Nuclear imaging for immune cell tracking in vivo – Comparison of various cell labeling methods and their application

to in vivo cell tracking and designing optimal treatment strategies for cell-based thera- pies. This review paper focuses on nuclear imaging and its role in the evaluation of immune cell-based therapies’ effectiveness - tracking cells and their ability to home to the target tissue. We compare differ- ent types of radiotracers, outline various ways of immune cell labeling, and provide the latest examples for the use of nuclear imaging techniques both in preclinical studies and clinical settings.


Introduction
Adoptive transfer of immune cells for inflammatory and neoplastic diseases is the subject of intense clinical research. So far, methods involving the administration of different types of autologous lymphocytes have shown great potential to treat different diseases. Such a therapeutic approach is especially promising in the treatment of cancer [1,2] as well as other disease conditions, including allergies, infections, and inflammation [3][4][5]. The type of immune cells to be selected for the adoptive therapy depends on the treated pathology. Among diverse immune cells tested so far, T cytotoxic lymphocytes (CTLs), T regulatory lymphocytes (Tregs), natural killer (NK) cells, monocytes, macrophages, eosinophils, neutrophils, and dendritic cells (DCs) all appear to play an emerging role.
Inarguably, the efficacy of cell-based therapies depends on the successful targeting of pathological lesions by the adoptively transferred cells. When considering novel approaches to this kind of treatment, the targeting efficiency of selected cells to the pathological lesion should be evaluated first. In this context, molecular imaging techniques, which enable noninvasive and real-time monitoring of the cell distribution in the tissues of interest, take center stage, as they provide key information on the long-term survival of the adoptively transferred immune cells. Among clinically relevant in vivo imaging modalities, the nuclear imaging techniques, comprising planar scintigraphy, SPECT, and PET, are considered to be of pivotal importance to in vivo cell-tracking as they can afford the design of optimal treatment strategies for cell-based therapies. Tracking of immune cells with nuclear imaging has been used for a long time in routine medical practice for diagnostic purposes in infections and inflammation [5]. Immune cell labeling for cell tracking with nuclear imaging can be accomplished with direct or indirect methods. Direct methods involve the labeling of immune cells with radiotracers in vitro before cell transfer, whereas indirect labeling relies on the introduction of a reporter gene into immune cells to induce an additional function that makes the cells uniquely targetable by a specific radiotracer in vivo after cell transfer. In this Review, we provide the description of various nuclear imaging modalities in the context of live tracking of immune cells, outline immune cell labeling methods, and underline the latest examples for the use of nuclear imaging techniques both in preclinical and clinical research.

Nuclear imaging overview
Nuclear imaging is routinely employed in medical practice to assess disease severity and treatment effect. It involves oral, intravenous, or inhalation administration of small doses of a radiopharmaceutical (radiotracer), which consists of a short-lived gammaemitting radionuclide conjugated to a chemical probe, capable of targeting the desired cellular or molecular process with high affinity and selectivity. The ionizing radiation thus emitted is then measured and imaged using an imaging device. The spatial resolution coupled to the capabilities of the probes to interact with specific cellular targets classify nuclear imaging among the molecular imaging techniques. Radioisotope-based imaging is commonly used for diagnostic purposes in cancer as well as in neurological, cardiovascular, gastrointestinal diseases, and cysts in kidneys [6][7][8][9]. In addition to diagnostic applications, nuclear imaging can also be selected to treat certain cancers such as non-Hodgkin lymphoma and thyroid-related tumors by radio-immunotherapy [10]. Nuclear imaging has become the most attractive and powerful modality for molecular imaging in personalized cancer therapy. Considering the type of image obtained, we can distinguish two groups of nuclear imaging techniques: planar (2D) gamma scintigraphy and spatial (3D) methods, including single-photon emission computed tomography (SPECT) and positron emission tomography (PET).
Nuclear imaging offers a range of important benefits in preclinical and clinical research, which lead to its routine use in many clinical applications. Primarily, nuclear imaging techniques are extremely sensitive to injected tracers compared to magnetic resonance imaging (MRI) and computed tomography (CT)-based techniques maintaining at least a similar level of sensitivity as optical imaging methods. Radiation detectors can detect picomolar concentrations of radiotracers. This sensitivity allows the use of subpharmacological quantities of radiotracers, avoiding the risk of biological side effects. Secondly, nuclear imaging is non-invasive and therefore can be applied in series to identify the longitudinal run of certain processes (e.g., reaction to the implemented therapy). Thirdly, thanks to suitable image analysis techniques, both PET and SPECT allow quantitative data to be obtained, such as the concentration of radiotracers in certain regions or volumes of interest (ROI/VOI). Fourthly, a great number and diversity of radiopharmaceuticals such as metabolites and their analogs, ligands for specific receptors, or antibodies are still being developed to provide an ever more accurate and relevant characterization of biological phenomena in situ [11][12][13]. However, nuclear imaging has its flaws. One of them is a moderate spatial resolution that ranges from~2 mm to~6 mm for PET or more for SPECT and is much worse than that of CT and MRI. Furthermore, nuclear imaging, by definition, is based on radioactive decay, and provides patients with low, but probably not insignificant, doses of radiation [14,15]. Finally, nuclear imaging delivers images with imprecise anatomical information, which may make their interpretation and analysis difficult. However, nowadays this limitation can be overcome by combining it with different imaging techniques, such as CT or MRI, which show anatomical structures in greater detail. Thanks to the widespread use of hybrid imaging methods such as SPECT/CT, PET/CT, and even PET/MRI, which are currently accessible at the clinical and preclinical level, it is feasible to obtain detailed functional and anatomical information during one study.

Nuclear imaging modalities
Planar gamma scintigraphy measures and images the location of the radiotracer in one 2D image. The signals from an array of photomultipliers in a gamma detector are transmitted to the positioncoding matrix and reconstructed on an oscilloscope screen. In such a manner, the distribution and accumulation of the radiolabeled probe in the organism are visualized [16]. For dynamic imaging, multiple ''frames" of images between 1 s (beginning) and 1 min (end) each, are usually obtained for 30-40 min following the radiotracer's injection [17]. Planar gamma scintigraphy, routinely employed in the last decades, is now commonly utilized in medical practice. In specific clinical scenarios, it is used for the diagnosis of patients with suspected infections and inflammatory processes. However, it has now been replaced to a large degree by the tomographic modalities: SPECT and PET.
SPECT is a modification of the classic gamma camera imaging. The SPECT principle is based on the detection of gamma rays with energy within a certain range (specific to each isotope), as opposed to PET, which only detects 511 keV gamma rays. The imaging is carried out by a gamma detector, which obtains a set of twodimensional projection images at different angles. Typically, one, two, or three detectors or heads are slowly rotated around the patient's body, while acquiring the information needed to reconstruct 3D images. Usually, clinical cameras have a resolution between 7 and 15 mm and preclinical devices between 0.8 and 1.5 mm. Usually, the SPECT tool is combined with CT and it applies a specific algorithm to reconstruct the projection data, to generate a spatial image of the administered radioisotope distribution. Pinhole collimators at a distance of several centimeters from the detector are often used for imaging small animals as they enable magnification of the image on the detector achieving very high resolutions, even sub-millimeter: 0.25-0.6 mm [18], at the expense of a reduction in detector sensitivity. An important advantage of SPECT is the possibility that multiple radioisotopes can be imaged at the same time thanks to the ability to detect photons of different energies. Therefore, several different cell populations, each labeled with a different radioisotope, can be tracked during the same imaging procedure [19].
PET imaging records gamma photons generated during the annihilation of positrons (b + ). The sources of positrons are the b +decayed radioisotopes. Emitted positrons, after a free path of not more than a few millimeters, collide with electrons in body tissues, leading to the annihilation of the positron and the electron [20]. The annihilation event produces two photons, each of energy 511 keV, moving in opposite directions from each other (at an angle of 180°). The angle can be slightly different due to the result of kinetic motion (like Fermi motion). This has a significant impact on resolution, especially with large-diameter (distances between opposite detectors) scanners [21]. The photons are detectable by the use of a PET camera, which is composed of an array of crystal sensors usually in a ring configuration surrounding the sample or patient. The coincidence detection mechanism searches for pairs of photons detected within about a 10 ns coincidence window on opposing sections of the detector at different angles around the object, enabling the precise determination of the positron formation point. After reconstruction to localize the points where the annihilation occurred, the 3D visualization of the isotope distributed in the organism is possible. Importantly, PET acquisition time is shorter than in the case of SPECT, due to a combination of the static ring detector sampling all radial lines of response simultaneously, and the fact that sensitivity is increased as collimation is not required. This feature allows reliable dynamic imaging to be performed, which allows pharmacokinetic modeling of the radiotracer to calculate rates of uptake, binding affinities, and receptor densities in tissues. The spatial resolution, which is measured as total width at half maximum, is usually about 2-6 mm, while preclinical animal scanners reach a resolution of 0.3-2 mm [16]. According to the quantitative ability of imaging modalities, PET is considered as the most valuable technique in this regard. A very high level of signal quantification accuracy is required especially in the case of cell-tracking studies, therefore in this context PET seems to be the preferred tool [22]. However, although SPECT scanners usually have lower spatial resolution and sensitivity in comparison to PET and clinical applications of quantitative SPECT imaging are lacking due to the insufficient availability of accurately calibrated SPECT reconstructions, currently produced SPECT instruments often exhibit quantitative ability comparable to PET. Routine corrections for resolution recovery, photon attenuation and scattering, radioactive decay, cross-calibration, and instrumental dead time are applied to enable quantitative SPECT analyses [23,24]. Nevertheless, detailed quantitative studies are nowadays performed mainly by using PET. A schematic presentation of the SPECT and PET scanners and their imaging modalities is shown in Fig. 1. In Table 1, the schematic mechanisms of action and main differences between SPECT and PET imaging techniques are given. For a thorough comparison of SPECT and PET imaging, the interested reader is referred to the additional review articles [11,19,25].
The technique called ''whole-body counter" also deserves mention, although it does not provide an image in the full sense of the word. In this method, high-sensitivity scintillation probes are moved along the patient's body, generating a quantitative profile of radioactivity from certain regions. The counter is isolated from the background radiation, which results in high sensitivity in the detection of photons, making it possible to administer radiotracers at very low activity.

Radioisotopes for nuclear imaging
There are a number of conditions that need to be satisfied for a radioisotope to be suitable for in vivo imaging. The unbound radioisotope and its decay products, which are commonly different elements from the parent isotope, must not exhibit any toxicity at clinical doses. The satisfaction of this criterion is helpfully aided by the extremely high sensitivity of nuclear imaging modalities, with just picomoles of radiotracer required for typical clinical doses. The radioisotope must also be capable of readily forming biologically active compounds or complexes that are soluble and chemically stable in biological conditions. In addition to these standard biochemical requirements for pharmaceutical agents, a radioisotope must also have favorable radioactivity properties. It must be possible to produce and isolate the radioisotope in high radiochemical purity with a high specific activity. The high radiochemical purity is important because any alternative radioisotopes present in the sample will cause spurious signals in the images and interfere with the interpretation of the images. The high specific activity is important because high levels of non-radioactive isotopes of the element can interfere with tracer synthesis and can lead to the inadvertent co-administration of non-radioactive versions of the tracer which are not detected by the scanner but act as 'cold' blocking agents hindering the uptake of the radiotracer. Also, the radioisotope must emit radiation that penetrates enough to leave the body to enable its external detection. These are most commonly single gamma rays in the case of planar gamma scintigraphy and SPECT imaging or the pair of 511-keV annihilation gamma rays emitted during positron (b + ) decay employed in PET imaging. Ideally, nonpenetrating radiation, i.e. particulate radiation, including electrons (apart from positrons relevant for PET) and bparticles, should only be emitted in small amounts by imaging radionuclides. This is because particulate radiation does not leave the body and cannot be measured with external detectors, but will still contribute to the radiation dose experienced by the tissues. Importantly, the imaging radioisotope should have a radioactive half-life comparable with the biological half-life of the probe, allowing for the radiotracer to reach the tissue of interest while still preserving an optimal imaging signal. In this way, patients can be exposed to minimal radiation dose. Table 2 summarizes the relevant physical properties of single-photon and positron-emitting radionuclides commonly used in nuclear imaging.
The ideal radioisotope for planar gamma scintigraphy and SPECT should emit gamma and X rays with 100-200 keV energies at 100% abundance. Additionally, it should emit only marginal amounts of particulate radiation or high energy gamma and X rays with energies above several hundred keV, which are not efficiently collimated and detected. By giving off only a 140-keV gamma-ray and few particulate radiations, technetium-99 m ([ 99m Tc]) is an almost ideal radioisotope for planar gamma scintigraphy and SPECT imaging. The latest devices allow obtaining fullwidth at half-maximum (FWHM) value of less than 0.4 mm for [ 99m Tc] in the center of Field of View (FOV) [18,26]. Unlike [ 99m Tc], iodine-131 ([ 131 I]), for example, emits high-energy (364 keV) gamma rays and high amounts of bparticles, allowing the compound to exhibit both diagnostic and therapeutic properties. The spatial resolution of SPECT images is influenced by many factors including the energy of photons emitted by the radioisotope, the number and shape of pinholes in collimators, and the type of detector. The use of radionuclides with energy values of gamma rays significantly different from technetium clearly affects the sensitivity of the device and parameters of the acquired image [26]. However, modern devices using collimators of greater thickness and varied geometry allow the use of isotopes with the higher energy of emitted gamma radiation, even exceeding the values used in PET technique. Thus, the proper selection of collimators makes it possible to keep the FWHM value at a relatively constant level when using radioisotopes emitting energies in the range 100-500 keV [27][28][29]. The relationship is not obvious when considering the influence of the emitted energy on the labeled cells. In addition to the value of the photon energy, the potentially harmful effect on cells is also influenced by the type of corpuscular radiation associated with the photon emission and the radio-sensitivity of the investigated cells.
The most frequently used SPECT radioisotopes have a half-life between 6 h and 8 days, which allows images to be acquired over a moderately long time scale. Additionally, SPECT radionuclides are more commonly available than PET radionuclides because SPECT radionuclides are usually eluted from a small portable generator. In turn, most PET isotopes can only be produced on a cyclotron, which means that large-scale infrastructure is required to produce them, making them much more expensive. The average cost of SPECT radiopharmaceuticals is about ten times lower than those for PET [30].
Numerous PET radionuclides may be incorporated into the imaging tracers. The ideal radionuclide for PET imaging should emit short-range, low-energy positrons with an abundance of 100%, and lack of gamma rays with energies close to the 511 keV, generated during positron annihilation. Fluorine-18 [ 18 F], for example, is such a PET radionuclide, and thus radiotracers tagged with this radioisotope exhibit the most favorable features for PET imaging, however, with a half-life of 109.8 min is only effective for imaging up to a few hours after administration. Thus, for specific applications that require longer term tracking of administered agents, including some which are routinely clinically applied for SPECT imaging, alternative isotopes with longer halflives are required. When considering the choice of an isotope, it is important to consider the radiation dose to the patient, as longer exposure times to a given amount of radioactivity will lead to a higher overall radiation dose. As a result, the best practice is to select radioisotopes with half-lives that are tailored to the optimum biodistribution time of the proposed radiotracer. Additionally, overall patient radiation dose for tracers based on longer lived isotopes can be reduced by minimizing the total amount of injected activity although this requires compensation through the use of modern measuring devices or extension of scan acquisition time. Also -unlike [ 18 [12,13,31,32]. Most commonly used PET isotopes are produced employing a cyclotron apparatus, and are usually short-lived isotopes in comparison to SPECT isotopes. Isotope lifetime may thus represent a key issue to be addressed as a function of the desired probe effect. Adequate time period must be provided for the radioisotopes in medicine for diagnostic tests in order to lose their radioactivity through natural decay, and the shorter the half-life the sooner this point will be achieved. In addition, it may be preferable to minimize the radiation dose delivered to the patient during the test or it might be desirable to repeating the test at intervals without concern for residual activity in the body from the previous test. On these accounts, a short-lived radioisotope might be preferable to a long-lived one. In contrast, in several specific cases, the short half-life of some of the most commonly exploited PET radioisotopes may hinder long-term monitoring of the radionuclide fate inside the body.

Labeling strategies of immune cells for tracking in vivo with nuclear imaging
Live tracking of immune cells using nuclear imaging involves the detection of radioisotope-labeled cells. There are two major cell labeling techniques for nuclear imaging: i) direct cell labeling with radiotracers or ii) indirect cell labeling involving the expression of an imaging reporter gene in target cells.

Direct cell labeling
Radiotracer-based direct labeling of immune cells is usually applied in the context of nuclear imaging. It is a quite straightforward technique that is performed under in vitro conditions with little influence on cell behavior, including cell viability and migration [8,18,20,24]. There are different types of radiotracers suitable for cell labeling, sharing one basic physical property consisting of a radionuclide that emits externally detectable radiation. The radiotracers, used for direct cell labeling, may be incorporated into the Only 511 keV radiation energy is detectable -only a single radiotracer can be applied Most of the available radiotracers are long-lived ✔ Most of the radiotracers are short-lived, however, several long-lived are available too Requires administration of higher radiotracer's radioactivity Requires administration of lower radiotracer's radioactivity ✔ Relatively cheap ✔ Relatively expensive Generator production of the most commonly used radioisotope ( 99m Tc) make them widely available ✔ Cyclotron production of the most commonly used radioisotopes (including 18 F-FDG) limit the availability Usually lower quantitative ability compared to PET Excellent quantitative ability ✔ 1) binding to a specific receptor 2) undergoing endo/phagocytosis 3) taken up via the pump or ion channel 4) undergoing passive transport across the membrane and trapping in the cytosol 5) being built-in the cell membrane 6) linking to cell membrane proteins Different radiotracers grouped by the mechanism of direct cell labeling are presented in Fig. 2 [37][38][39][40][41]. Subsequently, these complexes dissociate and the released radionuclides become trapped within the cell by binding to intracellular proteins. Direct labeling of immune cells with radiolabels can also be achieved through the uptake of radiolabeled probes via a transporter mechanism, including a transmembrane receptor, a channel, or a pump, located on the membrane of the cell. As a single transporter located on a plasma membrane can transport many molecules of a radiotracer, this method of direct cell labeling may lead to amplification of the signal. A typical example of radiotracer-based direct cell labeling through this mechanism is [ 18 F]FDG, which enters the cell through glucose transporter (GLUT) transmembrane proteins and is retained in cells by hexokinase-mediated phosphorylation, causing entrapment of this tracer in the cytoplasm [42,43]. Finally, direct cell labeling can be achieved through direct binding of the radiotracer to the cell membrane. One example of such radiotracers is [ 18 F]HFB, which binds to cellular membranes via a lipophilic long-chain ester [44]. Another approach of direct cell labeling through the radiotracer anchoring to the cell membrane involves the chemical coupling of [ 89 [46] to cell membrane-bound proteins.
Direct immune cell labeling provides several benefits, which include fairly simple labeling protocols performed without the requirement of genetic cell manipulation and high sensitivity, enabling detection of small numbers of cells due to the minor background signal. Given these properties, direct labeling methods are easier to translate into clinical practice than indirect labeling approaches. Although direct cell labeling is fairly simple, it has some limitations. One disadvantage of this method is that the radiotracer becomes diluted upon each cell division, which leads to decreased concentrations of the probe per individual cell. Additionally, the radiotracer may be transferred asymmetrically to the daughter cells following cell division or may be released from the cells (e.g., [ 111 In]oxine and [ 64 Cu]PTSM), eventually leading to the disappearance of the signal [12,39,47]. As a consequence, directly labeled cells can be tracked in vivo for only up to a few days. Furthermore, direct-labeling does not enable monitoring of cell biological functions, such as proliferation, activation, or viability, and may evoke changes of cellular characteristics after radiotracer uptake. Finally, the adoptively transferred, radiolabeled dead cells may be engulfed by the phagocytic cells leading to the generation of an unspecific signal. To overcome the limitations, reporter-based indirect labeling techniques for live-cell tracking have been developed. These approaches enable the detection of reporter geneexpressing, live immune cells with nuclear imaging techniques following administration of a specific radiotracer [12,13,48].

Tracking of directly labeled immune cells with planar gammascintigraphy and SPECT imaging
The use of immune cells directly tagged with radioisotopes for live tracking of immune cells with planar gamma scintigraphy and SPECT imaging is currently an accepted procedure, and many labeling techniques have been developed for this purpose [49]. Although live tracking of directly radiolabeled white blood cells with planar gamma scintigraphy is routinely used in clinical settings [6], recent progress has focused on the more technologically advanced SPECT imaging in the majority of preclinical and clinical immune cell tracking studies.
Immune cell labeling with direct methods for planar gamma scintigraphy and SPECT imaging relies on radiotracers that are mainly chemical complexes of two radionuclides, namely [   it has a high affinity for water making its complexes susceptible to hydrolysis. As a result, high coordinate polyamino acids such as EDTA, DTPA, DOTA, TACN are most frequently used as In(+III) chelators for radiochemical applications to maximize complex stability. In(+III) complexes can form with coordination numbers varying from 4 to 8 and can exhibit a wide variety of coordination complexes and geometries as detailed in these excellent reviews [50,51]. However, for cell labeling tight irreversible binding of the radiometal by the chelating ligands is not always required. One established labeling strategy is to employ metastable lipophilic complexes, which diffuse across the cell membrane, carrying the radiometal with them, and then dissociate inside the cells causing the radiometal to become trapped intracellularly [37,52,53] (Fig. 2). With this approach, highly stable complexes are not desired as there would be no trapping mechanism and the activity would leak out of the cells after labeling.
As described above, [ 111 In] complexes have been at the forefront of the development of cell labeling for in vivo imaging using this strategy. The most prominent example is [ 111 In]oxine [54], which in 1985 was approved by the FDA for leukocyte scintigraphy. The oxine ligand (8-hydroxyquinoline) is lipophilic but retains enough water solubility to react with metal ions in an aqueous solution. It reacts by losing a proton to become an anionic N, O-chelate enabling the formation of neutral lipophilic complexes, which can be readily purified by biphasic extraction into organic solvents followed by solvent evaporation. In the case of In(+III) a hexadentate tris(oxinato)indium(+III) complex is formed with a pseudo octahedral geometry [55] (see Fig. 3).
Initial approaches of immune cell tracking with [ 111 In]-based radiotracers took place in the 70-ties. Segal et al. radiolabeled leukocytes in vitro with [ 111 In]oxine and administered them to patients with various inflammatory diseases: gut abscesses, bacterial endocarditis, rheumatoid arthritis, pyogenic infection. In each case, a high signal was detected in the inflammatory areas using a whole-body counter, and inflammatory tissues were visualized using planar gamma-scintigraphy [54]. This early study gave rise to further research, which established [ 111 In]oxine leukocyte scintigraphy or SPECT/CT as a practical and very accurate technique identifying inflamed tissues. Currently, not only rodents are used as an animal model in cell tracking studies employing this tracer. Afzelius et al. labeled a fraction of leukocytes with [ 111 In]oxine, yielding 57.7% À 69.7%, and administered it intravenously to pigs with inflammatory lesions caused by Staphylococcus aureus. The aim of the experiment was to compare the efficacy of osteomyelitis detection by radiolabeled leukocytes and several pure radiotracers. Although 111 In-leukocytes accumulated in 79% of inflammatory affected regions, the effectiveness in the case of 18 F-FDG PET was 100%, which made this method more reliable [40]. Nowadays, the [ 111 In]oxine labeled leukocyte method is used in humans, for example, to diagnose inflammatory conditions of the urinary bladder [41].
The development of successful immune cell-based anticancer therapies requires a deep understanding of in vivo biodistribution and antitumor activity of immune cells upon administration. Nuclear imaging research has significantly improved our knowledge about immune cell-based therapies against cancer by noninvasive and real-time tracking of immune cells and by visualization of their anticancer activity in numerous preclinical and clinical studies [2]. Noninvasive tracking of directly labeled immune cells using [ 111 In]oxine has been used to trace target tumors. For example, in 20 patients with different lymphatic malignancies and a perceptible lymph node enlargement, in vivo migration of leukocytes labeled with [ 111 In]oxine was studied in planar scintigraphy. Targeting of those cells to lymph nodes was observed in all patients suffering from Hodgkin's lymphoma or high-grade lymphoma, although, in the case of patients with low-grade lymphomas, effective targeting was observed in less than half of the patients [56]. SPECT imaging can also be useful for the evaluation of targeting directly radiolabeled adoptive T lymphocytes to tumors. CD8 + Tc lymphocytes (CTLs) with affinity to the Melan-A melanoma antigen were labeled with [ 111 In]oxine, administered to patients, and their targeting was evaluated using whole-body counter and static gamma camera imaging. Nuclear imaging allowed the detection of CTLs at metastatic sites, but there was also a signal in the liver, lungs, and spleen [57]. In another study, Pittet et al. labeled mouse T-cells expressing the hyaluronan-characteristic T-cell receptor with [ 111 In]oxine and injected them intravenously to mice with HA + CT44 and HA -CT26 tumors. Labeled CTLs were detectable in HA + CT44 tumors within 24 h of administration, and their numbers increased throughout the study. By comparison, there was very little homing of the labeled CTLs to the HA -CT26 tumors, suggesting that the tumor-targeting of these T-Cells was mainly TCRs mediated [58]. In a different study, human CAR-T cells were radiolabeled with [ 111 In]oxine to evaluate their trafficking in mice using SPECT/CT when different routes of cell injection were used. Compared to subcutaneous (S.C.) and intraperitoneal (I.P.) injections, optimal biodistribution was obtained after intravenous (I.V.) administration [59]. Finally, cd T cells, which are a specific type of T cells capable of identifying and killing neoplastic cells without involvement of MHC molecules, were directly labeled with [ 111 In]oxine and adoptively transferred to mice with 4 T1 mammary adenocarcinoma (a murine model of triple-negative breast cancer). SPECT and CT study of adoptively-transferred cd T cells has revealed that these cells targeted mouse tumors, and efficiently reduced their size [60]. NK cells are another type of immune cell with substantial antitumor activity, and thus NK cell-based anticancer therapies have emerged as a promising therapeutic approach for cancer treatment. Nuclear imaging has contributed to our understanding of therapeutic approaches focused on the administration of tumorinfiltrating NK cells [61]. In a study by Meller et al. SPECT imaging was employed to investigate the biodistribution and kinetics of in vitro expanded allogeneic NK cells [62]. For this purpose, NK cells were labeled with [ 111 In]oxine and administered with a tenfold excess of unlabeled cells to three patients with renal cell carcinoma. SPECT imaging revealed that shortly after transfusion [ 111 In]oxine labeled NK cells were primarily detected in lungs, but within the first 24 h they were redistributed from lungs to liver, spleen, and bone marrow. Moreover, 24 h after transfusion [ 111 In]oxine labeled NK cells could be observed in two out of four large metastases, which were confirmed by high glucose uptake with [ 18 F]FDG-PET. In another study, SPECT imaging was used to compare the migration of [ 111 In]oxine-labeled autologous NK cells to liver metastasis in colon carcinoma patients upon administration either via the systemic (intravenous) or locoregional (intraarterial) routes [63]. SPECT imaging analysis revealed that NK cells injected intravenously localized primarily to the lung, whereas localization of intraarterially injected NK cells was limited to the spleen and liver. Moreover, migration of NK cells to liver metastasis was observed only after intraarterial injections. The [ 111 In]oxine radiotracer was also successfully used in the biodistribution and tumor targeting study of NK cells generated from umbilical cord blood hematopoietic progenitor cells (UCB-NK cells) and adoptively transferred in immunodeficient NOD/SCID/IL2Rgnull mice [64]. Using SPECT imaging, Cany et al. demonstrated that [ 111 In] oxine-labeled UCB-NK cells migrated to liver, spleen, and bone marrow within 24 h after transfusion, and most importantly a single administration of UCB-NK cell impaired the growth of bone marrow-residing human leukemia cells injected intra-femorally in mice, leading to increased mouse survival.
Dendritic cells (DCs) induce protective adaptive immunity and thus have been a focus of anticancer vaccines and immunotherapy treatments. SPECT imaging has contributed to our increased understating of DC biology and DC-based anticancer therapies. In a study by Prince et al. PET and SPECT imaging modalities were employed to track the in vivo migration of monocyte-derived nonmatured DC (nmDCs) or matured DC (mDCs). For this purpose, DCs were labeled with [ 18 F]FDG or [ 64 Cu]PTSM for PET imaging and [ 111 In] oxine for SPECT imaging and administered to patients with multiple myeloma via (I.V.), intradermal (I.D.), (S.C.), and intranodal routes [65]. Among tested radiotracers, [ 111 In]oxine showed reproducible tracking of both types of DCs to regional lymph nodes after either S.C. or I.D. administration, with mDCs showing superior migration to regional lymph nodes. [ 111 In]oxine-SPECT imaging was also successful in the in vivo monitoring of lymph node migration of [ 111 In]oxine labeled DCs in a murine breast cancer model (MMTV-Ras). Labeling of DCs with [ 111 In]oxine, had no adverse effect on DC phenotype or functionality, and most importantly SPECT imaging revealed the presence of [ 111 In]oxine-labeled DCs in both axillary and popliteal lymph nodes. These results were confirmed with immunohistochemistry and c-counting analysis [66]. [ 111 In]oxine binds to the iron transport protein transferrin in blood plasma and so the blood plasma must be removed from patient samples before the cell labeling step, however, this process may affect cell viability. To overcome this problem, other [ [67,68]. Although these radiotracers exhibited effective labeling of leukocytes in a plasma, they did not become commonly applied agents for nuclear imaging of leukocytes in vivo. Nevertheless,It is worth mentioning an experiment in which human blood-derived cd T cells were labeled with [ 111 In] tropolone and then injected I.V. into NSG mice with A375Pb6.luc cell tumors, inoculated I.V., S.C. and I.P. The I.P. model showed the highest cd T cell uptake of the three examined models [69].
Direct immune cell labeling with 99m Tc-based radiotracers. The promising results obtained with [ 111 In]-labeled leukocytes studies inspired efforts to develop analogous methods of cell radiolabeling with [ 99m Tc], which due to its unique physical properties is now the most commonly used radioisotope for gamma-based diagnostics [70]. . Moreover, this radioisotope guarantees a high signal-tonoise ratio, enabling the detection of even small numbers of cells. Stable isotopes of technetium do not exist and so most of the Tc used in laboratories is freshly generated from nuclear reactions. The most commonly used isotope [ 99m Tc] is produced in a generator by the decay of the longer-lived isotope -[ 99 Mo]. The half-life of [ 99m Tc] is 6 h and it decays with 12% internal conversion and 88% gamma photon emission at 140 keV, making it excellent planar gamma scintigraphy and SPECT imaging isotope ( Table 2). Technetium is available at oxidation states from (-I) to (+VII), but the most stable states are (+VII) and (+IV). The lower oxidation states (+I), (+II), and (+III) need to be stabilized with chelating ligands to prevent technetium hydrolysis products from forming, which interfere with subsequent labeling efficiencies.  Fig. 4. The resultant lipophilic complex is passively absorbed across the cell membrane and is trapped within cells via a combination of reduction to a hydrophilic complex and binding of the complex to non-diffusible proteins and cellular organelles [71].
Comparison of [ 99m Tc]HMPAO and [ 111 In]oxine biodistribution shows significant differences. [ 99m Tc]HMPAO has significant intestinal activity due to its biliary secretion route [71]. This makes [ 99m Tc]HMPAO less effective in the case of cell tracking within the abdomen. The low energy emission properties of the radionuclide [ 99m Tc] can be used at a higher dose than [ 111 In] which confers an advantage for imaging small parts of the body. The disadvantage is its relatively short half-life giving only a short period for the tracking of immune cells. [ 99m Tc]HMPAO is mainly applied in imaging of infections [6]. In 1990, 19 patients suffering from inflammatory diseases were injected with leukocytes labeled with [ 99m Tc]HMPAO and imaged with a gamma camera [72]. This study achieved 100% specificity, 93% sensitivity, and 95% accuracy. This was later repeated by Reynolds [75]. They observed a close correlation between the fecal excretion of [ 111 In] and the result of the [ 99m Tc] image analysis, which confirms the usefulness of these radiotracers in the context of determining inflammatory bowel activity. A diagnostic method involving the venous leukocyte fraction is also used today in the clinics. Migliari et al. reviewed 490 studies with labeled leukocytes in the context of infection diagnosing (mainly bone and orthopedic implant infections). They concluded, that this method is characterized by its ease of application in routine clinical practice, good cell labeling efficiency (72%) and high-quality images [76]. Jorgensen et al. have approached the radiolabeling of the pure lymphocyte fraction and confirmed the effectiveness of lymphocytes for targeting arthritic joints [38]. The research aimed at rheumatoid arthritis (RA) visualization was also performed with the use of autologous bloodderived monocytes which upon labeling with [ 99m Tc]HMPAO, were re-infused to RA patients. Scintigraphy [77] or SPECT imaging [78] of the hands and feet of RA patients revealed clear images of reinfused, radiolabeled monocytes localized to the inflamed joints. Importantly, the biodistribution of [ 99m Tc]HMPAO-labeled monocytes is similar to that of [ 99m Tc]HMPAO-labeled white blood cells and might therefore be employed for live monitoring of antirheumatic therapy in patients [79]. In 2011 Ruparelia et al. showed significant migration of [ 99m Tc]-labeled neutrophils to the target tissue in patients suffering from a chronic obstructive pulmonary disease (COPD). [ 99m Tc]-labeled neutrophils were administered and tracked using sequential SPECT imaging in six COPD patients and three volunteers. The obtained images demonstrated significant levels of neutrophil migration to the lungs in COPD patients. The results corresponded well with a whole-body counter quantitative analysis of [ 111 In]tropolone radiolabeled neutrophils [80]. Lukawska et al. compared the kinetics of autologous granulocytes and eosinophils in healthy human patients. These two populations of leukocytes were radiolabeled with [ 99m Tc]HMPAO, administered intravenously and SPECT images were taken after 1, 5, 15, 25, and 120 min. Granulocytes and eosinophils showed different distribution patterns to the liver, lungs, and mediastinum. Favorable trafficking of eosinophils to the spleen and bone marrow and neutrophils to the liver was proven [81]. The same team evaluated the targeting properties of neutrophils and eosinophils to the lungs of patients suffering from asthma [82]. The patients were divided into three groups: (1) early allergic responders, (2) early/ late allergic responders pre-treated with antigen, and (3) allergic patients treated with steroids. Using the gamma camera, a higher net retention time of eosinophils was observed in the first group of asthma individuals than the non-smoking group from the previous study. Further development of this technique may bring the discovery of a key tool for asthma phenotyping. This observation corresponds with the recent study, in which autologous venous blood eosinophils radiolabeled with [ 99m Tc]HMPAO were I.V. injected to healthy and obese asthmatic patients, showing higher eosinophil trafficking in the lungs of asthma group patients [83]. Interestingly, it was acknowledged that treating leukocytes with [ 99m Tc]HMPAO results in the preferential labeling of granulocytes, with the strongest selectivity of eosinophils over other leukocytes, which indicates the utility of this radiotracer application during eosinophillabeling studies [84]. Pillay et al. observed, that the trafficking of autologous neutrophils labeled with [ 99m Tc]HMPAO initially occurred in the lungs, over time their number in the lungs decreased and after 24 h they were detectable only in the liver and spleen. Interestingly, the administration of C-X-C chemokine receptor type 4 (CXCR4) antagonist did not affect neutrophils' distributions [85].
[ 99m Tc]HMPAO was also successfully used for in vivo tracking of specific populations of immune cells such as T cells, DCs, and NK cells in the context of cancer immunotherapy. Sharif-Paghaleh et al. employed SPECT/CT analysis to evaluate the efficacy of DC vaccination by early detection of [ 99m Tc]HMPAO-labeled CD4 + T cells in draining lymph nodes [86]. Another [ 99m Tc]HMPAO/SPECT imaging study investigated the lymph node migration of mDCs and immature DCs (iDCs) and the impact of administration route (intradermal versus subcutaneous) on this process. SPECT analysis showed that intradermal administration of [ 99m Tc]HMPAOlabeled DCs resulted in about a threefold higher migration to lymph nodes than subcutaneous administration, and mDC exhibited higher migration than iDC. DCs were first detected in lymph nodes just 20-60 min after inoculation and the maximum numbers were detected at 48-72 h post inoculation [87]. The [ 99m Tc]HMPAO radiotracer was also used for labeling and in vivo tracking of lymphokine (IL-2)-activated killer (LAK) cells induced from T cells (T-LAK) and NK cells (NK-LAK) isolated from patients with head and neck carcinoma [88]. [ 99m Tc]HMPAO-labeled LAK cells were administered back into the respective patients locally into the tumor tissue or via the superficial temporal artery. The administered LAK cells were tracked using a gamma camera. T-LAK cells were retained longer in the tissue than NK-LAK cells, because T-LAK cells were less adherent and less chemotactic to endothelial cells, and exhibited decreased migration through endothelium as compared with NK-LAK cells.
Another method of leukocyte labeling with [ 99m Tc] involves the phagocytic engulfment of a radiolabel being in the form of a colloid. An attempt has been made to label blood cells with [ 99m Tc]stannous oxide colloid ([ 99m Tc]-SnF 2 ). Although the labeling efficiency was between 80 and 90%, most of the radiolabel was accumulated in phagocytic cells: neutrophils and monocytes. The radiolabeled monocytes were reinjected to the patients with the tentative diagnoses of an abdominal abscess or septic loosening of an endoprosthesis, resulting in visualization of local inflammatory foci [89].  F]-based radiotracers generate images of high-resolution due to the high b + decay ratio (97%) and low positron energy (maximum 0.634 MeV). Its main drawback for cell tracking concerns its short half-life of 109.8 min, which impairs complex radiosynthesis, the performance of extended in vivo studies, and delivery to PET centers with no radiochemistry facilities. It entails the necessity of easy access to the freshly synthesized radiocompound. The most widely applied radio-compound both in medicine (tumor diagnosis) and cell labeling studies, is [ 18 F]FDG. This molecule is an analog of glucose with a positron-emitting [ 18 F] atom connected to the 2 carbon instead of a hydroxyl group. It is absorbed by live cells, via the GLUT transporters. The family of these membrane-bound proteins consists of 5 subtypes, however, GLUT-1 and GLUT-3 are considered to contribute to the in vivo accumulation of FDG in malignant tumors [90]. In the cytosol, it enters the glycolytic pathway, which leads to its phosphorylation by hexokinase to [ 18 F]FDG-6-phosphate. This molecule is not capable of leaking out from the cell, causing a metabolic trap. But elution of unphosphorylated [ 18 F]FDG which is initially taken up, is a major drawback that results in continuous retention and loss of cell radioactivity. For instance, the cell labeling efficiency decreased in two hours from 89.9% to 62.1% in human stem cells, and from 91.6% to 68.6% in porcine stem cells [91]. In a different study, the labeling efficiency of the fraction of human bloodderived leukocytes decreased from 64 to 75% to 39%-44% after 90 min and eventually to 19% after 240 min [92].
Many types of white blood cells as well as mixed fractions of leukocytes were radiolabeled with [ 18  F]FDG, it covalently binds to functional groups of the cell membrane surface and more specifically -to exofacial protein thiols (EPT). The limitation of this method is that was found to be a large variation in the labeling efficiency between cell types that couldn't be explained by the variation in EPT levels alone. Jurkat and SaOS-2 cells were radiolabeled at a fairly low level (7.7 and 7.1% respectively), whereas primary T lymphocytes at a relatively high level -60.5% [99]. The in vivo PET tracking of these cells was performed ; in SR images, green color represents PET signal; in the MIP image, PET signal intensity is reflected by the color bar; the tumor was not visualized, but strong, the asymmetric signal from lungs was detected; the signal from the bladder indicates that part of the radiotracer was eluted. C -schematic location of lung metastases (upper image) and representative histopathological image of the tumor focus from examined sample (lower image). The PET/ CT images were generated using PMOD software (PMOD Technologies LLC, Zurich, Switzerland). successfully, following intravenous injection of the cell suspension to the naïve C57BL/6 mice, and the signal was detected mainly in the spleen. Thus, this method has promise for certain cell types evaluation of labeling efficiency and label retention is required for each new cell type to be labeled. The most important feature of this method compared to [ 18 F]FDG labeling is distinctly lower radiotracer retention. There was also an approach of labeling murine spleen-derived leukocytes with [ 18 F]FBEM and evaluation of their recruitment to the lungs with fibrosis in mice. Interestingly, PET images showed greater and faster trafficking of leukocytes in the mice treated with bleomycin than control mice [100]. An efficient and quick procedure of automated radiosynthesis of [ 18 F] FBEM was recently introduced by Lim et al. [101] The [ 18 F]-containing radiolabel that demonstrates a cell incorporation mechanism similar to that observed for fluorescent dyes, is hexadecyl-4-[ 18 F]fluorobenzoate ([ 18 F]HFB). It is a novel, lipophilic ester derivative, which is incorporated into the cell membrane but does not enter the cytoplasm. The labeling efficiency was evaluated so far only in rodent mesenchymal stem cells -after 30 min it reached 25% and cell viability was found to be > 90%, furthermore, there was a retention of radiotracer in the cells > 90% over 4 h. The labeling procedure was quick and simple [44] .5% vs 7.6% ± 4.1%) and stable (88.4% ± 6.0% vs 26.6% ± 6.1% after 4 h). The CPC radiolabeling with [ 18 F]HFB resulted in a more effective, accurate, and stable way of quantifying cell migration as shown in a rat myocardial infarction model in vivo [103].
Direct cell labeling with [ 64 Cu]-based radiotracers. [ 64 Cu] is a theranostic radionuclide endowed with the remarkable property of releasing both b + (17.6%) decay for PET imaging and b -(38.5%) decay for therapeutic applications [104,105]. The maximum b + emission energy (656 keV) and the mean positron range (0.7 mm) compare favorably to [ 18 F] (250 keV, 0.6 mm) [106]. [ 64 Cu] has a half-life of 12.7 h allowing for the tracking of longerlived biological processes over 1 to 2 days. Copper is a group 11 transition metal element that is capable of accessing oxidations states (+I), (+II), and (+III) although in aqueous solutions Cu(+II) predominates. Copper is a borderline Lewis acid [107], meaning it is a versatile metal that can form stable complexes with a wide variety of donor atoms including hard Lewis bases such as nitrogen and oxygen, and soft Lewis bases such as phosphorus and sulfur. Cu(+II) complexes commonly take a distorted octahedral form with a coordination number of 6. Cu(II) complexes are typically very labile meaning ligands can rapidly exchange [108,109] leading to a breakdown of complexes and release of the metal ion from labeled constructs. As a result, careful design of multidentate and macrocyclic ligands are required for [ 64 Cu] radiolabeling strategies when metal ion release is not desired such as in the radiolabeling of antibodies, peptides, or other targeting moieties. This an extremely broad topic and has been the subject of many recent comprehensive reviews [52,34,105,[109][110][111][112][113][114] 18.9 h revealed that the lymphocytes initially traveled through the lungs and then accumulated in the spleen and liver [39]. Park and co-workers compared two radiolabels in the context of leukemia cells (K562-TL): [ 124 I]FIAU and [ 64 Cu]PTSM claiming that the second one is more efficient, because of its markedly higher labeling efficiency and lower in vitro but also in vivo efflux [115]. Griessinger and collaborators labeled in vitro murine OVA Th1 lymphocytes, reaching labeling efficiency of 6.5% -9%, however, retention studies indicated that only after 5 h, 47% of added [ 64 Cu]PTSM was still observed in the cells, however, after 24 h it was only 14%. They injected cells I.P. or I.V. to mice with airway hyperreactivity (AHR) induced by OVA and also to naïve control mice. The cells were targeted specifically to the hyperactive sites and remained visible for over 48 h. These authors were able to detect 60 000 [ 64 Cu]-labeled Th1 lymphocytes in a single lymph node 20 min after intraperitoneal administration [116]. In an alternative study [ 64 Cu]tropolone was evaluated as a radiolabel for the human WBC fraction. Although the initial labeling efficiency was encouragingly high (83%), the rapid elution of radionuclide (nearly 50% after 4 h and > 70% after 24 h) makes it impractical for potential application in this setting [117]. Similar approach to white blood cell tracking was introduced by Socan et al. Various radiotracers, including [ 64 Cu]oxinate and [ 64 Cu]tropolone were synthesized using specific anion-exchange cartridges. This novel method allows for the synthesis of oxinate and tropolonates in small volumes, suitable pH and the reaction is characterized by high yield (94.8% ± 2.4 for tropolone and 76% ± 20.3 for oxine) and high extraction efficiency (>94%). In this study, the animal model was BALC/c mouse and the targeted pathologies were muscle infection caused by intramuscular (I.M.) inoculation of P. aeruginosa and muscle sterile inflammation induced by I.M. injection of turpentine oil. WBC fraction isolated from BALB/c mice was successfully radiolabeled with [ 64 Cu]oxinate and [ 64 Cu]tropolone, reaching 57.1% ± 8.6 and 95.6% ± 2.6 labeling efficiency, respectively. The fractions of radiolabeled WBC were reto-orbitally injected and the series of PET scannings was performed. Clear signal indicating the presence of labeled leukocytes was detectable at sites of infection and inflammation [118].
To label cells with the [ 64 Cu]-containing radiolabel, another mechanism was employed. An organic polymer, polyethylenimine (PEI) was used to create a stable complex with [ 64 Cu]. Interestingly, this reaction was feasible without using a metal chelator. PEI binds to anionic heparin sulfate proteoglycans presented on the cell membrane and is transported into the cell via endocytosis [119]. Li  Cu]PTSM -initially, the cells were retained in the lungs, after 20 h redistributed to the liver but radioactivity was also found in the kidneys [35].
To overcome the problem of low radiotracers' retention, we developed a system in which [ 64 Cu] was incorporated within the apoferritin cage and this complex was subsequently loaded by endocytosis to macrophages to track their distribution in lungtumor-bearing mice. We called this diagnostic conjugate ''MRIC" -macrophage radioisotope conjugate. The [ 64 Cu] apoferritin incorporation procedure was previously proposed by Wang et al. [120]. Briefly, 1 ml apoferritin solution in acetate buffer 50 mM at 2 mg/ mL concentration, pH 6.0, was mixed with 1 ml of a [ 64 Cu] solution (about 1 GBq/mL, in 0.5 M HCl) neutralized with concentrated sodium carbonate (Na 2 CO 3 ). The solution was placed under stirring (Vortex) for 2 min and then 30 min in a shaking heater at 45°C. Thereafter, 100 ml of a 0.1 M sodium sulfite solution was added to the reaction and warmed again at 50°C for half an hour. Finally, the reaction mixture was exchanged for PBS buffer using a Vivaspin 500 concentrators PM30 in an Eppendorf centrifuge at 2000 g (4500 rpm for rotor MSC-6000) for 10 min. This was repeated two further times and the protein was recovered, diluted to the desired concentration for cell loading, and sterile filtered (overall protein recovery about 1 mg). Such method allowed for safe and fast labeling of the apoferritin protein with [ 64 Cu] with negligible leaking of the isotope, which was rendered water-insoluble inside the cavity after transformation onto its sulfide derivative. Further optimized human apoferritin mutants have been recently obtained that allow for faster and efficient [ 64 Cu] incorporation [121].
To be able to track RAW264.7 macrophages longitudinally in tumor-bearing mice (CT26 colon cancer cells injected on the thigh according to the procedure already described [122]), we loaded them with the complex of apoferritin and [ 64 Cu] and injected them into the tail vein. Using PET/CT we observed a specific accumulation of the radioactive signal in the tumor in a time-dependent manner showing macrophage trafficking to this site (Fig. 6). This study showed that even a single administration of autologous or allogeneic macrophages loaded with apoferritin-[ 64  Zirconium is a group 4 transition metal element, existing mainly in the (+IV) oxidation state. Because of its small size and high charge, Zr(+IV) is considered to be a hard Lewis acid meaning it preferentially coordinates to hard Lewis bases like oxygen and nitrogen [107]. Zr(+IV) has a preference for 8 coordinate complexes as demonstrated by Kathirgamanathan and co-authors who prepared and crystallized zirconium tetrakis(8-hydroxyquinolinolate) for the preparation of organic light-emitting diodes [124]. The formation reaction and structure of [ 89 Zr]oxinate 4 [125] is given in the Fig. 7.
This work was of interest to the nuclear medicine community as 8-hydroxyquinolinoline (oxine) has been employed since the mid-1970 s for radiolabeling cells with [ 111 In]oxine for scintigraphy studies and SPECT imaging. In the case of this compound, the cell labeling procedure is analogous to the other lipophilic agents (HMPAO, PTSM, tropolone). They passively diffuse across the cell membrane, and then the isotope-chelator complex breaks down in the cytosol in a reduction reaction and the isotope binds to intracellular proteins. Labeling efficiency of these cell lines was in the range of 40 -61%, the retention measured after 24 h upon labeling was high (from 71% to 90%) and cell viability > 90%, without a significant decrease after 24 h [126]. In another study, Sato et al. attempted to label BMDCs and naïve splenic T lymphocytes to evaluate their tumor-targeting properties in the mouse model. Interestingly, DC labeling efficiency was twice higher (>40%) than for lymphocytes (10% for naïve and > 20% for activated by TCR stimulation). PET tracking of intravenously injected cells was performed for 7 days. PET images showed that DCs were mainly distributed to the liver and spleen after passing through the lungs and Tc lymphocytes were mostly trafficked in the lymph nodes and spleen. The most interesting outcome from this study was that only a small fraction of these lymphocytes targeted the tumor [127]. A similar distribution was also observed in our studies [123,128], which corresponded well with recent outcomes. Watson et al.. isolated murine spleen and lymph node-derived CD8 + T cells and radiolabeled them with freshly synthesized [ 89 Zr]oxinate 4 reaching 18-20% labeling efficiency and > 90% cell viability. After intravenous injection of 8 -17 mln lymphocytes to C57BL/6J mice with subcutaneously inoculated B16F10 tumors, PET imaging was performed. The signal from the tumors was observed within the first-hour post-injection and was gradually increasing alongside the tumor growth until reaching the final time-point of imaging, which was 188 h. This shows immediate diapedesis of T cells from the vascular system and their migration to solid tumors. This phenomenon also gradually increases over time [123]. hNIS-GFP) via intramammary administration. Tumor targeting of lymphocytes was significantly increased in the case of injection of alendronate-loaded liposomes, suggesting that they increase leukocytes' migration to the tumor [129]. Adoptive therapy with chimeric antigen receptor T cells (CAR-T) has recently become a very exciting clinical concept in human oncology [130]. Weist et al. conducted a preclinical study involving the transplantation of [ 89 Zr]oxinate 4 -labeled human CAR-T cells to NSG mice, to evaluate their targeting of glioblastoma brain tumors (cells administered intraventricularly) and subcutaneously inoculated prostate tumors (cell administered intravenously). That was the first approach of human lymphocytes radiolabeling with this tracer. As expected, cell labeling efficiency was high (75%) and radionuclide retention measured after 6 days was satisfying (60%). The cells injected intravenously were distributed to the tumor over 6 days as shown with PET imaging, and the ones that were intratumorally injected were visible only in the region of the tumor for the whole study [131]. Concluding, the efficacy of this radiotracer in tracking many types of cells have been confirmed in numerous preclinical studies, not only in mice models but also in nonhuman primates [132].  4. Importantly, the labeling step did not affect the phenotype, function nor viability of the cells. The macaques were intravenously injected with radiolabeled NK cells, which were PET tracked for 7 days of the experiment's duration. Significantly, the monkeys were treated with a continuous I.V. infusion of deferoxamine, starting just before the cells' administration. The role of deferoxamine was to chelate the eluted zirconium what leads to the renal excretion of the radionuclide and prevents its accumulation in the bones. PET imaging revealed the initial accumulation of NK cells in the lungs and subsequently, they were distributed to the liver, spleen, and to a lesser extent, bone marrow. The most important observation from this study appears to be the confirmed safety of this type of adoptive cell transfer method. The fact of finding no side effects in a species close to humans allows us to consider this method as applicable in medicine [135].
The development of chelator chemistries for [ 89 Zr] to enable the radiolabeling of antibodies for immuno-PET imaging is an extremely fertile field of research and there have been several excellent reviews covering recent developments in this area [109,112,[136][137][138][139]. The most commonly used methodology is a two-step process whereby a deferoxamine based bifunctional chelator (DFO-Bz-NCS) is covalently bound to amine groups of lysine side chains on the surface of the antibody through NCS coupling followed by [ 89 Zr] chelation to the DFO in aqueous conditions at neutral pH [140]. Interestingly, DFO has also found application in the effective radiolabeling of human hemoglobin, where the chelator is bound to a cysteine residue at position b93 [141]. The formation reaction and structure of [ 89 Zr]DBN (or [ 89 Zr]DFO-NCS) is presented in the Fig. 8. Notice that as a hexadentate ligand DFO does not fully satisfy the 8 membered coordination sphere of Zr.
One interesting adaptation of this chemistry is to reverse the order of the process and form [ 89 Zr]DBN and then covalently bind this to the lysine residues of proteins presented on a cell's outer membrane as a novel approach to cell labeling [45]. In the first study applying this method, murine melanoma cells, murine DCs, and human mesenchymal stem cells (hMSCs) were radiolabeled with [ 89 Zr]DBN with an efficiency of 30% to 50%. Importantly, this method of labeling did not affect cell viability and their ability to proliferate, however, the most important outcome was no efflux observed for over 7 days post labeling. Radiolabeled hMSCs were intravenously injected into athymic mice and their migration was tracked by PET imaging for 7 days. They were also injected into the myocardium within the ischemic regions of a reperfusion mouse model, being retained in ischemic lesions for the whole imaging period. Lee et al. performed a similar study -they radiolabeled Jurkat/ CAR-T cells with [ 89 Zr]DBN, reaching 70 -79% of labeling efficiency and > 95% of cell viability after the procedure. Subsequently, the cells were I.V. injected into mice bearing Raji and K562 celltumors, inoculated on the left and right flanks. The mice were PET imaged for 7 days post-injection, revealing that the CAR-T cells progressively migrated from the lungs and trafficked mainly in the liver and in part the spleen. However, an unexpected result was, that cell targeting to the tumors was not observed on PET scans [142]. The radiolabeling of human cardiopoietic stem cells (CPs) using this radiotracer was also successfully performed [143]. This is the only one known radiolabel that does not exhibit significant leaking out of the cells and thus indicates a high potential and wide use in future studies. However, our research results suggest that lymphocytes and macrophages labeled with [ 89 Zr]DBN may non-specifically create in vivo focal shaped clusters in murine lungs, which may lead to false-positive results in PET studies concerning cell targeting to pathological lesions, such as lung tumors (Fig. 9).
[ 89 Zr] can also be applied in the form of chitosan-conjugate. Chitosan is a biocompatible copolymer of glucosamine and Nacetylglucosamine which can chelate metal ions. This nanoparticle can be incorporated through phagocytosis, therefore the phagocytic capacity of treated cells is essential. Nevertheless, Fairclough et al. performed the labeling of the blood-derived leukocyte fraction, reaching an efficiency of 82.7% after 10 min of incubation, however, continuous, gradual efflux was observed and it reached 53 [145] and T lymphocytes [146,147] in murine models.
Other radiotracers in direct PET imaging. An unusual approach to   their membrane thiols with 22%-62% labeling efficiency with extended radiotracer retention. Due to its fluorescent characteristics, the cell membrane localization of [ 124 I]FIT-(PhS) 2 Mal was proved by confocal fluorescence microscopy. Prolonged monitoring of the in vivo distribution of [ 124 I]FIT-(PhS) 2 Mal-labeled Jurkat cells showed that it can be monitored with PET for over 7 days [46]. The same radionuclide is also applied as a conjugate with deoxyuridine ([ 124 I]dU). The cell penetration of uridine is based on the principle of transporter uptake -via the concentrative nucleoside transporters (CNT1) [148] and it is incorporated into DNA [149,150]. Although the conjugates of deoxyurudine with various radioisotopes have been used in diagnostics for a long time, they have been mainly used for cancer detection due to their high uptake by the proliferating cells [151]. However, there was an approach in which OVA-specific CD8 + T cells were in vitro radiolabeled with [ 124 I]dU. Because the results showed a distinctive accumulation of injected radiolabeled T cells in B16-OVA-cell melanoma tumors of C57BL6 mice, it was confirmed that this method is effective for the labeling and long-term tracking of T cells [36]. Jung et al. reported the feasibility of a technique termed ''CellGPS", which allows a single radiolabeled cell to be tracked in vivo using PET imaging. Mesoporous silica nanoparticles (MSNs) concentrating high amounts of [ 68 Ga] were loaded into MDA-MB-231 breast cancer cells. Subsequently, an isolated single cell was I.V. administered into an athymic nude mouse. Using PET, it was possible to detect and visualize that single cell. As expected, the arrest of the cell was eventually observed in the lungs [152]. [ 68 Ga] was also used in connection with oxine ([ 68 Ga]oxinate 3 ) to radiolabel and track CAR-T lymphocytes in NSG mice, however, their targeting to CD19-K562-luc and Raji tumors was slow and no significant accumulation was reported. The authors claim, that although [ 68 Ga] allows performing shorter study comparing to [ 89 Zr], it has a better safety profile what suggests, that it could be safely translated into human studies [153].

Tracking of immune cells with radiolabeled antibodies -SPECT and PET
Another interesting approach of nuclear imaging is cell labeling with radionuclide-conjugated antibodies. In contradiction with the direct cell labeling methods, when tracking cells via this strategy the radioisotopes must be tightly bound to the targeting antibody, particularly if the antibody targets a non-internalizing cell surface receptor, as any released radiometal will no longer be cellassociated. As a result, different chemical strategies are required from the direct cell labeling methods discussed above. Radioisotopes may be connected to antibodies directly (e.g.,  [159]. Further developments and refine-ments of [ 99m Tc] direct labeling methodologies have been excellently explained and reviewed by Rhodes et al. [160] Bifunctional chelator-based labeling strategies rely upon the formation of coordination bonds which individually are weaker than covalent bonds and are prone to exchange in aqueous solution, hence appropriate chelator design is vital to minimize loss of radiotracer in vivo when employing this strategy. Guiding principles that apply to chelator design for all radiometals are the chelate and macrocyclic effects whereby multidentate and macrocyclic ligand complexes are vastly more stable than the equivalent monodentate complexes of the same coordination number and donor atom type. The strength of the chelate effect increases as chelate denticity increases, which explains the preponderance of high denticity chelates in radiochemistry applications. The primary driving force behind the chelate effect is the entropy increase caused by displacing multiple coordinated monodentate ligands with a single multidentate chelator ligand leading to an increase in the total number of individual molecules upon complex formation. Additionally, metal complexes are at high dilution compared to the competing coordinating solvent molecules meaning monodentate molecules will be readily diluted out during ligand exchange equilibria in solution whereas multidentate chelating ligands will remain tied in close vicinity to the metal ion maintaining competition until all the coordinating bases have been exchanged from the metal. As discussed by Martel et al. [161], numerous additional elements also influence the chelator effect to greater or lesser degrees depending on the characteristics of the chelating ligand and the metal ion. Each radiometal has its own unique set of chemical properties such as preferred-oxidation state, coordination number, coordination geometry, donor ligands, and ionic radius. Ignoring steric hindrance, small monodentate coordinating ligands are free to conform to the preferences of the coordinated metal ion whereas macrocyclic or multidentate coordinating ligands need to be pre-formed with the correct charge, number, and type of coordinating bases and have the conformational flexibility to satisfy the preferred coordination geometry of the metal to be coordinated. As a result chelators need to be designed and optimized specifically for each radiometal and it is crucial to match the appropriate chelator to the adequate radiometal. The design and development of chelators for individual radiometals is an extremely fertile field that has been the topic of many excellent recent reviews and the reader is directed to these for further details [50,51,109,111,112,138,153].
A unique aspect of the antibody-based labeling technique is the possibility to label cells in two different ways: in vivo -after intravenous administration of the antibody, and in in vitro conditionsbefore administration of the cells. Although the first method has the drawback of non-specific labeling of body cells that have the same receptor as the cells to be tracked, it is much more commonly applied. First attempts of using radiolabeled antibodies are dated back to the 1950s but the development of imaging methods allowed their effective detection only in the 70s [162]. Radiolabeled therapeutic or imaging antibodies approved by FDA and EMA are as follow:  [166]. In these groups, many compounds may be potentially used in cell radiolabeling and tracking trials. squamous carcinoma [192] Trastuzumab [ 111 In] HER2 n/a breast cancer [193,194,195] ABY-025 [ 111 In] n/a breast cancer [196] [46].
The examples of radionuclide-conjugated antibodies with their affinity to certain molecular targets and the isotopes that they are conjugated are given in Table 3.
In summary, many types of radiotracers are used in immune cells labeling. It is important to be aware that many molecules combined with the radioisotope to form the radiotracer were not originally designed to track cells, so some of them have significant limitations. Using radiotracers that are taken up by the cells that express the specific receptor, in the case of in vivo efflux of the radiotracer, may lead to the undesirable labeling of other cells of the organism. Some macromolecular radiotracers are characterized by a great capacity for radioisotope particles, what reduces their molar concentration in in vitro cell labeling step, however, they may be applied mainly in the case of phagocytic cells. Radiotracers that depend on transporter-mediated uptake are usually associated with their quick elution, furthermore, most of them have a relatively short half-life. In case of the radiotracers capable of passive diffusion across the cell membrane, low retention is also a big concern, however, their ability of unspecific labeling of different kind of cells regardless of their receptor profile is undoubtedly a big value. Fortunately, some compounds are devoid of the efflux effect, such as those labeling cells by absorption into the cell membrane, however with relatively low efficiency, or those binding to the cell membrane proteins, but with a risk of impairing cell  Typically rapid elution of a radiotracer [45,71,77,115,131] Absorption into the cell membrane The labeling process does not affect cell viability, phenotype, or migration potential Low elution of radiotracer Typically low labeling efficiency [44,34,103,117] Binding to the cell membrane proteins No cellular efflux of a radiotracer -stable retention In theory, binding of a radiotracer to membrane proteins, has the potential to affect cell functionality [45,46,99,100,200] Ł. Kiraga function. Nevertheless, the correct choice of radioactive agent should depend on the cell type intended to be labeled and the duration of the cell tracking study. The advantages and disadvantages of each direct labeling process are listed in Table 4.

Indirect cell labeling
Indirect labeling with reporters involves the introduction of a reporter gene to the immune cells under in vitro conditions before cell transplantation. Alternatively, immune cells isolated from transgenic animals that express a reporter gene of interest can be used [217]. These reporter genes are designed to induce an additional function to the cells to make them uniquely targetable in vivo by a chosen radiotracer. Reporter genes typically encode for enzymes or proteins capable of binding with high specificity and affinity to a particular radiotracer that may be administered repetitively at any time following cell transplantation. The indirect cell labeling provides several advantages as compared with the direct techniques, including the possibility of longitudinal monitoring of labeled immune cells over their entire lifetime, as well as tracking of proliferating and viable cell populations that express the imaging reporter genes. However, transcription of the reporter genes may be compromised by gene silencing through epigenetic mechanisms such as histone deacetylation or DNA methylation, leading to suppression of the reporter gene expression [218]. Treatment of cells with a DNA methyltransferase inhibitor may overcome this problem [219].
Both viral and non-viral methods have been used for the successful introduction of reporter genes to cells. Viral gene transfer makes use of viruses to introduce the reporter gene into the genome of the target immune cell and does not require any transfection reagents. Nowadays, lentiviruses and retroviruses are most commonly employed for genetic manipulation of immune cells [168,[220][221][222][223]. Reporter gene transfection with non-viral methods   involves the use of electroporation, polymers, chemical vectors, nanoparticles, or cationic transfection agents [224]. Gene transfer with non-viral techniques is rarely used for genetic manipulation of immune cells. In general, three major groups of imaging reporter genes can be distinguished [10,225,226]: 1) enzymes -resulting in metabolic entrapment of the radiotracer inside the cell, 2) transmembrane receptors -binding radiolabeled probes, 3) transporters -actively pumping the radiotracer into the cytoplasm.
Different radiotracers grouped by the mechanism of indirect cell labeling are presented in Fig. 10.
One of the most broadly evaluated imaging reporter enzymes for nuclear imaging is HSV1-TK and its many variants, which mediate the uptake of substrates tagged with radioisotopes suitable for SPECT or PET analysis. Radioactively  [227,228]. Several studies have shown that tumor-infiltrating lymphocytes can be successfully tracked with the HSV1-TK reporter enzyme in mice [229,230] or in humans [231]. However, immune cells expressing HSV1-TK may evoke potential immune reactions due to the viral origin of this reporter enzyme. Similar immune reactions are expected to occur in response to cells expressing other non-human reporter genes, such as eDHFR and VZV-tk. Immunogenicity against non-human reporter proteins may lead to major problems in cell kinetics monitoring studies due to eradication of cells expressing the reporter gene. Thus, to overcome this shortcoming and to reduce immunogenicity, various human-derived reporter genes have been developed. However, the exploitation of such endogenous human genes for in vivo cell tracking has three potential problems: first, the reporter probes may also accumulate in cells expressing the endogenous gene causing high tissue background; second, the imaging reporter gene may act like the endogenous gene and thus affect the functioning of the cell in which it is expressed; third, there is still limited knowledge concerning the immunogenicity of these human imaging reporter genes. To overcome some of these problems mutations have been introduced in some of these genes to either eliminate their function [232] or to increase their detection sensitivity with a specific reporter probe [233]. Two human-derived, enzyme-based reporter genes, hmTK2 and hdCK, represent attractive nuclear imaging reporter genes for live cell tracking imaged with pyrimidinebased radiotracers [229]. The second group of reporter genes used for tracking live immune cells with nuclear imaging are receptors, including human dopamine receptor type 2 (D2R) and hSSTR2, which bind their radiotracers on the outside of the cell. The hSSTR2 is a non-immunogenic protein that has been used as an imaging reporter receptor for immune cell tracking using its ligand [ 68 Ga] DOTATOC [234,235]. Selective radiotracer import through transporters, such as hNIS and hNET, is another method for live immune cell tracking with nuclear imaging. hNIS is a glycosylated ion channel present on the cell membrane and is expressed in salivary and thyroid glands, but also in the stomach, where it transports iodine to the cells for sodium exchange. Its expression in target immune cells enables the receptor-dependent uptake of numerous radiotracers, such as radioiodine ([ 123 [238,239]. It was also possible to synthesize a complex of different guanidines with 11 C as an indirect PET radiotracer [240]. A recently developed group of reporters used for indirect immune cell labeling are genetically introduced radiotracer-detectable surface tags. One such example is the DOTA antibody reporter 1 (DAbR1) that is a fusion of a single-chain fragment of a-Y-DOTA Ab 2D12.

Tracking of indirectly labeled immune cells with SPECT imaging
As described earlier, indirect cell labeling requires genetic modification of the cells for the expression of reporter genes encoding enzymes, receptors, or importers, that induce accumulation of the radiolabeled probe inside the cell. Unlike direct labeling methods, this strategy enables longitudinal immune cell tracking following cell administration. Various reporters, including hNIS, hNET, and SSTR2 have been tested for adoptive cell therapy monitoring with SPECT imaging. In a preclinical study by Emami-Shahri et al., the PSMA-specific P28f CAR-T cells were genetically modified to coexpress hNIS, and then administered to mice with PSMAexpressing tumors [221]. Following the adoptive transfer, repetitive imaging with SPECT was employed to track [ 99m TcO 4 ] uptake via the hNIS transporter in the CAR-T cells for up to 21 days. Tumor infiltration by T cells shown with SPECT imaging was confirmed by immunohistochemistry analysis of tumor tissue. This study showed that adoptive T cells expressing hNIS reporter gene could be tracked in vivo longitudinally with nuclear imaging, enabling analysis of the correlation between tumor infiltration by CAR-T cells and tumor growth. In vivo monitoring of Tregs seems to be of great importance for checking the biodistribution of these immune suppressive cells in patients with autoimmune disorders or after organ transplantation. To track the migration of Tregs in vivo, autologous Tregs were expanded under ex vivo conditions and then genetically modified with the hNIS reporter system for the [ 99m TcO 4 ] radiotracer detection by SPECT/CT imaging [241]. Imaging analysis revealed increased infiltration of the spleen by Tregs as compared with other organs. This preclinical study demonstrated that adoptively transferred Tregs can be tracked in vivo in a time-dependent manner with SPECT/CT imaging, as this approach does not affect Treg function and viability. Future research will show whether this approach may be exploited in the context of inflammatory or autoimmune diseases.
hNET, which binds and takes up clinically approved radiotracers such as MIBG or MFBG, is another reporter gene option for indirect immune cell tracking with SPECT. In a study by Doubrovin et al. human EBV-specific T lymphocytes were transduced to express hNET, and then injected intratumorally into EBV + tumors. After intravenous administration of radiotracers specific for the hNET reporter probe ([ 123 I]MIBG for SPECT or [ 124 I]MIBG for PET), as little as 10 4 hNET-expressing T cells injected intratumorally were detected by SPECT or PET imaging. For longitudinal studies, hNET-expressing EBV-specific T cells were administered intravenously and then tracked for accumulation in EBV + tumors for up to 28 days with SPECT or PET imaging. This study showed that the hNET reporter system is safe and non-immunogenic, and thus potentially may be used for longitudinal in vivo imaging of genetically modified immune cells in humans [238]. In another study, Moroz et al. directly compared various reporter systems for in vivo T cell detection. To this end, mice were injected subcutaneously with various numbers of T cells transduced with one of the following reporter genes hNET, hNIS, hdCKDM, and HSV-tk, followed by SPECT or PET detection using various radiotracers [242]. SPECT imaging with hNET and [ 123 I]MIBG tracer was the least effective as it required more than 10 7 T cells to generate a signal. The highest sensitivity, allowing for detection of only 40 000 injected cells, was observed for hNET and [ 18 F]MFBG radiotracer among various reporter gene-radiotracer combinations for PET that were evaluated.
Analogous to the reporter genes mentioned previously, efforts have been made to transduce immune cells for the expression of a surface tag-detected with a radiotracer for nuclear imaging. One such example is DAbR1 marker, which is made up of a single-chain fragment of the anti-lanthanoid-DOTA 2D12.5/G54C antibody linked to the CD4-transmembrane domain, which attaches covalently to AABD tracer. This tag has been recently successfully expressed in CAR-T cells and primary human T cells [168], and these cells not only showed efficient radiotracer uptake in vitro but also, could be tracked with AABD tracer labeled with [ 86 Y] (PET) or [ 177 Lu] (SPECT) upon intravenous administration to mice. Nuclear imaging analysis showing the highest radiotracer activity in the areas infiltrated by CAR-DAbR10-modified T cells was additionally corroborated with autoradiography and CD3 IHC staining.

Tracking of indirectly labeled immune cells with PET imaging
PET imaging has been successfully employed to track indirectly labeled immune cells with high sensitivity both in preclinical and clinical settings. Enzyme-mediated substrate alteration is commonly used for specific radiotracer entrapment in immune cells expressing enzymatic reporter genes. Up till now, various reporter enzymes and their substrates have been investigated in the context of in vivo immune cell tracking. The most extensively evaluated PET reporter gene, as mentioned in a previous paragraph, is HSV1-tk with its product HSV1-TK. Substrates of this enzyme are pyrimidine nucleoside analogs (e.g., FIAU: 5-iodo-2-fluoro-2deoxy-1-D-arabino-furanosyluracil; FEAU: 2-fluoro-2-deoxyarabi nofuranosyl-5-ethyluracil), or acycloguanosine derivatives (e.g., FPCV: fluoropenciclovir; FHBG: 9-[4-fluoro-3-(hydroxymethyl) butyl] guanine), which upon labeling with 18 [229]. Importantly, exposure of transduced T cells to the radiotracer did not affect their cytolytic activity. In another study, PET imaging of T cells upon [ 18 F]FHBG radiotracer administration allowed quantification of their homing during inflammation [244]. Clinical use of this technology is reported in the study by Yaghoubi et al. [230,245]. Using PET, Keu et al. studied glioblastoma homing of CD8 + CTLs expressing CAR IL-13, and HSV1-TK reporter enzyme that facilitated [ 18 F]FHBG uptake. The analysis of [ 18 F] FHBG signal on PET images performed before and after CLTinjection was very useful to show CTLs trafficking, survival, and proliferation [231].
The viral origin of the HSV1-tk reporter gene is its major limitation, as it can evoke potential immune reactions against cells carrying this gene. This limitation has been overcome by the generation of different mutants of human dCK, which selectively phosphorylate fluorinated thymidine analogs. Human prostatespecific membrane antigen (PSMA)-specific CAR-T cells have been successfully transduced with the human dCK double mutant (dCKDM) reporter gene before transplantation and then visualized in PSMA-positive prostate metastases in the lungs with [ 18 F]FEAU radiotracer and PET imaging within 6 h after T cell administration [233]. In another study, tumor infiltration by T cells co-expressing the triple mutant hdCK3mut with a melanoma antigen-specific T cell receptor was studied with [ 18 F]L-FMAU (1-(2-deoxy-2-fluorob-l-arabinofuranosyl) thymidine) radiotracer and PET imaging [246]. PET analysis showed that human leukocyte antigen (HLA)matched tumors had higher T cell-associated radioisotope signal as compared with HLA-mismatched tumors. Importantly, the expression of various mutants of human dCK did not evoke any changes in behavior and viability of transduced cells, supporting the utility of monitoring adoptive cell transplantation with human dCK mutants and PET reporter imaging.
Selective radiotracer import via importers encoded by reporter genes such as hNIS, hNET, and SSTR2 is another common approach for tracking live immune cells with PET imaging. Ahn et al. studied trafficking of bone marrow-derived DC (BMDC) transduced with luciferase/hNIS reporters for bioluminescence imaging (BLI) and PET imaging using [ 124 I] as a radiotracer, respectively [247]. After 7 days of BMDC injection into a footpad, both imaging modalities showed enhanced signal in the draining lymph nodes, suggesting DC trafficking. In another study, the lymph node homing of the murine DC line DC2.4, transduced with the same luciferase/hNIS reporter system, was successfully visualized using BLI and PET/CT imaging with another probe for hNIS reporter gene, [ 18 F]tetrafluoroborate ([ 18 F]TFB), in live mice [237], supporting the use of hNIS/[ 18 F]TFB reporter system in conjunction with PET imaging for monitoring immune cells in live animals. This approach has also found application in autologous human CAR-T cells tracking. Volpe et al. developed a novel platform to induce co-expression of both CAR and hNIS-RFP -a fused radionuclidefluorescence reporter gene. The introduction of this unique set of genes induces human T lymphocytes to express CAR, red fluorescent protein, and transporter for [ 18  It appeared that CAR-T targeting to tumors was inversely correlated with immune checkpoint expression in triple negative breast cancer (TNBC) models [248] SSTR2 is another human reporter gene that has been successfully employed to monitor in a quantitative and time-dependent manner the biodistribution and antitumor effects of CAR-T cells using the [ 68 Ga]-labeled somatostatin analog ([ 68 Ga]DOTATOC) and PET reporter imaging in a preclinical mouse tumor model [235]. This study showed that low numbers of CAR-T cells could be visualized with high sensitivity and specificity, and [ 68 Ga]DOTA-TOC uptake in the tumor tissue correlated with tumor development. This PET analysis was successfully validated by the immunohistochemistry analysis of tumor tissues showing the correlation between the uptake of [ 68 Ga]DOTATOC and percent of CAR-T cell homing into the tumor.
A very novel approach to indirect cell labeling was implemented by Minn et al. Taking advantage of the fact that PSMA has limited expression in human tissue almost only to the prostate, they created transgenic human blood-derived CD19-tPSMA (N9del) CAR-T cells. Prostate-specific membrane antigen (PMSA) expressed on the cell surface of those cells was exploited as a target for [ 18 F] DCFPyL, which acted as an indirect radiolabel. The experimental model was NSG mice with S.C. inoculated Nalm6-eGFP-fLuc cell tumors. On the 12th-day post CAR-T cells I.V. injection, a clear PET signal at the tumor side was detected, which indicated that most of the injected cells infiltrated tumor tissue [249].
The aforementioned labeling methods are powerful tools for monitoring immune cells in vivo, although they all have both advantages and disadvantages. The application of genetically modified cells in patients can lead to severe regulatory complications and may increase the cost of diagnostics. On the other hand, it is easier to perform direct labeling from a technical standpointthe method does not require an in vitro labeling step -instead, the radiolabel can be administered to the patient intravenously. Furthermore, the possibility of continuous monitoring of labeled immune cells over their entire lifetime is undoubtedly of great value. Hence, it is of great importance to appropriately match the cell labeling strategy with the purpose of the immune cell tracking study. Importantly, to date, there are no reports that would indicate impaired immune cell function caused by any of these approaches.

Cell tracking in clinical practice
Implementation of cell tracking methods in clinical practice still needs to face many regulatory and practical issues. Based on the preclinical data, indirect cell labeling has many advantages over the direct method. The possibility of longer cell tracking, the ability to monitor cell proliferation, and fate are the most important advantages. However, several issues need to be solved.
First of all, the range of available radioisotopes is limited. The development of reporter genes expressed in tracked cells that require already clinically approved and nontoxic radioisotope is an ongoing fast and efficient solution. Otherwise, the search for new labels will need to undergo the whole regulatory path.
Another aspect concerns the genetic engineering of the cells. The development of CAR-T technology gives the possibility to use gene-editing approaches in cell labeling for clinical use. Of course, the use of new vectors needs to face many regulatory issues, increasing cost and logistics of preparation of the cell-based product. However, one solution for that might be the co-delivery of the reporter gene in the same vector as the therapeutic gene [226].
The next aspect that cell labeling methods will face is the need for method tailoring. Different cells and different labels should be employed depending on the type of disease and its location. Different diseases attract different leukocyte populations. This is particularly true in oncology, as cancers are located at different sites, have different features (e.g. 'cold' vs. 'hot' tumors), and different neighboring tissues that may limit the selection of the proper radioisotope.
The unwanted accumulation of radioactivity in normal immune organs and tissues is also related to the previous comments as isotope carrying cells may certainly be endowed with preferred tropism towards selected tissues, independent of the tumor envi-ronments. This aspect is of particular concern in the case of cells carrying long half-life radionuclides.
A specific issue that must also be addressed concerns the stability of labeled cells. In the case of direct methods, the chemical stability of the isotope chelator moiety towards intralysosomal enzymes is an important experimental variable to be considered. In the case of indirect labeling, also the half life of the reporter gene product must be assessed with certainty. In fact, most reporter genes, even when fused with long lived protein moieties, may be targeted by the ubiquitin system and degraded at the proteasome intracellular sites. In both cases (direct and indirect), the degradation of the carrier inside the cell will lead ultimately to the leakage of the isotope outside the cell. As the last issue, the viability of the whole cell itself must be considered given possible intrinsic toxicity of the isotope label at a high concentration within the tracked cell.
Last but not least, before the cell labeling will be a routinely implemented method into clinical practice, the precise biological mechanisms of cell labeling and its effect on the cells and the whole organisms need to be studied [250].
Therefore, from a regulatory and technical point of view, direct cell labeling methods appear to be the most convenient ones in terms of costs and time and also the safest ones in clinical routine, especially when used for short-term cell tracking for diagnostic purposes.

Summary
In recent years nuclear imaging methods have been established for noninvasive and real-time monitoring of the long-term distribution and viability of adoptively transferred immune cells, including T cells, B cells, DCs, macrophages, and NK cells, both in preclinical and clinical settings. The most important studies, which were cited in this manuscript, are listed and summarized in Supplementary Table 1. Tracking of immune cells with nuclear imaging has been used for a long time in routine medical practice for diagnostic purposes in infections and inflammatory processes. Radioisotope-based imaging provides numerous advantages in basic research and clinical practice as compared with other molecular imaging modalities. Nuclear imaging modalities are exceptionally sensitive allowing for the use of sub-pharmacological amounts of radiotracers for cell labeling, that will not evoke any biological reactions. Furthermore, these imaging modalities provide quantitative analysis, e.g. in vivo concentration of the radiotracer in the tissue region of interest. Besides, the development of multimodal cameras (PET/CT, SPECT/CT, PET/MRI) provides detailed molecular information at both functional and anatomical levels, which overcomes the major shortcoming of nuclear imaging that is the absence of anatomical information. Finally, a great number of novel radiotracers for direct and indirect labeling strategies are being developed, and the growing availability of radiopharmaceuticals and imaging cameras will eventually broaden our understanding of immunological processes in living organisms. The summary of the currently used radiotracers in cell tracking studies containing valuable information concerning their dosage, normal systemic distributions and safety is provided in Table 6.
Immune cell labeling for nuclear imaging, which is essential for live-cell tracking, may be accomplished with direct or indirect labeling techniques. Direct labeling of immune cells, which is performed under in vitro conditions with radiotracers before cell transplantation, gives specific images of low cell numbers with no need for genetic manipulation. Hence, immune cell labeling with direct methods using radiotracers approved for clinical use may be potentially translated into the clinic. Indirect labeling, which involves the introduction of a reporter gene (e.g., HSV-TK, hNIS, hNET, and SSTR2) into host cells provides the highest sensitivity and can be employed for longitudinal studies of immune cell Table 6 The summary of the radiotracers used for immune cell tracking and data concerning their dosages, safety and systemic distribution.

Radiotracer
Cell type + radioactivity of dosage (administered in vitro in direct and in vivo in indirect cell labeling) with a total radioactivity of 2 lCi/10 6 T cells [75] cd T cells labeled with the standard nuclear pharmacy 111 In-oxine method at 5 pCi/cell [60] allogeneic NK cells [62] umbilical cord blood hematopoietic progenitor cells (UCB-NK cells) 2 MBq of [ 111 In]oxinate was added per 10 6 cells [64] monocyte-derived nonmatured DC (nmDCs) or matured DC (mDCs) -from 0,7 MBq to 20 MBq [82] it is reported that 111 In is secreted in human milk following administration [251] sensitivity reactions (urticaria) have been reported [251] recent studies have found no evidence of carcinogenicity in either rats or mice given oxyquinoline in feed at concentrations of 1500 or 3000 ppm for 103 weeks [251] it has been denoted, that human lymphocytes labeled with recommended concentrations of [ 111 In]oxine showed chromosome aberrations consisting of gaps, breaks and exchanges that appear to be radiation induced [252] administration of [ 111 In]oxine is followed by excessive accumulation of radioactivity in kidneys [253] [ 111 In]oxine bioaccumulation is observed mainly in liver, spleen, kidneys, bone marrow, stomach, intestine, lungs and muscles [254] [ 111  non-specific activity is oftentimes seen in kidneys, thus causing false-positive results [49] a very few cases of mild hypersensitivity evidenced by the development of an urticarial erythematous rash have been reported following direct intravenous injection; no serious adverse effects on animals or patients have been denoted [257] uptake in the brain reaches a maximum of 3.5-7.0% of the injected dose within one minute of injection; about 20% of the injected dose is removed by the liver immediately after injection and excreted through the hepatobiliary system; about 40% of the injected dose is excreted through the kidneys and urine over the 48 h after injection resulting in a reduction in general muscle and soft tissue background [257] endocytosis [ 99m Tc]SnF 2 human blood leukocytes (585 MBq, labeling efficiency 36% -99,7%) [89] no adverse effects on cells nor patients have been denoted high uptake in the liver of healthy rats (85.7% of injected dosage), lower in lungs (3.1%) and spleen (7.6%) [258] PET radiotracers for direct cell labeling through the mechanism: transporter uptake [ 18 [36] no adverse effects on animals or patients have been denoted however, it has been known, that the precorson -deoxyuridine exhibits dose dependent antifolate toxicity leading to perturbation of chromatine [149] in humans, marked accumulation of radioisotopes in the thyroid, salivary glands, intenstines, stomach, esophagus and along with elevated uptake in the bone marrow and liver and overall moderate enhancement in the mediastinum and throughout the abdominal cavity [150] (continued on next page)   [267] no toxicity from the diagnostic dose reported clearance through urinary excretion (clearance from blood was completed in 24 h); high radioactivity accumulation observed in kidneys and blood in the early phase (1, 4 h); the low radiotracer uptake in non-target tissue; the high tumour/blood ratios enables tumor imaging [267] [268] no clinically detectable pharmacologic effects caused by the administration of the diagnostic dose [268] clearance through urinary excretion; the highest radiotracer uptake in the thyroid, salivary glands, stomach, and urinary bladder [268] [ 123 I]MIBG intravenous injection of 400 MBq (10.8 mCi) of [ 123 I]MIBG in man [269] no clinically detectable pharmacologic effects caused by the administration of the diagnostic dose clearance through urinary excretion (~50% of the administered radioactivity appears in the urine within 24 h and 70-90% within 48 h after administration); the tracer accumulates in the liver, salivary glands, nasal mucosa, heart, lungs and bowel; high in vivo stability (metabolites that account for <10% of the injected dose are m-iodohippuric acid, m-iodobenzoic acid, and 4-hydroxy-3-iodobenzylguanidine and free radioiodide); accumulation of unbound 123 [274,275] clearance through urinary excretion [274,275] organs with the highest radiation exposure are kidney, liver, spleen, and urinary bladder (doses to most organs ranged from 0.11 to 0.76 mGy/MBq); the effective dose À 0.16 to 0.20 mSv/MBq [274] an exponential washout from the different organs during several hours followed by a late peak (>15 h) in the bladder no penetration across the blood-brain barrier (however some accumulation observed within intact glioma tumor, probably due to a compromised BBB [273] ? possible detection of HSV1-tk or hmTK2 expressing immune cells within these tumors)  [230,276] the phase I study showed the safety of, and lack of toxicity of intravenously injected [ 18 F]FHBG tracer for imaging purposes [276] rapid blood clearance (8.42% ± 4.76% of the peak blood activity remained at 30 min following injection); primary routes of clearance are renal and hepatobiliary (high activities observed in the bladder, gut, liver, and kidney ? possibly not suitable for imaging HSV1-tk gene expression in the lower abdomen); bladder absorbs the highest radiation dose; rapid clearance from all other tissues ? low background signal; high in vivo stability (in the urine, 83% of activity 180 min following administration was stable [ 18 F]FHBG); no penetration across the blood-brain barrier (however some accumulation observed within intact glioma tumors or tumor resection sites, probably due to a compromised BBB [230] ? possible detection of HSV1-tk expressing immune cells within these tumors) (continued on next page) Ł. Kiraga, P. Kucharzewska, S. Paisey et al. Coordination Chemistry Reviews 445 (2021) 214008  [277] no safety data in man both renal and hepato-biliary clearance; high activities observed in large intestine, gallbladder, small intestine and kidney [277] [ 18 F]L-FMAU bolus intravenous injection of~56 MBq (1.5 mCi) sterile l-[ 18 F]L-FMAU in man [278] no safety data in man biliary excretion of the radiotracer; in man high signals observed in liver, kidneys, gall bladder, bladder, the GI tract and myocardium ? the accumulation of the radiotracer in liver and myocardium may compromise its use for PET imaging of therapeutic cells at these sites [278] transporter reporter gene [ 18 F]TFB intravenous injection of [ 18 F]TFB (333-407 MBq) in healthy volunteers [279] safe and non-toxic probe -no immediate or delayed adverse reactions were observed in volunteers after the radiotracer administration; the estimated absorbed radiation doses are acceptable for clinical imaging purposes [279] a multi-phasic blood clearance of the radiotracer (two rapid clearance phases during the first 20 min, followed by a slower clearance phase); clearance through urinary excretion high uptakes observed in thyroid, stomach, salivary glands, and bladder; the highest radiation doses were observed in the thyroid, urinary bladder wall, intestine wall, heart wall, kidneys, liver, pancreas and spleen; high in vivo stability of the radiotracer (minor 18 F-labeled metabolites in the blood and urine during 4 h analysis) [ [280] safe and non-toxic probe [280] high uptake in heart and bladder before the 2 h time point; thyroid uptake increases after the 24 h time point, whereas uptake in liver, kidneys, and lungs decreases with time; the highest mean equivalent dose is observed in the thyroid; the estimated mean humanequivalent effective dose À 0. 25 [281] first-in-human study showed the safety of, and lack of toxicity of intravenously injected [ 18 F]MFBG tracer for imaging purposes (no side effects seen in any patients after [ 18 F]MFBG injection); the estimated absorbed radiation doses are acceptable for clinical imaging purposes [281] rapid and biexponential blood clearance (mean biologic T 1/2 of 18 min for rapid phase and 6 h for slower phase); monoexponential whole-body clearance, with a mean biologic T 1/2 of 1.95 h, enabling early imaging at 1 -2 h after injection; clearance through urinary excretion (45% of the administered activity in the bladder by 1 h after injection) high activity in blood, liver, and salivary glands, and mild uptake in kidneys and spleen, that decreases with time; urinary bladder receives the highest radiation dose; the effective dose À 0.023 ± 0.012 mSv/MBq [281] receptor reporter gene  [283] safe and non-toxic probe -no immediate or delayed adverse reactions were observed in volunteers during the one week follow up after the radiotracer administration; the estimated absorbed radiation doses are acceptable for clinical imaging purposes [283] clearance through urinary excretion (at 4 h after injection 18.8 ± 1.0% ID of the injected activity was eliminated in the urine) [283] the highest uptake at 1, 2, and 3 h after administration observed in spleen, kidneys and liver; the highest absorbed organ doses observed in urinary bladder, followed by spleen, kidneys, adrenals and liver [282] the effective dose from 2.1 [282] to 4.8 mSv [283] from 100 MBq injected activity distribution, activation, proliferation, and survival. Although indirect labeling of immune cells for reporter gene-based tracking in the clinical settings may be considered for adoptive cell therapies, such as CAR-T cells, that are already genetically modified, the genetic modification of primary immune cells for in vivo tracking purposes in humans probably will not be routinely used in the clinical practice. Limited use of reporter genes in man is caused by the fact that genetic modification of the target cell genome with viruses can lead to harmful effects owing to uncontrolled insertional mutagenesis. Genome editing methods, allowing for the reporter gene integration to precise genomic loci, may potentially circumvent this limitation and increase the safety of reporter gene imaging. Currently, available cell labeling techniques do not meet all expectations as they all have advantages and disadvantages. Key issues concerning the fate and distribution of labeled cells to unwanted sites in the body and consequent accumulation of radioactivity or undesirable leakage of radiotracer due to tracked cell premature degradation or simple biochemical degradation of the chelator molecule must be addressed. Consequently, a suitable strategy for immune cell labeling needs to be chosen for each experimental setup.
Despite the above-mentioned hurdles, nuclear imaging-based in vivo tracking of immune cells is a valuable technique that may provide us with a noninvasive way of studying the functioning of the immune system and contribute to the development of therapies using or targeting immune cells.
The room for improvement, besides the current biological and physical limits of the PET scanners, are proper approaches for data acquisition and algorithms for data analysis. The recent interesting study [152] showed that a very short acquisition timeframe and a reconstruction algorithm used for tracking single-cell radioactivity required a limited number of photon-coincidence detections and a spline method to track the cell. Importantly, the authors were able to track a single cell after i.v. injection. It was shown that a proper algorithm for a digital PET scanner with sensitive and excellent time-of-flight properties would limit the necessary radioactivity of the cell for its localization [286]. The cell tracking study utilizing PET, recently conducted by Sato et al. on the rhesus monkeys, was fully successful and importantly, no side effects were observed. That would give a potential clinical translation of this excellent preclinical method [135].
However, before the widely implementation of cell tracking methods in clinical practice will be possible, this method needs to face many regulatory and practical issues. The most important issues to solve seems to be the selection of the proper radioiso-topes, genetic engineering of the cells, tailoring of cell-labeling methods to particular disease and site, as well as validation of the safety of such approaches.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.  [285] safe and non-toxic probe -no immediate or delayed adverse reactions were observed in volunteers after the radiotracer administration [285] rapid blood clearance [285] clearance through urinary excretion [284,285] high in vivo stability [285] the highest activities were observed in the kidneys and bladder, followed by the salivary glands, liver, spleen and proximal small bowel [284] the highest radiation dose observed in kidneys (0.0945 mGy/MBq), followed by urinary bladder wall (0.0864 mGy/MBq), submandibular glands (0.0387 mGy/ MBq), and liver (0.0380 mGy/MBq); no radiotracer uptake in brain; the effective radiation dose À 0.0139 mGy/MBq from an injected dose of 370 MBq (10 mCi) [285]