https://www.nature.com/articles/s43587-026-01108-z
See also my second prompt
chatGPT:
Summary
This Nature Aging paper argues that reduced ULK1, a key initiator of macroautophagy and mitophagy, is a mechanistic contributor to Alzheimer’s disease rather than merely a marker. The authors report that ULK1 falls with age in serum and CSF in cognitively unimpaired people, and is lower in Alzheimer’s disease cohorts. They also find reduced ULK1 in human AD brain regions and cell types relevant to tau pathology, especially excitatory neurons and astrocytes. Higher baseline CSF ULK1 is associated with slower clinical progression by CDR-SB over follow-up.
In mouse and cell models, increasing ULK1 appears protective. ULK1-overexpressing neurons resist excitotoxic, mitochondrial and Aβ-related stress. In 5xFAD mice, ULK1 overexpression improves memory tests and reduces amyloid plaque burden, insoluble Aβ, astrocytosis and some inflammatory/pathological features. In tauopathy models, ULK1 overexpression reduces tau pathology, including acetylated tau at Lys174, and improves behavioural outcomes.
The proposed mechanism is: ULK1 upregulation → increased autophagy and mitophagy → improved mitochondrial quality → increased cellular NAD⁺ → SIRT1 activation → deacetylation of acTau-Lys174 → reduced tauopathy. The mitophagy arm is linked particularly to PINK1, FUNDC1 and AMBRA1 pathways. Pharmacological ULK1 activators, especially Rac-BL-918, reduce tau aggregation in tau-seeding assays and show benefit in a C. elegans tau model.
Claimed novelty
The main novelty is the paper’s attempt to connect human longitudinal biofluid data, post-mortem human brain data, mouse genetic overexpression, viral manipulation, cell culture, tau-seeding assays and worm tau models into one causal story about ULK1 decline in ageing and Alzheimer’s disease.
More specifically, the novel elements are:
-
ULK1 as a human ageing/AD-linked biomarker
The paper reports reduced ULK1 in serum and CSF with ageing in cognitively unimpaired participants, and lower ULK1 in AD. That gives the work a human clinical anchor rather than being only a mouse paper. -
Spatiotemporal and cell-type evidence in human AD brain
The authors examine ULK1 changes across Braak stages and report reductions in relevant cell types, especially excitatory neurons and astrocytes, rather than treating the brain as a single bulk tissue. -
ULK1 positioned as causal rather than correlative
By overexpressing ULK1 in AD mouse models and showing reductions in Aβ pathology, tau pathology and cognitive decline, they argue ULK1 loss is not just a bystander. -
Linking mitophagy to tau acetylation through NAD⁺–SIRT1
The mechanistic chain from ULK1-driven autophagy/mitophagy to NAD⁺, SIRT1 and acTau-Lys174 is a useful integrative model. -
Drugging ULK1
The use of ULK1 activators, especially Rac-BL-918, gives translational direction, although this remains early-stage.
Critique
This is a strong, ambitious paper, but the central causal claim still needs caution.
The human data are suggestive but not definitive. The cohorts show associations between ULK1, age, AD status and progression, but they do not prove that reduced ULK1 causes AD progression in humans. The authors themselves note that the cognitively unimpaired group was slightly older than the AD-dementia group and that stricter age-matching and larger clinical validation are needed. That matters because serum/CSF ULK1 could be influenced by systemic ageing, frailty, inflammation, cell turnover or blood–brain barrier changes, not just neuronal autophagy.
The mouse models are powerful but artificial. Ubiquitous ULK1 overexpression is not the same as restoring a physiological age-related decline in human neurons. Overexpression may create non-physiological effects, and whole-body ULK1 elevation may affect peripheral metabolism, immune cells or glia in ways that indirectly benefit the brain. The authors partly address this with hippocampal AAV experiments, but the broad overexpression model remains a limitation.
The 5xFAD model is heavily amyloid-driven and does not reproduce the full late-onset human AD process. Benefits in 5xFAD mice may reflect improved handling of aggressive transgenic amyloid pathology rather than proof that ULK1 activation would work in sporadic AD. Similarly, P301S tau models are useful but represent mutant tauopathy, not ordinary human AD tau spread.
The mechanistic chain is plausible but probably incomplete. ULK1 has many downstream effects beyond mitophagy. It can alter general autophagy, lysosomal function, inflammation, mitochondrial turnover and possibly kinase signalling. The authors acknowledge that the mechanism by which ULK1 reduces tau phosphorylation and acetylation remains obscure, and that ULK1 may act through additional substrates or autophagy-independent pathways.
The NAD⁺–SIRT1–acTau-Lys174 story is attractive, but it may be too linear. Increased NAD⁺ could result from better mitochondrial quality, altered metabolism, reduced stress or changed cell composition. SIRT1 effects on tau acetylation are plausible, but tau pathology involves many acetylation, phosphorylation, truncation, seeding and clearance events. It is unlikely that acTau-Lys174 alone explains the whole behavioural benefit.
The drug translation is very early. Rac-BL-918 looks interesting, but the data are mainly in cell systems, mitophagy assays and worms. Before it becomes a credible AD therapeutic lead, one would need pharmacokinetics, brain penetration, selectivity, chronic safety, dose–response, off-target kinase effects, and efficacy in mammalian AD/tau models after disease onset rather than mainly preventive/genetic designs.
Bottom line
The paper makes a credible case that ULK1 decline is linked to impaired autophagy/mitophagy in ageing and AD, and that boosting ULK1 can reduce amyloid and tau pathology in models. Its strongest contribution is integrating ULK1, mitophagy, NAD⁺/SIRT1 and tau acetylation into a single experimentally supported framework.
However, the evidence is still strongest in models, not in human causality. The translational claim — that ULK1 activation could be a therapy for human Alzheimer’s disease — is plausible but not yet proven. A fair reading is: ULK1 is a promising mechanistic node and therapeutic target, but the field now needs cleaner human validation, physiological restoration experiments, and post-onset pharmacological studies in mammalian models.
Second Prompt: what research points to a reduction in ULK1 activation as a result either of splicing changes or the introduction of nonsense mediated decay as a result of reduced histone acetylation
There is not yet a clean direct paper, as far as I can see, showing:
reduced histone acetylation → altered ULK1 splicing → ULK1 nonsense-mediated decay → reduced ULK1 activation
But there is a plausible evidence chain pointing in that direction.
1. The uploaded AD paper shows reduced ULK1 expression, but not the cause
The Nature Aging paper reports that ULK1 is reduced in ageing and AD: serum and CSF ULK1 fall with age, AD patients have lower ULK1, and human AD brain shows lower ULK1 mRNA/protein in relevant cell types. It also reports downregulation of other ULK1-complex genes — ATG101, FIP200/RB1CC1 and ATG13 — in excitatory neurons with advancing tauopathy.
However, that paper does not show that the reduction is caused by altered splicing, NMD, or reduced histone acetylation. It frames the cause of ULK1 decline as unresolved.
2. ULK1 is splice-capable, so mis-splicing is biologically possible
Ensembl currently annotates human ULK1 as having 10 splice-variant transcripts, while Human Protein Atlas notes only one protein-coding transcript from the gene. That combination is consistent with a gene where alternative transcripts exist but most may not encode stable canonical protein. (Ensembl)
This does not prove ULK1 undergoes disease-relevant NMD, but it means the architecture is compatible with nonproductive isoforms.
3. AS–NMD is a known mechanism for lowering protein output
Alternative splicing coupled to nonsense-mediated decay — AS-NMD — is a well-established mechanism by which cells reduce productive protein output. Reviews describe AS-NMD as a regulatory system that changes the ratio of productive to unproductive mRNA isoforms, often by generating premature termination codon-containing transcripts that are degraded by NMD. (PMC)
Ensembl’s own biotype definition says NMD is predicted where an in-frame termination codon lies more than 50 bp upstream of the final splice junction, i.e. the classic exon-junction-complex rule. (Ensembl)
So if ageing or AD shifted ULK1 pre-mRNA toward a premature-stop isoform, that could reduce ULK1 protein without needing a major drop in transcription initiation.
4. Histone acetylation can regulate splice-site choice
There is solid evidence that chromatin state and histone acetylation influence alternative splicing. A PLOS ONE study showed that HDAC inhibition changed splicing in about 683 genes, with effects linked to histone acetylation, RNA polymerase II processivity, and recruitment of splicing factors. (PLOS)
That paper’s mechanistic model is especially relevant to your hypothesis: histone acetylation around alternatively spliced elements can alter Pol II processivity and thereby change co-transcriptional recruitment of splicing regulators. (PLOS)
More generally, chromatin–splicing reviews support the idea that histone modifications help determine exon recognition by affecting chromatin structure, Pol II elongation and splicing-factor recruitment. (PMC)
5. Reduced histone acetylation could plausibly disturb ULK1 expression indirectly
A plausible model would be:
reduced nuclear acetyl-CoA / reduced histone acetylation → altered Pol II elongation and local chromatin state → altered ULK1 splice-site selection → increased nonproductive ULK1 isoforms → NMD → lower ULK1 protein → weaker autophagy/mitophagy initiation
The most relevant supporting link is that histone acetylation, especially marks such as H3K9ac, helps promote Pol II pause release and transcriptional elongation. (PMC) Since splicing is co-transcriptional, altered elongation kinetics can change exon inclusion or exclusion.
6. Why this matters for ULK1 activation
ULK1 activation is not just about the amount of ULK1 protein. It is also regulated post-translationally by AMPK, mTORC1, phosphorylation, acetylation and complex formation with ATG13, FIP200/RB1CC1 and ATG101. The uploaded paper describes ULK1 as the core initiator of autophagy and mitophagy, acting through the ULK1 complex and downstream Beclin1/VPS34/autophagosome pathways.
So reduced productive ULK1 mRNA would be expected to reduce the available pool of ULK1 that can be activated, but it would not by itself prove reduced kinase activation per molecule.
7. Current strength of evidence
I would grade the evidence like this:
| Link | Evidence strength |
|---|---|
| ULK1 is reduced in ageing/AD | Strong in the uploaded paper |
| ULK1 reduction impairs autophagy/mitophagy | Strong in models |
| Histone acetylation affects alternative splicing | Strong general evidence |
| AS-NMD can reduce protein output | Strong general evidence |
| ULK1 specifically is shifted to NMD isoforms in AD/ageing | Weak / not yet shown |
| Reduced histone acetylation specifically causes ULK1 AS-NMD | Not yet directly shown |
Best experimental test
The clean test would be to take aged/AD neurons or iPSC-neurons and measure:
- ULK1 isoform usage by long-read RNA-seq or targeted RT-PCR.
- NMD sensitivity by UPF1/SMG1 inhibition and asking whether hidden ULK1 premature-stop isoforms accumulate.
- Histone acetylation at the ULK1 locus by ChIP-seq/CUT&Tag for H3K27ac, H3K9ac and H4K16ac.
- Rescue with acetyl-CoA support, p300/CBP activation or HDAC inhibition, asking whether productive ULK1 transcript and ULK1 protein recover.
- Functional readout: p-ULK1, LC3 flux, p62 turnover, mitophagy reporters and mitochondrial quality.
So the answer is: the research supports the mechanism in principle, but the ULK1-specific splicing/NMD link remains a hypothesis rather than an established finding.