HN1 Promotes Proliferation, Metastasis and Attenuates the Chemosensitivity of HCC Cells to Oxaliplatin by Inhibiting Degradation of HMGB1 Through Interacting With TRIM28

Background: The oxaliplatin-based chemotherapy has revealed an encouraging therapeutic ecacy for advanced hepatocellular carcinoma patients. However, the development of resistance limits its clinical utilization. In addition, the chemotherapy resistance in HCC is usually accompanied with other malignant phenotypes, such as cell proliferation and metastasis, which together result in poor prognosis of HCC patients. Therefore, efforts should be made to explore potential regulators which fuel multiple events of HCC progression. Methods: The qRT-PCR, western blot, immunohistochemistry and immunouorescence were performed to measure mRNA and protein expression. MTT assay, colony formation and Transwell assay were performed to evaluate cell proliferation and metastasis. Flow cytometry was performed to test cell apoptosis. Alkaline Comet assay was performed to measure DNA lesions. Transmission electron microscope analysis provided potent testimony of autophagy. The role of HN1 on the malignant phenotypes of hepatoma carcinoma was demonstrated in vitro and in vivo. Results: The immunohistochemistry analysis of HCC patient tissues revealed that the expression of HN1 was higher in HCC tissues compared to adjacent tissues and was associated with worse prognosis. In vitro, HN1 knockdown inhibited proliferation and metastasis of HCC cells, whereas HN1 overexpression promoted their proliferation and metastasis. In addition, we found that HN1 knockdown sensitized HCC cells to oxaliplatin, which is companied with deteriorated DNA damage and increased cell apoptosis in oxaliplatin-treated HCC cells. In vivo, HN1 knockdown inhibited the tumor growth and metastasis, and promoted the anti-cancer eciency of oxaliplatin. Mechanically, HN1 prevented HMGB1 from ubiquitination and degradation via autophagy-lysosome pathway, which is related to its interaction with TRIM28, and overexpression of HMGB1 can restore the malignant phenotypes of HN1 knockdown in HCC cells. Furthermore, we found that HN1 can regulate cellular autophagy via HMGB1, which is important to tumor-promoting effect of

Background Hepatocellular carcinoma (HCC) is the most common malignant hepatic disease, ranked as the sixth in tumor incidence and the third in leading cancer-related mortality worldwide [1]. Delay diagnosis and limited therapy deteriorate poor prognosis of advanced HCC patients. In decades, several therapeutic strategies have been developed to against advanced unresectable HCC. The oxaliplatin-based chemotherapy has revealed encouraging therapeutic e cacy [2,3]. However, HCC usually variously responses to chemotherapeutic agents because of different sensitivity of cancer cells, which limits the utilization of oxaliplatin-based chemotherapy. Therefore, identi cation of sensitivity regulators in HCC cells is urgent to optimize current chemotherapeutic strategies. In addition, the low sensitivity of HCC cells is usually accompanied with aggressive and proliferative phenotypes which also contribute to the poor prognosis of advanced HCC patients [4]. Therefore, nding a therapeutically targetable central regulator which controls multiple process in HCC progression should be seriously considered.
Oxaliplatin is the third generation platinum drug which mainly targets DNA, generates DNA adducts, causes inter-and intra-strand crosslinking (ICLs) and ends up with DNA damage [5]. In this process, unwound, bended and destabilized DNA duplex structure which can recruit DNA repair-associated proteins and activate DNA damage repair pathway. Indeed, activity of DNA damage repair pathways are closely associated with chemosensitivity of HCC cells. Thus, DNA repair regulators have been extensively investigated to dissect mechanism of sensitivity modulation and resistance formation. Among these regulators, HMGB1, a highly abundant chromatin-associated non-histone protein, has received improving attention. HMGB1 was identi ed to regulate multiple DNA damage repair pathways and its expression was negatively correlated to the cellular sensitivity to DNA-damaging agents [6]. Specially, previous studies have demonstrated HMGB1 could promote processing of ICLs via multiple repair pathways [7][8][9].
De ciency of HMGB1 revealed increased genomic instability, DNA damage and cell death [8, 10,11]. In addition to regulating DNA damage repair, a series of studies found that HMGB1 could modulate drug sensitivity via regulating autophagy [12][13][14]. Besides, HMGB1 also played a crucial role in proliferation, metastasis in various types of human cancer [15][16][17]. Despite of deep investigation in biological functions of HMGB1, only a few studies were related to its own regulation mechanism. In macrophage, activation of CB2R promoted the degradation of intracellular HMGB1 via the autophagy-lysosome pathway [18]. In HCC cells, previous work has reported that HMGB1 was degraded via chaperone mediated autophagy pathway [19]. However so far, upstream regulator of HMGB1 protein remains unknown.
Hematological and neurological expressed (HN1), locating on chromosome 17q25.2, encodes a 16.5 kilodalton protein which is ubiquitously expressed and highly conserved [20]. Increasing evidences indicated that HN1 showed an elevated expression level and was associated with poor prognosis in many tumors [21][22][23]. HN1 has been proposed to act as a biomarker for diagnosis and prognosis of HCC [24]. In addition, previous work have also highlighted the associaltion between HN1 and DNA repair through bioinformatics analysis [24]. These observations intrigued us to further interrogate the multiple functions of HN1 in HCC, especially as a regulator of platinum drug sensitivity.

Patients and cell lines
We followed up 43

Immuno uorescence
Cells were seeded on glass coverslips to con uence, xed in methylalcohol permeabilized with 0.1% Triton X-100 diluted in PBS for 5 minutes and blocked by 10% goat serum (Sigma-Aldrich). Then cells were incubated with needed primary antibodies at 4°C overnight and then secondary antibodies being in tune with the manufacturer's introduction. The cells were stained by DAPI to highlight nucleus for 15min at room temperature and nally imaged with Fluorescence confocal microscope.
Immunohistochemistry HCC tissues of patients and nude mice were xed in 4% paraformaldehyde for at least 48 hours, and were dehydrated and embed in para n. Tissues were cut into sections for immunohistochemistry staining.
After depara nization with xylene, sections were subjected to antigen repair twice for 10 minutes at 100°C in citrate buffer (pH = 6.0). Tissue section were blocked by 10% goat serum for 20 minutes and incubated with diluted primary antibody overnight at 4°C. After being washed by phosphate buffer saline

Flow cytometry
The degree of apoptosis was assessed by PE Annexin V Apoptosis Detection Kit I (BD Pharmingen) followed by ow cytometric analysis. The cells were harvested and washed three times with pre-cooling PBS. Within 100µL cell suspension at the density of 1×10 6 cells/mL resuspended in 1× binding buffer, 5µL 7AAD and 5µL PE Annexin V were added and incubated for 15 minutes kept out of light. Then 400µL 1× binding buffer was added to terminate the reaction and analysis was conducted by FACS.

Cell proliferation and migration/invasion assays
Colony formation assay and MTT assay were performed to detect cell proliferation potency. Cells transfected with siRNA or plasmid were seeded onto 6-well plates at a density of 500 cells/well. After two weeks, when a certain well appeared colonies containing more than 50 cells, the colonies were xed with 5% methanol, stained with 0.1% crystal violet and nally counted. MTT assay is the assay that 1000 cells which were seeded on 96-well plates per well and cultured for 24,48,72,96 hours respectively, then treated with 0.5mg/mL MTT, incubated for 4 hours at 37°C, added 150µl DMSO for 10min and nally followed by measured at 560nm absorbance using the EnSpire Multimode Plate Reader (PerkinElmer). Transwell assay was conducted to assess cell migration and invasion. To measure cell invasion, 5×10 4 cells resuspended in serum-free medium were seeded into the upper chamber of 8µm pore size insert (Costar, Corning, Cambridge, MA, USA) pre-coated with Matrigel (BD Biosciences) and DMEM medium with 20% FBS was added to the lower chamber. For migration, the same procedures were carried out except for the pro-treatment of Matrigel. After incubated at 37°C for 24 hours, cells on the bottom surface of chamber were xed, stained and counted as the standard protocol.
Cell viability and drug sensitivity 2×10 3 cells/well cells were cultured on 6cm plates, added into drug and incubated for 24 hours as required at 37°C. The CCK8 kit (Dojindo Laboratories) was employed to measure cell viability and re ect drug sensitivity according to manufacturer's protocol. The absorbance at 450nm was measured.

Co-immunoprecipitation analysis
Cells were lysed in NP40 buffer containing the protease inhibitor. After added into the speci c antibodies and incubated at 4°C for 4 hours, cell lysis and activated protein A/G Plus-agarose beads (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) were co-incubated at 4°C overnight. The agarose beads were washed using RIPA for three times, added into SDS-PAGE loading buffer and immune complexes were subjected to subsequent western blot analysis.

Alkaline Comet assay
Alkaline comet assay is a sensitive genotoxicity tool typically for identi cation of DNA lesions. Cells were treated by oxaliplatin before embedded into 0.5% low melting point agarose gel in the single-cell suspension on clean slides precoated by 1.5% normal melting agarose. Then, cells protected from light undergone lysis for 1 hour at 4°C, electrophoresis (0.8v/cm, 20 minutes) and neutralization (0.5M Tris-HCl, pH 7.5). Finally, cells were stained with GoldView and subjected to acquire images by uorescence microscope. The cells with positive-tail were counted to measure DNA lesions.

Transmission electron microscope (TEM)
Cells were xed in 2.5% glutaraldehyde for 5 minutes and collected. Then cell mass was placed in fresh 2.5% glutaraldehyde for at least 2 hours. After treated by 1% osmic acid for 1 hour, rinsed once again and then dehydrated with gradient concentration of acetone, cells were nally made out into ultrathin sections. The samples were observed under an H-7650 transmission electron microscope (Hitachi, Tokyo, Japan).

Tumor xenografts
The laboratory animals used in this study were four-week-old male BALB/c nude mice raised in the Laboratory Animal Center (Xi'an Jiaotong University, China). The nude mice were randomly divided into 2 groups (n = 7 per group) and subcutaneously injected into inguina with MHCC-97L cells (5×10 6 in 200µL PBS) stably transfected with shHN1 or shNC. The sizes of tumor were measured every 3 days and all nude mice were sacri ced after 4 weeks when none of diameter of tumor exceeded 15mm. In lung metastasis model, above-mentioned cells (1× 10 6 in 100µL PBS) were injected into the lateral tail vein of nude mice (n = 3 per group). Eight weeks later, the mice were sacri ced and lungs were stripped out. For testing the effect of HN1 on drug sensibility, the mice were divided randomly into two groups (n = 8 per group) and subcutaneously injected with MHCC-97L cells (1×10 7 in 200µL PBS) stably transfected with shHN1 and shNC respectively. When volume of some tumors roughly reached 100mm 3 (V = length×width 2 ×0.52), every group were randomly divided into two subgroups. The mice in experimental group were infected intraperitoneally with oxaliplatin (10mg/kg) and the control groups with the same volume of normal saline (NS) every 3 days. After 24 days, the mice were sacri ced and stripped out of tumors. All animal experiments were approved by Animal Ethics Committee of Xi'an Jiaotong University.

Statistics
Data was statistically analyzed and graphed by GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA). 2-tailed student's test was conducted for two groups and one-way ANOVA for multicomponent groups. Overall survival (OS) and progression free survival (PFS) were calculated by the Kaplain-Meier method and log-rank test on the basis of offered survival time. p < 0.05 was supposed as statistically signi cant.

Results
HN1 was highly expressed in HCC and associated with poor prognosis of HCC patients.
First, the protein level of HN1 was analyzed by IHC in 43 HCC tissues and para-carcinoma tissues, and the protein level of Ki67 was also analyzed in HCC tissues to assess cell proliferation of malignant cells (Fig.  1a). We found that HN1 protein level was higher and positively correlated with Ki67 protein level in HCC tissues with Pearson's correlation of 0.6498 (Fig. 1b). The overexpression of HN1 in HCC tissues compared to para-carcinoma tissues was veri ed by western blot (Fig. 1c). Moreover, the protein and mRNA expression level of HN1 in human HCC cell lines were remarkably higher than that in the normal immortalized hepatic cell line LO2 (Fig. 1d, e). Notedly, Hep3B and Huh7, which have been proved to be sensitive to oxaliplatin [25], had lower level of HN1 than other cell lines in our study. The clinical data indicated that aberrant expression of HN1 was closely associated with the tumor size and the incidence of portal vein tumor thrombus (PVTT) of HCC (see additional le 1). Further analysis of the overall survival and progression free survival of the patients showed that patients with higher expression of HN1 had worse prognosis than those who had lower expression of HN1 (Fig. 1f, g). Overall, above data implied that HN1 was overexpressed in HCC tissues and closely associated with poor prognosis of patients.
HN1 knockdown inhibited HCC proliferation and metastasis in vitro.
To investigate the tumor-promoting mechanism of HN1 in HCC, two different siRNAs (siRNA#1 and siRNA#2) were transfected into HepG2 and MHCC-97L cell lines. The cells transfected with siHN1#2 had a lower HN1 mRNA expression and protein level than siHN1#1 (Fig. 2a, b). The MTT assay showed that downregulation of HN1 signi cantly inhibited HCC cell growth (Fig. 2c), and HN1 knockdown inhibited clonogenicity of HCC cells (Fig. 2d). In addition, c-Myc and cyclin D1 which re ect potency of cell proliferation also downregulated in HN1 knockdown cells (Fig. 2f). However, downregulation of HN1 had no effect on cell apoptosis (see additional le 2).
Given that higher expression of HN1 was related with PVTT patients, HN1 possibly promoted metastasis of HCC. As expected, the result of transwell assay showed that downregulation of HN1 decreased the migration and invasion of HCC cells (Fig. 2e). In addition, N-cadherin, E-cadherin and ZEB1 which re ect metastasis potency of HCC cells also altered following HN1 knockdown (Fig. 2g). Next, we overexpressed HN1 in Hep3B and Huh7 cell lines through pCMV-HN1 plasmid, and measured their potency of cell proliferation, migration and invasion. We found that these malignant phenotypes of HCC cells were consistently reinforced after HN1 overexpression (see additional le 3). In conclusion, gain and loss of function experiments indicated that HN1 could promote proliferation, migration and invasion of HCC cells.
HN1 knockdown increased sensitivity to oxaliplatin and aggravated DNA damage in HCC cells.
We established HN1 knockdown cells lines of HCC (HepG2 and MHCC-97L) through shRNA lentivirus vector and knockdown e ciency was con rmed by qRT-PCR and western blot (Fig. 3a, b). Taking the observation into account that patients with higher HN1 expression had worse prognosis after oxaliplatinbased chemotherapy treatment, we explored whether HN1 desensitized HCC to oxaliplatin. The CCK8 assay demonstrated that HepG2 and MHCC-97L cells with HN1 knockdown were more sensitive to oxaliplatin and had lower IC 50 compared to those with negative control HCC cells (Fig. 3c). Alkaline comet assay suggested that HCC cells with HN1 knockdown had higher tail-positive proportion compared with negative control group, which indicated that there were more DNA lesions in HN1 knockdown HCC cells after oxaliplatin treatment (Fig. 3d). In addition, HN1 knockdown also reduced the survival of HCC cells in low dose of oxaliplatin (1µM and 5µM, Fig. 3e). Next, we treated HCC cells with gradient concentration of oxaliplatin (1µM, 5µm and 10µM) for 1 hour or 24 hours (Fig. 3f, g). We measured the histone modi cation related to DNA damage (γ-H2AX) and chromatin accessibility (acetylation of H3 and H4) [26-28]. We found that after oxaliplatin treatment, γ-H2AX level was higher in HN1 knockdown HCC cells while acetylation of H3 and H4 were higher in negative control HCC cells. Then time-dependence assay (cells treated with 5µM oxaliplatin for 0h, 6h, 12h and 24h) were performed to verify that HN1 could affect DNA damage repair and chromatin modi cation (Fig. 3h). We also con rmed that immuno uorescence intensity of γ-H2AX foci is higher in HN1 knockdown HCC cells after oxaliplatin treatment (5µM and 10µM, Fig. 3i). Finally, ow cytometry analysis demonstrated that HN1 knockdown HCC cells had a signi cantly higher apoptosis rate compared to negative control cells after oxaliplatin treatment (5µM and 10µM, Fig. 3j). All these results indicated that HN1 knockdown substantially sensitized HCC cells to oxaliplatin treatment.
HN1 inhibited the degradation of HMGB1 through autophagy-lysosome pathway by interacting with TRIM28.
Given that the depletion of HN1 in HCC cells enhanced sensitivity to oxaliplatin-induced DNA damage, we hypothesized that deregulation of proteins in DNA repair pathway could be crucial to this sensitivity transformation. HMGB1, which can bind to damaged DNA, recruit other components and enhance DNA repair, was tested as a potential downstream effector of HN1. The result of qRT-PCR showed that mRNA expression level of HMGB1 was impervious followed by HN1 knockdown (Fig. 4a). The protein expression level of intracellular HMGB1, however, was obviously decreased both in nucleus and cytoplasm (Fig. 4b,   c). This result was veri ed by immuno uorescence staining (Fig. 4d). These results indicated that HN1 could regulate the expression of HMGB1 in post-transcriptional level and affect its degradation. To verify this speculation, after cells were transfected into pCMV-HMGB1 expression plasmid for 72 hours, the protein synthesis inhibitor, cycloheximide (CHX) was added at a concentration of 2µg/mL for another 24 hours, 48 hours, 72 hours, and 96 hours respectively. The result of western blot showed that degradation of HMGB1 is prominently faster in HN1 knockdown cells than negative control cells (Fig. 4e). Next, we aimed to investigate which pathway was responsible for HN1-related HMGB1 degradation. The autophagy inhibitor, chloroquine (10µM) or proteasome inhibitor, MG132 (1µM) was individually added into cells for 24 hours in the presence of CHX (2µg/mL). The result of western blot showed that autophagy inhibitor chloroquine could slower evidently the degradation rate of HMGB1 in HN1 knockdown cells (Fig. 4f). Ubiquitination is a common post-translational modi cation for substrate protein ready to be degraded [29]. Thus, we detected the endogenous ubiquitination of HMGB1. The result of IP showed that ubiquitination of HMGB1 was substantially increased followed by HN1 knockdown (Fig. 4g). These results indicated that HN1 prevent HMGB1 from autophagy-lysosome-depend degradation.
We next investigate how HN1 regulated the ubiquitination of HMGB1. Previous studies reported that TRIM28 could be essential to the ubiquitination and degradation of HMGB1 [30]. The result of coimmunoprecipitation assay revealed that HN1 interacted with TRIM28 (Fig. 4h), which is also supported by STRING database (https://string-db.org). In addition, from analysis of Immuno uorescence colocalization assay, we found the high degree of co-localization of HN1 and TRIM28 protein (Fig. 4i). Our pre-test indicated that TRIM28 knockdown could slightly affect the protein level of HMGB1 in the normal or HN1 knockdown cells (see additional le 4).
It has been well-established that MAGE proteins can bind to RING domain of E3 ligase like TRIM28 and form cancer-speci c ubiquitin ligase complexes which regulate ubiquitination and degradation of various tumor suppressors like p53 [30][31][32]. It's worth noting that MAGE-C2 and MAGE-A3/6 was specially upregulated in HCC [33][34][35]. Therefore, we hypothesized that MAGE-C2 or (and) MAGE-A3/6 could be involved in regulation of HMGB1 ubiquitination through TRIM28. We then knockdown mRNA expression of TRIM28, MAGE-A3/6 or MAGE-C2 with small interfering RNAs (siRNAs), and detected the protein level of HMGB1. We found that knockdown of TRIM28 and MAGE-A3/6 signi cantly restored the protein level of HMGB1, while knockdown of MAGE-C2 had no effect on HMGB1 protein level (Fig. 4j). As expected, knockdown of TRIM28 or MAGE-A3/6 could inhibit the ubiquitination of HMGB1 and restored its protein level in HCC cells with HN1 downregulation (Fig. 4k). Based on these observations, we drawn the conclusion that HN1 could inhibit the degradation of HMGB1 through autophagy-lysosome pathway via interacting with TRIM28-MAGE-A3/6 complexes.
Overexpression of HMGB1 restored the phenotypes resulted from HN1 knockdown in HCC cells.
To further investigate the essential roles of HMGB1 as a key mediator of HN1 in phenotypic regulation of HCC cells, we transfected pCMV-HMGB1 overexpression plasmid into HN1 knockdown cell lines and assessed their phenotype restoration. We con rmed the transfection e ciency of plasmid through qRT-PCR and western blot analysis (Fig. 5a, e). The results of MTT assay and clone formation assay showed that the potency of cell growth could be partly rescued by overexpression of HMGB1 in HN1 knockdown cells and so was potency of migration and invasion tested by Transwell assay (Fig. 5b-d). In addition, results of western blot showed that HMGB1 overexpression could partly reverse the protein expression level of the biomarkers of cell proliferation and metastasis in HepG2 and MHCC-97L cells (Fig. 5e). Finally, we observed that overexpression of HMGB1 could abolish the increased drug sensitivity induced by HN1 knockdown in HCC cells (Fig. 5f). The results of western blot revealed that DNA damage alleviated and chromatin accessibility increased followed by HMGB1 overexpression under oxaliplatin treatment (Fig. 5g).
Moreover, increasing evidences indicated that autophagy plays an essential role in regulation of malignant phenotypes of tumor and cancer resistance to chemotherapy agents. Given that HMGB1 is an acknowledged autophagy activator [36], we conjectured that HN1 could regulate autophagy via HMGB1 and further affect phenotypes of HCC cells. In contrast to shNC cells, the recognized autophagy stimuli HBSS employed in shHN1 cells failed to upregulate the expression of LC3-, which indicated that HN1 was essential to cell autophagy (Fig. 5h). To investigate the role of HMGB1 in this process, we transfected pCMV-HMGB1 overexpression plasmid into shNC and shHN1 HCC cells and results of western blot indicated that the overexpression of HMGB1 boosted the protein level of LC3-and lowered the level of p62 in shHN1 cells (Fig. 5i). In addition, cells were transfected by exogenous GFP-mRFP-LC3 adenovirus to monitor and measure cell autophagy. We found that the formation of GFP-mRFP-LC3 puncta markedly increased in shHN1 cells with HMGB1 overexpression compared with single shHN1 cells (Fig. 5j). Finally, transmission electron microscope analysis provided potent testimony of autophagy. The overexpression of HMGB1 in shHN1 cells exhibited signi cantly more autophagosomes compared with single HN1 knockdown cells (Fig. 5k). Above evidences supported a pivotal role of HMGB1 in process of autophagy regulated by HN1.
The above results veri ed that HMGB1 acted as a key downstream effector of HN1 in regulation of malignant phenotypes of HCC cells such like cell proliferation, metastasis and chemosensitivity.
Page 11/23 HN1 knockdown inhibited malignant phenotypes and enhanced the antitumor effect of oxaliplatin in vivo.
MHCC-97L cells stably transfected with shNC or shHN1 were subcutaneously injected into male nude mice. We found that knockdown of HN1 dramatically inhibited tumor growth compared with negative control group (Fig. 6a-c). From the result of IHC, we also found the protein level of Ki67 and HMGB1 were also decreased after HN1 knockdown (Fig. 6d). Additionally, the role of HN1 on HCC tumor metastasis was evaluated through caudal intravenous pulmonary metastasis model. The result of HE staining showed that HN1 knockdown reduced the number and size of visible pulmonary metastatic nodules and the Ki67 protein level in metastatic niche of lung (Fig. 6e, f). Brie y, HN1 knockdown inhibited tumor proliferation and metastasis in vivo. Furthermore, to investigate the effect of HN1 knockdown on HCC sensitivity to oxaliplatin treatment in vivo, the nude mice were treated with oxaliplatin or normal saline (NS) when some of tumors derived from HN1 knockdown or negative control MHCC-97l cells grew roughly to 100mm 3 . The growth curve, image and weight of tumor demonstrated that HN1 knockdown group was more sensitive to oxaliplatin treatment than the control group ( Fig. 6g-i).
Discussions HN1 reportedly promoted proliferation and metastasis of malignant cells in many human cancers. Recent research revealed that HN1 could increase STMN1 protein level by inhibiting its ubiquitination-mediated degradation to promote aggressiveness of anaplastic thyroid carcinoma [21]. In addition, previous work demonstrated that ectopic HN1 expression could increase degradation of β-catenin and resulted in disassociation of β-catenin and E-cadherin, which promoted progression of prostate cancer [22]. In our study, HMGB1 acted as an essential downstream target of HN1 to affect HCC cells malignant phenotypes. Similar to above-mentioned recognized mechanism, HN1 protected HMGB1 from the ubiquitination and subsequent degradation. Also, Chen Zhang and colleagues proved that HN1 could promote migration and invasion of breast cancer by enhancing the protein level of c-Myc and its target genes [23]. Consistently, we also found that HN1 could inhibit ubiquitination of endogenous c-Myc (see additional le 5). Base on above ndings, we could reach a conclusion that HN1 possibly play an extensively regulatory role at the post-translation level. Our data further indicated that TRIM28 acted as a crucial mediator in this process. However, whether TRIM28 is a common or speci c component to mediate oncoprotein degradation in context of HCC need more research.
Intranuclear HMGB1 mainly functions as DNA chaperone and can also be released into cytoplasm or extracellular context in response to diverse stimulus (eg, chemoradiotherapy and metabolic stress). Whether extracellular or intracellular HMGB1 has been proved to function as a crucial regulator of autophagy [36][37][38][39]. Here our study focused on intracellular HMGB1. The overwhelming evidences support that autophagy is important for cancer cells to survive from treatment of DNA damaging agents and ionizing radiation [40]. Indeed, we found that downregulation of HMGB1 followed by HN1 knockdown resulted in the suppression of autophagy, which was parallel with the enhancement of drug sensitivity. Another interesting bit of nding is that the regulatory mechanism of HN1-mediated HMGB1 degradation was dependent on autophagy-lysosome pathway. On the one hand, HN1 could inhibit the degradation of HMGB1 via the autophagy-lysosome pathway through interacting with TRIM28. On the other hand, HMGB1 in turn acted as the pivotal mediator of HN1 to promote autophagy in HCC cells. The mainstream view believes that unneeded cellular components can conduct selective autophagy to undergo precisely regulated and speci c degradation pathway [41]. Selective autophagic receptors and cargo labeled by ubiquitin are indispensable for this process [41]. Our nding indicated that the degradation of speci c molecule via autophagy-lysosome pathway was much more complex, subtle and context dependent.
TRIM28, a member of transcriptional intermediary factor 1 (TIF1) family, is a highly complex chromatinbinding protein with multiple domains structure and enzymatic activities [42]. Increasing ndings suggested that TRIM28 could function to affect genome stability by involvement in the chromatin relaxation, chromatin-context response to DNA damage and cell cycle regulation [43]. Studies found that TRIM28 was signi cantly upregulated across various cancers, such as glioma [44], ovarian [45], breast [46] and gastric cancer [47] and was associated with the dismal overall survival of these cancer patients. However, TRIM28 played dual roles of tumor promotor or suppressor in context of HCC where it was highly expressed compared to normal counterparts and predicted the worse prognosis, whereas it acted as a tumor suppressor to prevent tumorigenic conversion in murine HCC model [48,49].In our present research, ubiquitination of HMGB1 decreased and HMGB1 protein level restored prominently after TRIM28 knockdown in the HN1 knockdown cells. In addition, we found TRIM28 could interact with HN1. Based on these observations, we concluded that highly expressed HN1 probably sheltered the E3 ligase activity of TRIM28 through interacting with TRIM28 and inhibited degradation of HMGB1. MAGE proteins have been veri ed as the crucial function facilitator of E3 ligase activity of TRIM28 [31]. Conversely, our ndings indicated that HN1 acted as the function suppressor of TRIM28 in HCC cells. Maybe HN1 and MAGE are like two ends of a seesaw of TRIM28 function, and TRIM28 itself has no preference for being a promoter or a suppressor of cancer progression unless it associates with some certain speci c proteins like MAGE proteins or HN1. Given that TRIM28 serves as a large scaffold protein, whether there are other speci c proteins that can coordinate biological activity of TRIM28 need further exploration.

Conclusions
In conclusion, we found that HN1 expression promoted malignant phenotypes of HCC cells and     expressed stably for 2 days, cycloheximide (CHX) (2μg/mL) was added for incubation 24h, 48h, 72h, 96h respectively. The degradation rate of HMGB1 was measured by western blot. *p<0.05, **p<0.01. (f) The pCMV-HMGB1 expression plasmid was transfected into shNC and shHN1 cells. The protein level of HMGB1 was measured by western blot after chloroquine (10μM) or MG132 (1μM) with CHX (2μg/mL) were employed for 24h as indicated. (g) Ubiquitinated endogenous HMGB1 was detected by IP assay. (h) The co-IP analysis showed that HN1 interacted with TRIM28. (i) The immuno uorescence co-localization of HN1 and TRIM28 was illustrated in the merged images. Scatter plot indicated a high degree of colocalization. Rcoloc means Pearson's coe cient for image above thresholds. (j) The protein level of HMGB1 in HCC cells with TRIM28 or MAGE-C2 or MAGE-A3/6 knockdown was measured by western blot.
*p<0.05. (k) HMGB1 protein level and its endogenous ubiquitination level in HN1 knockdown HCC cells with knockdown of TRIM28 or MAGE-A3/6 was measured by western blot. **p<0.01. tested by western blot after pCMV-HMGB1 expression plasmid was transfected into shNC and shHN1 cells and cells were exposed to 0.1μM of ba lomycin A1 for 6 hours before total protein was collected. (j) pCMV-HMGB1 expression plasmid was transfected into shNC and shHN1 cells as indicated and then GFP-mRFP-LC3 adenovirus infected the cells. Ba lomycin A1 (0.1μM for 6h) was added before cells were xed. Representative images of LC3 puncta were acquired from confocal microscopy. The GFP-mRFP-LC3 puncta meaning the formation of autophagosome was counted and analyzed. *p< 0.05, **p <0.01. (k) Typical ultrastructural images of shNC or shHN1 cells transfected with pCMV-HMGB1 expression plasmid after treatment of ba lomycin A1 (0.1μM for 6h). More autophagosomes were discovered in shHN1 cells with HMGB1 overexpression compared with single HN1 knockdown cells.