That goes well into the realm of science fiction. Metal ions are too small to be antigenic (which is a good thing, else our immune system would wreak havoc). Simple peptides generally do not have good metal binding capacities (or high specificity), usually there are additional modifications metabolization steps needed (same for artificial metal-binding peptides).
But generally speaking, in order to get rid of something via chelation, you want high water solubility and small molecular weight, so that it can be excreted via urine. Invoking microbiota seems like yet another magic wand argument, especially as we have little success even in modifying composition in a lasting way, much less modify them to specific purposes.
But I think overall the argument is backward. First you would need to figure out a compound that actually has the specificity needed to sequester specific metals but leave essentials (which are generally present at much higher concentrations) alone, and then one could think about how to administer them. Genetic manipulation is a fairly high hurdle, especially when it comes to functional manipulation of mammals (much less the ethical issues with humans) so it would be the last, not the the first step.
In fact, one could simply take a look what kind of chelators are provided to treat metal poisoning. Typical compounds use are EDTA, succimer, D-penicillamine and succimer. They are used for similar ranges of metals (most commonly lead, mercury and arsenic) and most tend to bind a rather broad range of metals with various affinity. All have side effects and are generally not commended for preventive use. So unless a new compound comes along without the noted issues, we do not to dream about bioengineering something that would provide this marvelous compound.