SLAC1 is required for plant guard cell S-type anion channel function in stomatal signalling

LETTERS

SLAC1 is required for plant guard cell S-type anionchannel function in stomatal signallingTriin Vahisalu1,2*, Hannes Kollist1,3*, Yong-Fei Wang4*, Noriyuki Nishimura4, Wai-Yin Chan4, Gabriel Valerio4,Airi Lamminmaki1, Mikael Brosche1, Heino Moldau3, Radhika Desikan5{, Julian I. Schroeder4 & Jaakko Kangasjarvi1

Stomatal pores, formed by two surrounding guard cells in theepidermis of plant leaves, allow influx of atmospheric carbondioxide in exchange for transpirational water loss. Stomata alsorestrict the entry of ozone — an important air pollutant that has anincreasingly negative impact on crop yields, and thus global car-bon fixation1 and climate change2. The aperture of stomatal poresis regulated by the transport of osmotically active ions and meta-bolites across guard cell membranes3,4. Despite the vital role ofguard cells in controlling plant water loss3,4, ozone sensitivity1,2

and CO2 supply2,5–7, the genes encoding some of the main regula-tors of stomatal movements remain unknown. It has been pro-posed that guard cell anion channels function as importantregulators of stomatal closure and are essential in mediating sto-matal responses to physiological and stress stimuli3,4,8. However,the genes encoding membrane proteins that mediate guard cellanion efflux have not yet been identified. Here we report themapping and characterization of an ozone-sensitive Arabidopsisthaliana mutant, slac1. We show that SLAC1 (SLOW ANIONCHANNEL-ASSOCIATED 1) is preferentially expressed in guardcells and encodes a distant homologue of fungal and bacterialdicarboxylate/malic acid transport proteins. The plasma mem-brane protein SLAC1 is essential for stomatal closure in responseto CO2, abscisic acid, ozone, light/dark transitions, humiditychange, calcium ions, hydrogen peroxide and nitric oxide.Mutations in SLAC1 impair slow (S-type) anion channel currentsthat are activated by cytosolic Ca21 and abscisic acid, but do notaffect rapid (R-type) anion channel currents or Ca21 channel func-tion. A low homology of SLAC1 to bacterial and fungal organicacid transport proteins, and the permeability of S-type anionchannels to malate9 suggest a vital role for SLAC1 in the functionof S-type anion channels.

Stomatal aperture is regulated by light, plant water status, CO2

concentration, relative air humidity, and among other stresses,drought and ozone (O3)3,4. A number of signalling compounds,including abscisic acid (ABA), reactive oxygen species (ROS), nitricoxide (NO) and Ca21 ions are involved in the regulation of stomatalaperture3,4. Adjustment of stomatal apertures is achieved by con-trolled transport of osmotically active ions and organic metabolites,including potassium (K1), chloride (Cl–) and malate across guardcell membranes3,8,10,11, resulting in changes in osmotic potential.Anion channels have been proposed to function as central regulatorsof stomatal closure8,11 by mediating anion efflux and causing mem-brane depolarization, which controls K1 efflux through K1 channels.So far, none of the candidates for plant anion channels — the planthomologues to the animal CLC chloride channels — has been

localized to the plasma membrane10, and the first plant CLC channelthat was functionally characterized encodes a central vacuolarproton/nitrate exchanger12, rather than an anion channel. Thus,despite their proposed importance in several physiological and stressresponses in plants8,10,11, the molecular identity of the guard cellplasma membrane proteins that mediate anion channel activity hasremained unknown.

In a mutant screen for O3 sensitivity, a series of Arabidopsis ethylmethanesulphonate (EMS) mutants called radical-induced cell death(rcd) was identified13,14. One of them, a recessive mutant originallyreferred to as rcd3 (ref. 14) and here renamed slac1 (slow anionchannel-associated 1), showed constitutively higher stomatal conduc-tance than the wild type (Columbia, Col-0) (Fig. 1a). Interestingly,both rapid transient15 and long-term O3-induced decreases in sto-matal conductance were abolished in slac1 (Fig. 1a). Water loss fromexcised slac1 leaves resulted in 70–80% fresh weight loss after 90 min,whereas in the wild type, fresh weight loss was only 30% after 90 min(Fig. 1b). These differences in fresh weight loss were not a result ofvariation in stomatal number because slac1 and wild-type leaves havesimilar stomatal density (Supplementary Fig. 1). Microarray analysesusing messenger RNAs from 3-week-old rosette leaves did not revealany significant differences in gene expression between slac1 and thewild type when grown under optimal conditions. Furthermore, noother phenotypic differences have been observed between slac1 andthe wild type. Together these data suggested that the defect in slac1lies in defective stomatal regulation and that the O3 damage of slac1leaves (Supplementary Fig. 2) is a result of increased O3 flux intoleaves through more open stomata.

The slac1-1 mutation was identified in the gene At1g12480 by acombination of mapping, candidate gene expression in guard cellmicroarrays, and analyses of transfer DNA (T-DNA) insertionmutants (see Supplementary Information). SLAC1 encodes a pre-dicted membrane protein of 556 amino acids with a calculatedmolecular weight of 63.2 kDa and a predicted isoelectric point of9.58. SLAC1 has hydrophilic amino- and carboxy-terminal tails(189 and 60 amino acids, respectively) and 10 predicted transmem-brane helices (Fig. 1c, Supplementary Fig. 3), which contain a C4-dicarboxylate transporter/malic acid transport protein domain(InterPro: IPR004695) defined from the Escherichia coli TehA andSchizosaccharomyces pombe Mae1 proteins. Mae1 is involved in malateuptake16. TehA and Mae1 lack the long hydrophilic tail present in theN terminus of SLAC1, but show a weak, 15–20% amino-acid identityover the transmembrane region with SLAC1 (Supplementary Fig. 4a).SLAC1 shows no homology to the aluminium-activated malate trans-porters that function in plant aluminium resistance17. Homozygous

*These authors contributed equally to this work

1Plant Biology, Department of Biological and Environmental Sciences, University of Helsinki, FI-00014 Helsinki, Finland. 2Department of Botany, Institute of Ecology and Earth Sciences,University of Tartu, Tartu 51005, Estonia. 3Institute of Technology, University of Tartu, Tartu 50411, Estonia. 4Division of Biological Sciences, Cell and Developmental Biology Section,University of California San Diego, La Jolla, California 92093-0116, USA. 5Centre for Research in Plant Science, University of the West of England, Bristol BS16 1QY, UK. {Presentaddress: Division of Biology, Imperial College London, London SW7 2AZ, UK.

Vol 452 | 27 March 2008 | doi:10.1038/nature06608

487Nature Publishing Group©2008

T-DNA insertion lines (SALK_099139 and SALK_137265, referred toas slac1-3 and slac1-4, respectively; Fig. 1c) both showed similar reces-sive inheritance, and exhibited similar fresh weight loss from excisedleaves as slac1-1 (Fig. 1b). A genomic copy of SLAC1 complementedthe mutant phenotype in stably transformed slac1-1 (Fig. 1b).

SLAC1 belongs to a small family of five proteins in Arabidopsis.Three of the proteins, including SLAC1, have a long hydrophilicN-terminal tail, whereas two have only the transmembrane domains.Rice has nine orthologous proteins. The SLAC1 protein is moresimilar to its rice orthologue Os04g48530 than to the four other

g i jf kh

dNucleus

SLAC1::GFP FM 4-64Mergef + g

Mergef + g + i

Amino acid

cS456F

1 100 200

Hyd

rop

hob

icity

300 400 500

T-DNAslac1-4

3210

–1–2–3–4–5

T-DNAslac1-3b

0

20

40

60

80

100

0 30 60 90

WTslac1-1slac1-3

slac1-4slac1-1::SLAC1

e

Sto

mat

al c

ond

ucta

nce

(mol

m–2

s–1

)

WTslac1-1

0

0.1

0.2

0.3

0.4

0.5

100806040200 120

a

Time (min) Time (min)

Fres

h w

eigh

t lo

ss (%

)

O3

slac1-1

Figure 1 | Membrane protein SLAC1 controls leaf ozone and water-lossresponses. a, Stomatal conductance (n 5 10, 6 s.e.m.) of slac1-1 and wild-type plants after onset of 200 p.p.b. ozone, indicated by arrows. b, Weight lossfrom detached leaves of wild type (WT), slac1 alleles and slac1-1complemented with the SLAC1 gene (n 5 5, 6 s.e.m.). c, Membrane spanninghydrophobic regions in SLAC1 protein and the location of mutant alleles.

d, e, GUS activity in SLAC1 promoter uidA reporter lines. f, SLAC1::GFPtranslational fusion expressed in onion epidermal cells. g, Area as inf, membranes stained with FM 4-64. h, Overlay of f and g. i, Light micrographof h. j, Overlay of h, and i. k, SLAC1::GFP translational fusion in plasmolysedonion epidermal cells renders the Hechtian strands attaching the plasmamembrane to the cell wall visible. Scale bars: d–j, 100mm; k, 50mm.

0

0.1

0.2

0.3

Time of the day (hh)

Sto

mat

al c

ond

ucta

nce

(mol

m–2

s–1

)

WTslac1-1slac1-3

03 09 15 21 03

130 00

0

0.25

0.50

0.75

1.00

0 30 60 90Time (min)

0130 130

Sto

mat

al c

ond

ucta

nce

(rel

ativ

e un

its)

Sto

mat

al c

ond

ucta

nce

(rel

ativ

e un

its)

Sto

mat

al c

ond

ucta

nce

(rel

ativ

e un

its)

0.25

0.50

0.75

1.00

0 30 60 90 120

400 800 2,000 400CO2 concentration (µl l–1)

Photosynthetically active radiation(µmol m–2 s–1)

Photosynthetically active radiation(µmol m–2 s–1)

Relative humidity (%)4574

0.25

0.50

0.75

1.00

Time (min) Time (min)0 10 20 30 40

WTslac1-1slac1-3

WTslac1-1slac1-3

WTslac1-1slac1-3

a b

c d

Figure 2 | Mutations in SLAC1 impair stomatal responses to changes inenvironment. a, Diurnal dark/light stomatal conductance response in slac1and wild-type plants with 6 s.e.m. (n 5 3). b, Time courses of stomatalresponses to changes in light intensity. c, Time courses of stomatal response

to changes in air humidity. d, Time courses of stomatal response to changes inCO2 concentration. Stomatal responses were monitored with an Arabidopsiswhole-rosette gas-exchange system15, and values in b–d were normalized toconductances at 0 min and represent averages (6s.e.m.) of four rosettes.

LETTERS NATURE | Vol 452 | 27 March 2008

488Nature Publishing Group©2008

Arabidopsis SLAC1 homologues (Supplementary Fig. 4a, b). Thetransmembrane domains of Arabidopsis and rice SLAC1 homologuesand orthologues have several highly conserved amino acids. However,SLAC1 and Os04g48530 also differ from the rest of the proteins inseveral amino-acid residues (Supplementary Fig. 4a). For example, theamino acid that is mutated in slac1-1 is a serine in SLAC1 andOs04g48530, whereas other family members have an alanine residuein the same position. The serine mutated in slac1-1 is surroundedby three conserved threonine residues, suggesting that this region issignificant for either the structure, function or regulation of theprotein. Additionally, the predicted intracellular loops between thetransmembrane domains have several conserved, positively chargedamino-acid residues (Supplementary Figs 3, 4a), also suggesting func-tional significance.

When 1,582 base pairs (bp) of genomic sequence upstream of theSLAC1 translation start were fused to the reporter gene uidA, theresulting b-glucuronidase (GUS) activity in transgenic plants waslocalized predominantly to guard cells (Fig. 1d), and occasionallyto the vascular strands close to the leaf margins (Fig. 1e). No GUSactivity was detected in other parts of the plants. Expression data atthe Genevestigator database18 and comparison of gene expressionbetween guard cell and mesophyll cell microarrays also suggest strongpreferential guard cell expression of SLAC1.

To study the subcellular location of the SLAC1 protein, greenfluorescence protein (GFP) fused to the SLAC1 C terminus wastransiently expressed in onion epidermal cells (Fig. 1f–k) and intobacco protoplasts (Supplementary Fig. 5). Fluorescence andconfocal imaging showed that in onion epidermal cells, fluorescencefrom the SLAC1::GFP fusion protein (Fig. 1f) and the membrane-specific stain FM 4-64 (Fig. 1g) colocalized in merged images(Fig. 1h). GFP fluorescence was observed between the cell wall andthe nucleus (Fig. 1j; Supplementary Movie), and was connected tothe cell wall through Hechtian strands in plasmolysed cells (Fig. 1k),correlating with plasma membrane localization. Expression in

tobacco protoplasts showed results that are consistent with plasmamembrane localization (Supplementary Fig. 5).

Stomatal aperture is under environmental and hormonal control.We analysed stimulus responses in stomatal conductance by com-paring intact15 slac1 with wild-type plants. Stomatal conductance inslac1 was about 1.5-fold higher during the light period (Fig. 2a). Also,the decline in stomatal conductance at the beginning of the darkperiod took more than 1 h longer in slac1 compared with the wildtype (Fig. 2a). Light/dark transitions during the normal light periodcaused rapid changes in stomatal conductance in the wild type,whereas slac1 showed a slow and modest response (Fig. 2b). slac1exhibited a much slower response than the wild type to a decrease inthe relative air humidity (Fig. 2c), which is known to cause a rapidreduction of stomatal conductance19. Doubling of [CO2] from 400p.p.m. to 800 p.p.m. reduced stomatal conductance effectively in thewild type, whereas slac1 showed no responses (Fig. 2d). Thus, slac1stomata show only a slow and modest response to changes in lightand air humidity, and are completely insensitive to O3 stress (Fig. 1a)and elevated [CO2] (Fig. 2d).

The concentration of the plant stress hormone ABA increasesunder drought and induces stomatal closure through second mes-sengers, including ROS, cytosolic Ca21 and NO20–22. We measuredstomatal responses to ABA, hydrogen peroxide (H2O2), NO andrepetitive Ca21 pulses (Fig. 3). Stomata of slac1 mutants showed astrong insensitivity to ABA (Fig. 3a and Supplementary Fig. 6a).Similarly, they showed significantly reduced responses to H2O2

(Fig. 3b) and the NO donor sodium nitroprusside (SNP) (Fig. 3c).Transient addition and removal of Ca21 to the extracellular solutionbathing leaf epidermides, while shifting the K1 equilibrium poten-tial, allows experimental imposition of defined intracellular Ca21

transients in guard cells, resulting in stomatal closure23–25. Four repe-titive 5-min pulses of 1 mM external Ca21 were applied (Fig. 3d;top inset; Supplementary Fig. 7). The imposed intracellular Ca21

([Ca21]i) oscillation pattern of slac1-1 guard cells was similar to that

a b

–15 0 10 20 30 40 50 60 90

70

80

90

100

60

Time (min)

Time (min)

–10 0 10 20 403075

80

85

90

95

100

105

Ap

ertu

re (o

f 100

%)

Ap

ertu

re (o

f 100

%)

Ap

ertu

re (o

f 100

%)

Ap

ertu

re (o

f 100

%)

60

80

100

120

60

Time (min)

Time (min)

WTslac1-1slac1-3

80

100

120

60300

60300

c d

ABA H2O2

SNP Ca2+

WTslac1-1slac1-3

WTslac1-1slac1-3

WTslac1-1slac1-3

Figure 3 | Impaired stomatal responses to ABA, H2O2, NO and Ca21 inslac1. a, Time-course experiments of ABA-induced stomatal closure. ABA(1 mM) was added at time 5 0 (n 5 3 experiments, 28, 23 and 57 stomata forslac1-1, slac1-3 and wild type, respectively). Stomatal apertures at time 5 0(100%) corresponded to average stomatal apertures of 2.82 6 0.16 mm (wildtype), 2.88 6 0.12 mm in slac1-1 and 3.23 6 0.08 mm in slac1-3. b, c, Timecourse of stomatal closure induced by H2O2 (100mM) (b) and NO (derivedfrom 50 mM SNP) (c). n 5 3–5 independent experiments, 20 stomata perexperiment. d, Impairment in stomatal closure in response to four transient

5-min extracellular applications of 1 mM CaCl2 and 1 mM KCl (black stripsat top; n 5 3 experiments, 48, 24 and 32 stomata for wild type, slac1-1 andslac1-3, respectively). Imposed intracellular Ca21 transients (seeSupplementary Fig. 7) were followed by 5-min exposures to a depolarizingsolution containing 0 mM CaCl2 and 50 mM KCl (white strips at top) aspreviously described25. Stomatal apertures at time 5 0 (100%) correspondedto average stomatal apertures of 3.56 6 0.10 mm, 3.93 6 0.14 mm and3.56 6 0.10 mm in wild type, slac1-1 and slac1-3, respectively. Error barsdepict means 6 s.e.m.

NATURE | Vol 452 | 27 March 2008 LETTERS

489Nature Publishing Group©2008

of the wild type (Supplementary Fig. 7). The average amplitudes ofimposed [Ca21 ]i transients and the integrated total [Ca21]i increasesper period were statistically similar in wild-type and slac1-1 guardcells (see Supplementary Information). Imposed [Ca21]i transientscaused the typical downstream Ca21-induced reactive and pro-grammed23,24 stomatal closure in the wild type, whereas the responsewas greatly impaired in slac1-1 and slac1-3 (Fig. 3d). Thus slac1mutant guard cells do not abrogate imposed cytosolic Ca21 oscilla-tions, but show a strong impairment in downstream Ca21 oscillation-induced stomatal closing.

The activation of S- and R-type anion efflux channels, both ofwhich can transmit Cl– and malate efflux from guard cells8,9,11, isproposed to decrease guard cell osmotic potential, leading to stoma-tal closure3,4,8,10,11. This is consistent with Cl– and malate efflux occur-ring in response to ABA26,27. We therefore applied whole-cell patchclamp techniques to characterize the functioning of S-type andR-type anion channel activities. In wild-type guard cells, elevatedcytosolic Ca21 (2 mM) activated ion currents that were selective forCl– over caesium ions (Cs1) (n 5 16 guard cells) and showed a rela-tive permeability ratio for malate to chloride anions of 0.125 (n 5 12guard cells), consistent with previous anion selectivity analyses ofS-type anion channel currents9 (Supplementary Fig. 8).

S-type anion currents were readily recorded in wild-type guardcells (Fig. 4a, d). However, only very small combined backgroundwhole-cell membrane currents and patch-clamp seal currents wereobserved in slac1-1 and slac1-3 guard cells (Fig. 4b–d). R-type anioncurrents11 were activated as described25,28. Interestingly, no signifi-cant differences in R-type anion currents between wild-type and slac1

guard cells were observed (Fig. 4e, f). Similarly, ABA activation ofCa21-permeable ‘ICa’ channel currents21 was not disrupted in slac1guard cells (Supplementary Fig. 9). However, when ABA activationof S-type anion channels was analysed, slac1 mutants showed onlysmall whole-cell currents (Fig. 4h–j), whereas S-type anion currentswere recorded in wild-type guard cells (Fig. 4g, j).

Continuing increases in ozone concentrations in the troposphereowing to human activities are predicted to have a negative affect oncrop yields and global carbon sinks in the future1,2. The ozone sen-sitivity of slac1 leaves (Supplementary Fig. 2), the predominant guardcell expression of SLAC1 (Fig. 1d, e) and abolishment of O3-inducedstomatal closure in slac1 mutants (Fig. 1a) together provide directgenetic evidence for the importance of O3 sensing in guard cellsfor plant O3 tolerance. Only a few plant mutants are known thatshow CO2 insensitivity5,6 or a constitutive high CO2 response7 instomatal movements, but no recessive CO2-insensitive mutant genehas been isolated so far. All slac1 alleles are recessive and show acomplete lack of high CO2-induced stomatal closure (Fig. 2), illus-trating that the SLAC1 protein is a central positive mediator of CO2-induced stomatal closure.

Experiments with ABA, ROS, NO and Ca21 suggest that SLAC1 isan essential protein functioning downstream of these messengers inmediating stomatal closure (Figs 3, 4 and Supplementary Figs 6, 7, 9).The phenotype of slac1 differs from the ATP-binding cassette trans-porter mutant, atmrp5, which shows partial repression of ABA-induced stomatal closure, partial S-type anion current activity andimpaired Ca21 channel activation29. The strong impairment inS-type anion channel and normal Ca21 channel activity in slac1guard cells is consistent with SLAC1 being more closely associatedwith S-type anion channels than is AtMRP5, and provides directgenetic evidence for the model that these anion channels functionas a central control mechanism for stomatal closure8.

R-type anion channel activity was not disrupted in slac1 guard cells(Fig. 4e, f), providing genetic evidence for a molecular separation ofthe membrane proteins required for S- and R-type anion channels. Itremains possible that these anion channel types share other proteinsubunits30. R-type channels may be responsible for the slow stomatalconductance decrease observed in response to light/dark transitionsand decrease in relative humidity (Fig. 2a–c).

The data presented demonstrate that SLAC1 encodes an essentialsubunit for S-type anion channel function or regulation. The lowhomology of SLAC1 to bacterial and fungal organic acid transportersindicates a possible role for SLAC1 in contributing to formation ofan anion-transporting pore. Further research on SLAC1 and itshomologues should increase the general understanding of plasma-membrane anion channel structure and regulation in plants.

METHODS SUMMARY

Three- to six-week-old A. thaliana plants grown in a controlled environment

were used. slac1-1 was isolated from an O3-sensitivity mutant screen13. The

mapping population was generated by outcrossing to Ler, and an impaired

water-loss phenotype was used as a mapping trait. For water-loss analyses, the

weight of the detached leaves was followed. Whole-plant stomatal conductance

responses to O3, light/dark transitions, elevated CO2 and lowered humidity were

measured using the Arabidopsis whole-rosette gas-exchange system15. For GUS

activity and complementation analyses, transgenic SLAC1 promoter-driven

GUS expression lines and complementation lines with SLAC1 genomic DNA

were analysed. For transient gene-expression studies, a SLAC1::GFP fusion pro-

tein under the control of a 35S promoter was delivered into onion epidermides

by particle bombardment, and to tobacco protoplasts by electroporation. Images

were acquired by confocal microscopy. For stomatal responses to H2O2, NO and

ABA, stomatal apertures were measured from extracted epidermal fragments

after pre-incubation of leaves in opening buffer. Stomatal responses to Ca21

transients and Ca21 imaging experiments were analysed in intact leaf epider-

mides by imposing extracellular calcium pulses23,25. For electrophysiological

analyses, Arabidopsis guard cell protoplasts were isolated enzymatically, and

Ca21 activation of S- and R-type anion currents and ABA activation of S-type

anion and ICa Ca21 currents were recorded as described25,29.

WT

WT

WT

2 s

4 s

50 p

A

30 p

A

20 p

A

V (mV)

V (mV)

V (mV)

V (mV)

WT

WT

I (p

A)

I (p

A)

I (p

A)

slac1-1

slac1-1

slac1-1

slac1-3

slac1-3

slac1-3

slac1-3

slac1-1slac1-3

slac1-3

−120

−120

−180

−240

−180

−180

−120

−120

−120

−60

−60

−60

−60

−60

−40

−10

−20

−30

−80

−120

−160

0

0

0

00

0

a

e

g h

i

j

f

b d

c

Figure 4 | Ca21 and ABA activations of S-type anion channels are impairedin slac1 guard cells. a–d, Ca21 activation of S-type anion channels.a–c, Whole-cell recordings of S-type anion currents in wild type (a), slac1-1(b) and slac1-3 (c). d, Average current–voltage curves of S-type anionchannel currents recorded in wild type (n 5 7), slac1-1 (n 5 12) and slac1-3(n 5 10). e, f, Typical R-type anion channel recordings (e), and averagecurrent–voltage curves in wild type (n 5 3) and slac1-3 (n 5 6) (f). g–j, ABAactivation of S-type anion channels. g–i, Typical recordings in wild type(g), slac1-1 (h) and slac1-3 (i). j, Average current–voltage curves recorded inwild type (n 5 10), slac1-1 (n 5 8) and slac1-3 (n 5 8). Error bars depictmeans 6 s.e.m.

LETTERS NATURE | Vol 452 | 27 March 2008

490Nature Publishing Group©2008

Full Methods and any associated references are available in the online version ofthe paper at www.nature.com/nature.

Received 22 August; accepted 31 December 2007.Published online 27 February 2008.

1. Hopkin, M. Carbon sinks threatened by increasing ozone. Nature 448, 396–397(2007).

2. Sitch, S., Cox, P. M., Collins, W. J. & Huntingford, C. Indirect radiative forcing ofclimate change through ozone effects on the land-carbon sink. Nature 448,791–794 (2007).

3. MacRobbie, E. A. C. Signal transduction and ion channels in guard cells. Phil. Trans.R. Soc. Lond. B 353, 1475–1488 (1998).

4. Hetherington, A. M. & Woodward, F. I. The role of stomata in sensing and drivingenvironmental change. Nature 424, 901–908 (2003).

5. Webb, A. A. R. & Hetherington, A. Convergence of the abscisic acid, CO2, andextracellular calcium signal transduction pathways in stomatal guard cells. PlantPhysiol. 114, 1557–1560 (1997).

6. Young, J. J. et al. CO2 signaling in guard cells: calcium sensitivity responsemodulation, a Ca21 -independent phase, and CO2 insensitivity of the gca2 mutant.Proc. Natl Acad. Sci. USA 103, 7506–7511 (2006).

7. Hashimoto, M. et al. Arabidopsis HT1 kinase controls stomatal movements inresponse to CO2. Nature Cell Biol. 8, 391–397 (2006).

8. Schroeder, J. I. & Hagiwara, S. Cytosolic calcium regulates ion channels in theplasma membrane of Vicia faba guard cells. Nature 338, 427–430 (1989).

9. Schmidt, C. & Schroeder, J. I. Anion-selectivity of slow anion channels in Vicia fabaguard cells: large nitrate permeability. Plant Physiol. 106, 383–391 (1994).

10. Pandey, S., Zhang, W. & Assmann, S. M. Roles of ion channels and transporters inguard cell signal transduction. FEBS Lett. 581, 2325–2336 (2007).

11. Keller, B. U., Hedrich, R. & Raschke, K. Voltage-dependent anion channels in theplasma membrane of guard cells. Nature 341, 450–453 (1989).

12. De Angeli, A. et al. The nitrate/proton antiporter AtCLCa mediates nitrateaccumulation in plant vacuoles. Nature 442, 939–942 (2006).

13. Overmyer, K. et al. Ozone-sensitive Arabidopsis rcd1 mutant reveals opposite rolesfor ethylene and jasmonate signaling pathways in regulating superoxide-dependent cell death. Plant Cell 12, 1849–1862 (2000).

14. Kangasjarvi, J., Jaspers, P. & Kollist, H. Signalling and cell death in ozone-exposedplants. Plant Cell Environ. 28, 1021–1036 (2005).

15. Kollist, T. et al. A novel device detects a rapid ozone-induced transient stomatalclosure in intact Arabidopsis and its absence in abi2 mutant. Physiol. Plant. 129,796–803 (2007).

16. Camarasa, C. et al. Characterization of Schizosaccharomyces pombe malatepermease by expression in Saccharomyces cerevisiae. Appl. Environ. Microbiol. 67,4144–4151 (2001).

17. Sasaki, T. et al. A wheat gene encoding an aluminum-activated malatetransporter. Plant J. 37, 645–653 (2004).

18. Zimmermann, P., Hirsch-Hoffmann, M., Hennig, L. & Gruissem, W.GENEVESTIGATOR: Arabidopsis microarray database and analysis toolbox. PlantPhysiol. 136, 2621–2632 (2004).

19. Xie, X. et al. The identification of genes involved in the stomatal response toreduced atmospheric relative humidity. Curr. Biol. 16, 882–887 (2006).

20. McAinsh, M. R., Brownlee, C. & Hetherington, A. M. ABA induced elevation ofguard cell cytosolic calcium precedes stomatal closure in Commelina communis.Nature 343, 186–188 (1990).

21. Pei, Z. M. et al. Calcium channels activated by hydrogen peroxide mediate abscisicacid signalling in guard cells. Nature 406, 731–734 (2000).

22. Desikan, R., Griffiths, R., Hancock, J. & Neill, S. A new role for an old enzyme:Nitrate reductase-mediated nitric oxide generation is required for abscisic acid-induced stomatal closure in Arabidopsis thaliana. Proc. Natl Acad. Sci. USA 99,16314–16318 (2002).

23. Allen, G. J. et al. A defined range of guard cell calcium oscillation parametersencodes stomatal movements. Nature 411, 1053–1057 (2001).

24. Li, Y. et al. The parameters of guard cell calcium oscillation encode stomataloscillation and closure in Vicia faba. Plant Sci. 166, 415–421 (2004).

25. Mori, I. C. et al. CDPKs CPK6 and CPK3 function in ABA regulation of guard cellS-type anion- and Ca21-permeable channels and stomatal closure. PLoS Biol. 4,1749–1762 (2006).

26. Van Kirk, C. A. & Raschke, K. Release of malate from epidermal strips duringstomatal closure. Plant Physiol. 61, 474–475 (1978).

27. MacRobbie, E. A. C. Ion fluxes in ‘isolated’ guard cells of Commelina communis L.J. Exp. Bot. 32, 545–562 (1981).

28. Frachisse, J. M., Thomine, S., Colcombet, J., Guern, J. & Barbier-Brygoo, H. Sulfateis both a substrate and an activator of the voltage-dependent anion channel ofArabidopsis hypocotyl cells. Plant Physiol. 121, 253–261 (1999).

29. Suh, S. J. et al. The ATP binding cassette transporter AtMRP5 modulates anionand calcium channel activities in Arabidopsis guard cells. J. Biol. Chem. 282,1916–1924 (2007).

30. Linder, B. & Raschke, K. A slow anion channel in guard cells, activating at largehyperpolarization, may be principal for stomatal closing. FEBS Lett. 313, 27–30(1992).

Supplementary Information is linked to the online version of the paper atwww.nature.com/nature.

Acknowledgements We thank M. Uuskallio and I. Puzorjova for technical help. Thisresearch was supported by the Academy of Finland Centre of Excellenceprogramme and Helsinki University Environmental Research Centre (to J.K.), byEstonian Science Foundation and University of Tartu start-up grants (to H.K.), byNIH, NSF and, in part, DOE grants (to J.I.S.), and a Leverhulme Trust Early CareerFellowship (to R.D.)

Author Contributions T.V., H.K. and Y.-F.W. contributed equally to this work. J.K.and H.K. designed the experiments in Figs 1 and 2. A.L., H.K. and T.V. identified theSLAC1 gene. T.V. and M.B. performed the expression, complementation andsubcellular localization analyses in Fig. 1 and Supplementary Fig. 5. H.K. and H.M.performed experiments in Fig. 2. H.K. performed experiments in SupplementaryFigs 1 and 2. R.D. designed and performed experiments in Fig. 3b, c andSupplementary Fig. 6b. J.I.S. and J.K. designed experiments in Figs 3a and d, and 4,and Supplementary Figs 6a, 7, 8 and 9. W.-Y.C. and G.V. performed experiments inFig. 3d and Supplementary Fig. 6a. N.N. performed experiments in Fig. 3a andSupplementary Fig. 7. Y.-F.W. performed experiments in Fig. 4 and SupplementaryFigs 8 and 9. J.K. and J.I.S. wrote the paper. All the authors discussed the results,and commented on and edited the manuscript.

Author Information The primary microarray data reported has been depositedwith the ArrayExpress database under accession number E-MEXP-1388. Reprintsand permissions information is available at www.nature.com/reprints.Correspondence and requests for materials should be addressed to J.K.([email protected]).

NATURE | Vol 452 | 27 March 2008 LETTERS

491Nature Publishing Group©2008

METHODSPlant material and growth conditions. A. thaliana were grown in controlled

growth conditions in 1:1 peat/vermiculite mix, 150mmol m22 s21 light, 23 uC/

19 uC day/night, 12 h photoperiod. For patch clamping, Ca21 and ABA experi-

ments, plants were grown as described previously29 and at .70% humidity.

For whole-plant gas-exchange experiments, plants were grown as described

previously15.

Mapping. A mutant screen for increased O3 sensitivity13, in which approximately

14,000 EMS-mutagenized individual M2 plants (M2E-1A-4, Lehle Seeds) were

exposed to 250 p.p.b. of O3 for 6 h, was performed. Individuals showing different

patterns of O3-induced lesions were selected and named as rcd (radical-induced

cell death) mutants14. Segregation analysis of one of the mutants, originally called

rcd3-1, revealed that its O3-sensitive phenotype is conferred by a single mono-

genetic recessive allele, and the impaired water-loss phenotype co-segregated

with O3 sensitivity. For the identification of the mutation, a mapping population

was generated by outcrossing to Ler. Impaired water-loss phenotype was used as

a trait to identify 572 F2 mapping lines. The use of CAPS and dCAPS markers

established linkage to a 116.5-kb region with 34 genes between markers

SGCSNP9727 and SGCSNP9735. In this region, water loss from 15 T-DNA

insertion lines of different genes was analysed. Only homozygous T-DNA lines

SALK_099139 and SALK_137265 (later named as slac1-3 and slac1-4, respec-

tively), positioned in the exons of At1g12480, showed the impaired water-loss

phenotype (Fig. 1b) and only this gene had more than tenfold higher expression

in guard cell protoplasts than in mesophyll protoplasts31,32. The At1g12480 gene,

which encodes a C4-dicarboxylate transporter/malic acid transport protein

domain containing protein and RCD3, was renamed SLOW ANION

CHANNEL-ASSOCIATED 1 (SLAC1). Sequencing of genomic DNA and com-

plementary DNA of At1g12480 from slac1-1 revealed a C-to-T point mutation

resulting in a serine-to-phenylalanine conversion in amino acid 456 (Fig. 1c;

Supplementary Fig. 4a).

Water-loss measurements. The weight of detached leaves, incubated abaxial

side up under laboratory conditions, was followed at various time points.

Water loss was expressed as the percentage of initial fresh weight.

Stomatal density. Stomatal density was measured from images of second and

third true leaves of 2-week-old plants taken with an Olympus Provis AX70

microscope using 10 3 0.30 and 20 3 0.50 water objectives. Images were

acquired with an Olympus DP70 automated photomicrographic system.

SLAC1 promoter-driven GUS expression lines. SLAC1 promoter region

(1,504 bp) was cloned into the pMDC162 vector33 using Gateway technology

(Invitrogen). The floral dip method34 was used for obtaining transgenic lines.

Histochemical staining for GUS activity35 was performed on 3-week-old plants.

Transient gene-expression analyses in onion epidermis and tobacco proto-plasts. Full-length SLAC1 cDNA (1,671 bp) under the control of the 35S pro-

moter was cloned into the GFP fusion vector pMDC83, and delivered into onion

epidermal cells by particle bombardment using a Helios Gene Gun (BioRad),

and to tobacco protoplasts by electroporation36. Images were acquired using a

Leica SP2 AOBS confocal microscope (Leica Microsystems) 24 h after incuba-

tion using the HC PL APO 20x/0,7 Imm Corr (water) objective. A 488-nm laser

was used for GFP (emission 495–554 nm) and chlorophyll (emission 578–

682 nm), 364-nm laser for DAPI (emission 400–481 nm) and 561-nm laser for

FM 4-64 dye (emission 641–785 nm) imaging. Onion epidermis was stained with

20 mM FM 4-64 dye (in K Murashige and Skoog Medium), for 10 min to stain

only the plasma membrane or with DAPI (20mg ml21 in K MS) for 15 min to

stain the nucleus. Plasmolysis was generated by incubating onion epidermis in

5% NaCl for 20 min. Onion epidermis images were processed by Leica Confocal

Software Lite (Leica Microsystems). Tobacco protoplast images were deconvo-

luted by the Media Cybernetics AutoDeblur 3D Blind Deconvolution software.

For tobacco protoplast images and supplementary movie, a three-dimensional

reconstruction was created by Imaris 5.7.1 software, Bitplane. In the supple-

mentary movie, the nucleus is shown by DAPI channel as surface.

Complementation analysis. An SLAC1 genomic DNA fragment (4,075 bp) was

cloned into the pMDC100 vector33, and slac1-1 plants were complemented using

the floral dip method. Independent T2 plants were analysed for water-loss

phenotype.

Whole-plant stomatal conductance and stomatal aperture measurements.Details of Arabidopsis whole-rosette gas-exchange measurements are as

described15. For analysing stomatal responses to light/dark transitions, light

was removed for 30 min; for responses to O3, whole plants were exposed to

200 p.p.b. O3; for responses to different CO2 concentrations, whole plants were

exposed to 800 p.p.m. CO2 for 40 min, 2,000 p.p.m. CO2 for 40 min and

400 p.p.m. CO2 for 30 min. For responses to changes in air humidity, plants were

stabilized at 74 6 5% relative humidity for 20–40 min, and relative humidity was

reduced to 45 6 6% for 40 min.

Leaves were floated on stomatal opening buffer (10 mM MES, 5 mM KCl,

50 mM CaCl2, pH 6.15) for 2.5 h, then treated with 100mM H2O2 or 50 mM

SNP for 2.5 h. Data in Fig. 3b, c are from 3–5 independent experiments

(n 5 20 per experiment). For responses to ABA, intact leaf epidermis6 (Fig. 3a)

or leaves (Supplementary Fig. 6) were incubated for 3 h in stomatal opening

buffer (5 mM MES, 10 mM KCl, 50 mM CaCl2, pH 5.6), and exposed to the

indicated ABA concentrations. Thereafter stomatal apertures were measured.

Data in Fig. 3a were performed as genotype blind analyses (n 5 3 experiments,

28, 23 and 57 stomata in total for slac1-1, slac1-3 and wild type, respectively).

Data in Supplementary Fig. 6a are from double blind experiments in which the

ABA concentration and genotype of leaves were unknown to the experimenter

(n 5 4 experiments, 30 stomata per condition and experiment). All data are

presented as mean 6 s.e.m.

Stomatal movement responses to Ca21 pulses. Stomatal responses and Ca21

imaging experiments analysing responses to extracellular Ca21 pulses were per-

formed as described previously23,25. Intact leaf epidermis6 adhered to a coverslip

was preincubated in stomatal opening buffer with 0 mM added CaCl2 and depo-

larizing K1 (50 mM KCl, 10 mM MES-Tris (pH 5.6), 0 mM CaCl2) for 3 h in

200mmol m22 s21 white light. Four extracellular calcium pulses were imposed by

applying a 1 mM CaCl2 hyperpolarizing buffer for 5 min (1 mM CaCl2, 1 mM

KCl, 10 mM MES-Tris (pH 5.6)) followed by 5 min of the above 0 mM CaCl2depolarizing buffer25. Controls with extracellular Ca21 removed (using Ca21

chelator) showed no cytosolic [Ca21] elevations and no clear stomatal closing

responses to alternating KCl bath solutions (see supplementary information of

ref. 37). After four transient exposures to Ca21, the bath solution was returned to

the above stomatal opening buffer (0 mM CaCl2, 50 mM KCl) for the remaining

time. Data in Fig. 3d were genotype blind analyses: n 5 24 slac1-1 stomata,

n 5 32 slac1-3 stomata and n 5 48 wild-type stomata. Cytosolic Ca21 concen-

tration changes were monitored using the FRET reporter Yellow Cameleon 3.6

(Supplementary Fig. 7) as previously described25 in n 5 39 slac1-1 guard cells and

n 5 35 wild-type guard cells.

Patch clamp analyses. Arabidopsis guard cell protoplasts were isolated as

described previously29,38. S- and R-type anion currents were recorded in

Arabidopsis guard cells as described25, with bath and pipette solutions as below.

To analyse Ca21 activation of S-type anion channels, the bath solution contained

30 mM CsCl, 1 mM CaCl2, 2 mM MgCl2, 10 mM MES-Tris, pH 5.6; the pipette

solution contained 150 mM CsCl, 2 mM MgCl2, 5 mM Mg-ATP, 6.7 mM EGTA,

10 mM HEPES-Tris, pH 7.1, and 5.864 mM CaCl2 to result in 2 mM free Ca21.

The osmolalities of solutions were adjusted with D-sorbitol to 485 mmol kg21

for the bath solution, and 500 mmol kg21 for the pipette solution. Guard cell

protoplasts were extracellularly pre-incubated for 30 min in the same bath solu-

tion with 40 mM CaCl2 added before patch clamping39. The CaCl2 concentration

in the bath solution was reduced from 40 mM to 1 mM by perfusion before patch

clamping, whole-cell recordings were achieved within 30 min after the preincu-

bation39. To analyse ABA activation of S-type anion channels, the same pipette

and bath solutions as for Ca21 activation of S-type anion channels were used

(without 40 mM CaCl2 pre-incubation), and 5 mM Tris–GTP was freshly added

to the pipette solution each day. Arabidopsis guard cell protoplasts were pre-

incubated with 50 mM ABA for 20 min before patch clamping, and patch clamp

experiments were performed in the presence of 50 mM ABA in the bath solution.

S-type anion currents were measured 7–10 min after gaining access to whole-cell

configurations. The membrane voltage was stepped from 135 mV to –145 mV

with 30 mV decrements as described previously25, and the holding potential was

120 mV. For R-type anion channel recording, the bath solution contained

50 mM CaCl2, 2 mM MgCl2, 10 mM Mes-Tris, pH 5.6. The pipette solution

contained 75 mM K2SO4, 5 mM EGTA, 2.5 mM CaCl2, 10 mM HEPES-Tris,

pH 7.1, and osmolalities were adjusted with D-sorbitol to 485 mmol kg21 for

the bath solution, and 500 mmol kg21 for the pipette solution as described

previously25.

For analyses of plasma membrane Ca21-permeable channel currents, the bath

solution contained 100 mM BaCl2, 0.1 mM DTT, 10 mM MES-Tris, pH 5.6,

osmolarity was 485 mmol kg21 adjusted using D-sorbitol. The pipette solution

contained 10 mM BaCl2, 0.1 mM DTT, 4 mM EGTA, 5 mM b-NADPH, 10 mM

HEPES-Tris, pH 7.1, osmolarity was 500 mmol kg21 adjusted using D-sorbitol.

Solutions and experimental conditions were the same as previously described21,38.

Heterologous expression analyses. When SLAC1 was expressed in Xenopus

oocytes, it did not generate clear anion currents, nor did SLAC1 complement

malate-uptake mutations in E. coli deficient for four or five malate transporters

(AN387 derivatives IMW213b (AN387, but dctA::spcR dcuA::spcR dcuB::kanR

dcuC::miniTn10 camR) and IMW244 (AN387, but dctA::spcR dcuA::spcR

dcuB::kanR dcuC::miniTn10 camR dcuD::ampR))40, or tehA41, which indicates

that further research on the structure, regulation and possible complex forma-

tion of SLAC1 and its homologues is needed. Note, however, that lack of com-

plementation of the E. coli malate-uptake transporter mutants can be expected

doi:10.1038/nature06608

Nature Publishing Group©2008

because SLAC1 is required for anion channel function in guard cells. Anionchannels would energetically favour anion efflux and would thus be less well

suited for anion uptake into cells under most conditions owing to the electro-

chemical gradient for anions. Furthermore, activation of guard cell anion flux is

known to be regulated by phosphorylation25,42,43, for which the required mechan-

isms probably do not exist in either Xenopus oocytes or E.coli. Because ozone also

regulates guard cell K1 channels44, SLAC1 may be less likely to encode a direct

ozone target.

31. Leonhardt, N. et al. Microarray expression analyses of Arabidopsis guard cells andisolation of a recessive ABA hypersensitive protein phosphatase 2C mutant. PlantCell 16, 596–615 (2004).

32. Yang, Y., Costa, A., Leonhardt, N., Siegel, R. S. & Schroeder, J. I. Isolation of astrong Arabidopsis guard cell promoter and its potential as a research tool. BMCPl. Methods. (in the press).

33. Curtis, M. D. & Grossniklaus, U. A gateway cloning vector set for high-throughputfunctional analysis of genes in Planta. Plant Physiol. 133, 462–469 (2003).

34. Clough, S. J. & Bent, A. F. Floral dip: a simplified method for Agrobacterium-mediated transformation of Arabidopsis thaliana. Plant J. 16, 735–743 (1998).

35. Weigel, D. & Glazebrook, J. Arabidopsis: A Laboratory Manual (Cold Spring HarborLaboratory Press, Cold Spring Harbor, New York, 2002).

36. Suntio, T. M. & Teeri, T. H. A new bifunctional reporter gene for in-vivo tagging ofplant promoters. Plant Mol. Biol. Rep. 12, 43–57 (1994).

37. Allen, G. J., Chu, S. P. & Schroeder, J. I. A defined range of guard cell calciumoscillation parameters encodes stomatal movements. Nature 411, 1053–1057(2001).

38. Murata, Y., Pei, Z. M., Mori, I. C. & Schroeder, J. I. ABA activation of plasmamembrane Ca21 channels in guard cells requires cytosolic NAD(P)H and isdifferentially disrupted upstream and downstream of reactive oxygen speciesproduction in the abi1–1 and abi2–1 PP2C mutants. Plant Cell 13, 2513–2523(2001).

39. Allen, G. J., Murata, Y., Chu, S. P., Nafisi, M. & Schroeder, J. I. Hypersensitivity ofabscisic acid-induced cytosolic calcium increases in Arabidopsisfarnesyltransferase mutant era1–2. Plant Cell 14, 1649–1662 (2002).

40. Janausch, I. G., Kim, O. B. & Unden, G. DctA- and Dcu-independent transport ofsuccinate in Escherichia coli: contribution of diffusion and of alternative carriers.Arch. Microbiol. 176, 224–230 (2001).

41. Taylor, D. E. et al. Location of a potassium tellurite resistance operon (TehA TehB)within the terminus of Escherichia-coli k-12. J. Bacteriol. 176, 2740–2742 (1994).

42. Schmidt, C., Schelle, I., Liao, Y. J. & Schroeder, J. I. Strong regulation of slow anionchannels and abscisic acid signaling in guard cells by phosphorylation anddephosphorylation events. Proc. Natl Acad. Sci. USA 92, 9535–9539 (1995).

43. Li, J., Wang, X. Q., Watson, M. B. & Assmann, S. M. Regulation of abscisic acid-induced stomatal closure and anion channels by guard cell AAPK kinase. Science287, 300–303 (2000).

44. Torsethaugen, G., Pell, E. J. & Assmann, S. M. Ozone inhibits guard cell K1

channels implicated in stomatal opening. Proc. Natl Acad. Sci. USA 96,13577–13582 (1999).

doi:10.1038/nature06608

Nature Publishing Group©2008