Magnetic field-controlled gene expression in encapsulated cells

Abstract

Cell and gene therapies have an enormous range of potential applications, but as for most other therapies, dosing is a critical issue, which makes regulated gene expression a prerequisite for advanced strategies. Several inducible expression systems have been established, which mainly rely on small molecules as inducers, such as hormones or antibiotics. The application of these inducers is difficult to control and the effects on gene regulation are slow. Here we describe a novel system for induction of gene expression in encapsulated cells. This involves the modification of cells to express potential therapeutic genes under the control of a heat inducible promoter and the co-encapsulation of these cells with magnetic nanoparticles. These nanoparticles produce heat when subjected to an alternating magnetic field; the elevated temperatures in the capsules then induce gene expression. In the present study we define the parameters of such systems and provide proof-of-principle using reporter gene constructs. The fine-tuned heating of nanoparticles in the magnetic field allows regulation of gene expression from the outside over a broad range and within short time. Such a system has great potential for advancement of cell and gene therapy approaches.

Graphical abstract

Cells containing a heat inducible promoter construct (a) are encapsulated with magnetic nanoparticles (b+c). An alternating magnetic field produces heat (d), which allows controlled gene expression in patients (e).

1. Introduction

Most human diseases are based on the absence, misexpression or deregulation of gene products. Gene therapy is an approach to transfer DNA encoding therapeutic proteins into the patient to modulate the pathologic cellular pathways. Depending on the vector, transient expression or stable integration of the constructs is achieved. Integration into the host's genome results in long term production of the therapeutic protein, although random integration can result in severe problems (reviewed in Ref. [1]). Cell therapy represents a different approach, which is based on transplantation of cells producing the therapeutic protein. The classic approach has been to use autologous cells (reviewed in Ref. [2]); however, heterologous cells would have no limitation in their availability or for their manufacturing, i.e. in contrast to autologous products, would allow an off-the-shelf product to be generated. One limitation is however, that heterologous cells activate the immune system of the patient, which rapidly destroys them. To avoid this, microencapsulation can be used, as the semipermeable membrane protects the cells from the immune response. This allows their prolonged survival in vivo, making cell therapy one of the most exciting fields of translational medicine [3]. This approach has been shown to be viable in human clinical trials and it has also been demonstrated that such encapsulated cell products can be GMP manufactured at large scale [4,5].

Cell and gene therapies promise a wide range of applications in biomedicine. Similar to small molecules, also the effects of biologicals strongly depend on the dosage. Regulated expression therefore is essential for such strategies. In the last 20 years several inducible expression systems have been established, such as the tetracycline (TetR)-inducible system (reviewed in Ref. [6]) or the progesterone receptor/mifepristone (RU486)-inducible system [7]. These induction systems act via small activator molecules that are generally orally administered. However, the slow pharmacokinetics of these activators strongly limit the regulation of such systems in vivo. In contrast, one component systems use endogenous activation pathways and transcription factors. Most successful within this group are promoters reacting to the heat shock response.

The heat shock response represents the most important stress survival pathway of the cell. After exposure to different kinds of stress, like heat, heavy metals or radiation, heat shock factor 1 (HSF1), a key mediator of the pathway, is activated in the cytoplasm. It translocates to the nucleus where it binds to the heat shock elements (HSE), specific DNA motifs in the promoters of stress pathway-related genes (reviewed in Ref. [8]). Most of these genes encode heat shock proteins (HSP), which are acting as chaperones to prevent aggregation of denatured or partially unfolded proteins. Heat shock promoter regions show a complex architecture and integrate inputs from several different cellular pathways, limiting their application in therapeutic approaches. Consequently, modified natural promoters have been established (reviewed in Ref. [9]), showing low background activity and high inducibility. Activation is achieved by elevated temperatures, which can be provoked by external manipulation. Local hyperthermia, for example, can be induced by magnetic nanoparticles (NP) exposed to an alternating magnetic field (AMF). The magnetic cores of these NPs have a size in the nanometer range and are mainly composed of magnetite (Fe₃O₄), maghemite (γ-Fe₂O₃) or Co/Mn iron combinations [10]. By applying an AMF, a constant flow of energy is established in NPs, which is then transferred into thermal energy (reviewed in Ref. [11]). Biocompatibility of iron oxide NPs is largely affected by the coating [12–14], allowing different biomedical applications such as cancer treatment [15], gene transfer by magnetofection [16,17], targeted drug delivery [18] or their use in magnetic resonance imaging and related diagnostic techniques [19]. However, iron oxide NPs can also exhibit harmful properties and therefore surface coating, cellular targeting, and local exposure have to be considered for clinical applications [20].

In this work, we combined NP-mediated hyperthermia with a cell therapy approach. By co-encapsulating cells and NPs, the AMF-induced heat generation is restricted to the encapsulated cells, which contain the heat-inducible gene expression construct. With this concept we demonstrate highly regulated expression of two reporter genes in cells co-encapsulated with magnetic NPs in response to AMF treatment.

2. Materials and methods

2.1. Plasmids

The vector pSGH2luc [21] was used for expression of the reporter genes firefly luciferase (luc) and green fluorescent protein (GFP). Expression was driven by an artificial heat shock-inducible promoter composed of a core of eight idealised HSEs flanked by two minimal cytomegalovirus (CMV) promoter elements. To facilitate establishment of stably transfected cell clones, a puromycin resistance gene cassette was introduced into the construct (pSGH2luc puro).

2.2. Cell culture

For generation of a stable cell line, HEK293 cells were transfected with the pSGH2luc puro construct using polyethyleneimine (PEI) as a transfection reagent. Stable cell clones were selected with 1 μg/ml puromycin. HEK293 cells constitutively expressing luc or GFP were generated by transfection with CMV promoter-driven expression vectors pCMVluc and pCMVGFP [22]. All cells were grown in Dulbecco's modified Eagle's medium (DMEM), supplemented with 10% fetal calf serum (FCS), 100 U/ml penicillin and 100 U/ml streptomycin at 37 °C and 5% CO₂.

2.3. Cell viability assays

The effect of heat treatment on cell survival was determined applying a trypan blue assay. Briefly, 24 hours after heat treatment, detached cells contained in the culture medium as well as the trypsinated cell layer were collected by centrifugation (5 min and 116g) and resuspended in 1 ml PBS. 50 μl of the cell suspension was mixed with an equal volume of a 0.5% trypan blue solution and incubated for 2 min at room temperature. The percentage of dead cells was determined microscopically by counting the number of blue cells compared to the number of total cells using a haemocytometer.

For the analysis of the metabolic activity of encapsulated cells, an AlamarBlue assay (AbD Serotec) was performed. Metabolically active cells are able to convert the AlamarBlue substrate into resorufin, which can be quantified fluorometrically. Therefore, a defined number of capsules (n = 10) was pipetted in triplicates into a black 96-well plate. Wells containing the respective cell culture medium served as a sample blank. AlamarBlue reagent was pipetted into all samples and blanks. Plates were then incubated in a cell culture incubator at 37 °C (5% CO₂, 95% relative humidity) for 4 hours and subsequently analysed using a fluorometric plate reader (Tecan Genios™, excitation 520 nm and emission 590 nm). Different dilutions (37.5 μM, 12.5 μM, 4.17 μM 1.39 μM, 0.463 μM and 0.154 μM) of resurofin were used as positive control. The number of viable encapsulated cells was calculated according to a standard curve prepared with a given amount of non-encapsulated cells.

2.4. Heat induction

A defined number of cells was seeded into cell culture dishes and cultivated for 3 days at 37 °C. For induction of gene expression by heat treatment, cells were transferred to a cell culture incubator pre-adjusted to 43 °C (except experiments in Fig. 2 where the incubator was adjusted to 41 °C, 42 °C, 43 °C or 44 °C). To ensure an optimal temperature transfer, the cell culture plates were placed on metal plates kept in the incubator. After heat treatment for 0.5–2 hours, cells were transferred back to 37 °C and incubated further (up to 6 hours) to allow recovery.

Fig. 2.

Determination of optimal heat shock conditions for HSE promoter-driven expression. (a) The HEK293-C5 stable cell line was incubated at 41–44 °C for 1 hour to induce a heat shock response and luciferase activity was determined 6 hours after heat treatment. The luciferase activity levels were normalised to the values measured in the cells incubated at 37 °C. (b) To analyse cell viability, cells were prepared and heat-treated as described in (a). 24 hours after heat treatment, cells were stained with trypan blue and the percentage of viable cells was determined. Both experiments were performed in sextuplets; error bars: mean ± SEM, n = 6.

2.5. Reporter protein assay

To determine the activity of the expressed luciferase, cells or capsules were lysed in 50–100 μl luciferase lysis buffer (0.1 M Tris pH 7.5, 1% Triton X and 1 mM DTT); capsules were mechanically destroyed using a pestle and all samples were incubated for 15 min at room temperature. Samples were centrifuged for 5 min at 94g and all of the cleared cell lysate was used for a luciferase activity measurement in a LUMAT LB luminometer (Berthold).

GFP expression was analysed by flow cytometry (FACS Calibur, BD Biosciences). A proprietary method was used to dissolve the capsules. For this purpose 140 capsules at a time were mixed with 5× weight/volume of dissolving solution and shaken at 37 degrees for 1 hour. Then cell culture medium was added and the cells were plated out. The analysis by quantitative flow cytometry is described in Ref. [23]. GFP levels were determined by combining the number of gated cells with their mean fluorescence intensity. Relative induction was then calculated by comparing the GFP levels of induced versus non-induced samples.

2.6. Quantitative real-time PCR

To analyse reporter gene mRNA expression levels after heat stress, cells were incubated for different time spans at 43 °C and lysed either directly or after up to 48 hours of recovery at 37 °C. Total RNA was isolated according to the manufacturer's protocol using an RNA extraction Kit (Invisorb). Residual DNA was removed with DNAse I (Fermentas) and the mRNA was transcribed into cDNA (RevertAid™ H Minus First Strand cDNA Synthesis Kit, Fermentas) using random hexamer primers according to the manufactures protocol. For qPCR, Taqman probes were designed using Primer Express V2 (Applied Biosystems) and the cDNA was amplified in an Mx3000P (Stratagene) qPCR cycler. The human glyceraldehyde 3-phosphate dehydrogenase (GAPDH) gene was used as an endogenous control. The following oligonucleotide primers and probes were used: GAPDH forward: 5′-GGA AGG TGA AGG TCG GAG TCA A-3′, reverse: 5′-ACC AGA GTT AAA AGC AGC CCT G-3′, probe: 5′-HEX-ATT TGG TCG TAT TGG GCG CCT GGT C-BHQ1-3′; luciferase forward: 5′-TGG ATT ACG TCG CCA GTC AAG-3′, reverse: 5′-TTC GGT ACT TCG TCC ACA AAC A-3′, probe: 5′-Fam-CGC GAA AAG TTG CGC GGA GG-BHQ1-3′; Hsp72 forward: 5′-AAC CAG GTG GCG CTG AAC-3′, reverse: 5′-TGG AAA GGC CAG TGC TTC AT-3′, probe: 5′-Fam-AAC ACC GTG TTT GAC GCG AAG CG-BHQ1-3′; for GFP primers and probes see Ref. [24]. The expression levels of luciferase, GFP or Hsp72 were normalised to the expression levels of GAPDH.

2.7. Magnetic NPs

13 Different types of iron oxide NPs were tested for the efficiency of heat production. Additionally one commercially available formulation from Sigma-Aldrich (Iron(II,III)oxide nanopowder 98+%, Cat. No. 637106-25G) was used. All NPs were composed of a magnetic iron oxide core of different diameters (Supplementary Table S1). The particles S1, S7, and S8 with a core diameter from 30 to 80 nm were synthesised using oxidative hydrolysis method according to the procedure of Sugimoto and Matiievic [25] based on the precipitation of iron (II) salts in basic media and in the presence of nitrate ions as a mild oxidant, followed by addition of the coating component (palmitoyl dextran or branched polyethyleneimine 25 kDa). The particles with smaller average core diameter of 4 to 13 nm were synthesised by precipitation of Fe(II)/Fe(III) hydroxide by transformation into magnetite in an oxygen-free atmosphere with spontaneous adsorption of shell components as described elsewhere [16,26,27]. Composition of the particle coatings is given in Table S1.

2.8. Magnetic field induction

The magnetic field generator consisted of a power supply, a frequency generator and a power amplifier generating up to 27 A at 60 kHz. For measurements, an oscilloscope and a frequency counter were connected. The strength of the alternating magnetic field (AMF) generated in the middle of the induction coil was measured at 36 kA/m (at 27 A and 60 kHz). The induction coil was built from brass tubes permanently cooled using deionised water (20 °C). For the experiments without capsules, a similar coil with an additional water jacket around the reaction tube was used to further reduce heating of the samples (resulting magnetic field strength 33 kA/m at 27 A and 60 kHz).

To determine the activation of gene expression in response to AMF treatment, 10⁵ cells mixed with 0.5% (w/v) magnetic NPs or 140 capsules containing approximately 3.5 × 10⁵cells (≈ 2500 cells/capsule) either co-encapsulated with 0.5% magnetic NPs S8 or containing no NPs, were transferred with 200 μl of culture medium into a reaction tube (Eppendorf), pre-incubated at 37 °C and then placed into the induction coil for 30 min. Immediately after treatment, cells or capsules were removed from the coil, supplied with 2 ml medium and transferred to 37 °C for recovery. Luciferase activity was determined 6 hours after end of induction.

2.9. Encapsulation

For encapsulation, the IE-50R device from Inotech was employed, facilitating vibration-induced drop formation of a viscous polymer solution in a laminar jet. Cells and NPs were encapsulated using the biologically inert polyanion sodium cellulose sulphate (SCS, Fraunhofer, Potsdam) as a matrix. Capsule formation was enabled by gelating SCS in a solution of polycationic poly-diallyl dimethyl ammonium chloride (pDADMAC, Kaptol Chemie) according to the procedure initially described by Dautzenberg et al. [28]. For standard encapsulation experiments, 10⁶ cells/ml in phosphate-buffered saline (PBS) were mixed with 1.8% SCS. Co-encapsulation of cells with 0.5% magnetic NPs S8 was performed with 1.6% SCS. For gel formation, a solution of 1.3% pDADMAC (24 kDa) was used.

2.10. Histology

In order to analyse effects of magnetic field treatment on cell viability and integrity, encapsulated cells were fixed in pre-chilled (4 °C) 2% formalin in PBS for 1 hour at room temperature (RT). Subsequently, capsules were embedded in HistoGel (Richard-Allan Scientific) according to the manufacturer's instructions, with the exception that samples were allowed to cool for 3 hours at 4 °C. Then samples were dehydrated over night by incubation in an ascending series of ethanol using an automatic device (Thermo Scientific). The next day, samples were embedded in paraffin. Paraffin blocks were sectioned in 3 μm thick sections using a microtome (RM2235, Leica).

For further evaluation, sections were deparaffinised and rehydrated and either stained with hematoxylin/eosin or subjected to a TUNEL assay (ApoTag^R Red In Situ, Chemicon) according to manufacturer's instructions to determine the amount of apoptotic cells. Briefly, sections were incubated with digoxygenin-labelled nucleotides, which were transferred onto free 3′OH-termini of DNA present in apoptotic nuclei by a TdT-enzyme and subsequently incorporated digoxigenin-labelled nucleotides were detected by rhodamin-labelled antibodies. Finally, nuclei of encapsulated cells were counterstained with DAPI. For microscopic examination, samples were dehydrated by incubation in an ascending series of ethanol followed by incubation in xylol, finally mounted with DPX resin and covered with a cover slip.

For transmission electron microscopy (TEM), samples were fixed with 3% glutaraldehyde for at least 2 days. Subsequently, samples were washed three times with Soerensen buffer (pH = 7.4, 0.1 MNa₂HPO₄ and KH₂PO₄) for 15 min followed by incubation in 1% osmium in Soerensen buffer for 2 hours. After washing, encapsulated cells were dehydrated in an ascending ethanol series, i.e., in 30% for 5 min, 50% for 5 min, 70% for 60 min, twice in 80% 15 min each, twice in 96% ethanol for 15 min each, and twice in 96% (analysis grade) for 20 min each. Pure propylene oxide was added to the samples for 10 min twice and then a mixture of propylene and resin 1:1 was added for 60 min. Subsequently samples were put into a mixture of propylene and resin at a concentration ratio of 1:3 over night. The next day, samples were incubated in freshly prepared resin for 2 hours. Samples were allowed to polymerise by incubation at 60 °C for 3 days. Polymerised samples were cut using an ultratome (Ultracut S, Richard) into 100 nm thin sections, which were fixed on a copper grid (Plano). Samples were stained with uranyl acetate (2% in 80% methanol) for 8 min, washed three times for 10 s in ddH₂O, incubated in lead citrate for 5.5 min and subsequently washed three times for 10 s in ddH₂O.

3. Results

3.1. Experimental concept

As outlined in Fig. 1, the intention of our experiments was to combine encapsulation of cells with an efficient induction system. To enable full control from the outside, a heat shock-inducible system (Fig. 1a) was selected to regulate expression of reporter genes in genetically modified cells. After encapsulation of these cells together with magnetic NPs (Fig. 1b, c), heat can be induced within the capsules by applying an external AMF (Fig. 1d). Proof-of-principle is provided by successful activation of reporter gene expression in encapsulated cells. In possible clinical applications, capsules with cells containing therapeutic genes may be transplanted into human tissue and similarly activated by an external magnetic field (Fig. 1e). As a result, regulated expression of the therapeutic gene would be possible, according to the specific requirements of the therapy.

Fig. 1.

Schematic representation of the suggested therapeutic concept. A human cell line is generated containing the artificial, inducible heat shock promoter construct (a). Cells are then encapsulated together with magnetic nanoparticles (NP) (b, c). The permeable capsule membrane protects encapsulated cells from being attacked by the host's immune system (c). Application of an alternating magnetic field (AMF) excites co-encapsulated NPs resulting in elevated temperatures within the capsules, thereby inducing expression in encapsulated cells (d). Consequently, expression of a therapeutic gene within implanted capsules can be regulated from the outside, a concept which might be used as a new therapeutic strategy for cell therapy in humans (e).

3.2. Characterisation of the heat-inducible cell line

In a first step, a stable cell line was generated harbouring a heat-inducible promoter construct. For generation of the vector construct, a bidirectional artificial heat shock promoter containing multimerised heat shock elements (HSE) flanked by minimal promoter sequences was used. This promoter has been shown previously to be highly inducible with an extremely low background activity [21]. The bidirectional activity of the promoter was applied to simultaneously express luciferase and GFP. The construct was stably integrated into human embryonic kidney (HEK) 293 cells. A cell clone (C5) was selected, which showed the lowest background activity combined with a high inducibility of the artificial promoter. To test its inducibility in more detail, the cell clone was incubated for 1 hour at different temperatures and luciferase activity was measured 6 hours later (Fig. 2a). At mild heat shock conditions (41–42 °C) low levels of luciferase expression were detected, while a more than 2800-fold induction of luciferase expression was shown after incubation at 43 °C. The level of induction further increased to ~ 7000-fold for 44 °C (Fig. 2a). In parallel to luciferase measurements also survival rates of heat-treated cells were determined. At lower heat shock temperatures, i.e., 41 °C and 42 °C, cell viability (~ 95% living cells) did not significantly differ from the cells incubated at 37 °C. The percentage of living cells however decreased to 93.5% in samples incubated for 1 hour at 43 °C and dropped further to 91.5% after incubation at 44 °C (Fig. 2b). Consequently, 43 °C was chosen for the following heat shock experiments as an optimal compromise between high promoter inducibility and good survival rates.

For the application in regulated gene expression approaches, the kinetics of this inducible promoter in response to heat treatment was determined. Therefore, cells were incubated at 43 °C for different time spans (0.5, 1 and 2 hours) and protein and RNA levels were investigated at several time points (0.5 to 48 hours) thereafter (Fig. 3). Luciferase activity steadily increased within 2 hours, reaching a peak level between 4 and 6 hours depending on the duration of the exerted stress (reduced durations of heat treatment resulted in a shift of peak activity to earlier time points). Thereafter, luciferase activity continuously decreased to almost basal levels after 48 hours (Fig. 3a). Promoter kinetics in addition were analysed at the level of mRNA to exclude effects of protein stability (Fig. 3b). In cells that had been incubated for 30 min at 43 °C, maximum luciferase mRNA levels were detected 2 hours after heat treatment. After prolonged incubation (1–2 hours), luciferase mRNA revealed peak levels after 4 hours (Fig. 3b). The kinetics of luciferase mRNA did however not closely resemble that of luciferase activity, in particular effects of the various heat exposure times were more pronounced for the mRNA. To see whether these differences are based on effects of heat exposure at the protein level, we also determined the kinetics of the GFP mRNA, which appeared to be more stable, resulting in extended high levels up to 12 hours after heat treatment (Fig. 3c). The GFP mRNA kinetics closely resembled that of luciferase activity, indicating that the various exposure times did not differentially affect translation or protein stability of luciferase protein. These data clearly demonstrated functionality of the artificial bidirectional heat-inducible promoter and a regulation at the transcriptional level.

Fig. 3.

Kinetics of the heat shock response in HEK293-C5 cells. HEK293-C5 cells were incubated at 43 °C to induce a heat shock response and protein activity (a) or mRNA levels (b–d) were measured at different time points (0.5–48 hours) after treatment. (a) Luciferase activity was measured and normalised to the corresponding levels (set to 1) of cells incubated at 37 °C. These basal levels were only slightly above the background light units of the luminometer, indicating an extreme low background activity of the promoter. (b–d) mRNA was isolated, transcribed into cDNA and specific amounts of luciferase-encoding mRNA (b), GFP-encoding mRNA (c), or endogenous Hsp72-encoding RNA (d) were determined by quantitative PCR. The mRNA levels were normalised to the internal reference GAPDH and to the corresponding mRNA levels of cells incubated at 37 °C (set to 1). Error bars: mean ± SEM, n = 5.

To compare the expression kinetics of the artificial promoter to the natural heat shock response, endogenous Hsp72 expression was investigated (Fig. 3d). Hsp72 mRNA was activated at lower levels, but with similar kinetics as the reporter gene mRNA was initiated by the artificial promoter. Interestingly, between 6 and 12 hours the Hsp72 mRNA level dropped below that of untreated cells (37 °C control). This repression might be due to a negative feed-back regulation exerted by HSPs [29], which does not affect the artificial promoter. 24–48 hours after heat treatment Hsp72 mRNA returned to basal levels, indicating a normalisation of the cellular stress response.

In summary, the artificial promoter showed similar kinetics as the natural heat shock promoter and the peak activity of gene expression can be modulated by adjusting duration of the applied heat stress.

3.3. Regulation of expression by NPs and AMF treatment

The next step was to demonstrate magnetic NP-mediated activation of non-encapsulated HEK293-C5 cells. Preliminary experiments with NPs alone indicated an optimal heat response at an AMF of 60 kHz and 27 A. Since the induced temperature was shown to be directly linked to the concentration of NPs (Supplementary Figure S1), substantial differences in promoter activity were observed in dependence of the concentration of NPs (Fig. 4a, 0.3%–1%). Cells without NPs showed no increase in luciferase expression after magnetic field treatment, but reacted to incubation at 43 °C (black bars). Addition of 0.3% NPs enhanced heat-induced expression levels in HEK293-C5 cells and maximal luciferase activity was detected for 0.5% NPs in the AMF-treated cells (Fig. 4a, grey bars). This resulted in a more than 100-fold activation of the promoter (as compared to basal luciferase activity levels in AMF-treated cells without NPs). In the presence of 0.8%–1% NPs the luciferase levels decreased (Fig. 4a), which is most likely due to overheating resulting in a partial cell death. According to these results, a concentration of 0.5% magnetic NPs was chosen for further experiments.

Fig. 4.

AMF-induced regulated gene expression. (a) Effects of increasing nanoparticle concentrations. HEK293-C5 cells mixed with no, 0.3, 0.5, 0.8 or 1% (w/v) magnetic nanoparticles (SIGMA) were exposed to an AMF of 60 kHz and 27 A for 30 min (grey bars). As a positive control, cells were incubated for 45 min at 43 °C (black bars). Cells incubated at 37 °C were used as negative control (white bars). 6 hours after treatment, cells were lysed and the luciferase activity was measured. The graph shows the RLU of one representative experiment; the luciferase activity measured for the positive control of 0.5% nanoparticles was set 100%. Variation of magnetic field strength is shown in (b). HEK293-C5 cells mixed with 0.5% magnetic nanoparticles (SIGMA) were exposed for 30 min to an AMF of 60 kHz and 17–27 A, resulting in different field strengths. As positive control for luciferase expression (set as 100%), cells were incubated for 45 min at 43 °C (black bar). Cells incubated at 37 °C were used as negative control. 6 hours after induction, the cells were lysed and the luciferase activity was measured. Error bars: mean ± SEM, n = 5. Variation of exposure times is shown in (c). HEK293-C5 cells with 0.5% magnetic nanoparticles (SIGMA) were exposed for 1.8 min to 60 min to a magnetic field at 60 kHz and 23 A. As positive control for luciferase expression (set as 100%), cells were incubated for 45 min at 43 °C (black bar). Cells incubated at 37 °C were used as negative control. 6 hours after induction, the cells were lysed and the luciferase activity was measured and calculated as % of maximal activation at 43 °C. Error bars: mean ± SEM, n = 5.

The possibility to activate a promoter from outside a cell or even a human body is a major step towards a dosage-dependent expression of therapeutic genes. A further improvement above a simple on/off regulation would however be a fine tuned modulation of promoter activity over a wide range, enabling precise control of therapeutic protein delivery. This would allow even highly toxic proteins with a small therapeutic window to be considered for cell therapy applications. We therefore tested the effects of magnetic field strength variation by increasing the current from 17 to 27 A (Fig. 4b). The results revealed a continuous increase of promoter activity over more than two orders of magnitude in correspondence with the enhanced strength of the magnetic field.

A different option to modulate promoter activity in the established system is given by the duration of the treatment, i.e., AMF exposure. In order to evaluate this parameter, cells mixed with 0.5% NPs were subjected to an AMF of 60 kHz and 23 A for different time spans (1.8 to 60 min). A reduced current of 23 A was chosen to enable survival of the cells even at prolonged exposure times. Analysis of luciferase activity revealed a highly modulatable expression (more than three logs) in dependence of exposure time (Fig. 4c). These experiments demonstrate that two independent modes of control may be applied to effectively regulate HSE promoter activity.

3.4. Co-encapsulation of cells and NPs

As mentioned previously, magnetic NPs are used for a wide range of applications [15–18,30]. However, it has been suggested that non-coated magnetic iron oxide NPs are toxic, which might affect the viability of exposed cells [31,32]. For the presented approach therefore a panel of magnetic NPs was evaluated, coated with different polymers, such as polyethylenimine (Supplementary Table S1). The coating stabilises the particles against rapid degradation and aggregation. The particles used here were highly compatible with cell viability. For the proposed application heat induction capabilities were a major criterion. Accordingly, NPs S8, which showed the strongest increase in temperature in response to AMF treatment (Table S1), were chosen to establish co-encapsulation with the stably transfected HEK293 cell clone.

Co-encapsulation of cells with the selected NPs was established by modification of a standard procedure previously applied to encapsulate HEK293 cells into sodium cellulose sulphate (SCS) [3]. Due to the increased viscosity of the SCS-cell mixture now supplemented with 0.5% NPs, the concentration of SCS had to be reduced to 1.6%. This facilitated generation of a vibration-induced stable chain of droplets in the capsule formation process. The encapsulation mixture was dropped into a bath containing 1.3% of the counter ion (pDADMAC), with high voltage applied on an electrode. After a short gelling phase of 3 min the microcapsules were generated by hydrogel formation. The capsules were equally sized with a diameter of 700 μm and contained on average 600 cells per microsphere (Fig. 5a, b).

Fig. 5.

Biocompatibility of co-encapsulated magnetic NPs. HEK293-C5 cells were encapsulated without (a) or with 0.5% NPs S8 (b) and further cultivated. Pictures were taken at the day of encapsulation. (c) Encapsulated cells were cultivated for a period of 28 days and their metabolic activity was monitored over time. Therefore, samples were taken from the culture at day 4, day 11, day 18 and day 28 post encapsulation and equivalents of cell numbers (per capsule) were calculated by means of an AlamarBlue assay. (d–g) To determine the localisation of co-encapsulated NPs, capsules were analysed by electron microscopy. Samples were taken 18 days post encapsulation, fixed and embedded. 100 nm sections of encapsulated cells without NPs (d, e) or of cells co-encapsulated with S8 NPs (f, g) were analysed (N: nucleus, M: mitochondria, NP: nanoparticles). The membrane of one selected cell is indicated by a dotted line (d, f). Sections of the capsule membranes are shown in e and g. (a, b) Scale bar = 700 μm, (d–g) scale bar = 5 μm.

In order to further characterise the generated capsules, the internal localisation of encapsulated NPs was determined. To this aim, capsules containing cells and S8 NPs, as well as capsules containing cells only, were cultivated for two weeks to allow cell proliferation and were subsequently subjected to transmission electron microscopy (TEM). Uranyl-acetate-stained ultra thin sections revealed a dispersed distribution of the NPs within the membrane of the microcapsules (Fig. 5g), indicating that the level of aggregation was low during the encapsulation process. However, small aggregates of NPs were detected in the lumen of the microcapsules (Fig. 5f), most probably originating from the maturation process. The localization of the S8 NPs was mainly found within the cells (Fig. 5f). This is in good agreement with their polyethyleneimine coating, which has been shown previously to mediate polymer-assisted, magnetic-force enhanced transfer of nucleic acids into cells (magnetofection) [33].

3.5. Biocompatibility of magnetic NPs

To determine biocompatibility of the selected S8 NPs in detail, generated capsules were cultured for four weeks and the number of encapsulated cells monitored over time by determining their metabolic activity. For comparison, cells encapsulated without the addition of NPs were analysed as well. Four days after encapsulation, in average 1000 viable cells were determined in either capsules containing or lacking NPs. The encapsulated cells then started to divide, resulting in a continuously increasing cell number. Analysis revealed a slightly decreased metabolic activity of cells co-encapsulated with NPs until day 11 post encapsulation, which fully recovered after day 18 (Fig. 5c). Thus the presence of S8 NPs might initially retard cell proliferation, which is however compensated during cultivation of capsules for a period of four weeks.

Effects of magnetic-field treatment on cells co-encapsulated with NPs compared to those encapsulated without, were first analysed by histological examinations. Capsules that had been cultivated for two weeks were exposed to an AMF generated with 60 kHz and 27 A for 30 min and subsequently further cultivated for 8 days. Samples were taken immediately before treatment (day 0), as well as 2, 5 and 8 days post treatment, fixed in formalin and embedded in paraffin. Histological sections were prepared and stained with haematoxylin and eosin (Fig. 6). Microscopic analysis of capsules lacking NPs (Fig. 6 a–d) as well as AMF-treated capsules containing S8 NPs (Fig. 6 e–h) revealed the presence of a dense network of cells. Only few cells were showing aberrant nuclei (arrows) indicating necrotic cell death, which did not significantly increase with time (compare day 0, day 2, day 5 and day 8).

Fig. 6.

Cytological analysis of AMF-exposed encapsulated cells. HEK293-C5 cells were encapsulated without (a–d) or in the presence of 0.5% S8 NPs (e–h) and subjected to AMF treatment. Samples were taken before (day 0), 2 days (day 2), 5 days (day 5) and 8 days (day 8) after magnetic field treatment, fixed and embedded in paraffin. 3 μm thick sections were prepared and stained with hematoxylin and eosin. Aberrant nuclei are marked with an arrow (scale bar 25 μm).

Next, the presence of apoptotic cells was analysed by staining the prepared sections applying a TUNEL assay (Fig. 7). At all time points after AMF treatment only few cells were detected that exhibited fragmented DNA (Fig. 7, red fluorescent staining). In agreement with the previous finding (Fig. 6), no significant differences were observed in samples co-encapsulated with NPs (Fig. 7 d, h, l and p) compared to capsules containing cells only (Fig. 7 b, f, j and n). In addition, no significant increase in the number of apoptotic cells was observed when samples obtained before treatment (day 0) were compared to those after AMF treatment.

Fig. 7.

Analysis of apoptosis in AMF-treated encapsulated cells. HEK293-C5 cells were encapsulated without (a, b, e, f, i, j, m and n) or in the presence of 0.5% S8 NPs (c, d, g, h, k, l, o, p, s and t) and subjected to AMF treatment. Samples were taken before (day 0), 2 days (day 2), 5 days (day 5) and 8 days (day 8) after magnetic field treatment, fixed and embedded in paraffin. Sections were analysed by applying a TUNEL assay, revealing the presence of fragmented DNA as a red fluorescent staining (rhodamine). As control, total DNA was also stained (DAPI). DNaseI-treated sections served as positive control (q, r). Specificity of detection was shown by omitting digoxygenin-labelled nucleotides in the reaction (negative control: s, t).

Heat formation of the NPs strongly depends on their concentration (see Supplementary Figure S1). This might result in local overheating in cases of aggregated particles and the TEM data indeed indicate some aggregation (Fig. 5f). However, both the cytological analysis as well as the apoptosis assay show that even if localised heat formation might kill individual cells, the neighbouring cells rapidly compensate for this cell death. In summary, the different analyses revealed that co-encapsulation of 0.5% S8 NPs is well tolerated by HEK293 cells. Neither proliferation nor viability appeared to be significantly affected. Most importantly, excitation of NPs by exposing capsules to an AMF for 30 min did not result in increased levels of cell death demonstrating applicability of the proposed system.

3.6. NP-mediated, AMF-induced gene expression in encapsulated cells

Having demonstrated biocompatibility of NPs subjected to AMF treatment for encapsulated cells, the next question was whether they would produce sufficient heat to activate the inducible promoter. In order to provide a proof-of-principle, heat-inducible HEK293-C5 cells were co-encapsulated with magnetic S8 NPs (0.5% w/v). As a control, the HEK293-C5 cells were also encapsulated without the addition of NPs. Further, HEK293 cells stably transfected with either pCMVluc or pCMVgfp and therefore constitutively expressing luciferase or GFP were encapsulated as well. The generated capsules were then either subjected to magnetic field treatment (60 kHz, 27 A) for 30 min, incubated at 43 °C for 45 min (heat induction) or incubated at 37 °C (control). Subsequently, either luciferase activity was determined (Fig. 8a) or GFP expression was analysed by FACS (Fig. 8b).

Fig. 8.

Nanoparticle-mediated, AMF-induced gene expression in encapsulated cells. HEK293-C5 cells were encapsulated without (C5 caps.) or in the presence of 0.5% S8 NPs (C5 + S8 caps.) and either exposed to an alternating magnetic field (grey bars), heat-treated by incubation at 43 °C (black bars) or kept as control at 37 °C (white bars). In addition, encapsulated constitutively luciferase-expressing HEK293 cells (pCMVluc caps.) were also analysed. (a) Luciferase activity was determined 6 hours after treatment in total protein extracts prepared from the different samples. Relative values are indicated, the luciferase activity of cells kept at 37 °C was set to 1 (statistical significance p = 0.029, Mann–Whitney test, n = 5). (b) Quantitative flow cytometry analysis was performed 24 hours after treatment with cells that have been released from capsules. GFP fluorescence blots are shown from one representative experiment (x-axis FL1-H GFP fluorescence intensity, y-axis FL2-H; gates were set according to the controls). As controls HEK293 pCMVluc-transfected (negative) or pCMVgfp-transfected cells were used (absolute levels of luc activity were roughly 10-fold higher for CMV-luc-transfected compared to HEK293-C5 cells).

Heat treatment at 43 °C resulted in a robust induction of luciferase (Fig. 8a, black bars) as well as GFP expression (Fig. 8b) in capsules both containing or lacking S8 NPs. Induction levels of luciferase expression in comparison to controls kept at 37 °C were calculated to be more than 4500-fold. By contrast, magnetic field treatment (Fig. 8a, grey bars) caused a strong induction of reporter gene expression exclusively in capsules, which contained S8 NPs. Here, a more than 1700-fold increase of luciferase activity was determined in comparison to samples kept at 37 °C (Fig. 8a, C5 + S8 capsules), while capsules containing HEK293-C5 cells only (Fig. 8a, C5 capsules) showed a 26-fold increase in luciferase expression levels. Consistently, a more than 950-fold induction of GFP expression was calculated for C5 + S8 capsules (Fig. 8b, AMF) compared to a 13-fold induction in capsules lacking S8 NPs. The determined difference in luciferase expression is highly significant with a p-value of 0.029 (Mann–Whitney test). In contrast to HEK293-C5 cells, reporter gene expression of pCMVluc- (Fig. 8a) and pCMVgfp-transfected cells (Fig. 8b) was neither influenced by magnetic field treatment nor by incubation at 43 °C. In conclusion, in the presence of magnetic NPs, exposure of encapsulated HEK293-C5 cells to an AMF resulted in a significant induction of luciferase and GFP expression, both driven by the artificial bidirectional heat-inducible promoter. The coil heats up substantially during operation. Due to its small dimensions, it was not possible to completely shield the treated capsules from heating.

4. Discussion

Encapsulation of cells is a promising technique for cell therapy applications. In particular genetically modified heterologous cells can thus be protected from the immune system of the host [3,34]. The cells can extensively be selected and tested in cell culture and therefore are optimised for expression of a therapeutic protein. Basically all kinds of proteins and peptides can be produced by the transplanted cells [3]. Potential applications for cell therapy so far mainly comprise replacement strategies, which seek to compensate for gene deficiencies, yet a fine-tuned interference with genetic and biochemical pathways would open many other fields of applications. Critical regulatory constraints apply for example to hormones [35] or growth factors [36–38]. Similarly, expression of the pain alleviating proopiomelanocortin encoding gene ideally should be controlled in real-time by the patient [39]. Thus, promoters facilitating inducible rather than constitutive expression of therapeutic genes and, moreover, the possibility to control gene expression from outside a patient's body, would be of great advantage.

In our concept, heat shock promoters are applied representing a one-component inducible expression system. Two-component systems are more often discussed for such approaches; however, they exhibit a number of disadvantages for gene and cell therapy. Two-component systems use modified transcription factors which evoke an interfering immune response. Mostly these promoters are induced by small molecules, which reach their targets by diffusion and as such show slow pharmacokinetics. The TetR system, for example, was shown to facilitate inducible gene expression in encapsulated myoblasts that had been implanted into mice; however, it took several weeks to switch between the on and the off state of the promoter [40]. Furthermore, the inducer molecules are often themselves pharmacologically active and cross-react with other biological pathways [41].

Heat shock promoters in contrast use endogenous pathways and transcription factors and lack the above mentioned disadvantages of two-component systems. Heat shock promoters such as the HSP70 promoter have therefore repeatedly been used for gene therapy applications [42]. However, in addition to HSF1-mediated activity, natural promoters integrate multiple cellular emergency pathways and usually show a high background activity as well as tissue-specific variations. To overcome this problem, an artificial heat shock promoter built from multiple high affinity HSEs linked to a minimal CMV promoter element [21] was used in the presented study. Reporter gene expression driven by this artificial promoter in the generated HEK293 cell line was shown to be induced several 1000-fold by heat treatment, while exhibiting an extremely low background activity in the uninduced state (see Fig. 3A).

Temperature represents a well controllable physical parameter. The almost uniform temperature of mammals makes localised overheating an ideal tool to regulate gene expression. Several methods have been established to induce such a local hyperthermia in patients. The basis for these developments is the favourable combination of hyperthermia with radiotherapy or chemotherapy for the treatment of tumours [15]. For superficial hyperthermia, simple water baths or electromagnetic applicators emitting microwaves are suitable [43]. More demanding is a non-superficial heat production. Focused ultrasound has been demonstrated to induce small, well defined temperature peaks deep within the patients tissue [42]. This method has also been applied for activation of HSP70 promoters in gene therapy experiments [44]. Recently, another approach was introduced using magnetic NPs, activated by an AMF (reviewed in Ref. [45]). After injection of NPs, magnetic hyperthermia can precisely be induced by hysteresis heating within the tumour tissue [11], demonstrating the feasibility of such nanotechnology-based therapies.

In our approach, as a result of co-encapsulation, NPs and therefore also hyperthermia, are restricted to the expressing cells. We showed that when suspended with cells, the magnetic particles were biocompatible with the cells and also allowed a precise control of heat shock promoter activity over a wide range of expression (see Fig. 4). In particular, we were able to demonstrate regulation of promoter activity by varying either the magnetic field strength or the time of activation. In both cases, a control over several orders of magnitude was possible. Most importantly, this regulation worked within minutes, strongly contrasting the above mentioned two component systems activated by small molecules. Therefore, inducible gene expression mediated by magnetic NPs and an AMF is exceptional among the available inducible expression systems for therapeutic purposes.

The NPs at the applied concentration showed excellent biocompatibility both with and without AMF treatment. The AMF induced strong expression of the respective marker proteins in the encapsulated cells, providing a proof-of-principle for the concept. Currently, several clinical trials using modified encapsulated cells are in progress (reviewed in Ref. [46,47]). A next step for the establishment of our proposed therapeutic system would include in vivo experiments with larger, commercially available magnetic field generators. With respect to clinical applications, the company MagForce Nanotechnology AG in June 2010 received EU-wide regulatory approval for utilisation of their NPs and magnetic field generators for hyperthermia treatment of brain tumours. Consequently, magnetic field generators for treatment of patients will be available in the future allowing further evaluation of the novel cell therapy concept presented in this study.

5. Conclusion

Biologicals represent an indispensable part of modern medicine. A production of these molecules within the patient offers a number of advantages, in particular the concentrated expression directly in the affected tissues or organs. Cell therapy with encapsulated cells provides a solution for such a treatment. In order to enable dosage control we developed a method for regulated gene expression in encapsulated cells.

The presented method combines magnetic nanoparticles with a heat inducible promoter. An alternating magnetic field selectively activates the nanoparticles. This combination works highly efficient and allows a fine tuned regulation of gene expression over a broad range. Compared to other inducible systems the modulation of the expression by the magnetic field is fast and reproducible. The produced heat remains concentrated to the capsules, not affecting the surrounding tissue and is readily tolerated by the transplanted cells. Taken together our method allows the dose dependent application of biologicals by cell therapy.

The following are the supplementary materials related to this article.

Supplementary Figure S1

Dependence of the heat production on the magnetic nanoparticle concentration. For determination of the heat production (ΔT) in an AMF different concentrations of Sigma nanoparticles were incubated for 30 min at 27 A and 60 kHz and the temperature of the solution was measured immediately before and after AMF treatment.

Supplementary Table S1

Magnetic nanoparticles of the core-shell type. Palmitoyl-dextran 1 and 2 with 11 and 32 palmitoyl groups per 100 dextran units were synthesised by means of esterification of Dextran-10 (Amersham Biosciences) with palmitoyl chloride as described previously [42]. Polyethylenimine: 25 kD branched polyethylenimine; 1,9-NDT: 1,9- nonandithiol; FSA: fluorinated surfactant ZONYL FSA (lithium 3-[2-(perfluoroalkyl) ethylthio] propionate); FSE: fluorinated surfactant ZONYL FSE. Chitosan-OSL: chitosan oligosaccharide lactate Mn < 5000. For determination of the heat production (ΔT) in an AMF, 0.5% of the nanoparticles were incubated for 30 min at 27 A and 60 kHz and the temperature of the solution was measured immediately before and after AMF treatment.