Dissecting Abscisic Acid Signaling
One of the most
Pathways Involved in Cuticle Formation
ABSTRACT
The cuticle is the outer physical barrier of
aerial plant surfaces and an important
interaction point between plants and the
environment. Many environmental stresses
affect cuticle formation, yet the regulatory
pathways involved remain undefined. We
used a genetics and gene expression analysis
in Arabidopsis thaliana to define an abscisic
acid (ABA) signaling loop that positively
regulates cuticle formation via the core ABA
signaling pathway, including the PYR/PYL
receptors, PP2C phosphatase, and SNF1-
Related Protein Kinase (SnRK)
2.2/SnRK2.3/SnRK2.6. Downstream of the
SnRK2 kinases, cuticle formation was not
regulated by the ABA-responsive element -
binding transcription factors but rather by
DEWAX, MYB16, MYB94, and MYB96.
Additionally, low air humidity increased
cuticle formation independent of the core
ABA pathway and cell death/reactive
oxygen species signaling attenuated
expression of cuticle-biosynthesis genes. In
Physcomitrella patens, exogenous ABA
suppressed expression of cuticlerelated
genes, whose Arabidopsis orthologs were
ABA-induced. Hence, the mechanisms
regulating cuticle formation are conserved
but sophisticated in land plants. Signaling
specifically related to cuticle deficiency was
identified to play a major role in the
adaptation of ABA signaling pathway
mutants to increased humidity and in
modulating their immunity to Botrytis
cinerea in Arabidopsis. These results define
a cuticle-specific downstream branch in the
ABA signaling pathway that regulates
responses to the external environment.
Key words: cuticle, abscisic acid,
Arabidopsis, Physcomitrella, Botrytis, land
plants
INTRODUCTION
During early evolution of land plants,
aquatic organisms developed advanced
structures to adapt to the dramatic
environmenta changes of terrestrial life
(Kenrick and Crane, 1997).
important structures is the cuticle, a
hydrophobic layer protecting aerial plant
structures from desiccation (Riederer, 2006).
Cutin and wax are the main components of
the cuticle and are deposited on the surface
of epidermal cell wall from the inside out
(Domı´nguez et al., 2011).
Much effort has been made to understand the
cutin and wax biosynthesis pathways
(Riederer, 2006; Pollard et al., 2008; Yeats
and Rose, 2013; Kunst, 2015). Cutin
monomers are mainly C16 and C18 hydroxy
fatty acids (Kolattukudy, 2001). The
biosynthesis of these monomers starts with
fatty acids from the plastid and continues
with the following key enzymes: cytochrome
P450-family hydroxylase; acyl-activating
enzyme of the LONG-CHAIN ACYL-COA
SYNTHETASE (LACS) family; and
acyltransferases of the GLYCEROL-3-
PHOSPHATE SN-2
ACYLTRANSFERASE (GPAT) family
(Pollard et al., 2008). Loss of LACS2 or
GPAT8 function causes a significant
decrease in cutin monomers (Schnurr et al.,
2004; Li et al., 2007). Wax monomers are
mainly very-long-chain fatty acids
(VLCFAs; C24–C34) comprising alkanes,
secondary alcohols, ketones, primary
alcohols, and wax esters (Kunst and
Samuels, 2003). These compounds are also
synthesized from C16 and C18 long-chain
fatty acids by multiple elongases including
BETA-KETOACYL-COA SYNTHASE
(KCS), BETA-KETOACYL REDUCTASE
(KCR), ENOYL-COA REDUCTASE
(ECR), and BETA-HYDROXYACYL-CoA
DEHYDRATASE (Weng et al., 2010; Yeats
and Rose, 2013), and then transformed by a
series of ECERIFERUM (CER) enzymes
such as CER1, CER2, CER3, and CER4
(Javelle et al., 2011). WAX ESTER
SYNTHASE (WSD1) and MIDCHAIN
ALKANESHYDROXYLASE1 (MAH1)
catalyze the final steps of wax monomer
biosynthesis (Greeret al., 2007; Yeats and
Rose, 2013). Cutin and wax monomers are
transported out of epidermal cells by ATP-
binding cassette (ABC) transporters such as
ABC11 to form cuticle layers (Bird et al.,
2007).
Figure 1. Summary of Known Pathways
Regulating Cuticle Formation. Some
environmental factors and transcription
factors are known to regulate cuticle
biosynthesis. However, the relationships
between these pathways and the regulation
of these transcription factors remain
undefined (dashed lines).
Compared to the knowledge of
cuticle biosynthesis, the regulatory steps in
cuticle formation are less understood. To
date, only a few transcription factors
regulating cuticle formation have been
identified (Figure 1). In Arabidopsis, these
are mainly MYB transcription factors MYB-
16, -30, -41, -94, -96, and -106 (Cominelli et
al., 2008; Raffaele et al., 2008; Seo et al.,
2011; Oshima et al., 2013; Borisjuk et al.,
2014; Lee and Suh, 2015). Other
transcription factors include the
AP2/EREBP transcription factors WAX
INDUCER 1/SHINE 1 (WIN1/SHN1) and
DECREASE WAX BIOSYNTHESIS
(DEWAX; Aharoni et al., 2004; Broun et al.,
2004; Wu et al., 2011; Go et al., 2014), as
well as HD-ZIP IV transcription factor
HOMEODOMAIN GLABROUS 1 (HDG1).
MYB106 is upstream of WIN1/SHN1,
which directly promotes cutin synthesis and
subsequently wax synthesis. MYB106 also
functions together with MYB16 and
activates the synthesis of both cutin and wax
(Kannangara et al., 2007; Oshima et al.,
2013). MYB41 functions as a negative
regulator of cutin synthesis genes and HDG1
as a positive regulator of cutin and wax
synthesis genes (Cominelli et al., 2008; Wu
et al., 2011).
Wax synthesis is promoted by MYB-
30, -94, and -96, and suppressed by
DEWAX (Raffaele et al., 2008; Seo et al.,
2011; Go et al., 2014; Lee and Suh, 2015).
DEWAX expression is responsive to
photoperiod but expression of MYB96 is
regulated by water deficit (Seo et al., 2011;
Go et al., 2014). Other conditions or
treatments that may regulate these
transcription factors remain undefined.
Multiple environmental factors including
dehydration have a significant impact on the
expression of cuticle-related genes (Figure 1;
Shepherd and Wynne Griffiths, 2006; Yeats
and Rose, 2013; Zhu and Xiong, 2013).
However, the signaling pathways involved
in dehydration-regulated cuticle formation
are largely unknown (Figure 1), except for
the role of the plant hormone abscisic acid
(ABA).
Exogenous ABA increases the
cuticle wax content (Kosma et al., 2009);
impaired ABA biosynthesis in the aba2 and
aba3 mutants leads to cuticle deficiency in
Arabidopsis (L’Haridon et al., 2011), and the
ABA-induced transcription factor MYB96
positively regulates the expression of wax-
related genes (Seo et al., 2011). However,
the signaling components between ABA and
ABA-responsive transcription factors have
not been identified for cuticle formation
(Figure 1). The canonic ABA signaling
pathway has been well characterized. ABA
firstly binds to the receptor proteins,
PYRABACTIN RESISTANCE1 (PYR1)/
PYR1-LIKE (PYL)/REGULATORY
COMPONENTS OF ABA RECEPTORS
(RCAR), which then interact with TYPE 2C
PROTEIN PHOSPHATASEs (PP2Cs) to
release the SNF1- RELATED PROTEIN
KINASEs (SnRK2s, mainly SnRK2.2/
SnRK2.3/SnRK2.6) to activate downstream
ABSCISIC ACID RESPONSIVE
ELEMENT-BINDING FACTORs, such as
ABF2, ABF3, and ABF4 (Choi et al., 2000;
Fujii and Zhu, 2009; Fujii et al., 2009, 2011;
Gonzalez-Guzman et al., 2012).
ABA levels are regulated via both
biosynthesis and inactivation by
hydroxylation or conjugation and transport
within the plant (Merilo et al., 2015). In
addition to regulating responses to water
deficit, ABA is considered as a negative
regulator of plant immunity against Botrytis
cinerea (Botrytis), a model necrotrophic
fungal pathogen. This was mainly deduced
from the immunity phenotype of ABA
biosynthesis mutants (Audenaert et al., 2002;
Ton and Mauch-Mani, 2004; Asselbergh et
al., 2007). Intriguingly, plants with a
defective cuticle have increased Botrytis
immunity (Bessire et al., 2007). Since ABA-
deficient mutants also have permeable
cuticles (Curvers et al., 2010; L’Haridon et
al., 2011), it is difficult to dissect the roles of
ABA signaling and cuticle formation in the
response to Botrytis. In the evolutionary
history of plants, ABA signaling and the
cuticle appeared in the same timeframe in
early land plants (Riederer, 2006; Hauser et
al., 2011). ABA signaling components and
their regulation of stomata, which are
another key structure maintaining water in
plants, are evolutionarily conserved in the
basal land plant, Physcomitrella patens
(Physcomitrella) (Komatsu et al., 2009).
Recently, a few cuticle-related genes have
been identified in Physcomitrella, such as
the wax precursor transporter gene
PpABCG7 and the cutin biosynthesis gene
PpCUS1 (Buda et al., 2013; Yeats et al.,
2014). However, whether ABA regulation of
cuticle formation is evolutionarily conserved
has not been reported.
Figure 2. Core ABA Signaling
Regulates Cuticle Permeability. (A) Dye
exclusion assay. Fully elongated leaves of 3-
week-old plants were stained with 5 ml of
0.05% toluidine blue (TB) drops for 2 h.
Leaves with deficient cuticle stained dark
blue. Scale bar, 0.5 cm. (B) TB-stained areas
were quantified in ImageJ. Combined results
of four experiments (n = 12 in each
independent biological repeat) were
analyzed in a linear mixed model with
single-step p-value adjustment. Error bars
represent SE of means (N = 48 in total).
Letters above the bars indicate significance
groups (p < 0.05). (C) Young leaves were
not TB permeable in snrk2.236. Symptoms
of whole rosettes of 2-week-old plants
immersed in TB solution for 30 min. Scale
bar, 1 cm.
Here we used a genetic and gene
expression analyses in Arabidopsis thaliana
to define an evolutionarily conserved
cuticlespecific downstream branch in the
ABA signaling pathway that regulates
responses to the external environment.
RESULTS
The Core ABA Signaling Pathway
Regulates Cuticle Permeability
To test whether the core ABA
signaling components or other ABA-related
processes can regulate cuticle formation, the
following mutants were used: aba deficient3
(aba3), impaired in ABA biosynthesis
(Xiong et al., 2001); pyrabactin resistance1
(pyr1) pyr1-like1 (pyl1) pyl2 pyl4 pyl5 pyl8
(112458), a sextuple mutant lacking
PYR/PYL receptors (Gonzalez-Guzman et
al., 2012); aba insenstive1-1 (abi1-1), a gain-
of-function mutant of the ABA negative
regulator PP2C (Kanno et al., 2012); snf1-
related protein kinase2.2 (snrk2.2) snrk2.3
snrk2.6 (snrk2.236), an ABAinsensitive
triple mutant (Fujii and Zhu, 2009); and
abscisic acid responsive element-binding
factor2 (abf2) abf3 abf4 (abf234), a triple
mutant lacking three ABFs that are highly
induced by ABA and drought (Choi et al.,
2000). Furthermore, the following
mutantswith increased response or
accumulation of ABA were included: abo5
(aba oversentive5), and the double mutant
cyp707a1 cyp707a3 (cyp707a1a3) impaired
in ABA inactivation (Okamoto et al., 2006;
Liu et al., 2010). The cuticle integrity of
these mutants was tested with the classic
toluidine blue (TB) dye exclusion assay, in
which TB only permeates defective cuticles
to stain the cell walls (O’Brien et al., 1964;
Tanaka et al., 2004). Notably, the
ABAdeficient and -insensitive mutants aba3,
112458, abi1-1, and snrk2.236 exhibited
increased permeability (Figure 2A and 2B).
In contrast, abf234 did not show a
significant difference from wildtype (Figure
2A and 2B). This defined that ABA
signaling, from biosynthesis and perception
to the SnRK2 kinases, was required for the
formation of a proper cuticle, while the ABA
downstreampathways mediated by ABFs
were not required for cuticle formation. The
ABA-hypersensitive mutant abo5 and the
ABAhyperaccumulation mutant cyp707a1a3
exhibited a slightly, but not significantly,
lower permeability compared to wild-type
(Figure 2A and 2B). Overall, this illustrates
that ABA positively regulated cuticle
formation via its core signaling components.
Old leaves absorbed more TB than
the younger leaves. Even in the most
permeable mutant, snrk2.236, the youngest
four leaves were nearly free from stains
(Figure 2C). This suggests the ABA-
regulated cuticlen permeability is
developmental stage dependent.
Regulation of Cuticle-Related
Genes by ABA
To explore how ABA signaling
affects cuticle formation, the expression of
cuticle-related genes were examined in ABA
mutants using real-time quantitative PCR
(qPCR). All transcription factors
documented to be regulators of cuticle
formation and the key cuticle biosynthesis
and transporter genes were tested (Figure 3).
The qPCR data were subjected to cluster
analysis to find correlations among mutants
(Figure 3). The raw data used in making
Figure 3 are presented in Supplemental
Table 1. The ABA-deficient or -insensitive
mutants, with permeable cuticles, were
clearly distinct from the genotypes with
normal cuticle (Figure 3). Significantly
lower expression of most genes was
observed in the cuticle-permeable group.
These data and the TB staining results
(Figure 2) illustrate a positive role for ABA
signaling in the regulation of cuticle-related
genes.
The positive regulators of cuticle
formation, MYB16 and MYB96, had
significantly lower expression in all mutants
with a permeable cuticle compared with wild
type. MYB94 had decreased expression in
abi1-1 and 112458, while MYB106 had
decreased expression in 112458 (Figure 3;
Raffaele et al., 2008; Seo et al., 2011;
Lee and Suh, 2015). Although WIN1/SHN1
and MYB30 are important for cuticle
formation (Broun et al., 2004; Kannangara et
al., 2007; Raffaele et al., 2008), their
expression did not show any clear trends
among the tested mutants, suggesting that
WIN1/SHN1 and MYB30 regulate cuticle
formation independent of ABA signaling
(Figure 3). Interestingly, DEWAX and
HDG1 behaved oppositely to the other genes
tested (Figure 3). DEWAX, an AP2/ERF-
type transcription factor, is the only known
negative regulator of wax synthesis (Go et
al., 2014). The expression of DEWAX
increased in the mutants impaired in ABA
biosynthesis and signaling, thus ABA may
suppress DEWAX expression to release wax
formation. HDG1, encoding a class IV
homeodomain transcription factor that
promotes cuticle formation (Wu et al.,
2011), had increased expression in ABA-
signaling mutants. This implies that HDG1
may function in a feedback loop to promote
cuticle biosynthesis in the plants impaired in
ABA signaling (Figure 3).
The biosynthesis and transporter
genes were expressed in a pattern similar to
that of transcription factors, lower in
cuticlepermeable mutants compared with
mutants with normal cuticles. Among these,
GPAT8, ABCG11, LACS2, and CER1 had
higher expression in cyp707a1a3 and lower
expression in 112458 and snrk2.236 (Figure
3). In addition, LTPG1 and BDG also
exhibited a similar trend although their
expression was not significantly different
between wild-type and
ABAhyperaccumulation/- sensitive mutants
(Figure 3). Four of these six genes (GPTA8,
BDG, LACS2, and ABCG11) were involved
in cutin biosynthesis-related processes;
hence ABA signaling regulated the synthesis
and transport of both wax and cutin.
Interestingly, some cuticle-related genes had
lower expression in the ABA-hypersensitive
abo5. This included the key genes of wax
biosynthesis (CER4,WAX2, andWSD1) and
awax transporter (ABCG12; Figure 3; Yeats
and Rose, 2013). Similarly, the cutin
synthesis gene DCF (Rautengarten et al.,
2012) and a positive regulator of wax
biosynthesis, MYB94 (Lee and Suh, 2015),
exhibited decreased expression in the ABA-
hyperaccumulation mutant cyp707a1a3
(Figure 3). This could indicate feedback
effects when ABA signaling is disturbed.
Figure 3. Core ABA Signaling Regulates
Cuticle Formation via Modulation Cuticle-
Related Gene Expression. Gene expression
in 3-week-old plants (whole
rosettes) was examined with real-time
quantitative RT–PCR (qPCR). Expression of
various cuticlerelated genes (transcription
factors, biosynthesis enzymes, transporters)
were monitored in four independent
biological repeats. Combined means of
expression value (log10) from all
experiments were clustered with the R
function heatmap.2. Stars indicate
significant difference to Col-0 (t-test: ***p <
0.001; **p < 0.01; *p < 0.05). Gene
classifications are indicated with white-black
(Pathway) and color (Function) to present
their molecular roles in cuticle formation.
For the raw data used in creating this
heatmap, see Supplemental Table 1.
Humidity Can Affect Cuticle Formation
Independent of ABA
Dehydration promotes cuticle
synthesis (Kosma et al., 2009; Seo et al.,
2011). Since ABA signaling is also critical
in the response to water deficit (Hauser et
al., 2011), we tested the requirement for
ABA signaling in humidity-induced changes
in cuticle formation. TB staining clearly
showed that a shift to lower humidity
decreased the cuticle permeability of all
tested genotypes (Figure 4A). Notably, the
effects of humidity also took place in aba3,
112458, and snrk2.236, indicating an ABA-
independent pathway regulated by humidity.
The SnRK2 family contains 10
members in Arabidopsis (Hrabak et al.,
2003). In addition to SnRK2.2/3/6, which
are considered part of the core ABA
pathway, we tested the septuple mutant
snrk2.14578910 lacking all other SnRK2s
(SnRK2.1/4/5/7/ 8/9/10; Fujii et al., 2011).
The snrk2.14578910 mutant did not show
significant difference from wild-type in TB
staining (Figure 4A) or gene expression
(Figure 4B). Therefore, none of these
SnRK2s regulated cuticle formation.
Expression of cuticle-related genes
was analyzed to further confirm these
results. Plants were examined at 3 h after
exposure to low humidity. Altered gene
expression after a decrease in humidity was
similar between the wild type and snrk2.236,
the strongest TB-permeable ABA mutant
(Figure 4B). This further supports that
altered air humidity can regulate cuticle
formation through pathways independent of
ABA signaling.
Cuticle but Not Stomata Plays a Role in
Adaptation to High Humidity
Cuticle and stomata are structures gained in
land plants to control water loss (Yeats and
Rose, 2013; Lind et al., 2015). Both are
under control of the core ABA signaling
pathway (Figures 2 and 3; Hauser et al.,
2011; Lind et al., 2015). Evolutionarily,
ABA signaling is an advantage to the water-
limited life of terrestrial plants but may
come at a cost for plants when water is not
limited, such as under water-saturated
conditions, i.e. when well watered and under
high humidity. Weperformed an assay to test
the role of ABA and high humidity on plant
growth. In normal humidity (70%), seedlings
of both 112458 and snrk2.236 were smaller
than the wild type; however, under water-
saturated conditions, at 100% humidity
relative to their respective controls,
seedlings of 112458 and snrk2.236 were
larger than the wild type (Figure 5A).
Furthermore, in the wild type all leaves
except cotyledons and the first two true
leaves were epinastic and water soaked,
while leaves of 112458 and snrk2.236 did
not display these symptoms (Figure 5A,
lower panel). This suggests that ABA
signaling also regulates developmental
responses to increased humidity. To dissect
the role of stomata from permeable cuticle,
we tested slac1 (slow anion channel-
associated 1), in which stomata are more
open than in the wild type (Vahisalu et al.,
2008), and lacs2, which is impaired in
cuticle biosynthesis but has intact ABA
signaling (Bessire et al., 2007). Symptoms
from growth at high humidity in slac1 were
indistinguishable from the wild type (Figure
5A). However, leaves of lacs2 were less
epinastic than that of the wild type and were
not water soaked (Figure 5A).
Taken together, these results indicate that
cuticle permeability, but not stomata, plays a
role in the adaptation to watersaturated
conditions, and implicate the core ABA
signaling pathway as a negative regulator in
this process.
Cuticle Formation Genes Are Conserved
in Land Plants
Most of our knowledge of cuticle formation
is based on studies of higher plants. To
assess their evolutionary conservation and
origins, we analyzed the similarity of
cuticle-related proteins in various lineages.
Using genes selected from the literature and
annotation at The Arabidopsis Information
Resource (TAIR), the proteins encoded by
60 cuticle-related Arabidopsis genes were
analyzed in plant species including algae
(Coccomyxa subellipsoidea,
Chlamydomonas reinhardtii, and Volvox
carteri), bryophyte (P. patens), lycophyte
(Selaginella moellendorffii), gymnosperm
(Picea abies), and angiosperm lineages
(Oryza sativa, Amborella trichopoda,
Solanum lycopersicum, and A. thaliana).
Blast-P function was used to query these
species for the best protein hits to
Arabidopsis (Figure 5B); gene names, AGI
codes, and raw data used are presented in
Supplemental Table 3. The percent identity
of these blast hits were clustered using the
Pheatmap package in R. Most proteins were
conserved in land plants; with sequence
similarity first appearing in the moss
Physcomitrella (Figure 5B). However, there
were exceptions to this. Some proteins from
two clades in the middle were also
conserved in algae (Figure 5B). Additionally
some cuticle-related proteins, in a clade at
the bottom, were highly divergent; these
exhibited low sequence similarity between
Arabidopsis and all other species (Figure
5B).
Figure 4. Humidity-Responsive
Changes in Cuticle Formation Are
Independent of ABA Signaling. (A) Low-
humidity treatment decreased cuticle
permeability. Three-weekold plants grown
under 100% humidity (white bars) or shifted
into 70%humidity for 1 week (gray bars)
were stained with TB. Scale bar, 0.5 cm.
Combined results of three experiments (n =
12 in each independent biological repeat)
were analyzed in a linear mixed model with
single-step p-value adjustment. Error bars
represent SE of means (N = 36 in total).
Letters above the bars indicate significance
groups (p < 0.05). (B) Changes in humidity-
regulated expression of cuticle genes
independent of SnRK2s. Three-week-old
plants grown under 100% humidity or
treated with 3 h of 70% humidity were
examined with qPCR. Means of three
repeats were clustered into a heatmap. A
manual adjustment was subsequently applied
to order the presentation according to 100%
and 3-h 70% humidity treatments. Gene
classifications are indicated with white-black
(Pathway) and color (Function) to present
their molecular roles in cuticle formation.
Stars indicate significant differences
between the two humidity treatments (t-test:
*p < 0.05; **p < 0.01; ***p < 0.001). For
the raw data used in creating this heatmap,
see Supplemental Table 2.
Figure 5. Cuticle Function in
Humidity Adaptation. (A) Arabidopsis
mutants with permeable cuticle are more
adapted to water-saturated conditions. Well-
watered plants were covered with a lid to
maintain 100% humidity. Representative
pictures of 4-week-old plants are shown.
Control, normal watering and constant 70%
humidity; Saturated, well watered at 100%
humidity. Scale bar, 1 cm. Red arrows
indicate water-soaked leaves. (B) Similarity
analysis of Arabidopsis proteins involved in
cuticle formation. The color key (right)
shows the percentage identity ranging from
27 (lowest) to 100 (highest). The heatmap is
ordered based on the increasing average
percent identity from 60 cuticle genes with
the highest hits belonging to Arabidopsis
(rightmost column). The tree is clustered
based on the correlation matrix between
columns with Euclidean distance measure.
Gene classifications are indicated with
white-black (Pathway) and color (Function) related genes in basal land plants are also
to present their molecular roles in cuticle ABA responsive but, in contrast to higher
formation. Lyco., lycophyte. For the raw land plants, ABA is a negative regulator of
data used in creating this heatmap, see
these genes in Physcomitrella.
Supplemental Table 3. (C) Exogenous ABA
suppressed expression of cuticle-related Multiple Pathways Regulating
genes in Physcomitrella. Shoots of 2-month-
Cuticle Responses
old Physcomitrella were treated with the
indicated ABA concentrations. Gene To gain further insight into the
expression was examined with qPCR. Means relationship between ABA signaling, cuticle
and SD of three repeats are shown. Stars formation, and various stress and hormone
indicate significant differences between treatments, publicly available gene
mock (0 mM) and other ABA concentrations expression data were explored. We analyzed
(t-test: *p < 0.05; **p < 0.01; ***p < 0.001). the expression pattern of 136 ABArelated
ABA signaling and the cuticle
genes (Hauser et al., 2011), together with the
evolved in a similar time frame, which is
60 cuticlerelated genes using Bayesian
consistent with the fact that components of
hierarchical clustering (Figure 6A;
ABA signaling, as well as major cuticle-
Supplemental Table 4). Gene-expression
related genes, only exist in land plants
experiments were selected to include
(Riederer, 2006; Hartung, 2010; Hauser et
pathogen treatments, mutants undergoing
al., 2011; Figure 5B). It has been
cell death (acd11), hormone treatments
demonstrated that the ABA signaling
including ABA, abiotic stress treatment, and
components in Physcomitrella, the model
mutants with defective cuticle (bdg, lcr, fdh).
moss species, function similarly to that of
To simplify the interpretation, ABA-related
higher plants in response to ABA and
genes are marked in yellow and cuticle-
regulation of stomatal aperture (Marella et
related genes in blue (Figure 6A).
al., 2006; Rensing et al., 2008; Komatsu et
Experiments that lead to reactive oxygen
al., 2009; Chater et al., 2011; Lind et al.,
species (ROS) production, cell death, or very
2015). Since much of the cuticle-related
strong activation of defenses, including
machinery tested above was conserved, we
Botrytis infection (Figure 6A, group A), had
utilized Physcomitrella to test the ABA
the most striking gene-expression profile,
regulation of cuticle formation in basal land
where ABA-related genes had increased
plants. The function of only a few cuticle-
expression and cuticle-related genes had
related genes has been confirmed in
decreased expression (Figure 6A, clusters I
Physcomitrella. PpABCG7, an ABC
and II). In contrast, ABA and drought
transporter involved in wax precursor
resulted in increased expression of both
trafficking, regulates cuticle morphology
groups of genes (Figure 6A, group B). The
(Buda et al., 2013). PpCUS1 and PpCUS1-
mutants with defective cuticle also showed
like (PpCUS1-L) are two homologs of Cutin
slightly increased expression of both groups
Deficient 1, a cutin biosynthesis gene in
of genes (Figure 6A, group D), indicating a
tomato (Yeats et al., 2014). Unexpectedly,
feedback loop in the regulation of cuticle
exogenous ABA treatment led to decreased
biosynthesis. Other treatments resulted in
expression of these genes in the shoots of
minor changes in gene expression. Data for
Physcomitrella in a dose-dependent manner
the snrk2.236 triple mutant was available for
(Figure 5C). This indicates that cuticle-
both ABA and drought treatment (Umezawa
et al., 2013). This triple mutant was severely
impaired for ABA regulation of both cuticle-
and ABA-related genes, whereas it had a
wild-type-like response to drought (Figure
6A), suggesting again that drought activates
cuticle- and ABArelated genes independent
of SnRK2.2/3/6 (Figure 4B). Overall, the
analysis of public microarray data suggests
that defense and cell death regulate ABA-
and cuticle-related genes via distinct
signaling pathways.
Dissecting the Role of ABA Signaling in
Botrytis Immunity
We have revealed that multiple
independent pathways are involved in
regulating the cuticle in response to ABA,
ROS/ pathogens, and dehydration (Figure
6A). To further verify these results, the
physiological relevance of ABA-dependent
cuticle signaling on plant immunity was
tested. We assayed the Botrytis response of
all mutants used in this study (Figure 6B and
6C). It was found that ABA-insensitive
mutants, which have permeable cuticles,
exhibited significant immunity against
Botrytis, as measured in lesion sizes (Figure
6B). However, ABA hypersensitivity and
hyperaccumulation in abo5 and cyp707a1a3,
respectively, did not confer enhanced
Botrytis susceptibility (Figure 6C). The
abf234 mutant has impaired ABA signaling
but its cuticle permeability and Botrytis
phenotypes were normal (Figures 2A, 2B,
and 6C), indicating that the ABA signaling
pathway must be branched downstream of
the SnRK2s (Figure 7A) and that the branch
containing ABF2/3/4 was not involved in
cuticle formation or Botrytis immunity.
Figure 6. Regulation of ABA and Cuticle-
Related Genes and Botrytis Infection of
ABA Mutants.
(A) Expression analysis of cuticle- and
ABA-related genes in the indicated
experiments. Clustering was applied with
fold changes (log2) of the indicated
experiments. Cuticle-related genes
(indicated with a blue box) were collected
from TAIR (https://www.arabidopsis.org)
and publications; ABA-related genes
(indicated with a yellow box) were from
Hauser et al. (2011). Genes are separated
into four clusters vertically (I, II, III, and IV)
and experiments into five groups
horizontally (A, B, C, D, and E). Yellow and
blue boxes on the right side indicate ABA-
and cuticle-related genes, respectively. For
the raw data used in creating this heatmap,
see Supplemental Table 4. (B and C)
Mutants with impaired ABA signaling and
permeable cuticle were resistant to Botrytis
(B), while ABA mutants with normal
cuticles exhibited wild-type immunity to
Botrytis (C). Lesion diameter of 4-week-old
plants after Botrytis infection (2 3 106
spores ml1) were measured in ImageJ. Four
independent experiments (n = 15 in each
independent biological repeat) were
combined and analyzed using a linear mixed
model with single-step p-value
adjustment.Letters above the bars indicate
significance groups (p < 0.05). Error bars
represent SE of means (N = 60 in total).
DISCUSSION
Profile of ABA-Related Signaling in
Cuticle Formation
Previously, only ABA biosynthesis
mutants were shown to have permeable
cuticles (Asselbergh et al., 2007; L’Haridon
et al., 2011). Thus, the role of downstream
ABA signaling pathways in cuticle
formation remained undefined. Our results
that the sextupleABA receptor mutant
112458 was cuticle permeable genetically
demonstrates that ABA signaling was
positively involved in cuticle formation
(Figure 2A and 2B). his was further
confirmed with abi1 and snrk2.236; thus all
steps from ABA biosynthesis through the
core ABA signaling components regulated
cuticle formation (Figures 2, 3, and 7A).
Although we excluded the branch of ABA
signaling regulated by the ABF2/3/4
transcription factors and implicated several
other regulators, the cuticle-specific ABA
signaling between SnRK2.2/3/6 and the
cuticle-related transcription factors
(DEWAX, MYB16, MYB94, and MYB96)
remains to be defined.
The expression of all previously
defined cuticle-related transcription factors
were examined to identify potential
candidates involved in the ABA-cuticle
regulatory pathway (Figure 3). Based on our
results, we suggest that ABA signaling
promotes cuticle biosynthesis by decreased
expression of a negative regulator of wax
synthesis, DEWAX (Go et al., 2014) and
increased expression of the positive
regulators of cuticle biosynthesis MYB16,
MYB94, and MYB96 (Figures 3 and 7B).
HDG1, a positive regulator of cuticle
biosynthesis, was suppressed by ABA
signaling (Figures 3 and 7B). HDG1 might
regulate a negative feedback loop of ABA
signaling in cuticle biosynthesis.
Additionally the expression of some cuticle
biosynthesis genes is decreased in ABA-
hypersensitive and -hyperaccumulating
mutants; for example, CER4, WAX2,
WSD1, and ABCG12 in abo5, and DCF and
MYB94 in cyp707a1a3 (Figure 3). Thus in
Arabidopsis, although overall ABA
promoted cuticle formation, it acted as a
negative or a positive regulator of cuticle-
related gene expression depending on the
mutant used and the gene studied.
Figure 7. Signaling Pathways Regulating
Cuticle Formation.
(A) Core ABA signaling regulated both
cutin and wax formation (black lines) in
plants, which was independent but
intertwined with humidityinduced signaling
(gray lines) and possibly also pathogen
signaling (dashed gray lines). SnRK2.2/3/6
was a node upstream of three parallel
pathways: cuticle-related signaling, ABF-
regulated ABA signaling (gray), and stomata
signaling (gray). (B) ABA and humidity-
responsive transcription factors. Both ABA
and changes in air humidity suppressed the
expression of DEWAX and HDG1, but
increased the expression of MYB96.
MYB94 was oppositely regulated in
response to ABA and humidity changes.
MYB16 was regulated only by ABA.
Regulation of Cuticle Formation by ABA
in Land Plants
Cuticle, stomata, and ABA-signaling
components appeared concomitantly in early
land plants approximately 450 million years
ago (Kenrick and Crane, 1997; Riederer,
2006; Komatsu et al., 2009; Hauser et al.,
2011). Consistent with this fact, evidence of
the early cuticle can already be seen in the
similarity of cuticle-related proteins in the
moss Physcomitrella (Figure 5B). However,
some genes, especially transcription factors,
were also conserved in the three aquatic
algae lineages (Figure 5B, middle), implying
that the early terrestrial plants may have co-
opted pre-existing regulatory factors to
control the newly evolved cuticle genes. The
clade at the bottom of Figure 5B, consisting
of mainly cuticle biosynthesis and transport
enzymes, was divergent between
Arabidopsis and S. lycopersicum. This is not
surprising, as Arabidopsis is known to have
an atypical cuticle that has distinct
characteristics compared to tomato and other
species used as common models of cuticle
development (Pollard et al., 2008). Indeed,
proteins in this cluster were largely different
from those in all other species, possibly
indicating a complex evolutionary diversity
in cuticle formation that may reflect species-
specific cuticle characteristics. Since ABA
regulates stomatal aperture via the same
mechanism in basal and higher land plants
(Chater et al., 2011; Lind et al., 2015), it was
expected that ABA regulation of cuticle
formation is also conserved. Our results
showed that ABA decreased the expression
of cuticle-related genes in the moss
Physcomitrella, a model non-vascular plant,
in contrast to the complex mixture of mainly
positive but also negative regulation in
higher plants (Figures 3 and 5C). PpABCG7
and PpCUS1 exhibited decreased expression
in Physcomitrella (Figure 5C), while
expression of their Arabidopsis orthologs,
ABCG11 and LTL1, were positively
regulated by ABA (Figure 3), and ABCG11,
which is also known as WBC11, was
previously shown to be ABA induced
(Kosma et al., 2009). Although the
regulation of cuticle-related genes by ABA
is evolutionarily conserved, the nature of this
regulation has changed during the course of
evolution, likely by the addition of new
complex levels of regulation.
Pathogens, ABA, and Drought Signaling
Modulate
Cuticle-Related Genes via Distinct
Pathways The cuticle covers aerial parts of
plants primarily to protect them from
dehydration (Kenrick and Crane, 1997;
Riederer, 2006). Importantly, it also
functions as a barrier to pathogen entry and
other harmful agents, acting as the first
barrier between the plant and its
environment. This multifunctionality of the
cuticle suggests multiple interacting
signaling pathways involved in its
regulation. Clustered gene-expression data
demonstrate that the expression profiles of
cuticle-related genes were similarly
increased in response to ABA and drought,
and in plants with deficient cuticles (Figure
6A, groups B and D), but were decreased in
response to pathogens, cell death, and ROS
(Figure 6A, cluster II, group A). This
demonstrates that pathogens signal via
pathways distinct from these other
treatments. The pathogens used in this
analysis trigger both immune responses and
cell death, both involve ROS signaling
(Lamb and Dixon, 1997; Govrin and Levine,
2000). The other experiments (ozone, acd11,
and ethylene) in the same cluster with
pathogens were all ROS/cell death related
(Figure 6A, group A). Ozone causes
apoplastic ROS formation and defense-like
cell death in plants (Rao et al., 2000;
Overmyer et al., 2005). The acd11 mutant
has a continuous hypersensitive response
that is dependent on ROS burst in plants
(Brodersen et al., 2002). It is also known that
ethylene can activate defense-induced cell
death signaling (Mattoo and Handa, 2003;
Cohn and Martin, 2005). Thus, ROS/cell
death signaling is an additional signaling
pathway distinct from ABA and dehydration
in cuticle formation (Figures 6A and 7A).
The drought response of cuticle genes in the
snrk2.236 mutant (Figure 6A) indicates
additionally that there is an ABA
independent drought response pathway.
Further studies will be required to determine
the relationships and interactions between
these pathways.
Botrytis Immunity of ABA Mutants Is
Primarily Conferred by the Cuticle-
Specific Branch of ABA Signaling
Increased osmotic sensitivity, which
was considered to be an ABA-related
phenotype, is also present in many cuticle-
deficiency mutants (Fujii et al., 2011; Wang
et al., 2011). This indicates a complex
connection between ABA-related and cuticle
deficiency-related signaling. Thus,
dissecting the effects of deficient ABA
signaling, per se, versus its associated cuticle
deficiency is necessary when using mutants
with both of these phenotypes. Since ABA-
and cuticle-related genes were responsive to
many stress treatments (Figure 6A), this
might also influence some previous results
based on mutants that are impaired in the
core ABA signaling pathway.
Here, we used Botrytis infection as
an example to dissect the role of cuticle
deficiency in this ABA-related phenotype.
Previously ABA was considered as a
negative regulator of plant immunity against
Botrytis. Indeed, since aba3, 112458, abi1
and snrk2.236 all had smaller Botrytis
lesions, the simple interpretation is that ABA
is a negative regulator of Botrytis responses
(Figure 6B). However, the Botrytis
immunity of these mutants correlates with
the permeability of their cuticles rather than
their degree of ABA insensitivity. For
example, snrk2.236 had the most permeable
cuticle in this study (Figure 2A and 2B) and
the strongest immunity to Botrytis,
exhibiting significantly more immunity than
112458, which is the most ABA-insensitive
mutant identified so far and can germinate
and grow even on 100 mM ABA (Gonzalez-
Guzman et al., 2012). Thus, studies of ABA
responses must always consider the readouts
used in the isolation of individual ABA-
insensitive mutants. For most mutants, ABA
insensitivity was based on developmental
indexes, such as root growth or germination
inhibition on ABA medium. These
development-specific ABA pathways are at
least partially distinct from the cuticle-
specific branch of ABA signaling, which
controls Botrytis immunity. Finally, ABA-
hypersensitive and -hyperaccumulating
mutants did not exhibit enhanced Botrytis
susceptibility (Figure 6C), further illustrating
the incongruence between immunity and
ABA sensitivity. This suggests that ABA
regulation of Botrytis immunity is a
threshold phenomenon; increasing ABA
levels enough to reform the cuticle
undermines the immunity associated with
cuticle deficiency, but additional ABA
above that required to form a cuticle has no
dose-dependent effect.
The Botrytis-Arabidopsis system has
been well studied and some of the mutants
initially isolated for their Botrytis
phenotypes have subsequently been
characterized for their ABA responses.
These include the two ABA-sensitive
mutants botrytis susceptible1 (bos1) and zinc
finger ankyrin repeat1 (zfar; Mengiste et al.,
2003; AbuQamar et al., 2006; Cui et al.,
2013). Both bos1 and zfar1 are more
susceptible to Botrytis (Mengiste et al.,
2003; AbuQamar et al., 2006). The exact
role of BOS1 and ZFAR1 in ABA signaling
is not yet clear. Notably, bos1 exhibited
enhanced ABA-induced cell death (Cui et
al., 2013). Thus, in addition to regulating
cuticle formation, the role of ABA in
pathogen responses could be related to cell
death regulation and ROS signaling.
In this work we have shown that
ABA positively regulated cuticle formation
via its core signaling members and an
ABF2/3/4-independent sub-branch of ABA
signaling in Arabidopsis. This regulatory
mechanism exhibits signs of a complex
evolutionary history and is not required for
dehydration-induced cuticle formation, but is
required for ABA-mediated adaptation to
humidity and immunity to Botrytis in
Arabidopsis. These results provide a
framework and reference point for future
work defining the molecular mechanisms
regulating cuticle development. However, it
is important to note that although
Arabidopsis is an excellent genetic model, it
has an atypical cuticle (Pollard et al., 2008).
This was seen in our data as differences in
cuticle-related proteins in Arabidopsis
compared to tomato (S. lycopersicum;
Figure 5B), which is a more commonly used
model for cuticle development. Thus, care
should be taken in generalizing and there is a
need to retest these suggested pathways in
other species.
METHODS
Biological Material and Growth
Conditions
A. thaliana seeds were sown on
mixture of peat (B2; Kekkila¨ ; http://www.
kekkila.fi) and vermiculite (2:1) and
stratified for 2 days at
4 C prior to
growth in controlled environment growth
rooms or chambers with a 12-h light/12-h
dark photoperiod, 150–200 mE irradiance,
23 C/18
C (day/night),
and 70% relative humidity. Arabidopsis
Columbia (Col-0) accession was used in all
experiments. Protonemal tissue of
Physcomitrella, ecotype Gransden Wood
(Ashton and Cove, 1977), was grown on a
cellophane membrane (400P; Visella Oy,
Valkeakoski, Finland) placed on BCD
medium (Ashton and Cove, 1977)
supplemented with 1 mM CaCl2, 45 mM
EDTA, and 5 mM ammonium tartrate
[(NH4)2C4H4O6], and solidified with 0.8%
agar. The cultures were grown in a growth
cabinet (Model 3755; Forma Scientific,
Marietta, OH, USA) at
23 C, and 12-h
light/12-h dark photoperiod, light intensity
95 mmol m2 s1 (Valtavalo G3 4000K LED
tubes; Valtavalo Oy, Oulu, Finland).
Staining and Assay
TB solution (0.05%) was used in all staining
assays. Protocol was slightly modified from
published methods (Tanaka et al., 2004;
Bessire et al., 2007). In brief, for 2-week-old
plants whole rosettes were cut and immersed
in TB solution for 30 min; for 3-week-old
plants 5 ml of TB solution was dropped onto
fully expanded leaves, then covered with a
transparent cover for 2 h. Plants were then
washed with tap water and photographed
(Nikon D3100). The photos were used to
measure the sizes of staining areas in ImageJ
(http://rsb.info.nih.gov/ij/). Area data of all
the repeats were analyzed in R to quantify
the results (see Statistics). Similarity Heat
Map of Cuticle-Related Genes The protein
of the primary transcriptome of P. abies was
obtained from the Spruce Genome Project
website (http://congenie.org/start) while
other species were obtained from Phytozome
10.3(http://phytozome.jgi.doe.gov/pz/portal.
html). The set of cuticle-related protein
coding sequences from Arabidopsis was
downloaded from the TAIR website
(https://www. arabidopsis.org). For each
species a blast database of the protein sets
was built and the protein blast function was
used to search in all species for the best hits
to the Arabidopsis cuticle-related proteins.
The percentage identity of these blast hits
were clustered using the Pheatmap package
in R.
Gene Expression Assay
For Arabidopsis, 3-week-old plants
were harvested at 12:00, RNA extracted with
GeneJET Plant RNA Purification Mini Kit
(Thermo Scientific), and reverse
transcription performed with 5 mg of
DNAseI-treated RNA using RevertAid
Premium Reverse Transcriptase (Thermo
Scientific). The cDNA was diluted to 120
ml, and real-time reverse transcriptase qPCR
was performed with 1 ml and three technical
repeats with 53 HOT FIREPol EvaGreen
qPCR Mix (Solis Biodyne). Primer
sequences and amplification efficiencies,
determined with the Bio-Rad CFX Manager
program from a cDNA dilution series, are
provided in the Supplemental Information.
The raw cycle threshold values were put into
to Qbase+ 2.1 (Biogazelle; Hellemans et al.,
2007) using YLS8, PP2AA3, and TIP41 as
reference genes (Supplemental Table 5). For
Physcomitrella, 2-month old shoots were
treated with 1.5 ml 1/4-strength Murashige
and Skoog medium (pH 6.5) with a dose
series of ABA or solvent control in test
tubes, which were then shaken at 100 rpm
for 16 h in continuous light. RNA extraction
was done as for Arabidopsis. Reverse
transcription performed with 1 mg RNA and
cDNA was diluted to 30 ml. qPCR was
performed as for Arabidopsis except using
EF1a and L21 as reference genes
(Supplemental Table 5).
Humidity Treatment
Plants were covered with transparent
lids to maintain high humidity. For the TB
staining assay, two trays of plants were
grown under lids in parallel for 2 weeks,
followed by growth under high humidity
(with lid, 100%) or normal humidity (lid
removed, 70%) for 1 week. The droplet TB
staining was performed on these plants. For
the gene-expression assay, two trays of
plants were grown under lids in parallel for 3
weeks, then one lid was removed to give the
dehydration treatment for 3 h (09:00–12:00).
Whole rosettes of the plants were collected
for gene-expression assay. For high-
humidity adaption, two trays of plants were
grown under lids in parallel for 1 week, after
which one tray was uncovered and supplied
with adequate water. The other tray was
covered and watered to saturation at the third
week until photographed when the plants
were 4 weeks old.
Fungal Cultivation and Disease Assay
Botrytis cinerea (Botrytis) strain
Bo5.10 was grown on homemade potato
carrot agar medium (For 1 l: 300 g potato,
25 g carrot, and 200 g tomato were sliced
and boiled into broth then filtrated, and 10 g
dextrose and 15 g agar added) for conidia
production. Spores with mycelium were
collected using forceps into 1/2-strength
potato dextrose broth (Sigma P6685),
vortexed, filtered with microcloth
(www.merckmillipore.com), and diluted to 2
3 106 spores ml1. For lesion size assay, 3-ml
drops were inoculated onto fully expanded
leaves of 24-day-old plants, closed in a
covered tray at 100% humidity, and
transferred to a growth chamber, 180 mmol
m2 s1, 8 h light/16 h dark at
23 C/18
C (day/night).
Lesions were photographed at 3 days post
infection and diameters measured with
ImageJ. Data from four biological repeats
were pooled and analyzed in R (see
Statistics).
Transcriptome and Cluster Analysis
Raw data from RNA-seq,
Affymetrix, and Agilent Microarray
platform was obtained from several data
sources. From Array Express
(http://www.ebi.ac.uk/arrayexpress/) we
obtained E-ATMX-13 (MeJA); E-MEXP-
3713 (snrk2.236_ABA*, Col-0 ABA*,
drought*, snrk2.236_drought*, snrk2.236).
From Gene Expression Omnibus
(http://www.ncbi.nlm.nih.gov/geo/):RNA-
seq raw data for GSE61542 (ozone) and
GSE66290 (botrytis_RNA.seq) was
analyzed with several packages in the
JAVA-based client-server system Chipster the model fitting. The qPCR data were
(Kallio et al., 2011) and as previously log10-transformed and significance was
described (Xu et al., 2015) GSE5684 estimated by the two-tailed Student t-test
(botrytis_48hr); GSE18666 (heat); using equal variance.
GSE28800 (ABA); GSE14247 (ethylene 4
h); GSE14961 (SA 24 h); GSE5615 (Flg22);
GSE18978 (P. syringae); GSE5530(H2O2);
GSE15105 (bdg, lcr, dfh); GSE16765 (salt
stress); GSE22951(UV-B). Raw data for
acd11 was obtained from John Mundy
(Palma et al., 2010); raw data for high light
was obtained from Tikkanen et al. (2014).
Genes used for cluster analysis were selected
from genevestigator
(http://www.genevestigator.com/gv/). The
raw Affymetrix data were pre-processed
with ‘‘robust multiarray average’’
normalization using the affy package in R
(Gautier et al., 2004). The analysis of raw
Agilent microarray data was calculated with
the half-background correction method
followed by ‘‘quantile normalization’’ using
the Limma package in R (Ritchie et al.,
2015). Differential expression for each
experiment was calculated by log2-based
fold changes in a linear model between
treatment and control, or between mutants
and wild-type using the Limma package in
R. The false discovery rate of differentially
expressed genes for treatment/control and
mutant/wild-type comparisons was based on
the Benjamini and Hochberg method.
Statistics
Statistical analysis of lesion size data
was carried out with scripts in R (version
3.0.3). Using the nlme package, a linear
mixed model with fixed effects for genotype,
treatment, and their interaction was fitted to
the data, plus a random effect for biological
repeat. The model contrasts were estimated
with the multcomp package, and the
estimated p values were subjected to single-
step p-value correction. A logarithm of the
data was taken before modeling to improve