In vivo fate of free and encapsulated iron oxide nanoparticles after injection of labelled stem cells

Nanoparticle contrast agents are useful tools to label stem cells and monitor the in vivo bio-distribution of labeled cells in pre-clinical models of disease. In this context, understanding the in vivo fate of the particles after injection of labelled cells is important for their eventual clinical use as well as for the interpretation of imaging results. We examined how the formulation of superparamagnetic iron oxide nanoparticles (SPIONs) impacts the labelling efficiency, magnetic characteristics and fate of the particles by comparing individual SPIONs with polyelectrolyte multilayer capsules containing SPIONs. At low labelling concentration, encapsulated SPIONs served as an efficient labelling agent for stem cells. The bio-distribution after intra-cardiac injection of labelled cells was monitored longitudinally by MRI and as an endpoint by inductively coupled plasma-optical emission spectrometry. The results suggest that, after being released from labelled cells after cell death, both formulations of particles are initially stored in liver and spleen and are not completely cleared from these organs 2 weeks post-injection.


Introduction
Stem cell regenerative medicine therapies have been proposed for the treatment of a range of debilitating conditions. 1-4 Tracking of stem cells in pre-clinical models is a prerequisite to determine their safety and efficacy. [5][6][7] For this purpose, in addition to, and/or in combination with genetic reporters, nanoparticles are successfully used as contrast agents for cell labelling and tracking with various imaging modalities. [8][9][10][11][12] Cellular biodistribution, viability and proliferation can be monitored by bioluminescence imaging (BLI) the cells transduced with the genetic reporter firefly luciferase (Luc). [12][13][14] Cell death results in a decrease in bioluminescence signal, whilst cell division and tumor formation leads to an amplification of the signal. This is a highly sensitive and robust method of monitoring the biodistribution of cells though it has a low spatial resolution (intra and inter organ monitoring of cells distribution is not possible). 15 Magnetic resonance imaging (MRI) offers much higher spatial resolution (50 µm) than BLI allowing in vivo cell tracking combined with detailed anatomical information at the level of individual organs. 16 Appropriate MRI contrast agents 17,18 for cell tracking include superparamagnetic iron oxide nanoparticles (SPIONs) which generate a negative contrast. 11,[19][20][21][22][23][24] The uptake efficiency of SPIONs is strongly influenced by their functionalization. 21,25 Recently, positively-charged DEAE (diethylaminoethyl)-dextran coated SPIONs have been synthesized for enhanced cellular uptake and MRI contrast. 20,22 Upon cell uptake, the clustering and confinement of these particles in endosomal and lysosomal compartments affects their abilities to alter the relaxation rate and hence relaxivity of the surrounding water molecules. 26,27 The critical parameter for imaging is the contrast obtained after cell labelling rather than the solution relaxivity of the particles. 20,22 One potential approach to enhance SPIONs uptake by cells, is to assemble them inside polymeric capsules resulting in the packing of large amount of individual nanoparticles within a confined volume. [28][29][30] Polyelectrolyte multilayered (PEM) and multifunctional capsules are fabricated by layer-by-layer (LbL) assembly of oppositely charged polymeric layers around a template. 31 Their design can be tailored by multiple strategies and particle loading can be enhanced by including more layers of particles during the capsule's assembly. 32 The choice of the final polymer layer determines their cellular interaction, uptake efficiency and hence internalization. It has been observed that a final layer of positively charged polymers results in increased uptake of capsules. 33 For long term cellular imaging the walls of the capsules can be made of non-biodegradable materials. 34 We and others have shown that most cells die after injection. 11,15,35 Therefore, understanding the in vivo fate of cell-labelling nanoparticles, especially following the death of the labelled cells is important for the interpretation of imaging studies and to assess the risk of toxicity. In the current study, we focus on the effect of formulation on the fate of SPIONs after the in vivo injection of labelled cells. More specifically, we labelled mouse bone marrow derived mesenchymal stem cells (mMSCs) with free and PEM-encapsulated SPIONs. First the solution relaxivities, magnetic properties, toxicity, and cell labelling efficiency of both formulations of particles are compared. Then, after intra-cardially (IC) injecting the labelled cells in the left ventricle of mice, the animals were longitudinally imaged by BLI and MRI until 14 days post injection. The cellular bio-distribution and viability was monitored by BLI, whilst the MRI allowed visualization of the in vivo fate of particles after cell death. At the end point of experiments (2 weeks post-injection), the amount of elemental iron inside the mice organs was measured to analyze the accumulation and elimination of both formulations of SPIONs. Our studies demonstrate a similar accumulation and elimination pattern of particles injected via labelled cells to what has been reported for the particles directly injected in the blood stream. [36][37][38][39][40]

Preparation and characterization of free and encapsulated SPIONs
To evaluate the effect of incorporation of the SPIONs into capsules, magnetization curves were acquired for capsules and free particles; no difference was observed between the magnetic properties of the free and encapsulated SPIONs (Fig. 1). Whilst the properties of individual particles were not affected by encapsulation, the large number of particles in a single object resulted in higher magnetic forces that enabled magnetic separation of particles during purification from excess reactants. Thus, during LbL assembly of particles in polymeric capsules, as the number of particles per capsule increased, their magnetic separation by a bar magnet became possible. After deposition of 3 layers of SPIONs, for further layer deposition, the excess reactants were therefore removed by magnetic separation instead of centrifugation. The transmission electron micrographs ( Fig. 1 a & b) show that particles did not fuse inside the capsules; they retained their identity as individual particles (as indicated by the magnetization curve; Fig. 1 c). The crystalline structure and particle size (~ 8 nm) was confirmed by powder XRD (pXRD) before encapsulation (Fig. SI.1 a). Their hydrodynamic diameter in water was measured by dynamic light scattering (DLS, ~85 nm, Fig. SI.1 b). The hydrodynamic diameter of capsules cannot be measured accurately by DLS because of their dimension (µm size) and sedimentation behavior.

Effect of encapsulation of particles on the solution relaxivity
To determine the effect of encapsulation on the solution relaxivity, the water relaxation rates of particles at different concentrations (based on Fe content) were measured by MRI and plotted as function of concentration (Fig. 2). Encapsulation of SPIONs resulted in a drop in solution relaxivity due to hindered access to the solution water protons and is similar to the previous findings of the entrapment of free particles inside the lysosomes. 20,22,32 Cell labelling efficiency and toxicity of SPIONs mMSCs were chosen for this study because they are similar to human bone marrow derived mMSCs which are being used in clinical trials, 41,42 and it has already been shown that these cells uptake free SPIONs and other nanoparticle probes with good efficacy. 12,22 To evaluate the cell labelling efficiency and toxicity of free and encapsulated SPIONs, cells were labelled with suspensions of particles (encapsulated or free) having comparable Fe content. The cells were imaged by MRI to quantify the solution relaxation times (ms; Fig. 3 a & b) and the Fe content in labelled cells was determined with a ferrozine assay (Fig 3, c). This revealed a two-fold higher Fe uptake by the cells when labelled with encapsulated SPIONs as compared to the free particles.
The difference in Fe uptake is unlikely to be due to electrostatic interactions, 25 Fe uptake per cell (Fig 3, c). Depending upon the cell line/type and size/dimensions of capsules, the number of micro capsules phagocytosed by cells cannot exceed a limit of 2 -15 capsules/cell. 33 So for low labelling concentrations a higher Fe uptake is achieved by using encapsulated particles. Higher degrees of labelling have been reported with high labelling concentrations of free particles. 19 In addition to efficient labelling at low Fe concentrations, capsules can be used to combine nanoparticles in a single entity and hence can facilitate multiplexed imaging and measurements. 43 The toxic effect of free and encapsulated particles is comparable except at high doses of 125 and 250 capsules added per cell (Fig. 3 d). This may be due to the sedimentation of the excessive encapsulated particles (µm dimension; all particles cannot be endocytosed) resulting in the formation of a layer entirely covering the cells' surface leading to very high local concentration. In the following sections, the cells were labelled with particles below their

Bioluminescence of mMSCs
To verify the mMSCs Luc + bioluminescence signal, D-luciferin was added to the cells and the bioluminescence signal was detected after 15 min of incubation using an IVIS spectrum imaging system. It was noticed that the bioluminescence signal from ~200 cells was detectable (

Cell phenotype after labelling and re-seeding/re-spreading
To observe cell phenotype, the trypsinized cells suspended in ice cold PBS for 5 -6 hrs were plated (re-seeded). Three days post-seeding the morphology of the labelled cells observed by light microscopy was not distinguishable from unlabeled cells, irrespective of whether they were labelled with free or encapsulated SPIONs ( Fig. SI.3). Capsules can be seen inside the labelled cells (Fig. SI.3).

Animal injection and in vivo monitoring of the bio-distribution of labelled cells
When cells are injected into mice, a significant proportion of them die within a short timeframe. 5,42,44 Due to the impact on the interpretation of imaging results and particles' associated toxicity, it is important to see the fate of particles once labelled cells die. Hence, soon after injecting the labelled mMSCs into mice, their in vivo bio-distribution was monitored in parallel by BLI and MRI for 14 days. As expected, the vast majority of the cells were no longer detectable by day 1. Surprisingly, in some animals, a strong luminescence signal was detected in the cardiac regions throughout the course of imaging (Fig. 4); this was likely due to small numbers of cells from the needle tract engrafting in the cardiac muscles following IC administration. By day 14, luminescence signals were present in the hind regions of some mice.
As we have previously shown, these were likely to be osteosarcomas. 11 On culling the mice, no tumor was detected in the major body organs (liver, spleen, brain, heart, and lungs). A key objective of this study was to investigate what happened to the free and encapsulated SPIONs after day 1 when BLI showed that the majority of the cells had died. For this purpose, we require consecutive MRI and BLI to be performed on the same animals.

MRI complementary to BLI
The high resolution of MRI facilitates the visualization of the intra-organ bio-distribution of the contrast agent. 11 Here, MRI allowed the monitoring of in vivo clearing and accumulation pattern of particles ( To quantify the Fe accumulation and clearance pattern before and after injection, the relaxation times (ms) in some organs and tissues (liver, spleen, spinal muscles, kidney cortices and medullae) were calculated (Fig. 5) from the MRI scans by drawing regions of interests (ROIs) around these organs and tissues. Soon after injection, the relaxation time decreased in the liver for both free and encapsulated SPIONs and a further decrease can be seen in the following days post-injections. The BLI data indicates that 24 hours post-injection there was considerable cell death. Therefore, the decrease in relaxation time in the liver on day 1 and 2 post-injection is possibly due to SPIONs accumulating in the liver following their release from cells. The presence of SPIONs in the liver up to 48 hours post administration is in accordance to the recent findings of Scarfe et al. 11 On day 6 post-injection, the relaxation time in the liver started to increase but remained below the baseline value during the entire time course of imaging ( Fig. 5). This suggests that the SPIONs accumulated in liver were not completely cleared/eliminated by day 14. The tendency of accumulation and hence slow clearance of SPIONs from the liver of mice after cell death is similar to the fate of free particles systemically injected in mice. 36 The spleen has very short relaxation time baseline values, and no further reduction was observed until the last imaging day in the case of free SPIONs (Fig. 5). By contrast, relaxation times appeared to be lower for encapsulated SPIONs at day 14 post injection. Interestingly, the relaxation times for encapsulated SPIONs had increased in the liver at this time point, suggesting that after leaving the liver, encapsulated particles accumulated inside the spleen.
However, detailed biochemical pathways for Fe accumulation in these organs need to be explored. Recently, the long term in vivo fate of different hybrids of SPIONs with gold nanoparticles has been established after directly injecting the particles in the blood stream of mice which revealed a predominant accumulation of particles in liver and spleen for longer periods of time. 36 In similar studies, the accumulation and metabolism of particles in the liver and spleen and loss of magnetic characteristics as a function of their surface coating, inner core and formation of protein corona was evaluated. [37][38][39][40] The electron paramagnetic resonance (EPR) based measurements of iron oxide nanocubes showed their considerable elimination from liver on day 7 post administration, whereas half of the proportion of particles stored in the spleen was still detected on day 7 post injection. 39,40 In general, free SPIONs have faster lysosomal degradation and cellular elimination rate as compared to the aggregated particles. 45 The relaxation time and MRI contrast in spinal muscles did not change considerably throughout the time course of experiments (Figs. 5 and SI.4). The SPIONs detected in the kidney cortices and medullae immediately after administration were mostly cleared by day 1 post injection ( Fig. 5). This finding suggests that after cell death, the particles were rapidly cleared from the kidneys, in line with what has been reported for the intravenous and IC injected particles where particle accumulation inside kidneys is not observed. 11

Inductively coupled plasma-optical emission spectrometry (ICP-OES) based Fe quantification
To confirm the results of the MRI analysis, at the end point of imaging experiments the animals were culled and ex-vivo ICP-OES based Fe quantifications in 6 organs (brain, kidney, liver, lung, spleen & heart) were performed (Fig. SI.5). This analysis did not show a significant difference in the percentage of Fe per dry weight of organs at day 14 post injection from the control group of mice which did not receive any injection. This suggests that the majority of the particles released on cell death were cleared by 2 weeks. Although the sensitivity of such measurements is limited as the presence of high quantities of endogenous Fe and animal-to-animal variability renders their interpretation difficult. More elaborate measurement strategies, e.g. isotope labelling, 37, 46, 47 could be employed in future studies. Formvar/carbon-coated 200 mesh copper grids (#F077/100) were purchased from TAAB.

Characterization Tools
The hydrodynamic diameter and zeta potential measurements were performed in water using a

Synthesis of super-paramagnetic iron oxide (maghemite) nanoparticles (SPIONs)
SPIONs were synthesized following the published protocol from Barrow et al. 19,20 Briefly, maghemite (SPIONs) were synthesized by mixing DEAE-dextran (0.05 g, Mw 40 kDa), ferric chloride hexahydrate (0.03 g), and ferrous chloride tetrahydrate (0.015 g) in 25 mL water with non-magnetic (polytetrafluoroethylene) stirring under an air-tight connection. The mixture was purged with nitrogen on ice (30 minutes) and ammonium hydroxide (1 mL, 28 -30%) was added dropwise (in 2 min) under stirring (200 ± 5 rpm). This was followed by mixture transfer from ice to an oil bath (pre-set at 60 °C) and temperature elevation to 80 °C under nitrogen (in 15 min) and then left at 80 °C for 1 h. Afterwards, the product was brought under air, and refluxed for 5 hrs at 110 °C. SPIONs were purified at room temperature by dialysis (100 kDa membrane) until the pH became neutral (7). After concentrating by spin filter (1 -2 mL), the sample was passed through size exclusion beads (dextran-based G-100 Sephadex®) to remove excess free polymer.
Finally, the particles were washed thrice with deionized water by spin filteration. SPIONs were passed through 0.22 µm polyethersulfone (PES) membrane for sterilisation before use in experiments and saved at 4 °C.

Synthesis of SPIONs encapsulating PEM capsules
SPIONs were encapsulated by a modified co-precipitation method, 28 For SQUID measurements, free and encapsulated SPIONs were freeze-dried and SQUID measurements, TEM, pXRD (free SPIONs), zeta size and potential measurements were performed as mentioned in the characterization section.

Ferrozine assay and ICP-OES based Fe quantification
The samples (particles and particle-labelled cells) were dissolved in 1. (Note: the ferrozine reagent was prepared by dissolving 2.4 g ammonium acetate, 2.2 g ascorbic acid, 0.02 g ferrozine, and 0.02 g neocuproine (dissolved in small volume of ethanol) in 6.25 mL water, mixed, frozen in small aliquotes and protected from light).
For ICP-OES Fe quantifications after dissolution with 1.2 M HCl, the samples were diluted in water so that final acid content will not exceed 5%. ICP-OES measurements were performed using 3 readings per wavelength and 11 wavelengths per sample. The concentration determination was performed using calibration curve for Fe consisting of 6 measurement points of freshly prepared Fe concentrations (0 -10 µg/mL) derived from iron standard solutions from Inorganic Ventures (100 mg/L).

Cell viability measurements
CellTiter-Glo® reagent was used for cell viability measurements. For this assay, mMSCs were seeded in 96-well transparent bottomed plates at 10,000 cells per well in 100 µL complete cell growth medium (DMEM supplemmented with 10% FBS, 1% P/S, and 1% L-glutamine). After 24 hrs cells were incubated with free and encapsulated SPIONs suspended in fresh cell culture medium for 24 hrs. Serial dilution of particles was performed in complete cell growth medium.
Each condition was applied in triplicate. After incubation with particles, the cells were washed with PBS and 100 µL fresh cell growth medium was added to each well of assay plates, followed by the addition of 20 µL CellTiter-Glo® reagent. The plates were shaken at 600 -700 rpm in an orbital shaker for 1 min (to complete cell lysis), incubated at room temperature for 5 min and 100 µL mixture contents from each well of the plates were transfered in opaque bottomed 96 well plates. The luminescence from each well of 96 well plate was measured by a Fluostar Omega (BMG Labtech) plate reader. Viability was expressed as % of the untreated control.

Phantoms for MR measurements
Phantoms for MR measurments were prepared in 1% low melting agarose.  Table SI.1. The solution relaxivity of the particles entrapped in phantoms was calculated from their relaxation times which were acquired by multi gradient echo sequences (MGES) and the imaging was performed with a fast low angle shot (FLASH) sequence.

2-Phantoms of labelled cells
In 6 well plates mMSCs were seeded (2 x 10 5 cells per well) in 2 mL complete growth media.

Luc + activity of mMSCs
The cells were monitored for Luc + activity before in vivo experiments. For this mMSCs were seeded in 96 well plates in triplicate with serial dilutions (10,000, 5000, 2500, 1250, 625, 312, 156, 78, 39, 20, 7, 3, 2, 1, and 0 cells per well) in 100 µL complete growth media and cultured at constant (5%) supply of CO2 for 24 hrs at 37 °C. After refreshing the growth media, the plates were placed in IVIS spectrum imaging system, and noted the baseline bio-luminescnce signals.
Latter, their growth media was replaced by fresh growth media (100 µL) containing D-luciferin (15 µg/mL), left at room temperature for 15 min and their bio-luminescence was recorded and expressed as radiance (photons/second/cm 2 /steradian [p/s/cm 2 /sr]).

Cell labelling and administration in mice
mMSCs transfected with lentiviral vector (Luc + ) were IC injected in mice. The vector details are described in a recent study. 12 To label cells with particles, 1.5 X 10 6 mMSCs were seeded in 100 mm tissue culture dishes and incubated for 24 hrs at 37°C in humidified atmosphere with 5% CO2. Free (33.72 pg Fe/cell) and encapsulated (@15 capsules per cell equivalent to a dose of 13.1 pg Fe/cell) SPIONs were added to the cells considering the seeded number of cells and incubated for 24 hrs. The cells were washed twice with PBS, trypsinized, kept on ice (after removing trypsin), and resuspended in ice cold PBS after cell counting. To observe the morphology of cells after particles labelling, some of the trypsinized cells were re-seeded and imaged after 3 days. 10 6 cells in 100 µL PBS (having ~5 and ~4 pg Fe/cell for free and encapsulated SPIONs, respectively) were IC injected into the left ventricles of female BALB/c mice under ultrasound guided injection. Details of IC injections is provided in a recent study. 11 Briefly, mice were anesthetized with a mixture of isoflurane and oxygen, subcutaneous (SC) injected with an analgesic buprenorphine (0.1 mg/kg body weight) and were positioned supine above a heated platform. Their fur was removed and the limbs and abdomen were taped after extension of body (to hold body in extended position and the skin over the chest taut).
Ultrasound gel was applied to the chest, and ultrasound tranducer was positioned over the mice to have chest in view. Visualized the heart and labelled cells were injected inside the left cardiac ventricle with the help of 29G ½ inch insulin syringe. The cells were completely resuspended, and were injected in a slow and well controlled manner. Soon after IC injections mice were imaged by MR under the same anaesthesia session and latter in groups of 4 by IVIS based BLI, on 0 (injections day), 1, 2, 6, 10, and 14 days post injections. Where possible mice were imaged by both imaging modalities under same anesthetic session. The mice were recovered from anaesthesia in a heat box set at 32 °C, and closely monitored for the signs of any adverse effects during the course of studies.

In vivo MR imaging
The biodistribution of cells in the abdomenal region, i.e., kidneys, liver, spinal muscles, and spleen was imaged with a 9.4T horzontal bore MRI scanner (Bruker Avance III spectrometer) using a 40 mm transmit/receive volume coil. FLASH T2* weighted sequences were recorded and T2* maps were generated using MGES. A prescan before injections served as control to set baseline value for MR analysis. Regions of interests (ROIs) were drawn around the kidney cortices, medullae, livers, spleen and spinal muscles and T2* relaxation times were calculated from the T2* map images. MR acquisition parameters are listed in Table SI.3.

Bioluminescence imaging (BLI)
BLI complementary to each MR scan was performed until the last imaging day (day 14 post injection). Mice received D-luciferin (150 mg/kg body weight) intraperitoneally and were imaged 15 min post injections by IVIS spectrum imaging system. Imaging was performed by 1 -3 min luminescnce exposure and expressed as radiance (photons/second/cm 2 /steradian [p/s/cm 2 /sr]).

ICP-OES based Fe quantification in mice organs
For quantification of Fe content in the brain, kidneys, liver, lungs, spleen, and heart, at the end of in vivo experiments mice were killed under terminal anesthesia followed by cervical dislocation. After dissection, the above mentioned organs were collected, washed with PBS and preserved in 70% ethanol. Organs collected from 3 untreated mice served as control. The organs in 70% ethanol were dipped in 96% ethanol, crushed with pestle and mortar and left overnight for freeze-drying. The dried crushed organs mass was weighed. 0.5 mL HCl (1.2 M) was added to the dried and crushed organs, kept inside an oven at 70 °C for 3 hrs and brought the reaction contents at room temperature. Water was added to each sample vial to get 10 mL final voulme. Samples were filtered through 0.2 µm filters to remove organ debris. Fe content in each sample was determined by ICP-OES. The dried weight of crushed organs noted before acid digestion was used for normalizing Fe content per dried weight of each organ.

Conclusions
We have demonstrated that SPIONs can be encapsulated inside polymeric capsules while retaining their individual particle identities, and showed similar magnetization behavior as free particles. Although the solution relaxivity of free particles was very different from encapsulated ones, once internalized inside lysosomal compartments after cell uptake, this difference became negligible for similar concentrations (Fe) of particles. The encapsulation of particles has some potential advantages over free particles; e.g., (1) in case of cell labelling with small amount of particles, a higher uptake can be achieved, and (2) the addition of different types of particles in a single entity for multiplexed measurements is possible. Comparison of both formulations of SPIONs did not reveal any difference in the in vivo fate of particles upon cell death and was similar to what has been reported for particles directly injected in the blood stream of mice.