The Microbial Kinome

Eukaryotic protein kinases have a wealth of distant relatives in prokaryotes which share the same fold (Protein Kinase-Like or PKL) and catalytic mechanism. Only a few of these have been studied in depth, but recent genome and metagenome sequencing efforts now provide thousands of such relatives. We recently collaborated with the Venter Institute Global Ocean Survey project to predict Protein Kinase Like (PKL) genes in their metagenomic sequences and in public databases, and classified and analyzed over 45,000 PKL sequences:

Structure and function of the microbial kinome.
Natarajan, K, Taylor, S, Zhai, Y, Venter, C, Manning, G (2007)
PLoS Biol. 5(3): e17 (Press Release)

What did we find?

There is a huge diversity of PKL kinases in prokaryotes, and this diversity sheds light on the core aspects of even human protein kinases. We classified over 45,000 kinases into 20 major families, of which most human kinases are in just one family (ePK; see figure above). Several previously defined 'atypical' human kinases now fall into several other PKL families.
Several of these families are completely novel, and most are known only from a few examples. Most are ancient: two families are each found in all sequenced eukaryotes and archaeae, while several others are shared between eukaryotes and bacteria. Sequence analysis of hundreds of members of each family reveals the catalytic core and several other key features, and profiles for each family enable the annotation of thousands of previously novel proteins.
Just ten residues within the ~260-residue domains constitute most of what is conserved within each family, as well as virtually all of what is conserved between families. These include 6 previously known to be involved in catalysis, as well as four surprising new residues whose role is still unclear, but must be important in human as well as microbial kinases.
Even more surprising is that these 10 ultra-conserved residues are themselves mutable: all but one have been modified in one family or another, while still appearing to keep catalytic function. In the paper, we look at a few such cases and see that changes in one key residue are often coupled with modifications at or near other key residues, in ways that may compensate for the change. At least one of these compensations seen in an undescribed set of bacterial kinases is the same as one seen in a family of human kinases (the Wnk family).
Comparison of ePKs to all other kinases indicates reasons for their enormous success and flexibility, including the presence of the activation loop, and the flexibility of several key residues and regions.

Supplemental Data

Sequences of the 45,000 kinases, in fastA format (tarred, bzip'd archive).
A HMMer format file of 20 HMM profiles for the 20 kinase families.
Excel spreadsheet annotating key positions with the kinase domains of each of the 20 families.