In Situ Monitoring of Thermal Degradation of CH3NH3PbI3 Films by

Dec 27, 2018 - (14) The power conversion efficiency increased dramatically from 3.8(14) to 10.9%(15) in just three years and reached 23.3% recently, w...
0 downloads 0 Views 2MB Size
Journal of Pathology J Pathol 2018; 246: 508–518 Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/path.5164

ORIGINAL PAPER

HMGA2 promotes intestinal tumorigenesis by facilitating MDM2-mediated ubiquitination and degradation of p53 Yuhong Wang1,2† , Lin Hu3† , Jian Wang4 , Xiangwei Li1,2 , Sana Sahengbieke1,2 , Jingjing Wu1,2* and Maode Lai1,2* 1 2 3 4

Department of Pathology, Zhejiang University School of Medicine, Hangzhou, Zhejiang, PR China Key Laboratory of Disease Proteomics of Zhejiang Province, Hangzhou, Zhejiang, China Jiangsu Key Laboratory of Infection and Immunity, Institutes of Biology and Medical Sciences, Soochow University, Suzhou, Jiangsu, PR China Department of Surgical Oncology, Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, Zhejiang, PR China

*Correspondence to: J Wu or M Lai, Department of Pathology, Zhejiang University School of Medicine, 866 Yuhangtang Road, Hangzhou, Zhejiang 310058, PR China. E-mail: [email protected] or [email protected]

Co-first authors.

Abstract High mobility group A2 (HMGA2) is an architectural transcription factor that promotes human colorectal cancer (CRC) aggressiveness by modulating the transcription of target genes. The degradation of p53 is mediated by murine double minute 2 (MDM2) in a proteasome-dependent manner. Here we report that HMGA2 promotes cell cycle progression and inhibits apoptosis in CRC cells in vitro. We also developed an intestinal epithelial cell-specific Hmga2 knock-in (KI) mouse model. It revealed that the Hmga2 KI promoted chemical carcinogen-induced tumorigenesis in the intestine in vivo. In studying the underlying molecular mechanism, we found that HMGA2 formed a protein complex with p53. The tetramerization domain of p53 (amino acids 294–393) and the three AT-hook domains (amino acids 1–83) of HMGA2 were responsible for their direct interaction. We also found that HMGA2 directly bound to MDM2 and the central acidic and zinc finger domains of MDM2 (amino acids 111–360) were required for interaction with HMGA2. Furthermore, our results indicated that HMGA2 promoted MDM2-mediated p53 ubiquitination and degradation. Interestingly, Hmga2 overexpression in Hmga2 KI mice resulted in an increase in the accumulation of ubiquitinated p53. In addition, in two large CRC cohorts, it was demonstrated that high HMGA2 expression was predictive of an adverse outcome in the p53-negative subgroup of CRC patients. In summary, our data have established for the first time a novel mechanism by which HMGA2 functions with p53 and MDM2 to promote CRC progression. Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Keywords: colorectal cancer; high mobility group AT-hook 2; p53 Received 13 April 2018; Revised 30 July 2018; Accepted 28 August 2018

No potential conflicts of interest were declared.

Introduction Colorectal cancer (CRC) is one of the most common cancers. It remains a major cause of cancer-related deaths worldwide [1]. CRC carcinogenesis is a multistep process that is characterized by oncogene activation and tumor suppressor gene inactivation [2]. A better understanding of the somatic genetics of CRC is a prerequisite for a more accurate diagnosis and a more effective treatment. High mobility group A2 (HMGA2), as an oncoprotein, plays a very important role in embryogenesis and tumorigenesis [3–8]. It is highly expressed in different cancers, whereas it is almost absent in terminally differentiated tissues [9,10]. High levels of HMGA2 mRNA and protein correlate with poor outcomes [11,12]. In our previous work, we found that the oncogenic properties of HMGA2 were mediated by directly activating the Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. www.pathsoc.org

transcription of IL11 and FN1 and promoting CRC metastasis and epithelial–mesenchymal transition via a pSTAT3-dependent pathway [13]. Consistent with this, Wang et al [11] stated that high-level expression of HMGA2 predicted an adverse outcome in CRC from two large cohort studies. These findings suggest that HMGA2 impacts CRC progression by regulating the expression of target genes. However, it is still not clear whether HMGA2 requires other protein-interaction partners to exert its pro-tumor effect. As the most prominent tumor suppressor, p53 exerts its function by regulating the transcription of downstream genes that play important roles in diverse biological processes, including cell cycle, apoptosis, DNA damage response, and differentiation [14,15]. Under physiological conditions, p53 is maintained at a very low level by ubiquitination [16,17]. Murine double minute 2 (MDM2), an E3 ubiquitin ligase, interacts physically with p53 and promotes its ubiquitination J Pathol 2018; 246: 508–518 www.thejournalofpathology.com

HMGA2 facilitates MDM2-mediated ubiquitination of p53

to elicit proteasome-dependent degradation. Here, we verified a link between the HMGA2 protein and the MDM2-p53 signaling axis. In the present study, we defined the association between HMGA2, MDM2, and p53 in CRC. We observed that HMGA2 decreased the stability of the p53 protein and promoted MDM2-mediated p53 ubiquitination and degradation. As expected, findings from the conditional knock-in (KI) mouse model and clinical sample analyses supported that HMGA2 exerted its effects on CRC carcinogenesis through the p53 network.

Materials and methods Patients and tissue samples Two independent CRC cohorts were used in this retrospective study. The training cohort consisted of 121 cases diagnosed at the Sir Run Run Shaw Hospital of Zhejiang University between 2007 and 2009. The validation cohort of 90 cases was obtained from the National Engineering Center for Biochip at Shanghai between 2001 and 2008. The clinical characteristics of all of the patients in both of the cohorts are described in supplementary material, Table S1. The study was approved by the Ethical Committee at Zhejiang University.

Antibodies The primary antibodies used for western blot, co-immunoprecipitation (Co-IP), glutathione S-transferase (GST) pull-down, and ubiquitination assays were HMGA2 (59170AP, 1:1000 dilution, Biocheck, South San Francisco, CA, USA), p53 (ab179477, 1:1000 dilution, Abcam, Cambridge, MA, USA), p53 (mouse, ab31333, 1:1000 dilution, Abcam), p21 (ab109199, 1:1000 dilution, Abcam), MDM2 (ab178938, 1:2000 dilution, Abcam), β-actin (4970, 1:1000 dilution, Cell Signaling, Danvers, MA, USA), FLAG (F1804, 1:2000 dilution, Sigma), Myc (sc-40, 1:1000 dilution, Santa Cruz, Dallas, TX, USA), GST (M20007, 1:2000 dilution, Abmart, Arlington, MA, USA), HA (M20003, 1:5000 dilution, Abmart), His (M20001, 1:5000 dilution, Abmart), and Ub (sc-8017, 1:1000 dilution, Santa Cruz).

Generation of conditional Hmga2 KI mice and genotyping The experimental procedures were approved by the Review Committee of the Zhejiang University School of Medicine. Conditional Hmga2 KI mice were generated using a Cre-conditional ROSA26 KI strategy. Details are provided in the supplementary material, Supplementary materials and methods.

Co-immunoprecipitation (Co-IP) The cells were transfected with the indicated plasmids for 48 h and were lysed in NP-40 lysis buffer. Then, cell Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. www.pathsoc.org

509

lysates (1500 μg) were incubated with the indicated antibody or IgG control (1 μg) and with protein A/G beads (20 μl) overnight at 4 ∘ C. The precipitates were washed, separated by SDS-PAGE, transferred to a polyvinylidenedifluoride (PVDF) membrane and analyzed using the indicated antibodies by western blotting. All of the experiments were repeated at least three times.

Statistical analyses The data were presented as the mean ± SD and were analyzed by Student’s t-test or one-way ANOVA. The overall survival of patients was analyzed (various subset analyses) by univariate Cox regression analyses. SPSS 17.0 software (SPSS Inc, Chicago, IL, USA) was used for the statistical analyses. p < 0.05 was considered to be significant. Supplementary material, Supplementary materials and methods presents detailed information for cell culture and lentiviral infection, plasmid construction and transfection, siRNA transient transfection, cell cycle and apoptosis assays, RNA isolation and RT-qPCR analyses, western blotting analyses, GST pull-down assay, ubiquitination assays, and immunohistochemical (IHC) analyses.

Results HMGA2 promotes cell cycle progression and inhibited apoptosis in vitro To explore the roles of HMGA2 in cell cycle and apoptosis, we performed flow cytometry analysis in both the HMGA2 gain-of-function and loss-of-function cells. As previously described, we overexpressed HMGA2 in LoVo cells (with low endogenous levels of HMGA2), whereas we knocked it down in HCT116 and RKO cells (with high endogenous levels) [13]. Cell cycle assays showed that overexpression of HMGA2 significantly increased the proportion of cells in the S-phase in LoVo cells, whereas knockdown of HMGA2 inhibited cells’ entry into the S-phase for both HCT116 and RKO cells (see supplementary material, Figure S1A). As expected, similar results were obtained from the cell apoptosis assay; a higher proportion of cells was found to be apoptotic in HMGA2-knockdown cells, whereas the opposite effects were observed upon HMGA2 transfection in LoVo cells (see supplementary material, Figure S1B). These results showed that HMGA2 influenced the cell cycle and apoptosis in CRC cells in vitro.

Hmga2 overexpression accelerates chemical carcinogen-induced tumorigenesis in Hmga2KI/KI :PVillin-Cre T mice in vivo To investigate the in vivo functions of Hmga2, we used the Cre-conditional ROSA26 KI system to insert wild-type (WT) Hmga2 into the ROSA26 gene locus downstream of a loxP-flanked transcriptional stop J Pathol 2018; 246: 508–518 www.thejournalofpathology.com

510

cassette (Figure 1A). To remove the stop cassette, mice were crossed with PVillin-Cre transgenic mice to generate intestinal epithelial cell-specific KI mice (Hmga2KI/KI :PVillin-Cre T). We then evaluated Hmga2 expression in WT and KI mice by western blotting and IHC analysis. As shown in Figures 1D,E, and 5D, the Hmga2 protein was almost undetectable in the intestine tissues of WT mice, whereas it was markedly increased in the Hmga2KI/KI :PVillin-Cre T intestines. As no spontaneous tumors were observed in WT and KI mice we used the azoxymethane (AOM)- or AOM/dextran sodium sulfate (DSS)-induced mouse cancer models in WT and KI mice. AOM is a chemical carcinogen that induces alkylation of DNA and facilitates base mispairings. The AOM model is commonly utilized to study the pathogenesis of CRC [18]. DSS is a pro-inflammatory agent that exerts direct toxic effects on the mouse colonic epithelium, thereby inducing severe colitis. The combination of DSS with AOM is a well-established model for colitis-associated CRC studies [18]. For the AOM-induced model, mice were treated with AOM for six intraperitoneal injections of 10 mg/kg body weight until we evaluated tumor development at week 30 (Figure 1B). For the AOM/DSS-induced model, mice were given a single intraperitoneal injection of AOM and three cycles of 2.5% DSS in drinking water for 7 days, then drinking water for the subsequent 14 days. The development of tumors was monitored at week 10 after the initial treatment (Figure 1C). We found that WT and KI mice developed adenomas in both small and large intestines. Notably, we found that KI mice developed more and larger adenomas than WT mice in both the AOM-induced (p < 0.05, Figure 1D,F) and AOM/DSS-induced models (p < 0.05, Figure 1E,G). The tumors were confirmed by hematoxylin–eosin staining (Figure 1D,E). IHC analysis showed a remarkable upregulation of Hmga2 expression in tumors from KI mice compared with those from WT mice (Figure 1D,E). Collectively, the in vivo data suggested that the KI of Hmga2 accelerated intestinal carcinogenesis.

HMGA2 interacts directly with p53 in vivo and in vitro To demonstrate the protein–protein interaction between HMGA2 and p53, we first performed Co-IP experiments using the HCT116 cells that express HMGA2 and WT p53 endogenously. As shown in Figure 2A, HMGA2 was found to have a strong binding to p53. Next, the HEK293T cells were transiently transfected with pcDNA3.1, the Myc-tagged p53 and/or the FLAG-tagged HMGA2 expression plasmids, followed by dual Co-IP assays with either an anti-FLAG or anti-Myc antibody. The results showed that the HMGA2 and p53 proteins Co-IP reciprocally in these cells (Figure 2B,C). To further investigate whether the interaction between HMGA2 and p53 was direct, the GST pull-down assay was carried out. In this assay, a GST-fusion recombinant protein, containing full-length Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. www.pathsoc.org

Y Wang and L Hu et al

p53, was prepared and purified using glutathione beads. The Maltose Binding Protein (MBP)-tagged HMGA2 was incubated with affinity-purified GST or GST-p53. As shown in Figure 2D, the MBP-tagged HMGA2 bound to the GST-p53 but not the GST control beads. The HMGA2 protein consists of five main domains. The first three exons encode three AT-hook domains, whereas exons 4 and 5 encode the linker and the C-terminal acidic domain. To map precisely the HMGA2 domains responsible for the interaction with p53, we produced plasmids encoding the deleted mutant forms of the FLAG-tagged HMGA2 and expressed them in HEK293T cells, together with the Myc-tagged full-length p53 constructs. The results of the Co-IP assays revealed that the three AT-hook domains (amino acids 1–83) were required for binding to p53, but not the linker or the C-terminal acidic domain only (Figure 2E). The evolutionarily conserved p53 gene encodes a 393 amino acid protein, which consists of three highly conserved domains. The transactivation domain (TAD) is essential for the transcriptional activity and binding with MDM2. The central DNA binding domain (DBD) directly binds to the consensus DNA binding site. The C-terminal region contains a tetramerization domain (TD), which contributes to regulate the oligomerization status of p53. To delineate the specific HMGA2 binding region in p53, we expressed serial deletion mutants of Myc-tagged p53 together with full-length HMGA2 in HEK293T cells. As illustrated in Figure 2F, a p53 fragment comprising residues 294–393 successfully Co-IP full-length HMGA2. Collectively, these observations demonstrated that the three AT-hook domains of HMGA2 were found to interact with the TD of p53.

HMGA2 decreases the stability of the p53 protein We then asked whether HMGA2 modulated the mRNA or protein levels of p53 through their interaction with each other. To address this question, the endogenous HMGA2 levels were evaluated in LoVo, HCT116, and RKO cells, all of which had WT p53 (Figure 3A). The levels of HMGA2 expression were confirmed by western blotting (Figure 3A) and RT-qPCR analyses (Figure 3B). The results showed that HMGA2 overexpression inhibited p53 protein levels (Figure 3A) without altering TP53 mRNA level in LoVo cells (Figure 3B). In contrast, HMGA2 knockdown enhanced p53 protein expression (Figure 3A), but failed to affect TP53 mRNA expression in either HCT116 or RKO cells (Figure 3B). As a direct transcriptional target of p53, the cyclin-dependent kinase inhibitor p21 (CDKN1A) plays an important role in cell cycle arrest by interfering with CDKs [14]. Consistent with the results of p53 expression, p21 protein levels decreased in cells expressing HMGA2 and increased in HMGA2-knockdown cells (Figure 3A). These results indicated that HMGA2 functioned as an upstream regulator of p53 via a post-transcriptional mechanism, such as proteasomal degradation. J Pathol 2018; 246: 508–518 www.thejournalofpathology.com

HMGA2 facilitates MDM2-mediated ubiquitination of p53

511

Figure 1. Hmga2 increased AOM- or AOM/DSS-induced tumorigenesis in Hmga2KI/KI :PVillin-Cre T mice in vivo. (A) Strategy for the generation of conditional KI mice by inserting a WT Hmga2 into the ROSA26 gene locus. The targeting vector and the targeted allele are shown at the top and bottom, respectively. (B, C) Schematic diagram of (B) the AOM- and (C) the AOM/DSS-induced mouse cancer model. (D, E) Representative macroscopic images of the intestine tumors induced by (D, upper panel) AOM or (E, upper panel) AOM/DSS treatment. The arrows indicate tumors. Hematoxylin–eosin (HE) and Hmga2 IHC staining of tumors from (D, bottom panel) the AOMor (E, bottom panel) AOM/DSS-induced model. (F, G) The total number of intestine tumors from (F) AOM- or (G) AOM/DSS-treated mice. Mean ± SD. *p < 0.05.

Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. www.pathsoc.org

J Pathol 2018; 246: 508–518 www.thejournalofpathology.com

512

Y Wang and L Hu et al

Figure 2. The three AT-hook domains of HMGA2 interacted with the TD of p53. (A) Immunoprecipitation of the p53 protein by an anti-HMGA2 antibody in HCT116 cells. IgG was used as a negative control. (B) Immunoprecipitation of the Myc-p53 protein by an anti-FLAG antibody in HEK293T cells transfected with pcDNA3.1-Myc-p53 and/or pcDNA3.1-FLAG-HMGA2 as indicated. (C) Immunoprecipitation of the FLAG-HMGA2 protein by an anti-Myc antibody in HEK293T cells transfected with pcDNA3.1-Myc-p53 and/or pcDNA3.1-FLAG-HMGA2 as indicated. (D) Recombinant GST or GST-p53 proteins were incubated with purified MBP-HMGA2. The bound proteins in this GST pull-down assay were detected by an anti-HMGA2 antibody using western blotting. (E) Diagram showing the truncation mutants (1–35, 1–73, 1–83, and 84–109) of the FLAG-HMGA2 constructs (upper). Immunoprecipitation of the FLAG-HMGA2 protein by an anti-Myc antibody in HEK293T cells transfected with full-length pcDNA3.1-Myc-p53 and various truncation mutants of pcDNA3.1-FLAG-HMGA2 constructs as indicated (bottom). (F) Diagram showing the truncation mutants (1–102, 103–293, and 294–393) of the Myc-p53 constructs (upper). Immunoprecipitation of the Myc-p53 protein by an anti-FLAG antibody in HEK293T cells transfected with full-length pcDNA3.1-FLAG-HMGA2 and various truncation mutants of the pcDNA3.1-Myc-p53 constructs as indicated (bottom).

Next we examined whether HMGA2 attenuates the stability of p53 protein. Cycloheximide (CHX), a protein synthesis inhibitor, has been used to determine the effects of HMGA2 on p53 stability [19]. Cells with or without HMGA2 expression were treated with CHX for different times to block protein synthesis and the degradation rates of the existing p53 protein were measured by western blotting. The results showed that the overexpression of HMGA2 in LoVo cells accelerated p53 degradation compared with the control group (Figure 3C), whereas HMGA2 knockdown prolonged the half-life of endogenous p53 protein in both HCT116 (Figure 3D) and RKO cells (Figure 3E). To evaluate the relationship between p53 protein stability mediated by HMGA2 and the proteasome system we used MG132, a 26S proteasome inhibitor. Notably, treatment with MG132 allowed for an accumulation of p53 protein (Figure 3F–H). As shown in Figure 3G,H, p53 protein levels induced by proteasome Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. www.pathsoc.org

inhibition could not be further increased by HMGA2 knockdown. Furthermore, we also found that the p53 level did not differ significantly between the control and HMGA2-overexpressing cells following treatment with MG132 (Figure 3F). Thus, these data support that HMGA2 decreased the protein stability of p53 via proteasome-dependent degradation. To investigate whether HMGA2 had an impact on the target genes of p53, dual-luciferase reporter assays were conducted in LoVo (see supplementary material, Figure S2A–C) and RKO cells (see supplementary material, Figure S2D–F). Ectopically transfected p53 strongly stimulated the activity of the p21-Luc reporter. However, the p21 (CDKN1A) promoter activity was significantly repressed by the cotransfection of HMGA2 and p53 compared with that of p53 individually in LoVo cells (see supplementary material, Figure S2A) and RKO cells (see supplementary material, Figure S2D). Similar trends were found on the luciferase activities J Pathol 2018; 246: 508–518 www.thejournalofpathology.com

HMGA2 facilitates MDM2-mediated ubiquitination of p53

513

Figure 3. HMGA2 decreased the stability of the p53 protein. (A) Western blots showing p53, p21, and HMGA2 expression in LoVo cells transfected with the HMGA2 expression vector or control vector (LoVo-NC/A2) and HCT116 and RKO cells transfected with siRNAs against HMGA2 or control siRNA (HCT116-NC/siA2 and RKO-NC/siA2, respectively). Protein band intensities were measured using ImageJ and normalized to β-actin. Data were expressed as a fold-change relative to control. (B) RT-qPCR analysis of the p53 (TP53) and HMGA2 mRNA expression in CRC cells. **p < 0.01, ***p < 0.001. (C–E) Degradation of the p53 protein was measured after the treatment of 100 μg/ml CHX at the indicated time points in (C) LoVo-NC/A2, (D) HCT116-NC/siA2 and (E) RKO-NC/siA2 cells. Densitometry of p53 normalized to β-actin was plotted. Mean ± SD of three independent experiments. **p < 0.01, ***p < 0.001. (F–H) Western blotting analysis of p53 expression after treatment with 10 μM MG132 for 4 h in (F) LoVo-NC/A2, (G) HCT116-NC/siA2 and (H) RKO-NC/siA2 cells. Protein band intensities were measured using ImageJ and normalized to β-actin. Data were expressed as a fold-change relative to control.

of NOXA (PMAIP1) (see supplementary material, Figure S2B,E) and BAX promoters (see supplementary material, Figure S2C,F). Taken together, our data illustrate that HMGA2 downregulated p53 expression through decreasing the stability of the p53 protein.

HMGA2 interacts directly with MDM2 in vivo and in vitro Given that HMGA2 interacts directly with p53 and decreases its protein stability, we asked whether HMGA2 forms a complex with MDM2. We first Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. www.pathsoc.org

examined the interaction between endogenous HMGA2 and endogenous MDM2 by Co-IP in HCT116 cells. As shown in Figure 4A, HMGA2 co-precipitated with MDM2. Next, the reciprocal Co-IP analysis on HEK293T cells transfected with FLAG-HMGA2 and/or Myc-MDM2 revealed that HMGA2 co-precipitated with MDM2 (Figure 4B,C). In addition, the direct binding between HMGA2 and MDM2 was also confirmed by the GST pull-down assay. As shown in Figure 4D, MBP-tagged HMGA2 bound to GST-MDM2 but not to the GST control beads. All of these results demonstrated that HMGA2 interacted physically with MDM2. J Pathol 2018; 246: 508–518 www.thejournalofpathology.com

514

Y Wang and L Hu et al

Figure 4. The two or three AT-hook domains of HMGA2 interacted with the central acidic and zinc finger domains of MDM2. (A) Immunoprecipitation of the MDM2 protein by an anti-HMGA2 antibody in HCT116 cells. IgG was used as a control. (B) Immunoprecipitation of Myc-MDM2 protein by an anti-FLAG antibody in HEK293T cells transfected with pcDNA3.1-Myc-MDM2 and/or pcDNA3.1-FLAG-HMGA2 as indicated. (C) Immunoprecipitation of the FLAG-HMGA2 protein by an anti-Myc antibody in HEK293T cells transfected with pcDNA3.1-Myc-MDM2 and/or pcDNA3.1-FLAG-HMGA2 as indicated. (D) Recombinant GST or GST-MDM2 proteins were incubated with the purified MBP-HMGA2. The bound proteins from this GST pull-down assay were detected by an anti-HMGA2 antibody using western blotting. (E) Diagram of the truncation mutants (1–35, 1–73, and 1–83) of the FLAG-HMGA2 constructs (upper). Immunoprecipitation of the FLAG-HMGA2 protein by an anti-Myc antibody in HEK293T cells transfected with full-length pcDNA3.1-Myc-MDM2 and various truncation mutants of pcDNA3.1-FLAG-HMGA2 constructs as indicated (bottom). (F) Diagram of the truncation mutants (1–110, 111–360, and 361–491) of the Myc-MDM2 constructs (upper). Immunoprecipitation of the Myc-MDM2 protein by an anti-FLAG antibody in HEK293T cells transfected with full-length pcDNA3.1-FLAG-HMGA2 and various truncation mutants of the pcDNA3.1-Myc-MDM2 constructs as indicated (bottom).

MDM2 contains three different domains: an N-terminal p53 binding domain, followed by central acidic and zinc finger domains, and a C-terminal RING finger domain. To determine the specific domains of MDM2 binding to HMGA2, deletion variants of Myc-MDM2 were expressed together with full-length FLAG-HMGA2 in HEK293T cells. Co-IP results indicated that amino acids 111–360 of MDM2 were crucial for the direct interactions with HMGA2 (Figure 4F). Next, we investigated domains within HMGA2 that interacted with MDM2. As shown in Figure 4E, the two (amino acids 1–73) or three AT-hook domains (amino acids 1–83) of HMGA2 strongly interacted with MDM2. Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. www.pathsoc.org

HMGA2 promotes MDM2-mediated p53 ubiquitination To examine whether HMGA2 affects the physical interaction between MDM2 and p53, Co-IP experiments were carried out. HEK293T cells were transiently transfected with FLAG-p53 in the presence or absence of Myc-MDM2 and/or His-HMGA2, followed by Co-IP assays with an anti-FLAG antibody. The results showed that, in the presence of HMGA2, the binding affinity between MDM2 and p53 was induced (Figure 5A, compare lane 3 with lane 2). To study whether HMGA2 promotes p53 ubiquitination, we conducted ubiquitination assays. We first performed ubiquitination assays in LoVo cells that J Pathol 2018; 246: 508–518 www.thejournalofpathology.com

HMGA2 facilitates MDM2-mediated ubiquitination of p53

515

Figure 5. HMGA2 promoted MDM2-mediated p53 ubiquitination. (A) Immunoprecipitation of the Myc-MDM2 and FLAG-p53 proteins using an anti-FLAG antibody in HEK293T cells transfected with the indicated constructs. (B) Analysis of p53 ubiquitination was performed by immunoprecipitation using an anti-p53 antibody, followed by immunoblotting with anti-p53 in LoVo cells transfected with the indicated constructs. (C) Analysis of p53 ubiquitination was performed by immunoprecipitation using an anti-FLAG antibody, followed by immunoblotting with anti-HA or anti-FLAG antibody in HEK293T cells transfected with the indicated constructs. (D) Western blotting of p53, Hmga2, and Mdm2 proteins in intestine tissues of WT and Hmga2 KI mice (Left). Gapdh was used as the loading control (left). Analysis of p53 ubiquitination was performed by immunoprecipitation using an anti-p53 antibody, followed by immunoblotting with an anti-ubiquitin antibody in WT and KI mice (right).

endogenously express WT p53 and MDM2, but not HMGA2. We transfected with Ub and/or Myc-HMGA2 into LoVo cells and Co-IP assays were performed with p53 antibody. As shown in Figure 5B, overexpression of HMGA2 significantly promoted p53 ubiquitination in LoVo cells. In addition, we introduced exogenous FLAG-p53 and HA-Ub into HEK293T cells in the presence or absence of Myc-MDM2 and/or Myc-HMGA2. The cell extracts were immunoprecipitated by an anti-FLAG antibody, followed by an immunoblot analysis with anti-HA or anti-FLAG antibodies. As expected, p53 ubiquitination was apparent only in the presence of Myc-MDM2 (Figure 5C, lane 3–5). Furthermore, HMGA2 enhanced p53 ubiquitination in the presence of MDM2 in a dose-dependent manner (Figure 5C, lane 4–5). Taken together, our findings suggested that HMGA2 promoted MDM2-mediated p53 ubiquitination. To study the effects of the Hmga2-p53 interaction in vivo, we determined the protein levels of Hmga2, p53, and Mdm2 in intestinal tissues of WT and KI mice. Interestingly, overexpression of Hmga2 in KI mice significantly decreased the level of p53 protein relative to the WT controls, whereas the expression of Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. www.pathsoc.org

Mdm2 did not show notable differences (Figure 5D). Furthermore, ubiquitination assays were performed to detect the effect of Hmga2 on p53 ubiquitination. As shown in Figure 5D, Hmga2 overexpression in Hmga2 KI mice resulted in a substantial increase in the accumulation of ubiquitinated p53.

Correlation of HMGA2 expression with poor prognosis in the CRC subgroups To determine the prognostic power of HMGA2 in CRC we evaluated the expression of HMGA2 by IHC staining in both a training (n = 121) and a validation cohort (n = 90), whose clinical characteristics are shown in supplementary material, Table S1. As anticipated, the tumors with high HMGA2 were associated with a shorter overall survival (OS) rate in all of the patients in the training (p = 0.042) and validation cohorts (p = 0.032; see supplementary material, Table S2). Next, the association between HMGA2 expression and prognosis was investigated using various subset analyses by Cox proportional hazards model. As shown in supplementary material, Table S2, we observed that high HMGA2 had a worse predictive effect on prognosis in subgroups with tumor size ≤5 cm in the J Pathol 2018; 246: 508–518 www.thejournalofpathology.com

516

Y Wang and L Hu et al

Figure 6. A model depicting the association between the HMGA2-MDM2-p53 pathway and cell cycle and apoptosis in CRC.

validation cohort (p = 0.035). After stratifying by histologic grade, patients with high HMGA2 levels tended to show a shorter OS in the grade III subgroup in the training cohorts (P = 0.003; see supplementary material, Table S2). When stratified into AJCC stages I–II and III–IV, increased expression of HMGA2 was associated with worse outcome in subgroups of stage I–II in the training cohort (p = 0.040), but not in the validation cohort (P = 0.281; see supplementary material, Table S2). Stratifying our cases according to p53 staining, the patients in the training cohort with high HMGA2 had a worse OS in the p53-negative group (p = 0.032), but showed no significant difference in the p53-positive group (p = 0.551; see supplementary material, Table S2). Similar trends were found for the p53-negative (p = 0.042) and p53-positive subtypes (p = 0.312; see supplementary material, Table S2) in the validation cohort. Thus, our results suggested that HMGA2 showed a strong predictive value for the prognosis of patients with negative p53 and these two proteins were coordinated in the CRC tissues.

Discussion In this study, we reported that HMGA2 regulated cell cycle progression and apoptosis in vitro and promoted intestinal tumor development in Hmga2 KI mice in vivo. To our knowledge, we have, for the first time, generated intestinal epithelial cell-specific Hmga2 KI mice and characterized the role of Hmga2 in intestinal tumorigenesis by conditional mouse model. However, several lines of Hmga2-null and transgenic mice had been created and analyzed. Zhou et al [20] reported that Hmga2-null mice exhibited a pygmy phenotype. In contrast, the mice carrying truncated Hmga2 genes lacking its C-terminal domain developed an overabundance of adipose tissue and lipomas [21]. Furthermore, Fedele et al [22] found that transgenic mice overexpressing full-length Hmga2 genes under the control of the cytomegalovirus promoter developed pituitary adenomas which secreted prolactin and growth hormone. Consistently, we found that conditional KI of Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. www.pathsoc.org

Hmga2 developed intestinal adenomas in our study. Here, we also revealed a novel mechanism that linked HMGA2 to the MDM2-mediated p53 ubiquitination in CRC progression (Figure 6). HMGA2 directly interacted with p53 and accelerated p53 degradation in the presence of CHX, suggesting that HMGA2 decreased the protein stability of p53 in a proteasome-dependent manner. Then, we asked whether HMGA2 is involved in MDM2-mediated p53 ubiquitination. Interestingly, HMGA2 physically bound to MDM2. We observed that HMGA2 promoted MDM2-mediated p53 ubiquitination. It was intriguing that similar trends toward p53 ubiquitination were found in Hmga2 KI mice compared with WT mice. Furthermore, we observed that HMGA2 predicted poor prognosis in the p53-negative subgroup of CRC patients. It is well known that WT p53 protein is usually maintained at low levels [23]. However, the p53 mutant phenotype leads to abnormal p53 stabilization and mutant p53 protein overexpression [24,25]. Consistent with this, our study using two independent CRC cohorts suggested that HMGA2 was prognostic among p53-negative patients who were highly likely to have WT p53 (TP53) genes. Our findings provided additional support that HMGA2 mediated the function of WT p53 in CRC. Based on these findings, we reasoned that HMGA2 enhanced p53 protein degradation through MDM2-mediated ubiquitination. The tumor suppressor p53 has been studied extensively due to its role in diverse cellular processes [26–29]. Under physiological conditions, p53 is bound and negatively regulated by MDM2, which promotes its ubiquitination and degradation. At the same time, p53 promotes the transcription of MDM2 [23]. Taken together, p53 and MDM2 form a negative feedback loop which keeps p53 levels low in unstressed cells. Notably, MDM2 is the chief negative regulator of p53 activity and stability [23]. MDMX (also known as MDM4), an MDM2 homolog, is another negative modulator of p53. They share a high similarity in their p53 binding domains. However, MDMX has no intrinsic E3 ubiquitin ligase activity. It is capable of synergistic cooperation with MDM2 through RING–RING interaction, facilitating p53 degradation [30]. Abundant evidence J Pathol 2018; 246: 508–518 www.thejournalofpathology.com

HMGA2 facilitates MDM2-mediated ubiquitination of p53

demonstrates that heterodimerization of MDM2 and MDMX is crucial for p53 ubiquitination [31]. They work together and function as an integral complex in the control of p53 activity. Interestingly, Wang et al [32] reported that MDMX converted MDM2 from a monoubiquitin E3 ligase to a polyubiquitin E3 ligase for p53. How HMGA2 interacts with MDMX, however, remains unknown. Future studies may focused on it. As a transcription factor, p53 is involved in diverse biological processes by regulating the expressions of target genes, including cell-cycle arrest (e.g. CDKN1A, GADD45), apoptosis (e.g. BAX, PMAIP1, and BBC3) and DNA repair (e.g. DDB2, GADD45) [33] These ultimately contribute to the maintenance of genomic stability and tumor suppression. As mentioned above, the p53 protein has three main domains: TAD, DBD, and TD [26,34]. Previous studies demonstrated that the tetramer formation of p53, which was critical for p53 activity, was modulated by its TD [34,35]. Frasca et al [35] reported that HMGA1, another member of the HMGA family, directly bound to the COOH-terminal end of p53, thereby preventing its oligomerization in thyroid cancer. Several proteins, such as HERC2 and ARC, were also found to physically interact with the TD of p53, and this interfered with the formation of p53 tetramers [36,37]. Here, we reported that the Co-IP analysis revealed a physical association between HMGA2 and the TD of p53. We speculated that HMGA2 destabilized the tetrameric structure of p53 and postulated that HMGA1 and HMGA2 cooperated with each other and played an important role in carcinogenesis. The TAD of p53 mediates a direct interaction with MDM2, whereas MDM2 modifies p53 ubiquitination on six key lysine residues located at its C-terminus [38–40]. Sui et al [41] found that the transcription factor YY1 (Yin Yang 1) formed a ternary complex with p53 and MDM2, thereby promoting the MDM2-mediated p53 ubiquitination. In addition, Wu et al [42] reported that UBE4B (ubiquitination factor E4B) promoted p53 degradation in a similar manner. In this study, we found that HMGA2 negatively regulated the stability of p53 protein by stimulating MDM2-mediated p53 ubiquitination in vitro and in vivo. Notably, a similar trend was obtained from the WT and Hmga2 KI mouse models. Overall, these findings suggested that HMGA2-mediated CRC tumorigenesis might rely on the effect of p53. Post-translational modifications in p53 are critical processes that have effects on its stability [43–46]. In response to various genotoxic stresses, p53 is phosphorylated and acetylated at specific residues, thereby resulting in its stabilization and activation [43,47]. Hence, a number of questions remain unanswered, including how HMGA2 regulates p53 stability through other post-translational modifications (such as phosphorylation, acetylation, and neddylation). These can be exploited in future studies and will help yield new insights into the HMGA2 network underlying the process of CRC carcinogenesis. Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. www.pathsoc.org

517

In this study, we found that HMGA2 promoted cell cycle progress and inhibited apoptosis in CRC cells in vitro. Results from the Hmga2 KI mouse model indicated that intestine-specific Hmga2 overexpression accelerated chemical carcinogen-induced tumorigenesis in vivo. We observed that HMGA2 decreased the stability of the p53 protein and promoted MDM2-mediated p53 ubiquitination and degradation. In aggregate, our findings established HMGA2 as a potential candidate for the prevention and treatment of CRC by targeting the HMGA2-MDM2-p53 pathway.

Acknowledgements This study was supported by the National Natural Science Foundation of China (No. 81772527, 81672342, 81302073, 81302072 and 81572716), the 111 Project (B13026), the Major program of Zhejiang Province (2012C13014-3), Zhejiang Provincial Natural Science Foundation of China (LY17H160034) and Fundamental Research Funds for the Central Universities (2017QNA7004).

Author contributions statement JW and ML conceived and designed the study. YW, LH, JW, XL, and SS conducted experiments. YW, LH, JW, and ML analyzed data. JW and ML wrote and drafted the manuscript. All authors had final approval of the submitted versions.

References 1. Brenner H, Kloor M, Pox CP. Colorectal cancer. Lancet 2014; 383: 1490–1502. 2. Peters U, Bien S, Zubair N. Genetic architecture of colorectal cancer. Gut 2015; 64: 1623–1636. 3. Dong J, Wang R, Ren G, et al. HMGA2-FOXL2 axis regulates metastases and epithelial-to-mesenchymal transition of chemoresistant gastric cancer. Clin Cancer Res 2017; 23: 3461–3473. 4. Wang Y, Chen F, Zhao M, et al. The long noncoding RNA HULC promotes liver cancer by increasing the expression of the HMGA2 oncogene via sequestration of the microRNA-186. J Biol Chem 2017; 292: 15395–15407. 5. Yang F, Zhao L, Mei D, et al. HMGA2 plays an important role in Cr (VI)-induced autophagy. Int J Cancer 2017; 141: 986–997. 6. Macri S, Simula L, Pellarin I, et al. Hmga2 is required for neural crest cell specification in Xenopus laevis. Dev Biol 2016; 411: 25–37. 7. Wu J, Zhang S, Shan J, et al. Elevated HMGA2 expression is associated with cancer aggressiveness and predicts poor outcome in breast cancer. Cancer Lett 2016; 376: 284–292. 8. Yang E, Cisowski J, Nguyen N, et al. Dysregulated protease activated receptor 1 (PAR1) promotes metastatic phenotype in breast cancer through HMGA2. Oncogene 2016; 35: 1529–1540. 9. Wu J, Wei JJ. HMGA2 and high-grade serous ovarian carcinoma. J Mol Med 2013; 91: 1155–1165. 10. Hammond SM, Sharpless NE. HMGA2, microRNAs, and stem cell aging. Cell 2008; 135: 1013–1016. J Pathol 2018; 246: 508–518 www.thejournalofpathology.com

518

11. Wang X, Liu X, Li AY, et al. Overexpression of HMGA2 promotes metastasis and impacts survival of colorectal cancers. Clin Cancer Res 2011; 17: 2570–2580. 12. Zhao XP, Zhang H, Jiao JY, et al. Overexpression of HMGA2 promotes tongue cancer metastasis through EMT pathway. J Transl Med 2016; 14: 26. 13. Wu J, Wang Y, Xu X, et al. Transcriptional activation of FN1 and IL11 by HMGA2 promotes the malignant behavior of colorectal cancer. Carcinogenesis 2016; 37: 511–521. 14. Wang X, Simpson ER, Brown KA. p53: protection against tumor growth beyond effects on cell cycle and apoptosis. Cancer Res 2015; 75: 5001–5007. 15. Napoli M, Flores ER. The p53 family orchestrates the regulation of metabolism: physiological regulation and implications for cancer therapy. Br J Cancer 2017; 116: 149–155. 16. Wade M, Li YC, Wahl GM. MDM2, MDMX and p53 in oncogenesis and cancer therapy. Nat Rev Cancer 2013; 13: 83–96. 17. Kwon SK, Saindane M, Baek KH. p53 stability is regulated by diverse deubiquitinating enzymes. Biochim Biophys Acta 2017; 1868: 404–411. 18. Neufert C, Becker C, Neurath MF. An inducible mouse model of colon carcinogenesis for the analysis of sporadic and inflammationdriven tumor progression. Nat Protoc 2007; 2: 1998–2004. 19. Kao SH, Wang WL, Chen CY, et al. Analysis of protein stability by the cycloheximide chase assay. Bio Protoc 2015; 5: e1374. 20. Zhou X, Benson KF, Ashar HR, et al. Mutation responsible for the mouse pygmy phenotype in the developmentally regulated factor HMGI-C. Nature 1995; 376: 771–774. 21. Arlotta P, Tai AK, Manfioletti G, et al. Transgenic mice expressing a truncated form of the high mobility group I-C protein develop adiposity and an abnormally high prevalence of lipomas. J Biol Chem 2000; 275: 14394–14400. 22. Fedele M, Battista S, Kenyon L, et al. Overexpression of the HMGA2 gene in transgenic mice leads to the onset of pituitary adenomas. Oncogene 2002; 21: 3190–3198. 23. Junttila MR, Evan GI. p53--a Jack of all trades but master of none. Nat Rev Cancer 2009; 9: 821–829. 24. Bykov VJN, Eriksson SE, Bianchi J, et al. Targeting mutant p53 for efficient cancer therapy. Nat Rev Cancer 2018; 18: 89–102. 25. Brosh R, Rotter V. When mutants gain new powers: news from the mutant p53 field. Nat Rev Cancer 2009; 9: 701–713. 26. Bieging KT, Mello SS, Attardi LD. Unravelling mechanisms of p53-mediated tumour suppression. Nat Rev Cancer 2014; 14: 359–370. 27. Laptenko O, Prives C. p53: master of life, death, and the epigenome. Genes Dev 2017; 31: 955–956. 28. Meek DW. Regulation of the p53 response and its relationship to cancer. Biochem J 2015; 469: 325–346. 29. Fischer M. Census and evaluation of p53 target genes. Oncogene 2017; 36: 3943–3956.

Y Wang and L Hu et al

30. Karni-Schmidt O, Lokshin M, Prives C. The roles of MDM2 and MDMX in cancer. Annu Rev Pathol 2016; 11: 617–644. 31. Huang L, Yan Z, Liao X, et al. The p53 inhibitors MDM2/MDMX complex is required for control of p53 activity in vivo. Proc Natl Acad Sci U S A 2011; 108: 12001–12006. 32. Wang X, Wang J, Jiang X. MdmX protein is essential for Mdm2 protein-mediated p53 polyubiquitination. J Biol Chem 2011; 286: 23725–23734. 33. Riley T, Sontag E, Chen P, et al. Transcriptional control of human p53-regulated genes. Nat Rev Mol Cell Biol 2008; 9: 402–412. 34. Kamada R, Toguchi Y, Nomura T, et al. Tetramer formation of tumor suppressor protein p53: structure, function and applications. Biopolymers 2016; 106: 598–612. 35. Frasca F, Rustighi A, Malaguarnera R, et al. HMGA1 inhibits the function of p53 family members in thyroid cancer cells. Cancer Res 2006; 66: 2980–2989. 36. Cubillos-Rojas M, Amair-Pinedo F, Peiro-Jordan R, et al. The E3 ubiquitin protein ligase HERC2 modulates the activity of tumor protein p53 by regulating its oligomerization. J Biol Chem 2014; 289: 14782–14795. 37. Foo RS, Nam YJ, Ostreicher MJ, et al. Regulation of p53 tetramerization and nuclear export by ARC. Proc Natl Acad Sci U S A 2007; 104: 20826–20831. 38. Brooks CL, Gu W. p53 regulation by ubiquitin. FEBS Lett 2011; 585: 2803–2809. 39. Poyurovsky MV, Katz C, Laptenko O, et al. The C terminus of p53 binds the N-terminal domain of MDM2. Nat Struct Mol Biol 2010; 17: 982–989. 40. Kruse JP, Gu W. Modes of p53 regulation. Cell 2009; 137: 609–622. 41. Sui G, Affar el B, Shi Y, et al. Yin Yang 1 is a negative regulator of p53. Cell 2004; 117: 859–872. 42. Wu H, Pomeroy SL, Ferreira M, et al. UBE4B promotes Hdm2-mediated degradation of the tumor suppressor p53. Nat Med 2011; 17: 347–355. 43. Meek DW, Anderson CW. Posttranslational modification of p53: cooperative integrators of function. Cold Spring Harb Perspect Biol 2009; 1: a000950. 44. Charni M, Aloni-Grinstein R, Molchadsky A, et al. p53 on the crossroad between regeneration and cancer. Cell Death Differ 2017; 24: 8–14. 45. Abida WM, Nikolaev A, Zhao W, et al. FBXO11 promotes the Neddylation of p53 and inhibits its transcriptional activity. J Biol Chem 2007; 282: 1797–1804. 46. Kung CP, Khaku S, Jennis M, et al. Identification of TRIML2, a novel p53 target, that enhances p53 SUMOylation and regulates the transactivation of proapoptotic genes. Mol Cancer Res 2015; 13: 250–262. 47. Zheng S, Koh XY, Goh HC, et al. Inhibiting p53 acetylation reduces cancer chemotoxicity. Cancer Res 2017; 77: 4342–4354.

SUPPLEMENTARY MATERIAL ONLINE Supplementary materials and methods Figure S1. HMGA2 regulated cell cycle progression and apoptosis in vitro Figure S2. Effects of HMGA2 on the transcription and expression of target genes of p53 Table S1. Clinical characteristics of the training (n = 121) and validation cohorts (n = 90) Table S2. Stratification and univariate analysis for HMGA2 in CRC survival Table S3. Primers for RT-qPCR Table S4. Primers for construction of plasmids in Co-IP and GST pull-down assays Table S5. Primers for genotyping in generation of conditional Hmga2 knock-in mice

Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. www.pathsoc.org

J Pathol 2018; 246: 508–518 www.thejournalofpathology.com