RNA Processing Defects Associated with Diseases of the Motor Neuron

Abstract

Rapid progress in the discovery of motor neuron disease genes in amyotrophic lateral sclerosis, the spinal muscular atrophies, hereditary motor neuropathies, and lethal congenital contracture syndromes is providing new perspectives and insights into the molecular pathogenesis of the motor neuron. Motor neuron disease genes are often expressed throughout the body with essential functions in all cells. A survey of these functions indicates that motor neurons are uniquely sensitive to perturbations in RNA processing pathways dependent upon the interaction of specific RNAs with specific RNA-binding proteins, which presumably result in aberrant formation and function of ribonucleoprotein complexes. This review provides a summary of currently recognized RNA processing defects linked to human motor neuron diseases.

Introduction

The motor neuron is responsible for voluntary movement and is the target of a number of neuromuscular disorders that ultimately result in weakness. An appreciation of the selective vulnerability of the motor neuron in diseases characterized by weakness has existed for over 150 years, and the spectrum of motor neuron diseases covers a wide range of clinical severities and patterns. Spinal muscular atrophy (SMA), the most common genetic cause of infant mortality, is also the most common genetic form of motor neuron disease and is characterized by progressive proximal weakness and loss of lower motor neurons¹¹⁰. Clinically, progression is quite variable, and severity in SMA ranges from respiratory failure and early death in infancy to mild proximal weakness in adulthood¹¹⁵. Lethal congenital contracture syndrome (LCCM) and lethal arthrogryposis with anterior horn cell disease (LAAHD) are neurogenic forms of arthrogryposis congenita multiplex that are characterized pathologically by amyotrophy and a selective loss of lower motor neurons^141,47. Hereditary motor neuropathies (HMNs, also known as distal SMA or Charcot-Marie-Tooth disease type 2 subtypes) are a heterogeneous group of syndromes that typically result in weakness in the distal extremities and include demyelinating and axonal forms²⁶. Familial amyotrophic lateral sclerosis (fALS) is characterized by progressive loss of upper and lower motor neurons that occurs in adolescence or adulthood and leads to respiratory failure and death¹³⁸. While the clinical spectrum and course of these diseases is broad, they all share a common pathology: primary loss of the lower motor neuron that bridges the central nervous system to the peripheral musculature. What is the mechanism of motor neuron cell death in these diseases? What is the basis of motor neuron selectivity in these diseases? What explanations exist to account for the large number of genes that specifically affect the motor neuron?

The molecular genetics of motor neuron diseases is providing welcome new perspectives and insights into the molecular basis of motor neuron death and dysfunction by providing pieces to the puzzle of motor neuron-selective vulnerability. To date, there are 18 distinct genes that are mutated in primary diseases of the motor neuron (Table 1) (also see http://www.molgen.ua.ac.be/CMTMutations/Home/IPN.cfm and http://neuromuscular.wustl.edu/synmot.html). Some generalities may be gleaned from this list. For example, genes essential for axonal transport (DCTN1, NEFL) and oxidative cell stress response (SOD1) point to these pathways as critical for motor neuron cell survival.

Table 1.

Motor neuron disease genes. Mutations in these genes result in motor neuron disease syndromes. SBMA=spinal and bulbar muscular atrophy, ALS=amyotrophic lateral sclerosis, SMA=spinal muscular atrophy, CMT=Charcot-Marie-Tooth disease, SMARD=spinal muscular atrophy with respiratory distress, HMN=hereditary motor neuropathy, dSMA=distal spinal muscular atrophy, LCCS=lethal congenital contracture syndrome.

Gene	Symbol	Syndrome	Year	Refs
Androgen Receptor	AR	SBMA (X-Linked)	1991	73
Superoxide dismutase 1	SOD1	ALS1	1993	113
Survival Motor Neuron	SMN	SMA	1995	76
Neurofilament Protein, Light Polypeptide	NEFL	CMT2E	2000	88
Immunoglobulin Mu Binding Protein 2	IGHMBP2	SMARD1/HMN6	2001	39
Dynactin 1	DCTN1	HMN7B	2003	111
		ALS susceptibility	2004	95
αHeat Shock Protein 27-kD	HSPB1	CMT1F	2004	29
Glycyl tRNA Synthetase	GARS	CMT2D, dSMA5	2003	6
Vesicle-Associated Membrane Protein-Associated Protein B	VAPB	ALS8	2004	101
Heat Shock Protein 22-kD	HSPB8	CMT-2L/ HMN2A	2004	134
		CMT-2F/ HMN2B	2005	54
Seipin	BSCL2	Silver Spastic Paraplegia/HMN5	2004	147
Senataxin	SETX	ALS4	2004	18
Tyrosyl tRNA Synthetase	YARS	CMT dominant/intermediate type C	2006	57
Pleckstrin Homology Domain-Containing Protein, Family G, Member 4	PLEKHG5	dSMA4 (Lower Motor Neuron Syndrome)	2007	85
Tyrosine Kinase-Type Cell Surface Receptor HER3	HER3	LCCS2	2007	99
Phosphatidyl Inositol-4-Phosphate 5-Kinase, Type I, Gamma	PIP5K1C	LCCS3	2007	98
TAR DNA Binding Protein, 43-kD	TDP-43	ALS10	2008	124
GLE1	GLE1	LCCS1	2008	102
Fused in Sarcoma	FUS	ALS6	2009	72,139

This review will focus on motor neuron disease genes that appear to play a role in RNA processing pathways (Table 2). While there is no one cellular pathway that is exclusively responsible for these syndromes, it is instructive to place the ultimate molecular pathology responsible for motor neuron loss in this class of disease, including sporadic ALS, into the context of ubiquitous RNA processing events. We will begin with a brief review of RNA processing pathways.

Table 2.

Motor Neuron Disease Genes Involved in RNP formation

Gene	RNA processing pathway
SMN	UsnRNP assembly, RNP assembly
SETX	Transcription, processing
TDP-43	Splicing
FUS	Splicing
GLE1	mRNA transport
IGHMBP2	Translation
GARS/YARS	Amino-acyl tRNA assembly
HSPB1	AU-rich element mediated mRNA decay

Transcription to Translation

RNA is never naked within the cell. From the moment it is transcribed until it is degraded, RNA is bound by protein and assembled into ribonucleoprotein (RNP) complexes. The RNA-associated proteins in these complexes insure that each RNA transcript is properly spliced, processed, transported from the nucleus to the cytoplasm and translated by the protein synthesis machinery of the cell. The stages of this intricately choreographed process are complex and incompletely understood, although much has been learned over the past decade about the major pathways involved.

RNA processing begins while RNA polymerase II is in the process of transcribing each protein coding gene. The first step is the addition of an inverted guanosine to the 5’ end and its methylation to create a ‘cap’ that marks the beginning of the mRNA. Proteins binding to the cap structure coordinate subsequent steps in pre-mRNA processing, export, translation and decay. In pre-mRNA, the protein-coding sequences (exons) are interrupted by numerous non-coding sequences (introns) that are removed in the nucleus by pre-mRNA splicing. Several hundred proteins and some noncoding RNAs participate in the overall splicing reaction. Moreover, the inclusion or exclusion of selected exonic sequences in specific cells at specific developmental stages and in response to specific extracellular stimuli generates multiple mRNA transcripts from the same pre-mRNA sequence via alternative-splicing pathways. Perhaps nowhere is this more prevalent than in neurons, and this accounts for much of the remarkable variability in gene expression seen in the nervous system. The role of RNA splicing in neurological diseases is well recognized^36,78.

The pre-mRNA splicing process is essential for eukaryotic gene expression and is carried out by the nuclear spliceosome. The major components of the spliceosome are the U1, U2, U5, and U4/U6 small nuclear ribonucleoproteins (snRNPs), each of which is composed of one snRNA molecule, a set of seven common proteins, and several proteins that are specific to individual snRNAs^81,59,146. A critical aspect of snRNP biogenesis is that their assembly is highly regulated and requires both ATP hydrolysis and dedicated assembly factors^87,108. With the exception of U6, which never leaves the nucleus, the newly transcribed U snRNAs are initially exported to the cytoplasm, where the major assembly process of the snRNPs occurs. The common proteins, called Sm proteins, B/B’, D1, D2, D3, E, F, and G are arranged into a stable heptameric ring, the Sm core, on a highly conserved, uridine-rich sequence motif, the Sm site, of the snRNAs^59,1,125. The 5’ ends of snRNAs begin with the same 7-methyl guanosine cap structure as mRNA; however, once assembled, it undergoes additional methylation to a 2,2,7-trimethyl guanosine cap, which signals the re-import of the properly assembled and modified snRNPs into the nucleus where additional snRNP-specific proteins associate to form fully functional snRNPs^{83,30,46,31,84,146}. Mature snRNPs then carry out the process of pre-mRNA splicing. Splicosomal U snRNP biogenesis occurs in the cytoplasm and is mediated by the SMN (survival motor neuron) complex. The SMN complex directly recognizes and binds to both the protein and the RNA components of the RNP and facilitates their interaction, thereby ensuring a strict specificity of the snRNP assembly process (reviewed in^150,68).

Transcribing RNA polymerase continues past the end of the gene, and downstream sequences are removed by a precise processing event that involves assembly of a large protein complex guided by the conserved AAUAAA sequence and GU-rich sequences downstream of the processing site. This multiprotein complex contains an endonuclease that cleaves the pre-mRNA to create the 3’ end and poly(A) polymerase, which adds the poly(A) tail, a homopolymer of ~250 adenosine residues. These latter steps are coordinated by interaction between a splicing factor bound to the last intron of the pre-mRNA and poly(A) polymerase.

At this point the capped, spliced and polyadenylated mRNA is assembled into an mRNP complex that is competent for export from the nucleus to the cytoplasm. As with snRNP assembly, the export process is a highly coordinated event involving the formation and disruption of specific RNA-protein complexes and intermediaries. The objective in this case is to move mRNPs through nuclear pore complexes (NPCs), large protein assemblies that offer the only channel between the nucleus and cytoplasm^93,20,127. This transport requires that another protein complex, the Mex67:Mtr2 heterodimer, binds to the mRNP prior to transport¹²⁷. Once transported to the cytoplasmic side of the NPC, Mex67:Mtr2 is removed, which prevents retrograde transport back into the nucleus. This removal is catalyzed by the DEAD-box RNA helicase Dbp5. Dbp5 shuttles between the nucleus and cytoplasm, but it is on the cytoplasmic side of the nuclear pore complex, where it binds to and is activated by Gle1. The associated loss of Mex67:Mtr2 from the mRNP allows this to then complete transit through the NPC into the cytoplasm. DEAD-box helicases use the energy in ATP to unwind RNP complexes, and maximum ATPase activity of Dbp5 is achieved when it associates with InsP6, a molecule which facilitates and regulates the interaction between Gle1 and Dbp5¹⁵¹.

mRNA faces numerous potential fates after it arrives in the cytoplasm. It is currently thought that coincident with export, the mRNP undergoes a surveillance process where it is scanned by a ‘pioneer round’ of translation that determines whether the transcript will encode a full-length protein. This is an important, evolutionarily conserved process that protects the cell from the accumulation of C-terminally truncated proteins with potential dominant negative function that may arise from alternative splicing, processing errors or certain inherited mutations, all of which generate mRNAs with a termination codon upstream of the correct position. Splicing deposits a complex of proteins at each exon junction (the exon junction complex, EJC) that accompany the newly processed mRNP into the cytoplasm, and these provide positional information for the location of a stop codon. On a correct mRNA, ribosomes will scan through these. The ribosomes remove each of the complexes in the process until they reach the normal stop codon and are released. However, if the mRNA carries a termination codon that is upstream of an EJC, the ribosome will pause at that position, and this activates the degradation of that mRNA through a process termed ‘nonsense-mediated mRNA decay’ or NMD. NMD is ultimately responsible for the loss of protein product in many genetic disorders that result from the inheritance of one or more genes that carry a premature termination codon.

Once through the surveillance process, the mRNA can be bound by initiation factors and translated, silenced (for example by binding of microRNAs), stored or degraded. Each of these processes in itself is highly regulated, and the fate of each mRNA is determined by cis-acting sequence elements within the mRNA and by proteins (and microRNAs) that bind to these elements. A case in point is the large family of mRNAs that carry adenosine plus uridine-rich sequence elements (AU-rich elements, or AREs) in their 3’ untranslated regions. There are over 3000 mRNAs with AREs⁸, and their differential binding by a number of ARE-binding proteins (including HSPB1¹²¹) determine their overall accumulation, translation and decay. The role of mRNA stability in neurons is increasingly recognized and important in neuronal development and regeneration¹².

mRNA translation is mediated by the ribosome and is conventionally divided into the initiation, elongation and termination stages. During initiation, the ribosome and the tRNA for the first amino acid residue, methionine, recognize and bind to the first codon, AUG, of the mRNA transcript. The polypeptide encoded by the mRNA is assembled during the elongation phase, and termination occurs when the ribosome encounters a ‘stop’ codon that results in release of the completed polypeptide and dissociation of the ribosomal subunits. The entire process is, once again, mediated by the complex association of a host of specific RNP complexes, non-coding RNAs, and proteins.

If one considers the ribosome to be a protein-synthesizing machine, the raw materials are the mRNA template and the individual aminoacyl-tRNAs. Aminoacyl-tRNAs are covalently linked RNA-protein complexes that are produced by a number of aminoacyl-tRNA synthetases (ARSs) that work to catalyze the esterification of specific amino acids to their respective tRNAs. A number of additional functions for particular ARSs have recently come to light, including roles in transcription, splicing, apoptosis, inflammation and angiogenesis (reviewed in^67,104,105).

Ultimately all mRNAs are degraded, and their rate of decay is a controlling factor in the timing, quantity and perhaps even location of their encoded proteins. In general, mRNA decay begins with shortening of the poly(A) tail, followed by removal of the 5’ cap and degradation by exonucleases from both the 5’ and 3’ ends of the mRNA⁹⁶. This is a highly regulated process that involves the interplay between cis-acting sequences and the proteins that bind to them. Both translation and decay are targets of numerous signal transduction pathways, and we are just beginning to understand how they are controlled.

Motor neuron disease genes

A large number of known motor neuron disease genes are directly involved in RNA processing (Figure 1). While many of these genes theoretically impact upon all mRNAs as they are processed, this is not proven. In neurons, there may be unique susceptibility for distinct groups of mRNAs to mutations in the following genes (Table 2).

Figure 1.

RNA processing pathways involved in motor neuron disease.

Abbreviations: snRNA, small nuclear ribonucleic acid; ATP, adenosine triphosphosphate; mRNP, messenger ribonucleoprotein; InsP6, inositol hexakisphosphate 6; UsnRNP, uridine-rich small nuclear ribonucleoprotein; UPF1, Up Frameshift 1; hnRNP, heteronuclear ribonucleoprotein;

Survival Motor Neuron (SMN)

SMA is an autosomal recessive neurodegenerative disease that is the leading inherited cause of infant mortality. It affects 1 in 10,000 live births and has a carrier frequency of ~1 in 40^107,22. The hallmark features of SMA are progressive weakness, dysphagia, dyspnea and death and are the result of a selective loss of motor neurons at an early age^28,51,131. Clinically, SMA is classified based on age of onset and severity defined by major motor milestones. Children with the most severe type I (Werdnig-Hoffman) are never able to sit unassisted and usually die of respiratory insufficiency by age 2 years. Children with the intermediate type II are never able to stand unassisted and have a variable survival. The mild type III (Kugelberg-Welander) is characterized by onset of weakness after the age of 18 months, and survival is prolonged well into adulthood. However these patients often become wheelchair bound during youth or adulthood. There is currently no treatment for SMA.

SMA is caused by mutation of both alleles of the survival of motor neurons 1 (SMN1) gene⁷⁶, however all patients retain at least one copy of the SMN2 gene. The SMN1 gene produces full-length SMN mRNA, whereas the SMN2 gene produces full-length mRNA and mRNA lacking exon 7 (SMNΔ7), as well as small amounts of mRNA lacking exon 5 or exon 3 or combinations thereof^80,92. SMNΔ7 mRNA encodes an unstable, truncated protein⁷⁹. SMA is postulated to result from insufficient expression levels of full-length SMN protein in motor neurons, perhaps during a critical stage of motor neuron development, however, the exact mechanism by which SMN mutations result in SMA is unknown^86,114,34. Several studies have shown that SMN mRNA and protein are reduced in cell lines and tissues derived from type I SMA patients compared to controls^{21,77,37,64,123}.

The capacity of cells to assemble small nuclear ribonucleoproteins (snRNPs) is reduced when the expression of the SMN protein is reduced in cell culture, SMA patient cells and in SMA mouse model tissues^{150,143,34,68,153}. Thus, it appears that defects in the formation of specific RNPs can result in motor neuron disease¹⁰⁹. The consequences of decreased snRNP assembly include alterations of the relative expression of spliceosomal U snRNAs that subsequently alter specific mRNA splicing events^34,153. The precise alterations in the splicing of mRNAs that occur in motor neurons in SMA are not yet known. It is also unclear whether these alterations cause motor neuron death in SMA, and why the motor neuron is so vulnerable to these changes. One can imagine that nearly all cellular processes are vulnerable to splicing changes. For example, there is evidence that structural abnormalities at the neuromuscular junction precede the loss of motor neuron cell bodies in a mouse model of SMA⁶². One can speculate that genes involved in neuromuscular junction integrity or formation may be improperly spliced or otherwise processed as a result of reduced SMN levels.

Senataxin (SETX)

Several missense mutations in the senataxin gene (SETX) (L389S, R2136H, and T3I) cause a rare autosomal dominant form of juvenile ALS (ALS4) characterized by distal muscle weakness and atrophy with pyramidal signs, in which sensation remains unaffected¹⁸. Other mutations in SETX (L1976R, L1977F, N603D–Q653K) have been shown to cause oculomotor apraxia 2 (AOA2), an autosomal recessive cerebellar ataxia characterized by cerebellar atrophy, oculomotor apraxia, early loss of reflexes, late peripheral neuropathy, and slow progression leading to severe disability^94,32,10.

The senataxin protein is a member of the UPF1-like helicase within the DNA/RNA helicase superfamily. It contains 7 helicase motifs and is thought to unwind both DNA and RNA substrates^18,45. While little is known about the function of senataxin in motor neurons, Sen1p, the senataxin ortholog in Saccharomyces cerevisiae, has been well studied. Mutations in the helicase domain of Sen1p alter the cellular abundance of many diverse RNA species including intron-containing tRNA precursors, ribosomal RNA precursors and certain small nucleolar RNAs and snRNAs^{148,136,112,126}. In yeast, Sen1p plays a role in the normal formation of the U5 spliceosomal snRNA¹³⁷. It will be interesting to compare the changes in U5 snRNA abundance that result from SETX mutations to the U5 snRNA abundance changes that are now being reported as a result of reduced SMN expression.

TAR DNA binding protein (TDP-43)

TDP-43 was initially shown to be the ubiquitinated protein that comprises the ubiquitin-positive inclusions found in pathologic specimens from individuals with sporadic ALS and frontotemporal lobar degeneration with ubiquitin-positive inclusions (FTLD-U)¹⁰⁰. In one study, immunohistochemistry performed on postmortem spinal cord from individuals with sporadic ALS and familial ALS with and without the SOD1 mutation revealed TDP-43 positive inclusions in all of 59 sporadic and 11 SOD1-negative familial ALS specimens⁸². TDP-43 positive inclusions were NOT detected in 15 SOD1-positive familial ALS specimens⁸². A causal role of TDP-43 in ALS pathogenesis has been demonstrated by recent reports of TDP-43 mutations in familial ALS patients^{58,69,116,124,38}.

TDP-43 is an RNA-binding protein that contains two RNA recognition motifs (RRMs) that are similar to those found in well-characterized heteronuclear RNPs such as hnRNP A1 and poly A binding protein (PABP)^103,14,15. TDP-43 binds to (UG)n-repeated sequences near the 3’ splice site of the cystic fibrosis transmembrane conductance regulator (CFTR) exon 9 and, and its binding inhibits the inclusion of this exon¹⁴. TDP-43 also binds to UG dinucleotide repeats at the 3’ splice site of apolipoprotein A-II intron 3 and intron 2 of the human cardiac Na+-Ca2+ exchanger^35,7. The C-terminal domain of TDP-43 binds to several splicing inhibitors including hnRNP A1 and hnRNP A2/B1¹⁶. Thus, TDP-43 has splicing inhibitor activity and appears to form a molecular anchor to recruit hnRNP proteins at specific pre-mRNA dinucleotide sequences in order to effect splicing at that site.

Most TDP-43 mutations described in fALS (a mutation in RRM1) are found in the C-terminal domain of the protein. They are predicted to disrupt hnRNP binding, which would confer a functional consequence of reduced splicing inhibitory activity^58,124. fALS-causing TDP-43 mutations alter the subcellular distribution of the protein from the nucleus to the cytoplasm, and this may be a result of protein aggregation of the mutant protein that precludes import into the cytoplasm¹²⁴. Thus, the mutations would interfere with TDP-43 mediated splicing activity by directly decreasing the available amount of active protein. While these mutations are rare, the global presence of TDP-43 in sporadic ALS inclusions suggests that, in the case of sporadic ALS, TDP-43 mediated splicing inhibitory activity may be reduced secondarily via excessive ubiquitination and sequestration.

Fused in Sarcoma (FUS)

Mutations in FUS have been discovered in three families with fALS^72,139. In at least one of the reported families, the onset of weakness was in the proximal upper extremities with no bulbar involvement⁷². The FUS gene encodes a protein that contains three RGG repeat domains, an RNA-recognition motif (RRM) and a zinc-finger domain⁵². FUS associates with hnRNP A1 and C1/C2, suggesting a role in RNA splicing¹⁵⁴. FUS also binds to the actin-associated molecular motors KIF5 and Myosin Va, and colocalizes with RNA-transport granules^60,152.

A molecular consequence of fALS-causing FUS mutations is that FUS protein, which is predominantly localized in the nucleus, is redistributed and found within inclusions in the cytoplasm^72,139. Thus, FUS and TDP-43 share many of the same molecular properties (interaction with RNA and splicing factors) and are similarly altered when they are mutated (nuclear to cytoplasmic redistribution), resulting in fALS. The molecular overlap of these proteins in motor neuron disease may be further demonstrated should the FUS protein be found to co-localize with TDP-43 positive inclusions found in sporadic ALS patients.

GLE1

Mutations in GLE1 can lead to forms of lethal congenital contracture syndrome (LCCS), an autosomal recessive motor neuropathy characterized by early fetal hydrops, akinesia, agnathia, loss of anterior horn neurons, and various other features, which result in prenatal or immediately postnatal death¹⁰². LCCS has been divided into types 1, 2, and 3 based upon specific clinical criteria, and they are caused by mutations in GLE1, HER3, and PIP5K1C, respectively^98,99,102. Remarkably, the latter two proteins are required in the phosphatidyl inositol pathway. Both are involved in the synthesis of inositol hexakisphosphate (InsP6), which binds directly to the yeast homolog of GLE1.

GLE1 encodes the protein Gle1 which is a required mediator of mRNA export. Gle1 is found with the protein Dbp5 on the cytoplasmic side of nuclear pore complexes (NPCs) along nuclear pore filaments^5,145, where it activates DExD/H-box protein 5 (Dbp5) to affect mRNA export from the nucleus. Dbp5 is an ATP-dependent RNA helicase, although it is too inefficient alone to be considered active under biological conditions. However, binding of activated Gle1 (via the InsP6 pathway) increases the catalytic efficiency of Dbp5 by 27-fold¹⁴⁵.

Immunoglobulin µ-binding protein 2 (IGHMBP2)

Mutations in the gene that encodes immunoglobulin µ-binding protein 2 (IGHMBP2) have been demonstrated to cause distal spinal muscular atrophy with respiratory distress type 1 (SMARD1)^39,42. This severe form of SMA results in diaphragmatic paralysis with weakness in the upper limbs and distal muscles. The pattern of phenotypic features is somewhat variable with some involvement of sensory and autonomic nerves in some confirmed cases^40,110.

IGHMBP2 shares a 42% sequence homology with SETX and contains DExxQ-type helicase/ATPase domains that are conserved among DNA replication, repair, and recombination proteins, and a nuclear localization signal in the protein’s C-terminal region^89,91. IGHMBP2 immunoreactivity occurs in the cytosol and in axons of neuronal cells⁴¹. Recombinant IGHMBP2 directly interacts with ribosomes. It hydrolyzes ATP and is capable of unwinding duplex RNA or DNA in vitro⁴³. The functional consequences of each currently known disease-causing mutation are a loss of ATPase activity with resultant helicase function reduction or direct helicase impairment⁴³. One of the IGHMBP2 mutations, T493I, does not affect ATPase or helicase activity in vitro, however this mutation may lead to decreased stability of the protein. It produces a lower expression level, as has been demonstrated in DSMA1 patient cell lines⁴⁴.

Aminoacyl-tRNA synthetase family

Aminoacyl-tRNA synthetases attach amino acids to their cognate tRNAs. Mutations in a number of these enzymes result in neurological diseases, including motor neuron disease. Glycyl-tRNA synthetase (GARS) mutations result in CMT2D, an axonal form of CMT with upper limb predominance, and dHMNV, a very similar disease that lacks the distal sensory loss of CMT2D^{6,122,23,55,118,25}. Tyrosyl-tRNA synthetase (YARS) mutations cause CMT-dominant/intermediate type C, including G41R in a North American family, E196K in a Bulgarian family, and a 12-bp in-frame deletion (153–156delVKQV) in a Bulgarian individual⁵⁷.

Aminoacyl-tRNA synthetases are comprised of 3 domains: a WHEP-TRS domain that conjugates with other ARSs, a core catalytic domain for ligation, and an anticodon-binding domain to recognize the correct tRNAs³³. To become active, the GARS protein (GlyRS) forms a homodimer.¹⁴⁹. Mutations that affect the strength of the GlyRS dimer interaction also lead to GlyRS mislocalization within neurons^97,149. Building upon the findings that the related tyrosil ARS (TyrRS) localizes to sprouting neurites⁵⁷, researchers found that GlyRS also localizes to neurites, but mutant forms do so poorly⁹⁷. These dimer-localization studies suggest a potential role for a subset of tRNA synthetases in neuron development. In support of this hypothesis, a recent study in Drosophila has shown that the GARS ortholog (gars) is necessary for proper neuron outgrowth¹⁹. Additionally, mutant flies could be rescued with transgenic human GARS, but outgrowth improved to lesser degrees when flies were given two different human CMT2D mutant forms of GARS.

Aminoacyl-tRNA synthetases may also play a role in splicing. The fungal protein homologous to TyrRS in the mitochondria of Neurospora crassa (mt TryRS, encoded by cyt-18) is required not only for tyrosyl-tRNA synthesis, but also for efficient splicing of group I introns in vitro and in vivo^4,117,90,106. TyrRS may also be secreted and have cytokine properties under conditions of cell stress¹⁴².

Mutations in this family of enzymes are recognized in many human diseases of the nervous system, autoimmune disease and cancer¹⁰⁵. For example, mitochondrial aspartyl-tRNA synthetase mutation resulting in leukoencephalopathy with brain stem and spinal cord involvement¹¹⁹, and a missense mutation in the alanyl-ARS of mouse leads to a late onset neurological disease characterized by patchy fur and mild tremors progressing to ataxia and Purkinje cell loss (esp. in the rostral cerebellum)⁷⁵. Lastly, splicing errors in tRNA have been shown to cause pontocerebellar hypoplasia¹³.

α-crystallin type heat shock protein family

The α-crystallin type heat shock protein family (sHSPs) is comprised of ten multifunctional molecular chaperones that are induced in response to a wide variety of physiological and environmental factors and are numbered HSPB1 – HSPB10⁶¹. All sHSPs share an α-crystallin domain which is thought to mediate protein-protein interactions. Substitution mutations in three of the ten sHSP genes result in neuromuscular disease, including two (HSPB1 and HSPB8) which result in distal hereditary motor neuropathy^11,135,53,54. Mutations in HSPB1 have been shown in dominant, recessive and sporadic cases of hereditary motor neuropathy^{29,133,65,24,50,56}. In nearly every case, these mutations target positively charged arginines or lysines that are critical for the structural and functional integrity of the α-crystallin domain contained in each of these proteins and appear to alter sHSP chaperone-like activity^29,54,66. The functional consequences of these mutations as they relate to the motor neuron are not known, however, HSPB1 appears to be neuroprotective and have roles in the maintenance and regulation of the neuronal cytoskeleton and axonal transport^128,2,74.

HSPB1 plays a role in AU-rich element mediated mRNA decay¹²¹. As noted above, approximately 3000 mRNAs contain AU-rich elements in the 3’-UTR. As a group these mRNAs are inherently unstable, and it is this property that cells exploit to effect rapid changes in protein expression. Induction of an unstable mRNA results in a more rapid acquisition of a new steady state than a similar induction of a stable mRNA, and withdrawing the stimulus for this induction results in similarly rapid decrease in steady state, with matching decrease in protein production. While inflammatory cytokines are the mRNAs that are most commonly studied, they represent only a fraction of the mRNAs that contain AU-rich elements. The turnover of many of these mRNAs is controlled by a complex of RNA-binding proteins, many of which bind selectively to the different AU-rich elements. One of these, AUF1 (hnRNP D) is noteworthy, because differentially spliced isoforms act either to stabilize or destabilize these mRNAs. Recently, HSPB1 was identified as a constituent of a destabilizing complex with AUF1, Hsp70 and Hsc70¹²¹. HSPB1 binds to AU-rich elements, and its knockdown stabilizes a reporter mRNA that contains this destabilizing sequence. These findings raise the intriguing possibility that mutations in HSPB1 (and perhaps other members of this protein family) might result in motor neuron-specific changes in the stability of specific mRNAs.

Discussion/Convergence

The clustering of recognized motor neuron disease genes in various RNA processing pathways suggests that much more needs to be done to understand the mechanisms of RNA processing and metabolism as they relate to motor neurons (and associated cell types). The riddle of motor neuron selectivity, when one of a handful of RNA processing genes are mutated, suggests that a key to understanding sporadic ALS may be to focus on a characterization of the unique sets of mRNA, splicing isoforms, and the expansive repertoire of non-coding RNAs that a normal motor neuron is expected to produce throughout its lifetime. Defects in RNA processing may be subtle and have consequences that relate to other, relevant pathogenic pathways. For example, defects in the RNA editing pathway of GluR2 pre-mRNA have been described in the spinal cords of sporadic ALS patients which results in altered calcium permeability of the resultant AMPA receptor, presumably resulting in increased susceptibility to glutamate excitotoxicity^{129,63,71,130}. Ultimately, a key concept going forward is that motor neurons (and motor neuron-specific RNA expression patterns) appear to be susceptible to errant ribonucleoprotein formations.

The critical function of axonal transport mechanisms in motor neuron disease and neuropathies has been well-established^49,27. Mutations in DCTN1, which encodes dynactin, the principle protein responsible for dynein-mediated movement of organelles along microtubules, have been associated with a form of fALS^111,95. Two isoforms of dynactin exist in neuronal tissue, created by alternative splicing of the mRNA. While both isoforms bind dynein in the cytoplasm, only the p150 isoform binds microtubules, suggesting that the p135 isoform is functionally distinct⁴⁸. It is conceivable that primary errors in the expression of an RNA processing gene or of a non-coding RNA could produce a similar motor neuron disease by secondarily altering the alternative splicing of dynactin.

Dynactin may also be considered an important protein in the processing of RNAs in neuronal processes. Dynactin directly mediates the transport of RNAs along microtubules, in turn increasing the overall minus-end motility of RNA transport¹⁴⁰. There is also evidence that the seipin protein, which is encoded by the BSCL2 gene and mutated in yet another HMN, may be involved in RNA transport^3,147. Thus, a distinction between incorrect processing of RNAs destined to encode proteins essential for axonal transport and the incorrect transport of RNAs destined to travel along the axon may be a “chicken-or-egg” riddle that is not mutually exclusive in motor neuron disease.

It is well known that mutations in SOD1 result in fALS¹¹³. Oxidative damage is an established contributor to motor neuron pathology in ALS⁹. Remarkably, direct mRNA oxidation is also a common feature in ALS that may occur early in disease pathogenesis¹⁷. Moreover, mRNAs that encode genes linked to other motor neuron disease genes, such as SOD1 and DCTN1, were highly oxidized relative to other genes¹⁷. Since the oxidation of mRNA clearly interferes with subsequent processing steps^120,132, a convergence of errors in oxidative stress buffering capacity and RNA processing errors must also be considered in our riddle. For example, mutant SOD1 but not wild-type SOD1 interacts with lysyl-tRNA synthetase and translocation-associated protein δ, suggesting that SOD1 mutations have the ability to affect RNA processing by binding to and altering the normal function of key proteins⁷⁰. And it is, perhaps, no coincidence that oxidative stress inhibits the assembly of spliceosomal UsnRNPs by decreasing the activity of the SMN complex¹⁴⁴.

The production of mature proteins from the human genome is an extremely complex and wonderfully organized process that, at every step, presents an opportunity for error. Given the added complexity of gene expression required within the nervous system, which must respond to connectivity and neuronal activity cues, it is not surprising that primary disruptions of RNA processing might prove deleterious to specific populations of neurons. How RNA processing defects result in the selective loss of motor neurons is an important area of focus to gain another foothold in our understanding of the pathogenesis of motor neuron disease, including sporadic ALS. The emerging importance of RNA processing in motor neuron disease calls for: 1) the identification of alternative splicing events that are pathogenic in MND, 2) characterization of essential non-coding RNAs in the cells of the motor unit (motor neuron, glial cells and muscle), 3) characterization of the full complement of mature mRNAs and RNPs in the motor unit, 4) an understanding of RNA processing events in the axoplasm, 5) an inclusive approach to motor neuron disease gene pathogenesis with an appreciation for molecular pathway convergence and 6) facilitated collaboration between RNA biologists and motor neuron disease investigators.