The specific in or ex vivo transduction of cancerous, infected or malfunctioning cells with pro-apoptotic, defence or restoring genes is a promising approach for the treatment of patients. However, gene therapy continues to present obstacles, regarding safety, transduction efficiency and specificity. Most gene delivery systems are based on animal viral vectors, which are efficient on one hand, but may include the risk of intrinsic toxicity and are costly to manufacture. In addition to developing safer viral vectors, scientists began modifying bacteriophage particles for gene delivery purposes 12. For example, filamentous bacteriophage libraries, consist of up to 10¹⁰ unique phage particles; each displaying a different binding ligand on its surface and encapsulating the corresponding eukaryotic cDNA. These libraries are commonly used for the selection of novel antigen-specific binding molecules3. Furthermore, it is possible to select for ligand-bound phage being internalised by the target cell after receptor-binding 45. If a eukaryotic gene cassette is encoded in the phagemid vector, a small percentage of target cells will express this gene 267. Thus, bacteriophage have the potential to serve as specifically targeted gene delivery vectors. They have no natural tropism for eukaryotic cells, nor do they integrate into the cell's genome.

A filamentous bacteriophage containing a phagemid vector (herein after referred to as "phagemid particle") encodes only the gene for a surface fusion protein, an antibiotic resistance gene and an amplification and packaging signal, but lacks the genes for all other bacteriophage proteins. In order to produce functional phagemid particles, E. coli transformed with the phagemid vector have to be superinfected with helper phage (phage rescue), which contribute the missing proteins for the phagemid particle assembly in the periplasm of E. coli. Phagemid particles are easy to genetically modify in terms of targeting to a specific receptor-bearing cell population. As phagemid vectors are relatively small in size compared to full size phage vectors, they are capable of carrying larger foreign DNA sequences. However, a major obstacle of using bacteriophage as gene transfer vehicles is their low transduction efficiency, which relies on gene transfer and expression. Gene transfer efficiency generally depends on the binding affinity of the ligand and the intracellular trafficking of the internalized receptor-bound phage, including endosomal escape and capsid breakdown. Moreover, processing of the single-stranded phage DNA, including conversion into double-stranded DNA and nuclear localization, is a critical factor 8.

Commonly, the displayed ligand is fused to the minor coat protein pIII (5 copies per phage). Ligands can be displayed either monovalently or multivalently, depending on the helper phage system being used. The importance of this factor has been demonstrated by Larocca et al. (2001) 9 in comparing the transduction efficiency of epidermal growth factor receptor (EGFR)-positive target cells with bacteriophage containing a green fluorescence protein (GFP) gene and displaying either monovalently or multivalently the EGF molecule. The multivalency of the EGF, fused to the minor coat protein pIII on the phage particles, increased transduction efficiency by 5-fold. Similar effects have been shown by Poul and Marks (1999) using a pIII-displayed anti-ErbB2 scFv.

Moreover, it has been demonstrated that phages displaying ligands fused to their major coat protein pVIII (2700 copies per phage) are able to bind to cell surface receptors, be internalized and processed inside mammalian cells 10 leading to marker gene expression 7. It is generally believed that the wild type to fusion protein ratio of pVIII displaying phage is inversely related to the size of the displayed ligand, making it difficult to functionally express larger proteins on the phage surface 11. Nevertheless, it has been shown that this correlation is not so stringent. Malik et al. (1996) demonstrated that Fab fragments and enzymes can be displayed on pVIII, but with a frequency of less than 1% 12.

We were interested in developing a gene delivery vector for the specific transduction of Hodgkin/Reed-Sternberg cells, the malignant cell type of Hodgkin's lymphoma. The CD30 receptor has been exploited as Hodgkin-lymphoma associated antigen in recent experimental therapeutic strategies 1314. Antibodies bound to CD30 are internalized by clathrin-mediated endocytosis and traffic to the lysosome 15. Regarding this internal trafficking of CD30-bound complexes as a prerequisite for gene transfer, we generated a phagemid vector that encodes an anti-CD30 scFv-pVIII fusion protein and a eukaryotic CMV-EGFP gene cassette. Out of this vector we produced small, anti-CD30 phagemid particles, capable of delivering EGFP as a reporter gene for transduction. We chose the major coat protein pVIII as the fusion partner for the display of the anti-CD30 scFv Ki-4 14. Fusion with pVIII as an alternative to the widely used minor coat protein pIII should increase the multivalency of the scFv-display and demonstrate the feasibility of displaying relatively big molecules, like scFvs (27 kDa) on pVIII.

Use of bacteriophage as vectors for targeted gene transfer into eukaryotic cells is an alternative method to circumvent some of the problems occurring with the established viral vectors. Bacteriophage have no natural tropism for eukaryotic cells but can be targeted towards specific cell surface receptors by displaying appropriate binding ligands, such as peptides or antibody fragments. Internalized by receptor-mediated endocytosis, their packaged single-stranded DNA is further processed (mechanisms not fully understood), leading to the expression of encoded genes under eukaryotic promoters 6. The ease with which one can target, modify, produce and purify these phage makes them a valuable tool for antigen-specific cell transduction. Here, we describe the generation of anti-CD30 targeted phagemid particles, specifically binding and transducing CD30-positive cells. For this purpose, the anti-CD30 Ki-4 scFv gene has been genetically fused either to the major coat protein pVIII (pAK8-Ki-4(EGFP)) or the minor coat protein pIII (pC6-Ki-4(EGFP)) in a phagemid vector, which also encodes an EGFP gene under the control of a CMV promoter. The expression of the Ki-4 scFv-pVIII fusion protein was determined by Western blot analysis (Fig. 2) and ELISA experiments with recombinant CD30 antigen demonstrating the specific binding of the Ki-4 scFv displaying phagemid particles (Fig. 3). Additionally, phage displaying the Ki-4 scFv showed a specific binding to CD30-bearing cells in FACS analyses. Binding affinity of the M13-pAK8-Ki-4(EGFP) phagemid particles to CD30 antigen and CD30 receptor-bearing cells was always higher, compared to the M13-pC6-Ki-4(EGFP) phagemid particles. Provided that scFvs fused to the pIII coat protein have the same binding affinity as pVIII-fusion proteins, this difference could be explained by a multivalent display of the anti-CD30 Ki-4 scFv on the M13-pAK8-Ki-4(EGFP) phagemid particles indicating a binding avidity effect. The multivalent display of the scFv on the M13-pAK8-Ki-4(EGFP) phagemid particles is also supported by the fact that these phagemid particles are able to induce additional cluster formation of CD30-positive cells (Fig. 5). This phenomena was not seen with the supplement of M13-pC6-Ki-4(EGFP) phagemid particles or the addition of the M13-pAK8-Ki-4(EGFP) phagemid particles to the CD30-negative cell line BL38, but has been observed previously by the addition of monoclonal Ki-4 IgG antibody to CD30-positive cells (data not shown).

The transduction experiments with different cell lines revealed that anti-CD30 scFv-displaying phagemid particles are able to transduce CD30 receptor-bearing cells. The addition of the parental monoclonal antibody Ki-4 significantly reduced the amount of CD30-positive L1236 cells transduced by the M13-pAK8-Ki-4(EGFP) phagemid particles (Fig. 6a). There was no unspecific transduction of CD30-negative cells, nor did control phagemid particles M13-pAK8(EGFP) transduce the CD30-positive cells (Fig. 6d). This indicates that CD30 antigen on the targeting cell lines is required for transduction with anti-CD30 phagemid particles and establishes the high specificity of these gene transfer particles. Incubation of the target cells with the M13-pAK8-Ki-4(EGFP) phagemid particles led to a maximum transduction rate of 5%, a rate approximately 10-fold higher compared to the M13-pC6-Ki-4(EGFP) phagemid particles. This enabled a direct comparison of pVIII- and pIII- fusion protein display for receptor-specific gene transfer. Although there is a postulated avidity effect responsible for the stronger binding of the M13-pAK8-Ki-4(EGFP) phagemid particles to recombinant CD30 and CD30-positive cells compared to the M13-pC6-Ki-4(EGFP) phagemid particles, this significant difference in transduction efficiency was somewhat surprising. Other groups using monovalent pIII-display with transduction efficiencies of 1–4% were able to increase the efficiencies up to 10% by using multivalent pIII-display 69.

It has been described that CD30 receptor is being internalized by clathrin-mediated endocytosis, maturing to lysosomes and that cross-linking of CD30 receptors by monoclonal antibodies enhances this trafficking at least by 3-fold 15. In accordance with our results, this supports our hypothesis that pVIII-based display of scFvs is multivalent, leading to increased internalization and transduction efficiency of CD30-bearing cells. Although it has been shown that multivalent scFv-pIII display on bacteriophage is possible and increases transduction efficiency 9, our approach serves as an alternative or even superior system, especially if the internalization depends on multipoint 'zipper' contacts of the phage with the cell surface receptors. A multivalent display on pVIII may provide an increased binding and phage uptake because pVIII fusion proteins are spread over the surface of the phage, in contrast to pIII fusion proteins, which are focused on the tip of the phage. This so-called 'disseminated display' would also explain why M13-pAK8-Ki-4(EGFP), in contrast to the M13-pC6-Ki-4(EGFP) phagemid particles, are able to connect CD30-positive cells leading to cluster formation (Fig. 5). This also explains the stronger binding activity of the M13-pAK8-Ki-4(EGFP) phagemid particles (Fig. 4) and the higher transduction rate (Fig. 6d). However, this potential cross-linking of cells by single M13-pAK8-Ki-4(EGFP) phagemid particles does not prevent internalization of other M13-pAK8-Ki-4(EGFP) phagemid particles, because simultaneous binding of receptors on one cell is even more likely than cross linking of receptors on separate cells.

The feasibility of displaying molecules as large as scFvs (27 kDa) on the major coat protein raises the possibility of generating bispecific phagemid particles rescued from E. coli, encoding an additional pIII- or pVIII-fusion protein, thereby increasing specificity and avidity of the cell-targeting phagemid particle.

The unequal transduction rates of the different CD30-positive cell lines with the M13-pAK8-Ki-4(EGFP) phagemid particles are not directly explainable by their individual CD30 receptor densities. For example, cell line L1236 showed only a moderate binding of the anti-CD30 phagemid particles in FACS analysis compared to the L540 cell line(Fig. 4), but revealed the highest EGFP-expression rate (Fig. 6c). This demonstrates that the efficiency of the gene transfer seems to be dependent on cell line-specific trafficking of the bacteriophage after binding. Because the gene transfer efficiency of targeted bacteriophage is in general still very low, it seems to be necessary to further optimize trafficking, DNA processing and nuclear localization of the phagemid DNA. So far, satisfying transduction rates (30–45%) with bacteriophage have been reached only in vitro by genotoxic treatment 16.

To our knowledge, this is the first approach demonstrating a potential advantage of the major coat protein pVIII as fusion partner for the display of scFvs in gene transfer experiments, compared to the commonly used minor coat protein pIII. Furthermore, it is the first approach generating a CD30-specific gene transfer vector to be used for the transduction of Hodgkin-, Reed Sternberg cells. A possible application of this system would be the specific induction of apoptosis in these cells. It is believed that cancer cells are on the verge of apoptosis, containing processed effector caspases. At the same time, they also express inhibitor of apoptosis proteins (IAPs), which can not be neutralized due to malfunctioning apoptotic pathways 17. It has been demonstrated that overexpression of second mitochondrial activator of caspases (SMAC) in Hodgkin-derived cell lines blocks X-linked IAPs activation 18, and therefore would make a promising candidate gene for restoration of the apoptotic pathway in Hodgkin's disease via targeted gene therapy. Provided that the transduction efficiency of CD30-specific bacteriophage is further increased by additional molecular modifications, they can potentially serve as specific, safe, cheap and efficient gene therapy vectors for in and ex vivo applications. This is supported by the fact that tumor treatment with targeted bacteriophage in vivo has been demonstrated recently 19.

YSAC carried out most of the experiments, analyzed the results and helped to draft the manuscript. KS participated in the experiments and helped to draft the manuscript. MK contributed to the interpretation of the data and helped to draft the manuscript. AK conceived the study, designed and performed experiments and drafted the manuscript. All authors read and approved the final manuscript.