Genetic engineered molecular imaging probes for applications in cell therapy: emphasis on MRI approach

Abstract

Recent advances in stem cell-based regenerative medicine, cell replacement therapy, and genome editing technologies (i.e. CRISPR-Cas 9) have sparked great interest in in vivo cell monitoring. Molecular imaging promises a unique approach to noninvasively monitor cellular and molecular phenomena, including cell survival, migration, proliferation, and even differentiation at the whole organismal level. Several imaging modalities and strategies have been explored for monitoring cell grafts in vivo. We begin this review with an introduction describing the progress in stem cell technology, with a perspective toward cell replacement therapy. The importance of molecular imaging in reporting and assessing the status of cell grafts and their relation to the local microenvironment is highlighted since the current knowledge gap is one of the major obstacles in clinical translation of stem cell therapy. Based on currently available imaging techniques, we provide a brief discussion on the pros and cons of each imaging modality used for monitoring cell grafts with particular emphasis on magnetic resonance imaging (MRI) and the reporter gene approach. Finally, we conclude with a comprehensive discussion of future directions of applying molecular imaging in regenerative medicine to emphasize further the importance of correlating cell graft conditions and clinical outcomes to advance regenerative medicine.

Introduction

With the successful isolation of pluripotent stem cells and their maintenance in vitro, stem cell research has advanced dramatically over the past three decades. Various stem cell technologies, including isolation of human embryonic stem cells [1], directed stem cell differentiation [2-4], transdifferentiation [5,6], induced pluripotent stem cells (iPSC) [7], and successful derivation of patient-specific pluripotent stem cells by somatic cell nuclear transfer [8], underscore the tremendous pace at which stem cell technology is advancing. Stem cells have the potential to be used in fields ranging from, but not limited to, developmental biology, cancer biology, and genetics research to drug discovery and cell replacement therapy.

In the area of regenerative medicine, stem cell-based cell replacement therapy holds a high potential for curing diseases, such as diabetes, bone degenerative diseases, autoimmune diseases, myocardial infarction, brain and spinal cord injuries, and neurodegenerative diseases [9,16]. The primary goal of cell replacement therapy is to replenish damaged or degenerated cell populations. To develop and evaluate the efficacy of cell replacement therapies, it is essential to accurately assess cell survival, proliferation, migration, lineage differentiation, and functional integration at the graft site longitudinally [17,19]. Besides immune rejection and the functionality of cell grafts, one of the major obstacles to clinical translation is current lack of understanding of the fate of implanted cells and how this correlates with clinical outcomes. There is thus an urgent need for tools that allow for noninvasive and longitudinal evaluation of cell grafts. Interest in in vivo imaging of cell grafts has soared in the past decade (Figure 1A) on the heels of rapid advances in stem cell technology.

Figure 1.

The number of publications by year. A. The total number of publications by year. The PubMed search was conducted using the terms in vivo cell, imaging, tracking, or monitoring while excluding terms like reviews, methods, and drug delivery. The years when embryonic stem cells (1998) and induced pluripotent stem cells (2006) were developed are indicated by arrows. B. The number of publications broken down into each imaging modality. C. The number of publications using multimodal imaging methods. Abbreviations: PET-positron emission tomography, MRI-magnetic resonance imaging, BLI-bioluminescence imaging, CT-computed tomography, SPECT-single photon emission CT, CEST-chemical exchange saturation transfer.

The in vivo monitoring of grafted cells was reported first in 1976 [20]. In this inaugural study, leukocytes were extracted from patients, labeled with radioactive indium-111, reintroduced to patients, and followed for two days with a gamma camera [20]. With the development of lacZ (β-galactosidase) in 1980 [21] and green fluorescent protein (GFP) in 1994 [22], optical colorimetric and fluorescent reporter genes have since been used extensively in imaging of cellular events although the in vivo applications are limited. Today, there are a number of imaging modalities available for in vivo cell graft tracking leading to great interests and effort in developing cell tracking probes/reporters for respective imaging modalities, including positron emission tomography (PET) [23,24], computed tomography (CT) [24], single photon emission CT (SPECT) [25], ultrasound (US) [26,27], bioluminescence imaging (BLI) [28,29], fluorescence imaging (FLI) [30,32], magnetic resonance imaging (MRI) [17,23,33-39]. Among these available imaging modalities, MRI and PET are the most widely investigated and developed due to their relative greater potentials for human and clinical applications (Figure 1B). Recently, various combinations of imaging methods have been investigated for in vivo cell imaging (Figure 1C).

The focus of this review is on in vivo imaging and molecular imaging probes for applications in cell therapy. Therefore, in this review, we provide a brief discussion on the advantages and disadvantages of each imaging modality while giving a specific emphasis on MRI and the reporter gene approach. At the end of this review, we discuss future directions for applying molecular imaging in regenerative medicine and emphasize the importance of correlating cell graft conditions and clinical outcomes to advance regenerative medicine.

Literature search

In preparation for this review, we utilized search databases consisted of PubMed and Google Scholar. Search terms included but not limited to in vivo cell imaging, in vivo cell tracking, in vivo cell monitoring, molecular imaging, reporter gene, longitudinal monitoring, MRI reporter, PET reporter, and CT reporter while excluding drug delivery, patent, and agriculture. All the languages were included. The articles were systematically reviewed for relevance based on the title and abstract.

Basic requirements for an imaging probe/reporter for cell tracking

The characteristics and requirements of an ideal imaging probe/reporter were proposed by Frangioni and Hajjar more than a decade ago [40]. However, given the advancement in imaging technologies, emerging new applications and new imaging methods, natural progression, and paradigm shifts in the field, these information needs to be updated. We consider that the optimized imaging probe/reporters for cell tracking should have specific characteristics as summarized in Table 1. An ideal imaging probe/reporter should be biodegradable and safe for biological systems. Also, imaging probes/reporters should not impede the viability of the host cells. Although most imaging contrast materials used for cell labeling, such as nanoparticles, have shown promising results in tracking cell grafts, their long-term safety and biocompatibility are still under investigation. Furthermore, an imaging probe/reporter should have no or minimal impact on cell functions. In the cases of pluripotent stem cells or lineage-specific stem cells (i.e. neural stem cells), a probe/reporter should not affect the differentiation potential of the stem cell [41]. Currently, there is a need to establish a set of standardized functional assessment to evaluate the cell functions after the cell labeling with reporters. Some reports showed no effect on differentiation potential [41,44] while others reported a skewed preference for certain lineage-specific cell types [45,48] in the similar assessment. To enable tracking and monitoring cell grafts at the single-cell level and quantifying cell numbers, an ideal imaging probe/reporter and the accompanying imaging methods should also provide great sensitivity for the detection. High sensitivity is particularly important for monitoring migration of stem cells from the graft site in stem cell therapy. How far the grafted cells can migrate and whether they can localize to the targeted anatomic sites will be the critical evaluation points to assess the successful integration and functionality of the stem cells. To these days, the single cell detection is still limited in a few proof of principle studies, notably using micron-sized iron oxide particles (MPIOs) and superparamagnetic iron oxide (SPIO) particles to label cells for MRI tracking [49,50]. Since most cell-based therapies need longitudinal monitoring of transplanted cells, the signal persistence during the cell division and migration is another desirable property of an imaging probe/reporter. In the case of SPIO labeled cells, precise quantification remains a challenge because of the continued dilution of SPIO as cells divide, proliferate, and migrate. As required in any molecular and cellular imaging probes, cell tracking also needs to have high specificity with signals rising only from the grafted cells and not from other cells. When cells undergo division, apoptosis, or death, imaging probes used for labeling the cells can be released or lost from the cells and then picked up by adjacent cells or persist in the extracellular matrix and give false-positive signals. A genetically engineered imaging reporter can be expressed in the stem cell, and then stem cell grafts can be tracked longitudinally. Depending on the promoter used, the expression of the reporter gene can be restricted to a specific cell type or can be expressed constitutively. Many strategies have been developed, including enzymes, receptors, and iron-chelating proteins. While these approaches can provide information about the viability of the grafted cells and allows longitudinal tracking, epigenetic silencing or an immunogenic response by the host can occur [19]. Therefore, it is important that the signal or contrast is retained in the grafted cells or daughter cells.

Table 1.

Attributes of an ideal imaging probe/reporter for in vivo stem cell graft monitoring

1.	Nontoxic
2.	Maintenance of pluripotency
3.	High sensitivity
4.	Signal persistence
5.	Specificity
6.	Noninvasive

Imaging modalities for cell tracking

In vivo cell tracking prefers noninvasive imaging modalities, such as PET, CT, SPECT, US, BLI, FLI, and MRI. For each imaging modality, accumulation and amplification of a specific signal from the contrast materials in the cells make it possible to localize, track and quantify the cell grafts. The source of the signal and contrast, acquisition techniques, and instrumentation differ from one imaging modality to the other. Therefore, each imaging modality has its advantages and disadvantages regarding sensitivity, spatial and temporal resolution, and imaging depth. A brief summary of the characteristics of different imaging modalities is presented in Table 2. With the rapid advances in imaging methods and imaging probes/reporters, the parameters given here are only applicable to the current state of each technology.

Table 2.

Characteristics of imaging modalities

Imaging modality	Spectrum	Probe/reporter used	Type of visualization	Spatial resolution	Temporal resolution	Signal depth	Longi-tudinal	In clinic	Cost
SPECT	High-energy γ-rays	99mTc, 111In, 123I, NIS, NET	Whole-body	1~2 mm	min	Good	+	Yes	$$
PET	Low-energy γ-rays	18F, 124I, 64Cu, HSV-tk, NET	Whole-body	1~2 mm	10 sec~min	Good	+++	Yes	$$$$
CT	X-rays	125I, Gd	Whole-body	50~200 µm	min	Excellent	+	No	$$
US	High-frequency sound	Microbubbles, perfluorocarbons	Limited	1~2 mm	sec~min	mm~cm	+	No	$
BLI	Visible light	Luciferase	Whole-body*	3~5 mm, 3~5 µm*	min	1-2 cm	+++	No	$$
FLI	Near-infrared	QDs, Fluorescent proteins	Intravital microscope	3~5 mm, 2~3 µm*	sec~min	< 1 cm	++	No	$$
MRI	Radiowaves	Lanthanides, SPIO**, PEPE, Tyrosinase, β-galactosidase, LacZ, TFRC, FR, MagA	Whole-body	10~100 µm	min~hr	Excellent	+++	Yes	$$$
CEST-MRI	Radiowaves	HSV-tk, hPRM1, lanthanides, lipo-CEST	Whole-body	25~100 µm	min~hr	Excellent	+++	No	$$$

^*non-in vivo or small animal only;

^**including USPIO, MION, CLIO.

Abbreviations: NIS-sodium iodide symporter, NET-norepinephrine transporter, Gd-gadolinium, QD-quantum dots, SPIO-superparamagnetic iron oxide, PEPE-perfluoropolyether, TFRC-transferrin receptor, FR-ferritin, HSV-tk-herpes simplex virus type 1 thymidine kinase, hPRM1-human protamine-1.

Single-photon emission computed tomography (SPECT)

SPECT is widely available in clinical diagnostic imaging. It detects the γ-ray signal emitted from radioactive isotopes with long half-lives (t_1/2), such as ^99mTc, ¹¹¹In, and ¹²³I, using a rotating collimated gamma camera. The collected signal can be reconstructed as a 3-dimensional image. As a nuclear imaging method, SPECT has good sensitivity for imaging small numbers of cells with the ability to visualize up to 1 x 10⁴ labeled cells with a temporal resolution of minutes [51]. Moreover, SPECT has shown improved resolution of labeled cells in an anatomical context. Although SPECT is not as sensitive as PET [52], the longer halftime of radiotracers is beneficial for the applications of cell labeling and cell tracking. One of the advantages of SPECT is that it allows multi-spectral imaging using multiple radionuclides (e.g. ¹¹¹In and ^99mTC [53]) simultaneously detecting multiple biologic events.

In recent years, reporter genes, such as enzymatic conversion/retention and receptor-mediated targeting, have been developed for cell tracking applications with SPECT. Sodium iodide symporter (NIS) can be imaged with ¹²³I or ^99mTc for SPECT [54]. Norepinephrine transporter (NET), which can be labeled with ¹²⁴I-MIBG [55] as well as dopamine receptor and transporter, has also been used as a SPECT reporter [25]. These methods were applied to monitor neural stem cell [56] and cardiac stem cell [24] grafts, as well as to monitor neuronal differentiation [57].

While most studies reported no detrimental impact of isotopes used in SPECT imaging, one study found low labeling efficiency (32%), reduced viability, and complete impairment of proliferation and differentiation in CD34⁺ hematopoietic progenitor cells [58]. Another study involving human mesenchymal stem cells (hMSC) demonstrated that ¹¹¹In-oxiquinolon affected cell migration [59]. One major concern in using SPECT for cell tracking is that radioisotopes exhibit substantial efflux within 24 hours [53]. For applications in longitudinal monitoring and follow-up studies, the emitted γ-rays are potentially mutagenic and carcinogenic.

Positron emission tomography (PET)

PET is another popular clinical and preclinical imaging modality. It offers the most sensitive method for tracking relatively scarce cells with extraordinary sensitivity in the picomolar range (10^-11-10^-12 mol/l) [60,61]. The signal for PET is produced from positron-emitting radionuclides, such as ¹¹C, ¹³N, ¹⁸F, ¹²⁴I, and ⁸⁹Zr [52,62]. Upon the annihilation of a positron, the emission of two anti-parallel γ photons is detected by a sensitive photodetector. The signal is later computed for a spatial position with the intensity of the emission sources. Direct labeling and genetic reporter systems for tracking cells with PET have both been explored recently. In a recent study, mouse embryonic stem cells were labeled with widely available fludeoxyglucose ([¹⁸F]-FDG) to monitor retention of grafted cells in vivo [63]. Monitoring of neural stem cell migration has been reported with 3’-deoxy-3’-[¹⁸F]fluoro-L-thymidine [64].

Reporter gene and reporter probe paradigms have also been developed for PET. Herpes simplex virus type 1 thymidine kinase (HSV-tk) and probe (2’-fluoro-2’-deoxy-l-)3-D-arabinofuranosyl-5-iodouracil (FIAU) have been evaluated for myocardial gene therapy in pigs [65]. In addition to HSV-tk and FIAU, gene reporter and probe pairs, human sodium iodide symporter (NIS) with ¹²⁴I and human estrogen receptor ligand binding domain (hERL) with 16α-[¹⁸F] fluoro-17β-estradiol (¹⁸F-FES) have been evaluated for in vivo tracking of hMSC grafts in mouse [66,67].

However, applications of PET imaging are often limited by its substantial requirement in resource setup and complexity in the development of new tracers. Since it is a nuclear imaging method, there is a concern of possible mutagenic and carcinogenic effects of high energy γ photons. Comparing to MRI, another widely available clinically translatable cell tracking imaging modality, PET does not have as high a spatial resolution [61].

Bioluminescence imaging (BLI)

Bioluminescence imaging is one of the preclinical optical imaging methods that use luminescent light (peak wavelength ~450-600 nm) emitted from a product of an enzyme-mediated chemical reaction, i.e., luciferin oxidized by the enzyme luciferase in the presence of ATP and oxygen [68,69]. Genetic reporter luciferase and BLI are the most established and popular cell tracking imaging approach in rodent studies. For in vivo BLI imaging, animals are placed in a dark chamber with a sensitive photodetector, and D-luciferin is needed to be injected shortly before imaging. The emitted light is often detected by a charge-coupled device (CCD) camera. Among many luciferases identified and cloned, there are two isolated from two different organisms that are commonly employed: firefly (Photinus pyralis) and sea pansy (Renilla reniformis). Due to its structural and auto-oxidation properties, the luciferase isolated from the firefly is more broadly used for in vivo tracking. The feasibility of BLI relies basically on the ability to establish the cell lines that can incorporate and stably express the luciferase report gene for longitudinal monitoring [28]. BLI has made a great impact in the field of molecular and cellular imaging and continues to be one of the mostly applied imaging tools in preclinical studies, including cell tracking research. In a recent study, human neural progenitor cells (NPC) grafted in mouse brain was tracked in vivo for 12 weeks [29]. Also, the aforementioned study by Wolfs et al. [66] used BLI as one of the imaging modalities for their multimodal imaging of hMSC. BLI was also used to evaluate the engraftment efficiency, proliferation, and therapeutic potential of iPSC-derived cardiomyocytes in a mouse myocardial infarction model [70].

However, as an optical imaging method, the poor signal penetration and imaging depth limit the use of BLI mostly to rodents [40]. Since the wavelength used in typical BLI is 400-700 nm, the signal is highly susceptible to absorption and scattering in living tissue [40]. Even in mice, background signal can cause false-negative findings [71]. Also, the pharmacokinetics of luciferin has to be taken into account since each organ has a different absorption rate, catalysis rate, and elimination kinetics for luciferin [72]. Moreover, BLI requires injection of the high concentration of substrate compounds that are potentially immunogenic substances unlikely to be used in human [40].

Fluorescence imaging (FLI)

Unlike BLI using mostly genetic imaging reporter, FLI uses mostly organic/inorganic fluorophores (e.g., quantum dots) for labeling the cells but also can use genetically introduced reporters (e.g., green fluorescent protein, near-infrared fluorescent protein) [31]. The signal is produced by the fluorescence molecule when the molecule is excited by a specific incident wavelength and emits back light [31]. For in vivo monitoring of grafted cells, near-infrared (650-900 nm) fluorescent synthetic molecules and nanoparticles as well as near-infrared proteins, have shown great promise due to the relatively better signal penetration up to 10 mm comparing to other fluorescent light [32].

Comparing to BLI, FLI shows broad applications in preclinical studies given more choices of fluorescent probes available. For whole-body imaging, FLI suffers from the same limitations as BLI such as light scattering and signal absorption by surrounding tissue, which limits the depth of tissue to only the surface area. Even with tomographic imaging methods, spatial resolution is limited to approximately 1 mm [73]. However, the cytotoxicity, impact on host cell differentiation, signal penetrance, and light-scattering characteristics of such reporter need to be investigated.

Computed tomography (CT)

X-ray CT is the most available and affordable imaging modality. Yet, CT fell out of favor as an imaging modality for stem cell tracking because it has less soft tissue contrast for studying the soft tissue organs, and may require the use of a high concentration of high-density/high-atomic-number materials as contrast agents [40]. However, with the recent development of multi-imaging modality methods, CT has been used to track MSC grafts in rabbits [74].

Ultrasound (US)

The contrast for ultrasound is achieved by acoustic interfaces (e.g. microbubbles and perfluorocarbons) [40]. The advantages of ultrasound include low-cost, wide availability in most clinics and lack of long-term side effects. However, ultrasound has various limitations for cell tracking, such as poor anatomic coverage (unable to performing whole body imaging), limited contrast materials, the acoustic “shadowing” effect and limited signal penetrance depth [27]. Despite these limitations, US has been used in prostate stem cell monitoring with nanotubes [75] and in vivo neural progenitor cell tracking with microbubbles [26].

Magnetic resonance imaging (MRI)

Magnetic resonance imaging is a widely available clinical and preclinical imaging modality offering a superior spatial resolution, 3-dimensional imaging capability, and high soft tissue contrast for non-invasive in vivo tracking of cell grafts. Compare to SPECT and PET, MRI does not require the use of radioactive isotopes. Due to these advantages, MRI has become the most attractive imaging modality for tracking stem cells in vivo (Figure 1B).

Most commonly used MRI detects the signal that originates from mobile water protons (¹H) but it can be also used for tracking cells labeled with fluorinated molecules (¹⁹F) or other nuclei if sensitivity is sufficient. The contrast for MRI is typically generated by either manipulating pulse sequences to exploit differences in relaxation properties of water protons or introducing contrast agents (e.g., Gd³⁺ or iron oxide) that alter the relaxation properties of water protons in the region where contrast agents accumulate. When the external magnetic field is removed, magnetic moments re-align in the external magnetic field direction (B₀). The time it takes to recover net magnetization is called longitudinal relaxation time or spin-lattice (T₁) relaxation. The conventional MRI measures the relaxation of protons. The T₁ relaxation time depends on the mobility of the proton (spin-lattice) or the gyromagnetic ratio of the nucleus.

On the other hand, transverse relaxation time (T₂) describes the diminishing net transverse magnetization or the loss of spin coherence by dephasing of spins. MRI contrast agents are those paramagnetic materials that can accelerate the relaxation times of the protons affected, inducing either T₁ or T₂ relaxation of juxtapositioned protons generating hyper- or hypointense signal. T₁ contrast agents, such as lanthanide chelate Gd-DTPA, typically generates the hypointense signal from those water molecules recovered fast than background due to the presence of the contrast materials. The T₂ contrast agent, such as iron oxide nanoparticles, cause net transverse magnetization or signals diminishing fast under the effect of the contrast agents the, generating hypointense or “dark” contrast due to signal drop. Two factors affect T₂ relaxation: molecular interactions and local magnetic field inhomogeneity. Combinations of these artifacts result in hastened decay of transverse magnetization, referred to as T₂*.

Labeling cells for MRI tracking

There are two methods to label cells: direct and indirect (Figure 2, Table 3). The direct labeling method (Figure 3A) takes advantages of cell endocytosis of a contrast agent (e.g. SPIO) [149,160]. Stem cells are incubated in culture conditions with the contrast agent which can be taken up by the cells. The contrast agent can be surface modified with cell membrane receptors to enhance cellular uptake. Although the direct labeling method is relatively easy to employ, there are several limitations when used in cell tracking and longitudinal monitoring in vivo. First, the labeling agent is diluted as cells divide, and such dilution limits the duration of monitoring. Second, the contrast agent can be released from apoptotic cells and taken up by adjacent cells (i.e. false-positive signal) or even localize to the extracellular matrix. The first and second limitations of the direct labeling method were well demonstrated in a study involving tracking of lacZ expressing neural stem cells labeled with SPIO [76]. In this study, the rapid division of neural stem cell grafts resulted in the dilution of MRI contrast. In postmortem analysis, they further demonstrated a lack of MRI contrast in the stem cell-derived neuronal cell populations [76]. When transfection agents are sometimes used to assist the cellular uptake of the labeling agent, they can affect the motility, differentiation potential, viability, proliferation, and functionality [77].

Figure 2.

Different labeling strategies highlighting the main differences among different strategies.

Table 3.

Lists of direct and indirect probes/reporter

Probing method	Probe/reporter	Imaging modality	Toxicity reported	Research area	FDA approved
Direct	Gd3+ [149] or Mn2+ [150]	MRI-T1 (+)	Yes	Viability, migration	Yes
	SPIO [151,152], USPIO [153], CLIO [154], MION [155]	MRI-T2 (-)	Yes	Viability, migration	Yes
	PFC [156,157]	19F MRI	Yes	Viability, migration	Yes
	QD [158]	FLI	Yes	Immunology (homing)	Yes
	Fluorescent probe [159]	Intravital microscope	No	Migration, cell-cell interaction, infiltration, homing	Yes
	111In [51], 99mTc [53]	SPECT	Yes	Homing, cell therapy efficacy	Yes
	18F [64], 64Cu [160]	PET	Yes	Homing	Yes
Indirect	Ferritin/transferrin receptor [35,161]	MRI-T2/T2* (-)	No	Viability, migration, differentiation	No
	β-galactosidase [79]	MRI-T1/T2/T2* (-)	Yes	Viability, migration	No
	Tyrosinase [23]	MRI-T1 (-)	Yes	Viability	No
	MagA [37,38,116,117,119]	MRI-T2/T2* (-)	No	Viability	No
	Plasma membrane bound reporter peptide [162]	MRI-T1/T2/T2* (+/-)	Yes/No	Viability, differentiation	No
	Lysine-rich protein [131]	CEST	No	Viability, migration, pH sensing	No
	Fluorescent protein [163]	FLI	No	Migration, cell-cell interaction, infiltration, homing	No
	Luciferase [29,57,163]	BLI	No	Migration	No
	HSV-tk (18F) [65,66]	PET/SPECT	Yes	Viability, migration	No
	NET [55]	PET/SPECT	Yes	Migration, homing	No
	NIS [56,66]	SPECT	Yes	Migration, homing	No
	Dopamine 2 [164]	PET	Yes	Viability	No
	Somatostatin [165]	PET/SPECT	Yes	Viability	No
	DMT1 [100]	MRI-T1 (+)	No	Viability	No

For MRI- (-) for negative contrast and (+) for positive contrast. Abbreviations: MION-monocrystalline iron oxide, USPIO-ultrasmall superparamagnetic iron oxide, CLIO-cross-linked iron oxide, PFC-perfluorocarbon, PEPE-perfluoropolyether, SPIO-superparamagnetic iron oxide, NIS-sodium iodide symporter, NET-norepinephrine transporter, QD-quantum dots, DMT1-Divalent metal transporter 1.

Figure 3.

Direct and indirect labeling methods. A. In the direct labeling method, cells are incubated with a contrast agent (e.g. SPIO). Cells actively take up the contrast agent and are harvested after a specific incubation period. The harvested cells are injected into a study subject, and the cells are tracked using the imaging modality of choice. The major limitation of the direct labeling method is the dilution of contrast agent. B. For the indirect labeling method, a reporter gene is transduced into the cells, and cells expressing the reporter gene are injected into a study subject. The major advantage of the indirect labeling method is that the contrast agent does not get diluted by cell division, allowing longitudinal monitoring.

To address these limitations, the indirect labeling method (Figure 3B) employing molecular cloning strategies to produce endogenous MRI contrast materials, such as iron chelating or storing proteins intracellularly, or specific cell targets for MRI contrast probes, has been explored [161,165]. The indirect labeling method can be further divided into exogenous (reporter mediated) and endogenous (de novo) methods (Figure 4) [17,18]. The exogenous method requires administration of an imaging probe (Figure 4A). One limitation of the exogenous labeling approach is that the sufficient delivery of the probe for high contrast and sensitivity highly depends on the targeted accumulation and pharmacokinetics of the imaging probe [18]. In contrast, the endogenous method (Figure 4B) depends on developing and engineering a proper gene construct for a genetic imaging reporter. The endogenous reporters, however, have limited MRI contrast. There is also concern that a continuous expression of the reporter gene can result in unforeseen consequences including cytotoxicity. In one study, chronic overexpression of H-ferritin (FTH1) resulted in a neurodegeneration phenotype [78]. Ideally, the expression of a reporter gene could be regulated by an inducible promoter, such as Tet-On or Tet-Off switches (Figure 4C). The controllable expression approach adds safety by minimizing the impact (on proliferation, migration, and differentiation) of constitutive expression of the reporter. However, both exogenous and endogenous methods share some limitations that include low dynamic resolution because the lifetime of the signal depends on the clearance of the reporter and genetic manipulation of stem cells is required, making clinical application unlikely.

Figure 4.

Exogenous and endogenous labeling methods. A. Exogenous labeling methods require injection of a contrast agent. Either binding of the contrast agent or activation by enzyme generates contrast. B, C. Endogenous labeling methods. B. The constitutive expression does not require injection of contrast agent but lacks the mechanism to regulate the expression. C. An inducible promoter allows expression of the reporter gene when monitoring is required.

Exogenous MRI reporters

There are two types of exogenous reporters to date: enzymes and engineered cell surface peptides (Figure 5). The first exogenous MRI reporter gene was reported in 1997 [79]. In this study, a molecule, Egad, was synthesized containing gadolinium (III), Gd³⁺, in a cage composed of galactopyranose ring and tetraazamacrocycle. When the molecule was exposed to β-Galactosidase, galactopyranose was removed from the molecule, and the exposure of Gd³⁺ enhanced longitudinal relaxation of water molecules, resulting in positive T₁ contrast [79]. A contrast agent containing gadolinium, gadopentetate dimeglumine, was the first FDA-approved MRI contrast agent [80]. Gadolinium has been used to track mesenchymal stem cells [81], hematopoietic progenitor cells [82], and endothelial progenitor cells [83]. However, it has been reported that gadolinium could induce nephrogenic systemic fibrosis and even death [84]. Manganese (Mn²⁺) is another T₁ contrast agent with positive contrast. A recent study used silica-coated manganese oxide (MnO) nanoparticles to track mesenchymal stem cells [85]. The lacZ/β-galactosidase system has the versatility of providing T₁, T₂, or T₂* contrast depending on the contrast materials used as the probe. It has been demonstrated that β-galactosidase expressing MSC can be tracked in vivo by S-Gal™ as an imaging probe, which enhanced T₂ and T₂* MR contrast [86]; yet several limitations were observed for the lacZ/β-galactosidase system, including the efficient delivery of the probe which depends on the pharmacokinetics of the probe, the limited uptake of the probes by the targeted cells as probe concentration diluted rapidly once injected, the nonspecific cellular uptake of the injected imaging contrast agent, and the possible effects on cell viability [17]. Superparamagnetic iron oxide particles (SPIOs) offer ultrahigh T₂ relaxivity hence produce greater sensitivity, which make it optimal to track a small number of cells such as neural cells [87]. Since non-phagocytic cells do not internalize the SPIO nanoparticles, an alternative approach is used to engineered cell surface peptides. An engineered surface protein expressing hemagglutinin (HA), luciferase, and myelocytomatosis (myc) (i.e. HA-fluc-myc) was developed as an MRI reporter [88]. HA and myc serve as the molecular target for antibody conjugated with SPIO. Both human and mouse ESCs expressing HA-fluc-myc have shown significant hypointense signal in proliferating ESCs and teratoma [88]. Other surface receptors, such as a biotinylated transmembrane receptor (BAP-TM) and anti-polyethylene glycol (PEG) peptide, have also been developed as genetic MRI reporters and could be detected in cancer cells [89,90]. In a few early clinical trial studies, SPIO nanoparticles were used to track homing of mesenchymal stem cells to central nerve systems in individuals with multiple sclerosis and amyotrophic lateral sclerosis [91]. Also, Ferumoxytol®, an FDA approved ultrasmall superparamagnetic iron oxide nanoparticles (USPIOs) used as an iron supplement for treating anemia, was used to label and track human neural stem cells in mouse brain [92]. In 2010, Westmeyer et al. developed an interesting method using a secreted enzyme as a genetic MRI reporter. This study revealed that secreted alkaline phosphatase (SEAP) could be used as a genetic MRI reporter as it promotes aggregation of SPIO [93]. Most recently, ‘hot spot’ highly shifted proton (HSP) MRI reporter was developed [94]. The tumor cells were labeled with dysprosium (Dy)- or thulium (Tm)- 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetramethy-1,4,7,10-tetraacetic acid (DOTMA) and was imaged by the ultra-short echo time (UTE) sequence. Dramatically shortened T₁ contrast could be demonstrated [94]. This method has a detection limit of about 1 x 10⁴ cells [94].

Figure 5.

Exogenous and endogenous genetic reporters.

Endogenous MRI reporters

Endogenous MRI reporters use a gene or set of genes that do not require exogenous substrates to generate MRI contrast. The first reported endogenous MRI reporter was transferrin receptor (Tfrc) [33]. When Tfrc was overexpressed in cells, R₂* contrast was significantly enhanced [33]. Later, there was improved contrast when transferrin-associated SPIO was used, which takes advantage of an endogenous approach with the exogenous approach [95]. Although iron is crucial for normal cell function, internal free labile iron is strictly maintained by networks of genes [96]. Other genes involved in iron homeostasis, like transferrin (Trf) and ferritin, have also been explored. Transferrin is involved in transporting iron into the cell while ferritin is involved in storing iron in the cell. Of all the genes involved in iron homeostasis, ferritin has proved to be the most popular [17]. The overexpression of a heavy chain of ferritin (FTH1) is found to generate T₁ and T₂ contrast in vitro and in vivo [34]. One study has expressed human FTH1 as a reporter for in vivo tracking of cardiac stem cells [97]. Cell survival was monitored up to 4 weeks after grafting using T₂-weighted imaging and T₂* mapping [97]. To demonstrate the sensitivity of the system, human FTH1 and Tfrc were also coupled and overexpressed in murine neural stem cells. The overexpression resulted in enhanced transverse relaxivities observed in both R₂ (i.e., 1/T₁) and R₂* (i.e., 1/T₂*) [97]. Ten days after the graft, T₂*-weighted imaging yielded increased contrast only with 2,500 cells [97]. FTH1 has been used for stem cell tracking, such as in embryonic stem cells [35] and myoblasts [98]. These studies have shown significant MR signal enhancement between 14 to 21 days after injection. However, further investigation is needed to assess the impact of overexpressing FTH1 on iron homeostasis in different cell types. The overexpression of Tfrc is known to activate iron overload response, and the overexpression of FTH1 or FTL1 activates iron deficiency response [18].

In 1997, another endogenous MRI reporter gene, tyrosinase, was reported [99]. Tyrosinase is the primary enzyme involved in melanin production, and the overexpression of tyrosinase results in melanin accumulation. Melanin is excellent at sequestering paramagnetic ions, which results in T₁ contrast [99]. Although tyrosinase has not been used in stem cells, interest in using tyrosinase as an MRI reporter has risen due to its ability to be employed in multimodal imaging, combining MRI, photoacoustic, and PET [23].

A divalent metal transporter-1 (DMT1) was tested as a novel MRI reporter gene. The overexpression of DMT1 enhanced manganese uptake and resulted in a significant increase in R1 [100]. Unlike hypointense contrast caused by the attenuation of MR signal with iron homeostasis associated genes, DMT1 resulted in hyperintense contrast with signal enhancement, which was correlated with better sensitivity. However, intraperitoneal injection of MnCl₂ was needed to achieve better contrast. Also, the systemic and cellular impact of exposure to manganese must be further investigated.

Another possible endogenous MRI reporter candidates are the genes involved in magnetosome formation in bacteria. Magnetosome is a magnetite-enriched organelle. Magnetite crystals are considered to be an excellent MRI contrast agent because they can drastically shorten transverse relaxation time (T₂ and T₂*). The biosynthesis of magnetosomes was achieved in non-magnetosome-forming bacteria by an elaborate stepwise recombination method, which elucidated the genes responsible for magnetosome formation [101]. Magnetite nanocrystals are considered to be superparamagnetic, with much greater magnetic susceptibility than the paramagnets. Superparamagnetic iron oxide nanocrystals may form in mammalian cells through biosynthesis of magnetosomes while protecting the host cell by isolating magnetite in membrane-bounded magnetosome organelles (Figure 6) [52]. The formation of magnetosome starts with invagination of the inner plasma membrane. The newly formed vesicle is attached to the cytoskeletal structure in the cell. Next, iron is actively transported into the vesicle. Lastly, the iron in the vesicle is actively incorporated into the magnetite crystal, and the shapes and sizes of the crystals are determined by the specific species [102]. However, the lack of mammalian homologs of genes involved in magnetosome formation could lead to immune reactions. The formation of the magnetosome is controlled by a set of proteins with specific biological functions. These proteins are unique to magnetotactic bacteria (MTB) and are encoded by genes identified by genomic comparison of four species of MTB with nonmagnetotactic bacteria [103]. Among 28 genes identified, 18 were located within the magnetosome genomic island (MAI). There are some operons in the MAI: magnetosome membrane (Mam), magnetic particle membrane specific (Mms), and a monocistronic MamW [104]. Also, there are operons outside of the MAI, and these are magnetotaxis (Mtx) and magnetosome membrane (Mme) [102]. The functions of several operons are still under investigation. Some of these operons have an apparent association with the biosynthesis of magnetosomes while others have little or no association. Genes that are associated with magnetosome formation are listed in Table 4 with their putative functions [166,172]. In Magnetospirillum magneticum (strain AMB-1), only the genes in the mamAB operon are found to be essential for magnetosome formation [105]. The genes involved in biomineralization and iron transport are particularly interesting from the MR imaging perspective. There are a number of genes found to be involved in biomineralization: mms6, mad (10, 11, 12, 23, 25), and mam (G, F, D, C). Of these, mms6, identified in magnetite crystals, has been studied extensively [106]. mms6 regulates the mineralization of magnetites by binding to iron on the C-terminal domain [107]. Recently, a study reported using mms6 as an MRI reporter leading to a significant increase in R₂ and offering potential promise for in vivo tracking of tumor cell grafts [108]. However, the efficacy of mm6 in stem cell graft tracking regarding its impact on toxicity, proliferation, differentiation, and migration still need to be evaluated. If mms6 is similar to a ferritin system (iron chelator), magA and chapA act more like a transferrin receptor. When overexpressed, iron transporters might increase the iron content of the cell, allowing the cell to be monitored using MRI. For these reasons, chapA and magA have been investigated as MRI reporters. MagA is a membrane iron transporter protein involved in magnetosome formation in magnetotactic bacteria [109,111]. magA is homologous to Na⁺/H⁺ transporter, ferrous transporter, CPA2, KefC, and KefB (Table 5) [112]. A member of KefB superfamily, 2a37, protein domain is highly conserved in MagA (Table 6). KefB is a glutathione-regulated potassium efflux system [113] with a metal-binding domain (Figure 7A, 7B) [114,115], which suggests that MagA is a membrane protein with possible metal transportability. The MagA predicted model and c2k3ca (metal transporter templates) were aligned to create a superposition model to illustrate a possible iron-binding domain (Figure 7B). Although the complete reconstruction of a magnetosome in foreign cells has not been achieved, a number of studies have found that the overexpression of MagA increased MRI contrast [37,38,116,117].

Figure 6.

A current model of the magnetosome formation process. A. The first step of magnetosome formation is invagination and vesicle formation. The initial invagination process is activated by MamI, L, Q, and B. A vesicle is formed by recruiting additional proteins (e.g. MamK and E). B. After the vesicles are formed, they are aligned on a chain via interaction between MamK, cytoskeletal structure, and MamJ, a membrane-bound protein. Mm6 is involved in the biomineralization process. C. A magnetite crystal starts to form with an increase of iron content in the vesicle, with many proteins involved in the process (i.e. MamA, G, F, D, and C). MamP appears to be involved in the size restriction.

Table 4.

Magnetosome-associated genes

Gene name	Essential	Process involved
magA [110], feoB [104]	Somewhat [105]	Iron transporter
chpA [109]	Yes [105]	Copper-dependent, high-affinity iron transporter
mamB, M [166]	Yes [105]	Biomineralization and membrane assembly
mamH [167], N [103]	No [105]	Major facilitator superfamily of transporter
nir [168] l, nap [169]	Yes [105]	Reduction of nitrate to nitric oxide (oxidizing ferrous iron)
mamE, P, T [170], O [103]	Yes [105]	Magnetochrome/electron transport chain
mamX [167]	No [105]	Magnetochrome/electron transport chain/iron reductase
mamZ [167]	No [105]	Iron reductase
mms6 [106]	No [105]	Regulation of the mineralization of iron
mad25, 23, 10, 11, 12 [107]	No [105]	Shape and size regulation of biomineralization
mamG, F, D, C [171]	No [171]	Biomineralization/regulates size of the magnetite crystal
mamQ [172]	Yes [105]	Biomineralization
mamJ, K [172]	Yes [105]	Cytoskeletal structure
mamI, L [172]	Yes [105]	Vesicle formation

Table 5.

magA homology BLAST search result

Gene	Organism	Function	Max score	Identity
Na+/H+ exchanger	Magnetospirillum gryphiswaldense	Sodium/hydrogen exchanger	544	69%
Kef-type K+ transport system	Caenispirillum salinarum	K+ transporter	402	55%
Ferrous transporter	Candidatus Odyssella thessalonicensis	Fe2+ transporter	305	44%
Na+/H+ antiporter/ferrous transporter	Nitratireductor indicus	Sodium/hydrogen exchanger and ferrous transporter	304	44%
Na+/H+ antiporter/ferrous transporter	Fulvimarina pelagi	Sodium/hydrogen exchanger and ferrous transporter	254	40%
CPA2	Glaciecola chathamensis	monovalent cation/H+ antiporter	217	38%
Na+/H+ exchanger	Mariprofundus ferrooxydans	Sodium/hydrogen exchanger	206	35%
KefC	Francisella sp.	Glutathione-regulated potassium-efflux	214	31%
KefB	Nitritalea halalkaliphila	Potassium transporter	189	31%

Table 6.

MagA conserved domain

Conserved domain
TM PBP1	Transmembrane subunit of periplasmic binding protein (PBP)-dependent ATP-Binding Cassette (ABC) transporters
Na+/H+ exchanger	Sodium/hydrogen exchanger superfamily
KefC	Kef-type K+ transport system (inorganic ion transport and metabolism)
2a37	Transporter, monovalent cation/proton antiporter-2 (CPA2) family-transport and binding proteins, cations and iron-carrying compounds
RosB	Kef-type K+ transport system, predicted NAD-binding component (inorganic ion transport and metabolism)
KefB	Kef-type K+ system, membrane component (inorganic ion transport and metabolism)
NhaP	NhaP-type Na+/H+ and K+/H+ antiporters (inorganic ion transport and metabolism)

Figure 7.

MagA structure and longitudinal monitoring of stem cell graft using inducible MagA as a genetic MRI reporter. A. Phyre server-generated MagA 3D structure. B. Superimposed model of MagA and c2k3ca. C. A representative MRI image of a single mouse, with the status of MagA expression at the top of images. MRI images were taken in a 7-day interval. D. Regions of interest (ROIs) analysis of signal intensity showing significant hypointense signal with the “On” states, while no such difference was observed for the “Off” state [119].

Compared to the constitutive expression of an MRI reporter gene, an inducible system can have several advantages. First, controllable expression of an MRI reporter can minimize the adverse effects of constitutive expression of a reporter gene (e.g. toxicity, reduced proliferation rate, possible impact on differentiation potential). Another advantage is the use of an inducing agent with well-documented pharmacokinetics. One of the agents, doxycycline, has been widely used in controllable expression systems and has shown low toxicity and the ability to cross the blood-brain barrier, as well as the placenta barrier [118]. Inducible MRI reporters have been reported in in vivo monitoring of cell grafts. Cohen et al. have achieved in vivo monitoring of C6 glioma tumors expressing TET-EGFP-HA-ferritin [34]. By employing the Tet-Off system, they were able to illustrate that the overexpression of murine FTH resulted in significant R₂ relaxation up to 28 days post-transplantation [34]. Zurkiya et al. demonstrated the efficacy of monitoring tetracycline-inducible (Tet-On) MagA in human embryonic kidney cells (293FT) transplanted in the striatum of mice [37]. Tet-On MagA was also used for repetitive longitudinal monitoring of intracranial mouse embryonic stem cell (mESC) grafts in vivo [119]. This study took advantage of both the MagA system and inducible Tet-On system to regulate reporter expression and demonstrated repetitive monitoring correlation to the status of reporter expression [119].

Another genetic MRI reporter, ferritin, has also shown the efficacy of using an inducible promoter. Nasopharyngeal carcinoma cells expressing human ferritin heavy chain (FTH1) under the regulation of the Tet-Off system were used for in vivo monitoring of cell grafts [120]. Feng et al. observed a significant increase in transverse relaxivity (R₂) with the overexpression of FTH1. The proliferation, cytotoxicity, apoptosis, and migration of the cell could be assessed. The induction of gene expression in a cell graft was also detected by the genetic MRI reporter. In 2007, Cohen et al. generated transgenic mice with the TET:EGFP-HA-ferritin (tet-hfer) transgene [121]. By mating with mice expressing the tetracycline transactivator (tTA) under a tissue-specific promoter (e.g. vascular endothelial (VE) cadherin promoter and liver activator protein (LAP) promoter), expression of the tissue-specific genes was monitored with MRI [121]. In a more recent study, Rohani et al. compared the ability of MagA and FTH1+FTL1 to enhance the MR contrast in vivo. Here, MagA or FTH1+FTL1 lacking iron response element was expressed in human breast/melanoma (MDA-MB-435) cells and used for repetitive imaging of a tumor. MagA expression resulted in similar contrast to FTH1-FTL1-expressing cells and exhibited contrast enhancement up to 20 days [38]. Also, the same group was able to present similar results in vitro showing similar contrast from the cells expressing either MagA or FTH+FTL1 [117]. In our lab, we have demonstrated longitudinal and inducible monitoring of an intracranial stem cell graft (Figure 7C, 7D) [119]. The tetracycline-inducible promoter (Tet-On) was used to control the expression of MagA in mESC, hence controlling the level of MRI contrast. Our study has findings consistent with previous reports: increased iron content with MagA expression and enhanced relaxation rates in MagA-expressing tissue [119]. In one recent study, a MagA transgenic mouse was generated and showed iron accumulation and deposition of nanoparticles in various tissues by MRI [122]. This study also showed no apparent pathological symptoms in any organs and even showed attenuated oxidative damage induced by iron overload [122].

Chemical exchange saturation transfer (CEST)

Recently, chemical exchange saturation transfer (CEST), or CEST-MRI, has become an attractive imaging modality with the development of tailored synthetic peptides. Exogenous or endogenous compounds with different resonate frequency from the surrounding water molecules are selectively saturated using radiofrequency (RF) pulses. The saturated signal is subsequently transferred from the compound to surrounding water. The transfer of this saturation is detected through the water signal [123]. Saturation transfer was first measured by Forsen & Hoffman in 1963, by measuring proton transfer rates between salicylaldehyde and water [124]. In 2000, Ward and coworkers reported that many diamagnetic molecules and water protons can be used to enhance the sensitivity to detect low concentration solutes using the process of saturation transfer [125].

CEST-MRI can be explained in a simple two-pool model. Pool water (W) represents the high-concentration (about 110 M) water protons, and Pool solute (S) represents the molecules with the low-concentration (µM-mM range) exchangeable protons, such as hydroxyl, amine, or amide protons. A RF is applied at the resonance frequency of protons in Pool S. Due to difference in saturations, Pool W exchanges saturated protons with protons in Pool S leading to a minute decrease (µM to mM range) in the signal intensity of the protons in Pool W. It should be noted that a single transfer of saturation will not produce significant water signal saturation in Pool W. However, since the Pool W is much larger than the Pool S, each solute proton is replaced by a non-saturated water proton then again saturated. The detectable water signal reduction is generated when the exchange rate of the solute proton is sufficiently fast (more than tens of Hz), and T₁ of water proton is sufficiently long (seconds). This prolonged irradiation causes the accumulation of saturation effect on Pool W. To better understand the basic mechanism of CEST-MRI, we refer the readers to some recent reviews [126,129]. CEST-MRI opens the door for the discovery of endogenous and exogenous molecular contrast agents expanding MRI to multi-frequency detection or “multi-color” imaging [130].

Based on the principle of CEST-MRI, Gilad et al., developed an artificial gene, the lysine-rich protein (LRP), as a CEST-MRI reporter. The LRP provides a high density of amide protons, which can be detected via amide proton CEST-MRI. Through hydrogen bonding, amide proton exchange with bulk water molecules enables detection of LRP rich cells without requiring administration of a cognate probe [131]. LRP was overexpressed by the cytomegalovirus (CMV) promoter and gene-3 promoter (PEG-3 promoter), and an increase of CEST contrast was clearly visible in 9 L tumors model [132]. Using similar CSET contrast strategy, super positively charged mutants of green fluorescent protein (GFP) has also shown a dramatically improved CEST-MRI contrast compared to their wild type counterparts [133].

One of the potential applications of amide proton CEST imaging with the LRP reporter gene is to monitor the migration of the cancer cells as well as tracking stem cells. Oncolytic virus (OV) therapy has been developed in which a replicating virus is injected and selectively enters cancer cells. Therefore, replicated virus either kills the cells or induce an immune reaction capable of killing non-infected cells [134]. Farrar et al. demonstrated that engineered herpes simplex-derived oncolytic virus (G47Δ) carrying LRP reporter gene can be used for CEST imaging without affecting virus efficiency [135]. The process of G47Δ infecting the cell or replicating in tissue could be detected both in vitro and in vivo by using LRP reporter gene and CEST-MRI (Figure 8) [135]. The main strategy is to maximize the saturate-able protons in the Pool S. Therefore, amide group has been used in CEST-MRI. Amide proton transfer-weighted (APTw) image, a CEST-MRI method, has been utilized to show astrogliosis in vivo [136].

Figure 8.

CEST images assess cancer treatment response. Farrar et al., [134] evaluate whether the lysine-rich protein (LRP) MRI reporter gene can be engineered into G47Δ, a herpes simplex-derived oncolytic virus that is currently being tested in clinical trials, and can be in vivo tracked by CEST MRI. Representative MTRasym maps (color scale) acquired with a saturation frequency offset of 3.6 ppm and overlaid onto the associated T₂-weighted images at baseline (A and C) and 8 hours after injection of G47∆-LRP (B) and G47∆-empty virus (D). A significant increase in MTRasym is observed after virus injection (B) but not in G47∆-empty virus (D).

Despite the attractive features of CEST MRI and its potential to be employed for translational regenerative therapies without the need for exogenous contrast agents [18], it needs further development and investigations to address its limitations in complex post-processing methods and relative low signal-to-noise ratio.

Multimodal imaging approaches

Current developments have focused on molecular probes that allow multimodal imaging, which makes use of two or more imaging techniques (e.g. MRI and PET). The idea of developing multimodal imaging probes is to combine the strengths of the individual imaging modalities, for example the metabolic information or sensitivity from PET with anatomic details of MRI, to overcome the limitations associated with single imaging modalities. Using genetic engineering approaches, fusion proteins are developed as multimodal molecular imaging reporters. Several studies have investigated double or even triple multimodal imaging with fusion protein reporters. The early study reported the development of optical and PET-like hrl-mrfp-ttk [137] or HSV1-TF/GFP/Fluc [138] and showed promising result in monitoring tumor cells in vivo. A innovative approach was investigated by Lewis et al. using DMT1. A radioactive isotope of manganese, ⁵²Mn, was produced to track DMT1 over-expressing human neural progenitor cells [139]. However, PET was failed to produce image in vivo possibly due to poor biodistribution of ⁵²Mn in the brain while promising result was observed ex vivo via autoradiography [139]. Similarly, a tricistronic human norepinephrine transporter (hNET) combined with red-shifted firefly luciferase (Fluc) and a small marker gene RQR8 was used as a BLI/SPECT reporter [140]. The grafted cells were readily visible after administrating ¹²³I-meta-iodobenzyl guanidine (MIBG) [140].

Triple modality molecular imaging capability was demonstrated by Qin et al. [23] using tyrosinase derived genetic reporter. In this system, human breast cancer cells (MCF-7) was transfected with human tyrosinase (TYR), the key enzyme in melanin production. Overproduction of melanin could serve as an excellent imaging probe for photoacoustic imaging (PAI), MRI and PET. Melanin has a broad optical absorption spectrum with significant absorption at near-infrared (NIR) wavelengths, which allows for good tissue penetration. Therefore, it has been demonstrated as an excellent endogenous contrast agent for PAI [141]. Also, melanin has a high affinity for metal ions including Fe³ [142] thus serve as a promise MRI contrast agent. Finally, a special probe targeting melanin, N-(2-(diethylamino)ethyl)-¹⁸F-5-fluoropicolinamide, could be employed for PET imaging. Therefore, this probe can be effectively used to detect TYR reporter gene expression through melanin production [23]. This study provides an excellent example of rational design and use of fusion protein methods to make a genetic engineering reporter for imaging molecular and genetic events in vivo.

Perspectives

With multiple imaging modalities now available for molecular imaging and the unmet need in cell tracking for the development of cell-based therapy, development of molecular imaging reporters for cell tracking remains to be of great interest but with tremendous challenges. Compared to direct labeling methods, indirect endogenous labeling methods have advantages that are critical for stem cell monitoring, including the ability to longitudinally monitor the grafted cells and specificity for rapidly dividing cells. Considering recent stem cell therapy evaluations failed due to poor cell survival, immunological rejections, tumor development, and poor functional integration [143], a reliable genetically engineered MRI reporter will be an important tool, especially when using large animal models in preclinical research. The application of a genetic MRI reporter includes tracking cell migration, proliferation/viability imaging, neurogenesis imaging, myocardial stem cell imaging, cancer stem cell imaging, immune cell imaging, and monitoring of gene expression and differentiation [17,144,145]. The monitoring of grafted cells is important regarding evaluating therapeutic efficacy. Due to the limitations associated with current direct and exogenous monitoring methods, a genetic reporter will likely become a key to usher in the successful translation of regenerative medicine.

Other than the previously mentioned concerns (e.g. increasing cytotoxicity and affecting differentiation potential) of expressing a genetic MRI reporter, there are other pitfalls that might be associated with using a genetic MRI reporter. First, since the expression of a reporter gene requires genetic modifications, the integration site of the reporter gene might affect its expression, disrupt normal cell function, be epigenetically silenced, and even promote malignant transformation [17]. One way to overcome this problem is by using targeted gene insertion (knock-in). However, the efficiency of generating knock-ins is low. Also, inserting non-human gene such as MagA will not be feasible for human clinical applications. However, systems such as ferritin and DMT1 can be used in clinical settings if the safety of the cells expressing those proteins can be ensured. Second, tissue specificity has to be examined especially in stem cells because the expression of the reporter gene later in differentiated cells can be affected by epigenetic silencing. Also, overexpressing a gene, introducing a foreign gene, or increasing iron content in a cell might trigger a host immune response, and immune cells like macrophages might increase iron concentration at the graft site to create a nonspecific signal [17]. As of DMT1, different tissues might have different biodistribution of the contrast agent, which results in low contrast. Therefore, depending on the application and interested organs, the researcher might have to choose the best available method. Third, when an inducible promoter is used, its impact on temporal resolution due to the length of time associated with translating the protein and accumulating enough contrast agents, the turnover rate of the reporter and its clearance have to be investigated [17]. Since the intrinsic limitation of endogenous labeling method is the low sensitivity, multiple strategies can be explored to overcome this problem. The simplest method will be to use more powerful MR scanners. Studies using ferritin have demonstrated a linear increase in contrast enhancement with the increase in the field strength [146]. However, the high field strength scanners (i.e., higher than 7T) may not be available for imaging larger animals or humans. To increase the contrast signal for MRI at the current clinical field strength (i.e., 3T), multiple genes can be expressed. Previously, researchers expressed multiple ferritin/transferrin genes to increase the contrast [147,148]. Similar methods can be employed for MagA and DMT1 using different combinations of magnetosome-associated genes and iron homeostasis-associated genes. Also, genes can be engineered, so that the resultants proteins can uptake or retain more of contrast agents. Overcoming these limitations will require more investigation involving multi-disciplinary efforts.

Conclusions

With the recent growing interest and demand for in vivo monitoring of cell grafts, the field of molecular imaging for cell tracking is expected to grow further. With a variety of disease model systems and recent preclinical and clinical trials of stem cell replacement therapies, the tracking of therapeutic cells and monitoring of cell grafts are becoming critical for the future development of these new therapies and strategies. The effort to develop efficient and applicable MRI reporter and novel imaging methods, such as CEST reporter peptide, genetic MRI reporter, and multimodal monitoring methods, are expected to increase significantly. While the clinical translation of imaging reporter may take a long road with more obstacles to overcome, the development of such reporters will indeed enable a further understanding of the basic biology of grafted cells regarding proliferation, survival, migration, differentiation, and functional integration. The eventual development of a safe and effective in vivo reporter will help us design better experiments for preclinical studies, and hence facilitate the clinical translation of cell replacement therapy.

Disclosure of conflict of interest

None.