Introduction

The delivery of nucleic acids across the cell membrane in gene therapy is a hot topic in biomedical research (gene therapy).1, 2, 3, 4, 5, 6, 7, 8, 9, 10 As nucleic acids alone are not able to penetrate the cell membrane, they must be transported with a suitable carrier.11 Different strategies have been developed to address this problem, ranging from physical methods like electroporation12 over viral transduction13 to various kinds of nanosystems. Among the nanosystems, dendrimers, liposomes, polymeric and inorganic nanoparticles have been established.3, 5, 6, 7, 11, 14, 15, 16, 17 Typical nanoparticles for transfection comprise gold,18, 19, 20 iron oxide,21 silica,22 carbon nanotubes23 and many different polymers.2, 24, 25

Calcium phosphate nanoparticles for transfection were first proposed by Maitra,26 based on the classical calcium phosphate transfection method by Graham and van der Eb from 1973.27 They have gained some attention because of their inherent biocompatibility, their easy preparation and loading with biomolecules, their biodegradation/dissolution after cellular uptake in the lysosome.28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42 This is an advantage compared to non-biodegradable nanoparticles, for example, gold or carbon nanotubes. We have introduced the possibility to prepare them in a multishell manner that permits to add a number of biomolecules to the same nanoparticle below a protecting outer shell,43, 44 including a subsequent covalent functionalization with antibodies.45

It has been demonstrated for a number of cell lines46 that the transfection efficiency of calcium phosphate nanoparticles was lower than that of optimized cationic liposomal agents47, 48, 49 like Lipofectamine.50, 51 However, such cationic transfection agents are usually associated with a significant degree of cytotoxicity that restricts their application in vivo.46, 52 On the other hand, almost all cell types take up nanoparticles,46 but this does not always lead to an efficient processing of the transported nucleic acid. It has also been suggested that a successful transfection depends on the cell division,53, 54 although it has recently been demonstrated that a transfection can occur without mitosis.55, 56

To elucidate the intracellular processing of nucleic acids that are taken up by cells with the help of nanoparticles or liposomes, we have carried out live-cell imaging studies and monitored the time course of transfection and gene silencing, both with nanoparticles and Lipofectamine. This powerful in situ method57 allows a time-dependent insight into transfection and gene silencing.

Results and Discussion

Triple-shell nanoparticles of calcium phosphate–nucleic acid–calcium phosphate–poly(ethyleneimine), denoted as CaP/DNA/CaP/PEI and CaP/siRNA/CaP/PEI in the following, were prepared and purified from dissolved nucleic acids, PEI and synthesis by-products. Figure 1 shows scanning electron micrographs of both kinds of particles. The particles are mostly spherical with a typical diameter of 60 nm.

Figure 1
figure 1

Scanning electron micrographs of CaP/DNA/CaP/PEI nanoparticles (left) and of CaP/siRNA/CaP/PEI nanoparticles (right).

Full size image

The particles were well dispersable in water as indicated by dynamic light scattering and nanoparticle tracking analysis. The higher hydrodynamic diameter in dispersion in comparison to the diameter of the solid core indicates a moderate degree of agglomeration. The particles carried a positive charge as shown by the zeta potential due to the external layer of PEI. Under the assumption of spherical particles, it was possible to compute the particle concentration. The analytical data of both kinds of nanoparticles are given in Table 1. Note that it is not possible to exactly determine the position of the nucleic acid within the multishell nanoparticle, but clear evidence comes from colloid–chemical data where the surface charge of nanoparticles was measured after each shell.58

Table 1 Colloid–chemical data of the nanoparticles used
Full size table

For transfection, the stock solution (Table 1) was diluted 1:7. This gives a DNA concentration of 2.8 μg ml−1 and a particle concentration of 4.8 × 1010 ml−1. Clearly, the number of nanoparticles strongly exceeded the number of cells (1.5 × 104 cm−2). The concentration of DNA in the transfection with Lipofectamine was 0.14 μg ml−1.

First, the uptake of both kinds of nanoparticles by HeLa cells was studied. For this, the particles were fluorescently labeled with PEI-FITC (fluorescein isothiocyanate). After 4 h, many cells had taken up the particles (Figure 2).

Figure 2
figure 2

Localization of CaP/DNA/CaP/PEI-FITC nanoparticles after 4 h incubation of HeLa cells. Green represents CaP/DNA/CaP/PEI-FITC nanoparticles, blue represents nuclei stained by 4,6-diamidino-2-phenylindole (DAPI) and red represents the cell membrane. Scale bar 20 μm.

Full size image

The transfection with enhanced green fluorescent protein (eGFP)-encoding DNA was carried out both with nanoparticles and with Lipofectamine and followed by live-cell imaging. The high degree of transfection is shown in Figure 3. After 72, a high number of cells was transfected with both nanoparticles and Lipofectamine.

Figure 3
figure 3

Live-cell imaging microscopy of HeLa cells during transfection with CaP/DNA/CaP/PEI nanoparticles and Lipofectamine over 72 h. The increasing number of green fluorescent cells indicates the successful transfection by both Lipofectamine and DNA-loaded nanoparticles.

Full size image

The microscopic images were manually analyzed to quantify both the transfection efficiency and the cell viability. The ratio of the number of transfected (green) cells to the total number of cells gave the transfection efficiency. As the images always showed the same part of the cell culture, we were able to assess the cell viability by counting the total number of cells in the images. As absolute cell numbers were counted, this gave a better indication of the cell viability than relative assays like the MTT test. As control, HeLa cells were cultivated under the same conditions, but without transfection agent.

Figure 4 shows the results for the transfection with Lipofectamine. The expression of eGFP started after ~4 h with a steep increase, reaching almost 100% after 24 h. The number of cells first increased but then dropped to ~50% of the initial value where it remained. In the control, the number of cells steadily increased to more than 400% of the initial value. In total, after transfection with Lipofectamine for 72 h, 50% of the original cell number are still there, and all of them are transfected.

Figure 4
figure 4

Transfection efficiency and relative number of HeLa cells during 72 h transfection with Lipofectamine. The relative number of the cells at the first time point (0 h) corresponds to 100%. The control represents untreated HeLa cells.

Full size image

The transfection with nanoparticles showed a completely different picture (Figure 5). The expression of eGFP started slightly later (5 h) but with a gentler slope, reaching ~35% after 30 h. The transfection was shorter-lived than with Lipofectamine, dropping after ~50 h. However, the number of cells increased more steadily than with Lipofectamine with a plateau at the highest percentage of transfected cells. After 72 h, the number of cells was ~250% of the original value with 20% transfected cells. If we compare this to 100% of initially present cells, this corresponds to 50% transfected cells and 200% non-transfected cells after 72 h. Interestingly, this is the same total number of transfected cells as with Lipofectamine, but accompanied by the fourfold amount of untransfected cells.

Figure 5
figure 5

Transfection efficiency and relative number of HeLa cells during 72 h transfection with CaP/DNA/CaP/PEI nanoparticles. The relative number of the cells at the first time point (0 h) corresponds to 100%. The control represents untreated HeLa cells.

Full size image

A closer inspection of the images indicated that the transfection was associated with the cell division. Many examples were found where the expression of eGFP occurred 2–3 h after cell division, both for Lipofectamine and nanoparticles (Figure 6). This was also reported for the transfection of HeLa cells with Lipofectamine55 and also by us for the transfection of HeLa cells from a nanoparticle-coated surface.59

Figure 6
figure 6

Live-cell imaging microscopy of HeLa cells during transfection with CaP/DNA/CaP/PEI nanoparticles and Lipofectamine. Black arrows indicate divided cells and white arrows indicate the visible fluorescence of the daughter cells.

Full size image

Gene silencing by siRNA occurs in the cytoplasm, in contrast to transfection with DNA that has to enter the cell nucleus. Consequently, gene silencing is typically easier than transfection. We have compared the transfection efficiency of HeLa-eGFP cells both with nanoparticles and with Lipofectamine. In this case, the intracellular biosynthesis of eGFP is blocked and the green fluorescence is gradually vanishing.

For gene silencing, the stock solution (Table 1) was diluted 1:5. This gives an siRNA concentration of 4 μg ml−1 and a particle concentration of 6.8 × 1010 ml−1. Again, the number of nanoparticles strongly exceeded the number of cells (1.4 × 104 cm−2). The concentration of siRNA during the transfection with Lipofectamine was 0.2 μg ml−1.

Figure 7 shows corresponding data recorded over 72 h. It must be noted that the HeLa-eGFP cells are continuously dividing, that is, the number of green fluorescent cells is steadily increasing in the control.

Figure 7
figure 7

Live-cell imaging microscopy of HeLa-eGFP cells during gene silencing with CaP/siRNA/CaP/PEI nanoparticles and Lipofectamine. The control represents by untreated HeLa-eGFP cells.

Full size image

Lipofectamine caused a continuous increase in the gene-silencing efficiency up to 72 h, when almost all cells were silenced (Figure 8). Gene silencing started almost immediately after incubation. The rate of cell proliferation was about half of the rate of control cells, that is, after 72 h, the cell number had increased by a factor of four with Lipofectamine and by a factor of eight in the control. The HeLa-eGFP cells were dividing about two times faster than the HeLa cells. Altogether, the silencing with Lipofectamine resulted in 400% of silenced cells, compared to the original cell population.

Figure 8
figure 8

Gene silencing efficiency and relative number of HeLa-eGFP cells during the 72 h transfection with Lipofectamine. The relative number of the cells at the first time point (0 h) corresponds to 100%. The control represents untreated HeLa-eGFP cells.

Full size image

Nanoparticles also showed a steady increase of gene silencing, starting ~5 h after incubation (Figure 9). After 72 h, the gene-silencing efficiency was ~35%. The cell proliferation was almost not constrained by the presence of nanoparticles, leading to 700% of the original cell number after 72 h. Taken together, after 72 h incubation with nanoparticles, eGFP was silenced in ~240% of the original cell number, accompanied by ~460% of non-silenced cells.

Figure 9
figure 9

Gene-silencing efficiency and relative number of HeLa-eGFP cells during the 72 h transfection with CaP/siRNA/CaP/PEI nanoparticles. The relative number of the cells at the first time point (0 h) corresponds to 100%. The control represents untreated HeLa-eGFP cells.

Full size image

If we consider the cell proliferation in more detail, two separate effects have to be distinguished. As control, we always used untreated cells. If the cell proliferation in the presence of transfection agents was lower than in the control, this would be because of two effects: first, the effect of the transfection agent (that is, calcium phosphate nanoparticle or Lipofectamine), and second, the effect of the transfection itself (that is, the introduction and intracellular processing of the nucleic acid). It is not easy to separate these effects; however, in the light of earlier data on the effects of calcium phosphate nanoparticles (only cytotoxic at high doses60, 61, 62) and Lipofectamine (a moderate degree of cytotoxicity63, 64), also reported earlier in our comprehensive study with 10 different cell lines and these two transfection agents,46 it is highly likely that the processing of the nucleic acid does not have a significant role during the presented live-cell imaging experiments.

The gene-silencing efficiency probes the intracellular degradation of the corresponding protein (here, eGFP) because it essentially prevents the biosynthesis of the protein. This is a completely different mechanism than transfection where the biosynthesis of a specific protein is turned on. It is therefore not surprising that it is slower than the transfection. Of course, this will depend on the lifetime of a specific protein inside a cell. Bartlett and Davis65 have presented a kinetic model for the underlying processes. Both transfection and gene silencing will be transient, that is, after DNA or siRNA are consumed. Bartlett and Davis66 have also shown that gene silencing lasts ~1 week in rapidly dividing cells and ~1 month in slowly dividing cells. This has been ascribed to the dilution of siRNA by cell division, demonstrating a remarkable stability of siRNA in the cell after uptake.

The transfection is obviously related to cell division (mitosis). However, it has recently been shown by Kirchenbuechler et al.55 that a transfection is also possible in non-dividing cells, but with lower efficiency and on a longer timescale. Dividing HeLa cells showed a protein expression ~2–5 h after division, whereas non-dividing cells needed much more time to show a protein expression (up to 30 h). Lipofectamine was used for transfection. Durymanov et al.56 demonstrated that the protein expression started ~5 h after transfection of Cloudman S91 melanoma cells with a polycationic formulation in post-mitotic cells and ~15 h after transfection in non-divided cells. The probability for two transfected daughter cells was especially high in post-mitotic cells.

By comparison of Lipofectamine and and another liposomal formulation (DC-Chol/DOPE) for the transfection of CHO-K1 cells, Fiume et al.54 showed that Lipofectamine is especially successful to bring the plasmid DNA into both daughter cells after mitosis. This was related to the high number of DNA copies within the cell. Protein expression started ~3 h after incubation with the transfection agents. Bishop et al.67 showed that the degradation of plasmid DNA inside a cell by nucleases is rapid and occurs with a half-life time of ~2–3 h.67 However, Cardarelli et al.68 followed the intracellular pathway of Lipofectamine and showed that the DNA remains protected after cellular uptake by avoiding lysosomal degradation. The uptake of the ‘classical’ transfection agent PEI was studied by live-cell imaging by Bausinger et al.69 with high-end optical methods, also during the individual transfection steps.

Calcium phosphate nanoparticles enter the cell by endocytosis70 and are then dissolved by acidification in the lysosome.46, 71, 72 From our results, it is clear that nanoparticles follow the same delivery pathway as liposomal formulations for transfection, but with a lower efficiency. This may be related to the endolysosomal pathway that can lead to an intracellular degradation of nucleic acids,46 which is avoided especially for Lipofectamine.68 The onset of the transfection and gene-silencing processes after incubation is comparable with Lipofectamine. The potential of nanoparticles clearly lies not in creating a transfection agent superior to Lipofectamine (which works very well) but to offer a means to transport different agents in one particle, for example, for vaccination73 or in vivo targeting.74

Conclusions

A number of conclusions can be drawn from these results. First, Lipofectamine is significantly more efficient than calcium phosphate nanoparticles, leading to almost complete transfection and gene silencing after 24–72 h. The cell viability considerably decreases in the presence of Lipofectamine, but if the aim is a population of fully transfected cells, it is clearly superior to nanoparticles. Calcium phosphate nanoparticles lead to both lower and slower transfection and gene silencing, which starts 5 h after incubation, that is, later than with Lipofectamine. This may be because of the slower uptake and endolysosomal processing of nanoparticles compared to the liposomal Lipofectamine. It must also be noted that the concentration of nucleic acid in the cell culture medium was ~10–20 times higher than with Lipofectamine. However, the cells are proliferating much more easily in the presence of nanoparticles compared to Lipofectamine. Together with the fact that the absolute number of transfected or silenced cells is comparable to Lipofectamine, this suggests a better applicability in vivo where it should lead to a reasonable degree of gene transfer together with better cytocompatibility. Calcium phosphate nanoparticles are therefore especially suitable as a tool to create multifunctional nanosystems that carry more than one biomolecule (for example, a nucleic acid together with an antigen and a targeting moiety).

Materials and methods

Synthesis of calcium phosphate nanoparticles

Calcium phosphate nanoparticles were precipitated from aqueous solutions of calcium nitrate (6.25 mM) and diammonium hydrogen phosphate (3.74 mM) and then functionalized with DNA or siRNA (1 g l−1) and PEI (25 kDa, 2 g l−1) as described earlier75 with the following modifications: CaP/DNA/CaP/PEI nanoparticles were separated from nonadsorbed molecules (including excess nucleic acid or PEI) by centrifugation (5000 g, 4 °C, 15 min) and then dispersed in the half of the original volume of water by ultrasonication (UP50H, Hielscher, Teltow, Germany; Ultrasound Technology; sonotrode 2, cycle 0.8, amplitude 80%, 10 s). CaP/siRNA/CaP/PEI nanoparticles were isolated by centrifugation (20 000 g, 4 °C, 15 min) and then dispersed in the original volume of water by ultrasonication with the same conditions.

Model DNA leading to the expression of eGFP (pcDNA3-eGFP) was used. For the gene-silencing experiments the eGFP-siRNA: sense, 5′-GCAAGCUGACCCUGAAGUUCAU-3′; antisense, 5′-AUGAACUUCAGGGUCAGCUUGC-3′ (Invitrogen, Paisley, UK) was used. Lipofectamine 2000 was obtained from Life Technologies (Invitrogen, Carlsbad, CA, USA) and used according to the manufacturer's specifications.

Characterization

All nanoparticle dispersions were characterized by scanning electron microscopy (SEM), dynamic light scattering and nanoparticle tracking analysis. SEM was carried out with an ESEM Quanta 400 FEG instrument (FEI, Eindhoven, The Netherlands; gold-palladium [80:20]-sputtered samples). Size and zeta potential of the nanoparticles were determined by dynamic light scattering with a Zetasizer nanoseries instrument (Malvern Nano-ZS, Malvern, UK; laser: λ=532 nm) using the Smoluchowski approximation and by nanoparticle tracking analysis with a NanoSight LM10 Instrument (Malvern, UK). The calcium concentration was determined by atomic absorption spectroscopy (Thermo Electron Corporation, Thermo Fisher Scientific, Waltham, MA, USA; M-Series AA spectrometer) and then used to compute the calcium phosphate concentration in the nanoparticles dispersion.

The number of particles per volume was computed by taking the mass of calcium phosphate in the dispersion and assuming spherical nanoparticles with the diameter obtained by SEM:

with N(NP) the number of nanoparticles per m3, w(CaP) the mass concentration of calcium phosphate in kg m−3 (by assuming the stoichiometry of hydroxyapatite, Ca5(PO4)3OH), r(NP) the average radius of one nanoparticle from SEM, m(NP) the mass of one nanoparticle and ρ(CaP) the density of hydroxyapatite (3140 kg m−3).

The concentration of nucleic acid in the dispersion was determined by preparing the same particles with fluorescent Alexa555-labeled siRNA and quantitative analysis by UV-VIS spectroscopy after preparation of a calibration curve.

The concentration of PEI in the dispersion was determined by preparing the same particles with fluorescent FITC-labeled PEI and quantitative analysis by UV-VIS spectroscopy after preparation of a calibration curve.

All characterization data of the two nanoparticle stock solutions are listed in Table 1.

Cell culture

For the transfection experiments, we used HeLa cells (human transformed cervix epithelial cells). For the gene-silencing experiments, HeLa-eGFP cells were used that stably expressed eGFP.43 In all experiments, HeLa cells were cultured in DMEM, supplemented with 10% of fetal bovine serum, 100 U ml−1 penicillin, and 100 μg ml−1 streptomycin at 37 °C in humidified atmosphere with 5% CO2. HeLa-eGFP cells were cultured in DMEM, supplemented with 10% of fetal bovine serum and 50 μg ml−1 of the antibiotic Geneticin at 37 °C in humidified atmosphere with 5% CO2.

Light and fluorescence microscopy were performed on a Keyence Biorevo BZ-9000 instrument (Osaka, Japan), both equipped with filters for TRITC (excitation: 540 nm, emission: 605 nm), GFP (excitation: 470 nm, emission: 535 nm) and 4,6-diamidino-2-phenylindole (excitation: 360 nm, emission: 460 nm) channels. The cells were stained with 4,6-diamidino-2-phenylindole (nucleus; blue) and Cell mask (cell membrane; red) to indicate the cellular uptake of the nanoparticles.

Transfection with Lipofectamine was carried out according to the manufacturer's instructions. Fifty-microliter cell culture medium without fetal bovine serum were mixed with 2 μl Lipofectamine. One-microgram DNA or siRNA was given to 50 μl cell culture medium without fetal bovine serum. The two solutions were mixed and incubated for 20 min at room temperature. The mixture was then given to the cell culture at the appropriate dilution (1:7 for transfection and 1:5 for gene silencing). Transfection with nanoparticles was carried out by adding the appropriate amount of nanoparticle dispersion (stock solution as described in Table 1) to the cell culture medium and adding it to the well (dilution 1:7 for transfection and 1:5 for gene silencing). The concentration of nanoparticles was selected according to earlier experiments.76

Live-cell imaging was performed with an INUG2EF-KI4 incubation chamber (TokaiHit, Keyence, Mechelen, Belgium), coupled to the Keyence microscope. The instrument was controlled by the BZ-II-Viewer software of the microscope (Keyence, Osaka, Japan). For transfection, 1.5 × 104 cells cm−2 were cultivated in each well of a four-well chamber. For gene silencing, 1.4 × 104 cells cm−2 were cultivated in each well of an eight-well chamber. After cultivation for 24 h, transfection or gene silencing was started by adding Lipofectamine or nanoparticles. Untreated cells served as control. The recording time started immediately after incubation.

Images were taken every hour with the following parameters: × 20 objective, recording time 1/20 s (FITC filter) and 1/7 s (phase contrast), four images at different points of the sample. For gene silencing, the recording time for phase contrast was 1/35 s. After 7 h, the cell culture medium was replaced by fresh medium.

The transfection efficiency was determined microscopically by counting green cells and colorless cells.46 The gene-silencing efficiency was determined in the same way by comparing green fluorescent and silenced cells.77 Four images at each time point were taken and individually analyzed for cells (at least 100 cells per image). The error bars given in the graphs indicate s.d.'s. The analyses were carried out by an independent person who was not informed before about the nature of the experiment (that is, whether the images resulted from control cells of transfected cells).

Microscopic image analysis has been shown to be an efficient method to determine transfection efficiencies, although the results typically differ from fluorescence-activated cell sorting and quantitative PCR analyses.77 This is because of the different probed features in the cell (protein fluorescence versus mRNA).