Expression of DMP1-eGFP in Nicotiana benthamiana epidermis cells induces membrane remodeling
To investigate intracellular targeting of DMP1 we agroinfiltrated a 35S:DMP1-eGFP construct into tobacco leaves. Interestingly, the fusion protein displayed a highly dynamic and temporally changing fluorescence pattern (Figure ). Two to three days post infiltration (dpi), the first fluorescence signals became visible and labeled the boundaries of the cells and spherical structures inside the lumen of the central vacuole (Figure a). Until five dpi the fluorescence pattern changed and the cells underwent membrane remodeling to various degrees (Figure b). Two days later the majority of cells exhibited severely remodeled endomembranes, giving the cells a “foamy” appearance (Figure c). These membrane remodeling patterns and time courses were highly reproducible with only little fluctuation in severity. To investigate whether the observed membrane remodeling events are due to overexpression of DMP1-eGFP by the strong, constitutive 35S promoter, we also expressed the protein by the endogenous, senescence-specific DMP1 promoter. This promoter led to somewhat weaker expression levels in tobacco and induced a somewhat weaker membrane remodeling phenotype, that was reproducible and comparable to the 35S promoter-induced phenotype though. Thus, the observed membrane remodeling patterns are not merely overexpression artifacts.
Figure 1 Temporal dynamics of DMP1-eGFP fluorescence patterns in tobacco epidermis cells. Representative overviews of tobacco epidermis cells expressing DMP1-eGFP at 2 dpi (a), 5 dpi (b) and 7 dpi (c). Scale bar, 40μm.
We classified the course of endomembrane remodeling into five stages. Stage 1 is characterized by well-defined fluorescence signals along the cell walls (Figure a, arrow) and at spherical structures located inside the lumen of the vacuole (Figure a, arrowhead). Three to four dpi the cells typically enter stage 2 where they begin to display extended membrane sheets within the cytoplasm reminiscent of ER cisternae (Figure b, arrow) and bulbs (Figure b, arrowhead). Stage 3 is distinguished by large tubular and reticulated structures forming a network reminiscent of cortical ER (Figure c, arrow). Also spherical bodies are visible (Figure c, insets), but unlike the spherical structures in stages 1 and 2 they appear to be located in the cytoplasm, and large membrane sheets crossing and thereby compartmentalizing the central vacuole emerge (Figure c, arrowheads). Figure d shows a cell in transition from stage 3 with its distinctive tubular structures (Figure d, arrow) to stage 4 with its typical “foamy” membrane meshwork (Figure d, arrowheads). In stage 4 a great deal of the central vacuole is filled with this “foamy” membrane mesh (Figure e, arrow). Some residual tubular structures are still present, and occasionally enigmatic, sponge-like structures appear (Figure e, inset). In the terminal stage 5 the vacuole breaks down by vesiculation (Figure f). This stage was rarely observed because the cells appear to die rapidly after vacuole disintegration and only a minor fraction of stage 4 cells enter stage 5. Remarkably, in spite of strong membrane remodeling the cells seem to stay viable for a prolonged period of time without entering vesiculation. Figure g shows the approximate fractions of cells in stages 1 to 5 at different times after infiltration.
Figure 2 DMP1-eGFP induces membrane remodeling inNicotiana benthamiana. Transient overexpression of DMP1-eGFP in tobacco epidermis cells results in distinct fluorescence patterns classified into five stages: stage 1 (a), stage 2 (b), stage 3 (c), stage 3 to stage (more ...)
To characterize the membrane structures labeled by DMP1-eGFP we subsequently performed colocalization experiments with various membrane markers.
Stage 1: The tonoplast located DMP1-eGFP induces the formation of bulbs
The first DMP1-eGFP fluorescence signals were observed at the cell periphery and in spherical structures two days after infiltration (Figure a). Upon co-infiltration DMP1-eGFP clearly colocalized with the tonoplast marker TPK1-mRFP (Figure b,c, arrowheads), but not with the plasma membrane marker mRFP-MUB2 (data not shown). TPK1-mRFP was largely excluded from the spherical structures (Figure b,c, arrows) which supposedly are identical to the “bulbs” reported by Saito and colleagues [10
] as they are comparable in size, motility and fluorescence intensity. Overlap between DMP1-eGFP and TPK1-mRFP fluorescence at the bulbs was extremely rare and only partial. Some regions of the bulbs were labeled with either DMP1-eGFP or TPK1-mRFP (Figure d, arrows), suggesting different membrane properties and rapid exclusion of TPK1-mRFP from the bulbs. As γ-TIP-mCherry did not lead to proper fluorescence signals in tobacco [1
] it could not be used as an alternative tonoplast/bulb marker. We therefore studied DMP1-eGFP infiltrated tobacco leaf epidermis cells by transmission electron microscopy. In DMP1-eGFP expressing epidermis cells we observed a significantly higher number of bulbs (Figure l) than in mock-transformed cells, supporting the notion DMP1-eGFP induces formation of these bulbs. DMP1-eGFP was never observed in Golgi vesicles (Figure e,g,h) and was largely excluded from the ER (Figure e,f,h) which had a normal tubular morphology. The same result was obtained by using the integral fusion protein RFP-p24 instead of the luminal YFP-HDEL as ER marker ( Additional file 1
: Figure S1).
Figure 3 Stages 1 and 2. Co-expression of DMP1-eGFP (a) and the tonoplast marker TPK1-mRFP (b) shows in stage 1 colocalization at the vacuolar membrane (c) and occasional partial overlap at bulbs (d). Co-expression in stage 1 of DMP1-eGFP (e), YFP-HDEL labeling (more ...)
Stage 2: Reorganization of the ER - transition from tubular elements to cisternae
Stage 2 is characterized by the appearance of bulky cisternae in the cytoplasm that strongly resemble cortical ER observed under certain conditions (see Discussion), while the bulbs and tonoplast labeling from stage 1 are still retained (Figure b). The ER localization of DMP1-eGFP was verified by co-expression with RFP-p24 (Figure i,j,k). We also occasionally observed RFP-p24 signals in bulbs (Figure j,k, arrows). This might either indicate mislocalization of the ER marker due to overexpression or some dysfunction of the ER during stage 2.
Stage 3: De novo formation of a cortical ER-derived network inside the cytoplasm and vacuolar sheets inside the vacuole
Stage 3 is marked by different membrane remodeling events. Most conspicuously is the labeling of novel tubular structures which do not colocalize with the different markers used. In stage 3 DMP1-eGFP and RFP-p24 both decorate the whole ER network composed principally of cisternae (Figure a,b,c). DMP1-eGFP additionally decorates another tubular mesh from which RFP-p24 is excluded (Figure a,b,c, insets). However, both networks share the same overall pattern, indicating either physical connection or differential labeling of the same entity. Strikingly, over time DMP1-eGFP and RFP-p24 progressively segregate. While DMP1-eGFP initially colocalizes with RFP-p24 in the ER cisternae ( d,e,f, arrowhead), the tubular structures mostly dissociate from the ER network (Figure d,e, f, insets). In late stage 3, when first vacuolar sheets and “foamy” structures emerge (Figure c and g, arrows), DMP1-eGFP is almost undetectable in the ER network labeled by YFP-HDEL (Figure g,h). This time course suggests that the tubular structures derive directly from the ER and coincide with a progressive exclusion of DMP1-eGFP from the ER. The tubules labeled only by DMP1-eGFP form an interconnected network throughout the cytoplasm (Figure a,g,k and l), are homogeneous in diameter and show a smooth and relaxed appearance (Figure a,d,g,k,l and c), and are - in contrast to the repetitive polygonal structure of the cortical ER network - often tightly packed and peculiarly folded (Figure c, insets and 4
l, inset). Large swollen spherical formations reminiscent of ER cisternae are often observed at the intersection of DMP1-eGFP-labeled tubules (Figure g and k, arrows and inset). In late stage 3, isolated tubules are also found (Figure g, arrowheads and K, arrowhead) whose occurrence coincides with the presence of cytosol-located vesicles (Figure g, empty arrowhead, k, inset and l). These vesicles and the isolated tubules likely derive from the DMP1-eGFP-labeled network by fission events.
Figure 4 Stage 3. Co-expression of DMP1-eGFP and ER markers during stage 3 shows cells with partial colocalization (a-c and d-f) and cells lacking colocalization (g-j). Accordingly, in cells expressing DMP1-eGFP alone, ER with cisternal morphology can be distinguished (more ...)
As mentioned above, vacuolar sheets crossing the lumen of the vacuole and first “foamy” membranes appear in stage 3 and accumulate gradually (Figure c). The density of vacuolar sheets correlates with a progressive loss of the DMP1-eGFP labeled network. Moreover, the tubules were occasionally found tightly associated with these vacuolar sheets (Additional file 1: Figure S2
). These observations suggest a connection between these two structures. Golgi vesicles appeared to be unaffected during stage 3 (Figure i) suggesting proper ER-Golgi transport despite extensive remodeling of the ER.
Stage 4: Formation of “foamy” membrane structures inside the vacuole
Transition from stage 3 to 4 is indicated by the appearance of “foamy” membrane formations that coincide with a decrease in tubular structures (Figure d). The “foamy” membranes likely derive from accumulation of vacuolar sheets. At this time no DMP1-eGFP signals are detected in the ER anymore (Figure a,c,d and e,g,h1
) which appears to be compressed into interstices (Figure c, g, arrows) and junctions of the “foamy” membranes (Figure c,g, arrowheads). The junctions contain different organelles such as peroxysomes or mitochondria (Figure i, arrowhead and k) as found in transvacuolar strands [11
]. Confocal fluorescence microscopy (Figure h1
) and electron microscopy (Figure i,j,k) consistently revealed that the vacuolar sheets and “foamy” membranes are double membranes. DMP1-eGFP (Figure e) and TPK1-mRFP (Figure f) do not perfectly colocalize as shown by separation of the two fluorescence signals (Figure 5
). The distance between the two fluorescence peaks is about 200
nm to 300
nm (Figure 5
membrane segments 1, 2 and 3) which would allow small organelles to pass through. The double-membrane topology is corroborated by the observation of ER squeezed between the two membranes of a membrane sheet (Additional file 1
: Figure S3). Occasionally however, perfect colocalization is observed which might indicate localization of both fusion proteins at both membranes (Figure 5
membrane segment 4). Under electron microscopy the double membranes appear more closely stacked (Figure i and k). However, this may be a fixation artefact and not reflect the situation in vivo
. Membrane sheets consisting of a single membrane were never observed by EM. In 70
nm thin cross-sections the double membranes completely crossed the lumen of the vacuole, confirming that they correspond to the vacuolar sheets and not to transvacuolar strands (TVS) as the latter are unlikely straight and oriented in parallel to the section cut across the whole vacuole. TPK1-mRFP is often excluded from regions within foamy membrane structures (Additional file 1
: Figure S4). Interestingly, these areas are located at contact zones between adjacent sheets within foamy membrane structures.
Figure 5 Stages 4 and 5. Co-expression of DMP1-eGFP (a), TPK1-mRFP (b) and YFP-HDEL (C, false-colored) shows colocalization of DMP1-eGFP and TPK1-mRFP and dissociation from the ER (d). The ER is compressed into interstices and junctions formed by foamy membrane (more ...)
During stage 4 intriguing sponge-like flat structures arise (Figure e, inset, Additional file 1
: Figure S5). TPK1-mRFP is excluded from these areas (Additional file 1
: Figure S5) which is reminiscent of the observations in individual bulbs (Figure d) and within foamy structures (Additional file 1
: Figure S4). We hypothesize that these sponge-like structures represent residual TPK1-mRFP-free membrane domains derived from bulbs and vacuolar sheets. Additionally we observed the formation of crystalloid ER (Figure 5
Stage 5: Vesiculation of the vacuole and the ER leading to cell death
Six days post infiltration some cells with severe vesiculation of endomembranes also display overall intracellular disintegration, indicating the onset of cell death (Figure f). As in stage 4, DMP1-eGFP only labels the tonoplast and foamy membrane formations but not the ER (Figure l,m,n). The ER is not reticulated but highly vesiculated (Figure m,n,o, arrow). The vacuolar and foamy membranes also appear to vesiculate more heavily than in stage 4 and form smaller vesicles (Figure o, arrow and p). Despite the obvious breakdown of the ER, the integrity of the nuclear membrane (Figure m, arrowhead and o) and Golgi vesicles (Figure q) is still retained. The Golgi marker, which is partially secreted to the apoplast ( i), indirectly indicates in Figure p that the plasma membrane, not labeled by DMP1-eGFP, is still intact (Figure p,q,r, arrows). The massive vesiculation of endomembranes was confirmed by electron microscopy (Figure s).
Expression of DMP1-eGFP in transgenic Arabidopsis thaliana reveals dual ER/tonoplast localization
As dual tonoplast/ER localization of a protein is unusual we aimed to determine if dual tonoplast/ER localization and induction of membrane remodeling by DMP1 overexpression is conserved in transgenic plants. In Arabidopsis
plants carrying a 35S:DMP1-eGFP
transgene, seven days after sowing (DAS), bulbs and tonoplast localization is observed in young cotyledons (Figure
a). Five days later (12 DAS) the number of bulbs decreases (Figure b) and at 18 DAS no more bulbs were visible (Figure c). This time course of bulb development is consistent with previous observations using γ-TIP as marker [10
]. In addition to accumulation in bulbs strong DMP1-eGFP signals are observed in the ER as well as in ER bodies in all these stages (Figure a, b, c, arrows and inset). The ER bodies vanish as the cotyledons age, corroborating earlier reports [12
]. Accordingly, in cotyledons of Arabidopsis
DMP1-eGFP is dually targeted to the ER and the tonoplast, but overexpression of DMP1-eGFP does not affect the morphology and development of the ER and the tonoplast in this organ. ER bodies are also labeled by DMP1-eGFP in hypocotyl cells somewhat later in development (Figure g). In rosette leaves, we observe an intense, leaf-age independent accumulation of DMP1-eGFP in the ER (Figure d). In addition, protoplasts prepared from rosette leaves also show some tonoplast localization, confirming the dual localization seen in cotyledons (data not shown). During developmental leaf senescence and even more pronounced during dark induced leaf senescence (Figure h), individual cells or leaf areas show massive vesiculation reminiscent of the cellular breakdown process during stage 5 in tobacco. Thus, in Arabidopsis
leaves DMP1-eGFP appears to be similarly associated in disintegration of the ER and the vacuole by vesiculation as in tobacco (Figure h, arrows).
Figure 6 Dual ER/tonoplast localization of DMP1-eGFP in stably transformedA. thaliana plants. Overexpression of DMP1-eGFP in in young emerging Arabidopsis cotyledons (7 DAS) leads to labeling of the tonoplast and bulbs (a), the ER network (a, arrow and inset) (more ...)
Expression of DMP1-eGFP from the DMP1 promoter in Arabidopsis highlights formation of boluses within the ER and fragmentation of ER and tonoplast during senescence
To scrutinize whether dual localization in Arabidopsis
is an artifact by overexpression of DMP1-eGFP by the CaMV 35S
promoter, we expressed the same fusion protein from the native DMP1
promoter in transgenic plants. In accordance with the senescence-associated activity of the DMP1
], DMP1-eGFP fluorescence is only detectable in mature, early and late senescing rosette leaves, senescing cauline leaves, senescing silique walls and roots (Figure ). In mature-to-early senescing rosette leaves, DMP1-eGFP strongly accumulates in the ER and to a lesser extent in the tonoplast. However, the tonoplast signals are hardly distinguishable from the ER signals (Figure a). ER bodies are occasionally observed (Figure b). Formation of boluses resembling the eponymous protein aggregates reported by Griffing [13
] and vesiculation events are observed in rosette leaves (Figure c,e,f), cauline leaves (Figure g) and silique walls (Figure h) undergoing natural senescence. Darkening of single rosette leaves (Figure i) or whole plants (Figure d and j) lead to similar events. In individual cells disintegration of the ER is discernible (Figure c). In these cells the junctions of the ER tubules seem to swell (arrow) and vesiculate (arrowhead). We suggest that bolus formation precedes vesiculation of the ER, though it cannot be excluded that the two processes represent two different fates for cells undergoing senescence. Indeed, neighboring cells of the same type undergoing induced senescence can display different degrees of bolus formation and vesiculation (Additional file 1
: Figure S6). In other cells, fragmentation of the tonoplast is obvious (Figure d and e, arrows) with occasional persistence of residual ER network (Figure e, arrowhead), suggesting a close succession of the two vesiculation processes. Figure 7
show ER which already underwent vesiculation (arrowheads) and fragmentation/vesiculation of the tonoplast (arrow), indicating that ER breakdown precedes tonoplast breakdown. Tonoplast vesiculation is more rarely observed than ER vesiculation during developmental or dark induced senescence. Tonoplast breakdown is presumably only a short-lived phase as it rapidly and irreversibly leads to cell death. The persistence of the nuclear membrane (Figure 7
, open arrowhead) in spite of progressed ER breakdown is reminiscent of the events in tobacco during stage 5.
Figure 7 DMP1-eGFP fluorescence patterns during development inA. thaliana. In mature/early senescing rosette leaves DMP1-eGFP expressed from the native DMP1 promoter localizes in the ER (a) and occasionally in ER bodies (b). Vesiculation of the ER in rosette leaves (more ...)
In roots, vacuolar localization of DMP1-eGFP is obvious in the cortex of root tips (Figure k-o). In accordance with the current view of vacuole biogenesis, the emerging cells near the root tip contain several vacuoles differing in size (Figure d) whereas the older cells in the elongation zone have fewer vacuoles or a single central vacuole (Figure o). In these cells the plasma membrane is also labeled (Figure n, arrows), which is supposedly due to a truncated isoform of DMP1 (to be published elsewhere). In the phloem bundles, the subcellular localization could not be determined because of the small size of cells (Figure p). The ER network was also visible in roots, highlighting once more the ability of DMP1-eGFP to target multiple subcellular membrane systems (Figure q).