Mad1 destabilizes p53 by preventing PML from sequestering MDM2

Abstract

Mitotic arrest deficient 1 (Mad1) plays a well-characterized role in the mitotic checkpoint. However, interphase roles of Mad1 that do not impact mitotic checkpoint function remain largely uncharacterized. Here we show that upregulation of Mad1, which is common in human breast cancer, prevents stress-induced stabilization of the tumor suppressor p53 in multiple cell types. Upregulated Mad1 localizes to ProMyelocytic Leukemia (PML) nuclear bodies in breast cancer and cultured cells. The C-terminus of Mad1 directly interacts with PML, and this interaction is enhanced by sumoylation. PML stabilizes p53 by sequestering MDM2, an E3 ubiquitin ligase that targets p53 for degradation, to the nucleolus. Upregulated Mad1 displaces MDM2 from PML, freeing it to ubiquitinate p53. Upregulation of Mad1 accelerates growth of orthotopic mammary tumors, which show decreased levels of p53 and its downstream effector p21. These results demonstrate an unexpected interphase role for Mad1 in tumor promotion via p53 destabilization.

Fig. 1

Upregulated Mad1 accumulates into PML nuclear bodies (NBs) in cells and tumors. a–c Mad1-3xFLAG colocalizes with SUMO1 and SUMO2, which are markers of PML nuclear bodies. Mad1-3xFLAG colocalizes with Myc-SUMO1 (a) and Myc-SUMO2 (b) in MDA-MB-231 cells. Scale bars, 2.5 µm. c Mad1-3xFLAG colocalizes with endogenous SUMO1 in HeLa cells. Scale bar, 20 µm. d Mad1-3xFLAG colocalizes with endogenous PML in HeLa cells. Scale bar, 2.5 µm. e Mad1-3xFLAG colocalizes with HA-PML in HeLa cells. Scale bar, 2.5 µm. f Quantitation (±SD) of the percentage of cells with Mad1-3xFLAG puncta that colocalize with Myc-SUMO1, Myc-SUMO2, endogenous SUMO1, endogenous PML, and HA-PML, as shown in panels (a–e). n = 200 cells from each of three independent experiments. g HeLa cells expressing Mad1-3xFLAG in a tet-inducible manner were treated with tet for 24 h followed by treatment with 1 μM As₂O₃ to disintegrate PML nuclear bodies for the indicated time. Quantitation (±SD) of the percentage of PML foci that colocalize with Mad1 in the presence or absence of As₂O₃. n = 200 cells from each of three independent experiments. h Immunofluorescence showing loss of PML NBs and Mad1 puncta in response to arsenic. Scale bar, 5 µm. i, j Mad1 localizes to PML NBs in human breast cancer tumor tissue sections. i Quantitation of the percentage of Mad1 puncta in breast tumors that colocalize with SUMO1, a marker of PML NBs. n = 100 cells from each of 9 independent samples. j Representative images of Mad1 puncta colocalizing with SUMO1 in breast tumors from subjects 3, 5, and 9. Pan-cytokeratin is used as a marker of epithelial (tumor) cells. Insets show a magnified view of cells boxed in the first column. Scale bar, 100 µm

Fig. 2

The C-terminal domain (CTD) of Mad1 interacts with the N-terminus of PML. a Schematic of Mad1 and the Mad1 fragments tested. NIS nuclear import signal, NPD nuclear pore targeting domain, M2iD Mad2-interacting domain^,, CTD C terminal domain. b, c The CTD of Mad1 is necessary for Mad1 to interact with PML. 1-718(FL) indicates full length Mad1. b Only Mad1-3xFLAG constructs containing the CTD co-immunoprecipitate HA-PML from 293T cells. Blot is representative of 3 independent experiments. c HA-PML co-immunoprecipitates all FLAG-tagged Mad1 fragments that contain the CTD, but not fragments lacking the CTD. Blot is representative of 3 independent experiments. d–f The CTD of Mad1 is necessary for its localization to PML NBs. HeLa cells were co-transfected with constructs for full length wild type PML-HA and full length wild type Mad1-3xFLAG or the indicated FLAG-tagged Mad1 deletion mutants and analyzed by immunofluorescence microscopy with anti-HA and anti-FLAG antibodies. Scale bars, 5 µm. d Full length Mad1-3xFLAG localizes to PML NBs. e The CTD of Mad1 is necessary for localization to PML NBs. f The CTD also requires the Mad1 nuclear pore targeting domain (NIS + NES; aa 1–274) for localization to PML NBs. g–i The N-terminus of PML interacts with Mad1. g Schematic of PML-IV and the fragments used in this study. R RING-finger domains, B B-boxes, CC α-helical coiled-coil domain, NLS nuclear localization signal, S SUMOylation site. Only the N-terminal and CC domains were efficiently expressed. h Full length PML and an N-terminal fragment but not an internal fragment co-immunoprecipitate Mad1-3x-FLAG. Blot is representative of 3 independent experiments. i Mad1-3xFLAG co-immunoprecipitates full length and the N-terminus of PML, but not an internal fragment of PML. Blot is representative of 3 independent experiments

Fig. 3

A conserved SIM within the Mad1 CTD is necessary for Mad1 localization to PML NBs. a Alignment of a portion of the Mad1 CTD protein sequence showing a conserved putative SIM. Residues identical in all species are shown in yellow. Residues conserved in a majority of species are shown in blue. Functional components of SIMs are shown in the schematic underneath. Putative Mad1 SIM residues conserved between species are shown in bolded red font. b Sequence alignment of the putative SIM within the Mad1 CTD and known SIMs in PML and DAXX. Red lowercase p indicates the serine is phosphorylated. c Mutation of the SIM disperses Mad1 from PML NBs. HeLa cells co-transfected with FLAG-tagged wild type (WT) or SIM mutant Mad1 and HA-tagged PML were analyzed 36 h after transfection using Mad1 and HA antibodies. The leucine and isoleucine residues at positions 689 and 690 were mutated to lysines to generate SIM mutant Mad1. Scale bar, 2.5 µm. d Reciprocal immunoprecipitations showing the interaction between Mad1 and PML is dependent on the Mad1 SIM. 293T cells were co-transfected with HA-tagged PML and FLAG-tagged WT or SIM mutant Mad1. 48 h later, cells were lysed and FLAG or HA antibodies used for immunoprecipitation. A single film for both experiments is shown. Blot is representative of 3 independent experiments. e Schematic of proteins and protein fragments used in (f). Sumoylation of PML was simulated by fusing a 3xSUMO2 chain to the N terminus of PML. f MBP pull-down experiments showing both full length Mad1 and the CTD of Mad1 interact with PML directly and that this interaction is enhanced by sumoylation of PML

Fig. 4

Mad1 upregulation destabilizes p53 and impairs cell death in response to DNA damage. a MDA-MB-231 cells stably expressing Mad1-YFP in response to tet were treated ± tet for 24 h and then with the topoisomerase II inhibitor doxorubicin (2 µg/mL) to induce DNA damage for the indicated times. Blot is representative of 3 independent experiments. b The CTD of Mad1 is essential for its role in preventing stabilization of p53. HeLa cells stably expressing full length (FL) Mad1-3xFLAG or Mad1 lacking the CTD (aa 1–596) in a tet-inducible manner were treated with tet for 24 h and then with doxorubicin (2 µg/mL) for the indicated times. Blot is representative of 3 independent experiments. c MCF10A cells were infected with adenoviruses expressing mNeonGreen or Mad1-mNeonGreen for 1 h and then treated with 1 µg/mL doxorubicin to induce DNA damage for the indicated times. Expression of Mad1-mNeonGreen prevents stabilization of p53 and its downstream effector p21. d The SIM in the CTD of Mad1 is necessary for Mad1 to prevent stabilization of p53 and p21. HeLa cells stably expressing tet-inducible wild type Mad1-3xFLAG or Mad1-SIM mutant-3xFLAG were treated with tet for 24 h and then with 2 µg/mL doxorubicin for the indicated number of hours. Wild type Mad1 but not SIM mutant Mad1 prevents stabilization of p53 and p21. Blot is representative of 3 independent experiments. e Quantitation of cell death (±SD) in HeLa cells stably expressing tet-inducible 3xFLAG tagged full length or ΔCTD Mad1 ± 24 h tet, then ±2 µg/mL doxorubicin for 0, 12, and 24 h. Full length Mad1, but not Mad1 lacking the CTD, prevented cell death due to DNA damage. n > 250 cells from each of 3 independent experiments

Fig. 5

Mad1 destabilizes p53 by preventing sequestration of MDM2 into nucleoli. a–d In response to DNA damage, upregulated Mad1 replaces MDM2 in nucleoli. a, b MDM2 relocalizes to the nucleolus after DNA damage caused by doxorubicin in control cells (-tet) but not in cells expressing Mad1 in response to tet. HeLa cells stably expressing tet-inducible full length wild type Mad1 and GFP-nucleolin-P2A-3xFLAG-MDM2 were cultured for 24 h in the presence of tet and an additional 8 h with 2 µg/mL doxorubicin. a Immunofluorescence showing colocalization of MDM2 with GFP-nucleolin after DNA damage. Scale bar, 2.5 µm. b Quantification of the percentage of cells (±SD) exhibiting nucleolar MDM2. n > 200 cells from each of three independent experiments. c, d Upregulated Mad1 localizes to nucleoli in response to DNA damage. HeLa cells stably expressing GFP-nucleolin and tet-inducible wild type Mad1 were cultured with tet for 24 h and doxorubicin for another 8 h. c Immunofluorescence showing colocalization between Mad1 and GFP-nucleolin in response to DNA damage. Scale bar, 2.5 µm. d Quantitation of the percentage of cells (±SD) with nucleolar Mad1 staining. n ≥ 200 cells from each of three independent experiments. e, f HeLa cells stably expressing tet-inducible Mad1(SIM mutant)-3xFLAG were incubated with adenovirus expressing nucleolin-mNeonGreen-P2A-mScarlet-MDM2 for 1 h and then cultured with tet for 24 h and doxorubicin for another 8 h. e Immunofluorescence showing MDM2 localization to nucleoli after doxorubicin was unaffected by expression of SIM mutant Mad1. f Quantification of the percentage of cells (±SD) exhibiting nucleolar MDM2 in (c). n > 200 cells from each of three independent experiments. g, h HeLa cells stably expressing tet-inducible SIM mutant Mad1 were incubated with adenovirus expressing mNeonGreen-nucleolin for 1 h and then cultured for 24 h in the presence of tet and an additional 8 h ± 2 µg/mL doxorubicin. g SIM mutant Mad1 does not colocalize with nucleoli. Scale bar, 5 µm. h Rotated views of (g), showing the lack of colocalization between SIM mutant Mad1 and GFP-nucleolin. *p < 0.05. **p < 0.001. ns non-significant using t test. For specific p values, see Source Data file

Fig. 6

Upregulation of Mad1 prevents the interaction between MDM2 and PML, freeing MDM2 to bind p53. a Increased expression of full length (FL) Mad1 but not Mad1 lacking the CTD (aa 1–596; Mad1ΔCTD) impairs co-immunoprecipitation of PML with MDM2. 293T cells were co-transfected with HA-PML, 3xFLAG-MDM2, and increasing amounts of FL Mad1 or Mad1ΔCTD. b Reciprocal experiment showing that increased expression of FL Mad1 but not Mad1ΔCTD impairs co-precipitation of MDM2 with PML. 293T cells were co-transfected with 3xFLAG-PML, 2xHA-MDM2, and increasing amounts of FL Mad1 or Mad1ΔCTD. c Upregulation of Mad1 inhibits the interaction of MDM2 with PML, freeing MDM2 to interact with p53. 293T cells were co-transfected with 2xHA-PML, 3xFLAG-MDM2, and increasing amounts of FL Mad1 or Mad1ΔCTD. a–c 48 h after transfection, cell extracts were prepared and immunoprecipitated using beads coupled to anti-FLAG antibodies. Blots are representative of 3 independent experiments

Fig. 7

Mad1 upregulation promotes tumor initiation and accelerates tumor growth. a–f Upregulation of Mad1 is tumor promoting. Parental cells express the tet repressor but not Mad1 in response to tet. All mice were on a diet containing the tet analog dox (and a blue dye) to induce expression of Mad1 or control for the effects of dox. Arrows indicate tumors. a, b Upregulation of Mad1 permits orthotopic mammary tumor development by MCF10A cells. a Representative images after orthotopic mammary gland injection of MCF10A cells into nude mice. Parental MCF10A cells are nontransformed and failed to form tumors at any of 12 injection sites in nude mice (control). However, expression of Mad1-mNeonGreen was sufficient to induce orthotopic tumor formation at 9 out of 12 injection sites. b Growth rates of tumors after injection of MCF10A cells. Data from 6 parental and 6 MCF10A-Mad1-NeonGreen contemporaneous injections from a single experiment are shown. c, d Expression of Mad1-YFP promotes the growth of MDA-MB-231 orthotopic mammary tumors in nude mice. c Representative images of orthotopic mammary tumors in nude mice. d Volumes of each MDA-MB-231 tumor over time. n = 6 parental and 6 Mad1-YFP expressing tumors. e, f PML binding via the C-terminal SIM is necessary for Mad1 to promote mammary tumor growth. e Representative images of orthotopic mammary tumors. f Volumes of each MDA-MB-231 tumor. n = 6 parental, 6 Mad-YFP and 6 SIM mutant Mad1 expressing tumors. g Tumors expressing Mad1-YFP have lower levels of p53 and its effector p21. Protein lysates from tumors collected at day 40 post-injection were analyzed by immunoblot. h A proposed model for the role of upregulated Mad1 in destabilizing p53. In unstressed control cells (top), p53 levels are low due to continuous ubiquitination by MDM2 in PML NBs. In control cells exposed to DNA damage (middle), PML sequesters MDM2 into the nucleolus, physically separating MDM2 from p53. p53 protein accumulates and induces transcription, resulting in tumor suppressive effects. When Mad1 is upregulated (bottom), Mad1 competes with MDM2 to bind PML. PML sequesters Mad1 into nucleoli in response to DNA damage, permitting MDM2 to continue to bind and ubiquitinate p53 within PML NBs. p53 protein levels remain low, and downstream events of p53 stabilization including p21 accumulation and cell death do not occur. **p < 0.001 by Sen–Adichie test. For specific p values, see Source Data file