The plant Golgi apparatus is characterized by numerous individual cisternal stacks that appear more or less randomly distributed throughout the cytoplasm. While such a distribution is different to that commonly described for mammalian cells, it is more akin to that of insects such as Drosophila(1). However, in many plant cell types, it is apparent that Golgi stacks are motile and closely associated with the endoplasmic reticulum (ER) (2), while in others stacks can exist isolated from the ER (3). In this review, we consider the unique nature of the plant ER–Golgi interface and speculate as to how this is organized in a system that can be continually motile.
The interface between the endoplasmic reticulum (ER) and the Golgi apparatus is a critical junction in the secretory pathway mediating the transport of both soluble and membrane cargo between the two organelles. Such transport can be bidirectional and is mediated by coated membranes. In this review, we consider the organization and dynamics of this interface in plant cells, the putative structure of which has caused some controversy in the literature, and we speculate on the stages of Golgi biogenesis from the ER and the role of the Golgi and ER on each other’s motility.
Organization of the Plant ER–Golgi Interface
Anterograde protein transport from the ER to the Golgi takes place at specialized ER exit sites (ERES) and is mediated by the Sar1p guanosine triphosphatase (GTPase) and its exchange factor Sec12 plus the coat protein (COP)II coat comprising the heterodimeric Sec23/Sec24 and Sec13/31 complexes (4,5). Homologues of those proteins have been identified in plants (6–12), and the COPII machinery appears to be conserved in plants as overexpression of Sec12 and expression of Sar1 mutants disrupt anterograde protein transport (13–15).
What are the COPII Carriers?
COPI vesicles have been identified (12,16), yet the nature of COPII-coated carriers remains elusive. Although COPII proteins have been located (11,17–19) (Figure 1A–C and Table 1), to date, there is little hard evidence for the existence of COPII vesicles in plants. Such vesicles have been described in meristematic cells in high-pressure-frozen freeze-substituted material and in unicellular algae (20), but the specificity of the Sar1 antibody used to identify COPII components was not demonstrated. Budding profiles on the ER have been reported in BY2 cells, but immunolabelling failed to identify COPII components (21).
|Protein||Accession number||Reference||Function or putative function|
|Coat and related proteins|
|Sec12||At2g01470||(6)||Sar1p GTP exchange factor on ER membrane|
|AtSARA1a||At1g09180||(18,33)||Initiation of COPII coat assembly (Sar1 isoforms)|
|Sec23||At3g23660||(9,12)||Sar1p GTPase-activating protein on COPII coat|
|Sec24||At3g07100||(12)||Cargo binding protein on COPII coat|
|Sec13||At2g30050, At3g01340||(11,24)||COPII coat protein|
|Sec16||At5g47480, At5g47490||Not yet characterized (24)||Definition of ERES?|
|Regulatory proteins and fusion proteins|
|RabD2A||At1g02130||(31)||Regulation of ER–Golgi transport|
|Sed5 (Syp 31)||At5g05760||(30)|
|Matrix and movement|
|Myosin XIK||At5g20490||(82–84)||Tail domain severely perturbs Golgi movement|
|AtCASP||At3g18480||(46)||Putative tethering factor|
|Golgin-84 2 isoforms (GC1 and GC2)||At2g19950, At1g18190||(47)|
|AtP115 (GC6)||At3g27530||(47)||Tethering between ER and Golgi?|
|TRAPP1 and COG complexes||Various||(45)||Tethering between ER and Golgi. Organization of transferases?|
|ERD2||At1g29330 (L23573)||(2,90)||Putative H/KDEL receptor|
The possibility of membrane connections between the ER and the Golgi, be they permanent or transitory, tubular or direct, has been debated in several recent reviews and has been described in many electron micrographs (3,22–24). It does however need to be emphasized that the canonical view that all transport between ER and Golgi must be mediated by COPII vesicles is being challenged (reviewed in 4). Indeed, much of the early evidence of such vesicles came from in vitro reconstituted yeast systems (25), which may not reflect the in vivo state. In mammalian cells, it has now been suggested that the vesicular tubular carriers between the ER and the Golgi may be generated by the fusion of COPII vesicles or from tubules or specialized domains at the ERES (26). Such ER-to-Golgi carriers produced directly from the ER may be COPII dependent in formation but may not involve COPII vesicles (27). Thus, there is no a priori reason to assume that the COPII vesicles must be the carriers across what must be a very small divide in the plant ER–Golgi interface in many tissues such as leaves and hypocotyls. However, the question then needs to be answered as to what is the situation in cells where the Golgi bodies appear separate from the ER, such as the isodiametric less vacuolated cells in meristems? Is there a long-range transport of carriers from ER to Golgi or do the Golgi bodies dock onto ERES, retrieve cargo and then separate from the ER (28)?
Proteins Acting at the Plant ER–Golgi Interface
No matter what the physical nature of the ER-to-Golgi vector might be, transport between the two organelles requires, at some stage, fusion between the membranes of donor and acceptor compartments. A number of ER and Golgi SNARE proteins have been described for Arabidopsis(29), and four of these, Sec 22, Memb11, Bet 11 and Sed5 (Syp 31), have been suggested to play a role at the ER–Golgi interface (30). The whole transport process between the two organelles also appears to be regulated by the plant homologue of the Rab1 GTPase (RabD2A) (31). The Arabidopsis homologues of these regulatory proteins are summarized in Table 1 together with other key proteins that are implicated to act at the plant ER–Golgi interface. The putative functions and importance of COPII isoforms and coat-related proteins, Golgi matrix proteins and movement proteins will be discussed in more detail in the following sections of this review.
What is the Function of COPII Isoforms?
Plants possess genes encoding for multiple isoforms of COPII proteins (23,24,32,33), and analysis of their expression profiles showed tissue specificity for some of them, such as the Arabidopsis Sar1 isoform At1g09180 that appears to be expressed exclusively in stamen and pollen (24). COPII isoforms can have different intracellular locations and might also differ in their function, as the Arabidopsis Sar1 isoform AtSARA1a was found to be more cytosolic than AtSARA1b, and AtSARA1b affected ER export less when both isoforms were expressed in a GTP-locked form (18). Little is known about the function of other COPII isoforms, but it has been described recently that four isoforms of human Sec24 exhibited preferential binding to different cargo transport motifs, which could increase the complexity of cargo recognition (34).
As Sec12 and most Sar1 isoforms locate to the ER membrane, it has been suggested that the whole ER surface might be competent for ER export (24), as originally postulated by Boevink et al. (2). However, a combination of photobleaching studies on moving Golgi (19) and the colocalization of NtSar1Bt (19) and of Sec24 (12,17) with Golgi markers on the ER supports the motile export site complex hypothesis that involves the Golgi stacks plus ERES moving in synchrony on or with the ER surface (19,22), a situation very different to the more static exit site reported in mammalian cells (35). Tissue specificity of Sar1 isoforms might explain the contradictory evidence regarding the relationship of Golgi bodies to ERES as previous studies were undertaken with different Sar1 isoforms (24). The tobacco leaf isoform NtSar1Bt (13) was observed in the cytosol and in punctate structures colocating with the Golgi marker ERD2-green fluorescent protein (GFP) (19), whereas the NtSar1 isoform isolated from BY-2 cells (8) labelled the ER and punctate structures that only partially colocated with the Golgi marker ManI–red fluorescent protein (24). Perhaps, the discrepancy in location of the Sar1 isoforms reflects differences in the relationship between Golgi bodies and ER in leaf and root tissue, as Golgi bodies in tobacco leaf epidermal cells moved with the surface of the ER (36), whereas Golgi stacks in root cells seem to be able to dissociate from the ER (3).
Differentiation of ER Exit Sites
The exact processes leading to formation and differentiation of ERES are still unknown. The first step of COPII coat assembly is the recruitment of Sar1 by Sec12 to the ER (5). In Pichia pastoris, however, COPII coat formation remained restricted to a specific ER domain termed the transitional ER (tER) even when Sec12 was dispersed over the ER membrane (37). Therefore, it is likely that additional proteins are required to establish the identity of ERES, maybe by forming an ER membrane scaffold structure (35). It has been speculated that Sec16 could be part of such a scaffold (38) as upon expression of GTP-locked Sar1 in animal cells, Sec16 accumulated together with Sec24 and Sec31 not only on juxtanuclear membranes previously described as clustered ERES (39) but also on additional peripheral structures on the ER membrane (38). The authors suggested that those juxtanuclear membrane structures resembled clustered free COPII carriers and that Sec16 might constitute a more reliable ERES marker. Clearly, the choice of the marker protein is critical in studying the relationship between ERES and Golgi bodies as in animal cells, it has been suggested that the majority of total COPII proteins expressed labelled free tubules and vesicles (40). The Arabidopsis genome encodes two putative Sec16 isoforms (24), but their function in plants still needs to be established.
A putative ERES scaffold might also incorporate cis-Golgi matrix proteins that could play a role both in differentiation of ERES and in the nucleation and regulation of Golgi stack formation (41,42). Golgins are large coiled-coil proteins implicated in the tethering of vesicles or other membrane compartments and could provide a first level of vesicle recognition and specificity before SNARE-mediated vesicle fusion (43). Several golgin homologues have recently been identified in plants (44–47), and two of them, AtCASP and golgin-84, have been located to the cis-Golgi (47) (Figure 1D–F). A p115/Uso1p homologue was also described (47), and this is a matrix protein that has been implicated in a tethering role at the cis-Golgi (43,48). It has been shown that p115 tethers COPII vesicles to Golgi membranes (49). Another task of p115 is to form bridging tethers by linking giantin (present in recycling COPI vesicles) to GM130 (present on cis-Golgi membranes) (50). Recent results have shown that p115 forms, together with Rab1, a SNARE complex (51). These matrix proteins are therefore ideal candidates for tethering factors at the Golgi/export site complex to the ER membrane and also for organizing the export site complex during Golgi biogenesis (42).
Birth of a Golgi
A major step during the cell cycle is the partitioning of different organelles between daughter cells during the division process. Organelles with endosymbiotic origins such as mitochondria and chloroplasts cannot form de novo and are replicated by division. The Golgi apparatus however, with a close structural and functional relationship to the ER, displays different mechanisms of inheritance (52).
Different Models of Golgi Biogenesis
Two ways by which Golgi stacks could multiply are discussed in the literature: either by de novo formation from the ER or through fission of an existing stack (53). In animal cells, mitosis leads to a complete breakdown of Golgi stacks during prophase and remnant mitotic vesicular Golgi clusters may be formed (54); however, the relationship of these small clusters with the ER, and the importance of the ER in the organization of Golgi remnants and in the reconstruction of the Golgi, is a hotly debated topic (55,56). It has been reported that such Golgi clusters formed in telophase are segregated in pairs between daughter cells and fuse just before completion of cytokinesis (57). Alternatively, Golgi stacks may form de novo either from ERES, as in the yeast P. pastoris(58–61), or from mitotic vesicular clusters (56,62,63). A model for Golgi disassembly and reassembly during mitosis in mammalian cells has been proposed in which sequential inactivation of Sar1 and Arf1 leads to disruption of ERES and redistribution of Golgi enzymes to the ER, whereas sequential activation of those two proteins initiates Golgi reformation (55). In Toxoplasma gondii, an intracellular protozoan parasite, the Golgi apparatus is a single copy organelle that grows by lateral extension and undergoes medial fission during cell division (64). Studies on other protozoan parasites like Trypanosoma brucei have shown de novo synthesis, suggesting that both models of Golgi biogenesis can exist in protists (65–67).
Golgi Biogenesis in Plants and Algae
It is well documented in plant cells that Golgi bodies can reform on washout of Brefeldin A (BFA) from treated material (68), indicating that the ER has the capacity to generate Golgi de novo. Hanton et al. (17) have shown, using Sec24 as a marker, that de novo export site formation can be cargo induced, indicating that perhaps Golgi bodies can form in response to cargo production if export sites and Golgi stacks do behave as a single unit. During mitosis and cytokinesis in plants, Golgi bodies and membranes do not disaggregate as in mammalian cells. Whether secretion per se stops is not known, but from late anaphase onwards, the Golgi apparatus is highly active in producing new cell wall membrane and polysaccharide for the phragmoplast region (69,70). Data on Golgi inheritance in higher plant cells are contradictory. Golgi stacks were reported to double during metaphase in onion root meristems (71), while duplication was claimed to occur during cytokinesis in synchronized cultures of Catharanthus roseus(72). More recently, a tomographic analysis of Arabidopsis shoot meristem cells demonstrated a doubling of the number of Golgi stacks in G2 just prior to mitosis (73). Cells with high secretory activity such as pollen tubes and root hairs seem to produce large numbers of new Golgi stacks depending on their task and growth status, and this is not related to the cell cycle or division (74).
Recent studies in the single-celled alga Chlamydomonas noctigama, which has non-motile Golgi stacks around the nucleus (75), and in BY-2 cells with mobile Golgi (21) have shown that de novo Golgi biogenesis and Golgi fission can take place within the same system (Figure 2). Experiments were based on a complete deconstruction of Golgi stacks with BFA and reformation after BFA washout. Initially, in both systems, vesicle clustering was a first indication of Golgi reformation. After the first fusion events, mini-Golgi stacks were formed, starting at 200 nm diameter with up to five cisternae (Figure 2A–C,H,I). An increase in ERES number on the tER accompanied the early reformation events in C. noctigama. Mini-Golgi stacks displayed a very early cis-to-trans polarity, and in BY-2 cells, this could also be observed in Golgi stacks with a 250 nm diameter. In both studies, there was no clear indication that COPII-coated vesicles or membrane took part in early stages of biogenesis. Although budding sites on the ER were observed (Figure 2A), they did not label with antibodies to the Sec13 component of the COPII coat. From immunogold labelling, it was however shown that COPI proteins may play a role in the early membrane fusion events forming initial cisternae.
After stack formation, Golgi cisternae increase in size. The growth seems to be related to an increased number of budding sites on the ER in C. noctigama(75), and there appeared to be an increased formation of budding profiles on the ER in BY-2 cells with mobile Golgi stacks. In both Chlamydomonas and BY-2 cells, reforming Golgi stacks continued to grow to double the size of those in control cells and then divided vertically in a cis-to-trans direction (21,75). There is however no indication as to what triggers the overgrowth of the stacks or induced their division, but we have to hypothesize on the existence of molecular regulators of Golgi stack size. Could this be a putative role for some of the matrix proteins?
In mammalian cells, Golgi matrix proteins, mainly GM130 and p115, have been implicated in Golgi biogenesis (76), and as discussed earlier, homologues of Golgi matrix proteins have been described for plant cells (45,47), although a GM130 homologue does not exist. However, the p115 homologue is most likely situated towards the cis-Golgi and is a good candidate for a tether involved in early Golgi biogenesis. In Figure 3, we propose a sequence of events that may be involved in the birth of an individual Golgi stack from the ER. First, an exit site differentiates on the ER surface through interplay of Sec16, Sec12 and Sar1 (Figure 3A). This may also involve cis-Golgi matrix or tethering factors. A COPII-coated bud forms from the ER membrane and is tethered to the ER through the proto-Golgi matrix (Figure 3B). The bud or buds grow to form a tubulovesicular complex, which contains COPI buds, vesicles and SNARES, and is surrounded by a matrix (Figure 3C). Whether this is fed by direct membrane connections to the ER or by vesicles is still to be ascertained but quickly differentiates into a mini proto-Golgi stack with structural characteristics of both cis- and trans-faces, including clathrin-coated buds (Figure 3D). At some stage, membrane-bound Golgi enzymes are transferred into this structure from the ER and are anchored in the correct cisternae as the stack continues to mature. This whole complex is most likely motile with the ER surface (see subsequently).
Is the ER–Golgi Interface Implicated in Motility?
A large number of plant cells display highly dynamic organelle movement. However, the requirement for organelle motility is not completely understood, although environmental stresses such as light affecting chloroplast and nuclear positioning (77,78) and fungal infection affecting peroxisome location (79) have been implicated. Studies in tobacco epidermal and BY2 cells using GFP technology have shown that numerous individual Golgi bodies display a range of motilities from remaining stationary, slow to fast, plus uni- and bidirectional movements (2,80). These movements appear to occur over the ER, in what was coined ‘stacks on tracks’, where Golgi bodies are ‘stacks’ on the ER/actin ‘tracks’. Obviously, one key question to be asked is what is the additional contribution of moving Golgi stacks to the secretory process over static stacks considering that in undifferentiated meristematic cells, there appears to be less Golgi movement? A question that is as yet unanswered.
Movement of Golgi and ER
Owing to the intricate nature of ER-to-Golgi trafficking (antero- and retrograde transport) and the membrane equilibrium required to maintain Golgi homeostasis, both compartments are functionally and possibly structurally linked (see previously). Interesting questions raised from these observations relate to whether the Golgi body is then simply a subdomain of the ER (3) and not an organelle in its own right, and whether the movement of the two organelles is also intimately linked or co-ordinated.
Comparisons between ER and Golgi body movements have resulted in the following observations: Golgi bodies are associated with the three-way junctions and move along the tubular ER network (2); occasionally, Golgi bodies break free from the ER ‘track’, and in some cases, the ER tubules remodel and follow the path of the Golgi body (81). Photoactivation studies of ER membrane protein shows that the Golgi bodies move in a similar direction to the underlying activated ER membrane (36). These observations have resulted in the development of a model where ER–Golgi exit site and Golgi movement is a co-ordinated process and resulted in questions relating to potentially shared or distinct motors, whereby the ER drags the Golgi bodies or vice versa through the action of a motor protein either on the ER and/or on the Golgi bodies.
Cytoskeletal depolymerization experiments have indicated that both ER and Golgi body movements are in higher plants dependent on actin and not on microtubules (2,80,81). Comparative studies of ER and Golgi movements in cells expressing a fluorescent marker for the actin network confirm a close association between the organelles and the actin (81). These observations indicate that movement is driven either by myosins or by actin polymerization/depolymerization. Only recently, it was shown that expression of truncated variants of 3 of the 17 postulated Arabidopsis myosins (XIK, XIE and MYA2) severely perturbs Golgi, peroxisome and mitochondrial movement (82–84) (Figure 1I–L) in tobacco epidermal cells. XIK also plays a role in ER movement and remodelling (Sparkes et al. unpublished data), and a further myosin tail fragment partially labelled Golgi bodies (85). The effects of XIK on organelle movement were further corroborated through Arabidopsis T-DNA insertional mutant and RNA interference knockdown studies, although it was not discussed how the phenotype from overexpression of a tail domain having a dominant-negative effect could be the same as that from knockdown of the same protein (84). Interestingly, although XIK and XIE affect movement of several organelles, they do not appear to be completely colocated with these organelles and based on fluorescent markers do not appear to affect the global architecture of the actin cytoskeleton (83). It therefore remains to be seen how these motors control movement. While it appears that microtubules are not involved per se in Golgi body movement, other cytoskeletal interacting proteins such as kinesin 13A (86) and an actin-binding protein KATAMARI 1/MURUS3 (87) have been identified, which could be potential components of a complex required for an actin–microtubule linkage. Alternatively, kinesin 13A may be required for maintaining Golgi stack integrity or division as several studies in mammals have highlighted interplay between myosin and kinesins in Golgi motility and maintenance (88).
Another interesting question pertaining to Golgi stack movement is how the ER–Golgi linkage and the cisternae themselves are held together during rapid movement? Considerable shear forces must be exerted on the Golgi stack during the vectorial movement, which besides simple membrane connections might require tethering factors to hold the system together. Obvious candidates are Golgi matrix proteins as discussed above (47).
While the majority of motility and ER–Golgi interface studies have been carried out in tobacco epidermal and BY-2 cells, it is important to note that there are tissue and cell type differences; actively dividing cells in the root meristem display slower movement (Figure 1G) compared with elongating cells (Figure 1H), which could be because of cell volume restrictions, cell volume/surface area ratio requiring slower movement or altered metabolic demand in these cells or a different more remote relationship with the ER (see previously). Quantitative analysis of Golgi body movement in roots versus leaf epidermal cells in Arabidopsis also indicated reduced motility in roots compared with leaves (84). To compare potential effectors on movement, we have analysed microarray data to try and determine whether movement rates correspond to altered regulation of components involved in the ER–Golgi interface in specific tissue types. Analysis indicates that expression of the genes detailed in Table 1 (potentially) implicated in the ER–Golgi interface is fairly uniform in all the tissues assessed. There are a few exceptions: TMF and a Sec16 isoform are upregulated in pollen, Sed5 expression across all tissues is higher than other SNAREs (Table 1) and myosin XIE is upregulated in stamens. Whether these differences reflect functional attributes in Golgi motility and the requirement to tether or hold the cisternae together, for example, or have any bearing on the level of protein present in these tissues remains to be answered. It is important to note however that different cell types have different cell volume/cytoplasm ratios owing to the volume occupied by the vacuole. Therefore, a cell with a low ratio may require protein upregulation not necessarily in order for increased function but to maintain the protein/cytoplasm ratio for interaction with binding partners within the cytoplasm.
In 1996, we posed ‘stacks of questions’ on the working of the plant Golgi apparatus (89). Twelve years later, with the completion of genome sequences and the application of fluorescent proteins to live cell imaging, we are starting to answer some of these questions. For plants, it is becoming clear that although at the molecular level they express many of the proteins described at the yeast and mammalian ERES, the structural organization and dynamics of this interface may be very different. Plant Golgi stacks appear to have the ability to form ‘de novo’ at exit sites, which in many tissues are closely apposed to the stacks, and this may be in response to demands imposed by cargo and/or growth conditions, although they can also divide by fission. The exit site/Golgi complex is highly motile on the ER surface and may move with the ER membrane on an actin scaffold somehow driven by myosin motors. We still do not understand why this movement is necessary, but one could envisage a moving Golgi stack more readily shedding its secretory vesicles than a static stack. Perhaps, in a further 12 years, another stack of questions will be answered.
We thank Benoit Binctin for movies of Arabidopsis root tissue. Some of the work described in this study was supported by Biotechnology and Biological Sciences Research Council (BBSRC) and Leverhulme Trust grants to C. H.