Opposing Roles of apolipoprotein E in aging and neurodegeneration

This study investigates how APOE modulates neuronal function integrity during normal aging and in the context of amyloidosis. This work demonstrates that APOE is a necessary partner of Aβ-dependent neuronal dysfunction and synaptotoxicity but also preserves neuronal network during aging.

rescue of neuronal dysfunction in the context of AD Tg mice or leads to neuronal dysfunction in aged APOE-null mice in absence of amyloid pathology.
3. Even more concerning is the conclusion that "down-regulating ApoE expression as a possible therapeutic approach to alleviate Aβ-dependent neuronal impairment." Perhaps this would be an appropriate therapeutic approach for a FAD-Tg mouse. But without the interactions between h-APP/PS1 and the h-apoE isoforms, this is a potentially dangerous conclusion.

Answer:
We respectfully disagree with this statement. Indeed, our study shows that the absence of apoE improves Abeta-related neuronal dysfunction and synaptic loss, which clearly establish apoE as an important partner of Abeta to induce neurotoxicity. Based on those readouts, hypothesizing that down-regulating apoE expression in patients could be a way to alleviate Abeta-dependent neurotoxic effects and be beneficial to AD patients makes sense. A recent paper published in 2017 by the group of Dr. Holtzman at WashU has attempted such therapeutic approach in APOE4 AD Tg mice using antisense oligonucleotides targeting apoE (PMID 29216448). This work showed that a significant decrease of the levels of apoE mRNA and protein could be achieved after intracerebroventricular injection of ASO, and that lowering apoE after Aβ seeding modulates plaque size and toxicity (similarly as the present study).

Referee #2:
In this manuscript, the authors try to address the question whether loss of Apoe had any detrimental consequences (in wild type mice) and can reduce Abeta amyloid damage (in APP/PS1 mice). The authors use an experimental paradigm based on measuring neuronal activation in the visual cortex. They show that "the absence of APOE largely abrogates Aβdependent neuronal dysfunction in a mouse model of amyloidosis" and that the "complete lack of APOE itself can impact tuning performances". The data support well their conclusions, although the effects seen are not high. Further, they show that lack of Apoe is beneficial at the level of amyloid-induced synaptic damage (loss of Apoe rescues the decrease in synaptic density). A very interesting finding is in the last part of the paper, where the authors show that lack of Apoe abrogates astrogliosis and microgliosis around amyloid plaques. The study does not conclusively address whether Apoe loss is a viable therapy, but is an interesting report nonetheless adding to the knowledge of this field.

Remarks:
The authors should provide information about power calculations for the number of mice that they have decided to use.

Answer:
We have not performed a power calculation beforehand for the present work but evaluated the number of animals needed based on a previous work from our lab using a similar experimental setting in a model of tauopathy (in vivo recording of visual-evoked neuronal responses in wild-type and Tg4510 mice, Kuchibhotla KV et al. PNAS. PMID 24368848). Another paper published in 2012 reported visual-dependent decline of neuronal function in a mouse model of amyloidosis with comparable features as the one used in the present work (Grienberger C et al. Nat Comm. 2012, PMID: 22491322). The results of those seminal studies reached significance with a number of mice of 3-15 mice/group, as compared with 5 <n <11 animals per group in our work.
Authors should refrain from using "trend" when a comparison does not show a p value below their set threshold (e.g. page 4 when commenting Figure EV3, or page 5 last paragraph).

Answer:
We have now corrected this issue and disambiguate those statements for more clarity.
On page 5, first line, authors say that "non-responder" neurons were almost absent in all groups (except APP/PS1-Apoenull), however looking at Figure 2C there seem to be cells in all groups, particularly in Apoenull group. Authors should reword this sentence, and possibly have numbers of percentages of non-responders for each group in the text, as it is difficult to extrapolate from the plot in 2C.

Answer:
We have now rephrased this sentence and included the percentages of "off-responders" for each experimental group: "Intriguingly, the percentage of off-responding neurons was much lower in all the other groups considered (wild-type: 4.321.27%; APOEnull: 4.681.19%; APP/PSEN1/APOEnull: 3.20.84%), and especially in APP/PSEN1/APOEnull mice lacking apoe expression, suggesting that APOEnull mice normalizes an alteration from normal physiology observed in APP/PSEN1 mice." Loss of Apoe causes a drastic decrease in the number of microglia and astrocytes recruited at plaques. Could the authors add a quantification of microglia and astrocytes in wild type and wild type-Apoenull? It would be quite interesting to add the information about whether loss of Apoe in APP/PS1 mice results in a completely abolished recruitment (i.e. there is no difference with a general astrocyte / microglia distribution in the parenchyma of wild type mice, meaning that in absence of Apoe plaques are gliosis-inert).

Answer:
The gross evaluation of the density of microglial cells between wild-type and apoE-null mice did not reveal any striking difference, and therefore we did not perform a complete stereological evaluation to confirm those findings.

Referee #3:
Hudry et al. investigated apolipoprotein E (ApoE) effects on neuronal function and synaptic integrity in adult wild-type, Apoenull, APP/PS1 and APP/PS1;Apoenull mice. Using two-photon calcium imaging to record visually-evoked responses, authors first found that genetic removal of Apoe improved neuronal responses in 8-10 months of age APP/PS1 mice. In addition, these mice developed fewer parenchymal amyloid plaques with a reduced amount of synaptic loss -using high-resolution array tomography -and activated glial cells. Secondly, they also demonstrated that neuronal function is disrupted in aged mice (18-20 months of age) lacking Apoe, even in absence of amyloid. They concluded that ApoE has a dual effect with a neurotoxic component during early stages of amyloid brain load and a neuroprotective component in later stages of aging. The effect of lack of murine Apoe in models of amyloidosis has been extensively studied for the past two decades. Several concerns exist about the current study, as described below: 1. Several studies have shown that Apoe-/-mice exhibit a vascular phenotype including pronounced blood-brain barrier (BBB) leakage of blood-derived molecules, increased BBB permeability to Gadolinium contrast agent, perivascular IgG and fibrinogen deposits, loss of tight junctions, increased MMP-9 cerebrovascular expression, loss of pericytes, basement membrane degeneration, etc. Answer: We did not focus our analysis on the impact of the lack of apoE on the integrity of the neurovasculature, as we primarily analyzed its effect on neuronal function. We agree with the reviewer that apoE has been implicated in the maintenance of the BBB integrity, a function that could definitively contribute to the observed phenotype. Other important roles of apoE in brain lipid metabolism, myelin integrity, neuroinflammation, neuronal response to stress or via its direct interaction with Abeta could participate in apoE-dependent preservation of neuronal function in APP/PS1 mice or loss of neuronal functionality in aged apoE-null mice. The relationship between the vascular anomalies caused by the lack of Apoe and their effect on neuronal and synaptic integrity could potentially be the object of another study.
2. Plasma cholesterol is typically high in Apoe-/-mice. Has the same been observed after global Apoe silencing? Would silencing Apoe in the central nervous system have different effect from global Apoe silencing? Answer: We apologize if our manuscript was not clear enough and the description of the mouse model used in our study confusing. We have actually used mice presenting with a complete knock-down of the murine apoE gene in both the nervous system and the periphery. Unfortunately, no mouse model yet allows to conditionally ablate the apoE gene in the nervous system or the liver independently. We agree that it would be a very powerful approach to be able to do so.
3. How does Apoe genetic ablation rescue synaptic integrity near Abeta plaques? Is this a direct or indirect effect? Investigating some mechanistic aspect would greatly strengthen the present findings.

Answer:
We agree with the reviewer that this is a very interesting finding, especially because previous papers have demonstrated that Apoe may act as a chaperone for the recruitment of Abeta neurotoxic species at the synapse (Koffie RM et al. Brain 2012). Intriguingly, our array tomography data do not show that there is necessarily less Abeta at the synapse, which goes against this initial hypothesis. In our manuscript, we therefore discuss the possibility that the rescue effect toward Abeta-dependent synaptic loss observed in absence of apoE may be due to the fact that the association between Abeta and apoE changes the conformational state of those neurotoxic species and leads to more harmful impact on synapses. In vitro, apoE has been previously shown modulates the formation of neurotoxic Abeta oligomers (Hashimoto T et al. J. Neurosci. 2012), but we have not assessed this phenomenon in the present study.
4. The authors findings show "that the absence of APOE largely abrogates Aβ-dependent neuronal dysfunction in a mouse model of amyloidosis." Other groups have shown that Apoe-/mice have a phenotype with cognitive behavioral deficits and neuronal dysfunction (Nature. 2012 May 16;485(7399):512-6;Front Aging Neurosci. 2016 Nov 29;8:287;Behav Sci (Basel). 2018 Mar 3;8(3)). In the context of ApoE deficiency, how is neuronal function impaired in the absence of Abeta (refs above) and neuronal function improved in the presence of Abeta (authors' findings)? Please expand on this point and include a working model of how ApoE ablation improves specifically Abeta-dependent neuronal dysfunctions.

Answer:
The reported cognitive deficits in apoE-deficient animals remain somewhat controversial, as stated in the introduction of our manuscript (a few papers demonstrated cognitive impairment in those mice while others failed to do so: Grootendorst et al., 2008, Hartman, Wozniak et al., 2001. This is exactly why we initiated the present work, so that we could get a better understanding of the impact of the lack of apoE on neuronal function at the level of single cells (using multiphoton calcium imaging) and within a simple neuronal circuit (the visual network). Our findings do not disagree with the above references highlighted by the reviewer. Indeed, we also detected some impairment in the visually-evoked neuronal responses in aged Apoe-null mice, even without any amyloid pathology. In the discussion of our manuscript, we emphasize the possible molecular mechanisms that could explain why the lack of apoE alleviates Abeta-dependent neurotoxicity while at the same time sensitizing the brain to stress in older animals, even without amyloid pathology. 5. The rationale to examine neuronal function and synaptic integrity specifically in the visual cortex should be provided. Does neuronal dysfunction occur in other brain regions relevant to AD pathology?

Answer:
The study of neuronal dysfunction in the visual cortex is relevant to AD, as deficits in central sensory processing have been reported in the disease, particularly at advanced stages (PMID: 19713001, 1996878). In addition, the visual cortex is a convenient model to assess network integrity in vivo because it is easily accessible to perform either electrophysiology recordings or intra-vital calcium imaging (PMID: 22196337). The visual area V1 is also the output of a relatively simple circuitry, downstream of the retina and the lateral geniculate nucleus within the thalamus (PMID: 28772103), therefore facilitating the study of neuronal integration and response to well-controlled sensory stimuli.
We have now modified our manuscript and added this justification for our approach in the first paragraph of the results.
6. Fig.4A: The overall level of Abeta doesn't change in APP/PS1 vs. APP/PS1;Apoenull mice, however the distribution appears different between the groups (larger in diameter/size in