Analysis of alternative splicing associated with aging and neurodegeneration in the human brain

Abstract

Age is the most important risk factor for neurodegeneration; however, the effects of aging and neurodegeneration on gene expression in the human brain have most often been studied separately. Here, we analyzed changes in transcript levels and alternative splicing in the temporal cortex of individuals of different ages who were cognitively normal, affected by frontotemporal lobar degeneration (FTLD), or affected by Alzheimer's disease (AD). We identified age-related splicing changes in cognitively normal individuals and found that these were present also in 95% of individuals with FTLD or AD, independent of their age. These changes were consistent with increased polypyrimidine tract binding protein (PTB)-dependent splicing activity. We also identified disease-specific splicing changes that were present in individuals with FTLD or AD, but not in cognitively normal individuals. These changes were consistent with the decreased neuro-oncological ventral antigen (NOVA)-dependent splicing regulation, and the decreased nuclear abundance of NOVA proteins. As expected, a dramatic down-regulation of neuronal genes was associated with disease, whereas a modest down-regulation of glial and neuronal genes was associated with aging. Whereas our data indicated that the age-related splicing changes are regulated independently of transcript-level changes, these two regulatory mechanisms affected expression of genes with similar functions, including metabolism and DNA repair. In conclusion, the alternative splicing changes identified in this study provide a new link between aging and neurodegeneration.

Figure 1.

Overview of microarray and RT-PCR analyses of splicing profiles. (A) Hierarchical clustering of microarray data [(C) cognitively normal; (Ft) FTLD-tau; (Fp) FTLD-TDP with GRN mutation; (Fu) sporadic FTLD-TDP; (AD) Alzheimer's disease], with ages of individuals shown next to disease status. (B) Principal component analysis of RT-PCR data of 13 exons in 84 samples (data shown in Supplemental Fig. S2; Supplemental Tables S1, S2). PC1 values are compared to the age of the donor. The blue line shows the linear regression analysis of the cognitively normal individuals, indicating a trend for a more disease-like splicing pattern with increasing age.

Figure 2.

Analysis of age-related and disease-specific splicing changes. (A) Heatmap of percentage exon inclusion of the 2064 exons with age-related splicing changes (DAR < 1.2). (B) RT-PCR validation of a cohort of the age-associated splicing changes. (C) Heatmap of percentage exon inclusion of the 1551 exons with disease-specific splicing changes (DAR > 2). (D) RT-PCR validation of the disease-specific splicing changes. Exons regulated by TARDBP, NOVA, or PTB proteins are labeled on the side, with the enhancing or silencing activity of the protein indicated by an arrow. Age of individual donors and the disease type is shown on top; (Ft) FTLD-tau; (Fp) FTLD-TDP with GRN mutation; (Fu) sporadic FTLD-TDP.

Figure 3.

Analysis of the nuclear and cytoplasmic abundance of NOVA proteins. A total of six young healthy, five aging healthy, six FTLD-tau, and five AD samples were lysed into cytoplasmic (A) and nuclear (B) fractions. Immunoblot analysis of NOVA1, NOVA2 (detected in two different isoforms of ∼52 kDa and 70 kDa), and cytoplasmic (GAPDH) and nuclear (Histone H3) loading controls is shown. Mean and standard deviation of NOVA signal, normalized against the loading controls, is shown in the graphs. Significant difference in NOVA protein abundance is marked; (*) P-value < 0.05; (**) P-value < 0.01; two-sided Student's t-test, unequal variance. (C) Separation of cytoplasmic and nuclear fractions from human post-mortem samples was efficient. GAPDH was only present in the cytoplasmic fraction, whereas histone H3 was present only in the nuclear fraction.

Figure 4.

Analysis of cell-type-specific gene expression. The y-axis shows the enrichment of cell-type-specific genes (Cahoy et al. 2008) (see Methods) among the genes with increased or decreased age-related or disease-specific changes in transcript levels. (*) Significant difference in proportion of cell-type-specific genes (P-value < 0.0001, χ² test with Yates correction).