Pentatricopeptide repeat (PPR) proteins represent a large family of sequence-specific RNA-binding proteins that are involved in multiple aspects of RNA metabolism. PPR proteins, which are found in exceptionally large numbers in the mitochondria and chloroplasts of terrestrial plants, recognize single-stranded RNA (ssRNA) in a modular fashion. The maize chloroplast protein PPR10 binds to two similar RNA sequences from the ATPI-ATPH and PSAJ-RPL33 intergenic regions, referred to as ATPH and PSAJ, respectively. By protecting the target RNA elements from 5' or 3' exonucleases, PPR10 defines the corresponding 5' and 3' messenger RNA termini. Despite rigorous functional characterizations, the structural basis of sequence-specific ssRNA recognition by PPR proteins remains to be elucidated. Here we report the crystal structures of PPR10 in RNA-free and RNA-bound states at resolutions of 2.85 and 2.45 Å, respectively. In the absence of RNA binding, the nineteen repeats of PPR10 are assembled into a right-handed superhelical spiral. PPR10 forms an antiparallel, intertwined homodimer and exhibits considerable conformational changes upon binding to its target ssRNA, an 18-nucleotide PSAJ element. Six nucleotides of PSAJ are specifically recognized by six corresponding PPR10 repeats following the predicted code. The molecular basis for the specific and modular recognition of RNA bases A, G and U is revealed. The structural elucidation of RNA recognition by PPR proteins provides an important framework for potential biotechnological applications of PPR proteins in RNA-related research areas.

The tumor suppressor Merlin/NF2 functions upstream of the core Hippo pathway kinases Lats1/2 and Mst1/2, as well as the nuclear E3 ubiquitin ligase CRL4DCAF1. Numerous mutations of Merlin have been identified in Neurofibromatosis type 2 and other cancer patients. Despite more than two decades of research, the upstream regulator of Merlin in the Hippo pathway remains unknown. Here we show by high-resolution crystal structures that the Lats1/2-binding site on the Merlin FERM domain is physically blocked by Merlin's auto-inhibitory tail. Angiomotin binding releases the auto-inhibition and promotes Merlin's binding to Lats1/2. Phosphorylation of Ser518 outside the Merlin's auto-inhibitory tail does not obviously alter Merlin's conformation, but instead prevents angiomotin from binding and thus inhibits Hippo pathway kinase activation. Cancer-causing mutations clustered in the angiomotin-binding domain impair angiomotin-mediated Merlin activation. Our findings reveal that angiomotin and Merlin respectively interface cortical actin filaments and core kinases in Hippo signaling, and allow construction of a complete Hippo signaling pathway.

No abstract

Clustering and asymmetric distribution of receptors, ion channels and associated protein complexes are essential to the polarity of neurons and epithelial cells. Such asymmetric targeting of large molecular assemblies is thought to be governed in part by modular scaffold proteins. L27 domain, initially identified in the C. elegans Lin-2 and Lin-7 proteins, is a protein interaction module that exists in a large family of scaffold proteins 1 . Formation of the trimeric Lin-2-Lin-7-Lin-10 complex requires L27 domains 2,3 and has a central role in targeting receptor tyrosine kinase Let-23 signalsome to the basolateral surface of vulval precursor cells 4,5 . Mutations in Lin-2, Lin-7 or Lin-10 lead to a vulvaless phenotype, presumably by mistargeting of 5). The mammalian orthologs of Lin-2, Lin-7 and Lin-10 are known as mammalian Lin-2 (mLin-2)/CASK, mLin-7/Velis/Mals and mLin-10/X11α/Mint1, respectively 2,6-9 . The mLin-2-mLin-7-mLin-10 tripartite complex has also been observed in the brain 2,6 , and this evolutionarily conserved protein complex has been implicated in the targeting of NMDA receptors and β-catenin assemblies to membrane subdomains in epithelia and neurons 10,11 .A bioinformatics survey reveals that L27 domain-containing proteins are invariably scaffold proteins containing multiple proteinprotein interaction domains without intrinsic enzyme activities 12 . For example, SAP97 and mLin-2/CASK, which are members of a subset of membrane-associated guanylate kinases (MAGUKs), contain one and two L27 domains, respectively, in addition to their PDZ, SH3 and guanylate kinase-like domains. The N-terminal L27 domain of SAP97 can form specific heteromeric complexes with the N-terminal L27 domain of mLin-2, Dlg2 and Dlg3, respectively 3,13,14 . The C-terminal L27 domain of mLin-2 specifically binds to the L27 domain of 15). The discovery of L27 domains as specific protein-protein interaction modules capable of forming heteromeric complexes suggests that L27 domains can integrate multiple scaffold proteins into supramolecular assemblies 1,3,13,15,16 . However, the molecular basis of the L27 domain-mediated protein assembly formation remains unclear. The structures of isolated L27 domains and their cognate heteromeric complexes are not known.Here we show that the L27 domain of SAP97 and the N-terminal L27 domain of mLin-2 form a heterotetrameric complex. The structure of the SAP97-mLin-2 L27 complex was solved by NMR spectroscopy. The structure of the complex, together with data derived from various biochemical studies, establishes the roles of specific residues in the formation of the L27 heterotetrameric complex. The structure of the SAP97-mLin-2 L27 complex further suggests a mechanistic model for polymerization of L27 domain scaffold proteins. RESULTS L27 domains form specific heterotetrameric complexesWe first characterized the structural properties of the isolated L27 domains of SAP97 (referred to as L27S below), mLin-7, the N-terminal L27 domain (referred to as L27N below), and the C-terminal L27 domain (ref...