Engineering molecular imaging strategies for regenerative medicine

Abstract The reshaping of the world's aging population has created an urgent need for therapies for chronic diseases. Regenerative medicine offers a ray of hope, and its complex solutions include material, cellular, or tissue systems. We review basics of regenerative medicine/stem cells and describe how the field of molecular imaging, which is based on quantitative, noninvasive, imaging of biological events in living subjects, can be applied to regenerative medicine in order to interrogate tissues in innovative, informative, and personalized ways. We consider aspects of regenerative medicine for which molecular imaging will benefit. Next, genetic and nanoparticle‐based cell imaging strategies are discussed in detail, with modalities like magnetic resonance imaging, optical imaging (near infra‐red, bioluminescence), raman microscopy, and photoacoustic microscopy), ultrasound, computed tomography, single‐photon computed tomography, and positron emission tomography. We conclude with a discussion of “next generation” molecular imaging strategies, including imaging host tissues prior to cell/tissue transplantation.


| OVERVIEW
Regenerative medicine is a field that utilizes complex therapies comprised of cells and/or materials, which address failing tissues. Molecular imaging is a branch of radiology that focuses on imaging biology (receptors, biological pathways) rather than anatomy (anatomical imaging like computed tomography [CT] or magnetic resonance imaging [MRI]) or physiology (functional imaging). The goal of molecular imaging is noninvasive imaging, detection, or interrogation of biomolecular events in living subjects, to further understand biology, Abbreviations: ADME, absorption distribution metabolism excretion; ASC, adult stem cell; AuNP, gold (Au) nanoparticle; BLI, bioluminescence imaging; BRET, bioluminescence resonance energy transfer; CAG, chicken beta-actin/rabbit beta globin hybrid promoter; CAR-T, chimeric antigen receptor T cell; CCD, charged coupled device; CMV, cytomegalovirus; CSC, cancer stem cell; CT, computed tomography; ESC, embryonic stem cell 18F-FHBG 9-(4-18F-fluoro-3-[hydroxymethyl]butyl)guanine; Fluc, firefly luciferase; Gluc, Gaussia luciferase; GFP, green fluorescent protein; HSC, hematopoietic stem cells; HSV, herpes simplex virus; iPSC, induced pluripotent stem cell; IVM, intravital microscopy; MRI, magnetic resonance imaging; MaSC, mammary stem cells; MSC, mesenchymal stem cell; MPM, multiphoton microscopy; NIR, near infrared; NP, nanoparticle; PA, photoacoustic; PACT, photoacoustic computed tomography; PAM, photoacoustic microscopy; PSC, pluripotent stem cell; PET, positron emission tomography; QD, quantum dot; Rluc, Renilla luciferase; iRFP, bacteria phytochrome photoreceptor iRFP713; RG, reporter gene; SEAP, secreted alkaline phosphatase; SERS, surface-enhanced Raman scattering; siGNR, single gold nanorod; SPECT, single-photon emission computer tomography; SPIO, superparamagnetic iron oxide; SWNT, single-walled nanotube; TSTA, two-step transcriptional activation; TF, transcription factor; U/S, ultrasound; VEGR, vascular endothelial growth factor receptor to detect or diagnose a disease, or to monitor therapy. Molecular imaging has tended to receive more attention in the area of cancer imaging, but how molecular imaging can advance regenerative medicine still needs elucidation. Here, we will review the current state of regenerative medicine and offer new insights into applications of molecular imaging to regenerative medicine. The recurring theme of this review is that merging these regenerative medicine approaches in conjunction with molecular imaging can advance cell therapy in preclinical small animal models, large animal models, and in patients. Furthermore, based on the review these fields, we suggest strategies that will lead to the next generation of regenerative medicine.

| SUMMARY OF KEY CONCEPTS IN REGENERATIVE MEDICINE
Advances in surgery, 1 like skin grafting, 2 vascular anastomosis, 3 and organ transplantation 4 in part, motivated engineers in the development of artificial organs. 5 Further advances led to bioartificial organs, tissue engineering and biomaterials, 6 pluripotent stem cell (PSC) biology, 7,8 and the first cell therapy using bone marrow. 9 These various schools of thought share a common goal of treating the patient under conditions of tissue loss or tissue/organ failure. While there has been a focus on various types of impactful therapies, there has been less focus on advancing regenerative medicine through molecular imaging. In the following sections, we define various aspects of regenerative medicine, as they pertain to molecular imaging.

| Tissue engineering
Tissue engineering arose in the 1980s as an approach to generate human tissue equivalents for clinical tissue replacement. This creative field encompasses a wide array of approaches and methods involving cell biology, extracellular matrix, and biomimetic material scaffolds.
Tissue engineers focused on the transplantation of both cells and scaffolds to reverse tissue/organ failure. In certain cases, the isolation and function of cells were prioritized, 10 while in other cases, materials design was the major factor that impacted cell and tissue function. 11 These scaffold-based approaches involve generating tissue scaffolds using synthetic polymers of various configurations and naturally occurring or engineered biopolymers, 12 and most recently decellularized scaffolds, 13 all of which encompass tissue engineering approaches that address tissue loss. As tissues in the body can be broken down into connective tissue, muscle tissue, epithelial tissue, and neural tissue, tissue engineering products can be grouped in this way.
Along these lines, tissue engineering strategies have been established for: (a) connective tissues, 14 including cartilage and bone, 15 tendons, 16 and vasculature 17,18 ; (b) muscle [19][20][21] ; (c) epithelial (internal) organs, including the liver, 22,23 pancreas, 24 bladder, 25 lung, 26 and kidney 27 ; and (d) neural tissue. 28,29 Upon transplantation of an engineered tissue construct, many critical aspects affect its short-term and long-term fate. Vascularization, transport of nutrients and oxygen to the tissue of interest, maintenance of tissue architecture and function, restoration of normal organ function, and integration of the tissue into the whole body are all critical aspects. Conventional imaging can be used to monitor tissue anatomy (i.e., CT for bone regeneration, or MRI for soft tissue regeneration), and functional imaging (i.e., blood flow via MRI or ultrasound [Doppler]). However, another whole dimension of molecular information may be potentially ascertained by applying strategies in molecular imaging to tissue engineering, which could greatly affect outcomes in patients with tissue engineered constructs. These strategies will be further described in section of this review.

| Adult (and cancer) stem cells and regenerative biology
In the last 40 years, tremendous efforts in multiple areas of stem cell research have cemented their role in regenerative biology and medicine and helped fortify efforts to translate these findings towards human health. Through techniques developed to isolate adult stem cells (ASC) and assay their capacity for growth and differentiation in vitro and in vivo, scientists established many fundamental aspects of regenerative biology. Here, we will consider key aspects relevant for application of molecular imaging to ASC and regenerative biology.
ASC are rare (<1%), small, quiescent cells with a high nucleus to cytoplasm ratio. They are central to the tissue generation process by undergoing asymmetric cell divisions into multipotent progenitor cells, which then differentiate into multiple mature cell types. ASC can accomplish this because only ASC, but not their immediate multipotent progenitors, have the capacity to self-renew. Self-renewal is a specialized type of cell division that is biologically distinguishable from pure cell division. For example, if ASC divide, they can undergo symmetric self-renewal divisions into two new ASC, or undergo asymmetric cell divisions into a stem cell and a progenitor cell.
The immediate descendants of ASC are the multipotent progenitor cells, which proliferate and differentiate along different lineages, contributing to tissue homeostasis. These differentiated cells have a limited life span, whereas the ASC, because of their self-renewal property, have a continuous, unlimited, ability to regenerate themselves. In this way, ASC can both replace themselves and replenish downstream tissues. The corollary of this is that adult, parenchymal tissues are hierarchical with respect to cell type and cell state, and it has been shown that supporting cells can form tissue hierarchies as well. 30 Taken together, real tissues, as opposed to traditional tissue engineered tissues, are hierarchical and can be visualized as a triangle with horizontal layers (Figure 1). Within this triangle, one or more progenitor cells lie beneath the ASC, and these progenitors, with the appropriate spatiotemporal cues, can proliferate and differentiate into more mature cells. These mature, parenchymal, or functional cells make up the majority of tissue within the organ, and are at the base of the triangle. Two examples include hematopoietic stem cells (HSC), 31 which give rise to lymphoid versus myeloid lineages in the blood forming system, and mammary stem cells (MaSC), which select between myoepithelial versus luminal lineages in the mammary gland. 32,33 The activity of these stem cells depends on local or systemic factors, as well as the intrinsic ones, and ultimately the rate of tissue turnover. For example, the intestinal epithelium is renewed at a rate of 3-5 days, while the blood forming cells are renewed at a rate of 25-50 weeks. 34 ASC accomplish these divisions not only because of specialized molecular machinery but also because of specialized external microenvironments, termed niches, which support function of ASC ( Figure 2). Fundamentally, niches must protect ASC from loss, because if all ASC were lost, then the tissue and organism would not survive. Not surprisingly, niches are complex multidimensional environments that change in space and time, are located throughout an adult tissue where "ASC" are present, possess unique anatomical and functional dimensions, and have been reviewed in detail. 35 They were first experimentally identified in fruit fly (Drosophilla melanogaster) in solid tissue within the developing ovary models and have been studied in the hematopoietic and the hematopoietic and skin models and other model systems. 36 In the fruit fly ovary, this stem cell niche is maintained, in part, by intercellular interactions between germ stem cells and the somatic cap cells, 37 and signaling factors like decapentaplegic (Dpp), which are bone morphogenetic 2/4 protein (BMP 2/4) analogues. These niches are on the order of 5 μm × 0.5 μm × 2 μm and tend to contain niche cells with specialized functions, including unique expression of cell surface receptors, soluble extracellular matrix for supporting stem cells and the microenvironment for maintaining the state of ASC ( Figure 2). These highly specialized stem cell niches serve as a controlled microenvironment that, when altered due to physiological and pathological stress, control how ASC respond.
A major question is, how are functions of ASC evaluated? The development of functional assays has been critical for identifying unique markers for stem/progenitor populations. These assays have provided a framework for purifying stem cells and for understanding quantitative differences in the cell's in vivo differentiation properties. 38 To define the true properties of ASC, like self-renewal, stem cell hierarchy, and stem cell niches, in vivo assays are critical. Typically, this involves isolation of ASC from mouse or human donor tissue, clearing of endogenous tissue in host organ, which contain ASC within their niches, and orthotopic transplantation of donor ASC in the host organ. 38 The donor ASC self-renew, differentiate, and their functions can be assessed, typically by removing the host tissue and analyzing for tissue growth, differentiation, and self-renewal. A key assay for assessing stem cell fates is lineage tracing, 39 which is a method for understanding the descendants of an originating cell that is marked.
Using vital dyes, radioactive labels, genetically encoded reporters, or conditional Cre-Lox technology, stem cell scientists can track the fates of cells, including the number of cell divisions (low vs. high), their location, and the relative time in which they arose.
Differentiation of ASC occurs either spontaneously or due to a change in microenvironment. Stem cells leave their niche and differentiate into one or more progenitors, which then are committed to multiple lineages. From a molecular point of view, differentiation essentially suggests a change in cell state. The cell state is defined by cell-specific transcription factors (TF), which activate cell-specific genes at the DNA level, which result in the production of cell-specific proteins, which in turn confer cell identity and function. 40,41 The TF that control cell states are often controlled by developmental enhancers which contain binding sites for TFs from earlier states. Furthermore, expression of these TFs and access to these developmental enhancers are controlled epigenetically. 42 The presence of quiescent, primed, and active promoter/enhancers thus classifies each gene themselves in multiple states, with progenitor cells having primed states and more mature cells having active or open enhancers at differentiation genes. To establish a particular state, developmental or lineage-specific TF also often will have to repress opposing states (endothelial vs. cardiac differentiation of a cardiovascular progenitor cell), or previously committed states (activation of liver specific TF repressing previous states in endoderm progenitor cells). The former would guarantee that cells form the correct fate and repress alternate fates, whereas the latter enables cells to move "forward" with differentiation without dedifferentiating in the backwards direction. Ultimately, genes associated with mature differentiation are diverse, ranging from cell surface markers or receptors, secreted proteins, or cell-specific enzymes.
Scientists who support the cancer stem cells (CSC) hypothesis also believe that CSC (or tumorigenic stem cells), similar to ASC, sit at the top of a hierarchy, in which the cells within the hierarchy represent the tumor. 43 Therefore, cancer can be viewed through the eyes of regenerative biology. CSC theory proposes that the tumor hierarchy is a "caricature" of normal cellular hierarchy, with a CSC at the top of the hierarchy. Experimental evidence suggests that CSC are a rare population of cells within a tumor, which can undergo self-renewal and can give rise to the entire tumor, including recreation of the parent tumor's histology. This is consistent with the appearance of the differentiated state of tumors in biopsies. CSC theory explains tumor heterogeneity, or the fact that tumors are believed to be clonal, even though the tumor cells themselves are heterogeneous and nonidentical. CSC share the property of self-renewal with ASC. 44 CSC theory predicts that only a small number of cells of the tumor can in fact give rise to the tumor, whereas the remaining cells are more differentiated and are destined to die and proliferate less. CSC may arise from normal stem cells that have oncogenic mutations, as they are much longer-lived than their differentiated progeny. 45 CSC also may arise from tissue progenitors that have gained oncogenic mutations which enable the ability to self-renew, which can lead to tumor formation. FIGURE 1 Tissue hierarchy. To maintain tissue indefinitely, ASC undergo asymmetric cell divisions, in which they reform themselves (self-renewal) and differentiate to give rise to multipotent, and/or committed progenitor cells. These progenitors give rise to mature, differentiated cells, which sit at the bottom of the hierarchy, and provide the bulk of the tissue and organ functions, but have a limited lifespan. When tissues are injured, ASC can increase their activity, to stimulate tissue replacement, sometimes by dividing symmetrically to create more ASC that can replenish tissue

| The advent of and applications of PSCs
Although ASC represent evolution's approach for growing and maintaining the tissues in the body, PSC, including both ESC and induced iPSC, together represent a biotechnology with numerous health applications. The value of self-renewing PSC is that in theory, an infinite number of therapeutic cells can be generated from a single clone, and that these cells are personalized, that is, generated from an individual person and genome.
Through disparate studies of the regeneration of organisms like planarians, human tumors like teratomas and germ cell tumors, and experimental studies of the zygote (fertilized egg), scientists theorized and developed the concept of pluripotency. Supporting this notion, they found that portions (inner cell mass) of the developing zygote, or the pre-implantation blastocyst, can be cultivated to form a cell that meets the stringent criteria of pluripotency. 7,8 These ESC self-renew, differentiate in vitro, and could be introduced into the embryo to give rise to chimeric mice in which components of all three lineages are donor derived. Techniques have been developed such that the donor cells could be genetically modified using transgene or knock-in technologies. 46 Thus, the recipient mice can have donor cells, which are genetically modified for a particular disease phenotype, which can be passed through the germline to create new transgenic mice. When transplanted subcutaneously in immunodeficient mice, these cells give rise to teratomas, which are tumors derived from all three germ layers.
Techniques to grow the mouse-derived ESC (mESC) in vitro enabled understanding of self-renewal, differentiation toward germ layers, followed by specification and maturation using lineage-specific protocols based upon mouse development. [47][48][49] Scientists found that developmental gene networks function in mESC similarly to how they function in lower organisms. 40 Studies of the maturation of mESC and development of reversal of disease in mouse models of organ failure commenced next. [50][51][52] The development of human ESC was a huge step forward and used similar techniques and relied on similar differentiation approaches. Furthermore, the development and commercialization of cultivation techniques for hESC 53,54 and the differentiation and transplantation of hESC have helped to push ESC biology forward and demonstrated their potential in academic labs for both potential therapeutic applications 55 and clinical studies in patients. 56,57 Despite the promise of hESC, the ethical issues of handling discarded human fetuses generated tension in the field. In this environment, the advent of genetic technology to reprogram any mouse or human adult, somatic, cell into a human-iPSC was a huge discovery and a great boom to the field. 58  The adult stem cell niche. (a) Anatomy of the niche. The ASC niche is a dynamic, in vivo, microscopic microenvironment associated with a perivascular location, and/or have supporting cells which contribute cell-cell, cell-extracellular matrix (ECM), or cell-soluble factors to the niche to maintain ASC in a quiescent state. They are believed to be on the order of 1-2 μm in size within the area of the ASC. Changes in state present extrinsic changes which can give the ASC information needed regarding whether to divide and participate in homeostasis versus tissue repair. (b) Components of the niche. The niche includes: 1) soluble biochemical cues and ECM components 2) poorly understood physical signals, including possible roles for elasticity, stiffness, shear forces, and aspects of ECM such as topography 3) metabolic aspects such as oxygen, glucose, and calcium personalized approach enables rapid generation of personalized cell lines, particularly from patients with genetic diseases, or as donor cells that can be differentiated for transplantation. Importantly, iPSC bypass the ethical issues related to ESCs and ideally, the immune barrier of transplanting allogeneic cells for therapy. For example, a recent promising study demonstrated the use of iPSC for retinal transplantation and resulted in the first transplanted iPSC-derived cells in patients. 56 Despite this landmark, the results were halted due to mutations and heterogeneity in clones that occur in vitro during the reprogramming process. Challenges to PSC implementation include the genetic fitness of the cells (lack of mutations or chromosomal aberrations), the prevention of immune rejection, the avoidance of cancer, measuring functional maturity of cells, scale up, 61 and understanding in vivo cell fate. iPSCs have many other applications for in vitro disease modeling and drug development not mentioned here.

| Cell-based therapies are more complex than other established therapies
The clinical application of cell therapy, often using stem cell-derived products, has reached center stage. In the last 15 years, both ASC and PSCderived cells have proceeded through preclinical models, have paved the way for commercialization, and have motivated numerous clinical trials. 57 As patients continue to die waiting for a donor organ on organ transplantation lists, stem cell and tissue engineering-based approaches offer hope.

Examples of cell therapies include chimeric antigen receptor (CAR-T) cells
for cell-based cancer immunotherapy, 62 cardiac cell therapy, 63 islet cell therapy for type II diabetes mellitus, 61 retinal progenitor cell therapy for macular degeneration, 56 hepatocyte cell therapy, 64 cell therapy for the nervous system, 65 and mesenchymal stem cell (MSC) therapy for a wide range of diseases. 66 Consistent with this, forecasts suggest that the number of studies using adult and PSC will continue to grow because of the increasing need for treatments for chronic diseases.
To imagine the molecular imaging of cell therapies, it is important to distinguish cell therapies from traditional therapies like medicine and surgery in many nonobvious ways, (Table 1, Figure 3). For example, surgical therapy, for localized congenital or acquired disease, can be monitored visually within the operating room and has predictable complications such as bleeding, infection, and pain. Knowledge of these side effects is based upon knowledge of the coagulation system, immune system, and nervous system, respectively. The same can be said of many medical devices associated with surgical problems that are used in therapy (i.e., intra-aortic balloon pump). Pharmaceuticals or biopharmaceuticals have many aspects that are predictable. A smallmolecule drug, or even a therapeutic monoclonal antibody, having gone through $1 billion drug development process, has a known molecular target, a highly specific receptor or enzymatic target within the cell, with predetermined therapeutic doses and side effects. The absorption, distribution, metabolism, and excretion (ADME) of the drug occur by known, somewhat predictable pathways. Similarly, the routes of administration are well known (i.e., intravenous, oral) and result in predictable changes in ADME. In that sense, surgery, medical devices, and pharmaceuticals/biopharmaceuticals are quite predictable.
Cell and tissue-based therapy contrasts starkly with both traditional surgical (or medical device based) and pharmaceutical-based therapies (Table 1, Figure 3  objective is to image the biology of the process, rather than the anatomy (location of an organ) or function (blood flow). The advantage of imaging is that each living subject, in a preclinical (i.e., mouse) or clinical study, can serve as its own control. High sensitivity together with methods to generate and quantitate an imaging signal are needed in this field, which can allow detection of molecular events that occur either with a low mass of substance of interest, or at low concentrations. FIGURE 3 Hurdles for engineering cell and tissue therapies. Cell therapies, often using ASC, progenitor cells, or human pluripotent stem cellderived products, will face numerous hurdles compared to therapies like pharmaceuticals, biopharmaceuticals (monoclonal antibodies), medical devices, or even surgical treatment. Targets of cell therapy: most therapies have a highly specific target. However, cell therapies are complex, and their targets are numerous. They include increasing tissue mass, differentiation (with all its complex stages), morphogenesis, stimulating endogenous tissue repair and regeneration, establishing tissue micro-and macro-architecture, and reducing inflammation. Therapeutic cell fate: the cells being transplanted may have many fates which complicate cell therapy, because their fate may vary between patients. For example, proliferation, differentiation, migration, tissue engraftment and integration, apoptosis/necrosis, and senescence/aging. Therapeutic cell type: many cell types are available for therapeutic cell replacement, and it is unclear which therapeutic cells are the best for a particular case. Cardiac cell therapy is one example in which this is the case. For example, possible cell types include adult stem cell, tissue specific-progenitor cell, committed precursor cell, mature parenchymal cell, supporting cells, bone marrow-derived mononuclear cells, and mesenchymal stem cells. Host factors: Unlike other types of therapies, host factors play a major role in dictating the fate of cell therapy. However, many of these effects are unknown. Potential factors include extent of disease, age, genomic and epigenetic (intrinsic) factors, pre-existing local or systemic conditions, vascularity, immune status, metabolic status, and local biomechanical factors. Currently, these can only can be determined empirically, and many animal models do not take these factors into account in preclinical models A simple example is illustrative of molecular imaging. If a vascular endothelial growth factor receptor (VEGFR) inhibitor is being administered for cancer, then the therapist would need to assess the extent of the target VEGFR expression within the cancer prior to therapy.
Thus, a molecular image would first be generated prior to therapy. It would guide the decision to administer the molecular therapeutic, which in this case is the VEGFR inhibitor. Post-therapy, it would be important to assess whether the VEGFR inhibitor was targeted, which presumably resulted in loss of tumor and loss of target. Therefore, another molecular image would be generated, which could be used to assess the previous therapy, including factors such as dose and efficacy. Not only molecular imaging is tied to therapy and drug development, but it is also connected with early disease detection. It is believed that molecular changes must occur prior to anatomical changes and thus molecular imaging and thus molecular imaging can be used for early detection of disease.
Keeping this overview in mind, below we will identify potential targets of molecular imaging in regenerative medicine. In the subsequent section, we will separately review the state of cell imaging as it pertains to regenerative medicine.

| Molecular imaging targets in tissue engineering
The molecular targets for imaging in tissue engineering are dependent upon which cell type is being tissue engineered (i.e. epithelial, neural, Conventional anatomical imaging and functional imaging can be used with molecular imaging in unique ways, for example, imaging blood vessels, flow, and angiogenesis receptors.

| Molecular imaging targets in adult (cancer) stem cells and regenerative biology
The emerging details in regenerative biology, mentioned earlier, suggest that numerous molecular imaging targets exist for tissues in vivo, and that noninvasive molecular imaging and analysis of these targets could deepen our knowledge, but also hasten new diagnostics and therapeutics development. Molecular imaging the biology of ASC, such as cell receptors or self-renewal pathways, is particularly challenging, because of the scarcity (< 1%) and size of these cells. However, obtaining in vivo information about the biological aspects of multipotent progenitors would also be valuable. Multiplex (more than 1) imaging of key molecular targets of the cellular hierarchy present within tissues could provide valuable noninvasive assessment of tissues, which is currently not possible. This may include imaging of molecular interactions between key cell populations within tissues, and how they change with time and space. Molecular imaging of the stem cell niche has not been established, and not only the location of the niche but also the molecular composition of the niche, could be valuable for understanding tissue states. The simple integration of noninvasive molecular imaging with ASC assays, which are typically endpoint, can provide new information to stem cell scientists, and we have previously pursued this approach. 38 Noninvasive molecular imaging of ASC differentiation, one of the main purposes of ASC transplantation, would be valuable for understanding in vivo cell fate during various types of tissue insults. However, as ASC typically differentiate into more than cell type, targeting differentiation would involve imaging multiple cell types. This may involve with cell differentiation-specific transcription factors or lineage specific proteins, or other potential molecular targets. In vivo ASC differentiation is accompanied by morphogenesis, interactions with the tissue environment, and tissue remodeling, all of which represent molecular targets.
Because of isolation and characterization of CSC, scientists interested in regenerative medicine and molecular imaging can apply similar principles to imaging CSC, including the CSC niche, differentiation, morphogenesis, and tissue remodeling. In summary, studying ASC biology in the living subject using molecular imaging has enormous potential.

| Molecular imaging targets in cell therapy
Molecular imaging of cell therapies begins with cell imaging, which will be discussed further below. Furthermore, many of the applications for ASC and PSC also hold for cell therapies in general. Future directions for these cell therapies are discussed in the "Next generation regenerative medicine" section.     83 This information is summarized in Figure 4 and Table 2.  98 These NIR dye-containing NP have a typical core-shell structure in which the core is an NIR organic dye and the shell is a polymer or inorganic matrix-based particle. 99  in emphysema and liver failure models. 103 Consistent with this, using QD800, 1 × 10 5 cells, but not 1 × 10 4 cells, could be detected in subcutaneous transplantation experiments with ESC. 104 This information is summarized in Figure 4 and Table 2.

| Raman cell imaging with NP
The intersection of Raman imaging with regenerative medicine has recently been reviewed in detail. 105 112 This information is summarized in Figure 4 and   120 This information is summarized in Figure 4 and Table 2.

| U/S cell imaging with NP
U/S imaging is a real-time anatomical imaging tool which is easy to use, safe, and has a high temporal and spatial resolution. B-mode U/S uses differences in backscattered waves, due to the impedance of tissues, to generate an anatomical image. Here, ultrasonic (mechanical) waves are transduced across the tissue, a backscattered wave is generated and recorded, and an image is generated. Cells cannot be seen using conventional U/S, and contrast is needed. Microbubbles, a gasfilled bubble with about 5 μm lipid containing shell, are a clinically approved contrast agent, but cannot be used to image cells, as they remain extracellular and in the vasculature. 121 Scientists first generated U/S contrast by synthesizing silica 122 and mesoporous NP, 123 both of which eventually ended up being used for cell labeling and MSC imaging in the context of cell delivery to the heart. 124,125 These studies report a detection level between 7 × 10 4 and 5 × 10 5 cells.
A recent paper by Chen et al. report using an "exosome-like" NP for detection limits of 2 × 10 5 cells experimentally, but report that, in  theory, potentially higher levels of sensitivity were possible, with a theoretical limit of 5 × 10 2 cells. 126 This information is summarized in Figure 4 and Table 2.

| CT cell imaging with NP
CT is a technique that helps to visualize differences in tissue attenuation of x-rays, and it is advantageous because of cost effectiveness, higher spatial resolution, short scan time, and ease of imaging. CT is not considered to have a high sensitivity, and thus has a limited ability to image cells. NP-based cell imaging with CT is based on the principle that the higher atomic number within its solid-state structure. 127 129 In both cases described here, synchrotron radiation in a focused beam format is used, which greatly improves spatial resolution. However, it is unclear what the sensitivity is when using small animal imaging instrumentation, such as the microCT, for whole-body imaging. This topic was recently reviewed in great detail. 130 This information is summarized in Figure 4 and  134 This information is summarized in Figure 4 and Table 2.

| Positron emission tomography cell imaging with NP
In PET, the positron emitting radionuclide annihilates a nearby electron (100 μm) and emits two, anti-parallel, high energy, 511-keV gamma photons, which are detected in a coincident fashion. The drawbacks of PET are that the spatial resolution is low (mm), it is highly specialized and can be costly. A wide range of NP have been labeled with positron emitting radionuclides such as 18-F and 64-Cu, and these have recently been reviewed, 135 but it is unclear if these PET-labeled NP have been used for stem cell imaging. This information is summarized in Figure 4 and Table 2.

| Summary
A summary of all the imaging modalities and NP used is in Figure 4. In However, we did not see limit of detection studies in most cases, which suggests there is opportunity to do that. Many variables affect prelabeling, including concentration of NP per cell, duration of labeling, transfection reagent used, and cell type. Imaging parameters and host parameters may also affect imaging signal. Although these approaches have not necessarily contributed to our knowledge of stem cell biology, one can argue that cell prelabeling for imaging a population of cells after initial injection can be used to optimize and localize the initial aspects of cell therapy. Using a highly sensitive modality for NP labeling combined with an imaging modality with a high spatial resolution may provide further insight into the exact location of the injection, which can be used to further optimize cell delivery strategies.

| Fluorescent proteins
The discovery, 136 cloning, 137  The advantage here is that ASC, which are difficult and at times impossible to culture, can be studied in their intact environment. Furthermore, the response during regeneration, after injury, to disease, and so forth can be studied noninvasively using these approaches. It is important to note that, in some studies using IVM/MPM, the RG was expressed in the stem cell itself, while in other cases, a supporting cell or tissue structure expressed the RG. A dizzying array of biological fates or mechanisms have been explored using these techniques, including homing, trafficking, interstitial transport, differentiation, migration, stem cell-niche interactions, asymmetric versus symmetric cell divisions, stem cell heterogeneity, tissue homeostasis, spatial organization of the niche, differential growth, collective cell movements, and so forth. Importantly, not only can IVM/MPM enable single cell level imaging but also enable whole tissue/organ imaging. This versatile imaging tool thus facilitates concepts of tissue mapping, imaging differential tissue growth, imaging tissue regeneration, and imaging organ/tissue development. 38,159 Not only have ASC and the ASC niche been imaged using these approaches, but also CSC-mediated processes, like tissue remodeling and migration have been studied using IVM and MPM. 38,150,160 This has proved valuable, as the CSC tinue to provide some clues. This information is summarized in Figure 5 and Table 2.

| Bioluminescent RG-based cell imaging
Bioluminescence imaging (BLI) with RG has been a tremendous development, which has greatly impacted molecular imaging and in vivo and has more rapid signal kinetics. Rluc has been engineered extensively, including for enhanced stability in serum, 163 180 This information is summarized in Figure 5 and Table 2. 6.6 | Transgenic mice and promoter engineering for cell imaging Differentiation, on the other hand, requires strategies to engineer promoters and/or RG. Most differentiation promoters are too weak to see a change in signal over time. 183 We recently engineered a two reporter approach in which constitutive activity was measured with ubiquitin C and Fluc, while differentiation was measured with a cloned promoter and Rluc. 178 In this case, the differentiation promoter was Oct4 and was strong initially, but then was shut down during differentiation, although we observed complex kinetics that had previously not been appreciated. BRET involves the nonradiative transfer of energy between the donor and acceptor molecules by the FÖRSTER mechanism, in which energy from a donor chromophore is transferred to an acceptor chromophore through nonradiative dipole-dipole coupling and has a radiusdependence of 1/r 6 . BRET technology uses a fluorescence and bioluminescence protein pair. Here, a bioluminescence substrate is added to the living subject, exciting the bioluminescent protein to luminesce.

| Improving deep tissue cell imaging
This transfers the energy, nonradiatively, to the fluorescence protein.
Next, the fluorescent protein emits energy that reflects an interaction between the pair of proteins, often due to protein-protein interaction.
Manipulation of the donor protein, together with red-shifted acceptor fluorescent protein, and a new substrate that produced red shifting 188 have led to improvements in the ability to image 3 × 10 6 tumor cells entrapped and spread throughout the lungs (deep tissue) after tailvein injection of cells. However, thus far, these BRET systems have not been used to image stem cells in deep tissues, but this would be FIGURE 5 Reporter gene-based imaging strategies. While NP-based strategies are primarily focused on cell labeling, many reporter gene strategies have been used to understand underlying biology in addition to imaging cells in vivo. RG strategies for in vivo imaging include near infrared (NIR) and fluorescent RG, MRI RG (ferritin, transferrin, CEST, Tyrosinase, gas vesicles, aquaporin), photoacoustic (Lac Z, Tyrosinase, NIR proteins), PET/SPECT (hsv1tk, dopamine receptor, and sodium iodide transporter), and bioluminescence (Firefly luciferase (Fluc), Renilla luciferase (Rluc), and Gaussia luciferase (Gluc). In each modality, we demonstrate how the RG is expressed and how an imaging signal is generated valuable for imaging stem cells within internal organs like the intestines, liver, pancreas, and lungs. This information is summarized in Figure 5 and Table 2 and PET imaging signal (% injected dose per gram) demonstrated that this PET RG system was quantitative. A second PET RG system involves radiolabeled PET dopamine-based ligands that were developed for the dopamine receptor (D2R), which normally is on brain striatum and pituitary glands. 191,192 Regarding the hsv1-tk system, scientists improved expression levels by using mutant enzymes with improved reporter probe uptake and imaging signal, 193 identifying an improved reporter probe 9-(4-18F-fluoro-3-[hydroxymethyl] butyl) guanine (18F-FHBG), and testing 18F-FHBG pharmacokinetics and safety profile in patients. 194 This system was shown valuable for cell imaging in the setting of cardiac cell transplantation. 165,195 Importantly, this sr39tk-FHBG combination was tested in patients in a trial of gene therapy directed toward liver cancer, 196 and in the setting of T cell therapies expressing HSV1-tk in patients. 70,197 To determine the sensitivity of cell imaging in patients, we recently performed a study of MSC injection in swine hearts, followed by MRI and PET. 82 We found the detection levels in the heart, with the HSV1-tk sr39tk mutant, and 18F-FHBG, to be 2.5 × 10 8 cells. Although other studies did not perform a full, quantitative limit of detection study, reports have ranged between 50-fold and 2-fold less than what we reported. 198 The sodium iodide symporter RG is a third reporter system, which uses a cell membrane transporter as the RG and radiolabeled iodide as the probe of interest. 199 In this RG system, radioactive iodine is injected and selectively accumulates in cells that express the iodine symporter. The iodine symporter is normally present in thyrocytes in the thyroid gland for selective iodine transport in the process of thyroid hormone synthesis within the thyroid follicle. About 7.5 × 10 6 cells have been grown for 4 weeks and imaged using planar scintigraphy for 123 -I. 199 The advantage of this system is the complex aspects of PET radionuclide synthesis are not required, and a second advantage is that radiotherapy can be used to destroy labeled cells in the setting of cancer therapy or suicide therapy. Although studies demonstrate comparable uptake to the HSV1-tk system, the limit of cell detection is unclear. This information is summarized in Figure 5 and  207 This information is summarized in Figure 5 and Table 2. 6.10 | Photoacoustic reporter genes for cell imaging Photoacoustic (PA) RG are an active field of research leading to an explosion of research in the area of photoacoustic imaging. The first report of cell imaging of RG using PA was when 5 × 10 6 lac Zexpressing tumors were imaged beneath the scalp of rats after injection of X-gal, a substrate for Beta galactosidase which is expressed by the lac Z gene. The imaging time, however, was 25 min. Here, X-gal is cleaved to galactose and 5-bromo-4-chloro-3-hydroxyindole, which then dimerizes and is oxidized to 5,5 0 -dibromo-4,4 0 -dichloro-indigo. 208 Tyrosinase RG were evaluated and demonstrated a PA signal when expressed in nonmelanin containing cells. Tyrosinase (Tyr) is a rate limiting step in melanin production, and melanin is a pigment in the skin, hair, and eye. Tyr-expressing cells were imaged by PA with an estimated 45 cells at 3 mm in depth, 209 while another study 210 reported an in vitro imaging sensitivity of 2.5 × 10 3 cells. In this latter study, at least 1 × 10 7 Tyr-expressing tumor cells were imaged using PA, but an in vivo limit of detection study was not performed. Other studies also have produced cell imaging of 5 × 10 6 tyrosinase expressing cells in vivo, and 1 × 10 6 cells implanted at 6-8 mm were clearly visualized. 211 214 This information is summarized in Figure 5 and Table 2 and sensitivity of the assay is 50 pg/mL. 216 Mouse SEAP (mSEAP) has been engineered to avoid immune responses in murine models and has been shown to correlate with cell number and tissue growth of transplanted cells. 217 The SEAP assay however takes time and can be limiting in high throughput assays. Gaussia luciferase (Gluc) is a 480 nm emitting luciferase which oxidizes the substrate coelenterazine. Because the protein is small (19.9 kDa) and luminometry is highly sensitive and quantitative, the assay for Gluc is easy, fast, and highly sensitive (1000-fold more than SEAP). It has been widely used as a secreted protein for many applications in therapeutic monitoring. In one study, tumor cells constitutively expressing Gluc were implanted at various cell numbers and were imaged with bioluminescence imaging. Furthermore, samples of blood and urine were collected and an ex vivo assay for Gluc was performed. This study showed that Gluc in blood and urine was linear with cell number and correlated with in vivo imaging of cell number. 218 Despite its wide usage, Gluc has not yet routinely used for in vivo tracking of stem cell fates. This information is summarized in Figure 5 and  Furthermore, several orders of magnitude of improvement sensitivity are necessary for MRI and PET RG will improve the clinical utility of these approaches, such that stem cell therapies are further advanced.
In our final section of this review, we will consider what next generation regenerative medicine can be.

| "NEXT GENERATION" REGENERATIVE MEDICINE
Currently, the goal of cell and tissue-based therapies is to replace damaged or dysfunctional tissues. As many of these approaches have not yet been translated to patients, it would be useful to highlight how advanced molecular imaging strategies and tools may be used to improve regenerative medicine-based approaches or improve the translation of these approaches into patients.
The standard in the field is to use tissue sections to assess regenerative status. However, these are endpoint assays, which require animal sacrifice in preclinical models or biopsy in preclinical models or patients. These biopsies and their analysis are often not quantitative. Furthermore, as the tissue is being processed there can be a loss of information, and differences between individual subjects can be lost. Based on these concepts, below we discuss various ways, other than cell imaging, in which imaging can improve cell and tissue based studies. Figure 2 summarizes the variables when performing cell or tissue therapy which may influence outcome, and these areas represent opportunities for new imaging approaches.

| Integrating imaging with in vivo regenerative medicine assays
A simple area where molecular imaging can improve regenerative medicine potential is an improvement of in vivo ASC assays, reviewed in section 2.2. Stem cell assays involve transplantation of ASC into cleared tissues and are sometimes called a regeneration assay. Assay analysis is based on tissue sections is qualitative, although flow sorting of the tissues is quantitative, and both assays are endpoint assays. In vivo, noninvasive molecular imaging could investigate growth/regeneration at earlier time points in the assay, at lower cell numbers, and more quantitatively. In our studies of IVM, applied to mammary development, mammary stem cell regeneration, and cancer stem cell growth, we discovered several new findings, simply by incorporating in vivo imaging 38 into standard assays. While current assays can analyze bulk populations and standard endpoints, in vivo imaging can help obtain information based on serial imaging, and provide real time (temporal and spatial) information regarding a stem cell assay. The use of imaging also enables each mouse to serve as its own control, and could improve data and statistics. Small sources of anatomical and physiological variation can lead to variability in a cell therapeutic response. Using imaging to quantify this with improved controls may lead to more accurate observations and conclusions, determining which research directions are more promising. In vivo imaging may also help identify rare events which cannot be obtained by traditional assays, and as stem cell self-renewal and asymmetric divisions are rare events, developing ways to observe this in vivo can improve our knowledge of these processes. This imaging data can also be used to build quantitative silico models of tissue growth and predict growth changes due to intrinsic and extrinsic perturbations.

| Interrogating tissues prior to cell transplantation
With the continued development of regenerative medicine, for a particular regenerative injury, there will be several options (i.e., ischemic heart disease, Figure 2). An important question will be, how does one choose the correct regenerative medicine treatment option? Currently, this is determined by trial and error in clinical trials, averaging across large groups of patients. For example, in cardiac therapy, skeletal myoblasts, C kit + cardiac stem cells, Sca1 + progenitor cells, MSCs, and hPSC-derived progenitor cells are all used, including imaging (cardiac echo), function (cardiac catherization), and blood tests (coagulation profile). As cells are being placed within the heart, it would seem that molecular factors of the local heart tissue could dictate the success or failure of the therapy. Thus, one can conceive that molecular imaging tests or molecular diagnostics could be developed to reveal molecular information. Recently, Jokerst et al. 219 identified molecular biomarkers that are associated with patients that respond to cardiac therapy. Perhaps molecular imaging tests of not only the blood, but also the tissue, can also be used to predict cell therapy. These can potentially include molecular probes which identify angiogenesis, metabolism, inflammation, or other molecular aspects of the milieu that might predict successful therapy. Other host factors, including aspects of the local transcriptome and proteome, may heavily influence the outcome of cell therapy. Thus, molecular imaging can potentially be used as a diagnostic prior to cell therapy.

| Translating cell therapy to patients and multimodality imaging
A major problem is translating results from small animals to patients.
For example, there have been many successful cell therapies 220 in small animals, but how can this be translated to patients? Dosing with pharmaceutical can be done based on mass of the patient (or the amount of receptor available), but because molecular targets of cell therapies are complex, it is unclear how to perform exact dosing. In a recent set of publications, we demonstrated experimentally an idea of how molecular imaging in small animals and large animals could be connected, to improve the clinical translation of cell therapies. 82,180 In this case, multimodality RG, bearing eGFP (enhanced GFP), Fluc2, and hsv1 mutant (sr39tk), for fluorescence, bioluminescence, and PET RG imaging, were transduced into MSC. These MSCs were tested in small animal disease models and Fluc signal was associated with cell survival.
An endpoint of 14 days postmyocardial infarction was used, and MSC imaging demonstrated this endpoint was achieved. Because these MSC had reached the appropriate criteria and they also expressed the PET RG (fusion protein) under the same promoter, the same exact MSC cell line was tested with PET imaging in a large animal model. In this manner, one could imagine that a series of therapeutic cell candidates may be tested in a small animal model, and only the therapies that reach the specific molecular imaging endpoints could be tested in the large animal model. Furthermore, signals due to dosing in small animals (BLI) could be compared to signals in large animals (PET RG) to understand dosing-related issues.

| Summary
In this final section, we have summarized potential ways molecular imaging can be used to further inform regenerative medicine, in what we call next generation regenerative medicine. We predict that a wide range of imaging modalities and tools will continue to increase within molecular imaging for applications like cancer. What is needed is to shape molecular imaging for problems in regenerative medicine and to apply these tools to small animal and large animal preclinical models, and eventually, patients.