Traffic Control: BDNF Boosts SORLA, Reroutes APP
The sortilin-related receptor protein SORLA (aka SORL1) controls the trafficking and processing of the amyloid precursor protein (APP), and thus serves as a key regulator of amyloid-β (Aβ) production in the brain. But what regulates SORLA? A new report from Thomas Willnow of the Max Delbruck Center for Molecular Medicine, Berlin, Germany, and Anders Nykjaer, Aarhus University, Denmark, provides an intriguing answer to this question. In work published in the December 9 Journal of Neuroscience, Willnow, Nykjaer and colleagues show that brain-derived neurotrophic factor (BDNF) induces SORLA expression in mouse neurons in vitro and in vivo, which leads to a reduction in APP processing to Aβ. Both BDNF and SORLA are lacking in the brain in Alzheimer disease, and the study links the two, raising the possibility that some of the protective effects reported for BDNF in AD might stem from diminished Aβ production.
Willnow and Nykjaer had previously discovered that SORLA regulates the travels and processing of the amyloid precursor protein (APP), steering it away from cell compartments that favor amyloidogenic processing and thus reducing Aβ production (Andersen et al., 2005). Consistent with this role, knocking out SORLA worsens pathology in an AD mouse model (see ARF related news story on Dodson et al., 2008). Human studies implicated SORLA in AD. Expression of the receptor is reduced in brain tissue from AD patients (Scherzer et al., 2004; Dodson et al., 2006), and SORLA gene variants are associated with the risk of late-onset AD (see AlzGene entry for SORL1).
To explore possible regulators of SORLA, first author Michael Rohe took a screening approach, testing nine different trophic factors for their ability to elevate SORLA mRNA in newborn mouse primary cortical neurons in culture. The best inducer by far was BDNF, which caused a seven- to 10-fold elevation of SORLA mRNA and a 60 percent increase in protein levels within 48 hours of treatment. In vivo, too, BDNF appeared to be a major regulator of SORLA levels. Brain extracts from newborn BDNF knockout mice contained 40 percent less SORLA protein than did their normal littermates. Also, SORLA levels were down in a mouse model of Huntington disease, where loss of BDNF is part of the phenotype.
Given that SORLA restrains the amyloidogenic processing of APP, the researchers asked whether BDNF had any effect on Aβ production. They found that adding the growth factor to primary neurons slashed Aβ40 production by 40 percent. This depended on SORLA induction, as BDNF had no impact on neurons from SORLA knockout mice, despite the normal activation of the BDNF receptor and downstream signaling pathways. BDNF increased SORLA expression and reduced Aβ40 and 42 production in neurons from PDAPP mice, an AD model that overexpresses a mutant human APP gene. In vivo, intracranial injection of the factor into the hippocampi of mice caused a 40 percent drop in Aβ40 levels. Consistent with the in vitro work, no change in Aβ production was seen under the same treatment regimen in SORLA-deficient mice.
Finally, Willnow and colleagues looked at the signaling pathways involved in SORLA induction by BDNF. Processing of SORLA by γ-secretase has been shown to downregulate SORLA expression, but the German scientists found that treating cells with γ-secretase inhibitors did not interfere with the effects of BDNF. On the other hand, inhibiting the MEK/ERK pathway could prevent SORLA induction.
BDNF replacement has been proposed as a potential treatment for AD, based on its downregulation in AD brain and recent studies showing that chronic delivery of the factor either by lentiviral expression (see ARF related news story on Nagahara et al., 2009) or via stem cells (see ARF related news story on Blurton-Jones et al., 2009) improves synaptic density and cognition in two mouse models of AD. However, in neither study was there any reduction of plaque load in response to treatment, and the authors attribute the effects of BDNF to its better-known trophic actions.
How to reconcile the disparate results? The studies are difficult to compare with the new study, Willnow wrote in an e-mail to ARF, because models, treatment regimens, and endpoints measured were quite different. “In our work, we aimed at testing the acute effects of high doses of BDNF application in the hippocampus and in primary neurons on upregulation of SORLA gene expression and local Aβ formation. Therefore, we determined the amount of soluble Aβ formed acutely during seven days of BDNF infusion into the hippocampus. This experimental condition will not clarify whether chronic application of BDNF will also induce SORLA gene expression and reduce long-term Aβ formation and senile plaque deposition,” he noted.
“There is no doubt that BDNF has many pleiotropic effects on neuronal function and cognitive performance, some likely dependent, others independent of APP metabolism,” Willnow concluded. For SORLA aficionados, the work opens up a new set of interesting questions about whether and in what way their favorite protein may support the trophic actions of BDNF in neurons.—Pat McCaffrey
University of California, Irvine
This paper provides solid support for the idea that BDNF can regulate SORLA expression via ERK activation. It presents interesting findings showing that BDNF, acting via SORLA, can decrease Aβ generation in wild-type mice and primary neurons.
This is a tantalizing finding, as previous studies, including our own, did not see altered Aβ in aged 3xTg-AD mice following neural stem cell treatment, despite the fact that the NSCs produce and elevate levels of BDNF (Blurton-Jones et al., 2009). Another group led by Dr. Mark Tuszynski also did not observe any changes in Aβ in the J20 mouse model following viral BDNF delivery.
There are a couple of likely explanations for the differences between the effects of BDNF on Aβ generation in the study by Rohe et. al. and our own data:
Firstly, the concentration of BDNF used by Rohe et al. in vivo (40 ug/hippocampus) is substantially higher than the elevation of BDNF we see in the brain following NSC delivery. We find by ELISA that brain levels of BDNF increase from about 10.5 pg/mg of tissue to 15 pg/mg. That translates to an increase in total brain BDNF from 4.2 ng up to about 6 ng. Thus, the supraphysiological levels of BDNF used in vivo by Rohe et al. may complicate the interpretation of these results. It would be very interesting to know if the converse is true; that is, do mouse Aβ levels decrease in the Huntington model or BDNF knockout mice that they utilized? We think this would more clearly address the physiological effects of BDNF on APP metabolism.
Another important difference between our findings and the current study is that Rohe et. al. examined the effects of BDNF in wild-type mice. Our study utilized the 3xTg-AD mice, and Dr. Tuszynski's study utilized the J20 line. Both of these transgenic models harbor the Swedish mutation that enhances β-secretase cleavage of APP. Thus, the effects of the Swedish mutation on APP processing might override any influence of BDNF and SORLA that might have driven non-amyloidogenic processing of APP in our study. This suggests that it may be very interesting and important to perform these kinds of experiments in mice that express wild-type human APP.
Overall, this study adds intriguing information about the possible connections among BDNF, SORLA, and AD. Although we would respectfully argue that studies that examine the effects of more physiologic reduction or elevation of BDNF would help to more precisely define the relationship between BDNF and APP processing in vivo.
Blurton-Jones M, Kitazawa M, Martinez-Coria H, Castello NA, Müller FJ, Loring JF, Yamasaki TR, Poon WW, Green KN, Laferla FM. Neural stem cells improve cognition via BDNF in a transgenic model of Alzheimer disease. Proc Natl Acad Sci U S A. 2009 Aug 11;106(32):13594-9. PubMed.
The finding that BDNF reduces Aβ production by regulating the expression of SORLA is a potential link between degeneration of the locus coeruleus (LC), the main source of the neurotransmitter norepinephrine in the limbic system and forebrain, and the development of AD neuropathology. Although it is well established that LC neurons degenerate early in AD, the functional consequences are not well understood. In general, the LC appears to protect against Aβ neuropathology. For example, lesions of the LC enhance Aβ plaque formation in transgenic mice that overexpress mutant APP, a commonly used animal model of AD (Heneka et al., 2006).
Intriguingly, LC neurons express and release BDNF, and NE itself can promote BDNF expression in target neurons; thus, when LC neurons degenerate early during the early stages of AD, this source of BDNF is lost or greatly reduced. The newly described ability of BDNF to increase SORLA suggests that one consequence of LC degeneration could be a decrease in SORLA expression, leading to dysregulated sorting of Aβ and again resulting in greater amyloid plaque deposition in the LC denervated cortical and hippocampal target sites.
Heneka MT, Ramanathan M, Jacobs AH, Dumitrescu-Ozimek L, Bilkei-Gorzo A, Debeir T, Sastre M, Galldiks N, Zimmer A, Hoehn M, Heiss WD, Klockgether T, Staufenbiel M. Locus ceruleus degeneration promotes Alzheimer pathogenesis in amyloid precursor protein 23 transgenic mice. J Neurosci. 2006 Feb 1;26(5):1343-54. PubMed.