16 September 2005. For years, geneticists studying tau have been stymied by the conspicuous absence of mutations in this hallmark protein of Alzheimer disease pathology. This has held back research on tau’s role in AD even as the function of this microtubule-associated protein in normal neurons has become quite clear. (By contrast, research on the amyloid hypothesis of AD has flourished since AD-causing mutations in three different genes converged on APP metabolism in the early 1990s.) Having just about given up on autosomal-dominant mutations in tau, AD researchers are looking for more subtle genetic shenanigans by this protein, and their efforts are beginning to look promising.
In the August 15 Human Molecular Genetics, a collaborative team of geneticists in the U.K., Taiwan, and the U.S. report data suggesting the presence of a pathogenic tau variant in two separate series of AD patients. Led by John Hardy at the National Institute on Aging in Bethesda, Maryland, the scientists compared the tau locus in 260 people with AD and 252 without it. They identified a region between exon 1 and intron 9 that appears to be linked with AD in these patients.
The present paper extends earlier work on genetic variability at the tau locus, which identified two evolutionary branches, or clades, that is, haplotypes H1 and H2 (see St. Moritz conference report). H1 is linked with the sporadic tauopathies progressive supranuclear palsy (PSP) and corticobasal degeneration (CBD). Curiously, H2 appears to be a Neanderthal legacy, but, any assumptions about these forebears’ limited brainpower notwithstanding, H2 is underrepresented in PSP and CBD (see Hardy et al., 2005).
This spring, Hardy and colleagues at University College, London, reported that the H1 haplotype has great variability and identified one form, H1c, as the one that associated with the two tauopathies (Pittman et al., 2005). H1c now turned out to also be overrepresented in a U.K. and a U.S. series of autopsy-confirmed cases of sporadic, late-onset Alzheimer’s. However, the association was smaller than the one found for the tauopathies.
Clearly, this initial result needs confirmation in independent, and larger, populations before the field can make much of it. A whiff of an association is emerging from the ongoing AlzGene database meta analysis (see Q&A below). Previous studies failed to find an association for tau, but differed from this one in that they compared the H1 and H2 haplotypes rather than analyzing variants within H1.
For now, the study is intriguing, given that the field generally views amyloid toxicity pathways as involving tau. The authors speculate that a tau variant affects either tau expression or splicing in such a way that it makes neurons more vulnerable to amyloid-β. “We believe our data are most consistent with the view that, in the presence of an amyloid load, those individuals with [tau] loci which are either highly expressing or prone to express a more pathogenic species of tau through alternate splicing (Rapoport et al., 2002) are more prone to disease,” the authors write.—Gabrielle Strobel.
Myers AJ, Kaleem M, Marlowe L, Pittman AM, Lees AJ, Fung HC, Duckworth J, Leung D, Gibson A, Morris CM, de Silva R, Hardy J. The H1c haplotype at the MAPT locus is associated with Alzheimer's disease. Hum Mol Genet. 2005 Aug 15 ; 14(16):2399-404. Abstract
Q&A with AlzGene curator Lars Bertram
Q: What is the AlzGene meta
A: For AlzGene, we aim to collect all genetic association studies performed in
AD. Thus far, we have collected roughly 800 studies on almost 300 different genes that were tested as AD candidate genes. In studies using a case-control design, we can use the published genotype numbers to estimate a crude odds ratio, that is, an estimate of the associated effect (if there is any). For any specific
Polymorphism, we can then use a statistical procedure—the meta-analysis—to estimate summary odds ratios, that is, the overall effect and its significance level across all published studies.
Q: Does tau show a signal in it?
A: Across the 16 case-control studies investigating the H1/H2 haplotype in AD to date, the overall summary odds ratio shows a marginally significant association suggesting a protective effect of the H2 haplotype (reducing the risk by about 5-10 percent), which can also be regarded as a risk effect of the H1 haplotype.
However, when studies are included in the meta-analysis that investigate the Q7R polymorphism in the saitohin gene (STH), which is located in one of the introns of tau and is in complete linkage disequilibrium with the H1/H2 haplotype, these effects go away. We will upload this information to Alzgene shortly.
Q: Outside of AlzGene, are you seeing an association between H1c and AD in
the patient series your group analyzes?
A: We have not investigated the tau locus yet.
Q: Myers et al. suggest the presence of a tau variant within the H1c
haplotype that makes carriers more prone to developing AD. This is not an
autosomal-dominant mutation of the sort that identified APP and the
presenilin genes. Can current genetics methods eventually pinpoint such a
pathologic gene variant in the tau locus?
A: While the more refined analysis of the H1c haplotype within the H1 clade could indeed reconcile some of the previous conflicting findings, it is unlikely that the risk effect of this variant is large. That means that large sample
sizes are probably necessary to identify the underlying variant, if it exists.
Q: What should the next step be in testing or substantiating the H1c data?
A: Replication, replication, replication—and, of course, meta-analysis of all
available data. But first, the authors of the previous tau-haplotype (and STH)
studies may want to re-examine their data and genotype markers that
distinguish the H1c haplotype from the more general H1 haplotype.
If consistently positive signals emerge from these studies and the
meta-analyses show a stronger or more
significant effect, there could indeed be a chance that an as yet unidentified
AD risk variant is hidden on the H1c background. But certainly this needs to
be evaluated independently in sufficiently sized samples before any further
conclusions can be drawn.