16 October 2008. The unfortunate mouse is hardly an ideal model for human disorders, particularly those, such as Alzheimer disease (AD), that affect higher brain function. In addition to vastly different physiology and intelligence, we now have another reason to question the limits of mouse models. Researchers led by Bruce Yankner, Harvard Medical School, have compared brain transcriptomes among aging humans, rhesus monkeys, and mice. They identified an evolutionary divergence in age-related gene expression that suggests mice and humans grow old very differently. The finding might explain, among other things, why humans get AD, while mice do not. “Aging is the predominant risk factor for neurodegenerative disorders such as Alzheimer’s disease and Parkinson’s disease, and many of their clinical and pathological features are specific for humans,” Yankner told ARF. “The question is whether this reflects a different substrate of aging in humans versus other mammalian species.” That remains to be determined, but the researchers did find stark differences between the transcriptome of older humans and mice.
The study is reported in the October PLoS One. Joint first authors Patrick Loerch and Tao Lu compared age-related transcriptional changes in cortical samples from all three species. Though they found many transcripts that showed significant variation with age in humans (2,347 genes), mice (3,542 genes), and to a lesser extent rhesus monkeys (537 genes), only 154 genes were associated with aging in all three species. “That suggests to me that there aren’t common cohorts of genes or genetic programs associated with aging that are conserved between species,” Ben Barres of Stanford University in California, told ARF.
From those 154 genes, the researchers identified three main types. Thirty-five genes were phylogenetically conserved, in that they showed the same up or down age-related change in all three species; however, the bulk of the genes (110) were differentially expressed between mice and monkeys (only nine genes were differentially expressed between aged monkeys and aged humans). “This shows that there has been a major divergence in the transcriptome of the aging brain during mammalian evolution, and that this has occurred somewhere between mice and primates,” said Yankner.
To identify what cells are affected by these age-related changes, Loerch and colleagues used the Allen Brain Atlas of gene expression. The atlas correlates expression data to cell type in mice, but the researchers determined that it works equally well for human cell types. In both species more genes were upregulated with age in astrocytes and oligodendrocytes than would be expected by chance. The scientists confirmed this using a transcriptome database generated from homogenous populations of astrocytes, oligodendrocytes, and neurons (see ARF related news story).
Humans differed from mice, however, in showing a significant age-related downregulation of many neuronal genes. The authors confirmed the nature of those transcripts by cross-referencing to neuronal gene ontology (GO) groups, of which 24 were significantly enriched for age-regulated human genes while only five contained age-related mouse transcripts. On closer inspection, the researchers found that most neuronal transcripts changing with age in humans were downregulated, while in mice the opposite was true: most of them were upregulated.
For his part, Barres was surprised that there aren't more and bigger changes than those that were seen. “After all, human brains noticeably shrink with age. It may well be that because there are signaling mechanisms that closely link neuronal and glial cell number (for instance, they each support each other’s survival), with aging both populations concordantly shrink so that larger relative gene changes are not seen,” he suggested.
Could this cross-species difference in the age-related transcriptome explain why humans are susceptible to neurodegenerative diseases? It is too early to say, but Yankner noted that AD research to date has focused almost exclusively on the disease and is almost divorced from the fact that it occurs in old people. “The current trend has only gotten us so far,” said Yankner. “The hope is that this sets a framework for many labs to start thinking about this issue and to try to understand why the human brain undergoes this major change with age, and how this interfaces with neurodegenerative disease.”
One clue offered by Loerch and colleagues pertains to inhibitory neurotransmission. They found that several genes that are downregulated with age in humans, but not in mice, are involved in GABAergic transmission. These genes include GABA receptor subunits, transporters, and enzymes involved in GABA synthesis, and also markers of GABAergic interneurons. Reduced inhibitory circuits might be compensatory, the authors suggest, but the ensuing enhanced cortical activation might eventually become excitotoxic. Interestingly, Loerch and colleagues found no evidence for either age-related cell or synapse loss in human tissue samples.
As for specific AD-related genes identified in this analysis, presenilin-2 was found to be downregulated in aging humans by about 1.3-fold. APLP2 and the APP-binding protein APPBP1 are also age-downregulated, by 1.2- and 1.25-fold, respectively, but whether any of those changes pertain to AD susceptibility is not clear, said Yankner. PS1 and APP transcript levels were not age-dependent in all three species. Another gene that was significantly upregulated with age in all three species was ApoD, which protects against oxidative damage and is essential for flies to live out their full lifespan (see Sanchez et al., 2006). ApoD has been reported to be upregulated in AD brain also (see Kalman et al., 2000 and Belloir, 2001). “ApoD is normally made by oligodendrocytes, but is also made by astrocytes, and it becomes a major protein in reactive gliosis, so that is very interesting,” said Barres. ApoD is also strongly neuroprotective in a mouse model of viral encephalitis (see Do Carmo et al., 2008). Calmodulin-dependent protein kinase 4 (CAMK4) was significantly downregulated across the three species. CAMK4 is involved in cAMP signaling cascades that modulate synaptic plasticity, which is essential for learning and memory.
There have been recent questions about the validity of mouse models for human diseases and for preclinical study (see Alzforum Live Discussion and related Nature news coverage); however, Barres said that this study does not question the validity of those models. “The differences that were seen are interesting, but it is not really clear yet what the significance is. I think it is premature to make any major conclusions, but the findings of this study do suggest a very interesting new possibility for why mouse Alzheimer's models fail to mimic the dramatic degree of synapse loss observed in the human disease,” he said.—Tom Fagan.
Loerch PM, Lu T, Dakin KA, Vann JM, Isaacs A, Geula C, Wang J, Pan Y, Gabuzda DH, Li C, Prolla TA, Yankner BA. Evolution of the aging brain transcriptome and synaptic regulation. PLoS One 2008 October; 3: e3329. Abstract