Activation of Tripartite Motif Containing 63 Expression by Transcription Factor EB and Transcription Factor Binding to Immunoglobulin Heavy Chain Enhancer 3 Is Regulated by Protein Kinase D and Class IIa Histone Deacetylases

Rationale The ubiquitin–proteasome system (UPS) is responsible for skeletal muscle atrophy. We showed earlier that the transcription factor EB (TFEB) plays a role by increasing E3 ubiquitin ligase muscle really interesting new gene-finger 1(MuRF1)/tripartite motif-containing 63 (TRIM63) expression. MuRF 1 ubiquitinates structural proteins and mediates their UPS-dependent degradation. We now investigated how TFEB-mediated TRIM63 expression is regulated. Objective Because protein kinase D1 (PKD1), histone deacetylase 5 (HDAC5), and TFEB belong to respective families with close structural, regulatory, and functional properties, we hypothesized that these families comprise a network regulating TRIM63 expression. Methods and Results We found that TFEB and transcription factor for immunoglobulin heavy-chain enhancer 3 (TFE3) activate TRIM63 expression. The class IIa HDACs HDAC4, HDAC5, and HDAC7 inhibited this activity. Furthermore, we could map the HDAC5 and TFE3 physical interaction. PKD1, PKD2, and PKD3 reversed the inhibitory effect of all tested class IIa HDACs toward TFEB and TFE3. PKD1 mediated nuclear export of all HDACs and lifted TFEB and TFE3 repression. We also mapped the PKD2 and HDAC5 interaction. We found that the inhibitory effect of PKD1 and PKD2 toward HDAC4, HDAC5, and HDAC7 was mediated by their phosphorylation and 14-3-3 mediated nuclear export. Conclusion TFEB and TFE3 activate TRIM63 expression. Both transcription factors are controlled by HDAC4, HDAC5, HDAC7, and all PKD-family members. We propose that the multilevel PKD/HDAC/TFEB/TFE3 network tightly controls TRIM63 expression.

Rationale: The ubiquitin-proteasome system (UPS) is responsible for skeletal muscle atrophy. We showed earlier that the transcription factor EB (TFEB) plays a role by increasing E3 ubiquitin ligase muscle really interesting new gene-finger 1(MuRF1)/tripartite motif-containing 63 (TRIM63) expression. MuRF 1 ubiquitinates structural proteins and mediates their UPS-dependent degradation. We now investigated how TFEB-mediated TRIM63 expression is regulated.
Objective: Because protein kinase D1 (PKD1), histone deacetylase 5 (HDAC5), and TFEB belong to respective families with close structural, regulatory, and functional properties, we hypothesized that these families comprise a network regulating TRIM63 expression.
Methods and Results: We found that TFEB and transcription factor for immunoglobulin heavy-chain enhancer 3 (TFE3) activate TRIM63 expression. The class IIa HDACs HDAC4, HDAC5, and HDAC7 inhibited this activity. Furthermore, we could map the HDAC5 and TFE3 physical interaction. PKD1, PKD2, and PKD3 reversed the inhibitory effect of all tested class IIa HDACs toward TFEB and TFE3. PKD1 mediated nuclear export of all HDACs and lifted TFEB and TFE3 repression. We also mapped the PKD2 and HDAC5 interaction. We found that the inhibitory effect of PKD1 and

INTRODUCTION
Muscle mass is mainly regulated by an equilibrium of protein synthesis and degradation. Both muscle atrophy and hypertrophy occur physiologically and under disease conditions (Hershko and Ciechanover, 1998;Ciechanover, 2006;Schmidt et al., 2014;Wollersheim et al., 2014). The ubiquitin-proteasome system (UPS) is the principal protein-degrading system in muscle and largely responsible for the degradation of contractile proteins (Medina et al., 1995;Wing et al., 1995;Voisin et al., 1996;Ciechanover, 2006;Witt et al., 2008). Substrate protein ubiquitination is mediated by E3 ubiquitin-ligases that are specificity assuring and rate-limiting enzymes within the UPS (Hershko and Ciechanover, 1998;Ciechanover, 2006). Polyubiquitinated target proteins are then degraded by the 26S proteasome (Ciechanover, 2006). The UPS is activated during muscle atrophy leading to declining of structural and contractile proteins, most notably myosin heavy chain (MHC) (Fielitz et al., 2007), resulting in a reduction of muscle mass and function. The E3 ligase, muscle really interesting new gene-finger 1 (MuRF1), was identified as a major atrogene and is restricted to skeletal muscle and the heart (Bodine et al., 2001a). MuRF1 was shown to mediate ubiquitination and UPS-dependent degradation of structural proteins, such as alpha-actin, troponin I, troponin T, telethonin, titin, nebulin, the nebulin-related-anchoring protein, myosin light chain 2, myotilin, and T-cap (Kedar et al., 2004;Witt et al., 2005;Polge et al., 2011Polge et al., , 2018. Importantly, MuRF1 is also involved in muscular energy metabolism by degradation of non-structural proteins, such as muscle-type creatine kinase, glucocorticoid modulatory element binding protein-1, and 3-hydroxyisobutyrate dehydrogenase in striated muscles (McElhinny et al., 2002;Koyama et al., 2008). Numerous studies showed that the tripartite motif-containing 63 (TRIM63) gene expression, encoding MuRF1, is increased rapidly and strongly during various physiological and pathological atrophy conditions (Bodine et al., 2001a). The multitude of its target proteins, together with its strong transcriptional regulation, highlights the importance of MuRF1 in muscle homeostasis. To evaluate the transcriptional control of TRIM63, we performed a complementary DNA (cDNA)-expression screen to identify Abbreviations: HDAC, histone deacetylase-5; MuRF, Muscle really interesting new gene (RING)-finger containing protein; MiT/TFE, Microphthalmiaassociated transcription factor/transcription factor E; PKD, Protein Kinase D; TF, transcription factor; TFEB, transcription factor EB; TFE3, Transcription factor for immunoglobulin heavy-chain enhancer 3; TRIM63, Tripartite motif-containing 63 (mouse gene encoding MuRF1). regulators of TRIM63 expression and identified the basic helixloop-helix leucine zipper (bHLH-LZ) transcription factor EB (TFEB) as a novel TRIM63 regulator (Du Bois et al., 2015). We found that TFEB binds to specific, well-conserved, enhancer box (E-box) DNA motifs in the TRIM63 promoter that are close to its transcription start site. We reported that TFEB activity was inhibited via interaction with the class IIa histone deacetylase (HDAC) 5. We also showed that inhibition of TFEB by HDAC5 was attenuated by the stress-dependent serine/threonine kinase protein kinase D (PKD) 1, which interacted with HDAC5 and mediated its phosphorylation and 14-3-3 chaperone mediated nuclear export. This PKD1/HDAC5/TFEB axis controlled the expression of TRIM63 and was found to be important in angiotensin II-induced myocyte atrophy in vitro and muscle atrophy in vivo (Du Bois et al., 2015).
Recent reports, however, indicated that this pathway might not be as specific as we believed. For example, TFEB belongs to the microphthalmia/transcription factor E (MiT/TFE) family of bHLH-LZ transcription factors (TFs), which also includes TFE3, TFEC, and microphthalmia-associated transcription factor (MiTF). All MiT/TFEs recognize a unique E-box motif within the proximal promoters of lysosomal and autophagy genes (Sardiello et al., 2009;Palmieri et al., 2011;Martina et al., 2014) and regulate cellular catabolism and nutrient-dependent lysosomal response (Settembre et al., 2011;Slade and Pulinilkunnil, 2017). Importantly, TFEB and TFE3 were found to have partially redundant functions and to regulate overlapping sets of genes (Pastore et al., 2016). However, whether or not TFE3 or MiTF can activate TRIM63 expression is not known. In addition, the class IIa HDACs (HDAC4, HDAC5, HDAC7, and HDAC9) are highly expressed in the heart and skeletal muscle, and their interaction with the myocyte enhancer factor 2 (MEF2) TF decreases the expression of MEF2 target genes (Lu et al., 2000a,b;McKinsey et al., 2000;Zhang et al., 2002). If other class IIa HDACs inhibit the activity of TFEB or TFE3 is also unclear. Finally, PKD1 belongs to a family of calmodulin-calcium-dependent serinethreonine kinases (termed PKD1, PKD2, and PKD3). These kinases are important in cell growth, differentiation, migration, and apoptosis (Rozengurt et al., 2005). The close structural relationship of the PKD family members suggested that they might phosphorylate overlapping targets. Indeed, all three family members PKD1, PKD2, and PKD3 were shown to phosphorylate HDAC5 (Huynh and McKinsey, 2006). However, any effect on TFEB or TFE3-mediated TRIM63 expression is uncertain. Because PKD1, HDAC5, and TFEB belong to respective families with close structural, regulatory, and functional properties, we hypothesized that other family members could also be involved in the regulation of TRIM63 expression.

Statistics
All experiments were performed independently and at least three times using biological triplicates each and technical duplicates for each biological replicate. Data are shown as mean ± SEM. For all comparisons, one-way analyses of variance (ANOVAs) followed by a pairwise Student's t-test for independent groups were performed using GraphPad Prism 8.3.0 (GraphPad Software, Inc.) and Excel 2016 (Microsoft) software. P-values of less than 0.05 were considered statistically significant ( * P < 0.05; * * P < 0.01; * * * P < 0.001). Adobe Illustrator, version 16.0.0, was used for illustrations.

Class IIa Histone Deacetylases 4, 5, and 7 Inhibit Transcription Factor EB-Mediated Tripartite Motif-Containing 63 Expression
Previously, we showed that TFEB increases the expression of human TRIM63 by binding to specific E-box elements close to the transcription start site in its promoter (Du Bois et al., 2015). We showed that TFEB transcriptional activity was attenuated by HDAC5 that physically interacted and colocalized with TFEB in myocytes in vitro (Du Bois et al., 2015). Because HDAC4 and HDAC5 collectively inhibit the activity of the bHLH TF myogenin toward TRIM63 expression (Moresi et al., 2010), we hypothesized that other class IIa HDACs reduce the activity of TFEB on the TRIM63 promoter as well. Using luciferase assays, we found that HDAC4 ( Figure 1A), HDAC5 ( Figure 1B), and HDAC7 ( Figure 1C) inhibit TFEB-induced activity of the human TRIM63 luciferase reporter in a dose-dependent manner. These data indicate that TFEB-induced TRIM63 expression is inhibited by not only HDAC5 but also HDAC4 and HDAC7.

Protein Kinase D Family Members Attenuate Histone Deacetylase-Mediated Inhibition of Transcription Factor EB-Induced Tripartite Motif-Containing 63 Expression
Earlier, we reported that PKD1 associates with phosphorylates and facilitates 14-3-3-mediated nuclear export of HDAC5, which relives inhibition of TFEB and thereby increases TRIM63 Frontiers in Physiology | www.frontiersin.org expression (Du Bois et al., 2015). Because the three PKD-family members PKD1, PKD2, and PKD3 share a high degree of similarity in their functional domains (Avkiran et al., 2008) and were shown to phosphorylate HDAC5 (Huynh and McKinsey, 2006), we hypothesized that the PKD-family regulates TFEB activity by inactivating HDAC4, HDAC5, and HDAC7 as well.
To investigate the effect on HDAC-mediated inhibition of TFEB, we performed luciferase assays. As already shown, TFEB-induced TRIM63 activity was inhibited by HDAC4 (Figure 2A), HDAC5 ( Figure 2B), and HDAC7 ( Figure 2C). In contrast, the class I HDACs 1 and 3 did not decrease TFEB-induced TRIM63 activity (Supplementary Figure 1A). When we cotransfected PKD1 (Figures 2A-C, left panels), PKD2 (Figures 2A-C, middle panels), or PKD3 (Figures 2A-C, right panels), the repressive effects of all three HDACs were relieved. CamK IV, which was used as a positive control, also relieved the repressive effects of HDAC4, HDAC5, and HDAC7 in the same assay (Supplementary Figure 2A). Our data indicate that the PKD family converges on HDAC4, HDAC5, and HDAC7 to control TRIM63 expression. To investigate if the observed effects of the PKD family onto HDAC-mediated inhibition of TFEB are due to nuclear export of HDAC4 and HDAC7, we performed immunocytochemistry experiments. As expected, we observed colocalization of HDAC5 ( Figure 2D, left panel) and also HDAC4 ( Figure 2D, middle panel) and HDAC7 ( Figure 2D, right panel) with TFEB in the nucleus. This colocalization is a possible molecular basis for class IIa HDAC-mediated TFEB inhibition. When we cotransfected with PKD1, HDAC4, HDAC5, and HDAC7 translocated to the cytoplasm, whereas TFEB remained in the nucleus ( Figure 2D). Phosphoserines 246, 467, and 632 in HDAC4 , 259 and 498 in HDAC5 (Zhang et al., 2002), and 155, 181, 321, and 449 in HDAC7 (Dequiedt et al., 2005) serve as binding sites for the chaperone protein 14-3-3 and are known PKD1 phosphorylation sites. Previously, we confirmed that PKD1 binds to HDAC5 and phosphorylates its 14-3-3 consensus sites. To investigate if PKD1 has a similar effect on HDAC4 and HDAC7, we performed a mammalian-twohybrid assay using GAL4 upstream activator sequence (UAS)luciferase, as published recently (Chang et al., 2005;Du Bois et al., 2015). This assay was also used to investigate if PKD2 phosphorylates HDAC4, HDAC5, and HDAC7 as well. In this assay, the N-terminus of HDAC4, HDAC5, or HDAC7 is fused to the GAL4 DNA-binding domain, and 14-3-3 is fused to the VP16 transactivation domain. Under normal growth conditions, these HDACs are not phosphorylated in COS-7 cells. Thus, GAL4-HDAC4, GAL4-HDAC5, or GAL4-HDAC7 cannot interact with 14-3-3-VP16, and the GAL4-dependent UAS-luciferase cannot be activated ( Figure 3A). Expression plasmids encoding these fusion proteins, together with UAS-luciferase, were transfected into COS-7 cells together with increasing amounts of PKD1 or PKD2 expression plasmids. PKD1 and PKD2 increased UASluciferase activity in a dose-dependent manner. These data indicate that PKD1 and PKD2 create phospho-14-3-3 recognition motifs in HDAC4, HDAC5, and HDAC7, which recruit 14-3-3 proteins that mediate their nuclear export.
We had identified amino acids 360-601 as PKD1 binding region in HDAC5 (Du Bois et al., 2015). To test if PKD2 and HDAC5 physically interact as well, we performed coimmunoprecipitation experiments with PKD2 and wild-type and deletion mutants of HDAC5. We found that PKD2 interacted avidly with HDAC5 and that amino acids 360-601 in HDAC5 were responsible for this interaction (Figures 3B,C).
Transcription Factor Binding to Immunoglobulin Heavy-Chain Enhancer 3 but Not Microphthalmia-Associated Transcription Factor Increase Tripartite Motif-Containing 63 Expression Because TFEB and the closely related MiTF-family members TFE3 and MiTF elicit some degree of functional redundancy and bind to the same E-box elements (Steingrimsson et al., 2004;Martina et al., 2014;Pastore et al., 2016), we hypothesized that TFE3 and MiTF could regulate TRIM63 expression as well. Indeed, TFEB ( Figure 4A) and TFE3 ( Figure 4B) but not MiTF ( Figure 4C) increased TRIM63 activity in a dose-dependent manner. Thus, for further experiments, we focused on TFE3 and excluded MiTF. Using the wild-type and E-box mutated TRIM63 reporter constructs ( Figure 4D; Du Bois et al., 2015), we found that the TFE3-induced TRIM63 expression was strongly reduced in E-box 1 and E-box 3 mutants, whereas mutation of E-box 2 had a weaker effect ( Figure 4E). These data indicate that E-box 1 and E-box 3 mediate the TFEB-and TFE3-induced TRIM63 activity.

Transcription Factor EB and Transcription Factor Binding to Immunoglobulin Heavy-Chain Enhancer 3 Strongly Interact With Each Other
Because TFEB and TFE3 used the same E-box motives, we investigated if TFEB and TFE3 have additive functions on the TRIM63 promoter. We found that equimolar concentrations of TFEB and TFE3 activated the TRIM63 promoter to comparable degrees and that their combination did not result in a Luciferase activity was normalized to expression of CMV-LacZ and expressed as fold increase. Data are represented as mean ± SEM. ***P < 0.005 vs. control transfection with pcDNA3.1. (D) Schematic diagram of the human TRIM63-promoter. Positions of conserved E-box motifs (CANNTG) relative to the transcription start site (TSS) are indicated. Homo sapiens (mut.) indicates mutated nucleotides to inactivate individual E-boxes (mutated nucleotides are shown in bold). One-way ANOVA P < 0.0001 (A,B) and n.s. (C), respectively; *P < 0.05; **P < 0.01; ***P < 0.005. (E) COS-7 cells were transfected with a TFE3 expression plasmid and the indicated TRIM63-promoter constructs (-543 bp) harboring E-box mutations, as shown in (D). Data are represented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.005. (F) Luciferase assays (Hs_TRIM63-Luc; -543 bp) performed on cell extracts of COS-7 cells after a single transfection with either MYC-TFEB (TFEB) or FLAG-TFE3 (TFE3) or cotransfection with MYC-TFEB (TFEB) and FLAG-TFE3 (TFE3). Luciferase activity was normalized to expression of CMV-LacZ and expressed as fold increase. Data are represented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.005. (G), Co-immunoprecipitation (Co-IP) assay with lysates from COS-7 cells expressing FLAG-TFE3 and MYC-TFEB. TFE3 fusion proteins were immunoprecipitated (IP) with anti-FLAG M2 agarose, and TFEB fusion proteins were detected with an antibody directed against MYC. Input proteins were detected by Western blot (immunoblot, IB) with antibodies directed against the FLAG-or MYC-tag. n = 3. During cell lysis and IP increasing concentrations of sodium chloride (50, 100, 300, 500, 700, and 900 mM) were used as indicated.
further increase in TRIM63 activity ( Figure 4F). These data indicate that TFEB and TFE3 are possibly competing for the same E-box elements. As TFEB and TFE3 are known to heterodimerize, we tested the strength of this interaction. We performed cell lysis and co-immunoprecipitation with increasing sodium chloride buffer concentrations ranging from 50-to 900-mM sodium chloride. Only the highest sodium chloride concentration used attenuated the interaction between TFEB and TFE3, indicative of a strong interaction of both proteins (Figure 4G).

Class IIa Histone Deacetylases 4, 5, and 7 Inhibit Transcription Factor Binding to Immunoglobulin Heavy-Chain Enhancer 3-Mediated Tripartite Motif-Containing 63 Expression
We next hypothesized that the activity of TFE3 on the TRIM63 promoter is also inhibited by class IIa HDACs. Using luciferase assays, we found that HDAC4 (Figure 5A), HDAC5 (Figure 5B), and HDAC7 ( Figure 5C)  activity in a dose-dependent manner, indicating that not only TFEB but also TFE3 is collectively controlled by class IIa HDACs. In contrast, TFE3-induced TRIM63 activity was not reduced by HDAC1 or HDAC3 (Supplementary Figure 1B). We performed co-immunoprecipitation experiments and found that TFE3 and HDAC5 interacted with each other (Figure 5D). We next performed co-immunoprecipitation experiments with TFE3 and wild-type and deletion mutants of HDAC5 to determine which region within HDAC5 was required for its interaction with TFE3. We found that amino acids 51-175 of HDAC5 are required for physical interaction with TFE3 and, therefore, define a TFE3 binding site (Figures 5E,F). We performed immunocytochemistry using antibodies directed against endogenous TFE3 and endogenous HDAC4, HDAC5, and HDAC7 and found that these proteins colocalized in C2C12 myocytes indicating that TFE3 interacts with class IIa HDACs in myocytes (Supplementary Figure 4).

Protein Kinase D Family Members Attenuate Histone Deacetylase-Mediated Inhibition of Transcription Factor Binding to Immunoglobulin Heavy-Chain Enhancer 3-Induced Tripartite Motif-Containing 63 Activity
Based on our findings, we hypothesized that the PKD-family members could abolish the inhibitory effects of HDAC4, HDAC5, and HDAC7 toward TFE3. To test this hypothesis, we performed luciferase assays and found that the TFE3induced TRIM63 activity was inhibited by HDAC4 ( Figure 6A), HDAC5 (Figure 6B), and HDAC7 ( Figure 6C). Importantly, cotransfection of PKD1 (Figure 6, left panel), PKD2 (Figure 6, middle panel), or PKD3 (Figure 6, right panel) relieved the repressive effects of all three HDACs. CamK IV also relieved the repressive effects of HDAC4, HDAC5, and HDAC7 in the same assay (Supplementary Figure 2B). To investigate if the observed effects of the PKD family onto HDAC-mediated inhibition of TFE3 are due to nuclear export of HDAC4, HDAC5, or HDAC7, we performed immunocytochemistry experiments. We observed colocalization of all class IIa HDACs with TFE3 in the nucleus (Figure 6D). When we cotransfected with PKD1, HDAC5 ( Figure 6D, left panel), HDAC4 ( Figure 6D, middle panel), and HDAC7 ( Figure 6D, right panel) translocated to the cytoplasm, whereas TFE3 remained in the nucleus. These data show that PKD1, PKD2, and PKD3 attenuated the inhibitory effect of HDAC4, HDAC5, and HDAC7 onto TFE3-induced TRIM63 expression by nuclear export of these HDACs.

DISCUSSION
We report that TRIM63 expression is tightly regulated by a transcriptional network comprised of the PKD-family members PKD1, PKD2, and PKD3, the class IIa HDACs HDAC4, HDAC5, and HDAC7, as well as the MiT/TFE family members TFEB and TFE3. We show that HDAC4, HDAC5, and HDAC7 colocalize with TFEB and TFE3 in the nucleus of non-muscle cells and inhibit TFEB-and TFE3-induced TRIM63 expression. We demonstrate that PKD1, PKD2, and PKD3 phosphorylate HDAC4, HDAC5, and HDAC7. This state-of-affairs facilitates their binding to the 14-3-3 chaperon protein and their consecutive nuclearexport relieving repression of TFEB-and TFE3-induced TRIM63 expression. All PKD family members ameliorated HDAC-mediated inhibition of TFEB-and TFE3-mediated TRIM63 promoter activity. We conclude that TRIM63 expression, which is a key factor in UPS-dependent protein degradation in skeletal muscle during atrophy, is regulated at multiple levels. Skeletal muscle atrophy is characterized by a reduction in myofiber size, mainly due to a net loss of structural proteins leading to a reduced muscle mass and function. As it occurs in many pathological conditions, such as critical illness and inflammation, this results from increased UPSdependent protein degradation (Hershko and Ciechanover, 1998;Ciechanover, 2006). However, muscle atrophy is a complex process that is mediated by multiple factors. Indeed, the molecules, mediators, and cellular pathways contributing to muscle atrophy are still being discovered (Bodine and Baehr, 2014). In 2001, Bodine et al. (2001a) identified TRIM63/MuRF1 as an atrogene that was highly upregulated in atrophying muscle after denervation, immobilization, and unweighting-induced atrophy in rats. MuRF1/Trim63 knockout mice showed less denervation-induced muscle atrophy compared with wild-type mice (Bodine et al., 2001a). Thereafter, we and others confirmed the involvement of MuRF1 in muscle atrophy (Du Bois et al., 2015), described its increased expression in physiological and pathological muscle atrophy Wollersheim et al., 2014;Huang et al., 2017), and discovered MuRF1 targets (Fielitz et al., 2007;Witt et al., 2008;Labeit et al., 2010;Nowak et al., 2019).
Increased TRIM63 expression is a well-established and accepted read-out for skeletal muscle atrophy in general. The expression of MuRF1/TRIM63 is regulated by several signaling pathways converging onto multiple TFs, such as the forkhead box protein family (Stitt et al., 2004;Raffaello et al., 2010) and myogenin (Moresi et al., 2010). Glass and co-workers demonstrated that the insulin-like growth factor/phosphoinositide 3-kinase/protein kinase B/Akt pathway, which had previously been shown to induce hypertrophy (Bodine et al., 2001b), suppresses atrophy by downregulating MuRF1. Myogenin was also shown to regulate TRIM63, and mice that lack myogenin failed to upregulate TRIM63 expression. These mice were resistant to denervation-induced muscle atrophy (Moresi et al., 2010). Given the importance of transcriptional regulation of TRIM63, we previously performed a cDNA expression screen to identify regulators of TRIM63 expression. We identified TFEB as an important TF and described its regulation and importance in muscle atrophy in myocytes in vitro and in muscle in vivo (Du Bois et al., 2015).
TFEB belongs to the MiT/TFE family of bHLH-LZ TFs, including TFE3, MiTF, and TFEC (Steingrimsson et al., 2004). All MiT/TFE TFs recognize a unique E-box motif within the proximal promoters of lysosomal and autophagy genes (Sardiello et al., 2009;Palmieri et al., 2011;Martina et al., 2014) and regulate cellular catabolism and nutrient-dependent lysosomal response (Settembre et al., 2011;Slade and Pulinilkunnil, 2017). Within the MiT/TFE family, a close relationship exists between TFEB and TFE3. TFEB and TFE3 regulate cellular adaptation to stress by induction of lysosomal biogenesis and autophagy and immune responses, metabolism, mitochondrial homeostasis, and unfolded-protein responses. TFEB and TFE3 were shown to regulate the same genes (Pastore et al., 2016), such as CDH1, encoding E-cadherin (Huan et al., 2005), CD40L, encoding CD40 ligand (Huan et al., 2006), and PPARGC1a, encoding PGC1α (Settembre et al., 2013). However, whether or not TFE3 activates TRIM63 expression was not known. We report that not only TFEB but also TFE3 regulates TRIM63 expression and that both TFs are controlled by the PKD and HDAC enzyme families. To the well-described TFEB and TFE3 functions, we add that both TFs also regulate TRIM63 expression. We propose that they activate UPS-and autophagy-mediated protein degradation. We showed earlier that TFEB regulates TRIM63 expression via binding to E-box elements in the TRIM63 promoter, and mutation of these binding motifs abrogated TFEBdependent TRIM63 expression. However, although MiT/TFE family members bind specific E-box elements and use the same E-box motifs as TFEB to induce TRIM63 expression, we have not yet demonstrated direct binding of TFE3 to the native TRIM63 promoter. Such analyses need to be performed.
MuRF1 and the regulation of its expression are important for metabolic adaptations of striated muscles (Hirner et al., 2008;Witt et al., 2008;Rudolf et al., 2013). MuRF1 mediates the degradation of numerous metabolic enzymes, such as pyruvate dehydrogenase and its regulator pyruvate dehydrogenase kinase that are key mitochondrial enzymes regulating glycolysis. Interaction studies also demonstrated that MuRF1 interacts with pyruvate kinase 2, phosphorylase beta, and glycogenin, which are involved in glycolysis and glycogen metabolism, respectively (Hirner et al., 2008). Because TFEB (Mansueto et al., 2017), TFE3 (Iwasaki et al., 2012), and forkhead box (Schiaffino et al., 2013) are also involved in muscle metabolism and mitochondrial homeostasis, these effects are possibly mediated by their ability to regulate TRIM63 expression. As our conclusions are mainly based on data obtained from non-muscle cells, this hypothesis warrants further investigation to show its relevance in muscle metabolism.
Distinguished by their structures and expression patterns, HDACs can be divided into different classes. We focused on the heart and skeletal muscle enriched class IIa HDACs 4, 5, and 7. All class IIa HDACs repress the activity of myocyte enhancer factor 2 (MEF2), which reduces the expression of MEF2 target genes and suppresses the formation of slow-twitch, oxidative fibers in the muscle (Lu et al., 2000a,b;McKinsey et al., 2000;Zhang et al., 2002;Potthoff et al., 2007). HDAC4 and HDAC5 coordinately inhibited the activity of myogenin that is important for myogenesis (Moresi et al., 2010). Latter data implicated that HDAC4, HDAC5, and HDAC7 predominantly inhibit bHLH TFs contained in muscle to regulate muscular stress response. However, if other class IIa HDACs inhibit TFEB and if they can also reduce TFE3-mediated TRIM63 expression was unknown. We describe that HDAC4, HDAC5, and HDAC7 inhibit TFEB-mediated TRIM63 activity and show that all three HDACs inhibited TFE3. We propose that class IIa HDACs collectively control TRIM63 expression. However, as our experiments were mainly performed in non-myocytes, further studies are needed to investigate if this mode of action also occurs in myocytes.
Protein function is often regulated by reversible protein phosphorylation by protein kinases, such as the PKD family (Rykx et al., 2003;Rozengurt, 2011). This family of stressresponsive serine-threonine kinases of the calmodulin-calciumdependent-kinase family (termed PKD1, PKD2, and PKD3) play important roles in cell growth, differentiation, migration, and apoptosis (Rozengurt et al., 2005). Importantly, we and others showed that PKD1 mediates muscle atrophy (Du Bois et al., 2015), facilitates slow-twitch-fiber transformation in the muscle , and mediates cardiac stress response . However, the close structural relationship of the PKD family members especially in their functional domains (Avkiran et al., 2008) suggested that they might phosphorylate overlapping targets and might be involved in the observed phenotypes as well. Indeed, PKD1, PKD2, and PKD3 were all shown to phosphorylate HDAC5 (Huynh and McKinsey, 2006). This situation, in turn, relieves the repression of TFs such as MEF2 and TFEB Du Bois et al., 2015). The mechanism is involved in the pathophysiology of cardiomyocyte hypertrophy in vitro and cardiac hypertrophy and remodeling in vivo (Vega et al., 2004;Fielitz et al., 2008). These observations suggested that PKD family members redundantly control the activity of class IIa HDACs (Huynh and McKinsey, 2006). However, it was uncertain if this interaction affects TFEB-or TFE3-mediated TRIM63 expression. We confirmed this hypothesis for all three PKD family members and show that they phosphorylate HDAC4, HDAC5, and HDAC7 and promote their nuclear export. Because a high degree of functional redundancy was reported for PKD1 and PKD2, we investigated if PKD2 also interacts with HDAC5. We confirmed this interaction and mapped the binding of PKD2 to HDAC5 to the same region in HDAC5 that binds to PKD1, suggesting that TRIM63 expression is redundantly controlled. We also show that PKD3, which is normally not expressed in unstressed myocytes (Li et al., 2011), elicits comparable effects toward the control of TRIM63 expression by class IIa HDACs. This state-of-affair might especially be important during stress situations where PKD3 was shown to be strongly increased (Li et al., 2011). Our data are in line with previous reports showing that PKD3 can substitute for PKD1 as an HDAC5 kinase in non-muscle cells (Matthews et al., 2006). In summary, our data indicate that a diverse and multilevel pathway regulates muscle atrophy. Subcellular localization of class IIa HDACs is strongly controlled by several protein kinases, which are not restricted to the protein kinase D family but also involve CamK I, CamK IV (also shown here), MARK2, and others (Chang et al., 2005). Also, some kinases selectively target specific class IIa HDACs, i.e., CamK II specifically targets HDAC4 . These data, together with our findings and the known regulation of TFEB and TFE3 activity by posttranslational modification (Puertollano et al., 2018), indicate that TFEB-and TFE3-mediated transcriptional activity has a much higher level of complexity that could not be addressed here and warrants further studies.
Previous studies showed that TFEB (Settembre et al., 2012) and TFE3 (Martina et al., 2014) are predominantly localized in the cytoplasm of HEK-293T and ARPE-19 cells, respectively, and that their phosphorylation status and cytosolic-to-nuclear shuttling regulates TFEB and TFE3 activity. The kinases mammalian target of rapamycin complex 1 (mTORC1) and ERK are important in that regard (Martina et al., 2012(Martina et al., , 2014Settembre et al., 2012). In the presence of nutrients, mTORC1 phosphorylates TFEB and TFE3, thereby facilitating their binding to 14-3-3 chaperone proteins and mediating their retention in the cytoplasm. Conversely, reduced mTORC1 activity increases TFEB (Settembre et al., 2012) and TFE3 (Martina et al., 2014) shuttling into the nucleus. In contrast, in line with our previous report (Du Bois et al., 2015), we observed that TFEB and TFE3 were mainly localized to the nucleus of COS-7 cells when overexpressed. If these observations are attributable to differences in the cell type, culturing conditions, differentiation status, or transfection warrants further investigation. However, our observation that TFEB and TFE3 are localized to the nucleus is supported by other findings showing that 20-30% of TFEB is contained in the nucleus of several cell lines, such as patient-derived fibroblasts (Song et al., 2013), HeLa cells (Settembre et al., 2011), ARPE-19 cells (Martina et al., 2012), and mouse embryonic fibroblasts (Sardiello et al., 2009). Because we hypothesized that HDAC4, HDAC5, and HDAC7 regulate the activity of TFEB and TFE3 at the TRIM63 promoter, as reported for their inhibitory effects on MEF2 target genes (Lu et al., 2000a,b;McKinsey et al., 2000;Zhang et al., 2002), we used overexpression as a model system despite differences in subcellular localization of the participating proteins. Our data indicate that the activity of TFEB and TFE3 is regulated at least at two different levels, first by regulation of their subcellular localization and second by repression of their activity by class IIa HDACs.

LIMITATIONS
Most of the data shown in our study are based on overexpression experiments. Our results need further evaluation, especially in myocytes, myotubes, and skeletal muscle, and by working with the endogenous components of the signaling pathway described, such as co-staining of endogenous proteins in immunocytochemistry. Nevertheless, our results are in line with previously published work that focused on the regulation of the activity and subcellular localization of class IIa HDACs by PKD both in vitro (Zhang et al., 2002;Dequiedt et al., 2005;Backs et al., 2006) and in vivo Kim et al., 2008;Du Bois et al., 2015). Although we have shown that upon coexpression with PKD, HDAC4, HDAC5, and HDAC7 are localized in the cytoplasm, and PKD increased UAS-luciferase activity that depends on 14-3-3 binding, we have not proven that this export was mediated by CRM1. Likewise, to illustrate our findings, we display representative pictures of single nuclei and have not performed biochemical fractionation experiments (nuclear vs. cytosolic extracts) to support the nuclear export of the HDACs upon PKD1 overexpression for a larger subset of cells. Because such experiments have been performed previously, we would like to refer to this work Vega et al., 2004). We have used co-immunoprecipitation experiments to show the interaction of PKD1 and PKD2 with HDAC5, mapped the interacting domains, and showed functional consequences of these interactions. We have also shown that the activity of TFEB and TFE3 on the TRIM63 reporter is inhibited by class IIa HDACs. However, further studies are needed to prove that these proteins directly interact which each other by using cellfree assays, proximity ligation assays, coimmunostaining, or other techniques.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
CP and BF designed and performed experiments, prepared figures, and prepared the manuscript. YL performed experiments, prepared figures, and provided intellectual input. JR designed experiments and discussed data. FL discussed data, provided intellectual input, and rewrote the manuscript. JF supervised the project, analyzed data, and rewrote the manuscript. All authors contributed to the article and approved the submitted version.

ACKNOWLEDGMENTS
We thank Sibylle Schmidt and Lisa Lörzer for excellent technical assistance and Melanie Kny, Dörte Lodka, Philipp Du Bois, Niklas Dörmann, Sebastian Wundersitz, and Priyanka Giri for helpful discussions. We thank Stefan Groß for his insights and discussions regarding bioinformatic analyses.