Science makes yet another super-interesting discovery that started from a "wait, that's weird" moment. One which was initially misinterpreted, too.
So a while back there was a big splash in the news about some bacteria which had been isolated from a lake that was quite high in arsenic, where said bacteria—and other creatures—were thriving despite the arsenic.
It wasn't just that the bacteria could live in an environment containing arsenic, as there were already known bacteria which "breathe" arsenate; it was that the initial tests seemed to show that the bacteria could grow in the absence of phosphate, by substituting arsenate for phosphate.
This last conclusion turned out to be incorrect. However, the bacteria could still thrive in lab environments where the phosphate to arsenate ratio got completely absurd—as long as there was a trace of phosphate there. That in itself was pretty strange, because arsenate is a deadly poison.
Part of the reason this made such a splash was that, if correct, it would dramatically rewrite the rules of biochemistry and life as we know it. It's pretty rare that a discovery of that magnitude is made in any field, and even more unlikely that it'll happen in a field as intensely studied as biochemistry. There's still a lot we don't know, but the framework is pretty solid so far. And part of what that framework says is that arsenate compounds are highly unstable in water.
For a super-quick backgrounder of phosphate, it's a molecule that life absolutely depends on. It is in the backbone of our DNA, it is in the energy carrier ATP that powers our cells, and it's built into many other parts of our cells and thus our bodies.
Here's a picture: phosphate on the left, and arsenate on the right. They look pretty similar, except for that different central atom.
Based on the structure, it looks like arsenate should be able to substitute for phosphate, much like the science-fictionally popular silicon-based life instead of carbon-based life we know: silicon, like carbon, can make four covalent bonds, and silicon is directly below carbon on the periodic table, as arsenic is directly below phosphorus and bonds with four oxygens in a tetrahedral pattern, shown above. Periodicity says that elements in a given column have a lot in common with each other, but that doesn't mean they can directly substitute. In some cases, molecules that almost-but-not-quite match a molecule critical for life can be deadly poison, precisely because of this almost-but-not-quite matching characteristic. In creatures not resistant, arsenic's similarity to phosphorus is why it's poison—and carbon monoxide's similarity to oxygen is why it's poison.
However, the bonds between arsenate and carbon are much weaker (2nd page) than those between phosphate and carbon, so while a DNA analog using arsenate could theoretically form, it would tend to fall apart again almost instantly. DNA actually sticking around and being fairly stable is kind of important.
Anyhow, enough background. On to the cool part.
Arsenate is a deadly poison to most life. This bacteria not only thrives in an arsenic-rich lake, but it can grow well in a lab where the arsenic concentration is beyond unnaturally high and the phosphate concentration is extremely low. But, it does need phosphate, and it doesn't use the arsenate.
It turns out that this bacteria not only doesn't use arsenate in building its DNA, it has an extraordinarily efficient method of rejecting arsenate: in short, it hates arsenate and gets rid of the stuff even more strictly than life which is poisoned by arsenate does.
It is so incredibly efficient at rejecting arsenate and picking out phosphate because it has a protein which wraps completely around the arsenate and phosphate and spits out arsenate because it's just a shade too big and stresses and stretches the shape of the protein, while bringing phosphate inside the cell because it's exactly the right size. This arsenate rejection protein may even be why it's arsenic resistant in the first place: the poison never makes it fully into the cell. Well, at least not until the arsenate to phosphate ratio is so absurdly high that the system is overwhelmed.
It's not alien life, but that protein can do what most of our biological molecules can't, and that's distinguish between one deadly poison and its similar essential nutrient. Knowing how this bacteria does it, I wonder if eventually some protein could be designed which would capture and segregate arsenate while leaving phosphate alone? Something like that might have potential for treating arsenic poisoning.
(Some complete and utter speculation here: the scientists have all been careful to say that arsenate bonds are very unstable in water; what about places without liquid water? What is the bond stability on, say, Titan, with its methane atmosphere? Some other planet with some other atmosphere? Wouldn't that be something to see! Given what we know at the moment, however, it's more likely that arsenic would be used for respiration, not for building a DNA analog. We already know that happens, so in an appropriate environment, arsenic respiration wouldn't be surprising.)