Whole body tracking of superparamagnetic iron oxide nanoparticle-labelled cells – a rheumatoid arthritis mouse model

Current tissue engineering approaches focusing on restoring and regenerating articular cartilage damage are limited to the damage caused by trauma and osteoarthritis [1]. The chronic inflammatory environment of the rheumatic arthritic joint renders these techniques ineffective, as similarly to the original native cartilage, newly formed cartilage will again undergo destruction within the hostile environment [1]. Rheumatoid arthritis (RA, a chronic autoimmune disease) is characterised by pain, stiffness and inflammation of the synovial joint [1–3]. This results in the destruction of articular cartilage and affects approximately 1% of the global population [1, 2, 4, 5]. Current RA treatments involve a combination of drug regimens to alleviate symptoms, such as pain and inflammation, while preserving joint function and maintaining quality of life [1, 5]. Few patients have experienced complete drug free remission with little progress being made in restoring joint function and regenerating cartilage [1, 5, 6].

Advances in tissue engineering have emphasised the role of mesenchymal stem cells (MSCs) in treating autoimmune diseases, such as RA [1, 2, 7]. Their specific self-renewal, multipotent differentiation ability (osteoblasts, chondrocytes and adipocytes), migratory, anti-inflammatory and immunosuppresssive properties are all key characteristics linked to their success in stem cell-based therapies [1, 2, 8–10]. These are modulated by the secretion of bioactive molecules. The immunosuppressive properties of MSCs are of particular importance in treating autoimmune diseases, such as RA [1]. The release of cytokines and growth factors, such as IL-10, IL-6, IL-11 and transforming growth factor – β (TGF-β), acts to inhibit T cells and dendritic cells [7, 11] while the secretion of soluble antigens, such as human leukocyte antigen G (HLA-G), effectively disables natural killers and moderate dendritic cell and T cell activity. In addition, secreted immunosuppressive enzymes, such as indoleamine 2, 3-dioxygenase (IDO), suppress leukocytes such as B cells [7, 11]. The combined secretion of these factors, their role in tissue homeostasis and repair (governed by a signalling mechanism) [2] and the cartilage forming ability of MSCs provides a trophic regenerative environment, stimulating the proliferation and differentiation of tissues to achieve intrinsic repair while protecting the neo tissue in a localised immunosuppressive manner [1, 7, 11].

Very little is known of the in vivo events occurring post implantation. A means of imaging and tracking implanted MSCs could prove extremely valuable in evaluating and optimising mechanisms of cartilage repair within an inflammatory environment. Information linked to cell migration [12], rate of repair [12] and tissue integration are pivotal in optimising the therapy in terms of cell number [13], cell dose [12], dosage schemes [14] and delivery methods [12]. Traditional means of gathering such data have relied on histological tests on sacrificed animals [15–17]. This tends to be invasive and information limited. The shortcomings of these techniques render them unacceptable in assessing the success of the cellular therapies [15–17].

In response to this, superparamagnetic iron oxide nanoparticles (SPIONs) can be employed in conjunction with the use of magnetic resonance imaging (MRI) to track implanted cells in vivo[16, 18, 19]. To date, little work has been carried out to optimise SPION concentrations for specific tissue applications in vivo. To this end, we have optimised towards the current clinical MRI modalities commonly found in hospitals, these being 1.5 tesla (T) and 3 T scanners. In essence, stem cells are encouraged to internalise SPIONs; through this method, the magnetic properties of the particles are transferred to the cells. The intracellular iron disturbs the local magnetic field thus allowing cells to be visualised as a lack of signal with MRI [19–25].

Here we have adopted a murine model of RA to investigate the immunomodulating and anti-inflammatory properties of mouse MSCs (mMSCs) labelled with and without SPIONs and to optimise the in vivo imaging and tracking protocol. To our knowledge, this is the first time MSCs have been tracked using SPIONS within a rheumatic joint in this manner.

In response to the lack of pharmacological interventions to treat RA, cell therapies have been developed, offering new opportunities in tackling this disease. Stem cells, such as MSCs, have been used to regenerate and restore the function of damaged tissue, such as cartilage [7]. The striking ability of MSCs to migrate toward the site of injury, engraft with surrounding tissues [28] and differentiate into cartilage [29] whilst suppressing the immune system, highlights their applicability as a possible therapy for RA [2]. Despite the vast number of studies demonstrating the potential of MSCs in treating RA, the clinical translation and commercialisation of this therapy is hindered by additional unknown factors [7]. A non-invasive online method of monitoring the success of the therapy in terms of cellular bio distribution [30] and cell survival [31] is key in deciphering the mechanism of repair, evaluating the therapeutic effect, grafting location and ruling out potentially dangerous side effects [17]. This would encourage regulatory approval to be obtained and therapeutic potential optimised for clinical applications [28].

The precedent for labelling cells with magnetic nanoparticles (MNPs) or SPIONS comes from the technology employed in MRI and MRI contrast agents and offers an attractive way of monitoring cells in vivo[28, 32–34]. SPIONs are essentially composed of a magnetite (Fe₃O₄) or maghemite (γ-Fe₂O₃) core coated with a biocompatible polymer [35]. The range of particles commercially available is vast with particles ranging in size from the nanometer to the micron scale [17, 34] and differing in polymer coating and surface chemistries. An extensive range of studies utilising an assortment of MNPs have been used for a variety of tissue engineering studies. There is no clear consensus as to the ideal MNP size, concentration or magnet strength that would yield the greatest contrast in vivo. As previously mentioned, 1.5 and 3 T scanners are commonly found in clinical practise today. With this in mind, it is essential to design and optimise imaging and tracking protocols in terms of cell number and particle concentration to these clinically relevant scanners. Particle concentration ranging from 12.5 to 400 μg/ml have been reportedly used in studies focusing on 1.5 T scanners [16, 22, 28] while 2.8 to 168 μg/ml have been reported in studies using 3 T scanners [16, 24, 31]. In this study, a particle concentration of 10 μg/ml has been implemented within a 2.3 T scanner highlighting the clinical relevance of our system.

In recent years, a number of studies focusing on the development of strategies to image and track stem cells within the articular joint have emerged. These studies tend to adopt invasive surgical models of osteoarthritis in either rabbits or rats where cells (MSCs or chondrocytes) are labelled with MNPs, seeded onto scaffolds, such as collagen type I gel [25] or agarose [36], and implanted within osteochondral defects. Animals are then monitored over a period ranging from four to twelve weeks while being MR imaged [13, 25, 36, 37]. Particles used in these studies have ranged from very small superparamagnetic iron oxide nanoparticles (11 nm) [25] to Food and Drug Administration approved particles such as Endorem (150 nm) [37] and Feraheme (30 nm) [36]. Contrary to the above group of studies where MNP ranged in size between 11 nm and 150 nm in diameter, Saldanha et al. investigated the use of micron sized particles (1.63 μm) as a potential tracking agent to monitor cartilage regeneration. Rabbit MSC were labelled with a commercially available particle and implanted within an ex vivo knee model (bovine) and MR imaged using 3 T to monitor articular cartilage regeneration [19].

From these studies, it is evident that tracking studies have primarily focused on the rat or rabbit models where surgical techniques are used to mimic trauma or osteoarthritis. In addition, cells have generally been labelled with a maximum particle size of 150 nm and seeded onto hydrogel scaffolds prior to implantation. The end point of such animal studies is often histological assessment. Animal models, however, also allow for the investigation of functional outcomes, such as the presence of inflammation. Very few stem cell related studies actually monitor histopathological features while assessing functional outcomes and tracking cells in vivo.

Particle uptake is quantified by measuring the total intracellular iron (Fe) content. This value is dependent upon on a number of factors including particle size, surface charge, labelling protocol and the rate of cell proliferation [25, 38]. In this study, a 24 hour serum-free passive incubation method was applied. This method tends to be limited to cells with a high degree of phagocytosis, a property not exhibited by stem cells [39]. The addition of transfection agents (TA) may thus be required to stimulate phagocytosis [24]. Studies by Hill et al. and Saldanha et al. have demonstrated the efficient uptake of micron scale particles by MSCs via passive incubation without the use of TA [19, 23]. In this study, 1 μm, commercially available particles, SiMAG, have been used and resulted in a labelling efficiency of 80% with a total iron content of 20 pg of Fe/cell. This value is comparable to other studies where internal Fe content ranged from 1 to 30 pg of Fe/cell compared to a value of 0.1 pg of Fe/cell for unlabelled cells [18, 20].

Internalised Fe disrupts the local magnetic field causing a shortening of T₂ and T₂*, thereby allowing SPION-labelled cells to be visualised as signal voids (blackened marks) or hypointense areas on MRI scans [19–25]. It is essential to establish MRI visibility thresholds in terms of particle concentration and cell number, both in vivo and in vitro and to appreciate that the detection threshold is affected by magnetic field strength and MRI acquisition parameters [20]. In vitro detection threshold was set at a T₂ of 0.0840 ± 0.002 ms corresponding to a minimal cell dose of 100,000 (5 and 10 μg/ml) and particle concentration of 1 μg/ml (300,000 cells). These are considered highly acceptable values as similar studies by Jing et al. reported a minimal cell dose of 5 × 10⁵ MSCs when labelled with 25 μg/ml Feridex and the transfection agent protamine sulphate [16]. As expected, T₂ was seen to decrease with increasing cell numbers and particle concentrations corresponding to the increasing Fe content. T₁ values were also measured but little difference in T₁ was noticed. Detection thresholds were validated in vivo by implanting 300,000 cells labelled with 1, 5 and 10 μg/ml into each mouse knee and MR imaged using GEFI T₂ weighted sequences. Relaxivity measurements could not be taken due to the inhomogeneities associated with biological tissue. Signal intensity across the hypointense regions of the knee were read, plotted and compared. The magnetic susceptibility of an individual SPION results in a blooming artefact which extends beyond the size of the individual particles allowing small injections of particles to be amplified beyond the actual location making for practical identification [20, 23]. The effect of particle concentration on the blooming effect is clearly seen here. In mice injected with SiMAG (5 and 10 μg/ml), significant loss in signal (61 and 78%, respectively) was measured over a substantial area; however, this signal loss was experienced over a greater area in the mice injected with 10 μg/ml than in those injected with 5 μg/ml. A less obvious drop in SI was observed for mice injected with 1 μg/ml (20%) when compared to the control mice.

Toxicity and safety is a major concern in the implementation of SPIONs in any cell-based therapy [34]. It thus becomes necessary to investigate the viability, proliferation and differentiation potential of SiMAG-labelled cells. A vast number of studies have reported that the use of SPIONs in conjunction with stem cells has little or no effect on the proliferation and viability of cells [18, 20, 25]. SPIONs are considered to be inert and biocompatible given the nature of Fe [18]. Fe is a naturally occurring element in the human body playing an important role in cellular metabolic processes, such as DNA synthesis, oxygen transport and redox reactions [29, 34]. The body is therefore adapted for Fe metabolism and thus labelling MSCs with SPIONs is not likely to affect the biological properties of cells [29]. In high quantities, however, Fe can possibly impair cell viability by damaging cell membranes, proteins and DNA [28, 29]. Therefore, it is important to obtain a balance between Fe incorporation for the required role and cell function [28]. Cell viability assays revealed that cell viability was maintained in SiMAG-labelled cell populations up to seven days post labelling with no significant effect on cell proliferation capacity either. Furthermore, FACS data demonstrated similar profiles for SiMAG-labelled cells and unlabelled cells implying that the presence of SiMAG does not alter the MSC cell surface markers.

Currently, there is a debate regarding the differentiation ability of cells once they have been labelled with SPIONS [25]. Generally, studies have found that the osteogenic and apidogenic potential of MSCs was maintained post SPION labelling whilst there are conflicting reviews reported for the differentiation of SPION labelled MSCs to chondrocytes [25, 29]. Studies have shown that stem cells labelled with Feridex (an FDA approved contrast) inhibited their differentiation towards the chondrogenic lineage but this was not the case for osteogenic and adipogenic lineages [25, 28, 40]. Kostura et al. found that the presence of Feridex interfered with the signalling pathway responsible for driving chondrogenic differentiation [40]. In contrast, Jasmin et al. reported that MCSs labelled with Feridex underwent successful differentiation down all three lineages (adipogenesis, chondrogenesis and osteogenesis) [30]. In a study by Henning et al., it was shown that differentiation to chondrocytes was dose dependent which may explain the contrasting results [29]. Henning also suggested that the use of a TA could mitigate this effect by encouraging the internalisation of particles via an alternative mechanism and further compartmentalization which might cause less interaction with differentiation linked intracellular substrates [29]. We have not been able to demonstrate successfully the differentiation of BALB/c mMSCs to chondrocytes in either labelled or unlabelled cells. In a study by Chamberlain et al. BALB/c derived mMSCs were shown to differentiate down osteogenic and adipogenic, but not chondrogenic, lineages [26]. We cannot, therefore, contribute the lack of chondrogenic potential to the presence of SiMAG but to properties of the cells themselves. Importantly, we have shown that mMSCs labelled with 10 ug/ml of SiMAG retain their capability to differentiate successfully down osteogenic and adipogenic lineages.

The chosen model bears numerous similarities to the human mode of the disease and is thus suitable for predicating the efficacy and appropriateness of this therapy in humans [41]. It must be noted that the rodent version of this disease progresses at a faster rate than does the human version. Therefore, any interventions are monitored by assessing acute inflammation (macrophage, lymphocyte and neutrophil infiltration) in terms of joint swelling [3, 41]. As expected, a significant decrease in joint swelling was measured upon mMSC administration. In a similar way, the administration of SiMAG-labelled mMSC also resulted in a significant decrease in joint swelling with no statistical difference between the mMSC groups. This suggests that the immunomodulating properties of mMSC labelled with SiMAG are maintained. Biological variation between mice within groups accounts for the conflicting rates of joint swelling progression between the two studies where a significant drop in joint swelling was found on day two for the three day study but only on day three for the seven day study. A limitation of this study in terms of assessing the therapeutic effects of mMSCs in treating RA would be the low N numbers of the study.

The therapeutic effects of mMSCs in the same model of RA were assessed by Kehoe et al. (paper under review). This study concluded that mMSCs are therapeutic when injected into the joints of mice with AIA as demonstrated by reduced levels of cartilage destruction. Toluidine blue staining demonstrated far less cartilage depletion in groups receiving either SiMAG labelled or unlabelled mMSC when compared to control groups. This suggests that the therapeutic potential of mMSCs is not affected by SiMAG labelling. Although the mechanism of action is unknown, it is unlikely that the implanted cells are differentiating into chondrocytes and repopulating the damaged area as no fluorescently labelled cells could be found within the articular cartilage, corroborating findings by Kehoe et al. It has been suggested by Kehoe et al. that the mechanism of repair is attributed to a paracrine effect whereby the early MSC treatment acts to prevent proteoglycan loss via the secretion of factors influencing the activity of ADAMTS enzymes, an enzyme responsible for the cleaving of aggrecan (an abundant proteoglycan in articular cartilage).

In terms of imaging and tracking, the blooming effect of the particles and the application of T₂ weighted sequences (GEFI/FLASH) made for simple identification of the implanted SiMAG-labelled cell population and the location was easily identified as the synovial cavity up to seven days post implantation. This was shown graphically by comparing the SI of joints treated with SiMAG-labelled cells to unlabelled cells and to untreated knees, where a significant loss of signal was measured, further highlighting the position of cells. This was further verified via histological analysis where DiI labelled cells were clearly seen within the synovial joint up to seven days post implantation.

The use of MRI has a dual purpose. Not only can it be used to image and track implanted cells, but its ability to distinguish between cartilage and bone [3] can be used to assess the defect and determine the extent of cartilage repair. The amount of fill in the image could reflect the extent of repair while comparing the signal of the new graft with surrounding tissue could indicate the maturity of the graft [42]. As mentioned previously, the blooming effect of the cells allows for the simple identification of cells in vivo. However, when applied to such a small joint, it is impossible to analyse tissue repair as it blocks all surface anatomy.

Although this study has not assessed the extent of cartilage repair, there is no reason to believe that the presence of SiMAG will affect the repair process corroborating results seen in a similar study from Kehoe et al. (under review).

Regenerative medicine and tissue engineering thrive on the cross collaboration of multiple fields to aid in the application of stem cell based therapies. This often involves the development and implementation of enabling technologies (as described in the manuscript) in line with regulatory bodies to bring stem cell based therapies closer to the clinic. Feridex and Endorem are examples of FDA approved, iron-based MRI contrast agents that have previously been used in conjunction with TAs in similar studies. Unfortunately, these particles have recently been taken off the market [34] with no other suitable replacements. Therefore, the need to identify and implement a new particle is necessary. The ideal labelling agent should allow for the repetitive, non-destructive and non-invasive long term tracking of implanted cells while maintaining cellular function at a realistic dose within a clinically relevant system [22]. SiMAG has been shown successfully to maintain cellular function at realistic doses within a clinically relevant system; however, its application in long term tracking has yet to be investigated. This manuscript thus highlights the potential of SiMAG to be used and adapted as a suitable agent for the imaging and tracking of cell populations via MR imaging in the clinical translation of a wide range of cell therapies.