. Genetic suppression of transgenic APP rescues Hypersynchronous network activity in a mouse model of Alzeimer's disease. J Neurosci. 2014 Mar 12;34(11):3826-40. PubMed.

Recommends

Please login to recommend the paper.

Comments

Make a Comment

To make a comment you must login or register.

Comments on this content

  1. Born and colleagues present an elegant study demonstrating that juvenile overexpression of APP contributes to sharp wave discharges in EEG using a tetracycline-responsive APP transgenic mouse model with a humanized Aβ domain encoding both the Swedish and Indiana familial mutations. Aβ has been considered the likely seizure-promoting catabolite of APP, but the authors’ data demonstrate that Aβ overproduction is not responsible for observed EEG abnormalities. This study convincingly shows that Aβ is not the primary cause of epileptiform discharges in APP transgenic models.

    A caveat to these findings is that very few constructs used to generate APP transgenic mice contain the DNA sequences that code for regulatory elements in the 5’ and 3’ untranslated regions, which are involved in mRNA localization and translation. Thus, transgenic APP is likely not expressed in the same temporal or spatial pattern as endogenous APP. The authors controlled temporal expression in their tet-APP model, and future experiments that examined spatial expression would be quite interesting. For example, does Aβ contribute to seizure activity when APP is synthesized and processed in specific cell types or locally at synapses? In our work with a fragile X mouse model, we’ve found that genetic manipulation of the mice to either over- or under-express APP both resulted in increased seizures, suggesting that a balance of APP catabolites may be important to suppress seizure activity.

    This paper raises important questions from a therapeutic perspective as drug design to date has focused on reducing Aβ in Alzheimer’s disease. If intact APP is responsible for network hyperexcitability, then drugs that modulate synthesis instead of processing may be more advantageous.

  2. This study is quite comprehensive and shows quite nicely that APP, and not Aβ, is responsible for the epileptiform activity in APP-overexpressing mice. This has always been a bit of a conundrum for me, since our work shows that Aβ reduces excitatory transmission (see Kamenetz et al., 2003), which would most likely reduce epileptiform activity. This still raises the question regarding human AD patients, who generally don’t have increased APP, and why they would have a higher incidence of epilepsy.

    References:

    . APP processing and synaptic function. Neuron. 2003 Mar 27;37(6):925-37. PubMed.

  3. Unexplained mortality at a young adult age has plagued APP-transgenic mouse lines for as long as they have been bred. Accumulating reports have confirmed that several different APP-overexpressing mice, independent of the mutation or the promoter used, display frequent epileptiform EEG activity and spontaneous epileptic seizures with or without convulsions. The elegant study by Born et al. now sheds new light on the molecular underpinnings of hyperexcitability in APP-transgenic mice.

    The authors used tetracycline-responsive APP-transgenic mice (APP/TTA) expressing the Swedish and Indiana APP mutations. These transgenics were crossed with mice expressing tetracycline transactivator, allowing the researchers to turn the transgene off by doxycyclin (DOX) treatment. The EEG of APP/TTA mice is characterized by frequent sharp-wave discharges (SWD), while only occasional SWDs could be recorded in TTA-only or nontrangenic control mice. Turning off the transgene reduces SWD almost to the level of control mice, while a parallel microdialysis study confirmed that soluble Aβ levels were reduced by half and remained that low as long as DOX was administered. Intriguingly, however, it took five weeks to reach the full effect on SWDs, while the maximum reduction in APP levels (80 percent) and Aβ levels (50 percent) was observable after only two days. In contrast, chronic treatment with a γ-secretase inhibitor that reduced Aβ levels by more than 50 percent did not decrease the number of SWDs. This combination of findings strongly suggests that overexpression of humanized APP rather than Aβ is the primary cause for SWDs in the APP/TTA mouse model and plausibly in other human APP overexpressing mouse lines as well. 

    To provide further evidence for the primary role of APP overexpression as a cause for SWDs, the authors performed EEG-recordings in homozygous APPSwe/Lon/PS1M146V double knock-in mice that produce Aβ deposits at an old age but have no APP overexpression. These mice showed no increase in SWDs compared to their littermate controls. Unfortunately, these mice were recorded at the age of 23-24 months, whereas other studied mice were 8-9 months of age. This age difference is an important confound, since it is possible that aging-related degenerative changes make the brain less excitable. We have demonstrated by in-vivo microdialysis, for instance, a substantial decrease in synaptic glutamate release in C57Bl/6J mice (with or without APP overexpression) between 7 and 17 months of age (Minkeviciene et al., 2008), which could substantially reduce brain excitability at an old age. 

    It is obvious from the data that the APP does not directly provoke SWDs. First, the occurrence of SWDs was the highest during the animals’ light period when they are less active, whereas APP is released in an activity-dependent manner and shows the highest levels during the dark period (Cirrito et al., 2005). Second, the full effect of turning off the APP transgene on SWDs took five weeks.  Third, suppression of the APP transgene during early development had a suppressive effect on SWDs for more than six months. Finally, APP/TTA mice had robust changes in the cortical vGLUT and vGAT neurotransmitter transporter levels that were normalized by five-week suppression of the transgene. In addition, transgenic APP overexpression also associates with EEG abnormalities, indicative of the altered state of the network. Unfortunately, the authors failed to relate the EEG pattern with the movement data, although that was available. The high theta peak (7-8 HZ) in the power spectrum of APP/TTA mice compared to TTA mice most likely reflects the fact that APP/TTA mice were moving more that TTA and thus have a higher contribution of movement-related theta in the averaged power spectrum. Since turning off the APP transgene also reduced locomotor activity, this provides a parsimonious explanation for the “EEG-normalization” in this case.

    As any excellent paper, this one raises as many questions as it answers. The most puzzling finding at first glance was that despite a dramatic effect on spontaneous SWDs, suppression of the APP transgene did not affect the mice susceptibility to picrotoxin-provoked seizures. However, there is an important distinction between SWDs and seizures, which should have been made explicit in the paper. According to our experience, based on two- to three-week continuous video-EEG monitoring on about 50 APP/PS1 transgenic mice, there is no correlation between the occurrence of SWDs and seizures in individual animals. This suggests two different underlying mechanisms.  One obvious hypothesis to be tested is that APP overexpression per se induces network changes associated with SWDs, whereas the actual susceptibility for seizures may be due to some form of soluble Aβ oligomer/protofibril.

    What might be the clinical relevance of the network hyperexcitability (SWDs) observed in APP-transgenic mice? In contrast to what the Born et al. paper implies, this is highly unlikely to account for the high seizure incidence in early onset AD, since the overwhelming majority of these cases represent presenilin mutations or APP mutations without APP overexpression. Similarly, there is no evidence for APP overexpression in sporadic AD. Thus the cause for increased seizure incidence in these cases must be Aβ or some other cleavage product of APP. Nevertheless, there are two groups of human patients with increased incidence of epilepsy where APP overexpression is the most likely explanation. As Born et al. indicated in their introduction, these include patients with rare familial APP duplication and Down’s syndrome patients with dementia. 

    References:

    . Age-related decrease in stimulated glutamate release and vesicular glutamate transporters in APP/PS1 transgenic and wild-type mice. J Neurochem. 2008 May;105(3):584-94. PubMed.

    . Synaptic activity regulates interstitial fluid amyloid-beta levels in vivo. Neuron. 2005 Dec 22;48(6):913-22. PubMed.

  4. The paper by Born et al. describes an impressive body of work. We congratulate the authors on their study, which conclusively demonstrates that network hypersynchrony in APP/TTA transgenic mice is (1) reversible, (2) closely associated with behavioral abnormalities, and (3) not dependent on overexpressing APP/Aβ during early development.

    The authors also concluded that Aβ is not the main cause of epileptic activity in this line of APP transgenic mice and emphasized that their interpretation contradicts conclusions we have drawn from our previous studies in this field. Notably, in the most pertinent review they cited (Palop and Mucke, 2010), we summarized our view as follows: “Although much evidence supports a causal role of Aβ in the pathogenesis of Alzheimer’s disease, many other factors—other APP metabolites, tau, apoE4, α-synuclein, vascular alterations, glial responses, inflammation, oxidative stress, epigenetic determinants and environmental factors—may all have important co-pathogenic roles, especially in the most common forms of sporadic Alzheimer’s disease.” Thus, Born and colleagues’ conclusion would, in principle, not be quite as much at odds with our own view as their discussion seemed to suggest. For the following reasons, though, we think that they may have been too hasty in dismissing Aβ as a potential cause of epileptogenesis, even in the model they analyzed. 

    Separating out the contributions of APP from those of its metabolites in the generation of epileptic activity (or other functional abnormalities) is indeed very challenging. The main reason is that most manipulations of this system do not specifically isolate one species from the other. For instance, as expected, the treatment of mice with a γ-secretase inhibitor (GSI) in the Born et al. study greatly increased brain levels of APP C-terminal fragments (CTFs), some of which are known to cause epilepsy when overexpressed in transgenic mice (Vogt et al., 2011). Consequently, the GSI treatment may have simply replaced one cause of epilepsy with another. It therefore remains possible that Aβ accumulation is the main cause of epilepsy in untreated APP/TTA mice, whereas CTF accumulation is the main cause of epilepsy in GSI-treated APP/TTA mice. Because the GSI used by Born et al. (LY411575) is not selective for APP, the dysregulation of myriad other γ-secretase substrates may have sensitized the brain of APP/TTA mice to the epileptogenic effects of CTFs. 

    Another important caveat to consider in this study is the way in which Aβ was measured. The Aβ measurements Born et al. performed after each of the two main interventions were not strictly matched and thus are difficult to compare. After downregulating transgene expression, they measured Aβ levels in brain interstitial fluid, whereas after GSI treatment, they measured Aβ levels in brain homogenates. It is likely that the predominant Aβ species detected by these approaches are at least partly distinct. The paper also did not include a direct side-by-side comparison of Aβ levels in age-matched APP/TTA transgenic mice and APP/PS1 knockin mice. It therefore remains possible that the lack of epileptic activity in the latter model simply reflects relatively lower levels of soluble Aβ assemblies. Finally, it is possible that the two main interventions used (downregulation of transgene expression versus GSI treatment) may have differentially affected epileptogenic pools of Aβ oligomers. Aβ oligomers have emerged as the likeliest mediators of Aβ-induced neuronal dysfunction, but were not measured in this study. 

    In light of these considerations, we do not think that the study by Born et al. conclusively refutes the hypothesis that pathologically elevated levels of Aβ can cause epileptic activity. Consequently, this hypothesis deserves to be further tested in additional studies. Notwithstanding this conclusion, we fully embrace the notion that neuronal excitability may be affected not only by Aβ, but also by alternate APP metabolites, APP itself, and other AD-related factors such as tau and α-synuclein. 

    References:

    . Amyloid-beta-induced neuronal dysfunction in Alzheimer's disease: from synapses toward neural networks. Nat Neurosci. 2010 Jul;13(7):812-8. PubMed.

    . Abnormal neuronal networks and seizure susceptibility in mice overexpressing the APP intracellular domain. Neurobiol Aging. 2011 Sep;32(9):1725-9. PubMed.

  5. We are grateful to Drs. Westmark, Malinow, and Tanila for their positive comments on our study. We also appreciate the thoughtful letter from Drs. Mucke, Palop, Sanchez, and Johnson, and want to respond specifically to their comments, as they raise several important caveats to consider in the interpretation of data presented in our study. Each of these groups has made seminal contributions to our understanding of the relationship between Aβ and neuronal excitability and we value the insight they’ve shared here.

    We agree that there are limitations to the use of commercially available γ-secretase inhibitors (GSIs) as a means of selectively lowering Aβ production.  As Dr. Mucke and co-authors point out, any reduction in Aβ evoked by this treatment is accompanied by a corresponding increase in the corresponding C-terminal fragment (CTF).  Indeed, as they indicate, past work suggests that this fragment may have its own impact on neuronal activity (Vogt et al., 2011).   Further, because the GSI we used is not selective for amyloid-b precursor protein (APP), a change in any number of other γ-secretase substrates could have helped maintain network hypersynchrony in place of Aβ.  However, in this case, we might have seen the same unprocessed proteins raise the rate of sharp wave discharge (SWD) in GSI-treated control mice.  Either mechanism would explain our findings and cannot be ruled out with the experiments we’ve done.  

    We also appreciate Dr. Mucke’s apprehension about our comparison of Aβ reduction attained by GSI treatment and by transgene suppression, because these measurements were made by different methods that evaluate distinct pools of soluble peptide.  We conceded this point in our response to the paper’s referees, but neglected to make it explicit in our published discussion.  The comparison was meant to show that the two methods of lowering Aβ attain reductions that are within the same ballpark, making it less likely that a threshold effect explains the persistence of sharp wave discharge during GSI treatment.  If there is a precise cut-off in the level of soluble Aβ required to elicit this EEG phenotype, it must sit between the residual amount of Aβ overproduction left by GSI treatment and that left by transgene suppression.  

    We believe that there are several important experiments yet to be done that will help to illuminate the role of Aβ and APP in epileptiform activity.  First, we hope that future studies will test the effect of GSI in other APP transgenic lines where SWD is observed (i.e., J20, or APP/PS1) to determine whether our findings extend to other model systems.  Second, we are curious if the EEG phenotype is present in lines that don’t include the Swedish mutation and in which a BACE inhibitor could be tested as a complement to GSI studies.  We pursued this approach using the BACE inhibitor LY2811376 in our APP/TTA mice, but found that it would require doses in excess of 1 g/kg to attain the same reduction in  as the GSI LY411575.   Alternatively, the contribution of APP in the complete absence of exogenous Aβ might be examined in mice harboring the M671V mutation that prevents BACE cleavage, such as the tetO-hAPPmv line from Mark Albers’ group (Cao et al., 2012). Conversely, it will be important to re-examine whether Aβ overproduction is sufficient to elicit an EEG phenotype in the absence of APP overexpression using models that yield higher levels of Aβ than the APP/PS1 knockin line we tested. Potential experimental systems for this study are the BRI-40/42 mice from Todd Golde (McGowan et al., 2005), although this model may be complicated by the non-native route of Aβ production, or the complex knockin mouse from Takaomi Saido (not yet published).  We understand that some of these experiments are already underway, and look forward to their outcomes as new findings continue to move the Aβ discussion forward.   

    References:

    . Abnormal neuronal networks and seizure susceptibility in mice overexpressing the APP intracellular domain. Neurobiol Aging. 2011 Sep;32(9):1725-9. PubMed.

    . Aβ alters the connectivity of olfactory neurons in the absence of amyloid plaques in vivo. Nat Commun. 2012;3:1009. PubMed.

    . Abeta42 is essential for parenchymal and vascular amyloid deposition in mice. Neuron. 2005 Jul 21;47(2):191-9. PubMed.

This paper appears in the following:

News

  1. Could APP, Not Aβ, Explain Epileptic Activity in Alzheimer’s?