Tau is a candidate gene for chromosome 17 frontotemporal dementia

EXPEDITED PUBLICATION

Tau Is a Candidate Gene for Chromosome 17 Frontotemporal Dementia Parvoneh Poorkaj, PhD,*t Thomas D. Bird, MD,*$ Ellen Wijsman, PhD,$S" Ellen Nemens, MS,* Ralph M. Garruto, PhD,# Leojean Anderson, BS,* Athena Andreadis, PhD,** Wigbert C. Wiederholt, M D , t t Murray Raskind, MD,$&$ and Gerard D. Schellenberg, PhD*t$SS

Frontotempord dementia with parkinsonism, chromosome 17 type (FTDP-17), a recently defined disease entity, is clinically characterized by personality changes sometimes associated with psychosis, hyperorality, and diminished speech output, disturbed executive function and nonfluent aphasia, bradykinesia, and rigidity. Neuropathological changes include frontotemporal atrophy often associated with atrophy of the basal ganglia, substantia nigra, and amygdala. Neurofibrillary tangles (NFTs) are seen in some but not all families. Inheritance is autosomal dominant and the gene has been regionally localized to 17q21-22 in a 2- to 4-centimorgan (cM) region flanked by markers D17S800 and D17S791. The gene for tau, the primary component of NFTs, is located in the same region of chromosome 17. Tau was evaluated as a candidate gene. Physical mapping studies place tau within 2 megabases or less of D17S791, but it is probably outside the D17S800-Dl7S791 FTDP-17 interval. DNA sequence analysis of tau coding regions in affected subjects from two FTDP-17 families revealed nine DNA sequence variants, eight of which were also identified in controls and are thus polymorphisms. A ninth variant (vd279Me1was found in one FTDP-17 family but not in the second FTDP-17 family. Three lines of evidence indicate that the vd279Metchange is an FTDP-17 causative mutation. First, the mutation site is highly conserved, and a normal valine is found at this position in all three tau interrepeat sequences and in other microtubule associated protein tau homologues. Second, the mutation co-segregates with the disease in family A. Third, the mutation is not found in normal controls. Poorkaj P, Bird TD, Wijsman E, Nemens E, Garruto RM, Anderson L, Andreadis A, Wiederholt WC, Raskind M, Schellenberg GD. Tau is a candidate gene for chromosome 17 frontotemporal dementia. Ann Neurol 1998;43:8 15-825

Frontotemporal dementia (FTD) is an important cause FTD typically of dementia in middle and late occurs as an apparently sporadic disease without a strong family history of dementia. However, in some rare kindreds, FTD, which is often accompanied by parkinsonism, is inherited as an autosomal dominant trait.'-'' Although disease in these kindreds is clinically and neuropathologically heterogeneous, common behavioral, cognitive, and motor features are present." The behavioral findings often include a personality change sometimes associated with psychosis, hyperorality, and diminished speech output. The cognitive changes are usually disturbed executive function and nonfluent aphasia with memory preserved until relatively late in the illness. The motor changes are bradykinesia and rigidity without tremor, occasionally combined with muscle wasting and fasciculations. The

major neuropathological changes are frontotemporal atrophy often associated with atrophy of basal ganglia, substantia nigra, and amygdala with relative sparing of the hippocampus. Microscopically, there is neuronal loss and gliosis, Lewy bodies, Pick bodies, or ballooned neurons. Two of these rare families have neurofibrillary tangles ( N F T S ) ~ , ' ~ , 'composed * of microtubuleassociated protein (MAP) tau. Although the interfamily variability of the clinical and neuropathological findings makes it difficult to determine whether there is a single disease in these families, genetic studies demonstrate that for many of these kmdreds, the responsible gene is located at 17q21-22. This genetic linkage was first demonstrated in a family with social disinhibition, frontal lobe dementia, parkinsonism, and amyotrophy.' Subsequent linkage studies have shown that a gene in the same region is respon-

From the *Geriatric Research Education Clinical Center and f f D e partment of Psychiatry, Veterans Affairs Puget Sound Health Care System, Seattle Division; Divisions of t Gerontology and Geriatric Medicine, and SMedical Genetics, and Departments of $Neurology, SMedicine, "Biostatistics, SSPsychiatry, and SSPharmacology, University of Washington, Seattle, WA; #Department of Anthropology, Binghamton University, SUNY, Binghamton, NY; **E. K. Shriver Center for Mental Retardation, Walthem, MA; and TtDepartment of Neurosciences, University of California at San Diego, San Diego, CA.

Received Oct 2, 1997, and in revised form Jan 30, 1998. Accepted for publication Feb 1, 1998. Address correspondence to Dr Schellenberg, GFECC 182-B, Veterans Affairs Puget Sound Health Care System, 1660 S. Columbian Way, Seattle, WA 98108.

Copyright 0 1998 by the American Neurological Association 815

sible for F T D in seven other families (LOD scores, >3.0) and is probably the cause of disease in five more kindreds (LOD scores > 1 and < 3).'? The fact that the disease locus in most of these families is at the same relatively small region of chromosome 17 suggests that a single disease entity is responsible for F T D in these autosomal dominant families, and the phenotypic variability observed may be the result of allelic mutations at a single gene. The proposed name for this disorder is frontotemporal dementia with parkinsonism, chromosome 17 type (FTDP-17). An alternate but less likely hypothesis is that there are multiple genes at 17q21-22 in which mutations can cause FTD. It is presently unclear what the relationship is between FTDP-17 and the more common FTD, which frequently occur either without a family history or with affected relatives but no clear mode of inheritance. Once the FTDP-17 gene(s) is cloned, it will be possible to address questions of causative heterogeneity. T h e location of the FTDP-17 gene as determined by genetic mapping studies was recently reviewed, and the FTDP- 17 locus was placed between flanking markers D 17S800 and D 17S77 1. 1 3 This localization is based in part on observed obligate recombinants (recombinant events between 2 affected subjects) at D17S800 in a Dutch kindred' in which linkage analysis evidence clearly identifies FTDP-17 as the disease locus (LOD score, >3.0 for 17q21-22 markers using an affectedonly analysis). Marker D17S791 is defined as a flanking marker based on analysis of two families, one in which disease is conclusively linked to 17q21-22,9"6 and a second family in which suggestive but not conclusive evidence for linkage has been obtained.' The gene for MAP tau is located at 17q21-22 and is potentially in the D17S800-Dl7S771 interval (see Fig 1). Tau aggregates to form paired helical filaments (PHFs) and straight filaments found in Alzheimer's disease (AD) and in several other neurodegenerative diseases. At least two FTDP-17 families have N F T S . ~ , ' ~ , ' * I n one kindred (Family A; see Fig 2), the NFTs observed are structurally identical to AD NFTs in terms of the cypes of filaments formed, the tau subunits present, and the hyperphosphorylation sites detected. As in AD, tau-related pathology in this family is found only in neurons. I n another FTDP-17 kindred described by Spillantini and colleague^,^ NFTs (and glial fibrillary tangles), neuropil threads, and abnormally phosphorylated tau-containing inclusions are observed in both neurons and glial cells.9 However, the physical structure of the tau-related filaments and the tau subunits present are different from what is observed in A D and in Family A (see Fig 2). Still other FTDP-17 kindreds d o not appear to have NFTs but do have tau-positive rnclusions."x's Tau aggregates and NFTs are also a prominent neuropathological feature of Guam amyotrophic lateral sclerosis/parkinsonism

'

'

'*,"

816

Annals of Neurology Vol 43

No 6

June 1998

dementia complex (ALS/PDC) and progressive supranuclear palsy (PSP). T a u and rhe 17q21-22 region may also be involved in PSP; Conrad and associatesI9 recently reported an allelic association between PSP and a polymorphism within the tau gene. Thus tau is a candidate gene for both FTDP-17 and PSP. Here, we analyze tau as a candidate gene for FTDP17, by using physical mapping methods and by sequencing the gene in affected subjects from two potential FTDP-17 kindreds (Families A and B; see Fig 2). Linkage analysis indicates that the FTDP- 17 locus is definitely responsible for disease in the A kindred (maximum L O D score Z,,,, = 5.00 for D17S734, at a recombination fraction 8 = 0.001) and is probably responsible for dementia in the B kindred (Z,,, = 1.11 for D17S577, 8 = O.OO1).'o

Materials and Methods Genomic Clones Two tau-containing yeast artificial chromosome (YAC) clones (780F1 and 926F9) were identified by polymerase chain reaction (PCR) screening of the CEPH mega-YAC library*' arrayed essentially as described by Amemiya and coworkers.21 Primer pair 4F/4R (Table 1) was used for the screen. Other YACs adjacent to tau were identified from data in the Genethon (http://www.genethon.fr/) and Whitehead Institute (http://www.wi.mit.edu/) databases. The sizes of the YACs were determined by pulsed-field gel electrophoresis (PFGE). For each YAC, four individual clones were used to prepare agarose plugs for size determination; DNA in solution was also prepared from each growth.22,23After PFGE, YACs were detected by Southern blot analysis, using radiolabeled pBR322 DNA as a probe. If, for a given clone, the sizes of YACs from different growths varied, DNA from the growth producing the largest YAC was used for all further work. P1 clones for tau were identified by hybridization screening of high-density arrays of a human P1 library (Genome Systems, St Louis, MO). Filters were hybridized with probes generated from genomic DNA by PCR amplification, using primer pairs lBF/lBR, 4F/4R, and 14F114R (see Table 1). Hybridization conditions have been described p r e v i ~ u s l y . ~ ~ Two P1 clones were identified (320C5 and 226114) as containing exons from the tau gene, although neither clone contained the entire coding sequence. For mutation screening, an additional 250 nucleotides of intronic sequence between exons 6 and 7 were determined by direct sequencing of the PI clone 320C5. PI DNA was isolated from isopropylthiogalactoside-induced bacteria by using alkaline lysis followed by polyechyleneglycol precipitationz5 and spot dialysis, using VSWP 02500 Millipore filters. Ten micrograms of purified PI DNA was sequenced (primer 6BF; see Table I ) , using an ABI 373 DNA Sequencer (ABI, Foster City, CA) and TaqFS DNA polymerase with labeled fluorescent dye terminators. The sequencing reactions were performed according to manufacturer's instructions with 35 sequencing cycles (Perkin-Elmer, Foster City, CA).

Table 1. PCR Primer Sets Usedfor Tau Exon AmpliJication" Primer

Sequence

Mg*+/pH

Annealing Temperature ("C)

Product (bp)

Exon

1

1BF 1BR

CTCAGAACTTATCCTCTCC CAGTGATCTGGGCCTGC

1.5 mM/pH 8.5

60

22 1

150

2

2F 2R

CCCTTTGTGGGTTTGTTGC GTGAGCACATCTCTCAGCC

1.5 mMlpH 8.5

56

35 1

87

3

3BF 3BR

CACTGCAGCGTTTACACAGG CTGTCACAGGTCAGCTGGG

1.5 mMlpH 8.5

61

42 1

87

4

4F 4R

GAACTCCTCAGCAATGAC ACCAACTCCCTAAAATCTC

1.5 mM/pH 8.5

51

373

66

4A

4AF

GAAGGGTCCGGCCTTTCC GGAAGCTCAGTGGCAGTGC

1.5 mM/pH 10.0

61

883

753

4AR 5

5BF 5BR

TGGCTTTCTGTGAACAGTG CATAAAGCACAGCTTCTC

2.5 mM/pH 10.0

54

180

56

6

6BF 6INR

AGTTTGTTTCCCTCCTCC CACATTTGCAAACCACTGC

1.5 mM/pH 9.0

53

314

198

7

7CF 7BR

58

287

127

8

8BF 8BR

54

150

54

9

9BF 9BR

3.5 mM/pH 9.0 GTCACCCCAGTCTTAGCC AGCTTCAGCTTCCTCTAAGATTCAAG 2.5 mM/pH 10.0 GAAGGACTCATTAAG GCC CTGCTCCCAGCGAGTAG 2 mM/pH 10.0 CGTGCGCTTCCAACCTGG CACGCTCAACCGCCCACC

59

508

266

10

10F 1OR

CGAGCAAGCAGCGGGTCC GTACGACTCACACCACTTCC

1.5 mM/pH 8.5

58

200

93

11

11BF 11BR

CTCTCCTCCTCTCTCATC CACCAGGACTCCTCCACC

1.5 mMlpH 9.0

54

166

82

12

12FEXT 12RINT

CCACAGAACCACAGAAGATGATGGC CCAACCCACCCTACCC

1.5 mM/pH 10.0

55

190

113

13

13F 13R

ACTTCATCTCACCCTCCCTC CCTCTCCTTCTCCCTCTTCTAC

2 mM/pH 10.0

58

597

208

14

14F 14R

GCACTTCGATGATGACCTCC ACACACTCCAGAGATGCCAG

3.5 mM/pH 8.5

60

284

310

Exon

Additional intronic sequences have been submitted to GenBank with the following accession numbers: AF027491, AF027492, AF027493, AF027494, AF027495, and AF027496. "Primer pairs generated from sequences L35768, L35769, M93652, X61371, X61372, X61375, X61373, and X14474.

PCR

=

polymerase chain reaction.

Subject DNA DNA samples from FTDP-17 families were prepared from all available affected, at-risk subjects and spouses as previously described.'0325Control DNA is from 100 unaffected Caucasians. Guamainian Chamorro samples included 32 subjects with ALS (22 with autopsy confirmation), 51 subjects with PDC (26 with autopsy confirmation), and 94 Chamorro normal controls. AD samples were 153 cases (86 with autopsy confirmation) from 46 families with 2 or more affected subjects per family.

Mutation and Polymorphism Detection by DNA Sequence Analysis The tau gene was sequenced in 3 FTDP-17-affected individuals, 2 from Family A and 1 from Family B, and in 1 unaffected subject. Primer pairs for each exon from tau (see

Table 1) were used to amplie 200 ng of patient genomic DNA in 1OO-pl reactions (35 cycles) containing 1X PCR buffer, 200 ng of each primer, 2.5 U Taq DNA polymerase (Promega Corporation, Madison, WI), and 400 PM dNTP (Perkin-Elmer). PCR products were subjected to electrophoresis, using 2.5% Nusieve/O. 1X TAE gels, and the appropriate fragments purified by using a Gene-Clean kit (Bio 101). The purified fragments were either sequenced manualIy with a Sequenase PCR product sequencing kit (US Biochemical, Cleveland, OH) or sequenced automatically, using TaqFS DNA polymerase, fluorescent dye terminators, and an ABI 373 DNA Sequencer (ABI). All exons were sequenced completely in both the forward and the reverse directions. Additional sequencing primers are listed in Table 2. All sequence variants were initially detected by DNA sequence analysis. For genotype analysis of larger samples, restriction

Expedited Publication: Poorkaj et al: Tau and Frontotemporal Dementia

817

Results Chromosomal Localization of Tau A previous study by Neve and collaborators,26 using in situ hybridization of a tau cDNA clone to metaphasebanded chromosomes, placed the tau gene at 17q2122, which is in the vicinity of the FTDP-17 region. Other evidence that tau is near the FTDP-17 locus is that an expressed sequence-tagged site (STS) STSG2248, which is 9 1.3% homologous to a portion of tau exon 13, was mapped to the interval between D 17S930 and D 17S791 by Schuler and associatesz7 (Fig 1). To more precisely localize tau with respect to the D 17S800 -D 17S79 1 interval, the CEPH megaYAC library2' and a P1 library were screened for clones containing tau. Two YAC clones (780F1 and 926F9) and two P1 clones containing tau were identified. The YAC clones had not previously been assigned to a contig. However, Alu-PCR fingerprint data from Genethon (http://www.genethon.fr/)20 indicates that 780F 1 is adjacent to YACs 755B8, 845E2, 910E3, and 957F6, which are part of YAC contig WC17.6, constructed by the Whitehead Institute (http://www.wi. mit.edu/). Further, YAC 926F9 neighbors YAC 957F6. Two of the YACs from the WC17.6 contig were positive for the tau gene (910E3 and 771All; Table 3). Because contig WC17.6 includes both D17S800 and D 17S79 1 as well as flanking markers, additional experiments were performed with this contig, to attempt to determine whether tau is in the FTDP-17 interval.

Table 2. Additional Primers Used to Complete Exon Sequencing and Poolymorphism Screening Exon

Primer

Sequence

Exon 3

3FINT 3RINT

CTGG CATATGGCTGATCC GCACCCAGCAGGGCCTTG

Exon 4A

4ABR 4ABF 4ACF 4ACR 4ADF 4ADR

GCCTCTTTTGTGGTCCTCTCC GCCTGAGGGCCCCAGAGAGG GGTGGATGAAGACCGCGACG GGCCCGCCCTACACTGGGCC GGGCCCAGTGTAGGGCGGCC GGAGGACTCATCGACGTC G

Exon 6

6CF 6F 6R

TGC T G T GTG CCC AGA G C TTTCAACCATTACCTGCC CTTACCTTGAGTTTCATCTCC

Exon 9

9CFIN 9CR 9DF 13BR

TTCTGACCCCACCCACTCG CCACACGTCCACGCGCAGC GAGCCGCCTGCAGACAGCC

Exon 13

CTCTCCTCTCCACAATTATTG

digest assays were developed for most of the variants (see below). For these assays, PCR reactions were performed in 25-pl reaction volumes and restriction digests were performed directly in the PCR buffer with the addition of lox buffer (2.9 PI), bovine serum albumin (as per manufacturer's instructions), and 10 U of enzyme. All restriction digests were analyzed on 3 % Nusieve gels, using ethidium bromide staining.

Genethon Genetic Linkage Map

--

D17S800

0.8

0.5 0.1 -I 0.2

\

D17S934 Dl75932 D17S920 D17S930

0.5

RH

Map

2.4 6.8

/

// 3.8 0.1 2.4 1.9 2.6

818

Dl75934

D17S930

NIB-1996

D17S920

W 1-9656

, Wl-9656

WG 17.6

c:,:/

D17S943 \

Annals of Neurology

Vol 43

D17S791

D17S931 NIB-1996

D17S920

D17S791

D17S1834

Wl-7460

\

18-1321 D17S806

-

WI-6012

1D17S797

Wl-7460

Wl-6808

5.5

ci'7 o,

+ centromere

0175579

2.5

I-

D17S800

D17S932

6.8

0.1 : :

-

6.2

D17S791

D17S931 D17S806

D17S800

Wl-6523

-

-

__

Wl-4251

7

D17S797

No 6 June 1998

D17S931

D17S797

I3175806

ITau 17qter

+

Fig I . Genetic, radiation hybrid and yeast artificial chromosome (YAC) contig maps ofchromosome 17q21-22. The genetic map is the sex-average Genethon genetic map,42 with distances given in centimorgans (cM), and is oriented with the per end at the bottom. The Whitehead Institute radiation hybrid (RH) map has distances in centiRays ( C R ~ ~ ~Framework O). markers are in boldfaced, slightb larger type and are ordered with o& greater than 300:l. All other markers are located in intervals with odds less than 10:l. A subset ofthe markers placed in the Whitehead Institute YAC contig WC17.6 is shown on the right, with no indication of distance given.

Table 3. Regional Markers Placed on YACs in the FTDP Critical Region YAC

Size (kb)

741D2 957F6 910E3 755B8 930D5 767G11 655H11 930E1 813G10 793G3 926F9 771All 780F1 P1 clones 320C5 226114

610 (700) 400 (none) 260 (1,310) 560 (none) 464 (none) 1,160 (1,380) 1,159 (none) 265 (1,650) 1,209 (none) 880 (1,680) 1,900 (1,290) 1,550 (1,490) 330 (390) 72 86

WIG012 W n t 3 W7460 D17S920 NIB1996 Wl9656 D17S791 D17S1834 D17S931 Tau4 Tau8 Tau12 D17S943 D17S797 -

-

+ -

-

+ + + + -

-

-

-

D+ D+ F+

+ +

s-

-

-

-

-

-

-

-

-

-

-

+

+

+ + +

D+

+ S+ +

D+ D+ F+

D+ D+ F+

+

+ + + + D+ + + -

-

-

+ + + + F+ + + + -

-

-

-

-

-

-

-

-

-

-

+ D+ +-

-

-

+ +

-

-

D+

F

+

C+ F+

c

C

-

-

+

+ + + + + -

+

-

-

-

-

+

+ -

+

-

-

+

c

-

-

+ +

c

-

-

+

C

+ +

-

-

-

+

_

-

+

+

-

+

+

+

+

+

+

-

+

+

-

-

-

-

-

-

-

-

+ +

+

+

+

-

-

-

-

+

C+

+

+

-

C

+

-

-

-

+

+

-

+

-

-

-

Sizes are from pulsed-field gel electrophoresis experiments and from the CEPH database (in parentheses). Sequence-tagged site (STS) contenrs of yeast artificial chromosome (YAC) and P1 clones, as determined in this study: + = positive; - = negative. Datd from the Whitehead Institute and CEPH database are coded as follows: D = definite; F = ambiguous STS result resolved, using fingerprint data; C = verified hit reported by non-Whitehead Institute laboratory, primarily CEPH; c = unique, unverified hit reported hy non-Whitehead Insritute laboratory, primarily CEPH; S = ambiguous YAC/STS hit resolved, using STS content data. All above YACs wrre negative for markers "I6808 and IB1321 (see Fig 1).

FTDP

=

frontotemporal dementia with parkinsonism.

Additional YACs from contig WC17.6, which are neighboring the tau-positive YACs, were tested for STS markers, some of which have been used to construct genetic and radiation hybrid maps (see Fig 1). The STS content of each YAC or P1 clone was determined by PCR amplification of each marker. The marker order in Table 3 was chosen to minimize the number of YACs that appeared to be missing markers and thus have internal deletions. The orientation of the contig with respect to chromosome 17 was determined by comparing the STS-YAC derived order to the Genethon genetic map and the RH framework map, which is ordered with greater than 300-to-1 odds of marker inversion (see Fig 1). Markers on the left in Table 3 are centromeric and markers on the right are closer to the 17q telomere. The market order, derived from our data as well as the Whitehead data, is ambiguous. For the order given in Table 3, four YACs (767G11, 930D5, 655H11, and 793G3) are missing internal markers and no order is completely consistent with the STS content data. Presumably, some of the YACs have undetectable internal deletions. For example, YAC 910E3, which in the original CEPH library was characterized as being 1,310 kb, was only 260 kb in this study. Tau was placed adjacent to D17S943, because both markers are on YACs 780F1, 926F9, and 793G3, which do not have more centromeric markers. Tau was placed telomeric to D17S79 1, because three of the 1I YACs positive for D17S791 did not contain tau but did have more centromeric markers. Also, two YACs that did contain tau did not have D 17S791. However, because other marker orders are possible, tau could be centromeric to D17S791 and thus be in the FTDP-17 region. It is not possible to determine physical distances between markers from the YAC sizes, because all

YACs used here, except for 741D2, are chimeric (Genethon and Whitehead Institute databases). P1 clones 320C5 and 226114 contained tau exons 1 to 10 and 12 to 14, respectively, but were missing exon 11. Neither P I clone contained any of the other STSs used in this study. Thus, the P1 clones did not contribute to the marker order information.

D N A Sequence Analysis of Tau in FTDP-17 Subjects Because tau is near and possibly within the FTDP-17 critical region, the gene was sequenced in DNA from affected subjects from Families A and B (Fig 2). The genomic structure of the human tau gene, based on the sequence of human tau cDNA clones isolated from brain cDNA libraries, is comprised of at least 12 coding exons (exons 1-5, 7, and 9-14). In addition, cDNA clones from a bovine brain,28 a murine neuroblastoma cell line,23 and rat PC12 cells3' contain three additional possible exons (exons 4A, 6, and 8), which have not been observed in human cDNA clones. PCR primers were designed from intronic sequences flanking all 15 exons and were used to amplify fragments from genomic DNA from 3 affected subjects (2 from Family A and 1 from Family B) and from 1 control subject, 2 late-onset AD subjects, and a Guam ALS/ PDC subject. Both strands for each fragment were sequenced for all the coding regions and for between 7 and 70 bp of intronic sequence flanking each exon. The 50 bp of the 5'-untranslated region (UTR) sequence immediately upstream of the initiation codon, and the first 70 bp of the 3'-UTR immediately after the termination codon, were also sequenced. Nine sites were identified that were heterozygous in either Family A or Family B, or in both families (Table

Expedited Publication: Poorkaj et al: Tau and Frontotemporal Dementia

819

A Family 1

I 1

3

2

5

4

II 1

2

3

4

1

5

6

7

8

9

I

Ill

I

13

8

6ria 1

IV

62

60

-I- +I-

59

14 151

16

17

:fl I

7

6

23

24 251

26

27

57

47

+I- -I-

+I-

46

+I-

B Family I

1

I 1

2

3

4

43

77

14

72

5

6

I

7

a

II 88

Fig 2. (A) Pedigree of Family A, with fiontotemporal dementia with parkinsonism, chromosome 17 type (FTDP-17). Present age or age at death (in years) is under each symbol. +I- = All affected and at-risk individuals with the v a 1 2 7 pmutation; = = no children; -/- = individuals without the va'a"27p mutation; A = autopsy. Note that in Family A, all affected subjects sampled have the 2 7 p allele. (B) Pedigree of Farnib 3, with FTDP-17.

4). One of the variable sites was in the 5'-UTR, five were in either exon 4A (va1165Ala,A"p161A'p,and va'161va') or exon 6 (His47Tyr and ser53Pro),and are not presently known to be expressed in humans, one was in the 3'-UTR, and two (Asn197A""and v"'279Met) were in exons 9 and 12 and are known to be expressed in humans. To determine whether these variable sites were potential mutations or polymorphisms found in non-FTDP-17 subjects, assays were developed for each variable site and a panel of Caucasians were tested for the heterozygosity. Eight of the nine sites were also variable in controls and thus are polymorphisms. However, v"1279M"' was only found in affected subjects from Family A (Fig 3). Based on no examples of the v"1279Metallele in 96 control subjects, the 95% upper allele frequency in controls is 0.0155 (computed as previously described"). The 279Me' allele cosegregated

820

Annals of Neurology

Vol 43

No 6 June 1998

with FTDP-17 in this kindred; all affected subjects sampled from Family A had the 279Me' allele (see Fig 2). Affected subjects from the B kindred were not hetetozygous at amino acid 279 and the sequence variants identified in this family were also found in controls, indicating that each is a polymorphism. Also, the 27gMetallele was not observed in 153 familial AD subjects, and in 135 of their at-risk relatives, and in Chamorro subjects from Guam including 32 with ALS, 51 with PDC, 5 with ALYPDC, and 93 Chamorro controls.

Discussion The above study was undertaken to evaluate the role of tau in FTDP-17. Tau is also a candidate gene for PSP,19 and thus it is important to evaluate tau sequence variants and polymorphisms for genetic associ-

Fig 3. Screening of the va127p amino acid change in tau exon I 2 in Family A. Exon 12 was amplified from genomic DNA, digested with Bs tN4 and the resulting Ji-agments resolved by agarose gel electrophoresis, M = Size standards @om Boehringer Mannheim: lane I , set I 4 lane 14, marker set IX lane 15, marker set VIII); A = affected subjects; R = at-risk; N = spouses. A 126-bp fragment indicates the presence of the 2 7 p allele. A band at 63 bp (actually 2 X 63 bp j a p e n t s ) indicates the presence of the 279vqi allele,

ation studies of PSP. Obligate recombinants in FTDP- 17 kindreds define ' D17S800 as the centromeric flanking marker" and D17S791 as the probable telomeric flanking marker."13 Physical mapping studies described above show that tau is within 2 megabases of D17S791, and is possibly closer. Physical mapping using YACs suggests that tau may be outside this interval (see Table 3) and agrees with physical mapping studies of other researchers (see Fig 1). However, due to the problem of internal deletions in YACs, the precise location of tau awaits generation of small-insert (PI or bacterial artificial chromosome clone) contigs and the eventual complete sequence analysis of this region of chromosome 17. Because tau is physically close to, if not in, the FTDP-17 region, the gene was evaluated as a candidate gene by DNA sequence analysis. The expected pathogenic mutations for FTDP-17 should be rare and not present in controls, should cosegregate with the disease in the families, and should be present in a heterozygote state, because the inheritance of FTDP-17 is dominant. If the group of disorders that map genetically to 17q21-22 are a single disease entity, all families should have mutations in the same gene, although each family may have a different mutation. Nine DNA sequence variants were identified in DNA sequence analysis of affected subjects from the A and B kindreds (see Table 4). Eight of these variants are also found in controls and thus are polymorphisms rather than mutations. The ninth, Vd'279Mrr,is a potential disease-related variant, because it is not observed in any of the controls, and it cosegregates with affected

subjects in Family A (see Fig 2). This variant does not exist in Family B, and no other mutations in tau were found in this second kindred. The va'279Metsite is located in a region of tau that contains repeated sequences thought to be involved in microtubule binding. In the form of tau found in adults, four copies (Rl, R2,R3,and R4; R = "repeat"; Fig 4) of an 18-amino acid repeat sequence are found near the C terminus. These repeats are separated by divergent 14-amino acid spacers (IRI-2, IR2-3, and IR3-4; IR = "interrepeat"). In fetal tau, as a result of alternative splicing of exon 10, R2 is missing and IR3-4 separates R1 and R4.'2,'3 The va'279Mecsite is located at the second amino acid of IR3-4 and is thus present in both fetal and adult tau. Although the interrepeat sequences are not well conserved, the normal valine at position 279 is conserved in each interrepeat sequence. Also, in other members of the MAP family, including mouse MAP2, both human and mouse MAP4, and MAPU,34 a valine is present two amino acids after the last three glycines of the repeat sequences (see Fig 4). Thus, the va1279M"'site is highly conserved in this family of proteins. Functional analysis of this region of tau suggests that the repeats are individual microtubule-binding domains that interact with the microtubules through a flexible array of distributed weak site^.'^,'^ Although some analyses of the interrepeat binding domains of tau have shown that only the 1R1-2 contributes to the microtubule affinity, the possibility that IR3-4 may affect microtubule-binding efficiency cannot be excluded. More recent functional

Expedited Publication: Poorkaj et al: Tau and Frontotemporal Dementia

821

Table 4. Tau Gene Variants

Location

Family

5'-UTRd Exon 4A' Exon 4Af Exon 4Af Exon 6g Exon bh Exon 9' Exon 12' Exon 13k

B B A B BIA BIA B A B

Amino Acid Change"

Polymorphic Nucleotide

nt in

cDNAb

NG TIC CIT GIA CIT TIC TIC GIA TIC

nt in Exon' -

554 543 541 378

396 -

-

290

Number of Controls

Allele A Frequency

Allele B Frequency

88 95 92 92 74 87 22 96 12

0.79 0.78 0.85 0.77 0.82 0.8 1 0.79 1 .o 0.71

0.21 0.22 0.15 0.23 0.18 0.19 0.21 0 0.29

"For exons 4A and 6, amino acid numbering is from Andreadis and c o l l e a g ~ e s .For ~ ~ the remaining sites, amino acid numbering is from the translated human tau cDNA sequence (GenBank accession no. X14474). 'Tau cDNA nucleotide numbering is based on the human four-repeat isoform cDNA sequence32 (GenBank accession no. X14474). 'Nucleotide numbering is based on the total sequence given (GenBank accession nos. for exons 4A and 6 are M93652 and X61371, respectively), which includes both intron and exon sequences. dGenomic DNA was amplified by using primer pair 1BFllBR and the products were digested with ABI; allele A, the published sequencc (GenBank accession no. X14474), does not contain an AluI site and an undigested 221-bp fragment is produced, whereas allele B has an Ah1 site and yields 183and 38-bp fragments. 'The v"1165A'" site was assayed by introducing into the PCR amplification product a new restriction site for BsrDI at the polymorphic site. This was accomplished by amplifying genornic DNA with primers 4ABR and BsrDIF (5'CCTCAGAGCCCGACGGGCCCATTG; the nucleotide changed from the published sequence is indicated by boldfaced lettering in the primer). The 3' end of the latter primer ends immediately adjacent to the polymorphic base. For allele B, a BsrDI site results (GCAATGNN) and 19- and 186-bp fragments are generated by BsrDI digestion, whereas the allele A product is not digested by BsrDI and the uncut fragment is 205 bp. 'Both exon 4A polymorphisms were analyzed in all subjects by direct sequencing, using the primer pair 4ACF/4ABR. PCR products were subjected to electrophoresis by using a 2.5% Nuseive/O.lX TAE gel, the 448-bp fragment excised, purified by using a Gene-Clean kit (Bio IOl), and sequenced, using TaqFS DNA polymerase with fluorescently labrled dye terminators and an ABI 373 DNA sequencer. For each subject, the fragment was sequenced in both directions. "enomic DNA was amplified by using primer pair 6F/6R and the product digested with AfLK The A allele results in 6 5 , 117-, and 131-bp fragments, whereas the B allele yields 117- and 196-bp fragments. hGenomic DNA is PCR amplified by using primers 6F and 6R, and the product, when digested with Bsli,yields 136- and 177-bp fragments for allele A and 40-, 96-, and 177-hp fragments for allele B. 'Genomic DNA is PCR amplified with primer pair 9BF/9BR and the 509-bp product is digested with Tail.Allele A yields an uncut product, whereas allele B yields 162- and 347-bp fragments. jPrimer pair 12FEXT/12Rint (see Tables 1 and 2) was used to PCR amplify genomic DNA, to produce a 190-bp fragment. Digestion of this fragment with BstNI results in four fragments of 22, 42, 63, and 63 bp for allele A and 22-, 42-, and 126-bp fragments for allele B. kGenomic DNA is amplified with primer pair 13F113R and the resulting 597-bp product is digested with Tsp509I. Allele A yields 15-, 62-, 72-, 78-, 94-, and 276-bp fragmenta, whereas allele B yields 15-, 62-, 72-, 78-, and 370-bp products. n t = nucleotide; UTR = unrranslated region; PCR = polymerase chain reaction; NA = not applicable.

analysis of the repeat and interrepeat domains indicates that the sequences flanking the microtubule-binding domains are actually required for microtubule assembly. The flanking interrepeat regions are considered to be targeting domains responsible for positioning tau on the microtubule surface and allowing microtubule as~ e m b l y ? That ~ the second interrepeat amino acid is conserved in all the mammalian tau interrepeats suggests that a semi-nonconservative amino acid change in this position might alter some structural or functional property of tau-microtubule or tau-tau interaction. There are several possible interpretations of the significance of the 279"" allele in the A kmdred. First, the 279M" allele may simply be a benign private variant that is not responsible for FTDP-17 and does not have other pathological consequences. Evidence supporting this interpretation is that no potential tau mutations were observed in the B kindred. Although the

822

Annals of Neurology

Vol 43 No 6 June 1798

evidence that Family B is a FTDP-17 kindred is not as strong as that for Family A (as evidenced by positive but not conclusive LOD scores'"), the clinical features of this family overlap with FTDP-17, and thus the B kindred is probably a FTDP-17 kindred. Also, no tau candidate mutations have been found in affected subjects from other FTDP-17 kindreds (Clark L, Wilhelmsen K, personal communication), although these families have no documented N F T pathology. A second hypothesis is that there may be more than one neurodegenerative disease gene in this region and thus only some families will have mutations in tau. Challenges to this hypothesis are that most FTDP- 17 families have at least some overlapping clinical and neuropathological features, and that all map genetically to a relatively small region relative to the entire genome. However, kindred A is unique among FTDP-17 kindreds in that it is the only family reported, to date,

repeat 1 repeat 2 repeat 3 repeat 4

...

VRSKIGSTENLKHQPGGG

I I I I I IIIIIII . . . VQSKCGSKDNIKHVPGGG IIIIII I I I I I I I

KVQIINKKLDLSN . . . (IR1-2)

Ill I Ill SVQIVYKPVDLSK . . . I I

. . . VTSKCGSLGNIHHKPGGG QVEVKSEKLDFKD . . .

I I1 I l l I I I I I I I

. . . VQSKIGSLDNITHVPGGG

-3

M279

A

* KCGSKANIKHKPGGG DVKIESQKLNFK

IIII I I I I I I I I I I I I I I KCGSLGNIHHKPGGG QVEVKSEKLDFK OIII II I Ill1 I I IIIII KCGSLKNIRHRPGGG RVKIESVKLDFK IIII II I Ill1 IIIII II II KCGSKANIKHKPGGG DVKIESQKLNFK

MAP4 TAU

MAP2 MAPU

B that has neuron-exclusive NFTs that are physically indistinguishable from AD NFTs. Electron microscopy studies show that the tau filaments in Family A are primarily PHFs that are identical in diameter (8 -20 nm), periodicity of the twist (-80 nm), physical appearance, and immunoreactivity to those observed in AD.” As in AD, a minority of the filaments are straight. Biochemically, tau from Family A is indistinguishable from AD tau; in both diseases, the major tau bands are 60, 64, and 68 kd in size, and these bands react with the same set of antibodies against hyperphosphorylated sites. Also, for both diseases, dephosphorylation yields the same set of six bands representing the six possible isoforms generated by alternative splicing of exons 2, 3, and 10. The primary neuropathological difference between AD and Family A is the lack of P-amyloid protein pathology in Family A and a difference in the distribution of NFTs in the two diseases. The only other FTDP-17 family that has been reported to have NFTs is the “tauopathy” kindred described by Spillantini and co-workers.’ This kindred, in contrast to Family A, has NFT-like filaments in both neurons and glial cells and has tau in neuronal and glial granular deposits. Also, the PHFs observed in the tauopathy kindred, compared with AD and Family A filaments, are wider and have a different periodicity. Biochemically, only 64- and 68-kd tau bands are observed and these collapse on dephosphorylation into only two isoforms, both of which contain exon 10. Thus, based on neuropathological and biochemical evidence, Family A could represent a different disease entity relative to other FTDP-17 kindreds and the 279“4‘t allele is a pathogenic mutation. A third possibility is that the 279M“ allele predisposes FTDP- 17 gene carriers to develop AD-type NFTs. As discussed above, the A kindred is the only FTDP-17 kindred with AD-type tau pathology, and thus may be the only FTDP-17 kindred with a tau

Fig 4. Homology ofthe tau interrepeat (IR) domains within (IR2-3) the human tau protein (2) and protein sequence alignments of (IR3-4) the I& domains of dzferent microtubule-associated protein (MAP) species (B): human tau (amino acids 263-289), mouse tau (amino acids 221-247), moue MAP2 (amino acids 1,740-1,76@, human (amino acid 1,038-1,064) and mouse (amino acids I,Oll-l,037) MAP4, and bovine MAPU (amino acids 953-979). V and * = The amino acid changed with the “‘27F site.

variant affecting the pathology. In Family B, with four autopsies, no tau mutations and no tau-containing tangles are observed.” In support of this hypothesis of an interaction between tau variants and pathology are previous studies with PSP, another disease in which NFTs form: genetic variants in the tau gene may be associated with genetic susceptibility to PSP.19 Also, because the 279-amino acid site is highly conserved, a seminonconservative amino acid change, such as the methionine for valine substitution observed here, could potentially be functionally significant. A previous study by Conrad and associates’9 demonstrated a genetic association between PSP and a short-tandem repeat polymorphism (STRP) within tau. Because this STRP site itself is within an intron, it not likely to be the cause of susceptibility to PSP. A more likely interpretation is that a DNA sequence variant within tau or in a closely adjacent gene are the true PSP susceptibility locus. The polymorphisms described here are candidates for the PSP susceptibility site. The polymorphism in the 5’-UTR could potentially affect expression of tau. The four polymorphisms detected in coding sequences, which alter amino acids, are all found in exons (4A and 6) that may not be expressed in adult brain and thus may not be relevant to PSP. Two other polymorphisms in exons 4A and 9 are silent and do not change the protein sequence of tau. Although this type of variant probably does not affect gene or protein function, examples of silent mutations that alter translation efficiency3’ or cause exon skiphave been reported. Whereas the polymorphisms described here may not be the PSP susceptibility site, they will be useful for defining the region of PSP linkage disequilibrium around tau as distinct sites. These sites will also be useful for constructing haplotypes in PSP and control subjects to facilitate identification of the PSP susceptibility site. In conclusion, the findings do suggest that a muta-

Expedited Publication: Poorkaj et al: Tau and Frontotemporal Dementia

823

tion in the tau gene contributes to the pathogenesis of FTDP-17 in one family in which NFTs are found. However, additional work is needed to determine the relevance of tau variants in other FTDP-17 families and PSP. Affected subjects from additional FTDP-17 kindreds must be sequenced. Also, the polymorphisms described here must be tested in a PSP population.

Note Added in Proof After this manuscript was accepted, we screened a third newly ascertained FTDP-17 kindred (the D family) for mutations in tau. At nucleotide 728, a C to T change was found that results in a proline to leucine change. This mutation is at the end of tau repeat number 2 (. . . PGGG.. . t o . . . LGGG.. .). This Pro243Leu mutation co-segregates with frontotemporal dementia in this kindred and is not found in 91 normal controls. This additional mutation supports the hypothesis that mutations in tau cause disease in at least some FTDP-17 kindreds. This study was supported by NIA grant AG1176-03 (G.D.S.). Support for identification of Families A and B is from the NIA Alzheimer’s Disease Research Center (AGO5136; George Martin, director). Collection of Guam samples was supported in part by NIA grant PO1 AG 14382 (Wigbert Wiederholt, principal investigator). Informed consent was obtained from each subject of next-of-kin with approval of the University of Washington Human Subjects Review Committee.

0.

1.

12.

13.

14.

15.

16.

17.

18.

We thank Elaine Loomis, Kevin Farewell, and Darren Bisset for technical help with this manuscript and David Nochlin for autopsies.

20.

References 1. Gustafson L. Clinical picture of frontal lobe degeneration of non-Alzheimer type. Dementia 1993;4:143-148 2. Gearing M, Mirra SS, Hedreen JC, et al. The consortium to establish a registry for Alzheimer’s disease (CERAD). Part X. Neuropathology con firmation of the clinical diagnosis of Alzheimer’s disease. Neurology 1995;45:461-466 3. The I.und and Manchester Groups. Clinical and neuropathological criteria for frontotemporal dementia. J Neurol Neurosurg Psychiatry 1994;57:416-418 4. Brun A. Frontal lobe degeneration of non-Alzheimer type. I. Neuropathology. Arch Gerontol Geriatr 1987;6:193-208 5. Knopman DS, Mastri AR, Frey WH, et al. Dementia lacking distinctive histologic features: a common non-Alzheimer degenerative dementia. Neurology 1990;40:251-256 6. Neary D, Snowden JS, Northern B, Goulding P. Dementia of frontal lobe type. J Neurol Neurosurg Psychiatry 1988;51:353361 7. Wilhelmsen KC, Lynch T, Pavlou E, et al. Localization of disinhibition-dementia-parkinsonism-amyotrophycomplex to 17q21-22. Am J Hum Genet 1994;55:1159-1165 8. Wijker M, Wszolek ZK, Wolters ECH, et al. Localization of the gene for rapidly progressive autosomal dominant parkinsonism and dementia with pallido-ponto-nigral degeneration to chromosome 17q21. Hum Mol Genet 1996;5:151-154 9. Spillantini MG, Goedert M, Crowther RA, et al. Familial multiple system tauopathy with presenile dementia: a disease with

824

Annals of Neurology

19.

Vol 43

No 6 June 1998

21.

22.

23.

24.

25.

26.

27. 28.

abundant neuronal and glial tau filaments. Proc Natl Acad Sci USA 1997;94:4113-4118 Bird TD, Wijsman EM, Nochlin D, et al. Chromosome 17 and hereditary dementia: linkage studies in three non-Alzheimer families and kindreds with late-onset FAD. Neurology 1997;48: 949-954 Heutink P, Stevens M, Rizzu P, er al. Hereditary frontotemporal dementia is linked to chromosome 17q21-q22: a genetic and clinicopathological study of three Dutch families. Ann Neurol 1997;41:150-1 59 Yamaoka LH, Welsh-Bohmer KA, Hulette CM, et al. Linkage of frontotemporal dementia to chromosome 17: clinical and neuropathological characterization of phenotype. Am J Hum Genet 1996;59:1306-1 3 12 Foster NL, Wilhelmsen K, Sima AAF, et al. Frontotemporal dementia and parkinsonism linked to chromosome 17: a consensus conference. Ann Neurol 1997;41:706-715 Sumi SM, Bird TD, Nochlin D, Raskind MA. Familial presenile dementia with psychosis associated with cortical neurofibrillary tangles and neurodegeneration of the amygdala. Neurology 1992;42:120-127 Lynch T, Sano M, Marder KS, et al. Clinical characteristics of a family with chromosome 17-linked disinhibition-dementiaparkinsonism-amyotrophic complex. Neurology 1994;44: 18781884 Murrell JR, Koller D, Foroud T, et al. Familial multiple-system tauopathy with presenile dementia is localized to chromosome 17. Am J Hum Genet 1997;61:1131-1138 Spillantini MG, Crowther RA, Goedert M. Comparison of the neurofibrillary pathology in Alzheimer’s disease and familial presenile dementia with tangles. Acta Neuropathol 1996;92: 42-48 Sima AAF, Defendini R, Keohane C, et al. The neuropathology of chromosome 17-linked dementia. Ann Neurol 1996;39: 734-743 Conrad C, Andreadis A, Trojanowski J, et al. Genetic evidence for the involvement of tau in progressive supranuclear palsy. Ann Neurol 1997;47:277-281 Bellanne-Chantelot C, Lacroix B, Ougen P, et al. Mapping the whole human genome by fingerprinting yeast artificial chromosomes. Cell 1992;70:1059-1068 Amemiya CT, Alegria-Hartman MJ, Alslanidis C, er al. A twodimensional YAC pooling strategy for library screening via STS and Alu-PCR methods. Nucl Acids Res 1992;20:2559-2563 Sherman F, Fink GR, Hicks JB. Methods in yeast genetics: a laboratory manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press, 1986 Smith CL, Klco SR, Cantor CR. Pulse-field gel electrophoresis and the technology of large DNA molecules. In: Davies KE, ed. Genome analysis-a practical approach. Oxford: IRL Press, 1990:41-72 Levy-Lahad E, Poorkaj P, Wang K, et al. Genomic structure and expression of STM2, the chromosome 1 familial Alzheimer’s disease gene. Genomics 1996;34: 198-204 Sambrook J, Fritsch B, Maniatis T . Molecular cloning: a laboratory manual. 2nd ed. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press, 1989 Neve RL, Harris P, Kosik KS, et a]. Identification of cDNA clones for the human microtubule-associated protein tau and chromosomal localization of the genes for tau and microtubuleassociated protein 2. Mol Brain Res 1986;1:271-280 Schuler GD, Boguski MS, Stewart EA,et al. A gene map of the human genome. Science 1996;274:540-546 Himmler A, Drechsel D, Kirschner MW, Martin DW. Tau consists of a set of proteins with repeated C-terminal

29.

30.

31.

32.

33.

34.

35.

microtubule-binding domains and variable N-terminal domains. Mol Cell Biol 1989;9:1389-1396 Couchie D, Mavilia C, Georgieff IS, et al. Primary structure of high molecular weight tau present in the peripheral nervous system. Proc Natl Acad Sci USA 1992;89:4378-4381 Goedert M, Spillantini MG, Crowther RA. Cloning of a big tau microtubule-associated protein characteristic of the peripheral nervous system. Proc Natl Acad Sci USA 1992;87:19831987 Levy-Lahad E, Wasco W, Poorkaj P, et al. Candidate gene for the chromosome 1 familial Alzheimer’s disease locus. Science 1995;269:973-977 Goedert M, Spillantini MG, Rotier MC, et al. Cloning and sequencing of the cDNA encoding an isoform of microtubuleassociated protein tau containing four tandem repeats: differential expression of tau protein mRNA’s in human brain. EMBO J 1989;8:393-399 Kosik KS, Orecchio LD, Bakalis S, Neve RL. Developmentally regulated expression of specific tau sequences. Neuron 1989;2: 1389-1 397 Irminger-Finger I, Hurt E, Roebuck A, et al. MHP1, an essential gene in Succbaromyces cerevisiue required for microtubule function. J Cell Biol 1996;135:1323-1339 Goode BL, Feinstein SC. Identification of a novel microtubule

36.

37.

38.

39.

40.

41. 42.

binding and assembly domain in the developmentally regulated inter-repeat region of tau. J Cell Biol 1994;124:769-782 Butner KA, Kirschner MW. Tau protein binding to microtubules through a flexible array of distributed weak sites. J Cell Biol 1991;115:717-730 Preuss U, Mandelkow EM, Mandelkow E. The “jaws” model of tau-microtubule interaction examined in C H O cells. J Cell Sci 1997;1 10:789-800 Milland J, Christiansen D, Thorley BR, et al. Translation is enhanced after silent nucleotide substitutions in the A+T-rich sequences of the coding region of CD46 cDNA. Eur J Biochem 1996;238:221-230 Liu W, Qian C, Franke U. Silent mutation induces exon skipping of fibrillin-1 gene in Marfan syndrome. Nat Genet 1997; 16~327-328 Cooper TA, Mattox W. Gene regulation ’97. The regulation of splice-site selection, and its role in human disease. Am J Hum Genet 1977;61:259-266 Andreadis A, Brown WM, Kosik KS. Structure and novel exons of the human-tau gene. Biochemistry 1992;31:10626-10633 Dib C, Faure S, Fizames C, et al. A comprehensive genetic map of the human genome based on 5,264 microsatellites. Nature 1996;380: 152-154

Expedited Publication: Poorkaj et al: Tau and Frontotemporal Dementia

825