Research Collection

Doctoral Thesis

The Importance of Clathrin-Mediated Endocytosis in Adult Myelinating Schwann Cells

Author(s): Gerber, Daniel Paul

Publication Date: 2016

Permanent Link: https://doi.org/10.3929/ethz-a-010735322

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library ACHTUNG: ERSTE SETIE NICHT DRUCKEN (Druckbefehl von: 1-200)

DISS. ETH No. 23721

The Importance of Clathrin-Mediated Endocytosis in Adult Myelinating Schwann Cells

a thesis submitted to attain the degree of

DOCTOR OF SCIENCES of ETH ZURICH (Dr. sc. ETH Zurich)

presented by

DANIEL PAUL GERBER MSc ETH Biology

born 03.10.1987

citizen of Zurich (ZH)

accepted on the recommendation of

Prof. Dr. Ueli Suter (referee) Prof. Dr. Claire Jacob (co-referee) Prof. Dr. Bernd Wollscheid (co-referee) Dr. Axel Niemann (co-referee)

2016

Introduction

2

Summary

Mutations in dynamin 2 (DNM2) have been reported to cause Charcot-Marie-Tooth (CMT) disease, the most prevalent hereditary neuropathy of the peripheral nervous system. The ubiquitously expressed dynamin 2 has a wide range of cellular functions, being involved in various forms of endocytosis, vesicular trafficking and cytoskeletal remodelling. However, its physiological role in the CMT-relevant cell types is poorly understood, which has made it difficult to understand the pathomechanisms leading to CMT. Hence, we generated a -specific inducible knockout mouse (Dnm2iko) to investigate the physiological role of Dnm2 in adult myelinating Schwann cells. Four weeks after induction, Dnm2iko mice developed a severe demyelinating phenotype with paraparesis followed by spontaneous recovery, a clinical course reminiscent of an acute inflammatory demyelinating polyneuropathy (AIDP), the most common subtype of Guillain-Barré-Syndrom (GBS). At a histological level, the demyelination was accompanied by a strong inflammatory response with a prominent macrophage infiltrate. Interestingly, pharmacological ablation of macrophages led to a delayed clinical recovery, thus pointing towards a beneficial role of these inflammatory cells. We than took advantage of a YFP-reporter mouse line to determine the identity of the demyelinating and remyelinating SCs. To our surprise we could show that demyelination in Dnm2iko mice was followed by apoptosis of all the recombined Schwann cells (70 % of all the SCs), and that the remaining 30 % non- recombined SCs were able to repopulate and remyelinate the entire . Overall, these results underline the remarkable plasticity of SCs and show an absolute requirement of dynamin 2 for integrity and survival of adult SCs. A recent study suggested that impaired clathrin-mediated endocytosis (CME) is the major contributing factor in the Dnm2-related forms of CMT. In order to assess the contribution of impaired CME to the observed phenotype in Dnm2iko mice, we generated Schwann cell-specific inducible knockout mice for the µ-subunit (AP2iko) of the adaptor complex 2 (AP-2), which is absolutely required for and exclusively involved in CME. In contrast to Dnm2iko, AP2iko mice did not develop an acute demyelinating neuropathy within the first months after recombination. Instead, aged mice developed a chronic with features of de- and remyelination, the formation of outfoldings and onion bulbs, as well as secondary axonal loss. All of these are classical hallmarks of Charcot-Marie-Tooth disease. Finally, we found that the ablation of AP2µ2 or Dnm2 leads to an iron deficiency due to the impaired uptake of transferrin receptor, which could partially contribute to the observed demyelination. In conclusion, our data indicate that SCs rely on CME for long-term maintenance of peripheral , whereas other functions of dynamin 2 are additionally required to ensure SC survival.

Introduction

4

Zusammenfassung

Spezifische Mutationen im Dynamin 2-Gen führen zu einer Charcot-Marie-Tooth (CMT) Neuropathie. Das ubiquitär exprimierte Dynamin 2 hat eine Vielzahl von Funktionen. Diese reichen von verschiedenen Formen der Endozytose über den intrazellulären Transport von Vesikeln bis hin zur Modulation des Zytoskeletts. Trotzdem ist über die Funktion von Dynamin 2 in denen für CMT relevanten Zelltypen wenig bekannt. Deswegen generierten wir eine induzierbare Knockout-Maus, welche es uns ermöglicht die Expression von Dynamin 2 spezifisch in Schwann Zellen zu inhibieren (Dnm2iko). Vier Wochen nach der Induktion entwickeln diese Mäuse eine starke, aber transiente Neuropathie. Der klinische Phänotyp hat viele Gemeinsamkeiten mit der akuten inflammatorischen demyelinisierenden Polyradikuloneuropathie (AIDP), der häufigsten Unterart des Guillain-Barré-Syndrom (GBS). Wir konnten zeigen, dass die Schwann Zellen in den Dnm2iko Mäusen nach der Demyelinisierung in Apoptose gehen. Dieser Prozess wird von einer starken Entzündungsreaktion begleitet. Die hierbei infiltrierenden Makrophagen sind vorteilhaft für die Regeneration der beschädigten Nerven. Erstaunlicherweise kann der beobachtete Verlust von 70 % aller Schwann Zellen (alle rekombinierten Schwann Zellen) von den 30% nicht-rekombinierten Zellen kompensiert werden. Dies führt innerhalb von zwei Wochen nach der stärksten Beeinträchtigung zur Erholung der betroffenen Mäuse. Zusammenfassend können wir sagen, dass Dynamin 2 absolut notwendig für die Erhaltung der Myelinschicht, sowie für das Überleben von adulten myelinisierenden Schwann Zellen ist. Eine kürzlich erschienene Studie hat gezeigt, dass eine beeinträchtigte Clathrin-vermittelte Endozytose (CME) die Hauptursache für die von mutiertem Dynamin 2 hervorgerufene CMT ist. Um eine Verknüpfung des in den Dnm2iko Mäusen beobachteten Phänotyps mit einer beeinträchtigten CME herzustellen, haben wir eine weitere Knockout-Maus generiert. Hierbei wurde die µ-Untereinheit des Adaptorproteinkomplexes 2 (AP-2) spezifisch in adulten myelinisierenden Schwann Zellen abladiert (AP2iko). AP-2 ist ausschliesslich und exklusiv an CME beteiligt. Im Gegensatz zu den Dnm2iko entwickeln AP2iko Mäuse während der ersten Monate keine transiente Neuropathie. Stattdessen manifestiert sich in gealterten Mäusen eine chronische periphere Neuropathie. In der Pathologie offenbarten sich die klassischen Eigenschaften einer CMT, wie eine De- und Remyelinisierung, die Bildung von Outfoldings und Onion-Bulbs, sowie sekundärer Axonverlust. Weiter konnten wir zeigen, dass der Verlust von AP2µ2 oder Dnm2 aufgrund der fehlenden CME zu einem Eisenmangel führt. Dieser Eisenmangel könnte zu der beobachteten Demyelinisierung beitragen. Abschliessend können wir sagen, dass der Phänotyp in den Dnm2iko Mäusen nicht ausschliesslich auf eine verminderte CME zurück zu führen ist. Dennoch führt der Verlust von CME in Schwann Zellen zu einer mit CMT vergleichbaren Neuropathie.

Introduction

Table of Contents

1 Introduction ...... 1

1.1 The Peripheral Nervous System and its Myelin ...... 1 1.2 The Origin and Development of Schwann Cells ...... 2 1.3 The Process of De- and Remyelination in the PNS ...... 3 1.3.1 Wallerian Degeneration ...... 6 1.3.2 Charcot-Marie-Tooth Neuropathy...... 7 1.3.3 Inflammatory Neuropathies ...... 9 1.4 The Dynamin Superfamily ...... 10 1.5 Dynamin 2 ...... 11 1.6 Cellular Functions of Dynamin 2 ...... 12 1.7 Dynamin 2 in Disease ...... 13 1.8 The Heterotetrameric Adaptor Complex Family ...... 14 1.9 The Adaptor Protein Complex 2 ...... 15 1.10 Clathrin-Mediated Endocytosis ...... 16 1.11 Iron ...... 17 1.12 Objective of the Study ...... 19

2 Results ...... 21

2.1 SC-Specific Ablation of Dynamin 2 in Adult Mice Results in a Remitting Neuropathy ...... 21 2.1.1 Acute De- and Remyelination upon Dynamin 2 Ablation ...... 22 2.1.2 Ablation of Dynamin 2 Causes Schwann Cell Dedifferentiation ...... 24 2.1.3 Schwann Cells Coordinate an Acute Inflammation upon Dynamin 2 Ablation ...... 26 2.1.4 Dynamin 2-Depleted Cells are Replaced by Non-Recombined Schwann Cells ...... 29 2.2 SC-Specific Ablation of AP2µ2 in Adult Mice Results in a Late Onset Neuropathy ...... 31 2.2.1 Aged AP2iko Mice Develop Myelin Aberrations ...... 33 2.2.2 Ablation of AP2µ2 Leads to the Recruitment of Macrophages ...... 36 2.2.3 The Number of Schwann Cells Increases in Aged AP2iko mice ...... 37 2.3 The Surfaceome of CME-Impaired Primary Rat Schwann Cells ...... 38 2.4 Schwann Cells and Iron Deficiency ...... 40 2.5 Iron Deficiency in Different Knockout Mice ...... 41 2.6 The Contribution of Iron Deficiency to the Observed Phenotype in Dnm2iko Mice ...... 43 2.7 Transcriptome Analysis after Dynamin 2 or AP2µ2 Ablation ...... 46

3 Discussion ...... 49

3.1 SC-Specific Ablation of Dynamin 2 in Adult Mice Results in a Remitting Neuropathy ...... 50 3.2 SC-Specific Ablation of AP2µ2 in Adult Mice Results in a Late Onset Neuropathy ...... 54 3.3 Iron Deficiency is a Potential Contributor to the Observed Phenotypes ...... 57

4 Material and Methods ...... 63

4.1 Solutions and Buffers ...... 63 4.2 Schwann Cell-Specific Dynamin 2 Ablation in Adult Mice ...... 64 4.3 Schwann Cell-Specific AP2µ2 Ablation in Adult Mice ...... 64 4.4 Genotyping ...... 65 4.5 Gait Analysis ...... 66

6

4.6 Electron Microscopy ...... 66 4.6.1 Nerve Preparation for Electron Microscopy...... 66 4.6.2 Sample Preparation and Electron Microscopy ...... 67 4.7 Immunological Methods ...... 67 4.7.1 Nerve Preparation for Immunoblotting ...... 67 4.7.2 Protein Gel Electrophoresis ...... 68 4.7.3 Protein Transfer to PVDF Membrane ...... 68 4.7.4 Immunoblotting ...... 68 4.7.5 ELISA ...... 69 4.8 Histological Methods ...... 70 4.8.1 Nerve Preparation and Cryosectioning ...... 70 4.8.2 Immunohistochemistry ...... 70 4.8.3 Proliferation Assay ...... 71 4.8.4 Evens Blue Injections ...... 71 4.8.5 Iron Staining ...... 71 4.8.6 Fluorescence In Situ Hybridization ...... 72 4.8.7 Muscle Staining ...... 72 4.9 Quantitative Real Time – Polymerase Chain Reaction (qRT-PCR) ...... 72 4.9.1 RNA Extraction...... 72 4.9.2 Reverse Transcription ...... 73 4.9.3 qRT-PCR ...... 73 4.10 Cell Culture ...... 74 4.10.1 Commonly used Media and Solutions ...... 74 4.10.2 Isolation of Primary Rat Schwann Cells ...... 75 4.10.3 Culture of Primary Rat Schwann Cells ...... 75 4.10.4 Deferoxamine Treatment of Cells ...... 75 4.10.5 Dorsal Root Ganglia Explant Cultures ...... 76 4.10.6 Production of Low-Titer Lentivirus and Infection of Rat Schwann Cells ...... 76 4.10.7 Transferrin Uptake Assay ...... 77 4.10.8 Production of Inducible shRNAs ...... 77 4.10.9 Cell Surface Capturing ...... 78 4.11 Quantification and Statistics ...... 78 4.11.1 Morphometric Analysis and Quantification of EM Images ...... 78 4.11.2 Morphometric Analysis and Quantification of Immunological Methods ...... 78 4.11.3 Quantification of qRT-PCR ...... 79 4.11.4 Gait Analysis Statistics ...... 79 4.11.5 Statistical Analysis ...... 79

5 Appendix ...... 81

6 References ...... 95

7 Curriculum Vitae ...... 107

8 Acknowledgement ...... 109

Introduction

8 Introduction

1 Introduction

1.1 The Vertebrate Peripheral Nervous System and its Myelin

The nervous system of vertebrate can be subdivided into two parts, the central nervous system (CNS) and the peripheral nervous system (PNS). The CNS consists of the brain and the spinal cord, while the remaining nervous system is defined as the PNS. These two parts differ not only in their function but also in their cellular composition. The nervous system is composed of and glia cells. Most of the neuronal cell bodies lay within the CNS, from which they extend processes, the axons, into the periphery of the body where they connect peripheral targets to the CNS. Additionally to the neurons, the CNS consists of a variety of different glia cells, such as oligodendrocytes, astrocytes and microglia. In the PNS however, there is only one kind of glia cells – the Schwann cells (SCs). The CNS and the PNS work closely together to control movement and the response to external stimuli. Therefore, fast information exchange between PNS and CNS is crucial for appropriate interactions with the environment. According to the theories of Darwin, a fast reaction to an external stimulus is evolutionarily beneficial since it increases the chance that a predator successfully catches its prey or that the prey can avoid it´s predation. There are two ways how the nerve conduction velocity can be increased. One way is to increase the diameter of the axon, since resistance to electrical current is inversely proportional to the cross-sectional area of the axon. This happened for example in the squid, which developed a giant axon that can be up to 100 times thicker than conventional ones. In this way a single axon can reach a diameter of up to 1 mm, which is approximately the size of a murine sciatic nerve containing over 4000 axons [1]. The second possibility to increase nerve conduction velocity is to electrically insulate the axon. In the PNS, this is achieved by Schwann cells, wrapping multiple layers of -rich membrane around large calibre axons. These myelin sheaths are a poor conductor of electrical current. In myelinated axons, voltage-gated sodium channels are restricted to the gaps between two neighbouring Schwann cells called nodes of Ranvier. As a result, action potentials do not have to be generated in the regions between the nodes but jump all the way to the adjacent node. This so called saltatory nerve conduction leads to nerve conduction velocities even faster than that in a giant axon, with the second advantage of being much more space efficient. All , with the exception of the primitive lampreys and hagfishes, use myelin to increase their nerve conduction velocity [2]. However, ensheathing glia can also be found in different . Interestingly, myelin has convergently evolved multiple times in the clade of bilateria further demonstrating its evolutionary benefit [1]. The myelination of axons in the CNS is carried out by oligodendrocytes, which differ from Schwann cells in that a single oligodendrocyte myelinates multiple axons. Both, CNS and PNS myelin is composed of approximately 70 % and 30 % . The lipid composition of both structures is fairly similar, including mainly phospholipids and cholesterol [3]. There is a variety of proteins specific for myelin, such

1 Introduction as (MBP) [4] and myelin-associated (MAG). There are also proteins which are highly expressed in either the CNS, such as proteolipid protein (PLP) and myelin oligodendrocyte glycoprotein (MOG), or in the PNS, such as (MPZ / P0) and peripheral myelin protein 22 (PMP22) [7]. MBP is responsible for accumulating lipids in the myelin sheath by preventing the penetration of unintended proteins into the compact myelin [8]. In the CNS, the transmembrane protein PLP interacts with a neighbouring lipid bilayer in order to hold the two together. The same stabilizing function is carried out by P0 in the PNS, which can bind to another P0 protein inserted into the opposing membrane [9]. This interplay between lipids and proteins ensures a stable and functional myelin sheath.

1.2 The Origin and Development of Schwann Cells

During the process of neurulation, the neural plate invaginates and thereby forms the neural tube. In the course of this process a distinct cell population, the neural crest cells (NCCs), segregates from the most dorsal side of the newly formed neural tube. Depending on the rostrocaudal origin of this neural crest cells, they will develop into different structures. In the head region, these cells will form cartilage and bone, while those originating in the most anterior part of the trunk develop into fibroblasts and smooth muscle cells. NCCs originating from the trunk region migrate either in a lateral direction to give rise to melanocytes in the skin, in a ventro-lateral direction to generate neurons in the dorsal root ganglia (DRG) as well as glia cells, or in a ventral direction to give rise to autonomic neurons, chromaffin cells and glia cells [10, 11]. At around embryonic day (E) 12.5 in the mouse, neural crest cells committed to the Schwann cell (SC) linage undergo a transition towards Schwann cell precursors (SCPs). SCPs migrate together with the outgrowing axons in the developing PNS [12]. With time, SCPs undergo a second transition to become immature Schwann cells. These immature SCs surround bundles of axons and start depositing a common basal lamina. In a process called radial sorting, individual Schwann cells associated with a single large diameter axon segregate from a bundle. These pro-myelinating Schwann cells can now start to myelinate. In contrast, non-myelinating Schwann cells, that after radial sorting remain ensheathing several small calibre axons, form a Remak bundle [13] (Figure 1). During the process of myelination, the pro-myelinating Schwann cells start wrapping multiple layers of lipid-rich membrane around the axon. The number of wraps is determined by the diameter of the axon, or more precisely, by the amount of Neuregulin 1 (Nrg1) type III on the axonal surface [14]. Neuronal Nrg1 binds ErbB2/3, an obligate heteromeric receptor kinase pair, on the Schwann cell surface, activating downstream signaling pathways involved in myelination (reviewed in [15] and [16]). Since large calibre axons have more Nrg1 Type III on their surface, Schwann cells myelinating those axons will produce more myelin than those myelinating small calibre axons. This relationship between axon

2 Introduction

diameter and myelin thickness is very constant and can be described by the g-ratio [17, 18]. The g-ratio is defined as the quotient of the axon diameter and the diameter of the whole fibre (axon + myelin).

Figure 1) The origin and development of Schwann cells Schwann cells (SCs) develop from neural crest cells, which originate from the developing neural tube early in embryonic development. In their first differentiation state, the Schwann cell precursors migrate along outgrowing peripheral nerve axons. In the second differentiation state, the immature Schwann cells extend cytoplasmic processes to completely envelop bundles of axons. In a process called radial sorting, single axons with a diameter ≥ 1 μm are sorted out of the bundle by a single Schwann cell and are subsequently myelinated. The thickness of the myelin sheet is determined by the axon calibre and its surface- bound Neuregulin 1 (Nrg1) Type III, which is sensed by the Schwann cells through the ErbB2/ErbB3 receptor dimer. In contrast to myelinating Schwann cells, non-myelinating Schwann cells engulf several small-calibre axons in a so called Remak bundle. E: embryonic day, P: postnatal day.

1.3 The Process of De- and Remyelination in the PNS

Even though the differentiation of Schwann cells and the subsequent myelination is a tightly regulated process, myelinating Schwann cells are highly plastic. They have the ability to dedifferentiate when they lose axonal contact in injured nerves or when taken in vitro as well as in demyelinating neuropathies [19]. These dedifferentiated SCs have a phenotype in many ways similar to that of immature Schwann cells during development. They show similar expression patterns and are able to re-enter the cell cycle [19, 20]. Dedifferentiated Schwann cells highly upregulate the levels of c-Jun, a potent negative regulator of myelination, which in turn controls the molecular phenotype of these cells [21]. This molecular phenotype includes the upregulation of p75 neurotrophin receptor, the adhesion molecule , the Krox24 (Egr1) as well as genes important for proliferation such as cyclin D1

3 Introduction

[22, 23]. However, c-Jun is not the only negative regulator of myelination. Negative regulators of myelination should feature certain characteristics. They might be active in immature Schwann cells but they should be inactive in myelinating cells. Furthermore, they should be activated under dedifferentiating conditions (e.g. in injured nerves) and their inactivation should inhibit dedifferentiation. They would also be expected to oppose pro-myelinating signals (e.g. Krox20) [19]. Another example of a transcription factor negatively regulating myelination is Notch. This transmembrane receptor is cleaved upon ligand binding, freeing the Notch intracellular domain (NICD) which is then translocated to the nucleus where it acts as a transcriptional regulator [24]. Inactivation of Notch promotes myelination in vivo, while Notch activation delays it. This delay is inversely proportional to the levels of Krox20. The activation of the NICD in healthy myelinating Schwann cells results in severe demyelination [25]. Negative regulators of myelination do not necessarily have to be transcription factors in the first place. The complex intracellular signaling networks responsible for proper differentiation, myelination and myelin maintenance have to be in a fine balance to activate the appropriate transcriptional responses. Tampering with a single player may disturb this balance and induce demyelination. For example, Napoli and colleagues showed that activation of Erk using an inducible Raf-transgene leads to spontaneous demyelination. This work highlights a central role of the Erk-signaling pathway during Schwann cell dedifferentiation [26]. Demyelination not only includes dedifferentiation, but also the breakdown of myelin. Even though this process has mainly been studied during Wallerian degeneration, the mechanism to remove the myelin seems to be remarkably similar during a wide range of demyelinating conditions. During the first phase of myelin clearance, up to one week after injury, the Schwann cells themselves start to break down their myelin [27]. Recent studies have shown that this process involves a selective form of autophagy, myelinophagy, which is driven by an upregulation of c-Jun [28]. Moreover, Schwann cells secrete several chemokines and cytokines including CCL2 (Chemokine (C-C Motif) Ligand 2 / monocyte chemoattractant protein-1 (MCP1)), which is mediated by the Erk-signaling pathway, causing macrophage recruitment and activation [29]. There is a small population of nerve-resident macrophages which proliferates and is able to phagocytose some of the myelin debris. These resident macrophages might also account for some early molecular changes, such as the secretion of cytokines, in injured nerves [30]. However, approximately 80 % of the macrophages responsible for the phagocytosis of the myelin debris have to invade the injured nerve [31]. A more detailed description how the myelin is broken down can be found in the following chapter. Dedifferentiated Schwann cells have the potential to redifferentiate and remyelinate if the conditions are right. This process includes the Krox20-mediated suppression of c-Jun [23]. Krox20 acts as a master regulator of myelination and is sufficient to induce genes critical for myelin formation and maintenance.

4 Introduction

This includes genes encoding the myelin proteins myelin protein zero (P0/MPZ), peripheral myelin protein 22 (PMP22), myelin basic protein (MPB), myelin-associated glycoprotein (MAG), connexin 32 (Cx32/GJB1) and Periaxin (PRX) [32] (Figure 2).

Figure 2) The process of de- and remyelination Myelinating Schwann Cells (SCs) express markers including Krox20 (also called early growth response protein 2 or Egr2), P0 (protein zero), MBP (myelin basic protein) and Periaxin. Upon an injury or at disease onset, Schwann cells have the potential to dedifferentiate. To be able to then proliferate, they have to rid themselves of the myelin – they demyelinate. SCs secrete the potent macrophage chemoattractant CCL2. The recruited macrophages (MΦ) support the Schwann cells in the degradation of their myelin sheath. The dedifferentiated Schwann cells express markers including c-Jun, p75, Krox24 and Cyclin D1. In cases, in which only the axon but not the Schwann cell was injured/impaired, the dedifferentiated Schwann cells, which are in a state similar to pro-myelinating Schwann cells, have the capacity to engage regrowing axons and myelinate them again. If the underlying problem for the demyelination lies within the Schwann cells, other Schwann cells can dedifferentiate, proliferate and replace the lost cell.

5 Introduction

1.3.1 Wallerian Degeneration Upon a peripheral nerve injury or transection, the axonal parts distal to the injury site, which are no longer connected to the neuronal cell body, will degenerate and the myelin will be cleared by Schwann cells and invading macrophages. This distinct process is termed Wallerian degeneration (WD). The earliest events of WD occur within minutes at the site of injury [33]. Initiated by the damage to the axonal membrane, a calcium-mediated proteolytic activity is triggered, starting to break down the axon [34]. From there on, axonal degeneration progresses towards the periphery. Another main part of WD is the myelin removal, which is a prerequisite for efficient regeneration upon injury, since myelin contains molecules which inhibit axonal regrowth [35-38]. In a first phase of myelin clearance Schwann cells dedifferentiate to an immature Schwann cell-like state and start breaking down their myelin [27]. This dedifferentiated Schwann cells upregulate c-Jun, which integrates a broad collection of functions that support nerve regeneration [39]. Recent studies have shown that also the process of myelin breakdown involves the upregulation of c-Jun, which activates a selective form of autophagy termed myelinophagy [28]. Early expression of Schwann cell-derived CCL2, MIP-1α (Macrophage inflammatory protein 1 α), TNF-α ( α) and Il-1β (Interleukin 1 β) stimulate the expression of members of the phospholipase A2 (PLA2) family [39-41]. PLA2 hydrolyses phosphatidylcholine to lysophosphatidylcholine (LPC) and arachidonic acid. LPC can induce further myelin breakdown [42]. TNF-α and Il-1β have additionally been found to mediate myelin breakdown by a Schwann cell-mediated secretion of the extracellular matrix metalloproteinase 9 (MMP-9) [43]. In addition to the Schwann cells themselves, there are two different populations of macrophages important for the process of demyelination. A large fraction of the macrophages has to be recruited into the injured nerve. This migration begins within 2 to 3 days and peaks around 7 days after injury [44]. Many of the factors involved in macrophage recruitment, such as CCL2, MIP-1α, TNF-α and Il-1β, are directly produced by the demyelinating Schwann cells [41, 45]. Schwann cell-derived TNF-α and Il-1β stimulate fibroblasts to express the two cytokines, Il-6 and GM-CSF (Granulocyte-macrophage colony- stimulating factor) [46, 47]. The other, smaller population of nerve-resident macrophages starts proliferating shortly after injury and is able to phagocytose some of the initial myelin debris [30]. Furthermore, those macrophages express TNF-α and Il-1β, boosting the signal strength of those cytokines which are already expressed by Schwann cells, thus attracting additional macrophages and other leukocytes [48]. After successful phagocytosis of the myelin and axonal debris, the pro-inflammatory molecular response has to be terminated, since the continuance of the inflammatory response could otherwise lead to adverse effects such as neuropathic pain [49].

6 Introduction

1.3.2 Charcot-Marie-Tooth Neuropathy Charcot-Marie-Tooth disease (CMT), also termed hereditary motor and sensory neuropathy (HMSN), is the most prevalent hereditary neuropathy, with a prevalence of approximately one affected individual in 2500 [50]. CMT was first described by Jean-Martin Charcot and Pierre Marie in France and independently by the British neurologist Howard Henry Tooth in 1868 [51, 52]. Clinical hallmarks of this disorder include distal muscle weakness and atrophy predominantly in the lower extremities, diminished or absent deep tendon reflexes, distal sensory loss, and skeletal deformations such as pes cavus [53]. Historically, two classes of CMT have been differentiated. The demyelinating forms of Charcot-Marie-Tooth disease, CMT1, exhibit a decreased motor nerve conduction velocity (MCV) (< 38 m/s) whereas the MCV in axonal CMT2 forms is preserved [54, 55]. The main histological feature of CMT1 is de- and remyelination, while nerve biopsies of CMT2 patients show axonal degeneration and regenerative sprouting [56]. Some forms of CMT do not easy fit into one of the above categories. The MCV in this third group, termed intermediate CMT, lies between 25 and 45 m/s. Each category is further subdivided by the recessive, dominant or X-linked inheritance pattern [57]. Currently, there are more than 60 CMT-associated genes described. These genes encode proteins required for a variety of cellular functions, such as transcription factors, myelin proteins, factors required for organelle and vesicle morphology and transport, protein degradation and mRNA processing [58-60]. A summary of CMT- associated genes, the cellular localisation of their encoded proteins, as well as their inheritance pattern, can be found in Figure 3.

Figure 3) Known disease genes in Charcot-Marie-Tooth disease

7 Introduction

1.3.2.1 Brief look at Animal Models of CMT A variety of animal models have been produced in order to investigate the pathogenic roles of mutant forms of CMT-associated genes. These models showed that there are many different pathomechanisms leading to CMT. In 1991 it was discovered that a duplication in the short arm of 17, which contains the peripheral myelin protein 22 (PMP22) gene, is the most common cause of CMT [61]. It has been shown that the increased gene dosage is causative for CMT1A [62, 63]. Histopathological features of peripheral nerves in patients with CMT1A are demyelination, onion bulb formation, and secondary axonal loss [64]. Several Pmp22 transgenic mouse lines and one transgenic rat line have been generated by integrating extra copies of Pmp22 into the genome. CMT1A rats carry approximately three copies of the gene. Those rats develop hypomyelination and onion-bulbs, as well as axonal loss, rendering them an adequate animal model for CMT1A [65]. Two mouse lines, harbouring 7 or 16 additional copies of Pmp22 displayed severe developmental histological and behavioural phenotypes and a reduced life span. However, reduced life span is not a feature of CMT patients making these lines unfit for potential pre-clinical trials [62, 66]. In contrast, two other transgenic mouse lines, harbouring approximately 4 copies of Pmp22 develop normal myelin and show consecutive demyelination and axonal loss, thus making them appropriate mouse models to study CMT1A [66, 67]. Several mutations in the gene encoding P0, the most abundant myelin protein of the peripheral nervous system, have been described to lead to CMT1B [68]. Mouse models with a heterozygous null allele show a relatively mild phenotype, while complete P0-knockout mice exhibit a progressive behavioural phenotype and fail to establish compact myelin in a large proportion of nerve fibres [69]. However, human cases with a deletion of an entire P0 allele are so far not known, making these models well suited for basic research but less ideal for the study of CMT1B. One model, harbouring a S63del mutation, which leads to the retention of the protein in the endoplasmic reticulum, comes close to the human CMT1B, exhibiting distally pronounced demyelination, reduced nerve conduction velocity and signs of muscle atrophy [70]. In cases where the clinical phenotype in patients harbouring point mutations or a deleted gene appears relatively uniform, it is assumed that the disease mechanism is a loss of function [71]. CMT1X, caused by point mutations in connexin-32 (Cx32) or a deletion of the gene, is one such example [72]. Mice lacking both Cx32 alleles as well as mice transgenically expressing the R142W mutation develop a late onset demyelinating neuropathy, thereby phenocopying human CMT1X patients [73, 74]. Reconstitution of the protein in Cx32 null mice, by transgenic expression under a Schwann cell-specific P0 promotor rescued the demyelinating phenotype, suggesting that the Schwann cell-specific loss of functional Cx32 causes the demyelination in CMT1X [75].

8 Introduction

In many CMT models, immune cells were observed in demyelinating nerves. In mice lacking one P0 allele, the number of both T-lymphocytes and macrophages are increased [76]. Also in connexin-32- defcient mice, T-lymphocyte and macrophage numbers are significantly elevated in peripheral nerves [77]. Since both Cx32- and P0-deficiency lead to similar inflammatory processes, it was proposed that immune-mediated demyelination may be a feature common to many CMT-like neuropathies independent of their genetic origin [77].

1.3.3 Inflammatory Neuropathies Guillain-Barré Syndrome (GBS) is an acute inflammatory disorder of the peripheral nervous system. GBS manifests itself with numbness, weakness, and often paralysis of legs, arms, breathing muscles and facial muscles [78]. Due to the potential effects on the breathing muscles, about 25 % of patients develop a respiratory insufficiency and need to be artificially ventilated [79]. The reported mortality rate of patients suffering from GBS has varied widely in the past with rates between 1 and 18 % [80]. However, two more recent studies described mortality rates around 3 %, mainly due to respiratory complications [78, 81]. The most common underlying subtype of GBS, making up 60 to 80 % of the cases, is the acute inflammatory demyelinating polyneuropathy (AIDP) [82]. Another subtype, in which the neurological deficit is purely affecting motor axons, was named acute motor axonal neuropathy (AMAN) [83]. When sensory fibres are also affected, this axonal subtype is called acute motor and sensory axonal neuropathy (AMSAN) [84]. A less common subtype is Miller Fisher syndrome (MFS), which is characterized by ophtalmoplegia, ataxia and areflexia [85]. The clinical course, the severity and the outcome of Guillain-Barré Syndrome are highly variable. GBS is a monophasic disease, usually reaching maximum severity within 4 weeks after onset [86]. The progressive phase is usually followed by a plateau phase ranging from 2 days to 6 month, after which the patients start to recover [79]. The pathology of the different subtypes of GBS varies, depending on whether it is a demyelinating or an axonal form. During the acute inflammatory demyelinating polyneuropathy, macrophages invade the peripheral nervous system and attack intact myelin, resulting in demyelination. One hypothesis states that activated T-cells target macrophages to antigens such as P0 and PMP22 on the surface of the Schwann cells and their myelin sheath [87, 88]. According to an alternative, the complement system is activated by bound to epitopes on the outer surface of the Schwann cells, so that the resulting complement activation initiates the vesiculation of myelin [89]. In severe cases the axons can be damaged, probably as a secondary consequence of the toxic and radicals released by the immune-mediated inflammatory response directed against myelin [86]. The pathological process during the axonal forms of GBS, AMAN and AMSAN differs from that of AIDP. Probably targeted by the Fc-receptor-mediated binding of antibodies against ganglioside antigens, macrophages invade the nodes of Ranvier leaving the myelin sheath initially intact. This complement-

9 Introduction mediated attack can lead to the detachment of paranodal myelin and nerve conduction failure. Upon breakdown of the paranodal structures, macrophages invade from the nodes into the periaxonal space, scavenging the injured axons [86]. Chronic inflammatory demyelinating polyneuropathy (CIPD) is the chronic form of the Guillain-Barré Syndrome (GBS). CIPD typically presents as either a relapsing or progressive neuropathy, with proximal and distal weakness that develops over a period of at least 8 weeks [90]. Patchy regions of demyelination and oedema with variable inflammatory infiltrates occur everywhere in the PNS. Onion bulb formation and myelin abnormalities associated with macrophages belong to the morphological features of CIDP [91].

1.4 The Dynamin Superfamily

Members of the dynamin superfamily are distinguished from other GTPases by their common structure including a large GTPase domain (~300 AA), a middle domain, and a GTPase effector domain (GED). The latter two are involved in oligomerization and regulation of the GTPase activity. The dynamin superfamily can be divided into three sub-families; the classical dynamins, the dynamin-related proteins and the guanylate-binding protein (GBP)-related proteins [92].

The classical dynamins include the three mammalian isoforms dynamin 1 (Dnm1), dynamin 2 (Dnm2) and dynamin 3 (Dnm3). Theses dynamins have a highly conserved structure, consisting of an N-terminal GTPase domain (G-domain), a middle domain (MD), a pleckstrin (PH) domain, a GTPase effector domain (GED) and a /arginine-rich domain (PRD). Mammalian Dnm1 is highly enriched in the brain, where it is concentrated in the presynapses. Dnm2 is ubiquitously expressed, while Dnm3 is found predominantly in testis and in the brain, where it is located postsynaptically, as well as at lower levels in the lungs and the heart [92-95]. In invertebrates, such as Drosophila melanogaster and only one isoform of the classical dynamins is found, which presumably covers all the functions of the three mammalian isoforms [96, 97]. The triplication of the dynamin gene during evolution may be partially explained by a need to fine-tune overall dynamin levels in specific tissues [98]. On the other hand, different dynamin isoforms also engage in different molecular interactions. As an example, Liu and colleagues showed that differences in the lipid-binding characteristics of Dnm1 and Dnm2 cause differences in their membrane-fission activity [99]. However, most of the differences between the isoforms are quantitative rather than qualitative. They have different affinities for SRC homology 3 domain (SH3 domain)-containing proteins, and their rates of GTPase activity, their oligomerization efficiency, and their lipid-binding properties vary [100, 99].

10 Introduction

1.5 Dynamin 2

Like the other classical dynamins, dynamin 2 (Dnm2) comprises an N-terminal GTPase domain (G-domain), a middle domain (MD), a PH domain, a GTPase effector domain (GED) and a proline/arginine-rich domain (PRD) (Figure 4, A). The GTPase domain, extending to 300 amino acids (AA), contains the four GTP-binding motifs (G1-G4) that are required for guanine-nucleotide binding and hydrolysis. The catalytic activity of the GTPase can be stimulated by oligomerisation of multiple dynamins and is mediated by interactions of the GTPase domain with the MD and the GED [101]. The middle domain is crucial for tetramerisation and higher-order self-assembly. A dynamin tetramer is expected to exhibit a higher affinity for the target membrane than would a monomer or a dimer [102]. The pleckstrin homology domain of Dnm2 specifically interacts with phosphatidylinositol-4,5- bisphosphate (PI(4,5)-P2), which is enriched at the plasma membrane [103]. PH-domain mutants with impaired phosphatidylinositol binding capacity block clathrin-mediated endocytosis [104]. The PH domain of Dnm2 requires oligomerisation of the protein for high affinity phosphatidylinositol binding [105]. The GTPase effector domain is essential for stimulating the GTP hydrolysis, acting as an intra-molecular GTPase activating protein (GAP) when Dnm2 tetramers are assembled into higher-order structures [106]. The C-terminal proline/arginine-rich domain contains multiple PXXP amino acid motifs, which interact with Src-homology 3 (SH3) domain-containing proteins to localize dynamin at endocytic sites [107, 108]. The structural composition of Dnm2 allows it to generate mechanical force. Multiple dynamins work together to fulfil the task of membrane fission at the neck of a forming vesicle. It has been shown that a dynamin dimer is the basic assembly unit [109]. In a GTP dependent oligomerisation process, dynamin dimers are recruited to the bud neck where they form a right-handed dynamin helix [110]. The accumulation of a minimum of 26 dynamins, which make up one turn of the helix, can under certain conditions be sufficient for membrane scission [109]. Upon cooperative GTP hydrolysis, the helix rapidly expands and constricts, which leads to the fission of the bud neck and to the budding of the vesicle (Figure 4, B).

The DNM2 gene has two alternative splice sites in the MD coding region. Both of them exist as an “a” and a “b” variant. This results in four different splice forms, termed as “aa”, “ab”, “ba” and “bb”. Dnm2 “aa” and “ab”, differing by only 4 amino acids at residue 516-519 (GEIL), are associated with clathrin- coated vesicles at the plasma membrane and with the trans-Golgi network [111]. Dnm2s with a “b” variant of the first splice site are targeted to the trans-Golgi network [112].

11 Introduction

1.6 Cellular Functions of Dynamin 2

Dynamin 2 has a variety of different subcellular localisations and functions. Dnm2 contributes to multiple forms of endocytosis, such as clathrin-mediated endocytosis (CME), caveolae-dependent endocytosis, pinocytosis and phagocytosis. Its importance during CME is best understood [113, 114]. Dnm2 polymerizes around the neck of an endocytic bud, where a GTP-dependent conformational change leads to the fission of the membrane resulting in budding of the clathrin coated vesicle (CCV) [98]. Dynamin 2-dependent vesicle budding also occurs at other intracellular membrane structures. Dnm2 has been found at clathrin-coated buds of early and late endosomes [115, 116]. Furthermore, Dnm2 is targeted to the trans-Golgi network (TGN), where it is important for budding secretory vesicle [117].

Figure 4) Functions of dynamin 2 A) Dynamin 2 consists of 5 structural domains: GTPase domain (G-domain), middle domain (MD), pleckstrin homology domain (PH-domain), GTPase effector domain (GED) and proline/arginine rich domain (PRD). Mutations in the DNM2 gene may cause dominant-intermediate Charcot-Marie-Tooth disease type B (CMTDIB) or Charcot-Marie-Tooth diease type 2 (CMT2M). Corresponding mutations are indicated in blue (CMTDIB) or brown (CMT2M), respectively. Most of the CMT causing mutations are found in the PH-domain. The black numbers indicate the amino acid position. B) Dynamin 2 oligomerizes around the neck of a vesicle in a GTP-dependent fashion. The fast GTP hydrolysis leads to a conformational change which elongates the dynamin- spiral, leading to the fission of the vesicle. C) Dynamin 2 is involved in various cellular functions such as different forms of endocytosis, intracellular membrane trafficking, and cytoskeletal regulation. Dynamin 2 is depicted in purple. EE: early endosome, LE late endosome.

Another major property of dynamin 2 is its link to the . Historically, dynamin was first identified as a GTPase that co-purified with brain microtubules, and was suggested to mediate microtubule sliding in vivo [118]. Other studies suggest that Dnm2 may regulate the dynamic instability of microtubules, which is essential for organelle motility [119]. Moreover, Dnm2 co-localizes with γ-tubulin at the centrosome, suggesting a role in centrosome cohesion [120]. In addition to

12 Introduction

microtubules, Dnm2 has also been shown to interact with actin, more precisely with the actin-binding protein 1 (Abp1) and cortactin [121, 122]. Abp1 serves as a linker between the endosomal machinery and the actin network. The Dnm2-cortactin complex plays a role in actin filament remodelling [123]. This complex can assemble in response to growth factor stimulation and mediate the remodelling of actin in the leading edge of a migrating cell to facilitate lamellipodia protrusion [124] (Figure 4, C).

1.7 Dynamin 2 in Disease

Mutations in the DNM2 gene cause autosomal dominant centronuclear myopathy (CNM) or rare forms of Charcot-Marie-Tooth (CMT) disease [125, 126]. As the name indicates, the most predominant histopathological feature in CNM are centrally located nuclei in a large number of muscle fibres [127]. The Dnm2-related CNM is characterized by slowly progressive muscular weakness and wasting that usually begins during adolescence or early adulthood [128]. Other common clinical features of Dnm2-CNM include developmental delays, facial and generalized muscle weakness, ptosis, and ophthalmoparesis. Electrophysiological investigations frequently reveal signs of mild axonal peripheral nerve involvement, overlapping with observations made in patients suffering from DNM2-related Charcot-Marie-Tooth disease [129]. Other mutations in DNM2 can cause dominant intermediate Charcot-Marie-Tooth disease Type B (CMTDIB) or an axonal form of CMT (CMT2M). The motor nerve conduction velocities (MCVs) in Dnm2- related forms of CMT are variable, so that patients may be diagnosed with either intermediate CMT, with a MCV between 25 m/s and 38 m/s, or with axonal CMT with a MCV above 38 m/s. Some patients suffering from Dnm2-CMT show neutropenia in addition to distal muscle weakness and sensory impairment, which are common to all forms of CMT. Histopathological analysis of patients with CMTDIB show a loss of myelinated axons, rare segmental demyelination and remyelination with onion bulb formation, and focal hypermyelination [125]. The vast majority of CMT-associated DNM2 mutations lie within the pleckstrin homology (PH) domain (Figure 4, A). It has been shown, that PH-domain mutants with impaired phosphoinositide-binding capacity block clathrin-mediated endocytosis [104]. Furthermore, a more recent study by Sidiropoulos and colleagues demonstrated that CMT- but not CNM-causing mutations in Dnm2 have defects in clathrin-mediated endocytosis [114]. This suggests the possibility that altered endocytosis may be a major factor contributing to the disease mechanism in dynamin 2-related Charcot-Marie-Tooth diseases.

13 Introduction

1.8 The Heterotetrameric Adaptor Complex Family

Up to now, the heterotetrameric adaptor complex family consists of five adaptor protein (AP) complexes; AP-1, AP-2, AP-3, AP-4 and AP-5. Each complex consist of two large subunits (~ 100 kDa, β1- β5 and γ, α, δ, ε, ζ, respectively), one medium-sized subunit (~ 50 kDa, µ1-µ5) and one small subunit (~ 17 kDa, σ1-σ5) (Figure 5, A). The large subunits can be divided into three structural domains, namely a core-domain that is linked via a hinge region to an ear-domain [130]. The hinge domains of AP-1, AP-2 and AP-3 have a clathrin-binding motif, which is responsible for binding to the clathrin heavy chain. To fulfil their function, AP-1 and AP-2 depend on clathrin to build clathrin-coated vesicles, whereas AP-3, AP-4 and AP-5 are able to mediate clathrin-independent vesicle formation [131].

Figure 5) The heterotetrameric adaptor protein complex family A) Schematic drawing of the five different adaptor protein (AP) complexes. All complexes consist of two large subunits (β1-β5 and γ, α, δ, ε, ζ, respectively), a medium-sized subunit (µ1-µ5) and a small subunit (σ1-σ5). B) The different AP complexes function at different sites within a cell. AP-1 (red) is localized to the endosomes (E) and the trans-Golgi network (TGN), AP-2 (blue) to the plasma membrane (PM), AP-3 (green) to the endosomes, AP-4 (yellow) to the trans-Golgi network and AP-5 to the late endosomes / lysosomes (LE/L).

Each of the AP complexes is localized to a certain compartment within the cell. AP-1 is localized to endosomes and the trans-Golgi network (TGN), where it is involved in trafficking between the two organelles [132]. AP-2, the best characterized member of the family, is involved in clathrin-mediated endocytosis at the plasma membrane [133]. AP-3 has been shown to be involved in the transport of lysosomal membrane proteins form endosomes to lysosomes [134]. AP-4 has been shown to mediate sorting from the TGN towards early endosomes as well as to the plasma membrane [135, 136]. AP-5, the most recently discovered AP complex, was shown to localize to late endosomes [131]. A summary of the different localisations can found in (Figure 5, B). One important factor leading to the different subcellular localization is the binding capacity of the individual AP complexes for different phosphatidylinositols, which themselves are associated with different pools of membranes [137]. While AP-2 is able to directly bind to phosphatidylinositol-4,5-

14 Introduction

bisphosphate (PI(4,5)P2), AP-1 requires the interaction with an additional player, ADP-ribosylation factor 1 (ARF1), for binding to phosphatidylinositol-4-phosphate (PI4P)-rich membranes [138, 139]. Also AP-3 and AP-4 depend on activated ARF1 for membrane recruitment [140, 141]. Even so, an interaction between AP-3, AP-4, and AP-5 and a distinct phosphatidylinositol has not yet been formally demonstrated; co-localisation studies suggest that AP-3 interacts with phosphatidylinositol-3-phosphate

(PI3P), AP-4 with PI4P, and AP-5 with Phosphatidylinositol-3,5-bisphosphate (PI(3,5)P2) [131, 136, 137, 140].

1.9 The Adaptor Protein Complex 2

Nearly 40 years ago Keen and colleagues isolated the newly discovered clathrin-coated vesicles and showed that under physiological condition clathrin requires 100 kDa proteins to form clathrin baskets in vitro [142]. The 100 kDa proteins, turned out to be the two large subunits of the adaptor protein complex 2 (AP-2), which was originally termed assembly protein 2 [143]. As with the other AP complexes, AP-2 consists of 4 subunits. One of the two large (~ 100 kDa) subunits, the alpha subunit, mediates the binding to the plasma membrane by interacting with phosphatidylinositol-4,5-bisphosphate (PI(4,5)P2) [144, 139]. The ear domain of the α-subunit serves as a hub, linking various interaction partners such as epsin1, eps15 (Epidermal growth factor receptor substrate 15), AP180 and amphiphysin to the endocytic machinery [145, 146]. Most of these interactions are stabilized by the additional binding to the ear domain of the second large subunit (β) [147]. The β-subunit recruits clathrin through a clathrin-binding sequence (LΦXΦ[D/E], where Φ is a large hydrophobic amino acid), termed the clathrin box, lying within the hinge region of the protein [148]. Additionally, the core domain of the β-subunit can bind to transmembrane cargo proteins by interacting with dileucine-based motifs ([DE]XXXL[LI]) [149]. The other subunit of the complex directly interacting with cargo molecules at the plasma membrane is the middle-sized (~ 50 kDa) µ-subunit (formerly termed AP50). This subunit interacts with either the tyrosine-based YXXΦ or the dileuicine-based DXXLL sorting signal [149]. The fourth member of the complex, the small (~ 17 kDa) σ-subunit, is responsible for stabilizing the core of AP-2 [150]. AP-2 can be in a locked or an open conformation. The locked, inactive form of AP-2 is the predominant one in the absence of membrane interaction. The first step in activating AP-2 is its recruitment onto the plasma membrane by binding of the α- and β-subunits to PI(4,5)P2 [151]. Once attached to the membrane, the electrostatic attraction and subsequent binding of the µ-subunit to additional PI(4,5)P2 directs the complex into the open conformation. This open conformation is further stabilized by of the Thr156 residue of the µ-subunit by the α-appendage binding kinase (AAK1), leading to a tighter binding to the plasma membrane [152]. The stabilized open conformation of AP-2 allows the µ- and the β-subunit to bind to any YXXΦ or [ED]XXXL[LI] motifs in their vicinity. Because AP-2

15 Introduction is attached to the plasma membrane, potential binding partners will be transmembrane protein cargo [133] (Figure 6, activation of AP-2). From there on, clathrin-mediated endocytosis takes its course.

1.10 Clathrin-Mediated Endocytosis

Clathrin-mediated endocytosis (CME) is critical for the uptake of nutrients, membrane recycling and intercellular signaling. CME is used by all known eukaryotic cells, highlighting its importance [153]. During this process, clathrin-coated vesicles (CCV) bud off from the plasma membrane and are taken up into the cell. Even though the budding of CCVs in vitro requires only the presence of clathrin, AP-2, an additional adaptor protein and dynamin 2, CME in vivo is much more complex [154].

Figure 6) Clathrin-mediated endocytosis The adaptor protein complex 2 (AP-2) binds to PI(4,5)P2 at the plasma membrane. This opens its comformation, which enables the binding of cargo. This open conformation is further stabilized by phosphorylation of the µ-subunit. Cargo-bound AP-2, together with clathrin and clathrin-associated proteins (CLASPs), start invaginating the plasama membrane to form a clathrin- coated vesicle. Dynamin 2 is recruited to the neck of the forming vesicle. Upon GTP hydrolysis, the dynamin spiral elongates, which leads to the fission of the membrane. In the cytosol, the vesicle is uncoated with the help of auxilin and Hsc70.

In the first step, the adaptor protein complex 2 (AP-2) is recruited to the plasma membrane by binding to phosphatidylinositol-4,5-bisphosphate (PI(4,5)P2). Upon binding, AP-2 undergoes a conformational change from a locked into an open conformation, which is additionally stabilized by phosphorylation of the µ-subunit, allowing the complex to bind to YXXΦ or [ED]XXXL[LI] motifs present in transmembrane cargo proteins [151, 133]. A more detailed description of this initial step can be found in section 1.9. There are also CME-targets that are not directly bound by AP-2. Those targets bind to adaptor proteins, which in turn always bind to AP-2; for example, Stonin 2 recruits synaptotagmin, Dishevelled binds to

16 Introduction

Frizzled, and Epsin 1 interacts with ubiquitinated epidermal growth factor receptors [155-157]. After a cargo is selected and bound by AP-2 or by another cargo-specific adaptor protein, clathrin is recruited and bound to the clathrin box, lying within the hinge region of the β-subunit of the AP-2 complex. While multiple clathrin triskelia start polymerizing to form a clathrin coat, additional clathrin-associated sorting proteins (CLASPs) are recruited and bound to the ear domains of the α- and β- subunits of AP-2, thereby stabilizing the forming vesicle [153]. The dynamic polymerization of actin filaments creates a pulling force and helps invaginate the membrane [158]. Dynamin 2 (Dnm2) polymerizes around the neck of a forming CCV. A GTP-dependent conformational change then leads to the fission of the membrane, resulting in budding of the vesicle [98]. A more detailed description of the scission-step can be found in section 1.5. Once detached from the plasma membrane, the CVV is uncoated by the ATPase heat shock cognate 70 (Hsc70) and its cofactor, auxilin. After vesicle budding, auxilin is recruited to the terminal domains of clathrin triskelia. These are easily accessible at the zone where the neck was attached, since there the clathrin cage is not completely closed [159]. When bound to clathrin, auxilin recruits Hsc70, which initiates the ATP-dependent disassembly of the clathrin coat [160] (Figure 6).

1.11 Iron

Iron is an essential element for various synthetic and enzymatic processes within a eukaryotic cell, including oxygen transport, energy metabolism and DNA synthesis. However, iron is potentially toxic, due to its ability to donate and accept electrons. In our body, iron exists in a reduced ferrous (Fe2+) and an oxidized ferric (Fe3+) form. Ferrous iron can catalyse the conversion of hydrogen peroxide, produced by mitochondria and from other sources, into free radicals which cause damage to many cellular structures and molecules [161]. Due to limited solubility, the non-toxic ferric iron is poorly bioavailable. Thus, the acquisition and storage of iron and simultaneously minimizing the risk of toxicity poses a major challenge to cells and organisms. The majority of iron in our body (~ 1.8 g) is bound to haemoglobin in erythrocytes, where it is crucial for oxygen transport. Significant amounts of iron are also fund in macrophages (~ 0.6 g) and in the myoglobin of muscles (~ 0.3 g). Excess body iron is stored in the liver (~ 1 g) [162]. Other tissues contain lower, but not less important levels of iron. A healthy iron balance is maintained by the tight control of dietary iron absorption in the duodenum. In the intestinal lumen, nutritional iron (Fe3+) is reduced by ferric reductases, such as duodenal cytochrome b (Dcytb), and subsequently transported across the apical membrane of enterocytes by the divalent metal transporter 1 (DMT1) [163]. Enterocytes either temporarily store the iron or transport it to their basolateral membrane where Fe2+ is eventually exported into the bloodstream by ferroportin. This export is coupled to a reoxidation to Fe3+ by the membrane-bound ferroxidase hephaestin [164]. The exported iron is scavenged by transferrin (Tf), which maintains Fe3+ in a redox-inert state and delivers it to the different tissues [161].

17 Introduction

Most cell types acquire iron by taking up iron-loaded holo-transferrin present in the plasma. Holo-Tf has a high affinity for transferrin receptor 1 (TfR), which is present on the surface of the cell [165]. Once bound to Tf, the TfR undergoes clathrin-mediated endocytosis. Inside the cell, the vesicle is acidified by an ATP-dependent proton pump that lowers the luminal pH to ~5.5. The acidification leads to a conformational change in transferrin and TfR which in turn leads to the release of the iron [166, 167]. STEA3, a ferrireductase, converts the endosomal Fe3+ to Fe2+, which is then transported to the cytosol by the DMT1 [168]. If the cytosolic iron is not immediately used, it is stored in ferritin. Ferritin is a multimer of 24 subunits comprised of heavy (H) or light (L) chain subunits. The subunits form a sphere with a central core that can contain up to 4500 atoms of iron [169]. Depending on the cell type, a ferritin mutlimer can have a different compositon of subunits [170]. The acidified vesicles, in which iron-free apo-Tf remains bound to TfR, are shuttled back to the cell surface. After fusion with the plasma membrane, the more neutral pH leads to the dissociation of apo-Tf from TfR [171]. Apo-Tf is now ready to bind new iron and initiate further rounds of transferrin receptor-mediated endocytosis (Figure 7, A).

Figure 7) Cellular iron regulation A) Iron enters the cell coupled to transferrin (HOLO-Tf) which binds to the transferrin receptor (TfR) and is taken up via clathrin- mediated endocytosis. After acidification of the vesicle by a proton pump, the iron cations are transported into the cytosol by the divalent cation transporter 1 (DCT1) where they are either directly used or stored in ferritin. The iron-free transferrin (APO- Tf) is recycled to the plasma membrane, where it disassociates from the TfR and is ready for another cycle of iron import. B) During a state of iron deficiency, iron regulatory proteins (IRPs) bind on the one hand to the iron response elements (IREs) on the mRNA of the transferrin receptor (TfR), protecting it from degradation, and on the other hand to the IREs on the mRNA of ferritin, preventing its translation. In a state of iron overload, the abundant iron binds to IRPs, which inhibits the binding to IREs. Consequently, TfR mRNA is degraded and ferritin is synthesized.

Cellular iron homeostasis is transcriptionally regulated (Figure 7, B). Iron overload or depletion leads to compensating changes in the iron regulatory element (IRE) system. When iron is in excess, iron responsive proteins (IRPs) are in their inactive form and do not bind to IREs on the mRNAs of TfR and ferritin. Consequently, the mRNA of TfR is degraded by nucleases and the ferritin mRNA is translated. On

18 Introduction

the contrary, when a cell suffers from iron deficiency, IRPs bind the IREs on the 3’-untranslated region (UTR) of the TfR mRNA, protecting it from degradation and subsequently leading to an increased translation. Active IRPs also bind the 5’-UTR of the ferritin mRNA, preventing initiation of its translation [172]. The importance of adequate amounts of iron is well established, since a shift in this delicate equilibrium leads to a variety of disorders. It is well established that iron deficiency caused by malnutrition or chronic blood loss (e.g. menorrhagia) can cause anemia [173]. In severe cases, where an iron deficiency anemia (IDA) has led to a peripheral neuropathy, iron supplementation has been shown to be beneficial for the recovery of the patients, including their neuropathy [174]. Also in other diseases with iron deficiency, such as Willis–Ekbom disease, oral supplementation of iron has shown to be beneficial [175]. Not only iron deficiency, but also iron overload has negative effects on the human body. As an example, excess iron has been described as a common hallmark in many neurodegenerative disorders, including Parkinson’s and Alzheimer’s disease [176]. However, it is still unclear whether this is causative or a consequence of the degenerative processes.

1.12 Objective of the Study

Multiple mutations in dynamin 2 (DNM2) were found to cause Charcot-Marie-Tooth disease. Despite the ubiquitous expression of Dnm2, these disease mutations only seem to affect the peripheral nervous system. Up to now only little is known about the functional role of Dnm2 in the disease-relevant cell types. However, in order to eventually understand the pathomechanism underlying a disease, one prior has to gain knowledge about the physiological role of the involved proteins in the relevant cell types. Hence, we wanted to understand the physiological role of Dnm2 for Schwann cell maintenance. A previous study proposed that impaired clathrin-mediated endocytosis is a major contributing factor to the disease mechanisms in dominant intermediate Charcot-Marie-Tooth neuropathy type B, which is caused by a set of mutations in DNM2 [114]. Dnm2 has been described to have multiple functions, ranging from different forms of endocytosis to cytoskeletal regulations. In order to be able to appreciate the importance of Dnm2 for CME, one first has to understand the importance of CME in adult myelinating Schwann cells. To that end, we ablated the µ-subunit of the adaptor protein complex 2, which is absolutely required and exclusively involved in CME, specifically in adult myelinating Schwann cells. Having both mouse models available, allows us an extensive comparison in order to get a better understanding of the physiological role of the CMT-associated protein dynamin 2 as well as the overall importance of clathrin-mediated endocytosis for adult myelinating Schwann cells.

19 Introduction

20 Results

2 Results

2.1 SC-Specific Ablation of Dynamin 2 in Adult Mice Results in a Remitting Neuropathy

In order to investigate the role of dynamin 2 in myelin maintenance, we generated mice in which conditional Dnm2 ablation in Schwann cells could be induced upon tamoxifen administration. To this end, Dnm2fl/fl mice were bred with P0CreERT2+ mice. In order to investigate the fate of individual Schwann cells, for a subset of experiments the mice also carried a heterozygous ROSAYFP+/- Cre- reporter, which upon recombination leads to the expression of cytosolic YFP [177]. Dnm2 was ablated by injecting 10 week-old Dnm2fl/fl P0CreERT2+ (ROSAYFP+/-) mice with tamoxifen on five consecutive days. Tamoxifen-injected Dnm2fl/fl P0CreERT2+ mice will hereafter be referred to as Dnm2iko and tamoxifen-injected Dnm2fl/fl P0CreERT2- mice as CtrDnm2, whereas tamoxifen-injected Dnm2fl/fl P0CreERT2+ ROSAYFP+/- mice will be referred to as Dnm2iko(R) and tamoxifen-injected Dnm2wt/wt P0CreERT+ ROSAYFP+/- mice as CtrDnm2(R) (Figure 8, A). The loss of Dnm2 could be detected by Western blot analysis in sciatic nerve extracts 4 weeks post-tamoxifen (wpT) (Figure 8, B and C). Between 4 and 5 wpT, all mice that had lost dynamin 2 started to show signs of a peripheral neuropathy. We used an established scoring scheme to assess the development of their clinical performance [178- 180]. At the onset of the clinical phenotype, the Dnm2iko mice showed an impairment or absence of righting movements and ataxic gait. From there on, the phenotype worsened and developed into a moderate paraparesis around 6 wpT (Figure 8 E). Surprisingly, all Dnm2iko mice thereafter improved rapidly and reached a normal level of performance between 7 and 8 wpT (Figure 8 D). The recovered mice were indistinguishable from control mice and stayed healthy until 41 wpT, the last time point of analysis (Figure 8 F). Additional motor skill assessments can be found in the work of Dr. Christian Somandin, which precedes this study [181]. In order to present the complete picture, some of his findings are adapted, partially supplemented and incorporated into the results. These figure panels are marked with a “CHS”. To further characterize the clinical phenotype and link it to a predicted neuropathy, we performed electrophysiological measurements during the peak of impairment and after recovery. The examination of sciatic nerves of Dnm2iko at 6 wpT revealed a significant decrease in motor nerve conduction velocity (MCV). While applying a proximal stimulus, only a weak and disperse signal could be recorded. Furthermore, F-wave latencies were not assignable in mutant mice. F-waves are an electrophysiological artefact, produced by antidromic activation of motoneurons following distal electrical stimulation of motor nerve fibres. At 14 wpT the MCV was restored. Also at a late time point, 41 wpT, the MCV of Dnm2iko mice was undistinguishable from that of CtrDnm2 mice (Figure 8, G-I). Taken together, these results show that the loss of Dynamin 2 in myelinated Schwann cells leads to a rapid but transient loss of peripheral nerve function in vivo.

21 Results

Figure 8) Schwann cell-specific ablation of dynamin 2 in adult mice result in a remitting neuropathy A) Schematic drawing of the inducible-, Schwann cell-specific dynamin 2 (Dnm2) ablation and simultaneous YFP expression. Exon 2 of the Dnm2 allele as well as a transcription stop signal in the RosaYFP-reporter allele are flanked by loxP sites. Mice carrying the floxed alleles were crossed with mice carrying the P0CreERT2 allele, leading to P0CreERT2 Dnm2fl/fl ROSAYFP+/- mice. Cre recombinase is specifically expressed in Schwann cells under the control of the Schwan cell-specific P0 promotor, but is trapped in the by the ERT2 element until tamoxifen is administered. Cre is then shuttled into the nucleus where it leads to the excision of the floxed regions. B) Western blot analysis of Dnm2 protein levels in CtrDnm2(R) and Dnm2iko(R) sciatic nerve lysates at 4 wpT showed reduced protein levels in knockout mice. C) Quantification of Dnm2 protein levels (represented in C) (n=3 animals each genotype, two-tailed unpaired Student´s t-test). D) CtrDnm2 and Dnm2iko tamoxifen-injected mice were phenotypically assessed over a period of 14 weeks (n=5 mice each genotype). Clinical score: 0 = normal, 1 = less lively, 2 = impaired righting, 3 = absent righting, 4 = ataxic gait, 5 = mild paraparesis, 6 = moderate paraparesis, 7 = severe paraparesis, 8 = tetraparesis, 9 = moribund, 10 = death. Six weeks after tamoxifen treatment, all Dnm2iko mice exhibited a transient moderate paraparesis. CHS E) Representative photograph of a Dnm2iko and a CtrDnm2 mouse at 6 wpT. CHS F) Schematic drawing of the experimental timeline. 10 week-old Dnm2iko and CtrDnm2 mice were injected intraperitoneally with tamoxifen on 5 consecutive days. The mice were analysed at 4 weeks post-tamoxifen (wpT), 5 wpT, 6 wpT, 8 wpT, 14 wpT and 41 wpT. G) Schematic drawing of the setup for electrophysiological measurements. H) Representative electrophysiological profile of Dnm2iko and CtrDnm2 mice at 6 wpT and 14 wpT. At 6 wpT, only weak and disperse electric signals were assessed upon proximal stimulation in Dnm2iko mice. This deficit is restored at 14 wpT. CHS I) At 6 wpT Dnm2iko show decreased motor nerve conduction velocity (MCV). This deficit is restored at 14 wpT and stays unimpaired until 41 wpT. (n=6 animals each group, two- way ANOVA with Sidak´s multiple comparisons test) CHS| Mean ± SEM, *P < 0.05, ***P < 0.001

2.1.1 Acute De- and Remyelination upon Dynamin 2 Ablation Histological analysis of sciatic nerves revealed a severe demyelination followed by remyelination in Dnm2iko mice (Figure 9, A). Prior to the clinical onset at 4 wpT no morphological difference could be observed between nerves of Dnm2iko and CtrDnm2 mice.

22 Results

Figure 9) De- and remyelination upon dynamin 2 ablation A) Ultrathin sections of sciatic nerves from CtrDnm2 and Dnm2iko mice at 4 weeks post-tamoxifen (wpT), 5 wpT, 6 wpT, 8 wpT, 14 wpT and 41 wpT. White arrow: organelle accumulation in Schwann cells, black arrows: non-myelinated axons, white arrowheads: remyelinating Schwann cells, black arrowheads: macrophages. Scale bar: 10 µm. CHS B) Unmyelinated axons were counted on sciatic nerve ultrathin sections. There was a transient peak of unmyelinated axons around 6 wpT (two-way ANOVA with Bonferroni test). CHS C) G-ratio analysis of CtrDnm2 and Dnm2iko sciatic nerves revealed a slight hypomyelination at 14 wpT (n=3 animals each genotype, two-tailed unpaired Student´s t-test). D) Scatter plot of g-ratios of CtrDnm2 and Dnm2iko sciatic nerves at 14 wpT. (n=3 animals each genotype) CHS E) G-ratio analysis of CtrDnm2 and Dnm2iko sciatic nerves revealed a slight hypomyelination at 41 wpT (n=3 animals each genotype, two-tailed unpaired Student´s t-test). F) Scatter plot of g-ratios of CtrDnm2 and Dnm2iko sciatic nerves at 41 wpT. (n=3 animals each genotype) G) Teased osmicated sciatic nerve fibers of CtrDnm2 and Dnm2iko mice at 14 wpT. At 14 wpT, Dnm2iko mice have shorter internodes. Black arrowheads: node of Ranvier. Scale bar: 100 µm H) Internodal length in Dnm2iko mice is reduced compared to that in control mice (represented in G) (n=3 animals each genotype, 100 internodes per animal were measured, two-tailed unpaired Student´s t-test). I) Ultrathin sections showing whole (left) and magnified (right) dorsal and ventral roots of CtrDnm2 and Dnm2iko at 6 wpT. Both dorsal and ventral roots of Dnm2iko show signs of demyelination. Black arrows: non-myelinated axons, white arrowheads: remyelinating Schwann cells, black arrowheads: macrophages. Scale bar: 100 µm (left), 10 µm (right). CHS J) Hematoxylin/eosin (H&E), succinate-dehydrogenase (SDH) and ATPase (pH 4.6) staining of cross-sections of gastrocnemius muscles from CtrDnm2 and Dnm2iko mice at 14 wpT show no apparent difference. Scale bar: 100 µm | Mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001

23 Results

At 5 wpT the first Schwann cells in Dnm2iko mice showed indications of stress, such as swollen cytoplasm and an accumulation of organelles. At the peak of the clinical phenotype, at 6 wpT, many non- myelinated and rarely already remyelinated axons, as well as immune cells such as macrophages could be observed in the sciatic nerves of Dnm2iko mice. Within the next two weeks, remyelination started successfully on all the axons, so that besides some still present immune cells, hypomyelination was the main remaining feature (Figure 9, A). During the peak of the clinical phenotype at 6 wpT, which is in line with the peak of the transient demyelination, approximately 15 % of all the axons were unmyelinated (Figure 9, B). Even so remyelination successfully took place; at 14 wpT we could still observe a slight hypomyelination, which is persisted until 41 wpT, the last time point of analysis (Figure 9, C-F). After remyelination, myelin was not only thinner, but also the internodes were shorter as measured on teased osmicated fibres at 14 wpT (Figure 9, G and H). To exclude the possibility that only a subset of Schwann cells, namely those myelinating sensory or motor neurons, were affected, we analysed the L4 dorsal and ventral roots. Both dorsal and ventral roots showed signs of de- and remyelination as well as immune cells at 6 wpT (Figure 9, I). During the whole phase of de- and remyelination, we could not observe detectable histological signs of axonal loss. To validate this impression, we analysed the gastrocnemius muscle of Dnm2iko and CtrDnm2 mice for signs of reinnervation, which would occur after axonal loss and lead to fibre type grouping [182]. At 14 wpT, gastrocnemius muscles of Dnm2iko mice did not show detectable abnormalities (Figure 9, J). Overall, we could observe spontaneous de- and remyelination in the sciatic nerve upon Schwann cell- specific dynamin 2 ablation.

2.1.2 Ablation of Dynamin 2 Causes Schwann Cell Dedifferentiation The process of demyelination and myelin breakdown usually requires the dedifferentiation of the Schwann cells. To test if this model also applies in our context, we performed qRT-PCR analysis of RNA isolated from sciatic nerves of Dnm2iko and CtrDnm2 mice at 4, 6, 8 and 14 wpT. Already at 4 wpT, mRNA levels for myelin proteins, such as P0, MBP and Periaxin were reduced. Conversely, markers of dedifferentiated Schwann cells, p75 and Krox24, together with the proliferation marker cyclin D1, were significantly upregulated (Figure 10, A). During the phase of subsequent redifferentiation myelin genes were re-expressed and genes for dedifferentiated Schwann cells were supressed, so that at 14 wpT Dnm2iko show an expression pattern similar to that in CtrDnm2 mice. We examined sciatic nerve cross sections of Dnm2iko and CtrDnm2 mice for the expression of p75. Immunostaining of the sciatic nerve showed a large increase in the number of p75-positive cells at 6 wpT, confirming that Schwann cells had dedifferentiated back to an immature Schwann cell-like state (Figure 10, B). Interestingly, we found both recombined (YFP-positive) and non-recombined (YFP-negative) cells stained positive for p75, suggesting

24 Results

that also non-recombined cells dedifferentiated (Figure 10, C). In line with the observed dedifferentiation, the protein levels of the potent myelination inhibitor c-Jun in sciatic nerve lysates of Dnm2iko mice were highly upregulated from 5 wpT on (Figure 10, D).

Figure 10) Schwann cells dedifferentiate upon dynamin 2 ablation A) qRT-PCR analysis of Schwann cell differentiation and dedifferentiation markers in Dnm2iko sciatic nerves at 4 weeks post- tamoxifen (wpT), 6 wpT, 8 wpT and 14 wpT. The dashed line represents the means of the values for the corresponding CtrDnm2 mice. (n=3 animals each genotype, one-way ANOVA with Fisher´s LSD test) B) Immunostaining of CtrDnm2(R) and Dnm2iko(R) sciatic nerves cross-sections for p75 at 4 wpT, 6 wpT and 14 wpT. There is a transient increase of p75 staining at 6 wpT in Dnm2iko(R) mice. Scale bar: 200 µm C) Immunostaining of Dnm2iko sciatic nerves cross-sections at 6 wpT. Recombined (YFP+) as well as non- recombined (YFP-) Schwann cells (Sox10+) express p75. White arrow: recombined Schwann cell (Sox10+, YFP+) expressing p75, white arrowhead: non-recombined Schwann cell (Sox10+, YFP-) expressing p75. Scale bar: 10 µm D) Western blot analysis of c-Jun in sciatic nerves of CtrDnm2 and Dnm2iko mice at 4 wpT, 5 wpT, 6 wpT, 8 wpT and 14 wpT. (n=3 animals each genotype, two-tailed unpaired Student´s t-test) CHS E) Western blot analysis of P-Erk1/2Thr202/Tyr204 in sciatic nerves of CtrDnm2 and Dnm2iko mice at 4 wpT, 5wpT, 6 wpT, 8 wpT and 14 wpT. (n=3 animals each genotype, two-tailed unpaired Student´s t-test) CHS F) Immunostaining of CtrDnm2(R) and Dnm2iko(R) sciatic nerves cross-sections for P-Erk1/2Thr202/Tyr204 and Sox10 at 4 wpT. White arrowheads: Schwann cells (Sox10+) without P-Erk1/2Thr202/Tyr204 expression, white arrows: Schwann cells (Sox10+) expressing P-Erk1/2Thr202/Tyr204. Scale bar: 25 µm. | Mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001

25 Results

Recent studies described the activation of Erk1/2 as a potential upstream activator of c-Jun [26]. Indeed, the levels of P-Erk1/2 were already significantly increased at 4 wpT and increased even more thereafter, before returning to normal levels when remyelination was in progress at 8 wpT (Figure 10, E). Since at 4 wpT there were no obvious immune cells present, we assumed the activation of P-Erk1/2 to be a Schwann cell-specific feature. An immunostaining for P-Erk1/2 at 4 wpT verified this hypothesis (Figure 10, F). Taken together, these results suggest that the ablation of dynamin 2 causes a P-Erk1/2-driven dedifferentiation of recombined and non-recombined Schwann cells prior to the drastic morphological changes.

2.1.3 Schwann Cells Coordinate an Acute Inflammation upon Dynamin 2 Ablation In addition to the morphological changes of Schwann cells during the process of demyelination, we observed various immune cells on ultrathin sections of sciatic nerves from Dnm2iko mice. The most obvious ones were the highly phagocytic macrophages containing vesicles filled with myelin debris at 6 wpT (Figure 11, A, left panel). We therefore quantified the number of macrophages on cross sections of Dnm2iko and CtrDnm2 mice at 4, 6, 14 and 41 wpT. At 4 wpT no significant increase in the number of macrophages was observed, whereas their number peaked at 6 wpT. Even though between 6 and 14 wpT the number of macrophages in nerves of Dnm2iko mice was strongly reduced, we still detected a significantly higher number than in CtrDnm2 mice, which persisted until the last time point of analysis, 41 wpT (Figure 11, A). Besides macrophages, we also found T-cells, neutrophils and mast cells on ultrathin sections of sciatic nerves from Dnm2iko mice (left panels; Figure 11, B, C, D). T-cells show pattern similar to that seen with macrophages, with an increase starting at 4 wpT, a massive peak at 6 wpT and a subsequent decrease to a level that stayed significantly elevated also at late time points (Figure 11, B). Neutrophils and mast cells, which overall were less abundant, showed a long lasting peak from 6 to 14 wpT in nerves of Dnm2iko mice. The numbers of both cell types were reduced at 41 wpT, while only the number of mast cells stayed significantly elevated (Figure 11, C and D). In order for immune cells to invade the sciatic nerve, they need to receive activation cues and overcome the blood nerve barrier (BNB). It has been shown that the activation of the Erk-signaling pathway in Schwann cells is sufficient to open the BNB [26]. Since P-Erk1/2 levels were elevated already at 4 wpT, we injected Dnm2iko and CtrDnm2 mice at this time point with Evans Blue, a tracer that passes from blood vessels into the endoneurium and perineurium following breakdown of the BNB. Indeed, sciatic nerve cross-sections of Dnm2iko mice at 4 wpT were labelled with Evans Blue (Figure 11, E and F).

26 Results

Figure 11) Schwann cells coordinate an acute inflammation upon dynamin 2 ablation A) Left panel: Electron micrographs showing a macrophage (MΦ) in the Dnm2iko mice at 6 weeks post-tamoxifen (wpT). Scale bar: 5 µm. Middle panel: Immunostaining of CtrDnm2(R) and Dnm2iko(R) sciatic nerves cross-sections for the macrophage marker CD68 at 4 wpT, 6 wpT, 14 wpT and 41 wpT. Scale bar: 50 µm. Right panel: Quantification of the immunostaining represented in the middle panel (n=3 animals each genotype, two-tailed unpaired Student´s t-test). B-D) Similar representation and quantification for B) T-cells (CD3), C) Neutrophils (GR1) and D) Mast cells (CD117). E) Evans blue dye was injected into CtrDnm2(R) and Dnm2iko(R) mice 4 wpT 30 min before sacrifice. Representative fluorescent images of sciatic nerve cross-sections show a strong increase in staining in nerves of Dnm2iko(R) mice. Scale bar: 100 µm F) Quantification of E) (n=3 animals each genotype, two-tailed unpaired Student´s t-test) G) qRT-PCR analysis of chemokines in Dnm2iko sciatic nerves at 4 wpT, 6 wpT, 8 wpT and 14 wpT. The dashed line represents the means of the values for the corresponding CtrDnm2 mice. Levels of all the cytokines were significantly increased at 6 wpT. Only the chemokine (C-C motif) ligand 2 (CCL2) was already strongly expressed at 4 wpT. (n=3 animals each genotype, one-way ANOVA with Fisher´s LSD test) H) ELISA (-linked immunosorbent assay) for CCL2 showing an increase in protein levels in sciatic nerves of Dnm2iko animals at 4 wpT. (n=3 animals each genotype, two-tailed unpaired Student´s t-test) I) FISH (fluorescence in situ hybridization) for CCL2 mRNA in sciatic nerves of Dnm2iko animals at 4 wpT. CCL2 is expressed by Schwann cells (Sox10+). Scale bar: 10 µm | Mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001

27 Results

A variety of cytokines and chemokines has previously been described to be involved in immune cell recruitment into the sciatic nerve. We analysed a selection by qRT-PCR analysis of lysates from sciatic nerves of Dnm2iko and CtrDnm2 mice at 4, 6, 8 and 14 wpT. The mRNA levels of all the analysed cytokines were increased at 6 wpT in Dnm2iko mice. Interestingly, only for the chemokine (C-C motif) ligand 2 (CCL) the mRNA levels were already increased at 4 wpT (Figure 11, G). Also on the protein level, analysed by ELISA, CCL2 was increased in Dnm2iko sciatic nerves lysate (Figure 11, H). Following the established concept, that Schwann cells themselves can produce CCL2 upon Erk1/2 activation, we performed fluorescence in situ hybridization (FISH) to pinpoint the cell responsible for CCL2 production. Our analysis of sciatic nerve cross sections from Dnm2iko mice identified the Sox10-positive Schwann cells as the only cells expressing CCL2 at a detectable level (Figure 11, I).

Figure 12) The emerging macrophages are beneficial for the recovery after demyelination A) Schematic drawing of the experimental timeline. 10 week-old Dnm2iko and CtrDnm2 mice were injected intraperitoneally with tamoxifen on 5 consecutive days. 4 weeks post-tamoxifen (wpT) these mice were injected intravenously with clodronate liposomes (CL) or saline (Sal) three times (4 wpT, 4wpT + 3d, 4 wpT +6d). Mice were analysed at 6 wpT, 8 wpT and 14 wpT. B) CL and saline treated CtrDnm2 and Dnm2iko mice were phenotypically assessed over a period of 5 weeks. All Dnm2iko showed a transient paraparesis around 6 wpT. Clodronate-treated Dnm2iko recovered more slowly compared to saline-treated Dnm2iko mice. (n=5 mice each genotype). Clinical score: 0 = normal, 1 = less lively, 2 = impaired righting, 3 = absent righting, 4 = ataxic gait, 5 = mild paraparesis, 6 = moderate paraparesis, 7 = severe paraparesis, 8 = tetraparesis, 9 = moribund, 10 = death. C) Representative photograph of a spleen from a saline-treated CtrDnm2, a saline-treated and a clodronate-treated Dnm2iko mouse at 6 wpT. D) Blood monocytes [CD115+] were counted by FACS at 6 wpT, revealing strongly reduced levels in clodronate treated Dnm2iko mice. (n=3 mice each genotype, one-way ANOVA with Tukey´s multiple comparisons test) E) The macrophage marker CD68 was detected by immunostaining of sciatic nerve cross sections from saline- or clodronate-treated CtrDnm2 and Dnm2iko mice at 6 wpT, 8wpT and 14wpT. Scale bar: 25 µm F) Quantification of CD68 positive macrophages (represented in E) (n=3 animals each condition, two-way ANOVA with Tukey´s multiple comparisons test). | Mean ± SEM, *P < 0.05, ***P < 0.001

28 Results

In order to investigate whether the recruitment of macrophages by Schwann cells is beneficial or detrimental for the observed phenotype, we made use of clodronate liposomes (CL). After intravenous injection, liposomes are phagocytosed by macrophages and monocytes. This is followed by an intracellular release and accumulation of clodronate, which induces apoptosis. CL or saline, serving as a negative control, was injected three times starting at 4 wpT, before the massive influx of macrophages was observed. The intravenous injections were repeated three and six days later (Figure 12, A). A cohort of Dnm2iko and CtrDnm2 mice, where half of each group was injected with CL and the other half with saline, was phenotypically assessed as described above. CL- and saline-treated Dnm2iko mice started to develop the clinical phenotype at the same time. After a simultaneous worsening, the effect reached its maximum around 6 wpT. However, Dnm2iko mice treated with CL recovered with a one week delay (Figure 12, B). To ensure that the observed phenotypic change is indeed a result of fewer macrophages, we analysed these mice in more detail. Clodronate liposome-injected mice had an obviously reduced spleen, which is known to serve as a reservoir for monocytes (Figure 12, C). We also analysed the blood of these mice by FACS, where we could detect a strong reduction in the number of monocytes in clodronate treated mice (Figure 12, D). Finally, we analysed the sciatic nerves, where we quantified the number of macrophages on cross sections at 6, 8 and 14 wpT. Clodronate liposome-treated Dnm2iko mice had only half the number of macrophages compared to saline-treated mutants at 6 wpT. At the later time points the number of macrophages was comparable between both treatment groups of Dnm2iko mice (Figure 12, E and F). Overall, these results indicate that the emerging macrophages are beneficial for the recovery after the observed demyelination.

2.1.4 Dynamin 2-Depleted Cells are Replaced by Non-Recombined Schwann Cells In order to determine the fate of individual Schwann cells we made use of the RosaYFP-reporter mouse, which expresses a cytosolic YFP upon Cre-mediated recombination. To gain an overview, we performed an immunostaining for YPF of sciatic nerve cross-sections of Dnm2iko(R) and CtrDnm2(R) mice before (4 wpT), during (6 wpT) and after (14 wpT) the clinical phenotype. This staining indicated a recombination rate of approximately 70 %. More importantly, it showed the loss of nearly all recombined cells at 14 wpT in Dnm2iko(R) mice (Figure 13, A and B). Thereafter we performed a more detailed analysis of the sciatic nerve before and after the clinical phenotype, staining additionally for dynamin 2 and the Schwann cell-marker Sox10. In CtrDnm2(R) mice all Schwann cells expressed Dnm2 at all the analysed time points. However, in Dnm2iko(R) mice a loss in Dnm2 expression at 4 wpT was observed in recombined Schwann cells, whereas non-recombined Schwann cells still expressed Dnm2 at normal levels. At 14 wpT almost no recombined Schwann cells could be observed in mutant mice anymore, while all the present Schwann cells were expressing Dnm2 (Figure 13, C). In agreement with the loss of

29 Results recombined Schwann cells, we detected an increase in the levels of the apoptosis marker cleaved caspase 3 in sciatic nerve lysates of Dnm2iko(R) mice at 6 wpT (Figure 13, D). In order for the sciatic nerve to recover, the lost Schwann cells have to be replaced. We therefore performed a co-staining for Sox10 and EdU (5-ethynyl-2-deoxyuridine) to identify proliferating Schwann cells. Indeed, already at 4 wpT and more evidently at 6 wpT, we could identify proliferating Schwann cells on sciatic nerve cross-sections of Dnm2iko(R) mice (Figure 13, E). Recombined, as well as non-recombined Schwann cells were found to be proliferating (Figure 13, F). [183]

Figure 13) Dynamin 2-depleted cells are replaced by non-recombined Schwann cells A) Immunostaining of CtrDnm2(R) and Dnm2iko(R) sciatic nerves cross-sections at 4 wpT, 6 wpT and 14 wpT stained for YFP (recombined cells). YPF+ cell could no longer be detected at 14 wpT in the Dnm2iko(R) mice. Scale bar: 200 µm B) Quantification of YFP positive Schwann cells (represented in C). (n=3 animals each genotype, two-tailed unpaired Student´s t-test) C) Immunostaining of CtrDnm2(R) and Dnm2iko(R) sciatic nerve cross-sections at 4 wpT and 14wpT stained for dynamin 2 (Dnm2), Sox10 (Schwann cells) and YFP (recombined Schwann cells). CtrDnm2(R) mice showed no change in Dnm2 expression in Schwann cells independent of their recombination (white arrow: recombined Schwann cell expressing Dnm2, white arrowhead: non- recombined Schwann cell expressing Dnm2). In the Dnm2iko(R) mice at 4 wpT the recombined Schwann cells showed a loss of Dnm2 (black arrowheads). At 14 wpT exclusively non-recombined, Dnm2-expressing Schwann cells could be observed. Scale bar: 25 µm D) Western blot analysis of cleaved caspase 3 (cC3) in sciatic nerves of CtrDnm2 and Dnm2iko mice at 4 wpT, 6 wpT and 14 wpT. (n=3 animals each genotype, two-tailed unpaired Student´s t-test) E) Quantification of the EdU (5-ethynyl-2- deoxyuridine) -positive Schwann cells in whole sciatic nerve cross-sections of CtrDnm2 and Dnm2iko mice at 4 wpT, 6 wpT and 14 wpT (n=3 animals each genotype). F) At 6 wpT, YFP-positive (white arrow) and negative (arrowhead) Schwann cells positive for EdU were detected. Scale bar: 25 µm. | Mean ± SEM, ***P < 0.001

Combining these findings with the previously described morphological analysis we could show that the loss of Dnm2 leads to the apoptosis of recombined Schwann cells. To compensate for the loss of these cells, non-recombined Schwann cells dedifferentiate, proliferate and subsequently successfully remyelinate the axons.

30 Results

2.2 SC-Specific Ablation of AP2µ2 in Adult Mice Results in a Late Onset Neuropathy

In line with the current literature, we assumed that the observed phenotype in Dnm2iko(R) mice is a result of impaired clathrin-mediated endocytosis (CME). However, since dynamin 2 has also been described to have a variety of different functions, we decided to generate another mouse model with which we could investigate the role of clathrin-mediated endocytosis in myelin maintenance. To this end we ablated the µ-subunit of the adaptor protein complex 2 (AP-2), which has been described to be absolutely required and exclusively involved in CME. It has previously been shown, that the ablation of the µ2-subunit is sufficient to destabilize the whole complex [184]. [185]

Figure 14) AP2µ2 is ablated in Schwann cells of AP2iko A) Schematic drawing of the inducible, Schwann cell-specific AP2µ2 (µ-subunit of the adaptor protein complex 2) ablation and simultaneous YFP expression. Exons 4-11 of the Ap2m1 allele as well as the stop codon of the RosaYFP-reporter allele are flanked by loxP sites. Mice carrying the floxed alleles were crossed with mice carrying the P0CreERT2 allele, leading to P0CreERT2 Ap2m1fl/fl ROSAYFP+/- mice. Cre recombinase is specifically expressed in Schwann cells upon tamoxifen administration, causing the excision of the floxed regions. B) Schematic drawing of the experimental timeline. 10 week-old CtrAP2(R) and AP2iko(R) mice were injected intraperitoneally with tamoxifen on 5 consecutive days. The mice were analysed at 6 wpT, ½ ypT and 1 ypT. C) Western blot analysis of AP2µ2 protein levels in CtrAP2(R) and AP2iko(R) sciatic nerve lysates at 6 wpT showed reduced protein levels in knockout mice. D) Quantification of AP2µ2 protein levels (represented in C) (n=3 animals each genotype, two-tailed unpaired Student´s t-test). E) Immunostaining of CtrAP2(R) and AP2iko(R) sciatic nerve cross-sections at 6 wpT. Levels of the active form of AP-2, P-AP2µ2Thr156, were reduced in Schwann cells in the AP2iko(R) mice. White arrows: P-AP2µ2Thr156, Sox10 double-positive cells, white arrowheads: P-AP2µ2Thr156 negative, Sox10 positive cells. Scale bar: 20 µm. | Mean ± SEM, *P < 0.05

In the same manner as previously used to achieve Dnm2 ablation, we generated mice in which conditional AP2µ2 ablation in Schwann cells could be induced upon tamoxifen administration by crossing Ap2m1fl/fl with P0CreERT2+ mice. In order to investigate the fate of individual Schwann cells, for most of the experiments the mice also carried a heterozygous ROSAYFP+/- reporter, which upon recombination leads to the expression of cytosolic YFP (Figure 14, A) [177]. AP2µ2 was ablated by injecting 10 week-old Ap2m1fl/fl P0CreERT2+ (ROSAYFP+/-) mice with tamoxifen on five consecutive days. Tamoxifen-injected Ap2m1fl/fl P0CreERT2+ mice will further be referred as AP2iko and tamoxifen-injected Ap2m1fl/fl P0CreERT2- mice as CtrAP2, whereas tamoxifen-injected Ap2m1fl/fl P0CreERT2+ ROSAYFP+/- mice

31 Results will further be referred as AP2iko(R) and tamoxifen-injected Ap2m1wt/wt P0CreERT+ ROSAYFP+/- mice as CtrAP2(R). Since we assumed that the phenotype in Dnm2iko mice is a result of lacking CME, we were expecting to see a similar phenotype in the AP2iko mice. However, 6 wpT AP2iko mice did not develop a clinical phenotype and were not distinguishable from their CtrAP2 littermates. Therefore, we analyzed mice at 6 weeks post-tamoxifen (wpT) as an early time point, at an intermediate ½ year post-tamoxifen (ypT) and a late 1 ypT time point in order to properly assess myelin maintenance in the absence of clathrin- mediated endocytosis (Figure 14, B). The loss of AP2µ2 protein could be detected by Western blot analysis in sciatic nerve extracts of 6 wpT AP2iko(R) mice (Figure 14, C and D). Further we could also detect the loss of the active, phosphorylated form of AP2µ2 (P-AP2µ2Thr156) by an immunostaining of sciatic nerve cross-sections of AP2iko(R) compared to CtrAP2(R) mice at 6 wpT (Figure 14, E).

Figure 15) Ablation of AP2µ2 leads to gait impairments in aged mice A) Representative digital paw prints of CtrAP2 and AP2iko mice at 1 ypT. Red lines represent the measurement of the stride length. Green lines represent the measurement of the base of support (BOS). B) Quantification of the stride length of the hind paws of CtrAP2(R)* and AP2iko(R)* mice at ½ ypT and 1 ypT (n=5 animals each genotype, two-tailed unpaired Student´s t-test). C) Quantification of the duty cycle of the hind paws of CtrAP2(R)* and AP2iko(R)* mice at ½ ypT and 1 ypT (n=5 animals each genotype, two-tailed unpaired Student´s t-test). D) Representative diagram of the gait pattern of CtrAP2 and AP2iko mice at 1 ypT. Colored bars represent the time a paw is in contact with the floor plate. RF: right front paw, RH: right hind paw, LF: left front paw, LH: left hind paw, S: support (time at which three or four paws touched the floor plate at the same time). E) Quantification of the support as percentage of the whole run of CtrAP2(R)* and AP2iko(R)* mice at ½ ypT and 1 ypT, (n=5 animals each genotype, two-tailed unpaired Student´s t-test). F) Quantification of the base of support (BOS) of the hind paws of CtrAP2(R)* and AP2iko(R)* mice at ½ ypT and 1 ypT (n=5 animals each genotype, two-tailed unpaired Student´s t-test). G) Representative digital, complete paw print during a stance of a right hind paw of CtrAP2 and AP2iko mice at 1 ypT. Left panel: original image. Right panel: heat map of the print intensity; blue (low intensity) to red (high intensity). H) Quantification of the print area of the hind paws of CtrAP2(R)* and AP2iko(R)* mice at ½ ypT and 1 ypT (n=5 animals each genotype, two-tailed unpaired Student´s t-test). I) Quantification of the mean intensity of the hind paws of CtrAP2(R)* and AP2iko(R)* mice at ½ ypT and 1 ypT, (n=5 animals each genotype, two-tailed unpaired Student´s t-test). | Mean ± SEM, **P < 0.01, ***P < 0.001

32 Results

We were not able to detect an obvious abnormal clinical phenotype in the Ap2m1 knockout mice up to 1 ypT. As a consequence, we decided to assess the gait and locomotion using an automated quantitative gait analysis system, CatWalk XT. The performance of CtrAP2(R) and AP2iko(R) mice was assessed at ½ ypT and 1 ypT. We were able to detect an impaired gait pattern at 1 ypT but not at ½ ypT in AP2iko compared to CtrAP2 mice. The CatWalk analysis revealed a tendency towards a decrease in stride length of the hind limbs at 1 ypT in AP2iko compared to CtrAP2 mice (Figure 15, A and B). Further the analysis showed a deficit in balance- and posture-associated parameters. We could detect a significant increase in the duty cycle (percentage of a step cycle where the paw touches the glass plate), looking at the hind paws of AP2iko compared to CtrAP2 mice at 1 ypT (Figure 15, C). In addition, also the support (percentage of the whole run when the animal has three or four paws on the glass plate) was significantly increased at 1 ypT in AP2iko compared to CtrAP2 mice (Figure 15, D and E). However, we could not detect differences in the base of support (BOS) up to 1 ypT in the hind paws of these animals (Figure 15, F). Detailed examination of the paw prints displayed that the print area and mean intensity of the paw prints of the hind limbs of AP2iko were increased compared to CtrAP2 mice at 1 ypT (Figure 15, G-I).

Taken together, these results show that the loss of AP2µ2 in myelinated Schwann cells leads to a mild, late onset phenotype.

2.2.1 Aged AP2iko Mice Develop Myelin Aberrations Histological analysis of sciatic nerves revealed myelin aberrations in AP2iko(R)* mice at ½ ypT and 1 ypT (Figure 16, A). Aberrant myelin features included de- and remyelination, demyelinated fibers, onion bulbs, immune cells, degenerating axons, myelin infoldings, tomacula and myelin outfoldings (Figure 16, B, I, K, M). Measurements of the g-ratio at ½ ypT did not show differences between CtrAP2(R) and AP2iko(R) mice (Figure 16, C and D). However, in line with the ongoing de- and remyelination at 1 ypT, sciatic nerves of AP2iko mice were hypomyelinated (Figure 16, E and F). The number of demyelinated fibers increased in the AP2iko mice at 1 ypT (Figure 16, G). In order to assess axonal loss, all the sorted axons within a whole sciatic nerve of CtrAP2 and AP2iko mice were counted. This analysis showed a slight but significant reduction of axons in AP2iko compared to CtrAP2 mice at 1 ypT (Figure 16, H). Furthermore, we analyzed some of the observed myelin aberrations in more details by counting them on whole nerve ultrathin sections of CtrAP2(R)* and AP2iko(R)* mice at 6 wpT, ½ ypT and 1 ypT. There was no significant change in infoldings at either timepoint in AP2iko(R)* compared to CtrAP2(R)* mice, and only a slight trend to an increase in tomacula in AP2iko mice at 1 ypT (Figure 16, J and L). On the other hand, quantification of outfoldings revealed a mild increase at ½ ypT as well as a substantial increase at 1 ypT in AP2iko(R)* mice (Figure 16, N).

(R)* indicates an analysis for which not all the time points consisted of the same genotype with regard to the ROSAYFP Cre-reporter. 33 Results

Figure 16) AP2µ2 depleted mice develop late onset histopathological abnormalities A) Ultrathin sections of sciatic nerves of CtrAP2(R)* and AP2iko(R)* mice at 6 weeks post-tamoxifen (wpT), ½ year post-tamoxifen (ypT) and 1 ypT. Scale bar: 10 µm B) Electron micrographs showing pathological hallmarks in sciatic nerves of AP2iko mice at 1 ypT. Top panel from left to right: demyelinating Schwann cell, demyelinated fibre, remyelinating Schwann cell (white arrowhead). Bottom panel from left to right: Onion bulb, degenerating axon, macrophage. Scale bar: 5 µm. C) G-ratio analysis of CtrAP2(R) and AP2iko(R) sciatic nerves at ½ ypT showed no difference in myelin thickness (n=3 animals each genotype, two-tailed unpaired Student´s t-test). D) Scatter plot of g-ratios of CtrAP2(R) and AP2iko(R) sciatic nerves at ½ ypT. (n=3 animals each genotype, 150 axons measured each animal) E) G-ratio analysis of CtrAP2 and AP2iko sciatic nerves at 1 ypT revealed hypomyelination in mutant mice (n=3 animals each genotype, two-tailed unpaired Student´s t-test). F) Scatter plot of g-ratios of CtrAP2 and AP2iko sciatic nerves at 1 ypT. (n=3 animals each genotype, 150 axons measured each animal) G) Counting of demyelinated fibres on sciatic nerve ultrathin cross-sections showed an increase at 1 ypT (n=3 animals each genotype, two-way ANOVA with Sidak´s multiple comparisons test). H) Counting of axons on sciatic nerve ultrathin cross-sections revealed an axonal loss at 1 ypT (n=3 animals each genotype, two-tailed unpaired Student´s t-test). I) Representative electron micrograph showing a myelin infolding in an AP2iko mouse at 1 ypT. Scale bar: 5 µm. J) Quantification of myelin infoldings at 6 wpT, ½ ypT and 1 ypT (n=3 animals each genotype, two-way ANOVA with Sidak´s multiple comparisons test). K) Representative electron micrograph showing a tomaculum in an AP2iko mouse at 1 ypT. Scale bar: 5 µm. L) Quantification of tomacula at 6 wpT, ½ ypT and 1 ypT (n=3 animals each genotype, two-way ANOVA with Sidak´s multiple comparisons test). M) Representative electron micrograph showing a myelin outfolding in an AP2iko mouse at 1 ypT. Scale bar: 5 µm. N) Quantification of myelin outfoldings at 6 wpT, ½ ypT and 1 ypT (n=3 animals each genotype, two-way ANOVA with Sidak´s multiple comparisons test). | Mean ± SEM, *P < 0.05, ***P < 0.001 34 Results

In line with the observed de- and remyelination and the consequential hypomyelination, we found that the levels of some myelin proteins were reduced. Western blot analysis of sciatic nerve lysates showed a small but significant reduction in the levels of P0 and MBP in AP2iko compared to CtrAP2 mice at 1 ypT. However no significant change in the protein levels of MAG could be detected (Figure 17, A-D). As observed in the Dnm2iko mice, we hypothesised that also in the AP2iko mice the de- and remyelination would be accompanied by a dedifferentiation of Schwann cells. Therefore we performed a qRT-PCR analysis for markers known to be altered in dedifferentiated Schwann cells. Krox20 and Periaxin were significantly reduced in AP2iko compared to CtrAP2 mice at 1 ypT. However, P0 and MBP showed no significant changes at either time point. At ½ ypT we could detect a slight as well as at 1 ypT a substantial increase in the expression levels p75 and cyclin D1. Furthermore, we detected a mild increase in the levels of c-Jun and Krox24 at 1 ypT in AP2iko mice (Figure 17, E).

Figure 17) AP2iko mice show signs of dedifferentiation A) Western blot analysis of myelin-associated glycoprotein (MAG), myelin protein zero (P0) and myelin basic protein (MBP) levels in CtrAP2 and AP2iko sciatic nerves lysates at 1 ypT. P0 and MBP levels are reduced in AP2iko mice at 1 ypT. B) Quantification of MAG protein levels (represented in B) (n=3 animals each genotype, two-tailed unpaired Student´s t-test). C) Quantification of P0 protein levels (represented in B) (n=3 animals each genotype, two-tailed unpaired Student´s t-test). D) Quantification of MBP protein levels (represented B) (n=3 animals each genotype, two-tailed unpaired Student´s t-test). E) qRT-PCR analysis of Schwann cell differentiation and dedifferentiation markers in CtrAP2(R)* and AP2iko(R)* sciatic nerves at ½ year post-tamoxifen (ypT) and 1 ypT (n= 5 animals each genotype for ½ ypT, n=4 animals each genotype for 1 ypT, two-tailed unpaired Student´s t-test). | Mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001

35 Results

2.2.2 Ablation of AP2µ2 Leads to the Recruitment of Macrophages Beside the morphological changes of Schwann cells, we observed various immune cells on ultrathin sections of sciatic nerves from AP2iko mice at 1 ypT. The most prominent ones were the highly phagocytic macrophages containing vesicles filled with myelin debris (Figure 18, A). We therefore quantified the number of macrophages on cross-sections of AP2iko(R)* and CtrAP2(R)* mice at 6 wpT, ½ ypT and 1 ypT. At ½ ypT we could see a tendency and at 1 ypT a clear increase in the number of macrophages in sciatic nerves of AP2iko(R)* mice (Figure 18, B and C). We then analysed the relative mRNA expression of different macrophage chemoattractants. We could detect a significant increase in the levels of CCL2 and MIP1α, two cytokines that can be produced by Schwann cells, but no change in CSF1 levels in AP2iko(R)* compared to CtrAP2(R)* mice at ½ ypT and 1 ypT (Figure 18, D). In order for macrophages to invade the sciatic nerve, the blood-nerve barrier (BNB) has to break down. Western blot analysis of serum albumin protein levels in sciatic nerve lysates showed a tendency at ½ ypT and a clear increase at 1 ypT in AP2iko(R)* compared to CtrAP2(R)* mice, indicating an opening of the BNB (Figure 18, E and F).

Figure 18) Inflammation in sciatic nerve of AP2iko mice A) Electron micrograph of a macrophage (MΦ) from an AP2iko mouse at 1 ypT. Scale bar: 5 µm. B) Immunostaining of CtrAP2(R)* and AP2iko(R)* sciatic nerve cross-sections for the macrophage marker CD68 at 6 weeks post-tamoxifen (wpT), ½ year post-tamoxifen (ypT) and 1 ypT. Scale bar: 50 µm. C) Quantification of CD68-positive cells of whole sciatic nerve cross-sections (n=3 animals each genotype, two-way ANOVA with Sidak´s multiple comparisons test). D) qRT-PCR analysis of CCL2, MIP1α and CSF1 in CtrAP2(R)* and AP2iko(R)* sciatic nerves at ½ ypT (n=5 animals each genotype) and 1 ypT (n=4 animals each genotype) (two-tailed unpaired Student´s t-test). E) Western blot analysis of albumin protein levels of CtrAP2(R)* and AP2iko(R)* sciatic nerves lysates at ½ ypT and 1 ypT. F) Quantification of albumin protein levels (represented in E) (n=3 animals each genotype, two-way ANOVA with Sidak´s multiple comparisons test). | Mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001

36 Results

2.2.3 The Number of Schwann Cells Increases in Aged AP2iko mice On ultrathin sections we had the impression that the number of cells increased in aged AP2iko mice. This perception was confirmed by an immunostaining for the Schwann cell marker Sox10, which showed a significant increase in Sox10-postive cells in AP2iko compared to CtrAP2 mice at 1 ypT (Figure 19, A). This increase was further supported by Western blot analysis, showing an increase in the level of Sox10 in AP2iko mice at 1 ypT (Figure 19, B and C).

Figure 19) Non-recombined Schwann cells proliferate in AP2iko while the number of recombined Schwann cells stays unaltered A) Quantification of Sox10-positive cells on whole sciatic nerve cross-sections of CtrAP2(R)* and AP2iko(R)* at 6 wpT, ½ ypT and 1 ypT (n=3 animals each genotype, two-way ANOVA with Sidak´s multiple comparisons test). B) Western blot analysis of Sox10 protein levels in CtrAP2(R)* and AP2iko(R)* sciatic nerve lysates at ½ ypT and 1 ypT. C) Quantification of Sox10 protein levels (represented in B) (n=3 animals each genotype, two-way ANOVA with Sidak´s multiple comparisons test). D) Quantification of P-AP2µ2Thr156-negative Sox10-positive cells on whole sciatic nerve cross-sections of CtrAP2(R)* and AP2iko(R)* at 6 wpT, ½ ypT and 1 ypT (n=3 animals each genotype, one-way ANOVA with Tukey´s multiple comparisons test). E) Quantification of P-AP2µ2Thr156, Sox10 double-positive cells on whole sciatic nerve cross-sections of CtrAP2(R)* and AP2iko(R)* at 6 wpT, ½ ypT and 1 ypT (n=3 animals each genotype, one-way ANOVA with Tukey´s multiple comparisons test). F) Immunostaining to identify proliferating (Ki67+) cells on CtrAP2 and AP2iko sciatic nerve cross-sections at 1 ypT. White arrows: Ki67, Sox10 double-positive cells. Sscale bar: 25 µm G) Quantification of Ki67, Sox10 double-positive cells as percentage of all Sox10-positive cells (n=3 animals each genotype). H) Immunostaining to identify apoptotic (P-H2A.X+) cells on CtrAP2 and AP2iko sciatic nerve cross-sections at 1 ypT. White arrow: P-H2A.X, Sox10 double-positive cell, White arrowhead: P-H2A.X positive, Sox10-negative cell. Scale bar: 25 µm I) Quantification of P-H2A.X, Sox10 double-positive cells as percentage of all Sox10-positive cells (n=3 animals each genotype, two-tailed unpaired Student´s t-test). | Mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001

37 Results

In order to see whether the increased number of Schwann cells in the AP2iko mice was arising from the recombined (AP2µ2-negative) or non-recombined (AP2µ2-positive) Schwann cells, we analyzed a Sox10, AP2µ2 double-staining and subdivided the counted Sox10-positive nuclei into an AP2µ2-negativ (AP2µ-) and an AP2µ2-positive (AP2µ2+) fraction. This approach allowed us to identify the excessive Schwann cells at 1 ypT to be AP2µ2-positive, and thereby non-recombined Schwann cells (Figure 19, D and E). Since in healthy adult sciatic nerves no Schwann cell turnover takes place, it seemed likely, that the increase in Schwann cell number is due to proliferation. Indeed, by performing a Ki67-immunostaining at 1 ypT, we could see an increase in proliferating Schwann cells in AP2iko compared to CtrAP2 mice (Figure 19, F and G). Further we also performed a preliminary experiment using an immunostaining for P-H2A.X- to identify apoptotic cells, where we could detect a slight increase in apoptotic Schwann cells in AP2iko mice at 1 ypT (Figure 19, H and I).

Combining these findings with the previously described physiological and morphological analysis, we could show that the loss of AP2µ2 leads to a mild, late onset neuropathy. Mutant Schwann cells form myelin aberrations and eventually demyelinate. However in contrast to dynamin 2-depleted cells they do not necessarily undergo apoptosis, which overall leads to an increase in Schwann cell number.

2.3 The Surfaceome of CME-Impaired Primary Rat Schwann Cells

Impaired endocytosis has often been linked to changes in proteins on the plasma membrane [186]. By ablating dynamin 2 or AP2µ2 in Schwann cells, we therefore tamper with their surface proteins. While dynamin 2 plays a role in most forms of endocytosis, AP-2 is exclusively involved in CME. To analyse the changes on the cell surface upon Dnm2 or AP-2 depletion, we collaborated with the group of Prof. Bernd Wollscheid to perform mass spectrometry-based quantitative cell surface capturing (CSC). This technique allowed us to identify N-linked cell surface in a quantitative manner. We therefore generated primary rat Schwann cells (rSC) carrying two different inducible shRNAs for AP2µ2 and for dynamin 2, as well as control cells with an inducible non-silencing (NS) shRNA. Three days after doxycycline-induced expression of the respective shRNAs, the loss of the target protein could be seen by Western blot analysis (Figure 20, A). In order to link the loss of protein to an actual impairment in endocytosis, we performed a transferrin uptake assay. Cells expressing shRNAs for AP2µ2 and Dnm2 showed a significant reduction in transferrin uptake compared to cells expressing a NS shRNA (Figure 20, B and C). For these five sets of samples we performed a CSC to quantitatively assess the surface proteins upon AP2µ2 or Dnm2 knockdown. The mass spectrometer identified 405 glycoproteins present on the rSC surface (Figure 20, D, and Table 17 (Appendix)). We than quantitatively compared the surface glycoproteins of the cells expressing shRNAs for AP2µ2 or Dnm2 with the ones on cells expressing a NS shRNA. The only surface protein that was commonly regulated in all the knockdown conditions was the

38 Results

transferrin receptor (TfR) (Figure 20, E and F). In most cell types, the TfR is required for the uptake of iron and its proper function is therefore important for iron homeostasis.

Figure 20) The levels of TfR on the plasma membrane increase in dynamin 2 and AP2µ2-depleted Schwann cells A) Western blot analysis of AP2µ2 or Dnm2 protein levels in rat Schwann cells infected with lentivirus expressing different shRNAs. NS shRNA: non-silencing shRNA, shRNA AI + AII: shRNAs against AP2µ2, shRNA DI + DII: shRNAs against dynamin 2 (n=3 wells with individually induced knockdown, one-way ANOVA with Dunnett´s multiple comparison test) B) Transferrin (Tf) uptake of rat Schwann cells infected with lentivirus expressing different shRNAs (for details see A)). White: Transferrin, Alexa 568 conjugate, Blue: DAPI, Scale bar: 20 µm C) Quantification by fluorescence-activated flow cytometry revealed impaired transferrin uptake in knockdown cells. (n=3 wells with individually induced knockdown, one-way ANOWA with Sidak´s multiple comparison test) D) Representation of the cell surface gylcoproteome of primary rat Schwann cells. Each dot represents an identified glycoprotein. E) Volcano plots of differentially expressed cell surface proteins found by cell surface capturing (CSC). Cut-off: p-value <0.05, fold change >1.41 F) Venn-diagram of up- and downregulated surface proteins. The only protein that showed the same regulation for all the different shRNAs was the transferrin receptor (TfR). | Mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001

39 Results

2.4 Schwann Cells and Iron Deficiency

To investigate the role of iron in Schwann cells, we treated primary rat Schwann cells with deferoxamine (DFO), a potent iron chelator. DFO-induced iron deficiency in Schwann cells lead within 24 hours to an upregulation of the TfR and a downregulation of both ferritin subunits (Figure 21, A-D). The survival of rat Schwann cells cultured in the presence of DFO was strongly impaired, and their numbers were gradually reduced over a time course of 4 days when compared to vehicle-treated cells (Figure 21, E and F).

Figure 21) Iron deficiency in vitro leads to apoptosis and myelin instability A) Western blot analysis of transferrin receptor (TfR) and the iron storage protein ferritin (heavy (H) and light (L) chain) in primary rat Schwann cell lysates 0h, 6h, 24h and 48h after deferoxamine (DFO) treatment. Within 24 hours of iron depletion, TfR is upregulated and both ferritin subunits are downregulated. B) Quantification of TfR protein levels (represented in A) (n=3 individually treated wells, two-tailed unpaired Student´s t-test). C) Quantification of ferritin H protein levels (represented in A) (n=3 individually treated wells, two-tailed unpaired Student´s t-test). D) Quantification of ferritin L protein levels (represented in A) (n=3 individually treated wells, two-tailed unpaired Student´s t-test). E) DAPI staining of primary rat Schwann cells treated with vehicle or DFO over a 4 day time course. Iron depletion led to a continuous reduction in the number of cells. Scale bar: 200 µm. F) Quantification of cell survival upon DFO treatment (represented in E) (n=3 wells each condition). G) Western blot analysis of the apoptosis marker cleaved caspase 3 (cC3) and the phosphorylation of p38 in primary rat Schwann cell lysates at 0h, 6h, 24h and 48h of deferoxamine (DFO) treatment. Within 24 hours of iron depletion, caspase 3 is cleaved and p38 is phosphorylated. H) Quantification of cC3 protein levels (represented in G) (n=3 individually treated wells, two-tailed unpaired Student´s t-test). I) Quantification of p38 phosphorylation (P-p38Thr180/Tyr182 / total p38) (represented in G) (n=3 individually treated wells, two-tailed unpaired Student´s t-test). J) Immunostaining of dorsal root ganglia (DRG) co-cultures treated with vehicle or DFO for 4 day after 7 days of myelination. deferoxamine treated cultures showed signs of demyelination. Scale bar: 200 µm. | Mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001

40 Results

Since we assumed that the observed cell death is due to apoptosis, we analysed the protein levels of cleaved caspase 3 and phospho-p38 (P-p38) by Western blotting. Indeed, we could detect an upregulation of cleaved caspase 3 as well as an activation of p38 in DFO treated cells (Figure 21, G-I). In order to address the iron deficiency in a system that is closer to the physiological scenario, we made use of dorsal root ganglion (DRG) explant cultures. Mouse DRG explant cultures consist of sensory neurons, Schwann cells and fibroblasts. In these cultures, Schwann cells can differentiate and myelinate the axons which makes this the model that best mimics the situation in a peripheral nerve. To analyze the effect of iron deficiency in a maintenance situation, before adding DFO we cultivated DRG explant cultures to the point where the Schwann cells successfully myelinated the axons. Four days after treatment with either DFO or vehicle we stained these cultures for MBP and neurofilament. The neurofilament staining, which marks the neurons did not show obvious alterations. However, the Schwann cells, whose myelin was stained with MBP, showed signs of demyelination such as distorted and fragmented internodes in DFO-treated cultures (Figure 21, J). Taken together, these results suggest that Schwann cells sense iron deprivation and try to compensate this deficiency by an upregulation of TfR and downregulation of ferritin. Furthermore, these data show that massive iron loss is detrimental for Schwann cell survival as well as for myelin maintenance in vitro.

2.5 Iron Deficiency in Different Knockout Mice

We have previously demonstrated that Schwann cells with impaired clathrin-mediated endocytosis accumulate TfRs on their surface. Together with the fact that Schwann cells require a certain amount of iron to maintain proper myelination, we assessed the iron status of our mouse models with impaired CME. We therefore analysed the protein levels of TfR and ferritin H in sciatic nerve lysates from Dnm2iko(R) mice. At 4 wpT, prior to the onset of the clinical phenotype, Dnm2iko(R) mice showed an increase in TfR and a decrease in ferritin H protein levels compared to CtrDnm2(R) mice, indicating iron deficiency (Figure 22, A-C). An additional iron staining showed a reduced number of iron positive particles in sciatic nerves of Dnm2iko(R) mice (Figure 22, D and E). We further analysed the iron status of our AP2iko(R) mice, which in contrast to Dnm2iko mice have exclusively affected CME. At 6 wpT we were not able to detect alterations in the protein levels of TfR and ferritin H in sciatic nerve lysates of AP2iko(R) compared to CtrAP2(R) mice. However, at ½ ypT AP2iko(R) mice showed clear signs of iron deficiency with a strong upregulation of TfR and simultaneous downregulation of ferritin H (Figure 22, F-H). Additional to the changes in the protein levels, the TfR mRNA was slightly but significantly upregulated in AP2iko(R) mice at ½ ypT (Figure 22, I).

41 Results

Figure 22) Schwann cell-specific knockout mice with impaired clathrin-mediated endocytosis show signs of iron deficiency A) Western blot analysis of transferrin receptor (TfR) and ferritin heavy chain (ferritin H) in CtrDnm2(R) and Dnm2iko(R) sciatic nerve lysates at 4 weeks post-tamoxifen (wpT). The upregulation of TfR together with the simultaneous downregulation of ferritin indicates an iron deficiency in Dnm2iko(R) mice. B) Quantification of TfR protein levels (represented in A) (n=3 animals each genotype, two-tailed unpaired Student´s t-test). C) Quantification of ferritin H protein levels (represented in A) (n=6 animals each genotype, two-tailed unpaired Student´s t-test) D) Iron staining of CtrDnm2(R) and Dnm2iko(R) sciatic nerve cross-sections at 4 wpT showed reduced iron levels. Scale bar: 25 µm E) Quantification of iron-positive particles per sciatic nerve (SN) (represented in D) (n=3 animals each genotype, two-tailed unpaired Student´s t-test). F) Western blot analysis of transferrin receptor (TfR) and ferritin heavy chain (ferritin H) in CtrAP2(R) and AP2iko(R) sciatic nerve lysate at 6 wpT and ½ year post-tamoxifen (ypT). The upregulation of TfR together with the simultaneous downregulation of ferritin indicates an iron deficiency in AP2iko(R) mice at ½ ypT. G) Quantification of TfR protein levels (represented in F) (n=3 animals each genotype, two-tailed unpaired Student´s t-test). H) Quantification of ferritin H protein levels (represented in F) (n=3 animals each genotype, two-tailed unpaired Student´s t-test). I) qRT-PCR analysis of TfR showing an increase in AP2iko(R)* sciatic nerves at ½ ypT (n=5 animals each genotype) (two-tailed unpaired Student´s t-test). J) Western blot analysis of TfR and ferritin H levels from sciatic nerve lysate of different animals at postnatal day 5 (P5). P0Cre+Dnm2fl/fl mice showed signs of iron deficiency compared to their Cre-negative littermates (P0Cre-Dnm2fl/fl). Mice carrying a heterozygous point-mutation (K562E), which in humans leads to Charcot-Marie-Tooth disease, do not show disturbed iron homeostasis. DhhCre+Dicerfl/fl mice do not show disturbed iron homeostasis compared to their littermates (DhhCre-Dicerfl/fl). K) Quantification of TfR and ferritin H protein levels (represented in J) (n=3 animals each genotype, two-tailed unpaired Student´s t-test). L) Ultrathin sections of sciatic nerves of different animals at P5. Wildtype (wt) as well as Dnm2wt/K562E mice showed proper radial sorting and onset of myelination, whereas P0Cre+Dnm2fl/fl and DhhCre+Dicerfl/fl mice showed impaired radial sorting and an impaired onset of myelination. Scale bar: 10 µm | Mean ± SEM, *P < 0.05, **P < 0.01

42 Results

In order to check if iron deficiency is a general consequence of dynamin 2 depletion we additionally analysed sciatic nerve lysates from P5 P0Cre+Dnm2fl/fl and P0Cre-Dnm2fl/fl mice. Also during development a Dnm2-deficiency lead to an upregulation of the TfR and a downregulation of ferritin H, consistent with iron deficiency (Figure 22, J and K, left panel). We further analysed Dnm2wt/K562E mice. These mice carry a heterozygous point mutation, K562E, which in humans leads to CMT. The levels of TfR and ferritin H in these mice were unaltered compared to Dnm2wt/wt littermates at P5 (Figure 22, J and K, centre panel). To check whether the observed changes in the levels of TfR and ferritin H in P0Cre+Dnm2fl/fl mice are not just a result of the emerging morphological changes, we added DhhCreDicerfl/fl mice to our analysis. At P5, both P0Cre+Dnm2fl/fl and DhhCre+Dicerfl/fl mice showed an impairement in radial sorting and an impaired onset of myelination, while Dnm2wt/K562E were indistinguishable from wildtype (wt) mice (Figure 22, L). The protein levels of TfR and ferritin H were not changed in DhhCre+Dicerfl/fl compared to DhhCre-Dicerfl/fl control mice. Taken together, iron deficiency seems to be a common result of inhibition of CME and not a consequence of an emerging morphological change.

2.6 The Contribution of Iron Deficiency to the Observed Phenotype in Dnm2iko Mice

In order to elaborate the contribution of the iron deficiency to the observed phenotype of Dnm2iko(R) mice, we established a suitable in vitro system. To do so, we bred Dnm2fl/fl mice with mice carrying a tamoxifen-inducible Cre recombinase under control of the PLP promoter (PLPCreERT2), since our experience showed us that Schwann cells originating from P0CreERT2+Dnm2fl/fl embryos do not recombine in vitro upon hydroxy-tamoxifen (OHT) administration. Therefore, DRGs derived from PLPCreER2+Dnm2fl/fl embryos were cultured until the Schwann cells had successfully myelinated the axons. Then, OHT was added to induce recombination and cause dynamin 2 ablation. Six days after OHT- treatment, the cultures were stained for MBP and neurofilament. Cultures treated with OHT showed signs of demylination, such as distorted and fragmented internodes. This demyelination could be partially prevented by supplementing the OHT-treated cultures with iron dextran (FeDex) (Figure 23, A and B). Additionally to the myelin staining, we analysed Krox20 protein levels by Western blotting. The strong reduction in Krox20 protein levels was also partially prevented by the supplementation with FeDex (Figure 23, C and D). DRGs from P0CreERT2-Dnm2fl/fl did not show detectable changes upon FeDex treatment (Figure 25 (Appendix)). In a next step, we performed an in vivo rescue experiment in which Dnm2iko(R) mice were supplemented with iron (Figure 23, E). A cohort of Dnm2iko(R) and CtrDnm2(R) mice, where half of each genotype group received weekly injections of iron dextran (FeDex) and the other half of saline, was phenotypically assessed as described above. All Dnm2iko(R), independent of their treatment, showed the same clinical

43 Results phenotype, starting between 4 and 5 wpT and peaking at 6 wpT, the time point of analysis (Figure 23, F). A blood analysis at 6 wpT showed increased levels of serum ferritin in FeDex-treated mice, indicating a successful systemic iron overload (Figure 23, G). Histological analysis of sciatic nerve did not show detectable differences in Dnm2iko mice treated with FeDex or saline, including the number of demyelinated fibres (Figure 23, H and I). Further, we analysed the sciatic nerves of these mice on a transcriptional level. Iron supplementation did not have detectable effect in the CtrDnm2(R) mice. On the other hand, Dnm2iko mice treated with FeDex showed a milder upregulation of markers for dedifferentiated Schwann cells, such as p75, cyclin D1, Sox2 and c-Jun, when compared to saline-treated mutants. However, we could not detect changes in the pattern of downregulated myelination markers comparing FeDex- and saline-treated Dnm2iko mice (Figure 23, J). We then performed a Western blot analysis for p75, which showed the strongest change on the mRNA level. Indeed, we were able to detect a small but significant reduction in the protein levels of p75 in sciatic nerve lysates of FeDex- compared to saline-treated Dnm2iko mice (Figure 23, K and L). To rule out that the observed reduction in p75 is not just a result of fewer cells due to increased apoptosis, we performed a Western blot for cleaved caspase 3 (cC3), where we could not detect a difference between the differently treated Dnm2iko mice (Figure 23, M and N). Overall, iron supplementation in vitro can partially prevent demyelination, whereas in vivo it only seems to have an effect on the dedifferentiation of Schwann cells. These results demonstrated that iron deficiency might contribute to the observed phenotype in Dnm2iko mice.

Figure 23) Continued F) FeDex or saline treated CtrDnm2(R) and Dnm2iko(R) mice were phenotypically assessed over a period of 6 weeks. All Dnm2iko showed a transient paraparesis around 6 wpT with the same time of onset independent of FeDex treatment. (n=5 mice each treatment). Clinical score: 0 = normal, 1 = less lively, 2 = impaired righting, 3 = absent righting, 4 = ataxic gait, 5 = mild paraparesis, 6 = moderate paraparesis, 7 = severe paraparesis, 8 = tetraparesis, 9 = moribund, 10 = death. G) Blood serum ferritin levels of CtrDnm2(R) and Dnm2iko(R) mice treated with saline or FeDex were analysed at the timepoint of sacrification (6 wpT). All FeDex treated mice showed an increase in serum ferritin levels (n=5 mice each condition, one-way ANOVA with Tukey´s multiple comparisons test). H) Ultrathin sections of CtrDnm2(R) and Dnm2iko(R) mice treated with saline or FeDex. Demyelinated fibers are coloured in red. I) Quantification of demyelinated fibers on whole sciatic nerve sections (represented in H) (n=5 mice each condition, one-way ANOVA with Tukey´s multiple comparisons test). J) qRT-PCR analysis of Schwann cell differentiation and dedifferentiation markers in sciatic nerves of CtrDnm2(R) and Dnm2iko(R) mice treated with saline or FeDex. (n=4 animals each condition, one-way ANOVA with Tukey´s multiple comparisons test) K) Western blot analysis of p75 in sciatic nerves of CtrDnm2(R) and Dnm2iko(R) mice treated with saline or FeDex. L) Quantification of p75 protein levels (represented in K) (n=3 mice each condition two-tailed unpaired Student´s t-test). M) Western blot analysis of cleaved caspase 3 (cC3) in sciatic

44 Results

nerves of CtrDnm2(R) and Dnm2iko(R) mice treated with saline or FeDex. N) Quantification of cC3 protein levels (represented in K) (n=3 mice each condition two-tailed unpaired Student´s t-test) | Mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001

Figure 23) Iron deficiency might contribute to the observed phenotype resulting from dynamin 2 depletion. A) Immunostaining of dorsal root ganglia (DRG) explant cultures from PLPCreERT2+Dnm2fl/fl embryos. Hyroxy-Tamoxifen (OHT)- treated cultures began to demyelinate. This demyelination could be partially prevented by supplementing the OHT-treated cultures with iron dextran (FeDex). Degenerating MBP is false-coloured in yellow. Scale bar: 100 µm. B) Quantification of intact internodes over all internodes (represented in A) C) Western blot analysis of Krox20 and dynamin 2 (Dnm2) in DRG explant cultures (as described in A). Levels of dynamin 2 were reduced upon treatment with OHT. The observed reduction in the levels of Krox20 upon OHT treatment could be partially prevented by FeDex supplementation. D) Quantification of Krox20 protein levels (represented in C) (n=3 coverslips each condition, one-way ANOVA with Sidak´s multiple comparisons test). E) Schematic drawing of the experimental timeline for the in vivo rescue experiment. 10 week-old CtrDnm2(R) and Dnm2iko(R) mice were injected intraperitoneally with TMX on 5 consecutive days. From 1 week post-tamoxifen (wpT) onwards these mice received

45 Results weekly intraperitoneal injections of iron dextran (FeDex) or saline. Mice were analysed at 6 wpT.

2.7 Transcriptome Analysis after Dynamin 2 or AP2µ2 Ablation

In a global approach to find transcriptional changes after Dnm2 or AP2µ2 ablation we performed RNA deep sequencing on sciatic nerve lysates of the different mutants and the corresponding control mice. All mice were analysed 4 wpT, which is prior to the clinical phenotype in the Dnm2iko(R) mice. More than 4400 genes were significantly changed in Dnm2iko(R) compared to CtrDnm2(R) mice. On the other hand, the ablation of AP2µ2 did not seem to trigger major changes on the transcriptional level at 4 wpT. The heat map comparing all four genotypes revealed a clear cluster of the three Dnm2iko(R) mice which is massively different from all the other mice, which among themselves show a similar pattern of (Figure 24, A). A volcano plot comparing fold change and false discovery rate (FDR) identified 2928 significantly increased and 1540 significantly decreased mRNAs in Dnm2iko(R) mice at 4 wpT (Figure 24, B). A similar volcano plot comparing AP2iko(R) with CtrAP2(R) mice identified only 7 significantly increased mRNAs in AP2iko(R) mice 4 wpT (Figure 24, C). To validate the datasets, we compared the changes in differentiation and dedifferentiation markers of the Dnm2iko(R)-dataset with qRT-PCR results acquired previously with different samples (Figure 10, A). All markers were similarly regulated in both datasets. Since in the case of AP2µ2 ablation we only found 7 hits, we performed qRT-PCR for these (Figure 26, B, Appendix). When we analysed the same samples used for deep sequencing, we obtained comparable results (Figure 26, C, Appendix). However, using other samples from mice at 6 wpT and ½ ypT, none of the changes in these 7 mRNAs could be validated (Figure 26, D and E, Appendix). Trying to organize the 4468 genes significantly changed in Dnm2iko(R) compared to CtrDnm2(R), we used (GO). The majority of upregulated transcripts are involved in proliferation or inflammation, while the majority of downregulated transcripts are involved in biosynthesis of different lipids and small molecules (Figure 24, D and E). We were not able to extract any particular pathway that might be changed as a result of Dnm2 depletion. We further compared our dataset with another dataset comparing crushed and contralateral nerves of wildtype mice, which should display the early events after a sciatic nerve injury. This crush-dataset was generated by comparing crushed sciatic nerves, 3 days post crush (dpc), with the respective contralateral (uncrushed) nerves. The majority of transcripts in the Dnm2iko(R) dataset were similarly regulated in the crush-dataset. This shows that the transcriptional changes occurring around the onset of demyelination are surprisingly similar between our mouse model, involving loss of dynamin 2 and those seen upon Wallerian degeneration occurring after injury (Figure 24, F).

46 Results

Figure 24 ) RNA-Sequencing of Dnm2iko(R) and AP2iko(R) mice RNA of sciatic nerves of CtrDnm2(R) and Dnm2iko(R) mice, as well as from CtrAP2(R) and AP2iko(R) mice, all at 4 weeks post-tamoxifen (wpT) has been analyed. A) Heat map of differentially expressed genes. B) Volcano plot of RNA-sequencing results for Dnm2iko(R) in comaparison to CtrDnm2(R). Upregulated genes are depicted in red and downregulated genes in blue. Cut-off: p-value <0.05, fold change > 1,5. C) Volcano plot of RNA-sequencing results of AP2iko(R) in comparison to CtrAP2(R). Only 7 genes have been differently expressed. Cut-off: false discovery rate (FDR) <0.05, fold change > 1,5. D,E) Gene ontology (GO) analysis of differentially expressed (upregulated in D, downregulated in E) transcripts in Dnm2iko(R) compared to CtrDnm2(R). The majority of upregulated transcripts is involved in proliferation or inflammation, while the majority of downregulated transcripts is involved in biosynthesis of different lipids and small molecules. F) Venn diagram comparing the up- or down-regulated genes of the Dnm2iko(R)/CtrDnm2(R)-dataset with those in the dataset for crushed and contralateral nerves. The crush-dataset is the result of comparing crushed sciatic nerves, 3 days post crush (dpc), with the respective contralateral (uncrushed) nerve. The majority of transcripts in the the Dnm2iko(R)/CtrDnm2(R)-dataset is similarly regulated in the crushed nerves.

47 Results

48 Discussion

3 Discussion

For nearly 150 years people have worked to understand the pathomechanisms underlying Charcot- Marie-Tooth disease, the most common inherited peripheral neuropathy. Up to now, more than 60 CMT-associated genes have been described. Here, we try to shed light on the physiological role of one of these genes, dynamin 2 (DNM2), which has been described to cause either the dominant intermediate CMT Type B (CMTDIB) or the axonal form of CMT (CMT2M). Dynamin 2 (Dnm2) is a large GTPase which has been described to be involved in various cellular functions such as different forms of endocytosis, intracellular membrane trafficking and cytoskeletal regulation [113]. In vitro studies suggest that impaired clathrin-mediated endocytosis is a major contributing factor to the disease mechanism [114]. In the present study, we ablated dynamin 2 in adult mice in approximately 70 % of all the Schwann cells. We found that Dnm2 is crucial for Schwann cell survival, and that its ablation leads to demyelination and apoptosis of the affected cells within 6 weeks after recombination. Remarkably, the remaining 30 % of the Schwann cells have the potential to remyelinate the entire sciatic nerve, which leads to a fast and permanent recovery of the affected mice. In an attempt to pinpoint the observed phenotype to an impaired Clathrin-mediated endocytosis, we additionally generated Schwann cell-specific inducible knockout mice for AP2µ2. AP2µ2 is a crucial subunit of the adaptor protein complex 2 (AP-2) [184]. AP-2 was described to be absolutely required for, and exclusively involved in CME [130]. In contrast to our expectations, those mice did not develop a strong clinical phenotype 6 weeks after recombination as observed in mice upon Dnm2-depletion. However, AP2µ2-depletion led to a late onset peripheral neuropathy, which is in many ways comparable to CMT. In a global approach we identified the Transferrin receptor (TfR) to be accumulated on the surface of Dnm2- and AP2µ2-depleted cells. Proper TfR-uptake is important for cellular iron homeostasis [187]. Analysing the iron status of the sciatic nerves, we detected an iron deficiency in Dnm2iko and in AP2iko mice around their corresponding onset of the pathology. In vitro and in vivo experiments, in which Dnm2-deficient Schwann cells have been supplemented with iron, suggested that iron deficiency contributes to the observed phenotype.

49 Discussion

3.1 SC-Specific Ablation of Dynamin 2 in Adult Mice Results in a Remitting Neuropathy

To study the importance of dynamin 2 for myelin maintenance, we generated Dnm2fl/fl mice which were crossed with P0CreERT2+ mice. This system allowed us to ablate Dnm2 specifically in Schwann cells of adult mice [188]. In order to trace single Schwann cells, we bred our mice additionally with RosaYFP- reporter mice, which upon recombination leads to the expression of cytosolic YFP [177]. Four weeks after tamoxifen treatment, the Dnm2 protein levels in sciatic nerves of Dnm2iko dropped by half compared to control mice. The only partial reduction can be explained by two contributing factors. First, not all the Schwann cells recombine. Second, Dnm2 is ubiquitously expressed, leading to the detection of proteins with different cellular origin, such as from fibroblasts and axons [93]. Between 4 and 5 weeks post-tamoxifen (wpT), all mice that had lost Dnm2 started to show signs of a peripheral neuropathy. At the onset of the clinical phenotype, Dnm2iko mice started to show impaired righting movements and ataxic gait. This phenotype drastically worsened and developed into a moderate parapaersis around 6 wpT, the time point of maximum impairment. Surprisingly, all Dnm2iko mice improved rapidly thereafter and reached a normal level of performance between 7 and 8 wpT. This development looks remarkably similar to the monophasic disease progression during the acute inflammatory demyelinating polyneuropathy (AIDP), which is the most common subtype of Guillan-Barré syndrome (GBS) [82]. AIDP reaches a maximum of severity within 4 weeks after onset, followed by a plateau phase and a subsequent recovery [86]. During the AIDP, macrophages invade the PNS and attack intact myelin, resulting in demyelination. However, as discussed in more detail later, in our mouse model the inflammation seems to be a response to the Schwann-cell autonomous demyelination. The clinical observations were underlined with electrophysiological measurements. At 6 wpT, the time point of the most severe impairment, the motor nerve conduction velocity (MCV) decreased by two- third in Dnm2iko compared to CtrDnm2 mice. While applying a proximal stimulus, only a weak and disperse signal could be recorded. At 14 wpT the MCV was restored and was still unimpaired at 41 wpT, the last time point of analysis. The presence of a significant reduction of the MCV is considered one of the electrodiagnostic hallmarks of demyelinating neuropathies [189].

The morphological analysis of sciatic nerves from Dnm2iko mice confirmed the indicated remitting demyelinating neuropathy. We analysed sciatic nerves of Dnm2iko and CtrDnm2 mice prior to the impairment, at multiple time points during the course of the neuropathy as well as after recovery. Prior to the clinical onset of the phenotype, at 4 wpT, no morphological differences could be observed. At 5 wpT the first Schwann cells in sciatic nerves of Dnm2iko and mice showed indications of stress, such as swollen cytoplasm and an accumulation of organelles [190, 191]. At the peak of the clinical phenotype, at 6 wpT, many non-myelinated and rarely already remyelinated axons, as well as immune cells could be observed. The demyelination affected Schwann cells independently of the calibre of the axon they

50 Discussion

previously myelinated. In order to assess whether Schwann cells myelinating sensory and motor axons are similarly affected, we analysed the dorsal and ventral roots. During the peak of the phenotype, both roots showed signs of de- and remyelination comparable to those observed in the sciatic nerves. Even though on a morphological level, the demyelination in sciatic nerves was only obvious at 6 wpT, already at 4 wpT transcriptional changes could be detected in sciatic nerves of Dnm2iko mice. The mRNA levels of myelin proteins were reduced and markers for dedifferentiated Schwann cells as well as for proliferation were increased. In line with the increased mRNA levels of cyclin D1, we detected a slight increase in proliferation at 4 wpT, which increased to a substantial amount at 6 wpT. In order for a Schwann cell to proliferate, it first had to dedifferentiate and demyelinate. Looking at the protein levels for dedifferentiation markers, we could detect an increase in p75, a well-established marker for immature, dedifferentiated as well as non-myelinating Schwann cells, at 6 wpT [19]. Interestingly, we found recombined as well as non-recombined Schwann cells that were expressing p75, thus being in a dedifferentiated state. In line with this, we found proliferating recombined as well as non-recombined Schwann cells. The reason for the dedifferentiation and proliferation of healthy, non-recombined Schwann cells became obvious when we had a look at the recovered sciatic nerve. At 14 wpT, we could detect nearly no remaining recombined cell. A co-staining for YFP and dynamin 2 further confirmed that the recombined cells at 4 wpT had lost Dnm2 and that the remyelinating Schwann cells found at 14 wpT in sciatic nerves of Dnm2iko(R) mice were non-recombined and Dnm2 expressing. This observation demonstrates the remarkable plasticity of the peripheral nervous system. Here we show for the first time that 30 % of the Schwann cells are sufficient to compensate for the loss of the other 70 %. However, at this point we cannot rule out a contribution of non-myelinating Schwann cells to the proliferation and repopulation of the sciatic nerve. Nevertheless, we assume that the healthy myelinating Schwann cells somehow sense the loss of the neighbouring cells, which triggers their demyelination. This demyelination is part of the dedifferentiation process which is required in order to be able to proliferate. It would be of great interest to find the cues which command the healthy Schwann cell to dedifferentiate to an immature Schwann cell-like state in order to proliferate and compensate for the loss of other Schwann cells. In an attempt to further understand the observed de- and redifferentiation process, we analysed the protein levels of c-Jun, an important negative regulator of myelination [23]. We started to detect a strong increase in the levels of c-Jun at 5 wpT, which peaked at 6 wpT and decreased thereafter. Recent studies described the activation of Erk1/2 as a potential upstream activator of c-Jun [26]. Indeed, also in Schwann cells from sciatic nerves of Dnm2iko(R) mice, the levels of P-Erk1/2 increased prior to the total levels of c-Jun. We further analysed the recovery of Dnm2iko mice on a morphological level on sciatic nerves cross- sections. Already 2 weeks after the time point of maximum impairment, all axons were engulfed by

51 Discussion remyelinating Schwann cells. Remyelination proceeded rapidly, and reached almost normal levels at 14 wpT. To see, whether the observed slight hypomyelination would catch up, we analysed mice 41 wpT. Also at this late time point the slight hypomyelination was still present, which is in line with what has been described previously after a remyelination event following Wallerian degeneration. Next to an alteration in myelin thickness, the internodes have been described to be shorter after remyelination [192]. This was also observed in our mutants, comparing teased osmicated fibres of remyelinated sciatic nerves from Dnm2iko with those of CtrDnm2 mice at 14 wpT. To rule out an axonal loss during the course of this severe phenotype, we analysed the gastrocnemius muscle of Dnm2iko and CtrDnm2 mice after recovery at 14 wpT. In case of axonal degeneration followed by reinnervation of the muscle, one would expect to see fiber-type regrouping [193]. However, muscles of Dnm2iko mice did not show apparent differences compared to those of CtrDnm2 mice, thus rendering the possibility of axonal loss unlikely.

The observed de- and remyelination was accompanied by a transient influx of immune cells. We observed a large increase in the number of macrophages, mast cells, neutrophils, and T cells, all of which have been shown to be recruited into nerves following injury [194]. The most obvious ones were the highly phagocytic macrophages containing vesicles filled with myelin debris. It is well established that during Wallerian degeneration, macrophages are required to clear the axonal and myelin debris in order for the axons to regrow and the Schwann cells to remyelinate [31, 195]. During other forms of demyelination, such as in Guillain-Barré syndrome and in some CMT mouse models, macrophages have been described to be functionally related to myelin damage [196, 197]. In the acute inflammatory demyelinating polyneuropathy (AIDP), investigated by using the mouse model for experimental allergic neuritis (EAN), macrophages have been described to strip myelin and impair nerve function through a mechanism involving secretion of TNF-α and nitric oxide [198-200]. Making use of various CMT models (CMT1A, CMT1X and CMT1B), the group around Rudolf Martini described the activation of macrophages to be a common pathomechanism [201]. They characterized two important cytokine pathways, CCL2 and CSF1, to be involved in macrophage activation. Expressed by mutant Schwann cells, and induced downstream of the Erk-signaling pathway, CCL2 has been identified as a component to attract and activate pathogenic macrophages in peripheral nerves in CMT1 mouse models [29, 202-204]. In contrast to CCL2, CSF-1 has been suggested not to be expressed by Schwann cells, but rather by endoneurial fibroblasts [205]. By analysing the mRNA levels of those two and other additional cytokines, we made the surprising finding that at 4 wpT, prior to the observed invasion of macrophages, exclusively CCL2 was upregulated. We showed that Schwann cells produce CCL2 upon dynamin 2 depletion possibly downstream of the Erk-signaling pathway, as it was previously suggested in different CMT1 models. In contrast to the CMT1X model, we never observed a tight association of

52 Discussion

fibroblasts with the phagocytic macrophages [205]. We hypothesised that the Schwann cells, which at this time point are starting to dedifferentiate, secrete CCL2 that will attract macrophages in order to support the clearance of the shortly accruing myelin debris. We could strengthen this hypothesis, by demonstrating that the infiltrating macrophages are beneficial for the recovery after demyelination in our mouse model. This clodronate-mediated macrophage-deletion experiment showed that the onset of the phenotype is independent of the number of present macrophages, strongly supporting the point that the macrophages are not directly responsible for the phenotype. On the other hand, the delayed recovery associated with lack of macrophages indicates that they exert beneficial effects in the process of de- and remyelination upon Dnm2 depletion. A genetic depletion of the macrophages might further strengthen these points, since in our chemical-depletion approach we were only able to decrease the number of macrophages by half.

In summary, the ablation of dynamin 2 in adult Schwann cells leads to an acute de- and remyelination in which Schwann cells coordinate a beneficial inflammatory response.

Up to now, there is only one other mouse model presenting a phenotype comparable to the one observed in Dnm2iko mice, namely a tamoxifen-inducible, Schwann-cell specific Raf-transgene, in which tamoxifen activates the ERK-signaling pathway in adult myelinating Schwann cells [26]. These mice (in the following termed RafTR) have a temporally hyperactivated Erk1/2, which returns to normal levels within two weeks. Ten days after tamoxifen injection, RafTR mice show clinical hallmarks comparable to those observed in Dnm2iko mice at 6 wpT. As with Dnm2iko mice, RafTR mice recover at least on a physiological level within 3 weeks after the onset of the phenotype. Compared to the recovered sciatic nerve of Dnm2iko mice, RafTR mice still showed obvious myelin aberrations 90 days post-tamoxifen, which is a comparable time point to 14 wpT in our system. Additionally, axons appeared to have thicker myelin sheaths and the ratio of axon diameter to myelin thickness showed a greater heterogeneity. Unfortunately, the authors did not follow this up in the presented manuscript. As we observe in our Dnm2iko mice an almost perfect reconstitution of the sciatic nerve by healthy non-recombined Schwann cells, a plausible explanation for their observed phenotype would be that the temporal hyperactivation of the Erk-pathway leads to some downstream molecular changes which persist after remyelination by the same Schwann cells and subsequently lead to myelin abnormalities.

53 Discussion

3.2 SC-Specific Ablation of AP2µ2 in Adult Mice Results in a Late Onset Neuropathy

Dynamin 2 has been described to be involved in various cellular functions such as different forms of endocytosis, intracellular membrane trafficking and cytoskeletal regulation [113]. However, in vitro studies conducted in our laboratory suggested that impaired clathrin-mediated endocytosis is a major contribution factor to the disease mechanism in CMT [114]. In these previous studies, the authors transfected RT4 cells (a rat schwannoma cell line) with various forms of Dnm2 carrying point mutations which were described to cause either Charcot-Marie-Tooth disease or centronuclear myopathy. Making use of a transferrin-uptake assay, they showed that only Dnm2 CMT- but not CNM-associated disease mutants inhibited clathrin-mediated endocytosis [114]. In an attempt to pinpoint the phenotype observed in Dnm2iko mice to an impairment in CME, we generated Schwann cell-specific inducible knockout mice for AP2µ2. AP2µ2 is a crucial subunit of the adaptor protein complex 2 (AP-2), which has been described to be absolutely required for, and exclusively involved in CME [184, 130]. We used the same breeding setup as previously described for the Dnm2 deletion, crossing AP2µ2fl/fl with P0CreERT2+ and with RosaYFP+-reporter mice. Six weeks after tamoxifen injection, we could detect a partial loss of protein in AP2iko mice in whole sciatic nerve lysates. To obtain a cellular resolution we stained for the active, phosphorylated form of AP2µ2. As it was the case in the Dnm2iko mice, we observed a recombination frequency of approximately 70 %. Next to the 30 % non-recombined Schwann cells, there were other cells, presumably fibroblasts which had even higher levels of active AP2µ2 and could therefore account for a major portion of the residual protein detected by Western blotting. Surprisingly, AP2iko mice did not develop a clinical phenotype at 6 wpT, as we would have expected if the impaired CME was the main contributor to the phenotype observed in the Dnm2iko mice. We monitored these mice up to one year after tamoxifen injection without observing an obvious clinical phenotype. In order to be able to detect smaller alterations, we decided to characterize the gait of AP2iko mice at 1 ypT as well as at an intermediate time point (½ ypT) using the quantitative gait analysis system, CatWalk XT. At ½ ypT we found no detectable alteration in all of the analysed parameters. However, at 1 ypT we saw a tendency to a reduction in stride length. A reduction in stride length has been described in various peripheral neuropathies, such as diabetic peripheral neuropathy or CMT [206, 207]. Furthermore, an increase in the duty cycle and the support was detected, which has been described to be a compensatory response to impaired balance [208]. On the other hand, looking at the base of support (BOS), which is also often associated with impaired balance, we could not detect alterations in AP2iko mice 1 ypT. Since in the literature the BOS has been described to be increased in some but decreased in other injury models, we did not scrutinize this further [209-212]. Having a closer look at the single paw prints, we observed an increase in the print area as well as in the mean intensity. Those changes have been previously described as a consequence of a partial paralysis of the lower limbs [213]. Overall, the

54 Discussion

aged AP2iko mice develop abnormal gait and locomotion, which is usually associated with a defective peripheral nervous system. Subsequent morphological analysis confirmed this assumption. At 6 wpT, sciatic nerves of AP2iko(R) mice were indistinguishable from those of CtrAP2 mice. At ½ ypT, mutant mice started to show the first myelin aberrations. These aberrations were more pronounced and appeared with a higher frequency in aged AP2iko mice at 1 ypT. Aberrant myelin features included de- and remyelination, and myelin outfoldings. The de-and remyelination is in line with the increased g-ratio and the reduction in myelin proteins. Furthermore, markers for dedifferentiated Schwann cells, such as c-Jun, p75, Krox24 and cyclin D1, were upregulated in sciatic nerves of AP2iko mice at 1 ypT. Interestingly we found exclusively a significant increase in the number myelin outfoldings, while the number of infoldings and tomacula was not significantly changed in sciatic nerves of AP2iko compared to CtrAP2 mice. Myelin outfoldings, and sometimes infoldings, have been found in several types of inherited demyelinating neuropathies, such as CMT1B [214-216], CMT4A [217] and CMT4B [218, 219], as well as in various animal models of demyelinating CMT [108, 220-224]. It has been suggested that these shared pathological features in such a diverse group of disorders may represent a convergent pathogenic mechanism leading to a demyelinating peripheral neuropathy. In the literature the distinction between in- and outfoldings is not always made, while they are often grouped together as myelin aberrations. There are some studies describing the appearance of only myelin infoldings. For example, the inactivation of β4 was shown to cause an increase in myelin infoldings in aged mice [225]. Also in a murine CMT1C model (Litaf W166G) the authors described myelin infoldings as the exclusive myelin aberrations [226]. Taking into consideration our findings, in- and outfoldings do not necessarily emerge together, which makes a distinction relevant. The fact that the two structures can be present independently of each other suggests that they have a different origin. Additionally to the myelin aberrations, we observed onion bulb formation, immune cells, as well as a few degenerating axons. Onion bulbs are structures formed by concentric layers of Schwann cells surrounding either a naked or a myelinated axon [227]. The formation of onion bulbs is one of the histological hallmarks of demyelinating Charcot-Marie-Tooth disease, but has also been described in other hereditary neuropathies (Dejerine-Sottas disease, Refsum disease), in diabetic neuropathies, and in chronic inflammatory demyelinating neuropathies [227, 228]. Also the classical rodent-models for demyelinating CMT show onion bulb formation [71]. Since the presence of onion bulbs usually goes in hand with an increased number of Schwann cells, we assessed this point in more detail. Indeed, we found a 50 % increase in the number of Schwann cells in sciatic nerves of AP2iko mice at 1 ypT. We identified the surplus of cells to be non-recombined cells. We detected a small fraction of proliferative and an even smaller fraction of apoptotic Schwann cells in

55 Discussion sciatic nerves of AP2iko mice at 1 ypT. However, the total number of recombined cells did not change over time in AP2iko mice, demonstrating that the loss of AP2µ2 does not necessarily lead to cell death. As in the Dnm2iko mice, we also observed an increase in immune cells in the AP2iko mice during the process of de- and remyelination. At this point we hypothesized that the inflammation is a byproduct of the demyelination. To assess this further, we analysed the expression of Schwann cell-secreted CCL2 and fibroblast-secreted CSF1, the two best described cytokines in the process of demyelination in the PNS [29, 202-204]. We detected only an increase in CCL2, suggesting that the inflammation is initiated by the demyelinating Schwann cells themselves in order to support the myelin clearance. In line with this idea, we never observed a macrophage penetrating the basal lamina of a healthy Schwann cell with an intact myelin sheath. In contrast to the Dnm2iko mice, we observed slight axonal loss in aged AP2iko mice. Secondary axonal loss often occurs in severe or chronic demyelinating lesions [229]. Thus it has been described to be a typical histopathological feature in patients suffering from the demyelinating CMT1 [64]. Also the classical rodent models for CMT show secondary axonal loss [71].

In summary, the ablation of AP2µ2 in adult Schwann cells leads to a late onset peripheral neuropathy, with remarkable similarities to Charcot-Marie-Tooth disease. Therefore it might be interesting to analyse those mice on an electrophysiological level, since this is one of the main diagnostic criteria for CMT. Since AP2iko mice mimic the demyelinating form of CMT, we would expect to record a clearly reduced motor nerve conduction velocity.

We generated the AP2iko mice in order to get a better understanding of the phenotype observed in Dnm2iko mice. However, these two mice do not have much in common. Dnm2iko mice develop an acute transient neuropathy due to the loss of 70 % of Schwann cells, followed by fast and nearly complete recovery. The observed phenotype is in some ways similar to the acute inflammatory demyelinating polyneuropathy (AIDP), the most common form of Guillain-Barre syndrome (GBS). While the clinical phenotype of Dnm2iko mice mimics that found in AIDP, the underlying cell-biological origin is different. In contrast to the situation in AIDP, in Dnm2iko mice, the invading immune cells are not the cause for the observed demyelination, but only a byproduct in order to help clearing the accumulating myelin debris. On the other hand, AP2iko mice develop a late onset chronic peripheral neuropathy, perfectly mimicking Charcot-Marie-Tooth disease. These mice show all the classical hallmarks of CMT, such as de- and remyelination, onion-bulb formation and secondary axonal loss. Biology seems to have a sense of irony. We tried to understand the biological function of the CMT- related protein dynamin 2. Hypothesising that impaired clathrin-mediated endocytosis might be the cause of the phenotype; we generated a new mouse model ablating the µ-subunit of AP-2.

56 Discussion

Surprisingly by ablating AP2µ2 in Schwann cells of the adult peripheral nervous system, we generated a mouse model perfectly mimicking CMT1. This was unexpected, since the mouse model for the CMT- associated protein Dnm2 showed a phenotype comparable to AIDP rather than CMT, while AP2µ2 is a protein which has never been described in any form of neuropathy.

3.3 Iron Deficiency is a Potential Contributor to the Observed Phenotypes

By ablating dynamin 2 or AP2µ2, we tamper with different endocytic pathways, which in turn should lead to a difference in the cell surfaceome. While the ablation of AP2µ2 exclusively impairs clathrin- mediated endocytosis, the ablation of Dnm2 inhibits additionally several different forms of endocytosis. In order to investigate the resulting changes on the cell surface, we generated primary rat Schwann cells expressing different shRNAs leading to the knockdown of Dnm2 or AP2µ2, respectively. To functionally assess a reduction in endocytosis, we made use of the previously described transferrin uptake assay, which uses the uptake of labelled transferrin as a readout for CME [114]. Clathrin-mediated endocytosis was impaired in all of the generated Dnm2 and AP2µ2 knockdown cells to a similar extent. The subsequently performed cell surface capturing identified 405 glycoproteins on the surface of primary rat Schwann cells. Surprisingly, the only one that was similarly changed between the control and all the knockdown conditions was the transferrin receptor (TfR). On one hand it was self-evident to find an alteration in the levels of TfR, since it was also the one we used for functionally validating our cells. However, on the other hand it seemed rather unlikely for there to only be one commonly changed surface protein. This left us with three options. First, the knockdown in our cells is too weak to cause alteration in the levels of all the surface proteins while the TfR is the most susceptible one, which already reacts to minor impairments in CME. The second option considers the changes in TfR to be a downstream consequence of the partial endocytic impairment, which itself is not strong enough to cause mayor changes in the levels of cell surface proteins. Third, but most unlikely considering all the previous works and the current beliefs, Dnm2 and AP-2 are not strictly required for clathrin-mediated endocytosis [153]. The increase in the surface levels of TfR has been previously described as a response to cellular iron deficiency [230, 162, 161]. The iron regulatory mechanisms have been well characterized for many cell types, although up to the start of this project not for Schwann cells [187]. To make up for that omission, we characterized iron-deprived primary rat Schwann cells. The administration of deferoxamine (DFO), a potent iron chelator, to cultured cells mimics an acute iron deficiency. As described in many other cell types, iron deficiency induced the upregulation of TfR and the degradation of both ferritin subunits also in Schwann cells [161]. We further showed that a DFO-induced iron deficiency also leads to apoptosis of cultured Schwann cells. It has been previously shown in other cell types that DFO induces apoptosis in

57 Discussion an iron-dependent manner [161]. In line with that, recent in vivo data making use of different cell type- specific TfR knockout mice show an increased apoptosis in the affected tissues (cardiomyocytes [231], muscle [232], dopaminergic neurons [233]). Our artificial system with cultured Schwann cells nicely shows the regulation of the iron-related proteins TfR and ferritin, as well as an increase in apoptosis, as a result of a severe iron deficiency. In order to address the iron deficiency in a more physiological scenario, we treated myelinating DRG explant cultures with DFO. This led to demyelination, a feature we also observed in both of our CME-impaired mouse models. Therefore we analysed the iron status of these mice by using the protein levels of TfR and ferritin as readout. We detected an iron deficiency in Dnm2iko mice at 4 wpT as well as in AP2iko mice at ½ ypT. We additionally detected reduced iron levels in sciatic nerves of Dnm2iko mice at 4 wpT. In order to further pinpoint the observed iron deficiency to a loss of dynamin 2, we further analysed P5 mice with a Schwann cell specific Dnm2 ablation. Also these mice showed a strong iron deficiency, which was due to the loss of protein and not due to the emerging phenotype, as we could exclude by comparing the Dnm2iko to DhhCre+Dicerfl/fl mice [234]. The Dicer mutants have a similar morphological phenotype but not an iron deficiency. As a side note, recent work in our lab showed that also the Schwann cell-specific AP2µ2 ablation in development leads to an iron deficiency at P5 [235]. In order to investigate a potential disease relevance of our finding, we further analysed Dnm2wt/K562E mice, which carry a heterozygous point mutation which in humans leads to CMTDIB. The K562E mutation in Dnm2 does not lead to an iron deficiency in vivo, suggesting that it does not lead to a major impairment in CME. This stands in contrast to the results obtained by a previous study in vitro, in which Dnm2 disease mutants were overexpressed. In this study impaired CME was observed [114]. This might be explained by the gene dosage, which in the heterozygous knock-in mouse is at a physiological level, while the artificial overexpression results in a clear dominance of the mutated protein, leading to the observed impairments. Taken together, all analysed mice in which Schwann cells have an impaired CME show signs of iron deficiency. However, in contrast to our in vitro system, where massive iron deficiency was lethal, apoptosis of the recombined Schwann cells was only detected in Dnm2iko but not in AP2iko mice. This leaves us with two possible explanations. The first option takes into account the pleiotropy of dynamin 2 functions. Considering this scenario, iron deficiency is only a byproduct of impaired endocytosis, while the main problem leading to apoptosis lies elsewhere. The most likely problem would be at the cell surface, where AP-2 and thereby CME affects only a subset of surface proteins, whereas Dnm2 is involved in many additional forms of endocytosis, thereby affecting many more proteins [236]. An additional cell surface capturing experiment with complete knockout cells might further unravel these differences and possibly even suggest a potential mode of action. Clearly the correct timing would be of the essence in order to get the best possible knockout before the cells undergo apoptosis. Another

58 Discussion

option would be that the endocytosis-independent functions of Dnm2 are required for survival. As an example, it has been shown that cells with an inhibited dynamin function undergo apoptosis following cytokinesis failure [237]. Since we also observed that recombined cells proliferate in Dnm2iko mice, this might indeed be of relevance in order to understand the pathology. Unfortunately, with our current techniques we are not able to trace a single cell through all the chronological steps from dedifferentiation to apoptosis, which makes it nearly impossible to pinpoint the observed apoptosis exclusively to a defect in cytokinesis failure. Assuming that iron deficiency leads to apoptosis, one could explain the more severe impairment of the Dnm2 knockout cells by the fact that the loss of Dnm2 impairs not only CME, but also other forms of endocytosis. This, however, would suggest that some iron in AP2µ2-depleted cells can be taken up in a clathrin-independent manner, which is enough to keep the otherwise iron-deprived cell alive. It has been shown that in contrast to other receptors (e.g. EGFR and LDL receptor), the transferrin receptor is dependent on the interaction with AP-2 for its internalisation [184]. Combining these two arguments, one could speculate that there must be another way in which iron can be taken up by Schwann cells. It has been speculated that the divalent metal transporter 1 (DMT1) could play a direct role in the uptake of iron in Schwann cells [238]. However, DMT1, which is involved in the TfR-dependent uptake of iron, transports free ferrous (Fe2+) iron which under normal circumstances is not present extracellularly. Given additionally the fact that impaired endocytosis should not alter the function of ion channels per se, the possibility of an iron channel compensating for the loss of Tf-uptake is unlikely. Therefore we assume that there could be another receptor that is involved in the uptake of iron. For oligodendrocytes, Tim-2, a receptor for H-ferritin, was suggested to be involved in iron uptake [239]. However, this alternative seems unlikely for Schwann cells, since ferritin cannot cross the blood-nerve barrier and is therefore not available [240]. This leaves two possibilities for how the Schwann cell still could take up iron. First, there could be a backup mechanism for taking up the TfR. However, even though the TfR has been studied for decades, such a mechanism has never been proposed [241, 161]. It should also be noted that the uptake of the TfR is the exemplary model for CME, which makes it even more unlikely that an additional uptake mechanism would not yet have been described. The second option of how transferrin could be taken up is by binding to a different receptor, which is not strictly dependent on CME. This might also explain the different severities of the phenotype of the Dnm2iko and the AP2iko mice. Assuming such another transferrin receptor x (TfRx) exists, one could well imagine that the more severe impairments of the Dnm2iko mice is due to the fact that there additionally the uptake of this second receptor is blocked. Even though, such a TfRx has not yet been described; it would be of interest to follow this up. New techniques such as the Ligand-based receptor-capturing (LRC) would allow us to efficiently trace down a potential TfRx [161]. A similar experiment has already been conducted in the original publication by Frei and colleagues. Once more, the Tf binding to the TfR served

59 Discussion as an established model. For their experiments, the authors used a human osteosarcoma cell line in which they did not identify another potential receptor. Still, it is possible that such a receptor is not present on all cell types. Although the existence of a previously undiscovered TfRx is a matter of speculation, it still could be interesting to perform a LRC with transferrin on primary rat Schwann cells.

In order to elaborate the contribution of iron deficiency to the phenotypes of the Dnm2iko mice, we used inducible DRG-explant cultures, which allowed us to delete dynamin 2 after successful myelination in vitro. Upon recombination, we could observe a severe demyelination. These cultures demonstrated once more that the demyelination upon Dnm2 ablation is a Schwann cell autonomous process which does not require the presence of macrophages. Supplementation with iron dextran (FeDex), ameliorated the observed demyelination. In line with this, the reduction in the levels of Krox20, the master regulator of myelination, could be highly prevented. After a successful partial in vitro rescue, we tried to rescue the observed demyelination in Dnm2iko mice. We used the same compound as in vitro, FeDex, which has been previously successfully used by others to compensate for an iron deficiency [242, 243]. Even though we were able to detect a systemic iron overload, the course of the clinical phenotype could not be altered. The iron supplementation also had no effect on the morphological level. However, having a closer look at the differentiation state of our Schwann cells we saw small changes pointing in the direction of a delayed dedifferentiation. These changes observed at the mRNA level, were additionally confirmed on the protein level for p75, which is expressed by dedifferentiated Schwann cells [244]. By assessing cell death we ruled out that the reduced levels of dedifferentiation-markers are simply a result of more apoptosis resulting in fewer cells. The differences in the success of the in vitro and the in vivo rescue experiment might be explained by the different availability of the iron dextran. It has never been used to deliver iron to Schwann cells in a sciatic nerve before, and compared to most cell types in the body, Schwann cells are by far less accessible due to the blood-nerve barrier. In contrast, all cells in culture are in direct contact with the medium and therefore with the soluble FeDex, so that delivery of iron is by far more efficient. In order to link the too little iron uptake to the weak in vivo rescue, we would need to assess the iron status of the individual Schwann cells, which is not possible with our current techniques. One potential possibility to increase the sensitivity of the readout of the rescue experiment would be to perform the same experiment once more, but without sacrificing the mice at the peak of their phenotype. Having indeed less dedifferentiation should result in a slower redifferentiation and consequently in a delayed recovery.

Taken together, the observed iron deficiency might contribute to the observed phenotype in Dnm2iko mice. However, further experiments would be needed to fully understand the pathogenic role of iron deficiency in Dnm2-depleted Schwann cells.

60 Discussion

Overall, we were only able to link the dedifferentiation and not the apoptosis described in Dnm2iko mice to the observed iron deficiency. This makes sense when we compare these results with the phenotype of the AP2iko mice. In these mice, in which we also detected an iron deficiency, we only observed dedifferentiation but no obvious apoptosis. This points in the direction that iron is important for keeping a Schwann cell in a differentiated state. In line with these findings, it has previously been demonstrated that iron promotes differentiation in cultured Schwann cells [245]. One way to investigate this further would be to try to rescue the dedifferentiation-phenotype in the AP2iko mice. However, this would be difficult, since the phenotype arises only in aged mice. It would require a lifelong FeDex treatment, which might not be feasible. The best way to reinforce the link between impaired CME and impaired iron uptake resulting in iron deficiency, would be to analyse an inducible Schwann cell-specific TfR knockout mouse (TfRiko). If our hypothesis turns out to be valid, then the TfRiko mice should have a phenotype comparable to that observed in the AP2iko mice.

In an attempt to better understand the phenotypes of the Dnm2iko and the AP2iko mice on a more mechanistical level, we analysed the transcriptome of these mice at 4 wpT. Long story short, as it turned out the chosen time point was suboptimal. There was not a single transcript verifiably changed in the AP2iko(R) compared to CtrAP2(R) mice. By itself, this result is insofar surprising given that the deletion of AP2µ2 does not lead to an immediate cellular response. It might therefore be of great interest to analyse the transcriptome of AP2iko(R) mice at a later time point. Analysing the transcriptome at ½ ypT might be best suited for finding the initial changes, since morphologically the nerves of these mice show only minor alterations at this stage. Compared to no changes in the AP2iko(R) mice, the transcriptional differences in the Dnm2iko mice at 4 wpT were so massive that we were not able to pinpoint initial changes which might result from the Dnm2 depletion. Interestingly, most of the observed changes are similarly regulated in the initial phase of Wallerian degeneration. This shows that different causes initiating a demyelination process activate fairly similar downstream mechanisms.

Overall, the observed phenotype in Dnm2iko mice cannot be exclusively linked to impairments in CME. However, the inhibition of CME in Schwann cells, leads to a phenotype comparable to CMT. This would suggest that the underlying problem in CMTDIB is not a complete loss of dynamin 2 function, but rather a specific loss of CME. Our data further suggest that a resultant iron deficiency in Schwann cells might contribute to the pathology in CMTDIB. However, in a murine model for CMTDIB, in which the mice carry a heterozygous K562E point multination in the Dnm2 gene, iron deficiency was not evident, suggesting that CME is not heavily impaired. This might be different for other point mutations which are

61 Discussion associated with CMTDIB, as well as in humans carrying the K562E mutations. To this end, one should keep in mind the pleiotropic function of Dnm2, which might play a more important role in the pathology of CMTDIB than previously suspected.

62 Material and Methods

4 Material and Methods

4.1 Solutions and Buffers

Unless mentioned otherwise, deionized and sterile filtered water (ddH2O), purified by a water purifying apparatus (Milli-Q® Academic System, Millipore AG, Volketswil, Switzerland) was used.

Phosphate Buffer (PB) 1 M Na2HPO4, 0.03 M NaH2PO4, pH 7.4

EM Fix 4 % paraformaldehyde (PFA), 3 % glutaraldehyde (GA), 1 M phosphate buffer (PB), pH 7.4

Spurr 51 % nonenyl succinic anhydride (NSA) (EMS #19050), 36 % vinyl cyclohexene dioxide (ERL 4221) (EMS #15004), 12 % diglycidyl ether of polypropylene glycol (DER 736) (EMS #13000), 1 % dimethylamino- ethanol (DMAE) (EMS #13300)

Lysis Buffer (Genotyping) 25 mM NaOH, 0.2 mM ethylenediaminetetraacetic acid (EDTA), pH 12

Neutralizing Buffer 40 mM Tris-HCl, pH 5 (Genotyping)

Orange G loading dye 6x 0.25 % (w/v) Orange G, 15 % (w/v) Ficoll® PM 400 (Sigma #F4375-10G)

PN2 Lysis Buffer 25 mM Tris-HCl pH 7.4, 1 M NaCl, 2 % sodium dodecyl sulfate (SDS), 1 % proteinase inhibitors (Sigma #P8849) , 1 Tbl Phospho-Stop (per 10 mL) (Roche #04906837001)

4x Loading Dye 40 % Glycerol, 240 mM Tris-HCl pH 6.8, 8 % SDS, 0.04 % bromphenol- blue, 5 % β-mercaptoethanol

SDS Running Buffer 1 % SDS, 25 mM Trizma-Base, 190 mM glycine

Transfer Buffer 25 mM Trizma-Base, 190 mM glycine, 20 % methanol

Ponceau S 1 % acetic acid, 0.1 % Ponceau S

Tris Buffered Saline (TBS) 20 mM Tris-HCl pH 7.6, 150 mM NaCl, pH 7.6

TBS-Tween (TBS-T) TBS, 0.05 % Tween-20® (Sigma #P9416-50ML)

Blocking Buffer for 5 % non-fat dry milk in TBS-T Western Blot

AP Buffer 100 mM Tris-HCl, pH 9.5, 100 mM NaCl

Citrate Buffer 10 mM Citric acid pH 6, 0.5 % Tween-20® (Sigma #P9416-50ML)

Phosphate Buffered Saline 140 mM NaCl, 2.5 mM KCl, 6.5 mM Na2HPO4, 1.5 mM KH2PO4, pH 7.4 (PBS)

63 Material and Methods

AB Buffer 0.1 % Trition X-100, 0.1% BSA in PBS

Blocking Buffer for 10 % Donkey Serum, 90 % AB Buffer Immunostaining

Tamoxifen Solution (TMX) 20 mg/ml tamoxifen (Sigma #T5648), 10 % ethanol in sunflower oil (Sigma #S5007)

4.2 Schwann Cell-Specific Dynamin 2 Ablation in Adult Mice

In order to investigate the role of dynamin 2 in myelin maintenance, we generated mice featuring conditional Dnm2 ablation in Schwann cells upon tamoxifen administration. The generated Dnm2fl/fl mice were bred with P0CreERT2+ mice. Dnm2 was ablated by injecting 10 week-old Dnm2fl/fl P0CreERT2+ mice with tamoxifen (2 mg, i.p.) on 5 consecutive days. Tamoxifen-treated Dnm2fl/fl P0CreERT2- mice served as controls. Tamoxifen-injected Dnm2fl/fl P0CreERT2+ mice will further be referred as Dnm2iko and tamoxifen-injected Dnm2fl/fl P0CreERT2- mice as CtrDnm2. In order to investigate the fate of single Schwann cells, we bred the Dnm2fl/fl P0CreERT2+ mice with ROSAYFP+/+ mice, which upon recombination express cytosolic YFP [177]. In order to get YFP expressing controls and mutants, we administered tamoxifen (2 mg, i.p.) on 5 consecutive days to 10 week-old Dnm2fl/fl P0CreERT2+ ROSAYFP+/- and Dnm2wt/wt P0CreERT2+ ROSAYFP+/+ mice. Tamoxifen-injected Dnm2fl/fl P0CreERT2+ ROSAYFP+/- mice will further be referred as Dnm2iko(R) and tamoxifen-injected Dnm2wt/wt P0CreERT+ ROSAYFP+/- mice as CtrDnm2(R). (R)* indicates an analysis for which not all the time points consisted of the same genotype with regard to the ROSAYFP Cre-reporter.

4.3 Schwann Cell-Specific AP2µ2 Ablation in Adult Mice

In order to investigate the role of AP-2 in myelin maintenance, we generated mice featuring conditional AP2µ2 ablation in Schwann cells upon tamoxifen administration. The generated Ap2m1fl/fl mice were bred with P0CreERT2+ mice. AP2µ2 was ablated by injecting 10 week-old Ap2m1fl/fl P0CreERT2+ mice with tamoxifen (2 mg, i.p.) on 5 consecutive days. Tamoxifen-treated Ap2m1fl/fl P0CreERT2- mice served as controls. Tamoxifen-injected Ap2m1fl/fl P0CreERT2+ mice will further be referred as AP2iko and tamoxifen-injected Ap2m1fl/fl P0CreERT2- mice as CtrAP2. In order to investigate the fate of single Schwann cells, we bred the Ap2m1fl/fl P0CreERT2+ mice with ROSAYFP+/+ mice, which upon recombination express cytosolic YFP [177]. In order to get YFP expressing controls and mutants, we administered tamoxifen (2 mg, i.p.) on 5 consecutive days to 10 week-old Ap2m1fl/fl P0CreERT2+ ROSAYFP+/- and Ap2m1wt/wt P0CreERT2+ ROSAYFP+/+ mice. Tamoxifen-injected Ap2m1fl/fl P0CreERT2+ ROSAYFP+/- mice will further be referred as AP2iko(R) and tamoxifen-injected Ap2m1wt/wt P0CreERT+ ROSAYFP+/- mice as CtrAP2(R). (R)* indicates analyses for which not all the time points consisted of the same genotype.

64 Material and Methods

4.4 Genotyping

The genotypes of the mice were identified by polymerase chain reaction (PCR) on genomic DNA derived from ear tissue biopsies. The samples were incubated with 200 µl lysis buffer for 30 minutes at 99 °C and neutralized with the same amount of neutralizing buffer.

The forward and reverse primers for the wild-type (wt), floxed and ROSAYFP alleles are listed in Table 1. DNA Fragments were mixed with the corresponding PCR-Mastermix (Table 2 or Table 4) and amplified using a Thermocycler (T Professional Trio, Biometra) following the protocols of Table 3 or Table 5. The PCR amplification products were mixed with Orange G and separated in a 2 % agarose gel containing 0.001 % SYBR Green (iQ™ SYBR® Green Supermix #170-8880, Bio Rad) applying a linear electric field of 10 V/cm.

Table 1: Primers used for Genotyping Genomic Target Forward Primer (5’ – 3’) Reverse Primer (3’ – 5’) Product Size Dnm2 floxed allele GGGAATCCTGCTGGGGAAGCTCTC CTCTAGCACTTCCACTAAGCCCTCC 293 bp Dnm2 wt allele GGGAATCCTGCTGGGGAAGCTCTC CTCTAGCACTTCCACTAAGCCCTCC 243 bp Cre ATC GCC AGG CGT TTT CTG AGC ATA C GCC AGA TTA CGT ATA TCC TGG CAG C 387 bp Ap2m1 floxed allele CCA GCA CGA AGT TTT AGT CTT TGC TGG GAG GAC AAC CAA GGG ACC TAC AG 248 bp Ap2m1 wt allele CCA GCA CGA AGT TTT AGT CTT TGC TGG GAG GAC AAC CAA GGG ACC TAC AG 168 bp RosaYFP wt allele AAA GTC GCT CTG AGT TGT TAT (Primer A) GGA GCG GGA GAA ATG GAT ATG (Primer B) 500 bp RosaYFP targeted allele AAA GTC GCT CTG AGT TGT TAT (Primer A) GCG AAG AGT TTG TCC TCA ACC (Primer C) 250 bp

Table 2: PCR Mastermix Dnm2, Cre and Ap2m1 Table 3: PCR Protocol Dnm2, Cre and Ap2m1 Reagent Volume [μl] Cycles [#] Temperature [°C] Time [sec] 10X PCR Buffer 2.5 1 94 180 MgCl2 [50 mM] 1 94 60 dNTPs [10 mM each] 0.5 2 62 60

Forward Primer [10 μM] 2.5 72 60 Reverse Primer [10 μM] 2.5 94 30 Taq polymerase [5 U/μl] 0.2 30 62 30

ddH2O 13.8 72 30 DNA 2 1 72 180 1 4 ∞ total 25

Table 4: PCR Mastermix RosaYFP Table 5: PCR Protocol RosaYFP Reagent Volume [μl] Cycles [#] Temperature [°C] Time [sec] 10X PCR Buffer 3 1 94 120 94 45 MgCl2 [50 mM] 0.9 35 53 30 dNTPs [10 mM each] 0.66 72 60 Rosa A Primer [5 μM] 3 1 72 300 Rosa B Primer [5 μM] 3 1 4 ∞ Rosa C Primer [5 μM] 3 Taq polymerase [5 U/μl] 0.22

ddH2O 15.22 DNA 1 total 30

65 Material and Methods

4.5 Gait Analysis

Gait analysis was performed using the CatWalk XT automated gait analysis system (Noldus Information Technology, Wageningen, Netherlands). The system uses a 1.3 meter-long glass plate with a goal box on one end. The glass plate is illuminated by a green fluorescent light from the side. The light reflects downwards from the paws when the mouse touches the glass plate and is recorded by a high resolution camera underneath the walkway. The images from each run were converted into digital signals and processed with a threshold set at 30 arbitrary units (ranging from 0 to 225) by the CatWalk XT software (version 10.6). Only runs in which the mice were faster than 5 second and had a speed variation lower than 60 % were analyzed. Three such runs per mouse were considered for analysis. Each run was analyzed by identification and labelling of each paw print and a wide range of gait data was generated. We selected the stride length, base of support (BOS), duty cycle, support, print area and mean print intensity for our study. These parameters are commonly defined as seen in Table 6 [246].

Table 6 : Definition of Catwalk Parameters Name Definition Stance Time during which a certain paw is in contact with the glass plate Swing Time during which a certain paw is not in contact with the glass plate Step cycle Time of the stance and swing time together Stride length Distance between two consecutive paw placements of the same paw Distance between the mass-midpoints of two fore or hind paw prints at maximal contact during Base of support (BOS) each step cycle; the distance is measured perpendicularly to the direction of walking Duty cycle The proportion of the stance phase in one step cycle in % Support Time in % of the whole run where the mice have 3 or 4 paws on the glass plate at the same time Print area Total area where the paw is in contact with the glass plate during the stance Intensity at the moment where the paw is in maximal contact with the glass plate, parameter is Mean print intensity expressed in arbitrary units (a.u.)

4.6 Electron Microscopy

4.6.1 Nerve Preparation for Electron Microscopy Mice were sacrificed and sciatic nerves were removed, placed on a piece of paper (120 g/m2) and fixed overnight in EM-Fix at 4 °C. On the next day the nerves were placed in phosphate buffer and fixed with a surgical thread into small plastic baskets (Electron Microscopy Sciences), which were stacked on a screw and placed in a plastic beaker (Electron Microscopy Sciences). For the following steps, shown in Table 7, the nerves were kept in the plastic baskets and transferred into fresh beakers.

66 Material and Methods

Table 7: Preparation Protocol for Ultrathin Sections Compound Concentration Time [min] Temperature PB 0.1 M 20 RT PB 0.1 M 40 RT PB 0.1 M 80 RT

OsO4 in PB 1 % overnight 4 °C

PB 0.1 M 20 RT PB 0.1 M 40 RT PB 0.1 M 80 RT Dehydration

Acetone in ddH2O 30 % 20 RT Acetone in ddH2O 50 % 20 RT Acetone in ddH2O 70 % 20 RT Acetone in ddH2O 90 % 20 RT Acetone in ddH2O 96 % 20 RT Acetone in ddH2O 100 % 20 RT Acetone in ddH2O 100 % 20 RT Infiltration with Spurr Spurr : Acetone 1:2 60 RT Spurr : Acetone 1:1 60 RT Spurr : Acetone 2:1 60 RT Spurr pure overnight RT

Freshly prepared Spurr pure 120 RT

Afterwards the nerves were embedded in fresh Spurr in a flat embedding mold (Electron Microscopy Sciences) and hardened in the oven at 65 °C overnight.

4.6.2 Sample Preparation and Electron Microscopy Before cutting the sample blocks were trimmed with a rasp to create a narrow, trapezoidal surface. The EM blocks containing the nerve samples were cut into sections with a layer thickness of 65 nm with a diamond knife (Ultra 35°, Diatome) using an ultramicrotome (Ultracut E, Reichert Jung) and harvested in ultrapure water. The sections were then transferred onto ITO coverslips (Optics Balzers) and whole panoramas of the nerve sections were imaged by a FEG scanning electron microscope (, Zeiss) attached to ATLAS modules (Zeiss).

4.7 Immunological Methods

4.7.1 Nerve Preparation for Immunoblotting Mice were sacrificed, and sciatic nerves were dissected and separated from epi- and perineurium. Dissected sciatic nerves were snap-frozen in liquid nitrogen. Subsequently, the nerves were ground with a pre-cooled EPPI-Pistil (VWR, #431-0094) on liquid nitrogen. Then 75 µl of PN2 lysis buffer was added and homogenized with the nerves. Homogenate was boiled for 5 min at 95 °C. After 15 min of centrifugation at 17’000 g, the supernatant was transferred to a new Eppendorf tube and stored at -80 °C.

67 Material and Methods

4.7.2 Protein Gel Electrophoresis For the experiments precast polyacrylamide gels with a 4 – 15 % gradient (Mini-PROTEAN® TGX™ Gel, 15 well, 15 µl, Bio-Rad #456-1086) were used. The gel electrophoresis chamber (Mini-PROTEAN® Tetra Vertical Electrophoresis Cell, Bio-Rad #1658004) was assembled, placed in the container and filled with SDS-running buffer. The protein samples were mixed with 4X Loading dye, heated for 5 min at 95 °C and loaded into the wells of the gel. Precision Plus ProteinTM standard (Bio-Rad Laboratories, Inc., Hercules, CA, USA) was used as a molecular size marker. A current of 20 mA per gel was applied until the bromophenol blue marker dye reached the bottom of the gel.

4.7.3 Protein Transfer to PVDF Membrane With this method the electrophoretically separated proteins are transferred to a PVDF membrane (pore size 0.45 µm, Millipore Corporation, Bedford, USA) due to an electric field, which is oriented vertically to the gel and membrane surface. Per blot two sponges and six pieces of 3 mm thick Whatman chromatography paper (Sigma #3030917) were pre-soaked in transfer buffer and the PVDF membrane was activated by putting it in methanol for 1 min and afterwards in ddH2O for 5 min. The different components were arranged in the gel holder cassette from bottom to top in the following order: one sponge, three pieces of Whatman paper, SDS-gel, PVDF membrane, three pieces of Whatman paper and one sponge. Afterwards the blot basket was fixed and positioned in a chamber filled with transfer buffer and a cooling element. The transfer was performed at 90 V for 1 hour at 4 °C.

4.7.4 Immunoblotting Immunoblotting is a method that enables the specific and sensitive detection of a protein of interest. Protein samples are first separated by gel electrophoresis (4.7.2) transferred to a PVDF membrane (4.7.3) and afterwards incubated with a specific directed against the protein of interest. The PVDF membrane was incubated in blocking buffer for 1 hour at room temperature to block unspecific binding sites. The primary antibody (Table 8) or a combination of different antibodies was diluted in 5 % BSA in TBS-T and incubated on the membrane overnight at 4 °C. After washing the membrane three times for 5 min with TBS-T, the membrane was incubated for 1 hour at room temperature with the secondary antibody (Table 9) in diluted blocking buffer (1:5 in TBS-T). Finally, the membrane was washed three times for 5 min with TBS-T. If the secondary antibody was conjugated to alkaline phosphatase (AP), the membrane was additionally washed for 5 min with AP buffer prior to detection. To detect the AP conjugate, 1 ml of AP buffer containing 10 µl CDP-Star (Roche Diagnostics GmbH, Mannheim, Germany) was used as chemiluminescence substrate. To detect the horseradish peroxidase (HRP) conjugate, an equal volume of detection reagent A and B (Amersham™ ECL Prime Western Blotting Detection Reagent, GE

68 Material and Methods

Healthcare) was used as a chemiluminescence substrate. The Fusion FX7 image acquisition system (VILBER LOURMAT Deutschland GmbH) was used to detect the signal.

Represented Western blot bands have been adjusted in their orientations according to the contour of the running gel. Squares indicating the rotation are drawn on the full-length blots.

Table 8: Primary Antibodies used for Immunoblots Antibody Species Dilution Company Catalogue Number Albumin rabbit 1:1000 Cell Signaling 4929 AP2µ2 mouse 1:500 BD 611351 α-Tubulin mouse 1:1000 Sigma T5168 c-Jun rabbit 1:1000 Cell Signaling 2315 Cleaved caspase 3 rabbit 1:1000 Cell Signaling 9664 Dynamin 2 rabbit 1:1000 Pineda - Erk1/2 rabbit 1:1000 Cell Signaling 9102 Ferritin H rabbit 1:1000 Cell Signaling 3998 Ferritin L rabbit 1:1000 Abcam ab69090 GAPDH mouse 1:10000 Hytest Ltd 5G4 Krox20 rabbit 1:1000 Dies Meier - MAG rabbit 1:1000 Invitrogen 34-6200 MBP rat 1:1000 Serotec MCA409s P0 chicken 1:1000 Millipore ab9352 p38 rabbit 1:1000 Cell Signaling 9212 p75 rabbit 1:1000 Millipore ab1554 P-Erk1/2Thr202/Tyr204 mouse 1:1000 Cell Signaling 9106 P-p38 rabbit 1:1000 Cell Signaling 4631 Sox10 goat 1:500 R&D AF2864 Transferrin receptor mouse 1:500 Invitrogen 13-6800

Table 9: Secondary Antibodies used for Immunoblots Antibody Conjugate Species Dilution Company anti-mouse HRP goat 1:10000 Jackson Immuno Research anti-goat HRP donkey 1:10000 Jackson Immuno Research anti-rabbit HRP goat 1:10000 Jackson Immuno Research anti-chicken HRP goat 1:10000 Jackson Immuno Research anti-rat HRP goat 1:10000 Jackson Immuno Research anti-mouse AP goat 1:10000 Jackson Immuno Research

4.7.5 ELISA The ELISA for CCL2 was performed according to the manufacturer´s protocol (R&D, #MJE00).

69 Material and Methods

4.8 Histological Methods

4.8.1 Nerve Preparation and Cryosectioning Mice were sacrificed, the sciatic nerves were removed, placed on a piece of paper (120 g/m²) and fixed for 1 h in 4 % PFA at 4 °C. Subsequently, the nerves were washed in PBS. Afterwards the nerves were kept in 10 % sucrose for one hour, followed by 20 % sucrose over night at 4 °C. The nerves were placed vertically into a plastic embedding mold (T-8 Polysciences, Inc., U.S.) filled with OCT compound (Tissue- Tek, Sakura Finetek, Torrance CA, USA). The mold was put on an aluminium block surrounded by liquid nitrogen to freeze the OCT. The embedded nerves were stored at -80 °C. The frozen blocks were cut in sections with a layer thickness of 10 µm using a Cryostat (Microm HM 560, Thermo Fisher Scientific Inc., Walldorf Deutschland) and harvested on microscope slides (Menzel-Gläser, SUPERFROST® PLUS, Thermo Fisher Scientific). The slides were stored at -80 °C until further processing.

4.8.2 Immunohistochemistry To remove the OCT, the slides containing the cryo-sections (4.8.1) were washed for 10 min in PBS at room temperature. If required an antigen retrieval step was performed prior to permeabilization. In this case, the sections were post-fixed with 4 % PFA for 10 min, washed twice for 5 min in PBS and incubated in citrate buffer for 15 min at 95°C in a water bath. Afterwards samples were allowed to cool for 20 min at room temperature, and then washed three times for 5 min in PBS at room temperature. Samples free of OCT as well as the antigen-retrieved samples were permeabilized with 0.5 % (v/v) Triton X-100 in PBS for 20 min. After washing the samples twice with PBS for 5 min, they were blocked for 1 h at room temperature with blocking buffer in a wet chamber. Afterwards the samples were incubated with the primary antibody (Table 10) in AB buffer overnight at 4 °C in a wet chamber. On the next day, they were washed three times with PBS for 5 min and incubated for 45 min at room temperature with the secondary antibody (Table 11) in AB buffer in a wet chamber. The slides were washed once with PBS for 5 min and incubated with PBS containing 5 µg/ml 4’,6’-diamidino-2-phenylindole hydrochloride (DAPI; Molecular Probes, Basel, Switzerland) for 5 min. The samples were washed another two times with PBS for 5 min before mounting with Vecta Shield (Vector Laboratories, San Mateo, California). Finally, coverslips were sealed with transparent nail polish and dried for 1 hour at room temperature before imaging with a fluorescence microscope (Zeiss Axio Imager M2). Unless mentioned differently, one section per animal and three animals per genotype and/or condition were analysed.

70 Material and Methods

Table 10: Primary Antibodies used for Immunohistochemistry Antibody Species Dilution Company Catalogue Number CD3 rabbit 1:200 Dako A0452 CD68 rat 1:100 Serotec mca1957 CD117 rat 1:200 BD 562417 Dynamin 2 rabbit 1:200 GeneTex GRX109652 GFP chicken 1:1000 Aves Lab GFP-1020 GR1 rat 1:200 BD 553127 Ki-67 rabbit 1:200 abcam ab15580 MBP rat 1:200 Serotec MCA409s NF-160 mouse 1:200 Sigma N5264 p75 rabbit 1:200 Millipore ab1554 P-AP2µ2Thr156 rabbit 1:200 Cell Signaling 3843 P-Erk1/2Thr202/Tyr204 rabbit 1:1000 Cell Signaling 9101 P-H2A.X (Ser139) rabbit 1:800 Cell Signaling 2577 Sox10 goat 1:100 R&D AF2864

Table 11: Secondary Antibodies used for Immunohistochemistry Antibody Conjugate Dilution Company donkey-anti-goat Alexa Fluor 488 1:500 Invitrogen donkey-anti-rabbit Alexa Fluor 555 1:500 Life Technologies donkey-anti-rabbit Cy 3 1:500 Jackson Immuno Research donkey-anti-rat Cy 3 1:500 Jackson Immuno Research donkey-anti-chicken Alexa Fluor 647 1:500 Jackson Immuno Research goat-anti-mouse Alexa Fluor 488 1:500 Jackson Immuno Research goat-anti-rat Cy 3 1:500 Jackson Immuno Research

4.8.3 Proliferation Assay The cell proliferation in sciatic nerves was determined by EdU staining. Mice were injected with 50 mg/kg EdU (5-ethynyl-2´-deoxyuridine) one hour prior to euthanization. EdU is a nucleoside analogue of thymidine and is incorporated into DNA during active DNA synthesis. Sciatic nerves were prepared for cryosectioning and cut as in section 4.8.2. Then the samples were treated according to the manufacturer’s protocol using the Click-iT® EdU Alexa Fluor® 546 Cell Proliferation Assay Kit (Life Technologies, USA). Finally the samples were washed with PBS and further stained for Sox10, YFP and with DAPI (section 4.8.2).

4.8.4 Evens Blue Injections 200 µl of a 1% sterile solution of Evans Blue and 5% BSA (AppliChem, #A1391.0100) in PBS was injected into the tail vein 30 minutes before the animals were sacrificed.

4.8.5 Iron Staining A Prussian Blue-DAB staining was followed by a silver-gold-uranyl nitrate enhancement procedure as described in [249].

71 Material and Methods

4.8.6 Fluorescence In Situ Hybridization A pFLCI vector containing CCL2 cDNA (DNAFORM, Clone I830018H20) was linearized using EcoRI for producing the (+)-stand probe and with KpnI for the (-)-strand probe. After a phenol-chloroform extraction the pellet was resuspended in 10 µl TE-Buffer. In vitro transcription and hybridization was performed using the DIG RNA labeling mix (Roche, #11277073910) according to the manufacturers protocol’s using a T3 or a T7 RNA polymerase for (+)-strand or (-)-strand probe, respectively. For detection, the TSA Plus Fluorescence Kit (Perkin Elmer, #NEL744001KT) was used according to the manufacturers protocol’s. After the FISH protocol, an immunostaining was performed according to section 4.8.2, starting after the permeabilization step.

4.8.7 Muscle Staining Gastrocnemius muscles were extracted and frozen in liquid nitrogen-cooled 2-methylbutane. 10 µm thick cryo-sections were cut and stained with H&E and ATPase pH 4.6 in collaboration with Prof. Klaus Toyka according to his established protocol.

4.9 Quantitative Real Time – Polymerase Chain Reaction (qRT-PCR)

4.9.1 RNA Extraction Mice were sacrificed, and sciatic nerves were dissected and separated from epi- and perineurium. Dissected sciatic nerves were snap-frozen in liquid nitrogen. Subsequently, the nerves were ground with a pre-cooled EPPI-Pistill (VWR, #431-0094) on liquid nitrogen. After grinding, RNA was prepared using QIAzol Lysis Reagent (QIAGEN, #79306). The homogenized samples were incubated for 5 minutes at room temperature, after which 0.1 ml of chloroform was added to each sample and shaken vigorously by hand for 15 seconds followed by incubation for 3 minutes at room temperature. Centrifugation at 12’000 g for 15 minutes at 4 °C allows the mixture to separate into three phases; the upper aqueous phase containing RNA, the interphase containing DNA and the lower organic phase containing proteins. The aqueous, RNA-containing phase was then transferred into a new Eppendorf tube and 0.25 ml of isopropanol and 0.5 µl of GlycoBlue (Thermo Scientific, #AM9515) were added. The isopropanol lets the RNA precipitate and the GlycoBlue colours the pellet to facilitate the washing steps. After 10 minutes incubation at room temperature the samples were centrifuged at 12’000 g for 10 minutes at 4 °C. Then the supernatant was removed, leaving only the RNA pellet. The pellet was washed two times with 75 % ethanol and then resuspended in 15 µl of RNase-free water. The concentration and level of purity was measured using the NanoDrop™ Lite Spectrophotometer (Thermo Scientific).

72 Material and Methods

4.9.2 Reverse Transcription cDNA was synthesized using the Maxima First Strand cDNA Synthesis Kit for qRT-PCR (Thermo Fischer, # Ki642) according to manufacturer’s protocol. In short; RNA was mixed with reagents according to Table 12 and incubated according to Table 13.

Table 12: Reverse Transcription Mix Table 13: Reverse Transcription Incubation Program

Reagent Volume [μl] Temperature [°C] Time [min] 5x Reaction Mix 4 25 10 Maxima Enzyme Mix 2 50 15 RNA 50 ng 85 5 RNase-free H2O: fill up to 20 total 20

The synthesized cDNA was stored at -20 °C.

4.9.3 qRT-PCR Quantitative Real Time Polymerase Chain Reaction (qRT-PCR) was performed using the FastStart Essential DNA Green Master (Roche, # 06402712001) according to the manufacturer’s protocol. Forward and reverse primers for the qRT-PCR reaction were used according to Table 14.

Table 14: Primers for qRT-PCR Gene Forward Primer (5’-3’) Reverse Primer (5’-3’) β-Actin GTC CAC ACC CGC CAC C GGC CTC GTC ACC CAC ATA G c-Jun GCC AAG AAC TCG GAC CTT CTC ACG TC TGA TGT GCC CAT TGC TGG ACT GGA TG CSF-1 AGT ATT GCC AAG GAG GTG TCA G CAA TCT GGC ATG AAG TCT CCA TTT G Cyclin D1 TGT TCG TGG CCT CTA AGA TGA AG AGG TTC CAC TTG AGC TTG TTC AC GAPDH GGT GAA GGT CGG TGT GAA CG AAG GGG TCG TTG ATG GCA AC IL1b ATG GAT GCT ACC AAA CTG GAT ATA ATC CTG AAG GAC TCT GGC TTT GTC T Il-6 ATG GAT GCT ACC AAA CTG GAT ATA ATC CTG AAG GAC TCT GGC TTT GTCT Il-10 CCC TGG GTG AGA AGC TGA AG CAC TGC CTT GCT CTT ATT TTC ACA Krox20 ACA GCC TCT ACC CGG TGG AAG AC CAG AGA TGG GAG CGA AGC TAC TCG GAT A Krox24 CAG CGC CTT CAA TCC TCA AG AGC GGC CAG TAT AGG TGA TG MBP TTG GCT ACG GAG GCA GAG C GAG ATC CAG AGC GGC TGT C CCL2 GCA TCC ACG TGT TGG CTC A CTC CAG CCT ACT CAT TGG GAT CA MIP1α TTC TCT GTA CCA TGA CAC TCT GC CGT GGA ATC TTC CGG CTG TAG P0 GGC TGC CCT GCT CTT CTC GCA GTG CAG GGT CAC CTG p75 CCC CAC CAG AGG GAG AGA A GGC TAC TGT AGA GGT TGC CAT CA Periaxin AGG AAT CTT TGT CCG TGA GCT AGA ACA CAC GGG CAC TCA G Transferrin receptor GTG AAA CTG GCT GAA ACG GAG GGT CTG CCC AAT ATA AGC GAG A

The samples and primers were pipetted into a 384-well plate and the plate was centrifuged for 1 min at 1000 rpm (Centrifuge 5804 R, Eppendorf). The fragment amplification was done on a LightCycler 480 II (Roche, Switzerland) according to the protocol shown in Table 15.

73 Material and Methods

Table 15: qRT-PCR Protocol Process Cycles [#] Temperature [°C] Time [sec] Ramp rate [°C/s] Pre-Incubation 1 95 10 4.4 95 10 4.8 50 60 15 4.2 Amplification 72 10 2.8 1 95 5 4.8 1 65 60 2.2 Melting 1 up to 95 until 95°C is reached 0.11 Cooling 1 40 30 2.23

4.10 Cell Culture

4.10.1 Commonly used Media and Solutions D-Medium DMEM-Glutamax (Gibco #41965-039), 10 % heat-inactivated fetal calf serum (Gibco #12270-106)

Schwann cell growth medium DMEM-Glutamax (Gibco #41965-039), 10 % heat-inactivated fetal calf serum (FCS, Gibco #12270-106), 5 µg/ml bovine pituitary extract (BPE, BioConcept #BT-215-50), 2 μM Forskolin (Sigma, #F6886)

C-Medium For 500 ml of Medium: MEM-Glutamax (Gibco # 41090-028), 10 % heat-inactivated GOLD fetal calf serum (FCS-Gold, Gibco #16000-044, Lot #1619646), 10 ml D-glucose (0.2g/ml in MEM), 50 µl 2,5S NGF (0.5µg/µl, Millipore #480354), 50 µg/ml ascorbic acid (Sigma #A4403)

NB-Medium For 500 ml of Medium: Neurobasal (Gibco #21103-049), 10 ml B27 supplement (Gibco #17504-044), 10 ml D-glucose (0.2g/ml in MEM), 5 ml L-glutamine (200mM), 50 µl 2,5S NGF (0.5µg/µl, Millipore #480354)

L-15 Libovitz´s L-15 medium (Gibco #11415-064)

HBSS Hank´s balanced saline solution (Gibco #14170-112)

6x infection supplement 1 ml bovine pituitary extract (2,5 mg/ml inPBS, BioConcept #BT- 215-50), 100 µl forskolin (10mM in DMSO, Sigma, #F6886), 40,9 ml D-Medium

74 Material and Methods

4.10.2 Isolation of Primary Rat Schwann Cells Primary rat Schwann cells were prepared from sciatic nerves of 40 neonatal (P2) Sprague‐Dawley rats. Sciatic nerves were stripped of epineurium and collected in a 50 ml tube containing 25 ml L-15 medium. Once all nerves were collected, they were centrifuged for 10 min at 1000 rpm and the pellet was resuspended in 6 ml HBSS containing 1,25 mg/ml trypsin (Sigma, #T4665) and 2 mg/ml collagenase (Sigma, #C0130). The nerves were incubated for 70 min at 37 °C, flicking the tube every 10 min. After centrifugation for 10 min at 1200 rpm, the supernatant was removed and 2 ml D-medium was used to resuspend the pellet by pipetting 15 times. Subsequently the cells were plated on 4 PLL (poly-L-lysine, Sigma, #P5899)-coated 10 cm dishes. The next day, the medium was replaces with D-medium containing 10 µM Ara-C (Cytosine β-D-arabinofuranoside hydrochloride, Sigma C6645) to eliminate proliferating cells. Two days later, cells were gently washed twice with HBSS and the medium was replaced with D-medium. After two days, fibroblasts were lysed as follows: First, 2 ml of D-Medium containing 20 mM HEPES and 40 µl anti-Thy1.1 antibody (Serotec, MCA04G) was added to the cells. After incubating for 15 min at 37 °C, 400 µl of rabbit complement (Calbiochem, #234400) was added to the dish. The dish was swirled and placed at 37 °C for 30 min. Afterwards, the medium was removed and the cells were washed two times with HBSS. After two days in Schwann cell growth medium, the complement- mediated fibroblast killing was repeated. From here on, the primary rat Schwann cells were kept in Schwann cell growth medium.

4.10.3 Culture of Primary Rat Schwann Cells If not mentioned differently, cells were cultured in Schwann cell growth medium. When they reached confluency, they were split 1:3. For splitting a confluent 10 cm culture dish, the growth medium was removed and the adherent cells were incubated for 3 min with 3 ml of 0.25 % Trypsin (Gibco, # 25200- 056) to detach the cells from the culture dish. To inactivate the trypsin, 7 ml of pre-warmed D-medium was added. The cell solution was collected in a 15 ml tube and centrifuged for 5 min at 1200 rpm (Benchtop centrifuge MSE MISTRAL 1000, Kleiner AG, Wohlen, Switzerland). After removing the supernatant the cells were resuspended in fresh Schwann cell growth medium and seeded to PLL coated culture dishes.

4.10.4 Deferoxamine Treatment of Cells Cells were treated with a final concentration of 100 μM deferoxamine (DFO) (Sigma #D9533) which was added to the Schwann cell growth medium.

75 Material and Methods

4.10.5 Dorsal Root Ganglia Explant Cultures Mouse dorsal root ganglia (DRG) explant cultures consisting of sensory neurons, Schwann cells (SCs) and fibroblasts were prepared as follows: At embryonic day 13.5, the pregnant female was sacrificed and the embryos were extracted. Under the dissection hood, the surrounding ligaments of the uterus were removed and the embryos were rinsed in an L-15-filled dish. Embryos were transferred into a dish with fresh L-15, where they were kept until further dissection. In L-15, embryos were dissected by first cutting the head and the tail, then removing all the viscera. Then, the vertebras were cut open by inserting small scissor into the anterior part of the vertebral canal to expose the spinal cord. The spinal cord was removed and transferred together with the attached DRGs to a 35 mm petri dish containing 1 ml of L-15 medium. There, the DRGs were pinched off one after another. The DRGs were dissociated by removing the L-15 and adding 1 ml of 0.25 % Trypsin (Gibco, # 25200-056) and incubating for 45 min at 37 °C. The cell suspension was transferred to a 15 ml tube containing 10 ml of D-Medium to inactivate the trypsin and subsequently centrifuged for 5 min at 1200 rpm (Benchtop centrifuge MSE MISTRAL 1000, Kleiner AG, Wohlen, Switzerland). The supernatant was carefully discarded and the pellet was resuspended in 120 µl NB-medium per harvested DRG. 120 µl of the cell suspension was plated as a drop on matrigel-coated (BD, # 356234) glass coverslips (12 mm, placed in a 24-well plate). The next day, the medium was removed and the well filled with 300 µl NB-medium. From there onwards the medium was changed every other day. After two more changes of NB-medium cultures were switched to C-medium to induce myelination. 8 days after the change to C-medium, Schwann cells have properly myelinated the axons. In the case in which PLPCreERT2 DRGs were prepared to assess demyelination, recombination was induced 6 days after the switch to C-medium by adding 2 μM 4OH-tamoxifen (Sigma #H7904) for one medium cycle. These cultures were harvested 5 days after switching back to normal C-medium.

4.10.6 Production of Low-Titer Lentivirus and Infection of Rat Schwann Cells

For production of low-titer lentiviruses, HEK-293T cell were split to a PDL-coated (Sigma, #P7405) plate (ratio 1:7). In the evening of the following day, cells were transiently co-transfected over night with the lentiviral vector of choice (6 µg), the viral packaging constructs psPAX2 (3 μg) and VSV-G (3 µg) using Lipofectamine 2000 according to the manufacture’s protocol. 48h later the cell culture supernatant was collected and centrifuged for 5 min at 2000 rpm to pellet the cell debris. The supernatant was then filtered through a 0.45 μm syringe filter, aliquoted and stored at -80 °C until usage.

76 Material and Methods

Rat Schwann cells with approximately 70 % confluency were infected for 48 hours. For this the thawed viral aliquot was mixed with 6x infection supplement and added to the dish. After two days, the medium was changed back to Schwann cell growth medium.

4.10.7 Transferrin Uptake Assay To investigate clathrin-mediated endocytosis on a quantitative level, the transferrin uptake was analyzed by fluorescence-activated flow cytometry. Following serum starvation in DMEM GlutaMAX for 2 hours at 37°C, cells were incubated with 10 μg/ml Alexa Fluor 5687-labeled transferrin for 3 min at 37 °C. Subsequently, cells were kept on ice and washed twice with cold PBS, acidic stripped (0.2 M

Na2HPO4, 0,1 M citric acid) for 2 min to remove surface-bound transferrin, followed by two washing steps with PBS. After trypsinization with 1 ml for 10 min on ice in the dark, 2 ml of PBS were added before transferring them to a 15 ml falcon tube containing 3 ml of 8% PFA. After fixation for 20 min at RT, cells were washed twice with PBS by pelleting for 5 min at 600 g and finally resuspended in 500 µl PBS. Cells were kept in the dark at 4 °C until analysis by fluorescence-activated flow cytometry analysis, using a LSRFortessa (BD).

4.10.8 Production of Inducible shRNAs Doxycycline-inducible shRNAs were obtained by cloning the shRNA of interest into the pLKO-Tet-on vector. The cloning was done according to manufacturer’s protocol using the 21 bp sequences shown in Table 16.

Table 16: shRNA sequences shRNA name 21 bp sense sequence AP2 I CCG TAC TAC CAA GGA CAT CAT AP2 II GCC TGA GTG CAA GTT TGG AAT Dnm2 I GCC CTT GAG AAG AGG CTA CAT Dnm2 II GCA GTC CTA CAT CAA CAC AAA NS CAA CAA GAT GAA GAG CAC CAA

To induce the expression of the shRNAs, the first day 100 ng/ml doxycycline (Sigma, #D9891) and the second and third day 57 ng/ml were added to previously infected Schwann cells. 72 hours after the first administration, the cells were ready for analysis.

77 Material and Methods

4.10.9 Cell Surface Capturing Primary rat Schwann cells were infected with inducible shRNA-containing vectors and expression was initiated by treating them for 3 days with doxycycline as described above. Five confluent 15 cm dishes were used per condition. Cell Surface Capturing (CSC) technology was performed in collaboration with the Wollscheid-group according to their established protocol [247, 248].

4.11 Quantification and Statistics

4.11.1 Morphometric Analysis and Quantification of EM Images Quantification was performed on sciatic nerve cross-section panoramas (section 4.6.2). For the g-ratio analysis 150 axons (selected from 4 random fields) were quantified per animal using Aobe Photoshop CS5 with the polygonal magnetic lasso tool to determine the axonal area, and the ruler tool to measure the myelin thickness.

4.11.2 Morphometric Analysis and Quantification of Immunological Methods Quantification of Immunostainings (section 4.8.2) was performed on whole sciatic nerve cross-sections using the counting tool from Adobe Photoshop CS5.

Immunoblots (section4.7.4) were quantified using the relative density from each signal using ImageJ (version 1.48).

The fluorescence of Evens Blue (section 4.8.4) was determined using relative intensity measurements from Adobe Photoshop CS5.

For the analysis of DRG explant-cultures, at least 250 internodes were analysed per coverslip (selected from 4 random fields). N is the number of analysed coverslips per condition.

78 Material and Methods

4.11.3 Quantification of qRT-PCR

Samples were always measured as technical triplicates and averaged. Ct values calculated by the LightCycler 480 II were copied into Excel to calculate the difference between the gene of interest and the housekeeper gene (ΔCt). The ΔCt values were normalized to the average of all control samples according to equation 1 and further linearized according to equation 2. Statistics were performed on the normalized linear ΔCt values.

Equation 1: normalized ∆Ct= ∆Ct − average (∆Ct, controls)

- normalized ΔCt Equation 2: normalized linear ΔCt= 2

Melting curves were examined for amplicon confirmation.

4.11.4 Gait Analysis Statistics All gait analysis parameters were analysed using the CatWalk XT software (version 10.6).

4.11.5 Statistical Analysis For statistical analysis Prism 6 (GraphPad Software, Inc.) was used. Data show the mean ± SEM (standard error of the mean) and each data point represents one n. Significance was set at *P < 0.05, **P < 0.01, or ***P < 0.001. N indicating the number of animals per genotype, unless indicated otherwise.

Figures were assembled using Adobe Illustrator CS5.

79 Material and Methods

80 Appendix

5 Appendix

Figure 25) Iron supplementation does not affect control DRG-cultures. A) Immunostaining of dorsal root ganglia (DRG) co-cultures from PLPCreErt-Dnm2fl/fl and PLPCreErt+Dnm2fl/fl embryos. Hydroxy-Tamoxifen (OHT)-treated Cre+ cultures began to demyelinate. This demyelination could be partially prevented by supplementing the OHT-treated cultures with iron dextran (FeDex). FeDex did not alter Cre- cultures. Degenerating MBP is false-coloured in yellow. Scale bar: 100 µm. B) Quantification of intact internodes per field of view (FOV) (represented in A) C) Western blot analysis of Krox20 and dynamin 2 (Dnm2) of Cre- DRG co-cultures (as described in A). No changes were observed upon FeDex administration. D) Quantification of Krox20 protein levels (represented in C) (n=3 coverslips each condition, one-way ANOVA with Sidak´s multiple comparisons test).

81 Appendix

Figure 26) Follow-up study of the differently expressed transcripts found by RNA sequencing with AP2iko(R) mice A) Top panel: Schematic drawing of the inducible, Schwann cell-specific Ap2m1 (µ-subunit of the adaptor complex 2) ablation. Exons 4-11 of the Ap2m1 allele are flanked by loxP sites. Cre recombinase linked to an estrogen receptor variant is specifically expressed in Schwann cells, and activated upon tamoxifen administration, causing the excision of the floxed regions. Lower panel: Sashimi plot showing the Ap2m1 gene. Peaks represent the single exons. The connecting lines indicate reads, spanning to the neighbouring exon. The green line, reads spanning exon 3 to 12 is only present in recombined mice. B) Table of the 7 differentyl expressed genes. Cut-off: false discovery rate (FDR) <0.05, fold change > 1,5. C) qRT-PCR analysis of the differently expressed transcripts in the original samples used for the deep sequencing (n=3 animals each genotype, two-tailed unpaired Student´s t-test). D) qRT-PCR analysis of the differently expressed transcripts in sciatic nerve samples of AP2iko(R) mice at 6 weeks post-tamoxifen (n=3 animals each genotype, two-tailed unpaired Student´s t-test). E) qRT-PCR analysis of the differently expressed transcripts in sciatic nerve samples of AP2iko(R) mice at 1 year post-tamoxifen (n=3 animals each genotype, two-tailed unpaired Student´s t-test). | Mean ± SEM, *P < 0.05, **P < 0.01

82 Appendix

Figure 27) Full-length Blots, Part I A) Full-length Western blot, parts of which are shown in Figure 8 C). B) Full-length blots Figure 13 D). C) Full-length blot Figure 14 C). D) Full-length blot Figure 17 B). E) Full-length blot Figure 18 E). F) Full-length blot Figure 19 B).

83 Appendix

Figure 28) Full-length Blots, Part II A Full-length Western blots, parts of which are shown in Figure 10 D). B) Full-length blots Figure 10 E).

84 Appendix

Figure 29) Full-length Blots, Part III A) Full-length Western blot, parts of which are shown in Figure 20 A) left panel. B) Full-length blot Figure Figure 20 A) right panel. C) Full-length blot Figure 21 A). D) Full-length blot Figure 21 G). E) Full-length blot Figure 22 A). F+G) Full-length blot Figure 22 F).

85 Appendix

Figure 30) Full-length Blots, Part IV A) Full-length Western blots, parts of which are shown in Figure 22 J). B) Full-length blot Figure 23 C). C) Full-length blot Figure 23 K). D) Full-length blot Figure 23 M). D) Full-length blot Figure 25 C).

86 Appendix

Table 17: The surface glycoproteome of primary rat Schwann cells. This table updates and complements a dataset, previously published by Paris Sidiropoulos (PS) and colleagues [114]. We could identify 219 new previously not described n-glycosylated surface proteins on primary rat Schwann cells. The number of unique peptides and total number of independent spectra for each protein identified by mass spectrometry are listed. “Exclusively” refers to proteins that have been found in the previous screen but were not detected by our CSC. # of Total Identified by Protein name Uniprot ID unique indep. PS peptides spectra 4F2 cell-surface antigen heavy chain Q794F9 58 731 Yes 5'-nucleotidase P21588 1 3 Yes A disintegrin and metalloproteinase with thrombospondin motifs 20 D3ZN23 1 2 No Acid ceramidase Q6P7S1 2 45 No Activin receptor type-2A P27038 2 17 Yes Adenosine receptor A1 P25099 1 17 No Adenylate cyclase type 5 P84309 2 5 Yes Adenylate cyclase type 6 Q03343 9 135 Yes ADP-ribosyl cyclase 1 Q64244 2 21 Yes Agrin P25304 6 32 No ALK tyrosine kinase receptor F1LRZ0 2 9 No Alkaline phosphatase, tissue-nonspecific isozyme P08289 4 11 Exclusively Alpha-1,3-mannosyl-glycoprotein 4-beta-N-acetylglucosaminyltransferase B B2GV39 1 3 No Alpha-1,6-mannosyl-glycoprotein 2-beta-N-acetylglucosaminyltransferase Q09326 1 18 No Alpha-2-HS-glycoprotein P24090 2 28 No Alpha-mannosidase 2x D4A4J3 1 5 No Alpha-sarcoglycan D3ZDQ9 1 2 No Angiopoietin-related protein 2 G3V862 1 3 No Anoctamin-6 F1M5Z5 16 179 Yes Anthrax toxin receptor 1 Q0PMD2 2 67 Yes Anthrax toxin receptor 2 Q00IM8 1 23 No ATP-binding cassette sub-family A member 1 Q6MG08 18 183 Yes ATP-binding cassette sub-family A member 7 Q7TNJ2 2 52 Yes ATP-binding cassette sub-family A member 8-A D3ZXD2 2 28 Yes ATP-binding cassette sub-family A member 9 Q8K449 1 3 Exclusively Atrial natriuretic peptide receptor 2 P16067 3 42 Yes Attractin Q99J86 9 96 Yes Autophagy-related protein 9A Q5FWU3 1 4 No Basal molecule Q9ESS6 5 76 No Basement membrane-specific core protein Q05793 5 12 Exclusively Basigin P26453 15 261 Yes BDNF/NT-3 growth factors receptor Q63604 4 36 No Beta-hexosaminidase subunit alpha Q641X3 1 16 No Beta-type platelet-derived growth factor receptor Q05030 7 24 Exclusively Bone marrow stromal antigen 2 Q811A2 3 56 No Bone morphogenetic protein receptor type-2 Q91WY9 4 50 Yes EGF LAG seven-pass G-type receptor 1 Q9QYP2 3 7 No Cadherin EGF LAG seven-pass G-type receptor 2 Q9QYP2 12 79 Yes Cadherin EGF LAG seven-pass G-type receptor 2 (Fragment) Q9QYP2 12 79

Cadherin-13 Q8R490 21 192 No Cadherin-15 Q75NI5 3 77 No Cadherin-2 Q9Z1Y3 18 111 No Cadherin-6 P55280 10 105 Yes Calcium homeostasis modulator protein 2 Q5RJQ8 1 3 Exclusively Calumenin O35783 2 22 No Carboxypeptidase D O89001 8 103 No CAS1 domain-containing protein 1 M0R6Q0 1 1 No Cathepsin L1 P07154 2 5 No Cation-dependent mannose-6-phosphate receptor Q6AY20 6 134 Yes Cation-independent mannose-6-phosphate receptor Q63002 4 108 No CD151 antigen Q9QZA6 2 27 Yes CD166 antigen O35112 22 361 Yes CD276 antigen Q7TPB4 13 260 Yes CD320 antigen Q5HZW5 1 3 Yes CD44 antigen P26051 37 332 Yes CD48 antigen P10252 8 102 Yes CD59 glycoprotein P27274 3 53 Yes CD63 antigen P28648 9 87 Yes CD82 antigen O70352 3 49 Yes CD97 antigen Q5XI36 5 94 No

87 Appendix

Cell adhesion molecule 1 Q6AYP5 44 702 Yes 2 Q1WIM2 7 73 Yes Cell adhesion molecule 4 Q1WIM1 23 314 Yes Cell cycle control protein 50A Q6AY41 3 47 Yes Cell surface glycoprotein MUC18 Q9EPF2 20 191 Yes Ceruloplasmin P13635 4 98 Yes Chloride channel protein 2 P35525 1 3 Exclusively Chloride transport protein 6 D4A3H5 1 8 No Choline transporter-like protein 1 Q6X893 1 31 Yes Choline transporter-like protein 2 B4F795 16 597 Yes Ciliary neurotrophic factor receptor subunit alpha O88507 6 38 Yes CMP-N-acetylneuraminate-beta-galactosamide-alpha-2,3-sialyltransferase 4 P61131 3 26 No Collagen alpha-1(I) chain P02454 11 80 No Collagen alpha-1(V) chain O88207 1 4 No Collagen alpha-1(XII) chain F1LQC3 2 9 No Collagen alpha-1(XVI) chain F1LND0 1 34 No Collagen alpha-2(I) chain P02466 3 6 No Collagen alpha-2(IV) chain F1M6Q3 2 20 No Contactin-1 P12960 4 29 No Contactin-6 P97528 3 60 Yes C-type mannose receptor 2 Q4TU93 10 102 Yes C-X-C chemokine receptor type 7 O89039 1 2 No Death domain-containing NRADD Q8K5A9 2 34 No Deoxyribonuclease-1-like 1 Q2QDE7 3 59 Yes Dihydropyrimidinase-related protein 3 Q62952 13 29 No Disintegrin and metalloproteinase domain-containing protein 10 O35598 4 47 Yes Disintegrin and metalloproteinase domain-containing protein 10 Q10743 4 47

Disintegrin and metalloproteinase domain-containing protein 17 Q9Z0F8 3 17 Yes Disintegrin and metalloproteinase domain-containing protein 19 D3ZPM7 1 8 No Dolichyl-diphosphooligosaccharide--protein glycosyltransferase subunit 1 P07153 4 28 No Dolichyl-diphosphooligosaccharide--protein glycosyltransferase subunit 2 P25235 4 34 No Dolichyl-diphosphooligosaccharide--protein glycosyltransferase subunit STT3A B4F7C9 5 19 Yes Dolichyl-diphosphooligosaccharide--protein glycosyltransferase subunit STT3B Q7TQ74 7 93 No Dystroglycan D9IFS3 3 70 Yes E3 ubiquitin-protein ligase HECTD1 D3ZLS5 3 5 No E3 ubiquitin-protein ligase RNF13 Q66HG0 1 17 Yes Ectonucleoside triphosphate diphosphohydrolase 1 ENTP1 2 6 No Ectonucleoside triphosphate diphosphohydrolase 2 O35795 6 75 No Ectonucleoside triphosphate diphosphohydrolase 7 D3ZNB5 1 8 No Ectonucleotide pyrophosphatase/phosphodiesterase family member 1 P06802 2 37 Yes Embigin O88775 16 357 Yes EMILIN-1 D3Z9E1 3 36 Yes Endothelin B receptor P21451 2 31 No Endothelin-converting enzyme 1 P42893 28 346 Yes Eosinophil cationic protein 2 P97425 3 11 No Eosinophil cationic-type ribonuclease 3 O35290 2 6 Exclusively Ephrin type-A receptor 2 D3ZBN3 3 55 Yes Ephrin type-B receptor 2 P54763 2 6 Exclusively Ephrin type-B receptor 3 D3ZH39 1 4 No Ephrin type-B receptor 4 M0RDA4 2 37 Yes Ephrin-B1 P52795 5 45 Yes Epidermal growth factor receptor Q9QX70 5 53 No Epithelial membrane protein 1 P54848 6 97 Yes Epithelial membrane protein 2 Q66HH2 1 2 No Epithelial membrane protein 3 Q9QYW5 11 178 Yes Epsilon-sarcoglycan O70258 1 9 Yes Equilibrative nucleoside transporter 1 O54698 2 6 No Equilibrative nucleoside transporter 2 O54699 1 23 No ER membrane protein complex subunit 1 D4A994 1 36 No ER membrane protein complex subunit 10 Q6AYH6 1 10 No Excitatory amino acid transporter 1 P24942 2 7 Exclusively Fibroblast growth factor receptor 1 Q04589 3 74 No Fibroblast growth factor receptor-like 1 Q7TQM3 3 71 Yes Fibroleukin G3V7P2 2 13 No Fibronectin Q6LC76 13 117 Yes Fibronectin type III domain-containing protein 5 Q8K3V5 1 21 No Follistatin-related protein 1 Q62356 1 23 No Fructose-bisphosphate aldolase A P05064 13 37 No Gamma-aminobutyric acid type B receptor subunit 2 O88871 1 1 No

88 Appendix

GDNF family receptor alpha-1 Q62997 5 62 Yes GDNF family receptor alpha-2 O35977 20 377 Yes GDNF family receptor alpha-3 GFRA3 1 31 No GDP-fucose protein O-fucosyltransferase 2 D3ZUN5 2 12 No Gliomedin Q80WL1 9 27 Exclusively Glucosylceramidase B2RYC9 1 3 No Glutamate carboxypeptidase 2 P70627 3 7 Exclusively Glutamate receptor ionotropic, kainate 2 P42260 5 61 No Glycosyltransferase-like protein LARGE1 D4A390 1 1 No -1 P35053 12 171 Yes Glypican-4 Q642B0 4 53 Yes Golgi apparatus protein 1 Q61543 5 84 No Golgi membrane protein 1 D4AEL2 4 21 No GPI inositol-deacylase Q765A7 1 4 No GPI transamidase component PIG-S Q5XI31 1 3 No G-protein coupled receptor 124 D4ADC7 9 167 No G-protein coupled receptor 126 F1LUL6 5 119 Yes G-protein coupled receptor 56 Q8K3V3 13 246 Yes G-protein coupled receptor 64 Q8CJ11 3 3 Yes H(+)/Cl(-) exchange transporter 5 P51796 3 40 No Heparan sulfate 2-O-sulfotransferase 1 G3V7N0 2 33 No Hepatocyte cell adhesion molecule D3ZEI4 6 126 Yes Hepatocyte growth factor receptor P97523 1 5 No High affinity cationic amino acid transporter 1 P30823 5 162 Yes Hyaluronidase-2 Q9Z2Q3 2 2 No Hypoxia up-regulated protein 1 Q63617 5 115 No IgLON family member 5 F1LVR0 3 22 Yes Immunoglobulin superfamily DCC subclass member 4 F7EL93 2 7 No Immunoglobulin superfamily member 3 F1LZ40 2 2 No Immunoglobulin superfamily member 8 no rat 2 56 Yes Inactive tyrosine- 7 P32577 14 226 No Inositol monophosphatase 3 D4AD37 1 6 No Insulin receptor P15127 4 45 Yes Insulin-like growth factor 1 receptor P24062 10 106 Yes Integrin alpha V F1LZX9 24 176 No Integrin alpha-1 P18614 13 116 Yes Integrin alpha-3 F1M867 1 22 Yes Integrin alpha-4 Q00651 5 11 Exclusively Integrin alpha-5 D4ACU9 21 167 Yes Integrin alpha-6 Q924W2 19 264 Yes Integrin alpha-7 Q63258 15 133 Yes Integrin alpha-8 Q9WUP8 2 35 No Integrin beta-1 P49134 63 1175 Yes Integrin beta-3 Q8R2H2 3 30 Yes Intercellular adhesion molecule 1 Q00238 12 83 Yes Interleukin-1 receptor accessory protein Q63621 12 133 Yes Interleukin-6 receptor subunit beta P40190 23 246 Yes Junctional adhesion molecule C Q68FQ2 5 82 Yes KDEL motif-containing protein 2 Q566E5 1 2 No Keratin, type I cytoskeletal 28 Q6IFW7 2 5 No Kin of IRRE-like protein 1 F1M7V1 2 43 Yes Kit ligand P21581 1 2 Exclusively Lamina-associated polypeptide 2, isoform beta Q62733 4 16 No Laminin subunit alpha-4 F1LTF8 13 36 No Laminin subunit alpha-5 F1MAN8 4 29 No Laminin subunit beta-1 D3ZQN7 13 124 Yes Laminin subunit gamma-1 P02468 4 11 Exclusively Latrophilin-2 O88923 3 41 Yes Latrophilin-3 Q80TS3 8 113 No Leucine-rich repeat and fibronectin type-III domain-containing protein 4 D4ABX8 3 18 No Leucine-rich repeat and immunoglobulin-like domain-containing nogo receptor- G3V881 1 1 No interacting protein 1 Leucine-rich repeat transmembrane protein FLRT3 B1H234 3 29 No Leucine-rich repeat-containing G-protein coupled receptor 4 Q9Z2H4 3 12 No Leucine-rich repeat-containing protein 15 Q8R5M3 1 3 Exclusively Leucine-rich repeat-containing protein 4 Q45R42 2 2 No Leucine-rich repeat-containing protein 4B P0CC10 9 109 Yes Leucine-rich repeat-containing protein 8A Q4V8I7 4 36 Yes Leucine-rich repeats and immunoglobulin-like domains protein 1 F7F3K6 2 18 No

89 Appendix

Leucine-rich repeats and immunoglobulin-like domains protein 2 D3ZQV2 1 7 No Leucyl-cystinyl aminopeptidase P97629 14 115 Yes Leukocyte surface antigen CD47 P97829 13 157 Yes Limbic system-associated membrane protein Q62813 4 8 Exclusively Lipid phosphate phosphohydrolase 1 O08564 7 100 Yes Lipid phosphate phosphohydrolase 2 Q8K593 1 34 Yes Low-density lipoprotein receptor P35952 4 39 No Low-density lipoprotein receptor-related protein 4 Q9QYP1 3 12 No Lysophosphatidic acid receptor 3 Q8K5E0 2 13 Yes Lysophosphatidic acid receptor 4 D3ZHA3 1 34 Yes Lysophosphatidic acid receptor 6 Q4G072 2 54 Yes Lysosomal acid phosphatase P20611 2 31 No Lysosomal protein NCU-G1 Q68FV6 1 18 No Lysosome membrane protein 2 P27615 11 168 Yes Lysosome-associated membrane glycoprotein 1 P14562 27 310 Yes Lysosome-associated membrane glycoprotein 2 P17046 8 81 Yes Macrophage colony-stimulating factor 1 P07141 2 35 Yes Major facilitator superfamily domain-containing protein 8 no rat 1 17 No Major prion protein P13852 8 199 Yes Matrix metalloproteinase-15 D3ZCG5 2 35 Yes Matrix-remodeling-associated protein 8 Q5XI43 1 4 Yes Metal transporter CNNM4 P0C588 1 19 No Mucolipin-1 D3ZRF9 2 11 No Multidrug resistance protein 2 Q08201 1 3 Exclusively Multidrug resistance protein 3 Q08201 2 8 No Multidrug resistance-associated protein 1 Q8CG09 5 128 No Multiple epidermal growth factor-like domains protein 10 F1MAP4 3 4 No Multiple epidermal growth factor-like domains protein 8 Q9QYP0 2 4 Exclusively Multiple epidermal growth factor-like domains protein 9 D4A3L3 7 113 Yes Muscle, skeletal receptor tyrosine protein kinase Q61006 1 10 No Muscle, skeletal receptor tyrosine-protein kinase Q62838 1 10

Myelin protein P0 P06907 22 378 Yes Myelin protein zero-like protein 1 Q3TEW6 2 44 Yes Myelin-associated glycoprotein P07722 4 9 Exclusively Myotubularin-related protein 7 D3ZTB0 9 103 Yes N-acetylglucosamine-1-phosphotransferase subunits alpha/beta D3ZJS1 4 40 No Natural resistance-associated macrophage protein 2 O54902 2 10 No Neogenin (Fragment) P97603 3 17 Yes Neprilysin P07861 4 72 Yes Nestin P21263 26 45

Netrin receptor UNC5B O08722 3 62 Yes Neural cell adhesion molecule 1 P13596 54 808 No Neural cell adhesion molecule 2 O35136 2 4 Exclusively Neural cell adhesion molecule L1 Q05695 66 878 Yes Neural cell adhesion molecule L1-like protein Q05695 15 206 Yes Neurexin-1-alpha Q63372 1 3 Exclusively Neuroligin-2 Q62888 1 15 No Neuroligin-3 Q62889 1 3 Exclusively Neuronal cell adhesion molecule P97686 10 97 No Neuronal membrane glycoprotein M6-b P35803 7 187 Yes Neuropilin-1 Q9QWJ9 7 68 Yes Neuropilin-2 O35276 1 23 No Neuroplastin P97546 15 209 Yes Neutral cholesterol ester hydrolase 1 B2GV54 1 9 No Nicalin Q5XIA1 1 10 No Nicastrin Q8CGU6 8 83 Yes Nidogen-2 B5DFC9 3 36 Yes Noelin-2 Q568Y7 2 2 No NT-3 growth factor receptor Q03351 4 9 Exclusively Nucleotide exchange factor SIL1 Q6P6S4 1 6 No Opioid-binding protein/cell adhesion molecule P32736 5 30 No OX-2 membrane glycoprotein P04218 8 189 Yes P2X purinoceptor 4 P51577 2 38 No P2X purinoceptor 7 Q64663 1 21 Yes Palmitoyl-protein thioesterase 1 P45479 1 8 No Peptidyl-prolyl cis-trans isomerase FKBP9 Q66H94 3 4 No Periostin D3ZAF5 1 3 Yes Peripheral myelin protein 22 P25094 12 205 Yes Phospholipase D3 Q5FVH2 1 28 No

90 Appendix

Piezo-type mechanosensitive ion channel component 1 Q0KL00 4 19 No Platelet-derived growth factor receptor beta Q05030 7 124 No Plexin-A1 D3Z981 3 80 Yes Plexin-A2 D3ZWP6 3 8 No Plexin-A3 D3ZPX4 1 27 No Plexin-A4 D3ZES7 4 38 No Plexin-B1 D3ZDX5 5 63 Yes Plexin-B2 D3ZQ57 7 147 No Plexin-B3 D3ZLH5 1 23 Yes Plexin-D1 D4AA77 3 52 No Poliovirus receptor-related protein 1 F1LNP8 4 73 Yes Poliovirus receptor-related protein 2 Q5FVC5 4 42 Yes Poliovirus receptor-related protein 3 D4A5C0 2 19 Yes Pre-B-cell leukemia transcription factor-interacting protein 1 A2VD12 2 13 No Prenylcysteine oxidase Q99ML5 3 57 No Prenylcysteine oxidase-like B5DEI0 2 4 No Probable cation-transporting ATPase 13A3 no rat 2 5 No Probable G-protein coupled receptor 125 D3ZZF2 2 16 No Probable G-protein coupled receptor 176 Q64017 2 25 No Probable lysosomal cobalamin transporter Q6AZ61 7 109 No Procollagen galactosyltransferase 1 B1H282 4 79 No Procollagen-lysine,2-oxoglutarate 5-dioxygenase 1 Q63321 2 20 No Procollagen-lysine,2-oxoglutarate 5-dioxygenase 2 Q811A3 4 53 No Prolow-density lipoprotein receptor-related protein 1 Q6AYP5 28 388 Yes Prolyl 3-hydroxylase 1 Q9R1J8 3 11 No Prolyl 4-hydroxylase subunit alpha-1 P54001 2 21 No Prostaglandin F2 receptor negative regulator Q62786 9 92 Yes Protein CASC4 M0R605 3 23 No Protein FAM171A2 D3ZT47 1 21 No Protein FAM38A Q0KL00 4 9 Exclusively Protein ITFG3 Q5M7W6 5 103 Yes Protein jagged-1 Q63722 10 97 Yes Protein LSM14 homolog B D3ZU15 1 7 No Protein O-mannosyl-transferase 2 Q14U74 1 17 No Protein phosphatase 1L no rat 1 36 No Protein sel-1 homolog 1 Q80Z70 1 20 No Protein sidekick-2 D3ZDA6 8 118 Yes Protein tweety homolog 1 P0C5X8 1 1 No Protein tweety homolog 3 D4A383 2 36 No Proteinase-activated receptor 1 P26824 9 107 No alpha-4 O88689 1 9 Yes Protocadherin Fat 4 D3ZEH1 4 24 No Proton myo-inositol cotransporter Q921A2 1 22 No Proton-coupled folate transporter Q5EBA8 3 65 Yes Putative polypeptide N-acetylgalactosaminyltransferase-like protein 3 Q5CD99 3 56 No Receptor activity-modifying protein 3 Q9JJ73 1 12 No Receptor tyrosine-protein kinase erbB-2 P06494 7 60 Yes Receptor tyrosine-protein kinase erbB-3 Q62799 13 146 Yes Receptor-type tyrosine-protein phosphatase eta Q62884 2 44 Yes Receptor-type tyrosine-protein phosphatase F Q64604 1 3 Exclusively Receptor-type tyrosine-protein phosphatase mu F1LPJ1 3 24 Yes Receptor-type tyrosine-protein phosphatase S B0V2N1 3 73 No Receptor-type tyrosine-protein phosphatase zeta Q62656 4 61 Yes Retinoid-inducible serine carboxypeptidase Q920A6 1 9 No Reversion-inducing -rich protein with Kazal motifs D4ABJ4 7 86 No RT1 class I histocompatibility antigen, AA alpha chain P16391 2 37 No Sarcalumenin M0R570 1 25 No Scavenger receptor class B member 1 P97943 4 55 No Scavenger receptor class F member 2 D3ZBQ5 3 26 No Semaphorin-4C D4A9J3 5 73 Yes Semaphorin-6D D3ZDA2 1 1 No Semaphorin-7A D3ZQP6 10 131 Yes Serine incorporator 1 Q7TNK0 2 7 No Serine incorporator 5 Q63175 2 6 Yes Serum paraoxonase/arylesterase 2 Q6AXM8 1 2 No Sialomucin core protein 24 Q9QX82 3 36 Yes SID1 transmembrane family member 2 D3ZEH5 1 10 No Signal peptide peptidase-like 2B Q5PQL3 5 41 Yes SLIT and NTRK-like protein 2 D3ZK41 1 8 Yes

91 Appendix

SLIT and NTRK-like protein 6 D3ZP44 2 11 Yes Smoothened homolog P56726 1 2 No Sodium- and chloride-dependent creatine transporter 1 P28570 2 25 Yes Sodium- and chloride-dependent glycine transporter 1 P28571 1 2 No Sodium- and chloride-dependent taurine transporter P31643 17 170 No Sodium bicarbonate cotransporter 3 Q9R1N3 8 101 Yes Sodium channel protein type 5 subunit alpha P15389 3 5 No Sodium/hydrogen exchanger 1 P26431 3 48 Yes Sodium/potassium/calcium exchanger 3 (Fragment) Q9EPQ0 4 27 Yes Sodium/potassium-transporting ATPase subunit beta-1 P07340 2 9 Yes Sodium/potassium-transporting ATPase subunit beta-3 Q63377 5 118 Yes Sodium-coupled neutral amino acid transporter 2 Q9JHE5 4 72 Yes Sodium-dependent neutral amino acid transporter B(0)AT2 Q08469 1 1 No Solute carrier family 12 member 2 P55012 1 3 Exclusively Solute carrier family 12 member 4 Q63632 1 65 Yes Solute carrier family 12 member 7 Q5RK27 4 117 Yes Solute carrier family 12 member 9 Q66HR0 5 56 Yes Solute carrier family 15 member 4 O09014 3 42 No Solute carrier family 2, facilitated glucose transporter member 1 P11167 2 32 Yes Solute carrier family 2, facilitated glucose transporter member 3 Q07647 1 77 Yes Solute carrier family 22 member 17 Q9P290 1 7 No Solute carrier family 22 member 23 Q9QZG1 2 32 Yes Solute carrier family 22 member 5 O70594 4 59 No Solute carrier family 23 member 2 Q9WTW8 1 3 Exclusively Solute carrier family 52, riboflavin transporter, member 2 B5MEV3 3 45 No Solute carrier organic anion transporter family member 3A1 Q99N02 5 98 Yes Sortilin O54861 7 106 Yes Sphingosine 1-phosphate receptor 2 P47752 3 99 Yes Sulfate transporter O70531 8 43 No Sulfated glycoprotein 1 P10960 7 17 No SUN domain-containing protein 2 D3ZTT7 1 3 No Talin-1 Q498D4 3 4

T-cell immunomodulatory protein Q8R4E1 3 35 No Tectonic-2 Q3B7D3 4 22 No Tenascin Q62657 2 17 No -1 no rat 2 16 Yes Teneurin-3 Q9WTS6 10 28 Exclusively Testis-specific gene 10 protein Q9Z220 1 14 No -13 Q5FVL6 1 15 No Tetraspanin-15 M0R749 2 34 No Tetraspanin-3 Q66H06 2 62 Yes Tetraspanin-31 Q5U1V9 1 2 No TGF-beta receptor type-2 P38438 5 45 Yes Thioredoxin domain-containing protein 15 Q5BJT4 1 23 No Thrombospondin type-1 domain-containing protein 7A F1MA97 1 2 No Thrombospondin-1 P35441 2 5 Exclusively Thyrotropin-releasing hormone-degrading ectoenzyme Q10836 2 4 No Tissue alpha-L-fucosidase P17164 1 1 No Tissue factor P42533 6 15 Exclusively Tissue factor pathway inhibitor Q02445 1 3 Exclusively Tomoregulin-1 Q9QYV1 1 1 No Torsin-1A-interacting protein 1 Q5PQX1 2 6 No Torsin-1A-interacting protein 2 Q6P752 2 4 No Torsin-3A Q5M936 1 3 No Transferrin receptor protein 1 Q99376 5 67 No Transforming growth factor beta receptor type 3 P26342 3 12 Yes Translocon-associated protein subunit alpha Q7TPJ0 4 7 No Transmembrane 9 superfamily member 1 Q66HF2 1 34 No Transmembrane 9 superfamily member 3 D3ZCR2 1 36 No Transmembrane and TPR repeat-containing protein 4 B2RYC0 1 12 No Transmembrane channel-like protein 7 Q8C428 1 2 Exclusively Transmembrane emp24 domain-containing protein 9 Q5I0E7 8 85 No Transmembrane glycoprotein NMB Q6P7C7 3 20 No Transmembrane protease serine 5 F1LSP0 1 24 Yes Transmembrane protein 106B Q6AYA5 2 11 No Transmembrane protein 106C Q5RJK0 1 35 No Transmembrane protein 110 Q7TSW6 2 4 No Transmembrane protein 132A Q80WF4 2 41 Yes Transmembrane protein 150A Q9QZE9 2 6 Yes

92 Appendix

Transmembrane protein 158 Q91XV7 1 7 No Transmembrane protein 182 D3ZZT3 1 3 No Transmembrane protein 2 D3ZZ19 11 145 Yes Transmembrane protein 206 Q66H28 15 110 Yes Transmembrane protein 245 B1AZA5 2 15 No Transmembrane protein 248 Q6AY76 1 7 No Transmembrane protein 63B D4A105 6 52 Yes Transmembrane protein 87A M0R3Z3 1 1 No Transmembrane protein 9 B5DFJ7 1 22 No Triosephosphate isomerase P48500 4 14

Trophoblast glycoprotein Q5PQV5 5 56 No Tumor necrosis factor receptor superfamily member 3 Q5U2S8 2 13 No Tumor necrosis factor receptor superfamily member 6 Q63199 2 12 No Tyrosine-protein kinase Mer P57097 3 12 Yes Tyrosine-protein kinase receptor UFO E9PSY0 8 78 No Tyrosine-protein kinase RYK F1LQR5 2 27 Yes Tyrosine-protein phosphatase non-receptor type substrate 1 P97710 19 306 Yes UDP-GlcNAc:betaGal beta-1,3-N-acetylglucosaminyltransferase 2 D3ZEF9 1 24 No UDP-glucose:glycoprotein glucosyltransferase 1 Q9JLA3 1 1 No Uncharacterized family 31 glucosidase KIAA1161 D4AE63 1 20 Yes Uncharacterized protein KIAA1467 D3ZWJ9 1 2 No UPF0458 protein C7orf42 homolog Q6AY76 2 6 Exclusively UPF0606 protein KIAA1549 D3Z9D0 1 1 No Urokinase plasminogen activator surface receptor P49616 7 104 Yes Vascular cell adhesion protein 1 P29534 2 17 No Vasorin D3ZAE6 1 8 Yes core protein (Fragments) Q9ERB4 1 1 No Vesicular integral-membrane protein VIP36 B0BNG3 4 14 No Vitamin K-dependent gamma-carboxylase O88496 1 1 No Voltage-dependent calcium channel subunit alpha-2/delta-1 P54290 19 227 Yes VPS10 domain-containing receptor SorCS2 D3ZW09 5 48 Yes V-type proton ATPase subunit S1 O54715 1 6 No Zinc transporter ZIP10 D4A517 1 15 No Zinc transporter ZIP6 Q4V887 7 107 Yes Zinc transporter ZIP8 Q5FVQ0 1 3 Exclusively

93 Appendix

94 References

6 References

[1] D. K. Hartline and D. R. Colman, 2007, „Rapid conduction and the evolution of giant axons and myelinated fibers“, Current Biology, vol. 17, p. 29

[2] T. H. Bullock et al., 1984, „Evolution of myelin sheaths: both lamprey and hagfish lack myelin“, Neuroscience letters, vol. 48, p. 145

[3] G. Saher et al., 2005, „High cholesterol level is essential for myelin membrane growth“, Nature Neuroscience, vol. 8 (4), p. 468

[4] L. Steinman, 1998, „Multiple sclerosis: a coordinated immunological attack against myelin in the central nervous system“, Cell, vol. 85 (3), p. 299

[5] R. H. Quarles et al., 1972, „Demonstration of a glycoprotein which is associated with a purified myelin fraction from rat brain“, Biochemical and Biophysical Research Communications, vol. 47 (2), p. 491

[6] R. Martini and S. Carenini, 1998, „Formation and maintenance of the myelin sheath in the peripheral nerve: Roles of cell adhesion molecules and the gap junction protein connexin 32“, Microscopy research and technique, vol. 41, p. 403

[7] P. Morell and R. H. Quarles, 1999, „Myelin formation, structure and biochemistry“, book

[8] S. Aggarwal et al., 2011, „A size barrier limits protein diffusion at the cell surface to generate lipid-rich myelin- membrane sheets“, Developmental Cell, vol. 21, p. 445

[9] T. Coet et al., 1998, „New perspectives on the function of myelin galactolipids“, Trends in Neurosciences, vol. 21 (3), p. 126

[10] N. M. Le Douarin and M. Teillet, 1974, „Experimental analysis of the migration and differentiation of neuroblasts of the autonomic nervous system and of neurectodermal mesenchymal derivatives, using a biological cell marking technique“, Developmental Biology, vol. 41 (1), p. 162

[11] J. A. Weston, 1963, „A radioautographic analysis of the migration and localization of trunk neural crest cells in the chick“, Developmental Biology, vol. 6 (3), p. 279

[12] Z. Dong et al., 1995, „Neu differentiation factor is a -glia signal and regulates survival, proliferation, and maturation of rat Schwann cell precursors“, Neuron, vol. 15, p. 585

[13] H. D. Webster and J. R. Martin, 1973, „The relationships between interphase Schwann cells and axons before myelination: a quantitative electron microscopic study“, Developmental Biology, vol. 32 (2), p. 401

[14] G. V. Michailov et al., 2004, „Axonal neuregulin-1 regulates myelin sheath thickness“, Science, vol. 304, p. 700

[15] A. Citri et al., 2003, „The deaf and the dumb: the biology of ErbB-2 and ErbB-3“, Experimental cell research, vol. 284, p. 54

[16] K. R. Monk et al., 2015, „New insights on schwann cell development“, Glia, vol. 63 (8), p. 1376

[17] W. A. H. Rushton, 1951, „A theory of the effects of fibre size in medullated nerve“, The Journal of Physiology, vol. 115 (1), p. 101

[18] F. O. Schmitt and R. S. Bear, 193, „The optical properties of vertebrate nerve axons as related to fiber size“, Journal of Cellular and Comparative Physiology, vol. 9 (2), p. 261

[19] K. R. Jessen and R. Mirsky, 2008, „Negative regulation of myelination: relevance for development, injury, and demyelinating disease“, Glia, vol. 56, p. 1552

[20] C. Jopling et al., 2011, „Dedifferentiation, transdifferentiation and reprogramming: three routes to regeneration“, Nature Reviews Molecular Cell Biology, vol. 11, p. 79

[21] D. B. Parkinson et al., 2004, „Krox-20 inhibits Jun-NH2-terminal kinase/c-Jun to control Schwann cell proliferation and death.“, The Journal of Cell Biology, vol. 164 (3), p. 385

[22] P. J. Arthur, 2012, „c-Jun reprograms Schwann cells of injured nerves to generate a repair cell essential for regeneration“, Neuron, vol. 75 (4), p. 633

[23] D. B. Parkinson et al., 2008, „c-Jun is a negative regulator of myelination.“, The Journal of Cell Biology, vol. 181 (4), p. 625

95 References

[24] F. Schweisguth, 2004, „Regulation of notch signaling activity.“, Curr Biol, vol. 14 (3), p. R129

[25] A. Woodhoo et al., 2009, „Notch controls embryonic Schwann cell differentiation, postnatal myelination and adult plasticity“, Nature, vol. 12, p. 839

[26] I. Napoli et al., 2012, „A Central Role for the ERK-Signaling Pathway in Controlling Schwann Cell Plasticity and Peripheral Nerve Regeneration In Vivo“, Neuron, vol. 73 (4), p. 729

[27] V. H. Perry et al., 1995, „Radiation-induced reductions in macrophage recruitment have only slight effects on myelin degeneration in sectioned peripheral nerves of mice.“, Eur J Neurosci, vol. 7 (2), p. 271

[28] J. A. Gomez-Sanchez et al., 2015, „Schwann cell autophagy, myelinophagy, initiates myelin clearance from injured nerves.“, The Journal of Cell Biology, vol. 210 (1), p. 153

[29] S. Fischer et al., 2008, „Increase of MCP‐1 (CCL2) in myelin mutant Schwann cells is mediated by MEK‐ERK signaling pathway“, Glia, vol. 56 (8), p. 836

[30] M. Mueller et al., 2001, „Rapid response of identified resident endoneurial macrophages to nerve injury.“, The American Journal of Pathology, vol. 159 (6), p. 2187

[31] M. Mueller et al., 2003, „Macrophage response to peripheral nerve injury: the quantitative contribution of resident and hematogenous macrophages.“, Lab Invest, vol. 83 (2), p. 175

[32] R. Nagarajan et al., 2001, „EGR2 mutations in inherited neuropathies dominant-negatively inhibit myelin gene expression.“, Neuron, vol. 30 (2), p. 355

[33] R. George and J. W. Griffin, 1994, „Delayed macrophage responses and myelin clearance during Wallerian degeneration in the central nervous system: the dorsal radiculotomy model“, Experimental neurology, vol. 129, p. 225

[34] W. W. Schlaepfer and M. B. Hasler, 1979, „Characterization of the calcium-induced disruption of neurofilaments in rat peripheral nerve“, Brain Research, vol. 168 (2), p. 299

[35] Y. J. Shen et al., 1998, „Myelin-associated glycoprotein in myelin and expressed by Schwann cells inhibits axonal regeneration and branching“, Molecular and Cellular Neuroscience, vol. 12, p. 79

[36] M. Bähr and C. Przyrembel, 1995, „Myelin from peripheral and central nervous system is a nonpermissive substrate for axons“, Experimental neurology, vol. 134, p. 87

[37] M. Schäfer et al., 1996, „Disruption of the gene for the myelin-associated glycoprotein improves axonal regrowth along myelin in C57BL/Wld s mice“, Neuron, vol. 16, p. 1107

[38] G. Mukhopadhyay et al., 1994, „A novel role for myelin-associated glycoprotein as an inhibitor of axonal regeneration“, Neuron, vol. 13, p. 757

[39] M. Murakami et al., 1997, „Regulatory functions of phospholipase A 2“, Critical Reviews in Immunology, vol. 17 (3), p. 225

[40] S. Shamash and F. Reichert, 2002, „The cytokine network of Wallerian degeneration: tumor necrosis factor-α, interleukin-1α, and interleukin-1β“, Journal of Neuroscience, vol. 22 (8), p. 3052

[41] F. E. Perrin et al., 2005, „Involvement of monocyte chemoattractant protein-1, macrophage inflammatory protein-1α and interleukin-1β in Wallerian degeneration“, Brain, vol. 128 (4), p. 854

[42] S. De et al., 2003, „Phospholipase A2 plays an important role in myelin breakdown and phagocytosis during Wallerian degeneration.“, Molecular and Cellular Neuroscience, vol. 24 (3), p. 753

[43] S. Chattopadhyay et al., 2006, „Cytokine regulation of MMP-9 in peripheral glia: implications for pathological processes and pain in injured nerve.“, Brain Behavioural Immunology, vol. 21 (5), p. 561

[44] M. Bendszus and G. Stoll, 2003, „Caught in the act: in vivo mapping of macrophage infiltration in nerve injury by magnetic resonance imaging“, The Journal of neuroscience, vol. 24, p. 10892

[45] R. Martini et al., 2008, „Interactions between Schwann cells and macrophages in injury and inherited demyelinating disease“, Glia, vol. 56 (14), p. 1566

[46] F. Reichert and R. Levitzky, 1996, „Interleukin 6 in intact and injured mouse peripheral nerves“, European Journal of Neuroscience, vol. 8 (3), p. 530

96 References

[47] A. Saada et al., 1996, „Granulocyte macrophage colony stimulating factor produced in lesioned peripheral nerves induces the up-regulation of cell surface expression of MAC-2 by macrophages and Schwann cells“, The Journal of Cell Biology, vol. 133(1), p. 159

[48] I. Pineau and S. Lacroix, 2006, „Proinflammatory cytokine synthesis in the injured mouse spinal cord: multiphasic expression pattern and identification of the cell types involved.“, J Comp Neurol, vol. 500 (2), p. 267

[49] A. L. Clatworthy et al., 2995, „Role of peri-axonal inflammation in the development of thermal hyperalgesia and guarding behavior in a rat model of neuropathic pain“, Neuroscience letters, vol. 184, p. 5

[50] H. Skre, 1974, „Genetic and clinical aspects of Charcot‐Marie‐Tooth's disease“, Clinical genetics, vol. 6 (2), p. 98

[51] J. M. Charcot and P. Marie, 1886, „Sur une forme particulière d'atrophie musculaire progressive: souvent familiale débutant par les pieds et les jambes et atteignant plus tard les mains“, Revue de médecine, vol. 6, p. 96

[52] H. H. Tooth, 1886, „The peroneal type of progressive muscular atrophy: A thesis for the degree of MD in the University of Cambridge“

[53] J. M. Vance, 2006, „Mechanisms of disease: a molecular genetic update on hereditary axonal neuropathies“, Nature clinical practice Neurology, vol. 2 (1), p. 45

[54] P. J. Dyck and E. H. Lambert, 1968, „Lower motor and primary sensory neuron diseases with peroneal muscular atrophy: I. Neurologic, genetic, and electrophysiologic findings in hereditary polyneruopathies“, Archives of neurology, vol. 18 (6), p. 603

[55] P. J. Dyck and E. H. Lambert, 1968, „Lower motor and primary sensory neuron diseases with peroneal muscular atrophy: II. Neurologic, genetic, and electrophysiologic findings in various neuronal degenerations“, Archives of neurology, vol. 18 (6), p. 619

[56] R. Ouvrier, 2010, „What can we learn from the history of Charcot‐Marie‐Tooth disease?“, Developmental Medicine & Child Neurology, 2010.,

[57] M. M. Reilly et al., 2011, „Charcot‐Marie‐Tooth disease“, Journal of the Peripheral Nervous System, vol. 16 (1), p. 1

[58] A. Niemann et al., 2006, „Pathomechanisms of mutant proteins in Charcot-Marie-Tooth disease“, NeuroMolecular Medicine, vol. 8 (1), p. 217

[59] A. M. Rossor et al., 2013, „Clinical implications of genetic advances in Charcot–Marie–Tooth disease“, Nature Reviews Neurology, vol. 9, p. 562

[60] P. Berger et al., 2006, „Schwann cells and the pathogenesis of inherited motor and sensory neuropathies (Charcot‐ Marie‐Tooth disease)“, Glia, vol. 54, p. 243

[61] P. Raeymaekers et al., 1991, „Duplication in chromosome 17p11. 2 in Charcot-Marie-Tooth neuropathy type 1a (CMT 1a)“, Neuromuscular disorders, vol. 1 (2), p. 93

[62] J. P. Magyar et al., 1996, „Impaired differentiation of Schwann cells in transgenic mice with increased PMP22 gene dosage“, Journal of Neuroscience, vol. 16 (17), p. 5351

[63] U. Suter and S. S. Scherer, 2003, „Disease mechanisms in inherited neuropathies“, Nature Reviews Neuroscience, vol. 4 (9), p. 714

[64] M. E. Shy et al., 2005, „Hereditary motor and sensory neuropathies“, book

[65] M. Sereda et al., 1996, „A transgenic rat model of Charcot-Marie-Tooth disease“, Neuron, vol. 16, p. 1049

[66] A. M. Robertson et al., 2002, „Comparison of a new pmp22 transgenic mouse line with other mouse models and human patients with CMT1A“, Journal of Anatomy, vol. 200 (4), p. 377

[67] C. Verhamme and R. King, 2011, „Myelin and axon pathology in a long-term study of PMP22-overexpressing mice“, Journal of Neuropathology & Experimental Neurology, vol. 70 (5), p. 368

[68] L. E. Warner et al., 1996, „Clinical phenotypes of different MPZ (P 0) mutations may include Charcot–Marie–Tooth type 1B, Dejerine–Sottas, and congenital hypomyelination“, Neuron, vol. 17 (3), p. 451

[69] R. Martini et al., 1995, „Protein zero (P0)-deficient mice show myelin degeneration in peripheral nerves characteristic of inherited human neuropathies“, Nat Genet, vol. 11 (3), p. 281

[70] L. Wrabetz et al., 2006, „Different intracellular pathomechanisms produce diverse Myelin Protein Zero neuropathies in transgenic mice“, Journal of Neuroscience, vol. 26 (8), p. 2358

97 References

[71] R. Fledrich et al., 2012, „Murine therapeutic models for Charcot-Marie-Tooth (CMT) disease“, British medical bulletin, vol. 1 (25), p. 1

[72] O. Dubourg et al., 2001, „Clinical, electrophysiological and molecular genetic characteristics of 93 patients with X- linked Charcot–Marie–Tooth disease“, Brain, vol. 124, p. 1958

[73] S. S. Scherer et al., 1998, „Connexin 32-null mice develop demyelinating peripheral neuropathy“, Glia, vol. 24 (1), p. 8

[74] L. Jeng and R. J. Balice, 2006, „The effects of a dominant connexin32 mutant in myelinating Schwann cells“, Molecular and Cellular Neuroscience, vol. 32, p. 283

[75] S. S. Scherer et al., 2005, „Transgenic expression of human connexin32 in myelinating Schwann cells prevents demyelination in connexin32-null mice“, Journal of Neuroscience, vol. 25 (6), p. 1550

[76] S. Carenini et al., 2001, „The role of macrophages in demyelinating peripheral nervous system of mice heterozygously deficient in p0“, The Journal of Cell Biology, vol. 152 (2), p. 301

[77] I. Kobsar et al., 2003, „Preserved myelin integrity and reduced axonopathy in connexin32‐deficient mice lacking the recombination activating gene‐1“, Brain, vol. 126, p. 804

[78] B. van den Berg et al., 2014, „Guillain-Barré syndrome: pathogenesis, diagnosis, treatment and prognosis“, vol. 10 (8),

[79] C. Fokke et al., 2013, „Diagnosis of Guillain-Barré syndrome and validation of Brighton criteria“, Brain, vol. 137, p. 33

[80] A. B. Netto et al., 2011, „Mortality in mechanically ventilated patients of Guillain Barré Syndrome“, Annals of Indian Academy of Neurology, vol. 14 (4), p. 262

[81] B. van den Berg et al., 2013, „Mortality in Guillain-Barré syndrome“, Neurology, vol. 80 (18), p. 1650

[82] A. K. Asbury et al., 1969, „The inflammatory lesion in idiopathic polyneuritis.“, Medicine, vol. 294 (7613), p. 199

[83] G. M. McKhann et al., 1993, „Acute motor axonal neuropathy: a frequent cause of acute flaccid paralysis in China“, Annals of Neurology, vol. 33 (4), p. 333

[84] J. W. Griffin et al., 1996, „Pathology of the motor‐sensory axonal Guillain‐Barré syndrome“, Annals of Neurology, vol. 39 (1), p. 17

[85] M. Mori et al., 2001, „Clinical features and prognosis of Miller Fisher syndrome“, Neurology, vol. 56 (8), p. 1104

[86] R. Hughes and D. R. Cornblath, 2005, „Guillain-Barré syndrome“, Medicine, vol. 366, p. 1653

[87] P. Milner et al., 1987, „P0 myelin protein produces experimental allergic neuritis in Lewis rats“, Jorunal of the Neurological Sciences, vol. 79, p. 275

[88] C. M. Gabriel et al., 1998, „Induction of experimental autoimmune neuritis with peripheral myelin protein-22.“, Brain, vol. 121, p. 1895

[89] C. E. Hafer-Macko et al., 1996, „Immune attack on the Schwann cell surface in acute inflammatory demyelinating polyneuropathy“, Annals of Neurology, vol. 39 (5), p. 625

[90] P. Y. Van den Bergh et al., 2010, „Peripheral Nerve Society guideline on management of chronic inflammatory demyelinating polyradiculoneuropathy: report of a joint taskforce of the European Federation of Neurological Societies and the Peripheral Nerve Society“, vol. 15 (1)

[91] J. W. Prineas, 1971, „Demyelination and remyelination in recurrent idiopathic polyneuropathy“, Acta neuropathologica, vol. 18, p. 34

[92] G. Praefcke and H. T. McMahon, 2004, „The dynamin superfamily: universal membrane tubulation and fission molecules?“, Nature Reviews Molecular Cell Biology, vol. 5, p. 133

[93] J. M. Sontag et al., 1994, „Differential expression and regulation of multiple dynamins.“, Journal of Biological Chemistry, vol. 6 (11), p. 4547

[94] N. W. Gray et al., 2003, „Dynamin 3 is a component of the postsynapse, where it interacts with mGluR5 and Homer“, Current Biology, vol. 13, p. 510

[95] T. Nakata et al., 1993, „A novel member of the dynamin family of GTP-binding proteins is expressed specifically in the testis“, Journal of Cell Science, vol. 105 (1), p. 1

98 References

[96] S. G. Clark and D. L. Shurland, 1997, „A dynamin GTPase mutation causes a rapid and reversible temperature- inducible locomotion defect in C. elegans“, Proceedings of the National Academy of Sciences, vol. 94 (19), p. 10438

[97] A. M. van der Bliek and E. M. Meyerowrtz, 1991, „Dynamin-like protein encoded by the Drosophila shibire gene associated with vesicular traffic“, Nature, vol. 351 (6325), p. 411

[98] S. M. Ferguson and P. De Camilli, 2012, „Dynamin, a membrane-remodelling GTPase“, Nature Reviews Molecular Cell Biology, vol. 13, p. 75

[99] Y. W. Liu and S. Neumann, 2011, „Differential curvature sensing and generating activities of dynamin isoforms provide opportunities for tissue-specific regulation“, Proceedings of the National Academy of Sciences, vol. 108 (26), p. 234

[100] A. Raimondi et al., 2011, „Overlapping role of dynamin isoforms in synaptic vesicle endocytosis“, Neuron, vol. 70, p. 1100

[101] B. Marks et al., 2001, „GTPase activity of dynamin and resulting conformation change are essential for endocytosis“, Nature, vol. 410, p. 231

[102] R. Ramachandran et al., 2007, „The dynamin middle domain is critical for tetramerization and higher‐order self‐ assembly“, The EMBO Journal, vol. 26 (2), p. 559

[103] K. Salim et al., 1996, „Distinct specificity in the recognition of phosphoinositides by the pleckstrin homology domains of dynamin and Bruton's tyrosine kinase.“, The EMBO Journal, vol. 15 (22), p. 6241

[104] Y. Vallis et al., 1999, „Importance of the pleckstrin homology domain of dynamin in clathrin-mediated endocytosis“, Current Biology, vol. 9 (5), p. 257

[105] D. E. Klein et al., 1998, „The pleckstrin homology domains of dynamin isoforms require oligomerization for high affinity phosphoinositide binding.“, Journal of Biological Chemistry, vol. 273 (42), p. 27725

[106] S. Chakraborty et al., 2012, „NMR derived model of GTPase effector domain (GED) self association: relevance to dynamin assembly.“, PLoS ONE, vol. 7 (1), p. e30109

[107] P. M. Okamoto et al., 1997, „Role of the basic, proline-rich region of dynamin in Src homology 3 domain binding and endocytosis.“, Journal of Biological Chemistry, vol. 272 (17), p. 11629

[108] H. S. Shpetner et al., 1996, „A binding site for SH3 domains targets dynamin to coated pits.“, Journal of Biological Chemistry, vol. 271 (1), p. 13

[109] E. Cocucci and R. Gaudin, 2014, „Dynamin recruitment and membrane scission at the neck of a clathrin-coated pit“, Molecular Biology of the Cell, vol. 25 (22), p. 3595

[110] K. Faelber et al., 2012, „Structural insights into dynamin-mediated membrane fission“, Structure, vol. 20, p. 1621

[111] H. Cao et al., 1998, „Differential distribution of dynamin isoforms in mammalian cells.“, Molecular Biology of the Cell, vol. 9 (9), p. 2595

[112] Y.-W. Liu et al., 2008, „Isoform and splice-variant specific functions of dynamin-2 revealed by analysis of conditional knock-out cells.“, Molecular Biology of the Cell, vol. 19 (12), p. 5347

[113] A. C. Durieux et al., 2010, „Dynamin 2 and human diseases“, Journal of Molecular Medicine, vol. 88 (4), p. 339

[114] P. N. M. Sidiropoulos et al., 2012, „Dynamin 2 mutations in Charcot–Marie–Tooth neuropathy highlight the importance of clathrin-mediated endocytosis in myelination“, Brain, vol. 135 (5), p. 1395

[115] E. M. van Dam and W. Stoorvogel, 2002, „dynamin-dependent transferrin receptor recycling by endosome-derived clathrin-coated vesicles.“, Molecular Biology of the Cell, vol. 13 (1), p. 169

[116] P. Nicoziani et al., 2000, „Role for dynamin in late endosome dynamics and trafficking of the cation-independent mannose 6-phosphate receptor.“, Molecular Biology of the Cell, vol. 11 (2), p. 481

[117] O. Maier et al., 1996, „Dynamin II binds to thetrans-Golgi network“, Biochemical and Biophysical Research Communications, vol. 223 (2), p. 229

[118] H. S. Shpetner and R. B. Vallee, 1989, „Identification of dynamin, a novel mechanochemical enzyme that mediates interactions between microtubules“, Cell, vol. 59 (3), p. 421

[119] K. Tanabe and K. Takei, 2009, „Dynamic instability of microtubules requires dynamin 2 and is impaired in a Charcot- Marie-Tooth mutant“, The Journal of Cell Biology, vol. 185 (6), p. 939

99 References

[120] H. M. Thompson et al., 2004, „Dynamin 2 binds gamma-tubulin and participates in centrosome cohesion.“, Nature Cell Biology, vol. 6 (4), p. 335

[121] M. M. Kessels and Å. Engqvist, 2001, „Mammalian Abp1, a signal-responsive F-actin–binding protein, links the actin cytoskeleton to endocytosis via the GTPase dynamin“, The Journal of Cell Biology, vol. 153 (2), p. 351

[122] D. A. Schafer et al., 2002, „Dynamin2 and cortactin regulate actin assembly and filament organization“, Current Biology, vol. 12 (21), p. 1852

[123] O. L. Mooren et al., 2009, „Dynamin2 GTPase and cortactin remodel actin filaments“, Journal of Biological Chemistry, vol. 284 (36), p. 23995

[124] E. W. Krueger et al., 2003, „A dynamin-cortactin-Arp2/3 complex mediates actin reorganization in growth factor- stimulated cells.“, Molecular Biology of the Cell, vol. 14 (3), p. 1085

[125] S. Zuchner et al., 2005, „Mutations in the pleckstrin homology domain of dynamin 2 cause dominant intermediate Charcot-Marie-Tooth disease“, Nat Genet, vol. 37 (3), p. 289

[126] M. Bitoun et al., 2005, „Mutations in dynamin 2 cause dominant centronuclear myopathy“, Nat Genet, vol. 37, p. 1207

[127] P. Y. Jeannet et al., 2004, „Clinical and histologic findings in autosomal centronuclear myopathy“, Neurology, vol. 62 (9), p. 1484

[128] A. J. SPIRO et al., 1966, „Myotubular myopathy: persistence of fetal muscle in an adolescent boy“, Archives of neurology, vol. 14 (1), p. 1

[129] D. Fischer et al., 2006, „Characterization of the muscle involvement in dynamin 2-related centronuclear myopathy“, Brain, vol. 129 (6), p. 1463

[130] S. Y. Park and X. Guo, 2014, „Adaptor protein complexes and intracellular transport“, Bioscience Reports, vol. 34, p. 381

[131] J. Hirst et al., 2011, „The Fifth Adaptor Protein Complex“, PLoS Biology, vol. 9 (10), p. e1001170

[132] Y. J. Wang et al., 2003, „Phosphatidylinositol 4 phosphate regulates targeting of clathrin adaptor AP-1 complexes to the Golgi“, Cell, vol. 114 (3), p. 299

[133] T. Gaffry and S. Höning, 2010, „A Large-Scale Conformational Change Couples Membrane Recruitment to Cargo Binding in the AP2 Clathrin Adaptor Complex“, Cell, vol. 141 (7), p. 1220

[134] A. A. Peden et al., 2004, „Localization of the AP-3 adaptor complex defines a novel endosomal exit site for lysosomal membrane proteins“, The Journal of Cell Biology, vol. 164 (7), p. 1065

[135] P. V. Burgos et al., 2010, „Sorting of the Alzheimer's disease amyloid precursor protein mediated by the AP-4 complex“, Developmental Cell, vol. 18 (3), p. 425

[136] T. Simmen et al., 2002, „AP-4 binds basolateral signals and participates in basolateral sorting in epithelial MDCK cells“, Nature Cell Biology, vol. 4, p. 154

[137] T. G. Kutateladze, 2010, „Translation of the phosphoinositide code by PI effectors.“, Nature Chemical Biology, vol. 6 (7), p. 507

[138] Y. Zhu et al., 1998, „ADP-ribosylation factor 1 transiently activates high-affinity adaptor protein complex AP-1 binding sites on Golgi membranes“, Molecular Biology of the Cell, vol. 9 (6), p. 1323

[139] I. Gaidarov et al., 1999, „Phosphoinositide–Ap-2 Interactions Required for Targeting to Plasma Membrane Clathrin- Coated Pits“, The Journal of Cell Biology, vol. 146 (4), p. 755

[140] V. Faúndez et al., 1989, „A function for the AP3 coat complex in synaptic vesicle formation from endosomes“, Cell, vol. 93 (3), p. 423

[141] J. Hirst et al., 1999, „Characterization of a fourth adaptor-related protein complex“, Molecular Biology of the Cell, vol. 10 (8), p. 2787

[142] J. H. Keen et al., 1979, „Clathrin-coated vesicles: isolation, dissociation and factor-dependent reassociation of clathrin baskets“, Cell, vol. 16 (2), p. 303

[143] J. H. Keen, 1987, „Clathrin assembly proteins: affinity purification and a model for coat assembly.“, The Journal of Cell Biology, vol. 105, p. 1989

100 References

[144] J. Hirst and M. S. Robinson, 1998, „Clathrin and adaptors“, Biochimica et Biophysica Acta - Molecular Cell Research, vol. 1404, p. 173

[145] G. Praefcke et al., 2004, „Evolving nature of the AP2 α‐appendage hub during clathrin‐coated vesicle endocytosis“, The EMBO Journal, vol. 23, p. 4371

[146] D. J. Owen et al., 1999, „A structural explanation for the binding of multiple ligands by the α-adaptin appendage domain“, Cell, vol. 97, p. 805

[147] E. M. Schmid et al., 2006, „Role of the AP2 β-appendage hub in recruiting partners for clathrin-coated vesicle assembly“, PLoS Biol, vol. 4 (9), p. 1532

[148] D. J. Owen et al., 2004, „Adaptors for clathrin coats: structure and function“, Annu Rev Cell Dev Biol, vol. 20, p. 153

[149] J. S. Bonifacino et al., 2014, „Adaptor proteins involved in polarized sorting“, The Journal of Cell Biology, vol. 204 (1), p. 7

[150] W. Matsui and T. Kirchhausen, 1990, „Stabilization of clathrin coats by the core of the clathrin-associated protein complex AP-2“, Biochemistry, vol. 29, p. 10791

[151] S. Höning et al., 2005, „Phosphatidylinositol-(4, 5)-bisphosphate regulates sorting signal recognition by the clathrin- associated adaptor complex AP2“, Molecular cell, vol. 18, p. 519

[152] D. Ricotta et al., 2002, „Phosphorylation of the AP2 μ subunit by AAK1 mediates high affinity binding to membrane protein sorting signals“, The Journal of Cell Biology, vol. 156 (5), p. 791

[153] H. T. McMahon and E. Boucrot, 2011, „Molecular mechanism and physiological functions of clathrin-mediated endocytosis“, Nature Reviews Molecular Cell Biology, vol. 12 (8), p. 517

[154] P. N. Dannhauser and E. J. Ungewickell, 2012, „Reconstitution of clathrin-coated bud and vesicle formation with minimal components“, Nature Cell Biology, vol. 14 (6), p. 634

[155] V. Haucke and P. De Camilli, 1999, „AP-2 recruitment to synaptotagmin stimulated by tyrosine-based endocytic motifs“, Science, vol. 285 (5431), p. 5431

[156] A. Yu et al., 2007, „Association of Dishevelled with the clathrin AP-2 adaptor is required for Frizzled endocytosis and planar cell polarity signaling“, Developmental Cell, vol. 12 (1), p. 129

[157] M. Kazazic et al., 2009, „Epsin 1 is involved in recruitment of ubiquitinated EGF receptors into clathrin‐coated pits“, Traffic, vol. 10, p. 235

[158] M. Kaksonen et al., 2006, „Harnessing actin dynamics for clathrin-mediated endocytosis“, Nature Reviews Molecular Cell Biology, vol. 7, p. 404

[159] A. Fotin et al., 2004, „Structure of an auxilin-bound clathrin coat and its implications for the mechanism of uncoating“, Nature, vol. 432, p. 649

[160] I. Rapoport et al., 2008, „A motif in the clathrin heavy chain required for the Hsc70/auxilin uncoating reaction“, Molecular Biology of the Cell, vol. 19 (1), p. 405

[161] J. Wang and K. Pantopoulos, 2011, „Regulation of cellular iron metabolism“, Biochemical Journal, vol. 434 (3), p. 365

[162] N. C. Andrews, 1999, „Disorders of iron metabolism“, New England Journal of Medicine, vol. 341 (26), p. 1986

[163] H. Gunshin et al., 1997, „Cloning and characterization of a mammalian proton-coupled metal-ion transporter“, Nature, vol. 388, p. 482

[164] A. Donovan et al., 2000, „Positional cloning of zebrafish ferroportin1 identifies a conserved vertebrate iron exporter.“, Nature, vol. 403 (6771), p. 776

[165] P. Ponka et al., 1998, „Function and regulation of transferrin and ferritin.“, Seminars in hematology, vol. 35 (1), p. 35

[166] D. M. Sipe and R. F. Murphy, 1991, „Binding to cellular receptors results in increased iron release from transferrin at mildly acidic pH.“, Journal of Biological Chemistry, vol. 266 (13), p. 8002

[167] P. K. Bali et al., 1991, „A new role for the transferrin receptor in the release of iron from transferrin“, Biochemistry, vol. 30 (2), p. 324

[168] R. S. Ohgami et al., 2005, „Identification of a ferrireductase required for efficient transferrin-dependent iron uptake in erythroid cells“, Nat Genet, vol. 37 (11), p. 1264

101 References

[169] A. M. Koorts and M. Viljoen, 2007, „Ferritin and ferritin isoforms I: Structure–function relationships, synthesis, degradation and secretion“, Archives of physiology and biochemistry, vol. 113 (1), p. 30

[170] E. C. Theil, 2004, „Iron, ferritin, and nutrition“, Annual Review of Nutrition , vol. 24, p. 327

[171] A. Dautry, 1983, „pH and the recycling of transferrin during receptor-mediated endocytosis“, Proceedings of the National Academy of Sciences, vol. 80, p. 2258

[172] R. S. Eisenstein, 2000, „Iron regulatory proteins and the molecular control of mammalian iron metabolism“, Annual review of nutrition, vol. 20, p. 627

[173] C. Camaschella, 2015, „Iron-deficiency anemia“, New England Journal of Medicine, vol. 372 (19), p. 1832

[174] N. Kabakus et al., 2002, „Reversal of iron Deficiency Anemia‐induced Peripheral Neuropathy by iron Treatment in Children with iron Deficiency Anemia“, Journal of Tropical Pediatrics, vol. 48, p. 204

[175] D. Garcia et al., 2011, „Algorithms for the diagnosis and treatment of restless legs syndrome in primary care“, BioMed Central Neurology, vol. 11 (28), p. 2

[176] S. Altamura and M. U. Muckenthaler et al., 2009, „Iron toxicity in diseases of aging: Alzheimer‚s disease, Parkinson‘s disease and atherosclerosis“, Journal of Alzheimer´s Disease, vol. 16, p. 879

[177] S. Srinivas et al., 2001, „Cre reporter strains produced by targeted insertion of EYFP and ECFP into the ROSA26 locus“, BioMed Central Developmental Biology, vol. 1 (4)

[178] R. I. Craggs et al., 1984, „The effect of suppression of macrophage activity on the development of experimental allergic neuritis“, Acta neuropathologica, vol. 62 (316), p. 316

[179] H. P. HARTUNG et al., 1988, „Suppression of experimental autoimmune neuritis by the oxygen radical scavengers superoxide dismutase and catalase“, Annals of Neurology, vol. 23 (5), p. 453

[180] H. P. HARTUNG et al., 1988, „The role of macrophages and eicosanoids in the pathogenesis of experimental allergic neuritis“, Brain, vol. 111 (5), p. 1039

[181] C. Somandin, 2013, „Dynmin 2 Ablation in Adult Schwann Cells Affectts the Integrity of Peripheral Nerves“, Disseration ETH, vol. 20908

[182] C. Vleggeert-Lankamp and G. De Ruiter, 2005, „Type grouping in skeletal muscles after experimental reinnervation: another explanation“, Eur J Neurosci, vol. 21, p. 1249

[183] J. Jeker, 2014, „Dnm2 Deficient Adult Schwann Cells Undergo Apoptosis Due To iron Stress“, Semester thesis, ETH Zurich

[184] A. Motley, 2003, „Clathrin-mediated endocytosis in AP-2–depleted cells“, The Journal of Cell Biology, vol. 162 (5), p. 909

[185] J. Jeker, 2016, „The Role of the AP-2 Adaptor Protein Complex in Adult Myelinating Schwann Cells“, Master thesis, ETH Zurich

[186] B. D. Grant and J. G. Donaldson, 2009, „Pathways and mechanisms of endocytic recycling“, Nature Reviews Molecular Cell Biology, vol. 10, p. 597

[187] S. Levi and C. Taveggia, 2014, „Iron Homeostasis in Peripheral Nervous System, Still a Black Box?“, Antioxidants & Redox Signaling, vol. 21 (4), p. 634

[188] D. P. Leone et al., 2003, „tamoxifen-inducible glia-specific Cre mice for somatic mutagenesis in oligodendrocytes and Schwann cells“, Molecular and Cellular Neuroscience, vol. 22 (4), p. 430

[189] E. M. Raynor et al., 1995, „Differentiation between axonal and demyelinating neuropathies: identical segments recorded from proximal and distal muscles“, Muscle & Nerve, vol. 18 (4), p. 402

[190] G. Midroni and J. M. Bilbao, 1995, „Biopsy diagnosis of peripheral neuropathy“

[191] L. Lubińska, 1977, „Early course of Wallerian degeneration in myelinated fibres of the rat phrenic nerve“, Brain Research, vol. 130, p. 47

[192] J. M. Schro, 1972, „Altered ratio between axon diameter and myelin sheath thickness in regenerated nerve fibers“, Brain Research, vol. 45, p. 49

102 References

[193] M. Swash and M. S. Schwartz, 2013, „Neuromuscular diseases: a practical approach to diagnosis and management“, vol. 74 (11)

[194] S. Hall, 2005, „The response to injury in the peripheral nervous system“, Bone & Joint Journal, vol. 87 (10), p. 1309

[195] G. Stoll et al., 2002, „Degeneration and regeneration of the peripheral nervous system: from Augustus Waller's observations to neuroinflammation“, Journal of the Peripheral Nervous System, vol. 7 (1), p. 13

[196] M. Mäurer et al., 2002, „Role of immune cells in animal models for inherited neuropathies: facts and visions“, Journal of Anatomy, vol. 200 (4), p. 405

[197] D. Klein and R. Martini, 2015, „Myelin and macrophages in the PNS: An intimate relationship in trauma and disease.“, Brain Research, vol. 1641, p. 130

[198] M. Mäurer and R. Gold, 2002, „Animal models of immune-mediated neuropathies“, Current opinion in neurology, vol. 15 (5), p. 617

[199] G. Stoll and B. C. Kieseier, 2005, „Experimental autoimmune neuritis“, Periphearl Neuropathy, p. 609

[200] R. Kiefer et al., 2001, „The role of macrophages in immune-mediated damage to the peripheral nervous system“, Progress in neurobiology, vol. 64, p. 109

[201] J. Groh et al., 2015, „Inflammation in the pathogenesis of inherited peripheral neuropathies“, Neuroinflammation, p. 123

[202] B. Kohl et al., 2010, „MCP-1/CCL2 modifies axon properties in a PMP22-overexpressing mouse model for Charcot- Marie-Tooth 1A neuropathy“, The American Journal of Pathology, vol. 176 (3), p. 1390

[203] S. Fischer et al., 2008, „Monocyte chemoattractant protein-1 is a pathogenic component in a model for a hereditary peripheral neuropathy“, Molecular and Cellular Neuroscience, vol. 37, p. 359

[204] J. Groh et al., 2010, „Attenuation of MCP-1/CCL2 expression ameliorates neuropathy in a mouse model for Charcot– Marie–Tooth 1X“, Human Molecular Genetics, vol. 19 (18), p. 3530

[205] J. Groh et al., 2012, „Colony-stimulating factor-1 mediates macrophage-related neural damage in a model for Charcot–Marie–Tooth disease type 1X“, Brain, vol. 135, p. 88

[206] A. Veves et al., 2012, „ The diabetic foot: medical and surgical management“, book

[207] R. Don et al., 2007, „Foot drop and plantar flexion failure determine different gait strategies in Charcot-Marie-Tooth patients“, Clinical genetics, vol. 22 (8), p. 905

[208] A. E. Patla, 2003, „Strategies for dynamic stability during adaptive human locomotion“, Engineering in Medicine and Biology Magazine, vol. 22 (2), p. 48

[209] E. Joosten and W. B. Veldhuis, 2004, „Collagen containing neonatal astrocytes stimulates regrowth of injured fibers and promotes modest locomotor recovery after spinal cord injury“, Journal of Neuroscience, vol. 77, p. 127

[210] R. Deumens et al., 2007, „The CatWalk gait analysis in assessment of both dynamic and static gait changes after adult rat sciatic nerve resection“, Journal of Neuroscience Methods, vol. 164, p. 120

[211] W. Hendriks et al., 2006, „Profound differences in spontaneous long-term functional recovery after defined spinal tract lesions in the rat“, Journal of Neurotrauma, vol. 23 (1), p. 18

[212] F. Hamers et al., 2006, „CatWalk-assisted gait analysis in the assessment of spinal cord injury“, Journal of Neurotrauma, vol. 23 (4), p. 537

[213] T. J. van Laar and W. H. Gispen, 2001, „Automated Quantitative Gait Analysis During Overground Locomotion in the Rat: Its Application to Spinal Cord Contusion and Transection Injuries“, Journal of Neurotrauma, vol. 18 (2), p. 187

[214] A. Kochański et al., 2003, „Focally folded myelin in Charcot-Marie-Tooth type 1B disease is associated with Asn131Lys mutation in myelin protein zero gene: short report.“, European Journal of Neurology, vol. 10 (5), p. 547

[215] G. M. Fabrizi et al., 2000, „Focally folded myelin in Charcot-Marie-Tooth neuropathy type 1B with Ser49Leu in the myelin protein zero.“, Acta neuropathologica, vol. 100 (3), p. 299

[216] M. Iida et al., 2011, „A novel MPZ mutation in Charcot-Marie-Tooth disease type 1B with focally folded myelin and multiple entrapment neuropathies.“, Neuromuscul Disord, vol. 22 (2), p. 166

103 References

[217] K. W. Chung et al., 2011, „Two recessive intermediate Charcot-Marie-Tooth patients with GDAP1 mutations.“, Journal of the Peripheral Nervous System, vol. 16 (2), p. 143

[218] T. D. Bird et al., 1997, „Clinical and pathological phenotype of the original family with Charcot-Marie-Tooth type 1B: a 20-year study.“, Annals of Neurology, vol. 41 (4), p. 463

[219] A. Quattrone et al., 1996, „Autosomal recessive hereditary motor and sensory neuropathy with focally folded myelin sheaths: clinical, electrophysiologic, and genetic aspects of a large family.“, Neurology, vol. 46 (5), p. 1318

[220] A. Bolis et al., 2005, „Loss of Mtmr2 phosphatase in Schwann cells but not in motor neurons causes Charcot-Marie- Tooth type 4B1 neuropathy with myelin outfoldings“, Journal of Neuroscience, vol. 25 (37), p. 8567

[221] A. Bolino et al., 2004, „Disruption of Mtmr2 produces CMT4B1-like neuropathy with myelin outfolding and impaired spermatogenesis“, The Journal of Cell Biology, vol. 167 (4), p. 711

[222] K. Tersar et al., 2007, „Mtmr13/Sbf2-deficient mice: an animal model for CMT4B2“, vol. 16 (24), p. 2991

[223] M. Horn et al., 2012, „Myelin is dependent on the Charcot-Marie-Tooth Type 4H disease culprit protein FRABIN/FGD4 in Schwann cells.“, Brain, vol. 135 (12), p. 3567

[224] F. L. Robinson and I. R. Niesman, 2008, „Loss of the inactive myotubularin-related phosphatase Mtmr13 leads to a Charcot–Marie–Tooth 4B2-like peripheral neuropathy in mice“, Proceedings of the National Academy of Sciences, vol. 105 (12), p. 4916

[225] A. Nodari et al., 2008, „Alpha6beta4 integrin and dystroglycan cooperate to stabilize the myelin sheath.“, Journal of Neuroscience, vol. 28 (26), p. 6714

[226] S. M. Lee et al., 2013, „Motor and sensory neuropathy due to myelin infolding and paranodal damage in a transgenic mouse model of Charcot–Marie–Tooth disease type 1C“, vol. 22 (9), p. 1755

[227] I. Naba et al., 2000, „Onion-bulb formation after a single compression injury in the macrophage scavenger receptor knockout mice.“, Experimental neurology, vol. 166 (1), p. 83

[228] G. Lauria and R. Lombardi, 2007, „Skin biopsy: a new tool for diagnosing peripheral neuropathy“, British Medical Journal, vol. 334 (7604), p. 1159

[229] T. Chung et al., 2014, „Peripheral Neuropathy: Clinical and Electrophysiological Considerations“, Neuroimaging Clinics of North America, vol. 24 (1), p. 49

[230] L. Tacchini et al., 2002, „Transferrin receptor gene expression and transferrin‐bound iron uptake are increased during postischemic rat liver reperfusion“, Hepatology, vol. 36 (1), p. 103

[231] W. Xu et al., 2015, „Lethal Cardiomyopathy in Mice Lacking Transferrin Receptor in the Heart“, Cell Reports, vol. 13, p. 1

[232] T. Barrientos et al., 2015, „Metabolic catastrophe in mice lacking transferrin receptor in muscle“, EBioMedicine, vol. 2 (11), p. 1705

[233] P. Matak et al., 2016, „Disrupted iron homeostasis causes dopaminergic neurodegeneration in mice“, Proceedings of the National Academy of Sciences, vol. 113 (13), p. 3428

[234] J. A. Pereira et al., 2010, „Dicer in Schwann cells is required for myelination and axonal integrity“, Journal of Neuroscience, vol. 30 (9), p. 6763

[235] M. Ghidinelli et al., 2016, „Clathrin-mediated endocytosis is required for correct meyleination in peripheral nad central nervous system“, personal communication

[236] C. B. Harper et al., 2013, „Targeting membrane trafficking in infection prophylaxis: dynamin inhibitors“, Trends in Cell Biology, vol. 23 (2), p. 89

[237] M. Chircop et al., 2011, „Inhibition of dynamin by dynole 34-2 induces cell death following cytokinesis failure in cancer cells“, vol. 10 (9)

[238] R. Martinez-Vivot et al., 2015, „DMT1 iron uptake in the PNS: bridging the gap between injury and regeneration“, Metallomics, vol. 7, p. 1381

[239] B. Todorich et al., 2008, „Tim‐2 is the receptor for H‐ferritin on oligodendrocytes“, Journal of Neurochemistry, vol. 107 (6), p. 1495

104 References

[240] S. M. Hall and P. L. Williams, 1971, „The distribution of electron-dense tracers in peripheral nerve fibres“, Journal of Cell Science, vol. 8 (2), p. 541

[241] C. Harding et al., 1983, „Receptor-mediated endocytosis of transferrin and recycling of the transferrin receptor in rat reticulocytes.“, The Journal of Cell Biology, vol. 97, p. 329

[242] A. Donovan et al., 2005, „The iron exporter ferroportin/Slc40a1 is essential for iron homeostasis“, Cell metabolism, vol. 1 (3), p. 191

[243] H. Gunshin et al., 2005, „Slc11a2 is required for intestinal iron absorption and erythropoiesis but dispensable in placenta and liver“, The Journal of Clinical Investigation, vol. 115 (5), p. 1258

[244] J. Jung et al., 2011, „Transient lysosomal activation is essential for p75 nerve growth factor receptor expression in myelinated Schwann cells during Wallerian degeneration“, Anatomy & Cell Biology, vol. 44 (1), p. 41

[245] C. Salis et al., 2012, „Iron and holotransferrin induce cAMP-dependent differentiation of Schwann cells“, Neurochemistry International, vol. 61, p. 798

[246] M. Neumann et al., 2009, „Assessing gait impairment following experimental traumatic brain injury in mice“, Journal of Neuroscience Methods, vol. 176 (1), p. 34

[247] D. Bausch-Fluck et al., 2012, „Cell surface capturing technologies for the surfaceome discovery of hepatocytes.“, Methods Mol Biol, vol. 909, p. 1

[248] B. Wollscheid et al., 2009, „Mass-spectrometric identification and relative quantification of N-linked cell surface glycoproteins.“, Nat Biotechnol, vol. 27 (4), p. 378

[249] T. Moos and K. Mollgard, 1993, „A sensitive post-DAB enhancement technique for demonstration of iron in the central nervous system“, Histochemistry, vol. 99, p. 471

105 References

106 Curriculum Vitae

7 Curriculum Vitae

Personal Data Name Gerber First Name Daniel Address Wehntalerstrasse 521, 8046 Zürich Phone 0041 79 723 21 52 E-Mail [email protected]

Date of Birth October 3, 1987 Hometown Zürich, ZH, Switzerland Marital Status Single

Education 2012 - present Doctoral studies in biology, ETH Zürich, research group Prof. Dr. U. Suter 2010 - 2012 Master in biology, ETH Zürich Semester Thesis 1: „Biochemical Analysis of Membrane Integration of Human Litaf“ Semester Thesis 2: „Are the CPDs or the 6-4 photoproducts responsible for the Rad53 checkpoint response? “ Master Thesis: „Litaf Localization Studies / The Litaf Knockout Mouse Exhibits Increased Macrophage Migration Capabilities“ 2007 - 2010 Bachelor in biology, ETH Zürich 2002 - 2006 Kantonsschule Zürich Oerlikon

Additional Education 2014 – present Lehrdiplom für Maturitätsschulen 2010 - 2012 Head of Verkehrsgruppe fire brigade Wallisellen 2009 Feuerwehr Verkehrs-Unteroffiziers-Kurs 2007 Joining fire brigade Wallisellen / Fachkurs Verkehrssoldat 2006 - 2007 Rekrutenschule as Verkehrssoldat

Working Experience 2007-2009 Müllers Students-Coaching GmbH

Scientific Publications Somandin C., Gerber D., Pereira J. A., Horn M. and Suter U. (2012), LITAF (SIMPLE) regulates Wallerian degeneration after injury but is not essential for peripheral nerve development and maintenance: Implications for Charcot-Marie-Tooth disease. GLIA. doi: 10.1002/glia.22371

Languages German (Mother tongue) English (Fluent) French (Basic knowledge)

Computer skills Microsoft Office Products (MS Word, MS Excel, MS PowerPoint), Graphic Programs (Adobe Photoshop & Illustrator, advanced skills), Statistical Software (GraphPad Prism), Building and administrating web pages (Web design, html, basic flash, javascript and php knowledge), Operating various laboratory software of the ETH

Interests and Hobbies Skiing, Diving, Canyoning, Cooking, Traveling

107 Curriculum Vitae

108 Acknowledgement

8 Acknowledgement

This study would not have been possible without the support and help from many people over the last four years. My thanks belong to all of them. Some people I would like to acknowledge in particular. First of all I want to thank Prof. Dr. Ueli Suter for giving me the great opportunity to work in this group. I would like to thank him for his continuous support, advice and supervision, as well as for providing an outstanding working environment which left nothing to be desired. I would like to thank my whole thesis committee, Prof. Dr. Claire Jacob, Prof. Dr. Bernd Wollscheid and Dr. Axel Niemann for their constant support and advice. Special thanks belong to Dr. Christian Somandin, my supervisor and mentor during my semester and master thesis, who taught me patiently everything I needed to start a PhD. At this point I would like to thank him once more for the excellent shape of the dynamin 2 project which I had the privilege to take over [181]. I hope to have fulfilled his expectations as well as my student did the mine. Many thanks go to Joanne Jeker, who I had the pleasure to supervise during her semester and master thesis. She did not only brighten up my days with her cheerful nature, but also contributed substantially to the work presented in this thesis [183, 185]. Other special thanks go to my colleague and dear friend Gianluca Figlia. Not only for the outstanding scientific input but also for the priceless time we spent together in and outside of the lab. He started his PhD the same day as I did and soon became like a brother to me. Further I would like to thank all the current and past members of the Suter group for the great working environment – scientifically and socially. Especially I like to mention Dr. Jorge Pereira, not only for his direct contribution regarding the electron microscopy, but also for his intellectual inputs and his calm nature. Another lab-member that I would like to thank specially is Dr. Ned Mantei, for critically reading this thesis and for sharing his decades of experience and his always good moods with us. Further thanks go to Prof. Klaus Toyka and PD Dr. Carsten Wessig from the University of Würzburg, who performed the electrophysiology on the mice. I would like to thank Damaris Bausch-Fluck from the Wollscheid group for the fruitful collaboration analyzing the cell surface proteins. I also would like to acknowledge all the facilities that that provided indispensable services. Our mouse facility EPIC, especially Susanne Freedrich who made sure everything runs smoothly, the Functional Genomics Center Zurich, the Flow Cytometry Core Facility as well as our microscope facility ScopeM. Rolf Huber I would like to thank for maintaining the institute’s technical environment in great shape. Finally, I would like to thank my whole family and my girlfriend for their endless and unconditional support.

109