Rare but powerful gene variants help scientists understand how disease develops, but they are hard to find. By pooling data on people with different early onset dementias to get cohorts of a workable size, researchers pulled out an association between mutations in the DNA demethylase ten-eleven translocation 2 (TET2) and elevated risk for Alzheimer’s disease, frontotemporal dementia, and amyotrophic lateral sclerosis. In the May 7 American Journal of Human Genetics, Jennifer Yokoyama at the University of California, San Francisco, and Richard Myers at the HudsonAlpha Institute for Biotechnology in Huntsville, Alabama, and colleagues report that both coding and noncoding TET2 variants nearly doubled a person’s risk, and also meant faster decline in people who already had mild cognitive impairment.

  • Mutations in TET2 double the risk for Alzheimer’s and frontotemporal dementia.
  • Variants in coding and noncoding parts of TET2 were equally damaging.
  • TET2 mutations hastened decline in people with MCI.

The findings strengthen the idea that certain epigenetic changes can lead to neurodegeneration. The data also suggest that noncoding variants can have a powerful effect, and may indeed be the source of much of the still-unexplained heritability of these diseases, Yokoyama said.

Julie van der Zee at the VIB-University of Antwerp, Belgium, believes this last point is important. “An added value of their study design is that they demonstrate that one can have sufficient power to associate disease risk with noncoding regulatory variants,” she wrote to Alzforum (full comment below).

Accelerated Decline. Extrapolations predict that TET2 carriers will worsen faster on the CDR-SB than noncarriers. [Courtesy of Cochran et al., American Journal of Human Genetics.]

Geneticists search for rare disease variants in familial and early onset cohorts, but these are small, making it difficult to know whether a hit is statistically significant (Aug 2017 news). Or they sift through exome sequences, but these leave out noncoding variants.

Instead, Yokoyama and colleagues looked for variants linked to two different forms of early onset dementia. They capitalized on pleiotropy, or the ability of one gene to contribute to multiple diseases. The authors pooled data from clinically diagnosed AD and FTD patients seen at the UCSF Memory and Aging Center and the University of Alabama at Birmingham. The AD patients ranged from 45 to 84 years old, with an average age of 59, hence were enriched for early onset cases. Healthy age-matched controls came from UCSF, UAB, and a National Institute of Mental Health healthy aging cohort. People with known genetic mutations, or those who were related to each other, were excluded. The authors also excluded people with non-European ancestry, hoping to make the sample more homogeneous. They ended up with 671 controls and 435 cases, 227 of whom had EOAD and 208 FTD.

First author Nicholas Cochran at HudsonAlpha analyzed cases versus controls to find rare genetic variants enriched in the former. Cochran weeded out any that were predicted to be harmless. To do that, he used Combined Annotation Dependent Depletion (CADD) scores, a metric for assessing how likely a variant is to be pathogenic. The authors considered only variants with CADD scores above 10.

TET2 was the only gene that came up as significant across the whole cohort. Putative deleterious TET2 variants cropped up in 11 people with EOAD, eight with FTD, and one control. Two of them harbored the same variant, for a total of 19 different variants. Intriguingly, nine were in coding regions, 10 in noncoding regions.

Although the variants were equally enriched among participants with EOAD and FTD, the association with disease did not reach statistical significance in either of these smaller groups alone. This suggests that the strategy of pooling across diseases helped uncover the association.

To see if the finding held up, the authors turned to the much larger, late-onset AD cohorts of the Alzheimer’s Disease Sequencing Project (ADSP) and the Accelerating Medicines Partnership (AMP-AD), as well as early onset AD participants of non-European ancestry in the UCSF cohort. Together, this dataset comprised 2,849 EOAD, LOAD, and FTD cases, and 2,457 controls. Again, TET2 variants associated with disease in the whole cohort, at p=0.003, but missed statistical significance among AD or FTD patients alone.

The authors also saw an enrichment of TET2 mutations in an ALS cohort, Project MinE, suggesting the gene may contribute to this disease, as well. ALS and FTD form part of a biologically related disease spectrum (Sep 2011 newsApr 2018 news). 

The strength of the disease association varied widely across cohorts, with the greatest effect on early onset disease. In the original UCSF and UAB EOAD cohorts, TET2 mutations gave an odds ratio of 29 for developing early onset AD/FTD, while in the smaller, non-European ancestry UCSF cohort, variants increased the likelihood of early onset AD/FTD by 6.4-fold. In the other replication cohorts, which consisted of mostly late-onset AD cases, the odds ratio was less than two. Across all cohorts, this averaged to an odds ratio of 2.3. Rita Guerreiro at the Van Andel Institute in Grand Rapids, Michigan, noted that because of these group differences, it will be important to replicate the TET2 findings in additional cohorts (see comment below).

Do TET2 variants predict how fast a patient will deteriorate? Among 786 ADNI participants who have LOAD, TET2 mutation carriers worsened faster than noncarriers on the CDR-SB. This was particularly pronounced among people with mild cognitive impairment. In the MCI group, carriers declined by an additional 0.64 points per year on the CDR-SB compared with noncarriers. Extrapolated over 12 years for the whole cohort, this comes out to a 10-point decline for carriers, versus two points for noncarriers (see image above). TET2 mutation carriers with MCI also slid 0.43 points per year faster on the MMSE than did noncarriers. However, this cognition finding did not replicate in the UCSF cohort, which has less longitudinal follow-up.

How might TET2 affect disease risk? Seven of the nine coding variants were predicted to result in a loss of function, suggesting the protein is normally protective. As a group, these coding variants tripled a person’s risk of developing AD, FTD, or ALS.

The noncoding variants were even more powerful, with an odds ratio of 3.7. “We were quite surprised by that finding,” Yokoyama said. It suggests that noncoding variants might nearly abolish expression of the gene, resulting in haploinsufficiency, she noted. The authors are now expressing each variant in cultured cells to determine their effects on gene expression. They will also try to identify which genes TET2 demethylates. Typically, demethylation releases gene repression, allowing more transcription of its targets.

Curiously, global brain methylation drops in aging and AD. This would seem to be at odds with a protective effect for the TET2 demethylase (Jun 2010 conference news). But the effect of a specific demethylase depends on its targets; for example, demethylation boosts expression of reelin, which aids memory formation (Mar 2007 news). TET2 mutations have been found to pop up with high frequency in somatic cells in AD and PD blood and brain (Oct 2018 news). In a mouse model of amyloidosis, loss of TET2 accelerated plaque formation and memory problems (Li et al., 2020). 

Recently, TET2 was reported to promote a proinflammatory response in microglia, and to be elevated in microglia surrounding amyloid plaques in mouse models and AD brain (Carrillo-Jimenez et al., 2019). “The question remains if TET2 in plaque-associated microglia plays a deleterious or beneficial role in AD,” noted Miguel Burguillos, University of Seville, Spain, the senior author of that study, in an email to Alzforum (see comment below). “Further studies using conditional TET2 knockout mice crossed with an AD model may answer this.” Burguillos believes the varying levels of TET2 in neurons and microglia in mouse models hint that the gene could have multiple effects in AD brain.—Madolyn Bowman Rogers


  1. It is nice to see that large-scale genome-sequencing projects are starting to deliver new risk factors for the rarer, early onset dementia phenotypes like FTD and EOAD. In contrast to whole-genome sequencing, we can now zoom in to the impact of rare variants, with moderate to high penetrance and study how they impact one's genetic risk profile.

    An added value of their study design is that they demonstrate that one can have sufficient power to associate disease risk with noncoding regulatory variants.

  2. This paper by Cochran et al. reports on the potential role of TET2 variants in neurodegenerative diseases. The authors search for variants that contribute to disease via haploinsufficiency by focusing on loss-of-function variants or rare coding and noncoding variants after filtering conditions. Their approach builds upon the pleiotropy known to be part of the genetic architecture of dementias and neurodegenerative diseases by using a discovery cohort including EOAD and FTD and replication cohorts including LOAD, FTD, and ALS.

    The only significant association identified was between filtered TET2 variants and EOAD and FTD combined, compared to controls. TET2 is the type of gene one would expect to see associated with different forms of neurodegenerative disease given its known role in DNA methylation. However, the differences seen in the association across cohorts suggest that independent follow-up studies are needed to definitively establish TET2 as a gene associated with neurodegenerative disease risk. In fact, there is a wide range of odds ratios seen in the different cohorts, from very strong effects (OR=29) in the UCSF discovery series to completely null (OR=1) in the AMP-AD series, and when assessing risk association only in the larger cohorts (ADSP and AMP-AD), the results are not as significant.

    Important factors that may have contributed to biasing the results include the young age of the controls in the discovery set, the small size of the discovery cohort, and the impossibility of validating variants identified in non-UCSF samples. Still, if confirmed, the association of TET2 gene variants with AD and FTD suggests a possible role for epigenetic age clock signatures in neurodegenerative diseases.

  3. Using genome sequencing, the authors identified deleterious variations in TET2, with predicted loss-of-function outcomes, as a risk factor for multiple neurodegenerative disorders, including EOAD, LOAD, FTD, and ALS. This is an interesting paper where Cochran and colleagues hypothesized that non-functional TET2 enzymatic activity in neurons could explain the difference between normal physiological aging and neurodegenerative disease based on studies where increased expression of TET2 promoted neurogenesis (Li et al., 2017) and in a 2xTg AD mouse model where there is a decreased expression of Tet2 in the hippocampus (Li et al., 2020). 

    A common feature of neurodegenerative diseases is the presence of a chronic inflammatory response, which in many cases is considered to be detrimental for the surrounding neuronal population. Within the last five years, several studies have highlighted the role played by TET2 in the control of the inflammatory response from either the lymphoid (Ichiyama et al., 2015) or myeloid (Zhang et al., 2015) lineage, including our group's recent publication in microglia cells (Carrillo-Jimenez et al., 2019), where TET2 was necessary for the full proinflammatory response upon TLR-4 agonist treatment. In this context, TET2 is known to be able to regulate the inflammatory response dependent on (in the case of T-cells) or independently of (peripheral macrophages and microglia) its enzymatic activity. For instance, in our study of microglia cells, and in peripheral macrophages, TET2 plays an important activity-independent role upon TLR4 activation (Carrillo-Jimenez et al., 2019; Zhang et al., 2015). In our study, we also describe high protein expression of TET2 in microglia surrounding the amyloid plaque in a 5xFAD AD mouse model, but the function of TET2 in these microglia is yet to be elucidated. Therefore, the question remains if TET2 in plaque-associated microglia plays a deleterious or beneficial role in AD, and furthermore if the enzymatic activity of TET2 is required.

    To address the latter question, more in-depth studies of TET2 post-translational modifications could shed light on the role that TET2 has under different disease conditions. For instance, TET2 acetylation by P300 increases its enzymatic activity (Zhang et al., 2016). Regarding whether microglial TET2 plays a beneficial or deleterious role in AD, further studies using conditional TET2 KO mice crossed with an AD mouse model may answer this.

    In summary, the varying levels of TET2 expression in neurons versus microglia in AD disease models suggest multiple roles for TET2 in neurodegenerative disease (i.e., aging in neurons and modulation of microglia reactivity) where a loss-of-function variant of TET2 could be considered a risk factor for neurodegenerative diseases, including AD.


    . Ten-eleven translocation 2 interacts with forkhead box O3 and regulates adult neurogenesis. Nat Commun. 2017 Jun 29;8:15903. PubMed.

    . Reduction of Tet2 exacerbates early stage Alzheimer's pathology and cognitive impairments in 2 × Tg-ad mice. Hum Mol Genet. 2020 Jan 15; PubMed.

    . The methylcytosine dioxygenase Tet2 promotes DNA demethylation and activation of cytokine gene expression in T cells. Immunity. 2015 Apr 21;42(4):613-26. Epub 2015 Apr 7 PubMed.

    . Tet2 is required to resolve inflammation by recruiting Hdac2 to specifically repress IL-6. Nature. 2015 Sep 17;525(7569):389-393. Epub 2015 Aug 19 PubMed.

    . TET2 Regulates the Neuroinflammatory Response in Microglia. Cell Rep. 2019 Oct 15;29(3):697-713.e8. PubMed.

    . A Unique Microglia Type Associated with Restricting Development of Alzheimer's Disease. Cell. 2017 Jun 15;169(7):1276-1290.e17. Epub 2017 Jun 8 PubMed.

    . Tet2 is required to resolve inflammation by recruiting Hdac2 to specifically repress IL-6. Nature. 2015 Sep 17;525(7569):389-393. Epub 2015 Aug 19 PubMed.

    . Acetylation Enhances TET2 Function in Protecting against Abnormal DNA Methylation during Oxidative Stress. Mol Cell. 2017 Jan 19;65(2):323-335. PubMed.

  4. With great interest have we read this article by Cochran et al., who used peripheral blood cells as a resource for the germline genome. Recent developments in the field of hematology have shown that somatic mutations in the TET2 gene have been found to drive age-related clonal hematopoiesis (ARCH). Therefore, it may be possible that the effect observed by Cochran et al. is an age-related effect, and not an AD-related effect.

    Recent reports have shown that while humans age, our blood cells become more and more clonal (Genovese et al., 2014; Jaiswal et al., 2014; Xie et al., 2014). When we are young, our blood cells are generated by several thousands of hematopoietic stem cells (HSCs), which each produce offspring that contribute to the multiple hematopoietic lineages that constitute our peripheral blood, i.e., all myeloid and lymphoid cells.

    However, as we age, many of these HSCs die or become inactive, while a few survive and gain a competitive advantage over other HSCs. This competitive advantage is driven by the accumulation of somatic mutations in several genes, of which somatic mutations in the TET2 and DNMT3A genes are most common (Busque et al., 2012; Genovese et al., 2014; Jaiswal et al., 2017). Both genes are associated with epigenetic regulation, i.e., DNA-methylation and -demethylation processes.

    As we age, the HSC clone will expand and contribute to an increasing fraction of cells in our peripheral blood. This process has been coined “age-related clonal hematopoiesis,” ARCH (Shlush, 2018; Steensma et al., 2015). The somatic mutations accumulated by the stem cell over the course of a lifetime effectively “tag” each individual stem cell with a unique genetic “barcode” (Zink et al., 2017). 

    Acquired somatic mutations are heterozygous in the clone, and because the clone contributes to only a fraction of the total peripheral blood cells, these mutations can be recognized by their allele balance (between alternate allele and reference allele), which will be consistently lower than the 1:1 ratio observed for germline heterozygous mutations. With age, all somatic mutations accumulated in the HSC during aging will become more and more prevalent in the blood cells, and more and more detectable by genetic sequencing.

    In 2014, we reported that ~70 percent of the peripheral blood cells in a 115-year-old woman were generated by one hematopoietic stem cell “clone” and its subclones (Holstege et al., 2014). Recently, we found that one clone contributed to 75 percent of the peripheral blood cells in a 111-year-old woman, and that the clone generated almost all myeloid cells (van den Akker et al., 2020). Note that neither of these women suffered from hematopoietic malignancies.

    In the discovery analysis of this paper, the authors compared older AD and FTD cases with young controls. They find that mutations in the TET2 gene associated with a 29-fold increased risk of AD/FTD (OR: 28.9, 95%CI 4.5–1200, p= 4.9E-7; AD cases: N=227, median age 59, 23% >65; FTD cases: N=208, median age 65 year, 49% > 65; controls N=671; median age 40 years, 9% >65). However, in a replication analysis, in which controls were more age-matched with the cases (although still slightly younger), the authors found that the TET2 mutations associated with a 1.7-fold increased AD/FTD risk and with lower significance (OR= 1.7, 95%CI 1.2–2.6, p= 6.1-3; AD cases: N=2,530, median age 79, 90% >65; FTD cases: N=319, median age 76 year, 78% > 65; controls N=2,457; median age 74 years, 84% >65). This indicates an age relation in the burden of TET2 mutations in the samples, that is independent of the case-control association.

    A current debate in the hematology field is whether ARCH causes age-related disease or is a consequence of aging, and thereby indirectly associated with age-related diseases. Indeed, ARCH was previously associated with several age-related hematological diseases such as myeloid dysplasia or malignancies (Genovese et al., 2014; Jaiswal et al., 2014), and large prospective epidemiological studies have also established associations between ARCH and prevalent Type 2 diabetes, prevalent chronic obstructive pulmonary disease, incident cardiovascular disease, and vascular or all-cause mortality (Bonnefond et al., 2013; Genovese et al., 2014; Jaiswal et al., 2014; Jaiswal et al., 2017; Zink et al., 2017). However, ARCH has also been observed in many individuals with no apparent disease (van den Akker et al., 2016). 

    To confirm that germline variants in the TET2 gene are Alzheimer’s disease risk variants, the authors would need to separate the somatic signal from the germline signal. This could be done by investigating whether the allele balance of the heterozygous TET2 mutations is consistently lower than the allele frequency of heterozygous germline (common) variants. Moreover, the authors might investigate whether, in these datasets, a similar signal can be observed for somatic mutations in the DNMT3A gene and/or other ARCH-related genes (Jaiswal et al., 2017). 

    We encourage the authors to perform these follow-up analyses to confirm that the variants are germline, somatic, or a mix of both. The results might lead to confirmation of the conclusions in the paper, or they may provide evidence that AD should be included in the list of age-related diseases associated with carrying a TET2 mutation, i.e. ARCH. Alternatively, the results may indicate that somatic TET2 mutations lead to ARCH, but not to an increased risk of AD.


    . Association between large detectable clonal mosaicism and type 2 diabetes with vascular complications. Nat Genet. 2013 Sep;45(9):1040-3. Epub 2013 Jul 14 PubMed.

    . Recurrent somatic TET2 mutations in normal elderly individuals with clonal hematopoiesis. Nat Genet. 2012 Nov;44(11):1179-81. Epub 2012 Sep 23 PubMed.

    . Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence. N Engl J Med. 2014 Dec 25;371(26):2477-87. Epub 2014 Nov 26 PubMed.

    . Somatic mutations found in the healthy blood compartment of a 115-yr-old woman demonstrate oligoclonal hematopoiesis. Genome Res. 2014 May;24(5):733-42. Epub 2014 Apr 23 PubMed.

    . Age-related clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014 Dec 25;371(26):2488-98. Epub 2014 Nov 26 PubMed.

    . Clonal Hematopoiesis and Risk of Atherosclerotic Cardiovascular Disease. N Engl J Med. 2017 Jul 13;377(2):111-121. Epub 2017 Jun 21 PubMed.

    . Age-related clonal hematopoiesis. Blood. 2018 Feb 1;131(5):496-504. Epub 2017 Nov 15 PubMed.

    . Clonal hematopoiesis of indeterminate potential and its distinction from myelodysplastic syndromes. Blood. 2015 Jul 2;126(1):9-16. Epub 2015 Apr 30 PubMed.

    . Uncompromised 10-year survival of oldest old carrying somatic mutations in DNMT3A and TET2. Blood. 2016 Mar 17;127(11):1512-5. Epub 2016 Jan 29 PubMed.

    . Age-related mutations associated with clonal hematopoietic expansion and malignancies. Nat Med. 2014 Dec;20(12):1472-8. Epub 2014 Oct 19 PubMed.

    . Clonal hematopoiesis, with and without candidate driver mutations, is common in the elderly. Blood. 2017 Aug 10;130(6):742-752. Epub 2017 May 8 PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Searching for New AD Risk Variants? Move Beyond GWAS
  2. Corrupt Code: DNA Repeats Are Common Cause for ALS and FTD
  3. Genetics Tie ALS into the Frontotemporal Dementia Spectrum
  4. Bethesda: The Methylated Brain
  5. Heavy Methyl—DNA, Protein Modification Affect Memory, APP, and Tau
  6. Islands of Mutated Neurons Dot the Brain. Are They Bad for Us?

Paper Citations

  1. . Reduction of Tet2 exacerbates early stage Alzheimer's pathology and cognitive impairments in 2 × Tg-ad mice. Hum Mol Genet. 2020 Jan 15; PubMed.
  2. . TET2 Regulates the Neuroinflammatory Response in Microglia. Cell Rep. 2019 Oct 15;29(3):697-713.e8. PubMed.

External Citations

  1. ombined Annotation Dependent Depletion
  2. Project MinE

Further Reading

Primary Papers

  1. . Non-coding and Loss-of-Function Coding Variants in TET2 are Associated with Multiple Neurodegenerative Diseases. Am J Hum Genet. 2020 May 7;106(5):632-645. Epub 2020 Apr 23 PubMed.