CRISPR & Gene Drives — Engineering Life for Space

In December 2021, astronauts on the ISS cut yeast DNA with CRISPR-Cas9 and watched the cell try to fix it. Microgravity shifted the repair choice toward non-homologous end joining, the sloppy pathway that swaps accuracy for speed. The first deliberate genome edit performed off Earth was, in effect, an experiment in how badly biology repairs itself when gravity is gone.

That result matters because the alternative to engineering radiation resistance is not "we'll figure it out." It is heritable cancer rates and reproductive failure on any mission longer than Mars. A Mars round trip delivers roughly 0.6 Sv of effective dose. An interstellar mission at 0.2c without exotic shielding is in the tens of Sv range. Evolution on Earth has never had to solve this problem.

How the toolkit works

CRISPR-Cas9 is a guide RNA that points a nuclease at a specific DNA address. The cell, faced with a double-strand break, picks one of two repair routes. Homologous recombination uses a template and is faithful. Non-homologous end joining staples the ends back together and is mutagenic. The 2021 ISS experiment (Genes In Space-6, Boeing/miniPCR/New England Biolabs) found microgravity biased yeast toward the second.

Gene drives are CRISPR pointed at the inheritance machinery itself. A drive cassette copies itself onto the homologous chromosome in the germline, so offspring inherit it from both parents instead of one. Inheritance climbs from 50% to near 100% per generation. Anthony James's lab at UC Irvine demonstrated this in Anopheles stephensi mosquitoes in 2015. The technique works. The reversibility does not.

The radiation problem, specifically

Galactic cosmic rays are not X-rays scaled up. They are HZE particles, iron and carbon nuclei moving near light speed, that punch clustered lesions through cells. Five to twenty double-strand breaks within ten nanometers. Even Deinococcus radiodurans, which survives 5,000 Gy of acute gamma exposure, handles HZE damage worse than gamma damage because the lesion geometry is different.

Three independent failure modes stack:

Clustered lesions overwhelm repair fidelity.
Reactive oxygen species do most of the actual damage, and antioxidant pools deplete.
Microgravity independently suppresses the HR pathway, as the 2021 ISS work showed.

Standard shielding scales with mass. Mass scales with launch cost. Biology is the cheaper substrate to harden, if you can harden it.

What's being tried

Dsup transfer. Takuma Hashimoto and Takekazu Kunieda at the University of Tokyo reported in Nature Communications (2016) that the tardigrade protein Dsup, expressed in cultured human cells, cut X-ray-induced DNA breaks by about 40%. The protein physically coats DNA. No whole-organism human work has been published. The cross-kingdom transfer worked first try, which is itself surprising.

Deinococcus pathway grafts. The Mn²⁺-peptide antioxidant system that gives Deinococcus its tolerance is biochemically transplantable in principle. Michael Daly's group at USU has been mapping the relevant proteome since the early 2010s. Engineering it into mammalian mitochondria is active but not published in humans.

Repair-pathway tuning. Upregulating RAD51 or BRCA2 to push the repair choice back toward HR. The cancer-risk problem is real: too much RAD51 shows up in tumor phenotypes. Dose and tissue specificity are the open engineering questions.

Space crops. Less ethically loaded. CRISPR-edited wheat, potato, and lettuce lines for low-gravity root development and UV tolerance are being characterized in ground-based microgravity simulators. A closed-loop Mars agricultural system probably needs dozens of stacked edits per crop.

Engineered microbes for life support. Near-term and undercounted. Radiation-hardened algae and cyanobacteria for CO₂ scrubbing and O₂ production. The microbe is small, the population is large, replacement is cheap, and the ethics are familiar.

What's contested

Whether human germline editing for space is ever defensible. The He Jiankui case in 2018 produced three CRISPR-edited babies in Shenzhen and a three-year prison sentence, and no international governance framework has caught up since. Space missions reframe the argument: the consent problem applies symmetrically to not editing children born into a radiation environment with no evolutionary precedent. Neither side has a clean answer.

Whether gene drives can be contained. Kevin Esvelt at MIT, who co-described the CRISPR gene drive in 2014, has argued publicly that population-suppression drives may be irreversible at planetary scale, and has spent most of the years since trying to design "daisy-chain" drives that self-limit. The off-Earth case is worse, not better: small isolated populations on a generation ship reach fixation in a handful of generations and there is no wild-type reservoir to dilute a mistake.

Whether Dsup actually works in vivo. The cell-culture result has held up. The organism-level result, in mice, has been harder to reproduce cleanly. The protein may need specific chromatin context to function.

Why this has to do with other realms

The Deinococcus-to-human transfer path is structurally concept panspermia performed deliberately and on a fifty-year timescale instead of accidentally over billions. CRISPR collapses the horizontal gene transfer that prokaryotes have always done into something humans can direct. The space context makes this less optional than the Earth context, which is the part of the argument that defenders of germline editing usually skip: the environment, not the engineer, is forcing the choice. The same logic appears in concept convergent evolution — tardigrades, Deinococcus, and radiotrophic fungi independently arrived at radiation tolerance, and CRISPR asks whether their separate answers can be recombined into one organism.

An open question

If you can engineer a human lineage to survive 10 Sv of accumulated dose over a 50-year journey, are the descendants who arrive at Proxima Centauri b still members of the species that left Earth? See tech generation ship for the ship problem, and concept fermi paradox for whether anyone else has already had to answer this.

Key sources

Hashimoto, T. et al., "Extremotolerant tardigrade genome and improved radiotolerance of human cultured cells by tardigrade-unique protein," Nature Communications (2016) — the Dsup result.
Gambacurta, A. et al., Genes In Space-6 mission reports (2021) — first CRISPR DNA repair experiment on ISS. To verify: peer-reviewed publication versus mission summary.
Daly, M.J. et al., work on Mn²⁺-peptide antioxidants in Deinococcus radiodurans — the canonical reference for non-enzymatic radiation defense.
Esvelt, K.M. et al., "Concerning RNA-guided gene drives for the alteration of wild populations," eLife (2014) — the founding gene-drive paper and its author's later containment work.
NASA Human Research Program reports on galactic cosmic ray exposure for Mars-class missions. To verify: specific dose figures vary by mission profile and solar cycle.

Abhishek