The Pennsylvania State University The Graduate School

Eberly College of Science

MICROTUBULE DESTABLIZERS CONTROL

AND DENDRITE DISASSEMBLY AFTER INJURY

AND DURING PRUNING

A Dissertation in

Biochemistry, Microbiology, and Molecular Biology

by

Juan Tao

© 2014 Juan Tao

Submitted in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

December 2014

ii

The dissertation of Juan Tao was reviewed and approved* by the following:

Melissa M. Rolls Associate Professor of Biochemistry and Molecular Biology Dissertation Advisor Chair of Committee

Lorraine Santy Associate Professor of Biochemistry and Molecular Biology

Zhi-Chun Lai Professor of Biology Professor of Biochemistry and Molecular Biology

Richard W. Ordway Associate Professor of Biology

Scott Selleck Professor of Biochemistry and Molecular Biology Head of the Department of Biochemistry and Molecular Biology

*Signatures are on file in the Graduate School

iii

Abstract

A extends one axon and several dendrites far from its cell body, leaving these extremely thin processes subject to various insults, including physical damages, strokes, and neurodegenerative diseases during aging. Axon degeneration after injury, called Wallerian degeneration, was discovered in mice two decades ago, yet few molecular mechanisms that control the degeneration process have been identified. To study injury-induced and developmental degeneration, our lab used Drosophila dendritic arborization (da) . These neurons are classified into 4 types based on their dendrite branch complexity. I used the most simple ddaE classI and most complex ddaC classIV neurons in the experiments. DdaC neurons go through dendrite remodeling during the pupae stage. This characteristic of ddaC neurons allows me to compare three types of degeneration, including injury-induced axon degeneration, injury-induced dendrite degeneration and developmental pruning, in the same cell type on a single cell level of a living animal. Our lab developed a laser-cutting assay to precisely sever the dendrites or the proximal to the in whole, live Drosophila larvae. I observed that both axons and dendrites are able to initiate a rapid intrinsic program to clear the severely damaged part in a short period of time. Interestingly, I found that these processes looked similar to developmental pruning, in which dendrites of ddaC neurons are severed near soma and then quickly degraded within

18 hours after pupae formation. Due to this morphological similarity, I first wanted to know whether these processes share common machineries. I tested known pathways that have been reported to be involved in dendrite pruning. However, I discovered that none of these pathways had an effect on injury-induced degeneration. I concluded that axons and dendrites each have their own disassembly machinery that is activated by damage and is distinct from the ones employed in dendrite pruning.

iv

I next wanted to investigate the intracellular mediators involved in each injury-induced degeneration. I have observed fragmentation of microtubules during axon and dendrite degeneration.

Additionally, our lab has done extensive work on microtubule polarity research. I decided to focus on the regulation of microtubules during degeneration. As an important component of neuron cytoskeleton, microtubules play a crucial role in cargo transport and cell shape maintenance. The role of microtubules in neuronal degeneration remains mysterious. I performed a genetic screen of all the potential microtubule severing proteins and microtubule depolymerizing proteins, and I identified two proteins, klp59C and fidgetin, involved in neuron degeneration. Klp59C, a microtubule-depolymerizing protein, serves as a general microtubule regulator involved in all types of degeneration. In contrast, fidgetin, a microtubule severing protein, is specifically required in dendrite degeneration. I further confirmed the RNAi result by repeating the same cutting assay in mutant and deficiency flies.

Additionally, I investigated the working mechanism of klp59C and fidgetin through an EB1 comet assay. I found that EB1 dynamics lasted longer in klp59C knockdown neurons while EB1 dynamics disappeared around the same time in controls and fidgetin RNAi neurons. I thus concluded that fidgetin severs the microtubule in the middle to generate short fragments and klp59C further depolymerizes the microtubule fragments from the plus ends.

Through a large-scale screen of potential pathways responsible for protein and organelle turnover during degeneration, I also discovered a new role of Rab GTPase. Rab5 or Rab11 GFP, which are usually used as an endocytosis marker, can serve as an energy monitor in live cells. I first observed that Rab5 GTP altered its punctate morphology to a diffused distribution 3.5h after axon severing. I also observed that all the other Rab GTPase family members, including Rab7, Rab8 and Rab11, share the same pattern after axon injury. Furthermore, I confirmed the observation that Rab protein changes

v its morphology after ATP depletion in Hela cell cultures. This is the first time that the function of Rab as an energy indicator has been observed.

In my work, I also examined whether the autophagy pathway and the endocytosis pathway are required for neuron degeneration. Down regulating the key proteins in both pathways through RNAi did not delay injury- induced axon or dendrite degeneration. However, I discovered that a decrease in the protein level of VPS4 and shrb (ESCRTIII in fly) significantly affects normal developmental pruning. This result provides additional evidence that developmental pruning and injury –induced degeneration are distinct processes, which involve different machineries.

My work presented here demonstrated that microtubule regulators play important roles in degrading injured axons or dendrites. Klp59C plays a general role in all types of degeneration while fidgetin only acts in dendrite degeneration. Future work elucidating other pathways responsible for the degradation of other components of neurons will be crucial to dissect the comprehensive network of neuron disassembly after injury or in developmental settings.

vi

Table of Contents

List of Figures ...... x

Acknowledgements...... xiii

Chapter 1. Introduction…………………………………………………………………………….1

Overview of neuron degeneration………………………………...... 1

Comparison of injury-induced degeneration and developmental pruning………………...7

Ubiquitin-proteasom System (UPS) in neuron degeneration...... 7

Wlds distinguishes the two processes...... 8

Other players separate the two processes...... 9

Microtubule degradation in a degeneration neuron...... 10

Energy maintenance and neuron degeneration...... 12

Summary…………………………………………………………………………………..15

References………………………………...... 17

Chapter 2. Dendrites have an active program of injury-induced degeneration that is molecularly distinct from developmental pruning...... 21

Abstract...... 21

Introduction...... 21

Results...... 23

Dendrites undergo beading and clearance within 24 hours after severing...... 23

Dendrite degeneration is blocked by overexpression of Wlds or UBP2...... 27

Dendrite degeneration is independent of mitochondria...... 29

Dendrite degeneration is independent of apoptosis machinery...... 30

vii

The pruning machinery is not required for injury-induced dendrite degeneration...... 35 Discussion...... 37

Materials and Methods, Author Information, Acknowledgements...... 40

References...... 42

Supplementary Materials...... 44

Chapter 3. Kinesin-13 terminates microtubule dynamics in degenerating axons before energy collapse...... 46

Abstract...... 46

Introduction...... 47

Results...... 48

Rab proteins can be used to time energy collapse in individual axon...... 48

Microtubule growth terminates before energy collapse ...... 58

Wlds delays energy collapse, but not termination of microtubule dynamics, for day...... 59

Klp59C is required for normal timing of axon degeneration...... 63

Klp59C terminates microtubule dynamics after axon severing...... 64

Discussion...... 66

Materials and Methods, Acknowledgements...... 69

References...... 73

Supplementary Materials...... 75

Chapter 4. KLp59C and fidgetin control dendrite disassembly after injury and during pruning...... 80

Abstract...... 80

viii

Introduction...... 81

Results...... 83

Identification of microtubule regulators involved in degeneration...... 83

Fidgetin functions specifically in injury-induced dendrite degeneration...... 87

Klp59C function is required for timely dendrite degeneration, as well as dendrite pruning...... 90

Klp59C does not delay degeneration by activating known stabilization pathways...... 92

Increaased microtubule stability before injury is unlikely to account for delayed degeneration when Klp59C or fidgetin is reduced ...... 94

Dendrite structure is normal before injury in neurons with reduced fidgetin...... 97

Microtubule behaviors are normal in uninjured fidgetin knockdown neurons...... 98

Reduction of Klp59C but not fidgetin delays loss of dynamic microtubules in dendrite after injury...... 101

Targeting Klp59C and fidgetin together exacerbates delayed dendrite clearance...... 102

Discussion...... 105

Materials and Methods, Abbreviation, Acknowledgements...... 108

References...... 112

Supplementary Materials...... 114

Chapter 5. Conclusions and future avenues of research...... 117

Summary...... 117

Differences between injury-induced degeneration and developmental pruning...... 117

Neuronal functions of microtubule depolymerizing proteins and severing proteins……118

Regulation of microtubule severing proteins...... 119

ix

Further questions in neuron degeneration research...... 121

Closing remarks...... 123

x

List of Figures

Chapter 1

Figure 1. Schematic of major steps of injury-induced axon and dendrite degeneration.....4

Figure 2. Diagram of microtubule regulators required in axon and dendrite

degeneration...... 11

Table 1. Summary of experimental paradigms of neuronal injury...... 6

Table 2. Cellular processes that generate ATP and consume ATP...... 12

Chapter 2

Figure 1. Dendrite degenerate rapidly after severing...... ……………….26

Figure 2. Expression of Wlds or UBP2 blocks dendrite degeneration...... 28

Figure 3. Mitochondria are not required locally for dendrite degeneration...... 30

Figure 4. Pruning of ddaC is delayed by Wlds and UBP2...... 31

Figure 5. Apoptosis machinery is required for dendrite pruning, but not dendrite degeneration...... 33

Figure 6. Specific players required for dendrite pruning are not required for dendrite degeneration...... 36

Figure S1. Dendrite of ddaE neuron degenerate rapidly after severing...... 44

Figure S2. Targeting miro by RNAi reduces the number of mitochondria in dendrites...... 44

Chapter 3

Figure 1. Rab proteins redistribute rapidly in reponse to H2O2 induced ATP depletion………………………………………………………………………………….51

Figure 2. Rab proteins respond similarly to 2-Deoxy-D-glucose caused ATP deprivation...... 52

Figure 3. Energy collapse occurs 3 hours after axon severing...... 54

xi

Figure 4. Disturbance of endogenous Nmnat accelerates Rab diffuse after injury……...57

Figure 5. Cessation of microtubule growth is in early event in axon degeneration compared to energy collapse and membrane beading...... 59

Figure 6. Wlds delays energy collapse, but not termination of microtubule dynamics, for days...... 61

Figure 7. Klp59C is required for timely axon degeneration...... 65

Figrue 8. Klp59C is required for early termination of microtubule dynamics after axon injury...... 66

Figure S1. Rab5-YFP becomes diffuse about 3h after axon severing in the ddaC neuron...... 75

Figure S2. Rab proteins exchange off membrane in minutes...... 76

Figure S3. Mitochondria and LAMP remain punctate 4h after axon severing...... 77

Figure S4. Rab5 gain or loss of function does not affect degeneration timing...... 78

Figure S5. Microtubule regulators were tested for a role in axon degneration...... 79

Chapter 4

Figure 1. Identification of microtubule regulators required for clearing axons and dendrites...... 86

Figure 2. Fidgetin is specifically required for timely dendrite degeneration...... 89

Figure 3. Klp59C is required for normal dendrite degeneration and dendrite pruning...... 91

Figure 4. Delay degeneration in Klp59C knockdown neurons does not rely on activation of known stabilization pathways...... 93

Figure 5. Microtubule stabilization before injury does not account for delayed dendrite degeneration in neurons with reduced klp59C or fidgetin...... 96

Figure 6. Dendrite branch patterns are normal in complexity in neurons with reduced fidgetin...... 98

Figure 7. Loss of fidgetin does not affect microtubule behaviors in uninjured classIV ddaC neurons...... 100

xii

Figure 8. Microtubule dynamics persist longer in degenerating dendrites when Klp59C, but not fidgetin levels are reduced...... 102

Figure 9. Targeting Klp59C and fidgetin by RNAi slow degeneration more than targeting either alone...... 104

Figure S1. Extended candidate screen confirms that inhibition of the ubiquitin-proteasome system (UPS) and overexpresion of Wlds delay all types of degeneration, while other previously identified factors delay only pruning...... 114 Figure S2. Comparison of Klp59C reduction in combination with protective pathway knockdown to additional controls...... 115

Chapter 5

Figure 1. Summary of the relationship of microtubule disassembly and energy depletion in neuron degeneration...... 124

xiii

Acknowledgements Research acknowledgements are included in the individual chapters

First, my heartiest gratitude goes to my advisor Melissa Rolls. Thank you for taking me in your lab six years ago when I do not have a comprehensive background in and no experiences working with Drosophila. Since then, I have enjoyed almost every day working in this lab. You passion for science deeply motivated me to explore different possibilities in my research. Talking with you increases my confidence to continue my projects. Because of your optimistic altitude, I am no longer afraid of negative data and I am always able to keep up hope to go forward. You give fantastic scientific talks! They are great examples to teach us how to communicate science in a fun and exciting way.

Would also like to thank Floyd Mattie for introducing me to microscope, basic fly work and data analysis. You were always able to stay calm while perfectly solved the problems when I was paranoid about any unexpected situation. I also thank Michelle Stone, Michelle Nguyen, Kavitha Rao,

Li Chen, Meilssa Long,and Ricahrd Albertson for being patient with me. When I asked questions, you all explained the answers to me and made sure that I could understand. You all indulged me in scientific discussions and non-scientific conversations and encouraged me when I was stressed by all kinds of presentations and exams. Thank you all for making the lab a friendly and pleasant place to work at. Also, I thank my former Undergraduate student Kherlee Ng for your help on my project.

I would also like to thank my dissertation committee members: Lorraine Santy, Graham

Thomas, Zhichun Lai and Richard Ordway. Many thanks to Joyce Greslick and Linda Kunes for all their assistance with requirements and paperwork.

Also I really appreciate my English tutor/friend Marilyn Morrisson, and the Graduate Writing

Center for your enormous help with my English speaking and writing. I am also deeply thankful for my

xiv piano teacher Patricia Lloyd who always assigned the beautiful music that relaxed and soothed me. I had so much fun playing duets and Christmas carols with you.

My thanks also go to my dearest friend, Juan Wu. Thank you for all your warm encouragements when I encountered difficulties in life and in studies. Hope to join you soon again in Boston!

Loving thanks to my mother, Chunjiao Tao, an educator and a diligent learner herself, you have always been a source of encouragement and inspiration to me throughout my life. Thank you for being supportive and understanding, and for coming to the States to care for me when I had healthy issue.

1

Chapter 1: Introduction

Overview of Neuron Degeneration

Neurons, a group of highly compartmentalized cells, are comprised of axons, dendrites and soma. Each part has different functions: dendrites receive signals, the soma generates proteins and organelles, and axons send signals out. When neurons are damaged or not needed the whole cell can go through apoptotic cell death [1]. Besides whole cell death, the axons and dendrites themselves can be damaged by chemical toxins, physical trauma, neurodegenerative diseases, and so forth [2]. In these cases, instead of replacing the whole cell, neurons selectively clear the dysfunctional parts to avoid inflammation and create space for regeneration.

Though there are multiple types of neuron degeneration, my dissertation will focus on injury- induced degeneration and developmental pruning. After transection, neurons are able to initiate an active self-destruction program to clear the damaged part. In vertebrates, the process usually takes a few days, in which the damaged part first stays intact and electrically active and then degenerates rapidly along the [3]. In contrast, in invertebrates, neurons are able to initiate degeneration within a few hours after injury [4]. Based on the morphological changes, the degeneration process can be divided into two phases: the latent period and the executive period. The mechanisms responsible for the events occurring at each phase, such as microtubule degradation and protein/organelle turn over, remains unclear. Pruning is another type of neuron degeneration independent of cell body death [5].

During pruning, excessive axon and dendrite branches are removed to establish the precise neuronal connections for adult life [6].

Though degeneration has been widely studied, many questions still remain. It is debatable whether neurons use the same programs to prune their axon or dendrite trees during development. It is not clear which machineries are responsible for microtubule fragmentation during

2 degeneration. Furthermore, the relationship between energy requirements and degenerating activities is also elusive. An overlooked issue in neuron degeneration studies is that axon degeneration has been extensively studied whereas dendrite degeneration requires more attention and investigation. My thesis addressed these issues through a comparison study of injury-induced axon and dendrite degeneration and developmental pruning in one cell type on a single cell level. We also identified an energy indicator that can monitor the real-time energy collapse during degeneration in live cells. This discovery would help to distinguish the active energy consuming events from the passive events resulting from the ATP shortage in a degenerating neuron.

Injury-induced axon degeneration, also named Wallerian degeneration, is defined as a

“progressive degeneration of the distal portion of a transected axon” [7]. After transection in vivo, axons distal to the cut site appear normal for a period to time, and then go through rapid degeneration

[8, 9]. During the execution period, neurons become swollen, beaded and eventually completely engulfed by its surrounding cells [10]. Since transection of the axon (axotomy) separates the distal part from the cell body, axon degeneration was considered a passive dying away process due to the lack of nutrients and energy support from the soma. However, Wlds mutant mice showed that when axons are severed from the soma, they could stay intact for weeks. This discovery completely shattered the old theory and leaded to the hypothesis that Wallerian degeneration is an active self-destruction program

[11, 12]. Wlds, a chimerical protein, consist of two parts: the N terminus of Ube4b, an E4 ubiquitin ligase and the full sequence of nicotinamide mononucleotide adenylyltransferease 1 (Nmnat1) [13].

Wlds interference with Wallerian degeneration remains mysterious.

Our understanding of how Wlds protects transected axons has only started to emerge in recent years. As Wlds contains the complete sequence of Nmnat1, studying Nmnat provides insights into the working mechanism of Wlds [12]. Nmnat is required for NAD (nicotinamide adenine dinucleotide) biosynthesis. The oxidative form of NAD, NAD+, participates in ATP generation through accepting

3 and donating electrons [14]. Overexpression of either Nmnat1 or Nmnat3 (Golgi associated) is sufficient to protect injured axons in both mice and Drosophila [15, 16, 17]. Nmnat 2 (mitochondria associated) can protect injured neurons only when expressed at high levels [18]. Interestingly, knockdown of endogenous Nmnat 2 causes uninjured neurons to degenerate in mice superior cervical ganglia (SCG) explant culture, which cannot be compensated by endogenous Nmnat1 and 3 [19].

This data indicates that Wlds and Nmnat protect neurons at a non-nuclear site. The protection is more likely to be mediated by mitochondria through its ATP synthesis and its capability to buffer calcium. Wlds was found to be located on mitochondria in mice neuron [20]. Also mitochondria’s motility is well retained in these Wlds overexpression neurons. Thus, it is suggested that Wlds protects axons through maintaining ATP levels and increasing calcium-buffering capabilities in injured neurites

[20]. Another group observed that expression of Wlds inhibits roGFP2 (a reactive oxygen sensor) oxidation, and thus argued that mitochondria redox state but not mitochondria motility plays a more important role in Wlds exerted axon protection [21]. Also data supports the idea that Wlds depends on its intrinsic Nmnat activity to maintain energy homeostasis and thus protects injured axons [22].

Additionally, mitochondrial membrane potential is better maintained in Wlds neurons [23]. It is likely that mitochondria continue to generate ATP in Wlds neurons and this leads to the final protection. Our lab applied a novel strategy to detect energy collapse. We expressed this energy sensor with Wlds to test the spatial and temporal aspects of Wlds effect on maintaining the energy level in the severed axons and dendrites.

In contrast with Wallerian degeneration, dendrite degeneration is much less investigated.

Dendrite degeneration morphologically resembles Wallerian degeneration. Both include membrane reorganization and fragmentation along the neurites (Figure 1). The microtubule fragmentation is a hallmark of both types of neurite degeneration (Figure 1). A laser cutting assay and a pruning assay in

Drosophila ddaC peripheral sensory neurons can be used to study dendrite degeneration [24, 25].

4

Interestingly, Wlds also delays injury-induced dendrite degeneration and developmental pruning in

Drosophila ddaC neurons [4]. This indicates that Wlds has a conserved protecting effect on injury- induced neuron degeneration across species. Dendrites can also get injured by trauma and neurodegenerative diseases. Therefore, exploring the molecular machineries in dendrite degeneration is equally important for the development of treatment for degenerative diseases.

Figure1. Schematic of major steps of injury-induced axon and dendrite degeneration. During the latent period, axons and dendrites stay structurally intact. At the executive period, microtubules become disintegrated and the cell begins to swell, bead and eventually become fragmented in both axons and dendrites.

Both axon and dendrite degeneration appear to be molecularly distinct from apoptosis. Both processes usually occur before soma death [26]. Genetic manipulations and pharmaceutical drugs that strongly inhibit apoptosis do not suppress injury-induced neuron degeneration. For example, anti -

5 apoptotic factors Bax and Bak block the large-scale cell death during retinal cells development in mice, but not Wallerian degeneration of these axons [27, 28]. Our study in Drosophila peripheral sensory neurons found that down regulating apoptotic activator gene Dronc or overexpressing baculoviral caspase inhibitor p35 do not affect injury -induced axon or dendrite degeneration [4]. It was also shown that flies with mutations in genes that regulate cell death, such as grim, hid, and rpr prune their axons normally [29]. On the other hand, Wlds, which strongly delays injury-induced neuron degeneration, fails to block global NGF deprivation induced apoptosis in sympathetic neuron culture [30]. Similar to Wlds, dsarm/sarm1 (sterile alpha and TIR motif containing

1 and its fly homologue), blocks Wallerian degeneration in fly ORN and in cell cultures. However, this protein could not repress the wide spread apoptotic cell death in developing eye discs when a pro- apoptotic gene hid is over expressed in the Drosophila visual system [31-34]. Additionally, it is controversial whether neuron degeneration requires caspase activities. Though apoptotic caspases are not detected in Mushroom Body γ neuron pruning, our work found local caspases activities in dendrites of pruning da neurons [4, 35]. Additionally, the caspases involved in pruning might be different from the ones utilized in neuronal cell body death. Caspase 6, but not the classical apoptotic Caspases 3 and

9, is required in both in vitro and in vivo mice RGC pruning models. However, the axons of the same type of neurons degenerate after severing independent of Casp6 [36].

A wide range of models has been developed to investigate the molecular mechanisms in axon and dendrite degeneration. Both mice and Drosophila have proven powerful models for studying mechanisms underlying injury-induced degeneration and developmental pruning (Table1). Mouse superior cervical ganglia (SCG), sciatic nerve segments and dorsal root ganglia (DRGs) are widely used for in vitro studies [25, 37, 38]. Laser cutting is applied to induce axon degeneration in these cultured neurons. In in vivo studies, crushed optic nerves and sciatic nerves in mice are used for studying Wallerian degeneration [39, 40]. Axonal pruning studies can be performed in

6

(DG) in the hippocampus of a neonatal mouse [41]. In the adult fly, severed olfactory receptor neuron

(ORN) and Drosophila wings have been employed to study injury induced axon degeneration [42, 43].

In third instar larvae, peripheral ddaC neurons and crushed segmental axons have been used for the same purpose [4, 44]. Mushroom Body motor neurons and ddaC periphery sensory neurons are utilized for developmental pruning study [45]. Among all these different models, da neurons are the only model that can be used to study all three types of degeneration in the same cell.

Table1 Summary of experimental paradigms of neuronal injury

Animal Model Injury method Degeneration Major reference Type Mice In Superior cervical ganglia Laser cutting Wallerian [37] vitro (SCG) degeneration Sciatic nerve segments [38]

Dorsal root ganglia (DRGs) [25] In Sciatic nerve segments Crush with Wallerian [40] vivo Optic nerves needles degeneration [39] Dentate gyrus (DG) Natural Pruning [41] development Drosophila In Wing Cut with Wallerian [43] vivo scissors degeneration Motor neurons Crush with Wallerian [44] forceps degeneration DdaC Laser cutting, Wallerian [4] Natural degeneration, development Injury- induce, dendrite degeneration, Pruning Mushroom Body motor Natural Pruning [45] neuron development Olfactory receptor neurons Remove the Wallerian [42] (ORN) sensory organ degeneration

7

Comparison of injury-induced degeneration and developmental pruning

As I have already noted, during development neurons selectively eliminate exuberant neuronal processes without cell death. This process, referred to as pruning, is essential for the maturation of the and the establishment of the fine connections between neurons and their targets in both vertebrates and invertebrates [46, 47-49].

Developmental pruning morphologically resembles injury- induced degeneration of axons and dendrites. In addition to the morphological resemblances such as swelling, beading and debris clearance, the two processes have molecular similarities [27]. For example, both processes involve active self-destructive programs and are independent of apoptotic machineries. Microtubule fragmentation is an early sign in both processes [42, 50]. Calcium waves are observed after axon and dendrite severing, as well as in pruning dendrites [51-53]. These similarities suggest that the two processes may be governed by similar mechanisms. However, with distinct regulators discovered in each process, an alternative hypothesis has become more widely accepted: the mechanisms that initiate and execute the degeneration are specific to individual settings, but converge to a common downstream debris clearance mechanism in the two types of degeneration.

Ubiquitin-proteasome System (UPS) in neuron degeneration

Studies in different model animals indicate that local UPS activity is required for both pruning and injury-induced degeneration. Inhibiting UPS through overexpressing UBP2 (a deubiqutiation enzyme) or disrupting proteasome by mov34 RNAi and rpn RNAi causes severe MB γ neuron and ddaC pruning defects during metamorphosis [54, 55]. UBP2 overexpression and rpn RNAi also delay injury-induced axon and dendrite degeneration in Drosophila ddaC neurons [4, 56]. Recent studies also

8 showed that the disturbance of proteasome through pharmaceutical drug treatment could also delay

Wallerian degeneration in cultured neurons or the optic nerves in vivo [57].

These results demonstrate that UPS is involved in both types of degeneration. However, it remains controversial whether UPS generally cleans the damaged proteins or specifically targets and removes the negative regulators of degeneration. Thus it is critical to identify the relevant E2/E3 enzymes and their targets in different types of degeneration. So far, a few E1, E2, E3 enzymes were identified in different types of pruning in the Drosophila model. E1 uba1 ubiquitin-activating enzyme was found to inhibit Drosophila MB γ neuron pruning [55]. UbcD1, an identified E2 ubiquitin conjugating enzyme, acts in ddaC dendrite pruning [58]. DIAP3, an E3 enzyme, was found to negatively regulate Dronc caspase activities during pruning. Overexpression of DIAP leads to decreased caspase dronc activation and subsequent pruning defects [58]. Another cuilin-1 based SCF

E3 ubiqutin ligase was identified to facilitate ddaC dendrite and MB γ neuron pruning through the inactivation of InR/PI3K/TOR pathways [56]. On the other hand, so far, no E2 has been discovered in injury-induced degeneration. Hiw, an E3 ligase, is identified as an important regulator of injury- induced axonal and synaptic degeneration in the crushed nerve assays in Drosophila [59]. Hiw adds ubiquitin to Nmnat. A decrease in nmnat level in neurons after injury leads to faster degeneration [59].

Its mammalian homolog Phr1 (Pam, highwire, Rpm-1) functions similarly in axon degeneration after axotomy in mice [60]. In addition to Hiw, SkpA, a core component of SCF E3 ubiquitin ligases, was also revealed to down regulate Wnd-JNK pathway and Nmnat levels after nerve crushing to promote neuron degeneration [61].

Wlds distinguishes injury-induced degeneration and developmental pruning in certain contexts

Even the universal protective factor Wlds has different effects on injury-induced degeneration and developmental pruning. In mice, Wlds protects transected optic axons of developing retinal

9 ganglion cells (RGC) but has no effect on the naturally occurring developmental degeneration of these same axons [42]. In Drosophila, mouse Wlds protein suppresses degeneration of served ORNs and ddaC axons [21]. However, Wlds has no effect on Drosophila MB γ axon pruning. But interestingly, we found that Wlds expression does inhibit dendrite pruning of the peripheral ddaC neurons [4, 62]. This data suggests that different mechanisms exist not only between injury-induced degeneration and pruning and but also between axon and dendrite developmental pruning in different cell types of different systems. Other mediators besides UPS and Wlds need to be identified to further separate dendrite pruning and injury-induced neuron degeneration.

Other players separate injury-induced degeneration and developmental pruning

The recent identification of the Drosophila Toll receptor adapter dSarm adds more compelling evidence that injury-induced degeneration is different from pruning. Knocking out dSarm delays the degeneration of transected ORN neurons in fly. However, in dsarm mutant animals, axons and dendrites of the MB γ neuron prune normally [34].

In other studies of neuron pruning and injury-induced degeneration, different factors were identified to delay these two types of degeneration. However, due to that no comparison study was performed or no parallel pruning model was available, it is not clear whether the factors required for one type of degeneration are also required for the other. For example, the mutation of DLK /JNK signaling pathway was reported to suppress Wallerian degeneration, however no published data showed whether they are required for pruning in vivo. Our unpublished data showed that bsk RNAi or DN delays neither injury-induced axon nor dendrite degeneration. Additionally, bsk overexpression disrupts normal pruning in ddaC neurons. The mitochondrial permeability transition pore plays a role in

Wallerian degeneration. Depleting the mPTP protien cynD by shRNAi slowed axonal degeneration after transection in cultures of adult rat DRG neurons [8]. However, no parallel pruning assay is

10 available with DRG cultured neurons to test whether cynD RNAi also delays normal pruning. In

Drosophila, imbalanced voltage-gated sodium channel and potassium channel activities lead to either delay or accelerate degeneration of the distal stump of the after nerve crush [63].

However, similar comparison studies on whether sodium and potassium currents influence pruning have not been conducted and would be easy to test in the same model animal. These comparative studies would enormously advance our understanding of the similarities and differences of various types of degeneration in diverse experimental paradigms.

Microtubule degradation in degenerating neurons

Microtubules are crucial for transporting sorted proteins and organelles along axons and dendrites to their functional sites, which sometimes are distant from the soma [64]. Disassembly of microtubules is the earliest cellular event in both neuron pruning and injured axons or dendrites [55,

57]. Destabilization of microtubules by drugs, such as vinscritn or colchicine induces axonal degeneration[65], whereas stabilization of microtubules by paclitaxel can prevent neurons from injury- induced degeneration [66].

To date, the primary mechanisms and molecules disassembling microtubule are largely unknown (Figure2). It was recently reported that the UPS pathway regulates microtubule fragmentation at the very early stage of axon degeneration [57, 67]. Caspases 6 destabilizes microtubules through cleaving the microtubule-associated proteins, such as tau, in degenerating axons. In addition, has long been believed to function in neuron degeneration [51, 68]. Recently, Calpain has been found to be responsible for cytoskeleton breakdown in transected sciatic and optic nerves in vivo [69].

11

Figure2. Diagram of microtubule regulators required in axon and dendrite degeneration. Katanin p60-like and KIF-2A have found to be required for dendrite and axon pruning respectively. However, no parallel pathways are identified in injury-induced axon or dendrite degeneration. The morphological changes during the three types of degeneration resemble and all result in final neuritis clearance.

Besides proteases, microtubule depolymerizing and severing proteins play a role in neuron degeneration. Loss of a kinesin-13 family member and depolymerase (KIF2A) suppresses axotomy and

NGF deprivation induced axon degeneration in DRG neuron cultures. In those protected axons, microtubules also remain structurally continuous [66]. In vivo studies of Drosophila ddaC neurons also revealed that katanin p60-like1 severed the dendrites from the soma at the initial step of pruning [70].

To pursue whether any microtubule disassembly mechanism is also required in fly ddaC neurons after injury, our lab performed a genetic screen and identified Klp59C, a member of the microtubule depolymerizing proteins Kinesin 13 family, delayed all types of degeneration, whereas fidgetin, a microtubule severing protein, is specifically required for injury-induced dendrite degeneration.

Kinesin-13 proteins, instead of transporting cargos along the microtubule lattices like other kinesins, accumulate at microtubule ends and induce depolymerization of their ends in vitro [71]. In vivo studies showed that kinesin-13 proteins promote chromosome segregation through depolymerizing microtubules at kinetochore in mitotic or meiotic cytoplasm [72]. Fidgetin, a member of the AAA protein super family, has important roles in mammalian development. For example, mammalian fidgetin regulates the spindle architecture during mitosis [73]. However, very few of their neuronal functions have been identified so far.

12

Energy maintenance and neurite degeneration

Maintaining energy homeostasis is essential for neuron health. Managing membrane electrical potential, transporting cargos through motor proteins, recycling synaptic vesicles, and a myriad of phosphorylation reactions all require considerable ATP in neurons (Table2) [74]. Healthy neurons produce excess ATP to guarantee sufficient ATP supply for neuron to carry out normal functions.

Mitochondria are the major ATP suppliers in cells.

Table 2. Cellular processes that generate ATP and consume ATP.

ATP sources ATP consumers Mitochondrial Oxidation/Phosphrylation Na+/K+-ATPase Glycolysis Ca2+-ATPase F1/F0-mito ATPase Vesicle recycling Motor proteins Cytoskeletal dynamics Phosphorylation

Global changes in ATP levels in neurons strongly affect neuron health and their normal functions. All the processes that required ATP listed in Table 1 would be halted. A dramatic reduction in

ATP availability can initiate irreversible neuron degeneration. Mitofusin2 causes mitochondria to localize to the wrong positions and this causes axons to degenerate [75]. The energy deprivation of culture mouse cortical axons causes them to depolarize immediately and subsequently degenerate [76].

Even a brief period of energy restriction at a short length of nerve could trigger degeneration of the whole nerve fiber distal to the site of exposure [77]. How energy depletion leads to degeneration is not completely known. One possibility is that Na+/K+ pumps ceased to work properly upon energy deprivation [78]. The imbalance of ion concentration leads to hypodepolarization of neurons and

13 eventually irreversible neuron damage [78]. This suggests that local energy status might be an important factor regulating axonal stability and degeneration.

Energy decrease is often associated with neuron degeneration. In this case, instead of being the major trigger of degeneration, reduced energy production is secondary to the process that initiates neuronal degeneration. After injury, mitochondrial dysfunction, substrate deprivation and excessive energy usage by active machineries required for degeneration can all lead to the final ATP depletion

[76, 79]. ATP decreases along with NAD levels in the transected axons of cultured neurons [80, 81].

Attenuation of this ATP loss by Wlds or Nmnat overexpression protects axon from injury-induced degeneration [76]. Additionally, adding NAD or a NAD/pyruvate mixture to the sciatic nerve explant cultures retains ATP level and thus can prevent axon degeneration after axotomy [80, 82].

It is not clear what causes ATP to decrease in a degenerating axon and whether this decrease further promotes axonal degeneration. Mitochondria are the major source of cellular ATP.

Mitochondrial dysfunction might be responsible for the ATP decrease. Indeed, mitochondria become swollen and less mobile in degenerating axons [38]. Also, mitochondrial ATP production tightly links with the changes of cytoplasmic calcium levels. It's likely that the calcium spike after axotomy disturbs the membrane potentials of mitochondrial and interferes with its normal ATP production [38]. Another possibility is that the mitochondria can no longer be transported to the site of need due to the impaired axonal transport which is caused by microtubule depolymerizaton at the early stage [83]. The failure of the transport of mitochondria could compromise axon integrity and function and eventually causes degeneration. On the other hand, rescuing the impaired mitochondrial transport by Mfn1 (Mitofusins, essential for mitochondrial fusions in mammals) over expression protects axons from degeneration caused by Mfn2 mutant in DRG neurons [75].

The loss of intracellular ATP/GTP is an important demarcation point in neuron degeneration.

Similar to ATP, GTP is also an energy source for some enzymes, such as tubulin dimmers, RabGTPase

14 and etc. Additionally, GTP is readily interchanged with ATP. Pinning down the timing would provide insights into which events happen prior to energy collapse and are actively controlled, and which events simply result from unavailability of energy in the cytoplasm. So far, we know that in transected sciatic nerves, ATP depletion and microtubule fragmentation both occur during the latency period before degeneration becomes morphologically recognizable [82]. It would be beneficial to identify the order of energy failure and microtubule disassembly in a degenerating neuron [84]. Our study on the mechanistic relationship between energy failure and the alteration in microtubule dynamics may shed light on the molecular mechanisms of axonal degeneration.

A few methods have been applied to measure real-time ATP level. The amount of ATP can be quantified by a biochemical luciferease/luciferin assay kit. However, this biochemical measurement lyses the cells and is not sensitive enough to detect ATP level in a single neuron. Another method uses dyes, such as TMRM, to indicate the mitochondrial membrane potential of the cell [85]. However, these dyes have some limitations: they can only be used on living tissues and do not give a direct readout of energy availability in cells. Furthermore, TMRM dye could not reflect the energy status in certain cells such as cancer cells, in which glycolysis, not the mitochondria dependent oxidative phosphorylation, is the major energy source [86]. To better monitor ATP status during stress, death or degeneration, more reliable markers are needed.

Different labs have established novel technical approaches to study energy status in cells.

Currently, it is not possible to image ATP levels in different subcellular compartments. Fluorescent probes for ATP would greatly advance our understanding of the regulation of ATP concentrations and its functions in cellular physiology and disease [74]. Perceval is a fluorescent biosensor of ATP:ADP found by the lab of Gary Yellon. This sensor excites at different wavelengths depending on the ratio of

ATP binding versus ADP binding. However this sensor is saturated at a low ATP and ADP ratio, therefore the probe is limited from being applied to mammalian samples. An optimized version

15 percevalHR (high range) was later developed and responds more dramatically with a small ATP: ADP ratio change. Thus, this sensor enables the measurement of the ATP: ADP ratios in the healthy mammalian cells. However, the drawback of this sensor is that it needs to be normalized to magnesium levels and is complicated to use. We need a simple and straightforward readout of the energy availability in cells. Rab5-YFP can be used for this purpose. Membrane associated Rab5 appears in punctae and requires GTP to maintain this localization. Therefore in our assay, energy collapse is marked by the Rab protein changing from a punctate pattern to a diffuse one. So far we have tested only tagged Rabs in living cells, but we expect the assay to work equally well using anti-Rab antibodies for immunofluorescence in fixed cells. As the sensor can be modified for fixed samples, it would also have some potential for use in medical settings in which tissue must be fixed for analysis.

Summary

Neurons, due to their long and thin morphological features, are susceptible to various kinds of insults.

After they are impaired either at axons or dendrites, different programs within these structures are activated to clear the damaged parts. This process not clear up space for future regeneration and also avoid further inflammations caused by surrounding tissues. So far our understanding of the molecular pathways involved in this removal process is very limited. We have already known that it’s a active process based on the discovery of a mutant fusion protein Wlds. We also have known that it’s molecularly different from naturally happened developmental pruning and recently a few endogenous proteins have emerged to be involved in actively degrading the damaged parts. It still requires more collaborative efforts from labs all round world working on various model animals to map out the major players involved in this mysterious process. Our lab uses Drosophila as a model animal and characterized three types of degeneration. Using these systems, we investigated the potential molecular

16 players and the relationship between the energy depletion and a specific cellular event, microtubule plus end termination, within the degenerating axons.

17

References

1. Raff, M.C., A.V. Whitmore, and J.T. Finn, Axonal self-destruction and neurodegeneration. Science, 2002. 296(5569): p. 868-71. 2. Han, C., et al., Epidermal cells are the primary phagocytes in the fragmentation and clearance of degenerating dendrites in Drosophila. Neuron, 2014. 81(3): p. 544-60. 3. Wang, J.T., Z.A. Medress, and B.A. Barres, Axon degeneration: molecular mechanisms of a self-destruction pathway. J Cell Biol, 2012. 196(1): p. 7-18. 4. Tao, J. and M.M. Rolls, Dendrites have a rapid program of injury-induced degeneration that is molecularly distinct from developmental pruning. J Neurosci, 2011. 31(14): p. 5398-405. 5. Cusack, C.L., et al., Distinct pathways mediate axon degeneration during apoptosis and axon- specific pruning. Nat Commun, 2013. 4: p. 1876. 6. Maor-Nof, M. and A. Yaron, Neurite pruning and neuronal cell death: spatial regulation of shared destruction programs. Curr Opin Neurobiol, 2013. 23(6): p. 990-6. 7. Waller, A., Experiments on the Section of the Glossopharyngeal and Hypoglossal Nerves of the Frog, and Observations of the Alterations Produced Thereby in the Structure of Their Primitive Fibres. Philosophical Transactions of the Royal Society of London 1850. 140: p. 7. 8. Tsao, J.W., et al., Loss of the compound : an electrophysiological, biochemical and morphological study of early events in axonal degeneration in the C57BL/Ola mouse. Eur J Neurosci, 1994. 6(4): p. 516-24. 9. Lubinska, L., Patterns of Wallerian degeneration of myelinated fibres in short and long peripheral stumps and in isolated segments of rat phrenic nerve. Interpretation of the role of axoplasmic flow of the trophic factor. Brain Res, 1982. 233(2): p. 227-40. 10. Tsao, J.W., E.B. George, and J.W. Griffin, Temperature modulation reveals three distinct stages of Wallerian degeneration. J Neurosci, 1999. 19(12): p. 4718-26. 11. Perry, V.H., M.C. Brown, and J.W. Tsao, The Effectiveness of the Gene Which Slows the Rate of Wallerian Degeneration in C57BL/Ola Mice Declines With Age. Eur J Neurosci, 1992. 4(10): p. 1000-2. 12. Coleman, M.P., et al., An 85-kb tandem triplication in the slow Wallerian degeneration (Wlds) mouse. Proc Natl Acad Sci U S A, 1998. 95(17): p. 9985-90. 13. Mack, T.G., et al., Wallerian degeneration of injured axons and is delayed by a Ube4b/Nmnat chimeric gene. Nat Neurosci, 2001. 4(12): p. 1199-206. 14. Wang, J. and Z. He, NAD and axon degeneration: from the Wlds gene to neurochemistry. Cell Adh Migr, 2009. 3(1): p. 77-87. 15. Avery, M.A., et al., Wld S requires Nmnat1 enzymatic activity and N16-VCP interactions to suppress Wallerian degeneration. J Cell Biol, 2009. 184(4): p. 501-13. 16. Kitaoka, Y., et al., Axonal protection by Nmnat3 overexpression with involvement of autophagy in optic nerve degeneration. Cell Death Dis, 2013. 4: p. e860. 17. Sasaki, Y., et al., Transgenic mice expressing the Nmnat1 protein manifest robust delay in axonal degeneration in vivo. J Neurosci, 2009. 29(20): p. 6526-34. 18. Yan, T., et al., Nmnat2 delays axon degeneration in superior cervical ganglia dependent on its NAD synthesis activity. Neurochem Int, 2010. 56(1): p. 101-6. 19. Gilley, J. and M.P. Coleman, Endogenous Nmnat2 is an essential survival factor for maintenance of healthy axons. PLoS Biol, 2010. 8(1): p. e1000300. 20. Adalbert, R., et al., Intra-axonal calcium changes after axotomy in wild-type and slow Wallerian degeneration axons. Neuroscience, 2012. 225: p. 44-54.

18

21. O'Donnell, K.C., M.E. Vargas, and A. Sagasti, WldS and PGC-1alpha regulate mitochondrial transport and oxidation state after axonal injury. J Neurosci, 2013. 33(37): p. 14778-90. 22. Yahata, N., S. Yuasa, and T. Araki, Nicotinamide mononucleotide adenylyltransferase expression in mitochondrial matrix delays Wallerian degeneration. J Neurosci, 2009. 29(19): p. 6276-84. 23. Avery, M.A., et al., WldS prevents axon degeneration through increased mitochondrial flux and enhanced mitochondrial Ca2+ buffering. Curr Biol, 2012. 22(7): p. 596-600. 24. Williams, D.W. and J.W. Truman, Cellular mechanisms of dendrite pruning in Drosophila: insights from in vivo time-lapse of remodeling dendritic arborizing sensory neurons. Development, 2005. 132(16): p. 3631-42. 25. Miller, B.R., et al., A dual leucine kinase-dependent axon self-destruction program promotes Wallerian degeneration. Nat Neurosci, 2009. 12(4): p. 387-9. 26. Fischer, L.R. and J.D. Glass, Axonal degeneration in motor neuron disease. Neurodegener Dis, 2007. 4(6): p. 431-42. 27. Finn, J.T., et al., Evidence that Wallerian degeneration and localized axon degeneration induced by local deprivation do not involve caspases. J Neurosci, 2000. 20(4): p. 1333-41. 28. Whitmore, A.V., et al., The proapoptotic proteins Bax and Bak are not involved in Wallerian degeneration. Cell Death Differ, 2003. 10(2): p. 260-1. 29. Korhonen, L. and D. Lindholm, The ubiquitin proteasome system in synaptic and axonal degeneration: a new twist to an old cycle. J Cell Biol, 2004. 165(1): p. 27-30. 30. Deckwerth, T.L. and E.M. Johnson, Jr., Neurites can remain viable after destruction of the neuronal soma by programmed cell death (apoptosis). Dev Biol, 1994. 165(1): p. 63-72. 31. Adalbert, R., et al., The slow Wallerian degeneration gene in vivo protects motor axons but not their cell bodies after avulsion and neonatal axotomy. Eur J Neurosci, 2006. 24(8): p. 2163-8. 32. Beirowski, B., et al., The WldS gene delays axonal but not somatic degeneration in a rat glaucoma model. Eur J Neurosci, 2008. 28(6): p. 1166-79. 33. Conforti, L., et al., NAD(+) and axon degeneration revisited: Nmnat1 cannot substitute for Wld(S) to delay Wallerian degeneration. Cell Death Differ, 2007. 14(1): p. 116-27. 34. Osterloh, J.M., et al., dSarm/Sarm1 is required for activation of an injury-induced axon death pathway. Science, 2012. 337(6093): p. 481-4. 35. Williams, D.W., et al., Local caspase activity directs engulfment of dendrites during pruning. Nat Neurosci, 2006. 9(10): p. 1234-6. 36. Nikolaev, A., et al., APP binds DR6 to trigger axon pruning and neuron death via distinct caspases. Nature, 2009. 457(7232): p. 981-9. 37. Villegas, R., et al., Calcium release from intra-axonal endoplasmic reticulum leads to axon degeneration through mitochondrial dysfunction. J Neurosci, 2014. 34(21): p. 7179-89. 38. Barrientos, S.A., et al., Axonal degeneration is mediated by the mitochondrial permeability transition pore. J Neurosci, 2011. 31(3): p. 966-78. 39. Watkins, T.A., et al., DLK initiates a transcriptional program that couples apoptotic and regenerative responses to axonal injury. Proc Natl Acad Sci U S A, 2013. 110(10): p. 4039-44. 40. Vanderhaeghen, P. and H.J. Cheng, Guidance molecules in axon pruning and cell death. Cold Spring Harb Perspect Biol, 2010. 2(6): p. a001859. 41. Bagri, A., et al., Stereotyped pruning of long hippocampal axon branches triggered by retraction inducers of the semaphorin family. Cell, 2003. 113(3): p. 285-99. 42. Hoopfer, E.D., et al., Wlds protection distinguishes axon degeneration following injury from naturally occurring developmental pruning. Neuron, 2006. 50(6): p. 883-95.

19

43. Fang, Y., et al., A novel Drosophila model of nerve injury reveals an essential role of Nmnat in maintaining axonal integrity. Curr Biol, 2012. 22(7): p. 590-5. 44. Xiong, X. and C.A. Collins, A conditioning lesion protects axons from degeneration via the Wallenda/DLK MAP kinase signaling cascade. J Neurosci, 2012. 32(2): p. 610-5. 45. Lee, T., et al., Cell-autonomous requirement of the USP/EcR-B ecdysone receptor for mushroom body neuronal remodeling in Drosophila. Neuron, 2000. 28(3): p. 807-18. 46. Luo, L. and D.D. O'Leary, Axon retraction and degeneration in development and disease. Annu Rev Neurosci, 2005. 28: p. 127-56. 47. Weeks, J.C. and J.W. Truman, Independent steroid control of the fates of motoneurons and their muscles during insect metamorphosis. J Neurosci, 1985. 5(8): p. 2290-300. 48. Truman, J.W., Metamorphosis of the of Drosophila. J Neurobiol, 1990. 21(7): p. 1072-84. 49. O'Leary, D.D. and S.E. Koester, Development of projection neuron types, axon pathways, and patterned connections of the mammalian cortex. Neuron, 1993. 10(6): p. 991-1006. 50. Glass, J.D., B.L. Schryer, and J.W. Griffin, Calcium-mediated degeneration of the axonal cytoskeleton in the Ola mouse. J Neurochem, 1994. 62(6): p. 2472-5. 51. George, E.B., J.D. Glass, and J.W. Griffin, Axotomy-induced axonal degeneration is mediated by calcium influx through ion-specific channels. J Neurosci, 1995. 15(10): p. 6445-52. 52. Yang, J., et al., Regulation of axon degeneration after injury and in development by the endogenous calpain inhibitor calpastatin. Neuron, 2013. 80(5): p. 1175-89. 53. Kanamori, T., et al., Compartmentalized calcium transients trigger dendrite pruning in Drosophila sensory neurons. Science, 2013. 340(6139): p. 1475-8. 54. Kuo, C.T., L.Y. Jan, and Y.N. Jan, Dendrite-specific remodeling of Drosophila sensory neurons requires matrix metalloproteases, ubiquitin-proteasome, and ecdysone signaling. Proc Natl Acad Sci U S A, 2005. 102(42): p. 15230-5. 55. Watts, R.J., E.D. Hoopfer, and L. Luo, Axon pruning during Drosophila metamorphosis: evidence for local degeneration and requirement of the ubiquitin-proteasome system. Neuron, 2003. 38(6): p. 871-85. 56. Wong, J.J., et al., A Cullin1-based SCF E3 ubiquitin ligase targets the InR/PI3K/TOR pathway to regulate neuronal pruning. PLoS Biol, 2013. 11(9): p. e1001657. 57. Zhai, Q., et al., Involvement of the ubiquitin-proteasome system in the early stages of wallerian degeneration. Neuron, 2003. 39(2): p. 217-25. 58. Kuo, C.T., et al., Identification of E2/E3 ubiquitinating enzymes and caspase activity regulating Drosophila dendrite pruning. Neuron, 2006. 51(3): p. 283-90. 59. Xiong, X., et al., The Highwire ubiquitin ligase promotes axonal degeneration by tuning levels of Nmnat protein. PLoS Biol, 2012. 10(12): p. e1001440. 60. Babetto, E., et al., The Phr1 ubiquitin ligase promotes injury-induced axon self-destruction. Cell Rep, 2013. 3(5): p. 1422-9. 61. Brace, E.J., et al., SkpA Restrains Synaptic Terminal Growth during Development and Promotes Axonal Degeneration following Injury. J Neurosci, 2014. 34(25): p. 8398-410. 62. Schoenmann, Z., et al., Axonal degeneration is regulated by the apoptotic machinery or a NAD+-sensitive pathway in insects and mammals. J Neurosci, 2010. 30(18): p. 6375-86. 63. Mishra, B., et al., Sodium and potassium currents influence Wallerian degeneration of injured Drosophila axons. J Neurosci, 2013. 33(48): p. 18728-39. 64. Witte, H. and F. Bradke, The role of the cytoskeleton during neuronal polarization. Curr Opin Neurobiol, 2008. 18(5): p. 479-87.

20

65. Saxena, S. and P. Caroni, Mechanisms of axon degeneration: from development to disease. Prog Neurobiol, 2007. 83(3): p. 174-91. 66. Maor-Nof, M., et al., Axonal pruning is actively regulated by the microtubule-destabilizing protein kinesin superfamily protein 2A. Cell Rep, 2013. 3(4): p. 971-7. 67. Wakatsuki, S., F. Saitoh, and T. Araki, ZNRF1 promotes Wallerian degeneration by degrading AKT to induce GSK3B-dependent CRMP2 phosphorylation. Nat Cell Biol, 2011. 13(12): p. 1415-23. 68. Berridge, M.J., P.H. Cobbold, and K.S. Cuthbertson, Spatial and temporal aspects of cell signalling. Philos Trans R Soc Lond B Biol Sci, 1988. 320(1199): p. 325-43. 69. Ma, M., et al., mediate axonal cytoskeleton disintegration during Wallerian degeneration. Neurobiol Dis, 2013. 56: p. 34-46. 70. Lee, H.H., L.Y. Jan, and Y.N. Jan, Drosophila IKK-related kinase Ik2 and Katanin p60-like 1 regulate dendrite pruning of sensory neuron during metamorphosis. Proc Natl Acad Sci U S A, 2009. 106(15): p. 6363-8. 71. Desai, A., et al., Kin I kinesins are microtubule-destabilizing enzymes. Cell, 1999. 96(1): p. 69- 78. 72. Walczak, C.E., T.J. Mitchison, and A. Desai, XKCM1: a Xenopus kinesin-related protein that regulates microtubule dynamics during mitotic spindle assembly. Cell, 1996. 84(1): p. 37-47. 73. Mukherjee, S., et al., Human Fidgetin is a microtubule severing the enzyme and minus-end depolymerase that regulates mitosis. Cell Cycle, 2012. 11(12): p. 2359-66. 74. Mattson, M.P. and D. Liu, Energetics and oxidative stress in and neurodegenerative disorders. Neuromolecular Med, 2002. 2(2): p. 215-31. 75. Misko, A.L., et al., Mitofusin2 mutations disrupt axonal mitochondrial positioning and promote axon degeneration. J Neurosci, 2012. 32(12): p. 4145-55. 76. Shen, H., K.L. Hyrc, and M.P. Goldberg, Maintaining energy homeostasis is an essential component of Wld(S)-mediated axon protection. Neurobiol Dis, 2013. 59: p. 69-79. 77. Alvarez, S., M. Moldovan, and C. Krarup, Acute energy restriction triggers Wallerian degeneration in mouse. Exp Neurol, 2008. 212(1): p. 166-78. 78. Stirling, D.P. and P.K. Stys, Mechanisms of axonal injury: internodal nanocomplexes and calcium deregulation. Trends Mol Med, 2010. 16(4): p. 160-70. 79. Shiba-Fukushima, K., et al., PINK1-mediated phosphorylation of Parkin boosts Parkin activity in Drosophila. PLoS Genet, 2014. 10(6): p. e1004391. 80. Araki, T., Y. Sasaki, and J. Milbrandt, Increased nuclear NAD biosynthesis and SIRT1 activation prevent axonal degeneration. Science, 2004. 305(5686): p. 1010-3. 81. Wang, J., et al., A local mechanism mediates NAD-dependent protection of axon degeneration. J Cell Biol, 2005. 170(3): p. 349-55. 82. Park, J.Y., et al., Mitochondrial swelling and microtubule depolymerization are associated with energy depletion in axon degeneration. Neuroscience, 2013. 238: p. 258-69. 83. Millecamps, S. and J.P. Julien, Axonal transport deficits and neurodegenerative diseases. Nat Rev Neurosci, 2013. 14(3): p. 161-76. 84. Prokop, A., The intricate relationship between microtubules and their associated motor proteins during axon growth and maintenance. Neural Dev, 2013. 8: p. 17. 85. Scaduto, R.C., Jr. and L.W. Grotyohann, Measurement of mitochondrial membrane potential using fluorescent rhodamine derivatives. Biophys J, 1999. 76(1 Pt 1): p. 469-77. 86. Ros, S. and A. Schulze, Balancing glycolytic flux: the role of 6-phosphofructo-2-kinase/fructose 2,6-bisphosphatases in cancer metabolism. Cancer Metab, 2013. 1(1): p. 8.

21

Chapter Two:

Dendrites have a rapid program of injury-induced degeneration that is molecularly distinct from developmental pruning

This chapter was published in Journal of Neuroscience 2011, 31(14): 5398-5405

I was the first and the only author apart from my advisor

Abstract

Neurons have two types of processes: axons and dendrites. Axons have an active disassembly program activated by severing. It has not been tested whether dendrites have an analogous program. We sever

Drosophila dendrites in vivo and find that they are cleared within 24 hours. Morphologically this clearance resembles developmental dendrite pruning, and, to some extent, axon degeneration. Like axon degeneration, both injury-induced dendrite degeneration and pruning can be delayed by expression of Wld(s) or UBP2. We therefore hypothesized that they use common machinery.

Surprisingly, comparison of dendrite pruning and degeneration in the same cell demonstrated that none of the specific machinery used to prune dendrites is required for injury-induced dendrite degeneration.

In addition, we show that the active program of dendrite degeneration does not require mitochondria.

Thus dendrites do have an active program of degeneration, as do axons, but this program does not require the machinery used during developmental pruning.

Introduction

It is now well established that axons have an active program of degeneration that disassembles them after they are severed from the cell body [1-4]. This program, which is known as Wallerian

22 degeneration, facilitates clearance of badly damaged axons after injury, perhaps in preparation for regeneration [5]. The fact that distal regions of axons can be stabilized after severing by overexpression of proteins including Wld(s) led to the idea that endogenous machinery actively disassembles the axon, somewhat analogous to the way that caspases disassemble apoptotic cells [1-4].

It is known that dendrites can also be damaged, but it has not been clearly established whether they have an active program of degeneration in the same way that axons do. Dendrite beading has been described during excitotoxicity and after ischemia [6-10]. However, this beading may not be related to that seen after axon severing as it is reversible in at least some cases [6-10].

Both axons and dendrites have active programs of developmental disassembly known as pruning. Many axons generate extra projections that are removed during development [1, 3]. Large- scale remodeling of axons and dendrites also takes place during metamorphosis in insects including

Drosophila melanogaster [3, 11]. Morphologically, axon and dendrite pruning appear similar to injury- induced axon degeneration. Moreover, axon degeneration is characterized by involvement of the ubiquitin-proteasome system (UPS) [12]. Axon and dendrite pruning require the UPS and can be inhibited by expression of the ubiquitin protease, UBP2 [13, 14]. Thus most researchers treat pruning of axons and dendrites and injury-induced axon degeneration interchangeably. Indeed pruning is often used as a model for studying injury-induced axon degeneration [1-4, 12, 14, 15]. Only one study has suggested that injury-induced degeneration and pruning use different pathways. In this study, the

Wld(s) protein was shown to block injury-induced degeneration, but not developmental pruning in mammals or Drosophila [16]. Even in this work, the authors surmise that the two “may converge on a common execution pathway” [16]. Moreover, the generality of the conclusion that Wld(s) distinguishes developmental pruning and injury-induced degeneration has recently been challenged by the finding that Wld(s) does in fact block developmental pruning of dendrites in Drosophila [15]. Thus the

23 simplest hypothesis, and prevailing view in the field, is that there exists a common disassembly pathway that can remove axons after injury and axons and dendrites during developmental pruning.

In this study, we test whether dendrites also have an active injury-induced degeneration program that could feed into this common disassembly pathway. We show that dendrites do have an active program of degeneration, but that it uses different machinery than dendrite pruning.

Results

Dendrites undergo beading and clearance within 24 hours after severing.

We used Drosophila larval multidendritic neurons as a model system. These neurons lie under the cuticle and are responsible for mechanosensation and nociception [17-19]. They are multipolar neurons with several dendrites and a single axon that emerge from the cell body, and they have been used very successfully to study dendrite pruning, development and branching [14, 15, 20-27]

To test whether dendrites initiate an active program of degeneration after severing from the cell body, we used a pulsed UV laser to sever dendrites of two different types of multidendritic cells in

Drosophila larvae. We chose to use the class I cell ddaE because we have previously studied axon regeneration in this cell [28], and we used the class IV ddaC cell because it has been used as a model system in which to study dendrite pruning [14, 15, 25-27]. The class refers to the complexity of dendrite branching pattern: class I multidendritic neurons have the simplest dendritic arbors and class

IV have the most complex [22]. A pulsed UV laser can very effectively sever axons or dendrites of these cells [24, 28]. We labeled class I and class IV neurons by expressing a GFP-tagged membrane marker, mCD8-GFP, or the microtubule and soluble marker EB1-GFP, selectively in these cells with class I and class IV-specific Gal4 drivers.

In both cell types all dendrite fragments distal to the cut site were cleared within 24 hours

(Figure 1). To determine the time course of clearance we analyzed morphology of ddaC (Figure 1) and

24 ddaE (Figure S1) at multiple time points after dendrite severing. The degeneration process of both cells appeared similar.

Until 3-4 hours after severing the dendrite remained continuous and microtubules labeled with

EB1-GFP continued to grow in the severed part of the dendrite (Movie 1). Between 3 and 6 hours dendrite morphology began to change, and at 6h after injury, dendrite branches often had swollen regions and a beaded appearance. At 9 or 12 hours the region previously occupied by the degenerating dendrite was completely clear (Figures 1 and S1). The same general events were observed with both

EB1-GFP and mCD8-GFP.

In addition to the swelling and beading, we also observed extensive membrane protrusion before the dendrite fragmented. Around 4-5 hours after severing, blebs were seen to emerge from the dendrite shaft with both the mCD8-GFP marker (Figure 1 D and Movie 2) and EB1-GFP (Movies 3 and 4). At later timepoints, bright-internalized membranes were observed (Figure 1D and Movie 5). As the membrane protrusion seen at 4-5 hours has not been described during axon degeneration (for example [29], but looks similar to events during dendrite pruning [27], we wished to directly compare axon degeneration, dendrite degeneration and dendrite pruning in the same cell. Membrane protrusion, swelling and beading were all observed during dendrite pruning of ddaC (Figure 1 C and D). In ddaC axons we observed swelling, beading (Figure 1B) and internalized membranes similar to those seen at late stages of dendrite degeneration (Movie 6), however, we did not observe membrane protrusion. The time course of degeneration in the ddaC axon was slightly slower than in the dendrites. Axons could remain intact until 6-7 hours (Figure 1B), but were almost always fragmented by 12 hours after severing. Clearance of the severed pieces was more variable than that of dendrites. Most often all fragments were cleared by 24 hours after severing, but in some cases they could persist until 48-72 hours (not shown).

25

We conclude that some morphological features are shared between axon and dendrite degeneration, and that dendrite degeneration and dendrite pruning are morphologically indistinguishable at this level of resolution. Both axon degeneration and dendrite pruning are known to be active processes; we therefore hypothesized that dendrite degeneration after injury is an active process similar to Wallerian degeneration. As dendrite degeneration appears extremely similar to dendrite pruning, we also hypothesized that they may use the same machinery.

26

Figure 1. Dendrites degenerate rapidly after severing. A. The multidendritic neuron ddaC was labeled with mCD8-GFP. The binary Gal4-UAS system was used to express the fluorescent protein in a subset of neurons. Whole larvae were mounted on a microscope slide and a dendrite was severed with a pulsed UV laser. The cell was imaged after severing, and at additional later timepoints. Between imaging sessions the larva was recovered to food. The orange arrow indicates the site of severing and the bracket indicates the region of the dendrite separated from the cell body by the cut. B. The ddaC cell was labeled with mCD8-GFP, but the axon was severed rather than the dendrite. As in A, the larva was remounted at different times after severing to track degeneration. Arrow and brackets are as in A.

27

C. White prepupae expressing mCD8-GFP in ddaC under control of ppk-Gal4 were collected and aged until the indicated times after puparium formation (APF). At 14h dendrites were disconnected from the cell body and fragmented. In the same animal, dendrites were completely cleared at 18h APF. D. Closeup images of ddaE or ddaC neurons expressing mCD8-GFP at different times after dendrite severing or after pupariation. Movies 1-5 illustrate different aspects of dendrite degeneration. Movie 6 shows a late stage in axon degeneration.

Dendrite degeneration is blocked by overexpression of Wld(s) or UBP2.

As a first test of whether injury-induced dendrite degeneration might use an active mechanism similar to Wallerian degeneration, we tested whether it could be blocked by expression of the Wld(s) protein. Expression of Wld(s) in mammalian or Drosophila neurons delays axon degeneration by days or even weeks after axon severing [3, 16, 30]. In control ddaC cells approximately 98% of animals completely cleared all debris from severed dendrites by 18h after injury (Figure 2A and D). Two different genotypes were used as controls: one was a cross of our tester line (UAS-dicer2; ppk-Gal4,

UAS-mCD8-GFP) to yw flies, and one was a cross of the same tester line to UAS-rtnl2 hairpin RNA flies. We have not observed any phenotypes for expression of the hairpin RNA that targets rtnl2, and results from both genotypes were very similar. The cross to yw resulted in 2/111 animals with some dendrite remnants present at 18h after severing, and the cross to rtnl2 RNAi resulted in no animals out of 25 with any traces of dendrites left 18h after severing. In all figures rtnl2 RNAi images are shown, and the numbers in the tables as controls are crosses to yw.

In ddaC neurons that expressed the Wld(s) protein, severed dendrites were present for days after injury (Figure 2B). For quantitation of the effect, we assayed animals 18h after severing, as in control animals, and found that all animals still had dendrites present 18h after severing (Figure 2D). We conclude that, as for axon degeneration, the Wld(s) protein can dramatically delay dendrite degeneration.

To further test whether dendrites have an active program of degeneration similar to that in axons, we overexpressed the ubiquitin protease, UBP2. This manipulation has previously been shown

28 to block both axon and dendrite pruning [13, 14]. Expression of UBP2 also resulted in dendrites that remained for many days after severing (Figure 2C). In this case 83% of animals still had distal dendrites 18h after severing. Thus dendrites likely have an active program of degeneration that involves the ubiquitin proteasome system (UPS).

Figure 2. Expression of Wld(s) or UBP2 blocks dendrite degeneration. Whole larvae expressing mCD8-GFP under control of the class IV da neuron driver ppk-Gal4 were mounted and subjected to dendrite severing with a pulsed UV laser. Animals were then recovered to food, and remounted for

29 imaging at later time points. In A larvae expressed a hairpin RNA directed against rtnl2 (a control); in B larvae also expressed UAS-controlled Wld(s) [30], and in C they expressed UAS-controlled UBP2. For quantitation (D), WT larvae expressing only mCD8-GFP were assayed for the presence of any dendrite traces at 18h after severing. Similarly, larvae expressing mCD8-GFP and Wld(s) or UBP2 in the ddaC cell were assayed for dendrite traces 18h after severing. Both transgenes were extremely effective at blocking dendrite degeneration.

Dendrite degeneration is independent of mitochondria

As the UPS is required for both dendrite pruning and axon degeneration [12-14], and Wld(s) can also block both, we hypothesized that all three processes may use the same disassembly machinery.

Dendrite pruning has been shown to involve caspases [25, 26], and trophic factor withdrawal also induces a form of axon degeneration that is caspase-dependent [15], although it is not clear that injury- induced axon degeneration uses caspases [31].

To test whether dendrite degeneration might use a mitochondria-dependent pathway similar to apoptosis, we wished to determine whether dendrites completely devoid of mitochondria would degenerate with a similar timing to normal dendrites. Mitochondria are typically distributed throughout axons and dendrites, so we needed to reduce overall levels of mitochondria in dendrites to be able to find a dendrite without any mitochondria. The miro protein is required for both anterograde and retrograde mitochondrial transport [32], so we reasoned that it might also be required for transport of mitochondria into dendrites. Indeed, when we targeted the miro transcript with a hairpin RNA in the ddaE cell, levels of mitochondria were reduced by about 50% in the dorsal comb-like dendrite (Figure

S2).

In neurons expressing mito-GFP, which brightly labels mitochondria, and mCD8-RFP to label membranes as well as the hairpin RNA to target miro, we identified dendrites that lacked mitochondria in distal regions. These were then severed and followed over time. Dendrites lacking mitochondria were completely gone by 12-18h as usual (Figure 3). We performed this experiment in 20 ddaE neurons. In each case we identified a region of the neuron completely devoid of mitochondria and

30 severed it from the cell body as shown in Figure 3. In all instances the distal dendrite was completely cleared by 18h after severing.

We conclude that dendrite degeneration does not require local mitochondria. This suggests that the program of dendrite degeneration may not use apoptotic components, even though these are known to play a role in other forms of pruning and degeneration.

Figure 3. Mitochondria are not required locally for dendrite degeneration. The number of mitochondria in dendrites was reduced by expression of a hairpin RNA targeting the miro transcript. A pan-neuronal Gal4 driver (elav-Gal4) was used to express this hairpin, mito-GFP and mCD8-RFP in neurons. As ddaE cells are relatively separate from other da neurons, and also have shorter dendrites, we could identify dendrites that completely lacked mitochondria in this cell. Two examples (out of 20) are shown. Severed regions of dendrites that lack mitochondria are indicated by brackets; cut sties are indicated by orange arrows. In both cases the severed regions of the dendrite were cleared within the normal time frame. The “f” in the top middle panel indicates a region of the dendrite that is in a different focal plane.

Dendrite degeneration is independent of apoptosis machinery.

It has not been directly tested whether dendrite pruning is dependent on mitochondria, but dendrite pruning is known to depend on caspases [15, 25, 26]. We therefore also wished to determine

31 whether injury-induced dendrite degeneration was caspase dependent. To make sure that the manipulations we performed to alter the apoptotic pathway were effective, we established that we could assay dendrite pruning in the ddaC cell as shown in other studies. We were able to observe complete clearance of ddaC dendrites by 18h APF (Figure 1C and 4). We could also see that clearance of dendrites was delayed by expression of either Wld(s) or UBP2 in about 50% of animals (Figure 4).

Expression of either Wld(s) or UBP2 blocked degeneration of ddaC dendrites more effectively than either blocked pruning in the same cell (Figure 4), so we were reassured that we would be able to determine whether other machinery was shared between degeneration and pruning using this type of comparison.

32

Figure 4. Pruning of ddaC is delayed by Wld(s) and UBP2. Larvae expressing mCD8-GFP in ddaC were aged until 18h APF and then mounted for imaging. One ddaC cell per animal was scored for the presence of dendrite remnants; any dendrite fragments were scored as a positive. Dendrite fragments were frequently observed in larvae expressing UBP2 or Wld(s), but not control larvae. Quantitation is shown as black bars in the graph. Grey bars are the data from Figure 2 for comparison of the effectiveness of Wld(s) and UBP2 in pruning and degeneration

We used three previously tested manipulations to block caspase activity in ddaC cells, and then compared their effects on dendrite degeneration and pruning. Overexpression of the effector caspase inhibitor p35, the Drosophila inhibitor of apoptosis, DIAP1, or RNAi targeting the initiator caspase, dronc, all very effectively delayed clearance of dendrites during pruning (Figure 5). However, in the same cell, none of these manipulations had any effect on degeneration of dendrites after severing

(Figure 5). We conclude that apoptotic machinery is not involved in the program of dendrite degeneration.

33

34

Figure 5. Apoptosis machinery is required for dendrite pruning, but not dendrite degeneration. Animals of the same genotypes were assayed for both dendrite degeneration after severing and dendrite pruning during metamorphosis. Control genotypes are as in Figure 2. In addition to mCD8-GFP, experimental animals expressed Gal4-controlled p35, DIAP1 or a hairpin RNA to target the dronc transcript. For severing experiments larvae were mounted and a dendrite from ddaC was severing with a pulsed UV laser. After 18h recovery on food the presence of dendrite remnants was scored. For pruning experiments, animals were mounted for imaging 18h APF, and presence of ddaC dendrite remnants was scored.

35

The pruning machinery is not required for injury-induced dendrite degeneration.

Several additional proteins that contribute to removal of dendrites during pruning have been identified. The putative microtubule severing protein Kat-60L1 is required for an early step in pruning; the disconnection of dendrites from the cell body [33]. The kinase IK2 seems to play a role in the same pathway [33], and large multidomain protein Mical seems to function at a similar step in dendrite pruning [27]. These pruning studies were performed in the ddaC cell, so we confirmed that in our hands RNAi to target each of these players would delay ddaC pruning. In all cases we found a strong effect (Figure 6). When we severed dendrites in the same cells during larval life, dendrite degeneration proceeded as usual (Figure 6). Thus, while both dendrite pruning and dendrite degeneration can be blocked by expression of Wld(s) or UBP2, the specific machinery used to disassemble dendrites in these two cases seems to be entirely different. We conclude that dendrites have an active program of degeneration uses a different set of machinery from dendrite pruning.

36

37

Figure 6. Specific players required for dendrite pruning are not required for dendrite degeneration. The same strategy shown in Figure 5 was used to test the role of IK2, Kat-60L1 and Mical in dendrite degeneration, again with comparison to pruning in cells of the same genotype. RNAi to reduce each of these proteins was performed by expressing UAS-controlled hairpin RNAs. The presence of dendrite remnants in ddaC was scored 18h after severing or 18h APF.

Discussion

While it is well-established that axons have an active program of degeneration, it has not been directly tested whether dendrites have a similar program. We show that after severing, dendrites are rapidly cleared, and that they undergo morphological changes similar to those seen during dendrite pruning. Moreover, like dendrite pruning, and axon degeneration, dendrite degeneration can be blocked by overexpression or Wld(s) or UBP2.

Because axon degeneration, dendrite pruning and dendrite degeneration involve similar morphological changes and can all be blocked by overexpression of Wld(s) or UBP2, it seems logical that they might all involve the same disassembly machinery. Indeed it has previously been suggested that pruning might be a good model system to use to understand injury-induced degeneration [1-4, 12,

14, 15]. We have rigorously tested this idea by comparing dendrite pruning and injury-induced dendrite

38 degeneration in the same cell. We confirm that pruning uses caspases, IK2, Kat-60L1 and Mical, but none of these proteins is required for injury-induced dendrite degeneration. Mitochondria are also dispensable for injury-induced dendrite degeneration.

We propose that at least two pathways exist to disassemble axons and dendrites. One pathway is caspase-dependent, and is activated during pruning. Recent studies have also shown that caspases are involved in axon degeneration after trophic-factor withdrawal [15], so pruning and loss of trophic support could activate the same disassembly machinery. In contrast, none of the specific machinery used for dendrite pruning is used for injury-induced dendrite degeneration. Thus an entirely different, and as yet unidentified, set of machinery must be used in this process. Injury-induced axon degeneration may share this machinery as it also seems to be a caspase-independent process [31]. But testing this idea will first require identifying specific proteins required for either axon or dendrite degeneration after injury.

Wld(s) or UBP2 can block all disassembly pathways. It is still unclear how Wld(s) protects axons or dendrites, but this is likely to involve changes in NAD+ [34]. These changes must have a very general blocking effect as they inhibit both dendrite pruning and dendrite degeneration, which seem to be separate pathways. The UPS also seems to play a role in both pathways as UBP2 can inhibit both pruning and degeneration. The UPS could be a common effector of both pathways, or different specific substrates could be targeted by the UPS in each pathway.

IK2, Kat-60L1 and Mical seem to be involved in a specific early step in pruning in which the dendrite is clipped from the cell body [27, 33]. Typically dendrites are first disconnected from the cell body at their base (see Figure 5), and then the distal region fragments. When caspases are inhibited this early step can still occur (Figure 5), but when IK2, Kat-60L1 or Mical is targeted, dendrites typically remain connected to the cell body (Figure 6 and [27, 33]). It is unclear whether machinery is needed to

39 fulfill this early disconnection function after dendrite injury, as the injury itself may replace this step.

However, some other machinery must substitute for the later caspase-dependent steps of fragmentation.

In the current study we show that at least two disassembly pathways exist in the same compartment of a single cell, and that one is used to remove dendrites developmentally and one is used to remove them after injury. Additional pathways may exist in the axon of this cell or in dendrites and axons of different cells. Until now it has been assumed that all active disassembly of axons or dendrites would use ubiquitous machinery. As this does not seem to be the case, it will be extremely important to identify the specific machinery activated in each type of disassembly.

40

Materials and Methods

Systems to study dendrite degeneration and pruning.

To study the injury-induced degeneration, we expressed fluorescent proteins in ddaE and ddaC neurons in Drosophila larvae. ddaE neurons are visualized through EB1-GFP and mCD8-GFP driven by 221-Gal4. mCD8-GFP driven by ppk-Gal4 was used to monitor ddaC neurons. Responses to severing were studied in ddaE and ddaC, and pruning was studied in the ddaC neuron.

Dendrite severing was performed by aiming a pulsed UV laser (Photonic Instruments, St.

Charles, IL) on a region of the dendrite fairly close to the cell body. Images were acquired right after severing to make sure dendrites were completely cut and 18h after severing to exam successful removal of the transected part. All experiments were performed on an LSM510 confocal microscope Carl Zeiss,

Thornwood, NY. In time course experiment, the same larva was mounted on slides every 3h. In between imaging, larvae were returned to normal Drosophila media as described [28]

In the mitochondrial involvement assay, mitochondrial were labeled with mito- GFP [35] driven by elav Gal4 while membranes were labeled by mCD8-RFP [36]. To image the ddaC neuron during pupal stages, the pupal cases need to be removed as described [26] before mounting on slides. ImageJ software was employed to analyze and assemble images (http://rsb.info.nih.gov/ij/; NIH). Overviews of neurons are maximum projection Z stacks from confocal images.

Drosophila stocks and RNAi:

The tester lines for RNAi experiments were as follows: 1. UAS-dicer2; 221-Gal4, UAS-EB1-

GFP, UAS-dicer2; 2. UAS-dicer2; ppk-Gal4, mCD8-GFP/TM6, 3. UAS-mCD8-mRFP, UAS- dicer2/CyO; elav-Gal4, UAS-mito-GFP/TM6. Dicer-2 was included in all RNAi experiments to increase the effectiveness of neuronal RNAi [37]. Tester lines for overexpression studies were similar, but did not have dicer2.

41

RNAi or overexpression experiments were performed by crossing the tester lines to the following transgenic fly strains: UAS-dmiro-RNAi (VDRC 106683), UAS-Dronc-RNAi (VDRC

23035), UAS- Katanin-p60L1-RNAi (VDRC 31598), UAS-IK2-RNAi (VDRC 103748), pUAST-

Wld(s) [30], UAS-UBP2 [38], UAS-p35 (Bloomington Drosophila Stock Center, Bloomington, IN),

UAS-DIAP1 (Bloomington Drosophila Stock Center, Bloomington, IN). Crosses to UAS-rtnl2-RNAi

(VDRC 33318) and yw flies were used as controls.

Acknowledgements

We are very grateful to Wes Grueber, Bing Ye, Marc Freeman, the Bloomington Drosophila

Stock Center, and the Vienna Drosophila RNAi Center (VDRC) for Drosophila lines. This work was supported by R21 NS066216. MMR is a Pew Scholar in the Biomedical Sciences.

42

References 1. Saxena, S. and P. Caroni, Mechanisms of axon degeneration: from development to disease. Prog Neurobiol, 2007. 83(3): p. 174-91. 2. Raff, M.C., A.V. Whitmore, and J.T. Finn, Axonal self-destruction and neurodegeneration. Science, 2002. 296(5569): p. 868-71. 3. Luo, L. and D.D. O'Leary, Axon retraction and degeneration in development and disease. Annu Rev Neurosci, 2005. 28: p. 127-56. 4. Ehlers, M.D., Deconstructing the axon: Wallerian degeneration and the ubiquitin-proteasome system. Trends Neurosci, 2004. 27(1): p. 3-6. 5. Vargas, M.E. and B.A. Barres, Why is Wallerian degeneration in the CNS so slow? Annu Rev Neurosci, 2007. 30: p. 153-79. 6. Li, P. and T.H. Murphy, Two-photon imaging during prolonged middle cerebral artery occlusion in mice reveals recovery of dendritic structure after reperfusion. J Neurosci, 2008. 28(46): p. 11970-9. 7. Murphy, T.H., et al., Two-photon imaging of stroke onset in vivo reveals that NMDA-receptor independent ischemic depolarization is the major cause of rapid reversible damage to dendrites and spines. J Neurosci, 2008. 28(7): p. 1756-72. 8. Greenwood, S.M., et al., Mitochondrial dysfunction and dendritic beading during neuronal toxicity. J Biol Chem, 2007. 282(36): p. 26235-44. 9. Oliva, A.A., Jr., T.T. Lam, and J.W. Swann, Distally directed dendrotoxicity induced by kainic Acid in hippocampal of green fluorescent protein-expressing transgenic mice. J Neurosci, 2002. 22(18): p. 8052-62. 10. Zeng, L.H., et al., Kainate seizures cause acute dendritic injury and depolymerization in vivo. J Neurosci, 2007. 27(43): p. 11604-13. 11. Williams, D.W. and J.W. Truman, Remodeling dendrites during insect metamorphosis. J Neurobiol, 2005. 64(1): p. 24-33. 12. Zhai, Q., et al., Involvement of the ubiquitin-proteasome system in the early stages of wallerian degeneration. Neuron, 2003. 39(2): p. 217-25. 13. Watts, R.J., E.D. Hoopfer, and L. Luo, Axon pruning during Drosophila metamorphosis: evidence for local degeneration and requirement of the ubiquitin-proteasome system. Neuron, 2003. 38(6): p. 871-85. 14. Kuo, C.T., L.Y. Jan, and Y.N. Jan, Dendrite-specific remodeling of Drosophila sensory neurons requires matrix metalloproteases, ubiquitin-proteasome, and ecdysone signaling. Proc Natl Acad Sci U S A, 2005. 102(42): p. 15230-5. 15. Schoenmann, Z., et al., Axonal degeneration is regulated by the apoptotic machinery or a NAD+-sensitive pathway in insects and mammals. J Neurosci, 2010. 30(18): p. 6375-86. 16. Hoopfer, E.D., et al., Wlds protection distinguishes axon degeneration following injury from naturally occurring developmental pruning. Neuron, 2006. 50(6): p. 883-95. 17. Zhong, L., R.Y. Hwang, and W.D. Tracey, Pickpocket is a DEG/ENaC protein required for mechanical nociception in Drosophila larvae. Curr Biol, 2010. 20(5): p. 429-34. 18. Hughes, C.L. and J.B. Thomas, A sensory feedback circuit coordinates muscle activity in Drosophila. Mol Cell Neurosci, 2007. 35(2): p. 383-96. 19. Song, W., et al., Peripheral multidendritic sensory neurons are necessary for rhythmic locomotion behavior in Drosophila larvae. Proc Natl Acad Sci U S A, 2007. 104(12): p. 5199- 204. 20. Gao, F.B. and B.A. Bogert, Genetic control of dendritic morphogenesis in Drosophila. Trends Neurosci, 2003. 26(5): p. 262-8.

43

21. Gao, F.B., et al., Genes regulating dendritic outgrowth, branching, and routing in Drosophila. Genes Dev, 1999. 13(19): p. 2549-61. 22. Grueber, W.B., L.Y. Jan, and Y.N. Jan, Tiling of the Drosophila epidermis by multidendritic sensory neurons. Development, 2002. 129(12): p. 2867-78. 23. Grueber, W.B. and Y.N. Jan, Dendritic development: lessons from Drosophila and related branches. Curr Opin Neurobiol, 2004. 14(1): p. 74-82. 24. Sugimura, K., et al., Distinct developmental modes and lesion-induced reactions of dendrites of two classes of Drosophila sensory neurons. J Neurosci, 2003. 23(9): p. 3752-60. 25. Kuo, C.T., et al., Identification of E2/E3 ubiquitinating enzymes and caspase activity regulating Drosophila sensory neuron dendrite pruning. Neuron, 2006. 51(3): p. 283-90. 26. Williams, D.W., et al., Local caspase activity directs engulfment of dendrites during pruning. Nat Neurosci, 2006. 9(10): p. 1234-6. 27. Kirilly, D., et al., A genetic pathway composed of Sox14 and Mical governs severing of dendrites during pruning. Nat Neurosci, 2009. 12(12): p. 1497-505. 28. Stone, M.C., et al., Global up-regulation of microtubule dynamics and polarity reversal during regeneration of an axon from a dendrite. Mol Biol Cell, 2010. 21(5): p. 767-77. 29. Kerschensteiner, M., et al., In vivo imaging of axonal degeneration and regeneration in the injured . Nat Med, 2005. 11(5): p. 572-7. 30. MacDonald, J.M., et al., The Drosophila cell corpse engulfment receptor Draper mediates glial clearance of severed axons. Neuron, 2006. 50(6): p. 869-81. 31. Finn, J.T., et al., Evidence that Wallerian degeneration and localized axon degeneration induced by local neurotrophin deprivation do not involve caspases. J Neurosci, 2000. 20(4): p. 1333-41. 32. Russo, G.J., et al., Drosophila Miro is required for both anterograde and retrograde axonal mitochondrial transport. J Neurosci, 2009. 29(17): p. 5443-55. 33. Lee, H.H., L.Y. Jan, and Y.N. Jan, Drosophila IKK-related kinase Ik2 and Katanin p60-like 1 regulate dendrite pruning of sensory neuron during metamorphosis. Proc Natl Acad Sci U S A, 2009. 106(15): p. 6363-8. 34. Coleman, M.P. and M.R. Freeman, Wallerian degeneration, wld(s), and nmnat. Annu Rev Neurosci, 2010. 33: p. 245-67. 35. Pilling, A.D., et al., Kinesin-1 and Dynein are the primary motors for fast transport of mitochondria in Drosophila motor axons. Mol Biol Cell, 2006. 17(4): p. 2057-68. 36. Ye, B., et al., Growing dendrites and axons differ in their reliance on the secretory pathway. Cell, 2007. 130(4): p. 717-29. 37. Dietzl, G., et al., A genome-wide transgenic RNAi library for conditional gene inactivation in Drosophila. Nature, 2007. 448(7150): p. 151-6. 38. DiAntonio, A., et al., Ubiquitination-dependent mechanisms regulate synaptic growth and function. Nature, 2001. 412(6845): p. 449-52.

44

Supplementary Materials

Figure S1. Dendrites of ddaE neurons degenerate rapidly after severing. The multidendritic neuron ddaE was labeled with EB1-GFP. The binary Gal4-UAS system was used to express the fluorescent protein in a subset of neurons. The 221-Gal4 expresses at high levels in class I dendritic arborization (da) neurons including ddaE. Whole larvae were mounted on a microscope slide and the dorsal dendrite of the ddaE neuron was severed with a pulsed UV laser. The cell was imaged immediately after severing. The larva was then recovered to food and remounted for imaging at 3h intervals. The orange arrow indicates the site of severing and the bracket indicates the region of the dendrite separated from the cell body by the cut.

Figure S2. Targeting miro by RNAi reduces the number of mitochondria in dendrites. Animals expressed hairpin RNAs to target either rtnl2 (control) or miro in all neurons under control of elav- Gal4. Mito-GFP and mCD8-RFP were also expressed in all neurons by elav-Gal4. Total numbers of mitochondria in the dorsal comb-shaped dendrite of the ddaE neuron were counted (n=5 for control and n=4 for miro RNAi). The difference between the average number of mitochondria in this dendrite between control and miro RNAi was statistically significant using an unpaired t-test (p< 0.05).

45

Movie 1. Live imaging of EB1-GFP dynamics in ddaC 4.5 hours after dendrite severing. URL: http://www.jneurosci.org/content/suppl/2011/04/06/31.14.5398.DC1/Movie_S1.mov Frames were acquired every 2s; lookup table was inverted for the movie. Growing microtubules appear as dark spots that move in consecutive frames, and several examples are indicated with stars. The frame width in the zoomed in part of the movie is 105 microns.

Movie 2. Live imaging of mCD8-GFP in ddaC 4 hours after dendrite severing. URL: http://www.jneurosci.org/content/suppl/2011/04/06/31.14.5398.DC1/Movie_S2.mov Images were acquired every 2s. Some mCD8-GFP was also expressed in epithelial cells in this animal, and in some parts of the movie edges of these hexagonal cells can be observed. Two forming membrane blebs are indicated with stars. The frame width in the zoomed in part of the movie is 134 microns.

Movie 3. Live imaging of EB1-GFP in ddaC 3 hours after dendrite severing. URL: http://www.jneurosci.org/content/suppl/2011/04/06/31.14.5398.DC1/Movie_S3.mov Frames were acquired every 2s. Diffuse EB1-GFP labels the cytoplasm, and a protrusion can be seen extending from the side of one of the dendrite branches (star). The width of the movie is 210 microns.

Movie 4. Live imaging of EB1-GFP in ddaC 9 hours after dendrite severing. URL: http://www.jneurosci.org/content/suppl/2011/04/06/31.14.5398.DC1/Movie_S4.mov Frames were acquired every 2s. A dramatic example of blebbing (arrows) can be seen near the cut site in this dendrite. Frame width is 210 microns.

Movie 5. Membranes of ddaC labeled with mCD8-GFP are shown 9 hours after dendrite severing. URL: http://www.jneurosci.org/content/suppl/2011/04/06/31.14.5398.DC1/Movie_S5.mov Frames were acquired every 2 seconds. In contrast to mCD8-GFP labeled membranes at earlier time points, the brightest structures are now vesicles that look like they are contained internally. The width of the movie frame is 80 microns.

Movie 6. Live imaging of mCD8-GFP in ddaC 12 hours after axon severing. URL: http://www.jneurosci.org/content/suppl/2011/04/06/31.14.5398.DC1/Movie_S6.mov Frames were acquired every 2 seconds. Bright internal membranes are visible in the disintegrating axon. The width of the frame in the zoomed in part of the movie is 50 microns.

46

Chapter Three

Kinesin-13 terminates microtubule dynamics in degenerating axons before energy collapse

This chapter is part of a manuscript with the introduction, discussion and parts of the results written with Melissa Rolls.

Abstract

Wallerian degeneration of axons after injury is believed to be an active process. However, the events that require energy are largely unknown. To identify active events in degeneration, we used an energy collapse sensor to pinpoint the time at which ATP levels drop below the critical threshold for driving cellular behaviors. Energy collapse in individual axons occurred between 2.5 and 3 hours after axon severing in an in vivo Drosophila model. Only events occurring before this time can be actively controlled. A candidate early degeneration event is disassembly of microtubules. Termination of microtubule growth occurred 2 hours before energy collapse, indicating this is likely to be actively controlled. Expression of the Wallerian degeneration slow protein (Wlds) delayed energy collapse by several days, but did not delay termination of microtubule dynamics. This result confirms that energy loss is not responsible for stopping microtubule growth. Moreover, the kinesin-13 Klp59C was required for normal timing of microtubule growth termination. Reduction of Klp59C also delayed axon degeneration. We propose that kinesin-13s act in severed axons to terminate microtubule dynamics before energy collapse, and that active microtubule disassembly facilitates rapid degeneration of damaged axons.

47

Introduction

The discovery of the Wlds mouse in 1989 [1] suggested for the first time that degeneration of axons after injury might be an active process rather than a passive wasting away [2]. This observation, coupled with the fact that axon degeneration appears to proceed through a stereotyped series of events that end with membrane beading, has led to a model in which Wallerian degeneration is an active process analogous to apoptosis [2,3,4,5,6,7]. However, it has been difficult to identify key active events that contribute to degeneration. While the ubiquitin-proteasome system [8,9] and calpain proteases [10] seem to be involved in Wallerian degeneration, the process remains largely mysterious at the molecular level [7].

Active cellular processes, presumably including Wallerian degeneration, are driven by consumption of ATP. Supplies of ATP are primarily maintained by mitochondrial oxidative phosphorylation, and ATP and GTP levels are coupled though nucleoside diphosphate kinase. In axons severed from the cell body, local ATP/GTP levels could be maintained for some time by axonal mitochondria, as these organelles are abundant in axons. The model of active axonal disassembly predicts that cytosolic ATP and GTP are maintained at levels that can drive cellular processes, presumably through mitochondrial function, throughout degeneration. However, whether ATP levels are maintained or whether cytosolic energy collapses during axon degeneration has not been determined.

One event that is thought to be central to axon degeneration is deconstruction of the cytoskeleton, and on first principles this could be accomplished either by ATP-dependent proteins, or passively as a result of energy depletion. Specifically, microtubule (MT) disassembly has been proposed to be an early step in injury-induced axon degeneration [3,4]. The simplest model for disassembly of microtubules after axon injury is that it is passive and simply secondary to energy depletion. Microtubules are dynamic polymers that constantly grow and shrink at the plus end [11],

48 even in neurons where microtubules tend to be very stable [12]. Subunits are added to the plus end as alpha-beta tubulin dimers. The beta subunit is a GTPase, and exchanges GDP for GTP in the cytosol.

Only the GTP-bound form can be added to growing microtubule plus ends. If no GTP-bound tubulin is present, then growth is stopped, and only disassembly can occur. Depletion of GTP in the cytosol could therefore result in microtubule disassembly. Alternatively, microtubules could be actively disassembled by a microtubule severing protein or by a depolymerase that removes subunits from the end.

Microtubule severing proteins, which include the AAA ATPases spastin, fidgetin and katanin [13], and depolymerases, which include kinesin-13 proteins [14,15], require ATP to work on microtubules.

As microtubule disassembly could occur passively or actively during axon degeneration, we developed a strategy to position its timing relative to energy collapse. This analysis has led us to a new stepwise understanding of axon degeneration, in which microtubule disassembly occurs before energy collapse, which in turn occurs before membrane beading. Moreover, this approach allowed us to pinpoint when Wlds action plays a role in delaying degeneration.

Results

Rab proteins can be used to time energy collapse in individual axons

To determine when active processes become impossible during axon degeneration, we needed a way to interrogate energy status in individual axons in vivo. We reasoned that the most useful assay would measure not absolute ATP or GTP levels, but the ability of the cell to perform ATP- or GTP- dependent processes.

Rab proteins are small GTPases that are cytosolic in the GDP-bound form, and membrane- localized in the GTP-bound form [16]. In healthy cells Rab proteins are often used as markers of different endocytic compartments. Because Rabs are membrane bound only after binding GTP in the

49 cytosol, we hypothesized that their localization could be used to when GTP, and thus also tightly coupled ATP, levels fall below a critical threshold that can drive active processes.

To test whether Rab protein localization could be a useful indicator of energy collapse, we expressed a fluorescently tagged Rab protein, Rab11a-Venus [17,18], in HeLa cells and induced ATP depletion with either hydrogen peroxide (H2O2) or 2-Deoxy-d-glucose (2-DOG). H2O2 induces very rapid ATP decline in cells, with complete energy depletion after 15 minutes of 32 mM H2O2 [19].

Similarly, 2h treatment of 50mM 2-DOG severely decreases cellular ATP level [20]. We used the concentrations of H2O2 and 2-DOG to determine whether we could see rapid changes in Rab localization when ATP levels fall suddenly. We coexpressed an actin-binding fusion protein, LifeAct- mTurquoise2 [21], to report on overall cellular integrity, and incubated cells with the TMRM dye that accumulates in mitochondria when they are healthy and maintain membrane potential [22]. In control cells Rab11a-Venus was present in small punctae throughout the cytoplasm, and TMRM accumulated in mitochondria (Figure 1). We treated the cells with H2O2 at different time points from 3 minutes to

15 minutes. We also measured the ATP levels at each time point. After 3 minutes, TMRM and Rab11a-

Venus appeared indistinguishable from the untreated cells, and the ATP level was not significantly decreased (Figure 1). After 6 minutes, ATP level started to drop more severely; in 45% of the cells,

TMRM no longer accumulated in mitochondria and Rab11a-Venus was completely diffuse. The percentages of cells with decreased TMRM fluorescence and diffuse Rab increase with longer treatment time (Figure 1). The control, LifeAct- mTurquoise2, remained visible in stress fibers, and cell shape and overall appearance were unchanged at each time point (Figure 1). In addition to H2O2, we used another pharmaceutical drug, 2-Deoxy-d-glucose (2-DOG), to induce energy depletion in HeLa cells. We observed the similar changes in Rab11a distribution with 2-DOG treatment for 2h (Figure 2).

The cellular ATP level after 2h treatment of 2-DOG fell to similar levels as 15min H2O2 treatment

(Figure 2C). However, unlike H2O2 treatment, TMRM fluorescence tended to decrease slightly with a

50 large variance between individual cells after 2-DOG treatment (Figure 2B). The lack of TMRM change in 2-DOG treated cells is likely due to the fact that 2-DOG depletes cellular ATP through inhibiting glycolysis and does not interfere with mitochondria function [23][24]. Taken together, redistribution of

Rab11a from membranes to the cytosol occurred rapidly after acute ATP depletion. This result suggests that Rab protein localization might be a useful indicator to determine the time of energy collapse during axon degeneration.

51

Figure 1. Rab proteins redistribute rapidly in response to H2O2-induced ATP depletion. HeLa cells were transfected with plasmids encoding tagged Rab11a and LifeAct and were incubated with the mitochondrial membrane potential dye TMRM. Cells were either mock treated (top) or treated with 32 mM H2O2 (bottom). Rab11a-Venus distribution was scored as punctate or diffuse in individual cells. Statistical significance was calculated with a Fisher’s exact test.

52

Figure 2. Rab proteins respond similarly to 2-Deoxy-D-glucose-induced ATP deprivation. HeLa cells were transfected with Rab11a-venus and Lifeact for 24h and treated with TMRM dye for 15min. Then cells are exposed to PBS (top) or 50mM 2-DOG (bottom) for 120min, following immediate fixation and imaging. Same methods are used to score the Rab status. Statistical significance was calculated with a Fisher’s exact test.

53

We next tested whether Rab protein localization could be used to probe energy collapse in individual severed axons in vivo. Axon degeneration in Drosophila dendritic arborization (da) neurons is morphologically similar to that in mammalian neurons, beginning with a latent phase in which no overt changes in shape take place, followed by rapid beading [25]. In addition, like Wallerian degeneration of mammalian axons, axon degeneration in Drosophila is delayed by expression of the

Wlds protein [26,27]. We therefore monitored Rab localization during axon degeneration of Drosophila da neurons. These cells have their cell body and dendrite arbors in the body wall, and their axons transmit sensory information to the central nervous system [28,29]. We severed individual axons of two different types of da neurons expressing fluorescently tagged Rab proteins together with a plasma membrane marker. In both the ddaE neuron, which has a relatively simple dendrite arbor [29], and the ddaC neuron, which has a large, complex arbor [29], Rab11-GFP and Rab5-YFP switched from punctate localization to diffuse localization by 3.5 hours after severing with a pulsed UV laser in whole animals (Figure 3 and S1). As both markers behaved similarly, but Rab5-YFP labeled slightly larger membrane punctae that were a little easier to visualize, we continued the experiments with Rab5. In all experiments, similar results were obtained in ddaE and ddaC neurons.

To test how rapidly Rab5-localization might change in response to energy collapse, we timed normal exchange of Rab5 off the membrane with fluorescence recovery after photobleaching (FRAP).

We bleached individual punctae in axons, and monitored fluorescence recovery. Complete fluorescence recovery occurred within 6 minutes (Figure S2). Thus the population of Rab5-YFP on membranes turns over completely within 6 minutes, and so within 6 minutes of cytosolic GTP pools depleted below the critical threshold for driving intracellular processes, Rab5-YFP should become cytosolic. To confirm that membrane localization of Rab5-YFP requires GTP-binding in individual axons, we expressed mutant forms of Rab5 that are either constitutively GDP-bound (DN) or GTP-bound (CA). Rab5DN-

YFP was always diffuse in axons (Figure S2) and Rab5CA-YFP was punctate and did not exchange off

54 the membrane (Figure S2). Moreover, Rab5CA-YFP remained membrane-bound during axon degeneration (Figure 3).

Figure 3. Energy collapse occurs 3 hours after axon severing. Axons of ddaE neurons expressing tagged Rab proteins together with mCD8-RFP were severed with a pulsed UV laser in whole larvae (time 0). Larvae were remounted for imaging at different times after severing and Rab proteins were scored as diffuse or punctate. Example images at different times are shown, and quantitation of Rab5- YFP distribution after injury is shown in the graph. Numbers on bars in graph indicate number of cells analyzed.

55

If the cytosolic distribution of Rab5-YFP reflects energy collapse specifically and not general disruption of intracellular compartments, other organelle markers that do not require GTP should remain membrane-bound during axon degeneration. Lamp1-GFP, which is targeted to lysosomes, remained punctate at 3.5 hours after axon severing, as did the mitochondrial marker mito-GFP (Figure

S3). In addition, none of the different tagged Rab proteins, or RNAi of Rab5, altered the timecourse of axon degeneration (Figure S4), consistent with Rab proteins acting as passive reporters of intracellular events rather than key regulators of degeneration. We conclude that tagged Rab proteins behave as useful reporters of energy collapse during axon degeneration, and that this event occurs between 3 and

3.5 hours after axon degeneration.

To further prove that Rab diffusion correlates with energy depletion, we employed an in vivo method to make ATP decrease faster after injury. If Rab protein diffusion directly corresponds with low energy, we would expect to see Rab5-YFP detach from the membrane faster. To achieve a faster decrease in ATP, we applied Nmnat RNAi in ddaC neurons. Nmnat is the important enzyme involved in salvage pathway of NAD biosynthesis in Drosophila. The oxidative form of NAD, NAD+, plays a crucial role in ATP synthesis [30]. Decreased endogenous level of Nmnat though Nmnat RNAi presumably causes local Nmnat level to drop significantly after injury, which then affects normal NAD genesis and further decreases ATP levels [31]. Before injury, Rab5-YFP appeared normal in both control and Nmnat RNAi neurons, remaining at the membrane in punctae. After severing, Rab5-YFP became diffuse 3.5h after injury in control larvae. However, using Nmnat RNAi, we observed that

Rab5-YFP detached from membrane and became diffuse 2.5h after axotomy (Figure 4).

In transected axons, decreasing the basal level Nmnat has a more severe effect on ATP maintenance for the following three reasons: 1. Lack of Nmnat compensation from the soma; 2. ATP consumption to actively degrade cellular components; and 3. Decreased ATP synthetic capabity. It is possible that Rab proteins only become diffuse when around 30% of the cellular ATP is left as indicated

56 in the cell culture experiments (Figure 1B, Figure 2C). This likely explains why Rab proteins are able to maintain their membrane localization uninjured Nmnat RNAi cells, as the ATP level was not reduced below the threshold.

Additionally, we also tested whether dmiro RNAi, which was reported to decrease transport of mitochondria to the axon [32], can also accelerate detachment of Rab proteins after injury. However, dmiro RNAi only modestly accelerated the Rab diffusion rate (Figure 4). The result is not surprising as the we still observed many mitochondria in the proximal axon region that we were monitoring. The number of mitochondria in dmiro RNAi axons is likely sufficient to meet the ATP needs in both healthy and injured axons. As a result, Rab diffuses at normal rates in injured dmiro RNAi axons. Collectively, genetic manipulations that cause ATP to decrease faster can lead to more rapid Rab detachment in vivo.

These data provide further in vivo evidence that Rab protein is sensitive to cellular energy change and can serve as an energy monitor in neuron degeneration.

57

58

Figure 4. Disturbance of endogenous Nmnat accelerates Rab diffuse after injury. A. Axons of ddaE neurons were transected proximal to soma at 0h. The larva was recovered on the standard media at 20°C and remounted to microscope after 2.5h. B. Axons with diffuse Rab were scored and the statistic significance was calculated by a fisher’s exact test.

Microtubule growth terminates before energy collapse

Having defined the time of energy collapse, we could now identify active processes that precede this tipping point. Membrane beading is thought to be a relatively late event, so we first tested whether this occurred before or after energy collapse. Energy collapse in ddaC neurons is complete by

3.5 h (Figure S1), but none of these cells exhibited membrane beading at this time (Figure 5A and B).

Therefore membrane beading occurs after energy collapse.

Microtubule disassembly has been proposed to be an early event in degeneration. We focused on one aspect of microtubule disassembly, the termination of microtubule growth, as this could be mediated either by energy loss or by action of a depolymerase. To determine how long microtubule dynamics continued in severed axons, we cut axons from neurons expressing the plus end-binding protein EB1-GFP. In healthy axons, EB1-GFP binds to growing microtubules that extend away from the cell body and falls off when they shrink. Therefore presence of EB1-GFP comets in a cell or axon indicates microtubule growth is occurring.

We expressed EB1-GFP in subsets of neurons in Drosophila larvae and severed individual axons in whole animals. After injury, animals were returned to food and then remounted for imaging at different intervals. Immediately after severing, comets were visible in 60% of ddaE and ddaC neurons

(Figure 5C and D). By 30 minutes after injury about 30% of axons had comets, and by one hour only occasional severed axons still had growing microtubules. Thus microtubules ceased growing in injured axons several hours before Rab proteins lost the ability to associate with membranes. The early

59 termination of microtubule growth suggests that it could be a process that is actively controlled during

Wallerian degeneration.

Figure 5. Cessation of microtubule growth is an early event in axon degeneration compared to energy collapse and membrane beading. (A and B) Axons of ddaC neurons expressing mCD8-GFP were severed at time 0 and animals were remounted for imaging at different times after injury. Axons were scored as smooth and continuous, as in the images from 0h and 3.5h, or beaded, as in the 12h example. (C and D) EB1-GFP was expressed in ddaE and ddaC neurons and axons were severed at time 0 with a pulsed UV laser. Animals were remounted for imaging at different times after axon injury. Timeseries images were acquired and movies were analyzed for the presence of EB1-GFP comets and scored as comets present or absent. Numbers on the bars in the graphs indicate animals tested in each condition. In D statistical significance was calculated with a Fisher’s exact test.

Wlds delays energy collapse, but not termination of microtubule dynamics, for days.

To further test the hypothesis that microtubule growth is terminated before energy collapse in degenerating axons, we examined Rab5-YFP localization and EB1-GFP comets in neurons expressing the Wlds protein. The Wlds protein consists of the complete nmnat polypeptide, which is involved in

NAD biosynthesis, and an N-terminal extension derived from the non-enzymatic region of a ubiquitin

60 ligase and an intergenic sequence [2]. Wlds expression delays degeneration of severed axons for days to weeks in vertebrates including fish [33] and rodents [1], as well as a variety of cell types in

Drosophila [26,27]. It has been suggested that Wlds acts to delay degeneration by affecting NAD levels

[34], by affecting mitochondrial calcium handling and motility [35], by stabilizing mitochondrial redox state [33], or by acting as a chaperone [36].

In ddaE neurons expressing Wlds, severed axons remained continuous for more than two days after injury (Figure 6). Moreover, 48h after severing, Rab5-YFP retained a punctate distribution in the majority of detached axons (Figure 6B), this is much longer than in control axons in which all Rab5-

YFP was diffuse 3.5 hours after severing (Figure 2). This delay in Rab5-YFP delocalization in Wlds- expressing neurons suggests that Wlds keeps ATP and GTP available in severed axons.

If termination of microtubule dynamics was linked to availability of ATP and GTP, we predicted it should also be delayed for days by expression of Wlds. However, when we monitored EB1-GFP comets in axons severed from Wlds-expressing neurons, very few axons contained comets even 1.5 hours after injury (Figure 6C and D). Thus in Wlds-expressing neurons microtubules stop growing at least 2 days before Rab proteins lose the ability to associate with membranes. To further prove that the termination of microtubule dynamics and ATP/GTP depletion are separate processes, we examined the presence of EB1 comet in nmnat RNAi ddaE neurons. Nmnat RNAi causes Rab5 YFP to fall off the membrane after axotomy faster (Figure 4). However, we did not observe any alternations of termination of EB1 comet after axotomy. Similar to controls, EB1 comets were rarely observed in the transected nmnat RNAi axons. Temporal separation is consistent with microtubule dynamics being terminated by a mechanism that is directly triggered by injury, is not sensitive to Wlds or down regulation of Nmnat, and could require ATP or GTP.

61

62

Figure 6. Wlds delays energy collapse, but not termination of microtubule dynamics, for days. Decrease of Nmnat does not affect termination of microtubule dynamics. (A and B) Rab5-YFP, mCD8-RFP and Wlds were expressed in class I neurons. Axons of ddaE were severed and animals were remounted for imaging at different times after injury. Rab5-YFP was scored as punctate or diffuse at each time point. (C and D) EB1-GFP and Wlds were expressed in class I neurons, and whole animals were mounted for injury and imaging. Movies of EB1-GFP were obtained at different times after imaging. In the example image EB1-GFP comets (arrowheads) can be seen in axons from the neighboring ddaC and ddaD neurons, but EB1-GFP is entirely diffuse in the severed axon ddaE axon. Presence of comets was scored in severed axons at different times. Numbers on the bars in B and D indicate number of animals tested at each timepoint. In D statistical significance was calculated with a Fisher’s exact test. E. EB1-GFP and nmnat RNAi were expressed in class I neurons. Axons of ddaE neurons were transected proximal to the soma. Movies of EB1-GFP were obtained at different times after injury. EB1 comets are indicated in arrowhead and cut sites are indicated in arrows. Presence of comets was scored in severed axons at different times. Numbers on the bars indicate number of animals tested at each time point. Statistical significance was calculated with a Fisher’s exact test.

63

Klp59C is required for normal timing of axon degeneration.

If cessation of microtubule dynamics in degenerating axons is an active process, then we should be able to identify an enzyme responsible for it. Factors that promote microtubule disassembly have been identified and include severing proteins and depolymerases. We performed RNAi to reduce levels of known microtubule disassembly factors and found that targeting Klp59C, a kinesin-13 microtubule depolymerase, delayed degeneration (Figure 7 and S5). Expression of large RNA hairpins targeting other microtubule disassembly proteins had no effect (Figure S5). To confirm that the delay in degeneration was due to reduction of Klp59C, we tested animals with only one copy of the Klp59C gene (Klp59C Df/+). In these heterozygous animals, the number of intact axons 12 hours after severing was intermediate between control (rtnl2 RNAi) animals and animals in which Klp59C was targeted by

RNAi (Figure 7). In addition to the heterozygous deficiency, we also performed the same axon severing assay in homozygous Klp59C mutants. Due to the genetic availability, the severing experiment was only performed in ddaE class I neurons. Similar to Klp59C RNAi larvae and heterozygous mutants, the homozygous mutants showed a significant delay in degeneration after transection (Figure 7C). It’s surprising that the homozygous mutant does not have a more severe effect compared to heterozygous mutants and RNAi, but perhaps the complete depletion of Klp59C causes severe consequences and thus triggers the up-regulation of some unknown mechanisms to compensate for the loss. In this way, cells are able to avoid catastrophic cell death and survive. These results are consistent with two conclusions: the degeneration delay is due to reduction of Klp59C, and expression of the 48576 RNAi transgene results in strong reduction of Klp59C protein levels. However, in these experiments we were examining overall timing of axon beading, and did not directly examine microtubules.

64

Klp59C terminates microtubule dynamics after axon severing.

To determine whether Klp59C is the protein that is responsible for terminating microtubule dynamics in severed axons, we assayed EB1-GFP comets in neurons expressing the 48576 RNAi transgene. In neurons expressing a control RNAi hairpin, only 10% of axons had comets 1.5 hours after severing, but, when Klp59C was targeted by RNAi, 40% of axons retained comets at this timepoint

(Figure 8A and B). We conclude that Klp59C acts to terminate microtubule growth in severed axons.

We propose that this protein acts in the first hour to stop microtubules from growing, that this is followed by energy collapse two hours later, and finally after several more hours membranes bead and the axon breaks down (Figure 8C).

65

66

Figure 7. Klp59C is required for timely axon degeneration. EB1-GFP and dicer2 were expressed in ddaC and ddaE neurons with either the controls or RNAi targeting Klp59C. Alternately, EB1-GFP was expressed in animals with only one copy of the Klp59C gene (Df/+). Axons were severed at time 0 and scored as continuous or not at 12h after injury. Numbers on the bars in the graph indicate number of animals tested for each genotype. A Fisher’s exact test was used to calculate statistical significance.

Figure 8. Klp59C is required for early termination of microtubule dynamics after axon injury. (A and B) Either a control RNAi hairpin (rtnl2) or the 48576 hairpin targeting Klp59C was expressed together with EB1-GFP in ddaC neurons. Axons were severed and imaging of EB1-GFP comets was performed 1.5h later. Each severed axon was scored as either containing comets or not. Numbers on the bars in the graph indicate number of animals tested for each genotype. A Fisher’s exact test was used to calculate statistical significance. (C) A summary of timing of key events in degeneration is shown.

Discussion

Two features of Rab proteins, here tested for Rab5, make them potentially useful as sensors of available cytosolic energy: first, they are GTPases that require GTP to associate with membranes, and second, they cycle off the membrane and must acquire new GTP in the cytosol every few minutes. In

67 addition, Rab5 seems to act as a bystander rather than an active participant during axon degeneration, as neither over- or under-active forms affect the process. Thus we propose that the time at which Rab5 becomes cytosolic during axon degeneration represents the time at which the levels of cytosolic ATP and GTP can no longer drive cellular processes, and that any events after this time cannot be actively controlled. Monitoring Rab localization therefore provides a temporal landmark during degeneration; events that occur before this landmark could potentially be active, but ones that occur after it are unlikely to require ATP or GTP. Using this logic for axon degeneration, we find that termination of microtubule dynamics occurs before energy collapse, while membrane beading occurs after this time.

Thus microtubule dynamics could be arrested in an ATP- or GTP-dependent process, while, surprisingly, membrane beading is likely to be a process that is not actively controlled. Membrane beading, which is a classic step in Wallerian axon degeneration, could instead be a direct result of loss of cytosolic energy, and thus may be more similar to necrotic membrane blebbing than apoptotic blebbing, which occurs before energy collapse [19].

Expression of Wlds provided support for the idea that cellular degeneration events can be separated into those that require energy and those that occur after energy collapse. Axons severed from neurons that expressed Wlds maintained punctate Rab5 localization and remained unbeaded for over 2 days. In contrast, termination of microtubule dynamics occurred in almost all cells within 1.5 hours of injury. This result suggests that membrane beading is coupled to energy collapse, but that termination of microtubule dynamics is not. One important implication of this finding is that delaying degeneration by expressing Wlds only blocks some, but not all, aspects of axonal disassembly.

While tracking timing of Rab delocalization confirmed that termination of microtubule dynamics is an early and potentially active event during axon degeneration, it also suggested that energy collapse is a much earlier event than previously assumed. The prevailing model of Wallerian degeneration is that there is a latent phase during which the severed axon remains essentially

68 unchanged, and then the axon is rapidly disassembled [37]. Our data suggests that energy collapse occurs well before overt signs of degeneration appear. In Drosophila da neurons Rabs delocalize around 3 hours after severing, but membrane beading does not occur until 6-9 hours after injury [25].

Consistent with termination of microtubule dynamics being an active event during axon degeneration, we identified the kinesin-13 Klp59C as a factor required for normal timing of this event.

In addition, reduction of Klp59C delayed membrane beading suggesting termination of microtubule dynamics is functionally important for degeneration. Kinesin-13s are ATPases that act at microtubule ends to trigger depolymerization [14,15]. One of the mammalian kinesin-13s, KIF2A, was recently found to be required for normal timing of axon degeneration in vitro and in vivo [38], however, its role in terminating microtubule dynamics was not examined. We propose that kinesin-13s are conserved factors that terminate microtubule dynamics after injury to promote axon degeneration.

Mitochondrial function is required to maintain cytosolic ATP and GTP. We propose that a key step in axon degeneration is inactivating mitochondrial energy production. This step is blocked by

Wlds, while other degeneration events, like termination of microtubule dynamics, is not. It will be extremely interesting to determine how mitochondrial energy production is shut down in response to axon injury. The preservation of mitochondrial energy production by expression Wlds may be related to the stabilization of reactive oxygen species (ROS) levels in Wlds-expressing severed axons [33]. There is also evidence for increased motility of mitochondria after axon severing in Wlds-expressing axons

[35]. The kinesin and dynein motors that transport mitochondria are ATPases and so their ability to move mitochondria is likely to stop at the time of energy collapse, which is delayed in Wlds-expressing axons. Whatever the primary effect of Wlds in severed axons, the continued availability of cytosolic energy is likely to be important for delaying degeneration. However, the fact that some degenerative events occur independently of both energy collapse and Wlds indicate that unraveling the mechanism of action of Wlds will not unlock all the secrets of axon degeneration, or yield strategies that block all

69 aspects of degeneration. Understanding the active players that perform early steps in axon degeneration is critical for a complete picture of the process that can be used for both diagnostic and therapeutic design.

Materials and Methods

Drosophila Stocks and Genetic Backgrounds

The tester fly lines for RNAi experiments included the following: UAS-dicer2; ppk-Gal4, UAS- mCD8 GFP (for class IV neurons) and UAS-dicer2; 221-Gal4, UAS-mCD8 GFP (for class I neurons).

All the RNAi experiments included dicer2 to increase the knockdown efficiency of the neuronal RNAi

[39]. For experiments that did not involve RNAi, we used similar tester lines without UAS-dicer2.

Drosophila with UAS-Rab-GFP/YFP transgenes were crossed with 221-Gal4, UAS-mCD8-RFP and

477-Gal4, UAS-mCD8-RFP.

The following RNAi lines were obtained from the Vienna Drosophila RNAi Center (VDRC,

Vienna, Austria): UAS-rtnl2-RNAi (33320), UAS-fidgetin-RNAi (24746), UAS-Katanin 60-RNAi

(38368), UAS-katanin p60L1-RNAi (31598), UAS-spastin-RNAi(33110), UAS-stathmin-RNAi

(32370), and UAS-Klp59C-RNAi (48576), UAS-Klp59D-RNAi (100530), UAS-Klp10A-

RNAi(41534), UAS-Klp67A- RNAi(52105).

The following lines were provided by the Bloomington Drosophila Stock Center (BDSC,

Bloomington, IN): Klp59C Df(2R)Exel7177/CyO, UASp-YFP-Rab5, UASp-YFP-Rab5(S43N) [DN],

UASp-YFP-Rab5(Q88L)/TM3, Ser [CA], UAS-Rab11-GFP, UAS-GFP-LAMP [starting stock contained other transgenes that were crossed away], elav-Gal4, and UAS-mito-GFP. pUAST-Wlds #4 on III; pUAS-mCherry-zeus/CyO on II were generously provided by Marc Freeman and Clemens

70

Cabernard. Mutant and deficiency lines were rebalanced with Cyo, actinGFP. We performed standard

Drosophila genetics to combine transgenes when necessary.

Imaging and Axotomy Experiments

Imaging and injury assays were performed in ddaE and ddaC neurons in Drosophila larvae. mCD8-GFP and mCD8-RFP driven by 221-Gal4 were used to visualize the ddaE neurons while mCD8-GFP and mCD8-RFP driven by ppk-Gal4 were used for ddaC neurons. Embryos were collected overnight and then aged at 25°C for three days before imaging or injury. A whole larva was mounted on a slide with a dry agarose cushion and covered with a coverslip that was held in place by tape. Live imaging was performed on an FV1000 confocal microscope (Olympus) or an LSM510 confocal microscope (Zeiss) at a 1-second frame rate. Image J software generated maximum-intensity projections and performed image analyses (http://rsb.info.nih.gov/ij/; NIH). Overviews of neurons were maximum projection Z stacks from confocal images.

To study injury-induced axon degeneration, we used a micropoint pulsed UV laser (Andor

Technology) to sever a single axon close to the cell body. Images were acquired immediately after severing to ensure that axons were completely cut. The larva was then returned to the normal

Drosophila media to recover at 25 °C. The same larva was later remounted on the microscope to examine the severed axons. Axon beading was scored at 12h after severing in most experiments. In time course experiments, the same larva was mounted on slides every hour. In between imaging, larvae were returned to normal Drosophila media.

For FRAP analysis of different membrane markers, individual immobile punctae were bleached for 4s with a 405nm laser. Images were acquired immediately before and after imaging. Pilot experiments were performed to determine the rough turnover rate of Rab5-YFP. Most returned to the membrane by 6 minutes after bleaching, so for quantitative comparisons images were acquired at 6 minutes and 15 minutes after bleaching. The average intensity of the region of interest was analyzed

71 using ImageJ software. After subtracting background, the intensity was normalized such that intensity before bleaching equaled one and intensity after bleaching equaled zero.

Cell Culture and Transfection

HeLa cells were cultured on glass coverslips in 6-well plates with MEM medium supplemented with 10% fetal bovine serum (FBS) without antibiotics at 37°C under 95% air and 5% CO2. Transient transfections were performed using Lipofectamine LTX (Invitrogen) according to the manufacturer’s protocol. Equal amounts of Rab11a constructs and LifeAct were co-transfected one day after plating.

Venus-C2-Rab11a was a gift from James Goldenring, and pLifeAct-mTurquoise2 was obtained from

Addgene (#36201).

H2O2 and TMRM Treatment

24h after transfection, HeLa cells were washed with PBS and incubated with 20nM TMRM (Sigma) in the dark for 15 min at 37°C. Cells were then washed with PBS twice and treated with 32 mM H2O2 for

15 min at 37°C. After treatment, cells were washed twice in PBS. After washing, cells were immediately observed and photographed using an Olympus IX81 microscope equipped with

SlideBook5 software. The experiment was repeated 4 times. In each treatment, 4 images of different regions of the slide were acquired. Healthy cells with clear LifeAct signal were selected for analysis and were scored either as containing small punctae or only diffuse Venus signal.

EB1-GFP Comet Assay

To examine EB1-GFP comets after axon severing, we crossed control or RNAi flies with 221-

Gal4, UAS-EB1-GFP or 477-Gal4, UAS-EB1-GFP lines. Standard axon severing was performed on a

3-day old larva.1.5h after axon injury a 2X zoom video (one second/frame) was taken of the distal part of the transected axon. Mobile comets appearing in three consecutive frames were scored to determine whether a given axon contained dynamic microtubules.

72

Acknowledgements

We are grateful to the members of the Rolls and Santy labs for helpful discussions. We are also indebted to both the Vienna Drosophila RNAi Center and Bloomington Drosophila Stock Center for their invaluable service to researchers using Drosophila as a model organism. We would also like to acknowledge researchers who have deposited their reagents with the stock centers, as well as Marc

Freeman and Clemens Cabernard for sending us Drosophila stocks and James Goldenring for plasmids.

This work was made possible by funding from the Pew Scholars in the Biomedical Sciences (MMR) and the National Institutes of Health, grants GM085115 (to MMR) and NS066216 (to MMR) and

DK093729 (to LCS).

73

References

1. Lunn, E.R., et al., Absence of Wallerian Degeneration does not Hinder Regeneration in Peripheral Nerve. Eur J Neurosci, 1989. 1(1): p. 27‐33. 2. Coleman, M.P. and M.R. Freeman, Wallerian degeneration, wld(s), and nmnat. Annu Rev Neurosci, 2010. 33: p. 245‐67. 3. Saxena, S. and P. Caroni, Mechanisms of axon degeneration: from development to disease. Prog Neurobiol, 2007. 83(3): p. 174‐91. 4. Luo, L. and D.D. O'Leary, Axon retraction and degeneration in development and disease. Annu Rev Neurosci, 2005. 28: p. 127‐56. 5. Low, L.K. and H.J. Cheng, A little nip and tuck: axon refinement during development and axonal injury. Curr Opin Neurobiol, 2005. 15(5): p. 549‐56. 6. Raff, M.C., A.V. Whitmore, and J.T. Finn, Axonal self­destruction and neurodegeneration. Science, 2002. 296(5569): p. 868‐71. 7. Vargas, M.E. and B.A. Barres, Why is Wallerian degeneration in the CNS so slow? Annu Rev Neurosci, 2007. 30: p. 153‐79. 8. Zhai, Q., et al., Involvement of the ubiquitin­proteasome system in the early stages of wallerian degeneration. Neuron, 2003. 39(2): p. 217‐25. 9. Ehlers, M.D., Deconstructing the axon: Wallerian degeneration and the ubiquitin­ proteasome system. Trends Neurosci, 2004. 27(1): p. 3‐6. 10. Ma, M., Role of calpains in the injury­induced dysfunction and degeneration of the mammalian axon. Neurobiol Dis, 2013. 60: p. 61‐79. 11. Gardner, M.K., M. Zanic, and J. Howard, Microtubule catastrophe and rescue. Curr Opin Cell Biol, 2013. 25(1): p. 14‐22. 12. Stepanova, T., et al., Visualization of microtubule growth in cultured neurons via the use of EB3­GFP (end­binding protein 3­green fluorescent protein). J Neurosci, 2003. 23(7): p. 2655‐64. 13. Roll‐Mecak, A. and F.J. McNally, Microtubule­severing enzymes. Curr Opin Cell Biol, 2010. 22(1): p. 96‐103. 14. Su, X., R. Ohi, and D. Pellman, Move in for the kill: motile microtubule regulators. Trends Cell Biol, 2012. 22(11): p. 567‐75. 15. Ems‐McClung, S.C. and C.E. Walczak, Kinesin­13s in mitosis: Key players in the spatial and temporal organization of spindle microtubules. Semin Cell Dev Biol, 2010. 21(3): p. 276‐82. 16. Mizuno‐Yamasaki, E., F. Rivera‐Molina, and P. Novick, GTPase networks in membrane traffic. Annu Rev Biochem, 2012. 81: p. 637‐59. 17. Goldenring, J.R., et al., Enrichment of rab11, a small GTP­binding protein, in gastric parietal cells. Am J Physiol, 1994. 267(2 Pt 1): p. G187‐94. 18. Nagai, T., et al., A variant of yellow fluorescent protein with fast and efficient maturation for cell­biological applications. Nat Biotechnol, 2002. 20(1): p. 87‐90. 19. Barros, L.F., et al., Apoptotic and necrotic blebs in epithelial cells display similar neck diameters but different kinase dependency. Cell Death Differ, 2003. 10(6): p. 687‐97. 20. Mi, L., et al., Cancer preventive isothiocyanates induce selective degradation of cellular alpha­ and beta­tubulins by proteasomes. J Biol Chem, 2009. 284(25): p. 17039‐51. 21. Goedhart, J., et al., Structure­guided evolution of cyan fluorescent proteins towards a quantum yield of 93%. Nat Commun, 2012. 3: p. 751.

74

22. Scaduto, R.C., Jr. and L.W. Grotyohann, Measurement of mitochondrial membrane potential using fluorescent rhodamine derivatives. Biophys J, 1999. 76(1 Pt 1): p. 469‐77. 23. Papaconstantinou, J. and S.P. Colowick, The role of glycolysis in the growth of tumor cells. II. The effect of oxamic acid on the growth of HeLa cells in tissue culture. J Biol Chem, 1961. 236: p. 285‐8. 24. Herst, P.M., et al., Cell surface oxygen consumption by mitochondrial gene knockout cells. Biochim Biophys Acta, 2004. 1656(2‐3): p. 79‐87. 25. Tao, J. and M.M. Rolls, Dendrites have a rapid program of injury­induced degeneration that is molecularly distinct from developmental pruning. J Neurosci, 2011. 31(14): p. 5398‐405. 26. Hoopfer, E.D., et al., Wlds protection distinguishes axon degeneration following injury from naturally occurring developmental pruning. Neuron, 2006. 50(6): p. 883‐95. 27. MacDonald, J.M., et al., The Drosophila cell corpse engulfment receptor Draper mediates glial clearance of severed axons. Neuron, 2006. 50(6): p. 869‐81. 28. Bodmer, R. and Y.N. Jan, Morphological differentiation of the embryonic peripheral neurons in Drosophila. Roux's Arch Dev Biol, 1987. 196: p. 69‐77. 29. Grueber, W.B., L.Y. Jan, and Y.N. Jan, Tiling of the Drosophila epidermis by multidendritic sensory neurons. Development, 2002. 129(12): p. 2867‐78. 30. Wang, J. and Z. He, NAD and axon degeneration: from the Wlds gene to neurochemistry. Cell Adh Migr, 2009. 3(1): p. 77‐87. 31. Shen, H., K.L. Hyrc, and M.P. Goldberg, Maintaining energy homeostasis is an essential component of Wld(S)­mediated axon protection. Neurobiol Dis, 2013. 59: p. 69‐79. 32. Guo, X., et al., The GTPase dMiro is required for axonal transport of mitochondria to Drosophila synapses. Neuron, 2005. 47(3): p. 379‐93. 33. O'Donnell, K.C., M.E. Vargas, and A. Sagasti, WldS and PGC­1alpha regulate mitochondrial transport and oxidation state after axonal injury. J Neurosci, 2013. 33(37): p. 14778‐90. 34. Wang, J., et al., A local mechanism mediates NAD­dependent protection of axon degeneration. J Cell Biol, 2005. 170(3): p. 349‐55. 35. Avery, M.A., et al., Wld(S) Prevents Axon Degeneration through Increased Mitochondrial Flux and Enhanced Mitochondrial Ca(2+) Buffering. Curr Biol, 2012. 36. Zhai, R.G., et al., Drosophila NMNAT maintains neural integrity independent of its NAD synthesis activity. PLoS Biol, 2006. 4(12): p. e416. 37. Wang, J.T., Z.A. Medress, and B.A. Barres, Axon degeneration: molecular mechanisms of a self­destruction pathway. J Cell Biol, 2012. 196(1): p. 7‐18. 38. Maor‐Nof, M., et al., Axonal Pruning Is Actively Regulated by the Microtubule­Destabilizing Protein Kinesin Superfamily Protein 2A. Cell Rep, 2013. 3(4): p. 971‐7. 39. Dietzl, G., et al., A genome­wide transgenic RNAi library for conditional gene inactivation in Drosophila. Nature, 2007. 448(7150): p. 151‐6.

75

Supplementary Materials

Figure S1. Rab5-YFP becomes diffuse about 3h after axon severing in the ddaC neuron. Rab5- YFP and mCD8-RFP were expressed in class IV neurons and the ddaC axon was severed at time 0. Animals were remounted for imaging at different times after injury and the Rab5-YFP distribution was scored as diffuse or punctate in the severed axon. Numbers on graph bars indicate number of animals tested at each time.

76

Figure S2. Rab proteins exchange off membranes in minutes. Tagged proteins were expressed in ddaE neurons, in most cases together with mCD8-RFP. Individual punctae (in yellow squares) were bleached at time 0, and then images were acquired immediately after bleaching and at 6 and 15 mins.

77

Mito-GFP and Rab5CA-YFP were used as controls that are not expected to exchange off membranes. Rab5DN-YFP was entirely diffuse, and so was not bleached. Average fluorescent intensity, normalized to the level before bleaching, is shown in the graph.

Figure S3. Mitochondria and LAMP remain punctate 4h after axon severing. Axons of ddaE neurons expressing mCD8-RFP with Rab5-YFP, Lamp-GFP, or mito-GFP were severed and animals were remounted for imaging 4h later. The YFP/GFP markers were scored as punctate or diffuse in the cut off axon. Examples in A have clear puntae in the cut off axon. Quantitation is shown in B, with numbers on the bars indicating numbers of animals tested for each genotype.

78

Figure S4. Rab5 gain or loss of function does not affect degeneration timing. A hairpin RNA targeting Rab5 or a control (rtnl2) was expressed in ddaC neurons. Axons were severed at time 0 and were scored as intact/smooth 12h later (example in A, quantitation in D). Rab5DN-YFP or Rab5CA- YFP was expressed with mCD8-RFP in class I neurons. Axons of ddaE were severed and images were acquired 12h later (B and C). Axons were scored as intact or not (E). Numbers on bars indicate number of animals tested for each condition.

79

Figure S5. Microtubule regulators were tested for a role in axon degeneration. mCD8-GFP was expressed in ddaC neurons with dicer2 and RNAi hairpins targeting the indicated microtubule regulators. Axons were severed at time 0 and scored as intact (smooth) or not when animals were remounted for imaging 12h later. Numbers on the bars indicate number of animals tested for each genotype. When no bar is present, none of the axons was intact at 12h.

80

Chapter Four:

Klp59C and fidgetin control dendrite disassembly after injury and during pruning.

This chapter is part of a manuscript written in conjunction with Melissa Rolls.

Abstract

Regions of dendrites need to be cleared during developmental pruning as well as after dendrite injury.

It is not known whether developmental and injury context share a common set of clearance machinery.

Previous investigations of pruning in a model Drosophila neuron identified proteins required for pruning, but none have been shown to play a role in injury-induced degeneration in the same cell.

Because microtubule disassembly has been proposed to play a role in both pruning and injury-induced degeneration, we tested known microtubule regulators to determine whether any of them acted as universal clearance factors. We found that reduction of Klp59C delayed pruning as well as injury- induced dendrite clearance. Consistent with Klp59C acting in its role as a depolymerase, microtubule plus end dynamics continued longer in severed regions when klp59C was reduced. In addition to this general disassembly factor, we also identified fidgetin as a specific regulator of injury-induced dendrite degeneration.

We conclude that microtubule disassembly is a key step in all types of neurite clearance. We further demonstrate that both general and specific destabilizers act in different ways on microtubules during degeneration.

81

Introduction

During development, dendrites grow extensively and may need to be trimmed to generate specific mature connections [1]. A large-scale version of this pruning process occurs during metamorphosis in some insects. In this process, larval body parts are removed and adult parts are innervated [1]. Removing axon and dendrite regions is also important after neuronal damage. For example, after irreparable axonal damage, distal regions of the axon are cleared by Wallerian degeneration [2]. Previous studies have identified many of the key players in neuronal growth and regeneration; however, they have only identified a few players in the counterbalancing clearance pathways. Developmental pruning and injury-induced degeneration proceed through similar steps. For example, formation of filopodia-like structures occurs relatively early during dendrite pruning [3] and dendrite degeneration [4] in Drosophila. Membrane internalization also occurs during injury-induced degeneration and pruning [4]. Additionally, microtubule disassembly seems to be a relatively early step in dendrite pruning [3]. These early rearrangements are followed by beading and then clearance in all cases [3,4,5].

As developmental pruning and injury-induced degeneration proceed through a defined series of morphological changes, they may operate as programmed cellular pathways analogous to apoptosis.

One key aspect of apoptosis is that a core set of disassembly machinery including caspases can be activated by different upstream events and mechanisms. As injury-induced degeneration and pruning share some morphological features, these processes could use a common set of disassembly machinery.

In the past few years, a number of dendrite pruning regulators have been identified in

Drosophila. These include caspases [6,7], Sox14 and Mical [8], IK2 and katanin p60-like 1 [9] and headcase [10]. All of these factors are required for normal pruning of the sensory neuron ddaC during metamorphosis. However, when we tested whether they might be required for dendrite clearance after

82 injury in the same cell type, all proteins that we examined had no effect on injury-induced dendrite degeneration [4]. Several explanations account for this difference in requirements. Firstly, developmental pruning and injury-induced degeneration could possibly share no overlap in machinery despite the fact that both processes disassemble dendrites. Alternatively, pruning might be distinguished from injury-induced degeneration only in its initiation. Pruning requires an endogenous mechanism to initiate the degeneration process, whereas an external cut initiates the injury-induced degeneration.

Thus after initiation of degeneration, a common set of machinery could be activated to execute disassembly.

If the initiation factors constitute the only difference between pruning and injury-induced degeneration, then one might expect that any regulator identified in injury-induced degeneration would be part of the core clearance machinery, and also part of the pruning process. However, the regulators of injury-induced (Wallerian) axon degeneration are distinct from those that have emerged from pruning screens. Researchers have proposed endogenous factors involved in clearing damaged axons, including MORN4 [11], Sarm1 [12], and DLK [13]. However, no factors responsible for clearing damaged dendrites have been identified. It is not known whether injury-induced dendrite degeneration may use the same pathways as pruning.

In order to investigate whether the same or different sets of machinery are activated to execute clearance in developmental and injury contexts, we focused on the step of microtubule destabilization.

Destabilization is part of both injury-induced dendrite degeneration [5] and dendrite pruning [14] in

Drosophila, and has also been reported to be important for injury-induced degeneration and degeneration after growth factor withdrawal in mammals [15]. To determine whether a common set of microtubule destabilizers are used for dendrite clearance in the context of both development and trauma, we used a Drosophila model system.

83

Drosophila dendritic arborization (da) neurons are sensory neurons present in the body wall.

They are grouped into different classes based on dendrite branching patterns [16], and the most complex class, class IV, is responsible for nociception [17]. Dendrites from these neurons completely cover the larval body wall. During pupariation, the dendrites are pruned completely while the cell body and axons remain in place [3]. At the end of the pupal stage, dendrites regrow into the adult body wall

[18,19]. These cells are thus a useful model for large-scale developmental dendrite pruning study. We have also shown that these cells initiate a stereotyped clearance program after dendrite injury [4]. In the same cell type we can thus compare injury-induced dendrite clearance with developmental dendrite pruning.

We have taken advantage of this system to probe the role of microtubule destabilizers in developmental and injury-induced clearance programs. We have identified one destabilizing protein,

Klp59C, which plays a role in axon degeneration (unpublished paper). In this paper, we further revealed that klp59C is also involved in injury-induced dendrite degeneration and pruning. We also identified another putative severing protein, fidgetin, that plays a role only in injury-induced dendrite degeneration, not pruning. Thus our data supports the idea that there may be some common elements to all types of neurite clearance, but also that there are factors that are unique to each program, even at the level of microtubule disassembly.

Results

Identification of microtubule regulators involved in degeneration.

To identify microtubule regulators that might be involved in pruning or injury-induced degeneration, we conducted a three-pronged candidate screen. The screen was designed to identify only cell-autonomous regulators of degeneration that act in the neurons and not in the surrounding cells that

84 remove debris. Each candidate was screened for injury-induced axon degeneration, injury-induced dendrite degeneration and dendrite pruning in the class IV da neuron ddaC. Neurons were visualized with a membrane localized GFP (mCD8-GFP) expressed in class IV neurons with ppk-Gal4. Precise severing of axons or dendrites was accomplished with a pulsed UV laser as in previous studies [4,20].

Complete clearance of all dendrite remnants distal to the cut site was scored 18h after injury (Figure

1B). Because clearance of axon remnants was more variable than clearance of dendrite remnants, we scored axon beading at 12h after injury (Figure 1A), rather than complete clearance. Dendrite pruning was assayed by scoring clearance of dendrite remnants 18h after the onset of pupariation (Figure 1C).

To reduce levels of candidate microtubule regulators, a tester fly line (UAS-dicer2; ppk-Gal4,

UAS-mCD8-GFP) was crossed with lines that contained transgenes that encode large hairpin RNAs under UAS control [21]. Two candidates emerged from this screen: Klp59C and CG3326 (Figure 1D).

We have observed that a third of neurons expressing RNA hairpins targeting the klp59C had unbeaded axons 12h after severing in our previous study (Chapter 3). Surprisingly, expressing Klp59C

RNA hairpins also reduced clearance of dendrites after injury and during pruning (Figure 1D). Kinesin-

13 proteins mainly function to depolymerize microtubules from their ends [22]. In Drosophila, three kinesin-13s exist: Klp10A, Klp59C and Klp59D. Only reduction of Klp59C had a universal effect on the clearance assays.

In contrast to Klp59C, targeting CG3326 by RNAi affected only injury-induced dendrite degeneration (Figure 1D). CG3326 encodes a protein that contains an AAA ATPase domain, and is proposed to be the Drosophila fidgetin protein [23]. Fidgetin is part of a family of microtubule severing proteins that includes spastin and katanin. Like spastin and katanin, overexpression of Drosophila fidgetin completely disrupted microtubules in cultured Drosophila cells [23]. This phenotype is consistent with all three proteins having microtubule severing activity. In mitosis, fidgetin seems to have a similar role to spastin, controlling microtubule numbers and the turnover of tubulin at the ends

85 of microtubules [23]. We did not, however, find similar phenotypes for spastin in dendrite degeneration, although we have previously observed neuronal phenotypes with spastin RNAi [24]. We did not observe any phenotypes for a potential microtubule severing protein CG10793 either. Katanin p60-Like1 has been reported to be involved in dendrite pruning [9]; however, we did not find any defects in axon or dendrite degenerations in Katanin60like RNAi flies (Figure1D). Katanin60 has not yet been reported to be required for any types of degeneration. To completely rule out that katanin60 might play a role in degeneration, we used not only katanin60 RNAi but also a homozygous null mutant katnin60 75H to further decrease katnin60 level in ddaC neurons. However, in theses neurons, injury-induced axon or degeneration and dendrite pruning occurred at normal timing as controls

(Figure 1D).

86

Figure 1. Identification of microtubule regulators required for clearing axons and dendrites. Class IV ddaC neurons were labeled with mCD8-GFP under control of ppk-Gal4. A. The assay for injury-induced axon degeneration is shown. Larvae were mounted for imaging and the axon was severed at 0h with a pulsed UV laser. After injury animals were returned to food. They were remounted for imaging at 12h. Axons were scored as beaded (normal degeneration) or continuous. B. The assay for dendrite degeneration is shown. A ddaC dendrite was severed near its base with a pulsed UV laser at 0h. The neuron was scored for complete clearance of the dendrite 18h later. C. Dendrite pruning was assayed 18h after puparium formation (APF). An example image from a pupa is shown. Two adjacent cell bodies can be seen (arrows). Complete clearance of dendrites was scored as normal degeneration. D. Tester animals (UAS-dicer2; ppk-Gal4, UAS-mCD8-GFP) were crossed with UAS-RNAi lines to target the microtubule regulators indicated. We typically use an RNAi line targeting rtnl2 as a control as it has never generated a phenotype in any assay we have performed. Kat60-L1 was previously shown to play a role in dendrite pruning [9]. Klp59C RNAi resulted in abnormal axon degeneration and pruning, and fidgetin RNAi resulted in abnormal dendrite degeneration. Significance was calculated with a Fisher’s exact test, starred columns are all at least p<0.05 compared to the appropriate rtnl2 control. Numbers on each column are the number of animals tested for that condition. Additional screen results are shown in Figure S1.

87

In addition to microtubule regulators, we also screened some additional candidates (Figure S1).

These included proteins suggested to regulate kinesin-13s in mitosis [22,25], as well as proteins that have been shown to play a role in dendrite pruning [6,9]. None of these slowed axon or dendrite clearance after injury, although they had the expected effects on pruning (Figure S1). Thus our screen identified klp59C’s new role in injury-induced dendrite degeneration and dendrite pruning and fidgetin as a new player in injury-induced dendrite degeneration. We investigated each of these further to determine their specificity for different types of degeneration.

Fidgetin functions specifically in injury-induced dendrite degeneration.

In the candidate screen, fidgetin RNAi resulted in a small, but significant reduction in dendrite clearance 18h after severing. To probe this phenotype further, we took two approaches. First, we investigated dendrite degeneration at an earlier time point, and second, we used mutants in addition to

RNAi. We determined that assaying continuity of the severed dendrite at 9h after injury provided a more sensitive assay for dendrite degeneration (Figure 2). This time point was used in the rest of the experiments. Three different Deficiency (Df) lines that remove the fidgetin gene (Df (2L) Exel8008,

Df(2L)BSC163, and Df(2L)BSC164), and one P element insertion allele (CG3326SH1400) were used.

All trans-heterozygous combinations of the SH1400 insertional allele and Df lines, or homozygous SH1400 animals, had more dendrites that were continuous 9h after severing than controls

(Figure 2). The phenotype of SH1400/Df was not more severe than SH1400 homozygotes, indicating that SH1400 is a relatively strong loss-of-function allele. Nor were any of the mutant combinations significantly different that fidgetin RNAi, indicating that the RNAi also induces strong loss-of-function for fidgetin. All mutant combinations were also tested for dendrite pruning, but as in the initial screen, pruning was not affected by loss of fidgetin (Figure 2). We conclude that fidgetin is required for normal

88 timing of dendrite beading after injury, but not for dendrite clearance during pruning, and is thus a specific regulator of dendrite clearance after injury.

89

90

Figure 2. Fidgetin is specifically required for timely dendrite degeneration. RNAi and different mutant alleles of fidgetin were used to reduce fidgetin protein levels in neurons (RNAi) or globally (mutants). Axon degeneration and pruning were assayed as in the initial screen. Dendrite degeneration was scored by assaying beading at 9h after cutting. For all loss of function conditions dendrite degeneration was reduced, but axon degeneration and pruning were similar to controls. Numbers on each bar indicate the number of animals assayed. Significance was calculated with a Fisher’s exact test.

Klp59C function is required for timely dendrite degeneration, as well as dendrite pruning.

Our initial candidate screen indicated a role for Klp59C in axon degeneration and dendrite pruning, but not dendrite degeneration. However, our additional analysis of the fidgetin phenotype suggested that assaying dendrite clearance 18h after severing might be so stringent as to mask phenotypes. We therefore re-tested Klp59C RNAi at 9h after dendrite severing, and found that it reduced dendrite-beading 9h after severing (Figure3A). Thus Klp59C is required for all types of clearance in the ddaC neuron (Figure 3).

We also reexamined the pruning data to determine at which step in pruning Klp59C functions.

During dendrite pruning of ddaC, dendrites are first clipped from the cell body, and then the detached

91 pieces bead and are cleared. Some proteins seem to act specifically at the early clipping step [8,9].

Reduction of Klp59C seemed to affect both steps (Figure 3B).

Figure 3. Klp59C is required for normal dendrite degeneration and dendrite pruning. A. Dendrites of Klp59C or control RNAi neurons were severed, and dendrites were scored as intact or not (ie beaded or totally cleared) at 9h after severing. B. Clearance of dendrites during metamorphosis was assayed at 18h after puparium formation (APF) in control RNAi or Klp59C RNAi neurons. At 18h dendrites were scored as present and either attached the cell body, detached, or completely cleared.

92

Klp59C does not delay degeneration by activating known stabilization pathways.

As Klp59C globally delayed all types of degeneration in ddaC neurons, we considered the possibility that reduction of Klp59C may activate a stress response pathway in the neurons. We previously found that RNAi targeting a kinesin, -104, or expression of poly-Q proteins globally delays degeneration after injury by activating a stress pathway that involves JNK signaling [26]. The protein nmnat has also been identified as a general neuroprotective stress response protein that can delay degeneration [27,28,29]. We therefore performed double RNAi experiments to test whether the degeneration delay that we observed with Klp59C RNAi required either JNK (called bsk in Drosophila) or Nmnat. Rtnl2 RNAi was used as a control to pair with Klp59C. All three pairs of hairpins had similar delays in degeneration (Figure 4) that were in turn similar to reduction of Klp59C alone. Thus we did not find support for the idea that loss of Klp59C might delay degeneration indirectly, by activating a JNK- or nmnat- dependent stress response pathway. Instead, we considered that Klp59C might influence degeneration through its role as a microtubule regulator.

93

Figure 4. Delayed degeneration in Klp59C knockdown neurons does not rely on activation of known stabilization pathways. ddaC neurons expressing pairs of hairpin RNAs as well as mCD8-GFP

94 and dicer2 were subjected to axon severing, dendrite severing and dendrite pruning assays. Continuity of processes was assayed for axons and dendrites 12h and 9h after severing, respectively, and presence of dendrites was assayed 18h APF. Numbers on the graph bars indicate number of animals tested for each condition.

Increased microtubule stability before injury is unlikely to account for delayed degeneration when

Klp59C or fidgetin is reduced.

Both Klp59C and fidgetin were originally selected as candidates because of their roles as microtubule destabilizers. We therefore tested the hypotheses that loss of either Klp59C or fidgetin might stabilize microtubules before injury, and that the more stable microtubules might resist disassembly after injury and slow degeneration.

To determine whether microtubules were more stable in neurons with reduced levels of Klp59C or fidgetin, we immunostained larval body walls for acetylated tubulin, a modified form of tubulin associated with microtubule stability [30]. We expressed RNAi and GFP in class IV neurons. Other neurons, including the easily identifiable ddaE neuron, had normal levels of Klp59C and fidgetin in this experiment. We therefore calculated the ratio of acetylated tubulin fluorescence in ddaC and ddaE to normalize for antibody penetration.

Reduction of fidgetin did not result in an increased amount of acetylated tubulin in the ddaC cell body or dendrites compared to ddaE. However, reduction of Klp59C did lead to a relative increase in acetylated tubulin in the ddaC cell body and dendrites (Figure 5A and B). To test whether this microtubule stabilization might be related to the delay in degeneration, we expressed a microtubule- associated protein, futsch-N (an N-terminal region of the MAP1-related protein futsch [31]), as overexpression of MAPs tends to stabilize microtubules. We found that futsch-N overexpression stabilized microtubules 1.4-fold in ddaC dendrites (Figure 5A and B), and therefore tested whether it

95 would also delay injury-induced or developmental degeneration. However, dendrite pruning and axon and dendrite degeneration after injury was unaffected (Figure 5C).

We conclude that it is unlikely that loss of fidgetin acts to delay dendrite degeneration by stabilizing dendritic microtubules before injury, as we did not see an effect of fidgetin RNAi on microtubule acetylation. It is also unlikely that Klp59C reduction acts through this mechanism.

Although lowering levels of Klp59C did increase microtubule stability, this is unlikely to be sufficient to delay degeneration as overexpression of futsch-N stabilized microtubules without delaying injury- induced degeneration.

96

Figure 5. Microtubule stabilization before injury does not account for delayed dendrite degeneration in neurons with reduced Klp59C or fidgetin. Larvae expressing mCD8-GFP and dicer2 under control of the ppk-Gal4, which is specific to class IV neurons like ddaC, were crossed to control (rtnl2) RNAi, Klp59C RNAi, fidgetin RNAi or futsch-N overexpression (OE) lines. Larval progeny were fixed and stained for acetylated tubulin. A. Example images are shown. The top row shows the tubulin channel alone, and the bottom row also shows GFP, which marks the ddaC neurons. Cell bodies of ddaE and ddaC neurons are indicated with arrows. B. The intensity of acetylated tubulin staining was measured in cell bodies and dendrites of ddaE and ddaC neurons. The ddaE neuron is a control that does not express the RNAi hairpin or futsch-N and was used to normalize the ddaC staining. Ratios of fluorescence intensity in the ddaC to ddaE neuron are shown in the graph. A Student’s t-test was used to compare dendrite intensities in control and Klp59C RNAi and futsch-N OE animals. C. Degeneration assays were performed in ddaC neurons that overexpressed futsch-N. Control

97 numbers are shown for comparison. In B and C the numbers on the bars indicated numbers of animals tested for each condition. In B the error bars show the standard deviation.

Dendrite structure is normal before injury in neurons with reduced fidgetin.

Microtubule severing proteins have been reported to be involved in neuronal development during embryogenesis and mature neuron maintenance during adulthood in both Drosophila and mammalian models[32]. Knockdown of kataninp60-like1 causes dendrite branch retraction during later larval development in ddaC neurons[33]. To rule out the possibility that the delay phenotype is caused by unhealthy neurons in which active execution programs are already compromised, we carefully quantified and compared the total dendrite branch numbers in fidgetin RNAi and rtnl2 RNAi ddac neurons of three-day old larvae. The quantification results showed no significant difference of total branch number between figetin RNAi neurons and the rtnl2 RNAi controls (Figure 6). Thus we concluded that loss of fidgetin does not lead to any dendrite developmental defects and ruled out the possibility that fidgetin RNAi delays dendrite degeneration through compromising the normal function of a dendrite before injury.

98

Figure 6. Dendrite complexity is normal in neurons with reduced fidgetin. Neuron morphology was examined by expressing mCD8-GFP with ppk-Gal4. All the major and minor branches are counted in individual ddaC neurons, and the results were averaged. Error bars represent standard deviations. The significance is calculated by a t-test. Images were taken of each part of the neuron and combined using ImageJ software and so the background may not look even across the figures.

Microtubule behaviors are normal in uninjured fidgetin knockdown neurons.

We further wanted to test if reduced figetin would affect other aspects of microtubule behavior, such as EB1 comet dynamics, moving speed and polarity. We previously found that globally increased microtubule dynamics have a dendrite protection effect after injury[26]. To rule out the possibility that

99 the loss of fidgetin triggers a global increase in microtubule dynamics, which may cause the delay phenotype we observed, we performed a series of detailed microtubule behavior assays. To determine if the loss of fidgetin would result in an increase in microtubule dynamics, we counted the number of

EB1 comets passing through a 10 um length in 200 frames. We found no significant change in the EB1 comet number (Figure 7). Kymographs also showed that EB1 comets move at a similar speed in fidgetin RNAi and control dendrites (Figure 7). Finally, we examined microtubule polarity in the fidgetin RNAi dendrites and found that 90% of the EB1 comets were moving toward the soma, which is very similar to the controls (Figure 7). In sum, loss of fidgetin did not affect normal microtubule behaviors in dendrites. This further provides evidence that fidgetin does not function in uninjured neurons but is active after injury.

100

101

Figure 7. Loss of fidgetin does not affect microtubule behaviors in uninjured classIV ddaC neurons. EB1-GFP driven by 477-Gal4 was used to monitor the dynamics of microtubules. Fidgetin RNAi was expressed under the same driver. The total number of EB1 comets passing through a 10 um length in 200 frames was collected to indicate the dynamics of the growing microtubule. The arrows indicate an EB1-GFP comet moving in a non-typical direction. Comet speed was calculated based on kymographs, where EB1 travel distance was divided by the travel time. For these two assays, n indicates the number comets and p-values were calculated with a Fisher’s exact test. Microtubule polarity was quantified by the ratio of EB1 comets moving away from the soma over the number of total EB1 comets. For this assay, n indicates the number of neurons analyzed and a t-test was used to calculated p value.

Reduction of Klp59C but not fidgetin delays loss of dynamic microtubules in dendrite after injury.

If microtubule stabilization before injury does not account for the degeneration phenotype in

Klp59C and fidgetin RNAi neurons, perhaps these proteins act to disassemble microtubules after neuronal injury. To test this idea, we examined the behavior of microtubules in severed dendrites.

Although neuronal microtubules tend to be long and stable, they maintain dynamic plus ends that constantly switch between phases of tubulin dimer addition and loss. Dynamic plus ends can be visualized by expressing GFP-tagged plus-end binding proteins like EB1. In mammalian and

Drosophila, neurons EB1-GFP comets are seen throughout axons and dendrites [34,35]. In control neurons, EB1-GFP comets disappear from almost all severed dendrites 5h after injury (Figure 8).

Similarly, in fidgetin RNAi neurons, EB1-comets stop shortly after injury (Figure 6). However, at the same time points, almost one third of neurons with reduced Klp59C still had dynamic microtubules

(Figure 8 and Movie 1). Thus microtubule plus end growth continues longer after injury in neurons with lower levels of Klp59C, but not fidgetin.

102

Figure 8. Microtubule dynamics persist longer in degenerating dendrites when Klp59C, but not fidgetin, levels are reduced. A. Example images of EB1-GFP in ddaC dendrites 5h after severing. Arrowheads indicate cut sites near cell bodies. Arrows point to EB1-GFP comets. B. The number of dendrites in which moving EB1-GFP comets were seen 5h after severing was counted. Numbers on the bars indicate number of animals tested. Statistical significance was calculated with a Fisher’s exact test.

Targeting Klp59C and fidgetin together exacerbates delayed dendrite clearance.

Both Klp59C and fidgetin were required for timely dendrite clearance after injury, but only

Klp59C affected termination of microtubule dynamics. Therefore, we considered that fidgetin and

Klp59C might act in different ways to promote dendrite disassembly. To test this idea, we compared the time courses of dendrite clearance in double and single RNAi neurons.

Based on comparisons of the fidgetin RNAi, the SH1400 P element insertion homozygotes and

SH1400/Df phenotypes, fidgetin RNAi seems to result in a substantial loss of function (Figure 2). The

Klp59C RNAi also seems to produce a significant loss of function because its phenotype is approximately twice as severe as that of heterozygous Df (Figure 3A). Thus RNAi of fidgetin or

Klp59C should result in a strong inhibition of their pathways. In this case, if the two proteins act in the

103 same pathway, double RNAi should not result in a stronger phenotype. If, however, they act in separate pathways then double RNAi could increase the severity of the clearance defect.

When we analyzed dendrite clearance 9h and 12h after injury, we found that double RNAi of Klp59C and fidgetin resulted in a stronger defect than either RNAi alone (Figure 9). This result supports the hypothesis that these two proteins act in different pathways during dendrite degeneration.

104

105

Figure 9. Targeting Klp59C and fidgetin together by RNAi slows degeneration more than targeting either alone. An example double RNAi dendrite degeneration experiment is shown in A. Cells with either single or double RNAi expression were analyzed for dendrite degeneration 9 and 12h after injury and the results are plotted in C. The difference between either Klp59C or fidgetin single RNAi and the double RNAi was statistically significant (p<0.05) at both 9 and 12 hours using a Fisher’s exact test. C. A summary diagram of microtubule regulators required for different types of clearance is shown.

Discussion

In order to understand how axons and dendrites fragment after injury and during developmental clearance, we focused on microtubule breakdown. Specifically, we wondered whether injury and developmental clearance might share a common set of disassembly machinery. In our previous study, we identified one regulator required in axon degeneration by screening for regulators of microtubule breakdown. In this study, we found that this regulator is also involved in both injury and developmental dendrite degeneration. We also identified a specific regulator of injury-induced dendrite degeneration

(Figure 9C). We therefore propose that there are compartment-specific, as well as context-specific, regulators of neurite clearance. In addition to very specific regulators, there may also be a core set of machinery, which includes Klp59C, used by many different kinds of clearance.

Why are multiple microtubule disassembly proteins required?

Injury-induced dendrite clearance and dendrite pruning both require the microtubule depolymerase Klp59C. However, each of the two processes requires a different microtubule severing protein: kat-60L1 in the case of pruning and fidgetin for injury. We hypothesize a simple explanation: the severing protein generates microtubule plus ends for the depolymerase to disassemble. This model is consistent with all the data, although still speculative. Specifically, this model predicts that the most efficient microtubule disassembly would occur when both proteins are active. However, reducing either protein alone would still affect microtubules. The comparison of dendrite degeneration timing in single and double RNAi neurons is consistent with this conclusion (Figure 9B). This model also predicts that

106 only loss of Klp59C would affect the length of time that dynamic plus ends persist after injury (Figure

8). This model also predicts that a severing protein is necessary in axons for two reasons. First, axonal microtubules tend to be more stable and longer than dendritic ones [36]. Second, axons have fewer free ends in the absence of a severing protein. Our data shows that neither deduction of spastin nor

Kat60 delay injury-induced axon degeneration (Figure 1). It is possible that some other unidentified protein may serve this function in axons.

Are fidgetin and Klp59C active after injury?

Loss of microtubule destabilizers could delay degeneration in three fundamentally different ways. First, the delay could result from an overall stabilization of neuronal microtubules before injury or pruning. Second, it could be part of a stabilizing stress response triggered by mis-regulated microtubules. Third, after injury, the delay could be caused by proteins losing their active roles in neurite disassembly itself. We probed these different possibilities in several ways. Initially, we tested whether two known stress-response pathways, including one triggered by loss of a different kinesin, were responsible for the degeneration delay caused by Klp59C. We found they were not (Figure 4).

Because both the JNK-dependent pathway [26], and nmnat [27,28,29] seem to confer a broad-spectrum stabilization, while fidgetin loss affects only injury-induced dendrite degeneration, we did not test for the involvement of stress response pathways in the fidgetin phenotype, Next, to test whether the loss of fidgetin or Klp59C slowed degeneration by stabilizing microtubules before injury, we took two approaches. We used overexpression of a MAP to stabilize microtubules and tested whether this would delay degeneration; it did not (Figure 5). We also measured stable, or acetylated, microtubules in neurons with reduced Klp59C and fidgetin (Figure 5). Since fidgetin did not seem to increase microtubule stability in dendrites, this experiment suggests the relevant activity of fidgetin occurs after injury. For Klp59C the interpretation is a little more ambiguous as its reduction did increase microtubule stability in uninjured neurons. However, the fact that plus end dynamics continued longer

107 after injury when Klp59C was reduced (Figure 8) argues that Klp59C has a more specific role in controlling microtubule plus ends after injury. Thus, in total, the data is consistent with both fidgetin and Klp59C functioning to disassemble microtubules after injury or during pruning.

If both proteins act after injury, then future studies could investigate how their activities are upregulated in some way in response to the trauma. The MAP kinase kinase kinase DLK has been proposed to act upstream of a kinesin-13 in C. elegans to allow axon regeneration after injury [37]. It seems unlikely that DLK is the relevant regulator for degeneration. However, as reduction of DLK

(called wnd in flies) does not delay degeneration or pruning (Figure S1), it is proposed to be a negative, not positive, regulator of kinesin-13 in C. elegans [36]. Other kinases regulate kinesin-13s in mitosis

[22,25]. However, these kinases did not have any effect on dendrite degeneration or pruning (Figure

S1). Therefore, It is possible that different mechanisms regulate the activity of kinesin-13s in different contexts. Little is known about how severing proteins are regulated in time and space, although microtubule modification state seems to regulate the access of some severing proteins [38]. Our findings can help future studies determine whether the same upstream signal controls the activity of both the severing proteins and Klp59C, or whether several signals generated from the cell body independently control the two types of microtubule disassembly factors. If independent regulatory mechanisms are used, the regulators could serve as a quality control mechanism to ensure that fast microtubule destabilization occurs only when a portion of the cell is severed from the cell body. The fact that different severing proteins are used for dendrite pruning and dendrite degeneration after injury suggests that different sets of signals may be elicited in different scenarios. Perhaps Klp59C is activated by a signal common to all types of degeneration, while severing proteins are activated by different signals.

Taken together, we identified two microtuble regulators: Klp59C, a general regulator, required in both injury-induced dendrite degeneration and dendrite pruning, while fidgetin is specific to injury-

108 induced dendrite degeneration. Our study also revealed that Klp59C, but not fidgetin, terminates microtubule plus end growth in the injured regions of neurons. Each protein targets different parts of microtubule to efficiently disassemble microtubules during neuron degeneration.

Methods and Materials

Drosophila stocks

The tester fly lines for RNAi experiments included the following: UAS-dicer2; ppk-Gal4, UAS- mCD8-GFP and UAS-dicer2/CyO; 477-Gal4, UAS-EB1-GFP/TM6. All the RNAi experiments included dicer2 to increase the knockdown efficiency of the neuronal RNAi [21]. Overexpression studies and Klp59C heterozygous deficiency studies used similar tester lines without UAS-dicer2. For experiments in which both copies of fidgetin were mutant, or one copy was mutant with a deficiency on another copy, one of the mutant alleles was combined with ppk-Gal4, UAS-mCD8 GFP first.

The following RNAi lines were obtained from the Vienna Drosophila RNAi center (VDRC,

Vienna, Austria): UAS-rtnl2-RNAi (33320), UAS-fidgetin-RNAi (24746), UAS-katanin 60- RNAi

(38368), UAS-katanin-p60L1-RNAi (31598), UAS-spastin-RNAi (33110), UAS-stathmin-RNAi

(32370), CG10793 (31351), UAS-Klp59C-RNAi (48576), UAS-Klp59D-RNAi (100530), UAS-

Klp10A-RNAi (41534), UAS-Klp67A- RNAi (52105), UAS-bsk RNAi (34138), and UAS-nmnat-

RNAi (32255).

The following mutant and deficiency lines were provided by the Bloomington Drosophila Stock

Center: Klp59C Df(2R)Exel7177/CyO, fidgetinsh1400/CyO, fidgetin Df(2L)Exel8008/CyO, fidgetin

Df(2L)BSC164/CyO. We rebalanced all the mutant and deficiency lines with CyO, actinGFP in our lab.

For heterozygous Klp59C deficiency genotypes, Klp59C Df/Cyo, actin GFP was crossed to the tester

109 line without dicer2. For homozygous fidgetin mutants or fidgetin mutant over deficiency genotypes, lines were generated that contained fidgetinsh1400 mutant on the second chromosome balanced with

CyO, actinGFP, and ppk-Gal4, mCD8 GFP transgenes on the third chromosome balanced with TM6

(Tb). These lines were crossed to fidgetin mutant or deficiency lines balance with CyO, actinGFP.

Then, non-Tb progeny without actinGFP were analyzed. Katanin 60 75H mutant was a gift from Dr.

Nina Tang Sherwood (Duke University Department of Biology). We combined the mutant with

477gal4, EB1 GFP and cross it to the mutant. The homozygous Katanin60 mutant offspring were analyzed.

We also generated double transgenic fly lines for this study, including: UAS-nmnat-RNAi,

UAS-Klp59C RNAi; UAS-bsk RNAi, UAS-Klp59C RNAi; and UAS-fidgetin RNAi, UAS klp59C

RNAi. Additionally, UAS-rtnl2-RNAi; UAS-Klp59C-RNAi was generated as the control for all the experiments using double transgenic fly lines.

Axon and dendrite injury experiments

All the severing assays in this paper were performed in ddaC neurons in Drosophila larvae. mCD8-GFP driven by ppk-Gal4 was used to visualize the ddaC neurons. Embryos were collected overnight and then aged at 25°C for three days before imaging. A whole larva was mounted on a slide with a dry agarose cushion and covered with a coverslip that was held in place with tape. Live imaging was performed on an FV1000 confocal microscope (Olympus) or an LSM510 confocal microscope

(Zeiss). Image J software was used to generate maximum-intensity projections and perform image analysis (http://rsb.info.nih.gov/ij/; NIH).

To injure neurons, we used a pulsed UV laser (Andor Technology) to sever a single axon or dendrite close to the cell body. Images were acquired immediately after severing to make sure that axons or dendrites were completely cut. The larvae were then returned to normal Drosophila media to recover at 25 °C. The same larvae were later remounted on the microscope to examine the severed

110 axons or dendrites. Dendrites that remained continuous without signs of fragmentation were scored at

18h or 9h after injury. Axons without beading 12h after severing were used as the readout of axon integrity.

Dendrite pruning

Pruning was studied in ddaC neurons. White pupae were collected and maintained at 25°C. 18h after pupae formation, the pupal cases were first removed (as described by Williams et al., 2006) and then imaged on a confocal microscope. The presence of dendrites was scored as a pruning defect.

Microtubule dynamics after axon or dendrite injury

To examine the persistence of EB1-GFP comets after axon or dendrite severing, flies of different genotypes were crossed to the 477-Gal4, UAS-EB1-GFP/TM6 line. Standard axon or dendrite severing was performed on a 3-day old larva. After 1.5h for axon injury or 5h for dendrite injury, a time series (two seconds/frame) was taken of the distal part of the transected axon or dendrite. Mobile comets appearing in three consecutive frames were counted.

Immunostaining

For immunostaining of ddaC neurons, flies with different genotypes were crossed to ppk-gal4 mCD8 GFP. Third- instar larvae were dissected in Schneider’s medium and fixed in 4% paraformaldehyde for thirty minutes. After fixation, the larvae were washed four times in Blocking solution (PBS, 1% BSA, 0.2% Triton X-100, 10 mM glycine) and each wash lasted at least ten minutes.

Larvae were then incubated overnight at 4°C with primary antibodies (Sigma 042M4761) in the blocking solution. The next day, they were washed in the blocking solution several times for a few hours to remove the primary antibody. Then a Rhodamine-Red-X coupled secondary antibody (Jackson

ImmunoResaerch) was applied to the larval preparations for two hours at room temperature. The secondary antibody was removed with final four washes with the blocking solution, and larval

111 preparations were stored at 4°C in 85% glycerol/50 mM Tris, pH 8 until imaging. Images were obtained on a Zeiss LSM 510 confocal microscope.

Abbreviations

GFP: green fluorescence protein; Df: deficiency; EB: end-binding; FRAP:Fluorescence recovery after photobleaching; MAP: microtubule associated protein; DLK: dual leucine zipper kinase

Acknowledgements

We are very grateful to both the Bloomington Drosophila Stock Center and Vienna Drosophila RNAi

Center for maintaining and providing valuable reagents. We are also grateful to labs that shared resources with us, including Melissa Long, for helpful discussions, and to colleagues including

Avraham Yaron and Oren Schuldiner for their thoughts. Funding was provided by the NIH,

R01GM085115 and the Pew Charitable Trusts; MMR was a Pew Scholar in the Biomedical Sciences.

112

References

1. Luo, L. and D.D. O'Leary, Axon retraction and degeneration in development and disease. Annu Rev Neurosci, 2005. 28: p. 127‐56. 2. Coleman, M.P. and M.R. Freeman, Wallerian degeneration, wld(s), and nmnat. Annu Rev Neurosci, 2010. 33: p. 245‐67. 3. Williams, D.W. and J.W. Truman, Cellular mechanisms of dendrite pruning in Drosophila: insights from in vivo time­lapse of remodeling dendritic arborizing sensory neurons. Development, 2005. 132(16): p. 3631‐42. 4. Tao, J. and M.M. Rolls, Dendrites have a rapid program of injury­induced degeneration that is molecularly distinct from developmental pruning. J Neurosci, 2011. 31(14): p. 5398‐405. 5. MacDonald, J.M., et al., The Drosophila cell corpse engulfment receptor Draper mediates glial clearance of severed axons. Neuron, 2006. 50(6): p. 869‐81. 6. Williams, D.W., et al., Local caspase activity directs engulfment of dendrites during pruning. Nat Neurosci, 2006. 9(10): p. 1234‐6. 7. Kuo, C.T., et al., Identification of E2/E3 ubiquitinating enzymes and caspase activity regulating Drosophila sensory neuron dendrite pruning. Neuron, 2006. 51(3): p. 283‐90. 8. Kirilly, D., et al., A genetic pathway composed of Sox14 and Mical governs severing of dendrites during pruning. Nat Neurosci, 2009. 12(12): p. 1497‐505. 9. Lee, H.H., L.Y. Jan, and Y.N. Jan, Drosophila IKK­related kinase Ik2 and Katanin p60­like 1 regulate dendrite pruning of sensory neuron during metamorphosis. Proc Natl Acad Sci U S A, 2009. 106(15): p. 6363‐8. 10. Loncle, N. and D.W. Williams, An interaction screen identifies headcase as a regulator of large­scale pruning. J Neurosci, 2012. 32(48): p. 17086‐96. 11. Bhattacharya, M.R., et al., A model of toxic neuropathy in Drosophila reveals a role for MORN4 in promoting axonal degeneration. J Neurosci, 2012. 32(15): p. 5054‐61. 12. Osterloh, J.M., et al., dSarm/Sarm1 is required for activation of an injury­induced axon death pathway. Science, 2012. 337(6093): p. 481‐4. 13. Miller, B.R., et al., A dual leucine kinase­dependent axon self­destruction program promotes Wallerian degeneration. Nat Neurosci, 2009. 12(4): p. 387‐9. 14. Williams, D.W. and J.W. Truman, Remodeling dendrites during insect metamorphosis. J Neurobiol, 2005. 64(1): p. 24‐33. 15. Maor‐Nof, M., et al., Axonal pruning is actively regulated by the microtubule­destabilizing protein kinesin superfamily protein 2A. Cell Rep, 2013. 3(4): p. 971‐7. 16. Grueber, W.B., L.Y. Jan, and Y.N. Jan, Tiling of the Drosophila epidermis by multidendritic sensory neurons. Development, 2002. 129(12): p. 2867‐78. 17. Hwang, R.Y., et al., Nociceptive neurons protect Drosophila larvae from parasitoid wasps. Curr Biol, 2007. 17(24): p. 2105‐16. 18. Shimono, K., et al., Multidendritic sensory neurons in the adult Drosophila abdomen: origins, dendritic morphology, and segment­ and age­dependent programmed cell death. Neural Dev, 2009. 4: p. 37. 19. Yasunaga, K., et al., Dendrite reshaping of adult Drosophila sensory neurons requires matrix metalloproteinase­mediated modification of the basement membranes. Dev Cell, 2010. 18(4): p. 621‐32.

113

20. Stone, M.C., et al., Global up­regulation of microtubule dynamics and polarity reversal during regeneration of an axon from a dendrite. Mol Biol Cell, 2010. 21(5): p. 767‐77. 21. Dietzl, G., et al., A genome­wide transgenic RNAi library for conditional gene inactivation in Drosophila. Nature, 2007. 448(7150): p. 151‐6. 22. Ems‐McClung, S.C. and C.E. Walczak, Kinesin­13s in mitosis: Key players in the spatial and temporal organization of spindle microtubules. Semin Cell Dev Biol, 2010. 21(3): p. 276‐82. 23. Zhang, D., et al., Three microtubule severing enzymes contribute to the "Pacman­flux" machinery that moves chromosomes. J Cell Biol, 2007. 177(2): p. 231‐42. 24. Stone, M.C., et al., Normal spastin gene dosage is specifically required for axon regeneration. Cell Rep, 2012. 2(5): p. 1340‐50. 25. Knowlton, A.L., et al., ICIS and Aurora B coregulate the microtubule depolymerase Kif2a. Curr Biol, 2009. 19(9): p. 758‐63. 26. Chen, L., et al., Axon injury and stress trigger a microtubule­based neuroprotective pathway. Proc Natl Acad Sci U S A, 2012. 109(29): p. 11842‐7. 27. Ali, Y.O., et al., Nicotinamide mononucleotide adenylyltransferase is a stress response protein regulated by the heat shock factor/hypoxia­inducible factor 1alpha pathway. J Biol Chem, 2011. 286(21): p. 19089‐99. 28. Ali, Y.O., K. Ruan, and R.G. Zhai, NMNAT suppresses tau­induced neurodegeneration by promoting clearance of hyperphosphorylated tau oligomers in a Drosophila model of tauopathy. Hum Mol Genet, 2012. 21(2): p. 237‐50. 29. Xiong, X., et al., The Highwire ubiquitin ligase promotes axonal degeneration by tuning levels of Nmnat protein. PLoS Biol, 2012. 10(12): p. e1001440. 30. Hammond, J.W., D. Cai, and K.J. Verhey, Tubulin modifications and their cellular functions. Curr Opin Cell Biol, 2008. 20(1): p. 71‐6. 31. Roos, J., et al., Drosophila Futsch regulates synaptic microtubule organization and is necessary for synaptic growth. Neuron, 2000. 26(2): p. 371‐82. 32. Sherwood, N.T., et al., Drosophila spastin regulates synaptic microtubule networks and is required for normal motor function. PLoS Biol, 2004. 2(12): p. e429. 33. Stewart, A., et al., Katanin p60­like1 promotes microtubule growth and terminal dendrite stability in the larval class IV sensory neurons of Drosophila. J Neurosci, 2012. 32(34): p. 11631‐42. 34. Stepanova, T., et al., Visualization of microtubule growth in cultured neurons via the use of EB3­GFP (end­binding protein 3­green fluorescent protein). J Neurosci, 2003. 23(7): p. 2655‐64. 35. Stone, M.C., F. Roegiers, and M.M. Rolls, Microtubules have opposite orientation in axons and dendrites of Drosophila neurons. Mol Biol Cell, 2008. 19(10): p. 4122‐9. 36. Kollins, K.M., et al., Dendrites differ from axons in patterns of microtubule stability and polymerization during development. Neural Dev, 2009. 4: p. 26. 37. Ghosh‐Roy, A., et al., Kinesin­13 and tubulin posttranslational modifications regulate microtubule growth in axon regeneration. Dev Cell, 2012. 23(4): p. 716‐28. 38. Sudo, H. and P.W. Baas, Acetylation of microtubules influences their sensitivity to severing by katanin in neurons and fibroblasts. J Neurosci, 2010. 30(21): p. 7215‐26.

114

Supplementary Materials

Figure S1. Extended candidate screen confirms that inhibition of the ubiquitin-proteasome system (UPS) and overexpression of Wld(s) delay all types of degeneration, while other previously identified factors delay only pruning. Assays were performed as in Figure 1. The Wld(s), UBP2, bskDN, p35, DIAP, and IK2DN are overexpression transgenes. The rest are UAS-RNAi transgenes to target the gene indicated. The numbers above the condition indicate the number of animals tested for axon degeneration (top number), dendrite degeneration (middle number) and pruning (lower number).

115

Figure S2. Comparison of Klp59C reduction in combination with protective pathway knockdown to additional controls. This graph shows the data from additional genotypes analyzed as in Figure 4.

Movie 1. Microtubule dynamics continues longer after dendrite severing when Klp59C is reduced. EB1-GFP was expressed in the ddaC neuron with 477-Gal4. Dendrites were severed at 0h, and movies were acquired 4h (control) or 5h (Klp59C RNAi) after severing. The control movie is on the left and the Klp59C RNAi movie is on the right. EB1-GFP comets (some marked with stars) are visible only in the Klp59C RNAi movie. In both cases the images are shown with dorsal at the top, and the cell body at the bottom of the frame.

116

Materials and methods for the Supplementary data

Fly stocks

Additional lines were used in supplementary experiments: pUAST-Wld(s) [5], UAS-UBP2 [38],

UAS-p35, UAS-DIAP1, fidgetin Df(2L)BSC163/CyO, w1118, P{UAS-bsk.DN}2. The following

RNAi lines were provided by Vienna Drosophila RNAi Center: UAS-rpn6-RNAi (103942), UAS- auroraA-RNAi (108446), UAS-auroraB-RNAi (104051), UAS-polo-RNAi (20177), UAS-p38-RNAi

(52277), UAS- calcineurin-RNAi (32283), UAS-calpain-RNAi (35262), UAS-calpainD-RNAi (17740),

UAS-GSK3β-RNAi (101538), UAS-skittles-RNAi (101624), and UAS-toucan-RNAi (24083). UAS- rtnl2-RNAi (VDRC 33318) and yw flies were used as controls.

117

Chapter Five:

Conclusions and future avenues of research

Summary

Work over the past decades has greatly advanced our understanding of the cellular and molecular basis of neuron degeneration in different contexts. Several neural injury paradigms have developed in mice, zebrafish, C.elegans and Drosophila. Drosophila, with its rapid developmental cycle, simple nervous systems and signaling pathways strikingly similar to those of mammals, has emerged as a powerful tool for studying human diseases. Its olfactory system, brain explants, VNC of motor neurons, da neurons, and wing have been well characterized for identifying pathways required for injury–induced neuron degeneration or developmental pruning. This genetically approachable model organism has proven a useful tool for accelerating mechanistic insight into injury study and for developing potential therapeutic targets for neurodegenerative diseases.

Differences between injury-induced degeneration and developmental pruning

Despite the remarkable similarities in the cell biology of injury-induced degeneration and developmental pruning, the underlying molecular mechanisms are surprisingly different.

Comprehensive knowledge of how neurons initiate degeneration in different contexts would help translate the basic research data to understanding why neurons degenerate in different neurological diseases. This knowledge would also facilitate identifying new targets for therapeutic interventions in neurological conditions associated with axon and dendrite degeneration.

118

Neuronal functions of the microtubule depolymerizing proteins and severing proteins

We have examined the role of two microtubule destabilizers, Klp59C and fidgetin, in the regulation of microtubule disassembly in ddaC neurite degeneration. Our findings reveal a novel neuronal function for theses proteins, despite their conventional cellular function in mitotic and meiotic cytoplasm. Kinesin-13 proteins disassemble microtubules in various experimental systems in vivo or in vitro, however, only loss of klp59C suppresses neuron degeneration among the three Drosophila kinesin-13 proteins (klp10A, klp59C and klp59D)[1]. It’s possible that klp59C binds to the plus ends of growing microtubules and depolymerizes both growing and stable microtubules from the plus ends.

Besides our work in Drosophila, kinesin-13 also plays a role in C.elegans neurons. In mature C.elegans neurons, the worm kinesin family protein KLP-7 keeps microtubules growing at a constant rate. Lack of this protein caused the decrease in microtubule dynamics [2]. Another microtubule regulator that we found to be involved in dendrite degeneration, fidgetin, belongs to the AAA ATPase protein family.

AAA proteins participate in a wide range of cellular functions, such as protein and endosome sorting, membrane fusion and spindle formation during meiosis [3]. Katanin and spastin play important roles in nascent protrusion elongation during neuron development, structural maintenance in mature neurons, synapses plasticity in neuronal communication and etc [4-7]. Both human and Drosophila fidgetin depolymerize microtubules in tissue cell cultures [8, 9]. These observations indeed prove that fidgetin has a severing function. However, so far little research has reported a role of fidgetin in neurons. In mice, loss of fidgetin promotes axon elongation during neuron development. We did not observe dendrite pruning defects in ddaC neurons, but loss of fidgetin did affect injury induced -dendrite degeneration. Additionally, in Drosophila S2 cells, figetin localizes to the minus ends of microtubules in centrosomes [8]. Our observation that fidgetin RNAi does not affect EB1 dynamics after injury is

119 consistent with this observation from Sharp’s lab. Thus we hypothesized that fidgetin might work at the minus ends to cooperate with klp59C, which works at the plus ends, to accelerate the microtubule degradation in dendrite degeneration. Additionally, loss of fidgetin and klp59C only delays microtubule degradation for a few hours. Eventually, microtubules still become fragmented. It’s highly likely that in addition to the two proteins and loss of ATP/GTP would stop microtubules from growing, other microtubule regulating mechanisms work in parallel to break down microtubules.

To better understand the neuronal function of Klp59C and fidgetin, we could develop methods to test the protein subcellular localizatons and their endogenous levels. The changes in localization and protein expression level would indicate the exact function site and function time of each kinesin13 protein and microtubule severing protein in vivo. Klp10A and klp59C antibody as well as full-length

EGFP-tagged Klp10A and 59C are available for Drosophila S2 cells and Xenopus egg extracts in vitro

[2, 10]. Immunostaining or cell 4D imaging were able to provide detailed localization information of protein localization of Klp59C and fidgetin and thus revealed their different binding preferences [11].

According to our neuron degeneration data, it’s possible that klp59C- GFP is evenly distributed throughout the neuron or has a higher concentration in the axon while fidgetin is mostly concentrated in the dendrites. Thus, tagged proteins and antibodies are necessary for different models to detect the overexpression or endogenous protein levels throughout neuron degeneration.

Regulation of microtubule severing proteins

It is clear that tight regulation of microtubule breakdown is necessary for normal neuron degeneration in Drosophila neurons. Knockdown of fidgetin does not cause changes in microtubule architectures or dynamics before injury. However, depletion of these destabilizers leads to defects in neurites removal after injury and in developmental pruning. These findings all suggest that these

120 microtubule destabilizers need to be specifically activated shortly after the injury by the hormone signals or other cues. The exact signals that activate each microtubule regulator and how they are regulated through the degeneration process remains unclear.

Further investigation would consider how these microtubule regulators are activated upon injury and regulated during degeneration. Calcium signals have been considered important for microtubule breakdown. Increased calcium in cytoplasm activates serine-thronine protease calpain

[12]. Besides calcium signals, phosphorylation of the microtubule regulators is another possible regulation mechanism. JNK pathway has been considered as a promising candidate for this regulation.

Lack of the normal JNK function leads to axon and dendrite degeneration in the spinal cord and the hippocampus of JNK-/- mice. Through phosphorylation of the specific microtubule regulators, JNK indirectly controls the microtubule stability and dynamics in neurons. In cultured neurons, JNK2 expression causes MAP2 and MAP1B hyperphosphorylation. Hyperphosphorylated MAPs have a compromised ability to bind to microtubules. Without the protection of MAPs, microtubules are vulnerable to severing activities, which leads to microtubule disintegration [13]. Its upstream regulator

DLK also has been proposed to regulate kinesin13 protein KLP-7 in worms after axotomy. In this model, DLK1 is activated by axotomy and could phosphorylate KLP-7. Phosphorylated KLP-7 can no longer bind to and depolymerize microtubules. Thus, inhibition of depolymerization can facilitate growth cone formation and neurite extension [2]. In Addition to the JNK pathway, different phosphatases, which regulate the specific microtubule regulators during degeneration, are waiting to be discovered. A third possible regulation mechanism would be post-translational modifications of microtubule-associated protein and tubulin upon injury. A good deal of evidence indicates that this mechanism is present in non-neuronal cells. For example, MAP4 and tau can limit the severing proteins’ accessibility to the microtubules. Additionally, post-translational modifications of the C- terminal tails of tubulin greatly enhance or weaken the severing proteins’ binding abilities to the

121 surface of microtubules [14-16]. It would be interesting to explore whether injured neurons apply similar mechanisms to regulate the activities of microtubule destabilizers.

Further questions in neuron degeneration research

Neuron severing serves as a simple model for neuron degeneration study. So far, most factors delay neuron degeneration through a gain-of-function mechanism (overexpression of Nmnat and

Wlds). The delay mechanism is local and transcriptionally independent. To identify endogenous genes whose cellular functions are to promote neuron degeneration, efficient in vivo forward-genetic approaches are desired. As more endogenous players start to emerge, further questions need to be addressed; such as whether their functions are conserved across species and whether in vivo crosstalk exists between these identified mediators.

In addition to proteins, research on the function of the ER, Golgi and mitochondria in the degeneration process is also essential. Mitochondria have been at the center of this debate that whether these organelles play a role in neurite degeneration. Researchers are not sure about whether it is involved in neuron degeneration and what role it plays if it is. Our work showed that mitochondria are dispensable for dendrite fragmentation. Consistent with our observation, physical removal of mitochondria by mutation also did not affect Wallerian degeneration in worm ventral nerve cords and

GABA motor neurons[17]. However, the depletion of mitochondria causes normal neurons to initiate degeneration automatically without injury. Additionally, increased mitochondrial movement and its capability to balance against dramatic calcium increase after injuries are important for the protection of

Wlds [18, 19]. These results all suggest that instead of playing an active role in neuron degeneration, mitochondria are more likely to be involved in healthy neuron maintenance. How mitochondria protect injured axons in Wlds neurons and how they maintain cell homeostasis in healthy neurons are not

122 completely known. Growing evidence showed that in addition to its crucial role of generating ATP, mitochondria have other functions in neurons. Another organelle, the Endoplasmic Reticulum, has recently reported to play a role in axon degeneration [20]. As a primary cellular calcium sink, ER released calcium after injury. This process contributes to the intra-cellular calcium spike. This calcium increase leads to mitochondria dysfunction and axon degeneration [20]. It would be interesting to test if

ER is also the cause of calcium waves that are observed in fly neurons after transection and whether blocking ER calcium release could similarly delay the onset of neuron degeneration in fly neurons.

Along with the intrinsic factors that have been widely accepted to play important roles in neurites degeneration, accumulating evidence has emerged that extrinsic factors are also required in these processes. However, it’s controversial whether these extrinsic factors are required for the early initiation and the executive phase of degeneration or they only participate in cleaning the debris during the final stage. Some evidence showed that surrounding cells also act at the early step of neurite degeneration. For examples, time-lapse imaging demonstrated that phagocytic blood cells attack and sever intact dendrite arbors of dda neuron during metamorphosis [21]. Disrupting the phagocytosis and endocytosis function of glial cells through expressing a dominant negative form of dynamin (shibire) protein blocks axon pruning [22]. Glial cells were also observed to infiltrate the axon lobes before the onset of pruning of Mushroom Body neurons and were responsible for the disappearance of varicosities on the axon branches at the early phase of axon pruning [22, 23]. In addition to a role in the initial step of degeneration, surrounding cells have been widely accepted to be crucial for debris clearance in both axon and dendrite degeneration. This step is equally important for organisms to avoid further inflammations [24]. For instances, has also been reported to be responsible for debris clearance of a pruning axon of mushroom body neurons in Drosophila [22]. In this process, glial cells move to the proximity of the fragments, wrap these fragments and digest the

123 debris within cells. Distinct pathways have been reported to regulate the activation and phagocytosis function of glial cells after axotomy of CNS [25]. It is not surprising that glial cells are involved in all the steps of degeneration as the cells might be controlled through different molecular pathways.

Compared to axons, ddaC dendrites in Drosophila use different mechanisms to clear the fragments during developmental pruning and injury-induced degeneration. Using a novel tool (membrane- associate PH sensor), Jan’s group demonstrated that the dendrite debris are internalized and digested in different compartments of epithelial cells during both metamorphosis and severing caused dendrite degeneration [26]. On the other hand, the information of what surrounding cell types are involved in axon and dendrite degeneration and final debris clearance in vertebrate CNS and PNS are still lacking.

Due to the complexity of nervous system in vertebrate, there are more possible candidates that could interact with neurons to facilitate degradation of injured or unnecessary neurons. Questions also rise on how glial cells distinguish the injured neuron and healthy neuron and how glial cells accumulate and stop working at the right timing.

Closing remarks

This work has presented novel findings with regard to how microtubules are disassembled in injury-induced neuron degeneration and developmental pruning (Figure1). Obviously, microtubules become fragmented after injury and during pruning but it was not previously know what microtubule regulators mediates this process. Through a large genetic screen in the ddaC neurons of a Drosophila neuron degeneration model, we identified a general microtubule regulator klp59C, which is involved in all types of degeneration, and a specific microtubule scissor fidgetin, which only acts in injury-induced dendrite degeneration. Our experiments show that the tight regulation of stable and dynamic microtubules after injury and during pruning is essential for proper neurite removal. Our work further

124

provides the evidence to support the hypothesis that

injury-induced degeneration and pruning share certain

pathways but also involve distinct mediators.

We also identified a potential energy indicator

Rab5 (Figure1). Its morphological switch from

punctae to diffuse reflects energy collapse in a living

cell or a fixed cell/tissues sample. This discovery

allows us to compare the precise timing of energy

depletion and microtubule depolymerization during

degeneration. This energy indicator can be further applied to distinguish the active energy-consuming events in degenerating neurites.

125

References

1. Endow, S.A., F.J. Kull, and H. Liu, Kinesins at a glance. J Cell Sci, 2010. 123(Pt 20): p. 3420- 4. 2. Ghosh-Roy, A., et al., Kinesin-13 and tubulin posttranslational modifications regulate microtubule growth in axon regeneration. Dev Cell, 2012. 23(4): p. 716-28. 3. Patel, S. and M. Latterich, The AAA team: related ATPases with diverse functions. Trends Cell Biol, 1998. 8(2): p. 65-71. 4. Ahmad, F.J., et al., An essential role for katanin in severing microtubules in the neuron. J Cell Biol, 1999. 145(2): p. 305-15. 5. Sherwood, N.T., et al., Drosophila spastin regulates synaptic microtubule networks and is required for normal motor function. PLoS Biol, 2004. 2(12): p. e429. 6. Trotta, N., et al., The hereditary spastic paraplegia gene, spastin, regulates microtubule stability to modulate synaptic structure and function. Curr Biol, 2004. 14(13): p. 1135-47. 7. Yu, W., et al., The microtubule-severing proteins spastin and katanin participate differently in the formation of axonal branches. Mol Biol Cell, 2008. 19(4): p. 1485-98. 8. Zhang, D., et al., Three microtubule severing enzymes contribute to the "Pacman-flux" machinery that moves chromosomes. J Cell Biol, 2007. 177(2): p. 231-42. 9. Mukherjee, S., et al., Human Fidgetin is a microtubule severing the enzyme and minus-end depolymerase that regulates mitosis. Cell Cycle, 2012. 11(12): p. 2359-66. 10. Tournebize, R., et al., Control of microtubule dynamics by the antagonistic activities of XMAP215 and XKCM1 in Xenopus egg extracts. Nat Cell Biol, 2000. 2(1): p. 13-9. 11. Mennella, V., et al., Functionally distinct kinesin-13 family members cooperate to regulate microtubule dynamics during interphase. Nat Cell Biol, 2005. 7(3): p. 235-45. 12. Berridge, M.J., P.H. Cobbold, and K.S. Cuthbertson, Spatial and temporal aspects of cell signalling. Philos Trans R Soc Lond B Biol Sci, 1988. 320(1199): p. 325-43. 13. Chang, L., et al., JNK1 is required for maintenance of neuronal microtubules and controls phosphorylation of microtubule-associated proteins. Dev Cell, 2003. 4(4): p. 521-33. 14. McNally, K.P., D. Buster, and F.J. McNally, Katanin-mediated microtubule severing can be regulated by multiple mechanisms. Cell Motil Cytoskeleton, 2002. 53(4): p. 337-49. 15. Qiang, L., et al., Tau protects microtubules in the axon from severing by katanin. J Neurosci, 2006. 26(12): p. 3120-9. 16. Lacroix, B., et al., Tubulin polyglutamylation stimulates spastin-mediated microtubule severing. J Cell Biol, 2010. 189(6): p. 945-54. 17. Rawson, R.L., et al., Axons degenerate in the absence of mitochondria in C. elegans. Curr Biol, 2014. 24(7): p. 760-5. 18. Fang, Y., et al., A novel Drosophila model of nerve injury reveals an essential role of Nmnat in maintaining axonal integrity. Curr Biol, 2012. 22(7): p. 590-5. 19. Avery, M.A., et al., WldS prevents axon degeneration through increased mitochondrial flux and enhanced mitochondrial Ca2+ buffering. Curr Biol, 2012. 22(7): p. 596-600. 20. Villegas, R., et al., Calcium release from intra­axonal endoplasmic reticulum leads to axon degeneration through mitochondrial dysfunction. J Neurosci, 2014. 34(21): p. 7179‐89.

21. Williams, D.W. and J.W. Truman, Cellular mechanisms of dendrite pruning in Drosophila: insights from in vivo time­lapse of remodeling dendritic arborizing sensory neurons. Development, 2005. 132(16): p. 3631‐42. 22. Awasaki, T. and K. Ito, Engulfing action of glial cells is required for programmed axon pruning during Drosophila metamorphosis. Curr Biol, 2004. 14(8): p. 668‐77. 23. Watts, R.J., et al., Glia engulf degenerating axons during developmental axon pruning. Curr Biol, 2004. 14(8): p. 678‐84. 24. Frank‐Cannon, T.C., et al., Does neuroinflammation fan the flame in neurodegenerative diseases? Mol Neurodegener, 2009. 4: p. 47. 25. Ziegenfuss, J.S., J. Doherty, and M.R. Freeman, Distinct molecular pathways mediate glial activation and engulfment of axonal debris after axotomy. Nat Neurosci, 2012. 15(7): p. 979‐87. 26. Han, C., et al., Epidermal cells are the primary phagocytes in the fragmentation and clearance of degenerating dendrites in Drosophila. Neuron, 2014. 81(3): p. 544‐60.

Juan Tao

Education

Ph.D. in Biochemistry and Molecular Biology, 2008-2014 The Pennsylvania State University

B.A. in Biotechnology, 2004-2008 Wuhan University, China

B.A. in Literature, 2005-2008 Hua Zhong University of Science and Technology, China

Peer-Reviewed Publications

Chen, L.,Stone, M. C., Tao, J., Rolls, M. M., Axon injury and stress trigger a microtubule- based neuroprotective pathway. Proc Natl Acad Sci U S A, 2012. 109(29): p. 11842-7.

Stone, M. C., Rao, K., Gheres, K. W., Kim, S., Tao, J., La Rochelle, C., Folker, C. T., Sherwood, N.T., Rolls, M. M., Normal spastin gene dosage is specifically required for axon regeneration. Cell Rep, 2012. 2(5): p. 1340-50.

Tao, J. and M.M. Rolls, Dendrites have a rapid program of injury-induced degeneration that is molecularly distinct from developmental pruning. J Neurosci, 2011. 31(14): p. 5398-405.

Stone, M.C., Nguyen, M.M., Tao, J., Allender, D.L., and M.M. Rolls. Global up-regulation of microtubule dynamics and polarity reversal during regeneration of an axon from a dendrite. Mol Biol Cell, 2010. 21(5): p. 767-77.

Mattie, F. J., Stackpole, M. M., Stone, M. C., Clippard, J. R., Rudnick, D. A., Qiu, Y., Tao, J., Allender, D. L. ,Parmar, M.,Rolls, M. M.., Directed microtubule growth, +TIPs, and kinesin-2 are required for uniform microtubule polarity in dendrites. Curr Biol, 2010. 20(24): p. 2169-77.

Professional Organizations

Society for Neuroscience (SFN) American Society for Cell Biology (ASCB)

Presentations

Speaker: Mid-Atlantic Regional Meeting of the Society for Developmental Biology in 2011 Conference Poster Presentations: American Society for Cell Biology 2010, 2012 and 2011