The Prion Protein Is Embedded in a Molecular Environment That Modulates Transforming Growth Factor Β and Integrin Signaling

The Prion Protein is Embedded in a Molecular Environment that Modulates Transforming Growth Factor β and Integrin Signaling

Farinaz Ghodrati

A thesis submitted in conformity with the requirements for the degree of Master of Science Department of Laboratory Medicine and Pathobiology University of Toronto

The Prion Protein is Embedded in a Molecular Environment that Modulates Transforming Growth Factor β and Integrin Signaling

Farinaz Ghodrati

Master of Science

Department of Laboratory Medicine and Pathobiology University of Toronto

2018 Abstract

The prion protein (PrP) is known for its fundamental role in a group of neurodegenerative disorders, aptly called prion diseases. The function of the normal cellular prion protein (PrPC) as well as the underlying molecular mechanisms that lead to neurotoxicity in disease are still unresolved. The previously discovered evolutionary relationship between PrP and ZIP zinc transporters implicated PrPC in morphogenetic reprogramming events but also uncovered surprising differences when comparing signaling pathways downstream of PrPC in separate cellular paradigms. To extend this line of research, we combined CRISPR-Cas9-based genetic engineering to generate four relevant murine PrP-deficient cell models, with quantitative mass spectrometry-based analyses to compare the molecular environment of PrP in these models.

Interestingly, a unified theme emerged from these studies that placed PrPC in a specialized membrane domain that modulates TGF beta and integrin signaling.

Acknowledgments

First and foremost, I would like to wholeheartedly thank my supervisor, Dr. Gerold Schmitt-Ulms for his exceptional mentorship and unwavering support throughout my studies. I credit his genuine passion and dedication to research and his keen scientific mind with helping me grow both personally and professionally over these last two years. Though my time under his mentorship has now ended, I will take the lessons I’ve learned here with me onwards through my career.

I would also like to extend my thanks to the members of my advisory committee – Dr. Joel Watts and Dr. Sunit Das. Their advice and guidance have helped me immensely in my research. Additionally, many thanks to the tireless efforts of the Department of Laboratory Medicine and Pathobiology’s Graduate Coordinator, Dr. Harry P. Elsholtz, and the Graduate Administrator, Rama Ponda.

I am also tremendously grateful for the pleasure and privilege of collaborating with past and present members of the Schmitt-Ulms lab – Dr. Hansen Wang, Louisa Wang, and Michael Solarski. My time here would not have been as productive or enjoyable without them. Special thanks to Dr. Declan Williams, who helped move this project leaps and bounds with his invaluable guidance and expertise in the field of mass spectrometry. And finally, this entire project would not have been possible without the accomplishments of Dr. Mohadeseh Mehrabian. To her, I express my immense gratitude for her untiring mentorship, guidance and, above all, friendship.

Lastly, I would like to thank my friends and family for their love, patience, and support over the years, specifically, AS, NS, NR, NM and ABZ, for your unique sense of humour. I would also want to express my thanks to the colleagues at the Tanz CRND who provided me with tremendous scientific and personal support during my degree. And above all, my utmost appreciation to my parents for being such amazing sources of strength and support.

iii

Table of Contents

Acknowledgments ...... iii

Table of Contents ...... iv

List of Tables ...... vii

List of Figures ...... viii

List of Appendices ...... ix

Abbreviations ...... x

Chapter 1 ...... 1

The Prion Protein: What We Do and Don’t Know ...... 1 1.1 Introduction ...... 1 1.2 Prion diseases ...... 1 1.2.1 Overview of prion diseases ...... 1 1.2.2 Inherited human prion diseases ...... 2 1.2.3 Acquired human prion diseases ...... 3 1.2.4 Sporadic human prion diseases ...... 5 1.3 Prion propagation ...... 7 1.3.1 How the prion concept came to be ...... 7 1.3.2 PrPC expression is essential for prion infection ...... 8 1.3.3 Prion toxicity ...... 9 1.4 Structure of the prion protein ...... 10 1.4.1 Overview of the structural features of PrP ...... 10 1.4.2 Proteolytic processing of PrP ...... 11 1.5 Evolutionary origin of the prion protein ...... 12 1.5.1 Discovery of the PrP-ZIP connection ...... 12 1.5.2 How PrP evolved from ZIP ancestors ...... 13 1.5.3 Further proof on the family reunion of PrP and ZIPs ...... 14 1.6 Proposed functions of the prion protein ...... 15 1.6.1 Overview of the roles ascribed to PrPC ...... 15 1.6.2 Epithelial-to-mesenchymal transition ...... 16 1.6.3 Additional PrP-related phenotypes ...... 17 iv

1.7 Methodology background ...... 18 1.7.1 Overview of the experimental workflow ...... 18 1.7.2 Cell models ...... 19 1.7.3 CRISPR-Cas9 knockout technology ...... 20 1.7.4 Affinity capture matrix ...... 22 1.7.5 Mass spectrometry ...... 22

Chapter 2 ...... 25

Rationale, Hypothesis and Objectives ...... 25 2.1 Rationale for the ZIP study ...... 25 2.2 Rationale for the PrP interactome study ...... 26

Chapter 3 ...... 28

ZIP6-mediated NCAM1 Phosphorylation ...... 28 3.1 Introduction ...... 29 3.2 Results ...... 30 3.2.1 In vitro phosphorylation of the longest NCAM1 isoform by GSK3B ...... 30 3.3 Discussion ...... 32 3.4 Methods ...... 34 3.4.1 Sample preparation for immunoprecipitation ...... 34 3.4.2 Protein immunoprecipitation ...... 34 3.4.3 Active kinase assay ...... 35

Chapter 4 ...... 36

The Prion Protein is Embedded in a Molecular Environment that Modulates Transforming Growth Factor β and Integrin Signaling ...... 36 4.1 Introduction ...... 37 4.2 Results ...... 38 4.2.1 Design of comparative PrP interactome analysis in four mouse cell models ...... 38 4.2.2 Comparison of PrP interactome analyses across models ...... 41 4.2.3 Cell type-specific effects of PrP knockout on the global proteome reflect its molecular interactions ...... 48 4.2.4 PrP selectively interacts with Ece1 and Tfrc dimers ...... 51

4.2.5 Candidate PrP interactors exist in the secretory pathway or at the cellular membrane ..... 55 4.2.6 TGFβ1 profoundly affects steady-state levels of several PrP interactors but depletion of PrP only reduces NCAM1 ...... 58 4.3 Discussion ...... 61 4.4 Conclusion ...... 65 4.5 Methods ...... 65 4.5.1 Western blot analyses ...... 65 4.5.2 Cell culture and transfection ...... 66 4.5.3 Sample preparation for immunoprecipitation analyses ...... 66 4.5.4 Protein immunoprecipitation workflow ...... 67 4.5.5 Nanoscale HPLC-ESI tandem mass spectrometry ...... 67 4.5.6 Protein identification and quantification ...... 68 4.5.7 Data availability ...... 68

Chapter 5 ...... 70

Conclusion and Future Directions ...... 70

References ...... 76

Appendix ...... 87

List of Tables

Table 4.1: Comparison of PrP interactome in four mouse cell models (list of non-specific interactors truncated, see also Supplementary Table S1)

vii

List of Figures

Figure 3.1: In vitro phosphorylation of longest NCAM1 isoform by GSK3B.

Figure 4.1: Design of comparative PrP interactome study.

Figure 4.2: Validation of successful technical execution of quantitative interactome analysis.

Figure 4.3: The molecular environment of PrP is cell model-specific and comprises several novel candidate interactors.

Figure 4.4: PrP interacts selectively with the Ece1 dimer, not its more abundant monomer.

Figure 4.5: PrP’s molecular environment is enriched for proteins with known roles in TGFβ1 and integrin signaling.

Figure 4.6: Whereas TGFβ1 treatment causes divergent shifts in steady-state levels of a subset of PrP interactors, PrP-depletion in the same paradigm only affected NCAM1 protein levels.

viii

List of Appendices

Supplementary Figure S4.1: Consistent and selective enrichment of PrP contrasted to non-specific binding of Gapdh.

Supplementary Figure S4.2: Selective PrP co-enrichment of Cd109 and Tmem206.

Supplementary Figure S4.3: Evidence that Ece1 is not expressed in CAD5 cells at levels detectable by western blot analysis.

Supplementary Figure S4.4: PrP co-immunoprecipitates Tfrc from wild-type but not PrP knockout NMuMG cell lysates.

Supplementary Table S4.1: Comparison of the PrP interactome in four mouse cell models.

Supplementary Table S4.2: Global proteome analysis of NMuMG cells -/+ TGFB1 (datset I) in PrP-deficient and wildtype cells (dataset II).

Abbreviations

AA amino acid

AD Alzheimer’s disease

BCA bicinchoninic acid

BSE bovine spongiform encephalopathy

CFC cysteine-flanked core

CID collision induced dissociation

CJD Creutzfeldt-Jakob disease

CNS central nervous system

CRISPR clustered regularly interspaced short palindromic repeats

CWD chronic wasting disease

DSB double-stranded break

Ece1 endothelin converting enzyme 1

ECL enhanced chemiluminescence

EMT epithelial-to-mesenchymal transition

ESI electrospray ionization fCJD familial Creutzfeldt-Jakob disease

FOI Fear of Intimacy

FFI fatal familial insomnia

GAPDH glyceraldehyde 3-phosphate dehydrogenase

GPI glycosylphosphatidylinositol x

GSK3 glycogen synthase kinase 3

GSS Gerstmann-Straussler-Scheinker

HCD higher energy collisional dissociation

HDR homology-directed repair

HPLC high performance liquid chromatography iCJD iatrogenic Creutzfeldt-Jakob disease iTRAQ isobaric tag for relative and absolute quantitation

IP immunprecipitation

KD knockdown kDa kilo Dalton

KO knockout

Met Methionine

Mo mouse

MS2 tandem MS

MW molecular weight

NCAM1 neural cell adhesion molecule 1

NHEJ nonhomologous end joining

PIPLC phosphatidylinositol-specific phospholipase C

PK proteinase K

PL PrP-like

PTM post-translational modification

PrP prion protein

PrPC cellular isoform of the prion protein

PrPSc pathogenic isoform of the prion protein

PSM peptide-to-spectrum match

RNAi RNA interference sCJD sporadic Creutzfeldt-Jakob disease

SDS sodium dodecyl sulfate sFI sporadic fatal insomnia

SLC39 solute carrier 39

TGFβ1 transforming growth factor beta 1

TM transmembrane

TSE transmissible spongiform encephalopathy

Val Valine vCJD variant Creutzfeldt-Jakob disease

VPSPr variably protease-sensitive prionopathy

UK United Kingdom

ZIP Zrt-, Irt-like protein

xii

Chapter 1 The Prion Protein: What We Do and Don’t Know 1.1 Introduction

The prion protein (PrP) is central to the pathogenesis of prion diseases [1], also known as transmissible spongiform encephalopathies (TSEs). These fatal neurodegenerative disorders that can affect humans and other mammals [2] manifest with clinical symptoms of motor and cognitive dysfunction, often accompanied by the generation of PrP-specific deposits that can be stained with amyloid dyes [2]. Prions, the transmissible agents in prion diseases, possess unique characteristics that set them apart from other infectious pathogens, such as their lack of an informational nucleic acid genome [2,3]. The cellular isoform of PrP (PrPC) is widely expressed in healthy vertebrate cells of diverse lineages and has been linked to a wide range of cellular activities including, but not limited to, cell adhesion, neuritogenesis, circadian cycle regulation and ion transport [3,4]. In the diseased state, the central pathogenesis event occurs when PrPC undergoes a conformational change, leading to the alteration of its physiochemical properties and conversion into its pathogenic isoform, the scrapie prion (PrPSc) [2,3]. It is well documented that PrPC is not only the substrate of prion replication but also plays a key role in the prion-induced neurodegeneration that follows [2].

1.2 Prion diseases

1.2.1 Overview of prion diseases

The classic prion diseases affecting humans include Creutzfeldt-Jakob disease (CJD), Kuru, fatal familial insomnia (FFI) and Gerstmann-Straussler-Scheinker (GSS) disease [2]. Variant CJD (vCJD), sporadic fatal insomnia (sFI) and variably protease-sensitive prionopathy (VPSPr) have more recently been identified as new forms of these disorders [5]. Clinical symptoms of human prion diseases include rapidly progressive dementia, visual or cerebellar impairments, myoclonus, akinetic mutism and pyramidal as well as extrapyramidal signs [2]. The most common prion diseases affecting other mammals include bovine spongiform encephalopathy (BSE) also known

as “mad cow” disease, scrapie, which affects sheep and goat, and chronic wasting disease (CWD) affecting elk and deer [2]. Among the more recently recognized animal prion diseases are feline spongiform encephalopathy and TSE in non-human primates, affecting domestic cats and lemurs respectively [6,7]. The histopathological features of prion diseases observed in the central nervous system (CNS) consist of neuronal loss, astrogliosis as well as spongiform change, which is defined as the vacuolation of the neuropil [1,8]. These phenotypes are observed to variable degrees in different forms of the disease [8].

Worldwide, prion diseases were thought to affect about one person in a million every year [6] but numbers are trending upward with closer to 1 in 500,000 new cases tallied in more recent epidemiological surveys [9]. In spite of this low incidence rate, prion diseases have provoked remarkable interest both within the scientific community as well as the general public [2,6]. This is due to the unique pathobiology of these rare neurodegenerative diseases, owing to their initially unprecedented transmissible agents, or prions, as well as the public health threat that they could pose to the human population with the possibility of transspecies transmission from zoonotic sources, with BSE being the best known example of this risk [6,10].

1.2.2 Inherited human prion diseases

Human prion diseases are categorized as inherited, infectious or sporadic; this is based on the genetic, neuropathological, and clinical profiles associated with each case [1,2]. Patients with inherited forms of the disease, also called familial prion diseases, carry a germline mutation in the PRNP gene and follow an autosomal dominant inheritance pattern [1,2,11]. Pathogenic PRNP mutations can fall into three main categories: point mutations causing an amino acid (AA) substitution, point mutations leading to a premature stop codon, and additional octapeptide repeat insertions [10]. To date more than 30 different mutations of PRNP have been reported [10]. Familial prion diseases account for 10-20% of all cases and include FFI, GSS and familial CJD (fCJD) [1,2].

FFI has a mean age of onset of 51 years and a clinical duration of about 18 months [12]. FFI patients initially present with insomnia or disrupted sleep, with additional symptoms, including

hallucinations, autonomic hyperactivation as well as motor abnormalities, such as ataxia and myoclonus [5,10,12], presenting at some point. Interestingly, the disease was originally called thalamic dementia due to the severe degeneration of the thalamus that was observed in these patients [5,12,13]. FFI is genetically characterized by the D178N mutation in the PRNP gene, coupled with a methionine (Met) at the M129V human PRNP polymorphism [5,10,12]. The same D178N mutation is linked to fCJD when a Val is present at codon 129; however, recent data suggests that this association might not be as concrete as previously assumed [5,10]. This AA polymorphism, presenting with either a Met at codon 129 of PrP or a valine (Val) has also been shown to influence the etiology of sporadic and iatrogenic forms of the disease, with individuals homozygous at codon 129 being relatively more susceptible to prion diseases [2,14]. PrP codon 129 heterozygosity (Met/Val) provides relative protection against the disease and is associated with a lower risk and prolonged incubation time in various prion disorders [2,14].

GSS presents with slowly progressive ataxia followed by later onset dementia and has been linked to a number of PRNP mutations, including P102L [5,10,15]. GSS patients are usually 30 to 60 years of age at the onset of the disease, which has a course of 5 years on average [5,6]. Familial forms of CJD account for about 5-15% of all CJD cases [5]. Clinical symptoms of the disease include the rapid progression of dementia and myoclonus [5,10]. The E200K mutation, which is the most common mutation leading to fCJD, is also the most common cause of inherited prion disease worldwide [10]. Although the clinical expression of these disorders varies, it should be noted that in reality these conditions are symptomatically overlapping, with the above characteristic features recognized as extremes within a spectrum of clinical presentations [2,5].

1.2.3 Acquired human prion diseases

Acquired prion diseases, which include iatrogenic CJD (iCJD), Kuru and vCJD, are transmitted to patients as a result of environmental exposure to prions and, therefore, rely on the infectious nature of prion agents [5,6]. A feature of these disorders is the long silent incubation period following exposure to the transmissible agent, that precedes clinical manifestation [2]. However, it has been shown that different routes of transmission could lead to contrasting incubation periods with intracerebral cases exhibiting incubation periods ranging about 2 to 4 years, as opposed to

peripheral cases, displaying an average of approximately 15 years [6]. The iatrogenic form of CJD is transmitted to patients during surgical or medical procedures [2]. Documented cases of iCJD have occurred following corneal transplantation of a graft from a sporadic CJD (sCJD) carrier, as a result of surgical operations with contaminated equipment, and after neurosurgical implantation of contaminated EEG electrodes [2]. In addition, possible vCJD transmission following blood transfusion from incubating asymptomatic donors has been observed in a few cases [16]. CJD transmission has also been reported in patients that received human growth hormone preparations harvested from prion disease contaminated human pituitary [2]; notably, the majority of these patients that developed iCJD were homozygous at the 129th codon of PRNP for either Val or Met [2].

Because prions are unusually resistant to traditional sterilization and disinfection procedures and have long silent incubation times during which further transmission can occur, it is of great importance to find effective decontamination techniques [2,17]. Therefore, ongoing research in the field is in pursuit of prion inactivation methods that could be broadly and cost-effectively applied to medical instruments [17].

Kuru is another acquired prion disease that originated in the Fore linguistic population of Papua New Guinea Eastern Highlands [5,14]. The Kuru epidemic is thought to have arisen due to ritualistic endocannibalism of a carcass from a human sCJD patient [5,18]. Symptoms of Kuru include tremor, in addition to progressive cerebellar ataxia, which consequently led to this horizontally transmitted prion disease being named ‘Kuru’ meaning ‘to shiver’ in the Fore language [5,18]. At its peak, the Kuru epidemic killed up to 2% of the village populations annually and imposed a strong selection pressure on the PRNP gene in the Fore groups affected by the disease [5,18]. Similar to other prion diseases, heterozygosity at codon 129 of PRNP has been associated with a lower risk of Kuru and longer symptom-free disease incubations. This is supported by the high number of survivors carrying this polymorphism as well as longer than average -as long as 56 years- incubation times in M129V Kuru patients [2,5,14]. Intriguingly, this positive evolutionary pressure has led to development of a novel protective G127V PrP variant during the Kuru epidemic [14]. Remarkably, further investigation on this polymorphism has shown

that transgenic mice expressing human PrP V127 are as resistant to prion disease as PrP knockout (KO) mice, demonstrating complete resistance to sCJD, Kuru and vCJD prions [14]. However, transgenic mice that expressed both wild-type and variant human PrP only displayed resistance to sCJD and Kuru prions, which are very similar; these mice did not exhibit vCJD prion resistance, likely due to the fact that the Fore people were not previously exposed to this disease variant [14]. More insights into the mechanism was gained by studying transgenic mice expressing different ratios of wild-type to variant PrP, showing that V127 can inhibit wild-type prion propagation in a dose-dependent fashion [14].

In contrast to the classical (sporadic) form of CJD, vCJD presents clinicopathologically with psychiatric and behavioural symptoms accompanied by sensory disturbances in some patients [5,6]. As the disease progresses, patients exhibit ataxia and late-onset dementia, in addition to other typical sCJD symptoms, such as myoclonus [2,6]. The distinguishing pathological marks of vCJD include an abundance of a daisy-like vacuolation pattern, called “florid plaques”, made up of amyloidotropic fibrils [2]. It also has a relatively younger patient pool, with a mean age of onset of 29 years, and on average a longer clinical duration of 18 months [5,6]. This new variant of CJD, which was first reported in 1996 in the United Kingdom (UK), has been linked to the BSE epidemic of the mid-1980s and early 1990s that affected the UK, and to a lesser extent other European countries [2,5]. More specifically, it has been proposed that BSE prions infected the human population via dietary exposure to prion-contaminated tissue, leading to the emergence of this novel form of acquired human prion disease through zoonosis [2,6,10]. Transmission studies also support the notion that vCJD is caused by the same prion strain that leads to BSE, which is indeed distinct from the sCJD strain [19].

1.2.4 Sporadic human prion diseases

Sporadic prion diseases include sCJD, sFI and VPSPr. Given that these patients lack a germline PRNP mutation as well as exposure to a TSE agent, the underlying cause of sporadic prion diseases remains unknown [2,5]. However, it has been hypothesized that the sporadic forms of the disease could be due to somatic mutations in the PRNP gene as well as a spontaneous conversion of PrPC into PrPSc [5]. Accounting for 85% of cases, sCJD is the most common human prion disease [2]

with clinical and neuropathological features similar to those of the genetic forms of the disease [10]. The onset of sCJD typically occurs in patients aged 45 to 75 years old with a peak at the 60 to 65 age group [6]. The disease usually has a short clinical duration with about 70% of the subjects succumbing to illness in under 6 months [6].

In 1999 a sporadic form of fatal insomnia with clinical symptoms similar to FFI was reported in patients lacking PRNP mutations as well as a family history of the disease [5,12]. The neuropathological features of sFI also resemble those of the familiar form of the disease [5,12]. In line with the role of M129V polymorphism, all the patients diagnosed with sFI have thus far been Met homozygotes at PrP codon 129 [5].

A new type of atypical dementia was reported in 2008 in eleven 129VV homozygous patients with no other mutation in the PRNP gene [20]. The disease had characteristic neuropathological features of TSEs; however, a notable feature of the pathology was the reduced resistance of the insoluble PrPSc isoform to proteinase K (PK) digestion; as such, this variant of sporadic prion disease was named “protease-sensitive prionopathy” (PSPr) [20]. In the following years new cases of the disease with 129MV and 129MM genotypes were found [21]. Interestingly, the sensitivity to PK proteolysis is reduced in patients exhibiting a Met variant at the 129 codon, which led to the renaming of the disease to “variably protease-sensitive prionopathy” [21]. The patients present with aphasia, Parkinsonian signs as well as ataxia [21]. The 129MV heterozygotes have a later average age of onset of 72 years compared to 64 and 65 years in 129MM and 129VV subjects respectively. For a prion disease, VPSPr has a long clinical duration; cases involving the 129VV variant exhibit an average clinical duration of 2 years compared to approximately 3.5 years observed in the heterozygote and Met homozygote genotype groups [21]. Lastly, clinicians have noted that, thus far, the majority of affected individuals exhibit the 129VV genotype. As such, it has been proposed that the 129MM genotype could play a protective role in VPSPr disease formation [22].

1.3 Prion propagation

1.3.1 How the prion concept came to be

Prions are unlike any other infectious agent due to their unique propagation mechanism [23]. The transmissibility of the scrapie agent was accidentally discovered in 1937 in a now seminal study in which a population of sheep were exposed to scrapie during a vaccination, using formalin extracted sheep brain tissue, which was intended to immunize the sheep against a common virus [24]. In the following years, studies in the field went on to also illustrate the infectious nature of human prion diseases such as Kuru and CJD by demonstrating their transmissibility to animals [24]. A pivotal study by Alper and colleagues in 1967 showed that the TSE agent is highly resistant to treatments such as ultra violet and ionizing radiation, which are typically used to inactivate nucleic acids [25]. This finding prompted a number of hypotheses regarding the nature of the agent responsible for TSEs. One of the most remarkable conjectures was the ‘protein-only’ hypothesis, launched in 1967 by Griffith, which speculated that a self-replicating protein could be the responsible agent for TSEs [2,26]. Future work by Prusiner and colleagues provided further support for this hypothesis and the notion that TSE agents are free of nucleic acids [23]. Prusiner went on to coin the term ‘prion’ for these novel proteinaceous infectious particles [23].

During the same year in 1982, PrP was purified as the protein constituent of these infectious prions [27]. The protease resistant core of PrPSc that was isolated from infectious material had a molecular weight (MW) of about 27 to 30 kilo Daltons (kDa) and was therefore named PrP 27-30 [27]. This led to the identification of PRNP as the gene that encodes for PrP [28]. However, a surprising finding at the time was the constitutive expression of PRNP as well as its corresponding mRNA in healthy animals; this led to the idea that PrP must be present in alternative forms of PrPC and PrPSc [29]. Given that PrPC and PrPSc have identical AA sequences, it seemed plausible that there must be a post-translational modification (PTM) distinguishing these two PrP isoforms [29]. However, it was found that rather than a chemical modification, PrPC and PrPSc differ by their conformation [30]. The two PrP isoforms have vastly different secondary structures; while PrPC is predominantly an alpha helical molecule, PrPSc loses its alpha helical content and gains a more beta-sheet-rich structure [30]. Future studies solidified the notion that PrPSc is generated by a

template-driven conformational change of PrPC and propagates by triggering the misfolding of endogenous PrPC molecules [1–3].

1.3.2 PrPC expression is essential for prion infection

Although TSE agents are made up exclusively of PrPSc [1], it is important to recognize the pivotal role of PrPC in prion replication, as it is the required substrate for the propagation of prion infections [31–38]. The critical importance of PrPC expression in prion replication was documented by studies which highlighted that the presence of PrPSc alone fails to cause disease manifestation in mice lacking the Prnp gene [31,35]. It was also found that the relative susceptibility to prion diseases is determined by PrPC expression levels [31,32], again indicating that cellular PrPC is the limiting factor in the infection paradigm. Moreover it has been shown that disease progression and incubation time are inversely proportional to PrPC levels [31,32]. RNA interference (RNAi)-based knockdown (KD) of PrPC expression, or ablation of PrPC in the neurons of transgenic mice resulted in recovery from early neuronal dysfunction and extended the survival of mice, regardless of extraneuronal PrPSc accumulation [33]. In addition, a clever study design utilizing intraocular prion inoculation into PrP-deficient mice that contain PrPC only in a neural graft tissue of embryonic neuroectodermal origin, highlighted that prion spread within the CNS is dependent on PrPC expression, as the planted neural grafts remained uninfected and failed to show prion pathology [34].

Further support for the requirement of PrPC in infection was provided in two 2001 studies, demonstrating prevention of prion propagation via anti-PrP antibodies with selective affinity for PrPC and slight or no binding to the pathogenic isoform of the protein [36,37]. In these studies scrapie infection of mouse neuroblastoma Neuro2a (N2a) cells, which are one of the few cell lines susceptible to prion infection, was prevented using anti-PrP monoclonal antibodies or via phosphatidylinositol-specific phospholipase C (PIPLC), which cleaves cell surface-bound PrPC [36,37]. Treatment of infected cells with PIPLC or the antibodies resulted in rapid loss and degradation of PrPSc in the cultures [36,37]. PrPSc levels are therefore regulated by its rate of formation from PrPC as well as its degradation process [37]. The results of these in vitro studies

were replicated in vivo in a scrapie mouse model where it was shown that administration of anti- PrP monoclonal antibodies also inhibited prion replication [38].

1.3.3 Prion toxicity

Despite the prion field’s continuous interest in the precise nature of conditions that promote prion infection, the underlying molecular mechanisms leading to CNS pathology in these disorders are yet to be fully established [2,39]. Three different hypotheses have been put forth regarding the mechanism behind the neurodegeneration associated with the conversion of cellular PrPC into its pathogenic isoform:

1. The most widely accepted hypothesis proposes a toxic gain-of-function for PrP; more specifically, it implies that upon conversion to PrPSc, the conformational change brings about toxic properties in the protein unrelated to the usual physiological functions of PrPC [39]. For instance, PrPSc aggregates could trigger neurodegeneration by interfering with synaptic function, blocking axonal transport or activating apoptotic pathways [39].

2. The second hypothesis states that there is a biological function of PrP that is lost when the cellular isoform is converted to PrPSc and the presumed loss-of-function in turn leads to the prion- induced neurodegeneration [39]. Given that one of the many functions attributed to PrPC is its anti- apoptotic role, it is conceivable that a loss-of-function mechanism could at least contribute to neuronal death [39]. However, since prenatal or postnatal depletion of PrP expression causes no significant phenotypic change, rather than giving rise to prion disease symptoms, PrPC loss-of- function cannot solely explain the neurodegeneration that occurs in prion diseases [31,40]. Nevertheless, an aspect of PrP’s biology such as its cytoprotective function, which might be replaceable under normal cellular conditions, can possibly become indispensable under stressful conditions in the disease state [39].

3. A third hypothesis postulates that PrP’s normal cellular function is subverted upon interaction with the pathogenic PrPSc isoform [39]. Essentially, the third hypothesis proposes a paradigm in which the conversion into PrPSc transforms PrPC from a neuroprotective molecule to a neurotoxic agent [39]. It is suggested that this neurotoxicity then leads to neurodegeneration [39]. Given that

PrPC is a glycosylphosphatidylinositol (GPI)-linked cell surface glycoprotein and many GPI- anchored proteins serve as signal transducers, it could be hypothesized that misfolded PrP alters this signal transduction machinery, resulting in prion toxicity [39]. It has been shown that transgenic mice expressing GPI-negative PrP only show minimal neurological dysfunction and brain pathology despite PrPSc amyloid plaque accumulation [41,42]. Remarkably, in a more recent study, neuronal overexpression of anchorless PrP in wild-type mice induced late-onset neuropathology presenting with CNS amyloid depositions resembling that of GSS patients; however, this study further substantiated the crucial role of membrane-anchored PrPC in disease pathogenesis [42]. These findings point to the possible role of the GPI anchor in prion replication as well as prion toxicity, since both were reduced in GPI-negative PrPC-expressing mice [41,42]. The mechanism by which the interaction between PrPC and PrPSc culminates in neurotoxicity could involve aggregation of the cell surface PrPC, which blocks specific regions of the protein [39]. The overall effect of these events could then be the hindrance of the biological function of PrP through blockade of its protein interactions and signaling transduction [39].

1.4 Structure of the prion protein

1.4.1 Overview of the structural features of PrP

The PRNP gene is located on chromosome 20 in humans and Prnp on chromosome 2 in mice [43]. Both prion genes possess three exons with exon three encompassing the entire open reading frame [43]. The prion protein is 208 amino acids in length and has an approximate molecular weight of 35 kDa in mice and humans [3]. PrPC is located in lipid rafts, which are sphingolipid and cholesterol-rich microdomains, and attaches to the cell membrane’s outer leaflet via a GPI anchor [3]. GPI anchors are glycolipid structures that are attached to the C-terminus of eukaryotic proteins during PTM [44,45]. There are also two potential N-glycosylation sites located at the C-terminal of PrPC that could be glycosylated post-translationally, with un-, mono- and diglycosylated forms of PrPC present in the cell [3,43]. The prion protein structure is fairly conserved among different species presenting with an unstructured N-terminal domain and a C-terminus consisting of three alpha helices, two short beta strands and loop domains [3,43]. The C-terminal of PrPC contains a disulfide bond between two cysteine residues of helices B and C [43].

1.4.2 Proteolytic processing of PrP

As a protein’s function is dependent on its structure, this section will provide an overview of the functional relevance of the different domains of PrP as well as its proteolytic processing. The flexible N-terminal of the protein, among other features, is comprised of a neurotoxic domain, an octameric repeat region and a hydrophobic core [3]. The histidine residues present in the octarepeat region harbour a high affinity for binding copper and to a lower extent zinc [43,46]. The neurotoxic domain, located between AA 105 to 125 of mouse PrPC, has proven essential for the conformational conversion of PrPC into PrPSc [3].

PrPC can undergo three pivotal cleavage events: alpha, beta and ectodomain shedding [3]. Alpha cleavage occurs directly N-terminal of the hydrophobic core of mouse PrPC (AA 111-134) generating a soluble N1 fragment (~11 kDa) and a membrane-bound C1 fragment (~18 kDa) [3]. Since the N-terminal domain of PrPC possesses ligand binding properties, the alpha cleavage serves as a negative regulator of these functions [3]. The resulting N1 and C1 segments have displayed distinct physiological functions [3]. N1 has been implicated in neuroprotection and intercellular signalling [3]. As for C1, its role has proven controversial with some studies reporting a protective and myelinotrophic role while others have linked it to neurotoxic effects [3]. Overall, alpha cleavage serves a protective role by preventing prion propagation, as the cleavage site resides within the neurotoxic domain of PrPC [3]. Alpha cleavage has also been found to be impaired upon conversion of PrPC to PrPSc [3]. Since an intact neurotoxic domain is required for the misfolding of PrPC, transgenic mice expressing only C1 displayed no markers of neurodegeneration or PrPSc accumulation [3]. In addition, transgenic mice expressing both the C1 fragment and PrPC exhibited prolonged disease incubation and decreased production of PrPSc compared to wild-type mice [3]. The alpha cleavage event also inhibits the binding of PrPC N terminus to amyloid beta oligomers, which are the toxic species in Alzheimer’s disease (AD) [3]. The oligomer-induced neurotoxicity is further blocked by the binding of the N1 fragment to these toxic elements [3].

The beta cleavage event, which occurs at the end of the octapeptide repeat region (AA 51-90), results in the formation of N2 and C2 fragments with respective MWs of approximately 9 kDa and 20 kDa [3]. This cleavage event is less prominent under physiological conditions in comparison to

the alpha cleavage [3]. Unlike the alpha cleavage event, which is prevented in misfolded PrP, beta cleavage is actually the result of proteolytic processing during disease states [3]. The C2 fragment has been postulated to serve a pathophysiological role since it is the major PrP cleavage product in brains of CJD patients as well as prion infected neuroblastoma cells [3]. C2 also shares resemblance with the protease-resistant core of PrPSc, and as a result, has been called the in vivo homologue of PrP 27-30 by prion researchers [3]. Intriguingly, the beta site cleavage can also take place under oxidative stress conditions, linking it to the suggested protective role of PrPC against oxidative stress [3].

Lastly, ectodomain shedding takes place near the C-terminal GPI-anchor of the protein, giving rise to the release of a nearly full-length PrP from the cell membrane [3]. Notably this cleavage event differs from the phospholipase-induced cleavage of PrP’s GPI anchor [3]. It has been indicated that shedding of PrP occurs in neurons as well as lymphoid cells, as observed by the presence of soluble shed PrPC in human cerebrospinal fluid and blood [3]. As GPI anchors are employed in signal transduction [39], shedding of PrPC can affect its receptor properties by interfering with the signalling cascades involving PrPC. Similar to the N1 fragment, shed PrPC can also bind amyloid beta oligomers in AD and suppress their toxicity [3]. Prion disease-associated neurodegeneration seems to require membrane-anchored forms of PrPC, as models expressing anchorless PrPC present with longer incubation periods [3,42]. However, it should be noted that shed PrPC is capable of misfolding; and, as a matter of fact, GSS has been linked to PRNP mutations giving rise to anchorless forms of PrPC [3,42].

1.5 Evolutionary origin of the prion protein

1.5.1 Discovery of the PrP-ZIP connection

An interactome investigation of PrPC in N2a cells by Watts et al., documented two members of the Zrt- Irt-like protein (ZIP) family of metal ion transporters, ZIP6 and ZIP10, as novel interactors for PrP [47]. This finding provoked a deeper comparison of ZIP and PrP folds and sequences as well as examination of their membrane topology and other functional and phylogenetic characteristics [48] that culminated in the surprising discovery that the prion gene family has evolved from the ZIP family of zinc transporters [49]. It is worth mentioning that the evolutionary

origins of the prion protein were a mystery to the field until the aforementioned studies revealed the shared phylogenetic origin of ZIPs and PrP [48], a remarkable leap forward in the quest to find the function of PrP.

ZIPs are a family of Type-III transmembrane (TM) proteins that are involved in the import of zinc and other divalent cations into the cytosol and are encoded by an ancient gene family, called the solute carrier 39 (SLC39) family [49]. The SLC39 gene family is universally present in all organisms [50] and plays a major role in controlling the cellular zinc homeostasis [51]. The LIV- 1 subfamily of ZIP transporters is one of the four branches within this protein family that is distinguishable from other ZIPs by a conserved intramembrane metalloprotease motif [51] and includes ZIP5, ZIP6 and ZIP10 [49]. These three ZIP ion transporters form a distinct subbranch of the human ZIP protein family based on AA sequence similarities [48]. More specifically they contain an ectodomain similar to that of PrPC, which is accordingly termed the PrP-like (PL) domain [48] by our group. This PL domain contains a conserved AA sequence flanked by two cysteine residues, which is appropriately named the cysteine-flanked core (CFC) domain [48]. These two cysteine residues are ubiquitously present in all prion and PL domains and form a disulfide bridge in the prion family of proteins.

1.5.2 How PrP evolved from ZIP ancestors

Prion protein and its paralogs including Doppel and Shadoo are only present in vertebrates [48] but genes encoding ZIPs are present in all kingdoms of life. The Schmitt-Ulms group have performed a systematic bioinformatics investigation into the underlying mechanism with which the prion founder gene has evolved from ZIPs [48]. These findings pointed to two key genomic rearrangement events that took place half-billion years apart: the first event being the rise of the first ZIP gene encoding a PL ectodomain containing a CFC, and the second being the emergence of the prion founder gene itself [48]. To further elaborate, during the emergence of early metazoan, a CFC domain was inserted into an ancient ZIP transporter gene or was evolved de novo [48]. Then, in the course of early vertebrate speciation, a descendant of that ZIP transporter, which harboured a PL ectodomain resembling that of ZIP5, ZIP6 and ZIP10, produced a processed mRNA which was subsequently reverse transcribed and inserted into a region of the genome with

no synteny relationship with the parent gene [48]. The said retrocopy is thought to have become a functional retrogene by employing its nearby 5′ promoter element, giving rise to the prion subfamily founder gene [48].

1.5.3 Further proof on the family reunion of PrP and ZIPs

Other orthogonal pieces of evidence pointing towards the shared origin of the prion gene family and the LIV-1 branch of the ZIP gene family could be found by probing the function and structure of these proteins. Various studies in the literature have linked PrP and its closest molecular cousins within the ZIP transporter family to a morphogenetic reprogramming event called epithelial-to- mesenchymal transition (EMT). Independent lines of research have displayed converging results that point to both ZIP6 [52] and PrP [53] playing a role in cell migration during zebrafish gastrulation events. Particularly, developmental studies have shown that zebrafish embryos lacking ZIP6 display a gastrulation defect that resembles that of PrP1-deficient zebrafish [52]. This phenotype was later shown to also be mimicked by knocking down ZIP10 [54]. Interestingly, the aforementioned morpholino-based KD of PrP1, one of the two orthologs of PrP in zebrafish, leads to the stop of zebrafish embryogenesis at the gastrula stage, which involves cellular rearrangements relying on EMT [53]. Furthermore, the ZIP ortholog Fear of Intimacy (FOI) has also been implicated in EMT-like morphogenetic cell movements involved in trachea and gonad formation in Drosophila [55]. Mammalian cousins of these proteins have also been associated with EMT, as showcased by the upregulation of both ZIP6 and ZIP10 in particular cancers, and their involvement in the EMT process during certain human carcinomas [50]. It should then be apparent that there is ample evidence supporting the involvement of ZIP descendent proteins in EMT. Independent lines of investigation have also indicated the critical role that PrPC plays during EMT [55], which points to the functional overlap observed between PrP and its evolutionary cousins. Metal ion binding and transport is another example of the functional similarity between PrP and its relatives in the ZIP family. ZIPs are known to play a significant role in cellular zinc homeostasis and the mammalian prion proteins have also been documented to have an affinity for binding divalent cations such as copper and zinc. Therefore, it is not surprising that both PrP and ZIPs5/6/10 are capable of transporting zinc ions across the plasma membrane [49].

The protein structure and amino acid sequence of PrPC also shares resemblance with the ZIP family of metal ion transporters [49]. PrPC has a disordered N-terminal and a globular PL domain, which is similar to that of ZIPs5/6/10 [49]. In addition, the octarepeats present in the prion sequence resemble the histidine-rich repeat motifs of ZIPs and the GPI anchor attachment sequence of PrPC is reminiscent of the first TM domain of the ZIP proteins [49]. This structural overlap can in turn explain the recorded protein-protein interactions between PrPC and both ZIP6 and ZIP10 [47]. It could be that their similar PL ectodomain provides an affinity for the binding of these phylogenetically related proteins [49]. Given that the sole mechanism of prion replication pertains to the binding of PrPSc to PrPC, and PrPC dimers have been observed previously [50], it is indeed possible for proteins with PrP-like folds to be capable of interacting with each other. Other commonalities such as the shared localization of PrP and ZIPs5/6/10 to the plasma membrane [49] only further substantiate the evolutionary link between the two gene families.

Taken together, it is evident that the extent of functional and structural overlap observed between the prion and ZIP protein families as well as the phylogenetic evidence of their shared ancestry suggest the emergence of the prion founder gene from the ZIP family of metal ion transporters.

1.6 Proposed functions of the prion protein

1.6.1 Overview of the roles ascribed to PrPC

PrPC is ubiquitously expressed in all vertebrates and is highly conserved in mammals [6]. More specifically, PrPC is highly expressed in the CNS, neuromuscular junctions, as well as in lymphoreticular tissue [6]. An assortment of functions has been linked to PrPC and many hypotheses have been put forth regarding the role of this protein; however, it is conceivable that each of these reports represent a specific aspect of PrPC’s biology. Since PrP KO mice undergo gross normal development and only exhibit more subtle abnormalities under certain conditions, this line of investigation has not succeeded in shedding conclusive light on the molecular function of PrPC [2]. The most dramatic phenotype observed in PrP KO mice is their resistance to prion infection [2]. It is however puzzling for a protein that is highly expressed among various species, including those not known to be susceptible to prion diseases, to have no function other than determining the organisms’ susceptibility to prion diseases [2].

1.6.2 Epithelial-to-mesenchymal transition

As briefly mentioned above, EMT occurs during the developmental stages of organisms, organogenesis and wound healing [55]. EMT is also one of the main cellular events that take place during disease states, including cancer and fibrosis [55]. During these cellular reprogramming events, the cell-to-cell contacts of epithelial cells are replaced with cell-to-matrix connections present in mesenchymal cells [55]. In other words, as adherens junctions are dissociated, the cells gain a more fibroblastic morphology and focal adhesion complexes are formed, which facilitate the transition of cells with epithelial identity to a more dynamic network of mesenchymal cells. Focal adhesion complexes are integrin-mediated points of contact between the cells and their surrounding extracellular matrix that maintain the structure of mesenchymal tissues [56]. The prion protein had previously been linked to cell adhesion biology [3], with several independent observations implicating PrPC in diverse aspects of EMT [55]. Early downregulation of PrP in zebrafish embryos using morpholino technology led to gastrulation arrest due to a defect in the migration of cells [53], the most significant phenotype linked to PrP deficiency. Additionally, PrPC levels were correlated with metastasis and cell invasion propensities in the field of cancer research [55], which further supports a role of PrP in EMT, given that it is one of the main steps in tumour progression [56]. It is also interesting to note that proteins belonging to the ZIP family of metal ion transporters, from which the prion protein descended from, also play roles in EMT [49,50].

These findings prompted research in our group that investigated whether the reported role of PrP in EMT is conserved in mammalian paradigms [4]. These investigations made use of a well-known model of mammary gland epithelial cells that robustly undergoes EMT upon addition of transforming growth factor beta 1 (TGFβ1) [4]. Since TGFβ1 signaling is implicated in EMT [56], it is regularly used to induce EMT in cultured epithelial cells in various research paradigms. It was found that PrP transcript levels are more than tenfold upregulated during EMT and cells lacking PrP fail to complete this morphogenetic reprogramming [4]. Moreover, it was discovered that PrP controls the levels of its major interactor, neural cell adhesion molecule 1 (NCAM1) [57] as well as a specialized PTM, called polysialylation [4,57] that predominantly exists on NCAM1. Intriguingly, NCAM1 is a crucial regulator of EMT, with an increase in NCAM1 levels, followed by its recruitment to lipid rafts, constituting one of the initial steps of this cellular rearrangement

program [58]. This, in turn, activates a signalling cascade facilitating the formation of focal adhesion complexes [59] and promotes EMT. Remarkably, in neurons PrPC binds NCAM1 and recruits it to lipid rafts, inducing a similar downstream signalling pathway to drive neurite outgrowth [60]. The additional importance of NCAM1 polysialylation in EMT is highlighted with its critical influence on interactions at the cell surface, which in turn affect the balance between cell-matrix and cell-cell adhesion, as well as related processes, such as cell differentiation and migration [58]. Given the close next-neighbour relationship between PrPC and NCAM1, one would expect their functions to be intertwined. PrP’s role in NCAM1 polysialylation and EMT underscores this point and serves as one of the most striking phenotypes associated with PrP to date [4,58]. In fact, the role of PrP in cell adhesion and EMT is supported through independent avenues of research considering PrP’s evolutionary origin, protein interactors, as well as PrP KO studies.

1.6.3 Additional PrP-related phenotypes

The following section will briefly summarize a number of cellular mechanisms that PrPC has been linked to during the years as the prion research community pursued their quest of uncovering the enigma that is the function of the cellular prion protein. In a 1996 study, PrP KO mice were first shown to exhibit circadian activity alterations and changes in sleep patterns [61]. PrP’s potential role in circadian rhythmicity regulation is also consistent with the clinical symptoms of FFI, suggesting a loss-of-function mechanism might contribute to this prion disease [61]. Some other phenotypic abnormalities observed following deletion of the Prnp gene include a peripheral myelin maintenance deficiency [62], altered iron metabolism [63], deficits in long-term potentiation and synaptic function [64], as well as an infrapyramidal mossy fiber development defect [65]. There has also been evidence of PrP exhibiting cytoprotective activities with multiple reports demonstrating its inhibition of Bax-mediated apoptosis in neuronal cultures [39]. Other reports have pointed to a potential role of PrPC in the protection of cells from oxidative stress [39,66] . Given PrP’s evolutionary relationship to ZIPs, which are involved in divalent metal ion transport [49,50], it may not surprise that PrPC possesses the ability to bind copper. In fact, the octarepeat region within the PrPC amino acid sequence harbours four copper-binding sites [43]. However, this domain within PrPC is actually relatively poorly conserved, and merely shares with

ancestral ZIP transporters its disordered fold and an enrichment of pseudo-repeat motifs. Moreover, repeated attempts to makes sense of this ability of PrP to bind copper (and zinc) have been largely unsuccessful, with several studies failing to show a major involvement of PrP in mediating cellular copper trafficking (for an example, see [39]). Other studies have found PrPC to promote neurogenesis and neuronal differentiation [67]. In addition, PrPC has been implicated in the amyloid beta-induced synaptic plasticity impairment observed in AD [68]. More specifically, PrPC was found to act as a receptor of amyloid beta 42 oligomers, mediating their detrimental effects [68].

Although some of these PrP-related phenotypes have held up under close scrutiny, no consistent model that encompasses several of these observations has emerged from them to date. In addition, discrepancies observed by separate investigators and in distinct experimental paradigms have cast uncertainty on the degree to which differences in observations reflect biology versus methodology.

1.7 Methodology background

1.7.1 Overview of the experimental workflow

This chapter’s last section will provide essential background on the methodology of our comparative interactome analyses [69]. It is meant to initially provide a broad overview of the experimental workflow, followed by a more in-depth account of our methodological steps. Our study aimed to establish a database for the binding partners of PrP in different cellular paradigms in order to detect commonalities as well as cell-type specific patterns. It should be noted that comparative nature of the study dictated that we adhered to the same systematic experimental protocol for the different cell lines. Each of the interactome datasets were generated by a D18 antibody-driven immunoprecipitation (IP) of PrP as the bait protein, with PrP-deficient cells available for each of the four models serving as negative controls. Affinity-capture eluates were then prepared for mass spectrometry, ran on an Orbitrap mass spectrometer and analyzed further using sophisticated bioinformatics and statistical algorithms that could uncover and compare the proteins present in the vicinity of PrP across the four cell models of interest.

The following subsections will more deeply go over topics including the nature of the four cells types, the CRISPR-Cas9 knockout technology, affinity capture sample preparation, the characteristics of the D18 PrP-specific antibody and, finally, the mass spectrometry procedures used in our study.

1.7.2 Cell models

The PrP interactome was compared across four relevant mouse cell models in our study: NMuMG, C2C12, N2a and CAD5. The NMuMG cell line is a mammary gland epithelial cell population, which is a widely used mammalian model for studying EMT [4,70]. This murine-derived cell line was previously used by our group to establish the role of PrP in EMT [58]. NMuMG cells respond robustly to TGFβ1 exposure by transition from epithelial cells with a cobblestone morphology to more spindle-like mesenchymal cells [4]. PrP levels were reported to increase in a time-course treatment of TGFβ1 in NMuMG cells and we have therefore performed this treatment before the PrP capture was done. Although not exhaustively explored, NMuMG cells have so far not been successfully infected with PrPSc inoculations (unpublished results, Schmitt-Ulms laboratory). C2C12 cells are myoblasts that are capable of undergoing differentiation into post-mitotic myotubes with serum removal from their culture media [71]. These non-proliferative myotubes have high PrP expression and can replicate prions in culture to high titer [71], hence our use of the differentiated myotubes in our analyses. N2a neuroblastoma cells have been used for the past 20 years in the prion research field and can be host to many strains of PrPSc. Interestingly, CAD5 cells are known for their even higher permissibility to infection by a wider range of different prion strains and higher titer of infectious material [72].

Considering that all of our chosen cell lines share murine origins and have been explored side by side using an optimized methodology for PrP capture, elution and MS analysis paradigms, our study facilitates one of the deepest comparative analyses across different cell types that are performed so far in the field. It is hoped that these findings allow for further analyses relevant to the disease models and the toxicity that is observed downstream.

1.7.3 CRISPR-Cas9 knockout technology

The CRISPR (clustered regularly interspaced short palindromic repeats)-Cas (CRISPR-associated protein) machinery is an adaptive immune system found in various bacteria and archaea [73]. This ability was evolved in these microbial organisms in order to fight off repeated invasions by the same virus [73]. This defense mechanism first recognizes foreign DNA, originating from viruses or plasmids, as non-self and goes on to integrate short segments of the said foreign genetic material in designated locations within CRISPR loci in the host’s genome [73]. These virus-derived DNA fragments, called spacers, in addition to the host Cas proteins provide surveillance such that if a bacterium is consecutively invaded with the same bacteriophage, it can get recognized and disarmed by cleaving its genome [73,74]. More specifically, these incorporated spacer fragments undergo transcription and are processed into small noncoding RNAs, called guide RNAs (gRNAs) [73]. These gRNAs in conjunction with Cas protein complexes can base-pair with incoming foreign nucleic acids that are complementary to the gRNA sequence [73]. This binding in turn initiates the cleavage of the gRNA-foreign nucleic acid complex [73].

There have been three CRISPR systems (Type I, II, and III) identified in archaeal and bacterial hosts with Type II being the one utilizing Cas9 [75]. Type II Cas9 system is comprised of proto- spacer adjacent motifs (PAMs) flanking the 3′ end of the target DNA sequence, which direct the Cas9 cleavage process [76]. Cas9 nuclease has two nuclease domains and is able to make double- stranded breaks (DSBs) at specific loci within the genome, guided by small gRNAs that bind the target DNA [75]. Since the CRISPR mechanism is naturally serving as an antiviral defense, the sequence of the gRNA is usually corresponding to the phage genome sequences; however, Cas9 can easily be retargeted by substituting the gRNA with a different sequence of interest [76]. Following the Cas9 cleavage, one of the two main DNA damage repair pathways has to be activated in order to repair the target genomic locus. If a repair template is not present, the nonhomologous end joining (NHEJ) process is harnessed in order to re-ligate the DSBs [75]. NHEJ is an error-prone mechanism giving rise to insertion/deletion (indel) mutations, which can be exploited in order to produce gene knockouts since occurrence of indel mutations within a coding exon can induce premature stop codons or frameshift mutations [75]. The second DNA repair mechanism called the homology-directed repair (HDR) occurs when an externally introduced

repair template is present and can therefore be harnessed to make accurate modifications at the target genomic locus [75].

The microbial CRISPR adaptive immunity has now been harnessed to establish a novel genome editing technology, which provides a highly specific system capable of efficiently editing eukaryotic cells [75]. This system can be used in mammalian cells by heterologous expression of codon-optimized Cas9 enzyme in addition to the required RNA components used to guide the Cas9 to genomic loci of interest [75]. The PrP KO clones of our four cell lines were generated using the said CRISPR-Cas9 knockout technology [74]. The cells were transfected with plasmids expressing the desired gRNAs as well as the Streptococcus pyrogens Cas9 (SpCas9) nuclease [74]. The plasmids usually contain antibiotic resistance elements that are used to select for the cells that have successfully taken up the plasmids. The gRNAs were designed to target the beginning of the coding sequence on the Prnp gene within the third exon [74]. In this CRISPR-Cas9 system, the SpCas9 enzyme was codon-optimized for mammalian cell expression and harboured a mammalian nuclear localization sequence [74]. When designing the underlying DNA sequence of an expression vector one can use codon optimization in order to take advantage of the fact that each amino acid can be coded by multiple codons and the expression efficiency of a protein of interest can therefore be increased by avoiding rare codons in its sequence [77]. The nuclear localization sequence ensures that the SpCas9 enzyme is transported to the nucleus after being expressed in the mammalian cell, which facilitates eukaryotic gene editing. Our genome editing method was aimed to maximize the production of indel mutations at CRISPR target sites by a double stranded break followed by the NHEJ pathway [74]. While HDR is typically only active in diving cells [75], NHEJ is less dependent on the cell cycle and is also the more efficient of the two DNA damage mechanisms [74]. These indel mutations at the Prnp coding sequence can lead to the generation of PrP-deficient cells as the sequencing of genomic DNA has verified.

Following plasmid transfection for gRNA(s) and Cas9 nuclease, cells were diluted to single cells and expanded for screening to identify successfully engineered clones.

1.7.4 Affinity capture matrix

Each of the PrP pull down experiments start with in vivo mild formaldehyde crosslinking of wild- type and PrP deficient cells in order to preserve the protein-protein interactions [78]. After the crosslinking step, the cells are lysed using detergents. The samples undergo centrifugation steps to discard the cell debris. A bicinchoninic acid (BCA) colorimetric assay is used for protein level adjustment in each of the input samples at the beginning of capture. Lysates are then incubated overnight with the affinity capture matrix, which consists of the D18 anti-PrP antibody bound to the KappaSelect beads.

The D18 antibody is a recombinant protein produced in E. coli that recognizes mouse PrP (AA 132-158) and is conformation dependent [79]. It is solely comprised of the Fab portion of the antibody, based on which KappaSelect beads can be used as a resin by binding to the constant domain of the kappa light chain.

Following the overnight incubation, during which PrP along with its interacting protein partners bind the affinity capture matrix, the samples are thoroughly washed in order to remove nonspecific binders. Subsequently, bait-bound proteins are eluted from the resin by an acidic pH drop.

1.7.5 Mass spectrometry

Mass spectrometry (MS) is a powerful technology that can be employed to provide the relative levels of proteins present in different biological samples [80]. In the interactome analyses performed for this thesis, the relative quantitative comparison of proteins present in affinity- capture eluates derived from wild-type and PrP KO cells made it easy to discriminate specific interactors from non-specific binders to the affinity-matrix.

The primary goal of a mass spectrometer is to measure the mass or, more accurately, the mass-to- charge ratio of analytes [80,81]. Shotgun proteomics experiments typically begin with complex protein mixtures, which are then fragmented into peptides via sequence-specific chemical or enzymatic proteolysis [82]. The mass spectrometer isolates the peptides, breaks them into even smaller fragments and measures the masses of these fragments, which serve as a ‘fingerprint’ of the specific amino acid sequence of a given peptide. These tandem mass spectra are then matched

to the known amino acid sequences of proteins using a matching algorithm [82]. The following sections will provide a more in-depth description of the above steps leading to protein quantification in a mass spectrometer.

Prior to the injection of samples into the mass spectrometer, IP eluates are subjected to sample preparation steps, including denaturation, reduction, alkylation and trypsin digestion [83]. Thus, samples are first denatured in order to make all aspects of their primary structure equally accessible. Subsequently the disulfide bonds in the proteins are reduced, which further aids protein unfolding and facilitates the subsequent protein digestion step, in turn enhancing protein sequence coverage [83]. The free cysteine residues are alkylated in order to prevent the reformation of the disulfide bonds [83]. The protein samples can then be digested by a protease, such as trypsin, into peptides that can be analyzed by the mass spectrometer [83].

The samples have to then undergo a protein labelling procedure, in order for the MS apparatus to successfully provide simultaneous relative quantifications [80,84]. Our interactome studies took advantage of multiplex isobaric tags for relative and absolute quantitation (iTRAQ) labelling [80]. In this state-of-the-art technique peptides are covalently conjugated to chemical tags of identical structure that harbor chemical isotopes and yet are of identical total mass (i.e., isobaric). The latter contain labile components called reporter ions that can be distinguished by tandem MS (MS2) [80]. More specifically, each isobaric tag contains a reporter group, a chemically reactive group and finally a spacer to normalize the mass [80]. The tag works such that each sample’s differentially labelled proteolytic peptides remain isobaric upon mixing and provide a single peak in the initial MS1 scans. However, upon collision-induced dissociation (CID) the labelled peptides release distinct reporter ions, which unlike the overall isobaric tags have different mass-to-charge ratios and can be detected during MS3 scans [80]. The relative intensities of reporter ions determine the relative quantification of the component peptides and in turn their parent proteins within different samples [80].

The isobaric labelled samples are then pooled together and the sample mixtures are purified using strong cation exchange and reversed phase chromatography clean-up procedures [83]. Purification steps are important since excess salt and residual detergent can interfere with the subsequent

ionization step and additionally result in background or chemical noise in the mass spectra [83]. The purified peptides are then separated using high performance liquid chromatography (HPLC), prior to being introduced into the mass spectrometer.

Mass spectrometers consist of three modules: an ion source, a mass analyzer and a detector [81]. Tandem mass spectrometers are comprised of multiple mass analyzers utilizing back-to-back fragmentation and mass-to-charge separation steps to uniquely identify proteins [81]. In tandem MS, the first mass analyzer selects ions of a specific mass-to-charge ratio, the selected ions are then fragmented by CID and a second mass analyzer is used for the analysis of the resulting product ions [81].

One of the methods to ionize samples is electrospray ionization (ESI), which utilizes electrical energy to transfer ions from solution to the gaseous phase [85]. The peptides are required to undergo ionization since MS analyses necessitate the use of charged particles. After ions enter the mass spectrometer, mass analyzers construct an MS spectrum based on the measured mass-to- charge ratio [81]; however, because a given mass is not unique to a single peptide, further analyses are required to determine the AA sequence [82]. The peptides are fragmented into smaller pieces by collision with inert gas molecules [82]. The information on the AA sequence of each peptide can subsequently be inferred from a computational analysis of the MS2 fragment spectrum [82]. More precisely, the peptide fragmentation process generates several mass ladders by randomly fragmenting a given peptide at several possible positions within its peptide and side-chain bonds and by retaining the charge either at the front-end or back-end of the resultant fragments. Although there are many ways to match MS2 spectra to protein sequences, in one approach the collected tandem spectra are computationally matched to a database of predicted protein MS2 spectra, in order to initially identify the peptide sequences, and then assign them to their parent proteins [82].

Lastly, the relative quantitation of peptides across several samples will be based on an MS3 reporter ion scan that is populated with isotopically labeled iTRAQ reagent fragments following high-energy collision of the most intense MS2 fragments [80].

Chapter 2 Rationale, Hypothesis and Objectives 2.1 Rationale for the ZIP study

The most severe phenotype linked to PrP deficiency to date, aside from the inability to infect KO mice, was reported for one of two PrP orthologs in zebrafish, termed PrP1 [55]. Morpholino-based knockdown of PrP1 had shown to stop zebrafish embryogenesis at the gastrula stage, which involves cellular rearrangements relying on EMT [55]. Strikingly a similar phenotype has been observed in ZIP6-deficient zebrafish embryos [55]. Following up on these independent findings, we have previously reported a pivotal role of PrP in EMT. More specifically, in the NMuMG mouse cell model, which is a well-known epithelial cell line that robustly undergoes TGFβ1- induced EMT, PrP ablation disrupted EMT [4]. We also showed that during EMT PrP expression levels increased more than tenfold [4]. Our subsequent mechanistic studies revealed a link between PrP and NCAM1, which is one of the main interactors of PrP. Moreover, global proteome analyses, following TGFβ1 treatment of wild-type NMuMG cells, demonstrated that NCAM1 was amongst the proteins whose levels were profoundly increased during EMT and that PrP-deficient cells failed to show this upregulation to the same extent. A parallel line of study comparing wild-type and PrP KO NMuMG cells showed that PrP controls polysialylation of NCAM1 during EMT and that NCAM1 fails to undergo polysialylation in PrP-deficient clones we derived from this cell model.

A subsequent ZIP6 interactome study in TGFβ1-treated NMuMG cells in our lab identified ZIP10, NCAM1 and calreticulin to be the most enriched protein interactors of ZIP6, with ZIP10 being the main interactor. It is noteworthy that ZIP6 and ZIP10 had also been previously suggested to interact and have functional similarities [54,86]. Additional investigation into the relationship of ZIP6 and ZIP10 using both NMuMG and N2a cell models revealed that the expression of these two zinc transporters is closely co-regulated. Moreover, we were able to document that they form a functional heteromer that interacts with NCAM1 [55]. Given the evolutionary and functional links between the ZIPs and PrP, as well as the independently reported similarities in gastrulation arrest phenotypes of ZIP6 and PrP1 in zebrafish upon morpholino knockdowns, we investigated

whether PrP had inherited some of its biological functions from its ZIP ancestors. More specifically, given that PrP affects NCAM1 polysialylation in NMuMG cells by controlling polysialyltransferase ST8SIA2 transcript levels, we looked into whether ZIPs could similarly regulate this PTM in mammalian cell models. Thus, we hypothesized that ZIPs, like PrP, can influence the NCAM1 polysialylation machinery. Analyzing the levels of NCAM1 and its polysialylated form by western blotting revealed surprisingly that although ZIP6-deficient NMuMG cells also exhibit reduced NCAM1 expression, unlike PrP deficiency, ZIP6 ablation did not prevent NCAM1 polysialylation [55].

When we next compared the NCAM1 interactome in wild-type and ZIP6 KO NMuMG cells, we observed that ZIP6 ablation caused a reduction in NCAM1 phosphorylation. More specifically, levels of a specific phosphorylation event within a cluster of potential phosphoacceptor sites in the cytoplasmic, C-terminal domain of the longest isoform of NCAM1 (AA 945 to 974) were dramatically underrepresented in ZIP6 KO cells. Closer inspection of the phosphorylation consensus motif in this region pointed toward glycogen synthase kinase 3 (GSK3) as the most likely kinases responsible for NCAM1 phosphorylation at this site [55]. The notion that this phosphorylation event might be attributed to GSK3 gained independent support from a ZIP6 interactome investigation we undertook, which revealed the co-enrichment of both GSK3A and GSK3B kinase with ZIP6. Taken together, these observations led to the hypothesis that the ZIP6- ZIP10 heteromer binds and directs GSK3 kinases to NCAM1. Because these developments coincided with me joining the group, one of my initial tasks became to test this hypothesis by determining if the ZIP6-dependent NCAM1 phosphorylation is driven by GSK3 kinases on the basis of functional phosphorylation assays.

2.2 Rationale for the PrP interactome study

As mentioned above, previous work in our lab regarding the role of PrPC in EMT had led to the discovery that PrPC affects polysialylation of NCAM1 by regulating the transcript levels of ST8SIA2, one of two polysialyltransferases that can attach polysialic acids to N-glycans of NCAM1. Surprisingly, we observed that PrPC differentially regulates NCAM1 polysialylation in different cell models; thus, whereas in NMuMG cells ablation of PrPC prevented this PTM during

EMT, in PrPC-deficient C2C12 cells the opposite trend, namely a massive upregulation of NCAM1 polysialylation was observed [4]. This observation invoked our interest to dissect if cell type- specific differences in the molecular environment of PrPC might exist that could underlie these differences. In one scenario, the complete opposite effects of PrPC removal in NMuMG and C2C12 cells, might reflect a reality whereby PrPC could be part of separate microdomains in these cell models. Alternatively, we may find that PrPC is embedded in identical molecular environments and that differences in downstream signaling circuitry translates in distinct molecular phenotypes. Learning about these alternative realities is not merely an academic exercise but it is likely that cell-type specific differences in the molecular environment of PrPC, if they exist, could contribute to well-known differential susceptibilities of existing cell lines to prion infection.

Until now, the heterogeneity of the cellular paradigms and protocols used in the prion research field, precluded to answer to which extent differences in reported PrPC interactions reflected differences in methodology versus biology. Despite numerous studies that explored the function of PrPC in a wide range of experimental paradigms, no clear consensus has emerged. Thus, this thesis project was designed to address this unmet need by carrying out a first in depth comparative analysis of the PrP interactome based on four cell models of interest to the field. We hypothesized that insights into the molecular environment of PrPC will provide novel therapeutic angles by modulating PrPC levels, preventing its conversion to PrPSc, or interfering with PrPC-dependent toxicity.

Chapter 3 ZIP6-mediated NCAM1 Phosphorylation

Please note that this chapter incorporates material published in the following article [87] :

Brethour D, Mehrabian M, Williams D, Wang X, Ghodrati F, Ehsani S, et al. A ZIP6-ZIP10 heteromer controls NCAM1 phosphorylation and integration into focal adhesion complexes during epithelial-to-mesenchymal transition. Sci Rep. 2017;7: 1–19. doi:10.1038/srep40313

Candidate’s Role: Performed the NCAM1 immunoprecipitation experiments, analyzed the data and edited the manuscript.

Summary:

The prion protein evolved from the subbranch of ZIP metal ion transporters comprising ZIPs 5, 6 and 10, raising the prospect that the study of these ZIPs may reveal insights relevant for understanding the function of PrP. Building on data which suggested PrP and ZIP6 are critical during EMT, we investigated ZIP6 in an EMT paradigm using CRISPR-generated ZIP6 knockout cells, mass spectrometry technology and bioinformatic analyses. Reminiscent of PrP, ZIP6 levels are five-fold upregulated during EMT and the protein forms a complex with NCAM1. ZIP6 also interacts with ZIP10 and the two ZIP transporters exhibit interdependency during their expression. Additionally, ZIP6 contributes to the integration of NCAM1 in focal adhesion complexes but, unlike cells lacking PrP, ZIP6 deficiency does not abolish polysialylation of NCAM1. Instead, ZIP6 mediates phosphorylation of NCAM1 on a cluster of cytosolic acceptor sites. Substrate consensus motif features and in vitro phosphorylation data, point toward GSK3 as the kinase responsible, and interface mapping experiments identified histidine-rich cytoplasmic loops within the ZIP6/ZIP10 heteromer as a novel scaffold for GSK3 binding. Our data suggests that PrP and ZIP6 inherited the ability to interact with NCAM1 from their common ZIP ancestors but have since diverged to control distinct posttranslational modifications of NCAM1.

3.1 Introduction

In this chapter the focus is mostly on a specific portion of a 2017 study by our group, which examines the relationship between ZIPs and NCAM1 [87]. As discussed previously, PrP evolved from the ZIP family of ion transporters [49] and is particularly similar to the LIV1 subfamily of ZIPs, which include ZIPs 5, 6 and 10 [48]. NCAM1 on the other hand, is a close PrP interactor and both proteins have been shown to be involved in EMT [4,57], with PrP controlling the levels of polysialylated NCAM1 during EMT and related cellular reprogramming events.

Together, this body of work has raised several interesting questions: (1) Did PrP inherit its intimate involvement in a biology that modulates the expression of NCAM1 and controls its polysialylation from its ZIP6-like ancestor? and, if so, (2) are ZIP6 and PrP equivalent in this regard, or have they acquired specialized roles with respect to their influence on NCAM1?

To address these questions, we undertook a ZIP6 analysis in the EMT paradigm of NMuMG cells that showed ZIP6 transcript and protein levels are both dramatically upregulated upon treatment with TGFβ1. Moreover, in a follow-up interactome study, we observed that ZIP6 forms a heteromeric complex with ZIP10, whose predominant interactor is NCAM1. Subsequent in-depth analyses of NCAM1 interactors revealed a critical role for ZIP6 in the assembly of NCAM1-bound focal adhesion complexes. However, we also documented a minor influence of ZIP6 on the levels of polysialylated NCAM1 and instead uncovered a novel signaling module that has ZIP6/10 collaborate closely with GSK3B in its influence on NCAM1. When I took up my training in the laboratory, the key question whether the ZIP6/ZIP10 heteromer forms a novel hub that can sequester GSK3B to NCAM1 and thereby controls its direct phosphorylation by this kinase at a specific site had not been addressed. I therefore will address in this chapter this question before moving to the main body of my thesis, which naturally followed from differences in NCAM1 posttranslational effects we observed in separate experimental paradigms.

3.2 Results

3.2.1 In vitro phosphorylation of the longest NCAM1 isoform by GSK3B

Observations thus far raised the intriguing possibility that the ZIP6-ZIP10 heteromer may serve as a scaffold for binding and directing GSK3 kinases to their NCAM1 substrate. Note that the higher the number of peptides quantified and matched to a given protein, the more confidently and with higher statistical power, the differences in relative abundance levels can be assigned in this methodology. Thus, despite the enrichment levels of GSK3A and GSK3B in ZIP6-specific co-IP eluates by ~20% compared to the negative controls (Fig. 3.1a), this difference was significant and distinct from the peptide distribution of, for example, ATP synthase coupling factor 6 (ATP5J), a mitochondrial protein, which was observed at equal levels among all samples. This suggested that its binding to the affinity matrix had exclusively been through non-specific interactions (Fig. 3.1b).

To validate whether the longest NCAM1 isoform is indeed a GSK3 substrate, we examined whether recombinant GSK3B can phosphorylate immunoprecipitated NCAM1 from NMuMG cells after 48 h induction of EMT by TGFβ1. This design was selected over an alternative strategy, whereby both the kinase and putative substrate would be obtained from recombinant sources, because GSK3 typically requires a priming phosphorylation by casein kinase I, that would otherwise not be present in recombinant NCAM1. Corroborating the mass spectrometry results, this experiment revealed that a 180 kDa protein, corresponding in size to the longest isoform of NCAM1, was indeed the prime phosphorylation target of recombinant GSK3B in the NCAM1- directed immunoprecipitation eluate (Fig. 3.1c and d). Taken together, these data provided orthogonal evidence for the notion that GSK3B can preferentially phosphorylate the longest NCAM1 isoform in biological material relevant to this study.

Figure 3.1. In vitro phosphorylation of longest NCAM1 isoform by GSK3B. (a and b) Box

plots showing relative quantitations of glycogen synthase kinase 3 beta (GSK3B) and ATP synthase coupling factor 6 (ATP5J) in NCAM1 interactomes generated with wild-type or ZIP6 KO cells. Whereas GSK3B protein levels in the NCAM1 interactome correlated directly with the presence or absence of ZIP6, ASCF6 protein level ratios were similar in all samples, consistent with unspecific binding of this protein to the affinity capture matrix. (c and d) Evidence of NCAM- 180 in vitro phosphorylation by GSK3B. (c) Immunoprecipitation of NCAM1 from cellular extracts of NMuMG cells, which had been induced to undergo EMT by 48 h addition of TGFΒ1 to the cell culture medium. The OB11 antibody used in this experiment for immunoprecipitation and Western blot detection binds to a cytosolic epitope shared by NCAM-140 and NCAM-180, which gave rise to the characteristic band pattern in Lanes 5 and 6. Note that due to its relatively higher abundance in eluate fractions, signals for NCAM-140 exceeded maximum intensity levels in lanes 5 and 6, leading to a partially inverted ‘white’ signal. (d) Autoradiographic analysis of in vitro GSK3B phosphorylation of NCAM1immunoprecipitation eluates seen in Panel ‘c’. Note that although the signal detected at 180 kDa, corresponding to NCAM-180, has not been validated to consist of phosphorylated NCAM1, the Coomassie stain of the NCAM1 IP eluate fractions shown in Panel ‘c’ revealed no signals in the high mass region of the gel that could not be attributed to the expected NCAM-180 or NCAM-140 bands.

3.3 Discussion

Two independent observations placed a spotlight on GSK3 kinases in this work: First, the ZIP6- ZIP10 heteromer was observed to co-immunoprecipitate with both GSK3 paralogs. Second, we discovered a ZIP6-dependent phosphorylation site within the cytoplasmic domain of the longest isoform of NCAM1 that bears striking similarities to a previously known GSK3 substrate. Our data place GSK3 paralogs in close proximity to the site of ZIP6-ZIP10-mediated zinc influx— presumably one of only few cytoplasmic locations at which free zinc ion concentrations may exceed the generally very low levels of free zinc in the cytoplasm. This observation may be relevant in light of prior work by others which documented that the ability of GSK3 to phosphorylate its substrates is inhibited in the presence of free zinc [88]. The authors documented that there is specificity to this zinc-mediated GSK3 inhibition by showing its reliance on zinc, as opposed to other divalent cations, and by establishing that the zinc inhibition characteristic does

not extend to CDK2, a closely related kinase. It has long been known that NCAM1 can be phosphorylated within its intracellular domain by casein kinase 1 (CK1) and GSK3 but in these early studies the phospho-acceptor site was not determined [89]. Subsequent attempts to map the NCAM1 phosphorylation sites highlighted initially a domain in juxtaposition to the inner face of the plasma membrane but also a putative serine 761 phospho-acceptor site that conforms to the GSK3 consensus motif [90]. The design of the latter phospho-site mapping experiments precluded the detection of phosphorylation sites observed in this study. This was because analyses were limited to a truncated NCAM1 expression construct comprising the transmembrane and cytoplasmic domains of NCAM1 but lacking the Exon 18 coded alternatively spliced insertion that is only present in the longest NCAM1 isoform. However, a more recent global phospho-site analysis of developing mouse brain samples independently mapped amino acids 946 and 958 (NCBI accession number: NP_001106675) as NCAM1 phospho-acceptor sites [91]. These sites are identical to the first and fourth NCAM1 phosphorylation sites within the phosphorylation cluster (amino acids 945-974) that was repeatedly sequenced and quantified in this study. We subsequently validated the preferential GSK3B phosphorylation of NCAM-180 by in vitro phosphorylation assay, using immunoprecipitated NCAM-140 and NCAM-180 as substrates. Curiously, CRMP2, a member of the CRMP family, which is known to carry a highly similar GSK3 phosphorylation motif (Fig. 3.1i), had previously also been shown to be a binder of NCAM1 [92], but was not detected in the interactome dataset produced in this study.

What might be the functional consequences of the GSK3-dependent phosphorylation of NCAM1? One commonly observed scenario would see phospho-occupancy at this site alter NCAM1 interactions with phospho-serine/threonine (pSer/Thr) binding modules on other proteins. In fact, binding of NCAM1 to 14-3-3 proteins, the first signaling molecules recognized to engage in pSer/Thr-dependent interactions with other proteins, was also observed to be ZIP6-dependent in this study, making them excellent candidates for this scenario. While the broader physiological consequences are not known, a previous report established that inhibition of GSK3 can prevent NCAM1-induced neurite outgrowth [93].

3.4 Methods

3.4.1 Sample preparation for immunoprecipitation

Three biological replicates of NMuMG wild-type and ZIP KO clones, undergoing EMT, were used for the NCAM1 immunoprecipication. TGFβ1 (240-B; R&D Systems, MN, USA) was added for 48 hrs at a concentration of 6.4 ng/mL and replenished with fresh medium after 24 hours. After 48 hrs of control or treatment conditions, medium was removed, cells were washed with ice cold PBS, and crosslinking was completed with a 15 min incubation with 2% formaldehyde in PBS. The formaldehyde was removed, and the reaction was quenched with a 10 min incubation with 125 mM glycine in PBS. Cells were again washed with ice cold PBS before undergoing lysis in an ice cold buffer consisting of 5 mM EGTA, 10% glycerol, 1% sodium deoxycholate, 1% NP-40, 150 mM HEPES (pH 8.0) and 1× Complete Protease Inhibitor Cocktail (11836170001; Roche). Insoluble cellular debris was cleared by centrifugation for 30 min at 4000 RPM and 4 °C.

3.4.2 Protein immunoprecipitation

Immunoprecipitation with NCAM1 specific antibody was completed with Protein G sepharose (17-0618-01; GE Healthcare) beads. Beads were transferred to a microcentrifuge tube and washed twice with ultrapure water and twice with PBS before the addition of the respective antibody, which had been diluted in PBS to fill the tube. The smaller-scale NCAM1 immunoprecipitations for in vitro GSK3B phosphorylation was undertaken with the NCAM1-directed antibody OB11. The bead/antibody mixture was then gently agitated on a turning wheel for 4 hrs at room temperature (RT), before being equally divided into fresh tubes for individual samples. Beads were allowed to settle and the excess liquid was removed before protein samples were added. The samples were gently agitated on a turning wheel overnight at 4 °C, then washed thrice with 5 mM EGTA, 10% glycerol, 1% sodium deoxycholate, 1% NP-40, 150 mM HEPES (pH 8.0). Detergents were removed with two consecutive washes of 10 mM HEPES (pH 8.0), before samples were transferred to lo-bind 0.5 mL microcentrifuge tubes. Proteins were then eluted by acidification with 0.2% trifluoroacetic acid, 20% acetonitrile (pH 2.0).

The mapping of the GSK3 binding site to internal loop domains made use of affinity matrices based on the monoclonal capture antibody HA.11 (901501; BioLegend, CA, USA) bound to Protein G-sepharose (17-0618-01; GE Healthcare). The cell lysates were initially incubated with the antibody overnight and then added to beads for an overnight capture with gentle agitation at 4°C. The proteins were at the end eluted with sample buffer and heating at 45 °C for 10 minutes.

3.4.3 Active kinase assay

NCAM1 protein was immunoprecipitated with the OB11 antibody as described above and was in vitro phosphorylated with recombinant GSK3B while still attached to Protein G beads. More specifically, 4 µL (9 ng) of active GSK3B (catalog number 14-306, lot number WAA0024-B, Millipore, Burlington, Canada), 10 µL of hot ATP buffer, composed of 1 volume of [γ-32P]ATP (0.01 mCi) (catalog number NEG 502Z, Perkin-Elmer Inc., Woodbridge, ON, Canada) and 9 volumes of 75 mM MgCl2 and 500 µM ATP (catalog number A6559, Sigma-Aldrich, Oakville, ON, Canada), were added to 15 µL of NCAM1 immunoprecipitate to give rise to a reaction mix of 40 µL that was buffered by 8 mM MOPS, pH 7.0 and supplemented with 0.2 mM EDTA. After a 10 min incubation at 30 °C the reaction was stopped by the addition of Laemmli sample buffer. The analysis was undertaken in triplicate for wild-type and ZIP6 knockout cell extracts. As negative controls served samples that differed by the omission of GSK3B. Next, Protein G beads were pelleted by centrifugation, the superna- tant containing the eluted proteins boiled for 15 min and loaded onto a large format isocratic (7.5%) SDS-PAGE gel that was cast in a Hoefer gel cassette. Incorporated radioactivity was revealed by overnight exposure of the SDS-PAGE gel to a double-emulsion X-ray film (catalog number Z363006, Carestream Kodak BioMax MS film, Sigma-Aldrich).

Chapter 4 The Prion Protein is Embedded in a Molecular Environment that Modulates Transforming Growth Factor β and Integrin Signaling

Please note that this chapter incorporates material published in the following article [69] :

Ghodrati F, Mehrabian M, Williams D, Halgas O, Bourkas MEC, Watts JC, et al. The prion protein is embedded in a molecular environment that modulates transforming growth factor β and integrin signaling. Sci Rep. 2018; 8654. doi:10.1038/s41598-018-26685-x

Candidate’s Role: Conceived, designed and performed the experiments. Analyzed the data, wrote and edited the manuscript.

Summary:

At times, it can be difficult to discern if a lack of overlap in reported interactions for a protein-of- interest reflects differences in methodology or biology. In such instances, systematic analyses of protein-protein networks across diverse paradigms can provide valuable insights. Here, we interrogated the interactome of the prion protein (PrP), best known for its central role in prion diseases, in four mouse cell lines. Analyses made use of identical affinity capture and sample processing workflows. Negative controls were generated from PrP knockout lines of the respective cell models, and the relative levels of peptides were quantified using isobaric labels. The study uncovered 26 proteins that reside in proximity to PrP. All of these proteins are predicted to have access to the outer face of the plasma membrane, and approximately half of them were not reported to interact with PrP before. Strikingly, although several proteins exhibited profound co-enrichment with PrP in a given model, except for the neural cell adhesion molecule 1, no protein was highly enriched in all PrP-specific interactomes. However, Gene Ontology analyses revealed a shared association of the majority of PrP candidate interactors with cellular events at the intersection of transforming growth factor β and integrin signaling.

4.1 Introduction

Relatively little is known about how interactions of a given protein differ across cell models. Although there is no shortage of proteins whose binding partners have been studied in more than one paradigm, chances are such studies were done by separate investigators with different methodologies, precluding robust conclusions on the confounding effects of the paradigm itself.

In particular, proteins like the cellular prion protein, which lack catalytic domains and exhibit widespread expression [11,29,94], are prone to escape robust functional assignments. PrPC is central to the pathogenesis of prion diseases [23] and has been proposed to also act as a critical cell surface receptor in AD [68], raising the expectation that insights into the function of PrPC will provide useful angles for understanding the molecular underpinnings of these diseases.

To this end, the molecular interactions of PrPC have repeatedly been characterized [47,57,95] and many proteins have been reported to interact with PrPC in separate studies, including the laminin receptor precursor [96], the neural cell adhesion molecule 1 [57], the amyloid precursor like protein-1 [97], and the stress-inducible protein 1 [98]. Rather than suggest a common theme, this line of investigation has led to many hypotheses regarding the role of PrPC. Although it is to be anticipated that some of the reported interactions will not stand the test of time, other reasons for the diversity of observations need to be considered, including the likely existence of cell type- specific interactions.

In one study undertaken with neuroblastoma cells, ZIPs, a family of Type-III transmembrane proteins known to import zinc and other divalent cations into the cytosol, were initially observed as PrPC interactors [47]. This work then spurred the discovery that prion genes evolved from an ancient ZIP transporter and are members of the ZIP gene family, which comprises seventeen genes in humans [49,99]. Because PrP is homologous to the ectodomain present in a subset of ZIPs, studying the physiological function of this ectodomain may provide additional hints regarding PrP’s function. Interestingly, the deficiency of PrP, Zip6 or Zip10 causes a rare and phenotypically indistinguishable gastrulation arrest phenotype in zebrafish, apparently due to the ablation of a morphogenetic program known as epithelial-to-mesenchymal transition [52–54].

We recently documented that the expression levels of the aforementioned ZIPs, PrP, and NCAM1 are several-fold upregulated during EMT in mammalian cells [4], consistent with the interpretation that the interactions these proteins engage in change over time and depend on the cell lineage characteristics of the model. A hint that there may be additional differences in PrP interactions when comparisons are done across models came from an observation of model-dependent proteome shifts in PrP-deficient cells [100]. Not only did PrP-deficiency in distinct cell models cause the levels of members of the Marcks protein family to shift in opposite ways but it also prevented NCAM1 polysialylation in one cell model, yet caused a robust increase of this specific PTM in another model [4].

To address if these phenotypic differences reflect distinct, immediate PrP interactions or depend on downstream signaling, we undertook deep, quantitative PrP interactome comparisons in four cell models that made use of in vivo crosslinking and capitalized on recombinant Fabs for the selective capture of endogenous PrP-containing protein complexes. We identified cell lineage- specific sets of PrP interactors, including several novel interactors and a membrane protein of unknown function, whose PrP-dependent capture was supported by high-confidence peptide-to- spectrum assignments and quantitation. All of the 26 candidate PrP interactors we identified are known to be embedded in the plasma membrane or exist in the lumen/extracellular matrix. We demonstrate that NCAM1 is the only transmembrane protein that is a universal and robust interactor of PrP across the four models. The data comprise examples of homologous proteins interacting with PrP in a cell model-specific manner. Finally, we highlight that a majority of PrP interacting proteins observed in this study are known to play roles in EMT, either by acting as TGFβ1 signaling modulators, by facilitating the formation of NCAM1-dependent focal adhesion complexes, or by their association with integrin-mediated downstream cell signaling.

4.2 Results

4.2.1 Design of comparative PrP interactome analysis in four mouse cell models

The study made use of four mouse cell lines from which we had previously derived PrP knockout clones by CRISPR-Cas9 technology (Fig. 4.1a) [100]. The parental wild-type cell lines are familiar

to the prion research community due to their distinct properties with regard to PrP: 1) NMuMG cells exhibit a more than five-fold increase in their PrP protein levels when EMT was induced by the addition of TGFβ1 (attempts to infect these cells with prions have been unsuccessful but were also not exhaustive) [58]; 2) C2C12 cells are the only muscle cell model currently known to be susceptible to prion infection [71,101]; 3) N2a neuroblastoma cells may be the most often used cell model in prion research and can readily be infected with mouse-adapted Rocky Mountain Laboratory (RML) prions; and 4) CAD5 catecholaminergic cells exhibit susceptibility to infection with several prion strains [102]. To stabilize existing protein-protein interactions, cells were subjected to mild formaldehyde crosslinking prior to cell lysis (Fig. 4.1b). PrP-containing complexes were affinity-captured using a recombinant anti-PrP antigen binding fragment (Fab), designated as D18, that is known to bind to a non-linear epitope (comprising mouse PrP residues 133-157) within the globular domain of PrP [79]. Affinity-capture eluates were processed using a workflow that facilitated the relative quantitation of peptides in three biological replicates and three control samples obtained from wild-type and PrP knockout cells, respectively. The side-by- side binning of peptide-to-spectrum matches on the basis of cross-correlation (X corr) values computed by the SEQUEST score function [103] revealed similar stratifications for the four PrP interactome analyses, indicating that the respective datasets were comparable in regards to a key quality control benchmark and their depths of coverage (Fig. 4.1c).

Figure 4.1 Design of comparative PrP interactome study. (a) Models used in this study. (b) Flow-chart depicting workflow of quantitative interactome analyses. Note that for the NMuMG cell model, one additional step was inserted into the workplan, namely a 48-hour addition of TGFβ1 to the cell culture medium, which causes the cells to acquire mesenchymal morphology. (c) Similar SEQUEST X corr stratifications of the four cell type-specific PrP interactome datasets indicated comparable data quality

4.2.2 Comparison of PrP interactome analyses across models

A comparison of PrP-specific western blot signals obtained for cell lysates before (input) and after (unbound) the affinity capture step established that more than 50% of total PrP was captured (Fig. 4.2a, lanes 1 and 2). As expected, eluate fractions exhibited strong PrP-specific signals whose distribution matched the anticipated pattern, i.e., were comprised of low mass uncrosslinked bands characteristic for PrP and high mass crosslinked PrP-containing smears that were particularly concentrated in the 200-250 kDa range. When the same blot membranes were subsequently stained with Coomassie, equal total protein levels were detected in ‘input’ and ‘unbound’ samples. Eluate fractions contained protein levels below the Coomassie detection limit, except for one band whose apparent MW matched the known D18 mass (Fig. 4.2a, lanes 5-10). The band could be seen at equal intensity levels in all biological replicates of wild-type and control samples, consistent with the interpretation that it indeed represented small amounts of the Fab, which had detached during the pH 1.9 elution step. The subsequent mass spectrometry analysis confirmed the successful enrichment of PrP (Fig. 4.2b, Supplementary Fig. S4.1).

A total of 26 proteins were shortlisted as PrP candidate interactors. Gene Ontology enrichment analyses flagged categories within the ‘Molecular Function’, ‘Biological Process’ and ‘Cellular Component’ classes that made sense for a GPI-anchored molecule like PrP (Fig. 4.2c). For example, a cell-to-matrix binding category, namely binding to laminin was the most enriched ‘Molecular Function’ subcategory. Similarly, 23 of the candidate interactors had a prior annotation that identified them as members of the ‘Cellular Component’ subcategory ‘extracellular space’ (see below for details on membrane topology). Finally, ‘Biological Process’ annotations of these proteins are consistent with the notion that PrP is embedded in a membrane domain that serves as

a signaling hub, with the subcategory ‘positive regulation of response to stimulus’ being the most overrepresented annotation.

Figure 4.2 Validation of successful technical execution of quantitative interactome analysis. Analyses of benchmarks of PrP co-immunoprecipitation, iTRAQ quantitation and GO enrichment. (a) Western blot validation of co-immunoprecipitation of endogenous PrPC from NMuMG cell extracts. Strong depletion of PrP-related signals in the unbound fraction and its robust enrichment in wild-type eluate fractions. Asterisks denote weakly detected cross-reactive bands. (b) Box plot depicting selective detection of PrP in wild-type eluate samples of CAD5 cell-derived PrP immunoaffinity captures but not in negative control PrP knockout eluates (see Supplementary Figure S4.1. for the respective PrP box plots from NMuMG, C2C12 and N2a cells). The box plot depicts in log2 space enrichment ratios of individual PrP peptides used for quantitation. The computed Median peptide ratios and Inter Quartile Ranges (IQR) are shown above the graph. Note that a subset of PSMs (indicated with red circles) were automatically eliminated from the quantitation, either because their identification was redundant or as a consequence of mass spectrometry profiles underlying their identification not passing stringency thresholds. In this and other box plots in this report relative protein levels are depicted as ratios, with ion intensities of the heaviest isobaric labels within multiplex analyses (representing one of 3 PrP knockout biological replicates) serving as the reference (denominator). (c) GO enrichment analyses of the 26 shortlisted PrP candidate interactors identified in this study.

Although each of the four PrP interactome datasets comprised more than 200 proteins which had passed confidence thresholds for identification, approximately 80% of the proteins in each dataset were observed at levels that did not differ in wild-type versus PrP knockout samples, thereby revealing them to be non-specific interactors of the affinity matrix. An example which showcases proteins in this broad non-specific binder category is glyceraldehyde 3-phosphate dehydrogenase (Gapdh). This protein was robustly identified in all samples on the basis of dozens of peptide-to- spectrum matches (PSMs), yet was readily identified as a non-specific binder by its similar enrichment (< 1.5-fold) in all samples, including in eluates derived from PrP knockout samples (Supplementary Fig. S4.1). Of note, rather than manifesting as a hindrance, proteins in the non- specific binder category served as a useful internal control in these analyses that further validated the existence of comparable capture conditions across all samples. A closer look at the datasets revealed one additional group of highly abundant proteins, whose enrichment behavior suggested

them to be non-specific interactors. Like Gapdh, these proteins were observed in all samples, yet in contrast to Gapdh, they exhibited inconsistent and, when encountered, only modest co- enrichment with PrP. In this category fell a subset of ribosomal proteins, histones, tubulins and 14- 3-3 proteins. These proteins tended to exhibit the same trend in a given sample. For example, in the second (wt2) versus third (wt3) biological replicate of the PrP interactome analysis from NMuMG cells, most of these proteins were seen at levels that were consistently lower or higher, respectively, than those observed in the corresponding PrP knockout control sample (Supplementary Table S4.1). The distribution of these proteins suggested them to bind mostly non- specifically to the affinity matrix but also indicated that their binding was more responsive to sample-to-sample variations than the aforementioned bona fide non-specific interactors.

Based solely on their iTRAQ-enrichment characteristics, the 26 PrP candidate interactors could be sorted into three categories (Table 4.1), namely:

I) Proteins observed in all or a subset of the four interactome analyses that exhibited consistent and intermediate (1.5- to 7.5-fold) co-enrichment with PrP: This category encompassed the largest number of candidate PrP interactors (18 proteins), including protein disulfide isomerase (P4hb), galectin-1 (Lgals1), calreticulin (Calr), two gene products of the histocompatibility antigen gene cluster (H2-K1 and H2-D1), transmembrane emp24 domain-containing protein family members 2, 9 and 10 (Tmed2, 9 and 10), the sodium/potassium-transporting ATPase (Atp1a1) and the transferrin receptor protein 1 (Tfrc). The intermediate enrichment levels these proteins exhibited suggested that they bound, in addition to PrP, to the affinity matrix or other proteins that were non- specifically captured.

II) Proteins observed in a subset of samples that exhibited consistent and profound (> 7.5-fold) co- enrichment with PrP (7 proteins): This category included the ZIP10 zinc transporter (Zip10), 4F2 cell-surface antigen heavy chain (Slc3a2), the large neutral amino acid transporter small subunit 1 (Slc7a5), basigin (Bsg), endothelin-converting enzyme 1 (Ece1), thrombospondin type-1 domain containing protein 7A (Thsd7a), and insulin-like growth factor-binding protein 5 (Igfbp5).

III) Proteins observed in all samples that exhibited consistent and profound co-enrichment with PrP: The neural cell adhesion molecule 1 (NCAM1) was the only protein in this category.

Table 4.1. Curated list of PrP candidate interactors organized by cell type (please see Supplemental Table S4.1 for a complete account of shortlisted candidate interactors).

4.2.3 Cell type-specific effects of PrP knockout on the global proteome reflect its molecular interactions

We next analyzed cell type-specific differences of PrP candidate interactors (Fig. 4.3a), making use of a coordinate system that captured if PrP interactors were observed in a given cell model and also considered their cell model-specific levels of co-enrichment with PrP. Viewed in this manner, it was apparent that not only the identity of PrP interactors but also their levels of enrichment were most similar in N2a and CAD5 cells, with Calr, Atp1a1, NCAM1, Slc3a2 and Slc7a5 ending up in proximity to a trend line that bisected the quadrant bounded by the coordinate axes assigned to these cell models, a finding congruent with the shared neuronal origins of these cell models.

Often, next neighbor relationships of proteins are reflected in mutual effects on their expression levels. We therefore were curious to understand if and how the presence or absence of PrP might affect the steady-state expression levels of the proteins it is surrounded by in a given cell model. To answer this question, we were able to lean on data we had collected in a previous study that explored how PrP knockout affects the global proteome in an overlapping set of cell models (Fig. 4.3b) [100]. The data from this earlier work were based on three of the four cell models used in this study (i.e., NMuMG, C2C12 and N2a cells, however, instead of CAD5 it had explored 1C11 neuroectodermal cells) and had led to the relative quantitation of >1,500 proteins, per cell model, whose identification passed a 95% confidence threshold, including a dozen PrP candidate interactors revealed in the current study. Naturally, we were particularly interested in the possible influence of a PrP deficiency on the proteins we had assigned to categories II and III on account of their PrP-specific co-enrichment. The steady-state levels of four of these proteins, namely NCAM1, Slc3a2, Slc7a5, and Bsg, were quantified in all cell models on the basis of more than six independent iTRAQ signature ion ratios (counts >6) (Fig. 4.3b). As reported above, Slc3a2, Slc7a5 and Bsg were observed to co-immunoprecipitate with PrP on the basis of robust quantitations (counts >6) in N2a cells but not in NMuMG or C2C12 cells (Table 4.1). Interestingly, this cell model-specific relationship was reflected in the global proteome data, which documented that the

steady state levels of these proteins were pronouncedly perturbed in response to the PrP knockout only in the N2a cell model, and were only mildly affected in NMuMG cells or C2C12 cells (Fig. 4.3b). This correlation was observed despite the fact that the steady-state levels of these proteins across the cell models were similar. Interestingly, the direction of change observed in N2a cells was not consistent, i.e., whereas Bsg steady-state levels increased in the absence of PrP, levels of Slc3a2 and Slc7a5 were diminished. Taken together, this result corroborated the notion that the presence of PrP has a modulating effect on the steady-state levels of proteins in its immediate proximity

Figure 4.3. The molecular environment of PrP is cell model-specific and comprises several novel candidate interactors. (a) Graph depicting relative enrichment levels of candidate PrP interactors by cell type. The x-coordinate of each protein is its average WT/PrP KO ratio in CAD5 minus its average WT/PrP KO ratio in C2C12. Each PrP interactor is represented by a cross, whose position in the coordinate system is determined by its relative enrichment in the four cell models, i.e., the y-coordinate represents the average WT/PrP KO ratios observed in N2a versus NMuMG cell PrP interactomes. The average WT/PrP KO ratios used were normalized against the average WT/PrP KO ratio for PrP in the same cell line. The cell lines in which a given protein was quantified are indicated by shading in the corresponding cross arms, proteins identified in all cell types are therefore represented by fully shaded crosses while proteins identified in only one cell type are represented by crosses with one shaded arm. At least one protein was enriched to an exceptional degree in each cell line. Eng was quantified only in the N2a interactome, and Igfbp5 was similarly quantified only in the C2C12 dataset, each protein being second only to PrP in its level of enrichment. The figure demonstrates that the two neuron-like cell lines (N2a and CAD5) share several PrP interactors which were only weakly detected or undetected in NMuMG or C2C12 datasets. Conversely Ece1 and Cd109 were highly co-enriched with PrP in NMuMG and C2C12 datasets but not N2a or CAD5 PrP interactomes. (b) Global proteome analyses of wild-type versus PrP-deficient cells indicates that PrP depletion has reproducible effects on steady-state protein levels of shortlisted PrP candidate interactors within a given model but leads to inconsistent consequences of their relative abundance across cell models.

4.2.4 PrP selectively interacts with Ece1 and Tfrc dimers

In addition to several previously known PrP interactors (NCAM1, Slc3a2, Slc7a5, Bsg, P4hb, Itgb1, Lgals1, Tmem206, Gpc1, Calr, H2-K1 and H2-D1) [47], about half of the candidate interactors revealed in this work were not previously proposed to reside in immediate proximity to PrP. Amongst these were a few proteins that highly selectively co-purified with PrP in a subset of cell models, including Ece1, Thsd7a and Igfbp5. In particular, Ece1 stood out by a remarkable 127 count of quantified PSMs in PrP-specific immunoprecipitation eluates from NMuMG cells. The protein was also detected, albeit to a lesser extent, in the PrP interactome derived from C2C12 cells but not from N2a or CAD5 cells.

Ece1 is a Type II transmembrane protein that is best known for its endoproteolytic conversion of inactive big endothelin-1 (big ET-1) to active endothelin-1 (ET-1) and belongs to the family of zinc-dependent neprilysin-related endoproteinases. A western blot analysis of selected formaldehyde crosslinked cell extracts detected the predominant monomeric Ece1 (isoform D) signals at their expected apparent MW of near 120 kDa [104] and also revealed an SDS-resistant dimer band (most likely stabilized by the crosslinking reagent) in NMuMG and C2C12 cells (Fig. 4.4a). Consistent with its absence from the N2a and CAD5 interactome datasets, the Ece1-directed antibody picked up no signals for this protein in these cell models (see also Supplementary Fig. S4.3). The box plot of quantified spectra assigned to Ece1 exhibited the expected PrP-dependent enrichment of this protein in NMuMG- and C2C12-derived datasets (Fig. 4.4b). Strikingly, western blot results of eluate fractions from the interactome analyses revealed PrP to have exclusively co-immunoprecipitated the crosslinked Ece1 dimer (not the more abundant monomer) (Fig. 4.4c). To validate this interpretation, we next boiled aliquots of the eluate fractions for up to 30 minutes in the presence of reducing agents (a method known to revert formaldehyde crosslinks) [57]. As expected, this treatment reverted the crosslink, thereby leading to the appearance of the Ece1 monomer band through an intermediate band, which we interpreted to constitute residual levels of Ece1-monomer crosslinked to PrP (see red arrowhead in Fig. 4.4d). If PrP was indeed crosslinked to dimeric Ece1 in the PrP affinity capture eluate fractions, we reasoned that it might be possible to document this by separating Ece1-specific signals at higher resolution, next to Ece1 dimer bands seen in PrP knockout cell extracts. Indeed, this approach confirmed that the PrP affinity-captured Ece1 dimer migrated at a higher MW than the Ece1 dimer seen in PrP knockout cells, consistent with the interpretation that PrP had formed an SDS-resistant interaction with the Ece1 dimer in this cell model (compare levels of black and green arrowheads in the right panel of Fig. 4.4d).

Figure 4.4. PrP interacts selectively with the Ece1 dimer, not its more abundant monomer. (a) Ece1-specific western blot analysis of cellular extracts generated from wild-type and PrP knockout cell models (each sample shown with two biological replicates). Cells were subjected to mild formaldehyde crosslinking prior to their harvest. Arrowheads indicate signals derived from monomeric and SDS-stable dimeric Ece1. Note that consistent with the identification of Ece1 as a PrP interactor in NMuMG and C2C12 cells, but not in N2a cells, the protein is not observed in the latter cell model. The bottom panel depicts a Coomassie stain of the western blot membrane. (b) Box plot depicting relative quantitation of Ece1 in PrP interactome datasets of in vivo formaldehyde crosslinked wild-type and PrP KO NMuMG and C2C12 cells. Please see legend to Fig. 4.2b for a detailed description of graph elements. (c) PrP co-immunoprecipitation led to the co-enrichment of a slow migrating Ece1 antibody-reactive band in wild-type (but not in PrP knockout cell) eluates, suggestive of a selective interaction of PrP with the Ece1 dimer. The asterisk denotes a non-specific cross-reactivity of the antibody, most likely toward a component of the affinity matrix that is indicated with an empty arrowhead in the Coomassie stain depicted in the bottom panel. (d) Formaldehyde crosslink reversal treatment (90 °C, for 10-30 minutes) causes high mass Ece1 antibody-reactive band to shift to faster gel migration at level of monomeric Ece1. Note that the panels on the left (Lanes 1-3) and right (Lanes 4-7) were generated with SDS-PAGE gel systems of different resolution. Portions of the Ece1 western blots that showed no signals were cropped at the bottom and the corresponding Coomassie stains in this panel were trimmed accordingly. Arrowheads provide signal interpretation as follows: black, Ece1 dimer-PrP complex; green, Ece1 dimer; red, Ece1 monomer-PrP complex; blue, Ece1 monomer.

The transferrin receptor protein 1 (Tfrc) is another protein that according to the mass spectrometry data presented in this study (Table 4.1) resides in immediate proximity to PrP only in the NMuMG model. In contrast to Ece1, Tfrc-directed immunoblot analyses revealed robust steady-state expression levels of this protein in cell lysates of all four models tested. Yet, an intense Tfrc immunoblot signal was only observed in PrP co-immunoprecipitation eluates from wild-type NMuMG cells (Supplementary Fig. S4.4). However, weak Tfrc signals were also observed in PrP- specific eluates from wild-type C2C12 and N2a cells, presumably reflecting a slightly higher sensitivity of western blot analysis over mass spectrometry-based detection for this protein.

Taken together, the Ece1 and Tfrc validation experiments indicated that the absence of a PrP- interacting protein in a cell model-specific interactome data set may reflect dramatic differences in steady-state levels (Ece1) or more subtle differences in the cellular expression of a PrP binder (Tfrc) that push the amounts of a given protein residing in immediate proximity to PrP below the level necessary for its detection.

4.2.5 Candidate PrP interactors exist in the secretory pathway or at the cellular membrane

One other way to validate interactome data is to assess the known or predicted cellular localization of proposed binders of a protein-of-interest. In light of the known, predominant association of PrP with raft-like domains at the plasma membrane [105], PrP would be expected to encounter other proteins during its biogenesis in the secretory pathway and at the plasma membrane. A UniProt- based survey of cellular sites predicted to accommodate the identified PrP interactors established all 26 candidates as known or predicted residents of the membrane or lumen (Fig. 4.5), a finding consistent with the initial GO ‘Cellular Component’ enrichment analysis (Fig. 4.2c). Five of the candidate interactors (Calr, P4hb, as well as Tmed 2, 9 and 10) are predominantly found in the ER or Golgi, where they play critical roles for quality control and transport. Two, namely H1-K1 and H1-D1, play roles in antigen presentation and are therefore, context-dependently found in the Golgi or at the plasma membrane. Amongst the remaining 19 proteins, there are three secreted proteins, six Type-I, three Type-II and five Type III transmembrane proteins, as well as two GPI- anchored proteins. Of the latter, glypican-1 (Gpc1) is a well-known PrP interactor, that has been shown to promote both PrP’s association with detergent-resistant raft domains and its conversion to PrPSc [106–108], but CD109 (Supplementary Fig. S4.2) was to our knowledge not previously reported to reside in proximity to PrP.

One approach to learn about the function of a protein-of-interest is to deduce it from the known functions of the proteins it partners with. A survey of known key functions of the 19 plasma membrane-associated PrP candidate interactors revealed their involvement in a multitude of biological activities, including cell adhesion, zinc import, endothelin-1 cleavage, transport of neutral amino acids, insulin-growth factor signaling, angiogenesis and iron uptake. Strikingly

however, most of these proteins are known to also play critical roles (see Discussion for details) in either TGFβ1- or integrin-signaling (Fig. 4.5b).

Figure 4.5 e e e C 1 1 1 3 3 3 3 3 3 3 lik lik lik F P P P P P P P P P P - - -

S S S Extracellular N Laminin G-like S S S S S S S TSP C C Thsd7 a W T T T T T T T T T T

a GF GF GF V E E E N Thbs1 Ncam1 Lgal1 Cd10 9 C N N TSP1 N CBD

-S sAce

S CBD TSP1 Igfbp5 N A2M Itgb1 TSP1

-S TG1 N C

S C

e TSP1 Eng IGFBP-N s a 2 N d ti

Gpc1 M

Zip10 Zip6 p

-S TSP1 N βI domain Tfrc e S

N N P l

e TSP1 a

e v

A2M c t CD9 8 i Ece1 i

s p

-S C C

Acti Bsg A TSP1

n N Hybrid S PrP a h C (Cd147 ) p A2M PSI l r H2K1 N a e O c e

N i -S k l e e iv I-EGF1 N li TSP1 e S s sit

N e- a H s 2 d Act a ti

I-EGF2 e M t p 3

α1 -S o e 1 N A2M N r TSP1 3 S - P I-EGF3 P P N Z e e M 4f2hc F

s C e v S α2 t a I-EGF4 i TSP1 -S -S s -S d -S

-S S 3 ti Acti S S

S p C C A2Mr N - e F -S P P βTD TSP1 S Z C C S--S

C C C

D C C N N C C N C C N C C Lat1 C N N N-linked glycosylation -S IG-like domain C1-type Cd47 C S C Intracellular Tmem20 6

GPI anchor -S IG-like domain C2-type S Atp1a1

Heparan sulfate (GAG) -S IG-like domain V-type S b

Topology / Symbol Description localization Function Link to TGF-β and integrin signaling Ncam1 Neural cell adhesion molecule 1 (CD56) Type I TM (GPI) Cell adhesion Controls focal adhesion assembly during EMT

Slc39a10 Zinc transporter ZIP10 Type 3 TM Zinc ion import into the cytosol Binds Ncam1 and controls its Gsk3-dependent phosphorylaton. Ace Angiotensin-converting enzyme Type 2 TM Releases C-terminal dipeptde of its substrates Releases GPI-anchored proteins by cleaving mannose linkage Ece1 Endothelin-convertng enzyme 1 Type 2 TM Converts big endothelin-1 to endothelin-1 TGF-β regulates endothelin-1 signaling

Slc3a2 4F2 cell-surface antgen heavy chain (CD98 hc) Type 2 TM Transports large neutral amino acids Binds to and modulates integrin actvity

Slc7a5 Large neutral amino acids transporter small subunit 1 Type 3 TM Transports large neutral amino acids Binds to and modulates integrin actvity (CD98 lc) Thsd7a Thrombospondin type-1 domain-containing protein 7A Type 1 TM Promotes cell migraton via FAK-mediated Co-localizaton with integrin complexes in focal signaling. adhesions Thbs1 Thromobospondin-1 Secreted Mediates cell matrix interactions Binds integrin beta-1, TGF-β, CD47 and heparin.

Cd47 Leukocyte surface antigen CD47 Type 3 TM Role in cell adhesion, memory formaton and Integrin co-receptor that interacts with synaptc plastcity thrombospondin-1 Igfp5 Insulin-like growth factor-binding protein 5 Secreted Prolongs half-life of insulin growth factor Localizes to and actvates β1-containing integrin complexes Lgals1 Galectin-1 Secreted Lectin binding beta-galactoside and other Binds integrins and actvates focal adhesion carbohydrates complex signaling Itgb1 Integrin beta-1 Type 1 TM Regulates cell adhesion, motility and Controls TGF-β release from LAP complexes angiogenesis Tmem206Transmembrane protein 206 Type 3 TM Unknown Unknown

Gpc1 Glypican-1 GPI Cell surface proteoglycan that bears heparan Involved in Wnt and TGF-β receptor family sulfate signaling Atp1a1 Sodium/potassium-transportng ATPase subunit alpha-1 Type 3 TM Exchange of Na+ and K+ across plasma Inhibiton causes downregulaton of TGF-β membrane signaling Cd109 CD109 antgen GPI Modulates TGF-β receptor signaling Modulates TGF-β receptor signaling negatvely

Eng Endoglin Type 1 TM Regulates angiogenesis TGF-β co-receptor, interacts with integrins

Bsg Basigin (CD147) Type 1 TM Targets monocarboxylate transporters to the Clusters with CD98 and integrin beta-1 plasma membrane Tfrc Transferrin receptor protein 1 Type 2 TM Cellular iron uptake Facilitates TGF-β receptor family signaling signaling

Figure 4.5. PrP’s molecular environment is enriched for proteins with known roles in TGFβ1 and integrin signaling. (a) Cartoon depicting domain organization, as well as known or predicted mode of membrane association of PrP interactors based on UniProt annotations. (b) Functional annotations of shortlisted PrP interactors.

4.2.6 TGFβ1 profoundly affects steady-state levels of several PrP interactors but depletion of PrP only reduces NCAM1

TGFβ1 had previously come to the fore in relation to PrP, because this cytokine is understood to be a major inducer of EMT, which leads cells to acquire mesenchymal morphology and to shift their mode of attachment to focal adhesion complex-based cell-matrix contacts. A link to EMT was initially suggested by the aforementioned gastrulation arrest phenotype observed in zebra fish deficient for a PrP ortholog and was recently strengthened, when we observed mammalian PrP to be dramatically induced during EMT [4,58]. Here, we were interested in how the execution of this morphogenetic program affects steady-state levels of PrP candidate interactors and in learning to what extent PrP depletion influences changes in their steady-state levels. To address these questions, we capitalized on our access to previously reported global proteome data [4] of wild- type NMuMG cell extracts collected two days following mock- or TGFβ1-treatment (dataset I). A 2nd dataset which compared the global proteomes of TGFβ1-treated wild-type cells versus TGFβ1- treated PrP-deficient cells (dataset II) [4] could be harnessed to elucidate the effect of PrP on its nearest neighbors (Fig. 4.6a).

Of the 26 PrP candidate interactors revealed in the current study, 14 had been quantified in dataset I on the basis of more than three PSMs with associated iTRAQ reporter ion ratios. A closer look at the relative quantitation results led us to observe that except for Ece1, Cd109 and Tmem206 (Supplementary Fig. S4.2), all NMuMG-cell based PrP-candidate interactors had been quantified in the global proteome analyses (datasets I and II), revealing a striking dichotomy: Whereas the steady state levels of PrP candidate interactors with predominant plasma membrane localization were affected by the TGFβ1 exposure of cells, levels of the five PrP candidate interactors with predominant residence in the secretory pathway were not (Fig. 4.6b). Even more strikingly, the orientation of change, i.e., whether TGFβ1 addition resulted in an increase or decrease of a given

interactor, was profoundly divergent, with NCAM1, Itgb1, Lgals1 and Tfrc levels increased upon TGFβ1 exposure but levels of Cd47, Atp1a1 and subunits of the large neural amino acid transporter diminished upon exposure. When comparing PrP-depleted versus wild-type cells (dataset II), a selective reduction in NCAM1 steady-state levels was observed in contrast to the other PrP candidate interactors which showed no steady state changes. The results from these global proteome analyses speak to an intricate level of re-organization of the microenvironment of PrP during EMT (as opposed to mass effects influencing next neighbors in the same way) but also established that PrP depletion during this cellular program, except for its destabilizing effect on NCAM1 steady-state levels, has surprisingly little influence on the levels of other proteins it is surrounded by.

Figure 4.6 mesenchymal a

PrP kd proteins with relatively high levels of expression in: stable wt Heat map color code epithelial mesenchymal + TGFB1 + TGFB1 wt wt cells wt cells dataset I wt + TGFB1 dataset II mesenchymal mesenchymal stable PrP kd cells wt cells dataset II PrP kd +TGFB1 wt + TGFB1 dataset I wt (without TGFB1)

Global proteome data / NMuMG cells

epithelial dataset I dataset II -TGFB1 -TGFB1 -TGFB1 PrP kd1 PrP kd2 PrP kd3 b Accession Modified description Coverage +TGFB1 +TGFB1 +TGFB1 Count wt3 wt3 wt3 Count IPI00230665.3 Neural cell adhesion molecule 1 isoform 3 73.27% 0.367 0.353 0.378 28 0.762 0.757 0.751 57 IPI00132474.3 Integrin beta-1 75.56% 0.627 0.667 0.614 20 1.022 1.004 0.993 27 IPI00118413.2 Thrombospondin-1 67.55% 0.598 0.632 0.716 3 0.989 1.007 1.000 10 IPI00229517.5 Galectin-1 96.30% 0.826 0.802 0.796 35 1.047 1.133 1.156 41 IPI00124700.1 Transferrin receptor protein 1 74.31% 0.890 0.961 0.968 37 0.956 0.962 0.895 36 IPI00403079.4 Leukocyte surface antigen CD47 61.06% 1.169 1.101 1.082 12 1.051 0.967 1.018 13 IPI00311682.5 Sodium/potassium-transportng ATPase subunit alpha-1 61.68% 1.314 1.358 1.350 89 0.926 0.929 0.939 76 IPI00930882.1 4F2 cell-surface antgen heavy chain isoform a 76.46% 1.182 1.209 1.177 23 1.027 0.968 1.063 22 IPI00129395.2 Large neutral amino acids transporter small subunit 1 42.58% 1.514 1.438 1.425 11 1.121 1.023 1.165 9 IPI00133522.2 Protein disulfide-isomerase 96.86% 0.929 0.935 0.908 82 0.883 0.898 0.914 104 IPI00123639.1 Calreticulin 73.32% 0.987 0.961 0.964 52 0.935 0.952 0.978 34 IPI00127983.1 Transmembrane emp24 domain-containing protein 2 71.14% 0.890 0.867 0.904 13 0.939 0.981 1.010 12 IPI00473680.2 Transmembrane emp24 domain-containing protein 9 83.64% 0.944 1.000 0.885 15 0.893 0.980 1.055 16 IPI00466570.4 Transmembrane emp24 domain-containing protein 10 62.10% 0.980 0.943 0.934 12 1.004 0.893 0.985 25

Figure 4.6. Whereas TGFβ1 treatment causes divergent shifts in steady-state levels of a subset of PrP interactors, PrP-depletion in the same paradigm only affected NCAM1 protein levels. (a) Schematic depicting nature of global proteome datasets mined for this study. A more complete presentation of these datasets had been published before17. (b) Comparison of steady-

state protein levels of a subset of PrP interactors before and after TGFβ1 induced mesenchymal differentiation. The steady-state levels of proteins with predominant localization in the ER / Golgi compartments did not change upon two-day exposure of NMuMG cells to TGFβ1 (bottom half of table). However, proteins PrP is expected to be surrounded by at the plasma membrane underwent divergent changes in their steady-state levels. The stable PrP knockdown did not affect the protein levels of PrP candidate interactors, except for NCAM1, whose levels were diminished in PrP- deficient cells. Please see Supplementary Table S2 for a more complete presentation of underlying quantitations.

4.3 Discussion

We set out to investigate the causes for the notorious lack of overlap in the primary literature concerned with the identification of PrP interactors. To this end, we made use of four mouse cell models for which PrP knockout control lines were available, using identical tools and workflows to capture endogenous PrP and its co-purifying partners. The study uncovered 26 proteins that exhibit selective or partial co-enrichment with PrP. All of these PrP candidate interactors were known or predicted to reside in cellular locations PrP is understood to have access to, and about half of them had no previous occurrence in the PrP literature, including the metalloprotease Ece1 and the GPI-anchored molecule Cd109. Except for NCAM1, we observed none of the most co- enriched proteins in proximity to PrP in all cell models. Cumulatively, our results provide a powerful testament to the conclusion that the molecular environment of PrP is to a considerable degree cell type-specific. Yet, the data also establish that across all four cell models PrP is embedded in a specialized molecular environment that appears to be tasked with governing the crosstalk of TGFβ1 and integrin signaling.

To the best of our knowledge, this is the first report that systematically compares the molecular environment of PrP in more than one cell model. A conceptual advance in this study, relative to earlier reports, including our own work [47], was the targeting of endogenous PrP—as opposed to affinity-tagged or overexpressed protein—combined with the harnessing of PrP knockout models for the generation of negative control samples. On the basis of comparable numbers of PSMs (a parameter known to correlate with protein quantity) assigned to PrP, as well as comparable PrP

sequence coverages of biological replicates within and across the four interactome datasets, it can be concluded that similar amounts of this bait protein were present in the capture eluates of all cell-type specific interactome analyses. It is therefore remarkable that the levels of protein subsets which co-purified specifically with PrP differed dramatically across the cell models. A striking example of a protein that exhibited uneven co-enrichment with PrP in a cell-type specific manner was Ece1, which was not at all observed in N2a- or CAD5-based PrP interactome datasets, yet was identified and quantified on the basis of >100 PSMs and a cumulative 63.89% sequence coverage in the NMuMG cell model. We documented that this high level of Ece1 enrichment in the NMuMG cell model was not the consequence of an unusually high level of expression in this model, as Ece1 remained below the level of detection in our global NMuMG cell proteome analyses, and as we observed similar steady-state levels for this protein in the C2C12 model by western blot analysis. Results from additional validation experiments were consistent with the interpretation that PrP was directly crosslinked to an Ece1 dimer but did not reveal reciprocal influences of Ece1 or PrP on their maturation or steady-state expression levels. Although this example validated the conceptual choice to investigate the PrP interactome in more than one model, it also exemplified the limits of conclusions that can be drawn from this study in regards to 1) previously reported PrP candidate interactors, not observed in this work, whose identification was based on other cell models or paradigms, and 2) the completeness of the current list of PrP interactors. Whenever prior reports of PrP interactors in the same paradigm were available, our current data corroborated and complemented their findings. For instance, a prior report, which documented interactors of FLAG- tagged PrP in formaldehyde crosslinked N2a cells6, had—with the exception of endoglin (Eng)— identified all of the PrP interactors observed in the N2a cell model in the current study. The technical improvements underlying the current study, however, helped to flag a subset of previously reported PrP candidate interactors, including several 14-3-3 proteins, tubulin and ribosomal subunits, as most likely representing non-specific interactors.

In light of the striking differences in the molecular environments of PrP in the four cell models employed in this study, it warrants repeating that NCAM1 stood out by being prominent in all PrP interactome datasets. This unique relationship of PrP and NCAM1 was further emphasized by our observation that PrP deficiency in TGFβ1 exposed NMuMG cells had no effect on the steady-state

levels of the proteins found in PrP’s molecular environment, except for causing a relative depletion of NCAM1 levels. The increasingly special status of Ncam1 amongst PrP interactors corroborates previous observations: 1) formaldehyde crosslinking gives rise to SDS-stable NCAM1-PrP complexes that represent a majority of the high molecular mass PrP-crosslink signals in N2a cells [57]; 2) PrP controls NCAM1 polysialylation [4]; and 3) a recent interactome analysis of the closest PrP relatives amongst the ZIP transporters that documented a selective interaction with NCAM1 (and calreticulin) but not with other PrP interactors [87].

Despite the distinctness of PrP interactome data in the four cell models, a Molecular Function GO term analysis pointed toward an overarching similarity in the molecular environment of PrP, apparently populated by proteins with known roles in TGFβ1 and integrin signaling. A closer look at the primary reports underlying the relevant functional annotations, as well as the broader literature, further strengthened this conclusion. For some PrP interactors, the connection to the biology of TGFβ1 and integrin signaling seems less than obvious at first. For example, Tfrc is best known as a cellular importer of transferrin-bound iron. However, robust evidence has also linked this transporter to TGFβ1 signaling (a finding that is not yet reflected in its current GO annotations). More specifically, the neural crest cell-specific knockout of Tfrc in mice caused craniofacial abnormalities, including cleft palate and micrognathia (reduced mandible), and was associated with a suppression of TGFβ1 signaling [109]. The characteristic craniofacial abnormalities observed are pathognomonic of a disease known as Pierre Robin Sequence (PRS) caused by mutations in genes that impair TGFβ1 superfamily signaling. Moreover, Cd109 is a major negative inhibitor of TGFβ1 signaling [110] and Eng is a TGFβ1 co-receptor [111]. Notably, although these proteins share unequivocal and direct links to TGFβ1 signaling, this study did not observe their co-enrichment with PrP in the same cell model. Rather, Cd109 was detected in NMuMG- and C2C12-derived PrP interactome datasets, and endoglin was only seen in proximity to PrP in N2a cells.

Similarly, the Cd98 complex, composed of 4f2 (Slc3a2) and Lat1 (Slc7a5) subunits, is best known for its transport of large neutral amino acids. However, the primary literature and several reviews have drawn attention to the direct involvement of this complex in integrin signaling [112–114].

Cd98 exerts this involvement in part by its direct interaction with Itgb1, which is an integral component of integrin-based focal adhesion complexes and was observed in the N2a interactome dataset in this study. Additional PrP candidate interactors with connections to integrin biology were Cd47, which acts as an integrin co-receptor [115] and was named integrin-associated protein [116] after its discovery as a contaminant in integrin preparations, and Lgals1, which is known to bind to and activate integrin-dependent signaling of focal adhesion complexes [117–119]. As with TGFβ1-associated proteins, these integrin interactors were not all identified in the same PrP interactome dataset, but were revealed in different combinations in one or more of the cell-type specific datasets we presented. Moreover, it is increasingly understood that TGFβ1 and integrin signaling activities do not exist irrespective of one another. Rather, these signaling portals are considered sister complexes bound to each other through several connections. In fact, a subset of proteins shortlisted in this study as PrP interactors, including Cd98 and Thbs1, are known to facilitate this crosstalk [120,121].

How does PrP influence TGFβ1 and integrin signaling complexes? By its mere presence, PrP would be expected to influence the composition and architecture of its molecular environment in many subtle ways. As PrP is a GPI-anchored molecule with affinity for membrane domains enriched in cholesterol and sphingolipids, it seems plausible that one of these influences will relate to its effect on the balance between clathrin-mediated endocytosis and caveolae-based transport. Such a balance shift is unlikely to be innocuous. For example, TGFβ1 signaling is dependent on clathrin-mediated endocytosis (references in [109]), and it has been proposed that CD109 and Tfrc influence the canonical signaling of TGFβ1 in such manner [109,110]. A similar dichotomy of internalization pathways also exists for integrin complexes and Lgals1 has been proposed to influence their trafficking [119]. A more direct influence of PrP on the regulation of focal adhesion complexes may derive from its interaction with NCAM1 and its dramatic influence on NCAM1 polysialylation [4]. Although the details of how NCAM1 promotes the establishment of cell-to- matrix contacts based on focal adhesions have remained murky, and are likely to be complex, there is ample evidence for a TGFβ1-dependent expression of NCAM1 [122] and a critical role of this protein in the molecular rearrangements that govern this cellular program [59,123,124].

4.4 Conclusion

There is a tendency to ascribe differences in interactions of a protein-of-interest reported by separate investigators mainly to differences in methodology, even when studies are undertaken with different paradigms. Insights gained in this work caution that the molecular environment of a given protein-of-interest can be surprisingly diverse when comparing distinct models. The comparative interactome data we presented placed the evolutionarily conserved interaction between PrP and NCAM1 in the context of a plasma membrane microdomain tasked with modulating the crosstalk between TGFβ1 and integrin signaling. We anticipate that further investigations will substantiate roles of PrP in cellular programs relying on these sister signaling hubs. The results from this study further suggest that, unless PrP interactome studies are undertaken with brain cells, next-neighbor relationships central to the cellular toxicity that manifests in prion disorders may be overlooked.

4.5 Methods

4.5.1 Western blot analyses

Equal amounts of protein were separated on 4–12% Bis-Tris gels (Life Technologies, ON, Canada) and transferred to a 0.45 micron polyvinylidene fluoride membrane. The membranes were blocked with 10% skim milk in Tris-buffered saline with Tween 20 (TBST), and incubated overnight at 4°C with the respective primary antibody. Subsequently, the western blots were washed thrice with TBST and incubated with HRP-conjugated anti-mouse (catalog number 1706516; BioRad, Mississauga, ON, Canada), anti-rabbit (catalog number 1706515; BioRad), or anti-rat (31476; Thermo Fisher Scientific, Waltham, MA, USA) secondary antibodies for two hours at room temperature. Membranes were subjected to three washes with TBST and incubated with the ECL reagent (catalog number RPN2106; Sigma Aldrich, Oakville, ON, Canada). Signals were then visualized using either X-ray film or a LI-COR Odyssey Fc digital imaging system (LI-COR Biosciences, Lincoln, NE, USA). Immunoblotting was undertaken with the Sha31 antibody against PrP (catalog number A03213; Bertin Bioreagent, Montigny le Bretonneux, France), or antibodies directed toward Ece1 (catalog number ab71829; Abcam, Cambridge, United Kingdom) or Tfrc (catalog number ab84036; Abcam).

4.5.2 Cell culture and transfection

Mouse mammary gland NMuMG cells (catalog number CRL-1636; American Type Culture Collection (ATCC), Manassas, VA, USA) and Cath.a-differentiated (CAD-2A2D5; CAD5 for short) cells were a kind gift from Dr. Jeffrey Wrana (University of Toronto, Toronto, ON, Canada) and Dr. Charles Weissmann (The Scripps Research Institute, Jupiter, FL, USA), respectively. Mouse myoblast (C2C12) cells (catalog number CRL-1772) and mouse neuroblastoma Neuro-2a (N2a) cells (catalog number CCL-131) were purchased from ATCC. Cells were cultured in D(MEM) or opti-MEM supplemented with 10% heat inactivated fetal bovine serum (catalog number 12484028; Invitrogen Canada, Burlington, ON, Canada), 1% GlutaMAX (catalog number 35050061; Invitrogen Canada), and 1% antibiotic-antimycotic solution (catalog number 15240062; Invitrogen Canada). Human insulin solution (catalog number I9278; Sigma-Aldrich, Oakville, ON, Canada) was added at a concentration of 10 µg/mL for NMuMG cells.

To induce EMT, NMuMG cells were treated with TGFβ1 (catalog number 240-B; R&D Systems, Minneapolis, MN, USA) at a concentration of 6.4 ng/mL daily for 48 hours. C2C12 myoblasts were differentiated to myotubes by replacing the DMEM containing 10% FBS with medium supplemented with 2% horse serum.

The knockout of Prnp in NMuMG, C2C12, and N2a cells had been achieved using CRISPR-Cas9 genetic engineering and was described previously [74]. The CAD5 PrP knockout line was made with identical sgRNA reagents as the other PrP knockout lines but had not been previously described.

4.5.3 Sample preparation for immunoprecipitation analyses

Comparative interactome analyses were performed with three biological replicates for each condition (unless indicated otherwise), that were side-by-side expanded to large scales (average yield of 6 x 107 cells). Cells were washed with ice-cold PBS, and mildly crosslinked with 2% formaldehyde in PBS for 15 minutes at room temperature. Subsequently, residual crosslinking reagent was quenched during a 10-minute incubation with 125 mM glycine in PBS and lysed in ice-cold buffer consisting of 150 mM Tris (pH 8.3), 150 mM NaCl, 0.5% NP-40, 0.5% sodium

deoxycholate supplemented with protease inhibitor cocktail (catalog number 11836170001; Roche, Mississauga, ON, Canada) and phosphatase inhibitor tablet (catalog number 4906837001; Roche). Cellular debris were removed by centrifugation at 2,000 RPM for 5 minutes, followed by 30 minutes at 4,000 RPM (4 °C). Centrifugation supernatants were collected and protein levels were adjusted using a bicinchoninic acid colorimetric assay reagents A (catalog number 23228) and B (catalog number 1859078) from Thermo Fisher Scientific.

4.5.4 Protein immunoprecipitation workflow

PrP was captured using the D18 antibody, a humanized recombinant Fab developed against PrP that was provided by the laboratory of Dr. Emil F. Pai. The PrP antibody was conjugated to KappaSelect beads (catalog number 17-5458-01; GE Healthcare, Oakville, ON, Canada) under gentle agitation by a turning wheel at 4°C overnight. The affinity capture bead / antibody mixture was then equally aliquoted for individual samples and adjusted protein samples were added for an overnight capture at 4°C. The next day, the affinity matrices were stringently washed twice with lysis buffer and twice with lysis buffer containing 500mM NaCl to remove non-specific binders, followed by a pre-elution wash of 10mM HEPES, pH 8. Proteins were then eluted by acidification in 0.2% trifluoroacetic acid, 20% acetonitrile.

4.5.5 Nanoscale HPLC-ESI tandem mass spectrometry

Sample preparation for mass spectrometry was done as described previously [74]. Immunoprecipitation eluates were dried under vacuum then diluted in 9 M deionized urea. Reduction with tris (2-carboxyethyl) phosphine at 60°C was followed by room temperature sulfhydryl group alkylation with 4-vinylpyridine. The urea concentration was lowered to 1.25 M in 500 mM triethylammonium bicarbonate prior to the addition of mass spectrometry-grade trypsin (catalog number 90057; Thermo Fisher Scientific). Digestion occurred at 37°C overnight. Trypsin treated samples were covalently modified with 8-plex isobaric tags for relative and absolute quantitation (iTRAQ) (catalog number 4390811; Sciex, Concord, ON, Canada) according to the manufacturer’s protocol then mixed. Sample mixtures were purified on reversed phase resin in Bond Elut OMIX cartridges (catalog number A57003100; Agilent Technologies, Santa Clara, CA,

USA) alone and in combination with strong cation exchange cartridges (catalog number A57004100; Agilent Technologies).

All sample mixtures were analyzed over a four-hour reversed phase 300 nL/min gradient on an EASY-nLC 1000-Orbitrap Fusion Tribrid mass spectrometer platform (Thermo Fisher Scientific). The analytical column was a 25 cm long Acclaim PepMap RSLC 100 of 75 µm inner diameter with 2 µm C18 particles having 100 Å pores. Each liquid chromatography-mass spectrometry run was divided into scan cycles up to 3 seconds long, each including one orbitrap precursor ion MS scan and as many linear ion trap product ion (MS2) scans and orbitrap MS3 scans as possible within the 3 second time window. CID and higher energy collisional dissociation (HCD) were used for MS2 and MS3 respectively. The Orbitrap resolution was set to 60,000 for both MS and MS3.

4.5.6 Protein identification and quantification

MS2 data were converted to protein sequence information with Proteome Discoverer (version 1.4.0.288; Thermo Fisher Scientific) using the embedded Mascot and Sequest HT search algorithms with the mouse international protein index database (version 3.87). Up to two missed cleavages were allowed per peptide. The allowed peptide mass range was 400-6000 Da, with a precursor ion mass tolerance of 20 ppm and product ion mass tolerance of 0.4 Da. Variable modifications considered were asparagine and glutamine deamidation, methionine oxidation as well as serine, threonine and tyrosine phosphorylation. Cysteine pyridylethylation as well as iTRAQ 8-plex reagent labeling of peptide N-termini and lysines were defined as fixed modifications. False discovery rate estimation based on q-Value was performed with the Percolator algorithm. Relative protein quantification was produced from MS3 data by the Reporter Ions Quantifier built into Proteome Discoverer with the most confident centroid under a mass tolerance of 20 ppm.

4.5.7 Data availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository [125] with the dataset identifier PXD008781.

Chapter 5 Conclusion and Future Directions

Following up on the discovery of an ancient ZIP zinc transporter as the evolutionary ancestor of PrP [49], and the characterization of shared functional and structural similarities [50], an interactome study of ZIP6 implicated ZIPs in a signaling pathway that regulates the posttranslational modification of NCAM1 by serving as a hub for GSK3B [87]. Specifically, we established that ZIP6 and ZIP10 form a heteromer that binds to GSK3B kinase and controls the phosphorylation of the longest isoform of NCAM1 [87]. The ZIP protein family plays a role in regulating the cellular zinc homeostasis and, more precisely, is involved in the transport of zinc (and perhaps other divalent cations) into the cell [51]. Intriguingly, cellular zinc levels have previously been shown to directly affect GSK3 kinase activity, i.e., the presence of free zinc was shown to inhibit GSK3 phosphorylation activity [88]. This zinc-mediated inhibition is reported to be specific to GSK3 since its closely related kinase CDK2 does not share this property [88]. Furthermore, the inhibition of GSK3 is solely zinc-dependent and does not extend to other divalent cations [88]. Moreover, previous reports have singled out GSK3 as one of very few kinases whose activity might be influenced by zinc [88]. Based on our interactome data, the cellular location at which ZIP6- and ZIP10-mediated zinc influx occurs is in close proximity to where GSK3 paralogs are located, presumably leading to a micro-environment characterized by free zinc levels that may dramatically exceed the normally very low pico- to femtomolar cytoplasmic free zinc levels [87]. These findings could point to a regulatory mechanism in which the ZIP6-ZIP10 heteromer acts as a regulator of GSK3 kinase activity by controlling free zinc levels in immediate proximity to the kinase [87]. Thus, cells may have harnessed the ZIP metal ion transport ability to control specific GSK3-dependent substrate phosphorylation events and, by extension, the downstream signaling that depends on them. Although the functional relevance of the ZIP6/10 mediated, GSK3- dependent phosphorylation of NCAM1 is not yet known, there is ample precedent for single phosphorylation events modulating critical protein interactions. Interestingly, NCAM1’s ability to bind to a number of proteins, including integrins and 14-3-3 proteins, the latter being notoriously sequestered to phosphorylated interactors, was hindered in ZIP6-deficient cells [87]. It is therefore

plausible that the GSK3-dependent phospho-occupancy of NCAM1 could constitute a molecular switch by which ZIP proteins impact focal adhesion assemblies [87]. Thus, this angle of my work has generated a specific testable hypothesis. Although emerging from a study undertaken in a broad neurodegenerative disease research context, this type of focal adhesion biology is of broad interest in cancer research, where metastatic cells switch from cell-to-cell based adherence to cell- substrate-based adhesion [56,59]. Several avenues could be pursued to take this research to the next level. Follow-on experimentation could, for example, as a first step employ an in vitro assay format (similar to the one described in Chapter 3), to determine if recombinant 14-3-3 proteins can be co-immunoprecipitated with recombinant or affinity-purified NCAM1 that had been phosphorylated with recombinant GSK3B in the presence or absence of GSK3 specific inhibitors. An analogous experiment could be employed to clarify if phosphorylation of NCAM1 is strictly dependent on GSK3B or can also be mediated by GSK3A, a question our current data did not explicitly answer, although we observed both paralogs of this kinase in the ZIP6 interactome dataset.

The study was the first to investigate the relationship between ZIPs and NCAM1, highlighting that NCAM1 is the key interactor of the ZIP6-ZIP10 heteromeric complex [87]. Considering the evolutionary link between PrPC and ZIPs [49] and the fact that NCAM1 was already established to be a prominent molecular interactor of PrPC [4,47,57,60], it is in hindsight not surprising for NCAM1 biology to also be closely associated with ZIPs. The direct interaction of NCAM1 and PrP has been documented in a number of reports using different cellular and experimental paradigms [4,47,57,60]. Previous work by our group [4] had uncovered one physiological relevance of the interaction between PrPC and NCAM1 by demonstrating that PrPC controls NCAM1 polysialylation. A surprising finding of this line of work was the differential regulatory influence of PrPC on NCAM1 polysialylation in different cellular paradigms [4].

Numerous studies have examined the molecular environment of PrPC with the intent to deduce the biological function of this protein [47,57,95–98]. However, the use of inconsistent cell models and experimental protocols in these analyses caused different investigators to emphasize the importance of distinct PrPC interactors, leading to an abundance of hypotheses on the physiological

role of this protein. To dissect to which extent differences in paradigms and methods versus cell- type specific differences caused these inconsistencies in the literature, we employed an optimized pull-down methodology, followed by MS-based protein identification from mildly crosslinked lysates, to explore side-by-side the protein networks comprising PrPC in multiple cellular models [69]. CRISPR-generated knockout cells served as negative controls which revealed the identity of unspecific binders in each dataset [69].

The main theme that emerged from these comparative interactome analyses was that the majority of PrPC interactors are known to play roles in TGF beta as well as integrin signaling in the context of focal adhesion complex biology [69]. Strikingly, our results revealed differences between PrPC interactions in the four mouse cell models utilized in our investigations [69]. However, our analyses also revealed striking similarities between the PrPC interactome of N2a and CAD5 cells, which we attributed to their shared neuronal lineage [69]. Knowledge of the signaling pathways and the molecular context in which PrPC operates can open new rational avenues for intervening in prion diseases. Additionally, results from this study may be used to generate new hypotheses regarding the mechanisms that underlie differential susceptibility of distinct cell lines to prion infection.

More specifically, these findings may inform two approaches for prion disease intervention: the first of them would aim to take advantage of the known biology of these interactors in order to reduce PrPC levels. Removal of the PrPC substrate is widely considered a promising route for interfering with the conversion to PrPSc based on data which documented that there is an inverse correlation between prion disease incubation period and steady-state PrPC levels in mouse infections studies [31,33]. Considering that the majority of PrPC interactors are known to be embedded into the plasma membrane or to have access to it [69], treatments that trigger their endocytosis may be promising in this regard as it is plausible that PrPC might get passively co- internalized. A second approach might capitalize on data presented in this study to derive avenues for rationally blocking PrPSc-induced toxicity. Such a study would be strengthened if data generated in this work were complemented by results from co-immunoprecipitation analyses that compare the levels of interactors in proximity to PrPC versus PrPSc. The current data on PrPC

interactors would guide the selection of antibodies and serve as a reference. With this approach, differences in the levels of a given interactor in proximity to PrPC versus PrPSc could direct subsequent efforts to reveal if such a perturbation might underlie aberrant signaling that leads to toxicity observed in the disease. This knowledge could then suggest therapeutic angles to address prion disease neurodegeneration by correcting the perturbation or modulating the molecular circuitries involved.

In conclusion, the study emphasized the power of systematic side-by-side investigations of the molecular environment of a protein of interest in various cellular paradigms. Although interactome studies are not rare in the field of proteomics, to the best of our knowledge, a comparative, mass spectrometry-based analysis of the same protein target in more than two models has not been reported before. The methodology we employed represents a transferable platform for the characterization of protein-protein interactions as well as their cell type-specific differences. Had we restricted analyses to only one cell model, we might have uncovered specific interactions between PrPC and a subset of proteins that are involved in TGF beta and integrin signaling pathways but would also have added to a fragmented and highly diverse literature. Only by having interrogated interactions of PrPC in several cell lines, did a broader theme of PrPC involvement in focal adhesion complex biology emerge that united data from all cell models we studied. Thus, the systematic approach we took eliminated potential confounding variables introduced via methodological differences and allowed for a deeper understanding of the context in which PrPC operates. We are confident that the broader strokes of the study design and experimental platform can serve as a template also for the study of other proteins. Knowledge of the cellular context a protein of unknown function operates in can be a critical step toward elucidating its function. In particular in instances, when a large number of competing hypotheses cloud vision, this type of hypothesis-free discovery work can generate a much-needed framework for breaking through an impasse and for generating easily testable hypotheses that move the field forward.

Although the data generated in this study do not directly offer clinical insights, they serve as a stepping stone towards more translational research that can leave a mark in the clinic. To harness their potential diagnostic value, follow-on work could explore if focal adhesion complexes are

altered in prion diseases. In addition to the aforementioned experiments that aim to reduce PrPC levels or block toxicity by manipulating the TGFβ1 or integrin signaling hubs, also urgently needed is a line of investigation that evaluates whether the results observed with murine cells translate to human paradigms. Further analyses could utilize our optimized experimental workflow in human cell lines in order to compare the protein interactions of PrPC between the two species. It is noteworthy to mention due to the transmissible nature of prion diseases, use of human-derived infected samples brings about significant technical cost and requires more strict biosafety regulations, which limits the availability of research facilities that can undertake such investigative work.

While our study aimed at identifying cell type-specific features of the PrP interactome, another avenue could look into mouse brain tissue for these analyses. The underlying reason for our cell- based approach goes back to a previous global proteome study by our group, which highlighted that cell-based rather than tissue-based proteomics investigations can serve as more informative experimental paradigms when exploring the physiological function of PrP [100]. More specifically, we observed that proteome shifts detected in distinct cell models can escape detection in the brain [100], which is indeed plausible when one considers the diversity of cell types present in a particular tissue. If the expression of a certain protein follows opposite trends in individual cell models, one can see how the said effect can get cancelled out and therefore masked in a complex tissue sample [100]. However, this is not to undermine the importance of using more biologically relevant models, such as brain tissue for studying prion diseases. In fact, a deep interactome investigation of wild-type vs. prion infected mouse brain, could reveal novel insights for mechanism-driven drug design.

Our interactome analyses have mainly used dividing cell lines, however, since prion diseases are disorders of the nervous system, an improved study paradigm for cellular dynamics of prion biology could involve non-dividing or primary neuronal cultures. In order to look deeply into not only strong but also weak and even more transient interactions in a pull-down coupled with mass spectrometry design, a high protein yield and thus highly enriched bait levels are required. This criterion can be achieved more easily when using mitosis-competent cell lines. Having stated the

above, with recent advances in the sensitivity of cutting-edge mass spectrometers, in-depth analyses have become possible with less biological starting material, and an extension to non- diving cellular models, including primary neuron culture would nicely complement the data gained in this study.

The epitope of the D18 antibody that was used in our study is localized such that, in addition to full-length PrP, it can potentially also bind C1, C2 and shed PrP [3,79]. Since the majority of these PrP cleavage products are physiologically present in the body, binding partners of these fragments could serve a biological role in the cell or a functional role in the disease state. Additional work is needed in order to pinpoint which of the interactors, if any, are specific to a single cleavage product and whether or not there is a functional relevance of said interaction. Given the apparent protective role of the alpha cleavage and PrP shedding [3,42] as well as the toxic effects associated with beta cleavage [3], it could be relevant in this context to investigate if these proteolytic processing events lead to differential interactions within the molecular environment of PrP, which may trigger distinct downstream signaling pathways.

References

1. Prusiner SB. Prions. Proc Natl Acad Sci. 1998;95: 13363–13383. doi:10.1073/pnas.95.23.13363

2. Aguzzi A, Calella AM. Prions: Protein Aggregation and Infectious Diseases. Physiol Rev. 2009;89: 1105–1152. doi:10.1152/physrev.00006.2009

3. Altmeppen HC, Puig B, Dohler F, Thurm DK, Falker C, Krasemann S, et al. Proteolytic processing of the prion protein in health and disease. Am J Neurodegener Dis. 2012;1: 15– 31.

4. Mehrabian M, Brethour D, Wang H, Xi Z, Rogaeva E, Schmitt-Ulms G. The prion protein controls polysialylation of neural cell adhesion molecule 1 during cellular morphogenesis. PLoS One. 2015;10: e0133741. doi:10.1371/journal.pone.0133741

5. Imran M, Mahmood S. An overview of human prion diseases. Virol J. 2011;8: 559. doi:10.1186/1743-422X-8-559

6. Collinge J. Prion Diseases of Humans and Animals: Their Causes and Molecular Basis. Annu Rev Neurosci. 2001;24: 519–550. doi:10.1146/annurev.neuro.24.1.519

7. Imran M, Mahmood S. An overview of animal prion diseases. Virol J. 2011;8: 1–8. doi:10.1186/1743-422X-8-493

8. Budka H, Aguzzi A, Brown P, Brucher J -M, Bugiani O, Gullotta F, et al. Neuropathological Diagnostic Criteria for Creutzfeldt-Jakob Disease (CJD) and Other Human Spongiform Encephalopathies (Prion Diseases). Brain Pathol. 1995;5: 459–466. doi:10.1111/j.1750- 3639.1995.tb00625.x

9. Chen C, Dong X-P. Epidemiological characteristics of human prion diseases. Infect Dis poverty. BioMed Central; 2016;5: 47.

10. Mead S. Prion disease genetics. Eur J Hum Genet. 2006;14: 273–281. doi:10.1038/sj.ejhg.5201544

11. Puckett C, Concannon P, Casey C, Hood L. Genomic structure of the human prion protein gene. Am J Hum Genet. 1991;49: 320–329.

12. Montagna P, Gambetti P, Cortelli P, Lugaresi E. Familial and sporadic fatal insomnia. Lancet Neurol. 2003;2: 167–176. doi:10.1016/S1474-4422(03)00323-5

13. Capellari S, Strammiello R, Saverioni D, Kretzschmar H, Parchi P. Genetic Creutzfeldt- Jakob disease and fatal familial insomnia: Insights into phenotypic variability and disease

pathogenesis. Acta Neuropathol. 2011;121: 21–37. doi:10.1007/s00401-010-0760-4

14. Asante EA, Smidak M, Grimshaw A, Houghton R, Tomlinson A, Jeelani A, et al. A naturally occurring variant of the human prion protein completely prevents prion disease. Nature. 2015;522: 478–481. doi:10.1038/nature14510.A

15. Prusiner SB. Prion diseases and the BSE crisis. Science. 1997;278: 245–251. doi:10.1126/science.278.5336.245

16. Llewelyn CA, Hewitt PE, Knight RSG, Amar K, Cousens S, Mackenzie J, et al. Possible transmission of variant Creutzfeldt-Jakob disease by blood transfusion. Lancet. Elsevier; 2004;363: 417–421.

17. Rutala WA, Weber DJ. Guideline for Disinfection and Sterilization of Prion-Contaminated Medical Instruments. Infect Control Hosp Epidemiol. 2010;31: 107–117. doi:10.1086/650197

18. Collins S, McLean CA, Masters CL. Gerstmann-Sträussler-Scheinker syndrome, fatal familial insomnia, and kuru: A review of these less common human transmissible spongiform encephalopathies. J Clin Neurosci. 2001;8: 387–397. doi:10.1054/jocn.2001.0919

19. Bruce ME, Will RG, Ironside JW, McConnell I, Drummond D, Suttie A, et al. Transmissions to mice indicate that “new variant” CJD is caused by the BSE agent. Nature. 1997;389: 498–501. doi:10.1038/39057

20. Gambetti P, Dong Z, Yuan J, Xiao X, Zheng M, Alshekhlee A, et al. A novel human disease with abnormal prion protein sensitive to protease. Ann Neurol. 2008;63: 697–708. doi:10.1002/ana.21420

21. Zou W, Puoti G, Xiao X, Yuan J, Qing L, Cali I, et al. Variably Protease-Sensitive Prionopathy: A New Sporadic Disease of the Prion Protein. Ann Neurol. 2010;68: 162–172. doi:10.1002/ana.22094.Variably

22. Gambetti P, Cali I, Notari S, Kong Q, Zou W, Surewicz WK. Molecular biology and pathology of prion strains in sporadic human prion diseases. 2011;121: 79–90. doi:10.1007/s00401-010-0761-3.Molecular

23. Prusiner S. Novel proteinaceous infectious particles cause scrapie. Science (80- ). 1982;216: 136–144. doi:10.1126/science.6801762

24. Soto C, Castilla J. The controversial protein-only hypothesis of prion propagation. Nat Med. 2004;10: S63. doi:10.1038/nm1069

25. Alper T, Cramp WA, Haig DA, Clarke MC. Does the agent of scrapie replicate without nucleic acid? Nature. 1967;214: 764–766. doi:10.1038/214764a0

26. Griffith JS. Nature of the scrapie agent: Self-replication and scrapie. Nature. 1967;215: 1043–1044. doi:10.1038/2151043a0

27. Prusiner SB, Bolton DC, Groth DF, Bowman KA, Patricia Cochran S, McKinley MP. Further Purification and Characterization of Scrapie Prions. Biochemistry. 1982;21: 6942– 6950. doi:10.1021/bi00269a050

28. Oesch B, Westaway D, Wälchli M, McKinley MP, Kent SBH, Aebersold R, et al. A cellular gene encodes scrapie PrP 27-30 protein. Cell. 1985;40: 735–746. doi:10.1016/0092- 8674(85)90333-2

29. Basler K, Oesch B, Scott M, Westaway D, Wälchli M, Groth DF, et al. Scrapie and cellular PrP isoforms are encoded by the same chromosomal gene. Cell. 1986;46: 417–428. doi:10.1016/0092-8674(86)90662-8

30. Pan KM, Baldwin M, Nguyen J, Gasset M, Serban A, Groth D, et al. Conversion of alpha- helices into beta-sheets features in the formation of the scrapie prion proteins. Proc Natl Acad Sci. 1993;90: 10962–10966. doi:10.1073/pnas.90.23.10962

31. Büeler H, Aguzzi A, Sailer A, Greiner RA, Autenried P, Aguet M, et al. Mice devoid of PrP are resistant to scrapie. Cell. 1993;73: 1339–1347. doi:10.1016/0092-8674(93)90360-3

32. Fischer M, Rülicke T, Raeber a, Sailer a, Moser M, Oesch B, et al. Prion protein (PrP) with amino-proximal deletions restoring susceptibility of PrP knockout mice to scrapie. EMBO J. 1996;15: 1255–64. doi:10.1002/j.1460-2075.1996.tb00467.x

33. White MD, Farmer M, Mirabile I, Brandner S, Collinge J, Mallucci GR. Single treatment with RNAi against prion protein rescues early neuronal dysfunction and prolongs survival in mice with prion disease. Proc Natl Acad Sci. 2008;105: 10238–10243. doi:10.1073/pnas.0802759105

34. Brandner S, Raeber A, Sailer A, Blättler T, Fischer M, Weissmann C, et al. Normal host prion protein (PrPC) is required for scrapie spread within the central nervous system. Proc Natl Acad Sci. National Acad Sciences; 1996;93: 13148–13151.

35. Sailer A, Büeler H, Fischer M, Aguzzi A, Weissmann C. No propagation of prions in mice devoid of PrP. Cell. 1994. pp. 967–968. doi:10.1016/0092-8674(94)90436-7

36. Peretz D, Williamson RA, Kaneko K, Vergara J, Leclerc E, Schmitt-Ulms G, et al. Antibodies inhibit prion propagation and clear cell cultures of prion infectivity. Nature. 2001;412: 739–743. doi:10.1038/35089090

37. Enari M, Flechsig E, Weissmann C. Scrapie prion protein accumulation by scrapie-infected neuroblastoma cells abrogated by exposure to a prion protein antibody. Proc Natl Acad Sci. 2001;98: 9295–9299. doi:10.1073/pnas.151242598

38. White AR, Enever P, Tayebi M, Mushens R, Linehan J, Brandner S, et al. Monoclonal antibodies inhibit prion replication and delay the development of prion disease. Nature. 2003;422: 80–83. doi:10.1038/nature01457

39. Westergard L, Christensen HM, Harris DA. The cellular prion protein (PrPC): Its physiological function and role in disease. Biochim Biophys Acta - Mol Basis Dis. 2007;1772: 629–644. doi:10.1016/j.bbadis.2007.02.011

40. Mallucci GR. Post-natal knockout of prion protein alters hippocampal CA1 properties, but does not result in neurodegeneration. EMBO J. 2002;21: 202–210. doi:10.1093/emboj/21.3.202

41. Chesebro B, Trifilo M, Race R, Meade-White K, Teng C, LaCasse R, et al. Anchorless prion protein results in infectious amyloid disease without clinical scrapie. Science (80- ). 2005;308: 1435–1439. doi:10.1126/science.1110837

42. Stöhr J, Watts JC, Legname G, Oehler A, Lemus A, Nguyen H-OB, et al. Spontaneous generation of anchorless prions in transgenic mice. Proc Natl Acad Sci. National Acad Sciences; 2011;108: 21223–21228.

43. Watts JC, Westaway D. The prion protein family: Diversity, rivalry, and dysfunction. Biochim Biophys Acta - Mol Basis Dis. 2007;1772: 654–672. doi:https://doi.org/10.1016/j.bbadis.2007.05.001

44. Stahl N, Borchelt DF, Hsiao K. Scrapie Prion Protein Contains a Phosphatidylinositol G lycolipid. 1987;51: 229–240.

45. Paulick MG, Bertozzi CR. The glycosylphosphatidylinositol anchor: A complex membrane-anchoring structure for proteins. Biochemistry. 2008;47: 6991–7000. doi:10.1021/bi8006324

46. Walter ED, Stevens DJ, Visconte MP, Millhauser GL. The prion protein is a combined zinc and copper binding protein: Zn2+ alters the distribution of Cu2+ coordination modes. J Am Chem Soc. ACS Publications; 2007;129: 15440–15441.

47. Watts JC, Huo H, Bai Y, Ehsani S, Won AH, Shi T, et al. Interactome analyses identify ties of PrPC and its mammalian paralogs to oligomannosidic N-Glycans and endoplasmic reticulum-derived chaperones. PLoS Pathog. Public Library of Science; 2009;5: e1000608. doi:10.1371:/journal.ppat.1000608

48. Ehsani S, Tao R, Pocanschi CL, Ren H, Harrison PM, Schmitt-Ulms G. Evidence for retrogene origins of the prion gene family. PLoS One. 2011;6. doi:10.1371/journal.pone.0026800

49. Schmitt-Ulms G, Ehsani S, Watts JC, Westaway D, Wille H. Evolutionary descent of prion genes from the ZIP family of metal Ion transporters. PLoS One. 2009;4.

doi:10.1371/journal.pone.0007208

50. Ehsani S, Huo H, Salehzadeh A, Pocanschi CL, Watts JC, Wille H, et al. Family reunion - The ZIP/prion gene family. Prog Neurobiol. Elsevier Ltd; 2011;93: 405–420. doi:10.1016/j.pneurobio.2010.12.001

51. Jeong J, Eide DJ. The SLC39 family of zinc transporters. Mol Aspects Med. 2013;34: 612– 619. doi:10.1016/j.mam.2012.05.011

52. Yamashita S, Miyagi C, Fukada T, Kagara N, Che Y-S, Hirano T. Zinc transporter LIVI controls epithelial-mesenchymal transition in zebrafish gastrula organizer. Nature. Macmillian Magazines Ltd.; 2004;429: 298. Available: http://dx.doi.org/10.1038/nature02545

53. Málaga-Trillo E, Solis GP, Schrock Y, Geiss C, Luncz L, Thomanetz V, et al. Regulation of Embryonic Cell Adhesion by the Prion Protein. PLOS Biol. Public Library of Science; 2009;7: e1000055. Available: https://doi.org/10.1371/journal.pbio.1000055

54. Taylor KM, Muraina IA, Brethour D, Schmitt-Ulms G, Nimmanon T, Ziliotto S, et al. Zinc transporter ZIP10 forms a heteromer with ZIP6 which regulates embryonic development and cell migration. Biochem J. 2016;473: 2531–2544. doi:10.1042/BCJ20160388

55. Mehrabian M, Ehsani S, Schmitt-Ulms G. An emerging role of the cellular prion protein as a modulator of a morphogenetic program underlying epithelial-to-mesenchymal transition. Front Cell Dev Biol. 2014;2: 53. doi:10.3389/fcell.2014.00053

56. Xu J, Lamouille S, Derynck R. TGF-β-induced epithelial to mesenchymal transition. Cell Res. Nature Publishing Group; 2009;19: 156.

57. Schmitt-Ulms G, Legname G, Baldwin MA, Ball HL, Bradon N, Bosque PJ, et al. Binding of neural cell adhesion molecules (N-CAMs) to the cellular prion protein. J Mol Biol. 2001;314: 1209–1225. doi:10.1006/jmbi.2001.5183

58. Mehrabian M, Hildebrandt H, Schmitt-Ulms G. NCAM1 Polysialylation: The Prion Protein’s Elusive Reason for Being? ASN Neuro. SAGE Publications Inc; 2016;8: 1759091416679074. doi:10.1177/1759091416679074

59. Lehembre F, Yilmaz M, Wicki A, Schomber T, Strittmatter K, Ziegler D, et al. NCAM- induced focal adhesion assembly: a functional switch upon loss of E-cadherin. EMBO J. EMBO Press; 2008;27: 2603–2615.

60. Santuccione A, Sytnyk V, Leshchyns’ka I, Schachner M. Prion protein recruits its neuronal receptor NCAM to lipid rafts to activate p59fynand to enhance neurite outgrowth. J Cell Biol. 2005;169: 341–354. doi:10.1083/jcb.200409127

61. Tobler I, Gaus SE, Deboer T, Achermann P, Fischer M, Rülicke T, et al. Altered circadian

activity rhythms and sleep in mice devoid of prion protein. Nature. Nature Publishing Group; 1996;380: 639. Available: http://dx.doi.org/10.1038/380639a0

62. Bremer J, Baumann F, Tiberi C, Wessig C, Fischer H, Schwarz P, et al. Axonal prion protein is required for peripheral myelin maintenance. Nat Neurosci. Nature Publishing Group; 2010;13: 310.

63. Singh A, Kong Q, Luo X, Petersen RB, Meyerson H, Singh N. Prion protein (PrP) knockout mice show altered iron metabolism: A functional role for PrP in iron uptake and transport. PLoS One. 2009;4. doi:10.1371/journal.pone.0006115

64. Collinge J, Whittington MA, Sidle KCL, Smith CJ, Palmer MS, Clarke AR, et al. Prion protein is necessary for normal synaptic function. Nature. Nature Publishing Group; 1994;370: 295.

65. Colling SB, Khana M, Collinge J, Jefferys JGR. Mossy fibre reorganization in the hippocampus of prion protein null mice. Brain Res. Elsevier; 1997;755: 28–35.

66. Milhavet O, Lehmann S. Oxidative stress and the prion protein in transmissible spongiform encephalopathies. Brain Res Rev. Elsevier; 2002;38: 328–339.

67. Steele AD, Emsley JG, Özdinler PH, Lindquist S, Macklis JD. Prion protein (PrPc) positively regulates neural precursor proliferation during developmental and adult mammalian neurogenesis. Proc Natl Acad Sci U S A. 2006;103: 3416 LP-3421. Available: http://www.pnas.org/content/103/9/3416.abstract

68. Laurén J, Gimbel DA, Nygaard HB, Gilbert JW, M S, Strittmatter SM. Cellular Prion Protein Mediates Impairment of Synaptic Plasticity by Amyloid-β Oligomers. Nature. 2009;457: 1128–1132. doi:10.1038/nature07761.Cellular

69. Ghodrati F, Mehrabian M, Williams D, Halgas O, Bourkas MEC, Watts JC, et al. The prion protein is embedded in a molecular environment that modulates transforming growth factor β and integrin signaling. Sci Rep. 2018; 8654. doi:10.1038/s41598-018-26685-x

70. Miettinen PJ, Ebner R, Lopez AR, Derynck R. TGF-beta induced transdifferentiation of mammary epithelial cells to mesenchymal cells: involvement of type I receptors. J Cell Biol. 1994;127: 2021 LP-2036. Available: http://jcb.rupress.org/content/127/6/2021.abstract

71. Herbst A, Banser P, Velasquez CD, Mays CE, Sim VL, Westaway D, et al. Infectious Prions Accumulate to High Levels in Non Proliferative C2C12 Myotubes. PLoS Pathog. 2013;9. doi:10.1371/journal.ppat.1003755

72. Mahal SP, Baker CA, Demczyk CA, Smith EW, Julius C, Weissmann C. Prion strain discrimination in cell culture: The cell panel assay. Proc Natl Acad Sci. 2007;104: 20908– 20913. doi:10.1073/pnas.0710054104

73. Bhaya D, Davison M, Barrangou R. CRISPR-Cas systems in bacteria and archaea: versatile small RNAs for adaptive defense and regulation. Annu Rev Genet. Annual Reviews; 2011;45: 273–297.

74. Mehrabian M, Brethour D, Macisaac S, Kim JK, Gunawardana CG, Wang H, et al. CRISPR-Cas9-based knockout of the prion protein and its effect on the Proteome. PLoS One. 2014;9: 1–21. doi:10.1371/journal.pone.0114594

75. Ran FA, Hsu PD, Wright J, Agarwala V, Scott DA, Zhang F. Genome engineering using the CRISPR-Cas9 system. Nat Protoc. 2013;8: 2281–2308. doi:10.1038/nprot.2013.143

76. Hsu PD, Lander ES, Zhang F. Development and Applications of CRISPR-Cas9 for Genome Engineering. Cell. 2014;157: 1262–1278. doi:https://doi.org/10.1016/j.cell.2014.05.010

77. Kim CH, Oh Y, Lee TH. Codon optimization for high-level expression of human erythropoietin (EPO) in mammalian cells. Gene. Elsevier; 1997;199: 293–301.

78. Sutherland BW, Toews J, Kast J. Utility of formaldehyde cross-linking and mass spectrometry in the study of protein–protein interactions. J mass Spectrom. 2008;43: 699– 715.

79. Williamson RA, Peretz D, Pinilla C, Ball H, Bastidas RB, Rozenshteyn R, et al. Mapping the Prion Protein Using Recombinant Antibodies. J Virol. 1998;72: 9413–9418. Available: http://jvi.asm.org/content/72/11/9413.abstract

80. Dayon L, Sanchez J-C. Relative Protein Quantification by MS/MS Using the Tandem Mass Tag Technology BT - Quantitative Methods in Proteomics. In: Marcus K, editor. Totowa, NJ: Humana Press; 2012. pp. 115–127. doi:10.1007/978-1-61779-885-6_9

81. Haag AM. Mass Analyzers and Mass Spectrometers BT - Modern Proteomics – Sample Preparation, Analysis and Practical Applications. In: Mirzaei H, Carrasco M, editors. Cham: Springer International Publishing; 2016. pp. 157–169. doi:10.1007/978-3-319-41448-5_7

82. Marcotte EM. How do shotgun proteomics algorithms identify proteins? Nat Biotechnol. Nature Publishing Group; 2007;25: 755.

83. Gundry RL, White MY, Murray CI, Kane LA, Fu Q, Stanley BA, et al. Preparation of proteins and peptides for mass spectrometry analysis in a bottom-up proteomics workflow. Curr Protoc Mol Biol. 2010;90: 10–25.

84. Wiese S, Reidegeld KA, Meyer HE, Warscheid B. Protein labeling by iTRAQ: a new tool for quantitative mass spectrometry in proteome research. Proteomics. 2007;7: 340–350.

85. Ho CS, Lam CWK, Chan MHM, Cheung RCK, Law LK, Lit LCW, et al. Electrospray ionisation mass spectrometry: principles and clinical applications. Clin Biochem Rev. The Australian Association of Clinical Biochemists; 2003;24: 3.

86. Kong BY, Duncan FE, Que EL, Kim AM, O’halloran T, Woodruff TK. Maternally-derived zinc transporters ZIP6 and ZIP10 drive the mammalian oocyte-to-egg transition. Mol Hum Reprod. Oxford University Press; 2014;20: 1077–1089.

87. Brethour D, Mehrabian M, Williams D, Wang X, Ghodrati F, Ehsani S, et al. A ZIP6-ZIP10 heteromer controls NCAM1 phosphorylation and integration into focal adhesion complexes during epithelial-to-mesenchymal transition. Sci Rep. 2017;7: 1–19. doi:10.1038/srep40313

88. Ilouz R, Kaidanovich O, Gurwitz D, Eldar-Finkelman H. Inhibition of glycogen synthase kinase-3β by bivalent zinc ions: insight into the insulin-mimetic action of zinc. Biochem Biophys Res Commun. Elsevier; 2002;295: 102–106.

89. Mackie K, Sorkin BC, Nairn AC, Greengard P, Edelman GM, Cunningham BA. Identification of two protein kinases that phosphorylate the neural cell-adhesion molecule, N-CAM. J Neurosci. Soc Neuroscience; 1989;9: 1883–1896.

90. Little EB, Crossin KL, Krushel LA, Edelman GM, Cunningham BA. A short segment within the cytoplasmic domain of the neural cell adhesion molecule (N-CAM) is essential for N-CAM-induced NF-κB activity in astrocytes. Proc Natl Acad Sci. National Acad Sciences; 2001;98: 2238–2243.

91. Ballif BA, Villén J, Beausoleil SA, Schwartz D, Gygi SP. Phosphoproteomic analysis of the developing mouse brain. Mol Cell Proteomics. ASBMB; 2004;3: 1093–1101.

92. Karbe Y, Pollscheit J, Glaubitz N, Horstkorte R, Bork K. Phosphorylation of serine 774 of the neural cell adhesion molecule (NCAM) is involved in the interaction with collapsin response mediator protein-2. Neurochem Res. Springer; 2013;38: 1229–1235.

93. Korshunova I, Caroni P, Kolkova K, Berezin V, Bock E, Walmod PS. Characterization of BASP1-mediated neurite outgrowth. J Neurosci Res. Wiley Online Library; 2008;86: 2201– 2213.

94. Saeki K, Matsumoto Y, Hirota Y, Matsumoto Y, Onodera T. Three-exon structure of the gene encoding the rat prion protein and its expression in tissues. Virus Genes. 1996;12: 15– 20. doi:10.1007/BF00369996

95. Rutishauser D, Mertz KD, Moos R, Brunner E, Rülicke T, Calella AM, et al. The comprehensive native interactome of a fully functional tagged prion protein. PLoS One. 2009;4. doi:10.1371/journal.pone.0004446

96. Rieger R, Edenhofer F, Lasmézas CI, Weiss S. The human 37-kDa laminin receptor precursor interacts with the prion protein in eukaryotic cells. Nat Med. Nature Publishing Group; 1997;3: 1383. Available: http://dx.doi.org/10.1038/nm1297-1383

97. Yehiely F, Bamborough P, Da Costa M, Perry BJ, Thinakaran G, Cohen FE, et al.

Identification of Candidate Proteins Binding to Prion Protein. Neurobiol Dis. 1997;3: 339– 355. doi:https://doi.org/10.1006/nbdi.1997.0130

98. Zanata SM, Lopes MH, Mercadante AF, Hajj GNM, Chiarini LB, Nomizo R, et al. Stress- inducible protein 1 is a cell surface ligand for cellular prion that triggers neuroprotection. EMBO J. 2002;21: 3307 LP-3316. Available: http://emboj.embopress.org/content/21/13/3307.abstract

99. Taylor KM, Nicholson RI. The LZT proteins; the LIV-1 subfamily of zinc transporters. Biochim Biophys Acta - Biomembr. 2003;1611: 16–30. doi:https://doi.org/10.1016/S0005- 2736(03)00048-8

100. Mehrabian M, Brethour D, Williams D, Wang H, Arnould H, Schneider B, et al. Prion protein deficiency causes diverse proteome shifts in cell models that escape detection in brain tissue. PLoS One. 2016;11. doi:10.1371/journal.pone.0156779

101. Dlakic WM, Grigg E, Bessen RA. Prion Infection of Muscle Cells In Vitro. J Virol. 2007;81: 4615–4624. doi:10.1128/JVI.02628-06

102. Qi Y, Wang JKT, McMillian M, Chikaraishi DM. Characterization of a CNS Cell Line, CAD, in which Morphological Differentiation Is Initiated by Serum Deprivation. J Neurosci. 1997;17: 1217 LP-1225. Available: http://www.jneurosci.org/content/17/4/1217.abstract

103. Klammer AA, Park CY, Noble WS. Statistical Calibration of the SEQUEST XCorr Function. J Proteome Res. American Chemical Society; 2009;8: 2106–2113. doi:10.1021/pr8011107

104. Meidan R, Klipper E, Gilboa T, Muller L, Levy N. Endothelin-converting Enzyme-1, Abundance of Isoforms a-d and Identification of a Novel Alternatively Spliced Variant Lacking a Transmembrane Domain. J Biol Chem. 2005;280: 40867–40874. doi:10.1074/jbc.M505679200

105. Naslavsky N, Stein R, Yanai A, Friedlander G, Taraboulos A. Characterization of Detergent-insoluble Complexes Containing the Cellular Prion Protein and Its Scrapie Isoform. J Biol Chem. 1997;272: 6324–6331. doi:10.1074/jbc.272.10.6324

106. Taylor DR, Whitehouse IJ, Hooper NM. Glypican-1 mediates both prion protein lipid raft association and disease isoform formation. PLoS Pathog. 2009;5: 1–15. doi:10.1371/journal.ppat.1000666

107. Fang C, Josefin L, L. HC, R. BD, Katrin M. Copper-dependent co-internalization of the prion protein and glypican-1. J Neurochem. 2006;98: 1445–1457. doi:10.1111/j.1471- 4159.2006.03981.x

108. Kajsa L, Fang C, Lars-Åke F, Katarina B, Katrin M. Involvement of glypican-1 autoprocessing in scrapie infection. Eur J Neurosci. 2008;28: 964–972. doi:10.1111/j.1460- 9568.2008.06386.x

109. Lei R, Zhang K, Liu K, Shao X, Ding Z, Wang F, et al. Transferrin receptor facilitates TGF- β and BMP signaling activation to control craniofacial morphogenesis. Cell Death &Amp; Dis. The Author(s); 2016;7: e2282. Available: http://dx.doi.org/10.1038/cddis.2016.170

110. Li CD, Hancock MA, Seghal P, Zhou S, Reinhardt DP, Philip A. Soluble CD109 binds TGF-β and antagonizes TGF-β signaling and responses. Biochem J. 2015; Available: http://www.biochemj.org/content/early/2015/11/30/BJ20141488.abstract

111. Valluru M, Staton C, Reed M, Brown N. Transforming Growth Factor-β and Endoglin Signaling Orchestrate Wound Healing [Internet]. Frontiers in Physiology. 2011. p. 89. Available: https://www.frontiersin.org/article/10.3389/fphys.2011.00089

112. Nguyen HTT, Merlin D. Homeostatic and innate immune responses: role of the transmembrane glycoprotein CD98. Cell Mol Life Sci. 2012;69: 3015–3026. doi:10.1007/s00018-012-0963-z

113. Deves R, Boyd CAR. Surface antigen CD98 (4F2): not a single membrane protein, but a family of proteins with multiple functions. J Membr Biol. Springer; 2000;173: 165–177.

114. Yan Y, Vasudevan S, Nguyen HTT, Merlin D. Intestinal epithelial CD98: an oligomeric and multifunctional protein. Biochim Biophys Acta (BBA)-General Subj. Elsevier; 2008;1780: 1087–1092.

115. Isenberg JS, Roberts DD, Frazier WA. CD47: a new target in cardiovascular therapy. Arterioscler Thromb Vasc Biol. Am Heart Assoc; 2008;28: 615–621.

116. Brown E, Hooper L, Ho T, Gresham H. Integrin-associated protein: a 50-kD plasma membrane antigen physically and functionally associated with integrins. J Cell Biol. Rockefeller University Press; 1990;111: 2785–2794.

117. Moiseeva EP, Williams B, Goodall AH, Samani NJ. Galectin-1 interacts with β-1 subunit of integrin. Biochem Biophys Res Commun. Elsevier; 2003;310: 1010–1016.

118. Fischer C, Sanchez-Ruderisch H, Welzel M, Wiedenmann B, Sakai T, André S, et al. Galectin-1 interacts with the α5β1 fibronectin receptor to restrict carcinoma cell growth via induction of p21 and p27. J Biol Chem. ASBMB; 2005;

119. Fortin S, Le Mercier M, Camby I, Spiegl-Kreinecker S, Berger W, Lefranc F, et al. Galectin- 1 Is Implicated in the Protein Kinase C ε/Vimentin-Controlled Trafficking of Integrin-β1 in Glioblastoma Cells. Brain Pathol. 2010;20: 39–49.

120. Huang T, Sun L, Yuan X, Qiu H. Thrombospondin-1 is a multifaceted player in tumor

progression. Oncotarget. Impact Journals, LLC; 2017;8: 84546.

121. Adams JC, Lawler J. The thrombospondins. Cold Spring Harb Perspect Biol 3: a009712. 2011.

122. Roubin R, Deagostini-Bazin H, Hirsch M-R, Goridis C. Modulation of NCAM expression by transforming growth factor-beta, serum, and autocrine factors. J Cell Biol. Rockefeller University Press; 1990;111: 673–684.

123. Beggs HE, Baragona SC, Hemperly JJ, Maness PF. NCAM140 Interacts with the Focal Adhesion Kinase p125fak and the SRC-related Tyrosine Kinase p59fyn. J Biol Chem. ASBMB; 1997;272: 8310–8319.

124. Frame MC, Inman GJ. NCAM is at the heart of reciprocal regulation of E-cadherin-and integrin-mediated adhesions via signaling modulation. Dev Cell. Elsevier; 2008;15: 494– 496.

125. Vizcaíno JA, Csordas A, Del-Toro N, Dianes JA, Griss J, Lavidas I, et al. 2016 update of the PRIDE database and its related tools. Nucleic Acids Res. Oxford University Press; 2015;44: D447–D456.

Appendix

Supplementary Figure S4.1. Consistent and selective enrichment of PrP contrasted to non- specific binding of Gapdh. (a) Box plots of PrP-derived peptides in all four models. Please see legend to Fig. 4.2b for a detailed description of graph elements. (b) Box plots of Gapdh-derived peptides in all four models.

Supplementary Figure S4.2

NMuMG C2C12 CD109 TMEM206 CD109 Median 2.25 1.20 1.83 0.55 2.17 2.25 -0.49 2.80 1.10 2.69 2.11 -0.12 2.66 0.20 2.72 IQR 0.61 1.42 1.28 0.93 1.17 2.01 0.47 2.22 0.88 1.72 1.54 1.36 1.70 1.45 2.49 wt1 ko1 wt2 ko2 wt3 wt1 ko1 wt2 ko2 wt3 wt1 ko1 wt2 ko2 wt3 ko3 ko3 ko3 ko3 ko3 ko3 ko3 ko3 ko3 ko3 ko3 ko3 ko3 ko3 ko3

6 6

4 4

2 2 e e g g n n a a h h C C

0 0 d d l l o o F F

2 2 g g o o l -2 l -2

-4 -4

Supplementary Figure S4.2. Selective PrP co-enrichment of Cd109 and Tmem206. Box plots of CD109 and Tmem206 in the subset of datasets, in which these proteins were robustly identified and quantified. Please see legend to Fig. 4.2b for a detailed description of graph elements.

Supplementary Figure S4.3

Lysates Eluates NMuMG CAD5

MW [kDa] wt1 PrP ko1wt1 wt2 PrP ko1PrP ko2wt1 wt2 PrP ko1PrP ko2

191

97 anti-Ece1 64

191

97 Coomassie 64

39 1 2 3 4 5 6 7 8 9 10

Supplementary Figure S4.3. Evidence that Ece1 is not expressed in CAD5 cells at levels detectable by western blot analysis. Ece1 western blot analysis of formaldehyde crosslinked lysate and eluate fractions from the PrP-directed co-immunoprecipitation of wild-type and PrP knockout CAD5 cells (analogous to data shown in Fig. 4.4a). The results validate the PrP interactome data (Table 4.1) which failed to detect Ece1 in PrP-directed co-immunoprecipitations from CAD5 cell lysates. NMuMG cell lysates were loaded as a positive control for Ece1 detection,

and a Coomassie-stain of the western blot documents total protein levels in the respective lysates and eluates. Arrowheads indicate signals derived from monomeric and SDS-stable crosslinked dimeric Ece1.

Supplementary Figure S4.4

Lysates Eluates NMuMG C2C12 N2a NMuMG C2C12 N2a

MW MW wt1 wt2 PrP ko1PrP ko2wt1 wt2 PrP ko1PrP ko2wt1 wt2 PrP ko1PrP ko2 [kDa] wt1 wt2 PrP ko1PrP ko2wt1 wt2 PrP ko1PrP ko2wt1 wt2 PrP ko1PrP ko2 [kDa] anti-Tfrc anti-Tfrc 191 191 97 97

64 64

51 51

191 191 97 97 Coomassie Coomassie 64 64 51 51 39 39

28 28

1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12

Lysates Eluates NMuMG CAD5

MW wt1 PrP ko1wt1 wt2 PrP ko1PrP ko2wt1 wt2 PrP ko1PrP ko2 [kDa]

191 anti-Tfrc 97

191

97 Coomassie

28 1 2 3 4 5 6 7 8 9 10

Supplementary Figure S4.4 PrP co-immunoprecipitates Tfrc from wild-type but not PrP knockout NMuMG cell lysates. Validation of the transferrin receptor protein 1 (Tfrc) as a PrP binder. Consistent with the PrP interactome data, Tfrc is only prominently represented in PrP co- immunoprecipitation eluate fractions derived from formaldehyde-crosslinked wild-type NMuMG cells. However, whereas Tfrc escaped detection by mass spectrometry in PrP co- immunoprecipitation eluates derived from wild-type C2C12 and N2a cells (Table 1), weak Tfrc signals can be detected in the respective eluates by western blot analysis, presumably reflecting a slightly higher sensitivity of western blot analysis over mass spectrometry-based detection for this protein. No Tfrc was observed in PrP co-immunoprecipitation eluates from CAD5 wild-type cells. Coomassie stains of the western blot membranes are shown underneath the immunoblot panels to document protein amounts in the respective samples. Blue and green arrowheads point toward Tfrc monomer and formaldehyde-crosslinked dimer signals, respectively. A higher molecular mass band (possibly a crosslink of two Tfrc dimers) can also be detected. The empty arrowhead shown in the Coomassie images indicates the D18 recombinant Fab used for PrP-directed immunoprecipitation. Note also the subtle differences in Tfrc immunoblot signal intensities when comparing wild-type and PrP knockout lysates in NMuMG, C2C12 and N2a cell models, which could be indicative of molecular crosstalk between PrP and Tfrc modulating steady-state levels of Tfrc.

6 8 0 2 6 6 6 6 1 4 4 6 e 8 9 8 1 6 4 0 8 1 0 4 3 8 3 7 3 1 7 4 5 9 1 0 ...... r 3 3 0 7 8 5 3 5 9 8 9 3 2 6 4 1 3 9 4 7 2 9 1 6 0 5 7 4 3 0 2 2 0 3 8 o 1 1 8 6 0 5 1 0 5 2 4 1 6 3 9 3 6 1 7 7 9 5 2 7 4 6 9 5 3 0 9 3 5 3 c 2 3 3 1 2 1 2 2 2 1 5 3 8 3 1 3 S t 3 4 1 0 2 0 3 5 9 0 4 8 5 3 0 8 3 4 3 6 0 2 8 5 4 6 4 8 3 2 2 3 7 4 2 3 3 n 4 8 8 2 5 1 1 4 1 1 1 1 5 2 1 1 5 3 0 1 4 1 8 1 u o C 7 5 2 4 0 0 2 5 6 2 8 8 3 8 9 7 4 7 1 1 5 7 3 1 4 8 7 3 7 4 4 3 8 2 6 4 8 8 9 0 0 7 0 2 9 9 0 7 6 0 0 9 8 0 8 0 0 3 9 1 0 8 8 7 5 6 8 9 2 0 7 9 0 1 2 3 ...... o o 0 0 1 1 0 1 1 0 0 1 0 0 1 1 0 0 1 0 1 1 1 0 1 1 0 0 0 0 0 0 0 0 1 0 0 1 1 k k 6 0 0 6 3 5 8 3 5 1 8 2 9 5 4 8 2 1 6 0 7 8 5 4 8 1 4 9 0 8 6 1 3 7 1 4 4 5 1 2 1 1 1 0 5 3 0 7 2 5 2 0 5 4 2 3 2 2 2 0 5 6 4 7 2 2 8 0 1 3 1 5 5 7 1 3 2 ...... D t o 1 1 1 1 1 1 2 1 1 1 1 1 1 1 2 3 2 2 1 1 1 1 5 3 4 2 1 1 3 2 3 4 8 1 8 2 2 A w k 1 1 1 4 1 1 1 C 4 7 2 8 5 3 9 7 4 5 8 9 7 2 0 7 8 1 8 7 8 3 6 4 9 1 0 7 0 8 6 8 0 3 1 2 5 2 4 2 1 2 9 1 4 0 9 2 0 2 2 3 7 7 2 2 4 2 2 5 1 2 3 2 8 5 2 3 7 4 4 1 6 8 1 3 ...... o o 1 1 1 1 1 0 4 1 1 1 1 2 1 1 2 2 2 4 1 1 1 1 1 4 3 7 3 4 4 1 1 4 4 5 6 8 2 k k 1 1 1 4 1 1 1 5 4 2 4 1 4 6 4 2 0 5 0 5 7 9 1 0 7 5 4 1 7 6 4 0 7 5 7 4 7 3 5 4 8 3 0 7 0 1 6 0 0 5 8 1 0 5 6 0 9 0 9 8 5 8 6 1 7 0 0 9 5 3 0 7 5 3 5 6 0 7 0 7 9 3 1 ...... t o 1 1 0 1 1 0 2 1 1 1 0 3 0 1 1 1 3 2 0 1 0 4 1 4 3 3 0 2 2 7 4 0 1 8 4 5 1 w k 1 1 1 3 1 1 1 5 6 5 0 6 7 7 2 5 4 2 6 8 9 4 5 3 6 2 e 9 8 8 7 2 9 8 9 4 8 7 5 2 5 ...... r 8 3 9 8 2 9 2 5 2 8 6 1 5 5 9 9 0 9 5 9 7 5 5 1 1 6 9 2 5 8 5 5 3 o 8 4 7 4 1 9 0 4 1 4 7 8 4 6 7 7 0 2 9 9 7 5 9 3 0 7 3 6 6 0 3 5 3 c 5 1 3 8 6 1 3 6 3 1 2 5 2 7 1 7 6 2 3 5 7 2 4 S 1 1 1 t 0 5 4 0 8 1 4 4 5 6 3 0 2 0 8 8 3 1 2 2 1 3 0 5 5 0 6 5 8 3 3 0 9 n 2 3 4 7 2 3 6 1 6 1 1 2 4 6 0 3 3 5 4 8 3 1 6 8 1 4 4 8 1 u 1 3 1 1 2 1 2 1 1 1 1 1 o C 4 7 8 6 6 0 6 6 7 5 6 1 3 0 7 9 9 9 6 4 7 9 4 4 7 3 5 3 6 4 7 1 0 4 3 2 4 4 9 4 4 3 4 5 5 3 7 3 3 3 0 8 3 3 2 0 4 5 1 5 6 5 3 3 8 4 3 3 ...... t o 0 0 0 0 0 0 0 0 0 0 9 1 0 0 0 1 0 1 0 0 0 0 5 0 2 1 0 7 0 0 0 7 2 w k 2 1 8 5 6 2 6 1 1 4 9 4 1 1 2 5 6 4 3 2 6 2 5 6 9 3 8 8 1 0 9 2 7 5 3 8 5 4 8 8 7 5 7 5 8 3 9 4 7 5 9 8 0 8 4 5 5 2 8 6 1 0 7 7 4 5 3 0 2 3 ...... o o 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 1 0 0 1 1 1 0 0 0 1 1 k k 3 9 3 9 8 0 7 4 8 6 2 5 7 9 4 6 1 5 8 3 0 5 9 5 5 0 9 5 9 1 2 9 9 7 7 3 7 7 4 6 8 7 7 8 6 3 1 8 2 7 1 4 4 8 0 4 7 5 9 6 3 1 4 8 6 4 A 3 2 ...... t o 2 0 0 1 0 0 1 1 0 0 0 7 1 1 1 0 1 0 1 1 1 0 1 4 0 7 2 9 8 1 1 0 7 2 w k 1 N 5 1 9 9 6 9 0 2 4 9 0 6 3 3 1 8 6 0 9 1 1 8 4 7 0 5 6 8 2 5 9 4 4 2 6 6 2 2 3 6 6 5 2 9 9 7 3 5 7 2 0 3 7 6 6 0 2 6 2 2 3 3 7 5 3 7 1 3 ...... o o 1 1 1 1 1 1 1 1 1 1 1 0 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 k k 7 4 3 0 2 6 4 6 1 0 0 2 7 8 4 2 1 0 2 7 3 4 7 2 1 6 0 8 7 9 4 8 3 9 5 5 1 0 0 0 1 6 5 0 7 5 9 5 5 0 9 4 3 8 5 9 7 4 9 8 8 0 3 7 8 2 3 1 ...... t o 0 2 1 1 3 1 1 2 3 1 1 1 5 1 2 1 1 1 0 1 2 2 1 1 3 3 0 2 0 7 1 6 3 w k 1 1 7 9 4 8 9 1 6 1 9 5 2 6 1 8 4 9 1 4 8 e 3 6 3 3 8 1 0 6 9 6 1 7 6 ...... r 0 9 4 8 6 3 9 1 7 9 7 2 8 0 4 4 9 6 1 2 8 7 9 1 6 2 9 1 9 3 2 4 o 4 9 4 2 9 8 0 7 2 7 2 2 4 5 5 0 2 4 2 1 3 0 2 1 1 4 3 7 7 1 3 c 1 3 1 1 4 1 1 2 3 1 1 3 6 2 3 1 2 3 1 2 1 1 5 S t 2 8 2 8 3 2 6 8 2 8 9 0 2 2 5 0 2 6 8 1 5 7 7 4 5 8 6 2 2 6 5 6 n 2 3 3 2 9 3 1 5 1 5 2 2 2 1 4 2 6 2 3 2 2 3 3 5 u o C 9 6 3 8 8 1 2 6 5 8 8 3 1 7 5 4 8 2 2 6 8 1 2 5 8 1 5 5 1 9 7 8 3 7 6 3 9 0 4 5 3 6 8 6 7 3 2 6 3 9 9 1 5 1 5 6 6 5 6 9 5 9 3 3 ...... t o 0 0 0 0 0 0 1 1 0 0 0 0 6 0 0 6 0 0 0 2 3 0 6 0 0 6 0 0 0 0 0 w k 1 3 8 9 9 1 1 3 9 3 7 5 3 9 5 3 4 2 1 2 3 2 5 2 2 0 3 1 4 7 3 5 8 3 1 1 1 5 0 2 1 8 6 1 0 1 1 0 1 1 7 0 0 2 2 1 6 1 0 0 2 0 9 3 2 3 ...... o o 1 2 1 1 1 1 1 1 0 1 1 1 1 1 2 1 1 0 1 3 1 0 1 1 1 1 2 1 1 2 1 k k 2 3 0 4 9 1 9 5 7 6 6 9 5 2 9 6 6 5 9 2 7 9 4 2 5 9 6 6 6 4 9 8 1 1 7 9 8 4 8 0 5 9 1 8 8 3 8 5 2 7 4 9 3 8 3 1 4 8 7 7 9 8 8 1 3 2 ...... t o C 1 1 0 0 1 0 1 1 0 1 0 0 6 0 1 5 0 1 0 9 2 0 6 1 0 4 1 0 7 0 1 w k 2 2 C 8 6 0 0 9 4 2 6 3 9 6 0 2 1 2 0 6 3 4 2 0 5 1 8 4 2 5 7 9 4 3 3 9 2 2 5 1 5 0 9 5 1 2 9 0 9 2 1 2 1 4 4 3 2 9 2 5 9 5 5 1 9 1 3 ...... o o 1 1 4 4 1 1 1 1 0 2 4 1 0 1 1 1 2 1 1 2 1 0 1 1 4 1 1 2 1 1 0 k k 1 7 3 6 0 0 0 7 4 9 5 2 2 4 0 4 6 3 5 7 4 6 6 7 4 7 2 4 5 0 1 3 3 6 4 4 9 5 6 9 5 4 9 3 0 6 5 3 1 9 6 4 2 6 3 7 4 0 5 3 9 0 3 1 ...... t o 1 0 4 1 1 4 0 1 1 0 3 1 0 4 1 3 4 1 1 0 9 3 0 0 5 4 1 2 2 1 1 w k 1 2 5 3 6 7 4 6 2 0 1 7 9 6 2 0 0 3 3 0 5 6 7 9 6 5 0 6 2 9 4 5 e 5 0 1 0 0 0 9 1 ...... r 9 2 7 2 6 7 4 2 8 3 7 5 0 4 4 5 6 5 4 1 4 2 6 7 8 2 9 0 9 8 8 4 8 6 7 1 5 6 o 1 5 3 1 6 4 9 5 3 9 2 0 1 4 9 1 8 7 3 3 3 3 3 2 0 6 4 6 5 8 6 3 5 8 5 2 4 7 c 3 3 4 2 2 4 3 3 4 3 2 3 2 6 5 2 1 9 1 6 3 1 5 1 1 S t 8 7 2 5 6 3 1 4 4 6 1 8 7 5 5 1 9 0 9 2 4 9 9 1 6 4 8 7 9 9 9 1 8 2 9 6 n 6 7 8 5 5 2 9 1 7 2 8 5 6 1 1 1 1 1 1 4 1 1 2 2 7 4 4 3 4 1 u 1 1 1 1 2 1 o C 3 3 9 7 1 3 1 4 6 3 4 1 5 4 0 9 8 7 2 4 0 7 6 3 6 0 3 8 9 1 9 8 8 6 0 9 8 9 2 3 4 0 5 9 1 8 8 9 7 3 8 5 4 7 0 5 8 5 4 2 9 4 2 4 8 5 5 3 2 3 0 2 3 3 ...... (only a subset of non-specific interactors shown) t o 1 1 2 1 1 2 9 1 2 1 1 2 1 1 1 3 4 1 3 2 1 1 1 5 1 6 2 1 4 8 2 1 5 3 2 2 w k 1 1 2 5 3 5 9 3 3 4 4 3 3 2 8 2 9 6 7 7 2 3 9 9 8 3 9 4 2 4 2 3 3 6 0 7 3 3 2 2 7 8 8 6 0 2 2 2 3 7 9 9 1 2 4 8 9 7 8 9 8 0 1 1 3 1 0 4 7 8 9 1 4 7 3 2 ...... o o 1 1 0 0 0 0 1 1 1 1 1 0 0 0 1 1 1 0 0 0 0 0 0 1 1 2 1 1 2 1 0 0 0 1 1 0 k k G 2 6 9 0 3 8 5 7 1 2 0 4 5 2 6 5 6 5 7 0 6 0 2 9 1 6 6 4 7 1 0 2 2 8 2 3 0 8 2 5 5 3 2 8 5 0 2 3 2 5 8 2 5 5 6 3 5 8 5 2 0 8 9 0 9 1 3 5 5 3 0 3 3 2 M ...... t o 1 0 0 0 0 0 6 0 0 1 1 0 1 0 0 3 3 0 2 0 0 0 0 6 4 0 6 2 3 9 0 0 1 7 2 0 u w k 1 M N 5 2 7 5 3 9 9 3 7 6 9 7 2 3 8 2 4 4 1 3 1 5 9 0 9 9 1 8 5 7 8 3 5 6 6 6 0 0 7 3 3 7 0 0 8 0 8 7 9 3 0 0 4 4 9 8 4 1 2 8 6 0 7 7 7 6 7 3 1 1 7 5 3 1 ...... o o 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 0 1 0 1 1 1 1 0 0 1 0 0 0 0 1 1 1 1 0 k k 8 7 0 0 9 1 4 4 7 1 0 8 4 7 9 9 1 4 7 9 7 5 2 2 7 2 8 5 3 2 3 7 1 4 0 9 3 3 1 1 7 8 0 3 3 1 4 3 6 4 7 3 9 7 0 7 1 7 2 1 8 0 4 7 9 6 4 1 8 4 1 0 3 1 ...... t o 1 1 1 1 0 0 1 6 1 1 1 1 1 1 0 1 2 4 1 3 1 2 1 1 0 4 1 4 1 4 9 1 0 2 9 2 w k 1 1 9 6 6 3 6 7 0 3 0 4 9 9 5 0 8 0 3 5 0 0 4 1 6 9 3 6 8 7 9 s 7 2 9 6 8 5 2 5 4 4 2 1 8 5 8 5 9 3 2 2 7 4 7 3 0 1 7 2 2 9 5 8 6 1 2 4 9 3 6 6 6 9 9 8 3 0 7 2 1 5 3 4 3 6 4 5 7 4 5 3 6 3 9 0 8 8 3 6 M 7 7 7 0 3 4 3 9 7 4 3 4 7 1 5 2 1 1 4 1 6 3 9 6 1 7 8 4 5 0 3 4 1 2 S 1 1 1 1 1 P 0 6 4 2 9 9 5 7 9 6 0 1 9 6 3 0 4 9 0 7 1 0 4 1 0 3 8 8 3 3 6 6 2 2 3 5 5 8 6 0 8 3 3 5 5 0 6 s 0 9 3 3 2 1 1 2 1 3 2 3 1 1 1 2 3 2 3 7 3 4 3 6 1 2 1 1 1 1 2 4 1 1 3 1 2 3 6 2 5 5 1 1 3 1 1 e 5 d i t p e P e % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % g 0 8 1 9 1 2 7 5 1 6 8 8 2 5 1 1 1 0 4 7 6 2 0 7 9 1 6 7 0 4 9 2 9 5 8 6 4 2 2 4 8 9 3 7 7 3 4 2 0 a 4 7 7 7 3 8 8 9 6 9 1 7 9 8 9 1 7 5 7 8 2 9 5 5 5 6 0 3 8 2 5 3 9 7 7 2 1 4 5 4 3 8 5 9 7 1 0 7 8 r ...... e 9 9 3 2 4 1 0 9 1 1 9 7 6 9 5 9 3 2 7 3 8 7 7 5 0 3 9 4 5 3 9 1 2 2 9 0 5 5 5 3 0 3 3 7 1 6 6 8 0 v 6 7 9 7 6 8 7 7 7 3 6 7 7 7 4 5 7 6 4 6 6 3 6 4 4 6 6 2 6 5 7 5 6 4 7 4 8 6 7 8 5 6 7 4 6 3 5 5 4 o C 6 0 a b e a 0 0 1 a 4 2 3 3 b b b c a a 2 1 a 1 1 a 9 2 5 2 e q g z h 1 9 5 3 3 1 h h h h 1 2 5 1 2 1 1 5 1 7 x a a 7 d 9 d m e a a d a a a a a s 0 1 7 f 1 2 2 1 m 1 1 1 b 1 1 a b b a b 1 s a l p d D K 3 r 1 p 2 t t t t e 3 e e 3 7 e h h h h h c n s l 1 s l s 4 s a f - - g a h c e e g b b b b b b b s b a b p n s s s s c c c f r p e p a d p p p d p 2 3 i i 2 2 i c i w w n w l w t c c w l l s r 4 u u u m u u g h h m u m f m g t g T T H H Y Y H T E T T L R H H C Y T I C R T G T I H H S R R Y A G P T T C T P N R A E H Y S S B T T 1 - a 0 A 1

h 1 9 2 7 t

i p l n n n n n i i i i a u

e e e e t a t t t a t i b h o o o h o n u r r r r p p s u l l p p p

p l

a l a

g g g u g a 5 P s n n n K n

i i i i

- - m n n n n n e s i i i i D K i

s e a a a r a , , t t t t t a e n n n i o n n n t n P r e e r a o o o o T p g g o h c c c i i c

t A t - - - - p 7 c

g 1

1 n n n n n s n 4

e i i i

i g n y i a a n e a a a e D a

v m l i d a 1 y t y r a

y m C 6 u m m m m n t t r t a

i i i a z e n y

c l l 0 e o o o o o i i i 5 3 3 n b 7 z n s h e s 2 d d d d - p e l b

b 1 2 2 1 e

r i n t i e d a a s i t t

o n g e n r 4 S S 4 t L S 4 S 1 o e o i i l

c n

n n - t a

a g r e t k e 2 2 2 0 i i m i e a n e n n n n n a a n n c l e p g p n

p g i n t i i i i i i p p p 1 a a

i r i -

d m o a t p n t t n n n a a m o r a e e / e e e l h h 1 f i o i m i m P a r -

i o m t t m t t r m t y

t - n I o t h o h a c c n n a t i o a o r e s m s e i t e o o e t o o p i e o

h n

a c m i e a Z c

r i r r r c s r c h v t h e

c p a - t

e p e n A B u h m t r e e e e c p d i p v p p o c e p o n i e a e t g z e e w e B

t 1 1 f

a o e n l l n l l n n l n n d n s h o c n - - d r c 1 o e x

t s r s 2 5 2 3 i i n n n n l n i i . a a - a e s a a a r a a a c o n o d r a - - - - f i i i i i a a u e p r r r r l f 3 - e 1 a h r h a g a g c p o a o m a A . a B a a t . r e e e e s e h h i m m

m m r m

t t

b b - b

b u n t t t t t t t t

t t s I t l I i p n p n a 2 3 4 2 o 3 u p p l o o o o 1 o n e i

o e s

n e l e l e e o o o r o e o 1 i l n i s o s u i s m m - m m o s n t r r s r s s s r l e s r H H H H - s k s r H i b p b a b a b b - p

i u a d

b o e

n e

/ e n e r e r n l

c p p o p s o o o p e s o p n

i l i c -

n e e e e e o b n e a n a e p c n n n n n n i l n a b t a b b b h a b i 9 n i t m m

m m m f e r i i i i i i i i t l i i i l i n i 3 3 g 3 3 t 3 n n n n c n l l l l l r l t o r i a c r r r r i r s m m 0 s s s s c e - - o - - e u - g D l

c c e r e i o

o i

o o o o

o i u u u u u g u t k o

e r p t n o o 1 n n n n t t 3 3 t t 3 3 g 3 t k c j u l l u g d d d s 2 b b b b b r r e b - - S - S

S S - u r S - o 2 2 y o s s s a s a a s a a t s n a a l a D i i i i - - e i n n n o r a r i u u r u u u h h r r u r r 4 4 0 4 0 0 0 0 4 e a 0 4 F n n 1 H T T H 1 1 H T A T E 4 H G T C 1 H T C I 4 T T G 2 I 6 4 T H 1 N Z L T L E P S P M 4 T H 4 T T B 5 5 1 1 2 1 3 7 4 2 1 1 1 5 2 5 4 1 6 1 1 1 3 1 2 1 1 1 1 1 1 2 1 1 2 3 4 2 2 1 2 5 1 5 1 3 1 1 1 ...... 2 4 1 3 3 4 5 9 0 0 3 2 2 0 9 7 8 9 7 8 3 9 4 1 3 9 6 8 2 1 0 0 3 8 1 1 9 9 5 5 2 2 3 1 1 8 2 0 5 9 6 6 5 4 8 4 3 7 9 8 5 6 6 4 1 4 3 0 2 6 0 7 0 1 2 3 4 2 7 2 8 7 9 7 0 7 8 9 1 2 8 9 0 5 3 8 0 9 n 3 2 0 7 3 3 5 3 5 6 9 3 7 6 5 5 3 6 7 8 4 8 4 1 4 6 3 8 0 3 9 6 1 4 9 8 0 2 3 8 5 6 7 9 2 5 8 7 4 o i 7 0 9 0 5 8 4 7 6 2 7 7 9 0 6 9 7 3 0 5 9 3 2 8 8 7 7 2 4 6 2 3 5 6 2 3 3 7 9 2 3 1 0 5 2 3 0 4 8 s 2 3 0 1 8 1 5 0 6 7 2 1 2 3 7 2 1 2 3 8 6 5 3 8 1 4 3 8 1 8 7 7 1 1 2 7 0 8 2 6 3 1 2 2 1 5 3 2 0 s 2 2 1 1 7 1 7 4 4 2 1 1 9 2 8 2 1 1 2 9 1 1 1 9 1 9 1 2 1 9 9 4 5 1 1 2 4 1 1 7 1 3 1 1 1 5 9 1 4 e 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 c 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I c P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I A I I I I I I I Colors shading as for Table 1. In addition, grey shading of different intensities was used to visually separate non-specific interactors belonging to distinct protein families. Colors shading as for Table Supplementary Table 4.1. Comparison of the PrP interactome in four mouse cell models 4.1. Comparison of the PrP Supplementary Table Supplementary Table S4.1. Comparison of the PrP interactome in four mouse cell models.

019 .976 .977 0.938 1.112 0 0.955 0 0.999 0.898 1.103 1.006 0.961 0.931 0.757 1. 1.012 II wt 0 8 3 1 7 9 4 9 6 2 1 6 9 7 1 5 4 .45 I 0.94 0.80 0.94 0.97 0.88 0.64 0.92 0.63 1 0.36 1.18 0.9 1.3 1.11 TMT Averages wt cells wt cells -/+TGFB 6 1 6 4 2 0 4 7 9 7 2 5 6 3 mesenchymal mesenchymal 3 4 1 3 1 1 2 5 2 2 7 1 10 Count 5 6 5 8 4 0 0 3 5 1 3 5 9 8 3 0.89 1.15 1.05 0.97 0.91 1.01 1.00 0.99 1.16 0.75 1.06 0.98 0.93 1.01 wt PrP kd3 7 4 1 1 2 0 7 9 0 4 5 0 4 5 3 I I 0.93 1.05 1.05 1.00 1.01 0.91 0.88 1.00 1.20 0.94 1.00 0.97 1.01 0.96 wt2 wt

t e s 2 3 0 2 8 1 7 4 3 7 8 3 9 7 a t 3 a d 0.96 1.13 0.98 0.95 0.89 0.98 1.00 1.00 1.02 0.75 0.96 0.89 0.92 0.96 wt epithelial wt cells mesenchymal kd cells stable PrP proteins with relatively high levels of expression in: PrP kd2 3 4 3 3 4 8 5 1 4 1 6 0 2 5 3 96 . 1.02 0 1.00 0.96 0.96 0.99 1.05 0.95 0.97 1.10 0.98 1.06 0.98 1.02 wt1 wt wt 1 7 6 5 3 3 9 9 2 1 2 7 4 6 4 3 wt + TGFB1 + wt 3 wt + TGFB1 + wt 0 9 98 02 . . . . PrP kd +TGFB1 PrP 1.05 1 0.95 0 0.89 0.88 0.93 1 0 1.12 0.76 1.02 1.00 0.92 wt PrP kd1 t 2 5 7 2 5 2 3 0 3 1 8 3 2 9 1 3 3 5 1 8 1 2 1 2 2 1 8 Coun 2 6 8 4 5 8 4 4 6 5 8 7 1 4 0 9 6 1 1 7 9 6 7 . . . . dataset I dataset II Heat map color code 1.08 0 0.96 0 0.88 0.90 0.90 0 0 1.42 0.37 1.17 0.93 1.35 -TGFB1 +TGFB 7 8 2 2 1 0 9 8 7 1 3 6 1 9 3 3 7 5 0 1 0 99 9 8 . . . . . Global proteome analyses / NMuMG cells 1.02 1 0.97 0.98 1 0.89 0 0 0 1.07 0.94 0.95 1.00 1.04 +TGFB1 +TGFB I

1 2 1 0 1 7 7 5 2 8 3 9 1 3 8 t 6 0 6 6 3 3 e s 9 8 9 6 9 6 ...... a 0 0 1.00 0 0.86 0 1.10 0 0 1.43 0.35 1.20 0.94 1.35 t a -TGFB1 +TGFB d 9 9 8 0 9 3 7 3 3 3 9 4 1 8 0 4 1 5 1 6 6 9 9 9 0 9 9 ...... 0 0 0.99 0 0.91 1.10 1 0 0 1.12 0.98 1.04 0.99 1.02 +TGFB1 +TGFB 0 6 4 7 0 9 9 7 8 4 7 2 1 0 4 9 2 8 2 2 9 6 8 8 9 9 6 5 ...... 0 0 0.94 0 0.89 1.16 0 0 0 1.51 1.18 0.3 0.98 1.31 -TGFB1 +TGFB % % % % % % % % % % % % % % 1 0 2 6 6 5 6 3 64 3 3 14 8 5 5 ...... 4 6 3 6 5 7 7 83. 9 7 71. 61.06 9 7 6 42.58 73.27 62.10 61.68 76.4 overage C - 0 1 9 2 t a 3 7 1

n i e s e t a r o r e p

m r o o 1 s t - i - p n i e e d d c 1 i - n f e l a r o Table S4.2. Global proteome analysis of NMuMG cells -/+ TGFB1 (datset I) in PrP-deficient and wildtype (dataset II) Table

t u p n n s e i s i i 1 l b r - o d

u r

n b n c i e n i i t f i t r m s c e g e t o e n r l e l r o a t a a r r h n Transmembrane emp24 domain-containing protein 1 Leukocyte surface antgen CD4 surface Leukocyte T G Transmembrane emp24 domain-containing protein C Transmembrane emp24 domain-containing protein T P I Large neutral amino acidsLarge neutral transporter small subuni Neural cell adhesion molecule 1 isoform cell adhesion Neural 4F2 cell-surface antgen heavy chain isoform4F2 cell-surface antgen heavy Sodium/potassium-transportng ATPase subunit alpha Modiﬁed descripton 4 4 2 1 5 1 1 2 2 3 2 3 1 5 ...... 0 7 9 3 2 4 0 1 3 1 2 7 7 5 6 4 5 4 n 4 9 3 8 3 2 2 2 2 1 3 3 1 2 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 I I I I I I P P P P P P IPI00466570. IPI00403079. IPI00473680. I I IPI00127983. I I I I IPI00129395. IPI00230665. IPI00930882. IPI00311682. Accessio Supplementary

Supplementary Table S4.2. Global proteome analysis of NMuMG cells -/+ TGFΒ1 (datset I) in PrP-deficient and wildtype cells (dataset II).