Termination of asymmetric cell division
and differentiation of stomata
Lynn Jo Pillitteri1, Daniel B. Sloan1, Naomi L. Bogenschutz1& Keiko U. Torii1,2,3
Stomata consist of a pair of guard cells that mediate gas and water-vapour exchange between plants and the atmosphere.
Stomatal precursor cells—meristemoids—possess a transient stem-cell-like property and undergo several rounds of
asymmetric divisions before further differentiation. Here we report that the Arabidopsis thaliana basic helix–loop–helix
(bHLH) protein MUTE is a key switch for meristemoid fate transition. In the absence of MUTE, meristemoids abort after
excessive asymmetric divisions and fail to differentiate stomata. Constitutive overexpression of MUTE directs the entire
epidermis to adopt guard cell identity. MUTE has two paralogues: FAMA, a regulator of guard cell morphogenesis, and
SPEECHLESS (SPCH). We show that SPCH directs the first asymmetric division that initiates stomatal lineage. Together,
SPCH, MUTE and FAMA bHLH proteins control stomatal development at three consecutive steps: initiation, meristemoid
differentiation and guard cell morphogenesis. Our findings highlight the roles of closely related bHLHs in cell type
differentiation in plants and animals.
A key developmental innovation of land plants was the evolution
of specialized epidermal structures called stomata. In Arabidopsis, a
stomatal lineage arises from an undifferentiated protodermal cell
called a meristemoid mother cell (MMC), which divides asymmet-
rically to give rise to two daughter cells with distinct identities: a
pavement cell and a meristemoid. The meristemoid undergoes sev-
eral rounds of asymmetric division before differentiating into a
round, guard mother cell (GMC). The GMC divides symmetrically
to generate a pair of guard cells that surround a microscopic pore1.
factors as well as developmental programmes2,3. Recent studies have
current model suggests that the subtilisin-like protease STOMATAL
DENSITY AND DISTRIBUTION 1 (SDD1) generates a cell–cell sig-
nal that is interpreted by the transmembrane receptor TOO MANY
MOUTHS (TMM) and three ERECTA family receptor-like kinases
ERECTA, ERECTA-LIKE 1 (ERL1) and ERL2. This signalling path-
way is proposed to be mediated viamitogen-activated protein kinase
(MAP kinase) cascades4–8. Loss-of-function mutations in these loci
disrupt epidermal patterning with stomata produced adjacent to
each other or in clusters. In addition, two putative transcription
factors FAMA and FOUR LIPS (FLP) regulate the final differenti-
ation of guard cells: their loss-of-function mutations result in re-
iterative symmetric divisions of GMCs7,9.
Although our understanding of genes regulating stomatal devel-
opment has greatly advanced, an intrinsic factor that specifies differ-
exists, then the loss-of-function mutation may extend the lifespan of
meristemoids, such that mutant plants should exhibit excessive
rounds of asymmetric division but no GMC differentiation. On the
basis ofthis hypothesis, we conducted asensitized geneticscreen and
Key switch for meristemoid transition
of stomata. mute plants are small, pale and sterile, presumably owing
to severe defects in gas exchange and transpiration (Fig. 1a). Wild-
type meristemoids undergo 1–3 rounds (mean52.2360.13) of
asymmetric division before differentiating into a GMC (Fig. 1b, d)1.
In contrast, mute meristemoids undergo 3–6 rounds (mean54.466
an extended period of replication (Fig. 1c, d). The inward-spiral nat-
ureofthereiterative asymmetric divisions createdarosette patternin
the mute epidermis with an arrested, small triangular meristemoid at
the centre (Fig. 1).
mark stomatal lineage cells with the highest expression in meristem-
oids (Fig. 1e, f)6,8. The arrested mute meristemoids expressed high
levels of TMM::TMM–GFP, a translational fusion of TMM with a
driven by the ERL1 promoter6,8(Fig. 1i and j, respectively). In addi-
tion, the rosette of cells surrounding the arrested meristemoid exhib-
itedfaintexpressionofthese reporters(Fig.1e,f,i,j), suggestingtheir
identity as stomatal lineage ground cells (SLGCs)—larger daughter
cells derived from the reiterative asymmetric divisions of a single
To investigate whether arrested meristemoids progress into
GMCs, we further examined the expression of FAMA::FAMA–GFP,
a translational fusion of GFP and FAMA (Supplementary Methods).
in the nuclei of GMCs and early, immature guard cells, but not
in meristemoids (Fig. 1g). In mute, no FAMA–GFP expression was
detected (Fig. 1k). Likewise, the arrested meristemoids in mute did
controls differentiation of meristemoids to GMCs.
To place MUTE within the context of genetic pathways for stom-
atal patterning and differentiation, we next investigated genetic
1Department of Biology and2Institute for Stem Cell and Regenerative Medicine, University of Washington, Seattle, Washington 98195, USA.3CREST, Japan Science and Technology
Agency, Saitama, 332-0012, Japan.
Vol 445|1 February 2007|doi:10.1038/nature05467
interactions of MUTE with known regulators of stomatal devel-
opment. First, double and quadruple mutants were produced
between mute and cell–cell signalling mutants tmm, sdd1 and erecta;
erl1; erl2 (Fig. 2a–f). The leaf epidermis of tmm or erecta; erl1; erl2
mutants exhibits clusters of adjacent stomata (Fig. 2a, e), whereas
that of sdd1 shows increased overall stomatal density with a few
stomatal clusters (Fig. 2c). ‘Bubble-like’ clusters of small cells were
observed in the epidermis of mute; tmm and mute; erecta; erl1; erl2
high levels, indicating that they are stomatal lineages, most probably
meristemoids and SLGCs (Fig. 2b inset; Supplementary Fig. 1). The
cells in these clusters are oriented in random angles, indicating the
Figure 1 | MUTE is required for meristemoid differentiation. a, mute
phenotype. Unlike the wild type (WT), the mute plant is pale and stunted.
b, c, Abaxial leaf epidermis of WT (b) and mute (c), with meristemoids
indicated by arrowheads. Inset, tracing of a mute meristemoid with
n525 (t-test, P,0.00001). e–l, Expression of molecular markers in WT
(e–h) and mute (i–l). Arrested meristemoids in mute express high levels of
j), but does not express GMC/immature guard cell marker
FAMA::FAMA–GFP(k) andmature guard cell markerE994(l). For e, g–i,k,
bars, 20 mm.
erecta; erl1; erl2
erecta; erl1; erl2
Figure 2 | Genetic interactions between MUTE and known regulators of
stomatalpatterning anddifferentiation. a–l,Abaxialleaf epidermisoftmm
erl1; erl2 (f), erecta; erl2 (g), mute; erecta; erl2 (h), flp (i), mute; flp (j), fama
(k) andmute; fama(l).Solid brackets indicateclustersofstomata (a, c, e) or
meristemoids (b, d, f). Dashed brackets indicate stacks of symmetrically
divided GMC or guard cells (i, k). The bubble-like clusters of cells in mute;
tmm exhibit high expression of ERL1::GUS (b, inset). For the phenotypes of
wild type and mute, see Fig. 1b, c. For expression of ERL1::GUS and
TMM::GUS in wild type, mute and tmm; mute, see Supplementary Fig. 1.
Images are taken under the same magnification. Scale bar, 20 mm.
NATURE|Vol 445|1 February 2007
occurrence of randomized asymmetric division. This is consistent
with the roles of TMM and the three ERECTA-family genes in regu-
lating the frequency and orientation of asymmetric division in sto-
matal lineages8,11. The epidermis of mute; sdd1 has an increased
density of arrested meristemoids with two meristemoids adjacent to
each other observed occasionally (Fig. 2d). Overall, these results sug-
gest that MUTE and the cell–cell signalling genes interact additively,
whereas MUTE is epistatic with regard to guard cell differentiation.
erecta; erl2 double mutants exhibit an increased frequency of
asymmetric divisions with occasional patches of SLGCs that do not
include stomata (Fig. 2g). The phenotype is due to inhibitory effects
of ERL1 in the binary fate decision of a meristemoid to be a GMC
rather than an SLGC8. Interestingly, the erecta; erl2 double mutation
increased the number of SLGCs surrounding the single, arrested
mute meristemoid, resulting in a striking, ‘large rosette’ phenotype
(Fig. 2h). Therefore, ERL1 may enhance the longevity of meristem-
oids in the mute background.
In flp and fama, differentiation of a meristemoid to a GMC occurs
normally. However, subsequent guard cell differentiation is abnor-
mal; the GMC reiterates symmetric divisions and forms stacks of cell
clusters (Fig. 2i, k)7,9. The phenotypes of mute; flp and mute; fama
double mutants were indistinguishable from that of mute (Fig. 2i–l).
acting at an earlier step of stomatal differentiation.
MUTE triggers stomatal differentiation
By map-based cloning, we identified the MUTE gene as At3g06120,
which encodes the putative bHLH transcription factor bHLH45
(Supplementary Methods; Supplementary Fig. 2)12. mute possesses
a single GRA nucleotide substitution at the splice donor site of the
first intron (Fig. 3a). This abolishes the proper splicing of transcripts
and results in splicing variants, which confer premature truncations
domain is required for both DNA binding and dimerization13, and
therefore mute is most probably a null allele. Introduction of the
bHLH45 open reading frame with its native promoter into mute
plants fully restored the wild-type phenotype, confirming that
bHLH45 is MUTE (Supplementary Fig. 3).
To examine promoter activity and subcellular localization, we
used the native MUTE promoter to drive expression of the GUS
reporter (MUTE::GUS) and the translational fusion of a full-length
MUTE protein and GFP (MUTE::MUTE–GFP). The latter construct
complemented the mute phenotype, indicating that the MUTE–GFP
fusion protein is functional (Supplementary Fig. 3). MUTE::GUS
expression was highest in a subset of meristemoids that have under-
gone a few rounds of asymmetric division, with residual activity in
GMC and immature guard cells (Fig. 3d). The MUTE–GFP fusion
protein was localized in the nuclei of a subset of meristemoids that
have undergone a few rounds of asymmetric division and in recently
transitioned GMCs (Fig. 3e). No GFP signal was detected in other
epidermal cells, including newly formed meristemoids and both
immature and mature guard cells (Fig. 3e). These findings support
the hypothesis that MUTE acts as an intrinsic transcription factor
properties and trigger GMC differentiation.
We next ectopically overexpressed MUTE. CaMV35S::MUTE
plants exhibited a striking phenotype, converting all epidermal cells
into stomata (Fig. 3f). Many of these ectopic stomata expressed the
mature guard cell marker E994 (Fig. 3g). In less severe lines, a subset
bHLH domain C-terminal domain
mute v1, 2 32 + INFIFICSIPS*
mute v3 26 + WRSSFDHRRSDRVHQRVAAIGSSS*
mute v4 12 + EIKLRSSEE*
MUTE ( ) 609 bp
mute v1 1,132
Figure 3 | MUTE bHLH protein triggers stomatal differentiation. a, MUTE
gene structure. In mute, the GRA substitution occurred at the first intron
variants (see c) result in a truncation within the bHLH domain. Position of
frameshifts and the out-of-frame amino acid (aa) sequence up to the stop
codon (asterisk) are indicated. For wild-type MUTE protein sequence, see
Supplementary Fig. 4. c, PCR with reverse transcription (RT–PCR) analysis
of splicing variants. Size of each transcript indicated. d, MUTE::GUS marks a
subset of meristemoids and GMCs with residual activity in guard cells (1).
Young meristemoids (arrowhead) do not express GUS activity. e, A
functional, translational fusion protein MUTE–GFP is detected in the nuclei
of meristemoids that have undergone asymmetric divisions, but not in newly
guard cells. Insets: enlargements. f, g, Ectopic overexpression of MUTE
converts the entire epidermis to guard cells. f, Scanning electron microscope
image. g, Mature guard cell marker E994. h–j, Weak overexpression lines.
(asterisks, h), express E994 (i), and form a symmetric division plane with a
‘faux pore’ (h–j, arrowheads). Scale bars, 20 mm.
NATURE|Vol 445|1 February 2007
pavement cells contain chloroplasts, express the mature guard cell
marker E994, and produce a symmetric division plane with a ‘faux
pore’ (Fig. 3h–j). Together, these results demonstrate that MUTE is
capable of directing all protoderm cells to differentiate into guard
cellsand, depending onthestrength ofMUTE action, thesecells may
express characteristics of both pavement and guard cells.
Consecutive actions of three bHLH proteins
The genome-wide molecular phylogenetic analysis of Arabidopsis
bHLH family showed that MUTE belongs to subfamily III with
two closely related paralogues, SPEECHLESS (SPCH)/bHLH98
(At5g53210) and FAMA/bHLH97 (At3g24140)12,14–16(Supplemen-
and carboxy-terminal region. However, the MUTE protein lacks the
amino-terminal extension domain (Supplementary Fig. 4). FAMA is
expressed inGMCsandimmature guardcells(Supplementary Fig.5;
Fig. 1g), consistent with its known role in controlling the transition
from GMC to guard cells7.
To unravel the developmental function of SPCH, the closest para-
logue of MUTE, we next investigated its expression pattern as well as
loss-of-function and gain-of-function phenotypes. SPCH promoter
activity (SPCH::GUS) was observed broadly in the protoderm of leaf
function mutant exhibited an intriguing epidermal phenotype con-
sisting solely of jigsaw-puzzle-shaped pavement cells, thus lacking
sion of SPCH by the CaMV35S promoter produced a highly divided
epidermis with numerous small cells (Fig. 4e, g). These cells strongly
express TMM::TMM–GFP, indicating that they probably have the
identity of stomatal lineage cells (Fig. 4f, h). On the basis of these
results, we conclude that SPCH directs the first asymmetric division
that initiates the stomatal lineage. The spch mutant does not express
MUTE transcripts, consistent with SPCH acting before MUTE at the
initial step of stomatal development (Fig. 4c).
We have shown that MUTE is a key switch gene for meristemoids to
acquire GMC identity and that ectopic MUTE expression triggers
stomatal differentiation. Our findings highlight both similarities
and differences in the control of asymmetric division and cell type
differentiation between plants and animals. During Drosophila mel-
anogaster central nervous system development, a neuroblast divides
asymmetrically to produce another neuroblast and a differentiated
ganglion mother cell17. A transcription factor, Prospero, is unequally
MUTE protein is first detected in the nuclei of meristemoids just
before GMC differentiation. This suggests that plants are able to
monitor the age of meristemoids, possibly by tracking the rounds
of asymmetric division, and subsequently induce MUTE (Fig. 3e).
The longevity of meristemoids in mute; erecta; erl2 triple mutants
(Fig. 2h) implies that positional signals mediated by ERL1 might
intersect with this fate decision (Fig. 4i).
We and others19discovered that three closely related bHLH proteins
control three critical steps of stomatal development: initiation by
SPCH, meristemoid differentiation by MUTE, and final guard cell
differentiation by FAMA (Fig. 4i; ref. 19). SPCH triggers asymmetric
division whereas MUTE terminates asymmetrically dividing meriste-
moids. Therefore, these bHLH proteins may act in an opposing man-
ner in cell division control, but cooperate in stomatal differentiation.
The basal lineages of extant land plants, such as Selaginella sp., seem to
differentiate GMCs without reiterative asymmetric divisions of a pre-
cursor cell20,21. Diversification of this subfamily of bHLH proteins,
possibly by gene duplication, may have allowed plants to acquire com-
plex stomatal patterning as they conquered terrestrial environments.
Our discoveries emphasize the roles of bHLH proteins in cell type
determination and differentiation in plants and animals. During
myogenesis and neurogenesis in animals, a series of closely related
bHLH proteins, such as the four mammalian myogenic regula-
tory proteins, MyoD, Myf-5, myogenin and MRF4, as well as the
Drosophila achaete–scute complex, regulate sequential events in a
hierarchical fashion22–24. Muscle fibres and neurons are specialized
signals25. Similarly, proper spacing and density of stomata, which is
Figure 4 | SPCH, a paralogue of MUTE, initiates stomatal cell lineages. a,
b, SPCH::GUS expression in the first leaf primordia of a 4-day-old seedling
analysis. The T-DNA insertion spch mutant does not accumulate SPCH (1)
or MUTE (asterisk) transcripts. Note that mute produces aberrant splice
variants. ACTIN serves as a positive control. d, spch develops an epidermis
solelymade of pavementcells. e–h,Ectopic overexpression ofSPCH confers
a highly divided epidermis with small cells. e, Scanning electron microscope
image. f, These small cells express high levels of TMM::TMM–GFP.
g, h, Control WT epidermis. Scale bars, 20 mm. i, A model for consecutive
actionsofthreebHLHgenes.SPCH drivesthefirst asymmetricdivisionthat
initiates stomatal development. MUTE controls termination of asymmetric
division and differentiation of meristemoids into GMCs. FAMA regulates
differentiation of mature guard cells from GMCs. ERL1 affects the binary
decision of meristemoid versus GMC identity and hence inhibits GMC
differentiation. Red arrow, positive regulation; black T-bar, negative
NATURE|Vol 445|1 February 2007
critical for plant physiology and function, are controlled by posi-
Understanding the interplay between cell–cell signals and SPCH,
MUTE and FAMA function, as well as tracing evolutionary origins
of the three paralogous bHLH proteins, may elucidate the conser-
vation and innovation of asymmetric division and cell type differ-
entiation adopted by land plants.
a wild type. All mutants are in the Col background, except for flp-7, which is in
into Col twice. The mute mutant was backcrossed into Col three times before
analysis. T-DNA insertion alleles of FAMA (Salk_100073) and SPCH
(SAIL_36_B06) were obtained from SIGnAL (Salk Institute) and ABRC (Ohio
State University). For details, see Supplementary Methods.
meristemoid phenotype from an ethyl-methanesulphonate-treated M1popu-
lation of erecta-105; erl2. A subsequent test cross revealed that mute segregates
as a single recessivetrait:14 of 32 F1plants(43.8%)carried themute gene onthe
basis of the segregation in the next generation (x250.5, P50.48), and 207 of
851 F2plants (24.3%) showed the mute phenotype (x250.21, P50.65). F2
mapping populations were generated from a single cross of M3 mute/1;
erecta-105; erl23Ler. For details, see Supplementary Methods. A list of primer
sequences is available as Supplementary Tables 1 and 2.
study: SPCH::GUS (pLJP149), MUTE::GUS (pLJP138), FAMA::GUS (pLJP146),
MUTE–HA (pLJP158), CaMV35S::SPCH (pLJP152), and CaMV35S::MUTE
(pLJP151). See Supplementary Table 3 for a list of primer sequences used for
plasmid construction. For details, see Supplementary Methods.
Quantitative analysis of SLGCs. The number of asymmetric divisions per sto-
matal lineage between mute and wild-type plants was quantified using a
ture stoma (wild type) expressing b-glucuronidase were considered SLGCs and
counted for the analysis.
Microscopy. Detailed protocols, sample preparations, microscopy, and histo-
chemical staining for b-glucuronidase activity are available in Supplementary
Received 17 October; accepted 21 November 2006.
Published online 20 December 2006.
1. Nadeau, J. A. & Sack, F. D. Stomatal development in Arabidopsis. In The
Arabidopsis Book (eds Somerville, C. & Meyerowitz, E. M.) doi:10.1199/tab.0066
(American Society of Plant Biologists, 2002).
Hetherington, A. M. & Woodward, F. I. The role of stomata in sensing and driving
environmental change. Nature 424, 901–908 (2003).
Gray, J. E. et al. The HIC signalling pathway links CO2perception to stomatal
development. Nature 408, 713–716 (2000).
of stomatal density and distribution in Arabidopsis thaliana. Genes Dev. 14,
von Groll, U., Berger, D. & Altmann, T. The subtilisin-like serine protease SDD1
mediates cell-to-cell signaling during Arabidopsis stomatal development. Plant
Cell 14, 1527–1539 (2002).
Nadeau, J. A. & Sack, F. D. Control of stomatal distribution on the Arabidopsis leaf
surface. Science 296, 1697–1700 (2002).
Bergmann, D. C., Lukowitz, W. & Somerville, C. R. Stomatal development and
pattern controlled by a MAPKK kinase. Science 304, 1494–1497 (2004).
8. Shpak, E. D., McAbee, J. M., Pillitteri, L. J. & Torii, K. U. Stomatal patterning and
differentiation by synergistic interactions of receptor kinases. Science 309,
Lai, L.B.etal. TheArabidopsis R2R3 MYBproteins FOUR LIPSandMYB88 restrict
divisions late in the stomatal cell lineage. Plant Cell 17, 2754–2767 (2005).
10. Poethig, R. S. EnhancerTraps Æhttp://enhancertraps.bio.upenn.edu/æ (2006).
11. Geisler, M. J., Deppong, D. O., Nadeau, J. A. & Sack, F. D. Stomatal neighbor cell
polarity and division in Arabidopsis. Planta 216, 571–579 (2003).
12. Toledo-Ortiz, G., Huq, E. & Quail, P. H. The Arabidopsis basic/helix-loop-helix
transcription factor family. Plant Cell 15, 1749–1770 (2003).
13. Murre, C., McCaw, P. S. & Baltimore, D. A new DNA binding and dimerization
motifinimmunoglobulin enhancer binding, daughterless,MyoD,andmycproteins.
Cell 56, 777–783 (1989).
14. Bailey, P. C. et al. Update on the basic helix-loop-helix transcription factor gene
family in Arabidopsis thaliana. Plant Cell 15, 2497–2502 (2003).
15. Buck, M. J. & Atchley, W. R. Phylogenetic analysis of plant basic helix-loop-helix
proteins. J. Mol. Evol. 56, 742–750 (2003).
16. Heim,M.A.etal.Thebasichelix-loop-helix transcriptionfactor familyinplants:a
genome-wide study of protein structure and functional diversity. Mol. Biol. Evol.
20, 735–747 (2003).
17. Doe, C. Q. & Bowerman, B. Asymmetric cell division: fly neuroblast meets worm
zygote. Curr. Opin. Cell Biol. 13, 68–75 (2001).
18. Ikeshima-Kataoka, H., Skeath, J. B., Nabeshima, Y., Doe, C. Q. & Matsuzaki, F.
Miranda directs Prospero to a daughter cell during Drosophila asymmetric
divisions. Nature 390, 625–629 (1997).
asymmetric celldivisions thatestablishthestomatallineage.Naturedoi:10.1038/
nature05491 (this issues)
20. Dengler, N. The developmental basis of anisophylly in Selaginella martensii. II.
Histogenesis. Am. J. Bot. 70, 193–206 (1983).
21. Brown, R. & Lemmon, B. Development of stomata in Selaginella: Division polarity
and plastid movements. Am. J. Bot. 72, 1914–1925 (1985).
22. Weintraub, H. The MyoD family and myogenesis: redundancy, networks, and
thresholds. Cell 75, 1241–1244 (1993).
23. Olson, E. N. MyoD family: a paradigm for development? Genes Dev. 4, 1454–1461
24. Campuzano, S. & Modolell, J. Patterning of the Drosophila nervous system: the
achaete-scute gene complex. Trends Genet. 8, 202–208 (1992).
25. Olson, E. N. Interplay between proliferation and differentiation within the
myogenic lineage. Dev. Biol. 154, 261–272 (1992).
Supplementary Information is linked to the online version of the paper at
Acknowledgements We thank C. Doe, B. Wakimoto, J. McAbee, and T. Kakimoto
for commenting onthemanuscript; F.Sack andT.Altmann for agiftof flp,sdd1 and
TMM::TMM–GFP; S. Poethig and D. Bergmann for E994; T. Nakagawa and ABRC
for cloning vectors; SIGnAL and ABRC for T-DNA insertion lines; and P. Chan for
results and discussion about stomatal development. This work was supported in
part by the grants from DOE and NSF to K.U.T. L.J.P. was supported by the NRI
USDA/CSREES fellowship, and K.U.T. is a CREST JST investigator.
Author Contributions K.U.T. supervised the entire project. K.U.T. and L.J.P.
conceived and designed the experiments, and wrote the manuscript with
comments from co-authors. L.J.P. isolated the mute mutant and performed
characterization of mutants and transgenic plants with N.L.B. D.B.S. identified the
MUTE gene by map-based cloning.
Author Information The NCBI/GenBank accession numbers for the genes
described in this manuscript are: DQ863645 (MUTE mRNA), DQ864972 (MUTE
genomic) and DQ868373 (SPCH mRNA). Reprints and permissions information is
interests. Correspondence and requests for materials should be addressed to
NATURE|Vol 445|1 February 2007