25 January 2010. Called “Alzheimer’s Disease Beyond Aβ,” the Keystone Symposium held 10-15 January 2010 at Copper Mountain, Colorado, was convened to discuss therapeutic approaches to Alzheimer disease besides those that target amyloid-β (Aβ). Organized by Tony Wyss-Coray of Stanford University, California, and JoAnne McLaurin, University of Toronto, Canada, the meeting didn’t truly get away from Aβ, but it did feature novel presentations on alternative approaches to treatment.
Sheena Josselyn, Hospital for Sick Children, Toronto, Canada, raised the prospect of targeting cAMP response element binding protein, or CREB for short, to boost memory. CREB is known to be a key player in the formation of new memories, with early work led by Nobel laureate Eric Kandel, Columbia University, amongst others, showing that knocking out the protein disrupts memory. While in Alcino Silva’s lab at University of California, Los Angeles, Josselyn showed that poor memory in CREB knockout mice is not simply a result of neurodevelopment gone awry. She and colleague Satoshi Kida at Tokyo University of Agriculture, Japan, made a CREB conditional repressor mouse and showed that memory test performance suffered in adult mice even when CREB was suppressed just prior to training in a fear conditioning paradigm (see Kida et al., 2002). At the symposium, Josselyn showed how CREB might be useful to enhance adult memory as well.
Josselyn’s lab uses herpes simplex virus 1 (HSV1) to target CREB expression to specific sites in the mouse brain. In Copper Mountain, she showed that this approach can overcome reduced auditory fear conditioning (achieved by accompanying a tone with a mild foot shock) in CREB-negative mice. This works by injecting the CREB-carrying HSV1 into the amygdala, where the association of the tone and shock is thought to take place. Curiously, though, infection of only 12-15 percent of neurons in the amygdala was sufficient to restore performance in this test, leading Josselyn to wonder if a subset of neurons, specifically those expressing CREB, was adequate to establish the auditory fear memory trace. To test this idea, she looked at expression of the gene for Arc (activity-regulated cytoskeleton-associated protein), which is rapidly but transiently turned on when neurons are activated. She found that after the auditory fear conditioning, neurons that labeled positive for Arc mRNA, i.e., were active, were almost always the same neurons infected with the CREB viral vector, suggesting that neurons with higher levels of CREB preferentially participate in the memory trace.
Because the CREB/Arc connection was correlative, Josselyn tried an alternative approach to probe the link between CREB expression and memory traces. She used diphtheria toxin, which induces apoptosis, to selectively ablate CREB-overexpressing neurons and then looked to see if the memory trace still formed. Because mice do not normally express the diphtheria toxin receptor (DTR), Josselyn’s lab used mice that express a simian DTR gene that is silenced by a stop codon flanked by flox motifs. Into the amygdala of these animals, she injected a vector that carries CREB and cre recombinase genes, the latter excising the floxed stop codon and turning on DTR expression. Josselyn showed that the DTR mice carrying the cre/CREB vector learn better in the contextual fear paradigm, yet giving diphtheria toxin completely reversed this. The work, she said, suggests that cells expressing CREB are preferentially incorporated into memory traces in the brain (see Han et al., 2009).
How does this matter to Alzheimer disease? Josselyn has more recently taken up study of the hippocampus, testing the role of CREB in memory traces induced by training in the Morris water maze that is commonly used to test spatial memory in AD model mice. She showed that spatial memory in CREB-negative animals can be restored by targeting CREB expression to the CA1 regions of the dorsal hippocampus. Together with graduate student Adelaide Yiu, she tested this approach in CRND8 AD mice, which express human APP with Swedish and Indiana mutations. CRND8 mice do poorly in the water maze test and also have low levels of CREB activation in the hippocampus. Yiu found that targeting CREB expression to the hippocampus rescued spatial memory in these mice, but injecting the vector into the dentate gyrus did not. “We think that overexpressing CREB in this region might not rescue the memory deficit because the dentate gyrus likely encodes information in a very sparse manner and increasing CREB using viral vectors may interfere with this sparse encoding,” Josselyn said. The CREB effect may be downstream of Aβ, since the researchers found no difference in plaque load between treated and untreated mice. They did find that the CREB vector restored the number of dendritic spines on both apical and basal dendrites of hippocampal pyramidal neurons. Spinal loss is a key feature of these transgenic mice and probably accounts for their memory loss.
In conclusion, Josselyn noted that CREB is particularly important for memory, and that increasing it might treat memory loss in AD. But she cautioned that CREB is no magic bullet. The same CREB vector injection strategy failed to rescue memory loss associated with ablating calmodulin kinase II, which also results in poor spatial memory in mice.
One “druggable” way to boost CREB activity is to block phosphodiesterase enzymes that degrade the cyclic diester bond in cAMP. This strategy is being pursued by pharmaceutical companies (see ARF related news story). In his presentation, Menelas Pangalos, previously at Wyeth and now at Pfizer after the two companies’ merger last fall, noted that another cyclic nucleotide, cyclic GMP (cGMP), which can stimulate the cAMP/CREB pathway, is also tightly regulated at synapses and can reverse Aβ-induced impairment in long-term potentiation (see Puzzo et al., 2005). LTP is a synaptic strengthening phenomenon that supports formation of new memories. Pangalos noted that cGMP levels are elevated in mice lacking phosphodiesterase 9 (PDE9) and that those animals have a lower threshold for long-term potentiation (LTP), supporting the idea that cGMP helps boost memory. Pfizer developed a selective PDE9 inhibitor, PF 0447943, and is currently recruiting for a Phase 2 clinical trial to test the safety and efficacy of the compound in patients with mild to moderate AD (see ClinicalTrials.gov). In humans, a single acute dose of the inhibitor leads to an increase in cGMP in the CSF, peaking at six hours and then gradually declining, Pangalos reported. Preclinical studies are encouraging. In rodents the compound improves performance in a novel object recognition test, even when the animals are struggling under the influence of scopolamine, which blocks memory. And when given for 30 days via osmotic pump to four-month-old APP transgenic mice, PF 0447943 attenuates synaptic loss.
A related program, pursued initially by Wyeth, centers on selective agonists of the β form of the estrogen receptor (ERβ), which can activate CREB. Pangalos reviewed some of the data supporting the role of estrogens in protecting against AD, including that they promote survival, axon sprouting and repair, LTP, and cognition. Estrogen depletion also exacerbates Aβ pathology in mouse models of AD (see Carroll et al., 2007), suggesting that estrogens may be protective against the disease. To test this hypothesis, Wyeth developed WAY-200070, a selective estrogen receptor modulator, or SERM. Because the compound selectively activates ERβ, which is not expressed in sex organs, it might prove a safer alternative to straight hormone replacement therapy, which has been linked to life-threatening side effects such as cancer (see ARF Live Discussion).
Pangalos and colleagues previously reported that this compound can increase levels of glutamate receptor subunits and post-synaptic density 95 (PSD95), a synaptic marker, in ovariectomized (OVX) mice. The compound also enhanced LTP and improved spatial memory in OVX rats (see ARF related news story). At Copper Mountain, Pangalos reported that WAY-200070 seems to increase the interaction between GluR1 glutamate receptor subunits and stargazin, a protein that helps anchor glutamate receptors to the synapse. How the compound does this is unclear. However, Pangalos also reported that the compound may boost protein translation, since it decreases phosphorylation of eukaryotic initiation factor 2α (Eif2α), which is required to initiate translation. Blocking Eif2a dephosphorylation with the phosphatase inhibitor salubrinal abolished the increases of GluR1 and PSD95 seen when rodents are treated with the ERβ agonist, supporting the idea that WAY-200070 exerts its influence by dephosphorylation of Eif2α, suggested Pangalos. Incidentally, work from Bob Vassar’s lab at Northwestern University, Chicago, suggests that phosphorylation of Eif2α may lead to preferential translation of certain mRNAs, including that of β-secretase (see ARF related news story), which loops the story right back to...yes, the peptide that dared not speak its name at this symposium. By reducing Eif2α phosphorylation, WAY-200070 may not only improve synaptic transmission, but reduce, shall we say, amyloidogenic APP processing.—Tom Fagan.