New Understanding of the Effects of Microtubule- Associated Proteins on Dynamic Instability, Including the Role of Tau in Neuronal Function and Disease By

New understanding of the effects of microtubule- associated proteins on dynamic instability, including the role of tau in neuronal function and disease By

Timothy Ottum Austin

May 2018

A Dissertation Presented to the Faculty of Drexel University College of Medicine in partial fulfillment of the Requirements for the Degree of Doctor of Philosophy

Timothy Cunningham Peter W. Baas Ph.D. Ph.D. Neurobiology and Anatomy Neurobiology and Anatomy

Mark Black Nicholas M. Kanaan Ph.D. Ph.D. Lewis Katz School of Medicine Translational Science and Molecular Temple University Medicine Michigan State University

Liang Qiang Ph.D. Neurobiology and Anatomy

ACKNOWLEDGEMENTS

My deep appreciation to Dr. Peter Baas for acting as my mentor and advisor, supporting me through my studies and research and making this work a possibility. I couldn’t have asked for a better scientist or leader to have as my guide.

Thanks to my committee, Dr. Timothy Cunningham, Dr. Mark Black, and Dr. Nicholas

Kanaan, for their feedback, guidance, and scientific inquiry. An especial thanks to Dr.

Liang Qiang, whose constant and reliable input made this work a reality.

This research is a result of prodigious collaborative effort. My especial thanks to Andrew

Matamoros and Xiaohuan Sun for their hard work, dedication and scientific rigor.

Thanks to Dr. Wenqian Yu for scientific and administrative support that was critical to my training and work for the entirety of my tenure in the Baas laboratory.

Thanks to Dr. Shen Lin, Dr. Olga Kahn, Dr. Aditi Falnikar, Dr. Daphney Jean, Dr. Anand

Rao, and Dr. Lanfranco Leo for their training, teaching and support in the Baas lab.

Thanks to Dr. Emanuela Piermarini, Ankita Patil, Hemalatha Muralidharan, Silvia

Fernandes, and Philip Yates for being supportive and fantastically enthusiastic labmates.

My thanks to the administrators, staff, teachers, and professors at Drexel University.

Their hard work, professionalism and willingness to teach are a wonder.

Thanks to my parents, without whose support I never would have finished.

iii

TABLE OF CONTENTS

LIST OF ILLUSTRATIONS ……………………………………………………………………. v

ABSTRACT ……………………………………………………………………….…………... vii

BACKGROUND ……………………………………………………………………………...…. 1

NEURONAL MICROTUBULES ……………………………………………………………….. 1

MICROTUBULE COMPOSITION AND DYNAMICS……………………………………….... 3

FIDGETIN ………………………………………………………………………………...……... 7

TAU AND MAP6 ………………………………………………………………………...……… 8

CHAPTER 1: NANOPARTICLE DELIVERY OF FIDGETIN SIRNA AS A MICROTUBULE-BASED THERAPY TO AUGMENT NERVE REGENERATION ……..… 11

CHAPTER 1: INTRODUCTION ………………………………………………………….….... 12

CHAPTER 1: RESULTS ……………………………………………………….…………….… 13

CHAPTER 1: DISCUSSION …………………………………………………….………….….. 21

CHAPTER 1: METHODS ……………………………………………………………………… 23

CHAPTER 2: TAU DOES NOT STABILIZE AXONAL MICROTUBULES BUT RATHER ENABLES THEM TO HAVE LONG LABILE DOMAINS ……………………….. 35

CHAPTER 2: INTRODUCTION ………………………………………………………………. 36

CHAPTER 2: RESULTS AND DISCUSSION ………………………………………………… 37

CHAPTER 2: MATERIALS AND METHODS ……………………………………………..… 49

CONCLUSIONS AND RECOMMENDATIONS ……………………………………...……….69

LIST OF REFERENCES ……………………………………………………………………..… 74

LIST OF ILLUSTRATIONS

Figure 1. Validation of siRNA nanoparticles in adult DRG primary cultures ………………. 28

Figure 2. Microtubule mass is increased in the axons of adult DRG neurons as a result of fidgetin knockdown ……………………………………………...………. 29

Figure 3. Fidgetin knockdown decreases ratio of acetylated to total tubulin in the axons of DRG neurons ……………………………...………………………….... 31

Figure 4. Fidgetin knockdown positively affects axon outgrowth in a manner dependent upon un-acetylated tubulin …………………………………………………. 32

Figure 5. Effects of fidgetin knockdown on axonal growth onto an inhibitory substrate ……………………………………………………………………... 33

Figure 6. Schematic illustration of how fidgetin knockdown boosts axon regeneration ……………………………………………………………………………...34

Figure 7. Distribution of tau and effects of its depletion on axonal microtubules are inconsistent with tau’s purported role as a stabilizer of axonal microtubules …………. 55

Figure 8. MAP6 is recruited to and stabilizes neuronal microtubules of cortical neurons when tau is depleted ……………………………………………………………………. 58

Figure 9. Opposing effects of tau and MAP6 depletion on microtubule stability in the axon …………………………………………………………………………………….. 60

Figure 10. Mathematical analysis of microtubule decline as a result of exposure over time to nocodazole, and schematic summary of conclusions ………………………….. 62

Figure S1. Tau siRNA validation in immunostaining and Western blot and tau distribution ……………………………………………………………………………... 64

Figure S2. Alternation in tau distribution when MAP6 is depleted ……………….…… 65

Figure S3. MAP6 recruits in lamellipodia along the axons of cortical neurons after tau is depleted ………………………………………………………………………………… 67

Figure S4. Doublecortin is not recruited to microtubules after tau depletion from cortical neurons …………………………………………………………………………………. 68

ABSTRACT

New understanding of the effects of microtubule-associated proteins on dynamic instability, including the role of tau in neuronal function and disease Timothy Ottum Austin Peter W. Baas

As morphologically complex and polarized structures, neurons are critically dependent on cytoskeletal function. Neuronal processes are formed and maintained on a backbone of microtubules, which not only act as a structural support for axonal growth and formation, but also play important roles through their dynamic instability. Proper microtubule function, and by extension axonal function, relies upon distinct sub-populations of highly dynamic and highly stable microtubules. The functional difference between these domains has increasingly been shown to play a role in how microtubules promote axonal growth and interact with microtubule-associated proteins. Current dogma suggests that microtubule degeneration in neurological disease may best be averted by using microtubule-stabilizing drugs. In contrast, we demonstrate how siRNA knockdown of Fidgetin, a protein that specifically severs unstable microtubules, can promote neuronal growth by enhancing labile microtubule mass. We also investigate the role of tau in microtubule dynamics, demonstrating that tau does not act as a stabilizer of microtubules but rather promotes the growth and retention of dynamic microtubules, while MAP6, another microtubule- associated protein with true stabilizing characteristics, acts in opposite fashion. These findings open a new path for research into the mechanisms of tau and the possibilities of using labile microtubule manipulation to treat disease. vii

BACKGROUND

There are approximately 85 billion neurons in the human brain, besides those in the spinal cord and peripheral nervous system.1 Neurons are inherently polar cells, obvious not only in function but also morphology, with systems of dendrites funneling electrochemical signals from synaptic connections through (relatively short) neuritic processes to the cell body2. From there the electrical signal is carried by (relatively long) axonal processes to highly specified targets, triggering synaptic release and chemical signaling. These axons, which are the particular focus of this study, are responsible for making some neurons the longest cell in the human body, with single cells extending up to almost a meter in length.2

The intricate neural networks formed by such long and specific axonal connections are built upon a sophisticated system of intracellular architecture. Microtubules are one of the major components of the neuronal cytoskeleton, critical for the initial development of neuronal processes, the branching behavior of axons, the motility of the growth cone, the transport of intracellular materials, and the overall maintenance of neuronal processes.3

Neuronal Microtubules

Whereas microtubules function in most cells to regulate the processes of cell division and the cell cycle, including genome assembly during meiosis and mitosis, neurons are postmitotic and will no longer utilize microtubules for chromosome sorting following neuronal differentiation.4 Instead, microtubules play myriad roles related to the specialized functions of neurons. As noted above, neuronal function is unfailingly linked to neuronal

2 morphology and the polarization of the cell.3 Correct development of a differentiated neuron from an amorphous cell body depends on microtubule outgrowth. Disruption of microtubule sorting leads to malorientation of dendrites and axon (including switching characteristics of one neurite subtype to another), disturbing normal neuronal function.5,6

Extracellular signals activate downstream processes that sever microtubules in preparation for axonal or dendritic branching.7,8 As neurites extend along chemical or physical pathways to their destined target locations, microtubules act in concert with other cytoskeletal elements such as the actin lattice and extracellular binding proteins to facilitate adhesion and extension of the neurite.8 The growth cone is a key figure of neurite extension, being the splayed distal end of an axon containing a major microtubule-actin network that responds to turning cues.9,10 Microtubule depolymerization results in collapse of the growth cone, and if persisted, retraction of the axon. On the other hand, overt microtubule stabilization causes growth cones to ignore appropriate axonal guidance cues, potentially leading to missed or extraneous synaptic formations at inappropriate sites.11 Growth and function of axons up to and following synapse formation depends on axonal transport, as proteins appear to be synthesized almost exclusively within the cell body.12 The characteristics of microtubule transport have been well-illustrated, with vesicles and protein aggregates passed up and down the length of the neuron by kinesin and dynein motor proteins. The products carried to the distal regions of the axon include the neurotransmitters and peptides necessary for chemical communication at the synapse as well as any components necessary for growth and maintenance of the distal cell.12

Damaged proteins and organelles are carried back proximally for degradation within the cell body. Due to the required length of axons, disruption of the microtubule lattice and, by

3 extension, microtubule-based transport can be especially devastating to the cell.12

Degradation and loss of microtubules is a common component of many neuropathologies including Huntington’s, Parkinson’s and Alzheimer’s Disease. Microtubule loss is associated with a whole set of diseases labeled as tauopathies, diseases that specifically result from disruption of the microtubule-associated protein tau.13 These tau disruptions encompass both loss-of-function and gain-of-function effects. Loss-of-function effects include leaving microtubules vulnerable to severing proteins, while gain-of-function effects can involve much more complex and relatively uncharacterized signaling pathways leading to adverse phosphorylation, changes in microtubule post-translational modification, and toxic oligomerization.13-15 In these disease states, the removal of microtubules is sometimes attributed to a “destabilization” of the microtubule lattice that results in a depolymerization of tubulin from the cytoskeleton. However, intricacies of how and why microtubules are stabilized or destabilized requires a closer analysis of the innate characteristics of microtubule polymerization and tubulin enzymatic activity.

Microtubule Composition and Dynamics

Narrowing in on the arrangement of the neuronal architecture, microtubule polymers are assembled into cytoskeletal rods composed of 13 protofilaments arranged symmetrically into a hollow tube.16 Protofilaments are made up of individual subunits called tubulin, which are expressed ubiquitously through the cell body and processes. The globular tubulin subunits that make up microtubules are dimers composed of α- and β-tubulin. Dimers bind through a GTP-mediated mechanism, head-to-tail in a polarized fashion. This begins with

4 a site of initial polymerization in the centrosome with the aid of a third tubulin subtype, γ- tubulin. The nature of this tubulin-binding pattern ensures that microtubules are always generated as polarized structures. The nature of individual tubulin subunits having α- tubulin at one end and β-tubulin at the other is maintained throughout the entire microtubule polymer, with the round end of α-tubulin referred to as the “minus-end” and the β-tubulin referred to as the “plus-end.” Every microtubule maintains the plus- and minus-ends as characterized, whether polymers are growing, splitting or degenerating. The plus-end generally faces distally in axons while splitting evenly between distal- and proximal-facing in dendrites. The GTP-dependent assembly of microtubules plays a major role in the properties of microtubule growth and retraction.3 Tubulin dimers have intrinsic

GTPase activity. When dimers bind to GTP, they are prone to assemble onto developing protofilaments, leading to the growth and extension of microtubules. Upon being incorporated into a microtubule lattice, interactions within the lattice cause GTP to be hydrolyzed to GDP, changing the conformation of the tubulin subunit. This change in the tubulin conformation results in a tendency to depolymerize. As a microtubule grows, the leading end is prone to instances of rapid depolymerization (catastrophe) or subsequent recovery of polymerization (rescue). The tendency of a microtubule to undergo changes, whether for growth or reduction, is termed dynamic instability.3 Microtubule populations that do not undergo consistent growth or catastrophe cycles are termed “stable,” while those that do are termed “labile.” The relative stability of a particular portion of a microtubule depends both on the state of its bound GTP and on the relative association it with end-binding proteins and other microtubule-associated proteins. Landmark studies done in the 1990s revealed that within some sets of axons, individual microtubules within

5 the axon consisted of approximately 50% stable tubulin (located on the minus-end half) and 50% labile (located towards the plus end) in two surprisingly distinct domains.17,18 The stable or labile properties of a microtubule can be associated with various different measures which each have stronger or weaker correlation with tendency of tubulin turnover. These measures include relative distribution of post-translational modifications such as microtubule acetylation which tends to correlate with stable microtubule mass, and tyrosinated tubulin which tends to correlate with labile microtubule mass.3 Many other modifications exist which each correlate more or less to microtubules of a particular dynamicity. While useful, these modifications cannot be said to be completely exclusive to their preferred dynamic domain. Furthermore, the extent to which post-translational modification depends on stability state as opposed to stability state being affected by protein interactions with post-translational modifications is unclear. Certainly microtubule- associated proteins can be expected to play primary roles in conferring stability properties.3

Vocabulary clarifications become necessary when translating the details of microtubule dynamics to the symptoms seen in neurodegenerative diseases. Within the realm of microtubule dynamics, the terminology of “stable” and “labile” is used independently of the actual long-term fate of a particular tubulin chain, but these words can be used in multiple contexts and as such are not always used consistently to refer to the actual concept of dynamic instability. In other words, a set of microtubules may be considered

“destabilized” if a change in cellular conditions causes rapid disappearance of tubulin from the cytoskeletal lattice back into the pool of free-floating tubulin. Considered purely from the view of cellular symptoms, the microtubules were made “less stable” because they were no longer complete and whole within the cellular structure. From the viewpoint of dynamic

6 instability, the options are not exclusively stable survival and dynamic destruction. A microtubule could hypothetically have no change in actual dynamic properties and simply undergo more catastrophe than rescue in its labile portion due to a change in availability or binding capacity of protective and growth-promoting microtubule-associated proteins, producing the result of a diminishing microtubule mass.19 Stable domains could indeed be destabilized, making them more prone to change and therefore more susceptible overall to catastrophe, which could lead to microtubule loss as well. A third option involves the action of microtubule-severing proteins, which may have various preferences towards microtubules with labile or stable properties. Spastin is a microtubule-severing that preferentially cuts within microtubule domains with the traditionally stable-associated post-translational modification of poly-glutamylation.20 An overt stabilization of microtubules that leads to poly-glutamylation could theoretically lead to the loss of microtubule mass by inducing excessive severing by spastin. Thus, in this mechanism, the

“destabilized” microtubules that are preferentially removed undergo their destruction as a result of increased dynamic stability. For this current work, “stable” and “labile” will be used to refer to the dynamic properties of a microtubule and its propensity to change, rather than the actual capability of a given microtubule population to be removed long-term from the functional and polymerized neuronal cytoskeleton.

Far from being a mere semantic exercise, the distinction between labile and stable domains is an important question to consider for practical decisions regarding treatment of neurodegenerative disease. As noted above, microtubule loss and its associated detrimental effects could hypothetically be induced by a variety of changes in microtubule stability.

Stabilization of microtubules could potentially have unintended side effects in alteration of

7 microtubule interactions with their environment. The trade of microtubule responsiveness to cues (labile trait) for enhanced axonal growth (stable trait) may appear worth the investigation into microtubule-stabilizing drugs. Indeed, studies have begun looking into the use of microtubule-stabilizing drugs for treatment of neurodegenerative disease and injury, with promising results in the growth and viability of axons.21-24 These drugs have advantages in that are very available, well-studied and approved for non-nervous uses in human subjects, but the alteration in normal dynamic proportions leaves questions to their long-term effectiveness.

In particular, we would ask whether selective enhancement of labile microtubules can lead to the growth benefits of microtubule stabilization without altering the overall form and function of the stable domain. To this end, we investigated a novel treatment paradigm utilizing siRNA matched to the microtubule-severing protein fidgetin.

In addition to establishing the possibility of labile enhancement as another tool in the fight against neurodegeneration, we also wished to investigate the question of whether tau functions as a true stabilizer of microtubules, as many microtubule-stabilizing drugs are hypothesized to be potentially beneficial in that they would replace the presumed stabilizing properties of tau. We challenge this dogma of tau stabilization by investigating the direct effects of tau loss on microtubule dynamics.

Fidgetin

Fidgetin is a AAA protein (ATPases associated with diverse cellular activities) that was first discovered due to a spontaneous mutation that caused developmental defects evident

8 in head-shaking and circling behavior in mice.25 Subsequent analysis revealed that human fidgetin had the capacity in vitro to sever microtubules, and that depletion of fidgetin resulted in an increase of microtubule number and length.26 Detailed analysis of fidgetin within mouse and rat models revealed that fidgetin specifically severs labile microtubules, causing a reduction in microtubule mass within the cell but no change in microtubule number.27 Axonal length was also affected, increasing in cases of fidgetin depletion. Based on these results, we hypothesized that we could promote growth of axons relevant to spinal cord injury models using a nanoparticle-based delivery of siRNA against fidgetin. The details of this work comprise Chapter 1, wherein we outline enhanced axonal growth by promotion of labile microtubule mass alone, on both permissive and restrictive substrates.

Tau and MAP6

Tau is a microtubule-associated protein primarily found within the axons of the neurons.13

The exact function of tau, especially as it relates to the pathological loss- or gain-of- function effects, has been a matter of discussion for some time. It has been almost universally accepted that in the healthy neuron, tau acts as a microtubule stabilizer. This was based in part on experiments showing the capability of tau to stabilize microtubules in vitro, as well as the association of tau loss with depleted microtubule mass in tauopathies such as Alzheimer’s disease. However, overexpression of microtubule-associated proteins may lead to stabilization simply by nature of there being an overabundance of binding partners. As already outlined, absolute loss of microtubule mass as seen in Alzheimer’s disease does not necessarily imply that it was a destabilization of microtubules that led to

9 the reduction in polymerized cytoskeleton. Several previous lines of inquiry may in fact indicate that tau loss does not lead to an overall loss of stable microtubule mass, and tau’s effect on microtubules may not be to function as a stabilizer of microtubule dynamics, but rather as regulator of specific dynamic domain survival. One group of studies that most directly addresses this issue in actual human tissue revealed that brain samples from

Alzheimer’s patients had relatively diminished microtubule levels when compared to controls, but when analyzed for post-translational modifications, the relative levels of acetylated tubulin were higher than those of control samples, suggesting that the loss of tau may not actually have resulted in an overall destabilization of microtubule dynamics.28,29

Alternative explanations could result from the variability in correlation of post- translational modifications with dynamic instability, but the results do call into the question the notion of tau as a stabilizer. The actual molecular action of tau does not necessarily suggest the role of a stabilizer, considering a relatively rapid on-off activity of microtubule binding that only lasts about 40ms.30 Previous studies have also indicated that tau may be more prominent in distal portions of axon, where microtubules in general tend to be more labile.31 If in fact tau is associated with the labile domain, then treatments for tauopathy

(relevant at least to loss-of-function effects) would benefit not so much from microtubule stabilization as much as promotion of the labile domain. This would potentially restore not only healthy phenotypes of microtubule level, but also reduce the potential disparity of an excessive proportion of stable microtubules following neurodegeneration. We analyzed detailed microtubule dynamicity indicators following tau depletion in embryonic rat cells to ascertain what the actual effects of tau depletion on relative microtubule domain levels were. As part of this study, we also analyzed a known stabilizer of microtubules, MAP6.

MAP6 has been shown to bind microtubules and prevent destabilization and depolymerization when cells are exposed to microtubule-depolymerizing drugs and cold.32,33 It was thought that this could provide a good comparison of tau’s effects against another potential microtubule stabilizer, and lead to examinations of potential links between microtubule-associated proteins. The details of this study are analyzed in Chapter

In sum, microtubules are a critical component of the neuronal cytoskeleton. The unique structural demands in the nervous system and multiple pathological endpoints affecting the cytoskeleton make clear the necessity of proper microtubule function for neuronal health.

Strategies focusing on shoring up the cytoskeletal rigging of neurons have successfully promoted axonal integrity, but functional improvements remain elusive. This may in part be to an almost universal approach of hyperstabilization, wherein microtubule viability is enhanced while microtubule functionality is lost, due to a reduction in dynamic adaptability. Our study was to examine the intricacies of microtubule dynamics and how manipulations of microtubule associated proteins can target ever more elegantly the specific subsets of microtubules that may be lost in neuronal degeneration.

CHAPTER 1: NANOPARTICLE DELIVERY OF FIDGETIN SIRNA AS A MICROTUBULE-BASED THERAPY TO AUGMENT NERVE REGENERATION

Introduction

The regenerative capacity of injured adult axons is limited, particularly in the central nervous system. Injured axons tend to degenerate. If they do regrow, they encounter obstacles such as scar tissue and inhibitory molecules, and their growth rates simply do not match that of a juvenile axon.34,35 In recent years, the regeneration community has been intrigued by the idea of microtubule-stabilizing drugs as a therapy to augment nerve regeneration. Although microtubules in adult axons are already more stable than in juvenile axons, the premise is that perhaps the injured adult axon would retract less, grow better, and power its way through inhibitory environments if its microtubules were even more stable. Encouraging results with microtubule-stabilizing drugs have been obtained with preclinical rodent models for spinal cord injury.36,37 However, some of the results have been difficult to reproduce, with limited benefits more attributable to drug effects on scar tissue-forming cells rather than on neurons.38 In addition, the logic of microtubule stabilization as a therapy has been questioned for various reasons, including negative effects that could outweigh positive benefits.39,40 For example, an individual microtubule in the axon consists of a stable domain and a labile domain, with each domain having important work to do. 16 Hence, stabilizing the labile domain could incapacitate a portion of that work.

Rapidly growing axons tend to have a higher proportion of labile microtubule mass. For this reason, our premise is that we can augment nerve regeneration by increasing labile

13 microtubule mass in the axon. We posit that we can do so by reducing the levels of fidgetin, a microtubule-severing protein that normally exists to pare back the labile microtubule mass of the axon.27 To apply this strategy to adult neurons in animals and human patients, we need a method of delivery of siRNA that is effective, safe, and minimally invasive.

Here we propose to use a relatively new nanoparticle delivery system, which we have tested on cultures of adult rat dorsal root ganglion (DRG) neurons, a well-accepted in vitro model for spinal cord injury. Based on a hydrogel/sugar glass composite, our siRNA delivery platform is a hybrid nanoparticle capable of encapsulating and controllably releasing a broad range of therapeutically relevant materials ranging from gaseous nitric oxide to larger macromolecules such as chemotherapeutic agents and phosphodiesterase inhibitors.

The nanoparticles have been shown to be capable of delivering siRNA to tissues and organs.41

Results

Added to culture medium, the siRNA-encapsulated nanoparticles readily cross the cell membrane and dissolve, releasing the siRNA into the cytoplasm to interact with the RNA- induced silencing complex (Fig. 1A-schematic). Preliminary studies with nanoparticles conjugated to a fluorescent dye demonstrated that the nanoparticles enter neurons in adult

DRG cultures (Fig. 1B), which is not surprising given that they have previously been shown to effectively enter intact tissues and organs.41 It was our impression that neurons, especially ones with larger cell bodies, took up the nanoparticles better than the smaller and flatter cells in the culture. qPCR revealed that some batches of nanoparticles failed to knock down fidgetin (relative to control siRNA nanoparticles), while other batches

14 knocked down mRNA by 25-50% (Fig. 1C). The former batches, which presumably were defective at some point in the preparative procedure, were discarded. Knockdown of fidgetin (relative to control siRNA nanoparticles) with the qPCR-effective batches specifically in the neurons of the culture was confirmed by immunofluorescence (IF) for fidgetin (Fig. 1D). A reduction of IF signal intensity was particularly notable in neurons, perhaps due to their more efficient uptake of the nanoparticles.

Cultures exposed to fidgetin siRNA nanoparticles were markedly different in morphology, as assessed by phase-contrast microscopy, from cultures exposed identically to control siRNA nanoparticles, the latter of which were indistinguishable in appearance from cultures without siRNA treatment (not shown). Qualitatively, the cultures treated with fidgetin siRNA nanoparticles displayed a greater amount of axon mass than controls, which was also apparent in cultures IF-stained for neuron-specific β-III-tubulin to reveal microtubules (Fig. 2A and 2A’).

Nanoparticle delivery of fidgetin siRNA increases labile microtubule mass in the axons of cultured rat adult DRG neurons. Before using sparser cultures to quantify the morphological differences, we used denser cultures to ascertain microtubule levels and stability, as per our earlier work on fetal cortical neurons. Adult DRG cultures are notoriously heterogeneous, consisting of three classes of neurons as well as non-neuronal cells such as satellite cells and fibroblasts. Neurons are not as numerous compared to the other cell types, making Western blotting problematic as a means to assess the status of the microtubules specifically in the neurons. We suspected that the more modest level of

15 fidgetin knockdown achieved with the nanoparticle approach would result in a more modest elevation in microtubule mass than in our earlier study on fetal neurons, in which we were able to achieve near complete fidgetin knockdown by introducing the siRNA by nucleofection.27 For these reasons, we endeavored an IF approach that would acquire data from virtually every neuron in the culture, with sample numbers of roughly 600 for each experimental condition. We reasoned that a more modest effect would likely require a higher sample number to achieve statistical significance, especially in light of the greater heterogeneity of DRG neurons compared to cortical neurons.

Three separate dissections were performed to obtain primary cultures that were then cultured for three days with either control siRNA or fidgetin siRNA nanoparticles.

Experiments were conducted in duplicate. Axonal microtubule fluorescence values were not normally distributed (Shapiro-Wilk test, p<0.005 per group) for control siRNA treated neurons (n=1,726) with a skewness of 2.415 (standard error = 0.059) and kurtosis of 7.706

(standard error = 0.118), or for fidgetin siRNA treated neurons (n=1,888) with a skewness of 1.949 (standard error = 0.056) and kurtosis of 4.409 (standard error = 0.113). The data are positively skewed for both groups and therefore a non-parametric comparison of medians was performed. A Mann-Whitney U test was conducted to measure differences in microtubule fluorescence between control siRNA and fidgetin siRNA treated neurons.

Distributions of microtubule fluorescence values for control siRNA and fidgetin siRNA were similar upon visual inspection, permitting a comparison of medians. Median fluorescence value was statistically significantly higher in fidgetin siRNA treated neurons

(43.03) than in control siRNA treated neurons (36.91), U = 1,850,657.5, z = 1,850,657.5, p < 0.005.

As with the studies on fetal cortical neurons, we included a 30-minute treatment with nocodazole as well as a 2-hour treatment, in order to assess whether any increase in microtubule mass that we might document is primarily of the labile or stable microtubule fraction (Fig. 2B). In the adult DRG neurons treated with fidgetin siRNA (56.28 ±42.50), there was a 19% increase in mean microtubule mass relative to control siRNA (47.53

±36.56) per unit area of axon (Fig. 2C), which is about a third as much of an increase as we previously reported of the fetal cortical neurons with more complete fidgetin knockdown. After 30 minutes of nocodazole treatment, the microtubule levels in cultures treated with fidgetin siRNA were indistinguishable from those in cultures treated with control siRNA, which was also the case in our earlier study on fetal cortical neurons; therefore, no statistical testing was conducted. These results indicate that the microtubule mass added to the axon as a result of fidgetin depletion is predominantly or entirely labile, because very little of the stable microtubule fraction would be diminished after 30 minutes in drug. Interestingly, the standard deviation for the total microtubule mass was notably higher than for that remaining after 30 minutes of drug (Fig. 2D), which is consistent with the labile component of the microtubule mass being highly dynamic and the stable component being much less dynamic.

Finally, the ratio of acetylated to total tubulin was measured in control siRNA and fidgetin siRNA treated neurons, with and without tubacin, a histone deacetylase inhibitor that increases microtubule acetylation. Without tubacin, the ratio of acetylated to total tubulin

17 relative to control siRNA (Fig. 3A) was lower in axons of fidgetin siRNA treated neurons

(Fig. 3B). In the presence of tubacin, the ratio was heightened for both. Data (shown in Fig.

3C) are mean ± standard deviation. There were 28 control siRNA and 29 fidgetin siRNA distal axons measured. There were no outliers and the data was normally distributed per group, as assessed by histogram analysis and Shapiro-Wilk test (p>0.05 per group).

Levene’s Test for Equality of Variance confirmed that variance was equal (p=0.257). There was a statistically significant difference in mean acetylation ratios between control and fidgetin siRNA treatment groups, t(55) = 2.389, p = 0.020. Mean acetylation ratio for control siRNA treated axons (0.52 ±0.23) was significantly higher than mean fidgetin siRNA treated axons (0.39 ±0.18). Our results are consistent with fidgetin knockdown resulting in the addition of labile microtubule mass, because the ratio of acetylated to total tubulin is expected to be lower upon the addition of labile microtubules.27

Fidgetin knockdown results in increased axonal outgrowth in a manner dependent upon unacetylated tubulin. In order to quantify the morphological effects of the fidgetin knockdown, we grew a set of cultures at a 4-fold sparser density. For these experiments, we also treated some of the cultures with tubacin. In our previous studies on fetal cortical neurons, tubacin prevented fidgetin knockdown from increasing axonal length, which is consistent with fidgetin targeting labile domains of microtubules via a preference for unacetylated tubulin.27 We set out to determine if the same effect is observed in primary DRG cultures supplemented with fidgetin siRNA nanoparticles, as opposed to nucleofection of siRNA.

Data for axon length was not normally distributed, as assessed by Shapiro-Wilk Test

(p<0.005). Control siRNA treated axon length (n=222) had a skewness of 2.283 (standard error = 0.163) and kurtosis of 7.066 (standard error = 0.325). Fidgetin siRNA treated axon length (221) had a skewness of 3.444 (standard error = 0.164) and kurtosis of 18.238

(standard error = 13.980). Tubacin treated control siRNA axon length (114) had a skewness of 4.986 (standard error = 0.226) and kurtosis of 34.048 (standard error = 0.449). Tubacin treated fidgetin siRNA axon length (n=405) had a skewness of 4.986 (standard error =

0.121) and kurtosis of 38.187 (standard error = 0.242). The data are positively skewed for all groups and therefore a non-parametric comparison of medians was performed. A Mann-

Whitney U test was run to determine if there were differences between the control and fidgetin siRNA axon length. Distributions of axon length for control and siRNA treatment groups were similar based on visual inspection. Median axon length was statistically significantly higher for fidgetin siRNA treated axons (95.4µm) than control siRNA treated axons (79.04µm), U=21,828.5, z=-2.006, p=0.045. Fidgetin knockdown relative to control siRNA (Fig. 4A) resulted in increases in average axon length (Fig. 4B). As in our previous study on fetal cortical neurons,27 none of these parameters were different between control siRNA and fidgetin siRNA in the presence of tubacin. These results (data shown in Fig.

4C) are consistent with the effects of fidgetin knockdown on axonal growth being dependent upon the acetylation status of the microtubules.

All of the results presented thus far are consistent with the siRNA entering neurons in the culture, reducing fidgetin levels and producing the predicted results on microtubules and neuronal morphology (on the basis of our previous studies).27 This provides confidence

19 that the nanoparticle approach is an effective means for transfection of siRNA into adult neurons that are difficult to transfect by traditional means (i.e. nucleofection and lipofectamine have poor transfection efficiency for DRG neurons), and that the same principles of fidgetin knockdown previously reported of fetal neurons apply to adult neurons.

Fidgetin knockdown promotes axonal growth on non-permissive substrate. Axon regeneration in the adult central nervous system is not only a matter of axons growing faster in permissive environments, but also crossing into growth-inhibitory environments and beyond. The standard cell culture method for testing the capacity of a treatment regime to assist in this regard is to challenge axons to cross from a favorable polylysine-laminin substrate onto a stripe consisting of laminin together with aggrecan, a growth inhibitory protein associated with the glial scar tissue that develops in response to nerve injury. When axons growing on the favorable substrate encounter the aggrecan border, most turn away from their original projection path to avoid crossing onto the aggrecan (Fig. 5A,B). In the rare cases in which axons cross onto the aggrecan, they grow markedly more slowly than on the favorable substrate and often stop growing altogether42. Here, neurons were treated with control or fidgetin siRNA nanoparticles at the time of plating, and then assessed for crossing 2 days later. Double-labeling IF for neuron-specific β-III-tubulin and fidgetin shows an increase in microtubule invasion into the growth cone, concomitant with the expected decrease in fidgetin (Fig. 5C,D). Seventy-seven axons were identified approaching the gradient and documented as either crossing or not crossing the aggrecan stripe; 35 were from control siRNA treated cultures and 42 were fidgetin siRNA treated

20 cultures. Of the control siRNA treated axons assessed 29 (82.9%) did not cross and 6

(17.1%) crossed. Of the fidgetin siRNA treated axons assessed 27 (64.3%) did not cross and 15 (35.7%) crossed. There was no statistically significant association between nanoparticle treatment and axons crossing. Knockdown of fidgetin exhibits a trend toward increasing the frequency of axonal crossing (Fig. 4E).

Data for axon length of crossing axons were not normally distributed for fidgetin siRNA treated axons, as assessed by Shapiro-Wilk Test (p = 0.025). Whereas control siRNA treated axons were normally distributed (p = 0.584). Control siRNA treated axon length

(n=6) had a skewness of 2.298 (standard error = 0.845) and kurtosis of 5.388 (standard error = 1.741). Fidgetin siRNA treated axon length (15) had a skewness of 0.482 (standard error = 0.580) and kurtosis of -1.542 (standard error = 1.121). The data do not appear normally distributed for either group, and therefore a non-parametric comparison of medians was performed. A Mann-Whitney U test was run to determine the difference between the crossing-axon length of control and fidgetin siRNA treated axons. Axon length for fidgetin siRNA treated crossing axons were statistically significantly longer (mean rank

= 12.40) than control siRNA axons (mean rank = 4.80), U = 9, z = -2.488, p = 0.013.

Therefore, when the axons from knockdown neurons did cross, their growth rate was over three times greater than control axons that crossed into aggrecan, and this effect was statistically significant (Fig. 4F).

Discussion

Using adult primary DRG cultures, a broadly accepted in vitro model for evaluating cell biological hypotheses relevant to nerve regeneration, we set out to test whether partial knockdown of fidgetin has potential for providing therapeutic benefit. We took the nanoparticle approach both out of necessity (because adult DRG cultures do not transfect well by conventional methods) and because the approach can be translated to future in vivo and clinical work. The results on the adult DRG neurons were consistent with those on the fetal cortical neurons transfected by nucleofection,27 but the level of knockdown and the phenotype were somewhat more modest, with 19% increase in microtubule mass per unit length of axon, rather than 62%. Growth rates of axons were increased accordingly, and this was the case whether the axon grew on a favorable substrate or an unfavorable one composed of a growth-inhibitory protein of the glial scar tissue associated with spinal cord injury. Drug studies and tubulin acetylation studies confirmed that, like our earlier work on fetal cortical neurons, the increase in microtubule mass was due specifically to an increase in the labile microtubule fraction. This is consistent with fidgetin normally paring back the labile fraction by targeting regions of microtubules that are rich in tubulin that has not been post-translationally acetylated (Fig. 6). A greater level of knockdown theoretically could be achieved by manipulating the composition of the nanoparticles, but in fact, the best therapeutic is probably one that lowers the relevant protein modestly so as not to completely impede the normal work of that protein.

We posit that modest fidgetin knockdown offers a potentially superior microtubule-based approach for augmenting nerve regeneration than other microtubule-based approaches that have recently been tried. One of the challenges for the regenerating axon is navigation to its appropriate target tissue. Most treatments that enable axons to overcome inhibitory factors would also be detrimental to appropriate axonal navigation. This includes microtubule-stabilizing drugs as well as inhibitors of kinesin-5, a molecular motor that imposes growth-regulatory forces on the axon.42 However, axons navigate the best when they are richly endowed with labile microtubules extending into their distal regions,11,43 and thus fidgetin inhibition may provide an advantage over these other approaches in terms of enabling the axon to grow rapidly through inhibitory environments while being primed to navigate to its appropriate target.

At present, the most common experimental method for pursuing knockdown work in an animal model is to introduce RNA interference via a plasmid driven by a viral transduction system. Our nanoparticle approach may be a superior option; it is non-toxic, effective on cells, tissues and organs, and is controllable. The composition of the particles can be varied so that they release their load at different rates, so that treatment regimens can be refined accordingly. For example, it may be desirable to knock down a microtubule-related protein for a window of time that would enable a regenerating axon to grow through the glial scar, but only during the time required for that to happen. Such an approach may provide new hope for kinesin-5 as a target, if it could be knocked down only during a brief window of time so that its knockdown would not impede axonal navigation. Fidgetin, on the other hand, could be knocked down for longer periods of time through slower time-release of the

23 siRNA from the nanoparticles because its knockdown would presumably assist in axonal navigation. Also, unlike the viral approach, which continues to express indefinitely, the nanoparticles would mitigate adverse off-target effects because such effects would be reversible after the siRNA load has been fully released and degraded.

Thus, in conclusion, we present fidgetin as a promising therapeutic target for nerve injury

(Fig. 5), and we present a flexible nanoparticle-based platform as a promising mode of delivery for the siRNA (Fig. 1A).

Methods

RNA interference. A pool of siRNA consisting of four independent non-overlapping sequences for rat fidgetin was used in our previous study on rat fetal cortical neurons.27 In that study, because of the lack of effective fidgetin antibodies, efficiency of knockdown was assessed by Western blot analysis of levels of ectopically expressed GFP-fidgetin, using a GFP antibody. Appropriate control experiments were conducted on the specificity of the siRNA pool, including confirmation that the identical phenotype was obtained when the four siRNA sequences were used individually. Here we used the same siRNA pool as in the previous work, in order to capitalize on the extensive controls done in that study. For preparation of the fidgetin siRNA-fused nanoparticles (or control siRNA-fused nanoparticles), a volume of five hundred microliters of Tetramethyl orthosilicate (TMOS) was hydrolyzed in the presence of 100 μl of 1 mM HCl by sonication on ice for 15 minutes, until a single phase formed.41 The hydrolyzed TMOS (100 μl) was added to 900 μl of 20

μM of siRNA solution (either control or fidgetin pool) containing 10 mM phosphate, pH

7.4. The resulting gel, which formed within 10 minutes, was frozen at −80 °C for 15 minutes and lyophilized. siRNA-fused nanoparticles that had been re-suspended in water and sonicated were added to the culture medium at 2 μl/ml. Efficacy of fidgetin knockdown was ascertained by quantitative PCR and also qualitatively by IF-staining with a fidgetin antibody that we have found to be effective for IF (see below). Optical sections of identical thickness were obtained with a confocal microscope to compare fidgetin IF-staining levels in neuronal cell bodies. Some batches of siRNA-encapsulated nanoparticles proved ineffective at knockdown, and such batches were discarded.

Cell culture. Primary cultures of DRGs were prepared by a modification of our previously published method.42 DRGs were dissected from spinal cords of adult female rats (using protocols approved by Drexel University’s IACUC, and consistent with NIH regulations) and then exposed to 0.25% collagenase for 1 hour, followed by 0.25% trypsin for 15 minutes. Ganglia were then rinsed of the enzymes using Neurobasal A with 1% fetal bovine serum (FBS). Cells were suspended in culture medium containing Neurobasal A, B27,

NGF, Glutamax, and Pen/Strep, as previously described,42 and plated onto glass-bottomed

35 mm dishes in culture medium. The cell suspension was poured over a microsieve to remove debris. Prior to plating the cells, the glass-bottomed wells of culture dishes in which a 1 cm hole was covered with a glass coverslip were treated with poly-D-lysine, as previously described.27,42 DRG cultures were plated denser (10,000 cells/well) for microtubule analyses and sparser (2500 cells/well) for morphological analyses.

Sample preparation. On the third day of culture, cultures were prepared for IF or qPCR.

For IF on fidgetin, cultures were not pre-extracted but rather directly fixed with paraformaldehyde, then extracted in buffer containing TrixonX-100, and IF-stained with a commercially-available antibody (termed SC68343, obtained from Santa Cruz). For IF on microtubules, cultures were pre-extracted for 4 minutes in a microtubule-stabilizing buffer containing TritonX-100 to release free tubulin, and then fixed in a solution containing both paraformaldehyde and glutaraldehyde as previously described.17 For qPCR, cultures were not pre-extracted and RNA was isolated using the RNAqueous-Micro Kit (AM1931) from

Ambion.27,42,44 qRT-PCR was performed as previously described.45 Quantification of microtubule levels was conducted for IF using an antibody to β-III-tubulin (termed MMS-

435P, obtained from BioLegend), which is neuron-specific. In some experiments, cultures were double-labeled for acetylated tubulin (with an antibody termed 6-11b-1, obtained from Sigma) as well as β-III-tubulin. β-III-tubulin IF-staining was used for both microtubule quantification experiments as well as for morphological analyses on axon length and number. This was especially helpful for the DRG cultures, which are dominated by non-neuronal cells that can make distinguishing axons otherwise problematic.

Microtubule quantification, stability and acetylation analyses. For quantification of microtubule levels in neurons, cultures were exposed for 0, 0.5, or 2 hours to nocodazole

(2 mg/ml) or DMSO (vehicle control). For studies on the functional relevance of microtubule acetylation, some cultures were treated with tubacin (10 μM) or DMSO

(vehicle control) during the second and third days of culture, and then subjected to morphological analyses. Most imaging was conducted using a Zeiss Observer microscope,

100X oil objective (for microtubule quantification) or 40x oil objective (for morphometry),

Axiocam CCD, and Zen Blue software, except for confocal imaging, which was conducted with a Zeiss Pascal confocal microscope. For the microtubule quantification experiments, approximately 600-800 axons were imaged per dish (3 dishes per treatment condition).

Using ImageJ, a region of interest (ROI) was traced around the axon and thresholding was performed to remove background signal and quantify the average mean gray value of fluorescence (0-255). Fluorescence intensity from the single-label β-III-tubulin IF-staining was calculated per unit length of axon. In other studies, for control and fidgetin siRNA nanoparticle treated cultures, the ratio of fluorescence intensity for acetylated to total tubulin was acquired for the axonal shaft, at least 50 μm from the cell body or axon tip as well as in the distal region of the axon, within 15 μm of the tip, by previously described methods.27 Approximately 60 measurements were taken per dish, 3 dishes per treatment.

In other experiments, the fluorescence intensity of acetylated tubulin was expressed as a ratio to total tubulin, as an independent indicator of microtubule stability.

Morphological analyses. Morphological analyses were conducted on cultures treated with control siRNA nanoparticles or fidgetin siRNA nanoparticles, with or without tubacin. The following parameters were quantified: axon number per cell body, total axon length per cell body, average axon length per cell body, and length of the longest axon per cell body.

In a separate set of experiments, the ability of axons to cross from a laminin substrate onto aggrecan was assessed, using a modified version of a previously reported assay.42,46 In brief, glass-bottomed dishes were coated with 0.1 mg/ml poly-D-lysine, rinsed thoroughly with water and allowed to air-dry. Strips (1 mm x 1 cm) of filter paper that had been soaked

27 with 3 μl of aggrecan solution (prepared 75 μg/ml in water) were placed in the dried wells, and each strip was then allowed to dry, leaving stripes of aggrecan attached to the substrate upon the strip’s removal. Dishes were then coated with 10 μg/mL laminin. Adult DRG neurons were plated in thin lines of 250 cells on either side of the aggrecan stripes.

Immediately after settling, cells were treated with medium containing either control or fidgetin siRNA nanoparticles. Cells were fixed after 48 hours, IF-stained for CS-56

(antibody to aggrecan, Sigma) and β-III-tubulin, and analyzed for axons that had approached within 10 μm of the stripes.42 Axons that had turned so that a line drawn from the tip of the axon would point away from the stripe were defined as “not crossed,” whereas axons that had grown over the aggrecan border were counted as “crossed.” Axons grown from cell bodies that had landed within 20 μm of an aggrecan stripe or on the stripe itself were excluded.

Statistics. Statistical analyses were conducted using IBM SPSS 24 and detailed with the corresponding data in the results section. Data are mean ± standard deviation, unless stated otherwise. All data was checked for normality using the Shapiro-Wilk test. Parametric data was assessed using the t-test to compare means and non-parametric data utilized the Mann-

Whitney U test to compare medians. Two tailed tests were performed and all dependent values were continuous with dichotomous independent variables. A Chi square test was performed to test for an association between categorical data.

Figure 1. Validation of siRNA nanoparticles in adult DRG primary cultures. (A) Schematic detailing the production of siRNA hydrogel nanoparticles, nanoparticle entry and release into the cell, and siRNA targeting fidgetin mRNA. (B) Nanoparticles containing alexa488 were added to culture medium and a representative merged image of bright-field and green-channel fluorescence is shown; large diameter neurons displayed more uptake of nanoparticles compared to other cells.

Uptake is more visible in clusters of neuronal cell bodies. (C) qRT-PCR was performed on samples of RNA from siFIGN-nanoparticle treated DRG primary cultures; data from two separate primers are shown in the bar graph. (D) Representative fidgetin IF images (confocal z-stack images) of

DRG neuronal cell bodies indicating knockdown of fidgetin from neurons treated with siRNA nanoparticles.

Figure 2. Microtubule mass is increased in the axons of adult DRG neurons as a result of fidgetin knockdown. (A) Representative images of neuronal β-III-tubulin IF-staining show more extensive axon outgrowth after fidgetin knockdown compared to control, and also show denser microtubule mass within axons, and indicated by more intense IF-staining. Shown are inverted images with black and white reversed for enhanced clarity. (B) Representative IF displayed with quantitative pseudo color (standard fire-scale where purple is the least, white is the most, with shades of orange and red between them) to further accentuate the increased microtubule mass in axons of siFIGN

30 nanoparticle-treated neurons. Additionally, (B) shows that after 30 minutes or 2 hours in nocodazole, there is no discernable difference in microtubule levels between fidgetin and control siRNA. (C) Quantification of axonal tubulin IF shows a 19% increase in siFIGN nanoparticle treated axons compared to controls, Mann-Whitney U test, * = p<0,005. (D) Standard deviation of tubulin IF 8-bit grayscale values, higher when the labile fraction is intact and lower when it has been depolymerized, is consistent with the labile domains of the microtubules being more dynamic than the stable domains.

Figure 3. Fidgetin knockdown decreases ratio of acetylated to total tubulin in the axons of DRG neurons. (A,B), IF-staining for β-III-tubulin (green) and acetylated tubulin (red), shown as overlays of the two colors. Ratio of acetylated tubulin to β-III-tubulin (total tubulin) was significantly decreased as a result of fidgetin knockdown (A,B,C). The acetylated/total tubulin ratio of both treatments increased after tubacin treatment, with the control and fidgetin siRNA becoming statistically indistinguishable.

Figure 4. Fidgetin knockdown positively affects axon outgrowth in a manner dependent upon unacetylated tubulin. Cultured adult DRG cells were treated with control (A) or fidgetin-siRNA nanoparticles (B), in combination with vehicle (DMSO) or tubacin. Shown in the panels is IF- staining for β-III-tubulin. Fidgetin knockdown results in a significant increase in average axonal growth compared to control, while tubacin-siFidgetin showed no improvement over tubacin alone

(C).

Figure 5. Effects of fidgetin knockdown on axonal growth onto an inhibitory substrate. Cultured adult DRG neurons growing on laminin and treated with either fidgetin or control siRNA nanoparticles were challenged with an aggrecan border. Cultures were IF-stained for β-III-tubulin

(green) and aggrecan (red), as shown in panels A and B, or IF-stained for β-III-tubulin and fidgetin, as shown in panels C-D’. Axons were scored as a “cross” (A) or “no-cross” (B). Images were brightened above saturating levels to aid in border identification. (C,D) Growth cones of β-III IF- staining for siControl- and siFidgetin-treated cultures, respectively. Growth cones with siFIDG treatment display elongation. (C’,D’), Fidgetin IF-staining for siControl- and siFidgetin-treated growth cones, respectively. (E), No significant difference was found in percent axon crossing between the two groups (Chi-square, p>.05). However, among those axons that did cross, axonal growth was significantly greater (p<.05) in fidgetin siRNA cultures compared to control (F).

Figure 6. Schematic illustration of how fidgetin knockdown boosts axon regeneration. Fidgetin severs axonal microtubules in the labile domain of the microtubule, and thus has the function of paring back the labile domains, tamping back their length. Experimental partial depletion of fidgetin enables elongation of the labile domains. Labile domains are especially enriched in the growth cone, and hence their elongation promotes invasion of microtubules into filopodia, which in turns promotes greater axonal growth, even through inhibitory molecules associated with nerve damage. Theoretically, axons with greater invasion of labile microtubule domains into their growth cones should be better equipped to navigate to their targets, relative to their injured state or relative to injured axons treated with microtubule-stabilizing drugs.

CHAPTER 2: TAU DOES NOT STABILIZE AXONAL MICROTUBULES BUT RATHER ENABLES THEM TO HAVE LONG LABILE DOMAINS

Introduction

It is widely believed that tau stabilizes microtubules in the axon 47-49, and hence that disease-induced loss of tau from axonal microtubules leads to their destabilization 22,49,50.

An individual microtubule in the axon has a stable domain and a labile domain 17,18,51. We found that tau is more abundant on the labile domain, which is inconsistent with tau’s proposed role as a microtubule stabilizer. When tau is experimentally depleted from cultured rat neurons, the labile microtubule mass of the axon drops considerably, the remaining labile microtubule mass becomes less labile, and the stable microtubule mass increases. MAP6 (also called stable tubule-only polypeptide), which is normally enriched on the stable domain 52, acquires a broader distribution across the microtubule when tau is depleted, providing a potential explanation for the increase in stable microtubule mass.

When MAP6 is depleted, the labile microtubule mass becomes even more labile, indicating that, unlike tau, MAP6 is a genuine stabilizer of axonal microtubules. We conclude that tau is not a stabilizer of axonal microtubules but is enriched on the labile domain of the microtubule to promote its assembly while limiting the binding to it of genuine stabilizers such as MAP6. This enables the labile domain to achieve great lengths without being stabilized. These conclusions are contrary to tau dogma.

Results and Discussion

Tau depletion causes preferential loss of the labile microtubule fraction in the axon. An early indication that tau may not be responsible for the stability of axonal microtubules was reported almost twenty years ago, when cultured neurons were microinjected with a tau antibody prior to axon outgrowth. In that study, axons free of tau (which was trapped in the cell body) were then allowed to grow for a few hours, and these axons contained microtubules that were no less stable than controls 53. Here we delved deeper into the issue by comparing microtubule levels in the axons of cultured rat neurons after four days of tau depletion by RNA interference. Unlike those earlier studies which were only performed on cultured rat sympathetic neurons from the superior cervical ganglia (SCG), our studies were conducted on both rat cortical neurons and SCG neurons (with similar results obtained with both types of neurons). For this, we used siRNA introduced by nucleofection, which results in transfection of over 90% of the neurons in these cultures 54. Depletion of tau over those four days was confirmed by Western blot, with most of the tau depleted by the first day and no restoration in tau levels by the fourth day (Fig. 7A). On the fourth day, cultures were prepared for various microtubule-related analyses. As in previous studies using this approach on rat brain neurons 55-57, there was no obvious change in axon length or morphology with tau depletion with either type of neuron, except that branching frequency was slightly higher in tau-depleted axons of cortical neurons. Microtubule levels were assessed by quantitative immunofluorescence, comparing axons of neurons into which we introduced control siRNA or tau siRNA. Per unit length, the axons of tau- depleted cortical neurons displayed roughly 25% less microtubule mass than their control

38 counterparts (Fig. 7B and C), and roughly the same was true of SCG neurons (data not shown).

We next wished to investigate whether the microtubule loss due to tau depletion is chiefly of the stable domains of the axonal microtubules or their labile domains. For this, we exploited previous findings that the labile domains of axonal microtubules have much higher levels of tyrosinated tubulin than the stable domains 17,18,51. Detyrosination is a post- translational modification of alpha tubulin that accumulates in the microtubule over time, and hence becomes concentrated in the stable domain, although it is also present at lower levels in the labile domain 3,18. Levels of tyrosinated or detyrosinated tubulin can also vary in the labile domain, reflecting differences in how labile a labile domain is, with very labile ones containing fewer detyrosinated subunits than less labile ones 58. Conventional immunostaining revealed 25% higher levels of detyrosinated tubulin per unit length of axon (not just the percentage relative to total, but the absolute amount) in tau-depleted axons compared to control axons (Fig. 7B and D).

To delve into this further, we prepared cultures for immunofluorescence double-label visualization of total tubulin and tyrosinated neurons using an extraction/fixation protocol that causes microtubules to splay apart so that labile domains can be visualized in especially splayed regions along the axon’s length 51. [Particular care was taken to compare equally splayed regions, because the microtubule array along the tau-depleted axon was generally less compact than in control axons, consistent with a potential role for tau as a

39 microtubule bundler; see Figure S1A and S1B]. The number of labile domains (i.e., microtubule profiles rich in tyrosinated tubulin) visualized in this fashion per unit length of axon was reduced in the tau-depleted axons compared to control counterparts by 24%

(Fig. 7E and F).

Taken together, these results indicate that tau depletion results in the net loss of microtubule mass from the axon, and that the loss of microtubule mass is predominantly of the labile domains rather than the stable domains of the microtubules. In addition, there is an increase in microtubule stability in the remaining microtubule mass, relative to control, that is not explicable on the basis of the shortening of the labile domains. The increase in detyrosinated tubulin suggests either that the remaining portions of the labile domains are becoming less labile or that the stable domains are becoming more stable, or both.

For a rescue experiment, GFP-tagged full-length tau (human tau sequence, not affected by siRNA to rat tau sequence) was expressed in rat cortical neurons that had been depleted of tau, in order to ascertain whether the stability properties of the microtubules were restored to normal. We did this to confirm the specificity of the siRNA, but also because a number of previous studies have shown that ectopically expressing tau in non-neuronal cells results in greater microtubule stability 57,59-61, whereas our expectation is that restoration of tau to tau-depleted neurons should result in a decrease of microtubule stability (Fig. 7B).

Consistent with our expectation, restoration of tau to control levels resulted in a decrease in the levels of detyrosinated tubulin back to control levels, and an increase in total

40 microtubule levels back to control levels (Fig. 7C and D). These results confirm the specificity of the siRNA and also suggest that the previously published results on ectopic tau expression in non-neuronal cells showing enhanced microtubule stability do not reflect the actual role of tau in the axon.

Tau is enriched on labile domains of axonal microtubule relative to stable domains. The distribution of tau in growing axons is another hint, already in the literature, that tau’s role in the axon is contrary to dogma. Specifically, tau is enriched in the distal region of the axon, where labile domains of microtubules are known to predominate 31,62. However, several other proteins also concentrate in the distal region of the axon 63-65, so it remains unclear whether the distal enrichment of tau is a specialization of that region of the axon, or a manifestation of tau’s enrichment on labile domains of microtubules everywhere along the axon. In support of the latter, we found a high degree of correlation between tau staining and tyrosinated tubulin staining in the axons of both SCG (data not shown) and cortical neurons (Figure S1C and S1D). However, we could not use the splay technique to resolve the issue with finer resolution because the extraction procedure for that technique strips tau from the microtubules. Therefore, we used an indirect approach (Fig. 7G-R).

We reasoned that if there is more tau on labile domains than stable domains, then after 15 minutes in nocodazole (a drug that binds free tubulin subunits, thus inducing microtubule depolymerization) the levels of microtubule-associated tau in the axon should display a greater decrease than the decrease in microtubule mass. Our sample preparation procedure was previously optimized to release as much of the free tau as possible without stripping tau from the microtubules 31,66, but to further optimize staining for only tau that is

41 microtubule-associated, we used the tau1 antibody, which recognizes tau that is not phosphorylated at sites that when phosphorylated induce detachment of the tau from the microtubule 67-69. Similar results were obtained with generic tau antibodies, such as tauR1 and tau5 (data not shown).

Shown in Fig. 7P are the results with SCG neurons. Similar results were obtained with cortical neurons (data not shown). After 15 minutes in drug, there was a 70% diminution of microtubule-associated tau, while there was only a 50% decrease in microtubule mass

(Fig. 7H,K,N and P). This suggests that 70% of the microtubule-associated tau in the axon is normally on the labile domains of the microtubules while only 30% is on the stable domains.

To pursue this further, we conducted an experiment in which the nocodazole was washed from the culture to permit microtubule reassembly. Microtubule reassembly occurs from the plus ends of the stable domains of microtubules that remain after the nocodazole treatment 70. Reassembly was allowed to occur for just 3-5 minutes, in order that the newly assembled labile domains remain relatively short, so that they appear as short segments in the axonal shaft in immunostain analyses of tyrosinated tubulin 70. If tau is concentrated on labile domains relative to stable domains, we would expect a high degree of correlation between these fluorescent microtubule segments and tau immunoreactivity, which is exactly what we observed (Fig. 7I,L,O,Q to S).

Depletion of either tau or MAP6 causes changes in the other. Despite a 25% reduction in total microtubule mass in tau-depleted axons, there was a 25% increase in detyrosinated tubulin relative to control axons (Fig. 7B-D), suggesting that whatever protein(s) are stabilizing the stable domain of the microtubule take on a wider distribution along the microtubule mass of the axon in the absence of tau. Previously, it was reported that stable domains have more microtubule-associated-protein-6 (MAP6) (also called stable tubule- only polypeptide) bound to them than labile domains 52, prompting us to posit that MAP6 might be such a protein. Fortifying this speculation are studies on other cell types showing that MAP6 is mostly unbound to microtubules until the cells are treated with cold, after which significantly more MAP6 binds to the microtubules, thereby enhancing their stability 32. In addition, biochemical studies indicate that MAP6 can diffuse or “slide” along the microtubule 71, consistent with its potential to spread out along the microtubule after tau depletion, and thereby increase the length of the stable domain relative to the entire microtubule. Finally, MAP6’s ability to stabilize axonal microtubules has recently been confirmed by experimental studies on neurons 72. Thus, MAP6 is an appealing possibility for a MAP that can affect microtubule stability by attaching/detaching to the microtubule or sliding along its length in response to other events that might occur, such as tau depletion.

Initially, we performed Western blotting on the cortical cultures, which demonstrated that depletion of MAP6 significantly increased the levels of tau by 29%, while depletion of tau significantly increased the levels of MAP6 by 54% (Fig. 8A-D). We then ascertained whether depletion of either tau or MAP6 results in a change in the distribution of the other.

Shown in Figure S1C are cortical neurons double-labeled for immunofluorescence visualization of tyrosinated tubulin and microtubule-bound tau. When MAP6 was depleted by siRNA, tau took on a more widespread distribution in the axon (Figure S2A-I). In the case of tau depletion, we focused on the distal region of the axon contiguous with the growth cone, which is especially rich in labile microtubule domains and normally enriched with microtubule-bound tau. Immunofluorescence double-labeling for tubulin and MAP6 indicates that these microtubules are normally deficient in MAP6. However, in the case of tau-depleted axons, the microtubules along the axon and the growth cone become rich in

MAP6 (Fig. 8E-J). This same phenomenon was observed in the lamellipodia that extend from the axonal shaft, into which microtubules splay (Figure S3A-D). Thus, both the levels and distributions of these two proteins are intimately related.

At present, we do not know if MAP6 is unique with regard to the present observations, or if other stabilizers of microtubules have a similar relationship with tau. We reasoned that doublecortin might be a likely candidate, because doublecortin is normally enriched in tau- deficient regions of the axon 66, and because tau redistributes to bind microtubules previously bound to doublecortin when doublecortin is depleted 66. However, we found that microtubules do not become rich in doublecortin when tau is depleted (Figure S4A-

F).

Depletion of MAP6 has roughly opposite effects on microtubule stability as tau depletion.

To ascertain whether the growth cone microtubules of neurons depleted of tau were made more stable by MAP6 binding to them, we treated the cultures with nocodazole for just 10 minutes to deplete much of the labile microtubule mass while not causing notable axon retraction. We focused on stalled growth cones in the cortical neuronal cultures because they have a mixture of stable and labile microtubule domains, more similar to the axonal shaft than the growth cone of an elongating axon 73. In control neurons, about 53% of the microtubule mass in stalled growth cones was lost as a result of the drug treatment (Fig.

8K, L and Q). However, in the case of stalled growth cones of tau-depleted axons, almost none of the microtubule mass was lost (Fig. 8M, N and Q), indicating that the microtubules had become more stable in the absence of tau. In the case of stalled growth cones of MAP6- depleted axons, about 80% of the microtubule mass was lost from the growth cone as a result of the nocodazole treatment (Fig. 8O, P and Q), consistent with MAP6 being a genuine stabilizer of microtubules in the axon. [All of the microtubule mass quantifications were normalized to the growth cone areas. Growth cones are smaller in MAP6-depleted neurons compared to control or tau-depleted neurons (Fig. 8O-P and R)].

We further pursued changes in microtubule stability upon the depletion of tau or MAP6, next focusing on the axonal shaft. In a parallel study to the one we performed on tau- depleted axons (Fig. 7C, 1D and 1E), we double-labelled MAP6-depleted neurons for detyrosinated tubulin and βIII-tubulin. MAP6 depletion resulted in a 27% decrease in total microtubule mass per unit length of axon (Fig. 9A and B), which is just slightly higher than the 25% decrease observed with tau depletion. However, unlike tau depletion, which

45 resulted in a 25% increase in detyrosinated tubulin staining per unit length of axon, MAP6 depletion resulted in a 30% decrease (Fig. 9A and C). Western blots of cortical neuronal cultures depleted of either tau or MAP6, pre-extracted to release free tubulin, were entirely consistent with these immunofluorescence-based results (Fig. 9D-F).

We pursued an additional set of studies to investigate whether tau depletion causes the labile domains of microtubules to become less labile, while MAP6 depletion causes the labile domains to become more labile. Being more or less labile would mean more or less dynamic, as observed in live-cell imaging 3. We examined neurons expressing GFP-labeled end-binding protein-3 (EB3), in which excursions of microtubule assembly display as movements of fluorescent “comets” 3. Tau depletion substantially decreased the number of comets in the axon, consistent with a reduction in the number of labile domains that exist at any moment in time (Fig. 9G-I). MAP6 depletion did not alter the number of comets, consistent with labile domains becoming longer than in controls but not more numerous

(Fig. 9G and H). With regard to comet velocity, tau depletion caused a decrease, while

MAP6 depletion caused an increase (Fig. 9G and I), which is consistent with the labile domains being less labile after tau depletion and more labile after MAP6 depletion.

The observation of notably fewer EB3 comets in the tau-depleted axons is noteworthy in light of the fact that the stable domain is just as capable of elongating a new labile domain as a labile domain is of simply getting longer 70. Theoretically the tau-depleted axons and the control axons should have an equal number of EB3 comets unless the presence of tau

46 is having an additional effect, namely to promote the assembly of the labile domain. These results therefore indicate that labile domains are able to become long not only through tau’s capacity to prevent them from being stabilized but also through tau’s capacity to promote their elongation. Interestingly, when tau was first discovered, it was touted mainly as a promoter of microtubule assembly 74.

The fact that expression levels of tau and MAP6 are positively correlated instead of negatively correlated seems contrary to the expectation of their inverse functions. We are interested in exploring this further, especially in light of the potential implications for tauopathy.

Quantitative analyses on microtubule fractions in the axon assessed by rates of drug- induced depolymerization. In a final set of studies, we quantified the rates of microtubule disassembly induced by nocodazole. This strategy is based on the premise that nocodazole depolymerizes microtubules at rates reflective of the dynamic properties of the microtubules. For this, we used quantitative immunofluorescence to ascertain microtubule levels after various times in the drug, ranging from 5 to 60 minutes (Fig. 10A and B).

Previous studies established that most of the labile microtubule fraction of the axon depolymerizes by 15 minutes in nocodazole, while the stable fraction is diminished much more slowly over a period of hours 17,18,27,75. This manifests as a biphasic decay in microtubule mass when the data are graphed, with the first phase displaying a far sharper decline than the second phase. We previously reported that in the case of SCG neurons,

47 roughly 50% of the microtubule mass of the axon is labile and the other half is stable 17,18.

Here we show that in the control axons of rat cortical neurons, roughly ~64% of the microtubule mass is labile, while ~36% is stable, while in tau-depleted axons (in which roughly a fourth of the microtubule mass is lost, relative to control), ~38% of the microtubule mass is labile, while ~62% is stable. In addition, the decline of the first phase of the biphasic decay is less steep in the case of the tau-depleted axons compared to control axons, which is consistent with the labile domains of the microtubules being less labile

(i.e., more stable) after tau depletion. In the case of MAP6-depleted axons, there is a 27% reduction in total microtubule mass, but the proportions of stable and labile are not obviously different from control. However, the decline of the first phase of the biphasic decay is steeper than controls, which is consistent with the labile domains being more labile after MAP6 depletion. These changes are remarkably consistent with the results observed with the EB3 comets. The decline of the second phase was not detectably different from controls in the case of either tau or MAP6 depletion, suggesting that stabilizers other than

MAP6 are sufficient for the stable domain (while shorter in the absence of MAP6) to remain just as stable. A more detailed mathematical treatment of the data, including t1/2 values for the phases of decay is provided in the legend of Fig. 10.

These drug results combine with the other results presented here to indicate that: (i) after tau or MAP6 depletion, there is a net loss of microtubule mass from the axon; (ii) after tau depletion, the microtubules in the axon have shorter labile domains but longer stable domains; (iii) after MAP6 depletion, the microtubules in the axon have shortened both their stable and labile domains to achieve roughly the same proportion as in control axons; (iv)

48 the labile domains are less labile after tau depletion but more labile after MAP6 depletion.

These results are not consistent with the dogma that tau is responsible for stabilizing axonal microtubules, but rather suggest that tau enables the labile domains to grow longer than they otherwise would, without being stabilized. By contrast, MAP6 is a genuine stabilizer of axonal microtubules that regulates the properties of the labile domain in a manner opposite to tau.

How tau outcompetes genuine stabilizers for the microtubule lattice is unknown, but a simple competition for the same binding site on the microtubule seems unlikely. Tau is known to come on and off the microtubule at rapid rates of milliseconds 30, which may contribute to how it keeps genuine stabilizers from accessing the microtubule lattice. We posit that the means by which a labile domain or a portion of it becomes stabilized involves a signaling cascade that phosphorylates tau so that it detaches from the microtubule, thus allowing MAP6 and other stabilizing proteins to bind. Modifications of the stabilizing proteins may contribute as well 72.

Concluding remarks. Abundant data from test tube and overexpression studies indicate that tau has the capacity to stabilize microtubules 47,59,61,76, but whether it does so in the physiological context of the axon has undergone surprisingly little testing. There have been hints in the literature that tau is not a stabilizer of microtubules in the axon 3,19,30,31,53,62, but for the most part, these hints have not been pursued and have not impacted the field.

Hundreds of research papers, websites, and instructional materials continue to describe tau

49 as an important stabilizer of axonal microtubules 77,78, with this idea taken so seriously by the biomedical community 22,50 that microtubule-stabilizing drugs have been in clinical trials for diseases for tau-based neurodegenerative diseases 79. Based on our data, we posit that the real manner by which tau contributes to the stability properties of axonal microtubules is to permit them to have labile domains that can become long without being stabilized (Fig. 10C). Tau also promotes the elongation of the labile domains, and thereby supports the existence of a robust labile microtubule fraction in the axon throughout the life of the neuron.

Materials and Methods

Neuronal culture and siRNA-based depletion of tau or MAP6

All animals were used under the approval of Drexel University IACUC guidelines.

Cortical neurons were dissected from frontal cerebral cortex of 19-day Sprague Dawley rat fetuses of either sex, and then cultured as previously described 55. Superior cervical ganglia

(SCG) were acquired from P0-P2 Sprague Dawley rat pups of either sex and cultured as previously described 80. Control scrambled siRNA (Sigma, SIC001), tau siRNA (4 sequences of siRNAs of tau were combined for use; Sigma, SASI_Rn01_000548888,

SASI_Rn01_00054889, SASI_Rn01_00054890 and SASI_Rn02_00261575) or MAP6 siRNA (3 sequences of siRNAs of MAP6 were combined for use; Sigma,

SASI_Rn01_00121264, SASI_Rn01_00121265, SASI_Rn01_00121266) were delivered by nucleofection (NucleofectorTM 2b, Lonza) into neurons prior to plating. For all nocodazole studies, experiments are done on the fourth day after plating. The GFP-EB3

50 experiments were conducted on the third day after plating. For all other experiments, the cells were cultured for 2-3 days (to allow for siRNA-targeted proteins to be depleted) on

35 mm diameter petri dishes coated with 1 mg/ml poly-L-lysine (Sigma, P2636-25MG), and then re-plated for two days prior to experiments on glass-bottomed dishes (Cellvis,

#D35-14-1.5-N) coated with 1 mg/ml poly-L-lysine.

Immunofluorescence and Microscopy

For most experiments, cultures were co-extracted and fixed for 10 minutes in a solution containing 4% paraformaldehyde, 1x PHEM buffer (PIPES, HEPES, EDTA, MgCl2), 0.2% glutaraldehyde, and 0.1%Triton X-100. Glutaraldehyde was then quenched by treatment of the cultures twice for 15 minutes each with 2 mg/mL sodium borohydride. Cultures were then blocked for 1 hour with normal goat serum (Jackson ImmunoResearch #005-

000-121) and 10% bovine serum albumin (BSA) (Sigma#A7906-100G), followed by incubation with primary antibodies overnight at 4°C and then secondary antibodies for an hour at room temperature. For the nocodazole-based assay for measuring stable and labile microtubule fractions, the cultures were pre-extracted with a 0.05% Triton X-100 microtubule-stabilizing buffer 27 for 90 seconds (to release free tubulin), prior to fixation with the same fixation solution indicated above but without Triton X-100. For MAP6 immunostaining, cultures were fixed in ice-cold methanol for 8 minutes (without any pre- extraction), followed by the same procedure as above, but without the sodium borohydride step.

Primary antibodies were: rabbit anti-βIII-tubulin (1:1500; Biolegend #802001), mouse anti-βIII-tubulin (1:1500; Biolegend #801202), rat anti-tyrosinated tubulin (1:1500;

Millipore, MAB1864); rabbit anti-detyrosinated tubulin (1:1500; Millipore AB3201), rabbit anti-laminin (1:400, Sigma #L9393-2ML), rabbit anti-doublecortin (1:1000; Abcam,

#ab18723), rabbit anti-MAP6 (1:800) 81. Appropriate secondary antibodies were purchased from Jackson ImmunoResearch.

Microtubule splay assay

An immunofluorescence preparation procedure previously reported to splay apart the microtubules along the axonal shaft of SCG neurons 51 was modified for use on cortical neuron axons. For this, cortical neurons were re-plated on glass-bottomed dishes coated with 1 mg/ml Poly-L-lysine at a density of 8,000 cells per dish. The culture was washed once with warm 1x PBS and once 1x PHEM buffer (PIPES, HEPES, EDTA, MgCl2) and then exposed for 3 minutes to pre-extraction medium (1x PHEM buffer, 0.2% Triton X-

100, protease inhibitors, 10 μM taxol (Cytoskeleton, #TXD01). The cells were then fixed for 10 minutes with the same fixation solution indicated above but without Triton X-100.

Cells were stained for βIII-tubulin and tyrosinated tubulin, and various parameters were quantified using ImageJ and Zeiss blue edition software including number of individual microtubule profiles per 100 μm of the axon, percentage of the individual microtubule profiles double positive for βIII-tubulin and tyrosinated tubulin and average width of the axons.

Nocodazole-based assay for measuring stable and labile microtubule fractions

Cultures were treated for 5, 15, 30 or 60 minutes with DMSO (control; vehicle used to dissolve nocodazole) or 2 μg/mL nocodazole (Sigma# M1404-2MG). Immediately following nocodazole treatment, cells were pre-extracted and prepared for immunofluorescence staining for βIII-tubulin as indicated above. Axons were traced using

ImageJ and their intensities (averaged per unit length of axon) were plotted as percentage relative to control for various time points.

Nocodazole washout experiments

Cultures were treated for 15 minutes with nocodazole, and then the drug was washed out twice rapidly with culture medium, after which neurons were allowed to reassemble microtubules for 3-5 minutes. After preparation for immunofluorescence (see above), axons of neurons stained for tau, tyrosinated tubulin and βIII-tubulin were traced using

ImageJ and their staining intensities were plotted as percentage between nocodazole- treated and DMSO-treated (i.e., vehicle alone). Co-localization of tyrosinated tubulin and tau were represented as a scatterplot using ImageJ Coloc2 plugin.

Rescue experiment. Control siRNA or tau siRNA and MAP6 siRNA were introduced into cortical neurons by electroporation as described above. 3 days later, cells were collected by using accutase (Stem cell Technology 07920), transfected with GFP-tau-4R (human sequence; gift from Dr. Gloria Lee), by electroporation and re-plated on glass bottom

53 dishes coated with poly-L-lysine (1 mg/ml). The cultures were fixed 36 hours after re- plating and immunostained with antibodies against III tubulin and detyrosinated tubulin.

Western Blot

Protein lysates were extracted from transfected cultures using Pierce RIPA Buffer

(ThermoFisher Scientific, 89901) supplemented with protease inhibitor cocktail

(ThermoFisher Scientific, 78430) and phosphatase inhibitor Cocktail3 (Sigma#P0044-

1ML). Total protein concentration in the lysates was measured using BCA protein assay

(ThermoFisher Scientific, 23227). 20 μg of proteins was loaded per lane in 4-15% gradient

SDS-PAGE gel (Bio-RAD #456-1084), after which transfer was conducted onto PVDF membranes (Bio-RAD #162-0177) at 4oC for 16-18 hours at 30V. Membranes were then blocked for 1 hour in 5% fat-free milk (Lab Scientific # M0841) in 1xTBS-T (10x TBS

Bio-RAD #170-6435). Blots were treated with rabbit anti-tau (tauR1, 1:2000) anti-βIII- tubulin (1:20000; Biolegend#802001), mouse anti-βIII-tubulin (1:20000;

Biolegend#801202), rat anti-tyrosinated tubulin (1:500; Millipore, #MAB1864), rabbit anti-detyrosinated tubulin (1:500; Millipore ab3201), or mouse anti-GAPDH (1:5000; abcam#ab8245), followed by appropriate peroxidase-conjugated secondary antibodies.

Blots were visualized using Super Signal West Pico kit (ThermoFisher Scientific, 34580).

The intensity of the bands on the blots was measured using Image Lab (Bio-Rad).

EB3-GFP imaging

EB3-GFP DNA was co-transfected with control scrambled siRNA, tau siRNA or MAP6 siRNA, respectively. Live-cell imaging were conducted 3 days after transfection. Images were captured at an interval of 1 or 2 seconds for a total of 1-3 minutes using the Zeiss

100X/1.46 planachromat objective. The videos were quantified for the number and velocity of the comets, and kymographs were generated using Zeiss Blue edition software.

Statistical Analysis

All data analyses, statistical comparisons, and graphs were generated using Excel

(Microsoft), ImageJ, Image Lab and Zeiss blue edition software, with data represented as meanSEM of at least three separate experiments for each assay. Comparisons were performed using a two-tailed t test, one-way ANOVA or post-hoc Tukey test. For all statistical analysis, the mean difference was considered to be significant at 0.05 level.

Figure 7: Distribution of tau and effects of its depletion on axonal microtubules are inconsistent with tau’s purported role as a stabilizer of axonal microtubules.

(A) Western Blot result shows tau depletion efficiency on each day (days 1-4) after cortical neurons were transfected with tau-siRNA. Shown here is one such experiment, which was repeated thrice.

(B) Images of microtubule mass intensity and detyrosinated tubulin intensity in the axon of neurons which had been transfected with control siRNA or tau siRNA. Images are presented using fire- scale filter in ImageJ. Scale bar is 100 μm.

(C) Bar graph represents quantification of microtubule mean intensity along the length of the axon as percentage of control. (Control=1±0.1, n=14; tau siRNA=0.75±0.09, n=15; tau siRNA+GFP- tau 4R = 1.001±0.12, n=12) ∗, p < 0.05.

(D) Bar graph represents quantification of detyrosinated tubulin mean intensity along the length of the axon as percentage of control. (Control=1±0.05, n=14; tau siRNA=1.25±0.07, n=12; tau siRNA+GFP-tau 4R = 0.922±0.03, n=15) ∗, p < 0.05.

(E) Inverted immunofluorescence images of splayed individual microtubules from axons of control siRNA treated or tau siRNA treated neurons that were fixed and stained for βIII-tubulin and tyrosinated tubulin. Scale bar is 100 μm.

(F) Bar graph represents quantification of the percentage of individual microtubule profiles that splayed out from the axon that were immuno-positive for both βIII-tubulin and tyrosinated tubulin.

(Control=80%±4.5%, n=14; tau siRNA=56%±3.1%, n=15) ∗∗, p < 0.01.

(G-O) Inverted immunofluorescence images of axons from SCG neurons stained for tau and tyrosinated tubulin that had been subjected to DMSO (vehicle alone) for 15 minutes, nocodazole for 15 minutes, or nocodazole for 15 minutes followed by washout and recovery for 3-5 minutes.

(P) Bar graph represents quantification of percentage of mean fluorescence intensity of tau, tyrosinated tubulin and βIII-tubulin per unit length of axon with DMSO (vehicle alone) or nocodazole 15 minutes treatment. (DMSO: tau=100%±14%, tyrosinated tubulin=100%±9%, βIII-

57 tubulin=100%±12%, n=27; tau=31%±3%, tyrosinated tubulin=8%±0.9%, βIII-tubulin=45%±6%, n=15).

(Q) and (R) Pie graphs represent quantification of data from the experiments on axons of cultured

SCG neurons after nocodazole treatment for 15 minutes followed by washout and 3-5 minutes of recovery. (Q) is the newly assembled microtubule staining profile that stained for tyrosinated tubulin but not tau (+/-); or both tyrosinated tubulin and tau (+/+). (R) is the tau staining profile that stained also for tyrosinated tubulin (+/+); or did not stain for tyrosinated tubulin (+/-).

(S) Scatter plot represents the co-localization of tyrosinated tubulin and tau after nocodazole treatment 15 minutes with wash out 5 minutes, Pearson’s coefficient (p=0.89) was calculated using

ImageJ Coloc2 plugin.

Figure 8: MAP6 is recruited to and stabilizes neuronal microtubules of cortical neurons when tau is depleted.

(A-D) Western Blot and bar graphs show efficiency of depletion of target proteins of tau siRNA

(16%±4.0%) and MAP6 siRNA (31%±8.0%) and show increase in tau expression (129%±8.2%) when MAP6 is depleted (n=5) as well as increase in MAP6 expression (154%±2%) when tau is depleted (n=3) ∗, p < 0.05 and ∗∗, p < 0.01.

(E-H) Immunofluorescence (inverted) images of MAP6 and βIII-tubulin immunostaining in the growth cone after tau depletion. Scale bar is 10 μm.

(I) Bar graph represents quantification of the ratio of MAP6 immunostaining intensity to microtubule immunostaining intensity relative to control in the growth cone. (Control=1±0.1, n=23;

Tau siRNA=2.37±0.31, n=26) ∗∗, p < 0.01.

(J) Co-localization quantification of MAP6 and microtubule immunostaining intensities in the growth cone (Control, R=23%±5% n=10; tau siRNA, R =53%±5% n=10) using ImageJ and the

Pearson’s correlation coefficient is shown. ∗∗, p < 0.01.

(K-P) Immunofluorescence (inverted) image represents microtubule mass in the growth cone after nocodazole treatment for 10 minutes in control neurons, tau-depleted neurons and MAP6-depleted neurons respectively. Scale bar is 20 μm.

(Q) Bar graph represents microtubule mass quantification in the growth cone normalized to growth cone area, after nocodazole treatment for 10 minutes in control group, tau-depleted group or MAP6- depleted group respectively. (DMSO: Control=100%±14%, n=20, tau siRNA=107%±16%, n=20,

MAP6 siRNA=98%±21%, n=20; Nocodazole: Control=47%±6%, n=22, tau siRNA=92%±8%, n=22, MAP6 siRNA=18%±4%, n=23) ∗∗, p < 0.01.

(R) Bar graph represents quantification of growth cone size, after nocodazole treatment for 10 minutes in control group, tau-depleted group or MAP6-depleted group respectively. (DMSO:

Control=956.03±114.129, n=20, tau siRNA=868.02±89.39, n=20, MAP6 siRNA=366.3±39.28, n=20; Nocodazole: Control=918.31±113.49, n=36, tau siRNA=921.36±73.85, n=29, MAP6 siRNA=366.75±34.37, n=32) ∗∗, p < 0.01

Figure 9: Opposing effects of tau and MAP6 depletion on microtubule stability in the axon.

(A) Images of microtubule mass and detyrosinated tubulin mass in the axon of cortical neurons transfected with control siRNA or MAP6 siRNA. Images are represented using fire-scale filter in

ImageJ. Scale bar is 100 μm.

(B-C) Bar graph represents quantification of microtubule and detyrosinated tubulin mean intensity along the length of the axon as percentage of control. (Microtubule: Control=1±0.1, n=14, MAP6

61 siRNA=0.73±0.13, n=15; detyrosinated tubulin: Control=1±0.05, n=14, MAP6 siRNA=0.7±0.05, n=15) ∗, p < 0.05 and ∗∗, p < 0.01.

(D) Western blot results show microtubule mass and detyrosinated tubulin mass when tau or MAP6 is depleted.

(E-F) Bar graph represents the quantification of Western blots for microtubule and detyrosinated tubulin mass in respective siRNA transfected groups. (Microtubule: Control=1±0.062, tau siRNA=0.84±0.02, MAP6 siRNA=0.84±0.04, n=6; detyrosinated tubulin: Control=1±0.001, tau siRNA=1.45±0.14, MAP6 siRNA=0.83±0.07, n=4) ∗, p < 0.05 and ∗∗, p < 0.01.

(G) Kymographs represent EB3-GFP comets after tau or MAP6 depletion. Vertical scale bar is 10 seconds. Horizontal scale bar is 2 µm.

(H-I) Bar graphs represent quantification of EB3-GFP comet numbers (Control=18.5±0.7 n=15, tau siRNA=9.9±1.06 n=19, MAP6 siRNA=18.1±0.81 n=14) and comet velocity

(Control=0.21±0.01, n=148; tau siRNA=0.17±0.01, n=121; MAP6 siRNA=0.33±0.026, n=113) ∗, p < 0.05.

Figure 10: Mathematical analysis of microtubule decline as a result of exposure over time to nocodazole, and schematic summary of conclusions.

(A) Images of microtubule mass in the axon at different time points (0 minute, 15 minutes, 60 minutes) after nocodazole treatment of rat fetal cortical neurons transfected with control siRNA, tau siRNA or MAP6 siRNA. Images are presented using fire-scale filter in ImageJ. Scale bar is

10 μm.

(B) Quantification of microtubule mean fluorescence intensity along the axon relative to control after nocodazole treatment at different time points (0 minutes, 5 minutes, 15 minutes, 30 minutes,

60 minutes) in neurons transfected with control, tau or MAP6 siRNA. Significance markers compared to control siRNA indicate p<0.05, *, MAP6 siRNA, #, tau siRNA . Here, we applied a

3,18 previously published strategy to measure the t1/2 and percentages of the two phases of microtubule decline in control, tau-depleted or MAP6-depleted axons. Using this method, for control cortical axons, we calculated that the t1/2 for the first phase was ~5 minutes, while that for

the second phase was ~126 minutes. For tau-depleted axons, the t1/2 for the first phase was ~6 minutes, while that for the second phase was ~124 minutes. For MAP6 depleted axons, the t1/2 for the first phase was ~1 minute, while that for the second phase was ~125 minutes. In control axons,

64% of the total microtubule mass comprises the first phase while 36% comprises the second phase.

In tau-depleted axons, 38% and 62% comprise these two phases respectively; while in MAP6- depleted axons, 59% and 41% comprise these two phases respectively. Phase 1 and Phase 2 do not perfectly represent the labile and stable microtubule fractions because, for that to be the case, the stable and the labile microtubule fractions would have to be equally accessible to depolymerization.

Given that the labile domain of any given microtubule must depolymerize completely before the stable domain can start to depolymerize, the t1/2 for the first phase in this mathematical analysis is not directly synonymous with the labile fraction, which we suspect becomes even less labile (i.e. more stable) after tau depletion than indicated by the comparison of Phase 1 kinetics.

(C) Schematic represents the role of tau and MAP6 in regulating the stable (shown in green) and labile (shown in shades of red) domains of axonal microtubules. The two domains of an individual microtubule differ markedly from one another in their stability properties. The labile domain can vary in how labile it is, with lighter red indicating less labile and darker red indicating more labile.

Blue and yellow thread-like structures represent tau and MAP6, respectively. After tau depletion, there is loss of microtubule mass, with the microtubules in the axon having diminished labile domains but increased stable domains. After MAP6 depletion, there is loss of microtubule mass, but the proportions of stable and labile remain roughly similar to control. The labile domains are less labile after tau depletion but more labile after MAP6 depletion. We posit that an enrichment of tau on the labile domain prevents MAP6 and other genuine microtubule stabilizers from binding to the microtubule, thereby keeping the labile domain labile and allowing it to elongate without being stabilized.

Figure S1: Tau siRNA validation in immunostaining and Western blot and tau distribution.

(A) Bar graph depicts the number of individual microtubules splayed out from the axonal shaft in control and tau-depleted neurons. (Control=18.6±2, n=9; Tau siRNA=26.1.2±2.4, n=10) ∗, p <

0.05.

(B) Bar graph represents the quantification of width of the axon using microtubule splay assay.

(Control=1.99±0.11, n=17; tau siRNA=2.48±0.16, n=19) ∗, p < 0.05.

(D) Quantification of correlation between tyrosinated tubulin and tau shows positive correlation.

(R = 0.77, p < 0.05)

Figure S2: Alternation in tau distribution when MAP6 is depleted.

(A-F) Cortical neurons were transfected with control siRNA or MAP6 siRNA, fixed and stained for βIII-tubulin, tyrosinated tubulin and tau. Scale bar is 10 μm.

(G) Plot represents the distribution of tau along the length of the axon by using the analytic tool named “full width half maximum” in Excel. O represents the cell body and 1 represents the tip of the axon. The red and blue lines represent normalized fluorescence intensity profiles of axons after

66 treatment with control and MAP6 siRNA respectively. The black lines represent the height of the peak (i.e. the maximum tau intensity) and width of the peak (i.e. distribution of tau).

(H) Bar graph represents quantitative measures of spatial distribution (width of the peak in G) of normalized tau intensity along on the axon (Control=0.36±0.01, n=10; tau siRNA=0.56±0.02, n=10) ∗∗, p < 0.01.

(I) Bar graph represents the distance of maximum tau intensity (height of the peak in G measured from the cell body by full width half maximum respectively) (Control=0.84±0.03, n=9; tau siRNA=0.73±0.03, n=9) ∗, p < 0.05.

Figure S3: MAP6 recruits in lamellipodia along the axons of cortical neurons after tau is depleted.

(A-D) Immunofluorescence (inverted) images show microtubule and MAP6 staining in axonal lamellipodia when tau is depleted. Scale bar is 10 μm.

Figure S4: Doublecortin is not recruited to microtubules after tau depletion from cortical neurons.

(A-D) Immunofluorescence (inverted) images of the growth cone immunostained for doublecortin and βIII-tubulin under conditions of tau depletion. Scale bar is 10 μm.

(E) Bar graph represents the ratio of doublecortin intensity to microtubule intensity relative to control in the growth cone when tau is depleted. (Control=1±0.12, n=14; tau siRNA=1±0.20, n=13).

(F) Bar graph represents quantification of co-localization of doublecortin and βIII-tubulin using

Pearson’s correlation coefficient. (Control, R= 69%±3.6%, n=10; tau siRNA, R =63%±3.1%, n=10)

CONCLUSIONS AND RECOMMENDATIONS

The results of our studies have reemphasized the complicated nature of microtubule dynamics and interactions with associated proteins. The interactions of microtubule modifiers that come together to produce the cytoskeleton’s dynamic profile in the axon are still in need of study, and we have only just begun to be able commenting on the complications of disease in the currently elucidated systems.

Our work with fidgetin presents a viable option for therapy in a variety of regenerative contexts, though we chose to focus on a relatively simple model related to spinal cord injury. The convenience and effectiveness of nanoparticle delivery could hold at least as much potential for basic research as it does the clinically translatable purposes for which it was designed, as it allowed us to use a very simple in vitro model on cells that normally would not take an siRNA treatment. The unfortunate instances of individual batch failure among the nanoparticles may suggest that treatment with fidgetin knockdown would best be moved forward using other mechanisms, such as viral infection, if more reliable formulations of nanoparticle-fidgetin siRNA are not easily forthcoming. Control and stability of nanoparticles might be improved with the use of “sugar-glass” matrix types of nanoparticle formulation that allow for more efficient trapping of biomolecules and also a sustained release.82 Improvements in timed release nanoparticles would be a path to combinatorial treatments where carefully formulated nanoparticles containing microtubule-promoting factors are applied directly to nerve fibers as part of initial steps to promote neuronal regeneration. Certainly, the next steps with fidgetin knockdown would be to carry out in vivo experiments using well-established regenerative and functional models of nerve injury to analyze the effects of labile microtubule enhancement. Given our

70 study’s use of adult DRG neurons, this data would easily dovetail into a dorsal-root crush model with DRG injection of shRNA-loaded AAV5 and forepaw function as a functional measure of treatment effect.83,84 Fidgetin is developmentally regulated,27 and the effectiveness of labile microtubule enhancement inducing a more growth-promotive state within the axon may have differential effects when fidgetin knockdown is applied to injuries in central or peripheral fibers. Since fidgetin appears to be relevant to development, it may be a particularly good protein for enhancing regeneration in the intrinsically growth- resistant fibers of the central nervous system.85 In searching for more central applications, fidgetin knockdown might prove useful in treating neurotrauma to the hippocampus resulting from epilepsy and other neurodegenerative diseases, as fidgetin RNA expression has been shown to be a significant part of mouse neuron response to environmental stress.86

Beyond the intrinsic potential for promoting lamellipodial elongation and axonal growth, fidgetin knockdown has implications for the interaction of extraneuronal cells and factors.

The implications of careful experimental design to control mitotic and migratory cells in and around nervous wound areas may be worth further study, both to analyze helpful effects and stave off negative ones. Depletion of fidgetin-like 2, another member of the fidgetin family, has been shown to promote cell migration and wound healing in mice.41

The possibility of using fidgetin itself to regulate glial migration has also been suggested.87

These types of migration effects will be an important tool as the discussion of when where reactive astrocytes are beneficial to the spinal cord continues.88

The implications of cellular responsiveness to signaling after fidgetin manipulation are also important to consider. One advantage of maintaining a labile microtubule mass is the possibility of keeping growth cone turning intact. This can be helpful or a hindrance

71 depending on the desired conditions of untempered growth through inhibitory substrate or precise synaptic formations at target connections. We did not find a significant change in the turning rate of CSPG-challenged growth cones following fidgetin knockdown, though the relative levels of knockdown may have an effect on whether a phenotype may be induced. While overall growth cone turning appeared to remain similar to controls in this instance, it is unclear whether the ability of labile-enhanced axons to grow longer after crossing onto CSPGs is purely based on intrinsic motile forces or if perhaps there is also an endpoint of CSPG signaling lost with fidgetin knockdown. Microtubules certainly are a common endpoint for many signaling pathways, but whether extracellular aggrecan might directly influence growth cone guidance by acting on labile-severing proteins is another question.9,89 Solid information regarding the connection of extracellular signals to severing proteins has been difficult to produce, but one study has pinpointed fidgetin-like 1 as a key component of axon guidance, in fact critical to proper synaptic formation.7 As these results were found to affect microtubules not by polymer severing but by the catastrophe-inducing interaction of fidgetin-like 1 with end-binding proteins, it is unclear how common this mechanism of regulation may be among other fidgetin proteins. Regardless, it seems likely that fidgetin is a target molecule for inhibitory pathways that can be stimulated or depleted as required to promote robust axonal growth or sensitive axonal sprouting. As studies move further towards genetically-enhanced treatment of human disease, conditional regulation of fidgetin could be part of controlling progression of graft or neural progenitor transplantation following injury of the nervous system.90,91

Experiments in growth cone turning utilizing external cues also interest us as another mechanism of teasing apart the details of our tau depletion study. Considering the shift of

MAP6 to growth cones and the subtle changes in both quantity and quality of dynamic domains, however, the outcome and interpretation of such an experiment would not necessarily be clear-cut. As tau promotes retention of labile microtubules, we might predict that a loss of tau would lead to desensitization to external cues, while MAP6 depletion would have the opposite effect (e.g. MAP6 depleted neurons with a more dynamic microtubule profile turn and grow more quickly towards an NGF-coated bead).92 Having seen that tau is functionally polarized, favoring labile microtubules in both binding and effect, the question remains of exactly how this polarity is generated and maintained. We would be interested in performing experiments using extreme cold to analyze the interactions that seem to be at play between tau and MAP6. We might hypothesize that the induction of MAP6 binding brought on by cold would eventually favor the complete displacement of tau along surviving microtubules. The hypothesized role of MAP6 as a compensatory binding agent in absence of tau could also be tested in situations where tau is present or nonfunctional. Measuring changes of MAP6 binding in the presence of FTDP-

17 tau mutations would be one clinically-relevant model to ascertain whether phosphorylated tau permits MAP6 redistribution similarly to our knockdown study.93,94

MAP6 has been hypothesized to potentially have longitudinal contacts on the microtubule similar to tau, but the exact location of binding on the microtubule has not been determined, and so neither has the possibility of competitive binding sites as opposed to conformational changes in tubulin being responsible for distribution differences.33 Additional studies may be undertaken with single amino acid deletions or mutations within tau and MAP6 microtubule-binding domains to determine whether the direct effect of one protein binding affects the other.

Our studies with tau focused on highly simplified in vitro models for the sake of clarity in determining the basic function of tau. Moving forward to more clinically and physiologically relevant models will require reconsideration of tau function at each step.

Beyond considerations of varying tau phosphorylation states, complicated diseases where combinatorial therapy involving labile microtubule enhancement might be deemed beneficial, like Alzheimer’s, involve many confounding factors that may or may not produce positive treatment with something like fidgetin knockdown alone. For example, the effects of GSK-3β on tau and/or directly upon microtubule dynamics are not fully understood at this stage,95,96 though continued analysis of the effects of GSK-3β and other potentially pathological molecules on microtubules with the detailed analysis we have demonstrated would certainly be beneficial. Combined with our continuing studies of tau and MAP6 regulation, ongoing insights into changes in microtubule dynamics may quickly lead to suggestions for comprehensive microtubule preservation paradigms.

Ultimately our studies have emphasized the importance of understanding the dynamic nature of microtubules in the neuronal axon. While acknowledging the importance of maintaining microtubules as a treatment against overall microtubule loss, the elucidation of tau as a labile-promoting factor and the demonstration of fidgetin as a labile-based therapeutic target both suggest that microtubule-based treatments need to be considered as far more than an exclusive issue of stabilization. Microtubules are complex structures of varying dynamic domains, with tau acting in a manner opposed to previous supposition.

By clarifying the nature of tau and microtubule dynamics, we hope to enhance prospects for both basic research and clinical treatment through focus on the dynamic microtubule.

LIST OF REFERENCES

1 Herculano-Houzel, S. The human brain in numbers: a linearly scaled-up primate brain. Front Hum Neurosci 3, 31, doi:10.3389/neuro.09.031.2009 (2009).

2 Kandel, E. R. Principles of neural science. 5th edn, (McGraw-Hill, 2013).

3 Baas, P. W., Rao, A. N., Matamoros, A. J. & Leo, L. Stability properties of neuronal microtubules. Cytoskeleton (Hoboken) 73, 442-460, doi:10.1002/cm.21286 (2016).

4 Haendel, M. A., Bollinger, K. E. & Baas, P. W. Cytoskeletal changes during neurogenesis in cultures of avain neural crest cells. J Neurocytol 25, 289-301 (1996).

5 Kahn, O. I., Sharma, V., Gonzalez-Billault, C. & Baas, P. W. Effects of kinesin-5 inhibition on dendritic architecture and microtubule organization. Mol Biol Cell 26, 66-77, doi:10.1091/mbc.E14-08-1313 (2015).

6 Lin, S., Liu, M., Mozgova, O. I., Yu, W. & Baas, P. W. Mitotic motors coregulate microtubule patterns in axons and dendrites. J Neurosci 32, 14033-14049, doi:10.1523/JNEUROSCI.3070-12.2012 (2012).

7 Fassier, C. et al. Motor axon navigation relies on Fidgetin-like 1-driven microtubule plus end dynamics. J Cell Biol, doi:10.1083/jcb.201604108 (2018).

8 Sainath, R. & Gallo, G. Cytoskeletal and signaling mechanisms of neurite formation. Cell Tissue Res 359, 267-278, doi:10.1007/s00441-014-1955-0 (2015).

9 Kahn, O. I. & Baas, P. W. Microtubules and Growth Cones: Motors Drive the Turn. Trends in neurosciences 39, 433-440, doi:10.1016/j.tins.2016.04.009 (2016).

10 Nadar, V. C., Lin, S. & Baas, P. W. Microtubule redistribution in growth cones elicited by focal inactivation of kinesin-5. J Neurosci 32, 5783-5794, doi:10.1523/jneurosci.0144-12.2012 (2012).

11 Challacombe, J. F., Snow, D. M. & Letourneau, P. C. Dynamic microtubule ends are required for growth cone turning to avoid an inhibitory guidance cue. J Neurosci 17, 3085-3095 (1997).

12 Franker, M. A. & Hoogenraad, C. C. Microtubule-based transport - basic mechanisms, traffic rules and role in neurological pathogenesis. J Cell Sci 126, 2319-2329, doi:10.1242/jcs.115030 (2013).

13 Wang, J. Z. & Liu, F. Microtubule-associated protein tau in development, degeneration and protection of neurons. Prog Neurobiol 85, 148-175, doi:10.1016/j.pneurobio.2008.03.002 (2008).

14 Zempel, H. et al. Amyloid-beta oligomers induce synaptic damage via Tau-dependent microtubule severing by TTLL6 and spastin. EMBO J 32, 2920-2937, doi:10.1038/emboj.2013.207 (2013).

15 Zempel, H., Thies, E., Mandelkow, E. & Mandelkow, E. M. Abeta oligomers cause localized Ca(2+) elevation, missorting of endogenous Tau into dendrites, Tau phosphorylation, and destruction of microtubules and spines. J Neurosci 30, 11938-11950, doi:10.1523/JNEUROSCI.2357-10.2010 (2010).

16 Matamoros, A. J. & Baas, P. W. Microtubules in health and degenerative disease of the nervous system. Brain Res Bull 126, 217-225, doi:10.1016/j.brainresbull.2016.06.016 (2016).

17 Baas, P. W. & Black, M. M. Individual microtubules in the axon consist of domains that differ in both composition and stability. J Cell Biol 111, 495-509 (1990).

18 Baas, P. W., Slaughter, T., Brown, A. & Black, M. M. Microtubule dynamics in axons and dendrites. J Neurosci Res 30, 134-153, doi:10.1002/jnr.490300115 (1991).

19 Austin, T. O., Qiang, L. & Baas, P. W. in Neuroprotection in Alzheimer's Disease (ed I. Gozes) Ch. 4, 59-71 (Elsevier/Academic Press, 2017).

20 Lacroix, B. et al. Tubulin polyglutamylation stimulates spastin-mediated microtubule severing. J Cell Biol 189, 945-954, doi:10.1083/jcb.201001024 (2010).

21 Brunden, K. R. et al. Epothilone D improves microtubule density, axonal integrity, and cognition in a transgenic mouse model of tauopathy. J Neurosci 30, 13861-13866, doi:10.1523/JNEUROSCI.3059-10.2010 (2010).

22 Lou, K. et al. Brain-penetrant, orally bioavailable microtubule-stabilizing small molecules are potential candidate therapeutics for Alzheimer's disease and related tauopathies. J Med Chem 57, 6116-6127, doi:10.1021/jm5005623 (2014).

23 Zhang, B. et al. The microtubule-stabilizing agent, epothilone D, reduces axonal dysfunction, neurotoxicity, cognitive deficits, and Alzheimer-like pathology in an interventional study with aged tau transgenic mice. J Neurosci 32, 3601-3611, doi:10.1523/JNEUROSCI.4922- 11.2012 (2012).

24 Zhang, B. et al. Microtubule-binding drugs offset tau sequestration by stabilizing microtubules and reversing fast axonal transport deficits in a tauopathy model. Proc Natl Acad Sci U S A 102, 227-231, doi:10.1073/pnas.0406361102 (2005).

25 Cox, G. A., Mahaffey, C. L., Nystuen, A., Letts, V. A. & Frankel, W. N. The mouse fidgetin gene defines a new role for AAA family proteins in mammalian development. Nat Genet 26, 198- 202, doi:10.1038/79923 (2000).

26 Mukherjee, S. et al. Human Fidgetin is a microtubule severing the enzyme and minus-end depolymerase that regulates mitosis. Cell Cycle 11, 2359-2366, doi:10.4161/cc.20849 (2012).

27 Leo, L. et al. Vertebrate Fidgetin Restrains Axonal Growth by Severing Labile Domains of Microtubules. Cell Rep 12, 1723-1730, doi:10.1016/j.celrep.2015.08.017 (2015).

28 Cash, A. D. et al. Microtubule reduction in Alzheimer's disease and aging is independent of tau filament formation. Am J Pathol 162, 1623-1627 (2003).

29 Zhang, F. et al. Posttranslational modifications of alpha-tubulin in alzheimer disease. Transl Neurodegener 4, 9, doi:10.1186/s40035-015-0030-4 (2015).

30 Janning, D. et al. Single-molecule tracking of tau reveals fast kiss-and-hop interaction with microtubules in living neurons. Mol Biol Cell 25, 3541-3551, doi:10.1091/mbc.E14-06-1099 (2014).

31 Black, M. M., Slaughter, T., Moshiach, S., Obrocka, M. & Fischer, I. Tau is enriched on dynamic microtubules in the distal region of growing axons. J Neurosci 16, 3601-3619 (1996).

32 Delphin, C. et al. MAP6-F is a temperature sensor that directly binds to and protects microtubules from cold-induced depolymerization. J Biol Chem 287, 35127-35138, doi:10.1074/jbc.M112.398339 (2012).

33 Lefevre, J. et al. Structural basis for the association of MAP6 protein with microtubules and its regulation by calmodulin. J Biol Chem 288, 24910-24922, doi:10.1074/jbc.M113.457267 (2013).

34 Liu, K., Tedeschi, A., Park, K. K. & He, Z. Neuronal intrinsic mechanisms of axon regeneration. Annu Rev Neurosci 34, 131-152, doi:10.1146/annurev-neuro-061010-113723 (2011).

35 Mar, F. M., Bonni, A. & Sousa, M. M. Cell intrinsic control of axon regeneration. EMBO Rep 15, 254-263, doi:10.1002/embr.201337723 (2014).

36 Hellal, F. et al. Microtubule stabilization reduces scarring and causes axon regeneration after spinal cord injury. Science 331, 928-931, doi:10.1126/science.1201148 (2011).

37 Ruschel, J. et al. Axonal regeneration. Systemic administration of epothilone B promotes axon regeneration after spinal cord injury. Science 348, 347-352, doi:10.1126/science.aaa2958 (2015).

38 Popovich, P. G., Tovar, C. A., Lemeshow, S., Yin, Q. & Jakeman, L. B. Independent evaluation of the anatomical and behavioral effects of Taxol in rat models of spinal cord injury. Exp Neurol 261, 97-108, doi:10.1016/j.expneurol.2014.06.020 (2014).

39 Baas, P. W. Beyond taxol: microtubule-based strategies for promoting nerve regeneration after injury. Neural Regen Res 9, 1265-1266, doi:10.4103/1673-5374.137572 (2014).

40 Baas, P. W. & Ahmad, F. J. Beyond taxol: microtubule-based treatment of disease and injury of the nervous system. Brain 136, 2937-2951, doi:10.1093/brain/awt153 (2013).

41 Charafeddine, R. A. et al. Fidgetin-Like 2: A Microtubule-Based Regulator of Wound Healing. J Invest Dermatol 135, 2309-2318, doi:10.1038/jid.2015.94 (2015).

42 Lin, S. et al. Inhibition of Kinesin-5, a microtubule-based motor protein, as a strategy for enhancing regeneration of adult axons. Traffic 12, 269-286, doi:10.1111/j.1600-0854.2010.01152.x (2011).

43 Tanaka, E., Ho, T. & Kirschner, M. W. The role of microtubule dynamics in growth cone motility and axonal growth. J Cell Biol 128, 139-155 (1995).

44 Liu, K. et al. PTEN deletion enhances the regenerative ability of adult corticospinal neurons. Nat Neurosci 13, 1075-1081, doi:10.1038/nn.2603 (2010).

45 Barson, J. R. et al. Substance P in the anterior thalamic paraventricular nucleus: promotion of ethanol drinking in response to orexin from the hypothalamus. Addict Biol 22, 58-69, doi:10.1111/adb.12288 (2017).

46 Beller, J. A. et al. Comparison of sensory neuron growth cone and filopodial responses to structurally diverse aggrecan variants, in vitro. Exp Neurol 247, 143-157, doi:10.1016/j.expneurol.2013.02.012 (2013).

47 Kadavath, H. et al. Tau stabilizes microtubules by binding at the interface between tubulin heterodimers. Proc Natl Acad Sci U S A 112, 7501-7506, doi:10.1073/pnas.1504081112 (2015).

48 Dehmelt, L. & Halpain, S. The MAP2/Tau family of microtubule-associated proteins. Genome Biol 6, 204, doi:10.1186/gb-2004-6-1-204 (2005).

49 Kolarova, M., Garcia-Sierra, F., Bartos, A., Ricny, J. & Ripova, D. Structure and pathology of tau protein in Alzheimer disease. Int J Alzheimers Dis 2012, 731526, doi:10.1155/2012/731526 (2012).

50 Brunden, K. R., Trojanowski, J. Q., Smith, A. B., 3rd, Lee, V. M. & Ballatore, C. Microtubule-stabilizing agents as potential therapeutics for neurodegenerative disease. Bioorganic & medicinal chemistry 22, 5040-5049, doi:10.1016/j.bmc.2013.12.046 (2014).

51 Brown, A., Li, Y., Slaughter, T. & Black, M. M. Composite microtubules of the axon: quantitative analysis of tyrosinated and acetylated tubulin along individual axonal microtubules. J Cell Sci 104 ( Pt 2), 339-352 (1993).

52 Slaughter, T. & Black, M. M. STOP (stable-tubule-only-polypeptide) is preferentially associated with the stable domain of axonal microtubules. J Neurocytol 32, 399-413, doi:10.1023/b:Neur.0000011334.70648.87 (2003).

53 Tint, I., Slaughter, T., Fischer, I. & Black, M. M. Acute inactivation of tau has no effect on dynamics of microtubules in growing axons of cultured sympathetic neurons. J Neurosci 18, 8660- 8673 (1998).

54 Leo, L., Yu, W. & Baas, P. W. Using siRNA to study microtubule-related proteins in cultured neurons. Methods Cell Biol 131, 163-176, doi:10.1016/bs.mcb.2015.06.005 (2016).

55 Qiang, L., Yu, W., Andreadis, A., Luo, M. & Baas, P. W. Tau protects microtubules in the axon from severing by katanin. J Neurosci 26, 3120-3129, doi:10.1523/jneurosci.5392-05.2006 (2006).

56 Yu, W. et al. The microtubule-severing proteins spastin and katanin participate differently in the formation of axonal branches. Mol Biol Cell 19, 1485-1498, doi:10.1091/mbc.E07-09-0878 (2008).

57 Sudo, H. & Baas, P. W. Strategies for diminishing katanin-based loss of microtubules in tauopathic neurodegenerative diseases. Hum Mol Genet 20, 763-778, doi:10.1093/hmg/ddq521 (2011).

58 Ahmad, F. J., Pienkowski, T. P. & Baas, P. W. Regional differences in microtubule dynamics in the axon. J Neurosci 13, 856-866 (1993).

59 Baas, P. W. et al. Tau confers drug stability but not cold stability to microtubules in living cells. J Cell Sci 107 ( Pt 1), 135-143 (1994).

60 Baas, P. W., Pienkowski, T. P. & Kosik, K. S. Processes induced by tau expression in Sf9 cells have an axon-like microtubule organization. J Cell Biol 115, 1333-1344 (1991).

61 Lee, G. & Rook, S. L. Expression of tau protein in non-neuronal cells: microtubule binding and stabilization. J Cell Sci 102 ( Pt 2), 227-237 (1992).

62 Kempf, M., Clement, A., Faissner, A., Lee, G. & Brandt, R. Tau binds to the distal axon early in development of polarity in a microtubule- and microfilament-dependent manner. J Neurosci 16, 5583-5592 (1996).

63 Black, M. M., Slaughter, T. & Fischer, I. Microtubule-associated protein 1b (MAP1b) is concentrated in the distal region of growing axons. J Neurosci 14, 857-870 (1994).

64 Ferhat, L. et al. Expression of the mitotic motor protein Eg5 in postmitotic neurons: implications for neuronal development. J Neurosci 18, 7822-7835 (1998).

65 Errico, A., Claudiani, P., D'Addio, M. & Rugarli, E. I. Spastin interacts with the centrosomal protein NA14, and is enriched in the spindle pole, the midbody and the distal axon. Hum Mol Genet 13, 2121-2132, doi:10.1093/hmg/ddh223 (2004).

66 Tint, I., Jean, D., Baas, P. W. & Black, M. M. Doublecortin associates with microtubules preferentially in regions of the axon displaying actin-rich protrusive structures. J Neurosci 29, 10995-11010, doi:10.1523/jneurosci.3399-09.2009 (2009).

67 Mandelkow, E. M. et al. Glycogen synthase kinase-3 and the Alzheimer-like state of microtubule-associated protein tau. FEBS Lett 314, 315-321 (1992).

68 Sontag, E., Nunbhakdi-Craig, V., Lee, G., Bloom, G. S. & Mumby, M. C. Regulation of the phosphorylation state and microtubule-binding activity of Tau by protein phosphatase 2A. Neuron 17, 1201-1207 (1996).

69 Binder, L. I., Frankfurter, A. & Rebhun, L. I. The distribution of tau in the mammalian central nervous system. J Cell Biol 101, 1371-1378 (1985).

70 Baas, P. W. & Ahmad, F. J. The plus ends of stable microtubules are the exclusive nucleating structures for microtubules in the axon. J Cell Biol 116, 1231-1241 (1992).

71 Margolis, R. L., Job, D., Pabion, M. & Rauch, C. T. Sliding of STOP proteins on microtubules: a model system for diffusion-dependent microtubule motility. Annals of the New York Academy of Sciences 466, 306-321 (1986).

72 Tortosa, E. et al. Dynamic Palmitoylation Targets MAP6 to the Axon to Promote Microtubule Stabilization during Neuronal Polarization. Neuron 94, 809-825.e807, doi:10.1016/j.neuron.2017.04.042 (2017).

73 Suh, L. H., Oster, S. F., Soehrman, S. S., Grenningloh, G. & Sretavan, D. W. L1/Laminin modulation of growth cone response to EphB triggers growth pauses and regulates the microtubule destabilizing protein SCG10. J Neurosci 24, 1976-1986, doi:10.1523/jneurosci.1670-03.2004 (2004).

74 Witman, G. B., Cleveland, D. W., Weingarten, M. D. & Kirschner, M. W. Tubulin requires tau for growth onto microtubule initiating sites. Proc Natl Acad Sci U S A 73, 4070-4074 (1976).

75 Austin, T. O. et al. Nanoparticle Delivery of Fidgetin siRNA as a Microtubule-based Therapy to Augment Nerve Regeneration. Sci Rep 7, 9675, doi:10.1038/s41598-017-10250-z (2017).

76 Drubin, D. G. & Kirschner, M. W. Tau protein function in living cells. J Cell Biol 103, 2739-2746 (1986).

77 Bradshaw, R. A. & Stahl, P. in Encyclopedia of cell biology Vol. 2 539-547 (Elsevier/AP, Academic Press is an imprint of Elsevier, 2016).

78 Lodish, H. F. in Molecular cell biology Ch. 18, 757-800 (W.H. Freeman, 2008).

79 Zempel, H. & Mandelkow, E. M. Tau missorting and spastin-induced microtubule disruption in neurodegeneration: Alzheimer Disease and Hereditary Spastic Paraplegia. Mol Neurodegener 10, 68, doi:10.1186/s13024-015-0064-1 (2015).

80 Liu, M. et al. Kinesin-12, a mitotic microtubule-associated motor protein, impacts axonal growth, navigation, and branching. J Neurosci 30, 14896-14906, doi:10.1523/JNEUROSCI.3739- 10.2010 (2010).

81 Deloulme, J. C. et al. Microtubule-associated protein 6 mediates neuronal connectivity through Semaphorin 3E-dependent signalling for axonal growth. Nat Commun 6, 7246, doi:10.1038/ncomms8246 (2015).

82 Friedman, A. J. et al. Sustained release nitric oxide releasing nanoparticles: characterization of a novel delivery platform based on nitrite containing hydrogel/glass composites. Nitric Oxide 19, 12-20, doi:10.1016/j.niox.2008.04.003 (2008).

83 Cheah, M., Fawcett, J. W. & Andrews, M. R. Dorsal Root Ganglion Injection and Dorsal Root Crush Injury as a Model for Sensory Axon Regeneration. J Vis Exp, doi:10.3791/55535 (2017).

84 Wu, A., Lauschke, J. L., Morris, R. & Waite, P. M. Characterization of rat forepaw function in two models of cervical dorsal root injury. J Neurotrauma 26, 17-29, doi:10.1089/neu.2008.0675 (2009).

85 O'Donovan, K. J. Intrinsic Axonal Growth and the Drive for Regeneration. Front Neurosci 10, 486, doi:10.3389/fnins.2016.00486 (2016).

86 McAdams, R. M. et al. Dose-dependent effects of morphine exposure on mRNA and microRNA (miR) expression in hippocampus of stressed neonatal mice. PLoS One 10, e0123047, doi:10.1371/journal.pone.0123047 (2015).

87 Hu, Z. et al. Fidgetin regulates cultured astrocyte migration by severing tyrosinated microtubules at the leading edge. Mol Biol Cell 28, 545-553, doi:10.1091/mbc.E16-09-0628 (2017).

88 Lukovic, D. et al. Concise review: reactive astrocytes and stem cells in spinal cord injury: good guys or bad guys? Stem Cells 33, 1036-1041, doi:10.1002/stem.1959 (2015).

89 Gallo, G. The cytoskeletal and signaling mechanisms of axon collateral branching. Dev Neurobiol 71, 201-220, doi:10.1002/dneu.20852 (2011).

90 Bertero, A. et al. Conditional Manipulation of Gene Function in Human Cells with Optimized Inducible shRNA. Curr Protoc Stem Cell Biol 44, 5C 4 1-5C 4 48, doi:10.1002/cpsc.45 (2018).

91 Zhang, Z. et al. CRISPR/Cas9 Genome-Editing System in Human Stem Cells: Current Status and Future Prospects. Mol Ther Nucleic Acids 9, 230-241, doi:10.1016/j.omtn.2017.09.009 (2017).

92 Nadar, V. C., Ketschek, A., Myers, K. A., Gallo, G. & Baas, P. W. Kinesin-5 is essential for growth-cone turning. Curr Biol 18, 1972-1977, doi:10.1016/j.cub.2008.11.021 (2008).

93 Bardai, F. H. et al. A Conserved Cytoskeletal Signaling Cascade Mediates Neurotoxicity of FTDP-17 Tau Mutations In Vivo. J Neurosci 38, 108-119, doi:10.1523/JNEUROSCI.1550- 17.2017 (2018).

94 Combs, B. & Gamblin, T. C. FTDP-17 tau mutations induce distinct effects on aggregation and microtubule interactions. Biochemistry 51, 8597-8607, doi:10.1021/bi3010818 (2012).

95 Decker, H., Lo, K. Y., Unger, S. M., Ferreira, S. T. & Silverman, M. A. Amyloid-beta peptide oligomers disrupt axonal transport through an NMDA receptor-dependent mechanism that is mediated by glycogen synthase kinase 3beta in primary cultured hippocampal neurons. J Neurosci 30, 9166-9171, doi:10.1523/JNEUROSCI.1074-10.2010 (2010).

96 Ramser, E. M. et al. Amyloid-beta oligomers induce tau-independent disruption of BDNF axonal transport via calcineurin activation in cultured hippocampal neurons. Mol Biol Cell 24, 2494-2505, doi:10.1091/mbc.E12-12-0858 (2013).