Structural basis of ?-secretase inhibition and modulation by small molecule drugs

•Atomic structures of human ?-secretase bound to inhibitors and the modulator E2012•Differential recognition clinical candidates Semagacestat Avagacestat•Differential transition state analog (TSA) L685,458 non-TSA•Identification allosteric binding site E2012 Development (GSIs) modulators (GSMs) represents an attractive therapeutic opportunity for Alzheimer’s disease (AD) cancers. However, how these GSIs GSMs target has remained largely unknown. Here, we report cryoelectron microscopy (cryo-EM) individually two GSI candidates, Avagacestat, a L685,458, classic GSM E2012, at overall resolutions 2.6–3.1 Å. Remarkably, each occupies same general location on presenilin 1 (PS1) that accommodates ? strand from amyloid precursor protein or Notch, interfering with substrate recruitment. directly coordinates catalytic aspartate residues PS1. binds extracellular side, potentially explaining its modulating activity. Structural analysis reveals set shared themes variations inhibitor will guide development next-generation substrate-selective inhibitors. The membrane-embedded cleaves transmembrane domain (TM) multiple signaling proteins exemplified by (APP) Notch (Beel Sanders, 2008Beel A.J. Sanders C.R. Substrate specificity gamma-secretase other intramembrane proteases.Cell. Mol. Life Sci. 2008; 65: 1311-1334Crossref PubMed Scopus (219) Google Scholar; De Strooper et al., 1998De B. Saftig P. Craessaerts K. Vanderstichele H. Guhde G. Annaert W. Von Figura Van Leuven F. Deficiency presenilin-1 inhibits normal cleavage protein.Nature. 1998; 391: 387-390Crossref (1513) Scholar, 1999De Cupers Mumm J.S. Schroeter E.H. Schrijvers V. Wolfe M.S. Ray W.J. al.A presenilin-1-dependent gamma-secretase-like protease mediates release intracellular domain.Nature. 1999; 398: 518-522Crossref (1758) Scholar). APP is first cleaved ?-secretase in space. resulting C-terminal 99-residue fragment (APP-C99) further proteolyzed generate (AICD) 48- 49-residue peptide (A?48 A?49) (Takami 2009Takami M. Nagashima Y. Sano Ishihara S. Morishima-Kawashima Funamoto Ihara gamma-Secretase: successive tripeptide tetrapeptide beta-carboxyl terminal fragment.J. Neurosci. 2009; 29: 13042-13052Crossref (334) A?48 A?49 are then trimmed every three four ?-secretase, generating A? peptides varying lengths byproducts diverse tri- tetra-peptides Accumulation leads formation ?-amyloid plaques, hallmark (Kidd, 1964Kidd Disease–an Electron Microscopical Study.Brain. 1964; 87: 307-320Crossref (316) ?-Secretase contains subunits: subunit PS1 PS2, Pen-2, APH-1, nicastrin (NCT). vast majority familial AD-derived (FAD) mutations lesser extent PS2 (Citron 1992Citron Oltersdorf T. Haass C. McConlogue L. Hung A.Y. Seubert Vigo-Pelfrey Lieberburg I. Selkoe D.J. Mutation beta-amyloid increases beta-protein production.Nature. 1992; 360: 672-674Crossref (1490) Rogaev 1995Rogaev E.I. Sherrington R. Rogaeva E.A. Levesque Ikeda Liang Chi Lin Holman Tsuda al.Familial kindreds missense gene chromosome related type 3 gene.Nature. 1995; 376: 775-778Crossref (1736) 1995Sherrington Li al.Cloning bearing early-onset disease.Nature. 375: 754-760Crossref (3468) Suzuki 1994Suzuki N. Cheung T.T. Cai X.D. Odaka A. Otvos Jr., Eckman Golde T.E. Younkin S.G. An increased percentage long beta secreted (beta APP717) mutants.Science. 1994; 264: 1336-1340Crossref (1324) Nearly all result aberrant proteolysis (Borchelt 1996Borchelt D.R. Thinakaran C.B. Lee M.K. Davenport Ratovitsky Prada C.M. Kim Seekins Yager D. disease-linked variants elevate Abeta1-42/1-40 ratio vitro vivo.Neuron. 1996; 17: 1005-1013Abstract Full Text PDF (1308) These observations engendered hypothesis (Hardy Higgins, 1992Hardy J.A. Higgins G.A. disease: cascade hypothesis.Science. 256: 184-185Crossref (4535) Scholar), which suggests oligomers plaques be causative factor AD (Cleary 2005Cleary J.P. Walsh D.M. Hofmeister J.J. Shankar G.M. Kuskowski M.A. Ashe K.H. Natural amyloid-beta specifically disrupt cognitive function.Nat. 2005; 8: 79-84Crossref (1435) This predicts inhibition presumably reduces production decreases accumulation may constitute effective treatment (Golde 2010Golde Petrucelli Lewis J. Targeting Abeta tau disease, early interim report.Exp. Neurol. 2010; 223: 252-266Crossref (71) Wolfe, 2008Wolfe Gamma-secretase modulation disease.Curr. Alzheimer Res. 5: 158-164Crossref (58) Semagacestat, developed code-named LY-450139 Eli Lily (Henley 2009Henley D.B. May P.C. Dean R.A. Siemers E.R. semagacestat (LY450139), functional inhibitor, disease.Expert Opin. Pharmacother. 10: 1657-1664Crossref (134) was (GSI) entered phase III evaluation. In contrast encouraging earlier results (Siemers 2007Siemers Friedrich Ferguson-Sells Gonzales Farlow M.R. Safety, tolerability, effects plasma cerebrospinal fluid after gamma-secretase.Clin. Neuropharmacol. 2007; 30: 317-325Crossref (155) failed show improvement but produced severe side (Doody 2013Doody R.S. Raman Iwatsubo Vellas Joffe Kieburtz He Sun X. Thomas R.G. al.Alzheimer’s Disease Cooperative Study Steering CommitteeSemagacestat GroupA trial disease.N. Engl. Med. 2013; 369: 341-350Crossref (732) attributed substrates such as (Aster 2017Aster J.C. Pear W.S. Blacklow S.C. Varied Roles Cancer.Annu. Rev. Pathol. 2017; 12: 245-275Crossref (276) Doody 2015Doody Sperling Seimers E. Sethuraman Mohs StudyPeripheral central disease.Alzheimers Ther. 2015; 7: 36Crossref (22) To address this problem, substrate-specific have been Notch-sparing (Gillman 2010Gillman K.W. Starrett J.E. Parker M.F. Xie Bronson Marcin L.R. McElhone K.E. Bergstrom C.P. Mate Williams al.Discovery Evaluation BMS-708163, Potent, Selective Orally Bioavailable Inhibitor.ACS Chem. Lett. 1: 120-124Crossref (141) Mayer 2008Mayer Kreft A.F. Harrison Abou-Gharbia Antane Aschmies Atchison Chlenov Cole D.C. Comery begacestat, Notch-1-sparing disease.J. 51: 7348-7351Crossref (90) A representative candidate Avagacestat (known BMS-708163) preferentially inhibited over high doses still inhibit exhibit toxicity (Coric 2012Coric van Dyck C.H. Salloway Andreasen Brody Richter R.W. Soininen Thein Shiovitz Pilcher al.Safety tolerability avagacestat 2 study mild moderate disease.Arch. 2012; 69: 1430-1440Crossref (246) Coric 2015Coric Dubois Curtis al.Targeting Prodromal With Avagacestat: Randomized Clinical Trial.JAMA 72: 1324-1333Crossref (113) classified into non-TSA based mechanisms (Esler 2000Esler W.P. Kimberly W.T. Ostaszewski B.L. Diehl T.S. Moore C.L. Tsai J.Y. Rahmati Xia Transition-state analogue bind presenilin-1.Nat. Cell Biol. 2000; 2: 428-434Crossref (497) Tagami 2017Tagami Yanagida Kodama Takami Mizuta Oyama Nishitomi Chiu Y.W. Okamoto Ikeuchi al.Semagacestat Is Pseudo-Inhibitor ?-Secretase.Cell Rep. 21: 259-273Abstract (37) 2012Wolfe Neurochem. 120: 89-98Crossref (119) both non-competitive (Karran Hardy, 2014Karran Hardy critique drug discovery programs targeting disease.Ann. 2014; 76: 185-205Crossref (190) contrast, small molecule widely studied competitive TSA Henley Shearman 2000Shearman Beher Clarke E.E. H.D. Hunt Nadin Smith A.L. Stevenson Castro J.L. L-685,458, aspartyl mimic, potent activity.Biochemistry. 39: 8698-8704Crossref (362) active Tian 2010Tian Bassit Chau Y.M. inhibitory containing Flemish mutation residue modulates activity production.Nat. Struct. 151-158Crossref (64) addition GSIs, also recognized class molecules alleviating symptoms (Crump 2013Crump C.J. Johnson D.S. mechanism disease.Biochemistry. 52: 3197-3216Crossref (127) Kounnas 2010Kounnas M.Z. Danks A.M. Cheng Tyree Ackerman Zhang Ahn Nguyen Comer Mao al.Modulation deposition transgenic mouse model disease.Neuron. 67: 769-780Abstract (203) Liu 2014Liu Q. Waltz Woodruff Ouyang Israel Herrera Sarsoza Tanzi R.E. Koo Ringman J.M. al.Effect neurons derived 1-induced pluripotent stem cell mutant carriers.JAMA 71: 1481-1489Crossref (65) Wagner 2014Wagner S.L. Rynearson K.D. Wang Sisodia S.S. Mobley W.C. Soluble selectively 42-amino acid variant augment carboxy-truncated species.Biochemistry. 53: 702-713Crossref promote A?42, thus reducing amount aggregation-prone (Pozdnyakov 2013Pozdnyakov Murrey H.E. Crump Pettersson Ballard Am Ende C.W. Bales K.R. (GSM) photoaffinity probes reveal distinct sites presenilin.J. 288: 9710-9720Abstract (82) Unlike away Therefore, specific work achieve inhibition. used member heterocyclic family (Kimura 2005Kimura Kawano Doi Kitazawa Shin Miyagawa Kaneko Ito Takaishi Sasaki Hagiwara Preparation cinnamide, 3-benzylidenepiperidin-2-one, phenylpropynamide compounds PCT Int. Appl. 2005, WO 2005115990 A1 20051208.2005Google more necessitates mechanistic understanding well Improvement requires structural information their ?-secretase. Unfortunately, date, there no atomic any (Bai 2015aBai X.C. Rajendra Yang Shi Scheres S.H. Sampling conformational space ?-secretase.eLife. 4: e11182Crossref (345) 2014Li Lu Bohm Qamar Dodd R.B. Meadows Jeon McLeod Chen al.Structural interactions between docking give insight complexes.Structure. 22: 125-135Abstract (53) study, GSIs: L685,458. We structure Recombinant prepared described (Lu 2014Lu Bai Ma Yan Zhao Zhou S.H.W. Three-dimensional ?-secretase.Nature. 512: 166-170Crossref (233) examined ?-secretase-mediated APP-C99 (Figure 1). assay, exhibited similar potency 1A). Using AlphaLisa found display constant (IC50) 2.7 ± 0.9 ?M A?40 1B). shows clear preference 1C), IC50 16.7 2.2 nM 1D). (hereafter referred L458) behaves similarly 1E) 1F). excess incubated purified prior sample preparation. 1,978 micrographs were recorded using K2 Summit detector mounted Titan Krios microscope, yielding 1,044,389 particles S1A). Following two-dimensional (2D) three-dimensional (3D) classifications, 323,861 yielded final reconstruction Semagacestat-bound average resolution 3.0 Å (Figures 2A S1B; Table S1). isolated density lobe, located next ?2 PS1, assigned 2B S1C).Figure 2Structural basis ?-secretaseShow full caption(A) Overall cryo-EM map close-up view Semagacestat-binding shown inset. colored cyan magenta, respectively.(B) Chemical (upper panel) EM (lower panel, contoured 5.5?) Semagacestat.(C) deep cavity opens side. positions red (left panel). Binding induces strands (right electrostatic potential transparent surface.(D) Specific interacts Asp385 through mixture hydrogen bonds (H-bonds, left der Waals contacts H-bonds highlighted dashed lines, PAL motif orange.See Figures S1 S7 S1.View Large Image Figure ViewerDownload Hi-res image Download (PPT) (A) respectively. (B) Semagacestat. (C) surface. (D) orange. See S1. exactly Notch-100 (Yang 2019Yang Guo Ke Lei 2019; 565: 192-197Crossref (99) 2019Zhou Recognition ?-secretase.Science. 363: eaaw0930Crossref (129) Scholar) 2C). finding provides immediate satisfactory explanation Because hybrid sheet essential occupation hinder thereby cleavage. surrounded ?2, TM6a, portion TM8, ensuing (Pro433-Ala434-Leu435) important (Sato 2008Sato Takagi Tomita 9 involved pore gamma-secretase.J. 28: 6264-6271Crossref (118) makes (H-bonds) main chain groups Lys380 Leu432 just preceding 2D, regard, acts like strand. uses hydroxyl group donate H-bond carboxylate oxygen Intriguingly, chains Asp257 separated distance only 4.8 close activated protease, suggesting preservation conformation inhibitor-bound state. number surrounding hydrophobic right coordination leaves several crevices accommodate additional chemical S1D). For example, might fill around Val261 Val272 S1E). Such modification likely improved potency. prefers reported ranging 3-fold 193-fold (Chávez-Gutiérrez 2012Chávez-Gutiérrez Bammens Benilova Vandersteen Benurwar Borgers Lismont Cleynenbreugel Esselmann al.The dysfunction disease.EMBO 31: 2261-2274Crossref (358) 2012Crump S.V. Pozdnyakov BMS-708,163 targets lacks notch-sparing 7209-7211Crossref (75) Gillman How selectivity conferred remains Notably, identity quite different 3A, panel), raising possibility Avagacestat-bound collected 6,266 micrographs. initial dataset 835,806 subject 2D 3D 3.1 S2; S1).Figure S2Cryo-EM 3Show flowchart data processing. Please refer STAR methods details. estimated FSC 0.143 value. curves refined versus it against (black); independent maps calculation (red); second (green). difference green indicates refinement did not suffer overfitting. Local distribution RELION-2.0 (Kimanius 2016Kimanius Forsberg B.O. Lindahl Accelerated determination parallelisation GPUs RELION-2.eLife. 2016; e18722Crossref (204) B-factor sharpening ?108. (E) Representative local densities 4.5?. yellow.View