J. Phys. Chem. B 2010, 114, 7229–7244
7229
On the Ability of PAMAM Dendrimers and Dendrimer/DNA Aggregates To Penetrate POPC Model Biomembranes Marie-Louise Ainalem,*,† Richard A. Campbell,‡ Syma Khalid,§ Richard J. Gillams,§ Adrian R. Rennie,| and Tommy Nylander† Physical Chemistry, Center for Chemistry and Chemical Engineering, Lund UniVersity, S-221 00 Lund, Sweden, Institut Laue-LangeVin, 6 rue Jules Horowitz, BP 156, 38042 Grenoble Cedex 9, France, School of Chemistry, UniVersity of Southampton, UniVersity Road, Southampton SO17 1BJ, United Kingdom, and Department of Physics, Uppsala UniVersity, Box 530, S-751 21 Uppsala, Sweden ReceiVed: December 18, 2009; ReVised Manuscript ReceiVed: March 30, 2010
Poly(amido amine) (PAMAM) dendrimers have previously been shown, as cationic condensing agents of DNA, to have high potential for nonviral gene delivery. This study addresses two key issues for gene delivery: the interaction of the biomembrane with (i) the condensing agent (the cationic PAMAM dendrimer) and (ii) the corresponding dendrimer/DNA aggregate. Using in situ null ellipsometry and neutron reflection, parallel experiments were carried out involving dendrimers of generations 2 (G2), 4 (G4), and 6 (G6). The study demonstrates that free dendrimers of all three generations were able to traverse supported palmitoyloleoylphosphatidylcholine (POPC) bilayers deposited on silica surfaces. The model biomembranes were elevated from the solid surfaces upon dendrimer penetration, which offers a promising new way to generate more realistic model biomembranes where the contact with the supporting surface is reduced and where aqueous cavities are present beneath the bilayer. The largest dendrimer (G6) induced partial bilayer destruction directly upon penetration, whereas the smaller dendrimers (G2 and G4) leave the bilayer intact, so we propose that lower generation dendrimers have greater potential as transfection mediators. In addition to the experimental observations, coarse-grained simulations on the interaction between generation 3 (G3) dendrimers and POPC bilayers were performed in the absence and presence of a bilayer-supporting negatively charged surface that emulates the support. The simulations demonstrate that G3 is transported across free-standing POPC bilayers by direct penetration and not by endocytosis. The penetrability was, however, reduced in the presence of a surface, indicating that the membrane transport observed experimentally was not driven solely by the surface. The experimental reflection techniques were also applied to dendrimer/DNA aggregates of charge ratio ) 0.5, and while G2/DNA and G4/DNA aggregates interact with POPC bilayers, G6/DNA displays no such interaction. These results indicate that, in contrast to free dendrimer molecules, dendrimer/DNA aggregates of low charge ratios are not able to traverse a membrane by direct penetration. Introduction A potential approach to the treatment of genetic disorders is gene therapy for which it is essential to achieve efficient delivery of DNA into cells to replace defective genes.1 One of the barriers to overcome for the delivery of a hydrophilic macromolecule like DNA is the cell membrane, in which phospholipids are the main constituent. The use of nonviral gene delivery vehicles that induce DNA condensation is one way of enabling DNA to cross membranes, by reducing both the size and the charge of the DNA. The condensing agent used in this study, and the subject of several successful gene delivery studies, is the cationic poly(amido amine) (PAMAM) dendrimer with an ethylenediamine core and a well-defined shape and size.2-5 Dendrimers are hyperbranched structures with tree-like arms, where the internal structure depends on the type/number of functional groups and the degree of branching. This is described by the generation Gn which refers to 2n+2 functional groups.6-8 * Corresponding author. E-mail:
[email protected]. Phone +46 46 222 8175. Fax +46 46 222 4413. † Lund University. ‡ Institut Laue-Langevin. § University of Southampton. | Uppsala University.
Dendrimers of higher generations adopt a spherical conformation in bulk solution while those of lower generations are flatter.9-11 Electrostatic forces largely control the interactions of PAMAM dendrimers and we have previously shown that DNA condensation using PAMAM dendrimers is a cooperative process where the condensing agent has a preference to bind to preformed dendrimer/DNA aggregates rather than free DNA molecules.12-14 This cooperativity is due to a strong attractive interaction resulting from a correlation in the positioning of the counterions.15 As a result, free DNA will coexist in solution with dendrimer-condensed DNA.16 To understand the membrane penetration ability of DNA using amino-terminated PAMAM dendrimers, phospholipid bilayers have been used as model biomembranes in several studies. Klajnert et al. found using differential scanning calorimetry (DSC) that G3 preferentially interacts with the headgroups of zwitterionic lipids,17 and a similar study by Gardikis et al. showed that G4 also incorporates in the hydrophobic chain regions.18 Zhang and Smith have reported that vesicles containing the zwitterionic lipid phosphatidylethanolamine (PE) more effectively are disrupted in the presence of dendrimers if the vesicles also contain palmitoyloleoylglycerophosphate (POPA) and thereby have a net negative charge. This was proposed to
10.1021/jp9119809 2010 American Chemical Society Published on Web 05/11/2010
7230
J. Phys. Chem. B, Vol. 114, No. 21, 2010
be due to the electrostatic attractive interaction present for the charged system which induced bending of the membrane.19 Karoonuthaisiri et al. showed further that the ability of dendrimers to disrupt vesicles is related to how close the lipid system is to a phase transition. Vesicles with a net negative charge are for instance not destroyed if the nonbilayer-forming zwitterionic lipid, PE, is excluded.20 Atomic force microscopy (AFM) studies of model biomembranes composed of surfacedeposited bilayers have shown that G7 does not affect the gel phase (Lβ) domains but destroys fluid phase (LR) regions after initially inducing the formation of holes.21 This observation is consistent with another AFM study where G5 and G7 removed lipids from LR domains at a greater rate than from Lβ domains.22 Molecular dynamic simulations have in addition shown G3 to collapse upon binding to LR bilayers,23,24 which is in agreement with the conclusions of a coarse-grained molecular dynamic study by Lee and Larson.25 In vitro data that propose dendrimerinduced membrane holes,26 and cell uptake of dendrimers due to electrostatic (attractive) interactions and/or nonspecific fluid phase endocytosis,27 are also highly relevant for the present study. In summary, experimental and theoretical studies suggest that higher generations, with lower curvature and higher number of functional groups, traverse membranes more easily than lower generations. The mechanism and generation dependence for the transportation of dendrimers across model biomembranes needs, however, to be studied further, as dendrimers of lower generations recently have been suggested to penetrate cell membranes without the formation of membrane holes.28 In the present study specular neutron reflection from a solid-liquid interface is used to elucidate the nanostructure of complicated interfaces. The molecular distribution normal to a solid silica surface is established with Ångstro¨m resolution before and after G2, G4, and G6 dendrimers are exposed to model biomembranes composed of surface-supported single bilayers. The bilayers are composed of zwitterionic LR palmitoyloleoylphosphatidylcholine where one of the acyl chains, the palmitoyl chain, is deuterated (POPC-d31). In situ null ellipsometry, which is based on determining the changes in polarization of light upon reflection from a surface, has been used in addition to determine the kinetics of changes in surface excess and thickness of surface-deposited POPC bilayers in the presence of G2, G4, and G6 dendrimers, respectively. Coarsegrained simulations were performed to investigate the interaction between a POPC bilayer and G3 dendrimers in the presence and absence of a bilayer-supporting surface. These calculations have allowed us to gain further insight in the penetration process such as the effect of the negatively charged silica surface used in the experiments. Another aim of this study, of relevance to nonviral gene delivery, was to gain a better understanding of the interaction between model biomembranes and dendrimer-condensed DNA aggregates, using linearized plasmid DNA (4331 basepairs; bp).2-5 To the knowledge of the authors, very few studies have specifically addressed the interaction between dendrimer condensed DNA aggregates and phospholipid bilayers. Zhang and Smith have shown that G7/DNA aggregates of high charge ratios (rcharge > 1), defined as the ratio between the number of primary amine groups on the dendrimer surface and the number of phosphate groups on the DNA backbone (NH3+/PO4-), induce leakage from negatively charged vesicles.19 Although the highest transfection ability has been reported for high dendrimer generations of high rcharge, these are also the systems where the highest cytotoxicity has been reported.7,8 As a result, we have restricted our study to the membrane interaction using rcharge
