Magnetization transfer imaging (MTI) is based on the magnetization interaction between bulk water protons and macromolecular protons, such as proteins (24). MT contrast is generated by applying off-resonance RF irradiation designed to preferentially saturate the immobile, or bound, protons of macromolecules. This saturates the energy level of the bound protons, inducing energy transfer to the mobile, or free water, protons and consequently reducing their signal intensity (25). Thus, MTI indirectly measures the interaction between these two proton pools. The relative difference in signal intensity (SI) with and without magnetization transfer can be measured as the magnetization transfer ratio (MTR). The MTR has been very useful for assessing tissue pathology and most MTI has focused on intrinsic markers like myelin and collagen (26, 27). Much of the work evaluating the utility of the MTR as a marker of disease has focused on multiple sclerosis (MS) and the associated demyelination (28, 29). Demyelination results in a reduction in the MTR, as protons associated with macromolecules are lost. Several papers show a strong correlation between the MTR and myelin content (28–30). Schmierer et al. proved that the MTR is affected by myelin content in MS WM by measuring the MTR in fresh postmortem brains from MS patients (29).

The MTR has not been widely applied to study SCI. Franconi et al. measured the MTR of the cervical cord in a rat SCI model at 7 T and found the MTR to be most useful in defining the thin CSF surrounding the cord but less effective for distinguishing between WM and GM (31). McGowan et al. used MTI at 1.7 T for histogram analysis in a rat weight-drop injury model and showed that the WM MTR was predictive of myelination status and neurofilament damage (32). Gareau et al. used MTI to compare severe and mild SCI due to contusion in rats and showed that the WM MTR was lower for severe SCI on the first day postsurgery, but not by 1 week, and that the WM MTR was insensitive to differences between mild SCI and the control uninjured spinal cord (33). When using the MTR to evaluate SCI, or other CNS damage, it is important to consider the extent of other pathology in the tissue since the MTR can also be influenced by edema; the accumulation of fluid (and so more free water protons) in tissue will also lower the MTR (33).

Cellular MRI is a relatively young field of MRI research that combines the ability to obtain high-resolution image data with the use of magnetic contrast agents for labeling specific cells, thereby enhancing their detection (60). Superparamagnetic iron oxide (SPIO) nanoparticles represent a class of magnetic contrast agents used for cellular MRI that exhibit extremely high relaxivity (61). A variety of iron oxide-based labels are now available (62). Iron-labeled cells are usually imaged using either T2*-weighted GE or T2-weighted SE sequences (62). The presence of the magnetic label causes a distortion in the magnetic field and leads to abnormal signal hypointensities. Areas containing iron-labeled cells appear as regions of low signal intensity, creating negative contrast. The large magnetic susceptibility of these particles affects an area much larger than the actual particle size. This effect is known as a “blooming artifact,” and leads to an exaggeration of the region occupied by iron oxide (63).

Cellular MRI studies use one of two ways to label cells with iron particles, in situ or in vitro. In situ labeling is performed by injecting iron particles intravenously with the goal of labeling circulating immune cells or immune cells at the site of inflammatory lesions. This approach has been used to detect and monitor macrophages in many different preclinical disease models, including experimental autoimmune encephalomyelitis (EAE) (64, 65), ischemic stroke (66), atherosclerosis (67, 68), peripheral nervous system injury (69), arthritis (70), and rejection in a variety of transplantation models (71–73). To determine the contribution of blood-derived macrophages to the signal loss observed in MR images of inflammatory lesions after i.v. SPIO, Oweida et al. performed in situ cell labeling in a mouse model of MS using transgenic mice that expressed enhanced green fluorescent protein (EGFP) specifically in hematopoietic cells of the myelomonocytic lineage (64). MRI revealed the presence of discrete regions of signal loss throughout the brains of animals that were administered Feridex and histology showed that regions of signal loss on corresponded anatomically with regions of PPB- and GFP-positive cells. This experiment provided the first direct evidence that macrophages of hematogenous origin are labeled with SPIO after i.v. administration of SPIO, and contribute to the regions of signal loss detected in MR images (Fig. 3). Dunn et al. used cellular MRI to detect macrophages in a clip compression model of SCI in the rat ex vivo MRI at 1.5 T (74).

In vitro cell labeling involves prelabeling cells prior to their injection or transplantation and following their fate with MRI. Many types of cells will readily take up iron particles by simple coincubation with iron oxide particles. For example, macrophages, cancer cells, and stem cells, are relatively easy to label this way. Other cells, in particular nonadherent cells, are more difficult to label and require specialized techniques. Various methods have been developed to enhance the amount of iron taken up by cells such as the use of transfection agents (75), functionalized iron particles (76), magnetoelectroporation (77–79), magnetic fields, and cell penetrating peptides (80). SPIO has been used to label and track a wide variety of cell types including T lymphoctyes (81), macrophages (82), pancreatic islets (83), cancer cells (63), and stem cells (84) with minimal impact on cell function over a period of several weeks and over multiple cell divisions.

Sykova et al. showed that iron-labeled mesenchymal stem cells (MSC) administered i.v. in SCI rats can be detected as regions of signal loss in ex vivo images of the cord at 4 weeks after the cell injection (84). They also showed that iron-positive cells in the cord did not colocalize with macrophage staining, suggesting that iron-labeled MSC persist and can be tracked with MRI. Dunning et al. have used SPIO to label and track Schwann cells (SC) and olfactory ensheathing cells (OEC) in vivo for up to 4 weeks in SCI rats. They showed by histology transplanted cells remyelinating the demyelinated axons, demonstrating that the iron-labeled cells retained their biological function and that the majority of the label had remained within the transplanted cells (85). Lee et al. tracked SPIO labeled OEC in rat SCI for 2 months and found that MRI was able to show that the OEC were unable to bridge the transection lesion (86). Gonzalez-Lara et al. were the first group to use cellular MRI to track transplanted stem cells in the injured mouse cord (20, 87). The presence of transplanted cells in the cord was easily visualized as a large localized region of signal loss and could be monitored over time in vivo (Fig. 4). MRI was also useful for evaluating the accuracy of the transplantation technique since iron-labeled cells inadvertently placed outside of the spinal cord tissue could be readily identified and these mice were removed from the longitudinal study. This short-term cell tracking is valuable because endpoint histological analyses of these transplantation studies normally report on the presence or absence of stem cells in the tissue, and knowledge of the success/failure of the original transplantation is crucial for the interpretation of these data.

The biggest limitation of using iron nanoparticles and MRI for cell tracking in SCI models is that when iron-labeled cells die (or free iron is released) the iron may be taken up by bystander cells (macrophages), which will confuse the image interpretation (88–90). In any direct transplantation model a percentage of transplanted cells are expected to die due to the injection procedure itself (91). Cells labeled peripherally after i.v. SPIO administration, or the viable transplanted cells, will also eventually die at the SCI lesion site due to the harsh inflammatory conditions. This is compounded in SCI models by disrupted blood flow at the lesion site, which may result in an accumulation of iron particles due to cell death. Gonzalez-Lara et al. recently examined this issue in a longitudinal imaging study that followed SCI mice for 6 weeks after the direct transplantation of viable iron-labeled MSC, dead iron-labeled MSC, or free iron particles (87). The MRI appearance of the region of signal loss at the transplant site was similar for all three groups. In all transplanted SCI mice the volume of signal loss decreased over time. This decrease was most rapid in SCI mice that received live iron-labeled cells. The persistence of signal loss in the images of cords from mice that received dead MSC or free iron was attributed to iron particles lodged in the damaged cord tissue at 6 weeks postinjection, as shown by iron stained histology. The method of transplantation, host tissue, cell type, cell lifespan, and the extent of inflammation in the model system no doubt play important roles in the cell survival and clearance. The direct transplantation models for the injured spinal cord may represent the worst-case scenario for cell tracking with MRI. Long-term tracking of stem cells by MRI in models of SCI should be carefully validated through histology.