue) and five or more succinylated lysines (red). The distributions were significantly different by chi-square analysis (p = 0.002). Raw counts for each group are provided in Figure S2E. For all data shown, WT = Sucla2^+/+, CamKIIα-Cre positive; and KO = Sucla2^Flox/Δ, CamKIIα-Cre positive. Localizations were determined by the maximum confidence score (5) within Cytoscape’s STRING application.^21,22.

Of the hypersuccinylated proteome, 2,176 peptides from 640 proteins had a 2-fold or greater abundance in the SCS-deficient cortex (Figure 3E, red). Approximately 37% of all succinylated proteins quantified in both experimental groups have a single succinylation site; however, many proteins (63%) have multiple succinylated lysines (Figure 3F). This distribution is skewed (p < 0.0001), however, when only considering the proteins enriched for succinylation in the mutants, in which 38% of proteins with at least one hypersuccinylated lysine have five or more total succinylated lysines (Figure 3F). In other words, most proteins with a single succinylation site are not differentially succinylated in the Sucla2-deficient cortex. Succinylated proteins are enriched for mitochondrial localization (p < 0.0001, Figure S2C) as defined by cellular component analysis through Cytoscape’s STRING application.^21,22 However, while there is no difference in the distribution of subcellular localization between enriched (mutant vs. WT p < 0.05) and not enriched (p > 0.05) succinylated proteins (Figure 3G) there are more highly succinylated (five or more succinylation sites) proteins localized to the mitochondria and a higher proportion of proteins with only one succinylation site within the nucleus (p < 0.01, Figure 3H). When considering the entire hypersuccinylated proteome, 20% of SCS-deficient hypersuccinylated proteins are in the nucleus, and 11% are found within the cytosol. In addition, both the mitochondrial and nuclear localized desuccinylases, SIRT5 and SIRT7, respectively, are upregulated in the mutant cortex (Table S4). All together, these results suggest that SCS-related changes in the succinylome are certainly enriched but not restricted to the mitochondrion.

The Sucla2-related changes in the succinylome are associated with functional perturbations in the ETC

Given that the mitochondrial proteome is highly affected by SCS-related changes in post-translational succinylation, further investigation into the effects on mitochondrial pathways was pursued. Of the 20 most affected succinylation marks (with highest KO:WT fold change [FC]), 19 are in mitochondrial proteins, including subunits of the complexes of the TCA cycle (OGDH, MDG2, ACO2) and ETC (UQCRC2, NDUFC1, ATP5PD, ATP8) (Figure 4A). These pathways are also significantly enriched (FDR < 0.05) as determined by gene ontology (GO) enrichment analysis for biological processes (Figure 4B; Data S4). Four of the five complexes of the ETC and the TCA cycle were among the highest-enriched biological processes in the hypersuccinylated dataset. Plots of the FC of the abundance of succinylated peptides within TCA cycle enzymes and ETC complexes in mutant cortex show markedly increased succinylation (Figures 4C and 4D). Pathways in carbohydrate, amino acid, and fatty acid metabolism are also significantly (FDR < 0.05) enriched (Figures 4A and 4B).

Figure 4. Sucla2-related changes in the succinylome associated with functionally significant deficiency of complex I.

(A) Bar plot illustrating the 20 peptides with the highest fold change (FC) in succinyllysine abundance between genotypes. All peptides are annotated with the corresponding protein’s symbol with specific succinylation sites indicated.

(B) GO biological processes enrichment analysis of proteins hypersuccinylated in mutant cortex (KO vs. WT p < 0.05), with the most specific terms represented. Bar plot shows the fold enrichment of the 10 highest enriched biological processes, where statistical significance was determined via Fisher’s exact test and Bonferroni correction. Processes are grouped by color into metabolic pathways as in (A). Complete results of the GO analysis in PANTHER are provided in Data S4.

(C and D) Differentially succinylated peptides within the TCA cycle (C) and ETC (D). Data shown are KO vs. WT FC of differentially succinylated peptides (p < 0.05) on a log-scale.

(E) Complex I (NADH-ubiquinone oxidoreductase) activity in isolated mitochondria of WT (blue) and KO (red) murine cortex (n = 9).

(F) NADH/NAD+ ratio measured in cortex whole-cell lysates from WT (blue) and KO (red) mice (n = 9–10). Individual concentrations of NADH and NAD+ are provided in Figures S3C and S3D. WT = Sucla2^+/+, CamKIIα-Cre positive; and KO are either Sucla2^Flox/Flox or Sucla2^Flox/Δ and CamKIIα-Cre positive. Data in (E) and (F) are represented as mean ± SD. Statistical significance is indicated by asterisks, where *p < 0.05 by standard unpaired t tests.

To address how succinylation may affect the function of these enzymes, the catalytic activity of ETC complex I was measured in mitochondrial fractions isolated from the Sucla2^−/− murine cortex and WT controls. Global abundance of complex I subunits are equal or elevated in mutants compared with controls (Figure S3). However, in the mutant cortex, where 92 of 225 total identified peptides within complex I are significantly (p < 0.05) hypersuccinylated with an average FC of 6 (Figure 4D), mitochondrial complex I activity is reduced 40% on average (Figure 4E). This deficiency was further evidenced by a cellular increase in the NADH/NAD+ ratio, suggesting a functional decrease in the oxidation of NADH by complex I (Figures 4F, S3C, and S3D). These results demonstrate a respiratory complex dysfunction associated with severe hypersuccinylation. Interestingly, hypersuccinylation of complex V (CV) and TCA cycle enzyme citrate synthase (CS) are not associated with changes in enzyme activities (Figures S3E and S3F). Therefore, succinylation can lead to functional changes in enzyme activity. The changes occur in an enzyme-specific manner, however, indicating that the effect of succinylation on protein function is complex, similarly to many other post-translational modifications.^23–26

Sucla2-deficient hypersuccinylation is accompanied by perturbations in neuronal transcriptional regulation

In addition to functional changes within mitochondrial metabolic pathways, the potential consequence of hypersuccinylation of proteins outside of the mitochondria was also investigated, given the prevalence of hypersuccinylated proteins in other cellular compartments. Published literature has identified succinylation marks on histones that are associated with changes in chromatin structure and gene expression.²⁷ Therefore, the assay for transposase-accessible chromatin with high-throughput sequencing (ATAC-seq) was performed to compare the global chromatin landscape in Sucla2 mutant cortex to controls. Analysis of accessibility to enhancer regions was limited to enhancers known to be active in the cerebral cortex, while all defined promoter regions were analyzed. Multidimensional scaling of both datasets revealed stark genotype-specific clustering for changes in the chromatin landscape of both enhancer and promoter regions (Figures 5A and 5C). Of 5,495 enhancer regions analyzed, 431 showed significantly altered chromatin accessibility (p < 0.05) in the mutant cortex, with 144 having an absolute FC of 2 or higher (Figure 5B). For 22,317 promoter regions analyzed, nearly 22% (4,839) exhibited significantly altered chromatin accessibility (p < 0.05). Of those altered promoters, 436 had an absolute FC of 2 or higher, with slight skewing toward less-accessible promoter regions as a result of Sucla2 deficiency in the mutant cortex (Figure 5D). Analysis of both the promoter and enhancer regions are provided in Data S5.

Figure 5. SCS-deficient hypersuccinylation is accompanied by changes in global chromatin configuration and gene expression.

(A–F) Multidimensional scaling (MDS) plots reveal genotype-specific clustering of cortex-specific enhancer regions (A) and global promoter regions (C) measured by ATAC-seq, and differential gene expression measured by bulk RNA-seq (E). Volcano plots show genotype-specific changes in chromatin accessibility of enhancer (B) and promoter regions (C), and in global gene expression (F). Genes where FDR < 0.05 are shown in blue, and genes where FDR < 0.05 and absolute KO vs. WT FC R ≥ are red. n = 3–4 mice per genotype. Results from the ATAC-seq and RNA-seq experiments are provided in Data S5 and Data S6, respectively.

(G and H) Correlation between SCS-deficient gene expression and chromatin configuration at enhancer (G) and promoter (H) regions of respective genes. Colored points correspond to genes with significant (FDR < 0.05) changes in both chromatin accessibility and gene expression. Note that the promoter regions are positively correlated with a Pearson correlation coefficient (r) of 0.07 (p < 0.0001).

(I) Histogram of hypersuccinylated lysines (p < 0.05) within histone proteins (n = 6).

(J) Relative abundance of chromatin remodeling enzymes (n = 6). Data are presented in bar graphs as means ± SD, with WT data shown in blue and KO data in red. Statistically significant differences are depicted by asterisks, where *p < 0.05, **p < 0.01 by standard unpaired t tests.

(K) Relative regulatory activity of transcription factors (TF) on target genes (TGs) calculated by the statistical model PECA.²⁸ The 20 TF-TG relationships with the highest KO vs. WT FC in the mutant cortex (top) as well as the 20 regulatory relationships with the lowest KO vs. WT FC in the mutant cortex (bottom) are shown. Target genes are colored by the associated transcription factor. The full results from PECA are provided in Data S7.

(L) Bubble plot showing the correlation between Reactome pathway enrichment³¹ of target genes that have lost transcriptional regulation in the mutants (PECA FC < 0) and their associated regulatory transcription factors. Reactome pathways are annotated on the y axis, and transcription factors are shown on the x axis. The diameter and color of the bubbles represent the number of target genes within each pathway whose transcription is regulated by the corresponding transcription factors. The statistical results (fold enrichment and FDR values) of the pathway enrichment analysis are shown on the bar graph to the right. The top bar graph depicts the total number of target genes regulated by each transcription factor in all pathways (gray) as well as the number of genes that fall within the broad Reactome pathway “neuronal systems” (orange). For all data shown, WT = Sucla2^+/+, CamKIIα-Cre positive, and KO = Sucla2^Flox/Δ, CamKIIα-Cre positive.

To correlate the changes in chromatin landscape with alterations in the transcriptome, bulk RNA sequencing (RNA-seq) with differential gene expression (DE) analysis was performed, and all transcriptomic data have been deposited at GEO: GSE208692. The DE analysis shows a significant (FDR < 0.05) change in expression for 13% of the 24,421 total mapped genes, clearly distinguishing mutant cortex from controls (Data S6; Figures 5E and 5F). Interestingly, we see about 60% of the genes encoding subunits of ETC complexes are significantly (FDR < 0.05) upregulated in the mutant cortex (Figure S4), possibly as a compensatory response for the energy metabolism deficit resulting from SCS deficiency (Figure 2A) and the functionally significant secondary complex I deficit (Figures 4E and 4F). In further support of this hypothesis, GO Reactome pathway enrichment analysis was conducted on significantly upregulated genes within the RNA-seq data, with “metabolism” (FDR = 9.22 × 10⁻³²), “the citric acid (TCA) cycle” (FDR = 5.27 × 10⁻²⁷), and “respiratory electron transport” (FDR = 1.11 × 10⁻²³) among the most highly enriched pathways within the upregulated gene set (Data S4).

When the chromatin-accessibility dataset in promoters from the Sucla2 mutant cortex is compared with the differential gene expression dataset from the same animals, a positive correlation (p < 0.0001) between mutant-specific changes in accessibility of chromatin promoter regions and differentially expressed genes is observed (Figures 5G and 5H). The proteomic data were also interrogated to investigate whether the broad changes in global chromatin landscape may be attributed to histone succinylation. Increased mutant abundance of at least one succinylation mark was detected on each of the five major histone proteins (H1, H2A, H2B, H3, and H4) in the mutant cortex (Figure 5I). Furthermore, histone H2B had five hypersuccinylated marks detected and quantified in the mutant cortex. Interestingly, interrogation into the global (non-succinyl enriched) dataset showed an increase in abundance of a few chromatin remodeling enzymes (Figure 5J), suggesting that there is likely a complex combination of mechanisms leading to the mutant-specific changes in global chromatin configuration.

To better understand the relationship between changes in the chromatin landscape and changes in gene expression in the Sucla2 mutant cortex, the statistical model PECA (paired expression and chromatin accessibility) was employed²⁸ (Data S7). This model incorporates publicly available data on transcription factors (TFs), as well as experimental ATAC-seq and RNA-seq data, to infer differential activity of TF regulation on specific target genes (TGs). The regulatory relationships (TF-TG regulatory activities) that are more active in the Sucla2^−/− cortex (Figure 5K, top graph) have an average FC of around 1, while the regulatory relationships that are less active in the mutants (Figure 5K, bottom graph) have up to a 4-fold decrease in transcriptional regulation of specific target genes, suggesting an overall loss of transcriptional regulation in the Sucla2 KO model, which coincides with a skewing toward less-accessible promoter regions (Figure 5D). Notably, the TFs that had the largest reduction in TF-TG regulatory activities in the Sucla2 mutant cortex include members of the EGR and POU3 TF families, which play a crucial role in neuronal development and function^29,30 These data agree with results from GO Reactome pathway analysis from genes significantly downregulated in the RNA-seq data. In these datasets, not only were “chromatin organization” (FDR = 1.32 × 10⁻⁷) and “chromatin modifying enzymes” (FDR = 1.64 × 10⁻⁷) among the most enriched pathways, but “nervous system development” (FDR = 5.09 × 10⁻⁷) was also shown to be significantly enriched in the downregulated gene set (Data S4).

To further appreciate how these regulatory networks may impact neuronal function, the data from the PECA analysis were also subjected to Reactome pathway enrichment analysis,³¹ with specific focus on the TF-TG relationships determined by the PECA model to be statistically less active in the mutants (Figure 5L). Nearly all enriched pathways were related to neuronal functions, including neuronal signaling, neurotransmitter release, and nervous system development, which is further supported by an overall enrichment for genes associated with neuronal systems as shown in the bar graph at the top of Figure 5L. The EGR and POU families, as well as NEUROD2, were highly represented among TFs with lowered regulatory activity of TGs within these pathways. In total, the combined analysis of the ATAC-seq and RNA-seq datasets reveal, in the Sucla2 mutant cortex, changes in the global chromatin landscape, hypersuccinylation of histones, and a loss of regulation of transcriptional networks that may have a large impact on neuron development and neuronal function.

DISCUSSION

Spatiotemporal restriction of Sucla2 mutagenesis allows for the generation of a mouse model of postnatal SCS deficiency

Mutations in the subunits of SCS cause Leigh-like mitochondrial encephalomyopathy in human patients that typically manifests in early childhood. With no curative treatments currently available, these patients typically do not survive into adulthood.^11,13,19 To date, investigation into pathogenic mechanisms of SCS deficiency has largely relied upon in vitro, embryonic, or heterozygous models, as global recessive KO of Sucla2 results in lateembryonic or early life lethality in animal models.^{7,14,19,32,33} Using a Cre transgene driven by the CamKIIα promoter to drive recombination of a CRISPR-Cas9-generated Sucla2 floxed allele, the present research presents a viable adult mouse model of forebrain-specific homozygous Sucla2 deficiency. In contrast to previously described models, postnatal KO of Sucla2 in neurons results in post-transcriptional downregulation of both the α subunit, SUCLG1, and the GTP-forming β subunit, SUCLG2 (Figure 1). The tissue-specific loss of the entire enzymatic complex provides a unique opportunity to study consequences of SCS deficiency in the brain without compensatory elevation in the alternative β-isoform (SUCLG2) seen in previously reported models.¹⁴,³²,³³

Furthermore, the metabolic profile of the SCS-deficient model (Figures 2C and 2F) mirrors the molecular hallmarks of human SCS deficiency, including methylmalonic aciduria and elevations of C3 and C4-DC acyl-carnitines.^13,20,34 However, mtDNA depletion is not observed in the Sucla2-deficient mouse forebrain. The absence of depletion may be a tissue-specific phenotype, as mtDNA depletion has only been identified in cultured fibroblasts, and liver and skeletal muscle in patients,^7,12,13,19 and even so mtDNA depletion is not always present in human SCS deficiency with severe phenotype.³⁵ However, depletion has been observed in Sucla2-deficient cultured neurons and embryonic mouse brain.^14,33 An alternative hypothesis to consider is that SCS is required for mtDNA maintenance during development in actively dividing cells, and the postnatal onset of Sucla2 deficiency via CamKIIα-Cre recombination in post-mitotic neurons bypasses this critical period.¹⁶ In addition, the current model of Sucla2-associated mtDNA maintenance is centered upon its association with mitochondrial NDPK-D (or NME4).³⁶ However, research has shown that, while NDPK-D expression in the mouse cerebrum is high during development, it is markedly decreased postnatally³⁷ and was not detectable at the protein level in the global proteomic data in this report (Data S3). With molecular mechanisms yet to be determined, nonetheless, this conditional KO model allows for in vivo investigations of metabolic and molecular perturbations of Sucla2 deficiency without the confounding influence of OXPHOS defects caused by mtDNA depletion.

Global hypersuccinylation contributes to the pathogenesis of SCS deficiency

There is a growing body of evidence that pathogenic phenotypes associated with SCS deficiency are not necessarily dependent on the presence of mtDNA depletion. For example, a recent study reported siblings with biallelic missense pathogenic variants in SUCLA2 associated with sensorineural hearing loss, encephalopathy, and muscle-specific reduced activities of pyruvate dehydrogenase and ETC complexes with normal levels of mtDNA.³⁵ Furthermore, hypersuccinylation was recently characterized in SUCLA2^−/− patient fibroblasts and functional significance was subsequently studied in sucla2^−/− zebrafish larvae, where transgenic overexpression of mitochondrial desuccinylase Sirt5 partially rescued mitochondrial respiration without rescuing mtDNA depletion.⁷ Given the significant elevation of succinyl-CoA observed within the SUCLA2-deficient cortex, this study aimed to identify and characterize the changes in the SCS-deficient succinylome and determine potential pathogenic mechanisms associated with global hypersuccinylation. Similarly to other recent reports, mitochondrial proteins were highly enriched for hypersuccinylation, with nearly 50% of the total hypersuccinylated proteome localizing to the mitochondria.^7,9,38 GO biological pathway analysis of the hypersuccinylated proteome in Sucla2-deficient cortex demonstrated significant enrichment for multiple mitochondrial pathways, including the TCA cycle and ETC (Figure 4B). Furthermore, a large proportion of highly hypersuccinylated proteins with multiple succinylation sites are mitochondrial localized (Figure 3H), suggesting that hypersuccinylation has a significant effect on mitochondrial and metabolic processes.

Targeted metabolomic analysis demonstrating altered amino acid and intermediate energy metabolism provide additional evidence of altered metabolism in the model (Figures 2C–2F). Interestingly, the amino acids with significant changes in steady-state concentrations are directly involved in TCA cycle metabolism. In addition, energy metabolism seems to be further affected, as increased succinylation of mitochondrial complex I correlates with functionally significant deficiency of the enzymatic complex without global loss in protein abundance (Figures 4E, 4F, and S4). Hypersuccinylation could affect complex I activity through several mechanisms. On one hand, the formation and/or stability of the holo-complex could be disrupted by the massive hypersuccinylation of complex I subunits themselves (Figure 4D), or the hypersuccinylation of assembly factors such as NDUFAF1 and ACAD9 (Data S2). Variants in both assembly factors are associated with complex I deficiency in humans.³⁹ Alternatively, it is also likely that complex I activity is inhibited by hypersuccinylation of the N-module, the site of NADH dehydrogenase activity. The three core subunits of the N-module, NDUFS1, NDUFB1, and NDUFV2 all have at least three hypersuccinylated lysine residues in the mutant cortex, with NDUFS1 and NDUFV1 having 18 and 13 sites, respectively (Data S2). In addition, 4 of the 13 hypersuccinylated lysines in NDUFV1 are among the most significantly enriched succinylation sites within the Sucla2-deficient cortex (Figure 4A). These data suggest that inhibition of the N-module may be a likely mechanism of complex I deficiency; however, identifying the exact mechanism requires further investigation.

In addition, not all hypersuccinylated metabolic enzymes demonstrate functional changes in activity. Both ATP synthase (CV) and CS are highly hypersuccinylated in the mutant cortex (Figures 4C and 4D); however, there is no observed alteration in activity in either enzyme (Figures S3E and S3F). While there have been conflicting reports suggesting decreased activity of CV with hypersuccinylation as a result of Sirt5 KO in mouse heart,⁴⁰ the results presented in the mouse cortex agree with that of Sadhukhan et al.,³⁸ which demonstrated no change in either CS or CV activity in the mouse heart, despite hypersuccinylation in the absence of Sirt5 desuccinylase activity. This suggests that potential functional consequences of lysine-succinylation are diverse and require further exploration. The potential increase and consequences of other acylation modifications must also be considered, given the increase in both propionyl- and methylmalonyl-carnitine species. However, the succinyllysine antibody used within this study has been shown to exhibit little to no cross-reactivity to these other acyl-species⁴¹; thus, functional results herein can be specifically correlated to succinyl modifications. Therefore, the functional deficit in complex I activity provides further evidence that hypersuccinylation is associated with ETC dysfunction in the context of SCS deficiency.

Moreover, the proteomic and transcriptomic datasets provide strong evidence for further SCS-deficient metabolic perturbations to explore. With the high proportion of hypersuccinylated proteins within metabolic pathways such as the TCA and ETC (Figure 4), the data also show enrichment for transcriptomic upregulation of these pathways via GO enrichment analysis of the RNA-seq data (Data S4), with upregulation of over 60% of the genes found within the ETC (Figure S4). In addition, of 1,140 mitochondrial localized proteins identified by MitoCarta, 365 (32%) show differential (FDR < 0.05) gene expression, with 354 (97%) being upregulated (Data S6). On a proteomic level, 121 of the 145 differentially expressed mitochondrial proteins are upregulated in the mutant cortex (Data S3). These data, along with the increased mtDNA content (Figure 2B), suggest the possibility of increased mitochondrial biogenesis within the Sucla2 mutant cortex. The upregulation of mitochondrial pathways may provide evidence for a protective mechanism and may be an attempt at compensating for the impaired mitochondrial function observed within the SCS-deficient model herein. However, proliferation of dysfunctional mitochondria may also exacerbate the pathogenesis of SCS loss. In summary, this model provides a useful tool for future studies of the interplay of protein succinylation and mitochondrial energy metabolism and regulation.

Changes in the extra-mitochondrial succinylome are associated with global changes in neuronal transcriptional regulation

While it is clear that succinylation has a significant effect on metabolism, with nearly half of the identified SCS-dependent changes in the succinylome localized to mitochondrial proteins, the significant proportion of hypersuccinylated proteins outside of the mitochondria must also be considered. Furthermore, the distributions of subcellular localizations are the same for all succinylated proteins identified regardless of enrichment (Figure 3G), suggesting that the mechanisms for SCS-related changes in the succinylome occur across the cell. This idea is further evidenced by upregulation of both the mitochondrial and nuclear localized sirtuins with desuccinylase activity, SIRT5 and SIRT7 (Table S4). While the reported localization data of succinylated proteins agrees with the current literature in the field,^7,9,38 it is still a striking observation, as the increased succinylation within this model is likely reliant on perturbations in the steady-state levels of succinyl-CoA generated by the TCA cycle. Therefore, although succinyl-transferases have been identified outside of the mitochondria,^42,43 the question remains as to how negatively charged succinyl-CoA is transported across the mitochondrial membranes. One hypothesis holds that the increase of acyl-carnitines observed in the SCS-deficient cortex allows for increase in transport of the fatty acyl-CoA species across the membranes via the carnitine shuttle (Figure 2F). Alternatively, an emerging body of literature provides evidence for the presence of anaplerotic enzymes outside of the mitochondria, including SCS.^44,45 Assuming that extra-mitochondrial levels of SCS are also perturbed by the conditional Sucla2 recombination, the recessive mutations in Sucla2 presented herein would also lead to a buildup of succinyl-CoA in other cellular compartments. Nevertheless, lysine-succinylation outside of the mitochondria likely has biological effects beyond mitochondrial metabolism that warrant further investigation.

In particular, while the GO enrichment analysis of upregulated genes within the Sucla2 mutant cortex revealed increased metabolic pathways, pathway analysis of genes with reduced expression revealed that neuronal pathways including nervous system development are downregulated in the SCS-deficient mice, suggesting that, in addition to metabolic dysfunction, SCS loss may also affect neuronal pathways directly involved in encephalopathic pathogenesis (Data S4). Furthermore, IHC staining of neuronal marker NeuN and microglial marker IBA1 revealed neuronal cell loss accompanied by microgliosis within the SUCLA2-deficient hippocampal area (Figure S5). Although no significant differences in cell proportions were noted within the cortical area (data not shown), these results suggest that loss of SCS has significant consequences on neuronal cell health. Therefore, deeper analyses into the changes in global chromatin landscape and gene expression were conducted to investigate these pathways.

There is a growing body of literature investigating the relationship between the succinylation of histones, chromatin dynamics, and gene transcription.^27,42,46,47 In the vast majority of these studies to date, histone succinylation is associated with the opening of chromatin and increased gene transcription. The data described in this report reveal global changes in chromatin accessibility in both enhancer and promoter regions in the Sucla2^−/− cortex, with a skewing toward less-accessible promoter regions (Figures 5A–5D). In addition, the skewing toward more closed chromatin is accompanied by computational evidence for an overall loss in neuronal transcriptional networks in the mutant forebrain (Figure 5L). Given the observations of hypersuccinylated histone marks detected on all five major histone proteins, change in abundance of some chromatin remodeling enzymes, and dysregulation of neuronal transcriptional networks, changes in the transcriptome resulting from SCS deficiency appear to result from a complex and dynamic interaction of multiple regulatory mechanisms that require further study.

Limitations of the study

Although there is considerable evidence provided that the CamKIIα-driven Cre recombination of Sucla2 within the murine forebrain significantly affects neurons, this study is limited in that the data presented herein were collected within the whole cortex. Therefore, the contribution of and consequences to other cell types within the Sucla2 mutant forebrain cannot be specifically assessed with the datasets described. For example, mtDNA depletion is not present within the SUCLA2-deficient cortex; however, partial depletion within neurons may be masked by mtDNA content within other cell types. Therefore, it is possible that compensatory mechanisms, such as potentially increased mitochondrial biogenesis, may be occurring in other cell types. Future studies could include single-cell technologies to investigate cell-type-specific molecular, transcriptomic, and proteomic phenotypes within the SUCLA2-deficient brain. Furthermore, although there is histological and transcriptomic evidence for disruption of neuronal networks within the mutant forebrain (Figures 5 and S5), extended investigations on both the molecular and whole-animal level are required. Specifically, this study would significantly benefit from whole-animal behavioral phenotyping to link neuronal molecular mechanisms to potential cognitive impairment. Similarly, while the data presented clearly associate hypersuccinylation of mitochondrial complex I with functionally reduced enzymatic function, this work is limited in the investigation of how total mitochondrial respiration is affected. Thus, deeper analysis of mitochondrial function and whole-animal phenotype will help link the molecular and cellular data to clinical phenotypes observed in SCS deficiency.

Pathogenic variants in SCS cause a severe mitochondrial disorder for which the pathomechanisms are incompletely understood. The generation and validation of this mouse model presents a unique tool to study the molecular phenotypes and pathogenic mechanisms of SCS deficiency. This model also presents an opportunity to study the effects of a relatively recently identified PTM, lysine succinylation, and its role in both epigenetic regulation of gene expression and on primary protein structure and function in a tissue-specific context. Evidence suggests that increased succinylation is involved in the pathogenesis of SCS deficiency. In this model, hypersuccinylation is associated, at least in part, to perturbations in cellular energy metabolism. In addition, remodeling of the chromatin landscape and an altered transcriptomic profile links Sucla2 mutations to loss of neuronal transcriptional regulation, revealing another mechanism for potential neuronal dysfunction in SCS deficiency. Ultimately, this mouse model presents forebrain-specific SUCLA2 deficiency with metabolic, biochemical, and functional hallmarks reflective of clinical phenotypes in humans, and future investigations within this model hold the potential for further delineation of the mechanisms and role of protein succinylation, its relationship to SCS deficiency, and its potential as a therapeutic target for metabolic dysfunction.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Brett H. Graham (bregraha@iu.edu).

Materials availability

No new or unique reagents were generated within this study.

Data and code availability

• All raw proteomic data, both global and succinyl-enriched, and search results are available via massIVE: https://massive.ucsd.edu/, with the login username: MSV000089697_reviewer and password: succinyl. All transcriptomic raw data, both bulk RNA-sequencing results and ATAC-sequencing results, have been deposited in NCBI’s Gene Expression Omnibus (GEO)⁴⁸ and are accessible through the GEO Series SuperSeries accession number GSE208692. All other data reported are provided within supplemental datasets or will be shared by the lead contact upon request.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Animals

All experiments were approved by the Institutional Animal Care and Use Committee (IACUC) at Indiana University. Mice were bred and housed in the mouse facility at Indiana University School of Medicine in Indianapolis, Indiana. Animals were maintained in a 12/12-h light/dark cycle, consisting of 12hr-ON 7a.m.-7p.m., followed by 12hr-OFF. Room temperatures were maintained at 19^◦C–24^◦C (65–77^◦F) with 40–60% humidity. All mice were housed in positive pressure, individually ventilated cages (PIV). Standard autoclaved 6% fate diet (Purina Lab Dried 5K52) was available to the mice ad libitum, as was water with acidity regulated from pH 2.5–3.0. All breeder and experimental mice were housed in the same room and were aged together, with ages of mice for each experimental protocol appropriately outlined within figure legends. All mice were bred under the C57Bl/6J Mus musculus genetic background, with precise Sucla2 genotypes indicated within figure legends. Where both males and females were used, no statistical influence of sex was observed.

METHOD DETAILS

Generation of genetic alleles

The Sucla2 floxed mouse line was generated by the Genetically Engineered Rodent Models (GERM) Core at Baylor College of Medicine (BCM) under IACUC-approved protocols. The Sucla2 allele with LoxP sites flanking exons 3 and 4 (Figure 1A) was generated using the CRISPR/Cas9 strategy using 5′ and 3′ guide RNAs and single-stranded oligonucleotides (ssODN) for homology-based repair. The guide RNAs (complexed with Cas9 protein into ribonucleoproteins) were mixed with the ssODN repair templates and electroporated into C57BL/6J embryos as previously described.⁴⁹ Both the desired floxed allele as well as a constitutive deletion resulting from aberrant non-homologous end-joining of the Cas9-mediated double-stranded breaks were identified in progeny from founders and established as separate lines by backcrossing to C57BL/6J wild-type animals. The sequences for generating the guide RNA are as follows (PAM sequence underlined): 5′ guide = AAAGGGATAGACCTTTCATATGG; 3′ guide = 3′ guide = TCCACTGTAATACGTTTCATGGG. The genomic location of the 5′ guide is 14:73568447–73568469, Mouse genome build GRCm38, and the genomic location of the 3′ guide is 14:73569206–73569228, mouse genome build GRCm28. The sequences for the ssODNs are provided in the key resources table.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-IBA1	Abcam	RRID: AB_2636859
Anti-NeuN	Nobus	RRID: AB_11036146
Anti-succinyllysine	PTM Biolabs	RRID: AB_2937058
Anti-SUCLA2	GeneTex	RRID: AB_1952096
Anti-SUCLG2	GeneTex	RRID: AB_1952102
Anti-SUCLG1	GeneTex	RRID: AB_1952099
Anti-GAPDH	GeneTex	RRID: AB_11174761
Alexa Fluor 488 AffiniPure Donkey Anti-Rabbit	Jackson Laboratory	RRID: AB_2340619
Rhodamine Red-X-Donkey Anti-Mouse	Jackson Laboratory	RRID: AB_2340832
Anti-succinyllysine antibody beaded agarose	PTM Biolabs	Cat.No.: PTM-402
Chemicals, peptides, and recombinant proteins
TaqMan Fast Advanced Master Mix	ThermoFisher	Cat.No.: 4444557
OCT Compound	Tissue-Tek	Cat.No.: #4583
Hoechst 33342	Thermo Scientific	Cat.No.: PI62249
Aqua-Poly/Mount Reagent	Polysciences	Cat.No.: 18608
Trichloroacetic Acid	Sigma Aldrich	Cat.No.: T6399
Adenosine diphosphate	Sigma Aldrich	Cat.No.: A2754
Guanosine diphsophate	Sigma Aldrich	Cat.No.: G7127
5,5′-Dithiobis(2-nitrobenzoic acid)	Sigma Aldrich	Cat.No.: D8130
Succinyl-CoA	Sigma Aldrich	Cat.No.: S1129
Immobilon Western Chemiluminescent HRP Substrate	Millipore	Cat.No.: WBKLS0500
Tris(2-carboxyethyl)phosphine hydrochloride	Sigma Aldrich	Cat.No.: C4706
Chloroacetamide	Sigma Aldrich	Cat.No.: C0267
Trypsin/Lys-C	Promega	Cat.No.: V5072
NP-40	IBI Scientific	Cat.No.: 9016–45-9
Trifluoroacetic acid	Alfa Aesar	Cat.No.: LO6374
Tandem Mass Tag Pro Reagent, Isobaric Label Reagent Set	ThermoFisher	Cat.No.: 44520
Lysozyme, chicken egg white	Millipore	Cat.No.: 4403
Acetonitrile	ThermoFisher	Cat.No.: LS122500
EZ PREP	Sigma Aldrich	Cat.No.: NUK101
KAPA HiFi Hotstart ReadyMix	Roche	Cat.No.: KK2601
AMPure XP Beads	Beckman Coulter	Cat.No.: A63880
Critical commercial assays
Custom ND2 TaqMan Assay	ThermoFisher	(No Cat.) AR323VF
Predesigned B2M TaqMan Assay	ThermoFisher	Mm00542418_cn
Predesigned Sucla2 TaqMan Assay	ThermoFisher	Mm01310541_m1
Predesigned Suclg2 TaqMan Assay	ThermoFisher	Mm01182162_g1
Predesigned Suclgl TaqMan Assay	ThermoFisher	Mm00451244_m1
Predesigned Gapdh TaqMan Assay	ThermoFisher	Mm99999915_g1
Succinate Kit	Abcam	Cat.No.: Ab204718
Pierce BCA Assay	ThermoFisher	Cat.No.: 23227
Complex I Activity Assay	Sigma-Aldrich	Cat.No.: MAK359
NADH/NAD Kit	Sigma-Aldrich	Cat.No.: MAK037
Bradford Protein Assay	BioRad	Cat.No.: 5000006
Pierce Quantitative Colorimetric Assay	ThermoFisher	Cat.No.: 23275
Pierce High pH Reversed-phase Peptide Fractionation Kit	ThermoFisher	Cat.No.: A32993
KAPA mRNA Hyperprep Kit	Roche	Cat.No.:
Nextera DNA Library Preparation Kit	Illumina	Cat.No.: FC-131–1024
Deposited data
Proteomic Data	massIVE: https://massive.ucsd.edu/	Username: MSV000089697_reviewer Password: succinyl
Transcriptomic Data	Gene Expression Omnibus	Accession: GSE208692
Experimental models: Organisms/strains
CamKIIα-Cre Mouse Line	Jackson Laboratories	RRID: IMSR_JAX:027400
Oligonucleotides
5′ ssODN (LoxP Sites Underlined): TTGCCAAAAATTTTC AGTATACTATTTCAGAAATGATCTAAAATAAGTGCTTAA TCTTTTAAAGGGATAGACCTTTCTCGAGATAACTTC GTATAGCATACATTATACGAAGTTATCATATGG TGAATAATGTTTTAGTTTTCAAATCTAGAAAAATCTATG TATGTGGAATGACTTAGTCTT	BCM	N/A
3′ ssODN (LoxP Sites Underlined): GAAAAATAATATG GTAGAATTTGACATGTAATATAGTTTCTAAAAAGC GTAGGGGAATT CCACTGTAATACGTTCTCGAGATAAC TTCGTATAGCATACATTATACGAAGTTATTCATGGG GTTTTCAAATGATTAATTCTTTAGACGTACCTTAAAT CTCTGGAAATTGTATTTGGGTTA	BCM	N/A
Sucla2 Forward Primer 1: GCCTGGTTCCACTGGAGGAA	Sigma	N/A
Sucla2 Forward Primer 2: GACGTATTATCCATTACTCAC	Sigma	N/A
Sucla2 Reverse Primer: GCAGAAAAAGAACAGGATGC	Sigma	N/A
Cre Forward Primer: CCGGTCGATGCAACGAGTGAT	(Schuler et al.)50	N/A
Cre Reverse Primer: ACCAGAGTCATCCTTAGCGCC	(Schuler et al.)50	N/A
Software and algorithms
Fiji Software and ImageJ Densitometry v1.53t	NIH	RRID: SCR_003070
CellProfiler Program v4.25	Broad Institute	RRID: SCR_007358
Xcalibur Software v4.3	ThermoFisher	RRID: SCR_014593
Proteome Discoverer V2.5	ThermoFisher	RRID: SCR_014477
Cytoscape STRING-App	(Shannon et al.)21	RRID: SCR_003032
FastQC v.0.11.5	Babraham Bioinformatics	RRID: SCR_014583
STAR RNA-seq Aligner v.2.5	(Dobin et al.)62	RRID: SCR_005622
Bamutils from ngutils v.0.5.9	(Breese et al.)63	N/A
FeatureCounts v.1.5.1	(Liao et al.)64	RRID: SCR_012919
Limma v.3.38.3	(Ritchie et al.)65	RRID: SCR_010943
EdgeR v.3.30.3	(Robinson et al.)66	RRID: SCR_012802
Bowtie2	(Langmead et al.)68	RRID: SCR_016368
Picard	Broad Institute	RRID: SCR_006525
TrimGalore	Babraham Bioinformatics	RRID: SCR_014583
DeepTools	(Ramirez et al.)69	RRID: SCR_016366
MACS2	(Zhang et al.)70	RRID: SCR_013291
Sasquatch	(Schwessinger et al.)71	N/A
PECA	(Duren et al.)28	N/A
Other
CryoStat	LEICA Biosystems	Cat.No.: CM 1860
TintoRetriever Pressure Cooker	Bio SB	Cat.No.: BSB7008
THUNDER Imager Fluorescence Microscope	LEICA Biosystems	Cat.No.: DMi8
0.1mm Zirconium Oxide Beads	NextAdvance	Cat.No.: ZROB10-RNA
Bullet Blender StormPro	NextAdvance	Cat.No.: BT24M
C18 Sep-Pak Cartridges for Succinyl-CoA Measurement	Waters	Cat.No.: WAT020515
Sep-Pak Cartidges for Proteomic Peptide Purification	Waters	Cat.No.: WAT054955
Covaris cryoPREP	Covaris	Cat.No.:
Bioruptor Sonication System	Diagenode Inc.	Cat.No.: B01020001
EASY-nLC HPLC System	ThermoFisher	RRID: SCR_014993
Orbitrap Eclipse Mass Spectrometer	ThermoFisher	N/A
Aurora Columns	IonOpticks	Cat.No.: AUR2–25075C18A

Genotyping

DNA was purified from mouse tail snips via ethanol precipitation. Sucla2 genotypes were determined by a multiplexed PCR reaction with a common reverse primer and two allele-specific forward primers. All primer sequences are provided within the key resources table. Agarose gel electrophoresis was run on PCR products and Sucla2 alleles were distinguished by band size visualized by ethidium bromide staining (WT, 273bp; Flox, 313bp; Deletion, 373bp). The presence of absence of the transgene was determined by Cre-specific primers as published.⁵⁰

qPCR and RT-qPCR

Relative mtDNA content of cerebral cortex tissue was measured using real-time TaqMan qPCR with genomic DNA isolated via ethanol precipitation. A region of mitochondrial gene ND2 (mtDNA) was amplified in duplex with a region of β2-macroglobulin (B2M) used as a nuclear gene (nDNA) normalizer. Both TaqMan assays were purchased from ThermoFisher and are outlined in the key resources table. Relative mtDNA content (mtDNA/nDNA ratio) was calculated using the formula: mtDNA content = 1/2^ΔCt, where ΔCt = Ct_mtDNA – Ct_nDNA.⁵¹

Total RNA was isolated from mouse cortex using the Quick-RNA Miniprep Plus kit (Zymo Research R1057), and cDNA was prepared using the GoScript Reverse Transcriptase System kit (Promega: A5000). The TaqMan Fast Advanced Master Mix was used to run qPCR assays for all three SCS components. Pre-designed TaqMan assays for the SCS components with FAM-MGB fluorescent dyes were purchased from ThermoFisher as outlined within the key resources table. All assays were run in duplex with a predesigned assay for Gapdh with VIC-MGB fluorescent dye as a control. The fold change of SCS components was measured using the ΔΔCt method.

Immunohistochemistry methods

Sample collection

Mice were anesthetized via 250 mg/kg intraperitoneal injection with tribromoethanol (Avertin) and perfused with cold 0.1M PBS, and then the brain was collected. For each mouse, one hemibrain was fixed in 4% paraformaldehyde overnight at 4^◦C followed by: storage in 30% sucrose in PBS (pH 7.4) solution at 4^◦C for study groups. The other hemibrain was dissected, and cortical and hippocampal regions were flash-frozen, then stored at −80^◦C until further analysis.

Histological analysis

The fixed hemibrains were frozen embedded to OCT Compound, and then serial coronal sections (20 μm thickness) were obtained using a cryostat. Sections were stored in a cryoprotectant solution (50% glycerol in 0.1M PBS) at −20^◦C. Three sections spaced 460 μm apart were used for each staining procedure.

For comprehensive immunofluorescence staining, sections were blocked with PBS containing 5% normal goat serum (NGS) at room temperature for 1 h, incubated with polyclonal anti-IBA1 (1:1000), anti-NeuN (1:1000), and anti-Succinyllysine (1:200) antibodies in PBS containing 2.5% NGS at 4^◦C overnight. For SUCLA2 immunofluorescence staining, sections were steamed for 10 min in 10 mM sodium citrate buffer for antigen retrieval using a TintoRetriever Pressure Cooker. After antigen retrieval, sections were blocked with PBS containing 5% NGS at RT for 1 h, incubated with polyclonal anti-SUCLA2 (1:200) antibody in PBS containing 2.5% NGS at overnight.

The following day, sections were incubated with Alexa Fluor 488 AffiniPure Donkey Anti-Rabbit (1:400, for anti-SUCLA2, -Succinyllysine, and -IBA1) and Rhodamine Red-X-Donkey Anti-Mouse (1:400, for anti-NeuN) antibodies at RT for 1 h. Next, sections were incubated with 20nM Hoechst 33342 for nuclear counter staining at RT for 10 min, then mounted on slides with Aqua-Poly/Mount reagent. All images were obtained with THUNDER Imager fluorescence microscope.

Quantification of Image analyses

For each mouse, three sections were used to obtain a single average data per mouse. SUCLA2 localization of cell marker was analyzed using the Cell Counter Plugin and Succinyllysine-immunoreactivity (IR) was quantified using the Area Integrated Intensity analysis in the Fiji software.^52,53

Microglial, astrocytes, and neuronal positive (+) cells counting and percentage of each positive signal load were quantified using the CellProfiler program.⁵⁴ Quantification data were expressed as the percent of positive cell in cell type, immunoreactivity (mean fluorescence intensity, MFI), number of positive cells/mm², and percent of positive cell load. These metrics were then compared between the genotypes.

Measurement of metabolites

Whole cell tissue extracts were made via bead homogenization with 1.0mm Zirconium Oxide beads in the Bullet Blender in 1:6 w/v 80% methanol. Analyte extractions were amino acid, methylmalonic acid, and acylcarnitine analyses completed using stable isotope dilution liquid chromatography-tandem mass spectrometry (LC-MS/MS). For amino acid analyses, clarified supernatant was analyzed using a biphasic LC-MS/MS approach first with a HILIC separation (modified from Prinsen et al., 2016)⁵⁵ and next using a mixed-mode chromatographic separation.⁵⁶ Methylmalonic acid quantification was completed following the method described by Pedersen, T.L. et al., 2011),⁵⁷ and the measurement and quantification of acylcarnitines was completed as described by Luna et al.⁵⁸

Succinyl-CoA was measured as described by Smestad et al., 2018,⁵⁹ with modifications. Whole cell tissue lysates were generated from murine cortex homogenized in 4.5 w/v 10% trichloroacetic acid and immobilized on pre-conditioned C18 Sep-Pak cartridges. After 3 washes with 2mL sterile water, the immobilized metabolites were eluted twice with 750μL of pure methanol. Eluents were dried with a nitrogen evaporator and reconstituted in 100μL 0.1M sodium hydroxide. The reconstituted samples were then incubated at 40^◦C for 1 h to hydrolyze CoA esters, then neutralized by the addition of 10μL 1M hydrochloric acid (HCl) and 20μL 1M tris-HCl, pH 8 to stabilize pH. Concentrations of succinate were then quantified as a measurement of succinyl-CoA generated from succinate in vitro by succinyl-CoA synthetase in the presence of free CoA and ATP per manufacturer protocols. Details for commercial assay are provided in key resources table.

Measurement of enzyme activities and NADH concentrations

Succinyl-CoA synthetase (SCS) activity was measured in whole cell lysates generated from murine cortex tissue. Lysates were pulse-sonicated five times on ice for 10 s, and total protein content was quantified via Pierce BCA Assay. SCS activity was measured in the forward direction. A reaction mixture of 140 μL total volume consisted of 50mM potassium phosphate buffer, pH 7.2, 10mM magnesium chloride, 2mM ADP, 2mM GDP, or Water, and 20μg total protein lysate were mixed and incubated at 37^◦C. Each sample was run with both nucleotide mixes as well as a no nucleotide control. The reaction was started by adding 5,5′-Dithiobis(2-nitrobenzoic acid) (DTNB) and succinyl-CoA in quick succession to a final concentration of 0.2mM in solution. The absorbance at 412 nm was measured spectrophotometrically every 15 s for 5 min, and the enzyme activity was calculated in nmol TNB generated per minute per milligram of total protein. Complex I activity and subsequent NADH/NAD ratio were measured in mitochondrial preps (2.5ug total mitochondrial protein) according to manufacturers’ protocols. Mitochondria were isolated via differential centrifugation, with an initial slow spin at 1500 x g to pellet cellular debris and a secondary spin at 8000 x g to pellet the mitochondria. Isolations were completed in mitochondrial isolation buffer consisting of 210mM mannitol, 70mM Sucrose, 5mM HEPES, and 1mM EGTA.

Western Blotting

Whole cell lysates of mouse cerebral cortex were prepared in standard RIPA buffer. Total protein concentrations were determined via the Pierce BCA Protein Assay kit. 2.5μg of total protein was separated by size on a 12% SDS-polyacrylamide gradient gel. Proteins were transferred via electrophoresis to a polyvinylidene difluoride (PVDF) membrane for 90 min at 100 V. Membranes were blocked for 3 h in 5% milk-TBST and incubated overnight with primary antibodies. After three 15 min washes with TBST, membranes were incubated with HRP-conjugated goat anti-rabbit or goat anti-mouse secondary antibody diluted 1:30,000 in 5% milk-TBST. Secondary antibody was detected using the Immobilon Western Chemiluminescent HRP Substrate reagents, and band intensities were quantified using NIH ImageJ software. Primary polyclonal antibodies used for SUCLA2 (1:20,000), SUCLG2 (1:10,000), SUCLG1 (1:10,000) and succinyllysine (1:1,000) are outlined in the key resources table. Primary polyclonal mouse α-GAPDH (1:25,000) was used as a loading control.

Proteomic methods

Sample preparation, mass spectrometry analysis, bioinformatics, and data evaluation for quantitative proteomics and phosphoproteomics experiments were performed in collaboration with the Indiana University Center for Proteome Analysis at the Indiana University School of Medicine (IUSM) similarly to previously published protocols⁶⁰ unless otherwise stated.

Sample preparation

Brains (3 male WT, 3 male mutant, 3 female WT, 3 female mutant) were cryofractionated using a Covaris cryoPREP, and the resulting powder was sonicated in 8 M urea, 100 mM Tris-HCl, pH 8.5 using a Bioruptor sonication system with 30 s/30 s on/off cycles for 15 min in a water bath at 4^◦C. After subsequent centrifugation at 14,000 rcf for 20 min, soluble protein concentrations were determined by Bradford protein assay. The global proteomics workflow used 25 μg equivalent of protein from each sample and the succinyl-IP proteomics workflow utilized 1500 mg equivalent of protein from each sample. In both cases, proteins were reduced with 5 mM tris(2-carboxyethyl)phosphine hydrochloride (TCEP) for 30 min at room temperature and the resulting free cysteine thiols were alkylated with 10 mM chloroacetamide (CAA) for 30 min at room temperature in the dark. Samples were diluted with 50 mM Tris.HCl, pH 8.5 to a final urea concentration of 2 M for Trypsin/Lys-C based overnight protein digestion at 35^◦C (1:70 protease:substrate ratio, Mass Spectrometry grade).

Peptide purification and labeling

Digestions were acidified with trifluoracetic acid (TFA, 0.5% v/v) and desalted on Sep-Pak Vac cartridges with a wash of 1 mL 0.1% TFA followed by elution in 70% acetonitrile 0.1% formic acid (FA). Peptides for succinyl-peptide enrichment were lyophilized and then enriched.

Succinyl Peptide enrichment

An estimated 1500μg of Trypsin-digested peptides were resuspended in 200μL IP-buffer consisting of 100mM NaCl, and 1mM EDTA, 20mM Tris-HCl and 0.5% NP-40 substitute, pH 8.0. Separately, 20μL of drained anti-succinyllysine antibody beaded agarose were washed with 0.5mL cold PBS three times. The supernatant was then added to the pre-washed antibody conjugated beads and incubated overnight at 4^◦C with gentle end-to-end rotation. After incubation, the beads were pelleted by centrifugation at 500 x g for 30 s. Beads were washed twice with 0.5mL IP-buffer for 15 s by gentle inversion. Beads were washed a third time with IP-buffer that did not contain NP-40, and twice more with Milli-Q water. After the wash steps, the antibody-conjugated beads were pelleted by centrifugation at 500 x g for 30 s. Peptides were eluted off the beads with 100μL 0.1% trifluoroacetic acid. The elution step was repeated three times, and each elute was pooled together. All steps were repeated for each replicate sample.

TMT pro labeling

Peptides for the global experiment were dried by speed vacuum and resuspended in 24 μL of 50 mM triethylammonium bicarbonate pH 8.0 (TEAB). Peptide concentrations were checked by Pierce Quantitative colorimetric assay. The same amount of peptide from each sample was then labeled for 2 h at room temperature, with 0.25 mg of Tandem Mass Tag Pro (TMTpro) reagent (16-plex kit, manufactures instructions, Isobaric Label Reagent Set; see Table S5).⁶¹ Labeling reactions were quenched by adding 0.2% hydroxylamine (final v/v) to the reaction mixtures at room temperature for 15 min. Labeled peptides were then mixed and dried by speed vacuum.

Succinyl peptide samples were resuspended in TEAB, spiked with 15 ng chicken lysozyme peptides (digested and desalted in house), then labeled with 0.25 mg TMTpro reagent as above. These peptides were desalted on a Waters SepPak cartridge with a wash of 5% acetonitrile 0.1% FA and elution in 70% ACN 0.1% FA.

High pH Basic fractionation

Global peptides were reconstituted in 0.1% trifluoroacetic acid and fractionated on Sep-Pak Vac cartridges using methodology and reagents from Pierce High pH reversed-phase peptide fractionation kit (8 fractions including a 70% acetonitrile final fraction).

Nano-LC-MS/MS analysis

Nano-LC-MS/MS analyses were performed on an EASY-nLC HPLC system coupled to Orbitrap Eclipse mass spectrometer with a FAIMS pro interface. Approximately 1/8^th of each global peptide fraction and 1/8^th of the succinyl-peptide fraction was loaded onto a 25 cm Aurora column at 400 nL/min. Peptides were eluted from 5 to 30% with mobile phase B (Mobile phases A: 0.1% FA, water; B: 0.1% FA, 80% Acetonitrile) over 160 min, 30–80% B over 10 min; and dropping from 80 to 10% B over the final 10 min. The mass spectrometer was operated in positive ion mode with 3 FAIMS CVs (−40, −55, −70) and 1 s cycle time per CV. Data-dependent acquisition method with advanced peak determination and Easy-IC (internal calibrant) were used. Precursor scans (m/z 400– 1600) were done with an Orbitrap resolution of 120,000, RF lens% 30, maximum inject time 50 ms, standard AGC target, MS2 intensity threshold of 2.5e4, including charges of 2–7 for fragmentation with 30 s dynamic exclusion. MS2 scans were performed with a quadrupole isolation window of 0.7 m/z, 34% HCD CE, 50,000 resolution with the Orbitrap, 200% normalized AGC target, dynamic maximum IT fixed first mass of 100 m/z. The data were recorded using Xcalibur (4.3) software.

Proteome and succinyl-proteome analysis

Resulting RAW files were analyzed in Proteome Discover 2.5 with a Mus musculus UniProt FASTA (both reviewed and unreviewed sequences) plus common contaminants (49922 total sequences). Quantification methods utilized isotopic impurity levels available from Thermo Fisher Scientific for each TMT Pro lot used. For global data analysis SEQUEST HT searches were conducted with a maximum number of 3 missed cleavages; precursor mass tolerance of 10 ppm; and a fragment mass tolerance of 0.02 Da. Static modifications used for the search were carbamidomethylation on cysteine (C) residues. Dynamic modifications used for the search were oxidation of methionines, TMTpro label on the N-termini of peptides, TMTpro label on lysine (K) residues, succinyl on lysine (K) residues, and deamidation of N (max 3 dynamic mods were allowed for global and 4 for succinyl analysis). Dynamic protein terminus modifications allowed were: acetylation (N terminus), Met-loss or Met-loss plus acetylation (N terminus). Percolator False Discovery Rate was set to a strict setting of 0.01 and a relaxed setting of 0.05. IMP-ptm-RS node was used for calculation of modification site localization scores. In the consensus workflows, global data were normalized by total peptide amount with no scaling; succinyl enriched data were normalized by specific protein amount – chicken lysozyme that was used as a spike-in control for normalization. Co-isolation thresholds of 50% and average reporter ion S/N cutoffs of 7 were used for quantification of global peptides and 50% and 10 respectively for succinylpeptide data. Resulting normalized abundance values for each sample type, abundance ratio and log2(abundance ratio) values; and respective p values (ANOVA) from Proteome Discover were exported to Microsoft Excel. Global proteomics identified 6265 protein groups, 43441 peptide groups and 109945 PSMs. Succinyl Proteomics identified 4747 succinyl modified peptides out of a total of 21040 peptides identified in the experiment.

Gene ontology (GO) and cellular localization analysis

Analysis of GO terms for biological processes associated to succinylated proteins was performed within the Gene Ontology unifying biology PANTHER classification system. Gene IDs corresponding to accession numbers for the master proteins of succinylated peptides were input into the gene list for Mus musculus, and a statistical overrepresentation test for PANTHER GO-Slim Biological Processes was performed using Fisher Exact test with Bonferroni correction. Similar analysis was conducted for the transcriptomic data, using target genes that had lower transcriptional regulation in mutants as the input, and comparing to the Reactome Pathway enrichment. Cellular localization analyses were performed using Cytoscape’s STRING-App,^21,22 using the list of hyper-succinylated master protein symbols as the input, and data was analyzed as described by Yang, Y et al., 2022.⁹

RNA-sequencing

Total RNA was isolated from mouse cortex using the Quick-RNA Miniprep Plus kit and was first evaluated for its quantity and quality using Agilent Bioanalyzer 2100. All the samples had good quality with RIN (RNA Integrity Number) greater than 8. 100 nanograms of total RNA was used for library preparation with the KAPA mRNA Hyperprep Kit. Each resulting uniquely dual-indexed library was quantified and quality accessed by Qubit and Agilent Bioanalyzer, and multiple libraries were pooled in equal molarity. The pooled libraries were sequenced with 2 × 150bp paired-end configuration on an Illumina NovaSeq 6000 sequencer using the v1.5 reagent kit.

The sequencing data were first assessed using FastQC for quality control. All sequenced libraries were mapped to the human genome (UCSC mm10) using STAR RNA-seq aligner (v.2.5)⁶² with the following parameter: “–outSAMmapqUnique 60”. The read distribution across the genome was assessed using bamutils (from ngsutils v.0.5.9).⁶³ Uniquely mapped sequencing reads were assigned to mm10 refGene genes using featureCounts (subread v.1.5.1)⁶⁴ with the following parameters: “-s 2 –p –Q 10”. Each sample was analyzed independently and genes with read count per million (CPM) < 0.5 in more than the number of sample replicates in one group were removed from the comparisons. The data was normalized using TMM (trimmed mean of M values) method. Multi-dimensional scaling analysis was done with limma (v.3.38.3).⁶⁵ Differential expression analysis was performed using edgeR (v.3.30.3).^66,67 False discovery rate (FDR) was computed from p values using the Benjamini-Hochberg procedure.

ATAC-sequencing and PECA analysis

Fresh frozen mouse cortex tissues were cut into pieces <0.5 cm and were homogenized using a glass dounce tissue grinder (25 times with pastel A and 10 times with pastel B) in 2 mL of ice-cold EZ PREP and incubated on ice for 5 min. Dissociated tissues were centrifuged at 500 g for 5 min at 4^◦C, washed with 2 mL ice-cold EZ PREP and incubated on ice for 5 min. After centrifugation, the nuclei were washed in 2 mL nuclei suspension buffer (NSB) consisting of 1× PBS, 0.01% BSA. Isolated nuclei were resuspended in 2 mL NSB, filtered through a 30-μm cell strainer, and counted. The nuclei were pelleted and resuspended in Tn5 enzyme and transposase buffer. The Nextera libraries were amplified using the Nextera PCR master mix and KAPA biosystems HiFi hotstart readymix successively. AMPure XP beads were used to purify the transposed DNA and the amplified PCR products. The final dual-indexes libraries were quality assessed and quantified using Agilent Bioanalyzer. The libraries were sequenced on an Illumina NovaSeq 6000 and paired-end 50 bp reads were generated Illumina adapter sequences and low-quality base calls were first trimmed off the sequence reads with Trim Galore (http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). The resulting high-quality reads were aligned to the mouse reference genome mm10 using bowtie2.⁶⁸ Duplicate reads were marked with Picard (https://broadinstitute.github.io/picard/). Reads mapped to mitochondrial DNA together with low mapping quality reads (MAPQ<10) were excluded from further analysis. To assess the overall ATAC-seq signal enrichment, fractions of reads in peaks (FRiP) were calculated, and Fingerprint and average profile plots around transcription starting sites (TSS) were generated with deepTools.⁶⁹ MACS2⁷⁰ was used to identify peaks. To visualize the ATAC-seq signals, cutting sites were piled up with sasquatch⁷¹ and BigWig files of cutting sites were generated; Normalized bigwig files were also generated using the deepTools2 bamCoverage function. To identify differentially accessible regions, the reads fallen in the promoter regions and the mouse cortex-specific enhancers regions were quantified with featureCounts (subread v.1.5.1).⁶⁴ The mouse cortex enhancers were downloaded from EnhancerAtlas 2.0 (http://www.enhanceratlas.org/downloadv2.php). Differential chromatin accessibility analysis between the Sucla2 conditional KO and control samples was performed using edgeR (v.3.12.1).^66,67 False discovery rate (FDR) was computed from p values using the Benjamini-Hochberg procedure.

PECA²⁸ was next applied to infer the gene regulation relationships between transcription factors with target genes from our paired gene expression and chromatin accessibility data. The aligned filtered BAM files from the above ATAC-seq data and the gene expression count files from the RNA-seq data were used as inputs for the PECA analysis. The gene expression files included two columns with the official gene symbols in the first column and the FPKM normalized gene expression levels in the second column. Each pair of BAM and gene expression files from the same sample were input into PECA bash script ‘PECA.sh’ to infer the regulatory relationship with genome option mm10. The comparison between groups of networks was performed with PECA bash script ‘PECA_compare_dif_multiple.sh’ with the inferred results from the previous step.

QUANTIFICATION AND STATISTICAL ANALYSIS

All statistical results used can be found within figures and corresponding figure legends. Data in bar graphs are represented as means ± standard deviations. Where indicated, western blots, PCR experiments, and enzymatic assays were repeated at least twice independently, and data are pooled and normalized to the average of the experimental control group where indicated. All nonnormalized data are included within the supplemental files. In all experiments, the sample size (n), representing the number of biological replicates per experimental group, is also indicated within the figure legends. All datasets were subject to a Shapiro-Wilk test for normalization and Grubb’s outlier test with an alpha cutoff of 0.05, and any identified outliers were removed. Two-tailed unpaired t-tests were applied when comparing single measures between two groups. For data with multiple comparisons, one-way ANOVA was performed followed by a Tukey’s post-hoc test for discoveries. Pearson correlation analyses were conducted for transcriptomic datasets, and Fisher’s exact tests with Bonferroni corrections were performed in GO and pathway enrichment analyses. Correction for multiple testing via false discovery rates were also applied to the large -omics datasets. In all cases, statistical significance indicated within figures represents corrected p values. Relevant tests used are indicated in figure legends, and all statistical analyses were performed using GraphPad Prism version 9.1.2 unless otherwise noted within experimental methods.

Highlights.

• Tissue-specific Sucla2 mutagenesis yields mouse model of SCS deficiency

• Accumulation of succinyl-CoA in the mutant cortex leads to hypersuccinylation

• Hypersuccinylation of ETC complex I is associated with impaired enzymatic activity

• Nuclear hypersuccinylation is accompanied by changes in neuronal transcription

INCLUSION AND DIVERSITY

We support inclusive, diverse, and equitable conduct of research. One or more of the authors of this paper self-identifies as a gender minority in their field of research.