The BCKDH Kinase and Phosphatase Integrate BCAA and Lipid Metabolism via Regulation of ATP-Citrate Lyase

Abstract

Branched-chain amino acids (BCAA) are strongly associated with dysregulated glucose and lipid metabolism, but the underlying mechanisms are poorly understood. We report that inhibition of the kinase (BDK) or overexpression of the phosphatase (PPM1K) that regulates branched-chain ketoacid dehydrogenase (BCKDH), the committed step of BCAA catabolism, lowers circulating BCAA, reduces hepatic steatosis, and improves glucose tolerance in the absence of weight loss in Zucker fatty rats. Phosphoproteomics analysis identified ATP-citrate lyase (ACL) as an alternate substrate of BDK and PPM1K. Hepatic overexpression of BDK increased ACL phosphorylation and activated de novo lipogenesis. BDK and PPM1K transcript levels were increased and repressed, respectively, in response to fructose feeding or expression of the ChREBP-β transcription factor. These studies identify BDK and PPM1K as a ChREBP-regulated node that integrates BCAA and lipid metabolism. Moreover, manipulation of the BDK:PPM1K ratio relieves key metabolic disease phenotypes in a genetic model of severe obesity.

Keywords: ATP-citrate lyase; branched-chain amino acids; diabetes; lipid metabolism; obesity; systems physiology.

Figure 1. Metabolic effects of BT2 treatment or PPM1K expression in Zucker fatty rats

(A) BCKDH activity in liver, heart and skeletal muscle (Skm) tissue of BT2 (20mg/kg i.p.) or vehicle (Veh)-treated Zucker fatty rats (ZFR). (B) Representative immunoblots of total and phospho-ser 293 of BCKDH e1a. Effects of BT2 on circulating branched chain amino acids (BCAA) (C) and branched chain keto acids (BCKA) (D). Body (E) and tissue (F) weights measured at the end of the study period. (G) Liver triacylglyceride content in BT2- and Veh-treated ZFR. Glucose (H) and insulin (I) excursions during a 1g/kg IP glucose tolerance test. Data in panels A–I are expressed as the mean ± SEM, n=8–10 animals per group. * P<0.05, ** P<0.01, *** P<0.001. Recombinant adenoviruses expressing human PPM1K (Ad-CMV-PPM1K) or GFP (Ad-CMV-GFP) were administered to 14 week-old Zucker fatty rats (ZFR) via tail vein. (J) Expression of human and endogenous (rat) PPM1K mRNA in liver. (K) Effect of each adenovirus on BCKDH activity in liver and heart tissue. (L) Representative immunoblots of total and phospho-ser 293 of BCKDH e1a, PPM1K, and GFP in liver. Effects of each adenovirus on circulating BCAA (M) and BCKA (N). Body (O) and tissue (P) weights measured at the end of the study period. (Q) Liver triacylglyceride content. Glucose (R) and insulin (S) excursions during a 1g/kg i.p. glucose tolerance test. Data in panels J–S are expressed as the mean ± SEM, n=6–10 animals per group. * P<0.05, ** P<0.01, *** P<0.001. See Supplemental Figures 1 and 2 for related information.

Figure 2. Phospho-proteomics reveals additional targets of BDK and PPM1K in liver

(A) Study workflow. Panels (B) and (C) show flanking amino acid sequences of all phosphosites downregulated by BT2 or Ad-CMV-PPM1K treatments, respectively. Thresholds of ≥ −0.585 Log₂ fold change in phosphorylation and statistical significance of P < 0.05 were used (n=3 samples per group). The modulated serine in each phosphoprotein is highlighted in red. Consensus phosphosite motif sequences generated for BT2 (D) and Ad-CMV-PPM1K (E) modulated phosphosites. (F) Representative immunoblot for phospho-ser454 and total ATP citrate lyase (ACL), BDK, and GAPDH proteins in liver tissues from BT2- or Veh-treated Zucker fatty rats (ZFR). (G) Representative immunoblot for phospho-ser454 and total ATP citrate lyase (ACL), PPM1K, and GAPDH in liver tissues from Ad-CMV-PPM1K- or Ad-CMV-GFP-treated ZFR. Representative immunoblots in panels F and G are shown alongside densitometric analyses of pACL/total ACL. Data are expressed as mean ± SEM from n=5 animals per group. ** P<0.01. See Supplemental Table 1 for additional details.

Figure 3. Subcellular localization of ACL, BDK and PPM1K and effect of BDK overexpression on ACL phosphorylation in vitro

(A) Representative immunoblots of ACL, BDK, PPM1K, the mitochondrial markers ETFA and COXIV, and the cytosolic marker GAPDH in cytosolic and mitochondrial fractions of liver from lean Wistar rats sacrificed in the ad-libitum fed or overnight fasted states. (B–C) Volcano plots showing the subcellular location of proteins containing phosphopeptides found to be downregulated by BT2 or Ad-CMV-PPM1K treatments, respectively. (D) Effect of Ad-CMV-BDK overexpression in Fao cells on ACL phosphorylation on ser454, total ACL, BCKDH e1a phosphorylation on ser293, total e1a, and BDK protein abundance. Densitometric analysis of pACL/ACL ratio is shown below the representative blot. Data are mean ± SEM representing n=3 independent experiments. ** P<0.01. (E) Confocal images of Hek293 cells transfected with plasmid encoding a GFP tagged BDK lacking the mitochondrial targeting sequence, under control of a CMV promoter (CMV-ΔMTS-BDK-GFP) or CMV-GFP control constructs co-stained with MitoTracker (red) and Hoechst (blue). (F) Effect of Ad-CMV-ΔMTS-BDK overexpression in Fao cells on ACL phosphorylation on ser454, total ACL, BCKDH e1a phosphorylation on ser293, total e1a, and BDK protein abundance. Densitometric analysis of pACL/ACL ratio is shown below the representative blot, as mean ± SEM of 3 independent experiments. ** P<0.01. (G) Studies with purified ACL, protein kinase A (PKA), and BCKDH subunit proteins. The lower panel demonstrates direct phosphorylation of ACL and the e1a subunit of BCKDH by both BDK and protein kinase A (PKA). The Coomassie stain of the same gel is shown in the upper panel. See Supplemental Figure 3 for related experiments.

Figure 4. BDK phosphorylates ACL and activates de novo lipogenesis in vivo

(A) Representative immunoblot of phospho-454 and total ATP citrate lyase (ACL) in unfractionated liver samples from the same fasted or fed Wistar rats used for the fractionation study shown in Figure 3A. (B) ACL phosphorylation on ser454, total ACL, and BDK protein abundance in liver of Ad-CMV-BDK or Ad-CMV-βGAL-treated Wistar rats. (C) Densitometric analysis of pACL/ACL ratio. (D) Effect of Ad-CMV-BDK on rates of de novo lipogenesis (DNL) measured as incorporation of D₂O into newly synthesized palmitate in liver. (E) D₂O enrichment in plasma of Ad-CMV-BDK and Ad-CMV-βGAL-treated rats at sacrifice. (F) Body weights in Ad-CMV-BDK and Ad-CMV-βGAL-injected rats. Data in panels C–F are expressed as mean ± SEM, n=4–6 rats per group. ** P<0.01. (G) Dual localization of BDK and PPM1K in the cytosolic and mitochondrial subcellular compartments enables these enzymes to simultaneously modify the phosphorylation states of ACL and BCKDH, resulting in coordinated regulation of lipid and BCAA metabolism.

Figure 5. Transcriptional regulation of BDK and PPM1K by ChREBP

(A) Conservation across mammalian species of an enhancer containing a ChREBP binding site proximal to the human BDK gene. The red arrow locates the ChREBP binding site at a multicolored H3K27Ac “peak” which is indicative of an active regulatory element. Vertical black hatch marks to the right of each mammal indicates conserved sequence relative to the human genome. Note the absence of the element in mice, and its retention in rats. (B) ChREBP-β mRNA expression is positively correlated with BDK mRNA expression in liver biopsies taken from 86 overnight fasted human subjects with non-alcoholic fatty liver disease (NAFLD). (C) Effects of 4 hours of refeeding of high fructose (60% fructose) or standard chow diets to overnight fasted Wistar rats on hepatic transcript levels of known ChREBP response genes, as well as BDK and PPM1K. Data are mean ± SEM, n=5 rats per group. ** P <0.01, *** P<0.001, **** P<0.0001, ***** P<0.00001, ****** P<0.000001. (D) Mouse (m) ChREBP-β and rat (r) Bckdk (BDK), and PPM1K mRNA expression in liver of Ad-CMV-mChREBP-β or Ad-CMV-GFP-treated Wistar rats. Data are mean ± SEM, n=6–8 rats per group. ** P <0.01, *** P<0.001. (E) Schematic summary showing that fructose feeding activates ChREBP-β to drive transcription of the lipogenic program (component genes shown in burgundy), now including BDK as a post-translational activator of the pathway. ChREBP-β induction also leads to repression of PPM1K expression.