Here we have used comparative genomic analysis to demonstrate that some phosphorylation sites have evolved from acidic amino acids. The acidic amino acids Asp and Glu were found to be present at the positions of phosphoserines more often than they were at control serines (). Additionally, other charged and polar amino acids did not show increased incidence, indicating the importance of the negative charge in this enrichment. We did not see a similar enrichment of Asp and Glu at Thr or Tyr phosphorylation sites. This may have been simply due to the smaller number of Thr and Tyr phosphorylation sites in our database, as other evidence, discussed below, supports the hypothesis that all three types of phosphorylation sites (Ser, Thr, and Tyr) can evolve and have evolved from acidic residues.
For our high-confidence set of multiple sequence alignments—those with at least 60 homologs—6.8% of the 234 phosphosites had very high (>50%) incidences of Asp or Glu at the position of the phosphosite, while only 1.8% of the 16198 control serine sites did (). This suggests that ~5% of the phosphorylation sites we examined may have evolved from acidic residues.
Phylogenetic analysis demonstrates that the switch between acidic and phosphorylatable residues occurred for some proteins at the evolutionary division between eukaryotes and prokaryotes (). In the two examples examined in detail here (eEF2 and Topo II), most or all of the bacterial and archaeal species possessed a Glu residue, and most or all of the eukaryotic species possessed a Ser or Thr residue at the position of the phosphosite. If one assumes that the split between bacteria and archaea occurred prior to the split between prokaryotes and eukaryotes, the phosphorylation site probably evolved from the acidic residue rather than vice versa. This interpretation is supported by the maximum likelihood (FASTML) reconstruction of the ancestral sequences for the eEF2 tree. Reconstruction of the ancestors for the Topo II tree is also consistent with this direction of evolution (Glu –> Thr), although we have less confidence in the rooting of this tree. The phosphosite appears to have evolved once (Topo II) or possibly twice (eEF2), and to have then been maintained over long evolutionary times.
At other times the switch between an acidic amino acid and a phosphosite occurred later in evolution. For example, it occurred at the divergence of ascomycetes from basidiomycetes fungi (~400 mya) in the case of enolase (). In the GRK and Raf families, the switch occurred early in animal evolution, with modern species retaining Asp/Glu-residues in some paralogs (GRK2/3, B-Raf) and acquiring phosphorylatable residues in others (GRK1, 4–7, A-Raf, and C-Raf). It appears that Raf gene duplication has allowed a primordial DD-containing Raf to evolve new conditional regulation. For GRKs it is not clear whether the acidic residues or phosphosites came first.
With the GRKs, it appears that two phosphorylation sites have replaced four acidic residues, consistent with the observation that sometimes a pair of acidic amino acids better mimics a phosphorylation than a single amino acid does (Strickfaden et al., 2007
). In most of the examples examined here, however, phosphorylation sites appear to have replaced single amino acids, in line with Thorsness and Koshland’s classic study (Thorsness and Koshland, 1987
Our original rationale for hypothesizing that phosphorylation sites evolved from acidic residues provided an explanation for the fact that phosphorylation sometimes activates proteins. In two of the seven examples examined in detail here (Raf, GRK), the phosphorylation is in fact known to be activating. In the other cases the functional significance of the phosphorylations remains untested. We predict that many of these will prove to be activating phosphorylations.
Finally, we found examples where a phosphorylation site appears to have become fixed as a glutamate residue. In the GRK7 protein kinases, a glutamate residue appears to have evolved from a threonine phosphosite that remains present in the GRK1, 4, 5, and 6 proteins (). A second example is provided by the CMGC family of protein kinases, where one clade of the kinases (the CDK1-like kinases) appears to have evolved a glutamate residue in the place of an activating tyrosine phosphorylation site (Table S3H
Phosphorylation sites are sometimes found in conserved, structured regions of proteins, but more frequently they are found in poorly conserved, putatively unstructured regions (Holt et al., 2009
; Moses et al., 2007
). The prevailing hypothesis is that the former type of site is more likely to be involved in complicated allosteric regulation, and the latter type of site is more likely to be involved in simpler types of regulation such as bulk electrostatic effects or the generation of phosphoepitopes (Holt et al., 2009
; Moses et al., 2007
; Pawson et al., 2001
; Serber and Ferrell, 2007
). Our original rationale () seems more applicable to well-conserved phosphorylation sites in structured regions. Indeed, despite the fact that ~60–90% of phosphorylation sites are believed to be the poorly conserved type (Holt et al., 2009
; Kurmangaliyev et al., 2011
; Moses et al., 2007
), all five of the phosphosites examined in – are evolutionarily well-conserved and present in structured regions of the proteins. This raises the interesting possibility that phosphosites that evolved from Asp/Glu residues will be more likely than average to be involved in allosteric regulation.
The crystal structures of Topo II, enolase, and Raf show that in the Asp/Glu-containing versions of these protein, the acidic residue is involved in a salt bridge with a conserved basic residue, and in the Ser/Thr/Tyr-containing versions, the salt bridge is lost, with phosphorylation having the potential to restore it (). These findings provide strong support for the rationale that motivated these studies ().
Taken together, this work provides insight into the evolution of phosphorylation, and in particular provides a rationale for how well-conserved, activating phosphorylations have evolved.