Green Goo (Republished)

Synthetic biology, mirror life, and threats to global ecosystems

Note: I originally published this post in July 2022, but I removed it after being asked to do so by biosecurity experts who viewed it as an infohazard. However, recent discussions in the biosecurity field have resulted in a new consensus that it’s better to talk about this issue openly, so I am re-publishing this post. For further reading I recommend this perspective article, published today in Science, as well as a companion piece in Asimov Press.

The nanotechnologist Eric Drexler popularized the term “gray goo” to refer to self-replicating nanobots that go out of control and digest everything to make more copies of themselves.

But it turns out we already have self-replicating nanobots: they’re just called cells.

Bioreactor image, courtesy of Max Schubert.

The Azolla event

Around 49 million years ago, the Earth was a greenhouse, atmospheric CO2 was ~3500 ppm, and the Arctic Ocean was a freshwater lake. Then something interesting happened.

Aquatic duckweed-like ferns of the genus Azolla thrived in the Artic Ocean, creating massive blooms and sequestering about 0.9 – 3.5 trillion tons of carbon over about a million years in total. By the end of this period, atmospheric CO2 was only 650 ppm.¹1. Although not all of this reduction in CO2 is directly attributable to Azolla, these blooms certainly helped push the Earth towards its current glaciated state.

Notably, there have currently been calls to re-create a similar event to fight global warming. This might backfire horribly.

A thought experiment

March, 2053. Ajinomoto-Bayer-Cargill have just opened the first pilot plant for the environmentally sustainable production of L-glucose, a promising artificial sweetener. Decades of research projects on opposite-chirality biomolecules have finally come to fruition, leading to the engineering of a mirror-image variety of the marine photosynthetic microbe Synechococcus elongatus. Now, S. elongatus² will capture CO2 from the air, and fix it into valuable chemicals. Ajinomoto-Bayer-Cargill can sell not only artificial sweeteners, but also carbon credits! What could go wrong?

July, 2053. Algal blooms are common in the Gulf of Mexico around this time of year, but this one is different. The algae die from overcrowding, but don’t seem to be decomposing at all. And the standard PCR tests for the usual algal suspects are all coming up negative. Scientists quickly determine that the bloom is due to the mirror-image S. elongatus, which apparently escaped somehow from Ajinomoto-Bayer-Cargill’s plant. Fingers are pointed, but nobody is ever able to determine the cause of the leak.

October, 2053. A crash effort has produced a mirror-image virus targeted to kill the S. elongatus, but by now the rogue organism (with a doubling time of 2 hours under ideal conditions) have spread throughout all the world’s oceans. Since they can fix both carbon and nitrogen, and have no natural predators, minerals such as phosphorous and iron are their only limitations, and they quickly sequester all available nutrients. The virus is humanity’s only hope to bring the S. elongatus under control.

April, 2054. The virus has actually made things worse. It kills most of the S. elongatus, but some still persist at a lower population density. The dead cells release nutrients (phosphorous, iron, etc.) back into the water, and carbon fixation continues. Too bad the fixed carbon is all unusable by normal life forms.

July, 2054. The ocean still tastes salty, but by now it is also faintly sweet from all the L-carbohydrates. All marine life has died out, outcompeted by the S. elongatus. CO2 levels are already down to 350 ppm with no signs of stopping, and Earth is clearly headed for a snowball state.

September, 2060. The last survivors of humanity huddle around a steam generator in the Arctic while children work in the coal mines to keep the boilers fed.³

Key takeaways

The use of synthetic biology to create engineered human pathogens is scary, but even worse things are possible if synthetic organisms can outcompete natural ones in ecosystems. Mirror-image photosynthetic organisms are a good example of this. The technology is not developed enough to cause immediate concern today, but research on mirror life poses risks for the future.

Right now, efforts to optimize photosynthetic organisms have resulted in “domesticated” strains with improved performance under laboratory or agricultural conditions but worse performance in natural environments.However, as the field of synthetic biology progresses, we should be aware of potential harmful effects of our creations on ecosystems, especially if they involve something that, like mirror life, is highly distant from any natural precedent. The broad range of potential threats means that targeted mitigation efforts (such as testing for a particular species of concern) may be inadequate. The Nucleic Acid Observatory is a good example of an effort to detect biological risks in an unbiased manner. We should be funding this and things like it.

Developing robust containment systems (such as synthetic auxotrophy) may mitigate inadvertent leaks of engineered organisms, and these systems should also be a high priority for research. However, these might provide a false sense of security because even if inactivation by mutation or recombination is prevented (which is not trivial!), whatever safeguards are engineered may be removed by a bad actor.

We should also consider the downsides of dual-use technology that could have a high potential for harm if abused. This issue has already been raised for viral engineering, but also applies more broadly to efforts to engineer fitter organisms.

Finally, we should be especially wary of any plans to intentionally release engineered organisms to globally alter ecosystems. Gene drives to eliminate specific species of mosquitoes are fine,⁴ but a synthetic Azolla event is not something I ever want to see.

1. As of fall 2024, current levels are about 424 ppm.

2. Or, I should say, R. elongatus

3. Frostpunk is an excellent video game, by the way.

4. In fact, they’re a great idea that could permanently end huge amounts of human suffering. Gene drives are also inherently safe, in that they spread relatively slowly (compared to pathogens), are detectable by sequencing, and reversible with counter-drives.