Introduction
We recently reported the identification of mutations in the T-cell restricted intracellular antigen-1 gene (
TIA1) as a cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) [
19]. Similar to several other ALS/FTD related proteins (e.g. transactive response DNA-binding protein 43 (TDP-43), fused in sarcoma (FUS) and heterogeneous nuclear ribonucleoprotein A1 (hnRNPA1)), TIA1 is an RNA binding protein that contains a C-terminal, prion-like, low complexity domain (LCD) which promotes its self-assembly and the formation of membrane-less organelles through the process of liquid-liquid phase separation (LLPS) [
16,
22,
31]. Specifically, TIA1 plays a central role in the formation of stress granules (SG) that form in response to environmental stress to temporarily store and protect mRNA [
1,
9,
14,
25]. SG dysfunction has been implicated in the pathogenesis of a number of neurodegenerative conditions including ALS [
1,
30] and
TIA1 was previously identified as a candidate ALS gene in a yeast functional screen [
5]. Moreover, a founder mutation affecting the TIA1 LCD (E384K) has been reported in Swedish/Finnish patients to cause Welander distal myopathy (WDM) [
10,
15], a type of vacuolar myopathy with clinical and histopathological similarity to the myopathies caused by mutations a number of other genes that can also cause ALS/FTD (e.g. valosin containing protein and sequestosome-1) [
8,
12].
In the previous study, we identified a different heterozygous missense
TIA1 mutation (P362L) in affected members of a family with autosomal dominant ALS and FTD [
19]. This variant affects a highly conserved residue in the LCD and is predicted to be deleterious. Subsequent analysis of a large cohort of patients with ALS, with and without FTD, identified
TIA1 mutations in approximately 2% of familial ALS (fALS), and 0.4% of sporadic ALS (sALS), but not in neurologically normal controls [
19]. Autopsy material from five
TIA1 mutation carriers showed widespread TDP-43 immunoreactive (TDP-ir) pathology as a consistent feature. Biophysical and cell culture studies demonstrated that the disease associated mutations altered phase transition of TIA1 and resulted in SG that failed to normally disassemble following the removal of stress. It is known that TDP-43 is recruited into SG under a variety of stress conditions [
1] and we showed that prolonged localization of TDP-43 within persistent SG promotes TDP-43 aggregation and reduces its solubility. Based on these findings, we proposed that
TIA1 mutations are a cause of ALS and FTD; thus, reinforcing the central role of RNA metabolism and SG dynamics in the pathogenesis of this spectrum of disease [
19].
Whereas the original study focused on the genetic analysis and functional effects of TIA1 mutations, in this report we provide a more detailed description of the associated clinical features and neuropathology. In particular, we highlight phenotypic and pathological characteristics that distinguish cases with TIA1 mutation from other types of familial and sporadic ALS and FTD.
Discussion
The purpose of this report is to provide a detailed description of the clinical and neuropathological features associated with the recently identified TIA1 mutations that cause ALS ± FTD. Although the number of cases currently available is not extensive, it is sufficient to identify some features that may be helpful in distinguishing these cases from other ALS subtypes.
In the initial discovery series,
TIA1 mutations were identified in 2.2% of familial ALS and 0.4% of sporadic ALS cases [
19]; far less frequent than the most common genetic cause of ALS, which is the
C9orf72 repeat expansion (responsible for 37% of familial and 6% sporadic ALS [
26]), but similar to the frequencies reported for
VCP and
SQSTM1 mutations and more common than many other genetic variants that have been associated with ALS [
2]. The mean age of disease onset in our cases was 60 years, with an average disease duration of 3 years; similar to what has been reported for ALS caused by the
C9orf72 mutation and for cases of ALS with no identified mutation [
28]. Only three of our seven probands (43%) had a family history of neurological disease, which is significantly less than for
C9orf72 and
SOD1 mutations [
28] and suggests that
TIA1 mutations may be variably penetrant. Interestingly, all of our
TIA1 mutations carriers were female, despite there being no sex bias in the case series in which they were identified. This finding is all the more striking, given that ALS, in general, is reported to be more common in men with a ratio of males to females of 1.7:1 [
21] and could indicate that some factor associated with biological sex affects the penetrance of
TIA1 mutations, similar to what has been reported for some other genetic subtypes of ALS/FTD [
6]. This association will need to be investigated further in additional case series and we note that one affected member of UBCU2 was male, although no source of DNA was available to confirm his genetic status.
The initial clinical features of our
TIA1 mutation carriers varied, with five presenting with symptoms of ALS, three with early changes of FTD and one who was asymptomatic but found to have abnormal memory function of neuropsychological testing (Table
1). Of the eight cases that developed ALS symptoms, bulbar weakness was a prominent feature in six (75%), which is more common than in cases of sALS (26%) or those with the
C9orf72 mutation (44%) [
28]. Of the five patients who presented with, or later developed FTD, the initial and most prominent problem was expressive aphasia in four, whereas only one patient showed mainly behavioral abnormalities. The high frequency of primary progressive aphasia (PPA) in our
TIA1 mutation carriers is different from those with the
C9orf72 mutation whose FTD phenotype is more often bvFTD [
11]. It is important to note that the case series used to identify mutation carriers had a primary ALS diagnosis with only a small proportion (< 4%) also having FTD, which may underestimate the incidence of
TIA1 mutations in patients with mainly FTD. Interestingly, none of our
TIA1 mutation carriers developed significant parkinsonism or psychotic features, both of which are commonly reported in series with the
C9orf72 mutation [
11,
27].
It is intriguing that another founder mutation affecting the LCD of
TIA1 (E384K) was previously identified in Swedish and Finnish populations to cause WDM, which is characterized by late-onset slowly progressive weakness of hand and distal leg muscles and is associated with a rimmed vacuolar myopathy which is TDP-ir [
10,
15]. To our knowledge, no patients with WDM have been reported to also develop ALS or FTD and none of our ALS patients with other
TIA1 mutations had a personal or family history of muscle disease. It is possible that something about the specific amino acid substitution caused by the E384K mutation leads to a distinct and selective phenotype, or that the expression is influenced by other genetic factors common to the specific ethnic population in which the WDM
TIA1 mutation occurs [
15]. However, given the fact that the E384K mutation is reported to show similar effects to the ALS associated
TIA1 mutations in biophysical and cell culture studies [
19], it is also possible that greater overlap in the clinical features exist but has not yet been recognized due to referral or reporting biases. This might not be unexpected given the number of other genes (including
VCP,
SQSTM1,
HNRNPA1,
HNRNPA2B1 and
MATR3), in which mutations cause clinical syndromes, referred to as multisystem proteinopathies, with ALS, FTD, inclusion body myopathy and bone disease each showing variable penetrance [
29].
Overall, the neuropathology of our patients with
TIA1 mutations was characterized by chronic degenerative changes, primarily involving the pyramidal motor system, prefrontal neocortex and substantia nigra, with more anatomically widespread TDP-ir pathology (Table
2, Figs.
2 and
3). In all four cases with clinical FTD, the pattern of neocortical TDP-ir pathology fit best with FTLD-TDP type B and was similar to what is found in most patients with FTD combined with ALS, including those with the
C9orf72 mutation [
18]. None of the
TIA1 mutation cases showed severe degeneration of the caudate nucleus or hippocampal sclerosis which are both common in other genetic causes of FTLD-TDP (e.g.
C9orf72 and
GRN mutations) [
11,
20].
The most striking aspect of the pyramidal system pathology in the
TIA1 mutation carriers was the number of round, sometimes LBL inclusions seen in LMN of the spinal cord and medulla with HE stain (Fig.
2). Although similar round inclusions were also found in some cases of sALS and
C9orf72+ cases, they never averaged more than one per tissue section and were completely absent in most cases. In contrast, round and LBL inclusions were a consistent feature in the
TIA1 mutation carriers, with at least one, and often multiple examples, present in each section of spinal cord and medulla. Consistent with this was the finding of significantly more compact round TDP-ir NCI in LMN in the
TIA1 mutation cases, which were often of a similar size and shape as the round/LBL inclusions seen on HE stain. This finding suggests that frequent round/LBLI in LMN may represent a pathological signature of ALS-TDP caused by
TIA1 mutations and that the formation of these particular inclusion bodies may somehow be related to the altered SG dynamics that has been shown to be associated with expression of the mutant
TIA1 protein [
19]. Although this is somewhat speculative, it is interesting that cases of ALS caused by mutations in the RNA-binding protein FUS are also characterized by large round cytoplasmic inclusions in LMN that are visible with HE stain (although generally more basophilic) and that these have been proposed to form in persistent SG [
7]. Round, hyaline, LBLI have also been described in some (but not all) cases of ALS caused by
SOD1 mutations [
13,
23], where they are composed of misfolded SOD1, rather than TDP-43, and may be induced by ER stress [
33].
The pathomechanism that has been proposed for ALS associated with
TIA1 mutations is that the amino acid change in the LCD enhances its intermolecular interaction, which promotes LLPS and results in SG that are more persistent, thus creating an environment where the contents are more likely to begin to aggregate and become insoluble [
19]. Although TDP-43 is one of the most abundant and most aggregate prone constituents of SG [
1], this model raises the possibility that the resulting pathological inclusions might also contain TIA1, other SG markers, and other RBP that are typically stored in SG. However, our IHC and IF studies failed to demonstrate any co-localization of TIA1 with TDP-43 in the inclusions, and also failed to show any abnormal accumulation of other RBPs in cases with
TIA1 mutations. Although some other studies have suggested that TIA1 may co-localize with TDP-43 in ALS [
17,
32], others have refuted this finding [
4]. Consistent with our immunostaining results are previous biochemical studies that failed to show any enrichment of TIA1 in the insoluble protein fraction extracted from post mortem brain tissue of
TIA1 mutation carriers [
19]. This suggests that even if TDP-43 begins to form insoluble aggregates in SG, further protein aggregation may occur independent of the presence and function of TIA1. None-the-less, it is possible that the commercial TIA1 antibodies we used are not sufficiently sensitive for use in post mortem tissue that has undergone prolonged fixation and that further TIA1/TDP-43 co-localization studies are warranted.