Development of lipopolyplexes for gene delivery: A comparison of the effects of differing modes of targeting peptide display on the structure and transfection activities of lipopolyplexes

The design, synthesis and formulation of non‐viral gene delivery vectors is an area of renewed research interest. Amongst the most efficient non‐viral gene delivery systems are lipopolyplexes, in which cationic peptides are co‐formulated with plasmid DNA and lipids. One advantage of lipopolyplex vectors is that they have the potential to be targeted to specific cell types by attaching peptide targeting ligands on the surface, thus increasing both the transfection efficiency and selectivity for disease targets such as cancer cells. In this paper, we have investigated two different modes of displaying cell‐specific peptide targeting ligands at the surface of lipopolyplexes. Lipopolyplexes formulated with bimodal peptides, with both receptor binding and DNA condensing sequences, were compared with lipopolyplexes with the peptide targeting ligand directly conjugated to one of the lipids. Three EGFR targeting peptide sequences were studied, together with a range of lipid formulations and maleimide lipid structures. The biophysical properties of the lipopolyplexes and their transfection efficiencies in a basal‐like breast cancer cell line were investigated using plasmid DNA bearing genes for the expression of firefly luciferase and green fluorescent protein. Fluorescence quenching experiments were also used to probe the macromolecular organisation of the peptide and pDNA components of the lipopolyplexes. We demonstrated that both approaches to lipopolyplex targeting give reasonable transfection efficiencies, and the transfection efficiency of each lipopolyplex formulation is highly dependent on the sequence of the targeting peptide. To achieve maximum therapeutic efficiency, different peptide targeting sequences and lipopolyplex architectures should be investigated for each target cell type.

The design, synthesis and formulation of non-viral gene delivery vectors is an area of renewed research interest. Amongst the most efficient non-viral gene delivery systems are lipopolyplexes, in which cationic peptides are co-formulated with plasmid DNA and lipids. One advantage of lipopolyplex vectors is that they have the potential to be targeted to specific cell types by attaching peptide targeting ligands on the surface, thus increasing both the transfection efficiency and selectivity for disease targets such as cancer cells. In this paper, we have investigated two different modes of displaying cell-specific peptide targeting ligands at the surface of lipopolyplexes.
Lipopolyplexes formulated with bimodal peptides, with both receptor binding and DNA condensing sequences, were compared with lipopolyplexes with the peptide targeting ligand directly conjugated to one of the lipids. Three EGFR targeting peptide sequences were studied, together with a range of lipid formulations and maleimide lipid structures. The biophysical properties of the lipopolyplexes and their transfection efficiencies in a basal-like breast cancer cell line were investigated using plasmid DNA bearing genes for the expression of firefly luciferase and green fluorescent protein.
Fluorescence quenching experiments were also used to probe the macromolecular organisation of the peptide and pDNA components of the lipopolyplexes. We demonstrated that both approaches to lipopolyplex targeting give reasonable transfection efficiencies, and the transfection efficiency of each lipopolyplex formulation is highly dependent on the sequence of the targeting peptide. To achieve maximum therapeutic efficiency, different peptide targeting sequences and lipopolyplex architectures should be investigated for each target cell type.

| INTRODUCTION
The delivery of oligonucleotide or genetic material to specific cells has been a long-term goal for treatment of intractable diseases such as cancer, cystic fibrosis, retinal disorders, and cardiovascular disease. A range of potential gene delivery systems, both viral and non-viral, have been used for the delivery of pDNA, siRNA, mRNA, and miRNA. 1,2 Viral gene delivery systems have high efficiencies and generally show good transfection properties in vitro and in vivo, with several in clinical trials, and one (Glybera) recently approved in the EU. 3 However, in recent years, concerns about the potential safety of such approaches have led to a renewed interest in non-viral gene delivery vectors, as these have lower immunogenicity, have the potential to deliver large payloads, and can be functionalised to target specific cell types. A range of nanoparticle-based systems have been developed for gene delivery, 2 with the most common vectors being those based on cationic lipids (lipoplexes), cationic polymers (polyplexes), or a combination of cationic lipids and cationic polymers (lipopolyplexes). 1 Despite the advantages of non-viral gene delivery vectors, in the past they have been slow to progress to clinical use due to their generally lower efficiency of gene delivery. Recent understanding of the barriers to efficient non-viral vector delivery, such as nanoparticle instability in vivo, poor targeting to specific cells, and inefficient transport through biological barriers such as the cell membrane, has led to an increased number of candidate vectors currently in clinical trials. 2 However, further improvements in these areas are still needed to realise the potential of gene-based therapies, in particular in the treatment of cancers, where approaches such as suicide gene therapy, 4 regulation of gene expression by delivery of miRNA, 5 p53 replacement gene therapy, 6 and redirection of T-cell specificity towards cancer cells 7 have recently shown promise. Targeting of nanoparticles to tumors can be passive or active. Nanoparticles of 100 to 200 nm in diameter tend to accumulate in tumours, through a combination of leaky tumor endothelium and ineffective lymphatic drainage, a phenomenon known as the enhanced permeability and retention effect (EPR). 8 Whilst this passive targeting to tumors is undoubtedly important, inter-and intra-tumoral heterogeneity of the tumor microenvironment means that the EPR effect may be more pronounced in some tumors than others. 9 Moreover, it is also clear that accumulation in tumors is necessary, but accumulation on its own is not sufficient for cellular uptake. 10 This frequently needs to be enhanced by the presence of cell-specific targeting ligands on the surface of the nanoparticles. Such targeting ligands both enhance the selectivity of nanoparticles for cancer cells and may also trigger internalisation via mechanisms such as receptor-mediated endocytosis. For example, exploiting the fact that epidermal growth factor receptor (EGFR) is over-expressed on the surface of many cancer cell types, such as basal-like breast cancer cells, has been very effective in targeting nanoparticles to such tumors. 11 A range of preclinical studies have now demonstrated that active targeting improves the efficacy of nanoparticle-based therapies for several cancer types, 12,13 and several targeted nanoparticle therapies are now in clinical trials. 13 In addition, liposome-based delivery systems and other nanoparticles that are shielded from the reticuloendothelial system with a surface coating of poly (ethyleneglycol) (PEG) or n-ethylene glycol (n-EG) have a significantly longer half-life in vivo, allowing more of the nanoparticles to localise to the tumor. 14 Lipopolyplex gene delivery systems combine the desirable features of lipoplexes and polyplexes with high in vivo transfection efficiencies and a nanoscale size (100-200 nm). They are self-assembling nanoparticles which can be formulated from a wide range of components, enabling them to be tailored to many different applications and have multiple functionalities (reviewed in Rezaee et al 15 ).
For example: formulation of LPD nanoparticles using cationic lipids and peptide sequences derived from protamine or histone resulted in enhancement of cell transfection in vitro 16 ; in early work, RGDtargeted LPD revealed a 30-fold increase in cell transfection compared with the use of naked DNA 17 ; lipopolyplexes incorporating a fusion protein consisting of the carboxy-terminal domain of histone H1 and a nuclear localization signal gave transfection efficiencies up to 20-fold higher than lipofectin/DNA complexes. 18 We have previously developed targeted, environmentally responsive lipid:peptide:DNA (LPD) lipopolyplexes for gene delivery. These lipopolyplex formulations contain a bimodal peptide with a cationic sequence to bind and condense pDNA, 19 a linker sequence (RVRR) which can be cleaved by enzymes within the endosome, and a targeting sequence. 20 The formulation of the lipopolyplexes also includes cationic lipids such as DOTMA, and the helper lipid 1,2-dioleoyl-snglycero-3-phosphoethanolamine (DOPE). The latter is believed to mediate release of the nanoparticle components from the endosome by fusion to the endosomal membrane and perturbing the structure to a non-lamellar H II phase. 21 In the original paper describing the LPD vector 22 and in more recent work 23 the order of mixing of the lipid, peptide, and plasmid DNA were studied in detail, and it was shown that this order of mixing was crucial to ensure high transfection efficiencies. We have previously 24 used a combination of FCS, freeze-fracture electron microscopy, and fluorescence quenching experiments to prove the stoichiometry of the complex and demonstrate that the DNA is tightly condensed to the peptide in an inner core, which is surrounded by a disordered lipid layer, from which the integrin-targeting sequence of the peptide partially protrudes, mediating internalization through receptor-mediated endocytosis. Indeed, it has been shown by several groups (most recently Munye et al 25 ) that cationic peptides more efficiently condense and package DNA than cationic liposomes, a phenomenon attributed to the higher charge density of the peptide molecules. 26 We have recently developed lipopolyplexes which are sterically shielded by a shallow but even coverage of n-EG conferred by incorporating novel cationic lipids with short n-EG at the headgroup (n = 2-6) 20 and have studied the cellular uptake of the lipopolyplexes and the intracellular distribution of the components by confocal microscopy. We have also shown that liposomes formulated including these n-EG lipids form nanoparticles that are shielded with a shallow, homogeneous n-EG layer, and that these have much better cellular uptake than liposomes formulated with 1,2-distearoyl-snglycero-3-phosphoethanolamine-N-[carboxy (polyethyleneglycol) 2000 ] (DSPE-PEG2000). 27 We have recently used this approach to formulate lipopolyplexes that selectively transfected tumor cells with pDNA coding for a FRET biosensor and used this to monitor EGFR inhibition by tyrosine kinase inhibitors in vivo using quantitative FRET-FLIM imaging. 28 In this work, the bifunctional peptide incorporated peptide sequences targeting EGFR, conferring tumor selectivity and active targeting on these lipopolyplexes.
For these lipopolyplexes, the tumor selectivity and transfection efficiency both depend on how well the targeting moiety is displayed at the surface of the nanoparticle. A range of approaches for mounting the targeting moiety at the surface are possible 29 : as well as the bimodal peptide approach that we have adopted in previous work, other groups have successfully conjugated targeting peptides directly to the surface of liposomes, 30 lipoplexes, 31 and polyplexes. 32 However, it is imperative to understand how subtle changes in the structures of the toolbox components can affect both the macromolecular architecture of the nanoparticles and also the selectivity, stability, and effective transfection in vivo of the resulting imaging probe.
In this paper, we have for the first time directly compared two different modes of attaching the targeting moiety to the lipopolyplex, via the bimodal peptide approach versus direct conjugation to the lipid.
We have determined the effect that these two approaches have on the macromolecular structure of the lipopolyplex and its transfection efficiency. To develop this technology, we have used pDNA that has optical readouts through the expression of either firefly luciferase 33,34 or green fluorescent protein (GFP) 35 with the aim that the lipopolyplex formulations could then be applied to other targeted gene-based therapy approaches.  (Table S1) were synthesized via solid-phase peptide synthesis using Fmoc chemistry. The synthetic procedures, purification methods, and compound characterisations are reported in the Supporting Information.

| Plasmid DNA
The lentiviral transfer vector plasmid pSEW 38 was engineered for the transient expression of firefly luciferase (5x FLuc) 34 for bioluminescence, along with the enhanced GFP (eGFP) as a marker for fluorescence-activated cell sorting (FACS) analysis. 35 Plasmid DNA was amplified in bacteria (One Shot Top10 competent cells, Invitrogen) grown overnight in LB Broth with 100 μg/mL ampicillin, following heat-shock transfection with the plasmid DNA. DNA was extracted and purified using a Qiagen Plasmid Maxi kit, according to the manufacturer's instructions, and eluted in de-ionised water. DNA concentration and purity were measured using a spectrophotometer (NanoDrop 2000, Thermofisher), and the DNA stock was stored at −20°C until use.  Table 1, at a concentration of 1 mM in chloroform. The lipid mixture was slowly evaporated under reduced pressure to form a lipid thin film and further dried under high vacuum for at least 2 hours to ensure complete removal of organic solvents.

| Lipopolyplex formulation
The thin film was then hydrated with deionised water to give liposome solutions F1 to F6 and F9 (Table 1)

| Dynamic light scattering and zeta potential
The lipopolyplexes were characterised using dynamic light scattering and zeta potential measurements. Data were obtained using a Malvern Zetasizer Nano-ZS (Malvern, UK). Aliquots of 10 μL were diluted to 500 μL in deionised water and analysed in triplicates. A representative sample of the prepared liposomal formulations and the resulting lipopolyplexes are reported in the Supporting Information (Tables S2,   S3, and S4).  The lipopolyplex samples were diluted to a lipid concentration of 30 μM, and a total volume of 1.4 mL before fluorescence spectra were recorded and acrylamide was added. 24 Collisional quenching of the fluorescence was plotted using the Stern-Volmer equation:

| Formulation of targeted lipopolyplexes
The bimodal peptide targeted lipopolyplexes were formulated using our previously published procedures. 24 A mixture of lipids was first used to produce liposomes and give formulations F7 or F8 (Table 1).

FIGURE 1
Schematic representation of (top) bimodal peptide targeted lipopolyplexes formulated with a bimodal peptide consisting of an EGFR targeting sequence, a furin cleavable RVRR linker, a DNA condensing K 16  For both types of lipopolyplex, the initial liposome solution was sonicated to obtain an average particle size below 500 nm with a zeta potential of between +23 and +43 mV (Table S3). The liposomal formulation for bimodal targeted lipopolyplexes exhibited slightly higher surface charges than the surface targeted formulation despite using only 5 mol% more charged lipids in the bimodal formulation. The zeta potential difference is probably due to the shielding effect of the maleimide bearing lipids, analogous to the manner in which large PEG groups shield the charge of cationic lipoplexes. 48 After the addition of the K 16 containing peptides (P1-P3, P7) and plasmid DNA, the average size of the samples slightly increased while the zeta potential slightly decreased to values ranging from +15 to +33 mV (Table S4)  caveolae. 49 No studies on the contribution of particle size on the internalisation pathway of lipopolyplexes has been conducted so far, but it can be assumed that the broad size distribution of the presented lipopolyplex with an average size above 250 nm may trigger different endocytosis mechanisms. However, a major contribution of receptormediated endocytosis was expected due to the presence of EGFR targeting peptides. 20

| Transfection of HCC1954 human breast cancer cell line
We initially investigated the surface targeted lipopolyplexes, which were optimised by changing different parameters such as the lipid composition or the ratio between peptide and DNA. The relative merits of using either maleimide lipid 1 or 2 to conjugate the targeting peptides (P4-P6) to the exterior of the liposome were first examined.
As described above, formulation F1 (with DODSM 1) or formulation F9 (with DODEG3SM 2), respectively, were conjugated to P4 to P6, followed by formulation into lipopolyplexes by the addition of 5 μM K 16 peptide (P7) and then 0.01 μg/μL pDNA. As a control, formulation F7 was complexed with peptide P7 and pDNA to give the non-targeted lipopolyplex F7-(P7)-REF (Table S2) Lipopolyplexes containing DODSM 1 were therefore selected for further study. As lipopolyplexes bearing the EGFR targeting LARLLT peptide P5 at the surface appeared from these preliminary studies to have slightly superior transfection efficiencies, we used this peptide sequence in studies to further optimise the composition of the lipid bilayer ( Figure 2B). The ratios of DODSM, DOPE, and DOTMA were varied as shown in Table 1 (liposome formulations F1

| Characterisation of the macromolecular structure of the lipopolyplexes
In order to understand the differences in transfection efficiencies displayed by the bimodal and surface targeted lipopolyplexes, we conducted fluorescence quenching experiments to elucidate the location of the different components within the lipopolyplex. 24 In order to study the accessibility and location of the targeting sequence itself,