Mechanisms of MCL-1 Protein Stability Induced by MCL-1 Antagonists in B-Cell Malignancies

Abstract

Purpose: Several MCL-1 inhibitors (MCL-1i), including AMG-176 and AZD5991, have shown promise in preclinical studies and are being tested for the treatment of hematologic malignancies. A unique feature of these agents is induction and stability of Mcl-1 protein; however, the precise mechanism is unknown. We aim to study the mechanism of MCL-1i-induced Mcl-1 protein stability.

Experimental design: Using several B-cell leukemia and lymphoma cell lines and primary chronic lymphocytic leukemia (CLL) lymphocytes, we evaluated molecular events associated with Mcl-1 protein stability including protein half-life, reverse-phase protein array, protein-protein interaction, phosphorylation, ubiquitination, and de-ubiquitination, followed by molecular simulation and modeling.

Results: Using both in vivo and in vitro analysis, we demonstrate that MCL-1i-induced Mcl-1 protein stability is predominantly associated with defective Mcl-1 ubiquitination and concurrent apoptosis induction in both cell lines and primary CLL subjects. These MCL1i also induced ERK-mediated Mcl-1Thr163 phosphorylation, which partially contributed to Mcl-1 stability. Disruption of Mcl-1:Noxa interaction followed by Noxa degradation, enhanced Mcl-1 de-ubiquitination by USP9x, and Mule destabilization are the major effects of these inhibitors. However, unlike other BH3 proteins, Mule:Mcl-1 interaction was unaffected by MCL-1i. WP1130, a global deubiquitinase (DUB) inhibitor, abrogated Mcl-1 induction reaffirming a critical role of DUBs in the observed Mcl-1 protein stability. Further, in vitro ubiquitination studies of Mcl-1 showed distinct difference among these inhibitors.

Conclusions: We conclude that MCL-1i blocked Mcl-1 ubiquitination via enhanced de-ubiquitination and dissociation of Mcl-1 from Noxa, Bak and Bax, and Mule de-stabilization. These are critical events associated with increased Mcl-1 protein stability with AMG-176 and AZD5991.

Figure 1.. MCL-1 inhibitors induce Mcl-1 upregulation by increasing the stability of Mcl-1 protein.

A. Reverse Phase Proteomic Array (RPPA) for MEC1 and Mino Cell lines. 5X10⁶ cells were treated with DMSO, 500 nM AMG-176, or 500 nM AZD5991 for 12 hours. Proteins were extracted and analyzed by protein array as described in Materials and Methods. Protein levels were compared between drug-treated conditions and DMSO control from three biological repeat experiments. Unbiased hierarchical clustering was performed to identify proteins with significantly (*p ≤ 0.015) altered expression in MEC1 or in Mino cells (*p ≤ 0.01) after treatment with the inhibitors compared to expression in DMSO-treated cells. Mcl-1 protein level changes with MCL-1 inhibitors represented in box plot as in supplemental data (**p<0.002 in MEC1 or ****p<0.0001 in Mino cells, one-way ANOVA analysis). Protein concentrations were normalized and calculated as described in the Supplemental Methods. MEC1 is red and Mino is blue colored boxes. B. Immunoblot analysis of Mcl-1 protein levels in MEC1 and Mino cell lines treated with DMSO, AMG-176 or AZD5991. Left panel shows time-dependent changes in Mcl-1 protein levels after treatment with DMSO, AMG-176, and AZD5991 (both 1 μM). Right panel shows dose-dependent changes in Mcl-1 protein level following treatment with DMSO or AMG-176 and AZD5991 for 16 hours. C. qRT-PCR analysis of MCL-1 and cMyc mRNA expression at early (2 hours) and late (16 hours) following treatment with AMG-176 and AZD5991 in MEC1 and Mino cell lines and compared to the DMSO control. Each bar represents mean ± SEM (n=3 from separate experiments) and shown as relative to DMSO control. D. MEC1 cells were treated with DMSO, 500 nM AMG-176, or 500 nM AZD5991 for 12 hours followed by the addition of 10 μg cycloheximide (CHX). Cells were collected just before adding CHX and at 1, 2, 4, and 6 hours following CHX addition, immunoblotted, and probed for Mcl-1 (short half-life), HSP90 (long half-life), and BCL-2 (intermediate half-life) protein expression. GAPDH was probed for loading control. Mcl-1 protein expression after incubation with MCL-1 inhibitors for 12 hours was quantitated (left lower). Mcl-1 protein expression over time following CHX treatment was plotted and the rate of degradation of Mcl-1 was determined (right lower). Each bar represents mean ± SEM from n=3, separate experiments; *p<0.05; **p<0.001; ***p<0.0001 significant difference from DMSO control. E. MEC1 and Mino cells were either untreated or treated with DMSO, 500 nM AMG-176, 10 μg CHX, or combination of AMG-176 and CHX for the indicated time points and then processed for immunoblot analyses of Mcl-1, HSP90, and BCL-2. GAPDH was probed for loading control. LI-COR quantitation of Mcl-1 relative to GAPDH was calculated and expressed compared to DMSO control and indicated underneath Mcl-1 blots. AMG: AMG-176; AZD: AZD5991; CHX: Cycloheximide, CL: Cell line; TX, treatment.

Figure 2.. MCL-1 inhibitors induce defective ubiquitination of Mcl-1 protein.

A. Endogenous reverse co-immunoprecipitation (co-IP) analysis of MEC1 cells treated with DMSO, AMG-176 (500 nM), MG-132 (1.25 μM) for 12 hours or a 4-hour washout following AMG treatment (washout). Left panel, input; right panel, equal amounts of total protein (500 μg in each treatment) were subjected to IP with mouse anti-Mcl-1 and probed with rabbit-UB antibody (upper portion) or reverse co-IP with mouse anti-UB and probed with rabbit Mcl-1 antibody (lower portion). B. HEK293 cells were transfected with FLAG-WT-MCL1 and HA-UB (3 μg each in 10-cm plate). Forty-eight hours after transfection, cells were treated with DMSO, AMG (1 μM), or MG-132 (1.5 μM) for 12 hours or a 4-hour washout following AMG-176 treatment. Cells co-transfected with an empty FLAG vector and HA-UB served as vector control. Left panel, input. Right panel, Equal amounts of total protein (500 μg in each treatment) were subjected to IP with FLAG beads (upper portion) or HA beads (lower portion) followed by immunoblot analysis with a rabbit UB antibody or rabbit FLAG respectively. IB: immunoblot; IP: immunoprecipitated; AMG: AMG-176; AZD: AZD5991.

Figure 3.. MCL-1 inhibitors elicit ERK-mediated Mcl-1^Thr163 phosphorylation.

A. MEC1 and Mino cell lines were treated with the indicated concentrations of AMG-176 and AZD5991 for 24 hours and immunoblotted for total and Mcl-1^Thr163 phosphorylation. Vinculin was probed for loading control. Data were analyzed by LI-COR quantitation to determine the amount of total and Mcl-1^Thr163 phosphorylation and their ratios. Each bar represents mean ± SEM from three separate experiments (left graph: MEC1, right graph: Mino). B. Left panel immunoblot analysis of primary CLL lymphocytes showing total and Mcl-1^Thr163 phosphorylation following treatment of lymphocytes with 500 nM AMG-176 or 500 nM AZD5991 for 12 hours. Vinculin was blotted for loading control. Right upper panel: Bar diagram showing quantitation of total and phospho Mcl-1^Thr163 form CLL patients (n=10) immunoblots (*p<0.05; **p<0.001 significant difference from DMSO control); right lower panel: XY graph showing relationship between total and phospho-Mcl-1^Thr163 following treatment with AMG (left graph) or AZD (right graph). C. Immunoblot showing induction of phospho-ERK^{Thr202/Tyr204} in MEC1 cells treated with DMSO or 500 nM AMG-176 for 16 hours. Relative ERK or phospho-ERK was quantitated and plotted showing mean ± SEM (n=3, separate experiments); *p<0.05 significant difference from DMSO control. D. MEC1 and Mino cells were treated with DMSO, 500 nM AMG-176, 500 nM AZD5991, 10 μM trametinib (T-nib), AMG-176 plus T-nib, or AZD5991 plus T-nib for 16 hours. Immunoblot analysis of equal amount of proteins were probed with the indicated antibodies. Vinculin was probed for loading control. Because MCL-1 and GSK3β protein bands are from the same immunoblot for MEC1 cells, identical vinculin blot is shown in rows 3 and 9. Similar data obtained from three biological repeats. E. Immunoblot analysis of total and phospho-Mcl-1^Thr163 and total and phospho-ERK ^{Thr202/Tyr204} expressions from primary CLL samples treated with DMSO, 500 nM AMG-176, 500 nM AZD5991, trametinib (10 μM) alone or in combination, for 12 hours. F. Immunoblot analysis showing HEK293 cell line transiently transfected with WT- or T163A-FLAG-MCL-1. Forty-eight hours following transfection, cells were treated with DMSO, AMG-176 (500nM) or AZD5991 (500nM) for 24 hours. Equal amount of proteins was loaded and probed with indicated antibodies. FLAG expression was measured by LI-COR quantitation and represented in the bar graph (n=3) showing mean ± SEM. AMG: AMG-176; AZD: AZD5991; T-nib: Trametinib. Note. For Figures A-F, to optimally capture total and modified (phosphorylated) protein, we used one immunoblot and probed same protein band with two different species of antibodies. The membranes were imaged using Infrared Odyssey CLx machine. With different species of antibodies, we can optimally quantitate data because these are read at different wavelengths. This spectrally distinct fluorophores technique (for example, 700 nm (red) or 800 nm (green) channels) allows us to do multiplexing on the same protein band. Because immunogenicity of 2 separate antibodies are tested in the same protein blot, bands have similar appearance and features for each protein in both total and phospho-bands.

Figure 4.. MCL-1 inhibitors induced a transient decrease in Mule and Noxa expression.

A. Upper panel: Immunoblot for MEC1 cell line treated with DMSO, AMG-176 (1μM) or AZD5991 (1μM) for the indicated time points. Blot was probed with different E3 ligases (Mule, Trim-17–1, 17–2, β-TRCP) and Noxa. Lower panel: bar graph showing relative Mule expression (*p<0.05 significant difference from DMSO control). B. Upper panel: Immunoblot for MEC1 and Mino cell lines were treated with DMSO. AMG-176 (1μM) or AZD5991(1μM) for 16 hours. Equal amount of protein (500μg) was IP with rabbit Mule followed by IB of mouse Mcl-1 antibody. Lower panel, showing input. GAPDH was used for loading control. C. Left panel: Immunoblot for MEC1 cell line treated with DMSO, AMG-176 (500nM), AMG-176 (500nM) followed by wash out for four hours, or MG-132(1.5 μM) for 16 hours. Equal amount of proteins (500 μg) were immunoprecipitated by either mouse Mcl-1 antibody and probed with Noxa and Bax antibodies or immunoprecipitated with mouse BAK antibody and probed with Mcl-1 antibody respectively. Middle panel: showing input. Vinculin and GAPDH was used for loading controls. Right panel: HEK293 cells were co-transfected with HA-Ub and FLAG-MCL-1 for 48 hours followed by treatment by either DMSO, AMG-176, or AMG-176 washout (upper right panel) for 4 hours or MG-132 (1.5 μM) for 16 hours. Equal amount of proteins (500 μg) were immunoprecipitated by FLAG beads and probed with BAX. Bottom right panel shows input for FLAG. Vinculin was used for loading control. D. Immunoblot showing in vitro pull-down assay of His-tag recombinant Mcl-1 with either GST-tagged recombinant Mule (upper panel) or endogenous full-length Mule immunoprecipitated lysate from untreated HEK293 (lower panel) as described in methods section. Reaction was made for one hour in the presence of DMSO, AMG-176 (10mM) or AZD5991 (10mM) in a total volume of 1ml. A control sample without Mcl-1 protein was added. Samples were immunoblotted for Mule and Mcl-1. E. Immunoblot assay following in vitro ubiquitination assay of His-tag recombinant Mcl-1 protein in the presence of absence of Mule. Samples were untreated or treated with DMSO, AMG-176 (10mM), AZD5991 (10mM) or maritoclax (10mM) and incubated for two hours in ubiquitination buffer containing FLAG-tagged ubiquitin as described in method section. Samples were immunoblotted and probed with either anti-FLAG antibody to detect ubiquitination (upper panel) or rabbit Mcl-1 (lower panel). F. Immunoblot analysis for MEC1 cell line treated with either DMSO, AMG-176 (500nM), or AZD5991 (500nM) for 24 hours and probed with different BH3 proteins. Vinculin and GAPDH were probed for loading control. AMG: AMG-176; AZD: AZD5991.

Figure 5.. MCL-1 inhibitors target deubiquitinases for enhanced Mcl-1 de-ubiquitination

A. Immunoblot analysis for MEC1 cells treated with either DMSO, AMG-176 (1μM) or AZD5991 (1μM) for the indicated time points (upper panel). Bar graph showing relative USP9x expression (n=3; lower panel). B. Upper panel showing immunoblot analysis of MEC1 cell line treated with either DMSO, AMG-176 (1μM), or AZD5991 (1μM) for 16 hours. Equal amount of proteins (500μg) were immunoprecipitated with mouse Mcl-1 antibody and immunoblotted with rabbit USP9x and Mcl-1. Lower panel showing input control. GAPDH was used for loading control. C. Immunoblot analysis showing in vitro de-ubiquitination assay of purified full-length HA-Ub-FLAG-Mcl-1 protein by purified USP9x in the presence of either DMSO, AMG-176 (10mM), or AZD5991 (10mM) as described in method section followed by probing with anti-HA to detect FLAG-Mcl-1 ubiquitination. Relative ratio of AMG and AZD-induced Mcl-1 ubiquitination compared to DMSO control is shown underneath. D. Line graph showing the Relative Fluorescence Unit (RFU) of recombinant USP9x after incubation with either DMSO, AMG (1.25mM) or AZD (1.25mM) in DUB buffer containing the fluorogenic Ubiquitin-AMC substrate in 96 well plate as described in method section. E. Immunoblot analysis of MEC1 and Mino cell lines treated with DMSO, AMG-176 (1μM), AZD5991(1μM), WP1130 (5μM), either alone or in combination with AMG-176 or AZD5991 for six hours. All samples were also pre-treated by the pan-caspase inhibitor QVD (20μM) to inhibit apoptosis. F. Immunoblot analysis from primary CLL patient sample either treated with DMSO, AMG-176 (500nM), AZD5991 (500nM), WP1130 (10μM), either alone or in combination with AMG-176 or AZD5991 for six hours. All lanes are pre-treated by the pan-caspase inhibitor, QVD, (20μM), to inhibit apoptosis. G. Left panel; Immunoblot analysis of WT and USP9x knockout HCT-116 cell line treated with DMSO, AMG-176 (1μM) or AZD5991 (1μM) for 16 hours and probed with the indicated antibodies. Right panel; Immunoblot analysis of untreated WT and USP9X knockout HCT-116 cell line and probed with the indicated antibodies. H. Immunoblot of MEC1-shRNA-USP9x knockdown treated by DMSO, AMG-176 (1μM) or AZD5991 (1μM) for 16 hours and probed with the indicated antibodies. All lanes are pre-treated by the pan-caspase inhibitor QVD (20μM) to inhibit apoptosis. AMG: AMG-176; AZD: AZD5991.

Figure 6.. Conformation change upon inhibitor binding altered post-translation modification, interaction and stability in the Mcl-1 protein.

A. Mcl-1 apo structure after 10-ns molecular dynamics simulation. Left panel, binding efficacy (solvent-accessible surface area; SASA) without an MCL-1 inhibitor; right panel, complex of AMG-176 and Mcl-1 protein. Pink ribbon, Mcl-1; Blue, Ser¹⁵⁹ residue; Green, Thr¹⁶³; Cyan, AMG-176. B. Binding site of AMG-176 and AZD5991, Cyan: AMG176-Mcl-1 complex structure (PDB: 6OQC); Green: AZD5991 (PDB: 6FS0). C. Binding site for AMG-176 and BAX-BAK BH3 domains: Orange: BAXBH3 domain (PDB:3PK1); Yellow: BAKBH3 domain (PDB: 5FMK); Cyan: Mcl1; Magenta: AMG-176 (6OQC). D. Structural analysis of MCL1-Mule interactions (PDB: 5C6H). The Mule peptide (grey) binds to MCL1 (cyan) in the same region as BaxBH3, also partially overlapped with inhibitor binding (AMG-176). E. Left panel, MD simulations showed that the inhibitor binding can cause large conformational change of Mcl-1 protein and make the N-terminus more extended and less packed. Orange cartoon: USP9x; cyan cartoon: holo-MCL1; magenta: AMG-176. Right panel, Zoom-in view of MCL1-USP9x interactions. Several residues in Mcl-1 (cyan) including Ser155, Ser159 and Thr163 (cyan sticks) are spaced out and more exposed (see SASA data), thus prone to phosphorylation. Once phosphorylation occurs, the increased negative charges (phosphate groups) will enhance their binding to USP9x (orange) by interacting with several positively charged residues including Arg1936, Arg1940, Arg1941, and Lys1943 on USP9x. F. Schema summary showing mechanisms of MCL-1i induced stability of Mcl-1 protein. Anti- and pro-apoptotic proteins (Mcl-1, Bax, Bak, Noxa) are indicated by circles. Ubiquitinase and deubiquitinase proteins are oval shaped. Dissociation of Mcl-1 from BAX/BAK may be a trigger for apoptosis induction. Noxa disruption and degradation (e.g. AMG and AZD), along with MCL-1i induced conformation that does not favor ubiquitination and enhances de-ubiquitination by USP9x (e.g. AZD) can stabilize MCL-1. ERK phosphorylates Thr-163 on Mcl-1 which is facilitated by conformation change. Phosphorylated Mcl-1 is more stable. Mule adds ubiquitins on Mcl-1 protein however increased USP9x DUB activity removes these ubiquitins, increasing Mcl-1 protein stability. SASA: Solvent-Accessible Surface Area.