prp and its ancestral relatives zip6 and zip10 interact with … · 2017-11-21 · transition...
TRANSCRIPT
PrP and its ancestral relatives ZIP6 and ZIP10 interact with NCAM1, altering its molecular environment and post-
translational modifications during epithelial-to-mesenchymal transition
by
Dylan Edward Brethour
A thesis submitted in conformity with the requirements for the degree of Master of Science
Department of Laboratory Medicine and Pathobiology University of Toronto
© Copyright by Dylan Edward Brethour 2016
ii
PrP and its ancestral relatives ZIP6 and ZIP10 interact with
NCAM1, altering its molecular environment and post-translational
modifications during epithelial-to-mesenchymal transition
Dylan Edward Brethour
Master of Science
Department of Laboratory Medicine and Pathobiology
University of Toronto
2016
Abstract
The prion protein (PrP) was recently found to be evolutionarily linked to a subfamily of ZIP
transporters which possess a PrP-like domain. A member of this subfamily, ZIP6, is of particular
interest as separate studies have shown that morpholino knockdowns of ZIP6 or PrP in zebrafish
leads to an impairment in gastrulation, a process dependent on epithelial-to-mesenchymal
transition (EMT). Furthermore, the neural cell adhesion molecule (NCAM1), a known interactor
of PrP, has itself been described as a mediator of EMT. Based on these findings, we
hypothesized that both PrP and ZIP6 play crucial roles in the process of EMT by controlling the
environment surrounding NCAM. We determined that ZIP6 forms a heteromeric complex with
ZIP10 that affects NCAM1’s integration into adhesion complexes while also mediating its
phosphorylation during EMT. Meanwhile, PrP was found to have a unique role in controlling the
polysialylation of NCAM1 during EMT.
iii
Acknowledgments
First and foremost, I would like to thank my supervisor Dr. Gerold Schmitt-Ulms for allowing
me the opportunity to work in his lab, and for providing invaluable guidance and support
throughout my project. I would also like to thank Dr. Joel Watts and Dr. Lorraine Kalia for all of
the time, guidance, and constructive criticism they have provided me.
I would like to give special thanks to Mohadeseh Mehrabian for all of the time and hard work
that she also contributed to this project. Her assistance greatly helped with the progression of this
project, as well as with my development as a researcher. I would also like to thank my lab
members Declan Williams, Xinzhu Wang, Hansen Wang, and others for all of their assistance
and advice with my project.
Lastly, I would like to thank all of my friends and family for always supporting me, and helping
me to stay focused and positive whenever problems arose.
iv
Table of Contents
Abstract ........................................................................................................................................... ii
Acknowledgments.......................................................................................................................... iii
Table of Contents ........................................................................................................................... iv
List of Tables ................................................................................................................................ vii
List of Figures .............................................................................................................................. viii
Abbreviations ................................................................................................................................. ix
Project rationale ...............................................................................................................................1
Background .................................................................................................................................1
Hypothesis ...................................................................................................................................3
Specific aims ...............................................................................................................................3
Scientific impact..........................................................................................................................3
Chapter 1 Effects of PrP-deficiency on the global proteome, and its role in epithelial-to-
mesenchymal transition ...............................................................................................................5
1.1 Introduction ..........................................................................................................................6
1.2 Materials and Methods .........................................................................................................9
1.2.1 Antibodies and transforming growth factors ...........................................................9
1.2.2 Generation of gRNA Expression Vectors ................................................................9
1.2.3 Cell Culture and Transfection ................................................................................10
1.2.4 Generation of stable knockdown cell clones .........................................................10
1.2.5 Genetic analysis .....................................................................................................10
1.2.6 Western blot analyses ............................................................................................11
1.2.7 Enzymatic characterization of post-translational modifications of NCAM1 ........11
1.2.8 RT-PCR analysis ....................................................................................................11
1.2.9 Sample preparation for comparative global proteomics analysis ..........................12
1.2.10 Nanospray ionization tandem mass spectrometry ..................................................13
v
1.2.11 Post-acquisition analyses .......................................................................................13
1.3 Results ................................................................................................................................15
1.3.1 Strategy of CRISPR-Cas9-based PrP knockout in three mouse cell lines .............15
1.3.2 Validation and characterization of PrP knockout cell clones ................................17
1.3.3 Workflow of global proteome comparison of PrP knockout (or knockdown)
and wild-type NMuMG cells .................................................................................20
1.3.4 The global proteome of PrP-deficient NMuMG cells............................................22
1.3.5 PrP controls NCAM polysialylation during EMT .................................................27
1.4 Discussion ..........................................................................................................................32
1.4.1 CRISPR-Cas9 generated PrP-knockout cells are a viable model ..........................32
1.4.2 Benefits of a cell-specific, mass spectrometry based approach .............................33
1.4.3 Insights into the physiological role of PrP .............................................................34
1.5 Conclusions ........................................................................................................................36
Chapter 2 A ZIP6-ZIP10 heteromer interacts with NCAM1 controlling its phosphorylation
and integration into focal adhesion complexes during epithelial-to-mesenchymal transition ..37
2.1 Introduction ........................................................................................................................38
2.2 Materials and Methods .......................................................................................................39
2.2.1 Antibodies and siRNAs..........................................................................................39
2.2.2 Cell culture and transfection ..................................................................................39
2.2.3 Generation of Slc39a6 CRISPR-knockout clones .................................................40
2.2.4 Western blot analyses ............................................................................................40
2.2.5 RT-PCR analyses ...................................................................................................41
2.2.6 Sample preparation for immunoprecipitation ........................................................41
2.2.7 Protein immunoprecipitation .................................................................................42
2.2.8 Protein reduction, alkylation, trypsinization, and labelling ...................................42
2.2.9 Nanoscale HPLC-ESI tandem mass spectrometry .................................................42
2.2.10 Protein identification and quantification ................................................................43
vi
2.3 Results ................................................................................................................................44
2.3.1 Generation of a mammalian ZIP6 knockout EMT model .....................................44
2.3.2 ZIP6 forms a heteromeric complex with ZIP10 that interacts with NCAM1 ........46
2.3.3 NCAM1 serves as a hub for the assembly of focal adhesion complexes ..............52
2.3.4 ZIP6 influences the association of NCAM1 with specific interactors and
phosphorylation of NCAM1 at a GSK3 consensus site .........................................55
2.3.5 NCAM1 associates with integrins and actin-assembly complexes and is
polysialylated in a PrP and ZIP6-dependent manner during EMT ........................59
2.4 Discussion ..........................................................................................................................62
2.4.1 ZIP6 forms a heteromeric complex with ZIP10 that interacts with NCAM1 ........63
2.4.2 ZIP6 is critical for proper execution of EMT, presumably by facilitating
assembly of NCAM1 focal adhesion complexes ...................................................64
2.4.3 Binding of GSK3 to ZIP6 and phosphorylation of NCAM1 at GSK3
consensus sites .......................................................................................................64
2.5 Conclusions ........................................................................................................................66
Chapter 3 Future directions involving the investigation of the PrP-ZIP6/ZIP10-NCAM1
connection .................................................................................................................................69
References ......................................................................................................................................72
vii
List of Tables
Table 1.1: Subset of proteins observed in PrP 'ko' and 'kd' NMuMG global proteomes at levels
that deviated from 'wt' levels.
viii
List of Figures
Figure 1.1: Strategy for generation of mouse PrP knockout clones based on CRISPR/Cas9-
system.
Figure 1.2: Generation of Prnp knockout clones in three different mouse cell lines.
Figure 1.3: Flow-chart depicting experimental strategy for comparative analyses of the global
proteomes of Prnp knockout (or knockdown) and wild-type NMuMG epithelial cell clones.
Figure 1.4: PrP deficiency generated by CRISPR/Cas9-mediated gene knockout or stable
shRNA-mediated knockdown manifests in highly reproducible changes to the expression of
more than hundred proteins in NMuMG cell model.
Figure 1.5: PrP expression is upregulated during EMT, and PrP-deficiency decreases expression
levels of a subset of proteins undergoing pronounced expression levels changes during EMT,
including NCAM1 and its polysialylation.
Figure 2.1: Mouse NMuMG cell model for investigating role of ZIP6 during EMT.
Figure 2.2: The ZIP6 interactome.
Figure 2.3: Interaction of a ZIP6-ZIP10 heteromer complex with NCAM1 and GSK3.
Figure 2.4: The NCAM1 interactome.
Figure 2.5: Effect of ZIP6 on the NCAM1 interactome.
Figure 2.6: ZIP6 controls phosphorylation of the longest isoform of NCAM1 at a phospho-
acceptor site that conforms to a previously described GSK3 recognition site within members of
the Crmp protein family.
Figure 2.7: During EMT the shift of NCAM1-cytoskeletal interactions from a tubulin to an actin-
dominated environment is accompanied by its polysialylation.
Figure 2.8: Evolution of NCAM, ZIP, PrP, and polyST gene families.
ix
Abbreviations
4-VP 4-vinylpyridine
Aβ amyloid beta
AD Alzheimer’s disease
bp base pairs
BSE bovine spongiform encephalopathy
CFC cysteine-flanked core
CID collision-induced dissociation
CJD Creutzfeldt-Jakob disease
CRISPR clustered regularly interspaced short palindromic repeat
CWD chronic wasting disease
Dpl Doppel
EGTA ethylene glycol tetraacetic acid
EMT epithelial-to-mesenchymal transition
ESI electrospray ionization
FBS fetal bovine serum
FDR false discovery rate
FFI Fatal Familial Insomnia
GO gene ontology
GPI glycosylphosphatidylinositol
gRNA guide ribonucleic acid
HCD high-energy collision-induced dissociation
HDR homology-directed repair
HMM high molecular mass
HPLC high performance liquid chromatography
HRP horseradish peroxidase
Ig immunoglobulin
IP immunoprecipitation
IPI international protein index
iTRAQ isobaric tagging for relative and absolute quantitation
kd knockdown
KEGG Kyoto Encyclopedia of Genes and Genomes
ko knockout
Mm Mus musculus
MS mass spectrometry
MW molecular weight
N2a Neuro2a mouse neuroblastoma
NCAM1 neural cell adhesion molecule 1
NHEJ non-homologous end joining
ORF open reading frame
PBS phosphate buffered saline
PCR polymerase chain reaction
PL prion-like
PrP prion protein
PrPc cellular prion protein
PSA polysialic acid
PSM peptide spectrum matches
x
PVDF polyvinylidene fluoride
RT room temperature
RT-PCR real-time polymerase chain reaction
SDS sodium dodecyl sulfate
Sho Shadoo
shRNA small hairpin ribonucleic acid
siRNA small interfering ribonucleic acid
Slc solute carrier
TALEN transcription activator-like effector nucleases
TBST tris-buffered saline and Tween 20
TCEP tris(2-carboxyethyl) phosphine
TFA trifluoroacetic acid
TGFB1 transforming growth factor beta 1
TMT tandem mass tag
TPEN N,N,N’,N’-tetrakis (2-pyridylmethyl) ethylenediamine
ZFN zinc-finger nucleases
ZIP Zrt- Irt-like protein
1
Project rationale
This project builds upon the discovery of the evolutionary descent of prion genes from a ZIP
metal ion transport ancestor gene [38]. Its objectives are to investigate the physiological function
of PrP and its ZIP counterparts through a three-pronged approach in which knockout phenotypes
are observed, key interactors are examined, and comparisons between these ancestrally related
proteins are completed
Background
The prion protein (PrP) has been widely studied ever since its discovery as the causative agent in
mammalian prion diseases [3] such as bovine spongiform encephalopathy (BSE) in cattle,
chronic wasting disease (CWD) in deer, and most importantly human diseases such as
Creutzfeldt-Jakob disease (CJD) and Fatal Familial Insomnia (FFI). Though it is best known for
its role in disease, the cellular prion protein (PrPc) is widely expressed in healthy vertebrate cells
[4] and only develops its toxic effects when misfolding converts the original protein into its
diseased, ‘scrapie’ state (PrPSc) [3], prompting investigation into the physiological function of
the native PrPc. Despite extensive research, no clear function of PrPc has been uncovered,
however there has been no shortage of molecular and physiological phenotypes to which it has
been linked [4]. There have also been two mammalian PrPc paralogs identified, Doppel (Dpl)
[119] and Shadoo (Sho) [120], but characterizations of these prion protein family members have
also not yet pointed toward a conclusive biological role of PrPc [108]. Previously, our laboratory
established that the prion gene family originates from a partial duplication of an ancient ZIP (Zrt-
Irt-like Protein) metal ion transport gene [38]. Specifically, several lines of evidence converged
to the conclusion that a ZIP5/6/10-like ancestor gene must have given rise to the PrP founder
gene. Furthermore, it was determined that ZIP6 and ZIP10 were found to co-enrich with all three
members of the prion family in a neuroblastoma cell model [32].
The ZIP family of metal ion transporters consists of multi-spanning transmembrane proteins
which are known for their role in the import of divalent cations into the cytosol [94]. There are
fourteen ZIP proteins expressed in varying combinations in cells of humans and mice, and these
are coded by members of the solute carrier family 39a gene family (Slc39a1 to Slc39a14) [95].
Within this family of ZIP metal ion transporters, ZIP6 and ZIP10, as well as their nearest paralog
2
ZIP5, compose a distinct sub-branch that not only contains a prion-like (PL) domain but is most
closely related in sequence to prion genes [37]. The PL domains comprise their extracellular N-
terminal domains, contain a cysteine-flanked core (CFC), and have similar orientation and
positioning to their membrane anchorage sites as PrPc [37]. Furthermore, these ZIPs contain
histidine-rich imperfect repeat motifs N-terminal within their PL domains, similar to those seen
in PrPc [37]. Consistent with the evolution of PrPc from an ancestral ZIP gene, ZIP sequences
with ectodomains related to PrP are only found in the metazoan lineage, while members of the
broader ZIP protein family are highly conserved and can be found even in bacteria and plants
[37].
PrPc and ZIP metal ion transporters share the ability to bind divalent cations [94]. As the name
suggests, it appears that the main physiological function of the ZIP transporters is transporting
these divalent cations into the cytosol [133] while it remains unclear what the physiological
significance of PrPc’s ability to bind these metals may be. It has been suggested that PrP may
have a role in cellular responses to copper-induced oxidative stress [134] as well as a key role in
maintaining zinc homeostasis [135,136]. Another intriguing similarity between PrP and ZIP
transporters represent independent observations of gastrulation defects in zebrafish caused by
morpholino knockdown of either PrP [18,19] or ZIP6 [39]. Further characterization of this
phenotype revealed that the gastrulation defect was caused by an inability to complete the
morphogenetic program known as epithelial-to-mesenchymal transition (EMT) [18,19,39,98],
consistent with the interpretation that PrP and ZIP6 inherited at least a part of their functions
from a common ancestral gene.
Another protein that has been linked to EMT is the neural cell adhesion molecule 1 (NCAM1)
[36]. NCAM1 is a member of the immunoglobulin (Ig) super family [128], and is known to have
three predominant splice variants [129]: NCAM-180 and NCAM-140 which possess
transmembrane topology, and NCAM-120 which, like PrPc, is a glycosylphosphatidylinositol
(GPI)-anchored protein. The N-terminal ectodomain is conserved among all three NCAM1
isoforms, consisting of five Ig-like domains and two fibronectin type-III domains [130,131].
NCAM1 is also a known interactor of PrP [32,33,34], and of particular interest is that this
interaction stimulates neurite outgrowth [35], a process involving morphogenetic rearrangements
similar to those observed during EMT. Interestingly, a post-translational modification of
NCAM1 involving the addition of polysialic acid chain to N-glycan moieties within NCAM1’s
3
Ig-like domain 5 [132], leads to an altered hydrodynamic radius and embeds the respective
ectodomains in an overall negative charge cluster. These characteristics are critical for
facilitating cellular morphogenesis programs such as those associated with NCAM1 by altering
cell-cell and cell-matrix interactions. A prior investigation into the levels of polysialylated
NCAM1 in PrP-deficient mice displayed a reduction in polysialylation in the absence of PrP,
however this was attributed to altering the distribution of NCAM1 isoforms [35].
Hypothesis
The prion protein and ZIP6 contribute to epithelial-to-mesenchymal transition and related
morphogenetic programs by influencing the molecular environment and post-translational
modifications of NCAM1.
Specific aims
1. Generate knockout cell lines for PrP or ZIP6 in a cellular model, such as NMuMG mouse
epithelial cells, which can be induced to undergo EMT.
2. Examine global proteomic effects of PrP deficiency in relevant models and during EMT.
3. Characterize the expression of ZIP6 and its molecular interactions in an EMT model.
4. Examine the effects of PrP- or ZIP6- deficiency on the molecular environment and post-
translational modifications of NCAM1.
Scientific impact
North America is currently experiencing a rapid spread of one of the most aggressively
transmitted prion diseases in CWD. Though it appears highly unlikely that infectious cervid
prions can be transmitted to humans [137,138,139], there is still the underlying concern of
developing a subclinical disease or other novel manifestation of prion infection. Previous
outbreaks of BSE in Canada as well as the United Kingdom greatly raised public awareness of
prion diseases, illustrating its devastating effects not only on the population it infect, in this case
cattle, but also its grave effects on society and the economy. However, we have yet to see the full
effect of these outbreaks as we await the consequences of any variant CJD cases resulting from
the consumption of infected livestock. The uncertainty of the long-term effects of these prion
outbreaks, and incomplete understanding of the mechanisms by which the disease propagates
4
and crosses species barriers, underscores the importance of improving our understanding of prion
biology.
PrP has also been linked to Alzheimer’s disease (AD) in recent years due to the discovery that
PrP binds oligomeric amyloid beta, the key constituent of plaques in AD [9]. This emergence of
the cellular prion protein as a component of signaling cascades related to AD may present an
even more immediate need for understanding human prion biology. With AD and other
dementias becoming a growing concern for our aging population, it is estimated that in Canada
alone the costs associated with dementias will approach $33 billion a year while affecting nearly
1.4 million individuals by the year 2030 [140]. Though there is still controversy surrounding the
importance of PrP’s ability to bind amyloid beta [141,142,143], a thorough understanding of the
pathways downstream of this binding event and other interactors of PrP would clarify its
importance and role in AD.
With so many questions surrounding the prion protein remaining unanswered, this project will
contribute to the much needed understanding of the physiological function of PrP. Understanding
the function of PrP will provide insight into the pathways it affects as well as provide insight into
how these pathways are interrupted as a result of misfolded PrP. Determining the major players
in key pathways associated with PrP may elucidate novel therapeutic targets, while
understanding of how these pathways are altered in disease may clarify methods of disease
propagation and the associated phenotypic symptoms. The evolutionary descent of the prion
protein from a ZIP ancestor, as well as an apparent role during epithelial-to-mesenchymal
transition and related morphogenetic programming through its interactions with NCAM1 will
provide a number of new avenues to pursue, not only in the prion field, but also in the ZIP
transporter and NCAM fields of study. By connecting these fields of study, it will be possible to
look at the prion protein as it affects various pathways and diverse biological processes, both in
health and in disease.
5
Chapter 1 Effects of PrP-deficiency on the global proteome, and its role in
epithelial-to-mesenchymal transition
Please note that large parts of this chapter, excluding minor modifications, were published in the
following articles [1, 2]:
Mehrabian M, Brethour D, MacIsaac S, Kim JK, Gunawardana CG, Wang, H and Schmitt-
Ulms, G. (2014) CRISPR-Cas9-Based Knockout of the Prion Protein and Its Effect on the
Proteome. PLoS ONE 9(12): e114594.
Mehrabian M, Brethour D, Wang H, Xi Z, Rogaeva E, Schmitt-Ulms G (2015). The Prion
Protein Controls Polysialylation of Neural Cell Adhesion Molecule 1 during Cellular
Morphogenesis. PLoS ONE 10(8): e0133741. doi:10.1371/journal.pone.0133741.
Candidate’s role: Contributed to the generation and validation of knockout clones, and western
blot analysis.
Summary: A 1st generation CRISPR-Cas9 system was used to generate Prnp knockout cell lines
in mouse neuroblastoma Neuro2a cells, C2C12 mouse myoblast cells, and mouse mammary
gland NMuMG cells. Subsequently, deep quantitative global proteome analyses were undertaken
to begin to characterize the molecular consequences of PrP deficiency in NMuMG cells.
Expression levels for approximately 120 proteins were shown to correlate with the presence or
absence of PrP, with the majority belonging to extracellular components, cell junctions, or the
cytoskeleton. On the basis of its evolutionary link to ZIP metal ion transporters, we hypothesized
that PrP may contribute to the morphogenetic reprogramming of cells underlying epithelial-to-
mesenchymal transition (EMT). Consistent with this hypothesis, PrP transcription increased
tenfold during EMT, and PrP-deficient cells failed to complete EMT in NMuMG cells. Further
global proteomics analyses identified the neural cell adhesion molecule 1 (NCAM1) as a
candidate mediator of this impairment, which subsequently led to the observation that PrP-
deficient cells fail to undergo NCAM1 polysialylation during EMT.
6
1.1 Introduction
Ever since evidence mounted that the prion protein is the causative agent underlying prion
diseases [3], yet is widely expressed in healthy vertebrate cells of diverse lineages, scientists
have sought to uncover the physiological role of this protein [4]. The cellular prion protein
(PrPC) has been tied to diverse cellular activities ranging from cell adhesion to ion transport,
neuritogenesis, modulation of electrophysiological currents and circadian regulation (reviewed in
[5,6,7,8]) but the molecular mechanism of its proposed involvement in these and other activities
has remained largely enigmatic. It is a challenge to identify a prevailing theme in this body of
literature, a reality reflected in the widely held view that the role of this protein is complex,
multifaceted and context-dependent. The limitations of our current understanding of the
physiological role of PrP were further accentuated when it was proposed that PrPC plays a
critical role in a central signaling pathway in Alzheimer’s disease (AD) [9]. It is to be expected
that efforts to intervene with PrP’s pathogenic role in neurodegenerative diseases would benefit
from a thorough understanding of both the cellular programs that control its expression and the
principal signaling pathways that may contribute to toxic signals emanating from PrP.
When considering the relative merits of alternative approaches for determining the physiological
role of a given protein, three methods stand out, namely, one may choose to characterize the
phenotypic consequences of disease causing mutations, including gene knockouts [10], infer its
function from the function of its molecular interactors [11,12], or deduce it from the function of
its closest evolutionary relative [13].
To begin to address the physiological role of PrP, several Prnp knockout mouse models have
been generated and closely scrutinized for phenotypes [14]. At this time, more than a dozen
relatively subtle phenotypes have been reported in these PrP-deficient mice [4], yet it has been
difficult to connect observations because the study of the molecular underpinnings of these
phenotypes is hampered by the relative complexity of the experimental paradigms in which they
were observed.
One way of reducing complexity would be to investigate cell-specific phenotypes in different
cell models. The most often used and arguably best understood cell model for studying the
cellular biology of PrP is the mouse neuroblastoma cell line Neuro-2a (N2a) [15,16]. Recently,
mouse C2C12 cells, a cell line of myoblasts origins, were reported to provide an attractive
7
experimental paradigm for studying the cellular biology of PrP [17]. In light of previous reports
that document a role for PrP in morphogenetic rearrangements underlying epithelial-to-
mesenchymal transition (EMT) during zebrafish development [18,19], it would further be of
interest to explore the possible involvement of PrP in signaling pathways known to play a role in
EMT in a mouse epithelial cell line. Mouse mammary gland-derived NMuMG cells exhibit
epithelial morphology when cultured in standard growth medium but convert to a mesenchymal
phenotype upon prolonged exposure to transforming growth factor beta (TGFβ) and have
become a prime experimental paradigm for EMT-related studies. No PrP knockout models are
available for any of the aforementioned cell models.
Until a few years ago, genomic manipulations in mammalian cells posed a formidable challenge.
In recent years, genome editing methods such as those dependent on zinc-finger nucleases
(ZFNs) or transcription activator-like effector nucleases (TALENs) enabled the site-specific
generation of double-strand breaks. Once generated, powerful cell-encoded repair programs are
initiated that lead to the non-homologous end joining (NHEJ) of breaks or to their homology-
directed repair (HDR) in the presence of a template [20]. Despite the formidable advances ZFN
and TALEN technologies afforded, both methods proved somewhat cumbersome and require
considerable investment in time. A new technology was needed that addresses these
shortcomings.
Throughout their evolution several types of bacteria have acquired the ability to fight off
repeated attacks by the same virus by using an adaptive immunity. Whenever such a bacterium is
invaded by a virus, it deposits short genome segments of the attacking virus in a designated
location of its own genome. Should a consecutive attack by the same virus occur, the bacterium
can retrieve these virus-derived fragments not only to recognize the invader but also to direct a
molecular machine that can disarm the virus by cutting its genome [21,22,23]. Following more
than 10 years of work, during which the essential molecular components of clustered regularly
interspaced short palindromic repeat (CRISPR)-Cas9 machinery were defined, numerous reports
published in the past year documented how this adaptive immunity can be harnessed for genome
editing in a wide range of cells and organisms [24,25,26,27,28,29,30]. Utilizing this CRISPR-
Cas9 technology to generate PrP-knockout models of the aforementioned cell lines creates a base
upon which to study the physiological role of PrP in a cell-type specific context.
8
As previously mentioned, information regarding a protein’s physiological role can also be
elucidated by investigating its interacting partners and closest evolutionary relatives. Several
studies uncovered molecular interactors of PrPC or proteins residing in its spatial proximity
[31,32,33]. Cumulatively, these data suggest the prion protein is enriched in lipid raft membrane
domains and surrounded by several cell adhesion molecules, including NCAM1 and integrin or
non-integrin laminin receptors. NCAM1 seems particularly enriched amongst proteins residing
in proximity to PrP and this next-neighbor relationship can be captured by mild in vivo
formaldehyde crosslinking of cultured cells [32,34] or brain tissue [33]. The physiological
significance of this molecular proximity has remained largely unclear. One scenario sees PrP
cooperate with NCAM1 in its recruitment to lipid rafts and signaling to the tyrosine kinase FYN
[35]. In separate research, NCAM1 has been identified as a mediator of EMT [36].
Recently, several lines of evidence converged to reveal that the prion founder gene was derived
from an ancient ZIP (Zrt-, Irt-like protein) metal ion transporter gene [37] through a genomic
rearrangement that involved the genomic insertion of a spliced ZIP mRNA intermediate [38].
Several vertebrate genomes are known to code for more than a dozen ZIP transporter paralogs,
with ZIP5, ZIP6 and ZIP10 being most similar to PrP on the basis of their PrP-like ectodomains,
which share up to 30% sequence identity with PrP sequences in a subset of species [37]. The
morpholino-mediated knockdown of ZIP6 was observed to cause a gastrulation arrest phenotype
during zebrafish embryogenesis, characterized by a failure of cells to complete EMT and migrate
along the anterior-posterior axis [39]. Similar phenotypes of defective EMT-like cellular
migration programs, leading to perturbed gonad and trachea formation, have been reported to
occur in fruit flies deficient for a ZIP ortholog [40,41].
Due to the recurring presence of EMT as a key feature in each of these three strands of
observation, we hypothesized that PrP might play a role in EMT, and wondered if it affects the
execution of this process in mammalian cells, such as NMuMG cells.
This report describes the adaptation of this new CRISPR-Cas9 gene editing technology to ablate
PrP expression in N2a neuroblastoma cells, C2C12 myoblasts and NMuMG epithelial cells. To
begin to understand how PrP ablation changes the cellular proteome, a quantitative proteome
investigation of a PrP knockout NMuMG clone was undertaken, with cellular extracts from PrP
knockdown and wild-type NMuMG clones serving as positive and negative controls,
9
respectively. Our analysis revealed reproducibly altered the abundance levels of ∼120 proteins in
cells that exhibit no or reduced levels of PrP. A gene ontology analysis of these proteins strongly
indicates a role of PrP in cellular adhesion and differentiation and identifies a majority of these
proteins as belonging to the extracellular region, cell junction or the cytoskeleton. Furthermore,
we document that in NMuMG cells, PrP is more than tenfold upregulated during EMT and cells
deficient for PrP fail to complete EMT. Whereas wild-type cells undergo NCAM1
polysialylation during EMT, stable PrP-deficient cells fail to do so.
1.2 Materials and Methods
1.2.1 Antibodies and transforming growth factors
Immunoblotting made use of antibodies against NCAM1 (1:6666, 556324; BD Biosciences, ON,
Canada), PrP (1:2000, A03213; Bertin Pharma, France), and E-cadherin (1:4000, 3195; Cell
signaling, MA, USA).
Unless otherwise stated, cells undergoing EMT had TGFβ (240-B; R&D Systems, MN, USA)
added for 48 hrs at a concentration of 6.4 ng/ml and replenished with fresh medium after 24
hours.
1.2.2 Generation of gRNA Expression Vectors
The SpCas9 plasmid JDS246 (Plasmid 43861) and the gRNA expression plasmid MLM3636
(Plasmid 43860) were obtained from a non-profit plasmid share repository (Addgene,
Cambridge, MA, USA). Suitable CRISPR target sites within Prnp Exon 3 positive and negative
strands were identified using the ‘CRISPR Design Tool’ (http://crispr.mit.edu/). The respective
oligonucleotide pairs were obtained from Life Technologies (Burlington, ON, Canada) and were
customized to include overhangs compatible for ligation into MLM3636 linearized by digestion
with BsmB1 (R0580S; New England BioLabs, Ipswich, MA, USA), a cloning site located in this
vector on the 3′ side of a U6 promoter element. Oligonucleotides were phosphorylated with
polynucleotide kinase (EK0031; Fermentas, Ottawa, ON, Canada), annealed and inserted into the
gRNA plasmid using T4 DNA ligase (M0202S; New England Biolabs) and transformed into
Turbo competent E. coli (C2984H; New England Biolabs).
10
1.2.3 Cell Culture and Transfection
Mouse neuroblastoma Neuro-2a (N2a) (CCL-131) and mouse myoblast C2C12 cells (CRL-1772)
were sourced from the American Tissue Culture Collection (ATCC) (Manassas, VA, USA).
Mouse mammary gland NMuMG cells were, kindly, provided by Dr. Jeffrey Wrana (University
of Toronto, ON, Canada) but can also be sourced from the ATCC (CRL-1636). N2a and C2C12
cell lines were cultured as recommended by the ATCC distributor and NMuMG cells were
grown in Dulbecco's Modified Eagles medium (DMEM) supplemented with 10% FBS
(12484028, Life Technologies), 1% GlutaMAX (35050061, Life Technologies), 10 µg/mL
insulin (I9278-5ML, Sigma-Aldrich, Oakville, ON, Canada) and 1% antibiotic-antimycotic
solution (15240062, Life Technologies). Transfection with 3∶1 (w/w) mixtures of SpCas9 and
customized gRNA plasmids were carried out with the Lipofectamine 2000 or Lipofectamine
LTX (Life Technologies) according to the manufacturer's instructions and at cell culture plate
confluencies of approximately 75%. Forty-eight hours after transfection, cells were harvested,
diluted in cell culture medium to a level of 1 cell/mL and replated. Once individual colonies
formed, these were, initially, cultured in separate wells of 24-well plates and, subsequently,
further expanded in 6-well and 60 mm plates until cell numbers were sufficient for Western blot
analyses.
1.2.4 Generation of stable knockdown cell clones
The shRNA vectors (TRCN0000008471 and TRCN0000008472) against PrP were obtained from
the Thermo Scientific TRC Lentiviral shRNA Library (Ottawa, ON, Canada) and were co-
transfected into NMuMG cells using Lipofectamine LTX (Life Technologies). The selection
antibiotic, puromycin (P7255-25MG, Sigma Aldrich) was added to the medium 30h after
transfection to the final concentration of 1.1 µg/mL and medium changed every 1-2 days.
Following the selection, clonal isolation from the pool of stable cells was performed as described
above for the knockout cells. The stable cell clones were maintained and cultured with
puromycin at all times.
1.2.5 Genetic analysis
To sequence the genomic region targeted by CRISPR-Cas9-mediated Prnp editing, genomic
DNA was isolated from cells using QIAamp DNA Mini Kit (51304, Qiagen, Valencia, CA,
USA). DNA samples (20 ng/reaction) were amplified by polymerase chain reaction (PCR) using
11
primers (5′-TCTTTGTGACTATGTGGACTG-3′ and 5′-TGCCACATGCTTGAGGTTGGTT-3′)
that were predicted to anneal to Prnp gene sequences flanking the CRISPR target sites. The PCR
conditions were 94°C for 5 min, followed by 40 cycles of 94°C for 45 s, 59°C for 45 s, 72°C for
60 s, and 7 min at 72°C. The PCR products were analyzed on an ABI PRISM 3100 Genetic
Analyzer and visualized by DNA Sequencing Analysis 3.7 (Applied Biosystems).
1.2.6 Western blot analyses
Cells were grown to near confluency in 60 mm plates. Growth medium was removed and the
cells were rinsed twice with ice-cold phosphate buffered saline (PBS) before lysis with a buffer
consisting of 50 mM Tris, pH 8.0, 150 mM NaCl, 1% NP-40, and Roche Complete Protease
Inhibitor Cocktail. Insoluble cell debris was removed by 5 minute centrifugation at 14,000 rpm
and 4°C. The protein levels were adjusted based on a bicinchoninic acid colourimetric assay and
equal amounts of protein were separated on 4–12% or 12% Novex NuPAGE Bis-Tris (Life
Technologies) or 7% Tris-glycine gels (cast in-house), and transferred to a 0.45 micron PVDF
membrane. The blot membranes were blocked in 10% skimmed milk and probed overnight at
4°C with the respective antibody diluted in 5% skimmed milk. On the next day, the blots were
incubated with HRP-conjugated anti-mouse (1:5000, 170–6516; BioRad, ON, Canada) or anti-
rabbit (1:5000, 170–6515; BioRad) secondary antibodies and the ECL reagent (RPN2106; GE
Healthcare, Baie d'Urfe, QC, Canada). Signals were detected with either X-ray film or the Li-
COR Odyssey Fc digital imaging system (NE, USA).
1.2.7 Enzymatic characterization of post-translational modifications of NCAM1
Cell lysates adjusted for total protein content, were incubated overnight with 4 μL of PNGase F
(P0705; New England Biolabs), 1 or 4 μl of endo-N (AbC0020; ABC Scientific, CA, USA) or 4
μl of α2–3,6,8 neuraminidase (P0720; New England Biolabs) in a total reaction volume of 10 μl
at 37°C. In each case, the reaction proceeded in the presence of buffer solutions, which were
provided by the respective manufacturers together with enzymes.
1.2.8 RT-PCR analysis
RNA preparations were analyzed by a TaqMan gene expression assay targeting mouse Prnp
transcripts. Total RNA was extracted using the RNeasy Mini Kit (74104; Qiagen, ON, Canada)
12
and reverse transcribed to cDNA with oligo dT primers or random primers using the
AffinityScript Multiple Temperature cDNA Synthesis Kit (200436; Agilent Technologies, ON,
Canada). RNA integrity was checked on an Agilent 2100 Bioanalyzer (all samples were with
RIN>7). Real-time PCR analyses were then undertaken with these RNA preparations using
TaqMan Universal Master Mix II (4440038, Life Technologies) in triplicate to generate technical
amplification replicates of Hprt (Mm00446968_m1), Tfrc (Mm00441941_m1), and Prnp
(Mm00448389_m1) (Life Technologies) mRNAs. Amplification products were analyzed on an
ABI Prism 7500 system (Life Technologies). Relative quantifications were based on the qBASE
PLUS software (Biogazelle NV, Belgium) using the ddCt method after normalization to Hprt
and Tfrc mRNAs. The relative expressions of target transcripts were scaled to samples derived
from vehicle-treated NMuMG or C2C12 cells. Near-identical results were obtained with oligo
dT primers and random primers.
1.2.9 Sample preparation for comparative global proteomics analysis
For the comparative analyses of global proteomes NMuMG cell clones were employed, which
expressed wild-type levels of PrP, no PrP or stably reduced levels of PrP. For studying changes
during EMT, cells expressing wildtype levels of PrP with and without the addition of TGFβ, and
cells with stably reduced levels of PrP with TGFβ were employed. For each of these cell clones,
three 100 mm cell culture plates grown to near confluency served as starting material for
generating three biological replicates. Approximately, 5×106 cells were rapidly lysed with the
aid of 0.5 mm glass beads and a Mini-BeadBeater-8 (Biospec Products Inc., Oklahoma, USA) in
the presence of SDS-containing Lysis Buffer (2% SDS, 62.5 mM HEPES/NaOH, pH 8.0), which
had been preheated to 90°C. Following three cycles of 1 minute bead beading at
‘Homogenization Level’, samples were further incubated at 90°C to deactivate residual
enzymatic activities in the cellular extracts. Protein levels were adjusted by BCA colourimetric
assay (Thermo Scientific, Nepean, Ontario, Canada). Subsequently, disulfide bonds were first
reduced in 100 µg aliquots for 30 min at 60°C in the presence of 5 mM tris(2-carboxyethyl)
phosphine (TCEP), then alkylated for 1 hr at room temperature in the presence of 10 mM 4-
vinylpyiridine (4-VP). The samples were then acetone precipitated, the pellets washed with 90%
acetone and redissolved in 9 M urea. To ensure that the residual urea concentration did not
exceed 1.5 M, samples were diluted in Tetraethylammonium bromide buffer (TEAB) and then
digested with side chain-modified porcine trypsin overnight at 37°C. The covalent modification
13
of peptides with reagents from the 6-plex amine-reactive tandem mass tag (TMT) labeling kit
followed instructions provided by the manufacturer (Thermo Fisher Scientific, Waltham,
Massachusetts, USA).
1.2.10 Nanospray ionization tandem mass spectrometry
To ensure excellent proteome coverage, and to remove urea, buffer components and unreacted
TMT reagents, samples were ZipTip-purified on reversed-phase (C18) and strong cation
exchange resins. Eluents were adjusted to 0.1% formic acid and separated on 25 cm C18
nanocapillary columns (Acclaim PepMap RSLC with 100 Å pore size, 2 µm particle size, 75 µm
inner diameter) using an EASY-nLC 1100 system (Thermo Fisher Scientific). A linear four-hour
gradient of 2 to 95% acetonitrile in 0.1% formic acid was used to elute peptides and transfer
them by dynamic nanospray ionization to the ion transfer tube of an Orbitrap Fusion Tribrid
mass spectrometer. The acquisition method was designed with a view to maximize
parallelization and involved three scan types. First, a survey scan of the 400–2000 m/z range at
120,000 resolution was conducted in the orbitrap with the automated gain control target set to 2
e5. Next, the most intense precursor ion carrying two or more charges was isolated, subjected to
collision-induced dissociation (CID) and its fragments detected in the linear ion trap located at
the back end of the machine. Finally, the ten most intense fragments from this MS2 scan were
synchronously sent to the ion-routing multipole for higher-energy collision-induced dissociation
(HCD). The low m/z fragments obtained during MS3 fragmentation, including the TMT reporter
ions, were detected at 60,000 resolution in the orbitrap with the automatic gain control target set
to 1.0 e5 and the maximum injection time limited to 120 ms. The combined cycle time for this
method was set to 3 seconds during which as many precursors as possible, selected on the basis
of their intensity and in the order from most intense to lower intense, were subjected to this
processing scheme. Dynamic exclusion prevented the re-analysis of any precursor mass in a 20
ppm m/z window for the duration of 300 seconds.
1.2.11 Post-acquisition analyses
For the analyses of global proteome datasets without the addition of TGFβ, the international
protein index (IPI) mouse database (Version 3.87) was interrogated by Mascot (Version 2.4;
Matrix Science Ltd, London, UK) and Sequest HT search engines embedded in Proteome
Discoverer (Version 1.4; Thermo Fisher Scientific) and by integrated proteomics software
14
packages Scaffold (Version Q+; Proteome Software Inc., Portland, Oregon, USA) and PEAKS
Studio (Version 6.0; Bioinformatics Solutions Inc., Waterloo, Ontario, Canada). The percolator
algorithm [42] was applied to filter all CID spectra and select the subset of spectra that exceed a
false discovery rate target of 0.5. The mass spectrometry proteomics data have been deposited to
the ProteomeXchange Consortium [126] via the PRIDE partner repository [127] with the dataset
identifier PXD001301.
For the analyses of global proteome datasets with the addition of TGFβ, mass spectrometry raw
data files were analyzed by Mascot (Version 2.4; Matrix Science Ltd, London, UK) and Sequest
HT search engines embedded within Proteome Discoverer software (Version 1.4; Thermo Fisher
Scientific). All PSMs were based on queries of the mouse international protein index (IPI)
(release 3.87). Tolerance filters of 0.1 Da and 50 ppm were applied for searches of parent spectra
and tandem MS spectra, respectively (note that scatter plots depicting precursor mass errors as a
function of peptides scores indicated that observed masses of PSMs, which passed stringency
criteria, were largely within 10 ppm of theoretical masses). Database searches were configured to
allow for up to two missed tryptic cleavages. Because 4-vinylpyridine was used as the alkylating
agent, the search was configured to assume all cysteine side-chains were pyridylethyl-
derivatized. Variable modifications considered were TMT reagent modifications of primary
amines, phosphorylations of serines, threonines and tyrosines, deamidations of glutamine and
asparagine, and oxidations of methionines. For relative quantitation the low mass TMT signature
ion distributions were interpreted by an algorithm embedded in the Proteome Discoverer
software, which also generated raw data graphs depicting TMT ratios. Statistical analyses of
global proteomes were conducted with PEAKS (Version 6.0; Bioinformatics Solutions Inc.,
Waterloo, Ontario, Canada) and ProteinCenter (Thermo Fisher Scientific) software packages. A
stringent false discovery rate (FDR) of 0.5% was set as the initial filter, which had to be passed
by all PSMs. Because a reliable relative protein quantitation was critical for the interpretation of
data, a subsequent filtering process eliminated all proteins, which were not identified and
quantified on the basis of at least three PSMs for which TMT signature ion distributions were
available. The application of this filter eliminated false-positive identifications by exceeding
widely applied thresholds for protein group inclusion, which typically require assignments of
two PSMs per protein group for confident identifications. Because three biological replicates of
TGFB1-treated wt NMuMG cells were common to both proteomics analyses conducted, their
15
global proteomes served as a reference against which the other datasets were compared. The
Kruskal-Wallis test was applied to identify whether there is significant differences in the TMT-
based relative ratios of proteins amongst the two sets of three biological replicates within each
dataset. Since we could not assume Gaussian distributions for TMT ratios, this analysis was
based on a non-parametric version of the test (IBM SPSS Statistics, version 20, NY, USA). A
Spearman correlation analysis was conducted to determine whether there is a significant
correlation between the TMT levels of the 57 proteins whose relative expression levels were
most altered during EMT and following stable kd of PrP (Fig. 1.5). The mass spectrometry
proteomics data have been deposited to the ProteomeXchange Consortium [126] via the PRIDE
partner repository [127] with the dataset identifier PXD001875.
1.3 Results
1.3.1 Strategy of CRISPR-Cas9-based PrP knockout in three mouse cell lines
Any CRISPR-Cas9-based gene knockout experiment requires that consideration be given to the
choice of cell type, the reagents and strategy employed, the method for identifying positive
clones and the possibility of off-target effects. The selection of N2a cells, C2C12 and NMuMG
cells for this work was not only guided by the diverse systems biology these cell lines represent
as neuron-, muscle- and epithelial-like cell models but also by their shared murine origins
(enabling the use of identical reagents and facilitating comparative analyses across cell types)
and relative ease of transfection. Given the complex karyotype of N2a cells, which comprise
between 94 and 98 chromosomes in the stemline, and with anywhere between 59 and 193
chromosomes observed in individual subclones (American Type Culture Collection, Manassas,
Virginia), the generation of a PrP knockout in this cell line requires the concomitant genomic
editing of several Prnp copies. As of Spring 2013, the time at which this project was initiated, no
report documenting successful genome editing of this many alleles of a single gene was
available. Therefore, special consideration was given to the need to express all components of
the CRISPR-Cas9 system robustly. To this end, the two-plasmid CRISPR-Cas9 system,
consisting of plasmids for Streptococcus pyogenes Cas9 (SpCas9) and guide RNA (gRNA)
expression, was adapted from the Keith Joung laboratory (Massachusetts General Hospital, MA,
USA). This system drives the expression of a mammalian codon-optimized SpCas9 enzyme that
was engineered to carry a mammalian nuclear localization sequence from a cytomegalovirus
16
promoter. To further maximize the chance to produce the desired insertions or deletions (indels)
at CRISPR target sites, the genome editing step was intended to stimulate NHEJ, which is known
to proceed at a faster rate and is less cell cycle-dependent than genome editing approaches
relying on the HDR pathway [43]. For the design of gRNAs [44], consideration was given to
ensure CRISPR target sites map to an N-terminal stretch of the Prnp coding sequence within
Exon 3. Selected sites were predicted by the ‘CRISPR Design Tool’ (http://crispr.mit.edu/)
hosted by the Feng Zhang laboratory (Massachusetts Institute of Technology, MA, USA) [45] to
possess a minimum number of off-target sites in the mouse genome, with particular emphasis
placed on avoidance of cross-reactivity toward other coding sequences. Two sites were identified
which fulfilled these criteria and also targeted opposite strands of the coding sequence.
Following gRNA vector assembly using standard cloning procedures, host cells were co-
transfected with plasmids coding for SpCas9 and one of the two gRNAs derived from the two
target sites. In the absence of a suitable reporter or selection marker, the isolation of positive
clones had to rely on single cell isolations by the serial (limiting) dilution followed by both PrP-
specific Western blot analysis and DNA sequencing of genomic PCR products encompassing the
CRISPR target sites (Fig. 1.1).
17
Figure 1.1. Strategy for generation of mouse PrP knockout clones based on CRISPR/Cas9-
system.
1.3.2 Validation and characterization of PrP knockout cell clones
Whenever genomic double-strand breaks within exons are repaired by NHEJ, indels can be
generated that may shift the translational reading frame by 1, 2 or 3 nucleotides. In contrast to
single or double nucleotide frame shifts, the latter scenario most often merely gives rise to a loss
or gain of amino acids within the natural gene product, and only rarely produces translational
termination due to the insertion of a premature stop codon. Consequently, if no other limitations
would apply, no more than approximately 50% of the progeny derived from a susceptible diploid
18
parent clone could be expected to exhibit a complete loss of PrP expression. Perhaps not
surprisingly, the observed yield of PrP knockout progeny was considerably lower, amounting to
no more than ∼2% for C2C12 cells (1 clone amongst 44 clones tested) and ∼5% for NMuMG
cells (2 clones amongst 40 clones tested). No complete knockout of PrP was achieved in N2a
cells following a first round of transfection with the CRISPR-Cas9 plasmids and screening of 36
clonal isolates. A second round of transfection, however, led to ∼8.5% of N2a cell isolates (5
clones within 59 colonies tested) that exhibited no PrP expression by Western blot analysis (Fig.
1.2A).
19
Figure 1.2. Generation of Prnp knockout clones in three different mouse cell lines. (A)
Identification of clones that exhibit loss of PrP expression by Western blot analyses of cellular
lysates. Unspecific band are denoted with asterisks, while all other bands represent full-length or
cleaved forms of PrP. (B) Crude characterization of Prnp gene editing by genomic PCR analyses.
20
Note the appearance of additional slower or faster migrating PCR products in clones that exhibit
loss of PrP expression. C, control lane derived from genomic PCR analyses of cells not subjected
to CRISPR/Cas9 gene editing. The asterisk identifies the NMuMG clone that was employed for
the global proteome comparison described below. (C) Detailed insertion/deletion (indel) analysis
of one Prnp-deficient NMuMG cell clone by DNA-sequencing of genomic PCR products.
Consistent with the observed loss of detectable PrP by Western blot analysis, the analysis
established the deletion of 2 and 4 nucleotides within the two Prnp alleles present in this
NMuMG cell clone.
Consistent with the intended formation of indels, genomic PCR analyses of a segment of Prnp
Exon 3 comprising the CRISPR target sites revealed that a majority of PrP knockout clones gave
rise to products that were larger or smaller than the corresponding product seen in the respective
wild-type cell clones (Fig. 1.2B). Whenever such a change in the apparent molecular weight
could not readily be observed, DNA sequencing revealed the PrP knockout to be reliant on small
indels that are not expected to change the migration of the PCR product by agarose gel
electrophoresis (Fig. 1.2C). Taken together, this phase of the project validated the generation of
8 PrP knockout clones in the three mouse cell lines. Individual clones differed from each other
with respect to the CRISPR site that was targeted and the precise nature of their indels.
1.3.3 Workflow of global proteome comparison of PrP knockout (or knockdown) and wild-type NMuMG cells
To begin to explore how the loss of PrP expression may alter cellular protein homeostasis, a
global proteome analyses was conducted with one of the aforementioned CRISPR-Cas9-
generated NMuMG cell clones exhibiting a 47-nucleotide deletion and a 5-nucleotide deletion
(i.e., a 6-nucleotide deletion and a 1-nucleotide insertion) within its two Prnp Exon 3 alleles (Fig.
1.2B). These deletions are expected to give rise to truncated 62 and 76 amino acid Prnp gene
products due to the formation of premature translation stop codons generated by the two frame
shifts. To generate biological starting material for this experiment, cells were grown in standard
growth medium, a condition under which they are known to acquire an epithelial morphology,
and were rapidly lysed in the presence of pre-heated sodium dodecyl sulfate (SDS). Three
biological replicates were generated and as negative controls served naïve NMuMG cells. To
avoid possible confounders caused by run-to-run variance if samples and controls were to be
analyzed separately, trypsin digests of equal amounts of acetone-precipitated protein were
chemically labeled with isobaric tandem mass tags (TMT). This step ensured that all biological
21
replicates and controls could be combined and analyzed concomitantly. For the generation of
technical replicates, analyses were repeated three times for each of the samples. Eluates from the
separation column were directly transferred by dynamic nanospray ionization to an Orbitrap
Fusion Tribrid mass spectrometer which operated in a data-dependent fragmentation mode that
employed collision-induced dissociation (CID) to generate fragment ion spectra for the
identification of peptides and higher-energy CID (HCD) for the detection and relative
quantitation of TMT reporter ions.
A caveat of the strategy outlined above was that the secretion of N-terminal PrP fragments
derived from the expression of the CRISPR-edited Prnp gene with premature stop codons might
contribute to differences between the global proteome of PrP knockout and wild-type NMuMG
clones. Although there was a possibility that the cell autonomous nonsense-mediated decay
program [46] could recognize the respective mutated PrP mRNAs as faulty and target them for
destruction, this scenario seemed unlikely. It was dismissed largely because there is no exon-
intron boundary in the Prnp gene in 3′ proximity to the newly inserted nonsense codons, a
frequently observed requirement for triggering the nonsense-mediated decay quality control
program [47]. To nevertheless determine the extent to which changes to the global proteome
observed may have been caused by the loss of PrP (as opposed to be dependent on the abnormal
secretion of truncated PrP fragments or idiosyncrasies of the clone or knockout method) a
suitable control was needed. Note that the aforementioned caveat of N-terminally truncated PrP
constructs would also exist in other CRISPR-Cas9-derived PrP knockout clones produced by
employing the NHEJ pathway. To circumvent this confounder, an orthogonal set of three
biological samples was instead generated from an NMuMG cell clone, which had been derived
from the NMuMG parent cell line following its stable transfection with a plasmid coding for a
PrP-specific shRNA. This clone exhibited more than 75% reduction in PrP expression levels
when assessed by PrP-specific Western blot analysis, with relative band intensities measured on
a quantitative digital image scanner. As for the CRISPR-Cas9 PrP ko clone, the total proteome of
the stable PrP knockdown NMuMG cell clone was again compared to the total proteome of naïve
NMuMG cells using an identical workflow to the one described above (Fig. 1.3).
22
Figure 1.3. Flow-chart depicting experimental strategy for comparative analyses of the
global proteomes of Prnp knockout (or knockdown) and wild-type NMuMG epithelial cell
clones.
1.3.4 The global proteome of PrP-deficient NMuMG cells
The global proteome analyses of NMuMG cell extracts generated upward of 50,000 spectra for
each of the three technical replicates. To discriminate false positive hits from correct peptide
spectrum matches (PSMs) the percolator algorithm was used [42] and a false discovery rate
(FDR) target of 0.5% was applied, i.e., a peptide was only considered identified when it met this
stringent filter criterion. Indicative of consistent depth of coverage and data quality in PrP
knockout and knockdown datasets, a similar number of ∼33,000 PSMs passed this filter amongst
all acquired spectra in both datasets (Fig. 1.4A). For any protein to be considered for TMT-based
quantitation, more than three unique PSMs had to support its identification and provide reporter
ion profiles. Because a relatively wide isolation window of 2 m/z was applied during the
selection of parent ions and the tryptic proteome digests were of high complexity, a subset of
spectra were observed to be contaminated with fragments which resulted from co-isolated peaks.
Therefore, an additional filter was applied, which prevented PSMs from being considered for
quantitation when the intensity of an inadvertently co-isolated parent ion exceeded 30% of the
intensity of the selected parent ion. For a protein to be listed, the ratio of sample/control reporter
23
ion intensities further had to be consistently up or down in all three biological replicates. The
application of these criteria resulted in a total of 3254 protein groups that were confidently
identified and relatively quantified across all biological and technical replicates. Note that the
term ‘protein group’ refers to the fact that, frequently, peptides detected did not uniquely identify
a specific protein but were shared amongst several protein entries in the IPI database, which
caused them to be recorded as a single protein group. This scenario did not only arise from
protein isoforms but was occasionally also observed with gene products of paralogs or with
database entries that share highly conserved modular domains, suggesting that the true number of
proteins quantified must have exceeded the number of protein groups. Because the emphasis of
this work was not on identifying the largest number of proteins, no attempt was made to further
distinguish proteins within proteins groups, which hereafter are simply referred to as ‘proteins’.
Amongst all proteins identified in the PrP knockout dataset, 201 were consistently upregulated
(>1.1 ratio) or downregulated (<0.9 ratio) on the basis of the median ratio or their intensity levels
(PrP knockout levels/control levels). Similarly, the stable knockdown of PrP caused 200 proteins
to be shortlisted in this manner. Importantly, a majority of 120 protein groups were changed in
both of the two datasets, arguing that this subset represents the most confident candidates for
proteins whose abundance levels in NMuMG are influenced by the presence or absence of PrP
(Fig. 1.4B). Although the abundance level changes observed were modest for a majority of these
proteins, and in no instance exceeded a three-fold change, solid statistics obtained from the
recording of large numbers of reporter ions provided confidence that even smaller levels of
enrichment were real and not an artifact of the method. Furthermore, none of the proteins whose
abundance levels were observed to be changed in the PrP knockout clone relative to wild-type
cells could be matched to the list of genes that were predicted to harbor CRISPR off-target sites.
Finally, amongst the 120 shortlisted proteins the expression of only three proteins
(sodium/potassium-transporting ATPase beta-1, pantetheinase precursor and ferritin light chain
1) did not follow the same trend when knockout and knockdown datasets were directly compared
(Table 1.1).
24
Figure 1.4. PrP deficiency generated by CRISPR/Cas9-mediated gene knockout or stable
shRNA-mediated knockdown manifests in highly reproducible changes to the expression of
more than hundred proteins in NMuMG cell model. (A) Chart depicting number of peptide-
spectrum matches versus false discovery rate (FDR). 33,132 and 33,969 peptide-to-spectrum
matches (PSMs) passed the set FDR threshold of 0.5 in PrP knockout and knockdown datasets,
respectively. (B) Venn diagram depicting overlap in proteins (peptides) whose expression was
altered in clones made deficient for PrP expression by the two aforementioned methods. (C)
25
Several gene ontology (GO) classifiers were significantly overrepresented when their occurrence
was compared amongst the 120 proteins whose abundance levels were consistently altered by
PrP knockout (or knockdown). The list of all 3254 mouse NMuMG protein groups detected in
this study served as the reference data set for these analyses. Colour code: dark brown, proteins
whose abundance levels were changed in both PrP kd and ko data sets; grey, subset of proteins
within reference data set assigned to a given ‘GO’ classifier.
Table 1.1. Subset of proteins observed in PrP 'ko' and 'kd' NMuMG global proteomes at levels
that deviated from 'wt' levels.
26
27
From a cursory examination of the datasets, it is apparent that the levels of several intermediate
filament proteins (e.g., cytoskeletal keratins 7, 8, 14 and 19) and constituents of cell-cell
junctions (e.g., cadherin-1, periplakin, desmoplakin, plakoglobin, catenin alpha-1 and beta-1)
were higher in wild-type than in PrP-deficient NMuMG cells. A more systematic analysis of this
protein list on the basis of gene ontology (GO) annotations revealed ‘Biological Processes’
related to cell adhesion, epithelial cell differentiation and response to inorganic substance were
significantly enriched amongst proteins whose abundance levels were altered in PrP-deficient
NMuMG cells (Fig. 1.4C). GO assignments to ‘Molecular Functions’ and ‘Cellular
Components’ overrepresented amongst these proteins further characterized the phenotype of PrP-
deficient NMuMG cells as being compromised in its cell adhesion and cytoskeletal biology.
There are other protein expression profiles that stand out in the datasets collected. For instance,
protein abundance levels of calpains were higher in wild-type than in PrP-deficient cells and,
perhaps as a consequence, the levels of calponins (i.e., calponin 2 and 3), a well-known substrate
of calpains, followed the opposite trend. Finally, the annexin protein family was represented by
several members in the list of proteins whose abundance levels were changed, yet no consistent
trend was observed amongst these annexins, i.e., expression levels of annexins A13 and A1 were
observed to be robustly upregulated and, in striking contrast, annexin A6 was observed at lower
levels in wild-type cells than in PrP-deficient cells.
1.3.5 PrP controls NCAM polysialylation during EMT
Mouse mammary gland epithelial cells (NMuMG) [48] are a widely used model for studying
EMT because this cell line responds robustly to TGFβ exposure with morphogenetic
reprogramming that bears all hallmarks of EMT [49]. Consistent with prior reports, persistent
exposure of NMuMG cells to TGFβ stimulated the expected transdifferentiation, with cadherin
levels declining as cells lose their adherens junctions (Fig. 1.5A). As early as 6 h into the time-
course of TGFB1 exposure, PrP protein levels increased and continued to climb until 48 hrs,
when a peak level was reached that exceeded levels in mock-treated cells more than 6-fold (Fig.
1.5A).
To investigate if EMT-associated changes in PrP levels had arisen from increases in
transcriptional activity of the Prnp gene or changes to PrP translation or turnover, real-time PCR
(RT-PCR) analyses were undertaken. A robust accumulation of Prnp-transcripts was observed
28
during the time-course of TGFβ treatment, with fifteen-fold higher Prnp expression after three
days of treatment compared to transcript levels in untreated cells (Fig. 1.5B).
To address if transcriptional activation of PrP is essential for EMT or represents merely a
correlative phenomenon, we next compared the execution of this program in NMuMG wild-type
(wt) cells or derivative clones whose levels of PrP expression were diminished by CRISPR/Cas9
knockout (ko) technology or stable knockdown (kd) of PrP transcripts (Fig. 1.5C). When these
alternative PrP-deficiency models were monitored before and after exposure to TGFB1, several
observations were made: (i) PrP-deficient NMuMG cells exhibited defects in cell-cell contacts
even prior to EMT-induction. (ii) Upon addition of TGFB1, PrP-deficient cells acquired a less
pronounced fibroblastoid phenotype than wt cells.
29
30
Figure 1.5. PrP expression is upregulated during EMT, and PrP-deficiency decreases
expression levels of a subset of proteins undergoing pronounced expression levels changes
during EMT, including NCAM1 and its polysialylation. (A) Western blot analysis of E-
cadherin and PrPC protein levels in NMuMG cell extracts during 72 h of exposure to TGFB1.
(B) Profound upregulation of Prnp gene transcription accounts for increased PrPC protein levels
during EMT based on a time-course RT-PCR analysis of PrP transcripts in NMuMG cells
following addition of TGFB1 to the cell culture medium. (C) Comparison of E-cadherin and PrP
protein levels in wt NMuMG cells and PrP-deficient derivative cell clones obtained by CRISRP-
Cas9-based PrP knockout or stable shRNA-based kd. The ‘negative control’ represents a cell
clone which had been subjected to identical CRISPR-Cas9-based Prnp knockout procedures but
did not result in a PrP knockout. (D) List of proteins exhibiting >20% level differences in
comparison of global proteomes of TGFB1-treated stable PrP kd versus wt NMuMG cells
(dataset II). Coverage: percentages of primary structure of covered by peptide-to-spectrum
matches; # Peptides: number of peptides matched to a given protein entry (note that instances of
the same peptide being identified with different modifications counted separately in this tally);
Count: number of TMT signature ion distributions, which passed stringent filtering criteria and
were used for relative quantitation. (E) A post-translational modification of NCAM1 is missing
in cells expressing no or low levels of PrP. Western blot analysis of selected NMuMG cell
extracts revealed increased total levels of NCAM1 in all cell clones upon 48 h TGFB1 exposure.
Whereas cells expressing wt levels of PrP give rise to a continuous pattern of NCAM1 signals,
PrP-deficient cells exhibit more distinct NCAM1 bands, whose masses correspond to the
expected masses of the three predominant NCAM1 isoforms. Note that the PrPC band pattern
observed in NMuMG cells tends to be more complex than the corresponding pattern in, for
example, the Neuro2a cell model, possibly reflecting a greater heterogeneity of its N-glycans in
these cells. (F) Stable PrP-deficiency impairs polysialylation of NCAM1 at N-glycan acceptor
sites. To characterize the post-translational NCAM1 modification lacking in PrP-deficient cells,
extracts from wt or stable PrP kd NMuMG cells, which had been treated with TGFB1 for 48 h,
were subjected to enzymatic digestion with glycosylases known to remove terminating sialic
acids (exo-N), cut polysialic acid chains (endo-N) or hydrolyze the linkage of N-glycan groups
to asparagine side-chains within ‘NxS/T’ acceptor sites (PNGase F). Note that complete removal
of N-glycans abolishes the discriminating NCAM1 modification.
To identify proteins that contribute to EMT in NMuMG cells and are affected in PrP-deficient
clones, we next conducted two comparative global proteome analyses using similar logic to that
described in Figure 1.3. Whereas the first analysis was intended to reveal proteins whose levels
are altered in wt cells during EMT (dataset I), the second analysis addressed the question which
of the proteins, present in TGFB1- treated NMuMG cells, are affected following stable
knockdown of PrP transcripts (dataset II) (Fig. 1.5D). We determined the overlap amongst the
lists of proteins whose levels were most affected by TGFB1 induction or stable PrP kd, and
identified 57 proteins that appeared in both datasets top 200 most changed. Although this
established a highly significant connection of PrP to proteins whose levels changed in EMT, the
31
direction of protein levels changes in cells which underwent EMT (dataset I) was not correlated
to the protein level changes caused by stable PrP-deficiency in TGFB1 treated cells (dataset II)
(Spearman correlation; ƿ = 0.419). This result suggested that stable PrP kd does not impair EMT
by shifting the cellular proteome towards a more epithelial or mesenchymal phenotype but rather
disturbs the natural balance of EMT-related proteins. Close inspection of the datasets flagged
NCAM1 as a protein of interest, because it underwent profound induction during EMT, its levels
differed by more than 20% in a comparison of TGFB1-treated wt and stable PrP kd cells (Fig.
1.5D), and the protein was a next-neighbor of PrP also in this cell model (not shown) [34].
Validating the mass spectrometry results, TGFB1 treatment of wt NMuMG cells caused a
pronounced upregulation of NCAM1 protein levels in western blot analyses. Intriguingly, high
molecular mass (HMM) bands of diffuse appearance, which were reactive to the NCAM1
antibody and intensified during the course of EMT in wt NMuMG cells, were largely absent in
PrP-deficient cells (Fig. 1.5E).
NCAM1 is known to undergo several well-characterized post-translational modifications, and is
the predominant acceptor of polysialic acid modifications in the brain. Treatment of cellular
extracts with PNGaseF completely removed the HMM smearing, consistent with it originating
from N-glycan moieties (Fig. 1.5F). Treatment with endoneuraminidase (endo-N), a class of
enzyme specific for the endoglycolysis of (2→8)-α-sialosyl linkages, reduced the most slowly
migrating HMM signals only observed in PrP-expressing cells, indicating that polysialic acids
are responsible for their presence. A less pronounced but visible reduction of the HMM smear
was achieved by 90°C heat treatment, a method known to partially remove polysialic acid
residues [50]. Finally, treatment with an exo-neuraminidase, which removes terminal sialic acid
residues but does not remove longer polysialic acid chains, had only a minor effect on the
smeared HMM component of NCAM1 signals but caused a slight shift of the most prominent
NCAM1 bands to faster migrating species in both PrP-expressing and PrP-deficient cells (Fig.
1.5F). These data established that PrP does not affect the addition of N-glycan core structures,
including the addition of short terminal sialic acids. Instead, PrP’s influence on NCAM1
glycosylation relates specifically to its polysialylation.
32
1.4 Discussion
This may be the first report describing the successful knockout of PrP by CRISPR-Cas9
technology. Data presented established that the methodology applied can not only achieve a PrP
knockout in diploid C2C12 myoblasts and NMuMG epithelial cells but also in cells known to
possess a highly complex karyotype, such as the N2a neuroblastoma cell model. To begin to
characterize the impact PrP deficiency has on the cellular proteome, a deep global proteome
analysis was conducted that identified more than 120 proteins whose abundance levels are
modulated by PrP in the NMuMG cell model. In concordance with proposed roles of PrP in cell
adhesion, the proteins whose levels were most affected by its knockout or knockdown are known
to contribute to the organization of extracellular matrix, cell-cell junctions and the cytoskeleton.
Further investigations into the NMuMG model and its ability to undergo EMT uncovered that
PrP-deficient cells undergo a proteome shift which affects preferentially the levels of proteins
associated with this morphogenesis program, thereby precluding its proper execution and
perturbing NCAM1 polysialylation.
1.4.1 CRISPR-Cas9 generated PrP-knockout cells are a viable model
The study made use of a first-generation CRISPR-Cas9 system that relied on stimulating the cell
autonomous NHEJ program for the repair of double strand breaks and concomitant generation of
indels. During the course of this study, a number of reports were published that investigated
specificity constraints of this system and established a risk to generate off-target effects
[45,51,52,53,54]. Several measures were taken to address this risk: (i) CRISPR-target sites were
selected by an algorithm that suggests sites with minimal risk to generate off-target effects; (ii)
individual knockout clones were generated with one of two different CRISPR target sites; (iii)
the global proteomic analyses were undertaken with clones generated by CRISPR-Cas9-based
knockout and shRNA-based knockdown of PrP expression; and (iv) a comparison of the list of
proteins whose levels deviated between CRISPR-Cas9-based PrP-deficient and wild-type
NMuMG cells with the list of predicted CRISPR off-target sites revealed no match.
The current study is not the first to have generated in vitro models for studying how PrP
deficiency affects the biology of cells. Indeed, there is no shortage of reports that use one of
several alternative approaches, including the use of primary neurons [55] or cerebellar granule
cells [56] harvested from PrP-deficient mice, and the comparison of cells that are naturally
33
devoid of PrP (e.g., Cos7 and CHO cells) [57,58] or express low levels of PrP (e.g., NIH-3T3
and LM-TK cells) [59] versus their derivatives following PrP overexpression. The use of
CRISPR-Cas9-derived PrP-deficient models is not expected to replace all of the aforementioned
approaches but should complement them. In particular, the PrP-deficient N2a clones should not
only facilitate efforts to study the cellular biology of PrPC but also be helpful for ongoing
research aimed at defining the constraints underlying its conversion in prion diseases. In past
studies employing this cell line, researchers frequently had no choice but to over-express
heterologous PrP constructs equipped with epitope tags [60]. Often these studies were plagued
by the presence of endogenous PrP, which might have manifested, at best, as a nuisance or
technical challenge or, worse, prevented studies altogether by contributing to species barrier
effects [61].
1.4.2 Benefits of a cell-specific, mass spectrometry based approach
Previous attempts at identifying differences in the global proteome of wild-type and Prnp
knockout cells or tissue have led to mixed results. For instance, a comparison of the relative
abundances of 1131 brain proteins by two-dimensional differential gel electrophoresis revealed
no significant differences [62]. It is conceivable that existing differences between wild-type and
knockout cells may be masked when a multitude of cell types present in a tissue are
concomitantly analyzed by such a shotgun approach. A plausible scenario that could hinder the
discovery of PrP-dependent changes would, for instance, be the presence of opposite trends in
the expression of a given protein in different cell types. Similarly, a change could be masked if it
exists in only a small subset of cells. These limitations might be overcome if global proteome
comparisons are undertaken on homogenous preparations of cells. In fact, when cellular extracts
from PrP overexpressing and wild-type SH-SY5Y cells were compared, more than a dozen
changes in relative protein abundances were reported following 2D-gel electrophoresis analysis
and silver-staining of protein spots [63]. A similar number of differentially expressed proteins
were also observed in a recent comparison of the membrane proteomes of primary cerebral
granule cells derived from Prnp knockout or wild-type mice [64]. Perhaps surprisingly, there is
no overlap in the proteins whose expression was observed to be changed in these two paradigms
or in the current study. This could be attributed to cell type-specific effects of Prnp ablation or to
differences inherent to the methodologies employed.
34
The global proteomic analyses in this report benefited from mass spectrometry instrumentation
whose specifications exceeded experimental setups available for previous studies. It therefore is
not surprising that the current analyses afforded deeper coverage of the global proteome. It might
be expected that this mere increase in depth of coverage would lead to more candidates.
However, the higher proteome coverage alone is most likely not the reason for the relatively
large number of proteins whose abundance levels were observed to be changed in a PrP-
dependent manner in this study. This conclusion is based on the observation that a global
proteome analysis of PrP knockout mouse brains conducted in parallel and with identical
instrumentation led to a comparable depth of proteome coverage, yet failed to reveal a similar
number of differences in expression levels (not shown).
The current study avoided gel-based workflows in order to minimize the chance to introduce
inadvertent variances due to complex sample handling workflows. In contrast to 2D gel-based
approaches, which rely for relative quantitation on intensity comparisons of stained protein spots,
mass spectrometry-based quantitation offers critical advantages for quantitation, namely, (i) the
concomitant in-solution digestion of all proteins present in the cellular extract generates a
powerful reference against which the abundances of individual proteins can be compared; and
(ii) because each protein is identified on the basis of several PSMs, more than one quantitation
data point is generated per protein. As a result, even small changes in expression levels can
produce statistically significant hits when a given protein is identified and repeatedly quantified
on the basis of a large number of peptides.
1.4.3 Insights into the physiological role of PrP
The NMuMG model emerged as a promising paradigm for studying signaling upstream and
downstream of PrP in this work. For a protein whose expression has long been considered to be
under the control of ‘housekeeping’ promoter elements [65,66,67], an unexpected observation
was the more than tenfold transcriptional upregulation of endogenous PrP that accompanies
EMT in NMuMG cells. Insights into the cellular programs that govern PrP expression are
expected to provide novel angles for devising prion disease interventions [68]. Experimental
models that facilitate the dissection of these upstream signaling pathways are therefore urgently
needed. Although several transcription factors are known to act upstream of PrP in some
experimental paradigms [69], the details of the regulation of the Prnp promoter and the broader
35
context that drives its transcription have remained murky. Similarly, whereas the next-neighbor
relationship of PrP and NCAM had been established for some time [34], the coinciding
upregulation of their expression during EMT was not anticipated. Even more surprising is the
apparent PrP-dependent polysialylation of NCAM1 that occurs during EMT.
Though there are many functions attributed to PrP in the literature, the role of PrP in EMT has
some objective characteristics, which make it uniquely stand out: (1) Its identification was
precipitated by a research trajectory that built on insights gained from studying PrP’s relationship
to its evolutionary ancestors. (2) Its key components, NCAM polysialylation and PrP, evolved
around the same time and are restricted to vertebrates. (3) It reflects a role of PrP, which in
NMuMG cells is preceded by a more than tenfold transcriptional upregulation of its protein
levels, indicative that it might not represent a bystander phenomenon but is central to its
biological role. (4) It centers on NCAM, the predominant next-neighbor of mature PrP at the
plasma membrane. (5) Its consequences are profound, as it generates a major change to a post-
translational modification, which can by itself influence several aspects of cell biology, including
cellular migration, cell-cell adhesion, and others (discussed further below).
It is to be anticipated that the influence of PrP on NCAM polysialylation is not restricted to the
cell models or morphogenetic EMT program studied here. It is increasingly recognized that cells
put to use branches of this program also in other cell fate decisions. Prominent cellular activities
that have been associated with NCAM1 polysialylation are cell migration, neurite outgrowth
[70,71], including mossy fiber pathfinding [72], hematopoietic stem cell differentiation [73,74],
as well as AMPA [75] and NMDA receptor modulation [76,77]. NCAM1 polysialylation has
further been shown to play a role in circadian rhythm regulation [78,79,80,81,82], myelin repair
[83,84,85,86] and neurogenesis in both the subventricular zone and the dentate gyrus within the
hippocampal formation [87,88,89]. Readers versed in the literature on PrP function will
recognize that roles in all of these biological processes have also been attributed to PrP [4,90].
Careful further investigation will tell, which of the reported phenotypes will hold up upon close
scrutiny, and if the connection between PrP (and possibly other ZIPs with a PrP-like ectodomain)
and polysialylation is sufficient to explain them. Importantly, if it turns out that the ability of PrP
to regulate polysialylation is central to its function, we submit that it still would not constitute a
satisfying description of the function of PrP. In our view, a meaningful functional annotation of a
protein encompasses both the larger program its activity/presence contributes to and the
36
immediate molecular mechanism by which it contributes. On this account, regulation of
polysialylation might turn out to be the larger program PrP serves in several paradigms but the
detailed mechanism of its contribution remains to be elucidated.
1.5 Conclusions
It is anticipated that cell clones described in this report will be useful to a research community
studying the cellular biology of the prion and its involvement in neurodegenerative diseases. The
CRISPR-Cas9-based procedure described here can now easily be adapted for the generation of
additional PrP ko clones in other cell lines. In future studies, the risk of off-target effects could
be further reduced by employing a second generation Cas9-derived nickase for genome editing
and stimulation of the cell autonomous HDR pathway [51,52,53,91,92]. These new tools should
hopefully facilitate the identification of both the cellular function of PrP and signaling pathways
critical for neurotoxicity in AD and prion diseases. Our investigation into PrP-deficient NMuMG
cells has already proven to be a particularly intriguing paradigm for studying signaling
downstream of PrP, unveiling PrP as a mediator of NCAM1 polysialylation during EMT. With
investigations into PrP and polysialylation biology being mature fields, we hope that this
connection will stimulate progress in both fields of study.
37
Chapter 2 A ZIP6-ZIP10 heteromer interacts with NCAM1 controlling its
phosphorylation and integration into focal adhesion complexes during epithelial-to-mesenchymal transition
Please note that large parts of this chapter, excluding minor modifications, have been submitted
for publication in the following article [93]:
Brethour D, Mehrabian M, Ehsani S, Williams D, Xi Z, Rogaeva E, Schmitt-Ulms G (In prep).
A ZIP6-ZIP10 heteromer interacts with NCAM1 controlling its phosphorylation and integration
into focal adhesion complexes during epithelial-to-mesenchymal transition.
Candidate’s role: Contributed to the generation of knockout clones, interactome analyses, and
western blot analyses.
Summary: To investigate the mechanism by which ZIP6 affects the process of EMT, a 2nd
generation CRISPR-Cas9 system was utilized to generate a ZIP6 knockout in NMuMG cells. A
co-immunoprecipitation of ZIP6, using the knockout as a negative control, identified ZIP10 and
NCAM1 as key interactors during EMT. Knockdown studies of ZIP6 or ZIP10 further revealed
that these two transporters co-regulate each other’s expression, suggesting that a heteromer
forms within the cell. Due to its role as a mediator of EMT, and it’s binding to both ZIP6 and
PrP, a co-immunoprecipitation of NCAM1 was completed. It revealed a dramatic shift in
interactors during EMT, with interactions to focal adhesion complexes and actins increasing,
while interactions to proteins associated with microtubules decrease. Interestingly, the absence of
ZIP6 impairs NCAM1’s ability to integrate into these focal adhesions and prevents the
phosphorylation of NCAM1 on a cluster of cytosolic phosphoacceptor sites, but does not abolish
polysialylation of NCAM1. Combined with bioinformatics analyses, the data suggests a model
by which PrP inherited its ability to bind NCAM1 from its ZIP transporter ancestors, but
acquired a unique fitness advantage with its ability to control the polysialylation of NCAM1 with
the emergence of polysialyltransferases.
38
2.1 Introduction
ZIP metal ion transporters are an ancient family of multi-spanning transmembrane proteins
tasked with the import of divalent cations into the cytosol [94]. Humans and mice express
fourteen ZIP proteins that are coded by members of the solute carrier family 39a gene family
(Slc39a1 to Slc39a14) [95]. Beyond their role in divalent cation import, relatively little is known
about their function, regulation or protein interactions. However, two ZIP paralogs, namely ZIP6
and ZIP10, were observed to co-enrich with PrP [32]. As mentioned in Chapter 1, subsequent
multi-disciplinary analyses revealed that ZIP transporters and PrP not only bind to each other but
share a common ancestry [96,97,37]. More specifically, the PrP founder gene can be traced to an
evolutionary event that occurred during vertebrate speciation. Mechanistically, it can be deduced
to have involved the retro-insertion of a spliced and C-terminally truncated transcript of an
ancestral ZIP transporter with similarity to ZIP6 and ZIP10 [38]. Intriguingly, these discoveries
were preceded by independent observations, which indicated that both morpholino-based
knockdown of ZIP6 [39] or PrP [19] cause a similar gastrulation arrest phenotype in zebrafish
embryos. A closer characterization of the shared phenotype characterized it as a failure to fully
execute the morphogenetic program, EMT [39,98]. A parsimonious interpretation of this finding
was that ZIP6 and PrP are not only homologous but PrP may have inherited from its ZIP
ancestor at least part of its function. This led us to the work presented in Chapter 1, which
considered whether an involvement of the prion in EMT is conserved in mammalian cells. In
brief, consistent with this scenario, transcript and protein level of the prion protein are
profoundly upregulated during EMT in mouse NMuMG epithelial cells [1]. Moreover, the
CRISPR/Cas9-based knockout of PrP partially impaired EMT in this model [2]. Comparative
global proteome analyses of wild-type and PrP-deficient cells then highlighted NCAM1, a
known binder of PrP [34], as a key player in EMT, whose expression is affected in PrP-deficient
cells [2]. Closer analyses of NCAM1 revealed that PrP not only influences steady-state levels of
NCAM1 but also controls its polysialylation. Surprisingly, this effect of PrP on a key NCAM1
modification was not reliant on PrP directing the St8sia2 polysialyltransferase responsible for
NCAM1 polysialylation to its NCAM1 substrate but, instead, placed PrP upstream of a signaling
loop that controls St8Sia2 transcription [2]. Taken together, this body of work 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)
39
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 generated ZIP6 knockout clones, investigated its functional
involvement in EMT and conducted an unbiased ZIP6 interactome analysis. We will show that
ZIP6 transcript and protein levels are dramatically upregulated upon induction of EMT.
Moreover, we observed that ZIP6 forms a heteromeric complex with ZIP10, whose predominant
interactor is NCAM1. Subsequent in-depth interactome analyses of NCAM1 revealed a critical
role for ZIP6 in the assembly of focal adhesion complexes and suggest one mechanism by which
ZIP6 may regulate the assembly of these complexes. Finally, we will document a minor
influence of ZIP6 on NCAM1 polysialylation. The data will be discussed in an evolutionary
context, highlighting how different selective pressures may have led to specialized adaptations of
ZIP6 and PrP within the same overall developmental program centered on NCAM1
2.2 Materials and Methods
2.2.1 Antibodies and siRNAs
For immunoblotting, primary antibodies for ZIP6 and ZIP10 (1:1000, generated in-house), PrP
(1:2000, A03213; Bertin Pharma, France), NCAM1 (1:6666, 556324; BD Biosciences, ON,
Canada) and PSA-NCAM1 (1:1000, 556324; BD Biosciences) were used. For transient
knockdowns, SilencerSelect siRNAs targeting Prnp (s72188; Life Technologies) and Ncam1
(s70398; Life Technologies), and ON-TARGETplus SMARTpools targeting Slc39a6 (LQ-
053529-01-0005; GE Healthcare, ON, Canada) and Slc39a10 (LQ-059712-01-0005; GE
Healthcare) were used.
2.2.2 Cell culture and transfection
Mouse mammary gland NMuMG cells (CRL-1636) were a kind gift from Dr. Jeffrey Wrana
(University of Toronto, ON, Canada), but are also available commercially through the American
Tissue Culture Collection (ATCC) (Manassas, VA, USA). Mouse neuroblastoma Neuro-2a
(N2a) (CCL-131) cells were purchased through the ATCC. Cells were cultured as recommended
by the ATCC, with 10% heat inactivated FBS (12484028; Life Technologies), 1% GlutaMAX
(35050061; Life Technologies), and 1% antibiotic-antimycotic solution (15240062; Life
Technologies) in (Dulbecco’s) Modified Eagles medium. Human insulin solution (I9278; Sigma-
40
Aldrich, ON, Canada) was also added at a concentration of 10 µg/mL for NMuMG cells. To
transfect cells with siRNAs, Lipofectamine RNAiMAX (13778075; Life Technologies) was used
according to the manufacturer’s instructions.
2.2.3 Generation of Slc39a6 CRISPR-knockout clones
The gRNA expression plasmid MLM3636 (Plasmid 43860) and SpCas9 plasmid JDS246
(Plasmid 43861) were obtained from Addgene’s non-profit plasmid repository (Cambridge, MA,
USA). The SpCas9nD10A nickase plasmid, provided by Louisa Wang, was generated by site-
directed mutagenesis of the original JDS246 plasmid using a Q5 Site-Directed Mutagenesis Kit
(E0554S; New England Biolabs, Ipswich, MA, USA) and forward and reverse primers
AGTGCCGATGGCTAAACCAATAG and AATTCCGTTGGATGGGCTG, respectively.
Compatible CRISPR nickase target sites within Slc39a6 Exon 2 were identified using the online
‘CRISPR Design Tool’ (http://crispr.mit.edu/). Respective oligonucleotide pairs were
subsequently customized and integrated into the MLM3636 plasmid as described in Chapter 1.
NMuMG cells were transfected using Lipofectamine LTX (15338100; Life Technologies) and a
6:1:1 (w/w/w) mixture of SpCas9nD10A and customized gRNA plasmid pairs. Transfections
were completed according to the manufacturer’s instructions at a cell confluency of
approximately 75%. Forty-eight hours post transfection, cells were harvested, diluted to a
concentration of 1 cell/mL of culture medium, and replated. After one week, individual colonies
were picked and cultured in separate wells of a 24-well plate before being subsequently
transferred to 6-well and then 60 mm plates, when cell numbers became sufficient for Western
blot analyses.
2.2.4 Western blot analyses
Cells were grown to near confluency, growth medium was removed, and cells were rinsed twice
with ice-cold phosphate buffered saline (PBS) before being lysed in a buffer consisting of 50
mM Tris (pH 8.0), 150 mM NaCl, 1% NP-40, and 1× Complete Protease Inhibitor Cocktail
(11836170001; Roche, ON, Canada). Cellular debris was cleared by centrifugation for 5 minutes
at 14 000 RPM and 4°C, and protein levels were subsequently adjusted using a bicinchoninic
acid (BCA) colourimetric assay. Equal amounts of protein were separated on 4-12% or 12% Bis-
Tris (Life Technologies) or 7% Tri-Glycine (cast in-house) gels before being transferred to a
0.45 micron polyvinylidene fluoride (PVDF) membrane. The blot membranes were blocked with
41
10% skim milk in Tris-buffered saline and Tween 20 (TBST), and incubated overnight at 4°C
with the respective primary antibody, diluted in 5% skim milk in TBST. The blots were washed
thrice with TBST and were incubated with HRP-conjugated anti-mouse (1:5000, 170–6516;
BioRad), anti-rabbit (1:5000, 170–6515; BioRad), or anti-rat (1:5000, 31476; Thermo Fisher
Scientific, Waltham, MA, USA) secondary antibodies diluted in 5% skim milk in TBST for 1-2
hrs at RT. Blots were again washed thrice with TBST before being briefly incubated with the
ECL reagent (RPN2106; GE Healthcare). Signals were then visualized using either X-ray film or
a LI-COR Odyssey Fc digital imaging system (LI-COR Biosciences, NE, USA).
2.2.5 RT-PCR analyses
Real-time polymerize chain reaction was completed identical to the procedure described in
Chapter 1, with the exception that the mRNAs targeted were specific to Slc39a5
(Mm00511105_m1), Slc39a6 (Mm00507297_m1), and Slc39a10 (Mm00554174_m1) (Life
Technologies).
2.2.6 Sample preparation for immunoprecipitation
For comparative interactome analyses, NMuMG cell clones were employed, including control
wild-type clones, wild-type clones undergoing EMT, ZIP6 knockout clones undergoing EMT,
and siRNA knockdowns of NCAM1 undergoing EMT. For each experiment, three biological
replicates of each condition were maintained in parallel. For cells undergoing EMT, TGFβ (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 5mM
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.
42
2.2.7 Protein immunoprecipitation
Immunoprecipitation with ZIP6 and NCAM1 specific antibodies were completed with protein A
agarose (P3476; Sigma-Aldrich) and protein G sepharose (17-0618-01; GE Healthcare) beads,
respectively. 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 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 5mM 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).
2.2.8 Protein reduction, alkylation, trypsinization, and labelling
Eluted proteins were dried to a volume of 5 µL, washed with ultrapure water, and dried once
again. Samples were then denatured by the addition of 9 M urea for 30 min at room temperature,
before having disulfide bonds reduced for 30 min at 60°C in the presence of 5 mM tris(2-
carboxyethyl) phosphine (TCEP), and being alkylated for 1 hr at RT in the presence of 10 mM 4-
vinylpyiridine (4-VP). Samples were then diluted with ultrapure water to ensure that the residual
urea concentration of each sample was below 1.5 M to allow for digestion with side chain-
modified porcine trypsin overnight at 37°C. The covalent modification of peptides with reagents
from the 6-plex amine-reactive tandem mass tag (TMT) (Thermo Fisher Scientific), or 8-plex
isobaric tags for relative and absolute quantitation (iTRAQ) (4390733; AB Sciex, Concord, ON,
Canada) labeling kits followed instructions provided by the manufacturers.
2.2.9 Nanoscale HPLC-ESI tandem mass spectrometry
Immunoprecipitates were purified by C18 reversed-phase and strong cation exchange cartridges
(Agilent Technologies, ON, Canada). Eluates were dried in a centrifugal concentrator then
suspended in aqueous 0.1% formic acid and applied to C18 nanocapillary columns (25 cm long
Acclaim PepMap RSLC with 100 Å pore size, 2 µm particle size, 75 µm inner diameter) using
an EASY-nLC 1100 system (Thermo Fisher Scientific). Peptide separation was performed at 300
43
nl/minute using a binary mobile phase gradient with mobile phases containing 0.1% (v/v) formic
acid in water or acetonitrile. Over the gradient, the organic content of the mobile phase was
increased from 0 to 30% over 180 minutes then to 100% by 240 minutes. The nano-HPLC
system was coupled by online nanoscale electrospray ionization (ESI) to an Orbitrap Fusion
Tribrid mass spectrometer. The MS data acquisition method included three scan types. Each data
acquisition cycle began with a survey scan of the 400–2000 m/z range in the orbitrap at 120,000
resolution with automatic gain control set to 2e5 counts. Next, the most intense precursor ions
carrying two to seven charges were separately isolated, subjected to collision-induced
dissociation (CID) and their fragments detected in the linear ion trap. Finally MS3 with higher-
energy collision-induced dissociation (HCD) was performed on the ten most intense fragments
from each MS2 scan. The orbitrap was used to obtain MS3 scans at 60,000 resolution with an
automatic gain control target of 1e5 counts at a maximum injection time of 120 ms. The
combined cycle time for this method was 3 seconds during which as many precursors as possible
were analyzed. Dynamic exclusion prevented re-analysis of any precursor ion within a 20 ppm
m/z window for 300 seconds.
2.2.10 Protein identification and quantification
Peptide sequencing was performed using Mascot (Version 2.4; Matrix Science Ltd, London, UK)
and Sequest HT search engines embedded in Proteome Discoverer software (Version 1.4;
Thermo Fisher Scientific). The international protein index (IPI) mouse database (Version 3.87)
was used as the protein sequence source in peptide-spectral matching. Only tryptic peptides with
two or fewer missed cleavages were considered for assignment. Isobaric tags at peptide amino-
termini and lysine residues were specified as fixed modifications. Variable modifications
specified were methionine oxidation, asparagine and glutamine deamidation, cysteine
pyridylethylation as well as phosphorylation at serine, threonine and tyrosine. For the Sequest
HT searches, the number of identical modifications and dynamic modifications were limited to
three and four per peptide respectively.
The reliability of peptide-spectrum matches was assessed using the q-value determined by the
Percolator algorithm at a false discovery rate setting of 0.05. Peptide quantification was
performed using MS3 data, in which reporter ion peaks were prevalent.
The mass spectrometry data will be able to be accessed from the PRIDE Archive once published.
44
2.3 Results
2.3.1 Generation of a mammalian ZIP6 knockout EMT model
To investigate if mouse NMuMG cells represent a suitable model for studying the involvement
of Zip6 on EMT, we initially induced this morphogenetic program in wild-type NMuMG cells
by the addition of TGFB1 and measured transcript levels of ZIP6 and its closest paralogs ZIP10
and Zip5 (Fig. 2.1a) [37]. Reminiscent of the previously reported transcript and protein profile of
PrP in the same cell model [1], ZIP6 transcript levels increased five-fold upon exposure of cells
to TGFB1 during a 72 hour time window. The corresponding transcript profiles for ZIP10 and
ZIP5 gave rise to a more complex profile, with ZIP10 transcript levels undergoing a dramatic
drop during the first few hours following EMT induction, but then continuously recovering, and
eventually exceeding original levels. ZIP5 transcript levels, in contrast, were relatively low
throughout EMT progression and did not exhibit a consistent trend-line. Having established that
ZIP6 expression is induced during EMT, consistent with its potential involvement in this cellular
program, we next generated ZIP6 knockout clones by a 2nd generation CRISPR/Cas9 knockout
strategy (Fig. 2.1b). More specifically, NMuMG cells were co-transfected with an expression
construct coding for point-mutated Cas9 D10A nickase and a pair of guide RNAs that directed
the nickase to off-set target sites on opposite DNA strands within Exon2 of the Slc39a6 gene.
Using this strategy, the successful generation of nearby nicks led to a double-strand genomic
break, which triggered the cell-autonomous non-homologous end-joining (NHEJ) program.
Following clonal isolation by the dilution method, a western blot-based screen identified several
heterozygote and homozygote ZIP6 knockout clones. In this and subsequent experiments, the
detection of ZIP6 relied on an in-house ZIP6-directed polyclonal antibody raised against a
twelve amino acid ZIP6-specific sequence within the PrP-like ectodomain of this transporter
(Fig. 2.1c). The subsequent western blot analysis of wild-type NMuMG cells and a ZIP6
knockout clone led to the expected increase in ZIP6 protein levels upon TGFB1 treatment in
wild-type cells and confirmed the complete absence of this transporter in the ZIP6 knockout
clone (Fig. 2.1d).
45
Figure 2.1. Mouse NMuMG cell model for investigating role of ZIP6 during EMT. (a) ZIP6
transcript levels are more than five-fold increased during epithelial-to-mesenchymal transition
induced in NMuMG mouse epithelial cells upon exposure to TGFB1. (b) CRISPR-Cas9-based
knockout of ZIP6 protein expression in NMuMG cells following Cas9-D10A nickase-mediated
single-strand cleavages of Slc39a6 Exon2 at non-overlapping target sites. (c) Schematic of ZIP6,
46
indicating epitope of polyclonal antibody used in this work. (d) Evidence of ZIP6 protein level
increase in NMuMG cells upon 48 hrs exposure to TGFB1. The arrow indicates the ZIP6
specific band, while asterisks denote unspecific bands.
2.3.2 ZIP6 forms a heteromeric complex with ZIP10 that interacts with NCAM1
We next employed wild-type NMuMG cells that had been treated with TGFB1 for 48 hours as
biological source material to generate a first ZIP6 interactome dataset, with ZIP6 knockout
NMuMG cells serving as negative controls (Fig. 2.2a). To avoid the inadvertent breaking of
ZIP6 protein-protein interactions during cell lysis, cells were in vivo crosslinked by mild
formaldehyde crosslinking prior to their harvest [32]. ZIP6 and proteins covalently crosslinked
or loosely associated with it were subsequently co-immunoprecipitated with the aforementioned
anti-ZIP6 antibody, denatured in urea and trypsinized. To minimize run-to-run variance and
facilitate the direct relative quantitation of peptide levels, the analysis made use of a previously
established work-flow that relies on the isobaric labeling of peptide mixtures with tandem mass
tags (TMT) followed by the concomitant analyses of three biological samples and controls in a
six-plex format [2]. Tandem mass spectrometry analyses of peptide mixtures were undertaken on
a Thermo Orbitrap Fusion instrument using a configuration for the sampling of TMT signature
ion spectra and downstream data analysis we had previously reported [2]. Confirming successful
and reproducible immunoprecipitation of ZIP6, the analysis unequivocally identified ZIP6 in the
co-IP eluates derived from wild-type but not ZIP6 knockout cells (Fig. 2.2b and c) (note that
well-known restrictions on the dynamic range of the quantitation method [99] limited in this
experimental design the ability to distinguish enrichment ratios that exceeded eight-fold).
Interestingly, this analysis identified ZIP10 as a key interactor of the ZIP6 bait (Fig. 2.3a).
NCAM1 (Fig. 2.3b) and calreticulin were the only two other proteins in the dataset whose
detection appeared to strictly depend on ZIP6. However, several other proteins were also
identified in ZIP6 co-IP eluates at levels that exceeded their quantities in negative controls
samples, yet did not share the striking enrichment observed for the aforementioned binders. The
latter distribution is characteristic for proteins that bind not only to the co-IP bait and its
interactors but also to the affinity matrix itself, thereby leading to lower TMT enrichment ratios.
Pre-existing annotations of candidate ZIP6 interactors, which fell into this category identified
most of them as belonging to relatively abundant cellular pathways or protein complexes,
47
including (i) the endoplasmic reticulum-resident protein folding and quality control machine
composed of calreticulin (CALR), calumenin (CALU), protein disulfide-isomerase A3 (PDIA3)
and 78 kDa glucose-regulated protein (HSPA5), (ii) enzymes associated with the glycolytic
pathway, (iii) subunits of microtubules, (iv) a subset of highly abundant heat shock / chaperone
complex proteins, and (v) abundant dehydrogenase complex subunits. Additional ZIP6
interactors identified in this experiment, which shared relatively low enrichment levels, but
cannot readily be classified as highly abundant proteins, were a small number of membrane-
embedded transporters (ATP2B4, SLC25A5 and SLC1A4), tetraspanin-6 (TSPAN6), galectin-1
(LGALS1) and glycogen synthase kinase 3 alpha (GSK3A) and beta (GSK3B). The ability to
confidently assign even small differences in relative abundance levels increases in this
methodology with the statistical power inherent to the number of peptides quantified and
matched to a given protein. Thus, although GSK3A and GSK3B levels in the ZIP6-specific co-IP
eluates exceeded the levels of these proteins in the negative control sample by only ~20% (Fig.
2.3c), this difference was significant and distinct from the distribution of ATP synthase coupling
factor 6 (ATP5J), a mitochondrial protein, which was observed at equal levels in samples and
negative controls, presumably, because it bound to the affinity matrix exclusively through non-
specific interactions (Fig. 2.3d).
48
49
Figure 2.2. The ZIP6 interactome. (a) Chart depicting workflow of ZIP6 interactome analysis.
(b) Relative quantitation of ZIP6 levels in ZIP6 co-immunoprecipitations (co-IPs) from in vivo
formaldehyde crosslinked wild-type and ZIP6 ko NMuMG cells. Box plot depicts in log2 space
enrichment ratios of individual ZIP6 peptides used for quantitation, as well as the median peptide
ratio and Inter Quartile Range (IQR). (c) Truncated list of proteins enriched in ZIP6-specific co-
IPs. Colors identify proteins with related function and/or localization.
50
Figure 2.3. Interaction of a ZIP6-ZIP10 heteromer complex with NCAM1 and GSK3. (a)
Relative quantitation of ZIP10 levels in ZIP6 co-IPs from in vivo formaldehyde crosslinked
wild-type and ZIP6 ko NMuMG cells. Box plot depicts in log2 space enrichment ratios of
individual ZIP10 peptides used for quantitation, as well as the median peptide ratio and Inter
51
Quartile Range (IQR). See Fig. 2.2b for full legend. (b-d) Relative quantitation of neural cell
adhesion molecule 1 (NCAM1), glycogen synthase kinase 3 alpha (GSK3A) and ATP synthase
coupling factor 6 (ASCF6). Note the direct correlation between the ability to confidently detect
small differences in protein levels and the number of peptides quantified. (e) ZIP6 and ZIP10
regulate each other’s expression and maturation. Transient knockdown of ZIP6 or ZIP10, or both
zinc transporters in mouse neuroblastoma cells causes profound reduction of ZIP6 and ZIP10
protein levels. In contrast, knockdown of PrP affects maturation of ZIP6 but exerts no effect on
protein levels of ZIP6 or ZIP10. Note that the arrows indicate ZIP6 or ZIP10 specific bands,
while the asterisks denote unspecific bands. (f) Addition of copper or zinc to the cell culture
medium mimics siRNA-based knockdown with regard to its effect on ZIP6 protein levels. In
contrast, chelation of divalent cations by exposure of cells to TPEN causes an increase in steady-
state ZIP6 protein levels.
Several recent studies have revealed similarities and functional relationships between ZIP6 and
ZIP10 [100,101,102], suggesting that these two zinc transporters may interact. To further clarify
their relationship, we investigated if reciprocal effects on expression and maturation of these
proteins can be observed if the expression of one of them is transiently reduced (Fig. 2.3e). This
experiment was undertaken in the mouse neuroblastoma Neuro2a cell model, which expresses
both transporters at higher steady-state levels than NMuMG cells. Protein levels of ZIP6, ZIP10
and PrP were quantified by densitometric quantitation of western blot signals. The experiment
revealed a robust and significant reciprocal effect of ZIP6 and ZIP10 on each other’s expression,
evidenced by a more than 50% reduction in protein levels and higher mobility—possibly
reflecting incomplete maturation—of ZIP6 when ZIP10 expression was diminished by ZIP10-
specific siRNAs. In contrast, siRNA-mediated depletion of PrP did not significantly affect total
ZIP6 levels but also affected ZIP6 band migration, albeit in a less complete manner than ZIP10
depletion. To further explore the functional relationship of ZIP6 and ZIP10, we capitalized on
our earlier observation that exposure of cells to elevated Cu2+ concentrations, rather than Zn2+, in
the culture medium caused a pronounced reduction in ZIP10 protein levels [103]. If the
expression and maturation of ZIP6 and ZIP10 proteins is co-regulated, we would expect a similar
inhibition profile also for ZIP6. Indeed, the presence of Cu2+ caused the same profound
diminution of ZIP6 levels that we had previously observed for ZIP10 under identical
experimental conditions (Fig. 2.3f). Taken together, these data are consistent with the
interpretation that ZIP6 and ZIP10 form a functional heteromeric complex that interacts with
NCAM1 and whose expression is tightly co-regulated, possibly involving critical interactions
with calreticulin and other members of the ER-resident protein folding machinery.
52
2.3.3 NCAM1 serves as a hub for the assembly of focal adhesion complexes
To better understand the relationship between ZIP6 and NCAM1, we next undertook three
separate NCAM1 interactome analyses using essentially the same experimental design as
outlined above for the ZIP6 interactome analysis (Fig. 2.2a). The three NCAM1 datasets
investigated (i) the NCAM1 interactome in TGFB1-treated NMuMG cells with and without prior
knockdown of NCAM1, thereby revealing binders of NCAM1 in this differentiation state
(Dataset I), (ii) the influence of ZIP6 knockout on the NCAM1 interactome in TGFB1-treated
NMuMG cells (Dataset II), and (iii) changes the NCAM1 interactome undergoes in cells shifting
from an epithelial to a mesenchymal phenotype (Dataset III) (Fig. 2.4a). To facilitate the direct
comparison of samples contributing to analyses I and II, we made use of an 8-plex labeling
scheme based on isobaric tags for relative and absolute quantitations (iTRAQ) [104] that
afforded the ability to concomitantly obtain quantitation values for individual peptides from
iTRAQ signature ions of three wild-type, three ZIP6 knockout and two NCAM1 knockdown
samples (Fig. 2.4b). To gauge the reproducibility of biological replicates in this experiment, we
computed the correlation coefficient of protein ratios by pair-wise comparisons of samples that
gave rise to Datasets I and II (Fig. 2.4c). Indicative of high technical and biological
reproducibility these correlation coefficients were consistently >0.900 for pair-wise comparisons
of biological replicates. However, correlation coefficients dropped to values between 0.635 and
0.816 in pair-wise comparisons of protein ratios observed in NCAM1 co-IP samples observed
from Dataset I and II, suggesting that with regard to its influence on the NCAM1 interactome the
ZIP6 knockout does not simply mimic an overall reduction of NCAM1 and its interactions (seen
in Dataset I) but affects the NCAM1 interactome in a more complex way. Consistent with prior
knowledge of NCAM1 biology, the subsequent gene ontology (GO) ‘Cellular Component’
analysis of Dataset I indicated that the NCAM1 interactome is highly significantly enriched in
proteins that map to ‘focal adhesion’ complexes, ‘cell projections’ and the ‘plasma membrane’
(Fig. 2.4d). Perhaps less expected, the respective GO ‘Biological processes’ analysis of Dataset I
indicated also a significant enrichment of proteins with known roles in ‘glycolytic processes’ and
‘protein folding’. A meta-analysis of ZIP6 and NCAM1 interactomes generated in this work,
which made use of Cytoscape software to analyze and highlight shared features, revealed
considerable overlap but also differences in the composition of these two interactomes (Fig.
2.4e). More specifically, whereas the interactome of NCAM1 is consistent with its role as a hub
53
protein, which organizes proteins that are embedded in the plasma membrane or exist in its
proximity, the ZIP6 interactome comprised only relative few interactors not shared by NCAM1.
Notable exceptions to this trend were the exclusive interaction of GSK3A and B with ZIP6, as
well as a more pronounced representation in the ZIP6 interactome of the aforementioned ER-
resident proteins with known roles in protein folding and the quality control machinery.
54
Figure 2.4. The NCAM1 interactome. (a) Design of three multi-plex NCAM1 comparative
interactome analyses conducted in this study. Dataset I: identification of NCAM1 interactors in
55
TGFB1-treated NMuMG cells; Dataset II: effect of ZIP6 ko on NCAM1 interactome; and
Dataset III: effect of epithelial-to-mesenchymal transition on NCAM1 interactome. (b) The
experimental design followed the workflow outlined in Fig. 2a. However, eight-plex iTRAQ
labeling was employed to be able to generate Datasets I and II in one combined analysis. (c)
Comparison of interactomes generated in Datasets I and II by Pearson correlation analysis. Note
that biological replicates within Datasets I and II generated a Pearson coefficient near 1,
indicating excellent biological reproducibility of interactomes. A direct of comparison of
Datasets I and II also returned Pearson coefficients >0.5, indicating a direct correlation, which
suggests that the presence of ZIP6 promotes NCAM1 binding to a subset of its natural
interactors. (d) A Gene Ontology analysis of Dataset I indicated a highly significant enrichment
of ‘Cellular component’ categories ‘focal adhesion’, ‘cell projection’ and ‘plasma membranes’.
A similar query of ‘Biological processes’ categories revealed that proteins known to contribute
to ‘glycolytic process’, ‘protein folding’ and ‘biological adhesion’ were enriched in this dataset.
(e) Schematic depiction of ZIP6 and NCAM1 interactomes. The figure was originally generated
in Cytoscape but nodes were subsequently rearranged to indicate the known predominant cellular
compartments in which interactors reside. Grey shading was used to indicate relative enrichment
levels of interactors. Rectangular and oval node shapes identify NCAM1 and ZIP6 interactors,
respectively. Rectangular shapes with rounded edges were used to indicate the presence of a
given interactor in both NCAM1 and ZIP6 interactome datasets. Nodes with green boundaries
identify NCAM1 interactors whose presence in the NCAM1 interactome was most affected by
the presence or absence of ZIP6.
2.3.4 ZIP6 influences the association of NCAM1 with specific interactors and phosphorylation of NCAM1 at a GSK3 consensus site
A close comparison of enrichment ratios of individual proteins in NCAM1 interactome Datasets
I and II led to the following observations regarding the effect of ZIP6 knockout on NCAM1 and
its interactors (Fig. 2.5): (i) In wild-type NMuMG cells, NCAM1 levels observed were
approximately 50% higher than in ZIP6 knockout cells derived from them. (ii) The ZIP6
knockout affected primarily NCAM1 interactions with its key interactors (note the similar
overall trend in shading intensity for protein ratios observed in Datasets I and II). More
specifically, the ZIP6 knockout had a particularly pronounced effect on NCAM1 interactions
with enzymes that play a broader role in glycolysis and the cellular energy household, as well as
peptidyl prolyl cis-trans isomerase (PPIA), profilin (PFN1), heat shock 10 kDa protein 1
(HSPE1), protein S100-A6 (S100A6), and a subset of integrins. In contrast, interactions of
NCAM1 with ezrin (EZR), polyubiquitin (UBC), annexin A2 (ANXA2), and neuropilin-1
(NRP1) were to a lesser degree affected by ZIP6 knockout than what would expected if the
deficiency of this ZIP transporter affected NCAM1 interactions observed in wild-type NMuMG
cells equally.
56
Figure 2.5. Effect of ZIP6 on the NCAM1 interactome. Subset of interactome data from
Datasets I and II. NCAM1 interactors shown are sorted to reflect the influence of ZIP6 on the
interactome, i.e., interactors, whose co-enrichment with NCAM1 was most affected in a direct
comparison of NCAM1 interactome data generated from wild-type and ZIP6 ko are listed before
NCAM1 interactors influenced to a lesser degree by the presence or absence of ZIP6.
57
Because Dataset II led to 1,785 peptide-to-spectrum matches, which were assigned to the longest
isoform of NCAM1, cumulatively covering 68% of its sequence, it afforded the identification
and relative quantitation of several NCAM1 post-translational modifications. One of them was a
phosphorylation on the NCAM1 peptide (residues 945-974) with the amino acid sequence
‘ASPAPTPTPAGAASPLAAVAAPATDAPQAK’, which was repeatedly assigned to more than
a dozen independent tandem mass spectra both in its phosphorylated and unphosporylated state
(Fig. 2.6a-c). Intriguingly, in contrast to other NCAM1 phosphorylation sites observed in the
dataset, including a peptide which maps to a nearby stretch of amino acids with the sequence
NPPEAATAPASPK (residues 995-1007), phosphorylation of the 945-974 peptide within
NCAM1 was dependent on the presence of ZIP6 (Fig. 2.6d-f). The peptide harbors several
phospho-acceptor sites, of which as many as four might have been phosphorylated (there was
some ambiguity due to gaps in the y- and b-ion series of backbone fragments). Closer inspection
of the respective sites characterizes them as being four, two and six residues apart and embedded
within an amino acid sequence stretch rich in prolines and alanines that also harbors two lysine
residues (Fig. 2.6g). Clusters of nearby phosphorylation sites are also known in other proteins.
Amongst the known kinases, GSK3A and B are notorious for their ability to transfer phosphates
to sites of this nature, in particular, once a priming phosphate has been attached to the most C-
terminal acceptor site within such a cluster. We therefore compared the NCAM1
phosphorylation cluster both to the GSK3 substrate consensus sequence S/T-x(3-4)-S(P)/T(P) [105]
(Fig. 2.6h) and to validated GSK3 substrates. Surprisingly, a known GSK3 substrate sequence
within the family of collapsin response mediator proteins [106] showed striking similarity in
both its overall amino acid composition and sequence to the NCAM1 cluster identified in this
work, sharing with NCAM1 the core sequence A(S/T)PAP(S/T (Fig. 2.6i). Taken together, these
data are consistent with a model whereby GSK3 kinases contribute to NCAM1 phosphorylation
within this cluster, and the aforementioned co-enrichment of GSK3A and B with ZIP6 accounts
for the ZIP6-dependence of the observed phospho-occupancy at this site.
58
Figure 2.6. ZIP6 controls phosphorylation of the longest isoform of NCAM1 at a phospho-
acceptor site that conforms to a previously described GSK3 recognition site within
members of the Crmp protein family. (a) Extracted ion chromatograms of a specific NCAM1
59
peptide and its phosphorylated derivative that were repeatedly identified and quantified in
Datasets I and II. The relative peak intensities suggest that approximately 15% of the peptide was
phosphorylated. Panels (b) and (c) depict tandem mass spectra underlying the identification of
the non-phosphorylated and phosphorylated NCAM1 peptide, respectively. Note the
characteristic 40.5 mass shift observed in a subset of doubly-charged fragment ions of the b-
series, which identify the phosphorylation. (d) Phosphorylation of NCAM1 at the specific site
depends on the presence or absence of ZIP6. iTRAQ signature ion trace of tandem mass
spectrum depicted in panel (c). (e) Plot depicting the enrichment ratios of repeated analyses of
the phosphopeptide depicted in panel (c), documenting four-fold higher phospho-occupancy
levels for this peptide in NCAM1 co-IPs from wild-type cells than in ZIP6 ko cells. (f) Relative
levels of an unrelated NCAM1 phosphopeptide that may serve as an internal negative control,
because its phospho-occupancy levels are to a lesser degree dependent on ZIP6 than those of the
phosphopeptide quantified in panel e. Note that the two-fold higher levels of the phosphopeptide
in wild-type over ZIP6 ko cells only marginally exceed the relative ratios of non-phosphorylated
NCAM1-derived peptides, which were also observed at approximately 1.5-fold higher levels in
wild-type than in ZIP6 ko cells (see Fig. 2.5). (g) Multiple alignment of NCAM1 sequence
stretch encompassing the ZIP6-dependent phosphorylation site. The sequence stretch comprises
a cluster of candidate phospho-acceptor sites that conform to a GSK3 substrate consensus motif,
depicted in (h), which maps to a cytoplasmic insertion coded by Exon 18 present only in the
longest NCAM1 isoform. (i) The ZIP6 dependent NCAM1 phosphorylation site bears striking
resemblance to a previously reported GSK3-phosphorylation site present in a subset of Crmp
proteins.
2.3.5 NCAM1 associates with integrins and actin-assembly complexes and is polysialylated in a PrP and ZIP6-dependent manner during EMT
We next investigated changes to NCAM1 interactions during EMT by generating the
aforementioned NCAM1 interactome Dataset III (Fig. 2.4a) based on wild-type NMuMG cells
before and after 48 hour exposure to TGFB1. The analysis was undertaken at a smaller scale than
the NCAM1 interactome analyses underlying Datasets I and II and made use of six-plex TMT-
tags for relative quantitation of three TGFB1-treated biological replicates and three vehicle-
treated wild-type NMuMG negative control samples. The workflow for sample preparations, as
well as mass spectrometry analyses and data processing followed the steps outlined before for
generating the ZIP6 and NCAM1 interactome analyses. As expected for bait proteins in
successful co-immunoprecipitation experiments, NCAM1 was again identified on the basis of the
highest number of peptide-to-spectrum matches, and its levels were observed to increase
approximately two-fold during EMT (Fig. 2.7a). Interestingly, as NMuMG cells underwent
EMT, interactions of NCAM1 with tubulin beta-2A (TUBB2A) and alpha-1C (TUBA1C)—
presumably reflecting its interactions with the microtubule-based cytoskeleton—were reduced
60
and replaced by interactions with the actin cytoskeleton. The latter was evident by consistently
higher levels of actin (ACTB) itself, but also of moesin (MSN), integrins beta-1 (ITGB1), alpha
V (ITGAV) and alpha-2 (ITGA2) and cofilin (CFL1) in the NCAM1 interactome from wild-type
NMuMG cells of mesenchymal morphology, when compared to NMuMG cells of epithelial
morphology.
61
62
Figure 2.7. During EMT the shift of NCAM1-cytoskeletal interactions from a tubulin to an
actin-dominated environment is accompanied by its polysialylation. (a) Relative quantitation
of NCAM1 interactions before and after exposure of NMuMG cells to TGFB1. Changes in the
relative levels of proteins which co-immunoprecipitated with NCAM1 are depicted as ratios. (b)
Time-course study depicting increases in total NCAM1 levels and polysialylated NCAM1 during
EMT. (c) Western blot depicting ablation of NCAM1 polysialylation in PrP-deficient cells that
have underwent EMT. (d) Schematic depicting relative levels of NCAM1 and polysialylated
NCAM1 during EMT in wild-type, ZIP6-deficient, and PrP-deficient NMuMG cells.
To assess if ZIP6 deficiency exerts a similar effect on NCAM1 polysialylation as PrP deficiency
[2], we next harvested wild-type and ZIP6 knockout NMuMG cells at distinct time intervals
following the addition of TGFB1 to the cell culture medium and analyzed cell lysates by western
blot analyses with antibodies directed against total NCAM1 or its polysialic acid (polySia)
modification (Fig. 2.7b). This analysis revealed that ZIP6 deficiency reduces the overall amount
of NCAM1 in NMuMG cells but, unlike PrP deficiency (Fig. 2.7c), does not prevent NCAM1
polysialylation. In contrast, despite reduced overall NCAM1 protein levels in Zip6 knockout
cells, the intensity of PSA bands was increased in the absence of ZIP6. Taken together, these
data suggest that the presence of ZIP6 exerts, like the presence of PrP [2], a positive effect on
total NCAM1 levels but, whereas PrP deficiency abolishes NCAM1 polysialylation in this
model, Zip6 deficiency promotes it (Fig. 2.7d).
2.4 Discussion
This study set out to undertake a hypothesis-free investigation of the ZIP6 interactome in a
widely used mammalian cell model for studying EMT. It uncovered that ZIP6 forms a
heteromeric complex with ZIP10 that predominantly interacts with NCAM1. This discovery
triggered a 2nd set of interactome analyses, which first produced an in-depth analysis of the
molecular environment of NCAM1 in the same cell model, and next explored how the knockout
of ZIP6 affects NCAM1 interactions. The study established prominent interactions of NCAM1
with integrins and focal adhesion complexes but also revealed that ZIP6 critically influences
their assembly. Following up on the observation of a ZIP6-dependent phosphorylation of a
specific NCAM1 peptide, we identified a cluster of phospho-acceptor sites in the cytoplasmic
domain of the longest isoform of NCAM1, which conforms in its composition and core sequence
to a known GSK3 substrate cluster within the family of collapsin-response mediator proteins.
63
Finally, we documented that ZIP6—like its PrP cousin—promotes an increase in steady-state
NCAM1 levels during EMT but—unlike PrP—serves to limit NCAM1 polysialylation that
occurs during the execution of this morphogenetic program.
2.4.1 ZIP6 forms a heteromeric complex with ZIP10 that interacts with NCAM1
The ZIP6 interactome dataset generated in this work may represent the first investigation of
molecular interactions of any ZIP metal ion transporter that employed a mass spectrometry-based
discovery workflow. It has been known for some time that ZIP6 and ZIP10 co-purify under
certain conditions [32], are co-regulated in some paradigms [100,101,107], and share functional
similarities [101]. The functional relationship between these ZIP transporters was particularly
evident in the latter study, which documented that during meiotic maturation transcript levels of
ZIP6 and ZIP10 were correlated and exceeded transcript levels of other zinc transporter six- to
tenfold. The authors went on to show that the targeted disruption of maternally-derived ZIP6 or
ZIP10 during meiotic maturation perturbed intracellular zinc levels and resulted in cell cycle
arrest. The first evidence that ZIP6 and ZIP10 may, in fact, bind directly to each other emerged
recently from a functional analysis of ZIP10 in the zebrafish gastrulation paradigm [102]. Not
only mimicked ZIP10 knockdown in this recent work a previously reported gastrulation arrest
phenotype observed with ZIP6-deficient zebrafish embryos [39] but the two proteins were seen
to form an intimate interaction on the basis of proximity ligation assay data and co-
immunoprecipitation analyses. The current study corroborated these prior observations and
added to this body of research by showing that in the cell models we used (i) ZIP6 interacts with
ZIP10 but not with other ZIP transporters, (ii) ZIP6 and ZIP10 are not only co-regulated but
influence each other’s expression reciprocally, a property often observed amongst subunits of
functional protein complexes, and (iii) ZIP6 and ZIP10 share a rapid suppression of their steady-
state protein levels in cells exposed to increased copper. Currently unanswered remain questions
regarding the precise stoichiometry of the ZIP6/ZIP10 heteromer and the spatiotemporal
dynamics of its assembly. Rather than exhibiting perfect co-localization, our preliminary data
suggest that the cellular distribution of ZIP6 and ZIP10 is only partially overlapping in some
cells, consistent with a model that sees them cooperate only in specific cellular contexts and
paradigms.
64
2.4.2 ZIP6 is critical for proper execution of EMT, presumably by facilitating assembly of NCAM1 focal adhesion complexes
A central finding of this study was the highly reproducible and selective enrichment of NCAM1
in ZIP6 co-IPs. To our knowledge, NCAM1 had not been described to associate with any ZIP
metal transporter before. Yet, this discovery was not surprising for anyone aware of both (i) the
evolutionary relationship of ZIP zinc transporters and PrP, and (ii) the fact that NCAM1 has
repeatedly been shown to not only interact with PrP [33,34,35,108,109] but represented the only
prominent in vivo crosslinking partner of PrP in the neuroblastoma cell model [34]. The latter
study not only provided evidence for a direct interaction of PrP and NCAM1 but also identified
on the basis of in vitro binding data a tentative binding interface that encompassed Helix A and
the adjacent loop domain within PrP and β-strands C and C’ in the two fibronectin type III
(FNIII) domains within NCAM1. In light of the predicted structural similarity between PrP and
ectodomains of ZIP5, 6 and 10 [37,110] one may speculate that the interaction between the
ZIP6/ZIP10 heteromer involves a homologous binding interface but this question remains
unaddressed at this time. Consistent with the view that ZIP6 is not merely a bystander in
NCAM1-related cell biology, ZIP6 knockout cells did not only express lower levels of NCAM1
but were observed to exhibit profound and reproducible impairments in NCAM1 interactions
during EMT. In particular, the ability of NCAM1 to interact with glycolytic enzymes,
components of the actin cytoskeleton, integrins and 14-3-3 proteins was impacted in ZIP6
knockout cells. It will be of interest to resolve if the ZIP6-ZIP10 heteromer affects these
NCAM1 interactions in a manner that is dependent on the divalent cation import capabilities of
these ZIP transporters or operates independent of them.
2.4.3 Binding of GSK3 to ZIP6 and phosphorylation of NCAM1 at GSK3 consensus sites
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 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
65
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 [111]. 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. Putative sites for GSK3 binding
to the ZIP6-ZIP10 heteromer are found in a cytoplasmic loop connecting transmembrane
domains 3 and 4 (amino acids 453-667 and 518-688 in ZIP6 and ZIP10, respectively). Similar to
known docking sites of GSK3 on axin 1 (AXIN1) and 2 (AXIN2) [112,113], these cytoplasmic
loop segments within the two ZIP transporters are enriched in charged residues and C-terminally
flanked by a histidine-rich segment. It has long been known that NCAM1 can be phosphorylated
within its intracellular domain by casein kinase 1 (CK1) and GSK3 [114]. Subsequent attempts
to map the underlying NCAM1 phospho-acceptor sites highlighted initially a domain in
juxtaposition to the inner face of the plasma membrane but also mentioned a putative serine 761
phospho-acceptor site that conforms to the GSK3 consensus motif [115]. 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 [116]. These sites are identical to the first and fourth NCAM1
phosphorylation sites within phosphorylation cluster (amino acids 945-974) that was repeatedly
sequenced and quantified 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 [117].
66
2.5 Conclusions
The discovery of the evolutionary relationship between ZIP transporters and PrP a few years ago
[37] raised the possibility that the study of ZIP transporters may provide a promising angle for
uncovering the elusive function of PrP. Indeed, similarities in gastrulation defects in ZIP6- or
PrP-deficient zebrafish embryos precipitated a productive line of research, which revealed that
PrP modulates the expression of NCAM1 and controls its polysialylation during EMT [2].
Observations uncovered in this study remind us that relationships can work in more than one
direction. Prior knowledge of NCAM1 as the key interactor of PrP in several paradigms, placed
the discovery of NCAM1 as the most prominent ZIP6 interactor in a special light. Foremost, it
suggests that PrP inherited its NCAM1 binding properties from its ZIP ancestor. This conclusion
is consistent with bioinformatic evidence, which indicates that NCAM and ZIP transporters
carrying a PrP-like ectodomain go back in evolution approximately one billion years but the
emergence of the prion gene family and polysialyltransferases required for NCAM1
polysialylation represent more recent evolutionary inventions that occurred early during
vertebrate speciation (Fig. 2.8). Our data reveal the existence of an intricate relationship that sees
ZIP6 control the expression, post-translational modification and interactions of NCAM1 in the
context of larger morphogenetic rearrangements executed at various stages during development
and in cancers. It is tempting to speculate that the prion gene owes its survival to an adaptation
that allowed it to modulate the same overall NCAM1-centric biology but provided a fitness
advantage in form of its control of NCAM1 polysialylation.
67
68
Figure 2.8. Evolution of NCAM, ZIP, PrP, and polyST gene families. Schematic depicting
the number of paralogs of NCAM, ZIP, PrP, and polyST gene families for a range of species.
Figure provided by Gerold Schmitt-Ulms and Mohadeseh Mehrabian [118].
69
Chapter 3 Future directions involving the investigation of the PrP-
ZIP6/ZIP10-NCAM1 connection
Summary: This chapter outlines research angles that might represent a natural progression of
work described in this thesis and would expand on the PrP-ZIP6/ZIP10-NCAM1 connection
described in the previous chapters.
The previous two chapters have described an intricate relationship between PrP, ZIP6, ZIP10 and
NCAM1 in the context of epithelial-to-mesenchymal transition. We determined that PrP and a
ZIP6/ZIP10 heteromer both have previously unrecognized functions during the morphogenetic
rearrangements that underlie EMT, acting on this program by modulating the molecular biology
of both NCAM1 itself and its environment. Specifically, PrP was observed to control the
polysialylation of NCAM1 (Chapter 1), while the ZIP6/ZIP10 heteromer controls the
phosphorylation of NCAM1 and its inclusion into focal adhesion complexes (Chapter 2). It will
be of interest to investigate if these proteins, and other proteins containing PrP-like domains,
fulfill this and related roles also in cellular models not tested in this study. Furthermore, it will be
instructive to learn to which extent their influence on the cellular biology underlying EMT has
been adapted for related morphogenetic programs (e.g., neuritic extension and cell migration) in
order to more fully and universally characterize the scope of their influence.
One intriguing avenue of investigation would be to determine if the mammalian paralogs of PrP,
Doppel (Dpl) [119] and Shadoo (Sho) [120], exhibit similar, or opposing, functions with regards
to EMT and NCAM1 polysialylation. The Dpl protein is predominantly expressed in the male
reproductive tract and resembles the carboxy-terminal globular domain of PrPc [119,121], while
Sho is predominantly expressed in the brain and encompasses most conspicuously the central
hydrophobic disordered domain within PrPc [120]. As may be expected from its location in the
male reproductive tract, Dpl seems to play a role in spermatogenesis [121,122]. Similar to EMT,
spermatogenesis is a form of cellular morphogenesis. Determining if this process involves
modification to cellular adhesion molecules, such as the polysialylation of NCAM1, and whether
70
this process is dependent on Dpl would broaden the scope and importance of the PrP family of
proteins. To date there is some evidence that NCAM polysialylation coincides with mammalian
spermatogenesis [144,145], though the molecular mechanisms behind this process have yet to be
investigated. Furthermore, ectopically expressed Dpl in the brain leads to neurotoxic effects that
can be recovered with expression of PrPc or Sho [108]. Though the function of Sho is not well
understood, these observations combined with its expression in the brain suggest redundant and
protective roles of PrPc [108]. Investigating the precise roles of Dpl and Sho in the context of
cellular morphogenesis and polysialylation relative to PrP could shed light on the functional
significance of shared domains and interactions amongst these proteins.
Similarly, it would be useful to further investigate ZIP5, the closest paralog of ZIP6 and ZIP10,
in the context of EMT. Of the three ZIP transporters with PrP-like ectodomains, ZIP5 is the only
one not identified in the neuroblastoma cell model as an interactor of PrP through co-
immunoprecipitation [32]. However, ZIP5 has similar sequence identity to PrP as its molecular
cousins [110]. Of potential relevance, ZIP5 also has been independently linked to the bone
morphogenetic protein (BMP)/TGFB1 pathway [123], making it all the more intriguing to study
in this context. We did not observe the same increase in ZIP5 transcript levels during TGFB1
treatment of NMuMG cells as we did with ZIP6 and ZIP10 (Fig. 2.1a) and, in fact, expression
levels of this protein seem low in this model. It therefore might be necessary and instructive to
study ZIP5’s role on EMT following its siRNA-based knockdown or CRISPR-Cas9 knockout in
an experimental paradigm known to express this protein, including bone cells, enterocytes and
acinar cells [123,146]. Based on the well-established codependency and role of ZIP6 and ZIP10
in this model, characterizing ZIP5 in this manner would clarify for this subbranch of ZIP
transporters the scope of their involvement and level of specialization in cellular programs
described in this thesis.
Another intriguing and potentially rewarding avenue for further investigation would utilize the
PrP knockout models generated in C2C12 myocyte and N2a neuroblastoma cells described in
Chapter 1. To date, our lab has performed global proteomics on these knockout clones [124] and
investigated how PrP-deficiency in these cell models affects NCAM1 levels and its
polysialylation [2], furthering our understanding of the physiological function of PrP. Analogous
experiments in C2C12 and N2a cell models might not only reveal the degree to which
observations made in the NMuMG model are transferable to other models but would also open
71
up the possibility to study how signaling downstream of PrP is affected by prion infection. Both
cell models have been shown to be useful models for studying infection with the misfolded,
disease causing form of PrP, scrapie (PrPSc) [15,16,17].
Investigations in the C2C12 and N2a paradigms could be extended to explore more fully the
roles of ZIP transporters in the disease. Preliminary studies could be conducted with siRNA and
shRNA knockdown experiments to determine if there are any obvious effects caused by ZIP
deficiencies, or could be pursued directly with CRISPR-Cas9 based ZIP knockout clones
deficient for one or more transporter. This would provide a wider scope for determining the
physiological functions of the ZIP transporters, and allow examination into whether they have an
effect on prion disease propagation. The C2C12 model would be of particular interest not only
due to its ability to undergo cellular differentiation to myotubes [17], but also due to its
transformative response to treatment with members of the BMP/TGFB1 family [125].
Our data are consistent with prior data, which linked ZIP transporters and PrP to cancer biology.
Specifically, ZIP6 and ZIP10 and PrP have independently been found to be upregulated in a
subset of cancers and seem to play a role in metastasis [98,147-152]. Breast cancer seems to be
one of the most studied cancers with regard to ZIP transporters and PrP, which emphasizes the
importance of the findings from the NMuMG model as it is derived from mammary glands.
Understanding the roles that ZIPs and PrP have in healthy cells undergoing EMT may shed light
on the mechanisms involved with cancer metastasis. Further investigations into how metastasis
mechanisms in cancerous cells relate to those utilized during EMT could explain the role of ZIPs
and PrP in cancer and lead to novel therapeutic targets to interfere with these mechanisms.
In conclusion, our data are consistent with a model in which PrP has inherited and retained its
ability to bind NCAM1 from ZIP proteins containing a PrP-like domain, but developed a unique
subspecialization in its control of NCAM1 polysialylation. The model strongly points towards
physiological functions of PrP and ZIP transporters in epithelial-to-mesenchymal transition and
related morphogenetic programs. It is hoped that future work will more fully uncover the
molecular mechanisms by which PrP and ZIP transporters contribute cellular programs centering
on NCAM1.
72
References
1. Mehrabian M, Brethour D, MacIsaac S, Kim JK, Gunawardana CG, Wang, H and
Schmitt-Ulms, G. (2014). CRISPR-Cas9-Based Knockout of the Prion Protein and Its
Effect on the Proteome. PLoS ONE 9(12): e114594. doi: 10.1371/journal.pone.0114594
2. Mehrabian M, Brethour D, Wang H, Xi Z, Rogaeva E, Schmitt-Ulms G (2015). The
Prion Protein Controls Polysialylation of Neural Cell Adhesion Molecule 1 during
Cellular Morphogenesis. PLoS ONE 10(8): e0133741. doi:
10.1371/journal.pone.0133741
3. Prusiner SB. (1982). Novel proteinaceous infectious particles cause scrapie. Science
216: 136–144. pmid:6801762 doi: 10.1126/science.6801762
4. Aguzzi A, Baumann F, Bremer J (2008). The Prion's Elusive Reason for Being. Annu
Rev Neurosci 31:439–477. doi: 10.1146/annurev.neuro.31.060407.125620
5. Sakudo A, Ikuta K. (2009). Prion protein functions and dysfunction in prion diseases.
Curr Med Chem 16: 380–389. doi: 10.2174/092986709787002673
6. Nicolas O, Gavin R, del Rio JA. (2009). New insights into cellular prion protein (PrPc)
functions: the "ying and yang" of a relevant protein. Brain Res Rev 61: 170–184. doi:
10.1016/j.brainresrev.2009.06.002
7. Hu W, Kieseier B, Frohman E, Eagar TN, Rosenberg RN, Hartung HP, et al. (2008).
Prion proteins: physiological functions and role in neurological disorders. J Neurol Sci
264: 1–8. doi: 10.1016/j.jns.2007.06.019
8. Watts JC, Westaway D. (2007). The prion protein family: diversity, rivalry, and
dysfunction. Biochim Biophys Acta 1772: 654–672. doi: 10.1016/j.bbadis.2007.05.001
9. Lauren J, Gimbel DA, Nygaard HB, Gilbert JW, Strittmatter SM (2009). Cellular prion
protein mediates impairment of synaptic plasticity by amyloid-beta oligomers. Nature
457:1128–1132. doi: 10.1038/nature07761
10. White JK, Gerdin AK, Karp NA, Ryder E, Buljan M, Bussell JN, et al. (2013).
Genome-wide generation and systematic phenotyping of knockout mice reveals new
roles for many genes. Cell 154: 452–464. doi: 10.1016/j.cell.2013.06.022
11. Kocher T, Superti-Furga G. (2007). Mass spectrometry-based functional proteomics:
from molecular machines to protein networks. Nat Meth 4: 807–815. doi:
10.1038/nmeth1093
12. Mendelsohn AR, Brent R. (1999). Protein interaction methods—toward an endgame.
Science 284: 1948–1950. doi: 10.1126/science.284.5422.1948
13. Livstone MS, Oughtred R, Heinicke S, Vernot B, Huttenhower C, Durand D, et al.
(2011). Inferring protein function from homology using the Princeton Protein Orthology
Database (P-POD). Curr Prot Bioinform Chapter 6: Unit 6 11. doi:
10.1002/0471250953.bi0611s33
14. Flechsig E, Weissmann C (2004). The role of PrP in health and disease. Curr Mol Med
4:337–353. doi: 10.2174/1566524043360645
73
15. Butler DA, Scott MR, Bockman JM, Borchelt DR, Taraboulos A, et al. (1988). Scrapie-
infected murine neuroblastoma cells produce protease-resistant prion proteins. J Virol
62:1558–1564.
16. Race RE, Fadness LH, Chesebro B (1987). Characterization of scrapie infection in
mouse neuroblastoma cells. J Gen Virol 68:1391–1399. doi: 10.1099/0022-1317-68-5-
1391
17. Herbst A, Banser P, Velasquez CD, Mays CE, Sim VL, et al. (2013). Infectious Prions
Accumulate to High Levels in Non Proliferative C2C12 Myotubes. PLoS Pathog
9:e1003755. doi: 10.1371/journal.ppat.1003755
18. Malaga-Trillo E, Sempou E (2009). PrPs: Proteins with a purpose: Lessons from the
zebrafish. Prion 3:129–133. doi: 10.4161/pri.3.3.9651
19. Malaga-Trillo E, Solis GP, Schrock Y, Geiss C, Luncz L, et al. (2009). Regulation of
embryonic cell adhesion by the prion protein. PLoS Biol 7:e55. doi:
10.1371/journal.pbio.1000055
20. Gaj T, Gersbach CA, Barbas CF 3rd (2013). ZFN, TALEN, and CRISPR/Cas-based
methods for genome engineering. Trends Biotech 31:397–405. doi:
10.1016/j.tibtech.2013.04.004
21. Horvath P, Barrangou R (2010). CRISPR/Cas, the immune system of bacteria and
archaea. Science 327:167–170. doi: 10.1126/science.1179555
22. Wiedenheft B, Sternberg SH, Doudna JA (2012). RNA-guided genetic silencing
systems in bacteria and archaea. Nature 482:331–338. doi: 10.1038/nature10886
23. Bhaya D, Davison M, Barrangou R (2011). CRISPR-Cas systems in bacteria and
archaea: versatile small RNAs for adaptive defense and regulation. Annu Rev Genet
45:273–297. doi: 10.1146/annurev-genet-110410-132430
24. Hsu PD, Lander ES, Zhang F (2014). Development and Applications of CRISPR-Cas9
for Genome Engineering. Cell 157:1262–1278. doi: 10.1016/j.cell.2014.05.010
25. Mali P, Esvelt KM, Church GM (2013). Cas9 as a versatile tool for engineering biology.
Nat Meth 10:957–963. doi: 10.1038/nmeth.2649
26. Ebina H, Misawa N, Kanemura Y, Koyanagi Y (2013). Harnessing the CRISPR/Cas9
system to disrupt latent HIV-1 provirus. Sci Rep 3:2510. doi: 10.1038/srep02510
27. Cong L, Ran FA, Cox D, Lin S, Barretto R, et al. (2013). Multiplex genome engineering
using CRISPR/Cas systems. Science 339:819–823. doi: 10.1126/science.1231143
28. Mali P, Yang L, Esvelt KM, Aach J, Guell M, et al. (2013). RNA-guided human
genome engineering via Cas9. Science 339:823–826. doi: 10.1126/science.1232033
29. Niu Y, Shen B, Cui Y, Chen Y, Wang J, et al. (2014). Generation of gene-modified
cynomolgus monkey via Cas9/RNA-mediated gene targeting in one-cell embryos. Cell
156:836–843. doi: 10.1016/j.cell.2014.01.027
30. Cathomen T, Ehl S (2014). Translating the genomic revolution - targeted genome
editing in primates. N Engl J Med 370:2342–2345. doi: 10.1056/nejmcibr1403629
74
31. Rutishauser D, Mertz KD, Moos R, Brunner E, Rulicke T, Calella AM, et al. (2009).
The comprehensive native interactome of a fully functional tagged prion protein. PloS
One 4: e4446. doi: 10.1371/journal.pone.0004446
32. Watts JC, Huo H, Bai Y, Ehsani S, Jeon AH, Shi T, et al. (2009). Interactome analyses
identify ties of PrP and its mammalian paralogs to oligomannosidic N-glycans and
endoplasmic reticulum-derived chaperones. PLoS Pathog 5: e1000608. doi:
10.1371/journal.ppat.1000608
33. Schmitt-Ulms G, Hansen K, Liu J, Cowdrey C, Yang J, DeArmond SJ, et al. (2004).
Time-controlled transcardiac perfusion cross-linking for the study of protein
interactions in complex tissues. Nat Biotech 22: 724–731. doi: 10.1038/nbt969
34. Schmitt-Ulms G, Legname G, Baldwin MA, Ball HL, Bradon N, Bosque PJ, et al.
(2001). Binding of neural cell adhesion molecules (N-CAMs) to the cellular prion
protein. J Mol Biol 314: 1209–1225. doi: 10.1006/jmbi.2000.5183
35. Santuccione A, Sytnyk V, Leshchyns'ka I, Schachner M. (2005). Prion protein recruits
its neuronal receptor NCAM to lipid rafts to activate p59fyn and to enhance neurite
outgrowth. J Cell Biol 169: 341–354. doi: 10.1083/jcb.200409127
36. Lehembre F, Yilmaz M, Wicki A, Schomber T, Strittmatter K, Ziegler D, et al. (2008).
NCAM-induced focal adhesion assembly: a functional switch upon loss of E-cadherin.
EMBO J 27: 2603–2615. doi: 10.1038/emboj.2008.178
37. Schmitt-Ulms G, Ehsani S, Watts JC, Westaway D, Wille H. (2009). Evolutionary
descent of prion genes from the ZIP family of metal ion transporters. PLoS One 4:
e7208. doi: 10.1371/journal.pone.0007208
38. Ehsani S, Tao R, Pocanschi CL, Ren H, Harrison PM, Schmitt-Ulms G. (2011).
Evidence for retrogene origins of the prion gene family. PLoS One 6: e26800. doi:
10.1371/journal.pone.0026800
39. Yamashita S, Miyagi C, Fukada T, Kagara N, Che YS, Hirano T. (2004). Zinc
transporter LIVI controls epithelial-mesenchymal transition in zebrafish gastrula
organizer. Nature 429: 298–302. doi: 10.1038/nature02545
40. Mathews WR, Wang F, Eide DJ, Van Doren M. (2005). Drosophila fear of intimacy
encodes a Zrt/IRT-like protein (ZIP) family zinc transporter functionally related to
mammalian ZIP proteins. J Biol Chem 280: 787–795. doi: 10.1074/jbc.m411308200
41. Van Doren M, Mathews WR, Samuels M, Moore LA, Broihier HT, Lehmann R. (2003).
Fear of intimacy encodes a novel transmembrane protein required for gonad
morphogenesis in Drosophila. Development 130: 2355–2364. doi: 10.1242/dev.00454
42. Kall L, Canterbury JD, Weston J, Noble WS, MacCoss MJ (2007). Semi-supervised
learning for peptide identification from shotgun proteomics datasets. Nat Meth 4:923–
925. doi: 10.1038/nmeth1113
43. Saleh-Gohari N, Helleday T (2004). Conservative homologous recombination
preferentially repairs DNA double-strand breaks in the S phase of the cell cycle in
human cells. Nucleic Acids Res 32:3683–3688. doi: 10.1093/nar/gkh703
75
44. Jinek M, Chylinski K, Fonfara I, Hauer M, Doudna JA, et al. (2012). A Programmable
Dual-RNA–Guided DNA Endonuclease in Adaptive Bacterial Immunity. Science
337:816–821. doi: 10.1126/science.1225829
45. Hsu PD, Scott DA, Weinstein JA, Ran FA, Konermann S, et al. (2013). DNA targeting
specificity of RNA-guided Cas9 nucleases. Nat Biotech 31:827–832. doi:
10.1038/nbt.2647
46. Popp MW, Maquat LE (2013). Organizing principles of mammalian nonsense-mediated
mRNA decay. Annu Rev Genet 47:139–165. doi: 10.1146/annurev-genet-111212-
133424
47. Kuzmiak HA, Maquat LE (2006). Applying nonsense-mediated mRNA decay research
to the clinic: progress and challenges. Trends Mol Med 12:306–316. doi:
10.1016/j.molmed.2006.05.005
48. 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: 2021–2036. doi: 10.1083/jcb.127.6.2021
49. David G, Van der Schueren B, Bernfield M. Basal lamina formation by normal and
transformed mouse mammary epithelial cells duplicated in vitro. J Natl Cancer Inst
1981; 67: 719–728.
50. Hoffman S, Sorkin BC, White PC, Brackenbury R, Mailhammer R, Rutishauser U, et al.
Chemical characterization of a neural cell adhesion molecule purified from embryonic
brain membranes. J Biol Chem 1982; 257: 7720–7729.
51. Shen B, Zhang W, Zhang J, Zhou J, Wang J, et al. (2014). Efficient genome
modification by CRISPR-Cas9 nickase with minimal off-target effects. Nat Meth
11:399–402. doi: 10.1038/nmeth.2857
52. Cho SW, Kim S, Kim Y, Kweon J, Kim HS, et al. (2014). Analysis of off-target effects
of CRISPR/Cas-derived RNA-guided endonucleases and nickases. Genome Res
24:132–141. doi: 10.1101/gr.162339.113
53. Ran FA, Hsu PD, Lin CY, Gootenberg JS, Konermann S, et al. (2013). Double nicking
by RNA-guided CRISPR Cas9 for enhanced genome editing specificity. Cell 154:1380–
1389. doi: 10.1016/j.cell.2013.08.021
54. Fu Y, Foden JA, Khayter C, Maeder ML, Reyon D, et al.. (2013). High-frequency off-
target mutagenesis induced by CRISPR-Cas nucleases in human cells. Nat Biotech
31(9):822-6. doi: 10.1038/nbt.2623
55. Gauczynski S, Peyrin JM, Haik S, Leucht C, Hundt C, et al. (2001). The 37-kDa/67-
kDa laminin receptor acts as the cell-surface receptor for the cellular prion protein.
EMBO J 20:5863–5875. doi: 10.1093/emboj/20.21.5863
56. Herms JW, Korte S, Gall S, Schneider I, Dunker S, et al. (2000). Altered intracellular
calcium homeostasis in cerebellar granule cells of prion protein-deficient mice. J
Neurochem 75:1487–1492. doi: 10.1046/j.1471-4159.2000.0751487.x
57. Juanes ME, Elvira G, Garcia-Grande A, Calero M, Gasset M (2009). Biosynthesis of
prion protein nucleocytoplasmic isoforms by alternative initiation of translation. J Biol
Chem 284:2787–2794. doi: 10.1074/jbc.m804051200
76
58. Cortes CJ, Qin K, Norstrom EM, Green WN, Bindokas VP (2013). Early Delivery of
Misfolded PrP from ER to Lysosomes by Autophagy. Intern J Cell Biol 2013: 560421.
doi: 10.1155/2013/560421
59. Vorberg I, Raines A, Story B, Priola SA (2004). Susceptibility of common fibroblast
cell lines to transmissible spongiform encephalopathy agents. J Infect Dis 189:431–439.
doi: 10.1086/381166
60. Kaneko K, Zulianello L, Scott M, Cooper CM, Wallace AC, et al. (1997). Evidence for
protein X binding to a discontinuous epitope on the cellular prion protein during scrapie
prion propagation. Proc Natl Acad Sci U S A 94:10069–10074. doi:
10.1073/pnas.94.19.10069
61. Kang SG, Roh YM, Kang ML, Kim YS, Yoo HS (2010). Mouse neuronal cells
expressing exogenous bovine PRNP and simultaneous downregulation of endogenous
mouse PRNP using siRNAs. Prion 4:32–37. doi: 10.4161/pri.4.1.11218
62. Crecelius AC, Helmstetter D, Strangmann J, Mitteregger G, Frohlich T, et al. (2008).
The brain proteome profile is highly conserved between Prnp-/- and Prnp+/+ mice.
Neuroreport 19:1027–1031. doi: 10.1097/wnr.0b013e3283046157
63. Weiss E, Ramljak S, Asif AR, Ciesielczyk B, Schmitz M, et al. (2010). Cellular prion
protein overexpression disturbs cellular homeostasis in SH-SY5Y neuroblastoma cells
but does not alter p53 expression: a proteomic study. Neuroscience 169:1640–1650. doi:
10.1016/j.neuroscience.2010.06.013
64. Stella R, Cifani P, Peggion C, Hansson K, Lazzari C, et al. (2012). Relative
quantification of membrane proteins in wild-type and prion protein (PrP)-knockout
cerebellar granule neurons. J Prot Res 11:523–536. doi: 10.1021/pr200759m
65. Puckett C, Concannon P, Casey C, Hood L. (1991). Genomic structure of the human
prion protein gene. Amer J Hum Genet 49: 320–329.
66. Basler K, Oesch B, Scott M, Westaway D, Walchli M, Groth DF, et al. (1986). Scrapie
and cellular PrP isoforms are encoded by the same chromosomal gene. Cell 46: 417–
428. pmid:2873895 doi: 10.1016/0092-8674(86)90662-8
67. Saeki K, Matsumoto Y, Hirota Y, Onodera T. (1996). Three-exon structure of the gene
encoding the rat prion protein and its expression in tissues. Virus Genes 12: 15–20.
pmid:8879116 doi: 10.1007/bf00369996
68. Karapetyan YE, Sferrazza GF, Zhou M, Ottenberg G, Spicer T, Chase P, et al. (2013).
Unique drug screening approach for prion diseases identifies tacrolimus and astemizole
as antiprion agents. Proc Natl Acad Sci USA. doi: 10.1073/pnas.1303510110
69. Misiewicz M, Dery MA, Foveau B, Jodoin J, Ruths D, Leblanc AC. (2013).
Identification of a novel Endoplasmic Reticulum Stress Response Element regulated by
XBP1. J Biol Chem. doi: 10.1074/jbc.m113.457242
70. Kiss JZ, Muller D. (2001). Contribution of the neural cell adhesion molecule to
neuronal and synaptic plasticity. Rev Neurosci 12: 297–310. doi:
10.1515/revneuro.2001.12.4.297
77
71. Senkov O, Tikhobrazova O, Dityatev A. (2012). PSA-NCAM: synaptic functions
mediated by its interactions with proteoglycans and glutamate receptors. Int J Biochem
Cell Biol 44: 591–595. doi: 10.1016/j.biocel.2012.01.008
72. Angata K, Long JM, Bukalo O, Lee W, Dityatev A, Wynshaw-Boris A, et al. (2004).
Sialyltransferase ST8Sia-II assembles a subset of polysialic acid that directs
hippocampal axonal targeting and promotes fear behavior. J Biol Chem 279: 32603–
32613. doi: 10.1074/jbc.m403429200
73. Drake PM, Nathan JK, Stock CM, Chang PV, Muench MO, Nakata D, et al. (2008).
Polysialic acid, a glycan with highly restricted expression, is found on human and
murine leukocytes and modulates immune responses. J Immunol 181: 6850–6858. doi:
10.4049/jimmunol.181.10.6850
74. Drake PM, Stock CM, Nathan JK, Gip P, Golden KP, Weinhold B, et al. (2009).
Polysialic acid governs T-cell development by regulating progenitor access to the
thymus. Proc Natl Acad Sci USA 106: 11995–12000. doi: 10.1073/pnas.0905188106
75. Senkov O, Sun M, Weinhold B, Gerardy-Schahn R, Schachner M, Dityatev A. (2006).
Polysialylated neural cell adhesion molecule is involved in induction of long-term
potentiation and memory acquisition and consolidation in a fear-conditioning paradigm.
J Neurosci 26: 10888–109898. doi: 10.1523/jneurosci.0878-06.2006
76. Hammond MS, Sims C, Parameshwaran K, Suppiramaniam V, Schachner M, Dityatev
A. (2006). Neural cell adhesion molecule-associated polysialic acid inhibits NR2B-
containing N-methyl-D-aspartate receptors and prevents glutamate-induced cell death. J
Biol Chem 281: 34859–34869. doi: 10.1074/jbc.m602568200
77. Kochlamazashvili G, Senkov O, Grebenyuk S, Robinson C, Xiao MF, Stummeyer K, et
al. (2010). Neural cell adhesion molecule-associated polysialic acid regulates synaptic
plasticity and learning by restraining the signaling through GluN2B-containing NMDA
receptors. J Neurosci 30: 4171–4183. doi: 10.1523/JNEUROSCI.5806-09.2010
78. Glass JD, Lee W, Shen H, Watanabe M. (1994). Expression of immunoreactive
polysialylated neural cell adhesion molecule in the suprachiasmatic nucleus.
Neuroendocrin 60: 87–95. doi: 10.1159/000126723
79. Shen H, Watanabe M, Tomasiewicz H, Glass JD. (2001). Genetic deletions of NCAM
and PSA impair circadian function in the mouse. Physiol Behav 73: 185–193. doi:
10.1016/s0031-9384(01)00468-1
80. Glass JD, Watanabe M, Fedorkova L, Shen H, Ungers G, Rutishauser U. (2003).
Dynamic regulation of polysialylated neural cell adhesion molecule in the
suprachiasmatic nucleus. Neuroscience 117: 203–211. doi: 10.1016/s0306-
4522(02)00817-5
81. Prosser RA, Rutishauser U, Ungers G, Fedorkova L, Glass JD. (2003). Intrinsic role of
polysialylated neural cell adhesion molecule in photic phase resetting of the Mammalian
circadian clock. J Neurosci 23: 652–658.
82. Shen H, Watanabe M, Tomasiewicz H, Rutishauser U, Magnuson T, Glass JD. (1997).
Role of neural cell adhesion molecule and polysialic acid in mouse circadian clock
function. J Neurosci 17: 5221–5229. pmid:9185559
78
83. Jungnickel J, Eckhardt M, Haastert-Talini K, Claus P, Bronzlik P, Lipokatic-Takacs E,
et al. (2012). Polysialyltransferase overexpression in Schwann cells mediates different
effects during peripheral nerve regeneration. Glycobiology 22: 107–115. doi:
10.1093/glycob/cwr113
84. Carratu MR, Steardo L, Cuomo V. (1996). Role of polysialic acid in peripheral
myelinated axons. Microsc Res Tech 34: 489–491. doi: 10.1002/(sici)1097-
0029(19960815)34:6<489::aid-jemt1>3.0.co;2-f
85. Mehanna A, Mishra B, Kurschat N, Schulze C, Bian S, Loers G, et al. (2009). Polysialic
acid glycomimetics promote myelination and functional recovery after peripheral nerve
injury in mice. Brain 132: 1449–1462. doi: 10.1093/brain/awp128
86. Massaro AR. (2002). The role of NCAM in remyelination. Neurol Sci 22: 429–435. doi:
10.1007/s100720200001
87. Theodosis DT, Rougon G, Poulain DA. (1991). Retention of embryonic features by an
adult neuronal system capable of plasticity: polysialylated neural cell adhesion molecule
in the hypothalamo-neurohypophysial system. Proc Natl Acad Sci USA 88: 5494–5498.
doi: 10.1073/pnas.88.13.5494
88. Bonfanti L. (2006). PSA-NCAM in mammalian structural plasticity and neurogenesis.
Prog Neurobiol 80: 129–164. doi: 10.1016/j.pneurobio.2006.08.003
89. Gascon E, Vutskits L, Kiss JZ. (2010). The role of PSA-NCAM in adult neurogenesis.
Adv Exp Med Biol 663: 127–136. doi: 10.1007/978-1-4419-1170-4_8
90. Steele AD, Lindquist S, Aguzzi A. (2007). The prion protein knockout mouse: a
phenotype under challenge. Prion 1: 83–93. doi: 10.4161/pri.1.2.4346
91. Smith C, Gore A, Yan W, Abalde-Atristain L, Li Z, et al. (2014) Whole-Genome
Sequencing Analysis Reveals High Specificity of CRISPR/Cas9 and TALEN-Based
Genome Editing in Human iPSCs. Cell Stem Cell 15:12–13. doi:
10.1016/j.stem.2014.06.011
92. Veres A, Gosis BS, Ding Q, Collins R, Ragavendran A, et al. (2014) Low Incidence of
Off-Target Mutations in Individual CRISPR-Cas9 and TALEN Targeted Human Stem
Cell Clones Detected by Whole-Genome Sequencing. Cell Stem Cell 15:27–30. doi:
10.1016/j.stem.2014.04.020
93. Brethour D, Mehrabian M, Ehsani S, Williams D, Xi Z, Rogaeva E, Schmitt-Ulms G.
(2016). A ZIP6-ZIP10 heteromer interacts with NCAM1 controlling its phosphorylation
and integration into focal adhesion complexes during epithelial-to-mesenchymal
transition. In prep.
94. Guerinot ML. (2000). The ZIP family of metal transporters. Biochim Biophys Acta
1465: 190-198. doi: 10.1016/S0005-2736(00)00138-3
95. Eide DJ. (2004). The SLC39 family of metal ion transporters. Pflugers Arch 447: 796-
800. doi: 10.1016/j.mam.2012.05.011
96. Ehsani S, Huo H, Salehzadeh A, Pocanschi CL, Watts JC, Wille H, et al. (2011). Family
reunion--the ZIP/prion gene family. Prog Neurobiol 93: 405-420. doi:
10.1016/j.pneurobio.2010.12.001
79
97. Ehsani S, Mehrabian M, Pocanschi CL, Schmitt-Ulms G. (2012). The ZIP-prion
connection. Prion 6: 317-321. doi: 10.4161/pri.20196
98. Hogstrand C, Kille P, Ackland ML, Hiscox S, Taylor KM. (2013). A Mechanism for
Epithelial-Mesenchymal Transition and Anoikis Resistance in Breast Cancer Triggered
by Zinc Channel ZIP6 and Signal Transducer and Activator of Transcription 3
(STAT3). Biochem J 455(2): 229-37. doi: 10.1042/BJ20130483
99. Rauniyar N, Yates JR, 3rd. (2014). Isobaric labeling-based relative quantification in
shotgun proteomics. J Proteome Res 13: 5293-5309. doi: 10.1021/pr500880b
100. Croxford TP, McCormick NH, Kelleher SL. (2011). Moderate zinc deficiency reduces
testicular Zip6 and Zip10 abundance and impairs spermatogenesis in mice. J Nutr. 141:
359-365. doi: 10.3945/jn.110.131318
101. Kong BY, Duncan FE, Que EL, Kim AM, O'Halloran TV, Woodruff TK. (2014).
Maternally-derived zinc transporters ZIP6 and ZIP10 drive the mammalian oocyte-to-
egg transition. Mol Hum Reprod 20: 1077-1089. doi: 10.1093/molehr/gau066
102. Taylor KM, Muraina I, Brethour D, Schmitt-Ulms G, Nimmanon T, Kille P, et al.
(2016). Zinc transporter ZIP10 forms a heteromer with ZIP6 and stimulates embryonic
development and cell migration. Biochem J; Under review.
103. Ehsani S, Salehzadeh A, Huo H, Reginold W, Pocanschi CL, Ren H, et al. (2012). LIV-
1 ZIP ectodomain shedding in prion-infected mice resembles cellular response to
transition metal starvation. J Mol Biol 422: 556-574. doi: 10.1016/j.jmb.2012.06.003
104. Zieske LR. (2006). A perspective on the use of iTRAQ reagent technology for protein
complex and profiling studies. J Exp Bot 57: 1501-1508. doi: 10.1093/jxb/erj168
105. Sutherland C. (2011). What Are the bona fide GSK3 Substrates? Int J Alzheimers Dis
2011: 505607. doi: 10.4061/2011/505607
106. Cole AR, Knebel A, Morrice NA, Robertson LA, Irving AJ, Connolly CN, et al. (2004).
GSK-3 phosphorylation of the Alzheimer epitope within collapsin response mediator
proteins regulates axon elongation in primary neurons. J Biol Chem 279: 50176-50180.
doi: 10.1074/jbc.C400412200
107. Menezo Y, Pluntz L, Chouteau J, Gurgan T, Demirol A, Dalleac A, et al. (2011). Zinc
concentrations in serum and follicular fluid during ovarian stimulation and expression
of Zn2+ transporters in human oocytes and cumulus cells. Reprod Biomed Online
22(6): 647-652. doi: 10.1016/j.rbmo.2011.03.015
108. Watts JC, Drisaldi B, Ng V, Yang J, Strome B, Horne P, et al. (2007). The CNS
glycoprotein Shadoo has PrP(C)-like protective properties and displays reduced levels
in prion infections. EMBO J 26: 4038-4050. doi: 10.1038/sj.emboj.7601830
109. Prodromidou K, Papastefanaki F, Sklaviadis T, Matsas R. (2014). Functional cross-talk
between the cellular prion protein and the neural cell adhesion molecule NCAM is
critical for neuronal differentiation of neural stem/precursor cells. Stem Cells 32(6):
1674-87. doi: 10.1002/stem.1663
110. Pocanschi CL, Ehsani S, Mehrabian M, Wille H, Reginold W, Trimble WS, et al.
(2013). The ZIP5 Ectodomain Co-Localizes with PrP and May Acquire a PrP-Like Fold
80
That Assembles into a Dimer. PLoS One 8(9): e72446. doi:
10.1371/journal.pone.0072446
111. Ilouz R, Kaidanovich O, Gurwitz D, Eldar-Finkelman H. (2002). Inhibition of glycogen
synthase kinase-3beta by bivalent zinc ions: insight into the insulin-mimetic action of
zinc. Biochem Biophys Res Commun 295: 102-106. doi: 10.1016/S0006-
291X(02)00636-8
112. Thomas GM, Frame S, Goedert M, Nathke I, Polakis P, Cohen P. (1999). A GSK3-
binding peptide from FRAT1 selectively inhibits the GSK3-catalysed phosphorylation
of axin and beta-catenin. FEBS Lett 458: 247-251. doi: 10.1016/S0014-5793(99)01161-
8
113. Chen Y, Fu AK, Ip NY. (2013). Axin: an emerging key scaffold at the synapse. IUBMB
Life 65: 685-691. doi: 10.1002/iub.1184
114. Mackie K, Sorkin BC, Nairn AC, Greengard P, Edelman GM, Cunningham BA. (1989).
Identification of two protein kinases that phosphorylate the neural cell-adhesion
molecule, N-CAM. J Neurosci 9: 1883-1896.
115. Little EB, Crossin KL, Krushel LA, Edelman GM, Cunningham BA. (2001). A short
segment within the cytoplasmic domain of the neural cell adhesion molecule (N-CAM)
is essential for N-CAM-induced NF-kappa B activity in astrocytes. Proc Nat Acad Sci
USA 98(5): 2238-2243. doi: 10.1073/pnas.041597098
116. Ballif BA, Villen J, Beausoleil SA, Schwartz D, Gygi SP. (2004). Phosphoproteomic
analysis of the developing mouse brain. Mol Cell Proteomics 3(11): 1093-1101. doi:
10.1074/mcp.M400085-MCP200
117. Korshunova I, Caroni P, Kolkova K, Berezin V, Bock E, Walmod PS. (2008).
Characterization of BASP1-mediated neurite outgrowth. J Neurosci Res 86: 2201-2213.
doi: 10.1002/jnr.21678
118. Mehrabian M, Hildebrandt H, Schmitt-Ulms G. (2016). NCAM1 polysialylation: the
prion protein’s elusive reason for being? In prep.
119. Moore RC, Lee IY, Silverman GL, Harrison PM, Strome R, et al. (1999). Ataxia in
prion protein (PrP)-deficient mice is associated with upregulation of the novel PrP-like
protein doppel. J Mol Biol 292(4): 797-817. doi: 10.1006/jmbi.1999.3108
120. Premzl M, Sangiorgio L, Strumbo B, Marshall Graves JA, Simonic T, Gready JE.
(2003). Shadoo, a new protein highly conserved from fish to mammals and with
similarity to prion protein. Gene 314: 89–102. doi: 10.1016/S0378-1119(03)00707-8
121. Behrens A, Genoud N, Naumann H, Ru¨licke T, Janett F, Heppner FL, Ledermann B,
Aguzzi A. (2002). Absence of the prion protein homologue Doppel causes male
sterility. EMBO J 21: 1–7. doi: 10.1093/emboj/cdf386
122. Paisley D, Banks S, Selfridge J, McLennan NF, Ritchie AM, McEwan C, Irvine DS,
Saunders PT, Manson JC, Melton DW. (2004). Male infertility and DNA damage in
Doppel knockout and prion protein/Doppel double-knockout mice. Am J Pathol 164:
2279–2288. doi: 10.1016/S0002-9440(10)63784-4
81
123. Guo H, Jin X, Zhu T, Wang T, Tong P, Tian L, et al. (2014). SLC39A5 mutations
interfering with the BMP/TGF-beta pathway in non-syndromic high myopia. J Med
Genet 51, 518–525. doi: 10.1136/jmedgenet-2014-102351
124. Mehrabian M, Brethour D, Williams D, Wang H, Schneider B, Schmitt-Ulms G. (2016).
Global proteome comparisons of diverse mouse models point toward a role of the prion
protein in the stabilization of signaling hubs comprising members of the Marcks protein
family. Under Review.
125. Katagiri T, Yamaguchi A, Komaki M, Abe E, Takahashi N, et al. (1994). Bone
morphogenetic protein-2 converts the differentiation pathway of C2C12 myoblasts into
the osteoblast lineage. J Cell Biol 127: 1755-66.
126. Vizcaino JA, Deutsch EW, Wang R, Csordas A, Reisinger F, et al. (2014)
ProteomeXchange provides globally coordinated proteomics data submission and
dissemination. Nat Biotechnol 32:223–226. doi: 10.1038/nbt.2839
127. Vizcaino JA, Cote RG, Csordas A, Dianes JA, Fabregat A, et al. (2013) The
PRoteomics IDEntifications (PRIDE) database and associated tools: status in 2013.
Nucleic Acids Res 41:D1063–1069. doi: 10.1093/nar/gks126
128. Kolkova K. (2010). Biosynthesis of NCAM. Adv Exp Med Biol 663, 213‐25.
129. Edelman GM. (1984). Modulation of cell adhesion during induction, histogenesis, and
perinatal development of the nervous system. Annu Rev Neurosci 7, 339‐77.
130. Soroka V, Kasper C, Poulsen FM. (2010). Structural biology of NCAM. Adv Exp Med
Biol 663, 3‐22.
131. Nielsen J, Kulahin N, Walmod PS. (2010). Extracellular protein interactions mediated
by the neural cell adhesion molecule, NCAM: heterophilic interactions between NCAM
and cell adhesion molecules, extracellular matrix proteins, and viruses. Adv Exp Med
Biol 663, 23‐53. doi: 10.1007/978-1-4419-1170-4_2
132. Nelson RW, Bates PA, Rutishauser U. (1995). Protein determinants for specific
polysialylation of the neural cell adhesion molecule. J Biol Chem 270, 17171‐9. doi:
10.1074/jbc.270.29.17171
133. Lichten LA, Cousins RJ. (2009). Mammalian zinc transporters: nutritional and
physiologic regulation. Annu Rev Nutr 29: 153-76. doi: 10.1146/annurev-nutr-033009-
083312
134. Qin K, Zhao L, Ash RD, McDonough WF, Zhao RY. (2009). ATM-mediated
transcriptional elevation of prion in response to copper-induced oxidative stress. J Biol
Chem 284: 4582-93. doi: 10.1074/jbc.M808410200
135. Watt NT, Hooper NM. (2003). The prion protein and neuronal zinc homeostasis. Trends
Biochem Sci 28: 406-410. doi: 10.1016/S0968-0004(03)00166-X
136. Rachidi W, Chimienti F, Aouffen M, Senator A, Guiraud P, Seve M, Favier A. (2009).
Prion protein protects against zinc-mediated cytotoxicity by modifying intracellular
exchangeable zinc and inducing metallothionein expression. J Trace Elem Biol 23: 214-
23. doi: 10.1016/j.jtemb.2009.02.007
82
137. Kong Q, Huang S, Zou W, Vanegas D, Wang M, et al. (2005). Chronic wasting disease
of elk: transmissibility to humans examined by transgenic mouse models. J Neurosci
25(35): 7944-49. doi: 10.1523/JNEUROSCI.2467-05.2005
138. Wilson R, Plinston C, Hunter N, Casalone C, Corona C, et al. (2012). Chronic wasting
disease and atypical forms of bovine spongiform encephalopathy and scrapie are not
transmissible to mice expressing wild-type levels of human prion protein. J Gen Virol
93: 1624-29. doi: 10.1099/vir.0.042507-0
139. Kurt TD, Jiang L, Fernández-borges N, Bett C, Liu J, et al. (2015). Human prion protein
sequence elements impede cross-species chronic wasting disease transmission. J Clin
Invest 125(4): 1485-96. doi: 10.1172/JCI79408
140. (2012). A new way of looking at the impact of dementia in Canada. Alzheimer Society.
141. Balducci C, Beeg M, Stravalaci M, Bastone A, Sclip A, et al. (2010). Synthetic
amyloid-beta oligomers impair long-term memory independently of cellular prion
protein. Proc Natl Acad Sci 107: 2295-300. doi: 10.1073/pnas.0911829107
142. Cisse M, Sanchez PE, Kim DH, Ho K, Yu GQ, Mucke L. (2011). Ablation of cellular
prion protein does not ameliorate abnormal neural network activity or cognitive
dysfunction in the J20 line of human amyloid precursor protein transgenic mice. J
Neurosci 31: 10427-31. doi: 10.1523/JNEUROSCI.1459-11.2011
143. Benilova I, De Strooper B. (2010). Prion protein in Alzheimer's pathogenesis: a hot and
controversial issue. EMBO Mol Med 2: 289-90. doi: 10.1002/emmm.201000088
144. Simon P, Bäumner S, Busch O, Röhrich R, Kaese M, Richterich P, et al. (2013).
Polysialic Acid Is Present in Mammalian Semen as a Post-translational Modification of
the Neural Cell Adhesion Molecule NCAM and the Polysialyltransferase ST8SiaII. J
Biol Chem 288: 18825–18833. doi: 10.1074/jbc.M113.451112
145. Hänsch M, Simon P, Schön J, Kaese M, Braun BC, Jewgenow K, et al. (2014).
Polysialylation of NCAM correlates with onset and termination of seasonal
spermatogenesis in roe deer. Glycobiology 24: 488–493. doi: 10.1093/glycob/cwu023.
146. Geiser J, De Lisle RC, Andrews GK. (2013). The zinc transporter Zip5 (Slc39a5)
regulates intestinal zinc excretion and protects the pancreas against zinc toxicity. PLoS
One 8(11): e82149. doi: 10.1371/journal.pone.0082149
147. Roucou X, Giannopoulos PN, Zhang Y, Jodoin J, Goodyer CG, LeBlanc A. (2005).
Cellular prion protein inhibits proapoptotic Bax conformational change in human
neurons and in breast carcinoma MCF-7 cells. Cell Death Differ 12:783-795. doi:
10.1038/sj.cdd.4401629
148. Du J, Pan Y, Shi Y, Guo C, Jin X, et al. (2005). Overexpression and significance of
prion protein in gastric cancer and multidrug-resistant gastric carcinoma cell line
SGC7901/ADR. Int J Cancer 113:213-220. doi: 10.1002/ijc.20570
149. Meslin F, Conforti R, Mazouni C, Morel N, Tomasic G, et al. (2007). Efficacy of
adjuvant chemotherapy according to Prion protein expression in patients with estrogen
receptor-negative breast cancer. Ann Oncol 18:1793-1798. doi:
10.1093/annonc/mdm406
83
150. Taylor KM, Morgan HE, Smart K, Zahari NM, Pumford S, et al. (2007). The emerging
role of the LIV-1 subfamily of zinc transporters in breast cancer. Mol Med 13: 396-406.
doi: 10.2119/2007-00040.Taylor
151. Sollazzo V, Galasso M, Volinia S, Carinci F. (2012). Prion proteins (PRNP and PRND)
are over-expressed in osteosarcoma. J Orthop Res 30:1004-1012. doi:
10.1002/jor.22034
152. Yang X, Zhang Y, Zhang L, He T, Zhang J, Li C. (2014). Prion protein and cancers.
Acta Biochim Biophys Sin 46: 431-440. doi: 10.1093/abbs/gmu019