Integration of Arabidopsis thaliana stress-related transcript profiles, promoter structures, and cell-specific expression.

Abstract

Arabidopsis thaliana transcript profiles indicate effects of abiotic and biotic stresses and tissue-specific and cell-specific gene expression. Organizing these datasets could reveal the structure and mechanisms of responses and crosstalk between pathways, and in which cells the plants perceive, signal, respond to, and integrate environmental inputs.
We clustered Arabidopsis transcript profiles for various treatments, including abiotic, biotic, and chemical stresses. Ubiquitous stress responses in Arabidopsis, similar to those of fungi and animals, employ genes in pathways related to mitogen-activated protein kinases, Snf1-related kinases, vesicle transport, mitochondrial functions, and the transcription machinery. Induced responses to stresses are attributed to genes whose promoters are characterized by a small number of regulatory motifs, although secondary motifs were also apparent. Most genes that are downregulated by stresses exhibited distinct tissue-specific expression patterns and appear to be under developmental regulation. The abscisic acid-dependent transcriptome is delineated in the cluster structure, whereas functions that are dependent on reactive oxygen species are widely distributed, indicating that evolutionary pressures confer distinct responses to different stresses in time and space. Cell lineages in roots express stress-responsive genes at different levels. Intersections of stress-responsive and cell-specific profiles identified cell lineages affected by abiotic stress.
By analyzing the stress-dependent expression profile, we define a common stress transcriptome that apparently represents universal cell-level stress responses. Combining stress-dependent and tissue-specific and cell-specific expression profiles, and Arabidopsis 5'-regulatory DNA sequences, we confirm known stress-related 5' cis-elements on a genome-wide scale, identify secondary motifs, and place the stress response within the context of tissues and cell lineages in the Arabidopsis root.

Full text

Page 1

Genome Biology 2007, 8:R49
comment
reviews
reports
deposited research
refereed research
interactions
information
Open Access
2007Ma and BohnertVolume 8, Issue 4, Article R49
Research
Integration of Arabidopsis thaliana stress-related transcript profiles,
promoter structures, and cell-specific expression
Shisong Ma*† and Hans J Bohnert†‡
Addresses: *Physiological and Molecular Plant Biology Graduate Program, University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA.
†Department of Plant Biology, University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA. ‡Department of Crop Sciences, University
of Illinois at Urbana-Champaign, Urbana, IL 61801, USA.
Correspondence: Hans J Bohnert. Email: bohnerth@life.uiuc.edu
© 2007 Ma and Bohnert; licensee BioMed Central Ltd.
This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Stress response in Arabidopsis<p>The integration of stress-dependent, tissue- and cell-specific expression profiles and 5'-regulatory sequence motif analysis defines a common stress transcriptome, identifies major motifs for stress response, and places stress response in the context of tissue and cell line- ages in the <it>Arabidopsis </it>root.</p>
Abstract
Background: Arabidopsis thaliana transcript profiles indicate effects of abiotic and biotic stresses
and tissue-specific and cell-specific gene expression. Organizing these datasets could reveal the
structure and mechanisms of responses and crosstalk between pathways, and in which cells the
plants perceive, signal, respond to, and integrate environmental inputs.
Results: We clustered Arabidopsis transcript profiles for various treatments, including abiotic,
biotic, and chemical stresses. Ubiquitous stress responses in Arabidopsis, similar to those of fungi
and animals, employ genes in pathways related to mitogen-activated protein kinases, Snf1-related
kinases, vesicle transport, mitochondrial functions, and the transcription machinery. Induced
responses to stresses are attributed to genes whose promoters are characterized by a small
number of regulatory motifs, although secondary motifs were also apparent. Most genes that are
downregulated by stresses exhibited distinct tissue-specific expression patterns and appear to be
under developmental regulation. The abscisic acid-dependent transcriptome is delineated in the
cluster structure, whereas functions that are dependent on reactive oxygen species are widely
distributed, indicating that evolutionary pressures confer distinct responses to different stresses in
time and space. Cell lineages in roots express stress-responsive genes at different levels.
Intersections of stress-responsive and cell-specific profiles identified cell lineages affected by abiotic
stress.
Conclusion: By analyzing the stress-dependent expression profile, we define a common stress
transcriptome that apparently represents universal cell-level stress responses. Combining stress-
dependent and tissue-specific and cell-specific expression profiles, and Arabidopsis 5'-regulatory
DNA sequences, we confirm known stress-related 5' cis-elements on a genome-wide scale, identify
secondary motifs, and place the stress response within the context of tissues and cell lineages in
the Arabidopsis root.
Published: 4 April 2007
Genome Biology 2007, 8:R49 (doi:10.1186/gb-2007-8-4-r49)
Received: 27 September 2006
Revised: 2 January 2007
Accepted: 4 April 2007
The electronic version of this article is the complete one and can be
found online at http://genomebiology.com/2007/8/4/R49

Page 2

R49.2 Genome Biology 2007, Volume 8, Issue 4, Article R49 Ma and Bohnert http://genomebiology.com/2007/8/4/R49
Genome Biology 2007, 8:R49
Background
Knowledge about responses of the model plant Arabidopsis
thaliana to abiotic or biotic stresses has accumulated during
the past decade, based on large-scale mutant analyses and
genome-wide transcript profiles. In particular, random muta-
genesis combined with cell-specific or treatment-specific
reporter gene expression has identified many players in the
stress response, whereas microarray-based observations have
revealed transcriptional responses to stresses on a genome-
wide scale [1-4]. However, most analyses have been restricted
to individual genes or treatments. Plant-specific databases,
such as The Arabidopsis Information Resource (TAIR), Gen-
evestigator, and the Nottingham Arabidopsis Stock Centre
(NASC), have begun to collect data from various sources and
merge them with genome sequence-based features [5-8];
however, the data typically exist in isolation. Integrating
these diverse datasets remains a significant challenge in the
assembly of a unifying picture of plant responses to environ-
mental effects. For this purpose, various tools have been
developed, such as MapMan and STKE (Signal Transduction
Knowledge Environment), which begin to link individual
genes to pathways or coregulation circuits [9,10]. Here, we
present an alternative approach to integrating different data-
sets related to plant stress responses.
In Arabidopsis, as in all organisms, a variety of stress factors
that disturb homeostatic conditions bring about ubiquitous
as well as distinct responses at the transcription level. Identi-
fication of ubiquitous, cell autonomous responses is based on
monitoring the status of macromolecules in cells, gauging
DNA damage, protein degradation, or lipid membrane integ-
rity, and eliciting pathways that carry out repair functions
[11]. The degree of damage will trigger this common response,
which must be distinguished from a set of reactions that rec-
ognize and respond to specific stress conditions. Identifying
the genes that determine the specific responses and then sep-
arating them into distinct groups, functional categories, and
pathways is an important task that must be undertaken if we
are to elucidate how plants sense and recognize the environ-
ment, and then embark upon a meaningful defense that will
alleviate the stress condition. The approach presented here
aims to define the distinction between ubiquitous and specific
stress response categories. Very few transcript profiling stud-
ies, which did not include the majority of the Arabidopsis
genes, have addressed specificity and crosstalk of different
stress treatments [1,3,4].
Control over gene expression is in part determined by motifs,
cis-elements, within the promoter sequence of regulated
genes. In plants, distinct motifs have been correlated with
responses to individual treatments, resulting in discovery of a
number of motifs related to stress responses and develop-
mental or organ-specific regulation. Among these motifs,
those responding to light and osmotic and cold stress treat-
ments have been analyzed most intensely [12,13]. Databases
dedicated to plant promoter motifs have been established,
based on motif identification in single or, at most, a few genes
[14,15]. How their competence in regulating gene expression
is mirrored at the genome level has not been tested.
Here, we applied the fuzzy k-means clustering method [16] to
publicly available microarray data from the AtGenExpression
project to compare the response of Arabidopsis to a variety of
abiotic and biotic stresses that disturb homeostatic condi-
tions [17]. The results revealed common as well as distinct
pathways that govern changes in the expression of induced
and repressed genes in response to various treatments. Based
on the collection of motifs in the Plant cis-acting Regulatory
DNA Elements (PLACE) database [14], clusters of coregu-
lated genes were screened for over-represented cis-elements
within their promoters. In addition, gene expression profiles
identifying cell lineages in Arabidopsis roots were used to
correlate the cell type-specific response to various stresses in
the root [18,19]. Integration of information from previously
unconnected databases provided surprising insights about
genes and pathways that classify the evolutionarily conserved
cell-based common stress response, and the divergent path-
ways that organize abscisic acid (ABA)-dependent and ABA-
independent reactions to stress in a tissue-specific manner.
Results and discussion
An analysis of the Arabidopsis abiotic and biotic transcrip-
tome is presented in four sections (Figure 1). First, the overall
clustering pattern for 22,746 probes in response to different
environmental and chemical stress conditions was analyzed.
This was followed by analysis of a 'common stress transcrip-
tome', which unites genes that respond to any deviation from
homeostasis. Then, an analysis of 5'-motifs defined promoter
structures - cis-elements - that are characteristic for individ-
ual clusters of stress-responsive genes, focusing on clusters
containing induced genes (2,715 genes in total) and on the few
large clusters (5,998 genes) containing stress-repressed
genes. Finally, cell-specific and tissue-specific responses to a
variety of stresses were determined by integrating the clusters
defining stress specificity with the gene expression map
established for the Arabidopsis root [19]. This analysis pro-
vided intersections between stress and tissue or cell
specificity.
Clustering of different stress response categories
The fuzzy k-means clustering method [16,20,21] was applied
to the probe set (22,746 in total) printed on Affymetrix Ara-
bidopsis ATH1 chips, which corresponded to about 22,400
genes. In the following analysis, we treated each probe set as
a gene. The external conditions selected included treatments
with a variety of biotic and abiotic stresses included in AtGen-
Express [17], as outlined in a previous analysis that focused
on a subset of salt-responsive genes [21]. Additionally
included were results for different light conditions and expo-
sures of plants to chemicals and growth regulators such as t-
zeatin, tri-iodobenzoic acid, AgNO3, and cycloheximide. The

Page 3

http://genomebiology.com/2007/8/4/R49
Genome Biology 2007, Volume 8, Issue 4, Article R49 Ma and Bohnert R49.3
comment
reviews
reports
refereed research
deposited research
interactions
information
Genome Biology 2007, 8:R49
chemical treatments were included because we expected
them to add additional power of resolution to the analysis.
Considering the large number of genes to be analyzed, fuzzy
k-means clustering was conducted initially with a large cen-
troid parameter (k = 320). Subsequently, 10,490 genes with
significant membership values emerged from the dataset,
which, with the cutoff set at a membership value of 0.035,
most parsimoniously assembled into 180 clusters. The com-
position of 28 clusters (N0 to N27) is shown in Figure 2 and
the entire set is included in the Additional data files 2 and 3.
The 'limma' statistical program was applied to the Affymetrix
dataset to identify differentially regulated genes [22]. Of the
22,746 probe sets, 14,015 were differentially expressed in at
least one condition (P < 10-15). Among the 10,490 significant
genes included in the clustering analysis, 8,520 were differen-
tially expressed and 1,970 were not significantly regulated.
This nonregulated category includes 879 (out of 884) and 119
(out of 131) genes from clusters N6 and N53, respectively.
Genes in cluster N6 were not regulated under most condi-
tions, whereas genes in cluster N53 exhibited a very small
induction in osmotically-stressed roots only (see Additional
data file 4). The separation of clusters N6 and N53 reflects the
discriminative power of fuzzy k-means clustering and sensi-
tivity in identifying even minute differences in expression
patterns. The remaining nonregulated genes were mainly
found in downregulated clusters. In the following analysis of
common stress responses and promoter motifs, we focus our
attention on the 8,520 differentially expressed genes.
The majority of these 8,520 genes was concentrated in a few
large clusters. The most highly populated 15 clusters, each
including more than 100 genes, totaled 5,478 or more than
60% of all significantly clustered transcripts. The largest clus-
ters, namely N0, N2, N5, N18, included 699, 1,206, 705, and
430 genes, respectively. ABA, which acts as an important sig-
naling molecule under a variety of different stress conditions,
was implicated in and induced the expression of genes in clus-
ters N3, N9, N10, N12, N13 and N20, whereas genes in clus-
ters N0, N11, N16, N19 and N28 did not respond to ABA
(Figure 2). Genes in clusters N1 and N8 were induced by light,
and those in cluster N1 were additionally repressed in
response to biotic stress treatments. Genes in cluster N27
were induced by jasmonic acid (JA) treatment, as well as by
salt and wounding stresses. Large clusters in which gene
expression was generally repressed by environmental
stresses included N2, N4, N5, N7, N15, and N18. All genes are
identified in the Additional data files.
The 'universal stress response transcriptome': cluster N12
The 197 genes in cluster N12 (Figure 2) are induced by a broad
range of diverse stress conditions: cold, osmotic, salinity,
wounding, and biotic stresses (including treatments with elic-
itors). The 'limma' analysis indicated that approximately 80%
of these genes were significantly regulated under all treat-
ment conditions, whereas the rest of the included genes were
marginally regulated in one (mostly the wounding treatment)
but significantly regulated in all other conditions (P < 0.01;
Table 1; Additional data file 5). They appear to represent a
common or universal stress response transcriptome because
most of these genes are conserved among plants, animals and
fungi, and are stress regulated in all organisms, with the
inclusion of a few genes related to the plant-specific hor-
mones ABA and JA (Figure 3 and Table 1). Several Gene
Ontology (GO) categories were enriched among these genes:
GO:0009611 (response to
(response to pest, pathogen, or parasite), GO:0006970
(response to osmotic stress), GO:0009737 (response to ABA
stimulus), GO:0009651 (response
GO:0009723 (response to ethylene stimulus), GO:0009751
(response to salicylic acid stimulus), GO:0009753 (response
to JA stimulus), GO:0050832 (defense response to fungi),
GO:0006839 (mitochondrial transport), and GO:008270
(zinc ion binding). Signaling pathways related to mitogen-
activated protein kinase (MAPK), calcium, reactive oxygen
species (ROS), phospholipids, apoptosis, and protein degra-
dation were induced. Equally, part of this cluster of genes that
generally are upregulated by stress is functionally related to
vesicle transport and mitochondrial functions. N12 included
induced genes that had previously been identified as related
to or specific for biotic stresses, but these were also induced
by abiotic stresses, and vice versa. Past restrictions in the
scope of analyses, which typically focused on single treatment
conditions, and the resulting problem of annotation strin-
gency did not compromise the fuzzy k-means clustering anal-
ysis. We discuss these universal stress response genes by
organizing them into different pathways (Figure 3).
wounding), GO:0009613
to salt stress),
MAPK pathways
Several MAPK pathways, organized into signaling cascades,
are conserved in eukaryotic organism [23,24]. In Saccharo-
myces cerevisiae, for example, the high osmolarity glycerol
(HOG) signaling pathway is responsible for osmotic stress
Strategies to identify of Arabidopsis stress-regulated and tissue-regulated genes
Figure 1
Strategies to identify of Arabidopsis stress-regulated and tissue-regulated
genes.
Global gene expression profile
(22,746 genes; AtGenExpress)
Stress treatments:
abiotic, biotic,
elicitors, light
& hormones
Fuzzy k-means clustering identifies
8,520 genes in 180 clusters (N0-N179)
Common stress
response genes;
(cluster N12)
cis-elements
scanning
Tissue- & cell lineage-specific
reactions to st resses in roots
Root-specific
expression profile
Fuzzy k-means
clustering placed 12,360
genes into 19 clusters
Chemical treatments:
t-zeatin, AgNO3,
tri-iodobenzoic acid,
cycloheximide

Page 4

R49.4 Genome Biology 2007, Volume 8, Issue 4, Article R49 Ma and Bohnert http://genomebiology.com/2007/8/4/R49
Genome Biology 2007, 8:R49
sensing [25,26]. The Arabidopsis AtHK1, MEK1, MPK4, and
MPK6 can complement yeast deletion mutants of the HOG
pathways. Other examples of plant MAPKs are alfalfa stress-
induced MAPK (SIMK), tobacco salicylic acid-induced pro-
tein kinase (SIPK), wound-induced protein kinase (WIPK),
and Nicotiana Fus-3-like kinase6 (Ntf6).
Among common genes that are upregulated by stress, several
MAPK components were identified: MPK5, MKK9, and
MAPKKK14. The MAPK pathway has been suggested to be
involved in ethylene signaling [27-29]. Included among ubiq-
uitous stress-regulated genes is also ACS6, encoding the rate-
limiting enzyme of ethylene biosynthesis and a substrate for
MPK6 [30], together with six ERF/AP2 transcription factors
(AtERF). This implicates the ethylene signaling-mediated
engagement of a subset of the MAPK family as a component
of the common stress response.
However, the ethylene response transcriptome is not strictly
clustered in the stress transcriptome, notwithstanding its
importance in developmental processes such as fruit ripen-
ing. Incorporating the results from a study that measured
transcript changes in Arabidopsis Col-0 wild-type [31] into
the cluster structure obtained by fuzzy k-means, the signifi-
cantly ethylene-regulated genes identified in the study were
located in a large number of different clusters.
Clustering of genes in the Arabidopsis transcriptome
Figure 2
Clustering of genes in the Arabidopsis transcriptome. Out of 22,746 genes, 10,671 genes exhibited significant membership values in 180 clusters. The 17
most populated clusters include 7,039 genes (66% of total). Rows represent individual genes; columns (from left to right, as listed below) represent
treatment conditions. A total of 180 clusters emerged. Outlined is cluster 12 (216 genes) including genes that responded to all stress treatment conditions
(see Additional data files). (a) Time course experiments include cold (12 time points), osmotic (12), salt (12), drought (12), oxidative (12), and wounding
(14) treatments. (b) Hormone treatments include ABA (3), ACC (3) and MeJA (3). (c) Biotic stress treatments include bacteria-derived elicitors (12),
Pseudomonas syringae pt. tomato (Pst) DC300 (3), Pst avrRPM1 (3), Pst DC3000hrcC- (3), P. syringae pv. phaseolicola (3), Erysyphe oromoti (7), Phytophtera
infestans (3), P. syringae ES4325 avrRPT2 (5), and P. syingae ES4325 (5). (d) Different light conditions (14). (e) Chemical treatments included t-zeatin, tri-
iodobenzoic acid, AgNO3, and cycloheximide.
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
28
27
(a) (b) (c) (d)(e)
(a) (b) (c) (d)(e)
(a) (b) (c) (d)(e)
c os s d ox w
c os s d ox w
c oss d ox w

Page 5

http://genomebiology.com/2007/8/4/R49
Genome Biology 2007, Volume 8, Issue 4, Article R49 Ma and Bohnert R49.5
comment
reviews
reports
refereed research
deposited research
interactions
information
Genome Biology 2007, 8:R49
Table 1
Selected common stress response genes
Affymetrix probeAGIAnnotationMembership value
257053_at At3g15210ATERF-40.273508
261470_at At1g28370ERF/AP2 transcription factor 0.162002
248799_at At5g47230ATERF-5 0.086731
252214_at* At3g50260 ERF/AP2 transcription factor 0.083494
245250_atAt4g17490 ATERF-6 0.063712
248448_atAt5g51190 ERF/AP2 transcription factor 0.044611
254926_atAt4g11280 ACS6 0.109595
266832_atAt2g30040MAPKKK14 0.054407
245731_atAt1g73500 ATMKK90.165439
254924_at At4g11330ATMPK5 0.060749
247033_atAt5g67250SKIP20.052666
255872_atAt2g30360 CIPK110.093386
261648_atAt1g27730 ZAT10 0.458157
257022_at At3g19580AZF20.194905
248833_atAt5g47120 Bax inhibitor-1, AtBI-10.048683
246453_atAt5g16830 SYP210.089816
254422_atAt4g21560 VPS28 family protein 0.081642
264655_atAt1g09070 SRC20.119115
256238_at At3g12400 tumour susceptibility gene 101 (TSG101) family protein0.051677
265375_atAt2g06530 SNF7 family protein0.125775
262367_at* At1g73030SNF7 family protein0.037115
247204_atAt5g64990Ras-related GTP-binding protein, putative0.048757
260915_at At1g02660lipase class 3 family protein0.100258
254707_atAt4g18010inositol polyphosphate 5-phosphatase II (IP5PII)0.056767
251336_atAt3g61190BON1-associated protein 1 (BAP1)0.152337
262540_atAt1g34260phosphatidylinositol-4-phosphate 5-kinase family protein0.054866
247431_at*At5g62520 SRO5, similarity to RCD1 but without the WWE domain0.048374
247655_atAt5g59820zinc finger protein ZAT120.286685
259879_at* At1g76650calcium-binding EF hand family protein0.09479
266371_atAt2g41410 putative calmodulin0.072498
259137_atAt3g10300 calcium-binding EF hand family protein0.06951
247426_atAt5g62570 calmodulin-binding protein0.068879
247137_atAt5g66210CPK28, calcium-dependent protein kinase0.067785
251636_atAt3g57530CPK32, calcium-dependent protein kinase0.06706
253284_atAt4g34150C2 domain-containing protein0.056923
253915_atAt4g27280 calcium-binding EF hand family protein0.051136
265460_atAt2g46600calcium-binding protein, putative0.038761
249928_atAt5g22250similar CCR4-NOT transcription complex, subunit 7, CAF10.136686