Magnet-Bead Based MicroRNA Delivery System to Modify CD133+ Stem Cells

Aim. CD133+ stem cells bear huge potential for regenerative medicine. However, low retention in the injured tissue and massive cell death reduce beneficial effects. In order to address these issues, we intended to develop a nonviral system for appropriate cell engineering. Materials and Methods. Modification of human CD133+ stem cells with magnetic polyplexes carrying microRNA was studied in terms of efficiency, safety, and targeting potential. Results. High microRNA uptake rates (~80–90%) were achieved without affecting CD133+ stem cell properties. Modified cells can be magnetically guided. Conclusion. We developed a safe and efficient protocol for CD133+ stem cell modification. Our work may become a basis to improve stem cell therapeutical effects as well as their monitoring with magnetic resonance imaging.


Introduction
The transplantation of stem cells for the recovery of damaged tissue represents a promising strategy in regenerative medicine. Adult stem cells bear considerable potential since they are easily accessible from patients or healthy donors without posing ethical conflicts [1]. Among these, cells expressing the highly conserved transmembrane CD133 antigen (prominin-1) represent a heterogeneous stem and progenitor cell population capable of differentiating into haematopoietic, endothelial, and myogenic lineages [2][3][4]. We and others have shown that their regeneration potential is mainly based on cytoprotective effects and their contribution to neovascularization processes through differentiation into newly forming vessels and activation of proangiogenic signaling by paracrine mechanisms [5][6][7][8]. The high therapeutic relevance of these cells is reflected in more than 30 approved clinical trials (Clin-icalTrials.gov) using adult CD133 + stem cells for the treatment of various degenerative diseases. For instance, remarkably promising results have been reported for the treatment of chronic ischemic cardiomyopathy. An increase of left ventricular ejection fraction (LVEF) by 5-6% following intramyocardial application led to the approval of first phase III clinical studies in the field of CD133 research [9][10][11][12][13]. However, despite the rapid translation from bench to bedside, broad clinical application of stem cells is still hampered by low retention in the organ of interest and massive initial cell death [6,14,15]. Engineering of cells prior to transplantation can address all of the mentioned challenges. To accomplish this objective, the development of a safe and efficient way of CD133 modification for relevant improvement of their properties is required.
The retention of therapeutic cells at the site of interest is a fundamental prerequisite for an effective therapy [16]. Nevertheless, in highly perfused organs, such as the brain and the heart, 90-99% of transplanted cells are typically lost during the first 1-2 h, independently of the cell type and application route [17][18][19][20][21][22][23][24][25][26][27]. To overcome this immense washout, magnetic targeting of iron oxide labeled stem cells can be applied as an innovative noninvasive strategy to enhance immediate retention and therefore improve long-term engraftment and 2 Stem Cells International functional benefits [28][29][30][31][32]. In addition, cells which are magnetically labeled can be tracked and monitored using magnetic resonance imaging (MRI) [33] or the novel imaging modality magnetic particle imaging (MPI) [34,35].
Another major limitation of stem cell therapy is the massive cell death after transplantation due to the hostile, inflammatory environment of the target tissue [36]. Several strategies have been developed to improve the survival of grafted cells, including their preconditioning by hypoxia, heat shock, or cytokine treatment [37,38]. In addition, it has been successfully demonstrated in various studies that antiapoptotic microRNAs (miRs) both inhibit apoptosis in vitro and increase cell engraftment following transplantation in vivo [39]. MiRs have been proven to be important translational controllers of stem cell fate and behaviour, avoiding the hazard of stable integration into the genome [40]. Therefore, these small molecules are ideal candidates for cell engineering aiming at survival improvement.
Currently, delivery of nucleic acids into hard to transfect primary stem cells is almost exclusively based on recombinant viruses which are the most efficient vehicles so far [14,41]. However, uncontrollable gene expression, pathogenicity, immunogenicity, and insertional mutagenesis of viral vectors remain major obstacles for widespread clinical translation [41,42]. Consequently, the necessity of safer gene delivery methods has led to the development of various nonviral systems which are nonpathogenic, nonimmunogenic, and not limited by the size of delivered genetic material [43].
Nowadays, polyethylenimine (PEI) is one of the most efficient polymers for miR delivery, promoting nucleic acid protection against degradation, cellular uptake, and intracellular release [44]. The implementation of miR-PEI constructions in first clinical trials is demonstrating their high biocompatibility [45]. In our group, a vector has been designed, which consists of biotinylated PEI bound to streptavidin-coated iron oxide magnetic nanoparticles (MNPs) (Figure 1). During previously carried trials, the group worked on the adjustment of vector efficiency and safety. During these studies, it has been demonstrated that pDNA and miR can be efficiently delivered and processed in human mesenchymal stem cells (hMSCs) [46,47].
In this study, we worked on the development of an efficient strategy for magnet-bead based miR delivery into highly clinically relevant cell type, CD133 + stem cells. First, we have demonstrated that optimized transfection complexes are suitable for sufficient miR delivery into bone marrow (BM) derived CD133 + stem cells without affecting stem cell marker expression and haematopoietic differentiation capacity. Moreover, we showed that modified cells can be magnetically guided in vitro. These two achievements together form an essential prerequisite for further preclinical testing.
Calculation of cell viability and surface marker integrity was based on the following equations:  (CFU) assays, surface marker expression analysis, and intracellular visualization of complexes. Both miRs are doublestranded RNA molecules designed to mimic endogenous mature miRNAs. In order to ensure effective binding to streptavidincoated MNPs, branched PEI with a molecular weight of 25 kDA (Sigma-Aldrich, USA) was biotinylated using EZ-Link Sulfo-NHS-LC-Biotin (Thermo Fisher Scientific GmbH, Germany) according to the manufacturer's instructions and previous reports [49]. Minor changes were made in solvents (water instead of DMSO) and purification method: sizeexclusion chromatography with PD-10 Desalting columns (GE Healthcare, UK) was used instead of dialysis as described elsewhere [50]. Final concentration of amine groups was measured by ninhydrin assay (2% ninhydrin reagent, Sigma-Aldrich) and the biotinylation degree was determined as 1.585 ± 0.018 mmol biotin/mmol PEI in HABA assay using Pierce Biotin Quantitation Kit (Thermo Scientific).
Streptavidin MagneSphere5 Paramagnetic Particles (Promega Corporation, USA) were filtered through 0.45 m Millex-HV PVDF Syringe Filter Units (EMD Millipore Corporation, USA) to remove bigger aggregates and particles. Biotinylated PEI and filtered MNPs were stored at 4 ∘ C until usage.
For the formation of polyplexes, miR and PEI were diluted in equal volumes of 5% glucose solution (MP Biomedicals, Germany) in appropriate amounts to prepare different molar ratios of PEI nitrogen and miR phosphate (N/P ratios). After vigorous mixing, miR/PEI solution was incubated for 30 min at RT.
In order to form magnetic polyplexes, MNPs were incubated in an ultrasonic bath for 20 min every time before use to remove formed clusters. Prepared this way, MNPs were mixed with previously prepared miR/PEI complexes in different concentrations and the solution was incubated for 30 min at RT.
Freshly isolated 5 × 10 4 CD133 + cells were seeded for transfection in a 24-well plate and prepared miR/PEI or miR/PEI/MNP complexes were added immediately. After incubation for 18 h at 37 ∘ C and 5% CO 2 in Dulbecco's Modified Eagle Medium (DMEM, Pan Biotech GmbH) supplemented with 1% penicillin/streptomycin (PAA Laboratories GmbH, Germany) and 2% fetal bovine serum (FBS, Pan Biotech GmbH), cells were directly used for measurements or fresh medium was supplied. In order to further optimize culture conditions for CD133 + cells, in one set of tests, StemSpan6 H3000 culture medium (STEMCELL Technologies Inc., Canada) supplemented with StemSpan CC100 4 Stem Cells International (STEMCELL Technologies Inc.) was used for incubation (see Figure 8).

Uptake Efficiency and Cytotoxicity of Transfection Complexes.
For the quantification of uptake efficiency and cytotoxicity of different transfection complex formulations, CD133 + cells were stained 18 h after transfection for 10 min at 4 ∘ C with LIVE/DEAD5 Fixable Near-IR Dead Cell Stain Kit (Molecular Probes, USA) and fixed with 4% formaldehyde solution (FA) (Merck Schuchardt OHG, Germany). Samples were measured with LSR-II flow cytometer and data were analysed with FACSDiva software. The representative gating strategy is depicted in Figure S2.
Qualitative analysis of transfected CD133 + cells was carried out 18 h after transfection based on Cy3-labeled miR. For this purpose, cells were washed once with 2% FBS in phosphate buffered saline (PBS, Pan Biotech GmbH) in order to remove noninternalized particles and fixed with 4% FA for 20 min. Afterwards, cells were spun down to a coverslip and washed again with PBS. Then, the coverslip was mounted with Fluoroshield6 containing DAPI (Sigma-Aldrich) on microscope slides. These prepared samples were subjected to laser scanning confocal microscopy (40x oil immersion) in the tile-scan mode in order to acquire larger areas of 1062.33 m × 1062.33 m. In addition, in order to ensure that exclusively internalized Cy3 signal was analysed, -stacks were recorded of approximately 7 m depth.

Magnetic
Targeting. The verification of magnetic cell targeting was performed with CD133 + cells transfected at optimized conditions (20 pmol miR; N/P ratio 7.5; 3 and 5 g/mL MNPs). 18 h after transfection, cells were transferred to a 12-well plate and a magnetic field was applied locally using magnetic plate (OZ Biosciences, France; field strengths 70-250 mT and 50-130 T/m, resp. [51]). 24 h after incubation at 37 ∘ C and 5% CO 2 , cell numbers in the area with (+ ) and without (− ) magnet were counted by microscopic observation using the LSM 780 ELYRA PS.1 system and image files were analysed with ZEN 2011 software (Carl Zeiss Microscopy GmbH, Germany) and ImageJ 1.48 (NIH, USA). Magnetic targeting ratios were calculated by dividing obtained cell numbers of both areas (+M/ −M).

Magnetic Particle Spectroscopy to Define Magnetic
Loading of CD133 + Cells. Cell samples modified under the conditions defined as optimal for transfection and magnetic cell guidance were investigated by magnetic particle spectroscopy (MPS) to quantify the magnetic iron loading of CD133 + cells with miR/PEI/MNP complexes. MPS is a sensitive magnetic detection method that allows for the specific quantification of the magnetic nanoparticle iron content without being affected by cells or suspension medium [52,53]. After incubation of 18 h, untransfected (control) and transfected CD133 + cells were collected and washed with 2% FBS in PBS. Subsequently, cell samples of about 5 × 10 5 cells were resuspended in PBS, fixed with 4% FA, and transferred into 0.2 mL tubes for the measurements by a commercial MPS device (Bruker Biospin, Germany) operating at a magnetic field drive = 25 mT and a frequency 0 = 25 kHz. This device determines the nonlinear dynamic magnetic moments of the sample at higher (odd) harmonics of 0 . The iron quantification of the samples is carried out by the third harmonic 3 of the MPS spectrum which is normalized to the corresponding 3,ref of the MNP reference sample of known iron amount (3.2 g for MNPs only reference and 2.5 g for the CD133 MicroBeads reference) as described before [54]. The ratio of the fifth to third harmonic 5 / 3 is used as a characteristic fingerprint to identify the magnetic nanoparticle type in the sample.

Intracellular Visualization of Transfection Complexes.
For visualization of complexes, Pre-miR miRNA Precursor Molecules Negative Control #1 (Ambion) was labeled with Cy56 dye using Label IT 5 miRNA Labeling Kit (Mirus Bio LLC, USA) as recommended in the manufacturer's instructions. Redundant dye was removed using supplied purification column. The Cy5-labeled miR final concentration was measured spectrophotometrically in NanoDrop ND-1000 (Thermo Scientific, USA) and samples were stored at −20 ∘ C protected from light until further use.
PEI was stained using FluoReporter5 Oregon Green5 488 Protein Labeling Kit (Molecular Probes) according to the manufacturer's protocol. Therefore, PEI was mixed with 1 M sodium bicarbonate solution and incubated for 1 h with Oregon Green stock solution (10 mg/mL in DMSO). Unreacted dye was removed by size-exclusion chromatography using PD-10 Desalting columns (GE Healthcare, UK). 488labeled PEI concentration was defined in the ninhydrin assay and aliquots were further stored at 4 ∘ C protected from light.
MNPs were stained using Atto 565 dye conjugated to biotin (ATTO-TEC GmbH, Siegen, Germany). For this purpose, Atto 565 dye and MNPs (ratio of 1 : 1000 w/w) were mixed simultaneously with prepared miR/PEI complexes and incubated for 30 min in the dark.
To monitor the intracellular distribution of complexes, 5 × 10 4 CD133 + cells were transfected with fluorescently labeled complexes at previously optimized conditions (20 pmol miR; N/P ratio 7.5; 3 g/mL MNPs). 18 h after transfection, samples for microscopy were prepared as described above.
First, in order to assess the intracellular localisation of all delivery vector parts, confocal laser scanning microscopy (LSM) was performed using ELYRA PS.1 LSM 780 system. ZEN software (Carl Zeiss Microscopy GmbH) was applied for image processing.
Further, the localisation of miR/PEI/MNP was studied more in detail using 3-dimensional structured illumination microscopy (SIM). Four-colour imaging was performed using 100x alpha Plan-Apochromat 1.46 objective (oil immersion): 405, 488, 561, and 633 nm laser lines for excitation. SI raw data was acquired in -stacks with a 16-bit depth at 3 angles, 3

Haematopoietic CFU (CFU-H) Assay.
In order to monitor haematopoietic differentiation capacity of miR-modified CD133 + cells, CFU-H assays were performed. 18 h after transfection with Pre-miR miRNA Precursor Molecules Negative Control #1, cells were mixed with MethoCult H4434 Classic (STEMCELL Technologies) and 1 × 10 3 cells were seeded per 35 mm dish. Formed colonies were identified and counted for the assessment after 14 d incubation at 37 ∘ C and 5% CO 2 . As recommended by manufacturer, all samples were examined in duplicate.

Optimization of miR/PEI Complexes for Transfection.
In order to optimize the transfection of CD133 + cells, different compositions of miR/PEI complexes consisting of four different miR amounts (10, 20, 30, and 40 pmol per 5 × 10 4 cells) and three different N/P ratios (2.5, 5, and 7.5) were tested in terms of uptake efficiency and cytotoxicity (Figure 2). In all measurements, untransfected cells served as internal control representing the cytotoxic effect of the isolation and cultivation procedure.
Complexes with the smallest miR amount (10 pmol) showed the lowest uptake rates (ranging between ∼20 and 60% Cy3 + cells) and a minor increased cytotoxicity (∼40% dead cells) compared to the control (∼25% dead cells). As expected, complexes consisting of higher miR amounts led to a significantly increased uptake (up to ∼95% Cy3 + cells, 40 pmol miR; N/P 7.5) but also resulted in increasing cytotoxic effects (up to ∼80% dead cells, 40 pmol miR; N/P 7.5) because higher PEI amounts were required. Therefore, complexes composed of 20 pmol miR were considered as optimal for transfection of CD133 + cells representing a balance between increase in uptake rates (∼75-90% Cy3 + cells) and compromised cell survival.

MiR/PEI/MNP Complexes Are Suitable for CD133 + Stem
Cell Transfection. To achieve the possibility of magnetic targeting, previously selected polyplexes (20 pmol miR; N/P ratio 2.5, 5, and 7.5) were complemented by MNPs in six different concentrations (0, 1, 2, 3, 4, and 5 g per mL of prepared miR/PEI mixture). The influence of MNPs with respect to uptake efficiency and cytotoxicity was analysed by flow cytometry and confocal microscopy ( Figure 3). Again, untransfected cells served as internal control.
Uptake rates of miR/PEI/MNP complexes did not significantly differ from respective miR/PEI complexes (0 MNPs), whereas raising N/P ratios (2.5, 5, and 7.5) led to increasing uptake efficiencies (∼40%, ∼60%, and ∼80% Cy3 + cells, resp.) (Figure 3(b)). Representative images illustrate the intracellular localisation of labeled miR-Cy3 and its high uptake rates when a complex formulation of 20 pmol miR N/P ratio 7.5 combined with 3 g/mL MNPs was applied (Figure 3(a)). Furthermore, no significant changes in cytotoxicity were determined between the control (36 ± 11% dead cells) and transfected cells (ranging between ∼25 and 35% dead cells) (Figure 3(c)). In following experiments, magnetic complexes composed of 20 pmol miR and N/P ratios 5 and 7.5 combined with 3 and 5 g/mL MNPs were used with respect to their highest uptake rates and low cytotoxicity compared to the control.

MiR/PEI/MNP-Modified CD133 + Stem Cells Can Be
Efficiently Guided via a Magnetic Field. To investigate the influence of a magnetic field on the guidance of cells transfected with optimized miR/PEI/MNP complexes, a magnet was applied locally under the well plate for 24 h (Figure 4(a)).
In case of untransfected cells, almost the same cell numbers were detected in both areas (with and without magnet) ( Figure 4(c)). These results indicate that MACS CD133 MicroBeads alone are not sufficient for magnetic targeting of cells. In contrast, cells transfected with magnetic complexes composed of 20 pmol miR and N/P ratios 5 and 7.5 combined with 3 and 5 g/mL MNPs showed higher cell numbers in the area of the magnet (+ magnet) than in the area without magnet (− magnet). Correspondingly, transfected cells showed significantly higher magnetic targeting ratios (ratios between 1.6 and 2.6) than untransfected cells (ratio of 1±0.12) (Figure 4(b)). Moreover, transfection complexes with N/P ratio 7.5 led to significantly higher targeting ratios (2.6 ± 0.17 and 2.2 ± 0.08 with 3 and 5 g/mL MNPs) than complexes with N/P ratio 5 (1.6 ± 0.09 and 1.9 ± 0.02 with 3 and 5 g/mL MNPs). Therefore, complexes composed of 20 pmol miR and  29  22  56  32  55  28  62  25  53  23  31  32  49  31  45  23  57  24  69  34  33  40  41  29  51  28  46  16  54  24 5 g/ml MNPs 3 g/ml MNPs 5 g/ml MNPs 3 g/ml MNPs +magnet +magnet +magnet +magnet +magnet −magnet −magnet −magnet −magnet −magnet  Stem Cells International N/P ratio 7.5 combined with 3 or 5 g/mL MNPs are assumed to be most suitable for targeting of transfected cells in a magnetic field and were used for further experiments.

Intracellular Iron Concentration.
For the conditions defined as optimal for transfection and magnetic cell guidance, the amount of intracellular iron has been quantified using magnetic particle spectroscopy (MPS). As a result, we observed that a cellular uptake of 0.155 ± 0.0419 pg iron per cell could be achieved with our protocol for cells transfected by miR/PEI/MNP complexes. Importantly, the corresponding 5 / 3 values in the range 20.5 ± 3.1% allowed for a unique identification of successfully uptaken miR/PEI/MNP complexes, avoiding false positive signals emanating from magnetic cell isolation beads. MPS allows this discrimination due to the fact that untransfected as well as transfected cell samples yield signals with 5 / 3 values in the range of 27 to 30% which can be attributed exclusively to CD133 MicroBeads used for cell isolation.

Intracellular Visualization of Transfection Complexes.
For intracellular detection of the optimized transfection complexes, CD133 + stem cells were transfected with labeled miR/PEI/MNP complexes. The analysis of cellular complex uptake 18 h after transfection with confocal microscopy allowed for concluding that all vector parts are sufficiently delivered into cells ( Figure S3). This additionally validates the flow cytometry results: when proper washing steps are applied, complexes from the surface are removed, whereas intracellular uptake is being analysed. In order to monitor vector distribution with higher resolution, SIM was performed (lateral resolution up to 100 nm compared to 200-500 nm for confocal). As a result, we observed that 18 h after transfection miR/PEI/MNP complexes were mainly detected in the cytoplasm. All vector parts were located in the perinuclear region with no discernible differences between them ( Figure 5). The signal of PEI was always colocalised with MNP and miR. Formed structures 100-400 nm in size were located in the other focal plane compared to nucleus as recording of -stacks demonstrated.

miR/PEI/MNP Complexes Have No Negative Influence on Cell Viability and Surface Marker Pattern.
To evaluate the impact of transfection and subsequent 18 h culturing time on cell viability and surface marker pattern, flow cytometry measurements were performed and Boolean gating strategy was adapted on the ISHAGE guidelines for CD34 + cells. As described before, parameters defined as optimal for the targeting approach were used for this purpose and untransfected cells served as internal control.
A significant decrease in the viability of untransfected CD133 + cells from 95 ± 1% (0 h control) to 77 ± 3% (18 h control) was observed ( Figure 6). This represented cell death after 18 h culturing time most likely resulted from the suboptimal content of DMEM culture medium selected for the establishment of the transfection system. At the same time, no significant difference in viability was determined between control and cells transfected with 20 pmol miR, N/P 7.5, and 0, 3, and 5 g/mL MNPs (∼70% living cells) after 18  illustrating that the chosen transfection setup itself did not affect cell viability.

Modified CD133 + Stem Cells Retain Their Haematopoietic Differentiation
Potential. BM derived CD133 + cells are known to bear multipotent differentiation potential. In order to address a potential influence of our transfection system on their haematopoietic differentiation capacity, CFU-H assay was utilized. The comparison of untransfected cells (control) and cells modified with optimized miR/PEI and miR/PEI/MNP complexes (20 pmol miR; N/P ratio 7.5; 0, 3, and 5 L/mL MNPs) showed no significant difference in the amount of formed CFU-granulocyte, erythroid, macrophage, megakaryocyte (CFU-GEMM), CFU-granulocyte, macrophage (CFU-GM), Burst-forming unit-erythroid (BFU-E), and CFU-erythroid (CFU-E) (Figure 7(e)). Moreover, CFUs Stem Cells International formed by transfected cells had no morphological abnormalities (Figures 7(a)-7(d)). These results clearly demonstrate that transfection under optimized conditions has no influence on haematopoietic differentiation potential of CD133 + cells.

Modified CD133 + Cells Retain Cell Viability and Surface Marker Pattern in Supportive Cytokine-Supplemented Culture
Medium. In the initial transfection setup, 18 h was sufficient for transfection but cell viability was affected when DMEM medium was used for culturing. Therefore, we further have tested a different culture medium suitable for haematopoietic stem cells (HSCs) (StemSpan supplemented with StemSpan CC100) in order to apply the optimized transfection system in a clinically more relevant scenario (Figure 8). We have observed that this more complex medium did not impair the high transfection efficiencies of the system: 87 ± 2% and 82 ± 5% for 20 pmol miR, N/P 7.5 combined with 3 and 5 g/mL MNPs, respectively, in StemSpan versus 79±9% and 80±8% in DMEM. Yet, in StemSpan, cell viability remained unaffected 18 h and 24 h after transfection compared to 0 h, whereas in DMEM it was reduced by ∼25% dead cells after 18 h. Moreover, cultivation of CD133 + cells in StemSpan caused no changes in the expression of stem cell surface markers (CD45, CD34, and CD133) after an incubation of 18 h and 24 h of control as well as transfected cells.

Discussion
Several phases, I and II, clinical studies have shown that CD133 + stem cell application for the treatment of various diseases (e.g., liver disease, critical limb ischemia, amyotrophic lateral sclerosis, spinal cord injuries, cerebral palsy, chronic ischemic cardiomyopathy, and leukaemia) is safe and feasible [10,[55][56][57][58][59][60][61][62][63][64][65]. However, despite the large amount of collected clinical data, the benefit seen in trials is inconsistent and small. This underlines the urgent need of further optimization of the cells for an improved clinical outcome. Therefore, we developed a nonviral multifunctional transfection system which may become a basis to improve the properties of freshly isolated human CD133 + stem cells for innovative therapies. This multifunctional system enables efficient miR delivery (∼80%) with no significant cytotoxic effects. Moreover, the proposed system ensures magnetic guidance to the site of interest. Simultaneously, we demonstrated that our system has influence neither on stem cell marker expression nor on the haematopoietic differentiation capacity of cells.
The introduction of miRs into stem cells can improve cell survival and engraftment by posttranscriptional gene regulation [39,40,[66][67][68]. Furthermore, miR mediated cell modification can be used for programming of CD133 + stem cells into specific cell types. Recently, it was shown that miR-146a, miR-150, and miR-451 promote the differentiation of CD133 + stem cells into t-lymphoid and erythroid lineages, respectively [69,70]. This may become valuable in the production of artificial blood for avoiding allogenic blood transfusion after chemotherapy or in the treatment of blood disorders [69]. Yet, in all of these studies, the virus-based delivery of miR into CD133 + stem cells was used. However, clinical translation of viruses is still hampered by safety issues including potential immunogenicity and insertional mutagenesis [68].
To reach a safe and efficient way for nonviral miR delivery into CD133 + stem cells, we initially tested miR/PEI complex with different compositions. PEI is one of the most efficient reagents for gene delivery and has been used in several clinical trials proving its safety and biocompatibility [71]. Due to the high cationic charge, PEI forms stable complexes with miR which enter the cell through endocytosis [72]. Subsequently, PEI promotes endosomal escape by causing influx of protons and water (proton sponge effect) leading to endosome swelling and disruption to release complexes into the cytoplasm [73]. However, the main drawback of PEI is its potential cytotoxicity after its intracellular accumulation [74]. Keeping in mind this toxicity issue, we selected complexes composed of low miR (20 pmol per 5 × 10 4 cells) and PEI (N/P ratios 2.5-7.5) amounts for CD133 + stem cell engineering, although higher amounts led to improved uptake efficiencies ( Figure 2). In the next step, miR/PEI complexes were combined with MNPs via biotin-streptavidin bonds, which enabled efficient magnetic targeting in vitro (Figure 4). In addition, the presence of a magnetic core adds the potential for noninvasive MRI tracking [51]. The application of MNPs is a well investigated, multifunctional tool, which has been used for monitoring, magnetic drug targeting, magnetic fluid hyperthermia, and magnetic labeling and separation of cells [51,75]. Recently, it was shown that magnetic field based delivery of magnetically labeled CD133 + stem cells improved cell retention at the site of interest and enhanced the repair of skeletal muscle and spinal cord injury [76,77]. However, we have been able to demonstrate that MACS CD133 MicroBeads alone are not sufficient for magnetic targeting of cells in vitro (Figure 4). For this reason, we evaluate the influence of utilized MNPs. Therefore, CD133 + stem cells were transfected with previously selected miR/PEI complexes complemented with six different MNP amounts (0, 1, 2, 3, 4, and 5 g/mL) (Figure 3). Our current results showed no significant differences in uptake efficiencies and cytotoxicity of miR/PEI/MNP complexes compared to respective miR/PEI complexes. This result is in correspondence with previous studies, where no differences between magnetic and nonmagnetic complexes were found after transfection of freshly isolated CD105 + human mesenchymal stem cells [78]. In contrast, the amount of PEI influenced uptake efficiency: for each MNP concentration, higher NP ratios correlated with higher numbers of Cy3positive cells. This observation can be easily explained by the widely described fact that unbound PEI is necessary for the nucleic acid delivery [79][80][81]. Since we have detected no increase in toxicity of complexes formed by 20 pmol miR per 5 × 10 4 cells, N/P ratio 7.5, and 0-5 g/mL MNPs as compared to untransfected cells (control), these complexes were considered as optimal for transfection of CD133 + cells.
At the same time, however, it should be noted that remarkable variations in viability and transfection efficiency were detected among cells from different donors. This observation corresponds to previous findings of our group, where BM derived hMSCs were used for transfection [46,82]. Yet, it is not clear which inherent differences from patients lead to the individual variations and further studies should be considered prior to the clinical translation. Moreover, measuring uptake efficiency of Cy3 dye-labeled Pre-miR Negative Control #1 does not allow showing the miR processing within cells. However, previously obtained results by our group should be taken into account: the optimal Cy3-miR transfection conditions resulted in sufficient release and effect of miR once tested with functional miRs and evaluated by qPCR [47,83]. Therefore, we assume that miR delivery into CD133 + cells indicates also an efficient cell transfection.
Moreover, we intended to study the targeting potential of modified CD133 + stem cells in vitro. For this purpose, a magnet was attached under the culture plate. As a result, significantly higher amounts of cells were located in the area over the magnet versus the area without magnet (Figure 4). In contrast, untransfected cells showed almost the same cell numbers in both areas. This demonstrates that cells transfected with our magnet-bead based system can be efficiently guided to specific locations in the culture plate using a magnetic field. Therefore, a sufficient amount of incorporated iron was defined as 0.155 pg per cell by MPS. However, the guidance is limited to a certain degree (approximately one out of four cells was not targeted) and increased MNP amounts did not lead to improved targeting ratios (2.2 ± 0.08 with 5 g/mL MNPs). This observation can be explained by hypothesizing that nontargeted CD133 + stem cells include the percentage of untransfected cells (∼20%) and dead cells. This might also explain lower targeting ratios (between 1.6 and 1.9) after transfecting cells with complexes showing lower uptake rates (∼60% using N/P ratio 5). To conclude, performed in vitro targeting experiments demonstrate that the localisation of modified cells can be potentially controlled by applying a magnetic field. This potential might become crucial to overcome initial cell washout from the injured tissue and therefore be beneficial to reach better long-term engraftment and functional benefits in vivo. However, many factors in internal environment can affect cell retention in both negative and positive ways; therefore, future in vivo trials are clearly necessary.
In addition, successfully applied protocol for 3-colour labeling of complex components and delivered nucleic acid might be applied to track the vector fate in tissues after cell transplantation, when progressing to in vivo stage.
Importantly, we could demonstrate that our system influences neither the level of stem cell marker expression (CD34 and CD133 based on ISHAGE guidelines) (Figures 6(a) and 8(a)) nor the differentiation into several haematopoietic lineages (Figure 7). Despite this fact, initial culture conditions do affect cell viability. In particular, we discovered a notable decrease (∼25%) of CD133 + stem cell viability after 18 h cultivation time at 37 ∘ C and 5% CO 2 in DMEM. This observation corresponds with previous studies of our group, where a significant loss of CD133 + stem cell viability 72 h after isolation was shown, even under 4 ∘ C storage [84]. To date, many efforts have been devoted to the selection of proper supportive medium for HSCs. The combination of SCF and Flt3L with IL-11 or Tpo cytokines was defined as most supportive for HSC survival, proliferation, and maintenance in vitro and therefore represents the basic supplement in respective types of medium such as StemSpan [85,86]. However, very few information is available on suitability of such medium for transfection, in particular when cationic polymer based delivery vectors are applied [87]. Therefore, we have selected simple DMEM medium supplemented with FBS and antibiotics for the establishment of our transfection system. Thereby, we have proven the suitability of our miR/PEI/MNP system for effective miR delivery as well as for magnetic cell targeting. At the same time, the culture conditions were proven to be suboptimal for cell viability and expansion: already after 18 h, viability decreased to ∼70-75%. Therefore, we have examined the possibility to efficiently transfect CD133 + cells in a more clinically relevant medium supplemented with supporting cytokines. We have observed that in StemSpan H3000 (xeno-and serum-free) supplemented with StemSpan CC100 suitable for HSC culture, transfection worked equally as in DMEM in terms of high efficiency: 80-90% of cells had taken up Cy3-miR. Importantly, cell viability as well as expression of stem cell surface markers remained unaffected after 18 h and 24 h in culture. Taken together, our data prove that PEI/MNP based miR delivery is suitable for gentle and safe CD133 + stem cell engineering and for further preclinical investigation.

Conclusion
In this study, we developed a gentle and safe strategy for efficient modification of human CD133 + stem cells by miR. This strategy is based on the nonviral transfection system consisting of biotinylated PEI bound to streptavidin-coated iron oxide MNPs designed in our group. Optimized transfection complexes are suitable to reach high miR uptake rates (∼80-90%) without affecting CD133 + stem cell characteristics. In addition, using optimally supplemented supportive medium, cell viability remains unaffected after transfection. Furthermore, we demonstrated that modified cells can be magnetically guided to the site of interest in vitro. Hence, we expect our magnet-bead based miR delivery system to become an important tool for the engineering of stem cells prior to transplantation which can address certain challenges such as noninvasive cellular in vivo monitoring, low cell retention and initial cell death, and directed cell differentiation.

Ethical Approval
The authors state that they have obtained appropriate institutional review board approval or have followed the principles outlined in the Declaration of Helsinki for all human experimental investigations.

Consent
For investigations involving human subjects, informed consent has been obtained from the participants involved.