Arabidopsis gp91phoxhomologues AtrbohD and
AtrbohF are required for accumulation of
reactive oxygen intermediates in the
plant defense response
Miguel Angel Torres*, Jeffery L. Dangl*†‡, and Jonathan D. G. Jones§
*Department of Biology and†Curriculum in Genetics and Molecular Biology, CB 3280, 108 Coker Hall, University of North Carolina, Chapel Hill, NC
27599-3280; and§The Sainsbury Laboratory, John Innes Centre, Colney, Norwich NR4 7UH, United Kingdom
Edited by Klaus Hahlbrock, Max Planck Institute for Plant Breeding Research, Cologne, Germany, and approved November 6, 2001 (received for review
August 27, 2001)
Reactive oxygen intermediates (ROI) are strongly associated with
plant defense responses. The origin of these ROI has been contro-
versial. Arabidopsis respiratory burst oxidase homologues (rboh
genes) have been proposed to play a role in ROI generation. We
analyzed lines carrying dSpm insertions in the highly expressed
observed during incompatible interactions with the bacterial patho-
gen Pseudomonas syringae pv. tomato DC3000(avrRpm1) and the
cell death, visualized by trypan blue stain and reduced electrolyte
leakage, in the Atrboh mutants after DC3000(avrRpm1) inoculation.
However, enhanced cell death is observed after infection of mutant
lines with P. parasitica. Paradoxically, although atrbohD mutation
eliminated the majority of total ROI production, atrbohF mutation
exhibited the strongest effect on cell death.
the detection of O2
incompatible interactions between resistant plants and avirulent
pathogens (1–4). During the defense response, ROI can inhibit the
pathogen by strengthening cell walls via oxidative cross-linking of
cell-wall glycoproteins (5) or by directly killing the pathogen (3).
ROI also could act as signals to induce further defenses (6–9),
including the initiation of the hypersensitive response (HR). Nitric
oxide (NO), another reactive molecule that works synergistically
with ROI in driving mammalian cell death in macrophages (10),
also has emerged as an important mediator of plant defense
response and cell death signaling in plants (11, 12).
The likely source of ROI is an NADPH oxidase, originally
postulated to be membrane-bound and to use molecular oxygen to
make superoxide, based partly on inhibition studies using diphe-
partially similar to the one present in the mammalian phagocytes
(13, 14). In activated macrophages, the respiratory burst NADPH
oxidase (RBO) is responsible for generation ROI (14). Mutations
in gp91phox, encoding the catalytic subunit of the NADPH oxidase,
in which macrophages are unable to stop the spread of infection
(15). Several plant rboh genes, homologous to gp91phox, were
identified, although they carry a 300-aa N-terminal extension
compared with the mammalian proteins (16–19). Rac homologues
in rice have been implicated in pathogen-induced cell death occur-
ring in this plant (20), and Rac is required for assembly of an active
respiratory burst NADPH oxidase in animals (14). However, no
homologues to the p47 or p67 regulators of the mammalian
findings suggest that a superoxide-generating NADPH oxidase
does exist in plants, although the plant NADPH oxidase is most
likely regulated differently than the one present in mammalian
arly production of reactive oxygen intermediates (ROI) is a
?and?or its dismutation product, H2O2, during
also have been proposed invoking cell-wall-bound peroxidases as
the main ROI source (22, 23).
Using a dSpm insertion mutagenesis system (24), we isolated
mutants in eight Atrboh genes. We demonstrate that atrbohD and
atrbohF mutations largely eliminate ROI accumulation during
disease-resistance reactions of Arabidopsis to avirulent Pseudomo-
nas syringae and Peronospora parasitica (Pp). Hence, an NADPH
oxidase is responsible for ROI accumulation during some defense
responses in Arabidopsis.
Materials and Methods
Identification of the Atrboh Mutants. Insertions in AtrbohD and F
were identified through a PCR screen on genomic DNA
extracted from pools of Col-0 plants containing dSpm transpo-
son insertions (24). Primers used include dSpm11 and dSpm1
from the transposon (24) and specific primers from each Atrboh
gene: D122 (ATGAAAATGAGACGAGGCAATTC), D92b
(GGATACTGATCATAGGCGTGGCTCCA), F171 (CTTC-
CGATATCCTTCAACCAACTC), and F172 (GAGATTGC-
Insertions in each gene were confirmed by sequencing PCR
products spanning the insertion. All of the lines identified
contained a single transposon and were derived from gluforinate
ammonium (BASTA)-resistant heterozygote parents. Plants
were grown in a chamber under a 9-h photoperiod, 24°C day
and 20°C night temperatures, 60% relative humidity, and 250
Test with Pathogens. Bacterial strains used in this study were P.
syringae pv. tomato (Pst) DC3000, DC3000(avrRpm1), and
DC3000(avrRpt2). Four-week-old plants were inoculated in a stan-
dard manner (25). atrbohD?F plants that did not display necrotic
lesions were selected before infection. Various Pp isolates, Emco5,
Cala2, Emwa1, Noco2, or Ahco1, were sprayed on 11-day-old
Cell Death Measurements. Trypan blue stain, used to visualize dying
cells, was performed as described (27). The protocol for electrolyte
leakage was adapted from Dellagi et al. (28). Four-week-old plants
were injected with bacteria in 10 mM MgCl2. Ten minutes after
injection, 7.5-mm-diameter leaf discs were collected from the
injected area and washed extensively with water for 50 min, and
This paper was submitted directly (Track II) to the PNAS office.
Pp, Peronospora parasitica; HR, hypersensitive response; ONOO?, peroxynitrite; wt, wild
type; SA, salicylic acid; SAR, systemic acquired resistance; DAB, 3,3?-diaminobenzidine.
‡To whom reprint requests should be addressed. E-mail: email@example.com.
The publication costs of this article were defrayed in part by page charge payment. This
article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C.
§1734 solely to indicate this fact.
January 8, 2002 ?
vol. 99 ?
no. 1 ?
then four discs were placed in a tube with 6 ml of water. Conduc-
from the tubes over time by using an Orion (Boston) conductivity
cm refers to the distance between electrodes. No increase in the
Detection of ROI. To visualize H2O2in situ, 3,3?-diaminobenzidine
(DAB) staining was performed on Arabidopsis seedlings sprayed
with Pp spores as described (29). For bacterial experiments with
DC3000(avrRPm1), leaves were collected 2 h after injection of the
bacteria and vacuum-infiltrated with the DAB solution. Leaves
then were placed in a plastic box under high humidity until brown
precipitate was observed (5–6 h) and then fixed with a solution of
3:1:1 ethanol?lactic acid?glycerol. Catalase effectively eliminated
the DAB stain. Application of glucose?glucose oxidase or H2O2
directly to leaves was used to verify that atrboh mutants are not
impaired in the detection of ROI by this method. Quantification of
the staining was performed with QUANTISCAN (Biosoft, Milltown,
from two different experiments. The index of staining was calcu-
lated for each leaf injected as the average of the index of brown
pixels measured in three points inside the injected area minus the
average of three points in the opposite side of each leaf.
Identification of Mutants in AtrbohD and AtrbohF. AtrbohD and
AtrbohF are the highest expressed Atrboh genes in leaves (18), the
tissue infected by our test pathogens. Thus, we focused on mutants
in these two genes to test the role of NADPH oxidase in ROI
production. We identified atrbohD and atrbohF mutants through
PCR-based screening of pooled DNA from Arabidopsis plants
containing random dSpm transposons in their genome (24). Inser-
tion D3 is located in the fifth exon of AtrbohD after codon P535
(after nucleotide T1605 from the ATG). Insertions F3 and F4 are
A530) and I221 (after nucleotide A661). F5 is in the second intron
of AtrbohF, 11 nt downstream of the splice donor. Lines homozy-
gous for the respective insertions were identified by PCR after
gluforinate ammonium selection in the next generation. All three
independent atrbohF mutant lines gave similar phenotypes with
pathogens; therefore, we used only insertion F3 for these studies.
To verify that these mutant lines are null alleles, we performed
RNA gel blot analyses and reverse transcription–PCR with specific
primers on total RNA extracted from 10-day-old seedlings. These
analyses revealed no normal full-length transcript, some aberrant
transcripts, and chimeras between the transposon and the atrboh
mRNA that presumably would give rise to nonfunctional proteins.
atrbohD and atrbohF mutants are morphologically normal, al-
though they look slightly smaller than wild type (wt) (Fig. 1B). This
phenotype is enhanced in the atrbohD?F double mutant. Four-
week-old atrbohD?F plants display some necrotic lesions and cal-
lose deposition (Fig. 1 C and D). Some of these plants stop growing
and die before setting seeds. Importantly, in the experiments
described below, we used morphologically normal plants at ages
well before this phenotype developed.
Reduction of ROI Accumulation and Diminished Cell Death Symptoms
examined the contribution of AtrbohD and AtrbohF to ROI pro-
on contact with H2O2in a reaction requiring peroxidase. Thus,
H2O2is visualized in situ as a reddish-brown precipitate (29). We
observed a strong, brown precipitate in wt plants starting 3–4 h
postinjection of the avirulent strain DC3000(avrRpm1) (Fig. 2 A
and C), in accordance with previous observations of RPM1 func-
tion, by using CeCl3staining (30). No stain was observed after
injection of virulent DC3000 or MgCl2(data not shown).
We conducted similar analyses with the atrbohD, atrbohF, and
atrbohD?F double mutants. Although there was no change in DAB
is greatly reduced in the mutant atrbohD and in the double mutant
atrbohD?F (Fig. 2 A and C). Similar results were obtained with
DC3000(avrRpt2) (which produces a delayed HR compared with
DC3000(avrRpm1); data not shown). Quantification of the DAB
stain demonstrates that atrbohD and atrbohD?F double mutants
display levels of DAB precipitate comparable to control plants
inoculated with virulent DC3000 or MgCl2(Fig. 2B). Trypan blue
staining performed in additional leaves from the same experiment
indicates that the DAB-staining procedure does not interfere with
the progression of the HR (see below). Thus, Arabidopsis AtrbohD
is required for most of the ROI observed after inoculation with
avirulent Pst, whereas AtrbohF contributes little to this oxidative
We wanted to study the requirement of Atrboh-generated ROI
correlated with the early defense response, temporally preceding
either HR or cessation of pathogen growth (31), it is unclear
whether the oxidative burst is required for HR and?or stopping
exons of the genes, and red triangles mark the dSpm transposon insertions. (B)
Representative 4-week-old rosettes. (C) Trypan blue staining of an atrbohD?F
leaf stained with aniline blue to show callose deposition. Note that infection
experiments are performed on young plants that do not exhibit ectopic cell
death. [Bar ? 1 cm (B) and 1 mm (D).]
Arabidopsis atrbohD and atrbohF mutants. (A) Schematic representa-
www.pnas.org?cgi?doi?10.1073?pnas.012452499 Torres et al.
pathogen growth. Inoculation of avirulent DC3000(avrRpm1) or
virulent DC3000 bacteria under conditions in which the former
initiates a rapid hypersensitive reaction (?2.5 ? 107cfu?ml)
showed no clear differences in trypan blue staining between wt and
atrboh mutants (Fig. 2D and data not shown). However, the atrboh
mutants displayed less trypan blue stain than the wt after injection
reflects natural infection pressure. In particular, the atrbohD?F
double mutant displayed reduced localized cell death compared
with Col-0 (Fig. 2E). Measurement of in planta growth of both
avirulent and virulent bacterial strains revealed no significant
differences between the wt and the atrboh mutants (data not
We monitored electrolyte leakage to quantify cell death during
the HR (28). Leaf discs excised from control wt leaves infiltrated
with DC3000(avrRpm1) exhibited significantly increased ion leak-
(Fig. 3A). This corresponds to a time when leaves begin to exhibit
macroscopic HR. The atrbohD?F double mutant displayed signif-
icantly lower ion leakage in repeated experiments (Fig. 3A). Mu-
mutants exhibit reproducibly greater diminution of ion leakage
than atrbohD (Fig. 3B). Collectively, these data suggest that
AtrbohD is responsible for most of the ROI produced in response
to inoculation of avirulent Pst. Conversely, AtrbohF plays a key role
in HR, particularly at low-dose inoculation. By contrast, ROI
apparently are dispensable for mediating at least the RPM1-
dependent signals that limit bacterial growth.
Atrboh Genes Modulate the Response to Pp Emco5 Infection. We
evaluated whether the atrboh mutants have an effect on the
resistance response against the oomycete parasite Pp. Pp isolate
Emco5 is fully virulent on Col-0 cotyledons, but triggers resistance,
on wt Col-0 and atrboh mutant leaves 5 h postinoculation with DC3000(avrRpm1) at 2.5 ? 107cfu?ml. (B) Quantitative analysis of DAB on leaves 6 h postinoculation
with DC3000(avrRpm1) at 2.5 ? 107cfu?ml, DC3000 (same inoculum concentration), or 10 mM MgCl2. (Bars ? SD.) (C) Detail of leaves stained with DAB 5 h
stained with trypan blue 5 h postinoculation with DC3000(avrRpm1) at 2.5 ? 107cfu?ml. (E) Detail of leaves stained with trypan blue 12 h postinoculation with lower
dose of DC3000(avrRpm1), 106cfu?ml. (Bar ? 25 ?m.) All images in C–E have the same magnification.
Torres et al.
January 8, 2002 ?
vol. 99 ?
no. 1 ?
conditioned by a single locus, in the emerging true leaves (J. M.
McDowell, S. Williams, and J.L.D., unpublished data). This ‘‘adult
resistance’’ is associated with HR trailing the growing hyphae
around the growing Emco5 hyphae in a pattern that resembles the
appearance of HR revealed by trypan blue staining (Fig. 4). Thus,
this adult resistance is weaker than the resistance mediated by the
majority of R genes directed against Pp, because the latter typically
are associated with discrete HR.
Peroxide production, but not the typical trailing necrosis, trig-
gered by Pp Emco5 is greatly reduced in atrbohD mutants (Fig. 4).
Strikingly, both DAB staining and HR are enhanced, and focused
into discrete HR lesions, in atrbohF plants (Fig. 4). Peroxide
production is eliminated in the atrbohD?F double mutant but HR
is also enhanced. The enhanced HR surrounds the emerging
of the atrbohF and atrbohD?F double mutants compared with wt
adult R function directed toward Pp Emco5, atrbohD, and atrbohF
results in separable phenotypes. The atrbohD mutation eliminates
most of the peroxide produced, whereas atrbohF actually allows
enhanced cell HR and improved resistance toward the parasite.
We also infected the Atrboh mutants with two additional aviru-
lent isolates of Pp, Emwa1 and Cala2, that are recognized by the
Col-0 RPP4 and RPP2 genes, respectively. These interactions result
in typical HR limited to a discrete group of cells (32). atrbohD and
either wt Col-0 or atrbohD?F mutant inoculated with avirulent bacteria
MgCl2. (B) Detailed differences in conductivity between Col-0 and the atrboh
mutants during the incompatible interaction DC3000(avrRpm1) at 107cfu?ml.
Each value represents the mean and SD of three replicates (compatible interac-
tion and MgCl2) or four replicates (incompatible interactions) per experiment.
The experiment was repeated three times with similar results.
Reduced electrolyte leakage in atrboh mutants after inoculation with
Pp isolate Emco5. The experiment was repeated four times; ?10 leaves per genotype per experiment analyzed. (Bar ? 50 ?m for all.)
Reduced peroxide accumulation and enhanced cell death in the atrboh mutants after inoculation with Pp isolate Emco5. (A) DAB staining of leaves 3 days
that display ?20, 11–20, 1–10, or 0 sporangiophores 7 days postinoculation with
representing more than 200 total leaves evaluated for each genotype.
Reduced sporangiophore formation in the atrboh mutants. The histo-
www.pnas.org?cgi?doi?10.1073?pnas.012452499 Torres et al.
tion in these interactions (data not shown). However, depletion of
ROI in these experiments had no effect on either cell death or
resistance. Additionally, we observed no DAB staining and no
effect on disease progression after infection with virulent Pp
isolates Noco2 and Ahco1 in either the Col-0 or the atrboh mutants
(data not shown). This demonstrates that our data are not a result
of general induction of defense response.
We used reverse genetics in Arabidopsis to define functions for two
Pp. The AtrbohD gene is required for most of the ROI observed
limited contribution. In contrast, the atrboh mutants exhibit en-
hanced HR and less sporangiophore formation in response to the
minor diminution of ROI production, it expresses strongly en-
hanced cell death phenotypes. Finally, we demonstrate that the
Pst but enhances a weak resistance response against Pp. Thus, our
most important findings are: (i) extracellular ROI production in
Arabidopsis requires Atrboh function, (ii) AtrbohD and AtrbohF
share functions in ROI generation but have separable functions in
HR regulation, and (iii) atrbohF mutants have a limited effect on
ROI production compared with atrbohD but a much greater effect
on Pp-induced cell death.
We provide genetic evidence that AtrbohD and AtrbohF, encod-
ing probable components of a plant NADPH oxidase, are respon-
sible for the ROI produced in two widely used systems for the
analysis of plant defense responses. Controversy regarding the
origin of ROI in plant defense has existed since Doke (1) first
with Phytophthora infestans. Subsequent definition of similarities to
the oxidative burst in mammalian macrophages suggested that an
NADPH oxidase was responsible for this ROI production (13).
However, H2O2, and not superoxide, is the ROI detected in most
H2O2 is the proximal burst product or that rapid dismutation
systems that superoxide is, in fact, the proximal burst product and
suggested that H2O2 is not the key signal for HR or defense-
a novel activity gel assay, recently confirmed that a putative plant
plasma membrane NADPH oxidase can produce superoxide.
Cell wall-bound peroxidases were proposed as an alternate
source for ROI–H2O2generation (36). Induced gene expression
and enzymatic activity concomitant with the burst also implicated
interactions (23). However, these studies are based on correlations
derived from gene expression and protein accumulation. Addition-
ally, some of these conclusions rest on pharmacological studies that
require careful interpretation and specificity controls. There is, to
date, no direct genetic evidence supporting peroxidase or oxalate
oxidase as sources of the ROI produced after infection.
systems and cell culture systems (2, 3), our demonstration that an
NADPH oxidase subunit is required for ROI production confirms
Doke’s original suggestion that O2
least in interactions with avirulent Pst or Pp). Thus, our results
support previous findings identifying O2
molecule (31, 33, 37). However, we failed to detect O2
tetrazolium stain or cytochrome c reduction after infection (data
not shown). In fact, we observed less nitroblue tetrazolium precip-
itate directly in leaf panels injected with avirulent bacteria than in
the surrounding areas (data not shown). This might be a result of
rapid superoxide dismutase activity that increases after pathogen
?is the first ROI produced (at
?as the key regulatory
inoculation (38, 39). This suggestion begs the question of how
monstrable extracellular superoxide dismutase activity. Alterna-
tively, the O2
independent, nonenzymatic manner. Its rate of nonenzymatic
dismutation is close to 105M?1s?1, whereas its reaction rate with
nitroblue tetrazolium is 6 ? 104M?1s?1(40).
Our analysis of the atrbohD and atrbohF mutants strongly sug-
gests that these genes act together to produce ROI (AtrbohD) and
to control cell death (AtrbohF) in response to avirulent Pst
DC3000(avrRpm1) because the double mutant HR reduction phe-
notype is stronger than atrbohF. However, HR appears not to be
required for limiting bacterial growth. Other studies indicated
separation between cell death and resistance to pathogens. For
gene resistance to bacterial pathogens although the HR appears
The HR provoked by Pst DC3000(avrRpm1) is not suppressed
completely in the atrboh mutants. Both trypan blue stain and
electrolytic leakage measurements indicate residual HR in the
atrbohD?F double mutant that might be explained by residual ROI
below our limit of detection (Fig. 2 A–C). Alternatively, other
mechanisms may contribute to the cell death induced during
interaction with avirulent bacteria. Trypan blue stain after inocu-
lation of 106cfu?ml DC3000(avrRpm1) identifies localized lesions
in the atrbohD?F double mutant, compared with more spreading
staining in the wt at each infection point (Fig. 2E). Therefore, the
proximally produced ROI may act not as the initial trigger for HR
in the directly infected cells but, rather, as a local signal for HR in
ROI may play a role in the establishment of systemic acquired
resistance (SAR), a defense system that acts in distal parts of an
infected plant (42). Salicylic acid (SA) is required for this estab-
signal that mediates SAR (45, 46). SA and ROI metabolism are
interconnected because ROI accumulation is potentiated by very
small doses of SA (6, 47) and ROI induce SA accumulation (48).
H2O2has been proposed as a systemic signal (9). However, Dorey
et al. (49) indicate that H2O2is neither necessary nor sufficient
to drive the expression of defense markers in areas surround-
ing infection sites. We induced SAR by inoculation of
DC3000(avrRpt2) and assayed distal leaves for both protection
to mount an SAR response (data not shown). Thus, even in plants
devoid of an oxidative burst, SAR still can be induced.
Surprisingly, we identified two different roles in HR control for
atrbohD?F double mutant, display less HR after inoculation with
avirulent Pst, they exhibit enhanced HR after inoculation with Pp
Emco5. This unexpected phenotype was revealed only when we
and J.L.D., unpublished data). This weak recognition does not
completely block sporulation. The atrboh mutants display stronger
cell death around the growing hyphae that prevents sporulation.
Note that this is not simply enhancement of basal defense because
Pp isolates. We suggest that HR may have a clear mechanistic
relation to resistance in interaction with the Pp, as has been
demonstrated for the interaction of powdery mildew and barley
(50). Our failure to define an effect of atrboh mutants on stronger
R-mediated HR responses could reflect the fact that, despite ROI
depletion, sufficient R signaling was generated. This concept is in
occurrence and magnitude can be correlated directly to R and Avr
?is dismutated in the absence of de-
?is dismutated rapidly in a superoxide dismutase-
Torres et al.
January 8, 2002 ?
vol. 99 ?
no. 1 ?