With the right protein engineering approach, you could develop similar methods for almost any element you can think of - e.g. there are many millions of tons of

With the right protein engineering approach, you could develop similar methods for almost any element you can think of - e.g. there are many millions of tons of gold in the oceans, though at a ~1e3-1e6x lower concentration than uranium. The uranium binding peptide they're using has a remarkable affinity (7fM!?) which is about the same as the tightest natural binder we know about (biotin:streptavidin). The challenge in adapting this approach is first finding/designing proteins that can recognize the target ions, and then actually detecting whether a binding interaction is occurring. This has to be done on a very high throughput (1e6++) scale if you have hopes of evolving/optimizing your starting molecule to have higher affinity or to be more stable. Though for certain (toxic or life-essential) targets, nature may have already done the hard work and we simply have to find them. Towards deployment, the challenges are what they lay out in the article - producing lots and lots of functional protein on a suitable scaffold with every emphasis on stability and easy, low-energy recovery.