Mitotic MELK-eIF4B signaling controls protein synthesis and tumor cell survival

Significance

The work identifies the eukaryotic translation initiation factor 4B (eIF4B) as a substrate of maternal and embryonic leucine zipper kinase (MELK), a mitotic kinase known to be essential for aggressive types of malignancy. The MELK–eIF4B axis thus represents a previously unidentified signaling pathway that regulates protein synthesis during mitosis and, consequently, the survival of cancer cells.

Abstract

The protein kinase maternal and embryonic leucine zipper kinase (MELK) is critical for mitotic progression of cancer cells; however, its mechanisms of action remain largely unknown. By combined approaches of immunoprecipitation/mass spectrometry and peptide library profiling, we identified the eukaryotic translation initiation factor 4B (eIF4B) as a MELK-interacting protein during mitosis and a bona fide substrate of MELK. MELK phosphorylates eIF4B at Ser406, a modification found to be most robust in the mitotic phase of the cell cycle. We further show that the MELK–eIF4B signaling axis regulates protein synthesis during mitosis. Specifically, synthesis of myeloid cell leukemia 1 (MCL1), an antiapoptotic protein known to play a role in cancer cell survival during cell division, depends on the function of MELK-elF4B. Inactivation of MELK or eIF4B results in reduced protein synthesis of MCL1, which, in turn, induces apoptotic cell death of cancer cells. Our study thus defines a MELK–eIF4B signaling axis that regulates protein synthesis during mitosis, and consequently influences cancer cell survival.

Maternal and embryonic leucine zipper kinase (MELK) is a serine/threonine kinase with potential roles in mitosis. Similar to other established mitotic factors, such as Aurora kinases and cyclin B1, MELK demonstrates increased protein abundance during mitosis and is degraded when cells progress into G1 phase (1, 2). Our recent study proposed an essential role of MELK in the mitotic progression of specific cancer cell types, with MELK knockdown resulting in multiple mitotic defects, including G2/M arrest and mitotic cell death (2). Despite these advances, there is a lack of mechanistic understanding of the role of MELK during cell division. An immediate question is the identity of the MELK substrates that mediate its role in mitosis, such as promoting mitotic cell survival.

Myeloid cell leukemia 1 (MCL1) is an important negative regulator of apoptosis. Uniquely among the Bcl-2 family, it is turned over rapidly by ubiquitin-mediated proteolysis and must be continuously resupplied by translation (3, 4). Protein abundance of MCL1 decreases during prolonged mitotic arrest induced by antimicrotubule drugs, rendering arrested cells highly sensitive to inhibitors of other Bcl-2 family members (5). We thus speculate that synthesis of MCL1 is important for cell survival during normal and drug-arrested mitosis and, conversely, that a drug that decreases MCL1 synthesis during mitosis might have anticancer potential.

The rate of protein synthesis and other basic biological processes, such as DNA or RNA biogenesis, fluctuates throughout the cell cycle. Overall protein synthesis significantly decreases when cells enter mitosis (6, 7), consistent with the notion that macromolecule synthesis predominantly occurs in interphase before the segregation of cellular mass in mitosis.

A recent study based on ribosome profiling identified a set of genes that are translationally repressed in mitosis, and proposed that suppressed protein synthesis might provide a unique mechanism to complement the posttranslational inactivation of specific proteins (8). Nevertheless, protein synthesis still occurs during mitosis, although at an overall rate that is 30–65% of the overall rate in interphase cells (6, 8, 9). Moreover, the translation of certain mRNAs, such as c-Myc, MCL1, and ornithine decarboxylase (ODC), is even elevated during mitosis (8, 9). Together, these studies suggest that protein synthesis may be functionally important for mitotic cells and might be finely regulated by as yet unidentified signaling pathways.

In this study, we aimed to identify downstream effectors of the mitotic kinase MELK. Intriguingly, we found that MELK phosphorylates the eukaryotic translation initiation factor 4B (eIF4B). Our data implicate the MELK–eIF4B pathway as a previously unrecognized signaling mechanism regulating mitotic protein synthesis and tumor cell survival.

Results

Peptide Library Screen Identifies the Optimal Substrate Motif for MELK.

To identify kinase substrates of MELK that mediate its role in the regulation of mitosis and cell survival, we determined the consensus phosphorylation motif for MELK. Proteins carrying the optimal substrate motif are likely phosphorylated by MELK, and thus represent potential in vivo substrates. We expressed active full-length human MELK in insect cells and subjected the purified kinase to positional scanning peptide library screening, a technique that has been used extensively to identify optimal substrate motifs for kinases (10, 11). The profiling demonstrated that MELK is highly selective for its substrate, with a strong preference for arginine at the −3 position relative to the phosphoacceptor site (Fig. 1A). In addition, arginine was positively selected in the −2 and −4 positions, and hydrophobic residues were strongly selected against in the −3 position (Fig. 1A).

Fig. 1.

Consensus phosphorylation motif for MELK. (A) Positional scanning peptide library (PSPL) screen identified the optimal phosphorylation motif for MELK. (Top) Phosphor imaging of a PSPL screen after in vitro reaction with recombinant full-length MELK and radiolabeled ATP. Each peptide contains one fixed residue at one of nine positions relative to the centrally fixed phosphoacceptor (serine or threonine). Reactions were spotted onto a membrane and exposed to a phosphor storage screen. (Bottom) Sequence motif generated using quantified and normalized data from the above screen. (B) Immunoprecipitation (IP)/mass spectrometric (MS) assay identified MELK–eIF4B interaction during mitosis. MDA-MB-468 cells stably transduced with doxycycline (Dox)-inducible Flag-tagged MELK were left untreated or treated with Dox (100 ng/mL) for 3 d, followed by overnight treatment with nocodazole. Mitotic cells were harvested for IP using anti-Flag magnetic beads. The bound fractions were eluted and subjected to tandem mass spectrometric (MS/MS) analysis. (Left) Chart indicates the recovered peptide reads. (Right) Immunoblots show the interaction between eIF4B and Flag-MELK in mitotic lysates. (C) S406 and the upstream (−3) arginine are highly conserved across evolution. (D) EIF4B phosphorylation and Akt/MAPK signaling during cell cycling. MDA-MB-468 cells were either untreated [asynchronized (Async.)] or treated with nocodazole overnight to obtain floating mitotic cells (M) and attached cells that are enriched for cells in G2 phase. A sample of the mitotic cells was washed and incubated with nocodazole-free medium for 4 h to release cells into G1 phase. Cells were subjected to immunoblot analysis as indicated.

Because bioinformatics analysis revealed a large number of proteins carrying the optimal motif, we sought another approach to narrow down the hits. We reasoned that proteins that physically interact with MELK during mitosis might represent potential substrates. To identify such proteins, we stably introduced doxycycline-inducible Flag-tagged MELK into the breast cancer cell line MDA-MB-468 and immunoprecipitated Flag-MELK from lysates of mitotic cells obtained by nocodazole-induced cell cycle arrest at prometaphase. Tandem mass spectrometric analysis identified high enrichment of MELK together with eIF4B (Fig. 1B and Table S1), a translation initiation factor with roles in regulating cell survival (12). The MELK–eIF4B interaction was further confirmed by ectopic expression of MELK or eIF4B in HEK293T cells (Fig. S1A).

Table S1.

Peptides recovered from Flag-MELK IP

Total peptides	Total peptides	Matched protein
84	166	LRPPRC_IPI:IPI00783271.1
78	151	FLNA_IPI:IPI00302592.2
47	108	MELK_IPI:IPI00006471.3
35	91	EIF4B_IPI:IPI00012079.3
33	37	MYH9_IPI:IPI00019502.3
32	48	ACAP2_IPI:IPI00014264.5
29	49	HSPA5_IPI:IPI00003362.2
29	76	STK38_IPI:IPI00027251.1
27	28	ROCK1_IPI:IPI00022542.1
26	55	PRMT5_IPI:IPI00064328.3
26	73	LANCL2_IPI:IPI00032995.1
24	42	HSPA8_IPI:IPI00003865.1
24	34	HSPA9_IPI:IPI00007765.5
23	29	DHX15_IPI:IPI00396435.3
22	23	CCT8_IPI:IPI00302925.4
21	32	ANKFY1_IPI:IPI00159899.9
19	26	CCT2_IPI:IPI00297779.7
19	24	GLUD1_IPI:IPI00016801.1
18	22	TCP1_IPI:IPI00290566.1
18	23	CCT3_IPI:IPI00290770.3
18	20	DDX42_IPI:IPI00409671.3
18	18	LMNA_IPI:IPI00021405.3
18	18	THBS1_IPI:IPI00296099.6
17	23	TUFM_IPI:IPI00027107.5
17	23	EGFR_IPI:IPI00018274.1
16	21	VCP_IPI:IPI00022774.3
16	20	HSP90B1_IPI:IPI00027230.3
15	19	SERPINH1_IPI:IPI00032140.4
15	19	CCT7_IPI:IPI00018465.1
15	15	SUPT16H_IPI:IPI00026970.4
14	23	ILK-2_IPI:IPI00302927.6
14	16	CCT5_IPI:IPI00010720.1
14	15	PABPC1_IPI:IPI00008524.1
13	19	HSP90AA1_IPI:IPI00382470.3
13	17	RBM10_IPI:IPI00375731.1
12	16	CTTN_IPI:IPI00029601.6
12	24	C11orf84_IPI:IPI00106955.3
12	18	PDIA6_IPI:IPI00299571.5
12	12	ATP5A1_IPI:IPI00440493.2
12	12	MTHFD1L_IPI:IPI00291646.3
12	16	GSTK1_IPI:IPI00219673.6
12	12	ATP5B_IPI:IPI00303476.1
12	17	HSP90AB1_IPI:IPI00414676.6
12	13	PRKDC_IPI:IPI00296337.2
11	14	HTATSF1_IPI:IPI00013788.1
11	12	SMEK1_IPI:IPI00017290.4
11	11	HSPD1_IPI:IPI00784154.1
11	14	STK38L_IPI:IPI00237011.5
11	12	RPS3_IPI:IPI00011253.3
11	14	PPIB_IPI:IPI00646304.4
11	16	TRIM21_IPI:IPI00018971.8
11	12	MCM5_IPI:IPI00018350.3
10	11	SYNCRIP_IPI:IPI00018140.3
10	19	TUBB2C_IPI:IPI00007752.1
10	10	SNRNP200_IPI:IPI00420014.2
10	15	SMEK1_IPI:IPI00217013.2
10	11	CCT6A_IPI:IPI00027626.3
10	10	HDLBP_IPI:IPI00022228.2
10	10	HADHA_IPI:IPI00031522.2
10	11	SQRDL_IPI:IPI00009634.1
10	14	TUBA4A_IPI:IPI00007750.1
10	10	RPS9_IPI:IPI00221088.5

Fig. S1.

eIF4B phosphorylation in mitotic cells. (A) MELK interacts with eIF4B. HEK293T cells were transfected with Flag-tagged MELK (Left) or HA-tagged eIF4B (Right). At 36 h after transfection, cell lysates were prepared and incubated with magnetic beads conjugated with anti-Flag (Left), anti-HA (Right), or control IgG. Samples were analyzed by immunoblotting with anti-MELK or anti-eIF4B antibodies. (B) S422 and upstream (−3) arginine are highly conserved throughout evolution. (C) Phosphorylation of eIF4B and Akt/MAPK during the cell cycle. MDA-MB-468 cells were either left untreated [asynchronized (Async.)] or treated with paclitaxel (100 nM) for 18 h. Mitotic cells (M) were isolated by shake-off, and the attached cells enriched for G2 phase were harvested. A sample of the M was washed and released into G1 phase after 4 h of incubation in the absence of paclitaxel. Cell lysates were prepared in radioimmunoassay precipitation (RIPA) buffer and subjected to immunoblotting using the indicated antibodies.

We proceeded to examine whether eIF4B carries the optimal substrate motif for MELK. Two serine residues of eIF4B, Ser406 and Ser422, conformed to the consensus phosphorylation sequence (Fig. 1C and Fig. S1B). In addition, S406/S422 and their flanking residues are evolutionarily conserved across vertebrates (Fig. 1C and Fig. S1B), indicating that these residues might be subject to posttranslational modifications. Together, these findings suggest that eIF4B might be a protein substrate of MELK.

EIF4B Phosphorylation During Mitosis.

Given that MELK is a mitotic kinase, its potential role in regulating eIF4B phosphorylation drove us to investigate first how eIF4B phosphorylation occurs at different stages of the cell cycle. We harvested mitotic MDA-MB-468 cells at prometaphase through nocodazole- or paclitaxel-induced cell cycle arrest. As expected, the cells were enriched for mitotic factors, including MELK, cyclin B1, and Aurora kinase A, and, interestingly, exhibited strong phosphorylation of eIF4B at Ser406, but not at Ser422 (Fig. 1D and Fig. S1C). In contrast, phosphorylation of some of the known upstream signaling molecules, such as Akt or MAPK (13), was apparently decreased in mitotic cells (Fig. 1D and Fig. S1C).

MELK Directly Phosphorylates eIF4B at Ser406.

To investigate whether MELK phosphorylates eIF4B, we first performed in vitro kinase assays using purified recombinant MELK and eIF4B. MELK potently induced phosphorylation of eIF4B at Ser406, and to a lesser extent at Ser422, as indicated by immunoblot analysis with phospho-specific antibodies (Fig. 2A). We also examined the effect of chemical inhibition of MELK kinase activity using OTSSP167 (14), and found that the inhibitor efficiently suppressed MELK-driven eIF4B phosphorylation at Ser406 in a dose-dependent manner (Fig. 2A). To confirm that MELK is a bona fide kinase for eIF4B, we performed kinase reactions consisting of MELK (full-length or kinase domain) together with eIF4B (wild type or with mutations introduced at Ser406 or Ser422). The immunoprecipitated eIF4B was readily phosphorylated at Ser406 by full-length MELK or its kinase domain (Fig. 2B), and, as expected, the phosphorylation was abolished when serine was mutated to alanine (S406A) (Fig. 2B). In contrast, MELK was unable to catalyze the phosphorylation of eIF4B at S422 (Fig. S2A).

Fig. 2.

MELK regulates the phosphorylation of eIF4B. (A) MELK phosphorylates recombinant eIF4B selectively at Ser406. Recombinant GST-eIF4B (240 ng per reaction) was incubated without [negative control (Neg. Con), first lane] or with (1 μg per reaction) MELK kinase in the presence of OTSSP167 (0.64–10,000 nM; DMSO was used as a vehicle control, second lane). ATP was used at a final concentration of 300 μM. After incubation for 30 min at room temperature, the reaction was terminated and the samples were subjected to immunoblotting. Note that anti–p-eIF4B (S4222) also recognizes MELK. The relatively high concentration of OTSSP167 needed to suppress eIF4B phosphorylation is likely due to the high concentration of MELK protein (∼500 nM) used in the assay. (B) MELK phosphorylates immunoprecipitated eIF4B at Ser406 in vitro. Recombinant full-length or kinase domain of MELK was subjected to in vitro kinase assay using immunoprecipitated Flag-eIF4B [wild type (wt)] or Flag-eIF4B (S406A). Reactions were analyzed by immunoblotting. (C) MELK knockdown inhibits eIF4B phosphorylation at S406. MDA-MB-468 cells stably transduced with Dox-inducible short hairpin MELK (shMELK) were left untreated or treated with Dox. Mitotic cells were harvested, and lysates were subjected to immunoblotting. (D) Submicromolar doses of OTSSP167 inhibit eIF4B phosphorylation in mitotic cells. Mitotic MDA-MB-468 cells were harvested by nocodazole-induced prometaphase arrest and treated for 30 min with increasing concentrations of OTSSP167 in the presence of nocodazole (200 ng/mL) and MG132 (10 μM). Lysates were subjected to immunoblotting. Note that inhibition of eIF4B phosphorylation correlates with altered electrophoretic mobility of MELK. (E) OTSSP167 inhibits eIF4B phosphorylation in vivo. Mice with breast cancer cell xenografts were treated with vehicle (0.5% methylcellulose) or OTSSP167 (5 mg/kg) for 13 d. Tumors were isolated and radioimmunoprecipitation assay (RIPA) lysates were subjected to immunoblotting. Note that phosphorylation of eIF4B, but not Akt or MAPK, was inhibited by OTSSP167. Each lane represents one independent tumor sample.

Fig. S2.

MELK phosphorylates eIF4B specifically at Ser406. (A) MELK does not phosphorylate eIF4B at S422 in vitro. Recombinant full-length or kinase domain of MELK was subjected to in vitro kinase assay using immunoprecipitated Flag-eIF4B [wild type (wt)] or Flag-eIF4B (S422A). Reactions were analyzed by immunoblotting using the indicated antibodies. (B) Doxycycline (Dox)-inducible knockdown of MELK decreases mitotic phosphorylation of eIF4B at S406. BT549 cells stably transduced with tet-on-shMELK were either left untreated or treated with Dox (100 ng/mL). After 3 d, the cells were treated with nocodazole (200 ng/mL) for 20 h. Mitotic cells were harvested by shake-off. Lysates were prepared in RIPA buffer and subjected to immunoblotting. (C) MELK inhibition impairs eIF4B phosphorylation at S406 in nocodazole or paclitaxel-arrested mitotic cells. Indicated cell lines were prepared and treated with OTSSP167 at different concentrations for 30 min. Cell lysates were subjected to immunoblotting. (D) Asynchronized cells were treated with OTSSP167 at indicated concentrations for 30 min, and cell lysates were prepared for immunoblotting. Note that a high dose of OTSSP167 (100 nM) causes significant suppression of eIF4B phosphorylation. Con, control.

To test whether eIF4B phosphorylation in mitotic cells is regulated by MELK, we used a doxycycline-inducible shRNA platform to knock down MELK expression (15) and found that Ser406 phosphorylation of eIF4B in mitotic cells (MDA-MB-468, BT549) was strongly suppressed by the loss of MELK (Fig. 2C and Fig. S2B). Moreover, the MELK inhibitor OTSSP167 reduced eIF4B phosphorylation in mitotic cells at concentrations in the nanomolar range. Interestingly, OTSSP167 concurrently increased the electrophoretic mobility of MELK (Fig. 2D and Fig. S2C), suggesting possible dephosphorylation and inactivation of MELK. Together, data from in vitro and in vivo assays point to a predominant role of MELK in phosphorylating eIF4B at Ser406 in mitotic cells.

Both MELK expression and eIF4B phosphorylation, although most abundant during mitosis, were also observed in interphase, such as in cells in G1 phase (Fig. 1D), suggesting that MELK might also contribute to eIF4B phosphorylation outside of mitosis. Indeed, MELK inhibition in asynchronized cells (MDA-MB-468 and MDA-MB-231) caused a significant, albeit reduced, inhibition of eIF4B phosphorylation (Fig. S2D). In addition, treatment of mice bearing xenografted MDA-MB-231 tumors with OTSSP167 resulted in reduced phosphorylation of eIF4B at Ser406 in the absence of suppressed Akt/MAPK signaling (Fig. 2E). These observations indicate that the role of MELK in phosphorylating eIF4B is most prominent in mitosis but might extend to other cell cycle stages.

Chemical Library Screen Identifies MELK as the Unique Mitotic Kinase Phosphorylating eIF4B.

We next investigated whether MELK is the only kinase responsible for eIF4B phosphorylation during mitosis. A large number of kinases are expressed in mitotic cells, including Aurora kinases, CDK1, MPS1, and PLK1. Recent studies also documented hyperactivation of mTOR complex 1 during mitosis despite the fact that the protein abundance of mTOR does not fluctuate during the cell cycle, unlike the established mitotic kinases (7). We first tested whether inhibition of Aurora kinases affects eIF4B phosphorylation. Both MLN8045 and VX-680 (16, 17) suppressed the autophosphorylation of Aurora A and Aurora B (at a high dose for MLN8045) but failed to decrease, and in fact even enhanced, eIF4B phosphorylation (Fig. 3 A and B). Similarly, the highly potent ATP-competitive inhibitor of mTOR Torin 1 (18) was also unable to inhibit eIF4B phosphorylation (Fig. 3C).

Fig. 3.

MELK-specific regulation of eIF4B phosphorylation during mitosis. (A and B) Insensitivity of eIF4B phosphorylation (S406) to inhibition of Aurora kinases. Mitotic cells were harvested by shake-off following overnight treatment with nocodazole (200 ng/mL) and treated with increasing concentrations of MLN8045 (A) or VX-680 (B) for 30 min in the presence of nocodazole and MG132 (10 μM). Cells were harvested for immunoblotting. Con, control. (C) Insensitivity of eIF4B phosphorylation (S406) to mTOR inhibition. The assay was performed as in A and B, but in the presence of the mTOR inhibitor Torin 1. (D) Schematic diagram of the chemical library screen for small molecules that inhibit S406 phosphorylation of eIF4B. Mitotic cells were harvested by shake-off after treatment with nocodazole and exposed to 227 small inhibitors individually at a final concentration of 5 μM. RIPA lysates were subjected to fluorescent immunoblotting with anti–p-eIF4B (S406) and antivinculin (loading control) antibodies. (E) Normalized signals of eIF4B phosphorylation from the chemical library screen. Note that OTSSP167 was the strongest inhibitor of eIF4B phosphorylation among all of the chemicals tested.

To study systemically whether other kinases besides MELK are involved in regulating mitotic phosphorylation of eIF4B, we screened a chemical library for any compounds capable of suppressing eIF4B signaling. The Library of Integrated Network-Based Cellular Signatures (LINCS) project has collected and characterized more than 200 kinase inhibitors, most of which are commercially available (https://lincs.hms.harvard.edu/). We screened this panel at a single dose of 5 μM by incubation with nocodazole-arrested mitotic cells (MDA-MB-231) for 30 min, followed by immunoblot analysis of phospho-eIF4B (Ser406) (Fig. 3D). Of the 227 kinase inhibitors tested, only OTSSP167 showed robust inhibition of mitotic eIF4B phosphorylation at Ser406 (Fig. 3E and Table S2). Therefore, we concluded that MELK inhibition uniquely induced potent inhibition of mitotic phosphorylation of eIF4B.

Table S2.

Chemical library screen for inhibitors suppressing eIF4B phosphorylation

SM HMS LINCS ID	SM name	Alternative names	LINCS ID	PubChem CID	Normalized p-eIF4B (S406)/vinculin
10001-101	(R)-Roscovitine	CYC202; Seliciclib	LSM-1001	160355	1.197703546
10002-101	ALW-II-38-3		LSM-1002	24880028	1.796850516
10003-101	ALW-II-49-7		LSM-1003	24875320	1.784969631
10004-101	AT-7519		LSM-1004	11338033	1.601992176
10005-101	Tivozanib	AV-951	LSM-1005	9911830	2.038550714
10006-101	AZD7762		LSM-1006	67077825	2.063442535
10007-101	AZD8055		LSM-1007	25262965	1.902014416
10008-101	Sorafenib	BAY-439006	LSM-1008	216239	2.875109974
10009-101	CP466722		LSM-1009	44551660	2.121151961
10010-101	CP724714		LSM-1010	53401057	2.50132975
10011-101	Flavopiridol	Alvocidib; HMR-1275; L868275	LSM-1011	5287969	2.12465869
10012-101	GSK429286A		LSM-1012	11373846	1.556566278
10013-101	GSK461364		LSM-1013	15983966	0.509226884
10014-101	GW843682		LSM-1014	9826308	1.069507822
10015-101	HG-5-113-01		LSM-5982	25226483	2.717613168
10016-101	HG-5-88-01		LSM-5245	25226117	1.654874065
10017-101	HG-6-64-01	KIN001-206	LSM-1017		2.691833536
10018-101	Neratinib	HKI-272	LSM-1018	53398697	1.056430224
10019-101	JW-7-24-1		LSM-5674	69923936	1.308477581
10020-101	Dasatinib	BMS-354825; Sprycel	LSM-1020	3062316	1.447588823
10021-101	Tozasertib	VX680; MK-0457	LSM-1021	5494449	2.403164247
10022-101	GNF2		LSM-1022	5311510	1.533458882
10023-103	Imatinib	Gleevec; Glivec; CGP-57148B; STI-571	LSM-1023	5291	0.507833134
10024-101	NVP-TAE684		LSM-1024	16038120	0.650864909
10025-101	CGP60474	MLS000911536; SMR000463552	LSM-1025	644215	1.143196392
10026-101	PD173074		LSM-1026	1401	1.349794193
10027-101	Crizotinib	PF02341066	LSM-1027	11626560	2.35849665
10028-102	BMS345541		LSM-1028	9813758	2.284241161
10029-101	GW-5074		LSM-1029	5034	2.933896073
10030-101	KIN001-042		LSM-1030	6539732	2.034373016
10031-101	KIN001-043		LSM-1031	24906282	2.277225052
10032-101	Saracatinib	AZD0530	LSM-1032	10302451	1.909534768
10033-101	KIN001-055		LSM-1033	3796	1.713826382
10034-101	AS601245	JNK inhibitor V	LSM-1034	10109823	1.603949462
10035-101	Sigma A6730	KIN001-102; AKT inhibitor VIII; Akt1/2 kinase inhibitor	LSM-1035	10196499	1.236566245
10036-101	SB 239063		LSM-1036	5166	0.923668654
10037-101	AC220		LSM-1037	24889392	0.66036709
10038-101	WH-4-023		LSM-1038		0.705242221
10039-101	WH-4-025		LSM-1039		0.878242979
10040-101	R406		LSM-1040	11213558	1.290023399
10041-101	BI-2536	NPK33-1-98-1	LSM-1041	11364421	0.933762381
10042-101	Motesanib	AMG706	LSM-1042	11667893	0.982748388
10045-101	A443654		LSM-1045	10172943	0.33916124
10046-101	SB590885		LSM-1046	53239990	1.088672775
10047-101	GDC-0941		LSM-1047	17755052	1.260658806
10048-101	PD184352	CI-1040	LSM-1048	6918454	1.16452136
10049-101	PLX-4720		LSM-1049	24180719	1.176273559
10050-101	AZ-628		LSM-1050	11676786	1.04927208
10051-104	Lapatinib	GW-572016; Tykerb	LSM-1051	208908	0.429396894
10052-101	Rapamycin	Sirolimus	LSM-1052	5040	0.402750239
10053-101	ZSTK474		LSM-1053	11647372	0.41805056
10054-101	AS605240		LSM-1054	24906273	0.732759892
10055-101	BX-912		LSM-1055	11754511	2.605846607
10056-101	Selumetinib	AZD6244; Array142886	LSM-1056	10127622	0.553955666
10057-102	MK2206		LSM-1057	24964624	1.504894522
10058-101	CG-930	JNK930	LSM-1058	57394468	4.007666972
10060-101	TAK-715		LSM-1060	9952773	1.613382403
10061-101	NU7441	KU 57788	LSM-1061	11327430	1.569151414
10062-101	GSK1070916	KIN001-216	LSM-1062	46885626	2.337479347
10063-101	OSI-027		LSM-1063	56965966	1.942365366
10064-101	WYE-125132		LSM-1064	25260757	1.462093205
10065-101	KIN001-220	Genentech 10	LSM-1065	44139710	2.178706316
10066-101	MLN8054		LSM-1066	11712649	2.270808355
10067-101	Barasertib	AZD1152-HQPA	LSM-1067	16007391	2.241656229
10068-101	PLX4032	RG7204; R7204; RO5185426	LSM-1068	42611257	3.07767941
10069-101	Enzastaurin	LY317615	LSM-1069	176167	1.655733221
10070-101	NPK76-II-72-1		LSM-1070	46843648	3.842016417
10071-101	PD0332991		LSM-1071	5330286	1.627198217
10072-101	PF562271	KIN001-205	LSM-1072	11713159	3.899881066
10073-101	PHA-793887		LSM-1073	46191454	2.17518565
10074-101	KU55933		LSM-1074	5278396	1.21796861
10075-101	QL-X-138		LSM-5803		1.752769285
10076-101	QL-XI-92		LSM-5394		1.035369776
10077-101	QL-XII-47		LSM-6019	71748056	0.922741288
10078-101	THZ-2-98-01		LSM-1078		1.14616835
10079-101	Torin1		LSM-1079	49836027	0.878415444
10080-101	Torin2		LSM-1080	51358113	1.055414012
10081-101	KIN001-244		LSM-1081	49766501	0.81807548
10082-101	WZ-4-145		LSM-1082		0.91286562
10083-101	WZ-7043		LSM-5749		0.926100059
10084-101	WZ3105		LSM-5970	42628507	0.94546971
10085-101	WZ4002		LSM-1085	44607530	1.099049369
10086-101	XMD11-50	LRRK2-in-1	LSM-1086	46843906	1.018094245
10087-101	XMD11-85h		LSM-1087		1.101511287
10088-101	XMD13-2		LSM-1088		1.361331
10089-101	XMD14-99		LSM-6297		1.560871453
10090-101	XMD15-27		LSM-1090		1.528830129
10091-101	XMD16-144		LSM-1091		1.299223341
10092-101	JWE-035		LSM-1092	57340671	1.125954857
10093-101	XMD8-85		LSM-1093	46844147	0.701433806
10094-101	XMD8-92		LSM-1094	46843772	0.474119972
10095-101	ZG-10		LSM-1095		0.632899738
10096-101	ZM-447439		LSM-1096	9914412	0.871172197
10097-101	Erlotinib	OSI-774	LSM-1097	176870	1.370711224
10098-101	Gefitinib	ZD1839	LSM-1098	123631	1.062687263
10099-101	Nilotinib	AMN-107	LSM-1099	644241	1.126974966
10100-101	JNK-9L	KIN001-204	LSM-1100	59588070	1.096675185
10101-101	PD0325901	PD-325901	LSM-1101	9826528	1.296299084
10104-101	RO-3306		LSM-1104	71433937	0.613208707
10105-101	MPS-1-IN-1	HG-5-125-01	LSM-1105	25195352	1.347473378
10106-101	XMD-12		LSM-1106	54592204	1.051504315
10109-101	YM 201636	Kin001-170	LSM-1109	9956222	1.176291546
10110-101	FR180204	KIN001-230	LSM-1110	11493598	1.004015911
10111-101	TWS119		LSM-1111	9549289	1.006889712
10112-101	PF477736		LSM-1112	16750408	1.114963035
10113-101	Kin237	Kin001-237; c-Met/Ron dual kinase inhibitor	LSM-1113	16757524	2.009251213
10114-101	Pazopanib	GW786034	LSM-1114	10113978	2.716024512
10115-102	LDN-193189	DM 3189	LSM-1115	25195294	1.379353013
10116-101	PF431396		LSM-1116	11598628	1.083933499
10117-101	Celastrol		LSM-1117	5315765	1.352759395
10118-101	Amuvatinib	MP470	LSM-1118	11282283	1.17713665
10119-101	SU11274	PKI-SU11274	LSM-1119	53396327	1.204101689
10120-102	Canertinib	CI-1033; PD-183805	LSM-1120	156414	1.194695607
10121-101	SB525334		LSM-1121	9967941	0.875127226
10122-101	NVP-AEW541	AEW541	LSM-1122	46881851	0.291941727
10123-101	SGX523		LSM-1123	24779724	0.315938865
10124-101	MGCD265		LSM-1124	24901704	0.377790303
10125-101	PHA-665752		LSM-1125	66596670	0.919023711
10126-101	PI103		LSM-1126	9884685	1.045250399
10127-101	Dovitinib	TKI_258; TKI258	LSM-1127	49830557	1.115987595
10128-101	GSK 690693		LSM-1128	16048642	0.464314499
10129-101	PCI-32765		LSM-1129	16126651	0.463234881
10130-101	Masitinib	AB1010	LSM-1130	10074640	0.479937758
10131-101	Tivantinib	ARQ197	LSM-1131	11494412	0.465045725
10132-101	BMS-387032	SNS-032	LSM-1132	3025986	0.478334331
10133-101	Afatinib	BIBW-2992	LSM-5742	57519523	0.481581186
10134-101	GSK1904529A		LSM-1134	25124816	0.575227365
10135-101	OSI 906		LSM-1135	11640390	0.609908411
10136-101	TPCA-1		LSM-1136	9903786	0.786047558
10137-101	BMS509744	BMS-509744	LSM-1137	11467730	0.781029121
10138-101	Ruxolitinib	INCB018424	LSM-1139	25126798	0.722905845
10139-101	AZD-1480		LSM-1140	46861588	0.792774789
10140-101	CYT387		LSM-1141	25062766	1.029063396
10141-101	TG 101348		LSM-1142	16722836	1.108929975
10142-101	GSK-1120212	GSK1120212; JTP-74057	LSM-1143	11707110	1.035621271
10143-101	BMS 777607		LSM-1144	24794418	1.037536877
10144-101	Olaparib	AZD2281; KU-0059436	LSM-1145	23725625	0.909629031
10145-102	Veliparib	ABT-888	LSM-1146	11960529	0.842480273
10146-101	GSK2126458		LSM-1147	52914946	0.619658904
10147-101	NVP-BKM120		LSM-1148	16654980	0.600168165
10148-101	XL147	SAR245408	LSM-1149	52914932	0.645857739
10149-102	Y39983		LSM-1150	9810884	0.878862108
10150-101	Ponatinib	AP24534	LSM-1151	24826799	0.911583392
10151-101	BIBF-1120	Vargatef	LSM-1152	56965894	0.794713743
10152-101	MK 1775		LSM-1153	24856436	0.589353743
10153-101	KIN001-266		LSM-1154	44143370	0.769034963
10154-101	AT7867		LSM-1155	11175137	0.668385776
10155-101	KU-60019		LSM-1156	16117018	0.632957482
10156-101	JNJ38877605		LSM-1157	46911863	0.553079045
10157-101	Foretinib	XL880; GSK1363089	LSM-1158	42642645	0.763822095
10158-101	AZD 5438	KIN001-239	LSM-1159	16747683	0.528159223
10159-101	Pelitinib	EKB-569	LSM-1160	216467	0.819518458
10160-101	SB 216763		LSM-1161	176158	0.950210556
10161-101	NVP-AUY922		LSM-1162	53401173	0.801578531
10162-101	SP600125		LSM-1163	8515	0.861808171
10163-101	BIX 02189		LSM-1164		0.968534608
10164-101	AZD8330	ARRY-424704; ARRY-704	LSM-1165	16666708	0.90066059
10165-101	PF04217903		LSM-1166	17754438	0.950305021
10166-101	BAY61-3606		LSM-1167	10200390	1.22682713
10167-101	SB 203580	RWJ 64809; PB 203580	LSM-1168	176155	0.946537481
10168-101	VX-745		LSM-1169	3038525	0.65132132
10169-101	Doramapimod	BIRB 796	LSM-1170	156422	0.782130707
10170-101	JNJ 26854165		LSM-1171	60167550	0.861305329
10171-101	TGX221		LSM-1172	9907093	1.066374647
10172-101	GSK1059615		LSM-1173	71317162	1.694786056
10173-101	XL765	SAR245409	LSM-1174	49867926	2.278712817
10174-101	A 769662		LSM-1175	54708532	2.181401636
10175-101	Sunitinib	Sutent; SU11248	LSM-1176	3086686	2.205511276
10176-101	Y-27632		LSM-1177	5711	0.842124815
10177-101	Brivanib	BMS-540215	LSM-1178	11451527	0.951738314
10178-101	OSI-930		LSM-1179	9868037	1.032584208
10179-101	ABT-737		LSM-1180	11228183	1.126422454
10180-101	CHIR-99021	CT99021; KIN001-157	LSM-1181	9956119	0.796843315
10181-101	GDC-0879		LSM-1182	57519545	0.749396012
10182-101	Linifanib	ABT-869; AL-39324	LSM-1183	11485656	0.760089426
10183-101	BGJ398	KIN001-271; NVP-BGJ398	LSM-1184	53235510	1.017196562
10184-101	ON-01910		LSM-1185		1.663638422
10185-101	CC-401		LSM-1186	10430360	1.455918694
10186-102	Chelerythrine chloride		LSM-1187	2703	1.743156301
10187-101	Ki20227		LSM-1188	57345770	1.625687467
10188-101	BX795		LSM-1189	10077147	1.59013903
10189-101	Bosutinib	SKI-606	LSM-1190	5328940	2.405091896
10190-101	PIK-93		LSM-1191	6852167	2.203008586
10191-101	HMN-214		LSM-1192	54143018	2.603891467
10192-101	KW2449	KW-2449	LSM-1193	67089852	1.531038052
10193-101	Kin236	Tie2 kinase inhibitor	LSM-1194	23625762	2.429232222
10194-106	Cabozantinib	Cabozantinib S-malate; XL-184; BMS-907351	LSM-1195	25102847	2.444282396
10195-101	KIN001-269		LSM-1196	11654378	2.58049853
10196-101	KIN001-270		LSM-1197	66577006	2.393323376
10197-102	KIN001-260	Bayer IKKb inhibitor	LSM-1198	67135703	1.802186465
10198-101	Vandetanib	ZD6474; Zactima; Caprelsa	LSM-1199	3081361	1.551488714
10199-101	PF 573228		LSM-1200	11612883	0.995703518
10200-101	NVP-BHG712	KIN001-265	LSM-1201	16747388	0.514497652
10201-101	CH5424802		LSM-1202	49806720	0.556860557
10202-101	D 4476		LSM-1203	6419753	0.629024926
10203-101	A66		LSM-1204	42636535	0.790263551
10204-101	CAL-101		LSM-1205	11625818	0.907660675
10205-101	INK-128		LSM-1206	45375953	1.139088499
10206-101	RAF 265		LSM-1207	11656518	1.28101805
10207-101	NVP-TAE226	CHIR-265	LSM-1208	9934347	0.979705086
10208-101	JNK-IN-5A	TCS JNK 5a; KIN001-188	LSM-1209	766949	1.270556989
10209-101	BMS-536924, KIN001-126		LSM-1210	68925359	1.244170198
10213-101	KIN001-111/A770041		LSM-1214	9549184	1.325637216
10218-101	SU6656		LSM-1219	5309	1.333588108
10220-101	PKC412		LSM-1221	16760627	0.855362067
10221-101	GSK2334470		LSM-1222	46215815	0.837712491
10222-101	Dacomitinib	PF-00299804	LSM-1223	57519532	0.924157275
10223-101	AG1478	Tyrphostin	LSM-1224	2051	1.065735147
10224-104	AST1306		LSM-1225	24739943	0.9808901
10225-101	Regorafenib	BAY 73-4506	LSM-1226	11167602	1.36922362
10232-101	BEZ235	NVP-BEZ235	LSM-4255	11977753	0.766095857
10233-101	BYL719		LSM-4256	56649450	1.576765537
10234-101	GDC-0980		LSM-4257	66694608	1.446980047
10235-101	RAD001	Everolimus	LSM-4258	53398658	1.794965974
10282-101	Vorinostat	Suberoylanilide hydroxamic acid (SAHA); Zolinza	LSM-3828	5311	1.819556839
10285-101	PHA-767491		LSM-6304	11715767	1.848538985
10286-102	BS-181		LSM-6305	49867929	1.136561501
10287-101	Dinaciclib	SCH727965	LSM-6306	46926350	1.098502851
10288-101	SGI-1776		LSM-6307	24795070	1.210051596
10289-101	AZD4547		LSM-6308	51039095	1.396874436
10336-101	Epigallocatechin gallate	EGCG	LSM-5661	65064	1.450636465
10337-102	OTSSP167		LSM-6340	71543332	0.00332
10338-101	GDC-0068		LSM-6341	24788740	1.448246607
10340-101	HG-9-91-01		LSM-6343		1.327184634
10341-101	HG-14-8-02		LSM-6344	71496458	1.34586865
10342-101	HG-14-10-04		LSM-6345	56655374	1.297596531
10349-106	NVP-BGT226				1.214916579

CID, compound ID; HMS, Harvard Medical School; ID, identification; LINCS, Library of Integrated Network-Based Cellular Signatures; SM, small molecule.

A MELK–eIF4B Axis in Mitotic Cells.

MELK was recently implicated in the proliferation and survival of basal breast cancer (BBC) rather than luminal breast cancer cells, with MELK depletion causing mitotic errors and mitosis-associated cell death (2). We therefore examined whether eIF4B phosphorylation represents a major downstream event in cell proliferation and survival. Similar to the effect of loss of MELK expression (2), doxycycline-induced EIF4B knockdown induced apoptotic cell death and the appearance of G2/M markers, such as Aurora A (Fig. 4A and Fig. S3A); furthermore, EIF4B knockdown strongly impaired the proliferation of MDA-MB-468 BBC cells (Fig. 4B).

Fig. 4.

MELK–eIF4B signaling in cell growth and protein synthesis. (A and B) Effect of EIF4B knockdown on cell growth. MDA-MB-468 cells stably transduced with tet-on-short hairpin eIF4B (sh-eIF4B) were either left untreated or treated with Dox (100 ng/mL) for 3 d. (A) Mitotic lysates of nocodazole-arrested cells were used for immunoblotting. AURKA, Aurora A kinase. Cells transfected with control shRNA or two independent sh-eIF4B molecules were subjected to a cell proliferation assay with crystal violet staining (B, Top) and quantification of cell proliferation by absorbance of extracted stain (B, Bottom), ***P < 0.001. Note that sh-eIF4B-2 was used in A. (C and D) Phosphomimetic mutant of eIF4B (S406D) partially restores cell growth in cells with MELK depletion. MDA-MB-468 cells with tet-on-shMELK were stably transduced with a Dox-inducible construct encoding GFP or eIF4B (S406D)-HA. (C) Immunoblot shows that cells with ectopic expression of eIF4B (S406D) exhibited decreased cell death (indicated by PARP cleavage) and alleviated G2/M arrest (indicated by the protein abundance of Aurora A and cyclin B1) upon knockdown of MELK. (D) Crystal violet staining of cells (Top) and quantification of cell proliferation by absorbance of extracted stain (Bottom), ***P < 0.001. (E) Nonphosphorylatable mutant of eIF4B (S406A) shows decreased interaction with eIF3. HEK293T cells were transfected with the indicated vectors, and lysates were subjected to IP using anti-HA antibody. EV, empty vector; WB, Western blot. (F) De novo protein synthesis in mitotic cells with loss of MELK. Cells stably transduced with Dox-inducible shMELK were left untreated or treated with Dox for 3 d followed by 20-h treatment with nocodazole. Mitotic cells were harvested and assayed for incorporation of a methionine analog (31). Lysates were prepared for immunoblotting.

Fig. S3.

MELK–eIF4B signaling in cell growth and protein synthesis. (A) EIF4B knockdown induces cell death. MDA-MB-231 cells were stably transduced with tet-on-sh-eIF4B. Cells were left untreated or treated with Dox (100 ng/mL) for 4 d, and lysates were harvested for immunoblotting. Note that Aurora A kinase (AURKA) was used as a marker for the G2/M phase of the cell cycle and PARP cleavage at Asp214 was a marker of apoptotic cell death. (B and C) Ectopic expression of phosphomimetic eIF4B partially rescues MELK knockdown-induced G2/M arrest and suppression of cell growth. (B) BT-549 cells with tet-on-shMELK were stably introduced with tet-on-GFP or tet-on-eIF4B (S406D). (Left) Cells were left untreated or treated with Dox (100 ng/mL) for 4 d, and lysates were harvested for immunoblotting. (Right Top) Cells seeded in 12-well plates were treated for 7 d, fixed, and stained with crystal violet. (Right Bottom) Absorbance of extracted stain was measured for the quantification of cell growth. ***P < 0.01. (C) MDA-MB-468 cells with tet-on-shMELK were stably transduced with pWzl-GFP or -eIF4B (S406E). Cells were used for assays as in B. ***P < 0.001. (D) MDA-MB-468 cells with tet-on-sh-eIF4B were stably transduced with pWzl-GFP, -eIF4B (S406A), or -eIF4B (S406E). (Top) Cells were left untreated or treated with Dox (100 ng/mL) for 7 d, fixed, and stained with crystal violet. (Bottom) Absorbance of extracted stain was measured for quantification of cell growth. (E) MELK overexpression increases eIF4B phosphorylation and eIF4B-eIF3A interaction. HEK293T cells were transfected with HA-tagged eIF4B and either control vector or vector encoding MELK. At 36 h after transfection, cells were harvested for IP with anti-HA antibody and subjected to immunoblotting using the indicated antibodies. WB, Western blotting. (F) De novo protein synthesis in mitotic cells. Nocodazole-arrested mitotic cells (MDA-MB-468) were left untreated or treated with increasing concentrations of cycloheximide (CHX) for 30 min and then treated with puromycin (puro; 1.5 μg/mL) for 10 min. Lysates were subjected to immunoblotting using antipuromycin antibody. Lysate from cells without puromycin treatment was used as a negative Con (first lane). (Bottom) Histogram indicates the normalized signal of incorporated puromycin. (G) Cells with Dox-inducible shMELK or sh-eIF4B (tet-on-shRNA-neomycin vector) were left untreated or treated with Dox for 3 d, followed by 20 h of treatment with nocodazole. Mitotic cells were harvested and treated with puromycin for 10 min. Lysates were subjected to immunoblot analysis using antipuromycin antibody.

We then proceeded to investigate whether eIF4B phosphorylation at S406 is critical for the role of MELK or eIF4B in cell proliferation and survival. For this purpose, we introduced doxycycline-inducible phosphomimetic eIF4B (S406E or S406D) together with doxycycline-inducible shRNA targeting MELK or eIF4B into cells. Expression of the phosphomimetic form of eIF4B in both MDA-MB-468 and BT549 cells partially rescued the proapoptotic and antiproliferative effects caused by the loss of MELK or eIF4B (Fig. 4 C and D and Fig. S3 B–D), suggesting that eIF4B phosphorylation at Ser406 represents a major functional downstream event of MELK.

MELK–eIF4B Signaling Controls Protein Synthesis During Mitosis.

EIF4B is known to stimulate the RNA helicase activity of eIF4A responsible for unwinding the secondary structure of the 5′-UTR of mRNA (19). In addition, eIF4B promotes protein synthesis, in part, by interacting with eIF3 to facilitate the binding of mRNA with ribosomes (20). We found that mutating serine 406 to alanine (S406A) strongly decreased the interaction between eIF4B and eIF3 (Fig. 4E) and that MELK overexpression promoted S406 phosphorylation as well as the eIF4B–eIF3 interaction (Fig. S3E). These data suggest that MELK-mediated eIF4B phosphorylation might be important for efficient formation of the translation initiation complex.

To investigate a potential role of the MELK–eIF4B pathway in mRNA translation during mitosis, we harvested nocodazole-arrested mitotic cells (MDA-MB-468) that were pretreated with doxycycline to induce depletion of MELK or eIF4B and detected the synthesis of nascent protein using a puromycin-based approach (21). We confirmed active protein synthesis in mitotic cells (Fig. S3F) and found that MELK or EIF4B knockdown strongly reduced the synthesis of nascent proteins (Fig. S3G). MELK knockdown also resulted in a more than twofold decrease in global protein synthesis, as indicated by the incorporation of a methionine analog into nascent protein (Fig. 4F). Together, these data suggest a role of MELK–eIF4B signaling in regulating protein synthesis during mitosis.

MELK-eIF4B Regulates MCL1 Protein Abundance and Cancer Cell Survival.

The role of MELK-eIF4B in regulating cell survival during passage through mitosis (Fig. 4 A and B and Fig. S3A) led us to hypothesize that the synthesis of certain antiapoptotic protein(s) might be highly dependent on MELK–eIF4B signaling. To identify such factor(s), we analyzed the abundance of antiapoptotic proteins in mitotic cells depleted for eIF4B and found that the protein abundance of MCL1 and XIAP, but not Bcl2, Bcl-xL, or BIRC5/survivin, was reduced upon EIF4B knockdown (Fig. 5A and Fig. S4A). This finding is consistent with the faster turnover of MCL1 compared with other BH3 family proteins described previously (3). Also consistent with previous studies (4), active protein synthesis and degradation of MCL1, but not XIAP, were observed in mitotic cells (Fig. 5B and Fig. S4B). Therefore, we focused on the role of MCL1 in mediating MELK–eIF4B signaling and regulating cell survival.

Fig. 5.

MELK–eIF4B signaling regulates the protein synthesis of MCL1 during mitosis. (A) Analysis of antiapoptotic proteins in mitotic cells with loss of eIF4B. MDA-MB-231 cells stably transduced with tet-on-sh-eIF4B were left untreated or treated with Dox for 2 d, followed by 20-h treatment with nocodazole (200 ng/mL). Mitotic cells were harvested by shake-off and analyzed by immunoblotting. Note that the protein abundance of MCL1 and XIAP, but not Bcl-2, Bcl-xL, or BIRC5, was decreased upon loss of eIF4B. (B) Synthesis and degradation of MCL1 protein in mitotic cells. Paclitaxel-arrested mitotic cells from the indicated cell lines were left untreated (Con) or were treated with cycloheximide (CHX, 50 μg/mL) or MG132 (10 μM) for 2 h. Cell lysates were subjected to immunoblotting. (C) Loss of MELK reduces MCL1 protein abundance in mitotic cells. Cells with tet-on-shMELK were left untreated or treated with Dox for 2 d, and then treated with nocodazole for 20 h for the harvest of mitotic cells. (D) MELK inhibition does not alter the stability of MCL1. Mitotic cells were treated with CHX (100 μg/mL) or CHX combined with OTSSP167 (50 nM) for the indicated time. (Top) Lysates were prepared for immunoblotting. (Bottom) Signal of MCL1 was quantified and normalized to the signal of α-tubulin. (E) MCL1 transcription is not down-regulated upon MELK or EIF4B knockdown. Cells were treated as in C and harvested for total RNA extraction, followed by reverse transcription and quantitative PCR using primer pairs specific for MELK or eIF4B and two independent pairs of primers for MCL1. (F) MELK inhibition impairs the synthesis of MCL1. Mitotic cells were either left untreated or treated with CHX for 90 min. A sample of the CHX-treated cells was washed and incubated in the presence of MG132 (CHX to MG132) or MG132 plus OTSSP167 (50 nM) (CHX to MG132 + OTSSP167) for 90 min. Cell lysates were prepared for immunoblotting. (G) Ectopic expression of MCL1 results in partial rescue of growth inhibition in cells depleted for MELK. MDA-MB-468 cells with tet-on-sh-MELK were transduced with pWzl retroviral vector encoding GFP or MCL1. Cells were seeded in 12-well plates (20,000 cells per well), and either left untreated or treated with Dox (100 ng/mL) for 7 d. (Right) Cells were stained with crystal violet. (Left) Stain was extracted for quantification of cell growth. ***P < 0.001. (H) MCL1 overexpression alleviates cell death induced by MELK knockdown. (Left) Cells as in G were left untreated or treated with Dox (100 ng/mL) for 4 d before immunoblotting. (Right) Signals of PARP cleavage and MCL1 protein abundance were normalized to signals of vinculin. **P < 0.01. (I) Overexpression of MCL1 in BBC. Tumor samples in the Curtis et al. cohort (32) were classified into five distinct molecular subtypes (33). Note that BBC tumors have a higher expression level of MCL1 than luminal or HER2⁺ breast cancer. ****P < 0.0001. (J) BBC cells are sensitive to MCL1 inhibition. BBC (MDA-MB-468, BT549) and luminal breast cancer (MDA-MB-415, T47D) cells were treated with vehicle control or MCL1 inhibitor (MIM1). (Right) Cells were stained with crystal violet. (Left) Absorbance of extracted stain was measured for the quantification of cell growth.

Fig. S4.

Mitotic translation of MCL1. (A) Effect of EIF4B knockdown on the expression of antiapoptotic proteins. MDA-MB-231 cells stably transduced with tet-on-sh-eIF4B were left untreated or treated with Dox for 2 d, followed by 20 h of treatment with paclitaxel. Mitotic cells were harvested by shake-off, and cell lysates were subjected to immunoblotting. Note that the protein abundance of MCL1 and XIAP, but not Bcl-2, Bcl-xL, or BIRC5, was decreased upon loss of eIF4B. (B) MCL1 protein abundance in mitotic cells is sensitive to perturbation of protein synthesis or degradation. Nocodazole-arrested mitotic cells from the indicated cell lines were left untreated (Con) or treated with CHX (50 μg/mL) or MG132 (10 μM) for 2 h. Cell lysates were subjected to immunoblotting. (C) EIF4B knockdown does not accelerate MCL1 degradation. MDA-MB-231 cells stably transduced with tet-on-sh-eIF4B were left untreated or treated with Dox (100 ng/mL) for 2 d. Cells were treated with nocodazole for 20 h, and the resultant mitotic cells were treated with CHX (100 μg/mL) for the indicated time. (Left) Cell lysates were subjected to immunoblotting using the indicated antibodies. (Right) MCL1 signal was quantified and normalized to the signal of a loading control (α-tubulin).

Similar to eIF4B depletion, MELK knockdown significantly decreased the abundance of MCL1 in mitotic cells (Fig. 5C). To test whether MCL1 loss promoted by the loss of eIF4B and MELK was caused by an increase in proteolysis or a decrease in synthesis, we added cycloheximide to block mRNA translation and performed immunoblotting to measure protein stability during mitotic arrest. We found that the protein stability of MCL1 was not altered by MELK or eIF4B inhibition (Fig. 5D and Fig. S4C), indicating that loss of eIF4B and MELK affects MCL1 synthesis but not proteolysis. We next measured mRNA levels of MCL1 using two independent probes for quantitative PCR and found that mRNA expression was not reduced, or was even slightly increased, upon MELK or EIF4B knockdown (Fig. 5E). To address a direct role of MELK in regulating MCL1 synthesis, we pretreated cells with cycloheximide to deplete MCL1 protein and, after washing away the cycloheximide, treated the cells with MG132 to observe newly synthesized MCL1. MELK inhibition by OTSSP167 efficiently decreased the synthesis of MCL1 protein (Fig. 5F), suggesting a role of MELK in regulating the de novo synthesis of MCL1, presumably via regulation of eIF4B phosphorylation.

To study whether MCL1 synthesis represents a critical downstream effector of the MELK–eIF4B pathway, we introduced exogenous MCL1 into cells stably transduced with tet-on-short hairpin MELK. MCL1 overexpression in MDA-MB-468 cells partially rescued the growth inhibition and cell death resulting from MELK knockdown (Fig. 5 G and H). We previously found that MELK is selectively required by BBC, an aggressive and therapeutically recalcitrant type of breast cancer (2). Interestingly, the level of MCL1 transcript was higher in BBC than in other subtypes of breast cancer (Fig. 5I). Moreover, MCL1 inhibition using a small-molecule inhibitor, MIM1 (22), caused more prominent defects in cell proliferation in BBC cells than in luminal breast cancer cells (Fig. 5J). These observations suggest that BBC cells might be addicted to high MCL1 levels for successful passage through mitosis, which, in turn, depends on up-regulation of MCL1 synthesis during mitosis by MELK–eIF4B signaling.

Discussion

The cellular functions and regulatory mechanisms of mitotic translation are poorly understood. The Ser/Thr kinase MELK is one of the proteins that show increased abundance during mitosis. Our initial biochemical studies identified eIF4B as one of the kinase substrates of MELK, with selective phosphorylation at Ser406. A chemical library screen of more than 200 kinase inhibitors indicated that MELK is uniquely capable of phosphorylating MELK. We further demonstrated that the MELK–eIF4B signaling pathway regulates protein synthesis in mitotic cells and, in particular, synthesis of the antiapoptotic protein MCL1.

The control of protein synthesis plays a critical role in regulating cell growth, survival, and tumorigenesis (23). Aberrations in translational control frequently occur in cancer, often through activation of PI3K–MAPK–mTOR signaling (23, 24) but also through overexpression of translation initiation factors, such as eIF4E (25) and helicase protein DHX29 (26). Here, we identify the MELK–eIF4B pathway as a previously unidentified strategy used by tumor cells to regulate mitotic translation and promote survival. Given the widespread overexpression of MELK in high-grade tumors, our findings in breast cancer may be applicable to many other aggressive types of malignancy, such as high-grade serous ovarian cancer and glioblastoma.

How does eIF4B phosphorylation affect protein synthesis? Consistent with a recent study (27), we find that eIF4B phosphorylation at Ser406 is required for the interaction between eIF4B and the eIF3 translation initiation complex. Those observations suggest a model in which eIF4B phosphorylation promotes the recruitment of ribosomes to mRNA and, consequently, protein synthesis. Intriguingly, the same eIF4B phosphorylation suppresses neuronal protein synthesis because eIF4B binds to regulatory non–protein-coding brain cytoplasmic RNAs with high affinity, and is thus sequestered from initiating translation of other protein-coding transcripts (28). It will be extremely interesting to study how eIF4B phosphorylation regulates protein synthesis in different cell types.

Our work demonstrates a role of the MELK–eIF4B pathway in controlling the translation of MCL1, a short-lived antiapoptotic protein whose gene is frequently amplified in various tumor types (29). Notably, MCL1 gene amplification is observed in breast tumors treated with neoadjuvant chemotherapy (30). The identified mechanism for regulation of MCL1 synthesis also implicates the MELK–eIF4B pathway as a promising therapeutic target for cancers that display MCL1 gene amplification/overexpression and depend on MCL1 for cell survival.

As the only unstable antiapoptotic BH3 family protein, MCL1 plays a major role in determining the response of cancer cells to therapeutic intervention. Protein degradation of MCL1 is induced during prolonged mitotic arrest promoted by antimicrotubule drugs, such as paclitaxel (5), and it was proposed that sensitivity to this class of drugs depends, in part, on expression levels of the E3 ligases that promote MCL1 ubiquitination (4). Here, we show that a decrease in MCL1 synthesis also accounts for the selective killing of BBC cells by MELK inhibition. It is not clear why BBC cells are more sensitive to loss of MCL1 synthesis in mitosis than other breast cancer subtypes, although because BBC cells express more MCL1, they may be more addicted to its antiapoptotic effects. Given that the antiapoptotic roles of MCL1 are targeted by multiple modes of chemotherapy, our findings suggest clinical benefits of combination regimens using MELK inhibitors to repress mitotic phosphorylation of eIF4B and consequent MCL1 synthesis.

Although we find that the translation of MCL1 mRNA in mitotic cells is regulated by the MELK–eIF4B pathway, MCL1 overexpression only partially rescued the growth inhibition resulting from the loss of MELK. It is likely that the signaling may also have an impact on the protein synthesis of other prosurvival mRNAs, especially those prosurvival mRNAs with a structured 5′-UTR that consequently have a high dependence on eIF4B (12). The translation of these mRNAs, including MCL1, may collectively mediate the MELK–eIF4B signaling and regulate the mitotic survival of cancer cells.

Materials and Methods

Cell lines were cultured under standard conditions. Floating mitotic cells were harvested through nocodazole- or paclitaxel-induced metaphase arrest. Both MCL1 and eIF4B were cloned from cDNA libraries prepared from human BCC culture. Protein extraction, immunoblotting, and quantitative RT-PCR were performed using standard methods. A detailed description of reagents and protocols used in this study can be found in SI Materials and Methods.

SI Materials and Methods

Plasmids.

Human eIF4B was cloned from the reverse transcription products of total RNA extracted from human mammary epithelial cells using the following primers (forward: ATGGCGGCCTCAGCAAAAAAG; reverse: CTATTCGGCATAATCTTCTC). The 1.8-kb PCR product was used as a template for amplification of Flag-tagged or HA-tagged eIF4B with restriction sites. The resultant plasmid constructs (pWzl-Flag-eIF4B and pTrex-eIF4B-HA) were verified by sequencing. Site-directed mutagenesis of eIF4B was performed using QuikChange XL (Stratagene), and all mutant constructs were confirmed by sequencing.

To generate pLKO-tet-on-shRNA targeting human eIF4B, synthesized oligonucleotides were annealed and ligated with digested pLKO vector. The sequences for sh-eIF4B-1 and sh-eIF4B-2 were GGACCAGGAAGGAAAGATGAA and GCGGAGAAACACCTTGATCTT, respectively.

Retroviral and Lentiviral Gene Delivery.

Retroviruses were generated by transfecting HEK293T cells with retroviral plasmids and packaging DNA. Typically, 1.6 μg of pWzl DNA, 1.2 μg of pCG-VSVG (glycoprotein G of the vesicular stomatitis virus), 1.2 μg of pCG-gap/pol, and 12 μL of lipid of Metafectene Pro (Biontex) were used. DNA and lipid were each diluted in 300 μL of PBS and mixed; after 15 min of incubation, they were added to a 6-cm dish that had been seeded with 3 million HEK293T cells 1 d previously. Viral supernatant was collected 48 and 72 h posttransfection. The supernatant was filtered through a 0.45-μm membrane and added to target cells in the presence of 8 μg/mL polybrene (Millipore). Lentiviruses were generated with a similar approach, except the cells were transfected with 2 μg of pLKO DNA, 1.5 μg of pCMV-dR8.91, and 0.5 μg of pMD2-VSVG. Cells were selected with antibiotics, starting 72 h after the initial infection. Puromycin and blasticidin were used at final concentrations of 1.5 μg/mL and 4 μg/mL respectively.

Antibodies.

Antibodies to the following proteins were used for immunoblotting or immunoprecipitation (IP): puromycin (MABE343; Millipore), MELK (2916; Epitomics), p-eIF4B (S406) [8151; Cell Signaling Technology (CST)], p-eIF4B (S422) (3591; CST), eIF4B (3592; CST), MCL1 (5453; CST), XIAP (2045; CST), p-Akt (S473) (4060; CST), p-MAPK (T202/Y204) (4370; CST), cleaved PARP (Asp214) (9541; CST), Aurora A (4718; CST), and vinculin (V9131; Sigma). Anti-HA magnetic beads were from Pierce (88836), and anti-Flag magnetic beads were from Sigma (M8823). Secondary antibodies used were Alexa Fluor 680 goat anti-rabbit IgG (A-21109; Invitrogen) and IRDye800-conjugated anti-mouse IgG (Rockland).

IP.

Cells were lysed with buffer containing 50 mM Tris (pH 7.4), 100 mM NaCl, 1% Triton, protease inhibitor mixture (Roche), and phosphatase inhibitor mixture (Pierce). Cleared lysates were incubated with anti-mouse IgG magnetic beads (New England Biolabs) for the elimination of proteins with nonspecific binding to immunoglobins and/or beads. The lysates were then incubated overnight with anti-Flag (M8823; Sigma) or anti-HA magnetic beads (88836; Pierce). The beads were washed with lysis buffer five times before SDS sample buffer was added to prepare samples for immunoblotting. For mass spectrometry assay, proteins binding to anti-Flag beads were eluted using Flag peptides (F4799; Sigma) and precipitated by trichloroacetic acid. The sample was submitted to the Taplin Biological Mass Spectrometry Facility (Harvard Medical School) for mass spectrometry analysis.

Expression and Purification of MELK.

Full-length human MELK and its kinase domain (aa 1–340) were expressed using a baculovirus system. Briefly, vectors encoding MELK [pTriEx-His-GST-tobacco etch virus (tev)-MELK] were introduced into sf9 insect cells for generation of virus. BTI-TN-5B1-4 insect cells were infected with virus to express MELK. Lysates were subjected to His-tag affinity chromatography, GST-tag affinity chromatography, removal of His- and GST-tags using TEV protease, and size exclusion chromatography.

In Vitro Kinase Assay.

Flag-tagged eIF4B or Flag-eIF4B (S406A) was transfected into HEK293T cells (4 μg of DNA for cells in a 60-mm dish). At 36 h after transfection, the cells were lysed with IP buffer [100 mM NaCl, 50 mM Tris (pH 7.5), 0.5% Nonidet P-40, 0.5% sodium deoxycholate, supplemented with protease/phosphatase inhibitor mixture]. Lysates were cleared by incubation with anti-mouse IgG conjugated to magnetic beads (4 °C, 30 min) and then immunoprecipitated with anti-Flag M2 magnetic beads (Sigma; 4 °C, 120 min). The beads with bound antigens were washed five times with IP buffer. During the last wash, the beads were aliquoted into 1.5-mL microcentrifuge tubes. After removal of IP buffer, the beads were washed once with 1× kinase buffer without ATP [5 mM Tris (pH 7.5), 5 mM β-glycerophosphate, 2 mM DTT, 0.1 mM Na₃VO₄, 10 mM MgCl₂; CST], and 40 μL of 1× kinase buffer with 200 mM ATP was added to each tube, followed by 5 μL of buffer without or with 500 ng of recombinant MELK. The mixture was incubated at 30 °C for 30 min, and the reaction was terminated by the addition of 40 μL of 2× SDS sample buffer. The samples were boiled and subjected to immunoblotting. For the kinase reaction containing GST-tagged eIF4B (H00001975-P01; Abnova), 240 ng of GST-eIF4B, 1 μg of MELK (kinase domain), and 300 μM ATP were used in a total reaction volume of 50 μL.

Positional Scanning Peptide Library Screen.

The positional scanning peptide library screen was performed as described previously (11, 34) using active full-length human MELK purified from insect cells. Briefly, a set of 180 biotin-conjugated peptides with the following sequence, Y-A-X-X-X-X-X-S/T-X-X-X-X-A-G-K-K-biotin, was used, where S/T indicates an equimolar mixture of Ser and Thr, and one of the nine X positions is substituted with each of the 20 total amino acids. Thus, each amino acid surrounding the S/T can be systematically evaluated. Peptides were arrayed in 384-well plates in buffer [50 mM Hepes (pH 7.5), 20 mM MgCl₂, 0.02 mg/mL BSA, 0.01% Brij 35, 5 mM DTT, 0.5 mM EGTA]. Full-length recombinant MELK and ϒ-[³²P]-ATP were added to wells to final concentrations of 50 μM for peptide and 100 μM and 0.025 μCi/μL for ATP. After incubation for 2 h at 30 °C, aliquots of the reactions were spotted onto a streptavidin membrane. The membrane was quenched, washed extensively, dried, and exposed to a phosphor storage screen.