Gene Therapy (2000) 7, 2122–2131  2000 Macmillan Publishers Ltd All rights reserved 0969-7128/00 $15.00 www.nature.com/gt VIRAL TRANSFER TECHNOLOGY RESEARCH ARTICLE -like gene transfer into cells mediated by polyoma virus pseudocapsids

N Krauzewicz1, J Stokrova´2, C Jenkins3, M Elliott3, CF Higgins1 and BE Griffin4 1MRC Clinical Sciences Centre, Imperial College School of Medicine, Hammersmith Hospital, Ducane Road, London, W12 0NN, UK; 2Institute of Molecular Genetics, Czech Academy of Sciences, Prague, Czech Republic; 3Department of Physics and Astronomy, University of Wales Cardiff, Cardiff; and 4Viral Oncology Unit, Division of Medicine, Imperial College School of Medicine at St Mary’s, London, UK

Mouse polyoma virus-like particles (or pseudocapsids) are somes or calcium phosphate precipitates. Despite the fact composed solely of recombinant viral coat protein. They can that all cells appear to internalise pseudocapsid/DNA com- interact with DNA and transport it to cells, resulting in gene plexes, only a proportion show productive gene delivery. expression both in tissue culture and in mice. We demon- Bulk internalisation of complexes is dependent on actin strate that DNA transfer in vitro depends on partial packag- fibres, but not surface sialic acid or microtubules, indicat- ing of DNA within the virus-like . Cell surface sialic ing that a second transport pathway exists for pseudocap- acid residues and an intact microtubule network, required for sids which is nonproductive for gene transfer. The model viral infectivity, are also necessary for pseudocapsid- suggested by these data demonstrates the virus-like proper- mediated gene expression from heterologous DNA. Thus, ties of the pseudocapsid system, and provides a basis for gene delivery in this system requires pathways utilised by further development to produce a highly effective gene deliv- polyoma virions, rather than proceeding via the ‘nonspecific’ ery vehicle. Gene Therapy (2000) 7, 2122–2131. endosomal route typical of nonviral systems such as lipo-

Keywords: gene therapy; papovavirus; VP1; VLPs; AFM

Introduction thesised and partial packaging of plasmid DNA has also been observed on mixing the purified components in Gene therapy is being explored as treatment for human vitro.9,18,19 Furthermore, VLPs composed of mouse poly- diseases and a number of clinical studies have success- oma virus VP1 (called pseudocapsids) can transfer plas- 1 fully demonstrated proof of principle. Viral vectors have mid DNA to cells, resulting in heterologous gene been widely used in such studies as they are able to expression both in vitro and in vivo.20–22 DNA transfer in deliver genes efficiently and achieve long-term vitro has also been reported for a number of other VLPs, expression. However, viral vectors have the potential for including those derived from human polyoma and papil- introducing or generating infectious and the loma viruses.15,16,23 Thus, VLPs have the potential to be immune system rapidly clears modified cells expressing developed into a new family of gene therapy vectors with virally encoded gene products. These factors, in conjunc- different delivery characteristics and host ranges, depen- tion with problems of large-scale production, have driven dent on the nature of the parental virus. a search for alternative nonviral delivery systems. In vivo it has been observed that genes introduced with One approach being developed uses virus-like particles pseudocapsids are initially expressed at a low level, but (VLPs) to deliver genes. VLPs are composed of recombi- that expression is prolonged and may even increase with nant structural viral coat proteins, which spontaneously time.21,22 Furthermore, in vitro expression is sustained assemble into protein spheres closely resembling the par- even though initiating from relatively few copies of DNA 2 ental virus particle. Unlike viruses they are composed compared with transfection with calcium phosphate pre- only of one protein species and are devoid of nucleic cipitates.20,21 This pattern of expression is not typical of acid. Such structures were originally described for mouse gene transfer by nonviral methods, where high levels of 3–5 polyoma virus coat protein, VP1, and it has been gene product are detected, but expression is transient.24 shown that a number of different viral coat proteins can The quantity of pseudocapsid/DNA complex required to 6–17 behave in a similar manner. VLPs retain their ability achieve expression is, however, similar to that for cal- to package DNA. Particles composed of the coat protein cium phosphate-mediated transfer, with respect to the packaged with heterologous DNA have been isolated number of DNA molecules used. Thus, from cells in which the recombinant protein is syn- pseudocapsid/DNA complexes are considerably less active than the corresponding virus particles. To provide insight into why pseudocapsid-mediated Correspondence: N Krauzewicz DNA delivery appears to favour long-term expression, Received 28 March 2000; accepted 17 August 2000 and to identify ways of improving DNA transfer VLPs transfer DNA to cells in a virus-like manner N Krauzewicz et al 2123 efficiency, the interaction of pseudocapsid/DNA com- plexes with cells was studied. Pseudocapsids were found to be taken up into cells by at least two pathways. The first route resulted in gene expression and depended on cell surface sialic acid residues and intracellular microtu- bules, being sensitive to treatment of the cells with neura- minidase and nocodazole. The second route, accounting for the bulk of the material, was cytochalasin D sensitive and, therefore, actin fibre dependent, but did not lead to productive transfer of DNA to cells. Thus, a relatively minor population of pseudocapsid/DNA complexes which enter the cell appear to transfer DNA to the nucleus in a specific virus-like manner, which may result in delivery of DNA favouring long-term gene expression. Results Partial packaging of pseudocapsids with DNA is required for gene transfer Nonviral gene transfer agents, such as cationic polymers or liposomes, form aggregates with DNA that are taken up nonspecifically by cells.25 These aggregates are com- posed of relatively disordered complexes stabilised by ionic interactions between the carrier and the DNA phos- phate backbone. In contrast, pseudocapsids and empty from a natural polyoma virus infection appear to interact with DNA by partially packaging it.19,26 We therefore investigated whether pseudocapsid-mediated gene transfer depends on a specific packaging reaction. It has been demonstrated that empty capsids package DNA most efficiently when complexes are formed at a molar ratio of 5:1 capsids:DNA.26 To determine whether the degree of packaging influences the efficiency of gene transfer, pseudocapsid/DNA complexes were prepared at different molar ratios and the resulting complexes assayed for gene transfer activity. Pseudocapsids were incubated with a plasmid encoding the enhanced green fluorescent protein (EGFP) gene and the mixtures applied to monkey epithelial cos 7 cells in a transient expression Figure 1 Partial packaging of DNA by pseudocapsids is necessary for assay (see Materials and methods). The percentage of effective gene transfer (A) Cos 7 cells (2 × 105) incubated with cells expressing EGFP for each mixture, counted by flu- pseudocapsid/DNA complexes formed at increasing molar ratios from orescence microscopy, is shown in Figure 1A. As with 0.6:1 to 20:1 pseudocapsids:DNA (0.5 ␮g with respect to pEGFP DNA; empty viral capsids, the highest transfer activity was Clontech) were scored for EGFP expression 24 h after incubation by flu- obtained at the optimal molar packaging ratio of 5:1 (15 orescence microscopy using an FITC filter set. Results given are from two ␮ ␮ independent experiments performed in duplicate; error bars of standard g pseudocapsids mixed with 0.5 g DNA), either side deviations are shown. (B) Electron micrographs of heavy pseudocapsids of which DNA transfer activity was lower, demonstrating and plasmid DNA (panel a) empty pseudocapsids and DNA (panel b), or a correlation between packaging and DNA transfer. DNA alone (panel c), mixed at a molar ratio of 5:1, prepared using a To test this relationship further, the DNA transfer modified spreading technique as previously described.19 Bar, 250 nm. (C) activity of ‘heavy pseudocapsids’ was assayed. ‘Heavy Cos 7 cells (2 × 105) incubated with heavy or empty pseudocapsid/DNA ␮ pseudocapsids’ are also derived from VP1 recombinant mixtures, or DNA alone (0.5 g), prepared at a molar ratio of 5:1, scored for EGFP expression, as described above. (D) Cos 7 cells (3 × 106) incu- baculovirus-infected insect cells, but already have hetero- ␮ 18 bated for 48 h with DNA alone (10 g) (unbroken line, left hand panel), logous DNA packaged inside them. These particles do or pseudocapsid/DNA mixtures (10 ␮g with respect to DNA) (unbroken not interact with exogenously added DNA in gel shift line, right hand panel) were harvested by trypsinisation and fixed with assays and when observed by electron microscopy formaldehyde. Cells incubated with no additions were harvested similarly (Figure 1B, panel a) they do not form stable interactions (broken line, left and right hand panels). Cells were analysed for EGFP with plasmid DNA, in contrast to ‘empty pseudocapsids’ expression by flow cytometry as described in Materials and methods. A 19 significant background autofluorescence peak (AF peak) was observed in (panel b; plasmid DNA alone is shown in panel c). all cases. Cells showing fluorescence above this background level (indicated Heavy pseudocapsids exhibited little or no DNA transfer by the marker bar) were counted and are given as a percentage of total activity when mixed with EGFP DNA and tested in the events. Results shown are a representative example of three separate cos cell assay, in contrast to empty pseudocapsids, which experiments. gave more than a 10-fold increase in gene transfer efficiency compared with DNA alone (Figure 1C). These results confirm that a specific packaging reaction is required to form complexes of pseudocapsids and DNA capable of mediating DNA transfer.

Gene Therapy VLPs transfer DNA to cells in a virus-like manner N Krauzewicz et al 2124 DNA transfer in these experiments was only observed ever, more VP1 was detected by Western blotting in the for a relatively small percentage of cells. However, flow central fractions of the gradient (fractions 4–6) with cytometric analysis of similar cultures (Figure 1D) meas- refractive indices of 1.36–1.37, corresponding to the ured 8.33% of cells fluorescing above background auto- migration position of free pseudocapsids in gradients run fluorescence (AF peak) when treated with under similar conditions.21,27,28 Thus, more DNA transfer pseudocapsid/DNA complexes (right hand panel), com- activity was associated with VP1 derived from the heav- pared with 0.42% of cells treated with DNA alone (left ier fractions than the lighter small aggregates. hand panel). A wide range of fluorescence intensities was These results suggest a model for pseudocapsid–DNA observed by flow cytometry in this EGFP expressing cell interaction. Initially, a packaging reaction takes place, population, suggesting that, by microscopy, only the which is of high enough affinity to withstand the spread- most fluorescent cells are detected. However, as the ing forces in EM. Then in the presence of additional microscope assay provides a rapid and economical pseudocapsids or aggregation, complexes better able to means of comparing relative gene transfer, it was used productively transfer DNA into cells are formed. As gene further to study DNA transfer activity. transfer is reduced if pseudocapsids are mixed at molar ratios greater than 5:1, the greater excess of pseudocap- Gene transfer is enhanced on association of sids may drive the complexes into even larger aggregates pseudocapsid/DNA complexes into larger aggregates which the cells are unable to take up efficiently. The results indicate that specific packaging is necessary for efficient gene transfer of exogenous DNA by pseudo- Pseudocapsids enter cells by two routes, only one of capsids. However, a significant portion of DNA associa- which is productive for gene transfer ted with pseudocapsids appeared not to be packaged when observed by EM (Figure 1B, panel b), raising the Pseudocapsids are endocytosed by cells and transported question of how the DNA is protected from nucleases in a cytoskeleton-dependent manner: Although a packag- during gene transfer. Furthermore, although the optimal ing step is required for DNA transfer, large loosely asso- molar ratio was 5:1, on average only two to four pseudo- ciated complexes of pseudocapsids and DNA appear to capsids could be seen on each DNA molecule. One poss- be most efficient at mediating gene expression. To under- ible explanation for these observations is that the remain- stand how these complexes transfer DNA to cell nuclei, ing capsids interact with the DNA at a lower affinity, but early stages of cell entry and uptake of pseudocapsids are displaced by the spreading forces used to prepare were characterised by immunofluorescence. Initially, samples for EM.19 Evidence to support this hypothesis binding and entry of pseudocapsid/DNA complexes into was provided by examining the complexes using atomic cells was examined as a function of time. The assay was force microscopy (AFM). Sample preparation for this performed in both mouse 3T6 fibroblasts (Figure 4, method is less disruptive than for EM, as the samples are panels a–c) which support polyoma virus replication, and allowed to adsorb on to the hydrophobic surface of a cos 7 cells (panels d–f) which were used for the DNA mica sheet and then gently dried. Pseudocapsids or DNA transfer assays. Pseudocapsid/DNA complexes were alone observed by AFM distributed evenly across the allowed to adsorb on to the surface of cells at 0°C, cul- mica surface (Figure 2a and b). Pseudocapsid/DNA com- tures were shifted to 37°C and uptake observed after 0, plexes formed at a molar ratio of 0.08:1 resulted in struc- 30 min and 3 h by indirect immunofluorescence with an tures similar to those observed by EM (Figure 2c), in that antibody raised against VP1 (panels a and d, b and e, and the DNA was partially unwound and associated with c and f, respectively). Pseudocapsids were endocytosed occasional spherical pseudocapsid-like structures and transported within the cell, reaching steady state by (indicated by an arrow). However, when the ratio of 2–4 h. In 3T6 cells, most VP1-specific fluorescence was pseudocapsids to DNA was increased, the majority of the found near the nucleus, whereas in cos 7 cells it was seen material associated into larger particles, with dimensions both near the nucleus and at the cell periphery. Panels g of up to a micrometre. Figure 2, panels d and e show 10 and h (Figure 4) show images of pseudocapsid-derived ␮m and 1 ␮m scans of this sample, demonstrating associ- material in a single 3T6 or cos 7 cell after a 3 h incubation ation of pseudocapsid/DNA complexes both at low and at 37°C, captured as a series of planes through the Z axis high resolution. Some smaller structures were also of the cell and reconstructed in three dimensions (see observed in these samples (for example, Figure 2f; 0.25 Materials and methods). The cells were also stained with ␮m scan), with several pseudocapsids per DNA mol- Texas red-conjugated phalloidin to visualise actin fibres.

ecule. Thus, it appeared that in solution the majority of Two different XZ sections (panels g1 and 2, and h1 and 2) pseudocapsid/DNA complexes form much larger, were taken through each cell, at positions indicated by loosely associated structures, in which many molecules the blue lines in panels g and h, of 0.134 ␮m width. These of DNA and pseudocapsids aggregate together. sections demonstrate that much of the VP1 immmunore- To determine whether the smaller complexes and active material (green fluorescence) is located very close larger aggregates observed by AFM were capable of to the actin fibres (red fluorescence) and shows that a DNA transfer, pseudocapsids and DNA mixed at a molar significant proportion of pseudocapsids in each case have ratio of 5:1 were fractionated on a 10–40% sucrose gradi- migrated into the cell cytoplasm. Furthermore, a similar ent. Ten fractions were collected and tested for both DNA localisation was observed with polyoma virion particles transfer activity and VP1 content. The highest gene trans- (data not shown). Uptake to this internal location could fer activity was associated with the heavier fractions not be explained by membrane recycling, as labelling (Figure 3, fractions 1–3) and peaked in fraction 2. Con- plasma membranes with the lipophilic carbocyanin, siderable activity was also associated with fraction 10. DiIC16(3), resulted in accumulation of the dye into cellu- This fraction, being the last collected, also contained lar locations that differed from those seen for pseudocap- insoluble matter from the bottom of the gradient. How- sids (data not shown). Very little colocalisation was

Gene Therapy VLPs transfer DNA to cells in a virus-like manner N Krauzewicz et al 2125

Figure 2 Analysis of pseudocapsid/DNA complexes by atomic force microscopy. Pseudocapsids (panel a: 1.39 ␮m scan, 25 nm height range) or DNA (panel b: 2.00 ␮m scan, 8 nm height range) were diluted in water to 1 ␮g/ml and dropped on to the surface of freshly cleaved mica. Following overnight drying in a desiccator, samples were scanned using a silicon probe, in tapping mode. Complexes were formed between pseudocapsids and DNA at a molar ratio of 0.08:1 (panel c: 1.50 ␮m scan, 3 nm height range) and 1.3:1 (panels d, e and f: 10, 1 and 0.25 ␮m scans, 30, 20 and 10 nm height ranges, respectively) and visualised following dilution to 1 ␮g/ml with respect to pseudocapsids. The capsid-like structure indicated in panel c (arrow) appears flattened due to the low height range used for visualising the DNA. observed in double labelling experiments with antibodies prisingly, removal of cell surface sialic acid residues had raised against cathepsin B (Figure 4i), indicating that the no discernible effect on either the binding of pseudocap- compartment occupied by pseudocapsids was not pre- sids at 0°C to the surface of either type of cell (compare dominantly lysosomal. Also, VP1 did not colocalise with panels a and c, or e and g, for 3T6 or cos 7 cells, the microtubule organising centre and is, therefore, respectively), or their migration into the cell (panels b unlikely to be associated with Golgi bodies. The closely and d, or f and h). This suggests that uptake of the bulk related virus, SV40, has been shown to accumulate at the of pseudocapsids in this system is not receptor mediated, following cell entry.29 However, nor does it result in productive DNA transfer. Thus gene the asymmetric staining pattern observed for pseudocap- transfer resulting in gene expression appears to be exclus- sids is not typical for this structure. Thus, the identity of ively sialic acid/receptor dependent. Further evidence for the sites to which pseudocapsid/DNA complexes a specific route of delivery is given below. migrate remains unresolved. Intracellular transport of pseudocapsids requires the cyto- Pseudocapsid-mediated DNA transfer requires cell sur- skeleton: Occasionally, VP1-specific fluorescence was face sialic acid residues: Viral infection of cells depends observed in filamentous patterns in cells (examples are upon binding to a sialic acid component of the viral indicated with arrows in Figure 4c). Intracellular trans- receptor.30 To test whether sialic acid binding is necessary port of a number of viruses depends on cytoskeletal for pseudocapsid uptake, DNA transfer was carried out elements.32–34 To investigate whether this may also be the in the presence of neuraminidase, which cleaves sialic case for pseudocapsids, cells were incubated with acid residues from the surface of cells and blocks virus pseudocapsid/DNA complexes, fixed after 3 h, and co- binding and infection.31 As shown in Figure 5A (filled stained for either microtubules, or actin filaments, and bars), this treatment resulted in a complete block of both pseudocapsids. Some colocalisation of pseudocapsids virus and pseudocapsid-mediated gene transfer, as meas- was observed with both cytoskeletal structures. A net- ured by EGFP expression for pseudocapsids and early work of microtubules can be seen in cos 7 cells in Figure antigen (large-T) expression for the virus. However, neu- 6A (panel a, red fluorescence) with pseudocapsids (green raminidase treatment of the cells did not affect calcium fluorescence) in aggregates near the nucleus. 3T6 cell phosphate transfection of the EGFP gene. Thus, pseudo- microtubules were unstable in this assay, possibly due to capsids retain the need for binding to sialic acid to the adsorption step that was carried out at 0°C. A similar mediate DNA transfer. pattern would have probably been observed, however, The effect of neuraminidase on uptake of pseudocap- given the localisation of pseudocapsids in these cells, as sids, as measured by immunofluorescence, was also seen in Figures 4 and 5 and Figure 6A, panel c. More tested. 3T6 or cos 7 cells were treated with neuraminidase prominent colocalisation appeared to exist with actin and pseudocapsid uptake observed by immunofluoresc- fibres. In 3T6 cells (panel c) pseudocapsid complexes ence (Figure 5B, panels a–d and e–h, respectively). Sur- often appeared to be sited directly on the phalloidin

Gene Therapy VLPs transfer DNA to cells in a virus-like manner N Krauzewicz et al 2126 with DNA/calcium phosphate precipitate. Under the conditions used here, neither reagent reduced calcium phosphate transfection efficiency (Figure 6B, right-hand panel). Nocodazole actually appeared to result in an increase in transfection, possibly due to mitosis being retarded in the cells, increasing the chances of DNA entering the nucleus during nuclear envelope break- down. In contrast, DNA transfer by both pseudocapsids and virus (left and central panels) was almost completely abolished by treatment with nocodazole, but not by cyto- chalasin D. Thus, pseudocapsids and virus require intact microtubules and not actin fibres to mediate gene trans- fer. These data support the notion that pseudocapsids enter cells and deliver DNA by a regulated mechanism, similar to viral entry. Furthermore, consistent with the experiments using neuraminidase, they show that there are two routes of uptake of pseudocapsids in cells. One of these is sialic acid and microtubule dependent and results in gene transfer, whereas the other is actin fibre dependent and is not involved in gene transfer.

Discussion The major coat proteins of several papovaviruses have the ability to self-assemble into spherical particles with morphologies similar to those of the parental virus. We and others have demonstrated that these particles can interact with cells and carry heterologous DNA into them 15,16,20,21,23 Figure 3 Fractionation of pseudocapsid/DNA complexes by sucrose den- in a manner that results in gene expression. In sity gradient centrifugation. Pseudocapsid/EGFP DNA complexes, formed this article, we demonstrate that the mechanism of DNA at a molar ratio of 5:1, overlayed on to a 10–40% sucrose gradient and transfer by polyoma pseudocapsids exhibits many centrifuged for 1 h. Collected fractions (two-fifths of each) were added to characteristics in common with virus infection, rather cos 7 cells and cells expressing EGFP counted by fluorescence microscopy, than delivering DNA by nonspecific routes typical of 24 h later (bars). Inset: one hundredth of each fraction was run on 12% nonviral systems. We propose that VP1 is sufficient for SDS-PAGE, electroblotted on to Immobilon P membrane (Millipore) and probed with an anti-VP1 monoclonal antibody.5 Immune complexes were DNA binding and transfer of the pseudocapsid/DNA visualised with an HRP-conjugated secondary antibody and chemilumi- complexes to the nucleus via a receptor-mediated route, nescent ECL reagent (Amersham). Only the portion of the gel containing resulting in efficient gene expression. Thus, pseudocap- VP1 is shown. sids retain many of the advantages of a viral vector, and provide a useful class of delivery systems for gene transfer in vivo. stained fibres, whereas in cos 7 cells, where the fibres Pseudocapsid-mediated gene transfer depends on a were less well formed, colocalisation was less obvious. physical interaction of DNA with ‘empty’ pseudocapsids, As both capsids and cytoskeleton components were involving partial packaging of the DNA. It is, like virus abundant in the cells, it is possible that a certain amount infection, almost completely blocked by treatment of the of overlapping signal could be coincidental. To test this, cells with neuraminidase, indicating that DNA transfer cells were treated with nocodazole (cos 7 cells) or cytoch- by pseudocapsids is exclusively receptor mediated and alasin D (3T6 cells) that depolymerise microtubules and does not occur by passive, or nonspecific uptake. Sub- actin filaments, respectively. Figure 6A (panels b and d) sequent steps in the passage of the complexes to the show the results of pretreatment of cells with these nucleus also occur in a virus-like manner, since both reagents, followed by incubation with pseudocapsid/ virus and pseudocapsids require intact microtubules, but DNA complexes for 3 h. Cytochalasin D (panel d) not actin fibres, to deliver their DNA to the nucleus for appeared to have the greater effect, virtually abolishing expression. Whether this transport is regulated by infor- localisation of pseudocapsids to the site near the nucleus. mation encoded in VP1, or determined by the initial bind- A pattern of phalloidin stained material similar to that ing event, is not known. Microtubules regulate many found in cos 7 cells was observed, with lightly stained events concerned with transport of vesicles in cells. indistinct fibres and globular phalloidin stained material. Recent data have highlighted their role in movement of Also, capsid distribution was more reminiscent of that endocytic vesicles in both retrograde and anterograde found in cos 7 cells. Taken together, these data suggest directions and a number of viruses travel towards the that the state of the actin fibre network has a significant nucleus along microtubule structures.33–36 Polyoma virus influence on pseudocapsid localisation in this assay. and pseudocapsid/DNA complexes may, therefore, use To test the relevance of this observation to DNA trans- a similar mechanism to deliver DNA to the nucleus. fer leading to gene expression, cells were incubated in Alternatively, the cell surface accessibility of the receptor the presence of nocodazole or cytochalasin D and then for pseudocapsids may be microtubule dependent, as has treated either with pseudocapsid/DNA complexes, been described previously.37 Since the receptor for mouse infected with wild-type polyoma virus, or transfected polyoma virus has yet to be identified this cannot be

Gene Therapy VLPs transfer DNA to cells in a virus-like manner N Krauzewicz et al 2127

Figure 4 Uptake of pseudocapsid/DNA complexes into cells with time. Panels a–c: 3T6 cells (5 × 104) incubated with pseudocapsid/DNA complexes at a molar ratio of 5:1 (0.06 ␮g with respect to DNA) at 0°C for 30 min and shifted to 37°C, fixed in 3% formaldehyde at: (a) 0 h; (b) 30 min; (c) 3 h. Cells were permeabilised with 0.1% Triton X 100 and pseudocapsid uptake followed in cells by immunofluorescence with anti-VP1 monoclonal antibody and Oregon green-conjugated secondary antibody (Molecular Probes).5 Cell nuclei were counterstained (blue) with DAPI. Arrows in c indicate examples of VP1 containing complexes apparently aligning on filamentous structures within the cell. Panels d–f: VP1 immunoreactive material localised in cos 7 cells treated as above and fixed at: (d) 0 h; (e) 30 min; (f) 3 h. Panels g and h: a single 3T6 cell (g), or cos 7 cell (h), treated as above and harvested at 3 h, immunostained for VP1 (green), and stained for actin fibres with Texas red-conjugated phalloidin (Molecular Probes) (red). Stacks of images were taken through each cell at 0.3 ␮m intervals, reconstructed (as described in Materials and methods) and XZ orthogonal planes of 0.134 ␮m thickness were taken at two different places, as indicated by the blue lines in panels g and h (panels g1and g2 and h1 and h2, respectively). The pixels on the Z axis of panels g1,g2,h1 and h2 have been artificially increased three-fold to facilitate viewing of the images. Panel i: 3T6 cells treated as in above, harvested at 3 h, stained for VP1 (red) using an RITC-conjugated secondary antibody and for cathepsin B (green), with FITC-conjugated secondary antibody. Images for each fluorophore were captured separately and merged digitally (see Materials and methods). Bars, 20 ␮m.

ruled out, but it would still be consistent with receptor- can associate into large aggregates in the presence of mediated uptake. DNA, as, even in low ionic strength media, particles of In common with the accepted model of polyoma virus micrometre dimensions could be observed. Such aggre- uptake, pseudocapsid/DNA complexes have at least two gates could reduce the effectiveness of transfer by seques- modes of entry into the cell, one resulting in gene transfer tering material. However, as virtually all cells can be seen and the other apparently nonproductive.38 The majority to have taken up VP1, it seems more likely that the size of VP1 immunoreactive material was observed to migrate of the particle may impede specific receptor-mediated to an intracellular location near the nucleus, which was internalisation. Alternatively, differences in surface not lysosomal in nature. This movement was sensitive to characteristics or conformation induced by abnormal cytochalasin D treatment and, given that a number of packaging of plasmid DNA, or lack of the minor coat pro- VP1 containing particles colocalised with actin fibres, teins, VP2 and 3, might affect interactions with receptors may depend on regulated movement along the fibres or transport proteins. In support of this notion, pseudo- themselves. This population of complexes, however, was capsids made from VP1 and VP2 are more effective in apparently non-productive for DNA transfer since no blocking interaction of virus with the cell, than those reduction in gene expression was observed in the pres- composed of VP1 alone.39 Thus, a particular function may ence of cytochalasin D. It may be that this route rep- be missing from pseudocapsids, which, if replaced, could resents a default pathway, for instance by phagocytosis, restore the efficiency of transfer to levels nearer that achi- for material that did not follow the preferred viral route. eved with viruses. In this regard, virus infection (and in one case One of the major drawbacks in the field of gene ther- pseudocapsid/DNA transfer) in cells was more efficient apy is the lack of a suitably safe and efficient vector that in the presence of cytochalasin D. One explanation for can confer long-term expression in target tissues.1,40,41 The this observation could be that when the actin transport polyoma virus-like particle system has been shown to pathway was blocked, some particles were re-routed to deliver DNA to cells and tissues in a way that results in the productive pathway. prolonged gene expression.20,21 This property may be a It is not known why such a large proportion of VP1 consequence of the specific virus-like nature of uptake adopts this actin-dependent route, although it may and transport of DNA to the nucleus, as reported here. account for why pseudocapsids are much less effective The identification of two uptake pathways may provide at DNA transfer than virus. It is clear that pseudocapsids at least a partial explanation for the current inefficiency

Gene Therapy VLPs transfer DNA to cells in a virus-like manner N Krauzewicz et al 2128

Figure 5 Pseudocapsid DNA transfer in cells treated with neuraminidase. (a) Cells (2 × 105) treated with tissue culture medium containing 50 U neuraminidase/ml (filled boxes) or, as control, medium alone (open boxes) for 30 min, were incubated with pseudocapsid/DNA complexes (0.5 ␮g with respect to DNA) (cos 7 cells, left hand panel), with 0.1 p.f.u. per cell polyoma virus (3T6 cells, centre panel), or transfected with DNA/calcium phosphate precipitate (0.5 ␮g with respect to DNA) (cos 7 cells, right hand panel). Cells were incubated for a further 23 h in the presence or absence of neuraminidase, then scored for EGFP (left and right hand panels), or large T (LT) antigen expression (central panel). (b) Control 3T6 cells (panels a and b) and cos 7 cells (panels e and f), or cells treated with neuraminidase, as above (3T6 cells, panels c and d; cos 7 cells, panels g and h), analysed for pseudocapsid uptake at 0 h (panels a, c, e and g) or 3 h (panels b, d, f and h) performed as described in the legend to Figure 4. Pseudocapsid uptake was detected with anti-VP1 antibody, followed by an Oregon green-conjugated secondary antibody. Bar, 20 ␮m.

of pseudocapsids in DNA transport and provide a basis ating purified pseudocapsids, diluted 10-fold with water, for optimising the system. Once the relatively low with CsCl purified supercoiled plasmid DNA at a molar efficiency of DNA transfer is bettered, this new system ratio of 5:1, corresponding to a weight ratio of 30:1, for should prove a promising candidate for future gene 15 min at room temperature, unless stated otherwise. therapeutic applications. EGFP gene transfer assays Materials and methods For pseudocapsid-mediated transfer, cells washed with Dulbecco’s modified Eagle’s medium (DMEM) lacking All chemicals were obtained from Sigma Chemical Co. or serum were incubated with pseudocapsid/DNA com- BDH Chemicals, Poole, UK. Secondary and conjugated plexes in a small volume of DMEM for 90 min. Cultures antibodies (used at a 1/200 dilution, unless otherwise were then washed and grown in DMEM supplemented noted) were obtained from Dako (Ely, UK). Anti-cathep- with 5% FCS for 24 h. Cos 7 cells express SV40 LT and sin B antibody and FITC-conjugated secondary antibody plasmid pEGFP (Clontech, Palo Alto, CA, USA) contains were from Santa Cruz Biotechnology (Autogen Bioclear, an SV40 origin of replication, resulting in amplification of Calne, UK). DNA to facilitate detection of marker genes in transient cultures.42 Calcium phosphate transfections and virus Preparation of pseudocapsids and pseudocapsid/DNA infections were carried out as previously described.27,43 complexes Experiments were typically carried out on 2–3 × 105 cells VP1 was expressed in Sf9 insect cells from a recombinant plated on 13-mm diameter coverslips. Unless stated baculovirus.5 Pseudocapsids were purified to near hom- otherwise, 0.5 ␮g DNA was used per coverslip and at the ogeneity by differential gradient centrifugation, from end of the incubation period coverslips were rinsed in cells infected for 96 h, as previously described.21 PBS, inverted on to a slide and expressing cells counted Pseudocapsid/DNA complexes were prepared by incub- by fluorescence microscopy using an FITC filter set.

Gene Therapy VLPs transfer DNA to cells in a virus-like manner N Krauzewicz et al 2129 AFM: Samples were diluted in water to 1 ␮g/ml, dropped on to the surface of freshly cleaved mica, and allowed to adsorb for 2 min. The mica surface was then rinsed with water and the samples dried overnight under mild vacuum, in a desiccator. Silicon probe tips were used to scan the samples in tapping mode using a Multi- mode SPM and Nanoscope III control system (Digital Instruments, Santa Barbara, CA, USA).

Flow cytometry analysis Cos 7 cells, untreated, or treated with DNA or pseudocapsid/DNA mixtures, were harvested 48 h later by trypsinisation. Trypsin was inactivated by dilution in DMEM supplemented with 5% FCS, cells were washed, resuspended in PBS and sorted on a Becton Dickinson Vantage FACS (San Jose, CA, USA), using an argon laser at 488 nm wavelength. Fluorescent events were gated for forward and side scatter typical of live single cells. Ten thousand events were counted for each sample and posi- tive cells defined as having a fluorescence level above the background autofluorescence peak. Data were analysed using CellQuest version 3.1F. A small percentage of highly fluorescent events was observed in the pseudocapsid/DNA mixture samples. These events dis- tributed randomly throughout the bulk population of healthy cells confirming they were highly fluorescent cells.

Gradient analysis Pseudocapsid/DNA complexes (600 ␮g with respect to pseudocapsids), prepared as described above, were over- layed on to a pre-formed sucrose gradient (10–40% in 20 mm Tris, 150 mm NaCl, 0.01 mm CaCl2, pH 7.4) and sedi- mented by centrifugation at 100000 g for 1 h. Fractions were collected and immediately applied to cells diluted in DMEM for DNA transfer assays. Figure 6 Dependence of DNA transfer by pseudocapsids and polyoma virus on microtubules, but not actin fibres. (a) Panels a, b: Cos 7 cells Uptake assays and immunofluorescence incubated in medium alone (a), or medium containing 0.4 ␮g/ml nocoda- Cos 7 or 3T6 cells (5 × 104) were grown on glass coverslips zole (b) then incubated with pseudocapsid/DNA complexes, as described in 24-well dishes. For the uptake assays, cells were incu- above, in the absence or presence of nocodazole for 3 h. VP1 (green) and bated in DMEM, containing 20 mm Hepes, pH 7, and ␤ microtubules (red; Cy3-conjugated anti- -tubulin antibody), visualised by 0.5% BSA, for 30 min at 37°C, followed by incubation immunofluorescence. Some fluorescence emission from the Cy3 used for microtubule staining was detected with the FITC filter set, causing the with pseudocapsid/DNA complexes in ice-cold medium microtubules to appear orange in the overlayed images. In panel a the for 30 min on ice. Following several rinses with cold microtubule organising centre is indicated (MTOC). Panel c, d: 3T6 cells medium, cells were incubated in DMEM supplemented treated as for panels a and b, but with 1 ␮m cytochalasin D, and visualised with 5% FCS for the appropriate time at 37°C. Harvesting for VP1 (green), actin fibres (red; Texas red-conjugated phalloidin) and was by fixing in 3.7% formaldehyde, 250 mm sucrose in nuclei (blue). Images for each fluorophore were captured separately and PBS, for 15 min, and permeabilisation in 0.1% Triton X merged digitally (see Materials and methods for details). Panels a–d, bar, ␮ × 5 ␮ 100. Pseudocapsids were localised by incubating with a 20 m. (b) Cos 7 cells (2 10 ) incubated with either 0.4 g/ml nocoda- 5 zole (grey bars), 1 ␮m cytochalasin D (black bars), or culture medium VP1-specific monoclonal antibody for 1 h. Immune com- alone (open bars) for 30 min, then treated with pseudocapsid/DNA com- plexes were detected with either Oregon green-conju- plexes (0.5 ␮g with respect to DNA; left panel), 0.1 p.f.u. per cell polyoma gated secondary anti-mouse Ig polyclonal antibody virus (central panel), or DNA/calcium phosphate precipitate (0.5 ␮g with (Molecular Probes, Leiden, The Netherlands), or rabbit respect to DNA; right panel), in the presence of the appropriate depoly- anti-mouse Ig polyclonal antibody and an RITC-conju- merising reagent. Cells were washed, incubated for a further 24 h in the gated swine anti-rabbit Ig antibody. Microtubules were presence of reagent then scored for expression of EGFP or LT antigen, as ␤ described in the legend to Figure 4. Experiments were performed three detected with Cy3-conjugated anti -tubulin (Sigma, times in duplicate and error bars showing standard deviations are given. Poole, UK), and actin fibres with Texas red-conjugated phalloidin (Molecular Probes), according to the manufac- turer’s instructions. Cathepsin B was localised using an Electron microscopy and atomic force microscopy anti-cathepsin B antibody, at a 1/20 dilution. Cells were mounted in Vectorshield containing DAPI (Vector Lab- EM: Complexes of pseudocapsids and DNA, or DNA oratories, Burlingame, CA, USA). Virus infection was alone (50 ␮g/ml with respect to DNA) were prepared assessed by scoring the number of cells expressing the for EM using a modified aqueous spreading technique, early LT antigen after 24 h by immunofluorescence of as described.19 fixed cells, using a monoclonal antibody to LT.44

Gene Therapy VLPs transfer DNA to cells in a virus-like manner N Krauzewicz et al 2130 Cells were visualised by fluorescence microscopy using 13 Shishido Y et al. Assembly of JC virus-like particles in COS7 FITC, RITC or DAPI filter sets and Plan-apochromat × 63 cells. J Med Virol 1997; 51: 265–272. and × 100 lenses. Images were captured using the Digital 14 Unckell F, Streeck RE, Sapp M. Generation and neutralization Acquire function of the Metamorph Imaging program of pseudovirions of human papillomavirus type 33. J Virol 1997; 71: 2934–2939. (Universal Imaging Systems, West Chester, PA, USA). 15 Goldmann C et al. Molecular cloning and expression of major Exposure times were typically 3 s for pseudocapsids, 1.5 structural protein VP1 of the human polyomavirus JC virus: for- s for microtubules and actin fibres and 35 ms for nuclei. mation of virus-like particles useful for immunological and Images were overlayed using the Colour Align function. therapeutic studies. J Virol 1999; 73: 4465–4469. Where a wide dynamic range of signals was obtained, 16 Ou WC et al. The major capsid protein, VP1, of human JC virus stacks of section were taken 0.3 ␮m apart and images expressed in Escherichia coli is able to self-assemble into a capsid- were then deblurred and reconstructed using the Auto- like particle and deliver exogenous DNA into human kidney deblur and Metamorph programs. Images from different cells. J Gen Virol 1999; 80: 39–46. fluorophores were colour aligned and three-dimensional 17 Tellinghuisen TL et al. In vitro assembly of alphavirus cores by reconstructions performed. Reconstructed images are using nucleocapsid protein expressed in Escherichia coli. J Virol ° 1999; 73: 5309–5319. shown as viewed from the top (angle of 0 ). XZ sections 18 Gillock ET et al. Polyomavirus major capsid protein VP1 is cap- were taken using the Orthogonal Plane function, and ␮ able of packaging cellular DNA when expressed in the baculo- images of a single pixel (0.134 m) thickness shown. virus system. J Virol 1997; 71: 2857–2865. 19 Stokrova´ J et al. Interactions of heterologous DNA with poly- omavirus major structural protein, VP1. FEBS Letts 1999; 445: Acknowledgements 119–125. We thank M Stevens and J Forstova´ for many helpful dis- 20 Forstova´ J et al. Polyoma virus pseudocapsids as efficient car- riers of heterologous DNA into mammalian cells. Hum Gen Ther cussions, A Sardini and G Warnes for discussion and 1995; 6: 297–306. assistance with light microscopy and flow cytometry and 21 Soeda E et al. Enhancement by polylysine of transient, but not I Robinson for help with AFM. Support is acknowledged stable, expression of genes carried into cells by polyoma VP1 from The Wellcome Trust (Grant No. 048711/Z/96), the pseudocapsids. Gene Therapy 1998; 5: 1410–1419. Medical Research Council and the European Community 22 Krauzewicz N et al. Sustained ex vivo and in vivo transfer of a (No. BIO4-CT97–2147). The scanning force microscope reporter gene using polyoma virus pseudocapsids. Gene Therapy was purchased with assistance from the HEFCW Tech- 2000; 7: 1094–1102. nology Foresight Initiative. 23 Touze´ A, Coursaget P. In vitro gene transfer using human papil- lomavirus-like particles. Nucleic Acids Res 1998; 26: 1317–1323. 24 Prince HM. Gene transfer: a review of methods and appli- References cations. Pathology 1998; 30: 335–347. 25 Gao X, Huang L. Cationic liposome-mediated gene transfer. 1 Verma IM, Somia N. Gene therapy – promises, problems and Gene Therapy 1995; 2: 710–722. prospects. Nature 1997; 389: 239–242. 26 Aposhian HV, Thayer RE, Qasba PK. Formation of nucleoprot- 2 Krauzewicz N, Griffin BE. Polyoma and papilloma virus vectors ein complexes between polyoma empty capsids and DNA. J for cancer gene therapy. Adv Exp Med Biol 2000; 465: 73–82. Virol 1975; 15: 645–653. 3 Salunke DM, Caspar DL, Garcea RL. Self-assembly of purified 27 Krauzewicz N et al. Myristylated polyomavirus VP2: role in the polyomavirus capsid protein VP1. Cell 1986; 46: 895–904. life cycle of the virus. J Virol 1990; 64: 4414–4420. 4 Montross L et al. Nuclear assembly of polyomavirus capsids in 28 Soeda E. Developing polyomaviral pseudocapsids for gene insect cells expressing the major capsid protein VP1. J Virol 1991; transfer; delivery into cells with poly-l-lysine. London: Imperial 65: 4991–4998. College School of Medicine, 1999, PhD dissertation. 5 Forstova´ J, Krauzewicz N, Wallace S, Street AJ et al. Cooperation 29 Kartenbeck J, Stukenbrok H, Helenius A. Endocytosis of simian of structural proteins during late events in the life cycle of poly- virus 40 into the endoplasmic reticulum. J Cell Biol 1989; 109: omavirus. J Virol 1993; 67: 1405–1413. 2721–2729. 6 Hagensee ME, Yaegashi N, Galloway DA. Self-assembly of 30 Stehle T, Harrison SC. Crystal structures of murine polyoma- human papillomavirus type 1 capsids by expression of the L1 virus in complex with straight-chain and branched-chain sialyl- protein alone or by coexpression of the L1 and L2 capsid pro- oligosaccharide receptor fragments. Structure 1996; 4: 183–194. teins. J Virol 1993; 67: 315–322. 31 Fried H, Cahan LD, Paulson JC. Polyoma virus recognizes spe- 7 Rodgers RE, Chang D, Cai X, Consigli RA. Purification of recom- cific sialyloligosaccharide receptors on host cells. Virology 1981; binant budgerigar fledgling disease virus VP1 capsid protein 109: 188–192. and its ability for in vitro capsid assembly. J Virol 1994; 68: 32 Dramsi S, Cossart P. Intracellular pathogens and the actin cytos- 3386–3390. keleton. Ann Rev Cell Dev Biol 1998; 14: 137–166. 8 Volpers C, Schirmacher P, Streeck RE, Sapp M. Assembly of the 33 Kalicharran K, Dales S. Involvement of microtubules and the major and the minor capsid protein of human papillomavirus microtubule-associated protein tau in trafficking of JHM virus type 33 into virus-like particles and tubular structures in insect and components within neurons. Adv Exp Med Biol 1995; 380: cells. Virology 1994; 200: 504–512. 57–61. 9 Pawlita M et al. DNA encapsidation by viruslike particles 34 Sodeik B, Ebersold MW, Helenius A. Microtubule-mediated assembled in insect cells from the major capsid protein VP1 of transport of incoming virus 1 capsids to the B-lymphotropic papovavirus. J Virol 1996; 70: 7517–7526. nucleus. J Cell Biol 1997; 136: 1007–1021. 10 Touze´Aet al. Production of human papillomavirus type 45 35 Lippincott-Schwartz J, Smith CL. Insights into secretory and virus-like particles in insect cells using a recombinant baculo- endocytic membrane traffic using green fluorescent protein virus. FEMS Microbiol Letts 1996; 141: 111–116. chimeras. Curr Opin Neurobiol 1997; 7: 631–639. 11 Chang D et al. Self-assembly of the JC virus major capsid pro- 36 Scales SJ, Pepperkok R, Kreis TE. Visualization of ER-to-Golgi tein, VP1, expressed in insect cells. J Gen Virol 1997; 78: 1435– transport in living cells reveals a sequential mode of action for 1439. COPII and COPI. Cell 1997; 90: 1137–1148. 12 Sandalon Z, Oppenheim A. Self-assembly and protein–protein 37 Geppert TD, Lipsky PE. Association of various T cell-surface interactions between the SV40 capsid proteins produced in molecules with the cytoskeleton. Effect of cross-linking and acti- insect cells. Virology 1997; 237: 414–421. vation. J Immunol 1991; 146: 3298–3305.

Gene Therapy VLPs transfer DNA to cells in a virus-like manner N Krauzewicz et al 2131 38 Mackay RL, Consigli RA. Early events in polyoma virus infec- 42 Gluzman Y. SV40-transformed simian cells support the repli- tion: attachment, penetration, and nuclear entry. J Virol 1976; 19: cation of early SV40 mutants. Cell 1981; 23: 175–182. 620–636. 43 Graham FL, van der Eb AJ. Transformation of rat cells by DNA 39 An K et al. Use of the baculovirus system to assemble polyoma- of human adenovirus 5. Virology 1973; 54: 536–539. virus capsid-like particles with different polyomavirus struc- 44 Dilworth SM, Griffin BE. Monoclonal antibodies against poly- tural proteins: analysis of the recombinant assembled capsid- oma virus tumor antigens. Proc Natl Acad Sci USA 1982; 79: like particles. J Gen Virol 1999; 80: 1009–1016. 1059–1063. 40 Anderson WF. Human gene therapy. Nature 1998; 392: 25–30. 41 Alton EW et al. Towards gene therapy for cystic fibrosis: a clini- cal progress report. Gene Therapy 1998; 5: 291–292.

Gene Therapy