Background: The plant vacuole is a multifunctional organelle that has various physiological functions. The vacuole dynamically changes its function and shape, dependent on developmental and physiological conditions. Our current understanding of the dynamic processes of vacuolar morphogenesis has suffered from the lack of a marker for observing these processes in living cells.
Results: We have developed transgenic Arabidopsis thaliana expressing a vacuolar syntaxin-related molecule (AtVam3/SYP22) fused with green fluorescent protein (GFP). Observations using confocal laser scanning microscopy demonstrated that the plant vacuole contained a dynamic membrane system that underwent a complex architectural remodelling. Three-dimensional reconstitution and time-lapse analysis of GFP-fluorescence images revealed that cylindrical and sheet-like structures were present in the vacuolar lumen and were moving dynamically. The movement, but not the structure itself, was abolished by cytochalasin D, an inhibitor of actin polymerization. This moving structure, which sometimes penetrated through the vacuolar lumen, possessed a dynamic membrane architecture similar to the previously recognized ‘transvacuolar strand.’
Conclusion: We propose two possible models for the formation of the vacuolar lumenal structure. Membrane structures including protruding tubules and reticular networks have recently been recognized in many other organelles, and may be actively involved in intra- and/or inter-organelle signalling.
In plants, the expansion of body size is beneficial to the efficient acquisition of resources such as light, minerals and water from the environment. The expansion of a plant body is accomplished mostly by the enlargement of vacuoles rather than by cell division or by synthesis of cytosolic materials (Herman & Larkins 1999; Marty 1999). The importance of vacuolar enlargement is evident in various defects in plant development, such as a short and wavy shape in root hair morphology which accompanies a striking reduction in the vacuolar size (Galway et al. 1997; Wang et al. 1997).
It has long been known that the shapes and sizes of the plant vacuole vary during cell growth, differentiation and senescence (Boller & Wiemken 1986; Marty 1999). For example, in the stomatal cell, a globular form of the vacuole turns into a network of interconnected tubes and chambers during differentiation (Palevitz et al. 1981; Palevitz & O’Kane 1981). In the seed, the protein storage vacuoles (PSVs) fuse with the lytic vacuoles and are converted into a central vacuole (Inoue et al. 1995). In suspension-cultured cells, numerous active autophagic vacuoles are eventually expelled into a central vacuole (Aubert et al. 1996; Moriyasu & Ohsumi 1996). These observations led us to speculate that the vacuole may be a dynamic organelle with a constantly changing morphology. However, in spite of the availability of the above and other descriptions, our current understanding of such dynamic vacuole processes has suffered from a lack of sufficient information about their three-dimensional organization and real-time dynamics.
AtVam3p, the gene product of AtVAM3, belongs to the syntaxin family of Arabidopsis thaliana and is localized to the vacuolar membrane (Sato et al. 1997). Recently, a systematic nomenclature has been proposed; i.e. AtVam3 to be SYP22 (sytaxin of plants 22) (Sanderfoot et al. 2000). (In this study, ‘AtVam3’ was used, for historical reasons.) To visualize dynamic movements of the vacuole, we established transgenic A. thaliana lines expressing a full-length AtVam3 fused with green fluorescent protein (GFP). Observations by confocal laser scanning microscopy (CLSM) demonstrated that the GFP-AtVam3p was localized to the vacuolar membrane in various cell types. A real-time analysis and three-dimensional reconstitution of the CLSM images revealed highly complex architectures and dynamic movements of the vacuolar membrane in living cells.
cDNA encoding the fusion protein between GFP and AtVam3 was introduced into A. thaliana by Agrobacterium-mediated transformation. Twenty-two transgenic plants expressing GFP-AtVam3p were selected using a conventional fluorescence microscope, all of which exhibited an identical pattern of subcellular localization of GFP-fluorescence in different tissues. One transgenic line (At1/GFP-AtVam3p) showing the strongest GFP-fluorescence was chosen for further investigations.
A fluorescent probe for the vacuolar membrane
It has previously been demonstrated, by the use of immunoelectron microscopy, that AtVam3p is localized to the vacuolar membrane (Sato et al. 1997). To investigate whether addition of GFP affects the subcellular localization of AtVam3p, the crude membrane fraction prepared from At1/GFP-AtVam3p root tissue was subjected to sucrose-density-gradient centrifugation. In an immunoblot analysis of each fraction, the antibody against AtVam3p identified two polypeptides: GFP-AtVam3p (60 kDa) and AtVam3p (33 kDa) (Fig. 1A). The distribution patterns of both AtVam3p and GFP-AtVam3p were almost identical to that of a vacuolar marker protein, the H+-PPase (Fig. 1B), but were clearly different from that of the plasma membrane H+-ATPase (Fig. 1A and B). Moreover, the peak fractions of GFP-AtVam3p and endogenous AtVam3p were distinct from those of the markers for Golgi (RGP1; Dhugga et al. 1997) and ER (AtSar1; Takeuchi et al. 2000), as well as for the pre-vacuolar compartment (PVC) (SYP21; Bassham & Raikhel 1999). These results indicate that GFP-AtVam3p, as well as AtVam3p, is localized to the vacuolar membrane.
Consistent with the above data, the mesophyll protoplast cells prepared from At1/GFP-AtVam3p showed that the GFP-fluorescence was not localized to the plasma membrane (Fig. 2A). The chloroplasts (red fluorescence) were located outside the GFP-labelled membrane compartment, indicating that the GFP-fluorescence was localized to the central vacuolar membrane. Moreover, additional GFP-fluorescence was observed in the vacuolar lumen of the mesophyll cells (Fig. 2A and B), as well as in the vacuoles of other tissues such as root epidermal cells (Fig. 2C, D), root hair cells (Fig. 2E, F), the stomatal guard cells (Fig. 2G, H), and leaf epidermal cells (Fig. 2I). Thus, GFP-AtVam3 can be used as a fluorescent probe for the vacuolar membrane structure and its complexity in living cells.
Complexity of the vacuolar membrane system
Closer observations of GFP-fluorescence in At1/GFP-AtVam3p by CLSM revealed complicated vacuole architectures in various cell types. In root epidermal cells, internal membraneous structures (referred to as intravacuolar membrane structures; Fig. 2C, arrowheads) existed inside, in addition to the central vacuolar membrane (referred to as the outer-vacuolar membrane structure; Fig. 2C, arrows). In the stomatal guard cells, the vacuoles were separated into several compartments (Fig. 2G), and, in the leaf epidermal cells, the vacuole has a tubular structure invaginating into the lumen from the outside (Fig. 2I).
A three-dimensionally reconstructed image of the guard cell vacuole clearly demonstrated that several canal-like structures were invaginating from the outer-vacuolar membrane, by which the vacuole was separated into several parts (Fig. 2H). In the protoplast of the mesophyll cells, the central vacuole contained another enclosed membrane compartment in its lumen, which was also observed in the root epidermal cells (compare Fig. 3A and B with Fig. 2C). It should be emphasized that the intravacuolar membranes showed a stronger fluorescence than the outer-vacuolar membrane. A plot of intensity showed that the intravacuolar membranes had signals about twofold stronger than the outer ones (Fig. 3C, D). A three-dimensional reconstruction of the GFP-fluorescence images revealed that this compartment was not a vesicle but a cylindrical structure (Fig. 3A-1, A-2, B-1 and B-2), which was sometimes attached with another cylindrical structure (Fig. 3B-1 and B-2). In the leaf epidermal cells, the intravacuolar structure was not a simple tube, but it was rather a sheet-like structure (Fig. 4A, A-1 and A-2, arrows). This sheet-like structure seemed to be rounded and formed into a cylindrical structure, which was also observed in the mesophyll protoplast cells (Fig. 4B, B-1 and B-2).
Real-time vacuolar membrane dynamics
Time-lapse imaging of the living epidermal cells revealed that the sheet-like structures were constantly moving and changing in morphology (Fig. 5A, B, C, arrows; video materials are available at http://www.vacuole.net/video). The sheet-like structure changed its length and shape; when observed in the same focal plane, it extended towards the vacuolar lumen (t = 23.9 s) and then disappeared (t = 47.8 s; Fig. 5A, arrow). In another cell, a cylindrical structure was formed (t = 0; Fig. 5B, arrow), and then attached to another one extending from the opposite direction (t = 15.7 s; Fig. 5B, arrowhead). Meanwhile, the cylindrical structure disappeared and the sheet-like structure remained connected to another sheet-like structure (t = 31.6 s). Finally, another cylindrical structure was formed (t = 47.6 s). A cylindrical structure appeared in the middle of the lumen (t = 0; Fig. 5C, arrow) changed its structure to a Y-shape (t = 103 s), and then to a sheet-like structure (t = 271 s). This structure traversed through the vacuolar lumen and reached to the opposite side of the vacuole (t = 271 s; Fig. 5C). These dynamic structural changes were also observed in the root epidermal cells (data not shown).
Cytochalasin D and low osmotic pressure block vacuolar membrane dynamics
In the epidermal cells of the rosette leaf, the movement of the vacuolar membranes was blocked by cytochalasin D, which is known to disrupt the polymerization of actin filaments. The addition of cytochalasin D to the external solution greatly reduced the number of the moving sheet-like structures, in a dose-dependent manner (Fig. 6A). Cytoshalasin D also blocked the movement of the cylindrical structures, but did not disrupt its morphology. The microtuble-disrupting drugs, nocodazole and taxol, had no effect on the movement of any internal membrane structures (Fig. 6B). These data suggest that the dynamic movement of the internal vacuolar membrane is actin-dependent, but independent of microtubules.
Osmotic pressures also affected the intravacuolar structures. Whereas a complex intravacuolar structure was evident in the MS liquid medium (isotonic solution; Fig. 7A-1), it was hardly observable in the hypo-osmotic solution (distilled water; Fig. 7B-1). In the distilled water, a large single central vacuole occupied most of the cell volume. In contrast, in a hyper-osmotic solution (1 m mannitol), numerous complex intravacuolar structures were observed (Fig. 7C-1). A similar morphological change was observed in calli expressing the GFP-AtVam3p (Fig. 7A-2, B-2 and C-2). For a quantitative analysis of the morphological change, the cells of the calli were categorized into three types; cells without intravacuolar structures (Type 1), cells with the cylindrical structures (Type 2) and those with sheet-like structures (Type 3) (Fig. 7A-3, B-3 and C-3). The number of Type 1 cells in the hypoosmotic solution was greater than that in the isotonic solution (Fig. 7D). In contrast, in the hyper-osmotic solution, the number of Type 2 and 3 cells became greater than those in the isotonic solution (Fig. 7D). Thus, the intravacuolar membrane structures disappeared under hypo-osmotic conditions, whereas they became more complex under hyper-osmotic conditions.
The present study has demonstrated that the plant vacuole has complicated internal membranous architectures which include sheet-like and cylindrical structures. These structures are not static but dynamic membrane systems which are constantly being remodelled.
The sheet-like structure is a transvacuolar strand
Some sheet-like structures invaginated into the vacuolar lumen and even penetrated through the vacuole (Fig. 4). These structures were also observed using the GFP-δTIP or GFP-γTIP fusion proteins (Cutler et al. 2000; Hawes et al. 2001; Mitsuhashi et al. 2000). They are reminiscent of the transvacuolar strand, which was previously identified as a possible route of cytoplasmic streaming (Allen & Allen 1978; Nothnagel et al. 1981; Tominaga et al. 2000). This phenomenon is most pronounced in highly vacuolated cells and is generally assumed to be an efficient mixing and distribution process for solutes, vesicles and organelles (Kuroda 1994; Shimmen & Yokota 1994). Cytochalasin D blocked the streaming motion, but microtuble-disrupting drugs had no effect on it (Nebenführ et al. 1999). We showed that the movement of the sheet-like structure was also blocked by cytochalasin D, but not by nocodazole. Thus, the sheet-like structure is possibly the transvacuolar strand and involved in the cytoplasmic streaming.
The fluorescence intensity of the cylindrical structure observed in the mesophylle protoplast was twice as strong as that of the outer-vacuolar membrane (Fig. 3). This suggests that the intravacuolar membrane has a double membrane structure. It is unlikely that the GFP-AtVam3 fusion protein is just more densely localized to the intravacuolar membrane than to the outer-vacuolar membrane, because a rapid invagination of the intravacuolar membrane did not diminish the fluorescence intensity of GFP-AtVam3 (Fig. 5).
For the formation of the intravacuolar membrane structures, two putative models can be considered. In the first model (the autophagic model), a small vacuole is engulfed by a large vacuole, and remains enclosed in the vacuolar lumen (Fig. 8A). This model is supported by several pieces of evidence. In the developing seeds and the root tip cells of tobacco, small organelles (specifically recognized by anti-dark intrinsic protein (DIP) antibody) are predominantly located in the cytoplasm; however, as they develop, these organelles are taken up by PSVs (Jiang et al. 2000). In the sycamore suspension cells, autophagic vacuoles are formed in the peripheral cytoplasm under sucrose starvation, and are eventually expelled into a central vacuole (Aubert et al. 1996). In this context, the intravacuolar membrane shown in this study might be an intermediate structure of the autophagic process, being discontinuous to the outer vacuolar membrane.
Alternatively, the vacuolar membrane could be invaginated into the vacuolar lumen to form a sheet-like structure, which is then rounded (invagination model; Fig. 8B). The data from our time-lapse imaging strongly supports this notion. In addition, in the root epidermal cells, the intravacuolar membrane structure disappeared under hypo-osmotic conditions and became more complex under hyper-osmotic conditions (Fig. 7), indicating that the intravacuolar membrane and the outer-vacuolar membrane are continuous. In the guard-cell protoplasts, the plasma membrane is retrieved into the cytoplasm under hyper-osmotic conditions to regulate the membrane surface area (Homann & Thiel 1999; Kubitscheck et al. 2000). Thus, the intravacuolar membrane may function as a membrane reservoir and control the vacuolar surface area against various osmotic pressures.
Recently, the internal subregional structures of vacuoles designated as ‘bulbs’ have been reported in rapidly expanding cells of Arabidopsis (Saito et al. 2002). The bulb structures might be identical to the internal vacuolar structures reported here, although they were mainly observed in epidermal cells of cotyledon, while the exogenously expressed GFP-AtVam3p were detected in various cell types.
The appearance of the moving sheet-like structures raises a question of how such structures could be organized. One possibility is that the cytoskeletal filaments promote the formation of the moving structures from the cytoplasmic side. This seems likely because an actin-polymerization inhibitor, cytochalasin D, disrupts the movement of the sheet-like structures. Recently, some actin-interacting proteins have been characterized in various plants (Smertenko et al. 2001; Staiger 2000; Tominaga et al. 2000; Yokota & Shimmen 1999). One of these proteins, Rop GTPase—a member of the Rho family of Ras-related small GTP-binding proteins—is localized to the developing vacuoles. Rop GTPase is expected to regulate actin organization for the cell polarity and polar growth (Fu et al. 2001; Lin et al. 2001), and might be involved in the formation of the intravacuolar structures in an actin-dependent manner.
Inter- and intra-organelle signalling via membrane dynamics
Unexpected membrane structures, including protruding tubules and reticular networks, have recently been recognized in many other organelles. In the nucleus of animal and plant cells, deep tubular invaginations of the nuclear envelope have been suggested to play a possible role in nucleo-cytoplasmic transport and gene regulation (Collings et al. 2000; Fricker et al. 1997; Lui et al. 1998).
In tobacco mesophyll cells, the tubules interconnecting between chloroplasts have been identified and assumed to be involved in exchanging molecules within the interplastid communication system (Kohler et al. 1997; Shiina et al. 2000). Even the mitochondria have been shown to exist primarily as the reticular networks with tubular cristae (Gilkerson et al. 2000). Recently, the vacuole has been shown to contact with the nucleus via nuclear membrane protein, Nvj1p, in yeast (Pan et al. 2000). Finally, several types of tubular invaginations from the plasma membrane have also been observed (Fujimoto et al. 1995; Myers et al. 1993; van Deurs et al. 1996). All the available information leads to an implication that each organelle possesses distinct membrane tubing systems that may be actively involved in intra- and/or inter-organelle signalling.
Construction of GFP-AtVam3
The CaMV35S-sGFP(S65T)-NOS3′ GFP fusion vector and a T-vector, pMAT137, were kindly provided by Drs Y. Niwa (Shizuoka Prefectural University) and K. Matsuoka (Nagoya University), respectively. The coding region of the AtVam3 cDNA was amplified from an A. thaliana cDNA library by PCR with a set of oligonucleotides (5′-GGCTGTACAAGGGAGGTGGAGGTAGTTTTCAAGATTTAAATCAGGA-3′ and 5′-A TTGACGTCTTAAGATCATGAACACACACAAA-3′). The amplified fragment was sequenced and cloned into the BsrGI site of CaMV35S-sGFP(S65T)-NOS3′ at the downstream end of the GFP coding region in-frame to produce the GFP fusion protein. The resultant GFP(S65T)-AtVam3 construct was subcloned into pMAT137.
Syntaxin-related proteins have a single membrane-spanning region at the C-terminus, and the major part of the proteins located in the cytoplasm. Since we fused the GFP to the N-terminal end of AtVam3, GFP would not interfere with the topology of the fusion protein. In fact, it was reported that such GFP fusion does not interfere with the localization and the interaction of syntaxin and SNAP-25 (Xia & Liu 2001). It has also has been shown that GFP-Vam3p is fully functional in yeast vam3 mutant cells (N. Nakamura, personal communication). Therefore, GFP fusion is not expected to influence the topology, function and localization of AtVam3.
The constructed GFP(S65T)-AtVam3/T-DNA vector was introduced into Agrobacterium tumefaciens (strain C58C1 Rifr/pGV2260) by electroporlation. A. thaliana (ecotype Columbia) was infected with the A. grobacterium by in planta transformation (Bechtold & Pelletier 1998) and grown in Murashige–Skoog medium containing kanamycin (50 µg/mL). Transgenic plants expressing GFP-AtVam3p were selected under a conventional fluorescence microscope.
Sucrose gradient analysis
Cultured roots (15 g) of a selected transgenic A. thaliana line expressing GFP-AtVam3p were homogenized in a grinding buffer (12 mL of 50 mm HEPES-KOH, pH 7.5, 5 mm MgCl2, 5 mm EGTA, 250 mm sorbitol, 1 mm DTT, 1% polyvinilpyrolidone, 1% ascorbic acid and protease inhibitors [Roche]). The homogenate was centrifuged at 2000 g. The supernatant was collected and spun at 100 000 g for 40 min. The pellet was resuspended in a buffer (1 mL of 50 mm HEPES-KOH, pH 7.5, 2 mm MgCl2, 5 mm EGTA, 1 mm DTT and 13.7% sucrose) and loaded on to the top of a 20–50% sucrose gradient in 10 mL of 50 mm HEPES-KOH, pH 7.5, 2 mm MgCl2, 5 mm EGTA and 1 mm DTT. The gradient was centrifuged at 150 000 g for 13 h at 4 °C. Twenty fractions were collected from the top, and the sucrose concentration of each fraction was determined using a hand-refractometer. Equal volumes (10 µL) of each fraction were subjected to SDS-PAGE and subsequent immunoblotting. Quantitative analyses of the amounts of detected proteins were performed using image software (Scion Image). The Anti-SYP21 antiserum used in this study was generated against the specific peptide (Glu152-Lue-Asp-Thr-Glu-Ser-Leu-Arg-Ile-Ser-Gln162) coupled to the keyhole limpet haemocyanin. The other antibodies used were provided by our colleagues (see Acknowledgements).
Preparation of the protoplasts
Rosette leaves and roots (10 g) were cut into pieces and immersed in 0.5 m mannitol for 1 h at room temperature in the dark. The minced leaves and roots were then incubated in 60 mL of buffer (1% cellulase Onozuka R-10, 0.25% macerozyme R-10, 0.4 m mannnitol, 8 mm CaCl2 and 5 mm MES [pH 5.7]) for 2.5 h with gentle agitation, filtered through a nylon mesh, and resuspended in the protoplast medium (4.3 g/L Murashige and Skoog plant salt mixture [Sigma], 0.4 m sucrose, 100 mg/L inositol, 250 mg/L xylose, 460 mg/L CaCl2, 10 mg/L thiamine, 1 mg/L pyridoxine and 1 mg/L nicotinic acid. The pH was adjusted to 5.7 by the addition of KOH.
Confocal laser scanning microscopy (CLSM)
The microscope observation was performed using a confocal laser scanning microscope (Zeiss LSM5 PASCAL) equipped with green HeNe and argon lasers. A three-dimensional image was reconstructed from the continuous optical sections with a 0.20-µm step size, using the Zeiss LSM software, Image VisArt. For the treatments with cytochalasin D (Sigma), nocodazole (Sigma) and taxzol (Sigma), each stock solution (10 mm in DMSO) was prepared and added to the bath solution to give an appropriate final concentration.
Establishment of Arabidopsis calli expressing the GFP-AtVam3 fusion protein
Ten-day-old seedlings of transgenic Arabidopsis expressing the GFP-AtVam3 fusion protein were cut and placed on to an agar plate containing Murashige–Skoog (MS) medium supplemented with 2,4-D (1 mg/L) and kinetin (0.1 mg/L). After 30 days incubation at 23 °C, the calli derived from the seedlings were maintained in the MS medium; a small peace of the calli were transferred into a fresh medium every 3 weeks.
We thank Drs Y. Niwa (Shizuoka Prefecture University) and K. Matsuoka (Nagoya University) for kind gifts of the vectors listed in the Experimental procedures. We also thank Dr Y. Kasahara (Hokkaido National Agricultural Experiment Station) for providing an antibody against Oryza sativa plasma membrane H+-ATPase, Drs P. M. Ray and K. S. Dhugga for providing the anti-RGP-1 antibody, Dr A. Nakano for providing the Sar1 antibody, and Ms. H. Moritake for assistance with the picture drawing. This work was supported by grants from the Japanese Ministry of Education, Culture, Sports, Science and Technology: a Grant-in-aid for Encouragement of Young Scientist (to M.H.S.), and a Grant-in-aid for Basic Science Research (B) (to K.T.). S.H.Y. is a Research Fellow of the Japan Society for the Promotion of Science.