2023 Autumn Link Review
Significantly Enhancing Adult Intelligence With Gene Editing May Be Possible
by GeneSmith and kman
I spent many years reading IQ literature and following research developments. It's the strongest predictor of educational performance, income, socioeconomic mobility, and many more life outcomes deemed important in our culture. Full stop. It has a powerful influence over generations through assortative mating and its disconcerting origins in our DNA. Environmental interventions for IQ have failed again, and again, and again. The genes behind IQ are too numerous and gene editing is too expensive. In vitro doesn't begin to pay off for at least two decades, and the price tag and inconvenience don't help. The egregiously low fertility rate of high-IQ individuals only made the long-term worse. With the dawn of AI products, the near-term may be just as grim.
I figured because many of the doors to economic productivity, wealth, status, and family opportunity were jammed with this intractable problem, there was little hope of solving it. The culture would have to keep playing whac-a-mole with the many symptoms of this inequality. That was until I read GeneSmith and kman’s post.
My doomism has evaporated, and a glimmer of hope has worked its way into the imagination. The protocol proposed here is not limited to IQ but can be extended to other polygenic traits ranging from disease risks like diabetes, Alzheimer's, and schizophrenia to personality dispositions like Dark Triad, charisma, and musicality. It’s one thing to make these changes in an embryo fertilized in a lab, it’s wholly another to do it in a walking, breathing human. Not only can future generations benefit overwhelmingly, but our current population can as well. What was once a sci-fi trope beyond Gattaca’s wildest imaginations has crested over the horizon of possibility.
Sample size is everything
By increasing genome samples from 100k to 100m, we can uncover more and more genes and alleles and grow the effect size in adults
500 edits from 5 pts to 50 pts
2500 gene edits from 7 pts to 165 pts (you're reading that right)
23&Me has north of 12m genomes
More gene samples are only accompanied by IQ proxies like educational attainment instead of IQ recordings (22:1 ratio)
This is largely due to taboos in gov't and academia on genetics and cognition that impact funding availability
CRISPR, base editors, and prime editors
CRISPR was the original gene-editing technology, but it created lots of errors and raised alarms in the immune system to kill the cell
Base editing reduced CRISPR's error and immune response problems, but it was serial and therefore very slow
Prime editing inherited base editing's safety gains but did it in batches to improve speed
Targeting brain cells is the major hurdle
Option 1: lipid nanoparticles
The mRNA immunization technology from COVID vaccines has potential
mRNA is a set of instructions cells use to create proteins
It's very sensitive, so the vaccines delivered them to cells in containers
It's very powerful, so those containers only have molecular keys to certain cell membranes, like neuron membranes
The vast majority of neurons are not directly exposed to blood (filters at the liver & blood-brain barrier)
There's evidence from cancer treatment papers that allowed treatments to be administered with receptors used by cholesterol to bypass the blood-brain barrier
Option 2: adeno-associated viruses (AAVs)
The go-to for most current gene therapies
Viruses are very good at getting into cells, so they're viable candidates to deliver the new DNA
DNA floating in the nucleus can last weeks, even months
Undesirable since immunity will be developed, making repeat dosing harder
Option 3: engineered virus-like particles (eVLPs)
They have a similar protein structure to AAVs, but they typically carry proteins, not DNA
Proteins have much shorter lifespans in cells than DNA
Most edits occur shortly after the editor molecules are synthesized
The extra life-span is really just a risk that allows errors to be created
The envelope's proteins can be customized to bind and enter certain types of cells
Repeat dosing is also a problem
Developmental genes vs operable genes
Some genes are only expressed during development, so modifying them in adults is pointless
There are good recordings of the lifetime expression of most genes in the brain
We would know which are active in adulthood and which aren't
Off-target risk still exists for base and prime editors
Frameshift mutations are the real danger
Codons are 3 nucleotides used to create amino acids
If an extra 1 or 2 nucleotides are added to the gene, then the codon combinations change, and the proteins will come out very wrong
To avoid this, target non-coding regions such as regulatory regions
These regions influence production instead of the protein design
98% of intelligence-related genes are in non-coding regions
Neurons experience mutations regularly
20-40 per year, thousands by 45 years old
They do not appear catastrophic
How many edits to get a change in phenotype?
Simultaneous edits (multiplex editing)
Small scale for wheat with 8 genes at once
Evidence for thousands of simultaneous edits, but it was for one gene with thousands of copies
No publications on hundreds of edits to different genes yet
Don't have to do it all at once
Delivery vector with no immune response/immunosuppressants/rotating antigenic profile
Do X edits in the first round, X in the second, and so on
Why are there so few studies on multiplex editing?
Black box biology: you understand the gene affects the phenotype but don't know how. Many deem editing these to be dangerous
The wake of the eugenics era: many academics (and non-academics) are uncomfortable with the idea that genes influence life outcomes from diseases like Alzheimer's to lifetime traits like intelligence or personality
An unclear technical roadblock: qualified feedback has focused on the problem of the vector and the multiplex editing, but both appear addressable.
The opportunity cost of low-hanging fruit: easier problems can be solved with less time, more certainty, and less money. Publish-oriented, positive result academics and risk-averse investors and grants would rather spend that time on monogenic diseases and recurring revenue therapy.
High barrier to entry: this kind of editing is uncertain, technically difficult, and has political luggage. It requires a leap of faith that the end goal can be met.
Problems that will and won't be solved by markets
Will
Targeting specific organs
Might
Prevent immune response to editor proteins and delivery vectors
Off-target edits in multiplex editing
Won't
Create better intelligence predictors
Multiplex editing en masse
Edit efficacy in adult animals
Being a Single Man in Rural China
by Isabelle Attane and the DefiChine research team
The number of single men in rural China has reached astronomical heights. Many men in the country's interior have gone unmarried in the last few decades. This has reached extremes of 18 never-married men to every never-married woman at age 50. A few factors influenced this, including prenatal sex selection, sex-based internal migration policies that prohibit many men from migrating to cities, growing emphasis on expensive status markers (like a home), bride prices, and stigmatization of male suitors after 30. Soliciting prostitutes was more common among married men and wealthy single men.
I haven't read much into the survey methodology, so I can't attest to the quality of the data. I do have a few objections to their study, specifically on high-risk and socially disapproved practices:
They cover very few of these, and self-reporting crime in a very conservative, Big Brother society is probably not the best way to gather the evidence.
They only cover prostitution, alcohol abuse, the number of sexual partners, and depression.
Wealth is heavily tied to marriage, and you need spare money to purchase prostitutes. We expect married men to have the money to purchase prostitutes, and the rates of wealthy, single men being near identical suggests the underlying factor is money, not sex availability. They note this but for some reason thought poor, single men not soliciting prostitutes was suggestive of them not engaging in high-risk or socially disapproved practices. Same thing with alcohol consumption.
Unpaid sexual partners still perform mate selection to some degree. The fact that men who were deemed attractive enough to marry were also attractive enough to have extramarital sex with is no surprise. Impoverished, unmarried men probably score lower here because they just haven't had the same sexual opportunities that married men had.
Their brief evidence is too confounded to be indicative that unmarried men do not engage in risky, societally detrimental behaviors like violent crime.
DefiChine is focusing on a subject that I think is profoundly important: the behavioral impact of modern social environments on mating markets. It doesn't take a psychology and evolution nerd to know that sex is the primary driver of behavior, human or otherwise. China and India are great case studies due to the high sex asymmetry in some regions' demographics. However, I would argue that non-monogamous cultures create comparable markets. This is a topic I may explore in greater depth another time.
The Cycle of Violence
by Cremieux
In the wake of the October 7th attack on Israel, Cremieux covers an extensive history of the relationship between Jews and Muslims since the birth of Islam. He then goes on to look at analyses of the Israeli-Palestinian conflict that apply vector autoregressive (VAR) models to the conflict to reveal the effects of each agent's actions and the sequence of actions and reactions. VARs are time-series models used to test whether A reacts to B or B reacts to A.
The post is dense, long, and politically contentious, but it's among the best I have come across.