Gene Therapy for Amyotrophic Lateral Sclerosis: Therapeutic Transgenes

Introduction

As discussed in Chapter 4 , Molecular Mechanisms of Amyotrophic Lateral Sclerosis, amyotrophic lateral sclerosis (ALS) is a disease with a poorly understood etiology. There are a wide variety of abnormalities present, including mitochondrial dysfunction, glutamate excitotoxicity, oxidative stress, astrogliosis and microgliosis, loss of innervation of neuromuscular junctions (NMJs), and death of motor neurons (MNs). However, the relative contribution of these factors to disease initiation and progression is unclear at best. Therefore, a wide variety of approaches have been investigated to treat this disorder. This chapter will focus on the therapeutic transgenes that have been studied to date. These transgenes range from those that address a single abnormality, such as apoptosis, to those that are expected to alter several defects simultaneously. While this range of approaches has produced improvements in the ALS phenotype in animal models, none has become the silver bullet that will halt ALS in its tracks.

Antiapoptosis

One of the defining characteristics of ALS is the loss of upper and lower MNs. Since the death of these cells could be a direct cause of NMJ denervation, one obvious therapeutic approach would be to prevent cell death from occurring. Cells can be killed by a number of mechanisms ranging from a tidy programmed cell death, where the cell essentially breaks itself down to be phagocytized by nearby cells, to messy necrosis, where the cell spills its contents into the extracellular space, inducing inflammation. It is generally believed that MN death occurs via a process of programmed cell death known as apoptosis, or at the very least, something very similar to it.

Apoptosis was first described by Kerr et al. in 1972, and it involves a highly regulated series of steps ( Fig. 8.1 ). There are three main pathways in apoptosis: death receptor-mediated, mitochondrial-mediated, and endoplasmic reticulum-mediated. Based on cell culture models, animal models, and postmortem tissue from ALS patients, it appears that the mitochondrial-mediated pathway is likely the most relevant to MN death in this disease. In this pathway, the action of proteins like Bcl-2-associated X (Bax) and Bcl-2-homologous antagonist/killer (Bak) cause the mitochondria to release cytochrome c and other factors into the cytoplasm. This in turn activates a cascade of caspases, including caspases 3 and 9. This pathway is inhibited by a number of proteins, including X-linked inhibitors of apoptosis protein (XIAP), B-cell lymphoma 2 (Bcl-2), and B-cell lymphoma extra large (Bcl-X _L ).

During apoptosis, caspases break down the proteins of the cytoskeleton, causing cells to shrink and round. The nuclear chromatic condenses and endonucleases are activated, fragmenting the DNA. Finally, the plasma membrane blebs, breaking the cells into a host of small vesicles that are phagocytosed by macrophages. Importantly, in contrast to necrotic cell death, the contents of the cytoplasm are not spilled into the extracellular space. Thus, apoptosis does not activate inflammatory pathways through intracellular milieu release.

There are a number of signs that apoptosis occurs in both animal models of ALS and ALS patients. In mice expressing mutant SOD1 (SOD1 mice), the antiapoptotic Bcl-2 and Bcl-X _L proteins exhibit reduced expression, while the proapoptotic proteins Bcl-2-associated death protein (Bad) and Bax are upregulated during the symptomatic phase of the disease. Knocking out Bax in this model completely blocks MN death. Cytoplasmic cytochrome c levels increase in the spinal cords of SOD1 mice as they age, peaking when the mice reach the early symptomatic stage of the disease. Caspase 1 and caspase 3 mRNAs are also both upregulated, first in neurons and then in glial cells. Sequential activation of several downstream caspases is also observed.

In ALS patients, a similar pattern emerges. Bax and Bak are enriched in the mitochondrial membrane fraction, while Bcl-2 is decreased in this compartment. In addition, the binding of Bcl-2 to Bax, necessary for the suppression of Bax function, is also significantly reduced. Similar to the animal model, ALS patients also exhibit elevated activity of caspase 1 and caspase 9.

A number of studies have demonstrated that inhibiting apoptosis can slow or prevent some aspects of ALS. Inhibiting caspase activity is one potential approach. Overexpression of XIAP in neurons in a SOD1 mouse reduced caspase activity and extended life span. Intracerebroventricular infusion of N -benzyloxycarbonyl-Val-Ala-Asp-fluoromethylketone (zVAD-fmk), a broad-spectrum caspase inhibitor, significantly extended the life span of the SOD1 mouse when delivered by an osmotic pump beginning at 60 days of age. Disease onset was delayed by 20 days and survival was increased by 27 days. MN death was reduced by the greatest amount in the cervical spinal cord, possibly suggesting poor delivery of the drug to the lumbar cord. It should be noted that this therapy significantly reduced the production of mature interleukin (IL)-1β, a proinflammatory factor. Thus, part of the effect of this treatment could be due to dampening of inflammation.

Rather than targeting later steps in apoptosis, it may be more effective to target the initiating events. Overexpression of Bcl-2 in transgenic mice had previously been shown to block apoptosis due to a variety of factors, including the culling of excess neurons during normal development. This protein was also neuroprotective in other models of neuron death, including ischemia and axotomy.

To determine whether overexpression of Bcl-2 could impact ALS, transgenic mice expressing Bcl-2 from the neuron-specific enolase (NSE) promoter were crossed with SOD1-G93A mice to create double transgenics (Bcl-2;SOD1). When compared to SOD1 mice, Bcl-2;SOD1 mice exhibited a 33-day delay in disease onset, and life span was increased by 35 days. MN loss was also delayed, but end-stage animals in both groups had similar levels of MN death. The loss of NMJ innervation and axons in the phrenic nerve exhibited a similar pattern. Thus, Bcl-2 delayed, but could not prevent, MN loss. While Bcl-2 has a critical role in apoptosis, the authors suggested that the effects that they observed may not simply be due to a direct effect on apoptosis. The Bcl-2;SOD1 mice also showed smaller numbers of ubiquitin-positive MNs relative to age-matched SOD1 mice, something unexpected if Bcl-2 was only affecting apoptosis. The authors speculate that antioxidant properties of Bcl-2 could also be contributing to the delayed disease onset. It may be important to note that, while Bcl-2 can protect neurons from death, it may not prevent axon die back, a process that is critical in ALS.

Another member of the Bcl-2 family of proteins with antiapoptotic activity is Bcl-X _L . When tested in a tissue culture model of glutamate-induced cytotoxicity, overexpression of Bcl-X _L substantially reduced TUNEL positivity and increased cell viability in primary MNs. Ohta et al. studied a recombinant protein consisting of Bcl-X _L fused to the protein transduction domain (PTD) of the HIV TAT protein. The PTD is of interest because it can allow the coupled protein to cross cellular membranes, including those of the blood–brain barrier. Thus, one can use intravenous, intraperitoneal, or intrathecal delivery to get a recombinant protein into the cytoplasm of cells. When injected intraperitoneally in an ischemic brain injury model, this fusion protein (TAT-Bcl-X _L ) was able to reduce caspase-3 activation, substantially inhibit DNA fragmentation, and reduce neurological deficits. In SOD1 mice, TAT-Bcl-X _L was delivered by intrathecal infusion over a period of 28 days, beginning at 91 days of age. This therapy delayed disease onset by 10 days and increased life span by 13 days. Motor function showed improvement, and MN death was reduced relative to controls. Markers of apoptosis, including caspase 3 and caspase 9, activation and TUNEL positivity were also significantly reduced.

One other study should also be noted here. Kaspar et al. examined the effects of exercise in SOD1 mice. Ad libitum exercise increased life span by 24 days. Interestingly, exercise also elevated the expression of the antiapoptotic genes Bcl-2 and Bcl-X _L , suggesting that exercise could function, in part, by increasing MN resistance to apoptosis.

In the early stage of disease progression in SOD1 mice, activation of caspase 1, rather than caspase 8, leads to the cleavage of BH3-interacting domain death agonist (Bid), yielding truncated Bid (tBid), the active form of Bid that can initiate the mitochondrial apoptotic pathway. Thus, targeting caspase 1 is a promising target for an antiapoptotic therapy. Friedlander et al. created a transgenic mouse expressing a dominant negative form of caspase 1 by mutating an active site cysteine residue to glycine. Crossing this mouse with a SOD1 mouse substantially reduced the amount of cytochrome c release from mitochondria and delayed the activation of caspases 3 and 9. The dominant negative caspase 1 did not delay disease onset, but it did significantly increase survival by 21 days.

While there have been many successful studies using transgenes to inhibit apoptosis in animal models of ALS, only one study has evaluated a gene therapy approach. Azzouz et al. injected an adeno-associated virus (AAV) vector expressing Bcl-2 into the lumbar spinal cord of SOD1 mice. The therapy delayed spontaneous fibrillation potentials in the gastrocnemius muscle by approximately 10 days and increased MN survival in the injected region. However, no improvement in life span was detected.

The use of parenchymal injections might explain why the effect of this therapy is so muted. In contrast to the previous studies, Bcl-2 was expressed only in the region injected. Thus, disease progression in the rest of the spinal cord was free to proceed unchecked. In addition, even near the injection sites, it is unlikely that all of the target cells were transduced. A broader delivery of the transgene would likely increase its effectiveness. It would be interesting to see how Bcl-2 would perform using an AAV9 vector, which can provide robust expression throughout the central nervous system (CNS) when delivered intravenously.

Glutamate Signaling (Glutamate Transporter 1, ADAR2/AMPA)

Upper MNs, residing in the motor cortex, communicate with lower MNs in the spinal cord using the excitatory neurotransmitter glutamate. While it is the most prevalent neurotransmitter in the CNS, excess levels of glutamate can be toxic to neurons due to overstimulation of glutamate receptors. This aberrant signaling leads to excessive calcium influx into the cell and, possibly, to activation of other signaling pathways.

Glutamate-induced cytotoxicity is believed to contribute to the demise of MNs in ALS. In the spinal cord, excitatory amino acid transporter 2 (EAAT2, Glt1) is primarily responsible for glutamate clearance. Studies of ALS patients have demonstrated abnormally spliced EAAT2 transcripts in the spinal cord, loss of EAAT2 expression, reduced glutamate transport, and increased levels of glutamate in the cerebrospinal fluid, all pointing to a possible role for glutamate-induced excitotoxicity in the development and/or progression of ALS. Thus, increasing EAAT2 expression was hoped to help mitigate the disease.

To test this hypothesis, Guo et al. generated transgenic mice that overexpress EAAT2 in astrocytes. When this transgene was crossed with the SOD1 mouse to create EAAT2/SOD1 double transgenics, there was a small delay in the onset of motor decline, but the onset of paralysis was not delayed, and there was no improvement in life span. Interestingly, EAAT2/SOD1 mice did show increased MN survival relative to age-matched SOD1 mice. Astrogliosis appeared to be unaffected, but the double transgenic mice did exhibit lower levels of high-molecular-weight SOD1 aggregates. Similar results were found in a study evaluating an AAV8 vector expressing EAAT2 from the astrocyte-specific promoter Gfa2. Despite transducing 83% of GFAP ⁺ cells, no improvement was seen in motor function and MN loss was not reduced. The authors of both studies concluded that the loss of EAAT2 might contribute to, but probably does not cause, MN degeneration.

How glutamate contributes to the progression of ALS may be a more complicated story than a simple model of poor glutamate clearance. Glutamate signaling in MNs primarily occurs through α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors. In general, these receptors are heterotetramers, composed of four subunits. Four different AMPA receptor subunits have been described (GluA1–GluA4, also known as GluR1–GluR4). Depending on the cell type and other factors, different subunits are expressed and combine to form AMPA receptors. The GluA2 subunit is particularly interesting. The mRNA encoding this protein is the target of RNA editing by adenosine deaminase acting on RNA 2 (ADAR2). This enzyme induces a posttranscriptional modification, converting adenosine (A) to inosine (I) by deamination. During translation, inosine pairs with tRNAs as if it were guanosine. In the case of GluA2, this results in glutamine-to-arginine substitution in the protein. This change substantially alters the channel of AMPA receptors containing GluA2. When arginine is present (the edited form, GluA2(R)), the channel is permeable to sodium and potassium but not to calcium. However, in the absence of editing (GluA2(Q)), the presence of the glutamine residue also allows calcium to pass through the channel. In addition, the edited form of GluA2 affects assembly in the ER and slows trafficking to the plasma membrane. Thus, the presence of unedited GluA2 can increase conductance both through permitting calcium influx and increasing the number of channels on the cell surface (reviewed in refs and ).

Under normal conditions for most cells, including MNs, the vast majority of GluA2 subunits are of the edited variety. Interestingly, incomplete editing of GluA2 is present in some neurodegenerative disorders, including Huntington’s disease, Alzheimer’s disease, and schizophrenia. In patients with sporadic ALS, a fraction of MNs exhibit loss of expression of edited GluA2. In addition, a subset of spinal MNs from ALS patients lack expression of ADAR2, an abnormality not found in control cases. These ADAR2 ⁻ cells exhibit TDP-43-positive inclusions, a common finding in ALS. These observations have led some to suggest that ADAR2 deficiency could lead to the development of TDP-43 inclusions and thus represents a possible therapeutic target. Alternatively, it is equally possible that TDP-43 inclusions lead to loss of ADAR2 expression and GluA2 editing. A number of mouse models have been created to begin testing the former hypothesis.

Kuner et al. investigated mice expressing GluA2 with an asparagine inserted in place of the edited codon (GluA2(N)) in addition to the endogenous GluA2 alleles. Like GluA2(Q), this subunit creates an AMPA receptor that is permeable to calcium. These mice began to show deficits in motor function around 35 weeks of age, although they did not progress to paralysis by 2 years of age. By 1 year of age, there was a statistically significant decrease in the number of neurons in the ventral horns. At 2 years of age, about 30% of these cells had been lost. Crossing the GluA2(Q) mouse with the SOD1 mouse led to a significant increase in disease severity. In contrast, mice overexpressing GluA2(R) in MNs have decreased calcium influx into cells. When crossed with SOD1 mice, disease severity was reduced. These studies point to an important role for calcium influx through AMPA receptors in ALS disease progression.

To directly determine whether the failure to edit GluA2 might contribute to ALS disease progression, Hideyama et al. created a conditional ADAR2 knock-out by crossing a floxed ADAR2 mouse with a mouse expressing Cre recombinase from the vesicular acetylcholine transporter. This scheme significantly reduced ADAR2 expression in MNs and led to a significant fraction of MNs expressing unedited GluA2. This mouse exhibited a progressive loss of strength and motor function, although at a fairly protracted rate. Median life span was 82 weeks. The mice exhibited many symptoms of ALS, including fibrillation potentials and fasciculations, loss of NMJ enervation, astrogliosis and microgliosis, and MN loss. Importantly, crossing this mouse to one expressing a GluA2 mutated to only express the edited allele rescued the phenotype, demonstrating that the abnormalities observed in ADAR2-deficient mice were due to the loss of GluA2 editing.

Given the potential role that ADAR2 loss might play in the development or progression of ALS, Yamashita et al. examined gene therapy in the conditional ADAR2 knock-out mouse. They performed an intravenous injection of an AAV9 vector expressing ADAR2 into these mice before or at disease onset. This therapy significantly increased the amount of GluA2 that was correctly edited. Performance on the rotarod task, which deteriorated over time in untreated mice, stabilized with therapy for at least 20 weeks. At 39 weeks of age, MN survival and axon counts were higher in the treated group relative to controls.

While it is not yet clear whether defects in GluA2 editing represent an early or late event in the progression of ALS, the data from the mouse models strongly suggest that altering calcium influx via modulation of the AMPA receptor can at least slow the progression of ALS. Since reduced ADAR2 expression in MNs seems to be common in sporadic ALS patients, gene therapy to restore its lost function represents a plausible approach to treat this aspect of the disorder. However, given the mild presentation of disease in the conditional knock-out of ADAR2, it is likely that ADAR2 gene therapy would need to be given in the context of additional therapies that target other aspects of ALS.

Antioxidant Genes

Oxidative stress may also contribute to ALS disease progression. Reactive oxygen species (ROS), including hydrogen peroxide and the superoxide radical anion, are generated as a byproduct of aerobic metabolism. Unneutralized ROS can react with a number of molecules. For instance, the superoxide radical anion can combine with a nitric oxide radical, producing peroxynitrite which can go on to react with tyrosine residues producing nitrotyrosine. Other molecular changes include carbonylation and oxidative damage to DNA. A number of signs of oxidative damage have been found in postmortem tissue samples taken from ALS patients. These include higher 3-nitrotyrosine levels, evidence for an increased carbonylation, and elevated oxidative damage to DNA. In an in vitro model of ALS, application of catalase can increase cell viability, suggesting that targeting oxidative stress could be beneficial.

Metallothioneins (MTs) are a set of zinc binding proteins that likely have many functions, including protection from oxidative stress. They can also serve as zinc chaperones for SOD1. When zinc is depleted from SOD1, the enzyme more rapidly catalyzes the nitration reaction. Interestingly, mutant SOD1 also has reduced zinc binding and increased nitration activity, possibly pointing to one mechanism of its toxicity. Taken together, these observations suggest that MTs can play a role in slowing ALS disease progression.

Puttaparthi et al. hypothesized that inhibiting one or more MTs would exacerbate disease progression. Three MTs are expressed in the nervous system of mice: MT-I and -II are expressed in glial cells, while MT-III is expressed in neurons. When MT-I and MT-II were both knocked out in a SOD1 animals (MT-I ⁻ ;MT-2 ⁻ ;SOD1 mice), mean survival time was reduced by 32 days (229 vs 261 days). Disease onset was approximately 2 months earlier in the MT-I ⁻ ;MT-2 ⁻ ;SOD1 mice. Astrogliosis was more pronounced at 5.25 months in the MT-I ⁻ ;MT-2 ⁻ ;SOD1 mice than in SOD1 mice, but MN counts were similar between both groups at all time points evaluated. This suggested the glial component of the disease could be altering the function of the remaining MNs, resulting in a faster decline in motor function than would be predicted from the MN counts. When MT-III was knocked out in SOD1 mice, survival was reduced by 51 days (202 vs 253 days). In contrast to the MT-I/II knockout, disease onset was not significantly affected by the loss of MT-III. Instead, the decline in motor function was substantially steeper. In addition, MN loss was significantly greater at 5.25 months in MT-III knockouts, suggesting that MT-III protects MNs from death. These results suggest that altering the expression of MTs using gene therapy could also slow disease progression.

Hashimoto et al. investigated weekly injection of Ad5-MT-III into female SOD1 mice beginning at 20 weeks of age, just before the animals would become overtly symptomatic. The vector was injected into the lower limbs with the hope of getting retrograde transport of the vector to the MNs. Disease onset was not affected, which is not remarkable since treatment was so close to onset. MN counts were significantly higher at 160 days of age in Ad5-MT-III-treated animals. Life span was longer in treated mice as well. The authors found that the Ad vectors themselves were toxic and could accelerate aspects of the disease, and thus a different delivery system should be investigated.

Two other antioxidant genes have been evaluated in the context of gene therapy. Peroxiredoxin 3 (PRDX3) is a mitochondrial thioredoxin-dependent hydroperoxidase and is upregulated in response to oxidative stress. Nuclear factor erythroid 2-related factor 2 (NRF2) is involved in antioxidant response element-mediated gene expression and regulates many phase 2 detoxifying enzymes. Many of the genes controlled by NRF2 are downregulated in ALS. Overexpression of PRDX3 in NSC34 cells significantly increased cell survival under basal conditions and serum starvation, but did not have a significant effect in the face of oxidative stress induced by menadione. This was attributed to the severity of the menadione treatment and the susceptibility of the cells. Murine NRF2 showed protection against menadione in astrocytes and increased survival of NSC34 cells expressing mutant SOD1. These genes were evaluated in vivo using intramuscular injection of AAV6 vectors into 30-day-old SOD1 mice (facial muscles, tongue, intercostal muscles, diaphragm, and hindlimb). No improvement was seen with either vector. However, at endpoint, only 5% of the surviving neurons in green fluorescent protein controls were positive for the transgene. This suggested that poor transduction was to blame for the lack of efficacy.

Neurotrophic Factors

Neurotrophic factors are a set of signaling molecules that are involved in the development and maintenance of the nervous system. They bind to receptors on the cell surface, activating two major signaling cascades: the phosphoinositol-3 kinase (PI3K)/Akt pathway and the mitogen-activated protein kinase/extracellular signal-regulated kinase (MapK/Erk) pathway. In addition, IL-6-related cytokines and granulocyte-colony stimulating factor (G-CSF) can also activate the Janus kinase/signal transducer and activator of transcription (Jak/Stat) pathway ( Fig. 8.2 ). Since space does not permit a detailed description of the biology of each factor, the reader is directed to several excellent reviews: insulin-like growth factor 1 (IGF-1), vascular endothelial growth factor (VEGF), brain-derived neurotrophic factor, glial-derived neurotrophic factor (GDNF), G-CSF, and the IL-6 family of cytokines.

These signaling cascades allow neurotrophic factors to act on the ALS disease state in a number of ways. All three pathways play important roles in regulating axon growth and regeneration, maintaining and, possibly, restoring NMJ innervation. They can also block apoptosis by inhibiting proapoptotic factors and upregulating antiapoptotic factors. Activated Akt phosphorylates serine residues on Bad, caspase 9, and some forkhead family transcription factors, inactivating these proteins and inhibiting apoptosis. The Erk pathway similarly phosphorylates Bad, Bcl-2-interacting mediator of cell death (Bim), and caspase 9. All three pathways upregulate expression of the antiapoptotic proteins Bcl-2 and Bcl-X _L . Finally, both the Jak/Stat and MapK/Erk pathways have been implicated in resistance to ROS. Thus, neurotrophic factors have the potential to attack ALS on a number of fronts.

Recombinant neurotrophic factors have been evaluated in ALS patients. However, the results of these trials have been generally disappointing, with little efficacy shown. There are many potential causes for these failures, including subtherapeutic dosing, poor penetration into the CNS, and off-target effects. It is hoped that gene therapy approaches can overcome some or all of these hurdles.