With regard to the first question subcellular fractionation

With regard to the first question, subcellular fractionation studies revealed that BDK and PPM1K are clearly detectable in both the mitochondrial and cytosolic subcellular fractions, thus making it possible for these enzymes to interact with both the BCKDH and ACL substrates. The preferential presence of BDK in the cytosol is consistent with the rather low copy number of BDK bound to the 24-mer transacylase (E2) core of mitochondrial BCKDH from rat liver (Shimomura et al., 1990). We also show that a BDK variant lacking its mitochondrial targeting sequence is expressed in the cytosol, where it phosphorylates ACL in a BCKDH-independent manner. As to the second question, phosphorylation of Ser454 on ACL is clearly increased in the fasted to fed transition. Interestingly, this increase is accompanied by a decline in the level of PPM1K protein in the cytosolic, but not the mitochondrial compartment. Concerning the third question, analysis of the amino AZD7687 sequence of ACL and comparison with other peptides identified in our phosphoproteomics screen suggests that it could be directly regulated by both BDK and PPM1K due to the presence of a dual BDK-PPM1K motif surrounding the regulatory phosphosite. Moreover, studies summarized in Figure 3 with purified proteins demonstrate direct phosphorylation of ACL on Ser454 by BDK. Collectively, these data provide support for a previously unappreciated role for BDK and PPM1K in the regulation of hepatic lipid metabolism. The literature concerning ACL regulation is scattered over the past 20 years, and in many ways does not present a coherent picture. There is evidence that the enzyme becomes phosphorylated in response to catabolic signaling effectors (e.g., isoproterenol and PKA) as well as anabolic mediators (insulin and Akt) (Berwick et al., 2002, Potapova et al., 2000). However, these studies have been performed in different tissues, and the physiologic significance (or lack thereof) of multiple mechanisms for regulation of ACL has never been fully explored. For example, it is very unlikely that PKA, an enzyme activated by glucagon and other catabolic effectors associated with the fasted state, would play a physiologic role in increasing hepatic ACL phosphorylation and activity in anabolic, fed conditions. On the other hand, the increase in ACL phosphorylation that occurs in the transition from the fasted to the fed state could reasonably be mediated by insulin signaling through the Akt pathway. Our current findings demonstrate that modulation of the BDK/PPM1K ratio affects ACL phosphorylation in an Akt-independent fashion, both in isolated cells, and in liver of living animals. Importantly, just as Akt can be activated by insulin in the anabolic state, here we show that levels of cytosolic PPM1K protein decrease in response to feeding, consistent with a physiological role of this new mechanism. Obesity is a setting in which “selective insulin resistance” appears, a scenario where insulin fails to suppress hepatic glucose output, but continues to promote lipogenesis (Titchenell et al., 2016). Increases in the hepatic BDK:PPM1K ratio, as documented in rodent models of obesity (She et al., 2007, Lian et al., 2015), and here in response to fructose refeeding, may cause ACL to be constitutively phosphorylated, such that it no longer responds to fasting in the manner demonstrated in lean rats (Figure 4A). This model also aligns with our findings linking the global metabolic transcription factor ChREBP with expression of BDK and PPM1K. The ChREBP-β isoform is a particularly potent activator of lipogenesis in liver that is induced by excess consumption of sucrose as found in soft drinks and other sugar-containing foods common in western diets (Herman et al., 2012, Kim et al., 2016). These findings lead us to propose a disease pathogenesis model in which over nutrition, particularly when involving diets high in fructose, leads to activation of ChREBP in the liver, which drives increased expression of genes encoding classical enzymes of DNL, including PKLR, ACL, ACC, and FAS. We further posit that upregulation of BDK and downregulation of PPM1K by ChREBP stimulates the DNL pathway by phosphorylation and activation of ACL, thus adding BDK and PPM1K to the panel of genes regulated by ChREBP to enhance fatty acid synthesis and development of dyslipidemia (Figure 5E). Simultaneously, the increased BDK:PPM1K ratio leads to increased phosphorylation and inhibition of BCKDH, contributing to the obesity-linked rise in circulating BCAA and BCKA. These findings suggest that BDK and PPM1K represent a previously unidentified class of ChREBP-β-regulated, lipogenesis-activating genes that perform their function via post-translational modulation of a key enzyme rather than by playing a direct catalytic role in the metabolic conversion of glucose to lipids.