The 14-3-3 proteins are a family of highly conserved and abundant eukaryotic adapter proteins that regulate at least 300 binding partners in vital physiological processes  , including DNA damage response, transcriptional trafficking, apoptosis, and even neurodegenerative Alzheimer’s disease (AD)  . 14-3-3(s) form homo- and/or heterodimer that constitute of seven 14-3-3 genes (i.e., β, γ, ε, σ, ζ, τ, η) in mammals  and 15 genes in the plant  . All isoforms recognize two phosphorylated-dependent binding motifs with phosphorylation of serine or threnine residues: RSXpSXP (model 1) and RXXXp(S/T)X[PLM] (model 2)  . Nevertheless, 14-3-3(s) might also interact with non-phosphorylated peptide, such as ChREBP-α2 helix  , p190RhoGEF  , and metabolites, such as AMP generated during fatty acid metabolism  . Despite the targets contain phosphorylation modification or not, they all bind to the same amphipathic groove on 14-3-3  .
The function of 14-3-3(s) may usually be classified into three distinct modes. The more common one is the induction of conformation change from phosphorylated binding partner imposed mechanically by rigid 14-3-3(s), resulting in the inter-/intra-compartmental sequestration of target protein itself, which supports the “molecular anvil hypothesis”  . For instance, the cell-cycle regulator Cdc25B recruits a 14-3-3 monomer which binds to the high-affinity site named “gatekeeper”, allowing the other, low-affinity site engaging the second monomer  . Changing subcellular localization (i.e., nuclear/cytosol trafficking) is anther common mode. For example, 14-3-3ζ regulates nuclear trafficking of protein phosphatase 1α (PP1α) in human cell  . Also 14-3-3(s) can act as a scaffold protein to anchor target proteins to one another  . Despite 14-3-3(s) have been implicated in diverse diseases, the direct role of up- or down-regulated 14-3-3(s) function in human diseases is not well known yet. Examples are as follows: (i 14-3-3σ has been implicated in AD and breast cancer    ; (ii 14-3-3ζ/σ has been implicated in Parkinson’s disease  ; (iii 14-3-3γ in the cerebrospinal fluid (CSF) could be used as a marker for patient with Creutzfeldt-Jakob disease  and so on.
14-3-3β interacts directly with Ser196-phosphorylated ChREBP (inactive), an insulin-independent transcription factor, with high affinity to stabilize ChREBP under low circulating blood glucose and high glucagon level, resulting ChREBP retained cytosolically or exported out of the nucleus  . As the glucose level rises (e.g. taking a meal), the ChREBP was dephosphorylated by a specific protein phosphatase (PP2A-ABδC), which leads to dissociation of 14-3-3β, followed by nuclear localization by binding to importins, and transcriptional activation of glycolytic enzymes and all of the lipogenic enzymes. Hence 14-3-3β function as an “on/off” switch to control the nucler/cytoplastic trafficking of ChREBP. This pathway could origin back to increased concentration of the pentose shunt intermediate xylulose 5-P (Xu5P)  . In addition, a new metabolite, identified as AMP, was recently reported to specifically target the interaction between ChREBP and 14-3-3β  . Together with ketone bodies, AMP therefore inhibits lipogenesis by restricting localization of ChREBP to the cytoplasm. However, the structural details of the mechanism for glucose metabolites regulation are not clear. Yet, to our knowledge, no other metabolite intermediate from Glycolysis was reported to bind 14-3-3β in biochemistry or structure biology to date.
Here we reported the crystal structures of mammalian 14-3-3β protein in its apo and a Tricarboxylic Acid Cycle intermediate Malte bound form. The binding site of Malte is located in a non-canonical binding pocket surround by a complicated electrostatic and hydrogen interface. As a first step in the directly observed structure model of Glycolysis metabolite interaction with 14-3-3β, the work provides a structural frame for crosstalk of insulin-independent de novo fat synthesis in the liver and energy production in the citric acid cycle through acetyl-CaA formed during oxidation of fatty acids.
2. Materials and Methods
2.1. Protein Cloning, Expression and Purification
DNA for the full-length mouse 14-3-3β was amplified by polymerase chain reaction (PCR). The PCR genes were cloned into vector pHis-Parallel1 plasmid  and expressed in Escherichia coli strain BL21 (DE3) (Novagen) with an N-terminal 6X His tag. Harvested cells were lysed by a pressure homogenizer in a buffer containing 50 mM Tris, pH 8.0, 300 mM NaCl, 20 mM imidazole, 5 mM β-mercaptoethanol, 1 mM PMSF, and 0.03% (v/v) Brij-5. The clarified cell lysate was incubated with His60 Ni beads (Clontech) at 4˚C for one hour, and the bound protein was eluted with liner gradient of 20 - 500 mM imidazole. The pooled fractions were desalted into 50 mM Tris, pH 8.0, 200 mM NaCl, 2 mM DTT. The 6x His tags were removed by TEV protease overnight at room temperature. The cutted 14-3-3β was further purified using a HiTrap Q column (GE Healthcare Life Sciences) followed by Superdex75 gel filtration chromatography (GE Healthcare Life Sciences) and eluted in 50 mM Hepes pH7.5, 100 mM NaCl, and 2 mM DTT. The protein was then concentrated to 61.5 mg/ml, aliquoted, and flash frozen in liquid nitrogen and stored in −80˚C.
2.2. Crystallization and Data Collection
14-3-3β crystals were grown at 20˚C using the hanging drop vapor diffusion method. For crystallization, 14-3-3β was diluted to 44 mg/ml. The reservoir solution for Apo protein (spacegroup P212121) consisted of 2.2 M Sodim Malonate pH 7.0 and was mixed with protein in a 1:1 ratio. For Malate-bound crystal (spacegroup C2), the reservoir solution consisted of 2.1 M DL-Malic acid pH 7.0 and was mixed with protein in a 1:1 ratio. Crystal of Apo protein was transferred to a cryoprotectant solution of 50 mM HEPES pH 7.5, 100 mM NaCl, 2 mM DTT, 2.4 M Sodim Malonate and 25% Glycerol before flash frozen. Crystal of ligand-bound protein was harvested freshly and flash frozen in liquid nitrogen.
2.3. Structure Determination and Refinement
Both crystals were collected at Beamline 19-ID of Advanced Photon Source (APS), Argonne National Laboratory, Lemont, IL. The diffraction images were integrated, merged and scaled using HKL3000 software package  . Initial phases were obtained by the molecular replacement method performed using PHENIX Suite  . Manual model building was performed using Coot  . Further structure refinement was performed using REFMAC5  . The crystal data and final refinement statistics were summarized in Table 1. All figures were drawn using the program PyMOL (The PyMOL Molecular Graphics System, Version 1.7.4 Schrödinger, LLC). The coordinates of 14-3-3β in two forms have been deposited in the Protein Data Bank (Protein Data Bank code)  .
3.1. Structure of Apo 14-3-3β
The overall structural feature of the Apo protein14-3-3β is W-like homodimer with an aperture existing at the central dimeric interface (Figure 1(a)), similar
Figure 1. Structure of mouse 14-3-3β in Apo form. (a) Cartoon diagrams of the apo mouse 14-3-3β dimer looking down the canonical peptide binding groove (grey), showing classical “close-open” conformation; (b) The superimposition of m14-3-3βs (2BQ0, red); (c) The superimposition of two monomers in apo 14-3-3β emphasizes the C-terminal flexibility (gray, violet); (d) Close up of positively charged patch in groove that interacts with peptide. 14-3-3β is superimposed to PDB 5F74 (Cyan). Residues are shown as red sticks, peptide from PDB 5F74 is colored yellow.
to previously reported structures such as PDB ID 2BQ0 (Figure 1(b)). As expected, one monomer was captured in the closed conformation, whereas the opposite one adopts an open state following a 20˚ rotation of the α7 to α9 helices, leading to a shallow and exposed groove (Figure 1(a), Figure 1(c)). Each monomer displayed the typical 14-3-3 structural organization, which in general consist of nine helices (α1 - α9) arranged in a right-handed superhelical bundle, and organized into groups of two, two, two, and three helice (Figure 1(a)). The first four alpha helices contribute to formation of the dimer, and the subsequent helices always pack in an antiparallel manner. Helices α3, α5, α7, and α9 are essential to form the conserved peptide-binding groove located at the concave face of the protein, with dimension about 35 Å (broad) × 35 Å (wide) × 20 Å (deep) (Figure 1(c))  . One side of the groove contains positively charged patch formed by a conserved triad of Arg-58, Arg-129, Tyr-130 (compare to Arg-57, Arg-130 and Tyr-131 in ε isoform) that binds to the phosphate group or non- phosphorylated peptide (Figure 1(d)), whilst a hydrophobic patch exist on the other side. Upon superposition, 14-3-3β can be aligned with most homologous structures. For example, structural comparison of the root mean square deviations between Cα atoms of 14-3-3β with peptide-bound structure (PDB ID: 5F74) is only ~0.35 Å with the program CCP4MG  (Figure 1(c)). This indicates that ethier apo 14-3-3β or complex with peptide adopts essentially identical conformation.
3.2. Structure of Malate-Bound 14-3-3β
Tricarboxylic Acid Cycle intermediate malate was reported in a shuttle between mitochondrion and cytosol, resulting a far more effective pathway in producing NADPH in the cytosol than the pentose phosphate pathway   . We here crystallized 14-3-3β together with malate and obtained a 1.97 Å resolution structure (Table 1). Striking, 14-3-3β homodimer adopts a double-closed conformation in the malate bound state (Figure 2(a)), suggesting that 14-3-3β region encompassing helices α7 and α9 possess a degree of flexibility which contributes to the binding groove changing from an “open” to a “closed” state. Structural comparison of two monomers yields a root mean square deviation of ~0.96 Å for 228 Cα atoms (Figure 2(b)), indicating that helices α1 - α9 are largely superimposable. To analyze conformation change on malate binding, we compare the C terminal helices with other structurally elucidated 14-3-3 canonical closed-open state (PDB ID 2BQ0). Our structure showed that the helices α8 and α9 of monomer A slightly left shift, e.g. 1.9 Å of Leu218 residue on helix α9, but still keeping a closed conformation. However, the C-terminal helical orientation of monomer B displayed significantly flexible difference. The residues, such as Tyr180, Pro187, and Trp230, have been compacted to each other with respect to the superimposed model in open state. The conformational flexibiligty observed here help to explain the mechanism in recognition of diverse peptides with varying size and sequencing by 14-3-3 isoforms.
Table 1. Data collection and refinement statistics.
aValues in parentheses are for the highestresolution shell; bEvaluated by MolProbity  .
3.3. The Malate Binding Site
Generally, the protein 14-3-3 binds to targets through a highly conserved peptide- binding groove with a positively charged patch of Arg-58, Arg-129, Tyr-130 at one side and a hydrophobic patch on the other side (Figure 1(d)). Our malate- bound structure revealed an unexpected pocket located in the floor vicinity of 14-3-3β helices α2, α3, α4, which distinguishes from the above canonical binding site mainly arranged by helices α5, α7, α9 (Figure 3(a)). The observed cavity accommodates the malate perfectly. A close inspection of the interface between 14-3-3β and malate revealed the ligand interacts directly with amino acid resi-
Figure 2. Structure of mouse 14-3-3β in Malate-bound form. (a) Two views of 14-3-3β N-terminal dimer (green, blue); (b) The superimposition of two monomers; (c) Details of residue shift between C-terminal helices. Residues are shown as sticks. The red color model is PDB ID 2BQ0.
dues His36, Leu38 (loop between α2 and α3), Try106, Leu107, and Asn110 (α4) through mainly electrostatic and hydrogen interactions. The hydroxyl group on C1 of malate also interacts with Thr32 (α2) through a water molecule. Two more water forms classical hydrogen bonds with oxygen atoms on C4 of malate (Figure 3(b) and Figure 3(c)) with range 2.7 Å - 2.9 Å. The calculated protein-ligand buried interface is 106.9 Å2 per monomer.
4. Discussion and Conclusions
We determined the crystal structures of the mouse 14-3-3β in Apo and complex with a Tricarboxylic Acid Cycle intermediate malate at 1.65 Å and 1.97 Å, respectively. The structure of 14-3-3 in malate-bound form reveals a non canonical binding pocket, located in the floor vicinity of 14-3-3β helices α2, α3, α4, which was the first time observed in all 14-3-3 family proteins. Intriguingly, malate binding to 14-3-3β contributes to the conformation change from “close-open” state in Apo to “close-close” state in complex, raising the possibility that the alteration by malate might regulate the interaction between target proteins like ChREBP with 14-3-3s.
ChREBP is a large transcription factor to activate expression of all the lipogenic
Figure 3. Interaction between non-canonical binding pocket of m14-3-3β with Malate. (a) A ribbon diagram of m14-3-3 dimers with malate shown sticks; (b) Malate binds to a novel groove of 14-3-3β through electrostatic inerations; (c) Detailed view of the malate binding interface. Polar and hydrogen-bond contacts are depicted as purple dotted lines, and residues involved in contacts are shown as sticks.
enzymes at high glucose level, and in liver, independent to insulin effects  . 14-3-3β and/or importins-α binding to ChREBP play important roles in nuclear/cytosol trafficking pathway. Several metabolites were reported to inhibit nuclear localization of ChREBP, such as AMP, β-hydroxybutyrate, and acetoacetate in hepatocytes  . The finding that Tricarboxylic Acid Cycle intermediate malate binds to 14-3-3β in our structure highlighted a role for malate in regulating the subcellular localization of ChREBP. In support of this notion, the angle of three C-terminal helices of Apo structure which could superimpose perfectly with 14-3-3β: ChREBP-α2 complex (Figure 1(d)) undergoes a conformational change upon malate binding to “close-close” state.
The metabolism of glucose via mitochondrial pathways plays an essentially role in energy metabolism and redox homeostasis. A Pyruvate Malate shuttle operating system was reported across the mitochondrial membrane, where the malate exits the mitochondrion and in cytosol, and it is decarboxylated to pyruvate then re-enters mitochondrial pools in insulin cells  . How this shuttle interacts with insulin-independent ChREBP pathway in hepatocyte is not clear. Additionally, Glucose 6-phosphate was reported to be required for ChREBP activation in response to glucose in the liver  . So far, such interplay between signaling metabolite and ChREBP activation is still not fully understood. Knowledge of the structure interaction mechanism between 14-3-3β and metabolites accounting for ChREBP responsiveness to glucose is critical for understating glucose-dependent ChREBP activation. In hence, our Apo- and Tricarboxylic Acid Cycle metabolite malate-bound crystal structures with “on/off” switch function to control the ChREBP trafficking represent the first step toward this goal.
We thank Drs. Diana Tomchick and James Chen for help with synchrotron data collection. Results shown in this report are derived from work performed at Argonne National Laboratory, Structural Biology Center at Advanced Phonton Source. Argonne is operated by UChicago Argonne, LLC, for the U.S. Department of Energy, Office of Biological and Environmental Research under contract DE-AC02-06CH11357.
Atomic coordinates and structure factors for the reported crysal structures have been deposited in the Protein data Bank under accession numbers 5WFU and 5WFX.
Z.H. designed experiments and wrote the manuscript, X. L. performed experiments, L. S. analyzed the data.