THE PENNSYLVANIA STATE UNIVERSITY SCHREYER HONORS COLLEGE

DEPARTMENT OF BIOLOGY

REPROGRAMMING ASTROCYTES INTO FUNCTIONAL NEURONS FOR SPINAL CORD REPAIR

ALICE CAI SPRING 2017

A thesis submitted in partial fulfillment of the requirements for baccalaureate degrees in Biochemistry and Molecular Biology and Immunology and Infectious Disease with honors in Biology

Reviewed and approved* by the following:

Gong Chen Professor of Biology Thesis Supervisor

Timothy Jegla Associate Professor of Biology Faculty Reader

* Signatures are on file in the Schreyer Honors College. i

ABSTRACT

Spinal cord injury (SCI) and Amyotrophic Lateral Sclerosis (ALS), commonly known as

Lou Gehrig’s disease, are both highly debilitating afflictions of the spinal cord with limited available therapies. Pathology is comparable to that of other central nervous system (CNS) injury and disease, involving neuron degeneration and reactive gliosis, which prevents axonal recovery and results in a glial scar. Since the invention of induced pluripotent stem cell technology, many studies have demonstrated direct trans-differentiation across cell lineages through the upregulation of developmental transcription factors. Previous work, including ours, has shown that upregulation of neural NeuroD1 can directly convert astroglial cells into neurons in the brain. Through cell culture and mouse models for SCI and ALS, this technology is applied to the spinal cord to regenerate spinal cord neurons from the glial scar. This thesis aims to demonstrate the potential of improved functionality and quality of life after injury by reprogramming reactive astrocytes in the spinal cord into neurons.

ii

TABLE OF CONTENTS

LIST OF FIGURES ...... iii

LIST OF TABLES ...... iv

ACKNOWLEDGEMENTS ...... v

Chapter 1 Introduction ...... 1

Spinal Cord Injury ...... 1 Amyotrophic Lateral Sclerosis ...... 2 Reprogramming Astrocytes to Regenerate Neurons ...... 4

Chapter 2 Materials and Methods ...... 7

Plasmid Constructs and Transfection ...... 7 Cell Culture of Mouse Astrocytes ...... 8 Patch-Clamp Recording of Cell Cultures ...... 9 Mouse Models and Stereotaxic Viral Injection ...... 10 Sample Preparation and Immunocytochemistry ...... 12

Chapter 3 Results ...... 15

Characterization of Mouse Spinal Cord Cell Cultures ...... 15 Ngn2 and NeuroD1 Can Convert Spinal Cord Glia into Neurons in Vitro ...... 16 Transcription Factors Drive Differentiation into Specific Neuronal Subtypes ...... 20 Characterization of the Spinal Cord Stab Injury Model ...... 26 Characterization of the SOD1 Mouse Model ...... 28 Stereotaxic Injection of AAV9 in SOD1 Mice ...... 30

Chapter 4 Discussion ...... 34

Reprogramming Spinal Cord Glia into Neurons in vitro ...... 34 Efficacy of in vivo Models of Spinal Cord Injury and ALS ...... 36 Future Work and Conclusive Remarks ...... 37

BIBLIOGRAPHY ...... 40

iii

LIST OF FIGURES

Figure 1. Plasmid constructs encoding GFP and neurogenic factors for retroviral packaging.7

Figure 2. Characterization of primary cultures of mouse spinal cord glia...... 16

Figure 3. Live time-course images of morphological changes of primary spinal cord cell cultures transduced with retrovirus...... 17

Figure 4. Direct conversion of mouse spinal cord astrocytes into neurons in vitro...... 20

Figure 5. Converted neurons express neuronal subtype markers after 3 weeks...... 22

Figure 6. Co-expression of Ngn2, Lhx3, and Isl1 may promote differentiation of mouse spinal cord glia into a cholinergic fate...... 23

Figure 7. NeuroD1 and Dlx2 co-expression most efficiently reprograms mouse spinal cord glia into GAD65+ neurons...... 25

Figure 8. Characterization of the in vivo model for spinal cord injury...... 27

Figure 9. Characterization of SOD1 mice shows motor neuron loss and reactive astrocytes. . 29

Figure 10. Rotarod performance of SOD1 and age-matched wild-type mice...... 30

Figure 11. NeuroD1 expression through AAV9 vector successfully upregulates NeuroD1. ... 31

Figure 12. mCherry is expressed in some OPCs and astrocytes...... 33

iv

LIST OF TABLES

Table 1. Behavioral Scale for SOD1 Mouse Model ...... 12

v

ACKNOWLEDGEMENTS

First, I would like to acknowledge Dr. Gong Chen for welcoming me into his research lab for the duration of my undergraduate education, and for supervising the work that has culminated in this honors thesis. He has been an incredibly inspiring mentor by always aiming high and showing his students that anything can be achieved with hard work and ambition. I would also like to thank Dr. Hedong Li for his mentorship and leadership in our research on the spinal cord.

He not only brought his expertise and knowledge of the spinal cord, but was an encouraging teacher in guiding me through the start of a new project. Thank you also to postdoctoral researcher Dr. Lei Zhang for providing me mentorship when I entered the lab as a freshman. She patiently taught me various experimental techniques, scientific thinking, and great appreciation for research by involving me in all parts of the research process.

I would like to thank the spinal cord group - Dr. Yuan Liu, Matt Keefe, Xiaoyun Ding,

Brendan Puls, Yan Ding, Austin Redilla, and Michael Lai - for being great friends and research partners. I also wish to thank Dr. Zifei Pei for producing AAV9 viruses and providing cells for transformation, and Yuting Bai for supporting all of us in the lab. Thank you to the rest of Chen

Lab for their endless support and companionship, and my family and roommates for their care and support. Finally, I would like to acknowledge Dr. Timothy Jegla and Dr. Gong Chen for reading this honors thesis and providing feedback.

I would also like to note that while I conducted the cell culture work, which includes all work described in Chapter 2 unless otherwise noted, I have included others’ data generated from the cell cultures for a more complete telling of the results, including RT-PCR data collected by

Matt Keefe, and electrophysiology data collected by Lei Zhang. Dr. Hedong Li performed the vi stereotaxic injections described in the mouse models. I also worked side by side with Xiaoyun

Ding to characterize the mouse models, and have included results we generated together. 1

Chapter 1

Introduction

Spinal Cord Injury

The spinal cord is a major component of the central nervous system and is responsible for muscle reflexes and communicating information from the brain to the body. Spinal cord injury is permanent and can be highly debilitating with limited available therapies. Worldwide incidence of spinal cord injury is not well reported, but in the United States, about 282,000 people were living with para- or tetraplegia in 2016, and about 17,000 new cases occur each year, not including those who die shortly after injury (National Spinal Cord Injury Statistical Center, 2016

Data Sheet). Acute trauma of the spinal cord causes a primary and a secondary injury; primary injury is characterized by ischemia, blood vessel damage, and cell death due to mechanical disruption and swelling of the spinal cord, and secondary injury is characterized by inflammatory damage, glutamate excitotoxicity, and associated neurotoxicity and massive cell death (Vercelli et al., 2015, McDonald et al., 2002). The loss of neurons is partnered with hypertrophy and proliferation of glial cells, also known as reactive gliosis, and results in a glial scar. Reactive gliosis is a common pathological response to central nervous system injuries and diseases including brain trauma, Alzheimer’s disease, stroke, epilepsy, and glioma (Pekny et al., 2014,

Sofroniew, 2015). Reactive glia initially help to restrict the spread of neurotoxic inflammation to the damaged area, but they also secrete neuroinhibitory factors and become scar tissue that inhibits recovery of damaged neurons and neural networks in later stages (Fitch et al., 1999,

Sofroniew, 2015, McDonald et al., 2002). After neuron loss, some adult neurogenesis may be 2 triggered in endogenous neural stem cells of the dentate gyrus and forebrain subgranular zone stem cell niches (Nakatomi et al., 2002, Yamashita et al., 2006), but this low level of neurogenesis does not generate enough neurons to repair neural circuitry in humans, nor does it resolve neuroinhibition by the glial scar (Sohur et al., 2005). Additionally, there are no equivalent neurogenic regions in the spinal cord.

Amyotrophic Lateral Sclerosis

Amyotrophic Lateral Sclerosis (ALS), also known as Lou Gehrig’s disease, is a neurodegenerative disease primarily characterized by loss of motor neurons in the spinal cord.

Pathological responses include muscle weakness, atrophy, and paralysis starting from the lower extremities in early stages, which progresses to dyspnea and dysphagia in later stages, and eventually leading to death due to respiratory failure (Zarei et al., 2015). Patients are typically diagnosed between 50-65 years old survive for an average of about four years after disease onset

(Magnus et al., 2002, Zarei et al., 2015). There are two types of ALS: Familial ALS caused by an inherited genetic factor, and sporadic ALS, which has no inherited genetic factor. Most cases are sporadic, while only 5-10% are familial; various mutations have been identified in genetic screening of ALS patients, but only mutations of the encoding superoxide dismutase 1

(SOD1) lead to classically inherited ALS (Gros-Louis et al., 2006). About 20% of familial ALS cases are caused by a missense mutation in the gene encoding superoxide dismutase 1 (SOD1), which is a cytoplasmic protein typically responsible for neutralizing oxygen free radicals

(superoxide) produced by mitochondrial metabolism (Liu et al., 2004, Zarei et al., 2015).

Therefore, the compromised ability to neutralize reactive oxygen species can cause cell 3 cytotoxicity. Mutant SOD1 also forms aggregates that bind to mitochondrial membranes and associated proteins, which interferes with ATP-production, mitochondrial transport along motor neuron axons, and calcium regulation (Liu et al., 2004, Cleveland and Rothstein, 2001).

Additionally, glutamate excitotoxicity was observed in SOD1 mice due to a decreased level of astroglial glutamate transporter 1 (GLT1) in brain cells, causing an excess of extracellular glutamate, excessive neuronal firing, and calcium influx (Trotti et al., 1999). Although mutant

SOD1 is expressed in all cells, only spinal motor neurons undergo degeneration, implying unique properties or mechanisms in spinal motor neurons that cause disease progression. This is not to say that other cells are unaffected; microglia and astrocytes also become reactive and secrete toxic signals that facilitate motor neuron degeneration (Boillee et al., 2006, Cleveland and

Rothstein, 2001). In fact, astrocytes and microglia play important roles in disease progression because of their close interactions with each other and the neuronal network. Astrocytes have been shown to enhance microglial activation and cause faster disease progression (Yamanaka et al., 2008).

The only FDA-approved treatment for ALS is riluzole, which combats glutamate excitotoxicity and has extended the survival of patients with ALS by just 2-4 months (Riviere et al., 1998). Riluzole’s mechanisms of action are not completely understood, but it is thought to block sodium channels and activate a G-protein-dependent mechanism to inhibit release of glutamic acid from neurons. It also is thought to inhibit glutamatergic neurotransmission by noncompetitively blocking postsynaptic N-methyl-D-aspartate (NMDA) receptors, which are ion channels activated by glutamate binding (Riviere et al., 1998, Doble, 1996, Kretschmer, et al.,

1998). However, like other interventions that are currently in trial stages, it cannot regenerate lost motor neurons and regain muscle function. We predict that converting reactive glia into 4 motor neurons can help to rebuild neuronal circuitry and restore a healthy phenotype in surrounding glia.

Reprogramming Astrocytes to Regenerate Neurons

The discovery that pluripotent stem cells (iPSCs) can be induced from adult mouse and human fibroblasts (Takahashi 2006, 2007) has opened up cell reprogramming as a therapeutic approach for neurodegenerative disease in the last decade. The biggest challenge that researchers face in the road to curing neurodegenerative disorders is the loss of neurons and how to replace them in order to gain functional recovery. Previous studies have included grafting of stem cells or neurons generated from iPSCs into the central nervous system (Lu et al., 2012, Lu et al.,

2014). However, stem cell transplants or indirect reprogramming through a pluripotent stage run risks such as immune rejection and tumor development. Other studies have demonstrated direct reprogramming of fibroblasts into neural stem cells and neurons (Thier et al., 2012, Vierbuchen et al., 2010, Han et al., 2012), which overcome the risks associated with stem cells, but these cells must be converted and purified ex vivo and still do not address barriers imposed by glial scarring. Recent studies, including in my lab, have taken advantage of the glial scar as a cell source for neuron regeneration through direct reprogramming.

Direct reprogramming is based on changing the transcription profile of the original cell into the transcription profile of a new differentiated fate. Cell types are different from each other and perform different functions due because they express different proteins through cell-type- specific gene profiles. profile of neurons can be induced through upregulation of neurogenic transcription factors using viral vectors. Retroviral transduction of neurogenic 5 factors has converted glial cells into neurons in vitro (Heinrich et al,. 2010) as well as in vivo

(Guo et al., 2014). Small molecules can also be used to regulate cell-type-specific signaling pathways and transcription to induce neuronal conversion from mouse fibroblasts (Ladewig et al., 2012), human fibroblasts (Li et al., 2015), and human astrocytes, which have been generated in cell culture and shown to integrate into brain circuits of adult mice (Zhang et al., 2015).

Two specific transcription factors that have shown the most promise in converting glia into neurons are Neurogenin-2 (Ngn2) and NeuroD1. Both are basic helix-loop-helix (bHLH) transcription factors, which are important in the developing central nervous system. Ngn2, as well as Achaete-scute homolog 1 (Ascl1), is expressed early in embryogenesis and is important in determining a neural cell fate (Lee, 1997). In the context of spinal cord development, studies suggest that Ascl1 and Ngn2 are important in dorsal horn interneuron development (Helms et al.,

2005), and Olig2, Ngn2, Lhx3, Isl1, and Hb9 are important in ventral horn motor neuron development (Lee et al., 2005, Arber et al., 1999, Scardigli et al., 2001). NeuroD1 is also expressed during embryogenesis but plays a more important role in neuronal differentiation, and is expressed into adulthood (Lee, 1997). Analysis of NeuroD1 binding and activity in the genome showed that NeuroD1 restructures the epigenetic landscape of the genome to allow transcription of neuronal gene enhancers by facilitating removal of repressive histone methylations and introduction of activating histone acetylations in promotor regions (Pataskar et al., 2016, Ray et al., 2016). Guo et al. has demonstrated that NeuroD1 alone can directly reprogram reactive glia into functional neurons in brain injury and Alzheimer’s disease mouse models and in human cell cultures. They also show that distal-less 2 (Dlx2) co- expression with NeuroD1 can covert NG2 cells into a GABA-ergic fate, while NeuroD1 expression alone tends to drive a glutamatergic fate (Guo et al., 2014). 6 This honors thesis describes the novel conversion of mouse spinal cord glia into different neuronal subtypes in vitro through overexpression of NeuroD1 and Ngn2, and discusses further application of this technology in spinal cord injury and ALS mouse models. Reprogramming spinal cord glia into neurons has the potential to help spinal cord injury and ALS patients regain motor function, improve quality of life, and extend life expectancy.

7 Chapter 2

Materials and Methods

Plasmid Constructs and Transfection

Mouse NeuroD1, Neurogenin2, Lhx3, and Isl1 cDNA was inserted into a pCAG-GFP-IRES-GFP retroviral vector (gift of Dr. Fred Gage) to generate pCAG-NeuroD1-IRES-GFP, pCAG-Ngn2-

IRES-GFP, and pCAG-Dlx2-IRES-GFP. Mouse NeuroD1, Neurogenin2, and Dlx2 sequences were also inserted into an equivalent pCAG-DsRed-IRES-DsRed plasmid. The CAG promoter is a strong mammalian promoter consisting of a cytomegalovirus (CMV) enhancer, a beta-actin promoter, and a rabbit beta-globin splice acceptor. The internal ribosome entry site (IRES) is a translation initiator. Since expression of GFP and DsRed upstream of the IRES in the control but not in the neurogenic vectors, brightness may not be comparable between the two. However, this is not an issue because only positive and negative signals are quantified.

Figure 1. Plasmid constructs encoding GFP and neurogenic factors for retroviral packaging. Left: pCAG-GFP-IRES-GFP plasmid structure. Right: pCAG-NeuroD1-IRES-GFP, where the GFP sequence was replaced with the NeuroD1 sequence. Other plasmids such as pCAG-Ngn2-IRES-GFP and pCAG-DsRed-IRES-DsRed are similar but not shown here. This scheme was provided by Yi Hu. 8 HEK 293T cells were grown to 70% confluence in a 10 cm culture dish and exposed to

14 ug VSVG, 14 ug of the plasmid vector, and 130 uL of PEI in 1 mL OptiMEM for 5 hours.

The medium was collected and replenished 48, 72, and 96 hours after exposure. Virus particles were isolated from the collected medium by centrifugation at 25,000 rpm for 2 hours at 4˚C. The virus pellet was re-suspended in phosphate-buffered solution.

NeuroD1 and control mCherry DNA was also packaged in adeno-associated virus 9

(AAV9) vectors for gene expression in vivo. This work was done by Zifei Pei. Two different

AAV9 vectors were constructed and used simultaneously. A Cre loxP system under astrocyte promotor hGFAP was used to target Cre recombinase expression to GFAP+ cells. The other vector carries inverted (and therefore inactive) sequences of NeuroD1-mCherry or the control sequence, mCherry-mCherry, which are flanked by flip-excision (FLEx) loxP sites for Cre binding. Cre recombinase can bind the loxP sites to revert the to the correct orientation for expression. Theoretically, because Cre recombinase is needed for the expression of NeuroD1 and mCherry, and Cre is only expressed in GFAP+ cells, only GFAP+ cells can express the genes when co-infected by the two vectors. The two vectors were mixed for simultaneous injection in vivo.

Cell Culture of Mouse Astrocytes

Primary cultures of mouse spinal cord astrocytes were obtained from C57BL/6 mice at postnatal day four. The spinal cord tissue was isolated in cold Hanks’ Balanced Salt Solution

(HBSS) with 10% fetal bovine serum (FBS). The tissue was cut into small pieces and serum was removed through three washes with HBSS. The tissue was digested in 0.25% trypsin solution 9 and 100uL DNase for 20 minutes. The trypsin was stopped through three washes with 10% FBS in HBSS. The cells were then suspended in glial medium (MEM with 5% FBS, 1% GlutaMAX,

0.04% penicillin/streptomycin, and 0.04% NaHCO3) and centrifuged at 900 rpm for 5 minutes to wash. The cells were plated in fresh glial medium into 10-mL culture flasks and incubated at

37˚C with 5% CO2 and humidified air. After 16-24 hours, the culture was shaken vigorously and washed with glial medium to remove non-astrocytic cells. After another 24 hours, the medium was changed DMEM/F12 (GIBCO) supplemented with 10% FBS (GIBCO), 0.4% penicillin/streptomycin, 3.5 mM Glucose, B27 (GIBCO), 10 ng/mL epidermal growth factor

(EGF, Invitrogen), and 10 ng/mL fibroblast growth factor 2 (FGF2, Invitrogen). Upon confluence, the cells were subcultured on poly-D-lysine-coated coverslips in 24-well plates.

Virus Transduction

Once spinal cord astrocyte cultures were 90% confluent, the virus suspension was added to the medium. After 24 hours, the medium was replaced with DMEM/F12 supplemented with

0.5% FBS, P/S, N2 and B27 supplements, vitamin C, and Y27632. After 3 days for mouse cells and 5 days for human cells, BDNF (20 ng/mL), NGF (10 ng/mL), and NT3 (10 ng/mL) were added to support neuronal differentiation and survival, and were replenished weekly.

Patch-Clamp Recording of Cell Cultures

The Multiclamp 700A patch-clamp amplifier (Molecular Devices, Palo Alto, CA) was used for whole-cell recordings of the in vitro converted neurons. The cell cultures were submerged in 25 mM HEPES, 30 mM glucose, 128 mM NaCl, 5 mM KCl, 2 mM CaCl2, and 1 10 mM MgCl2 adjusted to pH 7.3 and 315-325 mOsm/L at room temperature. Patching pipettes pulled from borosilicate glass (3-5 MΩ) were filled with 10 mM HEPES, 5 mM Na- phosphocreatine, 135 mM KCl, 2 mM egtazic acid (EGTA), 4 mM MgATP, and 0.5 mM

Na2GTP adjusted to pH 7.3.

For voltage-clamp experiments, the potential across the cell membrane was held at -70 mV or -80 mV, and recorded using the pClamp9 software (Molecular Devices, Palo Alto, CA).

Sodium and potassium currents and action potentials were also analyzed using pClamp9

Clampfit software. The Minianalysis software (Synaptosoft, Decator, GA) was used to analyze spontaneous synaptic events.

Mouse Models and Stereotaxic Viral Injection

Stab Injury Model

Wild type mice were anesthetized with 20 mL/kg of 2.5% Avertin (25 mg/mL

Tribromoethylethanol and 25 uL/mL T-amylalcohol). The back of the mouse was shaved and disinfected using ethanol and betadine. Local anesthesia consisted of a 10 mL/kg subcutaneous injection of 4 mg/kg bupivacaine. After a midline incision along the spine, a laminectomy was performed at T11-12 to expose the dorsal surface of the spinal cord. The stereotaxic setup was used to guide a 27-gauge needle 1 mm into the exposed surface of the spinal cord to create a stab injury. The needle was withdrawn after 3 minutes.

Immediately after injury, 1.5 uL of virus suspended in 1xPBS was injected using a 5 uL

Hamilton syringe and a 34-gauge needle. The needle was guided 1 mm into the injury core, and 11 the virus was slowly injected at 0.1 uL per minute, while the needle was raised manually 0.05 mm/min to allow the virus to diffuse more evenly into the tissue and avoid virus spilling out.

After this procedure, the animals were administered 10 mg/kg carprofen as a post-surgery analgesic through drinking water and closely monitored in their cages, which were heated using a water-heated pad.

SOD1 Mouse Model

SOD1-G93A transgenic mice were purchased from The Jackson Laboratory. Behavioral assessment of SOD1 mice and age-matched controls was performed weekly using a rotarod test, weight measurement, and a behavioral scale.

The rotarod test was performed weekly beginning at 8 weeks of age. The animals were allowed to walk on the rotarod apparatus, which accelerated 3 rpm/min from 2-20 rpm. A cut-off point was set at 360 seconds. The time and speed of the rotarod at the time of fall was recorded, and three trials were recorded for each mouse. To minimize initial variability due to unfamiliarity to the test, the mice were trained on the apparatus by performing the test for three consecutive days in the week prior to the first meaningful data point.

A behavioral score was recorded weekly using the scale described in Table 1. For humane reasons, endpoint was set to when the mice reached a behavioral score of 1 (the paralyzed animal is unable to straighten up from its side within 15 seconds) or when the animal lost more than 20% of its greatest measured weight. Mice showing signs of paralysis were closely monitored and were given gel feed if necessary.

Stereotaxic injection of virus was performed in a similar manner as in the stab injury model. After anesthesia and a midline incision along the spine, a laminectomy was performed at 12 T11-12 to expose the dorsal surface of the upper lumbar spinal cord. 1.5 uL of virus suspended in 1xPBS was injected using a 5 uL Hamilton syringe and a 34-gauge needle. The needle was guided 1 mm into the spinal cord, about 1 mm away from the midline. The virus suspension was slowly injected at 0.1 uL per minute, while the needle was raised manually 0.05 mm/min to allow the virus to diffuse more evenly into the tissue and avoid virus spilling out. After surgery, the animals were administered carprofen through drinking water and closely monitored in their cages, which were heated using a water-heated pad.

Table 1. Behavioral Scale for SOD1 Mouse Model

Rank Behavior

5 Healthy, no symptoms

4 Slight paralysis of hind limbs and destabilized gait

3 Obvious paralysis and destabilized gait

2 Fully developed hind limb paralysis

Endpoint: Animal is unable to straighten up from its side within 15 seconds 1 AND/OR lost >20% of starting weight

Sample Preparation and Immunocytochemistry

For cell cultures

Cell cultures were fixed in 4% paraformaldehyde (PFA) for 10 minutes and washed 3 times in PBS. Non-specific binding sites were blocked with 2.5% normal donkey serum (NDS),

2.5% normal goat serum (NGS), and 0.1% Triton in 1xPBS for 1 hour. Primary antibodies were diluted in blocking solution before adding to the coverslips, and incubated overnight at 4˚C. The 13 cells were washed 3 times with 1xPBS to remove the primary antibody, and incubated for 2 hours in fluorophore-conjugated secondary antibodies (Alexa Fluor 647, 546, and 448, 1:800,

Molecular Probes) that were diluted in blocking solution. The cells were washed 3 times and mounted onto glass slides using ProLong Gold Antifade Mountant with DAPI (Invitrogen).

For tissue samples

Mice were anesthetized with 2.5% Avertin and perfused transcardially using artificial cerebral spinal fluid. After dissection, tissue samples were fixed in 4% PFA for 72 hours at 4˚C before dehydration in 30% sucrose in 1xPBS for 24 hours. The samples are then incased in

Cryomatrix embedding resin (Invitrogen) and stored in 80˚C, or immediately sectioned using the

Leica CM1950 Cryostat at 40 μm thickness. The samples can be stored in phosphate buffer for up to 2 weeks. The sectioned samples are washed 3 times in 1xPBS and permeablized in 2%

Triton in 1xPBS for 10 minutes. Non-specific binding sites are blocked with 5% NDS and 0.3%

Triton in PBS for 1 hour. Triton was removed through 3 washes in 1xPBS before incubating with primary antibody diluted in 5% NDS in 1xPBS for two nights. The samples were washed 3 times with 1xPBS to remove the primary antibody, and incubated for 2 hours in fluorophore- conjugated secondary antibodies that were diluted in 5% NDS in 1xPBS. The cells were washed

3 times and mounted onto glass slides using ProLong Gold Antifade Mountant with DAPI.

Imaging and Analysis

Immunofluoresence images were taken using the Nikon Eclipse TE2000-S microscope and the Zeiss ApoTome.2. Tiling and Z-stack functions were used to obtain images with a wider 14 field and range of depth. Cell counting was done using the cell counter plugin in the ImageJ software.

15 Chapter 3

Results

Before attempting to convert spinal cord glia into neurons in vivo, it is important first to screen for the ideal neurogenic transcription factors and potential combinations in vitro on primary cell cultures of spinal cord glia. The following work describes the cell cultures and the characteristics of neurons generated by the neurogenic factors. The goal is to identify factors that generate mature neurons of different subtypes. These factors will eventually be test in vivo to restore lost neuron populations as a result of spinal cord injury and ALS.

After screening for the ideal factors, the stab injury and SOD1-G93A mouse models were characterized to see if they indeed mimic the pathology of acute spinal cord injury and ALS.

This sets the stage for

Characterization of Mouse Spinal Cord Cell Cultures

Before transducing the primary cultures of spinal cord cells with neurogenic factors, we first identified the cell populations included in the cell cultures to better understand the targeted cells. The subcultures of mouse spinal cord glia were characterized upon 90% confluence in order to assess glial type populations at the time of viral transduction. Quantification after immunostaining for glial and neuronal markers for three different batches of primary culture clearly indicates a mixed culture of glia and some neurons (Figure 2). Astrocyte markers S100β

(51.8%) and GFAP (35.2%) suggest that astrocytes are an abundant cell type. The cells quickly grew to confluence after seeding in the culture plates within 1-2 days, and fraction of cells co- stained for GFAP and Olig2, suggesting that they are reactive glia (Chen et al., 2008). Other 16 Olig2+ cells (31.0%) are likely NG2 cells and oligodendrocyte precursor cells (OPCs), and

Iba1+ cells (7.59%) are likely microglia. There are also a small percentage of endogenous neurons labeled by DCX+ (1.38%) and MAP2+ (1.37%) in the subcultures, suggesting low numbers of neuroblasts and neurons in the culture. In summary, the majority of cells are astrocytes and OPCs with a lesser percentage of microglia. These results show that the primary culture is a mixed culture of spinal cord glia that mimics in vivo conditions.

Figure 2. Characterization of primary cultures of mouse spinal cord glia.

Quantification of glial and neuronal markers shows that astrocytes (S100B+, GFAP+) are most abundant. Images were taken at 200x using the Zeiss ApoTome.2 microscope. Imaging and cell counting was done by Xiaoyun Ding.

Ngn2 and NeuroD1 Can Convert Spinal Cord Glia into Neurons in Vitro

After characterizing the cell cultures, we began to test neurogenic transcription factors to find the best combination to later convert spinal cord glia into neurons in vivo. Initial testing of the factors was done in vitro on cell cultures of mouse spinal cord glia. NeuroD1 and Ngn2 retroviral vectors were added to the primary cultures at 90% confluence, and the morphological changes were monitored after transduction. Images were taken of the live culture during the conversion process to show the morphological changes (2, 6, and 10 days post-infection, shown in Figure 3). The fluorescing cells transduced by NeuroD1 and Ngn2 retroviruses initially show 17 large and flat astrocyte morphology at 2 days post-infection (DPI). By 5-6 DPI, the cells clearly show structural changes and become small and round neuron-like cells with extended neurites by

10 DPI. Cells transduced by the control GFP retrovirus showed no changes in morphology.

Figure 3. Live time-course images of morphological changes of primary spinal cord cell cultures transduced with retrovirus.

Primary mouse spinal cord glia cultures were transduced with GFP, NeuroD1 (ND1), and Ngn2 retroviruses, from left to right. A clear morphological difference is seen by 6 days post-infection (DPI) for ND1 and Ngn2 groups, and become more like mature neurons by 10 DPI.

To confirm the conversion of the glial cells into neurons, immunocytochemistry and quantitative reverse transcription polymerase chain reaction (qRT-PCR) was performed on the cultures (Figure 4). qRT-PCR was performed on cell RNA extracts of the retrovirus-treated cultures two weeks after retrovirus treatment to measure transcription levels of neuronal genes relative to GAPDH. Significant fold increases in immature neuron markers doublecortin (DCX) 18 and beta-3-tubulin (B3T) were observed after reprogramming by NeuroD1 (9.8 and 8.2-fold increases, respectively) and Ngn2 (11.2 and 7.1-fold increases, respectively).

Immunocytochemistry for mature neuron markers showed many converted neurons by two weeks of reprogramming. The cells were fixed in 4% PFA and stained for GFP and mature neuron markers microtubule-associated protein 2 (MAP2) and NeuN. Infection efficiency was quantified through the ratio of GFP+ cells among all cells (DAPI+). Conversion efficiency was quantified through the ratio of MAP2+ cells among GFP+ cells. Quantification of infection and conversion efficiency was obtained through cell counting of four different image fields for each group. Infection efficiency for NeuroD1 and Ngn2 was 17.5% and 37.0%, respectively, and conversion efficiency was 86.6% and 91.5%, respectively.

To further investigate the maturity of the converted neurons, whole-cell recordings were performed on the cultured cells using the Multiclamp 700 patch-clamp amplifier from Molecular

Devices. Repetitive action potentials were observed for the recorded Ngn2 and NeuroD1- reprogrammed cells, though the diminishing potentials suggest that sodium channels are present on the cell membrane, but are low in number. Voltage-clamp recordings to measure ion currents across the cell membrane show existing sodium and potassium currents. Spontaneous events were not detected, suggesting a lack of spontaneously active neural networks formed in the cell culture.

Immunostaining and RT-PCR results clearly support the conversion of mouse spinal cord glia into neurons after reprogramming by NeuroD1 and Ngn2. However, electrophysiology results indicate that the converted neurons are still functionally immature 2-3 weeks after virus transduction. Maintaining the culture for longer may produce fully mature neurons.

19

20 Figure 4. Direct conversion of mouse spinal cord astrocytes into neurons in vitro. A) NeuroD1 and Ngn2-reprogrammed cells (GFP+) show striking neuron morphology after 2 weeks in culture and are positive for MAP2, which is absent in GFP-transduced culture. B) Infection efficiency (GFP+/DAPI+) and conversion efficiency (MAP2+/GFP+) was quantified. Values are averages of four image fields of the cultures depicted in Panel A. C) qRT-PCR results show significant fold increase in transcription of DCX and B3T compared to GAPDH 12 days after reprogramming by NeuroD1 and Ngn2. Averages of three trials are shown ± standard deviation. qRT-PCR was performed by Matt Keefe. D) Neurons converted by Ngn2 (top) and NeuroD1 (bottom) fire repetitively but with diminishing strength. E) Low sodium and potassium currents are observed in neurons converted by Ngn2 (top) and NeuroD1 (bottom). F) A live image of a recorded neuron converted by Ngn2. G) No spontaneous synaptic events were observed in the NeuroD1-converted neurons at 3 weeks. Electrophysiology was performed by Dr. Lei Zhang.

Transcription Factors Drive Differentiation into Specific Neuronal Subtypes

After confirming the identity of our converted neurons, we further explored the specific neuronal subtypes included in the reprogrammed populations. Producing different neuronal subtypes is important, especially in spinal cord injury, to balance excitation and inhibition in reprogrammed neuronal networks. Previous studies demonstrated that NeuroD1 reprogramming can drive glial differentiation to both glutamatergic and GABA-ergic fates in the brain (Guo et al., 2014), and that Dlx2 drives differentiation to a GABA-ergic fate (Heinrich et al., 2010).

Studies also demonstrated Ngn2 reprogramming can drive differentiation to glutamatergic fates in the brain (Heinrich et al., 2010), as well as Ngn2 co-expression with Hb9, Isl1, and Lhx3 in spinal cord motor neurons (Scardigli et al., 2001, Sharma et al., 1998). Hu and Zhang (2009) also demonstrate that addition of retinoic acid (RA) and sonic hedgehog (SHH) to embryonic stem cells can drive differentiation towards spinal motor neuron fates (Hu and Zhang, 2009). Based on this information, Ngn2 reprogramming in combination with addition of RA and smoothened agonist (SAG), a hedgehog pathway agonist, to spinal cord glia cultures was analyzed for motor 21 neuron and glutamatergic fates, NeuroD1 reprogramming was evaluated for GABA-ergic and glutamatergic fates, and NeuroD1 and Dlx2 co-expression in spinal cord glia was evaluated for

GABA-ergic fate. Subtype markers chosen for immunostaining were vesicular glutamate transporter 1 (VGLUT1) to identify glutamatergic neurons, choline acetyltransferase (ChAT) to identify cholinergic and motor neurons, and glutamic acid decarboxylase 65/67 (GAD65,

GAD67) to identify GABA-ergic neurons. Neuron morphology and size was also observed through immunostaining results.

Reprogramming by Ngn2, RA, and SAG resulted largely in VGLUT1 and ChAT positive neurons, suggesting differentiation towards glutamatergic and motor neuron fates (Figure 5).

NeuroD1-reprogrammed neurons largely expressed VGLUT1 and GAD67, suggesting differentiation towards glutamatergic and GABA-ergic fates. Reprogramming by NeuroD1 and

Dlx2 yielded neurons positive for GABA-ergic markers GAD65 and GAD67. Furthermore, the cells reprogrammed by Ngn2 and NeuroD1 show some morphological differences. Some Ngn2- converted neurons seem to have very large cell somas with thick, multipolar process and more complex branching, while NeuroD1-converted neurons have small somas and look mostly bipolar. The large size and complex branching of the Ngn2-coverted neurons is consistent with the characteristics of spinal motor neurons. These results seem to be consistent with our hypotheses based on the previously mentioned findings on neuronal fate determination in the brain and the spinal cord. However, very low numbers of neurons were observed in the 4-week old cultures and some of the 3-week old cultures, making meaningful quantification difficult.

Additional experiments were conducted to further investigate the potential for generating motor neurons (Figure 6) and GABA-ergic neurons (Figure 7) using the spinal cord glia cultures. 22

Figure 5. Converted neurons express neuronal subtype markers after 3 weeks. Mouse spinal cord astrocytes were transduced virus and fixed with 4% PFA after 2-4 weeks. A) Ngn2 expression combined with retinoic acid (RA) and smoothened agonist (SAG) drive neuron differentiation into the cholinergic subtype within 3 weeks. Both acetyl-cholinetransferase (ChAT) and vesicular glutamate transporter 1 (VGLUT1) are expressed in these populations. B) NeuroD1 expression drives neuron differentiation into glutamatergic (VGLUT1) and GABA-ergic (GAD67) subtypes within 3 weeks. C) NeuroD1 and Dlx2 co-expression drives neuron differentiation into the GABA-ergic subtype expressing GAD65 and GAD67 by 2 weeks. 23 Ngn2, Lhx3, and Isl1 were expressed in spinal cord glia cultures, and RA and SAG were added to the culture medium 5 days after retroviral transduction (Figure 6). Cells were fixed 15 days after transduction. For this experiment, Isl1 and Lhx3 DNA sequences were cloned into separate

GFP retroviral vectors, and Ngn2 was cloned into the DsRed retroviral vector. Since both Isl1 and Lhx3 expression are detected by conjugation to the Alexa Fluor 488 secondary antibody, it is not possible to tell if cells are expressing Isl1 only, Lhx3 only, or both Isl1 and Lhx3; this causes

Figure 6. Co-expression of Ngn2, Lhx3, and Isl1 may promote differentiation of mouse spinal cord glia into a cholinergic fate. Culture media was supplemented with RA and SAG 5 days after retroviral transduction, and cultures were fixed after 15 days. Top row panels: Expression of GFP and DsRed controls produces negative results for ChAT. Second and third row panels: Single-factor expression of Isl1 and Lhx3 is insufficient to produce MAP2+ and ChAT+ cells. Bottom row panels: Ngn2, Isl1, and Lhx3 co-expression seems to stain positive for ChAT. However, other nonspecific antibody binding is observed. 24 a weakness in this experiment. This can be corrected in the future by cloning Isl1 and Lhx3 into a single retroviral vector so that they will be expressed together upon transduction.

No MAP2 and ChAT expression was detected in the GFP control group, nor in the Isl1 only and Lhx3 only groups. This suggests that Isl1 and Lhx3 alone cannot reprogram spinal cord glia into neurons. With the addition of Ngn2, cells co-expressing GFP and DsRed with neuron- like morphology and positive ChAT staining were observed. This result suggests that Ngn2, Isl1, and Lhx3 co-expression may reprogram spinal cord glia into cholinergic neurons. However, some cell characteristics are different from previously produced ChAT-positive neurons in

Figure 6; the cell somas are much smaller, suggesting they are not spinal motor neurons, and expression of the neurogenic transcription factors seems low. Also, non-specific binding was observed in many non-neuron-like cells. Further staining for MAP2 and other motor neuron markers may produce more conclusive results. Maintaining the cells longer in culture may also produce more convincing ChAT expression. Efficient conversion of spinal cord glia into motor neurons is important for regenerating neurons in ALS, which causes motor neuron degeneration.

Another experiment was performed to investigate the effect of Dlx2 co-expression with

Ngn2 and NeuroD1, and the best combination to drive reprogramming of spinal cord glia into

GABA-ergic neurons (Figure 7). Previous studies have shown that Dlx2 can help to drive

GABA-ergic differentiation (Heinrich et al., 2010). Primary cultures of mouse spinal cord glia were transduced with NeuroD1 and Dlx2, Ngn2 and Dlx2, NeuroD1 alone, or Ngn2 alone. Cells were fixed with 4% PFA after 2 weeks and stained with anti-GAD65 antibody. GAD65 was clearly expressed in a proportion of the cells in each group; GAD65 staining pattern in the neurons was observed in the soma and as puncta along the neuronal processes. Quantification of

GAD65+ cells among all neuron-like cells is included for the cultures co-transduced with Dlx2. 25

Figure 7. NeuroD1 and Dlx2 co-expression most efficiently reprograms mouse spinal cord glia into GAD65+ neurons. Cells were transduced with NeuroD1 only, Ngn2 only, NeuroD1 and Dlx2, or Ngn2 and Dlx2. Cells were fixed after 2 weeks and stained for GAD65. GAD65+ cells in each group are shown. Quantification of GAD65+ neurons is shown for the coverslips co-transduced with Dlx2 and NeuroD1 or Ngn2. No Ngn2- expressing cells (n=5), 20% of NeuroD1-expressing cells (n=30), 24% of Ngn2/Dlx2 co-expressing cells (n=17), and 60% of NeuroD1/Dlx2 co-expressing cells (n=5) were GAD65+. 26 All neuron-like cells expressing GFP (NeuroD1 or Ngn2 expression only) or both GFP and

DsRed (NeuroD1 or Ngn2 co-expression with Dlx2) were counted.

24% of neurons expressing Ngn2 and Dlx2 were GAD65+, and 60% of neurons expressing NeuroD1 and Dlx2 were GAD65+. No neurons expressing only Ngn2 were GAD65+, while 20% of neurons expressing only NeuroD1 were GAD65+. However, only five neurons were observed for the Ngn2-only and NeuroD1/Dlx2 groups. The number of neurons was very low in these particular cultures perhaps due to a low titer virus, low infection efficiency, or unfavorable culture conditions. This preliminary data suggests that NeuroD1 and Ngn2 can both produce GABA-ergic neurons, and the efficiency is likely enhanced with the addition of Dlx2 expression. Repeating this experiment and improving neuron density can increase confidence in these results and provide a better comparison of NeuroD1 and Ngn2 reprogramming efficiency.

Efficient conversion of spinal cord glia into GABA-ergic neurons is important for applications in spinal cord injury, which causes a loss of all types of neurons, including GABA-ergic interneurons of the dorsal horns.

Characterization of the Spinal Cord Stab Injury Model

It is important to first assess the characteristics of the stab injury mouse model before reprogramming with neurogenic transcription factors. The stab injury was created by performing a laminectomy on the spine of a wild type mouse at the T11-12 position. This exposes the dorsal surface of the spinal cord, which is punctured 1 mm in depth by a 27-gauge needle. The mouse was perfused transcardially after 1 week, and the injured spinal cord tissue was isolated and sectioned horizontally. Horizontal sections are useful in this model in comparison to coronal or 27 sagittal sections to generate more samples that capture the stab injury.

Figure 8. Characterization of the in vivo model for spinal cord injury. One week after stab injury using a 27-gauge needle 1 mm deep into the spinal cord. Immunostaining for neuronal nuclei marker NeuN (green) and glial markers GFAP (red) and Iba1 (cyan) show remarkable differences in cell populations in the non-injured spinal cord (top panels) and at the site of the stab injury (bottom panels). Imaging done by Xiaoyun Ding.

Immunostaining for neuron, astrocyte, and microglial markers showed clear differences between healthy spinal cord tissue and tissue at the site of stab injury. The stab injured tissue shows greatly decreased NeuN signal, which clearly indicates neuron loss at the injury core by one week after injury. Greatly upregulated GFAP and Iba1 signal at the injury site suggests reactive gliosis and hypertrophic activation of astrocytes and microglia around the injury. The morphology of the astrocytes and microglia also show a remarkable change. The enlarged, elongated, and polarized morphology of the GFAP+ astrocytes reflect hypertrophic activation in the process of forming a glial scar, and the abnormally enlarged and dense Iba1+ microglia suggest immune activation and inflammatory responses. These reactions are characteristic of 28 secondary damage described after acute spinal cord injury, indicating that our stab injury model successfully represents the pathology of spinal cord injury.

Characterization of the SOD1 Mouse Model

We have observed that SOD1-G93A mice show clear behavioral deficits as they age. The mice seem healthy when young, but their lower bodies begin to stiffen with the onset of paralysis, starting from their tails. As paralysis worsens, the animals begin to drag their hind limbs along the cage floor, and lose weight due to muscle atrophy. At endpoint, the animals often struggle to move around are not able right themselves easily from their sides. This obvious loss of motor function begins when the mice are around 16 weeks of age, but molecular changes and subtleties in behavioral deficits begin earlier and are better detected by immunostaining and behavioral tests. It is important to understand the pathology of the mice to assess the best timing for reprogramming intervention

The pathology of the SOD1-G93A mice was first assessed by immunostaining and rotarod performance in comparison to age-matched wild-type mice. SOD1 mice were perfused transcardially at behavioral endpoint, which is when animal has difficulty righting itself from its side within 15 seconds, or when it has lost 20% of its greatest body weight. Immunostaining for motor neuron marker ChAT and astrocyte marker GFAP show clear differences in expression levels (Figure 9). There are significantly less ChAT+ cells in the ventral horn of the SOD1 mouse compared to the wild-type mouse, indicating the loss of motor neurons due to disease progression. In contrast, GFAP staining is highly upregulated in SOD1 mice compared to the 29 control, suggesting increase in reactive astrocytes in the SOD1 spinal cord. GFAP is especially upregulated in the white matter of the SOD1 spinal cord.

Figure 9. Characterization of SOD1 mice shows motor neuron loss and reactive astrocytes. Immunostaining for ChAT (red) shows a loss of motor neurons in the ventral horn of the SOD1 mouse spinal cord. GFAP (green) is upregulated in the SOD1 spinal cord, especially in the white matter.

The rotarod test was performed weekly for 4 SOD1 mice and 3 age-matched wild-type mice (Figure 10). The mice were placed on the rotating rod that accelerated from 2 to 20 rpm in

3 minutes. The mice were allowed to run up to 6 minutes for three trials. The mice were trained for the test for three consecutive days at 7 weeks of age, and weekly recording began at 8 weeks of age. The results show that the latency times of SOD1-mutant mice by decrease slightly by 11 to 12 weeks of age, which gives a preliminary idea on the onset of motor deficits. The difference between SOD1-mutant mice and wild type is suggested to be small at best by 12 weeks of age, especially for the low sample sizes. Further work to include larger numbers of animals and 30 carrying out the analysis until the SOD1-mutant mice reach endpoint will provide more information on the significance of these results and show a greater difference as the mice age.

Figure 10. Rotarod performance of SOD1 and age-matched wild-type mice. Average latency of fall for SOD1 mice (n=4) and wild-type mice (n=3) was normalized to the longest latency for each mouse. 7-week old mice were trained on the apparatus for three consecutive days, and the time of fall was recorded from 8-12 weeks of age. The average of three trials per mouse was calculated for each week. The average latency of SOD1 and WT mice are shown.

Stereotaxic Injection of AAV9 in SOD1 Mice

SOD1 mice were injected with NeuroD1 and mCherry-expressing AAV9 vectors. A laminectomy was performed on the T11-12 vertebrae to expose the dorsal surface of the upper lumbar spinal cord. The virus suspension was injected slowly into the spinal cord starting at 1 mm in depth, and the needle was slowly raised during the injection to allow better distribution of the virus through the tissue. The mice were perfused transcardially 2 months after injection of virus.

Immunostaining for NeuroD1 confirms that NeuroD1 is upregulated in cells infected with

AAV9-GFAP::CRE+AAV9-flex-NeuroD1-mCherry virus (Figure 11). The widespread mCherry 31 signal in the spinal cord indicates high infection efficiency of the AAV9 vector, including in the ventral horn where motor neuron degeneration occurs. This confirms that our methods successfully target areas of degeneration due to ALS. Infected cells show much stronger staining for NeuroD1, and are also NeuN+ neurons. This clearly shows the AAV9 vector successfully upregulates NeuroD1 in infected cells. Further experiments may be needed to confirm reprogramming, since injection of the control AAV9-GFAP::CRE+AAV9-flex-mCherry- mCherry virus was also observed to infect neurons through an unknown mechanism. However, this can also be beneficial in helping endogenous neuron survival because of the neuroprotective effects of NeuroD1 in the central nervous system (Gao et al., 2009, Jahan et al., 2010).

Figure 11. NeuroD1 expression through AAV9 vector successfully upregulates NeuroD1. Virus-infected cells (red) have upregulated NeuroD1 (green) and are positive for NeuN (cyan) 2 months after virus injection into the SOD1 spinal cord. 32 To explore the populations infected by the virus, immunostaining for glial markers

GFAP, Iba1, and Olig2 was performed 2 months after injection of the control AAV9-

GFAP::CRE+AAV9-flex-mCherry-mCherry virus. mCherry expression was detected in Olig2+ cells but not clearly in GFAP+ or Iba1+ cells, demonstrating that largely OPCs are targets for reprogramming, and not astrocytes or microglia. There are also many other infected cells that do not express Olig2 and are not yet characterized. This result is important because it shows that

OPCs are the main targets for conversion by AAV9 in the spinal cord, which is different from in the brain, where astrocytes are the most targeted population (Guo et al., 2014).

33

Figure 12. mCherry is expressed in some OPCs and astrocytes. 2 months after injection of mCherry- expressing AAV9 into the SOD1 spinal cord, immunostaining for glial markers was performed. mCherry expression was observed Olig2+ but not GFAP+ or Iba1+ cells. White arrows indicate some (but not all) Olig2+ cells infected by AAV9.

34 Chapter 4

Discussion

Reprogramming Spinal Cord Glia into Neurons in vitro

The results demonstrate that primary cultures of mouse spinal cord glia can be converted into neurons in vitro by NeuroD1 or Ngn2, albeit being only partially functional two to three weeks after conversion. A clear morphological change from glial-like to neuron-like cells occurs within six days after retroviral transduction of the neurogenic factors. The converted neurons expressed mature neuronal proteins NeuN and MAP2 as well as subtype markers within two weeks. NeuroD1 alone can convert mouse spinal cord glia into both glutamatergic and GABA- ergic neurons, and the efficiency of GABA-ergic reprogramming was enhanced by co-expression of Dlx2. Ngn2 alone can also generate glutamatergic neurons in vitro, as well as spinal motor neurons with the addition of retinoic acid and smoothened agonist. Ngn2, Lhx3, and Isl1 co- expression may also be able to convert spinal cord glia into motor neurons at higher efficiency.

Although the converted neurons express mature neuron markers NeuN and MAP2, the pattern of action potentials, low sodium and potassium currents, and lack of spontaneous events in the converted neurons suggest the neurons were not fully functional. However, previous conversion studies of brain astrocytes show that conversion of cortical astrocytes achieves this full functionality within 26-30 days (Guo et al., 2014, Heinrich et al., 2010). Through this comparison, our converted spinal cord neurons are likely to be on their way to full maturation, and our patch-clamp recordings at 14-21 days were done too early to detect full functionality.

Full functionality may be achieved by maintaining healthy cultures for a longer time, which has been a challenge. In vitro converted neurons are highest in number two weeks after transduction, 35 but decrease greatly by four weeks after transduction. Cell survival may be improved by adding

BDNF, NGF and NT3 earlier on and replenishing them more often than once per week. We also observed a layer of unconverted astrocytes that support neuron growth; these astrocytes begin to die as well, leaving empty spaces in the culture where neurons cannot survive. This may be avoided by adding more mouse astrocytes into the culture to replace the lost cells, which may support the converted neurons to survive for a longer time.

Furthermore, cell death may be minimized by also introducing B-cell lymphoma 2 (Bcl-

2), an anti-apoptotic factor that may help cells to survive longer both in vitro and in vivo. In particular, Bcl-2 may be useful for treatment in the ALS model. While the wild-type SOD1 protein is known to associate with Bcl-2, mutant SOD1 can incorporate Bcl-2 into insoluble aggregates, and therefore interferes with the normal anti-apoptotic function of Bcl-2 (Pasinelli et al., 2004). However, caution must be used when overexpressing Bcl-2; although the reprogramming process has been shown to be through direct conversion, Bcl-2 is considered an oncogene and is associated with the risk of cancer (Cory, et al., 2003).

Improving the survival of converted neurons may support more functionally mature neurons not only by allowing more time for the cells to mature, but also by maintaining a sufficient cell density for synapses to form with neighboring neurons. Improved survival will also provide more data for quantification in the future, especially for quantifying subtype marker expression.

The ability of NeuroD1 and Ngn2 to produce neurons of specific subtypes has promising implications in the treatment of spinal cord injury and ALS. Since ALS is characterized by motor neuron loss and astroglial activation, converting glia into spinal motor neurons can not only replenish the population of motor neurons, but also decrease the harmful effects of reactive 36 gliosis. Motor neuron degeneration is localized to the ventral horns of the spinal cord, so is likely most effective to target that area when injecting viral vectors so the converted neurons can mature in the proper in vivo environmental niche. Likewise, in spinal cord injury, all subtypes of neurons are lost, and allowing the neurogenic treatment to infiltrate the entire injured tissue may allow the surrounding spinal cord environment to drive subtype differentiation.

Efficacy of in vivo Models of Spinal Cord Injury and ALS

Characterization of the spinal cord injury and ALS mouse models demonstrate the stab injury and SOD1-G93A mice pathology mimics that of acute spinal cord trauma and ALS. The stab injury inflicted on the spinal cord of wild type mice causes a loss of neurons at the site of the injury, as well as hypertrophic activation of astrocytes and microglia. This glial activation suggests the formation of an astroglial scar and intense immune response at the injury site. These reactions in combination with neuron loss are characteristic of well-described primary and secondary damage after acute spinal cord injury.

Characterization of the SOD1-G93A mouse spinal cords demonstrates an obvious loss of motor neurons in the ventral horns and hypertrophic activation of astrocytes compared to age- matched wild-type mice. The SOD1-G93A mice also show clear behavioral traits that reflect human ALS, including paralysis beginning from the lower extremities and consequential muscle atrophy. The rotarod test to assess motor function detected the onset of subtle behavioral deficits of SOD1-G93A mice compared to wild type between 10-11 weeks of age. This time point offers information for deciding the best time for reprogramming intervention. 37 Another option for modeling spinal cord injury is a contusion model, where injury is inflicted to the spinal cord by an impactor instead of a needle stab. Farooque (2000) has characterized contusion injuries of varying severity and have found severe behavioral deficits in injured mice despite slight recovery over time, and is an effective model for compressive spinal cord injuries (Farooque, 2000). Spinal cord contusion is an option to model a more widespread injury due to impact, which is more applicable to situations like car accidents and falls. Our lab has begun work on this model as well.

Additionally, fine motor deficits may be better assessed when complemented by Catwalk analysis, which is a sensitive machine with a variety of ways of measuring gait, including stride length, hindlimb and forelimb distance, swing speed, and print area. These precise measurements can give greater insight into the motor deficits of ALS mice, and may detect the deficits even earlier than rotarod tests.

One piece of information that is missing from our in vitro studies is the cell types that are targeted by reprogramming. However, in vivo injection of AAV9 was observed to target mostly oligodendrocyte progenitors and some astrocytes. This information in vitro is useful in refining our in vivo methods, as NG2 and GFAP-driven promoters can help to target specific glial cell types in vivo and increase the effectiveness of treatment.

Future Work and Conclusive Remarks

Our results on regenerating neurons through reprogramming in vitro have supported our hypothesis that NeuroD1 and Ngn2 can convert spinal cord glia into neurons, despite reaching only partial functional maturity. Both alone and in combination with other factors, NeuroD1 and 38 Ngn2 can generate glutamatergic, GABA-ergic, and motor neuron subtypes from spinal cord glia. However, more work should be done to extend the survival of the converted neurons to provide more reliable quantification of subtype differentiation. Additionally, preliminary characterization of the stab injury and SOD1-G93A mouse models show that they sufficiently mimic the pathology of human spinal cord injury and ALS. More work also needs to be done here to show statistically significant behavioral differences between the SOD1-mutant mice and wild type. Despite issues related to cell survival and small sample sizes, this preliminary work provides a launch point for further work on in vivo reprogramming in the spinal cord.

Although reprogramming was demonstrated in vitro, much work is yet to be done in vivo to demonstrate the potential of this technology to help patients regain motor function. First, it must be demonstrated that the neurogenic factors expressed through AAV9 vectors can efficiently convert spinal cord glia into neurons after injection in vivo. Although the in vitro work is done to best mimic in vivo conditions, the environments are still very different between a culture dish and the uncontrolled conditions of live tissue, making translation from in vitro to in vivo rather unpredictable. In both mouse models, the virus is injected 1 mm into the spinal cord, which is deep enough to allow the virus to reach as far as the ventral horn, and NeuroD1 was successfully upregulated in the ALS model after high infection efficiency by AAV9. This successfully paves the way for further testing of different factors in all areas of the spinal cord.

Since NeuroD1 and Ngn2 can both generate various types of neurons, we hope to observe that injecting just a single factor can generate the appropriate neuron subtypes in the correct spinal cord regions with the help of factors and cell contact in the local environmental niche.

Most importantly, we must demonstrate the ability for in vivo-reprogrammed neurons to survive long term and integrate into endogenous neuronal networks. This can be detected 39 through immunostaining for synaptic markers such as synaptic vesicle protein 2 (SV2), which allows visualization of synapses along axons. This reconstruction of neuronal networks is crucial for patients to recover motor function. The ideal timing of intervention is also important in optimizing the effectiveness of treatment; well-established glial scars and far-gone degeneration of neuron networks will not respond as well to reprogramming compared to earlier stages.

Ideally, human onset of ALS should be detected as early as possible through current genetic screening technology.

Furthermore, the injection of viral vectors is an invasive procedure that carries risks associated with the viruses and surgical procedures. The rise in research on using small molecules to reprogram cells through activation and inhibition of gene-expressing pathways offers potential for developing drug treatments for spinal cord injury and ALS. 40

BIBLIOGRAPHY

1. Ackery, A., Tator, C., & Krassioukov, A. (2004). A global perspective on spinal cord

injury epidemiology. Journal of Neurotrauma, 21(10), 1355–1370.

2. Boillée, S., Yamanaka, K., Lobsiger, C. S., Copeland, N. G., Jenkins, N. A., Kassiotis,

G., Kollias, G., Cleveland, D. W. (2006). Onset and progression in inherited ALS

determined by motor neurons and microglia. Science, 312(5778), 1389–1392.

3. Chen, Y., Miles, D. K., Hoang, T., Shi, J., Hurlock, E., Kernie, S. G., & Lu, Q. R. (2008).

The Basic Helix-Loop-Helix Transcription Factor Olig2 is Critical for Reactive Astrocyte

Proliferation after Cortical Injury. Journal of Neuroscience, 28(43), 10983–10989.

4. Cleveland, D. W., & Rothstein, J. D. (2001). From Charcot to Lou Gehrig: Deciphering

Selective Motor Neuron Death in ALS. Nature Reviews, 2, 806–819.

5. Cory, S., Huang, D. C. S., & Adams, J. M. (2003). The Bcl-2 family: roles in cell

survival and oncogenesis. Oncogene, 22, 8590–8607.

6. Doble, A. (1996) The pharmacology and mechanism of action of riluzole. Neurology, 47,

233-241.

7. Farooque, M. (2000). Spinal cord compression injury in the mouse: presentation of a

model including assessment of motor dysfunction. Acta Neuropathology, 100, 13–22.

8. Fitch, M. T., Doller, C., Combs, C. K., Landreth, G. E., Silver, J. (1999) Cellular and

Molecular Mechanisms of Glial Scarring and Progressive Cavitation: In Vivo and In

Vitro Analysis of Inflammation-Induced Secondary Injury after CNS Trauma. Journal of

Nueroscience, 19(19), 8182-8198. 41 9. Gao, Z., Ure, K., Ables, J. L., Lagace, D. C., Nave, K.-A., Goebbels, S., Eisch, A. J.,

Hsieh, J. (2009). Neurod1 is essential for the survival and maturation of adult-born

neurons. Nature Neuroscience, 12(9), 1090–1092.

10. Gros-Louis, F., Gaspar, C., & Rouleau, G. A. (2006). Genetics of familial and sporadic

amyotrophic lateral sclerosis. Biochimica et Biophysica Acta, 1762(11–12), 956–72.

11. Guo, Z., Zhang, L., Wu, Z., Chen, Y., Wang, F., & Chen, G. (2014). In vivo direct

reprogramming of reactive glial cells into functional neurons after brain injury and in an

Alzheimer’s disease model. Cell Stem Cell, 14(2), 188–202.

12. Han, D. W., Tapia, N., Hermann, A., Hemmer, K., Hoing, S., Arauzo-Bravo, M. J.,

Zaehres, H., Wu, G., Frank, S., Moritz, S., Greber, B., Yang, J. H., Lee, H. L.,

Schwamborn, J. C., Storch, A., Scholer, H. R. (2012). Direct reprogramming of

fibroblasts into neural stem cells by defined factors. Cell Stem Cell, 10(4), 465–472.

13. Heinrich, C., Blum, R., Gascón, S., Masserdotti, G., Tripathi, P., Sánchez, R., Tiedt, S.,

Schroeder, T., Gotz, M., Berninger, B. (2010). Directing astroglia from the cerebral

cortex into subtype specific functional neurons. PLoS Biology, 8(5).

14. Hu, B., & Zhang, S. (2009). Differentiation of spinal motor neurons from pluripotent

human stem cells. Nature Protocols, 4(9), 1295–1304.

15. Jahan, I., Kersigo, J., Pan, N., Fritzsch, B. (2010). Neurod1 regulates survival and

formation of connections in mouse ear and brain. Cell Tissue Research, 341(1), 95-110.

16. Kretschmer, B. D., Kratzer, U., & Schmidt, W. J. (1998). Riluzole, a glutamate release

inhibitor, and motor behavior. Naunyn-Schmiedeberg’s Archives of Pharmacology,

358(2), 181–190. 42 17. Lee, J. E. (1997). Basic helix-loop-helix genes in neural development. Current Opinion

in Neurobiology, 7, 13–20.

18. Li, H., & Chen, G. (2016). In Vivo Reprogramming for CNS Repair: Regenerating

Neurons from Endogenous Glial Cells. Neuron, 91(4), 728–738.

19. Li, X., Zuo, X., Jing, J., Ma, Y., Wang, J., Liu, D., Zhu, J., Du, X., Xiong, L., Du, Y., Xu,

J., Xiao, X., Wang, J., Chai, Z., Zhao, Y., Deng, H. (2015). Small-Molecule-Driven

Direct Reprogramming of Mouse Fibroblasts into Functional Neurons. Cell Stem Cell,

17(2), 195–203.

20. Liu, J., Lillo, C., Jonsson, P. A., Velde, C. Vande, Ward, C. M., Miller, T. M.,

Subramamiam, J. R., Rothstein, J. D., Marklund, S., Andersen, P. M., Brannstrom, O. G.,

Wong, P. C., Williams, D. S., Cleveland, D. W. (2004). Toxicity of familial ALS-linked

SOD1 mutants from selective recruitment to spinal mitochondria. Neuron, 43(1), 5–17.

21. Lu, P., Wang, Y., Graham, L., McHale, K., Gao, M., Wu, D., Brock, J., Blesch, A.,

Rosenzweig, E.S., Havton, L.A., Zheng, B., Conner, J.M., Marsala, M., Tuszynski, M.H.

(2012) Long-distance growth and connectivity of neural stem cells after severe spinal

cord injury. Cell, 150(6):1264-73.

22. Lu, P., Woodruff, G., Wang, Y., Graham, L., Hunt, M., Wu, D., Boehle, E., Ahmad, R.,

Poplawski, G., Brock, J., Goldstein, L. S. B., Tuszynski, M. H. (2014). Long-Distance

Axonal Growth from Human Induced Pluripotent Stem Cells After Spinal Cord Injury.

Clinical Lymphoma, 83(4), 789–796.

23. McDonald, J. W., & Sadowsky, C. (2002). Spinal cord injury. The Lancet, 359, 417–425.

24. Nakatomi, H., Kuriu, T., Okabe, S., Yamamoto, S. ichi, Hatano, O., Kawahara, N.,

Tamura, A., Kirino, T., Nakafuku, M. (2002). Regeneration of hippocampal pyramidal 43 neurons after ischemic brain injury by recruitment of endogenous neural progenitors.

Cell, 110(4), 429–441.

25. National Spinal Cord Injury Statistical Center, Facts and Figures at a Glance.

Birmingham, AL: University of Alabama at Birmingham, 2016.

26. Pasinelli, P., Belford, M. E., Lennon, N., Bacskai, B. J., Hyman, B. T., Trotti, D., &

Brown, R. H. (2004). Amyotrophic Lateral Sclerosis-Associated SOD1 Mutant Proteins

Bind and Aggregate with Bcl-2 in Spinal Cord Mitochondria. Neuron, 43, 19–30.

27. Pataskar, A., Jung, J., Smialowski, P., Noack, F., Calegari, F., Straub, T., & Tiwari, V. K.

(2016). NeuroD1 reprograms chromatin and transcription factor landscapes to induce the

neuronal program. EMBO Journal, 35(1), 24–45.

28. Pekny, M., Wilhelmsson, U., & Pekna, M. (2014). The dual role of astrocyte activation

and reactive gliosis. Neuroscience Letters, 565, 30–38.

29. Ray, S., Hui, J. L., Leiter, A. B. (2016). Analysis of Neurod1 Bound Sites Across the

Genome Reveals That Enteroendocrine Cell-Specific Gene Expression May Be

Associated With Epigenetic Regulatory Mechanisms Involving H3K9 Acetylation.

Gastroenterology, 150(4), 30.

30. Riviere, M., Meininger, V., Zeisser, P., & Munsat, T. (1998). An analysis of extended

survival in patients with amyotrophic lateral sclerosis treated with riluzole. Arch Neurol,

55(4), 526–528.

31. Sandoe, J., & Eggan, K. (2013). Opportunities and challenges of pluripotent stem cell

neurodegenerative disease models. Nat Neurosci, 16(7), 780–789. 44 32. Scardigli, R., Schuurmans, C., Gradwohl, G., & Guillemot, F. (2001). Crossregulation

between Neurogenin2 and Pathways Specifying Neuronal Identity in the Spinal Cord.

Neuron, 31(2), 203–217.

33. Sharma, K., Sheng, H.Z., Lettieri, K., Li, H., Karavanov, A., Potter, S., Westphal, H., and

Pfaff, S.L. (1998). LIM homeodomain factors Lhx3 and Lhx4 assign subtype identities

for motor neurons. Cell 95, 817–828.

34. Sofroniew, M. V. (2015). Astrocyte barriers to neurotoxic inflammation. Nature Reviews.

Neuroscience, 16(5), 249–63.

35. Sohur, S. U., Emsley, J. G., Mitchell, B. D., Kempermann, G., & Macklis, J. D. (2005).

Adult neurogenesis and repair of the adult CNS with neural progenitors, precursors, and

stem cells. Progress in Neurobiology, 75(5), 321–341.

36. Su, Z., Niu, W., Liu, M. L., Zou, Y., & Zhang, C. L. (2014). In vivo conversion of

astrocytes to neurons in the injured adult spinal cord. Nature Communications, 5, 3338.

37. Takahashi, K., Tanabe, K., Ohnuki, M., Narita, M., Ichisaka, T., Tomoda, K., &

Yamanaka, S. (2007). Induction of Pluripotent Stem Cells from Adult Human Fibroblasts

by Defined Factors. Cell, 131(5), 861–872.

38. Takahashi, K., & Yamanaka, S. (2006). Induction of Pluripotent Stem Cells from Mouse

Embryonic and Adult Fibroblast Cultures by Defined Factors. Cell, 126(4), 663–676.

39. Thier, M., Wörsdörfer, P., Lakes, Y. B., Gorris, R., Herms, S., Opitz, T., Seirferling, D.,

Quandel, T., Hoffmann, P., Nothen, M. M., Brustle, O., Edenhofer, F. (2012). Direct

conversion of fibroblasts into stably expandable neural stem cells. Cell Stem Cell, 10(4),

473–479. 45 40. Trotti, D., Rolfs, A., Danbolt, N. C., Brown, R. H., & Hediger, M. A. (1999). SOD1

mutants linked to amyotrophic lateral sclerosis selectively inactivate a glial glutamate

transporter. Nature Neuroscience, 2(5), 427–433.

41. Vercelli, A., & Boido, M. (2015). Spinal Cord Injury. In Neurobiology of Brain

Disorders (pp. 207–218).

42. Vierbuchen, T., Ostermeier, A., Pang, Z. P., Kokubu, Y., Sudhof, T. C., & Wernig, M.

(2010). Direct conversion of fibroblasts to functional neurons. Nature, 463, 1035–1041.

43. Yamashita, T., Ninomiya, M., Hernandez Acosta, P., Garcia-Verdugo, J. M., Sunabori,

T., Sakaguchi, M., Adachi, K., Kojima, T., Hirota, Y., Kawase, T., Araki, N., Abe, K.,

Okano, H., Sawamoto, K. (2006). Subventricular Zone-Derived Neuroblasts Migrate and

Differentiate into Mature Neurons in the Post-Stroke Adult Striatum. Journal of

Neuroscience, 26(24), 6627–6636.

44. Yamanaka, K., Chun, S. J., Boillee, S., Fujimori-Tonou, N., Yamashita, H., Gutmann, D.

H., Takahashi, R., Misawa, H., Cleveland, D. W. (2008). Astrocytes as determinants of

disease progression in inherited amyotrophic lateral sclerosis. Nature Neuroscience,

11(3), 251–3.

45. Zarei, S., Carr, K., Reiley, L., Diaz, K., Guerra, O., Altamirano, P. F., Pagani, W., Lodin,

D., Orozco, G., Chinea, A. (2015). A comprehensive review of amyotrophic lateral

sclerosis. Surgical Neurology International, 6(171).

46. Zhang, L., Yin, J. C., Yeh, H., Ma, N. X., Lee, G., Chen, X. A., Wang, Y., Lin, L., Chen,

L., Jin, P., Wu, G., Chen, G. (2015). Small Molecules Efficiently Reprogram Human

Astroglial Cells into Functional Neurons. Cell Stem Cell, 17(6), 735–747. 46 47. Zhang, M., Lin, Y. H., Sun, Y. J., Zhu, S., Zheng, J., Liu, K., Cao, N., Li, K., Huang, Y.,

Ding, S. (2016). Pharmacological reprogramming of fibroblasts into neural stem cells by

signaling-directed transcriptional activation. Cell Stem Cell, 18(5), 653–667.

ACADEMIC VITA

ALICE CAI [email protected]

EDUCATION BS The Pennsylvania State University University Park, PA The Schreyer Honors College May 2017 Biochemistry and Molecular Biology, Immunology and Infectious Disease Honors in Biology

HONORS/AWARDS Braddock Scholarship, Eberly College of Science 2013-2017 Provost Award, Pennsylvania State University 2013-2017 Ronald Venezie Scholarship in Science for Honors Education, Eberly College of Science 2015-2016

RESEARCH EXPERIENCE Research Assistant under Dr. Gong Chen, Penn State Department of Biology University Park, PA  Contributed to two projects: 1) Beneficial effects of in vivo reprogramming 2013–Present of astroglial cells into neurons for brain repair, 2) Reprogramming of astroglial cells into neurons for spinal cord injury repair  Performed primary cell culture from neonatal mice, mouse behavioral tests, immunostaining, tissue sectioning, perfusion, virus production, DNA isolation, restriction enzyme digestion, PCR CHOP Research Institute Summer Scholars Program Philadelphia, PA Research Assistant under Dr. William Peranteau, The Children’s Hospital of Philadelphia Summer 2016  Developing in utero gene therapy for patients with Friedreich’s ataxia using CRISPR/Cas9 Schreyer Honors College MD/PhD Summer Exposure Program Hershey, PA Research Assistant under Dr. James Connor, Hershey Medical Center Summer 2014  Project on finding iron transport-related biomarkers for Parkinson’s Disease

PUBLICATIONS/PRESENTATIONS Paper (under review), “Reversing glial scar back to neural tissue by a single transcription factor NeuroD1“

Poster, “In Utero Therapies for Friedreich’s Ataxia.” Cai, A., Hartman, H., Mejaddam, A., Clark, E., Lin, H., Zoltick, P., Lynch, D., Peranteau, W. Presented at the CHOP Research Institute Summer Scholars Program Poster Exhibition.

Poster, “Reprogramming astrocytes into functional neurons for spinal cord repair.” Cai, Alice; Ding, Xiaoyun; Zhang, Lei; Li, Hedong; Chen, Gong. Presented at the Penn State Undergraduate Exhibition, April 2016

Poster, “Investigating Iron-Handling Proteins in the Development of Parkinson’s Disease.” Cai, Alice; Snyder, Amanda M.; Huang, XueMei; Connor, James R. Presented at the Summer Undergraduate Research Symposium, Penn State School of Medicine, August 2014.

CLINICAL EXPERIENCE The Children’s Hospital of Philadelphia, Philadelphia, PA Summer 2016

Hershey Medical Center, Hershey, PA Summer 2014 Penn State Hershey Medical Group , State College, PA Summer 2012

EXTRACURRICULAR INVOLVEMENT Cofounder of the Undergraduate Research Society, Penn State University University Park, PA  Promoted undergraduate involvement in research through hosting research talks 2014 –Present and mentorship  Organized year-long lab rotation program for undergraduates in science and engineering Yoga Teacher, Yoga Lab State College, PA  RYT 200: completed a 200-hour Yoga Teacher Training Fall 2016 Global Medical Brigades, Penn State University University Park, PA  Service trip to Darien, Panama to set up clinic for locals with limited health care access 2013-Present  Assisted in patient in-take, triage, medication packaging, and health education

COMMUNITY SERVICE Centre County Youth Service Bureau, Bellefonte, PA Fall 2016-Present Hearts for Homeless, State College, PA Fall 2016-Present Centre County Promotion of Animal Welfare and Safety (P.A.W.S.) , State College, PA Spring 2015 Mount Nittany Medical Center, State College, PA 2011 – 2014