Two different cancer cell types were used. For a glioblastoma in vitro cancer model, the cell line U87 (ATCC^®, HTB-14™, derived from human) was compared with freshly isolated cortical neurons from E18 old rat embryos. F98 glioblastoma cells (ATCC^®, CRL-2397™, derived from rat) were used as folate receptor positive cancer cells for western blot analysis to characterize antibody cross reactivity for human and rat cells (Supplementary Figure S1). For a breast cancer in vitro model, the breast cancer associated cell line MCF-7 (ATCC^®, HTB-22™, derived from human) was utilized in comparison to primary human foreskin fibroblast (HFF) (ATCC^®, SCRC-1041™, derived from human).

All cell types were cultivated as indicated in Table 1 at 37°C in a humidified atmosphere containing 5% and CO₂. Cultivation of U87, MCF-7, HFF and F98 cells were performed by exchanging cell culture medium every 2 to 3 days. At a cell density of 80%–90%, cells were subcultured by incubation with 0.05% trypsin-EDTA solution (Sigma Aldrich, United States) at 37°C for 5–7 min. Isolation and cultivation of primary cortical neurons was performed as described previously (Fricke et al., 2011) and 1.2 × 10⁶ neurons were seeded on 0.01% (v/v) Poly-l-Lysin (PLL) (Sigma Aldrich, United States) coated 24-well plates (Thermo Scientific, United States). Neuronal network formation was performed under medium change every 2–3 days for 11 days until transfection. U87 (70.000 cells/cm²), MCF-7 (80.000 cells/cm²) and HFF (40.000 cells/cm²) cells were seeded on 0.01% (v/v) human fibronectin (FN) (VWR, United States) coated 24-well plates 24 h before transfection. At time of transfection, all cells were 70%–90% confluent. For western blot analyses all cell types as well as F98 cells (90.000 cells/cm²) were seeded as described above.

Immediately after live cell analysis, the flow cytometer (CytoFLEX S Flow Cytometer, Beckmann Coulter) was used to determine eGFP-pDNA transfer efficiencies for all transfections. For flow cytometric analysis, all cells were trypsinized with 250 μl 0.05% trypsin-EDTA solution (Sigma Aldrich, United States) at 37°C for 3 min and centrifuged with 300 g. To allow cell separation of neuronal networks, cells were centrifuged with 200 g for 5 min. Cell populations were gated separately based on cell granularity and size (forward (FSC) and side scatter (SSC)). At least 10.000 cells were analyzed by using appropriate filter settings to determine eGFP transfer efficiencies.

Evaluating the cell viability by flow cytometry 24 hpt the “LIVE/DEAD™ Fixable Violet stain fluorescence assay” (L34963, Invitrogen, United States) was used. In principal living cells react with the kit’s fluorescent reactive dye on their surface resulting in a weak fluorescent signal while dead cells with compromised plasma membrane incorporate the dye within the cytosol showing a bright signal. To measure all dead cells the supernatant of the substrate was involved and centrifuged before trypsinization. For precise differentiation between live and dead cells, a positive dead control (treated with 1 ml ethanol p.a. for 20 min at RT) was used. However, all cell viability stainings were performed according to the suppliers’ given protocols. During flow cytometry the cell populations were then gated separately based on cell granularity and size FSC and SSC . At least 10.000 cells were analyzed by using appropriate laser and filter settings to determine ca-Caspase3 activity.

Size- and zeta potential distribution were determined by dynamic and electrophoretic light scattering, respectively using a ZetaSizer (ZetaSizer Nano ZS from Malvern Instruments, Malvern, United Kingdom) equipped with a HeNe laser (633 nm). Scattered laser light was collected at a constant angle of 173°. Measurements were performed with both transfection reagents and their functionalized complexes for in situ long term stability. The preparation of lipoplexes for characterization of both transfection reagents and their functionalized complexes were performed as described previously. For in situ complex analyses, only the in vivo protocol was used with and without PEG-FA functionalization and prepared as shown before. The complexes were stored at 4°C after preparation and measured in an interval of 3 days for 4 weeks. For all characterization, the solutions were diluted 1/10 for size distribution and 1/100 for zeta potential measurements with purified, degassed, and sterile filtered water (Milli-Q Gradient A10, Merck Millipore, Germany). Size was determined by intensity and zeta potential measurements were performed with a refractive index of 1.33. All measurements were performed at 21°C and repeated three times at 1 min interval. Data were collected from three independently prepared samples and analyzed using the instrument software (DTS from Malvern Instruments).

Quantification of FolR1-α protein levels was performed via western blot analysis for all cell types. Therefore, cells were cultured in 12-well plates (Thermo Scientific, United States) to 80%–90% confluency. After washing with cold 1x PBS, pH 7.4 (ThermoFischer, United States), 250 μl of RIPA lysis-buffer (Sigma Aldrich, United States) supplemented with 1% protease inhibitor (v/v) (Sigma Aldrich, United States) was added to each well. Cell lysis was performed at room temperature for 20 min. Subsequently, the lysates were centrifuged at 16.000 g for 10 min at 4°C and 200 μL of each supernatant were heated to 95°C for 5 min after addition of 50 μl 5x loading dye (Bio-Rad, United States). 15 μL of each protein isolate was fractionated at 100 V for 2 h in ready-to-use 4%–20% gradient gels (Mini-Protean, Bio-Rad, United States) by SDS-PAGE and blotted to a nylon membrane. Using Anti-Tubulin Y1/2 clone (MAB 1864, Millipore, United States) as constitutively expressed marker, overall protein level of FolR1-α was determined with an anti-Folate Binding Protein/FBP antibody (ab221542, Abcam, United States). Secondary antibody Anti-Rat (A8438, Sigma Aldrich, United States) and Anti-Rabbit (A3812, Sigma Aldrich, United States) coupled to alkaline phosphatase were visualized by incubation of NBT/BCIP substrate solution (Thermo Scientific, United States) at room temperature for 5 min.

All data are given as mean including standard deviation of at least three independent measurements. Relative calculated data were generated from each experiment and combined accordingly from the different individual experiments to give a relative mean and the corresponding standard deviation. Statistical analysis (univariate ANOVA) was performed with Microsoft Excel (MS Office 2019) for multiple comparisons. Figures and graphs were created using Origin 2019 64 Bit (Origin Lab Graphing & Analysis). Only significant differences between individual probes were marked in graphs with asterisks. For this purpose, a p-value of ≤.05 was considered as significant. p-values of 0.05, 0.01 and 0.001 were labeled with one to three asterisks.

Cancer cell targeting depends on specific molecules present only in those cells. Since cancer cells are known to overexpress folate receptors (FR), we tested two different cancer cell lines for FR protein levels. Western blot analysis showed for breast cancer MCF-7 cells as well as for the glioblastoma U87 cells a clearly detectable FR signal. In contrast no signal was detectable in non-cancerous HFF cells and primary neurons which were chosen as controls (Figure 1A, B). For simplicity reasons only, cancer cells together with chosen controls were subsequently named as “breast cancer in vitro model” and “glioblastoma in vitro model”, respectively. Antibody cross-specificity between cells of different species was demonstrated and shown in Supplementary Figure S1. Since cancer cells (MCF-7 and U87) show a distinct and significant 18-fold increase in FR expressions compared to healthy primary cells (HFF and primary neurons) (Figure 1A), FA-functionalization of nucleic acid transfer systems could therefore lead to an improvement of transfection efficiencies that is supposed to be reduced by pure PEGylation in healthy and cancer cells.

Using fusogenic (Fuse-It-DNA) and endocytic lipoplexes (Lipofectamine^® 3000) (Figure 1), transfer efficiency of both reagents was analyzed in all cell types. Using Fuse-It-DNA, markedly enhanced transfer efficiencies for MCF-7 cancer cells and more biocompatible transfer in respective HFF control cells was demonstrated compared to lipofectamine^® 3000. For the glioblastoma model contrary results were found. Here, Lipofectamine^® 3000 showed higher efficiencies for U87 cancer cells and primary neurons. Both transfer systems were therefore used for further functionalization and characterized for targeting abilities in the respective in vitro model. The morphology of all cell types after transfection with Fuse-It-DNA (MCF7/HFF) and Lipofectamine^® 3000 (U87/pr. neurons) is shown in high resolution in Supplementary Figure S2.

Local in vivo delivery often depends on small transfer volumes that cannot be reached with standard transfection protocols. This is not only important for breast cancer, but even more essential for glioblastoma treatment (Hoffmann et al., 2022). We therefore additionally adapted standard cell culture in vitro protocols (Figure 2A, “in vitro”) with high transfection volumes of 50 µl per 1 µg of DNA in such a way that significantly reduced volumes of 4 µl (Lipofectamine^® 3000) and 5.5 μl (Fuse-It-DNA) could be applied for possible local applications in vivo. Although also for the adapted protocols all experiments very performed in cell culture, we termed it in the following “in vivo” for better differentiation (Figure 2A). As a result, both adapted in vivo protocols showed improved transfer efficiencies and significantly higher numbers of vital cells compared to standard protocols (in vitro) 24 h after DNA transfer.

Additional functionalization of lipoplexes of both systems with PEG-FA in concentrations between 0.4 mM and 12 mM showed different effects for the two transfection reagents. For Fuse-It-DNA, largely unaffected transfer efficiencies were observed at low and intermediate PEG-FA concentrations. Only functionalization with high 12 mM PEG-FA concentrations showed significant reductions in transfer efficiencies (Figure 2A, B). In contrast, for Lipofectamine^® 3000 already low concentrations of 0.4 mM PEG-FA resulted in a significant reduction in transfer efficiency. The effect was strongly enhanced at higher concentrations with up to 90% in U87 glioblastoma cells.

For lipoplex characterization and analysis of the zeta potential, both systems showed comparable reductions in zeta potentials upon functionalization with increasing PEG-FA concentrations (Figure 2C). Main differences between both systems were mainly an overall lower zeta potential by 15–20 mV for all Fuse-It-DNA modifications compared to the respective lipofectamine lipoplexes. The complex size of Fuse-It-DNA showed a steadily increasing trend in size distributions with increasing PEG-FA concentration, doubling in size from approximately 130 nm to complexes above 300 nm for 12 mM PEG-FA. In contrast, Lipofectamine^® 3000 complexes did not change significantly with increasing PEG-FA up to concentrations of 4 mM, whereas the size of resulting complexes with 12 mM PEG-FA significantly increased.

To analyze the specificity of PEG-FA modifications and the resulting cancer cell targeting abilities, PEG-modifications for both in vitro cancer models were used with and without the functional FA-group for NHS coupled-modifications. For Fuse-It-DNA, significantly higher transfer efficiencies by 30%–70% were achieved in MCF-7 cancer cells with 0.4 mM and 4 mM of PEG-FA functionalized complexes compared to pure PEGylation. Relative efficiencies increased continuously with increasing concentration of FA functionalization (Figure 3A, B). Moreover, up to a concentration of 4 mM PEG-FA, transfer efficiencies were comparable to those obtained in the unmodified in vivo protocol. In contrast, for healthy HFF cells, no significant improvements in transfer efficiency could be detected for FA-functionalized lipoplexes compared to pure PEGylation. Comparing transfer efficiencies with increasing PEG-FA modifications between healthy HFF cells and MCF-7 cancer cells (Figure 3C) showed significantly enhanced transfer efficiency in favor of MCF-7 cancer cells with 4 mM PEG-FA by 20%. Interestingly, higher concentrations of PEG-FA functionalization negate this effect.

For Lipofectamine^® 3000 lipoplexes, pure PEGylation massively impaired transfer efficiency for U87 glioblastoma cells by approximately 80% and for primary neurons by almost 70% already at PEG concentrations of 0.4 mM. Higher concentrations resulted in an almost complete inhibition of DNA transfer (below 10%). In contrast PEG-FA functionalization led to significantly higher transfer efficiencies in cancer cells with up to 31% eGFP positive cells (0.4 mM PEG-FA; Figure 4A). Compared to pure PEGylation, all PEG-FA modifications showed significant and more efficient transfer efficiencies in U87 glioblastoma cells with significant differences for 0.4 and 4 mM. Relative improvements of transfection efficiencies to pure PEGylation of 150%–700% could be achieved (Figure 4B).

Transfer of DNA into primary neurons showed even more significant reductions in transfer efficiencies. Here, a maximum efficiency of 2% was achieved by all modifications, whereby no significant improvement of the transfer could be detected by additional FA-functionalization (Figure 4A). In fact, there were even opposite effects, where the additional FA-modification showed reduced relative efficiencies of −20% to −40% compared to pure PEGylation (Figure 4B).

Modification of the Lipofectamine^® 3000 complexes with 0.4 mM PEG-FA significantly improvement the transfection efficiencies in favor of U87 cancer cells compared to healthy primary neurons of more than 800% compared to unmodified complexes. However, the effects of 0.4 mM PEG-FA modification still allowed efficient transfections into the FR-overexpressing cancer cells, whereas in primary neurons they only resulted in transfection efficiencies in the range of maximum 1%. The effect of favored and selective transfection for cancer cells is negated again with increasing PEG-FA concentration (4–12 mM), so that a more selective DNA transfer into cancer cells is reduced.

To verify folate receptor-dependent transfer of PEG-FA functionalized lipoplexes into cancer cells, surface binding capacity of FA-modified lipoplexes was reduced by addition of free FA to culture medium prior to transfection. As a result, FA supplementation already reduced transfer efficiencies of unmodified lipoplexes (Figure 5A, in vivo protocol) for healthy as well as cancer cells.

For PEG-FA functionalized lipoplexes, no additional reduction in transfer efficiency could be detected for healthy HFF cells upon FA supplementation (Figure 5A, B). In contrast, for MCF-7 cancer cells a significantly stronger reduction of transfer efficiencies was observed for 0.4 mM and 4 mM of PEG-FA modifications after reduction of the surface binding capacity of FA conjugated lipoplexes. Here, the specific uptake of PEG-FA functionalized lipoplexes into cancer cells was additionally reduced by 52% (0.4 mM PEG-FA) to 37% (4 mM PEG-FA) (Figure 5B).

Identical experiments were also performed on U87 cancer cells and healthy primary neurons. As before, also here free FA supplementation already reduced transfection efficiencies. Furthermore, upon use of PEG-FA functionalized lipoplexes, FA supplementation did not lead to pronounced reductions in transfer efficiencies into primary neurons independent on PEG-FA concentration (Figure 6A, C). In contrast, for U87 cells FA supplementation led to a strong additional reduction in transfer efficiencies of up to 85% (Figure 6B). These data therefore clearly confirm the folate receptor dependent and cancer cell targeted transfer of PEG-FA functionalized lipoplexes.

Finally, to analyze the PEG-FA conjugated lipoplexes optimized and characterized here in terms of their functionality and selectivity, the transfer of DNA was analyzed in both in vitro cancer models. PEG-FA conjugated variants of Lipofectamine^® 3000 and Fuse-It-DNA were used in comparison to non-modified reagents to enhance green labeling of cancer cells and to preferentially induce apoptosis of cancer cells with ca-Caspase3 constructs (Figure 7A–D).

As shown in Figure 7A–D, the influence of pure transfection on cell viability was first examined using a transfection control (analogous to previous experiments eGFP-plasmid transfection). Here, comparable transfer efficiency ratios with analogous reductions in eGFP expression of PEG-FA functionalized complexes compared to unmodified lipoplexes were shown. By using therapeutically effective ca-Caspase3 plasmids, an obvious change in cell morphology and confluence could be determined in MCF cells. In addition, a significantly improved Caspase activity was shown by using functionalized complexes. Healthy HFF cells showed no significant changes, with constant low activity of ca-Caspase3 (Figure 7C). In contrast, in the glioblastoma model, transfer of a ca-Caspase3 either did not show changes in cell morphology and density nor significant enhanced apoptosis in cancer cells was detected (Figure 7F).

To assess the stability of NHS chemistry-induced coupling of PEG-FA to the different liposomal transfection reagents, particle size and corresponding zeta potentials were measured over 4 weeks for both systems. There was no significant change in particle sizes for the two delivery systems after PEG-FA functionalization for 3 weeks. Only after 4 weeks, slight changes in complex size after storage at 4°C could be detected (Figure 8A). The zeta potentials showed reduced values for both functionalized variants of the lipoplexes compared to non-functionalized complexes of the same reagents. The PEG-FA functionalized variants showed no significant changes in zeta potentials in the Fuse-It-DNA reagent, while Lipofectamine^® 3000 PEG-FA complexes also had significantly changed values after 3 weeks (Figure 8B).