The relatively recent realization that neurogenesis is alive and kicking in the adult hippocampus has scientists wondering if the phenomenon can be exploited to protect against neurodegenerative diseases such as Alzheimer’s. That idea took on new momentum with the discovery that some risk factors for AD, including diabetes and lack of exercise, reduce adult neurogenesis in mammals. Recent studies in rodents may build on that momentum. In the February 17 Nature Cell Biology, Zhi-Cheng Xiao, Gavin Dawe, and colleagues at the National University of Singapore and elsewhere report that TAG1, a protein found in the outer plasma membrane, binds APP and suppresses neurogenesis in a manner dependent on presenilin and the APP intracellular domain. Though it is not clear if this pathway also impinges on adult neurogenesis, the finding is sure to raise interest among Alzheimer disease researchers given the links between adult neurogenesis and cognition. Also, a Nature Neuroscience paper out the same day shows that diabetes suppresses hippocampal neurogenesis and synaptic plasticity and learning in rats and mice. Mark Mattson and colleagues at the National Institute on Aging in Baltimore, Maryland, also report that these effects improve when corticosterone levels go down. The findings suggest that controlling stress hormones might be one strategy to mitigate the effects of diabetes on the brain.
Presenilin activity has been linked to neurogenesis before, but this work of Xiao and colleagues may be the best evidence yet that APP may be involved. Similarities between Notch and APP processing suggest that somewhere in the extracellular milieu lurks an APP ligand that can set off an intracellular response. This is what led Xiao and colleagues to make the connection between APP and TAG1. The latter is a member of the F3 family of cell adhesion molecules, and the researchers recently showed that F3, and its homolog NB-3, activate Notch signaling (see Hu et al., 2003 and Cui et al., 2004). Finding that TAG1-expressing Chinese hamster ovary (CHO) cells bound to APP-coated culture dishes, the researchers began to characterize the TAG1-APP interaction.
In immunoprecipitation experiments, first author Quan-Hong Ma and colleagues found that antibodies to APP could pull down TAG1 from mouse brain extracts and vice versa. The researchers also found that in embryonic brain tissue, APP and TAG1 co-localize with nestin, a marker of neural progenitor cells (NPCs), and that APP and TAG1 are coexpressed in the walls of the lateral ventricle, a stem cell niche. They found that the two proteins also co-localized in isolated NPCs.
To test whether APP and TAG1 binding might be linked to neurogenesis, Ma and colleagues assessed the neuron-forming potential of NPCs from knockout animals. Both APP-/- and TAG1-/- NPCs generated about 50 percent more new neurons than did wild-type progenitor cells, indicating that APP and TAG1 suppress neurogenesis. Adding just TAG1 to APP/TAG1 double negative NPCs had no effect on neurogenesis, suggesting that APP is needed to mediate TAG1 activity. On that point, expressing AICD59, a construct containing the APP intracellular domain, did suppress neurogenesis. Taken together, these experiments hint that TAG1 might set off a signal transduction pathway through APP that regulates the production of new neurons.
Several lines of evidence support this potential signaling pathway. TAG1 activates a reporter system based on APP processing and AICD release, for example. The authors also found that TAG1 promotes production of endogenous AICD in murine embryonic fibroblasts and that AICD levels in E15 brains isolated from TAG1+/- and TAG1-/- mice are reduced. As for regulating neurogenesis, the researchers found that the number of cells positive for neuronal markers was significantly elevated when ventricular wall NPCs from APP- and TAG1-negative mice were allowed to differentiate. “These results demonstrate that the interaction between TAG1 and APP may be involved in modulating neurogenesis during the early stages of CNS development,” write the authors.
If TAG1 and APP conspire to release AICD, then what lies between it and the suppression of neurogenesis? Fe65 is a well-known AICD-binding protein that has been linked to downstream effects. Guojun Bu and colleagues recently found that AICD suppresses lipoprotein receptor 1 (LRP1) expression, most likely in cooperation with Fe65 (see ARF related news story). It appears the AICD-Fe65 combination is important for modulating neurogenesis, too, because TAG1 failed to suppress neurogenesis from NPCs isolated from Fe65-negative mice. In addition, while AICD59 suppressed the enhanced neurogenic capabilities of TAG1-negative NPCs, mutating the AICD NPTY motif, needed for binding Fe65, abolished this suppression. All told, the findings suggest that Fe65 acts downstream of AICD and TAG1 in negatively regulating neurogenesis.
The finding by Xiao and colleagues may come as surprise, given that there is a small literature suggesting a role for presenilin function in maintaining neurogenesis. PS1 deficient mice are impaired in that respect (see Shen et al., 1997), and presenilin FAD mutations limit neurogenesis in the hippocampus (see ARF related news story). Just last month, Woo-Young Kim and Jie Shen, from Brigham and Women’s Hospital in Boston, reported that presenilins are essential for the maintenance of neural stem cells in the developing mouse brain. Kim and Shen conditionally knocked out presenilins in neural progenitor cells, and in the January 8 Molecular Neurodegeneration online reported that the number of NPCs plummeted in double presenilin knockouts. Not surprisingly, this led to severe morphological defects.
Interestingly, Kim and Shen discovered that the reason for the dearth of progenitor cells in the presenilin knockout mice is that these cells are leaving the cell cycle and differentiating into neurons. This suggests that presenilin prevents precursors from adopting a neural cell fate and would fit with Xiao and colleagues’ finding that APP processing suppresses neurogenesis in vitro. But Shen’s work also speaks to a fundamental caveat in interpreting effects on neurogenesis. Any manipulation that substantially reduces the progenitor pool in vivo could appear to be suppressing neurogenesis, even if it promoted differentiation into neurons. Precisely what effect the TAG1-APP pathway might have on neurogenesis in vivo remains to be determined.
There is also more to presenilin signaling than APP. “Gamma-secretase is involved in cleaving other proteins, including Notch. Actions via these other proteins may increase or decrease neurogenesis, and the balance of these effects may depend on the experimental system investigated,” Xiao noted in an e-mail to ARF (see ARF related news story). Xiao offered β-catenin as an example, as did Kim and Shen in their paper. Work from Eddie Koo’s lab showed that presenilin downregulates β-catenin signaling (see Soriano et al., 2001). This may be how presenilin regulates the cell cycle, suggest Kim and Shen.
Whether the TAG1-APP signaling pathway is relevant at all for adult neurogenesis remains unstudied. A new study out simultaneously proposes, however, that an entirely different and well-known risk factor for AD, namely diabetes, limits adult neurogenesis in rodents. The paper from Mark Mattson also suggests that this effect is related to an imbalance in corticosteroid biology.
First author Alexis Stranahan and colleagues examined the effects of diabetes on the adult hippocampus. They used models for both insulin-dependent diabetes (type 1, in this case streptozocin-treated rats) and non-insulin-dependent diabetes (type 2, db/db mice with inactivated leptin receptors). Though there were differences between the two models, they yielded substantially similar findings in that learning deficits in both were measurable but could be reversed by simply lowering plasma levels of corticosterone. Adrenalectomized diabetic animals supplemented with corticosterone to maintain normal levels performed just like wild-type animals in the Morris water maze, and better than sham-operated animals, which have elevated corticosterone.
The reason why lowering corticosterone rescued learning deficits in this model may be twofold. First, the authors report that restoring normal corticosterone also restored long-term potentiation (LTP) at perforant path synapses in the dentate gyrus of the hippocampus. LTP was impaired in both models but responded to normalizing corticosterone. “These findings suggest that diabetes causes a primarily postsynaptic deficit in dentate gyrus plasticity that is reversible by lowering corticosterone,” write the authors. The second effect was on adult neurogenesis. The researchers found that adult diabetic animals generated fewer new cells, both neurons and glia, in the hippocampus. Furthermore, the authors were able to differentiate LTP mediated by mature and new, adult-generated neurons by their sensitivities to picrotoxin, a GABA receptor antagonist. Under conditions that only allowed activation of newly generated neurons, the researchers found that LTP was impaired in diabetic animals but, again, restored when corticosterone levels dropped. “These results suggest that diabetes alters synaptic plasticity through multiple mechanisms involving both changes in new neurons and changes in the mature neuronal population,” they write.
Removing the adrenal gland is a drastic measure that affects a variety of circulating hormones. To confirm that corticosterone indeed is the culprit in learning deficits, Stranahan and colleagues fed diabetic adrenalectomized animals higher doses of corticosterone, and this treatment reinstated the learning and LTP deficits.
“Our findings demonstrate a pivotal role for the adrenal steroid corticosterone as a mediator of diabetes-induced impairments in hippocampal synaptic plasticity and neurogenesis, and associated cognitive deficits,” write the authors. Even so, diabetes is a highly complex disease, and effects on cognition might also be related to hyperglycemia and loss of insulin signaling or sensitivity, the authors acknowledge. It remains to be seen if this new revelation may lead to any new therapeutic approaches for diabetes-related cognitive deficits in humans. It is interesting, however, that Jonathan Seckl’s group in Edinburgh University showed that inhibiting 11-β-hydroxysteroid dehydrogenase 1 improves cognitive function in elderly diabetic men (see Sandeep et al., 2004). This enzyme is active in the brain, where it can convert inactive 11-keto metabolites of cortisol and corticosterone back into their active forms. Altogether, these recent papers suggest that links between neurogenesis and AD may be greater than expected.—Tom Fagan