Multiple Vacuoles in Plant Cells: Rule or Exception?


*Lorenzo Frigerio,


It is generally accepted that plant cells can contain multiple vacuoles with different functions, for example lytic vacuoles with lysosome-like properties and protein storage vacuoles for reserve accumulation. Recent data call into question the generality of this theory. In this study, we review the published evidence for the existence of multiple vacuoles. We conclude that the multivacuole hypothesis is valid for a number of cases, but care should be taken before assuming that it applies universally.

Plant vacuoles are defining compartments of plant cells. They display a variety of functions, ranging from the maintenance of turgor pressure, the degradation of cellular components, the sequestration of ions and secondary metabolites such as pigments and toxic compounds and the accumulation of reserve proteins in storage organs (1,2). Acidic, hydrolytic, lysosome-like vacuoles are generally termed lytic vacuoles (LV), whereas less acidic organelles with the capacity to accumulate reserve material are named protein storage vacuoles (PSV). Although the term ‘vacuole’ is used to indicate the large, late compartment of the anterograde plant secretory pathway, it is difficult to ascribe its multiple, and often contrasting, functions to a single organelle. How can the same vacuole be both a dustbin and a safe haven for storage proteins?

When, more then a decade ago, separate vacuoles were identified in the same cells (3), the idea of a single, multifunctional vacuole was supplanted by the view that contrasting vacuolar functions may coexist in distinct organelles within a cell. The existence of separate vacuoles conveniently explained why there exists more than one anterograde route to target proteins to vacuoles, as indicated by the existence of distinct vacuolar sorting signals (VSS) present on ‘lytic’ or ‘storage’ vacuolar cargo proteins (4). In recent years, therefore, the sentence ‘plant cells contain at least two types of vacuoles’ has become a favourite opener of many articles about vacuolar protein sorting.

There are now indications that the distinction between pathways, signals and ultimately between different vacuoles may not be as clean cut as initially hoped. Some recent data hint at the possibility that multiple vacuoles in plant cells may be the exception rather than the rule (5,6). This semi-heretical view has already sparked some controversy (7–9).

In this article, we review the multivacuole hypothesis in the light of the available evidence. We maintain that this hypothesis, and the working models stemming from it, needs to be operationally useful rather than just intellectually attractive. Our starting assumption is the following: in its most inflexible interpretation, ‘plant cells contain at least two types of vacuoles’ means that if we were to take a snapshot of a plant tissue, we would find that most cells within it do indeed contain functionally distinct vacuolar compartments. We posit that if this is not so, then the multivacuole hypothesis only applies to a relatively small number of examples or to well-documented developmental transitions, in which case its general applicability is limited.

Historically, the distinction between vacuoles with different functions such as PSV or LV has relied on the different pH of their lumen, as highlighted by staining with pH- or acidic protease-sensitive dyes (10–12), on the use of specific antibodies against particular isoforms of tonoplast intrinsic proteins (TIPs) present on their membranes (3,13) or on the presence of electron-opaque storage proteins in the lumen (14). The general consensus stemming from extensive TIP isoform analysis is that α-TIP is enriched in PSV, γ-TIP in LV (13) and δ-TIP in storage vacuoles present both in seed or in vegetative cells (15). More recently, the use of fluorescent proteins, fused to VSS and directed to the vacuolar lumen (5,10) or fused to TIPs and directed to the tonoplast (5,16,17), has been added to the arsenal of molecular tools, possibly, it can be argued, complicating the picture.

We begin our discussion of the available evidence for multiple vacuoles in plant cells by starting with very clear examples of such coexistence. We then focus on less clean-cut cases.

No Doubt

Multiple vacuoles in mature cells

There is limited but very clear evidence for the coexistence of separate vacuoles in specialized cells. The most visually striking example is found in mesophyll cells of Mesembryanthemum crystallinum (ice plant) (18). Under salt stress, these cells contain two separate, similarly sized large vacuoles: a neutral vacuole with the function of sequestering salt and an acidic, neutral red – accumulating one, with the capacity to store malate from CAM photosynthesis during the night and remobilize it during the day (18). As stressed by the authors, the ice plant is quite exceptional as it combines CAM physiology with the capacity to sequester salt (18). The presence of distinct vacuoles is a remarkable adaptation that guarantees the safe coexistence of two essential metabolic functions. Nothing is known about the TIP distribution on the membranes of these separate vacuoles.

A second example of visually distinguishable vacuoles in the same cell was found in protoplasts from barley aleurone cells (11). When protoplasts, which contain PSV, were treated with ABA or gibberellic acid, smaller vacuoles appeared, which were physically separate from PSV. These secondary vacuoles (SV) contained α-TIP in their tonoplast, like their neighbouring PSV, but no storage proteins. The use of fluorescent ratiometric probes showed that SV are slightly more acidic than PSV. It was argued that SV may be functionally equivalent to lysosomes and help degradation of PSV-stored protein. However, their protein content has not been elucidated.

Multiple vacuoles have also been detected at the end of the plant’s life cycle. In mesophyll and guard cells of senescing soybean and Arabidopsis leaves, small, acidic compartments have been identified that contain the senescence-specific cysteine protease SAG12 (12). These compartments are rather small (0.5–0.7 μm), but the presence of the vacuolar (H+)-pyrophosphatase on their membrane grants them the denomination of senescence-associated vacuoles (SAV). Interestingly, the SAV tonoplast contains no γ-TIP, indicating that SAV tonoplast biogenesis is separate to that of the central vacuole. The SAV have a lower pH than the central vacuole and, like the previously mentioned SV of aleurone (11), are probably involved in apoptosis.

Multiple vacuoles during developmental transitions

The parenchymatous cells in the cotyledons of many seeds, especially those of the legumes, synthesize and accumulate copious amounts of N-rich proteins during their development (19). These proteins – mainly of the globulin type – are deposited into a PSV, which shortly before desiccation subdivides many times to form numerous small protein bodies (14). Until the publication of Hoh et al. (20), it was generally considered that the PSV in these cotyledon cells arose simply through the transformation of previously existing LV-type vacuoles in much the same way as PSV seem to develop in vegetative tissues (21). However, in pea cotyledons, as monitored by subcellular fractionation and in situ immunogold electron microscopy (EM), there is a clear developmental sequence of vacuole populations with LV being succeeded by PSV. As a result, for a substantial period during pea cotyledon development, there are two easily identifiable vacuole types in the same cell. Interestingly, the PSV seems to have an autophagy-like ontogeny, starting out as tubules/cisternae that encircle and then engulf the pre-existing LVs. Storage protein deposits in these tubules contain a BiP cognate pointing to their endoplasmic reticulum origin (20).

Exactly the same kind of structures have been detected in developing embryos of Arabidopsis thaliana and Medicago truncatula (Figure 1). However, the time frame during which the two vacuole populations coexist is much shorter and does not extend into the late torpedo and bent cotyledon stage of embryo development. This is probably one of the reasons why Otegui et al. (22) did not detect LVs in their study of proteolytic processing of seed storage proteins in Arabidopsis embryos. However, another reason is that γ-TIP is not expressed in Arabidopsis embryos during seed maturation (Figure S1; see also subsequently), making it impossible to monitor for LV using γ-TIP antisera or fluorescent γ-TIP constructs, unless the latter are ectopically expressed.

Figure 1.

PSV develop in an autophagic-like manner. During embryo development in both Arabidopsis thaliana (A and B) and Medicago truncatula (C), the PSV is first recognizable in thin sections as a tubular structure containing flocculent electron-opaque deposits mainly attached to the outermost lying membrane. This structure is often seen encircling the pre-existing LV so that two adjacent tonoplasts (TP) can be distinguished (indicated by the TP bars). Scale bars: (A and C), 320 nm and (B), 120 nm.

The Jury’s Out

TIP markers

A paper that contributed greatly to the acceptance of the multivacuole hypothesis was that of Paris et al. (3). These authors performed a confocal immunofluorescence study on root tip squashes of barley and pea seedlings using antisera directed against α- and γ-TIPs and antisera against barley lectin and aleurain as luminal markers for PSV- and LV-type vacuoles. The data were internally consistent in the sense that α-TIP and barley lectin colocalized to punctate structures, as did γ-TIP and aleurain. In addition, the authors presented micrographs of two types of cells in their squash preparations: one showing separate α- and γ-TIP-marked punctate organelles and the other with larger ‘chambers’ harbouring both TIPs. The interpretation given for this was that the first type of cell originated nearer the meristem where two separate vacuole populations exist and the second came from a region further away from the meristem where the two vacuole types had fused to form a single type of vacuole. Recently, however, a reinvestigation of these two types of root tip by immunogold microscopy using a very similar fixation protocol (6) did not corroborate the data of Paris et al. (3). Cells in and around the meristem in both roots contained vacuoles with storage proteins (globulins in the case of peas and lectin in the case of barley) and both α- and γ-TIPs in their tonoplast. LV-type vacuoles, that is, vacuoles with only γ-TIP and lacking storage proteins, were not found. Is there an explanation for the contradiction between the Paris et al. (3) and Olbrich et al. (6) papers? A closer look at two sets of data in the Paris et al. paper raises issues as to both the origin of the cells investigated and the reliability of the TIP antisera in detecting vacuoles. First, the two cell types described above (one containing small vacuoles with different TIPs and the other with larger vacuoles and both TIPs) were of the same size, which one would not expect if increased vacuolation is an indicator of cell enlargement. Second, vacuoles in the cells in the stele reacted negatively towards both TIP antisera [an observation confirmed by Olbrich et al. (6)]. Despite these inconsistencies, it is clear from a subsequent paper from the Rogers group on pea root tips (13) that LV (i.e. with only γ-TIP) represent only about 1% of the vacuoles present. The great majority of the vacuoles (over 90%) are, as confirmed by Olbrich et al., of a hybrid type with combinations of α + δ-TIPs (28%), α- + g- + δ-TIPs (48%) and γ- + δ-TIPs (17%). Jauh et al. (15) had previously shown that δ-TIP is characteristic for the tonoplast of PSV in vegetative tissues (15). Thus, the observation that the storage globulins present in the PSV in pea roots are in an unprocessed form is in line with the notion that the primary vacuole in pea and barley root tips is a vegetative PSV marked by δ-TIP but together with α- or γ-TIP. The evidence for the central vacuole arising from the fusion of different vacuolar types in roots is therefore rather slight.

TIPs in Arabidopsis

The use of TIPs to identify different vacuoles, and their biogenesis, in A. thaliana is complicated by the fact that its genome encodes 10 TIP isoforms (23) (Figure S1): 3 γ-TIP, 3 δ-TIP and 1 of each, α-, β- and ɛ-TIP, and ζ-TIP. Two of these, ɛ-TIP (At2g25810) and δ-TIP3 (At5g47450) seem to be preferentially expressed in the root. Other TIPs also seem to follow a tissue-specific expression pattern, with γ-TIP3 (At4g01470) and ζ-TIP (At3g47440) specifically enriched in the flowers and floral organs. δ-TIP2 (At4g17340) and ɛ-TIP (At2g25810) have a second peak of expression in the flowers and floral organs (Figure S1). α- and β-TIP expression is specifically restricted to seed maturation, whereas δ-TIP1 and δ-TIP2 (At3g16240 and At4g17340) and γ-TIP1 and γ-TIP2 (At2g36830 and At3g26520) seem to be expressed only during earlier stages of seed development, but their transcript levels decrease sharply during seed maturation (Figure S1). It is not known whether these isoforms coexist on the same tonoplast.

With respect to the roots, the situation is rather intriguing. In contrast to what observed experimentally in pea and barley roots, α-TIP transcripts appear to be absent. Immunogold EM is negative when using an α-TIP polyclonal antibody raised against the HQPLAPEDY peptide sequence present on both α- and β-TIP (C. Viotti, University of Heidelberg, Germany, unpublished data). This negative result is in agreement with the results published by Hunter et al. (5) who could not find expression of α-TIP–yellow fluorescent protein (YFP) constructs under the control of their own promoter in Arabidopsis roots. Together, these results are in accordance with in silico expression studies (24). From the latter, two of three γ-TIP isoforms, γ-TIP1 and γ-TIP2 (At2g36830 and At3g26520) (Figure S1), are expressed at high levels in Arabidopsis roots, but neither western blotting nor immunogold EM with a polyclonal γ-TIP antiserum – which works positively on pea and barley roots and Arabidopsis leaves – succeeded in detecting γ-TIP protein (C. Viotti, unpublished data). This is surprising because this antiserum was raised against the carboxy-terminal peptide sequence THEQLPTTDY, which is conserved between γ-TIPs from Arabidopsis, pea, radish and rice (13,15). It should however be added that based on developmental transcriptomic data, no transcripts for the γ- or δ-TIP isoforms studied in Jauh et al. (13) and used to generate peptide antibodies seem to be present in root tips of Arabidopsis(24) This was also confirmed by expressing fusions between the same γ- and δ-TIP isoforms and YFP under control of their native sequences (5). The absence of γ-TIP expression in Arabidopsis root tips was also reported in a previous study (25). The remaining – and possibly Arabidopsis root-specific – isoforms mentioned above (Figure S1) still await investigation. It is therefore possible that the study of a different selection of TIP isoforms may well produce a more consistent picture of different vacuolar populations in roots.

Regarding the existence of separate vacuoles in photosynthetic tissues, α-TIP immunoreactive structures have been detected in mesophyll protoplasts of various plant species including Arabidopsis (but only when α-TIP was ectopically expressed) (26). This led to the conclusion that PSV are constitutively present in vegetative cells (26). Recently, however, fusions between α-TIP and YFP expressed in Arabidopsis leaves and protoplasts only seemed to localize to the large central vacuole (5), as previously shown in tobacco BY-2 cells (27). Moreover, when α-TIP–YFP was expressed under the control of its native promoter, expression was limited to seed tissue and the protein was never detected in leaves (5). This is in accordance with the localization of transcript levels for native α-TIP, which are restricted to the seed (24,28).

Fluorescent proteins in the vacuolar lumen

The first use of a fluorescent reporter for the plant vacuolar system was published in 1998 (10). In this study, it was shown that green fluorescent protein (GFP) fused to the C-terminal vacuolar sorting signal (ctVSS) of tobacco chitinase (GFP–Chi) localized either to the central vacuole or to smaller structures in transfected tobacco protoplasts, depending on the type of cells. In chloroplast-poor cells, the fluorescent puncta were clearly separate from the central vacuole, whereas in chloroplast-rich cells, most of the signal was detected in the large central vacuole. In both cases, the GFP-positive vacuoles did not stain with neutral red, hinting at the fact that a low pH and GFP fluorescence are mutually exclusive (10). Another reporter, based on the sequence-specific vacuolar sorting signal (ssVSS) of aleurain (Aleu–GFP), also gave punctate or central vacuolar patterns but with the opposite abundance ratio (29). The authors concluded that different VSS target GFP to either a neutral organelle (chitinase ctVSS) or an acidic one (aleurain ssVSS). This separation was clearly maintained during vacuole regeneration from evacuolated protoplasts where the neutral vacuole marker persisted in small structures distinct from the reforming central vacuole. In some cases, the neutral vacuoles eventually fused with the central vacuole ultimately producing a large hybrid vacuole (29). This is another clear example of separate vacuoles coexisting during a developmental, albeit artificially induced process.

The same GFP reporters were subsequently used in transgenic Arabidopsis plants (30). While Aleu–GFP tended to localize to the central vacuoles in most tissues, GFP–Chi marked central vacuoles to a lesser extent, more frequently highlighting smaller, punctate structures. This distinction was taken to indicate that in some cells, there may be different vacuoles as labelled by the two markers, although the authors acknowledged that relative GFP stability in the vacuolar lumen of different cell types may result in the observed phenotypes. Indeed, in the same year (2003), it was confirmed that GFP is degraded in the central vacuole by cysteine proteases at the acidic pH of its lumen when maintained in the light (31). Treatment with protease inhibitors, inhibitors of vacuolar proton pumps or incubation in the dark resulted in vacuolar GFP becoming visible (31). These data indicate that the initial observations performed in Arabidopsis tissues (30), and more recent studies using the same GFP-based markers (32), may have underestimated the extent of targeting to the large central vacuole. The small, bright organelles highlighted by these GFP constructs (30) may well be separate vacuoles, but the question remains as to whether an equally large proportion of the reporter has also reached the central vacuole. In addition, the lack of colocalization with known markers for post-Golgi compartments does not allow for informed guesses as to the nature of the labelled structures.

The recent finding that the monomeric red fluorescent protein (RFP) is stable and fluoresces both in the apoplast and in the vacuolar lumen (33) has led to the production of RFP-based vacuolar reporters. By fusing a ctVSS (from phaseolin) or a ssVSS (from proricin) to a secreted version of RFP, Hunter et al. (5) recently found that, regardless of the type of VSS present, the RFP reporters reach the same vacuole in seeds, leaves and roots. When RFP was replaced with GFP, the same pattern of localization was detected but only when transgenic plants were kept in the dark as instructed by Tamura et al. (31) (Figure 2). Therefore, we conclude that – as far as the targeting capacity of these two specific VSS is concerned – there is consistency between RFP- and GFP-based reporters, but care must be applied when working with the more degradation-prone GFP.

Figure 2.

Both RFP and GFP fused to VSS highlight central vacuoles in transgenic Arabidopsis tissues. Confocal images of tissues of transgenic Arabidopsis expressing either GFP or RFP fused to the ctVSS of phaseolin (xFP-AFVY) or the sequence-specific sorting signal of proricin (L-xFP) (5, 38). The plants expressing the GFP reporters were kept in the dark for 2 days before observation (31). Scale bars: (A–D), 15 μm; (E–H), 20 μm; (I–L), 40 μm and (M–P), 10 μm.

One strong argument in favour of the multivacuole hypothesis is that different sorting routes and mechanisms exist for vacuolar proteins (4,34) For example, there is recent, clear evidence of a separate pathway for sorting proteins that carry ctVSS. Elegant genetic screens in seeds (35) and vegetative tissues (36,37) have identified factors that are responsible for vacuolar targeting of ctVSS- but not ssVSS-bearing cargo. For example, using a genetic screen that cleverly exploits the secretion (or otherwise) of the signalling protein CLAVATA3, the Raikhel group identified a ctVSS-specific SNARE, VTI12 (36), and the shoot meristem identity gene TFL1(37) as factors responsible for ctVSS-mediated vacuolar sorting. It is tempting to speculate that this pathway leads to PSV-like organelles in vegetative tissues. Moreover, the colocalization of TFL1 with the δ-adaptin of AP3, which in mammalian cells is involved in the biogenesis of lysosome-related, pigment-storing organelles, further strengthens this hypothesis (37). However, such PSV-like organelles have not yet been identified as separate entities from the vacuole, and the central vacuole remains so far the location of reference for the mutant screen (36,37).


From the evidence presented, we draw the following conclusions:

  • • It is undeniable that more than one vacuolar type can be found in the same cell – but only in very particular tissues or plants and under particular physiological and developmental constraints.
  • • When multiple vacuoles are present during developmental transitions, their coexistence tends to be short lived. In developing/maturing embryos, for example, the PSV swiftly replaces the LV, and both structures only coexist for a few days or less.
  • • While it would be naïve to expect that phylogenetically related plants (e.g. dicots) differ widely in their subcellular compartmentalization, the use of multiple plant species to compile a unifying model of vacuolar distribution and biogenesis has not helped to provide firm handles for further research. The recently increased focus on Arabidopsis is likely to channel the efforts of the community and accelerate progress. This is of course assuming that Arabidopsis is a good model species for the study of vacuolar biogenesis and distribution!

In conclusion, a careful analysis of the evidence available questions the generality of the multivacuole hypothesis that, in its most literal reading, is restricted to a relatively small number of cases. Therefore, when working with model systems, it may be advisable to ascertain how many vacuoles are present before assuming (in a curious reversal of Occam’s razor) that there are more than one.


We thank Sara Di Benedetto for providing micrographs of plant cells expressing the GFP reporters and Corrado Viotti for sharing unpublished data. We thank Robert Spooner and Alessandro Vitale for critical reading of the manuscript.