The presented data demonstrate a dose-dependent inhibition of chondrogenesis of hMSCs that have been labeled with the SPIO Ferucarbotran. Thus, we confirm results of other groups that described inhibition of chondrogenesis by magnetic labeling with the SPIO Ferumoxides [11
]. With our study, we evaluated several variables that may cause the previously described inhibition of chondrogenesis: intracellular iron quantity and presence or absence of extracellular membrane-bound iron. The systematic analysis of spectrometric, histopathologic and ultrastructural data on cellular Ferucarbotran uptake and metabolism led to the development of an improved labeling protocol (100μg Fe/ml for 4h), which does not significantly interfere with chondrogenesis. All labeling protocols led to a strong contrast agent effect on T2 and T2* images that did not decrease after 14 days of chondrogenic differentiation. The reagents for the proposed labeling procedures are in principle readily clinically applicable in Europe.
Adult hMSCs have been successfully used in vivo in preclinical models for cartilage engineering [27
]. In a human clinical trial, Centeno et al. have recently reported an increased cartilage volume after transplantation of autologous mesenchymal stem cells [29
]. For such applications, noninvasive imaging of the implanted cells would be highly desirable. Experimental studies in animal models showed that direct localization of the engrafted cells allows for early diagnosis of graft failure or cell migration. For example Jing et al. labeled hMSC with Ferumoxides and Protamine sulfate and were able to track such magnetically labeled hMSC by MR imaging over 12 weeks [30
]. They discovered by MR imaging that labeled cells did not home to the cartilage defect but migrated to the subchondral bone and the synovium. Thus, they proved the ability to detect graft failure by MR imaging of labeled cells. Interestingly, this finding on MR images was also confirmed for unlabeled cells by histology. Thus, a possibly adverse effect of magnetic labeling can’t be the reason for failure to engraft at the site of the cartilage defect. Jing et al suggest a trapping of the injected cells in the exogenous artificial carrier matrix as a reason for the failure to engraft properly. However, intracellular iron content was not measured and can’t be eliminated as an additional cause.
Maintained viability of iron oxide-labeled hMSCs is a mandatory prerequisite for any application of stem cells for cell tracking studies. Iron has a well known fundamental role in many cellular metabolic processes, including electron transport, deoxyribonucleotide synthesis, oxygen transport and essential redox reactions involving hemoproteins [31
]. The incorporation of iron oxides into the normal iron metabolism makes them appealing as contrast agents for cell labeling [33
]. But previous studies also showed that iron oxides can impair the viability of stem cells when they are internalized in too high quantities [13
]. However, if applied in limited concentrations, iron oxides are slowly incorporated into the regular iron metabolism and do not change the physiology of the cells [8
]. None of the labeling procotols proposed in this study showed any impairment of cell viability of labeled cells.
Another prerequisite for successful in vivo studies is an unimpaired differentiation capacity of labeled stem cells. Other groups reported normal differentiation of embryonic or neuronal stem cells into neurons after labeling with iron oxides [37
]. In addition, normal differentiation of hMSCs into chondrocytes [19
] and osteocytes [8
] was noted after labeling with Ferumoxides [8
] or FITC-labeled iron oxide nanoparticles [41
]. However another group of investigators reported impaired chondrogenesis of hMSCs after labeling with 50 μg Fe/ml Ferumoxides-polylysine for a prolonged incubation time of 24 hours [10
]. This may have been related to the high quantity of internalized contrast agent into the cells, since the same authors found limited chondrogenesis of their cells after incubation with half of the initial contrast agent concentration [11
]. Our results – although conducted with a different contrast agent partly confirm these results: we found impaired differentiation of hMSC with an excessive iron load and unimpaired differentiation of hMSCs into chondrocytes, when the cells were labeled with limited doses of iron oxides.
We used the negatively charged iron oxide Ferucarbotran because it is spontaneously endocytosed by hMSC and does not require the use of additional transfection agents [42
]. However, since Frank et al. reported no effect on the rate of differentiation after the use of the transfection agent Protamine sulfate, we decided to use Ferucarbotran in combination with Protamine sulfate in order to create a comparable experimental setup [19
]. Similar to studies with Ferumoxides [19
], our data shows that the cellular uptake of Ferucarbotran can be significantly increased by the addition of Protamine sulfate.
The analysis of cell morphology of mesenchymal stem cells undergoing chondrogenic differentiation led to the findings that the rate of spindled cells increased during the observed 14 days of chondrogenic differentiation. This is a paradox, because undifferentiated mesenchymal stem cells in vitro are spindle-shaped and chondrocytes in vivo are round. A metaanalysis of histological images in other studies confirmed our findings [11
], however, these studies did not analyse cell morphology but staining. Right after trypsinization and at the beginning of chondrogenic differentiation, mesenchymal stem cells are round with a high nuclear to cytoplasmatic ratio. This is not the appearance of mature resting chondrocytes, which typically have ample cytoplasm and a small, central nucleus. The difference becomes more evident when looking at undifferentiated cells (). During the course of differentiation, we saw an increase in the spindle cell component. We conclude that hMSCs are maturing from a completely round, primitive phenotype, to a more differentiated but still mesenchymal phenotype (with increased cytoplasm but spindled in shape). The cells form a matrix, which is a sign of an intermediate stage of chondrocyte differentiation. However, the cells do not reach the morphology of mature resting chondrocytes with low nuclear to cytoplasmic ratio. It is known, that in the growth plate, the proliferation zone has cells that get compressed and flattened in multicellular clusters [45
]. Overall, we conclude that our somewhat paradox finding that is, differentiating chondrocytes appear spindled might be due to the growth in a pellet.
For all labeling protocols using simple incubation, there was an inversely proportional relation between intracellular iron oxide concentration and the rate of chondrogenic differentiation. In fact, the protocol yielding the lowest uptake or iron (100μg Fe/4 hours) showed no significant difference in GAG-synthesis towards the unlabeled control. This supports the theory of a dose-dependent inhibition of chondrogenesis, similar to what Bulte et al. suspected for Ferumoxides-polylysine complexes [11
]. However, it has to be considered that iron uptake is inhomogenous and that the amount of internalized iron per cell varies to a certain degree [21
]. Therefore, even when using the same labeling protocol, the rate of differentiation might vary for each individual cell and results have to be considered valid for a certain population and not for each individual cell.
Our data showed surface-bound iron oxide particles immediately after cell labeling by simple incubation with our standard protocol, whereas there was no surface-bound iron for all other protocol variations. This finding explains why freshly labeled cells with the standard protocol tend to form clumps and are more difficult to resuspend after centrifugation when compared to unlabeled controls. It is known that chondrogenic differentiation depends highly on surface-linked cellular interactions and needs to be conducted in a 3D culture. It seems likely that surface-bound iron oxide particles could interfere with the essential mechanisms or structures.
Transfection with protamine sulfate yielded the highest iron uptake, about 3 times higher than for our standard protocol. Surprisingly, the total amount of GAG synthesized during differentiation was significantly higher than for the standard protocol (). This contradicts a dose-dependency as the only mechanism of the inhibition of hondrogenesis. We offer two possible explanations for this: (1) After labeling with Protamine sulfate, there were no surface bound Ferucarbotran particles that could inhibit surface-linked cell-cell interactions because the transfection agent shuttles the iron oxides into the intracellular compartment more efficiently. (2) Iron oxide filled lysosomes are larger in Protamine sulfate labeled cells than for the other protocols suggesting a different mechanism of uptake or intracellular compartmentalization, which in turn might cause less interaction with differentiation-linked intracellular structures or substrates.
We recognize several limitations to this study. This study was performed on a well-characterized, commercially available cell line of hMSCs, but theoretically, each cell line can behave differently and results might vary for other primary cultures. Although we were able to quantify adverse effects of magnetic labeling on stem cell physiology, the underlying biological mechanisms remain unclear. We confirm similar findings of other groups [10
] and show that the effects of magnetic labeling will have to be investigated in more detail; this will have to be studied with regards to molecular biology, biochemistry, cell physiology and cell morphology before labeling techniques can be integrated into clinical practice. Furthermore, long-term in vivo studies will have to address the biocompatibility of labeling techniques and the feasibility of this approach for clinical applications.