In the late 1990s, linkage at chromosome 17q21 was found for multiple families in FTD and parkinsonism linked to chromosome 17 (FTDP-17). Three research groups found mutations in and around MAPT exon 10 and proved that dysfunction of the protein tau is sufficient to cause neurodegeneration and dementia [16–18]. Accumulation of tau occurs primarily as insoluble fibrils in neuronal cell bodies (neurofibrillary tangles (NFTs)) and processes (neuropil threads or dystrophic neurites), but can also accumulate in astrocytes and microglia. Tau deposits are characteristic for a large number of neurodegenerative diseases known as the “tauopathies,” which include Alzheimer’s disease (AD), Pick’s disease (PiD), PSP, corticobasal degeneration (CBD), and AGD [30]. In AD, all six brain isoforms of tau are found in inclusions [31], and in FTDP-17, inclusions are heterogeneous, with different mutations associated with a different isoform composition. For PSP, CBD, and AGD, inclusions consist mainly of 4-repeat tau [32–34], and PiD has predominantly 3-repeat tau deposits [35]. In populations of European descent, MAPT is characterized by two haplotypes, H1 and H2, which result from a 900-kb inversion polymorphism. Inheritance of the H1 haplotype is a risk factor for PSP, CBD, and idiopathic Parkinson’s disease. The H2 haplotype is associated with increased expression of exon 3, suggesting that inclusion of this exon might be protective [36–39].

Tau proteins are microtubule-associated proteins that are abundant in the central nervous system (CNS) and mainly concentrated in axons [40]. The MAPT gene is located on chromosome 17q21 and consists of 16 exons, as depicted in Fig. 5.1. In adult human brain, alternative mRNA splicing of exons 2, 3, and 10 produces six tau isoforms ranging from 352 to 441 amino acids. Three of these isoforms contain 4 microtubule-binding repeats (4R), whereas the other three contain only 3 microtubule-binding repeats (3R). The extra repeat domain is encoded by exon 10. The presence or absence of a 29 amino acid (exon 2) or a 58 amino acid (exons 2 and 3) insert in the N-terminal half of the protein divides the isoforms further in 0N, 1N, and 2N (Fig. 5.1). The ratio of 4R and 3R tau isoforms in the cerebral cortex of healthy adults is equal, while in fetal brain, only 3R0N is expressed. Most adult rodents express isoforms with four repeats [41]. Tau is enriched in neurons and has an important function in promoting microtubule assembly and stability [42–44]. The C-terminal repeat domains bind microtubules directly, and 4R tau promotes microtubule assembly 3-fold stronger than 3R tau [45, 46]. The N-terminal inserts are part of the projection domain, establishing the interaction of tau and microtubules with the plasma membrane [47, 48]. In addition to its role in microtubule assembly, tau mediates transport of vesicles and organelles along the microtubules by regulating attachment and detachment of motor proteins [49, 50]. The motility of motor proteins dynein and kinesin is differentially regulated and dependent on tau concentration and isoform, pointing to a balanced spatiotemporal regulation of axonal transport [51]. Literature also suggests a physiological role for tau in dendrites, which might be an important function mediating synaptic impairment early in the disease process [52, 53].

Phosphorylation is the major posttranslational modification of tau protein, and in diseased brain, tau is abnormally hyperphosphorylated, generating paired helical filaments (PHFs) and neurofibrillary tangles [54–56]. The degree of tau phosphorylation influences its interaction with microtubules. Abnormal hyperphosphorylated tau sequesters normal tau, leading to misfolding, co-aggregation, and inhibition of microtubule assembly [57]. Microtubule-associated proteins 1 and 2 are also sequestered from pre-assembled microtubules, leading to their disruption [58–61]. There are 80 serine/threonine and 5 tyrosine potential phosphorylation sites, and normal brain tau contains 2–3 mol of phosphate per mole of tau protein. Mass spectrometry and antibody staining have identified more than 40 serine/threonine hyperphosphorylation sites of tau in AD [62, 63]. The phosphorylation of tau itself is developmentally regulated, and fetal tau is more phosphorylated than adult brain tau [64]. Several protein kinases are known to phosphorylate tau, and hyperphosphorylation probably involves coordinated action of several of them [65, 66]. When tau hyperphosphorylation induces dissociation from microtubules, it is redistributed from axonal to somatodendritic compartments. The increased pool of hyperphosphorylated tau is thought to promote assembly into fibrillar aggregates. In addition to phosphorylation, tau undergoes other posttranslation modifications, such as acetylation, glycation, glycosylation with O-linked N-acetylglucosamine, nitration, ubiquitination, sumoylation, prolyl isomerization, and truncation, all of which affect tau phosphorylation and/or aggregation [30].

In a large number of families linked to chromosome 17q21, mutations in MAPT could not be identified. In 2006, systematic analyses of the 17q21 region led to the discovery of mutations in GRN as a cause of FTLD-U [19, 20]. All GRN-associated FTLD patients present with TDP-43-positive inclusions in the affected cortical regions and basal ganglia (FTLD-TDP), often accompanied by irregular and short dystrophic neurites [95, 96]. The inclusions are found in neurons and glial cells and can be cytoplasmic as well as intranuclear. NFTs, neuritic plaques, and Lewy bodies have occasionally been observed in GRN brains [97–100].

In humans, the GRN gene is located upstream to the MAPT gene and comprises 12 coding exons (Fig. 5.2) [101]. The gene encodes progranulin, a 593-amino-acid-long protein containing a secretory signal sequence and seven and a half repeats of a unique 10–12 cysteine-containing motif. This motif is highly conserved in evolution and a reminiscent of the epithelial growth factor [102–105]. Progranulin is expressed in several tissues of the periphery as well as in neurons and microglia in the central nervous system, and little or no expression has been detected in adult astrocytes and oligodendrocytes [102, 106–108]. A GWAS implicated sortilin (SORT1) as an important regulator of GRN levels in human plasma, and cellular studies showed that SORT1 is a major neuronal receptor for GRN [109, 110]. SORT1 facilitates extracellular GRN clearance via endocytosis and delivery to the lysosomes. When GRN interaction with SORT1 is impaired, it is possible to restore extracellular GRN levels in FTD patient-derived lymphocytes and induced pluripotent stem cells (iPSC) differentiated to neurons [111].

GRN plays a key role in several biological processes including cell growth, wound repair, inflammation, and neuron development [112]. Studies in different cellular models show that GRN expression enhances neuronal survival and promotes neurite outgrowth and branching, while siRNA-mediated GRN knockdown results in reduced neurite arborization and dendritic protrusions [113–117]. The exact mechanism through which GRN exerts its neurotrophic function is not completely understood; however, compelling evidence suggests the involvement of the phosphatidylinositol 3-kinase/Akt signaling pathway and glycogen synthase kinase-3 beta regulation [118–120]. GRN-deficient cells show increased caspase activation, decrease in cellular survival, and vulnerability to a number of cellular stressors, including oxygen and glucose deprivation, oxidative stress, and sublethal doses of N-methyl-D-aspartic acid, kinase inhibitors, and proteasomal inhibitors [118, 121–123].

It has been hypothesized that GRN performs a neuroprotective role in the CNS by stimulating production of anti-inflammatory Th2 cytokines and supporting neuroprotective inflammation by binding to TNF receptors [124, 125]. The anti-inflammatory effects of GRN are evident in Grn−/− mice that show an exacerbated inflammatory reaction [123]. Moreover, increased GRN expression has been observed in microglia following a variety of acute and chronic insults in several neurodegenerative disorders including Creutzfeldt-Jakob disease [126], ALS [127, 128] and FTLD [99]. It also attracts and activates microglia, enhancing endocytosis of extracellular peptides such as amyloid β (beta) 1–42 [129]. Interestingly, microglia of patients with GRN mutations are able to upregulate GRN despite reduced levels in blood, cerebrospinal fluid, and neurons in unaffected brain regions [99] suggesting GRN plays a central role in the inflammatory response.

In 2011, two independent consortia identified the pathological expansion of a noncoding GGGGCC hexanucleotide repeat in C9orf72 as a cause of familial FTD and ALS, explaining several linkages and GWAS performed since 2006 linking chromosome 9p to these disorders [21, 22, 153, 154]. The findings where confirmed by other studies in a wide series of populations [155, 156], and repeat expansions for C9orf72 have since then also been confirmed in families affected with AD, CBS, and ataxia syndromes [157, 158].

The C9orf72 gene is transcribed in three major transcripts and located on the short arm of chromosome 9. The location of the repeat is between noncoding exons 1a and 1b, which is the first intron following noncoding exon 1 of transcript variants 1 and 3 and the upstream regulatory region of transcript variant 2 (Fig. 5.3). In the general population, the median length of the hexanucleotide repeat sequence is two repeats (range 0 to ±20). The C9orf72 transcript variants 1 and 3 encode for a 481 amino acid protein, whereas transcript variant 2 encodes for a shorter 222 amino acid protein. These protein isoforms share the first 221 N-terminal amino acids but differ in their C-terminal region. C9orf72 is detected throughout the CNS, with the highest expression level observed within the cerebellum [21, 22, 154, 159, 160]. Currently, there are no specific antibodies for C9orf72, and therefore, the exact subcellular localization is not very well established. In neuronal cell lines transfected with C9orf72, the protein is detected both in the nucleus and cytoplasm and in the medium [161], and additional studies have suggested C9orf72 is also present in the membrane fraction of cells after subcellular fractionation experiments [162, 163]. Further experiments are needed to conclusively determine where the protein is localized.

C9orf72 is highly conserved in evolution, and bioinformatics approaches showed that the protein is a novel homologue of differentially expressed in normal and neoplastic (DENN) proteins [164, 165]. DENN proteins are guanine exchange factors (GEF) that activate Rab GTPases, and these findings suggest a role for C9orf72 in Rab GTPase-dependent membrane trafficking. Support comes from a study in which several Rab proteins involved in endocytosis and autophagy were found to co-localize or co-immunoprecipitate with C9orf72 [161]. Knockdown of C9orf72 in neuronal cell lines also led to defects in autophagic processing as well as endocytosis. Furthermore, human neurons differentiated from iPSC with the C9orf72 repeat expansion were more sensitive to autophagy-inhibiting drugs [166]. Functional studies are required in order to conclusively demonstrate that C9orf72 encodes a GEF and Rab proteins might be activated.

Mutations in two other genes, VCP and CHMP2B, lead to FTLD-TDP and account for a minority of familial cases [1]. Mutations in VCP were identified in 2004 by linkage analysis studies in families with inclusion body myopathy associated with Paget disease of bone and frontotemporal dementia (IBMPFD) [187]. VCP mutations are also a cause of familial ALS [188]. The protein is ubiquitously expressed and a highly conserved member of the AAA(+)-ATPase superfamily. The protein is associated with a wide variety of cellular activities, including cell cycle control and membrane fusion. It is also a multi-ubiquitin chain-targeting factor and is required in the degradation of many ubiquitin-proteasome pathway substrates [189]. IBMPFD mutations cluster in the N-terminal CDC48 domain, mediating substrate recognition and cofactor binding. Studies in patient tissue, transgenic mice, and cell models suggest that VCP is a key regulator at the intersection of autophagy and the ubiquitin-proteasome system [190, 191]. Mouse models expressing mutant VCP recapitulate aspects of the disease, including age-dependent degeneration in muscle, brain, and bone, and progressive widespread TDP-43 pathology [192, 193].

Mutations in CHMP2B were found in 2005 via linkage analysis in a large Danish FTD family [194], and mutations are also a cause of familial ALS [195]. The ubiquitinated neuronal cytoplasmic inclusions do not stain for tau, TDP-43, or FUS, and the classification is FTLD-UPS. CHMP2B is a component of the endosomal sorting complex required for transport III (ESCRT-III) and expressed in neurons of all major brain regions. The ESCRT-III complex is required for function of the multivesicular body (MVB), an endosomal structure that fuses with the lysosome to degrade endocytosed proteins. Mutations affect the C-terminal region of the protein due to aberrant splicing, and mutant CHMP2B disrupts the fusion of endosomes with lysosomes in cell culture models [196]. Mice expressing C-terminally truncated and mutant CHMP2B develop axonal swellings and have reduced survival. The mice also develop ubiquitinated protein inclusions negative for TDP-43 and FUS, mimicking the inclusions found in patients with CHMP2B mutations [197].

Mutations in TARDBP and FUS are found, although these are very rare in FTD and more common in ALS [24–27]. FTD associated with mutations in GRN, VCP, or C9orf72 is consistently characterized by the presence of TDP-43 pathology, and a considerable number of tau-/TDP-negative FTLD cases have inclusion of FUS, suggesting that dysregulation of both proteins might be important in the pathogenesis of FTD [10, 95].

TDP-43 is a 414 amino acid protein with two RNA recognition motifs and a carboxy-terminal glycine-rich domain, encoded by the TARDBP gene on chromosome 1. TDP-43 is a member of the hnRNP family of proteins and regulates RNA processing in a variety of ways. It is involved in mRNA transport, stability and turnover, splicing, and translation. The protein is highly conserved, is predominantly nuclear, and can shuttle between the nucleus and the cytoplasm [198–200]. Essentially, all ALS- and FTD-associated mutations are missense changes in the C-terminal glycine-rich region, involved in protein-protein interactions. In FTLD-TDP brains, TDP-43 is redistributed to the cytoplasm, hyperphosphorylated, ubiquitinated, and cleaved into the C-terminal fragments (CTFs) [95]. Experimental studies do not conclusively demonstrate whether TDP-43-mediated neurodegeneration results from a gain or loss of function of the protein. There is evidence for both: gains of functions mediated by aggregation and abnormal cytoplasmic function or loss of functions mediated by nuclear depletion, CTFs, and RNA dysregulation [201, 202].

FUS was first reported in 2009 to be the cause of ~3 % familial ALS cases and subsequently found to be the marker for the pathology of many remaining tau-/TDP-negative FTLD cases [10, 203, 204]. The FUS gene encodes a 526 amino acid protein and forms a gene family together with EWSR1 and TAF15 (FET). Increased insolubility of all FET proteins has been found in FTLD-FUS, but not in ALS-FUS [205]. It is highly conserved, ubiquitously expressed, and mainly localized to the nucleus. The protein has a high degree of functional homology with TDP-43 and is involved in multiple steps of RNA processing [206]. Most mutations are missense mutations affecting the C-terminus, disrupting the binding of FUS to Transportin. This nuclear import receptor shuttles proteins from the cytoplasm to the nucleus leading to an accumulation of mutant FUS in the cytoplasm [207]. Also for FUS, it is still unclear whether the cytosolic deposition causes loss of essential nuclear functions, a gain of toxic functions in the cytosol, or both.

Compared to Mendelian FTD genes, little is known about susceptibility genes contributing to the risk of developing the disease. In 2010, Van Deerlin and colleagues identified TMEM106B in a genome-wide association study as a risk factor for FTLD-TDP [28]. Three SNPs encompassing the TMEM106B gene were found significant in a series of 515 FTLD-TDP patients, and variants specifically increased the risk in GRN mutation carriers. The initial GWAS finding has been successfully replicated in several FTD cohorts, strongly supporting a key role for TMEM106B in FTLD-TDP pathogenesis [208–210]. TMEMB106B is a 274 amino acid transmembrane type 2 protein with a highly glycosylated luminal domain [211]. It is cytoplasmically expressed in neurons, glia, and endothelial cells/pericytes of human brain samples from normal individuals, while in neurons of FTLD-TDP cases, TMEM106B appears to extend beyond the cell body into neuronal processes. In multiple immortalized cell lines and primary cortical neurons, TMEM106B is localized to late endosomes or lysosomes, and increased levels affect endolysosomal and progranulin pathways [212–214]. A recent GWAS points to novel associations, and the immune system processes, lysosomal, and autophagy pathways are potentially involved in FTD. The new loci need to be replicated in future studies to determine their possible association with the disease [29].

In 2011, UBQLN2 was added to the list of ALS-FTD genes [215]. The gene encodes the ubiquitin-like protein ubiquilin 2, a member of the ubiquitin family, which regulates the degradation of ubiquitinated proteins. In ~20 % of UBQLN2 mutation carriers, progressive dementia similar to FTD was identified, but all patients eventually developed motor symptoms. P62, encoded for by the SQSTM1 gene, is another protein at the intersection of ALS and FTD. It was found following the observation of involvement of P62 in ALS and has been subsequently reported in ALS and in FTD [216–218]. Inclusion positive for p62 can be found in C9orf72 expansion mutation patients, both with and without TDP-43 pathology [219].

Another risk factor for FTLD-TDP comes from a variant located in the 3′UTR of GRN in a binding site for the microRNA mir-659 [220]. In a series of pathologically confirmed FTLD-TDP patients without GRN mutations, Rademakers and colleagues observed that homozygous T-allele carriers had a 3.2-fold increase risk to develop FTLD-TDP as compared to homozygote C-allele carriers. They hypothesized that this variant reduces GRN protein expression through mir-659-dependent translational inhibition. It is still unclear whether GRN levels play a direct role in general FTLD population as two other studies failed to replicate the original findings [221, 222]. However, the risk allele is overrepresented in cases with hippocampal sclerosis [223, 224] and was found associated with a lower GRN serum level in two independent studies [225, 226], suggesting that decreased GRN expression may be a risk factor for FTLD-TDP and other dementias.

In recent years, remarkable progress has been made regarding the understanding of the genetic causes and neuropathological features of FTD. The recent discovery of the C9orf72 repeat expansions as the most common cause of the FTD/ALS complex of diseases showed that there is still a lot to be discovered. We know, for example, very little about susceptibility genes contributing to the risk of developing FTD and not all proteinopathies are fully characterized. The biological significance of the identified FTD genes has been studied extensively in cellular and animal models, as described in this chapter. However, diagnosis of FTD remains challenging, and no cure is available yet. Interestingly, many of the FTD genes seem to be connected at the molecular level. TMEM106B has originally been found as a risk factor for GRN mutation carriers, and several studies now describe TMEM106B as a genetic modifier for FTD with the C9orf72 repeat expansions [227, 228]. Patients with mutations in C9orf72 in combination with mutations in other genes involved in FTD/ALS have also been reported [229, 230]. Furthermore, FTD cases associated with GRN and C9orf72 mutations both present with TDP-43 pathology, and several studies showed that TDP-43 binds the 3′-UTR of GRN and regulates its expression [199, 231]. Understanding the full biology and connecting all identified FTD genes and risk factors with clinical phenotypes are of high importance for developing therapeutic approaches and offering reliable genetic advice to patients.