Developing Gene Therapy Methods to Inhibit a BAG Family Co-Chaperone Protein in Rat Substantia Nigra

Stanley Li

A thesis submitted in conformity with the requirements for the degree of Master of Science Department of Laboratory Medicine & Pathobiology University of Toronto

Developing Gene Therapy Methods to Inhibit a BAG Family Co-Chaperone Protein in Rat Substantia Nigra

Stanley Li

Master of Science

Department of Laboratory Medicine & Pathobiology University of Toronto

2016 Abstract

Parkinson’s disease (PD) is a progressive neurodegenerative disorder affecting dopaminergic neurons in the substantia nigra (SN). Bcl2-associated athanogene 5 (BAG5) is a co-chaperone that promotes dopaminergic cell death in animal models of PD and dopaminergic degeneration. BAG5 also inhibits Hsp70 and parkin, which are protective in animal models of

PD, and interacts with several other proteins implicated in PD pathogenesis. Thus, BAG5 represents a promising therapeutic target for PD. This thesis describes the development of a viral vector that can deliver shRNA against BAG5 to rat SN. We identified shRNAs that efficiently knocked down BAG5 in two cell lines and packaged them into lentivirus (LV) and adeno- associated virus (AAV) 1/2 vectors. AAV-shBAG5 demonstrated superior expression in SN compared to LV. In rat SN, AAV-shBAG5 produced significant BAG5 knockdown and protected neurons from axotomy injury. This work provides the foundation for future studies of

BAG5 knockdown in rat models of PD.

Acknowledgments

I am pleased that my graduate studies have proven to be a challenging and worthwhile endeavour. I reflect on the past two years with great appreciation for the many skills and insights

I have gained, about science and about myself. As my graduate school journey comes to a close,

I would like to especially acknowledge the following individuals who have made my experience such a positive one:

• Dr. Suneil Kalia - For his endless patience and guidance in my pursuits both inside and

outside the lab. I have been extremely privileged to work with such an empathetic and

supportive supervisor.

• Hien Chau - For her technical expertise and advice, and for teaching me very nearly

everything I know about how to function in a lab.

• Drs. James Eubanks and Joel Watts, my committee members - For their constant

encouragement and valuable input on the direction of this project.

• Dr. Lorraine Kalia - For her help with developing the ImageJ method described in

Chapter 2, and for providing feedback on experiments.

• My parents and family - For their unwavering support and encouragement of my studies

in science.

iii

Contributions

Hien Chau - helped perform AAV injection and MFBx surgeries; contributed to Figs. 2, 3, 5, and 11

Drs. Darius Ebrahimi-Fakhari & Mustafa Sahin - contributed to Fig. 5

Christopher Lozano - contributed to Fig. 7

Erik Friesen - contributed to Fig. 11

Mitch de Snoo - helped perform AAV injection surgeries

Drs. Ornella Pellerito & Sherri Thiele - helped perform lentivirus injection surgeries

Table of Contents

ACKNOWLEDGMENTS ...... III

CONTRIBUTIONS ...... IV

TABLE OF CONTENTS ...... V

LIST OF ABBREVIATIONS ...... VIII

LIST OF FIGURES ...... IX

LIST OF TABLES ...... X

INTRODUCTION ...... 1

0.1 PARKINSON’S DISEASE OVERVIEW ...... 1

0.2 GENE THERAPY FOR PARKINSON’S DISEASE ...... 2

0.3 MEDIAL FOREBRAIN BUNDLE AXOTOMY ...... 3

0.4 BAG PROTEINS ...... 4

0.5 RATIONALE FOR TARGETING BAG5 ...... 6

0.6 RNA-TARGETING STRATEGIES FOR GENE INHIBITION THERAPY ...... 9 0.6.1 RNA interference ...... 9 0.6.2 Antisense oligonucleotides ...... 10

0.7 DNA-TARGETING STRATEGIES FOR GENE INHIBITION THERAPY ...... 11 0.7.1 Cre/loxP recombination ...... 11 0.7.2 TALEN ...... 12 0.7.3 Zinc finger nucleases ...... 13 0.7.4 CRISPR/Cas ...... 13

0.8 VIRAL VECTORS FOR GENE DELIVERY ...... 14 0.8.1 Lentivirus ...... 14 0.8.2 Adeno-associated virus ...... 15

0.9 RESEARCH OBJECTIVES ...... 16

CHAPTER 1: GENERATION OF VIRAL VECTORS EXPRESSING BAG5-TARGETING SHRNA ...... 17

1.1 INTRODUCTION ...... 17

1.2 MATERIALS AND METHODS ...... 17 1.2.1 BAG5 shRNA plasmid construction ...... 17 1.2.2 Cell culture and transfection ...... 18 v

1.2.3 Immunoblotting and analysis ...... 18 1.2.4 Adeno-associated virus generation ...... 19 1.2.5 Lentivirus generation ...... 19 1.2.6 Rat primary hippocampal neuron culture and infection ...... 20

1.3 RESULTS ...... 20 1.3.1 Screen of candidate BAG5 shRNA plasmids ...... 20 1.3.2 Confirmation of top three candidate shRNA plasmids ...... 27 1.3.3 Generation of AAV1/2 and lentivirus vectors expressing BAG5 shRNA ...... 29 1.3.4 AAV-shBAG5 knocks down BAG5 in rat primary hippocampal neurons ...... 29

1.4 DISCUSSION ...... 32 1.4.1 Choice of shRNA as gene inhibition strategy ...... 32 1.4.2 BAG5 antibody validation ...... 33 1.4.3 Transfection efficiency and interpretation of apparent knockdown ...... 33 1.4.4 Limitations of western blot and densitometry analysis ...... 36 1.4.5 Alternative techniques to quantify knockdown ...... 36

CHAPTER 2: CHARACTERIZATION OF BAG5 SHRNA VIRAL VECTORS IN VIVO ...... 37

2.1 INTRODUCTION ...... 37

2.2 MATERIALS AND METHODS ...... 38 2.2.1 Animals ...... 38 2.2.2 Stereotactic injection of AAV or LV into rat SN ...... 38 2.2.3 Medial forebrain bundle axotomy ...... 39 2.2.4 Processing of rat brains ...... 39 2.2.5 Immunofluorescence staining of rat brain sections for TH and BAG5 ...... 39 2.2.6 Immunostaining of rat brain sections for TH and NeuN ...... 40 2.2.7 Semi-quantification of BAG5 expression in SN using ImageJ ...... 41 2.2.8 Statistical analyses ...... 41

2.3 RESULTS ...... 43 2.3.1 AAV1/2 achieves greater expression in rat SN than LV ...... 43 2.3.2 AAV-shBAG5 vectors express throughout SN at level of MTN ...... 45 2.3.3 AAV-shBAG5 vectors at 1/2 dilution knock down BAG5 in vivo ...... 47 2.3.4 AAV-shBAG5 vectors at 1/3 dilution knock down BAG5 in vivo ...... 51 2.3.5 AAV-shBAG5 vectors protect against MFBx injury ...... 55

2.4 DISCUSSION ...... 57 vi

2.4.1 Rationale for selecting AAV1/2 serotype and injection titers ...... 57 2.4.2 Comparison of AAV to LV ...... 57 2.4.3 Spatial expression of AAV and analysis at level of MTN ...... 58 2.4.4 ImageJ fluorescence semi-quantification method ...... 60 2.4.5 Analysis of BAG5 knockdown in rat SN ...... 61 2.4.6 Alternative methods to measure BAG5 knockdown in vivo ...... 62 2.4.7 Potential toxicity from GFP overexpression ...... 63 2.4.8 Effects of AAV-shBAG5 on TH expression ...... 65 2.4.9 Effects of pre-injected AAV-shBAG5 following MFBx ...... 66

CHAPTER 3: GENERAL DISCUSSION AND CONCLUSION ...... 67

3.1 SUMMARY ...... 67

3.2 LIMITATIONS ...... 68

3.3 FUTURE DIRECTIONS ...... 69 3.3.1 Human A53T-SNCA overexpression model of PD ...... 69 3.3.2 VenusYFP-SNCA bimolecular protein complementation system ...... 69

3.4 CONCLUSION ...... 70

REFERENCES ...... 71

vii

List of Abbreviations

AAV = adeno-associated virus MOI = multiplicity of infection ANOVA = analysis of variance MTN = medial terminal nucleus AP = anteroposterior NHEJ = non-homologous end joining ASO = antisense oligonucleotide PBS = phosphate buffered saline BAG5 = Bcl2-associated athanogene 5 PD = Parkinson’s disease BGH = bovine growth hormone PEI = polyethyleneimine bp = base pairs PGK = phosphoglycerate kinase CBA = chicken beta actin PINK1 = PTEN-induced putative kinase 1 CHIP = Carboxyl-terminus of Hsp70 pre-mRNA = precursor mRNA interacting protein qPCR = quantitative polymerase chain CMV = cytomegalovirus reaction CRISPR = clustered regularly interspaced RISC = RNA-induced silencing complex short palindromic repeats RNAi = RNA interference DAB = 3,3′-Diaminobenzidine ROI = region of interest DIV = days in vitro SEM = standard error of the mean DSB = double stranded break shRNA = short hairpin RNA dsRNA = double-stranded RNA siRNA = small interfering RNA DV = dorsoventral SN = substantia nigra FACS = fluorescence activated cell sorting SNc = substantia nigra pars compacta GAPDH = glyceraldehyde 3-phosphate SNr = substantia nigra pars reticulata dehydrogenase SSB = single stranded break GDNF = glial cell line-derived neurotrophic ssDNA = single-stranded DNA factor ssRNA = single-stranded RNA GFP = green fluorescent protein TALE = transcription activator-like effector gp/mL = genomic particles per milliliter TALEN = transcription activator-like HDR = homology-directed repair effector nuclease IRES = internal ribosomal entry site TH = tyrosine hydroxylase LRRK2 = Leucine-rich repeat kinase 2 VTA = ventral tegmental area LV = lentivirus WPRE = woodchuck post-transcriptional MFB = medial forebrain bundle regulatory element MFBx = medial forebrain bundle axotomy ZFN = zinc finger nuclease miRNA = micro RNA αSyn = alpha synuclein ML = mediolateral

viii

List of Figures

Introduction

Figure 1: BAG5 interactors in Parkinson’s disease pathogenesis

Chapter 1: Generation of viral vectors expressing BAG5-targeting shRNA

Figure 2: Targeting location of candidate shRNAs along rat BAG5 mRNA transcript

Figure 3: Initial in vitro screen of knockdown efficiency of candidate BAG5 shRNAs

Figure 4: Confirmation of efficient BAG5 knockdown by top three shRNAs

Figure 5: Expression and BAG5 knockdown by AAV-shBAG5 in rat primary hippocampal neurons

Chapter 2: Characterization of BAG5 shRNA viral vectors in vivo

Figure 6: Demonstration of ImageJ semi-quantification method

Figure 7: GFP expression from AAV vs. LV vectors in rat SN

Figure 8: AAV-shBAG5 coverage of SN at level of MTN

Figure 9: BAG5 knockdown in SN by AAV-shBAG5 at 1/2 dilution (5.5×1011 gp/mL)

Figure 10: BAG5 knockdown in SN by AAV-shBAG5 at 1/3 dilution (3.7×1011 gp/mL)

Figure 11: Effect of AAV-shBAG5 on TH and NeuN expression at 7 days post-MFBx

List of Tables

Chapter 1: Generation of viral vectors expressing BAG5-targeting shRNA

Table 1: BAG5 shRNA targeting sequences and positions

x Introduction 0.1 Parkinson’s disease overview

Parkinson’s disease (PD) is a common and progressive neurodegenerative disorder that affects 1% of people over age 601. With the aging population and increasing life expectancy, PD is expected to become an even greater public health issue in the near future2.

PD is associated with the misfolding and aggregation of α-synuclein (αSyn) and the accompanying loss of dopaminergic neurons in the substantia nigra pars compacta (SNc). After a majority of these neurons are lost, motor symptoms including tremor, rigidity, bradykinesia, and postural instability begin to manifest. PD patients often also experience a variety of other non- motor symptoms, which include sleep disturbances, impulse control issues, and hallucinations2,3.

Most cases of PD are of unclear etiology (sporadic), but about 10% of cases are the result of mutations in specific genes. Autosomal dominant PD can be caused by mutations in SNCA

(which encodes αSyn), LRRK2, VPS35, EIF4G1, DNAJC13, CHCHD2, and the recently discovered TMEM2302,4. Parkin, PINK1, and DJ-1 are responsible for autosomal recessive PD.

The identification of these genes has shed light on the molecular events and systems involved in

PD pathogenesis, which include protein aggregation, the ubiquitin-proteasome system, the lysosome-autophagy pathway, mitophagy, and synaptic function2.

Although dopamine replacement drugs and deep-brain stimulation can effectively treat many of the motor symptoms, they do not change the progression of the disease2,5. Thus, the development of disease-modifying and neuroprotective therapeutics remains an important goal of

PD research2,5.

0.2 Gene therapy for Parkinson’s disease

Gene therapy is a therapeutic technique in which nucleic acids are introduced into cells in vivo to treat or prevent disease. It can be used to replace a mutated gene in monogenic disorders, add gene(s) to help treat complex disorders, or modify endogenous gene expression by targeting

RNA6.

Gene therapy is a promising therapeutic strategy for PD. One gene therapy strategy that has been explored in animal models of PD is to drive increased expression of neuroprotective proteins. Many of these studies have reported favourable effects. Gene transfer of Hsp70 into SN was found to protect dopaminergic neurons in mice treated with MPTP, a toxin that selectively kills dopaminergic neurons7. Rocha et al found that glucocerebrosidase overexpression in SN promoted clearance of αSyn aggregates and preserved dopaminergic neurons in both mice and rats overexpressing human αSyn8. Promotion of the expression of growth factors such as vascular endothelial growth factor9, glial cell-line derived neurotrophic growth factor (GDNF)10, and neurturin (a GDNF family member)11, has also demonstrated protection of dopaminergic neurons in SN in toxin-based animal models.

A different gene therapy strategy of reducing expression of harmful proteins using shRNA has been investigated as well. In rat models of PD, targets for silencing have included

αSyn12,13, ROCK214, and interleukin-1 receptor15. One study using a human αSyn overexpression model reported that knockdown of human αSyn rescued forelimb deficit, but also led to enhanced dopaminergic cell death, possibly due to chronic shRNA (or GFP) expression12.

Another study found that knockdown of endogenous rat αSyn in the rotenone model of PD preserved dopaminergic neurons and rescued forelimb deficit, without causing toxicity13. Saal et al experimented with toxin-based models of PD and reported that shRNA against ROCK2

protected midbrain neurons, upregulated pro-survival markers, and improved motor behaviour14.

However, interleukin-1 receptor knockdown failed to demonstrate any therapeutic effect in the 6- hydroxydopamine model of PD15.

Results from human clinical trials of gene therapy for PD are just beginning to emerge.

These trials have focused on expressing neurotrophic factors or neurotransmitter synthesis enzymes, as the complex etiology of PD makes it difficult to pinpoint one defective gene to target16. One study used AAV2 vector to bilaterally deliver glutamate decarboxylase I to the subthalamic nucleus. It was found to be a safe procedure and resulted in a significant improvement in clinical motor evaluations at 6 months after treatment17. A second clinical trial used AAV2 to deliver neurturin to the putamen bilaterally. An initial group of 12 patients at one- year follow-up demonstrated safety and some motor improvement18. A second group of patients showed no significant improvement at 12 months post-treatment, but mild improvement at 18 months19. When the brains of four patients were analyzed post-mortem, very little neurturin expression was detected in SN, thus highlighting a challenge of delivering gene therapy to a structure that may be severely degenerated by the time treatment is initiated20. In an attempt to improve expression in SN, additional patients were injected in both putamen and SN, but failed to demonstrate motor improvement after 2 years19. A third study is currently recruiting patients with advanced PD. It will aim to deliver GDNF bilaterally to the putamen using AAV2 and follow up the patients for 5 years21.

0.3 Medial forebrain bundle axotomy

The medial forebrain bundle (MFB) is an anatomical structure containing several different axon tracts, and in rat includes dopaminergic axons projecting from the SNc to the striatum22. Medial forebrain bundle axotomy (MFBx) is a surgical procedure in which the axons

of the medial forebrain bundle are mechanically severed, resulting in retrograde degeneration of dopaminergic neurons in SNc23. Weiser et al reported that, at seven days following unilateral

MFBx, the ipsilateral (transected side) SNc displayed 68% tyrosine hydroxylase (TH) immunoreactivity compared to the contralateral control24. A similar study by Brecknell et al reported a similar figure of 72%, again at 7 days post-MFBx25.

In the past, MFBx has been used to model PD in rats because it results in cell death of the same cell population affected in PD26-28. Recently however, αSyn-overexpression models have gained popularity, as they are believed to better represent some of the molecular features of PD

(i.e. presence of human αSyn)29-33. Nonetheless, MFBx remains a useful model of dopaminergic cell death.

0.4 BAG proteins

The Bcl2-associated athanogene (BAG) family is a group of co-chaperone proteins that function in cell growth and programmed cell death pathways34,35. The name comes from the discovery of the first BAG protein, BAG1, which binds to B-cell lymphoma/leukemia 2 gene

(Bcl2). Bcl2 itself promotes cell survival and has been found to increase neuronal survival, though its over-activation due to chromosomal translocations gives rise to cancer.

Six BAG proteins exist in humans: BAG1 through BAG634,35. Each BAG protein contains one evolutionarily conserved 78-amino acid BAG domain near the C-terminus36, except

BAG5, which has four34,35,36 or possibly five37-39 BAG domains spread along its length34,35. BAG proteins are nucleotide exchange factors for the Hsp70/Hsc70 family of molecular chaperones, and BAG domains mediate direct interaction with the ATPase domain of Hsp70/Hsc7034,36.

Several other domains are also found within the different BAG proteins, allowing them to interact with specific proteins or to be targeted to specific intracellular locales34,35.

There are four isoforms of BAG1 in humans: BAG1L, BAG1M, BAG1, and BAG1S, in order from longest to shortest36. Among the isoforms, BAG1L is the only one to have a nuclear localization signal at the N-terminus, so it is most likely found in the nucleus. BAG1 binding to

Hsc/Hsp70 stimulates nucleotide exchange, thus promoting chaperone activity. Each isoform contains one ubiquitin-like domain that allows it to bring Hsc/Hsp70 chaperones to the proteasome, suggesting a role for BAG1 in protein degradation36. BAG1 double knockout mice were found to have dramatic apoptosis of the fetal liver and nervous system40. Recently, BAG1 overexpression in the red nucleus was found to protect neurons from dorsal hemisection injury41.

Thus, there is strong evidence for an anti-apoptotic role of BAG1.

BAG2 is overexpressed in some cancers and may promote tumour formation by promoting the accumulation of other oncogenic mutations42. However, BAG2 may play a different role in other contexts such as PD. In HEK293 cells, BAG2 inhibits C-terminus of

Hsp70 interacting protein (CHIP), which normally ubiquitylates αSyn, PINK1, and other proteins for degradation43,44. In this fashion, BAG2 stabilizes PINK1 and promotes mitophagy43.

Although it inhibits CHIP-mediated ubiquitylation, BAG2 was found to promote ubiquitin- independent Tau degradation45.

BAG3 expression is induced by oxidative and proteasome stress. BAG3 knockout mice exhibit post-natal development issues, especially in muscle tissue, and die prematurely46. In various cancer cell lines, BAG3 expression has been found to promote proliferation and enhance resistance to anti-cancer drugs47,48. Recently, BAG3 was found to be important in autophagy pathways. As human fibroblast cells aged, they switched from predominantly BAG1 to BAG3 expression, corresponding to a switch from proteasome to selective macroautophagy based protein degradation49.

BAG4 is also called “silencer of death domain” and is well studied in cancer. It binds and inhibits “death domains” found in some tumour necrosis factor receptors, thus promoting cancer cell survival against radiation and anti-cancer drugs50,51.

BAG6 may be unique amongst the BAG proteins in that its BAG domain may lack nucleotide exchange factor activity for Hsc70. Instead its BAG domain may help BAG6 form a complex that regulates membrane protein degradation52. BAG6 inhibits Hsp70 protein refolding activity and appears to regulate Hsp70 induction after heat shock53. It has been implicated in lung cancer and multiple sclerosis54,55.

Unlike most other BAG proteins, several studies have suggested a pro-apoptotic function for BAG5. It has been shown to directly interact with several molecules associated with PD pathogenesis and it promotes dopaminergic cell death in two animal models of PD/dopaminergic injury56. However, one recent study showed that BAG5 stabilized PINK1 by inhibiting its ubiquitylation in HEK293 cells, and also protected mitochondria from oxidative stress by upregulating PINK157. Nonetheless, another study also showed that BAG5 and BAG2 stabilized pathogenic ataxin proteins by inhibiting their ubiquitylation, providing further evidence of the role of BAG proteins in neurodegeneration42. BAG5 may play a different role in cancer, where it was identified as a risk factor for non-Hodgkin’s lymphoma58. In prostate cancer patients, BAG5 was found to be overexpressed and inhibit ER-stress induced apoptosis38.

0.5 Rationale for targeting BAG5

BAG5 was originally discovered as a protein whose expression was upregulated in SNc at 24 hours following dopaminergic neuron injury, via medial forebrain bundle axotomy

(MFBx), in rat56. Adenovirus-mediated overexpression of BAG5 in rat SN significantly enhanced loss of TH immunoreactivity at 7 days post-MFBx56, suggesting a positive feedback

mechanism. Similarly, BAG5 overexpression in the MPTP mouse model of PD resulted in greater amounts of TH loss at 2 weeks post-MPTP treatment, compared to controls56. These animal experiments suggested that BAG5 could be a mediator of dopaminergic cell death.

BAG5 associates with several proteins that are mutated in familial forms of PD, including parkin56,59, PINK157, and LRRK260. BAG5 directly interacts with parkin, sequesters it into protein aggregates, and inhibits its E3 ligase activity, which is necessary for the controlled removal of damaged mitochondria (mitophagy)56,59. In addition, BAG5 forms a complex with the co-chaperone CHIP and αSyn, resulting in the inhibition of CHIP-mediated ubiquitylation of

αSyn, thus leading to increased αSyn oligomerization61. Furthermore, BAG5 has been identified as an interactor of LRRK2, though the functional consequences are unclear60,62 (also see Fig. 1).

By inhibiting Hsp70 chaperone function56 and inhibiting the E3 ligase activity of parkin56,59 and CHIP61, BAG5 negatively affects the molecular chaperone and ubiquitin- proteasome systems, as well as the control of mitophagy, which are considered to be protective in the contexts of PD and neurodegeneration63-66. Taken together with its role in mediating dopaminergic cell death in vivo, BAG5 seems to be a common element in several pathways relevant to PD, and thus represents a promising therapeutic target.

FIGURE 1.

Figure 1. BAG5 interactors and their involvement in pathways relevant to the pathogenesis of Parkinson’s disease. BAG5 directly interacts with Hsp70, parkin, PINK1, and LRRK2. It inhibits the chaperone activity of Hsp70 and also inhibits the E3 ligase activity of parkin56.

Through its interaction with Hsp70, BAG5 inhibits the E3 ligase activity of CHIP61. BAG5 also stabilizes PINK1 by inhibiting its ubiquitylation57. Finally, BAG5 is a director interactor of

LRRK2, though the functional consequences of this interaction are unclear60. Also depicted:

PINK1 interacts with parkin to control mitophagy. CHIP ubiquitylates αSyn and reduces its oligomerization (though this effect is inhibited by BAG5, through Hsp70)61. Molecular chaperones (such as Hsp70) refold misfolded proteins (such as αSyn) and prevent their aggregation.

0.6 RNA-targeting strategies for gene inhibition therapy

One approach to inhibiting gene expression in a biological system is to target mRNA transcripts before they can be translated into protein. With this approach, since the gene itself is not targeted, the risk of chromosomal mutation/damage is avoided, and inhibition is reversible.

The main technologies for RNA-targeting gene inhibition are short hairpin and small interfering

RNAs (via the endogenous RNA interference pathway) and antisense oligonucleotides.

0.6.1 RNA interference

RNA interference (RNAi) is a naturally occurring pathway that modulates gene expression through small RNA molecules, including the endogenous micro RNAs (miRNA)67.

The RNAi machinery can be exploited to reduce expression of a specific target gene by adding exogenous small RNAs, either in the form of short hairpin RNA (shRNA) or small interfering

RNA (siRNA).

Short hairpin RNAs consist of a 19-29 base pair stem region that is linked by a single- stranded RNA loop of 4 or more nucleotides, plus a 2-nucleotide 3’ overhang68. In the RNAi pathway, shRNAs are processed by two ribonucleases, Drosha and Dicer. Then, from the remaining 19-29 bp double-stranded RNA molecule (dsRNA) the sense (or passenger) strand is degraded and the antisense (or guide) strand is loaded into the RNA-induced silencing complex

(RISC)68. The guide strand directs RISC to bind and cleave mRNA transcripts that are fully complementary to the loaded guide strand. The cleaving is performed by the endonuclease

Argonaute, a component of RISC67,68. The similar siRNAs are artificial short dsRNA molecules designed to enter the RNAi pathway at the post-Dicer processing point and subsequently reduce expression of a target protein.

siRNAs are well suited for transient knockdown conditions in vitro because it is relatively cheap to synthesize and transfect siRNAs into cultured cells. For in vivo applications, siRNAs can struggle to avoid degradation and enter cells, and they must be constantly resupplied as they are (inefficiently) consumed69. Viral vector-delivered shRNA can be a more sustainable strategy in the long term, as it provides long-lasting silencing of a target gene, and the viral vector allows for more efficient and controlled delivery.

0.6.2 Antisense oligonucleotides

Antisense oligonucleotides (ASO) are synthetic nucleic acids, 8-50 nucleotides in length, which are designed to bind to complementary RNA in order to alter gene expression70. ASOs may operate through a variety of mechanisms, which may or may not lead to degradation of the target RNA. Unlike RNAi, ASOs can also target precursor mRNA (pre-mRNA) in the nucleus, before it has been exported and matured into mRNA. For instance, ASOs can bind near a splice site in pre-mRNA, causing exons to be skipped or included. They may also bind close to the 3’ end of the pre-mRNA and modulate polyadenylation70,71. ASO binding on mature mRNA can lead to translation inhibition, and binding on either pre-mRNA or mature mRNA can recruit various nucleases that lead to transcript degradation70,71.

One of the challenges facing in vivo use of ASOs is the natural instability of unmodified nucleic acids, due to the ubiquitous expression of nucleases. The chemical structure of ASOs may be modified in several different ways in order to improve nuclease stability71,72. Some modifications target the phosphate group in the backbone (swapping it for a sulfur-containing group) or the ribose/deoxyribose sugar (tagging it with additional functional groups or fusing it to a second ring that restricts conformation)71,72. A more extreme possibility is to produce nucleic acid mimics. These mimics can be either morpholinos (in which pentose sugars are replaced by

morpholine rings) or peptide nucleic acids (in which the sugar-phosphate backbone is replaced by a polypeptide backbone)70-72. Another challenge is to deliver ASOs to the target cells in the body. This is usually attempted by attaching the ASOs to a ligand of, or antibody against, a target cell-surface receptor to allow the ASO to enter via receptor-mediated endocytosis70.

0.7 DNA-targeting strategies for gene inhibition therapy

Rather than targeting RNA transcripts, genes themselves can be targeted in order to delete or silence the target gene, or to replace the target gene with an exogenously supplied DNA sequence. These alterations to the genome are mostly irreversible. The main technologies for gene editing are either recombinases (Cre/loxP system), or customizable nucleases (TALENs, zinc finger nucleases, CRISPR/Cas system) followed by endogenous DNA repair mechanisms.

0.7.1 Cre/loxP recombination

The Cre/loxP system, derived from machinery in P1 bacteriophage, uses Cre recombinase to create a targeted recombination event between two 34-bp loxP sites in the genome73.

Depending on the positions and directionality of the loxP sites, the recombination event may result in deletion or inversion of the interspacing DNA, or translocation (if the loxP sites are on different DNA molecules)74,75. Several variations on the Cre/loxP setup make it quite flexible.

Target genes can be silenced or overexpressed, either in the presence or absence of Cre expression. Genes can also be knocked in when Cre is present, via translocation or inversions.

Additionally, Cre expression can be made inducible or tissue-specific by placing it under the control of various promoters, or it can be supplied via viral vectors74. The major disadvantage of the Cre/loxP system is that it requires loxP sites to be inserted at the desired locations, resulting in relatively long turnaround times as cell lines or transgenic organisms (usually mice) are prepared74. In addition, it is difficult to accurately achieve multiple Cre/loxP recombination

events in the same cell since there would be many possible combinations of pairings between multiple loxP sites.

The FLP-FRT recombination system, derived from machinery in yeast, can also produce targeted recombination events. It is conceptually similar to Cre/loxP76.

0.7.2 TALEN

Transcription Activator-Like Effector (TALE) proteins are DNA-binding proteins found in several bacterial species. TALEs can be reprogrammed to bind any DNA sequence, as their binding specificity is conferred by just two amino acids in each DNA binding domain (DBD)67.

Customized TALEs can be fused to effector domains such as transcriptional repressors or endonucleases (creating TALE nucleases, or TALENs) to precisely direct the activity of the fused proteins. A particularly useful TALEN is the fusion of TALEs to FokI endonuclease, which functions only as a dimer and has separate binding and cleavage sites67,77. The combination of two different TALEs that target adjacent sites on DNA can bring together two

FokI to create a double-stranded break (DSB) at a highly specific target sequence67,77.

DSBs are repaired with one of two endogenous and competing repair mechanisms: non- homologous end joining (NHEJ) or homology directed repair (HDR). NHEJ is an error-prone repair mechanism, resulting in frameshift (functional deletion) 2/3 of the time, and potential insertions/deletions the other 1/3 of the time. Thus it can result in genetic and phenotypic heterogeneity, and the outcome may be difficult to predict67. On the other hand, HDR is a high fidelity repair system, but requires a template that shares homology with the area around the

DSB. This template may be the other allele or an exogenously supplied homologous donor DNA sequence. In the latter case, a precise sequence can be inserted to bridge the DSB. While

TALENs are very specific with low off-target effects, it is expensive and time-consuming to design TALEs for each custom target78.

0.7.3 Zinc finger nucleases

Zinc finger nucleases (ZFNs) are artificial restriction enzymes, formed by fusing eukaryotic transcription factors with a Cys2-His2 zinc-finger DNA binding domain and FokI67.

These DBDs bind DNA three nucleotides apart, and can be placed in tandem to confer better specificity for a target DNA sequence79. Otherwise, ZFNs are conceptually similar to TALENs, though generally more difficult to program and less specific78.

With either ZFNs or TALENs, one of the FokI domains can be mutated to inactivate its endonuclease activity. This creates a TALEN or ZFN “nickase” that generates a single-stranded break (SSB) at the target location67,77. SSBs are repaired by HDR and not NHEJ, thus avoiding some of the challenges with NHEJ such as unintentional mutations and increased risk of apoptosis77.

0.7.4 CRISPR/Cas

CRISPR (clustered regularly interspaced short palindromic repeats)/Cas (CRISPR associated protein) is a system found in the genome of bacteria and archaea that provides adaptive immunity against foreign nucleic acids80. Unlike the protein domain-based binding of

TALENs and ZFNs, the CRISPR/Cas system uses a guide RNA to target the nuclease to the target DNA site. It therefore has the major advantage of being easier to program to target custom

DNA sequences78,81.

The Cas9 endonuclease is most commonly used for CRISPR gene editing81. It contains two nuclease domains and functions as a monomer, causing a DSB at the target location. Though

ZFNs and TALENs are more specific than normal Cas967, Cas9 specificity can be improved in a few ways. For instance, one of its nuclease domains can be inactivated, producing a Cas9 nickase that may be paired with a second Cas9 nickase. Alternatively, both of Cas9’s nuclease domains can be inactivated to produce dCas9, which can then be fused to FokI domains. dCas9 can also be fused to transcriptional activators or repressors for additional control of gene expression81,82.

0.8 Viral vectors for gene delivery

In recent years, many viruses have been modified for in vivo gene therapy applications, including lentivirus and adeno-associated virus (AAV)83. Compared to physical or chemical delivery methods, viral vectors provide superior transduction efficiency, allow for better targeting, and naturally protect the gene product from nuclease degradation83. However, viral vectors have a genomic size limit, may induce an immune response, and can be costly.

0.8.1 Lentivirus

Lentivirus (LV) is an enveloped ssRNA virus that has the unique ability among retroviruses to integrate into non-dividing cells such as neurons. It has strong tropism (i.e. tendency/ability to transduce) toward neural stem cells83. Lentivirus tropism can be modified by pseudotyping, a process in which lentivirus particles are generated with glycoproteins from other enveloped viruses. One of the most popular lentivirus pseudotypes uses vesicular stomatitis virus glycoprotein (VSV-G), resulting in very broad tropism, including neurons84. Lentivirus has the advantage of being able to deliver a relatively larger genomic payload of 8kb of RNA to infected cells83. These features of lentivirus, combined with its low immunogenicity and lack of side effects83, have made it a popular and effective gene therapy vector in CNS applications in animal experiments85.

0.8.2 Adeno-associated virus

AAV vectors have been another popular choice for gene therapy. They are non- enveloped ssDNA viruses that can transduce a wide range of dividing and non-dividing cells and produce long-lasting transgene expression16. AAVs can deliver up to 4.8kb of DNA86. Upon infection, AAV can either persist as a double-stranded episome or integrate into the host genome, most often at the AAVS1 site on chromosome 1916,83,86. AAVs are not associated with any disease, exhibit very low immunogenicity, and cannot replicate except in the presence of a helper virus (usually adenovirus), making them safe for human therapeutic applications16,83.

AAV tropism is determined by its serotype, or type(s) of capsid protein expressed.

Different capsid proteins bind to specific mammalian cell surface receptors87. There are 11 natural AAV serotypes, but many hybrid serotypes (expressing capsid proteins from two different serotypes) and other variants have been generated, allowing for custom tropism16,87.

Most AAV serotypes, but especially AAV1, 2, and 5, show strong neuronal tropism when injected into the brain16. AAV4 preferentially transduces ependymal cells, while AAV9 exhibits both neuronal and astrocytic tropism and has the unique ability to cross the blood-brain barrier16.

To date, AAV2 has been the vector of choice in both human and animal studies of gene therapy in PD, mainly due to its historical safety record. For gene therapy applications, improved control of spatiotemporal expression can be achieved through the appropriate selection of AAV serotype, promoter, and injection strategy (i.e. systemic vs. targeted).

0.9 Research Objectives

The principal aim of this study is to develop and characterize an shRNA-expressing viral vector that can achieve efficient knockdown of BAG5 expression in the rat substantia nigra.

Chapter 1: Generation of viral vectors expressing BAG5-targeting shRNA.

The first chapter of this thesis describes the identification of several shRNAs that efficiently knock down rat BAG5 protein expression in two cell lines. Following this identification AAV and lentivirus vectors that express these shRNAs were generated for in vivo applications, as discussed in Chapter 2.

Chapter 2: Characterization of viral expression, target engagement, and biological consequences of BAG5 knockdown in vivo.

The second chapter of this thesis focuses on interrogating the effects of the viral vectors generated in Chapter 1 (especially the AAV) when injected into the rat substantia nigra. The effects of AAV-shRNA driven BAG5 knockdown in the context of SN dopaminergic neuron injury are also investigated.

16 1 Chapter 1: Generation of viral vectors expressing BAG5- targeting shRNA 1.1 Introduction

This chapter describes the process of designing and generating two viral vectors, adeno-

associated virus and lentivirus, which express BAG5-targeting shRNA and GFP as an expression

marker. It begins by detailing the initial screen of several candidate BAG5 shRNA target

sequences in two cell lines and concludes with the generation and confirmation of AAV

infectivity in primary neurons. Subsequent in vivo experiments are covered in Chapter 2.

1.2 Materials and Methods

1.2.1 BAG5 shRNA plasmid construction

All plasmids from Origene were constructed to express shRNA under control of the U6

promoter and GFP under the CMV (cytomegalovirus) promoter. The plasmids also contained the

puromycin resistance gene under the SV40 promoter.

All Creative Biogene plasmids were constructed to express shRNA under control of the

U6 promoter and GFP under the CMV promoter. The plasmids also contained the puromycin

resistance gene under the PGK (phosphoglycerate kinase) promoter.

All Genecopoeia plasmids were constructed to express shRNA under control of the U6

promoter and both GFP and puromycin resistance under the SV40 promoter, using an

intervening IRES (internal ribosomal entry site) sequence.

Targeting sequences for each of the shRNAs are found in Table 1.

1.2.2 Cell culture and transfection

Rat pheochromocytoma PC12 cells were maintained in RPMI (Roswell Park Memorial

Institute) 1640 medium supplemented with 10% FCS (fetal calf serum), penicillin (100 units/mL), streptomycin (100 µg/mL), and amphotericin B (250 ng/mL). Human embryonic kidney (HEK) 293T cells were maintained in DMEM (Dulbecco’s Modified Eagle Medium) supplemented with 10% FCS (fetal calf serum), penicillin (100 units/mL), streptomycin (100

µg/mL), and amphotericin B (250 ng/mL). All cells were incubated at 37°C in a humidified air atmosphere containing 5% carbon dioxide.

HEK and PC12 cells were plated on 6-well plates (1,000,000 cells/well for HEK;

2,000,000 cells/well for PC12) in antibiotic/antimycotic-free media. Cells were transfected with

1.5µg of plasmid DNA at 24 hours after plating (1.5µg each of shRNA and rat BAG5 plasmid for

HEK cells), using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol.

Media was replaced with fresh media at 24 hours post-transfection. Cells were harvested 48 hours post-transfection in RIPA (radioimmunoprecipitation assay) buffer containing protease inhibitor cocktail (Roche).

1.2.3 Immunoblotting and analysis

Cells were (re-) suspended and lysed in RIPA buffer containing protease inhibitor cocktail (Roche). Protein concentrations were determined using the DC Protein Assay (Bio-Rad).

Equal amounts of protein were loaded and run through sodium dodecyl sulfate (SDS) polyacrylamide gels by electrophoresis, followed by wet transfer to polyvinylidene fluoride

(PVDF) membranes. Membranes were probed with anti-BAG5 antibody from mouse (Santa

Cruz, cat. #ab56738) diluted at 1:200 (for endogenous BAG5 in PC12 cells) or 1:1000 (for exogenous rat FLAG-BAG5 in HEK cells), and with anti-GAPDH antibody from rabbit (Cell

Signaling, cat. #2118) diluted at 1:1000. Membranes were then probed with secondary anti- mouse IgG or anti-rabbit IgG linked to horseradish peroxidase (Cell Signaling) diluted at 1:5000.

Detection was performed with the enhanced chemiluminescence (ECL) method (Pierce) for

GAPDH and with ECL+ (Pierce) for BAG5. Relative protein expression was determined by western blot densitometry (with background subtraction), with GAPDH serving as the loading control. Transfection efficiencies were assumed to be the same across all conditions, thus the bands were not normalized for transfection efficiency.

1.2.4 Adeno-associated virus generation

Adeno-associated virus 1/2 vectors were custom ordered from GeneDetect Ltd.

(Auckland, New Zealand). Each vector expresses GFP under the control of the chicken beta actin

(CBA) promoter hybridized with the CMV immediate early enhancer sequence. These vectors also include a woodchuck post-transcriptional regulatory element (WPRE) and a bovine growth hormone (BGH) polyadenylation sequence to improve protein expression. Each vector also expresses either Origene non-targeting Control, Origene C, or Genecopoeia G1 shRNA against

BAG5, under the U6 promoter. Stock viral titers were determined by quantitative PCR to be

1.1×1012 genomic particles/mL for all AAVs.

1.2.5 Lentivirus generation

Lentivirus vectors were custom ordered from the UHN Vector Core (Toronto, Canada).

Each vector expresses either Origene non-targeting Control or Origene B shRNA under the U6 promoter, and also GFP under the CMV promoter. Stock viral titers were determined to be

9.0×107 IU/mL.

1.2.6 Rat primary hippocampal neuron culture and infection

Dissociated E18.5 rat hippocampal neurons were obtained and cultured as previously

88 described by Nie & Sahin . In brief, rats were euthanized with CO2 and embryos were recovered. Hippocampi and cortices were dissected and placed in chilled dissociation medium

(Ca2+-free HBSS with 100mM MgCl2, 10mM kynurenic acid (Sigma-Aldrich), and 100mM

HEPES). Following enzymatic dissociation with papain (Worthington Biochemical Corporation) and L-cysteine (Sigma-Aldrich), trypsin inhibitor (Sigma-Aldrich) was added, and cell clumps were dissolved by trituration. Neurons were suspended in Neurobasal medium supplemented with B-27, L-glutamine, and penicillin/streptomycin/primocin, and plated at a density of 1x106 per well in 6-well plates coated overnight with 20µg/ml poly-D-lysine (Sigma-Aldrich) and

3.5µg/mL laminin. At 2 days in vitro (DIV), neurons were infected with AAV vectors at an MOI of 2×104 in a volume of 1 mL of conditioned medium. Half the medium was replaced with fresh medium on DIV5. On DIV8, cells were washed with cold PBS, collected in 500uL PBS and pelleted by centrifugation at 500g for 5 minutes at 4°C. Cell lysates were prepared for western blot as described previously (Section 1.2.3). Unless stated otherwise, all reagents were purchased from Thermo Fisher Scientific. This experiment (except western blot) was performed by Dr.

Darius Ebrahimi-Fakhari in Dr. Mustafa Sahin’s lab at Boston Children’s Hospital, Harvard

Medical School (Boston, MA, USA).

1.3 Results

1.3.1 Screen of candidate BAG5 shRNA plasmids

Thirteen potential rat BAG5-targeting shRNA sequences and plasmids were generated from three vendors in order to increase our chances of identifying efficient shRNA clones. Based on the vendors, these shRNAs were termed Origene A through D, Creative Biogene 1 through 4,

and Genecopoeia G1, G4, G12, G13, and G14. Each candidate shRNA targeted a different site along the rat BAG5 mRNA transcript, within the open reading frame (ORF) (Fig. 2). None of the candidate shRNAs were located near a splice site; rat BAG5 has a single splice site located in the

5’ UTR (Ensembl genome browser89).

Candidate shRNAs were screened for ability to knock down BAG5 in two cell lines, and several promising shRNA sequences were identified. In rat pheochromocytoma PC12 cells

(adrenal gland tumour; expresses endogenous rat BAG5), all four shRNA plasmids from Origene

(clones A through D) generated substantial BAG5 knockdown 48h after transfection (Fig. 3A).

Compared to control shRNAs, transfection with these targeting clones resulted in a proportion of

BAG5 protein levels as follows (reported as mean ± standard deviation): Origene A (0.65 ±

0.18); Origene B (0.60 ± 0.23); Origene C (0.64 ± 0.27); Origene D (0.42 ± 0.07). All data are presented after removal of outliers using Grubbs’ test. Transfection efficiency for PC12 cells was observed to be roughly 30% (estimated proportion of GFP+ cells 48h post-transfection, based on visual inspection).

In human embryonic kidney (HEK) cells co-transfected with rat BAG5 and the candidate shRNA plasmids, substantial BAG5 knockdown was achieved 48 hours post-transfection by all four Origene clones, as well as Creative Biogene clone 2 and Genecopoeia clone G1 (Fig. 3B).

Compared to controls, transfection with these targeting clones resulted in the following proportions of BAG5 protein levels: Origene A (0.69 ± 0.23); Origene B (0.71 ± 0.28); Origene

C (0.51 ± 0.13); Origene D (0.35 ± 0.12); Creative Biogene 2 (0.59 ± 0.15); Genecopoeia G1

(0.46 ± 0.28). All data are presented after removal of outliers using Grubbs’ test. Transfection efficiency for HEK cells was observed to be roughly 70% (estimated proportion of GFP+ cells

48h post-transfection, based on visual inspection).

When the results from PC12 and HEK transfections are combined (Fig. 3C), again BAG5 knockdown was observed by all four Origene clones, Creative Biogene clone 2, and

Genecopoeia clone G1. Compared to controls, transfection with these targeting clones resulted in the following proportions of BAG5 protein levels: Origene A (0.67 ± 0.19); Origene B (0.66 ±

0.25); Origene C (0.57 ± 0.20); Origene D (0.37 ± 0.10); Creative Biogene 2 (0.79 ± 0.29);

Genecopoeia G1 (0.71 ± 0.48).

Origene D produces the most efficient knockdown in both PC12 and HEK cells. The other three Origene clones were also efficient at knocking down BAG5 in either cell line. Clone

C in particular was quite effective at knocking down exogenous BAG5 in HEK cells, though not quite as well as clone D. None of the four targeting shRNAs from Creative Biogene were able to produce reliable endogenous BAG5 knockdown in PC12 cells. In HEK cells, Creative Biogene clone 2 showed some knockdown of exogenous BAG5, though the magnitude of knockdown was less than that produced by Origene C, Origene D, and Genecopoeia G1. The other Creative

Biogene clones displayed either lack of efficiency or lack of reliability in HEK cells. None of the five Genecopoeia shRNAs were able to reliably knock down BAG5 in either PC12 or HEK cells, with the exception of G1 in the HEK condition.

Origene C, Origene D, and Genecopoeia G1 were selected as the three most promising candidate shRNAs. Although G1 showed similar overall efficiency (combined PC12 and HEK results) as Origene A and B, we chose to not select three shRNAs from a single vendor in order to avoid potential vendor-wide problems. Furthermore, despite the poor performance of G1 in

PC12 cells, it displayed the second greatest average knockdown efficiency in HEK cells, just behind Origene D. This suggested that it may be quite effective at knocking down BAG5 under some conditions, and may have potential for our subsequent in vivo applications.

TABLE 1.

BAG5 shRNA targeting sequences and positions

shRNA Targeting sequence (5’ to 3’) Target position on rat BAG5 mRNA 5’ 3’ Control GCACTACCAGAGCTAACTCAGATAGTACT No match

A ACTGATAAGAACTACATTCGTCTGGAGGA 1308 1336 B ACACAGAACGCCTTCTCAAAGAGTTGGAG 358 386

Origene C AGACTGGAGAGGATTCTGACGAAACAACT 258 286 D GTGCATCCTTCTGTCGCCAAGATCAATTC 672 700

Control GCTTCGCGCCGTAGTCTTA No match 1 GCAAGGTACCACACACTAACC 579 599 2 GCGTACAGAGATCAGAAATTA 839 859 3 GGAAGCTGACAATACACATGC 917 937

Creative Biogene Creative 4 GGGACAGAACCATTCCATTAT 947 967 Control GCTTCGCGCCGTAGTCTTA No match

G1 GTCGGATGACAAGAATTAC 236 254 G4 CACCGGATTGAAATCAAGA 405 423 G12 CGAAGTGAGTCTGGAGAAA 1091 1109

Genecopoeia G13 GAACAAGGCCAGAGGTACT 716 734 G14 GGAGAAGAGCAGTGATTGA 1132 1150

FIGURE 2.

Figure also contributed by H. C.

Figure 2. Schematic showing target location of each of the 13 candidate BAG5-targeting shRNAs along the rat BAG5 mRNA transcript.

FIGURE 3.

FIGURE 3 (CONT’D)

Figure 3. Initial in vitro screen of knockdown efficiency of candidate shRNAs against rat

BAG5, via western blot densitometry. (A) Knockdown of endogenous rat BAG5 in PC12 cells transfected with candidate shRNAs, 48h post-transfection. (B) Knockdown of exogenous rat

BAG5 (co-transfected with the shRNAs) in HEK cells transfected with candidate shRNAs, 48h post-transfection. (C) Combined data from (A) and (B). All data are reported as mean ± standard deviation.

1.3.2 Confirmation of top three candidate shRNA plasmids

After identifying BAG5 shRNA clones C, D, and G1 as the three most promising candidate shRNAs, the knockdown experiment in PC12 and HEK (Fig. 3) was repeated, focusing on these top three shRNAs, in order to confirm their ability to knock down BAG5 in vitro.

In PC12 cells, densitometry analysis showed that, compared to the respective control shRNA, transfection with targeting clones resulted in the following proportions of BAG5 protein levels: Origene C (0.68); Origene D (0.64); Genecopoeia G1 (0.84) (Fig. 4A).

In HEK cells, transfection with targeting shRNA resulted in the following proportions of

BAG5 protein levels, relative to the respective control shRNA: Origene C (0.73); Origene D

(0.74); Genecopoeia G1 (0.22) (Fig. 4B).

The results here are similar to those observed in the initial shRNA screen, with clones C and D being able to knock down BAG5 in both PC12 and HEK cells, while G1 resulted in strong

BAG5 knockdown in HEK cells, but not in PC12.

FIGURE 4.

Figure 4. Confirmation of efficient BAG5 knockdown by top three shRNAs identified in the initial screen. (A) Knockdown of endogenous rat BAG5 in PC12 cells transfected with candidate shRNAs at 48h post-transfection. Left: western blot. Right: western blot densitometry (n=1). (B) Knockdown of exogenous rat BAG5 (co-transfected with the shRNAs) in HEK cells transfected with candidate shRNAs at 48h post-transfection. Left: western blot. Right: Western blot densitometry (n=1).

1.3.3 Generation of AAV1/2 and lentivirus vectors expressing BAG5 shRNA

Having confirmed the knockdown activity of clones C, D, and G1, these clones were packaged into adeno-associated virus (AAV) 1/2 vectors co-expressing GFP. A non-targeting control virus was also generated, using the shRNA sequence from Origene non-targeting

Control. From here on, these vectors will be referred to as AAV-shBAG5 C, D, G1, and Control.

We also had access to an AAV1/2 expressing only GFP and no shRNA, but otherwise identical to our AAV-shBAG5 vectors, and termed this AAV-GFP.

Preliminary in vivo testing of the different AAV-shBAG5s showed that C and G1 produced more reliable knockdown of BAG5 in injected rat substantia nigra compared to D (data not shown). Thus, all subsequent experiments focused on investigating only AAV-shBAG5 C and G1 compared to the non-targeting Control AAV.

Lentivirus (LV) vectors co-expressing shRNA and GFP were also generated, using shRNA sequences from Origene B and Origene non-targeting Control. From here on, the LV vectors will be called LV-shBAG5 and LV-Control. The LV vectors were generated well before the AAVs. At the time of LV generation, fewer candidate shRNAs were available for testing. In addition, fewer replicates of the in vitro knockdown experiments (as in Fig. 3) had been performed. Based on the data available at the time, we chose to package Origene clone B shRNA into the LV vector, rather than one of the shRNA sequences used in the AAVs.

1.3.4 AAV-shBAG5 knocks down BAG5 in rat primary hippocampal neurons

AAV-shBAG5 vectors were used to infect rat primary hippocampal neurons. GFP expression and BAG5 knockdown activity were assessed by western blot (Fig. 5A). All three

AAV-shBAG5 vectors were all able to infect these neurons and drive expression of GFP. By 8

days in vitro, infection efficiency was observed to be approximately 75%, by visual inspection of the proportion of GFP+ neurons. Infection with AAV-shBAG5 Control resulted in considerably greater GFP expression compared to C (44% of Control) or G1 (68% of Control) (Fig. 5B). Both

AAV-shBAG5 C and G1 were able to produce substantial knockdown of BAG5. Infection with

AAV-shBAG5 C resulted in 28% BAG5 expression compared to Control AAV-infected neurons, while AAV-shBAG5 G1 resulted in 50% BAG5 expression compared to Control (Fig. 5C).

FIGURE 5.

Figure also contributed by D. E-F, M. S., and H. C.

Figure 5. GFP expression and BAG5 knockdown by AAV-shBAG5 vectors in rat primary hippocampal neurons at 8 days post-infection. (A) Western blot for BAG5, GFP, and GAPDH from lysate of primary rat hippocampal neurons infected with AAV-shBAG5 C, AAV-shBAG5

G1, and AAV-shBAG5 Control at MOI of 2×104. (B) Densitometry analysis of GFP expression, relative to GAPDH. (C) Densitometry analysis of BAG5 expression, relative to GAPDH. (n=1).

1.4 Discussion

1.4.1 Choice of shRNA as gene inhibition strategy

There are several advantages to using shRNA as a gene inhibition strategy in vivo. Once it is delivered to cells via viral transduction, it results in sustained knockdown of its target mRNA. Thus, shRNA treatment requires only a single intervention, which saves on costs and reduces risk to the animal. Virus-delivered shRNA is also easier to target to a specific area of the brain. This accuracy can be further refined to specific cell types by carefully selecting viral serotype/tropism. By comparison, ASOs would require regular infusions and constant production of the therapeutic agent. Since these agents are usually delivered by intrathecal or intraventricular infusion, it is also more difficult to target them to specific locations, potentially resulting in unwanted effects in other regions of the CNS (not to mention the inefficiency). The similar siRNA can also be delivered in vivo, but has traditionally struggled to overcome issues of low stability and cellular uptake69. Nonetheless, siRNAs and ASOs have shown promising results in animal models, and currently ASOs are being investigated in clinical trials for

Huntington’s disease90.

Another strategy to inhibit BAG5 could be to target the protein rather than the mRNA. A

BAG5 mutant, BAG5-DARA, has already been characterized. The DARA mutant is unable to interact with Hsp70, and prevents normal BAG5 from interacting with Hsp7056. The BAG5-

Hsp70 interaction is critical for some of BAG5’s functions, including its interaction with CHIP, which prevents CHIP from ubiquitylating αSyn61. However, the DARA mutant does not affect

BAG5’s interaction with parkin or its inhibition of parkin’s E3 ligase activity56. In addition,

BAG5 may have other, potentially deleterious functions that are unaffected by DARA. Thus, inhibiting the expression of BAG5 may result in a different phenotype that may have more comprehensive effects than DARA-based inhibition.

At the time of this project’s initiation, reports of in vivo CRISPR strategies were just beginning to emerge, so we chose a more traditional technology in shRNA. Going forward,

CRISPR will be an interesting alternative strategy to pursue, though gene knockdown and knockout may be differ in phenotype, and CRISPR has its own off-target effects and disadvantages to consider. For instance, it may be less efficient than RNAi given that it must successfully cleave and functionally knock outboth alleles of the target gene. Additionally, unrepaired double-stranded breaks can cause cell death, and off-target effects could result from cleaving at untargeted loci91.

1.4.11.4.2 BAG5 antibody validation

For application in western blot, our lab had previously tested several BAG5 antibodies from different vendors. Of these antibodies, the antibody used in Chapter 1 (Santa Cruz, cat.

#ab56738) provided the cleanest detection of endogenous BAG5 in PC12 cell lysate. It detected a single band at the predicted molecular weight of 51 kDa in HEK cell lysate expressing FLAG-

BAG5, and in the S2 (cytosolic) fraction of rat brain lysate (data not shown).

1.4.21.4.3 Transfection efficiency and interpretation of apparent knockdown In PC12 cells, we achieved approximately 30% transfection efficiency using 1.5µg of plasmid and Lipofectamine 2000, after trying several different cationic lipid-based transfection reagents that were no more efficient. Reports of PC12 transfection efficiency in the literature are quite variable. Lee et al reported 14% efficiency with 1 µg of DNA and Lipofectamine 2000, and

15% with polyethyleneimine (PEI)92. Cogli et al achieved a 30% transfection efficiency using a different cationic lipid product (Metafectene Pro)93. Covello et al achieved a transfection efficiency of 46% using 1 µg of plasmid Lipofectamine 200094. However, they were able to increase transfection efficiency to 90% when they used an electroporation method instead94.

In HEK cells, we were able to achieve a relatively high transfection efficiency of approximately 70% using 1.5µg of plasmid and Lipofectamine 2000. This value is similar to literature reports, which range from 60% efficiency with Lipofectamine 2000 and 1-2 µg of

DNA95, to 75% with PEI (with suspension cells)96.

Since we analyzed cell lysate from the entire population of cells and not just the transfected cells, the maximum expected knockdown in our population of cells should be equal to the transfection efficiency. The level of knockdown observed in our shRNA screen (Fig. 3) generally followed this rule, with the most efficient shRNA clones achieving roughly 30% knockdown in PC12 cells (Fig. 3A), and 70% knockdown in HEK cells (Fig. 3B). Thus, the actual knockdown produced by the most efficient shRNAs would be close to 100%. However,

Origene clone D produced a mean knockdown of 58% (i.e. 0.42 times the BAG5 band density compared to Control), which is substantially greater than the expected maximum knockdown of

30%. The reason for this is unclear. It is possible that transfection efficiency was underestimated, or that clone D produced an off-target effect increasing GAPDH expression, causing the amount of BAG5 in lysate to be underestimated. On the other extreme, we also observed that several less efficient shRNAs from Genecopoeia produced an apparent mean upregulation of BAG5 compared to non-targeting Control, in both HEK and PC12 cells (Fig. 3). This result may be attributable to the low sample size with high variance, poorly designed Control sequence, or off- target effects.

It is also important to consider protein half-life in knockdown experiments because shRNA only reduces mRNA levels without affecting existing protein. Previous studies of the proteome in various mammalian cell types have found that protein half-lives can be highly variable and depends on cell type, but are mostly on the order of hours to tens of hours97-99. The

half-life of BAG5 has not been studied, but if actual BAG5 knockdown were indeed close to

100% at 48h post-transfection, our results would suggest that BAG5 half-life would be on the order of several hours, such that the vast majority would be degraded within 48 hours. Protein half-life can be more accurately determined by pulse-chase assay or by cycloheximide blocking100.

It is unclear why some of the candidate shRNAs appeared to not produce appreciable knockdown, especially the clones from Creative Biogene (in PC12) and the non-G1 clones from

Genecopoeia (in PC12 and HEK). These shRNAs may simply not have been effective at binding the target sequence or associating with RISC. It is unlikely that there is an issue with transfection or with the plasmids’ stability or interaction with the lipofection reagent, since cells transfected with these plasmids demonstrated similar GFP expression levels and transfection efficiencies compared to shRNA clones that produced efficient knockdown. One possible explanation of poor knockdown efficiency is the dilution of plasmids as cells divide, since the plasmids are not designed to replicate in mammalian cells. The plasmids from Creative Biogene and Genecopoeia may produce efficient knockdown only at high plasmid copy numbers, perhaps because the shRNA targeting sequences bind less effectively to BAG5 mRNA. Different doubling times between PC12 and HEK cells (and thus plasmid dilution rates), and also different transfection efficiencies, may also explain the efficient knockdown produced by Creative Biogene plasmids in HEK cells, but not in PC12 cells. In addition, the Genecopoeia plasmids use a different GFP promoter than the plasmids from the other vendors. Therefore, although the GFP expression and transfection efficiency from Genecopoeia plasmids may seem comparable to that produced by other plasmids, the actual amount of plasmid delivered to each cell and the expression of shRNA may not be.

1.4.31.4.4 Limitations of western blot and densitometry analysis

Knockdown in vitro was quantified by western blot densitometry, which relies on the assumption that all bands of interest fall within the linear range (i.e. band intensity is directly proportional to amount of protein). We tried to adjust X-ray film exposure times to remain in the linear range, but did not use a protein standard to confirm. BAG5 bands were normalized to

GAPDH (glyceraldehyde 3-phosphate dehydrogenase) bands as a loading control. While this is standard practice, it assumes that all cells express equal amounts of GAPDH, and that this expression is not altered by the experimental treatment of shRNA transfection. While our shRNAs did not target GAPDH, there remains the risk of off-target effects. Recently, several groups have suggested that analyzing total protein (using a stain such as Ponceau S or Coomassie

Brilliant Blue) instead of a specific high-abundance protein such as GAPDH or actin would be a more accurate method of normalizing101-103.

1.4.41.4.5 Alternative techniques to quantify knockdown

There is some ambiguity in assessing knockdown in a heterogeneous population of transfected and non-transfected cells. New variations on transfection methods can achieve higher rates of transfection, but not 100% efficiency94,104. Instead, it is possible to isolate transfected cells (GFP+) from a population using fluorescence activated cell sorting (FACS), followed by analysis (or subculture) of this highly purified subpopulation105,106. Flow cytometry methods could also to used to obtain a more accurate measurement of transfection efficiency107. Another method of quantifying shRNA-mediated knockdown is RT-qPCR, though reported mRNA levels are not always reflective of the amount of protein available108. Unlike western blot, RT-qPCR is feasible even when good antibodies for the target protein are unavailable.

2 Chapter 2: Characterization of BAG5 shRNA viral vectors in vivo 2.1 Introduction

In the previous chapter, we identified shRNA sequences that can efficiently knock down

BAG5 in vitro, and subsequently generated AAV and LV vectors that co-express these shRNAs

along with a GFP expression marker. The next aim is to establish in vivo expression and BAG5

knockdown capabilities of these BAG5 shRNA viral vectors in rat SN. In this chapter, we aim to

identify a suitable viral titer for infection, characterize the spatiotemporal expression of these

vectors, and establish successful BAG5 knockdown in the rat SN. We also observe significant

differences in expression of AAV compared to LV. Finally we conduct a preliminary analysis of

some of the effects of BAG5 knockdown using the medial forebrain bundle axotomy (MFBx) rat

model of dopaminergic neuron injury.

Dopaminergic neurons, which express and positively stain for tyrosine hydroxylase (TH),

are present in the SNc and in the ventral tegmental area (VTA), a medial structure in the rat

ventral midbrain. It can be difficult to distinguish the boundary between the SNc and VTA in

certain coronal sections (“levels”) of the rat midbrain where these structures are continuous with

each other from left to right of the ventral region in coronal sections. However, another

anatomical structure called the medial terminal nucleus (MTN), which does not contain

dopaminergic neurons, exists between SNc and VTA at some levels of the midbrain. On TH-

stained coronal sections, the MTN appears as a small TH-negative area that extends

dorsolaterally from the ventral surface, separating SNc from VTA. Thus, at the level of the

MTN, the SNc is easily distinguishable from the VTA using TH staining, allowing subsequent

analysis to accurately examine SNc and not mistakenly include VTA in the analysis (or exclude

medial SNc). For this reason, all microscopy images and analysis included in this chapter have been taken/conducted at the level of the medial terminal nucleus (MTN).

Anatomically, SNc extends from -4.56mm to -6.60mm relative to bregma (Paxinos &

Watson, 1997). The VTA is particularly challenging to distinguish from SNc in more anterior sections, though VTA exists from -4.68mm to -6.24mm, nearly the entire anteroposterior extent of the SNc. MTN is a useful divider of SNc and VTA from -4.92mm to -5.28mm. Since the anteroposterior coordinate of injection is -5.2mm, the MTN-containing sections included in our analysis tend to be located very close but slightly anterior to the ideal injection site.

2.2 Materials and Methods

2.2.1 Animals

All animal experiments were conducted using female adult Sprague-Dawley rats (Charles

River), weighing 275-300g at the time of virus injection. Rats were co-housed in pairs on a regulated 12-hour light/dark cycle at standard temperature (21–22 °C) and allowed food and water ad libitum.

2.2.2 Stereotactic injection of AAV or LV into rat SN

All procedures were performed in accordance with the rules and regulations of the

Animal Resources Centre in the University Health Network. Animals were deeply anaesthetized with Anafen (Merial) at 5mg/kg delivered subcutaneously. Animals were mounted on a stereotactic frame and a craniotomy was made relative to bregma according to the brain atlas of

Paxinos and Watson (1997). Rats received a single 2µL, unilateral injection of AAV or lentivirus targeted to the SN by stereotactic injection. AAVs were injected at either 1/2 of stock titer

(resulting in a titer of 5.5×1011 gp/mL) or at 1/3 of stock (resulting in a titer of 3.7×1011 gp/mL).

AAV-GFP (used in Fig. 7) was injected at a titer of 5.1×1012 gp/mL. Lentivirus was injected at a

titer of 9.0×104 IU/µL (or 9.0×107 IU/mL). Viruses were delivered via microinjector (Harvard

Apparatus) at a rate of 0.5 µL/min, using the following coordinates relative to bregma: AP, -

5.2mm; ML, -2.0mm; DV, -7.5mm (Paxinos & Watson, 1997). After surgery, animals were housed individually on a regulated 12-hour light/dark cycle and allowed food and water ad libitum.

2.2.3 Medial forebrain bundle axotomy

Three weeks following AAV injection, rats received unilateral axotomy of the medial forebrain bundle (MFBx), ipsilateral to the side of injection. A retractable wire knife was inserted directly into the brain at the following coordinates relative to bregma: AP, -3.8mm; ML,

-2.4mm; DV, -8.0mm (Paxinos & Watson, 1997). The blade was extended, then raised 2.5mm, then lowered 3.0mm. The blade was retracted and removed. After surgery, animals were housed individually on a regulated 12-hour light/dark cycle and allowed food and water ad libitum. Rats were sacrificed 7 days post-MFBx (see Section 2.2.4).

2.2.4 Processing of rat brains

Rats were anaesthetized, and then sacrificed by exsanguination and transcardial perfusion with cold saline, followed by cold 4% paraformaldehyde (PFA). Brains were then removed and stored at 4°C, first in 4% PFA for 24 hours, then put through a 15% to 30% sucrose-PBS gradient. In order to generate sections, brains were frozen on a stage and then six series of 40µm coronal sections were generated using a sliding microtome. Each series of sections was stored in cryoprotectant (30% glycerol, 30% ethoxyethanol, 40% PBS) at -20°C.

2.2.5 Immunofluorescence staining of rat brain sections for TH and BAG5

Cryoprotectant was washed off with several rinses of 0.2% PBS-Tween 20 (PBS-T), then sections were blocked in 10% NGS + 2% BSA in 0.2% PBS-T for one hour at room temperature.

Sections were then incubated at room temperature overnight in primary antibody: rabbit polyclonal anti-BAG5, 1:1000 (Imgenex cat. #IMG-5678, identical to Novus Biologicals cat.

#NB100-56091); mouse anti-TH, 1:1000 (Millipore cat. #MAB318). After several washings in

PBS-T, sections were incubated in goat anti-mouse IgG biotin-SP at 1:1000 (Jackson

ImmunoResearch cat. #115-065-146) for 1 hour at room temperature, in order to amplify TH staining using the biotin-streptavidin system. After another several washings in PBS-T, sections were incubated for 1 hour at room temperature in either secondary fluorescent antibody (for

BAG5 staining) goat Alexa Fluor anti-rabbit 594, 1:500 (Invitrogen cat. #A-11037); and/or in either Alexa Fluor streptavidin 350, 1:500 (Molecular Probes cat. #S11249) or Alexa Fluor streptavidin 594, 1:500 (Molecular Probes cat. #S32356). Sections were again washed in PBS-T, then mounted on glass slides, allowed to dry, then covered by a coverslip with fluorescent mounting medium (Dako cat. #S302380-2).

2.2.6 Immunostaining of rat brain sections for TH and NeuN

Cryoprotectant was washed off with several rinses of 0.1% PBS-T. Endogenous peroxidase was quenched by washing sections in 3% H2O2 (Sigma cat. #H-1009) for 3 minutes, followed by another several rinses in PBS-T. Sections were blocked in 10% NGS + 2% BSA in

0.2% PBS-T for one hour at room temperature. Sections were then incubated at room temperature overnight in primary mouse anti-NeuN antibody, 1:1000 (Chemicon cat.

#MAB377). After several washings in PBS-T, sections were incubated in secondary antibody for

2 hours at room temperature: biotinylated anti-mouse IgG from goat, 1:400 (Vector cat. #BA-

9200). After washing in PBS-T, sections were incubated for 1 hour at room temperature in ABC

Elite mix according to the manufacturer’s instructions (Vector cat. #PK-6100). Sections were rinsed in 0.1M Dulbecco’s PBS (Sigma), and then stained with DAB (3,3′-Diaminobenzidine) according to the manufacturer’s instructions (Vector cat. #SK-4105). This was followed by

another several rinses in PBS-T and overnight, room temperature incubation in the second primary antibody, rabbit anti-TH, 1:2000 (Chemicon cat. #AB152). Sections were rinsed and incubated in secondary antibody as previously described, using alkaline phosphatase-conjugated anti-rabbit IgG from goat, 1:400 (Jackson ImmunoResearch cat. #111-055-144). This was followed by another series of PBS-T rinses, then by incubation in Vector Blue substrate according to the manufacturer’s instructions (Vector cat. #SK-5300). Sections were again washed in PBS-T, then mounted on glass slides, allowed to dry overnight, then covered by a coverslip with Vectamount (Vector cat. #H-5000).

2.2.7 Semi-quantification of BAG5 expression in SN using ImageJ

Paired images of rat SN stained for BAG5 were acquired at 100x magnification using a deconvolution microscope (Zeiss Axioplan 2). Paired images were captured using identical exposure times, set to the automatically determined optimal exposure time for the uninjected SN using AxioVision 4.8 software (Zeiss). These images were opened in grayscale using ImageJ software (Schneider et al., 2012)109. A freeform region of interest (ROI) was used to enclose the area of the SNc on the image taken contralateral to virus injection, and this ROI was mirrored onto the paired image. Using the “Find Maxima” tool with noise tolerance value arbitrarily set to

8, the number of points of local pixel intensity maxima within each ROI was identified, as a proxy for overall intensity (see discussion and Fig. 6).

2.2.8 Statistical analyses

Statistical analysis was conducted using GraphPad Prism 6.0 software (GraphPad

Software, La Jolla, California, USA, www.graphpad.com). Data containing three or more groups were analyzed by one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test.

Statistical significance was arbitrarily set to p < 0.05.

FIGURE 6.

Figure 6. Demonstration of semi-quantitative method of assessing BAG5 immunofluorescence in SN, using “find maxima” function in ImageJ.

2.3 Results

2.3.1 AAV1/2 achieves greater expression in rat SN than LV

As an initial step in our in vivo experiments, LV and AAV1/2 vectors were tested for expression in rat brain. We used AAV-GFP rather than AAV-shBAG5 vectors here because the latter were not yet available. Four weeks following unilateral injection of virus into the rat SN, coronal midbrain sections were examined for GFP expression (Fig. 7). Substantial GFP signal was observed in and around the SN in rats injected with AAV-GFP vector, but LV-Control and

LV-shBAG5 injected brains showed very poor GFP signal. These sections were also stained by immunofluorescence as an attempt to amplify any GFP signal that may be too weak to see by eye. However, even after signal amplification, GFP signal in the LV sections was poor. There was occasionally some GFP signal near the injection track, but not in SN, and signal was generally highly inconsistent. The reason for the LV vectors’ poor expression is unclear, as they efficiently transduced cultured cells (data not shown). Based on these results, we decided to use only the AAV vectors in subsequent experiments.

FIGURE 7.

Figure also contributed by C. L

Figure 7. GFP expression in SN-containing sections of LV- or AAV-injected rat brains, at four weeks post-injection. Top row: GFP signal without amplification, in green channel. Second row: immunofluorescence-amplified GFP signal, in red channel. Bottom row: merge of green and red channels.

2.3.2 AAV-shBAG5 vectors express throughout SN at level of MTN

All three AAVs provided excellent mediolateral coverage of the SNc, as visualized by

GFP expression throughout the SNc, in MTN-containing sections (Fig. 8). All three AAVs also provided fairly comprehensive SN coverage along the anteroposterior axis (images not shown), though GFP expression tended to appear slightly weaker in the most posterior sections. With the

BAG5-targeting AAVs, GFP expression was well localized to SNc, though some green signal was observed in the surrounding tissues, especially those immediately dorsal to the SN. By contrast, AAV-shBAG5 Control resulted in a large spread of strong GFP expression dorsal to the

SN, often covering nearly half the injected hemisphere.

FIGURE 8.

Figure 8. SNc-containing sections from rat brains injected with AAV-shBAG5 Control, C, or G1 at 1/2 stock dilution (resulting in 5.5×1011 gp/mL), at three weeks post-injection.

Immunofluorescence staining of tyrosine hydroxylase (TH) in red channel, merged with endogenous GFP in green channel, reveal the coverage of AAV expression in and around SNc at the level of the medial terminal nucleus (MTN), the area of non-TH immunoreactivity indicated by the white arrow.

2.3.3 AAV-shBAG5 vectors at 1/2 dilution knock down BAG5 in vivo

Having confirmed excellent AAV coverage at the level of the MTN at three weeks following a 1/2 of stock AAV dilution (resulting in 5.5×1011 gp/mL), we next sought to establish knockdown of BAG5 in these same animals by immunostaining for BAG5. Most MTN- containing sections from rats injected with either of the targeting AAV-shBAG5 vectors displayed clear BAG5 knockdown in the injected SN, by visual inspection and comparison against the uninjected SN as an internal control (Fig. 9A-F). Furthermore, the overall degree of

BAG5 knockdown was judged to be roughly 50% with both AAV-shBAG5 C and G1, again by inspection.

In order to confirm BAG5 knockdown and to eliminate potential observer bias, we developed a semi-quantitative image analysis method that could be used to compare BAG5 staining in the injected SN to the uninjected SN (see section 2.2.7). Based on this method, which uses an algorithm to identify the number of local pixel intensity maxima within a user-defined region of interest, it was confirmed that both AAV-shBAG5 C and G1 resulted in a significantly greater BAG5 knockdown in the injected SN than that produced by AAV-shBAG5 Control (Fig.

9G). Specifically, the ratio of total maxima in injected SN to total maxima in uninjected SN was found to be the following (reported as mean ± standard deviation): Control (0.62 ± 0.25); C (0.34

± 0.19); G1 (0.33 ± 0.21). Paired sections judged to have exceptionally poor BAG5 staining were excluded from the analysis. All data are presented after removal of outliers using Grubbs’ test.

FIGURE 9.

FIGURE 9 (CONT’D)

Figure 9. AAV-shBAG5 mediated BAG5 knockdown at three weeks post-injection of

AAVs at 1/2 dilution (5.5×1011 gp/mL). Representative immunofluorescence staining of BAG5 and TH in SNc of rats injected with AAV-shBAG5 Control, at low magnification (A) and high magnification (B); AAV-shBAG5 C, at low magnification (C) and high magnification (D); or

AAV-shBAG5 G1, at low magnification (E) and high magnification (F). Maxima-based semi- quantification of BAG5 staining in SNc of MTN-containing sections, expressed as the ratio between right side (RS; injected) and left side (LS; uninjected) total maxima (n=6 rats injected with each AAV; three MTN-containing sections analyzed for each rat). Statistical analysis performed by ANOVA with Tukey’s post-hoc test, * indicates p < 0.05.

2.3.4 AAV-shBAG5 vectors at 1/3 dilution knock down BAG5 in vivo

Since our AAV-shBAG5 vectors also express GFP, which has previously been shown to be toxic to neurons when overexpressed in rat SN33, we next assessed the expression and knockdown capabilities of our AAV-shBAG5 vectors when injected at a lower dilution (1/3 of stock, resulting in 3.7×1011 gp/mL) but otherwise using the same experimental parameters as before. At three weeks following AAV injection at this lower dilution, immunostaining for

BAG5 demonstrated substantial BAG5 knockdown in the injected SN, by visual inspection and comparison against the uninjected SN (Fig. 10A-F). Similar to observations with the 1/2 dilution, the degree of knockdown achieved by C and G1 was judged to be roughly 50% for both AAVs, by inspection. As before, the maxima-based semi-quantitative method was applied to confirm knockdown, and again showed that significantly greater BAG5 knockdown was achieved by the two targeting AAVs compared to AAV-shBAG5 Control (Fig. 10G). Specifically, the ratio of total maxima in injected SN to total maxima in uninjected SN was found to be the following

(reported as mean ± standard deviation): Control (1.03 ± 0.36); C (0.59 ± 0.19); G1 (0.57 ±

0.27). Paired sections judged to have exceptionally poor BAG5 staining were excluded from the analysis. All data are presented after removal of outliers using Grubbs’ test.

By inspection, the overall magnitude of BAG5 knockdown produced by the 1/3 dilution appeared to be slightly less than that produced by the 1/2 dilution. However, the difference between the magnitudes of knockdown achieved by the two dilutions is difficult to quantify, and the maxima-based method does not provide a suitable measure of this.

FIGURE 10.

FIGURE 10 (CONT’D)

Figure 10. AAV-shBAG5 mediated BAG5 knockdown at three weeks post-injection of

AAVs at 1/3 dilution (3.7×1011 gp/mL). Representative immunofluorescence staining of BAG5 and TH in SNc of rats injected with AAV-shBAG5 Control, at low magnification (A) and high magnification (B); AAV-shBAG5 C, at low magnification (C) and high magnification (D); or

2.3.5 AAV-shBAG5 vectors protect against MFBx injury

Following confirmation that our AAV-shBAG5 vectors are able to knock down BAG5 in rat SN at the tested dilutions, we next conducted a preliminary investigation into the neuroprotective potential of BAG5 knockdown in the MFBx model of dopaminergic injury23.

Injection of either targeting AAV three weeks prior to MFBx resulted in significant protection from loss of TH in the ipsilateral SN at 7 days post-MFBx (Fig. 11A). The percentages of TH+ profiles in the injected SN compared to the uninjected SN (reported as mean

± SEM) are as follows: Control (41.9 ± 11.7); C (70.9 ± 11.0); G1 (70.2 ± 10.1).

In addition, AAV-shBAG5 C significantly rescued the loss of mature neurons (NeuN+ cells) in the SN, while AAV-shBAG5 G1 trended towards rescue but did not reach statistical significance (Fig. 11B). The percentages of NeuN+ profiles in the injected SN compared to the uninjected SN (reported as mean ± SEM) are as follows: Control (30.8 ± 5.0); C (46.1 ± 3.6); G1

(42.5 ± 8.0). All data are presented after removal of outliers using Grubbs’ test.

FIGURE 11.

Figure 11. Effect of AAV-shBAG5 injected at 1/2 dilution (5.5×1011 gp/mL) on TH and

NeuN expression in SN of rats receiving medial forebrain bundle axotomy at 7 days post- axotomy (four weeks post-injection). Ratio of TH+ (A) and NeuN+ (B) profiles on injected vs. contralateral (i.e. uninjected) side at the level of the MTN (n=6 for Control, n=8 each for C and

G1; three MTN-containing sections analyzed per animal). All data are reported as mean ± SEM.

Statistical analysis performed by ANOVA with Tukey’s post-hoc test, * indicates p < 0.05.

2.4 Discussion

2.4.1 Rationale for selecting AAV1/2 serotype and injection titers

Although AAV2 has typically been the vector of choice for clinical gene therapy trials for

CNS disorders16, we chose to use AAV1/2 vector in our work for several reasons. Firstly, it combines the strong neuronal tropism of AAV2 with the widespread penetrance in brain tissue of

AAV133,110 to ensure full coverage of the SN. In addition, expertise about this specific vector was readily available from our collaborators, who developed a rat model of PD using AAV1/2 expressing the A53T mutant of human αSyn33,111. Based on their previous reports, we were able to choose an appropriate injection titer to minimize GFP-mediated toxicity by comparing titers to the previously characterized AAV1/2-GFP in rat SN111. Our first choice of titer (Fig. 9) of 1/2 stock (5.5×1011 gp/mL) was chosen to closely match the 5.1×1011 gp/mL titer of AAV1/2-GFP used by Koprich et al111, which showed no trend toward toxicity at either 3 or 6 weeks post- injection, as determined by TH expression and forepaw asymmetry111. Finally, using an AAV1/2 vector for shRNA delivery would enable us, as a future experiment, to test our vectors in the

AAV1/2-hA53T-SNCA model while ensuring consistency in spatiotemporal expression (see

Section 3.3.1).

2.4.2 Comparison of AAV to LV

It is unclear why the LV vectors failed to express any detectable amounts of GFP in the rat SN or around the injection site in the midbrain, even after signal amplification. Our injection titer of 9.0×104 IU/µL (or 9.0×107 IU/mL) falls within the range of titers used by other groups to achieve lentivirus-delivered GFP expression in rat SN. For instance, Harms et al. successfully expressed GFP with a 2µL injection of LV at a titer of 3.3×107 IU/mL, at both 2 and 5 weeks post-injection112. Similarly, Yin et al. found success with a 1µL injection of 1.0×108 IU/mL of

LV, and detected GFP expression in SN as early as 72 hours post-injection113. A third group was

successful using a higher titer of 9.0×108 IU/mL of LV, at 8 weeks post-injection114. Besides the injection titer, we also chose a well-established mammalian constitutive promoter (CMV) for protein expression, used standard stereotactic coordinates for SN injection, and allowed a reasonable time frame (4 weeks post-injection) for expression. Indeed, the LVs were injected using the same techniques and parameters as AAV-GFP (Fig. 7), which did show robust GFP expression at 4 weeks post-injection.

Given the lack of GFP expression following LV injection, it seemed likely that the shRNAs from the LV would also be poorly expressed. We therefore did not investigate BAG5 knockdown in the LV-injected brains. Based on the these results, we conclude that at least in our hands, AAV1/2 is superior to LV as a viral gene delivery vector in the rat SN, and we thus proceeded with AAV for the remainder of the experiments.

2.4.3 Spatial expression of AAV and analysis at level of MTN

The excellent coverage of the SN in sections containing MTN (Fig. 8) reinforces our decision to analyze SN at that level. We can be confident that our analysis of the SN includes

AAV-infected neurons along the full extent of the SN while it also excludes TH+ neurons of the

VTA based on the anatomy defined by the MTN. By narrowing the focus of analysis toward an area relatively dense in nigral dopaminergic neurons infected with AAV, differences between the injected and uninjected sides become easier to detect, since the suppressing effect of irrelevant background is minimized.

The widespread mediolateral SN coverage pattern in MTN-containing sections was observed at a relatively early time point of three and four weeks post-injection (as in Figs. 8-10, and Figs. 7 & 11, respectively). Although this is the only time window of AAV expression reported here, others have reported sustained protein expression with AAV1/2 at six33 and

twenty-four110,115 weeks after injection in rat brain. AAV expression in humans has been shown to last even longer: in one patient enrolled in a hemophilia gene therapy trial, AAV2 continued expressing after 10 years in non-dividing cells (skeletal muscle)116.

Based on the spatiotemporal expression of GFP, we expect that AAV-shBAG5 vectors will also be able to deliver widespread and sustained expression of shRNA well beyond the demonstrated three-week window. The ability to produce sustained BAG5 knockdown over several months would allow these vectors to be delivered in combination with one of several rat models of PD in which pathology and behavioural deficits may take several weeks to fully develop33,111,117 (also see section 3.3). It also better supports experimental designs where AAV- shBAG5 is injected prior to the induction of the PD model, rather than a simultaneous or delayed injection.

There are several limitations to narrowing the analysis of the SN to MTN-containing sections. Firstly, this approach ignores SN in non-MTN containing sections, which together represent a large part of the SN structure. We did not formally evaluate or confirm BAG5 knockdown on sections without MTN. If BAG5 is only knocked down in MTN-containing sections and not elsewhere, the functional consequences of BAG5 knockdown may be difficult to detect. Furthermore, it is possible that BAG5 knockdown in different regions of the SN can have different effects on the animal, and certain regions may be more critical for normal physiological functioning than others. Going forward, it may be important to analyze the other regions of the

SN to see how a single injection at our target coordinates affects BAG5 knockdown and pathology. If knockdown is insufficient in some areas of the SN, multiple injections and/or increased injection titer may be needed. Based on our observations however, most brains injected

with AAV-shBAG5 have displayed fairly comprehensive anteroposterior SN coverage, which should alleviate some of the concerns outlined above.

2.4.4 ImageJ fluorescence semi-quantification method

We have been unable to find methods in the literature to rapidly quantify or semi- quantify protein knockdown in the SN. In fact, many studies do not quantify in vivo knockdown at all, possibly due to the technical challenges involved.

We sought to develop a relatively rapid and unbiased approach to compare levels of

BAG5 protein in the injected SN versus the contralateral SN, free from observer bias and more quantitative than a simple visual assessment of staining intensity. The “find maxima” function in

ImageJ109 identifies local intensity maxima within a user-specified region of interest. This simple and efficient method fairly consistently produces results that match relative staining intensity as judged by eye (albeit subjectively). It is more reliable and less subject to human bias compared to sampling-based strategies, and is better able to capture small differences compared to mean intensity based methods, in which non-specific/background staining tends to wash out intensity differences between the two SN. This issue is especially problematic in the case of BAG5, which does not stain very strongly by immunofluorescence. In addition, unlike mean intensity based methods, the maxima method is mostly unaffected by technical imperfections such as tissue cracking or overlapping.

One caveat of the maxima method is that it becomes unreliable when the staining is extremely weak (i.e. almost indistinguishable from background), as this results in a low number of total maxima, which could more easily skew the ratio of maxima between the injected and uninjected SN. For this reason we have excluded such sections from the analysis. In addition, due to the relatively weak detection of BAG5 by immunostaining, SN images must be captured

at 100x magnification to ensure clarity. However, it is not possible to capture the entire SN in a single image at this magnification with the equipment available to us (only around 80% of the

SN can be captured at this magnification), thus excluding analysis of the most medial and most lateral portions of the structure. The maxima method somewhat relies on the assumption that the areas of high light intensity in a given area are spread out. For instance, if all the areas of intensity are very tightly clustered together, this method will report a low number of total maxima, even if the mean intensity in the area is very high.

2.4.5 Analysis of BAG5 knockdown in rat SN

At both the 1/2 and 1/3 titer, the BAG5 targeting AAVs had very similar injected/uninjected ratios, suggesting they have a similar knockdown efficiency. At both tested titers, the targeting AAVs resulted in significantly greater knockdown than the non-targeting

Control, according to the maxima analysis method. Though we did not quantify magnitude of knockdown, our roughly estimated knockdown of 50% for both targeting AAVs is similar to previous reports of in vivo intranigral delivery of shRNA against αSyn (35% knockdown with

AAV2)13 and interleukin-1 receptor (40% knockdown with lentivirus)15. Unlike our non- targeting Control AAV, control vectors in these two studies resulted in no appreciable knockdown of the target protein, as determined by single-cell densitometry analysis.

The maxima count does not necessarily scale linearly with the total brightness in a given image or region of interest, because it depends on the pattern of intensity distribution throughout the ROI. Thus, the maxima ratio between injected and uninjected SN derived from this method should not be interpreted as the magnitude of knockdown. Instead, it provides an unbiased method to support protein knockdown already observed by eye.

In theory the maxima counts in uninjected SN and SN injected with AAV-shBAG5

Control should be the same, since the Control shRNA should not produce BAG5 knockdown. As a result, the maxima ratio between injected and uninjected SN for Control AAV brains should theoretically be 1. This ratio was indeed very close to 1 when Control AAV was injected at 1/3 dilution (Fig. 10). At 1/2 dilution however, the mean injected to uninjected ratio across SN pairs of 0.62 suggests that BAG5 signal tended to be weaker in the injected SN vs. the uninjected SN

(Fig. 9). One potential explanation is that the Control shRNA had some off-target effect leading to down-regulation of BAG. Partial complementation between an shRNA targeting sequence and a stretch of mRNA can result in translation repression or mRNA degradation67,118. This type of

“specific off-target effect” can depend on the dose of shRNA119, which fits our observations of apparent knockdown in the higher dilution only. As well, cells that are overloaded with siRNA/shRNA may exhibit “non-specific off-target effects” as a result of endogenous miRNAs, which normally help control gene expression, being displaced from RISC120. Again, this would fit our observation that only the higher titer of Control AAV results in apparent BAG5 knockdown.

2.4.6 Alternative methods to measure BAG5 knockdown in vivo

It is also possible to quantitatively assess BAG5 knockdown in vivo, though these methods tend to be more complicated and time-consuming. One such method is single-cell densitometry analysis13,15, a technique in which many individual cells in each SN would be imaged followed by densitometry analysis of each cell. Single-cell densitometry has been used to quantify levels of αSyn13 and interleukin-1 receptor 115 in rat SN following delivery of shRNA against those molecules. However, it relies on having a strong antibody signal and may still be subject to selection bias when the individual cells are selected. Since BAG5 displays a fairly weak signal by immunostaining, single-cell densitometry may not be the most appropriate

method for our purposes. Another option for quantifying BAG5 knockdown in the SN is to dissect out the SN and western blot the tissue lysate121,122. This option surveys the entire SN without having to select a sample of cells, and may function without a strong antibody. It may be a viable way to further validate BAG5 knockdown in rat SN. As a third option is to use laser capture micro-dissection to excise individual cells from the SN, followed by protein (or RNA) analysis123,124. One benefit of this technique is the ability to analyze only infected neurons, avoiding other cell types and uninfected cells, which could dilute the measured knockdown effect.

As previously suggested, an alternate approach is to examine BAG5 mRNA rather than protein. Analyzing the levels of BAG5 mRNA more directly measures the effects of shRNA, and would be a valuable complement to protein analysis. One of the most popular techniques to quantify a specific mRNA molecule from a tissue sample is qPCR. Another option is in situ hybridization (followed by quantification), which can show the localization of the mRNA of interest, unlike traditional PCR methods that require RNA isolation. However, this technique is also limited by the nature of overlapping cells of mixed populations in tissue sections.

2.4.7 Potential toxicity from GFP overexpression

As shown by both western blot of infected primary hippocampal cells (Fig. 5) and by visual inspection of sections from AAV-shBAG5 injected SN (Figs. 8-10), the Control AAV tends to produce greater GFP expression than either of the two targeting AAVs, which appear to produce comparable levels of GFP expression compared to each other. The reason for the difference in GFP expression is unclear, given that all three AAVs are completely identical in sequence, aside from the shRNA targeting sequence. One possibility is that the Control AAV had a greater functional titer (i.e. more AAV able to infect cells) than the other AAVs, despite all

three AAVs having the same physical titer (in terms of genomic particles per mL). Another explanation could be that both shRNA targeting mRNAs exhibit off-target effects that reduce

GFP expression.

High levels of GFP expression have been reported to be toxic to dopaminergic neurons in the rat SN, in a dose-dependent fashion33,111,125. Koprich et al reported that delivery of AAV1/2-

GFP vector to the rat SN produced a 24% loss of TH+ neurons at three weeks post-injection using a titer of 5.1×1012 gp/mL, but no loss was observed when AAV titer was diluted down to

1.7×1012 gp/mL33,111. We chose to use relatively low injection titers in an attempt to avoid GFP- associated toxicity. Our higher titer (5.5×1011 gp/mL) was more than three times less concentrated than the “safe” titer of 1.7×1012 gp/mL reported by Koprich et al111. Despite this precautionary measure, it is possible that the reduction of BAG5 in SN injected with our higher titer of Control AAV is (at least partially) a result of GFP-mediated toxicity (Fig. 9). If this is indeed the case, the efficiency of BAG5 knockdown may be even greater than originally observed. The Control AAV (presumably causing increased cell death) would reduce the amount of BAG5 available for detection, but the targeting AAVs (presumably resulting in less cell death) still resulted in apparent knockdown compared to Control.

The intended purpose of the AAV-shBAG5 Control vector was to control for both shRNA and GFP expression. However, assuming the functional titers of all three AAVs are identical, the GFP expression difference makes it such that the Control vector is only

(presumably) appropriately controlling for shRNA load, but not GFP load. It is important for the

AAVs to co-express a marker in order to easily confirm and visualize infection and expression, as shRNA expression itself is difficult to detect. Ideally a less toxic expression marker could be used, but GFP already seems to be the marker of choice, with relatively low toxicity.

Unfortunately, dopaminergic neurons tend to be particularly susceptible to insult125, so other protein markers may present the same problem. AAV titer can always be reduced further, but this risks insufficient shRNA expression to achieve a detectable effect.

2.4.8 Effects of AAV-shBAG5 on TH expression

The effect of AAV-shBAG5 vectors on TH immunoreactivity was found to be highly variable across brain sections and animals, with either injection titer. We did not formally quantify TH expression in the immunofluorescence-stained sections (Figs. 9 & 10) partly for this reason. While many sections showed robust and equal levels of TH immunoreactivity in both SN

(with any of the three AAVs), several sections also showed very poor TH staining. The TH signal in these sections appeared ghostlike, with the fluorescence signal in apparently TH+ cells being significantly weaker than truly TH+ cells in the same section or same animal.

This issue may be related to tissue damage resulting from the injection of several microliters of liquid into the brain parenchyma. Indeed, the “ghosting” phenomenon was observed mainly in sections near the injection site and in the middle of the mediolateral extent of the SN structure, where the injections were targeted. The mediolateral and anteroposterior extremes of the SN structure tended to be relatively spared.

We stained TH with two fluorophores, Alexa 594 (red) and Alexa 350 (blue) to try and rule out problems with a specific fluorophore or fluorescence channel, but the results were identical in both channels. The pattern of TH immunoreactivity following AAV-shBAG5 injection will require further investigation with a larger sample size.

2.4.9 Effects of pre-injected AAV-shBAG5 following MFBx

Seven days following MFBx, AAV-shBAG5 Control-injected rats demonstrated a 42% preservation of TH immunoreactivity (Fig. 11). This figure is very similar to the report by

Crocker et al of 44% TH preservation using adeno-virus overexpressing β-galactosidase (as a control in their study)23. However, others have reported that in non-injected SNc at 7 days post-

MFBx, TH immunoreactivity only falls to around 70% of contralateral24,25, implying that virus injection and/or protein overexpression does result in some toxicity in rat SN.

Despite this apparent toxicity, both targeting AAVs significantly mitigated TH loss following MFBx, suggesting a protective effect of BAG5 knockdown (Fig. 11). In support of this, clone C significantly preserved loss of mature neurons (NeuN+) in the injected SN, while clone G1 trended toward reducing NeuN loss. This difference between the two targeting AAVs could be due to a lack of statistical power, in which case future studies may consider expanding the sample size. It is also possible that the two targeting AAVs have slightly different off-target effects that impact neuronal survival/NeuN expression, since the shRNA targeting sequences are different.

These results again confirm successful BAG5 knockdown and demonstrate that the levels of BAG5 knockdown produced by our AAVs, though not formally quantified, were sufficient to achieve measurable and relevant biological effects. Additionally, they show that the protective effects of the targeting AAVs in the MFBx model outweigh potential GFP-mediated toxicity at the 1/2 titer.

3 Chapter 3: General Discussion and Conclusion 3.1 Summary

The aim of this thesis was to develop and characterize a viral vector that could produce

shRNA-mediated BAG5 knockdown in rat substantia nigra. We set out to accomplish this in two

chapters. In chapter 1, we identified efficient shRNAs against BAG5 from a pool of candidates

and packaged them, along with a non-targeting Control, in lentivirus and AAV vectors. We

confirmed that the AAVs were able to infect primary neurons and knock down BAG5.

In chapter 2, we tested the LV and AAV vectors in rats, but discovered that the LVs did

not express in the rat brain, despite LV being an established vector for gene delivery in the rat

brain. By contrast, the AAV vectors achieved strong expression throughout the extent of the

substantia nigra. Upon analysis of sections close to the injection site, both clones of targeting

AAV-shBAG5 were able to knock down BAG5 at two different dilutions. Finally, pre-injection

of either targeting AAV-shBAG5 vectors resulted in significant protection from axotomy-

induced dopaminergic degeneration.

To our knowledge, this is work is among the first to study shRNA delivery to the rat

substantia nigra using the AAV1/2 serotype. The only published study to date using specifically

AAV1/2 to deliver shRNA in vivo targeted the red nucleus in rats41. Our work contributes to a

rapidly emerging body of research demonstrating the effectiveness of AAV1/2 as an shRNA

delivery vector in animal experiments. As gene therapy delivery research moves forward, it will

be important to understand the performance of different AAV serotypes in vivo as one of several

dimensions of spatiotemporal control.

3.2 Limitations

Firstly, it is worth noting that we studied knockdown but not knockout, and these two conditions may have different phenotypes67. We did not formally quantify BAG5 knockdown, and did not formally demonstrate knockdown throughout the entire SN, since we only analyzed

MTN-containing sections. Thus, it is possible that our current injection target and titer does produce sufficiently great or widespread knockdown to exert the full phenotypic effects of

BAG5 inhibition. Performing multiple injections or increasing titer may resolve some of these potential issues, but these modifications come with their own risks, such as increased toxicity from GFP load, or physical damage to the tissue from injecting larger volumes of liquid.

Despite having demonstrated some protective effects of BAG5 knockdown with histological analyses, we did not show that this effect translates to functional improvement with behavioural tests. Again, it is thus unclear how impactful our attained levels of BAG5 knockdown will be in terms of improving behavioural deficits when applied in animal models of

PD.

As with any animal model, the MFBx model is limited in how well it can represent human disease states. Other rat models that better recapitulate some of the molecular features of

PD, such as the presence of human αSyn, have recently gained favour over MFBx as models of

PD in rats. However, MFBx is still a useful model of dopaminergic degeneration, and we have shown that BAG5 knockdown does have biologically relevant consequences in the context of this particular injury. Accordingly, it has potential to protect dopaminergic neurons in other contexts of insult, including toxin and protein overexpression-based models of Parkinson’s disease, though this remains an open question for future investigation.

3.3 Future Directions

3.3.1 Human A53T-SNCA overexpression model of PD

Having established successful BAG5 knockdown using our AAV-shBAG5 vectors, and demonstrated protective effects in the MFBx model of dopaminergic injury, the next step would be to test this therapeutic strategy in a rat model of PD. The human A53T mutant αSyn overexpression model of PD (delivered by AAV1/2) is well characterized in terms of motor phenotype and timeframe of dopaminergic neuron loss, both of which are viral titer- dependent33,111. We attempted to test the effects of BAG5 knockdown in this model by co- injecting our AAV-shBAG5 vectors with AAV-A53T-SNCA. However, at 6 weeks post- injection, the hA53T-SNCA overexpression model did not behave as previously characterized in terms of behaviour and pathology33,111, so we were unable to interpret the influence of BAG5 knockdown in this system (data not shown). Thus some technical challenges remain to be addressed before testing our AAV-shBAG5 vectors in this model.

3.3.2 VenusYFP-SNCA bimolecular protein complementation system

The αSyn bimolecular protein complementation system can be used as another rat model of PD117. In this model, an AAV encoding the N-terminus half of VenusYFP fused to full-length wild-type human αSyn is co-injected with an AAV encoding αSyn fused to the C-terminus half of VenusYFP. When the protein products of both AAVs are expressed in the same cell, the oligomerization of αSyn joins the halves of Venus and produces a fluorescent signal, allowing for the visualization of αSyn oligomers in vivo. This complementation system has previously been implemented in rats using an AAV8 vector for delivery117. We are currently working on characterizing the behavioural and histopathological outcomes of this model of PD when delivered using AAV1/2 vector. After this, as a future experiment, we could pair this Venus-

αSyn model with our AAV-shBAG5 vectors (expressing RFP rather than GFP to avoid

confusion with Venus signal) to evaluate the effects of BAG5 knockdown on motor behaviour

(by rescue of forepaw asymmetry), dopaminergic cell death, and synuclein pathology.

3.4 Conclusion

We have created and characterized an AAV1/2 vector capable of selectively and quite comprehensively transducing neurons in the rat substantia nigra, producing substantial knockdown of BAG5 in neurons as early as three weeks post-injection. This work establishes the foundation for future investigations into the role of BAG5 in the SN, especially in relation to neurodegeneration and Parkinson’s disease.

70 References

1 Abdullah, R., Basak, I., Patil, K. S., Alves, G., Larsen, J. P. & Moller, S. G. Parkinson's disease and age: The obvious but largely unexplored link. Exp Gerontol 68, 33-38, doi:10.1016/j.exger.2014.09.014 (2015).

2 Kalia, L. V. & Lang, A. E. Parkinson's disease. Lancet, doi:10.1016/S0140- 6736(14)61393-3 (2015).

3 Szewczyk-Krolikowski, K., Tomlinson, P., Nithi, K., Wade-Martins, R., Talbot, K., Ben- Shlomo, Y. et al. The influence of age and gender on motor and non-motor features of early Parkinson's disease: initial findings from the Oxford Parkinson Disease Center (OPDC) discovery cohort. Parkinsonism Relat Disord 20, 99-105, doi:10.1016/j.parkreldis.2013.09.025 (2014).

4 Deng, H. X., Shi, Y., Yang, Y., Ahmeti, K. B., Miller, N., Huang, C. et al. Identification of TMEM230 mutations in familial Parkinson's disease. Nat Genet, doi:10.1038/ng.3589 (2016).

5 Connolly, B. S. & Lang, A. E. Pharmacological treatment of Parkinson disease: a review. JAMA 311, 1670-1683, doi:10.1001/jama.2014.3654 (2014).

6 Wang, D. & Gao, G. State-of-the-art human gene therapy: part II. Gene therapy strategies and clinical applications. Discov Med 18, 151-161 (2014).

7 Dong, Z., Wolfer, D. P., Lipp, H. P. & Bueler, H. Hsp70 gene transfer by adeno- associated virus inhibits MPTP-induced nigrostriatal degeneration in the mouse model of Parkinson disease. Mol Ther 11, 80-88, doi:10.1016/j.ymthe.2004.09.007 (2005).

8 Rocha, E. M., Smith, G. A., Park, E., Cao, H., Brown, E., Hayes, M. A. et al. Glucocerebrosidase gene therapy prevents alpha-synucleinopathy of midbrain dopamine neurons. Neurobiol Dis 82, 495-503, doi:10.1016/j.nbd.2015.09.009 (2015).

9 Tian, Y. Y., Tang, C. J., Wang, J. N., Feng, Y., Chen, X. W., Wang, L. et al. Favorable effects of VEGF gene transfer on a rat model of Parkinson disease using adeno- associated viral vectors. Neurosci Lett 421, 239-244, doi:10.1016/j.neulet.2007.05.033 (2007).

10 Du, Y., Zhang, X., Tao, Q., Chen, S. & Le, W. Adeno-associated virus type 2 vector- mediated glial cell line-derived neurotrophic factor gene transfer induces neuroprotection and neuroregeneration in a ubiquitin-proteasome system impairment animal model of Parkinson's disease. Neurodegener Dis 11, 113-128, doi:10.1159/000334527 (2013).

11 Herzog, C. D., Brown, L., Kruegel, B. R., Wilson, A., Tansey, M. G., Gage, F. H. et al. Enhanced neurotrophic distribution, cell signaling and neuroprotection following substantia nigral versus striatal delivery of AAV2-NRTN (CERE-120). Neurobiol Dis 58, 38-48, doi:10.1016/j.nbd.2013.04.011 (2013).

12 Khodr, C. E., Sapru, M. K., Pedapati, J., Han, Y., West, N. C., Kells, A. P. et al. An alpha-synuclein AAV gene silencing vector ameliorates a behavioral deficit in a rat model of Parkinson's disease, but displays toxicity in dopamine neurons. Brain Res 1395, 94-107, doi:10.1016/j.brainres.2011.04.036 (2011).

13 Zharikov, A. D., Cannon, J. R., Tapias, V., Bai, Q., Horowitz, M. P., Shah, V. et al. shRNA targeting alpha-synuclein prevents neurodegeneration in a Parkinson's disease model. J Clin Invest 125, 2721-2735, doi:10.1172/JCI64502 (2015).

14 Saal, K. A., Koch, J. C., Tatenhorst, L., Szego, E. M., Ribas, V. T., Michel, U. et al. AAV.shRNA-mediated downregulation of ROCK2 attenuates degeneration of dopaminergic neurons in toxin-induced models of Parkinson's disease in vitro and in vivo. Neurobiol Dis 73, 150-162, doi:10.1016/j.nbd.2014.09.013 (2015).

15 Walsh, S., Gavin, A., Wyatt, S., O'Connor, C., Keeshan, K., Nolan, Y. M. et al. Knockdown of interleukin-1 receptor 1 is not neuroprotective in the 6-hydroxydopamine striatal lesion rat model of Parkinson's disease. Int J Neurosci 125, 70-77, doi:10.3109/00207454.2014.904304 (2015).

16 Ojala, D. S., Amara, D. P. & Schaffer, D. V. Adeno-associated virus vectors and neurological gene therapy. Neuroscientist 21, 84-98, doi:10.1177/1073858414521870 (2015).

17 LeWitt, P. A., Rezai, A. R., Leehey, M. A., Ojemann, S. G., Flaherty, A. W., Eskandar, E. N. et al. AAV2-GAD gene therapy for advanced Parkinson's disease: a double-blind, sham-surgery controlled, randomised trial. Lancet Neurol 10, 309-319, doi:10.1016/S1474-4422(11)70039-4 (2011).

18 Marks, W. J., Jr., Ostrem, J. L., Verhagen, L., Starr, P. A., Larson, P. S., Bakay, R. A. et al. Safety and tolerability of intraputaminal delivery of CERE-120 (adeno-associated virus serotype 2-neurturin) to patients with idiopathic Parkinson's disease: an open-label, phase I trial. Lancet Neurol 7, 400-408, doi:10.1016/S1474-4422(08)70065-6 (2008).

19 Warren Olanow, C., Bartus, R. T., Baumann, T. L., Factor, S., Boulis, N., Stacy, M. et al. Gene delivery of neurturin to putamen and substantia nigra in Parkinson disease: A double-blind, randomized, controlled trial. Ann Neurol 78, 248-257, doi:10.1002/ana.24436 (2015).

20 Bartus, R. T., Kordower, J. H., Johnson, E. M., Jr., Brown, L., Kruegel, B. R., Chu, Y. et al. Post-mortem assessment of the short and long-term effects of the trophic factor neurturin in patients with alpha-synucleinopathies. Neurobiol Dis 78, 162-171, doi:10.1016/j.nbd.2015.03.023 (2015).

21 Kalia, L. V., Kalia, S. K. & Lang, A. E. Disease-modifying strategies for Parkinson's disease. Mov Disord 30, 1442-1450, doi:10.1002/mds.26354 (2015).

22 Nieuwenhuys, R., Geeraedts, L. M. & Veening, J. G. The medial forebrain bundle of the rat. I. General introduction. J Comp Neurol 206, 49-81, doi:10.1002/cne.902060106 (1982).

23 Crocker, S. J., Lamba, W. R., Smith, P. D., Callaghan, S. M., Slack, R. S., Anisman, H. et al. c-Jun mediates axotomy-induced dopamine neuron death in vivo. Proc Natl Acad Sci U S A 98, 13385-13390, doi:10.1073/pnas.231177098 (2001).

24 Weiser, M., Baker, H., Wessel, T. C. & Joh, T. H. Differential spatial and temporal gene expression in response to axotomy and deafferentation following transection of the medial forebrain bundle. J Neurosci 13, 3472-3484 (1993).

25 Brecknell, J. E., Dunnett, S. B. & Fawcett, J. W. A quantitative study of cell death in the substantia nigra following a mechanical lesion of the medial forebrain bundle. Neuroscience 64, 219-227 (1995).

26 Beck, K. D., Valverde, J., Alexi, T., Poulsen, K., Moffat, B., Vandlen, R. A. et al. Mesencephalic dopaminergic neurons protected by GDNF from axotomy-induced degeneration in the adult brain. Nature 373, 339-341, doi:10.1038/373339a0 (1995).

27 Chang, J. W., Wachtel, S. R., Young, D. & Kang, U. J. Biochemical and anatomical characterization of forepaw adjusting steps in rat models of Parkinson's disease: studies on medial forebrain bundle and striatal lesions. Neuroscience 88, 617-628 (1999).

28 Tseng, J. L., Bruhn, S. L., Zurn, A. D. & Aebischer, P. Neurturin protects dopaminergic neurons following medial forebrain bundle axotomy. Neuroreport 9, 1817-1822 (1998).

29 Lu, J., Sun, F., Ma, H., Qing, H. & Deng, Y. Comparison between alpha-synuclein wild- type and A53T mutation in a progressive Parkinson's disease model. Biochem Biophys Res Commun 464, 988-993, doi:10.1016/j.bbrc.2015.07.007 (2015).

30 Moloney, T. C., Hyland, R., O'Toole, D., Paucard, A., Kirik, D., O'Doherty, A. et al. Heat shock protein 70 reduces alpha-synuclein-induced predegenerative neuronal dystrophy in the alpha-synuclein viral gene transfer rat model of Parkinson's disease. CNS Neurosci Ther 20, 50-58, doi:10.1111/cns.12200 (2014).

31 Sanchez-Guajardo, V., Annibali, A., Jensen, P. H. & Romero-Ramos, M. alpha- Synuclein vaccination prevents the accumulation of parkinson disease-like pathologic inclusions in striatum in association with regulatory T cell recruitment in a rat model. J Neuropathol Exp Neurol 72, 624-645, doi:10.1097/NEN.0b013e31829768d2 (2013).

32 Van der Perren, A., Toelen, J., Casteels, C., Macchi, F., Van Rompuy, A. S., Sarre, S. et al. Longitudinal follow-up and characterization of a robust rat model for Parkinson's disease based on overexpression of alpha-synuclein with adeno-associated viral vectors. Neurobiol Aging 36, 1543-1558, doi:10.1016/j.neurobiolaging.2014.11.015 (2015).

33 Koprich, J. B., Johnston, T. H., Reyes, M. G., Sun, X. & Brotchie, J. M. Expression of human A53T alpha-synuclein in the rat substantia nigra using a novel AAV1/2 vector produces a rapidly evolving pathology with protein aggregation, dystrophic neurite architecture and nigrostriatal degeneration with potential to model the pathology of Parkinson's disease. Mol Neurodegener 5, 43, doi:10.1186/1750-1326-5-43 (2010).

34 Kabbage, M. & Dickman, M. B. The BAG proteins: a ubiquitous family of chaperone regulators. Cell Mol Life Sci 65, 1390-1402, doi:10.1007/s00018-008-7535-2 (2008).

35 Takayama, S. & Reed, J. C. Molecular chaperone targeting and regulation by BAG family proteins. Nat Cell Biol 3, E237-241, doi:10.1038/ncb1001-e237 (2001).

36 Behl, C. Breaking BAG: The Co-Chaperone BAG3 in Health and Disease. Trends Pharmacol Sci, doi:10.1016/j.tips.2016.04.007 (2016).

37 Arakawa, A., Handa, N., Ohsawa, N., Shida, M., Kigawa, T., Hayashi, F. et al. The C- terminal BAG domain of BAG5 induces conformational changes of the Hsp70 nucleotide-binding domain for ADP-ATP exchange. Structure 18, 309-319, doi:10.1016/j.str.2010.01.004 (2010).

38 Bruchmann, A., Roller, C., Walther, T. V., Schafer, G., Lehmusvaara, S., Visakorpi, T. et al. Bcl-2 associated athanogene 5 (Bag5) is overexpressed in prostate cancer and inhibits ER-stress induced apoptosis. BMC Cancer 13, 96, doi:10.1186/1471-2407-13-96 (2013).

39 Gupta, M. K., Tahrir, F. G., Knezevic, T., White, M. K., Gordon, J., Cheung, J. Y. et al. GRP78 Interacting Partner Bag5 Responds to ER Stress and Protects Cardiomyocytes From ER Stress-Induced Apoptosis. J Cell Biochem 117, 1813-1821, doi:10.1002/jcb.25481 (2016).

40 Gotz, R., Wiese, S., Takayama, S., Camarero, G. C., Rossoll, W., Schweizer, U. et al. Bag1 is essential for differentiation and survival of hematopoietic and neuronal cells. Nat Neurosci 8, 1169-1178, doi:10.1038/nn1524 (2005).

41 Challagundla, M., Koch, J. C., Ribas, V. T., Michel, U., Kugler, S., Ostendorf, T. et al. AAV-mediated expression of BAG1 and ROCK2-shRNA promote neuronal survival and axonal sprouting in a rat model of rubrospinal tract injury. J Neurochem 134, 261-275, doi:10.1111/jnc.13102 (2015).

42 Yue, X., Zhao, Y., Liu, J., Zhang, C., Yu, H., Wang, J. et al. BAG2 promotes tumorigenesis through enhancing mutant p53 protein levels and function. Elife 4, doi:10.7554/eLife.08401 (2015).

43 Che, X., Tang, B., Wang, X., Chen, D., Yan, X., Jiang, H. et al. The BAG2 Protein Stabilises PINK1 By Decreasing its Ubiquitination. Biochem Biophys Res Commun, doi:10.1016/j.bbrc.2013.10.086 (2013).

44 Arndt, V., Daniel, C., Nastainczyk, W., Alberti, S. & Hohfeld, J. BAG-2 acts as an inhibitor of the chaperone-associated ubiquitin ligase CHIP. Mol Biol Cell 16, 5891- 5900, doi:10.1091/mbc.E05-07-0660 (2005).

45 Carrettiero, D. C., Hernandez, I., Neveu, P., Papagiannakopoulos, T. & Kosik, K. S. The cochaperone BAG2 sweeps paired helical filament- insoluble tau from the microtubule. J Neurosci 29, 2151-2161, doi:10.1523/JNEUROSCI.4660-08.2009 (2009).

46 Konersman, C. G., Bordini, B. J., Scharer, G., Lawlor, M. W., Zangwill, S., Southern, J. F. et al. BAG3 myofibrillar myopathy presenting with cardiomyopathy. Neuromuscul Disord 25, 418-422, doi:10.1016/j.nmd.2015.01.009 (2015).

47 Shi, H., Xu, H., Li, Z., Zhen, Y., Wang, B., Huo, S. et al. BAG3 regulates cell proliferation, migration, and invasion in human colorectal cancer. Tumour Biol 37, 5591- 5597, doi:10.1007/s13277-015-4403-1 (2016).

48 Mani, J., Antonietti, P., Rakel, S., Blaheta, R., Bartsch, G., Haferkamp, A. et al. Knockdown of BAG3 sensitizes bladder cancer cells to treatment with the BH3 mimetic ABT-737. World J Urol 34, 197-205, doi:10.1007/s00345-015-1616-2 (2016).

49 Minoia, M., Boncoraglio, A., Vinet, J., Morelli, F. F., Brunsting, J. F., Poletti, A. et al. BAG3 induces the sequestration of proteasomal clients into cytoplasmic puncta: implications for a proteasome-to-autophagy switch. Autophagy 10, 1603-1621, doi:10.4161/auto.29409 (2014).

50 Eichholtz-Wirth, H. & Sagan, D. Altered signaling of TNFalpha-TNFR1 and SODD/BAG4 is responsible for radioresistance in human HT-R15 cells. Anticancer Res 22, 235-240 (2002).

51 Eichholtz-Wirth, H., Fritz, E. & Wolz, L. Overexpression of the 'silencer of death domain', SODD/BAG-4, modulates both TNFR1- and CD95-dependent cell death pathways. Cancer Lett 194, 81-89 (2003).

52 Mock, J. Y., Chartron, J. W., Zaslaver, M., Xu, Y., Ye, Y. & Clemons, W. M., Jr. Bag6 complex contains a minimal tail-anchor-targeting module and a mock BAG domain. Proc Natl Acad Sci U S A 112, 106-111, doi:10.1073/pnas.1402745112 (2015).

53 Corduan, A., Lecomte, S., Martin, C., Michel, D. & Desmots, F. Sequential interplay between BAG6 and HSP70 upon heat shock. Cell Mol Life Sci 66, 1998-2004, doi:10.1007/s00018-009-9198-z (2009).

54 Chen, J., Zang, Y. S. & Xiu, Q. BAT3 rs1052486 and rs3117582 polymorphisms are associated with lung cancer risk: a meta-analysis. Tumour Biol 35, 9855-9858, doi:10.1007/s13277-014-1912-2 (2014).

55 Saresella, M., Piancone, F., Marventano, I., La Rosa, F., Tortorella, P., Caputo, D. et al. A role for the TIM-3/GAL-9/BAT3 pathway in determining the clinical phenotype of multiple sclerosis. FASEB J 28, 5000-5009, doi:10.1096/fj.14-258194 (2014).

56 Kalia, S. K., Lee, S., Smith, P. D., Liu, L., Crocker, S. J., Thorarinsdottir, T. E. et al. BAG5 inhibits parkin and enhances dopaminergic neuron degeneration. Neuron 44, 931- 945, doi:10.1016/j.neuron.2004.11.026 (2004).

57 Wang, X., Guo, J., Fei, E., Mu, Y., He, S., Che, X. et al. BAG5 protects against mitochondrial oxidative damage through regulating PINK1 degradation. PLoS One 9, e86276, doi:10.1371/journal.pone.0086276 (2014).

58 Kelly, J. L., Novak, A. J., Fredericksen, Z. S., Liebow, M., Ansell, S. M., Dogan, A. et al. Germline variation in apoptosis pathway genes and risk of non-Hodgkin's lymphoma. Cancer Epidemiol Biomarkers Prev 19, 2847-2858, doi:10.1158/1055-9965.EPI-10-0581 (2010).

59 Hasson, S. A., Kane, L. A., Yamano, K., Huang, C. H., Sliter, D. A., Buehler, E. et al. High-content genome-wide RNAi screens identify regulators of parkin upstream of mitophagy. Nature 504, 291-295, doi:10.1038/nature12748 (2013).

60 Beilina, A., Rudenko, I. N., Kaganovich, A., Civiero, L., Chau, H., Kalia, S. K. et al. Unbiased screen for interactors of leucine-rich repeat kinase 2 supports a common pathway for sporadic and familial Parkinson disease. Proc Natl Acad Sci U S A 111, 2626-2631, doi:10.1073/pnas.1318306111 (2014).

61 Kalia, L. V., Kalia, S. K., Chau, H., Lozano, A. M., Hyman, B. T. & McLean, P. J. Ubiquitinylation of alpha-synuclein by carboxyl terminus Hsp70-interacting protein (CHIP) is regulated by Bcl-2-associated athanogene 5 (BAG5). PLoS One 6, e14695, doi:10.1371/journal.pone.0014695 (2011).

62 Zheng, X. Y., Yang, M., Tan, J. Q., Pan, Q., Long, Z. G., Dai, H. P. et al. Screening of LRRK2 interactants by yeast 2-hybrid analysis. Zhong Nan Da Xue Xue Bao Yi Xue Ban 33, 883-891 (2008).

63 Auluck, P. K., Chan, H. Y., Trojanowski, J. Q., Lee, V. M. & Bonini, N. M. Chaperone suppression of alpha-synuclein toxicity in a Drosophila model for Parkinson's disease. Science 295, 865-868, doi:10.1126/science.1067389 (2002).

64 Flower, T. R., Chesnokova, L. S., Froelich, C. A., Dixon, C. & Witt, S. N. Heat shock prevents alpha-synuclein-induced apoptosis in a yeast model of Parkinson's disease. J Mol Biol 351, 1081-1100, doi:10.1016/j.jmb.2005.06.060 (2005).

65 Klucken, J., Shin, Y., Masliah, E., Hyman, B. T. & McLean, P. J. Hsp70 Reduces alpha- Synuclein Aggregation and Toxicity. J Biol Chem 279, 25497-25502, doi:10.1074/jbc.M400255200 (2004).

66 Lorenc-Koci, E., Lenda, T., Antkiewicz-Michaluk, L., Wardas, J., Domin, H., Smialowska, M. et al. Different effects of intranigral and intrastriatal administration of the proteasome inhibitor lactacystin on typical neurochemical and histological markers of Parkinson's disease in rats. Neurochem Int 58, 839-849, doi:10.1016/j.neuint.2011.03.013 (2011).

67 Boettcher, M. & McManus, M. T. Choosing the Right Tool for the Job: RNAi, TALEN, or CRISPR. Mol Cell 58, 575-585, doi:10.1016/j.molcel.2015.04.028 (2015).

68 Ge, Q., Ilves, H., Dallas, A., Kumar, P., Shorenstein, J., Kazakov, S. A. et al. Minimal- length short hairpin RNAs: the relationship of structure and RNAi activity. RNA 16, 106- 117, doi:10.1261/rna.1894510 (2010).

69 Shim, M. S. & Kwon, Y. J. Efficient and targeted delivery of siRNA in vivo. FEBS J 277, 4814-4827, doi:10.1111/j.1742-4658.2010.07904.x (2010).

70 Evers, M. M., Toonen, L. J. & van Roon-Mom, W. M. Antisense oligonucleotides in therapy for neurodegenerative disorders. Adv Drug Deliv Rev 87, 90-103, doi:10.1016/j.addr.2015.03.008 (2015).

71 Bennett, C. F. & Swayze, E. E. RNA targeting therapeutics: molecular mechanisms of antisense oligonucleotides as a therapeutic platform. Annu Rev Pharmacol Toxicol 50, 259-293, doi:10.1146/annurev.pharmtox.010909.105654 (2010).

72 Goodchild, J. Therapeutic oligonucleotides. Methods Mol Biol 764, 1-15, doi:10.1007/978-1-61779-188-8_1 (2011).

73 Sternberg, N. & Hamilton, D. Bacteriophage P1 site-specific recombination. I. Recombination between loxP sites. J Mol Biol 150, 467-486 (1981).

74 Kratochwil, C. F. & Rijli, F. M. The Cre/Lox system to assess the development of the mouse brain. Methods Mol Biol 1082, 295-313, doi:10.1007/978-1-62703-655-9_20 (2014).

75 Sauer, B. Inducible gene targeting in mice using the Cre/lox system. Methods 14, 381- 392, doi:10.1006/meth.1998.0593 (1998).

76 Theodosiou, N. A. & Xu, T. Use of FLP/FRT system to study Drosophila development. Methods 14, 355-365, doi:10.1006/meth.1998.0591 (1998).

77 Luo, Y., Wang, Y., Liu, J., Cui, C., Wu, Y., Lan, H. et al. Generation of TALE nickase- mediated gene-targeted cows expressing human serum albumin in mammary glands. Sci Rep 6, 20657, doi:10.1038/srep20657 (2016).

78 Heidenreich, M. & Zhang, F. Applications of CRISPR-Cas systems in neuroscience. Nat Rev Neurosci 17, 36-44, doi:10.1038/nrn.2015.2 (2016).

79 Ul Ain, Q., Chung, J. Y. & Kim, Y. H. Current and future delivery systems for engineered nucleases: ZFN, TALEN and RGEN. J Control Release 205, 120-127, doi:10.1016/j.jconrel.2014.12.036 (2015).

80 Jinek, M., Chylinski, K., Fonfara, I., Hauer, M., Doudna, J. A. & Charpentier, E. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science 337, 816-821, doi:10.1126/science.1225829 (2012).

81 Barakate, A. & Stephens, J. An Overview of CRISPR-Based Tools and Their Improvements: New Opportunities in Understanding Plant-Pathogen Interactions for Better Crop Protection. Front Plant Sci 7, 765, doi:10.3389/fpls.2016.00765 (2016).

82 Tsai, S. Q., Wyvekens, N., Khayter, C., Foden, J. A., Thapar, V., Reyon, D. et al. Dimeric CRISPR RNA-guided FokI nucleases for highly specific genome editing. Nat Biotechnol 32, 569-576, doi:10.1038/nbt.2908 (2014).

83 Nayerossadat, N., Maedeh, T. & Ali, P. A. Viral and nonviral delivery systems for gene delivery. Adv Biomed Res 1, 27, doi:10.4103/2277-9175.98152 (2012).

84 Cronin, J., Zhang, X. Y. & Reiser, J. Altering the tropism of lentiviral vectors through pseudotyping. Curr Gene Ther 5, 387-398 (2005).

85 Kantor, B., Bailey, R. M., Wimberly, K., Kalburgi, S. N. & Gray, S. J. Methods for gene transfer to the central nervous system. Adv Genet 87, 125-197, doi:10.1016/B978-0-12- 800149-3.00003-2 (2014).

86 Deyle, D. R. & Russell, D. W. Adeno-associated virus vector integration. Curr Opin Mol Ther 11, 442-447 (2009).

87 Salganik, M., Hirsch, M. L. & Samulski, R. J. Adeno-associated Virus as a Mammalian DNA Vector. Microbiol Spectr 3, doi:10.1128/microbiolspec.MDNA3-0052-2014 (2015).

88 Nie, D. & Sahin, M. A genetic model to dissect the role of Tsc-mTORC1 in neuronal cultures. Methods Mol Biol 821, 393-405, doi:10.1007/978-1-61779-430-8_25 (2012).

89 Herrero, J., Muffato, M., Beal, K., Fitzgerald, S., Gordon, L., Pignatelli, M. et al. Ensembl comparative genomics resources. Database (Oxford) 2016, doi:10.1093/database/bav096 (2016).

90 Wild, E. J. & Tabrizi, S. J. Targets for future clinical trials in Huntington's disease: what's in the pipeline? Mov Disord 29, 1434-1445, doi:10.1002/mds.26007 (2014).

91 Barrangou, R., Birmingham, A., Wiemann, S., Beijersbergen, R. L., Hornung, V. & Smith, A. Advances in CRISPR-Cas9 genome engineering: lessons learned from RNA interference. Nucleic Acids Res 43, 3407-3419, doi:10.1093/nar/gkv226 (2015).

92 Lee, J. H., Ahn, H. H., Kim, K. S., Lee, J. Y., Kim, M. S., Lee, B. et al. Polyethyleneimine-mediated gene delivery into rat pheochromocytoma PC-12 cells. J Tissue Eng Regen Med 2, 288-295, doi:10.1002/term.94 (2008).

93 Cogli, L., Progida, C., Lecci, R., Bramato, R., Kruttgen, A. & Bucci, C. CMT2B- associated Rab7 mutants inhibit neurite outgrowth. Acta Neuropathol 120, 491-501, doi:10.1007/s00401-010-0696-8 (2010).

94 Covello, G., Siva, K., Adami, V. & Denti, M. A. An electroporation protocol for efficient DNA transfection in PC12 cells. Cytotechnology 66, 543-553, doi:10.1007/s10616-013- 9608-9 (2014).

95 Lin, C. Y., Huang, Z., Wen, W., Wu, A., Wang, C. & Niu, L. Enhancing Protein Expression in HEK-293 Cells by Lowering Culture Temperature. PLoS One 10, e0123562, doi:10.1371/journal.pone.0123562 (2015).

96 de Los Milagros Bassani Molinas, M., Beer, C., Hesse, F., Wirth, M. & Wagner, R. Optimizing the transient transfection process of HEK-293 suspension cells for protein

production by nucleotide ratio monitoring. Cytotechnology 66, 493-514, doi:10.1007/s10616-013-9601-3 (2014).

97 Cambridge, S. B., Gnad, F., Nguyen, C., Bermejo, J. L., Kruger, M. & Mann, M. Systems-wide proteomic analysis in mammalian cells reveals conserved, functional protein turnover. J Proteome Res 10, 5275-5284, doi:10.1021/pr101183k (2011).

98 Eden, E., Geva-Zatorsky, N., Issaeva, I., Cohen, A., Dekel, E., Danon, T. et al. Proteome half-life dynamics in living human cells. Science 331, 764-768, doi:10.1126/science.1199784 (2011).

99 Schwanhausser, B., Busse, D., Li, N., Dittmar, G., Schuchhardt, J., Wolf, J. et al. Global quantification of mammalian gene expression control. Nature 473, 337-342, doi:10.1038/nature10098 (2011).

100 Zhou, P. Determining protein half-lives. Methods Mol Biol 284, 67-77, doi:10.1385/1- 59259-816-1:067 (2004).

101 Gilda, J. E. & Gomes, A. V. Western blotting using in-gel protein labeling as a normalization control: stain-free technology. Methods Mol Biol 1295, 381-391, doi:10.1007/978-1-4939-2550-6_27 (2015).

102 Thacker, J. S., Yeung, D. H., Staines, W. R. & Mielke, J. G. Total protein or high- abundance protein: Which offers the best loading control for Western blotting? Anal Biochem 496, 76-78, doi:10.1016/j.ab.2015.11.022 (2016).

103 Welinder, C. & Ekblad, L. Coomassie staining as loading control in Western blot analysis. J Proteome Res 10, 1416-1419, doi:10.1021/pr1011476 (2011).

104 Zu, Y., Huang, S., Liao, W. C., Lu, Y. & Wang, S. Gold nanoparticles enhanced electroporation for mammalian cell transfection. J Biomed Nanotechnol 10, 982-992 (2014).

105 Basu, S., Campbell, H. M., Dittel, B. N. & Ray, A. Purification of specific cell population by fluorescence activated cell sorting (FACS). J Vis Exp, doi:10.3791/1546 (2010).

106 Sorensen, T. U., Gram, G. J., Nielsen, S. D. & Hansen, J. E. Safe sorting of GFP- transduced live cells for subsequent culture using a modified FACS vantage. Cytometry 37, 284-290 (1999).

107 Marjanovic, I., Kanduser, M., Miklavcic, D., Keber, M. M. & Pavlin, M. Comparison of flow cytometry, fluorescence microscopy and spectrofluorometry for analysis of gene electrotransfer efficiency. J Membr Biol 247, 1259-1267, doi:10.1007/s00232-014-9714- 4 (2014).

108 Holmes, K., Williams, C. M., Chapman, E. A. & Cross, M. J. Detection of siRNA induced mRNA silencing by RT-qPCR: considerations for experimental design. BMC Res Notes 3, 53, doi:10.1186/1756-0500-3-53 (2010).

109 Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat Methods 9, 671-675 (2012).

110 Kiyota, T., Yamamoto, M., Schroder, B., Jacobsen, M. T., Swan, R. J., Lambert, M. P. et al. AAV1/2-mediated CNS gene delivery of dominant-negative CCL2 mutant suppresses gliosis, beta-amyloidosis, and learning impairment of APP/PS1 mice. Mol Ther 17, 803- 809, doi:10.1038/mt.2009.44 (2009).

111 Koprich, J. B., Johnston, T. H., Huot, P., Reyes, M. G., Espinosa, M. & Brotchie, J. M. Progressive neurodegeneration or endogenous compensation in an animal model of Parkinson's disease produced by decreasing doses of alpha-synuclein. PLoS One 6, e17698, doi:10.1371/journal.pone.0017698 (2011).

112 Harms, A. S., Barnum, C. J., Ruhn, K. A., Varghese, S., Trevino, I., Blesch, A. et al. Delayed dominant-negative TNF gene therapy halts progressive loss of nigral dopaminergic neurons in a rat model of Parkinson's disease. Mol Ther 19, 46-52, doi:10.1038/mt.2010.217 (2011).

113 Yin, X. F., Xu, H. M., Jiang, Y. X., Zhi, Y. L., Liu, Y. X., Xiang, H. W. et al. Lentivirus- mediated Persephin over-expression in Parkinson's disease rats. Neural Regen Res 10, 1814-1818, doi:10.4103/1673-5374.170309 (2015).

114 Cordero-Llana, O., Houghton, B. C., Rinaldi, F., Taylor, H., Yanez-Munoz, R. J., Uney, J. B. et al. Enhanced efficacy of the CDNF/MANF family by combined intranigral overexpression in the 6-OHDA rat model of Parkinson's disease. Mol Ther 23, 244-254, doi:10.1038/mt.2014.206 (2015).

115 Kiyota, T., Ingraham, K. L., Swan, R. J., Jacobsen, M. T., Andrews, S. J. & Ikezu, T. AAV serotype 2/1-mediated gene delivery of anti-inflammatory interleukin-10 enhances neurogenesis and cognitive function in APP+PS1 mice. Gene Ther 19, 724-733, doi:10.1038/gt.2011.126 (2012).

116 Buchlis, G., Podsakoff, G. M., Radu, A., Hawk, S. M., Flake, A. W., Mingozzi, F. et al. Factor IX expression in skeletal muscle of a severe hemophilia B patient 10 years after AAV-mediated gene transfer. Blood 119, 3038-3041, doi:10.1182/blood-2011-09-382317 (2012).

117 Dimant, H., Kalia, S. K., Kalia, L. V., Zhu, L. N., Kibuuka, L., Ebrahimi-Fakhari, D. et al. Direct detection of alpha synuclein oligomers in vivo. Acta Neuropathol Commun 1, 6, doi:10.1186/2051-5960-1-6 (2013).

118 Jackson, A. L. & Linsley, P. S. Recognizing and avoiding siRNA off-target effects for target identification and therapeutic application. Nat Rev Drug Discov 9, 57-67, doi:10.1038/nrd3010 (2010).

119 Wang, X., Wang, X., Varma, R. K., Beauchamp, L., Magdaleno, S. & Sendera, T. J. Selection of hyperfunctional siRNAs with improved potency and specificity. Nucleic Acids Res 37, e152, doi:10.1093/nar/gkp864 (2009).

120 Khan, A. A., Betel, D., Miller, M. L., Sander, C., Leslie, C. S. & Marks, D. S. Transfection of small RNAs globally perturbs gene regulation by endogenous microRNAs. Nat Biotechnol 27, 549-555, doi:10.1038/nbt.1543 (2009).

121 Maraschi, A., Ciammola, A., Folci, A., Sassone, F., Ronzitti, G., Cappelletti, G. et al. Parkin regulates kainate receptors by interacting with the GluK2 subunit. Nat Commun 5, 5182, doi:10.1038/ncomms6182 (2014).

122 Salvatore, M. F., Pruett, B. S., Dempsey, C. & Fields, V. Comprehensive profiling of dopamine regulation in substantia nigra and ventral tegmental area. J Vis Exp, doi:10.3791/4171 (2012).

123 Parkinson, G. M., Dayas, C. V. & Smith, D. W. Age-related gene expression changes in substantia nigra dopamine neurons of the rat. Mech Ageing Dev 149, 41-49, doi:10.1016/j.mad.2015.06.002 (2015).

124 Parkinson, G. M., Dayas, C. V. & Smith, D. W. Increased mitochondrial DNA deletions in substantia nigra dopamine neurons of the aged rat. Curr Aging Sci 7, 155-160 (2014).

125 Klein, R. L., Dayton, R. D., Leidenheimer, N. J., Jansen, K., Golde, T. E. & Zweig, R. M. Efficient neuronal gene transfer with AAV8 leads to neurotoxic levels of tau or green fluorescent proteins. Mol Ther 13, 517-527, doi:10.1016/j.ymthe.2005.10.008 (2006).