Acox2 is a regulator of lysine crotonylation that mediates hepatic metabolic homeostasis in mice

Acyl-CoA oxidase 2 (Acox2) is an enzyme involved in peroxisomal bile acid synthesis and branched-chain fatty acid degradation. Acox2 knockout (−/−) mice spontaneously developed liver cancer with marked lymphocytic infiltrate. Tandem-affinity purification coupled with mass spectrometry analysis revealed that Acox2 interacted with methylcrotonoyl-CoA carboxylase followed by co-immunoprecipitation confirmation. Here we reported that non-histone lysine crotonylation (Kcr) levels were downregulated in Acox2−/− mice livers. Interestingly, Kcr signals were concentrated in the nucleus of tumor cells but mostly located in the cytoplasm of adjacent normal liver cells of Acox2−/− mice. Quantitative analysis of the global crotonylome further revealed that 54% (27/50) of downregulated non-histone Kcr sites were located in mitochondrial (11/50) and peroxisomal (17/50) enzymes including Ehhadh, Scp2, Hsd17b4, Crot, Etfa, Cpt1a, Eci1/2, Hadha, Etfdh, and Idh2. Subsequent site-directed mutagenesis and transcriptome analysis revealed that Ehhadh K572cr might have site-specific regulatory roles by downregulating TOP3B expression that lead to increased DNA damage in vitro. Our findings suggested Acox2 is a regulator of Kcr that might play critical role on hepatic metabolic homeostasis.


INTRODUCTION
Beta-oxidation chain shortening of very-long-chain fatty acids, 2-methyl-branched fatty acids, and the CoA esters of C27 bile acid intermediates di-and trihydroxycoprostanoic acid occur in the peroxisomes [1,2]. There are four enzymatic steps in each cycle of peroxisomal β-oxidation, including oxidation, hydration, dehydrogenation, and thiolytic cleavage [3]. Acyl-CoA oxidases (ACOXs) catalyze the first step. Three ACOX1/2/3 with different substrate specificities are found in humans [4]. ACOX1 is responsible for the oxidation of very long straight-chain fatty acids [5], ACOX2 is the only acyl-CoA oxidase involved in bile acid biosynthesis, and both ACOX2 and ACOX3 are involved in the degradation of branched-chain fatty acids [4,6]. ACOX2 is predominantly expressed in the liver and kidney, whereas ACOX3 has relatively lower expression levels in the liver. In the peroxisomes, D-bifunctional enzyme (HSD17B4) or L-bifunctional enzyme (EHHADH) catalyze the second and third steps of β-oxidation, respectively [7]. The last step in betaoxidation can be performed by multiple thiolases, including acetyl-CoA acyltransferase 1 (ACAA1), and the sterol carrier protein x (SCPx or SCP2) [8]. Then shortened fatty acids are transported to the cytosol for mitochondrial uptake through the action of carnitine octanoyltransferase (CROT) [9].
Lysine crotonylation (Kcr) has emerged as a key posttranslational modification that plays regulatory roles in many biological processes [10][11][12]. Interestingly, a recent study reported that both ACOX1 and ACOX3 were key crotonyl-CoA-producing enzymes, and deletion of ACOX3 in hESCs led to reduced Kcr, which impaired mesoderm or endoderm differentiation [13]. Short-chain enoyl-CoA hydratase (ECHS1), an enzyme involved in mitochondrial fatty acid beta-oxidation, controls intracellular crotonyl-CoA and maintains the maturity and homeostasis of cardiomyocytes through histone crotonylation [14]. These findings suggested that mitochondrial and peroxisomal fatty acid oxidation was important for crotonyl-CoA production. On the other hand, the classic histone acetyltransferases p300, pCAF, and MOF also have histone crotonyltransferase activity [15,16], whereas the classic histone deacetylases HDAC1/2/3 and SIRT1/2/3 can remove Kcr under different conditions [15,17]. Many non-histone proteins are also modified by crotonylation [18]. However, the functional elucidation of non-histone Kcr is in its infancy.
We previously reported that Acox2 deficiency is associated with a metabolic switch in primary malignant cardiac tumor progression [19,20]. ACOX2 homozygous stop-gain (p.Y69*) and missense (p.R225W) mutations have been identified in patients with transaminase elevation and liver fibrosis [21,22]. We generated an Acox2 knockout mice model that exhibited the typical clinical symptoms of fatty liver disease at the age of 4 months [23]. Notably, nicotinamide N-methyltransferase (NNMT), a critical regulator of metabolism and epigenetics [24,25], and its product 1-methylnicotinamide were uniformly downregulated in both serum and liver tissues of Acox2 −/− mice [23]. Since Acox2 and Acox3 have similar function in peroxisomal β-oxidation; therefore, hepatic Kcr was presumed to be downregulated in Acox2 −/− mice. In this study, we perform integrative analysis and find that Acox2 is a master regulator for protein post-translational modifications that involved in the metabolic homeostasis.

Acox2 loss induces liver cancer in mice
We generated an Acox2 knockout C57BL/6n mouse model using the CRISPR/Cas9 gene editing system. A deletion of 15,146 bp with a span from exon 3 to exon 13 was identified ( Supplementary Fig.  S1A) in Acox2 knockout (-/-) mice. Importantly, Hematoxylin and eosin (H&E) staining revealed that 28.57% (6/21) Acox2 −/− mice spontaneously developed liver cancer without high-fat diet induction or small molecule treatment at the postnatal age of 8-9 months (Fig. 1A-C and S1B), whereas wild type (WT) mice did not develop such tumors. Inflammatory infiltration is very popular in the liver tissues of Acox2 −/− mice. We performed immunohistochemical stains for immune cell markers (CD3, CD20, CD138, and CD68) to determine the immune cells types and immunocyte infiltration characteristics. The predominant cell-type is B-lymphocytes CD20, followed by T-lymphocytes CD3 and plasma cells CD138, but is almost negative for macrophages CD68 (Fig.  1D). There was a co-appearance of CD3 and CD20. Compared with WT mice, we also observed positive staining for Alpha-fetoprotein (AFP) in Acox2 −/− mice (Fig. 1E). Heterogeneity was identified in tumor lesion of Acox2 −/− mice liver (Fig. S1B). In addition, the ACOX2 expression level in humans is significantly downregulated in HCC than in normal liver specimens (fold-change = −3.19; pvalue = 1.91 × 10 −7 ) (Fig. 1F) and decreases gradually in malignant stages when compared with the early stages (p-value = 0.00096) (Fig. 1G). These findings are further supported by low ACOX2 expression in human HCC is correlated with a decrease in OS (overall survival, p-value = 0.036) (Fig. 1H) and DFS (disease free survival, p-value = 0.048) (Fig. 1I) when comparing to high ACOX2 expression as observed on analyzing data from TCGA portal.
Acox2 Interacts with methylcrotonoyl-CoA carboxylase We next used tandem-affinity purification coupled with mass spectrometry (TAP-MS) analysis to determine the interactome of ACOX2 and its functions in MHCC97H and HEK-293T cell lines, respectively (Supplementary Tables S1 and S2). Since ACOX2 works as a homodimer that is essential for biological function, ACOX2 was also identified as an interacting partner in both cell lines, which further validate the data reliability of TAP-MS analysis. Approximately 40% (12/30) of the top 30 interacting proteins for Acox2 overlapped in MHCC97H and HEK-293T cells ( Fig. 2A). In particular, four classic biotindependent carboxylases [26], including acetyl-CoA carboxylase 1 (ACACA) and 2 (ACACB), and methylcrotonoyl-CoA carboxylase subunit alpha (MCCA) and beta chain (MCCB), propionyl-CoA carboxylase subunit alpha (PCCA) and subunit beta (PCCB), and pyruvate carboxylase (PC), have protein-protein interactions with ACOX2. The interaction between ACOX2 and MCCC2 (MCCB) was further confirmed by co-immunoprecipitation in HEK-293T cells (Fig. 2B). MCCC1 and MCCC2 could form a heterodimer and catalyze the carboxylation of 3methylcrotonyl-CoA to 3-methylglutaconyl-CoA [27], and intracellular crotonyl-CoA could directly regulate Kcr [12]. In addition, ACOX2 has protein interaction with DNA topoisomerases including TOP2A and TOP2B in MHCC97H cells (Table S1).
Kcr changes are associated with liver cancer progression in Acox2 −/− mice To investigate the potential correlation between Kcr levels and HCC progression, we examined Kcr changes in tumor and adjacent normal cells of Acox2 −/− mice. A previously well-characterized rabbit pan-anti-Kcr antibody (PTM-501) was utilized to perform immunofluorescence staining [13,28]. To further clarify the specificity of this Kcr antibody, we firstly evaluated the distribution and levels of Kcr in two cultured cell lines including iPSC (induced pluripotent stem cells) and HepG2. We observed strong positive signals in both nucleus and cytoplasm of iPSCs (Fig. S2A), which is consistent with the findings from the previous study [13]. Meanwhile, we found that the Kcr signals are mostly concentrated in the nucleus of HepG2 cells (Fig. S2A). Since KATs and HDACs can modulate histone Kcr levels [29,30], HepG2 cells were then treated with p300 inhibitors and HDACIs, respectively. Compared to cells without treatment (Ctrl), we identified decreased Kcr levels in cells treated with C646 (2 μm) whereas Kcr signals were upregulated in cells treated with FK228 (0.25 μM) by immunoblotting (Fig. S2B) and immunofluorescence assay (Fig. S2C). These findings suggested that pan-anti-Kcr antibody provided adequate sensitivity for detecting Kcr changes in Acox2 −/− mice. In this study, we found that Kcr levels were more concentrated in the nucleus of tumor cells (Figs. 3A, B and S3) whereas adjacent normal liver cells that harbor most of Kcr signals in the cytoplasm through confocal immunofluorescence microscopy. By contrast, levels of Ehhadh were markedly downregulated in the tumor lesions of Acox2 −/− mice (Fig. 3A), which further validates the specificity of the fluorescence signal for the anti-Kcr antibody. In addition, Kcr fluorescence signal intensity in cytoplasm of normal liver cells in Acox2 −/− mice was significantly decreased when compared with WT mice (Fig. S4A). Since Acox2 −/− mice displayed bile duct proliferation (Fig. S4B), we also evaluate the possible Kcr changes in areas of ductular reaction. Staining for Sox9 and CK7 was used to outline bile ducts and ductular reaction as previous studies [31][32][33]. We found that Kcr levels were increased in areas of ductular reaction when compared with adjacent liver tissues (Figs. 3C and S4C). Co-localizations of Kcr signal and ductal markers (CK7 and Sox9) were observed by double immunostaining (Figs. 3C and S4D). These findings suggest that Kcr might play important roles in liver cancer progression.

Downregulation of non-histone Kcr in Acox2 mice
Since Acox2 and Acox3 have overlapped function on peroxisomal β-oxidation, we postulated that Kcr levels in Acox2 −/− mice might be changed. To address this concern, an overview of four different classic protein post-translational modifications, including ubiquitination (Fig. 4A), acetylation (Fig. 4B), succinylation (Fig. 4C), and crotonylation ( Fig. 4D), was obtained through western blot analysis using the corresponding pan antibody. We did not observe significant differences in acetylation, ubiquitination, and succinylation between liver tissues of Acox2 −/− mice and their WT littermates. However, non-histone lysine crotonylation (Kcr) was partially downregulated in the liver tissue of Acox2 −/− mice when compared with wild-type mice (Fig. 4D). By contrast, a mild increase in histone Kcr was observed in Acox2 −/− mice. In addition, decreased non-histone Kcr was also detected in the kidney tissues of Acox2 −/− mice (Fig. 4F) whereas acetylation ( Fig.  4E) was not significantly changed.
Histone (H2b) K86cr was upregulated in Acox2 −/− mice To precisely determine hepatic Kcr changes caused by Acox2 knockout, global crotonylome analysis was performed using the liver tissues of Acox2 −/− (n = 3) and WT littermates (n = 3) that were collected at four months of age (Fig. 5A). Principal component analysis (PCA) separated the six liver tissue samples into two distinct categories (Fig. 5B). All 2324 quantifiable crotonylation sites in 848 proteins were normalized according to Y. Zhang et al. the corresponding quantitative proteomics analysis for each sample. We found that Kcr is prevalent in mouse liver and almost every enzyme that catalyzes intermediate metabolism, including fatty acid oxidation, the TCA cycle, gluconeogenesis, the urea cycle, glycogen metabolism, and glycolysis, was crotonylated. We next analyzed the flanking sequences of these sites and found that aspartic acid (D) and phenylalanine (F) were overrepresented, whereas glutamine (Q), arginine (R), and cysteine (C) were less than expected at the −1 and + 1 positions surrounding the lysine crotonylation sites (Fig. 5C).

Validation of Kcr changes on Ehhadh and Crot
In our study, the most significantly changed Kcr sites were mainly located at key enzymes in peroxisomal beta-oxidation including Ehhadh, Crot, Scp2, Hsd17b4, Acaa1a and Abcd3. Interestingly, protein levels of these enzymes carrying a number of downregulated Kcr sites were significantly upregulated in Acox2 −/− mice (Fig. 6A, B). All six Kcr sites in Crot, including K 33 , K 69 , K 489 , K 384 , K 226 , and K 132 , were downregulated by 0.391-, 0.426-, 0.494-, 0.522-, 0.544-, and 0.664-fold, respectively. Ehhadh K 329 , K 344 , and K 572 decreased by 0.284-, 0.448-, and 0.621-fold, respectively, whereas K 335 did not significantly change (fold-change = 0.906) (Fig. 6C). A total of 17 Kcr sites were identified in Scp2. Sequencing alignment indicated that Ehhadh K 329 and K 572 (Fig. 6D) and Crot K 69 , K 384 and K 489 (Fig.  6E) were highly conserved across species, including zebrafish or xenopus. We then generated Crot or Ehhadh mutants in which lysine Kcr sites were simultaneously mutated to arginine by site-  specific mutagenesis to mimic a decrotonylated status. Crotonylation of Ehhadh (Fig. 6F, G) and Crot (Fig. 6H, I) was further confirmed by using immunoprecipitation assays in HEK-293T and HepG2 cell lines overexpressing Myc-tagged Ehhahdh or Crot recombinant proteins, respectively. Mutating lysine to arginine in Kcr sites of Crot (K 69 R and K 384 R) or Ehhadh (K 329 R, K 344 R and K 572 R) reduces Kcr levels to various degrees in vitro ( Fig. 6F-I), which further validate the data reliability of crotonylome analysis.
Site-specific regulatory role of Ehhadh Kcr on the transcriptome in vitro Ehhadh K329cr (fd = 0.284) and K335cr (fd = 0.906) were differentially regulated in Acox2 −/− mice, which implicates the  possible site-specific regulatory role of Ehhadh Kcr on biological processes. To mimic the upregulation of Ehhadh and investigate the site-specific effects of Kcr changes, HEK-293T cells were transfected with control (Ctrl) and Ehhadh WT or mutants (K to R) recombinant Myc-tagged plasmids, as indicated. Overexpression of Ehhadh was validated through western blotting. We next performed RNA-seq based transcriptome analysis with three biological replicates for each sample. Subsequent PCA analysis based on 3 biological replicates effectively separated K572R from Ehhadh WT or the other mutants (Fig. S5A). Hierarchical clustering results and heatmap representation of DEGs profiles are shown (Figs. 7A and S5B). Of note, Top3b (DNA topoisomerase III beta) was significantly downregulated in cells expressing Ehhadh K572R when compared with WT. GO enrichment analysis revealed that "DNA topoisomerase type I (single strand cut, ATP-independent) activity" was the most significantly enriched term on sitespecific DEGs from K572R probably due to the downregulation of DNA topoisomerase III beta (TOP3B) (Fig. 7B). Subsequent cDNA amplification (Fig. 7C), real-time (RT) quantitative PCR (Fig. 7D), and western blot (Fig. 7E) uniformly confirmed that TOP3B expression was downregulated in cells transfected with Ehhadh K572R when compared with Ehhadh WT and other mutants. These findings suggest that non-histone Kcr has sitespecific regulatory roles in biological processes.
Previous studies found that loss of TOP3B leads to increased DNA damage and impaired recovery after replication stress [34,35]. We utilized a anti-gammaH2AX antibody, a sensitive indicator of DNA damage and replication stress, to perform immunofluorescence staining (Fig. 8A) and Western blot (Fig.  8B). We found that gammaH2AX formation is enhanced in cells expressing Ehhadh K572R, when compared with the WT and the other mutants. ATM is the major kinase involved in the phosphorylation of H2AX in response to DNA double-strand breaks (DSBs) [36][37][38]. Thus, we next evaluate the phosphorylation levels of ATM-CHK2 pathway and found that phosphorylation of ATM (Ser1981) and CHK2 (Thr68) were significantly upregulated in cells expressing Ehhadh K572R when compared with the WT (Fig. 8B). Furthermore, compared with WT mice, Top3b protein levels were significantly downregulated in the liver of Acox2 −/− mice by Western Blot (Fig. 8C, D) (p-value < 0.01, n = 6). These findings suggested that non-histone Kcr changes caused by Acox2 knockout might be involved in DNA damage responses.

DISCUSSION
Our study highlights the oncogenic property of Acox2 deficiency in tumor progression as those of our previous [19] and other studies [39]. Acox2 could be a critical regulator for metabolic homeostasis through post-translational modifications, probably Kcr. Meanwhile, our results demonstrated that Kcr is very prevalent in liver and virtually every enzyme in glycolysis, the tricarboxylic acid (TCA) cycle, and fatty acid metabolism was crotonylated. Importantly, most of downregulated non-histone Kcr sites in Acox2 −/− mice were located in metabolic enzymes, which extend the scope of metabolism regulation by Kcr to an extent comparable to that of other major posttranslational modifications such as acetylation [40] and ubiquitination [41]. In our study, lymphocytes were diffusely positive for CD20 in Acox2 −/− mice. The characteristic of the tumor-infiltrating leukocytes could impact clinical outcome in HCC [42]. Although the role of tumor-infiltrating B cells remains controversial, two previous studies reported that margin-infiltrating CD20(+) B cells correlate with favorable prognosis in HCC [43] and CD20(+) showed a trend towards better DFS [44]. Interaction between tumor-infiltrating B cells and T cells controls the progression of HCC [45]. These findings support the anti-tumor activity of CD20(+) B cells in HCC.
A recent study found that histone crotonylation is crucial for endoderm differentiation by regulating metabolic switch [13]. Differentiation of pluripotent embryonic stem cells to mesoderm and endoderm triggers a metabolic switch from glycolysis to oxidative phosphorylation [46]. This metabolic switch is believed to be important to release stem cells from pluripotency and promote differentiation [47]. Interestingly, this switch is the reverse process for the classic Warburg effect [48]. Thus, Kcr changes caused by Acox2 deficiency might be directly involved in the metabolic switch that is essential for tumor transformation in liver tissues of Acox2 −/− mice.
Several key peroxisomal enzymes, including Ehhadh, Crot, Scp2, and Hsd17b4, carrying decreased Kcr sites were uniformly upregulated in Acox2 −/− mice. Studies have found that acetylation can directly affect enzyme stability [49]. For example, the PEPCK1 3KacR mutant was more stable than the wild type, whereas the PEPCK1 3KacQ mutant remained unstable [49]. Acetylation and crotonylation have similar mechanisms and are both involved in metabolic regulation. Since histone acetyltransferases p300, pCAF, and MOF also have histone crotonyltransferase activity [15,16], whereas the classic histone deacetylases HDAC3 and SIRT1/2/3 can remove Kcr under different conditions [15,17], we postulated that changes at non-histone Kcr levels were involved in stability regulation of metabolic enzymes. A recent study found that CDYL-regulated histone Kcr, mainly H2b (K 12 cr), controls the expression of sex chromosome-linked escaped genes in postmeiotic spermatogenic cells [12]. In the present study, we identified an upregulated Kcr site at H2b (K 86 cr) whereas H2b (K 12 cr) was not significantly changed in Acox2 −/− mice. Thus, investigating the site-specific biological function of H2b (K 86 cr) on liver homeostasis by generating point knock-in mutation would be a potential avenue for understanding these results.
In this study, many downregulated Kcr sites in Acox2 −/− mice were located in proteins involved in mitochondrial beta-oxidation, including Hadha and Eci1/2, as well as the TCA cycle, including Idh2. Thus, Kcr might provide a vehicle for the cross-talk between peroxisome and mitochondria, which is critical for metabolic regulation. A recent study found that CROT promotes fatty acid metabolism and is involved in mitochondrial dysfunction during vascular calcification [50]. Thus, dysregulated Kcr sites in peroxisomal enzymes could be developed as a potential therapeutic target for human disease.
In this study, Top3b expression is downregulated in the cells expressing Ehhadh K572R as well as in the liver tissues of Acox2 −/ − mice. Loss of TOP3B was recently reported to increase DNA damage and genome instability [34,35]. DNA damage sensor gamma-H2AX is increased in preneoplastic lesions of HCC [51]. Mutating the putative Ku70 phosphorylation sites in increased gamma-H2AX and spontaneous induction of HCC [52]. Phosphorylation of Chk2, a central effector of DNA damage response, at threonine68 (Thr68) leads to Chk2 activation and Patients with HCC had increased CHK2 mRNA [53]. Interestingly, a recent study by Carloni et al found that CHK2 overexpression enhances chromosomal instability and HCC progression [54]. In addition, a protein called TAZ in pre-tumor NASH-hepatocytes promotes DNA damage and thereby induce hepatocarcinogenesis [55]. These findings suggested that downregulated Top3b might cause genome instability that is benefit for tumorgenesis in the liver tissues of Acox2 −/− mice. On the other hand, non-histone Kcr was also downregulated in the kidney tissues of Acox2 −/− mice. Interestingly, quantitative proteomic analysis indicated that protein level of MRE11, a double-strand break repair nuclease, was significantly upregulated in the kidney tissues of Acox2 −/− mice (data not shown). Besides self-activation by autophosphorylation at S1981, ATM was activated via the MRE11 complex [56]. Interestingly, MRE11 complex could interact with both ATM and RPA1 [57,58]. A recent study by Yu et al. found that RPA1 Kcr plays a key role in homologous recombination DNA repair [12]. Thus, upregulated MRE11 might contribute to the DNA damages responses in the kidney tissues of Acox2 −/− mice. Over all, our findings provide additional evidences to support the role of nonhistone Kcr on DNA damage responses.
The main limitation of our study is that we could not precisely determine the mechanism of different changes in Kcr for histone and non-histone proteins in Acox2 −/− mice. Thus, some unknown mechanism might regulate this biological process. We postulated that the imbalanced distribution of crotonyl-CoA in the nucleus and cytoplasm is the main cause for this phenomenon. Overall, our results suggest that Acox2 is a regulator for protein posttranslational modifications, mainly Kcr, that play important roles on metabolic homeostasis and liver cancer progression in mice.
C57BL/6n female mice (7-8 weeks old) were used as embryo donors. Briefly, fertilized embryos (zygotes) were collected from the oviducts, and Cas9 mRNA (100 ng/μL) plus sgRNA (50 ng/μL) targeting Acox2 were mixed and injected into the cytoplasm of fertilized eggs with both pronuclei visible in Chatot-Ziomek-Bavister medium. The injected zygotes were then cultured in Quinn's Advantage cleavage medium (In Vitro Fertilization, Inc.) containing SCR7 (50 μm; TOCRIS Bioscience, Bristol, UK) for~24 h, and 18-20 two-cell-stage embryos were transferred into the oviduct of a pseudo-pregnant ICR female mouse at 0.5 dpc (days postcoitus). All mice had access to food and water. All experiments were performed in accordance with Health Guide for the Care and Use of Laboratory Animals. Genomic DNA was extracted from tail tips, and PCR-based genotyping was performed. Sequences of sgRNA and primers are provided in Supplementary data. For all studies, littermate WT and Acox2 −/− mice derived from heterozygous parents were used.

Tandem affinity purification
Tandem affinity purification (TAP)-MS analysis was performed as described [60]. Briefly, MHCC97H and HEK-293T cells (ATCC®ACS-4500™) were grown in DMEM High Glucose (4.5 g/L) (Gibco™), supplemented with 10% FBS (Gibco), 100 IU/mL penicillin and 100 μg/mL streptomycin at 37°C in a 5% CO 2 atmosphere. Cells were then transfected with pMCB-SBP-Flag-ACOX2 containing a puromycin resistance marker. Cells were lysed on ice in 0.1% NP40 buffer (20 mM Tris-HCl [pH 8.0], 150 mM NaCl) and centrifuged to remove cell debris; the resulting cell lysates were incubated with streptavidin-conjugated beads (GE Healthcare) for 3 h at 4°C. The precipitates were washed three times with 0.1% NP40 buffer, 2 times with ddH 2 O, and three times with 50 mmol/L NH 4 CO 3 and subjected to tryptic digestion at 37°C overnight. The peptides in the supernatant were collected by centrifugation, dried in a speed vacuum (Eppendorff), and dissolved in NH 4 CO 3 buffer containing 0.1% formic acid and 5% acetonitrile before being subjected to mass spectrometry.

Kcr peptide enrichment
To enrich Kcr modified peptides, tryptic peptides were dissolved in NETN buffer (100 mM NaCl, 1 mM EDTA, 50 mM Tris-HCl, and 0.5% NP-40 [pH 8.0]) and then incubated with antibody beads (PTM Bio Inc., Hangzhou) as described [12] at a ratio of 15-μL beads/mg protein at 4°C overnight. The antibody beads were washed four times with NETN buffer and two times with ddH 2 O. The Kcr peptides were then eluted by adding the elution buffer with 0.1% trifluoroacetic acid. For LC-MS/MS analysis, the resulting Fig. 6 Downregulated non-histone Kcr sites on metabolic enzymes. A Key peroxisomal enzymes including Ehhadh, Scp2, Crot, and Hsd17b4 carrying downregulated Kcr sites were significantly upregulated (fd > 1.5) in Acox2 −/− mice. B Upregulation of Ehhadh expression in Acox2 −/− mice was verified through western blotting. C Heatmap showing profiles of all quantifiable Kcr sites in Crot, Ehhadh, and Scp2 before or after normalization according to quantitative proteomics analysis. D, E Sequence alignment of identified Kcr sites in Ehhadh (D) or Crot (E) in different species including xenopus and zebrafish. F, G Replacing lysine with arginine at Kcr sites lead to decreased Ehhadh Kcr levels in 293T (F) and HepG2 (G) cells as indicated. Cells were transfected with Myc-tagged WT and point mutants of Ehhadh. Cell lysates were immunoprecipitated with an anti-Myc antibody followed by immunoblotting with a pan-Kcr antibody. H, I Replacing lysine with arginine at Kcr sites leads to decreased Crot Kcr levels in 293T cells (H) and HepG2 cells (I). Representative image from three independent experiments was displayed.
peptides were desalted with C18 ZipTips (Millipore) according to the manufacturer's instructions.

Tandem mass tagging proteomic analysis
Tandem mass tagging-based proteomics analysis was supported by Jingjie PTM Biolabs (Hangzhou, China) as described [61]. The detailed procedures are provided as Supplementary data. For GO analysis, proteins were classified using GO annotation into three categories: biological process, cellular compartment, and molecular function. For each category, a twotailed Fisher's exact test was employed to test the enrichment of differentially modified proteins against all identified proteins. GO with a corrected p < 0.05 was considered significant and GO term enrichment was performed using DAVID 6.8.

Site-directed mutagenesis and immunoprecipitation
pCMV6-Ehhadh (Myc-DDK-tagged) and -Crot (Myc-DDK-tagged) mouse ORF clones were purchased from OriGene Technologies Inc (Rockville, USA). The QuickChange site-directed mutagenesis kit (Stratagene; Agilent Technologies) was used to substitute the identified lysine Kcr sites with arginine and generate four mutant Ehhadh variants (K 329 R, K 335 R, K 344 R, and K 572 R) and five mutant Crot variants (K 33 R, K 69 R, K 143 R, K 384 R, and K 489 R), which were confirmed by Sanger sequencing and compared to a reference sequence (RefSeq accession number: NM_023733 for Crot and NM_023737 for Ehhadh). Then HepG2 or 293T cells were transfected with Myc-tagged Crot and Ehhadh constructs followed by anti-Myc immunoprecipitaion as above.

RNA sequencing
Transcriptome libraries were prepared using a TruSeq RNA Sample Preparation kit (Illumina) and then sequenced using the Illumina HiSeq PE platform at 2 × 151-bp read lengths as previously [62]. The expression level of each transcript was calculated according to fragments per kilobase of exon per million mapped reads. Differentially expressed genes (DEGs) between groups were selected based on a log fold-change >2 and a false discovery rate FDR < 0.05. Gene ontology (GO) functional-enrichment and Kyoto Encyclopedia of Genes and Genomes pathway analyses were performed using GOATOLLS and KOBAS. DEGs were considered significantly enriched when the p value was less than 0.05 following Bonferroni's correction. Fig. 7 Site-specific regulatory role of Ehhadh K572cr on the transcriptome in vitro. A Hierarchical clustering and Heatmap showing the profiles of differentially expressed genes (DEGs) in 293T cells expressing Ehhadh K572R when compared with Ehhadh WT. B Gene Ontology enrichment analysis revealed that "DNA topoisomerase type I (single strand cut. ATP-independent) activity" was the most significantly enriched term in the K572R vs WT group. C, D mRNA levels of TOP3B were further confirmed by cDNA PCR (C) and quantitative real-time PCR (D). *p < 0.05; ***p < 0.001 (one-way ANOVA test, n = 4). E Western blot showed protein levels of TOP3B was downregulated whereas expression of FH and HK2 was not affected in cells overexpression K572R when compared with WT and other mutants. Representative image from three independent experiments was displayed.

cDNA amplification and RT-qPCR
Total RNA was isolated from HEK-293T cells transfected with mouse Myctagged Ehhadh WT or mutant constructs using the RNAprep pure Cell Kit (DP430, Tiangen, Beijing, China) followed by first-strand cDNA synthesis. cDNA was then amplified using PCR with TOP3B primers followed by Sanger sequencing. RT-qPCR was performed using the FastFire qPCR PreMix (SYBR Green) (FP207, TIANGEN) on an ABI StepOnePlus instrument. The delta-delta-Ct (ddCt) algorithm was utilized to analyze the relative change in gene expression by using the housekeeping gene GAPDH as the internal reference. The data are presented as the mean ± SD of four independent experiments; each sample was assayed in triplicate in each experiment. RT-qPCR primer sequences are available in Supplementary data.

Immunoblotting
After undergoing electrophoresis and membrane transfer, the proteins were probed with one of the following primary antibodies in either 5% nonfat milk or 5% bovine serum albumin: anti-pan Kcr (PTM, 501, 502),

Statistics
Data are presented as individual samples and mean ± SEM. Student's twotailed unpaired t-test or one-way ANOVA was used to determine statistical significance of differences between WT and mutant groups using GraphPad Prism software v. 5.01 (GraphPad Software, Inc., La Jolla, CA, USA). Differences were significant when p values were <0.05.

DATA AVAILABILITY
The RNA-sequencing data have been deposited in the Gene Expression Omnibus (GEO) repository under accession number GSE194202. Other data that support the fndings of this study are available from the corresponding authors upon reasonable request. Fig. 8 Non-histone Kcr changes caused by Acox2 knockout might be involved in DNA damage responses. A Immunofluorescence staining with anti-gammaH2AX antibody in cells expressing Ehhadh WT or (K-R) mutants. B Immunoblot analysis of gammaH2AX and pan-or phospho-ATM and CHK2 in cells expressing Ehhadh WT or (K-R) mutants. Beta-actin served as a loading control. C, D Top3b protein levels were significantly downregulated in the livers of Acox2 −/− mice by western Blot (C). Quantification data of Top3b expression level when normalized against Beta-actin (D). **p value < 0.01 (two-tailed Student's t test; n = 6). A.U arbitrary units. Representative image from three independent experiments was displayed.