Life... uh... finds a way.
> Fourth paragraph:
> The microbes also do not swap their engineered DNA with natural counterparts because they no longer speak life’s shared biochemical language. “Establishing safety and security from the get-go will really enable broad and open use of engineered organisms,” says Farren Isaacs, a synthetic biologist at Yale University in New Haven, Connecticut, who led the other study.
Without knowing how critical the added genes are to the organism's metabolism or how they are integrated into its lifecycle, there is no way to tell if it is even possible for the organism to evolve an alternative in any practical time frame. Evolution does wonders over long time scales but even it is beholden to basic physics.
I think that a major difference here is that the length of a generation for bacteria is way shorter than for humans, so that it is feasible for them to evolve in unforeseen ways in an observeable amount of time.
Oh, I see. I incorrectly read the emphasis of your post as being the unlikelihood of change occurring "in a few generations", rather than on the fundamental nature of the changes.
Many single celled organisms can for example swap small segments of code between them. https://en.wikipedia.org/wiki/Horizontal_gene_transfer Thus the minimum change required for some information exchange is only transcription not survival and thus fairly simple. Sure, the original organism died, but as long as some information carries on that's what evolution is about.
A larger barrier is the value of such transcription being limited until it works. But again if there happens to be a simple way to do it then it might show up.
But humans no longer rely on only sexual reproduction, mutation, and horizontal gene transfer to evolve. We have this CRISPR/Cas9 mechanism now, which is putting us closer to doing something like flipping a known SNP--perhaps changing rs6152(G) to rs6152(A), to prevent baldness. Or we can splice the Bt gene into ourselves, so we won't ever be eaten by corn borer larvae.
In five human generations (110-160 years), humans will probably be able to replace/augment their mitochondria with captive subcellular oompa loompa biomachines. But even in 5000 human generations, we would not replace aerobic O2 metabolism with anaerobic N2 metabolism, simply because N2 is not as reactive as O2, thus is not as useful biologically in any environment where oxygen is available. It's more likely that we would add a supplementary anaerobic metabolism such that nitrogen asphyxiation results in a hibernation or torpor state, rather than death.
Complexity pays off sometimes.
[0] https://en.wikipedia.org/wiki/E._coli_long-term_evolution_ex...
The question isn't whether we can augment ourselves through advanced technology, it's whether or not we can design a microbe that is so different from existing life that we can control its reproduction indefinitely (in the face of evolution) without crippling the organism's usefulness to our civilization.