βTrCP regulates BMI1 protein turnover via ubiquitination and degradation

BMI1 undergoes proteasome-mediated degradation.

To examine the mechanism of posttranslational regulation of BMI1, we determined half life of BMI1 in MCF10A-BMI1 by inhibiting the de novo protein synthesis using cycloheximide (CHX) treatment for different time points (0–60 min). The BMI1 protein remaining after each time point was examined using western blot analysis and quantified by densitometry. Our data indicated that BMI1 is rapidly turned over with a half life of ∼40 min (Fig. 1A). On longer exposure of the western blot, we noticed several slow mobility bands of BMI1, which increased in intensity upon CHX treatment (Fig. 1A). We reasoned that these bands may represent phosphorylated forms of BMI1. Since, the multiple bands of BMI1 could complicate the half life analysis, we also determined half life of BMI1 after treating the extracts with calf-intestinal phosphatase (CIP) (Fig. 1A, lower part). The total cell extracts from each time point were treated with CIP in vitro, run on a gel, analyzed by western blot analysis and the half life of BMI1 was determined as above. The data indicated that CIP treatment results in disappereance of slow mobility bands and appearance of a faster mobility band, which represent the dephosphorylated form of the BMI1. Our data also indicated that the half life of CIP treated BMI1 is also ∼40 min (Fig. 1A and lower part).

The turnover of most cellular proteins involves either lysosomal or proteasomal degradation pathway. To determine whether BMI1 turnover involves lysosomal and/or proteasomal pathway, we used pharmacological inhibitors of these pathways. Our data indicated that treatment of cells with the 26S proteasomal inhibitor MG132 but not the lysosomal inhibitor chloroquine accumulates BMI1 in both MCF10A-BMI1 overexpressed as well as MCF10A-B0 cells. Thus exogenously expressed BMI1 (MCF10A-BMI1 cells) as well as endogenous BMI1 (MCF10A-B0 cells) are degraded via 26S-proteasome pathway (Fig. 1B).

As the proteasome degradation pathway is often mediated by the APC or the SCF ubiquitin ligase complexes,^18,19 and we did not observe any consensus APC degron motif in BMI1, we investigated the possibility of SCF-type E3 ligase complex regulating BMI1 turnover. To examine whether BMI1 undergoes degradation with the help of SCF type E3 ubiquitin ligase complex, we used a dominant negative (DN) mutant of the scaffolding protein Cullin 1 (ΔCul1), which has been shown to stabilize proteins known to be degraded by the SCF machinery. Our data indicated that the transient overexpression of ΔCul1-Flag accumulated BMI1 dose dependently in U2OS cells (Fig. 1C). These data indicate that the BMI1 is degraded via 26S-proteasome pathway involving the SCF complex.

βTrCP, the F-box component of SCF regulates BMI1 expression.

Since βTrCP is the most well characterized F box component of the SCF complex, we analyzed primary protein sequence of the BMI1 for the presence of a βTrCP recognition motif (βTrCP degron) (Fig. 2A). Our analysis indicated the presence of a DSGsdkanS sequence, which conforms to the consensus βTrCP recognition motif DSG(X)₂ + n(S) found in several SCF regulated proteins such as CDC25A, MCL1, SNAIL and β-catenin (Fig. 2A and right part). We further determined that the putative βTrCP degron site of BMI1 is highly conserved among various mammalian species (Fig. 2A left part). Interestingly βTrCP degron site is not present in Mel-18, which is highly homologus to BMI1, suggesting that Mel-18 is not regulated by βTrCP. Thus, the βTrCP degron site in BMI1 is likely to have functional relevance.

The presence of βTrCP degron in BMI1 prompted us to investigate whether βTrCP regulates BMI1 proteolysis via its ubiquitination and degradation. First, we examined the role of βTrCP in BMI1 degradation using RNA interference (RNAi) approach. U2OS cells overexpressing HA tagged wild-type βTrCP were transiently transfected with plasmids expressing a control shRNA, and two specific shRNA (TrCPi-1 and TrCPi-2), which have been shown to target βTrCP.²² Our data showed a modest increase of BMI1 levels after transient expression of shRNAs targeting βTrCP (Fig. 2B). Next, using a retroviral expression system, we overexpressed HA tagged full length wild-type βTrCP (TrCPWT-HA) and a DN mutant of it (TrCPΔF-HA) in MCF10A cells, and analyzed the levels of BMI1 in cells expressing wild-type or DN mutant TrCP proteins. Our results indicated that cells overexpressing wild-type βTrCP had decreased levels of BMI1 as compared to control cells, while cells overexpressing ΔF box mutant had increased levels of BMI1 protein (Fig. 2C). The biochemical activity associated with the BMI1 containing polycomb repressive complex 1 (PRC1) is monoubiquitination of K119 residue of the histone 2A (H2A K119Ub).²³ Accordingly, BMI1 overexpression has been shown to increase H2A K119Ub activity of the PRC1. Hence, we determined whether upregulation of BMI1 by βTrCP RNAi approach or exogenous expression of a DN mutant of βTrCP resulted in increase in H2A K119Ub levels. Our results indicated that indeed overexpression of DN mutant of βTrCP or downregulation of βTrCP by RNAi approach resulted in a modest increase in H2A K119 levels (Fig. 2D), suggesting that βTrCP is a physiological regulator of BMI1.

βTrCP promotes ubiquitination of BMI1.

Next, we determined whether βTrCP regulates BMI1 degradation via ubiquitination of BMI1. MCF10A cells overexpressing TrCPWT-HA and TrCPΔF-HA along with vector control cells were treated with 10 µM MG132 for 5 hr followed by immunoprecipitation with a BMI1 antibody. The immunoprecipitated samples were probed with an anti-ubiquitin antibody. The whole cell extracts were also probed for BMI1, and βTrCP using anti-HA antibody. The results of immunoblot assays indicated that overexpression of wild-type βTrCP increased ubiquitination of BMI1 as compared to control cells, while the DN mutant (βTrCPΔF) decreased relative ubiquitination of BMI1 (Fig. 3A). Based on these data, we conclude that the βTrCP promotes ubiquitination of BMI1.

To further examine the role of βTrCP in BMI1 ubiquitination, we constructed a BMI1 mutant in which the important βTrCP degron motif residues DSGS were changed to AAAA using site directed mutagenesis. We reasoned that this mutant of BMI1 (here in referred to as BMI Mut-FLAG) should be partially defective in ubiquitination. The Flag tagged wild-type (BMI1 WT-FLAG) and mutant BMI1 (BMI1 Mut-FLAG) were overexpressed in MCF10A and U2OS cells. The western blot analysis of wild-type and mutant BMI1 in U2OS and MCF10A cells suggested that the mutant BMI1 is expressed at higher levels than the wild-type BMI1 (Fig. 3B), possibly due to its inability to undergo βTrCP-mediated ubiquitination and degradation. To examine this possibility, we further studied ubiquitination of exogenously expressed wild-type and mutant BMI1 in MCF10A cells by immunoprecipitation using FLAG antibody followed by immunoblot analysis of ubiquitin using an anti-ubiquitin antibody. Our results indicated that indeed the mutant BMI1 is partially defective in ubiquitination (Fig. 3C).

βTrCP interacts with wild-type BMI1.

We reasoned that defective ubiquitination of mutant BMI1 protein may result due to inability of the βTrCP to interact with it and recruit SCF ubiquitination machinery. To examine this possibility, first we determined whether the wild-type BMI1 interacts with βTrCP. We developed U2OS cells stably overexpressing wild-type BMI1 and HA tagged wild-type βTrCP (TrCP WT-HA) or mutant βTrCP (TrCPΔF-HA). The interaction of BMI1 with the wild-type or mutant βTrCP was studied by co-immunoprecipitation (co-IP) assays. The cells were treated with 10 µM MG132 for 5 hr followed by IP of BMI1 and βTrCP using anti BMI1 and anti HA antibodies respectively. The immunoblot analysis of the IPs indicated that BMI1 co-IPed both wild-type βTrCP and ΔF mutant of βTrCP, and conversely βTrCP co-IPed BMI1 (Fig. 3D). Thus our data indicated that BMI1 physically interacts with βTrCP in an F-box-independent manner. These data are consistent with published reports, which have shown that βTrCP interacts with its substrates via WD repeats and not the F box.^24,25

Next, we examined whether wild-type BMI1 but not the mutant BMI1 with an altered βTrCP degron motif interacts with βTrCP. For this purpose, we exogenously expressed HA-tagged wild-type βTrCP in U2OS cells expressing FLAG tagged wild-type or mutant BMI1 protein, and carried co-IP assays using anti FLAG and anti HA antibodies. The results indicated that βTrCP specifically co-IPed wild-type but not the mutant BMI1 protein, and conversely wild-type BMI1 but not the mutant BMI1 co-IPed βTrCP (Fig. 3E). Thus, our data indicate that the βTrCP interacts with BMI1 via its recognition motif present in BMI1.

Mutant BMI1 with an altered βTrCP degron motif exhibits increased oncogenic activity.

As the mutant BMI1 with an altered βTrCP degron motif does not interact with βTrCP, and is partially defective in ubiquitination and degradation, we reasoned that the mutant BMI1 should be biologically more active than the wild-type BMI1. We have previously shown that BMI1 overexpression leads to increased proliferation and upregulation of AKT and ERK activities in HMECs.^3,8,11 To determine whether mutating βTrCP degron motif of BMI1 augments proliferative activity of BMI1, we examined the expression of AKT, GSK and ERK in control, and wild-type and mutant BMI1 expressing MCF10A cells. MCF10A-derived cells were made quiescent by EGF and serum deprivation for 72 hr, and cells were then stimulated by EGF for 20 min. Consistent with the role of AKT, GSK and ERK in cell signaling, expression of P-ERK (phospho-ERK) was very low whereas P-AKT (phospho-AKT) was undetectable in quiescent MCF10A cells (Fig. 4A). EGF stimulation of cells caused substantial upregulation of P-AKT, P-GSK (Phospho-GSK), and P-ERK in all control as well wild-type and mutant BMI1 expressing MCF10A cells. Furthermore, compared to wild-type BMI1, mutant BMI1 expression always led to a modest increase of P-ERK, P-GSK and P-AKT (Fig. 4A). We also examined the rate of proliferation of control MCF10A, and wild-type- and mutant BMI1-expressing cells. Our results indicated that MCF10A cells expressing mutant BMI1 proliferated more rapidly than control or wild-type BMI1 expressing cells (Fig. 4B).

To determine whether the mutation in βTrCP degron motif of BMI1 impacts its oncogenic potential in HMECs, we performed invasion and wound healing assays. First, we studied invasion potential of MCF10A-derived cells using transwell invasion assay. The results suggested a modest but significant increase in invasion potential of MCF10A cells expressing mutant BMI1 compared to wild-type BMI1 expressing cells (Fig. 4C). Next, we determined the migration potential of control, and wild-type and mutant BMI1 expressing MCF10A cells using wound healing assay. The relative migration of cells was computed by measuring the average distance remaining free of cells in each case. The result showed a significant increase in migration potential of mutant BMI1 expressing cells compared to wild-type BMI1 expressing and control MCF10A cells (Fig. 4D).

The mutant BMI1 promotes proliferation and bypass senescence in HDFs more efficiently than the wild-type BMI1.

Next, we determined whether the overexpression of wild-type and mutant BMI1 resulted in increased H2A K119Ub activity of PRC1. The levels of H2A K119Ub in control, and wild-type and mutant BMI1 overexpressing IMR90 fibroblasts were analyzed by western blot analysis. The results indicated a modest two fold increase in H2A K119Ub levels in cells expressing mutant BMI1 as compared to wild-type BMI1 expressing IMR90 cells (Fig. 5A). Next, we determined whether the increased stability of the mutant BMI1 is reflected in increased binding of it to the known targets of BMI1, and that whether the mutant BMI1 expressing fibroblasts proliferated more rapidly as compared to the wild-type BMI1 expressing and control fibroblasts. Using chromatin-immunoprecipitation linked PCR (ChIP) assay, we examined the binding of wild-type and mutant BMI1 to two well known PcG targets-p16^INK4a,²⁶ and HOXA9.²³ The results of ChIP assay showed that compared to wild-type BMI1, the mutant BMI1 immunoprecipitated two-fold more p16^INK4a chromatin (Fig. 5B). The mutant BMI1 also exhibited increased ChIP activity when primers for HOXA9 promoter were chosen for the ChIP assay (not shown). As expected, increased binding of the mutant BMI1 to p16^INK4a locus resulted in proportionally decreased expression of p16^INK4a mRNA as determined by the qRT-PCR (Fig. 5C), as well as decreased levels of p16^INK4a as determined by the western blot analysis (Fig. 5D).

Next, we determined whether the mutant BMI1 expressing cells proliferated more rapidly. A short term proliferation assay indicated that the mutant BMI1 expressing HDFs proliferated more rapidly compared to wild-type BMI1 expressing cells, which also proliferated better than the control vector infected cells (Fig. 5E). Because p16^INK4a is an important determinant of cellular senescence, and its downregulation by BMI1 overexpression has been shown to bypass senescence,^12,14 we determined whether the mutant BMI1 bypasses senescence more efficiently. Senescence bypass in IMR90 cell was studied using SA-β-gal marker.²⁷ We compared the number of SA-β-gal positive (senescent) cells in control, and wild-type and mutant BMI1 expressing cells at similar passage number. The results indicated that mutant BMI1-expressing IMR90 culture has the least number of senescent cells (Fig. 5F). Thus our data indicate that the mutant BMI1 protein is much more efficient in bypassing senescence in HDFs.

Cells and cell culture methods.

Immortal HMEC strain MCF10A and a human osteosarcoma cell line U2OS were obtained from the American Type Culture Collection (ATCC) (Manassas, VA USA). The MCF10A cells were cultured as described previously in reference ³⁶. IMR90 strain of HDFs was obtained from the National Institute on Aging (NIA), Aging Cell Repository, Coriell Institute for Medical Research, Camden, NJ. HDFs and U2OS were cultured in DMEM containing 10% FCS, 2 mM L-glutamine at 37°C under 10% CO₂ and 90% atmosphere as described in reference ²⁷.

Expression vectors and expression methods.

A retroviral vector overexpressing wild-type BMI1 has been described earlier in reference ¹¹ and ¹². The cDNA of βTrCP degron site mutant of BMI1 was generated by PCR, sequence verified and cloned into pBabe-puro and pMSCV-neo retroviral vectors using standard cloning methods. Where indicated FLAG or HA epitopes were added to cDNA of interest using PCR. For βTrCP knockdown experiments, shRNA vectors pU6-shRNA-βTrCP1-1, pU6-shRNA-βTrCP1-2 and pU6-shRNA-hHOS (mutant control) were obtained from Dr. Serge Fuchs (Univ. of Pennsylvania, PA). To express human βTrCP1, its cDNA was amplified from Flag-β TrCP1 vector (Addgene, Cambridge, MA) and cloned into pBabe-puro vector. An F-box deletion mutant of βTrCP1 was also generated by PCR and cloned into pBabe-puro vector. The DN mutant of Cullin 1 was also obtained from Addgene (Cambridge, MA).

Retroviruses were generated by transfecting retroviral vectors together with a packaging plasmid pIK into tsa54 cells and collecting supernatant as described in reference ³⁷. Stable cell lines expressing gene(s) of interest were generated by infection of the retroviral vector(s) expressing a particular gene and selecting cells in either puromycin or G418 as described in reference ^{11, 14} and ³⁷. For transient expressions, U2OS cells were transfected using different concentrations of plasmids using lipofectamine-2000 (Invitrogen, Carlsbad, CA).

Chromatin immunoprecipitation and quantitative real time RT-PCR assays.

Chromatin immunoprecipitation-linked PCR (ChIP) assays were performed as described in reference ³⁰ and ³⁸. Briefly, cells were treated with 1% formaldehyde for 20 min at room temperature. The cross-linked chromatin was isolated and sonicated to yield 200–500 bp fragments. Immunoprecipitations (IPs) were carried out using a custom made rabbit polyclonal antibody raised against BMI1, and a control IgG. The BMI1- and IgG-bound chromatin were amplified using p16^INK4a-specific primers by quantitative real time PCR (qPCR) in a Rotor-Gene 6000 series PCR machine (Corbett Life Sciences-Qiagen Inc., Valencia, CA). The primers used for amplification of p16^INK4a were- Forward (F) 5′-GGG CTC TCA CAA CTA GGA AAG-3′ and Reverse (R) 5′-GGT GTT TGG TGT CAT AGG G-3′. To perform quantitative real time reverse transcript PCR (qRT-PCR), total RNA was isolated using Trizol (Invitrogen, Carlsbad, CA), and cDNA synthesis and PCR amplification were carried out using SensiMix Plus SYBR kit (Quantace Inc.). The primers used for p16^INK4a cDNA amplification were- F-5′-CCC CTT GCC TGG AAA GAT AC-3′ and R-5′-AGC CCC TCC TCT TTC TTC CT-3′. Real time PCR data were analyzed using Rotor-Gene Q software package (Corbett Life Sciences-Qiagen Inc., Valencia, CA).

Antibodies, western blot analyses and protein half life determination.

BMI1 was detected either using 1H6B10G7 monoclonal antibody (mAb) (Invitrogen, Carlsbad, CA), or F6 mAb (Millipore (Billerica, MA). For the analysis of AKT pathway, phospho-AKT 1/2/3 (Ser-473) and total AKT-1/2/3 were obtained from Santa Cruz Biotech (Santa Cruz, CA). Antibodies against pRb, phospho-ERK, ERK, p16^INK4a, p21 and α-tubulin were obtained from Santa Cruz Biotech (Santa Cruz, CA). Among other antibodies, Phospho-GSK3β (Ser 9) and GSK antibodies were obtained from Cell Signaling Technologies (Danvers, MA), and β-actin, HA and FLAG mAbs were obtained from Sigma-Aldrich (St. Louis, MO). For immunostaining, a pAb against H2AK119Ub were obtained from Millipore (Billerica, MA). Recombinant protein A- and protein G-sepharose beads were obtained from GE Healthcare Biosciences (Pittsburgh, PA).

When required, MCF10A cells were growth factor deprived using D3 medium for 72 hrs and/or stimulated for 20 min by addition of D medium, which contains 12.5 ng/ml epidermal growth factor (EGF) as described in reference ³⁶. Western blot analyses of total cell extracts from control and wild-type or mutant BMI1 expressing cells, and different combinations of BMI1 and βTrCP were performed using respective antibodies as described in reference ¹⁴. For the analysis of H2A ubiquitination, total histones were prepared using acid extraction method as described in reference ³⁹. The half life of wild-type and mutant BMI1 protein was determined using cycloheximide (CHX) treatment of MCF10A or U2OS cells expressing BMI1 proteins as described in reference ¹⁴.

Immunoprecipitation and immunostaining methods.

For immunoprecipitation (IP) cells were scraped off and lysed in IP buffer (20 mM Tris HCl pH 8, 137 mM NaCl, 10% glycerol, 1% Triton X-100, 2 mM EDTA, 1 mM PMSF, 1 ug/ml Aprotinin, 1 mM NaF, 1 mM sodium orthovanadate, x100 diluted protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO). IPs were performed using specific antibodies and protein A- or Protein G-sepharose as described in reference ³⁶. The immunoprecipitates were analyzed for interacting proteins using western blot analyses as described above. Immunostaining assays were performed as described in reference ³⁶. After staining, cells were photographed (40x) using a Nikon Eclipse 80i confocal microscope.

Senescence, proliferation, invasion and wound healing assays.

Senescence induction or bypass assays in HDFs expressing wild-type or mutant BMI1 proteins was determined using histochemical staining for senescence-associated beta galactosidase (SA-β-gal) marker as described in reference ^{12, 14, 27} and ³⁷. Similarly, proliferation assays were done as described in reference ^{12, 14, 27} and ³⁷. Invasion and migration assays were performed as described in references 8 and 36.