|Home | About | Journals | Submit | Contact Us | Français|
Molecular imaging has undergone an explosive advancement in recent years, due to the tremendous research efforts made to understand and visualize biological processes. Molecular imaging by definition assesses cellular and molecular processes in living subjects, with the targets of following metabolic, genomic, and proteomic events. Furthermore, reporter gene imaging plays a central role in this field. Many different approaches have been used to visualize genetic events in living subjects, such as, optical, radionuclide, and magnetic resonance imaging. Compared with the other techniques, magnetic resonance (MR)-based reporter gene imaging has not occupied center stage, despite its superior three-dimensional depictions of anatomical details. In this article, the authors review the principles and applications of various types of MR reporter gene imaging technologies and discuss their advantages and disadvantages.
The major goal of molecular imaging is the spatiotemporal imaging of genomic and proteomic events. Molecular imaging encompasses the fields of chemistry, biology, physics, and medicine, and brings experts in these scientific fields together to determine means of visualizing molecular and cellular events 1, 2
Molecular imaging includes several imaging modalities, such as, bioluminescence, fluorescence, positron emission tomography (PET), single photon emission computed tomography (SPECT), computed tomography (CT), optical imaging, ultrasound, and magnetic resonance imaging (MRI). The advantages and disadvantages of these modalities have been discussed extensively in the literature 1. MRI is well known for its superior three-dimensional resolution, and can be used to acquire physiological and ultra-fine anatomical information using different pulse sequences. Unlike other cross-sectional modalities, such as, PET, SPECT, and CT, MRI is free from the issue of ionizing radiation with arbitrary imaging planes and provides multiplanar imaging capabilities. Ultrasound and optical imaging are limited in their ability to detect signals through deep tissue. CT provides anatomical maps for PET/SPECT; however, its role in molecular imaging is restricted due to its low sensitivity and limited contrast resolution of soft tissue. One of the biggest advantages of MRI is its ability to provide images of deep tissues in a background of superb anatomical detail. Furthermore like CT and ultrasound, MR scanners are widely available that can accommodate large animals.
The disadvantages of MRI are difficult interpretation among complex background signal intensity and its relatively low sensitivity. According to Massoud et al, the sensitivity of MRI probe detection is 106-109 times lower than PET and 1010-1014times lower than bioluminescence. However, thanks to the development of high field MRI and improved hardware and software designs, the signal to noise ratio of MRI has been improved significantly 1, and novel approaches used for reporter gene imaging may further increase its sensitivity and specificity.
In a typical MRI, the signal provided by the smallest image element is a function of mobile proton in hydrogen molecule and the relaxation time. There are two types of relaxation times, that is, a T1 spin-lattice relaxation time, and a T2 spin-spin relaxation time. Different pulse sequences exploit localized variations in these relaxation times, so that a given element (a voxel) takes on different signal intensities according to physiological conditions. “Enhancement” is used to increase voxel contrast by perturbing the local environment. Gadolinium, the most commonly used MR contrast “enhancement” material, shortens T1 and produces bright voxels corresponding to gadolinium-containing region on T1-weighted images. On the other hand, iron, the second most common contrast agent, shortens T2, and produces dark voxels on T2-weighted images. T1-shortening agents are more useful clinically because target lesions are imaged at high signal-to-background ratios, whereas T2-shortening agents have higher sensitivity at the molecular level 3 .
Two types of labeling mechanisms can be used to target a biological process or a cell using MRI. Direct labeling involves the binding of gadolinium or an iron-containing compound to the cell surface or internalized intracellularly. The other mechanism involves reporter gene techniques that induce specific genetic cascades. The major benefits of reporter gene techniques are that cells must be viable to produce an imaging signal, and that imaging signal does not dilute with cell division 4. Research is being undertaken to find safe transgene strategies for reporter gene imaging.
Reporter gene imaging by MRI can be grouped into four types: enzyme-based, spectroscopy-based, iron-related, or chemical exchange saturation transfer (CEST)-based.
Enzyme-based MR can be divided into two modalities, one for MR spectroscopy (MRS) and the other for MR imaging (MRI).
Several approaches to enzyme-based MRI have been devised. One pioneering example involved the use of ß-galactosidase (Figure (Figure1).1). Louie et al. developed a gadolinium-based substrate that contains a galactose group, which conceals the central gadolinium atom. In the presence of galactosidase (introduced by lacZ transfection), the galactopyranose moiety is enzymatically cleaved, which allows a water molecule to access the gadolinium, and increase the T1 signal 5. More recent study combined lacZ-transfected tumor with 3,4-cyclohexenoesculetin b-D-galactopyranoside and ferric ion, which result in T2* relaxation on MRI 6.
Another example is provided by the tyrosinase gene (Figure (Figure2).2). Tyrosinase is required for the production of melanin, which binds paramagnetic iron ions to produce metallomelanin, a T1-shortening agent. Cells overexpressing tyrosinase thus exhibit high signal intensity on T1-weighted images 7, 8.
Nevertheless, enzyme-based MR reporter gene imaging faces many challenges. One problem associated with the ß-gal-based approach is that even when the enzyme has not cleaved off galactose, outer surface relaxation changes of the gadolinium-containing compound are non-zero. Thus, while one might assume that the background signal would be zero without any endogenous ß-galactosidase, some enhancement may actually occur in practice. Another problematic issue concerns the delivery barrier, which restricts substrate access to the transfected cell. In addition, the persistence of melanin and metallomelanin in cells, even when the reporter gene is not activated, cause false high MRI signal intensities 9.
MR spectroscopy or MRS techniques take use catalytic enzymes, such as, creatine kinase and arginine kinase, and their conversion of ATP to ADP, using 31P MR spectroscopy (MRS) 10. For example, the arginine kinase gene from Drosophila can be transfected into vertebrate muscle using a viral vector, and the byproduct of the enzymatic conversion of ATP, phosphorarginine, can be detected by 31P MRS. The expression and activity of the transfected kinase can be followed in living mammalian muscle for several months after injection 11. Another approach involves the use of a reporter enzyme to follow a prodrug during its internalization and conversion by a target cell, for example, the cytosine deaminase gene catalyzes the conversion of the prodrug 5-fluorocytosine into 5-fluorouracile in tumor cells, which can be monitored by 19F MRS 12. The fluorouracile and fluorocytosine, being anions, cannot cross the cell membrane and are retained in the cell, therefore serve as a means to develop contrast. Furthermore, the cytosine deaminase result in toxic metabolite formation in the transfected tumor, suggesting that reporter gene technology can provide cancer gene therapy in the future 12-14.
Another enzymatic reporter gene technique for MRS utilizes b-galactosidase. Here, 19F MRS monitors the transformation of a fluorinated version of the traditional ß-D-galactopyranoside (p-fluoro-o-nitrophenyl-ß-D-galactopyranoside; PFONPG, or its isomer o-fluoro-p-nitrophenyl-ß-D-galactopyranoside: OFPNPG), to aglycone (p-fluoro-o-nitrophenol: PFONP), by cells transfected with the lacZ gene. The substrate PFONPG may be superior to EgadMe since it penetrates the cell membrane to be accessible to the intracellular enzyme, thus alleviating the delivery barrier issue and facilitating the in-vivo translation 15-17.
The limitations of MRS are the kinase-based study relies on energy metabolism, and that its images are of relatively low spatial resolution. It remains to be seen whether such a reporter gene method can be translated to multicellular organisms or to multidimensional imaging 9, 18, 19.
Recent reports have described iron-binding proteins other than metallomelanin. Weissleder et al. used an engineered transferrin receptor (ETR, Figure Figure4),4), which generated contrast by the receptor-mediated internalization of ironbound transferrin. This technique was used to image ETR-transfected (ETR+) gliosarcoma xenografts in mice after the systemic injection of a receptor-targeted probe, and it was found that ETR+ tumors showed significantly lower signal intensity on T2 gradient-echo images than control-transfected (ETR-) tumors 20. However, the overexpression of ETR or tyrosinase could lead to iron-catalyzed free radical formation via the Fenton reaction, and be potential toxic to cells 21, 22.
Ferritin is a ubiquitous protein which stores and releases the iron, and is activated by excess free intracellular iron to maintain iron homeostasis. The pioneering work by Cohen et al was based on the hypothesis that overexpression of ferritin would trap excess intracellular iron, and provide a signal without injection of exogenous contrast agent. It was shown that the induced expression of ferritin caused significant T2 relaxation time shortening 23. In a C6 glioma xenografts transfected with TET-responsive enhanced green fluorescent protein (EGFP) and heavy-chain transferrin (C6-TET-EGFP-HA-ferritin), the absence of tetracycline in drinking water was found to increase the spin-spin relaxation rate (Figure (Figure5,5, ,6).6). It should be noted that exogenous iron, such as SPIO (superparamagnetic iron oxide), was not administered for the signal detection. Genove et al. described the effect of the injection of viral vector carrying ferritin into mouse brain. Localized ferritin overexpression was found to result in decreased T2 time and thereby generating MR contrast by increasing R1 and R2' (R1=1/T2). It was also found that ferritin-transduced cells showed no adverse effect over control cells 24. In fact, various reports, including a review of ferritin by Arozio et al, have suggested that ferritin plays key roles in the following process: the reduction of reactive oxygen species levels, anti-angiogenesis, and in the reduction of cellular apoptosis from oxidative stress 25-27.
Questions regarding anomalies of ferritin-related relaxation remain to be answered. The phenomenon of T2-weighted image darkening by iron near protons is well known, but the measurements of relaxation rates and of iron quantities are sensitive to experimental protocols and researchers, for example, they are affected by different tissues, different solutions, molecular aggregation, MR field strength, and the quantification method used 28. One unique feature of MRI is the linear relationship between R2 and the external magnetic field; although a quadratic relationship has been suggested 29, 30. For example, the relaxation rate induced by deoxyhemoglobin, another iron-associated protein, shows a quadratic increase with respect to R2. To complicate the story further, the relaxation of iron, normalized by iron concentration, has different values in different tissues and even in the same tissues from different researchers. These variances suggest that relaxation is not caused by iron but rather by ferritin and hemosiderin. It has been suggested that differently sized aggregations of ferritin in tissues explain these discrepancies 28, 31. Furthermore, the signal change induced by ferritin is dependent on the redistribution of intracellular iron, rather than on net iron content 23. These confounding factors suggest that the main drawback of ferritin reporter gene imaging is its semiquantitative nature 30. More recently, Choi et al suggested using the R50 values than the T2* values themselves for in vivo iron quantification 32. By using melanoma cell line which stably expresses ferritin (myc-hFTH), they were able to evaluate the tumor burden in metastatic lymph node (Figure (Figure7).7). The drop in T2* value was noted in the myc-hFTH cells, but the histogram overlapped with those of control cells, making it difficult to draw a threshold for tumor burden. The cumulative T2* histogram analysis and its representative median value, R50, and may hold promise in future studies.
Recently, the MagA gene, which synthesizes magnetosomes in some bacteria, has attracted research attention. The gene product has properties in common with SPIO nanoparticles. In a study by Zukiya et al., the T2*-weighted images of magA-positive cells showed significant signal drops as compared with those of control cells. Electron microscopy revealed uniform particles magnetosomes. Cytosolic G6PD and MTT assays did not show cytotoxicity or change in cell proliferation. Particles produced in the magA-transfected animal cell do not cause immune response trigger on cell surfaces (Figure (Figure8,8, ,9).9). Accordingly, it has been suggested that magA-based imaging may provide a new means of noninvasive cellular tracking 33.
The newest class of MR contrast agents utilize a process called chemical-exchange saturation transfer (CEST). Here, frequency-selective irradiation, or saturation is used to label protons noninvasively in lysine-rich protein (LRP). The spin saturation is then transferred to bulk water resonance, and thus, changes the signal intensity of the water and provides on and off contrast. Gilad et al. transfected LRP-encoding mammalian vector into glioma cells, which were then inoculated into murine brains. LRP-expressing cells were selective saturated at the exchangeable proton-resonance frequency, and the signal intensity-difference map obtained was found to differentiate between LRP-expressing tumors and control tumors using a high field magnet (11.7 Tesla). Furthermore, tumor xenografts extracted at 10 days after inoculation showed LRP gene expression (Figure (Figure10)10) 34. These results show that CEST imaging to detect gene expressions by MR, but energy deposition in tissue (specific absorption rate : SAR) and the millimolar sensitivity of the technique are limitations 35. SAR could be described as the rate at which energy is absorbed by the body when exposed to a radiofrequency electromagnetic field, which is sufficient to cause tissue heating. Furthermore, SAR is dependent on magnetic field strength, which implies that the ultra-high magnet field needed for CEST signal detection is likely to cause significant tissue heating 36. Further studies are required to validate the applicability of CEST-MRI in a clinical setting.
Progress toward the development of a good MR reporter continues to be made. MRI continues to lag behind other molecular imaging modalities in this field because of its intrinsic sensitivity and the associated necessity of high-field magnet for signal detection, because of immunologic and biologic issues, or simply because it does not provide an instant 'hot-spot' signal like PET or bioluminescence. Until recently, no report on therapeutic gene delivery mediated by a solitary imaging-and-suicide reporter, such as HSV1-sr39tK for PET imaging, had been reported for MRI 37, 38. But researchers have showed that MR reporter gene may provide tumor-specific cytoxicity, for example, by cytosine deaminase and its metabolite 5-FU. Some approaches alleviate the need for external contrast injection, such as ferritin-related MRI, implicating its application in brain with blood-brain barrier 39. MR is indeed slowly gaining its foothold in the field of reporter gene imaging. There is no single imaging molecular imaging modality that will meet every purpose. Fusion reporters and fusion imaging modalities will probably be devised to take advantage of the three-dimensional microanatomical imaging offered by MRI, and the combined biologic and chemical efforts being made will undoubtedly accelerate the application of molecular genetic MRI.