Gene Therapy (2005) 12, 865–872 & 2005 Nature Publishing Group All rights reserved 0969-7128/05 $30.00 www.nature.com/gt REVIEW Delivery into cells: lessons learned from plant and bacterial toxins K Sandvig1 and B van Deurs2 1Institute for Cancer Research, The Norwegian Radium Hospital, University of Oslo, Montebello, Oslo, Norway; and 2Structural Cell Biology Unit, Department of Medical Anatomy, The Panum Institute, University of Copenhagen, Denmark A number of protein toxins of bacterial and plant origin have Some toxins enter from endosomes in response to low cytosolic targets, and knowledge about these toxins have endosomal pH, whereas others, including Shiga toxin and provided us with essential information about mechanisms that ricin, are transported all the way to the Golgi apparatus and can be used to gain access to the cytosol as well as detailed the ER before they are translocated to the cytosol. Plant knowledge about endocytosis and intracellular sorting. proteins such as gelonin that are without a binding moiety are Such toxins include those that have two moieties, one (the taken up only by fluid-phase endocytosis, and normally they B-moiety) that binds to cell surface receptors and another have a low toxicity. However, they can be used to test for (the A-moiety) with enzymatic activity that enters the cytosol, disruption of endosomal membranes leading to cytosolic as well as molecules that only have the enzymatically active access of internalized molecules. Similarly to toxins with moiety and therefore are inefficient in cell entry. The toxins a binding moiety they are highly toxic when reaching the discussed in the present article include bacterial toxins such cytosol, thereby providing the investigator with an efficient as Shiga toxin and diphtheria toxin, as well as plant toxins tool to study endosomal disruption and induced transport to such as ricin and ribosome-inactivating proteins without a the cytosol. In conclusion, the protein toxins are useful tools binding moiety, such as gelonin. Toxins with a binding moiety to study transport and cytosolic translocation, and they can can be used as vectors to translocate epitopes, intact be used as vectors for transport to the interior of the cell. proteins, and even nucleotides into the cytosol. The toxins Gene Therapy (2005) 12, 865–872. doi:10.1038/ fall into two main groups when it comes to cytosolic entry. sj.gt.3302525; Published online 7 April 2005 Keywords: toxin; Shiga toxin; diphtheria toxin; ricin; endocytosis Introduction and as indicated in Figure 2, the toxins use two main pathways to gain access to the cytosol. They can be Bacteria produce a large variety of protein toxins with transferred from endosomes (early or late) in response intracellular targets, and also plants contain similar toxic to low pH, or they can enter the Golgi and then the ER proteins (for recent reviews about toxins containing both and be transferred to the cytosol from this destination. a binding moiety and an enzymatically active moiety, Toxins normally entering from acidic endosomes can as well as toxic proteins without a binding moiety, see even in some cases, as first shown for diphtheria toxin,7,8 Sandvig and van Deurs,1 Hartley and Lord,2 Sandvig be induced to enter directly from the cell surface in et al,3,4 Turton et al5). The schematic structure and the response to external low pH. Not only do the toxins activity of some of these toxins are shown in Figure 1. reveal the different intracellular mechanisms that can be All of the plant toxins studied so far and several of the used to enter the cytosol, it turns out that they can in bacterial toxins (the Shiga toxins) block protein synthesis themselves be used as vectors to bring in epitopes to be by removing one adenine from the 28S ribosomal RNA presented by MHC class I,9 they can bring in intact of the 60S subunit, thereby blocking protein synthesis. proteins,10,11 and in some cases it turns out that they Diphtheria toxin and Pseudomonas exotoxin A inhibit might even be used to transfer nucleotides into the protein synthesis by inactivation of elongation factor 2. cytosol and the nucleus of cells.12–19 These aspects of the Bacterial toxins have a number of different target toxins are described in the paragraphs below. molecules, ranging from kinases to actin and Rho proteins.1,6 Common to all these toxins, regardless of the target molecule, is that they have to enter the cytosol to exert their effects. Therefore, studies of these toxins Endocytic mechanisms: ports of entry used have provided us with knowledge of how an external by protein toxins molecule might reach the cytosol. As described below, The variety of endocytic mechanisms in a cell has partly been revealed by studies of protein toxins that seem to Correspondence: Professor K Sandvig, Institute for Cancer Research, 1,20 The Norwegian Radium Hospital, University of Oslo, Montebello, exploit all of them in order to attack a given cell. The 0310 Oslo, Norway plant toxin ricin, which binds all over the cell surface Published online 7 April 2005 since it binds to both glycoproteins and glycolipids with Lessons learned from plant and bacterial toxins K Sandvig and B van Deurs 866 Figure 1 Schematic overview of some bacterial and plant toxins. The enzymatically active subunit is denoted A. However, as indicated, in some cases this subunit is cleaved into A1 and A2, where A1 is the one with enzymatic activity. The B moiety is the part of the toxin responsible for binding to cell surface receptors. In the figure, also the enzymatic activity and the intracellular targets are listed. The arrows indicate sites where cellular enzymes cleave and activate the toxins. For the properties and structure of more toxins, please see recent reviews, such as Sandvig and van Deurs.1 terminal galactose, provided some of the early evidence clathrin-dependent and -independent processes operate for clathrin-independent endocytosis.21,22 In fact, cla- in the same cell, and as described below in the paragraph thrin-dependent endocytosis was during many years about transcytosis, the clathrin-independent mechan- believed to account for all uptake from the cell surface. ism(s) can be subjected to regulation by a number of It is now known that several mechanisms exist, both signalling pathways. Furthermore, when it comes to Gene Therapy Lessons learned from plant and bacterial toxins K Sandvig and B van Deurs 867 can induce pinching off of some caveolae, or whether some cholera toxin becomes internalized by a small fraction of caveolae which for unknown reasons are not stabilized at the plasma membrane. The internalization of caveolae may at least in some cases be dependent on tyrosine phosphorylation and actin reorganization33 or regulated by addition of cholesterol and glycolipids.34 Importantly, it has recently been published that cholera toxin taken up from caveolar structures ends up in normal acidified endosomes,35 simlarly as we have found for diphtheria toxin bound to a receptor with a GPI anchor (see below).36 That endocytic compartments are able to exchange contents makes it less critical in the present context whether a ligand is taken up from caveolae or clathrin-coated pits. Diphtheria toxin is ideal to investigate whether the receptor molecule it binds to is internalized into an acidic compartment. For instance, when diphtheria toxin was bound to its receptor modified to contain a GPI-anchor instead of the normal transmembrane/cytosolic domain, the toxin was, in Figure 2 Transport of protein toxins in cells. Toxins can be endocytosed by contrast to when bound to its normal receptor, inter- 36 clathrin-dependent (1) or clathrin-independent endocytosis (2), they can be nalized by clathrin-independent endocytosis. This GPI- internalized by macropinocytosis (3), recycled (4), transported to the Golgi anchored receptor was internalized into an acidified apparatus (5) and the ER (6) from where some of them are translocated to organelle, presumably an endosome, since the toxin was the cytosol (open arrow). Alternatively, after endocytosis, some of the translocated to the cytosol and intoxicated the cell. Thus, toxins are transported to the cytosol from acidified endosomes (early/late; although there are – as with other approaches – pitfalls open arrows). A variable, cell-type-dependent fraction of the toxin is transported from early to late endosomes (7) and toxins may then continue when it comes to studies of endocytosis using protein to lysosomes (8) for degradation. EE, early endosomes; SE sorting toxins, they have proven very important for our under- endosomes; LE, late endosomes; LY, lysosomes. standing of the endocytic mechanims. clathrin-independent processes, different molecules can Direct transport from endosomes be involved (perhaps in a cell-type specific manner?). For to the cytosol instance, it has been described that uptake of GPI- anchored proteins can be dependent on cdc42 in CHO To which extent can transport of proteins and nucleo- cells,23 whereas this does not seem to be the case in HeLa tides to the cytosol occur from endosomes? It is now cells.24 Furthermore, the clathrin-independent endo- 25 years since it was discovered that diphtheria toxin cytosis can be dependent on lipid rafts (domains rich enters from endosomes in response to the relatively low in cholesterol and sphingolipids),25 or it can be raft- pH found in this organelle, and that direct entry from independent.26,27 Clathrin-independent endocytosis can the cell surface can be induced by exposing cells with also operate with25 and without dynamin26,27 as an surface-bound toxin to low pH, thereby mimicking the essential molecule. Thus, clathrin-independent uptake conditions in the endosomes (see Table 1).7,8 of the IL2 receptor is dependent on lipid rafts,25 whereas Thus, it became clear that low pH is sufficient to drive ricin and cholera toxin can be internalized under the translocation of the receptor-bound toxin across the conditions where clathrin-dependent uptake is inhibited membrane.