(1)
Asheville, NC, USA
Frontotemporal lobar degeneration (FTLD) is the most common cause of dementia in individuals under the age of 60 years, and overall, follows Alzheimer’s disease and dementia with Lewy bodies in its frequency of occurrence in the industrialized world. This term encompass a number of syndromes that affect neurons in the frontal and temporal lobes of the brain. These disorders are characterized by deficits in behavior, mood, emotion, or language. The most common form of the disease is frontotemporal dementia (FTD). This form includes three subtypes—behavioral variant FTD (bvFTD), semantic dementia (SD), also known as temporal variant FTD, which affects the ability to assign meaning to words, and progressive non-fluent aphasia (PNFA) that affects the ability to speak. In some instances of FTD, motor neuron disease (MND) is present, as well. In those situations, the disorder is known as FTD-MND. Two other neurological disorders, progressive supranuclear palsy (PSP) and corticobasal syndrome (CBS), complete the enumeration of FTLD disorders. These disorders have a prevalence of 15 per 100,000 with incidences ranging from 2 to 9 per 100,000 per year depending on the age of the individual.
Although it is less well known to the general public than Alzheimer’s disease, FTLD has been known and studied for just as long. Arnold Pick (1851–1924) described the disease in 1892, noting the presence of protein tangles, which appeared as large bodies in the cells he had examined. Those deposits along with extracellular plaque s and abnormal neurites were studied by Alois Alzheimer, Oskar Fischer (1876–1942), Frederic Lewy and Pick over the next 20 years. The deposits described by Pick are known as Pick bodies. They, like the tangles discussed earlier in the chapter on Alzheimer’s disease, are composed mainly of the microtubule protein tau. Since that time an expanded list of disorders characterized by abnormal deposits of tau have been identified, and are known as the tauopathies .
Amyotrophic lateral sclerosis (ALS) is referred to as Lou Gehrig’s disease in the USA and as Charcot’s disease elsewhere. It is an adult-onset, fatal disorder producing paralysis and primarily affecting adults 45–60 years old. Its progression, typically 3–4 years, is far more rapid than Parkinson’s disease, which has a 20 year progression. Its prevalence in the USA is roughly 30,000 individuals with an annual incidence of about 6000. Most cases are sporadic (sALS), but approximately 10 % are inherited (fALS). A landmark finding was the discovery in 1993 by Rosen that missense mutations in the Cu/Zn superoxide dismutase gene, SOD1, were associated with fALS. Shortly thereafter, in 1997, Bruijn reported that ubiquitin-positive inclusions containing ALS-linked mutated SOD1 gene products appear first in astrocytes and then in motor neurons and these inclusions escalate as the disease progresses.
Superoxide dismutase catalyzes the conversion of the dangerous superoxide free radical to far less-dangerous hydrogen peroxide. This enzyme had gained prominence following its initial discovery by McCord and Fridovich in 1969 when its existence was taken as evidence in support of the free radical (oxidative damage) theory of aging . This theory has as its central tenet that free radicals, small molecules containing unpaired electrons that render them highly reactive, accumulate over time in cells and attack macromolecules, thereby placing a limit on an organism’s life span. The groundwork for this theory was laid by several discoveries. The first of these was the groundbreaking realization in 1954 by the Argentinian biochemist Rebeca Gerschman (1903–1986) that potentially damaging free radicals are present in cells. She noted that free radicals produced as by-products of oxidative metabolism caused damage to cells in a manner similar to the actions of ionizing radiation such as X-rays. In the same year, Commoner, Townsend, and Pake using an electron-spin resonance spectrometer, which can detect the presence of unpaired electrons, found that reactive oxygen species (ROS) were present in biological materials.
The discovery that free radicals can damage lipids, DNA, and proteins led Denham Harman to advance the free radical theory of aging in a highly influential paper in 1956. The basic idea is that free radicals are continually generated by the mitochondrial electron transport chain (ETC) as a byproduct during cellular respiration. These accumulate over time and induce cellular damage that is responsible for aging . The most prominent of the free radicals is superoxide (O2 •). The primary sources of superoxide production from the viewpoint of aging theory are ETC complexes I and III. Small amounts of superoxide are released into the mitochondrial matrix from Complex I and into both the matrix and intermembrane space from Complex III. Under healthy cellular conditions, when the ETC is operating normally and is not excessively stressed, production of superoxide is low and is readily converted to a far less dangerous molecule, hydrogen peroxide (H2O2), by resident SODs. That changes over time and under stressed conditions. Considerable effort was expended, once SOD1’s involvement in ALS emerged, in exploring this line of reasoning with results that were disappointing and perplexing.
This situation began to evolve in the 2006–2009 time period when a series of discoveries radically changed the conceptual landscape of ALS and FTLD. The discovery in 2006–2007 that the tar-binding protein 43 (TDP-43), a 43-kDa RNA processing protein, was a main component of the ubiquitinated intracellular inclusion s seen in the majority of ALS cases began the shift in understanding. These inclusions were found in glial cells and neurons, in many instances of sALS, and also in a large number of cases of FTLD. These studies were followed by several others that established unambiguously that mutated forms of TDP-43 could produce ALS. These findings were followed in 2009 with the discovery that another RNA processing protein, fused in sarcoma (FUS), was present in a different set of inclusions, and that mutated forms of that protein too has a causal connection to ALS. As shown in Table 11.1 mutated forms of these two proteins can not only generate ALS but also produce FTLD, thereby linking the two very different disorders.
Table 11.1
Amyotrophic lateral sclerosis and frontotemporal lobar degeneration genetics and pathology
fALS (%) | sALS (%) | FTLD (%) | Protein | Gene (locus) | Function | Inclusion |
---|---|---|---|---|---|---|
20 | 2 | Superoxide dismutase | SOD1 (21q22.1) | Detox enzyme | SOD1 | |
5 | <1 | <1 | Tar-binding protein 43 | TDP-43 (1p36.2) | RNA processing | TDP-43 |
5 | <1 | <1 | Fused in sarcoma | FUS (16p11.2) | RNA processing | FUS |
<1 | Angiogenin | ANG (14q11.2) | RNA processing | TDP-43 | ||
<1 | Ataxin 2 | ATXN2 (12q24) | RNA processing | TDP-43 | ||
<1 | <1 | Valosin-containing protein | VCP (9p13.3) | Protein quality control | TDP-43 | |
5 | <1 | Optineurin | OPTN (10p15-p14) | Protein quality control | TDP-43 | |
<1 | <1 | Ubiquilin 2 | UBQLN2 (Xp11.21) | Protein quality control | TDP-43, FUS | |
40 | 5 | 10 | C9orf72 | C9ORF72 (9p21.3-p13.3) | (unknown) | TDP-43 |
10 | Progranulin | PGRN (17q21.31) | Protein quality control | TDP-43 | ||
<1 | <1 | Charged multivesicular protein 2B | CHMP2B (3p11.2) | Protein quality control | p62, TDP-43 | |
10 | Microtubule-associated protein tau | MAPT (17q21) | Cytoskeleton | tau |
Yet another discovery further linked together ALS and FTLD. In 2011, an unstable GGGGCC hexanucleotide repeat expansion in the noncoding region of the C9ORF72 gene was shown to cause to fALS and FTLD. This mutation was found to be the single most widespread cause of fALS and FTLD. Disease-causing mechanisms arising from unstable repeat expansions were discussed in the last chapter. These same processes come into play in C9ORF72-associated ALS and FTLD along with several new ones.
In examining the different forms of FTLD several nonoverlapping classes of inclusions can be discerned. One grouping consists of tau-positive, ubiquitin-positive inclusions (FTLD-tau), a second as TDP-43-positive, ubiquitin-positive inclusions (FTLD-TDP-43), and a third as FUS-positive, ubiquitin positive inclusions (FTLD-FUS). Fourth, some inclusions are negative for tau, TDP-43, and FUS; instead, they may contain p62 and are labelled as FTLD-UPS. Lastly, in some rare instances no discernable inclusions can be detected. That group is designated as FTLD-ni. These molecular/histological characterizations underlie the six clinical syndromes named in the first paragraph of the chapter.
In this chapter, the four main genetic constellations underlying ALS and FTLD—SOD1, TDP-43 and FUS, C9ORF72, and tau—are explored starting with SOD1 and then turning to TDP-43 and FUS, and from there to the others. Dynamic, membrane-less organelles such as stress granule s , P-bodies , and nucleoli are important loci of action in ALS and FTLD. Their formation and role in these disorders is scrutinized along with a continuation of the discussion of RNA-mismanagement from the previous chapter. Unstable repeat expansions were discussed in the last chapter with Fig. 10.1 depicting the types of problems these repeats cause. That discussion picks up again in an examination of the disease-causing GGGGCC repeat expansion in C9ORF72, and its involvement of non-ATG translation and G-quadruplex es . The chapter concludes with a look into the tauopathy landscape in which a variety of disorders have misfolded and aggregated tau as a common link. Significantly, that family of disorders includes chronic traumatic encephalopathy (CTE ) brought on by repeated concussive and sub-concussive injuries to the head suffered by military personnel and athletes involved in violent sports.
11.1 Mutations in the SOD1 Gene Cause Amyotrophic Lateral Sclerosis
Dismutases are enzymes that catalyze the dismutation of their substrates, that is, they simultaneously generate oxidized and reduced forms of a chemical species. The superoxide dismutase (SOD) family consists of three enzymes—cytoplasmic Cu/Zn-SOD (SOD1), mitochondrial Mn-SOD (SOD2), and extracellular EC-SOD (SOD3). These are regarded as the most important antioxidants in the body; they, most notably, convert superoxide anions to molecular oxygen and hydrogen peroxide. Starting with two superoxide molecules the SODs oxidize the first one and reduce the second, as discovered by McCord and Fridovich in 1969.
Later, in 1993, Rosen found that mutations in SOD1 can produce amyotrophic lateral sclerosis. More than 150 mutations in SOD1 have been uncovered to date. The mutations are situated throughout the protein and produce a variety of structural alterations that result in (1) loss of stability, (2) misfolding, and (3) aggregation. Unlike Aβ and α-synuclein the mutations do not, in general, give rise to amyloids; instead, they tend to produce amorphous aggregates. The most common mutation is the Ala4Val (A4V) mutation. It accounts for 50 % of all the fALS cases in the USA. Like most other SOD1 missense mutations it is inherited in an autosomal dominant manner.
SOD1 contains eight antiparallel β-strands, binds two metal atoms, and functions as a dimer (Fig. 11.1). Depending upon the specific mutation, one of several different kinds of functional defects can be produced. They do not, however, produce a loss-of-enzymatic, antioxidant function. Attempts to demonstrate that this was the dominant mode of failure produced negative results as noted in the introductory remarks. Instead, the mutations are widely regarded as primarily generating a gain-of-toxic-function, the specific nature of which has not yet been determined, conclusively, from the underlying loss of stability, misfolding, and loss of solubility/aggregation.
Fig. 11.1
Crystal structure of the SOD1 homodimer. The zinc and copper ions are shown in cyan and orange, respectively. The G93 mutation site and C111 oxidation site are shown as well. The two events, G93A mutation and C111 oxidation, generate the same misfolded structure (from Bosco Nat. Neurosci. 13: 1396 © 2010 Reprinted by permission from Macmillan Publishers Ltd)
Some of the mutations disrupt the folding of the apoSOD1 monomers. These mutations destabilize the caps that form at the extrema of the folded protein and protect the β-sheet structures from undesirable side reactions. As a result the proteins become “sticky.” This is similar to the pathology seen elsewhere where the protective role of edge strands becomes compromised resulting in aggregation. Other mutations prevent monomers from acquiring their metal cofactors ; these mutations have a greater destabilizing effect than mutations that leave this capability alone. Monomers lacking their Cu and Zn ions are more prone to unfold and refold into aggregation-prone conformations. Disulfide bonds are another major factor. Incorrect disulfide bonds prevent dimerization and, instead, enable multimeric aggregation to occur via destabilized monomeric folding intermediates.
There is evidence that oxidative damage can lead to misfolding and aggregation. The central observation is that fully metalized, properly disulfide linked SOD1 is extraordinarily stable. In order to dispose of the enzyme by proteasomes, it first has to be destabilized. This is accomplished through a negative feedback loop in which a buildup in hydrogen peroxide leads to a loss of metal ions as a preliminary step, but the resulting metal-free monomers are sensitive to oxidative misfolding and aggregation. The most significant aspect of this mode of failure is that it ties together SOD1-induced fALS and the far more prevalent sALS in a common pathology, and provides an explanation for the presence of SOD1 inclusions in both forms of the disease.
Lastly, the disease causing mutations in SOD1 produce varying effects on its stability. What these mutations have in common is a tendency to alter SOD1’s folding kinetics. Small single domain proteins such as SOD1 should fold fairly rapidly into their native state. In contrast to this expectation, mutant SOD1 proteins take orders of magnitude longer than expected to find their native state. The energy landscape s for these mutants are rugged and the folding pathways are populated by kinetic trap s. The formation of a variety of folding intermediates and apo monomers provides ample opportunity for the sampling of aggregation-prone conformers.
11.2 Gain-of-Toxic Function by Mutant SOD1 (mSOD1)
ALS is a disease of upper and lower motor neurons (MNs). It is widely believed (but by no means unanimously) that it begins with the destruction of the postsynaptic apparatus, followed by a dying back of axons at the neuromuscular junction (NMJ) leading to a loss first of lower motor neurons and then of upper ones. This idea is supported by a growing body of evidence typically obtained using the G93A SOD1 transgenic mouse model of ALS. Within this framework, two distinct sets of actions seem to be needed. First, the deterioration of the synapse itself, perhaps due to a single toxic event or small set of events, or alternatively, arising through multiple accumulating events that pile up and initiate the destruction of the NMJ. Secondly, inflammation occurs involving neighboring glial cells activated by extracellular mSOD1. The sustained release of inflammatory molecules by the glia, in turn, potentiates the damage to the neurons and NMJ.
Inflammation plays an important effector role in mutant SOD1-generated ALS. This conclusion is supported by several observations. First, mutant forms of SOD1 are secreted from neurons in secretory vesicles; they are detected in CSF, and are responsive to immunization protocols. Secondly, intracellular SOD1-bearing inclusions are not limited to neurons, but instead appear in nearby glia; these cells have become damaged at some point by mutant SOD1. Astrocytes exposed to mutant SOD1 are toxic to neurons. Paralleling the situation encountered in other neurological disorders, extracellular mutant SOD1 hyperactivates microglia and astrocytes, which, in turn, secrete a variety of toxic inflammatory mediators. Expanding on this notion, astrocytes taken from sporadic ALS sufferers are shown to be toxic to motor neurons. These findings all place astrocytes and microglia downstream of mutant SOD1, in the role of key effectors.
Mutant SOD1-ALS arises primarily through gain-of-toxic-function mechanisms rather than through loss-of-normal-function means. A variety of ways that this can occur have been uncovered through studies of transgenic mice expressing one or more of the predominant mutations with some variations associated with the specific mutation(s) being introduced. Arguably the best established of these disease modes are glutamate excitotoxicity and mitochondrial dysfunction. These two are familiar points of failure, having been encountered in Alzheimer’s disease and Parkinson’s disease and Huntington’s disease.
A seminal observation on the importance of glutamate handling, the first of these widely encountered modes, was made by Rothstein in 1996 who noted that proper uptake of glutamate by astrocyte -resident transporters was critical—it prevents excessive glutamate stimulation of neurons. That these transporters are negatively impacted by mutant SOD1 was established a few years later. More recently, it was found that the presynaptic terminals of the tripartite synapse regulate this activity. Breakdowns in signaling from neuron to astrocyte trigger reductions in transcription and expression of astrocyte-resident glutamate transporters such as GLT1/EAAT2 (excitatory amino acid transporter 2).
One of the earliest signs of a neurological disorder is the appearance of mitochondrial and mitochondrial network abnormalities. The assertion that mSOD1 produces a gain-of-toxic function is based on two broad sets of findings. First, there is the preservation of normal SOD1 dismutase activity in spite of the presence of disease-causing mutations. Secondly, at the same time, mSOD1 localizes to mitochondria that develop a number of striking abnormalities. These include the appearance of numerous membrane-containing vacuoles and a deformed morphology along with the emergence of electron transport chain defects and reduced Ca2+ buffering capabilities, and, at the network level, impaired mitochondrial fusion and transport.
There is also evidence that SOD1 misfolding can spread from cell to cell in a prion-like manner. Two papers published in 2011 showed that (1) templated conversion can occur, that is, aggregates composed of misfolded SOD1 can serve as seeds that template the conversion of normal SOD1 to the misfolded form, and (2) propagation can occur, i.e., once seeding has taken place, removal of the introduced seeds does not prevent further conversions, thereby demonstrating that new templates have formed.
11.3 Mutations in TDP-43 and FUS Genes Cause Amyotrophic Lateral Sclerosis and Frontotemporal Lobar Degeneration
The proper maintenance of protein quality control (PQC) requires the participation in one way or another of a substantial fraction of the cellular proteins. Some participants ensure that proteins fold properly into their native conformations, and that they are expressed at the proper levels so that their concentrations are neither too low nor too high. Others transport newly synthesized proteins and mRNAs to locales in the cell where they are needed, and recycle and remove unwanted and damaged proteins and organelles. One of the most dramatic set of findings in the last few years has been discoveries that aberrations in mRNA processing can produce ALS and FTLD.
Mutations in the genes encoding the mRNA processing proteins TDP-43 and FUS give rise to ALS and FTLD. These disorders are characterized by the presence of inclusions bearing one or the other of these proteins, but not SOD1. As a result, these disorders are regarded as having an etiology distinct from that of SOD1-ALS. TDP-43 is a 414-amino-residue-long protein widely expressed throughout the body. It contains two RNA recognition motifs, RRM1 and RRM2, in its central region and a C-terminal glycine-rich domain that mediates protein–protein interactions (Fig. 11.2). As noted in the introduction to the chapter, the involvement of TDP-43 in neurodegeneration was discovered in 2006. Those finding were followed in 2008 by the discovery of a cluster of disease-causing mutations, all but one in the C-terminal region and the lone exception in RRM1 (Fig. 11.2).
Fig. 11.2
Structure of TDP-43 and FUS along with their disease-causing mutations. These are concentrated in the glycine-rich regions of the two proteins along with a cluster in the C-terminal-most 17 amino acid residues of FUS [see text for details] (from Lagier-Tourenne Hum. Mol. Genet. 19, Rev. Iss 1: R46 © 2010 Reprinted by permission from Oxford University Press)
Soon after TDP-43’s involvement of ALS was uncovered another ALS-causing protein with a structure similar to TDP-43 was found. Discovered in 2009 by two groups, the fused in sarcoma/translocated in liposarcoma (FUS/TLS) gene encodes a 526-amino-acid-residue protein FUS that, like TDP-43, can aggregate in ALS-associated inclusions. As shown in Fig. 11.2 FUS contains an N-terminal region enriched in glutamine, glycine, serine and tyrosine (QGSY) residues and a C-terminal zinc-finger motif. Situated in between are a glycine-rich region, an RRM, and an arginine–glycine (RG)-rich region. Most of the disease-causing mutations are concentrated in two regions—the glycine-rich regions, and the extreme C-terminal segment in which there are five targeted arginine residues.
11.4 TDP-43 Localization, Aggregation, and Prion-Like Spread
TDP-43 is found in the nucleus and cytoplasm. It shuttles between the two compartments but under normal conditions is localized primarily in the nucleus. Disease-causing mutations to TDP-43, or other perturbations that disturb its predominant nuclear localization, lead to the formation of cytoplasmic aggregates and produce characteristic disease phenotypes. Consistent with this finding, inclusions are generated when the TDP-43 cytoplasmic concentration is artificially elevated.
TDP-43-bearing inclusions are the preeminent hallmark of frontotemporal lobar dementia and also familial and sporadic forms of amyotrophic lateral sclerosis (Table 11.1). They are also formed in a substantial fraction of Alzheimer’s disease cases. Recapitulating a widely encountered phenomenon, the C-terminal region of TDP-43 is often proteolytically cleaved, hyperphosphorylated, and ubiquitinated, operations that enhance its tendency to form insoluble aggregates. Again, as was the case for the other neurodegenerative diseases, the exact role of the inclusions in the disease progression is unclear, but there is a feeling that these deposits are part of a protective response to mislocalization, misfolding, and excessive quantities of the protein.
TDP-43 deposition, like tau pathology, α-synuclein deposition, and SOD1-aggregation spreads from region to region in the brain in a stereotypic manner, giving rise to a progressively worsening sequence of disease stages. It is surmised that aggregates of TDP-43 spread out along the brain’s network of anatomical connections, perhaps via exosomes or tunneling nanotube s . This places TDP-43 in an expanding group of proteins that possess some prion-like properties but perhaps lack the latter’s ability to convey infectivity between individuals.
11.5 TDP-43 and FUS Are RNA-Processing Proteins
Unlike the situation for SOD1, cellular damage caused by TDP-43 primarily involves a loss-of-normal function. It turns out that three families of proteins have prominent roles in RNA processing and, when malformed, in neurodegeneration. These are, as follows:
The heterogeneous nuclear ribonucleoprotein hnRNPs family of RNA-binding proteins which includes TDP-43. These proteins carry out a variety of tasks in RNA metabolism upon binding to nascent RNAP II transcripts. These include alternative splicing and regulation of translation.
SR (serine–arginine) repeat-containing RNA-binding proteins. These proteins usually possess two domains—an RNA-recognition motif (RRM) in their N-terminal and a stretch of SR dipeptides (hence their name) that mediates protein–protein interactions in their C-terminal region.
The FET (TET) family of proteins is less well known than the other two families but does include FUS/TLS as a member. Proteins in this family contain motifs for binding nucleic acids, and participate in, and perhaps coordinate, RNA Poly II mediated transcription and pre-mRNA splicing.
11.5.1 Transcription
TDP-43 and FUS participate in several stages of RNA processing . First, TDP-43 and FUS bind ss and ds DNA, RNA, and proteins. As shown in Fig. 11.3 TDP-43 preferentially binds ss TG-rich sequences in promoters, while FUS has been found to associate with the general transcription machinery as well as with certain gene-specific transcription factors. Transcriptional actions carried out at these sites by TPD-43 and FUS at these sites are primarily inhibitory (Fig. 11.3b–d).
Fig. 11.3
Roles of TDP-43 and FUS in RNA processing . (a) Steps in RNA processing; (b) TDP-43 binds ss TG-rich promoter sequences and blocks transcription; (c) FUS associates with the general transcription machinery; (d) in response to DNA damage, FUS is recruited by sense and antisense noncoding RNAs (ncRNAs) and represses transcription; (e, f, g) participation of TDP-43 and FUS in pre-mRNA splicing; (h) Possible role of TDP-43 and FUS in micro-RNA (miRNA) processing; (i) TDP-43 and FUS shuttle between the nucleus and the cytoplasm; (j) TDP-43 and FUS are found in stress granule s , as will be discussed in the text; (k) TDP-43 and FUS are also found in transport granules used to ferry mRNAs to remote sites for local translation (from Lagier-Tourenne Hum. Mol. Genet. 19, Rev. Iss 1: R46 © 2010 Reprinted by permission from Oxford University Press)
11.5.2 Splicing and miRNA Processing
Secondly, TDP-43 and FUS bind to thousands of mRNAs in their 3′ UTR and contributes to RNA splicing (Fig. 11.3e–g). TDP-43 has a strong preference for GU-rich intronic motifs, especially those occurring in long pre-mRNAs, that is, pre-RNA molecules possessing exceptionally long introns. It binds to noncoding RNAs, introns, and 3ʹ UTRs of mRNAs, thereby supporting multiple roles. FUS targets a set of mRNAs that have little overlap with those selected by TDP-43. It has some preference for GUGGU motifs but, in general, recognizes multiple RNA-binding motifs resulting in a sawtooth binding pattern and preferentially binds towards the 5ʹ ends of long introns. Thirdly, TDP-43 and FUS participate in microRNA processing through their membership in the Drosha complex (Fig. 11.5h).
11.5.3 Autoregulation
Expression levels of proteins essential for RNA processing are tightly controlled. One common method of control is through establishment of negative feedback loops. Proteins such as TDP-43 bind to alternatively spliced sites in the 3ʹ UTR of their own mRNAs resulting in the appearance of a premature STOP codon. That event triggers nonsense-mediated RNA decay (NMD), a quality control operation whereby transcripts containing premature STOP codons are eliminated to prevent the generation of incomplete proteins. When placed within the abovementioned negative feedback loop, the expression levels are maintained at a constant value. Mutations in TPD-43 that disturb this feedback loop can alter the TDP-43 expression levels.
11.6 TDP-43 and FUS Are Found in Stress Granules, Processing Bodies, and mRNA Transport Granules
11.6.1 Stress Granules and P-Bodies
Cells respond to elevated stress conditions by reallocating resources to operations needed for survival and eventual recovery. In order to conserve energy, translation of mRNAs is halted. The translation-halted mRNAs and their associated RNA binding proteins are then collected and formed into highly dynamic structures called stress granule s . These dynamic compartments lack a lipid membrane, and can be rapidly assembled and disassembled according to need. A second kind of dynamic, membrane-less aggregate known as a processing body, or P-body , may also be formed. These are aggregates of translational-stalled mRNAs plus the machinery for halting translation and for degradation.
Both of these compartments contain, in addition to RNAs and RNA-processing proteins, signaling proteins that mediate the response to the stress signals sent from metabolic and other cellular signaling pathways. In addition, the assembly of the stress granule s is mediated by TIA-1 (T-cell intracellular antigen 1) proteins. These are RNA-binding proteins that route stalled initiation complexes to nascent stress granules and utilize their prion-like domain s to promote their formation. TDP-43 and FUS shuttle between the nucleus and cytoplasm and are found within stress granules when they form. In those locales, they partner with mRNAs and other RNA processing proteins.
11.6.2 mRNA Transport and Local Translation
TDP-43 and FUS also contribute to mRNA transport and local translation. A third type of transiently formed mRNA granule called an mRNA transport granule is used by neurons to deliver mRNAs to distal sites such as dendrites where they undergo local protein synthesis (Fig. 11.3k). These granules inhibit protein synthesis while undergoing microtubule-based anterograde transport. TDP-43 and FUS associate with proteins involved in mRNA transport. Some mutations in TDP-43 impair this bidirectional movement of mRNA-bearing granule, perhaps providing an explanation for the observed morphological abnormalities in distal axons sometimes seen in ALS-affected neurons.
11.6.3 Angiogenin and Ataxin-2
TDP-43 and FUS are not the only causal agents of ALS and/or FTLD found to have a role in stress granule dynamics. For example, angiogenin (Ang) promotes their assembly. It is a small 14 kDa ribonuclease that is secreted under conditions of stress. It is taken up by angiogenin receptors on neighboring cells where it stimulates angiogenesis, proliferation, and pro-survival responses in target epithelial cells, cancer cells, and motor neurons, respectively. A major way that this protein promotes angiogenesis is through its translocation to the nucleus where it promotes rRNA transcription, an action that is permissive for angiogenesis. Since 2006, 17 missense mutations in this protein have been uncovered. In stressed motor neurons, Ang helps reprogram global translation to ensure neuronal survival. It does so by cleaving tNRAs to generate tRNA halves, stress-induced fragments termed tiRNAs. The resulting 5ʹ-tiRNAs subsequently inhibit protein synthesis and promote the assembly of stress granules where the translation stalled mRNAs are collected.
Mutations in the ATXN2 gene, which encodes the ataxin-2 protein, give rise to two distinct neurological disorders—spinocerebellar ataxia 2 (SCA2) and ALS/FTLD. As noted in the last chapter SCA2 is caused by expanded CAG repeat s in the protein-coding region of the ATXN2 gene. Normally, this gene contains from 15 to 24 CAG repeats. However the gene is unstable and the repeat number can increase from generation to generation. When the number of repeats is 34 or greater, SCA2, which affects cerebellar Purkinje cells, ensues. Intermediate numbers of CAG repeats, from 28 to 33, primarily affects a different set of neurons and produces a different disorder—ALS/FTLD. Several other polyQ-disorder-associated genes have been examined for this propensity, but all failed the test. The implication from this finding is that ability to produce ALS/FTLD is specific to the folding and functions of this particular protein and not to some general property polyQ tracts.
Intermediate repeat-length mutated forms of ataxin-2 associate in an RNA-dependent manner with both wt and mutated TDP-43, the latter more strongly than the former. It also forms associations with FUS and, most importantly, interferes with stress granule and P-body dynamics. One possible result of this interference is to cause these proteins to relocate elsewhere. This protein appears to function as an accelerant, i.e., it enhances the disease-causing propensity of other mutated proteins co-expressed along with it. It not only enhances the harmful effects of mutated TDP-43 and FUS but also that of C9orf72 (soon to be discussed), increases sensitivity to cellular stresses, and alters the mix of ALS and FTLD clinical features being exhibited.
11.7 Phase Transitions Are Mediated by Depletion Attraction, Multivalency, and Low Complexity
11.7.1 Phase Transition Enable the Formation of Liquid Droplets
Recall from elementary chemistry and physics that nonbiological materials exist in one of three phases—gaseous, liquid, or solid. In the gas phase, the constituents are free to move and they undergo little or no interactions with one another. In a liquid phase, weak transient forces are felt that somewhat limit movement, while in a solid phase the interactions are stable and movements are small. Parameters such as temperature govern the transitions between phases. Something similar can be observed in biological materials. Recall from Chap. 8 that lipid membranes exist in one or more phases depending on the amount of cholesterol contained therein giving rise to specialized microdomains such as lipid raft s and caveolae . A similar partitioning of the cytoplasm (Fig. 11.4) may well occur, and which helps makes order out of the crowded intracellular environment. The self-assembly of proteins and RNA molecules into non-membrane bound compartments such as stress granule s and P-bodies is increasingly understood as arising from a phase transition s between a gas-like phase consisting of diffuse soluble molecules and a liquid phase in which the soluble molecules have condensed into a liquid droplet.
Fig. 11.4
Phase transition s by non-biological molecules and intracellular biomolecules. (a) Phase transitions in non-biological materials driven by temperature, and (b) phase transition s in cells driven by multivalency and low complexity (from Weber Cell 149: 1188 © 2012 Reprinted by permission from Elsevier)
That the cytoplasm and its macromolecular constituents might be organized as multi-phase mixtures of liquid droplets (“granules”) that self-organize or condense out of a gas-phase diffuse background (“continuous substance”) was first posited by Edmund Beecher Wilson (1856–1939), the father of American cell biology. That idea languished for a century, but has recently seen a revival with the recognition that at the mesoscale (100 nm to 1 μm [bacteria] or 10 μm [eukaryotes]) level the cytoplasm and its macromolecular constituents undergo liquid phase separations. This process generates a variety of structures where proteins, lipids, and nucleic acids that need to work together can be rapidly brought into close proximity and then kept there for as long as necessary. In particular, membranes-less dynamic compartments are widely utilized to facilitate protein–RNA interactions. Examples of these structures include cytoplasmic stress granule s and P-bodies along with transport granules, Cajal bodies, nucleoli, nuclear speckles, and germ-line P-granules.
11.7.2 Depletion Attraction, Multivalency and Low Complexity Underlie Liquid Droplet Phase Transitions
The ability to form membrane-less compartments containing the correct constituents is made possible by a confluence of physical and chemical properties. First, the structures so formed exhibit surface tension and fluidity; they also fuse together and drip in the presence of shear. These are properties of liquid droplets. Their formation is promoted by macromolecular crowding , which was discussed earlier in Chap. 2. The formal name given to the tendency, or force, arising from macromolecular crowding is depletion attraction . This attraction is an entropic one arising from the exclusion of small bodies from the spaces in between large macromolecules. A mathematic formula for the force so generated using a hard-sphere model was presented by Asakura and Oosawa in 1954, and largely confirmed in experiments carried out using optical tweezer s . In the case of proteins and RNA, two other properties of the protein constituents—their multivalency and low complexity —further drive the liquid droplet phase transition .
In a multivalent interaction, multiple contacts of the same kind are established between a protein and its binding partner. The advantage of this type of non-covalent interaction is that it enables multiple weak and rather nonspecific interactions (i.e., depletion attraction , hydrogen bonds, and van der Waals forces) to produce strong specific bonds, which can be easily broken and the partners separated when necessary.
The other key property needed for liquid droplet formation is low complexity , that is, the presence of amino acid sequences that have little diversity in their composition. Repeats are, of course, a prime example of amino acid stretches with little (or no) diversity. (They are also natively disordered.) For example, FUS contains 27 repeats of the tripeptide (G/S)Y(G/S) sequence. In examining the composition of stress granule s , it emerges that there is a practical purpose for the repeat sequence s . The proteins containing low complexity (LC) domains found in the stress granules can undergo a phase transition to a hydrogel state in which they form polymerized fibers. Although they are amyloid-like these fibers can readily assemble and disassemble, and accommodate heterotypic polymerization. However, when disease-causing mutations are present these same properties can promote the formation of pathological aggregates (Fig. 11.4b).
11.7.3 Prions Are Low Complexity Proteins
Prions are a good example of low complexity proteins. Prions and prion strains were introduced and discussed in Chap. 7, and again in the subsequent chapters with regard to their remarkable ability to spread misfolding and aggregation from cell to cell. The chief biochemical characteristic of the underlying prion domain was not the details of its primary sequence but simply its overall amino acid composition. Prion-like domain s (PrLDs) are low-complexity sequences composed of glycine (G) and the uncharged polar amino acids asparagine (N), glutamine (Q), tyrosine (Y), and serine (S). The structures formed by these amino acids (1) can switch rapidly between noninfectious and infectious conformations, the latter giving rise to different prion strains, and (2) can template the conversion of noninfectious to infectious forms in original or new host cells. These domains are found throughout the human genome . It is estimated that 1 % of all the genes contain domains of this type. These domains are even more common among RNA-binding proteins where 20 % of these proteins contain PrLDs, and that grouping includes both TDP-43 and FUS.
Prion-like domain s serve a useful function in their host RNA-processing proteins. They are needed for alternative splicing activities; they stabilize binding to RNAs, and promote optimal RNA hydrogen bond pairing (annealing). They may also mediate the phase transition that underlies the assembly of stress granule s , driving the assembly of TDP-43 and FUS into oligomers and linear polymers within the hydrogels. However, when mutated, they tend to drive self-complementary beta-strand assembly.
11.8 Stress Granules Are Cleared by Autophagy
11.8.1 VCP and p62
Several of the entries in Table 11.1 have roles in protein quality control and, more specifically, in autophagic clearance of unwanted and damaged proteins and protein aggregates. One of the listed proteins whose mutated forms lead to ALS is valosin-containing protein (VCP). Mutated forms of this protein also causes inclusion-body myopathy, Paget disease, and Frontotemporal dementia (IBMPFD)/multisystem proteinopathy (MSP). Recall from Chap. 6 that VCP is a versatile AAA-ATPase involved in maintaining protein quality control. It (1) unwinds DNA, (2) extracts proteins from membranes by unfolding and disaggregating them, (3) chaperones proteins to proteasomes, and (4) assists in autophagy. Mutated forms of VCP associate with TDP-43, hnRNPA1, hnRNPA2B1 and other stress-granule-associated proteins. In the presence of VCP mutations, autophagic clearance of stress granule s as well as the clearance of damaged mitochondria is impaired.
Another protein involved in autophagy and whose mutated forms are associated with increased susceptibility to ALS is the ubiquitin-binding protein p62/SQSTM1. This scaffold protein plays a role in both proteasomal degradation and through its interactions with LC3 in autophagic pathways (see Fig. 6.8). Inclusion bodies containing p62 are encountered in FTLD, Pick’s disease, and several other neurological disorders. Mutant SOD1 associates with p62, and mutations in SQSTM1, the gene encoding p62, are found in ALS and FTLD. The p62 protein itself is degraded by autophagy. Impairment through mutations in p62 or in other autophagy-essential proteins results in an excessive accumulation of proteins needing clearance, and a shift in the cellular balance towards stress responsive/cell death pathways.
11.8.2 Ubiquilin 2 and Optineurin
Two other proteins linked to ALS and FTLD and whose mutated forms are tied to defective proteasomal and autophagic-lysosomal pathways are ubiquilin 2 encoded by the UBQLN2 gene and optineurin encoded by the OPTN gene. The ubiquilin 2 protein possesses a ubiquitin-binding motif and a heat-shock chaperone binding domain that stabilizes proteasome substrates. ALS-associated mutations in this protein promote aggregation, formation of cytoplasmic inclusions, and localization to vesicles containing optineurin and p62. Optineurin contains a ubiquitin-binding domain, a motif for interacting with LC3, and a coiled-coil domain that recognizes protein aggregates. It interacts with p62 and LC3 to form an autophagy receptor complex and induces the formation of autophagosomes. It has been found to play a key role in the autophagic removal of ubiquitin-coated Salmonella enterica and in the mitophagic removal of parkin-labelled damaged mitochondria.
11.9 Mutations in the C9ORF72 Gene Are the Leading Cause ALS and FTLD
11.9.1 The C9ORF72 Gene Contains a Disease-Causing Unstable GGGGCC Hexanucleotide Repeat
It had been known for some time that an important locus for autosomal dominant ALS and FTLD resided on the short arm of chromosome 9. Finally, in 2011, two groups reported that the responsible gene, located at 9p21, was C9ORF72. In this gene, a GGGGCC hexanucleotide expansion in the noncoding region between exons 1a and 1b (Fig. 11.5) causes ALS and FTLD. This defect is the single most common cause of familial instances of ALS. As shown in Table 11.1 it accounts for roughly 40 % of all fALS cases and also accounts for about 10 % of sALS and FTLD disease situations. Normally, there are from 2 to 23 GGGGCC hexanucleotide repeats, but in the disease-causing C9ORF72 genes the number of repeats has expanded to the neighborhood of 700–1000.
Fig. 11.5
Possible mechanisms of toxicity from GGGGCC repeat expansions contained within the C9ORF72 gene [see text for details] (from Ling Neuron 79: 416 © 2013 Reprinted by permission from Elsevier)
Unstable repeat expansions in either protein coding or noncoding regions were discussed at length in the last chapter. Continuing that discussion, recall that fragile X syndrome and monotonic dystrophy 1 (MD1) were among the diseases brought on by unstable repeat expansions in non-protein-coding regions. As discussed in Chap. 10, aberrant repeat-generated DNA, RNA, and DNA–RNA hybrid structures such as hairpin and R-loop s interfere with transcription, translation, and replication. The hairpins in MD1 sequester essential proteins, while the R-loops in fragile X syndrome silence genes.
Those same classes of defects—loss-of-protein function in the case of gene silencing and gain-of-toxic-RNA function in the case of protein-sequestering R-loop s —occur in cases of GGGGCC repeat expansions in C9ORF72. That commonality is depicted in Fig. 11.5. The first panel of this figure (Fig. 11.5a) emphasizes the loss of the allele containing the excessively expanded repeat. (In haploinsufficiency, the mutant allele is dominant over the wt allele.) The result is reduced mRNA levels and loss-of-protein function. The second panel (Fig. 11.5b) depicts the formation of RNA foci about hairpin structures. The third panel (Fig. 11.5c) is new and to discuss it two additional sets of recent discoveries need to be introduced. The first of these is the emergence of a phenomenon first observed a century ago, in 1910, and now known to be caused by G-quadruplex es , and the second is the discovery of an unconventional form of translation that bypasses the requirement for an ATG [AUG] (Met) start codon.
11.9.2 GGGGCC Repeats Generate G-Quadruplexes
In 1910, at the same time that Harriette Chick was exploring the heat coagulation of egg albumen and hemoglobin, the Scandinavian medical biochemist Ivar Bang (1869–1918) was studying guanylic acid. He had just published a paper noting its puzzling tendency at high concentrations to form a gel. The mystery of how this could happen lasted for over 50 years, but was finally explained in 1962 when Gellert, Lipsett, and Davies examined the X-ray diffraction pattern generated by the material. They found that sets of four guanines self-assembled into planar squares with open centers, and these structures then stacked one on top of the other to produce the exceptionally stable guanine gels observed by Bang.
Interestingly, these stable structures are formed through Hoogsteen-type hydrogen bonding , an alternative type of hydrogen bonding first described by Karst Hoogsteen in 1963. In this type of bonding, A•T (or G•C) base pairs form with a different geometry from that of Watson–Crick base pairing. Additional stability is provided by the presence of a cation in the central cavity of each planar structure, thereby reducing the electrostatic repulsion of the nearby oxygen atoms. That these structures were more than laboratory curiosities became apparent with the discovery that they cap telomeres. Recall that telomeric ends of eukaryotic chromosomes are capped by repeating guanines interspersed with short A/T-rich sequences. These stretches form an overhang that protects the ends of chromosomes from attack by DNA repair proteins. It emerges that these overhangs form G-quadruplex es .
Interest in these alternative DNA and RNA structures increased further with realization that they are functionally relevant for transcriptional regulation and DNA replication as well as genomic stability. These structures are found in many regions where repetitive sequences are present including upstream promoters and, like the hairpins previously discussed, serve as barriers to replication fork progression. They are, in fact, an alternative to the hairpins. These two types of structures can interconvert into each other, and into other conformations, according to environmental conditions such as pH, metal ions, or small-molecule metabolites.
11.10 Repeat-Associated Non-ATG (RAN) Translation Causes Neurodegeneration
11.10.1 RAN Translation Generates Toxic Proteins and RNAs
Repeat-associated non-ATG (RAN) translation, the second of the two new features, was first described in 2011 by Zu and coworkers in Laura Ranum’s group. They had been studying the CAG•CTG repeat expansion in the ATXN8OS/ATXN8 genes associated with spinocerebellar ataxia 8 and also the CTG repeat expansion in the DMPK gene that causes myotonic dystrophy 1 . Both sets of microsatellite expansions occur in noncoding regions of those proteins. What they found was that these microsatellite expansions did not follow the normal rules for translation. First, translation of these sequences was being initiated without the presence of a start ATG [AUG] codon, and secondly, the result of translating the non-protein-coding repeats in this way was a protein, specifically, a number of different homopolymeric peptides. Thus, one needs to consider an expanded repertoire of potentially toxic RNA and protein species and mechanisms. Among the unstable repeats subsequently found to undergo RAN translation were the expanded CGG repeats in the FMR1 gene responsible for fragile X tremor ataxia syndrome, and the expanded GGGGCC repeats in the C9ORF72 gene that produce ALS and FTLD.
In more detail, each set of malformed transcripts generates a specific collection of RAN translation proteins:
SCA8—From expanded CAG repeat s in the ATXN8 gene: utilization of all three reading frames (CAG, AGC, and GCA) produces polyglutamine, polyserine, and polyalanine homopolymeric peptides. Similarly, translation from the antisense (opposite strand) CUG, UGC, and GCU reading frames generates polyleucine, polycysteine, and polyalanine, respectively.
FXTAS—From expanded CGG repeats in the FMR1 gene: The primary toxic product is a protein containing a long polyglycine stretch in its N-terminal designated by the discoverers as FMRpolyG. That RAN translation product is normally degraded but long repeats resist that process leading to their accumulation in intracellular inclusion s , and resulting in cytotoxicity.
ALS/FTD—From expanded GGGGCC repeats in the C9ORF72 gene: utilizing all reading frames, six RAN translation dipeptide polymers can be produced. Three of these can be generated from transcription carried out in the sense direction—poly glycine–alanine (GA)N, poly glycine–proline (GP)N, and poly glycine–arginine (GR)N, and another three RAN translation products can be produced from transcription performed in the antisense direction—poly proline–alanine (PA)N, poly proline–glycine (PG)N, and poly proline–arginine (PR)N.Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree