Repairing the Brain: Gene Therapy.

In vivo gene therapy for neurodegenerative disorders has turned out to be a formidable challenge. It is a field not much older than twenty years, but we were many who would have predicted a much easier path towards the clinic using this treatment modality. For Parkinson's disease patients, this has meant a frustrating wait, seeing many promising therapies being forgotten after a few pre-clinical proof-of-concept studies. The reasons for this are both scientific and economical. However, this is slowly but surely changing and over the next two decades we will see a very exciting development in this field. In a foreseeable future, gene therapy will be a very natural component of many clinical therapies, not least in Parkinson's disease.

THE RISE AND FALL OF THE FIRST GENERATION OF GENE THERAPY IN PARKINSON’S DISEASE

Since the dawn of in vivo gene transfer, the treatment of Parkinson’s disease (PD) has always been listed as one of its prime targets [1, 2]. Many attempts have been made, each one with its own twist. However, in retrospect, with very few exemptions, they can be clustered into two groups; neuroprotective/restorative therapies [3-5] and symptomatic therapies modulating the dopamine system/basal ganglia [6-10]. It should be noted that all implemented therapies have had symptomatic relief as the primary endpoint in Phase II and not neuroprotection even when that is theoretically possible. There are many good reasons for this, not least that the disease heterogeneity and slow progression would require clinical trials an order of magnitude larger than what is feasible with gene therapy to prove disease modification.

The scientific rational for the implemented approaches is outside the scope of this opinion piece but is very well covered in previous reviews [2, 11–13]. In the enthusiasm around early gene therapy, multiple companies were formed including Neurologix [7], Ceregene [4, 5], Oxford Biomedica [14], and Voyager Therapeutics [6], all with the focus on gene therapy in PD. Now a decade later, the results are less than encouraging. The two first companies failed to show convincing benefits over current therapeutic options to remain viable and have disappeared. Oxford Biomedica have had signs of efficacy intermingled with aggravated L-DOPA induced dyskinesias (LIDs) [14, 15] and have gone back to the drawing board with the aim to refine the vector [16]. The latest news is that this approach is licensed out to Axovant and thus its future remains unpredictable. Voyager therapeutics has been very successful in attracting funding and has expanded its portfolio significantly. However, the AADC pro-drug approach remains in their active pipeline and is planned to enter a Phase 2 double-blind placebo-controlled trial. It is with this trial we will see if there is adequate therapeutic potential of this approach to warrant further development. The published results from the phase 1 trials have been puzzling in that they show greater therapeutic efficacy OFF compared to ON L-DOPA [6, 17] which does not fit well with the current understanding of the proposed action of this therapy which is aimed to potentiate the conversion of peripherally delivered L-DOPA into DA.

The Ceregene sponsored Neurturin trials failed to meet the primary endpoint of symptomatic relief in Phase II. It has been broadly debated why this happened [12, 18–20], was the degeneration of the DA system too severe in the selected patient population? Was the dosage too low or the AAV serotype suboptimal? Is it so that we do not fully understand the mechanism of action of neurotrophic factors in PD and thus targeted a neuroprotective strategy more suitable for disease models of PD than the real thing? Most likely it is a combination of all factors. The ongoing AAV-mediated GDNF trial conducted at the NINDS [21] will hopefully shine some light on the feasibility of the approach.

Prediction

The current generation of clinical gene therapy will not be the prevailing therapy for PD in 20 years. They will be replaced by refined alternatives described below. It is possible that the heterogeneity of PD will mean that the cause will not be identified in every case and that preventive measures cannot be taken for some patients. Those patients will hopefully be served well by restorative measures described below. However, the foundation which these pioneering gene therapy trials in PD have provided with regards to safety data, viral vector production and clinical trial design should not be underestimated. It is on this foundation all approaches described below will stand.

THE MERGER OF CELL AND GENE THERAPY

In 20 years, the front-line pre-clinical research for PD will have moved beyond symptomatic relief and circuit reconstruction and will focus on curative prevention of disease. This will however not mean that symptomatic therapies and circuit repair will become obsolete. On the contrary, it is expected that such clinical therapies will peak in complexity and potential about that time. Cell repair will be covered in detail in elsewhere in this issue, but in essence it aims to reconstruct the dopamine neurons in the Parkinsonian brain, originally residing in the substantia nigra. Cell sources differ significantly between the approaches from fetal ventral mesencephalon to replenishable sources such as embryonic stem cells and induced pluripotent stem (IPS) cells.

If nothing exceptional has occurred between the time of writing of this review and publication, the first PD patient will have received an IPS-based DA neuroblast transplant in September 2018 [22]. This is the result of a national fast track approach in Japan and everyone in the field does hope that the outcome will be positive. Otherwise this trial could again put an unfortunate delay on the development of brain repair therapies [23].

Until the day when we can direct and regulate the circuit integration of de novo neurons, extrinsic regulation of their activity will be an attractive option [24-26]. The Japanese trial as well as those planned elsewhere using human embryonic stem cells do all pursue an ectopic graft placement, i.e., in the putamen instead of the substantia nigra pars compacta where the somata of the nigral neurons normally reside. This is currently a necessity as the post-synaptic dopamine stimulation is required in the putamen for the recovery of motor impairment in PD [27]. Transplanted neuroblasts do not yet show efficient axonal outgrowth from the substantia nigra (SN) to the putamen [28], at least not efficient enough to cover the distance in the human brain. Due to this fact, the ectopic DA transplant does not readily have access to the afferent regulation of the SN and is subject to the highly inhibitory environment of the putamen. We have found that this results in a state where the transplanted cells don’t show their full therapeutic potential [24]. Fortunately, this can be remedied using gene therapy. Using chemogenetic receptors expressed only in the transplanted DA neurons, we were able to independently introduce an excitatory and an inhibitory regulation of the transplant. We and others have found that the behavioural recovery can be significantly increased with the excitatory stimulation of the DA neurons while it is diminished if the DA neurons is silenced [24-26]. This provides a powerful bidirectional paradigm for clinical regulation of cell transplants, regardless of cell source, which will not only provide an attractive safety mechanism for handling graft-induced dyskinesias [29] but one which could also provide significant benefits for the patient.

Gene transfer will be a key component of clinical cell transplantation within the next 20 years. All transplanted neurons will contain genetic alterations in one way or the other. Chemogenetics will be one such alteration providing a significant potentiation of the transplanted neurons but will not be the only change. We will see genome editing repairing disease-causing mutation e.g., in LRRK2 enabling personalized approaches but we will also see genetically engineered cells evading an immune response and thus allowing broadly utilized cell sources.

TAILORED AAV CAPSIDS HOMING FROM THE PERIPHERY TO THE CNS

Gene therapy for PD has been pursued using almost all recombinant viral vehicles developed, including Retrovirus [30], Adenovirus [31], Herpex simplex virus [32], Lentivirus [14] and of course the Adeno-associated virus (AAV) [5-8]. The AAV vector has such distinct advantages over the other vectors with regards to efficacy and low immunogenicity in the CNS that it has become the de facto standard vehicle for CNS gene transfer despite its intrinsic limitation in the size of packaged genetic material [33]. One key property of the AAV which has been especially attractive for CNS delivery has been the broad diversity of naturally available AAV serotypes and their varied tropisms [34, 35]. Furthermore, the AAV capsid has turned out to be extremely malleable. Parts of the capsid can be exchanged between different serotypes to generate new variants with different functions [36] and random peptides can be displayed on the capsid surface to result in even bigger changes in vector function [37]. We can even utilize our knowledge of other virus types to engage in molecular mimicry where peptides from other viruses (or other proteins) are used to transfer desirable functions onto the clinically suitable AAV such as the retrograde infectivity of neurons from the terminal [38]. This development has been greatly facilitated by recent advances in parallel gene synthesis, deep sequencing and computational modelling [39] where novel capsid variants can now be systematically screened millions at the time also in humanized modelling systems or non-human primates. In the coming decade we will see an explosion in synthetic capsid variants with tailored properties for each individual therapy.

Gene therapy in PD will in 20 years be conducted solely using tailored synthetic vectors. Intraparenchymal injections will be replaced with intrathecal or systemic injections. Each therapy will depend on its own tailored vector with optimized function such as transport over the brood brain barrier or homing to the targeted neuronal population based on either connectivity, gene expression profile or disease state.

BEYOND BRAIN REPAIR: PREVENTION OF DISEASE THROUGH GENE REPAIR

In 20 years, we will have moved away from the focus on monogenetic familiar forms of PD as the focus for genome editing and repair and all emphasis will be on the idiopathic forms [40]. This is not because of financial or technical constraints but it will be because all monogenetic forms will have seen novel therapies become available in the form of gene repair.

Gene repair has gone through multiple steps towards reaching that point, but many remain before we have suitable and safe therapies. Gene disruption techniques such as the CRISPR/Cas9-mediated NHEJ [41-43] are filling our journals with very interesting proof-of-concept approaches for therapy based on gene removal [44, 45]. At the time of writing, the first rouge application of CRISPR/Cas9 applied to In vitro fertilization appears to have resulted in the first birth of a genome edited child in China [46]. In this case the target was the CCR5 receptor for very weak scientific reasons and with even weaker regulatory oversight. While NHEJ will no doubt find suitable clinical applications over the next 10 years, the real revolution for in vivo genome editing comes when we can perform efficient and safe gene repair in situ.

This is a field that is seeing active development as well. In its simplest form, gene replacement can be conducted without the use of genome editing. An example can be seen in a recent paper on the development of AAV-mediated therapy for autosomal dominant retinitis pigmentosa. With this approach, the disease-causing gene (mutant rhodopsin) is silence using short hairpin (sh) RNA interference [47]. In the same AAV vector an intact rhodopsin gene (with silent mutations allowing escape from RNA interference) is expressed from a short opsin promoter. Such approach has shown promise in canine models and could become suitable for clinical translation. This approach may also be suitable for gene replacement in PD. However, the challenge there will be the regulation of gene dosage. While the jury is still out on how detrimental it is for the adult dopamine neurons to have α-Synuclein totally removed (e.g., using shRNA) [48, 49], it is very clear that an increase of any form of α-Synuclein is not very desirable. Thus, going forward, gene dosage for PD gene repair will be a necessity. With α-Synuclein, multiple approaches are being studied to reduce α-Synuclein load either by targeted methylation of the locus [50] or exchanging regulatory sequences using knock-in [51]. Knock-in approaches are very attractive for gene repair and gene dosage. Such techniques will allow the chromatin to control tightly the gene and thus ensure perfectly regulated gene expression in every cell regardless of viral titer and local concentration gradients.

Knock-in approaches in the CNS mediated by genome editing have now started to see proof-of concept with the HITI technique [52, 53]. It is using the fact that AAV vectors are very attractive for template delivery and that a double-strand break in the genome opens up for insertion of genetic material [54]. However, the “passive” nature of this insertion together with the requirement of a double-strand break (which will also introduce NHEJ) makes this approach intrinsically ineffective and not very suitable for clinical translation, at least not in PD. It is also not yet shown to be capable of repairing a gene but has only been used to insert a coding sequencing flanking the endogenous allele.

Potent editing approaches also require potent means of regulation. Gene editing ex vivo and even in some peripheral organs in vivo can be achieved using transient expression of nucleases using either mRNA or protein delivery [55, 56]. This is not yet possible for the post mitotic cells of the CNS and thus we still have to depend on constitutive transgene delivery for the foreseeable future. However, attempts have been made to make the Cas9 protein Self-inactivating after viral delivery. One example is the KamiCas9 approach presented by Deglon and collaborators [45]. This approach ensures that the Cas9 is itself inactivated as soon as the cell starts to produce all components required for the NHEJ. Through the use of PolIII promoters of different strength, the authors could drive the equilibrium so that most cells edit the target gene before the Cas9 is disrupted. The first application of this approach was gene editing of a virally derived huntingtin gene in the mouse striatum.

Base editing approaches show much greater therapeutic potential but do have other challenges. In base editing, no double strand break is made and there is no requirement for a template DNA strand to be used for insertion and repair. Instead, a cytidine or adenine deaminase enzyme chemically alters a nucleotide on one of the DNA strands in situ [56, 57]. Through a DNA repair process, the other strand will be modified to adhere to the Watson-Crick base pairing. The deaminase will have to be accurately and efficiently brought to the correct nucleotide to allow for the editing to occur but also to restrict the editing of other bases in the genome [58]. This is a formidable challenge, but fusion proteins with the Cas9 protein and tailored sgRNA have shown initial promise [57, 59, 60]. Further restrictions of its applicability relate to the available enzymes only permitting A to G or C to U mutation and the available Cas9 proteins restricting the possible positions of Cas9 binding in the targeted genes. However, the ongoing and historically successful mutation efforts will very likely result in a battery of base editing tools [61] which will allow for the editing of any base at any position of the gene and this will happen well within the next 20 years.

The tool of the future for monogenetic familiar forms of PD will be a synthetically designed viral vector capsid with the capacity to transport the repair machinery from the blood into the CNS delivering the cargo throughout the body and the CNS. This will be performed in the early neonatal stage and will be seen as severe as a flu shot. The cargo of this viral vector will be a base editing enzyme which will accurately edit the erroneous base and will then self-delete when the mission is accomplished.

Path and milestones predicted for gene therapy in Parkinson’s disease over the next 20 years. Clinical gene therapy is currently aiming for symptomatic relief as its primary end point. This will gradually shift to brain repair when it is merged will cell therapy or in vivo re-programming for circuit repair. Over the next decade we hope for a significant increase in mechanistic insights into the idiopathic Parkinson’s disease which will then pave the way for gene therapy to move into prevention of disease. As a final step we will see these measures being refined and validated to warrant their applications in normal healthy aging. Through prevention and reversal of detrimental processes in the brain such as DNA damage, protein accumulation and oxidative stress we may see the first attempts towards brain rejuvenation using gene therapy before the year 2040. It should be noted that most patients will receive a combination of these therapies and the most successful therapies will most likely target two or more of the modalities, e.g., both symptomatic and restorative.

NEURONAL REJUVENATION

Aging is the principal risk factor for PD [62]. It is also a major contributor to most other neurodegenerative disorders. In addition, even for those of us fortunate enough to evade a clinical diagnosis, aging will with time endow us with subclinical manifestations of many of the hallmarks of PD such as rigidity, bradykinesia, postural instability and tremor. To add insult to injury we will also see our cognitive performance gradually decline. If we could choose, most of us would like to avoid these changes altogether and remain neurologically top performing until the inevitable end. In this endeavour gene therapy will play an important role. Neuronal aging and its prevention or reversal is still a largely unexplored field [63]. However, we can learn a lot from the “accelerated aging” seen in PD but also from the healthy aging brain [62]. The genome of our neurons is continuously attacked through our lifespan inducing random mutation throughout [64]. Most mutations are corrected in the young nerve cells using the base excision repair (BER) pathway but with time the uncorrected mutations accumulate and the BER pathway becomes less and less active. These changes are thought to contribute negatively to the aging, not only of the brain. Another cellular event which has attracted significant interest in the neuronal aging field is the Nicotinamide adenine dinucleotide (NAD+) biosynthesis [65]. In the pathophysiology of aging, NAD+ is both linked to the afore mentioned DNA repair through its requirement in the poly-ADP-ribose polymerase (PARP) function and in histone methylation through Sirtuins. Reduction in NAD+ levels are observed with age in most tissues and thus has detrimental effects on both DNA repair and Sirtuin function. Interestingly overexpression of the brain-specific SIRT 1 reverts many age-related changes, prolongs lifespan [66] and protects against neurodegeneration in AD [67]. Thus, the NAD+ biosynthesis, e.g., through the Namnpt gene is an interesting target for neuronal rejuvenation.

Neurons have to survive many challenges throughout a lifetime. Due to the lack of self-renewal, every dysfunctional protein or organelle may become a treat to survival of the neuron and gradually they accumulate [68]. Therefore, the self-eating or macroautophagy is a crucial housekeeping event, not least in dopamineric neurons [69, 70]. Macroautophagy is impaired with aging and with its decrease comes protein aggregation and neurodegeneration [70, 71]. Controlling macroautophagy may thus also be a potent component in neuronal rejuvenation. However, it is also so that more is not always better in this case [72] so much more knowledge is required.

In 20 years, gene therapy will have become so mainstream and safe that it will be used to also counteract the detrimental effects on the dopamine system of normal aging. Through potentiation of DNA repair, careful chromatin remodeling, stimulation of neurogenesis, plasticity and macroautophagy, we will be able to maintain a much better brain health throughout the lifespan. This will not only protect again clinically manifested neurodegenerative disorders such as PD and AD but will allow all of us remain neurologically young.

FINAL WORDS

Gene therapy will no doubt see many challenges in the coming decades. With the increasing complexity of the approaches comes significant safety and ethical concerns. It is my belief however that these hurdles are surmountable. Recent unauthorized shortcuts to the clinic with genome editing in China must not become the norm. We all need to decide on an ethical development route which protects and benefits the patients without stifling innovation. If this balancing act is executed successfully, the future for clinical gene therapy in PD will be very bright even within as short time frame as 20 years

CONFLICT OF INTEREST

The authors have no conflict of interest to report.