20 December 2010. Whenever scientists start looking at ways to maintain memory, some of the same names keep cropping up. Cyclic AMP-response element binding protein (CREB) and brain-derived neurotrophic factor (BDNF) are two of the perennial stars of the memory game. What’s not clear, however, is how to adapt their roles into a therapy for Alzheimer’s patients. A paper in the December 13 PNAS suggests a new approach. Researchers led by Salvatore Oddo at the University of Texas in San Antonio found that plying Alzheimer’s disease model mice with CREB binding protein (CBP) rescued learning in these mice without altering Aβ or tau levels. Whether this strategy will work in people no one knows, but the new data keep the spotlight firmly planted on the CREB pathway and on the potential of viral gene delivery.
CREB is a transcription factor that affects many genes, among them the neurotrophic factor BDNF. Researchers have uncovered several ways to modulate CREB signaling and thereby improve memory, for example, by activating the CREB-regulated transcription coactivator 1 (CRTC1), or with the sirtuin SIRT1 (see ARF related news story on España et al., 2010 and Gao et al., 2010). Direct application of BDNF has also shown promise in restoring learning and memory in animal models
(e.g., see ARF related news story on Blurton-Jones et al., 2009 and ARF related news story on Nagahara et al., 2009).
To explore the alteration of the CREB pathway in AD, first author Antonella Caccamo used 3xTgAD mice at six months old, an age when they show early learning problems. She trained the mice four times per day for either three or five days in the Morris water maze, then sacrificed them within 30 minutes of the last training session and snap-froze the brains to capture the immediate transcriptional changes due to learning. Normal mice boasted twice as much phosphorylated (i.e., activated) CREB as did transgenic mice at baseline and after training, which quadrupled pCREB levels in all animals. Caccamo and colleagues used several methods to show that this difference was due to Aβ. Injection of Aβ oligomers lowered pCREB still further, while injection of antibodies to Aβ increased pCREB in the AD mice. The authors also used a genetic approach, replacing the mutant PS1 gene in the 3xTg mice with its wild-type counterpart, which has been shown to abolish Aβ accumulation. Mice tinkered with in this way showed no Aβ deposits, and higher levels of pCREB.
To treat these mice, Caccamo and colleagues focused on CBP, because it is an essential coactivator that helps recruit transcriptional machinery to CREB. Caccamo and colleagues injected a lentivirus carrying the CBP gene into the lateral ventricle, from where the virus diffused into the hippocampus. In this manner, overexpression of CBP restored CREB phosphorylation in the hippocampus and rescued learning and memory in treated animals, but did not alter Aβ or tau levels. Not surprisingly, BDNF levels were also low in 3xTg mice, but restored in the hippocampus in treated mice. The authors also found evidence that higher BDNF levels potentiated signaling through NMDA-type glutamate receptors, which can further phosphorylate CREB, possibly creating a positive feed-forward loop.
The paper “reinforces the importance of CREB signaling in models of AD,” said Michael Shelanski of Columbia University Medical Center in New York City. “It’s important because it shows a different approach to increasing the activity of CREB.” Since we don’t yet know what way will work, Shelanski said, “the more we know about this pathway, the better.”
The use of a viral system to deliver therapeutic agents is a promising approach, said Philippe Marambaud at the Albert Einstein College of Medicine in Manhasset, New York. For human therapy, using an adeno-associated virus (AAV) is more appropriate than a lentivirus, Marambaud said, because the viral integration is better characterized and it is expected to have fewer side effects. Several current clinical trials in Parkinson’s patients are using AAVs to deliver various factors (e.g., neurturin, glutamic acid decarboxylase, and human aromatic L-amino acid decarboxylase). This type of system allows the delivery of large proteins that could not normally get into the brain.
The question in AD is what factor to deliver. CBP is a candidate, Marambaud said, but there are some contradictory data on whether CBP is protective or detrimental in AD (for example see Marambaud et al., 2003). Marambaud pointed out another potential problem with CBP: It is involved in several other transcriptional pathways, including some that can cause cancer. For these reasons, it might make more sense to go downstream of CBP and simply deliver BDNF through a viral vector, Marambaud suggested. This approach has shown some success in primates (see ARF related news story).
Shelanski points out, however, that most mouse models in these studies represent early stages of AD. In these animals, restoring CREB activity reverses harmful dendritic changes. “We have no idea whether that would happen in the human. By the time we make a diagnosis of AD, it may be too late to restore the synaptic activity.” If Alzheimer’s disease could be diagnosed at much earlier stages, Shelanski said, such interventions might have much greater potential to restore memory. A variety of new imaging and fluid biomarker analyses are helping to push diagnosis earlier (e.g., see ARF related news story; ARF Live Discussion from 2008; and ARF story on the Alzheimer’s Disease Neuroimaging Initiative).—Madolyn Bowman Rogers.
Caccamo A, Maldonado MA, Bokov AF, Majumder S, Oddo S. CBP gene transfer increases BDNF levels and ameliorates learning and memory deficits in a mouse model of Alzheimer’s disease. Proc Natl Acad Sci USA. 2010 Dec 13. Abstract