Review

Correspondence *Department of Neurology, University of Michigan Medical School, Ann Arbor, MI 48109, USA

Email henry-paulson@uiowa.edu

Introduction

Andrew Fire and Craig Mello won last year's Nobel Prize in Physiology or Medicine for their discovery of RNA interference (RNAi) less than a decade after their seminal publication.^{1, 2} This remarkably short time from discovery to recognition by the Nobel Committee speaks volumes about the importance of RNAi. Not only did their discovery reveal a novel, evolutionarily conserved pathway for gene regulation, it also led to the generation of powerful new tools for biological research and suggested novel therapeutic approaches for human disease.

Clinicians with an interest in scientific advances will probably have already read much about RNAi. In a PubMed search, the term "RNAi or RNA interference" returned 3,227 articles published in 2006 alone. By contrast, there were only 31 such articles in 1999. This difference illustrates both the remarkable growth of interest in RNAi and its widespread adoption as a laboratory method.

In this article, we provide an overview of RNAi, focusing on its promise as a therapeutic strategy for neurological disorders.^{3, 4, 5} We pay particular attention to neurodegenerative diseases, as many of these conditions represent strong candidate disorders in which RNAi could be applied as therapy.

Principles of RNA interference

In their landmark paper,¹ Fire et al. reported the discovery of a powerful RNA-based mechanism that silences genes in a sequence-specific manner. In nematodes, they found that very small amounts of double-stranded RNA (dsRNA) complementary to a particular gene could suppress the expression of that gene by virtually eliminating its messenger RNA (mRNA). This dsRNA-mediated silencing proved to be much more potent than that caused by the single-stranded, antisense reagents that had long been used by scientists. In this newly discovered mechanism, only a few dsRNA molecules per cell were needed to silence the targeted gene. This observation led the authors to predict—correctly—that dsRNA-mediated silencing must be a catalytic (i.e. enzymatic) process rather than simply a stoichiometric (i.e. binding) process.

The discovery of this RNA-based gene-silencing in organisms as diverse as plants, worms and humans suggested that cells must possess evolutionarily conserved machinery that mediates the process. In other words, there must be a collection of cellular proteins that recognize and process dsRNAs, then incorporate these small effector molecules into an enzymatic complex that can bind and cleave complementary (target) mRNAs. This machinery has now been well described through the work of many groups, as highlighted in several recent reviews.^{6, 7, 8, 9, 10, 11, 12, 13, 14} Cells possess two dedicated ribonuclease (RNase) III enzymes, Drosha and Dicer, which process precursor dsRNAs into smaller intermediates known as small interfering RNAs (siRNAs).⁷ At approximately 21 nucleotides in length and containing 3' overhangs, siRNAs possess both a guide (or antisense) strand that is complementary to the target mRNA, and a passenger strand. The guide strand is incorporated into the macromolecular RNA-induced silencing complex (RISC). RISC contains the endonuclease Argonaute 2 (AGO2), which cleaves bound mRNA provided that it is perfectly complementary to the guide strand.¹⁵ The cleaved mRNA is rapidly destroyed, leaving the RISC–guide strand complex available for additional cycles of mRNA binding and cleavage (Figure 1, purple arrows denote siRNA pathway). As a result of this process, expression of the targeted gene is markedly suppressed.

Figure 1 The process of RNA interference and its manipulation

Whereas some nuclear events are unique to the endogenous or the exogenous RNA interference (RNAi) pathway, many nuclear processing steps (1,2) and all cytoplasmic processing steps (3,4) are shared by both pathways, presenting potential sources of saturation. Exogenous RNAi-mediating molecules can follow the miRNA pathway (black arrows) or the siRNA pathway (purple arrows), entering the pathways at different points depending on the specific molecule and delivery vehicle employed. Most endogenous miRNA genes are transcribed by RNA polymerase II before nuclear processing by the Drosha–DGCR8 complex (1). Exogenous miRNAs can be engineered as pri-miRNAs, thereby following the same steps, or as shRNA templates that enter the nucleus but are transcribed by RNA polymerase III, thereby bypassing the nuclear transcriptional and processing machinery employed by the endogenous pathway. All nuclear RNAi-mediating constructs converge in the nuclear export event (2), a possible rate-limiting step and source of saturation. Once in the cytoplasm, dsRNAs still joined by a loop, whether derived from nuclear processing events or from exogenous delivery as synthetic Dicer substrates, require processing by TRBP–Dicer (3), resulting in separation into two strands. Exogenous siRNAs do not need Dicer processing, and therefore bypass steps 1–3. All dsRNAs converge in loading the guide strand into RISC (4), whereas endogenous miRNAs mostly lead to translational repression, and exogenous dsRNAs to target cleavage. Abbreviations: DGCR8, DiGeorge syndrome critical region 8; dsRNAs, double-stranded RNAs; Exp5, exportin-5; miRNA, microRNA; mRNA, messenger RNA; pre-miRNA, precursor microRNA; pri-miRNA, primary microRNA; pol, RNA polymerase; RISC, RNA-induced silencing complex; shRNA, short hairpin RNA; siRNA, small interfering RNA; TRBP, trans-activation-responsive RNA binding protein.

Full figure and legend (47K)Figures & Tables indexDownload Power Point slide (105K)

As a result of its potency and selectivity, RNAi has rapidly become standard methodology in biological labs.¹⁶ Scientists can exploit this phenomenon to suppress the expression of virtually any gene of interest, simply by introducing exogenous siRNAs that are perfectly complementary to the target mRNA into cells or organisms. To help scientists design and test their RNAi effector molecules, a growing array of reagents and web-based informatics tools, from both commercial and academic sources, is available. By employing this new technology, neurobiologists can dissect physiological pathways and characterize disease-linked genes more rapidly than ever before.¹⁷

This conserved RNAi machinery did not, however, evolve to help scientists turn off their favorite genes. Rather, it serves various important biological purposes, which differ depending on the organism in question.¹⁸ In humans and other mammals the genome encodes hundreds of endogenous, small dsRNA molecules known as microRNAs (miRNAs), which regulate the expression of a substantial proportion of the protein-coding genes in the mammalian genome.^{19, 20, 21, 22} Unlike exogenous siRNAs, which are designed to be perfectly complementary to the target mRNA and, therefore, result in RISC-mediated cleavage of the target mRNA, endogenous miRNAs in mammals typically are not fully complementary to their targets. Rather than directing cleavage of target mRNAs, miRNAs suppress gene expression by blocking translation of the target mRNA and facilitating its degradation at specific cellular sites known as processing bodies or P-bodies.^{23, 24} The synthesis of endogenous miRNAs proceeds through several steps beginning with their transcription as larger, primary miRNA transcripts that are often polycistronic. These primary miRNAs are processed in the nucleus by Drosha into 70 nucleotide stem-loop hairpin structures known as precursor miRNAs, which are then translocated from the nucleus via an exportin-5-dependent mechanism, and processed in the cytoplasm by Dicer into mature, 22 nucleotide, miRNAs that are loaded into RISC (Figure 1, black arrows denote miRNA pathway).^{6, 8, 9, 10, 14} It is important to mention miRNA biosynthesis here because, as illustrated in Figure 1, the same pathway also serves to process exogenous RNAi effector molecules when they are delivered by plasmid or virus to the cell and expressed intracellularly as short hairpin RNAs (shRNAs).

This previously underappreciated world of noncoding RNAs is one of the most exciting areas of research in developmental neuroscience.²⁵ Specific miRNAs have been implicated in cellular processes that are critical for brain development, including neuronal differentiation and dendritic spine organization,²⁶ and are likely to have a major role in neurological disease processes. For example, dysfunction in the RNAi pathway might underlie various neurodevelopmental disorders, as suggested by the fact that fragile X mental retardation protein (FMRP), which is deficient in fragile X syndrome, the most common inherited form of mental retardation, is itself a component of the RNAi machinery.²⁷ Although we will not explore miRNAs in depth (this topic is covered in more detail in another recent review),²⁵ any discussion of RNAi as therapy must acknowledge the importance of the endogenous miRNA pathway. Co-option of this pathway to suppress a specific disease gene for therapeutic purposes could have adverse effects by interfering with the normal biological function of miRNAs in neurons and other cell types.

Therapeutic RNA interference and neurological disease

In many chronic diseases, conventional 'druggable' targets are still lacking. Such diseases pose a major public health problem if they are common, incurable disorders that result in significant disability or death. In these cases, the exploration of alternative therapeutic strategies like RNAi becomes a medical priority.²⁸

For some such disorders, a growing understanding of the molecular events underlying pathogenesis has implicated specific gene products for which suppression would be expected to lead to a therapeutic benefit. This is precisely the scenario in which RNAi should be explored as a potential treatment strategy.^{29, 30} At least three major groups of human diseases fit this profile: neoplasias,³¹ infections,³² and neurodegenerative disorders.^{3, 4} In all three, development of therapeutic RNAi is now underway. Although cancer and viral infections often involve the nervous system and are, therefore, important to the clinical neuroscientist, we will focus on the application of therapeutic RNAi to diseases caused by neuronal degeneration.

If neurologists are to succeed in developing RNAi-based strategies for neurodegenerative disorders, what types of targets should be sought? In certain diseases the answer is obvious. When a specific gene product triggers disease through a toxic mechanism, RNAi-mediated downregulation of this gene product offers a simple, direct route to therapy. Many dominantly inherited neurodegenerative diseases fit this model, including those that are caused by expansion of polyglutamine tracts in the disease protein. In the polyglutamine disorders, which include Huntington's disease (HD) and many spinocerebellar ataxias (SCAs), the expanded repeat mutation promotes misfolding of the disease protein.^{33, 34} This in turn initiates a complicated, still poorly understood, molecular cascade leading to neuronal dysfunction and cell death. Whereas other therapeutic strategies proposed for polyglutamine diseases act more distally in this molecular cascade, RNAi-based efforts to prevent expression of the pathogenic protein attack the disease close to its root. Many other dominant neurodegenerative disorders should also be amenable to RNAi-mediated downregulation of a toxic gene product, including familial forms of Alzheimer's disease (AD; caused by mutations in genes encoding amyloid precursor protein or the presenilins), amyotrophic lateral sclerosis (ALS; e.g. caused by mutations in the superoxide dismutase 1 [SOD1] gene) and frontotemporal dementia (e.g. mutations in the gene encoding the microtubule-associated protein tau).⁵

A similar scenario is presented by dominantly inherited diseases attributable to mutations that act in a dominant-negative manner. In this setting, the mutated protein not only fails to function, but also prevents normal functioning of the protein encoded by the wild-type allele. A recently discovered case in point is DYT1 dystonia, the most common inherited dystonia. In this disorder, a three-base-pair deletion mutation in the DYT1 gene, which encodes the protein torsinA, causes redistribution of the mutant protein from its normal location in the endoplasmic reticulum to the nuclear envelope.^{35, 36} Wild-type torsinA, which forms a complex with mutant torsinA, is probably co-transported with the mutant form to the nuclear envelope, resulting in an overall loss of torsinA function.

Silencing both alleles by RNAi clearly would not offer therapeutic benefit in DYT1 dystonia or other disorders caused by a dominant-negative mechanism. Preventing the expression of the mutant gene product while sparing the normal allele, however—an RNAi strategy known as allele-specific silencing³⁷—could rescue the normal function of the wild-type protein. Indeed, RNAi-mediated 'rescue' of the normal DYT1 gene product has been achieved in cell-based models that employed RNAi reagents specific for the mutant allele.^{38, 39} In this particular case, the three-base-pair difference between normal and mutant alleles simplified the task of designing allele-specific reagents—the deletion virtually ensures that an siRNA targeting the site of mutation will not also act on the wild-type allele. It is also possible, however, to design RNAi molecules that distinguish even subtler differences between two alleles—including pathogenic alleles that differ from wild-type alleles by as little as a single nucleotide.^{40, 41} As more is learned about how guide strands and RISC bind to and cleave complementary target mRNA, the strategies employed to design successful allele-specific reagents are likely to become even more sophisticated.

The dominant diseases described above are logical candidates for RNAi therapy, but what about diseases in which the pathogenic molecular trigger is unknown, dominant diseases resulting from haploinsufficiency, or even recessive diseases in which loss of function is the pathogenic mechanism? In these cases, the proximal cause of disease is either unknown or the implicated gene is not a logical target for RNAi. Even when the initiating event or gene is not a potential target, however, it might trigger a pathogenic cascade involving downstream targets that do represent good candidates for downregulation by RNAi. One neurological disease belonging to this category is tuberous sclerosis (TS). TS is caused by an inactivating mutation in either one of two genes, TSC1 and TSC2, which encode hamartin and tuberin, respectively. These two proteins form a functional complex that represses a common signaling pathway that is known to stimulate cell proliferation. In TS, this pathway is left unchecked because of inactivating mutations in hamartin or tuberin. As a result, people with TSC1 or TSC2 mutations develop the CNS and dermatologic lesions characteristic of this disease. RNAi-mediated downregulation of one or more unchecked, downstream components of this signaling pathway might help restore normal regulation and thereby counteract the fundamental pathogenic process in TS.⁴²

In this Review, we have focused on disease examples in which we know plausible and attractive molecular targets. It is important to keep in mind, however, that continuing efforts to identify the genetic basis of neurodegenerative diseases, including the definition of genetic risk factors for common disorders such as AD and Parkinson's disease (PD), are expected to uncover additional molecular targets. As these targets are identified, the range of neurodegenerative diseases that might be candidates for RNAi therapy will expand. For example, the recent finding that increased levels of -synuclein expression might be associated with increased risk of developing PD suggests this gene as a potential RNAi target in this common disorder.⁴³

The RNA interference therapeutic pipeline

Several human phase I and II therapeutic RNAi trials have already been completed or are nearing completion.²⁸ These trials have targeted common conditions including age-related macular degeneration, and respiratory syncytial virus and hepatitis C infection. In all cases to date, the delivered therapeutic reagents have been synthetic, chemically modified siRNAs. Although none of these early trials has targeted a brain disease, they should provide important information regarding the optimal dosage and frequency of administration of RNAi-mediating compounds. These trials might also shed light on potential safety issues arising from exogenous manipulation of intrinsic RNAi pathways in humans.

In view of the unique features of the brain, the targeting of CNS diseases poses additional obstacles not faced in current RNAi trials. Perhaps the biggest challenge facing the field of RNAi research is achieving safe and efficient delivery of therapeutic RNAi agents to neurons—highly specialized, postmitotic cells protected by the blood–brain barrier. The chronic nature of many brain disorders also means that sustained delivery of an RNAi reagent, or its repeated administration, will be required to achieve a long-term benefit.

What is the best way to proceed? A valid therapeutic strategy for a given neurological disease should address the following four questions. What is the most appropriate therapeutic RNAi agent? How should it be delivered? Where is the principal anatomical target? When in the course of disease should the therapy be applied?⁵ Before beginning human trials, the chosen strategy will also need to undergo preclinical testing in validated animal models. Although common sporadic diseases such as AD or PD would seem to be high priority targets, the fact that most cases of these diseases are not attributable to single genetic defects will make RNAi therapy difficult. Human RNAi clinical trials for neurological disorders are likely to be initiated in unequivocally inherited disorders. Despite the rarity of many of these disorders, they are attractive candidate diseases because they often have a well-defined genetic etiology and a relatively well-understood pathobiology and anatomical substrate. Moreover, appropriate animal models exist in which to carry out the necessary preclinical tests.

A candidate disease that fulfills these various criteria is SCA type 1 (SCA1). In this dominantly inherited polyglutamine expansion disorder, the mutant ataxin-1 protein acts via a toxic, gain-of-function mechanism that causes neuronal dysfunction and cell death in the cerebellum and certain brainstem nuclei.^{44, 45} Mice lacking the Atxn1 gene show only subtle behavioral abnormalities, suggesting that RNAi-mediated knockdown of both alleles in humans would be unlikely to result in adverse effects caused by loss of gene function.⁴⁶ Studies in an inducible transgenic mouse model of SCA1 have also demonstrated that the disease phenotype can be reversed by suppressing expression of mutated ataxin-1;⁴⁷ this promising result suggests that RNAi-mediated knockdown of ataxin-1 expression in mildly to moderately symptomatic humans might actually lead to some recovery of function, not simply to a slowing of disease progression.

The first successful RNAi trial for a neurodegenerative disease was completed in a transgenic mouse model of SCA1. Using adeno-associated virus to deliver an shRNA specific for Atxn1, Xia et al. were able to suppress ataxin-1 expression in the cerebellum of SCA1 mice, leading to cytological, histological and motor improvement in the treated animals.⁴⁸ A negative control virus encoding missense shRNA showed no effects and, importantly, the ataxin-1 shRNA-expressing virus did not cause any adverse effects in wild-type mice. SCA1 seems, therefore, to be well positioned to take the lead in the RNAi therapeutic pipeline for neurodegenerative diseases. Although the overall public health impact of slowing down disease in patients with SCA1 would be rather small because of the rarity of the condition, clinical success in a disease like SCA1 would be extremely important. It would demonstrate the feasibility of the RNAi approach, and might, therefore, pave the way for future studies in more-prevalent, and perhaps more-challenging, neurodegenerative diseases.

As with any disease, SCA1 is not without its hurdles as a candidate RNAi target. The pattern of brain degeneration in SCA1 is fairly widespread, involving both the cerebellum and brainstem. The brainstem in particular does not present an easy target for RNAi delivery. Moreover, the successful preclinical RNAi study was performed in transgenic mice in which the mutant gene was only expressed in Purkinje cells. Additional studies in SCA1 knock-in mice, which more faithfully recreate the widespread pattern of neurodegeneration in humans and express the disease allele at physiological levels, might be helpful before a clinical trial is initiated in humans.

Since the first report of successful therapeutic RNAi in SCA1 mice, many successful preclinical trials in mouse models of other neurodegenerative diseases have been completed and have had their results published (Table 1). These studies have silenced the disease-causing gene products in animal models of familial ALS (SOD1),^{49, 50, 51, 52} familial PD (-synuclein),⁵³ and HD (huntingtin).^{54, 55, 56, 57} Other studies have suppressed the expression of key molecules in the pathogenic cascade underlying familial and sporadic forms of AD (-site APP-cleaving enzyme 1 or BACE1),^{58, 59} and acquired Creutzfeldt–Jakob disease (endogenous prion protein).⁶⁰ These trials have used different RNAi effector molecules, explored various delivery methods (mostly virus-mediated delivery), and targeted diverse anatomical areas ranging from the hippocampus to motor neurons and muscle.

Table 1 Preclinical RNA interference trials in rodent models of neurodegenerative diseases

Full tableFigures & Tables indexDownload Power Point slide (153K)

These successful preclinical studies suggest several other candidate diseases that could take the lead in initiating the clinical application of RNAi therapy. For example, the polyglutamine disorder HD has generated significant interest as a promising candidate for therapeutic RNAi. HD has been successfully targeted with RNAi by various groups that employed different transgenic mouse models and diverse approaches.^{54, 55, 56, 57}

Another disease considered to be an attractive early target for RNAi therapy is familial ALS caused by dominant-acting mutations in SOD1. SOD1-mediated ALS accounts for only a small fraction of all familial cases, which in turn comprise only about 10% of all cases of ALS. Nevertheless, the clear-cut success of RNAi in different preclinical mouse studies of SOD1-mediated ALS^{49, 50, 51, 52} argues compellingly that RNAi should be developed for this subset of disease. The rapidly progressive, fatal course of this disease and the lack of successful alternative therapies make the need all the more pressing.

Finally, the polyglutamine disorder SCA7 presents a special case because, in addition to progressive ataxia, most patients develop a retinopathy leading to blindness.⁴⁴ Trials in this disease might start by exclusively targeting the retina. Although not yet published, preliminary reports of phase I and II trials of RNAi for age-related macular degeneration, in which synthetic siRNAs have been directly injected into the eye, are promising.²⁸ Using the same approach, retinal expression of ataxin-7 could be silenced, with the expectation that the disabling visual loss would be halted or reversed. A successful intraocular trial would facilitate the design of subsequent trials attempting to silence the mutant gene in the nervous system.

Challenges ahead

The most important information gained from preclinical studies has been the demonstration that an RNAi approach, using various delivery strategies, is feasible for a wide variety of neurodegenerative diseases (Table 1). Studies have also, however, highlighted some of the challenges that the RNAi field faces before it is in a position to conduct human trials. These challenges include achieving efficient delivery, reducing possible off-target effects and adverse consequences of therapeutic RNAi, and developing reliable measures of efficacy in human studies.

It is no coincidence that all but one of the in vivo preclinical studies mentioned above were performed with intracellularly expressed RNAi delivered via recombinant virus. Although synthetic, chemically modified siRNAs have worked well in systemic delivery to other organs, the blood–brain barrier makes this kind of delivery a more daunting task in brain disorders. Direct intraparenchymal injection of neurotropic virus circumvents the blood–brain-barrier issue.⁶¹ In addition, virally delivered shRNAs provide continued intracellular expression of the RNAi effector molecule, something that will probably be required in chronic neurodegenerative disease states. Although viral delivery of therapeutic RNAi has been highly effective in rodent models, the reader should keep in mind that the mouse striatum or cerebellum can be transduced with one or a few small-volume injections. The human brain, being orders of magnitude larger, will require a more complex administration approach. Additional studies in the nonhuman primate brain, or possibly porcine brain, should help bridge the gap between preclinical rodent studies and human trials. Furthermore, improvements in the large-scale production of clinical-grade viral vectors are likely to be required if this technology is to successfully enter the clinical arena.

Although viral delivery of RNAi for specific neurodegenerative diseases is likely to be developed first, continuing advances in modifications of synthesized siRNAs could eventually lead to successful CNS delivery.^{28, 62, 63} Chemical modifications at specific points along the siRNA ribose backbone greatly stabilize siRNAs without unacceptably reducing potency. Likewise, conjugation to cholesterol or incorporation into liposomes stabilizes siRNAs and facilitates delivery to tissues. Tethering the siRNA to antibody fragments or to RNA aptamer motifs that bind specific cell-surface receptors can both enhance and restrict delivery to specific cell populations in the CNS. In addition, encasing siRNAs in customized nanoparticles that are coated with specific ligands can promote delivery to specific cell populations. The distinct advantage of synthetic siRNAs for many applications is their transient nature; any unexpected adverse consequences can be countered by halting further administration. The same cannot be said of RNAi-expressing virus, which would be expected to continue expressing shRNA against the target gene. For this reason, efforts are underway to adapt existing regulable expression systems for RNAi viral delivery so that expression in humans might be titrated up or down by varying the administration of a simple compound.

To date, preclinical studies in animal models have focused on efficacy, with less attention having been paid to the potential adverse consequences of exogenous manipulation of the RNAi pathway. In a recent report, virus-mediated RNAi delivered to hepatocytes in vivo caused significant toxicity by impairing aspects of the endogenous miRNA pathway.⁶⁴ Similar effects could occur in neurons—postmitotic cells in which endogenous miRNAs play a major role.^{25, 65} Broadly speaking, adverse consequences of RNAi can derive from the abnormal presence of a dsRNA in the target cell, the utilization of the endogenous RNAi pathway by exogenous dsRNA, and off-target effects caused by the specific sequences employed in the therapeutic construct (Table 2).³⁰

Table 2 Potential adverse effects of RNA interference

Full tableFigures & Tables indexDownload Power Point slide (93K)

One possible adverse consequence of RNAi is the induction of an immunostimulatory response. This possibility is greater with exogenously delivered siRNAs than with intracellularly expressed shRNAs. The interferon response is a naturally occurring cellular defense mechanism against the entry of exogenous nucleic acids.⁶⁶ Double-stranded RNAs longer than 30 bases are recognized by this protective pathway, triggering a molecular cascade that ends in global translational repression. Although dsRNA shorter than 30 bases was not expected to trigger this response, initial studies in various cellular systems yielded conflicting results,^{67, 68, 69, 70, 71, 72} suggesting the possibility of cell-type specificity. Studies in cultured mammalian neurons transduced with virally encoded shRNAs did not detect activation of this protective response.³⁹ Furthermore, assessment of the interferon response in mice that were overexpressing an shRNA that targets mutant SOD1 failed to show activation.⁵² Moreover, even if triggered, the interferon response is usually transient and might not have notable long-term effects if an RNAi-mediating construct was to be administered only once. Repeated administration of synthetic siRNAs, however, could lead to periodic activation of this protective response. In future RNAi studies in the CNS, a systematic analysis of possible immunostimulatory responses, including but not limited to the interferon response, will be important to clarify the extent to which immune responses will act as a barrier to effective therapy.

A second class of potential adverse effects are those derived from saturation of the RNAi machinery by exogenously administered dsRNAs, which could interfere with the normal function of endogenous miRNAs that are critical for neuronal function. The need to carefully assess this possibility was underscored by a recent study that reported shRNA-induced abnormalities in dendritic arborization of cultured rat neurons.⁶⁵ Minor alterations in the miRNA pathway could lead to functional defects such as abnormal neuronal polarization or altered synaptic function. Although these defects might not lead to overt behavioral abnormalities in rodents, they could have profound effects in the more complex human brain. Development of appropriate measures of endogenous RNAi function that could be employed in the brains of different mammalian species would help to ensure that scientists advance with an acceptable level of safety toward human trials.

A third area of concern is potential off-target effects resulting from the specific sequence contained in the RNAi reagent. In choosing a target sequence and designing an RNAi molecule, informatics-based searches are routinely performed to exclude identical or highly similar sequences in other genes, which might inadvertently be affected. Even in the presence of significant sequence mismatches (particularly in the 3' half of the antisense strand), however, other unintended mRNAs could be suppressed.^{68, 73, 74} The RNAi field still lacks a completely reliable algorithm to predict unintended effects. Before applying a specific dsRNA to the human brain, researchers should consider performing experiments in cultured human neural cells with gene microarray analysis to identify the pattern of neuron-expressed genes that are unexpectedly suppressed by the specific sequence used.

It can be assumed that most RNAi reagents will have identifiable off-target effects. From the perspective of therapeutic RNAi, however, what really matters is whether these off-target effects cause in vivo toxicity or otherwise impede the beneficial effect of the RNAi reagent in suppressing the target disease gene. To lessen the risk of these potential adverse consequences, we must strive to use the lowest clinically effective doses of RNAi reagents that have been engineered to be processed as efficiently as possible and that also favor incorporation of the guide strand into RISC. This goal can be accomplished now that the rules governing strand selection are better understood,^{75, 76} the engineering of shRNAs to mimic endogenous miRNA-like structures has improved,⁷⁷ and the capacity to control expression with tissue-specific and regulable promoters is at hand.⁷⁸

Conclusions

The possibility of exploiting the RNAi pathway to treat incurable neurodegenerative diseases has generated high expectations both in the medical community and among the lay public. The established potency and selectivity of RNAi, coupled with ongoing advances that are enhancing the efficiency and delivery of RNAi molecules, lead us to be cautiously optimistic that RNAi as therapy will reach the neurology clinic. While initial studies in cellular and animal models of neurodegenerative disease have produced encouraging results, the safety of this therapy in the diseased human CNS has not, however, been established. Furthermore, efficient delivery of RNAi-mediating molecules and the development of reliable methods to monitor the influence of this intervention on the endogenous miRNA pathway still remain challenging obstacles. The rapid progress in the field of RNAi research suggests, however, that these obstacles are not insurmountable. Advances have derived from the collaborative efforts of neuroscientists working with RNAi researchers, who are defining how the endogenous miRNA pathway functions and are thereby facilitating the design of effector molecules, and with gene therapists, who are expanding the quantity and quality of delivery systems. It is remarkable that merely a decade after the landmark discovery by Fire and Mello, treatment of neurodegenerative diseases through RNAi is becoming a plausible scenario. In the next few years we might witness initial human clinical RNAi trials for neurodegenerative diseases such as ALS, HD and the SCAs, most probably as joint efforts between academia and the biotechnology industry. Only then will we know how far we are from applying RNAi to more-common neurodegenerative diseases.

Acknowledgments

Work on RNA interference in the authors' laboratories has been funded by grants from American Federation for Aging Research (HLP), Dystonia Medical Research Foundation (PG) and NIH (PO1 NS050210, directed by Beverly Davidson, University of Iowa).

References

1. Fire A et al. (1998) Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature 391: 806–811 | Article | PubMed | ISI | ChemPort |
2. Paulson H and Gonzalez-Alegre P (2006) RNAi gets its prize. Lancet Neurol 5: 997–999 | PubMed |
3. Gonzalez-Alegre P (2007) Therapeutic RNA interference for neurodegenerative diseases: from promise to progress. Pharmacol Ther 114: 34–55 | PubMed | ChemPort |
4. Davidson BL and Paulson HL (2004) Molecular medicine for the brain: silencing of disease genes with RNA interference. Lancet Neurol 3: 145–149 | Article | PubMed | ISI | ChemPort |
5. Rodriguez-Lebron E and Gonzalez-Alegre P (2006) Silencing neurodegenerative disease: bringing RNA interference to the clinic. Expert Rev Neurother 6: 223–233 | Article | PubMed | ChemPort |
6. Bartel DP (2004) MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116: 281–297 | Article | PubMed | ISI | ChemPort |
7. Carmell MA and Hannon GJ (2004) RNase III enzymes and the initiation of gene silencing. Nat Struct Mol Biol 11: 214–218 | Article | PubMed | ISI | ChemPort |
8. Cullen BR (2004) Transcription and processing of human microRNA precursors. Mol Cell 16: 861–865 | Article | PubMed | ISI | ChemPort |
9. Du T and Zamore PD (2005) microPrimer: the biogenesis and function of microRNA. Development 132: 4645–4652 | Article | PubMed | ISI | ChemPort |
10. Kim VN (2005) MicroRNA biogenesis: coordinated cropping and dicing. Nat Rev Mol Cell Biol 6: 376–385 | Article | PubMed | ISI | ChemPort |
11. Meister G and Tuschl T (2004) Mechanisms of gene silencing by double-stranded RNA. Nature 431: 343–349 | Article | PubMed | ISI | ChemPort |
12. Murchison EP and Hannon GJ (2004) miRNAs on the move: miRNA biogenesis and the RNAi machinery. Curr Opin Cell Biol 16: 223–229 | Article | PubMed | ISI | ChemPort |
13. Tomari Y and Zamore PD (2005) Perspective: machines for RNAi. Genes Dev 19: 517–529 | Article | PubMed | ISI | ChemPort |
14. Tomari Y and Zamore PD (2005) MicroRNA biogenesis: drosha can't cut it without a partner. Curr Biol 15: R61–R64 | Article | PubMed | ISI | ChemPort |
15. Meister G et al. (2004) Human Argonaute2 mediates RNA cleavage targeted by miRNAs and siRNAs. Mol Cell 15: 185–197 | Article | PubMed | ISI | ChemPort |
16. Miller VM et al. (2005) RNA interference in neuroscience: progress and challenges. Cell Mol Neurobiol 25: 1195–1207 | Article | PubMed |
17. Gunsalus KC and Piano F (2005) RNAi as a tool to study cell biology: building the genome–phenome bridge. Curr Opin Cell Biol 17: 3–8 | Article | PubMed | ISI | ChemPort |
18. Dykxhoorn DM and Lieberman J (2005) The silent revolution: RNA interference as basic biology, research tool, and therapeutic. Annu Rev Med 56: 401–423 | Article | PubMed | ISI | ChemPort |
19. Bentwich I et al. (2005) Identification of hundreds of conserved and nonconserved human microRNAs. Nat Genet 37: 766–770 | Article | PubMed | ISI | ChemPort |
20. Berezikov E et al. (2005) Phylogenetic shadowing and computational identification of human microRNA genes. Cell 120: 21–24 | Article | PubMed | ISI | ChemPort |
21. Xie X et al. (2005) Systematic discovery of regulatory motifs in human promoters and 3[prime] UTRs by comparison of several mammals. Nature 434: 338–345 | Article | PubMed | ISI | ChemPort |
22. Lim LP et al. (2005) Microarray analysis shows that some microRNAs downregulate large numbers of target mRNAs. Nature 433: 769–773 | Article | PubMed | ISI | ChemPort |
23. Jackson RJ and Standart N (2007) How do microRNAs regulate gene expression? Sci STKE 2007: re1 | Article | PubMed |
24. Liu J et al. (2005) MicroRNA-dependent localization of targeted mRNAs to mammalian P-bodies. Nat Cell Biol 7: 719–723 | Article | PubMed | ISI | ChemPort |
25. Kosik KS (2006) The neuronal microRNA system. Nat Rev Neurosci 7: 911–920 | Article | PubMed | ISI | ChemPort |
26. Giraldez AJ et al. (2005) MicroRNAs regulate brain morphogenesis in zebrafish. Science 308: 833–838 | Article | PubMed | ISI | ChemPort |
27. Jin P et al. (2004) RNA and microRNAs in fragile X mental retardation. Nat Cell Biol 6: 1048–1053 | Article | PubMed | ISI | ChemPort |
28. Bumcrot D et al. (2006) RNAi therapeutics: a potential new class of pharmaceutical drugs. Nat Chem Biol 2: 711–719 | Article | PubMed | ISI | ChemPort |
29. Shankar P et al. (2005) The prospect of silencing disease using RNA interference. JAMA 293: 1367–1373 | Article | PubMed | ISI | ChemPort |
30. Kim DH and Rossi JJ (2007) Strategies for silencing human disease using RNA interference. Nat Rev Genet 8: 173–184 | Article | PubMed | ISI | ChemPort |
31. Izquierdo M (2005) Short interfering RNAs as a tool for cancer gene therapy. Cancer Gene Ther 12: 217–227 | Article | PubMed | ISI | ChemPort |
32. Ketzinel-Gilad M et al. (2006) RNA interference for antiviral therapy. J Gene Med 8: 933–950 | Article | PubMed | ChemPort |
33. Paulson HL (2000) Toward an understanding of polyglutamine neurodegeneration. Brain Pathol 10: 293–299 | PubMed | ChemPort |
34. Paulson HL et al. (2000) Polyglutamine disease and neuronal cell death. Proc Natl Acad Sci USA 97: 12957–12958 | Article | PubMed | ChemPort |
35. Gonzalez-Alegre P and Paulson HL (2004) Aberrant cellular behavior of mutant torsinA implicates nuclear envelope dysfunction in DYT1 dystonia. J Neurosci 24: 2593–2601 | Article | PubMed | ISI | ChemPort |
36. Goodchild RE and Dauer WT (2004) Mislocalization to the nuclear envelope: an effect of the dystonia-causing torsinA mutation. Proc Natl Acad Sci USA 101: 847–852 | Article | PubMed | ChemPort |
37. Rodriguez-Lebron E and Paulson HL (2006) Allele-specific RNA interference for neurological disease. Gene Ther 13: 576–581 | Article | PubMed | ChemPort |
38. Gonzalez-Alegre P et al. (2003) Toward therapy for DYT1 dystonia: allele-specific silencing of mutant TorsinA. Ann Neurol 53: 781–787 | Article | PubMed | ISI | ChemPort |
39. Gonzalez-Alegre P et al. (2005) Silencing primary dystonia: lentiviral-mediated RNA interference therapy for DYT1 dystonia. J Neurosci 25: 10502–10509 | Article | PubMed | ChemPort |
40. Miller VM et al. (2003) Allele-specific silencing of dominant disease genes. Proc Natl Acad Sci USA 100: 7195–7200 | Article | PubMed | ChemPort |
41. Schwarz DS et al. (2006) Designing siRNA that distinguish between genes that differ by a single nucleotide. PLoS Genet 2: e140 | Article | PubMed |
42. Marcotte L and Crino PB (2006) The neurobiology of the tuberous sclerosis complex. Neuromolecular Med 8: 531–546 | Article | PubMed | ChemPort |
43. Singleton AB et al. (2003) -Synuclein locus triplication causes Parkinson's disease. Science 302: 841 | Article | PubMed | ISI | ChemPort |
44. Paulson H and Ammache Z (2001) Ataxia and hereditary disorders. Neurol Clin 19: 759–782, viii | Article | PubMed | ChemPort |
45. Humbert S and Saudou F (2002) Toward cell specificity in SCA1. Neuron 34: 669–670 | Article | PubMed | ISI | ChemPort |
46. Matilla A et al. (1998) Mice lacking ataxin-1 display learning deficits and decreased hippocampal paired-pulse facilitation. J Neurosci 18: 5508–5516 | PubMed | ISI | ChemPort |
47. Zu T et al. (2004) Recovery from polyglutamine-induced neurodegeneration in conditional SCA1 transgenic mice. J Neurosci 24: 8853–8861 | Article | PubMed | ISI | ChemPort |
48. Xia H et al. (2004) RNAi suppresses polyglutamine-induced neurodegeneration in a model of spinocerebellar ataxia. Nat Med 10: 816–820 | Article | PubMed | ISI | ChemPort |
49. Ralph GS et al. (2005) Silencing mutant SOD1 using RNAi protects against neurodegeneration and extends survival in an ALS model. Nat Med 11: 429–433 | Article | PubMed | ISI |
50. Raoul C et al. (2005) Lentiviral-mediated silencing of SOD1 through RNA interference retards disease onset and progression in a mouse model of ALS. Nat Med 11: 423–428 | Article | PubMed | ISI |
51. Saito Y et al. (2005) Transgenic small interfering RNA halts amyotrophic lateral sclerosis in a mouse model. J Biol Chem 280: 42826–42830 | Article | PubMed | ChemPort |
52. Xia X et al. (2006) Allele-specific RNAi selectively silences mutant SOD1 and achieves significant therapeutic benefit in vivo. Neurobiol Dis 23: 578–586 | Article | PubMed | ChemPort |
53. Sapru MK et al. (2006) Silencing of human -synuclein in vitro and in rat brain using lentiviral-mediated RNAi. Exp Neurol 198: 382–390 | Article | PubMed | ChemPort |
54. Harper SQ et al. (2005) RNA interference improves motor and neuropathological abnormalities in a Huntington's disease mouse model. Proc Natl Acad Sci USA 102: 5820–5825 | Article | PubMed | ChemPort |
55. Rodriguez-Lebron E et al. (2005) Intrastriatal rAAV-mediated delivery of anti-huntingtin shRNAs induces partial reversal of disease progression in R6/1 Huntington's disease transgenic mice. Mol Ther 12: 618–633 | Article | PubMed | ISI | ChemPort |
56. Machida Y et al. (2006) rAAV-mediated shRNA ameliorated neuropathology in Huntington disease model mouse. Biochem Biophys Res Commun 343: 190–197 | Article | PubMed | ChemPort |
57. Wang YL et al. (2005) Clinico-pathological rescue of a model mouse of Huntington's disease by siRNA. Neurosci Res 53: 241–249 |