Misexpression of FATTY ACID ELONGATION1 in the
Arabidopsis Epidermis Induces Cell Death and Suggests
a Critical Role for Phospholipase A2 in This Process
Jose ´ J. Reina-Pinto,aDerry Voisin,aSergey Kurdyukov,a,1Andrea Faust,a,2Richard P. Haslam,b
Louise V. Michaelson,bNadia Efremova,aBenni Franke,cLukas Schreiber,cJohnathan A. Napier,b
and Alexander Yephremova,3
aMax-Planck-Institut fu ¨r Zu ¨chtungsforschung, 50829 Ko ¨ln, Germany
bDepartment of Biological Chemistry, Rothamsted Research, Harpenden, Herts AL5 2JQ, United Kingdom
cInstitut fu ¨r Zellula ¨re and Molekulare Botanik, Universita ¨t Bonn, D-53115 Bonn, Germany
Very-long-chain fatty acids (VLCFAs) are important functional components of various lipid classes, including cuticular lipids
in the higher plant epidermis and lipid-derived second messengers. Here, we report the characterization of transgenic
Arabidopsis thaliana plants that epidermally express FATTY ACID ELONGATION1 (FAE1), the seed-specific b-ketoacyl-CoA
synthase (KCS) catalyzing the first rate-limiting step in VLCFA biosynthesis. Misexpression of FAE1 changes the VLCFAs in
different classes of lipids but surprisingly does not complement the KCS fiddlehead mutant. FAE1 misexpression plants are
similar to the wild type but display an essentially glabrous phenotype, owing to the selective death of trichome cells. This
cell death is accompanied by membrane damage, generation of reactive oxygen species, and callose deposition. We found
that nuclei of arrested trichome cells in FAE1 misexpression plants cell-autonomously accumulate high levels of DNA
damage, including double-strand breaks characteristic of lipoapoptosis. A chemical genetic screen revealed that inhibitors
of KCS and phospholipase A2 (PLA2), but not inhibitors of de novo ceramide biosynthesis, rescue trichome cells from death.
These results support the functional role of acyl chain length of fatty acids and PLA2 as determinants for programmed cell
death, likely involving the exchange of VLCFAs between phospholipids and the acyl-CoA pool.
In plants, very-long-chain fatty acids (VLCFAs; fatty acids with
chain lengths >18 carbons) form important structural compo-
nents of membranes and epidermal surfaces. VLCFAs are es-
sential for the normal growth and development of Arabidopsis
thaliana (Bach et al., 2008). They also appear to provide lipid
signals involved in mediating rapid, localized death of plant cells
at the site of pathogen invasion, a process known as hypersen-
sitive response (HR), contributing to resistance (Raffaele et al.,
2008; Wang et al., 2008).
process that occurs on the cytosolic face of microsomal mem-
branes. FAs are initially activated by esterification with CoA,
catalyzed by acyl-CoA synthase. The first step in fatty acid
elongation is catalyzed by a b-ketoacyl-CoA synthase (KCS),
that FAs esterified to a glycerolipid or a phospholipid rather than
CoA may also serve as substrates for elongation reactions
(Hlousek-Radojcic et al., 1998). In yeast and animal species, the
enzymes with KCS activity are encoded by the ELO family of
genes, the genomes of higher plants contain a surprisingly large
unrelated to the ELOs). Based on sequence similarity to the seed-
1995; David et al., 1998), this family comprises 21 members in
Arabidopsis (which, by contrast, has only four ELO genes),
suggesting that the FAE genes take part in a number of plant-
specific pathways. The KCS-catalyzed condensation is the rate-
limiting step in microsomal fatty acid elongation, and current data
reactions in the progressive elongation of fatty acids and deter-
mines the VLCFAs produced. By contrast, the other three core
enzyme activities, which are required for the elongation, play no
direct role in the control of VLCFA synthesis (Millar and Kunst,
1997; Paul et al., 2006). Therefore, manipulation of KCSs via the
use of mutants or overexpressor lines provides a means for
deciphering the functional roles of VLCFAs in cellular responses
and developmental processes.
Among different Arabidopsis KCSs, FAE1 is the best-charac-
terized example (Ghanevati and Jaworski, 2002). It directs two
rounds of elongation of C18 (and probably C16) FAs to produce
1Current address: ARC Centre of Excellence for Integrative Legume
Research, School of Environmental and Life Sciences, University of
Newcastle, University Drive, Callaghan, NSW 2308, Australia.
2Current address: Institut fu ¨r Zellula ¨re and Molekulare Botanik,
Universita ¨t Bonn, Kirschallee 1, D-53115 Bonn, Germany.
3Address correspondence to firstname.lastname@example.org.
The author 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) is: Alexander
WOnline version contains Web-only data.
The Plant Cell, Vol. 21: 1252–1272, April 2009, www.plantcell.org ã 2009 American Society of Plant Biologists
the C20 and C22 species that constitute 13.0 to 21.2% of total
fatty acids in the triglycerides of Arabidopsis seed oil (O’Neill
et al., 2003). These C20 and C22 species are absent in the seeds
of fae1 mutants, which contain only C16 and C18 FAs in their oil
(James et al., 1995). Although VLCFAs occur in sphingolipids in
the plasma membrane, trans-Golgi network, and in recycling
endosomes, C20 and longer fatty acids account for <1% of the
total fatty acids in the leaves of wild-type Arabidopsis plants
(Millar and Kunst, 1997), and most of them accumulate in the
epidermis, as components of cuticular waxes and polyesters.
Characterization of CaMV35S:FAE1 transgenic Arabidopsis
plants (CaMV35S is a cauliflower mosaic virus 35S promoter
sequence) revealed, however, that the plants are capable of
accumulating high levels (>30%) of VLCFAs in leaf membrane
lipids. The transgenic plants with relatively low levels of VLCFAs
(less than ;8.5% [w/w] of total fatty acids in 6-week-old plants)
appeared wild-type but the transgenic plants with high levels of
VLCFA (from ;9.0 to ;13.5%) exhibited a wide range of
morphological changes and some failed to survive (Millar et al.,
1998). To investigate the molecular mechanism by which VLCFA
exert their effects, it can be helpful to use tissue-specific pro-
moters to target the expression of a KCS, such as FAE1,
accurately to appropriate cell types. The epidermis not only
offers a model to study cell-type differentiation and provides the
major physical barrier to invading pathogens and water perme-
ation but also mediates a broad set of defense responses. The
epidermis-specific FIDDLEHEAD(FDH)geneencodes aputative
shows strong expression in vegetative and floral meristems
(Yephremov etal.,1999;Efremova etal.,2004).The fdhmutation
has a deleterious effect on cuticle quality, plant morphology, and
aspects of trichome differentiation. We reasoned that, if VLCFA
biosynthesis affects these aspects of plant development, the
FDH promoter is very suitable for driving expression of a well-
In this article, we report the misexpression phenotype of FDH:
FAE1 transgenic Arabidopsis plants. Most surprisingly, while the
phenotype, which is due to the specific execution of developing
trichome cells. This phenotype contrasts sharply with the pheno-
initiation, morphogenesis, or both. Phospholipase A2 (PLA2) in-
hibitors and chloracetamide herbicides that inhibit KCSs were
found in pharmacological screening to rescuethe development of
trichome cells in the FDH:FAE1 plants. Cytochemical studies also
provide support for the view that the death of trichomes in the
transgenic plants may result from a process akin to lipoapoptosis
in animal cells (Listenberger and Schaffer, 2002).
The FDH:FAE1 Transgenic Plants Show a
To investigate the effects of VLCFA biosynthesis in the epider-
mis, Arabidopsis (Columbia ecotype) plants were transformed
with a construct expressing the FAE1 cDNA from the epidermis-
specific FDH promoter (Efremova et al., 2004). All 27 indepen-
dent FDH:FAE1 T1 transgenic plants were clearly distinct from
and cauline leaves, stems, and sepals (Figures 1A to 1F). Tri-
chome initials clearly formed in young developing leaves (;0.5
cm),but no evidence of the presence of trichomes was observed
and mainly at leaf edges, rudimentary or mature trichomes were
ableto develop.Incontrastwithtrichomes, thepavementcellsin
the epidermis, trichome subsidiary cells, and stomata were not
noticeably affected. The glabrous phenotype of T1 plants seg-
regated as a semidominant Mendelian trait in T2 and T3 gener-
ations, showing that one copy of the transgene was sufficient to
confer the phenotype.
With respect to plant architecture, organ morphology, flower-
ing time, and other visible phenotypes, including those affected
in the most extreme examples of CaMV35S:FAE1 transgenic
plants described by Millar et al. (1998), all FDH:FAE1 transgenic
plants appeared to be normal in the T1 and subsequent gener-
Closer examination of the FDH:FAE1 plants showed that
trichome differentiation was often arrested at the initiation stage
or soon afterwards. Several transcription factors are known to
play essential roles during trichome differentiation. GL1, GL3,
and TTG constitute the primary transcription protein complex
and regulate GLABRA2 (GL2), which is also involved in trichome
initiation. GL2 is required for trichome morphogenesis, and its
expression marks developing trichomes throughout develop-
ment (Szymanski et al., 1998; Ohashi et al., 2002). Histochemical
staining for b-glucuronidase (GUS) activity in double transgenic
showed that, in most trichomes, which were arrested at very
early stages of development in these plants, GUS expression
was not maintained. Only a few trichomes that had reached the
stage at which they formed branches (three, as in the wild-type
plants, or fewer) were stained (Figures 1G and 1H). GL2 is a
homeodomain transcription factor, but it has a so-called StAR-
related lipid transfer (START) domain, raising the possibility that
the activity or cell localization of GL2 may be regulated by a lipid
ligand. Therefore, we also used the native GL2 promoter to drive
expression of a GL2-green fluorescent protein (GFP) fusion in
transgenic GL2-GFP plants and double transgenic FDH:FAE1
GL2-GFP plants. However, the GFP fluorescent signal, when
present, was localized to nuclei (see Supplemental Figure 1 and
Supplemental Movie 1 online), and no evidence for mislocaliza-
tion of the GL2-GFP protein was observed in the FDH:FAE1
plants. These results suggest that either FDH:FAE1 acts on
trichome differentiation upstream of GL2 or that newly initiated
trichomes undergo rapid cell death.
Enzymic Activity of FAE1 Is Required to Suppress
Ectopic expression of the FAE1gene in the epidermis could
affect fatty acid elongation and cell differentiation in two alter-
native ways. The enzymic activity of FAE1, by elevating levels of
longer fatty acids (and accordingly depleting levels of shorter
Lipoapoptosis-Like Process in Plants1253
cell differentiation. Alternatively, FAE1 transcripts might reduce
the efficiency of FA elongation by inducing cosuppression of
protein level by competing for interacting proteins (i.e., other
components of the elongase), cofactors, and substrates.
To study whether the fatty acid elongation activity of FAE1 is
responsible for the suppression of trichome differentiation, we
sprayed FDH:FAE1 transgenic plants with sublethal dosages of
herbicidal chloroacetamides and oxyacetamides (alachlor and
flufenacet, respectively) that are known to inhibit KCSs through
the covalent modification of the active site (Boger et al., 2000;
Trenkamp et al., 2004). Rescue of trichome development be-
came apparent within several days after multiple applications of
these chemicals (Figures 1I and 1J). These results confirm that
enzymatically active FAE1 is required to suppress trichome
development in the transgenic plants. Some rescued trichomes
were arrested at intermediate stages, but most developed three
lateral branches and ultimately reached the normal cell size.
However, they typically showed a flattened appearance, sug-
gesting that the trichome morphogenesis program constitutively
depends on the elongation of fatty acids in FDH:FAE1 plants.
FAE1 Misexpression Differentially Affects Various Fatty
Acid Classes in Acyl-CoAs, Membranes, and
To further prove functional expression of FDH:FAE1, we ana-
lyzed acyl-CoA esters, the key substrates for elongation by the
microsomal system and central metabolic intermediates of lipid
metabolism. The fatty acyl-CoA esters were extracted from leaf
samples, derivatized to fluorescent acyl-etheno-CoAs (Larson
and Graham, 2001), and fractionated by HPLC. As expected, in
FDH:FAE1 leaves,the longer fattyacids wereoverrepresented in
the acyl-CoA pool (Figures 2A and 2C). In particular, the level of
C20 acyl-CoA esters is significantly increased in FDH:FAE1
plants: to 239% of the wild-type level for C20:0, 580% for C20:1,
and 596% for C20:2 (Figure 2A). Leaves of the FDH:FAE1 plants
contained approximately ninefold higher levels of C22:1 acyl-
transgenic plants were able to synthesize C20:3, which were not
found in wild-type leaves and accumulated fewer C18:1, C18:2,
and C18:3 acyl-CoA esters (68, 65, and 66% of the wild-type
were not present in detectable levels in the FDH:FAE1 or wild-
type plants, though this may reflect the poor resolution of
VLCFA-CoAs with >22 carbons under the chromatographic
conditions used. This experiment was repeated twice using
plants from different cultivation setups, and similar results were
obtained. These results prove that FAE1 is functional in the
epidermis and are consistent with the previously proposed
Figure 1. Glabrous Phenotype of FDH:FAE1 Transgenic Plants.
(A) to (F) Comparison of 3.5-week-old wild-type (A) and FDH:FAE1 (B)
rosettes (grown under short-day conditions), wild-type (C) and FDH:
FAE1 (D) inflorescences, and stems from 5-week-old wild-type (E) and
FDH:FAE1 (F) plants carrying cauline leaves and axillary shoots.
(G) and (H) Expression of the GL2:GUS reporter in developing leaves of
2-week-old wild-type (G) and FDH:FAE1 (H) seedlings. Note that most
trichomes remained undeveloped and were not histochemically stained
for GUS in the double transgenics (arrows). Bars = 500 mm.
(I) and (J) Chemical rescue of trichome development with 0.005%
alachlor. (I) and (J) show the same plant before and 1 week after
exposure to alachlor, a potent inhibitor of KCSs. The inhibitor was
applied to 3-week-old FDH:FAE1 seedlings by spraying, to inactivate
FAE1 (I). Leaves were numbered from the base to the apex of the
plant. Young leaves, which developed 1 week after spraying, exhibit
1254 The Plant Cell
function of FAE1 (James et al., 1995; Millar and Kunst, 1997;
Blacklock and Jaworski, 2006).
C20 fatty acids also appeared in total leaf lipids extracted from
the same batches of the FDH:FAE1 plants (Figure 2B), albeit in
CoA fraction), suggesting that incorporation of atypically long FA
chains into membrane lipids in the epidermis is a controlled
process. Given that the epidermis constitutes a minor proportion
of the total volume of the leaf, the expression of FDH:FAE1 is
likely to cause a more pronounced increase in VLCFA levels in
the epidermis while depleting shorter-chain fatty acids.
Elongation of fatty acids up to C34 is required for biosynthesis
of the epi- and intracuticular wax in Arabidopsis. Therefore,
based on the results of acyl-CoA profiling, one would anticipate
that the chain lengths of fatty acids in wax, or the amount of wax
synthesized, should be increased in the FDH:FAE1 plants if the
level of KCS expression plays a significant role in controlling wax
deposition (as it has been shown for CER6, another epidermis
specific KCS; Hooker et al., 2002). However, contrary to this
expectation, biochemical analysis of soluble lipids from epider-
mis revealed that the FDH:FAE1 plants had significantly (P <
the wild type; Figure 3A). Among the functional classes of
compounds in wax, the alkane fraction dropped to ;57% of
that in the wild type (Figure 3A). These wax analysis data are in
the control of the constitutive CaMV35S promoter reduced wax
load by up to 50% (Millar and Kunst, 1997; Millar et al., 1998). In
many eceriferum mutants of Arabidopsis, a severe reduction of
1995), but the FDH:FAE1 plants displayed neither of these
The cuticular polyester cutin is the other major constituent
responsible for the protective functions of the epidermis. It is
difficult to purify cutin from Arabidopsis leaves and stems, but it
has been demonstrated that the lipids left in leaves after ex-
tended exposure to organic solvents reflect the distribution seen
in pure cutin (Bonaventure et al., 2004; Franke et al., 2005). To
Figure 2. Analysis of the Fatty Acid Composition of the Acyl-CoA and Total Lipid Pools and of LCBs in FDH:FAE1 and Wild-Type Plants.
(A) HPLC analysis of VLCFAs in acyl-CoA esters (mean 6 SE; n = 5). Acyl-CoAs were extracted from young rosette leaves of 6-week-old plants (0.5 to
1.0 cm), derivatized to fluorescent acyl-etheno-CoAs, and analyzed by HPLC.
(B) Representative HPLC profiles of acyl-CoAs in FDH:FAE1 and wild-type plants obtained as summarized in (A).
(C) Gas chromatography (GC) analysis of total fatty acids isolated from the same tissues as in (A) (mean 6 SE; n = 5). Fatty acids were transesterified and
analyzed as methyl esters (FAMEs).
(D) LCBs were extracted and analyzed as dinitrophenyl derivatives by reverse-phase HPLC/MS (mean 6 SE; n = 4). In the figure, “d” denotes dihydroxy
LCBs, and “t” denotes trihydroxy LCBs; the numbers designate the length and degree of desaturation of the acyl chain, respectively; t18:1(Z),
4-hydroxy-8-(cis)-sphingenine; t18:1(E), 4-hydroxy-8-(trans)-sphingenine; t18:0, 4-hydroxysphinganine; d18:0, dihydrosphinganine.
Lipoapoptosis-Like Process in Plants1255
study whether the expression of FAE1 might play a role in the
biosynthesis of cutin, we analyzed the residual lipids that remain
bound after exhaustive extraction with chloroform/methanol in
the FDH:FAE1 plants.
Misexpression of FAE1 elicits essentially no change in the
typical cutin constituents, v-hydroxylated fatty acids and a,v-
dicarboxy fatty acids, which are only 16 to 18 carbons in length.
Compared with the wild type, no atypical fatty acids appeared in
the fraction of residual bound lipids in FDH:FAE1 plants (Figure
3B). Supporting these results, a toluidine blue (TB) staining test
did not reveal any overall defects in the cuticle of the FDH:FAE1
transgenic plants compared with wild-type plants (Figure 4U).
implied that cuticle permeability was unchanged in FDH:FAE1
leaves (Figure 4V). A third method used to distinguish cuticular
mutants, based on measuring the chlorophyll extraction rate
(Lolle et al., 1997), also failed to uncover any significant changes
in the FDH:FAE1 plants (Figure 4T), indicating that their cuticle is
expression induced an increase in the levels of saturated and
monounsaturated 2-hydroxylated fatty acids in residual bound
lipids (Figure 3B). The total amount of these compounds in-
creased to 132% of that in the wild type. Increases in the
individual length classes ranged from 110% for C24 to 180% for
C26:1 (Figure 3B). Given that enzymatically isolated cuticles
Figure 3. Composition Analysis of Epicuticular Waxes and Residual-Bound Lipids in Leaves.
(A) Wax was extracted by rapid dipping in chloroform and BSTFA derivatized and analyzed by GC and GC-MS (mean 6 SE; n = 5). Leaves of 6-week-old
plants were harvested for analysis in (A) and (B).
(B) For cell wall–bound lipid analysis, leaves were first extensively defatted with chloroform-methanol. Bound lipids were analyzed using GC and GC-
MS as described previously (Franke et al., 2005; Kurdyukov et al., 2006a) (mean 6 SE; n = 5).
1256 The Plant Cell
Figure 4. Characterization of the Trichome Cell Death Phenotype in FDH:FAE1 Plants.
(A) and (B) Scanning electron micrographs showing the adaxial epidermis of rosette leaves. Note the fully developed trichomes on the wild-type leaf (A)
and the arrested trichomes characteristic of FDH:FAE1 plants (B). Bars = 500 mm.
(C) Close-up view of FDH:FAE1 trichomes, including some that arrested or underwent collapse and degeneration at various developmental stages.
Bar = 100 mm.
(D) and (E) Micrographs of young leaves (left panels) stained with PI and viewed using the DsRed filter set (right panels). Note that PI labels trichomes in
the FDH:FAE1 leaf (E) but not the wild type (D). Bars = 500 mm.
(F) and (G) Micrographs of young leaves (left panels) stained with DCFH-DA and viewed using the GFP filter set (right panels). The DCFH-DA
fluorescence reveals the presence of ROS in FDH:FAE1 trichomes (G) but not in wild-type trichomes (F). Bars = 500 mm.
(H) to (J) Trypan blue staining detects cell death in differentiating TEs in wild-type (H) and FDH:FAE1 (I) leaves and trichome death in FDH:FAE1 ([I] and
[J]). Note that some mature trichomes in (I) are viable and are not stained. (J) shows detail of a FDH-FAE1 trichome. Bars = 250 mm in (H) and (I) and 50
mm in (J).
(K) to (M) Wild-type (K) and FDH:FAE1 ([L] and [M]) leaves stained with DAB to detect hydrogen peroxide. A brown reaction product indicates the
presence of H2O2in trichomes in FDH:FAE1 and in the vasculature of both wild-type and FDH:FAE1 leaves. (M) shows detail of a FDH-FAE1 trichome.
Bars = 250 mm in (K) and (L) and 50 mm in (M).
(N) to (S) Aniline blue fluorescence test to detect callose. Bright blue-white fluorescent spots indicate the presence of callose in FDH:FAE1 trichomes
([O] and [P]) but not in wild-type leaves ([R] and [S]). (N) and (Q) are visible-light images, and (O) and (R) are UV-light images. (P) and (S) show enlarged
Lipoapoptosis-Like Process in Plants1257
bound lipids of chloroform/methanol extracted leaves (Franke
et al., 2005), only a minor fraction of them seems to be derived
from apoplastic polyesters; however, these results do show that
accumulation of 2-hydroxy-VLCFAs is elevated in FDH:FAE1
leaves. Because 2-hydroxylated VLCFA are known to be present
in the sphingolipids, these results also suggest that the fatty acyl
composition of sphingolipid molecular species is affected in
FDH:FAE1 plants. Although this was not analyzed in further
detail, we found no significant difference between transgenic
and wild-type samples either in the absolute amounts of sphin-
as determined by analysis of sphingolipid-specific long-chain
bases (LCBs). Specifically, the ratios of the predominant LCBs
(t18:1, t18:0, and d18:0) in wild-type and FDH:FAE1 plants are
very similar (Figure 2D). These LCBs (and the absence of d18:1)
are diagnostic for glycosylinositol-phosphoceramides (GIPCs),
rather than the glucosylceramides (GluCers). Thus, our data
FDH:FAE1 plants nor perturbation to the overall levels of sphin-
golipids. Collectively, the data presented here show that FDH:
FAE1 misexpression increases carbon chain lengths of acyl-
CoAs but differently affects carbon chain lengths in various fatty
The FDH:FAE1 Phenotype Is Independent of FDH, and FAE1
Is Not Able to Complement fdh
Mutations in FDH, the epidermis-specific promoter of which
drives expression of FAE1 in our transgenic lines, result in
epidermal organ fusions in rosette leaves and inflorescences, a
feature which is easy to recognize (Lolle et al., 1992). The fdh
et al., 1999). The FDH gene encodes a putative KCS and is
required for the formation of cuticle (Yephremov et al., 1999;
Pruitt et al., 2000), but its precise metabolic function is not
known. In fdh plants, we found relatively low levels of C18 and
C20 fatty acids in acyl-CoAs, compared with wild-type plants,
suggesting a deficiency in a C16-elongase, but the changes
observed werenot particularly pronounced. Thedirect biochem-
ical effect of the fdh mutation is difficult to analyze (due in part to
the restricted expression of FDH in leaf tissue) and also because
it induces a strong response in the mutant plants (A. Yephremov
and D. Voisin, unpublished data). FAE1 is active both with
saturated and unsaturated FAs; in particular, the highest activity
has been demonstrated in a condensation assay with 16:0, 16:1,
and 18:0-CoAs (Blacklock and Jaworski, 2006). Since the enzy-
matic function of FAE1 as a KCS is well established, we decided
to determine whether FAE1 could fully or partially complement
the fdh mutation upon expression in the epidermis. If successful,
this experiment would provide good biochemical support for the
contention that FDH is a bona fide KCS with similar substrate
specificity to FAE1. In addition, we sought to determine whether
the fdh mutation might rescue the trichome growth.
We therefore created transgenic plants that express FAE1
under the control of the FDH promoter in the fdh mutant
background. Because the recessive fdh mutant is sterile, the
transgenic FDH:FAE1 plants were crossed with FDH/fdh
heterozygotes, and five heterozygous FDH:FAE1 FDH/fdh F1
plants were selected by progeny testing. In total, 196 F2 plants
were classified on the basis of phenotype, without prior knowl-
edge of their individual genotypes. Based on subsequent PCR
and DNA polymorphism analyses, we then defined six genetic
classes in this segregating population: three nontransgenic and
three transgenic classes (Table 1). In the case of FAE1 providing
complementation of fdh, the transgenic fdh plants should not
exhibit organ fusions or any other features of the fdh mutant.
However, as summarized in the Table 1, all 25 fdh/fdh FDH:FAE1
transgenic plants showed a typical fdh phenotype, as did all nine
nontransgenic fdh/fdh plants. These results show that FAE1 is
not able to compensate for the loss of FDH.
Furthermore, the trichomeless phenotype was not rescued in
the fdh/fdh FDH:FAE1 transgenic plants. On the contrary, be-
cause fdh mutants arepartially glabrous, trichome differentiation
appeared to be further inhibited in the fdh/fdh FDH:FAE1 double
mutant (P < 0.001 by nonparametric Mann-Whitney test), sug-
gesting that fdh and FDH:FAE1 separately contribute to sup-
pressing the development of trichomes.
Cell Type–Specific Cell Death Accounts for the Glabrous
Appearance of FDH:FAE1 Plants
Closer examination of the FDH:FAE1 plants using scanning
electron microscopy revealed that many arrested trichomes
are shrunken and ultimately degenerate, suggesting that cell
death can explain the general lack of mature trichomes and the
loss of GL2 expression (Figures 4A to 4C). To characterize the
process in detail, wild-type and FDH:FAE1 leaves were exposed
to trypan blue (Figures 4H to 4J) and the fluorescent DNA
stain propidium iodine (PI), to which intact cells are imperme-
able (Figures 4D and 4E). FDH:FAE1 trichomes, unlike other
Figure 4. (continued).
Bars = 500 mm in (N), (O), (Q), and (R) and 50 mm in (P) and (S).
(T) Kinetics of chlorophyll leaching from leaves of the indicated genotypes into ethanol solution. Leaves were incubated in 70% ethanol for the times
indicated. Note that FDH:FAE1 leaves behave like the wild type, whereas fdh shows abnormal cuticle permeability.
(U) Staining of leaves of wild-type, FDH:FAE1, and fdh plants with TB to reveal cuticular defects. Adaxial (top panel) and abaxial (bottom panel) sides of
leaves are shown. Note that only fdh tissues were positively stained. Bar = 5 mm.
(V) Calcofluor white test for cuticle permeability. Leaves of wild-type, FDH:FAE1, and fdh plants (top panel) were stained with calcofluor white and
examined under UV light (bottom panel). The permeable fdh leaf was used as positive control; wild-type and FDH:FAE1 leaves were not stained;
however, FDH:FAE1 dead trichomes could be visualized following an a more extended TB staining (see Supplemental Figure 3 online). Bar = 5 mm.
1258The Plant Cell
epidermal and nonepidermal cells, were clearly stained, indicat-
ing that their eventual loss is due to cell-autonomously triggered
death. As well as selectively staining trichome cells in FDH:FAE1
leaves, trypan blue also, but to a lesser extent, reacted with leaf
veins in FDH:FAE1 and wild-type plants (Figures 4H to 4J).
Staining was more often observed in the basal region of devel-
oping leaves where trichomes normally continue to develop
(Szymanski et al., 1998), indicating that the trichome cell mem-
brane becomes abnormally permeable during the early post-
initiation stage. As in the case of pGL2:ICK1/KRP1 plants, which
misexpress the cyclin-dependent kinase inhibitor ICK1/KRP1
(Schnittger et al., 2003), the distal portions of young leaves,
which lacked living trichomes, were hardly stained at all by either
dye. Using PI staining, we also observed that the number of
trichome cells showing irregular chromatin condensation pat-
terns tended to increase from the base toward the distal end of
the FDH:FAE1 leaf. Since cell death does not contribute to the
trichome phenotype of fdh (Yephremov et al., 1999), fdh tri-
chomes were indistinguishable by PI staining from those in wild-
type plants (Figure 4D).
Cell death is often preceded by a burst of reactive oxygen
species (ROS) or reactive oxygen intermediates (Kroemer et al.,
1995), such as hydrogen peroxide, oxyradicals, or organic hy-
droperoxides. Therefore, we assessed intracellular ROS levels in
the FDH:FAE1 plants by staining them with 2,7-dichlorodihydro-
fluorescein diacetate (DCFH-DA). This nonfluorescent com-
pound is able to cross the cell membrane freely and can
be hydrolyzed by cellular esterases to yield the membrane-
impermeable dichlorodihydrofluorescein, which is then oxidized
by ROS in the cell to generate the fluorescent dichlorofluores-
cein. Staining with DCFH-DA (Figures 4F and 4G) gives rise to
intense fluorescentsignalsin developing trichomes in FDH:FAE1
but not in the wild-type plants (unless the latter were damaged).
Nontrichome cells, including adjacent subsidiary cells, were not
stained by DCFH-DA. As with PI staining, living trichomes in the
basal portion of the leaf displayed an intense fluorescent signal,
staining revealed that hydrogen peroxide accumulated in devel-
oping or mature FDH:FAE1 trichomes but not in wild-type
trichomes or other cells except for differentiating tracheids
(Figures 4K to 4M), further supporting the idea that ROS may
be involved in the induction of cell death in FDH:FAE1 leaves.
These findings show that developing FDH:FAE1 trichomes spe-
cifically accumulate increased amounts of intracellular ROS. The
role of ROS has not been studied further, but treatment with the
antioxidant N-acetyl cysteine (sprayed on leaves at 50 mM or
added to the medium at 2 to 10 mM) did not rescue trichomes in
FDH:FAE1 plants (see Supplemental Table 1 online).
To determine the levels of callose deposition, young leaves
from FDH:FAE1 and wild-type plants were stained with the
callose binding dye aniline blue (Figures 4N to 4S). Fluorescent
staining indicative of callose was observed exclusively in the
trichomes of FDH:FAE1 leaves. This result agrees with previous
formation of papillae, preventing invasion by pathogens, in
response to abiotic stresses and during cell death (Hardham
specifically induced in FDH:FAE1 trichomes.
It should be noted that, while the FDH:FAE1 transgenic plants
have overall the same quality of cuticle as wild types, the TB
staining test reveals a cuticular deficiency in their trichomes as
they collapse (see Supplemental Figure 3 online). We noticed
that living trichomes on emerging rosette leaves and some
arrested trichomes at the basal part of bigger rosette leaves
remain unstained, suggesting that it is not a defect of cuticle
barrier that initiates the cell death program. Supporting this view,
we have also observed that cultivation of FDH:FAE1 plants in
vitro under high relative humidity did not rescue trichomes (data
Double-Strand Breaks in FDH:FAE1 Trichomes
In many cell death models, the hallmark of the active, genetically
controlled cell death process is fragmentation of DNA that
nucleases (Walker etal.,2002). Tostudy thisissue,weemployed
a method that detects both single- and double-strand breaks
using 39-hydroxyl ends as primers for terminal deoxynucleotidyl
transferase (TdT). This approach is known as terminal deoxynu-
cleotide transferase dUTP nick end labeling (TUNEL) (Gavrieli,
1992), and it involves the direct labeling of the fragmented DNA
with a nucleotide to which a fluorochrome, biotin, or digoxigenin
has been conjugated (Walker et al., 2002). We developed a
modification of this method in which unlabeled poly-dT tails
produced by TdT were detected by in situ hybridization with a
digoxigenin-labeled poly-dA oligonucleotide probe modified to
contain locked nucleic acid (LNA) residues (LNA-A32-Dig). One
advantage of this method is that the incubation time with TdT
may be greatly extended to increase sensitivity. The other
advantage is that, in principle, various fluorophore-labeled LNA
gene probes may be combined to relate the appearance of DNA
strand breaks to gene expression. We named this method
TUNEL-LID (TUNEL-LNA in situ detection). Generally speaking,
TUNEL-LID labeled cell nuclei as expected and showed little
unspecific background staining. In sections of young leaves,
FDH:FAE1 trichomes could be labeled with this technique, in
5A to 5C), suggesting the abnormal accumulation of unrepaired
Table 1. Complementation Analysis of fdh/fdh FDH:FAE1
aScoring was based on a blind evaluation of trichome phenotypes,
which were ranked according to overall trichome appearance from 3
(completely pubescent leaves of the wild type) to 0 (completely glabrous
Lipoapoptosis-Like Process in Plants1259
DNA strand breaksin these cells.In olderrosette leaves, staining
was also detected in the immature tracheal elements (TEs),
which are known to undergo programmed cell death (PCD)
(Turner et al., 2007) (Figures 5D and 5E). The specific labeling of
FDH:FAE1 trichomes became obvious after typical incubation
periods with alkaline phosphatase (30 to 60 min).
In contrast with single-strand breaks, spontaneously arising
double-strand breaks (DSBs) cannot be repaired efficiently and
could result in the cell cycle arrest and death. Presence of DSBs
et al., 2003).
To detect DSBs specifically, we used in situ ligation, which
relies on T4 DNA ligase-mediated attachment of labeled hairpin-
forming oligonucleotide probes with specific ends to the ends of
DNA in tissue sections (Didenko, 2002). We also developed this
method further, taking advantage of a new fluorescent label
(ATTO 590) with an emission maximum at 624 nm. This com-
pound was used instead of a nonfluorescent biotin tag to
internally label two probes: the LINS probe for blunt ends and
the 3LINS probe for single 39 dN overhangs. Both types of DNA
ends are characteristic of the cleavage products produced by
(Didenko et al., 2003). In addition to allowing the application of
stringent washing conditions at 708C (which was necessary after
DSB labeling), the method has the advantage that red fluores-
cent labeling of apoptotic DNA can be combined with blue or
green fluorescent labeling of nuclear DNA. For dual or triple
labeling experiments,thenucleicaciddyeEvaGreen,because of
its high fluorescence yield and stability, and 4’,6-diamidino-2-
phenylindole (DAPI) were used. Both nucleic acid dyes gave
similar fluorescent patterns, although the DAPI staining some-
times showed higher contrast.
In sections of young rosette leaves, intense ATTO 590 labeling
was observed in most FDH:FAE1 trichome nuclei (Figures 6B
and 6C, arrows), whereas no labeling was seen in wild-type
trichome nuclei (Figure 6A, arrows). Omitting ligase from the
reaction resulted in the absence of ATTO 590 labeling. As
evidenced also by staining with EvaGreen (Figure 6B), some
FDH:FAE1 trichomes displayed chromatin condensation at the
periphery of the nucleus; however, most apparently possessed
highly deformed or collapsed nuclei (Figure 6C). EvaGreen and
DAPI also stained cytoplasmic RNA, although less efficiently,
and this contributed to the background signal in very young leaf
primordia, thereby interfering with chromatin visualization. How-
DAPI were not labeled with ATTO 590 in FDH:FAE1 or wild-type
plants. In particular, cells in the leaf vasculature did not show
nuclear labeling with ATTO 590 (Figure 6D), confirming that
nuclear fragmentation does not occur in the course of differen-
tiation of procambium into TEs, although TUNEL-positive nuclei
are observed there (Fukuda, 2000).
We also attempted to detect the presence of nucleosomal
fragmentation, leading to the appearance of a characteristic
fragment ladder starting at;200 bp upon gel electrophoresis of
nuclear DNA. Because only a small proportion of cells (i.e., the
trichomes) seemed to be affected in FDH:FAE1 plants, we used
linker-mediated PCR (LM-PCR) amplification (Staley et al.,
cells (McLachlan et al., 2000), to screen leaf samples. After
amplification, the expected fragment ladders were observed in
our positive controls (allowing detection of as little as 1.2%
fragmented DNA), but not in FDH:FAE1 or wild-type control leaf
samples. It is worth noting that nucleosomal fragmentation of
DNA does not appear to be indispensable for PCD, as some cell
types produce only high molecular weight fragments (Walker
et al., 2002). Therefore, such nucleosomal DNA fragments must
be essentially absent in FDH:FAE1 plants, or their frequency
must be so low that it is below the detection threshold of the
technique. We concluded that DNA fragmentation that occurs in
the course of cell death in FDH:FAE1 trichomes is detectable by
both TUNEL-LID and the in situ ligation assay.
Endogenous Nuclear DNA Fragmentation in the Epidermis
Using TUNEL-LID, we observed that longer staining periods
resulted in visualization of DNA strand breaks associated with
differentiating TEs (beginning at the time of leaf primordium
formation) and, surprisingly, with nuclei in the epidermis; other
cell types in leaves and apical meristems were not labeled
(Figures 5D and 5E). Although trichomes are not present on the
abaxial surfaces of rosette leaves, the nucleus-specific labeling
(Figures 5F and 5G), and it disappeared nearly totally when the
probe concentration used in the hybridization was reduced by
50%,clearly indicating that the signalis specific to the LNA-A32-
Dig probe. Because many (if not all) nuclei in wild-type epidermal
cells showed the DNA strand break-specific labeling, the breaks
appear to arise during the normal differentiation program (i.e., in
the absence of cell death). When a lower hybridization temper-
ature (618C) and longer staining periods (up to overnight) were
used, TUNEL-LID signals were observed in essentially all nuclei,
Chemical Genetic Screens of FDH:FAE1 Identify Cell Death
Inhibitors in the PLA2 Pathway
Screening for secondary mutations or chemical compounds that
suppress a mutant phenotype are genetic and chemical genetic
approaches, respectively, toward a pathway-based analysis.
Successful rescue of trichomes after spraying with KCS in-
hibitors (Figure 1) shows that the suppression of cell-specific cell
death in the FDH:FAE1 epidermis offers a powerful system for
the latter approach. To initiate the candidate-based chemical
genetic screening, we analyzed the literature regarding potential
targets to be screened and inhibitors of cell death, including
lipoapoptosis. It appears that two pathways in ceramide and
lysophosphatidylcholine (LPC) biosynthesis, both involving acyl-
CoA, converge to determine critical cell fate decisions (Figure 7).
Although the precise mechanisms remain to be determined,
in several animal systems, while their respective biosynthesis
inhibitors suppressed PCD (Chalfant and Spiegel, 2005; Han
et al., 2007). In plants, the strong evidence for involvement of
ceramide metabolites in PCD has been provided by the molec-
ular identification of the ACCELERATED CELL DEATH5 gene as
1260 The Plant Cell
one that encodes a ceramide kinase (Liang et al., 2003) and
FUMONISIN B1 RESISTANT11, coding for an LCB1 subunit of
serine palmitoyltransferase (SPT) (Shi et al., 2007). LPC is pro-
duced by the breakdown of phosphatidylcholine (PC), catalyzed
by PC-specific PLA2. The effective appearance of cell death
symptoms following Botrytis cinerea infection or herbicide par-
aquat treatment in transgenic Arabidopsis plants that overex-
press PLA2 (coded by the patatin-like PLP2 gene) (La Camera
et al., 2005) suggests the pro-PCD role of the PLA2 in plants as
To study whether foliar application of bioactive chemicals can
plants at various concentrations (mostly in the low micromolar
range) and monitored the trichomes daily for rescue of the
lethality. The list of test chemicals included enzyme inhibitors of
phospholipid and sphingolipid synthesis, protein kinase, and
phosphatase inhibitors, lipids, hormones, etc. (>30 compounds
in total; seeSupplemental Table1 online). Although all chemicals
caused visible symptoms at higher concentrations (growth re-
tardation, formation of local lesions, hyponasty, etc.), their ap-
plication did not generally prevent cell death (see Supplemental
Table 1 online).
Out of all compounds tested against the ceramide and LPC
pathways, only a few compounds, including aristolochic acid
(ARA) and bromoenol lactone (BEL; also referred to as haloenol
lactone suicide substrate) were found to definitely possess anti-
PCD activity on FDH:FAE1 trichomes (Figure 8). The effect was
concentration dependent in the range of 10 to 40 mM. Both
compounds are known PLA2 inhibitors (Figure 7); however, they
differ with regard to their mode of action and chemical structure.
While ARA is a noncompetitive, reversible inhibitor of PLA2, BEL
isapotent,irreversible,mechanism-based inhibitor ofPLA2.The
identification of two structurally and mechanistically distinct
inhibitors of PLA2 blocking cell death (ARA and BEL) strongly
suggests that PLA2 is indispensable for inducing and/or execut-
in control and ARA-treated FDH:FAE1 plants (see Supplemental
Figure 4 online). The VLCFA levels were not suppressed by ARA
but elevated significantly (by 1.5-fold) compared with mock-
treated controls, showing that ARA does not inhibit enzymes
involved in the VLCFA biosynthesis.
Figure 5. Detection of DNA Fragmentation in Leaves by in Situ
Relatively thick sections (14 mm) were cut to enhance the visibility of
trichomes. DNA strand breaks were detected by enzymatically labeling
the free 39-OH, followed by hybridization of a specific digoxigenin-
labeled probe to the added tag. Hybridization signals were visualized as
blue-purple intranuclear precipitates. Bars = 200 mm in (A), (B), and (E) to
(G) and 50 mm in (C) and (D).
(A) and (B) Cross sections of shoot apexes at a nonflowering stage,
showing leaf primordia in the wild type (A) and FDH:FAE1 (B). Note
specific TUNEL labeling in FDH:FAE1 trichomes (B).
(C) and (D) High-resolution views of the areas boxed in (A) and (B),
respectively. In (B), the labeling of the FDH:FAE1 trichome (arrow)
indicates the presence of DNA strand breaks.
(E) Differentiating TEs in older wild-type leaves (arrows) labeled under the
same conditions as used in (A) and (B). Nuclear fragmentation is a
normal part of TE formation, which involves PCD.
(F) and (G) Longer staining periods (up to overnight) result in a notable
increase in sensitivity and reveal DNA strand breaks in the epidermis and
developing vasculature of both wild-type (F) and FDH:FAE1 (G) leaves.
Lipoapoptosis-Like Process in Plants1261
On the other hand, blocking de novo sphingolipid synthesis
N-acyltransferase (ceramide synthase) inhibitor Fumonisin B1
(Figure 7) was not effective in preventing the death of trichomes
at the concentrations tested (10 to 40 mM; see Supplemental
Table 1 online), suggesting that the canonical sphingolipid bio-
synthesis pathway is not a major regulator of PCD in FDH:FAE1.
Profiling Membrane Lipids Using Mass Spectrometry
Since the above results suggested the role of membrane lipids
and PLA2s in trichome death in FDH:FAE1, we performed
electrospray ionization (ESI) and tandem mass spectrometry
(MS/MS) analysis to determine whether the levels of mem-
brane lipids are altered in plant tissues. ESI-MS/MS allowed
identification of alterations in >140 diverse polar glycerolipids,
including six head-group classes of phospholipids (PC, phos-
phatidylethanolamine [PE], phosphatidylinositol, phosphatidyl-
serine [PS], phosphatidic acid [PA], and phosphatidylglycerol)
and two head-group classes of galactolipids (monogalactosyl-
diacylglycerol and digalactosyldiacylglycerol). In each class,
molecular species varying in the lengths and desaturation levels
of acyl groups were identified according to their mass spectra
and with previous references (Welti et al., 2002; Devaiah et al.,
Major phospholipids, in which the proportions of VLCFAs are
usually low, accumulated higher amounts of VLCFAs in FDH:
FAE1 plants in a qualitatively similar fashion (Figure 9). For
example, VLCFAs are present in a small proportion in PC (1.9%)
in wild-type plants, but FDH:FAE1 plants accumulate VLCFAs in
Figure 6. Detection of Blunt-End and Single 39 dN Overhang DNA Fragments by in Situ Ligation.
Confocal fluorescence images of ATTO 590 labeling (red) and EvaGreen staining (green) and transmission images (gray) ofcross sections were obtained
from FDH:FAE1 ([B] and [C]) and wild-type plants ([A] and [D]) at a nonflowering stage. EvaGreen preferentially stained the nuclear DNA and, to a lesser
extent, cytoplasm due to indiscriminate staining of RNA. Bars = 20 mm.
(A) Wild-type leaf primordium. Trichome nuclei are not labeled by the ATTO 590 oligonucleotide probe (arrow).
(B) FDH:FAE1 leaf primordium showing bright ATTO 590 labeling of fragmented chromatin (arrow), which has undergone condensation at the periphery
of the trichome nucleus. Note the absence of ATTO 590 labeling in other epidermal cells in (A) and (B).
(C) FDH:FAE1 leaf showing disorganized chromatin labeled with the ATTO 590 probe in two trichomes (arrows).
(D) ATTO 590 fluorescence and transmission images of vascular bundles in a wild-type leaf. The thickened cell walls of TEs display background
fluorescence signals. No evidence for labeling of double-stranded DSBs is seen in this or any other sections of younger or older leaves from wild-type or
1262 The Plant Cell
12.6% of all PC species. Similar proportions are 4.3 and 15.0%
for PE, and 1.3% and 4.7% for phosphatidylinositol, respec-
8.7%), while the lowest increase (76.2 to 86.6%) was detected in
PS species, which already contain a significant fraction of
VLCFAs. Thus, these results are in agreement with our results
Millar et al., 1998), consistent with the interpretation that over-
expression of FAE1 leads to an increase in the VLCFAs levels in
phosholipids. Higher levels of VLCFAs in galactolipids, such as
agreement with the previous results (Millar et al., 1998).
of lysophospholipids, which are minor phospholipid species in
Arabidopsis, in the three classes analyzed (LPC, LPE, and LPG)
are not essentially changed in FDH:FAE1 plants. Since lyso-
phospholipids are obtained when a FA is enzymatically released
from phospholipids by phospholipase A, this suggests that no
change in overall activity of PLA occurs in plant tissues in FDH:
Also interesting is that while levels of VLCFA phospholipids
were increased at the expense of non-VLCFA species in most
classes, C34 PA species (i.e., phosphatidic acid containing 18:0
and 16:0), which do not contain VLCFAs, also accumulated to
been shown to induce cell death in leaves through the ROP-
the interpretation that the PA formation, increased in response
to FAE1 expression, potentiates the PCD. However, inhibitors
of two enzymes that produce PA (Figure 7), phospholipase D
and R59949), were not effective in preventing the death of
trichomes (see Supplemental Table 1 online), warranting further
Cell Type–Specific Death in FDH:FAE1: A Case of a
Lipoapoptosis-Like Process in Plants
This study presents the unexpected finding that epidermal
misexpression of FAE1 in Arabidopsis leads to the death of a
specific cell type, trichomes, via a lipid metabolic pathway. Loss
of viability and membrane integrity, generation of ROS, and DNA
fragmentation lend support to the conclusion that trichomes
undergo PCD. Using the new TUNEL-LID method, we detected
DNA strand breaks in the nuclei of epidermal cells and in
differentiating vascular bundle cells (which can thus serve as a
positive internal control) in wild-type and FDH:FAE1 plants.
Under certain conditions, FDH:FAE1 trichomes could clearly be
distinguished from wild-type trichomes by specific labeling,
Figure 7. Simplified Scheme of Cell Death–Related Acyl-CoA Metabolism and Chemical Inhibitors.
Biosynthesis of ceramides requires two acylation steps catalyzed by SPT and ceramide synthase, which are membrane-bound enzymes active at the
cytosolic face of the endoplasmic reticulum. The eukaryotic glycerolipid pathway in the endoplasmic reticulum involves three acylation steps from G3P
to TAG. PC is the main constituent of cellular membranes. For simplicity, PE and PS are not shown here. C1P, ceramide 1-phosphate; DAG,
diacylglycerol; DGAT, DAG acyltransferase; DGK, diacylglycerol kinase; DHAP, dihydroxyacetone phosphate; FFA, free fatty acid; G3P, glycerol
3-phosphate; GPAT, glycerol-3-phosphate acyltransferase; LPA, lysophosphatidic acid; LPCAT, acyl-CoA:lysophosphatidylcholine acyltransferase;
LPAAT, lysophosphatidic acid acyltransferase; PAP, phosphatidic acid phosphohydrolase; PLD, phospholipase D; S1P, sphingosine 1-phosphate;
SPT, serine palmitoyltransferase; TAG, triacylglycerol.
Lipoapoptosis-Like Process in Plants1263
suggesting that DNA strand breaks could not be efficiently
repaired, leading to chromosomal breaks. It is known that if
DNA strand breaks are not repaired prior to replication, they are
generally harmful, prevent cell cycle progression, and appear to
be sufficient to initiate the events leading to PCD (Nelson and
Kastan, 1994). However, in addition to being primarily respon-
sible for cell death, DNA strand breaks may be a consequence of
a PCD-activated nuclease action or result from ROS attack on
DNA in FDH:FAE1 trichomes.
In situ ligation revealed that the DNA strand breaks in FDH:
FAE1 trichome nuclei give rise to blunt-end and single 39 dN
overhang DNA fragments. It is remarkable that we did not
observe DSB-specific labeling in the vasculature in the same
experiment. It was previously reported that nucleosomal DNA
fragmentation, nuclear shrinkage, and noticeable chromatin
condensation do not occur during TE differentiation when col-
lapse of the vacuole triggers cell death, whereas these charac-
teristic features are displayed during apoptosis-like PCD in
plants when nuclei appear to be the first target of degradation
(Fukuda, 2000). It is noteworthy that, in FDH:FAE1 trichomes, we
detected intracellular ROS, including H2O2, which are also char-
acteristic of the HR, while oxidative bursts have not been
detected during TE formation (Jones and Dangl, 1996) pointing
to a marked difference between these PCDs.
We have shown that levels of long-chain acyl-CoAs esters are
increased in the FDH:FAE1-expressing plants, suggesting a
possible link with lipoapotosis, a process that is induced by
Figure 8. Results of the Chemical Genetic Screen Designed to Identify Chemicals That Rescue Trichomes from Lipid-Induced PCD.
Images were taken within 4 days post-treatment (dpt) as indicated. The images in (C), (F), (I), and (L) show blown-up portions of the images in (B), (E),
(H), and (K), respectively. Leaves are numbered from the base to the apex of the plant. Plants shown here are representative of multiple samples (;100
(A) Mock-treated FDH:FAE1 plant.
(B) and (C) The FDH:FAE1 plant exposed to a high concentration of myriocin (125 mM), an inhibitor of the de novo ceramide synthesis pathway. Note
that myriocin suppresses the growth of leaf primordia but cannot block trichome death.
(D) to (F) Chemical rescue of trichome development in FDH:FAE1 with a PLA2 inhibitor, 30 mM ARA. Trichomes that were dead at the time of spraying
were not recovered.
(G) to (I) Chemical rescue of trichome development in FDH:FAE1 with a mechanistically different PLA2 inhibitor, 30 mM BEL. Note rescue of trichome
death after spraying.
(J) to (L) Chemical rescue of trichome development in FDH:FAE1 with 0.02% clofibrate, a lipid-lowering drug showing hepatoprotective effect in
animals. Clofibrate is a putative LPLAT inhibitor, but this function has not been confirmed in vitro and in plants.
1264 The Plant Cell
Figure 9. Lipid Species Profiling to Detect Changes in FDH:FAE1 Leaf Tissues.
Lipids were extracted from young leaves and analyzed by ESI-MS/MS as described in Methods. The y axis of each plot indicates the amount of lipids
(nmol/mg dry weight); y axis values are the means 6 SE (n = 6). The x axes indicate lipid molecular species (total acyl carbons: double bonds). Acyl chain
composition of major lipid molecular species could be found in Welti et al. (2002). L and H indicate that the value in FDH:FAE1 is lower or higher than that
of wild-type plants; P < 0.01, Mann-Whitney test. MGDG, monogalactosyldiacylglycerol; DGDG, digalactosyldiacylglycerol; PI, phosphatidylinositol;
PG, phosphatidylglycerol; LPG, lysophosphatidylglycerol; LPE, lysophosphatidylethanolamine.
Lipoapoptosis-Like Process in Plants1265
acids, via an acyl-CoA–dependent pathway, in a variety of cell
to mention that lipoapoptosis was not defined purely on mor-
phological grounds but rather as a fatty acid–induced PCD in
animals. Bearing in mind that the course of PCD in plants is
different from that of apoptosis (e.g., cell fragmentation and
phagocytosis of apoptotic bodies do not occur) (Danon et al.,
2000; van Doorn and Woltering, 2005), the term “lipoapoptosis-
here to emphasize the fact that both cell death programs are
possibly triggered by the presence of a critical threshold of
certain fatty acids or lipids.
molecules, they are classically viewed as precursors to numerous
other metabolites containing very long aliphatic chains, including
molecules in animals and plants (Chalfant and Spiegel, 2005).
However, the identification of the responsible death effectors in
lipoapoptosis remains controversial with evidence both for and
against the involvement ofsphingolipids.One explanation may be
that cell type–specific processes for channeling FAs to particular
metabolic fates may define the mechanism by which PCD is
activated (Listenberger et al., 2001). Since the FDH promoter is
active in all epidermal cells (Yephremov et al., 1999; Efremova
diverse cell types are likely to be the causes for the observed
trichome death in FDH:FAE1 plants, in turn mediated by differ-
ences in the metabolic configuration of acyl-exchange pathways
in different cells. It is reasonable to hypothesize that GL2 or other
trichome-specific genes play some role in eventual manifestation
of the death process.
There is an interesting parallel between our findings and data
on palmitate- and stearate-induced lipoapoptosis in Chinese
hamster ovary cells and in human coronary artery endothelial
(HCAE) cells that was also found to be independent of ceramide
synthesis (Listenberger et al., 2001; Staiger et al., 2006). In these
studies, inhibition of ceramide synthase and specific inhibition of
SPT by fumonisin B1 and L-cycloserine, respectively, did not
rescue cells from lipoapoptosis (see Supplemental Table 1 on-
line). Our data are also in a good agreement with the results of
Han et al. (2007), who reported a critical role for the PLA2-
catalyzed formation of LPC in hepatocite lipoapoptosis. In com-
mon with these studies, we found that myriocin did not prevent
PCD (Figure 8), while selective PLA2 inhibitors were quite effec-
tive. The fungal metabolite myriocin is a potent inhibitor of SPT,
which catalyzes the first rate-limiting step in the de novo sphin-
golipid biosynthesis (Figure7)(Kolter andSandhoff, 1996).When
this step is chemically or genetically blocked in plants, the
sphingolipid-mediated pathway of cell death is interrupted
(Spassieva et al., 2002; Shi et al., 2007). Treatment with a major
metabolite of ceramide, ceramide-1-phosphate, which is known
to partially block PCD induction by C2 ceramide (Liang et al.,
2003), also did not promote cell survival in FDH:FAE1 (see
Supplemental Table 1 online). Collectively, these experiments
further support a link between PLA2 and lipoapoptotis-like PCD
and suggest that the initiation of cell death in FDH:FAE1 may not
require de novo sphingolipid synthesis or can occur through a
Fatty Acid–Mediated Cell Death Pathway and PLA2 Operate
in Hypersensitive and Stress Response
Also unexpected was the finding that selective PLA2 inibitors
prevent PCD in FDH:FAE1 plants. While PCD could be a
component of normal development in plants, it occurs during
plant–pathogen interactions and can play a dual role in the path-
ogenesis, both promoting and counteracting the spread of
disease through a HR around the site of infection (Rogers,
2005; Van Breusegem and Dat, 2006; Hofius et al., 2007). The
involvement of lipid-mediated signaling in the HR cell death is
supported by the identification of ENHANCED DISEASE SUS-
CEPTIBILITY1 (EDS1), PHYTOALEXIN DEFICIENT4 (PAD4),
SUPPRESSOR OF SA INSENSITIVITY2 (SSI2), SUPPRESSOR
OF FATTY ACID DESATURASE DEFICIENCY1, and MYB30
genes in Arabidopsis. EDS1 and its interacting partner, PAD4,
encode lipase-like proteins, important activators of salicylic acid
signaling that is involved in cell death control, but their putative
enzymic functions have not been proven (Wiermer et al., 2005).
SSI2/FAB2, which encodes a plastidial stearoyl-acyl carrier
protein-desaturase, preferentially converts stearoyl-ACP (18:0-
ACP) to oleoyl-ACP (18:1D9-ACP). The recessive ssi2 mutation
results in spontaneous lesion formation but confers constitutive
more resistant against bacterial and oomycete pathogens (Shah
et al., 2001). The misregulation of cell death control in ssi2 and
apoptotic effect of saturated FAs and the anti-(lipo)apoptotic
effect of unsaturated FAs in animal cells (Busch et al., 2005;
Shah, 2005). The sfd1 mutant is a ssi2 suppressor, defective in
dihydroxyacetone phosphate reductase/G3P dehydrogenase,
which catalyzes the interconversion of dihydroxyacetone phos-
phate to G3P (Nandi et al., 2004).
is a transcription factor, positive regulator of PCD, and a key
transcriptional activator of the HR (Vailleau et al., 2002). Most
interestingly, MYB30 putative target genes, identified by DNA
microarrays, encode four enzymes involved in VLCFA biosyn-
thesis: KCS, hydroxy-CoA dehydratase, trans-2,3-enoyl-CoA
reductase, b-ketoacyl-CoA reductase, and other lipid-related
proteins (Raffaele et al., 2008). These genes are significantly
avirulent strains 147 of Xanthomonas campestris pv campestris
and DC3000 (avrRpm1) of Pseudomonas syringae pv tomato.
The expression of three KCS genes, KCS1, KCS2, and FDH, is
controlled by MYB30. Based on these findings, a model was
proposed in which MYB30 acts as a strong, positive regulator of
biosynthesis of VLCFAs and that, as yet unidentified, VLCFA-
derived signaling molecules are capable of regulating the HR cell
death and defense responses (Raffaele et al., 2008). Our results
are indicative of possible involvement of PLA2 in this VLCFA-
triggered PCD process.
like PLA2s (PAT-PLA), which are related in amino acid sequence
to calcium-independent PLA2s in animals. Studies with animal
PLA2 inhibitors, such as ARA, BEL, ETYA (5,8,11,14-eicosate-
traynoic acid), PACOCF3, and AACOCF3, showed that they too
can inhibit plant PLA2s in vitro and in planta (Holk et al., 2002;
1266The Plant Cell
Ryu, 2004). BEL is deemed to be not capable of inhibiting
secreted PLA2, which uses a different catalytic mechanism;
therefore, it is most probable that a PAT-PLA enzyme is
responsible for executing the PCD process in FDH:FAE1 plants.
However, this remains to be established in further studies.
Arabidopsis possesses a PAT-PLA family containing nine (La
Camera et al., 2005) or 10 proteins (Holk et al., 2002), which are
localized to the cytoplasm or associated with the plasma mem-
brane. PAT-PLA PLP2 is induced in response to biotic and
was sharply induced in tobacco (Nicotiana tabacum) infected
with tobacco mosaic virus, it was proposed that PAT-PLAs
reaction products contribute to HR cell death against pathogens
(Dhondt et al., 2000). PLP2 expression in transgenic plants was
found to potentiate the appearance of PCD symptoms imposed
by the superoxide radical-generating herbicide paraquat and B.
cinerea and also promoted higher multiplication of avirulent P.
syringae (avrRpt2) (La Camera et al., 2005). Of interest is also the
diverse as bacteria, fungi, and animals. Furthermore, Pseudo-
monas aeruginosa type III secretory toxin ExoU, which is a factor
directly responsible for cell death resulting in acute lung injury, is
a potent PAT-PLA (Tamura et al., 2004).
PAT-PLAs, which appeared to combine phospholipase A1
(PLA1) and PLA2 activities, potentially generate free fatty acids
and a range of lysoglycerolipids, such as LPC, LPE, LPI, LPS,
monogalactosyl monoacylglycerol, and digalactosyl monoacyl-
glycerol. Although LPC has been identified as a death effector in
an animal model (Han et al., 2007), it is unclear at this stage
whether the release of free fatty acids or a lysoglycerolipid, or
both, could give rise to a pro-PCD signal (or an anti-PCD signal
depleted in FDH:FAE1 plants) and whether there is a role for
perturbation and destabilization of membrane structure that
reduces cell viability. It is also plausible that PAT-PLA–induced
reorganization of plasmamembranetriggers relocalization ofraft
proteins and stimulates a signal transduction cascade that
eventually culminates in cell death. Although the focus of this
report is on the unexpected, lipoapoptosis-like death of tri-
altered disease susceptibility when infected with bacteria and
fungi, suggesting that all epidermal cell types may be competent
in responding to the putative PLA2-generated lipid mediators.
Further studies are currently underway to elucidate the mecha-
nism of this phenomenon, but it would be beyond the scope of
this article to address them.
in Fatty Acid–Induced Cell Death
Another, yet not excluding, possibility is that PLA2 inhibitors
block a remodeling pathway (Lands’ cycle) involved in the
dynamic turnover of acyl groups in glycerophospholipids. The
pathway is thought to involve the PLA2-catalyzed hydrolysis of
the sn-2 acyl residue, followed by reacylation of the resulting
lysophospholipid mediated by acyl-CoA:lysophospholipid acyl-
transferases (LPLATs), such as acyl-CoA:lysophosphatidylcho-
line acyltransferase (Yamashita et al., 1997). This metabolic
process has been studied in a variety of tissues in animals,
has not been determined. It is interesting that one of a few
compounds capable of interfering with trichome cell death
appeared to be the antihyperlipidaemic and hepatoprotective
drug clofibrate (Figure 8). Clofibrate [ethyl 2-(4-chlorophenoxy)-
proliferator-activated receptors, which are absent in the genome
of plants, was also reported to act as an LPLAT inhibitor (Riley
and Pfeiffer, 1986). Given that plant LPLAT genes were recently
identified (Sta ˚hl etal.,2007; Hishikawaetal.,2008),acellular site
of action for clofibrate in plants would be interesting to study.
Although details of the acyl remodeling pathway remain to be
determined, the model predicts that both PLA2 and LPLAT
chemical inhibitors should be effective in controlling acyl com-
position of membranes and acyl-CoAs. In such a scenario, their
anti-PCD action in FDH:FAE1 plants may be related to the
protective effects on cellular membranes, which maintain the
obtained from quantitative analysis of acyl-CoA esters extracted
treatment with a PLA2 inhibitor increased VLCFA levels in the
full confirmation of this hypothesis would require an additional
analysis of membrane lipid composition and the use of other
with reduced activity of PLA2 or LPLAT in the future should give
us insight into the mechanisms underlying acyl remodeling and
its potential link to PCD.
Arabidopsis thaliana Plant Material and Growth Conditions
Arabidopsis, Columbia (Col-0) ecotype, was used as a wild type in all
Ohio State University (Columbus, OH). The fdh-3940S1 allele used for
complementation analysis (referred here as fdh) was described previously
(Yephremov et al., 1999). Arabidopsis plants were grown for analyses in a
greenhouse at 22 to 238C and 50 to 60% humidity and kept on an 8-h
photoperiod (shortday)for thefirst6to7weeksaftersowingandona 16-h
or renewal of seeds in a controlled environment chamber.
Generation of Transgenic Plants and Transgenic Complementation
To construct the binary vector pBPF:FAE1, the FAE1 cDNA from
pT7T318U (Millar and Kunst, 1997) was cloned as a SalI-XbaI fragment
into the XhoI/XbaI sites of the binary vector pBPF-SX (Efremova et al.,
2004), placing it under the control of the FDH promoter. Wild-type
Arabidopsis plants (Col-0) were transformed with pBPF:FAE1 by vacuum
infiltration as described previously (Bechtold et al., 1993), and BASTA
was employed for selection of transgenic plants. More than 30 indepen-
dent transformants, all showing a similar trichome phenotype, were
obtained. Homozygous lines were established in four families, and
plants were backcrossed to the Col ecotype to test segregation and
identify lines carrying a single copy of the transgene. Expression of FAE1
driven by the FDH promoter was confirmed by RT-PCR using pAnos
(59-TATTACATGCTTAACGTAATTCAACAG-39) and FAE-T (59-AAAACG-
GTCGGTCCTAATTTGATG-39) as primers.
Lipoapoptosis-Like Process in Plants1267
To determine whether FDH:FAE1 can complement the fdh mutant, fdh
ing five different families. The progeny of double heterozygotes was then
blind experiment, progeny plants were tested for the presence of the
transgene by PCR with the pAnos and FAE-T primers.
To construct the GL2-GFP protein fusion, a 5.6-kb genomic portion
of the GL2 gene was PCR-amplified using GL2XP (59-GATAAAG-
appended with XmaI recognition sequences (underlined). This fragment
comprised 2124-bp sequence upstream the initiation codon and GL2
open reading frame excluding the termination codon. It has been cloned
in frame to GFP into the unique XmaI site in the polylinker of the
pBctGFP binary vector (A. Yephremov, unpublished data). Wild-type
and gl2 plants were transformed with the GL2-GFP construct, and
transgenic plants were selected with BASTA as described above.
Expression the GL2-GFP protein was confirmed by complementation of
to FDH:FAE1 plants, and double transgenic plants were selected by
epifluorescence microscopy and PCR.
automated DHPLC instrument equipped with a DNASep column (WAVE;
Transgenomic) as described previously (Efremova et al., 2004). This
approach relies on different melting behavior of PCR amplicons derived
from FDH, fdh-3940S1, which carries a GT insertion, and their heterodu-
Three-week-old FDH:FAE1 transgenic and wild-type plants with seven to
eight rosette leaves were sprayed three times daily with mock solutions
(controls) and the solutions to be tested. Plants were examined using a
Leica MZFL-III microscope prior to the first treatment and on days 4, 5,
and 7 after treatment. The following concentrations of chemicals were
used if not otherwise indicated: 10, 20, 30, and 40 mM (see Supplemental
targets, and effects).
Microscopy and Staining Techniques
Cryoscanning electron microscopy was performed as described previ-
ously (Yephremov et al., 1999).
Young rosette leaves (;1 cm) or rosettes of 5- to 6-week-old plants
were stained with PI and DCFH-DA (both from Sigma-Aldrich) by incu-
bation for 15 to 30 min and then washed with water for 20 min. PI was
used at 0.1 mg/mL in PBS. The 50 mM stock solution of DCFH-DA was
prepared in dimethylformamide and diluted to 5 mM in water before
staining. Uptake of DCFH-DA and PI by living cells was examined using a
Leica MZFL-III fluorescence microscope equipped with GFP3 (excitation
470/40 nm; emission 525/50 nm) and DsRED (excitation 546/10 nm;
emission 600/40 nm) filters (AHF-Analysentechnik), respectively.
Staining with trypan blue lactophenol, aniline blue, and DAB was
performed as described by Koch and Slusarenko (1990), Adam and
Somerville (1996), and Schraudner et al. (1998), respectively.
Tissues from 6-week-old plants grown on an 8-h/16-h day/night cycle
were fixed in methanol:acetic acid (3:1) at 2208C overnight, and chloro-
phyll was then removed by extraction with 70% ethanol at 2208C
overnight (longer incubation times and several changes of the ethanol
solution may be necessary to ensure extraction). The fixed tissues were
dehydrated in an ethanol series (70, 85, 95, and 23 100% for 5 min each),
transferred from 100% ethanol through an ethanol-Histoclear series to
100% Histoclear (Histoclear: ethanol 1:2, Histoclear: ethanol 2:1, 100%
Histoclear for 5 min each), and embedded into Paraplast plus (Tyco
Healthcare) at 608C. Sections (14 mm) were cut with a microtome and
mounted on SuperFrost Plus adhesion microscope slides (Menzel-
Gla ¨ser). Before the TUNEL treatment, sections were deparaffinized (33
100% Histoclear, Histoclear:ethanol 2:1, Histoclear:ethanol 1:2, 23
100% ethanol), rehydrated in a graded series of ethanol (95, 85, 70, 50,
and 30% for 2 min each), washed in a 23 SSPE buffer at 708C for 20 min,
incubated with 1 mg/mL Proteinase K solution for 20 min at 378C, and
carefully washed again in 23 SSPE at 708C and in 0.85% NaCl at room
temperature. A 200-mL aliquot of tailing reaction mixture, comprising 160
units of TdT (Fermentas) and 1 mM dTTP in the supplied TdT buffer was
applied toeachslide.Slideswereincubatedat378Cfor 1h,washedin23
SSPE and 0.85% NaCl, and dehydrated using an ethanol series (70, 85,
95, and 100% for 5 min each). The probe LNA-A32-Dig used to detect
poly-dT tails was a digoxigenin-labeled 32-nucleotide poly-dA oligo-
nucleotide containing 11 LNA residues (aaAaaAaaAaaAaaAaaAaa-
AaaAaaAaaAaaAaa; LNA residues are shown in upper case). The probe
was synthesized, labeled, and purified by anion-exchange HPLC by the
manufacturer (Exiqon). Its melting temperature is predicted to be 618C
(http://lna-tm.com). This temperature was optimal for dot blots with oligo
(dT)-labeled l phage DNA digests; however, we found that the best
results were obtained when in situ hybridization assays were conducted
at 648C using 5 nM probe. The hybridization buffer contained 50%
0.5 mg/mL yeast tRNA, 10% dextran sulfate, and 13 Denhardt’s solution
(Sigma-Aldrich). A 100-mL aliquot of hybridization mix was applied to
eachslide,covered withacoverslip, andincubated overnight at648Cina
humid chamber. Slides were then washed with 33 SSPE (twice for 5 min
each at room temperature and twice for 20 min at 648C) and NTE (0.5 M
NaCl, 13 TE, pH 7.5; twice for 5 min at room temperature). Slides were
for 5 min at room temperature), 1.53 SSPE (once for 30 min at 648C), and
0.33SSPE (twice for 30 min at 648C). Digoxigenin-labeled hybrids were
detected using antidigoxigenin antibody conjugated to alkaline phos-
phatase diluted 1:3000 in buffer 1 (100 mM Tris-HCl, pH 7.5, 150 mM
in buffer 1 containing 0.1% Tween 20 (for 10 min, 15 min, and 23 20 min)
and equilibrated in the staining buffer (100 mM Tris-HCl, pH 9.5, 150 mM
NaCl, 50 mM MgCl2, and 0.1% Tween 20). For staining, the slides were
incubated in the staining buffer containing the color substrates 5-bromo-
4-chloro-39-indolyphosphate p-toluidine salt and nitro-blue tetrazolium
chloride as recommended by the manufacturer (Roche). The signal
appeared after 30 to 60 min. The reaction was stopped by washing the
slides in deionized water.
In Situ DNA Ligation Assay
The in situ ligation assay was performed using the same plant materials
and preparation techniques as for the TUNEL-LID tailing reaction. Sec-
tions were first preincubated with 13 ligation buffer containing 40 mM
Tris-HCl, 10 mM MgCl2, 10 mM DTT, and 5 mM ATP, pH 7.8 (Fermentas),
slide) containing 13 ligation buffer, 5% PEG 4000 (Fermentas), T4 DNA
ligase (0.15 units/mL) (Fermentas) and the labeled hairpin-forming oligo-
nucleotides LINS and 3LINS (5 pmol/mL of each). LINS (GAATTCCCGG-
GATCCtGGATCCCGGGAATTC) and 3LINS (NGAATTCCCGGGATCC-
tGGATCCCGGGAATTC) were labeled internally during synthesis with
1268The Plant Cell
ATTO 590 (t denotes dT with the attached fluorescent label, N denotes
any of the four nucleotides) and HPLC purified by the manufacturer (IBA).
The slides were covered with cover slips, and the ligation was allowed to
proceed overnight at room temperature in a moist, light-protected
chamber. The slides were washed twice for 10 min with 23 SSPE at
708C and stained with 0.13 EvaGreen fluorescent DNA stain (Biotium) in
23 SSPE at 708C to reveal nuclei or/and RNA. Unbound EvaGreen was
and the slides were examined on a Leica TCS SP2 AOBS spectral
confocal microscope. Fluorescence was observed with excitation at 561
nm and emission at 570 to 651 nm for ATTO 590 and excitation at 488 nm
and emission at 495 to 545 nm for EvaGreen. To reveal nucleic acids,
some slides were also stained with DAPI (2.5 mL/mL in PBS) for 5 min and
washed in PBS for 5 min at room temperature. The DAPI signal was
excited with UV light and recorded at 415 to 470 nm.
DNA Fragmentation Analysis Using LM-PCR
For LM-PCR, genomic DNA, isolated from Arabidopsis using the DNeasy
plant mini kit (Qiagen), was ligated to blunt-end linkers consisting of
oligonucleotides LM24 (59-AGCACTCTCGAGCCTCTCACCGCA-39) and
LM12 (59-TGCGGTGAGAGG-39). Ligation products were purified using a
Microcon 30 device (50-bp cutoff for double-stranded DNA; Millipore)
and amplified using theAdvantage2PCRkit (Clontech Laboratories)with
LM24 as a primer. The PCR program was 948C for 1 min, followed by 15
cyclesof 948C for 20 s, 568C for 30 s,688C for 1min 30 s, and 20 cycles of
948C for 20 s, 568C for 30 s, and 688C for 1 min 30 s with a 5-s increment
PCR products were analyzed by electrophoresis on 1.2% agarose gels.
Positive control ladders were produced from l phage DNA (Fermentas)
digested with SspI (Biolabs) and ligated to the same linkers.
Sphingoid base analysis was conducted essentially as described previ-
ously using sphingosine (d20:1) as an internal standard (Borner et al.,
2005; Sperling et al., 2005). Briefly, 500-mg mixtures (fresh weight) of 22
to 24 4-week-old seedlings of wild-type and transgenic plants were
hydrolyzed in 10% (w/v) Ba(OH)2for 20 h at 1108C, and LCBs were
extracted with chloroform:dioxane:water (6:5:1; v/v/v). LCBs were then
converted to their dinitrophenyl derivatives with methanolic 1-fluoro-2,4-
dinitrobenzene, extracted with chloroform:methanol:water (8:4:3; v/v/v),
and purified by thin layer chromatography on silica plates. The samples
were analyzed by reverse-phase HPLC/MS, and individual compounds
were quantified with respect to the internal standard.
HPLC Analysis of Acyl-etheno-CoA Derivatives and Analysis of the
Total Fatty Acids
Twenty-milligram portions of leaf material were collected from 4-week-old
plants grown under short-day conditions, frozen in liquid nitrogen, and
extracted for subsequent quantitative analysis of fluorescent acyl-etheno-
CoA derivatives by HPLC and GC analysis of the total fatty acids. HPLC
[Agilent 1100 LC system; Phenomenex LUNA 150 3 2 mm C18(2) column]
was performed using the methodology and gradient conditions described
acyl-etheno-CoA derivatives in ARA-treated plants was conducted simi-
leaves was performed as described previously (Sayanova et al., 2007).
Profiling Membrane Lipids
Samples were prepared from young leaves (<1 cm) of 6-week-old
Arabidopsis plants grown under short-day conditions. Each sample
contained tissues from seven plants with a pooled dry weight of 5 to 10
mg, and six replicates for each genotype were analyzed. To inhibit
lipolytic activities, tissues were transferred immediately into 3 mL of
isopropanol with 0.01% butylated hydroxytoluene at 758C and extracted
several times with chloroform/methanol as a recommended by Kansas
Lipidomics Research Center (http://www.k-state.edu/lipid/lipidomics/
leaf-extraction.html). Automated ESI-MS/MS analysis was performed in
theKansasLipidomics ResearchCenter Analytical Laboratory essentially
as described previously (Welti et al., 2002; Devaiah et al., 2006).
Characterization of Cuticle
The following experiments were employed to characterize the properties
of the cuticular barrier and were performed as described previously: TB
staining and evaluation of chlorophyll leaching from rosette leaves into
ethanol (Kurdyukov et al., 2006b), fatty acid composition analysis of wax,
and residual bound lipids in leaves (Franke et al., 2005; Kurdyukov et al.,
The Arabidopsis Genome Initiative locus identifiers for the genes char-
acterized in this study are as follows: FDH (At2g26250), FAE1
(At4g34520), and GL2 (At1g79840).
The following materials are available in the online version of this article.
Supplemental Figure 1. Localization of the GL2-GFP Protein Fusion
in Trichome Nuclei in FDH:FAE1 Plants.
Supplemental Figure 2. Aniline Blue Staining for Callose.
Supplemental Figure 3. Toluidine Blue Staining for Cuticular Defects
in FDH:FAE1 Trichomes.
Supplemental Figure 4. Effect of Aristolochic Acid Treatment on
VLCFAs in Acyl-CoA Esters in FDH:FAE1 Plants.
Supplemental Table 1. Summary Data Set for Compounds Tested in
the Cell Death Assay.
Supplemental Movie 1. Localization of the GL2-GFP Protein Fusion
in Trichome Nuclei in FDH:FAE1 Plants.
We thank Ljerka Kunst for the FAE1 cDNA clone, Martin Hu ¨lskamp for
bringing the cell death phenotype of the FDH:FAE1 plants to our
attention, Elmon Schmelzer for his help with fluorescence microscopy,
Ruth Welti and Mary Roth for the ESI-MS/MS identification of lipid
molecular species in our samples, Klaus Tietjen for a sample of
flufenacet, and Paul Hardy for his helpful comments and critical reading
of the manuscript. We would especially like to thank Isa Will for technical
Received January 12, 2009; revised March 9, 2009; accepted March 31,
2009; published April 17, 2009.
Adam, L., and Somerville, S.C. (1996). Genetic characterization of five
powdery mildew disease resistance loci in Arabidopsis thaliana. Plant
J. 9: 341–356.
Lipoapoptosis-Like Process in Plants 1269
Bach, L., Michaelson, L.V., Haslam, R., Bellec, Y., Gissot, L., Marion,
J., Da Costa, M., Boutin, J.P., Miquel, M., and Tellier, F. (2008). The
very-long-chain hydroxy fatty acyl-CoA dehydratase PASTICCINO2 is
essential and limiting for plant development. Proc. Natl. Acad. Sci.
USA 105: 14727–14731.
Bechtold, N., Ellis, J., and Pelletier, G. (1993). In-planta Agrobacterium-
mediated gene-transfer by infiltration of adult Arabidopsis thaliana
plants. C. R. Acad. Sci. Paris Life Sci. 316: 1194–1199.
Bessire, M., Chassot, C., Jacquat, A.C., Humphry, M., Borel, S.,
Pete ´tot, J., Me ´traux, J.P., and Nawrath, C. (2007). A permeable
cuticle in Arabidopsis leads to a strong resistance to Botrytis cinerea.
EMBO J. 26: 2158–2168.
Blacklock, B.J., and Jaworski, J.G. (2006). Substrate specificity of
Arabidopsis 3-ketoacyl-CoA synthases. Biochem. Biophys. Res.Com-
mun. 346: 583–590.
Boger, P., Matthes, B., and Schmalfuss, J. (2000). Towards the
primary target of chloroacetamides - New findings pave the way. Pest
Manag. Sci. 56: 497–508.
Bonaventure, G., Beisson, F., Ohlrogge, J., and Pollard, M. (2004).
Analysis of the aliphatic monomer composition of polyesters associ-
ated with Arabidopsis epidermis: Occurrence of octadeca-cis-6, cis-
9-diene-1,18-dioate as the major component. Plant J. 40: 920–930.
Borner, G.H.H., Sherrier, D.J., Weimar, T., Michaelson, L.V.,
Hawkins, N.D., MacAskill, A., Napier, J.A., Beale, M.H., Lilley,
K.S., and Dupree, P. (2005). Analysis of detergent-resistant mem-
branes in Arabidopsis. Evidence for plasma membrane lipid rafts.
Plant Physiol. 137: 104–116.
Busch, A.K., Gurisik, E., Cordery, D.V., Sudlow, M., Denyer, G.S.,
Laybutt, D.R., Hughes, W.E., and Biden, T.J. (2005). Increased fatty
acid desaturation and enhanced expression of stearoyl coenzyme A
desaturase protects pancreatic b-cells from lipoapoptosis. Diabetes
Chalfant, C.E., and Spiegel, S. (2005). Sphingosine 1-phosphate and
ceramide 1-phosphate: Expanding roles in cell signaling. J. Cell Sci.
Danon, A., Delorme, V., Mailhac, N., and Gallois, P. (2000). Plant
programmed cell death: A common way to die. Plant Physiol.
Biochem. 38: 647–655.
David, D., Sundarababu, S., Gerst, J.E., and Gerst, J.E. (1998).
Involvement of long chain fatty acid elongation in the trafficking of
secretory vesicles in yeast. J. Cell Biol. 143: 1167–1182.
Devaiah, S.P., Roth, M.R., Baughman, E., Li, M., Tamura, P.,
Jeannotte, R., Welti, R., and Wang, X. (2006). Quantitative profiling
of polar glycerolipid species from organs of wild-type Arabidopsis and
a PHOSPHOLIPASE Da1 knockout mutant. Phytochemistry 67: 1907–
Dhondt, S., Geoffroy, P., Stelmach, B., Legrand, M., and Heitz, T.
(2000). Soluble phospholipase A2activity is induced before oxylipin
accumulation in tobacco mosaic virus-infected tobacco leaves and is
contributed by patatin-like enzymes. Plant J. 23: 431–440.
Didenko, V.V. (2002). Detection of specific double-strand DNA breaks
and apoptosis in situ using T4 DNA ligase. Methods Mol. Biol. 203:
Didenko, V.V., Ngo, H., and Baskin, D.S. (2003). Early necrotic DNA
degradation: Presence of blunt-ended DNA breaks, 39 and 59 over-
hangs in apoptosis, but only 59 overhangs in early necrosis. Am. J.
Pathol. 162: 1571–1578.
Efremova, N., Schreiber, L., Bar, S., Heidmann, I., Huijser, P.,
Wellesen, K., Schwarz-Sommer, Z., Saedler, H., and Yephremov,
A. (2004). Functional conservation and maintenance of expression
pattern of FIDDLEHEAD-like genes in Arabidopsis and Antirrhinum.
Plant Mol. Biol. 56: 821–837.
Franke, R., Briesen, I., Wojciechowski, T., Faust, A., Yephremov, A.,
Nawrath, C., and Schreiber, L. (2005). Apoplastic polyesters in
Arabidopsis surface tissues - A typical suberin and a particular cutin.
Phytochemistry 66: 2643–2658.
Fukuda, H. (2000). Programmed cell death of tracheary elements as a
paradigm in plants. Plant Mol. Biol. 44: 245–253.
Gavrieli, Y. (1992). Identification of programmed cell death in situ
via specific labeling of nuclear DNA fragmentation. J. Cell Biol. 119:
Ghanevati, M., and Jaworski, J.G. (2002). Engineering and mechanis-
tic studies of the Arabidopsis FAE1 b-ketoacyl-CoA synthase, FAE1
KCS. Eur. J. Biochem. 269: 3531–3539.
Han, M.S., Park, S.Y., Shinzawa, K., Kim, S., Chung, K.W., Lee, J.H.,
Kwon, C.H., Lee, K.W., Park, C.K., and Chung, W.J. (2007). Lyso-
phosphatidylcholine as a death effector in lipoapoptosis of hepato-
cytes. J. Lipid Res. 49: 84–97.
Hannoufa, A., Mcnevin, J., and Lemieux, B. (1993). Epicuticular waxes
of eceriferum mutants of Arabidopsis thaliana. Phytochemistry 33:
Hardham, A.R., Jones, D.A., and Takemoto, D. (2007). Cytoskeleton
and cell wall function in penetration resistance. Curr. Opin. Plant Biol.
Hishikawa, D., Shindou, H., Kobayashi, S., Nakanishi, H., Taguchi,
R., and Shimizu, T. (2008). Discovery of a lysophospholipid acyl-
transferase family essential for membrane asymmetry and diversity.
Proc. Natl. Acad. Sci. USA 105: 2830–2835.
Hlousek-Radojcic, A., Evenson, K.J., Jaworski, J.G., and Post-
Beittenmiller, D. (1998). Fatty acid elongation is independent of
acyl-Coenzyme A synthetase activities in leek and Brassica napus.
Plant Physiol. 116: 251–258.
Hofius, D., Tsitsigiannis, D.I., Jones, J.D.G., and Mundy, J. (2007).
Inducible cell death in plant immunity. Semin. Cancer Biol. 17:
Holk, A., Rietz, S., Zahn, M., Quader, H., and Scherer, G.F.E. (2002).
Molecular identification of cytosolic, patatin-related phospholipases A
from Arabidopsis with potential functions in plant signal transduction.
Plant Physiol. 130: 90–101.
Hooker, T.S., Millar, A.A., and Kunst, L. (2002). Significance of the
expression of the CER6 condensing enzyme for cuticular wax pro-
duction in Arabidopsis. Plant Physiol. 129: 1568–1580.
Hu ¨lskamp, M., Kopczak, S.D., Horejsi, T.F., Kihl, B.K., and Pruitt, R.
E. (1995). Identification of genes required for pollen-stigma recogni-
tion in Arabidopsis thaliana. Plant J. 8: 703–714.
James, D.W., Lim, E., Keller, J., Plooy, I., Ralston, E., and Dooner, H.
K. (1995). Directed tagging of the Arabidopsis FATTY ACID ELON-
GATION1 (FAE1) gene with the maize transposon Activator. Plant Cell
Jenks, M.A., Tuttle, H.A., Eigenbrode, S.D., and Feldmann, K.A.
(1995). Leaf epicuticular waxes of the eceriferum mutants in Arabi-
dopsis. Plant Physiol. 108: 369–377.
Jones, A.M., and Dangl, J.L. (1996). Logjam at the Styx: Programmed
cell death in plants. Trends Plant Sci. 1: 114–119.
Koch, E., and Slusarenko, A. (1990). Arabidopsis is susceptible to
infection by a downy mildew fungus. Plant Cell 2: 437–445.
Kolter, T., and Sandhoff, K. (1996). Inhibitors of glycosphingolipid
biosythesis. Chem. Soc. Rev. 25: 371–381.
Kroemer, G., Petit, P., Zamzami, N., Vayssiere, J.L., and Mignotte, B.
(1995). The biochemistry of programmed cell-death. FASEB J. 9:
Kunst, L., Taylor, D.C., and Underhill, E.W. (1992). Fatty acid elonga-
tion in developing seeds of Arabidopsis thaliana. Plant Physiol.
Biochem. 30: 425–434.
Kurdyukov, S., Faust, A., Nawrath, C., Bar, S., Voisin, D., Efremova,
N., Franke, R., Schreiber, L., Saedler, H., Metraux, J.P., and
1270The Plant Cell
Yephremov, A. (2006b). The epidermis-specific extracellular BODY-
GUARD controls cuticle development and morphogenesis in Arabi-
dopsis. Plant Cell 18: 321–339.
Kurdyukov, S., Faust, A., Trenkamp, S., Bar, S., Franke, R., Efremova,
N., Tietjen, K., Schreiber, L., Saedler, H., and Yephremov, A.
(2006a). Genetic and biochemical evidence for involvement of HOT-
HEAD in the biosynthesis of long-chain a-,v-dicarboxylic fatty acids
and formation of extracellular matrix. Planta 224: 315–329.
La Camera, S., Geoffroy, P., Samaha, H., Ndiaye, A., Rahim, G.,
Legrand, M., and Heitz, T. (2005). A pathogen-inducible patatin-like
lipid acyl hydrolase facilitates fungal and bacterial host colonization in
Arabidopsis. Plant J. 44: 810–825.
Larson, T.R., Edgell, T., Byrne, J., Dehesh, K., and Graham, I.A.
(2002). Acyl CoA profiles of transgenic plants that accumulate
medium-chain fatty acids indicate inefficient storage lipid synthesis
in developing oilseeds. Plant J. 32: 519–527.
Larson, T.R., and Graham, I.A. (2001). A novel technique for the
sensitive quantification of acyl CoA esters from plant tissues. Plant J.
Liang, H., Yao, N., Song, L.T., Luo, S., Lu, H., and Greenberg, L.T.
(2003). Ceramides modulate programmed cell death in plants. Genes
Dev. 17: 2636–2641.
Listenberger, L.L., Ory, D.S., and Schaffer, J.E. (2001). Palmitate-
induced apoptosis can occur through a ceramide-independent path-
way. J. Biol. Chem. 276: 14890–14895.
Listenberger, L.L., and Schaffer, J.E. (2002). Mechanisms of lipo-
apoptosis: Implications for human heart disease. Trends Cardiovasc.
Med. 12: 134–138.
Lolle, S.J., Berlyn, G.P., Engstrom, E.M., Krolikowski, K.A., Reiter,
W.D., and Pruitt, R.E. (1997). Developmental regulation of cell
interactions in the Arabidopsis fiddlehead-1 mutant: A role for the
epidermal cell wall and cuticle. Dev. Biol. 189: 311–321.
Lolle, S.J., Cheung, A.Y., and Sussex, I.M. (1992). Fiddlehead - An
Arabidopsis mutant constitutively expressing an organ fusion program
that involves interactionsbetweenepidermal cells. Dev. Biol.152: 383–392.
McLachlan, C.S., Yin, J.L., Driussi, C., Jusuf, P.R., Hambly, B., and
McGuire, M.A. (2000). A semi-quantitative PCR method for the
detection of low levels of apoptotic DNA fragmentation in a heart
failure model. Jpn. J. Physiol. 50: 281–284.
Millar, A.A., and Kunst, L. (1997). Very-long-chain fatty acid biosyn-
thesis is controlled through the expression and specificity of the
condensing enzyme. Plant J. 12: 121–131.
Millar, A.A., Wrischer, M., and Kunst, L. (1998). Accumulation of very-
long-chain fatty acids in membrane glycerolipids is associated with
dramatic alterations in plant morphology. Plant Cell 10: 1889–1902.
Nandi, A., Welti, R., and Shah, J. (2004). The Arabidopsis thaliana
dihydroxyacetone phosphate reductase gene SUPPRESSOR OF
FATTY ACID DESATURASE DEFICIENCY1 is required for glycerolipid
metabolism and for the activation of systemic acquired resistance.
Plant Cell 16: 465–477.
Nelson, W.G., and Kastan, M.B. (1994). DNA strand breaks: The DNA
template alterations that trigger p53-dependent DNA damage re-
sponse pathways. Mol. Cell. Biol. 14: 1815–1823.
Ohashi, Y., Oka, A., Ruberti, I., Morelli, G., and Aoyama, T. (2002).
Entopically additive expression of GLABRA2 alters the frequency and
spacing of trichome initiation. Plant J. 29: 359–369.
O’Neill, C.M., Gill, S., Hobbs, D., Morgan, C., and Bancroft, I. (2003).
Natural variation for seed oil composition in Arabidopsis thaliana.
Phytochemistry 64: 1077–1090.
Park, J., Gu, Y., Lee, Y., Yang, Z., and Lee, Y. (2004). Phosphatidic
acid induces leaf cell death in Arabidopsis by activating the Rho-
related Small G Protein GTPase-mediated pathway of reactive oxygen
species generation. Plant Physiol. 134: 129–136.
Paul, S., Gable, K., Beaudoin, F., Cahoon, E., Jaworski, J., Napier,
J.A., and Dunn, T.M. (2006). Members of the Arabidopsis FAE1-like
3-ketoacyl-CoA synthase gene family substitute for the Elop proteins
of Saccharomyces cerevisiae. J. Biol. Chem. 281: 9018–9029.
Pruitt, R.E., Vielle-Calzada, J.P., Ploense, S.E., Grossniklaus, U.,
and Lolle, S.J. (2000). FIDDLEHEAD, a gene required to suppress
epidermal cell interactions in Arabidopsis, encodes a putative lipid
biosynthetic enzyme. Proc. Natl. Acad. Sci. USA 97: 1311–1316.
Raffaele, S., Vailleau, F., Leger, A., Joubes, J., Miersch, O., Huard,
C., Blee, E., Mongrand, S., Domergue, F., and Roby, D. (2008). A
MYB transcription factor regulates very-long-chain fatty acid biosyn-
thesis for activation of the hypersensitive cell death response in
Arabidopsis. Plant Cell 20: 752–767.
Riley, W.W., and Pfeiffer, D.R. (1986). The effect of Ca2+and acyl
coenzyme A: Lysophospholipid acyltransferase inhibitors on perme-
ability properties of the liver mitochondrial inner membrane. J. Biol.
Chem. 261: 14018–14024.
Rogers, H.J. (2005). Cell death and organ development in plants. Curr.
Top. Dev. Biol. 71: 225–261.
Ryu, S.B. (2004). Phospholipid-derived signaling mediated by phos-
pholipase A in plants. Trends Plant Sci. 9: 229–235.
Sayanova, O., Haslam, R., Venegas Caleron, M., and Napier, J.A.
(2007). Cloning and characterization of unusual fatty acid desaturases
from Anemone leveillei: Identification of an acyl-Coenzyme A C20 D5-
desaturase responsible for the synthesis of sciadonic acid. Plant
Physiol. 144: 455–467.
Schnittger, A., Weinl, C., Bouyer, D., Schobinger, U., and Hu ¨lskamp,
M. (2003). Misexpression of the cyclin-dependent kinase inhibitor
ICK1/KRP1 in single-celled Arabidopsis trichomes reduces endore-
duplication and cell size and induces cell death. Plant Cell 15:
Schraudner, M., Moeder, W., Wiese, C., Camp, W.V., Inze, D.,
Langebartels, C., and Sandermann, H. (1998). Ozone-induced ox-
idative burst in the ozone biomonitor plant, tobacco Bel W 3. Plant J.
Shah, J. (2005). Lipids, lipases, and lipid-modifying enzymes in plant
disease resistance. Annu. Rev. Phytopathol. 43: 229–260.
Shah, J., Kachroo, P., Nandi, A., and Klessig, D.F. (2001). A recessive
mutation in the Arabidopsis SSI2 gene confers SA- and NPR1-
independent expression of PR genes and resistance against bacterial
and oomycete pathogens. Plant J. 25: 563–574.
Shi, L., Bielawski, J., Mu, J., Dong, H., Teng, C., Zhang, J., Yang, X.,
Tomishige, N., Hanada, K., and Hannun, Y.A. (2007). Involvement of
sphingoid bases in mediating reactive oxygen intermediate produc-
tion and programmed cell death in Arabidopsis. Cell Res. 17: 1030–
Spassieva, S.D., Markham, J.E., and Hille, J. (2002). The plant disease
resistance gene Asc-1 prevents disruption of sphingolipid metabolism
during AAL-toxin-induced programmed cell death. Plant J. 32:
Sperling, P., Franke, S., Luthje, S., and Heinz, E. (2005). Are gluco-
cerebrosides the predominant sphingolipids in plant plasma mem-
branes? Plant Physiol. Biochem. 43: 1031–1038.
Sta ˚hl, U., Sta ˚lberg, K., Stymne, S., and Ronne, H. (2007). A family of
eukaryotic lysophospholipid acyltransferases with broad specificity.
FEBS Lett. 582: 305–309.
Staiger, K., Staiger, H., Weigert, C., Haas, C., Haring, H.-U., and
Kellerer, M. (2006). Saturated, but not unsaturated, fatty acids induce
apoptosis of human coronary artery endothelial cells via nuclear
factor-kB activation. Diabetes 55: 3121–3126.
Staley, K., Blaschke, A.J., and Chun, J. (1997). Apoptotic DNA
fragmentation is detected by a semiquantitative ligation-mediated
PCR of blunt DNA ends. Cell Death Differ. 4: 66–75.
Lipoapoptosis-Like Process in Plants 1271
Szymanski, D.B., Jilk, R.A., Pollock, S.M., and Marks, M.D. (1998). Download full-text
Control of GL2 expression in Arabidopsis leaves and trichomes.
Development 125: 1161–1171.
Tamura, M., Ajayi, T., Allmond, L.R., Moriyama, K., Wiener-Kronish,
J.P., and Sawa, T. (2004). Lysophospholipase A activity of Pseudo-
monas aeruginosa type III secretory toxin ExoU. Biochem. Biophys.
Res. Commun. 316: 323–331.
Trenkamp, S., Martin, W., and Tietjen, K. (2004). Specific and differ-
ential inhibition of very-long-chain fatty acid elongases from Arabi-
dopsis thaliana by different herbicides. Proc. Natl. Acad. Sci. USA
Turner, S., Gallois, P., and Brown, D. (2007). Tracheary element
differentiation. Annu. Rev. Plant Biol. 58: 407–433.
Vailleau, F., Daniel, X., Tronchet, M., Montillet, J.-L., Triantaphylide `s, C.,
and Roby, D. (2002). A R2R3-MYB gene, AtMYB30, acts as a positive
regulator of the hypersensitive cell death program in plants in response to
pathogen attack. Proc. Natl. Acad. Sci. USA 99: 10179–10184.
Van Breusegem, F., and Dat, J.F. (2006). Reactive oxygen species in
plant cell death. Plant Physiol. 141: 384–390.
van Doorn, W.G., and Woltering, E.J. (2005). Many ways to exit? Cell
death categories in plants. Trends Plant Sci. 10: 117–122.
Walker, P.R., Carson, C., Leblanc, J., and Sikorska, M. (2002).
Labeling DNA damage with terminal transferase: Applicability, spec-
ificity, and limitations. Methods Mol. Biol. 203: 3–19.
Wang, W., et al. (2008). An inositolphosphorylceramide synthase is
involved in regulation of plant programmed cell death associated with
defense in Arabidopsis. Proc. Natl. Acad. Sci. USA 20: 3163–3179.
Welti, R., Li, W., Li, M., Sang, Y., Biesiada, H., Zhou, H.-E.,
Rajashekar, C.B., Williams, T.D., and Wang, X. (2002). Profiling
membrane lipids in plant stress responses. Role of phospholipase Da
in freezing-induced changes in Arabidopsis. J. Biol. Chem. 277:
Wiermer, M., Feys, B.J., and Parker, J.E. (2005). Plant immunity: The
EDS1 regulatory node. Curr. Opin. Plant Biol. 8: 383–389.
Yamashita, A., Sugiura, T., and Waku, K. (1997). Acyltransferases and
transacylases involved in fatty acid remodeling of phospholipids
and mtabolism of bioactive lipids in mammalian cells. J. Biochem.
Yephremov, A., Wisman, E., Huijser, P., Huijser, C., Wellesen, K.,
and Saedler, H. (1999). Characterization of the FIDDLEHEAD gene of
Arabidopsis reveals a link between adhesion response and cell
differentiation in the epidermis. Plant Cell 11: 2187–2201.
1272The Plant Cell