A mouse model that mimics both signature pathologies of Alzheimer disease develops a day-to-day forgetfulness at four months, a young adult age when the animals' brains accumulate the Aβ peptide inside their neurons but don’t yet harbor plaques or tangles. What’s more, a treatment proffered for years by an Israeli scientist appears to reverse this forgetfulness and both pathologies. If these findings hold up, they will stand out from among the more than 900 AD-related presentations at the 34th annual conference of the Society for Neuroscience, held last week in San Diego, California.

The mice in question are Frank LaFerla’s triple transgenics (3xTg-AD), which carry clinical mutations in the human APP, presenilin 1, and tau genes (see Oddo et al., 2003). These mice have generated widespread attention; indeed, LaFerla’s laboratory at University of California, Irvine, has responded to continuing requests from investigators by breeding and shipping the mice to investigators from nearly 20 countries and all across the United States. At the same time, other researchers caution that the model is artificial in its own way, as no humans are known to carry mutations in all three genes. Humans also do not have such high levels of overexpression as do most other AD mouse models.

In San Diego, the LaFerla lab reported their analysis of these mice in regard to lipids, oxidative stress, calcium channels, microglial activation, inflammation, and other aspects in 12 separate presentations. This news summary will focus on three of those dealing with cognition and treatment.

Lauren Billings described her search for a molecular change that marks the earliest cognitive deficit in the mice. She used the Morris water maze, a hippocampal task, and a fear-conditioning task testing function of the amygdala. At two months, the mice learned and remembered normally, indicating that the mice were not born with cognitive impairments, and whatever deficit they developed later was progressive and related to the transgenes. An ongoing longitudinal study has reached the 12-month point to date.

At six months, the 3xTg-AD homozygous mice needed more time to learn the task at hand, but they did learn it. An intriguing clue emerged when Billings broke down the data from this group effect to a trial-by-trial basis. The mice had four training trials per day, spaced apart by 30 seconds. The wild-type controls remember what they’ve learned the day before and therefore start off at a higher performance level every subsequent day of training. The 3xTg-AD also learn from test to test on a given day but by the next day have forgotten and repeat their learning curve from the day before. Compared to previously described learning deficits in AD-related mouse strains, this early result is a seductive one because it echoes a memory problem well known to AD clinicians, where a patient will remember a standard word list for a few minutes, but an hour later has no recollection of it at all.

But is this truly the earliest deficit in this mouse strain? Probably not, the scientists reasoned, because six-month-old homozygous animals showed a deficit during the 1.5-hour time course plus one from day to day, whereas hemizygous animals at this age were defective only in the latter measure. This meant that perhaps younger homozygous animals have only this 24-hour deficit, and indeed the four-month-olds did.

At this age, the researchers can detect only intraneuronal Aβ, not yet plaques or tangles. “This is correlation. We want to be able to say cause,” LaFerla said. The scientists then injected an anti-Aβ antibody (the monoclonal 6E10, which recognizes amino acids 1-17 of human Aβ) into the mice’s third ventricle, and found that it not only clears away intraneuronal pathology from the hippocampus, but also rescues the memory retention deficit. This worked only for the water maze task, however, not for the fear-conditioning task, as the pathology was not cleared in this brain region. This finding may reflect the amygdala’s distance from the injection site, the researchers assume.

“The bottom line is that early cognitive deficits follow closely with intraneuronal Aβ,” LaFerla said. “My personal belief is that the first cognitive deficits in AD are a functional change, not structural. At the point where memory retention first declines, there is not yet loss of synapses, no plaques, no tangles, no dystrophic neurites, no inflammation. Those all come later.”

When asked about the relevance of their model, LaFerla and Salvatore Oddo noted that it did predict accurately some observations of the AN-1792 trial, namely that it failed to remove mature tangles but appeared to reduce soluble tau (unpublished, but see Ferrer et al., 2004). They say this jibes with their recently published data (see ARF related news story), which supports the hypothesis that Aβ accumulation leads to tau pathology in vivo.

Oddo’s talk in San Diego builds on his prior finding that injection of an anti-amyloid antibody clears extra- and intraneuronal Aβ. This again highlights the open question of whether intraneuronal Aβ accumulation contributes to the extracellular plaques (see ARF Live Discussion and ARF Philadelphia news story). To address it, Oddo injected antibody into the mice’s brains once and analyzed the brains not days and weeks later, as in their August paper, but 6, 12, and 18 hours after the shot. Again, he saw that extracellular Aβ disappears first, followed by intraneuronal Aβ. Hours later, the intraneuronal Aβ aggregates first return, then the extracellular ones. This implies that the two pools are connected by a dynamic equilibrium. The finding that intracellular Aβ reappears before extracellular Aβ suggests that intraneuronal accumulation may be a precursor to the plaques, said LaFerla. Other scientists were impressed by this study, but noted that they would like to see follow-up work formally rule out the possibility that the intraneuronal Aβ represents the Aβ sequence within APP.

Last but not least, these data establish a basis on which to test the prowess of potential treatments, and this was the topic of the Antonella Caccamo’s poster. Caccamo injected into the 3xTg-AD mice’s intraperitoneal cavity—every 24 hours over two months, a tiring total of 2,800 times—the compound AF267B. Aficionados of the field may recognize it by its name as one of Abraham Fisher’s. Fisher and his colleagues, at the Israel Institute for Biological Research in Ness Ziona, have synthesized and studied series of small molecules in an effort to prove Fisher’s hypothesis that agonists that are highly selective for M1 muscarinic acetylcholine receptors could treat the symptoms of AD, as well as change its course, (see, for example Fisher et al., 2003 and Fisher et al., 1998). Some muscarinic agonists have been tested for years in vitro and in humans, but none have made it all the way to a useful AD drug. Some early muscarinic agonists have failed in clinical trials. That has made Fisher’s quest seem quixotic to some, despite his insistence that the failure was due to the compound's inadequate M1 selectivity and poor pharmacokinetics. Caccamo put Fisher’s hypothesis to the test, and her data appear to vindicate him. “Everything Abe Fisher has written has come true in our study,” LaFerla said.

Caccamo presented immunocytochemistry data suggesting that AF267B diminished the mice’s plaque pathology, intraneuronal pathology, and tau pathology in cortex and hippocampus. ELISA and Western blots also indicate decreases in soluble and insoluble Aβ formation. The Western blot shows a decrease in C99 (the product of β-secretase cleavage) and an increase in C83 (the APP fragment released by α-secretase cleavage.) Measuring steady-state levels, Caccamo and colleagues found a decrease in BACE, an increase in ADAM-17, and no change in ADAM-10 (see ARF related news story). AF267B also diminished phospho-reactive tau as stained with the antibody AT8. Finally, the compound reversed the memory retention deficit in the water maze.

“This is the first in-vivo evidence for Abe’s prediction of how this compound would shift APP processing and affect tau,” said LaFerla. At the Neuroscience meeting, and also in Neurobiology of Disease this month (see Farias et al., 2004), Fisher and colleagues at the Catholic University of Chile in Santiago laid out a mechanism for how this compound might counteract Aβ toxicity. In short, they propose that AF267B, via activation of the M1 receptor, inhibits the tau kinase GSK3-β and restores a downregulation of the wnt signaling pathway caused by Aβ.

Many questions remain. One fly in the ointment is that AF267B reversed neither the fear-conditioning deficit nor AD pathology in the amygdala. LaFerla hopes that an ongoing collaboration with memory researcher James McGaugh, also at UC Irvine, will shed light on this issue. Moreover, it's not yet clear that AF267B will meet the brain penetration and pharmacological requirements to become an AD drug.

Even so, this early data proposes to show for the first time a small molecule that can cross the blood-brain barrier, is bioavailable, and reverses Aβ and tau pathology as well as some behavioral deficits at doses that cause no adverse effects in mice. A biotechnology company in California has licensed the compound.—Gabrielle Strobel.


  1. If this data holds true, it is very good news in the field. Abe Fisher has been working for years to develop highly specific muscarinic (M1) agonists. He is more advanced on that than anyone else, as the major pharmaceutical companies have abandoned that front for AD therapeutics, mostly due to low efficacy and undesirable side effects. There is a rationale for a "good" M1 agonist in AD. First, there is the clear effect of the M1 receptor-driven switch towards a non-amyloidogenic APP metabolism, i.e., stimulation of ADAM secretases. Second, there is the intrinsic cognitive effect of muscarinic agonists. Third, and less well proven, there is the possibility that the muscarinic stimulation favors endogenous production of neurotrophic factors.

  2. This data confirms our work, done in collaboration with Abraham Fisher, showing that AF267B and two other of his M1 agonists (AF102B, AF150S) all lower CSF and cortical Aβ concentrations in normal rabbits [1]. This confirms many years of in vitro work going back to 1992, when Roger Nitsch showed that M1 receptor activation shifts APP processing into the non-amyloidogenic pathway [2]. We have also demonstrated the opposite effect, in vivo, that decreasing cortical M1 receptor activation by lesioning the nucleus basalis magnocellularis (nbm) results in increased amyloidogenic processing of APP and Aβ deposition [3] and that treatment with AF267B prevents this deposition [4]. The aggregate data suggest a fusion of the cholinergic and amyloid hypotheses: cortical cholinergic deafferentation occurs during preclinical AD [5-7] and leads to Aβ deposition and AD through decreased M1 receptor activation. If this is true, then cholinergic therapy should be preventative, if given early enough. Treatment begun after dementia has been diagnosed is too late, as Aβ deposition has already reached a plateau by this stage and tangle formation has usually proceeded to at least Braak stage IV. The recent ACDS trial results showing that Aricept slows conversion of MCI to AD supports a preventative role for cholinergic therapy in AD. Both muscarinic agonists and acetylcholinesterase inhibitors should now proceed to primary prevention trials.


    . Reduction of cerebrospinal fluid amyloid beta after systemic administration of M1 muscarinic agonists. Brain Res. 2001 Jun 29;905(1-2):220-3. PubMed.

    . Muscarinic agonists as preventative therapy for Alzheimer's disease. Curr Opin Investig Drugs. 2002 Nov;3(11):1633-6. PubMed.

    . Cholinergic deafferentation of the rabbit cortex: a new animal model of Abeta deposition. Neurosci Lett. 2000 Mar 31;283(1):9-12. PubMed.

    . Immunotoxin lesion of the cholinergic nucleus basalis causes Ab deposition: towards a physiologic animal model of Alzheimer's disease. Immunol Endocr Metab Agents Med Chem. 2003;3(1):57-75.

    . Cholinergic fibre loss associated with diffuse plaques in the non-demented elderly: the preclinical stage of Alzheimer's disease?. Acta Neuropathol. 1997 Feb;93(2):146-53. PubMed.

    . The cholinergic deficit coincides with Abeta deposition at the earliest histopathologic stages of Alzheimer disease. J Neuropathol Exp Neurol. 2000 Apr;59(4):308-13. PubMed.

    . Clinical, pathological, and neurochemical changes in dementia: a subgroup with preserved mental status and numerous neocortical plaques. Ann Neurol. 1988 Feb;23(2):138-44. PubMed.

  3. The triple transgenic animals are a fascinating addition to the tools available. Especially exciting is this new data indicating that these genes have to work together in order to convert mice to an AD model. Obviously, these mice remain an animal model of AD, but are likely a big leap forward from the standard amyloid model mice we used to work with. There might be other animal models that do the same without the need for mutations in three different genes, but to be able to study this in mice will speed up the necessary research enormously.

    Intracellular Aβ accumulation sheds light on what might be expected. Intracellular Aβ accumulations had been found in several studies previously, including human brain. However, in transgenic mice this does not appear to be a consistent feature. For a “perfect” model of AD, that’s just strange. If it exists in human AD brains, it should be present in all transgenic mouse models. The conclusion that comes to mind is that the triple transgenic mice enhance important aspects of the pathology that would otherwise be easily missed. Do we then have to conclude that tau plays a role in intracellular Aβ toxicity? This may very much be the case as intracellular Aβ has all the properties to interfere with intracellular trafficking, including the role of the Aβ domain in APP as essential mediator of axonal APP transport.

    The other intriguing item is Abraham Fisher's M1 agonist that interferes in a complex manner with APP processing and Aβ production. Again, previous transgenic approaches might simply not have been able to model AD closely enough to reveal the strong effect AF267B has, as these new data indicate. The more we try to shape and optimize approaches to find a cure, the more complex the models will have to be to gain the necessary insights. If that needs to be done with three mutated genes, it's fine with me.

  4. I think this is among the most important observations shown at the meeting. I had seen Frank LaFerla’s behavior data and antibody reversal earlier in September at a meeting on cognition. It supports and is consistent with the results from most of the mice that cognitive function and Aβ correlate. We have not been able to detect the intracellular Aβ in our Tg2576-based mice, but that may be a technical difference. In any event, it seems very likely that Aβ can cause the memory deficits in the APP-transgenic animals.

    This is the first time for the M1 data in vivo to my knowledge. There is a long history of muscarinic cholinergic regulation of APP processing (see Nitsch et al., 1992 or Buxbaum et al., 1992), in addition to the mechanism suggested by Abe. The one control, however, that I know Frank will run if he hasn't already is to evaluate the levels of the transgene mRNAs. Because the transgenes are driven by an autologous promoter (Thy-1, in this case), changes in gene expression might be trivial due to regulation of Thy-1 expression (which is presumably irrelevant for AD).

    I emphasize these are exciting data and I can’t wait to see how they translate into the clinic.


    . Release of Alzheimer amyloid precursor derivatives stimulated by activation of muscarinic acetylcholine receptors. Science. 1992 Oct 9;258(5080):304-7. PubMed.

    . Cholinergic agonists and interleukin 1 regulate processing and secretion of the Alzheimer beta/A4 amyloid protein precursor. Proc Natl Acad Sci U S A. 1992 Nov 1;89(21):10075-8. PubMed.

  5. The ability of the M1 agonist AF267B to decrease amyloid plaque load, decrease tau phosphorylation, and enhance memory function in the "triple transgenic" mice is indeed encouraging. Selective muscarinic agonists are among the few therapeutic approaches that could help alleviate the symptoms (memory deficits, cognitive dysfunction) of Alzheimer disease and have an impact on the underlying disease process. Over the years, Dr. Fisher has been a strong proponent of using selective muscarinic agonists to treat Alzheimer disease. Although several muscarinic agonists have failed in clinical studies, most of the compounds tested lacked selectivity for M1 receptors or appreciable activity at M1 receptors in the CNS.

    We also presented data at the 2004 Society for Neuroscience meeting (1) on the potential neuroprotective effects of a selective M1 agonist CDD-0102. In the studies presented in San Diego, CDD-0102 promoted activation of α-secretase, (as measured by elevated levels of soluble APP-α) and decreased levels of Aβ in HEK 293T cells expressing human M1 receptors, wild-type APP695 and mutant PS1. In PC12 cells treated with NGF to promote differentiation into a neuronal phenotype, CDD-0102 protected cells from apoptosis and elevated caspase-3 cleavage induced by staurosporine.

    CDD-0102 is functionally selective for M1 receptors with minimal activity at other muscarinic receptor subtypes (2,3). It also enhances memory function in rats treated with 192IgG saporin to deplete cortical and hippocampal acetylcholine (4). Further studies are necessary to determine whether CDD-0102 produces similar effects to those observed in transgenic mice for AF267B.

    See also:

    Messer WS, Jr, Tang B, Hoss WP, and Ghosh D. Neuroprotective effects of the selective M1 agonist CDD-0102: Stimulation of α-secretase, inhibition of Aβ and prevention of apoptosis. Society for Neuroscience, 29, no. 673.19, 2004.


    . Synthesis and biological characterization of 1,4,5,6-tetrahydropyrimidine and 2-amino-3,4,5,6-tetrahydropyridine derivatives as selective m1 agonists. J Med Chem. 1997 Apr 11;40(8):1230-46. PubMed.

    . Tetrahydropyrimidine derivatives display functional selectivity for M1 muscarinic receptors in brain. Drug Dev Res. 1997 Feb;40(2):171-84.

    . Development of CDD-0102 as a selective M1 agonist for the treatment of Alzheimer’s disease. Drug Dev Res. 2002 Dec;57(4):200-13.

  6. I concur with the other comments: The in-vivo M1 agonist treatment effect on Aß is very encouraging. It has long been clear that different muscarinic receptor subtypes have opposing actions on amyloidogenesis and other physiological processes, and that selective M1 agonists may provide a major step forward from current nonselective cholinergic therapies. M1 is the predominant muscarinic receptor involved in cognition, neuronal excitability, synaptic plasticity and likely regulation of amyloidogenesis.

    However, the hypothesis has never been adequately tested since highly selective and potent M1 agonists have been so difficult to develop. Hopefully, Dr. Fisher's persistence in developing M1 agonists will pay off and add a significantly improved therapeutic approach that targets cognition, behavior, and amyloidogenesis. The unanticipated benefits of cholinergic therapies on behavorial problems in AD, including psychosis, have also renewed the interests of big pharma in developing M1 agonists given their potential for schizophrenia (and pain). Hence, we may finally see some progress in the pharmacology.

  7. Given the recent interest in the M1 agonist formerly designated AF267B, we thought it would be useful to provide information concerning the basic pharmacologic properties and future development plans for this compound at Neurogenetics, Inc. (licensee) in La Jolla, CA.

    Currently designated as NGX267B, this compound is an orally active, rigid analog of acetylcholine. Its pharmacological properties partially mimic the actions of acetylcholine through a stimulation of neurons that are generally spared in the neurodegenerative processes characterizing Alzheimer disease (AD). Consistent with this hypothesis, animal models with predictive utility for the symptomatic treatment of AD have demonstrated efficacy at dosages of NGX267B below those associated with nonspecific effects. NGX267B has also demonstrated disease modification properties involving reduction of β-amyloid and tau deposition in the LaFerla triple transgenic mice, thereby offering insights into mechanisms with long-term clinical implications.

    One mechanism of action for NGX267B involves direct stimulation of specific muscarinic (M1) receptors that exist on intact cholinergic neurons within the CNS; therefore, this proposed therapy could complement currently available AD treatments (e.g., acetylcholine esterase inhibitors). It is specific to memory and cognitive behaviors, and largely devoid of adverse effects such as dizziness, salivation, nausea, vomiting, or diarrhea at clinically relevant doses. We consider both monotherapy and concomitant use with existing products plausible at this juncture.

    The known biological and pharmaceutical properties of NGX267B should permit an exploration of efficacy and safety across a range of disease severity, co-morbidities, and concomitant medications. For example, no clinically important adverse events have been detected in animals at dosages that enhanced memory and cognitive behaviors; this suggests that the therapeutic index (a measure of the safety margin) may be attractive for the elderly or debilitated patient. The major metabolite of NGX267B has a biological profile similar to that of the parent compound. This makes it less likely that we will experience the disappointing clinical results that occurred in the evaluation of other agents as a result of poor oral absorption or extensive metabolic breakdown into products with less specificity. Finally, an anticipated long duration of action may enhance both efficacy and convenience.

    The clinical development program for NGX267B is scheduled to begin approximately mid- to late 2005. Although a variety of cognitive disorders are potential therapeutic targets based upon known pathophysiology, Neurogenetics has identified the symptomatic treatment of mild to moderate Alzheimer disease as an immediate therapeutic goal. The first few clinical studies will explore safety, tolerance, pharmacokinetics, and biodisposition in both asymptomatic young and older volunteers in order to define the permissible clinical dose range for subsequent "proof-of-concept" studies in AD. The influence of dosing frequency on cognitive performance will be evaluated in light of the known neuropsychological effects of cognitive enhancing agents. Additionally, clinical measures such as neuroimaging, computerized neuropsychological test batteries, and cerebrospinal fluid biomarkers may be added as outcome measures to help fully characterize the biological properties of NGX267B before we begin a more traditional dose-ranging paradigm in mild to moderate AD.

  8. I concur that the work with the triple transgenic mouse and also the possibility of M1 agonists as therapy are exciting, but I specifically want to comment on a technical issue that is being brought up regarding intraneuronal Aβ, something that the triple transgenic mouse is providing intriguing new insights into. There is no evidence, to my knowledge, that AD mutant mice exist that develop plaques but never show intraneuronal Aβ. One comment mentioned not observing intraneuronal Aβ in a Tg2576-based mouse. I understand the difficulty with convincingly detecting intracellular Aβ. I use, as an analogy, doing a Western blot of brain extract with an anti-Aβ antibody. If you do a short exposure, you can see a faint band for full-length APP but no Aβ band. If you stop there, you can convince yourself that there is no Aβ in brain. But if you expose your gel longer, the APP band will become more pronounced while APP CTFs and Aβ also eventually appear. Similarly with Aβ42 immunohistochemistry, if you do a brief reaction time, you can have a clean image only of plaques. But if you wait longer, you can start to see intraneuronal Aβ42, especially if there are not too many plaques around. If you think this is all non-specific background, consider performing the optimal control, comparing staining in equally aged/processed APP knockout mouse sections. Or consider using pre-embedding immuno-gold electron microscopy (thereby avoiding a reaction product and endogenous peroxidase), since in our experience intraneuronal Aβ42 especially accumulates in AD transgenic mice in processes/synapses, which cannot easily be seen by light microscopy. If you have difficulty with seeing intraneuronal Aβ42, please feel free to contact me.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Tackling Alzheimer’s from the Outside In
  2. Philadelphia: The Enemy Within—Neurodegeneration From Intraneuronal Aβ
  3. α-Secretase Returns to Center Stage

Paper Citations

  1. . Triple-transgenic model of Alzheimer's disease with plaques and tangles: intracellular Abeta and synaptic dysfunction. Neuron. 2003 Jul 31;39(3):409-21. PubMed.
  2. . Neuropathology and pathogenesis of encephalitis following amyloid-beta immunization in Alzheimer's disease. Brain Pathol. 2004 Jan;14(1):11-20. PubMed.
  3. . M1 muscarinic agonists can modulate some of the hallmarks in Alzheimer's disease: implications in future therapy. J Mol Neurosci. 2003;20(3):349-56. PubMed.
  4. . M1 muscarinic agonist treatment reverses cognitive and cholinergic impairments of apolipoprotein E-deficient mice. J Neurochem. 1998 May;70(5):1991-7. PubMed.
  5. . M1 muscarinic receptor activation protects neurons from beta-amyloid toxicity. A role for Wnt signaling pathway. Neurobiol Dis. 2004 Nov;17(2):337-48. PubMed.

Other Citations

  1. monoclonal 6E10

Further Reading


  1. . Neuropathology and pathogenesis of encephalitis following amyloid-beta immunization in Alzheimer's disease. Brain Pathol. 2004 Jan;14(1):11-20. PubMed.
  2. . M1 muscarinic agonists can modulate some of the hallmarks in Alzheimer's disease: implications in future therapy. J Mol Neurosci. 2003;20(3):349-56. PubMed.
  3. . M1 muscarinic agonist treatment reverses cognitive and cholinergic impairments of apolipoprotein E-deficient mice. J Neurochem. 1998 May;70(5):1991-7. PubMed.
  4. . M1 muscarinic receptor activation protects neurons from beta-amyloid toxicity. A role for Wnt signaling pathway. Neurobiol Dis. 2004 Nov;17(2):337-48. PubMed.
  5. . Treatment with the selective muscarinic m1 agonist talsaclidine decreases cerebrospinal fluid levels of A beta 42 in patients with Alzheimer's disease. Amyloid. 2003 Mar;10(1):1-6. PubMed.