Peroxide Sensors for the Fission Yeast Stress-activated Mitogen-activated Protein Kinase Pathway

Abstract

The Schizosaccharomyces pombe stress-activated Sty1p/Spc1p mitogen-activated protein (MAP) kinase regulates gene expression through the Atf1p and Pap1p transcription factors, homologs of human ATF2 and c-Jun, respectively. Mcs4p, a response regulator protein, acts upstream of Sty1p by binding the Wak1p/Wis4p MAP kinase kinase kinase. We show that phosphorylation of Mcs4p on a conserved aspartic acid residue is required for activation of Sty1p only in response to peroxide stress. Mcs4p acts in a conserved phospho-relay system initiated by two PAS/PAC domain-containing histidine kinases, Mak2p and Mak3p. In the absence of Mak2p or Mak3p, Sty1p fails to phosphorylate the Atf1p transcription factor or induce Atf1p-dependent gene expression. As a consequence, cells lacking Mak2p and Mak3p are sensitive to peroxide attack in the absence of Prr1p, a distinct response regulator protein that functions in association with Pap1p. The Mak1p histidine kinase, which also contains PAS/PAC repeats, does not regulate Sty1p or Atf1p but is partially required for Pap1p- and Prr1p-dependent transcription. We conclude that the transcriptional response to free radical attack is initiated by at least two distinct phospho-relay pathways in fission yeast.

INTRODUCTION

The mitogen-activated protein (MAP) kinase (MAPK) signaling pathways are critical for the response of cells to changes in their environment (Marshall, 1994; Herskowitz, 1995; Waskiewicz and Cooper, 1995; Treisman, 1996). They serve to transduce signals generated at the cell surface or in the cytoplasm to the nucleus, where changes in gene expression result. In mammalian cells, multiple distinct MAP kinases have been identified, including a large subset whose members are activated by a variety of environmental stress conditions, DNA-damaging agents, inflammatory cytokines, and certain vasoactive neuropeptides (Dérijard et al., 1994; Freshney et al., 1994; Galcheva-Gargova et al., 1994; Han et al., 1994; Kyriakis et al., 1994; Lee et al., 1994; Rouse et al., 1994; Sluss et al., 1994). These stress-activated MAP kinases (SAPKs) fall into two distinct classes, termed the C-Jun N-terminal kinase (JNK) and p38 kinases, based on their sequences (Davies, 1994; Waskiewicz and Cooper, 1995). A number of transcription factors are phosphorylated in response to SAPK activation; for example, the c-Jun factor is regulated by JNK (Hibi et al., 1993; Derijard et al., 1994; Kyriakis et al., 1994) but not by p38, whereas ATF2 is phosphorylated and regulated by both JNK (Gupta et al., 1995; Livingstone et al., 1995; van Dam et al., 1995) and p38 (Raingeaud et al., 1995). Although a number of MAPK kinases (MAPKKs) and MAPKK kinases (MAPKKKs) that activate the SAPKs have been identified in mammalian cells, very little is known about how these are regulated by stress stimuli (reviewed in Ichijo, 1999; Tibbles and Woodgett, 1999). This is probably due to the multiplicity of SAPK pathways in mammalian cells and the difficulties of genetic analysis in these organisms.

Recently, a single member of the SAPK family, called Sty1p (also known as Spc1p or Phh1p), has been identified in the fission yeast Schizosaccharomyces pombe (Millar et al., 1995; Shiozaki and Russell, 1995; Kato et al., 1996). The Sty1p MAP kinase stimulates gene expression via the Atf1p and Pap1p transcription factors, homologs of human ATF2 and c-Jun, respectively, suggesting that the transcriptional targets of the S. pombe pathway are closely related to those in mammalian cells (Toda et al., 1991; Takeda et al., 1995; Wilkinson et al., 1996; Shiozaki and Russell, 1996; Gaits et al., 1998; Toone et al., 1998; Wilkinson and Millar, 1998). Stress-induced nuclear translocation of activated Sty1p causes it to bind and phosphorylate Atf1p, although the precise mechanism by which this phosphorylation induces transcriptional activation is not known (Wilkinson et al., 1996; Shiozaki and Russell, 1996; Gaits et al., 1998). Sty1p also binds to Pap1p but in this case the mechanism of activation appears to be distinct. In contrast to Atf1p, the bulk of the Pap1p protein is cytoplasmic in unstressed cells and is translocated to the nucleus only in response to an oxidative stress (Toone et al., 1998). This translocation requires Sty1p, although activation of Pap1p may not involve phosphorylation, because Pap1p is not a phosphorylation target of Sty1p in vitro. Instead, interaction of Pap1p with the nuclear export factor Crm1p is disturbed by oxidation of the cysteine residues in the C terminus of Pap1p. The role of Sty1p in this process is not understood (reviewed in Wilkinson and Millar, 1998). Nevertheless, many of the genes induced after Sty1p activation in S. pombe are similar to those induced by SAPK activation in mammals.

Sty1p is activated by a similar range of environmental stimuli to the mammalian SAPKs, including oxidative stress, UV light, DNA-damaging agents, osmotic stress, physical stresses, heat shock, and the protein synthesis inhibitor anisomycin (Millar et al., 1995; Shiozaki and Russell, 1995; Degols et al., 1996; Degols and Russell, 1997; Shieh et al., 1997, 1998). Thus, one might postulate that the mechanisms controlling SAPK activation are similar in S. pombe and mammalian cells. However, we recently demonstrated that activation of Sty1p is regulated by Mcs4p, a protein that is homologous to the Saccharomyces cerevisiae Ssk1p response regulator protein, which is similar to a number of two-component systems that function in bacteria but have yet to be identified in mammals (Stock et al., 1989; Parkinson, 1993). Ssk1p acts in a multistep phospho-relay system to control the activity of the Hog1p MAPK (Maeda et al., 1994, 1995; Posas et al., 1996; Shieh et al., 1997; Posas and Saito, 1998). Although Hog1p is structurally related to the SAPK family, it is activated only by increases in external osmolarity (Brewster et al., 1993; Schüller et al., 1994). In hypotonic medium, a transmembrane histidine kinase, Sln1p, initiates a phospho-relay system in which a phosphate group is transferred from the Sln1p response-regulator domain to a histidine residue in Ypd1p and then to a conserved aspartate residue in Ssk1p (Posas et al., 1996). In response to hypertonic stress, Sln1p is inactivated and Ssk1p is dephosphorylated. In this state, Ssk1p binds and activates one of two functionally overlapping MAPKKKs, Ssk2p and Ssk22p (Maeda et al., 1994, 1995; Posas and Saito, 1998). Ssk2p and Ssk22p kinase activation leads to the sequential phosphorylation and activation of the Pbs2p MAPKK and Hog1p MAPK (Brewster et al., 1993; Maeda et al., 1994, 1995). Activation of Hog1p leads to the induction of a number of genes whose products protect the cells in hypertonic medium.

It has been shown previously that in S. pombe, Mcs4p acts upstream of the Wak1/Wis4p and Win1p kinases, homologs of the Ssk2p and Ssk22p MAPKKKs, which in turn transmit the stress signal to the Wis1p MAPKK, the direct activator of Sty1p (Warbrick and Fantes, 1991; Millar et al., 1995; Shiozaki and Russell, 1995; Shieh et al., 1997, 1998; Samejima et al., 1998). Recently, an S. pombe homolog of Ypd1p, Mpr1p, was found to be required for the activation of Sty1p in response to an oxidative but not osmotic stress (Nguyen et al., 2000). This suggests that the Sty1p MAPK pathway may be regulated by a phospho-relay system similar to the Sln1p-Ypd1p-Ssk1p pathway in S. cerevisiae. However, the sensors that initiate this signal in S. pombe have not been identified. Moreover, because deletion of mcs4 abrogrates the activation of Sty1p in response to multiple environmental stresses, it is not clear what role Mcs4p phosphorylation may play in transmitting these signals.

In this article, we describe the role of Mcs4p in stress-induced activation of Sty1p in more detail. We present evidence that S. pombe contains at least two phospho-relay signaling systems that are specifically involved in sensing peroxide stress and that one of these directly controls the activity of the Sty1p MAPK. We discuss the implications of these results both in relation to the ability of microbial pathogens to survive free radical attack and the mechanisms controlling SAPK activation in metazoans.

MATERIALS AND METHODS

Media and General Techniques

Media and genetic methods for studying S. pombe were as described by Moreno et al. (1991). Standard DNA methods were used (Sambrook et al., 1989). Cell length measurements were made by using log-phase cells with a Nikon filar eyepiece drum micrometer at 1200× magnification. Transformations were regularly performed by the lithium acetate method (Moreno et al., 1991) or by electroporation (Prentice, 1991) with a Bio-Rad gene pulser.

To isolate DNA S. pombe cells were cultured in YES medium to stationary phase. Chromosomal DNA isolated (Moreno et al., 1991) from a 10-ml culture was dissolved in 25 ml of TE, and one-fifth was digested and subjected to electrophoresis and Southern blot hybridization. To isolate RNA, S. pombe cells were cultured in YES to exponential phase. Approximately 10 μg of total RNA was isolated (Moreno et al., 1991) and resolved by agarose gel electrophoresis before transfer to nitrocellulose for hybridization as described previously (Wilkinson et al., 1996). Probes for pyp2, gpx1, ctt1, and cdc2 were prepared as described previously (Wilkinson et al., 1996; Toone et al., 1998).

Detection of Mcs4p–Wak1p Interaction and of Activated Sty1p

The catalytic domain of wak1 was cloned by polymerase chain reaction (PCR) amplification from S. pombe genomic DNA. The 5′ oligonucleotide TAACTAGATCTATGGCTTTCTGTTAACGCAT, incorporating a BglII site (italicized), hybridized to sequences 5′ to the catalytic domain, whereas the 3′ oligonucleotide TATTAGCGGCCGCGGTCAACACTATAGTTTATTGTG, incorporating a NotI site (italicized), hybridized to sequences surrounding the TGA termination codon. The fragment generated was digested with BglII and NotI and cloned into the BglII and NotI sites of pBSSK-ura4 (Millar, unpublished data) to form pBSSK-ura4-wak1. A tandem 9-myc epitope was amplified by PCR from plasmid pC3280 (a gift of K. Nasmyth, Research Institute of Molecular Pathology, Vienna, Austria) by using the 5′ oligonucleotide GAAAAAGGGCGGCCGCATGGTTCAC and the 3′ oligonucleotide ATATATATGCGGCCGCCTTATGTCGGCATATTCGAG, which incorporate NotI sites (italicized). The resulting fragment was cloned into pBSSK-ura4-wak1 that had been digested with NotI to form pBSSK-ura4-wak1(9myc). pBSSK-ura4-wak1(9myc) was linearized with NdeI before being transformed into CH429 cells. Stable integration of the tagged wak1–9myc gene at the genomic wak1 locus was confirmed by Southern blot and PCR.

The mcs4 open reading frame (ORF) was amplified by PCR from S. pombe genomic DNA. The 5′ oligonucleotide CCCGGATCCATGCGCATTTGGTTTAAAAAAGTT incorporates a BamHI site (italicized) and hybridized to sequences near the start codon, whereas the 3′ oligonucleotide CCCGGATCCTCATCGACCGCGAAAACGGCA incorporates a BamHI site (italicized) and hybridized to sequences surrounding the TGA termination codon. The amplified fragment was digested with BamHI and cloned into the BamHI site of pDS472a (Forsburg and Sherman, 1997) to form pDS472a-mcs4, which expresses GST-Mcs4 from an attenuated version of the thiamine-repressible nmt promoter (nmt41) (Basi et al., 1993). mcs4::his7 wak1–9myc cells (HM1830) transformed with either pDS472a or pDS472a-mcs4 were grown in minimal medium lacking leucine and thiamine for at least 24 h. Pelleted cells were lysed by using glass beads into lysis buffer (0.5% NP-40, 150 mM NaCl, 50 mM NaF, 10% glycerol, 2 mM Na-orthovanadate, 10 mM β-mercaptoethanol, 10 μg/ml aprotonin, 10 μg/ml benzamidine, 2 mM phenylmethylsulfonyl fluoride, 10 μg/ml pepstatin A, 10 μg/ml leupeptin, 50 mM HEPES; pH 7.4) and proteins were partially purified by affinity precipitation on glutathione-agarose beads. Precipitated proteins were eluted in the presence of 10 mM reduced glutathione, resolved by SDS-PAGE, and blotted to nitrocellulose membranes. Membranes were probed with a monoclonal antibody to the myc epitope (9E10; Covance, Richmond, CA) according to the manufacturer's instructions. Phosphorylated and activated Sty1p was detected by Western blot exactly as described previously (Shieh et al., 1997; Gaits et al., 1998).

Construction of mcs4::his7 and mcs4(D412N) Alleles

mcs4 was disrupted with his7 as follows. First, the EcoRI/SphI fragment of pURB1-Mcs4-1 (Shieh et al., 1997) was cloned into pT7T318u (Amersham-Pharmacia, Little Chalfont, Buckinghamshire, United Kingdom) to create pT7T318u-mcs4. Next, a 642-bp XbaI/KpnI fragment of pT7T318u-mcs4 was replaced with an XbaI/KpnI fragment from pEA2 (a gift from C. Hoffman, Boston College, MA) that carries his7. The mcs4::his7 fragment was then excised from pT7T318u-mcs4::his7 by digestion with EcoRI and SphI, gel purified, and transformed into CH429 cells. Haploid disruptants were isolated on minimal medium lacking histidine and stable integrants were confirmed by Southern blot.

The wild-type mcs4 genomic locus was replaced with mcs4(D412N) as follows. A 1.56-kb fragment carrying the mcs4(D412N) allele was excised from pREP41HM-Mcs4(D412N) (Shieh et al., 1997) by using NdeI and BamHI and cotransformed with plasmid pIRT2(LEU2) (Basi et al., 1993) into a mcs4::ura4 haploid (Shieh et al., 1997). Transformants were selected on minimal medium lacking leucine and then replica-plated onto complete minimal medium containing 5-fluoro-orotic acid to select for uracil auxotrophs. Genomic DNA from these strains was subjected to PCR amplification by using the oligonucleotides Mcs4-N1 and Mcs4-C1 (Shieh et al., 1997), which generated a single 1.56-kb band in strains carrying the mcs4(D412N) allele. Successful gene replacement was confirmed by sequencing of PCR-amplified DNA.

Isolation and Sequencing of mak1

PCR amplification was performed on genomic S. pombe DNA by using the 5′ oligonucleotide CTI(A/G/C)TIGTIGA(G/A)GA(C/T)(A/G/C)A and the 3′ oligonucleotide GGCAT(C/T)TGI(A/C)I(A/G)TCCATIA, where I is inosine. The reaction was performed for 30 cycles at an annealing temperature of 40°C. The PCR products were separated by PAGE and a 160-bp fragment was purified and ligated directly into pCRII (Invitrogen, San Diego, CA) before being transformed into Escherichia coli DH5α. Plasmid inserts from nine resulting colonies were sequenced by using M13 and T7 primers. Six were found to contain the same sequence, which has homology to the conserved region of all response regulators. The160-bp fragment was then used to screen a pURB1 genomic library (Barbet et al., 1992) and three overlapping clones were identified. Sequencing of the largest clone, pURB1-mak1, was performed by using custom-synthesized oligonucleotide primers, after constructing unidirectional deletions by using exonuclease III and S1 nuclease (Henikoff, 1984).

Sequence Alignments

GAF domain alignments were identified by using PSI-BLAST (Altschul et al., 1997) and Mak3p as the query sequence. Reiterative searching identified Mak2p with an e-value of 2e-06 after round 1. Round 4 picked up the C.a. HK1 with an e-value of 1e-37. Round 5 picked up E. coli FhlA with an e-value of 5e-05. Round 7 picked up the ethylene receptor (Etr1) with an e-value of 2e-11. The PSI-BLAST converged before the PHYE sequences were picked up with e-values below the threshold. The program THREADER (Jones et al., 1992) was run on the putative PAS/PAC domains Mak1a, Mak1b, Mak2, and Mak3. The top hits with high p were either the FixL or the ERG potassium channel PAS/PAC domains. The serine/threonine kinase sequences could also be picked up using PSI-BLAST with low e-values (<0.05) and columns of sequence identity can be seen in the alignment.

Deletion of mak1, mak2, mak3, and prr1 Sequences

To delete mak1 sequences, the HindIII site in pBluescript II (Stratagene, La Jolla, CA) was first destroyed by digesting with HindIII, filling the 5′ overhangs with Klenow polymerase and religating to form pBSSK*. Next, pURB1-mak1 was digested with BamHI and PstI, and the resulting 4.0-kb fragment carrying 60% of the mak1 ORF was ligated into pBSSK* that had been linearized with BamHI and PstI, thus forming pBSSK*-mak1. pBSSK*-mak1 was digested with HindIII to remove ∼2.0 kb of the mak1 ORF, including sequences critical for histidine kinase function, which were replaced with a 1.6-kb HindIII fragment bearing the ura4 marker from pREP42 (Basi et al., 1993). The resulting construct, pBSSK-mak1::ura4, was digested with BamHI and PstI and transformed into a leu1-32/leu1-32 ura4-D18/ura4-D18 ade6-210/ade6-216 his7-366/his7-366 h−/h+ diploid strain (JM1429) (Table 1). Transformants were selected on minimal medium lacking uracil and stable heterozygous diploids were sporulated. Haploid colonies segregated 2:2 with respect to uracil prototrophy, indicating that mak1 is a nonessential gene. Disruptants were confirmed by Southern blot. A similar strategy was used to disrupt mak1 sequences with the LEU2 marker.

Table 1.

Strains used in this study

Strain	Genotype	Source
CH429	h−	C. Hoffman
HM1830	mcs4::his7 wak1-9myc: ura4 h−	This study
VB1692	mcs4 (D412N) h−	This study
JM2280	mcs4 (D412N) sty1::ura4 h−	This study
JM1468	mcs4::his7 h+	This study
KZ1535	mak1::ura4 h+	This study
KZ1536	mak1::LEU2 h+	This study
VB1530	mak2::ura4 h−	This study
VB1834	mak2::LEU2 h−	This study
VB1932	mak3::kanR h−	This study
VB1930	mak2::LEU2 mak3::kanR h−	This study
VB1971	mak1::LEU2 mak2::ura4 mak3::kanR h+	This study
VB1925	mak1::ura4 mak2::LEU2 mak3::kanR mcs4::his7 h−	This study
JM1521	sty1(HA6His):ura4 h+	Shieh et al., 1997
VB1700	sty1(HA6His):ura4 mcs4 (D412N) h−	This study
VB1828	sty1(HA6His):ura4 mak1::LEU2 h+	This study
VB1829	sty1(HA6His):ura4 mak2::LEU2 h−	This study
VB1934	sty1(HA6His):ura4 mak3::kanR h−	This study
KS1479	atf1(HA6His):ura4 h−	Shiozaki and Russell, 1995
TS2029	atf1(HA6His):ura4 mak1::LEU2 h+	This study
TS2030	atf1(HA6His):ura4 mak2::LEU2 h−	This study
TS2031	atf1(HA6His):ura4 mak3::kanR h−	This study
JM1516	prr1::his7 h−	This study
JM1546	prr1::his7 atf1::ura4 h−	This study
JQ51	prr1::his7 mak2::LEU2 mak3::kanR h−	This study
JM1429	h+/h−	This study

All strains are leu1-32 ura4-D18 ade6-M210 (or ade6-M216) his7-366. 

To delete mak2 sequences, a 8.7-kb PvuII fragment carrying 98% of the mak2 ORF was first excised from cosmid clone c27E2 (Ressourcenzentrum, Berlin, Germany) and ligated into the large PvuII fragment of pBSSK to create pBSSK-mak2. pBSSK-mak2 was digested with HindIII to remove 4.6 kb (67%) of the mak2 ORF, including the histidine kinase domain, which was replaced with a 2.2-kb HindIII fragment bearing the LEU2 marker from pREP41 (Basi et al., 1993) to create pBSSK-mak2::LEU2. The resulting construct, pBSSK-mak2::LEU2, was digested with PvuII and the resulting 6.8-kb fragment was transformed into the diploid strain JM1429. Haploid mak2 disruptants were isolated as described above for mak1. A similar strategy was used to disrupt mak2 sequences with ura4.

To delete mak3 sequences, the kanamycin-resistance gene was amplified from pFA6a-kanMX6 (Bahler et al., 1998) by PCR with the 5′ oligonucleotide GTTCTTCCTTATGAAAATTGAAGTTCAATA-ACTTATAACATGGGAGTAAATGATAATTAACAATTGTTGCC-CGGATCCCCGGGTTAATTAA, which is homologous both to sequences immediately upstream of the mak3 ORF and to the 5′ region of the kanMX6 cassette (italics), and the 3′ oligonucleotide TGCCAAAATGTAACGAGCTCTTCTATGGTTTAGCATAGACATTC-ACGAGATTTTCCAAGAAAACAAAACAGAATTCGAGCTCGTT-TAAAC, which is homologous both to sequences downstream of the mak3 termination codon and to the 3′ region of the kanMX6 cassette (italics). The resulting fragment was transformed into CH429 cells and G418-resistant colonies were isolated as described (Bahler et al., 1998). mak3::kanR haploid deletion strains were confirmed by PCR and Southern blot.

To delete prr1 sequences, part of the prr1 gene was amplified by PCR from cosmid SPAC8D9 (Ressourcenzentrum) by using the 5′ oligonucleotide Prr1.1, GTGTTGGATGTAAGAATTCCTTGAAAA-TTTGTTTGCTGG and the 3′ oligonucleotide Prr1.2, AAAGTAGGTAACTGCAGAATTCATCGATATATTTGAAGTTATAGC (EcoRI sites in italics). The resulting 2.2-kb fragment was digested with EcoRI and ligated into pBR322 digested with the same enzyme to form pBR-prr1. pBR-prr1 was then digested with NsiI and XbaI and a PstI/XbaI fragment containing his7 was isolated from pBSSK-his7(R) (Millar, unpublished data) and cloned into these sites to generate pBR:prr1::his7. This cloning removes codons 67 to 209 of the prr1 ORF. pBR:prr1::his7 was digested with EcoRI and transformed into CH429 cells. Stable histidine prototrophs were selected and disruption of prr1 confirmed by PCR.

RESULTS

Phosphorylation of Mcs4p Controls Sty1p Kinase Only in Response to Peroxide

Previous genetic data indicate that the Mcs4p cell cycle regulator acts upstream of two functionally redundant MAPKKKs, Wak1p/Wis4p and Win1p, to control the activity of the Sty1p MAP kinase (Shiozaki et al., 1997; Shieh et al., 1998; Samejima et al., 1998). To examine whether Mcs4p interacts with either of these MAPKKKs, a GST-Mcs4p fusion protein was expressed in cells lacking endogenous Mcs4p but expressing Wak1p MAP fused to a 9-myc epitope tag. Proteins associating with GST-Mcs4p were identified after affinity precipitation from cell extracts. Wak1p associated with GST-Mcs4p but not glutathione S-transferase (GST) itself, even in unstressed cells (Figure 1A). Binding of Mcs4p to Wak1p was not altered when cells were challenged with various environmental stresses (our unpublished data). This result indicates that Mcs4p associates with Wak1p and is thus an integral component of the Sty1p MAPK complex. In this respect, Mcs4p acts in a similar manner to the Ssk1p response regulator in S. cerevisiae, which binds directly to the N termini of the Ssk2p and Ssk22p MAPKKKs.

Figure 1.

Mcs4p is required for activation of Sty1p MAPK specifically in response to peroxide stress. (A) Mcs4p interacts with the Wak1p MAPKKK. Cells expressing Wak1p-9myc and either GST or GST-Mcs4 were grown and extracts were prepared and processed as described in MATERIALS AND METHODS. Proteins in the extract or those purified on glutathione sepharose beads were analyzed by SDS-PAGE and Western blotting by using a monoclonal antibody to the myc epitope. (B) D412 of Mcs4p is required specifically for activation of the Sty1p MAPK in response to peroxide stress. Wild-type (JM1521) or mcs4(D412N) cells (VB1700) bearing an integrated, six-histidine and hemagglutinin-tagged version of Sty1p growing in YEPDA (2% bacto-peptone, 1% yeast extract, 2% glucose, 100 mg/l adenine) medium at 30°C were incubated for the times indicated in the same medium containing 1 mM sodium arsenite, 3 μg/ml 4-nitroquinoline oxide (4NQO), 1 mM hydrogen peroxide (H₂O₂), 8 mM paraquat or 0.5 M NaCl, or subjected to a mild heat shock at 42°C. Approximately 2 × 10⁸ cells were harvested at each time point and lysed under nondetaturing conditions before the Sty1p protein was precipitated by using Ni²⁺-nitriloacetic acid agarose. Precipitated proteins were separated by SDS-PAGE and Western blots were probed for the presence of phosphorylated Sty1p by using anti-phospho-p38 antibodies (see MATERIALS AND METHODS). (C) Dose response for activation of Sty1p in response to an oxidative or osmotic stress. Wild-type (JM1521) or mcs4(D412N) cells (VB1700) growing in YEPD at 30°C were incubated for 10 min in the same medium containing increasing doses of either hydrogen peroxide or NaCl. After this time, cells were harvested and phosphorylation of Sty1p was measured as described above.

In S. cerevisiae, inactivation of the Sln1p histidine kinase in response to changes in external osmolarity results in dephosphorylation of Ssk1p on a conserved aspartate residue (D554) in the response-regulator domain (Maeda et al., 1994). The equivalent residue in Mcs4p is aspartate D412. In a previous article, we reported that deleting mcs4 prevented activation of Sty1p in response to multiple environmental insults (Shieh et al., 1997). Because the total absence of a protein could disrupt the integrity of multiprotein complexes, we reexamined the role of Mcs4p with respect to signal transmission via phosphorylation of D412 by constructing a strain in which mcs4⁺ is replaced by a mutant allele bearing a nonphosphorylatable asparagine at residue 412. These mcs4(D412N) cells were viable but divided at a smaller size (11.4 ± 0.4 μm) relative to wild-type cells (14.3 ± 0.2 μm), although growth rate was unaffected (Figure 2B; our unpublished data). This phenotype is similar to that of cells lacking the Pyp1p MAP kinase phosphatase, suggesting that it is due to partial activation of Sty1p (Millar et al., 1992). Consistent with this we observed that mcs4(D412N) sty1::ura4 cells divided at 22.9 ± 1.8 μm, a size similar to that of sty1::ura4 cells (23.1 ± 2.0 μm).

Figure 2.

Identification of three nonessential histidine kinases in fission yeast. (A) Diagrammatic representation of the domain architecture of S. pombe (S.p.) Mak1p, Mak2p, and Mak3p compared with the S. cerevisiae (S.c.) Sln1p and C. albicans (C.a.) HK1 histidine kinases. The positions of the following domains are shown in these molecules: putative serine/threonine kinase domain (diagonally hatched bar), GAF domain (gray bar), PAS/PAC motifs (open bar), histidine kinase (black bar), and response-regulator domain (vertically hatched bar). The Mak1p, Mak2p, and Mak3p sequences appear in the SWISSPROT and EMBL databases under the accession numbers AL157734, Z98978, and AL031543, respectively. (B) S. pombe histidine kinases control cell size at cell division via Mcs4p. Wild-type (mcs4⁺) (JM 1059), mcs4(D412N) (VB1692), Δmcs4 (JM1468), Δmak1 Δmak2 Δmak3 (VB1971), or Δmak1 Δmak2 Δmak3 Δmcs4 (VB1925) cells were grown to log phase in minimal medium and photographed by using Normarski optics.

Wild-type or mcs4(D412N) strains bearing a six-histidine and hemagglutinin-tagged sty1 allele were subjected to various stresses, including 1 mM sodium arsenite, 3 μg/ml 4-nitroquinoline oxide, 1 mM hydrogen peroxide, 8 mM paraquat, 0.5 M NaCl, or subjected to a mild heat shock at 42°C. Phosphorylation of Sty1p was monitored by Western blot with an antibody that recognizes only the phosphorylated and, by inference, activated form of Sty1p (see MATERIALS AND METHODS). With the sole exception of hydrogen peroxide, which failed to activate Sty1p in a mcs4(D412N) background, phosphorylation of Sty1p was observed after all other stresses in both the wild-type and mcs4(D412N) cells (Figure 1B). To confirm this result, wild-type or mcs4(D412N) cells were subjected to increasing doses of either NaCl or hydrogen peroxide and the phosphorylation of Sty1p was monitored after 10 min of incubation. With NaCl, no difference was observed between wild-type and mcs4(D412N) cells in the activation of Sty1p, indicating that phosphorylation of Mcs4p D412 is not required for activation of Sty1p by hypertonic stress (Figure 1C). In contrast, activation of Sty1p by hydrogen peroxide was almost completely abolished in mcs4(D412N) cells irrespective of the dose (Figure 1C), suggesting that phosphorylation of Mcs4p D412 is required only for the activation of Sty1p MAP kinase to peroxide stress. This contrasts sharply with the results obtained with mcs4-deletion cells, suggesting either that Mcs4p is required for the structural integrity of the Sty1p MAP kinase complex or that modifications other than D412 phosphorylation of Mcs4p are required for signal transmission in response to other environmental stimuli.

Identification of a Family of Histidine Kinases in Fission Yeast

We next sought the putative peroxide receptor(s) that might control Mcs4p phosphorylation, postulating that they might be structurally related to Sln1p, which contains both histidine-kinase and response-regulator domains. Thus, degenerate primers were designed to regions corresponding to the amino acid sequences L(I/V/L)VEDN and (L/F)MD(I/L/V)QMP and used in low-temperature PCR amplification of S. pombe genomic DNA. Six of nine individual PCR clones contained the same nucleotide sequence, which translated to a polypeptide with homology to response regulators. One of these was used to isolate a full-length clone from a genomic library. By this means, an uninterrupted 4.8-kb ORF was identified that gives rise to a predicted protein of 1615 amino acids. The protein was found to contain, in addition to a response-regulator domain, a region most closely related to members of the histidine-kinase family (Figure 2A). We termed this gene mak1, for Mcs4-associated kinase. By screening the S. pombe genome sequence for similar sequences, we identified two additional histidine-kinase–like ORFs. These intronless genes translate to predicted proteins of 2310 and 2344 amino acids, which we designated Mak2p and Mak3p, respectively (Figure 2A). Comparison of these proteins revealed a high degree of homology to each other throughout, including both histidine-kinase and response-regulator domains in the C termini. The Mak1p, Mak2p, and Mak3p histidine kinases contain several additional domains in their N termini that are absent in Sln1p. Adjacent to the histidine-kinase domain are two (in the case of Mak1p) or one (in the case of Mak2p and Mak3p) copy of a PAS/PAC domain (Crews and Fan, 1999), which is found throughout the evolutionary spectrum in proteins associated with light reception and regulation (plant phytochromes), oxygen or redox (E. coli Aer, ArcB, and FixL), oxygen-regulated transcription (human HIF1α), and regulation of circadian rhythms (such as mouse PER and CLOCK and Drosophila SIM) (Figures 2A and 3A). Adjacent to this region in Mak2p and Mak3p, but not Mak1p, is a GAF domain, which has also been identified in phytochromes (such as Arabidopsis PhyE), redox-regulated transcription factors (E. coli FhlA), and plant ethylene receptors (e.g., Arabidopsis Etr1) (Aravind and Ponting, 1997) (Figures 2A and 3B). Finally, at the extreme N termini of Mak2p and Mak3p, we identified a domain that has considerable homology to a family of atypical serine/threonine kinases from prokaryotes, in particular Mycobacterium tuberculosis PknB (Figures 2A and 3C). Although this domain is absent in Mak1p and Sln1p, a similar domain is found in the Candida albicans HK1 histidine kinase, which has a similar overall domain architecture to S. pombe Mak2p and Mak3p (Figures 2A and 3C).

Figure 3.

Sequence alignment of Mak1p, Mak2p, and Mak3p. (A) Alignment of two PAS/PAC domains of S. pombe (S.p.) Mak1p (a and b) with similar domains in S. pombe (S.p.) Mak2p and Mak3p, Mus musculus (M.m.) PER3 and CLOCK, D. melanogaster (D.m.) SIM, human (H.s.) HIF1α, Rhizobium meliloti (R.m.) FixL, and E. coli (E.c.) Aer proteins. Homologies were identified by reiterative PSI-BLAST searches. Amino acid similarities are represented in color: yellow for proline, turquoise for hydrophobic residues (I, L, V, F, M, C), purple for acidic residues (E, D), red for basic residues (K, R), dark blue for aromatic residues (H, Y), green for polar residues (Q, S, T, N), and orange for glycine. Sequences are colored automatically on the basis of an alignment consensus that is calculated automatically by the CLUSTAL program. Residues that are completely conserved (*), highly conserved (:) or mostly conserved (.) are indicated. (B) Alignment of the GAF domain in S. pombe (S.p.) Mak2p and Mak3p with those in C. albicans (C.a.) HK1, Arabidopsis thaliana (A.t.) Etr1 and PhyE, E. coli (E.c.) FhlA, and Pharbitus nil (P.n.) and Pinus sylvestris (P.s.) PhyE proteins. Homologies are displayed as in A. (C) Atypical serine/threonine-kinase domains of S. pombe (S.p.) Mak2p and Mak3p, C. albicans (C.a.) HK1, M. tuberculosis (M.t.) PknB, and two uncharacterized proteins from Streptomyces coelicolor (St. c.; CAB45215) and Lactococcus lactis (L.l.; CAA10713). Homologies are displayed as in A.

To investigate the cellular functions of mak1, mak2, and mak3, internal sequences were replaced with auxotrophic markers. Disruption of mak1 with ura4 gave rise to viable haploids that divided at a marginally (∼10%) smaller size than wild-type (our unpublished data). mak2::LEU2 and mak3::kanR haploid strains were also viable but had no discernible growth or cell-size defect. We also made doubly and triply deleted strains. The triple-deletion cells were viable and divided at 11.2 ± 0.5 μm (as opposed to 14.3 ± 0.2 μm for wild-type cells) and were phenotypically indistinguishable from mcs4(D412N) cells (Figure 2B). All double deletion strains divided at an intermediate size (our unpublished data). To test whether the Mak histidine kinases may function through Mcs4p, we constructed a mak1::ura4 mak2::LEU2 mak3::kanR mcs4::his7 quadruple mutant. These cells divided at 19.3 ± 1.2 μm, a size similar to that of the mcs4::his7 single mutant (19.1 ± 1.4 μm) (Figure 2B). This result suggests that the Mak1p, Mak2p, and Mak3p histidine kinases act upstream of Mcs4p.

Mak2p and Mak3p Regulate Sty1p Activity in Response to Peroxide

To determine whether any of the three histidine kinases is involved in signaling to Sty1p, wild-type cells or cells individually disrupted for mak1, mak2, or mak3 were incubated in the presence of 1 mM hydrogen peroxide and phosphorylation of Sty1p was monitored after various lengths of time. Whereas a rapid increase in Sty1p phosphorylation was observed in both wild-type and Δmak1 cells, the response to peroxide was almost completely absent in Δmak2 or Δmak3 cells (Figure 4A). The Sty1p MAP kinase is cytoplasmic in unstressed cells and translocates to the nucleus, where it associates with and phosphorylates the Atf1p transcription factor, only in response to stress (Shiozaki and Russell, 1996; Wilkinson et al., 1996; Gaits et al., 1998). Phosphorylation of Atf1p can be monitored by a mobility shift on SDS-PAGE (Shiozaki and Russell, 1996). Cells expressing a six-histidine and hemagglutinin-tagged atf1 allele (KS1479) or derivatives independently deleted for mak1 (TS2029), mak2 (TS2030), or mak3 (TS2031) were incubated in the presence of 1 mM hydrogen peroxide for 5 min and phosphorylation of Atf1p was monitored by SDS-PAGE. Although a rapid hyperphosphorylation of Atf1p was observed in both wild-type and Δmak1 cells, no change in Atf1p mobility was observed in Δmak2 or Δmak3 cells (Figure 4B). These results imply that Sty1p is not fully phosphorylated, activated, or translocated to the nucleus in response to peroxide stress in the absence of Mak2p or Mak3p. Given the similarity to the Sln1p-Ypd1p-Ssk1p pathway in S. cerevisiae, we infer that Mak2p and Mak3p are responsible for the phosphorylation of Mcs4p (presumably via Mpr1p) and consequently the activity of the Sty1p MAP kinase.

Figure 4.

Mak2p and Mak3p are required for Sty1p activation in response to peroxide stress. (A) Mak2p and Mak3p are required for Sty1p phosphorylation. Log-phase cultures of wild-type (JM1521), Δmak1 (VB1828), Δmak2 (VB1829), or Δmak3 (VB1934) cells bearing a hemagglutinin- and six-histidine–tagged version of Sty1p growing in YEPD medium at 30°C were incubated in the same medium containing 1 mM hydrogen peroxide for the times indicated. Approximately 2 × 10⁸ cells were harvested at each time point and phosphorylation of Sty1p was measured as described in MATERIALS AND METHODS. (B) Mak2p and Mak3p are required for Sty1p-dependent Atf1p phosphorylation. Log-phase cultures of wild-type (KS1479), Δmak1 (TS2029), Δmak2 (TS2030), or Δmak3 (TS2031) cells bearing a six-histidine– and hemagglutinin-tagged version of Atf1p growing in YEPD medium at 30°C were incubated in the same medium containing 1 mM hydrogen peroxide for 5 min. Atf1p was precipitated from each sample using Ni²⁺-nitriloacetic acid agarose and analyzed by probing Western blots with the 12CA5 anti-hemagglutinin antibody.

Mak1p, Mak2p, and Mak3p Control the Transcriptional Response to Peroxide

In S. pombe, a number of genes, including those encoding the tyrosine-specific MAP kinase phosphatase (pyp2) and glutathione peroxidase (gpx1), are induced in response to peroxide in a Sty1p- and Atf1p-dependent manner (Millar et al., 1995; Degols et al., 1996; Wilkinson et al., 1996; Toone et al., 1998; Yamada et al., 1999). Induction of other genes such as those for thioredoxin reductase (trr1) and catalase (ctt1) are also dependent on Sty1p but require one of two distinct transcription factors, Pap1p and Prr1p (Toone et al., 1998; Ohmiya et al., 1999). To determine whether peroxide-induced gene expression is affected by loss of one or more of the Mak histidine kinases, wild-type cells or those deleted for mak1, mak2, or mak3 were incubated in the presence of 1 mM hydrogen peroxide and the levels of various transcripts were assessed by Northern blot analysis. Induction of both pyp2 and gpx1 was severely reduced in Δmak2 and Δmak3 cells but not in Δmak1 cells (Figure 5). Conversely, induction of ctt1 was significantly reduced in Δmak1 cells but unaffected in Δmak2 and Δmak3 cells (Figure 5). These results indicate that Mak2p and Mak3p are required for Atf1p-dependent, but not Pap1p/Prr1p-dependent, gene expression in response to peroxide stress. This is presumably because Sty1p is not activated and therefore unable to phosphorylate and activate Atf1p. On the other hand, Mak1p appears to be required for effective transcription through a Pap1p- or Prr1p-dependent pathway. Deletion of mak1, mak2, or mak3 had no effect on the activation of Sty1p or induction of gene expression in response to all other stresses tested (our unpublished data). We conclude that at least two distinct phospho-relay systems control the SAPK-dependent transcriptional response to peroxide stress in S. pombe.

Figure 5.

Mak2p and Mak3p control Atf1p-dependent gene transcription in response to peroxide stress. Log-phase cultures of wild-type (CH429), Δmak1 (KZ1535), Δmak2 (VB1530), or Δmak3 (VB1932) cells growing in YES at 30°C were incubated in the presence of 1 mM hydrogen peroxide for the times indicated. Total RNA was extracted and equal quantities were separated by electrophoresis. Northern blots were then probed sequentially by using DNAs specific to the pyp2, gpx1, ctt1, and cdc2 genes.

Two Phospho-relay Systems Are Required for Protection against Free Radical Attack

Previous studies have shown that the Atf1p and Pap1p transcription factors control distinct but complementary roles in the protection of S. pombe cells from free-radical attack. Cells lacking both Atf1p and Pap1p are more sensitive to peroxide attack than cells lacking either protein alone (Nguyen et al., 2000). In S. cerevisiae, Yap1p (the Pap1p homolog) plays a functionally overlapping role with the response-regulator protein Skn7p in the transcriptional response to peroxide stress (Morgan et al., 1997). Because Prr1p, a homolog of S. cerevisiae Skn7p, is also required for peroxide-induced catalase expression in S. pombe, Prr1p and Pap1p may share a similar relationship (Ohmiya et al., 1999). To determine the role of various two-component signaling molecules in the protection of S. pombe from peroxide attack, wild-type cells or cells lacking Atf1p or Prr1p were tested for their ability to survive exposure to an acute dose of hydrogen peroxide either in the presence or absence of the Mak2p and Mak3p histidine kinases. Whereas cells lacking Atf1p alone showed little sensitivity to peroxide (our unpublished data), cells lacking Prr1p were more sensitive than wild-type but not as sensitive as cells lacking both Atf1p and Prr1p (Figure 6). These results suggest that Atf1p and Prr1p control distinct aspects of the response to peroxide in S. pombe. Because Mak2p and Mak3p appear to control Atf1p-dependent gene expression, we next tested the sensitivity to peroxide of cells lacking Mak2p or Mak3p. We found that cells lacking both Mak2p and Mak3p were no more sensitive to peroxide than wild-type when Prr1p was present but displayed increased sensitivity when Prr1p was absent (Figure 6). These results suggest that at least two distinct phospho-relay pathways contribute to the protection of S. pombe cells from peroxide attack. The role of Mak1p in this process will be dealt with in detail in a subsequent study.

Figure 6.

Mak2p and Mak3p protect cells against peroxide attack. Exponentially growing cultures of wild-type (CHP429), Δprr1 (JM1516), Δprr1 Δatf1 (JM1546), Δmak2 Δmak3 (JM1930), and Δmak2 Δmak3 Δprr1 (JM2232) cells growing in YES medium at 30°C were incubated in the same medium containing 50 mM hydrogen peroxide. Cells were removed at the times indicated, washed, and plated on YES agar plates to determine the number of viable cells. The data shown represent the mean values from three independent experiments.

DISCUSSION

As reported previously, we and others have identified Mcs4p as an upstream regulator of the S. pombe Sty1p MAPK pathway (Shiozaki et al., 1997; Shieh et al., 1997). Here we show that Mcs4p associates with Wak1p, one of two MAPKKKs that control phosphorylation of Sty1p. Thus, Mcs4p acts in a manner similar to that of the S cerevisiae Ssk1p response-regulator protein, which binds the Ssk2p and Ssk22p MAPKKKs to regulate the Hog1p MAP kinase (Posas and Saito, 1998). The activity of Ssk1p is controlled by phosphorylation on aspartate 554, an event that is initiated by a transmembrane histidine kinase, Sln1p. Phosphotransfer between Sln1p and Ssk1p requires phosphorylation of an intermediate protein, Ypd1p, on histidine 64. In this article, we report that phosphorylation of the equivalent residue in Mcs4p (D412) is required for activation of the Sty1p MAP kinase, but only in response to hydrogen peroxide and not to a battery of other environmental insults, including osmotic stress. Surprisingly, paraquat, which induces superoxide radicals and, as a consequence, the production of hydrogen peroxide, still activates Sty1p in mcs4(D412N) cells. We conclude that either the superoxide radical is sensed by a distinct mechanism or, at the concentration used, paraquat activates stress signals other than the overproduction of peroxide. Similarly, heavy metal toxicity, also regarded as an oxidative stress, appears to require a distinct pathway to activate Sty1p that does not involve phosphorylation of Mcs4p. Together, these results suggest that Mcs4p acts in a two-component system that specifically senses peroxide stress. Indeed, Nguyen et al. (2000) have recently shown that an S. pombe homolog of S. cerevisiae Ypd1p, Mpr1p, is also required for activation of Sty1p by peroxide and that this is mediated through a conserved histidine residue. In this article, we present evidence that deletion of either the Mak2p or Mak3p histidine kinase, but not Mak1p, prevents phosphorylation of Sty1p in response to hydrogen peroxide. As a consequence, Sty1p is not activated as judged by the absence of Atf1p phosphorylation and failure of Atf1-dependent gene expression. These results strongly suggest that Sty1p is controlled by a phospho-relay system that is similar to the Sln1p-Ypd1p-Ssk1p pathway in S. cerevisiae, but which is initiated by two histidine kinases, Mak2p and Mak3p (see model in Figure 7).

Figure 7.

Regulation of the Sty1p MAP kinase cascade by a two-component peroxide receptor. We propose that peroxide stress regulates the activity of a Mak2p/Mak3p histidine kinase heterodimer to control Sty1p activity. According to this model, activation of the Mak2p/Mak3p kinase via alterations in the PAS/PAC domain leads to sequential phosphorylation of histidine 221 in Mpr1p and aspartate 412 in Mcs4p. Mcs4p interacts directly with the N terminus of the Wak1p MAPKKK both before and after stress stimulation. Phosphorylation of Mcs4p alters the activity of Wak1p (and probably also the Win1p MAPKKK), leading to sequential phosphorylation of the Wis1p MAPKK and Sty1p MAPK. Other stresses control the activity of Wak1p and Win1p by an as-yet-unidentified mechanism(s).

An important difference between the S. pombe and S. cerevisiae histidine kinases is that Mak2p and Mak3p specifically sense oxidative stress, whereas Sln1p senses changes in external osmolarity. The domain architecture of the Mak2p and Mak3p proteins provides some clue as to why this may be. In particular, both proteins contain a single repeat of a PAS/PAC domain, which has been identified in a variety of prokaryotic sensors of oxygen or redox, including other members of the histidine-kinase superfamily such as E. coli Aer and ArcB and Rhizobium meliloti FixL, but not Sln1p (Crews and Fan, 1999). Importantly, the PAS/PAC domain of FixL binds heme as a prosthetic group. Changes in the availability of oxygen to the heme moiety induce a conformational change in the PAS/PAC domain that causes an alteration in the activity of the adjoining histidine-kinase domain (Gong et al., 1998; Pellequer et al., 1999). It is possible that a similar mechanism may regulate the activity of Mak2p and Mak3p. The PAS/PAC domain may also explain why both of these histidine kinases are required for signal transmission to Sty1p, because this domain has been shown to mediate protein–protein interaction in other species (Huang et al., 1993, 1995). Although Mak2p and Mak3p appear to be exclusively cytoplasmic in S. pombe, we have not been able to demonstrate an interaction between them (Quinn, Buck, and Millar, unpublished data). Thus, we cannot yet rule out the possibility that Mak2p and Mak3p initiate independent signals that are both required for activation of Wak1p.

How is the peroxide signal actually transduced to the Wak1p MAPKKK? It is possible that the peroxide signal activates the histidine-kinase domains of Mak2p and/or Mak3p. In this manner, the response-regulator domains of Mak2p, Mak3p, or both would become phosphorylated and the phosphate group would, in turn, be transferred to Mpr1p. This conclusion is supported by the observation that peroxide induces Mpr1p to associate with Mcs4p and mutation of histidine 221 in Mpr1p abolishes this association (Nguyen et al., 2000). Conversely, it is also formally possible that the peroxide signal inhibits the activity of Mak2p and/or Mak3p, leading to dephosphorylation of Mpr1p and Mcs4p, and that the dephosphorylated form of Mcs4p activates Wak1p. Notably, mcs4(D412N) cells show a higher basal level of Sty1p phosphorylation than mcs4⁺ cells, suggesting that the dephosphorylated form of Mcs4p is active. Analysis of the in vivo phosphorylation status of Mak2p, Mak3p, Mpr1p, and Mcs4p will be necessary to distinguish between these possibilities.

In contrast to Mak2p and Mak3p, Mak1p does not control Sty1p activity or Atf1p-dependent gene expression. Instead, we find that Mak1p is partially required for the induction of catalase in response to peroxide stress. It is intriguing in this respect that Mak1p also contains two PAS/PAC repeats, suggesting that Mak1p may also function as a peroxide sensor. Induction of catalase requires both the Pap1p transcription factor and Prr1p, a response-regulator protein that is homologous to the S. cerevisiae Skn7p transcription factor (Ohmiya et al., 1999). Skn7p shares an overlapping function with the Yap1p transcription factor in the response to oxidative stress in S. cerevisiae (Brown et al., 1993; Morgan et al., 1995; Krems et al., 1996; Morgan et al., 1997). It is possible that Mak1p regulates the activity of Prr1p via phosphorylation of an independent pool of Mpr1p, but the details of such a pathway are unclear because deletion of mpr1 causes superinduction of catalase expression (Nguyen et al., 2000). Nevertheless, these results suggest that the transcriptional response to free radical attack is regulated by at least two distinct phospho-relay pathways in fission yeast.

In addition to the PAS/PAC motif, Mak2p and Mak3p contain a GAF domain, which has been identified in a diverse array of phototransducing proteins, including cGMP-specific and -stimulated phosphodiesterases, Anabena adenylate cyclase, and the E. coli FhlA transcriptional regulator (Aravind and Ponting, 1997). One function of the GAF domain that has been well characterized is that of cGMP binding (Charbonneau et al., 1990). Intriguingly, the chromophore attachment site in plant and cyanobacterial phytochromes occurs on a conserved cysteine residue within their single GAF domain (Figure 3B; Yeh and Lagarias, 1998). Although the overall domain architecture of Mak2p and Mak3p is similar to that of the phytochrome family, their GAF domains lack this residue. Thus, the function of this domain in Mak2 and Mak3 is at present unknown. Intriguingly, Mak2p and Mak3p are most closely related to a histidine kinase in the human pathogen C. albicans, C.a. HK1 (Calera et al., 1998). In addition to histidine-kinase, response-regulator and GAF domains, all three proteins possess an N-terminal domain that is strikingly similar to a series of atypical serine/threonine kinase in prokaryotes, in particular M. tuberculosis PknB. To our knowledge, these are the only known proteins to have two separate kinase domains of apparently distinct specificity. Our initial analysis suggests that these N-terminal domains in Mak2p and Mak3p are not required for signal transmission to Sty1p (Quinn, Buck, and Millar, unpublished data). Nevertheless, it is tempting to speculate that Mak2p, Mak3p, and C.a. HK1 are derived from a fusion of distinct ORFs in an ancestral prokaryotic operon that functioned in a common signaling pathway to sense and respond to free radical attack. Although it is not clear whether C.a. HK1, which lacks a PAS/PAC motif, acts as a peroxide sensor, homozygous null diploids for HK1 are nonvirulent in a murine model (Calera et al., 1998, 1999). Furthermore, a homolog of Ssk1p and Mcs4p is also required for virulence in this system, suggesting that C.a. HK1 may function in a phospho-relay system similar to that described here (Calera and Calderone, 1999; Calera et al., 2000).

The absence of structural homologs of Mak2p and Mak3p in S. cerevisiae could explain why the Hog1p MAP kinase is activated only in response to osmotic stress and not to hydrogen peroxide. Indeed, S. cerevisiae has only one histidine kinase (Sln1p), whereas at least three have been identified in C. albicans and S. pombe. In addition, the C. albicans and S. pombe histidine kinases are not identical to each other. Although an osmosensing Sln1p homolog exists in C. albicans (C.a. Sln1), it is unlikely that Sty1p is controlled by a Sln1p-like molecule in S. pombe, because activation of Sty1p in response to osmotic stress does not require phosphorylation of Mcs4p. Conversely, no homologs of Mak1p or C.a. Nik1 have been identified in C. albicans and S. pombe, respectively. It seems possible, therefore, that all fungi are derived from a common eukaryotic ancestor that possessed multiple two-component regulatory systems that have been selectively lost by evolutionary pressure.

To date, no structural homologs of the phospho-relay proteins have been identified in mammals. Thus, control of SAPK pathways by two-component systems is likely to be unique to lower eukaryotes. Importantly, we have shown that the phospho-relay system is not required for activation of the S. pombe Sty1p MAPK in response to environmental stresses that activate the mammalian SAPKs, including heat shock, the protein-synthesis inhibitor anisomycin, agents that cause oxidative damage such heavy metals (e.g., arsenic or cadmium), superoxide anion generators (e.g., paraquat), or UV light mimetics (e.g., 4-nitroquinoline oxide). We tentatively propose that each environmental stress is sensed by a specific cellular receptor(s) that triggers activation of Sty1p, possibly by regulating the activity of the Wak1p/Wis4p and/or Win1p MAPKKKs (Figure 7). We will need to identify other upstream activators of the Sty1p pathway to determine whether these are structurally or functionally conserved in mammals.