It turns out that your classic experimental trick to mimic protein phosphorylation by mutating serines and threonines to aspartate or glutamate at phosphorylation sites was not first discovered by humans. Pearlman, Serber and Ferrell argue that many phosphorylation sites in proteins evolved from negatively charged amino acid residues, which means that phosphorylation evolved to mimic the effects of glutamate and aspartate. This, of course, occurred long before human scientists discovered in 1987 that you could replace phosphorlated serines and threonines with negatively charged amino acids and still get a functional protein.
“A Mechanism for the Evolution of Phosphorylation Sites”, Samuel M. Pearlman, Zach Serber, James E. Ferrell Jr., Cell Volume 147, Issue 4, 11 November 2011, Pages 934–946
It is less intuitive how activating phosphorylations might evolve, since presumably the nonphosphorylated ancestor of the phosphoprotein would be inactive.
One possible mechanism is suggested by mutational studies of isocitrate dehydrogenase. Thorsness and Koshland showed that substituting an acidic aspartate residue for the serine phosphorylation site mimicked the phosphorylated state of the protein. The rationale behind this observation is that acidic residues are negatively charged, like pSer, pThr and pTyr (although Glu and Asp are singly charged whereas pSer, pThr, and pTyr are, nominally, doubly charged at physiological pH, and Asp and Glu are approximately isosteric with pSer and pThr (Figure 1B). This “trick” of substituting Asp or Glu for pSer or pThr turns out to be widely applicable. Although there are examples where two acidic residues are needed to mimic one phosphorylation event, as well as examples where the activity of the Asp or Glu mutant is more like that of the nonphosphorylated form than the phosphorylated form, often the Asp or Glu mutants do faithfully mimic the phosphorylated state.
Perhaps nature has been using this same trick in reverse…
The Finch and Pea 