This article is a Plant Cell Advance Online Publication. The date of its first appearance online is the official date of publication. The article has been
edited and the authors have corrected proofs, but minor changes could be made before the final version is published. Posting this version online
reduces the time to publication by several weeks.
Integrated Systems View on Networking by Hormones
in Arabidopsis Immunity Reveals Multiple Crosstalk
for Cytokinin
W
Muhammad Naseem,a Nicole Philippi,a,1 Anwar Hussain,b,2 Gaby Wangorsch,a Nazeer Ahmed,c
and Thomas Dandekara,3
a Department
of Bioinformatics, Biocenter, D-97074 Wuerzburg, Germany
of Microbiology and Molecular Genetics, University of the Punjab, Lahore 54590, Pakistan
c Department of Biotechnology and Informatics, Balochistan University of Information Technology, Engineering and Management
Sciences, Quetta 87300, Pakistan
b Department
Phytohormones signal and combine to maintain the physiological equilibrium in the plant. Pathogens enhance host
susceptibility by modulating the hormonal balance of the plant cell. Unlike other plant hormones, the detailed role of
cytokinin in plant immunity remains to be fully elucidated. Here, extensive data mining, including of pathogenicity factors, host
regulatory proteins, enzymes of hormone biosynthesis, and signaling components, established an integrated signaling
network of 105 nodes and 163 edges. Dynamic modeling and system analysis identified multiple cytokinin-mediated
regulatory interactions in plant disease networks. This includes specific synergism between cytokinin and salicylic acid
pathways and previously undiscovered aspects of antagonism between cytokinin and auxin in plant immunity. Predicted
interactions and hormonal effects on plant immunity are confirmed in subsequent experiments with Pseudomonas syringae
pv tomato DC3000 and Arabidopsis thaliana. Our dynamic simulation is instrumental in predicting system effects of individual
components in complex hormone disease networks and synergism or antagonism between pathways.
INTRODUCTION
Plant pathogen interactions are the consequence of a dynamic coevolution. Mounting resistance on the part of the host is balanced
by altered virulence on the part of the pathogen. This monolayer
paradigm, known as the gene-to-gene concept, turned into a multiphase zigzag logical model when sufficient data became available
on the innate components of plant immunity (Jones and Dangl,
2006). Both pathogen-associated molecular patterns (PAMPs)
and effectors are pathogenicity factors, which provoke immune
triggers in the host (Schneider and Collmer, 2010). No matter
whether these triggers lead to resistance or susceptibility, plant immune networks and the ultimate phenotype (Robert-Seilaniantz
et al., 2011a) are often critically determined by small molecules
with hormonal properties (salicylic acid [SA], jasmonic acid [JA],
auxin, ethylene [ET], gibberellic acid [GA], abscisic acid [ABA],
and various cytokinins). With the spotlight more on other phytohormones, the detailed effects of cytokinin in plant immunity in
this context remain unclear. By modeling and experiments, we
1 Current
address: Max Planck Institute for Biology of Aging, Bioinformatics Group, Robert-Koch-Strasse 21, D-50931 Cologne, Germany.
2 Current address: Department of Botany, Shankar Campus, Abdul Wali
Khan University, Mardan 23200, Pakistan.
3 Address correspondence to [email protected].
The authors responsible for distribution of materials integral to the
findings presented in this article in accordance with the policy described
in the Instructions for Authors (www.plantcell.org) are: Muhammed
Naseem ([email protected]) and Thomas Dandekar ([email protected]).
W
Online version contains Web-only data.
www.plantcell.org/cgi/doi/10.1105/tpc.112.098335
highlight the role of hormone molecules in plant immunity. In
particular, we provide a new perspective on auxin-cytokinin antagonism and insights into cytokinin-SA crosstalk. Our validated
dynamic model is a starting point in revealing hormonal implications for plant immunity.
In plants, pathogen attack causes hormonal perturbations
both at the host-pathogen interface (local response; Melotto
et al., 2006) and beyond (systemic response; Mishina and Zeier,
2007). Depending upon the nature of pathogen, either SAmediated (biotrophs; Glazebrook 2005) or JA/ET-mediated
(necrotrophs; Lai et al., 2011) defense pathways are operative in
plants (Grant and Jones, 2009). Antagonism of JA to SA and
synergism to ET has long been elucidated (reviewed in RobertSeilaniantz et al., 2011b). ET-based synergism is not limited
only to JA; rather, there exists a synergistic crosstalk between
pathways of ET and SA as well (Pieterse et al., 2009). Recent
investigations show that SA-JA/JA signaling pathways are interconnected, and if nodes are mutated or inhibited, the response is either shifted toward one or the other counterpart
(Sato et al., 2010). Growth regulatory hormones, such as auxin,
add to JA-mediated responses and suppress the SA pathway
(Wang et al., 2007). Likewise, ABA antagonizes SA-dependent
defense signaling both upstream and downstream of SA biosynthesis, whereas SA abolishes ABA responses (de Torres
Zabala et al., 2009; Cao et al., 2011). On the contrary, GA expedites SA accumulation (Navarro et al., 2008; Alonso-Ramírez
et al., 2009) and also promotes resistance against Pseudomonas syringae pv tomato DC3000 (hereafter referred to as Pst) by
degrading DELLA proteins (Navarro et al., 2008). Taking this
background carefully into account, we established an integrated
The Plant Cell Preview, www.aspb.org ã 2012 American Society of Plant Biologists. All rights reserved.
1 of 22
2 of 22
The Plant Cell
network of hormonal interactions to perform dynamic simulations on various aspects of plant immunity.
Regarding immunity, it is the collective temporal dynamic of
hormones during infection dictated by the ability of the host to
mitigate (Bari and Jones, 2009) and the pathogen to propagate
(Robert-Seilaniantz et al., 2007) that determines the outcome
of the plant–microbe interactions. Successful infections propagate through multiple phases, the simplest deconstruction being
entry (Melotto et al., 2008), suppression of basal defense (Boller
and He, 2009), reconfiguration of host metabolism (Rico and
Preston, 2008), and persistence in an equilibrium related to
a metabolic balance between host and pathogen (Rico et al.,
2011). However, it is not sufficient to collate a wealth of literature
and infer processes from this. We hence advocate and develop
a dynamic network model. By virtue of this, the zigzag model of
immunity (Jones and Dangl, 2006) can be studied in the context
of its hormonal dynamics, including semiquantitative information
on activity and the temporal sequence of activated nodes.
Moreover, hormonal poise is fine-tuned through various phases
of infection, and all of the hormones mentioned above participate in this process. Thus, JA, ABA, ET, and GA are equally
relevant in the infection process (Verhage et al., 2010) and can
be analyzed within the context of our simulation (see Methods;
see Supplemental Figures 1 and 2 and Supplemental Methods 1
online), but their detailed interplay is not our focus here (including, for example, their biochemical production as new output). We intensively analyzed our model with special reference
to cytokinin and the impact cytokinin has on various components of plant immunity.
The potential role of cytokinin in plant immunity in general
and a link to the SA-JA/ET backbone in particular is still unclear.
Apart from their role in tumor and gall formation (Tzfira and
Citovsky, 2006; Siemens et al., 2006) and generation of green
islands around the infection of fungal biotrophs (Walters and
McRoberts, 2006), cytokinin has not been assigned a conclusive role in plant pathogen interaction. Nevertheless, many
biotrophic and hemibiotrophic pathogens pattern their in planta
cytokinin secreting capability to increase uptake of nutrients and
host cell cycle regulation for self-establishment (Walters and
McRoberts, 2006; Pertry et al., 2009). Moreover, there are
reports about enhanced plant resistance to viral pathogens
in the context of increased cytokinin levels of the plant (Clarke
et al., 1998; Fayza and Sabrey, 2006). It has been shown that
cytokinin promotes resistance in Arabidopsis thaliana to the
infection of Alternaria brassicicola as well as Pst (Choi et al.,
2010). Moreover, increased resistance to infection by P. syringae
pv tabaci in tobacco (Nicotiana tabacum; Grosskinsky et al.,
2011) and reduction in the performance of herbivores in poplars
(Populus spp; Dervinis et al., 2010) have been associated
with higher levels of cytokinin. Recently, Argueso et al. (2012)
demonstrated enhanced and reduced susceptibility against
infection by Hyaloperonospora arabidopsidis (Hpa Noco 2)
in Arabidopsis with lower and higher cytokinin levels, respectively. Therefore, itis plausible that in plant cellular circuitry
cytokinin signaling has multiple interaction possibilities and
that each interaction has its own dynamics in plant pathogen
immune networks with responses that optimize plant defense
against the respective pathogen.
Broadly, biological networks are mathematical representations of biological structure where nodes are connected via
edges and thus constitute a graph (Albert, 2005). Based on the
type of interacting nodes, the following networks can be distinguished: metabolic (Schuster et al., 2000), protein–protein
interaction (Li et al., 2006), transcriptional regulation (Sato et al.,
2010), and signaling networks (Liu et al., 2010). Depending upon
the network, edges are either nondirectional or directed from
one node to the other. Edges depict processes, which require
time, context, and kinetics to occur (Pritchard and Birch, 2011).
Nodes of maximum connectivity are called hubs. They are of
different functional types, for instance “party” or “date” hubs
accumulating general or specific interactions regarding time and
type of interaction (Han et al., 2004) Depending on the exact
case, they can be of central importance for network structure as
well as biological function (Mukhtar et al., 2011). SA and DELLA
proteins are examples of functionally important hub nodes in our
network topology. Indicator nodes are densely connected, but
unlike a hub, indicator nodes such as pathogenesis-related
protein1 (PR-1) have minuscule impact on structural and functional orientation of the network but give an indication of the final
outcome of input stimuli.
Network-associated complexity can sometimes be captured
with parametric mathematical approaches, such as ordinary
differential equation (ODE) models. However, these require detailed kinetic data and other parameters (Wangorsch et al.,
2011). Alternatively, parameter-free qualitative approaches,
such as Boolean networks, can also model complex dynamic
behavior (Ay et al., 2009; Pomerance et al., 2009). “Boolean”
refers to dynamic models in which each node is characterized
by two qualitative states (often referred to as on or off) (Philippi
et al., 2009). Boolean network models have an advantage over
ODE-based kinetic models regarding complex networks including immune and pathogen responses (Wittmann et al.,
2009). In contrast with ODE models, Boolean network models
can also work when kinetic information is scarce and many
nodes are involved (Schlatter et al., 2011). SQUAD (Standardized Qualitative Dynamical systems; Di Cara et al., 2007) is
a powerful modeling package that combines Boolean and ODE
models. This approach is an extension of Boolean modeling.
It creates a system of exponential functions that allows interpolation between the step function of Boolean models according to the sum of activating and inhibitory input (Philippi
et al., 2009). It allows qualitative modeling of networks with the
added possibility of quantitative information. Using standardized
qualitative dynamic modeling, we analyzed plant hormone disease networks and performed simulations on pathogen- mediated perturbations in Arabidopsis.
In this study, we use dynamic modeling to better understand
the complex network interactions around plant immunity after
pathogen attack. We model the interactions following infection
by pathogen Pst in host plant Arabidopsis. Modeled responses
are consistent with reported experimental data. Predictions are
next extended to cytokinin and experimentally validated to show
points where cytokinin enhances plant immunity. Further analysis shows that protection mediated by cytokinin is due to its
promoting effect on SA signaling as well as inhibiting behavior to
auxin. We predict and show experimentally that cytokinin does
Modeling Plant Immunity
not influence the early events of recognition between host and
pathogen. Furthermore, looking at the established molecular
sequence of events, we study the promoting influence of cytokinin on camalexin accumulation. Taken together, a model and
experiments highlight in detail the challenged hormonal balance
for plant immune defense. This includes specific events around
modulation by cytokinin and implications for crop protection.
The model is instrumental in elucidating complex hormonal
crosstalk in plant pathogen interactions.
RESULTS
Boolean Formalism, Network Topology, and Initial
Conditions and Parameters for Model Simulations
Exploring their sequenced genomes, we modeled hormonal
attributes of interactions between Arabidopsis and Pst (phenotype is shown in Figure 1A). Known pathogenicity factors of Pst,
such as PAMPs and effectors (Boller and He, 2009), and their
interactions (Schneider and Collmer, 2010) with cellular components resulted in a network of 105 nodes and 163 edges with
logical (Boolean) connections (Figure 1B; see Supplemental
Table 1 online). Biosynthetic pathways of phytohormones
were integrated with defense signaling, immunologically important receptors, transcription factors, repressor molecules,
and degradative complexes. The model was implemented in
CellDesigner (Funahashi et al., 2008). We also considered the
primary literature for each and every edge and thus defined the
nature of interactions between nodes in our network topology
(see Supplemental Table 1 online). The behavior of an individual
node in our network is defined by a Boolean formalism, where
node is specified by a variable x, which denotes its state of
activation (x = 1) or inactivation (x = 0). However, the activation
of each component x depends on its overall interactions and
regulatory relationship within network cross-linking. Logical
operators OR (∨), AND (∧), and NOT (¬) in the SQUAD algorithm
cumulate the effect of edges: either only activating or only inhibiting or both activating and/or inhibiting (see Methods and
Supplemental Methods 1 online).
The static representation of the network was converted into
a discrete dynamical system for steady state analysis by applying the SQUAD software package and then into a continuous
dynamic system in the form of time-dependent ODEs (see
Supplemental Figure 1 and Supplemental Methods 1 online). To
assess the behavior of nodes across the network, we considered various conditions, such as presence or absence of the
pathogen (Pst; Figure 2A), wild-type or mutant nature of Pst (see
Supplemental Figure 2 online), presence or absence of a particular hormone (Figure 3A), full and partial activation (see below
and Supplemental Figure 3 online), and so forth. Simulation results on the carefully refined network (Figure 1B) well reflected
systems behavior according to literature (see Supplemental Table
1 online for nodes and type of interactions; see Supplemental
Table 2 online for simulation validation). The SQUAD simulation
approximates for this complex dynamic of system responses to
stimuli in a simplified way: It looks at system equilibria and its
changes. Simulation parameters were adjusted such that input
3 of 22
stimuli (changes of equilibrium) were set to be fully active [for the
input signal of virulent Pst infection: xt0 (Pst) = 1], while initial
states of other nodes (nodei) were set to no activation [being
in equilibrium: xt0 (nodei) = 0]. Moreover, SQUAD transforms the
original discrete step function into a sigmoid response curve.
The magnitude of the sigmoid function depends on the gain
(h) and decay (gi) in exponential terms (see Methods and
Supplemental Methods 1 online). Having fixed the initial conditions and parameters, a signal is injected as input node in the
form of invading pathogen (Pst), elicitors (flagellin or EF-Tu),
effector (HopIA1 or AvrRps2), or hormonal stimulus (auxin or
cytokinin). The route of the injected signal can be followed
through the network gates and then reaches the marker node of
PR-1. We defined PR-1 as marker node for the output as it reflects (Mukhtar et al., 2011) the resulting immunological response of the host plant to the invading pathogen (Pst).
Wherever appropriate, targeted experiments were performed to
support the outcome of modeling simulations. After careful
validation (see Supplemental Table 2 online), the established
network topology and dynamic model allowed us to predict the
impact of specific hormonal stimuli in plant pathogen interactions as detailed in the following.
Modeling of the Infection of Pst in Arabidopsis and Systems
Analysis for the Components of Plant Immunity
Plant pathogenic bacteria (Pst) have evolved strategies to keep
the host susceptible (Nishimura and Dangl, 2010). Whether
guarding the cell surface or being used for surveillance inside
the plant cell, receptors are instrumental in the recognition of
pathogenic PAMPs and effectors (Figure 1B, red). These recognition events evoke changes in system states of the cell.
Therefore, we modeled virulent infection of Pst (fully active input
node; see network topology in Figure 2A) and determined resulting activation states for individual nodes across the whole
network (Figure 2A and see below; see Supplemental Figure 2A
online). However, a less virulent pathogen shows substantially
lower activation states (see Supplemental Figure 2B online).
Furthermore, immunity is the sum total of various components
(Jones and Dangl, 2006), such as PAMP-triggered immunity
(PTI) and effector-triggered immunity (ETI). To model PTI, we
performed SQUAD simulations such that flagellin and EF-Tu
(nodes of network topology; flagellin and elongation factor;
Figure 1B) being input nodes were given full activation (1.0 as
activation value during simulations). This resulted in the activation of NADPH-oxidase, mitogen-activated protein kinases,
WRKY transcription factors, miR393 (suppressor of auxin responses; Navarro et al., 2006), PR-1, and closure of stomata
(Figure 2B). Moreover, activation (concentration/response) of
nodes representing phytohormones, such as SA (Mishina and
Zeier, 2007) and ABA (ABA responses; see Melotto et al., 2006;
Nishimura and Dangl, 2010; Zeng and He, 2010), as well as
activation of DELLA proteins (flagellin treatment leads to the
stabilization of DELLA protein complexes; Navarro et al., 2008)
is evident in PTI modeling (Figure 2B; see Supplemental Figure
4B online). To model ETI, we assigned full activation to nodes of
bacterial effectors (Figure 1B, red), which subsequently resulted
in the activation of relevant nodes across the whole network
4 of 22
The Plant Cell
Figure 1. Network Topology for Small-Molecule Hormone Disease Networks.
(A) Phenotype: Pst (104 colony-forming units/mL) infection; red arrows indicate pathogen inoculation; black arrow indicates mock (10 mM MgCl2)
inoculation. Symptoms were photographed 3 d postpathogen inoculation (DPPI).
(B) Network: Topology of Pst-mediated hormone disease networks in Arabidopsis. Connectivity among nodes is based either on activation (→) or
inhibition (─┤). Node designation: blue, enzymes of hormone biosynthesis and degradation; yellow, active hormone molecules; green, host regulatory
factors; red, Pst-originated pathogenicity factors responsible for triggering immunity in Arabidopsis; pink, PR-1, marker node for immunity against the
infection of Pst in Arabidopsis. All nodes are denoted by abbreviations (see Supplemental Figure 8 online for full names and Supplemental Table 1 online
for internodal interactions and literature on the behavior of these interactions).
(Figure 2B; see Supplemental Figure 4B online). Network analysis shows extensive overlap (70%; see Supplemental Figure 5
online) between PTI and ETI, whereas system stability analysis
(difference in states for ETI and ETI* over time) revealed robustness as well as redundancy in ETI (Figure 2B, middle and
bottom panels). Besides inhibiting PTI (see Supplemental Figure
4B online), the modeling suggests that bacterial effectors alter
the hormonal profile of the plant cell (Figure 2B; see
Supplemental Figure 4A online). Pst modeling results agree with
current literature (see Supplemental Table 2 online). These results demonstrate the behavior of the components of plant immunity and their innate properties and quantify their mutual
interactions across the hormone disease networks. Moreover,
our model envisions the behavior of Pst infection as a source of
hormonal modulation in plants.
Small-Molecule Hormones Modulate Plant
Immunity: Cytokinin Crosstalk
Our dynamic model for Pst showed activation for nodes of SA,
JA, ET, ABA, and auxin, while demonstrating lack of activation
for GA and cytokinin (Figure 2A; see Supplemental Figure 4A
online). Lack of activation is not equivalent to repression. Repression requires a signal to go from on to off. To model the
Modeling Plant Immunity
5 of 22
Figure 2. Infection Modeling of Virulent Pathogen Pst and Systems Analysis for Components of Plant Immunity.
(A) Pst as fully active input node (with activation value of 1.0, red dotted in each four graphs): The SQUAD simulation assigns activation states to nodes
using our assembled network (Figure 1B) to calculate response curves. Fully virulent pathogen (Pst DC3000) besides other nodes activates SA, auxin,
JA, ET, ABA, and PR-1, while cytokinin received no activation. x axis, arbitrary units of time (SQUAD simulation is shown in 12.5 steps); y axis, activation
states ranging from no (0) to full (1) activation. Each trajectory represents an individual node. Color codes with their corresponding node names are
shown at right side of each graph (full names are shown in Supplemental Figure 8 online; see Supplemental Data Set 2 online for a comprehensive list of
nodes and their respective activation over time upon the infection of Pst).
6 of 22
The Plant Cell
impact of these hormones on plant immunity, we performed
SQUAD simulations by taking individual hormones as input activating node with and without Pst. The activation of PR-1 over
time was used as an index of plant immunity. We found in the
simulations that the three hormones ET, SA, and GA activate
PR-1 in contrast with JA, ABA, and auxin (Figure 3A, left panel).
Interestingly, SA and GA further enhance the signal of the activity of PR-1 in the presence of Pst, whereas JA, auxin and ABA
diminish even the residual activity of PR-1 manifested by Pst
alone (Figure 3A, right panel). This supports the fact that unlike
the promoting effect of SA and GA on immunity, auxin, JA, and
ABA mediate susceptibility of Arabidopsis against infection by
Pst. The simulation data in fact allow us to distinguish whether
a signal goes from on to off (repression in an active sense) or
whether there is simply lack of activation (see Supplemental
Figure 4A online). In most cases, we observed only lack of activation (e.g., mitigated pathogen for most system nodes modeled). We validated these and further outcomes of SQUAD
simulations on hormonal stimuli by detailed comparison with
published data (see details in Supplemental Table 2 online).
Although implicated both in resistance and susceptibility
(Pertry et al., 2009; Argueso et al., 2012), the phytohormone
cytokinin seems to have multifaceted signaling roles in plant
diseases. Regarding their connection to the SA-JA/ET backbone
of plant immunity, Choi et al. (2010) reported positive crosstalk
between cytokinin and SA pathway, whereas Argueso et al.
(2012) found negative crosstalk among them. However, their
interaction with the JA pathway is still not known. Our own
observations from both experiments and modeling intrigued us
to investigate the specific crosstalk by cytokinin as this seemed
to enhance immunity in plants. We further investigated all these
observations by setting up and applying our integrated model
and can now provide insights into cytokinin-mediated resistance in Arabidopsis to infection with Pst. Modeling infection by a virulent pathogen gave no indication of a direct
activation of the node representing cytokinin (Figure 2A; see
Supplemental Figure 2A and Supplemental Figure 4A online).
However, when set to be fully activated together with Pst as
input activating nodes (modeling the increased cytokinin
status of the plant during pathogen infection), the SQUAD
analysis (besides Pst-based node activations; Figure 2A) resulted in the activation of nodes such as Arabidopsis Histidine
Kinases, Arabidopsis homologs of HPt Proteins, Arabidopsis
Response Regulators (ARRs), Cytokinin Oxidase, and PR-1
(Figure 3B; see Supplemental Figure 3A online). Subsequent
simulations also highlighted that cytokinin (without Pst) alone
can cause the activation of PR-1 and hence promote plant immunity (see Supplemental Figure 3B online). To verify modeling
predictions, we applied exogenously kinetin solution (10 µM)
and observed clearly reduced Pst disease symptoms compared
with mock-fed pathogen-infected Arabidopsis leaves (Figure
3C, left panel). Moreover, reduced in planta bacterial multiplication (Figure 3C, right panel) in cytokinin treated versus control
leaves nicely confirms the simulation. Regarding the cytotoxic
effect of cytokinin on Pst, we demonstrated experimentally that
unlike the antibacterial activity of tetracycline, cultures of Pst
grew equally well with and without the addition of cytokinin (see
Supplemental Figure 6 online). Additionally, feeding of kinetin
(without subsequent pathogen infection) solution to detached
leaves of Arabidopsis PR1:GUS (for b-glucuronidase) plants
showed higher GUS activity compared with mock feeding (see
below). These results substantiate the fact that enhanced plant
cytokinin levels boost immunity against infection of Pst and that
cytokinin has no direct cytotoxic effect on bacterial multiplication.
Cytokinin Overlay with SA in Regulating Genes of
Arabidopsis Implicated in Plant Immunity
Cytokinin-mediated protection of Arabidopsis against infection
by Pst (Figure 3C) and its activation of PR-1 (see Supplemental
Figure 3B online; see below) pinpoint crosstalk between SA and
cytokinin pathways. To explore the mechanism of cytokininmediated protection, we investigated genome-wide transcriptional overlay between cytokinin and SA and the impact they
have on the expression of genes related to immune defense in
plants. We analyzed microarray data from the Gene Expression
Omnibus (GEO) on hormonal stimuli of SA (GSE3984; ThibaudNissen et al., 2006) and cytokinin (GSE 6832; Hitoshi Sakakibara
RIKEN Plant Science Center). After normalization and statistical
analysis, the up- and downregulated genes of both these hormonal stimuli are shown in a Venn diagram (Figure 4, left side).
We found overlap (an indication of cytokinin-SA crosstalk) between SA- and cytokinin-regulated (regarding both up- and
downregulated) genes. The overlapping genes are shown in the
heat map representation, indicating their expression together
with their annotation (Figure 4, right side). Immunity-relevant
genes include those encoding glutaredoxins, chitinase, and the
well-known marker PR-1 as an integral part of SA-mediated
defense (Bari and Jones, 2009). These were found in the gene
Figure 2. (continued).
(B) Modeling of PTI and ETI. Top, node names; red arrows: elicitors. PTI system states: Giving full activation (1.0) to elicitors, flagellin and elongation
factor, downstream components, such as FLS2, BAK1, MAPKs, SA, NPR1, TGA-TF, and WRKYs, and PR-1 achieved various states of activation in
SQUAD analysis. These are shown as a heat map in terms of activity over time (early time points above and corresponding later time points follow in
each heat map). For details on global activation of nodes across the network as a consequence of PTI, see Supplemental Data Set 2 online. ETI system
states: All effectors (top left nodes with black asterisks) in the network are given full activation in SQUAD analysis. Robustness in ETI*: HopAI1 (red
asterisk at the bottom left) was deleted from the network. This did not significantly change the overall dynamics of components involved in ETI and
particularly the marker node PR-1 time course. ETI is a stable form of the immunity. Comparison between ETI and PTI reveals that both operate through
overlapping components. See Supplemental Data Set 2 online for details on states during ETI and ETI* and Supplemental Figures 5 and 7 online for
system stability and robustness.
Modeling Plant Immunity
7 of 22
Figure 3. Modeling the Impact of Plant Hormones on the Pathogenicity of Pst.
(A) Plant hormones change the immunity of the host. States of activation of PR-1 over arbitrary units of time (top to bottom) are shown as a heat map for
plant hormones with (right panel) and without (left panel) the infection of Pst. Fully activated (by assigning 1.0 as level of activation) nodes of Pst and
plant hormones as activating input signal change the state of immunity in the host. System states around the marker node of PR-1, as consequence of
Pst infection with and without hormone application, are shown in detail in Supplemental Data Set 6 online.
(B) Impact of cytokinin on infection by Pst in Arabidopsis. Simultaneous activation of the Pst and cytokinin nodes (top left corner; assigned full activation
of 1.0 to Pst and cytokinin during SQUAD simulation) results in the activation of nodes of the assembled network. Left y axis: the states of input signals
of Pst infection alone (red line with black dots), with cytokinin (green line with red dots) and with auxin (black line with red dots). Right y axis: activity of
PR-1 as output of Pst (red curve), Pst with cytokinin (green curve), and Pst with auxin (black curve). The x axis denotes arbitrary units of time. Additional
nodes activated due to the effect of cytokinin are given with their maximum activation states in Supplemental Data Set 6 online.
(C) Higher cytokinin levels enhance resistance to infection by Pst. Left: Arabidopsis leaves treated with kinetin (10 mM) 24 h prior to Pst inoculation (105
colony-forming units [CFU]/mL) manifest resistance to the infection (green leaves in left panel; bottom row); control leaves received mock feeding of 10
mM MgCl2 (affected leaves in left panel; top row). Symptoms were photographed 4 DPPI. Similar results were obtained in three independent experiments. Right: Higher cytokinin status (green line) reduced the growth of bacteria (red arrow). Hormone feeding was done as in the left panel. For
bacterial quantification (logarithmic y axis), samples were collected at 0, 1, 2, and 3 DPPI. Asterisks indicate statistically significant response to
pathogen (P < 0.05, t test, and 6SE, n = 5). The experiment was repeated three times with similar results.
8 of 22
The Plant Cell
Figure 4. Positive Interaction between SA and Cytokinin Signaling and Its Impact on Plant Immunity.
SA and cytokinin treatment were compared using Affymetrix whole-genome Gene Chip arrays for gene expression analysis: (1) Arabidopsis ecotype
Col-0 3-week-old leaves were sprayed with 1 mM SA (left column, top panel, box plots of log intensities; obtained from GEO; gene expression omnibus
database; ID GSE 3984) and (2) 2-week-old aerial parts of Arabidopsis ecotype Col-0 plants were treated with 20 mM t-zeatin (left column, bottom panel;
GEO ID GSE6832). GEO-deposited respective replicates of raw data for both these experiments were independently analyzed. Synergism: SA-cytokinin
coregulated genes are presented in a Venn diagram (left column, middle panel; tool: http://bioinfogp.cnb.csic.es/tools/venny/index.html). SA-cytokinin
coregulated upregulated genes (URGs) and downregulated genes (DRGs), their gene IDs and their potential implications in plant immunity are shown in
the right column. SA- and cytokinin-regulated expression levels (see Supplemental Data Set 7 online for statistical confidence: t-values, P values, and
fold changes) are represented as a heat map (Expression Browser Software; http://bbc.botany.utoronto.ca). The scale bar (bottom of right column)
shows gene expression (up- and downregulation) in terms of t-values.
expression analysis to be mutually upregulated both by SA and
cytokinins. Furthermore, our analysis revealed the RPS4 gene
involved in resistance to Pst to be downregulated by methyl
jasmonate upregulated by both cytokinin and SA. The mRNA
for a chitinase-type broad-spectrum mildew resistance protein
MLO11, as well as mRNAs for defensin-related genes, were
found to be upregulated. Similarly, a couple of Toll/interleukin-1
receptor domain–containing genes were shown to be mutually
upregulated by SA and cytokinin. The gene for JAZ1, a positive
regulator of SA responses and CONSTITUTIVE DISEASE
Modeling Plant Immunity
RESISTANCE1, is also upregulated by both these hormones.
Interestingly, more than one and half dozen disease resistance
protein genes belonging either to the coiled-coil–nucleotide
binding site–leucine-rich repeat or Toll/interleukin-1 receptor–
nucleotide binding site–leucine-rich repeat class (Mukhtar et al.,
2011) were found to be mutually upregulated by SA and cytokinins.
The genes encoding the MYC2 transcription factor, a negative
regulator of SA responses (Figure 4, bottom right panel; LaurieBerry et al., 2006), and MYB28, which gives protection against
necrotrophic pathogens (Stotz et al., 2011), are downregulated
by cytokinin and SA. Auxin responses promote susceptibility
(Navarro et al., 2006) against biotrophic pathogens; one such
auxin-responsive protein gene (AT2G37030) and a gene
encoding an enzyme of the auxin biosynthetic pathway
(Moubayidin et al., 2009) were found to be repressed both by
cytokinin and SA. ABA mutants impaired in ABA biosynthesis or
sensitivity are thought to be resistant to infection with Pst
DC3000 (de Torres Zabala et al., 2009), and ABA1 is mutually
downregulated by cytokinin and SA. DELLA proteins are negative regulators of SA-mediated resistance against Pst DC3000
(Navarro et al., 2008), and two DELLA-like genes, RGL3 and
RGL2, are found to be repressed both by cytokinin as well as
SA. The axr1 Arabidopsis mutant showed increased resistance
to Pst (Kazan and Manners, 2009). AXR1 was analyzed and
found to be mutually downregulated by SA and cytokinin. Taken
together, these data demonstrate that cytokinin-SA positive
crosstalk is not limited to a disease marker gene (Choi et al.,
2010) but has wide transcriptional overlap.
Modeling the Behavior of Cytokinin Regarding the SA
Pathway and Camalexin Accumulation
Conventionally, resistance to (hemi)biotrophic pathogens has
often been associated with elevated SA levels coupled to the
upregulation of PR-1 in plants (such as tomato [Solanum lycopersicum], Arabidopsis, tobacco; Nawrath and Métraux, 1999).
Therefore, we investigated the impact of enhanced cytokinin
levels on the accumulation, as well as downstream signaling, of
SA. SQUAD simulations for cytokinin (without Pst) as an input
node resulted in the activation of PR-1 without activating the
node of SA (see Supplemental Figure 3B online). However, when
cytokinin was combined with Pst as input activating node, the
activation of PR-1 increased beyond their individual effects,
while node of SA received a minor (Pst alone 0.28 to Pst + cytokinin 0.32) activation (Figures 2A and 3B; see Supplemental
Figure 3A online). Our analysis of free SA levels did not reveal
any significant difference for cytokinin- versus mock-treated
leaves (Figure 5A). However, early time points (12 h) prior to
cytokinin treatment followed by infiltration with Pst resulted in
a higher SA accumulation in comparison to mock, as well as
cytokinin, treatments. Moreover, the difference in free SA accumulation between treatments of Pst with and without cytokinin was not significant (Figure 5A). We also analyzed the
synthesis of camalexin in Arabidopsis Columbia-0 (Col-0) plants
and found that cytokinin-treated plants accumulated significantly
more camalexin than did control plants (Figure 5B). Enhanced
camalexin accumulation upon cytokinin treatment occurred in
9 of 22
late time points (72 h after cytokinin application). These results
signify that cytokinin does not cause higher SA accumulation
to protect Arabidopsis against the infection with Pst and that
cytokinin treatment has a positive effect on camalexin accumulation.
We performed SQUAD simulations to investigate to what degree cytokinin-mediated protection against Pst is independent of
the accumulation of SA and to investigate whether the node of
interaction is downstream from the synthesis of SA. We used
cytokinin as the activating input node in combination with Pst in
the presence (+SA; to mimic overaccumulation of SA) and absence (2SA; to mimic deficiency in SA accumulation) of SA,
while activation of PR-1 over time was used as the output
marker node for the immune response (Figure 5C). Individually,
activation of SA in comparison to cytokinin strongly activated
PR-1 during infection by Pst, while combined activation of SA,
cytokinin, and Pst further augmented the PR-1 signal (Figure 5C,
top three rows). Interestingly, deletion of SA from the network
(this mimics the state of the sid2 mutant) brought the PR-1
signal back to the level of Pst with cytokinin. However, deletion
of TGA3-TF efficiently reduced this signal (Figure 5C). To verify
the computational analysis, experiments were performed on
sid2 mutants (SA-deficient plants; Wildermuth et al., 2001). Prior
to the inoculation with Pst, exogenous application of kinetin to
sid2 plants resulted in a less severe disease phenotype in
comparison to mock treatment (Figure 5D). Quantification of the
protective effect (Figure 5D) of cytokinin on sid2 plants showed
that this happens only at a low initial pathogen load (see below).
Thus, both modeling and experimental data substantiate that
the interaction between cytokinin and SA is down the node of
SA biosynthesis, while higher SA levels only fortify the cytokininmediated system state with better protection against infection
with Pst.
Modeling the Impact of Cytokinin on PTI
Cytokinin augments PTI in Arabidopsis by enhancing callose
deposition upon flagellin treatment (Choi et al., 2010). Moreover,
cytokinin upregulates PR-1, which is regarded as a downstream
event (Mishina and Zeier, 2007) in PAMP–Pattern Recognition
Receptor interaction. However, it is still unknown if cytokinin
influences early events of pathogen recognition (Zipfel et al.,
2006) in plants. To model the impact cytokinin has on early
events of PAMP recognition, we performed SQUAD simulation
while taking flagellin as input activation node in the presence
and absence of cytokinin as well as cytokinin alone as input
stimuli. Our dynamic model contains all relevant nodes and
models the time evolution of stimulatory input regarding the
whole network. We can thus predict and distinguish early versus
late events in the activation as well as differences in activation
by flagellin and/or cytokinin and resulting responses for any
other network node of interest. Figure 6A shows for the aforementioned input stimuli the state of key nodes representing
PR-1, NADPH-Oxi, and reactive oxygen species (ROS) with regard to their direct relevance to recognition events (Panstruga
et al., 2009). Cytokinin alone does not cause the activation of
NADPH-Oxi and ROS but activates marker node PR-1 (Figure
6A). Furthermore, together with flagellin, cytokinin changes the
10 of 22
The Plant Cell
Figure 5. Effects of Cytokinin on the SA Pathway of Resistance in Arabidopsis.
(A) Accumulation of free SA: Leaves of Arabidopsis Col-0 plants were compared with and without Pst and cytokinin feeding. The 12-, 24-, and 72-h time
points indicate differences in SA accumulation: Different letters denote statistically different values (P < 0.05, t test, and 6SE, n = 3). The experiment was
repeated twice with similar results. FW, fresh weight.
(B) Accumulation of camalexin: Leaves of Col-0 plants after 10 mM kinetin feeding were analyzed for camalexin accumulation at 12, 24, and 72 h. Kinetin
treatment achieved significant camalexin accumulation at the 72-h time point (asterisk, P < 0.05, t test, n = 3, 6 SE); similar results were obtained in two
independent experiments.
(C) Impact of cytokinin on the SA sector of resistance. Modeled activation of PR-1as resistance marker node (see text) is shown as a heat map over
arbitrary units of time (from left to right) comparing (top to bottom) Pst with cytokinin, Pst with SA, Pst with both cytokinin and SA, Pst with cytokinin and
removal of the SA node (to mimic the effect of a SA-deficient mutant, like sid2 mutant), and Pst with cytokinin and removal of the TGA-TF node (to mimic
the effect of mutations such as a TGA mutant). Details on these effects are also shown in Supplemental Data Set 8 online.
(D) Impact of cytokinin and auxin on Pst infection in SA-deficient mutants. sid2 mutant plants were sprayed with 10 mM kinetin (top, left panel), mock
treatment to sid2 plants (top right; control to hormone treatments), and 10 mM NAA, and sid2 (bottom left) is compared with sid2 sprayed with mock
solution (right, top panel) as well as to Col-0 with kinetin treatment (bottom, right). All plants were spray inoculated with Pst (5 3 105 colony-forming
units/mL) in 10 mM MgCl2 containing 0.02% Silwet L-77, 24 h after kinetin or mock treatments. Symptoms were photographed 3 DPPI. The experiment
was repeated three times.
Modeling Plant Immunity
activation state of the PR-1 marker node without influencing the
states of NADPH-Oxi and ROS (Figure 6A). Moreover, the corresponding calculated detailed activation states over time for
the impact of cytokinin on flagellin-mediated ROS production
are shown in Supplemental Data Set 1 online.
To validate these simulations regarding the impact of cytokinin on early events of flagellin/PAMP recognition (Figure 6A),
we performed experiments on flagellin elicitation of ROS production in Arabidopsis leaf fragments. Upon treatment with 10 nM
flagellin, leaf fragments produced ROS within 3 min, reached
a maximum level, and then returned back to the initial state
(Figure 6B). Addition of t-zeatin (1 µM) did not change the
flagellin-mediated amplitude of ROS production (Figure 6B) in
Arabidopsis leaf fragments. To get further experimental insight,
we performed pH experiments (Kunze et al., 2004) on Arabidopsis cell suspension cultures and showed that ΔpH (change in
medium alkalinization) for flagellin did not change with the addition of t-zeatin. Furthermore, we did not notice any elicitation
effect of t-zeatin alone on the cultures used in such experiments
(Figure 6C). These experiments were repeated with various
types and concentrations of cytokinin and similar results were
obtained (data not shown). Taken together, both modeling and
experiments underpin the importance of cytokinin in promoting
PTI but show that they seem not to influence the early (PAMP–
Pattern Recognition Receptor) events of pathogen recognition
via ROS production and medium alkalinization.
Contrasting Behavior of Auxin and Cytokinin in Plant
Immunity: Transcriptional and Physiological Evidence
Pst DC3000 injects its effector weaponry and hormonal
mimicry to render the host more susceptible (Pieterse et al.,
2009). Compromised virulence occurs when Pst is kept deprived of both these tools (Thilmony et al., 2006). Our virulent
infection model revealed that during the course of infection,
Pst causes the activation of auxin concentration/response but
not that of cytokinin (Figure 2A; see Supplemental Figure 4A
and Supplemental Data Set 2 online). A Pst mutant (deficient in
Type three secretion system) grew well in cytokinin-deficient (35S:
Cytokinin Oxidase4,) plants compared with higher cytokinin–
producing (35S:IPT3) and wild-type Col-0 Arabidopsis plants
(Choi et al., 2010). We therefore hypothesized that virulent and
avirulent strains of Pst differentially manipulate auxin and cytokinin pathways in plants. The mutated pathogen could be considered to represent a mitigated infection, and we compared our
modeling predictions with the gene expression data. To validate
this notion at the transcriptional level, we retrieved raw but
complete (genome-wide) microarray data on the infection of the
Pst wild type (Pst DC3000) and Pst coronatine and hp mutant
(Pst Cor– hp2; Thilmony et al., 2006) in Arabidopsis from the
GEO database (ID GSE5520; Thilmony et al., 2006) (Figure 7A).
Among the significantly regulated genes, we focused only on
genes related to the biosynthesis, signaling, transport, or degradation of auxin or cytokinin.
The auxin pathway genes downregulated upon the infection
with Pst included auxin transporters (PIN3, PIN4, PIN7, AUX1,
LAX1, and LAX3) auxin repressors (AUX/IAAs), and small auxin
upregulated (SAUR) genes. However, upon the infection with the
11 of 22
mutant pathogen Pst Cor– hp–, the above genes were either
slightly upregulated or remained unchanged (Figure 7A). The
upregulated genes included those that convert amino acid
conjugates of indole-3-acetic acid (IAA) to free IAA (e.g., genes
encoding IAA–amino acid hydrolases [ILR1, ILL5, and ILL6] and
genes for IAA-amido synthases [DFL2, GH3.3, and GH3.12]). In
contrast with wild-type infection, infection with the mutant Pst
strain caused either repression of these genes or left them unchanged. Genes implicated in the biosynthesis of IAA, such as
IAA18 and NIT3, are derepressed upon the infection with Pst
(Figure 7A). Genes that are negative regulators (Muller and
Sheen, 2007) of cytokinin signaling, such as type A response
regulators (ARR4, 5, 6, 7, 9, 15, and 16), are repressed during
the infection with Pst. However, the contrary is the case for the
Pst Cor– hp– pathogen in terms of expression. Moreover, the
IPT3 gene responsible for cytokinin biosynthesis is downregulated by Pst and not regulated by the mutant bacterial
counterpart (Figure 7A). In contrast with the infection by Pst Cor–
hp–, we find that genes encoding enzymes of cytokinin degradation are upregulated by wild-type Pst in Arabidopsis (Figure
7A). Taken together, during infection, Pst promotes auxin accumulation by upregulating biosynthesis and deconjugation
enzymes genes, as well as by repressing transporters. Auxin
signaling is enhanced by the repression of auxin repressors. On
the other hand, repressing a gene encoding an enzyme of cytokinin biosynthesis and derepression of the genes for cytokinin
degradation enzymes reduces the levels of cytokinin upon
pathogen infection.
We next studied the impact of the infection of Pst on Arabidopsis promoter-reporter lines (AtDR5:GUS and AtARR5:GUS),
designed to show responses/concentrations of auxins and cytokinin, respectively (Siemens et al., 2006; Liu et al., 2010). Upon
inoculation of AtDR5:GUS plants with Pst, we found substantially stronger GUS activity at the host-pathogen interface,
followed by faint staining in the surrounding tissue of the same
leaf and compared this to mock-inoculated leaves (Figure 7B).
Pst-inoculated leaves of the AtARR5:GUS line manifested low
GUS activity at the site of infection, while mock-inoculated
leaves showed residual background staining (Figure 7B). We
determined the accumulation of auxin and cytokinin with gas
chromatography–mass spectrometry and found significantly
higher free IAA levels at the site of infection of Pst compared
with the control (Figure 7C). On the other hand, the level of
t-zeatin at the site of infection was significantly lower than in
mock-inoculated leaves (Figure 7C). These experiments complement the virulent pathogen infection modeling (Figure 2A)
and gene expression with Pst infection in Arabidopsis (Figure
7A) and substantiate that both elevated auxin and repressed
cytokinin are part of the infection process that Pst mediates in
Arabidopsis.
SQUAD simulation provided further insight into the contrasting behavior of auxin and cytokinin in plant immunity.
Auxin together with Pst as input activating nodes resulted in
the reduction of PR-1 signal in comparison to Pst alone (Figure
3A). The simulation predicted reduced immunity upon the
application of auxin (Figure 7D). Prior to Pst inoculation,
1-naphthaleneacetic acid (NAA)–fed (10 mM) Col-0 Arabidopsis
leaves manifested severely enhanced symptoms compared with
12 of 22
The Plant Cell
Figure 6. The Impact Cytokinin Has on PTI and Early Events of Recognition.
(A) Modeling the impact of cytokinin on PTI. SQUAD simulations for the effect of flagellin treatment with (middle panel) and without (left panel) cytokinin
as well as cytokinin alone (right panel) were performed to study the impact of cytokinin on PTI and early events of pathogen recognition. States of
activation over time (from top to bottom) for nodes NADPH-oxidase, PR-1, and ROS are shown as a heat map; also see Supplemental Data Set 1 online
for data on activation of more nodes of interest in our network on flagellin treatment, with and without cytokinin.
(B) Cytokinin does not change dynamics of the oxidative burst achieved with flagellin in Arabidopsis leaves. ROS production in leaf fragments of
Arabidopsis with/without the incubation of 1 nM of flag 22 as well as with and without 1 mM t-zeatin (four conditions) was measured as luminescence in
counts per second (y axis) over time (x axis). Results are averages 6 SE (n = 5).
(C) Flagellin-based media alkalinization. Induction of medium alkalinization in Arabidopsis cell culture due to flagellin does not change with the addition
of 1 mM t-zeatin (see Methods). Cultures with/without flagellin as well as with and without 1 mM t-zeatin were assayed for alkalinization-inducing activity,
and the change in extracellular pH in Arabidopsis cells is shown on the y axis and over time on the x axis. Results are averages 6 SE (n = 5).
corresponding controls (Figure 7E and see below). This substantiates that auxin promotes susceptibility of Arabidopsis to
Pst, and cytokinin does not have this effect (Figure 3C). By virtue
of their positive crosstalk to SA (down the node of synthesis)
cytokinin promotes resistance against the infection with Pst
(Figures 5C and 5D). To study how auxin influences the SA
pathway of resistance, we modeled this interaction with SQUAD
simulation. We monitored the activation of the marker node PR1 in the presence of pathogen as well as modulated (partial or
full activation) conditions of SA versus auxin. Together with Pst,
the signal for PR-1 activation was higher when SA was given full
activation and auxin was kept either partially active or inactive
(Figure 7D, middle panel). Decay in the PR-1 signal resulted
when SA was either partially or completely inactivated while
auxin was given full activation (Figure 7D, right panel). This implies that a decrease in the level of plant SA and increase in
auxins further enhance vulnerability. To support this, we demonstrated that sid2 mutants exhibited severe Pst disease phenotypes when given exogenous NAA (Figure 5D). These results
demonstrate that unlike cytokinin, which promotes the SA
sector of immunity, auxin counterregulates SA responses in
favor of susceptibility to infection by Pst in Arabidopsis.
Auxin–Cytokinin Interaction: A New Perspective of
Antagonism in Plant Immunity
Auxin and cytokinin have long been known to interact in growth
and development of the plant (Liu et al., 2010). Furthermore,
auxin promotes Arabidopsis susceptibility to infection with Pst
by its positive and negative crosstalk with JA and SA, respectively (Navarro et al., 2006; Chen et al., 2007; Wang et al.,
2007; Llorente et al., 2008). On the other hand, cytokinin
Modeling Plant Immunity
13 of 22
Figure 7. Signaling of Plant Hormones Auxin and Cytokinin during the Course of Infection by Pst in Arabidopsis.
(A) Raw microarray data: Genes related to the biosynthesis, degradation, and signaling of auxin and cytokinin are presented here; gene IDs (left row),
gene names (middle row) and t-values (color code) in the form of a heat map are given (right row). Data on the infection of Pst versus mock and PstCOR.Hrp versus mock were obtained from GEO (database of the National Center for Biotechnology Information: ID GSE5520). Four-week-old leaves of
ecotype Col-0 of Arabidopsis were infected with the wild type and mutants of Pst DC3000. GEO2R-based analysis led to the identification of significantly regulated genes. Further details on expression analysis, t-values, and P values are in Supplemental Data Set 7 online.
(B) Auxin and cytokinin reporter lines: Auxin (top panel) and cytokinin (bottom panel) reporter lines were infiltrated with Pst (106 colony-forming units/mL,
red arrows). The 3 DPPI leaves were GUS stained and photographed afterwards. As a control, plants were treated with 10 mM MgCl2 (black arrows).
Similar results were obtained in five independent experiments.
(C) Status of auxin and cytokinin during infection with Pst DC3000. Leaves were infiltrated with Pst (106 colony-forming units/mL). Three DPPI, infected
areas surrounding healthy tissue were sampled for hormone analysis. As a control, plants were treated with 10 mM MgCl2. The values represented are
the average for three replicate samples (6SE). Asterisks indicate statistically significant hormone accumulation: The difference between Pst treatment
and mock is significant. For IAA, Pst treatment causes significant increase in IAA accumulation compared with mock treatment. For cytokinin, Pst
14 of 22
The Plant Cell
promotes plant immunity against Pst (Figure 3). With the spotlight on their individual roles, the combined effect of auxin and
cytokinin in plant immunity is still not known. To address this
more closely, we modeled the combined effects of auxin and
cytokinin on plant immunity. To mimic modulation in levels of
auxin and cytokinin, we chose various input activation values
(partial and full activation) for nodes of auxin and cytokinin
without Pst (Pst is also capable of increasing in planta auxin
concentrations; Figures 7B and 7C) while performing SQUAD
simulations. Partially and fully activated nodes of auxin and
cytokinin as a combined input signal did change system states
across the whole network (see Supplemental Data Set 3 online).
Furthermore, we monitored states of marker node PR-1 as an
indicator of immunity status and presented it as trajectory
over time for each hormonal combination (Figure 8A). Input
activations of higher (1.0 as input activation) auxin and lower (0.2
as input activation) cytokinin resulted in low PR-1 activity. When
this combination was reversed, PR-1 gained more activation
(Figure 8A). Full to partial input activation of auxin resulted in
low PR-1 activity, whereas partial to full input activation of
cytokinin resulted in an increase in the state of PR-1 (Figure 8A).
Furthermore, using SQUAD simulation, we determined values
of PR-1 activation for combination of auxin and cytokinin at
various input activating values (ranging from 0 to 1). Threedimensional visualization of the data provided an overview
showing that a relative shift of activation from auxin to cytokinin
resulted in the activation of PR-1, whereas the reverse is observed when the trend of activation was directed toward auxin
(Figure 8A, bottom panel). More precisely, immunity is compromised when auxin supersedes the activation of cytokinin,
and immunity is achieved when cytokinin attains higher activation than auxin.
Pst inoculation enhances de novo auxin synthesis in Arabidopsis and reduces cytokinin levels at the site of infection
(Figures 7B and 7C). However, to demonstrate the impact of
interaction between auxin and cytokinin on plant immunity,
we performed experiments (exogenous application of hormones) on PR1:GUS Arabidopsis transgenic plants. PR1:GUS
plants were induced (via petiole feeding of hormones) with kinetin and NAA alone, as well as in combination without inoculation of Pst. Following hormonal induction, leaves were
subjected to histochemical GUS staining. Kinetin-treated PR1:
GUS leaves showed higher GUS activity, while NAA treatment
alone did not induce any GUS activity (Figure 8B). However, the
combination of kinetin and NAA manifested low GUS activity
compared with kinetin alone. Taken together, our model and
experimental data substantiate that a balance between auxin
and cytokinin modulates plant immunity against infection with
Pst in Arabidopsis.
To confirm that a balance between auxin and cytokinin
affects plant immunity during infection of Arabidopsis by Pst,
we next performed simulations using auxin and cytokinin (in
various input activations) as input nodes in the presence of Pst.
Activation of PR-1 as function of time (arbitrary units) is shown
by a heat map (Figure 9A). In this simulation, the combination
of auxin and Pst (auxin together with Pst) resulted in decay of
the PR-1 signal generated by Pst alone (Figure 9A, top three
rows). However, together with Pst, cytokinin increased the
signal of PR-1. Full activation of both auxin and cytokinin in the
presence of Pst as input nodes diminished the signal of PR-1
activation entirely. However, in the presence of Pst, the signal
was regained when auxin was kept partially activated and
cytokinin was used as a fully active input node (Figure 9A,
bottom and middle rows). Experimentally, exogenous application of kinetin led to significantly reduced in planta bacterial
growth, whereas NAA enhanced the sensitivity of the plant to
infection with Pst. When administered exogenously, the combination of both auxin and cytokinin led to intermediate bacterial growth in comparison to auxin and cytokinin alone
(Figure 9B). These results suggest that the balance between
auxin and cytokinin modulates in planta Pst multiplication in
Arabidopsis.
We tested how the interaction between auxin and cytokinin
influences the SA pathway of resistance against infection with
Pst. Simulations were conducted in the presence (+SA; enhanced levels) and absence (2SA; mimic SA biosynthesis
mutants) of SA together with different auxin and cytokinin
combinations, as well as Pst as activating input nodes. We
analyzed further a modified network in which the node of SA
was completely deleted (to mimic a sid2 mutant). In the presence of Pst, the activation of auxin as input node abolished the
signal of PR-1 (Figure 9C). However, the activity of PR-1 was
regained when auxin was replaced with cytokinin. Experimentally, we found higher bacterial growth in sid2 mutant plants
when treated with NAA compared with nontreated and kinetintreated plants (Figure 9D). Intriguingly, application of auxin and
cytokinin in combination on the sid2 mutant resulted in intermediate bacterial growth in comparison to the individual application of auxin or cytokinin (Figure 9D). These results
substantiate that fine-tuning between auxin and cytokinin has
an impact on the SA pathway of resistance against infection
with Pst.
Figure 7. (continued).
treatment significantly reduces cytokinin level compared with mock treatment (*P < 0.05, ***P < 0.001; t test). The experiment was repeated three times
and similar results were observed. fgw, fresh gram weight.
(D) Modeling the impact of auxin and SA on infection by Pst. In SQUAD simulations, the node of Pst was kept as a fully active (1.0; virulent pathogen)
input node, while input activations of auxin and SA were compared for full (1.0) and partial (0.5) levels of activation. Activation of PR-1 over arbitrary units
of time (left, top to bottom) is shown as a heat map (scale: right) at various activation levels of auxin and SA in combination with Pst. Further details on
activation of other nodes are given in Supplemental Data Set 9 online.
(E) Auxin promotes sensitivity in Arabidopsis to infection by Pst. Arabidopsis Pst (105 colony-forming units/mL)-inoculated leaves, pretreated with 10
mM NAA. Symptoms were photographed 4 DPPI. Similar results were obtained in three independent experiments.
Modeling Plant Immunity
15 of 22
Figure 8. The Interaction between Auxin and Cytokinin Modulates Plant Immunity.
(A) SQUAD analysis on the balance between auxin and cytokinin (CK) regarding plant immunity. Without activating the node of Pst, only auxin and
cytokinin were used as input nodes in two combinations. Very top, case 1: Auxin was partially activated, while cytokinin was fully activated (auxin, 0.5;
cytokinin, 1). Directly below, case 2: Auxin was fully activated while cytokinin was partially activated (auxin, 1; cytokinin, 0.5) How these ratios modulate
plant system states is shown on the left y axis regarding key network nodes, while immunity in terms of the activity of marker node PR-1 is shown on the
right y axis. Activation states for other nodes in the network are shown as trajectories (see color codes below the x axis; detailed data for a given auxin to
cytokinin ratio are in Supplemental Data Set 3 online). Bottom left panel: Auxin/cytokinin balance and marker node PR-1 activation. SQUAD simulation
for various activation levels of auxin versus cytokinin, including the corresponding activation values of PR-1. y axis, left: Activation of cytokinin as input
signal. x axis, right: Activation value of auxin as input signal. The z axis depicts the activation of PR-1 at a given value of auxin and cytokinin.
(B) Auxin/cytokinin balance in PR1:GUS Arabidopsis transgenic plants. Four-week-old leaves of PR1:GUS Arabidopsis transgenic plants were fed via
the petiole with kinetin, two combinations of kinetin and NAA, NAA alone, and water (mock) for 12 h. Leaves were then subjected to GUS staining. For
visibility of GUS staining, leaves were treated with ethanol as per standard protocol. The experiment was repeated twice.
16 of 22
The Plant Cell
Figure 9. Antagonistic Interaction between Auxin and Cytokinin and Its Impact on Infection by Pst in Arabidopsis.
(A) Modeling of infection by Pst in various combinations of auxin and cytokinin. SQUAD simulations were performed such that the input node of Pst was
kept fully activated (1.0 value of activation) with no other input node, with auxin as a fully activated input node, with cytokinin, with combination of auxin
(partially activated) and cytokinin (fully activated), with combination of auxin (fully activated) and cytokinin (partially activated), and with a fully activated
combination of auxin and cytokinin. For these combinations of inputs, activation states of marker node PR-1 over arbitrary units of time (from left to
right) are shown as a heat map. How these combinations change system states of plant immunity across the whole network is shown in Supplemental
Data Set 3 online.
Modeling Plant Immunity
DISCUSSION
Networking of small-molecule hormones is a newly emerging
aspect of plant immunity where crosstalk between growth regulators and stress-specific hormones modulate plant pathogen
interactions. Pathogen infection profoundly alters the hormonal
profile and network dynamics of the plant cell (Grant and Jones,
2009). Therefore, dynamic modeling is required for better understanding of hormonal implications in plant immunity. However, kinetic data on complex hormonal interactions are scarce.
A solution is standardized qualitative dynamic modeling in which
connectivity (Sankar et al., 2011) is emphasized more than
kinetics. Our modeling illustrates the impact of integration
between bacterial effectors and host hormonal cues and
sampling over the whole network and documents the eventual
outcome for the host-pathogen system. Cytokinin is among
the most important, but immunologically least deciphered, plant
hormones. With modeling and targeted experiments, we have
now identified previously undiscovered crosstalk in cytokinin
signaling.
Pst infection modeling and subsequent systems stability
analysis revealed profound overlap between PTI and ETI.
However, redundancy and robustness was exclusively found in
ETI (Figure 2B; see Supplemental Figures 5 and 7 online).
Robustness in ETI restricts incompatibilities but allows compatible pathogens (Tsuda et al., 2009; Nishimura and Dangl,
2010) to establish susceptibility of the host. Moreover, Type
three secretion system–deficient mutants failed to mount an
immune response equivalent to that of virulence and effectordeficient mutants (see Supplemental Figure 5 online) and were
therefore subject to suboptimal multiplication (Hauck et al.,
2003). Such mutants fail to breach the first line of defense (PTI;
Figure 2B; see Supplemental Figure 4B online), which is otherwise circumvented by effectors of the virulence system (see
Supplemental Figure 4B online) in favor of optimal multiplication. Our modeling envisions zigzag immune responses (Jones
and Dangl, 2006) in dynamic terms and provides relative
quantification to various components of plant immunity.
17 of 22
Regarding the first line of defense, SQUAD modeling highlighted the promoting effect of cytokinin on PTI (Figure 6A).
However, the early events of recognition (ROS production and
media alkalinization; Zipfel et al., 2006) seem to be cytokinin
independent. It remains to be elucidated if cytokinin influences
PTI by modulating nitric oxide signaling. However, during the
advanced phase of the infection, repression of the cytokinin
response/concentration (Figure 7) at the host-pathogen interface seems to be a virulence strategy Pst has adopted for
optimal in planta bacterial multiplication. Pst, unlike other biotrophic pathogens (Walters and McRoberts, 2006; Pertry et al.,
2009), does not produce cytokinin during infection. Cytokinin
response reporter lines (AtARR5:GUS) demonstrate the repression of host cytokinin during pathogen infection in detail
(Figure 7B). Additionally, the sequenced genome of Pst (http://
www.pseudomonas-syringae.org/pst_DC3000_gen.htm) contains
no genes related to biosynthesis or degradation of cytokinin
otherwise present in Arabidopsis, Agrobacterium tumefaciens,
Pseudomonas savastanoi, and Ralstonia fascians, as well as
other plant growth-promoting bacteria. The lack of increase in
cytokinin concentration/response (measurements in Figures 7B
and 7C) during the course of infection suggests the absence of
potential effectors in Pst that would otherwise enhance the level/
response of cytokinin in the host.
Modeling the impact of phytohormones on Pst infection revealed that JA, auxin, and ABA reduced Pst-mediated PR-1
activation and promoted susceptibility. SA and GA further increased PR-1 activation and enhanced resistance (Figure 3A).
Predicted node switching and model outputs are supported in
detail according to current literature (see Supplemental Table 2
online). Modeling of increased plant cytokinin with/without the
addition of Pst revealed a boosting effect on plant immunity
(Figures 3B and 9A; see Supplemental Figure 3B online). Experiments support this by revealing reduced disease symptoms
with abrogated in planta bacterial multiplication (Figure 3C) and
higher expression of PR-1 (Figure 8B) upon cytokinin treatment.
Moreover, increased cytokinin levels enhanced resistance in
transgenic plants to Pst (Choi et al., 2010) and Pst tabaci
Figure 9. (continued).
(B) Bacterial growth: This was quantified using (color code given on top) mock, 10 mM kinetin, 10 mM NAA, or 10 mM kinetin and 10 mM NAA in
combination. Bacterial reisolation was performed 2 DPPI. Statistically significant differences between treatments (a to c) and SE were calculated by
analysis of variance. Error bars represent SE and letters on the top represent significant differences (P < 0.05, n = 5) in response to pathogen (however,
cytokinin brings IAA-mediated susceptibility back to the wild-type situation, no significant difference, hence, again letter a). The experiment was
repeated three times, and similar results were observed. CFU, colony-forming units.
(C) Relative impact of auxin and cytokinin on the SA pathway during infection by Pst. To mimic the behavior of SA-deficient mutants, the node of SA
was removed from the network and simulations were performed such that Pst as the input node was kept fully activated (activation value 1) with full
activation of auxin and cytokinin being input nodes, only with fully activated auxin, with fully activated cytokinin alone, with fully activated auxin and SA
(wild-type situation; i.e., with deletion of SA from the network), and with cytokinin and SA (wild-type situation). Activity of PR-1 is shown over arbitrary
units of time (from left to right) for these combinations of inputs. For changes in system states across the whole network for these combinations, see
Supplemental Data Set 10 online.
(D) Quantification of bacterial growth. This was quantified (color code given on top) for Col-0, for sid2 alone, or with 10 mM cytokinin, 10 mM NAA, and
10 mM kinetin + 10 mM NAA in combination. Col-0 Arabidopsis leaves and sid2 mutants were infiltrated with a suspension of Pst (0.5 3 103 colonyforming units [CFU]/mL). Statistical differences (a to d) between treatments and SE were calculated by analysis of variance (P < 0.05, n = 5). Error bars
represent SE, and letters on the top represent significant differences in response to pathogen. The experiment was repeated three times, and similar
trends were observed.
18 of 22
The Plant Cell
(Grosskinsky et al., 2011), whereas decreased cytokinin level
reduced the expression of PR-1 (Uchida and Tasaka, 2010).
We took up the challenge to create a dynamic model of these
various influences. Thus, modulation in plant cytokinin
changes infection dynamics such that an increase in cytokinin
favors protection, while its decline adds to vulnerability.
Argueso et al. (2012) found that comparatively low exogenous
cytokinin application lead to higher susceptibility against infection with Hpa Noco2 in Arabidopsis. However, (1) unlike
Hpa, Pst reduces the level of in planta cytokinin at the site of
infection (Figure 7C); (2) Pst weakens cytokinin responses
(Figures 7A and 7B) to avoid cytokinin-mediated defense (Choi
et al., 2010) for self-multiplication; and (3) in contrast with infection with Hpa Noco2, where cytokinin operates upstream
(EDS16) of SA production, cytokinin-mediated protection
against Pst operates downstream (TGA3) of the SA synthesis
(Figure 5D); (4) finally, Hpa Noco2 uses type A response regulators to suppress SA-mediated defense responses, whereas
Pst infection downregulates A type ARRs (Figure 7A; Thilmony
et al., 2006), and in the presence of cytokinin, B-type ARR
(ARR2) is implicated in immunity (Choi et al., 2010). Trophic
nature, in planta cytokinin secreting capabilities, host cell inoculation, and multiplication are features that further differentiate Hpa Noco2 from Pst. We thus argue that infection
dynamics of Hpa Noco2 and Pst regarding cytokinin responses
are out of phase and, hence, not directly comparable.
Cytokinin-mediated increases in the expression of marker
gene PR-1 (Figure 8B) made SA-cytokinin positive crosstalk
quite plausible. However, our simulations on the mode of interaction between SA and cytokinin (Figures 4 and 9C) revealed
important biological inferences. SQUAD modeling and subsequent experimental analysis on cytokinin-treated Arabidopsis
leaves demonstrated no significant increase for the accumulation of SA (Figures 3B and 5A). By contrast, without pathogen
inoculation, cytokinin treatment led to significantly higher camalexin accumulation in treated versus control leaves (Figure
5B). These results prompted experiments on SA-deficient mutants for the underlying SA-cytokinin crosstalk. However, the
literature is mired with contrasting conclusions in this regard.
Cytokinin-treated, Pst-inoculated NahG Arabidopsis plants (Choi
et al., 2010) failed to show rescued susceptibility. By contrast,
Pst tabaci–inoculated prior cytokinin-treated NahG tobacco
(Grosskinsky et al., 2011) plants showed a completely rescued
susceptible disease phenotype. We analyzed this dichotomy
with SQUAD simulations and found that in Arabidopsis, interaction between SA and cytokinin positively influenced plant
immunity against Pst (Figure 5C). However, deletion of SA in the
presence of cytokinin slightly affected the activity of PR-1,
whereas deletion of TGA substantially reduced this activity
(Figure 5C). In corresponding experiments, we partially restored
resistance against Pst by the application of exogenous cytokinin
to sid2 mutants (Figures 5D and 9D). This seems to be in contrast with a previous finding where prior cytokinin treatment
could not prevent the severe disease phenotype in NahG plants
(Choi et al., 2010). This may be due to the nature of the mutant
we investigated. In addition, according to our results, the system
response differs according to pathogen load. At low Pst density, cytokinin treatment does prevent the severe phenotype.
Enhanced pathogen resistance by elevated cytokinin is thus
partly a consequence of direct crosstalk between SA and cytokinin. However, SA biosynthesis adds up to better protection
by enhancing system states already in place (Figure 5C), and
cytokinin-driven phytolaxin (Figure 5B) accumulation and
weakening of auxin responses (Figures 8 and 9C) are additional factors contributing to immunity in Arabidopsis against
Pst.
The Pst infection model determined activation states for
nodes representing hormonal stimuli. In contrast with the lack
of activation found for cytokinin, auxin was activated over time
(Figure 2A; see Supplemental Figure 4A online). Analysis of
large-scale gene expression data on Pst infection in Arabidopsis also mirrored the notion of repression and derepression
for cytokinin and auxin responses, respectively (Figure 7A).
Moreover, Pst infection experiments on response regulators of
auxin and cytokinin reporter lines (Figure 7B) nicely complemented the infection modeling. Significantly higher auxin
and reduced cytokinin accumulation at the host-pathogen interface due to the infection with Pst (Figure 7C) further supported the outcome of experiments on reporter lines. These
different and complementary approaches adequately highlighted the activation of auxin and inhibition of cytokinin as
a consequence of infection with Pst in Arabidopsis. However,
SQUAD simulations and subsequent experimental analysis
support that both auxin and cytokinin change the infection
dynamics of Pst (Figures 5C, 5D, 7D, and 7E). Cytokinin increased protection, whereas auxin enhanced susceptibility of
Arabidopsis to infection by Pst. Our findings are in agreement
with previous findings. Increased endogenous cytokinin (35S:
IPT; Choi et al., 2010) restricted Pst multiplication, whereas
enhanced auxin levels (Wang et al., 2007; Crabill et al., 2010) in
Arabidopsis led to enhanced sensitivity to infection with Pst.
However, rather than investigating the role of auxin (Wang
et al., 2007) and cytokinin (Choi et al., 2010) as independent
events in plant immunity, we captured them in concert and
substantiated that activation of the former with concomitant
inhibition of the latter combine to create optimal infection
conditions for Pst in Arabidopsis.
Crosstalk between auxin and cytokinin is quite prevalent in
biological processes related to growth and development of
plants (Moubayidin et al., 2009; Liu et al., 2010), but the literature
is scarce regarding its impact on plant immunity. SQUAD
modeling and subsequent experimental demonstration elucidated the interplay between auxin and cytokinin and revealed
that immunity and susceptibility are a carefully controlled balance (Figure 8A). States of activation of cytokinin over auxin
divert the trend toward resistance, and the opposite holds true
for susceptibility (Figure 8A). Experimentally, in comparison to
auxin, addition of cytokinin alone resulted in higher PR-1 expression, while their combination minimized this effect (Figure
8B). Moreover, higher auxin levels led to severe disease symptoms coupled with fast bacterial growth (Figures 7E and 9B).
Auxin and cytokinin fed to Arabidopsis leaves in combination
and in concentrations as above resulted in an intermediate state
of response (Figure 9B). Intermediate bacterial growth resulted
from the combined effect of cytokinin and auxin in sid2 mutants
(Figure 9D). Mutual regulation between auxin and cytokinin from
Modeling Plant Immunity
the perspective of host-pathogen interaction has not been explored so far. The antagonistic effect of auxin on cytokinin levels
has already been investigated for morphogenesis (Nordström
et al., 2004). However, via type B response regulators, cytokinin
decreases auxin responses by inducing auxin repressor protein
(AUX/IAA; Moubayidin et al., 2009). Cytokinin also affects auxin
flux and gradients by decreasing the level of auxin efflux carriers
(PIN1) through a transcription-independent mechanism (Stepanova
and Alonso, 2011). Taken together for the host-pathogen system,
our modeling and experimental analysis substantiates an antagonistic interaction between auxin and cytokinin. Stronger resistance
is impaired by auxin and increased sensitivity is diminished by
cytokinin. Likely mechanisms to promote resistance involve (1)
cytokinin stabilizing the auxin repressors AUX/IAA through
positive interaction with SA and (2) cytokinin influencing transport of auxin away from the site of infection via auxin transporters. Likewise, auxins might promote (1) degradation of
cytokinin through oxidation by their activation enzymes or (2)
conversion of cytokinin to inactive storage forms. Molecular
mechanisms that influence these interactions need detailed
further experimentation and more resolution.
Tiny and simply structured yet complex in cell signaling,
cytokinin is actively involved in vital processes, such as cell
division, tissue differentiation, lateral bud growth, photosynthetic activity, and floral development (Kudo et al., 2010). They
thus have immense potential in enhancement of crop yield. By
virtue of their implications in delaying aging and retaining plant
chlorophyll (Gan and Amasino, 1995) cytokinin is instrumental
in increasing crop shelf life, reduction of post-harvest losses,
and thus ensuring food security. Elevated cytokinin promotes
antioxidative forces, keeps the cellular redox potential high,
and protects plants from abiotic environmental extremes
(Rivero et al., 2007). Adding further to this list, our systems
biology–led experiments and modeling suggests crosstalk
possibilities for cytokinin in complex hormonal disease networks with implications for new yield protection strategies in
agriculture.
METHODS
General Network Setup
Data mining and an extensive literature survey established the
plant hormone disease network with all undisputed key nodes (see
Supplemental Table 1 online) and inhibitory and/or activatory edges (type
of interactions, references, and evaluation; see Supplemental Tables 1
and 2 online). The network was implemented in CellDesigner version 3.5.1
and stored in SBML format.
Modeling Strategy
The SQUAD suite (Di Cara et al., 2007) performed steady state analyses
and dynamic simulations (see Supplemental Figure 1 online), recognizing
nodes and interacting edges automatically. For the discrete dynamic
analysis, SQUAD uses generic assumptions: The presence of any one of
the activators can effectively activate a target node, and the presence of
any one of the inhibitors can successfully inactivate the target node. For
more complex scenarios, different inputs are each considered and integrated according to the network topology (see details in Supplemental
19 of 22
Methods 1 online; in particular, equation 2). Thereby, SQUAD identifies
the number of stable steady states located in the network. Furthermore, it
calculates activity values of each node at a given steady state. SQUAD
simulations and steady state analyses highlight system equilibria and their
changes upon signaling stimuli (pathogen and/or hormonal signals). The
activity value of each node was set to zero (trivial steady state) as
a starting point for continuous dynamic simulations to bring the network at
equilibrium. To monitor the global system response semiquantitatively
toward pathogen or hormonal stimuli, an input signal of the system was
initially set to 1. To study the mutual impact of two or more stimuli, we
simultaneously set their initial activation to 1. For partial activation, input
stimuli were initialized with values between 0 and 1. SQUAD converted
the effect of these perturbations into time-resolved activity curves
(sigmoid) for each and every node of the network which then was visualized by means of heat maps and trajectory plots. Supplemental
Methods 1 online shows the exact equations and resulting curve shape
used for this and integration of activating as well as inhibitory input (or
both) according to logical connectivity including partial activation of
nodes. Supplemental Data Set 4 online shows for illustration the detailed
modeling of auxin and cytokinin during the infection of Pst DC3000 in
Arabidopsis thaliana (see also results above). For validation, this can be
compared with Supplemental Data Set 5 online, for measured gene
expression values of auxin and cytokinin pathways during infection of
Arabidopsis by Pst DC3000 versus mock (see also results above).
To asses the robustness of the network, we randomly added and
removed nodes from the network and performed simulations as detailed
above (see Supplemental Figure 7 online).
Analysis of Microarray Data
We analyzed original submitter-supplied microarray data in the GEO
database using the interactive Web tool GEO2R (http://www.ncbi.nlm.nih.
gov/geo/geo2r/). This allows users to compare two or more samples
within the GEO submitted microarray series. GEO2R uses Bioconductor
R packages GEOquery (Sean and Meltzer, 2007) and limma (Linear
Models for Microarray Analysis; Smyth, 2004). GEOquery parses GEO
submitted array data into R data structures, while limma applies multiple
testing corrections on P values to correct the occurrence of false
positives and identifies differentially (P value # 0.05) expressed genes.
We entered GEO accession numbers of the above microarray experiments and then defined groups as treated and control with given
number of replicates found in GEO. We calculated the distribution of
the values for selected samples via value distribution option in GEO2R
and graphically presented them as box plots to view their suitability
for comparison. For multiple-testing corrections, the Benjamini and
Hochberg false discovery rate method (Benjamini and Hochberg, 1995)
was applied. GEO2R provided limma-generated statistical analysis of the
data, such as adjusted and raw P values, t and B values, as well as fold
changes.
Plant Pathogen Experiments
Four- to five-week-old Arabidopsis thaliana Col-0 (see http://www.
Arabidopsis.org/) and sid2 mutant plants were used in experiments
(Raacke et al., 2006). Plants were grown in soil with 9 to 16 h of light and
periods of 5 weeks in a growth chamber. The bacterial strain Pseudomonas syringae pv tomato DC3000 was cultured in King’s broth medium
containing 50 mg/L rifampicin. Harvested cells were resuspended in 10
mM MgCl2. Bacterial inoculations were performed by syringe infiltration.
Furthermore, bacterial growth was quantified through a plate counting
method. Arabidopsis Col-0 and sid2 mutants were sprayed with kinetin
and/or NAA solutions. For pathogen inoculation, the spray method was
also used. PR-1:GUS plants (Gust et al., 2007) were also investigated.
GUS staining and subsequent destaining were performed according to
20 of 22
The Plant Cell
standard procedures. Free SA and phytoalexin were analyzed according
to Mishina and Zeier (2006), nad t-zeatin and IAA were analyzed as described by Hussain et al. (2010). ROS and pH experiments were performed as described by Ahmed (2010).
ACKNOWLEDGMENTS
We thank the German Research Council (Deutsche Forschungsgemeinshaft: Da 208/10-2; 12-1), Bundesministerium für Bildung und Forschung
(0315395B), and Land Bavaria for funding.
Accession Numbers
AUTHOR CONTRIBUTIONS
Data analyzed in this study are available in the GEO database under
accession numbers GSE3984, GSE6832, and GSE5520.
Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure 1. Key Steps in the Implementation of SQUAD
Methodology for Investigating Plant Hormones and Disease Signaling
Networks.
M.N. and T.D. conceived and planned the project. M.N. and N.P. set up
simulations and M.N., N.P., G.W., and T.D. analyzed them. M.N. (main),
A.H., and N.A. performed experiments for verification. T.D. supervised
and guided the study. All authors wrote the article and approved the final
version.
Received March 15, 2012; revised April 26, 2012; accepted May 10,
2012; published May 29, 2012.
Supplemental Figure 2. Infection Modeling of Virulent Pathogen Pst
and Fitness of the Pathogen.
Supplemental Figure 3. Modeling the Impact of Plant Hormone
Cytokinin on the Pathogenicity of Pst.
Supplemental Figure 4. Effectors of Pst DC3000 Change the
Hormonal Profile in Arabidopsis and Inhibit PTI.
Supplemental Figure 5. Robustness and Redundancy in ETI and Its
Operational Overlap with PTI.
Supplemental Figure 6. Cytokinin Has No Cytotoxic or Antimicrobial
Effects on the Multiplication of Pst DC3000.
Supplemental Figure 7. System Stability Analysis and Resulting
Robustness of the Network Topology.
Supplemental Figure 8. Network Topology for Hormones and
Disease Networks in Arabidopsis.
Supplemental Table 1. Node Interactions and Literature Support.
Supplemental Table 2. Input Signals and SQUAD-Analyzed Outputs
with Experimental Validation.
Supplemental Methods 1. Computation of the Dynamic Behavior of
the Network Components.
Supplemental Data Set 1. Cytokinin Promotes PTI.
Supplemental Data Set 2. SQUAD Modeling for Infection of Fully
Virulent Pst DC3000 in Arabidopsis.
Supplemental Data Set 3. Auxin-Cytokinin Antagonism in Plant
Immunity.
Supplemental Data Set 4. Modeling the Combination of Auxin and
Cytokinin during the Infection of Pst DC3000 in Arabidopsis.
Supplemental Data Set 5. Measured Gene Expression Values of
Auxin and Cytokinin Pathways during Infection of Arabidopsis by Pst
DC3000 versus Mock.
Supplemental Data Set 6. How Plant hormones Influence the
Infection of Pst DC3000 in Arabidopsis.
Supplemental Data Set 7. Impact of SA and Cytokinin Signaling on
Genes of Immunity in Arabidopsis.
Supplemental Data Set 8. Modeling of the Impact of Cytokinin on SA
Pathway of Resistance.
Supplemental Data Set 9. Modeling the Impact of SA and Auxin on
Infection by Pst DC3000 in Arabidopsis.
Supplemental Data Set 10. Relative Strength of Auxin and Cytokinin
Effects on the SA Pathway.
REFERENCES
Ahmed, N. (2010). Physiological and Molecular Basis of AzospirillumArabidopsis Interaction. PhD dissertation (Würzburg, Germany:
University of Wuerzburg).
Albert, R. (2005). Scale-free networks in cell biology. J. Cell Sci. 118:
4947–4957.
Alonso-Ramírez, A., Rodríguez, D., Reyes, D., Jiménez, J.A.,
Nicolás, G., López-Climent, M., Gómez-Cadenas, A., and
Nicolás, C. (2009). Evidence for a role of gibberellins in salicylic
acid-modulated early plant responses to abiotic stress in Arabidopsis seeds. Plant Physiol. 150: 1335–1344.
Argueso, C.T., Ferreira, F.J., Epple, P., To, J.P., Hutchison, C.E.,
Schaller, G.E., Dangl, J.L., and Kieber, J.J. (2012). Two-component
elements mediate interactions between cytokinin and salicylic acid in
plant immunity. PLoS Genet. 8: e1002448.
Ay, F., Xu, F., and Kahveci, T. (2009). Scalable steady state analysis
of Boolean biological regulatory networks. PLoS ONE 4: e7992.
Bari, R., and Jones, J.D. (2009). Role of plant hormones in plant
defence responses. Plant Mol. Biol. 69: 473–488.
Benjamini, Y., and Hochberg, Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing.
J. R. Stat. Soc. Ser. 57: 289–300.
Boller, T., and He, S.Y. (2009). Innate immunity in plants: An arms
race between pattern recognition receptors in plants and effectors
in microbial pathogens. Science 324: 742–744.
Cao, F.Y., Yoshioka, K., and Desveaux, D. (2011). The roles of ABA
in plant-pathogen interactions. J. Plant Res. 124: 489–499.
Clarke, S.F., Burritt, D.J., Jameson, P.E., and Guy, P.L. (1998). Influence of plant hormones on virus replication and pathogenesis
related proteins in Phaseolus vulgaris L. infected with white clover
mosaic potexvirus. Physiol. Mol. Plant Pathol. 53: 195–207.
Crabill, E., Joe, A., Block, A., van Rooyen, J.M., and Alfano, J.R.
(2010). Plant immunity directly or indirectly restricts the injection of
type III effectors by the Pseudomonas syringae type III secretion
system. Plant Physiol. 154: 233–244.
Chen, Z., Agnew, J.L., Cohen, J.D., He, P., Shan, L., Sheen, J., and
Kunkel, B.N. (2007). Pseudomonas syringae type III effector
AvrRpt2 alters Arabidopsis thaliana auxin physiology. Proc. Natl.
Acad. Sci. USA 104: 20131–20136.
Choi, J., Huh, S.U., Kojima, M., Sakakibara, H., Paek, K.H., and
Hwang, I. (2010). The cytokinin-activated transcription factor ARR2
promotes plant immunity via TGA3/NPR1-dependent salicylic acid
signaling in Arabidopsis. Dev. Cell 19: 284–295.
Modeling Plant Immunity
Dervinis, C., Frost, C.J., Lawrence, S.D., Novak, N.G., and Davis,
J.M. (2010). Cytokinin primes plant responses to wounding and
reduces insect performance. J. Plant Growth Regul. 29: 289–296.
de Torres Zabala, M., Bennett, M.H., Truman, W.H., and Grant,
M.R. (2009). Antagonism between salicylic and abscisic acid reflects early host-pathogen conflict and moulds plant defence responses. Plant J. 59: 375–386.
Di Cara, A.D., Garg, A., De Micheli, G., Xenarios, I., and Mendoza,
L. (2007). Dynamic simulation of regulatory networks using SQUAD.
BMC Bioinformatics 8: 462.
Fayza, A.F., and Sabrey, Y.M.M. (2006). Induction of resistance in
Phaseolus vulgaris against TNV by salicylic acid and kinetin. Int. J.
Agric. Biol 8: 47–51.
Funahashi, A., Matsuoka, Y., Jouraku, A., Morohashi, M., Kikuchi,
N., and Kitano, H. (2008). CellDesigner 3.5: A versatile modelling
tool for biochemical networks. Proc. IEEE 96: 1254–1265.
Gan, S., and Amasino, R.M. (1995). Inhibition of leaf senescence by
autoregulated production of cytokinin. Science 270: 1986–1988.
Glazebrook, J. (2005). Contrasting mechanisms of defense against
biotrophic and necrotrophic pathogens. Annu. Rev. Phytopathol.
43: 205–227.
Grant, M.R., and Jones, J.D.G. (2009). Hormone (dis)harmony
moulds plant health and disease. Science 324: 750–752.
Grosskinsky, D.K. et al. (2011). Cytokinins mediate resistance
against Pseudomonas syringae in tobacco through increased antimicrobial phytoalexin synthesis independent of salicylic acid signaling. Plant Physiol. 157: 815–830.
Gust, A.A., Biswas, R., Lenz, H.D., Rauhut, T., Ranf, S.,
Kemmerling, B., Götz, F., Glawischnig, E., Lee, J., Felix, G., and
Nürnberger, T. (2007). Bacteria-derived peptidoglycans constitute
pathogen-associated molecular patterns triggering innate immunity
in Arabidopsis. J. Biol. Chem. 282: 32338–32348.
Han, J.D., Bertin, N., Hao, T., Goldberg, D.S., Berriz, G.F., Zhang,
L.V., Dupuy, D., Walhout, A.J., Cusick, M.E., Roth, F.P., and
Vidal, M. (2004). Evidence for dynamically organized modularity in
the yeast protein-protein interaction network. Nature 430: 88–93.
Hauck, P., Thilmony, R., and He, S.Y. (2003). A Pseudomonas syringae type III effector suppresses cell wall-based extracellular defense in susceptible Arabidopsis plants. Proc. Natl. Acad. Sci. USA
100: 8577–8582.
Hussain, A., Krischke, M., Roitsch, T., and Hasnain, S. (2010).
Rapid determination of cytokinins and auxin in cyanobacteria. Curr.
Microbiol. 61: 361–369.
Jones, J.D.G., and Dangl, J.L. (2006). The plant immune system.
Nature 444: 323–329.
Kazan, K., and Manners, J.M. (2009). Linking development to defense: Auxin in plant-pathogen interactions. Trends Plant Sci. 14:
373–382.
Kudo, T., Kiba, T., and Sakakibara, H. (2010). Metabolism and
long-distance translocation of cytokinins. J. Integr. Plant Biol. 52:
53–60.
Kunze, G., Zipfel, C., Robatzek, S., Niehaus, K., Boller, T., and
Felix, G. (2004). The N terminus of bacterial elongation factor Tu
elicits innate immunity in Arabidopsis plants. Plant Cell 16: 3496–
3507.
Lai, Z., Wang, F., Zheng, Z., Fan, B., and Chen, Z. (2011). A critical
role of autophagy in plant resistance to necrotrophic fungal
pathogens. Plant J. 66: 953–968.
Laurie-Berry, N., Joardar, V., Street, I.H., and Kunkel, B.N. (2006).
The Arabidopsis thaliana JASMONATE INSENSITIVE 1 gene is required for suppression of salicylic acid-dependent defenses during
infection by Pseudomonas syringae. Mol. Plant Microbe Interact.
19: 789–800.
21 of 22
Li, D., Li, J., Ouyang, S., Wang, J., Wu, S., Wan, P., Zhu, Y., Xu, X.,
and He, F. (2006). Protein interaction networks of Saccharomyces
cerevisiae, Caenorhabditis elegans and Drosophila melanogaster:
Large-scale organization and robustness. Proteomics 6: 456–461.
Liu, J., Mehdi, S., Topping, J., Tarkowski, P., and Lindsey, K.
(2010). Modelling and experimental analysis of hormonal crosstalk
in Arabidopsis. Mol. Syst. Biol. 6: 373.
Llorente, F., Muskett, P., Sánchez-Vallet, A., López, G., Ramos, B.,
Sánchez-Rodríguez, C., Jordá, L., Parker, J., and Molina, A.
(2008). Repression of the auxin response pathway increases Arabidopsis susceptibility to necrotrophic fungi. Mol. Plant 1: 496–509.
Melotto, M., Underwood, W., and He, S.Y. (2008). Role of stomata
in plant innate immunity and foliar bacterial diseases. Annu. Rev.
Phytopathol. 46: 101–122.
Melotto, M., Underwood, W., Koczan, J., Nomura, K., and He, S.Y.
(2006). Plant stomata function in innate immunity against bacterial
invasion. Cell 126: 969–980.
Mishina, T.E., and Zeier, J. (2006). The Arabidopsis flavin-dependent
monooxygenase FMO1 is an essential component of biologically
induced systemic acquired resistance. Plant Physiol. 141: 1666–
1675.
Mishina, T.E., and Zeier, J. (2007). Pathogen-associated molecular
pattern recognition rather than development of tissue necrosis
contributes to bacterial induction of systemic acquired resistance in
Arabidopsis. Plant J. 50: 500–513.
Moubayidin, L., Di Mambro, R., and Sabatini, S. (2009). Cytokininauxin crosstalk. Trends Plant Sci. 14: 557–562.
Mukhtar, M.S. et al, European Union Effectoromics Consortium
(2011). Independently evolved virulence effectors converge onto
hubs in a plant immune system network. Science 333: 596–601.
Muller, B., and Sheen, J. (2007). Arabidopsis cytokinins signalling
pathway. Sci. STKE 407: cm5.
Navarro, L., Bari, R., Achard, P., Lisón, P., Nemri, A., Harberd, N.P.,
and Jones, J.D.G. (2008). DELLAs control plant immune responses
by modulating the balance of jasmonic acid and salicylic acid signaling. Curr. Biol. 18: 650–655.
Navarro, L., Dunoyer, P., Jay, F., Arnold, B., Dharmasiri, N., Estelle,
M., Voinnet, O., and Jones, J.D.J. (2006). A plant miRNA contributes to antibacterial resistance by repressing auxin signaling.
Science 312: 436–439.
Nawrath, C., and Métraux, J.P. (1999). Salicylic acid inductiondeficient mutants of Arabidopsis express PR-2 and PR-5 and accumulate high levels of camalexin after pathogen inoculation. Plant
Cell 11: 1393–1404.
Nishimura, M.T., and Dangl, J.L. (2010). Arabidopsis and the plant
immune system. Plant J. 61: 1053–1066.
Nordström, A., Tarkowski, P., Tarkowska, D., Norbaek, R., Astot,
C., Dolezal, K., and Sandberg, G. (2004). Auxin regulation of cytokinin biosynthesis in Arabidopsis thaliana: A factor of potential
importance for auxin-cytokinin-regulated development. Proc. Natl.
Acad. Sci. USA 101: 8039–8044.
Panstruga, R., Parker, J.E., and Schulze-Lefert, P. (2009). SnapShot: Plant immune response pathways. Cell 136: 978.e1–978.e3.
Pertry, I. et al. (2009). Identification of Rhodococcus fascians cytokinins and their modus operandi to reshape the plant. Proc. Natl.
Acad. Sci. USA 106: 929–934.
Philippi, N., Walter, D., Schlatter, R., Ferreira, K., Ederer, M.,
Sawodny, O., Timmer, J., Borner, C., and Dandekar, T. (2009).
Modeling system states in liver cells: Survival, apoptosis and their
modifications in response to viral infection. BMC Syst. Biol. 3: 97.
Pieterse, C.M.J., Leon-Reyes, A., Van der Ent, S., and Van Wees,
S.C. (2009). Networking by small-molecule hormones in plant immunity. Nat. Chem. Biol. 5: 308–316.
22 of 22
The Plant Cell
Pomerance, A., Ott, E., Girvan, M., and Losert, W. (2009). The effect
of network topology on the stability of discrete state models of
genetic control. Proc. Natl. Acad. Sci. USA 106: 8209–8214.
Pritchard, L., and Birch, P. (2011). A systems biology perspective on
plant-microbe interactions: Biochemical and structural targets of
pathogen effectors. Plant Sci. 180: 584–603.
Raacke, I.C., von Rad, U., Mueller, M.J., and Berger, S. (2006).
Yeast increases resistance in Arabidopsis against Pseudomonas
syringae and Botrytis cinerea by salicylic acid-dependent as well as
-independent mechanisms. Mol. Plant Microbe Interact. 19: 1138–
1146.
Rico, A., McCraw, S.L., and Preston, G.M. (2011). The metabolic
interface between Pseudomonas syringae and plant cells. Curr.
Opin. Microbiol. 14: 31–38.
Rico, A., and Preston, G.M. (2008). Pseudomonas syringae pv. tomato DC3000 uses constitutive and apoplast-induced nutrient assimilation pathways to catabolize nutrients that are abundant in the
tomato apoplast. Mol. Plant Microbe Interact. 21: 269–282.
Rivero, R.M., Kojima, M., Gepstein, A., Sakakibara, H., Mittler, R.,
Gepstein, S., and Blumwald, E. (2007). Delayed leaf senescence
induces extreme drought tolerance in a flowering plant. Proc. Natl.
Acad. Sci. USA 104: 19631–19636.
Robert-Seilaniantz, A., Grant, M., and Jones, J.D. (2011a). Hormone
crosstalk in plant disease and defense: More than just jasmonatesalicylate antagonism. Annu. Rev. Phytopathol. 49: 317–343.
Robert-Seilaniantz, A., MacLean, D., Jikumaru, Y., Hill, L.,
Yamaguchi, S., Kamiya, Y., and Jones, J.D.G. (2011b). The microRNA miR393 re-directs secondary metabolite biosynthesis away
from camalexin and towards glucosinolates. Plant J. 67: 218–231.
Robert-Seilaniantz, A., Navarro, L., Bari, R., and Jones, J.D.G.
(2007). Pathological hormone imbalances. Curr. Opin. Plant Biol. 10:
372–379.
Sankar, M., Osmont, K.S., Rolcik, J., Gujas, B., Tarkowska, D.,
Strnad, M., Xenarios, I., and Hardtke, C.S. (2011). A qualitative
continuous model of cellular auxin and brassinosteroid signaling
and their crosstalk. Bioinformatics 27: 1404–1412.
Sato, M., Tsuda, K., Wang, L., Coller, J., Watanabe, Y., Glazebrook,
J., and Katagiri, F. (2010). Network modeling reveals prevalent
negative regulatory relationships between signaling sectors in
Arabidopsis immune signaling. PLoS Pathog. 6: e1001011.
Schlatter, R., Philippi, N., Wangorsch, G., Pick, R., Sawodny, O.,
Borner, C., Timmer, J., Ederer, M., and Dandekar, T. (2011).
Integration of Boolean models exemplified on hepatocyte signal
transduction. Brief. Bioinform. 13: 365–376.
Schneider, D.J., and Collmer, A. (2010). Studying plant-pathogen
interactions in the genomics era: Beyond molecular Koch’s postulates to systems biology. Annu. Rev. Phytopathol. 48: 457–479.
Schuster, S., Fell, D.A., and Dandekar, T. (2000). A general definition
of metabolic pathways useful for systematic organization and analysis of
complex metabolic networks. Nat. Biotechnol. 18: 326–332.
Sean, D., and Meltzer, P.S. (2007). GEOquery: A bridge between the
Gene Expression Omnibus (GEO) and BioConductor. Bioinformatics
23: 1846–1847.
Siemens, J., Keller, I., Sarx, J., Kunz, S., Schuller, A., Nagel, W.,
Schmülling, T., Parniske, M., and Ludwig-Müller, J. (2006).
Transcriptome analysis of Arabidopsis clubroots indicate a key role
for cytokinins in disease development. Mol. Plant Microbe Interact.
19: 480–494.
Smyth, G.K. (2004). Linear models and empirical Bayes methods for
assessing differential expression in microarray experiments. Stat.
Appl. Genet. Mol. Biol. 3: article3.
Stepanova, A.N., and Alonso, J.M. (2011). Bypassing transcription: A
shortcut in cytokinin-auxin interactions. Dev. Cell 21: 608–610.
Stotz, H.U., Sawada, Y., Shimada, Y., Hirai, M.Y., Sasaki, E.,
Krischke, M., Brown, P.D., Saito, K., and Kamiya, Y. (2011). Role
of camalexin, indole glucosinolates, and side chain modification of
glucosinolate-derived isothiocyanates in defense of Arabidopsis
against Sclerotinia sclerotiorum. Plant J. 67: 81–93.
Thibaud-Nissen, F., Wu, H., Richmond, T., Redman, J.C., Johnson,
C., Green, R., Arias, J., and Town, C.D. (2006). Development of
Arabidopsis whole-genome microarrays and their application to the
discovery of binding sites for the TGA2 transcription factor in salicylic acid-treated plants. Plant J. 47: 152–162.
Thilmony, R., Underwood, W., and He, S.Y. (2006). Genome-wide
transcriptional analysis of the Arabidopsis thaliana interaction
with the plant pathogen Pseudomonas syringae pv. tomato DC3000
and the human pathogen Escherichia coli O157:H7. Plant J. 46:
34–53.
Tsuda, K., Sato, M., Stoddard, T., Glazebrook, J., and Katagiri, F.
(2009). Network properties of robust immunity in plants. PLoS
Genet. 5: e1000772.
Tzfira, T., and Citovsky, V. (2006). Agrobacterium-mediated genetic
transformation of plants: Biology and biotechnology. Curr. Opin.
Biotechnol. 17: 147–154.
Uchida, N., and Tasaka, M. (2010). Intersections between immune
responses and morphological regulation in plants. J. Exp. Bot. 61:
2539–2547.
Verhage, A., van Wees, S.C., and Pieterse, C.M.J. (2010). Plant
immunity: It’s the hormones talking, but what do they say? Plant
Physiol. 154: 536–540.
Walters, D.R., and McRoberts, N. (2006). Plants and biotrophs: A
pivotal role for cytokinins? Trends Plant Sci. 11: 581–586.
Wang, D., Pajerowska-Mukhtar, K., Culler, A.H., and Dong, X.
(2007). Salicylic acid inhibits pathogen growth in plants through
repression of the auxin signaling pathway. Curr. Biol. 17: 1784–
1790.
Wangorsch, G., Butt, E., Mark, R., Hubertus, K., Geiger, J.,
Dandekar, T., and Dittrich, M. (2011). Time-resolved in silico
modeling of fine-tuned cAMP signaling in platelets: Feedback
loops, titrated phosphorylations and pharmacological modulation.
BMC Syst. Biol. 5: 178.
Wildermuth, M.C., Dewdney, J., Wu, G., and Ausubel, F.M. (2001).
Isochorismate synthase is required to synthesize salicylic acid for
plant defence. Nature 414: 562–565.
Wittmann, D.M., Krumsiek, J., Saez-Rodriguez, J., Lauffenburger,
D.A., Klamt, S., and Theis, F.J. (2009). Transforming Boolean
models to continuous models: Methodology and application to
T-cell receptor signaling. BMC Syst. Biol. 3: 98.
Zeng, W., and He, S.Y. (2010). A prominent role of the flagellin receptor FLAGELLIN-SENSING2 in mediating stomatal response to
Pseudomonas syringae pv tomato DC3000 in Arabidopsis. Plant
Physiol. 153: 1188–1198.
Zipfel, C., Kunze, G., Chinchilla, D., Caniard, A., Jones, J.D.G.,
Boller, T., and Felix, G. (2006). Perception of the bacterial PAMP
EF-Tu by the receptor EFR restricts Agrobacterium-mediated
transformation. Cell 125: 749–760.
Integrated Systems View on Networking by Hormones in Arabidopsis Immunity Reveals Multiple
Crosstalk for Cytokinin
Muhammad Naseem, Nicole Philippi, Anwar Hussain, Gaby Wangorsch, Nazeer Ahmed and Thomas
Dandekar
Plant Cell; originally published online May 29, 2012;
DOI 10.1105/tpc.112.098335
This information is current as of June 14, 2017
Supplemental Data
/content/suppl/2012/05/22/tpc.112.098335.DC1.html
Permissions
https://www.copyright.com/ccc/openurl.do?sid=pd_hw1532298X&issn=1532298X&WT.mc_id=pd_hw1532298X
eTOCs
Sign up for eTOCs at:
http://www.plantcell.org/cgi/alerts/ctmain
CiteTrack Alerts
Sign up for CiteTrack Alerts at:
http://www.plantcell.org/cgi/alerts/ctmain
Subscription Information
Subscription Information for The Plant Cell and Plant Physiology is available at:
http://www.aspb.org/publications/subscriptions.cfm
© American Society of Plant Biologists
ADVANCING THE SCIENCE OF PLANT BIOLOGY