Biochemistry and Cell Biology Structure and Function of the Fgd Family of Divergent FYVE Domain Proteins Journal: Biochemistry and Cell Biology Manuscript ID bcb-2018-0185.R1 Manuscript Type: Mini Review Date Submitted by the 03-Aug-2018 Author: Complete List of Authors: Eitzen, Gary; University of Alberta Faculty of Medicine and Dentistry Smithers, Cameron C.; University of Alberta, Biochemistry Murray, Allan; University of Alberta Faculty of Medicine and Dentistry Overduin, Michael; University of Alberta Faculty of Medicine and Dentistry Draft Fgd, Pleckstrin Homology domain, FYVE domain, Dbl Homology Domain, Keyword: Rho GEF Is the invited manuscript for consideration in a Special CSMB Special Issue Issue? : https://mc06.manuscriptcentral.com/bcb-pubs Page 1 of 37 Biochemistry and Cell Biology Title: Structure and Function of the Fgd Family of Divergent FYVE Domain Proteins Authors: Gary Eitzen1, Cameron C. Smithers2, Allan G Murray3 and Michael Overduin2* Draft 1Department of Cell Biology, 2Department of Biochemistry, 3Department of Medicine, University of Alberta, Edmonton, Alberta, Canada *Corresponding author. Michael Overduin Telephone: +1 780 492 3518 Fax: +1 780 492-0886 E-mail: [email protected] https://mc06.manuscriptcentral.com/bcb-pubs Biochemistry and Cell Biology Page 2 of 37 Abstract FYVE domains are highly conserved protein modules that typically bind phosphatidylinositol 3-phosphate (PI3P) on the surface of early endosomes. Along with pleckstrin homology (PH) and phox homology (PX) domains, FYVE domains are the principal readers of the phosphoinositide (PI) code that mediate specific recognition of eukaryotic organelles. Of all the human FYVE domain-containing proteins, those within the Faciogenital dysplasia (Fgd) subfamily are particularly divergent, and couple with GTPases to exert unique cellular functions. The subcellular distributions and functions of these evolutionarily conserved signal transducers, which also include Dbl homology (DH) and two PH domains, are discussed here in order to better understand the biological range of processes that such multidomain proteins engage in. Determinants of their various functions include specific multidomain architectures, post-translational modifications including PIP stops that have been discovered in sorting nexins, PI recognition motifs and phospholipid binding surfaces as defined by the Membrane Optimal Docking Area (MODA) program. How these orchestrate Fgd function remains unclearDraft but has implications for developmental diseases including Aarskog-Scott syndrome, which is also known as faciogenital dysplasia, and forms of cancer that are associated with mutations and amplifications of Fgd genes. Keywords: Cdc42, Dbl, FYVE, Fgd, GEF, GTPase, PH, phosphoinositide, pleckstrin, cancer, faciogenital dysplasia, Aarskog-Scott syndrome , lipid signaling, membrane trafficking, MODA, PIP, Rho. Text: Phosphoinositide Code Stimulation and differentiation of cells involve dramatic alterations of phosphorylation patterns in proteins, lipids and nucleotides. Defining how these molecular events converge to change cellular behaviour remains a central challenge that, if achieved, promises to unlock novel targets for drug discovery. The molecules that transduce biological signals undergo reversible addition of phosphates, thus mediating the flow of cellular information. In the case of protein modifications, these signals are binary. https://mc06.manuscriptcentral.com/bcb-pubs Page 3 of 37 Biochemistry and Cell Biology That is, amino acid residues are either phosphorylated or unphosphorylated. Mononucleotides are represented by multiple forms such as the cGMP, GMP, GDP and GTP derivatives of guanosine present in eukaryotes, and regulate the activities of proteins such as GTPases. The phosphatidylinositol headgroup is differentially phosphorylated at the 3-, 4- and/or 5- positions to yield 7 distinct PI lipids, which are distributed to various organelles where they are recognized by FYVE, PH, PX and other domains. These modules serve as principal readers of the PI code and direct membrane-dependent signaling and trafficking in eukaryotes (Overduin, Cheever, and Kutateladze 2001). Together, these protein, nucleotide and lipid modifications offer a rich tapestry of signaling states that encodes biological information to determine cellular fate. Our understanding of how this actually works is in its infancy, as illustrated here by the complexity of Fgd proteins. This family is unique in terms of its architecture, integrating strands of information through multiple domains that recognize specific phospholipids and proteins including GTPases to drive cellular development. Here we review what is known about the family and explore the roles of the component domains and family members. Architecture of the Fgd Family Draft The human genome encodes six Fgd genes (Fig 1). Homologues are present in most metazoans, and diverged early in chordate evolution while maintaining the same modular architecture. They act as Rho guanine-nucleotide exchange factors (Rho GEFs) that are thought to be guided to membranes by PI- binding FYVE and PH domains. They contain the typical Rho GEF subdomain structure of a DH domain adjacent to the first PH domain (PH1). The DH and PH1 domains function together to mediate the regulated activation of Rho proteins by binding to and catalyzing the exchange of GDP for GTP nucleotides. All Fgd proteins contain a divergent FYVE domain followed by a second PH domain (PH2) at their C-termini. The long N-termini of Fgd proteins are poorly conserved and apparently disordered. Fgd1 contains an N- terminal proline-rich element which is phosphorylated (Hornbeck et al. 2015), influences peripheral membrane localization dynamics (T Oshima et al. 2010), and harbours a critical point mutation linked to Aarskog-Scott syndrome (Orrico et al. 2004). Fgd4 interacts with F-actin via its N-terminal 150 residues (Obaishi et al. 1998). The structural domains of Fgd proteins are discussed individually below, including descriptions of structural and regulatory features. FYVE Domains https://mc06.manuscriptcentral.com/bcb-pubs Biochemistry and Cell Biology Page 4 of 37 FYVE domains are found in 29 human proteins, which exhibit a wide range of architectures, with that of the Fgd subfamily (Fig 1) being most recurrent. This short module is named after four proteins in which its 70 residue sequence was first identified: Fab1p, YOTB, Vac1p and EEA1. The fold of the FYVE domain consists of two double-stranded antiparallel β sheets and a C-terminal α-helix (T. G. Kutateladze and Overduin 2004). Two zinc ions are gripped tightly by four cysteines in a tetrahedral arrangement to provide essential structural stabilization (T. G. Kutateladze et al. 1999). FYVE domains represent the most tightly conserved lipid recognition module, and are thought to almost always recognize PI3P as their cognate ligand (T. Kutateladze and Overduin 2001). Organelle recognition involves electrostatic approach and dipping of a membrane insertion loop (MIL) into a bed of PI3P, phosphatidylserine (PS) and phosphatidycholine (PC) molecules for firm anchoring and stereospecific lipid headgroup recognition (T. G. Kutateladze et al. 2004). This bilayer binding is strengthened by the acidic microenvironments found along the endocytic route, thus keeping proteins more stably tethered to low pH endosomal membranes (Lee et al. 2005) rather than the other PI3P pools found in the cell . Almost all FYVE domains contain three functionallyDraft critical signatures, all of which diverge in the Fgd family (Fig. 2). The aspartic acid in the N-terminal motif WxxD (where x is any residue) excludes binding of PIs that bear phosphates at the 4- or 5-positions, as seen in EEA1, while the tryptophan underpins the binding site (Dumas et al. 2001; T. Kutateladze and Overduin 2001). This highly conserved motif is replaced with PIRE, TEED and LVPV sequences in Fgd1, Fgd3 and Fgd5, respectively, suggesting different ligand specificity profiles and altered accommodation of the inositol ring. The central R+HHCR motif (where + is Arg or Lys) is the core PI3P code reader, and provides direct recognition of the 3-phosphate as well as ensuring pH-dependent ligand recognition. The last arginine of the reader motif mediates a pair of hydrogen bonds to coordinate this phosphate, but is replaced by a lysine in Fgd1 and Fgd3 and a histidine in Fgd5, reinforcing the notion that these Fgd proteins possess altered PI selectivity. The arginine of the C-terminal RVC motif provides electrostatic interactions with the bilayer, but is replaced with a lysine in the case of Fgd4 and Fgd5. The substitution for arginine for lysine is relatively conservative as both have bulky and cationic side chains. Based on analysis of experimentally mapped peripheral membrane protein interfaces (Kufareva et al. 2014), arginines are more common in membrane binding surfaces than lysines, reflecting their more stable and specific bidentate hydrogen bonding of phospholipid ligands. Anomalies in PI binding motifs could be expected to influence the ligand specificities of the Fgd family of FYVE domains. The FYVE domain of Fgd1 binds not only to PI3P but also to PI5P, albeit more weakly, in a https://mc06.manuscriptcentral.com/bcb-pubs Page 5 of 37 Biochemistry and Cell Biology dot-blot lipid interaction assay, as performed at pH 8.0 (Sankaran et al. 2001). Both Fgd3 and Fgd4, as expressed in colon cancer cells, are pulled down by PI3P bait molecules. Moreover, Fgd4 binds preferentially