Brefeldin A: Deciphering an Enigmatic Inhibitor of Secretion

The fungal macrocyclic lactone brefeldin A (BFA) has proved to be of great value as an inhibitor of protein trafficking in the endomembrane system of mammalian cells (Sciaky et al., 1997). BFA has also often been used as an inhibitor of secretion and vacuolar protein transport in plant cells, but just exactly how BFA achieves these effects has been a matter of debate for some time (e.g. Satiat-Jeunemaitre et al., 1996;Staehelin and Driouich, 1997). The apparently broad spectrum of BFA responses, combined with a lack of understanding of the primary target of BFA, has made it difficult to develop a coherent explanation of BFA effects, which in turn has led to a series of misconceptions that riddle the plant literature. In the last few years, there has been a tremendous increase in our understanding of the molecular targets and primary effects of BFA in mammalian and yeast systems (see below). On the plant side, several papers have recently been published that, by applying new tools and technologies, have shed fresh light onto the BFA problem (Baluska et al., 2002; Brandizzi et al., 2002; Emans et al., 2002; Ritzenthaler et al., 2002; Saint-Jore et al., 2002). In thisUpdate, we wish to evaluate these new findings and explore whether we are any closer to solving the dilemma of “what BFA really does.” It is now well established that the target of BFA in mammalian cells is a subset of Sec7-type GTP-exchange factors (GEFs) that catalyze the activation of a small GTPase called Arf1p (Jackson and Casanova, 2000). Arf1, in turn, is responsible for the recruitment of coat proteins (coatomer, also called COPI; as well as clathrin via the adaptor complex AP-1) to membranes, resulting in the formation of transport vesicles (Scales et al., 2000). Arf1p and BFA-sensitive GEFs are localized to the Golgi apparatus of mammalian and yeast cells (Spang et al., 2001). Thus, one of the earliest and best characterized effects of BFA in nonplant organisms is the loss of COPI coats from the Golgi apparatus (Kreis et al., 1995). In recent years, it has become apparent that the molecular targets for BFA also exist in plant cells. The Arabidopsis gene GNOMencodes an Arf GEF that falls into the subgroup of BFA-sensitive Sec7-like proteins (Steinmann et al., 1999). Although the intracellular localization of GNOM is not yet known, most data are consistent with its Golgi localization, in accordance with other members of the Gea/GNOM/GBF subfamily (Jackson and Casanova, 2000). Plants also possess Arf1p, coatomer, AP-1, and clathrin, all of which also localize to the Golgi apparatus (Pimpl et al., 2000), suggesting that the machinery that mediates vesicle transport in the endomembrane system is basically the same in all eukaryotes (Nebenführ and Staehelin, 2001). Moreover, the expression of a dominant-negative mutant form of Arf1p inhibits protein transport to and through the Golgi apparatus of Arabidopsis protoplasts in a BFA-like manner (Lee et al., 2002). Thus, it is reasonable to assume that BFA has similar primary effects in plants as in animals. This prediction has been borne out by a recent study in which the effects of BFA on the intracellular localization of COPI coat proteins in tobacco (Nicotiana tabacum) bright yellow-2 (BY-2) suspension-cultured cells were examined (Ritzenthaler et al., 2002). In untreated cells, about one-half of the protein detected by an antibody raised against the γ subunit of Arabidopsis COPI (Movafeghi et al., 1999) was associated with Golgi stacks. This situation changed rapidly after BFA addition, and after only 5 min, essentially no anti-Atγ-COPI labeling was detected on the Golgi (Ritzenthaler et al., 2002). Thus, a rapid release of COPI coats from the Golgi into the cytosol in response to BFA is a universal feature of all eukaryotic cells, and suggests that the molecular target and mode of action of BFA is highly conserved. In other words, all BFA effects in plants most likely are consequences of an initial inhibition of Arf1p and its BFA-sensitive GEF. As documented extensively for mammalian cells, a striking effect of BFA is the complete redistribution of Golgi enzymes into the ER (e.g. Sciaky et al., 1997). Recent work with different sets of Golgi markers has firmly established that this BFA response also occurs in plants (Boevink et al., 1998; Lee et al., 2002; Ritzenthaler et al., 2002; Saint-Jore et al., 2002). In these studies, green fluorescent protein (GFP)-tagged Golgi enzymes have been used to examine the fate of this organelle in living cells. This has allowed researchers to follow the fate of Golgi stacks in real time and show convincingly the redistribution of Golgi proteins into the ER network (for example, see the following video clips:http://www.brookes.ac.uk/schools/bms/research/molcell/hawes/BFAonGolgi.aviand http://web.utk.edu/∼nebenfur/pubs/BFA-video.mov). This phenomenon suggests, at face value, a stimulation of retrograde transport from the Golgi to the ER. However, it is well established that COPI-coated vesicles are involved in the normal recycling of escaped ER residents back from the Golgi (Letourneur et al., 1994). Therefore, BFA-inhibition of COPI recruitment should actually prevent Golgi→ER transport, rather than stimulate it. This apparent contradiction can be resolved by focusing on other proteins involved in Golgi→ER transport, the so-called tethering factors and SNARE complexes (Shorter et al., 2002). Whereas tethering involves a variety of peripheral membrane proteins and the activity of Rab-GTPases, fusion is dependent upon the interaction of a v-SNARE on the vesicle with its cognate t-SNARE on the target membrane (Rein et al., 2002). In vivo and in vitro data point strongly to SNAREs being concentrated in COPI-coated vesicles at the time of budding (Hay et al., 1998). However, when COPI vesicle formation is blocked by BFA, the population of v-SNAREs in the Golgi membranes presumably will rise temporarily, thereby increasing the chances of direct and uncontrolled fusion between neighboring Golgi and ER membranes (Elazar et al., 1994). Thus, we postulate that in plant as in mammalian cells, BFA leads to a breakdown of the physical separation of ER and Golgi compartments by allowing membrane fusion to occur in the absence of prior vesicle formation and not by a stimulation of the normal retrograde vesicular transport pathway. Effects of BFA on membrane trafficking between ER and Golgi. In untreated cells (left), export from the ER (pink) is mediated by COPII-coated vesicles (green). Fusion of these anterograde vesicles depends on anterograde v-SNAREs on the vesicles (red triangles) and t-SNAREs on the cis Golgi (red lines). Anterograde transport (at least to the PVC) continues on the trans side of the Golgi with clathrin-coated vesicles (crossed blue lines). Retrograde transport within the Golgi and from the Golgi to the ER depends on COPI coats (red) and retrograde v- and t-SNAREs (black triangles and lines). In BFA-treated cells (right), all vesiculation at the Golgi stops due to the inhibition of Arf1. As a result, retrograde v-SNAREs remain exposed on the Golgi and the cisternae fuse directly with the ER. Cisternal maturation continues in the presence of BFA so that early Golgi compartments assume a more trans-like morphology. At the same time, the later cisternae and the TGN are lost to the cytoplasm and, eventually, to the BFA compartment. COPII vesicle formation at the ER is initially not inhibited by BFA, but anterograde vesicles may no longer be able to fuse with the maturing cis cisterna, thus effectively blocking ER-to-Golgi transport. It is interesting that the events lying between the BFA-induced loss of COPI coats from the Golgi on the one hand, and fusion of Golgi membranes with the ER on the other hand, are strikingly different between mammalian and plant cells. These distinct, secondary responses most likely reflect the different organization of the Golgi apparatus in the two organismal groups. In mammalian cells, the loss of COPI coats is followed by extensive formation of membrane tubules along microtubules (Sciaky et al., 1997; Hess et al., 2000), which eventually fuse with the ER. In contrast, BFA does not lead to extensive tubulation of the plant Golgi apparatus. Instead, at least in tobacco BY-2 cells, Golgi stacks initially maintain their morphology except for an apparent loss of cis cisternae (Ritzenthaler et al., 2002; see also below). The remaining Golgi cisternae eventually fuse with the ER, forming hybrid ER-Golgi stacks, or in some cases, they fuse laterally with other Golgi cisternae, resulting in oversized Golgi complexes that are continuous with the ER (Ritzenthaler et al., 2002). These unusual hybrid structures display ER and Golgi characteristics, are unique to plants, and presumably result from a heightened coherence of the plant Golgi cisternae that is not found in their mammalian counterparts. The recent description of a BFA-induced ER-Golgi hybrid compartment in some plant cells (see above) is apparently at odds with the older literature that described a disintegration of the Golgi apparatus into a conglomerate of tubules and vesicles termed the “BFA compartment” (Satiat-Jeunemaitre et al., 1996). It is possible that these two responses are tissue (or even species) specific, but it is more likely that the apparently different fates of the Golgi membranes in these cell types are simply a question of which markers were monitored. In particular, those markers demonstrating Golgi-ER fusion all reside in earlier parts of the Golgi: Gm-Man1-GFP is limited to the cis Golgi (Nebenführ et al., 1999), and AtERD2-GFP has been shown to be distributed over the entire Golgi stack (Boevink et al., 1998). In contrast, JIM84 antibodies that recognize the Lewis-a epitope, a complex oligosaccharide that is added to glycoproteins very late in the Golgi (Fitchette et al., 1999), do not label the ER in BFA-treated cells (Henderson et al., 1994). It is interesting that two sialyltransferase-based markers seem to give contradictory results. The sialyltransferase (ST)-based GFP marker, which consists of a GFP fused to the transmembrane domain of rat sialyltransferase, has been used to demonstrate Golgi-ER fusion in tobacco leaf epidermal and BY-2 cells (Boevink et al., 1998; Saint-Jore et al., 2002). The ST-myc version, on the other hand, where the small myc epitope had been fused to the full-length enzyme, accumulates in BFA compartments in Arabidopsis callus (Wee et al., 1998). This apparent discrepancy might be explained by the slightly different distribution of these heterologous proteins in plant Golgi stacks. The ST-GFP protein is broadly distributed over the trans one-half of the stack (Boevink et al., 1998), whereas the ST-myc protein is found only in the trans-most cisterna and the trans-Golgi network (TGN) (Wee et al., 1998). In support of this, we point to the recent data of Jin et al. (2001) who have demonstrated that the Golgi-associated dynamin homolog ADL6, a protein probably involved in fission of clathrin-coated vesicles at the TGN, colocalizes exactly with the JIM 84 epitope, but only partially overlaps with ST-RFP (red fluorescent protein)-labeled Golgi stacks. In Arabidopsis root tips, ADL6 accumulates in BFA compartments, as would be expected from this distribution. We would also expect the GDP-Man transporter GONST1 from Arabidopsis to reside in the trans-most Golgi cisterna or the TGN because it is also present in BFA compartments of onion (Allium cepa) epidermal cells (Baldwin et al., 2001). The above analysis of cisternal markers in plants reflects very much the situation in mammalian cells where TGN residents, e.g. ST and TGN38, have been shown not to redistribute to the ER upon BFA treatment (Chege and Pfeffer, 1990; Ladinsky and Howell, 1992). Instead, the remaining TGN membranes fuse/aggregate with elements of the endocytic pathway (Wood et al., 1991; Reaves and Banting, 1992). Taken together, these examples from the plant and animal literature strongly suggest that in response to BFA, the Golgi apparatus essentially splits horizontally, with most cisternae being absorbed into the ER, whereas the TGN, and possibly also the trans-most Golgi cisterna, contribute to the BFA compartment. This means that BFA compartments as well as ER-Golgi hybrids are always formed as a result of BFA treatment. Because Golgi-ER fusion is dependent upon SNARE-SNARE interactions, it also means that only those cisternae having the appropriate SNAREs will be fusion competent. As mentioned above, these SNAREs are normally concentrated into COPI vesicles, so one might speculate that the position of the “Golgi split” is indicated by cisternae no longer engaged in COPI vesicle budding (Fig. 1). In animal cells, “nonclathrin”-type vesicles have been shown to bud off every cisterna except for the last plus the TGN, where clathrin coats are found exclusively (Ladinsky et al., 1999). It is tempting to speculate that the difference between these two cisternae marks the “point of no return”: Up to can be via COPI vesicles back to cisternae eventually the and after this, it is only a matter of into clathrin or vesicles and As a this would also be the point where the Golgi is BFA treatment. It should be possible to this in plants by following the action of BFA on cells GFP that reside at different in the The effect of BFA to BFA compartments and ER-Golgi hybrids in the same plant cell has actually been described in a few example, the of the ER which in this followed the formation of BFA compartments, has been by labeling root cells for ER proteins and JIM84 (Henderson et al., 1994). In a similar a subset of BY-2 cells formed BFA compartments in to the ER-Golgi hybrid (Ritzenthaler et al., 2002). new of BFA responses the question the formation of compartments has not been described more and was only in a small of the BY-2 cells examined (Ritzenthaler et al., 2002). As indicated this may have simply from the of markers in most In addition, it is possible that of the effects be small and difficult to Moreover, a in the of trans-Golgi elements to the BFA compartment is likely to be the of cisternal maturation within the Golgi This the TGN that would into the BFA compartment by of trans cisternae and and Cisternal to the early of BFA treatment (Ritzenthaler et al., 2002). In their BFA compartments would only of the TGN and a trans cisterna or two that were off early These may be small and may It is also that very Golgi stacks have a of cisternae and off more of the remaining fuse with the ER. It is interesting that a of apparently cell markers also in after BFA treatment. These of the transport machinery et al., et al., 2002; and et al., as well as a membrane and (Baluska et al., 2002). Although the intracellular membranes these have thus been characterized only at the light it is highly likely that they are BFA These new data strongly suggest that the endocytic in to the TGN, also membranes to this unusual compartment (Baluska et al., 2002). Therefore, one might speculate that the and of a BFA compartment reflects the between and endocytic of the cell in In this it is interesting to that in root tips, BFA compartments have not been detected in root cells or in cells (Baluska et al., of which can be as being engaged in A found in a of is that BFA export from the ER. This apparently its in the of et al. where it is in BFA-treated cells, protein secretion is inhibited at an early in the pathway the in a Recent in plant cells support this et al., 2002). These have of a called after which the of in a living cell by all fluorescent markers in a with light and the of of in that Because of is an new in the has to from that are or the from of the this it has been possible to show that Golgi enzymes are in and out of the because to normal within about 5 of et al., 2002). It is interesting that this of Golgi enzymes can be blocked in about one-half of the Golgi stacks in a cell by the of BFA to the et al., 2002). In a this would to that BFA export out of the ER. However, it be out that these do not directly for release from the ER, but at the Golgi apparatus. In other words, the can only that transport is blocked in BFA-treated plant cells. The question that from this is at which in ER to Golgi transport does BFA its The is at the of ER However, vesiculation at the ER is by a different protein machinery than at the Golgi. In particular, from the ER, at least in yeast and mammalian cells, the two COPII coat and as well as the et al., 2001). is by a different of which does not to the Although COPII vesicles remain to be and characterized from plants, and have been (e.g. et al., 1999). Moreover, for transport in plants has now been published et al., 2001). with the absence of and Sec7-type GEFs from the ER, there are no in the literature of BFA directly with the of COPII In just the has recently been demonstrated by the of BFA to prevent recruitment to ER export in mammalian normal rat cells and the of after in the presence of BFA et al., 2001). In the effect of BFA on transport most likely does not occur at the of export from the ER (see also 1). of ER-Golgi trafficking in and In mammalian cells COPII-coated vesicles fuse with in to ER export as recycling from which retrograde COPI vesicles escaped proteins to the ER. The remaining membranes as transport along microtubules to the Golgi complex where they form new cis In plant cells COPII vesicles directly fuse at the cis Golgi of stacks because the compartment of does not transport is by of Golgi stacks along Retrograde transport from the Golgi to the ER may occur ER export as or of Golgi stacks. The between ER export and Golgi stacks in plant cells is not do these about the plant In we should assume that ER export also occurs in BFA-treated plant cells. In support of this, have been on the ER-Golgi hybrid compartment of BFA-treated BY-2 cells (Ritzenthaler et al., 2002). The lack of a should result in the direct transport of these vesicles to the cis which is the compartment in plants (Fig. This has been in the of Brandizzi et al. only for about one-half of the Golgi stacks The question is the other one-half of the stacks in the not new proteins from the ER. is different in these stacks that of transport vesicles at the the cis Although we do not have a to this we would to a that is consistent with our understanding of the eukaryotic endomembrane In particular, the described difference between plants and with to BFA effects may simply be a result of the different of the target organelle for ER In plants, this target organelle is the cis whereas in it is the (see analysis of BFA effects in BY-2 cells has that cisternae are lost BFA treatment (Ritzenthaler et al., 2002). This loss has been as the maturation of Golgi cisternae combined with a lack of formation of new This is on the that the Gm-Man1-GFP (Nebenführ et al., 1999) is lost from the stacks at a later time when the Golgi cisternae (Ritzenthaler et al., 2002). This of a cis Golgi protein to and trans cisternae in BFA-treated cells the of COPI vesicles in recycling Golgi enzymes to cisternae, as by the cisternal (e.g. 2001). This of transport also that new Golgi cisternae form by the fusion of anterograde vesicles and retrograde A lack of these recycling vesicles with a maturation of cis may prevent the transport vesicles from at the Golgi. This the lack of in about one-half of the Golgi stacks et al., whereas the other stacks presumably a normal transport through the presence of cis Therefore, our that transport in the presence of BFA be for some plant Golgi stacks simply because the target organelle is no longer The situation in mammalian cells is different in that the compartments are the which do not in the presence of BFA and to as a target for ER This leads to several that can be a of ER export by means of a dominant-negative mutant of should in all Golgi stacks because it the in the transport et al., 1998). the same effect should occur when vesicle fusion at the Golgi is by means of a dominant-negative mutant of et al., 2000). the of Golgi stacks that do not show in the presence of BFA should increase over time because the loss of cis cisternae has been found to be and, in BY-2 cells, to over to (Ritzenthaler et al., 2002). This last point in should be because the other possible transport can be expected to work much more rapidly and on an The of membranes and in BFA compartments suggests that there is an of BFA action the one that This should not as a because endocytic and in mammalian and yeast cells are by expression of et al., 1998). Arf1p is also for the retrograde transport of et al., 2001). It has been shown in mammalian cells that the formation of a transport between early and late involves an recruitment of a subset of COPI vesicle coat and 2000). with these it has been for some time that BFA in mammalian cells a tubulation of similar to that in the Golgi apparatus et al., The endocytic pathway in plants, over of to this The plant of early and late have yet to be and the of compartments is et al., 2000). However, the recent that BFA the of the to the in BY-2 cells et al., might that an along the plant endocytic pathway. It has to be that it is not whether activation of Arf1p at any of these depends on BFA-sensitive GEFs (Jackson and Casanova, 2000). effect of BFA on trafficking in the endocytic pathway in also be a secondary effect resulting from of the Golgi apparatus. The complex between anterograde and retrograde transport at the ER-Golgi (see above) as an for the effects of one in a network of membrane In addition, coat protein recruitment is not the only effect of Arf1p on the endomembrane of Arf1p and and 2000), that BFA can also have effects on membrane which in turn might be expected to a variety of membrane an effect has recently been demonstrated to occur in as well as BY-2 cells et al., 2002). on recent in our understanding of the system in plants and other we suggest the following of BFA The molecular target for BFA to be the same in all eukaryotic cells, a Sec7-type GEF that is for activation of effects are the to COPI coat proteins onto Golgi As a result, the of Golgi cisternae fuse directly with the ER, to the formation of an ER-Golgi hybrid compartment. However, trans-Golgi elements and the TGN from the Golgi stack and with of the endocytic pathway to form “BFA The of these two responses depends on the of the and endocytic Therefore, BFA effects are not limited to the Golgi apparatus and early The apparent in transport due to BFA treatment is more likely to result from the of transport vesicles to fuse with Golgi cisternae than to an inhibition of ER export