Abstract
Frontotemporal lobar degeneration (FTLD) is a clinically heterogeneous neurodegenerative disease with a strong genetic component. In this review, we summarize most common mutations in MAPT, GRN, and C90RF72, as well as less common mutations in VCP, CHMP2B, TARDBP, FUS gene and so on. Several guidelines have been developed to help gene testing based on genotype–phenotype correlation, the underlying histopathological subtypes, and the neuroanatomic associations. Furthermore, we also summarize molecular pathways implicated by genes and novel targets for FTLD prevention and management in recent years.
FRONTOTEMPORAL LOBAR DEGENERATION
Frontotemporal lobar degeneration (FTLD) is a collective term depicting a class of a clinically, pathologically, and genetically heterogeneous group of diseases, which is associated with either symmetric or asymmetric selective degeneration of the frontal and temporal lobes in our brain. It generally gives rise to frontotemporal dementia (FTD), an early onset dementia with progressive deficits in behavior, executive function, and language [1].
Clinically, FTLD can be classified into three main subtypes: behavioral variant frontotemporal dementia (bv-FTD), nonfluent variant primary progressive aphasia (nfv-PPA), and semantic variant primary progressive aphasia (sv-PPA). Bv-FTD shows deficits early in behavioral and executive function. NFV-PPA typically exhibits progressive speech, grammar, and word output deficits. SV-PPA is characterized by damage in word and object meaning but with spared repetition and speech production. However, we should note that the diagnostic criteria outline is changing along with time in view of the new discovery of mutations and the use of progressed imaging technologies. In addition, FTLD may overlap clinically with motor neuron disease, the Parkinsonian syndromes, and corticobasal syndrome (CBS) as well as progressive supranuclear palsy (PSP). On the whole, these various phenotypes well highlight the clinical heterogeneity of FTLD [2].
Generally, consensus opinion currently recognizes FTLD-tau, TDP-43, and FUS as the vast majority of pathological sources of disease. Patients with inclusions containing hyperphosphorylated tau protein are referred to as FTLD–tau, which was the first pathology to be recognized [3]. Inclusions consisted of Trans-activating DNA binding protein are referred to as FTLD-TDP [4], which can be subdivided as types A–D depending on the distribution and relativeabundance of the inclusions [5]. Fus and other protein family members (regarded as FET proteins) are referred to as FTLD–FUS. They are the main ubiquitinated proteins that are TDP-43-negative [6].
GENETICS
Up to 40% (varying from center to center) of FTLD cases show a positive family history of dementia, indicating a powerful genetic component. Mutations in the MAPT, GRN, and C9orf72 hexanucleotide repeat expansion account for nearly half of familial FTLD. In contrast, other genes are responsible for a minority of familial cases [7]. (Table 1)
Common mutations
MAPT
Microtubule-associated protein tau gene (MAPT) was identified as the first causative gene in the FTD with Parkinsonism (FTDP-17) in 1998 [8]. MAPT encodes the tau protein critical for the microtubule stabilization and axonal transport. Six tau isoforms are expressed in adult human including three microtubule-binding repeats (3R-tau) and four microtubule-binding repeats (4R-tau). The expressing level of 3R-tau is equal approximately to that of 4R-tau in healthy adult cerebral cortex. So far, more than 50 causal MAPT mutations have been described and its frequency is up to 11% in familial FTD cases [9]. MAPT mutations are mainly clustered in exons 9–13 and they are classified into two broad types: missense mutations and splice site mutations. The former alters the amino acid code of the protein which is generally in or adjacent to the 4R-tau region of the microtubule-binding domains. The latter changes the inclusion of exon 10, which may change the ratio of 4R-tau to 3R-tau [10].
The clinical manifestation of FTLD patient with MAPT mutations may be in accordance with bvFTD, and/or language syndromes and often times have extra-pyramidal features (Parkinsonism), with a mean onset (48 years; range, 38–58 years), Mean survival (9 years; range, 3.5–14.5 years) [7]. MAPT mutations have also been linked with other tau pathologies including PSP, CBS, argyrophilic grain disease (AGD), and so on [11].
GRN
In 2006, mutations in the progranulin gene (GRN) were recognized as one that was responsible for those familial FTLD cases lacked MAPT mutations and were demonstrated tau-negative histopathology [12]. GRN is located approximately 6.2 Mb from the MAPT on chromosome 17q21 coincidentally. Progranulin (PGRN), the protein product, is predominantly associated with frontal dysfunction [13] but its roles in the brain are still largely unclear apart from the role in neurite outgrowth, synapse structure and function, inflammation, response to stress and lysosomal function [14]. GRN mutation can explain about 5% –11% of familial case and more than 150 variants of GRN have been reported so far [15]. These mutations includes frameshift, splice-site, and nonsense mutations, deletion and defective protein sorting [9].
Histopathologically, GRN-mutated FTLD cases are consistently with FTLD-TDP type A [5]. In term of clinical features, GRN variants carriers present with extremely heterogeneous phenotypes, even among individuals or families with the identical mutation, ranging from typical FTLD presentations such as bvFTD and nfvPPA, to mild cognitive impairment [16] and psychiatric disorders [17]. Age at disease onset is extremely wide with a mean average in 58 years and disease duration (7 years, range, 3–8 years) [7]. In other neurodegenerative disorders, such as AD, PD, CBS, [18], and Lewy body dementia [19], GRN mutations are also described. It is likely that the alterations of RAP1GAP mRNA levels may explain the clinical variability of GRN-FTLDpatients [20].
C9ORF72
A hexanucleotide GGGGCC repeat located either upstream the exon 1 or in the first intron of the C9ORF72 gene (depending on the transcript variant) was identified in c9FTLD/ALS in 2011 [21]. In the normal population, the size of the G4C2 repeat typically lower than 25 repeats whereas repeat expansions in patient can reach several hundreds to thousands repeats. Although most studies agree that up to 60 repeats are considered pathogenic, the smallest risky hexanucleotide expansion remains unknown [22]. The size of the expanded repeats even varies from tissues to tissues in a single individual, and a hypothesis of multiple origins for the expansion is supported. Small expansions might be regarded “pre-mutations” to pathologic expansions and is prone to expand in the next generation [23]. We are recommended to combine repeat primed-PCR and southern-blotting to assess the number of the expanded repeat now [24]. Although frequency varies geographically, G4C2 hexanucleotide repeat expansion can explain approximately 40% familial ALS, 8–25% of familial FTLD, and 30% familial FTD-ALS, thus representing the biggest single genetic cause of familial FTLD and ALS worldwide [7].
The C9ORF72 gene is expressed as three major transcripts variants (V1, V2, and V3) but only two different protein isoforms are produced as two mRNAs undergo identical translation [21]. C9ORF72 protein is a full-length homolog of DENN proteins [25], which act as GDP-GTP exchange factors for Rab-GTPases. A proposal that C9ORF72 regulate autophagy and endosomal trafficking because of the co-location with Rab proteins and membrane vesicle is supported in one study. This study assumes the function of C9ORF72 in membrane trafficking is disturbed in C9ORF72 disease because the interaction between mutant C9ORF72 and Rab7 and Rab11 is enhanced [26]. In addition, another study indicates a role for C9ORF72 in nuclear-cytoplasmic transport on account of its interaction with hnRNP A1 and hnRNP A2/B1 [27]. The normal and pathological role of the C9ORF72 protein is still known little and the study of modifiers of C9ORF72-disease may help us to get the detail about the protein. To date, we have discovered a number of potential modifiers of age at onset, survival and phenotype such as ATXN2, TMEM106B, and C9ORF72 repeat size [29] and its hypermethylation [30].
C9ORF72 mutations carriers were claimed with a range of neuropathology (TDP-43 pathology, p62 pathology, microglial pathology, ubiquilin (UBQLN) pathology, RNA-binding protein pathology and pathology related with dipeptide-repeat (DPR) proteins) [31]. Nevertheless, cases with C9ORF72 expansions are predominantly consistent with TDP-43 type A or type B [5]. In addition, the pathology with p62 positive, TDP-43-negative inclusions is a highly specific feature of C9ORF72 pathology [31].
Clinically, C9ORF72 repeat expansions carriers present with bvFTD as predominant phenotypes. Besides typical symptoms, there are more recent studies concerning the psychotic manifestations. The C9ORF72 expansion is supposed to be a rare genetic cause for schizophrenia and bipolar disorder disorders [32, 33]. In particular, we should assess the prevalence of the C9ORF72 expansion in the late-onset psychosis patients carefully compared to the early-onset psychosis [34]. Psychiatric symptoms involve early disinhibition, lack of insight, hallucinations, paranoid ideation delusion, anxiety, hyperorality, early apathy, loss of empathy, and obsessive-compulsive behaviors [35]. Among them, paranoid ideation, hallucinations, and delusions are more frequently identified in C9orf72 hexonucleotide repeat patients [36]. Some cases expand phenotype to rare manifestations, including eating disorder and constructional apraxia, which further manifest multiple clinical syndromes of C9orf72 mutation carriers [37, 38]. Regretfully, the relationship between schizophrenia and C9orf72 expansions remains unclear. One study claimed that a 10 base pair gene deletion influence the occurrence of the psychosis related with the C9orf72 gene via the modified pathological process [39]. The second most common subtype is nfvPPA but svPPA has unusually been seen in C9ORF72 expansions [40]. C9ORF72 expansion carriers developed FTLD at an early age (average, 55 years; range, 33–75 years), mean survival (4.5 years; range, 3–10 years) [41]. C9ORF72 mutations carriers mainly show classical clinical presentations of FTD in the form of earlier onset age, earlier death, and a shortened survival as well as a higher frequency of positive family history [42]. In addition, the disease-associated expansions were also discovered in some motor diseases (ALS, CBS, Parkinsonism, HD phenocopies, Olivopontocerebellar and Demyelinating disorders) and some other non-motor syndromes (AD, CJC-like phenotype, and so on) [35]. One reason that may be responsible for the high phenotypic variability of C9orf72 repeat expansions is that additional genetic factors including UNC13A, ATXN2 and NIPA1determinephenotype [43].
Uncommon mutations
VCP
The valosin containing protein gene (VCP) mutation was identified in inclusion body myopathy associated with Paget disease of bone and frontotemporal dementia (IBMPFD), and the term of ‘multisystem proteinopathy’ is recommended to describe this phenotype recently [44]. It encodes a member of ATPase family on chromosome 9p13.3, which participate in multiple cellular processes like quality control of mitochondria, suppression of apoptosis, stress response, mRNA–protein complexes (mRNPs) remodeling and so on [45]. Today, 17 different mutations have been described, and all IBMPFD mutant residues localize in a region at the interface between the N- and D1- domains [46]. However, VCP mutations are rare and account for less than 1% of the family frontotemporal lobar degeneration (F-FTLD) [9].
FTLD patients with VCP mutations present with TDP-43 proteinopathy type D. Some inclusions also stain for VCP protein p97 [47]. Myopathy and Paget’s disease are found respectively in about 90% and 50% of affected subjects, whereas classical symptoms, mainly bvFTD and SD, is found in about 30%, usually many years after the occurrence of muscle symptoms [48]. VCP mutation phenotypes may also contain motor neuron degeneration in the shape of familial ALS and Parkinsonism [49].
CHMP2B
A few Danish and Belgian FTLD families show variants in the chromatin modifying 2B gene (CHMP2B) at chromosome 3p11.2. The gene codes a component of the heteromeric endosomal sorting complex required for transport (ESCRT III complex), associated with the endosomal–lysosomal and autophagic protein degradation pathway [50]. CHMP2B containing ESCRT-III complexes can also regulate synaptic plasticity at dendritic spines, and may act as a new element of the submembrane cytoskeleton of spines [51]. However, its functions are still not established well. At present, only several missense variants have been described [52]. Syntaxin 13 serves as a genetic modifier of mutant CHMP2B in FTLD and participates in autophagosome maturation [53]. In addition, we uncover the small endosomal GTPase Rab8, regulators of synaptic growth responses, as another distinct modifier of the FTLD-associated CHMP-2BIntron5 protein in a Drosophila model [54]. Neuropathologically, CHMP2B mutations carriers are consistent with a hereditary form of frontotemporal dementia (FTD-3) as FTLD-UPS [55]. The main clinical manifestation in FTLD-CHMP2B mutation carriers is bvFTD. Clinical signs of motor neuron impairment (parkinsonism, dystonia, myoclonus, and pyramidal signs) develop in the late of disease procession [56]. The age at onset ranges between 46 and 65 years with an average of 58 year and the mean disease duration is nearly 10 years with great variability in the Danish family [57].
TARDBP
TARDBP encoding for a 43-kDa ubiquitously is expressed nuclear DNA- and RNA- binding protein (TDP-43) on chromosome 1p36.22 and has been recognized as a causative gene for ALS. TDP-43 played a role in regulating transcription, splicing, microRNA biogenesis, and RNA transport [58]. So far, more than 30 TARDBP mutations have been found and the pathogenic mutations may have partial loss-of function properties [59]. TARDBP mutations explain approximately 4% –6% of familial ALS cases and <1% of F-FTLD cases with differences in prevalence evident geographically [7]. The main clinical feature associated with TARDBP mutations is ALS, FTD-ALS, or isolated FLTD is rare instance. TARDBP mutation carriers develop the degeneration in an average of 54 year (ranging from 35–74 years), and the mean disease duration is 3 years (1–6 years) [60].
FUS
Mutations in the FUS gene were originally described on chromosome 16p11.2 in familial ALS. The protein production, fused in sarcoma protein, has a crucial role in the regulation of GluA1 mRNA stability, function of post-synaptic, and FTLD-like animal behaviors [61]. Missense, but splicing, in-frame insertions, and deletions are the main mutations identified [62]. Mutations in FUS are the cause of 4% familial ALS and rarer familial FTLD cases [7]. There is no strong evidence that FUS is genetically involved in FTLD but an accumulation of FUS protein in inclusion bodies was associated with three clinicopathological subtypes of FTLD.
UBQLN2
Ubiquilin2 (UBQLN2) is located on chromosome Xp11.21, which is responsible for an X-linked form of ALS and FTLD/ALS and also rare sporadic ALS [63]. It encodes ubiquilins that mediate the degradation of proteasome-dependent protein. Neuropathologically, ubiquilin 2 might in some respects be involved in the TDP-43 pathway or Ubiquilin-2. Clinically, bvFTD often occurs ahead of motor symptoms [7].
CSF1R, SQSTM1, hnRNPA2B1 and hnRNPA1 genes
Colony stimulating factor 1 receptor (CSF1R) gene mutations on chromosome 5q32 were reported in North American families with hereditary diffuse leukoencephalopathy with spheroids (HDLS) [64]. The mutations carriers mainly present with bvFTD, CBDS, and multiple sclerosis [65]. The sequestosome 1 gene (SQSTM1) is located on 5q35 and codes for protein p62. Its mutations are consistently correlated with p62 and TDP-43 neuropathology [66]. Mutations in SQSTM1 mainly lead to Paget’s disease of bone and behavioral disorder and can explain about 2% –4.4% of F-FTD [67]. Similar to SQSTM1, mutations in heterogeneous nuclear ribonucleoprotein A2/B1 (hnRNPA2B1) and heterogeneousnuclear ribonucleoprotein A1 (hnRNPA1) were also identified in families presenting with multisystem proteinopathy (MSP) [18]. While the frequency of mutations in hnRNPA2B1 and hnRNPA1 is still unknown.
TBK1
Most recently, the TRAF family member-associated NF-kappa-B activator (TANK)-binding kinase 1 (TBK1) gene was reported as an ALS and FTD gene [68]. TBK1 is a vital serine/threonine kinase of the IB kinase family involved in autophagy and inflammation. It consists of a series of substrates such as optineurin (OPTN) and p62 (SQSTM1) [69]. In a whole-genome sequencing research, we identified five variants in FTLD-TDP patients who are negative for C9ORF72 and GRN [70]. Neuropathologically, TBK1 carriers are consistent with TDP-43 pathology. Clinically, in a Belgian cohort, patients of FTD with TBK1 mutation show bvFTD especially disinhibition as the most frequent feature, psychiatric and extrapyramidal symptoms occur rather often in the TBK1 carriers. In addition, loss in memory and disorientation in time and/or space were common early in the initial disease course. In all, TBK1 carriers have a later age at onset (63.3 years on average in FTD patients) and the disease duration is relatively long, particularly in patients with FTD (8.2 ± 4.9 years) [71].
Furthermore, mutations in APOE [48], TMEM106B [72], KIF24 [73], UBAP1 [74], TREM2 [75], CHCHD10 [76], ATXN2 [77], novel N-terminal domain mutation in prion protein [78], DCTN1 [79], and so on, also represent a risk factor for FTLD. However, susceptibility genes contributing to the development of FTLD are still little known and are required to be explored further.
GENETIC AND NEUROANATOMIC ASSOCIATIONS
We have identified several genetic and imaging associations in living FTLD patient with novel multivariate tools, eigenanatomy, and sparse canonical correlation analysis (SCCAN) lately. Mutations in MAPT, GRN, C9ORF72, and the less common genetic cause of FTLD show different patterns of atrophy. In MAPT mutation carriers, symmetrical atrophy can be observed in the anteromedial temporal lobes, which is also observed in the orbitofrontal cortex and insula, but rarely observed in the anterior cingulate or parietal lobes. GRN mutations are related with asymmetrical anterior temporal and inferior frontal atrophy [80]. The neuroimaging profile of the C9ORF72 expansion links with symmetrical frontotemporal changes but also posterior cortical, cerebellar, thalamic volume reduction, and it is significantly more symmetrical than GRN mutations and involves significantly less temporal lobe than MAPT mutations [81]. Furthermore, the rate of volume change in GRN mutation carriers is greatest among the three main mutations and the rate of volume change in MAPT is similar to that in C9ORF72 mutation carriers [82]. Atrophy patterns of VCP mutations is in frontal, temporal, and parietal lobes, especially prefrontal cortex and superior temporal gyrus; hippocampus, caudate nucleus, amygdala. TARDBP mutation carriers show symmetrical frontal, orbitofrontal cortex and temporal atrophy. Frontal and temporal atrophy with striking striatal atrophy is the feature of FUS mutations [2]. CHMP2B mutations generally involve cortical atrophy, most marked in frontal, parietal–occipital lobes. MRI finding of SQSTM1 is frontotemporal atrophy and may be asymmetric [7]. Brain MRI of CSF1R reveals extensive, predominantly frontal white matter T2 hypersignals, but a reduced white matter involvement was also reported in some cases [65]. As to TBK1 gene, both various patterns of atrophy and hypoperfusion or hypometabolism can be discovered [71]. While the rest gene and neuroanatomic associations of FTLD remains to be researched. Moreover, a recent study has identified some associations between brain regions and SNPs. For example, rs8070723 (in MAPT) is related with gray matter variance in temporal cortex, which may be used to improve the diagnostic accuracy of FTLD [83].
MECHANISMS IMPLICATED BY GENES
MAPT mutations
Potential mechanisms in MAPT-mediated neurodegeneration mainly include tau phosphorylation, tau aggregation, mitochondrial dysfunction, oxidative damage, amyloid pathogenic hypothesis, and neuroinflammation. The hyperphosphorylated tau damages microtubule stabilization and axonal transport (Fig. 1) [84]. Axonal dysfunction is suggested to be the main source of tau-induced detrimental effects, which can be intensified by impaired retrograde transport via the dynein/dynactin [85]. Mitochondrial dysfunction caused by the hyperphosphorylated tau, the truncated tau, and tau oligomers could affect the ATP production and lead to substantial oxidative damage. Consequently, it results in neuron death and aggravates neurodegeneration [84]. Tau aggregates can be released from neurons and extracellular tau could activate microglia and astrocytes, which cause neuroinflammation. Recently, a study suggested that amyloid was also associated with familial FTLD and its pattern was more like AD [86]. Tau and Aβ synergistically cause tauopathies. Tau could participate in the Aβ-induced pathological process and cause loss of synapse, damage in mitochondrion, and even death of neuron finally. On the contrary, Aβ could also result in aberrant alternative splicing of tau and high level of tau phosphorylation [84].
GRN mutations
All pathological GRN mutations identified so far are confirmed or predicted loss-of-function alleles, which can be reflected by a nearly 50% reduction in the level of mRNA and 33% reduction in the level of protein in heterozygous GRN mutation carriers [14]. This suggests the loss of functional progranulin or haploinsufficiency is the disease mechanism. The underlying mechanisms of GRN mutations effects on reducing constitutive PGRN levels include subjecting PGRN to degradation, affecting the localization and secretion of the protein, nonsense-mediated RNA decay as well as the methylation of the GRN promoter [87, 88]. Regretfully, how haploinsufficiency leads to frontotemporal dementia is not unclear well and some hypotheses are put forward as followed. The loss of progranulin in neurons reduces neurite outgrowth. Extracellular progranulin is endocytosed via the endosomal-lysosomal pathway through the sortilin receptor. In neurons, progranulin co-localizes with the transmembrane protein TMEM106B in late endosomes and early lysosomes as well as with brain-derived neurotrophic factor that go through both anterograde and retrograde transport along axons (Fig. 1). Haploinsufficiency also causes an exaggerated neuroinflammatory, which increases the production of progranulin and the release of multiple cytokines in microglia such as interleukin-6 (IL-6) [14]. The altered cytokines and chemokines in cerebrospinal fluid can be signature of GRN mutation carriers [89]. Our laboratory indicates that PGRN haploinsufficiency has an association with impaired response to serum stimulation or cell cycle alterations in lymphoblasts bearing a null GRN mutation [90]. A complete GRN absence because of a homozygous GRN loss-of-function mutation leads to neuronal ceroid lipofuscinosis (NCL), which suggests FTLD-GRN may share common feature with lysosomal storage disorders [91]. Furthermore, progranulin deficiency increases sensitivity to H2O2, amyloid beta 1–42 (Aβ1–42) and NMDA-mediated toxicity in animal experiments and some kinase inhibitors such as the phosphatidylinositide 3-kinase (PI3K)/protein kinase B (Akt) [14]. In addition, PGRN haploinsufficiency is also demonstrated to cause inefficient cortical neuron generation [92] and contribute to phosphorylation and intraneuronal accumulation of tau in P301L tau transgenic mice [93]. GRN haploinsufficiency also could elevate levels of filamin C (FLNC), which involved in aging, neurodegeneration, and synaptogenesis [94]. Finally, cleaved granulin fragments can increase TDP-43 levels through a post-translational mechanism and may accelerate the disease in TDP-43 proteinopathies [95].
C9ORF72 expansions
So far, there are mainly three mechanisms proposed for the repeat expansion including haploinsufficiency due to the loss of transcription, RNA toxicity due to sequestration of RNA-binding proteins in RNA foci, and dipeptide protein accumulation from non-AUG translation. Nevertheless, how these mechanisms drive the process of the disease and which plays pivotal role in the development of neurodegeneration need further investigations (Fig. 2). In addition, some special structures should be taken into account when studying C9ORF72-linked neurodegeneration. For example, RNA*DNA hybrids (R-loops) participate in the instability of the repeat, dysregulation of transcription, epigenetic modification, antisense-mediated gene regulation [96].
Haploinsufficiency
There are convincing evidences that the expanded repeat can result in haploinsufficiency due to loss of transcription in blood, lymphoblasts, and fibroblasts, as well as in postmortem brain in C9ORF72 expansion carriers [97]. Furthermore, the fact that the reduction of C9ORF72 mRNA produces reduced axon lengths of motor neurons and locomotion deficit in zebrafish experiments and in Caenorhabditis elegans experiments supports the mechanism [28]. The reduced C9ORF72 expression of all transcriptsrather than just those containing the repeat sequence could be explained by epigenetic changes such as the hypermethylation of the CpG island [98] and the trimethylation of lysine residues of histones H3 and H4 [99]. G4C2-repeat itself could also be the site of methylation in large repeat length [100]. Additionally, silencing of the C9ORF72 transcript by aberrant DNA methylation and stable structural features formed by the expanded G4C2 repeats may also at least in part contributes to changes in C9ORF72 transcription [101]. However, reduction of C90RF72 homologue in mice does not produce motor phenotype suggesting the insufficiency of this mechanism to produce FTLD/ALS and we need wider range of specific antibodies to ensure the disease relevance of the C9ORF72 haploinsufficiency[102].
RNA-mediated toxicity
Spurred by the fact that RNA foci has been deemed to an important disease mechanism in other repeat expansion disorders and the fact that RNA foci have been observed in the C9ORF72 repeat expansion carriers, the similar mechanisms was supposed to exert an effect on C9ORF72 related FTLD/ALS [21]. GGGGCC expansions produce both mutant sense and antisense RNAs that form G-quadruplex structures [28], which can isolate intracellular heme and facilitate oxidative damages in ALS and FTLD [103]. In addition, G4C2 repeat-induced nucleolar stress could be thought as a possible additional factor in aberrant RNA processing [104]. One possible mechanism for RNA-mediated toxicity is RNA-binding proteins (RBP) sequestration and dysregulation, ultimately leading to cell death. The depletion of RBP, which is available for normal RNA metabolism, causes the reduction of specific transcripts and unconventional splicing patterns [105]. This mechanism is supported by the identification of different RBPs co-location with C9ORF72 RNA foci and the discovery of several proteins (hnRNPs, Pur-alpha, ADARB2, nucleolin) binding to various G4C2 or C4G2 repeat structural features such as G-quadruplexes or hairpins [101, 104]. We need further elucidate the correlation between RNA foci burden and neurodegeneration but this mechanism is thought to be a central mechanism in disease procession because overexpression of Pur-alpha rescued the locomotor phenotype and neurodegeneration in a fly model with a 30 repeats expansion presenting [28].
Dipeptide-repeat protein accumulations
Besides the recognized mechanisms in C9ORF72-associated FTLD mentioned above, accumulations of dipeptide repeats proteins (DPRs) have been uncovered as another potentially toxic mechanism. The non-ATG initiated translation induced by the stable RNA G-quadruplex structures was originally discovered in spinocerebellar ataxia type 8 (SCA8) [106]. Both sense and antisense transcripts of the expanded repeat can give rise to the production and aggregation of DPRs: poly-glycine-alanine (GA), poly-glycine-proline (GP) and poly-glycine-arginine (GR) in the sense (G4C2)exp transcript, poly-proline-arginine (PR), poly-glycine-proline (GP), and poly-proline-alanine (PA) in the antisense (C4G2)exp transcripts [107]. The hypothesis of the unconventional translation of the C9ORF72 locus is supposed by that matching genomic sequence of the C9ORF72 locus is revealed via sequencing of DNAse 1 treated pre-mRNA from post mortem tissue [108]. These DRPs are cytotoxic possibly via ubiquitin–proteasome system (UPS) dysfunction [109] and may be related with nucleolar stress and impaired stress granule formation [110]. The modifiers of DPRs suggest damages in nucleocytoplasmic transport, both import and export, may be another mechanism of DPR toxicity [111]. Moreover, the neurotoxicity of poly-GA expression is associated with loss of function of Unc119, contributing to selective vulnerability of neurons with DPR protein inclusions [112]. Poly-GA expression accompanied by caspase-3 activation also can impair neurite outgrowth, surpress proteasome activity, and induce endoplasmic reticulum (ER) stress [113]. DPRs distribution correlates with neuropathological subtypes rather than neurodegeneration spatially and links to transcriptional silencing [114]. Besides, Drosha was identified as a new DPR protein in c9FTLD-TDP and Drosha-positive neuronal cytoplasmic inclusions was more likely to colocalized with p62 and ubiquilin-2 than with TDP-43 pathology, which indicated its unique pathogenic role in the onset or progression in C9orf72-mediated FTLD-TDP [115].
Others
Other potential mechanisms were implicated by other genes or locus in the pathology of FTLD in addition to what we have discussed above. Patients carrying the APOE ɛ4 have a reduction in neuronal repair mechanisms. Moreover, in vitro and animal studies indicate APOE could affect FTLD via modulation of neuroinflammation, neuronal toxicity, and synaptic plasticity [116]. VCP is relevant to degenerative diseases via its functions in proteostasis and signaling pathways and ATPase hyper-activation may lead to IBMPFD [46]. FTLD/ALS-related proteins TDP-43 and FUS pathologies perhaps affect lncRNA-based mechanisms either by regulating their transcription and transcript stability or making lncRNAs exert their cellular function via binding to them [117]. Pathogenic UBQLN2 mutations affect its interaction with ubiquitin regulatory X domain-containing protein 8 (UBXD8), leading to the disruption of endoplasmic reticulum-associated protein degradation (ERAD) [118]. hnRNPA2B1 and hnRNPA1 are ribonucleoproteins involved in splicing, RNA trafficking, and stabilization, and they interact with TDP-43 [67]. A DNAJB6 mutation may cause FTD through the dysfunction of autophagy, as well as SQSTM1, UBQLN2, VCP, and OPTN [119]. 5-HTTLPR polymorphism on SCLA4 gene modulated the serotoninergic system and its impairment has been revealed as a key factor in FTD pathogenesis [120]. Possible pathways of TREM2 is inflammation, while CHCHD10 may linked with mitochondrial dysfunction [119]. Nevertheless, efforts are still imperatively needed to shed light on the pathomechanisms contributing to FTD.
CLINICAL APPLICATION
Potential genetic testing
FTLD genetic testing is still complicated due to its high heterogenicity and its overlap with other neurodegenerative disorders, though we can sequence numerous genes simultaneously with the advent of exome sequencing and whole-genome sequencing. So, we develop several guidelines here on the basis of genotype–phenotype correlation and strict correspondence between causative gene and histopathological changes to help decide which gene or genes to test first (Table 2).
In our practice, MAPT should be analyzed first in cases with onset before age 50, Parkinsonism, and familial PSP. In patients with both FTLD and ALS, C9ORF72 expansions should be analyzed first followed by testing first the TBK1gene in negative cases [68]. Occasionally, motor neuron disease can also occur in families with GRN mutations. Cerebellar involvement is another pointer to testing for C9ORF72 mutations [121]. GRN sequencing should thus be included in patients who show unusual white matter involvement [122]. For families with corticobasal syndrome, we suggest considering both GRN and MAPT. But, plasma progranulin levels is recommended to measure before GRN sequencing because it has been shown to be a good predictor of GRN mutations and is more cost-effective than GRN sequencing [123]. Both GRN and C9ORF72 mutations should be searched for if psychotic symptoms or familial PPA is part of the presentation [121, 124]. Moreover, TARDPB screening might be undertaken even in young patients with “pure” neuropsychiatric disturbances and with no evidence of neurodegenerative disease in the parental generation [125]. Mutations in VCP are more frequent than hnRNPA2B1 and SQSTM1 mutations in patients with multisystem proteinopathy, while Danish and Belgian ancestry is indicative of CHMP2B [121]. In terms of pathological data, MAPT should be analyzed first in cases with FTLD-tau, and GRN, C9ORF72, and VCP in cases of FTLD-TDP [124]. In all, a comprehensive family history, clinical presentation, underlying histopathological subtypes and imaging will all help prioritize which gene or genes to test.
Genetic counseling is useful in both symptomatic genetic testing and predictive testing, but it is essential for people who experience predictive testing to have a formal counseling with a geneticist [121]. Regretfully, despite these techniques transform the process of genetic testing and bring us closer to the ideal of personalized medicine, it will possible be confusion for us because of interpretation of rare and novel variants and discussion of all the tested genes in genetic counseling.
Modified therapeutics
Nowadays, potential drug targets are increasingly being uncovered with the improved understanding of molecular genetics and advancements in technology, which may make pharmacological strategies shift from a symptom-based approach towards treatment of the underlying disease process (Fig. 3).
Tau-directed therapeutics
Given the strong genetic associations of tau with FTLD, it is rationale to believe interventions that target tau protein abnormalities (hyperphosphorylation, aggregation, and overall expression) will be effective treatments for this disorder. Both glycogen synthase kinase-3beta (Temsirolimus, suppression of beta2-adrenergic receptors) and CDK5 (cyclin-dependent kinase-5) inhibitors (Tolfenamic acid, p5 peptide) can prevent tau hyperphosphorylation [126 –128]. The microtubule-stabilizing agent davunetide (NAP) may provide microtubule stability [129]. In addition, we have also investigated antioxidants and other mitochondrial targeted therapies in PSP and proved some promise in transgenic tauopathy models [129]. Methylene blue, as an antioxidant, plays its neuroprotective effects by preventing tau protein aggregation and improving energy metabolism through upregulation of Nrf2/ARE genes. The molecular chaperones of the heat shock protein 70 (HSP70) family can inhibit tau aggregation potently and alleviate tau-mediated toxicity on fast axonal transport [128]. Besides, tau is hyperphosphorylated and aggregates in both AD and FTLD, Therefore, drugs developed for tau in AD may be equally or more efficacious for treating Tau-FTLD. For example, inhibition of DYRK1A (dual specificity tyrosine phosphorylation-regulated kinase 1A) plays a key role in Aβ-mediated tau hyperphosphorylation in AD [130].
In addition, recent studies have uncovered several potential therapeutic methods in targeting tauopathies such as enhancing functions of NMDAR (NMDA receptors) [131], reducing tau levels or blocking neuroinflammatory pathways [132]. Besides, we can suppress human tau expression by evaluating the effectiveness of small interfering RNAs (siRNAs) [133] and by the use of Tolfenamic acid. The later promotes the degradation of the transcription factor Specificity protein 1 (Sp1) and ultimately reduced tau mRNA and protein, as well as its phosphorylated form and CDK5 levels [134]. Recently, there is increased interest in antibodies that could be used to prevent pathological tau protein abnormalities from transneuronal spreading and the development of tau vaccines [135, 136].
Progranulin-directed therapeutics
Based on the haploinsufficiency mechanism by which it contributes to disease, it is a promising therapeutic approach to restore PGRN levels either by increased production or reduced clearance. Amiodarone, one of the known drugs capable to selectively raise PGRN levels, however, was ineffective in monogenic GRN mutated patients in a pilottrial [137]. Researchers have recently tried to increase PGRN levels with the use of micro-ARN-29b in vitro [138]. TMEM106B presumably sequester progranulin from secretion and/or impair the endo-lysosomal pathway to decrease the rate of progranulin degradation and increase intracellular progranulin eventually [139]. Besides, inflammatory cascade is a potential pharmacological target and activation of nicotinic acetylcholine receptors (nAChRs) by nicotine or specific alpha7 nAChR agonists reduces neuroinflammation [140]. Some molecules such as the histone deacetylase inhibitor suberoylanilide hydroxamic acid, alkalizing compounds, and inhibitors of the vacuolar ATPase are able to alleviate the development of the disease related with GRN mutation through increasing intracellular and secreted progranulin concentrations [141]. FTD-PGRN would be an ideal population for testing the potency of drugs aimed at normalizing PGRN levels since FTD is the only known phenotype related with reduced PGRN. The development of progranulin-modulating therapies will be facilitated by a better comprehension of the complexity of progranulin biology in the brain.
C90RF72-directed therapeutics
Based on the assumption that the toxic RNA foci is at the crux of C9FTLD pathogenesis, we mainly pursue the therapeutic strategies that focus on neutralizing or degrading transcripts of the expanded repeat to alleviate downstream neurotoxic cascades. Antisense oligonucleotides (ASOs) were able to significantly reduce RNA foci and they perform their function through binding to the RNA and accelerating the degradation of the target RNA [142]. However, C9orf72 repeats transcribed in the antisense (C4G2) direction are not influenced by sense strand-targeting ASOs [102]. In addition to ASOs, small molecules also can abrogate RNA-toxicity via preventing foci formation or the interaction between these transcripts and RBPs depending on their tight binding affinities. Small molecules are capable of inhibiting RAN translation as well [143]. In addition, some of them are able to target the ER to protect against poly(GA)-induced toxicity [113] and disrupt RNA G-quadruplex structures to interfere in protein interactions [144]. Furthermore, we identified a network of effective modifiers of C9orf72 DPR toxicity such as karyopherins and effectors of Ran-mediated nucleocytoplasmic transport, which could antagonize DPR toxicity in c9FTLD/ALS [111]. Finally, C9ORF72 elevating drugs, reducing the TDP-43 protein and modifiers of disease progression (TMEM106B, C9orf72 promoter hypermethylation, the number of hexanucleotide repeats) are another viable therapeutic avenues aiming at C9ORF72 expansion carriers [145]. Whether levels of C9RAN proteins could evaluate disease activity and assess efficacy of therapies that target the expansions requires to be verified further.
In all, there are advantages to pursue drug development in FTLD due to its clinical and molecular features and the success in FTLD management may help to accelerate drug development for other neurodegenerative disorders. Although there are a large number of genetic, proteomic, clinical, and pathological data available, only few targets have been identified and FTLD is still an orphan disorder without disease-modifying therapies. Therefore, in addition to a thorough understanding of disease-related genetics and molecular mechanisms as well as the process of illness, advance in specific tools (animal models, compound libraries, etc.) and academic-industry collaboration also have a significant influence on facilitating drug development.
CONCLUSION
FTLD are highly complex and heterogeneous group of disorders in terms of its clinical features, neuropathologies, and genetic. Clinical manifestations range from behavioral and executive injury to language disorders and overlap with motor dysfunction as well as parkinsonian syndromes. The progresses in identification of FTLD genes especially MAPT, GRN, and C90RF72 have brought better understandings and classifications of the pathogenesis in FTLD. Moreover, they also facilitated the introductions of various underlying disease mechanisms implicated in the both main genetic cause (MAPT, GRN, and C9ORF72) and rare genes (VCP, CHMP2B, SQSTM1, CFS1R, and so on). Now we are deeply aware the importance of axonal transport, DNA/RNA metabolism, neuroinflammation, lysosomal pathway, and autophagy in the pathological processes of this neurodegeneration disorder. Consequently, the corresponding therapeutic agents are established though majority of them need to be clinical verifications. Additionally, several genetic and imaging associations have been found out, which plays an increasingly important role in genetic testing accompanied by the clinical presentation and underlying histopathological subtypes. Although remarkable increase in knowledge of the histopathology, mechanisms, the genetic diagnostic accuracy and treatments of the complicated disease, further studies are needed to completely to figure out several fundamentally questions such as the detailed relationship between lower progranulin and TDP-43 accumulation, the mainly mechanisms of C90RF72 responsible for degeneration, other risky genes or locus involved in FTLD and so on. More important, future studies should pay more attention to the presymptomatic stage of the disease. A recent cross-sectional analysis has already revealed that structural imaging changes around 10 years and cognitive changes 5 years before expected onset of symptoms could be identified in asymptomatic adults with a risk of genetic frontotemporal dementia. Here, the insula is supposed to be an early core of pathology because it was the first atrophic cortical area in the whole mutation group and was one of the earliest atrophic areas in the analyses of genetic subgroups. Moreover, the analyses confirmed and extended previous neuroimaging studies that there exists differences between each individual genetic group. In the MAPT group, temporal lobe and medial temporal structures (the hippocampus and amygdala) were the earliest areas affected. In the GRN group, the insula was affected initially, followed by the temporal and parietal lobes. In the C9orf72 group, the thalamus and more posterior cortical areas were influenced originally [146]. Such findings could help to facilitate the insight into disease biomarkers that can stage presymptomatic disease and track disease process, which will be helpful to the cures and preventions of the disease.
Footnotes
ACKNOWLEDGMENTS
This work was supported by grants from the National Natural Science Foundation of China (81471309, 81371406, 81171209, 81571245, and 81501103), the Shandong Provincial Outstanding Medical Academic Professional Program, Qingdao Key Health Discipline Development Fund, Qingdao Outstanding Health Professional Development Fund, and Shandong Provincial Collaborative Innovation Center for Neurodegenerative Disorders.
