REPORT◥

PLANT SCIENCE

Mobile MUTE specifies subsidiarycells to build physiologicallyimproved grass stomataMichael T. Raissig,1* Juliana L. Matos,1 M. Ximena Anleu Gil,2 Ari Kornfeld,3

Akhila Bettadapur,2 Emily Abrash,1 Hannah R. Allison,1 Grayson Badgley,3

John P. Vogel,4 Joseph A. Berry,3 Dominique C. Bergmann1,2*

Plants optimize carbon assimilation while limiting water loss by adjusting stomatalaperture. In grasses, a developmental innovation—the addition of subsidiary cells (SCs)flanking two dumbbell-shaped guard cells (GCs)—is linked to improved stomatalphysiology. Here, we identify a transcription factor necessary and sufficient for SCformation in the wheat relative Brachypodium distachyon. Unexpectedly, the transcriptionfactor is an ortholog of the stomatal regulator AtMUTE, which defines GC precursor fatein Arabidopsis. The novel role of BdMUTE in specifying lateral SCs appears linked to itsacquisition of cell-to-cell mobility in Brachypodium. Physiological analyses on SC-lessplants experimentally support classic hypotheses that SCs permit greater stomatalresponsiveness and larger range of pore apertures. Manipulation of SC formation andfunction in crops, therefore, may be an effective approach to enhance plant performance.

When plants colonized land and formeda cuticle to prevent desiccation, the evo-lution of adjustable pores—stomata—inthe epidermis allowed plants to balancecarbon dioxide uptake with water loss.

Stomata in today’s dicots resemble those of ~400million years ago, consisting of two kidney-shaped guard cells (GCs) (1, 2). Grasses, however,have four-celled stomatal complexes with dumb-

bell shaped GCs flanked by two paracytic (lin-eally unrelated) subsidiary cells (SCs) (Fig. 1, Aand B) (1–4). This cellular organization may en-able fast adjustments of pore aperture at a lowenergetic cost, while allowing higher gas-exchangecapacity (1, 2, 5), and may have contributed tothe successful diversification and dispersal ofthe grass family during global aridification 30to 45 million years ago (1, 2, 6).

Stomatal formation is limited to certain cellfiles in the grass leaf epidermis (Fig. 1A, stage 1)(3, 7). In Brachypodium distachyon, a stomatal ini-tiation module—consisting of three basic helix-loop-helix (bHLH) transcription factors, BdICE1,BdSPEECHLESS1 (BdSPCH1), and BdSPCH2—guides the smaller daughters of asymmetric divi-sions toward guard mother cell (GMC) identity(Fig. 1A, stage 2) (7). The GMC then induces itslateral neighbor cells to acquire subsidiary mothercell (SMC) identity (Fig. 1A, stage 3) and to divideasymmetrically to form the SCs (Fig. 1A, stage 4).Finally, the GMC divides symmetrically (stage 5)and the complex matures (Fig. 1A, stage 6). Severalaspects of stomatal development are conserved:BdICE1 and BdSPCH1/2 are orthologs of stomatalinitiation factors in Arabidopsis (8, 9), and forma-tion of mature GCs in rice requires the stomatalmaturation bHLH OsFAMA (10). Recruitmentof lateral SCs, however, is unique to grasses, anddespite use of these cells to study polarity estab-lishment (11, 12), factors defining SMC and SCidentity are unknown. Here, we identify a locusresponsible for the formation of SCs in Brachy-podium and confirm that this morphologicalinnovation improves physiological performance.A forward genetic screen in Brachypodium tar-

geting genes responsible for the unique morphol-ogy of grass stomata yielded the subsidiary cellidentity defective (sid) mutant, which fails to re-cruit SCs and insteadproduces dicot-like two-celled

RESEARCH

Raissig et al., Science 355, 1215–1218 (2017) 17 March 2017 1 of 3

Fig. 1. BdMUTE is required for subsidiary cell formation. (A) Stomataldevelopment in Brachypodium; in a stomatal file (1) the smaller cell ofan asymmetric division (2) becomes a GMC (purple) and laterally in-duces SMC fate (yellow) (3). SMCs divide asymmetrically (4) beforeGMCs divide symmetrically (5) and the complex matures (6). (B to D)Differential interference contrast (DIC) images of the epidermis in WT(Bd21-3) (B), sid (C), and sid complemented with BdMUTEp:BdMUTE(D) [first leaf, 7 days after germination (dag)]. H, hair cells. Color key is at bottom right. (E) Stomatal density of four-celled (red), two-celled (green), aborted(blue), and three-celled (purple) complexes in WT (Bd21-3), sid, and complemented sid lines (n = 6 individuals per genotype, first leaf, 7 dag). (F) Gene model ofBdMUTE with position and nature of sid/bdmute-1 and CRISPR-Cas9–induced (bdmute-2, bdmute-3, and bdmute-4) mutations indicated. Scale bars, 10 mm.

1Department of Biology, Stanford University, 371 Serra Mall,Stanford, CA 94305, USA. 2Howard Hughes Medical Institute(HHMI), Stanford University, 371 Serra Mall, Stanford, CA 94305,USA. 3Department of Global Ecology, Carnegie Institution forScience, 260 Panama Street, Stanford, CA 94305, USA.4U.S. Department of Energy Joint Genome Institute, 2800Mitchell Drive, Walnut Creek, CA 94598, USA.*Corresponding author. Email: raissig@stanford.edu (M.T.R.);dbergmann@stanford.edu (D.C.B.)

Erratum 7 April 2017. See Erratum. on M

arch 12, 2020

http://science.sciencemag.org/

Dow

nloaded from

stomata (Fig. 1, C and E). sid plants also exhibitsomemisoriented GMC divisions and aborted GCs(31.1% ± 7.0%; n = 599) (Fig. 1, C and E) but areviable and fertile. Bulk segregant mapping com-bined with whole-genome sequencing of sid un-covered a 5–base pair (bp) deletion in BdMUTE, anortholog of Arabidopsis MUTE, which encodes abHLH transcription factor associated with GMCidentity (Fig. 1F and fig. S1) (13). Three independentclustered regularly interspaced short palindromicrepeat (CRISPR)/CRISPR-associatedprotein9 (Cas9)mutationswithin theBdMUTE gene recapitulatedthe sid phenotype (Fig. 1F and fig. S2).We complemented sidwith BdMUTEp:BdMUTE

(Fig. 1, D and E) and a fusion of BdMUTE andyellow fluorescent protein (BdMUTEp:YFP-BdMUTE)(Fig. 1E) and confirmed that BdMUTE is requiredfor SC formation and proper GMC division.BdMUTEp:YFP-BdMUTE is detected in GC andSC precursors, as well as in young GCs andSCs (Fig. 2A). Overexpressing BdMUTE inducedpolarized SMC-like divisions in epidermal pave-ment cells, resulting in multiple SC layersaround stomata and SCs around hair cells (Fig.2B), which suggested that BdMUTE is sufficientto specify SMC identity throughout the leaf.It was mysterious how BdMUTE acquired the

ability to specify SMC identity, whereas its Ara-bidopsis ortholog specifies GMCs. Hints to themechanism came from detecting YFP-BdMUTEsignal not only in GMCs and early GCs but alsoin the neighboring SMCs and early SCs (Figs. 2Aand 3A). All other described Brachypodium stoma-tal bHLH reporters are expressed only withinstomatal files (7). Either the BdMUTE promoteris active in SMCs or the BdMUTE peptide movesfrom GMCs to neighboring cell files. In supportof the latter scenario, YFP expression from a tran-scriptional reporter (BdMUTEp:3xYFPnls) was re-stricted to GMCs and GCs (Fig. 3B). Similarly,protein encoded by BdSPCH1-YFP (7) expressedwith the BdMUTE promoter was restricted toGMCs (Fig. 3C), whereas YFP-BdMUTE expressedby the BdSPCH2 promoter [active only in thestomatal lineage (7)] appeared in both GMCsand SMCs (Fig. 3D); this indicated that BdMUTEprotein is mobile. Punctate patterns of callosestaining—a marker for plasmodesmata (14)—around GMCs (Fig. 3E) suggest that secondaryplasmodesmata connecting the nonsister GMCsand SMCs are an available route of transport. AsBdMUTE is small (237 amino acids, ~25.5 kDa),either those symplastic connections allow for itspassive transport or, alternatively, domains withinBdMUTE define its mobility. We took advan-tage of evidence that the even smaller AtMUTE(202 amino acids, ~22.9 kDa) is nonmobile inArabidopsis (13) and cloned interspecific re-porter hybrids to test whether movement capacitywas an intrinsic feature of BdMUTE or whether itreflected differences between Brachypodium andArabidopsis cells. BdMUTEp:AtMUTE-YFP flu-orescence was restricted to Brachypodium GMCsand GCs, and although this version of AtMUTE-YFP is functional in Arabidopsis, it failed to res-cue the sid SC and GMC defects (Fig. 3F). In thereciprocal experiment, YFP-BdMUTE expressed

in Arabidopsis with the GMC-specific AtMUTEpromoter (13) was visible in the neighbor cells, un-like GMC-only expression of AtMUTE-YFP (Fig. 3, G

and H, and fig. S3), and BdMUTE-YFP signal inneighbors is not just a consequence of differentialprotein stability (fig. S3E). Together, our data

Raissig et al., Science 355, 1215–1218 (2017) 17 March 2017 2 of 3

Fig. 2. BdMUTE is expressed duringsubsidiary cell recruitment andoverexpression induces subsidiarycell–like divisions. (A) BdMUTEp:YFP-BdMUTE expression in sid during stomataldevelopment. YFP-BdMUTE starts to beexpressed in young GMCs and showsstrong signal in mature GMCs and weaksignal in SMCs. Expression is maintaineduntil after GMC division in both GCsand SCs and disappears duringcomplex maturation. All images fromsecond leaf, 6 to 7 dag, T1 generation.(B) Epidermal development in Ubip:YFP-BdMUTE shows ectopic SMC-like divisions before, during, andafter GMC division (arrows), resulting in stomata with multiple rows of SCs (see box in third panel) andSCs around hair cells (rightmost panel; H, hair cells). All images from about the sixth leaf, T0. Cell wallsstained with propidium iodide (PI, purple). Scale bars, 10 mm.

Fig. 3. BdMUTE is a mobile transcription factor. (A) BdMUTEp:YFP-BdMUTE signal is detected inSMCs and SCs (arrows) in addition to GMCs before and after SC recruitment. (B) BdMUTEp:3xYFPnlssignal is detected only in GMCs and GCs before and after SC recruitment. (C) BdMUTEp:BdSPCH1-YFPis detected only in GMCs and GCs before and after SC recruitment. (D) BdSPCH2p:YFP-BdMUTE isdetected in SMCs and SCs (arrows) in addition to GMCs before and after SC recruitment. (E) Aniline-blue(AB) staining of callose indicates that secondary plasmodesmata connect nonsister GMCs and SMCs.(F) BdMUTEp:AtMUTE-YFP does not rescue SCs or GMC divisions in sid (arrowhead) nor does it move.(G) AtMUTEp:AtMUTE-YFP is observed exclusively in Arabidopsis GMCs. (H) AtMUTEp:YFP-BdMUTE isobserved in GMCs and neighboring cells (asterisk) in Arabidopsis. Brachypodium images from second orthird leaves of T1 individuals, 6 to 7 dag or 11 to 12 dag, respectively, except in (F) (about the fifth leaf,T0).Arabidopsis images from T2 cotyledons, 3.5 dag. Cell walls stained with PI (purple). Scale bars, 10 mm.

RESEARCH | REPORT

Erratum 7 April 2017. See Erratum. on M

arch 12, 2020

http://science.sciencemag.org/

Dow

nloaded from

suggest that BdMUTE is indeed mobile and thatmobility is a protein-intrinsic feature of BdMUTE(fig. S4).The recruitment of lateral SCs in grass stomata

is hypothesized to permit enhanced gas exchangecapacity (gsmax) and stomatal responsiveness(dgs/dt) relative to dicots (5). SCs might serveas ion source and sink, facilitating rapid open-ing and closing, and could mechanically accom-modate and restrict GC movement or morphology(1, 2, 5). Before the identification of SC-less sid,however, it was not possible to test the contribu-tion of SCs to stomatal physiology directly. sidGCs can be forced open by the toxin fusicoccin(15) or forced closed by the hormone abscisicacid (ABA) (16), but maximal pore area is onlyhalf that of GCs from intact complexes (Fig. 4, Aand B). Similarly, gsmax of sid stomata—whenmeasured under conditions favoring maximumstomatal opening [low [CO2], high humidity,and saturating light intensity (fig. S5)] andcorrected for lower density of functional stomatain sid (fig. S6)—is also only half of wild type(WT) (sid, 0.20 ± 0.01 mol H2Om–2 s–1; WT, 0.38 ±0.03 mol H2O m–2 s–1) (Fig. 4C). Consequently,5-week-old sid plants produced less biomass(Fig. 4D). Lifetime stomatal diffusion capacitycan be assessed using the tissue–carbon isotope

ratio (d13C), because diffusion-limited leaves willincorporate a larger proportion of heavy isotopes(17). d13C in sid mutants is higher than in WT(–28.57‰ ± 0.59‰ and –32.99‰ ± 0.47‰,respectively; P = 1.67–6) confirming that thesemutants are diffusion limited (fig. S7).Finally, we assessed stomatal responsiveness

to changing light conditions in WT and sid byacclimating the plants at near-saturating light,then simulating shade, increasing light again,and finally mimicking sudden nightfall (Fig. 4E).WT stomata quickly responded to changing lightconditions, whereas both reaction times andamplitudes of the response were dampened insid (Fig. 4E). In summary, sid stomata lackingSCs cannot open as wide (gsmax) (Fig. 4C), donot close as tightly in darkness (Fig. 4E), andare slow to respond to changing light (Fig. 4E).The developmental innovation of SC-containing

stomatal complexes is accompanied by changesin functional aspects and mobility of one of thecore stomatal identity bHLH transcription fac-tors. Whereas AtMUTE halts the asymmetric stemcell–like divisions and specifies GMC identity inArabidopsis (13), BdMUTE is mobile, promotesdivisions, and defines SMC identity in Brachy-podium. Mobile transcription factors in plantsmaintain stem cell niches, pattern root hairs and

trichomes, define the root endodermis, and evenregulate flowering (18). Transcription factor trans-port, however, seems to be mostly associated withdevelopmental plasticity, where different degreesof movement of orthologs fine-tune organ size(19–21). In the stomatal context, we find species-specific mobility for BdMUTE associated with adevelopmental innovation. BdMUTE mobilitymight be an elegant mechanism to coordinatethe development of GCs and SCs to enable forma-tion and function of intimately connected four-celled complexes. Analysis of leaf-level gas exchange,stable isotopes, and biomass indicate that thepresence of SCs correlates with improved stomatalresponsiveness and greater aperture range, whichenhanceplantperformanceparticularlywhenwateris limited and/or temperatures are high. Therefore,engineering SC properties might allow for tuningof stomatal responsiveness and, thus, boost car-bon assimilation andwater use efficiency in crops.

REFERENCES AND NOTES

1. A. M. Hetherington, F. I. Woodward, Nature 424, 901–908(2003).

2. Z.-H. Chen et al., Trends Plant Sci. 22, 124–139 (2017).3. G. L. Stebbins, S. S. Shah, Dev. Biol. 2, 477–500 (1960).4. P. J. Rudall, J. Hilton, R. M. Bateman, New Phytol. 200,

598–614 (2013).5. P. J. Franks, G. D. Farquhar, Plant Physiol. 143, 78–87 (2007).6. E. A. Kellogg, Plant Physiol. 125, 1198–1205 (2001).7. M. T. Raissig, E. Abrash, A. Bettadapur, J. P. Vogel,

D. C. Bergmann, Proc. Natl. Acad. Sci. U.S.A. 113, 8326–8331(2016).

8. M. M. Kanaoka et al., Plant Cell 20, 1775–1785 (2008).9. C. A. MacAlister, K. Ohashi-Ito, D. C. Bergmann, Nature 445,

537–540 (2007).10. T. Liu, K. Ohashi-Ito, D. C. Bergmann, Development 136,

2265–2276 (2009).11. H. N. Cartwright, J. A. Humphries, L. G. Smith, Science 323,

649–651 (2009).12. M. R. Facette et al., Nature Plants 1, 14024 (2015).13. L. J. Pillitteri, D. B. Sloan, N. L. Bogenschutz, K. U. Torii, Nature

445, 501–505 (2007).14. A. Vatén et al., Dev. Cell 21, 1144–1155 (2011).15. G. R. Squire, T. A. Mansfield, New Phytol. 73, 433–440 (1974).16. R. J. Jones, T. A. Mansfield, J. Exp. Bot. 21, 714–719 (1970).17. G. Farquhar, J. Ehleringer, K. Hubick, Annu. Rev. Plant Physiol.

Plant Mol. Biol. 40, 503–537 (1989).18. K. L. Gallagher, R. Sozzani, C. M. Lee, Annu. Rev. Cell Dev. Biol.

30, 207–233 (2014).19. W. J. Lucas et al., Science 270, 1980–1983 (1995).20. J. A. Long, E. I. Moan, J. I. Medford, M. K. Barton, Nature 379,

66–69 (1996).21. S. Wu et al., Proc. Natl. Acad. Sci. U.S.A. 111, 16184–16189 (2014).

ACKNOWLEDGMENTS

This work is supported by the Swiss National ScienceFoundation (P2ZHP3_151598 to M.T.R.) and through a grantfrom The Gordon and Betty Moore Foundation (GMBF2550.05) tothe Life Science Research Foundation (to M.T.R.). The workconducted by the U.S. Department of Energy (DOE) JointGenome Institute is supported by the Office of Science of theDOE under contract no. DE-AC02-05CH1123. E.A. was aNSF graduate research fellow and D.C.B. is a GBMF Investigatorof the HHMI. Supplement contains additional data.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/355/6330/1215/suppl/DC1Materials and MethodsFigs. S1 to S7Table S1References (22–32)

3 November 2016; accepted 16 February 201710.1126/science.aal3254

Raissig et al., Science 355, 1215–1218 (2017) 17 March 2017 3 of 3

Fig. 4. Subsidiary cells enhance stomatal physiology. (A) WT and SC-less sid stomata close and openupon incubation with 50 mM ABA and 4 mM fusicoccin, respectively. Cell walls stained with PI. (B) Stomatalaperture (mm2) is significantly reduced in SC-less complexes (sid) upon fusicoccin treatment (n = 84 for WT;n = 123 for sid). (C) Maximum stomatal conductance (gsmax) is significantly reduced in sid (n = 8 pergenotype, corrected for reduced density of functional stomata in sid). (D) Fresh weight of 5-week-old sidplants is reduced compared with WT (n = 10 per genotype). (E) Stomatal conductance (gs) in response tochanging light conditions [acclimation at 1200 units of photosynthetically active radiation (PAR), dropped to300 PAR, increased to 1200 PAR, dropped to 0 PAR]. The magnitude and speed of response was sig-nificantly greater in WT (left side, black dots) compared with sid individuals (right side, red dots; correctedfor reduced density of functional stomata). Vertical dotted lines indicate light intensity changes. Each pointequals the mean value of five independent experiments; error bars represent SEM.

RESEARCH | REPORT

Erratum 7 April 2017. See Erratum. on M

arch 12, 2020

http://science.sciencemag.org/

Dow

nloaded from

Mobile MUTE specifies subsidiary cells to build physiologically improved grass stomata

Grayson Badgley, John P. Vogel, Joseph A. Berry and Dominique C. BergmannMichael T. Raissig, Juliana L. Matos, M. Ximena Anleu Gil, Ari Kornfeld, Akhila Bettadapur, Emily Abrash, Hannah R. Allison,

DOI: 10.1126/science.aal3254 (6330), 1215-1218.355Science 

, this issue p. 1215Sciencestomatal subsidiary cells and grew poorly.

whose MUTE protein could not move between cells lackedBrachypodiumthem to become subsidiary cells. Mutant . The extension in the grass protein promotes its movement into adjacent cells, promptingArabidopsisbroad-leaved plant

is a little bigger than the equivalent protein in the modelBrachypodiumtranscription factor in the wheat-like grass found that the MUTEet al.stomata on broad-leaved plants, which are made up only of two guard cells. Raissig

Stomata on grasses are made up of two guard cells and two subsidiary cells, and they perform better thanMaking more of your stomata

ARTICLE TOOLS http://science.sciencemag.org/content/355/6330/1215

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2017/03/16/355.6330.1215.DC1

CONTENTRELATED

http://science.sciencemag.org/content/sci/356/6333/eaan3164.fullhttp://stke.sciencemag.org/content/sigtrans/4/201/re4.fullhttp://stke.sciencemag.org/content/sigtrans/8/392/ra89.fullhttp://stke.sciencemag.org/content/sigtrans/6/280/ra48.fullhttp://stke.sciencemag.org/content/sigtrans/7/333/ra65.fullhttp://stke.sciencemag.org/content/sigtrans/4/173/ra32.fullhttp://stke.sciencemag.org/content/sigtrans/7/342/ra86.fullhttp://science.sciencemag.org/content/sci/355/6330/1110.full

REFERENCES

http://science.sciencemag.org/content/355/6330/1215#BIBLThis article cites 32 articles, 10 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Copyright © 2017, American Association for the Advancement of Science

on March 12, 2020

http://science.sciencem

ag.org/D

ownloaded from