1.INTRODUCTION

MRI has proven to be a powerful imaging modality for non-invasive, whole body imaging, with meaningful image resolution for studying cellular dynamics. MRI contrast is generated by capitalizing on differences in water molecule microenvironment, including diffusion rate and direction, magnetic field differences, and water content. In order to use MRI to distinguish a unique population of cells from other cells in the body, such as a cell transplant, or the infiltration of immune cells to a disease, one of these properties needs to be altered. This will generate contrast from these cells.

2.CONTRAST AGENTS FOR MRI - A REALLY CONCISE HISTORY

The use of paramagnetic, metal-based chemicals to purposefully affect NMR signal characteristics is nearly as old as the field of NMR itself. The utility of iron based chemicals to decrease relaxation times of water was demonstrated as early as 1946 (1). Jumping ahead to 1973, MRI, or NMR Zeugmatographic Imaging as it were, was reported for the first time (2). Soon after, in 1978, the first demonstration of a contrast agent for in vivo MRI was reported, with manganese chloride used as an agent to delineate the boundaries of cardiac infarct in dog by exclusion of manganese by the infarct (3). 1978 also saw a report on the use of dextran-magnetite for T2 relaxation enhancement for NMR (4). In the mid-1980’s iron oxide nanoparticles were concurrently demonstrated to be useful for specific in vivo targeting applications, such as MRI tumor detection by way of antibody coated particles (5), and passive targeting studies, such as liver MRI by way of reticulendothelial system (RES) macrophage accumulation of iron oxide nanoparticles (6, 7).

The first reports of magnetic cell labeling for the purpose of specific cell tracking by MRI were published in 1992-3. Bulte, et al used two different versions of dextran coated magnetite to specifically label B- and T-lymphocytes by way of antibody mediated uptake (8) and various blood cells by viral envelope coated magnetoliposomes (9). Similarly, Yeh, et al used hydrophilic dextran coated magnetite nanoparticles to label T-lymphocytes, however without the use of antibodies (10). Two additional papers reported on magnetic cell labeling of neural stem cells with Hawrylak, et al using viral encapsulated magnetite (11) and Norman, et al using magnetite nanoparticles coated with wheat germ agglutinin (12).

In 1995, Ferumoxides was released as an FDA approved iron oxide nanoparticle. This dextran coated iron oxide nanoparticle was 100 – 150 nm hydrodynamic radius and contained a 5-10 nm iron oxide core (13). This standard, well characterized iron oxide nanoparticle allowed for numerous investigators to embrace magnetic cell labeling for MRI, however, low labeling amounts and efficiencies were often the case (14). In 2002, it was reported that combining simple transfection agents plus ferumoxides enabled robust labeling for a number of cell types (15). This discovery greatly accelerated the pace of MRI-based cell tracking as evidenced by a steep rise in number of publications per year investigating iron oxide labeled cells by MRI.

2.1 IRON OXIDE BASED CONTRAST AGENTS

2.1.1 Particles

Iron oxide nanoparticles have been the workhorse for MRI-based cell tracking for nearly 30 years. Iron oxide is a general term for molecules containing iron and oxygen, with several molecules such as Fe3O4, Fe2O3, and FeO easily synthesized. In bulk, these materials are magnetic, but as nanomaterials ~ <30-40 nm, these materials become superparamagnetic. This term describes the property of a material to achieve magnetism when placed inside a magnetic field, but lose it when removed from the field, that is, there is no or very little magnetic hysteresis. Facile synthetic approaches can be accomplished in water (16) or organic solvents (17). Aqueous magnetite is usually formed by the base catalyzed precipitation of magnetite from Fe(II) and Fe(III) salts, followed by citrate or oleic acid stabilization. Hydrophobic magnetite is most often formed by thermal decomposition of either Fe(II) salts or from Fe(CO)5 with oleic acid capping. Nanocrystal sizes range in diameter from 5-20 nm, with lower polydispersity and higher saturation magnetization achieved by the thermal decomposition route. Magnetite formation and properties are extensively reviewed in (16).

Iron oxide is potentially harmful to cells, so polymer coating strategies are employed to protect the cores from biology – and biology from the cores. An enormous array of natural and synthetic polymers has been used to coat particles, with varying degradation kinetics and functionalities. These have been reviewed extensively numerous times. In general, particles are encased within polymers as either single or few cores within polymer shells in a core:shell fashion, or are encapsulated en mass throughout larger polymer particles via emulsion methodologies, such as PLGA or cellulose (18, 19). These are reviewed here (20).

The superparamagnetic nature of these nanomaterials enables their detection via T2 and T2* weighted sequences. The diffusion of water through inhomogeneous magnetic field gradients caused by the nanoparticles enables T2 contrast while the static magnetic field caused by ensembles of nanoparticles or by microparticles or MPIOs allows T2* weighted contrast (21). Experimental MRI sequences have been developed to specifically detect iron oxide labeled cells by providing bright contrast. These include IRON (inversion-recovery with ON-resonant water suppression) (22), Off‐resonance saturation as a means of generating contrast with superparamagnetic nanoparticles (23), and GRASP (GRadient echo Acquisition for Superparamagnetic particles with positive contrast) (24).

2.1.2 Genetics

There are biological mechanisms for forming iron oxide nanocrystals. These include the protein ferritin, the magnetosomes found in magnetotactic bacteria and melanin. In general, ferritin contains a rather weak iron oxide nanocrystal, mainly due to the non-uniform crystal properties of the iron oxide. Interesting approaches have been reported to modify the ferritin macromolecule to improve its relaxivity, reviewed in (25), with the most recent report achieving 50% improvement in relaxivity by enhancing the biomineralization of iron (26). Various aspects of magnetosome formation have been parsed out and recapitulated in mammalian cells, including the iron transporter MagA (27). While contrast enhancement from iron accumulation is possible, fully reconstituting iron oxide nanocrystals themselves via magnetosome machinery remains elusive still. Lastly, melanin is synthesized by tyrosine hydroxylase from tyrosine containing monomers. Importantly, melanin acts like a metal sponge, and so an abundance of melanin will accumulate metals and yield MRI contrast, especially if those metals are iron. Thus, overexpression of tyrosine hydroxylase could be used via melanin synthesis to recruit iron locally (28).

2.1.3 Fluorine

Fluorine based contrast agents have emerged as a viable cell labeling strategy. Fluorine has no abundance in the body, and so cells can be labeled with fluorine based agents can be detected unambiguously following transplant. The most promising agents are perfluorocarbon based agents which are readily taken up by cells by simple incubation (29). This class of agents is extensively reviewed in (30).

3.1 CELL LABELING

3.1.1 In vitro

Magnetic cell labeling is mostly performed in vitro. Cells are plated in culture vessels and incubated with particles. Some particles, such as MPIOs (31) and PLGA encapsulated iron oxide particles (18), sink onto adherent cells and can be endocytosed within hours. Other particles which remain in suspension require assistance for efficienct labeling. Chemical tools that were originally developed for transfecting genes into cells have been adapted for cell labeling, and have the ability to increase the efficiency of particle uptake into cells in vivo or in vitro. These include the use of poly-L-lysine (PLL) (15) or protamine sulfate (32) which are co-incubated with particles and cells for in vitro labeling and co-injected with particles for in vivo labeling. A recent report on the complexation of heparin, protamine sulfate and feromoxytol has shown good uptake properties (33). More complex methods for in vitro labeling include electroporation (34) or sonoporation (35) and these have been used with some success. For cells which grow in suspension, transfection agents are entirely necessary, or other affinity based tagging such as antibodies (36).

3.1.2 In vivo

In vivo cell labeling has been demonstrated in a number of systems. In the brain, endogenous neural progenitor cells can be labeled by directly injecting magnetic particles either into the lateral ventricles (37) or into the rostral migratory stream (38). These cells endocytose the particles in situ and carry them as they migrate. In vivo labeling of dendritic cells has been performed in the context of vaccine development (39). In vivo labeling of immune cells, macrophages in particular, has been demonstrated in a wide variety of models across the biomedical spectrum. Whether cells are labeled as circulating monocytes or as homed macrophages remains to be fully discerned.

4.1 MRI METHODS AND DATA ANALYSIS

4.1.1 Iron quantification

Quantification of MRI-based cell tracking experiments is perhaps the least developed aspect of the experiment. In general, the data analysis remains phenomenological, that is, results are reported in vague terms of “cells are detected”. But in order to realize the true power of this technique, methodologies are required to quantify cell numbers, migration rates and other quantitative parameters. One approach to deriving quantitative metrics from this data is iron quantification. The position here, is that if we know how much iron cells contain, and we can quantify the iron, then we can quantify the number of cells. Two methods for iron quantification have emerged. The first is relaxometry. Reports detailing the use of T2 measurements to quantify iron content allow the measurement of cell numbers (40, 41). A second technique is based on SWIFT imaging (42). The challenge with these techniques is that the iron oxide nanoparticles can degrade over time, which affects their magnetic properties, and hence their relaxation properties. So, over time, these methods would be dependent on these changes. Further, relaxation times depend on clustering of particles, so as cells migrate, this clustering is reduced.

4.1.2 Cell counting

An alternative method for quantifying cell numbers is to use MRI-based single cell detection to enumerate cells. If cell can be labeled with enough iron oxide >10 pg/cell, and with high enough resolution, ~100 micron, individual cells can be detected in vivo. Spot detection paradigms can then be used to identify these cells and enumerate them. This has been nicely demonstrated by (43). The challenge, of course, is to provide this high resolution in scans of animals larger than rodents. This methodology also provides a nice mechanism for monitoring migration rates of cells (37).

5.1 CLINICAL APPLICATIONS

There are few reports of clinical application of cell tracking by MRI. This has been driven in part by the lack of FDA-approved contrast agents and lack of dedicated image analysis platforms for analyzing the data. Mostly, clinical applications of MRI-based cell tracking have employed off-label use of iron oxide nanoparticles (44, 45), though a recent paper describes a clinical study with perfluorocarbon tracer for detecting dendritic cells in cancer patients.