This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: 18/4/935    most recent tpc.105.038950v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (18) Citing Articles via Google Scholar Google Scholar Articles by Sutter, J.-U. Articles by Blatt, M. R. Search for Related Content PubMed PubMed Citation Articles by Sutter, J.-U. Articles by Blatt, M. R. Agricola Articles by Sutter, J.-U. Articles by Blatt, M. R.

Selective Mobility and Sensitivity to SNAREs Is Exhibited by the Arabidopsis KAT1 K⁺ Channel at the Plasma Membrane^[W]

Laboratory of Plant Physiology and Biophysics, Institute of Biomedical and Life Sciences–Plant Sciences, University of Glasgow, Glasgow G12 8QQ, United Kingdom

¹ To whom correspondence should be addressed. E-mail m.blatt{at}bio.gla.ac.uk; fax 44-0141-330-4447.

	ABSTRACT

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Recent findings indicate that proteins in the SNARE superfamily are essential for cell signaling, in addition to facilitating vesicle traffic in plant cell homeostasis, growth, and development. We previously identified SNAREs SYP121/Syr1 from tobacco (Nicotiana tabacum) and the Arabidopsis thaliana homolog SYP121 associated with abscisic acid and drought stress. Disrupting tobacco SYP121 function by expressing a dominant-negative Sp2 fragment had severe effects on growth, development, and traffic to the plasma membrane, and it blocked K⁺ and Cl^– channel responses to abscisic acid in guard cells. These observations raise questions about SNARE control in exocytosis and endocytosis of ion channel proteins and their organization within the plane of the membrane. We have used a dual, in vivo tagging strategy with a photoactivatable green fluorescent protein and externally exposed hemagglutinin epitopes to monitor the distribution and trafficking dynamics of the KAT1 K⁺ channel transiently expressed in tobacco leaves. KAT1 is localized to the plasma membrane within positionally stable microdomains of 0.5 µm in diameter; delivery of the K⁺ channel, but not of the PMA2 H⁺-ATPase, to the plasma membrane is suppressed by Sp2 fragments of tobacco and Arabidopsis SYP121, and Sp2 expression leads to profound changes in KAT1 distribution and mobility within the plane of the plasma membrane. These results offer direct evidence for SNARE-mediated traffic of the K⁺ channel and a role in its distribution within subdomains of the plasma membrane, and they implicate a role for SNAREs in positional anchoring of the K⁺ channel protein.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Eukaryotic cells maintain a traffic of vesicles to shuttle membrane material, proteins, and soluble cargo between endomembrane compartments, the plasma membrane, and the extracellular space. Vesicles are formed by budding and constriction at the formative membrane surface, and their delivery is achieved by fusion and intercalation with the lipid bilayer of the target membrane (Pratelli et al., 2004; Surpin and Raikhel, 2004). These processes sustain membrane turnover and thereby contribute to cellular homeostasis, differentiation, and growth. Vesicle traffic contributes to neurotransmitter release and nervous signal transmission across the synaptic junctions of nerves, to cell wall delivery and budding in yeast (Chen and Scheller, 2001; Jahn et al., 2003), and to cell polarity, growth, and development in plants (Blatt and Thiel, 2003).

Central to the process of vesicle fusion is a family of membrane trafficking proteins, Soluble NSF (N-ethylmaleimide–sensitive factor) Attachment Protein Receptors (SNAREs), that are conserved among all eukaryotes (Jahn et al., 2003; Pratelli et al., 2004; Surpin and Raikhel, 2004). Complementary SNAREs, identified by their core residues (either Arg or Glu), are localized to different membrane compartments and interact to form a tetrameric bundle of coiled helices that draws the membrane surfaces together and facilitates fusion. In reconstituted membrane preparations, this complex forms a minimal set of proteins required for fusion (Parlati et al., 1999; Weber et al., 1998; Hu et al., 2003). In vivo, other cytoplasmic factors, including the N-ethylmaleimide–sensitive factor Sec1 and its homologs, control interaction between the SNARE elements (Jahn et al., 2003; Pratelli et al., 2004; Surpin and Raikhel, 2004). Nonetheless, the combinatorial specificity of SNARE interactions is thought to contribute significantly to membrane recognition and vesicle targeting (Paumet et al., 2004; Varlamov et al., 2004). Of the core SNAREs in plants, the syntaxins (members of the Q-SNARE subclass; Pratelli et al., 2004) are the best characterized and, functionally, most intriguing. In Arabidopsis thaliana, the large number of syntaxin-like SNAREs far outnumber the identifiable membrane compartments, and there are hints of an associated functional diversity that extends beyond the canonical roles in membrane targeting and vesicle fusion (Pratelli et al., 2004).

Additionally, a growing body of kinetic and physiological data is available for vesicle fusion, at least at the plasma membrane where physical access is possible in vivo. In plants, as in mammalian tissues, exocytotic and endocytotic events at the plasma membrane have been identified with stepwise changes in capacitance that accompany the increase or decrease of membrane surface area during vesicle membrane fusion and removal, respectively (Thiel and Battey, 1998; Blatt and Thiel, 2003). These changes in capacitance are consistent in size with the predicted vesicle dimensions derived from ultrastructural studies (Picton and Steer, 1983; Phillips et al., 1988) and from imaging studies using fluorescent styryl dyes to label internalized membrane (Meckel et al., 2004). Factors shown to affect vesicle traffic in plants include cytosol-free Ca²⁺ concentration ([Ca²⁺]_i), guanosine nucleotides (Homann and Tester, 1997; Carroll et al., 1998), and osmotic changes (Kubitscheck et al., 2000). Furthermore, evidence for [Ca²⁺]_i-dependent and -independent exocytotic pathways underscores the complexity of secretory processing that must occur in parallel within individual cells (Homann and Tester, 1997; Sutter et al., 2000).

By contrast, information remains scarce that bears on partitioning of specific membrane proteins to the plasma membrane, much less on roles for SNAREs in these processes. SNARE-related vesicle traffic has been implicated in the spatial distribution of the auxin efflux carrier Pin1 and its sensitivity to the ARF-GEF inhibitor brefeldin A (Steinmann et al., 1999; Geldner et al., 2001) that disrupts Golgi structure and trafficking (Nebenfuhr et al., 2002). Traffic of the H⁺-ATPase has also been suggested to underpin auxin-stimulated H⁺ extrusion and parallel increases in H⁺-ATPase protein that take place over a similar time scale (10 min) at the plasma membrane (Hager et al., 1991). Details of any mechanism is lacking in this case, although recent evidence points to a concerted targeting to the plasma membrane (Lefebvre et al., 2004). Finally, fast and reversible exchange of K⁺ channels occurs in guard cells during osmotically driven changes in cell volume (Homann and Thiel, 2002). Again, details of the molecular mechanism(s) are lacking. However, this exchange appears nonselective among K⁺ channel species and therefore differs fundamentally from the physiological regulation of channel activity during stomatal movements (Schroeder et al., 2001; Dreyer et al., 2004); instead, it has been suggested to maintain overall channel density when the membrane surface area changes (Hurst et al., 2004).

Previously, we identified a Q-SNARE, SYP121/Syr1 from tobacco (Nicotiana tabacum), associated with abscisic acid and drought stress (Leyman et al., 1999). SYP121 is localized to the plasma membrane, and although originally cloned from a leaf cDNA library, the protein is expressed throughout the plant (Leyman et al., 2000). We found that disrupting SYP121 function, by expressing a dominant-negative, cytosolic (so-called) Sp2 fragment in vivo, had severe effects on growth, tissue development, and traffic to the plasma membrane (Geelen et al., 2002). Intriguingly, SYP121 was cleaved by the Clostridium neurotoxin BotN/C, and both the neurotoxin and the Sp2 fragment blocked K⁺ and Cl^– channel responses to abscisic acid when loaded directly into guard cells (Leyman et al., 1999). The observations indicated a role for the Q-SNARE in traffic to the plasma membrane. They also raised questions about additional functions in regulating ion channel activities, either through selective exocytosis and endocytosis of the channel proteins or through direct signaling associations and regulation within the plane of the membrane.

To explore these questions further, we tagged the KAT1 K⁺ channel of Arabidopsis with hemagglutinin (HA) epitopes and a photoactivatable green fluorescent protein (paGFP) construct to monitor channel distribution and trafficking dynamics and to relate its characteristics to those of the PMA2 H⁺-ATPase in vivo when expressed in tobacco leaves. We report here that the KAT1 K⁺ channel, like the H⁺-ATPase, is localized primarily to the plasma membrane, and both proteins appear positionally stable within the membrane. However, unlike PMA2, KAT1 distribution is localized in discrete puncta. Furthermore, only KAT1 mobility and delivery to the plasma membrane are affected by the Sp2 fragments of tobacco and Arabidopsis SYP121, although the traffic of both proteins to the plasma membrane is disrupted by a dominant-negative mutant of Rab1b that affects exchange between the endoplasmic reticulum and the Golgi apparatus. These and additional results offer primary evidence for selective SNARE-mediated traffic of the K⁺ channel to the plasma membrane, they support a view that distribution of the channel is targeted to subdomains of the plasma membrane, and they implicate a role for the SNAREs in these processes as well as in positional anchoring of the K⁺ channel protein.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Transient Expression of Epitope-Tagged KAT1 Yields a Functional K⁺ Channel That Is Photoactivatable
Initially, we cloned the full coding sequence for the KAT1 K⁺ channel (Anderson et al., 1992), introducing sequences encoding the influenza virus HA epitope (YPYDVPDYA) with additional linkers of 9 and 11 residues (see Methods) from the modified Kir6.2 K⁺ channel clone of Zerangue et al.(1999). The HA epitopes and flanking sequences were inserted within the extracellular loops of the K⁺ channel, one between the first two transmembrane helices and the second between the S5 helix and the pore loop (Figure 1A ), after first introducing corresponding NheI and BglII restriction sites within these loops. Insertions resulted in loss of 8-mer and 6-mer peptide sequences native to KAT1 between T85-I93 and S242-V248, respectively, to yield the modified haKAT1 coding sequence. paGFP was constructed by generating a T207H site mutation in mGFP5, lacking cryptic introns (Haseloff et al., 1997), as described by Patterson and Lippincott-Schwartz (2002). Primers for both mGFP5 and paGFP constructs included NstI and PstI restriction sites, and fusion constructs were generated after replacing the haKAT1 stop codon with a PstI site to generate pHaKAT1:GFP and pHaKAT1:paGFP in a pBS cloning vector. These constructs were used to replace the mGFP5 reporter gene in pCambia1302 between sites BamHI within the multicloning site of the vector and PmlI in front of the Nos terminator. Finally, a tandem 35S promoter of Cauliflower mosaic virus (CaMV) was inserted at SacI in the multicloning site of the binary vector to generate the final expression cassette. pHaKAT1:paGFP was similarly subcloned into a second binary vector, pEN6, with six serial 35S promoter sequences (Batoko et al., 2000) as shown in Figure 1A.

View larger version (26K): [in this window] [in a new window]   Figure 1. The KAT1 K+ Channel Coding Sequence Yields a Functional Channel Protein as a Fusion Construct with Photoactivatable GFP and Tagged with HA Epitopes. (A) Schematic of the haKAT1:GFP (top) and haKAT1:paGFP (bottom) constructs with HA epitopes, spacer sequences, and insertion points as indicated. Underlined amino acid residues and flanking residue numbers relate to the KAT1 channel sequence; residues in boldface compose the HA epitopes. (B) Protein gel blot analysis of haKAT1:paGFP and KAT1 after expression in Xenopus oocytes. SDS-PAGE of solubilized microsomal membrane fractions (8 µg total protein/lane) from oocytes 3 d after injections probed with monoclonal antibody to GFP (GFP), then stripped and reprobed with monoclonal antibody to HA (HA). The antibodies register a single band near 110 kD, consistent with the molecular mass of 105.6 kD for the fusion construct. (C) Steady state current voltage curves for oocytes expressing haKAT1:paGFP (closed circles) and control (water-injected) oocytes (open circles) recorded in 10 mM KCl. Data are means ± SE of three recordings and show the characteristic appearance of inward current at voltages negative from –80 mV associated with the KAT1 K+ channel. Inset shows current relaxations from one control-injected oocyte and one oocyte expressing haKAT1:paGFP recorded under voltage clamp showing activation of the haKAT1:paGFP current. Voltage clamp cycles (data not shown): holding voltage, –50 mV; test voltage steps (eight), –90 to –160 mV. (D) and (E) Cross-sectional confocal fluorescence images of one Xenopus oocyte expressing haKAT1:paGFP before (D) and after (E) photoactivation of paGFP with 351/364-nm light from the UV laser. Excitation, 488 nm; emission, 505 to 530 nm.

To determine the localization in the plant, we used infiltration and Agrobacterium tumefaciens–mediated transfection to express haKAT1:paGFP and haKAT1:GFP in tobacco leaves. Leaves transfected with haKAT1:GFP were also used for cellular fractionation and (protein gel blot) analysis. We found that the GFP and HA epitopes were recovered in microsomal pellet and, following two-phase partitioning, segregated with the plasma membrane H⁺-ATPase (see Figure 12). Again, protein gel blot analysis yielded a single band of 105 kD expected of the haKAT1:GFP fusion, indicating its predominant localization to the plasma membrane. Leaf sections were examined by confocal microscopy 48, 72, and 96 h after transfection. All three time points gave similar results, with strongest expression at 72 h. Confocal images of leaves transfected with haKAT1:GFP showed GFP fluorescence that was confined to the cell periphery and was absent from cytosolic strands and perinuclear regions (Figures 2A to 2D ; see also Figure 8). Leaves transfected with haKAT1:paGFP were indistinguishable from nontransfected tissue (Figures 2E and 2F). We used the 351- and 364-nm lines of a UV laser to activate the paGFP in vivo. After preirradiation at these wavelengths, a similar distribution of fluorescence was observed with 488-nm excitation in haKAT1:paGFP transfected leaves (Figures 2G to 2I) but not in the cells of control leaves that were not transfected or were transfected with the empty vector (data not shown). Comparison of the fluorescence signals before and after photoactivation indicated an 30- to 50-fold increase in fluorescence yield when excited at 488 nm, consistent with previous reports (Patterson and Lippincott-Schwartz, 2002, 2004), and a maximum photoactivation with 4- to 5-µs irradiation at the combined 351/364-nm UV laser lines (Figure 2J).

View larger version (28K): [in this window] [in a new window]   Figure 12. Sp2 Fragment of Tobacco SYP121 Selectively Suppresses KAT1 K+ Channel Distribution to the Plasma Membrane. Protein gel blot analysis of the H+-ATPase (PMA2:GFP) and KAT1 K+ channel (haKAT1:GFP) localization after expression in tobacco leaves alone (Control), with the Sp2 fragment of tobacco SYP121 (+Sp2), and with the mutant Rab1b-N121I. SDS-PAGE (20 µg total protein/lane) of solubilized microsomal membranes after separation by two-phase partitioning to isolate plasma membrane (PM) and endomembrane (IM) fractions. After blotting, nitrocellulose filters were probed with polyclonal antibody to GFP and visualized by chemiluminescence (H+-ATPase) and radiotracer phosphor imaging (K+ channel).

View larger version (54K): [in this window] [in a new window]   Figure 2. Tobacco Leaves Transfected with haKAT1:GFP and haKAT1:paGFP Yield a Peripheral Distribution of Punctate GFP Fluorescence. Confocal images of tobacco leaf epidermis expressing haKAT1:GFP at 72 h after transfection at low ([A] and [B]) and high ([C] and [D]) magnification with ([A] and [C]) and without ([B] and [D]) the corresponding bright-field overlay. Corresponding confocal images of tobacco leaf epidermis expressing haKAT1:paGFP before ([E] and [F]) and after ([G] to [I]) 4-µs photoactivation of paGFP within the boxed region indicated in (F). (E) and (H) are bright field only, and (H) and (I) are high-magnification images of the top portion of the photoactivated region in (G). Again, note the punctate, peripheral distribution of the fluorescence. Relative fluorescence activation time course (J) as a function of total pixel dwell time with 351/364-nm light indicated a 4- to 5-µs irradiation for maximum efficiency in photoactivation and a 30- to 50-fold increase in fluorescence yield when excited with 488-nm light. Data are means ± SE of three experiments. Bars = 50 µm in ([A], [B], and [E] to [G]) and 5 µm in ([C], [D], [H], and [I]).

View larger version (45K): [in this window] [in a new window]   Figure 8. Cytosolic (Dominant-Negative) Sp2 Fragment of Tobacco SYP121 Selectively Suppresses KAT1 K+ Channel Traffic and Affects Its Distribution at the Plasma Membrane. (A) Three-dimensional reconstructions from confocal fluorescence images of tobacco epidermal cells expressing haKAT1:GFP ([a] to [c]) and the H+-ATPase PMA2:GFP ([d] to [f]) fusion constructs. GFP fluorescence is in green, and chloroplast fluorescence is overlayed in red. Excitation, 488 nm; emission, 505 to 530 nm (GFP) and 560-nm long-pass filter (chloroplasts). Expression of the fusion constructs only ([a] and [d]), together with the Sp2 fragment of NtSyp121 ([b] and [e]), and with the dominant-negative Rab1b-N121I mutant ([c] and [f]) to block export from the endoplasmic reticulum to the Golgi apparatus is shown. Note the punctate distribution of haKAT1:GFP (a) compared with the essentially uniform distribution of the PMA2:GFP marker (d) over the cell surfaces. A reticulate network, cytoplasmic strands, and a nuclear ring were evident in coexpression with the Rab1b-N121I mutant in both cases ([c] and [f]) and for the K+ channel in coexpression with the Sp2 fragment (b). For the H+-ATPase, coexpression with the Sp2 fragment had no visible effect on its distribution to the cell periphery (e). Bars = 20 µm. (B) Kymographic analysis of data from Figure 9 for haKAT1:paGFP. The fluorescence signal was taken over the time course of the experiment from pixels along a line traced around the edge of the cell, averaging over a width of 10 pixels (1.4 µm), as shown in the bright-field image (dotted line, inset). Position along the line determines the horizontal axis, time progression determines the vertical axis, and fluorescence intensity is color-coded (scale inset). Photoactivated (P) region is indicated above the kymograph, and time of photoactivation is indicated at the left (lollipop). The rapid lateral dispersal of the fluorescence signal after photoactivation doubled the predominant fluorescence spread over a period of 40 to 50 s and gives rise to the diagonal pattern of fluorescence intensities seen here (cf. Figure 5C). (C) and (D) Confocal dual-labeling experiments using protoplasts from tobacco leaf tissue previously transfected with haKAT1:GFP together with either the Sp2 fragment of tobacco SYP121 (C) or the Rab1b-N121I mutant (D). Protoplasts in each case were bound with Alexa594-HA for 5 min prior to imaging and are shown here in tangential surface views as (left to right): bright-field composite, Alexa594, GFP, and chloroplast fluorescence. Imaging parameters were as in Figure 6. Colocalization analysis ([C], bottom inset) was determined as the relative fluorescence intensities of Alexa594 and GFP labeling along the dotted lines (tail to head = position left to right) as shown. Note the absence of Alexa594-HA labeling in coexpression with the Rab1b-N121I mutant (D) and the presence of both Alexa594 and GFP signals with Sp2 fragment coexpression (C). With Sp2 coexpression, Alexa594-HA and GFP labeling do not colocalize but do overlap, as expected with retention of a portion of the K+ channel protein in the endoplasmic reticulum. Bars = 5 µm.

View larger version (122K): [in this window] [in a new window]   Figure 3. haKAT1:paGFP Is Nonmobile at the Cell Periphery. Bright-field (top left frame) and GFP confocal fluorescence images from one experiment with tobacco epidermal cells expressing the fusion construct. Images are sections within the epidermal cell layer taken parallel to the leaf surface. Images were collected at intervals before and after 4-µs photoactivation with 351/364-nm light within the area circled in the first two frames. Time in seconds relative to photoactivation is as indicated for each frame. See Supplemental Movie 1 online for the full set of images. Note the spatial stasis of the punctate fluorescent pattern around the lateral surface of the adjoining epidermal cells as viewed in this optical section and the apparent lack of any lateral movement or dispersion of the fluorophore within the plane of the cell surface beyond the photoactivation boundary. See Figure 5C for quantitative analysis. Bar = 5 µm.

View larger version (34K): [in this window] [in a new window]   Figure 5. haKAT1:paGFP Is Stationary at the Cell Periphery. Kymographic analysis of data from Figures 3 and 4 for the H+-ATPase PMA2:GFP (A), GFP:HDEL (B), and haKAT1:paGFP (C). In each case, the fluorescence signal was taken over the time course of the experiment from pixels along a line traced around the edge of the cell, averaging over a width of 10 pixels (1.4 µm), as shown in the bright-field images (dotted lines, insets). Position along the line determines the horizontal axis, time progression determines the vertical axis, and fluorescence intensity is color-coded (see inset in [A]). Photobleached (b) and photoactivated (p) regions are indicated above each kymograph and are color-coded for the two lines in (B). Times of photobleaching/photoactivation are indicated at the left (lollipop) in each case. Photobleaching of the H+-ATPase (A) led to a virtually complete loss of fluorescence within the two bleached regions that showed no evidence of filling or lateral movement near the edges of the two regions during the time course of the experiment. By contrast, photobleaching of GFP:HDEL (B) was followed by a rapid filling and recovery of fluorescence and was associated with diagonal patterns of fluorescence intensity, indicative of lateral movement within the cell. Kymographic analysis of data from Figure 3 for haKAT1:paGFP (C) gave an effective image-inverse of the results for the H+-ATPase, as expected of the complementarity of photoactivation with photobleaching and confirming the absence of any measurable displacement of the fluorescence laterally over the surface of the cell. Arrows (color-coded with the bright-field inset) mark two of the brighter puncta; vertical lines of fluorescence intensity indicate the absence of local mobility.

To quantify the mobility of the haKAT1:paGFP construct, we used kymographic analysis, recording over the time course of each experiment the fluorescence signal from pixels along a line traced around the edge of the cells. A line width of 10 pixels (=1.4 µm) was averaged in most cases, although similar results were obtained with line widths of 3, 5, and 8 pixels (0.5 to 1 µm). This analysis supported the interpretation from visual inspection of data such as shown in Figure 3. Kymographic analysis for this experiment is shown in Figure 5C with a 10-pixel width. Here, position along the line (Figure 5C, inset) determines the horizontal axis, time progresses down along the vertical axis, and fluorescence intensity is color-coded (see Figure 5A, scale inset). The analysis shows stationary boundaries at the limits of photoactivation; it also exposes the punctate fluorescence distribution, confirming the absence of any significant lateral movement or relocations affecting this pattern as evident in the longitudinal striation in fluorescence intensities over the 300-s time course after photoactivation.

For comparison, we examined the fluorescence distribution and lateral mobility of the GFP-tagged PMA2 H⁺-ATPase from Nicotiana plumbaginifolia. Three days after transfection, the tobacco leaves showed extensive GFP fluorescence in the epidermis that was restricted to the cell surface and distributed uniformly, as reported previously (Lefebvre et al., 2004; see also Figure 8); within the limits of resolution, no evidence was found for heterogeneous H⁺-ATPase accumulation or exclusion zones such as has been reported for the yeast H⁺-ATPase (Malinska et al., 2003, 2004). To test the mobility of the H⁺-ATPase, we used fluorescence recovery after photobleaching (FRAP) with the 458- and 488-nm argon laser lines while focusing either on the cell in cross section or on the cell surface. Images collected before and after photobleaching in each of five separate experiments showed a loss of fluorescence within the photobleached regions that did not recover, even over periods of 8 to 10 min. The experiment in Figure 4A included two regions subjected to photobleaching and illustrates the lack of fluorescence recovery in these regions. Much the same conclusion was drawn from kymographic analysis, shown in Figure 5A for the data of Figure 4A. The analysis yielded stationary boundaries in fluorescence at the limits of the photobleached regions and thus no evidence of an appreciable displacement of the H⁺-ATPase marker laterally around the cell over the time course of these measurements (see Supplemental Movie 2 online).

View larger version (93K): [in this window] [in a new window]   Figure 4. Comparative Analysis of Fluorophore Mobilities by FRAP with H+-ATPase PMA2:GFP, Which Labels the Plasma Membrane, and GFP:HDEL, Which Labels the Endoplasmic Reticulum. (A) Plasma membrane labeled by H+-ATPase PMA2:GFP. (B) Endoplasmic reticulum labeled by GFP:HDEL. In each case, bright-field (top left frames) and GFP confocal fluorescence images from one experiment with tobacco epidermal cells expressing the fusion constructs are shown. Images were collected at intervals before and after photobleaching with 458/488-nm light within the areas circled in the first two frames. Images are cross sections through the epidermal cell layer taken parallel to the leaf surface. Time in seconds relative to photoactivation is as indicated for each frame. See Supplemental Movies 2 and 3 online for the full sets of images. Note the absence of any significant fluorescence recovery in (A) and its rapid recovery within the photobleached areas in (B). The nuclear ring, characteristic of labeling in the endoplasmic reticulum, is clearly visible in the bottom part of each frame in (B) and is seen to migrate slowly to the left. See Figures 5A and 5B for quantitative analysis. Bars = 5 µm.

KAT1 Is Localized to Subdomains within the Plasma Membrane
One striking feature of the paGFP-tagged K⁺ channel was its apparent lack of mobility. Hurst et al. (2004) noted similar structures in guard cells and protoplasts from Vicia faba expressing a 35S-driven KAT1:GFP construct (see also Meckel et al., 2004) and suggested these to be endocytic vesicles and larger endosomal compartments that form at, and subsequently recycle to, the plasma membrane. Such vesiculation events imply a significant mobility to the structures, so we were intrigued to find that the fluorescent puncta remained stationary at the cell surface rather than relocating with cytosolic streaming and internal membranes.

To explore the association of the KAT1 label with the cell surface, we performed dual-labeling experiments using protoplasts from leaf issue previously transfected with the haKAT1:paGFP and haKAT1:GFP constructs. After 5-min incubation with Alexafluor594-tagged HA monoclonal antibody (Alexa594-HA), protoplasts were examined for Alexafluor594 fluorescence and for GFP fluorescence with haKAT1:paGFP after photoactivation. Each of 10 independent experiments gave similar results, and representative data and their analysis are shown in Figure 6 . As in the intact tissue, experiments with haKAT1:GFP and haKAT1:paGFP in the protoplasts showed the characteristic punctate pattern of fluorescence at the protoplast surface both without (Figures 6A to 6C) and with Alexa594-HA exposures (Figures 6E to 6P). By contrast, control experiments with protoplasts from nontransfected leaves (Figures 6Q to 6T) and from leaves transfected with KAT1:GFP (data not shown) gave no measurable labeling with Alexa594-HA, even after prolonged antibody treatment (see Figure 6 legend). Fluorescence intensity analysis confirmed the colocalization of Alexafluor594 and GFP signals over the protoplast surface: Alexafluor594 fluorescence was always associated with that of GFP (Figures 6G and 6M), although occasionally GFP fluorescence was observed independent of the HA tag, consistent with haKAT1:GFP in transit to the plasma membrane (data not shown).

View larger version (53K): [in this window] [in a new window]   Figure 6. The KAT1 K+ Channel Resides in Clusters at the Plasma Membrane Surface. Confocal fluorescence images from dual-labeling experiments using protoplasts from tobacco leaf tissue previously transfected with haKAT1:GFP and bound with Alexa594-HA. Bright-field composite (A), GFP (B), and chloroplast (C) fluorescence from one protoplast in tangential surface view without prior Alexa594-HA treatment. The images show characteristic fine punctate fluorescence of the fusion constructs and an appreciable intrinsic fluorescence from the chloroplasts frequently observed, especially in the GFP channel (see [Q] to [T]). Analysis of one point near the center of this image set ([D]; analysis area boxed in inset at left) shows the fluorescence intensity pseudo-color-coded (bottom) and in three-dimensional surface representation (top). Fluorescence intensity was well-fitted to a two-dimensional Gaussian distribution function as , where A is the maxiumum fluorescence amplitude, x and y are the image coordinates, xo and yo are the corresponding coordinates of the fluorescence maximum, and bx and by reflect the corresponding dimensional spread of the fluorescence signal away from the maximum. As shown, the analysis yielded visually satisfactory and statistically best fittings with an isodiametric surface and FWHM peak height of 528 ± 13 nm (gray surface overlay). Dual-labeling after 5-min exposure to Alexa594-HA of protoplasts expressing haKAT1:GFP in tangential surface view ([E] to [J]) and in equatorial cross section ([K] to [P]) as composites with ([E] and [K]) and without ([F] and [L]) bright-field overlay shows colocalization ([G] and [M]) of the Alexa594 ([H] and [N]) and GFP ([I] and [O]) fluorescence signals distinct from that of the chloroplasts ([J] and [P]). Colocalization analysis ([G] and [M]) was determined as the relative fluorescence intensities of Alexa594 and GFP labeling along the dotted lines (tail to head = position left to right) shown in (F) and (L). Bright-field composite (Q), Alexa594 (R), GFP (S), and chlorophyll (T) fluorescence images of a protoplast from an untransfected tobacco leaf after Alexa594-HA treatment are shown. This tangential surface view shows the absence of Alexa594-labeled surface structure, although the wide-spectrum fluorescence of the chloroplasts is evident, especially in the GFP channel. Fluorescence excitation: 543 nm (Alexa594-HA), 488 nm (GFP, chloroplasts); emission: 585 to 615 nm (Alexa594-HA), 505 to 530 nm (GFP), 560-nm long-pass filter (chloroplasts). Bars = 5 µm.

View larger version (24K): [in this window] [in a new window]   Figure 7. KAT1 K+ Channels Distribute in Clusters with Prevalent Diameters of 0.5 to 0.6 µm. Relative fluorescence intensities were fitted to a two-dimensional Gaussian distribution function as in Figure 6, and the FWHM peak height was averaged over the two dimensions used as a measure of the cluster diameter. (A) Data comparing the frequency distribution of cluster diameters measured from protoplasts treated without (GFP) and with Alexa594-HA (HA), with dimensions determined using the corresponding fluorescence signals. Diameter frequencies in each case were well fitted to a single, log-normal distribution function (solid curve) with a maximum of 532 ± 22 nm and peak spread coefficient of 0.329 ± 0.003. (B) Data, including those in (A), comparing the frequency distribution of haKAT1:GFP and haKAT1:paGFP cluster diameters in intact tobacco epidermal cells (see Figures 2, 3, and 8) with haKAT1:GFP cluster diameters in protoplasts as in (A). Diameter frequencies for expression in epidermal cells were well fitted to single, log-normal distribution function (solid curve) with a maximum of 509 ± 18 nm and peak spread coefficient of 0.359 ± 0.003.

We examined the effects of the Sp2 fragment of tobacco SYP121, both by cotranfecting tobacco leaves with haKAT:paGFP or haKAT1:GFP together with a 35S-driven Sp2 construct and by transfecting with the K⁺ channel constructs in leaves of the stably transformed Sp2-14 tobacco line that expresses the Sp2 fragment under control of a dexamethasone-inducible promoter (Geelen et al., 2002). For comparison, similar experiments were performed by cotransfection with Sp2 fragments of tobacco SYP121, of its Arabidopsis Q-SNARE homolog (Leyman et al., 1999), and with the dominant-negative Rab1b-N121I mutant that blocks traffic between the endoplasmic reticulum and Golgi apparatus (Batoko et al., 2000). We also examined the mobility of haKAT1:paGFP with Sp2 coexpression. Finally, we tested the effects of the Sp2 fragments on plasma membrane delivery of the GFP-tagged PMA2 H⁺-ATPase.

Figures 8 to 11 show results from experiments directed to each of the above points in turn. In every case, we obtained similar results from each of at least eight independent experiments. Three days after transfection, tobacco leaves expressing haKAT1:GFP alone showed the characteristic punctate distribution of channel clustering around the cell periphery (Figure 8A, a), while coexpression of the K⁺ channel marker and the Sp2 fragments of tobacco (Figure 8A, b) or of Arabidopsis (data not shown) SYP121 gave GFP fluorescence both at the cell periphery as well as in an internal reticulate network and a ring around the nucleus, characteristic of retention in the endoplasmic reticulum. Similar results were obtained when haKAT1:GFP was expressed together with the Rab1b-N121I mutant (Figure 8A, c), although in this case, labeling of the internal structures was generally more pronounced. Remarkably, we were unable to affect the delivery of the GFP-tagged PMA2 H⁺-ATPase. Coexpression with the Sp2 fragments gave peripheral distributions (Figure 8A, e), much as was observed when the H⁺-ATPase was expressed on its own (Figure 8A, d). Only when coexpressed with the Rab1b-N121I mutant did we observe labeling of an internal reticulate network and a ring around the nucleus expected for marker retention within the secretory pathway (Figure 8A, f).

View larger version (133K): [in this window] [in a new window]   Figure 9. haKAT1:paGFP Is Mobile at the Epidermal Cell Periphery When Coexpressed with the Sp2 Fragment of NtSyp121. Bright-field (top left frame) and GFP confocal fluorescence images from one experiment with tobacco epidermal cells expressing the fusion construct. Images were collected at intervals before and after photoactivation with 351/364-nm light within the boxed area in the first two frames. Images are sections within the epidermal cell layer taken parallel to the leaf surface. Time in seconds relative to photoactivation is as indicated for each frame. Note that the nuclear ring, characteristic of labeling in the endoplasmic reticulum, was clearly visible after photoactivation along with two chloroplasts, just visible in the bright-field image, that enter the fluorescence focal plane later in the frame series. See Figure 8C for quantitative kymographic analysis and Supplemental Movie 4 online for the full sets of images. Imaging parameters were as in Figures 3 and 4. Bar = 5 µm.

View larger version (76K): [in this window] [in a new window]   Figure 10. haKAT1:GFP Is Mobile at the Plasma Membrane When Coexpressed with the Sp2 Fragment of NtSyp121. (A) Bright-field (top left frame) and Alexa594 confocal fluorescence images from one experiment with tobacco protoplasts expressing the fusion construct. Protoplasts were bound with Alexa594-HA for 5 min prior to imaging and are shown here in tangential surface view. Images were collected at intervals before and after photobleaching with 543-nm light within the circled area (a) in the second frame. Time in seconds relative to photobleaching is as indicated for each frame. Bar = 5 µm. (B) Fluorescence recovery analysis shows fluorescence intensities in areas (cross-referenced to circles [a], [b], and [c] in [A]) fitted by nonlinear least squares (Marquardt, 1963) to single exponential functions (solid line) for time points after photobleaching (t > 0 s). Note the recovery in the photobleached area (a) is accompanied by a loss in fluorescence in the neighboring area (b), but no appreciable change is seen in the more outlying area (c). Fitted time constants were as follows: (a), 6.5 ± 0.4 s; (b), 6.8 ± 0.3 s. See Supplemental Movie 5 and Figure1 online for the full image sequence, composite, and overlay color images. Imaging parameters were as in Figure 6.

View larger version (14K): [in this window] [in a new window]   Figure 11. Dominant-Negative SNARE Fragments Affect Surface Clustering of the KAT1 K+ Channel as Well as Its Trafficking to the Plasma Membrane. (A) Frequency distributions of the K+ channel taken from x20 (objective) magnification images at random from all 211 experiments expressing haKAT1:GFP or haKAT1:paGFP alone (Control), together with the Sp2 fragments of tobacco or Arabidopsis SYP121 (+Sp2), and together with the Rab1b-N121I mutant (+Rab1b-N121I). Localizations scored as peripheral only, reticulate, and reticulate with nuclear ring. Difficulties in identifying peripheral labeling at low magnification mean that a peripheral distribution cannot be excluded from the latter two categories. (B) Frequency distributions of cluster diameters for the KAT1 K+ channel from protoplasts of tobacco leaves expressing haKAT1:GFP alone (-Sp2) or together with the Sp2 fragments of tobacco or Arabidopsis SYP121 (+Sp2). Cluster diameters were determined as in Figure 6 using the Alexa594-HA fluorescence signal. Data were fitted to log-normal functions (solid lines) as in Figure 7. Fitted parameters were as follows: diameter at maximum frequency, 532 ± 22 nm (-Sp2) and 3080 ± 70 nm (+Sp2); peak spread coefficient, 0.329 ± 0.003 (-Sp2) and 0.40 ± 0.02 (+Sp2). (C) Frequency distributions of cluster diameters for the KAT1 K+ channel from intact epidermal cells of tobacco leaves expressing haKAT1:GFP alone (-Sp2) or together with the Sp2 fragments of tobacco or Arabidopsis SYP121 (+Sp2). Cluster diameters were determined as in Figure 6 using the GFP fluorescence signal. Data were fitted to a log-normal function (solid lines) as in Figure 7. Fitted parameters were as follows: diameter at maximum frequency, 509 ± 18 nm (-Sp2) and 4300 ± 100 nm (+Sp2); peak spread coefficient, 0.359 ± 0.003 (-Sp2) and 0.43 ± 0.02 (+Sp2). Note that the inability to distinguish internal labeling from true localization to the plasma membrane in this case almost certainly skews the results with the Sp2 fragments to larger diameter values.

Finally, we examined the mobility of the K⁺ channel in conjunction Sp2 fragment coexpression, again using the facility for local photoactivation of the haKAT1:paGFP fusion protein. Figure 9 illustrates one such experiment (see also Supplemental Movie 4 online). In this case, the region for photoactivation was deliberately chosen to catch both the cell periphery and nucleus of one epidermal cell. The images after photoactivation clearly show a rapid dispersal and loss of the fluorescence signal along the cell periphery and a somewhat slower decay in signal from the nuclear ring. Kymographic analysis for these data (Figure 8B) gives a quantitative picture of this process, showing that the diffusive expansion of the signal led to a doubling of the predominant fluorescence spread along the transverse cell surface over a period of 40 to 50 s (cf. Figure 5C). Similar analyses with Sp2 fragment coexpression yielded a mean time of 48 ± 6 s to doubling of the fluorescence signal distribution. To confirm that the mobility of paGFP in this case truly reflected the K⁺ channel within the plasma membrane, we also performed FRAP experiments on protoplasts coexpressing haKAT1:GFP and the Sp2 fragments after exposures to Alexa594-HA. Results from one of 16 separate experiments are shown in Figure 10 (see also Supplemental Movie 5 and Figure 2 online). After photobleaching of the Alexa594 dye locally, the fluorescence signal was observed to recover rapidly within the same region of the cell (Figure 10A). Analysis of the fluorescence recovery (Figure 10B) indicated a simple exponential process with a time constant, , of 6.5 ± 0.4 s and similar kinetics for the decay of the Alexa594 fluorescence signal in adjacent regions of the membrane surface. Analysis of all 16 FRAP experiments yielded a mean of 7.5 ± 0.9 s for the fluorescence response to photobleaching. These results are consistent with haKAT1:GFP movement (Figures 8 and 9) and confirm that the K⁺ channel is highly mobile over the plasma membrane surface when the Sp2 fragments are expressed. They also indicate a similar degree of mobility within the K⁺ channel clusters in this situation. Thus, the effects of Sp2 fragment expression on K⁺ channel and H⁺-ATPase membrane distributions lead us to propose that the Q-SNAREs contribute selectively to the delivery of the channel to the plasma membrane and also influence its anchoring at the cell surface. We return to these points below.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Traffic of membrane proteins between endomembrane compartments and the plasma membrane is generally recognized to depend on the activities of SNARE proteins as part of the vesicle fusion machinery of all eukaryotes. These activities clearly play a part in regulating ion channels and other transporters at the plasma membrane, although many details are only now beginning to come to light. Vesicle trafficking must impact on ion channel turnover, the steady state balance of ion channel populations during cell development, and their modulation by hormones as well as by environmental factors (Battey and Blackbourn, 1993; Blatt and Thiel, 2003; Muday et al., 2003; Pratelli et al., 2004). However, there has been remarkably little information available that bears on traffic, partitioning, and mobility of specific membrane proteins to the plasma membrane, much less on roles for plasma membrane SNAREs in these processes. Our results now address these questions in context of the KAT1 K⁺ channel, taking advantage of the facility for tracking channel protein mobility over subcellular distances. (1) We show that K⁺ channel expression leads to its appearance in stationary clusters within the plane of the plasma membrane. (2) We demonstrate that traffic of the KAT1 K⁺ channel, but not a GFP-tagged H⁺-ATPase, is sensitive to the dominant-negative Sp2 fragments of the SYP121 plasma membrane Q-SNAREs from tobacco and Arabidopsis. (3) We report that expressing the Sp2 fragments affects the organization of the KAT1 clusters, as is manifest by their mobility, loss of integrity, and diffusive spread. These findings indicate a selective trafficking of the K⁺ channels at the plasma membrane and implicate a role for the SNAREs in this process. They also point to a targeted insertion and/or removal of the K⁺ channels and suggest that the plasma membrane SNAREs are important for localizing and anchoring the K⁺ channels within microdomains of the plasma membrane.

K⁺ Channel Clustering
Our evidence for clustering of the KAT1 K⁺ channels lends substance to previous work suggesting that many ion channels, as well as other integral membrane proteins, partition to form microdomains at the plasma membrane surface. From the biophysical and cellular standpoints, a common feature of K⁺ and other ion channels is the nonhomogeneous distributions of their activities over the cell surface (Tester, 1990; Hille, 2001). Patch clamp studies frequently yield a smaller percentage of recordings that include channel openings, and those that do often exhibit several open current levels consistent with a patch containing multiple channels. Indeed, the frequency of such channel recordings has been used to estimate the density and distribution of K⁺ channels within the plasma membrane of hair cells (Roberts and Haigler, 1990), of cation permeable channels in Fucus (Taylor et al., 1996), and of K⁺ and Cl^– channels in Chara (Thiel et al., 1993; Homann and Thiel, 1994). Of course, these analyses take account of functional channels and cannot distinguish patches with bone fide but nonactive channels from ones that actually lack the channel proteins. Hurst et al. (2004) and Meckel et al. (2004) have approached one aspect of this issue, relating single-channel recordings and ensemble current densities to protoplast surface area. They also observed a GFP-tagged KAT1 K⁺ channel to cluster when expressed in Vicia guard cells and protoplasts, although these observations were interpreted to reflect K⁺ channel densities associated with exocytotic and endocytotic vesicles rather than stable K⁺ channel microdomains per se.

From a biochemical standpoint, evidence is mounting for raft-like subdomains in plants (Mongrand et al., 2004; Borner et al., 2005), as it is in animals (Lang et al., 2001; Anderson and Jacobson, 2002; Lai, 2003), associated with distinct, detergent-resistant lipid complexes, integral membrane protein compositions, and accompanying cytoskeletal linkages. In neuromuscular and epithelial tissues, both K⁺ channels and SNAREs are known to colocalize within such domains (Bravo-Zehnder et al., 2000; Chamberlain et al., 2001; Martens et al., 2001; Martens et al., 2004), and similar biochemical associations are evident also in plants (Borner et al., 2005). Whether lipid composition per se is a factor in determining the population (Romanenko et al., 2004) and mobility of proteins within these microdomains remains unclear (Lippincott-Schwartz, 2002; Kenworthy et al., 2004). Of interest, recent studies (Malinska et al., 2003, 2004) identified a segregation of two different lipid-raft domains in Saccharomyces, although both associated with membrane proteins that were biochemically isolated together in detergent-resistant fractions. When fluorescently tagged for confocal microscopy, one domain formed puncta of 300-nm diameter and incorporated amino acid–H⁺ symporters, including CAN1p; the second domain, incorporating the yeast PMA1p H⁺-ATPase, showed exclusion holes at the same sites and covered much of the rest of the membrane.

The situation for the KAT1 K⁺ channel and H⁺-ATPase expressed in tobacco differs from that of yeast on at least two counts. First, within the limits of resolution, both PMA2 (Figures 4 and 8) and PMA4 H⁺-ATPases (Lefebvre et al., 2004) appear uniformly spread over the plasma membrane. If these H⁺-ATPases are found primarily in lipid rafts, like their Arabidopsis counterparts (Borner et al., 2005), it would appear that the dimensions of these raft domains are well below the limit of microscopic resolution (Varma and Mayor, 1998; Pralle et al., 2000) and are homogenously distributed over the membrane. The same may be true of the yeast PMA1p H⁺-ATPase, although the H⁺-ATPase domains in yeast are evidently excluded from some regions of the plasma membrane (Malinska et al., 2003). Second, we found that the PMA2 H⁺-ATPase was positionally stable within the plasma membrane surface (Figures 4 and 5). Thus, if there is lateral movement, it must be at least an order of magnitude slower than that of the yeast H⁺-ATPase (Malinska et al., 2004). The KAT1 K⁺ channel may also accumulate within lipid rafts. However, expression of plasma membrane ion channels generally is insufficient to permit their biochemical detection at this time (Mongrand et al., 2004; Borner et al., 2005). We did find KAT1 to be associated with a moderately detergent-resistant fraction when expressed in tobacco (see Supplemental Figure 1 online), although these observations may not reflect its native situation in Arabidopsis. Whether or not the K⁺ channel is a raft protein, the plant plasma membrane clearly does support its clustering within the bilayer surface, even in the absence of the HA monoclonal antibody (cf. Figures 2, 3, 6, and 7), and this distribution differs from that of the PMA2 H⁺-ATPase.

It is worth noting that plant ion channels normally express at levels that are too low for detection using GFP and standard fluorescence confocal microscopy when driven by their own promoters. We used KAT1 constructs driven by the 35S CaMV promoter and so cannot exclude the possibility that overexpression misrepresented KAT1 localization and clustering. Nonetheless, for a large majority of membrane proteins, 35S-driven expression leads to distributions that align closely with that of the native protein first (Lippincott-Schwartz et al., 2001; Tian et al., 2004; Uemura et al., 2004), and only with extreme overexpression are these proteins mistargeted and accumulate in other membranes. Significantly, we did not observe haKAT1:paGFP or haKAT1:GFP to accumulate in the endoplasmic reticulum, except when export was suppressed by coexpression with the Sp2 fragments or with the mutant Rab1b-N121I. These observations lend some validity to the idea that the clustering of the K⁺ channels is physiologically relevant. How big are these clusters? An estimate of the confocal resolution suggests that at least 200 chromophores must reside in close proximity to give a detectable fluorescence signal (Kubitscheck et al., 1996). KAT1, like other Kv-type K⁺ channels (Yellen, 2002), is thought to coalesce as a tetramer to form functional channel units, implying a minimum size for the clusters of the order of 50 functional K⁺ channels. It is of interest that the KAT1 amino acid sequence includes a C-terminal domain that may contribute to the formation of channel clusters (Ehrhardt et al., 1997). Thus, in principle, these clusters might arise de novo, without requirements for additional proteins, such as cytoskeletal elements for nucleation, scaffolding, or anchoring.

In fact, one of the most striking observations was of the spatial constancy of the KAT1 clusters. From photoactivation studies, we found these K⁺ channel clusters to remain stationary at the cell surface (Figures 3 and 5), and even over periods of 20 min no evidence of lateral displacement or dispersal could be identified. This behavior contrasts with the lateral mobilities of several model proteins, including glycosylphosphatidyl-inositol–anchored and raft-associated proteins as well as non-raft-associated integral membrane proteins (Kenworthy et al., 2004; Malinska et al., 2004). Many of these proteins exhibit (two-dimensional) diffusion constants on the order of 10^–9 cm²/s and FRAP time constants of 2 to 10 s over distances of a few micrometers (Kenworthy et al., 2004). These diffusion constants compare favorably with values for lateral diffusion of the mammalian inositol trisphosphate receptor Ca²⁺ channel (Fukatsu et al., 2004) and rhodopsin (Jacobson et al., 1987). By contrast, calculated diffusion constants for the major Band 3 protein and spectrin, which are associated with the erythrocyte cytoskeleton, are 30- to 100-fold lower (Jacobson et al., 1987; Tomishige et al., 1998). In the absence of any measurable fluorescence displacement of haKAT1:paGFP, it is difficult to estimate a diffusion constant, but this value must certainly lie an order of magnitude or more lower still. Thus, one explanation is that the KAT1 K⁺ channels are firmly anchored in place or are corralled within the microdomains through protein associations, possibly tying the channels into an underlying cytoskeletal framework. Protein complexes that function in trafficking, scaffolding, and anchoring have been identified to associate with Kir inward-rectifier K⁺ channels (Schwappach et al., 2000; Leonoudakis et al., 2004a, 2004b