DOI: 10.1126/science.1167324
, 1684 (2009);324Science
et al.Gabriela C. Pagnussat,
Gametophyte
FemaleArabidopsisSpecification in the
Auxin-Dependent Patterning and Gamete
This copy is for your personal, non-commercial use only.
.clicking herecolleagues, clients, or customers by
, you can order high-quality copies for yourIf you wish to distribute this article to others
.herefollowing the guidelines
can be obtained byPermission to republish or repurpose articles or portions of articles
(this information is current as of February 5, 2010 ):
The following resources related to this article are available online at www.sciencemag.org
http://www.sciencemag.org/cgi/content/full/324/5935/1684
version of this article at:
including high-resolution figures, can be found in the onlineUpdated information and services,
http://www.sciencemag.org/cgi/content/full/1167324/DC1
can be found at:Supporting Online Material
found at:
can berelated to this articleA list of selected additional articles on the Science Web sites
http://www.sciencemag.org/cgi/content/full/324/5935/1684#related-content
http://www.sciencemag.org/cgi/content/full/324/5935/1684#otherarticles
, 12 of which can be accessed for free:cites 29 articlesThis article
http://www.sciencemag.org/cgi/content/full/324/5935/1684#otherarticles
5 articles hosted by HighWire Press; see:cited byThis article has been
http://www.sciencemag.org/cgi/collection/evolution
Evolution
:subject collectionsThis article appears in the following
registered trademark of AAAS.
is aScience2009 by the American Association for the Advancement of Science; all rights reserved. The title
CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
onFebruary5,2010www.sciencemag.orgDownloadedfrom
Auxin-Dependent Patterning and
Gamete Specification in the
Arabidopsis Female Gametophyte
Gabriela C. Pagnussat,1
* Monica Alandete-Saez,1
John L. Bowman,1,2
Venkatesan Sundaresan1,3
†
The female reproductive unit of flowering plants, the haploid female gametophyte, is highly
reduced relative to other land plants. We show that patterning of the Arabidopsis female
gametophyte depends on an asymmetric distribution of the hormone auxin during its syncitial
development. Furthermore, this auxin gradient is correlated with location-specific auxin
biosynthesis, rather than auxin efflux that directs patterning in the diploid sporophytic tissues
comprising the rest of the plant. Manipulation of auxin responses or synthesis induces switching of
gametic and nongametic cell identities and specialized nonreproductive cells to exhibit attributes
presumptively lost during angiosperm evolution. These findings may account for the unique egg
cell specification characteristic of angiosperms and the formation of seeds with single diploid
embryos while containing endosperm that can have variable numbers of parental haploid genomes.
T
he highly reduced female gametophyte of
flowering plants (1, 2), with specialized
gametes to enable double fertilization, constitutes
a microcosm of pattern formation and
gamete specification for which the underlying
mechanisms have remained elusive (3–7). In
Arabidopsis and most other flowering plants, the
female gametophyte, referred to as the embryo
sac, first develops as a syncitium, a structure that
contains eight nuclei, which are later partitioned
into seven cells consisting of four cell types.
These are the egg cell and central cell that form
the gametes for double fertilization, two accessory
synergid cells positioned next to the egg cell,
and three antipodal cells. Manifestation of cell
identity is concomitant with cellularization, suggesting
that cell fates are programmed before this
stage. Moreover, it has been shown that unusual
positioning of the nuclei during embryo sac development,
as observed in the eostre mutant, can
lead to both morphological and functional defects,
including the production of an extra functional egg
cell in place of a synergid (8). These observations
suggested that, in angiosperms, the determination
of egg cell fate may depend on a location-specific
mechanism within the syncitial female gametophyte.
Additional support for this conjecture comes
from observations of the maize indeterminate
gametophyte mutant in which extra cells are generated
that appear to assume specific cell fates
correlated with their position (9). On the basis of
these findings, we hypothesize that positional information
within the syncitium might rely on the
asymmetric distribution of a morphogenetic determinant
as described in animals such as the development
of the Drosophila embryo (10).
Auxin, auxin response, and cell specification.
The phytohormone auxin regulates cell division,
elongation, and differentiation in plants. Local
accumulation of auxin provides positional signals
for multiple developmental processes, such as establishment
of the shoot and root meristems along
the apical-basal axis, organogenesis, and vascular
differentiation (11–14). In order to determine
whether auxin acts as a positional determinant
regulating cell specification during megagametogenesis
in Arabidopsis, we followed the distribution
of auxin by DR5-driven expression in the
synthetic reporters DR5::GFP and DR5::GUS
(GFP is green fluorescent protein, and GUS is bglucuronidase)
(14, 15). Auxin signaling output
was traced to megasporogenesis, the stage at which
a signal was detectable at the distal tip of the nucellus
(fig. S1A). At gametophyte developmental
stage FG1, when a functional megaspore is visible,
the signal is strong in the nucellus, outside the
RESEARCH ARTICLE
1
Department of Plant Biology, University of California, Davis,
CA 95616, USA. 2
School of Biological Sciences, Monash University,
Melbourne, Victoria 3800, Australia. 3
Department of
Plant Sciences, University of California, Davis, CA 95616, USA.
*Present address: Instituto de Investigaciones Biológicas, Universidad
Nacional de Mar del Plata, 7600 Mar del Plata,
Argentina.
†To whom correspondence should be addressed. E-mail:
sundar@ucdavis.edu
Fig. 1. Expression of
the synthetic reporters
DR5::GFP during female
gametophyte development.
(A) Scheme showing
the developmental
stages during wild-type
female gametophyte development
in Arabidopsis.
Ant indicates antipodal
cells; Cc, central cell; Ec,
egg cell; Fg, female gametophyte;
Fm, functional
megaspore; ii, inner
integument;nu,nucellus;
oi, outer integument; Pn,
polar nuclei; and Syn, synergid.
(B), At FG1, the
signal is strongly detected
in the nucellus, outside
the embryo sac. Scale bar
indicates 25 mm, and the
developing embryo sac is delimited by a dashed line. (C) At FG3 stage, the
signal is detectable inside the embryo sac at the micropylar pole. (D) As
the female gametophyte continues to develop, a strong DR5::GFP signal
is localized to the micropylar end of the embryo sac by stage FG4. Small
black arrowheads indicate nuclei inside the developing embryo sac. (E) A
DR5::GFP activity maximum at the micropylar end of the embryo sac could
be detected up to FG5 stage. (F) At FG6 stage the distribution of the DR5::GFP
signal becomes less polarized. Signals outside the developing embryo sac in
(C), (E), and (F) correspond to vascular tissues in the sporophytic ovule, particularly
the funiculus.
A
FG1 FG3 FG4 FG5 FG6 FG7
iioi
ii
oi
ii
oi
B C D E F
oi
Syn
Ec Pn Ant Syn Ec Cc
Fg
Fg
Fm
nuii
oi
ii
Fg
Syn
Ec Cc
26 JUNE 2009 VOL 324 SCIENCE www.sciencemag.org1684
onFebruary5,2010www.sciencemag.orgDownloadedfrom
developing embryo sac (Fig. 1, A and B, and fig.
S1B). At the FG3 stage, after the first mitotic
division and when the two resulting nuclei are
separated by a small vacuole, the signal is detectable
at the micropylar pole inside the embryo
sac where one of the two nuclei is located (Fig. 1,
A and C, and fig. S1C). As the female gametophyte
developed, a strong signal was localized
to the micropylar end of the developing embryo
sac by stage FG4, after the second mitotic division,
with one pair of nuclei located at the
chalazal end of the gametophyte and the other
pair at the micropylar end (Fig. 1, A and D). The
DR5::GFP activity maximum at the micropylar
end of the female gametophyte could be detected
up to the FG5 stage, in which the third mitotic
division was completed and eight nuclei were
localized along the gametophyte (Fig. 1, A and E).
These observations suggest that an auxin maximum
is formed at the micropylar pole of the developing
embryo sac. At late stages, that is, after
cellularization, the distribution of the DR5::GFP
signal appeared less polarized than in earlier stages
and was detectable in all cells of the mature embryo
sac (stage FG6, Fig. 1, A and F).
To determine whether the auxin maximum observed
in the developing syncitial embryo sac conveys
positional information for cell specification,
we down-regulated genes involved in the auxin
response with an artificial microRNA (miRNA)
called amiR-ARFa targeting a subset of the auxin
response factor genes (ARFs), a family of transcriptional
regulator proteins that mediate responses
to auxin by using a sequence highly conserved in
ARFs from two major clades (fig. S5). Because
many ARFs have essential functions in sporophyte
development, we constructed transgenic plants carrying
a pOp/LhG4 system to restrict the expression
of amiR-ARFa to the embryo sac (16). This system
was driven by the embryo sac promoter pES1 (8),
which is uniformly expressed during early female
gametophyte development (stage FG1) to the mature
embryo sac (stage FG7). To confirm that
expression of amiR-ARFa by pES1 results in
down-regulation of the auxin response in the embryo
sac, these transgenic plants were crossed to
plants carrying the DR5::GFP auxin reporter. By
scoring for DR5::GFP expression in the unfertilized
pistils of the progeny, we showed that
GFP-expressing embryo sacs were reduced by the
predicted fraction, indicating down-regulation of
the auxin response by amiR-ARFa (fig. S2 and
table S1).
We observed that ARF down-regulation resulted
in F1 plants with defective embryo sacs at
ratios close to the expected 0.25 for co-inheritance
of the driver and miRNA transgenes (table S2).
Examination of the lines showing the greatest penetrance
of the miRNA constructs revealed that
defective embryo sacs developed to maturity but
were unfertilized. Differential interference contrast
(DIC) microscopy of mature defective embryo
sacs revealed that the three cells at the micropylar
end had identity defects (i.e., could not be recognized
as egg cells or synergids) or had more
than one cell exhibiting features characteristics of
an egg cell (i.e., nucleus toward the chalazal end of
the embryo sac and a large vacuole toward the micropylar
end of the embryo sac) (Fig. 2, A to D).
We introduced cell-type specific GUS markers
into the pES driven amiR-ARFa background to
further examine cell identity. We observed that,
although two micropylar cells were GUS positive
for the expression of a synergid-specific marker
[ET884 (6)] in wild-type embryo sacs, this marker
was not detected in abnormal embryo sacs (fig.
S2, A and B, and table S3). Concordantly, none
of the mutant embryo sacs attracted pollen tubes,
which is consistent with the loss of synergid cell
identity (fig. S3E). For the egg cell–specific marker
[ET119 (6)], we could detect expression of the
marker in all three micropylar for 10 out of 489
embryo sacs examined (where 61 embryo sacs
are expected to inherit all three transgenes; table
S3), suggesting that synergid identity had been
Ec
Ccn
A B
WT
Syn
C D
Pn
Ec
EcSyn
Ccn
R-ARFa
Syn
Ab.Ec
Ccn Syn
Ab. Syn
pES1>>amiR
E F
Ccn
Ab. Ec
Ab. Syn
>>amiR-ARFa
Ccn
EcL
Syn
Ec
G H
pES1S
pES1>>amiR-ARFaWT
Es
Es
Mc
EC:GUS
Fig. 2. When several ARF genes were down-regulated with an artificial miRNA expressed specifically in
the embryo sac identity, defects in micropylar cells were observed. Scale bars, 25 mm. (A) DIC image of a
wild-type (WT) embryo sac. Ccn, central cell nucleus; Ec, egg cell; and Syn, synergid. (B) Scheme of a WT
female gametophyte. (C) Mature embryo sac showing an abnormal egg cell (Ab. Ec) with a centrally
located nucleus. (D) Embryo sac showing unfused polar nuclei (Pn) and an abnormal synergid (Ab. Syn)
cell, with a nucleus located toward the chalazal end of the embryo sac. (E) Embryo sac exhibiting two
egglike cells. EcL, egg cell–like cell. (F) Embryo sac showing three cells morphologically similar to egg
cells. (G) Expression of an egg cell–specific marker in a WT embryo sac. The developing embryo sac (Es) is
delineated by a dashed line. (H) Expression of an egg cell–specific marker in all three micropylar cells of
an ARF–down-regulated embryo sac. Mc, micropylar cell.
www.sciencemag.org SCIENCE VOL 324 26 JUNE 2009 1685
RESEARCH ARTICLE
onFebruary5,2010www.sciencemag.orgDownloadedfrom
replaced with egg cell identity (Fig. 2, E and F).
Markers for the central cell [pMEA-GUS (6)]
and the antipodal cells [pAt1g36340-GUS (7)]
were also tested in transgenic embryo sacs; however,
GUS-staining patterns for these markers
were similar in both phenotypically wild-type
and abnormal embryo sacs (fig. S2, C and D, and
table S3). Importantly, we observed no abnormalities
in nuclear positioning at earlier stages of
embryo sac development, and the overall positioning
of nuclei at maturity was similar to that
of wild-type embryo sacs (Fig. 2 and fig. S3).
Therefore, it appears that auxin may not regulate
nuclear positioning during embryo sac development
but instead regulates cell fate specification
at cellularization.
Auxin efflux and biosynthesis in developing
female gametophytes. In order to determine the
origin of the discrete auxin maxima observed, the
expression of different auxin efflux facilitators
(PINs) was studied in the developing embryo sac.
Plants carrying fusions of PIN1, PIN2, PIN3,
PIN4, and PIN7 to GFP were used to study PIN
expression in wild-type female gametophytes.
No PIN expression was observed in the female
gametophyte, but PIN1 was expressed in the
nucellus until the FG1 stage (fig. S4, A to C). The
pattern of localization of PIN1 suggests that auxin
flux may occur in the nucellus, establishing an
auxin maximum at the distal tip during the
earliest stages of embryo sac development (fig.
S4D). Later in development, PIN1 expression
was not observed in the gametophyte or the
immediately adjacent sporophytic ovule tissues
(fig. S4E).
YUCCA (YUC) genes encode key enzymes in
auxin biosynthesis (17–19), and transgenic plants
carrying YUC::GUS constructs (17, 18) showed
that YUC1 and YUC2 expression overlapped with
the auxin response signal in the ovules during
megagametogenesis (Figs. 1 and 3). For both
genes, YUC expression appears first at the micropylar
region of the nucellus outside the embryo
sac at FG1 and then localizes to the micropylar
pole of the gametophyte from FG2 through the
later stages, while being undetectable in the nucellus
(Fig. 3). This suggests that auxin might be
synthesized initially at the nucellus and subsequently
at the micropylar pole of the embryo sac
during early embryo sac development, by at least
these YUC genes.
To distort the polarized distribution of auxin
observed in the embryo sac toward higher auxin
levels, we expressed YUC1 in the whole megagametophyte
with the Op-LhG4 transactivation
system (16). Because YUC1 expression was driven
by pES1 throughout developing embryo sacs from
FG1 to FG7 (20), we predicted that auxin would be
distributed throughout the developing embryo sac if
YUC1 substrates were present in the embryo sac.
As expected YUC1-overexpressed female gametophytes
did not establish a discernible auxin gradient,
and DR5::GFP activity was detected in the
entire female gametophyte from stages FG1 to FG5
(Fig. 4A and fig. S5). F1 plants overexpressing
YUC1 in the embryo sacs had about 25% defective
ovules as identified by embryo sacs with micropylar
cells with abnormal polarities as well as
unfused polar nuclei and collapsed embryo sacs
(Fig. 4, B and C, and table S4). Furthermore,
antipodal cells that did not degenerate were detected
in 34 of out of 268 overexpressed-YUC
ovules examined (Fig. 4J).
We examined expression of a synergid-specific
marker [ET884 (6)] normally expressed in wildtype
embryo sacs only in the synergid cells. In
YUC-overexpressing embryo sacs, the synergid
marker was expressed in multiple cells, including
those normally specified as egg cell, central cell,
and the antipodals at the chalazal pole of the
embryo sac (Fig. 4, D to F, and table S5). Equally
frequently, in the YUC-overexpressing embryo
sacs, an egg cell–specific marker [ET119 (6)] was
also detected in the antipodal cells (Fig. 4H and
table S5). In some cases this marker was exclusively
expressed at the chalazal pole of the embryo
sac, where the antipodal cells are usually
located (Fig. 4I). It may be that, in these YUC1overexpressing
embryo sacs, the cell at the position
of the WT egg cell has assumed synergid
attributes (Fig. 4, C and F).
Markers for the central cell and the antipodal
cells in YUC-overexpressing embryo sacs showed
similar GUS staining patterns in both wild-type
and YUC-overexpressing embryo sacs with the
pMEA-GUS central cell marker (6), indicating
that overall central cell identity is maintained
(table S5). Thus, the central cell seems to be less
responsive to changes in local auxin concentrations
than other female gametophytic cells, although,
as noted, some central cells did express synergid
markers. In addition, after fertilization we detected
isolated embryo sacs that lacked endosperm development
and had a morphological structure resembling
a zygote at the position of the central
cell, suggesting that the central cell might have
switched to an egg cell fate (Fig. 4L). In contrast,
the antipodal-specific marker pAt1g36340-GUS
(20) showed expression only in a small fraction
of the expected YUC-overexpressing embryo sacs
carrying the GUS reporter (~10%, table S5). These
observations were consistent with the reduction
in the characteristic cell death after maturity that
typifies antipodal cells.
We also ascertained whether the extra synergidlike
cells observed in YUC-overexpressing embryo
sacs interfered with normal pollen tube growth.
When YUC1-overexpressing pistils were pollinated
with wild-type pollen, two pollen tubes approaching
and entering the female gametophyte were
detected in 44 ovules out of 428 examined, of
which 107 were predicted to contain embryo sacs
carrying the two transgenes required to overexpress
YUC1. By comparison, only 0.5% of wild-type
ovules had more than one pollen tube enter the
female gametophyte (21). These observations
show that high auxin levels throughout the embryo
sac alters cell identities, resulting in the conversion
of chalazal cell identities to micropylar
Fig. 3. Expression of YUCCA
genes in transgenic plants carrying
YUC1::GUS and YUC2::GUS
constructs suggests a role for
auxin biosynthesis in gametophyte
development. (A to F)
Expression of YUC1 (A to C) and
YUC2 (D to F) in wild-type embryo
sacs. Scale bars, 25 mm.
(A) YUC1 expression in the nucellus
at FG1 stage. (B) YUC1
expression at the micropylar end
of the embryo sac at stage FG3.
(C) YUC1 expression at the micropylar
end of the embryo sac
at stage FG4. (D) YUC2 expression
in the nucellus at FG1 stage.
(E) YUC2 expression at the micropylar
end of the embryo sac
at stage FG3. (F) YUC2 expression
at the micropylar end of the
embryo sac at stage FG4.
A B C
Fm
Es
Es
Es
Mi
D
Mi
Ch
Ch
Ch
Mi
E F
Fm Es
Ch
ChCh
Mi
Mi
Mi
Ch
26 JUNE 2009 VOL 324 SCIENCE www.sciencemag.org1686
RESEARCH ARTICLE
onFebruary5,2010www.sciencemag.orgDownloadedfrom
cell identities. As with the ARF–down-regulation
experiments, we observed no abnormalities in
nuclear positioning during embryo development
arising from YUC1 overexpression (fig. S5), further
confirming that nuclear positions within the
syncitium are insensitive to auxin.
Auxin signaling and embryo sac development.
We then examined mutants in auxin signaling for
defects in cell specification. The tir1 afb1 afb2
afb3 quadruple mutants have mutations in four
genes of the TIR auxin receptor family and a significantly
attenuated auxin response (22). From a
total of 399 tir1-1 afb1-1 afb2-1 afb3-1 embryo
sacs examined, 82 showed defects in egg cell and
synergid morphology, whereas 26 aborted at the
1 nuclear FG1 stage. The defective embryo sacs
contained an additional egg cell–like cell replacing
one of the two synergids, suggesting a switch
of synergid to egg-cell fate (fig. S7A). We confirmed
that at least some of these egg cell–like
cells can function as gametic egg cells that form
zygotelike structures after fertilization. By using
embryonic markers introduced through the pollen,
we demonstrated that the fertilized embryo sacs
contain two embryos, together with an absence of
fertilized endosperm, indicating that the two
sperm cells have fertilized two egg cells (fig. S7,
B and C).
Models for patterning of the female gametophyte
in Arabidopsis by auxin. Thus, our results
suggest that cell-type specification arises
from the positional information conferred by the
distance of the different nuclei from a micropylar
auxin source in the syncitial embryo sac. The
early auxin flux in the nucellus, as indicated by
the location of PIN1 expression, might be involved
in establishing an initial sporophtyic auxin
source, which could then provide the trigger for
events establishing gametophyte patterning. The
trigger could be auxin itself or another patterning
factor that is generated within the nucellus at that
location. In either case, we can consider two alternative
models for the subsequent events. In
one model, the sporophytic signal leads to auxin
sources specified within the gametophyte at the
micropylar pole that result in a gradient of auxin
along the micropylar-chalazal axis (Fig. 5, A to
C). The outcome is the graded specification of
different cell types, with the highest auxin concentrations
specifying synergids, followed by egg
cells, and the cells receiving the lowest amount of
auxin developing into antipodal cells (Fig. 5D). In
this model, all cells in the embryo sac might have
the competence to differentiate into any of the different
cell types composing the embryo sac, but the
gradient of auxin directs the fate of each of the cells
along the female gametophyte according to position.
The positioning is determined by nuclear
migration, which does not appear to be auxindependent,
but may use other mechanisms (8).
This proposed function of auxin in the embryo
sac implies that it acts as a morphogen, that is, a
source-derived gradient directing a concentrationdependent
specification of different cell types, a
concept for which support has been found in
sporophytic root development (23, 24). One difference
is that the gametophytic auxin appears to
be synthesized locally and in this respect resembles
more closely the classical morphogen systems
in animals.
Alternatively, gametophytic auxin might not
be the primary instructive signal but might instead
Fig. 4. Phenotype of YUC1overexpressing
embryo sacs and
two alternative models for auxindependent
cell specification in the
female gametophyte. Scale bars,
25 mm. (A) DR5::GFP activity is
localized in the whole embryo sac
(Es) of YUC1-overexpressing embryo
sacs at FG5. (B) Polarized DR5::GFP
activity detected in a WT embryo sac
at FG5. Ccn, central cell nucleus; Syn,
synergid cells; and Syn LC, synergidlike
cell. (C) A YUC1-overexpressing
embryo sac with a synergid-like cell
at a position in which the egg cell is
usually located. (D) WT embryo sac
showing the expression of a specific
synergid marker. Ch, chalazal end of
the embryo sac, and mi, micropylar
end of the embryo sac. (E) Expression
of the synergid marker in a
YUC1-overexpressing embryo sac
shows signal in the positions of both
the synergid cells and the antipodal
cells. (F) Expression of the synergid
marker is detected in cells at the
positions of synergids, egg cell, and
central cell in a YUC1-overexpressing
embryo sac. (G) Wild-type embryo
sac showing the expression of an
egg cell (E.C)–specific marker. (H)
YUC1-overexpressing embryo sac exhibiting
expression of the egg cell
marker in a cell at the position of the
egg cell and a cell at the antipodal
position. (I) YUC1-overexpressing embryo
sac showing expression of the
egg cell marker only in a cell at the
antipodal position. (J) Mature and unfertilized YUC1-overexpressing embryo
sac from an emasculated flower showing surviving antipodal cells (S Ant) 2 days
after emasculation. (K) Two pollen tubes (Pt) enter the micropyle of a YUC1overexpressing
embryo sac and continue to grow within the embryo sac. F,
funiculus. (L) Fertilized YUC1-overexpressing embryo sac showing no endosperm
development and a structure that morphologically resembles a zygote
at the position of the central cell. EcL, egg cell–like cell; Z, zygote, and ZL,
zygotelike cell.
Ccn
Syn LC
DR5:GFP
PES1>>YUC1
Es
WTPES1>>YUC1A B C
Syn
D E F
Ch
Ch
SYN:GUS
PES1>>YUC1 PES1>>YUC1WT
Es
G H I
Syn
mi
mi
mi
Ch
Ch
pSUS
WT PES1>>YUC1PES1>>YUC1
E.C
E C
mi
mi
mi
Ch
Ch
pEC:GUC1
Ccn
S Ant
E.C
ZLZ
Pt
F
PES>>YUC
J LK
F
www.sciencemag.org SCIENCE VOL 324 26 JUNE 2009 1687
RESEARCH ARTICLE
onFebruary5,2010www.sciencemag.orgDownloadedfrom
act downstream of an undiscovered patterning
factor “X” acting at the micropylar pole, which
then induces YUC expression in the micropylar
nuclei, resulting in the observed asymmetric
auxin distribution (Fig. 5, E to H). In this model,
high auxin might promote synergid cell fate,
whereas egg-cell fate might be the consequence
of lower levels of X that then result in lower
auxin (Fig. 5H). The central cell and antipodals
could then be specified by secondary auxinindependent
mechanisms. The observation that
central cell specification appears to be less susceptible
to auxin levels than the other cells is
consistent with the latter model. However, the
relative stability of central cell fate to auxin perturbations
can also be reconciled with the auxin
as morphogen model. Because the central cell
nucleus normally results from the fusion of two
polar nuclei initially located at opposite sides of
the female gametophyte, and these nuclei are hypothesized
to be exposed to quite different auxin
concentrations that are nonetheless intermediate
between the egg apparatus and antipodals, specification
of central cells might occur within a
broader concentration range. Cells at this position
are in fact capable of assuming synergid or egg
cell attributes in high auxin (Fig. 4, F and L).
Further studies will be needed to discriminate
between these alternative models for patterning.
The maintenance of an auxin gradient within
the syncitial embryo sac is unexpected. The localization
of YUC gene expression, as well as the
auxin GFP reporter, might be explained by the
observation that the embryo sac syncitium is
partitioned into cytoplasmic domains (25, 26);
such partitioning may restrict translation of messenger
RNAs to the domain containing the nucleus
where the transcript originated. However,
auxin has a diffusion constant two orders of
magnitude greater than GFP and is likely to diffuse
rapidly even with cytoplasmic partitioning.
Thus, it is probable that other mechanisms maintain
a polarized auxin distribution, such as nonPIN
auxin carriers or through the inactivation of
auxin by conjugation or degradation at the chalazal
pole.
Implications. This study provides support for
the modular hypothesis of female gametophyte
evolution (1, 2, 27). If the seven-cell/eight-nucleate
Polygonum-type embryo sac that characterize the
majority of the angiosperms is the result of a
four-nuclei/cell module duplication, then cells in
each of the modules might have conserved the
capacity to differentiate into egg cells and synergids
despite the loss of these functions at time
of divergence of the basal angiosperm lineages
>120 million years ago (2). Our observation that
synergid and egg cell attributes can be generated
at both poles of the embryo sac is consistent with
this prediction. We hypothesize that, after the
initial duplication, the potential egg apparatus
located at the chalazal pole would be within a
field of low auxin because of its distal position
relative to the micropylar auxin source, leading to
acquisition of a default cell fate as antipodal cells
as a side effect (2). This study may also provide
insights into the variation observed among angiosperm
embryo sacs. Although the eight nuclear
Polygonum-type embryo sac predominates,
a wide variety of embryo sacs with 4 to 16 nuclei
have been documented; yet in all such cases a
unique egg cell develops and results in monozygotic
seed when fertilized, although the number
of haploid nuclei within the endosperm varies
(28, 29). This can be explained if the egg apparatus
is always specified by its proximal location to
the micropylar auxin source, which itself may be
initially specified by the sporophyte. The formation
of a single embryo in fertilized seeds, combined
with variation in maternal genotype and
ploidy of endosperm tissue, may determine fitness
by affecting both the paternal genome dosage
and the effects of imprinting that control seed
size, as well as the relatedness between the endosperm
and embryo (29).
Lastly, it is interesting to consider our results
within the framework of the ancestral function of
auxin in gametophyte patterning. Analyses of extant
land plants and paleobotanical evidence indicate
that the reduced female gametophyte of
angiosperms is derived from an ancestral, complex,
free-living gametophyte (30). Although little
is known about the genetic basis of pattern formation
of extant land plants with complex gametophytes,
pharmacological experiments implicate
that auxin is an important factor (31), despite evidence
that polar auxin transport may be restricted to
the sporophyte in several mosses studied (31, 32).
We hypothesize that localized auxin synthesis,
along with the induction of sequential sources
combined with auxin degradation/conjugation,
rather than PIN-mediated transport might have
been more important in establishing the patterning
of lower plant gametophytes. Given their
evolutionary history, it is likely that genetic programs
patterning the female gametophytes of angiosperms
were inherited from those patterning
the complex three-dimensional tissues of ancestral
gametophytes.
References and Notes
1. M. W. Frohlich, M. W. Chase, Nature 450, 1184 (2007).
2. J. H. Williams, W. E. Friedman, Plant Cell 16, S119
(2004).
3. C. A. Christensen, E. J. King, J. R. Jordan, G. N. Drews,
Sex. Plant Reprod. 10, 49 (1997).
4. C. A. Christensen, S. Subramanian, G. N. Drews, Dev. Biol.
202, 136 (1998).
5. J. M. Moore, J. P. V. Calzada, W. Gagliano, U. Grossniklaus,
Cold Spring Harbor Symp. Quant. Biol. 62, 35 (1997).
6. R. Gross-Hardt et al., PLoS Biol. 5, e47 (2007).
7. G. C. Pagnussat et al., Development 132, 603 (2005).
8. G. C. Pagnussat, H. J. Yu, V. Sundaresan, Plant Cell 19,
3578 (2007).
9. M. M. S. Evans, Plant Cell 19, 46 (2007).
10. D. Umulis, M. B. O'Connor, H. G. Othmer, Curr. Top. Dev.
Biol. 81, 65 (2008).
11. E. Benkova et al., Cell 115, 591 (2003).
12. J. Friml et al., Nature 426, 147 (2003).
13. J. Mattsson, W. Ckurshumova, T. Berleth, Plant Physiol.
131, 1327 (2003).
14. S. Sabatini et al., Cell 99, 463 (1999).
15. T. Ulmasov, J. Murfett, G. Hagen, T. J. Guilfoyle, Plant Cell
9, 1963 (1997).
16. Materials and methods are available as supporting
material on Science Online.
FG4 FG5
A B C D
FG1
E G H
X
YUC YUC
FG1
Cell fates
Syn E.C Central C. Ant. C.
F
FG5FG4FG3
Fig. 5. Models for patterning of the female gametophyte. (A to D) Auxin gradient model for patterning.
(A) At FG1 stage, the auxin source is sporophytic and is derived from the nucellus. (B) Subsequently, a
secondary auxin source is specified locally within the gametophyte at the micropylar pole. (C) Auxin
gradient with a maximum at the micropylar pole of the embryo sac. (D) Auxin concentration determines
cell fates, with the highest auxin concentrations specifying synergids (Syn), followed by egg cells (E.C), and
the lowest auxin resulting in antipodal (Ant.) cells. Central C., central cell. (E to H) Model for patterning by
a nonauxin molecule acting through auxin. (E) Nucellus generates a signal (block arrow), which might
originate from the auxin source. (F) This signal induces a factor X at the micropylar pole, which then
induces auxin synthesis by activating YUCCA gene expression in the nucleus located at that pole. (G)
Continued synthesis of auxin at micropylar pole. (H) Two of the micropylar nuclei are in the high auxin
zone and differentiate into synergids, whereas the third micropylar nucleus in lower auxin differentiates
into the egg cell. In this model, the central cell and antipodal cell fates could be specified by auxindependent
or auxin-independent mechanisms.
26 JUNE 2009 VOL 324 SCIENCE www.sciencemag.org1688
RESEARCH ARTICLE
onFebruary5,2010www.sciencemag.orgDownloadedfrom
17. Y. Cheng, X. H. Dai, Y. D. Zhao, Genes Dev. 20, 1790
(2006).
18. Y. Cheng, X. H. Dai, Y. D. Zhao, Plant Cell 19, 2430
(2007).
19. Y. Zhao et al., Science 291, 306 (2001).
20. H. J. Yu, P. Hogan, V. Sundaresan, Plant Physiol. 139,
1853 (2005).
21. N. Huck, J. M. Moore, M. Federer, U. Grossniklaus,
Development 130, 2149 (2003).
22. N. Dharmasiri et al., Dev. Cell 9, 109 (2005).
23. O. Leyser, Cell 121, 819 (2005).
24. C. Galinha et al., Nature 449, 1053 (2007).
25. R. C. Brown, B. C. Lemmon, EMSA Bull 22, 48 (1992).
26. R. C. Brown, B. C. Lemmon, Protoplasma 165, 155
(1991).
27. W. E. Friedman, J. H. Williams, Evolution 57, 216 (2003).
28. V. Brukhin, M. D. Curtis, U. Grossniklaus, Curr. Sci. 89,
1844 (2005).
29. W. E. Friedman, E. N. Madrid, J. H. Williams, Int. J. Plant
Sci. 169, 79 (2008).
30. P. Kenrick, P. R. Crane, The Origin and Early
Diversification of Land Plants: A Cladistic Study
(Smithsonian Series in Comparative Evolutionary Biology,
Smithsonian Institution, Washington, DC, 1997).
31. T. J. Cooke, D. Poli, A. E. Sztein, J. D. Cohen, Plant Mol.
Biol. 49, 319 (2002).
32. T. Fujita et al., Evol. Dev. 10, 176 (2008).
33. We thank U. Grossniklaus for providing the cell-specific
marker lines used in this study; B. Sheres, J. Friml, and
M. Tasaka for providing plants carrying PIN::GFP fusions;
Y. Zhao for providing YUC::GUS reporter lines and YUC1
cDNA; N. Dharmasiri for seeds of the tir1afb quadruple
mutant; M. Yu and A. Krishnan for technical assistance; and
G. Ruvkun, D. Hanahan, members of the Gasser, Bowman,
and Sundaresan labs, and three anonymous reviewers for
valuable suggestions. Funded by NSF grants IOS-0313501
(V.S.), IOS-0745167 (V.S.), and IOB-0515435 (J.L.B.).
Supporting Online Material
www.sciencemag.org/cgi/content/full/1167324/DC1
Materials and Methods
Figs S1 to S7
Tables S1 to S5
References
17 October 2008; accepted 28 April 2009
Published online 4 June 2009;
10.1126/science.1167324
Include this information when citing this paper.
REPORTS
Extending Universal Nodal Excitations
Optimizes Superconductivity
in Bi2Sr2CaCu2O8+d
Aakash Pushp,1,2
* Colin V. Parker,1
* Abhay N. Pasupathy,1
† Kenjiro K. Gomes,1
‡
Shimpei Ono,3
Jinsheng Wen,4
Zhijun Xu,4
Genda Gu,4
Ali Yazdani1
§
Understanding the mechanism by which d wave superconductivity in the cuprates emerges and is
optimized by doping the Mott insulator is one of the major outstanding problems in condensedmatter
physics. Our high-resolution scanning tunneling microscopy measurements of the
high–transition temperature (Tc) superconductor Bi2Sr2CaCu2O8+d show that samples with different
Tc values in the low doping regime follow a remarkably universal d wave low-energy excitation
spectrum, indicating a doping-independent nodal gap. We demonstrate that Tc instead correlates
with the fraction of the Fermi surface over which the samples exhibit the universal spectrum.
Optimal Tc is achieved when all parts of the Fermi surface follow this universal behavior. Increasing
the temperature above Tc turns the universal spectrum into an arc of gapless excitations, whereas
overdoping breaks down the universal nodal behavior.
C
entral to the current debate on the mechanism
underlying high-temperature superconductivity
is the question of whether
pairing strength in the cuprates is diminished
as these systems approach the Mott insulator
limit with reduced hole density. The panoply of
physical phenomena in lightly doped cuprates
near the Mott state uncovered over the last two
decades—from observation of the pseudogap
behavior (1, 2) to fluctuating superconductivity
(3) above the transition temperature (Tc) to the
possibility of other competing orders (4–6)—
have made addressing this question challenging.
In a simple d wave superconductor, a single
energy scale suffices to completely describe
the excitation spectrum, the associated pairing
energy gap (including its angular and temperature
dependence), as well as the Tc of the sample.
In the underdoped cuprates, however, there
is increasing evidence (7–12) that a single energy
scale is insufficient to describe the anisotropy
of the energy gap because different behavior
is seen near the node (45° to the Cu–O bond
direction) and the anti-node (along the Cu–O
bond direction). The temperature evolution of
the spectroscopic measurements has also shown
a dichotomy between nodal and anti-nodal
gaps, showing different temperature dependence
(8, 10, 13). Theoretical proposals for addressing
these phenomena include those based on phase
fluctuations of preformed pairs (14–16), incipient
order (5), breakup of the Fermi surface due to
umklapp scattering (17), and incoherence of antinodal
quasi-particles (18). Although it is clear that
the gap near the anti-node increases as one approaches
the Mott insulator (19), the behavior of
the gap near the node still remains debated, with
different measurements showing both increasing
(19–21) and decreasing (7, 8, 22) trends with underdoping.
Whether pairing gaps associated with
nodal excitations track the samples’ Tc, as expected
for simple d wave superconductors, is an unresolved
question that deeply affects our understanding
of superconductivity in the cuprates. The
answer to this question can determine if the pairing
is derived from the strong electronic correlations
of the Mott state and can identify the mechanism
by which d wave superconductivity is optimized
in proximity to an insulating ground state.
To elucidate the nature of the nodal and antinodal
gaps as a function of doping and temperature,
we performed atomically resolved scanning
tunneling microscopy (STM) measurements on
Bi2Sr2CaCu2O8+d (BSCCO) in the doping range
of 0.07 < x < 0.24 (where x is the nominal hole
doping) and a temperature range from 5 to 120 K.
Spectroscopic measurements in our home-built
scanning tunneling microscope can be performed
with sub–millielectron volt energy resolution on
the same atomic location as a function of temperature
(9, 23). Although STM spectroscopy
does not have intrinsic angular resolution, information
about the nodal gap can be obtained from
the spectrum near the Fermi energy, whereas
the anti-nodal excitations occur at higher energy.
We can use this information to extract the behavior
of the nodal gap.
The complexity of the excitation spectrum in
underdoped BSCCO samples (Fig. 1A) is seen in
lattice-tracking spectroscopy (9, 23) measurements,
in which we track the temperature evolution of
tunneling spectra at a given atomic location. This
spectrum (typical for this sample) shows a higher
energy gap, D0, (black arrow in Fig. 1A), a smaller
“kink” within the higher energy gap (red arrow in
Fig. 1A), and an overall background, all of which
are position dependent. D0, determined by the maximum
conductance on the positive side, shows
strong spatial inhomogeneity (24, 25) on the sample
(Fig. 1C). The spatial average of the higher
energy gap compares well with angle-resolved
1
Joseph Henry Laboratories and Department of Physics,
Princeton University, Princeton, NJ 08544, USA. 2
Department
of Physics, University of Illinois at Urbana-Champaign,
Urbana, IL 61801, USA. 3
Central Research Institute of Electric
Power Industry, Komae, Tokyo 201-8511, Japan. 4
Condensed
Matter Physics and Materials Science, Brookhaven National
Laboratory (BNL), Upton, NY 11973, USA.
*These authors contributed equally to this work.
†Present address: Department of Physics, Columbia University,
New York, NY 10027, USA.
‡Present address: Department of Physics, Stanford University,
Stanford, CA 94305, USA.
§To whom correspondence should be addressed. E-mail:
yazdani@princeton.edu
www.sciencemag.org SCIENCE VOL 324 26 JUNE 2009 1689
onFebruary5,2010www.sciencemag.orgDownloadedfrom