## Introduction

Safety and efficacy are the two main pillars of any therapeutics and cell-based therapies and imaging are no exception. Not much is known, to effectively assess the biodistribution, clearance and efficacy of cell-based therapies due to the absence of an appropriate noninvasive imaging tool. In vivo cell tracking could provide information about distribution, localization, and clearance of various cell-based therapies including immune cells (CAR-T cells), stem cells and hepatocytes post-administration in the body. There are various non-invasive molecular imaging modalities that could be employed to track cell based therapies including optical imaging via fluorescence imaging (FLI)1,2, bioluminescence imaging (BLI)3,4, and ultrasound-guided photoacoustic imaging (PA)5,6,7. Radiology imaging including magnetic resonance imaging (MRI)8,9,10, computed tomography (CT)11,12,13, and nuclear medicine imaging such as positron emission tomography (PET)14,15,16,17,18 and single photon emission computed tomography (SPECT)19,20, could also be employed to effectively measure the distribution, localization, and clearance of various cell-based therapies over time and to shed light on safety and efficacy.

Among various imaging modalities, optical imaging modalities are restricted to small animals due to limited tissue penetration (1–2 mm) in humans. MRI and CT provide high resolution anatomical information, but have low sensitivity in both animals and humans. Both PET and SPECT are advantageous over other techniques and are often integrated with CT and MRI. The PET/CT or SPECT/CT or PET/MRI provide quantitative and temporal distribution of immune and stem cells in animals and patients with no limitation of tissue penetration due to high energy gammas21,22,23,24.

Cell radiolabeling using a direct radiolabeling approach with various SPECT radiopharmaceuticals such as [99mTc]Tc-HMPAO (t1/2 = 6.01 h)27,28,29, and [111In]In-oxine (t1/2 = 68.2 h)30,31,32,33 have been used to track leukocytes for infection and inflammation imaging over the past four decades. SPECT is a powerful clinical imaging tool with lower usage cost than PET since an onsite cyclotron is not needed. PET, however, has many advantages over SPECT including two to threefold higher sensitivity, superior spatial resolution in the clinical setting, and with its quantitative nature it is a preferred imaging modality for tracking a single cell or small number of administered radiolabeled cells with more precise quantification and hence, requires lower radiation exposure34. Examples of commercial PET probes used to label cells include [18F]FDG (t1/2 = 109.7 min, β+  = 97%)35, [64Cu]Cu-PTSM (t1/2 = 12.7 h, β+  = 17.9% )36 and [68 Ga]Ga-oxine (t1/2 = 68 min, β+  = 88.8%)37,38.

Recently, among various PET radioisotopes, zirconium-89 (β+  = 22.3%) is gaining popularity for cell tracking due to its well established cyclotron-mediated production, longer half-life of 3.27 days and low average positron energy (Eβ+ = 0.395 MeV). This enables monitoring of radiolabeled cells up to 3-weeks, either through direct cell labeling (also called non-specific cell labeling agents)39,40 or indirect labeling mediated through antibodies41,42, peptides43, proteins44 and nanoparticles45,46,47.

Various chelators used for the radiolabeling of cells with 89Zr are tropolone, malonate, hydroxamates, and oxine (8-hydroxyquinoline). Among these, oxine forms a lipophilic complex with 89Zr and enters the cells passively. To date, [89Zr]Zr-oxine is a commonly used radiotracer to label various cells including tumor cell lines48,49, bone marrow cells50,51, T cells52, NK cells53, white blood cells (WBCs)54, stem cells (SCs)55 and leukocytes56. However, efflux of 89Zr from cells labeled with [89Zr]Zr-oxine remains a challenge. Recently, Friberger et al. reported a one-step clinically translatable method of synthesis of [89Zr]Zr-oxine with a cell labeling efficiency of 61–68% with human decidual stromal cells (hDSCs), bone marrow-derived macrophages (rMac) and human peripheral blood mononuclear cells (hPBMCs). However, a 29–38% apparent efflux of 89Zr from the labeled cells raised a further concern of radiotoxicity and non-specificity of the signal57.

Besides [89Zr]Zr-oxine, the other reported method of cell labeling was covalent attachment of radiolabeled [89Zr]Zr-DFO-Bn-NCS complex with the primary amines present on the cell surface proteins to form stable thiourea bonds, which has solved the efflux problem observed with [89Zr]Zr-oxine57,58,59,60. The [89Zr]Zr-DFO-Bn-NCS has been successfully used to radiolabel mouse melanoma cells, mouse dendritic cells and human mesenchymal stem cells with insignificant efflux of free 89Zr from [89Zr]Zr-DFO-Bn-NCS over time (7 days-post radiolabeling)58. Additionally, a better version of the DFO chelator as DFO* has been developed to further strengthen the stability of 89Zr complexation and has shown lower bone uptake over time61. Various other chelators are also being developed to address in vivo stability of the 89Zr complex over time62,63,64,65,66.

In this work, we have optimized and compared the radiolabeling yields of WBCs and SCs using three different ready-to-use labeling synthons [89Zr]Zr-Hy3ADA5-NCS66, [89Zr]Zr-Hy3ADA5-SA66 and [89Zr]Zr-DFO-Bn-NCS57,58,59,60 (Fig. 1), and evaluated their applications in cell trafficking to better understand the biodistribution/pharmacokinetics of cell based therapies. This approach could be extended to various other cell-based therapies like CAR-T cell therapy.

## Results and discussion

### Production of [89Zr]ZrCl4 and radiosynthesis of [89Zr]Zr-DFO-Bn-NCS, [89Zr]Zr-Hy3ADA5-NCS and [89Zr]Zr-Hy3ADA5-SA

The PET isotope 89Zr was produced and purified in-house using a cyclotron as described earlier by Pandey et al.67,68,69,70 in a high apparent molar activity of 17.0–23.13 GBq/µmol, as assessed by complexing purified [89Zr]ZrCl4 with different amounts of DFO-Bn-NCS (Fig. S1, supplementary figure). All three synthons were successfully conjugated with 89Zr at 37 °C; pH 7.5–8.0 for 30 min in 72–98% radiolabeling yield. The DFO-Bn-NCS showed the highest complexation yield of 97.76 ± 0.31% (n = 3) followed by Hy3ADA5-NCS, 88.85 ± 0.05% (n = 3) and Hy3ADA5-SA, 71.58 ± 0.47% (n = 3) (Table 1, Fig. 2). These results indicate that acyclic chelator DFO-Bn-NCS imparts faster binding kinetics as compared to hybrid “cyclic-acyclic’ chelators Hy3ADA5-NCS and Hy3ADA5-SA. This is consistent with the complexation yield reported in our previous work, where a higher complexation yield was observed for DFO derivatives as compared to Hy3ADA derivatives66.

### Small animal PET imaging and biodistribution of 89Zr labeled WBCs

Overall, the in vivo stability of [89Zr]Zr-DFO-Bn-NCS as demonstrated here is promising and superior over other synthons [89Zr]Zr-Hy3ADA5-NCS and [89Zr]Zr-Hy3ADA5-SA. Of these, [89Zr]Zr-Hy3ADA5-SA showed the lowest in-vivo stability but had considerably higher in vitro stability66. The biodistribution of radiolabeled WBCs in the rest of the major organs are presented in Figs. 6 and 7 and Table 2, indicating mild uptake in lung, heart, muscle, pancreas, and skin at 7 days post injection.

### Small animal PET imaging of un-chelated [89Zr]ZrCl4

The in vivo characteristics of un-chelated [89Zr]ZrCl4 was also investigated (Fig. 11). The small animal PET imaging showed a high accumulation of free 89Zr in the bones at 4 h and did not distribute to the lung, liver, spleen, or any other organs at any other time points as were noted with radiolabeled WBCs and SCs. The radioactivity increased significantly on day 2 and remained in the bones until day 7, attributed to the entrapment of osteophilic 89Zr and its poor clearance from the bones (Fig. 12, Table 4). This observation is consistent with findings by Abou et al. 2011 demonstrating that [89Zr]ZrCl4 is a bone-seeking species and accumulates in bones and joints post-administration in mice71.

## Materials and methods

### General

The 89Zr used in this study was produced on a PETtrace cyclotron (GE Healthcare, Waukesha, WI) using 89Y target foil (0.1 mm; 50 X 50 mm, 99.9%), which was purchased from Alfa-Aesar, Haverhill, MA. The trace metal grade nitric acid (67–70%) and hydrochloric acid (34–37%) were purchased from Thermo Fisher Scientific, Waltham, MA. Sodium bicarbonate, oxalic acid dehydrate (TraceSELECT® ≥ 99.9999% metal basis), sodium carbonate, sodium citrate dihydrate and HPLC grade acetonitrile were purchased from Sigma Aldrich, St. Louis, MO. The silica gel iTLC was purchased from Agilent Technologies, Santa Clara, CA. The chelator p-SCN-Bn-Deferoxamine or DFO-Bn-NCS (≥ 94%) was purchased from Macrocylics, Plano, TX, whereas the other two chelators Hy3ADA5-NCS and Hy3ADA5-SA were synthesized as described by Klasen et al.66. The empty Luer-Inlet SPE cartridges (1 mL) with frits (20 µm pore size) were purchased from Supelco Inc (Bellefonte, PA) and Chromafix® 30-PS-HCO3 PP cartridges (45 mg) were purchased from Macherey–Nagel, Duren, Germany. The Millex ®-GV filter (0.2 µm) was purchased from Millipore Sigma, Burlington, MA. The hydroxamate resin was synthesized in-house as demonstrated by Pandey et al.67,68,69,70 The Thermomixer was purchased from Eppendorf, Hamburg, Germany.

### Production and purification of [89Zr]ZrCl4

The 89Zr was produced using yttrium foil on a solid target through a 89Y(p,n)89Zr nuclear reaction in a PETtrace cyclotron as described previously by Pandey et al.68. 89Zr was purified first as [89Zr]Zr-oxalate and then converted to [89Zr]ZrCl4 using activated Chromafix 30-PS-HCO3 SPE as demonstrated by Pandey67,68,69,70 and Larenkov et al.72, respectively. The final [89Zr]ZrCl4 was eluted in ⁓ 0.5 mL of 1.0 N HCl and then dried using a steady flow of nitrogen gas in a V-vial at 65 °C.

### Apparent molar activity of [89Zr]ZrCl4

The apparent molar activity of 89Zr was estimated using a DFO-Bn-NCS titration method. In this method, 10µL [89Zr]ZrCl4 (36.4 MBq) was added to 90µL de-ionized H2O. To this, 4 µL of 0.5 M Na2CO3 was added to neutralize and adjust the pH to 7.5–8.0. To the neutralized mixture, 0.01–10 µg of DFO-Bn-NCS in 4µL of DMSO was added and mixed. The complexation mixture was then incubated at 37 °C for 1 h. After 1 h, the degree of 89Zr complexation was determined with respect to the DFO-Bn-NCS concentration using radio-TLC with 20 mM sodium citrate (pH 4.9–5.1) as a mobile phase. The complexed 89Zr as [89Zr]Zr-DFO-Bn-NCS showed at the origin with Rf = 0, whereas free or un-complexed 89Zr had an Rf of 0.99 (solvent front). The half maximal inhibitory concentration (IC50) of DFO-Bn-NCS in mg/mL was calculated using non-linear ﻿regression curve fitting analysis.

The analysis was performed using analysis software—GraphPad Prism 9 (GraphPad Software, San Diego, CA). The minimum ligand concentration for which 100% complexation occurred was estimated by multiplying the IC50 by 2, and the apparent molar activity (GBq/µmole) and the apparent specific activity (GBq/mg) of 89Zr were calculated by correcting for the total activity divided by µmoles or mg of DFO-Bn-NCS needed for 100% 89Zr complexation.

The radiosynthesis of the different synthons [89Zr]Zr-DFO-Bn-NCS, [89Zr]Zr-Hy3ADA5-NCS and [89Zr]Zr-Hy3ADA5-SA were performed using a modified procedure demonstrated in our previous work.58 The purified [89Zr]ZrCl4 was resuspended in appropriate volume of 0.1 N HCl and then neutralized to pH ⁓ 8.0 with 0.5 M Na2CO3. The neutralized [89Zr]ZrCl4 solution (70 -100 µL) containing ~ 21 MBq of 89Zr was used in the case of DFO-Bn-NCS, whereas ~ 61 to 68 MBq of 89Zr was used in the case of Hy3ADA5-NCS and Hy3ADA5-SA. To this neutralized [89Zr]ZrCl4 solution, 4 nmoles of DFO-Bn-NCS or Hy3ADA5-NCS or Hy3ADA5-SA (prepared in DMSO) were added in separate reactions. The resultant reaction mixtures were stirred for 30 min at 37 °C in a thermomixer at 500 rpm. The radiolabeling efficiency was determined at different time points by silica radio-TLC using 20 mM sodium citrate (pH 4.9–5.1) as a mobile phase.

### Cell preparation

The human mesenchymal SCs were gifted by Dr. Atta Behfar from the Department of Cardiovascular Medicine, Mayo Clinic, Rochester, MN, USA, and WBCs were isolated from the peripheral blood provided by the Division of Transfusion Medicine, Mayo Clinic, Rochester, MN, USA. The isolation of WBCs from the peripheral blood was performed using Lymphoprep™ (STEMCELL Technologies Inc., Canada) gradient centrifugation method as per manufacturer instructions. The final WBC solution was washed with Hank’s Balanced Salt Solution.

### Trypan blue exclusion assay cellular viability test

The effect of radiolabeling on cellular viability was assessed using the trypan blue exclusion assay test within 1 h of labeling.

#### Animals

8–10 week old athymic nude mice (male and female, 1:1) were obtained from Charles Rivers Laboratories or Taconic Biosciences, Inc.

### PET imaging and ex vivo biodistribution studies

After radiolabeling, the radiolabeled WBCs (0.1–0.6 × 106; 0.03–0.11 MBq); and SCs (0.1–1 × 106; 0.1–0.15 MBq) were injected via tail vein into a group (n = 3) of athymic nude mice. PET images were acquired at 4 h, 2 days, 4 days and 7 days post-injection (p.i.) using a small animal PET scanner. The free [89Zr]ZrCl4 with radioactivity (0.15–0.19 MBq) was also injected intravenously via the tail vein. The small animal PET images were visualized, analyzed, and scaled to SUV using image analysis software, MIM 7 software (MIM Software Inc., Cleveland, OH, USA). The PET images are shown as maximum intensity projection (MIP) images in the coronal and sagittal plane. The animals were euthanized at 7d p.i., and organs/tissues collected to measure the standardized uptake value (SUV) in major organs. Animals were euthanized via cardiectomy under anesthesia using isoflurane as approved by the Institutional Animal Care and Use Committee (IACUC) of the Mayo Clinic Rochester MN USA. SUV was calculated using following formula:

$$Standardized \,Uptake\, Value =\frac{Radioactivity \,concentration\, in\, tissue\, (\upmu Ci/g)}{{\frac{Injected \,dose\, (\upmu Ci)}{Body\, Weight (g)}}}$$

### Statistics

The obtained data were analyzed using Microsoft Excel program and the results were compared using unpaired Student’s t-test analysis. Differences were regarded as statistically significant for p < 0.05.

### Ethical standards

Studies were conducted with proper use and care of animals as approved by the Institutional Animal Care and Use Committee of the Mayo Clinic Rochester MN USA. Additionally, all methods were also performed in accordance with Institutional Animal Care and Use Committee’s guidelines and regulation. These guidelines are equivalent to the ARRIVE guidelines and therefore all methods were also performed in accordance with ARRIVE guidelines.