Iron oxide labelling of human mesenchymal stem cells in collagen hydrogels for articular cartilage repair

Available online at www.sciencedirect.com

Biomaterials 29 (2008) 1473e1483 www.elsevier.com/locate/biomaterials

Iron oxide labelling of human mesenchymal stem cells in collagen hydrogels for articular cartilage repair Andrea Heymer a, Daniel Haddad b,c, Meike Weber a, Uwe Gbureck d, Peter M. Jakob b,c, Jochen Eulert a, Ulrich No¨th a,* a

Division of Tissue Engineering, Orthopedic Center for Musculoskeletal Research, Ko¨nig-Ludwig-Haus, Julius-Maximilians-University of Wu¨rzburg, Brettreichstrasse 11, 97074 Wu¨rzburg, Germany b Research Center Magnetic Resonance Bavaria, 97074 Wu¨rzburg, Germany c Department of Physics, EPV, Julius-Maximilians-University of Wu¨rzburg, 97074 Wu¨rzburg, Germany d Department for Functional Materials in Medicine and Dentistry, Julius-Maximilians-University of Wu¨rzburg, 97070 Wu¨rzburg, Germany Received 4 September 2007; accepted 4 December 2007 Available online 21 December 2007

Abstract For the development of new therapeutical cell-based strategies for articular cartilage repair, a reliable cell monitoring technique is required to track the cells in vivo non-invasively and repeatedly. We present a systematic and detailed study on the performance and biological impact of a simple and efficient labelling protocol for human mesenchymal stem cells (hMSCs). Commercially available very small superparamagnetic iron oxide particles (VSOPs) were used as magnetic resonance (MR) contrast agent. Iron uptake via endocytosis was confirmed histologically with prussian blue staining and quantified by mass spectrometry. Compared with unlabelled cells, VSOP-labelling did neither influence the viability nor the proliferation potential of hMSCs. Furthermore, iron incorporation did not affect hMSCs in undergoing adipogenic, osteogenic or chondrogenic differentiation, as demonstrated histologically and by gene expression analyses. The efficiency of the labelling protocol was assessed with high-resolution MR imaging at 11.7 T. VSOP-labelled hMSCs were visualised in a collagen type I hydrogel, which is in clinical use for matrix-based articular cartilage repair. The presence of VSOP-labelled hMSCs was indicated by distinct hypointense spots in the MR images, as a result of iron specific loss of signal intensity. In summary, this labelling technique has great potential to visualise hMSCs and track their migration after transplantation for articular cartilage repair with MR imaging. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Mesenchymal stem cell; MRI (magnetic resonance imaging); Collagen hydrogel; SPIO nanoparticle; Cartilage tissue engineering

1. Introduction Cell-based therapies have shown great potential in modern orthopaedic surgery. Technologies such as autologous chondrocyte transplantation (ACT) for articular cartilage repair are already in clinical use for more than 10 years. Mesenchymal stem cells (MSCs) have received considerable attention in the field of cell therapy because of their ability to differentiate into various tissues of mesenchymal origin (e.g. bone, cartilage, fat, muscle, marrow stroma, tendon, ligament, and other * Corresponding author. Tel.: þ49 (0)931 803 0; fax: þ49 (0)931 803 1129. E-mail address: [email protected] (U. No¨th). 0142-9612/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2007.12.003

connective tissues) [1]. Recently, we have shown that human MSCs are able to undergo chondrogenic differentiation in a collagen type I hydrogel, which is in clinical use for matrix-based ACT [2,3]. The collagen hydrogel is currently tested for a future clinically approved MSC-based therapy. However, an important issue to successfully develop stem cell-based therapies is to analyse cell behaviour in vivo, in particular the localisation, proliferation, differentiation, and migration of the transplanted cells in the repair tissue. Today, cellular imaging is an established technique to monitor cell behaviour in vivo, which avoids invasive and irreversible tissue removal procedures. The main advantage of this non-invasive imaging technique is that it is repeatable [4].

1474

A. Heymer et al. / Biomaterials 29 (2008) 1473e1483

Among different imaging modalities, magnetic resonance imaging (MRI) is already in clinical use for injury and disease diagnosis on the anatomic scale. In combination with an efficient labelling method, MRI allows imaging at the cellular and even molecular level [5]. Superparamagnetic iron oxide (SPIO) particles can be used as such labels. They have been applied as magnetic contrast agents for over 20 years and lead to reduced signal intensities in T2*-weighted MR images by decreasing the transverse relaxation time constant T2*. SPIO particles are biocompatible because of their biodegradable iron oxide core and their appropriate surface coating, which allows intracellular incorporation [4]. SPIO-based imaging of macrophage activity was the initial and is still the most significant clinical application in this field, in particular for tumour staging of the liver and lymph nodes [6,7]. As a valuable tool in preclinical research, various non-phagocytic cells, including MSCs, were previously labelled in culture with SPIO particles [8e13]. The SPIO-labelling showed no negative influence during investigations of the viability and proliferation of magnetically labelled MSCs [14]. The differentiation of MSCs, however, especially the chondrogenic differentiation, has been controversially discussed in previous studies. No alteration in the differentiation capacity of human MSCs was demonstrated by Arbab et al. [15,16]. In contrast, Kostura et al. showed similar results concerning adipogenic and osteogenic differentiation, but a marked inhibition of chondrogenesis in human MSCs, using the same SPIO particles [17]. The particle size, as well as the surface coating material, has an influence on the cellular uptake of the particles [18e 20]. SPIO particles with a diameter more than 60 nm are incorporated to a higher degree in different cell lines than ultrasmall SPIO (USPIO) particles with a diameter of approximately 10e40 nm, using particles with carboxydextran as a coating material [19,20]. Dextran is the most common used coating, although a wide variety of SPIO particles with other coating materials have been developed showing different uptake efficiencies [5]. However, dextran-coated particles are only poorly incorporated into the cell because of a relatively inefficient fluid-phase endocytosis pathway. Therefore, several strategies have been developed to optimise SPIO particle uptake including the improvement of surface coating by functionalisation with specific ligands, such as antibodies, transferrin, or the membrane-translocating signal peptide HIV-1 Tat [22e24]. Another possibility to improve particle incorporation is the use of transfection agents, which form highly charged complexes with the SPIO particles, facilitating the interaction with anionic sites on the cell membrane thereby stimulating endocytosis [8,25e27]. The so-called very small superparamagnetic iron oxide particles (VSOPs) used in the present study have a total diameter of 11 nm including an iron oxide core of 5 nm [27]. They are coated with citrate bearing negative surface charges. Citrate is a natural occurring substance in mammals, which can be metabolised by them, thus providing high biocompatibility. In contrast to transfection agent-supported particle uptake, those anionic nanoparticles are probably incorporated by the cells

via adsorptive endocytosis, mediated by strong and nonspecific electrostatic interactions of the particles with cationic sites on the plasma membrane [18]. So far, VSOPs were successfully employed to label and monitor embryonic stem cells in the rat brain [28] and spleen-derived mononuclear cells in the ischemic mouse brain [29]. The aim of this study was to systematically evaluate the influence of magnetic labelling with VSOPs on human mesenchymal stem cells (hMSCs) in terms of viability, proliferation and differentiation capacity. We focused particularly on VSOPs because of their pronounced incorporation into cells compared to conventional dextran-coated USPIO particles [30]. In contrast to other SPIO particles, no addition of transfection agents is necessary, which are mostly not clinically approved and possibly capable to alter cell function. The small particle size of VSOPs should minimise possible negative effects on the stem cell function of hMSCs, especially on the chondrogenic differentiation capacity. Another important objective of this study was to visualise the cells in a matrix, which is in clinical use for articular cartilage repair. Therefore, high-resolution MRI of VSOP-labelled hMSCs embedded in a clinically used collagen type I hydrogel was performed. With this labelling technique, an important issue in the development of a stem cell-based therapy, which is the non-invasive assessment of the transplanted cell distribution and migration in the target tissue, can be analysed. 2. Materials and methods 2.1. Isolation and culture of human mesenchymal stem cells (hMSCs) After approval of the Institutional Review Board of the University of Wu¨rzburg, isolation of hMSCs was performed from the femoral head of patients undergoing total hip arthroplasty (all because of osteoarthritis) using a protocol first described by Haynesworth et al. [31], and modified by No¨th et al. [32]. Briefly, using a bone curet, trabecular bone plugs were harvested from the cutting plane of the femoral head and were transferred to 50 ml polypropylene conical tubes containing hMSC medium consisting of DMEM/Ham’s F-12 supplemented with 10% foetal bovine serum (FBS), 100 U/ml penicillin, 100 mg/ml streptomycin (all PAA, Linz, Austria), and 50 mg/ml L-ascorbic acid 2-phosphate. Marrow cells from the bone plugs were extracted and collected until the bone plugs appeared yellowish-white. The collected cells were resuspended in hMSC medium, counted with a haemocytometer, and plated at a density of 3  106 cells/cm2 in tissue culture flasks (Greiner BioOne, Frickenhausen, Germany). After 2e3 days of cultivation non-adherent cells were removed by aspiration with a pasteur pipette and attached cells were washed twice with phosphate-buffered saline (PBS). The culture medium (hMSC medium) was changed every 3e4 days. In all labelling experiments hMSCs were used directly after primary culture.

2.2. Labelling of hMSCs with very small superparamagnetic iron oxide particles (VSOPs) When the cells reached 70e80% confluency (after 10e14 days) they were used for labelling experiments with VSOPs C200 (Ferropharm, Teltow, Germany). Therefore, hMSC medium containing iron oxide particles at a concentration of 1.5 mM was added to the adherent cells and incubated for 90 min at 37  C and 5% CO2, according to the manufacturer’s instructions. No additional transfection agent was used. After incubation, hMSCs were

A. Heymer et al. / Biomaterials 29 (2008) 1473e1483 washed three times with PBS to remove any VSOPs not endocytosed by the cells. Adherent cells were detached with 0.25% trypsineEDTA (PAA, Linz, Austria), passed through a 70-mm cell filter to exclude cell clusters, counted with a haemocytometer, and used for further experiments.

2.3. Prussian blue staining and measurement of iron content Prussian blue staining for iron detection was performed from an aliquot of hMSCs for each labelling experiment. Cells were seeded on poly-L-lysine coated slides and allowed to adhere overnight before fixation with 4% phosphate-buffered paraformaldehyde. After washing with distilled water the cells were incubated with 1% potassium ferrocyanide in 1% hydrochloric acid for 30 min, washed again and counterstained with nuclear fast red for 5 min [33]. Quantification of the cellular iron content was performed with inductively coupled plasma-mass spectrometry (ICP-MS, Varian, Darmstadt, Germany) against standard solutions of 500 and 1000 ppm Fe in 0.65% nitric acid. The standard solutions were prepared from a multi-cation standard VI (Merck, Darmstadt, Germany) using a dilution factor of 100 and 200, respectively. The cell suspensions, with known cell density, were centrifuged at 250  g and the supernatant was removed. The remaining cell pellet was resuspended in 500 ml of 65% nitric acid and 10 ml 30% hydrogen peroxide to lyse the cells. Subsequently, the solution was diluted with deionised water to a final concentration of 0.65% nitric acid.

2.4. Cellular viability, apoptosis, and proliferation capacity The viability of VSOP-labelled and unlabelled cells was evaluated with a trypan blue dye exclusion test immediately after labelling. To detect if labelling with VSOPs led to cellular apoptosis, the CaspaseGloÒ 3/7 Assay (Promega, Mannheim, Germany) was used. This assay, based on luminescence, measures caspase-3 and -7 activities, key effectors in apoptosis in mammalian cells. VSOP-labelled and unlabelled cells were grown in 96-well culture plates for 24 h. After adding the Caspase-GloÒ 3/7 Reagent, the cells were incubated in the dark for 30 min at room temperature. The luminescence was measured with a plate-reading luminometer (Berthold Detection Systems, Pforzheim, Germany). The influence of VSOP-labelling on the cell proliferation capacity in longterm culture was analysed acquiring proliferation curves. Both unlabelled cells and VSOP-labelled cells were seeded at 2.3  103 cells/cm2 in tissue culture flasks. A cell count was performed after reaching 70e80% confluency at the end of each passage. Living cells were counted using trypan blue dye as a viability stain. From each passage 2.3  103 cells/cm2 were reseeded in tissue culture flasks and the procedure described above was repeated for six passages. The proliferation capacity was modelled by the standard exponential proliferation equation: N ¼ N0  2n 5logN ¼ logN0 þ n  log25n ¼

logN  logN0 log2

with N0 being the number of seeded cells on day 0 (4  105 cells) and N the number of harvested cells at the end of a passage. The cell doubling time was determined by dividing the total number of days in culture by the number of doublings (n).

2.5. Differentiation of hMSCs Both VSOP-labelled and unlabelled hMSCs were subjected to adipogenic, osteogenic, and chondrogenic differentiation, to determine whether magnetic labelling has any effect on their differentiation potential. For each lineage differentiation VSOP-labelled and unlabelled cells from three donors were used. On day 28 of cultivation, histological staining for extracellular matrix production was performed and total cellular RNA was extracted to assess gene expression of differentiated cells. Adipogenic differentiation was induced according to the method described by Pittenger et al. [34], and modified by No¨th et al. [35]. Confluent monolayer cultures of hMSCs were grown in adipogenic differentiation medium, consisting of DMEM high glucose (PAA, Linz, Austria), 10% FBS, 100 U/ml

1475

penicillin, and 100 mg/ml streptomycin which was supplemented with 100 mM insulin, 500 mM 3-isobutyl-1-methylxanthine, 1 mM dexamethasone, and 100 mM indomethacin. Cells maintained in hMSC medium served as a negative control. Medium changes were performed three times a week. Osteogenic differentiation was assessed in confluent monolayer cultures grown in DMEM high glucose containing 10% FBS, 100 U/ml penicillin, 100 mg/ml streptomycin, 50 mg/ml L-ascorbic acid 2-phosphate, 10 mM b-glycerophosphate, and 100 nM dexamethasone, according to Jaiswal et al. [36]. Cells maintained in hMSC medium served as negative controls. The medium was changed three times per week. Chondrogenic differentiation of hMSCs was induced using a high-density pellet cell culture system, modified from Johnstone et al. [37]. The cells were washed in serum-free chondrogenic differentiation medium consisting of DMEM high glucose, 100 U/ml penicillin, 100 mg/ml streptomycin, 50 mg/ml L-ascorbic acid 2-phosphate, 40 mg/ml proline, 100 mg/ml sodium pyruvate, 100 nM dexamethasone, and ITS-plus (final concentration: 10 mg/ml bovine insulin, 5.5 mg/ml transferrin, 5 mg/ml sodium selenite, 4.7 mg/ml linoleic acid, and 0.5 mg/ml bovine serum albumin). Aliquots of 250,000 cells were resuspended in chondrogenic differentiation medium, centrifuged at 250  g and 10 ng/ml TGF-b1 (R&D Systems, Wiesbaden, Germany) were added. Pellets maintained in chondrogenic differentiation medium without addition of TGF-b1 served as negative controls. Medium changes were performed twice a week.

2.6. Fabrication and culture of hMSC collagen hydrogels To visualise VSOP-labelled hMSCs in a three-dimensional matrix with the MR spectrometer, both, labelled and unlabelled cells were embedded individually in collagen type I hydrogels. The collagen type I stock solution, extracted from rat tails and dissolved in acetic acid at a concentration of 6 mg/ml collagen type I, was provided by Arthro Kinetics AG (Esslingen, Germany) together with a gel neutralisation solution consisting of HEPES-buffered, 3fold concentrated DMEM, and 30% FBS. Different concentrations of hMSCs (0.5  105, 1  105 and 1.3  105 cells/ ml gel) were embedded in collagen type I hydrogels with a final collagen concentration of 4 mg/ml. Briefly, the gel fabrication was carried out on ice by mixing two parts of collagen type I stock solution with one part of gel neutralisation solution: VSOP-labelled hMSCs were suspended in 0.5 ml gel neutralisation solution using three times the required cell density of the final gels. After adding 1 ml of collagen type I stock solution, the two components were carefully mixed to avoid air inclusions. The resulting collagenecell suspension was pipetted into custom-made Teflon inserts in 12-well culture plates. Polymerisation was allowed for 20 min at 37  C in a 5% CO2 humidified atmosphere. The resulting collagen type I hydrogels had a diameter of 17 mm and a thickness of 6 mm. After polymerisation, 3 ml of hMSC medium was added. Collagen hydrogels with unlabelled cells were used as controls. HMSC collagen hydrogels were cultivated for 5 days before MR imaging to allow air bubbles to disappear, which could give similar signal voids as the iron oxide particles.

2.7. High-field MR imaging MR imaging studies of hMSC collagen hydrogels were performed at 11.7 T on a 500-MHz Bruker Avance 500 MRI system (Bruker BioSpin GmbH, Rheinstetten, Germany) with a maximum gradient strength of 0.66 T/m. The MR images were acquired using a 3D FLASH sequence with the following parameters: echo time TE ¼ 20 ms, repetition time TR ¼ 250 ms (32 averages), and field-of-view (FOV) of 40  20  5 mm3. With a matrix size of 512  256  24 image points, this leads to a nominal spatial resolution of 78  78  208 mm3. During the image post-processing, a zero filling by a factor of 2 was applied in every dimension.

2.8. Histology To assess adipogenic differentiation, cells were stained for 10 min at room temperature with 0.3% oil red O, as an indicator for intracellular lipid accumulation, and counterstained with haematoxylin [34].

A. Heymer et al. / Biomaterials 29 (2008) 1473e1483

1476

Osteogenesis was determined using alizarin red S staining (1%) to detect calcium mineralisation [38]. Chondrogenic differentiation of pellet cultures was confirmed histologically after alcian blue staining for sulphated proteoglycans and immunohistochemical staining for collagen type II. The constructs were fixed for 2 h with 4% phosphate-buffered paraformaldehyde, dehydrated using a graded series of ethanol, infiltrated with amyl acetate, embedded in paraffin, and sectioned at a thickness of 4 mm. For histological staining, the sections were incubated in 1% alcian blue at pH 1.0 for 30 min and counterstained with nuclear fast red for 5 min [39]. Immunohistochemical analyses were performed with a primary antibody for collagen type II (Acris, Hiddenhausen, Germany) at a concentration of 714 ng/ml. The specimens were pre-digested with pepsin (1 mg/ml in TriseHCl, pH 2.0) for 15 min at room temperature, blocked for non-specific binding, and incubated with the primary antibody overnight at 4  C. Immunohistochemical staining was detected colorimetrically using BioGenex Super SensitiveÔ Link-Label IHC Detection System (DCS, Hamburg, Germany), which is based on alkaline phosphatase staining. The specimens were counterstained with haematoxylin. Negative control slides were incubated with mouse serum as a substitute for the primary antibody. To detect the iron oxide particles in the VSOP-labelled hMSCs embedded in collagen type I hydrogels, the constructs were fixed immediately after the MR experiments with 4% phosphate-buffered paraformaldehyde, dehydrated using a graded series of ethanol, infiltrated with amyl acetate, embedded in paraffin, and sectioned at a thickness of 4 mm. Prussian blue staining was performed as described above.

2.9. RNA isolation and RT-PCR analyses The total amount of cellular RNA was extracted on day 28 using Trizol reagent according to the manufacturer’s instructions. Further purification was performed using a RNA isolation kit and treatment with DNase I (Marcherey-Nagel, Du¨ren, Germany). The RNA was quantified spectrophotometrically and reverse transcribed using random hexamers (GE Healthcare, Munich, Germany) and Bioscript reverse transcriptase (Bioline, Luckenwalde, Germany). PCR amplification of cDNA was carried out using Taq DNA Polymerase (Bioline, Luckenwalde, Germany) and the human-specific primer sets listed in Table 1. These genes included the adipose-specific genes lipoprotein lipase (LPL) and peroxisome proliferator-activator receptor-g2 (PPARg2), the bonespecific genes alkaline phosphatase (ALP) and bone sialoprotein (BSP), and the cartilage-specific genes aggrecan (AGN) and collagen type II (Col II). The housekeeping gene elongation factor-1a (EF1a) was included to monitor RNA loading. The PCR products were electrophoretically separated on a 1.5%

agarose gel containing ethidium bromide and visualised using the Bio Profile software (LTF, Wasserburg, Germany). The intensity of the PCR products was analysed densitometrically with the Bio 1D/Capt MW software (LTF, Wasserburg, Germany) and the mean ratio (x-fold change) to the EF1a housekeeping gene was calculated.

2.10. Statistical analysis All quantitative data are expressed as mean  standard deviation (SD) and were verified by analysis of variance. The ManneWhitney test was applied to identify significant differences. P values of less than 0.05 were considered statistically significant.

3. Results 3.1. Prussian blue staining and iron content Confirming efficient labelling, prussian blue staining clearly showed the presence of various small iron oxide particles within the cytoplasm of VSOP-labelled hMSCs (Fig. 1A). No positive staining could be detected in unlabelled cells (Fig. 1B). Furthermore, ICP-MS confirmed the cellular uptake of iron oxide nanoparticles following VSOP-labelling and revealed a significant higher average iron concentration of 4.59  2.01 pg iron per cell (n ¼ 8, p < 0.001). The high standard deviation refers to relatively large differences in the iron content of labelled hMSCs from different donors. In contrast, unlabelled cells exhibited an iron content of 0.22  0.15 pg per cell, which corresponds to the sum of the natural intracellular iron content. 3.2. Cellular viability, apoptosis, and proliferation capacity of VSOP-labelled hMSCs Cellular viability, determined with a trypan blue dye exclusion test immediately after labelling, showed no significant difference. VSOP-labelled hMSCs exhibited a viability of

Table 1 PCR primer sets for gene expression analyses Gene Housekeeping gene EF1a

Primer sequences: sense/antisense

50 -AGGTGATTATCCTGAACCATCC-30 50 -AAAGGTGGATAGTCTGAGAAGC-30 Adipose-specific genes LPL 50 -GAGATTTCTCTGTATGGCACC-30 50 -CTGCAAATGAGACACTTTCTC-30 PPARg2 50 -GCTGTTATGGGTGAAACTCTG-30 50 -ATAAGGTGGAGATGCAGGCTC-30 Bone-specific genes OP 50 -ACGCCGACCAAGGAAAACTC-30 50 -GTCCATAAACCACACTATCAG-30 BSP 50 -AATGAAAACGAAGAAAGCGAAG-30 50 -ATCATAGCCATCGTAGCCTTGT-30 Cartilage-specific genes AGN 50 -GCCTTGAGCAGTTCACCTTC-30 50 -CTCTTCTACGGGGACAGCAG-30 Col II 50 -TTTCCCAGGTCAAGATGGTC-30 50 -CTTCAGCACCTGTCTCACCA-30

Product size (bp)

Position (bp)

Annealing temp. ( C)

235

1090e1324

54

276

1261e1536

51

351

128e478

51

483

140e622

51

450

620e1069

54

392

1814e2205

54

374

1498e1871

58

A. Heymer et al. / Biomaterials 29 (2008) 1473e1483

1477

Fig. 1. Representative prussian blue staining of VSOP-labelled hMSCs (A) and unlabelled cells (B). Intracytoplasmatic iron particles (blue) were only visible in the labelled cells. Bar ¼ 50 mm.

96.8  3.3%, unlabelled cells a viability of 97.2  3.4% (n ¼ 42). Therefore, we conclude that the viability was not influenced by the labelling procedure. Furthermore, VSOP-labelling of hMSCs did not result in increased apoptosis compared to unlabelled cells. The measured luminescent signal, which is proportional to the amount of caspase activity, showed similar levels in both, labelled and unlabelled cultures (n ¼ 5). Since VSOP-labelling did not lead to an observable influence on viability and/or apoptosis of hMSCs, we examined the cell proliferation of labelled and unlabelled cells for an extended culture period. No loss of proliferative activity was detectable for VSOP-labelled cells compared to control cells. The population doubling times, as determined over six passages of cultivation, showed no significant differences (n ¼ 8). 3.3. Differentiation of VSOP-labelled hMSCs Following adipogenic differentiation, oil red O staining at day 28 showed similar amounts of fat vacuoles in both, VSOP-labelled and unlabelled cells (Fig. 2A,B). Also, the gene expression levels of LPL and PPARg2 in VSOP-labelled hMSCs and unlabelled cells revealed no differences after appropriate stimulation (Fig. 3A). Cells maintained in hMSC medium as a negative control showed no intracellular fat vacuole staining (not shown). Osteogenic induced differentiation led to a strong matrix mineralisation in both cultures at day 28, as shown histologically by alizarin red S staining (Fig. 2C,D), which was not seen in negative control cells (not shown). In the presence of osteogenic supplements, VSOP-labelled and unlabelled cells revealed a similar gene expression pattern of ALP and BSP (Fig. 3B). Chondrogenic differentiation in high-density pellet cell cultures over 28 days exhibited chondrocyte-like cells embedded in a proteoglycan-rich extracellular matrix, as evident in the alcian blue staining (Fig. 2E,F). Immunohistochemical analysis showed a homogeneous and in most cases similar distribution of collagen type II in the extracellular matrix of both

groups. In two out of six differentiation experiments, however, a slightly more intense staining in the pellets of unlabelled cells was detected, as shown in Fig. 2G,H. This slight difference in staining intensity was also seen in the corresponding alcian blue staining of the pellets. On the other hand, the gene expression levels of AGN and Col II revealed in all analysed high-density pellet cell cultures no difference between VSOP-labelled and unlabelled hMSCs (Fig. 3C). Control pellet cultures cultivated without TGF-b were considerably smaller and showed no evidence of chondrogenic matrix accumulation (not shown). 3.4. MR imaging of VSOP-labelled cells in collagen type I hydrogels To visualise VSOP-labelled hMSCs in collagen type I hydrogels, high-resolution MR imaging at 11.7 T was performed. Hydrogels containing different concentrations of labelled hMSCs were cultivated for 5 days and imaged subsequently (Fig. 4). The presence of iron oxide particles was indicated by distinct hypointense spots in the MR images. In comparison, control collagen gels with unlabelled cells showed a homogeneous appearance with no iron specific loss of signal intensity. Isolated dark spots in these control gels and on their surface were attributed to the presence of air bubbles. 3.5. Histological analysis of imaged collagen type I hydrogels Histological analysis of the imaged hydrogels revealed a homogeneous cell distribution (Fig. 5). Confirming the MR data, prussian blue staining demonstrated the presence of iron oxide particles within the VSOP-labelled hMSCs, whereas no positive staining could be detected in control collagen gels. 4. Discussion The development of new clinical cell therapies for articular cartilage repair requires methods to monitor these cells

1478

A. Heymer et al. / Biomaterials 29 (2008) 1473e1483

Fig. 2. Differentiation of VSOP-labelled and unlabelled hMSCs. Oil red O staining revealed lipid vacuoles in labelled (A) and control hMSCs (B), as a result of adipogenic differentiation. Osteogenic differentiation was not affected by labelling with VSOPs (C) as compared to control cells (D). Alizarin red S staining showed calcium deposition in both cultures. Following chondrogenic differentiation, alcian blue staining revealed a proteoglycan-rich extracellular matrix in labelled (E) and unlabelled pellets (F). Cartilage-specific collagen type II staining was observed in both, VSOP-labelled (G) and unlabelled cultures (H), but more pronounced in the latter one. (AeD) Bar ¼ 100 mm and (EeH) bar ¼ 50 mm.

A. Heymer et al. / Biomaterials 29 (2008) 1473e1483

1479

Fig. 3. Gene expression of differentiated VSOP-labelled and unlabelled hMSCs. (A) Analysis of adipogenic differentiation showed the presence of LPL and PPARg mRNA. Evaluation by densitometry exhibited no difference in the gene expression levels between adipogenic induced labelled and unlabelled cells. (B) In cultures with osteogenic differentiation medium, mRNA expression of ALP and BSP was clearly detectable in both cultures. Analysis of the intensity of PCR products showed no differences in the gene expression levels of VSOP-labelled and unlabelled cells. (C) Expression of AGN and Col II mRNA was visible in both, VSOP-labelled and unlabelled pellet cultures. No difference in the gene expression levels of these cartilage-specific genes could be evaluated.

non-invasively and repeatedly. MR imaging of cells, labelled with a MR contrast agent in vitro prior to transplantation, offers a possibility to accomplish this task. In the study presented here, we were able to adapt an efficient and robust labelling protocol for human MSCs and evaluate its biological impact and labelling performance. Furthermore, we visualised the labelled cells in a collagen type I hydrogel, which is in

clinical use for articular cartilage repair. As MR contrast agent, commercially available very small superparamagnetic iron oxide particles were used. These VSOPs have been developed as a blood pool contrast agent for MR angiography and were recently evaluated in a phase I clinical trial [40]. The development of labelling procedures with commercially available and clinically approved MR contrast agents allows

1480

A. Heymer et al. / Biomaterials 29 (2008) 1473e1483

Fig. 4. MR images of hMSCs in collagen type I hydrogels. VSOP-labelled cells showed the typical signal decrease due to iron particles, present as dark spots. Control collagen gels with unlabelled cells showed a homogeneous appearance without hypointense spots. 3D FLASH experiment, nominal spatial resolution: 78  78  208 mm3. TE ¼ 20 ms and TR ¼ 250 ms.

a fast translation of these methods into clinical practice. In contrast to various previously described labelling methods [8,21,24], cell labelling with VSOPs is very quick and easy to apply since it needs only an incubation time of 90 min and no additional reagents and techniques with their possibly adverse biological effects.

In order to assure the MR detectability of transplanted cells, the applied labelling protocol has to assure a sufficient iron uptake. The iron content of the VSOP-labelled hMSCs was determined with ICP-MS and yielded an average amount of 4.59  2.01 pg iron per cell. This is a relatively low value, compared to other studies described in the literature which

Fig. 5. Histological analysis of imaged collagen type I hydrogels (cell concentration of 1  105 cells/ml gel) confirmed iron containing single cells (stained for prussian blue), homogeneously distributed in the construct (A). The inset in (A) shows a higher magnification (bar ¼ 20 mm) of a single cell containing blue stained iron particles. Control collagen gels with unlabelled cells showed no positive staining (B). Bar ¼ 100 mm.

A. Heymer et al. / Biomaterials 29 (2008) 1473e1483

found up to 64.5 pg iron per cell [10,14], depending on the contrast agent, cell type, and labelling protocol used [8]. Nevertheless, the labelled cells could easily be detected in the MR images. This was confirmed by histological analysis with iron specific prussian blue staining. The variability in the iron concentration per cell in the present study may be due to a varying endocytotic activity of hMSCs from different donors. According to Stroh et al., the low iron content can be attributed to the relatively low VSOP concentration (82 mg of iron/ml) used for the labelling procedure [41]. In the context of efficient cell labelling, the effect of high intracellular iron content on the metabolism of the cells has to be considered. It is well known, that an intracellular iron overload can induce free radical formation and oxidative stress, which may also lead to cytotoxicity and cell death [42]. A transient increase of oxidative stress in rat macrophages immediately after labelling with VSOPs was indeed ascertained, but 24 h after incubation, this induction of oxidative stress could no longer be observed [41]. Supporting these results, we assessed no increased rate of apoptosis 24 h after labelling hMSCs with low VSOP concentrations. Furthermore, proliferation assays in long-term cell culture studies indicated no cytotoxic effects due to the presence of intracellular iron oxide particles neither in our study, nor in a study by Stroh et al. [41]. Moreover, these results are consistent with those reported in other studies using different iron oxide particles [14,24,33,43]. Concerning the amount of intracellular iron during the proliferation of labelled cells, several studies have already shown the dilution of the magnetic label upon cell division [13,19,33,43]. Prussian blue staining still reveals homogeneous staining of all cells, but fewer iron within each cell, suggesting that the iron oxide particles are split between the daughter cells [43]. We have tracked the presence of iron in VSOP-labelled hMSCs histologically over several passages in monolayer culture and observed a distinct reduction of VSOP-labelling after 5e9 cell divisions, relating to passages 2e3. Loss of label is certainly also related to the biodegradation of SPIO particles and the subsequent recycling of iron by the cells. Regarding long-term tolerability, in a study by Arbab et al., a long-term retention of intracellular iron in nondividing hMSCs over 43 days with no increase in the production of reactive oxygen species was revealed, suggesting that the iron was not released into the cytoplasm as a free radical form [44]. In our study, however, we only determined the time sequence of VSOP-labelling reduction, but we didn’t investigate the underlying mechanisms of iron loss. Human MSCs are good candidates for future matrix-based cartilage repair technologies, where they are expected to differentiate into chondrocytes and regenerate missing cartilage tissue. Therefore, a possible adverse effect of the labelling procedure on their function has to be excluded. Since previous studies assessed the influence of magnetic labelling on the differentiation ability of MSCs predominantly by histological and immunohistochemical analyses, RT-PCR analyses were performed in our study to evaluate possible changes also on the mRNA level. Concerning adipogenic and osteogenic differentiation, the results are in accordance with previous studies

1481

[15,16,43]. Both VSOP-labelled and unlabelled cells showed similar matrix staining after appropriate stimulation. These qualitative results were confirmed by semi-quantitative analyses of the gene expression levels. This suggests that iron incorporation did not affect mesenchymal stem cells when undergoing these differentiation pathways. The chondrogenic differentiation of hMSCs has been discussed controversially in previous studies, which present opposing results about the influence of magnetic labelling. At first, Arbab et al. [15] reported no effect of FeridexÒ-labelling on chondrogenesis of hMSCs. However, a critical revision of this publication [45] assumed the failure of chondrogenic differentiation, based on the lack of glycosaminoglycan and collagen type II staining. In contrast, Kostura et al. [16] reported a marked inhibition of chondrogenesis in hMSCs following FeridexÒ-labelling and referred this inhibition to the iron oxide particles themselves, rather than to the transfection agent. Both studies used the same MR contrast agent (FeridexÒ, particle size of 80e150 nm), but different transfection agents. Arbab et al. [17] demonstrated in a second publication again no inhibition of chondrogenic activity in both, FeridexÒlabelled and unlabelled hMSCs. The authors speculated that the failure of chondrogenic differentiation in the study by Kostura et al. [16] referred to residual extracellular iron particle complexes on the surface of the cells. We used a different type of SPIO particles and investigated their influence on the chondrogenic differentiation ability of hMSCs. The results showed a very slight inhibition of cartilage-specific matrix accumulation in two out of six VSOP-labelled pellet cultures, as evidenced histologically, but no differences in the semiquantitative analyses of the AGN and Col II gene expression could be detected. One could speculate that the reason for the inhibition at the protein level only but not at the mRNA level might be related to degradation products, released from VSOPs in endosomes, which interacts with enzymes involved in the protein synthesis in the cytoplasm of the cells. In this study, we have deliberately chosen only one labelling condition, based on findings of previous studies, which used the same particles. It was shown that labelling of rat macrophages with a 2-fold higher VSOP concentration led to a pronounced increase in oxidative stress compared to our conditions [41]. Furthermore, no influence on the viability and differentiation capacity of embryonic stem cells was assessed, when using our labelling protocol [28]. These findings led us to suppose that a small particle size and/or a low iron content of the cells may minimise possible negative effects on the stem cell function of hMSCs, especially on the chondrogenic differentiation capacity. Moreover, the approximately 10-fold smaller particle size of VSOPs compared to FeridexÒ and the lower iron content of the labelled cells are two characteristics of this particular labelling technique, which is quite different form the study of Kostura et al., who showed a marked inhibition of chondrogenic differentiation of hMSCs [16]. So far we cannot assess if the absence of a marked inhibition of differentiation is related to either the small particle size, the surface coating material, the low iron content of the cells, or the absence of a transfection agent. To understand

1482

A. Heymer et al. / Biomaterials 29 (2008) 1473e1483

the mechanism of a possible impact of SPIO-labelling on stem cell function, comparative studies with different SPIO particles and/or increasing iron internalisation have to be performed. After we had ascertained no adverse influence of VSOPlabelling on the differentiation of hMSCs, we were ready to visualise the labelled cells with MRI. In many tissue engineering approaches, especially in the field of orthopaedics, cells are transplanted in a three-dimensional scaffold to achieve a homogeneous distribution of the cells within the defect. Therefore, the ability to track cells in a matrix was mandatory for the development of a successful monitoring technique. In the present study hMSCs were embedded in a collagen type I hydrogel, which is already in clinical use for matrix-based ACT [3]. In a first step, we were able to visualise different concentrations of VSOP-labelled hMSCs in the hydrogel using high-resolution MRI at 11.7 T. This allows us to state that a sufficient amount of VSOPs was incorporated in the cells to act as an efficient cellular label for MRI experiments. In further studies, the correlation of MR monitoring results with findings from conventional histology has to reveal, whether a single hypointense spot represents a single cell or small cell clusters. These experiments should help to find a threshold cell concentration for an easy and distinct identification of labelled hMSCs in collagen hydrogels. An additional important prerequisite for the translation into clinical practice is the detection of labelled MSC-laden hydrogels on a clinical whole body MR scanner at 1.5 T, which is the currently used clinical standard. Taken together, these investigations should reveal, whether a monitoring technique relying on MRI only is feasible to track cells after transplantation in the repair tissue. 5. Conclusions Our study demonstrates a simple and efficient labelling protocol for human mesenchymal stem cells applying VSOPs as a commercially available MR contrast agent. By using a short incubation time and a low iron concentration, we achieved a sufficient intracellular magnetic label for cell imaging without significantly influencing cell metabolism. No effect on the cell viability, apoptosis, proliferation, and differentiation potential of hMSCs could be ascertained. In contrast to other studies, the chondrogenic differentiation capacity of hMSCs was not affected. Furthermore, the technology allows the visualisation of labelled stem cells in a clinically used collagen type I hydrogel, which is an important feature for monitoring cell-based therapies in the field of articular cartilage tissue engineering. Acknowledgments The authors thank Arthro Kinetics AG (Esslingen, Germany) for providing the collagen type I stock solution and the gel neutralisation solution. Viola Monz, Martina Regensburger, Christa Amrehn, and Nadja Karl are acknowledged for their excellent technical assistance. We also thank Dr. Nicole Wollmerstedt for performing the statistical analyses.

This work was supported by the Interdisciplinary Centre for Clinical Research (IZKF) of the University of Wu¨rzburg and the Bavarian Ministry of Economic Affairs, Infrastructure, Transport and Technology. References [1] Caplan AI. Mesenchymal stem cells. J Orthop Res 1991;9:641e50. [2] No¨th U, Rackwitz L, Heymer A, Weber M, Baumann B, Steinert A, et al. Chondrogenic differentiation of human mesenchymal stem cells in collagen type I hydrogels. J Biomed Mater Res A 2007;83:626e35. [3] No¨th U, Siebenlist S, Rackwitz L, Schreiber B, Steinert A, Barthel T, et al. Matrix-based autologous chondrocyte transplantation for the treatment of large osteochondral defects. In: McKenna J, editor. European musculoskeletal review. London: Touch Briefings; 2006. p. 62e4. [4] Bulte JWM, Kraitchman DL. Iron oxide MR contrast agents for molecular and cellular imaging. NMR Biomed 2004;17:484e99. [5] Thorek DLJ, Chen AK, Czupryna J, Tsourkas A. Superparamagnetic iron oxide nanoparticle probes for molecular imaging. Ann Biomed Eng 2006;34:23e38. [6] Harisinghani MG, Jhaveri KS, Weissleder R, Schima W, Saini S, Hahn PF, et al. MRI contrast agents for evaluating focal hepatic lesions. Clin Radiol 2001;56:714e25. [7] Anzai Y, Piccoli CW, Outwater EK, Stanford W, Bluemke DA, Nurenberg P, et al. Evaluation of neck and body metastases to nodes with ferumoxtran 10-enhanced MR imaging: phase III safety and efficacy study. Radiology 2003;228:777e88. [8] Arbab AS, Bashaw LA, Miller BR, Jordan EK, Bulte JWM, Frank JA. Intracytoplasmatic tagging of cells with ferumoxides and transfection agent for cellular magnetic resonance imaging after cell transplantation: methods and techniques. Transplantation 2003;76:1123e30. [9] Frank JA, Anderson SA, Kalsih H, Jordan EK, Lewis BK, Yocum GT, et al. Methods for magnetically labeling stem and other cells for detection by in vivo magnetic resonance imaging. Cytotherapy 2004;6:621e5. [10] Bos C, Delmas Y, Desmoulie`re A, Solanilla A, Hauger O, Grosset C, et al. In vivo MR imaging of intravascularly injected magnetically labeled mesenchymal stem cells in rat kidney and liver. Radiology 2004;233:781e9. [11] Jendelova´ P, Herynek V, DeCroos J, Glogarova´ K, Andersson B, Ha´jek M, et al. Imaging the fate of implanted bone marrow stromal cells labeled with superparamagnetic nanoparticles. Magn Reson Med 2003;50:767e76. [12] Daldrup-Link HE, Rudelius M, Piontek G, Metz S, Bra¨uer R, Debus G, et al. Migration of iron oxide-labeled human hematopoietic progenitor cells in a mouse model: in vivo monitoring with 1.5-T MR imaging equipment. Radiology 2005;234:197e205. [13] Rivie`re C, Boudghe`ne FP, Gazeau F, Roger J, Pons J-N, Laissy J-P, et al. Iron oxide nanoparticle-labeled rat smooth muscle cells: cardiac MR imaging for cell graft monitoring and quantitation. Radiology 2005; 235:959e67. [14] Ju S, Teng G, Zhang Y, Ma M, Chen F, Ni Y. In vitro labeling and MRI of mesenchymal stem cells from human umbilical cord blood. Magn Reson Imaging 2006;24:611e7. [15] Arbab AS, Yocum GT, Kalish H, Jordan EK, Anderson SA, Khakoo AY, et al. Efficient magnetic cell labeling with protamine sulfate complexed to ferumoxides for cellular MRI. Blood 2004;104:1217e23. [16] Arbab AS, Yocum GT, Rad AM, Khakoo AY, Fellowes V, Read EJ, et al. Labeling of cells with ferumoxideseprotamine sulfate complexes does not inhibit function or differentiation capacity of hematopoietic or mesenchymal stem cells. NMR Biomed 2005;18:553e9. [17] Kostura L, Kraitchman DL, Mackay AM, Pittenger MF, Bulte JWM. Feridex labeling of mesenchymal stem cells inhibits chondrogenesis but not adipogenesis or osteogenesis. NMR Biomed 2004;17:513e7. [18] Wilhelm C, Billotey C, Roger J, Pons JN, Bacri J-C, Gazeau F. Intracellular uptake of anionic superparamagnetic nanoparticles as a function of their surface coating. Biomaterials 2003;24:1001e11.

A. Heymer et al. / Biomaterials 29 (2008) 1473e1483 [19] Sun R, Dittrich J, Le-Huu M, Mueller MM, Bedke J, Kartenbeck J, et al. Physical and biological characterization of superparamagnetic iron oxide- and ultrasmall superparamagnetic iron oxide-labeled cells. Invest Radiol 2005;40:504e13. [20] Matuszewski L, Persigehl P, Wall A, Schwindt W, Tombach B, Fobker M, et al. Cell tagging with clinically approved iron oxides: feasibility and effect of lipofection, particle size, and surface coating on labeling efficiency. Radiology 2005;235:155e61. [21] Gupta AK, Gupta M. Synthesis and surface engineering of iron oxide nanoparticles for biomedical applications. Biomaterials 2005;26:3995e4021. [22] Lewin M, Carlesso N, Tung C-H, Tang X-W, Cory D, Scadden DT, et al. Tat peptide-derivatized magnetic nanoparticles allow in vivo tracking and recovery of progenitor cells. Nat Biotechnol 2000;18:410e4. [23] Ahrens ET, Feili-Hariri M, Xu H, Genove G, Morel PA. Receptormediated endocytosis of iron-oxide particles provides efficient labeling of dendritic cells for in vivo MR imaging. Magn Reson Med 2003;49: 1006e13. [24] Bulte JWM, Douglas T, Witwer B, Zhang S-C, Strable E, Lewis BK, et al. Magnetodendrimers allow endosomal magnetic labeling and in vivo tracking of stem cells. Nat Biotechnol 2001;19:1141e7. [25] Frank JA, Miller BR, Arbab AS, Zywicke HA, Jordan EK, Lewis BK, et al. Clinically applicable labeling of mammalian and stem cells by combining superparamagnetic iron oxides and transfection agents. Radiology 2003;228:480e7. [26] Montet-Abou K, Montet X, Weissleder R, Josephson L. Cell internalization of magnetic nanoparticles using transfections agents. Mol Imaging 2007;6:1e9. [27] Pilgrimm H. Superparamagnetic particles with increased R1 relaxivity, process for producing said particles and use thereof. Patent no. US6638494; 2003. [28] Stroh A, Faber C, Neuberger T, Lorenz P, Sieland K, Jakob PM, et al. In vivo detection limits of magnetically labeled embryonic stem cells in the rat brain using high-field (17.6 T) magnetic resonance imaging. NeuroImage 2005;24:635e45. [29] Stroh A, Zimmer C, Werner N, Gertz K, Weir K, Kronenberg G, et al. Tracking of systemically administered mononuclear cells in the ischemic brain by high-field magnetic resonance imaging. NeuroImage 2006;33:886e97. [30] Fleige G, Seeberger F, Laux D, Kresse M, Taupitz M, Pilgrimm H, et al. In vitro characterization of two different ultrasmall iron oxide particles for magnetic resonance cell tracking. Invest Radiol 2002;37:482e8. [31] Haynesworth SE, Goshima J, Goldberg VM, Caplan AI. Characterization of cells with osteogenic potential from human marrow. Bone 1992;13:81e8. [32] No¨th U, Tuli R, Osyczka AM, Danielson KG, Tuan RS. In vitro engineered cartilage constructs produced by press-coating biodegradable

[33]

[34]

[35]

[36]

[37]

[38]

[39] [40]

[41]

[42] [43]

[44]

[45]

1483

polymer with human mesenchymal stem cells. Tissue Eng 2002; 8:131e44. Ittrich H, Lange C, Dahnke H, Zander AR, Adam G, Nolte-Ernsting C. Labeling of mesenchymal stem cells with different superparamagnetic particles of iron oxide and detectability with MRI at 3T. Fortschr Ro¨ntgenstr 2005;177:1151e63. Pittenger MF, Mackay AM, Beck SC, Jaiswal RK, Douglas R, Mosca JD, et al. Multilineage potential of adult human mesenchymal stem cells. Science 1999;284:143e7. No¨th U, Osyczka AM, Tuli R, Hickok NJ, Danielson KG, Tuan RS. Multilineage mesenchymal differentiation potential of human trabecular bone-derived cells. J Orthop Res 2002;20:1060e9. Jaiswal N, Haynesworth SE, Caplan AI, Bruder SP. Osteogenic differentiation of purified, culture-expanded human mesenchymal stem cells in vitro. J Cell Biochem 1997;64:295e312. Johnstone B, Hering TM, Caplan AI, Goldberg VM, Yoo JU. In vitro chondrogenesis of bone marrow-derived mesenchymal progenitor cells. Exp Cell Res 1998;238:265e72. La˛czka-Osyczka A, La˛czka M, Kasugai S, Ohya K. Behavior of bone marrow cells cultured on three different coatings of gel-derived bioactive glass-ceramics at early stages of cell differentiation. J Biomed Mater Res 1998;42:433e42. Lev R, Spicer SS. Specific staining of sulphate groups with alcian blue at low pH. J Histochem Cytochem 1964;12:309. Taupitz M, Wagner S, Schnorr J, Kravec I, Pilgrimm H, BergmannFritsch H, et al. Phase I clinical evaluation of citrate-coated monocrystalline very small superparamagnetic iron oxide particles as a new contrast medium for magnetic resonance imaging. Invest Radiol 2004;39: 394e405. Stroh A, Zimmer C, Gutzeit C, Jakstadt M, Marschinke F, Jung T, et al. Iron oxide particles for molecular magnetic resonance imaging cause transient oxidative stress in rat macrophages. Free Radic Biol Med 2004;36:976e84. Emerit J, Beaumont C, Trivin F. Iron metabolism, free radicals, and oxidative injury. Biomed Pharmacother 2001;55:333e9. Terrovitis JV, Bulte JWM, Sarvananthan S, Crowe LA, Sarathchandra P, Batten P, et al. Magnetic resonance imaging of ferumoxide-labeled mesenchymal stem cells seeded on collagen scaffolds e relevance to tissue engineering. Tissue Eng 2006;12:2765e75. Arbab AS, Bashaw LA, Miller BR, Jordan EK, Lewis BK, Kalish H, et al. Characterization of biophysical and metabolic properties of cells labelled with superparamagnetic iron oxide nanoparticles and transfection agent for cellular MR imaging. Radiology 2003;229:838e46. Bulte JWM, Kraitchman DL, Mackay AM, Pittenger MF. Chondrogenic differentiation of mesenchymal stem cells is inhibited after magnetic labeling with ferumoxides. Blood 2004;104:3410e2.