RNA Knockdown As a Potential Therapeutic Strategy in Parkinson's

Home , Gene knockdown, Hammerhead ribozyme

FP Manfredsson1,2, AS Lewin2,3 and RJ Mandel1,2 1Department of Neuroscience, McKnight Brain Institute, University of Florida College of Medicine, Gainesville, FL, USA; 2The Powell Gene Therapy Center, University of Florida College of Medicine, Gainesville, FL, USA and 3Department of Molecular Genetics and Microbiology, University of Florida College of Medicine, Gainesville, FL, USA

Parkinson’s disease is a prevalent progressive degenerative interference can theoretically be applied to Parkinson’s disease disorder of the elderly. There is a current need for novel since over-expression of various proteins is known to kill the therapeutic strategies because the standard levodopa phar- dopamine neurons of the substantia nigra in animal models macotherapy is only temporarily efﬁcacious. Recently, there and in familial forms of Parkinson’s disease. Potential RNA have been some high-proﬁle successful preclinical results interfering strategies and caveats are discussed in this review. obtained in animal models of neurological disorders using Gene Therapy (2006) 13, 517–524. doi:10.1038/sj.gt.3302669; small interfering RNAs delivered by viral vectors. RNA published online 3 November 2005

Keywords: adeno-associated virus; ribozyme; interfering RNA

Introduction eukaryotic kingdoms. The major enzymes involved are highly conserved amongst species, indicating that the Parkinson’s disease (PD) is one of the most common response is an important cellular mechanism. Work into neurological disorders, second only to Alzheimer’s dis- the world of naturally occurring siRNA has shown that ease, affecting approximately 2% of adults. Although the this response is not only involved in protecting the cause of the idiopathic form of the disease is yet to be genome from threats such as viruses,1 but is also established, recent advances in the study of certain familial mechanistically related to microRNA (miRNA) path- forms of the disease have shed some light on the overall ways that provide crucial regulation of temporal control process of neurodegeneration in PD. Still, much remains to of development. The understanding of the mechanisms be understood about the disease. The recent development involved in RNA interference has progressed dramati- of new techniques such as the use of short RNA molecules cally in the past several years. Briefly, long double- like small interfering RNAs (siRNA) have gained increas- stranded RNA are processed by an RNase III enzyme ing popularity as tools to knockdown certain cellular named dicer to shorter (21–23 nt) siRNA complexes. The mRNAs, and these tools have been valuable in elucidating actual gene-silencing is processed by a group of proteins molecular pathways and players involved in PD. Still, the collectively termed the RISC (RNA-induced silencing ultimate goal is to treat the disease, and the current complex), which serves to bring the siRNA to the vicinity understanding of the potential etiology of PD makes it an of a complimentary mRNA and provides the catalytic attractive target for a siRNA therapy. One common feature enzymes involved in the cleavage of the mRNA.2–4 amongst the various forms of the disease is protein miRNA works along a similar pathway, however, naturally accumulation and aberrant protein clearance. Hence, a occurring miRNA precursors are short (70 bp) non-coding reduction in translation of specific proteins may suffice to hairpin sequences that are also processed by dicer into relieve the cellular stresses. In addition, a number of cell short stretches of single-stranded RNA. Unlike siRNAs, death signaling events likely to be involved in all forms of miRNAs contain short mismatches in the sequences PD are also potential siRNA targets. complementary to their target mRNAs, and miRNAs are thought to inhibit translation rather than lead to the degradation of mRNA. miRNAs are extremely important RNA interference in gene regulation, and certain neuronal disorders such as fragile X are linked to the miRNA pathway.5 RNAi is a response to double-stranded RNA that has The siRNA system has been commandeered to create been identified in numerous organisms in all three a valuable tool in order to study the effects of knocking down certain transcripts. Experimentally, one scans a Correspondence: Dr RJ Mandel, Department of Neuroscience, target gene for suitable target sequences 21–23 nt long McKnight Brain Institute, University of Florida College of Medicine, and designs double-stranded RNAs to target these sites. PO Box 100244, Gainesville, FL 32610-0024, USA. Scanning is typically assisted by one of several public- E-mail: [email protected]fl.edu Received 21 June 2005; revised 16 September 2005; accepted 19 access computer programs that incorporate experimental September 2005; published online 3 November 2005 ‘rules’ for designing effective siRNAs. These rules RNA knockdown as a potential therapeutic strategy in Parkinson’s disease FP Manfredsson et al 518 include avoiding runs of four or more guanosines or an alternative to siRNA lipid complexes, electroporation adenosines, keeping the base composition between 30 of ‘naked’ RNA has been shown to be effective in and 55% G+C, and identifying sites with asymmetric delivering siRNA to a limited number of target organs, thermal stability so that the antisense strand is preferen- including the visual cortex and the CA1 region of the tially incorporated into RISC.6–9 As an alternative for hippocampus.20 As with lipid complexes and RNA- rational design of siRNAs, several groups have devel- nanoparticles, however, electroporation leads to only a oped strategies for creating and screening siRNA transient reduction of gene expression, and the prospect libraries targeted to a gene of interest.10–13 This approach of sequential re-administration to the brain is unattrac- is valid because some of the most effective siRNAs that tive in a therapeutic setting. have been described in the recent literature would have Viral vectors provide the most suitable means of been avoided according to the current design algorithms. delivering and expressing siRNA for long-term therapy. Several recombinant viruses have been used for gene Delivery of siRNA therapy in the brain, but the short duration of transduc- Short interfering RNAs can be introduced to mammalian tion and the proinflammatory properties of vectors based cells and experimental animals either as RNA or as DNA on adenovirus and herpes simplex virus have restricted clones that code for the double-stranded RNA. When their use to antitumor applications. Lentivirus vectors DNA clones are used, they are often designed to produce and recombinant adeno-associated virus (rAAV) vectors short hairpin RNA (shRNA) that is processed to form have been used for long-term transduction of the brain siRNA in the cell. For cultured cells and for temporary and particularly for gene therapy in animal models of dosing to tissue, direct delivery of siRNA has distinct neurodegenerative disorders such as Huntington’s dis- advantages: short RNA molecules are commercially ease (HD),21,22 spinocerebellar ataxia23 and PD.24,25 Both available in pure form. Double-stranded RNA can be virus types can infect nondividing cells and lead to long- stabilized by chemical modifications in both strands of term gene expression. Lentiviral vectors give a more the duplex.14–16 Certain ribose modifications in the sense localized response and can contain a larger ‘payload’ in strand have the added advantage of increasing the terms of passenger genes, but delivery of siRNA does not incorporation of the antisense strand of the duplex into exceed the 4.7 kb carrying capacity of AAV. While rAAV RISC, therefore increasing cleavage of the intended does not provoke an inflammatory response, infection mRNA. Using RNA also permits accurate dosing of the does stimulate humoral immunity, and most people have cells, since a known quantity of RNA is added. The circulating antibodies to AAV.26,27 Indeed, pre-immuni- disadvantage of direct RNA treatment is that a large zation of rats with AAV2 prevents striatal transduc- amount of siRNA is required (typically 10–100 nM)and tion with rAAV2 vectors.28 Several rAAV serotypes the effect is temporary (3–4 cell generations). Using a high have been used to deliver marker genes or therapeutic dose of siRNA may saturate the nuclear-cytoplasmic genes to the midbrain, but to date, rAAV1 and rAAV5 transport of miRNAs and thus lead to non-specific effects. appear to provide the widest area of transduction, while Plasmids and viral vectors can be used to deliver gene expression mediated by rAAV2 infection is more shRNAs, which are usually produced under the control precisely located near the site of injection.29 siRNA of RNA polymerase III (pol III) promoters. Pol III therapeutics will almost certainly need to specifically promoters generate large amounts of transcript, but target the SN. rAAV2 has the serendipitous feature that it much of it is sequestered in the nucleus away from the transduces the nigral DA neurons with higher efficiency cytoplasmic RISC. Consequently, some investigators have than other neurons in the anatomical vicinity providing switched to pol II promoter systems for the expression of specific nigrostriatal transduction.29–31 shRNA17 or have disguised their siRNA hairpins as RNA interference holds a valuable potential as a miRNA precursors to facilitate transport to the cyto- future therapeutic, it could be used to simply knock- plasm.18 Delivery with viral vectors has the advantages down a specific disease gene. siRNA is selective for fully that hard-to-transfect primary cells can often be infected complementary mRNA targets, and short mismatches and that expression of the RNA hairpin may remain stable has been shown to significantly lower the efficacy of in nondividing cells or for many generations in dividing target mRNA degradation.32 Nevertheless, as few as 11 cells, if an integrating virus is employed. sequential matches between the 30 end of the target RNA While the delivery options are plentiful in the and the 50 end of the antisense strand are sufficient to laboratory, bringing siRNA to the clinic reduces these mediate cleavage by RISC, so that allele specificity is options. Obvious problems arise from RNA stability as thought to require three or more mismatches with well as access to target organs. Nonviral delivery nontarget RNAs. Allele specificity would prove valuable methods to the mammalian brain have included both as it would enable treatment of dominant mutations by non-specific lipid and polyethylenimine RNA complexes targeting only the mutant sequence. If specificity could and cell-specific methods. For example, receptor-specific not be achieved by targeting the degradation of an RNA pegylated immunoliposomes (PIL) have proved success- bearing a point mutation, RNA replacement might be ful when injected directly into tissue. PILs are ‘nano- achieved by concomitantly expressing a ‘hardened’ containers’ with the ability to store plasmid DNA and are mRNA, in which several silent nucleotide substitutions heavily coated with 2000-Da polyethyleneglycol (PEG). have been introduced. Thus, this approach essentially Tethered to a fraction of the PEG are antibodies specific replaces the defective gene with a healthy one. for those cell types to be targeted. For instance, in a Recent work on an animal model for spinocerebellar mouse model for intracranial human brain cancer, ataxia type 1, a polyglutamine expansion disease plasmid-encoded siRNA-targeting human epidermal characterized by progressive neurodegeneration, demon- growth factor receptor was injected with this vehicle, strates the potential for pre-translational silencing of significantly increasing the survival of the animal.19 As a gene.23 Polyglutamine diseases are characterized by

Gene Therapy RNA knockdown as a potential therapeutic strategy in Parkinson’s disease FP Manfredsson et al 519 the presence of intracellular inclusions containing the Nonetheless, there is no protective treatment or a cure expanded protein as well as numerous components of for PD. the machinery involved in ubiquitin-mediated proteaso- Neuropathologically, the disease is defined as a mal processing. It has been shown in vitro that lengthier progressive depletion of striatal dopamine (DA) with repeats cannot be processed through ubiquitin-mediated the concomitant loss of the dopaminergic neurons of the proteasomal degradation. These uncleaved stretches of substantia nigra pars compacta (SN). Similar to the amino acids are thought to further inhibit the proteasome polyglutamine repeat disorders mentioned previously, and are also very prone to aggregation which may be a hallmark feature of PD is the presence of intracyto- toxic to the cell.33 However, when Davidson et al.23 plasmic inclusions, termed Lewy Bodies, which are highly injected a rAAVexpressing a siRNA targeting the ataxin- immunoreactive to components of the proteasomal 1 mRNA into the cerebellum, they demonstrated that machinery such as ubiquitin and proteasomal subunits.43 transduced cells had a lower level of inclusions and that The cause of the idiopathic form of the disease is still the animals displayed a significant improvement in unknown and is probably multifactorial. Numerous motor performance. Similar positive results were ob- hypotheses have been put forth claiming that environ- tained delivering shRNAs to huntingtin in polygluta- mental factors such as pesticides or metals may be mine-expanded HD mice.21,22 exogenous factors in the disease.44,45 However, it appears Similar to spino-cerebellar ataxia, several other likely that subsequent steps in the disease process neurodegenerative diseases are also characterized by include mitochondrial impairment and increased oxida- the accumulation and aggregation of protein. In certain tive stress.46 At the point in time when motor functions familial forms of Amyotrophic Lateral Sclerosis, a muta- are impaired, approximately 80% of striatal DA is tion in the gene encoding the antioxidant enzyme Cu,Zn already gone, and the actual mechanism leading to cell superoxide dismutase, is thought to cause protein death in the idiopathic form of the disease is still misfolding, oligomerization and/or proteasome inhibi- unknown. Therefore, the need for a more complete tion.34,35 siRNA treatment in transgenic animals carrying understanding of the disease progression is obvious. these mutations targeting the proapoptotic gene prostate Additional clues to the disease etiology have come with apoptosis response-4, a protein shown to be upregulated the discovery of familial forms of PD and the genes in the lumbar spinal cord samples of ALS patients, involved in these forms. protected from caspase 3-mediated apoptosis and pre- served mitochondrial function in a staurosporine regi- 36 men. RNAi treatment has shown some promise in a-synuclein transmissible spongiform encephalopathies where the transmittable conformational change of an aberrant The first gene to be linked to PD was PARK1 that was isoform of the prion protein causes neuronal cell death. identified in 1997 by Polymeropoulos et al.47 by studying When transfected into a scrapie-infected neuroblastoma several families in southern Europe. The mutation cell line, siRNA targeting the prion mRNA reduced the identified in the a-synuclein gene was a single point levels of the protease-resistant isoform of the protein.37 mutation (A53T), and subsequently, several other In PD, several familial forms have been identified and mutations were also identified. In addition, a gene tripli- linked to disease genes, several of which are thought to cation was discovered in a US family,48 suggesting be inhibitory to proteasomal processing, or prone to that overexpression of the gene is detrimental as well. accumulation and aggregation. It is also likely that this a-synuclein-familial PD differs from the sporadic disease common theme of protein accumulation and aberrant in that it is generally earlier in onset, and has a greater protein degradation also share common downstream component of dementia. The parkinsonism is often effectors, such as common cell death triggers which also considered a subset in a more general disorder termed become attractive targets in a siRNA-based therapy diffuse Lewy body disease. The function of a-synuclein is (Figure 2). still unknown, although there are many indications that it is involved in the maintenance of synaptic DA vesicles and pre-synaptic function. Parkinson’s disease a-synuclein is considered a natively unfolded protein with a high propensity to aggregate, with the end The average age of onset of the idiopathic form of the product being a highly insoluble polymer termed fibrils. disease is at the age of 60 years, and the disease It is still unclear which is the toxic event involved in symptoms usually appear on one side of the body with a-synuclein-associated disease, but what all mutations a subsequent bilateral progression. Noticeable motor have in common is that they enhance the formation of an dysfunctions include a highly characteristic resting oligomer intermediate (protofibril) to fibril formation, tremor, rigidity, bradykinesia, loss of postural reflexes and this protofibril is thought to possibly have the ability and a distinguishable freezing phenomenon.38 Diagnosis to damage cellular membranes.49 cannot be complete without establishing a responsive- a-synuclein is a major component of LBs, the presence ness to levodopa, a neurotransmitter replacement strat- of misfolded a-synuclein and proteasomal machinery led egy, and currently the standard pharmacotherapy for to the idea that LBs are, in fact, neuroprotective, a way PD.39 This treatment is limited, however, by the for the cell to harness damaging misfolded, insoluble appearance of side-effects including reduced duration proteins into a contained ‘compartment’. of therapeutic response to a single dose and peak-dose The a-synuclein deficient mouse is relatively absent of dyskinesias 3–5 years after the commencement of pathology and also experiences an attenuated response therapy.40 There are numerous experimental treatments to certain neurotoxic stimuli such as a 1-methyl-4- that are currently undergoing clinical evaluation.24,41,42 phenyl-1,2,3,6-tetrahydropyridine (MPTP) regimen, a

Gene Therapy RNA knockdown as a potential therapeutic strategy in Parkinson’s disease FP Manfredsson et al 520 common model toxin for nigro-striatal degeneration. The cells would lead to cell division, but in post-mitotic MPTP metabolite 1-methyl-4-phenylpyridinium (MPP+) neurons increased cyclin activity is believed to trigger is inhibitory to mitochondria leading to decrease in apoptosis.57 Cyclins have been shown to accumulate in respiration and increase in the production of oxidative response to proapoptotic stimuli such the excitotoxin, free radicals, subsequently leading to cell death through kainate. Furthermore, cyclin E has been shown to apoptosis or necrosis.50 Considering the tendency for accumulate in parkin-deficient cells and tissue, and more a-synuclein to aggregate the animal model data, and the importantly, in the brains of AR-PD patients. Parkin toxicity involved in overexpression, targeting this gene overexpression in cerebellar granule cells was shown to with siRNA could hold therapeutic potential. relieve toxicity and apoptosis induced by kainite. However, siRNA-mediated parkin reduction was shown to sensitize the cells to, and enhanced the effects of, 54 Parkin kainite toxicity. The PARK2 gene was identified in 1998 to be involved in an early onset form of PD; autosomal recessive juvenile DJ-1 and PINK-1 PD (AR-JP).51 A distinguishable feature of this form of PD is that onset happens as early as in the teens. However, Additional early onset familial forms of PD were following the identification of this gene, several research- discovered recently with the identification of two disease ers identified PARK2 polymorphisms as a potential risk- genes involved in mitochondrial function. The first one factor in sporadic PD.52 Interestingly, patients with this to be identified was the gene encoding DJ-1.58 The form of the disease lack LBs, suggesting that parkin is function of this gene remains unknown, and numerous required for their formation/maintenance. The gene hypotheses have been put forth. Probably, the most product parkin is an E3 ligase involved in targeting certain interesting proposed function in terms of PD pathology substrate proteins for degradation by the proteasome.53 is the suggested role of DJ-1 in improved protection Briefly, ubiquitin is activated through an ATP-dependent against oxidative stresses, perhaps acting as a cellular process by an E1 ubiquitin-activating enzyme. The sensor for oxidative stress and modulator of gene activated ubiquitin is then transferred to an E2 ubiquitin- expression.49 This was demonstrated in vitro when a conjugating enzyme. E2 enzymes act in combination with siRNA targeting DJ-1 was expressed in the SH-SY5Y E3 accessory subunits that bind to specific degradation neuroblastoma cell line. These cells became more signals in protein substrates. susceptible to the MPP+ and 6-OHDA toxins, similar to Parkin has two RING (really interesting new gene) cells overexpressing the disease mutation.59 domains in the carboxy-terminus of the protein which The second new familial PD gene identified was serves as a binding region for the E2. There is a unique PINK-1, a mitochondrial kinase, which was shown to parkin domain adjacent to the ring domains which is protect cells against proteolytic stresses, although it is thought to serve as a substrate recognition region. Thus, unsure by which mechanism this protection occurs.60 parkin serves to facilitate the transfer of activated ubiquitin from the E2 to the substrate creating a substrate ‘tagged’ with a poly-ubiquitin chain, thereby targeting it UCHL1 for degradation in the 26S proteasome. The substrates that have been identified for parkin are CDC-rel 1 and An additional gene that has been linked to PD is PARK5 CDC-rel 2 (septin family, associated with synaptic coding for ubiquitin C-terminal hydroxylase L1 vesicles), cyclin E, Pael-Receptor (G protein-coupled (UCHL1).61 As the name suggests, this protein is also receptor), p38 tRNA synthase, synaptotagmin XI, syn- involved in ubiquitin-dependent protein degradation by phillin-1, as well as parkin itself.49 The cyclin E substrate acting as a deubiquitination enzyme removing ubiquitin is unique in that parkin is part of a multiprotein complex from processed amino-acid fragments, and, as such, fits which includes Hsel-10 and Cullin-1.54 siRNA was used with the general idea of impaired protein degradation. to evaluate the toxicity of some of these substrates. In However, only a single family has ever been identified Drosophila, Pael-R overexpression causes a selective with this gene allele, and more work needs to be done to degeneration of dopaminergic neurons, a condition that assess its relevance to PD. can be repressed through coexpression of a fly homo- The genetic linkages that have been made in PD logue of parkin (dparkin). However, when expressing a provide valuable insight and clues to the pathology siRNA designed against dparkin, the Pael-R mediated involved in familial forms as well as in sporadic disease; degradation is accelerated, indicating that the endo- after identifying the a-synuclein form of the disease the genous dparkin is involved in the degradation of field soon realized that a-synuclein may also play a role Pael-R.55 The inverse experiment would obviously be a in idiopathic disease. The identification of parkin natural extension of this test; using siRNA to target mutations and the lack of LBs hints to the idea that Pael-R in a parkin-deficient animal should inhibit cell proteasomal processing play an integral part of the death due to Pael-R accumulation. However, parkin molecular pathology, and the DJ-1 and PINK mutations knockout mice display an incomplete pathology with also involves the mitochondria and oxidative stress in little connection to the human disease, and no accumu- the disease. Although these are distinct molecular lation of substrates.56 processes, they are likely to be interconnected and not Similarly, Staropoli and co-workers used a siRNA necessarily exclusive. Given these ideas it is also likely against parkin to study the effects on cyclin E accumula- that the end result of these various forms of the disease tion. Cyclin E is involved in the regulation of G1/S converge on a common path of cell death. It is inherently cell-cycle transition, and when upregulated in mitotic difficult to study cell death in the human PD brain, given

Gene Therapy RNA knockdown as a potential therapeutic strategy in Parkinson’s disease FP Manfredsson et al 521 the slow rate of cellular death occurring over an may be a consequence of mitochondrial failure and extended period of time (decades). oxidative stress. Further support for this idea comes from Even so, there are numerous reports of TUNEL- reports that show that dopaminergic neurons from PD positive nigral neurons from patients with PD, suggest- brains show an upregulation of proapoptotic genes, and ing that cell death is mediated through apoptosis, which the fraction of neurons containing activated caspase 3 is

Figure 1 Ribozyme against parkin delivered via rAAV2 in substantia nigra with GFP coexpression This figure shows a representative animal that received unilateral nigral injections of rAAV-rbz131-IRES-GFP 4 weeks earlier (a–d). (a) Direct GFP fluorescence as a result of AAV transduction. (b) Tyrosine hydroxylase (TH) was labeled with Cy5 (shown in false color blue here). (c) Parkin is shown in red (Cy3). (d) Merger of a–c.In(a–d), yellow arrows point to GFP+ cells that are also TH+ but do not show parkin expression. White arrows show GFP+ cells that are also TH+ but do show parkin expression. In raw images, parkin expression in the injected hemisphere was barely detectable. On the uninjected side (e–g), all TH+ neurons had robust parkin expression. This figure shows that there is a variable amount of parkin knockdown in transduced neurons. We interpret this to mean that different nigral dopamine neurons received differing copy numbers of the ribozyme cDNA. This is a feature of the use of rAAV that cannot be controlled. Also, knockdown of parkin apparently caused no loss of TH positive neurons on the injected side. (h) In situ hybridization from an oligo-probe against rat parkin from another animal in this experiment, showing little parkin knockdown. (i) Similar in situ hybridization using a different oligo-probe in the same animal showing almost complete knockdown. We interpret these data to demonstrate that different parkin isoforms were differentially knocked down, and this should be considered for each gene in future siRNA experiments. Scale bar in (c)is50mm and applies to a–g. Scale bar in (h) is 1 mm and applies to (i). (j–m) distribution of a-synuclein (j, k) and parkin (l, m) in normal rat SN. (j) Low-power photomicrograph of a-synuclein staining in normal rat SN. There is relative overexpression of a-synuclein in the substantia nigra. The white triangle shows the area of enlargement shown in (k) and also shows that there is intense fiber staining in the striatonigral fibers. Scale bar ¼ 1 mm and applies to (l). (k) Magnification of the nigral neurons expressing a-synuclein. The white triangle is pointing to striatonigral fibers that also have relative overexpression of a-synuclein. Scale bar ¼ 100 mm and applies to (m). (l) Low-power photomicrograh of parkin staining demonstrating more ubiquitous and homogenous parkin staining in the SN. The white triangle shows the area of enlargement shown in (m). (m) Magnification of parkin staining. Note that some cells resembling microglia show some specific staining. (j–m) are intended to illustrate the problem when using siRNA strategies for specific cell populations like nigral DA neurons, Western blots or RT-PCR for a-synuclein will be relatively informative after taking punches from SN, while these same methods will be too diluted by other cell phenotypes when dealing with parkin knockdown.

Gene Therapy RNA knockdown as a potential therapeutic strategy in Parkinson’s disease FP Manfredsson et al 522 significantly higher than in healthy individuals.62 This RNA interference has been and will continue to play increase in apoptosis immediately raises the possibility an important role in PD research. However, the ultimate to target proapoptotic markers and activators such as goal of researchers is to bring treatments and cures to the caspase 3 using RNA knockdown techniques. When clinics. Given the relative ease by which one can design a expressed in hippocampal cultures, ribozymes designed siRNA to target a certain gene, there is no need for against pro-caspase 3 successfully protected against lengthy pharmacological analyses, and drug targeting staurosporine-induced apoptosis.63 Similarly, in an ani- issues. One can simply target the specific circuits in the mal model for apoptosis; rat optic nerve transection, brain using viral vectors expressing an siRNA. The pretreatment of siRNAs targeting c-jun and apoptotic potential targets in PD are numerous. First of all, one protease-activating factor-1 (Apaf-1) resulted in a signi- could target a-synuclein, which seems to be intimately ficantly improved survival of retinal ganglion cells.64 tied to the disease. The substrates of parkin are also Blockade of apoptotic enzymes has been shown to be attractive targets. Blocking the production of various beneficial in other model systems as well. For example, apoptotic activators also become potential treatment siRNA designed against FAS-ligand expressed in a basal strategies, but this approach also raises some concern cell carcinoma, was shown to reduce the level of about the risk for cancer if the siRNA is expressed in glia. apoptosis of infiltrating immune cells and led to a more For example, can a delivery system be refined enough, efficient clearing of the tumor cells.65 to target only dopaminergic cells? What happens if an Animal models of PD that completely recapitulate antiapoptotic siRNA is expressed in the surrounding the human disease are not available. Toxins such as glia? What if there is a gain of function mutation of an MPTP, 6-hydroxydopamine, and rotenone have been otherwise required protein? Lessons learned from used to create an acute level of oxidative stress in the SN, ribozyme studies demonstrate that one can target the causing DA neurons to die. Models such as these have mutated sequence specifically, but if there is extensive mainly been used in cell-replacement/survival studies. allelic heterogeneity among disease mutations, this Other models include a-synuclein transgenic animals, approach becomes less feasible. In the event that the which display a relatively wide array of pathology, but mutant allele cannot be targeted, one can target a not resembling that of PD. In addition, overexpression of common wild-type sequence, and attempt to replace wt-a synuclein or A53T using nigral-specific rAAV the gene with a siRNA or ribozyme-resistant copy. vectors induces over 50% nigral cell death and approximately an equal amount of striatal DA depletion in rats and monkeys.66,67 Parkin knockout mice have been created, but again they do not mimic the real disease. As a matter of fact, parkin-deficient animals display no significant pathology at all, which is surprising for a protein that is so ubiquitously expressed throughout the body. It is possible that utilizing siRNA to regulate certain genes may actually provide us with better models to study PD. For instance, in the case of parkin, this would allow one to study cellular impairment with spatio-temporal precision, and also allow one to avoid any ‘genomic compensation’ that may have occurred throughout the development of an organism. Previous studies in our laboratory tried to test this hypothesis. We designed a number of ribozymes against parkin. Ribozymes are enzymatic RNAs that detect certain sequences in the mRNA and cleave them. These ribozymes were then packaged into rAAV and unilat- erally injected into the SNc of adult rats. Immunohisto- chemistry, 4 weeks following the injections, clearly shows transduced cells (as detected by green fluorescent Figure 2 Schematic of potential etiological factors of PD and protein), with a significant loss of parkin expression potential siRNA targets (red symbols). Accumulated a-synuclein is (Figure 1a–g). However, attempts to identify accumula- a common feature in PD, and as such it serves as a potential target tion of substrates and quantifying levels of parkin for pre-translational silencing (1). In parkin-associated disease, a expression were unsuccessful. This was the result number of substrates are thought to accumulate, some may be of an inherent difficulty with RNA and protein levels detrimental due to function (e.g. cyclin E) (2), or simply due specifically in the SN. Proteins such as parkin are to accumulation (3). The newly identified genes (DJ-1 and PINK) may be involved in protection against oxidative stresses, which ubiquitously expressed in the brain, but the rAAV may be induced by endogenous factors such as by-products of transduction pattern that we experience is highly specific DA metabolism, or exogenous factors such as environmental toxins for the SNc; thus it becomes very difficult to isolate impairing mitochondrial function. Increases in oxidative stress also these nigral neurons, and to analyze their content acts as to create a general increase in misfolded proteins, thus without an overwhelming background from surround- potentially triggering apoptosis directly through an unfolded ing tissue. Whereas studying other gene products such protein response (3, 4), or indirectly through mitochondrial signaling events (4). Finally, since apoptosis is suspected as the as a-synuclein, which is highly upregulated in the final common pathway in DA neuron cell death, blockade of SNc as compared to surrounding cells, is less difficult proapoptotic proteins in neurons late in the apoptotic cascade may (Figure 1j–m). be therapeutic (5).

Gene Therapy RNA knockdown as a potential therapeutic strategy in Parkinson’s disease FP Manfredsson et al 523 Another concern is the efficacy with which you can 7 Schwarz DS, Hutvagner G, Du T, Xu Z, Aronin N, Zamore PD. knockdown a certain mRNA. Even in the most favorable Asymmetry in the assembly of the RNAi enzyme complex. Cell circumstances 100% reduction is not feasible. Thus, 2003; 115: 199–208. experiments must be designed to determine if residual 8 Khvorova A, Reynolds A, Jayasena SD. Functional siRNAs and expression of the gene is still therapeutic. miRNAs exhibit strand bias. Cell 2003; 115: 209–216. For instance, one of the earliest successful in vivo 9 Dorsett Y, Tuschl T. siRNAs: applications in functional genomics siRNA experiments targeted tyrosine hydroxylase (TH) and potential as therapeutics. Nat Rev Drug Discov 2004; 3: expression in the SNc of mice using rAAV-mediated 318–329. expression. TH is the rate limiting enzyme in the 10 Akashi H, Matsumoto S, Taira K. Gene discovery by ribozyme production of DA, the main signaling molecule in this and siRNA libraries. Nat Rev Mol Cell Biol 2005; 6: 413–422. circuit. Although, the experiment showed a significant 11 Luo B, Heard AD, Lodish HF. Small interfering RNA production Proc Natl Acad Sci reduction in TH levels, the behavioral testing was not by enzymatic engineering of DNA (SPEED). USA 2004; 101: 5494–5499. entirely consistent with complete DA depletion in the 68 12 Sen G, Wehrman TS, Myers JW, Blau HM. Restriction enzyme- nigrostriatal tract. These results emphasize the point generated siRNA (REGS) vectors and libraries. Nat Genet 2004; that in some instances siRNA may not be enough to 36: 183–189. achieve the intended biochemical result. An additional 13 Seyhan AA, Vlassov AV, Ilves H, Egry L, Kaspar RL, Kazakov concern comes with that of ‘off-targeting’. Although SA et al. Complete, gene-specific siRNA libraries: production great care is practiced when designing the siRNA, you and expression in mammalian cells. RNA 2005; 11: 837–846. may still be affecting other genes, effects that may or may 14 Czauderna F, Fechtner M, Dames S, Aygun H, Klippel A, Pronk not have detrimental results. GJ et al. Structural variations and stabilising modifications of synthetic siRNAs in mammalian cells. Nucleic Acids Res 2003; 31: 2705–2716. Conclusion 15 Hamada M, Ohtsuka T, Kawaida R, Koizumi M, Morita K, Although in its infancy, applied siRNA technology has Furukawa H et al. Effects on RNA interference in gene had great success in the laboratory, both to study expression (RNAi) in cultured mammalian cells of mismatches 0 pathology and to effect a cure. However, as of yet, only and the introduction of chemical modifications at the 3 -ends of one clinical trial is underway, targeting vascular Endo- siRNAs. Antisense Nucleic Acid Drug Dev 2002; 12: 301–309. thelial Growth Factor Receptor-1 in age-related macular 16 Prakash TP, Allerson CR, Dande P, Vickers TA, Sioufi N, Jarres R degeneration. In PD, there are numerous potential targets et al. Positional effect of chemical modifications on short for RNA knockdown that when reduced could slow or interference RNA activity in mammalian cells. J Med Chem 2005; 48: 4247–4253. even halt the progression of the disease (Figure 2). 17 Zhou H, Xia XG, Xu Z. An RNA polymerase II construct Given the recent high-profile successes of virally synthesizes short-hairpin RNA with a quantitative indicator delivered siRNA, this strategy is clearly poised to burst and mediates highly efficient RNAi. Nucleic Acids Res 2005; onto the PD research scene. All of the PD strategies to 33: e62. impact survival of DA neurons require nigral DA neuronal 18 Boden D, Pusch O, Silbermann R, Lee F, Tucker L, Ramratnam B. expression of the siRNA. Currently, the rAAV2 vector Enhanced gene silencing of HIV-1 specific siRNA using micro- system has shown the most specificity for nigral DA RNA designed hairpins. Nucleic Acids Res 2004; 32: 1154–1158. neurons. In this review, we have pointed out some caveats 19 Pardridge WM. Intravenous, non-viral RNAi gene therapy of when dealing with ubiquitously expressed proteins in the brain cancer. Expert Opin Biol Ther 2004; 4: 1103–1113. nigra (Figure 1). These potential problems in demonstrating 20 Akaneya Y, Jiang B, Tsumoto T. RNAi-induced gene silencing by knockdown must be taken into account in future studies. local electroporation in targeting brain region. J Neurophysiol 2005; 93: 594–602. 21 Harper SQ, Staber PD, He X, Eliason SL, Martins IH, Mao Q et al. Acknowledgements RNA interference improves motor and neuropathological This work was supported by a Fast-track grant from the abnormalities in a Huntington’s disease mouse model. Proc Natl Michael J Fox Foundation to RJM and ASL. Acad Sci USA 2005; 102: 5820–5825. 22 Rodriguez-Lebron E, Denovan-Wright EM, Nash K, Lewin AS, Mandel RJ. Intrastriatal rAAV-mediated delivery of anti-hun- References tingtin shRNAs induces partial reversal of disease progression in R6/1 Huntington’s disease transgenic mice. Mol Ther 2005; 12: 1 Ketting RF, Haverkamp TH, van Luenen HG, Plasterk RH. Mut- 618–633. 7ofC. elegans, required for transposon silencing and RNA 23 Xia H, Mao Q, Eliason SL, Harper SQ, Martins IH, Orr HT et al. interference, is a homolog of Werner syndrome helicase and RNAi suppresses polyglutamine-induced neurodegeneration in RNaseD. Cell 1999; 99: 133–141. a model of spinocerebellar ataxia. Nat Med 2004; 10: 816–820. 2 Meister G, Tuschl T. Mechanisms of gene silencing by double- 24 Bjorklund A, Kirik D, Rosenblad C, Georgievska B, Lundberg C, stranded RNA. Nature 2004; 431: 343–349. Mandel RJ. Towards a neuroprotective gene therapy for 3 Filipowicz W, Jaskiewicz L, Kolb FA, Pillai RS. Post-transcrip- Parkinson’s disease: use of adenovirus, AAV and lentivirus tional gene silencing by siRNAs and miRNAs. Curr Opin Struct vectors for gene transfer of GDNF to the nigrostriatal system in Biol 2005; 15: 331–341. the rat Parkinson model. Brain Res 2000; 886: 82–98. 4 Tomari Y, Zamore PD. Perspective: machines for RNAi. Genes 25 Mandel RJ, Burger C. Clinical trials in neurological disorders Dev 2005; 19: 517–529. using AAV vectors: promises and challenges. Curr Opin Mol Ther 5 Jin P, Alisch RS, Warren ST. RNA and microRNAs in fragile X 2004; 6: 482–490. mental retardation. Nat Cell Biol 2004; 6: 1048–1053. 26 Blacklow NR, Hoggan MD, Sereno MS, Brandt CD, Kim HW, 6 Reynolds A, Leake D, Boese Q, Scaringe S, Marshall WS, Parrott RH et al. A seroepidemiologic study of adenovirus- Khvorova A. Rational siRNA design for RNA interference. Nat associated virus infection in infants and children. Am J Epidemiol Biotechnol 2004; 22: 326–330. 1971; 94: 359–366.

Gene Therapy RNA knockdown as a potential therapeutic strategy in Parkinson’s disease FP Manfredsson et al 524 27 Erles K, Sebokova P, Schlehofer JR. Update on the prevalence of 49 Cookson MR. The Biochemistry of Parkinson’s Disease. Annu serum antibodies (IgG and IgM) to adeno-associated virus Rev Biochem 2004; 74: 29–52. (AAV). J Med Virol 1999; 59: 406–411. 50 Drolet RE, Behrouz B, Lookingland KJ, Goudreau JL. Mice 28 Peden CS, Burger C, Muzyczka N, Mandel RJ. Circulating anti- lacking alpha-synuclein have an attenuated loss of striatal wild-type adeno-associated virus type 2 (AAV2) antibodies dopamine following prolonged chronic MPTP administration. inhibit recombinant AAV2 (rAAV2)-mediated, but not rAAV5- Neurotoxicology 2004; 25: 761–769. mediated, gene transfer in the brain. J Virol 2004; 78: 6344–6359. 51 Kitada T, Asakawa S, Hattori N, Matsumine H, Yamamura Y, 29 Burger C, Gorbatyuk OS, Velardo MJ, Peden CS, Williams P, Minoshima S et al. Mutations in the parkin gene cause autosomal Zolotukhin S et al. Recombinant AAV viral vectors pseudotyped recessive juvenile parkinsonism. Nature 1998; 392: 605–608. with viral capsids from serotypes 1, 2, and 5 display differential 52 West A, Periquet M, Lincoln S, Lucking CB, Nicholl D, Bonifati V efficiency and cell tropism after delivery to different regions of et al. Complex relationship between Parkin mutations and the central nervous system. Mol Ther 2004; 10: 302–317. Parkinson disease. Am J Med Genet 2002; 114: 584–591. 30 Mandel RJ, Spratt SK, Snyder RO, Leff SE. Midbrain injection 53 Shimura H, Hattori N, Kubo S, Mizuno Y, Asakawa S, of recombinant adeno-associated virus encoding rat glial cell line- Minoshima S et al. Familial Parkinson disease gene derived neurotrophic factor protects nigral neurons in a progressive product, parkin, is a ubiquitin-protein ligase. Nat Genet 2000; 25: 6-hydroxydopamine-induced degeneration model of Parkinson’s 302–305. disease in rats. Proc Natl Acad Sci USA 1997; 94: 14083–14088. 54 Staropoli JF, McDermott C, Martinat C, Schulman B, Demireva E, 31 Klein RL, Meyer EM, Peel AL, Zolotukhin S, Meyers C, Abeliovich A. Parkin is a component of an SCF-like ubiquitin Muzyczka N et al. Neuron-specific transduction in the rat ligase complex and protects postmitotic neurons from kainate septohippocampal or nigrostriatal pathway by recombinant excitotoxicity. Neuron 2003; 37: 735–749. adeno-associated virus vectors. Exp Neurol 1998; 150: 183–194. 55 Yang Y, Nishimura I, Imai Y, Takahashi R, Lu B. Parkin 32 Miller VM, Xia H, Marrs GL, Gouvion CM, Lee G, Davidson BL suppresses dopaminergic neuron-selective neurotoxicity in- et al. Allele-specific silencing of dominant disease genes. Proc duced by Pael-R in Drosophila. Neuron 2003; 37: 911–924. Natl Acad Sci USA 2003; 100: 7195–7200. 56 Perez FA, Palmiter RD. Parkin-deficient mice are not a 33 Venkatraman P, Wetzel R, Tanaka M, Nukina N, Goldberg AL. robust model of parkinsonism. Proc Natl Acad Sci USA 2005; Eukaryotic proteasomes cannot digest polyglutamine sequences 102: 2174–2179. and release them during degradation of polyglutamine-contain- 57 Liu DX, Greene LA. Neuronal apoptosis at the G1/S cell cycle ing proteins. Mol Cell 2004; 14: 95–104. checkpoint. Cell Tissue Res 2001; 305: 217–228. 34 Bruijn LI, Houseweart MK, Kato S, Anderson KL, Anderson SD, 58 Bonifati V, Rizzu P, van Baren MJ, Schaap O, Breedveld GJ, Ohama E et al. Aggregation and motor neuron toxicity of an Krieger E et al. Mutations in the DJ-1 gene associated with ALS-linked SOD1 mutant independent from wild-type SOD1. autosomal recessive early-onset parkinsonism. Science 2003; 299: Science 1998; 281: 1851–1854. 256–259. 35 Kabashi E, Agar JN, Taylor DM, Minotti S, Durham HD. Focal 59 Taira T, Saito Y, Niki T, Iguchi-Ariga SM, Takahashi K, Ariga H. dysfunction of the proteasome: a pathogenic factor in a mouse model DJ-1 has a role in antioxidative stress to prevent cell death. of amyotrophic lateral sclerosis. J Neurochem 2004; 89: 1325–1335. EMBO Rep 2004; 5: 213–218. 36 Xie J, Awad KS, Guo Q. RNAi knockdown of Par-4 inhibits 60 Valente EM, Abou-Sleiman PM, Caputo V, Muqit MM, neurosynaptic degeneration in ALS-linked mice. J Neurochem Harvey K, Gispert S et al. Hereditary early-onset Parkinson’s 2005; 92: 59–71. disease caused by mutations in PINK1. Science 2004; 304: 37 Daude N, Marella M, Chabry J. Specific inhibition of patholo- 1158–1160. gical prion protein accumulation by small interfering RNAs. 61 Leroy E, Boyer R, Auburger G, Leube B, Ulm G, Mezey E et al. J Cell Sci 2003; 116: 2775–2779. The ubiquitin pathway in Parkinson’s disease. Nature 1998; 395: 38 Klockgether T. Parkinson’s disease: clinical aspects. Cell Tissue 451–452. Res 2004; 318: 115–120. 62 Alberi L, Sgado P, Simon HH. Engrailed genes are cell- 39 Nutt JG. Clinical pharmacology of levodopa-induced dyskinesia. autonomously required to prevent apoptosis in mesencephalic Ann Neurol 2000; 47: S160–S164; discussion S164–S166. dopaminergic neurons. Development 2004; 131: 3229–3236. 40 Nutt JG. Motor fluctuations and dyskinesia in Parkinson’s 63 Bantounas I, Glover CP, Kelly S, Iseki S, Phylactou LA, Uney JB. disease. Parkinsonism Relat Disord 2001; 8: 101–108. Assessing adenoviral hammerhead ribozyme and small hairpin 41 Dunnett SB, Bjorklund A, Lindvall O. Cell therapy in Parkinson’s RNA cassettes in neurons: inhibition of endogenous caspase-3 disease – stop or go? Nat Rev Neurosci 2001; 2: 365–369. activity and protection from apoptotic cell death. J Neurosci Res 42 Lozano AM, Mahant N. Deep brain stimulation surgery for 2005; 79: 661–669. Parkinson’s disease: mechanisms and consequences. Parkinson- 64 Lingor P, Koeberle P, Kugler S, Bahr M. Down-regulation of ism Relat Disord 2004; 10 (Suppl 1): S49–S57. apoptosis mediators by RNAi inhibits axotomy-induced retinal 43 Eriksen JL, Wszolek Z, Petrucelli L. Molecular pathogenesis of ganglion cell death in vivo. Brain 2005; 128: 550–558. Parkinson disease. Arch Neurol 2005; 62: 353–357. 65 Ji J, Wernli M, Mielgo A, Buechner SA, Erb P. Fas-ligand gene 44 Tanner CM, Langston JW. Do environmental toxins cause silencing in basal cell carcinoma tissue with small interfering Parkinson’s disease? A critical review. Neurology 1990; 40 RNA. Gene Therapy 2005; 12: 678–684. (Suppl): 17–30; discussion 30–31. 66 Kirik D, Rosenblad C, Burger C, Lundberg C, Johansen TE, 45 Tanner CM, Ottman R, Goldman SM, Ellenberg J, Chan P, Muzyczka N et al. Parkinson-like neurodegeneration induced by Mayeux R et al. Parkinson disease in twins: an etiologic study. targeted overexpression of alpha-synuclein in the nigrostriatal JAMA 1999; 281: 341–346. system. J Neurosci 2002; 22: 2780–2791. 46 Greenamyre JT, Hastings TG. Biomedicine. Parkinson’s – diver- 67 Kirik D, Annett LE, Burger C, Muzyczka N, Mandel RJ, gent causes, convergent mechanisms. Science 2004; 304: 1120–1122. Bjorklund A. Nigrostriatal alpha-synucleinopathy induced by 47 Polymeropoulos MH, Lavedan C, Leroy E, Ide SE, Dehejia A, viral vector-mediated overexpression of human alpha-synuclein: Dutra A et al. Mutation in the alpha-synuclein gene identified in a new primate model of Parkinson’s disease. Proc Natl Acad Sci families with Parkinson’s disease. Science 1997; 276: 2045–2047. USA 2003; 100: 2884–2889. 48 Singleton AB, Farrer M, Johnson J, Singleton A, Hague S, 68 Hommel JD, Sears RM, Georgescu D, Simmons DL, DiLeone RJ. Kachergus J et al. Alpha-Synuclein locus triplication causes Local gene knockdown in the brain using viral-mediated RNA Parkinson’s disease. Science 2003; 302: 841. interference. Nat Med 2003; 9: 1539–1544.

Gene Therapy