Novel 89Zr cell labeling approach for PET-based cell trafficking studies

Background With the recent growth of interest in cell-based therapies and radiolabeled cell products, there is a need to develop more robust cell labeling and imaging methods for in vivo tracking of living cells. This study describes evaluation of a novel cell labeling approach with the positron emission tomography (PET) isotope 89Zr (T1/2 = 78.4 h). 89Zr may allow PET imaging measurements for several weeks and take advantage of the high sensitivity of PET imaging. Methods A novel cell labeling agent, 89Zr-desferrioxamine-NCS (89Zr-DBN), was synthesized. Mouse-derived melanoma cells (mMCs), dendritic cells (mDCs), and human mesenchymal stem cells (hMSCs) were covalently labeled with 89Zr-DBN via the reaction between the NCS group on 89Zr-DBN and primary amine groups present on cell surface membrane protein. The stability of the label on the cell was tested by cell efflux studies for 7 days. The effect of labeling on cellular viability was tested by proliferation, trypan blue, and cytotoxicity/apoptosis assays. The stability of label was also studied in in vivo mouse models by serial PET scans and ex vivo biodistribution following intravenous and intramyocardial injection of 89Zr-labeled hMSCs. For comparison, imaging experiments were performed after intravenous injections of 89Zr hydrogen phosphate (89Zr(HPO4)2). Results The labeling agent, 89Zr-DBN, was prepared in 55% ± 5% decay-corrected radiochemical yield measured by silica gel iTLC. The cell labeling efficiency was 30% to 50% after 30 min labeling depending on cell type. Radioactivity concentrations of labeled cells of up to 0.5 MBq/106 cells were achieved without a negative effect on cellular viability. Cell efflux studies showed high stability of the radiolabel out to 7 days. Myocardially delivered 89Zr-labeled hMSCs showed retention in the myocardium, as well as redistribution to the lung, liver, and bone. Intravenously administered 89Zr-labeled hMSCs also distributed primarily to the lung, liver, and bone, whereas intravenous 89Zr(HPO4)2 distributed to the liver and bone with no activity in the lung. Thus, the in vivo stability of the radiolabel on the hMSCs was evidenced. Conclusions We have developed a robust, general, and biostable 89Zr-DBN-based cell labeling strategy with promise for wide applications of PET-based non-invasive in vivo cell trafficking.


Background
With the growth of interest in cell-based therapies, there is a need to develop more sensitive, robust, and quantitative imaging methods for in vivo tracking of living cells. A number of radioisotopic cell labeling methods have traditionally been used for single-photon emission computerized tomography (SPECT) and positron emission tomography (PET) imaging-based cell tracking [1]. However, a PET-based approach would offer superior quantification and imaging sensitivity characteristics over a SPECT-based approach, which are critical for tracking of small numbers of administered cells [1]. In this regard, 89 Zr has emerged as an attractive PET radionuclide for cell labeling applications due to its high spatial resolution and 78.4-h half-life that may allow monitoring of administered cells up to a 2-to 3-week period.
However, this radiotracer yielded poor in vivo imaging characteristics, possibly due to insufficient CD45 molecules on the plasma membrane of stem cells [8].
In this study, we propose a novel cell labeling strategy that covalently binds a 89 Zr-DFO-labeled agent to cell surface proteins independent of cell type. The novel method employs the two-step process ( Figure 1): 1) preparation of 89 Zr-labeled p-isothiocyanato-benzyldesferrioxamine ( 89 Zr-DBN) and 2) random labeling of primary amines of cell surface proteins with 89 Zr-DBN. We have evaluated this labeling strategy in three cell types: mouse melanoma cells (mMCs), human mesenchymal stem cells (hMSCs), and mouse dendritic cells (mDCs). The labeled cells were evaluated for 7 days post-labeling for label retention and probable changes in cell proliferation, cell viability, and degree of apoptosis in radiolabeled cells as compared to their

Incorporation of 89 Zr-DBN in protein fraction
To understand the subcellular localization of the label, incorporation of 89 Zr-DBN into different protein fractions in mMCs, hMSCs, and mDCs was evaluated using a subcellular protein fractionation kit (Piercenet Thermo Scientific, Waltham, MA, USA) at days 1, 4, and 7 postlabeling. The cytosolic proteins, hydrophobic membrane proteins, nuclear proteins, and cytoskeletal proteins were isolated, and each protein fraction was counted for radioactivity using a 2480 Wizard 2 automatic gamma counter (PerkinElmer, Waltham, MA, USA).

Efflux of 89 Zr-DBN from labeled cells
To determine cellular efflux, 0.3 × 10 6 89 Zr-labeled cells were plated into each well of a six-well culture plate.
The medium was replaced with fresh medium daily for 7 days, and radioactivity in the replaced medium was counted. For mDCs with mix of adherent and suspension cells, the plate was centrifuged at 1,000 rpm for 10 min before replacing the medium to avoid loss of unattached 89 Zr-labeled cells.

CyQUANT cellular proliferation assay
The effect of radiolabeling on cellular proliferation was assessed by the CyQUANT DNA content assay (Thermo Fisher, Waltham, MA, USA). A known number of unlabeled and 89 Zr-labeled cells (approximately 10 4 cells/ well) were plated in 21 wells of a 96-well culture plate and maintained at 37°C in a CO 2 incubator. The amount of DNA in each well was quantified from absorbance values as a surrogate marker of the number of cells present. The culture medium was replaced daily. The CyQUANT assay was performed for three wells per day over 5 days.

Trypan blue exclusion assay cellular viability test
The effect of labeling on cellular viability was assessed using trypan blue exclusion assay test within 1 h of labeling, third and seventh day post-labeling. The culture medium was replaced daily and maintained at 37°C in a CO 2 incubator. Unlabeled cells served as control.

ApoTox-Glo viability/cytotoxicity and apoptosis assay
The effect of radiolabeling on cellular viability was also assessed using the ApoTox-Glo viability, cytotoxicity, and caspase 3/7 apoptosis assay (Promega Corporation, Madison, WI, USA). Unlabeled cells served as control. A known number of unlabeled and 89 Zr-labeled cells (approximately 10 4 /well) were plated in a 96-well culture plate. The culture medium was replaced daily and maintained at 37°C in a CO 2 incubator. At day 7, cell viability, cytotoxicity, and apoptosis were quantified in triplicate using the ApoTox-Glo assay. As positive controls, cells were incubated with 30 μg/mL digitonin for 30 min for the viability and cytotoxicity assays, while 2 μM staurosporine was added for 16 h for the caspase 3/7 dependent apoptosis assay.

PET imaging and ex-vivo biodistribution of 89 Zr-labeled cells and 89 Zr(HPO 4 ) 2
Experiments were performed with 2-month-old athymic nude Foxn1 nu mice (Harlan Laboratories, Inc., Indianapolis, IN, USA). 89 Zr(HPO 4 ) 2 (approximately 0.074 MBq) or 89 Zrlabeled cells (2 × 10 5 cells with radioactivity concentration approximately 0.37 MBq/1 × 10 6 cells) were injected intravenously through a tail vein. On days 2, 4, and 7, the mice were anesthetized under 1% to 2% isoflurane and underwent PET imaging using a small animal PET/X-RAY system (Sofie BioSystems Genesys4, Culver City, CA, USA). At day 7, the mice were sacrificed and tissues were extracted and radioactivity counted using a gamma counter to evaluate the biodistribution of 89 Zr radioactivity. PET images were normalized to units of standardized uptake value (SUV) = (activity concentration in tissue / (injected dose/g whole body wt.)) and presented as a coronal sectional images.
In vivo tracking of stem cell engraftment in ischemia/ reperfusion mouse model Athymic nude Foxn1 nu mice (2 months old) were anesthetized under 1% to 2% isoflurane and placed on a heating pad maintained at 37°C. Respiratory and heart rates were monitoring continually. After intubation, mechanical ventilation and intercostal block of bupivacaine and lidocaine, an incision was made in through the fourth or fifth intercostal space for access into the thoracic space, the heart was exposed and the pericardium was incised anterior and parallel to the phrenic nerve. With visualization of the coronary vasculature, the left coronary artery was ligated to induce myocardial ischemia at the anterior wall of the left ventricle. One hour after the coronary ligation, the suture was untied for reperfusion. Myocardial reperfusion was confirmed by color change of the left ventricle and electrocardiographic changes. During reperfusion, 89 Zr-labeled cells (2 × 10 5 cells with radioactivity concentration approximately 0.37 MBq/10 6 cells) were injected at four sites within the ischemic region. After myocardial injection, the intercostal space, the chest musculature, and the skin were closed with a 7-0 Ethilon suture. The animals were imaged at day 2, day 5, and day 7 using small animal PET/X-RAY system (Sofie BioSystems Genesys4, Culver City, CA, USA). At day 7, the mice were sacrificed and tissues were extracted and radioactivity counted using gamma counter to evaluate cell trafficking. PET images were normalized to units of SUV and presented as coronal sectional images.

Statistical analysis
The data were compared using unpaired Student's t-test analyses. Differences were regarded as statistically significant for p < 0.05.

Results
Synthesis of 89 Zr-DBN and cell labeling studies 89 Zr hydrogen phosphate was readily chelated by DFO-NCS to form 89 Zr-DBN, with radiolabeling efficiency of 55% ± 5% after 1 h of reaction. This reaction mixture was then used directly for labeling of cells. The cell labeling efficiency using 89 Zr-DBN was approximately 30% to 50% as determined by cell-bound radioactivity. Radioactivity concentrations of 0.50 ± 0.10, 0.47 ± 0.10, and 0.39 ± 0.20 MBq/10 6 cells were achieved when 6 × 10 6 cells were incubated for 30 min with approximately 6 MBq 89 Zr-DBN with mMCs, hMSCs, and mDCs, respectively. In contrast, no cell labeling was observed using 89 Zr(HPO 4 ) 2 .

Cellular proliferation and viability studies
The CyQUANT proliferation assay showed no difference in proliferation rate between unlabeled and 89 Zr-labeled cells ( Figure 2). Trypan blue cell viability tests were performed on radiolabeled cells immediately after labeling and up to 7 days post-labeling and compared with unlabeled cells. No change was observed in number of dead cells (blue-stained cells) over live cells (unstained cells) in both 89 Zr-labeled and unlabeled cells, with percentage of dead cells <5% in all days tested.

ApoTox-Glo viability/cytotoxicity and apoptosis assay
The ApoTox-Glo assay showed no loss in cellular viability and no increase in cytotoxicity or apoptosis in radiolabeled cells as illustrated in Figure 3. Viability was lost, and cytotoxicity enhanced when 30 μg/mL digitonin was added to cells (positive control), and apoptosis was increased with the addition of 2 μM staurosporine (positive control).

Subcellular distribution of 89 Zr radioactivity
At days 1, 4, and 7 after 89 Zr labeling of mMCs, hMSCs, and mDCs, subcellular protein fractionation of the cells was performed. 89 Zr radioactvity was incorporated predominantly (>99%) in hydrophobic membrane protein fraction of all cell types studied, strongly supporting the proposed mechanism of reaction of 89 Zr-DBN with cell surface membrane protein to form a stable covalent bond.

Efflux of 89 Zr radioactivity from labeled cells
Retention of 89 Zr radioactivity by 89 Zr-DBN-labeled cells was found to be stable in all the cells studied with negligible efflux observed over 7 days post-labeling ( Figure 4).

PET imaging and biodistribution studies in mice with intravenous injections
PET images and biodistribution data of intravenously administered 89 Zr-labeled hMSCs and 89 Zr(HPO 4 ) 2 in healthy mice are shown in Figure 5. The 89 Zr-labeled hMSCs were concentrated primarily in the lung and liver, followed by the bone. On the other hand, 89 Zr(HPO 4 ) 2 accumulated in the bone and liver and did not distribute to lung.

In vivo tracking of stem cell engraftment in ischemia/ reperfusion mouse model
Following myocardial delivery, 89 Zr-labeled hMSCs (approximately 19.5% ± 9.5%) were retained for 7 days in the heart (Figure 6). The remaining cells were concentrated in the lung, followed by the bones and liver. The higher uptake in the lung relative to the liver is consistent with the biodistribution of 89 Zr-labeled hMSCs released into the circulation ( Figure 5).

Discussion
Various strategies have been employed in the past to label cells with imaging isotopes for non-invasive in vivo cell tracking for cell-based therapies and infection imaging. Among them, 18 F-FDG (for PET) [12][13][14][15][16] and 111 In-oxine (for SPECT) [2][3][4][5][6], are the most widely used. Although 18 F-FDG is useful for assessment of immediate delivery of cells and early fate of cells (approximately first few hours), it is not suited for in vivo cell tracking after 24 h post-injection due to its short half-life and poor retention in cells. Inability of 18 F-FDG to allow cell tracking after 24 h limits its utility in cell-based therapies. For cell-based therapies, early engraftment period of 2 to 5 weeks post cell delivery is the most critical time period [20]. Therefore, imaging-based methods should be robust over this time frame to allow evaluation of various interventions for improving cell engraftment. The ability to monitor cells in vivo beyond 24 h is also of high importance for evaluation of infection using radiolabeled leukocytes. Conventional infection imaging protocols perform imaging at 1, 4, and 24 h post-injection to differentiate between inflammatory, acute infection, and chronic infection loci; however, in some patients, 48 h was necessary for reliable detection of infected lesions [21].  The use of 111 In (T 1/2 = 2.8 days) as a radiolabel for cell labeling allows longer observation periods for cell tracking but with lower spatial resolution of SPECT imaging. Cell labeling with 111 In typically requires a lipophilic carrier molecule (e.g., oxine) for transporting the radiometal into cells [2][3][4][5][6]. After entering the cells, the radiometal then dissociates and gets trapped in the cell by binding to non-specific intracellular metal-binding proteins. The two major disadvantages of this approach are chemotoxicity of the lipophilic carrier molecule [22] and efflux of radiolabel from cells [2][3][4][5][6].
Recently, two groups (Charoenphun et al. [9] and Davidson-Moncada et al. [10]) reported synthesis of 89 Zroxinate or 89 Zr-oxine as a cell labeling reagent for PETbased cell tracking. As expected, both groups faced the problem of chemotoxicity and significant efflux of radioactivity from the cells post-labeling commonly associated with oxine-based labeling. Charoenphun et al. [9] showed reduced viability of 89 Zr-oxine-labeled 5 T33 myeloma cells (from 93% to 76.3% ± 3.2% in the first 24 h) and significant efflux of radioactivity post-labeling (29% effluxed in 24 h). Davidson-Moncada et al. [10] also reported similar results with 89 Zr-oxine-labeled human and rhesus macaques' natural killer cells. They observed a broad range of viability of 60% to 100% in the radiolabeled cells over the first 24 h, which declined to 20% to 30% after 6 days. A significant efflux of radioactivity was also observed in these viable 89 Zroxine-labeled cells, approximately 20% to 25% effluxed in the first 24 h and 70% to 80% of radioactivity was effluxed after 7 days of culture. These drawbacks associated with the 89 Zr-oxine labeling method compromises its utility for PET-based monitoring of in vivo cell trafficking.
To improve the stability of the 89 Zr radiolabel on cells, we proposed 89 Zr-DFO-NCS ( 89 Zr-DBN) as a labeling entity capable of forming covalent bonds with primary amines of cell surface protein (Figure 1). Since all cells express cell surface protein with exposed lysine residues and other primary amines, this strategy also provides a general labeling method to label a broad array of cells. The new strategy exploits both the strength of chelation of 89 Zr by DFO with three hydroxamate groups (qualitative Zr-binding constant = approximately 10 31 ) [23][24][25][26] as well as the inherent biostability of the thiourea bond that conjugates NCS group in 89 Zr-DBN to primary amines of protein [27,28]. Furthermore, the labeling agent, 89 Zr-DBN, is also expected to be well-tolerated by cells as opposed to toxic lipophilic carrier molecules like oxine, relying on the fact that DFO-NCS has been routinely used to conjugate DFO to IgG and IgM antibodies with no loss of antibody protein function [23,[26][27][28][29]. The generality of the labeling target, along with the multiplicity of primary amines available on the cell surface, also avoids the specific targeting of highly sensitive processes that might affect cellular function or viability. In contrast to the previously noted 89 Zr-oxine results [9,10], no efflux of radiolabel was observed from cultured cells labeled with 89 Zr-DBN after repeated washing and culture with medium with 10% fetal bovine serum (FBS) out to 7 days. These data strongly argue for a covalent bonding of the radiolabel to the cells. Furthermore, in a subcellular fractionation study, essentially all 89 Zr radioactivity was incorporated into the membrane bound (See figure on previous page.) Figure 3 Assessment of (A) viability, (B) cytotoxicity, and (C) apoptosis in 89 Zr-labeled and unlabeled cells. No statistically significant differences were observed between 89 Zr-labeled and unlabeled cells after 7 days of culture with regard to viability, cytotoxicity, or apoptosis. As positive controls, 30 μg/mL digitonin was used for assays (A) and (B), and 2 μΜ staurosporine for (C). *p < 0.05 versus assessments in 89 Zr-labeled and unlabeled cells using unpaired t-test. Values are shown as mean ± standard deviation, n = 3. protein fraction of the cells confirming the anticipated targeting of membrane protein. The targeting of cell-surface membrane protein also has the potential benefit of distancing the labeling agent from potentially sensitive sites in the cell. We found no evidence of chemotoxicity or radiotoxicity effects of 89 Zr-DBN labeling of three cell types in the present study.
In this study, 30% to 50% labeling efficiencies were achieved with 89 Zr-DBN in several cell types. The cell labeling yield for mMCs, hMSCs, and mDCs were 0.50 ± 0.10, 0.47 ± 0.10, and 0.39 ± 0.20 MBq/10 6 cells for mMCs, hMSCs, and mDCs, respectively. This is the maximum load of radioactivity per 10 6 cells that we could achieve in this study. In the clinical setting, approximately 10 8 cells  Zr-labeled human MSCs (2 × 10 5 cells with radioactivity concentration approximately 0.37 MBq/10 6 cells) and 89 Zr(HPO 4 ) 2 (approximately 0.074 MBq radioactivity) were intravenously injected in athymic mice. Most of the radioactivity was distributed in the lung, liver, and bones following injection of 89 Zr-labeled hMSCs whereas most of the radioactivity was distributed in the liver and bones following injection of 89 Zr (HPO 4 ) 2 . Values in graphs are shown as mean ± standard deviation, n = 3.

Figure 6
Representative PET images and biodistribution data of 89 Zr-labeled hMSCs following myocardial delivery. 89 Zr-labeled hMSCs (2 × 10 5 cells with radioactivity concentration approximately 0.37 MBq/10 6 cells) were delivered to myocardium of an ischemia/reperfusion mouse model. Most of the radioactivity was distributed in the heart (arrow), lung, liver, and bones following myocardial delivery of 89 Zr-labeled hMSCs. Values in graph are shown as mean ± standard deviation, n = 5.
are typically delivered to patients. Based on the labeling yield obtained in this study, the amount of 89 Zr radioactivity administered to a patient would be in the range 30 to 50 MBq (0.8 to 1.4 mCi), which is in the range of 89 Zr radioactivity that is currently being used in patients with 89 Zr-labeled antibodies. After encouraging in vitro validation tests, we performed in vivo validations by investigating the biodistribution of 89 Zr-labeled hMSCs after intravenous injection in athymic nude mice for 7 days post-injection. Trapping of MSCs in the lungs following intravenous injection is well documented [30,31]. Therefore, we expected major accumulation of 89 Zrlabeled human MSCs in mouse lungs following intravenous injection with slow clearance. With time, cells were expected to dislodge from this physical pulmonary entrapment and distribute to other organs. As expected, the majority of intravenously injected 89 Zr-labeled hMSCs were trapped in the lung (50% ± 27%) and the remainder was found in the liver (27% ± 19%) and bones (16% ± 5%) after 7 days postinjection, which are expected homing sites for injected mesenchymal stem cells after dislodging from the lung [32,33]. This was in contrast to the biodistribution of 89 Zr(HPO 4 ) 2 , which distributed primarily in the bone (59% ± 13%) and liver (32% ± 15%) but did not accumulate in lungs. The distinct biodistributions of 89 Zr-labeled hMSCs and 89 Zr (HPO 4 ) 2 , together with the stability of radiolabel and lack of cytotoxicity, strongly support the robustness of the 89 Zr-DBN-based cell labeling approach.
To further test the application of 89 Zr-labeled hMSCs, we performed a stem cell engraftment study using a myocardial acute ischemia/reperfusion mouse model. 89 Zr-labeled hMSCs were delivered to the myocardium of athymic mice following an acute myocardial ischemia/reperfusion insult. After 7 days post-delivery, 89 Zr-labeled hMSCs were found in the heart (20% ± 7%), lung (40% ± 16%), bone (29% ± 11%), and liver (7% ± 5%). The observed retention in the heart is in accordance with previously published work on hMSC engraftment estimated by invasive quantitative PCR method in a similar rodent model [34].
Low levels of in vitro demetalation of 89 Zr-DFO complexes (2% to 3%/week) in serum at 37°C have been reported [23,29], and clinical studies using 89 Zr-DFOlabeled antibodies out to 7 days have yet to show significant bone uptake of 89 Zr indicative of demetalation [35][36][37][38][39]. Our initial findings of distribution of 89 Zr-labeled hMSCs in mouse models confirm the biostability of the radiolabel bound to the DFO moiety supporting further exploration of the 89 Zr-DBN labeling method for monitoring stem cell engraftment and cell trafficking. Extension of this approach with the use of an alternative zirconium chelator, such as 3,4,3-[LI-1,2-HOPO] [40] may further improve the biostability of the labeling agent.

Conclusions
The 89 Zr-DBN labeling agent is shown to be a robust, general, and biostable cell labeling strategy for PET-based noninvasive in vivo cell tracking. To explore the full potential of this approach, more work is needed to test this strategy in various model systems and disease processes that are germane to cell trafficking and stem cell therapies. We have ongoing efforts to define the imaging sensitivity, biostability, and toxicity parameters as the limits are pushed toward higher 89 Zr radioactivity loading of cells and longer observation periods in vivo.
Compliance with ethical standards 1. Disclosure of potential conflicts of interest: no authors have affiliations that present financial or non-financial competing interests for this work. 2. Research involving human participants and/or animals: -All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The patient-derived mesenchymal stem cells were obtained in compliance with institutional ethical review board guidance. -All applicable international, national, and/or institutional guidelines for the care and use of animals were followed. All procedures performed in studies involving animals were under approval and in accordance with the Ethical Standards of Mayo Clinic Institutional Animal Care and Use Committee. 3. Informed consent: informed consent was obtained from all individual participants included in the study.