5 April 2013. To discover new risk genes for Alzheimer’s disease, researchers typically look for genetic markers overrepresented in people with the disease. This method provides no clue as to how a genetic variant might promote AD. In the April 4 Neuron, researchers led by Alison Goate at Washington University, St. Louis, Missouri, took a different tack. They performed a genomewide association study (GWAS) to find markers associated with elevated levels of tau and phosphorylated tau in cerebrospinal fluid (CSF). By starting from a known biomarker, the researchers hoped to find AD risk genes while gleaning clues to their underlying disease mechanism. The study yielded four hits, two known—namely the risk genes ApoE and TREM2—and two new. In follow-up studies using additional datasets, the researchers realized that one of the new genetic signals was also linked to having more tau pathology in the brain and a faster rate of cognitive decline. “The fact that we found a signal associated with CSF tau that also associates with various AD phenotypes validates this approach,” Goate told Alzforum. The research made it into the BBC News.
Other researchers agreed that the strategy shows promise. “I love that this study links genetic risk with specific disease phenotypes,” said Monica Carson at the University of California, Riverside. “GWAS often lump so many things together that they can miss correlations.”
In recent years, GWAS have turned up a number of genetic loci linked to AD, including markers near the genes BIN1, clusterin, and CR1. For most of these, it has proven challenging to find the functional genetic variant and figure out how it contributes to disease. Goate and colleagues thought it might be easier to do this if genetic hits were already tied to a phenotype, such as pathologic tau levels, that could be easily modeled in a cell culture system. This would allow researchers to test various candidate alleles in cultured cells and look for those that affect tau. A recent GWAS took a similar phenotypic approach, looking for genes linked to amyloid imaging results (see ARF related news story).
First author Carlos Cruchaga and colleagues performed a GWAS on almost 1,300 participants from several large AD studies. About half the participants had the disease; the rest were healthy elderly controls. The researchers found numerous single nucleotide polymorphisms (SNPs) linked to high CSF tau and ptau levels. They clustered in four genetic regions. To explore a link to AD, the authors tested these SNPs in three other datasets: a case-control study to look for increased AD risk; a longitudinal study to examine rates of cognitive decline; and a neuropathological series that revealed neurofibrillary tangles in the brain.
One of the implicated regions was the triggering receptor expressed on myeloid cells (TREM) gene cluster. In 2012, scientists reported that the R47H allele of TREM2 strongly increases the risk of developing AD (see ARF related news story). This microglia receptor plays a role in phagocytosis, and its appearance last summer in the human genetics of AD intensified interest in the inflammatory aspects of the disease. The new study tagged not only R47H, but also found evidence for two additional independent signals in this region of the genome. The stronger signal came from an SNP located near a gene called natural cytotoxicity triggering receptor 2 (NCR2), another member of the immunoglobulin superfamily. This finding further implicates the immune system in AD pathogenesis (see ARF Webinar).
Carson, whose research had implicated TREM2 with Alzheimer’s already a decade ago (see Schmid et al., 2002), pointed out that TREM2 expression surges during many types of brain damage, suggesting the common allele is neuroprotective. “We are going to see TREM2 coming up as a risk factor for a variety of neurodegenerative diseases,” she predicted. Researchers still need to determine if the R47H allele leads to loss of function, or a gain of inappropriate function, she added.
In addition to the TREM family cluster, the GWAS fingered several SNPs in the ApoE region. ApoE4, the major genetic risk factor for late-onset AD, is thought to affect Aβ levels. However, even after the researchers corrected for the peptide, they still saw a signal from ApoE, suggesting that this gene also affects tau levels through mechanisms unrelated to Aβ. “That may help explain why ApoE is such a strong risk factor for AD. There are two different mechanisms in play,” Goate suggested.
Stuart Pickering-Brown at the University of Manchester, U.K., found this result particularly interesting. Some researchers have found a link between ApoE and frontotemporal dementia (FTD). Around 45 percent of FTD cases are tauopathies, but this frequency varies in different populations, Pickering-Brown pointed out. Perhaps ApoE’s connection to tau explains why the gene seems to be linked to FTD in some studies but not in others that may have included populations with fewer cases of tauopathy-based disease, he speculated.
Another intriguing finding was a new signal from a chromosome 3 region. It is wedged between the genes geminin coiled-coil domain-containing protein 1 (GEMC1) and osteocrin (OSTN). These genes themselves do not seem to account for the association. The SNP linked to high CSF tau also associated with higher risk for AD, more tau tangles, and faster global cognitive decline. Finding the disease-causing variant in this region may prove a challenge, however, as several genes and non-coding RNAs lie nearby. “We have a strong genetic signal here with multiple phenotypes, but we don’t yet know which of these genes is likely to be driving the association,” Goate noted.
The fourth signal, also new, comes from SNPs in an intron of the GLIS family zinc finger 3 (GLIS3) gene. Goate is cautious about this one, because these SNPs showed no linkage to Alzheimer’s risk or tangle pathology. “It’s harder to tell what relevance this might have for AD,” she noted. The GLIS3 gene has been implicated in type 2 diabetes, which itself is a risk factor for AD. A pathway analysis on the whole GWAS dataset, including SNPs that fell below genomewide significance, found numerous signals in diabetes-related genes.
The researchers estimate that the four loci they found account for about 22 percent of the genetic variability in ptau and 9 percent of that in tau, implying that more genes related to tau levels remain to be discovered. In ongoing work, they have doubled their sample size to about 2,500 people to see if they can find more of these genetic ties. They are also following up on the TREM findings by sequencing genes in the cluster to pin down the disease-causing variants. Carson suggested that an important next step will be to discover where in the brain these candidate genes are expressed, and by what cell types. “This is going to stimulate a lot of cell type-specific functional studies that we can then bring back to the whole brain and see how everything pulls together,” Carson said.
Goate believes that some of these GWAS hits could point to new targets for tau-based drugs. Much previous tau research has focused on its kinases and phosphatases, without great therapeutic success to date. “This study identifies pathways that are completely separate from kinases and may be more amenable as therapeutic targets. It’s important to have targets that relate to a variety of different processes,” Goate said.—Madolyn Bowman Rogers.
Cruchaga C, Kauwe JS, Harari O, Jin SC, Cai Y, Karch CM, Benitez BA, Jeng AT, Skorupa T, Carrell D, Bertelsen S, Bailey M, McKean D, Shulman JM, De Jager PL, Chibnik L, Bennett DA, Arnold SE, Harold D, Sims R, Gerrish A, Williams J, Van Deerlin VM, Lee VM, Shaw LM, Trojanowski JQ, Haines JL, Mayeux R, Pericak-Vance MA, Farrer LA, Schellenberg GD, Peskind ER, Galasko D, Fagan AM, Holtzman DM, Morris JC, GERAD Consortium, Alzheimer’s Disease Neuroimaging Initiative (ADNI), Alzheimer Disease Genetic Consortium (ADGC), Goate AM. GWAS of cerebrospinal fluid tau levels identifies risk variants for Alzheimer’s disease. Neuron. 2013 Apr 24;78(2). Abstract