ArticlePDF AvailableMutations in Disordered Regions Can Cause Disease by Creating Dileucine MotifsSeptember 2018Cell 175(1)DOI:10.1016/j.cell.2018.08.019Project: Identification of disease mechanisms related to intrinsically disordered proteinsAuthors: Katrina MeyerUniversity of Zurich Marieluise KirchnerBerlin Institute of Health (BIH) at Charité Bora UyarMax-Delbrück-Centrum für Molekulare Medizin Jing-Yuan ChengJing-Yuan ChengThis person is not on ResearchGate, or hasn t claimed this research yet.Show all 19 authorsHide Download full-text PDFRead full-textDownload full-text PDFRead full-textDownload citation Copy link Link copied Read full-text Download citation Copy link Link copiedCitations (56)References (48)Figures (3)Abstract and FiguresMany disease-causing missense mutations affect intrinsically disordered regions (IDRs) of proteins, but the molecular mechanism of their pathogenicity is enigmatic. Here, we employ a peptide-based proteomic screen to investigate the impact of mutations in IDRs on protein-protein interactions. We find that mutations in disordered cytosolic regions of three transmembrane proteins (GLUT1, ITPR1, and CACNA1H) lead to an increased clathrin binding. All three mutations create dileucine motifs known to mediate clathrin-dependent trafficking. Follow-up experiments on GLUT1 (SLC2A1), the glucose transporter causative of GLUT1 deficiency syndrome, revealed that the mutated protein mislocalizes to intracellular compartments. Mutant GLUT1 interacts with adaptor proteins (APs) in vitro, and knocking down AP-2 reverts the cellular mislocalization and restores glucose transport. A systematic analysis of other known disease-causing variants revealed a significant and specific overrepresentation of gained dileucine motifs in structurally disordered cytosolic domains of transmembrane proteins. Thus, several mutations in disordered regions appear to cause \"dileucineopathies.” Differential Interactors of Wild-Type and Mutant IDRs…  Adaptor Proteins Bind to Mutated GLUT1 and Cause Cellular Mistrafficking (A) Adaptor protein complexes AP-1, AP-2, and AP-3 show increased colocalization to GLUT1 due to P485L mutation in replicates of BioID experiment from Figure 4. Identified subunits of APs are shown in red. See also Figure S4. (B) P485L mutant but not wild-type Glut1 C-terminal tail interacts with AP-1 and AP-2. Tails were tagged with GST to pull-down interaction partners from mouse brain lysate. Talin is shown as a negative control and is not pulled down from either of the two variants. (C and D) Mutant but not wild-type GLUT1 extensively colocalizes with endocytosed transferrin. HEK cells stably expressing FLAG-GLUT1 are incubated with fluorescently labeled transferrin for 10 min before fixation. Scale bar, 10 mm (C). Data are represented as mean ± SD (D). (E and F) Western blot against AP-2 a and m subunits shows downregulation after two rounds of siRNA transfection against AP-2 m. AP-2 knockdown leads to relocalization of GLUT1_P485L to the plasma membrane and hence rescue of the mutation phenotype. Scale bar, 10 mm.…  GLUT1 P485L Mislocalizes in Patient-Derived iPSCs and Endothelial Cells of the Blood-Brain Barrier in Mice (A) A skin sample was taken from a GLUT1 deficiency patient with a heterozygous GLUT1_P485L mutation. Fibroblasts were grown and reprogrammed to iPSCs. See also Figure S5. (B) Heterozygous GLUT1_P485L mutation leads to partial mislocalization of GLUT1 in patient-derived iPSCs. Scale bars, 10 mm. (C and D) GLUT1_P485L colocalizes with the post-Golgi SNARE VTI1A. Scale bar, 10 mm (C). Data are represented as mean ± SD (D). (legend continued on next page)… Figures - uploaded by Katrina MeyerAuthor contentAll figure content in this area was uploaded by Katrina MeyerContent may be subject to copyright. Discover the world s research20+ million members135+ million publications700k+ research projectsJoin for freePublic Full-text 1Content uploaded by Katrina MeyerAuthor contentAll content in this area was uploaded by Katrina Meyer on Apr 29, 2020 Content may be subject to copyright. ArticleMutations in Disordered Regions Can Cause Diseaseby Creating Dileucine MotifsGraphical AbstractHighlightsdA peptide-based screen detects how mutations affectprotein-protein interactionsdSeveral pathogenic mutations create dileucine motifs andrecruit clathrindA dileucine motif gain in GLUT1 causes mistrafficking inGLUT1 deficiency syndromedProtein mistrafficking via dileucine motif gains is a recurrentcause of diseaseAuthorsKatrina Meyer, Marieluise Kirchner,Bora Uyar, ..., Sebastian Diecke,Juan M. Pascual, Matthias SelbachCorrespondencematthias.selbach@mdc-berlin.deIn BriefIntrinsically disordered regions (IDRs)may serve as functional hubs to regulateprotein functions. In this issue of Cell,Meyer et al. showed that disease-causingmissense mutations in IDRs createdileucine motifs, which mediate clathrin-dependent trafficking that underliesdisease etiology.Meyer et al., 2018, Cell 175, 1–15September 20, 2018 ª2018 Elsevier Inc.https://doi.org/10.1016/j.cell.2018.08.019 ArticleMutations in Disordered RegionsCan Cause Disease by Creating Dileucine MotifsKatrina Meyer,1Marieluise Kirchner,1Bora Uyar,2Jing-Yuan Cheng,1Giulia Russo,3Luis R. Hernandez-Miranda,4Anna Szymborska,5,7Henrik Zauber,1Ina-Maria Rudolph,6Thomas E. Willnow,6Altuna Akalin,2Volker Haucke,3Holger Gerhardt,5,7,8Carmen Birchmeier,4Ralf Ku¨ hn,8,9Michael Krauss,3Sebastian Diecke,7,8,10Juan M. Pascual,11and Matthias Selbach1,12,13,*1Proteome Dynamics, Max Delbru¨ck Center for Molecular Medicine in the Helmholtz Association (MDC), Robert-Ro¨ssle-Str. 10, 13125 Berlin,Germany2Bioinformatics Platform, Max Delbru¨ck Center for Molecular Medicine in the Helmholtz Association (MDC), Robert-Ro¨ssle-Str. 10,13125 Berlin, Germany3Molecular Pharmacology and Cell Biology, Leibniz-Forschungsinstitut fu¨r Molekulare Pharmakologie, Robert-Ro¨ssle-Str. 10, 13125 Berlin,Germany4Developmental Biology/Signal Transduction, Max Delbru¨ck Center for Molecular Medicine in the Helmholtz Association (MDC),Robert-Ro¨ssle-Str. 10, 13125 Berlin, Germany5Integrative Vascular Biology Laboratory, Max Delbru¨ck Center for Molecular Medicine in the Helmholtz Association (MDC),Robert-Ro¨ssle-Str. 10, 13125 Berlin, Germany6Molecular Cardiovascular Research, Max Delbru¨ck Center for Molecular Medicine in the Helmholtz Association (MDC), Robert-Ro¨ssle-Str.10, 13125 Berlin, Germany7DZHK (German Centre for Cardiovascular Research) partner site, 13347 Berlin, Germany8Berlin Institute of Health (BIH), 10178 Berlin, Germany9Core Facility Transgenics, Max Delbru¨ck Center for Molecular Medicine in the Helmholtz Association (MDC), Robert-Ro¨ssle-Str. 10,13125 Berlin, Germany10Core Facility Pluripotent Stem Cells, Max Delbru¨ck Center for Molecular Medicine in the Helmholtz Association (MDC), Robert-Ro¨ssle-Str.10, 13125 Berlin, Germany11Department of Neurology and Neurotherapeutics, UT Southwestern Medical Center, 5323 Harry Hines Blvd. Dallas, TX 75390, USA12Charite´-Universita¨tsmedizin Berlin, 10117 Berlin, Germany13Lead Contact*Correspondence: matthias.selbach@mdc-berlin.dehttps://doi.org/10.1016/j.cell.2018.08.019SUMMARYMany disease-causing missense mutations affectintrinsically disordered regions (IDRs) of proteins,but the molecular mechanism of their pathogenicityis enigmatic. Here, we employ a peptide-based pro-teomic screen to investigate the impact of mutationsin IDRs on protein-protein interactions. We findthat mutations in disordered cytosolic regions ofthree transmembrane proteins (GLUT1, ITPR1, andCACNA1H) lead to an increased clathrin binding. Allthree mutations create dileucine motifs known tomediate clathrin-dependent trafficking. Follow-upexperiments on GLUT1 (SLC2A1), the glucose trans-porter causative of GLUT1 deficiency syndrome, re-vealed that the mutated protein mislocalizes to intra-cellular compartments. Mutant GLUT1 interacts withadaptor proteins (APs) in vitro, and knocking downAP-2 reverts the cellular mislocalization and restoresglucose transport. A systematic analysis of otherknown disease-causing variants revealed a signifi-cant and specific overrepresentation of gaineddileucine motifs in structurally disordered cytosolicdomains of transmembrane proteins. Thus, severalmutations in disordered regions appear to cause‘‘dileucineopathies.’’INTRODUCTIONGenome sequencing technologies have greatly facilitated thediscovery of human protein variants. In many cases, it is notknown whether such variants cause disease, and even when as-sociations have been established, determining the molecularmechanisms remains a major challenge (Cooper and Shendure,2011). Most disease-causing missense mutations affect evolu-tionarily conserved amino acids within structured regions of pro-teins and destabilize their structure (Subramanian and Kumar,2006; Yue et al., 2005). However, over 20% of human diseasemutations occur in so-called intrinsically disordered regions(IDRs) (Vacic et al., 2012). Contrary to the traditional understand-ing of protein structure and function, it is now clear that IDRsrepresent a functionally important and abundant part of eukary-otic proteomes (Uversky et al., 2008; Wright and Dyson, 2015).Yet, because IDRs lack a defined tertiary structure and are typi-cally poorly conserved, the classical structure-function para-digm cannot explain how mutations in IDRs cause disease.We set out to investigate the mechanism of these mutations byanalyzing protein-protein interactions (PPIs), which can help tounderstand how mutations cause disease (Ryan et al., 2013;Cell 175, 1–15, September 20, 2018 ª2018 Elsevier Inc. 1Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 Wang and Marcotte, 2010). The role of PPIs in disease is high-lighted by the enrichment of missense mutations in interactioninterfaces of proteins associated with the corresponding disor-ders (Wang et al., 2012). Moreover, comparing the interactionpartners of wild-type proteins and their disease-associated var-iants can reveal disease mechanisms (Hosp et al., 2015; Zhonget al., 2009). We therefore sought to systematically investigatehow mutations in IDRs affect PPIs.IDRs often harbor short linear motifs (SLiMs) that mediate theirfunction (Fuxreiter et al., 2007; Van Roey et al., 2014). TheseSLiMs typically fall into two major classes—motifs that mediateinteractions with globular domains and/or motifs that harborposttranslational modification sites (Tompa et al., 2014). Muta-tions in IDRs can cause disease by disrupting such motifs orby creating novel ones. A number of examples of such patho-genic changes in motifs have been reported (Cordeddu et al.,2009; Kadaveru et al., 2008; Silvis et al., 2003; Vogt et al.,2005). Additionally, computational studies have revealed thatpathogenic mutations often occur in SLiMs (Narayan et al.,2016; Radivojac et al., 2008; Uyar et al., 2014). Despite these in-sights, there has not yet been a systematic experimental analysisof how disease-causing mutations in IDRs affect interactions.One reason for this is that the small binding area between SLiMsand cognate domains results in low binding affinities, whichmakes it difficult to study these interactions (Neduva and Rus-sell, 2005).Here, we developed a scalable proteomic screen using syn-thetic peptides to assess the impact of missense mutationsin disordered regions on PPIs. By applying this screen to over120 known disease-causing mutations, we obtained a networkof PPIs that are lost or gained as the result of mutations in disor-dered regions. Within this network, we identified a subnetworkcomprising three mutations and five interacting proteins en-riched in terms related to clathrin-dependent trafficking. Intrigu-ingly, all three mutations in this subnetwork create novel dileu-cine motifs in cytosolic tails of transmembrane proteins.Because dileucine motifs mediate clathrin-dependent traf-ficking, our findings provide a mechanistic explanation abouthow these mutations cause disease. Indeed, experiments onthe glucose transporter GLUT1 confirmed that the gained dileu-cine motif causes protein mislocalization by recruiting adaptorproteins and inducing clathrin-dependent endocytosis. In sum-mary, we show that our scalable proteomic screen can revealthe functional consequences of mutations in disordered regions.The data suggest that dileucine motif gains in disordered cyto-solic tails of transmembrane proteins are a relatively frequent—and potentially druggable—cause of disease.RESULTSA Peptide-Based Interaction Screen of Disease-CausingMutationsWe reasoned that quantitative interaction proteomics with im-mobilized synthetic peptides should enable us to systematicallyassess the impact of mutations in IDRs. Such peptide pull-downs can maintain specificity even in the setting of low-affinityinteractions (Schulze and Mann, 2004). Peptides can also bedirectly synthesized on cellulose membranes and used for inter-action screens via mass spectrometry (Dittmar et al., 2017;Frank, 2002; Okada et al., 2012). We followed this strategy todesign a scalable proteomic screen (Figure 1A): pairs of peptideswith 15 amino acids that correspond to IDRs in both the wild-type and mutant form are synthesized on cellulose membranes.These membranes are incubated with cell extracts to pull-downinteracting proteins. After washing, peptide spots are excisedand the proteins associated with them are identified and quanti-fied by shotgun proteomics.The main challenge in such interaction screens is to distin-guish specific interaction partners from non-specific contami-nants (Gingras and Raught, 2012; Gstaiger and Aebersold,2009; Meyer and Selbach, 2015; Smits and Vermeulen, 2016).We addressed this challenge through the use of two levels ofquantification. First, two replicates of a pull-down with a specificpeptide sequence are compared to all other peptide pull-downsvia label-free quantification (LFQ) (Cox et al., 2014). This LFQ-fil-ter selects proteins that bind specifically to a given peptide. Sec-ond, the screen employs stable isotope labeling by amino acidsin cell culture (SILAC)-based quantification (Mann, 2006) to iden-tify differential interaction partners of the wild-type and disease-causing form of a peptide. This strategy requires incubating tworeplicates of the membrane with cell lysates that have beendifferentially SILAC-labeled. Wild-type peptide spots from theheavy pull-down are combined with spots from the light pull-down that correspond to the mutant forms of the same peptideand vice versa. SILAC ratios give a measure of the degree towhich each particular mutations affects a specific interaction.For the screen, we selected 128 mutations in IDRs that areknown to cause neurological diseases (Figure S1;Table S1).We included a peptide from an IDR in the SOS1 protein that con-tains a proline-rich motif by which it is known to recruit severalspecific binders via their SH3 domains (Schulze and Mann,2004). We analyzed the 2 3129 pull-down samples by usinghigh-resolution shotgun proteomics in 45-min runs, resulting ina total measurement time of 8 days. Replicates of the samepeptide clustered with a median correlation coefficient (Pear-son’s R) of 0.87, indicating good reproducibility (Figure S2A).The LFQ data identified nine specific interactors of the SOS1peptide, including four of the five that were previously known(Figure 1B). In the corresponding SILAC data, seven of the nineLFQ-specific binders show preferential binding to the wild-typecompared to the mutant, which contains a disrupted proline-rich motif (Figure 1C). Importantly, all interactors that are bothspecific (LFQ) and differential (SILAC) contain SH3 domains.To further assess the relationship between peptide motifs andcognate domains, we analyzed all pull-downs combined. Wefound that mutations that disrupt a predicted SLiM in the peptidetend to reduce binding of proteins with cognate domains (Fig-ure S2B). Conversely, the gain of a SLiM in a peptide tendsto increase binding to proteins with matching domains. Weconclude that our screen efficiently detects how mutations inIDRs affect interactions mediated by SLiMs.A Quantitative Interaction Network forDisease-Associated IDRsIndividualpull-downs typically led to the identificationof 400 pro-teins. If all of these proteins were specific binders, this would2Cell 175, 1–15, September 20, 2018Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 correspond to 400 binary interactions per pulldown and morethan 100,000 interactions in total. However, because many pro-teins are background binders, we applied our quantitative filterswith cut-offs derived from the SOS1 control peptide (Figure 2A).Approximately half of the 2 3128 peptides showed at least onespecific binder according to the LFQ-filter (Figure S2C). ApplyingtheLFQ-filterdramaticallyreducedthe totalnumberof interactionsto 618. All of these 618 interactions are specific for the wild-typeand/or the mutant form of a peptide as compared to all other pep-tides in the screen (Table S2). However, not all of these specificABCFigure 1. Quantitative Interaction Screen with Disease-Associated Disordered Regions(A) Cellulose membranes with synthetic wild-type (circles) and mutated (stars) peptides are incubated with lysate from light (light blue) or heavy (dark blue) SILAC-labeled cells to pull-down interacting proteins. Spots are excised, corresponding wild-type/mutant pairs are combined and analyzed by quantitative shotgunproteomics (represented by an Orbitrap). Middle: label-free quantification (LFQ) identifies specific interactors by comparing both replicates to all other pull-downs. Volcano plots depict protein enrichment in in the two replicate pull-downs of a given peptide over all other peptide pull-downs, separately for the wild-type(left) and mutant peptide (right). The threshold (red lines) was derived from the benchmark experiment with the SOS1 peptide (B, for details see STAR Methods).LFQ-specific interactors are depicted in red. SILAC-based quantification identifies differe ntial binders by directly comparing corresponding wild-type and mutantpairs. Differential binders of the wild-type and mutant peptide appear in the upper-right and lower-left quadrants, respectively (for detailed selection criteria seeSTAR Methods, see also Figure S1 and Table S1).(B and C) Results for a SOS1-derived peptide with a SH3 domain-binding PxxP motif as a benchmark.(B) Volcano plot from LFQ data for wild-type SOS1. Specific binders are shown as red dots. 4 out of 5 known binders (red gene names) are detected.(C) Differential binders of the wild-type and mutant SOS1 peptide. Proteins with SH3 domains are shown with black outlines.Cell 175, 1–15, September 20, 2018 3Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 interactions are differential (i.e., affected by the mutation). There-fore, we next applied the SILAC-filter, which led to a final list of180 differential interactions (Table S3). 111 of these interactionsare lost through mutations in the peptide, while 69 are gained. Ofnote, because pull-downs can capture indirect binders, not all ofthese interactions are necessarily direct.To provide an overview of the data, we displayed the differen-tial interactions as a network (Figure 2B). This revealed thatseveral wild-type or mutant peptides shared differential interac-tors, suggesting functional similarities. Moreover, subnetworkswere enriched in specific gene ontology terms (Data S1). Fig-ure 2B highlights two subnetworks that we find particularly101,765detectedinteractions618specificinteractions180differentialinteractionsgainedlostBACSNK1DZNF768MORF4L2CCDC137DDOSTRPL39P5 C19orf53LANCL1PDIA3HNRNPKMAPRE1YWHAGPDIA4HDGFYWHAHH1FXSRP14HIST1H1DCHMP2B_D148YFAM91A1 RBMXDARS2DDX49G3BP2MAP2FTSJ3SFPQBCKDKPDCD11NXF1AURKAL1CAM_S1194LNONORPUSD3HNRNPA0CSNK1A1HNRNPA1regulation of mRNA splicing,via spliceosome (p = 3.9e-12) SRSF10SRSF2FUS_R521CSRSF5SRSF3SRSF1TRA2BSRSF9SRSF6EARS2_E96KALYREFCACNA1A_R2135CRPN2EMG1SLTMCOPAFLOT2FLOT1ERLIN2GJB1_R264CDKC1_G402RANP32BANP32EADAR_K999NTCF4_R565WTMEM240_P170LSETX_R1294CWWOX_P47TANP32A SETACADMClathrin-coated vesicle (p = 1.4e-05)ITPR1_P1059LCACNA1H_P648LDECR1CLTB CLTASLC2A1_P485LCLTCCLINT1IPO9CACNA1H_A748VCD2APCAPZA1AAAS_Q15KIPO7PYCR1RPUSD4ZC2HC1ACHTOPWDR11G3BP1NME4AP2A1CC2D1APRIC295RRP12XPO1CLN6_R6TDEPDC5_S1073REGR2_I268NERHMECP2_G161VGPHNGCH1_P23LVCPIFT140_E664KTRAF2SMPD1_A196PRCN2CDKL5_N399TMYCBP2MATR3_T622ASPTAN1TYMP_R44QSSBP1SETBP1_I871TDNCL1GJB1_C280GPOLDIP2POLRMTCC2D2A_T1114MUBR4P4HBZEB2_Q1119RCKMT1AMDN1HUWE1UBL5AF1QPHGDGSOD1_I152TDYNC1H1ERCC6_P1042LGJB1_R230CC1QBPNOC2LAP3B1TCOF1EARS2_R168GPANK2_E134GPTMAFUS_R216CCASR_R898QSPAST_P293LHSP90B1PPM1GTRPV4_R315WNAP1L1CALRSUPT5HRUVBL1 RUVBL2TUBB2BCALUUBA1_S547GIPO11MAP1BSSR4DARS2_Y629CTPM3PCNAPDIA6RBBP4CKBNASPYWHAZRTL1CCDC47SEC63HDGFL2CTNNBL1 ZC4H2_R213WTINF2_K280ESRRM2DDX39BHSP90AA1Peptide-protein interactionspeptidebait-5.73 4.91log2FC SILAC lostgainedFigure 2. Differential Interactors of Wild-Type and Mutant IDRs(A) Quantitative filters to select specific and differential interactions. Only a minor fraction of all detected interactions is specific (LFQ filter). Moreover, only afraction of specific interactions are differential (SILAC filter), i.e., show preferential binding to the wild-type or mutant form of a peptide. Mutation-inducedinteraction losses are more frequent than mutation-induced gains. See also Figure S2.(B) Network of all differential interactions. Peptides (rectangles) and interacting proteins (ovals) are presented as nodes. The edges indicate preferential binding tothe wild-type (blue) or mutant (red) form of a peptide (edge darkness indicates SILAC ratios). Highlighted subnetworks are enriched in splicing regulators andclathrin-coated vesicle proteins (see text).See also Tables S2 and S3 and Data S1.4Cell 175, 1–15, September 20, 2018Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 interesting (insets): one is enriched in proteins connected to cla-thrin-coated vesicles (see below). The other is enriched in splicingfactors that interact with an IDR corresponding to amino acids512–526 of fused in sarcoma (FUS). These interactions are dis-rupted by the R521C mutation. FUS is an RNA-binding proteinimplicated in amyotrophic lateral sclerosis (ALS) (Deng et al.,2014). R521C and other mutations in the C-terminal region ofthe protein are thought to cause disease by disrupting a nuclearlocalization signal (Dormann et al., 2010). Our data suggest thatimpaired binding of splicing factors could be an additional/alter-native explanation for the pathogenicity of this mutation. Interest-ingly, FUS has already been implicated in splicing (Ishigaki et al.,2012; Qiu et al., 2014; Rogelj et al., 2012). In fact, the C-terminalregion of the protein was found to interact with serine/arginine-rich splicing factor 10 (SRSF10) even before pathogenic muta-tions in this region were identified (Yang et al., 1998).Recruitment of Clathrin through Gains of DileucineMotifsThe finding we considered most interesting is that mutated IDRsfrom CACNA1H, GLUT1/SLC2A1, and ITPR1 lead to specific in-teractions with clathrin (Figure 3A). The corresponding SILACdata revealed that, in all three cases, clathrin exhibited a strongpreference for the mutant form of the peptides over the wild-type(Figure 3B). Because clathrin mediates endocytosis and intracel-lular trafficking of transmembrane proteins, our finding suggeststhat these mutations might affect protein trafficking. Intriguingly,the three mutations share other features beyond increased cla-thrin binding: First, all three mutations affect transmembraneproteins—a calcium channel (CACNA1H) and a glucose trans-porter (GLUT1) residing in the plasma membrane and an inositol1,4,5-trisphosphate receptor (ITPR1) located mainly in the endo-plasmic reticulum (ER) (Figure 3C). Second, all three mutationsaffect disordered regions exposed to the cytosol, which makesthem accessible to cytosolic adaptor proteins that mediate cla-thrin recruitment. Third, all three mutations involve the changeof a proline to a leucine residue and thereby result in the appear-ance of a novel dileucine motif (‘‘LL’’) in the IDR (Figure 3D). Suchmotifs are known to recruit clathrin to the plasma membrane orintracellular locations (Pandey, 2009). The classical dileucinemotif is [D/E]XXXL[L/I] (Dinkel et al., 2016), but variations of thistheme are common (Kozik et al., 2010; Pandey, 2009; Staudtet al., 2017; Traub, 2009).A Dileucine Motif Gain Causes Mislocalization of theGlucose Transporter GLUT1To assess the functional significance of the dileucine motif gainsand clathrin recruitment, we selected the P485L mutation inGLUT1/SLC2A1. This mutation causes GLUT1 deficiency syn-drome (G1DS), a disorder characterized by seizures and intellec-tual disability with onset in early infancy (De Vivo et al., 1991;Leen et al., 2010; Pascual et al., 2008). GLUT1 is mainly ex-pressed in endothelial cells that form the blood-brain-barrierand in astrocytes, facilitating glucose entry into the brain. Path-ogenic mutations in GLUT1 impair cerebral glucose flux, leadingto permanent encephalopathy.To determine the impact of the P485L mutation on the subcel-lular localization, we first generated stable inducible cell lines ex-pressing epitope tagged full-length wild-type or mutant GLUT1.While the wild-type protein mainly localized to the plasma mem-brane, the P485L mutant displayed an intracellular pattern (Fig-ure 4A). Hence, the mutation indeed causes protein mislocaliza-tion. Colocalization experiments with several Markers confirmedthat mutated GLUT1 localizes to endocytic compartments (Fig-ures 4B and S3). To more systematically characterize the cellularcompartment in which GLUT1_P485L resides, we used BioID asa proximity labeling method (Roux et al., 2012)(Table S4). Weperformed this experiment in a comparative manner for bothwild-type and mutant GLUT1 using SILAC-based quantification(see STAR Methods). Mutated GLUT1 colocalized with proteinsinvolved in membrane trafficking, clathrin-mediated endocytosis,and post-Golgi trafficking (Figure 4C). In contrast, wild-typeGLUT1 colocalized with plasma membrane-associated proteins.Adaptor Proteins Bind to Mutant GLUT1 and CauseCellular MistraffickingThe BioID experiment identified several subunits of heterotetra-meric vesicular transport adaptor proteins (APs). This finding isparticularly relevant because AP-1, AP-2, and AP-3 directlybind to both dileucine motifs and clathrin to mediate cellulartransport (Traub and Bonifacino, 2013). Binding of AP com-plexes to the cargo triggers a conformational change, whichopens up the AP complex. It now exposes a ‘‘Clathrin box motif’’that leads to recruitment of clathrin, which begins to surround theemerging vesicle bud as a second protein layer.All subunits of AP-1, AP-2, and AP-3 that we identified showedincreased co-localization with mutated GLUT1 when comparedto the wild-type in both the forward and reverse (that is, SILAClabel swap) BioID experiment (Figure 5A). We therefore testedwhether the cytosolic tail of GLUT1 can interact with APsin vitro and found that mutated, but not wild-type, GLUT1 pulleddown both AP-1 and AP-2 (Figure 5B). Together, these resultsshow that the mutation causes association of GLUT1 with APs,providing a molecular explanation for mistrafficking. Becausewe did not detect APs as hits in our original screen, we designedtargeted assays against peptides from several APs based ontheir known fragmentation spectra (Zauber et al., 2018).Repeating the peptide pull-downs for GLUT1, CACNA1H andITPR1 with targeted proteomics as readout, confirmed thatseveral APs preferentially interact with the mutated peptides(Figure S4).APs localize to different intracellular compartments andmediate membrane trafficking in distinct pathways (Park andGuo, 2014). For example, AP-2 mediates clathrin-dependentendocytosis at the plasma membrane (McMahon and Boucrot,2011; Pandey, 2009; Staudt et al., 2017; Traub, 2009). The inter-action of GLUT1_P485L with AP-2 thus suggests that internaliza-tion of the protein from the plasma membrane contributes toits mislocalization. To test if mutant GLUT1 is taken up via endo-cytosis, we added fluorescently labeled transferrin to GLUT1expressing cells. Mutant, but not wild-type, protein extensivelycolocalized with endocytosed transferrin (Figures 5C and 5D). Ifthe P485L mutation causes GLUT1 endocytosis via AP-2, inhib-iting AP-2 function should restore the correct subcellular localiza-tion. We therefore used small interfering RNAs (siRNAs) to knockdown AP-2 (Figure 5E). Consistent with our prediction, loss ofCell 175, 1–15, September 20, 2018 5Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 AP-2 expression rescued the mislocalization of mutated GLUT1(Figure 5F). We also tested the ability of stable cell lines induciblyexpressing GLUT1 to take-up radiolabeled glucose. Cells ex-pressing the mutated protein showed significantly increasedglucose uptake when AP-2 was knocked down (Figure 5G).GLUT1 P485L Mislocalizes in Patient-Derived InducedPluripotent Stem CellsThe experiments described so far are based on in vitro assays orcell line models expressing tagged variants of GLUT1. We there-fore sought to validate our findings by analyzing the behavior ofendogenous GLUT1 in patient cells. To this end, we obtained fi-broblasts from a GLUT1-deficient patient harboring the P485Lmutation via a skin punch biopsy (see STAR Methods). Becausefibroblasts do not express significant amounts of GLUT1, wereprogrammed them into induced pluripotent stem cells (iPSCs)by RNA-based transfer of pluripotency factors (Figure 6A). Theclones obtained showed characteristic expression of pluripo-tency markers (Figure S5). We then analyzed the subcellulardistribution of GLUT1 in these cells. While GLUT1 was mainlyCLTB CLTCCLTACLINT1DECR10.00.51.01.5-8 -6 -4 -2 0 2 4 6 8CLTC CLTACLTB0.00.51.01.5-8 -6 -4 -2 0 2 4 6 8GLUT1CACNA1HITPR1LLLLLLcytoplasm ERP485L P648LP1059L0.01.02.0TPGPGTSEGE5LGENHFTHGLL10SGDALNDDSQPH15VGD0.01.02.0bitsTPGPGTSEGE5LGENHFTHGPL10SGDALNDDSQPH15VGDPGTGGHGPLSLNSPD PGTGGHGLLSLNSPDTPEELFHPLGADSQV TPEELFHLLGADSQVGGSEENTPLDLDDHG GGSEENTLLDLDDHGtnatumepyt-dliwCACNA1HGLUT1ITPR1CLTA0.00.51.01.5-8 -6 -4 -2 0 2 4 6 8Log2FC Mut/other peptides-log[10]pvalueLog2FC Mut/other peptides Log2FC Mut/other peptides-log[10]pvalue-log[10]pvalueABCDCACNA1H P648L GLUT1 P485L ITPR1 P1059LCLTCCLTA-8 -6 -4 -2 0 2 4 6 8Ratio L/H Log2FC Wt/MutRatio H/L Log2FC Wt/Mut-2-4-6-8CLTBDECR1CLTCCLTACLINT1-8 -6 -4 -2 0 2 4 6 8Ratio L/H Log2FC Wt/MutRatio H/L Log2FC Wt/Mut-2-4-6-8CLTBCLTCCLTA-8 -6 -4 -2 0 2 4 6 8Ratio L/H Log2FC Wt/MutRatio H/L Log2FC Wt/Mut-2-4-6-8Figure 3. Recruitment of Clathrin by Recurrent Gains of Dileucine Motifs(A) Volcano plots for pull-downs with mutated peptides derived from CACNA1H, GLUT1, and ITPR1. Specific binders (relative to all other pull-downs) arehighlighted in red. All three peptides specifically interact with clathrin.(B) Corresponding SILAC plots show that clathrin and related proteins preferentially bind to the mutant form of peptides (relative to the wild-type).(C) Graphical representation of the mutation sites. All three mutations affect cytosolic regions of transmembrane proteins. CACNA1H and GLUT1 are locatedmainly in the plasma membrane and ITPR1 mainly in the ER.(D) Aligning the three peptide sequences reveals a common gain of a dileucine motif.6Cell 175, 1–15, September 20, 2018Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 localized in the plasma membrane of control iPSCs, patient-derived cells showed characteristic intracellular accumulations(Figure 6B). Co-staining with a post-Golgi SNARE revealedextensive colocalization (Kreykenbohm et al., 2002)(Figures6C and 6D). Of note, the patient-derived iPSCs also showedGLUT1 signal at the plasma membrane. This is consistent withthe fact that only one GLUT1 allele in the patient is affected(Leen et al., 2010; Pascual et al., 2008; Slaughter et al., 2009).In summary, these data show that the mislocalization observedin HEK cells can be reproduced in patient cells.ACDELateendosomeEndocyticdegradationMVELysosome NucleusCIEEarlyendosomeTGNGolgiEndoplasmic reticulumRecyclingendosomeSlow recyclingFast recyclingCytosolPlasma membraneC-3-2-10123-3 -2 -1 0 1 2 3Ratio M/H Log2FC Wt/MutRatio H/M Log2FC Wt/MutFLAG-GLUT1 FLAG-GLUT1_P485L0510 15 20 25 30adherens junction(GO:0005912)cell-cell junction (GO:0005911)cell leading edge (GO:0031252)septin complex (GO:0031105)actin cytoskeleton (GO:0015629)extrinsic component of membrane (GO:0019898)synapse part (GO:0044456)cell projection membrane (GO:0031253)membrane raft (GO:0045121)basal part of cell (GO:0045178)clathrin-coated vesicle (GO:0030136)endosome membrane (GO:0010008)vacuole (GO:0005773)trans-Golgi network (GO:0005802)SNARE complex (GO:0031201)early endosome (GO:0005769)perinuclear region of cytoplasm (GO:0048471)Ragulator complex (GO:0071986)AP-3 adaptor complex (GO:0030123)trans-Golgi network transport vesicle (GO:0030140)-log10(P)0.10.00.10.20.30.4EEA1 Rab4 Rab9 LAMP1 VTI1A VTI1BPearson s correlation coefficientWTP485LBFigure 4. A Mutation-Induced Dileucine Motif Gain Causes Mislocalization of the Glucose Transporter GLUT1(A) Confocal images of GLUT1 localization in HEK cells, stably expressing FLAG-GLUT1, reveal that the wild-type is localized mainly at the cell membrane whilethe P485L mutant is mislocalized to endocytic compartments. (green, FLAG-GLUT1; blue, DAPI). Scale bars, 10 mm.(B) Colocalization analysis shows extensive colocalization of mutant, but not wild-type GLUT1 with markers of several endocytic compartments. Pearson’sthresholded coefficients (as implemented in the Imaris software) were determined for GLUT1 variants with the indicated proteins. Data are represented asmean ±SD. See Figure S3 for example images.(C) Comparison of proteins colocalizing with wild-type and mutant GLUT1 by proximit y labeling (BioID). The upper left panel shows SILAC log2 fold changes fromtwo replicate experiments with swapped isotope labels. Blue and red labeled proteins are enriched by wild-type GLUT1 or mutant GLUT1, respectively. The tenmost significant cellular component GO-terms reveal that mutated GLUT1 is involved in clathrin-dependent processes and endosomal trafficking. In contrast,wild-type GLUT1 colocalizes with plasma membrane-associated proteins. The lower panel is colored according to the top three enriched GO-terms and showsthe variants typical subcellular compartments. Figure adapted from Raiborg and Stenmark (2009).See also Table S4.Cell 175, 1–15, September 20, 2018 7Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 FLAG-GLUT1 P485LTransferrinTransferrinOverlayOverlayAP-1 AP-1 AP-2 AP-2 AP-2 AP-3 AP-3 AP-3 AP-1 AP-1 AP-2 AP-2 AP-2 AP-3 AP-3 AP-3 AP-1 AP-2 InputGSTGST-GLUT1GST-GLUT1_P485LAP-2 AP-2 Profilin1siRNAGLUT1 GLUT1P485LAP-2 AP-2 controlcontrolsiRNAGLUT1 GLUT1_P485LAP-2  controlGLUT1AP-2 αOverlayAP-2  control-3 -2 -1 0 1 2 3Log2 SILAC ratiosGLUT1/GLUT1_P485LGLUT1_P485L/GLUT1111AP-1222AP-2333AP-3ABCFTalinDE0.1 0.0 0.1 0.2 0.3 0.4TfPearson s correlation coefficientWTP485LGLUT1TfOverlayGLUT1 GLUT1_P485LG% glucose uptake100806040200120doxycycline ++-AP-2  siRNA -+-+-++-- +-+cytochalasin B +-WTP485Ln.s. **Figure 5. Adaptor Proteins Bind to Mutated GLUT1 and Cause Cellular Mistrafficking(A) Adaptor protein complexes AP-1, AP-2, and AP-3 show increased colocalization to GLUT1 due to P485L mutation in replicates of BioID experiment fromFigure 4. Identified subunits of APs are shown in red. See also Figure S4.(B) P485L mutant but not wild-type Glut1 C-terminal tail interacts with AP-1 and AP-2. Tails were tagged with GST to pull-down interaction partners from mousebrain lysate. Talin is shown as a negative control and is not pulled down from either of the two variants.(C and D) Mutant but not wild-type GLUT1 extensively colocalizes with endocytosed transferrin. HEK cells stably expressing FLAG-GLUT1 are incubated withfluorescently labeled transferrin for 10 min before fixation. Scale bar, 10 mm (C). Data are represented as mean ±SD (D).(E and F) Western blot against AP-2 aand msubunits shows downregulation after two rounds of siRNA transfection against AP-2 m. AP-2 knockdown leads torelocalization of GLUT1_P485L to the plasma membrane and hence rescue of the mutation phenotype. Scale bar, 10 mm.(legend continued on next page)8Cell 175, 1–15, September 20, 2018Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 GLUT1_P485L Localization at the Blood-Brain-Barrier IsPerturbed In VivoTo study the P485L mutation in vivo, we used CRISPR/Cas9 togenerate the corresponding mutant in the mouse (Figure 6E).Heterozygous mice were viable, fertile, and did not display anyobvious phenotype. However, out of 6 heterozygous crossings,we failed to detect any born homozygous mutant pup. A detailedinspection of maternal delivery illustrated that homozygousmutant pups come to term but die immediately after birth andare removed from the litter by the dams. To histologically analyzeGLUT1 distribution in the endothelial cells of the blood-brain-barrier, we dissected embryonic (E) day 14.5–15.5 mice andstained their cerebral cortex with antibodies against GLUT1.We co-stained with anti-ICAM2 to label the luminal plasma mem-brane and isolectin B4 (IB4) to label the entire endothelial plasmamembrane. We observed an overall reduction in GLUT1 stainingin heterozygous and homozygous animals with otherwise largelynormal vascular morphology (Figure 6F). High resolution STEDimaging revealed a strong reduction in the density of GLUT1clusters in endothelial plasma membranes of homozygousmutant mice compared to wild-type littermates (Figures 6Gand 6H). Hence, the P485L mutation also reduces GLUT1 levelsin the plasma membrane in vivo.Gains in Dileucine Motifs as a General DiseaseMechanismWe next investigated whether dileucine motif gains represent amore general disease mechanism. To test this, we conducteda search of missense mutations associated with disease thatoccur within disordered cytosolic regions of transmembraneproteins. We found four additional pathogenic dileucine motifgains in the Humsavar database (Figure 7A; Table S5). Thesame search in the ClinVar database returned four additionalmutations. For example, two dileucine motif gains affectcytosolic regions of the cystic fibrosis transmembrane conduc-tance regulator (CFTR) and cause cystic fibrosis (Figure S6A;Table S5). In total, 11 dileucine motif gains in 8 different proteinscause a wide range of diseases.Because we focused our follow-up experiments on GLUT1,we cannot state with certainty that the other dileucine motif gainsalso cause protein mistrafficking. Alternatively, the mutationscould cause disease by different mechanisms and might justcreate dileucine motifs as a by-product. If that was the case,dileucine motifs should be homogeneously distributed betweendisease-causing mutations and non-pathogenic polymor-phisms. In contrast, if dileucine motif gains are responsible forpathogenesis, they would be predicted to occur more often indisease than in non-pathogenic variants. Moreover, pathogenicdileucine motif gains should be specific for cytosolic regions oftransmembrane proteins because this is where they exert theirfunction. To test these predictions, we compared the frequencyof dileucine motif gains found in disease-causing mutations totheir appearance in non-pathogenic polymorphisms (fromHumsavar). A global survey of all disordered regions of the entireproteome revealed that dileucine motif gains occurred at aboutthe same rate in disease and non-pathogenic variants (odds ratio[OR] = 0.81, p value = 0.319, two-sided Fisher’s exact test).In the cytosolic tails of transmembrane proteins, however, weobserved a 3.7-fold enrichment of dileucine motifs implicatedin disease (OR = 3.7, p value = 0.017, two-sided Fisher’s exacttest, Figure 7B). Disordered extracellular regions of transmem-brane proteins do not show this enrichment. Performing theequivalent analysis with the ClinVar database yielded similarresults (Figure S6B). We conclude that dileucine motif gains indisordered regions of the cytosolic segments of transmembraneproteins are significantly and specifically enriched in disease. Tofurther assess the significance of this finding, we systematicallysearched within cytosolic regions of transmembrane proteins forall other annotated SLiMs contained in the ELM database (Dinkelet al., 2016). Intriguingly, of all 263 SLiMs tested, the dileucinemotif (LIG_diLeu_1) was the only significantly enriched motif indisease (Figure 7C).Finally, to test if some of the additional dileucine motifgains can cause mistrafficking, we performed antibody feedingexperiments. We created chimeric proteins consisting of theIL-2 receptor alpha chain (TAC) fused to mutated and wild-type cytosolic regions of the respective disease protein (seeSTAR Methods). Cells expressing these constructs are incu-bated with antibodies against the extracellular region of TACand allowed to endocytose the chimeric proteins together withbound antibodies. A specific staining protocol is then used toexclusively detect internalized antibodies (Diril et al., 2009). Wegenerated fusion proteins for GLUT1 (as positive control) andseven additional dileucine motif gains. Four of the sevenmutations resulted in increased internalization relative to thecorresponding wild-type sequences (Figure 7D). In addition,we observed mutant-specific interaction of AP-1 and/or AP-2for several selected cytosolic regions in in vitro interactionassays (Figure S6D). Collectively, these results indicate thatseveral additional pathogenic dileucine motif gains cause proteinmistrafficking.DISCUSSIONUnderstanding the functional relevance of protein variants isa major challenge in the era of personal genomics–especiallyfor missense mutations in disordered regions. Our proteomicscreen provides a first systematic experimental analysis of howmutations in disordered regions affect protein-protein interac-tions. Our results show that the method can (1) capture knowninteractions, (2) detect how mutations in SLiMs affect bindingof cognate domains, and (3) provide mechanistic insights intopathogenesis. The peptide-based method is especially usefulfor mutations in proteins that are otherwise difficult to study,(G) AP-2 knock down leads to rescue of glucose uptake in GLUT1_P485L expressing cells. GLUT1 expression was induced by doxycycline, cytochalasin Binhibition was used as control. % glucose uptake is relative to GLUT1 wild-type, +doxycycline, cytochalasin B. Mean values of technical triplicates from threeindependent experiments are shown. We only compared glucose uptake within and not between cell lines to avoid possible differences between clones. Errorbars, SEM. **p value 0.01 from a paired, one-sided t test.Cell 175, 1–15, September 20, 2018 9Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 HLLGLUT1VTl1AOverlayGLUT1 GLUT1_P485LControl GLUT1-DS patientAB CDEControlPatient0.0 0.1 0.2 0.3VTI1APearson s correlation coefficient+/++/P485LP485L/P485LGLUT1 DAPI GLUT1 IB4 ICAM2GLUT1FIB4 ICAM2+/+N=3n=44+/P485LN=3n=43P485L/P485LN=3n=47GCas9LLLLSkin biopsy Reprogrammingc-MYCOCT4 GLIS1SOX2 KLF-4RNA-basedFibroblasts iPSCsZygotegRNArecombinationtemplateGLUT1 DAPIexon1 exon2 exon3-8exon9-10SLC2A15 CCCGAGGAGCTCTTCCACttgtTGGGGGCGGACTCCCAAGTGTGA3 3 GGGCTCCTCGAGAAGGTGaacaACCCCCGCCTCAGGGTTCACACT5 5 CCCGAGGAGCTCTTCCACCCTCTGGGGGCGGACTCCCAAGTGTGA3 3 GGGCTCCTCGAGAAGGTGGGAGACCCCCGCCTGAGGGTTCACACT5 P LL LgRNA PAM+/++/P485LP485L/P485Llog2(GLUT1/IB4)p = 3.7e-18n.s.543210-1-2-3Figure 6. GLUT1 P485L Mislocalizes in Patient-Derived iPSCs and Endothelial Cells of the Blood-Brain Barrier in Mice(A) A skin sample was taken from a GLUT1 deficiency patient with a heterozy gous GLUT1_P485L mutation. Fibroblasts were grown and reprogrammed to iPSCs.See also Figure S5.(B) Heterozygous GLUT1_P485L mutation leads to partial mislocalization of GLUT1 in patient-derived iPSCs. Scale bars, 10 mm.(C and D) GLUT1_P485L colocalizes with the post-Golgi SNARE VTI1A. Scale bar, 10 mm (C). Data are represented as mean ±SD (D).(legend continued on next page)10 Cell 175, 1–15, September 20, 2018Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 such as large transmembrane proteins. Nevertheless, it is alsoimportant to consider the intrinsic limitations of the approach.Most importantly, in vitro pull-downs do not necessarily reflectphysiological interactions in vivo. For example, artifactual bind-(E) A mouse carrying the GLUT1 P485L mutation was created by CRISPR/Cas9-targeted method. PAM sequence and gRNA are marked in targeted region ofSLC2A1 (GLUT1 gene). Sanger sequencing confirmed insertion of mutation (chromatogram: A = green, T = red, C = blue, G = black).(F) Immunohistological analyses of cortical slices of wild-type, heterozygous, and homozygous GLUT1 mutant mice using antibodies against GLUT1 (red) andDAPI (blue) as counterstain (left panels); a higher magnification of a vessel stained by antibodies against GLUT1 (red), IB4 (green), and ICAM2 (blue) is shown in theright panels.(G) Representative STED images of transverse cross-sections through brain vessels of wild-type (+/+), heterozygous (+/P485L) and homozygous (P485L/P485L)mutant mice stained with isolectin B4 and antibodies against GLUT1 and ICAM2. Insets show a fragment of abluminal membrane (IB4 positive, ICAM2 negative)indicated with a black box. Scale bars, 2 mm (main panels); 0.25 mm (insets).(H) Quantification of GLUT1 signal relative to IB4 signal in vessel membranes (n, number of vessels per genotype; N, number of animals per genotype). Boxplotcentral line indicates the median, the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively.All proteins TransmembraneproteinsCytosplasmicdomainsoftransmembraneproteinsExtracellulardomainsoftransmembraneproteins1n.s.n.s.p = 0.017CACNA1HL1CAMRHBDF2RETcytoplasmERLLLLLLLLS1149LP618LP1039LP189Ldisease mutations vs polymorphismsOdds Ratio dileucine gain23n.s.ABCGLUT1_P485ITPR1_P1059 CFTR_P5CFTR_P750KCNQ1_R452 L1CAM_S1149RHBDF2_P189 KCNQ1_R591wt mut wt mutD-log10(pval)log2 oddsRatiopolymorphism diseaseLIG_diLeu_11.510.50-2 02Figure 7. Mutation-Induced Gains of Dileu-cine Motifs Are a Significant Cause ofDisease(A) A systematic bioinformatic search revealedfour additional pathogenic mutations in cytosolicsegments of transmembrane proteins that createdileucine motifs. See also Table S5.(B) Relative frequency of dileucine motif gainsin disease mutations and polymorphisms indifferent disordered regions (IUPred score R0.4)of the proteome. Dileucine motif gain is signif-icantly enriched only in disordered regions of thecytoplasmic domains of transmembrane proteins(two-sided Fisher’s exact test).(C) Comparison of all gained motifs (disease-associated versus polymorphism) in disorderedregions of cytoplasmic tails of transmembraneproteins reveals the dileucine motif to have themost significant and specific enrichment.(D) Antibody feeding indicates that four out of seventested mutations with gain of dileucine motiflead to a gain in endocytosis (candidates fromHumsavar and Clinvar, for details on selection seeSTAR Methods). Fluorescence signal comes frominternalized antibody. Surface exposed anti-TACantigen was blocked prior to permeabilization.Scale bars, 10 mm.See also Figure S6 and Table S6.ing can occur when combining peptidesand proteins that never meet each otherin the cell (Gibson et al., 2015). Moreover,taking IDRs out of the context of the full-length protein and immobilizing them asshort peptides can affect interactions.Finally, amino acids within IDRs oftencarry posttranslational modifications—apossibility that we did not considerhere. In the future, it will be interesting toinclude modified peptides, especiallybecause mutations often affect modifica-tion sites (Narayan et al., 2016; Radivojacet al., 2008).Our screen revealed that three muta-tions in cytosolic tails of transmembraneproteins create dileucine motifs and leadto increased binding of clathrin. Follow-up experiments demon-strated that the dileucine motif gain in GLUT1 causes mislocali-zation from the plasma membrane to endocytic compartments.We also observed that mutated GLUT1 recruits several adaptorCell 175, 1–15, September 20, 2018 11Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 proteins, thus providing a direct link between the motif gain,clathrin recruitment, and GLUT1 mistrafficking. Knocking downAP-2 rescued mislocalization of mutated GLUT1 and restoredglucose transport. This finding shows that the aberrant traf-ficking is—at least partially—due to increased endocytosis.However, it should be kept in mind that AP-1 and AP-3 mediateother trafficking events such as transport between endo-somes and the trans-Golgi network or transport to lysosomes(Dell’Angelica, 2009; Park and Guo, 2014). The observationthat mutated GLUT1 also interacts with these APs thereforesuggests that other trafficking events may also be perturbed.The finding that pathogenic mutations in cytosolic tails of othertransmembrane proteins also create dileucine motifs is particu-larly intriguing. We term diseases caused by such motif gains‘‘dileucineopathies’’. Whether the other dileucine motif gainscause protein mislocalization similar to GLUT1 remains to beinvestigated. The observation that these mutations are signifi-cantly and specifically enriched in cytosolic domains suggeststhat at least some of them are functional. Also, we have foundthat four out of seven tested mutations increase internalizationof chimeric proteins in antibody feeding experiments. Thisfurther supports the view that at least some pathogenic dileucinemotif gains cause disease by inducing protein mistrafficking.However, more detailed follow-up experiments are required totest this hypothesis for individual mutations. It is also interestingto note that a pathogenic mutation in the cystic fibrosistransmembrane conductance regulator (CFTR) has been re-ported to generate a tyrosine-based internalization motif (Silviset al., 2003).Why are dileucine-motif gains a recurrent cause of disease?We think this is due to a combination of several factors: First, di-leucine motifs are not very complex and can thus easily arise bychance. Second, proline codons can mutate to leucine codonsby changing a single nucleotide. Third, proline is overrepre-sented in IDRs, which are also the sites where the motif needsto be located in order to be functional. Because the geneticcause of many diseases has not yet been identified (Boycottet al., 2013), we expect that more pathogenic dileucine motifgains will soon emerge. Knowing that such gains can be patho-genic will make it easier to classify them as disease-causingamong the many variants present in the human population(Cooper and Shendure, 2011). The observation that GLUT1 mis-localization and glucose transport can be rescued suggests thatpathogenic dileucine motif gains may be druggable. Whetherpatients with ‘‘dileucineopathies’’ might benefit from inhibitingspecific clathrin-dependent trafficking events remains to beinvestigated.Bioinformatic studies have established that pathogenicmutations in disordered regions often affect SLiMs (Narayanet al., 2016; Radivojac et al., 2008; Uyar et al., 2014). However,whether these predicted motif changes really affect protein-protein interactions has not yet been investigated systematically.Moreover, many motifs have not yet been defined and thusescape computational predictions (Tompa et al., 2014). Thebiochemical approach presented here provides a useful comple-mentary strategy to computational studies. Key advantagesof our setup are its scalability (by using synthetic peptides)and specificity (by employing two quantitative filters). While wefocused on neurological disorders here, the approach can alsobe applied to other types of Mendelian disorders, somatic muta-tions in cancer and also to non-pathogenic polymorphisms.STAR+METHODSDetailed methods are provided in the online version of this paperand include the following:dKEY RESOURCES TABLEdCONTACT FOR REAGENT AND RESOURCE SHARINGdEXPERIMENTAL MODEL AND SUBJECT DETAILSBCell linesBFlp-In T-Rex GLUT1BPatient-derived iPSCsBAnimal modeldMETHOD DETAILSBPeptide-protein interaction screenBPRMBBioIDBFLAG-GLUT1 localizationBTransferrin uptakeBFLAG-GLUT1 localization under AP-2 mknockdownBAntibody feeding assayBGLUT1 localization in iPSCsBFluorescence microscopy from cell cultureBImmunofluorescence in mouse tissueBRadioactive glucose uptake under AP-2 mknockdownBGST pulldown assayBAnalysis of human missense variants and short linearmotifs (SLiMs)BAnalysis of gain of SLiMs via missense variants indisordered regionsBPeptide-Protein Interaction Network AnalysisdQUANTIFICATION AND STATISTICAL ANALYSISBColocalization analysisdDATA AND SOFTWARE AVAILABILITYSUPPLEMENTAL INFORMATIONSupplemental Information includes six figures, six tables, and one data file andcan be found with this article online at https://doi.org/10.1016/j.cell.2018.08.019.ACKNOWLEDGMENTSWe are indebted to the patient and her family for providing fibroblasts. We alsothank members of the Gunnar Dittmar and Achim Leutz labs (MDC) and Ulf Re-imer (JPT) for helpful discussions in setting up the screen. We thank Philip Kim(University of Toronto) for useful suggestions on peptide selection and RussHodge (MDC) for valuable comments on the manuscript. Markus Landthaler(MDC) kindly provided stable cell lines, and the Advanced Light MicroscopyTechnology Platform (MDC) helped with microscopy. We also like to thankMartha Hergeselle and Sven Buchert (both MDC) for excellent technical assis-tance during cloning, cell culture and mouse experiments as well as Jane Re-znick (MDC) for help in setting up the glucose uptake assay. G.R. was fundedby the German Research Foundation (DFG, SFB958/A11 to M. Krauss). B.U.acknowledges funding by the German Federal Ministry of Education andResearch (BMBF) as part of the RNA Bioinformatics Center of the GermanNetwork for Bioinformatics Infrastructure (de.NBI) (031 A538C RBC). J.M.P.is supported by NIH (NS077015 and NS094257).12 Cell 175, 1–15, September 20, 2018Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 AUTHOR CONTRIBUTIONSK.M. and M. Kirchner established the screen and selected the candidates.K.M. performed most of the wet-lab experiments with contributions fromM. Kirchner and J.-Y.C. K.M. processed and analyzed the mass spectrometricdata. B.U. carried out most of the remaining bioinformatic analyses (mainlymotif based analyses) supervised by A.A. G.R. carried out the GST-pulldownassays supervised by M. Krauss and V.H. H.Z. set-up and analyzed targetedproteomic experiments. J.M.P. obtained fibroblasts from the patient. S.D.generated iPSCs from the patient fibroblasts. I.-M.R. performed experimentswith patient-derived iPSCs supervised by T.E.W. R.K. generated the mouse.L.R.H.-M. prepared samples and performed fluorescent microscopy of mousebrain sections supervised by C.B. A.S. performed STED imaging and analysissupervised by H.G. M.S. conceived and supervised the work. K.M. and M.S.wrote the manuscript with input from all authors.DECLARATION OF INTERESTSThe authors declare no competing interests.Received: January 9, 2018Revised: June 9, 2018Accepted: August 8, 2018Published: September 6, 2018REFERENCESAlexa, A., and Rahnenfuhrer, J. (2016). topGO: Enrichment Analysis for GeneOntology. R package version 2.24.0. CRAN.Boycott, K.M., Vanstone, M.R., Bulman, D.E., and MacKenzie, A.E. (2013).Rare-disease genetics in the era of next-generation sequencing: discoveryto translation. Nat. Rev. Genet. 14, 681–691.Briatte, F. (2016). ggnetwork: Geometries to Plot Networks with ‘‘ggplot2’’,R package version 0.5.1. CRAN.Cooper, G.M.,and Shendure, J. (2011). Needles in stacks of needles:finding dis-ease-causal variantsin a wealth of genomic data. Nat. Rev. Genet.12, 628–640.Cordeddu, V., Di Schiavi, E., Pennacchio, L.A., Ma’ayan, A., Sarkozy, A., Fo-dale, V., Cecchetti, S., Cardinale, A., Martin, J., Schackwitz, W., et al. (2009).Mutation of SHOC2 promotes aberrant protein N-myristoylation and causesNoonan-like syndrome with loose anagen hair. Nat. Genet. 41, 1022–1026.Costes, S.V., Daelemans, D., Cho, E.H., Dobbin, Z., Pavlakis, G., and Lockett,S. (2004). Automatic and quantitative measurement of protein-protein colocal-ization in live cells. Biophys. J. 86, 3993–4003.Couzens, A.L., Knight, J.D.R., Kean, M.J., Teo, G., Weiss, A., Dunham, W.H.,Lin, Z.-Y., Bagshaw, R.D., Sicheri, F., Pawson, T., et al. (2013). Protein interac-tion network of the mammalian Hippo pathway reveals mechanisms of kinase-phosphatase interactions. Sci. Signal. 6, rs15.Cox, J., and Mann, M. (2008). MaxQuant enables high peptide identificationrates, individualized p.p.b.-range mass accuracies and proteome-wide pro-tein quantification. Nat. Biotechnol. 26, 1367–1372.Cox, J., Hein, M.Y., Luber, C.A., Paron, I., Nagaraj, N., and Mann, M. (2014).Accurate proteome-wide label-free quantification by delayed normalizationand maximal peptide ratio extraction, termed MaxLFQ. Mol. Cell. Proteomics13, 2513–2526.Csardi, G., and Nepusz, T. (2006). The igraph software package for complexnetwork research. InterJournal. Complex Syst. 1695, 1–9.De Vivo, D.C., Trifiletti, R.R., Jacobson, R.I., Ronen, G.M., Behmand, R.A., andHarik, S.I. (1991). Defective glucose transport across the blood-brain barrier asa cause of persistent hypoglycorrhachia, seizures, and developmental delay.N. Engl. J. Med. 325, 703–709.Dell’Angelica, E.C. (2009). AP-3-dependent trafficking and disease: the firstdecade. Curr. Opin. Cell Biol. 21, 552–559.Deng, H., Gao, K., and Jankovic, J. (2014). The role of FUS gene variants inneurodegenerative diseases. Nat. Rev. Neurol. 10, 337–348.Dinkel, H., Van Roey, K., Michael, S., Kumar, M., Uyar, B., Altenberg, B.,Milchevskaya, V., Schneider, M., Ku¨hn, H., Behrendt, A., et al. (2016). ELM2016–data update and new functionality of the eukaryotic linear motifresource. Nucleic Acids Res. 44 (D1), D294–D300.Diril, M.K., Schmidt, S., Krauss, M., Gawlik, V., Joost, H.-G., Schu¨rmann, A.,Haucke, V., and Augustin, R. (2009). Lysosomal localization of GLUT8 in thetestis–the EXXXLL motif of GLUT8 is sufficient for its intracellular sorting viaAP1- and AP2-mediated interaction. FEBS J. 276, 3729–3743.Dittmar, G., Perez-Hernandez, D., Kowenz-Leutz, E., Kirchner, M., Kahlert, G.,Wesolowski, R., Baum, K., Knoblich, M., Muller, A., Wolf, J., et al. (2017).Protein interaction screen on peptide matrix (PRISMA) reveals interactionfootprints and the PTM-dependent interactome of intrinsically disorderedC/EBPb. BioRxiv. https://doi.org/10.1101/238709.Dormann, D., Rodde, R., Edbauer, D., Bentmann, E., Fischer, I., Hruscha, A.,Than, M.E., Mackenzie, I.R.A., Capell, A., Schmid, B., et al. (2010). ALS-asso-ciated fused in sarcoma (FUS) mutations disrupt Transportin-mediated nu-clear import. EMBO J. 29, 2841–2857.Doszta´nyi, Z., Csizmok, V., Tompa, P., and Simon, I. (2005). IUPred: webserver for the prediction of intrinsically unstructured regions of proteins basedon estimated energy content. Bioinformatics 21, 3433–3434.Famiglietti, M.L., Estreicher, A., Gos, A., Bolleman, J., Ge´hant, S., Breuza, L.,Bridge, A., Poux, S., Redaschi, N., Bougueleret, L., and Xenarios, I.; UniProtConsortium (2014). Genetic variations and diseases in UniProtKB/Swiss-Prot: the ins and outs of expert manual curation. Hum. Mutat. 35, 927–935.Finn, R.D., Coggill, P., Eberhardt, R.Y., Eddy, S.R., Mistry, J., Mitchell, A.L.,Potter, S.C., Punta, M., Qureshi, M., Sangrador-Vegas, A., et al. (2016). ThePfam protein families database: towards a more sustainable future. NucleicAcids Res. 44 (D1), D279–D285.Frank, R. (2002). The SPOT-synthesis technique. Synthetic peptide arrayson membrane supports–principles and applications. J. Immunol. Methods267, 13–26.Fuxreiter, M., Tompa, P., and Simon, I. (2007). Local structural disorder im-parts plasticity on linear motifs. Bioinformatics 23, 950–956.Gibson, T.J., Dinkel, H., Van Roey, K., and Diella, F. (2015). Experimentaldetection of short regulatory motifs in eukaryotic proteins: tips for good prac-tice as well as for bad. Cell Commun. Signal. 13, 42.Gingras, A.-C., and Raught, B. (2012). Beyond hairballs: the use of quantitativemass spectrometry data to understand protein-protein interactions. FEBSLett. 586, 2723–2731.Gstaiger, M., and Aebersold, R. (2009). Applying mass spectrometry-basedproteomics to genetics, genomics and network biology. Nat. Rev. Genet. 10,617–627.Herna´ndez-Miranda, L.R., Cariboni, A., Faux, C., Ruhrberg, C., Cho, J.H.,Cloutier, J.-F., Eickholt, B.J., Parnavelas, J.G., and Andrews, W.D. (2011).Robo1 regulates semaphorin signaling to guide the migration of cortical inter-neurons through the ventral forebrain. J. Neurosci. 31, 6174–6187.Hosp, F., Vossfeldt, H., Heinig, M., Vasiljevic, D., Arumughan, A., Wyler, E.,Landthaler, M., Hubner, N., Wanker, E.E., Lannfelt, L., et al.; Genetic andEnvironmental Risk for Alzheimer’s Disease GERAD1 Consortium (2015).Quantitative interaction proteomics of neurodegenerative disease proteins.Cell Rep. 11, 1134–1146.Ishigaki, S., Masuda, A., Fujioka, Y., Iguchi, Y., Katsuno, M., Shibata, A.,Urano, F., Sobue, G., and Ohno, K. (2012). Position-dependent FUS-RNAinteractions regulate alternative splicing events and transcriptions. Sci. Rep.2, 529.Ittner, L.M., and Go¨tz, J. (2007). Pronuclear injection for the production oftransgenic mice. Nat. Protoc. 2, 1206–1215.Kadaveru, K., Vyas, J., and Schiller, M.R. (2008). Viral infection and human dis-ease–insights from minimotifs. Front. Biosci. 13, 6455–6471.Keilhauer, E.C., Hein, M.Y., and Mann, M. (2015). Accurate protein complexretrieval by affinity enrichment mass spectrometry (AE-MS) rather than affinitypurification mass spectrometry (AP-MS). Mol. Cell. Proteomics 14, 120–135.Cell 175, 1–15, September 20, 2018 13Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 Ko¨hler, S., Vasilevsky, N.A., Engelstad, M., Foster, E., McMurry, J., Ayme´, S.,Baynam, G., Bello, S.M., Boerkoel, C.F., Boycott, K.M., et al. (2017). The hu-man phenotype ontology in 2017. Nucleic Acids Res. 45 (D1), D865–D876.Kozik, P., Francis, R.W., Seaman, M.N.J., and Robinson, M.S. (2010). A screenfor endocytic motifs. Traffic 11, 843–855.Kreykenbohm, V., Wenzel, D., Antonin, W., Atlachkine, V., and von Mollard,G.F. (2002). The SNAREs vti1a and vti1b have distinct localization and SNAREcomplex partners. Eur. J. Cell Biol. 81, 273–280.Landrum, M.J., Lee, J.M., Benson, M., Brown, G., Chao, C., Chitipiralla, S., Gu,B., Hart, J., Hoffman, D., Hoover, J., et al. (2016). ClinVar: public archiveof interpretations of clinically relevant variants. Nucleic Acids Res. 44 (D1),D862–D868.Leen, W.G., Klepper, J., Verbeek, M.M., Leferink, M., Hofste, T., van Engelen,B.G., Wevers, R.A., Arthur, T., Bahi-Buisson, N., Ballhausen, D., et al. (2010).Glucose transporter-1 deficiency syndrome: the expanding clinical andgenetic spectrum of a treatable disorder. Brain 133, 655–670.Mann, M. (2006). Functional and quantitative proteomics using SILAC. Nat.Rev. Mol. Cell Biol. 7, 952–958.McLaren, W., Gil, L., Hunt, S.E., Riat, H.S., Ritchie, G.R.S., Thormann, A.,Flicek, P., and Cunningham, F. (2016). The Ensembl variant effect predictor.Genome Biol. 17, 122.McMahon, H.T., and Boucrot, E. (2011). Molecular mechanism and physiolog-ical functions of clathrin-mediated endocytosis. Nat. Rev. Mol. Cell Biol. 12,517–533.Meyer, K., and Selbach, M. (2015). Quantitative affinity purification mass spec-trometry: a versatile technology to study protein-protein interactions. Front.Genet. 6, 237.Narayan, S., Bader, G.D., and Reimand, J. (2016). Frequent mutations in acet-ylation and ubiquitination sites suggest novel driver mechanisms of cancer.Genome Med. 8, 55.Neduva, V., and Russell, R.B. (2005). Linear motifs: evolutionary interactionswitches. FEBS Lett. 579, 3342–3345.Okada, H., Uezu, A., Soderblom, E.J., Moseley, M.A., 3rd, Gertler, F.B., andSoderling, S.H. (2012). Peptide array X-linking (PAX): a new peptide-proteinidentification approach. PLoS ONE 7, e37035.Pandey, K.N. (2009). Functional roles of short sequence motifs in the endo-cytosis of membrane receptors. Front. Biosci. 14, 5339–5360.Park, S.Y., and Guo, X. (2014). Adaptor protein complexes and intracellulartransport. Biosci. Rep. 34, 381–390.Pascual, J.M., Wang, D., Yang, R., Shi, L., Yang, H., and De Vivo, D.C. (2008).Structural signatures and membrane helix 4 in GLUT1: inferences from humanblood-brain glucose transport mutants. J. Biol. Chem. 283, 16732–16742.Qiu, H., Lee, S., Shang, Y., Wang, W.-Y., Au, K.F., Kamiya, S., Barmada, S.J.,Finkbeiner, S., Lui, H., Carlton, C.E., et al. (2014). ALS-associated mutationFUS-R521C causes DNA damage and RNA splicing defects. J. Clin. Invest.124, 981–999.Radivojac, P., Baenziger, P.H., Kann, M.G., Mort, M.E., Hahn, M.W., andMooney, S.D. (2008). Gain and loss of phosphorylation sites in human cancer.Bioinformatics 24, i241–i247.Raiborg, C., and Stenmark, H. (2009). The ESCRT machinery in endosomalsorting of ubiquitylated membrane proteins. Nature 458, 445–452.Rappsilber, J., Ishihama, Y., and Mann, M. (2003). Stop and go extraction tipsfor matrix-assisted laser desorption/ionization, nanoelectrospray, and LC/MSsample pretreatment in proteomics. Anal. Chem. 75, 663–670.Rogelj, B., Easton, L.E., Bogu, G.K., Stanton, L.W., Rot, G., Curk, T., Zupan, B.,Sugimoto, Y., Modic, M., Haberman, N., et al. (2012). Widespread binding ofFUS along nascent RNA regulates alternative splicing in the brain. Sci. Rep.2, 603.Roux, K.J., Kim, D.I., Raida, M., and Burke, B. (2012). A promiscuous biotinligase fusion protein identifies proximal and interacting proteins in mammaliancells. J. Cell Biol. 196, 801–810.Ryan, C.J., Cimermancic, P., Szpiech, Z.A., Sali, A., Hernandez, R.D., and Kro-gan, N.J. (2013). High-resolution network biology: connecting sequence withfunction. Nat. Rev. Genet. 14, 865–879.Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch,T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., et al. (2012). Fiji: anopen-source platform for biological-image analysis. Nat. Methods 9, 676–682.Schulze, W.X., and Mann, M. (2004). A novel proteomic screen for peptide-protein interactions. J. Biol. Chem. 279, 10756–10764.Schwanha¨usser, B., Busse, D., Li, N., Dittmar, G., Schuchhardt, J., Wolf, J.,Chen, W., and Selbach, M. (2011). Global quantification of mammalian geneexpression control. Nature 473, 337–342.Shannon, P., Markiel, A., Ozier, O., Baliga, N.S., Wang, J.T., Ramage, D., Amin,N., Schwikowski, B., and Ideker, T. (2003). Cytoscape: a software environmentfor integrated models of biomolecular interaction networks. Genome Res. 13,2498–2504.Shi, J., and Kandror, K.V. (2008). Study of glucose uptake in adipose cells.Methods Mol. Biol. 456, 307–315.Silvis, M.R., Picciano, J.A., Bertrand, C., Weixel, K., Bridges, R.J., andBradbury, N.A. (2003). A mutation in the cystic fibrosis transmembraneconductance regulator generates a novel internalization sequence and en-hances endocytic rates. J. Biol. Chem. 278, 11554–11560.Slaughter, L., Vartzelis, G., and Arthur, T. (2009). New GLUT-1 mutation in achild with treatment-resistant epilepsy. Epilepsy Res. 84, 254–256.Smits, A.H., and Vermeulen, M. (2016). Characterizing protein-protein interac-tions using mass spectrometry: challenges and opportunities. Trends Bio-technol. 34, 825–834.Staudt, C., Puissant, E., and Boonen, M. (2017). Subcellular trafficking ofmammalian lysosomal proteins: an extended view. Int. J. Mol. Sci. 18, 47.Subramanian, S., and Kumar, S. (2006). Evolutionary anatomies of positionsand types of disease-associated and neutral amino acid mutations in the hu-man genome. BMC Genomics 7, 306.Tompa, P., Davey, N.E., Gibson, T.J., and Babu, M.M. (2014). A million peptidemotifs for the molecular biologist. Mol. Cell 55, 161–169.Toprak, U.H., Gillet, L.C., Maiolica, A., Navarro, P., Leitner, A., and Aebersold,R. (2014). Conserved peptide fragmentation as a benchmarking tool for massspectrometers and a discriminating feature for targeted proteomics. Mol. Cell.Proteomics 13, 2056–2071.Traub, L.M. (2009). Tickets to ride: selecting cargo for clathrin-regulated inter-nalization. Nat. Rev. Mol. Cell Biol. 10, 583–596.Traub, L.M., and Bonifacino, J.S. (2013). Cargo recognition in clathrin-medi-ated endocytosis. Cold Spring Harb. Perspect. Biol. 5, a016790.Tripathi, S., Pohl, M.O., Zhou, Y., Rodriguez-Frandsen, A., Wang, G., Stein,D.A., Moulton, H.M., DeJesus, P., Che, J., Mulder, L.C.F., et al. (2015).Meta- and orthogonal integration of influenza ‘‘OMICs’’ data defines a rolefor UBR4 in virus budding. Cell Host Microbe 18, 723–735.UniProt Consortium (2012). Reorganizing the protein space at the UniversalProtein Resource (UniProt). Nucleic Acids Res. 40, D71–D75.Uversky, V.N., Oldfield, C.J., and Dunker, A.K. (2008). Intrinsically disorderedproteins in human diseases: introducing the D2 concept. Annu. Rev. Biophys.37, 215–246.Uyar, B., Weatheritt, R.J., Dinkel, H., Davey, N.E., and Gibson, T.J. (2014).Proteome-wide analysis of human disease mutations in short linear motifs:neglected players in cancer? Mol. Biosyst. 10, 2626–2642.Vacic, V., Markwick, P.R.L., Oldfield, C.J., Zhao, X., Haynes, C., Uversky, V.N.,and Iakoucheva, L.M. (2012). Disease-associated mutations disrupt function-ally important regions of intrinsic protein disorder. PLoS Comput. Biol. 8,e1002709.Van Roey, K., Uyar, B., Weatheritt, R.J., Dinkel, H., Seiler, M., Budd, A.,Gibson, T.J., and Davey, N.E. (2014). Short linear motifs: ubiquitous and func-tionally diverse protein interaction modules directing cell regulation. Chem.Rev. 114, 6733–6778.14 Cell 175, 1–15, September 20, 2018Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 Vizcaı´no, J.A., Csordas, A., Del-Toro, N., Dianes, J.A., Griss, J., Lavidas, I.,Mayer, G., Perez-Riverol, Y., Reisinger, F., Ternent, T., et al. (2016). 2016 up-date of the PRIDE database and its related tools. Nucleic Acids Res. 44, 11033.Vogt, G., Chapgier, A., Yang, K., Chuzhanova, N., Feinberg, J., Fieschi, C.,Boisson-Dupuis, S., Alcais, A., Filipe-Santos, O., Bustamante, J., et al.(2005). Gains of glycosylation comprise an unexpectedly large group of path-ogenic mutations. Nat. Genet. 37, 692–700.Wang, P.I., and Marcotte, E.M. (2010). It’s the machine that matters: predictinggene function and phenotype from protein networks. J. Proteomics 73,2277–2289.Wang, X., Wei, X., Thijssen, B., Das, J., Lipkin, S.M., and Yu, H. (2012). Three-dimensional reconstruction of protein networks provides insight into humangenetic disease. Nat. Biotechnol. 30, 159–164.Wefers, B., Bashir, S., Rossius, J., Wurst, W., and Ku¨hn, R. (2017). Geneediting in mouse zygotes using the CRISPR/Cas9 system. Methods 121-122, 55–67.Wilkinson, L. (2011). ggplot2: elegant graphics for data analysis by WICKHAM.H. Biometrics 67, 678–679.Wright, P.E., and Dyson, H.J. (2015). Intrinsically disordered proteins in cellularsignalling and regulation. Nat. Rev. Mol. Cell Biol. 16, 18–29.Yang, L., Embree, L.J., Tsai, S., and Hickstein, D.D. (1998). Oncoprotein TLSinteracts with serine-arginine proteins involved in RNA splicing. J. Biol.Chem. 273, 27761–27764.Yue, P., Li, Z., and Moult, J. (2005). Loss of protein structure stability as a majorcausative factor in monogenic disease. J. Mol. Biol. 353, 459–473.Zauber, H., Kirchner, M., and Selbach, M. (2018). Picky: a simple online PRMand SRM method designer for targeted proteomics. Nat. Methods 15,156–157.Zhong, Q., Simonis, N., Li, Q.-R., Charloteaux, B., Heuze, F., Klitgord, N., Tam,S., Yu, H., Venkatesan, K., Mou, D., et al. (2009). Edgetic perturbation modelsof human inherited disorders. Mol. Syst. Biol. 5, 321.Cell 175, 1–15, September 20, 2018 15Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 STAR+METHODSKEY RESOURCES TABLEREAGENT or RESOURCE SOURCE IDENTIFIERAntibodiesMouse monoclonal anti-FLAG Sigma-Aldrich Cat# F3165; RRID:AB_259529Rabbit polyclonal anti-GLUT1 Merck Millipore Cat# 07-1401; RRID:AB_1587074Rat monoclonal anti-ICAM2 BD Biosciences Cat# 553326; RRID:AB_394784Isolectin GS-IB4 Alexa Fluor 488 conjugate (IB4) Thermo Fisher Scientific Cat# I21411; RRID:AB_2314662Mouse monoclonal anti-Vti1a BD Biosciences Cat# 611220; RRID:AB_398752Mouse monoclonal anti-Vti1b BD Biosciences Cat# 611405; RRID:AB_398927Rabbit polyclonal anti-EEA1 Cell Signaling Technology Cat# 2411S; RRID:AB_2096814Rabbit monoclonal anti-Rab4 Abcam Cat# ab13252; RRID:AB_2269374Rabbit monoclonal anti-Rab9 Cell Signaling Technology Cat# 5118S; RRID:AB_10621426Rabbit monoclonal anti-LAMP1 Cell Signaling Technology Cat# 9091; RRID:AB_2687579Mouse monoclonal anti-gamma Adaptin (AP-1 g) BD Biosciences Cat# 610385; RRID:AB_397768Mouse monoclonal anti-AP50 (AP-2 m) BD Biosciences Cat# 611351; RRID:AB_398873Mouse monoclonal anti- alpha Adaptin (AP-2 a) Thermo Fisher Scientific Cat# MA3-061; RRID:AB_2056321Mouse monoclonal anti-alpha Adaptin (AP-2 a) Abcam Cat# ab2730; RRID:AB_303255Mouse monoclonal anti - IL-2 R alpha (TAC) Santa Cruz Biotechnology Cat# sc-65258; RRID:AB_631112Chemicals, Peptides, and Recombinant ProteinsDeoxy-D-glucose, 2-[1,2-3H (N)]-, Specific Activity:5-10 Ci (185-370 GBq)/mmol, 250 mCi (9.25 MBq)Perkin Elmer NET328250UCL-arginine-HCl (Arg0) Sigma-Aldrich A6969; CAS: 1119-34-2L-arginine-HCl(13C6) (Arg6) Sigma-Aldrich 643440; CAS: 201740-91-2L-arginine-HCl(13C6,15N4) (Arg10) Sigma-Aldrich 608033; CAS: 202468-25-5L-lysine-HCl (Lys0) Sigma-Aldrich L8662; CAS: 657-27-2L-lysine-2HCl(4,4,5,5-D4) (Lys4) Cambridge Isotope Laboratories DLM-2640-PK; CAS: 657-26-1L-lysine-HCl(13C6,15N2) (Lys8) Silantes 211604102Cytochalasin B Sigma-Aldrich C2743; CAS: 14930-96-2Deposited DataPeptide-protein interaction screen dataset This paper PXD010027PRM dataset This paper PXD010005BioID This paper PXD010061Experimental Models: Cell LinesFlp-In T-Rex GLUT1 This paper N/AFlp-In T-Rex GLUT1_P485L This paper N/AHuman: Patient-derived iPSCs This paper https://hpscreg.eu/: BIHi037-(A-E)Experimental Models: Organisms/StrainsMouse: C57BL/6N: GLUT1_P485L This paper N/AOligonucleotidesON-TARGETplus Human AP2M1 (1173)siRNA - SMARTpoolDharmacon Cat# L-008170-00-0005ON-TARGETplus Non-targeting Pool Dharmacon Cat# D-001810-10-05Primers TAC chimera constructs This paper See Table S6Recombinant DNASLC2A1 (GLUT1) Harvard Plasmid Repository HsCD00378964(Continued on next page)e1 Cell 175, 1–15.e1–e9, September 20, 2018Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 CONTACT FOR REAGENT AND RESOURCE SHARINGFurther information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, MatthiasSelbach (matthias.selbach@mdc-berlin.de).EXPERIMENTAL MODEL AND SUBJECT DETAILSCell linesSH-SY5Y and T-REx-293 cells were cultured under standard cell culture conditions. In brief, cells were cultured in DMEM (LifeTechnologies) complemented with 10% fetal calf serum (Pan-Biotech).Cells used for SILAC based experiments were cultured in SILAC DMEM (Life Technologies) complemented with glutamine(Glutamax, Life Technologies), Pyruvate (Life Technologies), non-essential amino acids (Life Technologies) and 10% dialyzed fetalcalf serum (Pan-Biotech). The SILAC DMEM was supplemented with standard L-arginine (Arg0, Sigma-Aldrich) and L-lysine(Lys0, Sigma-Aldrich) (‘‘light’’) as in Schwanha¨usser et al. (2011). Alternatively, Arg6 and Lys4 (‘‘medium-heavy’’) or Arg10 andLys8 (‘‘heavy’’) were added in place of their light counterparts. Cells were cultured at 37C and 5% CO2.Flp-In T-Rex GLUT1We purchased SLC2A1 (GLUT1) from Harvard Plasmid repository (HsCD00378964). A stop codon has been added to the gene withthe following primers Fw:TCCCAAGTGTAATTGCCAACTTTCTTGTACAAAGTTG, Rev:ATCAGCCCCCAGGGGATG.P485L Mutation has been introduced by changing c.1454 C T (Slaughter et al., 2009) with Q5Site-Directed Mutagenesis Kit(NEB) Fw:CTGTTCCATCtCCTGGGGGCT, Rev:CTCCTCGGGTGTCTTGTCAC.SLC2A1 and SLC2A1 mutant have been further cloned into a destination vector with a N-terminal BirA-FLAG Tag (pDEST-pcDNA5-BirA-FLAG N-term (Couzens et al., 2013)) with Gateway cloning strategy (Thermo Fisher Scientific). HEK293 Flp-InT-Rex cells (Invitrogen) that exhibit tetracycline-inducible expression of BirA-FLAG-GLUT1 or BirA-FLAG-GLUT1_P485L were gener-ated using the Flp-In system developed by Life Technologies according to the manufacturer’s protocol.Patient-derived iPSCsFibroblasts were obtained from a GLUT1 deficient patient with the P485L mutation. The voluntary informed consentprocess was documented in writing as approved in advance by the University of Texas Southwestern Medical Center Institu-tional Review Board. This included information regarding the de-identification of the sample and the adherence to HIPAAregulations.A 4 mm single-use, sterile skin punch was applied to the lateral surface of the left shoulder after the skin had been cleansed withiodine solution in aseptic fashion followed by injection of 0.5 mL of 1% unbuffered lidocaine with a vasoconstrictor. Prilocaine andlidocaine cream had been previously applied to the area. The punch was advanced by rotation under pressure and the explant wassevered from its base and harvested in culture medium containing complete DMEM plus 20% fetal bovine serum and placed on iceuntil the explant was divided for culture the same day. The explant was divided into 12-15 evenly sized pieces and each piecemaintained in a 10 cm dish at 37C until fibroblast confluence was reached. The cells were then treated with trypsin and passagedinto a T-25 flask for further expansion. Fibroblasts were grown to approximately 50% confluence in the T-25 flask. They were thensuspended with trypsin and frozen over dry ice in complete DMEM medium with 10% DMSO at a density of 106cells/mL per vial priorto storage and shipment on dry ice.The patient fibroblast were reprogrammed using the mRNA reprogramming kit ReproRNA-OKSGM from Stem Cell Technologiesaccording to the instructions. In brief, 1x 105fibroblast cells were plated on Geltrex coated 6-well plate using regular DMEM mediawith 10% FBS. The day after the cells were transfected with the ReproRNA- OKSGM construct using the ReproRNA transfectionreagents and growth Media with B18R. The next 5 days the growth media was changed every day and supplemented with B18Rand 0.8 mg/mL Puromycin. After 8 days the growth media was exchanged by ReproTeSR and first colonies appeared after14 days. In total 5 clones were picked and established using mTESR-1 media. As a control for the experiments the following fibro-blasts (NHDF-Ad-Der Fibroblasts, C-2511, LONZA) were reprogrammed using the Epi5 Episomal iPSC Reprogramming Kit fromContinuedREAGENT or RESOURCE SOURCE IDENTIFIERSoftware and AlgorithmsImaris v8.4.1 Bitplane N/AMaxQuant v1.5.2.8 http://www.biochem.mpg.de/5111795/maxquant Cox and Mann, 2008Cell 175, 1–15.e1–e9, September 20, 2018 e2Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 Thermo Fisher Scientific following the vendor’s instructions. The established lines and clones were registered and named using theHuman pluripotent stem cell registry (https://hpscreg.eu/): BIHi037-(A-E).The iPSCs used for the experiments were characterized using the PSC 4-Marker immunocytochemistry kit from Thermo FisherScientific following the instructions of the protocol. In addition to the 4 markers (OCT4, SOX2, TRA-1-60 and SSEA4) included inthe kit, the expression of another pluripotency marker NANOG (Nanog PA1-097, Thermo Fisher Scientific) was analyzed.Human iPSC cultures were maintained on plates coated with hESC-Qualified Matrigel (Corning) in mTESR-1 medium (Stem CellTechnologies) following the manufacturer’s instruction. All cells were cultured at 37C in humidified atmosphere containing 5% O2and 5% CO2. Cells were passaged using StemPro Accutase (Thermo Fisher) and replated in mTESR-1 medium with the addition of10 mM ROCK inhibitor Y-27632 (LC Laboratories).Animal modelGLUT1 P485L mice were produced by microinjection of C57BL/6N zygotes with Cas9 protein (IDT), synthetic guide RNA (IDT)(50GAGGAGCTCTTCCACCCTCT30) and a mutagenic single stranded deoxyoligonucleotide (IDT) (50TAGCTGCCTGTGCTCCAGAGAGATCCTTGGGCTGCAGGGAGCAGGCCGGGCTGGGTGTGGGGCTCCTCACACTTGGGAGTCCGCCCCCAacaaGTGGAAGAGCTCCTCGGGTGTCTTGTCACTTTGG30) as recombination template, as described (Wefers et al., 2017). Reagents were diluted inmicroinjection buffer (10 mM Tris, 0.1 mM EDTA, pH 7.2), filtrated through a centrifugal filter (Millipore, UFC30LG25) and stored insingle use aliquots at 80C. For microinjections, zygotes were obtained by mating of C57BL/6N males with superovulatedC57BL/6N females (Charles River, Sulzbach, Germany). Zygotes were injected into one pronucleus following standard procedures(Ittner and Go¨tz, 2007). Injected zygotes were transferred into pseudo-pregnant NMRI female mice to obtain live pups. All miceshowed normal development and appeared healthy. Mice were handled according to institutional guidelines under experimentationlicense no. G0162/12 approved by the Landesamt fu¨r Gesundheit und Soziales (Berlin, Germany) and housed in standard cages in aspecific pathogen-free facility on a 12 h light/dark cycle with ad libitum access to food and water.METHOD DETAILSPeptide-protein interaction screenCandidate selectionDisease mutations in humans were taken from UniProt annotations (UniProt Consortium, 2012) of Online Mendelian Inheritance inMan, OMIM, McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University (Baltimore, MD), https://omim.org/.This dataset consists of experimentally validated missense mutations that contribute to inherited diseases. Inherited disease muta-tions were downloaded from UniProt (https://www.uniprot.org/docs/humsavar.txt, Release: 2015_07 of 24-Jun-2015) (Famigliettiet al., 2014). Only mutations that were associated to ‘Disease’ were kept. ‘Unclassified’ mutations or ‘Polymorphisms’ wereexcluded. The 26,649 disease mutations were further filtered by applying a disorder cut-off. Disorder tendencies of 15 amino acids(AAs) long peptides, with the AA mutated in disease if possible located at position eight, were predicted using IUPred (Doszta´nyi et al.,2005) using the ‘SHORT’ profile considering sequential neighborhood of 25 residues. IUPred disorder scores above 0.5 denoteregions of the proteins that have 95% likelihood to be disordered. For filtering, the mean disorder score for all 15 AA as well asthe mutation position were required to be 0.5. This resulted in 1,878 disease mutations in disordered regions. Next we assigneddisease classes to 3,119 different diseases included in the humsavar database by combining a manual approach with automaticannotation with the Human Phenotype Ontology database, HPO (Ko¨hler et al., 2017). We selected 305 mutations causing neurolog-ical diseases. After manual inspection, we remained with 128 mutations causing 124 distinct neurological diseases that were used forthe peptide-protein interaction screen.Experimental setupPeptides of 15 AAs, in total 128 wild-type peptides and 128 related peptides containing the disease causing mutation (256 peptides)plus one control peptide pair were synthesized in situ on cellulose membrane using PepTrack techniques (JPT Peptide Technologies,Berlin, Germany). Two of those peptide filters were moistened in cell lysis buffer [50 mM HEPES pH 7.6 at 4C, 150 mM NaCl, 1 mMEGTA, 1 mM MgCl2, 10% Glycerol, 0.5% Nonidet P-40, 0.05% SDS and 0.25% sodium deoxycholate, supplemented with proteaseinhibitor (Roche) and benzonase (Merck)]. In order to reduce nonspecific binding the membrane was incubated with 1 mg/mL yeastt-RNA (Invitrogen) for 10 min and then washed twice with cell lysis buffer without detergents. The entire peptide libraries wereincubated with 15 mL of light or heavy SILAC labeled cell lysate (5 mg/mL) from SH-SY5Y cells for 2 h. Membranes were washedthree times and air-dried.Sample preparation for mass spectrometric analysisSingle spots were punched out from cellulose membrane with a 2 mm diameter ear punch (Carl Roth) and SILAC pairs were placedtogether in a 96-well plate (Thermo Fisher Scientific) prepared with 30 ml of denaturation buffer [6 M urea (Sigma-Aldrich), 2 M thiourea(Sigma-Aldrich), 10 mM HEPES, pH 8]. Samples were reduced by incubating with 10 ml of 3.3 mM DTT (Sigma-Aldrich) for 30 minat RT, followed by an alkylation step using 10 ml of 18.3 mM iodoacetamide (IAA) (Sigma-Aldrich) for 60 min at RT. The sampleswere first digested using 1 mg endopeptidase LysC (Wako, Osaka, Japan) for 4 h. The samples were diluted by adding 100 mlofe3 Cell 175, 1–15.e1–e9, September 20, 2018Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 50 mM ammonium bicarbonate (pH = 8.5), and finally digested with 1 mg trypsin (Promega) for 16 h. The digestion was stopped byacidifying each sample to pH 2.5 by adding 10% trifluoroacetic acid solution. The peptide extracts were purified and stored onstage tips according to Rappsilber et al. (2003).LC-MS/MS analysisPeptides were eluted using Buffer B (80% Acetonitrile and 0.1% formic acid) and organic solvent was evaporated using a speedvac(Eppendorf). Samples were diluted in Buffer A (5% acetonitrile and 0.1% formic acid). Peptides were separated on a reversed-phasecolumn with 45 min gradient with a 250 nl/min flow rate of increasing Buffer B concentration on a High Performance Liquid Chroma-tography (HPLC) system (Thermo Fisher Scientific). Peptides were ionized using an electrospray ionization (ESI) source (ThermoFisher Scientific) and analyzed on a Q-exactive plus Orbitrap instrument (Thermo FisherScientific). Dynamic exclusion for selectedprecursor ions was 30 s. The mass spectrometer was run in data dependent mode selecting the top 10 most intense ions in the MS fullscans, selecting ions from 300 to 1700 m/z (Orbitrap resolution: 70,000; target value: 1,000,000 ions; maximum injection time of120 ms). The resulting MS/MS spectra from the Orbitrap had a resolution of 17,500 after a maximum ion collection time of 60 mswith a target of reaching 100,000 ions.Data analysisThe resulting raw files were analyzed using MaxQuant software version 1.5.2.8 (Cox and Mann, 2008). Default settings were keptexcept that ‘match between runs’ and ‘re-quantify’ was turned on. Lys8 and Arg10 were set as labels and oxidation of methioninesand N-terminal acetylation were defined as variable modifications. Carbamidomethyl of cysteines was set as fixed modification. Thein silico digests of the human Uniprot database (2015-12), a FASTA file containing all peptides used for pull-down and a databasecontaining common contaminants were done with Trypsin/P. The false discovery rate was set to 1% at both the peptide and proteinlevel and was assessed by in parallel searching a database containing the reversed sequences from the Uniprot database. Followingstatistics and figures were done using R (R version 3.2.1, RStudio Version 1.0.143). The resulting text files were filtered to excludereverse database hits, potential contaminants, and proteins only identified by site. We imputed missing LFQ-intensity values withrandom noise simulating the detection limit of the mass spectrometer (Keilhauer et al., 2015). To this end, imputed values are takenfrom a log normal distribution with 0.25 3the standard deviation of the measured, logarithmized values, down-shifted by 1.8 stan-dard deviations. In this way, we obtained a distribution of quantitative values for each protein across samples. For determination ofspecific interactions, two replicated pull-downs for the same peptide were tested against all other pull-downs, excluding the corre-sponding variant peptide, by the nonparametric Mann–Whitney U test. Resulting p values and fold-changes (log2 space) have beenplotted as volcano plots to determine cut-offs. We used an approach that uses a graphical formula to combination a fold-change andp value cut-off (Keilhauer et al., 2015): log10ðpÞRcjxjx0with x: enrichment factor of a protein, p: p value of the Mann–Whitney U testcalculated from replicates, x0: fixed minimum enrichment, c: curvature parameter. The curvature parameter c determines themaximum acceptable p value for a given enrichment x.The parameters c and x0can be optimized based on prior knowledge of known true and false positives (Keilhauer et al., 2015;Schulze and Mann, 2004). Here, cut-offs were chosen according to known interaction partners of the SOS1 control peptide (Keilhaueret al., 2015). This resulted in x0=0,c=8.This cut-off was applied to all other pull-downs to separate specific binders from background. SILAC ratios were normalized bysubtracting the median SILAC ratio of every experiment from all SILAC ratios in that experiment. To define interaction partnersthat bind differentially to wild-type and mutant peptide, a SILAC cut-off has been defined. For wild-type specific interaction partners,the mean log2 SILAC ratio of the two replicates needed to be 1 and none of the two ratios 0 (mutant specific mean log2 SILACratio 1 and none of the two ratios 0). Resulting figures were modified in Inkscape (0.91).PRMExperimental procedure was identical to general peptide-protein interaction screen. Only peptide variants from GLUT1_P485L,ITPR1_P1059L, CACNA1H_P648L and CACNA1H_A748V (control peptide) have been used for experiment.LC-MS/MS analysisPeptides were separated by reverse phase chromatography on an effective 150 min gradient (0, 2, 100, 30, 15, 1 and 5 min with 2, 4,20, 30, 60, 90 and 90% of buffer B with 90% acetonitrile) and analyzed on a Q-Exactive HFx (Thermo Fisher Scientific). The PRMsettings were: 30,000 resolution; 5e5 AGC target; 1.6 m/z isolation window; 60 ms max ion injection time. The inclusion list for thePRM method was generated using Picky (Zauber et al., 2018) with SILAC option enabled and a retention time window of 30 min.Predicted retention-times were calibrated in Picky with a complex sample of 100 ng Pierce HeLa Protein standard (Thermo FisherScientific) immediately before the PRM measurements.Analysis of PRM dataTraces of all fragments from precursors in the spectral library (as exported from picky) were extracted from all rawfiles using theThermo MSFileReader and the MSFileReader.py bindings written by Francois Allen. For each light or heavy scan the normalizedspectral contrast angle (SCA) was calculated (Toprak et al., 2014). Peaks were manually selected and required a SCA 0.4 and Frag-ment Matches 4 in the light or heavy channel. Further Peaks needed to be within a similar retention time range across all differentmeasurements. Ratios for each fragment using the maximum intensity of each peak were calculated. The median log2 transformedCell 175, 1–15.e1–e9, September 20, 2018 e4Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 ratio (log2FC) for each peptide in each raw-file was calculated from selected fragment ratios: The five highest abundant fragmentswere selected from the peak with the highest detected SCA. Peptide log2FC were plotted as boxplot distributions in a protein centricmanner across the different experiments (Figure S4A).BioIDMedium-heavy and heavy labeled T-REx-293 cells have been induced for 24 h with 0.1 mg/mL Doxycycline to induce expressionof GLUT1 (wild-type, wt) or GLUT1_P485L (mutant, mut). Light labeled cell lines from both GLUT1 and GLUT1_P485L havebeen left uninduced and served as a control for background binding. SILAC labeling allowed for quantitative comparison ofproteins that have been proximity labeled by the transiently expressed constructs (Forward experiment, B_1: Light - Control,Medium-heavy - wt, Heavy - mut; Label swap experiment, B_2: Light - Control, Medium-heavy - mut, Heavy - wt). During theinduction period all cell lines have been incubated for 24 h in cell culture medium containing biotin. BioID experiment has beenperformed essentially as in Couzens et al. (2013), with minor adaptations.Mass spec setup and analysis was done similarly as to samples from peptide pull-downs, but on bead digested peptides wereseparated on a 2,000 mm monolithic column with a 100 mm inner diameter filled with C18 material that was kindly provided by YasushiIshihama (Kyoto University) using a 4 h linear gradient with a 300 nl/min flow rate of increasing Buffer B concentration on a HighPerformance Liquid Chromatography (HPLC) system (Thermo Fisher Scientific). The resulting raw files were analyzed usingMaxQuant software version 1.5.2.8 (Cox and Mann, 2008). Default settings were kept except that ‘match between runs’ and‘re-quantify’ was turned on. Lys4 and Arg6 or Lys8 and Arg10 were set as labels and oxidation of methionines and N-terminalacetylation were defined as variable modifications. Carbamidomethylation of cysteines was set as fixed modification. The in silicodigests of the human Uniprot database (2015-12), a fasta file containing the sequence of BirA-FLAG-GLUT1 and a database contain-ing common contaminants were done with Trypsin/P. The false discovery rate was set to 1% at both the peptide and protein level andwas assessed by in parallel searching a database containing the reversed sequences from the Uniprot database.Biotinylated proteins with a wild-type to mutant enrichment ratio (log2FC) 1 or 1 have been considered as significant. Theseproteins have been analyzed for gene ontology enrichment of cellular components with http://metascape.org (Tripathi et al., 2015).FLAG-GLUT1 localizationHEK293 Flp-In T-Rex cells with BirA-FLAG-GLUT1 or BirA-FLAG-GLUT1_P485L have been seeded on coverslips coated withpoly-L-lysine (Sigma-Aldrich). After induction for 24 h in doxycycline (0.1 mg/mL) containing media, cells were fixed with 4% PFA(paraformaldehyde). Standard procedures were used for immunostaining. Cells have been stained against FLAG 1:200 (F1804,Sigma-Aldrich). Nucleus has been stained with DAPI (Sigma-Aldrich). FLAG staining was accompanied by staining to one of thefollowing endosomal markers and with the following dilutions: anti-EEA1 (Cell Signaling Technology, 1:100); anti-Rab4 (Cell SignalingTechnology, 1:100); anti-Rab9 (Cell Signaling Technology, 1:100); anti-LAMP1 (Cell Signaling Technology, 1:100). Mouse anti-FLAGstaining was substituted by rabbit anti-GLUT1 (Merck Millipore, 1:500) to costain mouse monoclonal antibodies: anti-VTI1A (BDBiosciences, 1:100); anti-VTI1B (BD Biosciences, 1:100). Secondary antibodies all come from Invitrogen. For colocalization analysisthree z stacks of 5-10 cells each have been quantified for each marker with Imaris v8.4.1 (see ‘‘QUANTIFICATION AND STATISTICALANALYSIS’’ for details).Transferrin uptakeEssentially as in ‘‘FLAG-GLUT1 localization.’’ Additionally, after 24 h cells were serum-starved for 1 h and used for Transferrin (Tf)uptake. For Tf uptake, cells were treated with 10 mgmL1Tf-Alexa568 (Life Technologies) for 10 min at 37C. For colocalization anal-ysis three z stacks of more than 15 cells each have been quantified with Imaris v8.4.1 (see ‘‘QUANTIFICATION AND STATISTICALANALYSIS’’ for details).FLAG-GLUT1 localization under AP-2 mknockdownTo rescue the GLUT1_P485L phenotype, clathrin mediated endocytosis (CME) has been inhibited by knocking down AP-2 mandhence the adaptor complex responsible for recognition of cargo for CME.On day 1, cells were seeded in 6-well plates. On day 2, cells were transfected with 25 nM final siRNA concentration (AP-2 m:ON-TARGETplus Human AP2M1 (Dharmacon) and non-target: ON-TARGETplus Non-targeting Pool (Dharmacon)) according toDharmaFECT (Dharmacon) transfection protocol. 24 h after the transfection, medium was replaced with complete medium to reducecytotoxicity and incubated for another 24 h. On day 4, siRNA transfection was repeated as described for day 2. On day 5, cells havebeen seeded in a 24-well plate onto coverslips coated with poly-L-lysine (Sigma-Aldrich) for microscopy and into a 6-well plate forwestern blot analysis. Doxycycline (0.1 mg/mL) has been added to the medium to induce expression of the GLUT1 constructs. Afterinduction for 48 h, cells in 24-well plates were fixed with 4% PFA. Standard procedures were used for immunostaining. Cells werestained with rabbit polyclonal GLUT1 antibody 1:200 (Merck Millipore) and co-stained with mouse monoclonal anti-alpha adaptinantibody [AP6] 1:200 (Abcam). Secondary antibodies with Alexa fluorophores have all been purchased from Invitrogen. Nucleushas been stained with DAPI (Sigma-Aldrich).Lysate from cells in 6-well plates has been used for western blotting, aand m2 subunits of AP-2 were detected using mouse mono-clonal antibodies from Thermo Fisher Scientific and BD transduction, respectively. Profilin 1 was stained as a loading control withe5 Cell 175, 1–15.e1–e9, September 20, 2018Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 polyclonal rabbit antibody from CST. Horseradish peroxidase coupled secondary antibodies were purchased from GE Healthcare.Proteins were detected with chemiluminescence substrate (Perkin Elmer) on a ChemiDoc MP Imaging System (Bio-Rad) andquantified with Image Lab 5.2.1.Antibody feeding assayAn antibody feeding assay was used to study the gain of endocytosis by gain of dileucine motifs. For antibody internalization assay,genes and cytoplasmic regions were chosen according to the following criteria: All 11 disease mutations from Humsavar and Clinvar(‘Pathogenic’ or ‘Conflicting interpretations of pathogenicity’, in case ‘pathogenic’ or likely pathogenic was included in the differentinterpretations) that lead to a gain of a dileucine motif have been considered. All regions ±7 AAs of the mutation have been analyzedaccording to Eukaryotic Linear Motif (ELM) database (Dinkel et al., 2016). CACNA1H_P618L and RET_P1039L were not consideredfor the assay since wild-type variants of the peptides already harbor trafficking motifs. For GLUT1_P485L the whole cytoplasmic Cterminus was amplified via PCR adding EcoRV 50and NotI 30. All other seven constructs were generated by inserting the region sur-rounding the mutation position with Q5Site-Directed Mutagenesis Kit (NEB) resulting in a 15 AA insert (we were not able to generatea construct for CACNA1H_P648L). Chimeras consisting of one of the cytoplasmic regions and the human TAC antigen (interleukin-2receptor achain, CD25) were constructed based on a TAC construct (Diril et al., 2009). HeLa cells were transiently transfected withthe TAC chimera constructs using jetPRIME (Polyplus-transfection). Two days after transfection, cells were labeled with anti-TAC IgG(Santa Cruz Biotechnology) (1: 1000 diluted in Opti-MEM; Invitrogen) for 30 min at 4C. After one change of medium (to Opti-MEM at37C), plasma membrane antigens were allowed to internalize for 30 min at 37C. The cells were then fixed with 4% paraformalde-hyde (Sigma-Aldrich) for 10 min on ice, and surface-bound TAC antibody was blocked using goat anti-mouse serum [goat anti-mouseIgG (Thermo Fisher Scientific at a 1: 5 dilution in goat serum dilution buffer, consisting of 30% normal goat serum (Sigma-Aldrich),450 mm NaCl in 20mM sodium phosphate buffer pH 7.4] for 2 h at room temperature. Cells were permeabilized and blocked with goatserum dilution buffer containing 0.2% saponin for 10 min. For detection of internalized TAC antibody, a goat anti-mouse Alexa Fluor488-conjugated IgG (Invitrogen) was added for 1 h. Cells were then washed three times for 10 min each with sodium phosphate buffercontaining 0.02% saponin. For total TAC staining, the specimens were incubated for 1 h with TAC antibody diluted 1: 1000. Assecondary antibody, an Alexa Fluor 594-conjugated goat anti-mouse IgG (Invitrogen) was added for 30 min, and nuclei were stainedusing DAPI (Sigma-Aldrich). Cells were washed, and coverslips were mounted in ProLongGold antifade reagent (Invitrogen). Forimaging, cells with positive signal in the 594 channel were chosen. All cells in Figure 7D are positive for total TAC staining (594). Wehave seen that the level of internalised TAC chimera (488) does not correlate with the amount of total TAC staining (594), betweensamples and in the same sample, and hence we have decided to exclude this channel from visualization.GLUT1 localization in iPSCsHuman iPSCs were seeded on coverslips coated with hESC-Qualified Matrigel (Corning). Cells were fixed with 4% PFA, stainedwith rabbit polyclonal GLUT1 antibody 1:200 (Merck Millipore) and costained with mouse monoclonal VTI1a antibody 1:100 (BDBiosciences). Secondary antibodies with Alexa fluorophores have all been purchased from Invitrogen.Fluorescence microscopy from cell cultureImages from FLAG-GLUT1 localization were acquired by Leica DMI6600 confocal laser scanning microscope with an HCX PL APO63.0/1.40 oil objective. Transferrin uptake, GLUT1-localization under AP-2 mknockdown, antibody feeding assay and GLUT1 iniPSCs were acquired by a Zeiss LSM 700 confocal laser scanning microscope with an EC Plan-Neofluar/NA1.3 40x oil objectiveor a EC Plan-Apochromat/NA1.4 63x oil objective. Images were further processed with Fiji (Schindelin et al., 2012). For colocalizationanalysis see ‘‘QUANTIFICATION AND STATISTICAL ANALYSIS.’’Immunofluorescence in mouse tissueE14-E15.5 embryos were obtained by Caesarian section from pregnant dam on day 14-15.5 post-coitus. Whole-mount embryoswere dissected in ice-cold phosphate buffer and fixed for 2 h with a solution of 4% PFA in ice-cold phosphate buffer and cryopro-tected overnight in 30% sucrose in phosphate buffer at 4C. Whole embryo heads were sectioned in a horizontal plane using a cryo-stat to obtain 12-16 mm sections.Sample preparation for confocal microscopyEssentially as in Herna´ndez-Miranda et al. (2011). Brain sections were incubated in blocking buffer 1 (5% horse serum and 0.1%Triton X-100 made in phosphate buffer) for 1 h at room temperature. Then, sections were incubated overnight in blocking buffer 1containing the following antibodies: rabbit anti-GLUT1 (1:200; Merck Millipore #07-1401), rat anti-ICAM2 (1:100; BD Biosciences#553326) and Isolectin GS-IB4 Alexa Fluor 488 conjugate (1:100, Thermo Fisher Scientific #I21411) at 4C. Next, sections werewashed three times in ice-cold phosphate buffer and incubated for 3 h in blocking buffer 1 containing Cy3 horse anti-rabbit(1:500; Jackson Lab), Cy5 horse anti-rat (1:500; Jackson Lab) and DAPI at room temperature. Fluorescence was imaged on a ZeissLSM 700 (Jena, Germany) confocal microscope in a non-blind manner.Sample preparation for STED microscopySections were washed twice for 5 min with PBS to remove the embedding resin and incubated in 0.2% Triton X-100 in blockingbuffer 2 (1% bovine serum albumin, 1% fetal calf serum in PBS) for 1 h at room temperature. Samples were incubated with primaryCell 175, 1–15.e1–e9, September 20, 2018 e6Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 antibodies anti-GLUT1 (rabbit anti-human, Merck Millipore #07-1401) and anti-ICAM2 (rat anti-mouse CD102, BD Biosciences#553326) at 1:100 dilution in blocking buffer 2 overnight at 4C. Next, samples were washed three times for 5 minutes in PBS andincubated with STAR Red goat anti-rabbit antibody (Abberior, #2-0012-011-9), Alexa Fluor 594 donkey anti-rat antibody (ThermoFisher Scientific #A21209), and Isolectin GS-IB4 Alexa Fluor 488 conjugate (Thermo Fisher Scientific #I21411), all diluted at 1:500in blocking buffer 2. Subsequently, samples were washed three times for 5 min in PBS and mounted in Abberior Mount Solid Antifademounting reagent (Abberior, #4-0100-007-4) under #1.5 coverslips (22x50 mm, VWR #631-0138) and allowed to cure overnight atroom temperature.STED imaging and image analysisIB4 and ICAM2 signals were used to assess the positions of luminal (IB4 and ICAM2 positive) and abluminal (IB4 positive, ICAM2negative) vessel membranes. Cross-sections of vessels (10-20 per animal) were selected for imaging in areas where luminal andabluminal membranes were clearly distinguishable, typically in the vicinity of the endothelial cell nucleus.STED images were acquired using Abberior STED microscope equipped with 640 nm, 561 nm and 485 nm pulsed excitation lasers,775 nm and 595 nm pulsed depletion lasers, UPlanSApo 100x/1.40 Oil objective (Olympus), 509/22 (GFP), 605/50 (Cy3) and 685/70(Cy5) bandpass emission filters and spatial light modulators for STED beam shaping and alignment. Emitted light was collected withavalanche photodiode detectors using 8 ns-wide detection time gates. 120 ms total pixel dwell time per channel was used. Cy3 andCy5 channels were acquired by line switching followed by the acquisition of the GFP channel.STED images in 488 nm and 640 nm channels were aligned using reference images of fluorescent beads (Tetraspeck, 100 nm,ThermoFisher Scientific #T7280). Centers of beads were determined by centroid fit and resulting positions were used as controlpoints to calculate an affine transformation between the 488 nm and 640 nm channels.To quantify the average amount of membrane-localized GLUT1 per vessel, a measurement area containing all pixels within 300 nmof manually segmented abluminal membrane was created. Luminal membranes were not included in the analysis due to frequentcollapse of vessels during sample preparation. The ratio between mean GLUT1 and mean IB4 signal was used as a measure ofGLUT1 to account for the amount of membrane in the measurement area, imaging depth and antibody penetration differences be-tween samples. The ratio of GLUT1 to IB4 was further corrected using images of Tetraspeck beads for relative intensity fluctuationsbetween Cy5 and GFP detection channels between imaging sessions. Statistical significance was assessed using unpaired Stu-dent’s t tests of log2 transformed data. The analysis was performed using ImageJ and MATLAB 2015 (Mathworks, Inc).Radioactive glucose uptake under AP-2 mknockdownAP-2 knockdown was performed as described before (FLAG-GLUT1 localization under AP-2 mknockdown). Only that cells wereseeded in triplicates in a 24-well plate without coverslips. Radioactive glucose uptake was performed mainly as in Shi and Kandror(2008). Radioactive glucose cocktail was prepared by adding 10 mL of 3H-2-deoxy-D-glucose in ethanol:water solution (specific ac-tivity, 5–10 Ci (185–370 GBq) / mmol) (Perkin Elmer) to a 2.0-mL tube and left open for 5 min to evaporate ethanol. 1.6 mL of KRH()glucose and 16 mL of cold 2-DOG (100X) stock solution (100 mM 2-deoxy-D-glucose in KRH () glucose) (Sigma-Aldrich) were addedto the tube. Cells in each well were rinsed with DMEM (without serum, SFM) warmed to 37C and SFM was added to cells slowly andcarefully by the side of the well in order to avoid detachment of cells. Cells were incubated with 0.5 mL of SFM (in case of +dox con-taining 1mg/ml doxycycline) per well for 2 h at 37C. Cells in each well were washed twice with 2 mL of KRH() glucose buffer (121 mMNaCl, 4.9 mM KCl, 1.2 mM MgSO4, 0.33 mM CaCl2, 12 mM HEPES, pH 7.4) at 37C. 225ml of KRH(-) containing 25 mM final Cyto-chalasin B (dissolved in DMSO) or 0.5% DMSO were added to each well. Immediately after, 25 mL of radioactive glucose cocktail wasadded to all wells. Samples were incubated at 37C for 1 h and then transferred on ice. Radioactive glucose cocktail was aspirated,and ice-cold KRH (+) glucose (121 mM NaCl, 4.9 mM KCl, 1.2 mM MgSO40.33 mM CaCl2:12 mM HEPES, 25 mM D- (+)- Glucose,pH 7.4.) was added to terminate the reaction. Cells were washed once more with ice-cold KRH(+) glucose. Plate was transferred toroom temperature, and 400 mL of 0.1% SDS in KRH () glucose were added to each well, incubated at room temperature for 10 min,and thoroughly resuspend to homogeneity. 100 ml of the lysate was kept to measure protein concentration with DC protein assay kit I(BioRad). 300 mL of lysates were transferred in scintillation vials containing 4 mL of Rotiszint eco plus scintillation fluid (Carl Roth) andcount in a Liquid Scintillation Analyzer (Tri-Carb 2800TR, PerkinElmer) for 1 min per vial. These numbers represent ‘‘Counts in thesamples.’’ In parallel, 10 mL of the radioactive glucose cocktail were mixed with 290 mL of 0.1% SDS in KRH () glucose and thismixture was measured under the same conditions. This number represents ‘‘Counts in the cocktail.’’ The amount of intracellular2-deoxyglucose was calculated using the following formula:½Counts in the sample31000½Counts in the cocktail30:033½C3tpmol/mg 3min, where [C] is protein concentration in mg/mL and t is the total time of incu-bation with radioactive glucose in min. All resulting values have been divided by the overall mean value from all wild-type GLUT1(+ doxycycline, - cytochalasin B) and multiplied by 100 to receive relative values for glucose uptake (%). For test of statistical signif-icance, the mean values of three technical replicates were calculated from three biological replicates. Statistical significance wasdetermined with a paired, one-tailed t test. Depicted values are mean values over all replicates and error bars show standard errorof mean (SEM) over all replicates.e7 Cell 175, 1–15.e1–e9, September 20, 2018Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 GST pulldown assayGLUT1 cytoplasmic C-terminal tail was amplified from pDEST_pcDNA5_FLAG_BirA GLUT1 or GLUT1_P485L with Fw: tatatcGAATTCGTTCCTGAGACTAAAGGC, Rev: aacaatGCGGCCGCTTACACTTGGGAATCAGCC. This resulted in C-terminal tail aminoacids 451-492 (UniProt P11166). Added EcoRI and NotI restriction sites have been used to insert the PCR product into pGEX6P1.Other cytoplasmic regions have been ordered as gBlocks Gene Fragments (IDT) from the region ±20AA of the mutation position,with an additional 50- EcoRV restriction site and 30- Stop codon - NotI restriction site. After restriction, the gene fragments have beeninserted into pGEX6P2.Expression of GST-tagged proteins was induced for 5 h at 22C by addition of isopropyl thio-b-d-galactoside (0.5 mM) to E.coliBL21 in 2X YT medium (0.8 OD). To lyse the cells, bacterial pellets were resuspended in PBS and left on ice for 15 min in presenceof PMSF (1mM), cyanase (4U/mL) and lysozyme (1 mg$mL1). Then, Triton X-100 was added to 0.5% and cells were sonicated for2 min. Lysates were centrifuged for 15 min at 50 000 x g. 300 mL of glutathion-coupled beads were added to the supernatant androtated end-over-end for 2 h at 4C. Beads were washed three times with PBS / 0.1% Triton X-100 and once with PBS.Pulldown experiments were performed using mouse brain extracts. Mouse brains were homogenized in buffer (20 mM HEPES,320 mM sucrose, pH 7.5) containing protease inhibitor cocktail (Sigma-Aldrich). The homogenate was centrifuged at 1000 x g for10 min and the supernatant was supplemented with 1% Triton X-100, 50 mM KCl, 2mM MgCl2, and kept on ice for 10 min with oc-casional vortexing. The lysate was cleared by centrifugation at 17,000 x g for 15 min and at 178,000 x g for 15 min. The supernatantwas recovered and used at a concentration of 7.5 mg protein/mL.The pulldown experiments were performed using 85 mg of GST fusion proteins and 0.6 mL protein extract by end-over-end rotationfor 3 h. The samples were washed four times with buffer containing 20 mM HEPES, 50 mM KCl, 2 mM MgCl2, Triton X-100 (1%) andonce in the same buffer without detergent. Proteins were eluted from the beads twice with Laemmli buffer and analyzed by westernblotting. The following antibodies and dilutions were used: mouse anti-talin 1:1000 (Sigma-Aldrich), mouse anti-g1 adaptin of AP-11:500 (BD Biosciences), mouse anti a-adaptin of AP-2 1:200 (BD Biosciences), horseradish peroxidase-conjugated goat anti-mouse1:2000 or 1:5000 (Jackson labs).Analysis of human missense variants and short linear motifs (SLiMs)SLiM regular expression patterns262 annotated SLiM class definitions (regular expression patterns) were downloaded from the Eukaryotic Linear Motif (ELM) data-base (Dinkel et al., 2016). In order to analyze dileucine motifs, an additional motif ‘.LL.’ was added to this compilation and named‘LIG_diLeu_1’ in order to conserve the naming convention followed by the ELM database.Pathogenic and non-pathogenic missense variantsHumsavar dataset: For the analysis of the missense variants that lead to de novo SLiM instances in protein sequences Uniprot Hum-savar dataset (version 12-Apr-2017) (Famiglietti et al., 2014) was downloaded and filtered for missense variants. Variants that areclassified as ‘Disease’ or ‘Polymorphism’ in this dataset were selected.ClinVar dataset: Clinically relevant genomic variation data annotated in the ClinVar database (Landrum et al., 2016)wasdownloadedfrom the ftp server (ftp.ncbi.nlm.nih.gov/pub/clinvar/tab_delimited/variant_summary.txt.gz) in tab-delimited format (latest update on25th of March, 2017). The downloaded table was filtered for assembly version GRCh38, and variants of type ‘single nucleotide variant’were kept. Inorder to integrate the ClinVar annotations with other kinds of annotationsavailable from the Uniprot database, these nucle-otide variants were translated to the Uniprot protein sequences to obtain single amino-acid substitutions using the Ensembl VariantEffect Predictor (version 82) (VEP) (McLaren et al., 2016). The output of VEP tool was filtered to only keep missense variants suchthat the translated amino-acid substitution occurs at exactly the same amino-acid at the same position of the Uniprot sequence withthe same genename as those of the annotation in the ClinVar dataset (‘Name’ field). Thus, 98,219 unique single amino-acid substitutions(missense variants) from 4,298 Uniprot sequences were obtained. Variants primarily annotated with clinical significance levels‘Pathogenic’, ‘Pathogenic/Likely pathogenic’, or ‘Likely pathogenic’ were grouped as ‘Disease’ variants, while variants annotatedwith ‘Benign’, ‘Benign/Likely benign’, or ‘Likely benign’ were grouped into the ‘Polymorphism’ variants.Protein domains: PFAM domain annotations of proteins were downloaded from the PFAM database (ftp://ftp.ebi.ac.uk/pub/databases/Pfam//releases/Pfam30.0/proteomes/9606.tsv.gz)(Finn et al., 2016).SLiM - PFAM associations: PFAM domains and SLiM classes that are known to interact were downloaded from the ELM database(http://elm.eu.org/interactiondomains).Analysis of gain of SLiMs via missense variants in disordered regionsFor each reviewed human protein from Uniprot (20,191 proteins), the disorder scores of each residue were calculated using IUPred(using the ‘short’ setting). Using a IUPred disorder score cut-off of 0.4, the missense variants in disordered regions were selected. Themissense variants that overlap PFAM domains were further filtered out based on the PFAM domain annotations found in the proteinfeature files downloaded from Uniprot in GFF format (e.g., the link to the GFF file for GLUT1 is https://www.uniprot.org/uniprot/P11166.gff). These protein feature files were also used to detect the transmembrane proteins and their cytoplasmic/extracellularregions. The missense variants in disordered regions and not overlapping any PFAM domains were further classified as variantsfrom 1) the whole proteome, 2) the transmembrane proteins (only those that have annotation of at least one cytoplasmic domainor an extracellular domain, in total 3836 proteins), 3) the cytoplasmic domains of transmembrane proteins, and 4) extracellularCell 175, 1–15.e1–e9, September 20, 2018 e8Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 domains of transmembrane proteins. For each of these classes, the number of disease-causing variants and the number of polymor-phisms that lead to a gain of SLiMs was counted and a two-sided Fisher’s Exact Test was applied to see if there is a statistically sig-nificant difference for the likelihood of a given class of SLiMs to be gained via disease-causing variant compared to that ofpolymorphisms.Peptide-Protein Interaction Network Analysis180 peptide-protein interactions that passed the strict LFQ filter and showed significant differential SILAC ratios between wild-typeand mutant forms of the peptides were used to compose a peptide-protein interaction network. The network was visualized usingCytoscape 3.5.1 (Shannon et al., 2003). Sub-graphs of the significant interactions were generated using R package igraph (version1.0.1) (Csardi and Nepusz, 2006) (using fastgreedy.community function) and visualized using the R packages ggnetwork (Briatte,2016) and ggplot2 (Wilkinson, 2011). Enriched GO terms for each sub-graph were calculated using the topGO R package (Alexaand Rahnenfuhrer, 2016).QUANTIFICATION AND STATISTICAL ANALYSISThe type of statistical test (e.g., Mann–Whitney U test or t test) is annotated in the Figure legend and/or in the Methods segment spe-cific to the analysis. In addition, statistical parameters such as the value of n, mean/median, standard error of mean (SEM), standarddeviation (SD) and significance level are reported in the figures and/or in the figure legends. When * are used to signify the significancelevel the key is reported in the respective figure legend. Statistical analyses were generally performed using R as described inMETHODS AND DETAILS for each individual analysis, if not stated differently.Colocalization analysisImaris v8.4.1 was used for the quantitative colocalization analysis. The original z stack images were adjusted by adding an adequatemask on the respective red channel to subtract background noise (Costes et al., 2004). The threshold for the mask was uniformlyadjusted in each staining experiment. Automatic thresholding was used to define the area where a colocalization would be deter-mined and the statistics was calculated for each colocalization channel (Costes et al., 2004). For the images whose observed cor-relation was not statistically significant in comparison to randomized images, the colocalization channel was built without additionalthresholding on the masked dataset. The resulting thresholded Pearson’s coefficients were exported. The number of images andcells in the analyses is stated in the respective Method sections.DATA AND SOFTWARE AVAILABILITYThe mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Vizcaı´no et al.,2016) partner repository. The accession numbers for the peptide-protein interaction screen dataset, the PRM dataset, and the BioIDdataset reported in this paper are ProteomeXchange: PXD010027, PXD010005, and PXD010061, respectively.e9 Cell 175, 1–15.e1–e9, September 20, 2018Please cite this article in press as: Meyer et al., Mutations in Disordered Regions Can Cause Disease by Creating Dileucine Motifs, Cell (2018),https://doi.org/10.1016/j.cell.2018.08.019 Supplemental Figures1,878305128Mutation databaseHumsavarUniProt annotations of OMIMDisorder predictionIUPred Disease classificationNeurological diseasesHuman Phenotype Ontology, Manual Manual selection26,649Figure S1. Candidate Selection for Peptide-Protein Interaction Screen, Related to Figure 1Candidates were selected from missense disease mutations in the Humsavar database (Uniprot) by selecting mutations in disordered regions that causeneurological diseases (see STAR Methods). n = 136 n = 120peptides with specific interactors peptide pairs with differential interactionsAll peptides Peptide pairs with specific interactorsn = 32 n = 44p 2.2e-16AllPearson s RReplicates0.60.000640.0523.4e11N: 99526 N: 38 N: 15 N: 112 N: 4050510All interactionsGained interactions (unspecific (LFQ))Gained interactions (specific (LFQ))Lost interactions (unspecific (LFQ))Lost interactions (specific (LFQ))Median.SILAC.ratio.Wt/Mut0.90.80.70.60.5ABC(legend on next page) Figure S2. Differential Interactors of Wild-Type and Mutant IDRs, Related to Figure 2(A) Reproducibility of technical replicates Pearson’s R shows significantly higher correlation of technical replicates than correlations between all pull-downsaccording to Welch’s two sample t test.(B) SILAC ratio distributions of detected interactions that can be explained by presence of SLiMs in the peptides and PFAM domains in the interaction partnersPeptide-protein interactions detected in the screen were classified as ‘gained’ or ‘lost’ according to the following criteria: An interaction is classified as ‘gained’ ifthe mutant peptide sequence matches a SLiM pattern that does not match the wild-type peptide sequence and the mutant peptide has an interaction partner thatcontains a compatible PFAM domain to bind that SLiM instance. On the other hand, an interaction is classified as ‘lost’ if the wild-type peptide sequence matchesa SLiM pattern that is not matched in the mutant peptide sequence and the wild-type peptide has an interactio n partner that contains a compatible PFAM domainto bind that SLiM instance. Gained and lost interactions are further sub-classified as ‘LFQ positive’ and ‘LFQ negative’ depending on whether the peptide-proteininteraction has an LFQ value that passes a looser version of the LFQ cut-off (i.e., p value 0.05 and log2 fold enrichment 1). The median SILAC ratio distributions(wild-type versus mutant) of each of these four categories of interactions (‘Gained interactions - LFQ negative’, ‘Gained interactions - LFQ positive’, ‘Lostinteractions - LFQ negative’, and ‘Lost interactions - LFQ positive’) are compared with the median SILAC ratio distributions of all detected interactions from thearray using a Wilcoxon-Mann-Whitney test. Compared to the background distribution of median SILAC ratios (in red), the gained interactions that pass the LFQfilter (in green) show a significant negative skew while the lost interactions that pass the LFQ filter (in purple) show a significant positive skew.(C) Impact of specificity cut-off (LFQ) and differential cut-off (SILAC) on peptide candidates After applying the specificity cut-off (derived from control peptide, seeSTAR Methods) on all interactions, only about half of the 256 (128 variant pairs) peptides showed at least one specific binder according to the LFQ-filter (left piechart, red). These 120 peptides relate to 76 peptide pairs with specific interactions of wild-type and/or mutant peptide. More than half of all 76 peptide pairs withspecific interactors show differential interaction between the variants after applying the SILAC cut-off (see STAR Methods) (right pie chart, red). Rab9 VTI1AFLAGEEA1 LAMP1 Rab4VTI1BGLUT1_Wt GLUT1_P485L GLUT1_Wt GLUT1_P485L GLUT1_Wt GLUT1_P485LFLAGFLAG FLAGFLAG FLAGFigure S3. Intracellular Localization of GLUT1_P485L, Related to Figure 4Quantification of Figure 4B is based on these example images for colocalization study of GLUT1 wild-type and P485L mutant to markers of endocyticcompartments. GLUT1 (green), Marker (red), DAPI (blue). AP1G1 in sample GLUT1 forwardRT [min]MS2 Intensity82.0 82.2 82.4 82.6 82.801000020000300004000050000lightVLAINILGRSCA:0.45RT [min]MS2 Intensity82.0 82.2 82.4 82.601000020000300004000050000heavyVLAINILGRSCA:0.79AP1G1 in sample GLUT1 reverseRT [min]MS2 Intensity79.2 79.4 79.6 79.8 80.0050001000015000200002500030000 lightVLAINILGRSCA:0.81RT [min]MS2 Intensity79.4 79.5 79.6 79.7 79.8 79.9050001000015000200002500030000 heavyVLAINILGRSCA:0.55AP1G1 in sample ITPR1 forwardRT [min]MS2 Intensity92.0 92.1 92.2 92.3 92.4010000200003000040000lightVLAINILGRSCA:0.72RT [min]MS2 Intensity91.8 91.9 92.0 92.1 92.2 92.3 92.4 92.5010000200003000040000heavyVLAINILGRSCA:0.84AP1G1 in sample ITPR1 reverseRT [min]MS2 Intensity90.8 90.9 91.0 91.1 91.2 91.3 91.4010000200003000040000lightVLAINILGRSCA:0.73RT [min]MS2 Intensity90.9 91.0 91.1 91.2 91.3 91.4 91.5010000200003000040000heavyVLAINILGRSCA:0.66AP1G1 in sample CACNA1H_PL forwardRT [min]MS2 Intensity94.7 94.8 94.9 95.00500010000150002000025000lightVLAINILGRSCA:0.68RT [min]MS2 Intensity94.6 94.8 95.0 95.20500010000150002000025000heavyVLAINILGRSCA:0.79AP1G1 in sample CACNA1H_PL reverseRT [min]MS2 Intensity93.9 94.0 94.1 94.2 94.3 94.405000100001500020000lightVLAINILGRSCA:0.82RT [min]MS2 Intensity93.8 94.0 94.2 94.405000100001500020000heavyVLAINILGRSCA:0.63AP1G1 in sample Cntrl forwardRT [min]MS2 Intensity97.0 97.2 97.4 97.6050001000015000200002500030000 lightVLAINILGRSCA:0.74RT [min]MS2 Intensity97.2 97.3 97.4 97.5050001000015000200002500030000 heavyVLAINILGRSCA:0.62AP1G1 in sample Cntrl reverseRT [min]MS2 Intensity95.6 95.8 96.0 96.2 96.4050001000015000lightVLAINILGRSCA:0.7RT [min]MS2 Intensity95.6 95.8 96.0 96.2050001000015000heavyVLAINILGRSCA:0.68AP2A1 in sample GLUT1 forwardRT [min]MS2 Intensity115.0 115.2 115.4 115.60200004000060000800001e+05120000 lightDFLTPPLLSVRSCA:0.74RT [min]MS2 Intensity114.8 115.0 115.2 115.4 115.6 115.80200004000060000800001e+05120000 heavyDFLTPPLLSVRSCA:0.75AP2A1 in sample GLUT1 reverseRT [min]MS2 Intensity114.0 114.2 114.4 114.6 114.801000020000300004000050000lightDFLTPPLLSVRSCA:0.73RT [min]MS2 Intensity114.0 114.2 114.4 114.601000020000300004000050000heavyDFLTPPLLSVRSCA:0.59AP2A1 in sample ITPR1 forwardRT [min]MS2 Intensity120.7 120.8 120.9 121.0 121.1 121.20200004000060000800001e+05lightDFLTPPLLSVRSCA:0.58RT [min]MS2 Intensity120.6 120.8 121.0 121.20200004000060000800001e+05heavyDFLTPPLLSVRSCA:0.73AP2A1 in sample ITPR1 reverseRT [min]MS2 Intensity120.2 120.4 120.6 120.8020000400006000080000 lightDFLTPPLLSVRSCA:0.74RT [min]MS2 Intensity120.3 120.4 120.5 120.6 120.7020000400006000080000 heavyDFLTPPLLSVRSCA:0.49AP2A1 in sample CACNA1H_PL forwardRT [min]MS2 Intensity121.4 121.5 121.6 121.7 121.8 121.90100002000030000400005000060000lightDFLTPPLLSVRSCA:0.64RT [min]MS2 Intensity121.3 121.5 121.7 121.90100002000030000400005000060000heavyDFLTPPLLSVRSCA:0.69AP2A1 in sample CACNA1H_PL reverseRT [min]MS2 Intensity121.0 121.1 121.2 121.3 121.401000020000300004000050000 lightDFLTPPLLSVRSCA:0.6RT [min]MS2 Intensity121.0 121.1 121.2 121.301000020000300004000050000 heavyDFLTPPLLSVRSCA:0.61AP2A1 in sample Cntrl forwardRT [min]MS2 Intensity122.1 122.2 122.3 122.4 122.5 122.6 122.70100002000030000400005000060000 lightDFLTPPLLSVRSCA:0.55RT [min]MS2 Intensity122.2 122.3 122.4 122.5 122.60100002000030000400005000060000 heavyDFLTPPLLSVRSCA:0.24AP2A1 in sample Cntrl reverseRT [min]MS2 Intensity121.9 122.0 122.1 122.2 122.30100002000030000lightDFLTPPLLSVRSCA:0.57RT [min]MS2 Intensity121.9 122.0 122.1 122.2 122.30100002000030000heavyDFLTPPLLSVRSCA:0.43AGLUT1 P485L Mut/WtWt/MutCACNA1H P648LCLTCAP1G1AP2A1AP2B1AP2M1AP2S1AP3B1AP3D1AP3M1AP3S1TUBA1A42024median log2 FC H/LCLTCAP1G1AP2A1AP2B1AP2M1AP2S1AP3B1AP3D1AP3M1AP3S1TUBA1AITPR1 P1059LCLTCAP1G1AP2A1AP2B1AP2M1AP2S1AP3B1AP3D1AP3M1AP3S1TUBA1A42024median log2 FC H/LCLTCAP1G1AP2A1AP2B1AP2M1AP2S1AP3B1AP3D1AP3M1AP3S1TUBA1A42024median log2 FC H/LCLTCAP1G1AP2A1AP2B1AP2M1AP2S1AP3B1AP3D1AP3M1AP3S1TUBA1ACLTCAP1G1AP2A1AP2B1AP2M1AP2S1AP3B1AP3D1AP3M1AP3S1TUBA1Acontrol peptideCLTCAP1G1AP2A1AP2B1AP2M1AP2S1AP3B1AP3D1AP3M1AP3S1TUBA1A42024median log2 FC H/LCLTCAP1G1AP2A1AP2B1AP2M1AP2S1AP3B1AP3D1AP3M1AP3S1TUBA1ABFigure S4. Adaptor Proteins Bind Preferentially to Mutant Variant Peptides Carrying a Dileucine, Related to Figure 5(A) A highly sensitive, targeted mass spectrometry technique (parallel reaction monitoring, PRM) reveals that adaptor proteins (AP-1, AP-2, AP-3) bindpreferentially to peptides carrying a dileucine.(legend continued on next page) (B) Raw elution profiles of transitions from two example peptides from the proteins AP1G1 and AP2A1 for all forward and label swap (reverse) experiments. Ratioswere calculated based on extracted intensity informations from the light and heavy channel in each experiment for each transition. Different transitions aredisplayed with different colors. The spectrum contrast angle (SCA) indicates similarity to the corresponding spectrum in the spectrum library and is displayed forevery peak. Pluripotency Immunocytochemistry BIHi037-A25 mOverlayDAPINANOGOverlayDAPIOverlayDAPISOX2OCT4OverlayDAPISSEA4OverlayDAPITRA-1-60Figure S5. iPSCs Show Pluripotency Markers, Related to Figure 6iPSCs (BIHi037-A) were reprogrammed from a G1DS patient carrying a heterozygous GLUT1 P485L mutation and were stained for known pluripotency markers(see STAR Methods). inputGST-CACNA1HGST-CACNA1H P618LGSTGST-RHBDF2GST-RHBDF2 P189LGST-L1CAMGST-L1CAM S1194LGST-ITPR1GST-ITPR1 P1059LTalinAP-1 AP-2 p = 0.069p = 1KCNQ1 R452L/R591LLLCFTR P5L/P750LcytoplasmLLLLLLp = 0.305p = 0.04p = 0.041p = 11234All proteins TransmembraneproteinsCytosplasmic domainsoftransmembraneproteinsExtracellular domainsoftransmembraneproteinsdisease mutations vs polymorphismsOdds Ratio dileucine gainLIG_LIR_Gen_1MOD_NEK2_1LIG_diLeu_1LIG_BRCT_BRCA1_1LIG_SH2_STAT501220 2 4log2oddsRatiolog10(pval)diseasepolymorphismABCDFigure S6. Mutation-Induced Gains of Dileucine Motifs Are a Significant Cause of Disease, Related to Figure 7(A) A systematic bioinformatic search from the Clinvar Database revealed four additional mutations, with pathogenic indications, in cytosolic segments oftransmembrane proteins that create dileucine motifs. (Depicted here are only mutations from Table S4/Clinvar that are classified ‘Pathogenic’ or ‘Conflictinginterpretations of pathogenicity’, in case ‘pathogenic’ or ‘likely pathogenic’ was included in the different interpretations, that had not been found with theHumsavar analysis before.) Overlap between Humsavar and Clinvar is: RET_P1039L, L1CAM_S1194L, ITPR1_P1059L , RHBDF2_P189L).(B) Relative frequency of dileucine motif gains in disease mutations (Clinical significance levels ‘Pathogenic’, ‘Pathogenic/Likely pathogenic’, or ‘Likely patho-genic’) and polymorphisms (Clinical significance levels ‘Benign’, ‘Benign/Likely benign’, or ‘Likely benign’) in different disordered regions (IUPred Score = 0.4) ofthe proteome. Dileucine motif gain is significantly enriched in disordered regions of transmembrane proteins. Enrichment becomes more significant when lookingonly at disordered cytoplasmic domains (two-sided Fisher’s exact test).(C) Comparison of gained motifs (disease-associated versus polymorphism) in disordered regions of cytoplasmic tails of transmembrane proteins reveals thedileucine motif is one of the most significant, specific enriched motifs. (ELM Motif LIG_LIR_Nem_3 (-log10(pval) = 2.723722, log2oddsRatio = 2.63761) has beentaken out of graphical representation because it is functional only in nematodes and represents a less specific form of LIG_LIR_Gen_1.)(D) Dileucine containing peptides interact with adaptor proteins. Mutant tails of CACNA1H, L1CAM and ITPR1 show increased binding of AP-1 compared towild-type tails. Mutant tails of CACNA1H and RHBDF2 show increased binding of AP2 compared to wild-type tails. Tails were tagged with GST to pull-downinteraction partners from mouse brain lysate. Talin is shown as a negative control and is not pulled down from any of the constructs.Citations (56)References (48)... The mass spectrometric measurement of interacting proteins has increased in popularity due to the increased sensitivity and possibilities of modern mass spectrometry-based proteomics [6,[8][9][10]. While many studies use whole proteins for interaction studies, the use of peptides in such studies has increased over time [11][12][13][14][15][16]. In this article, we will focus on the newly developing field of peptide array-based interaction studies. ...... While peptide matrices have been used to find the optimal binding sequence [67][68][69], the true power of the approach emerges when it is combined with a proteomics readout. Meyer and coworkers use this principle to analyze mutations that cause neurological diseases [12]. One hundred twenty known disease-causing mutations were selected in extensive bioinformatics analysis. ...... Three of the interaction nodes created a dileucine motif which is necessary for clathrin-dependent transport. In case of the glucose transporter GLUT1, the mutated version was wrongly localized to the endocytic compartment [12]. ...Peptide array–based interactomicsArticleFull-text availableMay 2021ANAL BIOANAL CHEM Daniel Perez Hernandez Gunnar DittmarThe analysis of protein-protein interactions (PPIs) is essential for the understanding of cellular signaling. Besides probing PPIs with immunoprecipitation-based techniques, peptide pull-downs are an alternative tool specifically useful to study interactome changes induced by post-translational modifications. Peptides for pull-downs can be chemically synthesized and thus offer the possibility to include amino acid exchanges and post-translational modifications (PTMs) in the pull-down reaction. The combination of peptide pull-down and analysis of the binding partners with mass spectrometry offers the direct measurement of interactome changes induced by PTMs or by amino acid exchanges in the interaction site. The possibility of large-scale peptide synthesis on a membrane surface opened the possibility to systematically analyze interactome changes for mutations of many proteins at the same time. Short linear motifs (SLiMs) are amino acid patterns that can mediate protein binding. A significant number of SLiMs are located in regions of proteins, which are lacking a secondary structure, making the interaction motifs readily available for binding reactions. Peptides are particularly well suited to study protein interactions, which are based on SLiM-mediated binding. New technologies using arrayed peptides for interaction studies are able to identify SLIM-based interaction and identify the interaction motifs.Graphical abstractViewShow abstract... In recent years, several technologies have attempted to bridge this gap between high-throughput, qualitative assays, and low-throughput, quantitative assays. Peptide microarrays 11 and Luminex bead-based assays 12 allow multiplexed measurements of ∼10−1000 interactions in parallel, but do not take place at equilibrium and therefore cannot return thermodynamic binding constants. Microscale thermophoresis and holdup assay approaches can facilitate such thermodynamic measurements and have successfully been applied toward a variety of systems, including PDZ-and chromo domain peptide interactions, antibody/antigen interactions, and receptor/ligand interactions. ...MRBLE-pep Measurements Reveal Accurate Binding Affinities for B56, a PP2A Regulatory SubunitArticleFull-text availableJul 2021Jamin B. Hein Martha S CyertPolly M. FordyceSignal transduction pathways rely on dynamic interactions between protein globular domains and short linear motifs (SLiMs). The weak affinities of these interactions are essential to allow fast rewiring of signaling pathways and downstream responses but also pose technical challenges for interaction detection and measurement. We recently developed a technique (MRBLE-pep) that leverages spectrally encoded hydrogel beads to measure binding affinities between a single protein of interest and 48 different peptide sequences in a single small volume. In prior work, we applied it to map the binding specificity landscape between calcineurin and the PxIxIT SLiM (Nguyen, H. Q. et al. Elife2019, 8). Here, using peptide sequences known to bind the PP2A regulatory subunit B56α, we systematically compare affinities measured by MRBLE-pep or isothermal calorimetry (ITC) and confirm that MRBLE-pep accurately quantifies relative affinity over a wide dynamic range while using a fraction of the material required for traditional methods such as ITC.ViewShow abstract... In recent years, several technologies have attempted to bridge this gap between high-throughput, qualitative assays and low-throughput, quantitative assays. Protein microarrays (Meyer et al. 2018) and Luminex bead-based assays (Cook et al. 2019) allow multiplexed measurements of ~10-1000 interactions in parallel, but do not take place at equilibrium and therefore cannot return thermodynamic binding constants. Microscale thermophoresis and hold-up assay approaches can facilitate such thermodynamic measurements and have successfully been applied towards a variety of systems, including PDZ-and Chromo domain peptide interactions, antibody/antigen interactions, and receptor/ligand interactions (Wienken et al. 2010;Plach, Grasser, and Schubert 2017;Vincentelli et al. 2015). ...MRBLE-pep measurements reveal accurate binding affinities for B56, a PP2A regulatory subunitPreprintFull-text availableDec 2020Jamin B. Hein Martha S CyertPolly M. FordyceSignal transduction pathways rely on dynamic interactions between protein globular domains and short linear motifs (SLiMs). The weak affinities of these interactions are essential to allow fast rewiring of signaling pathways and downstream responses, but pose technical challenges for interaction detection and measurement. We recently developed a technique (MRBLE-pep) that leverages spectrally encoded hydrogel beads to measure binding affinities between a single protein and 48 different peptide sequences in a single small volume. In prior work, we applied it to map the binding specificity landscape between calcineurin and the PxIxIT SLiM (Nguyen et al. 2019). Here, using peptide sequences known to bind the PP2A regulatory subunit B56, we systematically compare affinities measured by MRBLE-pep or isothermal calorimetry (ITC) and confirm that MRBLE-pep accurately quantifies relative affinity over a wide dynamic range while using a fraction of the material required for traditional methods such as ITC.ViewShow abstract... A substantial number of disease-causing variants are, however, located in regions of predicted disorder and are predicted to affect for example SLiMs (Vacic et al., 2012). Thus, it is becoming clear that missense variants in IDPs can also lead to disease via perturbed interactions that either cause loss or gain of function (Meyer et al., 2018;Li et al., 2019;Wong et al., 2020), including promoting the formation of fibrils and toxic oligomeric species. ...On the potential of machine learning to examine the relationship between sequence, structure, dynamics and function of intrinsically disordered proteinsPreprintJun 2021Kresten Lindorff-Larsen Birthe B. KragelundIntrinsically disordered proteins (IDPs) constitute a broad set of proteins with few uniting and many diverging properties. IDPs-and intrinsically disordered regions (IDRs) interspersed between folded domains-are generally characterized as having no persistent tertiary structure; instead they interconvert between a large number of different and often expanded structures. IDPs and IDRs are involved in an enormously wide range of biological functions and reveal novel mechanisms of interactions, and while they defy the common structure-function paradigm of folded proteins, their structural preferences and dynamics are important for their function. We here discuss open questions in the field of IDPs and IDRs, focusing on areas where machine learning and other computational methods play a role. We discuss computational methods aimed to predict transiently formed local and long-range structure, including methods for integrative structural biology. We discuss the many different ways in which IDPs and IDRs can bind to other molecules, both via short linear motifs, as well as in the formation of larger dynamic complexes such as biomolecular condensates. We discuss how experiments are providing insight into such complexes and may enable more accurate predictions. Finally, we discuss the role of IDPs in disease and how new methods are needed to interpret the mechanistic effects of genomic variants in IDPs.ViewShow abstract... Cancer mutations were shown to occur within linear motif sites located in IDRs [14]. In a specific case, the creation of IDR-mediated interactions was suggested to lead to tumorigenesis [15]. However, it has not been systematically analyzed whether mutations of IDRs can have a direct role driving cancer development or what the main molecular functions and biological processes altered by such events are. ...Mutations of Intrinsically Disordered Protein Regions Can Drive Cancer but Lack Therapeutic StrategiesArticleFull-text availableMar 2021 Bálint MészárosBorbála Hajdu-SoltészAndrás Zeke Zsuzsanna DosztányiMany proteins contain intrinsically disordered regions (IDRs) which carry out important functions without relying on a single well-defined conformation. IDRs are increasingly recognized as critical elements of regulatory networks and have been also associated with cancer. However, it is unknown whether mutations targeting IDRs represent a distinct class of driver events associated with specific molecular and system-level properties, cancer types and treatment options. Here, we used an integrative computational approach to explore the direct role of intrinsically disordered protein regions driving cancer. We showed that around 20% of cancer drivers are primarily targeted through a disordered region. These IDRs can function in multiple ways which are distinct from the functional mechanisms of ordered drivers. Disordered drivers play a central role in context-dependent interaction networks and are enriched in specific biological processes such as transcription, gene expression regulation and protein degradation. Furthermore, their modulation represents an alternative mechanism for the emergence of all known cancer hallmarks. Importantly, in certain cancer patients, mutations of disordered drivers represent key driving events. However, treatment options for such patients are currently severely limited. The presented study highlights a largely overlooked class of cancer drivers associated with specific cancer types that need novel therapeutic options.ViewShow abstract... More specific studies are needed in the future to generalize the pathogenic dysregulation of T cells as well as the role of CD100-related pathways in other patients with PSC. Elucidating the disease-related functional consequences of missense mutations in disordered regions is challenging because such mutations can lead to gain of functions that are hard to predict (37,38) or the predicted functions are not disease-relevant. We observed that the K849T mutation resulted in more protein binding, suggesting a gain-of-function mechanism. ...A heterozygous germline CD100 mutation in a family with primary sclerosing cholangitisArticleFeb 2021Sci Transl Med Xiaojun Jiang Annika BergquistBritt-Sabina LöscherEspen MelumPrimary sclerosing cholangitis (PSC) is a chronic inflammatory liver disease without clear etiology or effective treatment. Genetic factors contribute to PSC pathogenesis, but so far, no causative mutation has been found. We performed whole-exome sequencing in a family with autosomal dominant inheritance of PSC and identified a heterozygous germline missense mutation in SEMA4D , encoding a K849T variant of CD100. The mutation was located in an evolutionarily conserved, unstructured cytosolic region of CD100 affecting downstream signaling. It was found to alter the function of CD100-expressing cells with a bias toward the T cell compartment that caused increased proliferation and impaired interferon-γ (IFN-γ) production after stimulation. Homologous mutation knock-in mice developed similar IFN-γ impairment in T cells and were more prone to develop severe cholangitis when exposed to 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) diet. Transfer of wild-type T cells to knock-in mice before and during DDC exposure attenuated cholangitis. Taken together, we identified an inherited mutation in the disordered cytosolic region of CD100 resulting in T cell functional defects. Our findings suggest a protective role for T cells in PSC that might be used therapeutically.ViewShow abstractA universal peptide matrix interactomics approach to disclose motif dependent protein bindingArticleFull-text availableAug 2021Mol Cell ProteonomicsProtein-protein interactions (PPIs) mediated by intrinsically disordered regions (IDRs) are often based on short linear motifs (SLiM). SLiMs are implicated in signal transduction and gene regulation, yet remain technically laborious and notoriously challenging to study. Here, we present an optimized method for a PRotein Interaction Screen on a peptide MAtrix (PRISMA) in combination with quantitative mass spectrometry. The protocol was benchmarked with previously described SLiM based PPIs using peptides derived from EGFR, SOS1, GLUT1 and CEBPB and extended to map binding partners of kinase activation loops. The detailed protocol provides practical considerations for setting up a PRISMA screen and subsequently implementing PRISMA on a liquid handling robotic platform as a cost effective high-throughput method. Optimized PRISMA can be universally applied to systematically study SLiM based interactions and associated post translational modifications (PTMs) or mutations to advance our understanding of the largely uncharacterized interactomes of intrinsically disordered protein regions.ViewShow abstractUse of viral motif mimicry improves the proteome-wide discovery of human linear motifsPreprintFull-text availableJun 2021 Bishoy WadieVitalii KleshchevnikovElissavet SandaltzopoulouEvangelia PetsalakiLinear motifs have an integral role in dynamic cell functions including cell signalling, the cell cycle and others. However, due to their small size, low complexity, degenerate nature, and frequent mutations, identifying novel functional motifs is a challenging task. Viral proteins rely extensively on the molecular mimicry of cellular linear motifs for modifying cell signalling and other processes in ways that favour viral infection. This study aims to discover human linear motifs convergently evolved also in disordered regions of viral proteins, under the hypothesis that these will result in enrichment in functional motif instances. We systematically apply computational motif prediction, combined with implementation of several functional and structural filters to the most recent publicly available human-viral and human-human protein interaction network. By limiting the search space to the sequences of viral proteins, we observed an increase in the sensitivity of motif prediction, as well as improved enrichment in known instances compared to the same analysis using only human protein interactions. We identified 8,400 motif instances at various confidence levels, 105 of which were supported by all functional and structural filters applied. Overall, we provide a pipeline to improve the identification of functional linear motifs from interactomics datasets and a comprehensive catalogue of putative human motifs that can contribute to our understanding of the human domain-linear motif code and the mechanisms of viral interference with this.ViewShow abstractAntagonistic regulation controls clathrin-mediated endocytosis: AP2 adaptor facilitation vs restraint from clathrin light chainsArticleJun 2021Lisa RedlingshöferFrances M BrodskyOrchestration of a complex network of protein interactions drives clathrin-mediated endocytosis (CME). A central role for the AP2 adaptor complex beyond cargo recognition and clathrin recruitment has emerged in recent years. It is now apparent that AP2 serves as a pivotal hub for protein interactions to mediate clathrin coated pit maturation, and couples lattice formation to membrane deformation. As a key driver for clathrin assembly, AP2 complements the attenuating role of clathrin light chain subunits, which enable dynamic lattice rearrangement needed for budding. This review summarises recent insights into AP2 function with respect to CME dynamics and biophysics, and its relationship to the role of clathrin light chains in clathrin assembly.ViewShow abstractPRISMA and BioID disclose a motifs-based interactome of the intrinsically disordered transcription factor C/EBPαArticleFull-text availableJun 2021 Evelyn RambergerValeria SapozhnikovaElisabeth Kowenz-Leutz Achim LeutzC/EBPα represents a paradigm intrinsically disordered transcription factor containing short linear motifs and post-translational modifications (PTM). Unraveling C/EBPα protein interaction networks is a prerequisite for understanding the multi-modal functions of C/EBPα in hematopoiesis and leukemia. Here, we combined arrayed peptide matrix screening (PRISMA) with BioID to generate an in vivo validated and isoform specific interaction map of C/EBPα. The myeloid C/EBPα interactome comprises promiscuous and PTM-regulated interactions with protein machineries involved in gene expression, epigenetics, genome organization, DNA replication, RNA processing, and nuclear transport. C/EBPα interaction hotspots coincide with homologous conserved regions of the C/EBP family that also score as molecular recognition features. PTMs alter the interaction spectrum of C/EBP-motifs to configure a multi-valent transcription factor hub that interacts with multiple co-regulatory components, including BAF/SWI-SNF or Mediator complexes. Combining PRISMA and BioID is a powerful strategy to systematically explore the PTM-regulated interactomes of intrinsically disordered transcription factors.ViewShow abstractShow moreSubcellular Trafficking of Mammalian Lysosomal Proteins: An Extended ViewArticleFull-text availableDec 2016INT J MOL SCI Catherine StaudtEmeline Puissant Marielle BoonenLysosomes clear macromolecules, maintain nutrient and cholesterol homeostasis, participate in tissue repair, and in many other cellular functions. To assume these tasks, lysosomes rely on their large arsenal of acid hydrolases, transmembrane proteins and membrane-associated proteins. It is therefore imperative that, post-synthesis, these proteins are specifically recognized as lysosomal components and are correctly sorted to this organelle through the endosomes. Lysosomal transmembrane proteins contain consensus motifs in their cytosolic regions (tyrosine-or dileucine-based) that serve as sorting signals to the endosomes, whereas most lysosomal acid hydrolases acquire mannose 6-phosphate (Man-6-P) moieties that mediate binding to two membrane receptors with endosomal sorting motifs in their cytosolic tails. These tyrosine-and dileucine-based motifs are tickets for boarding in clathrin-coated carriers that transport their cargo from the trans-Golgi network and plasma membrane to the endosomes. However, increasing evidence points to additional mechanisms participating in the biogenesis of lysosomes. In some cell types, for example, there are alternatives to the Man-6-P receptors for the transport of some acid hydrolases. In addition, several \"non-consensus” sorting motifs have been identified, and atypical transport routes to endolysosomes have been brought to light. These \"unconventional” or \"less known” transport mechanisms are the focus of this review.ViewShow abstract2016 update of the PRIDE database and its related toolsArticleFull-text availableSep 2016Nucleic Acids Res Juan Antonio VizcainoAttila Csordas Noemi Del Toro Ayllón Henning HermjakobThe authors wish to correct the affiliation of one of the authors. GerhardMayer s only affiliation should be ³Medizinisches Proteom Center (MPC), Ruhr-Universität Bochum, D-44801 Bochum, Germany. He does NOT have a second affiliation at: ¹European Molecular Biology Laboratory, European Bioinformatics Institute (EMBL-EBI), Wellcome Trust Genome Campus, Hinxton, Cambridge, CB10 1SD, UK. The authors apologise to the readers for this error.ViewShow abstractThe Ensembl Variant Effect PredictorArticleFull-text availableJun 2016GENOME BIOLWilliam McLaren Laurent GilSarah HuntFiona CunninghamThe Ensembl Variant Effect Predictor is a powerful toolset for the analysis, annotation, and prioritization of genomic variants in coding and non-coding regions. It provides access to an extensive collection of genomic annotation, with a variety of interfaces to suit different requirements, and simple options for configuring and extending analysis. It is open source, free to use, and supports full reproducibility of results. The Ensembl Variant Effect Predictor can simplify and accelerate variant interpretation in a wide range of study designs.ViewShow abstractClinVar: Public archive of interpretations of clinically relevant variantsArticleFull-text availableNov 2015Nucleic Acids ResMelissa J Landrum Jennifer M LeeMark BensonDonna MaglottClinVar (https://www.ncbi.nlm.nih.gov/clinvar/) at the National Center for Biotechnology Information (NCBI) is a freely available archive for interpretations of clinical significance of variants for reported conditions. The database includes germline and somatic variants of any size, type or genomic location. Interpretations are submitted by clinical testing laboratories, research laboratories, locus-specific databases, OMIM®, GeneReviews™, UniProt, expert panels and practice guidelines. In NCBI s Variation submission portal, submitters upload batch submissions or use the Submission Wizard for single submissions. Each submitted interpretation is assigned an accession number prefixed with SCV. ClinVar staff review validation reports with data types such as HGVS (Human Genome Variation Society) expressions; however, clinical significance is reported directly from submitters. Interpretations are aggregated by variant-condition combination and assigned an accession number prefixed with RCV. Clinical significance is calculated for the aggregate record, indicating consensus or conflict in the submitted interpretations. ClinVar uses data standards, such as HGVS nomenclature for variants and MedGen identifiers for conditions. The data are available on the web as variant-specific views; the entire data set can be downloaded via ftp. Programmatic access for ClinVar records is available through NCBI s E-utilities. Future development includes providing a variant-centric XML archive and a web page for details of SCV submissions.ViewShow abstractThe Human Phenotype Ontology in 2017.ArticleNov 2017Sebastian Köhler Nicole A Vasilevsky Mark Engelstad Peter RobinsonDeep phenotyping has been defined as the precise and comprehensive analysis of phenotypic abnormalities in which the individual components of the phenotype are observed and described. The three components of the Human Phenotype Ontology (HPO; www.human-phenotype-ontology.org) project are the phenotype vocabulary, disease-phenotype annotations and the algorithms that operate on these. These components are being used for computational deep phenotyping and precision medicine as well as integration of clinical data into translational research. The HPO is being increasingly adopted as a standard for phenotypic abnormalities by diverse groups such as international rare disease organizations, registries, clinical labs, biomedical resources, and clinical software tools and will thereby contribute toward nascent efforts at global data exchange for identifying disease etiologies. This update article reviews the progress of the HPO project since the debut Nucleic Acids Research database article in 2014, including specific areas of expansion such as common (complex) disease, new algorithms for phenotype driven genomic discovery and diagnostics, integration of cross-species mapping efforts with the Mammalian Phenotype Ontology, an improved quality control pipeline, and the addition of patient-friendly terminology.ViewShow abstractPicky: A simple online PRM and SRM method designer for targeted proteomicsArticleFeb 2018Br J Pharmacol Henrik Zauber Marieluise KirchnerMatthias SelbachViewGene editing in mouse zygotes using the CRISPR/Cas9 systemArticleMar 2017METHODS Benedikt Wefers Sanum BashirJana Rossius Ralf KühnThe generation of targeted mouse mutants is a key technology for biomedical research. Using the CRISPR/Cas9 system for induction of targeted double-strand breaks, gene editing can be performed in a single step directly in mouse zygotes. This article covers the design of knockout and knockin alleles, preparation of reagents, microinjection or electroporation of zygotes and the genotyping of pups derived from gene editing projects. In addition we include a section for the control of experimental settings by targeting the Rosa26 locus and PCR based genotyping of blastocysts.ViewShow abstractReorganizing the protein space at the Universal Protein Resource (UniProt)ArticleJan 2012NUCLEIC ACIDS RES Philippe Le MercierThe mission of UniProt is to support biological research by providing a freely accessible, stable, comprehensive, fully classified, richly and accurately annotated protein sequence knowledgebase, with extensive cross-references and querying interfaces. UniProt is comprised of four major components, each optimized for different uses: the UniProt Archive, the UniProt Knowledgebase, the UniProt Reference Clusters and the UniProt Metagenomic and Environmental Sequence Database. A key development at UniProt is the provision of complete, reference and representative proteomes. UniProt is updated and distributed every 4 weeks and can be accessed online for searches or download at http://www.uniprot.org.ViewShow abstractCharacterizing Protein–Protein Interactions Using Mass Spectrometry: Challenges and OpportunitiesArticleMar 2016TRENDS BIOTECHNOLArne H. Smits Michiel VermeulenDuring the past decades, mass spectrometry (MS)-based proteomics has become an important technology to identify protein-protein interactions (PPIs). The application of a quantitative filter in protein enrichments from crude lysates to discriminate bona fide interactors from background proteins has proved to be particularly powerful. Recently, many different approaches to identify PPIs have been developed, including proximity-ligation technology and global interactome profiling based on the co-behavior of protein complexes in biochemical purification or perturbation experiments. Furthermore, methodologies have been introduced that provide information regarding the stoichiometry and topology of detected PPIs. We review these novel methodologies and emphasize the need to miniaturize workflows to analyze protein interactions in biological and pathological contexts where sample amounts are limited.ViewShow abstractCorrigendum: Global quantification of mammalian gene expression controlArticleMar 2013NatureBjörn Schwanhäusser Dorothea Busse Na LiMatthias SelbachGene expression is a multistep process that involves the transcription, translation and turnover of messenger RNAs and proteins. Although it is one of the most fundamental processes of life, the entire cascade has never been quantified on a genome-wide scale. Here we simultaneously measured absolute mRNA and protein abundance and turnover by parallel metabolic pulse labelling for more than 5,000 genes in mammalian cells. Whereas mRNA and protein levels correlated better than previously thought, corresponding half-lives showed no correlation. Using a quantitative model we have obtained the first genome-scale prediction of synthesis rates of mRNAs and proteins. We find that the cellular abundance of proteins is predominantly controlled at the level of translation. Genes with similar combinations of mRNA and protein stability shared functional properties, indicating that half-lives evolved under energetic and dynamic constraints. Quantitative information about all stages of gene expression provides a rich resource and helps to provide a greater understanding of the underlying design principles.ViewShow abstractShow moreAdvertisementRecommendationsDiscover moreProjectPipelines for Genomics Analysis using GNU Guix Alexander Blume Altuna Akalin Katarzyna Wreczycka[...]Jonathan RonenIn bioinformatics, as well as other computationally-intensive research fields, there is a need for workflows that can reliably produce consistent output, from known sources, independent of the soft ware environment or configuration settings of the machine on which they are executed. Indeed, this is essential for controlled comparison between different observations or for the wider dissemination of workflows. Providing this type of reproducibility and traceability, however, is often complicated by the need to accommodate the myriad dependencies included in a larger body of software, each of which generally come in various versions. Moreover, in many fields (bioinformatics being a prime example), these versions are subject to continual change due to rapidly evolving technologies, further complicating problems related to reproducibility. Here, we propose a principled approach for building analysis pipelines and managing their dependencies with GNU Guix. As a case study to demonstrate the utility of our approach, we present a set of highly reproducible pipelines called PiGx for the analysis of RNA-seq, ChIP-seq, Bisulfite-seq, and single-cell RNA-seq. All pipelines process raw experimental data, and generate reports containing publication-ready plots and figures, with interactive report elements and standard observables. Users may install these highly reproducible packages and apply them to their own datasets without any special computational expertise beyond the use of the command line. We hope such a toolkit will provide immediate benefit to laboratory workers wishing to process theirown data sets or bioinformaticians seeking to automate all, or parts of, their analyses. In the long term, we hope our approach to reproducibility will serve as a blueprint for reproducible workflows in other areas. Our pipelines, along with their corresponding documentation and sample reports, are available at http://bioinformatics.mdc-berlin.de/pigx PiGx: Reproducible genomics analysis pipelines with GNU Guix | Request PDF. Available from: https://www.researchgate.net/publication/328027339_PiGx_Reproducible_genomics_analysis_pipelines_with_GNU_Guix [accessed Oct 31 2018]. ... [more]View projectProjectIdentification of disease mechanisms related to intrinsically disordered proteins Anna SzymborskaView projectArticleNovel URAT1 mutations caused acute renal failure after exercise in two Chinese families with renal h...October 2012 · Gene Zongzhe Li Hu none Ding Chen Chen[...]Yongman LvRenal hypouricemia (RHUC), as an infrequent hereditary disease, is associated with severe complications such as exercise-induced acute renal failure (EIARF). Loss-of-function mutations in urate transporter gene URAT1 (Type 1) and in glucose transporter gene GLUT9 (Type 2) are major causes of this disorder. In this study, URAT1 and GLUT9 were screened in two uncorrelated families from mainland ... [Show full abstract] China and a total of five mutations were identified in exons, including two novel heterozygous URAT1 mutations. In four members of the first family, c.151delG (p.A51fsX64) in exon 1 was detected, which resulted in a frameshift and truncated the original 553-residue-protein to 63 amino acid protein. A missense mutation c.C1546A (p.P516T) in exon 9 in GLUT9 was revealed in the second family, which caused a functional protein substitution at codon 516. These two novel mutations were neither identified in the subsequent scanning of 200 ethnically matched healthy control subjects with normal serum UA level nor in a 1000 genome project database. Thus our report identifies two novel loss-of-function mutations (c.151delG in URAT1 and p.P516T in GLUT9) which cause RHUC and renal dysfunction in two independent RHUC pedigrees.Read moreArticleFull-text availableIdentification of a Novel Homozygous Mutation, TMPRSS3: c.535G A, in a Tibetan Family with Autosomal...December 2014 · PLoS ONEDongyan FanWei ZhuDejun Li[...] Ping WangDifferent ethnic groups have distinct mutation spectrums associated with inheritable deafness. In order to identify the mutations responsible for congenital hearing loss in the Tibetan population, mutation screening for 98 deafness-related genes by microarray and massively parallel sequencing of captured target exons was conducted in one Tibetan family with familiar hearing loss. A homozygous ... [Show full abstract] mutation, TMPRSS3: c.535G A, was identified in two affected brothers. Both parents are heterozygotes and an unaffected sister carries wild type alleles. The same mutation was not detected in 101 control Tibetan individuals. This missense mutation results in an amino acid change (p.Ala179Thr) at a highly conserved site in the scavenger receptor cysteine rich (SRCR) domain of the TMPRSS3 protein, which is essential for protein-protein interactions. Thus, this mutation likely affects the interactions of this transmembrane protein with extracellular molecules. According to our bioinformatic analyses, the TMPRSS3: c.535G A mutation might damage protein function and lead to hearing loss. These data suggest that the homozygous mutation TMPRSS3: c.535G A causes prelingual hearing loss in this Tibetan family. This is the first TMPRSS3 mutation found in the Chinese Tibetan population.View full-textArticleFull-text availableBoth sequence and context are important for flagellar targeting of a glucose transporterMarch 2012 · Journal of Cell Science Khoa D Tran Dayana Rodriguez-Contreras Ujwal Shinde Scott LandfearMany of the cilia- and flagella-specific integral membrane proteins identified to date function to sense the extracellular milieu, and there is considerable interest in defining pathways for targeting such proteins to these sensory organelles. The flagellar glucose transporter of Leishmania mexicana, LmxGT1, is targeted selectively to the flagellar membrane, whereas two other isoforms, LmxGT2 and ... [Show full abstract] LmxGT3, are targeted to the pellicular plasma membrane of the cell body. To define the flagellar targeting signal, deletions and point mutations were generated in the N-terminal hydrophilic domain of LmxGT1, which mediates flagellar localization. Three amino acids, N95-P96-M97, serve critical roles in flagellar targeting, resulting in strong mistargeting phenotypes when mutagenized. However, to facilitate flagellar targeting of other non-flagellar membrane proteins, it was necessary to attach a larger region surrounding the NPM motif containing amino acids 81-113. Molecular modeling suggests that this region might present the critical NPM residues at the surface of the N-terminal domain. It is likely that the NPM motif is recognized by currently unknown protein-binding partners that mediate flagellar targeting of membrane-associated proteins.View full-textArticleMutations in the clathrin-assembly gene Picalm are responsible for the hematopoietic and iron metabo...August 2003 · Proceedings of the National Academy of SciencesMitchell L. KlebigMelissa D WallMark D. Potter[...]E M RinchikRecessive N-ethyl-N-nitrosourea (ENU)-induced mutations recovered at the fitness-1 (fit1) locus in mouse chromosome 7 cause hematopoietic abnormalities, growth retardation, and shortened life span, with varying severity of the defects in different alleles. Abnormal iron distribution and metabolism and frequent scoliosis have also been associated with an allele of intermediate severity (fit14R). ... [Show full abstract] We report that fit14R, as well as the most severe fit15R allele, are nonsense point mutations in the mouse ortholog of the human phosphatidylinositol-binding clathrin assembly protein (PICALM) gene, whose product is involved in clathrin-mediated endocytosis. A variety of leukemias and lymphomas have been associated with translocations that fuse human PICALM with the putative transcription factor gene AF10. The Picalmfit1-5R and Picalmfit1-4R mutations are splice-donor alterations resulting in transcripts that are less abundant than normal and missing exons 4 and 17, respectively. These exon deletions introduce premature termination codons predicted to truncate the proteins near the N and C termini, respectively. No mutations in the genes encoding Picalm, clathrin, or components of the adaptor protein complex 2 (AP2) have been previously described in which the suite of disorders present in the Picalmfit1 mutant mice is apparent. These mutants thus provide unique models for exploring how the endocytic function of mouse Picalm and the transport processes mediated by clathrin and the AP2 complex contribute to normal hematopoiesis, iron metabolism, and growth.Read moreDiscover the world s researchJoin ResearchGate to find the people and research you need to help your work.Join for free ResearchGate iOS AppGet it from the App Store now.InstallKeep up with your stats and moreAccess scientific knowledge from anywhere orDiscover by subject areaRecruit researchersJoin for freeLoginEmail Tip: Most researchers use their institutional email address as their ResearchGate loginPasswordForgot password? Keep me logged inLog inorContinue with GoogleWelcome back! Please log in.Email · HintTip: Most researchers use their institutional email address as their ResearchGate loginPasswordForgot password? Keep me logged inLog inorContinue with GoogleNo account? Sign upCompanyAbout usNewsCareersSupportHelp CenterBusiness solutionsAdvertisingRecruiting© 2008-2021 ResearchGate GmbH. All rights reserved.TermsPrivacyCopyrightImprint