Glioblastoma mutations alter EGFR dimer structure to prevent ligand bias.

The epidermal growth factor receptor (EGFR) is frequently mutated in human cancer1,2, and is an important therapeutic target. EGFR inhibitors have been successful in lung cancer, where mutations in the intracellular tyrosine kinase domain activate the receptor1, but not in glioblastoma multiforme (GBM)3, where mutations occur exclusively in the extracellular region. Here we show that common extracellular GBM mutations prevent EGFR from discriminating between its activating ligands4. Different growth factor ligands stabilize distinct EGFR dimer structures5 that signal with different kinetics to specify or bias outcome5,6. EGF itself induces strong symmetric dimers that signal transiently to promote proliferation. Epiregulin (EREG) induces much weaker asymmetric dimers that drive sustained signalling and differentiation5. GBM mutations reduce the ability of EGFR to distinguish EREG from EGF in cellular assays, and allow EGFR to form strong (EGF-like) dimers in response to EREG and other low-affinity ligands. Using X-ray crystallography, we further show that the R84K GBM mutation symmetrizes EREG-driven extracellular dimers so that they resemble dimers normally seen with EGF. By contrast, a second GBM mutation, A265V, remodels key dimerization contacts to strengthen asymmetric EREG-driven dimers. Our results argue for an important role of altered ligand discrimination by EGFR in GBM, with potential implications for therapeutic targeting.

Structural Insights into Pseudokinase Domains of Receptor Tyrosine Kinases.

Despite their apparent lack of catalytic activity, pseudokinases are essential signaling molecules. Here, we describe the structural and dynamic properties of pseudokinase domains from the Wnt-binding receptor tyrosine kinases (PTK7, ROR1, ROR2, and RYK), which play important roles in development. We determined structures of all pseudokinase domains in this family and found that they share a conserved inactive conformation in their activation loop that resembles the autoinhibited insulin receptor kinase (IRK). They also have inaccessible ATP-binding pockets, occluded by aromatic residues that mimic a cofactor-bound state. Structural comparisons revealed significant domain plasticity and alternative interactions that substitute for absent conserved motifs. The pseudokinases also showed dynamic properties that were strikingly similar to those of IRK. Despite the inaccessible ATP site, screening identified ATP-competitive type-II inhibitors for ROR1. Our results set the stage for an emerging therapeutic modality of "conformational disruptors" to inhibit or modulate non-catalytic functions of pseudokinases deregulated in disease.

Structural basis for ligand reception by anaplastic lymphoma kinase.

The proto-oncogene ALK encodes anaplastic lymphoma kinase, a receptor tyrosine kinase that is expressed primarily in the developing nervous system. After development, ALK activity is associated with learning and memory1 and controls energy expenditure, and inhibition of ALK can prevent diet-induced obesity2. Aberrant ALK signalling causes numerous cancers3. In particular, full-length ALK is an important driver in paediatric neuroblastoma4,5, in which it is either mutated6 or activated by ligand7. Here we report crystal structures of the extracellular glycine-rich domain (GRD) of ALK, which regulates receptor activity by binding to activating peptides8,9. Fusing the ALK GRD to its ligand enabled us to capture a dimeric receptor complex that reveals how ALK responds to its regulatory ligands. We show that repetitive glycines in the GRD form rigid helices that separate the major ligand-binding site from a distal polyglycine extension loop (PXL) that mediates ALK dimerization. The PXL of one receptor acts as a sensor for the complex by interacting with a ligand-bound second receptor. ALK activation can be abolished through PXL mutation or with PXL-targeting antibodies. Together, these results explain how ALK uses its atypical architecture for its regulation, and suggest new therapeutic opportunities for ALK-expressing cancers such as paediatric neuroblastoma.

Structural basis for EGFR ligand sequestration by Argos.

Members of the epidermal growth factor receptor (EGFR) or ErbB/HER family and their activating ligands are essential regulators of diverse developmental processes. Inappropriate activation of these receptors is a key feature of many human cancers, and its reversal is an important clinical goal. A natural secreted antagonist of EGFR signalling, called Argos, was identified in Drosophila. We showed previously that Argos functions by directly binding (and sequestering) growth factor ligands that activate EGFR. Here we describe the 1.6-A resolution crystal structure of Argos bound to an EGFR ligand. Contrary to expectations, Argos contains no EGF-like domain. Instead, a trio of closely related domains (resembling a three-finger toxin fold) form a clamp-like structure around the bound EGF ligand. Although structurally unrelated to the receptor, Argos mimics EGFR by using a bipartite binding surface to entrap EGF. The individual Argos domains share unexpected structural similarities with the extracellular ligand-binding regions of transforming growth factor-beta family receptors. The three-domain clamp of Argos also resembles the urokinase-type plasminogen activator (uPA) receptor, which uses a similar mechanism to engulf the EGF-like module of uPA. Our results indicate that undiscovered mammalian counterparts of Argos may exist among other poorly characterized structural homologues. In addition, the structures presented here define requirements for the design of artificial EGF-sequestering proteins that would be valuable anti-cancer therapeutics.

Structural insights into the role and targeting of EGFRvIII.

The epidermal growth factor receptor (EGFR) is a well-known oncogenic driver in lung and other cancers. In glioblastoma multiforme (GBM), the EGFR deletion variant III (EGFRvIII) is frequently found alongside EGFR amplification. Agents targeting the EGFR axis have shown limited clinical benefits in GBM and the role of EGFRvIII in GBM is poorly understood. To shed light on the role of EGFRvIII and its potential as a therapeutic target, we determined X-ray crystal structures of a monomeric EGFRvIII extracellular region (ECR). The EGFRvIII ECR resembles the unliganded conformation of EGFR, including the orientation of the C-terminal region of domain II. Domain II is mostly disordered, but the ECR structure is compact. We selected a nanobody with preferential binding to EGFRvIII relative to EGFR and structurally defined an epitope on domain IV that is occluded in the unliganded intact EGFR. These findings suggest new avenues for EGFRvIII targeting in GBM.

Rational design of photosynthetic reaction center protein maquettes.

New technologies for efficient solar-to-fuel energy conversion will help facilitate a global shift from dependence on fossil fuels to renewable energy. Nature uses photosynthetic reaction centers to convert photon energy into a cascade of electron-transfer reactions that eventually produce chemical fuel. The design of new reaction centers de novo deepens our understanding of photosynthetic charge separation and may one day allow production of biofuels with higher thermodynamic efficiency than natural photosystems. Recently, we described the multi-step electron-transfer activity of a designed reaction center maquette protein (the RC maquette), which can assemble metal ions, tyrosine, a Zn tetrapyrrole, and heme into an electron-transport chain. Here, we detail our modular strategy for rational protein design and show that the intended RC maquette design agrees with crystal structures in various states of assembly. A flexible, dynamic apo-state collapses by design into a more ordered holo-state upon cofactor binding. Crystal structures illustrate the structural transitions upon binding of different cofactors. Spectroscopic assays demonstrate that the RC maquette binds various electron donors, pigments, and electron acceptors with high affinity. We close with a critique of the present RC maquette design and use electron-tunneling theory to envision a path toward a designed RC with a substantially higher thermodynamic efficiency than natural photosystems.

De novo protein design of photochemical reaction centers.

Natural photosynthetic protein complexes capture sunlight to power the energetic catalysis that supports life on Earth. Yet these natural protein structures carry an evolutionary legacy of complexity and fragility that encumbers protein reengineering efforts and obfuscates the underlying design rules for light-driven charge separation. De novo development of a simplified photosynthetic reaction center protein can clarify practical engineering principles needed to build new enzymes for efficient solar-to-fuel energy conversion. Here, we report the rational design, X-ray crystal structure, and electron transfer activity of a multi-cofactor protein that incorporates essential elements of photosynthetic reaction centers. This highly stable, modular artificial protein framework can be reconstituted in vitro with interchangeable redox centers for nanometer-scale photochemical charge separation. Transient absorption spectroscopy demonstrates Photosystem II-like tyrosine and metal cluster oxidation, and we measure charge separation lifetimes exceeding 100 ms, ideal for light-activated catalysis. This de novo-designed reaction center builds upon engineering guidelines established for charge separation in earlier synthetic photochemical triads and modified natural proteins, and it shows how synthetic biology may lead to a new generation of genetically encoded, light-powered catalysts for solar fuel production.

Biochemical and structural basis for differential inhibitor sensitivity of EGFR with distinct exon 19 mutations.

Tyrosine kinase inhibitors (TKIs) are used to treat non-small cell lung cancers (NSCLC) driven by epidermal growth factor receptor (EGFR) mutations in the tyrosine kinase domain (TKD). TKI responses vary across tumors driven by the heterogeneous group of exon 19 deletions and mutations, but the molecular basis for these differences is not understood. Using purified TKDs, we compared kinetic properties of several exon 19 variants. Although unaltered for the second generation TKI afatinib, sensitivity varied significantly for both the first and third generation TKIs erlotinib and osimertinib. The most sensitive variants showed reduced ATP-binding affinity, whereas those associated with primary resistance retained wild type ATP-binding characteristics (and low KM, ATP). Through crystallographic and hydrogen-deuterium exchange mass spectrometry (HDX-MS) studies, we identify possible origins for the altered ATP-binding affinity underlying TKI sensitivity and resistance, and propose a basis for classifying uncommon exon 19 variants that may have predictive clinical value.

ROR and RYK extracellular region structures suggest that receptor tyrosine kinases have distinct WNT-recognition modes.

WNTs play key roles in development and disease, signaling through Frizzled (FZD) seven-pass transmembrane receptors and numerous co-receptors including ROR and RYK family receptor tyrosine kinases (RTKs). We describe crystal structures and WNT-binding characteristics of extracellular regions from the Drosophila ROR and RYK orthologs Nrk (neurospecific receptor tyrosine kinase) and Derailed-2 (Drl-2), which bind WNTs though a FZD-related cysteine-rich domain (CRD) and WNT-inhibitory factor (WIF) domain respectively. Our crystal structures suggest that neither Nrk nor Drl-2 can accommodate the acyl chain typically attached to WNTs. The Nrk CRD contains a deeply buried bound fatty acid, unlikely to be exchangeable. The Drl-2 WIF domain lacks the lipid-binding site seen in WIF-1. We also find that recombinant DWnt-5 can bind Drosophila ROR and RYK orthologs despite lacking an acyl chain. Alongside analyses of WNT/receptor interaction sites, our structures provide further insight into how WNTs may recruit RTK co-receptors into signaling complexes.

N-terminal domains elicit formation of functional Pmel17 amyloid fibrils.

Pmel17 is a transmembrane protein that mediates the early steps in the formation of melanosomes, the subcellular organelles of melanocytes in which melanin pigments are synthesized and stored. In melanosome precursor organelles, proteolytic fragments of Pmel17 form insoluble, amyloid-like fibrils upon which melanins are deposited during melanosome maturation. The mechanism(s) by which Pmel17 becomes competent to form amyloid are not fully understood. To better understand how amyloid formation is regulated, we have defined the domains within Pmel17 that promote fibril formation in vitro. Using purified recombinant fragments of Pmel17, we show that two regions, an N-terminal domain of unknown structure and a downstream domain with homology to a polycystic kidney disease-1 repeat, efficiently form amyloid in vitro. Analyses of fibrils formed in melanocytes confirm that the polycystic kidney disease-1 domain forms at least part of the physiological amyloid core. Interestingly, this same domain is also required for the intracellular trafficking of Pmel17 to multivesicular compartments within which fibrils begin to form. Although a domain of imperfect repeats (RPT) is required for fibril formation in vivo and is a component of fibrils in melanosomes, RPT is not necessary for fibril formation in vitro and in isolation is unable to adopt an amyloid fold in a physiologically relevant time frame. These data define the structural core of Pmel17 amyloid, imply that the RPT domain plays a regulatory role in timing amyloid conversion, and suggest that fibril formation might be physically linked with multivesicular body sorting.

X-ray structure of a water-soluble analog of the membrane protein phospholamban: sequence determinants defining the topology of tetrameric and pentameric coiled coils.

Phospholamban (PLB) is a pentameric transmembrane protein that regulates the Ca(2+)-dependent ATPase SERCA2a in sarcoplasmic reticulum membranes. We previously described the computational design of a water-soluble variant of phospholamban, WSPLB, which reproduced many of the structural and functional properties of the native membrane-soluble protein. While the full-length WSPLB forms a pentamer in solution, a truncated variant forms very stable tetramers. To obtain insight into the tetramer-pentamer cytoplasmic switch, we solved the crystal structure of the truncated construct, WSPLB 21-52. This peptide has a heptad sequence repeat with Leu residues at a- and Ile at d-positions from residues 31-52. The crystal structure revealed that WSPLB 21-52 adopted an antiparallel tetrameric coiled coil. This topology contrasts with the parallel topology of an analogue of the coiled-coil of GCN4 with the same Leu(a) Ile(d) repeat. Analysis of these structures revealed how the nature of the partially exposed residues at e- and g-positions influence the topology formed by the bundle. We also constructed a model for the pentameric form of PLB using the coiled-coil parameters derived from a single monomer in the tetrameric structure. This model suggests that both buried and interfacial hydrogen bonds are important for stabilizing the parallel pentamer.

Analysis and design of turns in alpha-helical hairpins.

Although the analysis and design of turns that connect the strands in antiparallel beta-hairpins has reached an advanced state, much less is known concerning turns between antiparallel helices in helical hairpins. We have conducted an analysis of the structures and sequence preferences of two types of interhelical turns, each of which connects the two helices by a two-residue linker in an alphaL-beta conformation. Based on this analysis, it became apparent that the turn introduced into a designed four-helix bundle protein, DF1, did not occur within an optimal structural context. DF1 is a dimeric model for the diiron class of proteins. A longer loop with a beta-alphaR-beta conformation was inserted between two helices in the protein, and a sequence was chosen to stabilize its conformation. X-ray crystallography and NMR analysis of the protein showed the structure to be in excellent agreement with design.

X-ray structure of a maquette scaffold.

Maquettes are de novo designed mimicries of nature used to test the construction and engineering criteria of oxidoreductases. One type of scaffold used in maquette construction is a four-alpha-helical bundle. The sequence of the four-alpha-helix bundle maquettes follows a heptad repeat pattern typical of left-handed coiled-coils. Initial designs were molten globular due partly to the minimalist approach taken by the designers. Subsequent iterative redesign generated several structured scaffolds with similar heme binding properties. Variant [I(6)F(13)](2), a structured scaffold, was partially resolved with NMR spectroscopy and found to have a set of mobile inter-helical packing interfaces. Here, the X-ray structure of a similar peptide ([I(6)F(13)M(31)](2) i.e. ([CGGG EIWKL HEEFLKK FEELLKL HEERLKKM](2))(2) which we call L31M), has been solved using MAD phasing and refined to 2.8A resolution. The structure shows that the maquette scaffold is an anti-parallel four-helix bundle with "up-up-down-down" topology. No pre-formed heme-binding pocket exists in the protein scaffold. We report unexpected inter-helical crossing angles, residue positions and translations between the helices. The crossing angles between the parallel helices are -5 degrees rather than the expected +20 degrees for typical left-handed coiled-coils. Deviation of the scaffold from the design is likely due to the distribution and size of hydrophobic residues. The structure of L31M points out that four identical helices may interact differently in a bundle and heptad repeats with an alternating [HPPHHPP]/[HPPHHPH] (H: hydrophobic, P: polar) pattern are not a sufficient design criterion to generate left-hand coiled-coils.