Histidine kinases in signal transduction pathways of eukaryotes

Receptor coupled protein phosphorylation is a common cellular response to external signals. Protein kinases that catalyze the transfer of phosphate from ATP to serine, threonine, tyrosine or histidine residues are wide spread in eubacteria, archaebacteria and eukaryotes (Burret et al., 1991; Saier, 1993; Hunter and Cooper, 1985; Hunter, 1995). Some of these enzymes autophosphorylate while others phosphorylate specific proteins in signal transduction pathways. The histidine kinases autophosphorylate a histidine moiety in the active site and then transfer the phosphate to an aspartate moiety. Such phosphotransfer systems have been found to mediate dozens of bacterial responses and have recently been recognized in several eukaryotes (Alex and Simon, 1994; Alex et al., 1996; Ota and Varshavsky, 1993; Maeda et al., 1994; Chang et al., 1993; Hua et al., 1995; Kakimoto, 1996; Schuster et al., 1996; Wang et al., 1996; Shaulsky et al., 1996). Conserved motifs in the sequences of these protein kinases including the H motif that encompasses the phosphorylated histidine clearly show that they form a family of related proteins (Hess et al., 1988; Alex and Simon, 1994). Many but not all of the histidine kinases have two transmembrane domains separated by an extracellular loop that positions them in the plasma membrane. Their genes appear to be the descendents of an ancient protein kinase gene that used ATP and had already diverged from those encoding histidine kinases that accept phosphate from phosphoenolpyruvate (Saier, 1993). The histidine kinases are only distantly related to the serine/threonine/tyrosine protein kinases that also use ATP as the phosphate donor such as the ones found in MAP kinase cascades.

Activation of histidine kinases results in phosphorylation of histidine but the phosphate must be transferred to an aspartate before the signal is transduced. The first analyses of such pathways were caried out in bacterial systems where the histidine is found in one protein and the aspartate in another and so were called ‘two-component’ systems. These included the regulators of nitrogen metabolism, NtrB and NtrC, and the chemotactic regulators, CheA and CheY (Ninfa and Magasanik, 1986; Keener and Kustu, 1988; Hess et al., 1988). When other histidine kinases were discovered, they were sometime found to be ‘one-component’ systems with independent autokinase and a phosphate acceptor domains such as FrzE which controls motility and development in Myxococcus xanthus (Acuna et al., 1995), while at other times they were found to be multi-component systems such as the KinA, KinB, KinC, Spo0F, Spo0B, Spo0A components that are used in Bacillus subtilis to regulate the choice between growth and sporulation (Hoch, 1993). However, in all cases the phosphate ends up on an aspartate before affecting the output. In some cases the histidine and the aspartate are present in the same protein and in others they are present in different proteins. Likewise, some of these signal transduction pathways are complete when the first aspartate phosphate has been generated while in others the phosphate is passed along to another histidine before coming to its resting place on a separate aspartate. All of the eukaryotic histidine kinases that have been characterized to date have been found to have a carboxy-terminal domain with the D motif that surrounds the aspartate to which the phosphate is transferred (Chang et al., 1993; Maeda et al., 1994; Schuster et al., 1996; Wang et al., 1996; Alex et al., 1996; Kakimoto, 1996). In two cases it is now clear that this is not the end of the pathway because separate response regulators have been found.

The genome of Saccharomyces cerevisiae carries a single gene encoding a histidine kinase, SLN1, but its product is the most well characterized of all such eukaryotic enzymes (Maeda et al., 1994; Posas et al., 1996). The protein Sln1p has an extracellular sensor domain, a cytoplasmic histidine kinase catalytic domain and an aspartate relay domain. It is active under normal growth conditions and autophosphorylates his₅₇₆ from where the phosphate is relayed to asp₁₁₁₄. The phosphate is transferred to his₆₄ of Ypd1p and then passed along to asp₅₅₄ of Ssk1p (Fig. 1). Phosphorylation of Ssk1p renders it unable to activate the MAP kinase cascade that ends in Hog1p. Under conditions of high osmolarity the histidine kinase Sln1p is no longer active and Ssk1p becomes active such that the cells can continue to grow in the hyperosmotic environment. Alterations of the critical histidine or aspartate in Sln1p by site-directed mutagenesis destroy its ability to phosphorylate Ypd1p both in vitro and in vivo (Posas et al., 1996). Similar experiments showed that his₆₄ of Ypd1p and the asp₅₅₄ of Ssk1p are critical for their phosphorylation. The product of a truncated version of SLN1 in which the portion encoding the relay aspartate was deleted was shown to autophosphorylate his₅₇₆ and to be able to pass the phosphate to asp₁₁₁₄ in a construct that lacked the catalytic domain as well as the H motif. This experiment created an artificial twocomponent system out of Sln1p.

Genetic studies have shown that SLN1, YPD1, and SSK1 are all necessary to keep the MAP kinase cascade in check. The complete phosphorelay system has been selected such that the cells can grow under varying conditions of osmolarity. Using a modified version of Snl1p in which the histidine was replaced with an asparagine so that it was not phosphorylated, Posas et al. (1996) were able to show that a phosphate on asp₁₁₁₄ of Snl1p could be transferred in vitro to Ypd1p and then to Ssk1p. These experiments have shown that the purified proteins are sufficient to catalyze the phosphorelay from Sln1p to Ypd1p to Ssk1p. By comparison with the multi-component systems in Bacillus subtilis and Bordetella pertussis, Posas et al. (1996) suggest that alteration of phosphohistidine and phosphoaspartate is a universal signal transduction apparatus used in both prokayotes and eukaryotes (Fig. 1). Multistep phosphorelay allows different inputs and checks to be applied along the way. Cross-talk between partially independent pathways could integrate the response to a wide variety of conditions. Moreover, the aspartates in some systems can accept phosphate from metabolites such as acetyl-phosphate which would provide other routes into the signal transduction pathway (Lukat et al., 1992; Lukat and Stock, 1993).

Recently, two genes that encode homologs of histidine kinases have been characterized in Dictyostelium discoideum. One of these, dokA, is expressed during growth and is necessary for protection from hyperosmotic conditions (Schuster et al., 1996). The other, dhkA, is not expressed until 8 hours after the initiation of development and is necessary for proper terminal differentiation of both stalk cells and spores (Wang et al., 1996). Both of the genes encode large proteins that carry the H motif, all the motifs of the histidine kinase catalytic domain, and the D motif. Autophosphorylation of the histidine in the H motif from ATP has recently been demonstrated with DokA that was expressed in bacteria while the aspartate in the D motif has been shown to accept phosphate from acetyl-phosphate (Schuster et al., 1996; S. C. Schuster, personal communication). Site directed mutations in DhkA that change either histidine₁₃₉₆ to glutamine or aspartate₂₁₇₆ to asparagine have been found to affect its ability to function (Shaulsky, Wang and Loomis, unublished). The cognate response regulator of DhkA was discovered in a suppressor screen for strains that would sporulate in the absence of signaling from prestalk cells (Shaulsky et al., 1996). The N-terminal portion of the protein encoded by the suppresser gene, regA, is homologous to response regulators found in bacterial two-component systems while the C-terminal portion is homologous to cyclic nucleotide phosphodiesterases. The three aspartates that bind magnesium in CheY and position the acceptor aspartate for phosphorylation (Stock et al., 1993) are all present in conserved sequences. Disruption of regA in a dhkA⁻ null mutant corrects the defects in stalk formation and suppresses the block to sporulation. These genetic studies indicate that when DhkA is activated during culmination, it phosphorylates RegA and inactivates it.

RegA has been expressed in bacteria with a his₆ tag and purified on a metal column (Shaulsky, Wang and Loomis, unpublished). cAMP phosphodiesterase activity could be demonstrated with the purified RegA and the Km for cAMP was found to be 5 μM. RegA had no activity on cGMP, unlike the major cyclic nucleotide phosphodiesterase in Dictyostelium that is used during aggregation.

It appears that this signal transduction pathway functions during development of Dictyostelium to integrate terminal differentiation of prespore and prestalk cells. When culmination is initiated, prestalk cells release a signal that activates DhkA in prespore cells. Autophosphorylation of DhkA positions a phosphate on his₁₃₉₆ that is relayed to asp₂₁₇₆ before being passed to asp₂₁₂ of RegA. It is not yet clear if a histidine phosphate intervenes in the relay between the two aspartate phosphates as would be expected from the precedent in yeast and bacteria (Fig. 1).

Using the sequences of conserved motifs in bacterial histidine kinases to design PCR primers, Alex et al. (1996) amplified portions of two genes from the filamentous fungus, Neurospora crassa. One of these was used to clone and characterize the nik-1 gene. The predicted product shows homology with both the kinase and the aspartate relay domains of the other eukaryotic histidine kinases. Deletion of the nik-1 gene resulted in cells that grow well in minimal medium but have restricted colonial growth when 1 M sorbitol is added as an osmolyte. Neither aerial hyphae nor conidia are formed under these hyperosmotic conditions. It appears that osmoregulation is mediated in a variety of eukaryotic micro-organisms by histidine kinase signal transduction pathways.

There are a half dozen or more members of the histidine kinase family in the genome of the mustard plant Arabidopsis thaliana (Chang et al., 1993; Kakimoto, 1996). Several of them are involved in the response to the plant hormones ethylene or cytokinin. The ETR1 gene was first recognized among mutant seedlings that failed to respond to etheylene. The gene was shown to be dominant and mapped to chromo-some 1. Positional cloning and subsequent characterization of the gene showed that ETR1 encodes a putative membrane protein with all the conserved motifs of the histidine kinase domain as well as those of the aspartate domain. Four of the ethylene insensitive mutants were shown to have missense mutations in one or another of the transmembrane domains or in the intervening region. This region has been directly shown to bind ethylene (Schaller and Bleeker, 1995). The dominant mutations in the receptor region have been suggested to result in constitutive kinase activity that represses the ethylene response (Chang et al., 1993). Transformation of wild-type plants with the mutant gene results in ethylene insensitivity as expected (Chang et al., 1993). However, when the histidine predicted to accept the phosphate is changed to glutamine, transformation with the mutant gene still gives the ethylene insensitive phenotype (E. Meyerowitz, personal communication). This result raises doubts that ETR1 is an active histidine kinase. It is conceivable that ETR1 may have originally been a histidine kinase but has since evolved a separate function. Perhaps the kinase domain directly phosphorylates the aspartate in the D motif without forming a histidine phosphate intermediate when ethylene is bound to the extra-cellular domain.

The other well characterized member of the Arabidopsis histidine kinase family, CKI1, also has two transmembrane domains flanking a putative extracellular domain as well as all of the conserved motifs of the catalytic domain and the aspartate relay domain (Kakimoto, 1996). It was isolated from a population of mutated cells that grew in the absence of cytokinin. This plant hormone is a derivative of adenine and regulates many physiological functions in plants such as rapid proliferation, greening, stimulation of shoot production and inhibition of root production. Overexpression of CKI1 gives a phenotype characteristic of the cytokinin response even in the absence of added cytokinin (Kakimoto, 1996). It was suggested that high levels of CKI1 might increase the sensitivity of cells to endogenously produced cytokinin that is below the level of detection in untransformed cells (Kakimoto, 1996).

Since the D motif expected to be phosphorylated is found in the last 100 amino acids of both ETR1 and CKI1 and they do not appear to have downstream effector regions, it is likely that the phosphate is passed to a separate protein before the pathway is activated. However, such response regulators have yet to be discovered.

Signals in the environment are detected directly or indirectly by the N-terminal domain of the sensor kinases. As might be expected by their diverse roles, the primary sequences of these regions in the eukaryotic histidine kinases are unrelated. Likewise, the output domains of the response regulators have been extensively modified or commandeered from other gene families to suite the purposes of the specific signal transduction pathways.

The extracellular loop between the two transmembrane domains in the N terminus of Sln1p may somehow sense changes in osmolarity of the environment but it is not clear how this might be done since the osmotic response is not specific to a given class of compounds. At high osmolarity the histidine kinase activity of Sln1p is inhibited and Ssk1p accumulates in its unphosphorylated form that can activate the MAP kinase kinase kinases, Ssk2 or Ssk22. These kinases phosphorylate the MAP kinase kinase, Pbs2, thereby activating it such that it will phosphorylate the MAP kinase, Hog1p (Maeda et al., 1995). Since each of the components of this MAP kinase cascade is an enzyme that can use ATP to phosphorylate multiple downstream molecules, the initial signal can be amplified to generate an all-or-none response. Moreover, parallel pathways can feed into the MAP kinase cascade to further stimulate activation of Hog1p. The end result of this pathway is that transcription of several stress response genes is induced. Transcription of the gene encoding glycerol-3-phosphate dehydrogenase is also induced and can lead to accumulation of glycerol to counter the effects of hyperosmolarity.

Genetic evidence in Dictyostelium indicates that the DhkA/RegA pathway is used when encapsulation of prespore cells is triggered by a peptide signal secreted by prestalk cells (Shaulsky et al., 1996, Shaulsky, Wang and Loomis, unpublished). A 300 amino acid domain separates the two hydrophobic regions of DhkA that hold the histidine kinase in the membrane. This putative extracellular loop may directly bind the signalling peptide or could interact with a separate receptor to activate autophosphorylation on the histidine. Modification of the loop by an insertion of a myc epitope renders DhkA inactive while deletion of the loop and flanking transmembrane regions renders it constitutive (Shaulsky, Wang and Loomis, unpublished). Phosphorelay to RegA is likely to inhibit the cAMP phosphodiesterase activity which would be expected to lead to an increase in the internal concentration of cAMP. A cAMP dependent protein kinase (PKA) functions during terminal differentiation and can be artificially activated by adding high levels of the membrane permeant analog 8-bromo-cAMP. As predicted by the pathway leading from the extracellular signal through the DhkA/RegA pathway to PKA, adding 8-bromo-cAMP can induce sporulation in mutants that fail to produce the signal or that lack DhkA (Wang et al., 1996). Thus, PKA activation in prespore cells appears to act as a check-point to avoid sporulation before the signal is received from prestalk cells. Strains that carry mutations that inactivate the negative regulatory subunit of PKA (pkaR) as well as those that inactivate the phosphodiesterase gene regA sporulate precociously even in the absence of the peptide signal released from prestalk cells (Simon et al., 1992; Shaulsky, Wang and Loomis, unpublished). Moreover, strains in which the prestalk cells fail to produce the encapsulation signal due to inactivation of either of the prestalk genes, tagB or tagC, are able to form spores if they also carry a pkaR mutation that renders PKA activity independent of cAMP (Shaulsky, Wang and Loomis, unpublished).

A feedback loop that depends on the DhkA/RegA two-component system in prestalk cells may amplify production of the signal. There is evidence that the protein precursor of the signaling peptide must be phosphorylated by PKA to be active (C. Anjard, M. van Bemmelen, C. D. Reymond and M. Véron, personal communication). The loop starts with binding of the phosphopeptide to DhkA on the surface of prestalk cells, activation of this histidine kinase such that RegA becomes phosphorylated and thereby inactivated, a rise in internal cAMP resulting from inactivation of the phosphodiesterase, and further stimulation of phosphorylation of the precursor protein by PKA. Cleavage and secretion of the phosphopeptide by the prestalk specific TagB/C membrane complex finishes the loop (Shaulsky, Wang and Loomis, unpublished). Such positive feedback would result in a sharp rise in the prestalk signal and rapid induction of encapsulation of prespore cells as they ascend the stalk.

One of the advantages of information processing by phosphorelay is that the steps are readily reversible. Unlike kinase cascades in which phosphate enters from ATP at each step, phosphorelay between histidine and aspartate phosphates is stochiometric and involves little change in free energy. Rapid phosphorylation and dephosphorylation of proteins will result in a kinetic equilibrium that can shift in response to changes in the environment. Both kinases and phosphatases can be separately activated to affect the overall flow of information. A ligand activated kinase might act as a phosphatase in the absence of the ligand thereby amplifying the effects of the ligand. In the case of the histidine kinase pathways in which acyl-phosphates can donate phosphate to an intermediate, transfer of the phosphate to the other components might serve as a buffering mechanism. This might become particularly important in the case of convergent or branched pathways in which several different kinases affect a single response.

The prevalence of histidine kinase/response regulators in organisms from all three Kingdoms attests to the early evolution of this signal transduction mechanism. Subsequent duplication of the pertinent genes may have lead to multicomponent pathways in which the phosphate is alternatively transferred to histidines and aspartates. Such pathways are innately more adaptable to the needs of complex organisms. While histidine kinases have not yet been isolated from vertebrate tissues, there is evidence in the nucleotide sequence data banks indicating that genes encoding members of this family are present in the human genome and are likely to play significant roles.