Protein Toxins: Intracellular Trafficking for Targeted Therapy

Home , Caveolae, Cholera toxin, Diphtheria toxin, Pseudomonas exotoxin, Shiga toxin, STX2

L Johannes1 and D Decaudin2 1Laboratoire ‘Traﬁc et Signalisation’, UMR144 Curie/CNRS, Institut Curie, Paris, France; and 2Service d’He´matologie Clinique, Institut Curie, Paris, France

The immunotoxin approach is based on the use of tumor- roles are inverted: nontoxic receptor-binding toxin moieties targeting ligands or antibodies that are linked to the catalytic are used for the targeting of therapeutic and diagnostic (toxic) moieties of bacterial or plant protein toxins. In this compounds to cancer or immune cells. The cell biological review, we ﬁrst discuss the current state of clinical develop- basis of these novel types of toxin-based therapeutics will ment of immunotoxin approaches describing the results be discussed, and we will summarize ongoing preclinical and obtained with the two toxins most frequently used: diphtheria clinical testing. and Pseudomonas toxin-derived proteins. In the second part Gene Therapy (2005) 12, 1360–1368. doi:10.1038/ of the review, a novel concept will be presented in which the sj.gt.3302557; published online 19 May 2005

Keywords: immunotoxin; Shiga toxin B-subunit; retrograde transport; immunotherapy; tumor targeting

Introduction proteins, and we will indicate their limitations. In the second part of the review, a novel concept will be Management of unresectable or metastatic tumors is presented in which the roles are inverted: nontoxic based on the cytotoxic effect of radiotherapy and/or receptor-binding toxin moieties are used for the targeting chemotherapy. Numerous toxin-based therapies have of diagnostic and therapeutic compounds to cancer cells been developed in this field for their potent cell-killing or immune cells. This approach exploits many of the activity. However, because of the nonspecific effects of characteristics that receptor-binding toxin moieties have toxins on normal and tumor cells, targeting of cancer acquired throughout evolution, such as stability, capacity cells generally remains a prerequisite for such therapeu- to cross tissue barriers, and their unique intracellular tic applications. Targeting of toxins to cancer cells has distribution pathways. The cell biological basis of these used a variety of modalities, namely (1) a direct novel types of toxin-based therapeutics will be dis- interaction between the toxin and its natural receptor cussed, with a presentation of ongoing preclinical and expressed on tumor cells; (2) vectorization of the toxin by clinical testing. a natural ligand or a monoclonal antibody that specifically recognizes cancer cells that express their natural receptors; and (3) gene therapy allowing cellular penetration of a viral vector expressing the toxin. In addition Recombinant immunotoxins to these direct cancer cell cytotoxic approaches, a few publications have reported anticancer immune activation To be used in clinical practice for cancer therapy, toxins via T-lymphocytes, natural killer cells, or dendritic cells must meet a number of stringent criteria: (1) the toxin with tumor-directed targeting by toxin compounds. must possess a very high inherent toxicity, in the low nanomolar range,1 because expression of the targeted The most developed class of targeted cytotoxic 5 treatments is constituted by immunotoxins. The immuno- receptors is usually less than 10 molecules per cell. For toxin approach is based on the use of tumor targeting example, one molecule of diphtheria toxin (DT), which is one of the two toxins most often used for cancer ligands or antibodies that are linked to the catalytic 2 moieties of bacterial or plant protein toxins. The targeted therapeutics, is sufficient to kill a cell; (2) the toxin toxin induces apoptosis by reaching the cytosol and compound, which includes the ligand of the target inactivating vital cell processes or by modifying the receptor, the toxin and, in most cases, a linker, should tumor cell surface membrane. In this review, we will preserve the main characteristics of the toxin; (3) delivery first discuss the current state of clinical development of of the toxin compound requires internalization of the immunotoxin approaches, focusing on clinical trials toxin after binding to the surface of cancer cells, as using diphtheria and Pseudomonas exotoxin (PE)-derived internalization leads to the apoptotic effect. Indeed, unlike radioisotope immunoconjugates, most toxins only kill cells once they are inside the cytosol;3 (4) the toxin Correspondence: Dr L Johannes, Laboratoire ‘Trafic et Signalisation’ UMR144 Curie/CNRS, Institut Curie, 26 rue d’Ulm, F-75248 Paris must be able to survive proteolytic processing and the Cedex 05, France acidic environment of lysosomes. Intracellular trafficking Published online 19 May 2005 of the toxin therefore appears to be a checkpoint for the Intracellular trafficking for targeted therapy L Johannes and D Decaudin 1361 4 desired cytotoxic effect; and (5) in most, but not all, A phase I trial of a ligand fusion-protein DAB389IL-2 situations, the toxin must be delivered from its targeting was performed in 73 patients with lymphoma expressing compound to be in an active form.1 the interleukin-2 receptor (35 cutaneous T-cell lympho- There are three classes of immunotoxins: class I toxins, mas (CTCL), 17 non-Hodgkin’s lymphomas (NHL), and mainly used for hematologic malignancies, are intra- 21 Hodgkin’s disease).9 Most side effects consisted of cellular enzymes blocking cytosolic protein synthesis flu-like symptoms (fever/chills, nausea/vomiting, and either by modifying elongation factor 2 (DT and PE) myalgias/arthralgias), acute infusion-related events or ribosomal RNA (ricin, gelonin, saporin, etc), or by (hypotension, dyspnea, chest pain, and back pain), and degrading total RNA (angiogenin, ribonuclease);4 class a vascular leak syndrome (hypotension, hypoalbumine- II toxins such as the Fab fragment of the antiCD25 mia, edema). Titers of DT antibodies were 38% at monoclonal antibody fused to phospholipase C5 bind to baseline and reached 92% after two courses. Five the cell surface and trigger intracellular signaling path- complete responses and eight partial responses were ways; class III toxins are pore-forming peptides leading obtained (ORR 18%) with durations of response ranging to leaks in the plasma membrane; these include mellitin,6 between 2 and 39 months. The presence of antibody did the plant Pyrularia thionin peptide,7 and the cobra venom not preclude clinical responses. Various phase II trials cationic 63 amino-acid membrane lytic peptide toxin.8 were subsequently performed, showing an overall Various toxins have been used, but the two bacterial response rate of 30% in CTCL (10% of complete toxins most frequently exploited for cancer treatments responses),10 11% in fludarabine-refractory chronic lym- are DT and PE. Numerous targets have been used to phocytic leukemia with no complete responses,11 and accumulate immunotoxins on cancer cells, namely 24% in B-cell NHL, including 7% of complete res- cluster designation differentiation antigens, cytokine ponses.12 A phase I trial with DT/GMCSF fusion protein receptors (interleukin-2 receptor, IL-3R, IL-4R, IL-6R, (DT388GMCSF) on 31 patients with refractory or relapsed IL-7R, IL-9R, IL-13R, IL-15R, granulocyte–macrophage acute myeloid leukemia13 also revealed dose-related colony-stimulating factor receptor (GM-CSFR), granulo- toxicity such as liver injury observed in two out of eight cyte colony-stimulating factor receptor (G-CSFR), trans- cases receiving 4.5–5 mg/kg/day. Other toxicities were ferrin receptor, and urokinase receptor), and a hetero- mild, namely fever, chills, hypoxemia, and transient geneous group of cell membrane receptors (human postinfusion hypotension. After 15–60 days of treatment, leukocyte antigen-DR, immunoglobulin idiotype, IgMFc, 21 of 25 patients that could be evaluated had a significant etc).4 Clinical trials performed with DT- and PE-derived increase of antibody titers. Only three clinical responses proteins will be presented. were obtained (10%) with only one complete response.

DT-derived proteins PE-derived proteins A very large number of DT fusion proteins have been As for DT, various ligands have been fused to PE, mainly developed and tested in preclinical models: (1) DT/GM- IL-4, IL-13, and EGF, and tested in various tumor cell

CSF fusion protein (DT388GMCSF) for acute myeloid lines or xenografted tumors expressing the correspond- leukemia; (2) DT/IL2 fusion protein (Denileukin Difti- ing receptors. In contrast, various heterogeneous clinical tox) for lymphoid malignancies; (3) DT/IL3 for acute trials have been performed with PE-derived protein, as myeloid leukemia and glioblastoma multiforme; (4) DT/ shown in Table 1. In malignant gliomas, targeted toxin IL4 for acute myeloid leukemia; (5) DT/epidermal was directly delivered to tumors using stereotactic growth factor (EGF) for acute myeloid leukemia, catheters.14–16 In these four studies, neurological toxicity glioblastoma multiforme, and pancreatic cancers; (6) was observed with increases of intracranial pressure DT/vascular endothelial growth factor with an antian- most probably related to infusion volume, recurrent giogenic effect for the treatment of breast and pancreatic tumor, or stereotactic catheter placement; a few cases of carcinomas; (7) DT/alpha-fetoprotein for hepatoblasto- fatigue and emesis were also described. The most ma; (8) DT/human chorionic gonadotropin for ovarian frequent toxicity observed in clinical trials performed cancers; (9) DT/gastrin-releasing peptide for small cell on benign brain tumors was vascular leak syndrome,17 lung cancer cells; and (10) DT/transferring conjugate for nausea/vomiting,17,18 diarrhea,17 pain and inﬂammation malignant gliomas. However, very few compounds have at injection sites,19 and elevation of transaminase levels.18 been tested in clinical studies, namely Denileukin The response rates are presented in Table 1, showing

Diftitox and DT388GMCSF. extensive tumor necrosis deﬁned on MRI scans in

Table 1 Clinical trials with Pseudomonas exotoxin-derived proteins

Ligands Indications Number of patients Responses References

LeY-related Ag Carcinomas 46 0 Posey (2002) HER2 Breast and colorectal cancers, melanoma 11 Two RC and four PR Azemar (2003) IL-4 Malignant gliomas 31 71% tumor necrosis (29% ETN) Weber (2003) IL-4 High-grade gliomas 9 6/9 tumor necrosis (1 CR) Rand (2000) IL-13 Malignant gliomas 12 Two tumor necrosis Lang (2002) TGF-a Malignant brain tumors 20 One CR and one PR Sampson (2003) CD25 CD25+ hematologic malignancies 35 One CR and seven PR Kreitman (2000)

LeY-related Ag ¼ Lewis-related antigen; TN ¼ tumor necrosis; ETN ¼ extensive tumor necrosis deﬁned as a near complete disappearance of contrast-enhancing lesion in MRI scans.

Gene Therapy Intracellular trafficking for targeted therapy L Johannes and D Decaudin 1362 malignant glioma trials, and a 23% response rate after rapidly clear immunotoxins from the bloodstream and anti-CD25-PE38 in patients with CD25-positive hemato- have a markedly reduced area under the curve. For DT- logic malignancies.18 derived compounds, most people are immunized with Immunogenicity of the recombinant immunotoxin DT leading to a significant pretreatment blood antibody anti-CD25-PE38 (LMB-2) was well evaluated.18 Neutra- titer in many patients and the development of an lizing antibodies were defined by at least 50% neutrali- amnestic immune response after DT conjugate therapies. zation of cytotoxicity when serum was incubated with Decreased immunogenicity has been observed with LMB-2. By this criterion, 22 of 35 patients (63%) had toxins that are foreign antigens to which patients have neutralizing antibodies after one injection cycle, nine of not been previously exposed, but antitoxin antibodies 12 patients (75%) after two cycles, and four of five appear after a lag of 3–4 weeks.26 (4) Cytotoxicity of the patients (80%) after three cycles. In all patients, neutra- immunotoxin depends on its intratumor cell internaliza- lizing anti-LMB-2 antibodies were directed against the tion and intracellular trafficking. In particular, DT can toxin portion of LMB-2, and in eight of 10 patients with escape from early endosomes to the cytosol, while PE strongly neutralizing antibodies had also developed and ricin toxin only reach the cytosol after passage human anti-mouse antibodies. The authors concluded through the Golgi apparatus and endoplasmic reticulum that LMB-2 was not very immunogenic (at least for the (see below), and this clearance by intracellular compo- production of neutralizing antibodies) in patients im- nents influences the cytotoxicity of the toxins.4 No data munosuppressed because of their tumor and previous concerning saturation of extravascular tumor sites have therapy, such as patients with heavily pretreated been obtained on toxin-derived compound internaliza- malignant NHL. tion and intracellular trafficking in reported clinical trials using immunotoxins. All these comments emphasize the difficulties involved in the development of a toxin- Lessons from immunotoxins derived therapeutic approach for the treatment of cancer. Recombinant immunotoxins used in cancer therapy should address a number of general rules whose Intracellular toxin trafficking pathways importance has become apparent in clinical studies as and mechanisms those summarized above: tumor specificity and favorable biodistribution between normal and tumor tissues, The study of immunotoxins has paid much attention to acceptable levels of nontumoral toxicity with maximal the catalytic (toxic) moieties of protein toxins and the cancer cell cytotoxicity, lack of significant immunogeni- associated membrane translocation domains. However, city allowing repeated administrations, and favorable the receptor-binding parts of these toxins are more than pharmacological profile in term of extra- and intracel- simple devices to hook the toxins onto cells. In some lular compartment half-life. From this point of view, the cases, they follow unique intracellular trafficking path- use of immunotoxins in clinical trials has identified ways and have been instrumental in the discovery of various limitations to their development, namely: (1) these pathways. The most prominent examples are specific toxicities of immunotoxins have been observed. clathrin-independent endocytosis and retrograde trans- The raised intracranial pressure observed after intra- port (Figure 1). In this paragraph, we will summarize the tumoral delivery of toxin-derived compounds in glioma current knowledge on unusual toxin trafficking path- patients is not due to the toxin.14–16 In contrast, vascular ways. The following paragraph illustrates how this leak syndrome secondary to vascular injury may be due knowledge may guide the development of novel types to selective receptors for peptide toxins on endothelial of targeting strategies in which receptor-binding toxin cells.20,21 A modification of the endothelial binding moieties are combined with diagnostic or therapeutic residues on toxins may reduce this toxicity. Reaction of compounds. immunotoxins with macrophages and lymphocytes leads to the release of cytokines such as IL-6 and TNF- Toxins enter cells by various endocytotic mechanisms a with constitutional symptoms.13,22–24 Corticosteroids The most extensively studied endocytosis mechanism is appear to reduce or eliminate this side effect.21 (2) still the clathrin-dependent pathway. Clathrin is an Pharmacologic problems have been observed. Small ubiquitous cytosolic protein that can be recruited to immunotoxins of less than 50 kDa, such as DT-derived membranes, directly or via so-called adaptor proteins.27 proteins, are rapidly cleared by the kidney glomeruli Clathrin has the remarkable capacity of forming different leading to a short half-life of 1–5 h,9,13,23 reduced area types of lattices, flat or basket-like, and it is commonly under the curve, and renal tubular injury.13 In contrast, believed that the latter favour membrane deformation penetration into extravascular sites of disease is better during the budding process of vesicles, at the plasma with small toxin-derived compounds, reducing endo- membrane or on internal membranes. Some protein thelial toxicity and increasing tumor targeting efficacy.4 toxins, such as DT28 and anthrax toxin,29 exclusively use However, there are no reports to date regarding the the clathrin pathway (Figure 1). This is achieved by saturation of extravascular tumor deposits in clinical binding to specific transmembrane domain protein trials with recombinant immunotoxins. In others words, receptors – heparin-binding EGF-like growth factor the proximal tumor tissue is exposed to high therapeutic precursor30 and anthrax toxin receptor,31 respectively. concentrations, whereas the middle portion is to ther- In other cases, it has clearly been established that entry apeutic concentrations, and the distal portion of the can also occur in a clathrin-independent manner. tumor is to subtherapeutic concentrations.25 (3) Efficacy In fact, protein toxins were the first cargoes that were of immunotoxins is influenced by their immunogenicity, used to identify clathrin-independent endocytosis me- as patients with antibodies directed against toxins chanisms. Cholera toxin, tetanus toxin, and ricin were

Gene Therapy Intracellular trafﬁcking for targeted therapy L Johannes and D Decaudin 1363

Figure 1 Intracellular transport pathways of protein toxins (adopted from Johannes and Goud113). Black arrows indicate the classical transport pathways. Diphtheria toxin and anthrax toxin enter cells by clathrin-dependent endocytosis. Shiga and cholera toxins can enter cells at least in part through clathrin- independent endocytosis, in addition to clathrin-dependent endocytosis (red arrows). The blue circle represents the interface between early and recycling endosomes and the trans-Golgi network. Shiga toxin and some other toxins such as cholera toxin, ricin, and Pseudomonas exotoxin, reach this interface and are then transported to the Golgi apparatus via the retrograde route (red arrow). Notably the use of Shiga toxin has allowed to identify the first molecular players at this interface. The retrograde pathway is distinct from the well known retrieval pathway used by the mannose 6-phosphate receptor (MPR), even tough recent evidence suggests that the MPR may also use the early and recycling endosomes–TGN interface.35 Shiga and cholera toxins are then transported from the Golgi apparatus to the endoplasmic reticulum (red arrow), bypassing the classical KDEL-receptor/COPI recycling pathway (blue arrow). Sulfation and glycosylation on corresponding Shiga toxin B-subunit variants are used to detect and quantify the retrograde transport of the protein to the trans-Golgi network and to the endoplasmic reticulum, respectively.64 From the lumen of the endoplasmic reticulum, catalytic toxin subunits are then retrotranslocated to the cytosol. found in smooth noncoated plasma membrane invagina- is true for Shiga toxin, which binds to Gb3.41,42 Just how tions, and interfering with clathrin-dependent endo- raft-type microdomains are involved in endocytosis cytosis using potassium depletion or cytosol acidification and whether cellular proteins are also involved have did not prevent their entry into cells.28,32–34 These initial yet to be established. It should also be pointed out that observations, based on the use of relatively nonspecific the dichotomy between clathrin-dependent endocytosis inhibitors, were subsequently confirmed by using tools and endocytosis involving raft-type microdomains may that allow a direct inhibition of clathrin function. Thus, not be as stringent as initially thought (see Johannes and when clathrin expression was downmodulated, Shiga Lamaze43 for a review on cellular proteins). As for other toxin still entered cells efficiently.35,36 For cholera toxin, toxins, cholera toxin is raft-associated in neurons, but similar conclusions were reached using dominant-nega- enters these cells via the clathrin pathway,44 and anthrax tive mutants of clathrin interacting proteins37 and toxin behaves in a similar way in several cell lines.29 morphological approaches.38 While it is now well The best-studied rafts are caveolae (for a review, see established that these toxins and others can use Parton45 and van Deurs et al46). It is often assumed that clathrin-independent mechanisms to enter cells, it must these are internalization structures that bud off from the be kept in mind that very often several pathways, plasma membrane, similar to clathrin-coated vesicles. In clathrin-dependent or not, can serve as entry portals favour of this hypothesis, it has been described that the (see eg Sandvig et al39 and Saint-Pol et al35 for Shiga GTPase dynamin was involved in caveolar endocyto- toxin), and endocytic plasticity, that is, the capacity to use sis,47,48 much as it has been implicated in the budding of several endocytic pathways, is a common trait to many clathrin-coated pits. Other work indicated, however, that of the known protein toxins (Figure 1). the role of caveolae in endocytosis may be more complex The mechanisms of clathrin-independent endocytosis than initially assumed. In one study, cholera toxin used remain elusive. The best correlation has been established the clathrin pathway in cells that express low levels of with the association of toxins with membrane micro- caveolin-1, but shifted to clathrin-independent endo- domains of the raft-type, enriched in glycosphingolipids, cytosis when caveolin-1 was high.49 Other reports have cholesterol, and long-chain phospholipids. Cholera toxin shown that caveolae are very poorly dynamic;50 that cells binds to a glycosphingolipid, ganglioside GM1, and as lacking caveolae can still internalize cholera toxin;51–53 such has become an archetypical raft marker.40 The same and that caveolin-1 overexpression inhibits rather than

Gene Therapy Intracellular trafficking for targeted therapy L Johannes and D Decaudin 1364 stimulates endocytosis.54,55 It appears that caveolae have endoplasmic reticulum uses the retrieval receptors and to be stimulated to become dynamic, as has been nicely the COPI membrane coat. This is the case for resident shown for the cellular entry of simian virus.40,56 proteins of the endoplasmic reticulum that are recycled The complexity of clathrin-independent endocytosis from Golgi cisternae after escaping from their steady- is further illustrated by a very recent study in which state localization in the endoplasmic reticulum (for a cholera toxin was again used as the tool.38 In this study, review, see, Pelham78). The retrieval signals are basic it was observed that the toxin could enter cells in a amino acids found at the C- or N-termini of type I or type clathrin-, caveolin-, and dynamin-independent manner. II trans-membrane proteins, respectively, or the KDEL This new pathway had previously been observed for tetrapeptide at the C-terminus of luminal proteins. When GPI-anchored proteins.57 It appears to operate via retrieval signal-tagged resident endoplasmic reticulum uncoated tubular or ring-shaped structures. Its impor- proteins encounter sorting receptors in the intermediate tance for intoxication of cells remains to be formally compartment or in Golgi cisternae, their retrograde demonstrated. transport to the endoplasmic reticulum is induced. The retrograde sorting machinery at the Golgi–endoplasmic Intracellular toxin distribution pathways reticulum interface includes components of the COPI Once they have entered cells, protein toxins are found in coat, which interact directly with retrieval signals, and early and recycling endosomes (Figure 1). For some mutation of which has been correlated with retrieval toxins, such as diphtheria and anthrax toxin, acidifica- phenotypes (for a review, see, Cosson and Letourneur79). tion allows translocation of the catalytic A-subunits to Surprisingly, many protein toxins can reach the the cytosol. Other toxins, such as Shiga toxin, cholera endoplasmic reticulum in a COPI/KDEL-receptor in- toxin, ricin, and PE, do not have this capacity. They dependent manner. This question was directly addressed follow a recently discovered pathway – the retrograde for the first time by using the B-subunit of Shiga toxin. route – from early and recycling endosomes to the The B-subunit does not carry a KDEL signal, and adding endoplasmic reticulum, via the Golgi apparatus,58–60 the signal to the protein did not increase its retrograde from where their catalytic A-subunits are translocated transport efficiency, as determined by a quantitative to the cytosol, using the cellular retrotranslocation assay.80 However, it was not determined how cell machinery (Simpson et al61 and Wesche et al62; see Lord intoxication was influenced by the addition of KDEL. et al63 for a review) (Figure 1). This was recently performed using cholera toxin.81 The The critical step of the retrograde pathway is escape catalytic A-subunit of cholera toxin carries a KDEL from endosomes. Quantitative tools based on Shiga toxin signal, but not the receptor binding B-subunit. As B-subunit64 have been instrumental for the molecular described for Shiga toxin B-subunit,80 adding the KDEL description of retrograde sorting at the interface between signal to the B-subunit directly, or via the A-subunit, early and recycling endosomes and the trans-Golgi did not enhance retrograde transport. Intoxication, how- network (TGN), a tubular membrane network at the ever, was slightly increased by the KDEL signal. These trans-side of Golgi cisternae. It was shown that this glycosphingolipid-binding toxins therefore appear to transport step depends on the following proteins (Figure reach the endoplasmic reticulum via an as yet ill-defined 1): SNARE fusion proteins that are found on early and mechanism bypassing the classical COPI/KDEL-receptor recycling endosomes or the TGN,65,66 clathrin,35,36 the recycling loop. However, once they reach the endoplas- clathrin adaptor epsinR,35 the GTPases dynamin36 and mic reticulum, they become substrates for the KDEL- Rab6a’,65 GPP130 protein,67 and the GRIP-domain receptor system, which increases their steady-state protein Golgin-97.68 In addition to this protein machin- concentration in this compartment, and therefore the ery, lipid microcompartmentalization has also been probability for retrotranslocation. hypothesized as a sorting factor on early/recycling This hypothesis explains earlier findings concerning endosomes.42,69 Nonassociation with lipid microdomains the relatively small effect on cell intoxication of KDEL and the resulting failure to enter the retrograde pathway signals, when deleted from various toxins (see Johannes may indeed explain why some cells, such as mono- and Goud60 for a review). It also explains why the cytes,70 macrophages,71 and bovine intestinal epithelial various methods used to interfere with COPI/KDEL- cells,72 are resistant to Shiga toxin, despite expression receptor machinery had no effects on toxin retrograde of Gb3. transport and/or cell intoxication by Shiga toxin,82 Apart from Shiga toxin, ricin has also been used to cholera toxin,83 and ricin.84 demonstrate the implication of dynamin,73 cholesterol,74 calcium,75 and protein kinase A76 in retrograde sorting. Some recent studies have also indicated the possibility Therapeutic applications of toxin trafficking that endosomes may be present in the early endocytic pathway that are different from the ‘conventional’ The trafficking characteristics of some of the above- transferrin receptor-positive early and recycling endo- mentioned protein toxins have inspired the development somes. These endosomes have been called ‘caveosomes’, of biomedical research aimed at targeting diagnostic or and they were found to be involved in the cellular entry therapeutic compounds not only to specific cells but even of simian virus 40.77 They also contain cholera toxin. to specific compartments within these cells. However, their role in cholera toxin trafficking still A prominent example is the use of protein toxins for remains uncertain. the targeting of antigens to the MHC class I pathway of From the TGN/Golgi apparatus, the protein toxins dendritic cells. These cells are currently considered to be that use the retrograde route are delivered to the the most potent inducers of cytotoxic T-CD8+ lympho- endoplasmic reticulum. The ‘classical’ recycling pathway cyte responses. They have the remarkable capacity to at the interface between the Golgi apparatus and the present not only antigens that are produced by these

Gene Therapy Intracellular trafficking for targeted therapy L Johannes and D Decaudin 1365 cells but also exogenous antigens.85,86 This process, intracellular targeting of the B-subunit via the retrograde termed cross-presentation, is remarkable as it involves pathway for tumor-specific activation and targeted the transfer of antigens from the outside where they are delivery of compounds to selected intracellular sites, released, for example by necrotic cancer cells, to the while at the same time avoiding degradation and cytosol where class I presentation starts with the clipping inactivation in late endosomes/lysosomes. Such devel- of antigenic proteins by proteasomes. From this point opments are currently under investigation in a number of view, cross-presentation has similar requirements to of laboratories around the world and are expected to protein toxins: an exogenous protein – antigen or yield promising results in the near future. catalytic A-subunit – needs to cross membranes to reach cytosolic machinery – proteasomes or molecular targets such as ribosomal RNA. Various protein toxins have Conclusions been used for the delivery of antigens into the class I pathway of dendritic cells (see Smith et al87 for a review). In this review, we have described how the initial More recently, two of them have been more particularly immunotoxin concept may evolve to take advantage of followed up because of their capacity to specifically nontoxic, receptor-binding, and trafficking subunits of target dendritic cells: the nontoxic B-subunit of Shiga protein toxins for the targeted delivery of tumor toxin, and a detoxified variant of Bordetella pertussis diagnostics and therapeutics. Focusing on Shiga toxin adenylate cyclase toxin (CyaA). B-subunit as an example, it was shown that this protein Shiga toxin B-subunit binds to the cellular toxin exclusively binds to a cellular receptor, the glycolipid receptor, the glycosphingolipid globotriaosyl ceramide Gb3, with a markedly restricted expression pattern that (Gb3 or CD77). It was observed that Gb3 is specifically allows designing compositions, which specifically target expressed on human and mouse dendritic cells and that certain cells types, such as antigen-presenting cells the B-subunit is able to cross membranes into the cytosol (exploitation for immunotherapy) or tumor cells (ex- in these cells.42,88 By incubating dendritic cells with B- ploitation for tumor targeting). In addition to cell subunit-coupled antigenic peptides or full-size proteins, specificity, the B-subunit also has intracellular trafficking it was shown that antigenic peptides were presented characteristics that allow it to avoid recycling and by both MHC class I and II molecules.89,90 Vaccination degradation and to reach specific intracellular sites, such of mice with B-subunit coupled to various antigenic as the Golgi apparatus, the endoplasmic reticulum, and peptides or full-size antigenic proteins induced a in some instances the cytosol. These targeting and T-CD8+ response without the use of adjuvant.88,90 The trafficking characteristics are currently exploited for the B-subunit-induced immune response was TH1 pola- development of new immunotherapy and cancer cell rized.90 Since TH1 responses play an important role in targeting approaches. protection against intracellular pathogens and tumors, Apart from targeting and trafficking, Shiga toxin B- the induction of TH1 polarization is highly desirable for subunit also has other characteristics that may be of vaccines designed to target these diseases. Recent data importance in the development of the above-mentioned show that prophylactic or therapeutic vaccination of vectorization approaches, such as small size, resistance mice with B-subunit-vectorized antigens leads to tumor to the intestinal milieu (pH and proteases), crossing protection (LJ and Eric Tartour, unpublished). of the intestinal mucosa, and distribution in the body. Similar characteristics have been observed for These characteristics are the result of its evolution with CyaA,91–95 which binds to dendritic cells via alpha(M)- mammalian hosts as a product of intestinal pathogens. In beta(2) integrin.96 mice, the B-subunit has been described as a low A second example for the use of receptor targeting immunogenic protein with major histocompatibility toxin subunits for biomedical applications is found in complex dependence.107 The same seems to hold true the domain of cancer cell targeting. The Shiga toxin in human. Indeed, in one study, among 30 patients with receptor, Gb3, has been reported to be overexpressed by Shiga-like toxin (or verotoxin (VT); note that VT1, or a certain number of human tumors, such as Burkitt Shiga-like toxin 1, is essentially the same protein as Shiga lymphomas,97 acute nonlymphocytic leukemia,98 ovar- toxin) induced hemolytic and uremic syndrome, no one ian99 and breast100 carcinomas, testicular seminomas,101 had antibodies to VT1 or VT2.108 Ludwig et al109 reported and various cancer cell lines. Shiga toxin B-subunit that 31% of patients with hemolytic and uremic could therefore possibly be used to target diagnostic syndrome had antibodies to Shiga-like toxin 2, while or therapeutic compounds to these tumors. This only 5% had antibodies to Shiga-like toxin 1. Similarly, approach has been pioneered at the preclinical stage among healthy patients, 10% had antibodies to Shiga-like using Shiga holotoxin itself. It was shown that intra- toxin 2, and only 1.8% to Shiga-like toxin 1. In another tumoral injection of Shiga toxin inhibits tumor study, 6/21 patients with Shigella dysenteriae induced growth.102–104 Another potential application of Shiga diarrhea had antibodies to Shiga holotoxin, but no one to toxin is the control of tumor neo-angiogenesis105 and B-subunit derived peptides.110 However, it cannot be purging of residual tumor burden from a human B-cell excluded that in a therapeutic setting that uses Shiga lymphoma xenograft, while sparing normal hematopoie- toxin B-subunit as a vectorization tool, repetitive injec- tic precursor cells.106 tion of the protein leads to a neutralizing antibody As a further refinement of the cancer cell targeting response. If that was the case, it may be achieved to approach, it might be suggested to replace the catalytic maintain Gb3 targeting while changing the targeting A-subunit of Shiga toxin by contrast agents for non- vector, as various Shiga-like toxins with limited sequence invasive tumor imaging or by therapeutic compounds identity have been described that all have Gb3 binding that have preferential effects on cancer cells. These capacity.111 Furthermore, a peptide was recently identi- strategies should also aim to exploit the specific fied with significant affinity for Gb3.112

Gene Therapy Intracellular trafficking for targeted therapy L Johannes and D Decaudin 1366 As described in this review, Shiga toxin B-subunit has mutated form of the Pseudomonas exotoxin termed PE-38 (TP- naturally evolved to transport an exogenous protein, the 38) for the treatment of malignant brain tumors. J Neurooncol catalytic A-subunit, from the outside of cells into the 2003; 65: 27–35. cytosol, and it can also be used for the delivery of 17 Posey JA et al. A phase I trial of the single-chain immunotoxin antigenic proteins to the cytosol of dendritic cells.88 In SGN-10 (BR96 sFv-PE40) in patients with advanced solid this context, it is noteworthy that Shiga toxin B-subunit tumors. Clin Cancer Res 2002; 8: 3092–3099. binds to a nucleolar protein that shuttles between cytosol 18 Kreitman RJ et al. Phase I trial of recombinant immunotoxin and nucleoplasm.114 Whether these different activities anti-Tac(Fv)-PE38 (LMB-2) in patients with hematologic malig- can be exploited to increase the efficiency of nonviral nancies. J Clin Oncol 2000; 18: 1622–1636. vector delivery to specific cells remains to be explored. 19 Azemar M et al. Regression of cutaneous tumor lesions in In conclusion, receptor-binding protein toxin moieties patients intratumorally injected with a recombinant single- chain antibody-toxin targeted to ErbB2/HER2. Breast Cancer Res hold out the promise of better therapeutics and diag- Treat 2003; 82: 155–164. nostics that exploit many of their remarkable character- 20 Baluna R et al. The effect of a monoclonal antibody coupled to istics, notably their trafficking to selected intracellular ricin A chain-derived peptides on endothelial cells in vitro: sites. insights into toxin-mediated vascular damage. Exp Cell Res 2000; 258: 417–424. 21 Siegall CB et al. Characterization of vascular leak syndrome References induced by the toxin component of Pseudomonas exotoxin- based immunotoxins and its potential inhibition with 1 Dyba M, Tarasova NI, Michejda CJ. Small molecule toxins nonsteroidal anti-inflammatory drugs. Clin Cancer Res 1997; 3: targeting tumor receptors. Curr Pharm Des 2004; 10: 2311–2334. 339–345. 2 Vallera DA et al. Targeting urokinase-type plasminogen 22 Amlot PL et al. A phase I study of an anti-CD22-deglycosylated activator receptor on human glioblastoma tumors with ricin A chain immunotoxin in the treatment of B-cell diphtheria toxin fusion protein DTAT. J Natl Cancer Inst 2002; lymphomas resistant to conventional therapy. Blood 1993; 82: 94: 597–606. 2624–2633. 3 Morokoff AP, Novak U. Targeted therapy for malignant 23 Kreitman RJ et al. Efficacy of the anti-CD22 recombinant gliomas. J Clin Neurosci 2004; 11: 807–818. immunotoxin BL22 in chemotherapy-resistant hairy-cell leuke- 4 Frankel AE et al. Immunotoxin therapy of hematologic mia. N Engl J Med 2001; 345: 241–247. malignancies. Semin Oncol 2003; 30: 545–557. 24 Sausville EA et al. Continuous infusion of the anti-CD22 5 Chovnick A et al. A recombinant, membrane-acting immuno- immunotoxin IgG-RFB4-SMPT-dgA in patients with B-cell toxin. Cancer Res 1991; 51: 465–467. lymphoma: a phase I study. Blood 1995; 85: 3457–3465. 6 Choo AB, Dunn RD, Broady KW, Raison RL. Soluble expression 25 Fenstermacher J, Kaye T. Drug ‘diffusion’ within the brain. Ann of a functional recombinant cytolytic immunotoxin in insect NY Acad Sci 1988; 531: 29–39. cells. Protein Exp Purif 2002; 24: 338–347. 26 Hall PD et al. Antibody response to DT-GM, a novel fusion 7 Gasanov SE, Rael ED, Gasanov NE, Vernon LP. In vitro toxin consisting of a truncated diphtheria toxin (DT) linked to evaluation of Pyrularia thionin-anti-CD5 immunotoxin. Cancer human granulocyte–macrophage colony stimulating factor Immunol Immunother 1995; 41: 122–128. (GM), during a phase I trial of patients with relapsed or 8 Hinman CL, Tang HP. A membrane-lytic immunoconjugate refractory acute myeloid leukemia. Clin Immunol 2001; 100: selective for human tumor T-lymphocytes. Int J Immunophar- 191–197. macol 1998; 20: 467–478. 27 Brodsky FM et al. Biological basket weaving: formation and 9 LeMaistre CF et al. Phase I trial of a ligand fusion-protein function of clathrin-coated vesicles. Annu Rev Cell Dev Biol 2001; (DAB389IL-2) in lymphomas expressing the receptor for 17: 517–568. interleukin-2. Blood 1998; 91: 399–405. 28 Moya M et al. Inhibition of coated pit formation in Hep2 cells 10 Olsen E et al. Pivotal phase III trial of two dose levels of blocks the cytotoxicity of diphtheria toxin but not that of ricin denileukin diftitox for the treatment of cutaneous T-cell toxin. J Cell Biol 1985; 101: 548–559. lymphoma. J Clin Oncol 2001; 19: 376–388. 29 Abrami L et al. Anthrax toxin triggers endocytosis of its 11 Frankel AE et al. A phase II study of DT fusion protein receptor via a lipid raft-mediated clathrin-dependent process. denileukin diftitox in patients with fludarabine-refractory J Cell Biol 2003; 160: 321–328. chronic lymphocytic leukemia. Clin Cancer Res 2003; 9: 30 Naglich JG, Metherall JE, Russell DW, Eidels L. Expression 3555–3561. cloning of a diphtheria toxin receptor: identity with a heparin- 12 Dang NH et al. Phase II study of denileukin diftitox for binding EGF-like growth factor precursor. Cell 1992; 69: relapsed/refractory B-Cell non-Hodgkin’s lymphoma. J Clin 1051–1061. Oncol 2004; 22: 4095–4102. 31 Bradley KA et al. Identification of the cellular receptor for 13 Frankel AE et al. Phase I trial of a novel diphtheria toxin/ anthrax toxin. Nature 2001; 414: 225–229. granulocyte macrophage colony-stimulating factor fusion 32 Montesano R, Roth J, Robert A, Orci L. Non-coated mem- protein (DT388GMCSF) for refractory or relapsed acute brane invaginations are involved in binding and internali- myeloid leukemia. Clin Cancer Res 2002; 8: 1004–1013. zation of cholera and tetanus toxins. Nature 1982; 296: 14 Weber F et al. Safety, tolerability, and tumor response of IL4- 651–653. Pseudomonas exotoxin (NBI-3001) in patients with recurrent 33 Tran D et al. Ligands internalized through coated or noncoated malignant glioma. J Neurooncol 2003; 64: 125–137. invaginations follow a common intracellular pathway. Proc Natl 15 Rand RW et al. Intratumoral administration of recombinant Acad Sci USA 1987; 84: 7957–7961. circularly permuted interleukin-4-Pseudomonas exotoxin in 34 Sandvig K, Olsnes S, Petersen OW, van Deurs B. Acidification patients with high-grade glioma. Clin Cancer Res 2000; 6: of the cytosol inhibits endocytosis from coated pits. J Cell Biol 2157–2165. 1987; 105: 679–689. 16 Sampson JH et al. Progress report of a phase I study of the 35 Saint-Pol A et al. Clathrin adaptor epsinR is required for intracerebral microinfusion of a recombinant chimeric protein retrograde sorting on early endosomal membranes. Dev Cell composed of transforming growth factor (TGF)-alpha and a 2004; 6: 525–538.

Gene Therapy Intracellular trafficking for targeted therapy L Johannes and D Decaudin 1367 36 Lauvrak SU, Torgersen ML, Sandvig K. Efficient endosome-to- 59 Sandvig K, van Deurs B. Entry of ricin and Shiga toxin into Golgi transport of Shiga toxin is dependent on dynamin and cells: molecular mechanisms and medical perspectives. EMBO J clathrin. J Cell Sci 2004; 117: 2321–2331. 2000; 19: 5943–5950. 37 Nichols BJ et al. Rapid cycling of lipid raft markers between the 60 Johannes L, Goud B. Surfing on a retrograde wave: how does cell surface and Golgi complex. J Cell Biol 2001; 153: 529–541. Shiga toxin reach the endoplasmic reticulum? Trends Cell Biol 38 Kirkham M et al. Ultrastructural identification of uncoated 1998; 8: 158–162. caveolin-independent early endocytic vehicles. J Cell Biol 2005; 61 Simpson JC et al. Ricin A chain utilises the endoplasmic 168: 465–476. reticulum-associated protein degradation pathway to enter the 39 Sandvig K et al. Endocytosis from coated pits of Shiga toxin: a cytosol of yeast. FEBS Lett 1999; 459: 80–84. glycolipid-binding protein from Shigella dysenteriae 1. J Cell Biol 62 Wesche J, Rapak A, Olsnes S. Dependence of ricin toxicity on 1989; 108: 1331–1343. translocation of the toxin A-chain from the endoplasmic 40 Harder T, Scheiffele P, Verkade P, Simons K. Lipid domain reticulum to the cytosol. J Biol Chem 1999; 274: 34443–34449. structure of the plasma membrane revealed by patching of 63 Lord JM et al. Retrograde transport of toxins across the membrane components. J Cell Biol 1998; 141: 929–942. endoplasmic reticulum membrane. Biochem Soc Trans 2003; 31: 41 Katagiri YU et al. Activation of src family kinase yes induced 1260–1262. by Shiga toxin binding to globotriaosyl ceramide (Gb3/CD77) 64 Mallard F, Johannes L. Shiga toxin B-subunit as a tool to study in low density, detergent-insoluble microdomains. J Biol Chem retrograde transport. In: Philpott D, Ebel F (eds). Methods 1999; 274: 35278–35282. in Molecular Medicine. Shiga Toxin Methods and Protocols, 42 Falguie`res T et al. Targeting of Shiga toxin B-subunit to Vol 73, Chapter 17. Humana Press: Totowa, New Jersey, 2002, retrograde transport route in association with detergent pp 209–220. resistant membranes. Mol Biol Cell 2001; 12: 2453–2468. 65 Mallard F et al. Early/recycling endosomes-to-TGN transport 43 Johannes L, Lamaze C. Clatrin-dependent or not: is it still the involves two SNARE complexes and a Rab6 isoform. J Cell Biol question? Traffic 2002; 3: 443–451. 2002; 156: 653–664. 44 Shogomori H, Futerman AH. Cholera toxin is found in 66 Tai G et al. Participation of syntaxin 5/Ykt6/GS28/GS15 detergent-insoluble rafts/domains at the cell surface of SNARE complex in transport from the early/recycling endo- hippocampal neurons but is internalized via a raft-independent some to the TGN. Mol Biol Cell 2004; 15: 4011–4022. mechanism. J Biol Chem 2001; 276: 9182–9188. 67 Natarajan R, Linstedt AD. A cycling cis-Golgi protein mediates 45 Parton RG. Caveolae – from ultrastructure to molecular endosome-to-Golgi traffic. Mol Biol Cell 2004; 15: 4798–4806. mechanisms. Nat Rev Mol Cell Biol 2003; 4: 162–167. 68 Lu L, Tai G, Hong W. Autoantigen Golgin-97, an effector of Arl1 46 van Deurs B, Roepstorff K, Hommelgaard AM, Sandvig K. GTPase, participates in traffic from the endosome to the trans- Caveolae: anchored, multifunctional platforms in the lipid Golgi network. Mol Biol Cell 2004; 15: 4426–4443. ocean. Trends Cell Biol 2003; 13: 92–100. 69 Wolf AA, Fujinaga Y, Lencer WI. Uncoupling of the cholera 47 Oh P, McIntosh DP, Schnitzer JE. Dynamin at the neck of toxin-G(M1) ganglioside receptor complex from endocytosis, caveolae mediates their budding to form transport vesicles by retrograde Golgi trafficking, and downstream signal transduc- GTP-driven fission from the plasma membrane of endothelium. tion by depletion of membrane cholesterol. J Biol Chem 2002; J Cell Biol 1998; 141: 101–114. 277: 16249–16256. 48 Henley JR, Krueger EW, Oswald BJ, McNiven MA. Dynamin- 70 Ramegowda B, Tesh VL. Differentiation-associated toxin mediated internalization of caveolae. J Cell Biol 1998; 141: 85–99. receptor modulation, cytokine production, and sensitivity to 49 Singh RD et al. Selective caveolin-1-dependent endocytosis of Shiga-like toxins in human monocytes and monocytic cell lines. glycosphingolipids. Mol Biol Cell 2003; 14: 3254–3265. Infect Immun 1996; 64: 1173–1180. 50 Thomsen P, Roepstorff K, Stahlhut M, van Deurs B. Caveolae 71 Tesh VL, Ramegowda B, Samuel JE. Purified Shiga-like toxins are highly immobile plasma membrane microdomains, which induce expression of proinflammatory cytokines from murine are not involved in constitutive endocytic trafficking. Mol Biol peritoneal macrophages. Infect Immun 1994; 62: 5085–5094. Cell 2002; 13: 238–250. 72 Hoey DE et al. Verotoxin 1 binding to intestinal crypt epithelial 51 Orlandi PA, Fishman PH. Filipin-dependent inhibition of cells results in localization to lysosomes and abrogation of cholera toxin: evidence for toxin internalization and activation toxicity. Cell Microbiol 2003; 5: 85–97. through caveolae-like domains. J Cell Biol 1998; 141: 905–915. 73 Llorente A et al. Expression of mutant dynamin inhibits toxicity 52 Torgersen ML, Skretting G, van Deurs B, Sandvig K. Inter- and transport of endocytosed ricin to the Golgi apparatus. J Cell nalization of cholera toxin by different endocytic mechanisms. Biol 1998; 140: 553–563. J Cell Sci 2001; 114: 3737–3747. 74 Grimmer S, Iversen TG, van Deurs B, Sandvig K. Endosome to 53 Nichols BJ. A distinct class of endosome mediates clathrin- Golgi transport of ricin is regulated by cholesterol. Mol Biol Cell independent endocytosis to the Golgi complex. Nat Cell Biol 2000; 11: 4205–4216. 2002; 15: 15. 75 Lauvrak SU, Llorente A, Iversen TG, Sandvig K. Selective 54 Le PU, Guay G, Altschuler Y, Nabi IR. Caveolin-1 is a negative regulation of the Rab9-independent transport of ricin to the regulator of caveolae-mediated endocytosis to the endoplasmic Golgi apparatus by calcium. J Cell Sci 2002; 115: 3449–3456. reticulum. J Biol Chem 2002; 277: 3371–3379. 76 Birkeli KA et al. Endosome-to-Golgi transport is regulated by 55 Sharma DK et al. Selective stimulation of caveolar endocytosis protein kinase A type II alpha. J Biol Chem 2003; 278: 1991–1997. by glycosphingolipids and cholesterol. Mol Biol Cell 2004; 15: 77 Pelkmans L, Burli T, Zerial M, Helenius A. Caveolin-stabilized 3114–3122. membrane domains as multifunctional transport and sorting 56 Pelkmans L, Kartenbeck J, Helenius A. Caveolar endocytosis of devices in endocytic membrane traffic. Cell 2004; 118: 767–780. simian virus 40 reveals a new two-step vesicular-transport 78 Pelham HR. The dynamic organisation of the secretory path- pathway to the ER. Nat Cell Biol 2001; 3: 473–483. way. Cell Struct Funct 1996; 21: 413–419. 57 Sabharanjak S, Sharma P, Parton RG, Mayor S. GPI-anchored 79 Cosson P, Letourneur F. Coatomer (COPI)-coated vesicles: role proteins are delivered to recycling endosomes via a distinct in intracellular transport and protein sorting. Curr Opin Cell Biol cdc42-regulated, clathrin-independent pinocytic pathway. Dev 1997; 9: 484–487. Cell 2002; 2: 411–423. 80 Johannes L, Tenza D, Antony C, Goud B. Retrograde transport 58 Lord JM, Roberts LM. Retrograde transport: going against the of KDEL-bearing B-fragment of Shiga toxin. J Biol Chem 1997; flow. Curr Biol 1998; 8: R56–R58. 272: 19554–19561.

Gene Therapy Intracellular trafficking for targeted therapy L Johannes and D Decaudin 1368 81 Fujinaga Y et al. Gangliosides that associate with lipid rafts 98 Cooling LL, Zhang de S, Naides SJ, Koerner TA. Glyco- mediate transport of cholera and related toxins from the plasma sphingolipid expression in acute nonlymphocytic leukemia: membrane to the endoplasmic reticulum. Mol Biol Cell 2003; 14: common expression of Shiga toxin and parvovirus B19 4783–4793. receptors on early myeloblasts. Blood 2003; 101: 711–721. 82 Jackson ME et al. The KDEL retrieval system is exploited by 99 Arab S et al. Expression of the verotoxin receptor glycolipid, pseudomonas exotoxin A, but not by Shiga-like toxin-1, during globotriaosylceramide, in ovarian hyperplasias. Oncol Res 1997; retrograde transport from the Golgi complex to the endoplas- 9: 553–563. mic reticulum. J Cell Sci 1999; 112: 467–475. 100 LaCasse EC et al. Shiga-like toxin-1 receptor on human breast 83 Chen A, Hu T, Mikoryak C, Draper RK. Retrograde transport of cancer, lymphoma, and myeloma and absence from CD34(+) protein toxins under conditions of COPI dysfunction. Biochim hematopoietic stem cells: implications for ex vivo tumor Biophys Acta 2002; 1589: 124–139. purging and autologous stem cell transplantation. Blood 1999; 84 Chen A, AbuJarour RJ, Draper RK. Evidence that the transport 94: 2901–2910. of ricin to the cytoplasm is independent of both Rab6A and 101 Kang JL, Rajpert-De Meyts E, Wiels J, Skakkebaek NE. COPI. J Cell Sci 2003; 116: 3503–3510. Expression of the glycolipid globotriaosylceramide (Gb3) in 85 Guermonprez P, Amigorena S. Pathways for antigen cross testicular carcinoma in situ. Virchows Arch 1995; 426: 369–374. presentation. Springer Semin Immunopathol 2005; 26: 257–271. 102 Farkas-Himsley H et al. The bacterial colicin active against 86 Ackerman AL, Cresswell P. Cellular mechanisms governing tumor cells in vitro and in vivo is verotoxin 1. Proc Natl Acad Sci cross-presentation of exogenous antigens. Nat Immunol 2004; 5: USA 1995; 92: 6996–7000. 678–684. 103 Salhia B et al. The treatment of malignant meningioma with 87 Smith DC et al. 1st class ticket to class I: protein toxin as verotoxin. Neoplasia 2002; 4: 304–311. pathfinders for antigen presentation. Traffic 2002; 3: 697–704. 104 Ishitoya S et al. Verotoxin induces rapid elimination of human 88 Haicheur N et al. The B-subunit of Shiga toxin fused to a tumor renal tumor xenografts in SCID mice. JUrol2004; 171: antigen elicits CTL and targets dendritic cells to allow MHC 1309–1313. class I restricted presentation of peptides derived from 105 Heath-Engel HM, Lingwood CA. Verotoxin sensitivity of exogenous antigens. J Immunol 2000; 165: 3301–3308 (*principal ECV304 cells in vitro and in vivo in a xenograft tumour model: investigators). VT1 as a tumour neovascular marker. Angiogenesis 2003; 6: 89 Lee R-S et al. Major histocompatibility complex class I 129–141. presentation of exogenous soluble tumor antigen fused to the 106 LaCasse EC et al. Shiga-like toxin purges human lymphoma B-fragment of Shiga toxin. Eur J Immunol 1998; 28: 2726–2737. from bone marrow of severe combined immunodeficient mice. 90 Haicheur N et al. The B-subunit of Shiga toxin coupled to full- Blood 1996; 88: 1561–1567. size protein elicits humoral and cellular immune responses 107 Bast DJ et al. Murine antibody responses to the verotoxin 1 B associated with a TH1 dominant polarization. Int Immunol 2003; subunit: demonstration of major histocompatibility complex 15: 1161–1171 (*principal investigators). dependence and an immunodominant epitope involving 91 Fayolle C et al. In vivo induction of CTL responses by Infect Immun 65 recombinant adenylate cyclase of Bordetella pertussis carrying phenylalanine 30. 1997; : 2978–2982. viral CD8+ T cell epitopes. J Immunol 1996; 156: 4697–4706. 108 Chart H, Law D, Rowe B, Acheson DW. Patients with 92 Saron MF et al. Anti-viral protection conferred by recombinant haemolytic uraemic syndrome caused by Escherichia coli O157: adenylate cyclase toxins from Bordetella pertussis carrying a absence of antibodies to Vero cytotoxin 1 (VT1) or VT2. J Clin CD8+ T cell epitope from lymphocytic choriomeningitis virus. Pathol 1993; 46: 1053–1054. Proc Natl Acad Sci USA 1997; 94: 3314–3319. 109 Ludwig K et al. Antibody response to Shiga toxins Stx2 93 Fayolle C et al. Therapy of murine tumors with recombinant and Stx1 in children with enteropathic hemolytic–uremic Bordetella pertussis adenylate cyclase carrying a cytotoxic T cell syndrome. J Clin Microbiol 2001; 39: 2272–2279. epitope. J Immunol 1999; 162: 4157–4162. 110 Levine MM et al. Antibodies to Shiga holotoxin and to two 94 Schlecht G et al. Antigen targeting to CD11b allows efficient synthetic peptides of the B subunit in sera of patients presentation of CD4+ and CD8+ T cell epitopes and in vivo with Shigella dysenteriae 1 dysentery. J Clin Microbiol 1992; 30: Th1-polarized T cell priming. J Immunol 2004; 173: 6089–6097. 1636–1641. 95 Fayolle C, Bauche C, Ladant D, Leclerc C. Bordetella pertussis 111 Lingwood CA. Role of verotoxin receptors in pathogenesis. adenylate cyclase delivers chemically coupled CD8(+) T-cell Trends Microbiol 1996; 4: 147–153. epitopes to dendritic cells and elicits CTL in vivo. Vaccine 2004; 112 Miura Y et al. Peptides binding to a Gb3 mimic selected from a 23: 604–614. phage library. Biochim Biophys Acta 2004; 1673: 131–138. 96 Guermonprez P et al. The adenylate cyclase toxin of Bordetella 113 Johannes L, Goud B. Facing inward from compartment shores: pertussis binds to target cells via the alpha(M)beta(2) integrin how many pathways were we looking for? Traffic 2000; 1: (CD11b/CD18). J Exp Med 2001; 193: 1035–1044. 119–123. 97 Wiels J, Fellous M, Tursz T. Monoclonal antibody against a 114 Falguie`res T, Johannes L. Shiga toxin B-subunit binds to Burkitt lymphoma-associated antigen. Proc Natl Acad Sci USA the chaperon BiP and the nucleolar protein B23. Biol Cell 1981; 78: 6485–6488. (in press).

Gene Therapy