Design, expression, purification and crystallization of human 14-3-3ζ protein chimera with phosphopeptide from proapoptotic protein BAD

Highlights

• 14-3-3ζ protein chimera with the BAD peptide is designed, purified and crystallized.

• BAD peptide SRHSS⁷⁵YPAGTE is phosphorylated upon co-expression with PKA in E.coli.

• Tethered phosphorylated BAD peptide occupies the amphipathic grooves of 14-3-3ζ.

• Efficient purification of the untagged 14-3-3ζ-BAD75 chimera is demonstrated.

• The assembled chimera readily crystallizes in different chemical environments.

Abstract

14-3-3 protein isoforms regulate multiple processes in eukaryotes, including apoptosis and cell division. 14-3-3 proteins preferentially recognize phosphorylated unstructured motifs, justifying the protein-peptide binding approach to study 14-3-3/phosphotarget complexes. Tethering of human 14-3-3σ with partner phosphopeptides via a short linker has provided structural information equivalent to the use of synthetic phosphopeptides, simultaneously facilitating purification and crystallization. Nevertheless, the broader applicability to other 14-3-3 isoforms and phosphopeptides was unclear. Here, we designed a novel 14-3-3ζ chimera with a conserved phosphopeptide from BAD, whose complex with 14-3-3 is a gatekeeper of apoptosis regulation. The chimera could be bacterially expressed and purified without affinity tags. Co-expressed PKA efficiently phosphorylates BAD within the chimera and blocks its interaction with a known 14-3-3 phosphotarget, suggesting occupation of the 14-3-3 grooves by the tethered BAD phosphopeptide. Efficient crystallization of the engineered protein suggests suitability of the “chimeric” approach for studies of other relevant 14-3-3 complexes.

Introduction

The eukaryotic 14-3-3 protein family comprises ~30-kDa polypeptides that assemble into homo- and heterodimers performing various roles in the regulation of apoptosis, cell division, transcription, signal transduction and other important cellular processes [1,2]. Such rich functionality is achieved due to rather promiscuous binding of 14-3-3 to hundreds of target proteins with phosphorylated serine or threonine in 14-3-3 binding motifs [1,3,4]. These motifs tend to be located in the intrinsically disordered regions, which facilitate accommodation of the phosphopeptides in the amphipathic grooves (AGs) of 14-3-3 [5,6]. Although pioneering studies outlined binding preferences of certain phosphorylated peptide sequences, revealing at least three canonical 14-3-3 binding motifs (I-III) [4,7], the constantly expanding set of experimentally confirmed 14-3-3 binding sequences helped realize that those sequence determinants are not as strict as previously thought. Indeed, there are now many confirmed 14-3-3 binders that lack the canonical binding sequences, which complicates the prediction of 14-3-3 binding sites [8]. As the interactome of 14-3-3 proteins is estimated to exceed 1,000 proteins [6,9], and some of them may be polyvalent 14-3-3 binders [10], there is urgent need to systematically analyze as many as possible 14-3-3-bound phosphopeptide conformations.

Not surprisingly, the wide interactome implicates 14-3-3 proteins in the development of many human disorders, including neurodegenerative, developmental and oncological abnormalities. In this sense, the modulation of the 14-3-3 interaction with phosphorylated target proteins is recognized as a big opportunity for drug discovery and therapy [11,12]. Toward this goal, 14-3-3/phosphotarget complexes must be well characterized structurally. Yet the 14-3-3 interactome remains disproportionally explored [8] – among hundreds of experimentally confirmed 14-3-3 protein-protein interactions (PPIs) [13], only a minority has been structurally characterized with phosphopeptides, and even fewer have been with full protein partners (e.g., AANAT [14], HSPB6 [15], Hd3a [16], Nth1 [17], ExoS/ExoT [18] and B-RAF [19]).

Several 14-3-3 isoforms are usually expressed per organism and there are seven 14-3-3 isoforms in humans (β, ζ, γ, ε, σ, τ, η) [20]. Although sharing a conserved and unique α-helical fold [21,22], various 14-3-3 isoforms reveal distinct preferences towards dimerization with other isoforms and may also display some specialization [6,20,23]. For example, 14-3-3σ almost exclusively homodimerizes, whereas 14-3-3ε prefers to heterodimerize with other 14-3-3 isoforms [21,24,25]. The rest isoforms are intermediate in this ability [21,25]. Likewise, 14-3-3 isoforms appear to differ in terms of their regulation by posttranslational modifications [26,27]. For example, phosphorylation of a semi-conserved residue Ser58 located in the dimer interface by a range of protein kinases including PKA was shown to promote 14-3-3 monomerization [[28], [29], [30], [31], [32]], which in turn may interfere with the phosphopeptide-binding ability directly [33] or indirectly [34]. Ser58 is replaced by alanine in 14-3-3σ and 14-3-3τ, which makes them unphosphorylatable at this position.

The remarkable structural element of 14-3-3 proteins is their intrinsically disordered C-terminal tail of 12–23 residues. This element is the most variable segment in 14-3-3 proteins. One of its functions in 14-3-3ζ was proposed to be an autoinhibitory role based on its transient interaction with the AGs of 14-3-3, potentially excluding undesired binding of weak and unspecific ligands [22,35]. Curiously, the C-terminal segment of Cryptosporidium parvum 14-3-3 was found spontaneously phosphorylated and bound in the AG (PDB ID 3EFZ [36]) (Fig. 1A). Later on, this serendipitous discovery inspired the design of the first artificial chimera of human 14-3-3σ with phosphorylated fragments from partner proteins, to streamline purification and crystallization of 14-3-3/phosphopeptide complexes at a fixed protein-peptide stoichiometry. These advantages remove many challenges associated with traditional peptide co-crystallizations, such as low peptide solubility, weak binding affinity and incomplete occupancy [37]. In addition, the 3EFZ crystal structure provided information on the length of the linker, because all residues of the phosphorylated C-terminal segment of C. parvum 14-3-3 were crystallographically resolved, with no apparent constraints for the phosphopeptide binding in the AG [36] (Fig. 1A). The prototype 14-3-3 chimera was successfully used to determine crystal structures of at least six 14-3-3/phosphopeptide complexes [37,38]. However, it was based on 14-3-3σ only and the question remained whether studies on other 14-3-3 isoforms can benefit from this approach, which would open up access to all the 14-3-3 isoform-specific client interactions.

Here, we expanded this approach by designing a chimera of a heterodimerizable 14-3-3 isoform, 14-3-3ζ. As a partner peptide we have chosen the conserved Ser75 (human BAD numbering, equivalent to Ser112 in mouse) phosphopeptide of the death agonist protein BAD. BAD is a well-known 14-3-3 partner phosphorylatable at several positions by PKA [39], PAK1 [40], PKB/Akt [41,42] and RSK1 [43] kinases. Phosphorylated BAD is sequestered by 14-3-3 in the cytoplasm until pro-apoptotic stimuli dissociate 14-3-3 from BAD, releasing it for translocation to the mitochondrial membrane, association with other members of the BCL-2 family and triggering of apoptosis [44,45]. The 14-3-3/phospho-BAD complex thus represents a gatekeeper of the pro- and anti-apoptotic regulation [[46], [47], [48], [49], [50], [51], [52], [53]] and an attractive target of drug discovery and therapy to modulate the balance in apoptotic and pro-survival signaling in a broad spectrum of pathological conditions [[54], [55], [56], [57]]. Intriguingly, despite its huge significance for biomedicine, the structure of the 14-3-3/BAD complex has not been resolved. As the first step in this direction, here we report on successful design, bacterial expression, untagged purification and efficient crystallization of the new 14-3-3ζ-BAD75 chimera and discuss further improvements and application of 14-3-3 chimeras in structural studies.