Genetic Convergence Brings Clarity to the Enigmatic Red Line in ALS

Genetic Convergence Brings Clarity to the Enigmatic Red Line in ALSNeuron
Review
Genetic Convergence Brings Clarity to the Enigmatic Red Line in ALS
Casey Cook1,2 and Leonard Petrucelli1,2,* 1Department of Neuroscience, Mayo Clinic, Jacksonville, FL, USA 2Neurobiology of Disease Graduate Program, Mayo Clinic Graduate School of Biomedical Sciences, Jacksonville, FL, USA *Correspondence: [email protected] https://doi.org/10.1016/j.neuron.2019.02.032
Amyotrophic lateral sclerosis (ALS) is an aggressive neurodegenerative disorder that orchestrates an attack on the motor nervous system that is unrelenting. Recent discoveries into the pathogenic consequences of repeat expansions in C9ORF72, which are the most common genetic cause of ALS, combined with the iden- tification of new genetic mutations are providing novel insight into the underlying mechanism(s) that cause ALS. In particular, the myriad of functions linked to ALS-associated genes have collectively implicated four main pathways in disease pathogenesis, including RNA metabolism and translational biology; protein quality control; cytoskeletal integrity and trafficking; and mitochondrial function and transport. Through the identification of common disease mechanisms on which multiple ALS genes converge, key targets for potential therapeutic intervention are highlighted.
Overview of ALS: Clinical Perspective and Features Amyotrophic lateral sclerosis (ALS), also known as Lou Gehrig’s
disease, is themost commonmotor neuron disease, with an esti-
mated prevalence of 5 in 100,000 people affected in theUS alone
(Mehta et al., 2018). While the defining clinical feature of ALS is
involvement of both upper and lower motor neurons, significant
heterogeneity is observed across patients in terms of age of
onset, initial site of onset, anatomical pattern and rate of disease
progression, involvement of upper versus lower motor neuron
signs, and presence of behavioral and/or cognitive symptoms
(Ravits et al., 2007; Statland et al., 2015). However, despite the
wide variability observed in clinical presentation, ALS is typically
a rapidly progressive, fatal disorder with an average duration of
2.5–3 years (Statland et al., 2015).
As the main goal of this Review is to examine disease mecha-
nisms in ALS, it is important to first consider the anatomy and
connectivity of the neuronal tracts susceptible to degeneration
(Figure 1), combinedwith clinical observations of typical patterns
of disease onset and progression. Dysfunction of upper motor
neurons (UMNs) (large pyramidal neurons called Betz cells that
are located in layer V of the primary motor cortex and project
to lower motor neurons [LMNs] in the spinal cord) is charac-
terized by increased muscle spasticity and brisk reflexes. In
contrast, as LMNs project out of the spinal cord to directly inner-
vate muscle, LMN signs include muscle weakness, fascicula-
tions, and atrophy (Statland et al., 2015). While LMN involvement
can be determined by electrophysiologic tests, assessment of
UMN symptoms currently relies on clinical examination and
can be masked by LMN degeneration (Eisen and Swash, 2001;
Simon et al., 2014; Statland et al., 2015). To address this limita-
tion, the utilization of high-resolution imaging techniques to
quantitatively measure UMN degeneration in the primary motor
cortex has been explored with promising results (Cosottini
et al., 2016; Donatelli et al., 2018), which will likely facilitate map-
ping the onset and spread of disease through the nervous sys-
tem in future studies.
N
The onset of UMN and/or LMN signs can occur in any part of
the body, with a cross-sectional study finding that motor symp-
toms were localized to one body region at the time of initial diag-
nosis in 98% of ALS patients (Ravits et al., 2007). This focal
nature of initial symptoms was noted regardless of site of onset.
In patients with bulbar onset, typically one major muscle group
was affected first (pharyngeal, laryngeal, masticator, or tongue
muscles), while limb onset was usually unilateral (Ravits et al.,
2007). In addition, the evolution of symptoms tends to follow
several intriguing patterns that may support the proposedmech-
anism that disease progression is mediated by propagation or
spreading of pathology through anatomically connected path-
ways and/or neighboring regions. For example, the contralateral
limb is most frequently affected next in patients with unilateral
limb onset, while patients with bulbar onset typically report
symptoms in the cervical region as opposed to the more distant
lumbosacral region of the spinal cord (Walhout et al., 2018).
There also appears to be a preferential direction of disease pro-
gression through the spinal cord, with lumbosacral involvement
most often following onset in cervical regions (rostrocaudal di-
rection) rather than bulbar region involvement following cervical
onset (Figure 1) (Ravits and La Spada, 2009; Walhout et al.,
2018). Progression to non-contiguous regions of the body is
observed in some patients, but most often with UMN signs,
which may reflect the topography of the primary motor cortex
(Ravits and La Spada, 2009; Walhout et al., 2018). Therefore,
although the simultaneous involvement of UMN and LMN sys-
tems increases the complexity of tracking disease progression,
the combination of rapidly evolving imaging capabilities and
recent scientific discoveries will both simplify this process and
provide the means to evaluate the relationship with pathological
features and genetic factors.
Genetics of ALS The most common form of ALS is sporadic, but approximately
5%–10% of ALS cases are considered familial (Nguyen et al.,
euron 101, March 20, 2019 ª 2019 Published by Elsevier Inc. 1057
1058 Neuron 101, March 20, 2019
Neuron
Review
2018; Taylor et al., 2016). While 60%–70% of familial ALS (fALS)
and 10% of sporadic ALS (sALS) cases can be attributed to
mutations in six genes (SOD1, TARDBP, FUS, VCP, C9ORF72,
and OPTN) (Taylor et al., 2016), additional rare genetic variants
have also been discovered that collectively appear to incrimi-
nate four major pathways in the pathogenesis of disease. In
particular, disruption of RNA metabolism and translational
biology (C9ORF72, TARDBP, FUS, TIA1, MATR3, HNRNPA1,
HNRNPA2/B1, EWSR1, TAF15, ANG), aberrant regulation of
protein quality control (UBQLN2, VCP, OPTN, VAPB, TBK1,
SQSTM1), cytoskeletal defects and trafficking abnormalities
(PFN1, TUBA4A, KIF5A, ANXA11), and mitochondrial dysfunc-
tion and oxidative stress (SOD1, CHCHD10) have all been impli-
cated in ALS based on genetic variants and functional studies
(Kapeli et al., 2017; Nguyen et al., 2018; Nicolas et al., 2018; Tay-
lor et al., 2016), which will be discussed in greater detail below.
As this broad classification is not exclusive in the sense that
pathogenic mutations likely disrupt multiple pathways and
cellular functions, it is intriguing to note that there is a higher inci-
dence of multiple ALS-associated genetic variants in the same
ALS patients and families than what would be expected based
on chance alone (van Blitterswijk et al., 2012). Since the G4C2
hexanucleotide repeat expansion in the intronic region of the
C9ORF72 gene is the most common genetic cause of ALS
(although frequency varies in different regions of the world;
Majounie et al., 2012), this mutation is most frequently observed
in combination with other ALS genetic variants in the same indi-
vidual. This finding might indicate that coordinated and simulta-
neous disruption of multiple pathways and cellular functions is in
fact required for the ALS disease phenotype, but additional
studies assessing the relationship with clinical presentation are
warranted.
Neuropathological Features of ALS In patients with pure ALS, marked degeneration of anterior nerve
roots leaving the spinal cord is observed in the absence of gross
atrophy in the brain (Saberi et al., 2015). However, as 15%of ALS
patients are also diagnosed with frontotemporal dementia (FTD)
(Robberecht and Philips, 2013), which is characterized by
behavioral and cognitive abnormalities, patients with comorbid
ALS and FTD also exhibit degeneration of frontal and temporal
lobes. The seminal discovery that the main protein component
of ubiquitinated inclusions in the majority of ALS patients (and
up to 50% of FTD patients) is TAR DNA-binding protein 43
(TDP-43) (Neumann et al., 2006), which is encoded by the
TARDBP gene, provided significant insight into the biology of
ALS and helped to establish the current view that ALS and
FTD are part of the same disease spectrum. Finally, although
TDP-43 pathology is observed in approximately 97% of ALS
cases (Nguyen et al., 2018), patients with genetic mutations in
either SOD1 or FUS lack TDP-43 inclusions and instead develop
pathology containing mutant SOD1 or FUS protein, respectively
(Saberi et al., 2015).
is also associated with depletion of nuclear TDP-43 in affected
neurons (Giordana et al., 2010). While the mechanism behind
loss of nuclear TDP-43 has not been conclusively determined,
Neuron
Review
plasmic trafficking, such as dysregulation of nuclear export or
import pathways, followed by excessive posttranslational mod-
ifications (ubiquitination and phosphorylation) and sequestration
within cytosolic aggregates (Saberi et al., 2015). However, Braak
and colleagues discovered that despite the classic redistribution
of TDP-43 from the nucleus to cytosolic aggregates detected in
LMNs, nuclear clearing of TDP-43 was actually observed in the
relative absence of TDP-43-positive cytosolic inclusions in
some Betz cells in the primary motor cortex of sALS patients
(Braak et al., 2017). This indicates that accumulation of TDP-43
within protein aggregates in the cytosol does not entirely ac-
count for lack of nuclear TDP-43, at least in Betz cells.
Given the pathological abnormalities in TDP-43 (i.e., subcellu-
lar localization, posttranslational modifications, and solubility),
two recent studies evaluated the consequence of TDP-43 loss
of function, discovering that reduced expression of the TDP-43
target stathmin-2 (STMN2) due to aberrant splicing is a new
biochemical feature of ALS (Klim et al., 2019; Melamed et al.,
2019). Since STMN2 regulates microtubule stability (Morii
et al., 2006; Riederer et al., 1997), future studies will need to
assess whether reduced expression of either STMN2 or another
TDP-43 target contributes to the morphological defects
observed in Betz cells in both sALS and fALS patients. These de-
fects include a reduction in cell soma size, degeneration of axons
in the corticospinal tract, andmarked deterioration of apical den-
drites with vacuolation and significant reduction in synapses
(Genc et al., 2017; Kiernan and Hudson, 1991; Saberi et al.,
2015). As incoming synaptic signals at the apical dendrite
converge to modulate activity and output of Betz cells, the
degeneration and synaptic loss observed in ALS patients could
represent a key step in disease pathogenesis, essentially dis-
connecting the motor system from CNS control. This is particu-
larly intriguing given that cortical hyperexcitability appears to be
a prominent and early feature of ALS, which may represent an
attempt by the CNS to compensate for degenerative processes
and regain control of motor function (Menon et al., 2015; Vucic
and Kiernan, 2006; Vucic et al., 2008).
Efforts to define neuropathological stages of ALS also seem to
support early cortical involvement in the disease process. Spe-
cifically, using an antibody specific for hyperphosphorylated
TDP-43 (pS409/410) to label pathology, four main stages of
ALS disease progression were identified (Brettschneider et al.,
2013). Although involvement was primarily restricted to the pri-
mary motor cortex and motor neurons in the brainstem and spi-
nal cord in the initial stage, pathology gradually spreads through
the nervous system with each consecutive stage, culminating
with involvement of the hippocampus by stage 4. Intriguingly,
direct cortical innervation appears to impact susceptibility of spi-
nal cord motor neurons, as oculomotor and visceromotor neu-
rons that receive only indirect cortical input are relatively spared
in ALS (Braak et al., 2013). These findings, combined with the
observation that oligodendrocytes in close proximity to axons
(but not cell bodies) of affected neurons also develop TDP-43-
positive inclusions, supports a model in which pathology
spreads anterogradely from the cortex to spinal cord through
axonal transport (Braak et al., 2013; Brettschneider et al.,
2013). Finally, as C9ORF72 repeat expansion carriers with ALS
exhibited an elevated burden of TDP-43 pathology in all regions
despite a similar pattern of distribution (Brettschneider et al.,
2013), including frontal and temporal lobes, this could provide
insight into why the C9ORF72 mutation also increases risk of
FTD (Nguyen et al., 2018). These ideas will be discussed and
re-examined in greater detail below in the context of potential
molecular mechanisms of disease.
derived from the sense (G4C2) or antisense (C4G2) repeat expan-
sion are detected in C9ORF72 mutation carriers. Specifically,
sense and antisense repeat-containing RNA transcribed from
C9ORF72 expansions can form nuclear RNA foci (DeJesus-Her-
nandez et al., 2011; Lagier-Tourenne et al., 2013; Mizielinska
et al., 2013) and subsequently undergo repeat-associated non-
ATG (RAN) translation (Ash et al., 2013; Gendron et al., 2013;
Mori et al., 2013; Zu et al., 2013). This atypical form of translation,
which occurs in all reading frames in the absence of an initiating
start codon, produces dipeptide repeat proteins (DPRs): poly
(GA), poly(GR) and poly(GP) proteins from sense G4C2 repeat
RNA, and poly(PA), poly(PR) and poly(GP) proteins from anti-
sense G2C4 repeat RNA. Inclusions containing DPRs generated
from the sense transcript are more abundant than antisense-
derived DPRs, with the highest burden of DPR aggregates
observed in cerebellum, hippocampus, and neocortical regions
(Mackenzie et al., 2014, 2015). Intriguingly, while RNA foci are
detected in LMNs in the spinal cord (DeJesus-Hernandez
et al., 2011; Lagier-Tourenne et al., 2013; Mizielinska et al.,
2013), DPR inclusions are rarely observed in LMNs and do not
correlate with TDP-43 burden or disease progression (Gomez-
Deza et al., 2015; Mackenzie et al., 2015).
Pathogenic Mechanisms Implicated in ALS As discussed above, the myriad of functions linked to ALS-asso-
ciated genes have collectively implicated four main pathways in
disease pathogenesis, including RNA metabolism and transla-
tional biology; protein quality control; cytoskeletal integrity and
trafficking; and mitochondrial function and transport (Figure 2).
Given that mutations in SOD1 (encodes an enzyme called Cu-
Zn superoxide dismutase) were the first genetic cause of ALS
discovered in 1993 (Rosen et al., 1993), the majority of studies
investigating disease mechanisms in ALS and preclinical testing
of potential therapeutic strategies were performed in mutant
SOD1 models (reviewed in Benatar, 2007; Taylor et al., 2016).
However, considering that alterations in the enzymatic function
of SOD1 to convert superoxide to hydrogen peroxide and oxy-
gen do not appear to explain how ALS-linked mutations in
SOD1 cause disease (Bruijn et al., 1998; Ratovitski et al.,
1999), alternative pathogenic mechanism(s) have been and
continue to be explored. In particular, the finding that aberrant
folding and aggregation of mutant SOD1 correlates with disease
duration in patients (Abu-Hamad et al., 2017; Prudencio et al.,
2009; Wang et al., 2008) implicates an impairment in protein
quality control, with further insight provided by the demonstra-
tion that mutant SOD1 pathology can propagate in vivo and is
associated with degeneration of motor neurons and paralysis
(Ayers et al., 2016). Additional disease mechanisms that have
been investigated include the contribution of glial cells to mutant
SOD1 toxicity (reviewed in Taylor et al., 2016), as well as the
Neuron 101, March 20, 2019 1059
Figure 2. Pathogenic Mechanisms Implicated in ALS Schematic representation depicting how putative disease mechanisms might converge to disrupt local translation at the synapse in affected neurons in ALS. Altered stress granule (SG) dynamics favoring more stable assemblies would be expected to sequester both mRNAs and RNA-binding proteins (RBPs), limiting the availability of mRNAs available for translation in the synapse. Sequestration of ribosomal subunits would lead to reduced protein translation throughout the cell, including at the synapse. Impaired proteasomal activity due to sequestration within inclusions would exacerbate deposition of aggregation-prone proteins, such as RBPs known to associate with SGs, compromising both SG dynamics and the ability of aggregation-prone RBPs to deliver their mRNA targets to the synapse. Decreased nucleocytoplasmic trafficking would lead to inappropriate nuclear/cytoplasmic ratios for critical RBPs, exacerbating perturbations in SG dynamics and protein aggregation, and inhibiting mRNA transport to the synapse. Impaired axonal transport would clearly inhibit translation at the synapse through reduced local concentrations of RNA granules, ribosomal subunits and translational machinery, and mitochondria.
Neuron
Review
endoplasmic reticulum (ER) stress, and mitochondrial trans-
port and trafficking abnormalities, all of which will be discussed
below.
redistribution as a characteristic feature of the majority of ALS
cases, with the exception of SOD1 and FUS mutation carriers
(Saberi et al., 2015), has now led to a dramatic shift in research
efforts to focus on TDP-43 models and pathophysiology. This
is due in part to the potential concern that SOD1-linked ALS
and the more common forms of sALS and fALS associated
with TDP-43 pathology may be mediated by distinct mecha-
nisms, which might explain the poor translation of therapeutic
candidates that cure SOD1 mice to an efficacious treatment
for ALS patients (Benatar, 2007), although it remains to be
seen whether preclinical testing in a model with TDP-43 pathol-
ogy will improve translation to an effective human therapy.
Regardless, the transformative discovery that C9ORF72 repeat
expansions are a common genetic cause of ALS (DeJesus-Her-
nandez et al., 2011) has further galvanized the field, leading to a
remarkable pace of scientific breakthroughs that are increasing
hope for an effective treatment. As an accurate picture of the
1060 Neuron 101, March 20, 2019
dysfunction is needed to develop a cure, the multiple pathol-
ogies associated with the C9ORF72 repeat expansion collec-
tively implicate the four main pathways linked to other ALS genes
and, in converging on common disease mechanisms with other
genetic causes of ALS, highlight key targets for potential thera-
peutic intervention.
pathogenic consequences associated with the G4C2 hexanu-
cleotide repeat expansion in C9ORF72, including loss of endog-
enous C9ORF72 protein expression, accumulation of sense and
antisense repeat-containing RNA in nuclear foci, and DPR depo-
sition. While a combination of all three is believed to be involved
in disease onset/progression to some extent (Taylor et al., 2016),
the finding that accumulation of TDP-43 inclusions in mice ex-
pressing expanded G4C2 repeats via adeno-associated virus
(AAV) or bacterial artificial chromosome (BAC) clearly estab-
lishes that TDP-43 pathology is downstream of the repeat
expansion (Chew et al., 2015; Liu et al., 2016b). Since the 66
G4C2 repeats and 2 G4C2 repeat control were cloned into the
AAV vector outside the context of the endogenous mouse
C9orf72 gene (although 119 base pairs of the 50 flanking region
Neuron
Review
and 100 base pairs of 30 flanking region of the human C9ORF72
genewere included in the construct; Chew et al., 2015), the pres-
ence of TDP-43 pathology in the 66 repeat G4C2-AAV mouse
model is likely mediated by repeat-containing RNA or DPR
burden rather than reduced C9orf72 protein levels. However,
this does not exclude the possibility that loss of C9ORF72
expression could contribute to the disease phenotype in
C9ALS/FTD patients, in particular given that C9orf72 knockout
mice exhibit inflammatory and autoimmune phenotypes
(Burberry et al., 2016; O’Rourke et al., 2016). Loss of C9ORF72
has also been shown to impact lysosomal biogenesis and vesic-
ular trafficking, as well as exacerbate toxicity to poly(GR) and
poly(PR) exposure in human induced motor neurons (Shi et al.,
2018), whichmay indicate thatC9ORF72 haploinsufficiency syn-
ergizes with DPR and repeat RNA toxicity to drive neurodegen-
eration in C9ORF72 expansion carriers.
The G4C2 repeat has been shown to form highly-stable,
G-quadruplex structures (Conlon et al., 2016; Haeusler et al.,
2014) and assemble into RNA foci by undergoing phase separa-
tion in a repeat-length-dependent manner (Jain and Vale, 2017).
Moreover, the RNA-binding proteins (RBPs) hnRNP H and
MBNL1 were colocalized and potentially sequestered by G4C2
RNA foci (Conlon et al., 2016; Jain and Vale, 2017; Lee et al.,
2013), which is notable for a number of reasons. First, transcrip-
tional alterations in both C9ORF72-associated ALS (C9ALS) and
sALS patients implicate a loss of function of hnRNP H (Conlon
et al., 2016; Prudencio et al., 2015). Second, splicing defects in
hnRNP H targets correlate with hnRNP H insolubility in sALS pa-
tients (Conlon et al., 2018). Finally, the synthesis of toxic RAN
translation products is inhibited byMBNL1-mediated sequestra-
tion of CCUG-repeat-containing RNA in myotonic dystrophy
type 2 (Zu et al., 2017), a disorder caused by a CCTG repeat
expansion in the CNBP gene (Liquori et al., 2001). While it re-
mains to be determined whether MBNL1 levels regulate DPR
production by sequestration of G4C2-repeat-containing RNA in
nuclear foci in…