Downregulation of Keap1 Confers Features of a Fasted Metabolic State

Abstract

Transcription factor nuclear factor erythroid 2 p45-related factor 2 (Nrf2) and its main negative regulator, Kelch-like ECH-associated protein 1 (Keap1), are at the interface between redox and intermediary metabolism, allowing adaptation and survival under conditions of oxidative, inflammatory, and metabolic stress. Nrf2 is the principal determinant of redox homeostasis, and contributes to mitochondrial function and integrity and cellular bioenergetics. Using proteomics and lipidomics, we show that genetic downregulation of Keap1 in mice, and the consequent Nrf2 activation to pharmacologically relevant levels, leads to upregulation of carboxylesterase 1 (Ces1) and acyl-CoA oxidase 2 (Acox2), decreases triglyceride levels, and alters the lipidome. This is accompanied by downregulation of hepatic ATP-citrate lyase (Acly) and decreased levels of acetyl-CoA, a trigger for autophagy. These findings suggest that downregulation of Keap1 confers features of a fasted metabolic state, which is an important consideration in the drug development of Keap1-targeting pharmacologic Nrf2 activators.

Keywords: Human Metabolism; Molecular Biology; Omics.

Graphical abstract

Figure 1

Proteomic Analyses of Mitochondria-Enriched Preparations from Liver and Early-Passage Intestinal Organoids from Wild-Type (WT), Nrf2-Knockout (Nrf2-KO), and Keap1-knockdown (Keap1-KD) Mice (A and B) Coomassie-stained SDS-PAGE gel of protein samples prepared from mitochondria-enriched fraction from liver (A) or intestinal organoids (B) from mice with the indicated genetic alterations. (C) Principal-component analysis (PCA) of 906 proteins with intensities reported in all samples in all MS runs in both experimental systems. Normalized LFQ intensities were log₂ transformed and Z-scored by average log₂ LFQ before PCA. (D and E) Summary of the quantitative data from two MS runs each for the enriched mitochondria samples from liver (D) and intestinal organoids (E). Proteins are colored by statistical criteria for outlier status (see text for statistical methods). Red entries are described further in Tables S1 and S2. The indicated proteins were found in all four MS runs to be significantly different between Nrf2-KO and Keap1-KD genotypes. Gene names: Cbr1: carbonyl reductase [NADPH] 1, Ces1: liver carboxylesterase 1, Ces1f: carboxylesterase 1F, Creg1: protein CREG1, Ephx1: epoxide hydrolase 1, Gstm1: glutathione S-transferase Mu 1, H6pd: GDH/6PGL endoplasmic bifunctional protein, Htatip2: oxidoreductase HTATIP2, Ugt2b35: UDP-glucuronosyltransferase. All data can be found in Data S1, Quantitative proteomics data. See also Figure S1.

Figure 2

Clusters of Metabolic Proteins Identified by STRING Functional Group Enrichment Analyses (A and B) Network of proteins identified by STRING with both functional enrichments and quantitative relationships in mitochondria-enriched preparations from livers (A) and early-passage intestinal organoids (B) from Nrf2-knockout (Nrf2-KO) and Keap1-knockdown (Keap1-KD) mice. Networks created by STRING “Proteins with values/ranks” tool (Szklarczyk et al., 2019) were rendered in Cytoscape (Shannon et al., 2003) to overlay ratio values (colors). Gray proteins were not identified in the experiments but were included by STRING. For edges, a minimum interaction score of 0.4 (medium confidence) was applied, with disconnected nodes and subnetworks hidden. Some node positions have been adjusted in highly interacting regions to show names. This type of analysis identified clusters of metabolic proteins within the Ces1 and Ugt families in livers (A) and proteins within the Ces1 and Cyp families in intestinal organoids (B) as statistically significantly different between the Nrf2-KO and Keap1-KD genotypes. For other protein clusters, see Figures 3, S2, and S3.

Figure 3

Genetic Interference with Keap1/Nrf2 Affects the Abundance of Glycolytic Enzymes and Metabolites (A–E) (A) Networks created by STRING “Proteins with values/ranks” tool (Szklarczyk et al., 2019) were rendered in Cytoscape (Shannon et al., 2003) to overlay ratio values (colors). Gray proteins were not identified in the experiments but were included by STRING. For edges, a minimum interaction score of 0.4 (medium confidence) was applied, with disconnected nodes and subnetworks hidden. Some node positions have been adjusted in highly interacting regions to show names. ∗Edited networks have had proteins not identified in the proteomics analysis removed for brevity. Full details of STRING data can be found in Data S2. In addition to clusters of proteins shown in Figures 2B and S3, this type of analysis identified clusters of proteins involved in glycolysis and the pentose phosphate pathway as statistically significantly different between the Nrf2-KO and Keap1-KD genotypes of organoid preparations. (B–E) Concentration of glucose-6-phosphate (B), glucose-1-phosphate/fructose-6-phosphate (C), dihydroxyacetone phosphate (DHAP) (D), and glyceraldehyde 3-phosphate (E) in colon tissue of C57BL/6 mice. Green bars represent wild-type (WT) mice, red bars Nrf2-knockout (Nrf2-KO) mice, and blue bars Keap1-knockdown (Keap1-KD) mice. ∗ 0.05 > p > 0.01; ∗∗ 0.01 > p > 0.001. See also Figure S4 and Table S3.

Figure 4

Ces1 and Acox2 Are Transcriptional Targets of Nrf2 (A) mRNA levels for Ces1g in cultures (n = 3) of intestinal organoids from wild-type (WT), Nrf2-knockout (Nrf2-KO), and Keap1-knockdown (Keap1-KD) C57BL/6 mice. (B) mRNA levels for Nqo1, Gstp, Gclc, and Ces1g in cultures (n = 3) of intestinal organoids from WT C57BL/6 mice that had been treated with vehicle (0.1% acetonitrile) or TBE-31 (10 nM) for 16 h. (C) mRNA levels for Ces1g, Ces1f, and Nqo1 in colon tissue of WT, Nrf2-KO, and Keap1-KD C57BL/6 mice (n = 3). (D) Protein levels for Ces1g in colon tissue of WT, Nrf2-KO, and Keap1-KD C57BL/6 mice (n = 3). (E) mRNA levels for Ces1g in colon tissue of male C57BL/6 WT mice (n = 3–4) that had been treated with vehicle (1% DMSO in corn oil) or RTA-408, per os, 3 times, 24 h apart; colon tissue was harvested 6 h after the last dose. (F) mRNA levels for Acox2 in colon tissue of WT, Nrf2-KO, and Keap1-KD C57BL/6 mice (n = 5). (G) mRNA levels for Acox2 in colon tissue of male WT C57BL/6 mice (n = 3–4) that had been treated with vehicle (1% DMSO in corn oil) or RTA-408, per os, 3 times, 24 h apart; colon tissue was harvested 6 h after the last dose. (H) mRNA levels for Acox2 in cultured intestinal organoids (n = 3) from WT and Nrf2-KO C57BL/6 mice that had been treated with vehicle (0.1% acetonitrile, white bars) or TBE-31 (10 nM, black bars) for 16 h. (I) mRNA levels for Acox2 in the colon of female WT and Nrf2-KO C57BL/6 mice (n = 4–5). The animals were treated with TBE-31 (5 nmol/g body weight, 3 times, at 24-h intervals, per os, black bars) or vehicle (1% DMSO in corn oil, white bars) and fasted for 4 h before tissue harvesting 24 h after the last dose. ∗p < 0.05. See also Figures S5, S6, S7, and S11.

Figure 5

Nrf2 Alters the Lipidome (A and B) Levels of triglycerides in liver (A) and colon (B) of wild-type (WT), Nrf2-knockout (Nrf2-KO), and Keap1-knockdown (Keap1-KD) C57BL/6 mice (n = 4). ∗p < 0.05. (C) Orthogonal projection to latent structure-discriminant analysis (OPLS-DA) score plot from GC-MS data for total fatty acids. Seven independent biological replicates of colon tissue from WT (green), Keap1-KD (blue), and Nrf2-KO (red) C57BL/6 mice were included. (D) Plot showing the individual fatty acids driving the separation among the genotypes (loadings). The variables (i.e., fatty acids) are represented as green circles, while the direction of the discriminate variable from the classification matrix for each genotype (WT $M1-DA(1), Keap1-KD -$M1-DA(2), and Nrf2-KO $M1-DA(3)) is in blue.

Figure 6

Downregulation of Keap1 Decreases the Hepatic Levels of Acetyl-CoA at Fed State and Increases the Acetylation of α-tubulin Following Fasting (A) Concentration of phosphoenolpyruvate, fructose bis-phosphate, and acetyl-CoA in colons of wild-type (WT, green bars), Nrf2-knockout (Nrf2-KO, red bars), and Keap1-knockdown (Keap1-KD, blue bars) C57BL/6 mice. ∗∗ 0.01 > p > 0.001. (B) mRNA levels for Acly in organoids from wild-type (WT), Nrf2-knockout (Nrf2-KO), and Keap1-knockdown (Keap1-KD) C57BL/6 mice. ∗p < 0.01. (C and D) mRNA levels for Acly in livers (C) and colons (D) of wild-type (WT) and Keap1-knockdown (Keap1-KD) female C57BL/6 mice (n = 8) that were either fed ad libitum or fasted for 18 h; 18S used as a reference gene; ∗p < 0.01, in relation to fed state in respective genotype; ^$p < 0.01 and ^$$0.01 < p < 0.05, relative to respective WT. (E) Protein levels for Acly, AcK40-α-tubulin, and α-tubulin in livers from wild-type (WT) and Keap1-knockdown (Keap1-KD) female C57BL/6 mice (n = 7–8) that were either fed ad libitum or fasted for 18 h. (F) Levels of acetyl-CoA in livers of fed wild-type (WT) and Keap1-knockdown (Keap1-KD) female C57BL/6 mice (n = 8). See also Figures S8–S12.

Figure 7

Deletion of Nrf2 in the Context of Mutant Keap1, Which Does Not Suppress Nrf2, Decreases the Acetylation of α-tubulin and Autophagic Flux (A) Levels of NQO1, AcK40-α-tubulin, and α-tubulin in whole-cell lysates of A549 and Nrf2-KO A549 cells. (B) Levels of LC3B-I (non-lipidated form) and LC3B-II (lipidated form) in whole-cell lysates of A549 and Nrf2-KO A549 cells that had been treated with vehicle (0.1% DMSO, VEH) or 10 nM bafilomycin A1 (BAF) for 16 h. GAPDH served as a loading control. See also Figure S12. (C) Downregulation of Keap1 has features of a fasted metabolic state. Nrf2 channels glucose through the pentose phosphate pathway by upregulating glucose-6-phosphate dehydrogenase (G6pd) and the enzymes of the pentose phosphate pathway (PPP) and enhances fatty acid oxidation (FAO) in part by upregulating Ces1 and Acox2, as well as the fatty acid transporter Cpt1, while inhibiting fatty acid synthesis (FAS) by downregulating Acly, and thus decreasing the levels of cytosolic acetyl-CoA. These features of a fasted metabolic state channel acetyl-CoA into the mitochondria for ATP synthesis and increase autophagic flux. The upregulation of the PPP, isocitrate dehydrogenase-1 (Idh1), and malic enzyme-1 (Me1) provides reducing equivalents (NADPH) for redox reactions and regeneration of reduced glutathione (GSH), which is catalyzed by the Nrf2-target enzyme glutathione reductase (Gr).