Peter J. Kennelly

Inspection of the genomes for the bacteria Bacillus subtilis 168, Borrelia burgdorferi B31, Escherichia coli K-12, Haemophilus influenzae KW20, Helicobacter pylori 26695, Mycoplasma genitalium G-37, and Synechocystis sp PCC 6803 and for the archaeons Archaeoglobus fulgidus VC-16 DSM4304, Methanobacterium thermoautotrophicum delta H, and Methanococcus jannaschii DSM2661 revealed that each contains at least one ORF whose predicted product displays sequence features characteristic of eukaryote-like protein-serine/threonine/tyrosine kinases and protein-serine/threonine/tyrosine phosphatases. Orthologs for all four major protein phosphatase families (PPP, PPM, conventional PTP, and low molecular weight PTP) were present in the bacteria surveyed, but not all strains contained all types. The three archaeons surveyed lacked recognizable homologs of the PPM family of eukaryotic protein-serine/threonine phosphatases; and only two prokaryotes were found to contain ORFs for potential phosphatases from all four major families. Intriguingly, our searches revealed a potential ancestral link between the catalytic subunits of microbial arsenate reductases and the protein-tyrosine phosphatases; they share similar ligands (arsenate versus phosphate) and features of their catalytic mechanism (formation of arseno-versus phospho-cysteinyl intermediates). It appears that all prokaryotic organisms, at one time, contained the genetic information necessary to construct protein phosphorylation-dephosphorylation networks that target serine, threonine, and/or tyrosine residues on proteins. However, the potential for functional redundancy among the four protein phosphatase families has led many prokaryotic organisms to discard one, two, or three of the four.

For many years, the regulation of protein structure and function by phosphorylation and dephosphorylation was considered a relatively recent invention that arose independently in each phylogenetic domain. Over time, however, incidents of apparent domain trespass involving the presence of 'eukaryotic' protein kinases or protein phosphatases in prokaryotic organisms were reported with increasing frequency. Today, genomics has provided the means to examine the phylogenetic distribution of 'eukaryotic' protein kinases and protein phosphatases in a comprehensive and systematic manner. The results of these genome searches challenge previous conceptions concerning the origins and evolution of this versatile regulatory mechanism.

Bacteria play host to a wide range of protein phosphorylation-dephosphorylation systems (Fig. 1). As little as five years ago the known systems were thought to be late-emerging and absolutely prokaryote specific. Today we know that most protein kinases and protein phosphatases are descended from a set of common, and possibly quite ancient, prototypes. Prokaryote- and eukaryote-specific protein kinases and protein phosphatases are rare and represent exceptions, not the rule as previously thought. Commonality suggests that a dynamic and versatile regulatory mechanism was first adapted to the modulation of protein function as early if not earlier than more "basic" mechanisms such as allosterism, etc. The existence of common molecular themes confirms that the microbial world offers a unique, largely untapped library and a powerful set of tools for the understanding of a regulatory mechanism which is crucial to all organisms, tools whose diversity and experimental malleability will provide new avenues for exploring and understanding key modes of cellular regulation.

A synthetic peptide modeled after the calmodulin (CaM)-binding domain of rabbit skeletal muscle myosin light chain kinase, Lys-Arg-Arg-Trp-Lys5-Lys-Asn-Phe-Ile-Ala10-Val-Ser-Ala-Ala-+ ++Asn15-Arg-Phe-Glycyl amide (M5), inhibited the CaM-independent chymotryptic fragment of the enzyme, C35 (Edelman, A. M., Takio, K., Blumenthal, D. K., Hansen, R. S., Walsh, K. A., Titani, K., and Krebs, E. G. (1985) J. Biol. Chem. 260, 11275-11285), with a Ki of 3.2 +/- 2.1 microM. Inhibition was competitive with respect to the peptide substrate Lys-Lys-Arg-Ala-Ala5-Arg-Ala-Thr-Ser-Asn10-Val-Phe-Ala and was of the noncompetitive linear mixed type with respect to ATP. M5 and homologues with a serine residue substituted at positions 9, 13, or 14 were phosphorylated with the following order of preference: M5(Ser9) greater than M5(Ser13) much greater than M5(Ser14) greater than M5. The order of preference observed agreed with that predicted by comparison of the sequence of these peptides with the phosphorylation sites of myosin P-light chains. Both inhibition of C35 by M5 and phosphorylation of M5 and its serine-substituted homologues were severely curtailed by the addition of a stoichiometric excess of CaM over peptide. Thus, synthetic peptides modeled after the CaM-binding domain of skeletal muscle myosin light chain kinase can function as calmodulin-regulated active site-directed inhibitors of the enzyme.