Action Potentials for April
1: New study predicts the electrophysiology and transcription patterns of inhibitory neurons based on electron microscopy connectivity data. By integrating morphologic (M), electrophysiologic (E), and transcriptomic (T) properties of cells measured with Patch-seq, they defined 28 inhibitory MET-types in the mouse primary visual cortex:
The cell types exhibit distinct ultrastructural features and synapse connectivity patterns, such as axon myelination patterns, synaptic features, and target cell subclass connectivity profiles. They hypothesize that certain features measured in Patch-seq correlate with the connectivity patterns observed in electron microscopy for different MET-types:
2: New study explores how the C. elegans nervous system adapts during the dauer stage, a unique phase in their development where these worms can survive in harsh environments. Specifically, they reconstructed the connectome of the dauer nerve ring and compared this to public connectomes from other developmental stages.
They tried to discover the connectomic differences responsible for dauer-specific behavior. For example, they found more feedback connections from motor to sensory neurons in the dauer connectome, which might help these animals have quicker responses to environmental changes:
3: Study explores how developing neurons in the Drosophila visual system select their synaptic partners by integrating the synapse-level connectome with the expression patterns and binding specificities of cell adhesion molecules (CAMs).
This means that the choice of synaptic partners seems to be determined by the expression patterns and binding specificities of particular cell adhesion molecule proteins:
This is an example of a biomolecular pattern that could be used to infer what the original connectome state was if the connectome were partially damaged, but the biomolecules were still present.
4: Nice data set with volumetric immunolabeling for pre- and post-synaptic proteins in formalin-fixed human brains. From 2020, but new to me. I found this helpful to get a better intuition of how synapses are preserved in fixed human brains in 3d:
5: Synchrotron-based Xray nano-holography (XNH) to segment axons:
6: Preliminary results evaluating the effects of low-dose whole-brain radiation therapy for Alzheimer’s disease. One year after treatment, of the 5 patients tested, 3 had improved cognition scores and 1 had stable scores. It probably won’t replicate based purely on historical trends, but interesting.
7: Meta-analysis finds that the evidence for a beneficial effect of exercise on cognition in aging is very small.
8: Alison Abbott reports on current clinical trials in Alzheimer's disease. The main trials discussed:
- Tau NexGen: An international trial testing combination therapy using lecanemab and a tau-reducing antibody (E2814) on 168 participants with a genetic predisposition to early-onset Alzheimer's.
- ATP Trial: A planned US-based trial for sporadic, late-onset Alzheimer's, testing different tau therapies in combination with lecanemab on around 900 participants with early signs of plaques and tau tangles. Awaiting NIH funding decision.
- Early clinical trial of γ-secretase modulator: A trial testing an oral compound that modifies γ-secretase enzyme behavior, making it produce shorter, non-toxic amyloid proteins. Sponsored by Acta Pharmaceuticals and funded by the NIH.
- TREM2-targeting antibody trial: An early-phase trial testing an antibody that binds to TREM2 and activates microglia to clear plaques, as a solo treatment.
- Anti-tau and anti-amyloid vaccines: Several vaccines are in preparation or early-phase clinical trials that aim to train the brain's immune system to recognize and destroy tau, amyloid-β, or both.
- APOE gene therapy trial: An open-label trial by Lexeo Therapeutics testing the delivery of the protective APOE ε2 gene variant to mitigate the harmful effects of the higher-risk APOE ε4 variant in 15 volunteers with mild Alzheimer's symptoms.
- Alkahest’s trial: A small clinical trial testing whether factors in the blood of young people could replace those lost in the aging process and help with Alzheimer's.
Overall, most recent drug developments, like the antibody lecanemab, focus primarily on lowering amyloid plaques or tau tangles and thereby attempting to slow cognitive decline. The big money in the field remains firmly in the amyloid hypothesis regime.
9: New meta-analysis of Alzheimer's disease clinical trial data reignites debate over whether ventricular expansion in patients receiving anti-amyloid immunotherapy represents cellular degeneration or a more benign effect. Reminds me of similar debates in the neuroleptic treatment literature.
10: A phase I trial by Biogen finds clearance of tau tangles in the brains of people with Alzheimer's disease for the first time. They used an antisense oligonucleotide drug, which reduced soluble tau by 60%. Phase II will test for effects on cognition.
11: Are older people more at risk for suicide in part due to an increased burden of neuropathologic diseases? This article suggests that argyrophilic grain disease is a risk factor for suicide in Japan.
12: Review of rapamycin in anti-aging. Claims that before 2009, the consensus was that aging could not be treated or would require a youth factor. In 2009, the NIA’s Intervention Testing Program, which set out to find compounds that can extend lifespan in mice, successfully identified rapamycin. It is now considered the only drug that has been consistently demonstrated to increase mammalian longevity. And a new study at Columbia is planning to test its ability to extend women’s reproductive health and fertility.
13: Localized use of senolytics fails once again, this time in a Phase II trial to treat wet age-related macular degeneration.
14: Study on the physiological benefits of senescent cells in salamander limb regeneration.
15: In a new placebo-controlled LSD microdosing study (n = 40 in each group), there were benefits found for mood on dose vs non-dose days, but 10% dropout in the microdosing group due to treatment-related anxiety. Seems like there might be an anxiety vs anhedonia trade-off here, which is a common trade-off in psychopharmacology.
16: In a large prospective study (n = 6980), a higher baseline cardiovascular health score and an improvement in cardiovascular health over a 7-year period were both linked to a reduced risk of experiencing depressive symptoms. The odds ratio of depressive symptoms for each additional cardiovascular health metric at intermediate or ideal levels at baseline was 0.87 (95% CI, 0.84-0.91), and for each improvement over the 7-year period, the odds ratio was 0.91 (95% CI, 0.86-0.96). More evidence that cardiovascular health is important for preventing depression, and vice versa.
17: KarXT, an M1/M4-preferring muscarinic agonist, reported positive Phase III trial results reducing symptoms of schizophrenia. Compared to placebo, there was a reduction on the PANSS scale of 8.4 points at five weeks: -20.6 KarXT vs. -12.2 placebo; p<0.0001; Cohen’s d effect size of 0.60.
If this is approved, as well as the more widespread availability of GLP-1 agonists (like Ozempic or Wegovy) to treat the metabolic side effects of neuroleptics, the upcoming years may provide some much-needed improvements in the treatment of schizophrenia.
18: A 20-year follow-up of an RCT of early intervention services for first-episode schizophrenia spectrum disorder patients finds no significant differences in long-term outcomes between the early intervention and treatment as usual groups.
19: Speaking of GLP-1 agonists, here is an article predicting that they might significantly increase lifespan in certain people. A group of forecasters at the Swift Centre predicts that if they are widely used in the UK, this will increase the UK’s average lifespan by 1.2 years:
20: Spatial ATAC-RNA-seq analysis of the human hippocampus:
They find distinct ATAC and RNA clusters that align with major anatomical landmarks and marker genes, such as the granule cell layer (GCL). Because ATAC-seq maps open chromatin regions across the genome, this allows for combination spatial chromatin and gene expression studies.
Regarding sample prep, the tissue was initially frozen (without cryoprotectant). After rewarming, the tissue was fixed with formaldehyde, but only for a very short time (5 minutes) and a very low concentration (0.2%, while the typical concentration is 4%).
21: New study on the evolution of APOE using Eurasian ancient DNA samples. This study found that the ε4 allele was more common in Western Hunter-Gatherers than in the first farming populations or modern Europeans. For context, ε4 is the main risk factor for Alzheimer's disease.
The ε4 allele seems to have been an adaptation to the hunter-gatherer lifestyle, offering protection against certain pathogens, improved cognitive development, and increased fertility in high-pathogen environments. In contrast, the Early Farmers showed the lowest ε4 frequency, with a possible selective sweep in the APOE region, as the transition from hunting-gathering to farming may have resulted in a mismatch between a pro-inflammatory environment and the pro-inflammatory ε4 allele.
22: Both cognitive and noncognitive skills make contributions to academic achievement. Noncognitive ones include emotional stability, attention regulation, impulse control, and motivation. A new study finds that noncognitive genetic contributions become increasingly predictive of academic achievement as children get older:
23: Large language models are pretty well-calibrated:
This is underappreciated. Back in the olden days (2022) when I was mostly using Open AI playground, it was really useful to be able to visualize the probabilities of the outputted tokens. This could be a helpful part of the user interface for language models, to help assess the extent of confabulation.
24: Cal Newport on LLM consciousness: "ChatGPT is absolutely not self-aware, conscious, or alive in any reasonable definition of these terms. The large language model that drives ChatGPT is static. Once it’s trained, it does not change; it’s a collection of simply-structured (though massive in size) feed-forward neural networks that do nothing but take in text as input and spit out new words as output. It has no malleable state, no updating sense of self, no incentives, no memory. It’s possible that we might one day create a self-aware AI… but if such an intelligence does arise, it will not be in the form of a large language model."
25: My current take on AI safety. Even though I have some disagreements, I continue to agree with Robin Hanson that it is very unlikely that one AI will so rapidly become so much more powerful than others so as to be able to take over the world. As a result, it seems to me that the best way to address the possibility of AI harm is with standard methods of managing communities, like laws, policing, norms, liability insurance, rewarding honesty, and social approval or disapproval. I might be wrong, though. I find the topic confusing.
26: Jacob Cannell: "Underlying much of the rationalist groupthink on AI safety is a set of correlated incorrect anti-connectivist beliefs which undermines much of the standard conclusions."
27: José Luis Ricón started a job at the anti-aging company Retro Biosciences. He also describes why he’s not directly doing AI safety research — he thinks that AI-related cybersecurity is more important than alignment research, and he’s not an expert in cybersecurity.
28: New in situ perfusion fixation protocol for human brain banking. They first cannulate the bilateral common carotid arteries (comCa):
They then perfuse saline for washout and fix the brain with 4% paraformaldehyde. They report high-quality confocal microscopy and electron microscopy in the preserved brains:
29: Max More on the Jameson Satellite by Neil Jones. Published in 1931, this is the story that inspired Robert Ettinger to propose cryonics in 1962.
30: Please consider signing this petition from the Brain Preservation Foundation: “Allow global access to high-quality brain preservation as an option rapidly after death”.
Very interesting !