TMEM106B risk variant is implicated in the pathologic presentation of Alzheimer disease

TDP-43 protein is the major component of the ubiquitin-positive inclusions in neurons and glia of patients with amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD-TDP).¹ TDP-43 pathology has also been detected in as many as 56% of patients with Alzheimer disease (AD) and in 70% of patients with AD and concomitant hippocampal sclerosis (HpScl).² Importantly, clinical, neuropsychological, and imaging studies suggested that the presence of TDP-43 pathology in AD may be associated with a modified phenotype.³ A better understanding of what factors predispose to TDP-43 pathology in AD is therefore critical and could have important clinical implications.

Last year, a genome-wide association study identified the uncharacterized transmembrane protein 106B (TMEM106B) as a novel risk factor for FTLD-TDP.⁴ Follow-up studies confirmed the importance of TMEM106B in FTLD and suggested that TMEM106B may influence risk for FTLD-TDP by modulating the levels of the secreted growth factor progranulin (GRN).⁵ Here, we study the role of TMEM106B in the pathologic presentation of AD using TMEM106B SNP rs1990622, previously associated with reduced levels of GRN in human plasma.⁵

Methods.

We studied a cohort of 907 white AD cases (57% female) from the Mayo Clinic Brain Bank. The neuropathologic diagnosis of AD was made according to NIA–Reagan criteria and mean age at death was 80.3 ± 9.4 years. The presence of HpScl was diagnosed if there were neuronal loss and gliosis in the subiculum and CA1 regions of the hippocampus that were disproportionate to the degree of neurofibrillary degeneration. TDP-43 immunoreactivity was assessed in a standardized section of medial temporal lobe using TDP-43 immunohistochemistry (rabbit polyclonal antibody; ProteinTech Group, Chicago, IL; n = 167)² or an affinity-purified C-terminal specific polyclonal antibody to TDP-43⁶ (n = 740). Genotyping of TMEM106B rs1990622 was performed using an inventoried Taqman SNP genotyping assay (Applied Biosystems).

PLINK software (http://pngu.mgh.harvard.edu/purcell/plink/) was used to perform logistic regression analysis of TMEM106B rs1990622 under an additive, dominant, and recessive model adjusting for age, sex, and presence of the APOE ϵ4 allele.

Results.

Out of a total of 907 pathologically confirmed AD cases, 301 cases (33.2%) showed abnormal TDP-43 immunoreactivity. HpScl was present in 88 AD cases (9.7%). Association analyses of TMEM106B rs1990622 in this pathologically confirmed cohort showed a highly significant decrease in the frequency of the rs1990622 C-allele in AD cases with TDP-43 pathology compared to AD cases without TDP-43 pathology (C-allele frequency of 37.8% vs 46.5%; p = 5.0 × 10⁻⁴) (table 1). More specifically, there were fewer homozygous carriers of this minor C-allele in the subgroup of AD cases with TDP-43 pathology (CC genotype frequency of 13.0% vs 20.7%; p = 5.0 × 10⁻³ in a recessive model). Association analyses further showed a highly significant association of rs1990622 with the presence of HpScl (p = 1.95 × 10⁻⁶) (table 1). AD cases carrying at least 1 copy of the rs1990622 C-allele were significantly less likely to develop HpScl (odds ratio [OR] = 0.39; 95% confidence interval [CI] = 0.27–0.57; p = 8.36 × 10⁻⁷ in an additive model). The association of rs1990622 with HpScl persisted when all patients with TDP-43 immunoreactivity were excluded from the analyses (OR = 0.42; 95% CI = 0.18–0.97; p = 0.04 in an additive model) (table e-1 on the Neurology® Web site at www.neurology.org). Similarly, when all patients with HpScl were excluded from the analyses, rs1990622 continued to show a significant association with TDP-43 pathology (p = 0.04 in a recessive model; table e-2), suggesting the associations of rs1900622 with TDP-43 pathology and HpScl are, at least in part, independent.

Table 1.

Association analyses of TMEM106B in pathologically confirmed AD series

Abbreviations: AD = Alzheimer disease; ADD = additive; CI = confidence interval; DOM = dominant; HpScl = hippocampal sclerosis; OR = odds ratio; REC = recessive.

Discussion.

We evaluated the contribution of the TMEM106B rs1990622 risk variant to the development of TDP-43 pathology and HpScl. In AD cases with TDP-43 pathology we showed significantly reduced frequencies of homozygote carriers of the minor C-allele of rs1990622 compared to AD cases without TDP-43 pathology. Since the minor C-allele of rs1990622 was previously associated with increased GRN levels,⁵ we speculate that reduced levels of GRN may increase the risk to develop TDP-43 pathology in AD. The mechanisms by which low levels of GRN lead to TDP-43 pathology are not completely understood; however, activation of programmed cell death pathways may be involved.⁷

HpScl is common in elderly subjects with dementia. Interestingly, HpScl can be detected in more than 75% of FTLD-TDP cases and up to 83% of GRN mutation carriers, suggesting a link between HpScl, TDP-43 pathology, and GRN levels. In our AD series, we observed a highly significant association of rs1990622 with HpScl. We showed that AD cases carrying at least 1 minor C-allele were significantly protected from the development of HpScl. These data suggest that increased levels of this neurotrophic factor in the hippocampus may protect against neurotoxic insults which would otherwise lead to hippocampal damage.

Together, these data implicate TMEM106B in the pathologic presentation of AD.