As of 2004, 34 pathogenic tau mutations have been described in 101 FTDP-17 families [25]. They include 21 missense mutations, three silent mutations, two in-frame single codon deletions, and nine intronic mutations (Table 1). In addition, 17 coding polymorphisms have been reported [25]. The mutations are clustered in exons 9-13 which encode the microtubule-binding domain and flanking regions [39,40]. Most coding polymorphisms are with in exons 4A, 6 and 8, that are not expressed in any of the major brain isoforms [39,41,42]. The most frequently observed mutations are a C.T substitution corresponding to P301L* in exon 10 (identified in 25 families) and IVS-10+16 (C.T) in intron 10 (identi-fied in 22 families). In contrast, 21 mutations were reported in a single family [25]. Almost all mutations are heterozygous and segregate in a dominant manner within families; however, homozygous N296del* and S352L* mutations have been observed in siblings of consanguineous marriages [25,43,44].

Mutations in tau proteins may affect alternative splicing of exon 10 and lead to changes in the proportion of 4R- and 3R-tau isoforms, or modify tau interactions with microtubules [21,25]. The former group includes intronic mutations (+3, +13, +14,+16) and some missense mutations. Intronic mutations destabilize a stem loop structure in the 5’ splice site of exon 10 that stabilizes this region of the premRNA [45-47]. Without this stem loop, access of U1snRNP to this site may be facilitated, increasing the formation of exon 10+ tau mRNAs, and thus, the 4R-tau isoform [21]. In FTDP-17 families that carry these mutations, abnormally phosphorylated 4R-tau isoforms aggregate into filaments and display an electrophoretic profile similar to the major tau doublet of 64 and 69 kDa seen in PSP and CBD [21]. Some missense mutations also modify the splicing of exon 10 (Fig. 4). The change in nucleotide that results in N279K* and S305N* mutations also creates an exon-splicing enhancer sequence. The silent mutation L284L* increases the formation of tau mRNAs containing exon 10, presumably by destroying an exon-splicing silencing element. Families with one of these three missense mutations show the same tau electrophoretic profile as those having intronic mutations [25,45]. The latter group of tau mutations consists of missense mutations that affect microtubule assembly and polymerization. When located in the tau regions outside exon 10 they affect all tau iso forms so that none can properly bind to microtubules. These proteins aggregate into PHF and straight filaments in neuronal cells. Conversely, when these mutations are located in exon 10, 4R-tau isoforms are affected which do not bind to microtubules and aggregate into twisted ribbon filaments. This type of filamentous inclusion has been described in both neuronal and glial cells [21,48,49].

Mutations of the tau gene and their involvement in FTDP-17 emphasize the fact that abnormal tau proteins may play a central role in the pathogenesis of neuro degenerative disorders, without involvement of the amyloid cascade [1,21]. The functional effects of the mutations suggest that a reduced ability of tau to interact with micro tubules may be the first event upstream of hyperphos phorylation and aggregation. These mutations may also lead to an increase in free cytoplasmic tau (especially 4R-tau isoforms which facilitates their aggregation into filaments [50,51].

Even before tau mutations were identified in FTDP-17 families, over representation of the most frequent allele (A0) of a dinucleotide repeat polymorphism, located in intron 9 of tau in PSP patients, had been reported [52]. This association has been confirmed in four independent populations [53-56]. Sequencing of the tau gene in FTDP-17 patients identified a series of polymorphisms within intronic sequences flanking tau exons [25,54]. Analysis of these polymorphisms revealed them to be in complete linkage disequilibrium with the intron 9 microsatellite, and led to identification of two major tau haplotypes (H1 and H2) extending over the complete coding tau region and the 5’ promoter region of exon –1 (minus one) [7,25,57-59]. Recombination events between H1 and H2 have not been observed. Because the A0 allele is inherited as part of H1, increased frequencies of H1 in PSP populations have been identified [52,53]. Moreover, association studies in two CBD populations demonstrated similarly increased H1 frequencies, supporting a common genetic susceptibility for PSP and CBD [60,61]. Genetic investigation of the saitohin gene (STH) located within tau intron 9, revealed a coding polymorphism changing a glutamine (Q) to an arginine (R) at codon 7. Since the Q allele of STH is part of H1, it is no surprise that the positive association of PSP with tau is replicated by genotyping the Q7R polymorphism [62,63].

In most populations, genotypic (H1H1) association is more significant than allelic (H1) association, suggesting the existence of a recessive mutation or a dosage-sensitive risk allele on H1 [25]. Both PSP and CBD are characterized by parkinsonism and neurodegeneration.Although they are pathologically distinct, similar tau deposits, predominantly composed of 4R-tau isoforms, are found in the brains of these patients. It remains unclear how the presence of H1 would result in 4R-tau pathology. Although H1 and H2 differ within tau coding regions, missense mutations have only been identified in exons 4A and 6, which are not expressed in the major human brain isoforms. Since H1 is present in approximately 80% of the Cauca sian population, extensive sequencing analysis of the H1 carriers is necessary to identify genetic variations that define disease-specific H1 subhaplotypes [25].

Tau association studies in other tauopathies (FTD, Pick’s disease and AD) were negative or inconsistent in independent studies [1,25]. Although H1H1 frequencies were found to be moderately increased in all FTD populations, they did not reach statistical significance [25]. The majority of tau association studies in AD populations were negative. Two studies, containing neuropathologically diagnosed Pick’s disease patients, gave slightly increased association with the H2H2 genotype [64,65].

Figure 4. Stem loop structure of intron 10, mutations in intron 10 and exon 10, destabilizing this structure, and some other missense mutations in exons 9, 12 and 13 [1].