ARTICLE IN PRESS

European Journal of Cell Biology 86 (2007) 489–500

0171-9335/$ - se

doi:10.1016/j.ej

�Correspondof Queensland,

fax: +617 3365

E-mail addr1These autho2Present add

Bohrweg 2, 2333Present add

University Brem

www.elsevier.de/ejcb

Distinct fluorescent pattern of KAT1::GFP in the plasma membrane of

Vicia faba guard cells

Ulrike Homann1, Tobias Meckel1,2, Jennifer Hewing, Marc-Thorsten Hutt3,Annette C. Hurst�

Institute of Botany, University of Technology Darmstadt, Schnittspahnstrasse 3-5, 64287 Darmstadt, Germany

Received 18 May 2006; received in revised form 15 May 2007; accepted 15 May 2007

Abstract

The organisation of membrane proteins into certain domains of the plasma membrane (PM) has been proposed tobe important for signalling in yeast and animal cells. Here we describe the formation of a very distinct pattern of theK+ channel KAT1 fused to the green fluorescent protein (KAT1::GFP) when transiently expressed in guard cells ofVicia faba. Using confocal laser scanning microscopy we observed a radially striped pattern of KAT1::GFPfluorescence in the PM in about 70% of all transfected guard cells. This characteristic pattern was found to be cell typeand protein specific and independent of the stomatal aperture and the cytoskeleton. Staining of the cell wall of guardcells with Calcofluor White revealed a great similarity between the arrangement of cellulose microfibrils and theKAT1::GFP pattern. Furthermore, the radial pattern of KAT1::GFP immediately disappeared when turgor pressurewas strongly decreased by changing from hypotonic to hypertonic conditions. The pattern reappeared within 15minupon reestablishment of high turgor pressure in hypotonic solution. Evaluation of the staining pattern by amathematical algorithm further confirmed this reversible abolishment of the radial pattern during hypertonictreatment. We therefore conclude that the radial organisation of KAT1::GFP depends on the close contact between thePM and cell wall in turgid guard cells. These results offer the first indication for a role of the cell wall in the localisationof ion channels. We propose a model in which KAT1 is located in the cellulose fibrils intermediate areas of the PM anddiscuss the physiological role of this phenomenon.r 2007 Elsevier GmbH. All rights reserved.

Keywords: Potassium inward rectifier; KAT1; Cell wall; Cellulose fibres; Cytoskeleton; Latrunculin B; Propyzamide; Turgor

pressure; Membrane domains

e front matter r 2007 Elsevier GmbH. All rights reserved.

cb.2007.05.003

ing author. School of Biomedical Sciences, University

St. Lucia, QLD 4072, Australia. Tel.: +61 7 3346 1224;

1766.

ess: a.hurst@uq.edu.au (A.C. Hurst).

rs contributed equally to the work.

ress: Institute of Physics, Leiden University, Niels

3 CA Leiden, The Netherlands.

ress: School of Engineering and Science, Jacobs

en, Campus Ring 1, D-28759 Bremen, Germany.

Introduction

Structural conditions which cause a heterogeneousdistribution of membrane proteins are believed to beimportant factors for the regulation of numeroussignalling and transport events, especially at the plasmamembrane (PM). So far a number of different mechan-isms that induce a heterogeneous distribution of

ARTICLE IN PRESSU. Homann et al. / European Journal of Cell Biology 86 (2007) 489–500490

membrane proteins have been described mainly inmammalian cell lines and yeast.

These mechanisms include the lateral separation ofspecific membrane lipid species (i.e. mainly cholesteroland glycosphingolipids) which leads to the formation ofspecialised microdomains, so-called lipid rafts. Certainproteins accumulate in these microdomains (e.g. glycosylphosphatidyl inositol (GPI)-anchored proteins) whileothers are not affected. This accumulation can beexplained by a slowdown of protein mobility by a factorof �2 (Dietrich et al., 2002) which results in aheterogeneous distribution of proteins in the PM. Manyproteins involved in signalling cascades are found in lipidrafts (Brown and Rose, 1992). This provided the first hinton the physiological significance of microdomain forma-tion. The slowdown and accumulation of proteins in lipidrafts is believed to increase the probability of dimer,multimer and cluster formation which is important formany signalling events at the PM. Nevertheless, the exactrole of lipid rafts in cellular signalling, trafficking, andstructure has yet to be determined.

Lipid rafts and raft-associated proteins have also beenidentified in plants (Borner et al., 2005; Mongrand et al.,2004). However, like for lipid raft formations inmammalian and yeast cells, the physiological meaningof these microdomains remains to be confirmed.

A second factor that was found to limit free diffusionof membrane proteins is the cytoskeleton. It candetermine the localisation of PM proteins via directattachment to the protein or indirectly. A direct connec-tion to the cytoskeleton is for example of particularimportance for the localisation and functioning of thecellulose synthase complex in the PM of plant cells(Gardiner et al., 2003). In the case of certain mechan-osensitive ion channels the attachment to the cytoske-leton is also proposed to function as a signallingcomponent in mammalian cells (Barritt and Rychkov,2005; Ghazi et al., 1998). Apart from the directconnection to PM proteins it is believed that the actincytoskeleton can confine the movement of proteins withenlarged cytosolic domains by generating ‘‘fenced’’microenvironments without the direct attachment tothe diffusing components (Ritchie and Kusumi, 2004).While such a mechanism has so far not been described inplant cells, a cortical cytoskeleton—the only determi-nant of this effect—is also present in plant cells (Staigerand Lloyd, 1991).

The third factor that can contribute to heterogeneousdistribution of PM proteins is the extracellular matrix(ECM). It is able to directly influence the distribution ofproteins in the PM of eukaryotic cells (Arnold et al.,2004). For plant cell walls, which can be viewed as theplant ECM, a role in distribution of PM proteins hasnot yet been described.

PM ion channels which play an important rolein signal transduction have been implicated to be

distributed non-homogenously in the PM of plant andanimal cells (Tester, 1990; Deutsch, 2002). In animalcells lipid rafts have been described as an importantfactor for the localisation of ion channels in certaindomains in the PM (Martens et al., 2004). For plant ionchannels mechanisms which determine their localisationin microdomains have not been identified.

Recently Sutter et al. (2006) demonstrated that theK+ inward rectifier KAT1 from Arabidopsis thaliana islocalised in clusters in the PM when transientlyexpressed in tobacco epidermal cells. The authorsdetected KAT1 protein in a ‘moderately’ detergent-resistant fraction, indicating its association with lipidrafts. The KAT1 cluster showed nearly no lateralmobility. Investigations of the role of SNAREs (solubleNSF [N-ethylmaleimide-sensitive factor] attachmentprotein receptors) on trafficking of KAT1 indicatedthat SNAREs are involved in cluster formation andmobility of KAT1 (Sutter et al., 2006). However, themechanism which anchors KAT1 in the PM remains tobe determined.

KAT1 plays an important role in guard cell function-ing. We therefore analysed turgid guard cells transientlyexpressing KAT1 fused to green fluorescent protein(GFP). KAT1::GFP was organised in clusters in the PMsimilar to what we previously described for guard cellprotoplasts (Hurst et al., 2004). In addition we found aradial distribution of KAT1::GFP clusters which wasdependent on a close contact between the PM and thecell wall. In animal cells contacts of the PM with theECM are mediated by substrate adhesion moleculessuch as fibronectin, vitronectin, collagen, and others, viathe short amino acid sequence Arg-Gly-Asp (RGD)(D’Souza et al., 1991) that interacts with integrins. Theintegrins in turn link the ECM to the cytoskeleton(Ruoslahti, 1996). For plant cell walls plant biologistsare just beginning to understand how cell wall-to-membrane interactions are established to acquire cell-and tissue-specific characters and how this affects cellfunction and polarity and cell-to-cell interactions. Sofar, in plants only few homologues of classical adhesionmolecules, e.g. b-integrin or fibronectin that revealed aRGD-mediated membrane matrix adhesion have beenidentified (Canut et al., 1998; Faik et al., 1998; Genset al., 1996; Pellenc et al., 2004). This points to a similarinteraction between the PM and the ECM or cell wall ofanimal and plant cells, respectively. In addition anumber of plant PM proteins have been proposed todirectly bind to both the PM and extracellular carbohy-drates and may thus anchor the cell to the cell wall (forreview see (Kohorn, 2000)). Among these is the cellulosesynthase complex (Kohorn, 2000), which also requirescortical microtubule arrays for normal localisation inthe PM (Gardiner et al., 2003). This indicates that acontinuous cytoplasm–cell wall scaffold is essential tocontrol key events in plant development and growth.

ARTICLE IN PRESSU. Homann et al. / European Journal of Cell Biology 86 (2007) 489–500 491

Our data clearly demonstrate the crucial role of theguard cell wall in the pattern formation of KAT1. Wepropose a model in which KAT1 clusters are located inthe intermediate areas of the cellulose microfibrils anddiscuss the physiological role of such a localisation.

Materials and methods

Construction/design of the KAT1::GFP fusion

protein

The kat1 cDNA was amplified by PCR for directedcloning into pAVA393 (NcoI restriction site) in framewith the mGFP5 (Haseloff et al., 1997) for expression ofthe fusion protein KAT1::GFP under the control of twoP35S promoters, as described in detail by Hurst et al.(2004). The plasmid was then cloned in Escherichia coli/DH5a, followed by preparation of plasmid DNA(Qiagen high speed Midi-Kit, Qiagen, Germany). Thepurified vector was used for ballistic bombardment ofguard cells and epidermal cells as described by Hurst etal. (2004).

The construct GFP::TM23 was kindly provided byNadine Paris (University of Rouen, France). It repre-sents the transmembrane domain of human LAMP1(lysosome-associated membrane protein-1) fused to thegreen fluorescent protein in the vector pSGFP6K(Brandizzi et al., 2002).

The P-ATPase PMA4 from Nicotiana plumbaginifolia

fused to GFP in the vector pTZ19U-gfp (Lefebvre et al.,2004) was a kind gift from M. Boutry, (UniversiteCatholique de Louvain, Belgium). The talin::YFPconstruct (Kost et al., 1998) was kindly provided by B.Kost (University of Heidelberg, Germany), and theplasmid with TOR1::GFP (Buschmann et al., 2004) byA. R. Schaffner (GSF Research Center, Germany),respectively. All fusion constructs were expressed underP35S.

Transfection of guard cells

Guard cells were transfected via ballistic bombard-ment as described by Hurst et al. (2004). Briefly, wholeleaves of Vicia faba L. cv. Bunyan were placed upsidedown on solid Murashige Skoog Medium and bom-barded with 2mg gold (1 mm particle diameter) coatedwith 10 mg DNA according to the manufacturer’sinstructions (BioRad, Munich, Germany), at a pressureof 650 Psi, a distance of 6 cm, and a vacuum of 25 in Hg.

Confocal microscopy

Confocal microscopic analysis of transfected guardcells was carried out 16 to 24 h after ballistic bombard-

ment as described by Meckel et al. (2004). Briefly,abaxial epidermal peels from bombarded leaves wereplaced in small dishes in a standard buffer solutionconsisting of 10mM MES-KOH (pH 6.1), 45mM KCl,and 100 mM CaCl2. Analysis was carried out using aconfocal laser scanning microscope (Leica TCS SP,Leica Microsystems GmbH, Heidelberg, Germany),equipped with a 63� water immersion objective (planapo, N.A. 1.2). For excitation of mGFP5, the 488-nmline of a 25-mW Ar/Kr-Ion-Laser was used; emissionwas detected at 505–535 nm.The confocal aperture wasadjusted to give optical sections with a full-width at half-maximum of around 0.68 mm. Images were processedusing the Leica Confocal Software 2.00 (LCS, LeicaMicrosystems GmbH, Heidelberg, Germany).

Analysis of guard cells stained with Calcofluor White

For staining of cellulose fibrils Calcofluor White(Sigma, Munich, Germany) was added to epidermalpeels bathed in standard buffer to a final concentrationof 0.1% (w/v). Epidermal peels were incubated inCalcofluor White for 5min and washed 5 times withstandard buffer. Epidermal peels were analysed usingthe confocal laser scanning microscope Leica TCS SP2AOBS (Leica Microsystems GmbH, Heidelberg,Germany) with a 63� water immersion objective. Forexcitation the 405-nm line of a 50-mW Ar-UV-Laserwas used, emission was detected at 415–440 nm.

Cytoskeleton inhibitors

Latrunculin B (Calbiochem, Darmstadt, Germany)and propyzamide (Sigma, Munich, Germany) wereprepared in dimethyl sulphoxide (DMSO) as 25 and50mM stocks, respectively. The toxins were diluted withstandard buffer solution to a final concentration of10 mM latrunculin B and 50 mM propyzamide. The finalconcentration of DMSO was 0.04% (v/v) and 0.05%(v/v), respectively.

Hyper- and hypotonic treatment of guard cells

Epidermal peels were placed in standard buffer (seeabove) which was hypotonic (100mosmol/kg), ensuringthat cells were fully turgescent. For hypertonic treat-ment of cells the standard buffer plus sorbitol(1000mosmol/kg) was added until plasmolysis wasvisible just at the tips of guard cells. The final osmolarityof the bath solution was between 500 and 800mosmol/kg. To restore turgor pressure hypotonic conditionswere induced by replacing the bath solution withstandard buffer.

ARTICLE IN PRESS

DC

BA

Fig. 1. Distribution of KAT1::GFP in guard cells and

epidermal cells. Maximum projection of cells expressing

KAT1::GFP (green). (A) Guard cell of stomata displaying a

clear radial staining pattern (type 3). (B) Guard cell of stomata

displaying radial staining pattern only in part of the cell

(type 2). (C) Guard cell displaying random staining pattern

(type 1). (D) Epidermal cell displaying random staining

pattern. Scale bars: 10 mm.

U. Homann et al. / European Journal of Cell Biology 86 (2007) 489–500492

Determination of radial distribution versus random

distribution via mathematical methods

The algorithmic challenge in quantifying the degreeof radial organisation in these fluorescence dis-tribution patterns is due to three features of thedata: (i) inspection by eye shows that the degreeof radial organisation varies gradually between theimage sets, (ii) noise content (i.e. randomly excitedpixels in the fluorescence images) varies stronglywithin the data and (iii) the image sizes are toosmall for an informative spectral analysis. Thesecircumstances are calling for a quantification of thepattern to attach a reproducible value to each imagewhich is independent of the observer. Our methodexploits the fact that the radially organised patternsand the random spots transform differently under anIsing algorithm (i.e. a well-defined domain-formingdynamical process). For all analysis steps we use abinarised fluorescence image, obtained by substitutingthe value of each image point by 1, if it exceeds acertain threshold and 0 otherwise. In all cases we havechosen the average fluorescence (without background)as binarisation threshold. The Ising algorithm is anearly thermodynamic model of ferromagnetism. Inthe original model of a ferromagnet the implementa-tion of this algorithm leads to domain formation.Here it emphasises existing radial patterns. The result-ing transformed image is then evaluated using astandard method from information theory, namely themutual information. This process assigns a number(the information content of the image, as given bythe mutual information) to this transformed image.The value is related to the radial organisation ofthe pattern, due to the different transformation proper-ties of radial stripes and random dots. Whilethis procedure is sufficient to assess the organisa-tional features of a pattern at fixed noise intensityof the image, the varying noise content of the datarequires an additional analysis step. We estimatethe noise content of each (binarised) image with thehelp of a melting algorithm. The application ofthis algorithm gives the percentage of image pointswhich are more likely to be noise rather thansignal. Consequently, application of our algorithmsto a given fluorescence image yields two numbers,namely the noise content of the original (binarised)image and the information content of theIsing-transformed image as a measure of stripeformation.

We calibrated the full algorithm on simulated datawhere the stripe-to-dot ratio and the noise content havebeen varied systematically. This led to the referencecurves shown in Fig. 7G. A more detailed descriptionand application of the Ising-algorithm in this context isgiven by Hutt (2001).

Results

Radial distribution of KAT1::GFP in the PM of

guard cells

Figs. 1A–C show intact turgid guard cells of V. faba

transfected with the K+ channel KAT1 fused to GFP(KAT1::GFP). The expression of KAT1::GFP resultedin a distinct staining pattern of the PM which in manycells appeared as stripes originating from the dorsal sideof the guard cells and radially centred towards theirventral side (Fig. 1A). For further analysis we cate-gorised the stripe formation as follows: guard cellsshowing a clear radial pattern over the whole PM wereclassified as type 3 (Fig. 1A); guard cells with lessprominent stripe formation only visible in part of thePM or displaying shorter stripes not forming acontinuous line from the dorsal to ventral side wereclassified as type 2 (Fig. 1B); guard cells without anyclear stripe formation with KAT1::GFP appearing inlarge randomly distributed dots in the PM wereclassified as type 1 (Fig. 1C). According to thisclassification 36% out of the 157 guard cells analysedwere classified as type 3 and 32% each as type 2 and type1. In addition to staining of the PM, guard cells oftenexhibited labelling of intracellular compartments(mainly ER and nuclear envelope). This most likelyresulted from protein expression under the strong 35Spromoter which can lead to accumulation of the fusionprotein in the ER (Hawes et al., 2001).

ARTICLE IN PRESS

15 type 1

U. Homann et al. / European Journal of Cell Biology 86 (2007) 489–500 493

Distribution pattern of KAT1::GFP is cell type and

protein specific

In addition to guard cells we also analysed thedistribution of KAT1::GFP fluorescence in transfec-ted epidermal cells. Epidermal cells expressing theKAT1::GFP fusion protein displayed a strikinglydifferent staining pattern compared to guard cells.The stripe formation typical of guard cells was neverobserved in epidermal cells (n4 50). All epidermal cellsdisplayed random distribution of KAT1::GFP fluores-cence in large dots in the PM (Fig. 1D). This implicatesthat the radial distribution of KAT1::GFP is cell typespecific. To examine whether PM proteins in guard cellsare generally distributed in a radial pattern we analysedthe expression of two different PM proteins fused toGFP. These proteins included the 23 amino acid longsingle-pass membrane domain from human LAMP1(lysosome-associated membrane protein-1; GFP::TM23;Brandizzi et al., 2002) and the PM ATPase fromNicotiana plumbaginifolia (PMA4::GFP; Lefebvreet al., 2004). Fig. 2 shows typical expression patternsfrom guard cells transfected with these fusion con-structs. Both fusion proteins tested displayed an evendistribution of the fluorescence in the PM without anyformation of radial stripes.

Together these results implicate that the radialdistribution of KAT1::GFP in the PM of transfectedguard cells is cell type and protein specific.

5

10

Fre

qu

en

cy

type 2

type 3

Distribution pattern of KAT1::GFP in guard cells is

independent of stomatal aperture and cytoskeleton

inhibitors

Investigations of the cytoskeleton in guard cells ofopen stomata revealed that both actin filaments andmicrotubules are radially distributed (Fukuda et al.,1998; Kim et al., 1995). In closed stomata the radialdistribution pattern disappears because actin filaments

A B

Fig. 2. Distribution of different GFP fusion proteins in guard

cells. (A) Maximum projection of guard cell expressing

GFP::TM23. (B) Maximum projection of guard cell expressing

PMA4::GFP. Scale bars: 10 mm.

and microtubules depolymerise. The distribution pat-tern of KAT1::GFP in transfected guard cells is at firstglance similar to the distribution of actin filaments andmicrotubules in open guard cells indicating that thecytoskeleton may be involved in the distribution ofKAT1::GFP. This would implicate that the radialstaining pattern can only be found in guard cells ofopen stomata.

However, analysis of the relationship between stoma-tal aperture and stripe formation revealed no correlationbetween the type of KAT1::GFP pattern and opening ofthe stomatal pore (Fig. 3). Fig. 3 shows the distributionof the stomatal aperture from type 1, type 2 and type 3cells. Stomatal aperture was determined by measuringthe inner width of the stomatal pore from maximumprojections of 157 guard cells. All distributions werefitted with a Gaussian function displaying a peak valueat 9.7, 9.8 and 10.4 mm for type 1, type 2 and type 3,respectively. Kolmogorov–Smirnov two-sample testsrevealed no significant difference between the threedistributions (a ¼ 0.05). This suggests that the distribu-tion pattern is unrelated to the stomatal aperture andthus independent of the cytoskeleton.

To further investigate the participation of thecytoskeleton in the formation of the radial distribu-tion of KAT1::GFP, we applied latrunculin B

0 8 12 16 20

0

Aperture [µm]

4

Fig. 3. Stripe formation is independent of stomatal aperture.

Distribution of stomatal aperture of guard cells displaying

either clear radial staining pattern (type 3), radial staining in

parts of cell (type 2) or no radial staining (type 1). Stripe

formation was categorised according to text (see also Fig. 1).

The distributions were each fit to a Gaussian function. The

peak values were 9.7, 9.8 and 10.4 mm for type 1, type 2 and

type 3, respectively. Distributions were not significantly

different (a ¼ 0.05, Kolmogorov–Smirnov two-sample test).

ARTICLE IN PRESS

A

B

C

D

Fig. 4. Filamentous structures of actin and microtubules are

destroyed by latrunculin B and propyzamide. (A) Actin

cytoskeleton in guard cells visualised by talin::YFP. (B) Same

guard cell as in (A) after 20min incubation in 10 mMlatrunculin B. (C) Microtubules in guard cell visualised by

TOR1::GFP. (D) Same guard cell as in (C) after 15min

incubation in 50 mM propyzamide. Note the complete disap-

pearance of smaller filamentous structures. Scale bars: 10mm.

A

C D

B

Fig. 5. Distribution of KAT1::GFP in guard cells is not

affected by cytoskeleton inhibitors. Maximum projection of

guard cells expressing KAT1::GFP. (A) and (C) Guard cells

expressing KAT1::GFP before addition of cytoskeleton

inhibitors. (B) KAT1::GFP-transfected guard cell 70min after

addition of 50 mM propyzamide. (D) KAT1::GFP-transfected

guard cell 140min after addition of 10mM latrunculin B. Scale

bar: 10 mm.

U. Homann et al. / European Journal of Cell Biology 86 (2007) 489–500494

(final concentration: 10 mM), an actin-depolymerisingagent and propyzamide (final concentration: 50 mM) as amicrotubule-destabilising drug. The effect of theseinhibitors on the cytoskeleton was tested in guard cellstransfected with talin::YFP (Kost et al., 1998) orTOR1::GFP (Buschmann et al., 2004) to visualise thedistribution of the actin or microtubules, respectively.Incubation of guard cells in10 mM latrunculin B for20min was sufficient to completely destroy the actinmeshwork (Figs. 4A and B). The radial distribution ofthe microtubule marker significantly changed alreadyafter 10min in 50 mM propyzamide with completedisappearance of smaller filamentous structures(Figs. 4C and D). This demonstrates that the arrange-ment of the cytoskeleton in guard cells is indeeddestroyed by the inhibitors applied.

However, neither the application of latrunculin B northe incubation in propyzamide had an appreciable effecton the radial distribution of KAT1::GFP. The stripedpattern remained unchanged even after prolongedtreatment for 70 and 140min in propyzamide orlatrunculin B, respectively (Figs. 5B and D). Togetherthese results demonstrate that the cytoskeleton is notinvolved in maintenance of the radial distribution ofKAT1::GFP.

Possible involvement of the cell wall in the radial

distribution of KAT1::GFP

In guard cells not only actin filaments and micro-tubules but also cellulose fibrils are radially distributed(Raschke, 1979). Fig. 5 shows Calcofluor White stainingof cellulose fibrils in the cell wall of guard cells andepidermal cells. Guard cells displayed a parallel andradial organisation of cellulose fibrils (Fig. 6A). Inepidermal cells such a pattern could not be observed,

and cellulose fibrils were found to be arranged in adisordered manner (Fig. 6B). The arrangement of thecellulose fibrils in guard cells shows the same radialorientation as KAT1::GFP stripes in the PM, indicatinga causal relation between the two staining patterns.

Further indications for the participation of the cellwall in formation of the radial distribution ofKAT1::GFP can be derived from images of transfectedguard cell protoplast. In contrast to turgid guard cellsprotoplasts lacked the radial fluorescence pattern in thePM (Hurst et al., 2004).

To further investigate the participation of the cell wallin the distribution of KAT1::GFP, we incubatedepidermal peels in strong hypertonic solution todecrease turgor pressure of guard cells. Under thiscondition the PM is no longer tightly pressed against thecell wall. In order to avoid infolding of the PM weincreased the osmolarity of the bath solution until thePM started to retract from the cell wall only at the celltips. At this stage there was no visible detachment of thePM from the cell wall in the rest of the cell. The changesin staining pattern from three guard cells before andafter hypertonic treatment are displayed in Figs. 7A–F.

ARTICLE IN PRESS

A B

C

Fig. 6. Radial distribution of cellulose fibrils in guard cells.

Confocal images of cells stained with Calcofluor White.

(A) Partial view from a maximum projection of four

consecutive stacks of a guard cell showing radial distribution

of cellulose fibrils. (B) Partial view from a guard cell (left) and

epidermal cell (right) showing radial and random distribution

of cellulose fibrils, respectively. (C) Overview of the epidermis

utilizing Normarski optics. The selected area corresponds to

image (B). Scale bars: 10 mm.

G

info

rmation c

onte

nt of tr

ansfo

rme

d im

age

noise intensity

0.2

0.3

0.4

0.1 0.15 0.2 0.25

(1)

(2)

B

C

D

E

A

F

A

B

C

D

E

F

Fig. 7. Loss of radial pattern of KAT1::GFP after hypertonic

treatment. (A), (C) and (E) Partial view from maximum

projections of four consecutive stacks of guard cells expressing

KAT1::GFP before hypertonic treatment. (B), (D) and (F)

Partial view from guard cells shown in (A), (C) and (E),

respectively, 3 to 8min after hypertonic treatment. (G) Result

from mathematical analysis of images shown in (A)–(F);

reference grid has been obtained by analysing simulated data;

values above and below dashed reference line correspond to

stripe and random formation, respectively, of KAT1::GFP;

(1) and (2) indicate increasing noise and dottiness, respectively,

in reference data. Scale bar: 10 mm

U. Homann et al. / European Journal of Cell Biology 86 (2007) 489–500 495

Hypertonic treatment resulted in an immediate loss ofthe radial pattern in the PM (Figs. 7B, D and F).

To quantitatively analyse the systematic features ofthe distribution patterns in guard cells we havedeveloped an image analysis algorithm (see Materialsand methods). The radial organisation seems to exist ingradually varying degrees, ranging from cases, where itis a very prominent feature of the image, to cases, whereit is almost completely masked by other features andhard to distinguish from randomly distributed fluores-cent spots. This fact particularly emphasises the need fora quantitative algorithmic look at these patterns. Thisalgorithm allows distinguishing cells with a radialstaining from cells with a more random distribution ofKAT1::GFP. Results from analysis of the cells shown inFigs. 7A–F are represented in Fig. 7G (circles markedwith characters). The reference grid in Fig. 7G has beenobtained by analysing simulated data, where noisecontent and the stripe-to-dot ratio have been varied(see Materials and methods). Note that the length scales(i.e. the ‘‘thickness’’ of typical stripes) determines theabsolute position of this grid. This additional depen-dence on other (but constant) features than stripe-to-dot-ratio and noise has two important consequences forour interpretation of the data: (1) for a given experi-mental data set, individual reference grids have to besimulated and (2) only the relative position in the plane

for images within a single experiment can be interpreted.Thus, for the further examples given in Fig. 8 only therelevant region of the plane is shown without anadditional reference grid.

The guard cells shown in Figs. 7A, C and E are wellabove the reference line in this quantitative analysis and,therefore, represent stripe formation whereas Figs. 7B,D and F are clearly below this reference line and,therefore, lack this feature of radial organisation. Thisquantitative analysis confirms that loss of turgor isassociated with a decrease in the degree of stripe-likedistribution of KAT1::GFP.

ARTICLE IN PRESS

info

rmation c

onte

nt

of tr

ansfo

rmed im

age

noise intensity

0.15 0.2 0.25

A

C

B0.3

0.4

0.35

0.45

D

A B C

Fig. 8. Recurrence of radial pattern of KAT1::GFP upon

hypotonic treatment. (A) Partial view from maximum projec-

tions of four consecutive stacks of guard cell expressing

KAT1::GFP before hypertonic treatment. (B) Partial view of

guard cell shown in (A) 4min after hypertonic treatment.

(C) Partial view of guard cell shown in (A) and (B) 15min after

hypotonic treatment. (D) Result from mathematical analysis

of images shown in (A)–(C); lower information content

corresponds to loss of ordered stripe formation of

KAT1::GFP. Scale bar: 10 mm.

U. Homann et al. / European Journal of Cell Biology 86 (2007) 489–500496

To test for the reversibility of the loss in stripeformation we replaced the hypertonic solution by thehypotonic standard bath solution. Figs. 8A–C show partof the cortical section of a KAT1::GFP-transfectedguard cell before and after hypertonic treatment, and15min after replacing the hypertonic solution by thestandard bath solution. The radial fluorescent patternwhich vanished upon loss of turgor (Fig. 8B) started torecur when high turgor pressure was restored instandard solution (Fig. 8C). Quantitative analysis ofthe images confirmed that the decrease in radialorganisation observed during hypertonic treatment isreversible (Fig. 8D).

Together the results demonstrate that the formationof the radial fluorescence distribution in guard cells isassociated with a close contact of the PM to thesurrounding cell wall.

Discussion

In this study we heterologously expressed the K+

inward rectifier KAT1 from Arabidopsis thaliana fused

to the green fluorescent protein (GFP) in V. faba guardcells, and observed a cell type- and protein-specificfluorescent pattern. In most guard cells KAT1::GFP wasorganised in distinct radial stripes in the PM. Our resultslead to the hypothesis that this staining pattern is due tointeraction of KAT1 with the cell wall of guard cells.

Distinct KAT1::GFP pattern in the PM of guard

cells does not refer to overexpression artefact

Ever since GFP in fusion with functional proteins wasapplied in plant cell biology, artefacts resulting fromoverexpression have been a critical point to take intoaccount while interpreting such data. This is especiallytrue for the expression of constructs under a strongpromoter like the 35S promoter used in our studies. Atfirst sight the observed KAT1::GFP distribution maythus be explained by a non-physiological association offusion proteins into clusters mediated by GFP. How-ever, our results from two other GFP fusion proteinsexpressed in guard cells under the same strong promotercontradict this hypothesis. When fused to GFP neitherthe single-pass transmembrane domain TM23 fromLAMP1 nor the functional protein PM-ATPase showedthis striped fluorescent pattern in transfected guard cells.In addition KAT1::GFP was always found in an evenlydotted distribution in transfected epidermal cells. GFPfluorescence intensity per PM area did not differ withcell type or expressed fusion protein. Hence, varyingprotein amounts cannot account for the differentexpression patterns observed (for the quantificationprocedure see Meckel et al., 2007). We thereforeconclude that the expression pattern of KAT1::GFP iscell type and protein specific. In addition, guard cellprotoplasts transfected with KAT1::GFP never showeda striped pattern of KAT1::GFP. Taken together theresults demonstrate that the striped orientation ofKAT1::GFP is clearly not the result of an overexpres-sion artefact, but rather points to a distinct mechanismthat determines the distribution of KAT1 in the PM ofturgid guard cells.

Stomatal aperture and the cytoskeleton are not

involved in pattern formation

From animal cells, a PM-cytoskeleton-ECM contin-uum is proposed in which various membrane proteinsare functionally included (Gumbiner, 1996). The role ofthe cytoskeleton in guard cell function is somewhatcontroversial. Marcus et al. (2001) demonstrated adiurnal cycle in microtubule formation in guard cellsthat was dependent on stomata aperture. Furthermorethis cycle could be blocked by application of propyza-mide, an inhibitor of microtubule formation (Marcuset al., 2001). In contrast, Assmann and Baskin (1998)

ARTICLE IN PRESSU. Homann et al. / European Journal of Cell Biology 86 (2007) 489–500 497

found no role for microtubules in guard cell functioning.Also actin filaments seem to have a role in guard cellfunction. Their distribution changes from a more radialpattern in open stomata to a less organised distributionin closed stomata (Hwang et al., 1997). Hwang et al.(1997) also propose that actin is involved in theregulation of K+ channels in guard cells. Despitethe striking similarity of the KAT1::GFP distributionto the arrangement of the cytoskeleton (microfilamentsand microtubules) observed in this study, we found noevidence for the involvement of the cytoskeleton in thedistribution of KAT1. The staining pattern was notdependent on the stomatal aperture. Additionallyneither the application of propyzamide as an inhibitorof microtubule formation nor latrunculin B as aninhibitor of microfilament formation could abolish theradial pattern of KAT1::GFP. The concentrations oflatrunculin B and propyzamide used in our study werein the same range of what has been shown to destroymicrofilaments in inner cortex cells of maize root apices(Baluska et al., 2004) and cortical microtubules inA. thaliana (Naoi and Hashimoto, 2004), respectively.Furthermore, sequence predictions from KAT1 do notreveal an ankyrin motif which would allow directbinding of actin to KAT1 (Nakamura et al., 1995). Wetherefore conclude that neither microtubules nor actinmicrofilaments are direct participants in the formationof the striped pattern of KAT1::GFP in the PM ofguard cells.

Pattern formation is related to a KAT1–cell wall

interaction

Our analysis of guard cells suggests that a tightcontact between the PM and the cell wall is essential forthe striped distribution of KAT1. When turgid guardcells were incubated in strong hypertonic solutionresulting in a loss of turgor and consequently the PMbeing no longer pressed against the cell wall the radialfluorescent pattern disappeared immediately and dis-solved into randomly distributed dots. Restoration ofthe turgor pressure by exchange of the hypertonicsolution with hypotonic bath solution led to thereappearance of the radial pattern. Using a mathemat-ical algorithm the pattern of fluorescence before andafter loss of turgor could clearly be separated by ‘‘gradeof their homogeneity’’ (with the fluorescence being moreordered before loss of turgor independent of noise andbackground). Together with the observation thatKAT1::GFP-expressing guard cell protoplasts displayedno radial staining pattern (Fig. 1B; Hurst et al., 2004)this demonstrates that the formation of stripes dependson a close contact of KAT1 with the cell wall. The radialarrangement of KAT1 is very similar to the organisationof cellulose microfibrils in the guard cell wall. In

epidermal cells cellulose fibrils displayed no parallelpattern but showed a rather disordered arrangement.The observed cell type-specific difference in the fluor-escent pattern of KAT1::GFP in epidermal and guardcells (random versus radial distribution) may thus be theresult from the disordered or radial arrangement ofcellulose fibrils in the respective cell type. We thereforesuggest that KAT1 is directly or indirectly associatedwith the cellulose fibrils in the cell wall. The stripesformed by KAT1::GFP show the same orientation asthe cellulose fibrils but are much broader than singlecellulose fibrils which have a diameter of only about10 nm. This can be explained by an interaction of severalparallel arranged cellulose fibrils with KAT1 or by therestriction of KAT1 to the wider gaps found betweenbundles of cellulose fibrils.

The observation that about 30% of transfected guardcells showed no stripe formation supports the hypothesisthat KAT1 is not linked directly to the cellulose fibrilsand that this link is not obligatory. A direct binding ofKAT1 to the cellulose should have led to a radialdistribution in all transfected guard cells because allfully developed guard cells show a radial distribution ofthe cellulose microfibrils. The KAT1–cell wall associa-tion is thus most likely mediated by additional protein(s)which are apparently not active in all cells.

In animal cell PM proteins, such as integrins, linkintracellular proteins to the ECM. Recently, it has beenshown that integrins are also physically and functionallyconnected to some classes of ion channels and that thisassociation is of general importance for cell physiology(for review see (Arcangeli and Becchetti, 2006)). For theGIRK K+ channel the link between the channel andintegrins most likely occurs via an RGD motif in theextracellular loop (McPhee et al., 1998). Other K+

channels do not contain an RGD motif. Some of thesechannels may interact with integrins via their N-terminaldomain (Cherubini et al., 2005).

Even though only few homologues of classicaladhesion molecules like integrins have been identifiedin plants, for a number of plant PM proteins links to thecell wall have been suggested on the basis of theirmolecular interaction with cell wall carbohydrates.Among these are the membrane intrinsic cellulosesynthase complex and cell wall-associated kinases(WAK), and further arabinogalactan proteins (AGP),which are bound to the outer leaflet of the PM via a GPIanchor (Kohorn, 2000). Also, the formation of Hechtianstrands during plasmolysis shows that attachment sitesbetween the PM and the cell wall exist. These attach-ment sites cannot be restricted to plasmodesmata sinceguard cells that in general lack plasmodesmata alsodevelop Hechtian strands (Oparka et al., 1996). How-ever, the nature of these attachment sites is not clear.Results from Lang et al. (2004) demonstrated thatattachment sites of Hechtian strands are lost during

ARTICLE IN PRESS

plasma membranecellulose

microfibril

R1R2

R3 R1

atmospheric

pressure = 0.1 MPa

force

turgor pressure =

up to 4.5 MPa

Fig. 9. Model for ion accessibility at plasma membrane–cell

wall interface of turgid guard cells. At sites where the PM is

firmly pressed against the cellulose fibrils the resistance for ion

movement (R2) is high; in between cellulose fibrils the

resistance for ion movement (R3) is low and ions can move

freely. R1 corresponds to resistance of the plasma membrane.

U. Homann et al. / European Journal of Cell Biology 86 (2007) 489–500498

digestion of cellulose which implicates that they areformed by a tight connection between PM proteins andcellulose. Likewise, digestion of cellulose diminished thespatial pattern of KAT1::GFP (Hurst et al., 2004). Wesuggest that some of the KAT1 channels may becontributing to these attachment sites. This is alsosupported by the observation that the proposedKAT1–cell wall association and the formation ofHechtian strands are both independent of the cytoske-leton (Lang-Pauluzzi, 2000; Lang-Pauluzzi andGunning, 2000).

In animal cells there is evidence that the interactionbetween ion channels and integrins is accompanied bythe formation of macromolecular complexes that arelocated in microdomains in the PM (Cherubini et al.,2005). Recently, Sutter et al. (2006) analysed thedistribution of KAT1 transiently expressed in tobaccoleaf tissue. In tobacco epidermal cells KAT1 waslocalised in small microdomains in the PM similar towhat we observed in epidermal cells of Vicia faba. Usingphotoactivatable GFP Sutter et al. (2006) were able toshow that KAT1 was largely immobile in the PM. Theysuggest that SNAREs which are known to be involvedin vesicle fusion may also participate in microdomainformation and mobility of KAT1 in the PM ofepidermal cells. Our results show that the radialdistribution of KAT1 but not the microdomain forma-tion was dependent on a close contact between the PMand the cell wall. This implies that microdomainformation and radial distribution of KAT1 are twoseparate processes. The protein(s) that mediate theassociation of KAT1 with cell wall components yetremain to be determined.

Possible physiological relevance of the distinct

localisation of KAT1 in the PM

So far we can only speculate on the physiologicalimportance of the protein- and cell type-specificdistribution of KAT1 in the PM of guard cells. Rapidfreezing methods on plant cells showed that the distancebetween the cell membrane and the cell wall is smallerthan previously observed. More recent electron micro-graphs revealed that the PM is appressed against the cellwall, thus they are apparently in tight contact (Roberts,1990). This is particularly true for guard cells as in thisspecialised cell type the turgor is extremely high(4–5MPa; Franks et al., 2001; Raschke, 1979). Sincethe cell wall serves as an external K+ store the tightcontact between the cell wall and the PM may affect theaccessibility of K+ for ion channels in the PM. Wetherefore propose the following model for the physio-logical role of the observed K+ channel distribution inthe PM (Fig. 9). In membrane areas where the PM ispressed tightly to the cellulose fibrils, the high resistance

for ion flow reduces K+ fluxes, whereas in PM areasthat reach into the intermediate space of the cellulosefibrils, K+ can move freely and therefore resistance forion flow is low. Hence in these areas the ion accessibilityfor K+ channels is sufficient for guard cell function. TheK+ inward rectifier in guard cells KAT1 plays a key rolein K+ uptake during stomatal movement. The radialdistribution may reflect the arrangement of KAT1 in theintermediate areas of cellulose fibrils and the attachmentto the fibrils would assure the stable localisation of theprotein in the PM.

However, electrophysiological measurements on pro-toplasts show that the cell wall link is not essential forKAT1 function. Gutknecht et al. (1978) proposed aturgor sensor in the PM of guard cells. Considering thetight association of KAT1 with the cell wall found in ourstudy, KAT1 may participate in turgor sensing. Thisunderlines the physiological importance of an intimateinteraction between PM and cell wall. The identificationof the molecular nature of the link between KAT1 andthe cell wall is therefore of great importance to furtherdissect stomatal functioning.

Acknowledgements

The authors would like to thank Prof. G. Thiel fordiscussion and helpful comments on the manuscript.We are grateful to N. Paris, M. Boutry, B. Kost andA. R. Schaffner for committing the constructs with thefusion proteins and Susanne Liebe for assistance withthe CLSM at Leica, Bensheim, Germany. The work wassupported by Grants of the Deutsche Forschungsge-meinschaft to U. Homann (SPP 1108 HO-2046/3-2 andHO-2046/5-2).

ARTICLE IN PRESSU. Homann et al. / European Journal of Cell Biology 86 (2007) 489–500 499

References

Arcangeli, A., Becchetti, A., 2006. Complex functional

interaction between integrin receptors and ion channels.

Trends Cell Biol. 16, 631–639.

Arnold, M., Cavalcanti-Adam, E.A., Glass, R., Blummel, J.,

Eck, W., Kantlehner, M., Kessler, H., Spatz, J.P., 2004.

Activation of integrin function by nanopatterned adhesive

interfaces. Chem. Phys. Chem. 19, 383–388.

Assmann, S.M., Baskin, T.I., 1998. The function of guard cells

does not require an intact array of cortical microtubules. J.

Exp. Bot. 49, 163–170.

Baluska, F., Samaj, J., Hlavacka, A., Kendrick-Jones, J.,

Volkmann, D., 2004. Actin-dependent fluid-phase endocy-

tosis in inner cortex cells of maize root apices. J. Exp. Bot.

55, 463–473.

Barritt, G., Rychkov, G., 2005. TRPs as mechanosensitive

channels. Nat. Cell Biol. 7, 105–107.

Borner, G.H.H., Sherrier, D.J., Weimar, T., Michaelson, L.V.,

Hawkins, N.D., MacAskill, A., Napier, J.A., Beale, M.H.,

Lilley, K.S., Dupree, P., 2005. Analysis of detergent-

resistant membranes in Arabidopsis: evidence for plasma

membrane lipid rafts. Plant Physiol. 137, 104–116.

Brandizzi, F., Frangne, N., Marc-Martin, S., Hawes, C.,

Neuhaus, J.-M., Paris, N., 2002. The destination for single-

pass membrane proteins is influenced markedly by the

length of the hydrophobic domain. Plant Cell 14,

1077–1092.

Brown, D., Rose, J.K., 1992. Sorting of GPI-anchored

proteins to glycolipid-enriched membrane subdomains

during transport to the apical cell surface. Cell 68, 533–544.

Buschmann, H., Fabri, C.O., Hauptmann, M., Hutzler, P.,

Laux, T., Lloyd, C.W., Schaffner, A.R., 2004. Helical

growth of the Arabidopsis mutant tortifolia1 reveals a

plant-specific microtubule-associated protein. Curr. Biol.

14, 1515–1521.

Canut, H., Carrasco, A., Galaud, J.-P., Cassan, C., Bouyssou,

H., Vita, N., Ferrara, P., Pont-Lezica, R., 1998. High

affinity RGD-binding sites at the plasma membrane of

Arabidopsis thaliana link the cell wall. Plant J. 16, 63–71.

Cherubini, A., Hofmann, G., Pillozzi, S., Guasti, L., Crociani,

O., Cilia, E., Di Stefano, P., Degani, S., Balzi, M., Olivotto,

M., Wanke, E., Becchetti, A., Defilippi, P., Wymore, R.,

Arcangeli, A., 2005. Human ether-a-go-go-related gene 1

channels are physically linked to beta1 integrins and

modulate adhesion-dependent signaling. Mol. Biol. Cell

16, 2972–2983.

Deutsch, C., 2002. Potassium channel ontogeny. Annu. Rev.

Physiol. 64, 19–46.

Dietrich, C., Yang, B., Fujiwara, T., Kusumi, A., Jacobson,

K., 2002. Relationship of lipid rafts to transient confine-

ment zones detected by single particle tracking. Biophys. J.

82, 274–284.

D’Souza, S.E., Ginsberg, M.H., Plow, E.F., 1991. Arginyl-

glycyl-aspartic acid (RGD): a cell adhesion motif. Trends

Biochem. Sci. 16, 246–250.

Faik, A., Laboure, A.M., Gulino, D., Mandaron, P., Falconet,

C., 1998. A plant surface protein sharing structural

properties with animal integrins. Eur. J. Biochem. 253,

552–559.

Franks, P.J., Buckley, T.N., Shope, J.C., Mott, K.A., 2001.

Guard cell volume and pressure measured concurrently by

confocal microscopy and the cell pressure probe. Plant

Physiol. 125, 1577–1584.

Fukuda, M., Hasezawa, S., Asai, N., Nakajima, N., Kondo,

N., 1998. Dynamic organization of microtubules in guard

cells of Vicia faba L with diurnal cycle. Plant Cell Physiol.

39, 80–86.

Gardiner, J.C., Taylor, N.G., Turner, S.R., 2003. Control of

cellulose synthase complex localization in developing

xylem. Plant Cell 15, 1740–1748.

Gens, J.S., Reuzeau, C., Doolittle, K.W., McNally, J.G.,

Pickard, B.G., 1996. Covisualization by computational

optical-sectioning microscopy of integrin and associated

proteins at the cell membrane of living onion protoplasts.

Protoplasma 194, 215–230.

Ghazi, A., Berrier, C., Ajouz, B., Besnard, M., 1998.

Mechanosensitive ion channels and their mode of activa-

tion. Biochimie 80, 357–362.

Gumbiner, B.M., 1996. Cell adhesion: the molecular basis of

tissue architecture and morphogenesis. Cell 84, 345–357.

Gutknecht, J., Hastings, D.F., Bisson, U.A., 1978. Ion

transport and turgor pressure regulation in giant algal

cells. In: Giebisch, G.etal. (Ed.), Membrane Transport in

Biology, Transport Across Multi-Membrane Systems, vol.

3. Springer, Berlin–Heidelberg–New York, pp. 125–137.

Haseloff, J., Siemering, K.R., Prasher, D.C., Hodge, S., 1997.

Removal of a cryptic intron and subcellular localization of

green fluorescent protein are required to mark transgenic

Arabidopsis plants brightly. Proc. Nat Acad. Sci. USA 94,

2122–2127.

Hawes, C.R., Saint-Jore, C., Martin, B., Zheng, H.Q., 2001.

ER confirmed as the location of mystery organelles in

Arabidopsis plants expressing GFP. Trends Plant Sci. 6,

245–246.

Hurst, A.C., Meckel, T., Tayefeh, S., Thiel, G., Homann, U.,

2004. Trafficking of the plant potassium inward rectifier

KAT1 in guard cell protoplasts of Vicia faba. Plant J. 37,

391–397.

Hutt, M.-T., 2001. Datenanalyse in der Biologie: Eine

Einfuhrung in Methoden der nichtlinearen Dynamik,

fraktalen Geometrie und Informationstheorie. Springer,

Heidelberg, pp. 194–217.

Hwang, J.U., Suh, S., Yi, H., Kim, J., Lee, Y., 1997. Actin

filaments modulate both stomatal opening and inward K+-

channel activities in guard cells of Vicia faba L. Plant

Physiol. 115, 335–342.

Kim, M., Hepler, P.K., Eun, S.O., Ha, K.S., Lee, Y., 1995.

Actin filaments in mature guard cells are radially distrib-

uted and involved in stomatal movement. Plant Physiol.

109, 1077–1084.

Kohorn, B.D., 2000. Plasma membrane-cell wall contacts.

Plant Physiol. 124, 31–38.

Kost, B., Spielhofer, P., Chua, N.H., 1998. A GFP-mouse talin

fusion protein labels plant actin filaments in vivo and

visualizes the actin cytoskeleton in growing pollen tubes.

Plant J. 16, 393–401.

Lang, I., Barton, D.A., Overall, R.L., 2004. Membrane-wall

attachments in plasmolysed plant cells. Protoplasma 224,

231–243.

ARTICLE IN PRESSU. Homann et al. / European Journal of Cell Biology 86 (2007) 489–500500

Lang-Pauluzzi, I., 2000. The behaviour of the plasma

membrane during plasmolysis: a study by UV microscopy.

J. Microsc. 198, 188–198.

Lang-Pauluzzi, I., Gunning, B.E.S., 2000. A plasmolytic cycle:

the fate of cytoskeletal elements. Protoplasma 212,

174–185.

Lefebvre, B., Batoko, H., Duby, G., Boutry, M., 2004.

Targeting of a Nicotiana plumbaginifolia H+-ATPase to

the plasma membrane is not by default and requires

cytosolic structural determinants. Plant Cell 16, 1772–1789.

Marcus, A.I., Moore, R.C., Cyr, R.J., 2001. The role of

microtubules in guard cell function. Plant Physiol. 125,

387–395.

Martens, J.R., O’Connell, K., Tamkun, M., 2004. Targeting of

ion channels to membrane microdomains: localization of

KV channels to lipid rafts. Trends Pharmacol. Sci. 25,

16–21.

McPhee, J.C., Dang, Y.L., Davidson, N., Lester, H.A., 1998.

Evidence for a functional interaction between integrins and

G protein-activated inward rectifier K+ channels. J. Biol.

Chem. 273, 34696–34702.

Meckel, T., Hurst, A.C., Thiel, G., Homann, U., 2004.

Endocytosis against high turgor: intact turgid guard cells

of Vicia faba constitutively endocytose fluorescently

labelled plasma membrane and GFP-tagged K+-channel

KAT1. Plant J. 39, 182–193.

Meckel, T., Gall, L., Semrau, S., Homann, U., Thiel, G., 2007.

Guard cells elongate: relationship of volume and

surface area during stomatal movement. Biophys. J. 92,

1072–1080.

Mongrand, S., Morel, J., Laroche, J., Claverol, S., Carde, J.P.,

Hartmann, M.A., Bonneu, M., Simon-Plas, F., Lessire, R.,

Bessoule, J.J., 2004. Lipid rafts in higher plant cells:

purification and characterization of Triton X-100-insoluble

microdomains from tobacco plasma membrane. J. Biol.

Chem. 279, 36277–36286.

Nakamura, R.L., McKendree Jr., W.L., Hirsch, R.E.,

Sedbrook, J.C., Caber, R.F., Sussman, M.R., 1995.

Expression of an Arabidopsis potassium channel gene in

guard cells. Plant Physiol. 109, 371–374.

Naoi, K., Hashimoto, T., 2004. A semidominant mutation in

an Arabidopsis mitogen-activated protein kinase phospha-

tase-like gene compromises cortical microtubule organiza-

tion. Plant Cell 16, 1841–1853.

Oparka, K.J., Prior, D.A.M., Crawford, J.W., 1996. Mem-

brane conversation during plasmolysis. In: Smallwood, M.,

Knox, J.P., Bowles, D.J. (Eds.), Membranes: Specialized

Functions in Plants. Bios Scientific Publishers, Oxford,

pp. 39–56.

Pellenc, D., Schmitt, E., Gallet, O., 2004. Purification of a

plant cell wall fibronectin-like adhesion protein involved in

plant response to salt stress. Protein Expr. Purif. 34,

208–214.

Raschke, K., 1979. Movement of stomata. In: Haupt, W.,

Feinleb, M.E. (Eds.), Encyclopedia of Plant Physiology.

Physiology of Movements, vol. 7. Springer, Berlin,

pp. 382–441.

Ritchie, K., Kusumi, A., 2004. Role of the membrane skeleton

in creation of microdomains. Subcell. Biochem. 37,

233–245.

Roberts, K., 1990. Structures at the plant cell surface. Curr.

Opin. Cell Biol. 2, 920–928.

Ruoslahti, E., 1996. RGD and other recognition sequences for

integrins. Annu. Rev. Cell Dev. Biol. 12, 697–715.

Staiger, C.J., Lloyd, C.W., 1991. The plant cytoskeleton. Curr.

Opin. Cell Biol. 3, 33–42.

Sutter, J.U., Campanoni, P., Tyrrell, M., Blatt, M., 2006.

Selective mobility and sensitivity to SNAREs is exhibited

by the Arabidopsis KAT1 K+ channel at the plasma

membrane. Plant Cell 18, 935–954.

Tester, M., 1990. Plant ion channels: whole-cell and single-

channel studies. New Phytol. 114, 305–340.