Molecular characterization of functional domains in the protein kinase SOS2 that is required for plant salt tolerance.

Abstract

The SOS3 (for SALT OVERLY SENSITIVE3) calcium binding protein and SOS2 protein kinase are required for sodium and potassium ion homeostasis and salt tolerance in Arabidopsis. We have shown previously that SOS3 interacts with and activates the SOS2 protein kinase. We report here the identification of a SOS3 binding motif in SOS2 that also serves as the kinase autoinhibitory domain. Yeast two-hybrid assays as well as in vitro binding assays revealed a 21-amino acid motif in the regulatory domain of SOS2 that is necessary and sufficient for interaction with SOS3. Database searches revealed a large family of SOS2-like protein kinases containing such a SOS3 binding motif. Using a yeast two-hybrid system, we show that these SOS2-like kinases interact with members of the SOS3 family of calcium binding proteins. Two-hybrid assays also revealed interaction between the N-terminal kinase domain and the C-terminal regulatory domain within SOS2, suggesting that the regulatory domain may inhibit kinase activity by blocking substrate access to the catalytic site. Removal of the regulatory domain of SOS2, including the SOS3 binding motif, resulted in constitutive activation of the protein kinase, indicating that the SOS3 binding motif can serve as a kinase autoinhibitory domain. Constitutively active SOS2 that is SOS3 independent also was produced by changing Thr(168) to Asp in the activation loop of the SOS2 kinase domain. Combining the Thr(168)-to-Asp mutation with the autoinhibitory domain deletion created a superactive SOS2 kinase. These results provide insights into regulation of the kinase activities of SOS2 and the SOS2 family of protein kinases.

Full text

Page 1

The Plant Cell, Vol. 13, 1383–1399, June 2001, www.plantcell.org © 2001 American Society of Plant Physiologists
Molecular Characterization of Functional Domains in the
Protein Kinase SOS2 That Is Required for Plant Salt Tolerance
Yan Guo, Ursula Halfter, Manabu Ishitani, and Jian-Kang Zhu
Department of Plant Sciences, University of Arizona, Tucson, Arizona 85721
1
The SOS3 (for SALT OVERLY SENSITIVE3) calcium binding protein and SOS2 protein kinase are required for sodium
and potassium ion homeostasis and salt tolerance in Arabidopsis. We have shown previously that SOS3 interacts with
and activates the SOS2 protein kinase. We report here the identification of a SOS3 binding motif in SOS2 that also
serves as the kinase autoinhibitory domain. Yeast two-hybrid assays as well as in vitro binding assays revealed a 21–
amino acid motif in the regulatory domain of SOS2 that is necessary and sufficient for interaction with SOS3. Database
searches revealed a large family of SOS2-like protein kinases containing such a SOS3 binding motif. Using a yeast two-
hybrid system, we show that these SOS2-like kinases interact with members of the SOS3 family of calcium binding pro-
teins. Two-hybrid assays also revealed interaction between the N-terminal kinase domain and the C-terminal regula-
tory domain within SOS2, suggesting that the regulatory domain may inhibit kinase activity by blocking substrate
access to the catalytic site. Removal of the regulatory domain of SOS2, including the SOS3 binding motif, resulted in
constitutive activation of the protein kinase, indicating that the SOS3 binding motif can serve as a kinase autoinhibitory
domain. Constitutively active SOS2 that is SOS3 independent also was produced by changing Thr
vation loop of the SOS2 kinase domain. Combining the Thr
created a superactive SOS2 kinase. These results provide insights into regulation of the kinase activities of SOS2 and
the SOS2 family of protein kinases.
168
to Asp in the acti-
168
-to-Asp mutation with the autoinhibitory domain deletion
INTRODUCTION
Many extracellular stimuli, including hormones and such en-
vironmental signals as abscisic acid, gravity, light, salinity,
drought, cold, oxidative stress, anoxia, and mechanical per-
turbation, cause changes in cytosolic free Ca
tion (Poovaiah and Reddy, 1993; Bush, 1995; Trewavas and
Malho, 1998; Knight, 2000). In addition, changes in cytosolic
calcium also occur during the tip growth of root hairs and
pollen tubes (Pierson et al., 1996). Cytosolic calcium
changes could be brought about by release from internal
stores and/or entry from outside the cell (Gilroy et al., 1993;
Knight, 2000). The change can vary in magnitude, duration,
frequency, and spatial arrangement within the cell, depend-
ing on the nature of the stimulus. These variables may con-
tribute to the specificity of cytosolic calcium signals. In the
cytosol, calcium functions as a second messenger, coupling
the stimulus to cellular responses. The targets of second-
messenger calcium in the cytosol are calcium-modulated
proteins (Roberts and Harmon, 1992; Zielinski, 1998).
The superfamily of EF-hand helix-loop-helix calcium bind-
ing proteins represents the largest group of calcium-modu-
lated proteins (Moncrief et al., 1990). More than 250 such
2
?
concentra-
proteins belonging to nearly 40 subfamilies have been found
in various eukaryotic cells (Celio et al., 1996). Some of these
proteins function in the buffering of cytosolic calcium con-
centration (Cox, 1990). Most, however, are presumed to
function in the mediation of calcium signals in response to
extracellular stimuli (Cohen and Klee, 1988; Cox, 1990).
Upon binding to calcium, these proteins undergo conforma-
tional changes that subsequently regulate the activities of
target proteins (Cohen and Klee, 1988; Cox, 1990; Roberts
and Harmon, 1992; Celio et al., 1996; Zielinski, 1998). Vari-
ous calcium binding proteins have been shown or sug-
gested to be involved in nearly every aspect of cellular
function, including cell division, expansion, and metabolism
(Cohen and Klee, 1988; Cox, 1990; Celio et al., 1996). De-
spite the apparent importance of EF-hand calcium binding
proteins, no specific function has been established for the
majority of them (Celio et al., 1996). Elucidation of their func-
tion would require genetic studies of loss-of-function muta-
tions and cell biological identification of their direct and
indirect target proteins.
Calcium signaling often is coupled with protein phosphor-
ylation. Phosphorylation by protein kinases is one of the most
common and important regulatory mechanisms in signal
transduction (Hardie, 1999). In plants, the calcium-depen-
dent, calmodulin-independent protein kinases are modu-
lated directly by calcium via their intrinsic calmodulin-like
1
ag.arizona.edu; fax 520-621-7186.
To whom correspondence should be addressed. E-mail jkzhu@

Page 2

1384The Plant Cell
domain (Harper et al., 1991; Roberts and Harmon, 1992).
The coupling between calcium and protein phosphorylation
also can be indirect. For example, in calmodulin kinases,
calcium binds to calmodulin, which in turn interacts with and
activates the kinase catalytic subunit (Hardie, 1999). In the
protein phosphatase type 2B (calcineurin), calcium controls
the activity of the protein phosphatase catalytic subunit cal-
cineurin A through two EF-hand calcium binding proteins,
calmodulin and calcineurin B (Klee et al., 1988).
In Arabidopsis, the
SOS2
(for
and
SOS3
genes are required for potassium and sodium ion
homeostasis and plant salt tolerance (Liu and Zhu, 1997;
Zhu et al., 1998).
SOS3
encodes a myristoylated EF-hand
calcium binding protein (Liu and Zhu, 1998; Ishitani et al.,
2000) that may sense the calcium signal elicited by salt
stress (Lynch et al., 1989; Knight et al., 1997).
codes a serine/threonine protein kinase with an N-terminal
kinase domain similar to that of SNF1/AMPK (Hardie et al.,
1998) and a novel C-terminal domain that is presumably
regulatory (Liu et al., 2000). Recently, we found that SOS3
interacts physically with SOS2 in the yeast two-hybrid sys-
tem as well as in vitro (Halfter et al., 2000). In the presence
of calcium, SOS3 activates SOS2 protein kinase activity
(Halfter et al., 2000). The interaction between SOS3 and
SOS2 also is supported by
sos2 sos3
sis, which indicates that the two genes function in the same
pathway (Halfter et al., 2000). Salt stress upregulation of the
SOS1
gene encoding a Na/H
the control of the SOS3-SOS2 regulatory pathway (Shi et
al., 2000).
SALT OVERLY SENSITIVE2
)
SOS2
en-
double mutant analy-
??
antiporter is partially under
In this article, we further characterize the interaction be-
tween SOS2 and SOS3 and identify SOS2 domains involved
in kinase activation and autoinhibition. We show that a 21–
amino acid motif in the regulatory domain of SOS2 is neces-
sary and sufficient for the interaction with SOS3. We extend
these protein interaction studies to a family of SOS2-like
protein kinases and the SOS3 family of calcium binding pro-
teins. We also found interaction between the N-terminal ki-
nase domain and the C-terminal regulatory domain within
the SOS2 protein. A constitutively active SOS2 kinase could
be generated either by deletion of the C-terminal portion, in-
cluding the SOS3 binding motif, or by a Thr
in the activation loop of the SOS2 kinase domain. When the
two constitutively active mutations were combined, a super-
active SOS2 was created. Our results provide an important
structural basis for understanding the regulation of the ki-
nase activities of SOS2 and SOS2-like protein kinases.
168
-to-Asp change
RESULTS
Identification of an SOS3 Binding Motif in SOS2
To identify amino acid residues in SOS2 that might be nec-
essary for interaction with SOS3, we cloned various parts of
SOS2 into the prey vector pACT2 and then tested for interac-
tion with the bait pAS-SOS3 in the yeast two-hybrid system.
Figure 1 shows that full-length SOS2 interacted strongly
with SOS3 and that the Lys
-to-Asn mutation, which abol-
40
Figure 1. Deletion Analysis of SOS2 to Identify an SOS3 Binding Motif.
Depicted are yeast two-hybrid constructs of SOS2 kinase with various deletions that were tested in combination with bait pAS-SOS3 in yeast for
LacZ activation and quantified as ?-galactosidase activity (n ? 3). The C-terminal regulatory domain has been shown previously to be sufficient
for binding to SOS3, whereas the N-terminal kinase catalytic domain is not required (Halfter et al., 2000). N-terminal deletion of 320 or 332 amino
acids results in the loss of ?-galactosidase activity, whereas the N-terminal deletion of 303 amino acids leads to very strong interaction. Amino
acid residues from Met309 to Arg330 are sufficient and necessary to activate the LacZ reporter and are at a level comparable to that of the intact
kinase protein.

Page 3

Functional Domains in SOS2 1385
ishes SOS2 kinase activity (Halfter et al., 2000), had little ef-
fect on this interaction. SOS2T1, which contains only the
regulatory domain of SOS2, interacted more strongly with
SOS3 (Figure 1). These results confirm our previous obser-
vation that the regulatory domain of SOS2 mediates its in-
teraction with SOS3 (Halfter et al., 2000). The significantly
stronger interaction exhibited by SOS2T1 is interesting and
suggests that the kinase domain interferes with SOS3 bind-
ing, perhaps by blocking access to the regulatory domain.
Serial deletions of the SOS2 regulatory domain were
made and then tested for interaction with SOS3. The results
show that residues between Asp
for SOS2 interaction with SOS3 (Figure 1). SOS2T4, which
spans from Gly
to Arg , was found to be sufficient to in-
teract with SOS3. We further delimited the region sufficient
for SOS3 interaction to a stretch of 21 amino acids between
Met
and Arg of SOS2 (Figure 1). SOS2D1, in which this
21–amino acid motif was deleted, failed to interact with
SOS3 at all (Figure 1). Therefore, we conclude that this 21–
amino acid motif is necessary and sufficient for interaction
with SOS3. This SOS3 binding sequence is designated the
FISL motif for its conserved amino acid residues (see below)
and for convenience.
304
and Tyr
321
are required
257 368
309330
The FISL Motif Binds to SOS3 in Vitro
To determine whether the 21–amino acid FISL motif defined
in the yeast two-hybrid system is sufficient to bind SOS3 in
vitro, we expressed the peptide in
tathione
S
-transferase (GST) fusion protein. The GST-FISL
fusion protein was labeled with phosphorus-32, and then
the GST portion was removed by protease cleavage (Figure
2A). The labeled FISL peptide was overlaid onto a protein
blot of SOS3 and control proteins. Figures 2B and 2C show
that FISL was able to bind to GST-SOS3. This binding is not
dependent on Ca
, because similar results were obtained
when the reaction was depleted of Ca
Ca
was added (Figure 2C). In contrast, FISL did not bind
to the control protein GST, GST-SOS2, or molecular mass
markers. We also found that FISL was not capable of bind-
ing to the cytosolic tail of the sodium/proton antiporter
SOS1 (Shi et al., 2000), which was expressed as a GST fu-
sion protein. These results show that FISL binds specifically
to SOS3.
Escherichia coli
as a glu-
2
?
2
?
(Figure 2B) or when
2
?
SOS2-like Protein Kinases
We searched the GenBank database and identified a large
number of plant sequences that are highly similar to SOS2
in both the kinase catalytic domain and the regulatory do-
main. These SOS2-like proteins (protein kinase S [PKS] pro-
teins) all contain a putative FISL motif. It seems that the FISL
motif is found only in the SOS2 subfamily of protein kinases.
Table 1 lists plant protein kinases that appear to contain an
Figure 2. The FISL Motif Binds to SOS3 in Vitro in Gel Blot Overlay
Assays.
(A) The 21–amino acid FISL motif identified in the yeast two-hybrid
assay was produced as GST-FISL. GST-FISL was labeled with
phosphorus-32 and, after cleavage with thrombin, which removed
the GST portion, yielded 32P-FISL to be used as a probe.
(B) GST-SOS3 and control proteins. Molecular mass markers, both
prestained (MM, pre) and unstained (MM), GST-SOS3, GST-SOS1,
and GST-SOS2 were separated on a 7.5% SDS-PAGE gel, electro-
blotted onto a membrane, and stained with Ponceau S to reveal all
proteins (left). The membrane was overlaid with 32P-FISL in the pres-
ence of 10 mM EGTA and subjected to autoradiography (right).
(C) As given in (B), except that the hybridization solution included
1 mM free Ca2?.

Page 4

1386The Plant Cell
FISL motif. Sequence analyses of the putative FISL motifs
indicate that residues A, F, I, S, L, and F are conserved ab-
solutely (hence the name FISL motif). Twenty-three of the
PKS proteins are from Arabidopsis and are named PKS2 to
PKS24. PKS2 to PKS5 correspond to SIP1 (for SOS3-inter-
acting proteins) to SIP4 (Halfter et al., 2000), respectively.
PKS6 to PKS8 represent three other SOS3-interacting pro-
tein kinases that had been obtained from a two-hybrid li-
brary screening (Halfter et al., 2000) but were not included.
Except for PKS3, which was cloned based on sequence
homology with serine/threonine protein kinases (Mizoguchi
et al., 1994), cDNAs were obtained for PKS2 to PKS8 by re-
verse transcription–polymerase chain reaction (PCR). Align-
ment of the amino acid sequences of SOS2 and seven PKS
proteins shows that the kinase domains of these proteins
are very conserved (Figure 3). In the superfamily of protein
kinases, these proteins can be placed in the CaMKII/SNF1/
AMPK family (Hanks and Hunter, 1995). Like SOS2, the PKS
proteins contain the conserved FISL motif located near the
kinase domain. When the seven PKS proteins were identi-
fied as SOS3-interacting proteins, all clones were truncated.
Interestingly, these N-terminal truncations never extended
into the conserved SOS3 binding motif (Table 2; Halfter et
al., 2000). This is consistent with the FISL motif being nec-
essary for the PKS proteins to interact with SOS3.
Expression of PKS Genes under Salt Stress
Steady state transcript levels of
shoots of young Arabidopsis seedlings were analyzed. Be-
cause of our interest in plant salt stress responses, potential
PKS
regulation by salt stress also was examined. Two-
week-old seedlings were subjected to various durations of
PKS
genes in roots and
Table 1.
List of Plant Protein Kinases Containing the FISL Motif
a
a
A 21–amino acid SOS3 binding sequence called the FISL motif was identified in SOS2. Other serine/threo-
nine protein kinases appear to contain an FISL motif, with A, F, I, S, L, and F being conserved absolutely.
The cDNAs for PKS2 to PKS8 were determined experimentally, and together with their predicted amino acid
sequences, they have been submitted to GenBank.

Page 5

Functional Domains in SOS21387
Figure 3. Amino Acid Alignment of SOS2 with PKS2 to PKS8.
The N-terminal kinase catalytic domain is highly conserved. The C-terminal regulatory domain contains the conserved FISL motif (marked). Also
marked is the activation loop between the conserved DFG and APE motifs (dots) and the Thr residue, which may be phosphorylated by an up-
stream protein kinase (asterisk). Dashed lines represent spaces that were introduced to maximize alignment.