(Received for publication, January 13, 1997)
From the Department of Biological Sciences,
University of Iowa and the
Department of Biochemistry,
University of Iowa, Iowa City, Iowa 52242
Two-component signal transduction systems involving histidine autophosphorylation and phosphotransfer to an aspartate residue on a receiver molecule have only recently been discovered in eukaryotes, although they are well studied in prokaryotes. The Sln1 protein of Saccharomyces cerevisiae is a two-component regulator involved in osmotolerance. Phosphorylation of Sln1p leads to inhibition of the Hog1 mitogen-activated protein kinase osmosensing pathway. We have discovered a second function of Sln1p by identifying recessive activated alleles (designated nrp2) that regulate the essential transcription factor Mcm1. nrp2 alleles cause a 5-fold increase in the activity of an Mcm1-dependent reporter, whereas deletion of SLN1 causes a 10-fold decrease in reporter activity and a corresponding decrease in expression of Mcm1-dependent genes. In addition to activating Mcm1p, nrp2 mutants exhibit reduced phosphorylation of Hog1p and increased osmosensitivity suggesting that nrp2 mutations shift the Sln1p equilibrium toward the phosphorylated state. Two nrp2 mutations map to conserved residues in the receiver domain (P1148S and P1196L) and correspond to residues implicated in bacterial receivers to control receiver phosphorylation state. Thus, it appears that increased Sln1p phosphorylation both stimulates Mcm1p activity and diminishes signaling through the Hog1 osmosensing pathway.
Two-component regulators are a family of signal transduction molecules prevalent in prokaryotic organisms that play a major role in the adaptation of microorganisms to changes in the extracellular environment (1, 2). The response of cells to chemotactic agents, sporulation conditions, and changes in osmolarity, for example, are each governed by two-component regulators in which information about external conditions is converted via a sensor-associated kinase into high energy phosphoryl groups on histidine and aspartate side chains ultimately activating an appropriate set of genes. The "two-component" designation derives from the observation that in many cases the pertinent activities are divided between two polypeptides. One is designated the "sensor/kinase" and contains an extracellular domain (sensor) and a histidine autokinase activity (transmitter). The second is called the response regulator and is made up of a receiver domain containing a conserved aspartate residue and, in some cases, an output domain.
Two-component regulators have been recently identified in eukaryotes including Arapidopsis, Saccharomyces, Dictyostelium, and Neurospora (3). In most of these eukaryotic examples, the transmitter and the receiver domains are present in a single hybrid molecule. Examples of hybrid kinases are also known in prokaryotes, and in certain cases, the attached receiver domain is thought to regulate phosphotransfer to a cytoplasmic receiver containing an output domain (4, 5). The strength of the signal through a given pathway is determined by the stability of the phosphoryl aspartate whose susceptibility to hydrolysis depends on the tertiary or higher order structure of the receiver domain. Stability of the aspartyl phosphate may be regulated by "phosphatase" activities present either in the sensor/kinase, receiver domain, or in auxiliary factors (6, 7). Whether the role of these phosphatases is enzymatic or rather to trigger a conformational alteration that alters the exposure of the residue to water is unclear.
The yeast SLN1 gene is a member of the two-component family that encodes a hybrid kinase including sensor, histidine kinase, and receiver domains (8). Its role is to sense and respond to changes in osmolarity by regulating a MAP1 kinase pathway composed of Hog1p (MAP kinase), Pbs2p (MAP kinase kinase), and Ssk2p/Ssk22p (two MAP kinase kinase kinases) (9). Sln1p phosphorylates a second two-component molecule, Ssk1p consisting of a receiver domain, via the phosphorelay intermediate, Ypd1p (10). Under conditions of increased osmolarity, the equilibrium between phosphorylated and unphosphorylated Sln1p shifts to favor the unphosphorylated form (10), and the resulting accumulation of unphosphorylated Ssk1p leads to activation of the Hog1 MAP kinase pathway (9). Phosphorylated Hog1p, in turn, activates a set of hyperosmotic response genes necessary for survival under increased external osmotic pressure (11). Also contributing to Hog1p phosphorylation in response to hyperosmolarity is Sho1p, a putative membrane-associated molecule that interacts with the Hog1p kinase Pbs2p via its Src homology 3 domain (12).
Alleles of the SLN1 gene that increase expression of an Mcm1-dependent reporter were previously isolated in a screen for mutants that increased the activity of the CYC1-lacZ reporter gene under the control of a high affinity Mcm1 binding site (referred to as a P site) (13). Deletion of the SLN1 gene causes a 10-fold reduction in Mcm1-dependent reporter gene expression without affecting Mcm1p levels, thus Sln1p is an activator of Mcm1p or of an Mcm1p auxiliary protein, and the so-called nrp2 (negative regulator of P site) mutants are recessive activating alleles of SLN1 (13). Since deletion of the HOG1 gene has no effect on the activity of an Mcm1-dependent reporter, we suggested that Mcm1p is not directly regulated by Hog1p and that Sln1p has two separate regulatory functions (13). It is not clear, however, how Sln1p accomplishes both functions. Although the mechanism by which Sln1p regulates the Hog1 MAP kinase pathway involves phophorylation and phosphotransfer, whether the same activities are required for Sln1p regulation of Mcm1p is an open question. In addition, although we have shown that Hog1p is not required for Sln1p activation of Mcm1p, other intermediates in the Sln1-Hog1 pathway have not been tested so the point at which the two "pathways" branch has not been defined.
Our present results indicate that Sln1p activation of Mcm1p uses regulatory mechanisms in common with Sln1p activation of the Hog1 pathway and indicates that signaling through the two pathways may depend on the same phosphorelay capabilities of Sln1p. The hypothesis that the nrp2 mutations affect the Sln1p phosphorylation state is further supported by our finding in two instances that the amino acid changes responsible for the nrp2 mutant phenotype map to positions likely to have an effect on the equilibrium between phosphorylated and unphosphorylated Sln1p. Based on these results, we propose a model in which shifting the equilibrium of Sln1 to the phosphorylated state results in inactivation of the Hog1 pathway and activation of the Mcm1 pathway. Given the multiple roles of Mcm1p in growth control and cell physiology, this may represent an integrated mechanism by which cells respond to stress.
Yeast strains used in this study
are listed in Table I. All strains were grown at
30 °C unless indicated otherwise. Diploids containing the
sln1::LEU2 or
sln1
::TRP1 mutations were
constructed by transformation with fragments isolated from plasmids
pGY143 and pGY148 as described previously (13). The SHO1 and
SSK1 loci were disrupted by single step gene replacement
(14) using, respectively, a SalI-NotI fragment
from plasmid pDOS84 in which TRP1 is inserted between the
two EcoRV sites of the SHO1 gene (gift of H. Saito) (12) and a PstI-XbaI fragment from the
ssk1
::LEU2 plasmid, pDSS14 (gift of
H. Saito) (9). The presence of the disruption alleles was confirmed by
Southern hybridization analysis.
|
The media were prepared as described by Sherman et al. (15).
Plates for the detection of -galactosidase contained 50 µg of
5-bromo-4-chloro-3-indolyl
-D-galactopyranoside/ml and
were prepared as described by arson et al. (16).
Plasmids used in this work are listed in
Table II. The Mcm1-dependent reporter,
P-lacZ, is carried on the previously described plasmid,
pGY48 (17), a derivative of the 2-µm URA3
CYC1-lacZ reporter, pLG670Z (18). pHL589 consists of
the entire SLN1 gene inserted into pRS314 (19). It was
constructed by a recombination strategy using KpnI-digested
pHL566 as the gapped recipient vector. pHL566 consists of a 2.95-kb
EagI/KpnI fragment of SLN1 inserted into the EagI and KpnI sites of pRS314 (19).
KpnI-digested pHL566 was cotransformed into yeast strain
JF1455 with a 2.86-kb polymerase chain reaction fragment of
SLN1 generated using oligonucleotides SLN 1588F
5-ACTATGACAGACGAATTCGACCAACATTAT-3
) and OLI196
(5
-GCTCGGAATTAACCCTCACTAGGAACAAAAGCTGGTACCATCCAAATTTGTGGTTCATGCTCTC-3
). OLI196 is complementary to sequences in the 3
-untranslated
region of SLN1 and contains sequences at the 5
end
homologous to pRS314. Generation of a full-length functional pRS314
SLN1 clone (pHL589) was confirmed by restriction and
sequence analysis, as well as by demonstrating the ability to support
SLN1-dependent growth when JF1455 was
sporulated. Construction of pHL592 containing a SLN1 gene
lacking the receiver domain (deletion of aa 1078-1220) was done by
cutting pHL589 with StuI/MluI and inserting a
2.18-kb StuI/MluI fragment obtained by digesting
a polymerase chain reaction fragment generated using oligonucleotides
SLN 1057F (5
-CTAGCACACTGGGAATTCCAACCAATTGTA) and SLN 3212R
(5
-GATCGTACGCGTCTCAGTCTCGAGTTATCAGCGGCCGCCTGTGGGGATGTTTCTACTTG).
|
pBG95 containing an HA epitope-tagged Sln1 receiver domain under the control of UASG was constructed by first constructing a galactose-inducible expression vector, pSV146, by insertion of a 1.4-kb EcoRV-BamHI fragment from YEp55 (20) containing the GAL10 UAS, promoter, and ATG into the SmaI and BamHI sites of pRS313 (19). The HpaI-HinDIII fragment containing the Sln1 receiver domain was introduced into the polylinker of pSP72 (Promega), and an HA epitope tag was introduced at the SLN1 ScaI site to create pBG94. The HA-tagged receiver domain was then introduced as a HpaI-PvuII fragment into the blunted BamHI site of pSV146. pWT5 is an nrp2-2 integrating vector constructed by inserting the 2.1-kb EagI-XbaI fragment (XbaI at SLN1 1624-end of SLN1) utilizing a polylinker EagI site into EagI-XbaI of pRS406 (19). DNA was digested with BclI which recognizes a unique site in SLN1 downstream of the nrp2-2 mutation prior to transformation into the appropriate strains.
Genetic Methods and-Galactosidase measurements were performed as described
previously using cleared glass bead extracts from log phase cells (17).
Transformation was by a modified LiOAc method (21, 22).
Yeast strains containing the UASG-regulated Sln1 receiver domain (pBG95) were grown in 2% raffinose to early log phase at which time 3% galactose or 2% glucose was added. Cultures were incubated for approximately 5 h before being harvested for preparation of protein extracts. Extracts were prepared by a glass bead disruption protocol (22) and subjected to Western blot analysis with HA antibody. Following electrophoresis on 12% SDS-polyacrylamide gels and transfer to nitrocellulose, filters were blocked for 15 min at 37 °C in 5% milk powder dissolved in 0.1% Tween/phosphate-buffered saline. Filters were subsequently incubated (6 h) with HA antibody (1:1000) (Babco) followed by secondary antibody anti-mouse IgG horseradish peroxidase (1:2500) (Sigma). Immune complexes were detected by enhanced chemiluminescence according to protocols supplied by the manufacturer (Amersham Corp.).
Gap Rescue and Reconstruction of nrp2 MutationsThree
SLN1 gapped plasmids (A, B, and C) were generated by
digesting pHL589 with EcoRI/StuI (A),
StuI/ClaI (B), or
BstEII/MluI (C). Gap rescue (23) was performed by
transforming gap A, B, or C plasmids into mutant yeast strains
harboring the nrp2-1 (JF1356), nrp2-2 (JF1357),
or nrp2-3 (JF1358) mutations. The presence of the mutation
on a rescued gap plasmid was assessed by the blue color of colonies on
5-bromo-4-chloro-3-indolyl -D-galactopyranoside plates.
Plasmids were rescued from yeast (24) and subjected to sequence
analysis. The sequence of both strands was determined. In
vitro mutagenesis (nrp2-1) or subcloning of restriction
fragment (nrp2-2 and nrp2-3) encompassing the
presumed missense mutation was done to confirm the identity of each
mutation.
Cells were cultured to early log phase (1 × 107/ml) and serially diluted 1:10 in YPD or YPD + 0.9 M NaCl. Two µl of the appropriate dilutions were spotted onto YPD and YPD + 0.9 M NaCl plates. Plates were incubated at 30 °C for up to 3 days.
Protein Extracts and Western AnalysisCells were cultured
to early log phase (7 × 106/ml) in YPD at which time
solid NaCl was added to 0.4 M. 10-ml aliquots were taken at
indicated times following the addition of salt. The t = 0 time point was taken prior to the addition of salt. Cells were
pelleted at 4 °C at low speed and frozen in dry ice prior to storage
at 70 °C. Cell pellets were washed once with 5 ml of cold protease
inhibitor mixture (50 µg/ml leupeptin, 23 µg/ml aprotinin, 1 µg/ml pepstatin, 20 µg/ml chymostatin, 2 mM
phenylmethylsulfonyl fluoride in water). Extracts were processed for
antiphosphotyrosine immunoblots essentially as described by Maeda
et al. (9). Proteins were resolved by electrophoresis on
10% SDS acrylamide gels (6% stacking gel) and electroblotted to 0.45 µm nitrocellulose (Protrans, Schliecher & Schuell). The
nitrocellulose filter was blocked overnight at 4 °C in TBST (10 mM Tris, pH 8.0, 150 mM NaCl, and 0.05% Tween 20) + 4% bovine serum albumin, washed twice for 10 min in TBST, incubated for 2 h at 4 °C with RC20 recombinant
antiphosphotyrosine antibody conjugated to horseradish peroxidase
(Transduction Laboratories, Lexington, KY) diluted 1:2500 in TBST + 4%
bovine serum albumin, and washed twice with TBST. Immune complexes were
visualized by enhanced chemiluminescence (Amersham Corp.).
Activation of
the Hog1 MAP kinase pathway depends on the phospho-relay capabilities
of the Sln1p hybrid kinase (9). Mutation of either the phosphorylated
histidine or the phosphorylated aspartate confers a phenotype
equivalent to the null, indicating the requirement of these residues
for signaling (9). Activation of Mcm1-dependent transcription may likewise involve Sln1p phosphorylation and
phosphotransfer to downstream molecules. Alternatively, activation of
Mcm1p may be independent of the phosphorylation state of Sln1p.
Assessment of Mcm1p activity in the null or in phosphorylation
defective mutants is difficult since the SLN1 deletion
mutation causes lethality on rich media and extremely slow growth on
minimal media in some genetic backgrounds (13) and unconditional
lethality in other genetic backgrounds (9). To circumvent these
difficulties, we tested the involvement of Sln1p phospho-intermediates
in Mcm1 signaling by examining the effect of the nrp2
mutations on the ability of cells to respond to osmotic stress. If the
stimulating effect of nrp2 mutations on
Mcm1-dependent transcription is due to a shift in the Sln1p
equilibrium to the phosphorylated state, for example, the resulting
decrease in unphosphorylated Sln1p might limit the amount of
unphosphorylated Ssk1p available to trigger a normal osmotic
response. To test this idea, we examined the osmotic sensitivity of our
nrp2 mutants. To render the cell completely dependent on
Sln1p signaling for its response to hyperosmolarity (12), it was
necessary to first delete the SHO1 gene which encodes a
second osmosensor. The SHO1 gene was deleted by one-step
disruption in nrp2-1 and nrp2-3 strains. The
osmotolerance of isogenic nrp2-1 and nrp2-1
sho1- strains and isogenic sho1-
and wild type
strains was compared by spotting serial dilutions of cells on 0.9 M NaCl plates or on unsupplemented YPD media.
SHO1+ SLN1+ cells
exhibited normal viability on 0.9 M NaCl although colony size was reduced. Strains lacking the Hog1 MAP kinase, on the other
hand, were extremely sensitive to this condition (Fig.
1). Both nrp2-1 and sho1-
single mutants grew with wild type viability on 0.9 M NaCl
plates at 30 °C; however, the plating efficiency of the
nrp2-1sho1 double mutants was reduced approximately 100-fold (Fig. 1). Likewise, the nrp2-3 mutant was partially
sensitive to 0.9 M NaCl, and SHO1 deletion
exacerbated the effect (Fig. 1). The enhanced osmosensitivity of the
double nrp2 sho1 mutant is consistent with nrp2
mutations shifting the Sln1 equilibrium to favor the phosphorylated
form, thus diminishing or inhibiting the normal effect of increased
osmolarity on the activity of the sensor/kinase.
To examine in more detail the basis for the osmosensitivity of
nrp2 mutants, we examined the tyrosine phosphorylation of
Hog1p in the nrp2 sho1 double mutant and sho1 and
nrp2 single mutants. Strains were grown in rich media, and
NaCl was added to 0.4 M at time 0. Aliquots were taken over
time for immediate processing, denaturing gel electrophoresis, and
immunoblot analysis with antiphosphotyrosine antibody. As expected, a
50-kDa band corresponding to Hog1p was apparent in wild type strains at
the earliest point assayed (5 min) after addition of salt (Fig.
2). Likewise, in sho1 deletion and
nrp2 single mutants, Hog1p phosphorylation was rapidly
induced and persisted for 15-20 min. In contrast, in the nrp2
sho1 double mutants, Hog1 phosphorylation was not detectable (Fig.
2). Taken together, these studies indicate that
Mcm1-dependent transcription is stimulated by
phosphorylation of Sln1p.
The nrp2 Mutations Map to the Receiver Domain (of Sln1p) and to the Linker Region Between the Transmembrane and Histidine Kinase Domain of Sln1p
To determine if the proposed effects of the nrp2
mutations on Sln1p phosphorylation state could be understood at the
molecular level, we used gap rescue to localize the mutations and
sequence analysis to pinpoint the alterations. Plasmids expressing the nrp2 mutant alleles were constructed by in vitro
mutagenesis or by swapping the appropriate restriction fragments into
the context of wild type SLN1 sequences carried on pHL589.
The phenotype of the reconstructed nrp2 alleles was
evaluated in a sln1- background following transformation
and sporulation of a sln1-
/SLN1+
heterozygous diploid (JF1455). Two of the nrp2 alleles,
nrp2-2 and nrp2-3, were found to be point
mutations leading to changes in conserved proline residues just
downstream of the aspartate residue in the Sln1p receiver domain which
is the target for phosphorylation (Fig. 3).
nrp2-2 is a change to serine at P1148 and nrp2-3
is a change to leucine at P1196. The third nrp2 mutation,
nrp2-1, is a T550I change in the non-conserved region
upstream of the histidine kinase domain in the region linking the
kinase to the presumed membrane spanning domains (Fig. 3). The
significance of the base pair changes in the nrp2-2 and
nrp2-3 mutants can be evaluated by comparison of the Sln1p
receiver domain to the bacterial receiver, CheY, for which a
crystallographic structure is available. As described in detail below
(see "Discussion"), the alignment of Sln1p and CheY receiver
domains shows that the nrp2-2 and nrp2-3
mutations map to positions that are likely to affect Sln1p function and
are consistent with changes in Sln1p phosphorylation state.
The Sln1 Pathways Branch Prior to Ssk1p
Since phosphorylated Sln1p inhibits the activity of the Hog1 pathway but appears to stimulate the activity of the Mcm1 pathway, it seemed possible that phosphorylated and unphosphorylated Ssk1p might likewise play opposing roles in regulating the two pathways. We tested the importance of Ssk1p in Sln1-Mcm1 signaling by measuring the effect of an SSK1 deletion on the activity of the Mcm1-dependent reporter, P-lacZ (P = Mcm1 binding site). Unlike the SLN1 deletion which causes constitutive activation of the Hog1 pathway, and perhaps, as a result, lethality, SSK1 deletion is not lethal. In the absence of Ssk1p, the Hog1 pathway fails to be activated by the Sln1 pathway but is under normal regulation via Sho1p-Pbs2p interactions. If the presence of Ssk1p is required for signaling to Mcm1p, deletion of SSK1 would be expected to prevent the activation of the Mcm1-dependent reporter by nrp2 mutations. To examine the involvement of Ssk1p in Sln1-Mcm1 signaling, we deleted the SSK1 gene by one-step transformation of wild type and nrp2-1 mutant strains. We found that the activity of our reporter was virtually unchanged as a result of SSK1 deletion in either wild type or nrp2 backgrounds (Table III).
|
The absence of an effect of the Ssk1p
receiver on Sln1-Mcm1 signaling suggested the possibility that Sln1p
itself interacts with downstream non-two-component signaling molecules.
In this case, the receiver domain of the Sln1p hybrid kinase may be an important signaling intermediate rather than an inhibitory domain as
postulated for other hybrid kinases. We examined this possibility by
deleting the receiver domain from Sln1p and separately expressing it
under the control of an inducible promoter. Before evaluating the
effect of expressing the transmitter in the absence of the receiver, it
was necessary to establish that the two separated domains were
functional. We tested whether the two half-molecules could function in
trans by simultaneously introducing plasmid pHL592 carrying
a receiverless Sln1p (aa 1-1077) and plasmid pBG95 carrying the
galactose-inducible receiver (aa 1037-1220, end) into a diploid strain
(JF1705) heterozygous for a LEU2 marked sln1-
allele. The sln1-
heterozygote was then sporulated and dissected on media prepared with galactose instead of glucose. The
appearance of both Leu+ (sln1-
) and
Leu
(SLN1+) viable spores at equal
frequencies (Table IV) and indistinguishable growth
rates suggests that the two plasmids fully complemented the
sln1-
mutant phenotype. Next, we asked whether either
domain alone was able to complement the sln1-
lethality.
To this end we compared the frequency of plasmid loss in
Leu+ (sln1-
) and Leu
(SLN1+) spore colonies. After 8 h of growth
in non-selective media, 40% of the cells in a culture from a
Leu
colony had lost either the plasmid carrying the
N-terminal (marked with TRP1) or the C-terminal (marked with
HIS3) SLN1 portion. In contrast, no
His
or Trp
segregants were present in a
culture from a Leu+ (sln1-
) colony
(Table V). This result indicates that both the transmitter kinase and receiver domains are required for Sln1 function.
Neither the C terminus nor the N terminus alone is sufficient for
complementation. Further evidence that the receiver domain is essential
for viability comes from the failure of a
sln1
::LEU2 strain carrying the
UASG receiver and transmitter plasmids to survive when
shifted from galactose to glucose media (data not shown).
|
|
To further examine the role of the receiver domain in signal transduction, the Sln1 receiver domain was overexpressed in both wild type (SLN1+) and nrp2 strains. If the receiver plays an inhibitory role, overexpression had the potential both to diminish phosphorylation of Ssk1 and therefore activate the Hog1 pathway (causing lethality) and to diminish signaling to Mcm1. We found that galatose-induced overexpression of the UASG-regulated Sln1 C-terminal receiver domain had no effect on doubling time nor on the P-lacZ reporter gene activity of SLN1+ cells (Table VI). Taking advantage of a 9-aa hemagglutinin epitope tag (YPYDVPDYA) inserted at aa 1065, it was possible to confirm by Western analysis the presence of high levels of the receiver domain (a single species migrating at 22 kDa) in extracts prepared from galactose grown cells and not in extracts prepared from glucose grown cells (data not shown). A second indication that the liberated receiver molecule was indeed capable of interacting and functioning with the intact Sln1p molecule, albeit inefficiently, was the observation that the overexpressed receiver domain partially complemented the activation phenotype of the nrp2-2 mutant (Table VI). Thus, the absence of an effect of overexpressing the Sln1p receiver on viability and Mcm1-dependent transcription again argues that the Sln1 receiver domain is likely to play a positive and not a negative role in signal transduction.
|
The phosphotransfer capabilities of Sln1p have been shown to be important for regulation of the Hog1 MAP kinase osmotic response pathway, and it is clear that hyperosmotic conditions promote the accumulation of unphosphorylated Sln1p (9). We previously showed that Sln1p separately regulates the activity of Mcm1-dependent genes and suggested that this function of Sln1p might involve the phosphorylated form of Sln1p (13). To provide evidence for this assertion and to determine whether activation of the Mcm1 pathway, like activation of the Hog1 pathway, involves the phosphotransfer capabilities of Sln1p, we investigated the effect of nrp2 mutations on the capacity of the cell to respond to osmotic stress. The osmotic phenotype of nrp2 mutants suggests that activation of the Mcm1 pathway by the nrp2 mutations decreases the unphosphorylated Ssk1p that is available to stimulate the Hog1 pathway. The absence of Hog1p phosphorylation in the nrp2 strains is consistent with this interpretation. These results demonstrate that phosphorylation of Sln1p has at least two functions: one is to turn off the Hog1 pathway through the action of the receiver, Ssk1p, and the other is to activate Mcm1p. This is accomplished in an Ssk1p-independent manner. These results do not distinguish whether Mcm1p activation is dependent on the phosphotransfer capabilities of Sln1p or if it simply depends on histidine phosphorylation of Sln1p. For example, Sln1p phosphorylation may be required to induce a conformational change which in turn mediates a particular protein-protein interaction. However, the location of the nrp2-2 and nrp2-3 mutations in the receiver domain at positions likely to affect the stability of the aspartyl phosphate suggests that Sln1p-H to Sln1p-D phosphotransfer as well as the histidine autokinase activities of Sln1p are important for Mcm1p activation. In vitro phosphorylation and phosphotransfer experiments are underway that will further test this idea.
nrp2 Mutations Are Localized in Both the Transmitter Kinase and Receiver Domains of Sln1pTwo of three nrp2 alleles
are changes in proline residues in positions highly conserved between
other two-component receivers. In CheY, these residues are postulated
to be critical for the conformation of the CheY receiver. The
nrp2-2 mutation is a P S mutation at residue 1148 of
Sln1p. This position, four amino acids downstream of the phosphoryl
aspartate residue, corresponds to P61 of CheY. This residue is part of
a structurally important
-turn loop. The proximity of this base pair
to the phosphorylation domain, as well as its solvent accessibility and
high degree of conservation, has led to the suggestion that this loop
region of response regulators is likely to be important for recognition by kinases (25).
The nrp2-3 mutation is a P L mutation at residue 1196 of
Sln1p. This position corresponds to the highly conserved P110 of CheY
which makes a cis-peptide bond with the absolutely conserved Lys-109
residue to form a rigid
5-
5 loop that is postulated to be
important in propagating conformational changes as a result of
activation (25). The side chain of Lys-109 interacts with Asp-57 in the
acidic pocket in the structure of unphosphorylated CheY (26). Although
detailed structural information is unavailable for the phosphorylated
form of CheY, phosphorylation is expected to perturb the
Lys-109-Asp-57 interaction, initiating the switch to the active
conformation. The
5-
5 loop also contributes to the solvent
accessible face that interacts alternately with CheA and the flagellar
motor proteins (27). Interestingly, the K109R mutation in CheY blocks
the interaction of P-CheY with the phosphatase, CheZ, and therefore
results in substantial decreases in dephosphorylation rates as well as
decreased phosphorylation rates (25, 28).
In contrast to nrp2-2 and nrp2-3, the
nrp2-1 mutation is not located in the receiver domain.
nrp2-1 is a T I mutation located in the linker region
between the histidine kinase domain and the transmembrane domains. The
position of this mutation is reminiscent of activating mutations in
bvgS, a hybrid two-component regulator involved in virulence
from Bordetella pertussis and narX, an
Escherichia coli sensor involved in regulation of anaerobic
respiratory genes. Constitutive mutations in these proteins have been
mapped to a flexible linker region between the transmembrane domains
and the histidine kinase domain (29, 30). This type of mutation may trigger signal transmission in a signal-independent fashion by mutational simulation of an appropriate conformational state. Alternatively, in the case of Sln1p, phosphorylation of the affected threonine may normally modulate the activity of the histidine kinase.
An explanation of the activating phenotype of nrp2 mutations
we have identified must account for the fact that they are recessive. An increase in receiver phosphorylation can occur by either increased rate of phosphorylation or a decreased rate of dephosphorylation. We do
not believe that the nrp2 mutations increase kinase activity because it is not apparent how an increase in kinase activity could be
recessive (13). The recessiveness of the nrp2 mutations is
more consistent with a defect in a postulated Sln1p-associated aspartyl-directed phosphatase-like activity. Perhaps the receiver domain mutations nrp2-2 and nrp2-3 cause
decreased sensitivity to a phosphatase. These mutations lie within the
exposed and
5-
5 loops which are postulated to define a
surface involved in protein-protein interactions. On the other hand,
since many bacterial transmitter kinases possess inherent phosphatase
activity toward their respective receivers, the nrp2-1
mutant may be defective in a putative Sln1p transmitter
kinase-associated phosphatase activity (7, 31, 32). Since Sln1p, like
other two-component regulators appears to function as a dimer (9),
functional interactions with a phosphatase may be provided in
trans and thus account for the recessiveness of the
nrp2 activating phenotype.
Since SSK1 has no apparent role in signaling
to Mcm1p, we considered a possible role for the Sln1p receiver domain
in signaling to the Mcm1p branch of the pathway. The role of the
attached receiver domain in hybrid kinases is not clear. In the case of
VirA and ArcB and others, it appears to have at least in part an
inhibitory role, since deletion of the domain leads to increased
signaling (4, 5). Thus the attached receiver domain appears to prevent phosphotransfer to the cytoplasmic response regulator and to downstream targets until signal-mediated histidine autophosphorylation and internal phosphotransfer leads to a change in conformation that allows
phosphotransfer to a second substrate. An alternate, but not
necessarily mutually exclusive, role for the attached receiver is in
signal transmission. Phosphorylation of the attached receiver may be
required for subsequent phosphorylation of the cytoplasmic receiver or
other downstream molecules. The observation that mutation of the Sln1p
aspartate target from aspartate to asparagine (D1144N) has a phenotype
equivalent to the SLN1 null (9) indicates that Asp-1144 is
important but does not distinguish between the two models. If the role
of Sln1p Asp-1144 is in inhibiting phosphotransfer to a second
molecule, removal of that domain might be expected to relieve the
inhibition and promote phosphorylation of Ssk1p. Hyperphosphorylation
of Ssk1p might inhibit the activation of the Hog1 pathway in response
to salt but would not be lethal. Hence the requirement for the Sln1p
receiver domain to support viability in a sln1- mutant
argues that the Sln1p receiver domain must have a role in signal
transmission. The signaling role of the Sln1p receiver is consistent
with conclusions from a very recent study by Saito and colleagues (10)
in which an in vitro system was used to demonstrate transfer
of the histidine phosphate from the transmitter kinase to the receiver
aspartate followed by transfer to the histidine of a phosphorelay
intermediate, Ypd1p, and finally to the aspartate of Ssk1p (10).
In conclusion, our analysis of the nrp2 alleles of SLN1 has shown that phosphorylation of Sln1p has two effects in the cell, inhibition of the Hog1 MAP kinase pathway and activation of Mcm1-dependent transcription. It is interesting to speculate that the linkage of Mcm1p activation to the Hog1 osmosensing pathway through Sln1p may indicate that the cell must regulate the activity of Mcm1p in response to changes in environmental osmolarity. Since Mcm1p is suspected or known to be involved in regulating a large number of different physiological processes including cell cycle, cell wall structure, metabolism, and cell type (33-37), regulation of Mcm1p activity may coordinate the physiology of the cell under hyperosmotic conditions.
Many thanks to our generous colleagues, H. Saito and Fred Winston for strains and plasmids, to J. Hoch for insightful discussions, and to P. Geyer for critical comments on the manuscript.