Action Potentials for August
1: Driving progress towards a long-term goal of neuroscience, a new microscopy method combines serial section electron microscopy with fluorescence imaging using molecular labels derived from antibodies. This enables the multiplexed mapping of cell types, proteins, and synaptic connections in the same sample. The images speak for themselves.
2: Memories are thought to be stored in specific groups of neurons called neuronal ensembles. When a memory is formed, these neurons become active together, and then undergo synaptic modifications, allowing for long-term storage of the memory. Re-activation of these neuronal ensembles has been found to be both necessary and sufficient for memory recall.
How are certain neurons — and not others — grouped together to form the neuronal ensemble? One factor that determines which neurons are allocated to a memory ensemble is its intrinsic excitability — basically, the extent to which it tends to fire action potentials — at the time of learning. Highly excitable neurons are more likely to be recruited into the ensemble representing a new memory, allowing memories formed close in time to share populations of neurons that can link the memories.
A new study corroborates this paradigm and takes it in an interesting direction, finding that linking a neutral memory to a fear memory days later involves reactivating the neutral memory's cell ensemble during an offline period:
While recording from the brains of the mice, the researchers also tracked their movement. They reported an interesting finding — that mice decreased their locomotion about one second before and after the bursts of hippocampal activity wherein the neutral and fear memory ensembles were co-active:
I find it interesting to speculate about why the mice might have slowed down their locomotion during this time. Could it be due to the cognitive load of emotional processing?
3: Recent experiments have shown that memory engrams exist as neural ensembles simultaneously across many brain areas. For example, one study found that 117 out of 247 brain areas mapped were significantly reactivated during fear memory recall, suggesting that neuronal ensembles for memories are not localized in a single area but dispersed across the brain. A new theoretical study suggests that bioelectric fields — ephaptic coupling — may act as a conductor to tie together distributed memory engrams across multiple brain areas.
4: A new study finds that the proper modification of receptive fields in cerebellar Purkinje cells requires both synaptic plasticity mechanisms such as long-term potentiation (proxied by CaMKII inhibitory autophosphorylation) and intrinsic plasticity mechanisms that control neuronal excitability (proxied by the regulation of SK2 potassium channels). They suggest that intrinsic plasticity acts as a gate that enables synaptic plasticity to effectively alter neuronal output.
5: Study finds that vinegar flies will voluntarily spend time on a spinning platform — i.e. a carousel. The authors suggest that this may be because this “play-like behavior” helps them to strengthen their neuronal circuits for proprioception and self-representation.
6: Study identifies three genetic clusters of dopamine neurons. Another good example of how molecular neuron type classification can help in predicting electrophysiology.
7: New study performs ex situ perfusion fixation of the postmortem human brain, followed by immersion fixation and then graded immersion cryoprotection in sucrose. They report good preservation of histology, seen here in the cerebellum.
8: New cryo-EM structure of CALHM1 (calcium homeostasis modulator 1), a voltage-dependent channel involved in neuromodulation, reveals its lipid-binding sites. When people think of “structure”, they often think of “things you can see under the microscope”, often forgetting that biomolecules are structures too.
9: New study using resting-state fMRI finds that higher g-factor scores — i.e. general capacity across cognitive domains — is associated with more stable functional connectivity states (i.e. potentially more “deeply-entering” states) and a lower idiosyncrasy of those states relative to the population.
10: Among Royal Navy officers, variability in motivation is a better predictor of advancement to leadership positions than variability in mental ability. This has been my experience as well. The people who are most successful in a given endeavor seem to be the ones who are most motivated.
11: The FDA approves zuranolone as the first oral medication for postpartum depression. In the two trials described, it was used for 14 days once an evening and improved symptoms compared to placebo. I would have liked to see a more active comparison such as a benzodiazepine. The obvious confound is that just helping new moms to sleep is clearly going to be helpful for mood. But psychiatric medication trials are not known for having active comparisons.
12: MDMA and psilocybin can now be prescribed by psychiatrists in Australia.
13: Researchers analyzed rare genetic variants in UK Biobank participants without psychiatric disorders and found that damaging variants in genes associated with schizophrenia were linked to lower general cognitive ability. The strongest effects were seen for variants predicted to fully inactivate genes that are intolerant of loss-of-function mutations.
14: I agree with Kelsey Piper that DNA synthesis organizations should be required by law to “know their customer” and check the order for safety prior to shipment.
15: Paul Christiano on why he is encouraged by developments in large language models in re: AI safety.
16: Jason Crawford on AI safety: “[T]here is so far neither a strong theoretical nor empirical basis for power-seeking... Of course, that doesn’t prove that we’ll never see it. Such behavior could still emerge in larger, more capable models — and we would prefer to be prepared for it, rather than caught off guard.”
17: One thing I admire about anti-aging researcher Steve Horvath is that he collects large, heterogeneous data sets and then analyzes them all together, allowing him to identify the robust trends that are seen across individual data sets. For example, this is what he did in the development of the first epigenetic clock of aging, analyzing 82 public Illumina methylation data sets, in a 2013 study now cited 4800+ times. It sounds simple but I don’t know anybody else in the world who does it like he does. It’s not flashy, it doesn’t get you into Nature, but it does get you results that replicate.
This month he is the first author on a similar study, analyzing DNA methylation profiles of 2400 tissues derived from 37 primate species to build clocks predictive of age:
They found that as tissues age, there are two main types of DNA regions that change in methylation: (a) those specific to the tissue and (b) those that are common across all tissues, which are less dependent on cell composition changes. Interventions would theoretically be needed on both sets to slow or reverse body-wide aging.
18: The anti-amyloid therapy donanemab has shown promise in Alzheimer's trials, slowing cognitive decline by 33-40%, and seeming to help more in early disease stages. Over the past year, the amyloid hypothesis has been vindicated to a degree that I don't think everyone is reckoning with. This is a point in favor of expert consensus. However, amyloid still seems to also not be the whole story, as donanemab showed less efficacy in patients over 75, and there are probably more contributing neuropathologic causes of dementia than people realized years ago.
19: Apparently the Alzheimer’s disease community is also considering retesting BACE inhibitors, after their previous failures in clinical trials that actually led to worsening of cognition in some cases. If they are retried, my guess is that they will not work, because I think they have the side effect of hindering myelination. My reasoning for this is that (a) BACE is known to be highly expressed in oligodendrocytes and (b) the cognitive impairment they have caused in trials is known to be reversible on the timescale — a few months — that it takes for myelin to regenerate. But I could be wrong!
20: A larger choroid plexus — the structure that produces and secretes most of the brain’s CSF — seems to be a useful biomarker for frontotemporal lobar degeneration.
21: Jake Selinger has been inspiring me this month. Here is his essay on how he cannot access potentially life saving treatments for squamous cell cancer because of the slowness of the FDA. Here is an essay by his wife Bess on how he is suffering due to a thick glob of mucus continuously collecting in the back of his throat. Here is his essay on his regrets.
22: The power of nematode cryptobiosis is demonstrated again, as a new species of nematode is found that has been stored in the Siberian permafrost for ~46,000 years.
23: Mapping genealogy with ancient DNA from 7,000 to 4,000 years ago in present day France. And ancient DNA identifies currently living ancestors of enslaved people who lived in Maryland.
24: In cryonics podcasts this month, new Alcor Co-CEO James Arrowood was interviewed on the Cryonics Underground Podcast and Mike Perry was interviewed on the Open Dewars podcast.
25: On cryonics and regret.
26: Anna L discusses the mission of Nectome. She also shares her experience of stumbling into her dream job.
27: New 3 minute explainer video from the Brain Preservation Foundation:
28: High school student Franco Pieri has a far more reasoned take on cryonics than most of what I read about the topic online. The conclusion: “Because cryonics is scientifically and legally backed and has the potential for advancement within decades, it is clear that the practice is something that society must accept and maintain legally.”