Kynurenine analog 3-HAA is a ligand for transcription factor YY1

Kynurenine, a metabolite of tryptophan, promotes immune tolerance in development and tumor evasion by binding to the aryl hydrocarbon receptor (AHR). However, the IDO inhibitors, blocking kynurenine generation, fail in stage III of clinical trials in several tumors for unknown reasons. Here, we report that 3-hydroxyanthranilic acid (3-HAA) induces apoptosis and synergizes with IDO inhibitors by increasing the suppression of IDO inhibitors on HCC xenograft growth. The content of 3-HAA, a catabolite of kynurenine, is lower in tumor cells by downregulating its synthetic enzyme KMO/KYNU and/or upregulating its catalytic enzyme HAAO. Overexpression of KMO suppresses tumor formation and tumor growth by increasing endogenous 3-HAA while adding exogenous 3-HAA also inhibits tumor growth. metabolite mobility shift assay (EMSA). The synthesized oligonucleotide was labeled at the 5’ terminus with FAM, and YY1 was purified using a Hitrap heparin column.

The essential amino acid tryptophan is catabolized mainly through the serotonin pathway in the brain or the kynurenine pathway in the liver (Chen and Guillemin, 2009;Schrocksnadel et al., 2006).
Tryptophan metabolism is enhanced in various tumors by upregulating the expression of indoleamine 2,3dioxygenase 1/2 (IDO1/2), the rate-limiting enzyme in the kynurenine pathway. Kynurenine, a catabolite of tryptophan, increases immune tolerance in development and disease by directly binding to the aryl hydrocarbon receptor (AHR) (Opitz et al., 2011;Sharma et al., 2007), and enhances tumor immune evasion to promote tumor growth (Opitz et al., 2011) (Li et al., 2011) (Schwarcz et al.). The 3hydroxyanthranilic acid (3-HAA), a derivative of kynurenine was reported to exert anti-inflammatory effects by selectively inducing the apoptosis of activated T cells (Krause et al., 2011;Lee et al., 2010).
However, the function of other kynurenine derivatives largely remains unclear. Here, we report 3-HAA, selectively decreased in various tumors, regulates the activity of transcription factor YY1 by directly binding and consequently induces tumor cell apoptosis and suppresses HCC growth in vitro and in vivo.

3-HAA is decreased in tumor cells
To comprehensively understand the effect of kynurenine derivatives on tumor, tryptophan catabolites were first analyzed in clinical HCCs using liquid chromatography-tandem mass spectrometry (LC-MS/MS) and gas chromatography-mass spectrometry (GC-MS). The concentration of kynurenine catabolite 3-HAA decreased in 37 cases of HCC (p< 0.01) and 42 cases of esophageal carcinomas (p<0.01) compared to the matched paratumor tissues (Figs. 1A and S1A). Conversely, the concentration of tryptophan and kynurenine was higher in these HCCs and esophageal carcinomas than in the matched paratumor tissues, respectively. There was no significant difference in 3-hydroxykynurenine (HK) between tumors and adjacent non-cancerous tissues. Consistent with this observation, the concentration of 3-HAA was also lower in seven HCC cell lines tested than in normal hepatic LO2 cells, whereas the content of tryptophan and kynurenine increased in these tested HCC cell lines (Fig. 1B). Immunohistochemical analysis further confirmed lower 3-HAA content in clinical HCC tissues than in adjacent non-cancerous tissues (Fig. 1C).
Metabolic flux analysis using labeled tryptophan revealed catabolism into kynurenine but not HK or 3-HAA in HCC cells, and the newly generated kynurenine was secreted into the culture medium (Fig.   1D). Although the expression of 3-HAA degrading enzyme hydroxyanthranilate 3,4-dioxygenase (HAAO) varied in HCC cells, both western blotting and immuno-histochemical analysis showed that kynurenine 3-monooxygenase (KMO) and kynureninase (KYNU), enzymes converting kynurenine into 3-HAA, were downregulated in HCC cells (Fig. 1E &S1B). The analysis of HCC expression profile in the TCGA database also supported our finding (Fig. S1C), suggesting tumor cells reprogram the tryptophan metabolism by modulating the upstream/downstream catalytic enzymes.
Furthermore, overexpression of KMO or knockdown of HAAO significantly increased the concentration of 3-HAA in HCC SMMC7721 cells, 6 times more than control cells, but not the HK, picolinate (PA), or quinolinate (QA) (Fig. 1F & S1D). These observations suggest that 3-HAA is selectively downregulated in cancer cells by downregulating KMO expression and/or upregulating HAAO.

3-HAA inhibits tumor growth by inducing apoptosis
To determine whether 3-HAA could inhibit tumor growth, exogenous 3-HAA was added to HCC cultures, and cell proliferation was assessed. Indeed, 3-HAA significantly inhibited HCC cell growth and colony formation (Figs. 2A and S2A). In contrast, the other tryptophan metabolites Kyn, PA, and QA did not substantially affect growth or colony formation of HepG2 or SMMC7721 cells. To verify that this effect was caused by 3-HAA and not its substrate, the cellular concentrations of 3-HAA and kynurenine were analyzed by LC-MS/MS in 3-HAA-treated tumor cells. Although the content of endogenous 3-HAA varied among different HCC cell lines, 3-HAA treatment increased the cellular concentration of 3-HAA, particularly in SMMC7721 and HepG2, and the downstream substrate but not kynurenine (Fig. S2B).
To understand the mechanism by which 3-HAA inhibited HCC cell growth, gene expression profiling of 3-HAA-treated HCC cells was analyzed by the gene ontology (GO) enrichment. The GO analysis revealed that the cell death-related pathways were highly activated in 3-HAA-treated HCC cells (Fig. 2B). To further determine what signal mediates 3-HAA-induced cell death, tumor cells were treated with 3-HAA and then analyzed for markers of apoptosis, autophagy, and necrosis (Fig. S2C). Apoptosis was increased in a dose-dependent manner by TUNEL assay in the SMMC7721 cells (Fig. 2C). Levels of cleaved caspase 3 and cleaved PARP were increased in a dose-and time-dependent manner in the SMMC7721 and HepG2 cells (Fig. 2D). However, 3-HAA treatment did not significantly alter levels of autophagy markers LC3-II and p62 or necrosis marker RIP3 in the SMMC7721 or HepG2 cells (Fig.   S2C).
This effect was observed at ZVAD doses of 50 and 100 μM. 3-HAA treatment also increased apoptosis in HCC xenografts, as observed based on TUNEL foci and flow cytometry of Annexin V (Figs. 2F and 2G).
To determine the effect of endogenous 3-HAA on HCC tumor formation, the cell growth was analyzed in SMMC7721 cells overexpressing KMO or knocking down HAAO enzymes. Either overexpression of KMO or knock down of HAAO inhibited cell growth of HCC cells (Fig. 2H & S2E). Moreover, the tumor formation assay showed KMO overexpression suppressed tumor formation and tumor growth in HCC xenograft model, (9/10) versus (4/10) (Fig. 2I, S2F & S2G). The flow cytometry analysis showed that overexpression of KMO increased apoptosis of HCC cells, two times more than control cells (Fig. 2J & S2H). Remarkably, the Kaplan-Meier survival analysis showed that HCC patients with high KMO expression had longer disease-free survival than patients with low KMO expression (Fig.   2K). These observations suggest that 3-HAA is a negative regulator for tumor formation and tumor growth.

3-HAA suppressed tumor growth by upregulating DUSP6 expression
To determine the mechanism by which 3-HAA induces tumor cell apoptosis, RNA sequencing was used to profile gene expression in SMMC7721 or HepG2 cells after 0, 1, 8, or 24 h treatment with 3-HAA according to the screening strategy previously applied (Opitz et al., 2011). At all three time points after the start of 3-HAA treatment, the top 21 upregulated genes were selected (Fig. 3A). The expression of the top seven of these genes was individually verified by real-time PCR. Immunoblotting of the top two upregulated genes also showed elevated levels of DUSP6 and IGFBP1 in SMMC7721 and HepG2 cells (Fig. 3B). These results suggest that 3-HAA alters the gene expression profile of HCC cells, with DUSP6 and IGFBP1 as two of the most upregulated genes. However, the clinical data showed that the overall survival of HCC patients was only associated with the expression level of DUSP6 (p < 0.05), but not IGFBP1 (p > 0.05). Patients expressing a high level of DUSP6 were longer than patients expressing a low level of DUSP6 (p < 0.05) ( Fig. 3C and S3A). The multivariate analysis showed that DUSP6 was a major factor in the HCC patients' survival among the top 4 upregulated genes (Fig. S3B), and the corrective analysis with the clinical characteristics also supported this finding (Fig. S3C), indicating that DUSP6 is important to 3-HAA-induced apoptosis.
To demonstrate whether DUSP6 mediates 3-HAA-induced tumor cell apoptosis, the effects of DUSP6 on HCC cell growth were first analyzed in HCC cells. DUSP6 knockdown restored growth of HepG2 and SMMC7721 cells inhibited by 3-HAA (Figs. 3D and S3D), whereas it did not affect the growth of untreated HCC cells (Fig. S3E). 3-HAA induced apoptosis to a smaller extent in DUSP6depleted SMMC7721 cells than in the control cells, based on flow cytometry using Annexin V (Fig. 3E).

3-HAA upregulated DUSP6 expression by binding with transcription factor YY1
To determine which transcription factor or co-activator mediates 3-HAA regulation of gene expression, the 38 common transcription factors or co-activators were first selected from those proteins potentially bind to the promoter region (-5000 to +1) of top 4 genes (http://gtrd.biouml.org) (Yevshin et al., 2019) (Fig. 4A and S4A). Moreover, levels of chromatin proteins in 3-HAA-treated SMMC7721 and HepG2 cells were quantified by tandem mass-tagged quantitative proteomics analysis (TMT). The 91 proteins consistently and increasingly bound to chromatin at the 1 st hour and the 8 th hour post 3-HAA treatment (Fig. S4B), and YY1 was the only protein overlapped with the predicted transcription factors that potentially bind to the promoter region of the top 4 genes (Figs. 4B). Both the immunoblotting of nuclear fraction and immunofluorescent staining also showed that 3-HAA increased YY1 nuclear accumulation ( Fig. 4C & S4C). Moreover, analysis on gene expression profile showed that about 16% of 3-HAA-regulating genes were the same with YY1-targeting genes, including DUSP6 (Fig. S4D), suggesting transcription factor YY1 could regulate DUSP6 expression after 3-HAA treatment.
Since YY1 is a transcription factor, we speculated that 3-HAA could regulate YY1 transcription activity on DUSP6 gene promoter. Closer analysis of the DUSP6 promoter region using online-based prediction tools (jaspar.genereg.net and ecrbrowser.dcode.org) (Khan et al., 2018a;Khan et al., 2018b) revealed a novel potential YY1 binding DNA fragment at positions -1,145 to -1,134 (TCCATCCGGCTT), which is distinct from the reported consensuses binding sequence (CAANATGGCGGC) (Kim and Kim, 2009). To determine whether YY1 regulates DUSP6 expression by binding this novel sequence, each DNA fragment was added to a luciferase reporter gene, and YY1driven luciferase expression was measured by its enzyme activity. Higher luciferase activity was observed with the full length or partial DUSP6 promoter containing this novel specific sequence as well as the consensus YY1 binding sequence. Luciferase activity decreased when mutations involving this novel binding site (mut2) occurred (Fig. 4H). 3-HAA increased YY1 binding to this novel binding sequence in a dose-dependent manner, as evidenced by an in vitro electrophoretic mobility shift assay (Fig. 4I).
Moreover, the quantitative PCR analysis following chromatin immunoprecipitation of YY1 revealed that 3-HAA promoted YY1 binding to the consensus sequence of p53 promoter region (positive control) and the novel binding sequence in the DUSP6 promoter region, as reflected by 3-HAA-induced YY1 enrichment (Figs. 4J and S4H). ChIP sequencing analysis further confirmed that 3-HAA induced the union peak formation of YY1 on the promoter region of DUSP6, IGFBP1, NR0B2 and IER3 genes ( To determine whether 3-HAA promotes DUSP6 expression by binding YY1, we first tested whether 3-HAA directly associates with YY1 in vitro using nuclear magnetic resonance. Dose-dependent signal attenuation was observed in the T1r NMR spectrum, and positive saturation transfer difference (STD) signals were detected in the STD spectrum, suggesting that YY1 interacts with 3-HAA (Fig. 4L). Surface plasmon resonance experiments suggested a moderate binding affinity with KD of 121.7 μM (Fig. 4M).

PKCζ phosphorylates YY1 at Thr 398 in response to 3-HAA
Previous studies showed that phosphorylation regulated YY1 DNA binding activity (Daub et al., 2008;Kaludov et al., 1996). Thus, the phosphorylation of YY1 were first analyzed 2 hours after 3-HAA treatment to determine whether 3-HAA induced YY1 phosphorylation. The immunoblotting following phosphorylated protein enrichment showed that 3-HAA treatment also increased YY1 phosphorylation, and mass spectrometry analysis revealed that the T398 of YY1 were phosphorylated by 3-HAA (Fig. 5A).
The function analysis displayed that the T398A mutation of YY1 suppressed 3-HAA-upregulated DUSP6 expression and reduced the level of cleaved Caspase 3/cleaved PARP, whereas the mimic phosphorylation of T398E mutation promoted DUSP6 expression even without 3-HAA treatment (Fig.   5D). The TUNEL assay and the flow cytometry analysis demonstrated that the T398A mutation of YY1 suppressed 3-HAA-induced apoptosis ( Fig. 5E & S5A), suggesting T398 phosphorylation of YY1 is critical for 3-HAA-induced apoptosis.
To identify the kinase for the 3-HAA-induced T398 phosphorylation of YY1, the kinase screening assay was performed on the peptide of FAQSTNLK. Three candidate enzymes AKT, mTOR, and PKCζ were chosen from predicted kinases based on the online software (www.cbs.dtu.dk; gps.biocuckoo.cn) (Mok et al.;Xue et al., 2011). Furthermore, proteomics analysis by mass spectrometry following YY1 immunoprecipitation showed that 3-HAA increased the association of YY1 with PKCζ, which were further confirmed by immunoblotting, suggesting 3-HAA recruits PKCζ to phosphorylate YY1 (Figs. 5F, S5B & 5G). The kinase PKCζ significantly increased the peptide phosphorylation, reflected by the autoradiogram on dot blot (Fig. 5H). Also, only the PKCζ inhibitor markedly decreased YY1 phosphorylation (Fig. 5I), suggesting PKCζ is the kinase for T398phosphorylation of YY1 induced by 3-

HAA.
Moreover, the PKCζ inhibitor markedly decreased the mRNA level and protein level of DUSP6 in the SMMC7721 cells depleted of endogenous YY1 and expressing exogenous wild type YY1, but not in the cells expressing T398A/T398E mutant YY1 ( Fig. 5J & 5K). Eventually, the PKCζ inhibitor decreased the level of cleaved Caspase 3 and cleaved PARP in the cells expressing exogenous wild type YY1 but not T398A/T398E mutant YY1 (Fig. 5K). The clinical data that the PKCζ expression level is closely correlated with the overall survival of the grade I HCC patients (Fig. S5C), further supports these findings.
Using computational software, we propose a binding model that 3-HAA binds to the site encompassing Gln396 to Thr398 of YY1 (Fig. 5L), which is consistent with our above finding that the Thr398 phosphorylation increasing the YY1 binding to the promoter of DUSP6 gene.

T398 phosphorylation of YY1 is critical for 3-HAA-suppressed HCC xenografts growth
To further evaluate the effects of T398 phosphorylation of YY1 on 3-HAA-suppressed HCC cell growth, the apoptosis and tumor growth were determined in HCC xenograft mice. In mice with control HCC xenografts, 3-HAA (100 mg/kg•day) slowed xenograft growth, leading to smaller tumors, whereas the same dose of 3-HAA had no remarkable effect on YY1 depleting xenograft growth (Fig. 6A). Both the TUNEL assay and flow cytometry revealed a noticeably higher percentage of apoptotic cells only in YY1 expressing HCC xenografts, but not in YY1 depleting HCC xenografts ( Fig. 6B-C & S6A). This result was obtained with SMMC7721 and HepG2 xenografts.
Moreover, in mice with HCC xenograft depleting endogenous YY1 and expressing T398A mutant YY1, 3-HAA (100 mg/kg•day) had no effect on tumor growth. Whereas, the same dose of 3-HAA significantly decreased xenograft growth expressing wild type YY1 (Fig. 6D). The TUNEL assay and flow cytometry also demonstrated a markedly higher percentage of apoptotic cells in xenografts expressing wild type YY1, but not in T398A mutant xenografts (Fig. 6E, 5F & S6B). In mice with HCC xenografts expressing T398E mutant YY1, the growth of xenografts was obviously slower than the xenografts expressing wild type YY1, and 3-HAA had no effects on these xenograft growth (Fig. 6E).
And, the PKCζ inhibitor (0.8mg/kg.day) treatment restored the growth of xenografts suppressed by 3-HAA ( To determine the pharmacokinetic of 3-HAA, the concentration of 3-HAA was analyzed at the time course of administration in the plasma of mice. As shown in Fig. S6D, the half-time of 3-HAA was 3.89 hours in the serum of mice. The concentration of 3-HAA reached 71.3 µmol per gram in tumors post 7 days treatment. Moreover, 3-HAA did not disrupt liver/renal function in Balb/c mice at a dose of 800 mg/kg.day, which was six-fold higher than the tested dose (Fig. S6E). These results highlight the promise of 3-HAA as a potential HCC therapy.

Discussion
Tryptophan metabolism plays a very important role in development and tumor progression. It not only provides critical intermediates for anabolism but also regulates cell signaling. IDO converts tryptophan into kynurenine, which is a well-known functional metabolite of tryptophan and could be further catabolized to 3-HAA. Kynurenine binds AHR to induce immune suppression (tolerance), which is associated with successful embryo implantation but also poor prognosis in various malignancies. The inhibition of IDO1/2 suppresses tumor formation in animal models but failed in stage III of clinical trials. This is probably due to the expression of the other kynurenine generating enzyme TDO, which may not be suppressed by the IDO inhibitor. However, the 3-HAA treatment reversed the tumor-promoting effect of kynurenine and significantly improved the efficacy of IDO1/2 inhibitors on HCC xenografts, suggesting 3-HAA as a negative feedback regulator reverses the tumor cell-hijacked tryptophan metabolism.
The 3-HAA exerts anti-inflammatory and neuroprotective effects by selectively inducing the apoptosis of activated T cells or suppressing microglia/astrocytes that express cytokines and chemokines (Krause et al., 2011;Lee et al., 2010). The 3-HAA induced the expression of cytoprotective enzyme hemeoxygenase-1 in astrocytes and microglia, the latter is an enzyme with proven anti-inflammatory and cytoprotective activities (Brusko et al., 2005;Ryter and Choi, 2009). The 3-HAA-induced apoptosis of activated T lymphocytes are linked to oxidative stress and induction of caspases (Hayashi et al., 2007;Hiramatsu et al., 2008). These results indicate that 3-HAA-has a distinct biological function in different type of cells. However, the biological function of 3-HAA largely remains unclear.
In this study, we report that 3-HAA is downregulated in HCC cells and HCC tissues due to the downregulation of KMO and KYNU enzymes as well as the upregulation of the HAAO enzyme, which are regulated by kynurenine signal. The excessive kynurenine generated from tryptophan in HCCs is exported. We also observed that the HCC cell sensitivity to 3-HAA, to some extent, were correlated with the cellular 3-HAA concentration post exogenous 3-HAA treatment. However, the variation of the Moreover, we also discovered that 3-HAA is a ligand of transcription factor YY1 and that binding of the two molecules leads to 3-HAA-mediated alterations in gene expression. 3-HAA up-or downregulated a number of genes via YY1 binding, including DUSP6, IGFBP1, and NR0B2 etc. YY1 may bind to two motifs in the DUSP6 promoter; the known consensus binding site, as well as a novel binding site, was identified in this study. The T398 phosphorylation of YY1 may change its structure conformation associating with DNA binding, promoting YY1 binding to its target sequence.
Although the in vitro measurement showed a higher dissociation constant of YY1 with 3-HAA (the KD50 of 3-HAA was 121.7 µM), this may not reflect the binding affinity of YY1 with 3-HAA in cells since the cellular contents of 3-HAA were less than 20 µM, even in HCC cells treated by 3-HAA or expressing KMO. Moreover, the expression level of solute carriers determines the cellular uptake of 3-HAA, thus, the contents of 3-HAA in 3-HAA-treated HCC cells were only 5-10 times more than untreated cells, similar to the level of endogenous 3-HAA in cells expressing KMO (Fig. 1F & 2B).
Taken together, our results have determined that 3-HAA is a functional metabolite regulating tumor cell fate by binding to and activating the transcription factor YY1 (Fig. 6K). Selective downregulation of 3-HAA by kynurenine signal appears to be essential for HCC growth. Exogenous 3-HAA induces tumor cell apoptosis and inhibits HCC growth, especially improves the efficacy of IDO1/2 inhibitors, suggesting its potential use in HCC therapy.      H. Transcription activity of YY1 on the DUSP6 promoter, as determined in a luciferase reporter assay.
The schematic depicts the plasmid encoding luciferase under the control of the DUSP6 promoter, which was truncated or mutated as indicated.
I. 3-HAA binding to YY1 determined by electrophoretic mobility shift assay (EMSA). The synthesized oligonucleotide was labeled at the 5' terminus with FAM, and YY1 was purified using a Hitrap heparin column.
K. The ChIP-sequencing analysis of YY1 on the DUSP6 gene. The HCC cells were treated with the indicated dose of 3-HAA prior to ChIP-sequencing.
L. NMR measurement of direct binding between 3-HAA and YY1. T1r NMR spectra for 6b (red) alone or in the presence of SPOP at 2.5 mM (blue), 5 mM (cyan), or 10 mM (green). The STD spectrum for 6b was recorded in the presence of 5 mM SPOP.
M. 3-HAA bound YY1 protein as shown by surface plasmon resonance. Graphs of equilibrium unit response versus concentrations are shown. The estimated Kd was 121.7 μM.

Figure 5 PKCζ phosphorylates YY1 at Thr 398 in response to 3-HAA
A. The YY1 phosphorylation was analyzed by the immunoblotting and the mass-spectrometry. The YY1 was blotted on the enriched phospho-proteins from SMMC7721 cells. The YY1 modification was analyzed by the mass-spectrometry following YY1 immunoprecipitation.
B. The YY1 mutation of T398A attenuates the 3-HAA-increased YY1 nuclear accumulation. The concentration of 3-HAA was 100 μM and treated for 24 h.
C. The T398A but not S247A mutation abolished 3-HAA-induced YY1 phosphorylation. The YY1 was conjugated with HA tag. The YY1 phosphorylation was detected by the T398 phospho-specific antibody.
D. The mutation of T398E in YY1 promoted DUSP6 expression, whereas the T398A mutation suppressed DUSP6 upregulation. The YY1 were fused with HA tag.
E. The YY1 mutation of T398A reduces the 3-HAA-induced apoptosis, analyzed by the TUNEL assay.
L. The proposed 3-HAA binding model with YY1. Note: The mouse xenografts were generated by the inoculation of 1.5 x 10^6 of SMMC7721 cells into the armpit of rear limb. The 3-HAA (100 mg/kg.day) was administered by intraperitoneal injection for 7 days. Tumor volumes are presented as mean ± SD (*: P < 0.05; **: P < 0.01.). Photographs on the right show representative xenografts in different groups.

Metabolism flux analysis by LC-MS/MS
For the flux experiment of tryptophan catabolism, tryptophan in medium was replaced by fully 13

Quantitative proteomics
Cell samples were sonicated three times on ice using a high-intensity ultrasonic processor (Scientz, Ninbo, Zhejiang, China) in lysis buffer (8 M urea, 1% Protease Inhibitor Cocktail). The supernatant was collected and proteins were reduced with 5 mM dithiothreitol for 30 min at 56 °C, then alkylated with 11 mM iodoacetamide for 15 min at room temperature in darkness. Following addition of 100 mM TEAB to dilute the urea to < 2 M, trypsin was added to the protein samples first at a trypsin-to-protein mass ratio of 1:50 for digestion overnight, then at a ratio of 1:100 for a second digestion lasting 4 h.
After trypsin digestion, peptides were desalted on a Strata X C18 SPE column (Phenomenex) and vacuum-dried. Peptides were reconstituted in 0.5 M TEAB and processed according to the manufacturer's protocol for TMT kit/iTRAQ kit. The tryptic peptides dissolved in 0.1% formic acid (solvent A) were directly loaded onto a custom-made reverse-phase analytical column (15-cm length, 75 μm i.d.) on an EASY-nLC 1000 UPLC system. The gradient to solvent B (0.1% formic acid in 98% acetonitrile) increased from 6% to 23% over 26 min, then from 23% to 35% in 8 min and then to 80% in 3 min, after which it remained at 80% for 3 min. The flow rate was constant at 400 nL/min.
Peptides were subjected to NSI source followed by tandem mass spectrometry (MS/MS) in Orbitrap Fusion TM Tribrid TM (Thermo, CA, USA) coupled online to the UPLC. Intact peptides were detected in the Orbitrap at a resolution of 60,000. Peptides were selected for MS/MS by NCE at 35; ion fragments were detected in the Orbitrap at a resolution of 30,000. A data-dependent procedure that alternated between one MS scan followed by 10 MS/MS scans was applied for the top 10 precursor ions above a threshold intensity, which were greater than 5 x 10 3 in the MS survey scan with dynamic exclusion of 30.0 s. The electrospray voltage applied was 2.0 kV. Automatic gain control was used to prevent the orbitrap from overfilling; 5 x 10 4 ions were accumulated to generate MS/MS spectra. For MS scans, the m/z scan range was from 350 to 1550. The fixed first mass was set as 100 m/z. MS/MS data were processed using the Maxquant search engine (version 1.5.2.8). Carbamidomethyl on cysteine was specified as a fixed modification, while oxidation on methionine was specified as a variable modification. FDR was adjusted to < 1% and the minimum score for peptides was set to > 40.

NMR detection of 3-HAA-YY1 interaction
NMR experiments based on the Carr-Purcell-Meiboom-Gill (CPMG) sequence and saturation transfer difference (STD) were applied to investigate 3-HAA-YY1 interactions. All NMR spectra were acquired at 25 °C on a 600 MHz Bruker Avance III spectrometer equipped with a cryogenically cooled probe (Bruker Biospin, Germany). Samples containing 200 µM 3-HAA and 200 µM 3-HAA in the presence of 5 µM protein were dissolved in Tris-HCl buffer (50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 5% DMSO, 95% D2O) and then used in NMR data acquisition. CPMGs were recorded using the pulse sequence of solvent-suppressed 1D 1 H CPMG (cpmgPr1d). The 90° pulse length was adjusted to approximately 11.82 μs. A total of 4 dummy scans and 64 free induction decays (FIDs) were collected into 13 K acquisition points, covering a spectral width of 8 kHz (13.3 ppm) and giving an acquisition time of 3 s. STD data were acquired using 4 dummy scans and a relaxation delay of 3 s, followed by 40-dB pulsed irradiation at a frequency of -1.0 ppm or alternatively 33 ppm. Total time to acquire the STD spectrum was 21 min with 128 FIDs.

Surface plasmon resonance and microscale thermophoresis
To measure the affinity of 3-HAA to YY1 protein, the dissociation constants were measured using a BIAcore T200 instrument (GE Healthcare, CA, USA) with a CM5 sensor chip (GE Healthcare). The YY1 protein was immobilized on a CM5 sensor chip in sodium acetate buffer (1 μg/mL, pH 5.5), and 3-HAA was gradually titrated at the indicated concentrations. The 3-HAA was injected at a flow rate of 30 μL/min. The association time was 120 s and the dissociation time was 420 s. The binding constant was calculated using a 1:1 Langmuir binding model via the BIAevaluation software.
The dissociation constant of 3-HAA for YY1 was also measured by microscale thermophoresis assay according to manufacturer's protocol (Nanotemper Technologies Ltd., Shanghai, China). YY1 protein was labeled using NT650-NHS labeling kit (Nanotemper Technologies Ltd., Shanghai, China), and the concentration of labeled YY1 protein was set to 50 nM. The unlabelled 3-HAA was gradually diluted from 2500 μM to 76.3 nM as indicated. YY1 and 3-HAA interacted in PBS buffer containing 2.5% DMSO, 0.05% Tween-20, 85mM NaCl. Measurements were made using the Monolith NT.115 (NanoTemper), and data were analyzed using MO.Affinity analysis (×64) software. Graphs were blotted using Prism 5 software.

TUNEL assay
The cover glass was placed in the 24-well plate, and the HCC cells were inoculated on the cover glass overnight. DMSO or 100 μM 3-HAA was added to culture medium for 48 hours. The cells were washed with PBS for 3 times. Add 0.5mL of 4% paraformaldehyde and fix cells at room temperature for 10 minutes. Cells were treated with 0.4%Triton X-100 for 5 minutes and washed with PBS. Cells on the cover glass were treated with TUNEL staining solution and incubated in a wet box at 37℃ for 1 hour.
DAPI staining solution was used to stain the nuclear for 5 minutes in dark. Cells were observed and photographed under a fluorescence microscope.

Xenograft HCC mouse model
Six-week-old male BALB/c mice were purchased form Lingchang, Shanghai, China. Xenograft mouse model of HCC was generated by injecting SMMC-7721 HCC cells (1.5×10^6) subcutaneously into the armpit of rear limb. Mice were injected intraperitoneally with 100 mg/kg 3-HAA or equal amount of DMSO every day. when tumors reaching 1 cm 3 , the PKC inhibitor Go6983 was injected intravenously every two days and the dose of PKC inhibitor was administered at the dose of 0.8 mg/kg.day.

DMSO 3HAA
Input IgG YY1 Ab D U S P 6 D U S P 6 D U S P 6 D U S P 6 D U S P 6 D U S P 6 M a r k e r  Figure S4 A. The common transcription factors potentially binding to the promoter region (-5000 to +1) of top 4 genes, which was up-regulated in 3-HAA-treated SMMC7721 cells. B. Proteins increasely binding to chromatin after 3-HAA treatment, analyzed by quantitative proteomics analysis. C. The localization of YY1 with or without 3-HAA treatment. D. The common genes regualted by both 3-HAA and YY1. E. The apoptosis analysis by TUNEL assay in SMMC7721 cells depleted of YY1 and/or overexpressing DUSP6.

RuvBL2
PKCζ  . Following a single Intraperitoneal dose, the mean plasma T1/2 of 3-HAA were 3.89 hrs and 4.13 hrs, respectively. T1/2, elimination half-life; Cmax, maximum concentration; AUC0-t, area under the curve of a plasma concentration from time zero to time t; AUC0-inf, area under the plasma concentration-time curve from time zero to infinity; Vz, volume of distribution; CL, total plasma clearance of drug; MRT, mean residence time.

CTRL PI PE-
F. The effect of 3-HAA on the liver/renal function and body wight of Balb/c mice. Figure S6 The T398 phosphorylation of YY1 is critical for 3-HAA-suppressed HCC xenografts growth. Related to Figure 6