Diterpene Molecular Glue Stabilizes Protein–Protein Interactions of a Disordered Phosphoprotein that Controls Translational Repression

Abstract

Chemical stabilization of protein–protein interactions involving intrinsically disordered phosphoproteins is an emerging yet challenging problem. Synthetic agents capable of achieving this goal are desirable for advancing understanding of the biological roles of these transient interactions and for developing new therapeutics. 12-Deoxyfusicoccin (12-dFC), a semisynthetic derivative of the phytotoxin fusicoccin A, exhibits significant antitumor activity in animal models while its parent compound and 12-hydroxyfusicoccin (12-hFC) are inactive, although the molecular mechanism underlying this differential activity has remained unclear. Here, through tandem-affinity proteomics, we identified GRB10-interacting GYF protein 2 (GIGYF2), a largely disordered scaffold protein in mRNA translational repression complexes as a 14-3-3 binding partner selectively stabilized by 12-dFC but not 12-hFC. Biochemical analyses revealed that 12-dFC promotes cooperative binding of 14-3-3 to the mode-1 consensus motif KGVpS₅₄₆IP in GIGYF2, enhancing binding affinity by approximately 50-fold. This stabilization enhanced GIGYF2’s translational repressor function, suppressing protein synthesis and inhibiting cell proliferation. We further demonstrated that activation of AMP-activated protein kinase (AMPK), the cellular energy sensor kinase, enhanced 14-3-3 binding to wild-type GIGYF2 but not the S546A mutant, indicating that AMPK mediates S₅₄₆ phosphorylation. Importantly, S₅₄₆ phosphorylation increased significantly under starvation conditions, suggesting a GIGYF2-regulated stress-responsive pathway that suppresses protein synthesis to conserve ATP through reversible 14-3-3 binding. These findings demonstrate that 12-dFC is a potent molecular glue stabilizer of 14-3-3 interactions with intrinsically disordered phosphoproteins and offers a robust chemical basis for developing synthetic agents that target 14-3-3-mediated control of metabolic homeostasis.

Introduction

Chemical stabilization of protein–protein interactions (PPIs) represents an important yet challenging task. Molecular glues that induce biomolecular associations and proximity are emerging as a transformative modality to address this challenge. , While numerous examples of targeting degraders and inducers of neo-protein interactions have been developed for applications including protein degradation, cell therapies, and immune modulation, strategies to enhance existing PPIs remain underexplored. Such stabilizers are particularly needed given the critical roles of transient PPIs involving intrinsically disordered proteins in controlling spatiotemporally regulated biological processes, including formation of biological condensates and pathways dynamically modulated by posttranslational modifications. − Chemical agents that can selectively modulate these transient interactions would provide valuable tools for understanding and manipulating cellular processes.

Fusicoccin-A (FC-A, Figure A) is a 5-8-5 tricyclic diterpene glycoside produced by the plant pathogenic fungus Phomopsis amygdali, an example of natural product-inspired molecular glues. FC-A possesses a unique ability to stabilize PPIs between 14-3-3 proteins and their phosphorylated binding partners through formation of stable ternary complexes. In its canonical target, plasma membrane H⁺-ATPase, FC-A forms significant van der Waals contacts with the C-terminal valine residue, establishing a hydrogen bond with the carboxylate group via a water molecule (Figure B). − This cooperative binding results in approximately 2 orders of magnitude increase in affinity for both FC-A and the phospholigand compared to their individual binding, despite the inherently weak affinity of FC-A for 14-3-3 (submillimolar K _d).

1.

(A) Chemical structure of fusicoccin A, cotylenin A, and 12-deoxy (12-dFC) and 12-hydroxy (12-hFC) fusicoccin derivatives. (B) Left: Overall crystal structure (1O9F) of the ternary complex of plant 14-3-3 (white surface) bound to C-terminal phosphorylated motif of PM H⁺-ATPase (QSYpTV, green surface) and FC-A (yellow surface). Right: A close-up view of the interface (the area enclosed by the dotted line in the overall structure) between FC-A (pink stick) and PM H⁺-ATPase (green stick). A water molecule is highlighted in a blue sphere.

The therapeutic potential of FC-A and its derivatives stems from the ubiquitous expression and highly conserved structure of 14-3-3 proteins across eukaryotic cells. Humans express seven 14-3-3 isoforms (β, ε, δ, γ, τ, ζ, and σ) that bind thousands of phosphorylated client proteins involved in virtually every cellular process, , including stress responses mediated by nonequilibrium processes such as phase separation. , Dysregulation of 14-3-3 PPIs has been linked to cancers, neurodegenerative diseases, and neuropsychiatric disorders, making FC-A derivatives potentially promising therapeutic tools.

Several groups, including ours, have demonstrated biological activity of FC-A and related analogues in mammalian cells. The FC-A analogue cotylenin A (CN-A, Figure A) induces the differentiation of HL-60 human myeloid leukemia cells and suppresses lung carcinoma growth in interferon-α-pretreated mice. Similarly, FC-A induces cell death in combination with interferon-α, , stimulates central nervous system axon regeneration, , and suppresses glioblastoma cell migration. However, FC-A alone is inactive in inducing cancer cell differentiation or inhibiting their proliferation, highlighting critical structure–activity-relationships.

Structure–activity-relationship studies led by Kato revealed that removal of the 12-hydroxy group from FC-A, inspired by the chemical structure of CN-A, converts the phytotoxin into a potent antiproliferative agent. The resulting semisynthetic derivative 12-deoxyfusicoccin (12-dFC, Figure A) has been shown to inhibit proliferation of MIAPaCa-2 pancreatic carcinoma cells with particular efficacy under hypoxic conditions and significantly suppresses tumor growth in xenograft mouse models without apparent side effects. Importantly, the corresponding 12-hydroxyfusicoccin (12-hFC, Figure A) exhibited no antiproliferative activity despite retaining the ability to stabilize 14-3-3 interactions with mode-3 C-terminal phospholigands, a property shared with the parent compound FC-A. Moreover, labeling of endogenous 14-3-3 with a reactive fluorescent probe derived from 12-dFC in U937 lymphoma cells occurred only under hyperphosphorylation conditions. The collective evidence suggests that the mechanism of action of 12-dFC involves formation of ternary complexes with 14-3-3 and specific phosphorylated proteins distinct from those targeted by FC-A or 12-hFC.

Development of FC-based pharmaceuticals requires a comprehensive understanding of the molecular mechanism in human cells. To this end, we sought to identify the phosphorylated binding target(s) of 12-dFC involved in ternary complex formation with 14-3-3. Using tandem-affinity purification coupled with quantitative mass spectrometry–based proteomics, and employing 12-hFC as a negative control, we report the identification of GRB10-interacting GYF protein 2 (GIGYF2), a largely disordered scaffold protein of the mRNA translational repression machinery, as a key intracellular 14-3-3 binding partner selectively stabilized by 12-dFC, but not by 12-hFC. Our biochemical characterization revealed that this stabilization occurs through cooperative binding at a 14-3-3 consensus motif containing S₅₄₆ near the GYF domain of GIGYF2, resulting in enhancement of translational repression and inhibition of cellular proliferation. Furthermore, we demonstrate that the cellular energy sensor AMP-activated protein kinase (AMPK) mediates S₅₄₆ phosphorylation, which was significantly enhanced under starvation conditions, suggesting an intriguing link between cellular energy stress and translational control via GIGYF2-14-3-3 interactions. These findings highlight the potential of 12-dFC as a robust chemical scaffold for developing molecular glues that target 14-3-3-mediated control of metabolic homeostasis, particularly for mode-1 phosphoproteins, the most commonly found class of 14-3-3 binding partners in the human interactome.

Results

Tandem-Affinity Purification of 14-3-3 Led to the Identification of GIGYF2, to Which Binding of 14-3-3 Is Upregulated by 12-dFC but Not 12-hFC

Isolation of the ternary complex of 14-3-3 bound to phosphorylated protein(s) and a FC derivative is challenging due to the relatively low stability of the resulting complex. Previous attempts to isolate the ternary complex using conventional coimmunoprecipitation (co-IP) with FLAG-14-3-3 were unsuccessful due to the high intensity of background signals in mass analysis, which prevented the identification of target proteins (data not shown).

In this study, we employed a tandem-affinity purification (TAP) methodology using 14-3-3ζ, an essential isoform for human cells, which was conjugated to consecutive FLAG and Strep Tag II peptides at the N-terminus. Tagged 14-3-3ζ was then subjected to two rounds of affinity purification in the presence of either DMSO (control), 12-dFC, or 12-hFC. Briefly, HEK293T cells were transfected with 3× FLAG–Strep Tag II-14-3-3ζ and treated with the respective compounds. Cell lysates were prepared and subjected to immunoprecipitation using agarose beads conjugated to an anti-FLAG antibody. Proteins that eluted with the 3× FLAG peptide were further purified using Strept-Tactin–conjugated sepharose beads and analyzed by SDS–PAGE (Figure A). In the control, a number of proteins aside from 14-3-3 were detected on the gel (lane 2), suggesting successful purification of the phosphorylated binding partners of 14-3-3. Importantly, certain protein bands appeared to be upregulated in the presence of 12-dFC compared with 12-hFC (lanes 3–4), such as the band at approximately 150 kDa (indicated by the black arrow).

2.

Tandem-affinity purification of 14-3-3ζ-bound proteins in the presence of FC derivatives. (A) SDS–PAGE analysis of 14-3-3ζ-bound proteins isolated by tandem-affinity-purification in the presence of a DMSO control or FC derivatives. HEK293 cells were transfected with EGFP or N-terminal 3× FLAG-Strept II-tagged 14-3-3ζ and treated with either DMSO, 12-dFC, or 12-hFC before cells were lysed. Obtained lysates were first treated with agarose beads linked to a monoclonal anti-FLAG antibody, and bound proteins were eluted with 3× FLAG peptide. The proteins were subsequently treated with Strept-Tactin Sepharose beads and eluted with D-desthiobiotin. The purified proteins were separated by SDS-PAGE gel (5–20%) and visualized by silver staining. 14-3-3 (white arrow) and the protein band upregulated in the presence of 12-dFC (black arrow) are highlighted. (B) A plot of “12-dFC12hFC”, the normalized mass intensity (emPAI) in the presence of 12-dFC which was subtracted by the intensity in the presence of 12-hFC of the top 30 proteins (shown with the accession numbers). The data of GIGYF2 is highlighted in red. (C) Validation co-IP of 14-3-3ζ-bound GIGYF2. Cell lysates transfected with 3× FLAG-Strep Tag II-14-3-3ζ and treated with 0.1% DMSO, 12-dFC, or 12-hFC were subjected to TAP purification, and the bound proteins were analyzed by Western blot using anti-GIGYF2 and anti-14-3-3 antibodies, respectively. *Endogenous 14-3-3 (Figure S1).

Quantitative mass analysis of proteins obtained from the control (0.1% DMSO treatment) detected 2,227 bound proteins (Table S1). A comparison of the mass intensity of these proteins in the presence of 12-dFC versus 12-hFC revealed elevated levels in several proteins upon 12-dFC treatment (Figure B). The most abundant protein was 14-3-3ζ, most likely due to stabilization of 14-3-3 dimers bound to phosphorylated ligands in the presence of 12-dFC. Other notably elevated proteins included histone H4, tuberin, and ubiquitin E3 ligase MYCBP2, each predicted to contain mode-1 14-3-3-binding motifs. Among these proteins, we focused on GRB10-interacting GYF protein 2 (GIGYF2), which emerged as the most prominent 14-3-3ζ-interacting protein in the presence of 12-dFC (Figure B, highlighted in red). GIGYF2 is a 150-kDa intrinsically disordered scaffold protein that serves as a central hub in the mRNA translational repression machinery. − This protein has been shown to associate with diverse translational regulatory factors, including cap-binding translation repressor 4EHP, , mRNA-binding proteins such as tristetraprolin (TTP), ZNF598, RNA helicase DDX6, and deadenylase complex CCR4-NOT. GIGYF2 plays critical roles in mammalian development, pro-inflammatory responses, and miRNA-mediated translational repression. Western blot analysis of the proteins bound to the FLAG–Strep-Tag II-14-3-3ζ complex using an anti-GIGYF2 antibody confirmed that 12-dFC significantly promotes the PPI in comparison with the DMSO control, while 12-hFC does not (Figures C and S1).

12-dFC Upregulates the GIGYF2–TTP Interaction and Suppresses Protein Synthesis

Next, we examined how stabilization of the PPI between 14-3-3 and GIGYF2 affects the interactions of GIGYF2 with other binding proteins. Western blotting was used to examine the effect of 12-dFC or 12-hFC on the binding of different proteins to Halo-GIGYF2. These analyses revealed that the PPI between the RNA-binding protein TTP and GIGYF2 was significantly enhanced in the presence of 12-dFC but not 12-hFC (Figure A), in contrast to the effects observed for 4EHP (Figure S2A). Although the signal intensity for DDX6 also increased upon 12-dFC treatment (Figure S2B), the enhancement observed for TTP was more pronounced. These results suggest that the binding of 14-3-3 to GIGYF2 has a marked effect on the recruitment of TTP to GIGYF2, with 12-dFC potentially stabilizing the entire mRNA translational repression complex by enhancing the 14-3-3–GIGYF2 interaction.

3.

12-Deoxy-fusicoccin 12-dFC upregulates GIGYF2–TTP interaction and suppresses protein synthesis. (A) HEK293 cells were transfected by Halo-GIGYF2 and treated with either 0.1% DMSO, 12-dFC, or 12-hFC before cells were lysed. Obtained lysates were treated with magnet beads linked to a Halo-tag, and bound proteins were eluted after cleavage by TEV protease. Proteins were separated by SDS-PAGE gel (5–20%) and analyzed by Western blot using anti-TTP and anti-GIGYF2 antibodies, respectively. (B) BONCAT assay was performed using HEK293 cells grown in methionine-free medium, which were treated with 0.1% DMSO, 12-dFC, or 12-hFC, and incubated. Cells were further treated with azidohomoalanine, and obtained lysates were reacted with biotin-alkyne, separated by SDS–PAGE gel (5–20%). Gels were analyzed by CBB staining and Western blot using streptavidin conjugated with horseradish peroxidase (HRP). These experiments were independently repeated four times, and the average protein band intensities from the Western blots are plotted (Figure S3). The significance of the difference is analyzed using one-way ANOVA test (**P < 0.01). (C) BONCAT assay was performed using HEK293 cells in which GIGYF2 was knocked down or treated with nontargeting control RNA (NTC-siRNA). Proteins were also analyzed by Western blot using anti-GIGYF2 and β-actin antibodies. These experiments were independently repeated four times (Figure S3), and the average values of the intensity of protein bands are plotted. The significance of difference is analyzed using one-way ANOVA with Dunnett’s test (**P < 0.01).

As the zinc finger protein TTP acts as a translational repressor by binding to AU-rich elements (ARE) located in the 3′–untranslated region of target mRNAs and promotes repression and degradation of mRNAs by binding to GIGYF2, we next examined the global effect of 12-dFC on protein synthesis using bioorthogonal noncanonical amino acid tagging (BONCAT), which enables the detection of nascent proteins within a defined time period. Briefly, cells were treated with 0.1% DMSO, 12-dFC, or 12-hFC for 1 h, followed by treatment with the methionine surrogate azidohomoalanine for 2 h. The noncanonical amino acid was incorporated into newly synthesized proteins, which were then conjugated to an alkynyl biotin molecule via azide–alkyne Huisgen cycloaddition and visualized by Western blotting. Although total protein levels in cell lysates remained unchanged upon compound treatment, treatment with 12-dFC resulted in a significant reduction in nascent proteins (Figures B and S3A–C). By contrast, no such reduction was observed in cells treated with 12-hFC, consistent with its lack of an effect on the TTP-GIGYF2 interaction. The inhibitory effect of 12-dFC on protein synthesis was clearly abolished by knockdown of GIGYF2 expression (Figures C and S3D–F). These results demonstrate that 12-dFC–mediated stabilization of the PPI between 14-3-3 and GIGYF2 leads to translational suppression.

12-dFC Downregulates the Expression of Glycolysis-Related Proteins

Having confirmed that 12-dFC suppresses protein synthesis through GIGYF2, we next sought to investigate which group of proteins are affected by 12-dFC. To this end, the global effect of 12-dFC on protein expression was examined through proteomic profiling analysis using two-dimensional difference gel electrophoresis (2D-DIGE). First, the antiproliferative activity of 12-dFC against HeLa cells was tested, revealing an IC₅₀ value of approximately 10 μM (Figure inset), consistent with previously reported activity against MIAPaCa-2 cells (IC₅₀ of ∼11 μM under normoxic conditions). HeLa cells were treated with 20 μM 12-dFC, and lysate proteins were then fluorescently labeled and separated using 2D gel electrophoresis. Changes in the expression levels of proteins between drug-treated and control cells were evaluated. Heatmap analysis was used to compare the expression levels of 296 common protein spots (Table S2) derived from cells treated with 12-dFC and cells treated with 41 well-characterized antiproliferative agents (Figure S4). Hierarchical clustering analysis revealed that 12-dFC exhibited a distinctive regulatory profile, with no apparent similarity to the patterns observed with the other compounds (Table S3), suggesting that the mechanism of action of 12-dFC is unique. Gene ontology analysis of spots showing significant differences in expression (P < 0.05, Student’s t-test, Table S4) using DAVID , identified various glycolytic proteins among those that were downregulated.

4.

Changes in the expression levels of glycolysis-related proteins caused by drug treatment in HeLa cells. Proteins in HeLa cells treated by 12-dFC were analyzed by a similar proteomics profiling method (2D-DIGE) that reported previously (Figure S4). Eight glycolysis-related proteins (24 spots) were selected, and the log ratios of the amount of each spot between drug-treated and control cells were estimated. The sum of the fold ratios was calculated as described in literature, and the data were plotted as a waterfall plot together with the data of each 150 compounds. The bars corresponding to 12-dFC (red), collismycin A, Dp44 mT, and rapamycin (light blue) are highlighted. Inset: Dose–response curve of viability of HeLa cells treated with various concentrations of 12-dFC. Error bars represent SEM (n = 3).

Further comparisons of changes in the expression of selected glycolytic proteins between cells treated with 12-dFC and those treated with a pool of 150 bioactive compounds (Table S5) showed that HIF-1 activators, such as Dp44 mT and collismycin A, significantly upregulated the expression of glycolytic proteins (Figure ). In addition, several kinase inhibitors, abemaciclib, AT9283, and CGP60474, as well as tubulin inhibitors such as nocodazole and TH287, were identified among the top compounds that significantly upregulate glycolytic protein expression (Table S5), suggesting that perturbations in cell cycle regulation and microtubule dynamics affect the expression of glycolysis-related proteins.

In contrast to the above, rapamycin and heat shock protein (HSP) inhibitors notably downregulated the expression of glycolysis-related proteins (Table S5). It is of particular interest that 12-dFC also downregulated glycolytic protein expression, comparable to rapamycin (Figure ), further indicating that 12-dFC downregulates a specific subset of proteins.

12-dFC and 14-3-3 Bind to the GIGYF2 Consensus Motif Involving Phosphorylated S₅₄₆

We next sought to identify the specific binding site of 12-dFC within GIGYF2 that facilitates formation of the ternary complex with 14-3-3. Typically, 14-3-3–binding motifs are categorized into three classes: mode-1 (-RSXpS/TXP-), mode-2 (-RXFXpS/TXP-), and the C-terminal mode-3 (-pS/TX_0–2-CO₂H), in which X represents any amino acid, pS/T represents phosphoserine or threonine, and F refers to any aromatic residue. The mode-1 consensus binding motif is the most common within the human 14-3-3 interactome, as approximately 50% of client proteins contain a proline residue at the i+2 position (where ‘i’ corresponds to the phosphorylation site, pS/T). By contrast, relatively few C-terminal mode-3 motifs have been identified in humans to date.

GIGYF2 consists of 1320 amino acids and harbors a large number of S/T phosphorylation sites. GIGYF2 includes a structured GYF domain that interacts directly with TTP. Database analyses predicted the presence of three mode-1 motifs near the GYF domain: MENS₄₃₈LP, KGVS₅₄₆IP, and RSVS₇₀₅VP (Figures A and S5). The i+1 residues of these motifs are L, I, and V, respectively, which are favorable for formation of the ternary complex with 12-dFC. , Although S₄₃₈ and S₅₄₆ are known phosphorylation sites, phosphorylation of S₇₀₅ has yet to be confirmed. To identify the common binding site between 12-dFC and 14-3-3, GIGYF2 was divided into four regions, and Halo-tagged deletion mutants consisting of pairs of adjacent regions (regions 1-2, 2-3, and 3-4; Figure A) were subjected to pull-down experiments in the presence of 12-dFC and 12-hFC. As shown in Figure B, the amount of 14-3-3 that bound to regions 1-2 and 2-3 in the presence of 12-dFC was higher than the amount that bound in the presence of DMSO or 12-hFC. By contrast, no significant effect was observed for region 3-4. These results suggest that the targeted binding site is located within region 2 and indicated that either the MENS₄₃₈LP, KGVS₅₄₆IP, or both motifs are binding targets for the 12-dFC and 14-3-3 interaction.

5.

Phosphorylated Ser546 of GIGYF2 is the binding site of 12-dFC and 14-3-3. (A) Design of deletion mutants of GIGYF2. Serine residues, S438, S546, and S705, in the 14-3-3 mode-1 motifs predicted by 14-3-3-Pred are highlighted in red. (B) Pull-down experiments from HEK293 cell lysates transfected with Halo-GIGYF2 and treated with 0.1% DMSO, 12-dFC, or 12-hFC (20 μM), respectively. Halo-GIGYF2 and the bound 14-3-3 were detected by Western blot using anti-Halo and anti-14-3-3 antibodies, respectively. The experiment was repeated three times (Figure S6). (C) Pull-down experiments from HEK293 cell lysates transfected with Halo-WT-GIGYF2 or Halo-S438A-S546A-GIGYF2 and treated with 0.1% DMSO or 12-dFC (20 μM). Proteins were separated by SDS–PAGE and detected by Western blot using anti-GIGYF2 and anti-14-3-3 antibodies, respectively. These experiments were repeated four times, and the relative intensity of each band was plotted. The differences were analyzed using Mann–Whitney U test and Student’s t test (*P < 0.05). (D) Co-IP experiments from HEK293 cells cotransfected with 3xFLAG-Strept-Tag II-14-3-3ζ and Halo-WT-, Halo-S438A-, or Halo-S546A-GIGYF2, and treated with 12-dFC (20 μM). Tandem Co-IP from the lysates was performed and bound proteins were detected by Western blot using anti-FLAG and anti-Halo antibodies, respectively. (E) Competitive fluorescence polarization titration of FAM-RGRSWpTY (100 nM, K _d = 0.06 ± 0.01 μM, Figure S7) and 14-3-3 (80 nM) in the absence of (blue circles) and the presence of (red circles) 12-dFC (10 μM) with different conc. of - Left) MENpS₄₃₈LP, and Right) KGVpS₅₄₆IP in HEPES buffer (10 mM, pH 7.3 containing 150 mM NaCl, 0.05% (v/v) Tween20 and 0.1% DMSO). The data were fitted by a 1:1 equilibrium competitive model. Ex = 470 nm, Em = 525 nm, n = 3. Mean values ± SEM are shown.

To confirm that these mode-1 motifs indeed function as binding sites, residues S₄₃₈ and S₅₄₆ were each substituted with alanine. Wild-type Halo-GIGYF2 and the resulting doubly mutated Halo-S438A-S546A-GIGYF2 were subjected to Halo-tag pull-down analysis in the presence of 12-dFC, and the amount of bound 14-3-3 was determined. By expressing Halo-GIGYF2, substantial level of 14-3-3 in the Halo-tag pulldown was detected, which was further increased by 12-dFC (Figures C and S6). These results confirmed that the PPI between 14-3-3 and GIGYF2 is stabilized by 12-dFC. The double mutations markedly diminished the stabilization effect of 12-dFC on the binding of 14-3-3 to GIGYF2 (Figures C and S6), indicating that phosphorylation of S₄₃₈ and/or S₅₄₆ is essential for the stabilization effect. Single-mutation experiments revealed that replacing S₅₄₆ with alanine had more impact on the stabilization effect of 12-dFC than the alanine substitution of S₄₃₈ (Figure D), suggesting that pS₅₄₆ is the primary site mediating cooperative binding to 14-3-3 with 12-dFC.

To further investigate the stabilization effect of 12-dFC on the binding of motifs containing pS₄₃₈ and pS₅₄₆, an in vitro competitive fluorescence polarization binding assay was performed. Briefly, recombinant 14-3-3ζ was first bound to a noncanonical FAM-labeled SWpTY peptide (FAM-RGRSWpTY, K _d = 0.06 ± 0.01 μM, Figure S7). This complex was then titrated with various concentrations of the phosphohexapeptides MENpS₄₃₈LP or KGVpS₅₄₆IP in the presence of 12-dFC, and the dissociation constant (K _d) was determined (Figures E and S7). Notably, no significant change in fluorescence polarization of the MENpS₄₃₈LP peptide was observed, even in the presence of 12-dFC, suggesting that MENpS₄₃₈LP exhibits minimal binding affinity for 14-3-3, which was barely detectable in this assay. By contrast, the KGVpS₅₄₆IP peptide exhibited moderate affinity for 14-3-3 (blue solid line, K _d = 68.7 ± 20.2 μM), and binding of the peptide to 14-3-3 was significantly enhanced in the presence of 12-dFC (10 μM), increasing by nearly 50-fold (red solid line, K _d = 1.41 ± 0.09 μM). These results support that the KGVpS₅₄₆IP mode-1 motif serves as the primary binding site for 12-dFC and 14-3-3 in GIGYF2, resulting in significant stabilization of the PPI via ternary complex formation.

AMPK Mediates Phosphorylation-Dependent GIGYF2–14-3-3 Interaction

Given that 14-3-3 binds to the KGVpS₅₄₆IP consensus motif in GIGYF2, which plays a role in translational repression, identifying the upstream kinase responsible for S₅₄₆ phosphorylation is a critical next step. A comprehensive kinase database analysis of the 12-amino acid sequence surrounding S₅₄₆ in GIGYF2 (RAKGVS₅₄₆IPLMHE) identified AMPK as the top-ranking candidate kinase (Table S6). AMPK functions as a ubiquitous cellular energy sensor that responds to changes in the AMP/ATP ratio. , ADP or AMP binding to the γ subunit induces conformational changes that facilitates T₁₇₂ phosphorylation within the activation loop, thereby promoting catabolic pathways while inhibiting anabolic processes to restore energy homeostasis. Previous studies demonstrated that AMPK depletion in mouse embryo fibroblasts has shown to elevate HIF-1α levels, consistent with the hypothesis that AMPK-mediated GIGYF2 phosphorylation conserves cellular energy through translational suppression.

To test this hypothesis, we investigated whether AMPK activation enhances the GIGYF2–14-3-3 interaction through S₅₄₆ phosphorylation by treating cells with the AMPK activator 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR). As expected, AICAR treatment induced AMPK phosphorylation at T₁₇₂, a hallmark of AMPK activation, and also phosphorylated acetyl-CoA carboxylase (ACC), a well-characterized AMPK substrate (Figures A and S8). Importantly, tandem co-IP of 14-3-3 demonstrated increased GIGYF2 binding to 14-3-3 upon AICAR treatment (Figure B). This AICAR-mediated enhancement was abolished in the S546A mutant (Figures B and S8), confirming the critical role of S₅₄₆ phosphorylation. These results demonstrate that AMPK mediates S₅₄₆ phosphorylation, facilitating 14-3-3 recruitment to the consensus motif and enhancing translational repression, although further investigation is required to determine whether AMPK directly phosphorylates S₅₄₆ or acts through downstream effectors.

6.

Activation of AMPK enhances binding between 14-3-3 and GIGYF2 and mutation of S546A cancels the effect. (A) Upregulation of phosphorylation of AMPK and ACC was observed upon the treatment of AICAR, whereas no significant effect of AICAR was observed in the case of S546A-GIGYF2. Cells were transfected by halo-WT- or -S546A-GIGYF2 and were incubated in the presence of AICAR (1 mM) for 48 h before lysis. Proteins were analyzed by Western blot using anti-pAMPK (Thr172), anti-AMPK, anti-pACC, anti-ACC, and anti-β-actin, respectively. (B) Co-IP of 14-3-3-bound GIGYF2 in the presence of or absence of AICAR. In the case of WT-GIGYF2, AMPK activation by AICAR treatment resulted in upregulation of binding of 14-3-3 to GIGYF2, whereas no significant effect was observed in the case of the S546A mutant. Cells were cotransfected by FLAG-Strep Tag II-14-3-3ζ and halo-WT-, or halo-S546A-GIGYF2. The lysates were subjected to tandem-purification, and the input and bound proteins were analyzed by Western blot. Relative intensity of the bands corresponding to Halo-GIGYF2/14-3-3 were plotted. These experiments were independently repeated three times (Figure S8). (C) HEK293 cells transfected with Halo-WT-GIGYF2 were grown in either normal or amino acid-depleted media for the indicated time periods. Cell lysates were separated by SDS-PAGE and analyzed by Western blot using anti-pS₅₄₆ GIGYF2, anti-Halo, and anti-β-actin antibodies, respectively. The relative intensity of each band was quantified and plotted. These experiments were repeated three times (Figure S12). Statistical differences were analyzed using one-way ANOVA with Dunnett’s test (*P < 0.05, **P < 0.01).

Discussion

Our interactome analysis identified a previously uncharacterized interaction between 14-3-3 and GIGYF2, a critical scaffold in translational repression complexes. The antiproliferative FC derivative 12-dFC significantly stabilized this interaction, while the inactive analogue 12-hFC did not. Through mutagenesis and titration assays, we identified the GIGYF2 mode-1 consensus motif KGVpS₅₄₆IP as the primary binding site for both 12-dFC and 14-3-3.

The molecular basis of this interaction aligns with previous predictions that mode-1 phosphopeptides containing branched aliphatic residues at the i+1 position act as cooperative binding partners for 12-dFC due to favorable van der Waals interactions. , In those studies, isoleucine, valine, and leucine were the best ligands, whereas bulkier or smaller residues such as tryptophan, tyrosine, phenylalanine, and alanine were significantly less effective. These results indicated that precise molecular recognition between 12-dFC and the adjacent i+1 residue enables highly selective complex formation. Indeed, the identified site contains an i+1 isoleucine, and 12-dFC increased GIGYF2–14-3-3 binding affinity by approximately 50-fold at 10 μM (Figure E). The crystal structure of the ternary complex further predicts that the 12-hydroxy group of 12-hFC likely causes steric repulsion with the i+2 proline of the mode-1 motif. Thus, a plausible explanation for the differential activities of 12-dFC and 12-hFC is that 12-dFC stabilizes the ternary complex with 14-3-3 and the mode-1 motif by accommodating both the i+1 isoleucine and the i+2 proline, whereas 12-hFC fails to do so due to steric hindrance from its 12-hydroxy group (Figure A).

7.

Schematic representation of a plausible model for the mechanism of action of antiproliferative activity of 12-dFC. (A) 12-dFC accommodates the i+1 isoleucine and the i+2 proline of the mode-1 consensus motif KGVpS₅₄₆IP in GIGYF2 and forms the ternary complex with 14-3-3, whereas 12-hFC hampers complex formation due to the steric repulsion between the 12-hydroxy group and i+2 proline. (B) Stabilization of 14-3-3 binding to GIGYF2 at pS₅₄₆ by 12-dFC enhances GIGYF2 interaction with TTP, resulting in translational inhibition of TTP-targeted transcripts, including HIF-1α.

Recent efforts have intensively focused on chemical stabilization of 14-3-3 PPIs, and a number of covalent , and noncovalent , synthetic 14-3-3 stabilizers have been developed. While these molecules exhibit high potency for mode-3 interactions, few effectively stabilize mode-1 interactions in cells. An example is the semisynthetic FC derivative DP-005, which has been shown to enhance binding to a mode-1 motif by approximately 120-fold at 100 μM. Direct comparison of 12-dFC and DP-005 on their stabilization effect for the pS546 GIGYF2–14-3-3 interaction was performed by FP titration assay, and 12-dFC was shown to be as potent as DP-005 (Stabilization index: (1.9 ± 0.9) × 10² for DP-005, and (1.6 ± 0.4) × 10² for 12-dFC at 100 μM, Figure S9). These results further validate that 12-deoxy FC derivatives can serve as robust stabilizers of mode-1 motifs. Surprisingly, despite their similar stabilization ability, parallel comparison of their antiproliferative activity revealed that DP-005 exhibits much weaker cytotoxicity (71 ± 2% viability at 90 μM) than that observed for 12-dFC (4 ± 1% viability at 30 μM, Figure S10). These results suggest that the better cellular activity of 12-dFC may result from enhanced stabilization of full-length GIGYF2 (rather than the mode-1 peptide model used in the FP assay) or better cell permeability.

The cellular activity of 12-dFC within a physiologically relevant micromolar concentration range highlights its potential for further development. Notably, 12-dFC is readily derived from the naturally occurring but inactive FC-A by removing the 12-hydroxy group, a simple structural modification that enables accommodation of the mode-1 consensus binding motif. Given that 14-3-3 plays central roles in virtually every biological process, 12-dFC offers a robust and accessible chemical scaffold for developing molecular glues targeting diverse mode-1 14-3-3 binding partners.

Stabilization of the 14-3-3–GIGYF2 interaction by 12-dFC enhanced TTP recruitment to GIGYF2, and BONCAT analysis confirmed that 12-dFC significantly suppressed protein synthesis, an effect abolished by GIGYF2 knockdown. In addition, 2D-DIGE proteomic analysis revealed that 12-dFC treatment significantly reduces glycolytic protein expression. These results demonstrate that stabilizing the 14-3-3–GIGYF2 interaction inhibits translation, likely underlying the antiproliferative effect of 12-dFC against cancer cells (Figure B). As HIF-1α transcripts are targeted by TTP, this model is consistent with the previously reported downregulation of HIF-1α expression by 12-dFC under hypoxic conditions.

Although several protein translation inhibitors have been developed, these agents are often cytotoxic because they target essential components of the eukaryotic translational machinery, including elongation factors and ribosomes. To our knowledge, 12-dFC is the first antiproliferative agent that suppresses translation by selectively stabilizing PPIs within mRNA translational repression complexes. The absence of apparent toxicity in mouse models suggests that stabilizing the translational repression machinery may offer a promising approach for developing less-toxic antitumor therapeutics.

The critical role of the 14-3-3–GIGYF2 interaction involving pS₅₄₆ in translational repression prompted us to identify the kinase that phosphorylates S₅₄₆. Strikingly, our results indicate that AMPK mediates S₅₄₆ phosphorylation (Figure A,B) and that the level of S₅₄₆ phosphorylation is elevated in response to starvation stress (Figure C). These results suggest that GIGYF2-mediated translational repression is associated with the cellular response to energy depletion, potentially bridging metabolic stress with translational regulation.

GIGYF2 and TTP are highly disordered proteins, as 76 and 58% of the respective sequences are predicted to be disordered (Figure S14). Computational modeling of 14-3-3, GIGYF2, and TTP suggested that the KGVpS₅₄₆IP motif of GIGYF2 is located near the GYF domain, which binds the “PPPPG” motif in TTP (Figure S15). Moreover, TTP harbors two 14-3-3–binding motifs located adjacent to the PPPPG motif. , While these 14-3-3 binding motifs are unlikely to accommodate 12-dFC binding, the dimeric nature of 14-3-3 suggests that one monomer could bind GIGYF2 in a 12-dFC-stabilized manner while the other engages TTP, potentially serving as a molecular scaffold to facilitate their interaction and enhance translational repression.

Notably, p38-dependent MAP kinase–activated protein kinase 2 (MK2) has been reported to be activated by UV-induced stress and phosphorylates S₁₅₇ and S₆₃₈ of GIGYF1, a GIGYF2 homologue. This phosphorylation promotes 14-3-3 binding and excludes GIGYF1 from processing bodies (P-bodies), which are key sites of mRNA decay and translational regulation. Given that these phosphorylation sites are conserved in GIGYF2 (S₁₆₀ and S₇₀₅), it is possible that MK2 may similarly phosphorylate GIGYF2 in response to oxidative stress, thereby promoting 14-3-3 binding.

Moreover, studies have revealed that under oxidative stress conditions, MK2 phosphorylates S₅₂ and S₁₇₈ of TTP, promoting 14-3-3 binding and sequestration of TTP from stress granules (SGs), thereby inhibiting TTP-mediated degradation of ARE-containing mRNAs, , which are critical for the post-transcriptional regulation of pro-inflammatory cytokine expression. Together, these observations highlight the differential regulation of GIGYF2 and TTP by distinct kinases under different stress conditions.

Based on these observations together with our findings, we propose a model in which distinct stress signals differentially modulate translational repression via GIGYF2, 14-3-3, and TTP. Under energy-depleted conditions, AMPK phosphorylates S₅₄₆ of GIGYF2, recruiting 14-3-3 to facilitate TTP binding and enhance translational repression. This mechanism would conserve cellular ATP by suppressing energy-intensive protein synthesis. In contrast, during oxidative stress, MK2 phosphorylates TTP and potentially GIGYF2 at different sites, promoting 14-3-3 binding that may sequester these proteins from SGs/P-bodies, thereby potentially enabling stress-response gene translation while inhibiting degradation of ARE-containing mRNAs. This context-dependent regulation suggests that cells fine-tune translational control through stress-specific kinase activation and 14-3-3-mediated protein interactions, though further validation will be necessary to fully elucidate this proposed mechanism.

Our identification of GIGYF2 as a 12-dFC target with clear functional consequences (Figures –) provides a foundation for future comprehensive target profiling, which will be essential for understanding the broader biological effects of FC-based molecular glues and guiding their development.

Conclusions

In conclusion, this study revealed 12-dFC as a molecular glue stabilizer that suppresses translation by stabilizing the 14-3-3-GIGYF2 interaction within mRNA translational repression complexes, leading to antiproliferative activity. Furthermore, the study identified a stress-responsive phosphorylation site in GIGYF2 that is mediated by the cellular energy sensor AMPK, supporting a regulatory link between metabolic stress sensing and translational control. These results shed light on the critical roles of intrinsically disordered phosphoproteins in dynamic cellular processes regulating energy homeostasis and suggest that chemical stabilization of 14-3-3 PPI interfaces could provide promising avenues for developing new therapeutic approaches targeting stress-responsive signaling pathways. Work toward this goal is currently underway in our laboratory.

Glossary

Abbreviations

FC

fusicoccin

CN-A

cotylenin-A

GIGYF2

GRB10-interacting GYF protein 2

AMPK

AMP-activated protein kinase

PPIs

protein–protein interactions

co-IP

coimmunoprecipitation

TAP

tandem-affinity purification

TTP

tristetraprolin

BONCAT

bioorthogonal noncanonical amino acid tagging

2D-DIGE

two-dimensional difference gel electrophoresis

AICAR

5-aminoimidazole-4-carboxamide ribonucleotide

ACC

acetyl-CoA carboxylase

hypoxia-inducible factor-1

HIF-1

MK2

p38-dependent MAP kinase–activated protein kinase 2

SG

stress granules

P-body

processing bodies.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.5c01403.

• General procedures; plasmid construction; cell culture and transfection; tandem-affinity purification of 4× FLAG-Strep-Tag II-14-3-3ζ; quantitative mass spectroscopy; pull-down experiments of Halo-GIGYF2; immunoblot; BONCAT assay; proteomic profiling using two-dimensional difference gel electrophoresis (2D-DIGE); co-immunoprecipitation of deletion mutants of Halo-GIGYF2; co-immunoprecipitation of Halo-S438A-S546A-GIGYF2; co-immunoprecipitation of 4× FLAG-Strep-Tag II-14-3-3ζ from cells co-transfected with Halo-S438A-GIGYF2 or Halo-S546A-GIGYF2; expression and purification of recombinant 14-3-3ζ; fluorescence polarization titration assay; co-immunoprecipitation of 4× FLAG-Strep-Tag II-14-3-3ζ from AMPK-activated cells cotransfected with Halo-WT-GIGYF2 or Halo-S546A-GIGYF2 AM; anti-pS546 GIGYF2 antibody time-course changes in the phosphorylation level at S546 in GIGYF2 under starvation conditions; Figures S1–S17; Table S6; and references and notes (PDF)

• List of proteins identified in quantitative affinity purification mass analysis using 3× FLAG-Strep Tag II-14-3-3ζ in the presence of 12-dFC, 12-hFC, or 0.1% DMSO (Table S1) (PDF)

• Expression of 296 spots of 12-dFC treated HeLa cells (PDF)

• Top 10 compounds similar to 12-dFC against 150 reference compounds in proteomic profiling (PDF)

• Spots showing significant differences (p < 0.05, Student’s t test) in 2D-DIGE analysis of 12-dFC treated HeLa cells (PDF)

• Changes in the expression levels of glycolysis-related proteins caused by drug treatment in HeLa cells (PDF)

• ID mapping of enriched proteins by 12-dFC (PDF)

N.O., R.M., and S.I. contributed equally to this work. N.O. and R.M. conducted coimmunoprecipitation and validation. N.O. and S.I. performed BONCAT assays and siRNA experiments. N.O. conducted experiments using the AMPK activator. S.I. prepared deletion mutants and conducted coimmunoprecipitations. M.A. and A.M. conducted fluorescent polarization (FP) titration and coimmunoprecipitation. I.K. designed and constructed DNA plasmids and conducted quantitative mass analysis-based proteomics analysis. R.N. performed quantitative mass proteomics analysis. Y.H. synthesized fusicoccin derivatives and assisted FP titration assays and data analysis. M.M. and H.O. conducted 2D DIGE and analyzed data. K.M. and M.Y. assisted with siRNA experiments. M.A., A.M., and Y.I. performed assays under energy depletion. M.Y. and K.M. assisted with editing manuscript. J.O. conceptualized, designed, and supervised the research and wrote the manuscript with input from all the authors. The manuscript was reviewed and edited by all the authors.

This work was supported by the JSPS KAKENHI (21H02077, 18H02106, 20H04769, 26102727, and 24102519 to J.O., 24K08739 and 20H05620 to M.M., 21H02434 and 22H04922 to K.M., 21H04720 and 23H04885 to H.O., and 23H05473 and 23H04882 to M.Y.), Daiichi Sankyo Foundation of Life Science, Astellas Foundation for Research on Metabolic Disorders, and Uehara Memorial Foundation (to J.O.).

The authors declare no competing financial interest.