Ivermectin represses Wnt/β-catenin signaling by binding to TELO2, a regulator of phosphatidylinositol 3-kinase-related kinases

Abstract

Ivermectin (IVM), an avermectin-derivative anthelmintic, specifically binds to glutamate-gated chloride ion channels (GluCls), causing paralysis in invertebrates. IVM also exhibits other biological activities such as Wnt/β-catenin pathway inhibition in vertebrates that do not possess GluCls. This study showed that affinity purification using immobilized IVM B1a isolated TELO2, a cofactor of phosphatidylinositol 3-kinase-related kinases (PIKKs), as a specific IVM B1a-binding protein. TELO2 knockdown reduced cytoplasmic β-catenin and the transcriptional activation of β-catenin/TCF. IVM B1a bound to TELO2 through the C-terminal α-helix, in which mutations conferred IVM resistance. IVM reduced the TELO2 and PIKK protein levels and the AKT and S6 kinase phosphorylation levels. The inhibition of mTOR kinase reduced the cytoplasmic β-catenin level. Therefore, IVM binds to TELO2, inhibiting PIKKs and reducing the cytoplasmic β-catenin level. In conclusion, our data indicate TELO2 as a druggable target for human diseases involving abnormalities of the Wnt/β-catenin pathway and PIKKs, including mTOR.

Keywords: Biochemistry; Molecular biology; Small molecule.

Graphical abstract

Figure 1

Ivermectin (IVM) suppresses the Wnt/β-catenin pathway in zebrafish embryos and mammalian cells (A) IVM was identified as a chemical suppressor of the eyeless phenotype. Zebrafish embryos were pretreated with 100 μM IVM at 50% epiboly. They were then treated with 6-bromo-indirubin-3′-oxime (BIO), which is a GSK3 inhibitor that leads to the accumulation of β-catenin, at the shield stage and were incubated for 24 h. Images were obtained at 30 h postfertilization. Scale bar = 200 μm. See also Figure S1. (B) IVM reduced the β-catenin/TCF-dependent transcriptional activity. Human embryonic kidney 293 (HEK293) cells were transiently cotransfected with Super 8x TOPFlash—a firefly luciferase reporter plasmid—to monitor the β-catenin/TCF-dependent transcriptional activity and with pRL-SV40—a renilla luciferase reporter plasmid—to normalize the transfection efficiency. The cells were pretreated with IVM at the indicated concentrations for 1 h and then treated with 50 ng/mL of Wnt3A for 18 h in 1% fetal bovine serum (FBS)–supplemented Dulbecco’s Modified Eagle Medium (DMEM). Transcriptional activation in the cells was assayed by measuring the firefly and renilla luciferase activities. Normalized relative luciferase activities were calculated by dividing firefly luciferase activities by those of renilla. Transcriptional activation levels are indicated as values relative to the control (dimethyl sulfoxide [DMSO]-treated and bovine serum albumin [BSA]–treated cells). Data are presented as the means ± standard errors of the means (n = 4 biological replicates). ∗p < 0.05, ∗∗∗p < 0.001, n.s.: not significant, one-way ANOVA with Tukey’s test. (C) IVM downregulated the target proteins involved in Wnt/β-catenin signaling. Human colorectal cancer DLD-1 cells were treated with IVM at the indicated concentrations for 18 h in 1% FBS-supplemented RPMI 1640 medium. Subsequently, the cytoplasmic fractions of the cells were probed for Axin2, cyclin D1, β-catenin, and actin. (D) IVM reduced the cytoplasmic β-catenin levels in the presence of a proteasomal inhibitor, MG132. HEK293 cells were first treated with 25 μM MG132 for 15 min, followed by treatment with 10 μM IVM for 1 h and 50 ng/mL of Wnt3A for 2 h in 1% FBS/DMEM. The cytoplasmic proteins were probed with anti-β-catenin and anti-actin antibodies (the left panel). The band intensities were quantified, normalized to the actin levels, and reported as values relative to the control (DMSO-treated and BSA-treated cells in the absence of MG132; right panel). The open triangle indicates the bands of ubiquitinated β-catenin. Data are presented as the means ± standard deviations (SDs; n = 3 biological replicates). ∗p < 0.05, one-way ANOVA with Tukey’s test. See also .Figure S2

Figure 2

Identification of telomere length regulation protein 2 homolog (TELO2) as a mammalian target of IVM (A) Chemical structure of avermectins and immobilized dihydroavermectin B1a (IVM B1a) for target identification. (B) IVM B1a-binding proteins were identified through a pull-down assay with chemically immobilized IVM B1a. HEK293 cell lysates were incubated with the affinity beads in the absence or presence of soluble IVM as a competitor at 33- or 100-folds. The bound fractions were eluted, resolved by SDS-PAGE, and visualized with silver staining. The arrows indicate the bands that were competed out by the competitor. (C) Mascot analysis of the data was summarized. TELO2 was identified as an IVM B1a-binding protein. Cluster of K22E, keratins, type II cytoskeletal 2 epidermal, 5, 6A, 6B, and 75; VIME, vimentin; TTI1, TELO2-interacting protein 1 homolog; K2C1, keratin, type II cytoskeletal 1; IPO11, importin-11; SGPL1, sphingosine-1-phosphate lyase 1; K1C9, keratin, type I cytoskeletal 9; cluster of TBB4B, cluster of tubulin beta, beta-2A, beta-4B, and beta-6 chains; K1C10, keratin, type I cytoskeletal 10. See also Data S1. (D) Binding of TELO2 to the immobilized IVM B1a was confirmed through western blotting with an anti-TELO2 antibody. Free IVM was added as the competitor.

Figure 3

TELO2 is essential for the maintenance of β-catenin levels (A) Knockdown of TELO2 with siRNAs reduced β-catenin/TCF-dependent transcriptional activation in HEK293 cells. The cells transfected with control or TELO2 siRNAs, and then transiently cotransfected with Super 8x TOPFlash (a firefly luciferase reporter plasmid) to assess the β-catenin/TCF-dependent transcriptional activity and with pRL-SV40 (a renilla luciferase reporter plasmid) to normalize the transfection efficiency. The cells were treated with 50 ng/mL Wnt3A for 18 h. Firefly and renilla luciferase activities were measured. Normalized relative luciferase activities were calculated by dividing firefly luciferase activities by those of renilla and indicated the level of transcriptional activation. Data are presented as the means ± SDs (n = 3 biological replicates). ∗∗∗p < 0.001, one-way ANOVA with Tukey’s test. (B–D) Knockdown of TELO2 with siRNAs reduced the β-catenin levels in HEK293 and HT-29 cells. HEK293 (B and C) or HT-29 (D) cells were transfected with siRNAs for TELO2 for 72 h. The cell lysates were analyzed via western blotting with the indicated antibodies (B and D). (C and D) Band intensities were quantified, normalized to the actin levels, and reported as values relative to those obtained from the cells transfected with a control siRNA. Data are presented as the means ± SDs (C, n = 3 biological replicates; D, n = 4 biological replicates). ∗∗p < 0.01, ∗∗∗p < 0.001, one-way ANOVA with Tukey’s test.

Figure 4

TELO2 requires the C-terminal α-helix integrity for its binding to IVM B1a (A) TELO2 deletion mutants lacking α-helices in the C-terminal region. (B) Glutathione S-transferase (GST)-fusion TELO2 deletion mutants Δ5 and Δ6 lost the binding affinity for IVM B1a. Lysates from E. coli expressing GST-fusion TELO2 mutants were subjected to the binding assay. The lysates and the bound fractions were analyzed through western blotting with an anti-GST antibody. Arrows indicate the bands corresponding to TELO2 mutants. (C) Sites of point mutations in the C-terminal α-helix are indicated as triangles. (D) Site-directed mutagenesis in the C-terminal helix of TELO2 inhibited the binding of TELO2 to IVM B1a. Binding affinities of the point mutants were analyzed by probing the lysates and bound fractions with an anti-GST antibody. (E) Chemical structures of IVM B1a-SNIPERs (specific and nongenetic IAP-dependent protein erasers) 3 and 4. (F) IVM B1a-SNIPER promotes the proteasomal degradation of TELO2. (G) IVM B1a-SNIPERs 3 and 4 reduced endogenous TELO2. HEK293 cells were treated with IVM B1a-SNIPERs 3 or 4 for 18 h in 1% FBS/DMEM. Then, the cell lysates were probed with anti-TELO2 and anti-actin antibodies (the left panel). The band intensities were quantified, normalized to the actin levels, and indicated as values relative to the control (DMSO) (the right panel). Data are presented as the means ± SDs (n = 3 biological replicates). ∗p < 0.05, ∗∗p < 0.01, one-way ANOVA with Tukey’s test. See also Figure S3. (H) Interaction of IVM B1a with TELO2 depended on K749 in cells. HEK293 cells were transfected with vectors encoding FLAG-tagged WT or K749T TELO2 and treated with 1 or 10 μM IVM B1a-SNIPER 3 for 5 h in 1% FBS/DMEM. Cell lysates were probed with anti-FLAG and anti-actin antibodies (the left panel). The band intensities were quantified, normalized to the actin levels, and indicated as values relative to the control (DMSO; the right panels). Data are presented as the means ± SDs (n = 3 biological replicates). ∗∗p < 0.01, ∗∗∗p < 0.001, one-way ANOVA with Tukey’s test. See also Figure S4.

Figure 5

IVM suppresses Wnt/β-catenin signaling via binding to TELO2 (A and B) Reconstitution of the TELO2 restored β-catenin levels in TELO2-knockdown cells. HEK293 cells were transfected with TELO2 siRNA for 4 days and transfected with an siRNA-resistant FLAG-tagged WT TELO2 expression vector. The cells were treated with 10 μM IVM for 1 h and then with 50 ng/mL Wnt3A for 2 h in the presence of 25 μM MG132. (A) Cell lysates were analyzed through western blotting with anti-β-catenin, anti-TELO2, anti-FLAG, and anti-actin antibodies. The open triangle indicates the bands corresponding to ubiquitinated β-catenin. (B) Band intensities of TELO2 and β-catenin were quantified in control siRNA-transfected (circles), TELO2 #1 siRNA-transfected (triangles), or TELO2-reconstituted (squares) cells in the absence of IVM, normalized to the actin levels, and indicated as values relative to the control siRNA-transfected cells. The X- and Y axes of the left panel indicate relative TELO2 and β-catenin levels, respectively (n = 4 biological replicates). Correlation coefficient (r) = 0.85. Data of the right panels represent the means ± SDs (n = 4 biological replicates). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, one-way ANOVA with Tukey’s test. (C) TELO2 K749T reconstitution conferred IVM resistance. TELO2-knockdown HEK293 cells were transfected with vectors expressing siRNA-resistant FLAG-tagged WT TELO2 or TELO2 K749T. The cells were treated with 25 μM MG132 for 15 min, 10 μM IVM for 1 h, and 50 ng/mL Wnt3A for 2 h in 1% FBS/DMEM. Protein levels were analyzed through western blotting with specific antibodies (the left panel). The open triangle indicates bands corresponding to ubiquitinated β-catenin. The band intensities were quantified, normalized to actin levels, and indicated as values relative to the control (DMSO) (the right panels). Data are presented as the means ± SDs (n = 3 biological replicates). ∗∗p < 0.01, n.s.: not significant, Welch's t-test. See also .Figure S5

Figure 6

IVM reduces TELO2, phosphatidylinositol 3-kinase-related kinase (PIKK) levels and mTOR substrate phosphorylation levels (A) Short-term IVM treatment reduced TELO2 and β-catenin, but not PIKK levels. HEK293 cells were treated with 5 or 10 μM of IVM for 3, 9, or 24 h in 1% FBS/DMEM. Subsequently, the cell lysates were probed for TELO2, cytoplasmic β-catenin, mTOR, ataxia telangiectasia mutated (ATM), ATM-related and Rad3-related (ATR), DNA-dependent protein kinase (DNA-PK), and actin through western blotting with specific antibodies. The band intensities were quantified, normalized to the actin levels, and indicated as values relative to the control (0 h). Data are presented as the means ± SDs (n = 3 biological replicates). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, one-way ANOVA with Tukey’s test. (B) Long-term IVM treatment reduced PIKKs. HEK293 cells were treated with 5 μM IVM for 72 h in 1% FBS/DMEM. The cell lysates were probed through western blotting with the indicated antibodies (the left panels). The band intensities were quantified, normalized to the actin levels, and indicated as values relative to the control (DMSO) (the right panels). Data are presented as the means ± SDs (n = 3 biological replicates). ∗p < 0.05, Welch’s t-test. (C) Short-term IVM treatment reduced AKT and S6 kinase phosphorylation levels. HEK293 cells were treated with 10 μM IVM for 3 h in 1% FBS/DMEM. The cell lysates were probed through western blotting with the indicated antibodies (left panels). The band intensities were quantified, normalized by the total protein levels, and indicated as values relative to the control (DMSO) (the right panels). Data are presented as the means ± SDs (n = 3 biological replicates). ∗∗p < 0.01, Welch’s t-test. See also Figures S6 and S7. (D) Torin2 reduced the cytoplasmic level of β-catenin. HEK293 cells were treated with 0.1 μM Torin2 or rapamycin for 3 h in 1% FBS/DMEM. The cytoplasmic proteins were probed through western blotting with the indicated antibodies (the left panels). The band intensities were quantified, normalized to the actin levels, and indicated as values relative to the control (DMSO; the right panels). Data are presented as the means ± SDs (n = 3 biological replicates). ∗∗p < 0.01, Welch’s t-test. See also Figure S8.

Scheme 1

Preparation of 4″-O-(ethoxycarbonylmethyl)-ivermectin B1a (1) and 5-O-(ethoxycarbonylmethyl)-ivermectin B1a (2)

Scheme 2

Preparation of 4″-O-(carboxymethyl)-ivermectin B1a (P1)

Scheme 3

Preparation of 5-O-(carboxymethyl)-ivermectin B1a (P2)

Scheme 4

Preparation of 4″-O-(azido-PEG3-carbonylmethyl)-ivermectin B1a (IVM B1a-P1-longN₃)

Scheme 5

Preparation of 4″-O-[2-((3-azidopropyl)amino)-2-oxoethyl]-ivermectin B1a (IVM B1a-P1-shortN₃)

Scheme 6

Preparation of 5-O-(azido-PEG3-carbonylmethyl)-ivermectin B1a (IVM B1a-P2-longN₃)

Scheme 7

Preparation of 5-O-[2-((3-azidopropyl)amino)-2-oxoethyl]-ivermectin B1a (IVM B1a-P2-shortN₃)

Scheme 8

Preparation of N-[(2S,3R)-3-(Fmoc-amino)-2-hydroxy-4-phenylbutyryl]-L-leucine (Ubenimex-Fmoc)

Scheme 9

Preparation of N-[(2S,3R)-3-(Fmoc-amino)-2-hydroxy-4-phenylbutyryl]-L-leucine propargyl ester (Ubenimex-Fmoc-alkyne)

Scheme 10

Preparation of N-[(2S,3R)-3amino-2-hydroxy-4-phenylbutyryl]-L-leucine propargyl ester (Ubenimex-alkyne)

Scheme 11

Preparation of ubenimex 4-[1-(4″-O-IVM B1a-PEG3-linker)-1H-1,2,3-triazolyl]-methyl ester (IVM B1a-SNIPER 1)

Scheme 12

Preparation of ubenimex 4-[1-(5-O-IVM B1a-PEG3-linker)-1H-1,2,3-triazolyl]-methyl ester (IVM B1a-SNIPER 2)

Scheme 13

Preparation of ubenimex 4-[1-(4″-O-IVM B1a-amide linker)-1H-1,2,3-triazolyl]-methyl ester (IVM B1a-SNIPER 3)

Scheme 14

Preparation of ubenimex 4-[1-(5-O-IVM B1a-amide linker)-1H-1,2,3-triazolyl]-methyl ester (IVM B1a-SNIPER 4)