03 Jan 2025

[Editor’s note: this page closes an Open Letter debate between Scott Small, Columbia University, New York City, and Dennis Selkoe, Brigham and Women’s Hospital, Boston. Selkoe’s replies are interspersed with Small’s arguments posted on 31 December 2024, which therefore reappear on this page. To ease the reader’s experience, Small’s prior points are in regular font, Selkoe’s responses are italicized.]

Dear Dennis,

I thank you for your response, for its thoughtfulness and thoroughness.

The first round of a debate is most useful for isolating the crux. I therefore kindly ask for one more public round. This would be most beneficial, not just for me, but also for the many Alzforum readers who are following our conversation.

Dear Scott, Although I believe our exchange heretofore covered our points of agreement and disagreement well, I understand your wish to respond to what I have written and your request for a final response from me. This is that response.

The Crux
It seems that it all boils down to “when and where.”

The When
The easiest way to track down “when” in a chronic disorder like AD is with the causal quartet, those carrying disease-causing mutations in APP, PSEN1, PSEN2, and SORL1, and, as you expand, including those with trisomy 21.

I believe some of our colleagues with AD expertise would not necessarily speak of a “causal quartet.” Beyond the three autosomal-dominant genes for which there are decades-long evidence of an Aβ-generating mechanism underlying their causality of early onset AD, there is also a sizeable and growing number of genetic risk factors for what was originally called “sporadic” AD. Possible causative mechanisms have been reported for some of these. This iterative exercise of unearthing plausible mechanisms is one of the most important thrusts of further explanatory research in AD genetics.

Yes, in them, their neuronal endosomes start overprocessing APP and secreting Aβ. at birth, if not earlier. And it is biologically just as true that this coincides with when endosomes begin the jamming process.

Here, you and I separate our formulations of AD’s earliest cell biological mechanisms. I am less convinced than you that the constitutive production of Aβ peptides, which are continuously secreted from recycling endosomes that fuse with the plasma membrane (as in many other protein secretory events), coincides with a “jamming process.” In my view, concluding the latter requires more cell biological evidence as to what portion of newly generated Aβ peptides is secreted and what portion may be incapable of being secreted and thus retained in endosomes in a way that compromises endosomal function. Such comparative quantifications need to be done in both normal, healthy cells and in cells expressing various AD-causing gene defects, to prove that impaired endosomal secretion of Aβ peptides occurs more in the latter than the former. Thus, while you and I agree that Aβ production is initiated intracellularly, the peptides are largely destined to be secreted, as are other products of presenilin/γ-secretase substrate cleavages. Because we clearly observe abundant extracellular Aβ in brain interstitial fluid (e.g., using brain microdialysis in healthy animals) and in the conditioned media of all healthy cultured cells, and we also find Aβ normally in human CSF and plasma, it is my view that the quantitative default pathway for newly generated Aβ is extracellular release.

In this context, you rightly say below that inhibitng β-secretase or modulating substrate cleavages by γ-secretase (i.e., with GSMs) are targeting intracellular events, but their immediate (and demonstrated) consequence is to reduce extracellular Aβ levels. So, as I stated in my original response, I fully support targeting intracellular processes therapeutically, and I particularly favor GSMs. But the cellular secretion process of peptides (including Aβ) is intimately related to extracellular levels of the secretory products. So, I think focusing our field on this tenuous distinction is not a compelling separation of a continuous process that follows the rules of cell biology. In short, lowering excess Aβ42 in AD is worthwhile by both intra- and extracellular approaches.

This pathophysiological coincidence has been empirically established and is unsurprising, because endosomal APP processing and endosomal traffic jams appear to be in a vicious cycle.

We differ about the centrality of “endosomal traffic jams” as necessarily coincident with physiological endosomal APP processing and Aβ secretion. This difference of perspectives calls for more quantitative cell biological experimentation to measure the degree and timing of endosomal secretion of Aβ versus the degree and timing of its retention within endosomes in neurons that express genetic defects which are known to cause AD. This balance could also be quantified in living animals.

So, if coincidental, if a vicious cycle, can’t we simply substitute the times you reference “Aβ production” as the earliest event, with “endosomal traffic jams” as the earliest event?

As argued above, I do not think we can, at this juncture in our mechanistic knowledge, simply make the substitution you suggest to explain the earliest events in AD, whether in familial or “sporadic” forms of the disease.

The Where
It is the “where” question, inside or outside of neurons, that is more pressing with immediate relevancy for drug development.

As I opine above, I don’t think this distinction of targeting events inside or outside neurons is so clearly separable cell biologically and pharmacologically. As regards your suggestion that focusing inside neurons “is more pressing with immediate relevancy for drug development” in AD, I could argue instead that the therapeutic approach we both agree has achieved some success to date is targeting extracellular Aβ. So, one might conclude that following up rigorously on something that has already worked to slow AD in patients—i.e., with better, safer agents in that therapeutic class—is at least as “pressing” as discovering new approaches which are not yet fully validated in patients. Obviously, both are important and relevant; it does not need to be a competition.

I agree with what I think you imply, that compared to drugs that clear extracellular amyloid, those that target the production of Aβ and other APP fragments inside neurons are better, because it is inside neurons where the process starts.

I did not mean to imply that targeting production of Aβ inside neurons is “better.” Intracellular production relates immediately to extracellular levels, and targeting extracellular levels (e.g., via antibody- or vaccine-mediated Aβ clearance) is neither “better” nor “worse.” This is especially true when AD chronology is taken into consideration: Once patients have substantial amyloid deposits extracellularly (whether without or with neurological symptoms), targeting just a so-called intracellular process (e.g., β-secretase inhibition) may not be quantitatively sufficient to correct Aβ dyshomeostasis and thus its downstream neurodegenerative consequences.

Because either those, or drugs that unjam endosomes, will break the vicious cycle, a cycle that leads to Aβ secretion and neurodegeneration through intraneuronal mechanisms.

I am not convinced of an obligatory “vicious cycle that leads to Aβ secretion and neurodegeneration through intraneuronal mechanisms.” The abundant secretion of Aβ (particularly the strongly self-aggregating Aβ42 peptide) has been shown to lead to the binding back of a portion of it (and its oligomers) to cells and resultant alterations of cell membranes, with neuronal—as well as astrocytic and microglia—consequences.

Genetics support this intracellular mechanistic conclusion. Whether as shown with protective BACE1 variants, which break the pathogenic cycle by reducing intraneuronal APP processing in endosomes.

But reducing intraneuronal APP processing also lowers extracellular Aβ levels and protects cells from extracellular Aβ-binding events such as plasma membrane injury, with subsequent tau alteration and neuritic dystrophy (as we and others have demonstrated).

Or, as more recently reported, by protective SORL1 haplotypes, which break the pathogenic cycle by increasing SORL1 expression which is known to unjam the endosome.

(I refrain from referencing recent observations suggesting that APOE protective variants might unjam endosomes because other papers propose neuroimmune mechanisms. The neuroimmune network does appear to mediate neurodegeneration and is the current focus of our ADRC and my lab.)

Your acknowledgment here of “neuroimmune mechanisms” in AD (including as modulated by ApoE variants), i.e., progressive inflammation due to microgliosis and astrocytosis, has been shown to include activation of these cells by extracellular oligomers and the diffuse-to-compacted plaque transition, all of which further support extracellular contributions to AD pathogenesis.

The therapeutic advantage of intracellular drugs
I am hesitant to “argue by analogy." But since you invoke analogies, shouldn’t we be developing the equivalent of “statins for the brain”?

Indeed, many AD experts, including you and I, have written that GSMs may represent just such an analogy: an oral agent administered chronically over years to reduce amyloidogenic Aβ42 accumulation and its extracellular deposition. But this may need to be combined with amyloid-clearing vaccines or antibodies to get the job done, especially once abundant plaques have accrued.

That is, targeting the most upstream intracellular pathogenic trigger, which we agree starts inside neurons. Even if clearing amyloid plaques does confer benefits.

Again, dividing intra- from extracellular events in a disorder of an obligatory secretory product like Aβ represents, in my view, a forced separation of a continuous cell biological process.

So, Dennis, in an attempt at forging synthesis from our opening statements, can we agree on this point: Our field should be developing drugs that target AD’s intracellular pathologies.

I’m afraid we don’t fully agree on this emphasis, Scott. While intracellular targets are highly worthy of pursuit (as I stated in my original response and again here), they should go hand in hand with extracellular approaches to diminish and neutralize the overarching problem of abnormal Aβ42 oligomerization and cytotoxicity.

Intraneuronal drugs might include the new generation of γ-secretase modulators, safe BACE1 inhibitors, endosomal recycling enhancers, or any other drug designed to break the intraneuronal vicious cycle. It can also, however, extend to include intraglia drugs, since the neuroimmune network appears to mediate neurodegenerative process.

As regards the latter, there is abundant evidence that extracellular Aβ accumulation can activate microglia and astrocytes and thus alter the “neuroimmune network.”

May I wish you, and the rest of the field who are listening, a productive 2025 in our collective crusade, battling one of the most dreaded diseases of our generation.

Scott

I heartily agree, Scott, and wish you and your colleagues all the best in your further studies of AD. Warm regards,

One Last Round in an Open Debate: Final Exchange

Quick Links

Tools

Comments

Make a Comment

References

News Citations

Further Reading

Annotate