In Vivo PET Imaging of Monocytes Labeled with [89Zr]Zr-PLGA-NH2 Nanoparticles in Tumor and Staphylococcus aureus Infection Models

Simple Summary Immune cells are increasingly used for therapy in cancer and other diseases. To better understand immune-cell kinetics, cell-tracking with highly sensitive imaging modalities is required. The aim of this study was to develop a new strategy for the in vivo tracking of a small number of cells, using positron emission tomography (PET). We labeled poly(lactic-co-glycolic acid) nanoparticles containing a primary endcap (PLGA-NH2) with the radionuclide zirconium-89. The nanoparticles were characterized for size, polydispersity index, zetapotential and radiolabel retention. Subsequently, they were used for the ex vivo radiolabeling of a monocyte cell line (THP-1). We demonstrated that these radiolabeled monocyte cells can be traced in vivo in mouse tumor and infection models. Abstract The exponential growth of research on cell-based therapy is in major need of reliable and sensitive tracking of a small number of therapeutic cells to improve our understanding of the in vivo cell-targeting properties. 111In-labeled poly(lactic-co-glycolic acid) with a primary amine endcap nanoparticles ([111In]In-PLGA-NH2 NPs) were previously used for cell labeling and in vivo tracking, using SPECT/CT imaging. However, to detect a low number of cells, a higher sensitivity of PET is preferred. Therefore, we developed 89Zr-labeled NPs for ex vivo cell labeling and in vivo cell tracking, using PET/MRI. We intrinsically and efficiently labeled PLGA-NH2 NPs with [89Zr]ZrCl4. In vitro, [89Zr]Zr-PLGA-NH2 NPs retained the radionuclide over a period of 2 weeks in PBS and human serum. THP-1 (human monocyte cell line) cells could be labeled with the NPs and retained the radionuclide over a period of 2 days, with no negative effect on cell viability (specific activity 279 ± 10 kBq/106 cells). PET/MRI imaging could detect low numbers of [89Zr]Zr-THP-1 cells (10,000 and 100,000 cells) injected subcutaneously in Matrigel. Last, in vivo tracking of the [89Zr]Zr-THP-1 cells upon intravenous injection showed specific accumulation in local intramuscular Staphylococcus aureus infection and infiltration into MDA-MB-231 tumors. In conclusion, we showed that [89Zr]Zr-PLGA-NH2 NPs can be used for immune-cell labeling and subsequent in vivo tracking of a small number of cells in different disease models.


Introduction
Cell-based therapy is maturing into clinical practice and holds great promise for treating cancer, as well as immune-related diseases. In vivo cell tracking is desired to better understand the complex cell targeting mechanism and cell-cell interactions. For example, such tools could guide the development of treatment strategies to increase tumor-targeting and minimize off-target accumulation and associated toxicity [1][2][3][4]. The effector immunecell populations involved in these therapeutic interventions reach peripheral tissues in relatively small numbers, in the order of a few thousand [5][6][7][8][9]. Therefore, stable labeling of immune cells and a highly sensitive imaging system with adequate tissue penetration depth for whole-body imaging are required.
For direct cell labeling, specific cell (sub)types are isolated from patients and labeled ex vivo. For example, T cells can be labeled with highly derivatized crosslinked iron oxide nanoparticles (NPs) and detected with magnetic resonance imaging (MRI) [10]. Although magnetic particle imaging (MPI) is a highly sensitive technique (detection of ∼200 cells and a resolution of ∼1 mm), there are currently no clinically available scanners [11][12][13][14]. In contrast, positron emission tomography (PET) is commonly applied in clinical practice for diagnosis, staging and response monitoring in cancer and other diseases. PET tracking of radiolabeled cells is traditionally performed with lipophilic compounds, including oxine and hexamethylpropyleneamine oxime (HMPAO), that passively diffuse across the cell membrane [15]. In clinical settings, [ 111 In]In-oxine and [ 99m Tc]Tc-HMPAO are used in combination with single-photon emission computed tomography (SPECT) imaging [16,17]. Although this labeling strategy is fast and results in a good labeling efficiency, it has several drawbacks. In particular, it has been shown that cells labeled with [ 111 In]In-oxine release the radionuclide up to 50% within 48 h [18], and the oxine carrier appears to be chemotoxic [19]. Furthermore, clinical SPECT systems have a low sensitivity and resolution, which hamper the tracking of low numbers of cells distributed over a large volume. In contrast, PET has a higher resolution, sensitivity and more accurate quantification compared to SPECT. New developments in PET technology hold promise for even higher sensitivity compared to current PET scanners [20]. Therefore, PET-based in vivo cell tracking is preferred over SPECT in the clinical setting. In this study, we introduced the radiometal zirconium-89 ( 89 Zr) as a positron emitter, with an ideal half-life (78.41 h) for cell tracking.
More recently, [ 89 Zr]Zr-oxine has been introduced for PET-based cell tracking [21,22], but it experiences similar limitations to those observed with [ 111 In]In-oxine [23,24]. A multitude of 89 Zr-labeled NPs for different applications have been reported previously [25][26][27]. For example, chitosan NPs have also been used for 89 Zr-labeling of leukocyte cells [28]. The intrinsic labeling of the chitosan NPs was suggested to be possible via the interaction of the 89 Zr with the OH and NH 2 groups. High cell-labeling efficiency was achieved with up to 73% after 24 h. However, the 89 Zr-release from the cells was also rapid, with up to 79% after 24 h. No specific activity per number of cells was reported, making it difficult to compare to other studies. In another study, in vivo tumor-associated macrophages were targeted and imaged by using 89 Zr-desferroxiamine-NCS (DFO) conjugated dextran NPs in colon carcinoma (CT26) tumor xenograft mice [25]. Here, also some release was detected in the bones with PET images. Radiolabel release by ex vivo labeled cells is a hurdle for sensitive and specific in vivo cell tracking, as free radionuclides accumulate in off-target tissue and could lead to higher background signal and potential misinterpretation of images, while also exposing tissue to unnecessary radiation dose [29,30].
In our previous work, we have shown that poly(lactic-co-glycolic acid) NPs with amine groups (PLGA-NH 2 NPs) can be used to radiolabel cells and demonstrates improved radiolabel retention compared with the oxine labeling method [31]. Here, we report the intrinsic labeling capacity of these NPs with [ 89 Zr]ZrCl 4 under various conditions. In vitro, immortalized human monocytes (THP-1) were labeled with [ 89 Zr]Zr-PLGA-NH 2 NPs and the retention of 89 Zr in the cell was studied over time. Finally, we show that it is feasible to image ex vivo labeled THP-1 cells with PET in mice with Staphylococcus aureus (S. aureus) inflamed muscles or human breast adenocarcinoma MDA-MB-231 tumors.

Characterization of Nanoparticles
PLGA-NH 2 NPs were analyzed for size, polydispersity index (PDI) and zeta potential, in the same way as in our previous study [31]. The NPs were dissolved at 0.1 mg/mL in MilliQ, and both size and PDI were measured by using a NANO-flex (Microtrac, Inc., Duesseldorf, Germany), and the data were analyzed by using Microtrac software (Microtrac FLEX 11.1.0.2, Duesseldorf, Germany). The zeta potential was measured by using Zetasizer Nano ZS (Malvern Instruments, Worcestershire, United Kingdom), where similar NP concentrations were dissolved in NaCl (5 mM, pH 7.4). Encapsulation efficiency of PFCE was measured by using a nuclear magnetic resonance (NMR, Bruker Avance III 400 MHz, Bruker BioSpin, Ettlingen, Germany) spectrometer coupled with a Broad Band Fluorine Observation (BBFO) probe. NPs,~5 mg, were dissolved in 500 µL deuterium oxide (D 2 O) containing 100 µL 1 volume% trifluoroacetic acid (TFA) in D2O. For quantification, the interscan relaxation delay (D1) was set at 5 times the relaxation time (T1) of TFA, at 20 s. The data were evaluated with Mestrenova 10.0.2 (Mestrelab Research, Escandido, CA, USA).

Characterization of Nanoparticles Stability in Human Serum and PBS over Time
PLGA-NH 2 and Zr-PLGA-NH 2 NPs' size and PDI were measured in 100% human serum (human male AB plasma, Sigma-Aldrich, USA) and PBS at 0, 1, 2, 4, 6, 24, 48, 72, 168 and 336 h. First, the NPs were labeled with non-radioactive zirconium (932 µg zirconium/mg NP in 0.05 M HCl, pH 1.1-1.4, MO, USA) in metal-free 0.5 M ammonium acetate (NH 4 Ac, pH 5.5), which is similar to 89 Zr-labeling (see Zirconium-89 labeling of PLGA and PLGA-NH 2 NPs). Second, both PLGA and Zr-PLGA-NH 2 NPs were dissolved at a concentration of 10 mg/mL in PBS or 100% human serum. The samples were incubated at 37 • C, in a thermomixer, for the indicated timepoints. Last, 10 µL of NP solution was transferred to 990 µL MilliQ (0.1 mg/mL), and both size and PDI were measured as explained above.

[ 89 Zr]ZrCl 4 Preparation from 89 Zr-Oxalate
In order to obtain [ 89 Zr]ZrCl 4 , we removed oxalate by using a Sep-Pak Light Accell Plus QMA Cartridge (Waters, Dublin, Ireland). The Sep-Pak was activated with 10 mL acetonitrile and then washed with 10 mL 0.9% NaCl, 10 mL 1 M HCl and 10 mL water. This experiment was performed in the same manner as described in our previous study [31]. For intrinsic labeling, 1 mg NPs were dissolved in 0.5 M NH 4 Ac and incubated with 1-5 MBq [ 89 Zr]ZrCl 4 , at 37 • C, for 30 min. After washing the NPs 3 times with PBS, the labeling efficiency and radiochemical purity were determined with instant Thin-Layer Chromatography (iTLC). Labeling efficiency was calculated as the fraction of radioactivity at the origin to the total amount of radioactivity. Unless otherwise stated, the NPs were washed until a radiochemical purity of >95% was obtained. All radioactive labeling was performed in 0.5 M NH 4 Ac, pH 5.5, unless stated otherwise.

Cell Culture
The immortalized human monocyte cell line THP-1 (ATCC ® TIB-202 TM , VA, Gaithersburg, MD, USA) was used for cell labeling (passage of <20). The cells were maintained in culture as described previously [31].
The human adenocarcinoma cell line (MDA-MB-231, passage 46, ATCC ® HTB-26 TM , Gaithersburg, MD, USA) was cultured under the same conditions. Subsequently, cells were washed to remove NPs which were not taken up by the cells. After labeling and washing, cells were incubated at culture conditions for 1, 2, 4, 6, 24 and 48 h. At every timepoint, the cells were first measured for radioactivity for 1 min with a γ-counter (wizard 2480 Automatic Gamma Counter, PerkinElmer, Downers Grove, IL, USA). The cells were then centrifuged at 300× g for 5 min, the supernatant was removed and the cells were resuspended in fresh PBS before another radioactivity measurement. The percentage retained radioactivity in the cells was calculated by dividing the activity measured after removal of supernatant by total amount of radioactivity before centrifugation, multiplied by 100.

Cell Counting
Cell numbers after an experiment were counted with Luna-II Automated Cell Counter (Logos Biosystems, Inc., Anyang, South Korea). The mixture of cells with trypan blue (1:1) was transferred to a counting cassette (Logos Biosystems, Inc., Korea) before automated counting. Living cells were used for calculating the specific activity per number of cells by dividing the total activity associated with the pellet with the number of living cells times hundred.

CellTiter-Glo Assay
For ATP content measurement, 80,000 cells were diluted with PBS to a volume of 350 µL and mixed with 350 µL of premixed substrate and buffer CellTiter-Glo (Promega, Madison, WI, USA). After a short vortex, the samples were incubated for 10 min, at room temperature (RT). From each sample, 200 µL in triplicate was transferred to a 96-wells plate (black flat bottom), and luminescence was measured by using a Tecan Infinite M200 PRO and software Tecan i-control (attenuation automatic, integration time 1000 millisecond, settle time 0 millisecond, Tecan, Grödig, Austria). Controls were set to 100%, and sample results were compared to this.

Animal Experiments
For animal experiments, the guidelines set by the Nijmegen and European Animal Experiments Committee (CCD application 2018-0011 and 2020-0007) were followed. The animals were housed in groups in individually ventilated Blue line cages. To determine [ 89 Zr]Zr-PLGA-NH 2 NPs biodistribution and blood clearance, 6 female C57BL/6JRj mice (Janvier Labs) were used (age 6-8 weeks, weight 18.4 ± 1.2 g). For PET and MRI studies with [ 89 Zr]Zr-PLGA-NH 2 NPs labeled THP-1 cells in Matrigel, 12 female BALB/cAnNRj-Foxn1nu/Foxn1nu mice (Janvier Labs) were used (age 6-8 weeks, weight 20.0 ± 0.9 g). In vivo tracking of [ 89 Zr]Zr-THP-1 cells in S. aureus and MDA-MB-231 tumor models were performed in 11 female BALB/CAnN.Cg-Foxn1nu/Crl mice (Charles River) (age 6-8 weeks, weight 16.5 ± 2.3 g). The mice were allowed to acclimate for 1 week before the start of the experiments. Upon arrival, the mice were randomly identified with tattoos by biotechnicians who were blinded to the experimental setup.
Immediately following the PET scan, mice were transported by using the same imaging bed to the MRI scanner for anatomical imaging, where they were scanned for a duration of 32 min. A birdcage body coil (Bruker BioSpin, Ettlingen, Germany) with 86 mm inner diameter was used for image acquisition. After a localizer scan, the following settings were used for the 3D gradient echo scans: acquisition time = 32 min; repetition time = 30 ms; echo time = 1.3 ms; flip angle = 15 degrees; field of view = 120 × 56 × 32 mm; and matrix = 576 × 270 × 160, resulting in an isotropic resolution of 0.20 mm 3 .
The PET/MRI images were merged, and Inveon Research Workplace software (version 4.1) was used to create maximum-intensity projections. For the overlay, a reference tube with 89 Zr in PBS was placed on the scan bed. NPs (3400 ± 2194 Bq, n = 4) in PBS was s.c. injected as a control. For blood kinetics, blood samples were collected via saphenous vein or heart puncture (after sacrifice) at 30 min (4 mice), 1 h (4 mice), 4 h (4 mice) and 24 h (6 mice). For ex vivo biodistribution, organs and Matrigel were harvested and measured as described previously.
For in vivo PET/MRI imaging, mice were imaged at 1 h (4 mice) and 24 h (4 mice) with PET and followed immediately by MRI (Bruker BioSpec 117/16 11.7T, Bruker BioSpin, Ettlingen, Germany). For PET, the same settings were used as previously described.
Immediately following the PET scan, mice were transported by using the same imaging bed to the MRI scanner for anatomical imaging, where they were scanned for a duration of 45 min. A birdcage body coil (Bruker BioSpin, Ettlingen, Germany) with 72 mm inner diameter was used for image acquisition. After a localizer scan, the following settings were used for the 3D gradient echo scans: acquisition time = 42 min; repetition time = 100 ms; echo time = 2.2 ms; flip angle = 30 degrees; field of view = 100 × 40 × 40 mm; and matrix = 400 × 160 × 160, resulting in an isotropic resolution of 0.25 mm 3 . For PET/MRI imaging and processing of images, the same settings were used as in PET and MRI imaging of [ 89 Zr]Zr-PLGA-NH 2 NPs labeled THP-1 cells in Matrigel.

Statistical Analysis
Statistical analysis was performed in GraphPad Prism software. A Gaussian distribution with a two-tailed unpaired Student's t-test was used to analyze differences in two groups. A p-value of <0.05 was considered statistically significant. When comparing three or more groups, a one-way ANOVA test with a Tukey test correction of multiple comparisons was performed. A two-way ANOVA test with a Sidak correction of multiple comparisons was used in the biodistribution experiments (Figure 2A), while a Tukey's correction was used to compare the groups in the Matrigel experiments ( Figure 5A). All comparisons were based on a minimum of three replicates.

Characterization of Particles
The PLGA-NH 2 particles, prepared as described previously [31], were labeled with non-radioactive zirconium (Zr) and characterized for diameter, polydispersity (PDI) and zeta potential (Table 1). The diameter increased slightly when the PLGA-NH 2 NPs were labeled with zirconium from 189 ± 1.9 nm to 196 ± 4.1 nm. The PDI remained the same while the zeta potential increased to a more neutral charge, from −2.3 ± 0.9 mV to −0.3 ± 0.4 mV. These results indicate that NP characteristics were altered by the Zrlabeling, with increased size and zeta potential. Table 1. Characterization of PLGA-NH 2 and Zr-PLGA-NH 2 NPs: size (n = 3), PDI (n = 3) and zeta potential (n = 3).

Stability of the PLGA-NH 2 and Zr-PLGA-NH 2 NPs in PBS and Human Serum
To assess whether the characteristics of the particles changed over time, we incubated the PLGA-NH 2 and Zr-PLGA-NH 2 NPs in PBS and 100% human serum at 37 • C for a period of 2 weeks. The diameter of the NPs remained stable (~200 nm) in PBS for 72 h and was increased (~300 nm) at 336 h ( Figure S1A). Similarly, the PDI of both NPs remained stable (~0.08) for 72 h and was increased (~0.2) at 336 h. In human serum, the diameter of the NPs was increased (>200 nm) at 336 h ( Figure S1B). The PDI of both samples showed similar fluctuations as observed for the diameter over time.

[ 89 Zr]ZrCl 4 Labeling of PLGA and PLGA-NH 2 NPs
PLGA and PLGA-NH 2 NPs were radiolabeled with [ 89 Zr]ZrCl 4 , where a labeling efficiency of 7.1 ± 0.9% and 101.5 ± 1.1% for PLGA NPs and PLGA-NH 2 NPs (p < 0.0001, Figure 1A) was observed, respectively, showing efficient 89 Zr-labeling of PLGA-NH 2 NPs, without the need for additional chelator. To evaluate the effect of buffer on labeling efficiency, the PLGA-NH 2 NPs were labeled in 0.5 M HEPES, MES and NH 4 Ac buffer at a pH of 5.5 ( Figure 1B). Labeling efficiency was highest for the NH 4 Ac buffer (76 ± 2%, p < 0.0001 compared to HEPES and MES buffers). We therefore continued to label PLGA-NH 2 NPs in NH 4 Ac buffer. The retention of the 89 Zr by the NPs was measured in PBS and 100% human serum. In PBS and 100% human serum, the 89 Zr-retention was ∼85 ± 15% after 336 h ( Figure 1C). In addition, [ 89 Zr]Zr-PLGA-NH 2 NPs were challenged with EDTA at 37 • C, for 336 h. After an initial release of 89 Zr from the NPs during the first 6 h, a gradual and EDTA concentration-dependent release of 89 Zr was observed for up to 336 h ( Figure 1D). From these results, we can conclude that 89 Zr was interacting with the PLGA-NH 2 NPs and retained by the NPs in PBS and human serum. However, the 89 Zr-label could be challenged by EDTA.
Spleen, liver and bone marrow were the main organs for NP accumulation, as demonstrated by the result of the ex vivo measurements and PET/MRI scans ( Figure 2B and Figure 3 and Table S1). In addition, we also observed accumulation in femur (5.9 ± 0.1% ID/g at day 14) and knees (7.2 ± 1.8% ID/g at day 14). Taken together, these results show that the particles are cleared from the blood in the first 24 h after injection and that the spleen, liver and bone marrow are the main accumulation sites.  Spleen, liver and bone marrow were the main organs for NP accumulation, as demonstrated by the result of the ex vivo measurements and PET/MRI scans (Figures 2B and 3  and Table S1). In addition, we also observed accumulation in femur (5.9 ± 0.1% ID/g at day 14) and knees (7.2 ± 1.8% ID/g at day 14). Taken together, these results show that the particles are cleared from the blood in the first 24 h after injection and that the spleen, liver and bone marrow are the main accumulation sites.      Figure 5). All three Matrigel depositions were visible on the PET scans ( Figure 6). From the biodistribution data, we can see that the blood and organ signals were low, indicating that the radioactive signal remains at the Matrigel for over 24 h ( Figure 5 and Table S2). In all experiments, controls are THP-1 cells which were treated in the same way as other conditions without [89Zr]Zr-PLGA-NH2 but with PBS. Moreover, the controls did not change in value over time and therefore were set to 100%, and the remaining samples were compared to the controls. The mean and standard deviation of at least three independent experimental datasets are shown.

[ 89 Zr]Zr-THP-1 Cells for in Vivo PET/MRI Imaging
To determine PET sensitivity for the detection of low numbers of [ 89 Figure 5). All three Matrigel depositions were visible on the PET scans ( Figure 6). From the biodistribution data, we can see that the blood and organ signals were low, indicating that the radioactive signal remains at the Matrigel for over 24 h ( Figure 5 and Table S2).     Table S3). Furthermore, 0.46 ± 0.10% of the total administered [ 89 Zr]Zr-THP-1 cells (corresponding to 26.89 ± 5.79 × 10 3 cells) accumulated in the S. aureus-infected muscle, compared with 0.04 ± 0.02% (2.14 ± 0.89 × 10 3 cells) in the control muscle. A PET signal was detected as early as 4 h post-injection in the S. aureus-infected muscle, which increased at 24 h post-injection (Figure 8 and Videos S1 and S2).
the radiolabeled cells at 24 h ( Figure 7C,D and Table S3). The number of [ 89 Zr]Zr-THP-1 cells accumulated at the tumor site was 0.11 ± 0.05% (corresponding to 5.55 ± 2.53 × 10 3 cells) of the total injected cells, compared with 0.02 ± 0.00% (0.96 ± 0.19 × 10 3 cells) in the muscle of the foreleg. Moreover, a low PET signal in the tumor was detected at 4 h, which slightly increased at 24 h post-injection (Figure 8 and Videos S3 and S4).
Together, these sets of experiments show that it was feasible to track ex vivo [ 89 Zr]Zr-PLGA-NH2 NPs labeled THP-1 cells in two disease models.   Table S3). The number of [ 89 Zr]Zr-THP-1 cells accumulated at the tumor site was 0.11 ± 0.05% (corresponding to 5.55 ± 2.53 × 10 3 cells) of the total injected cells, compared with 0.02 ± 0.00% (0.96 ± 0.19 × 10 3 cells) in the muscle of the foreleg. Moreover, a low PET signal in the tumor was detected at 4 h, which slightly increased at 24 h post-injection (Figure 8 and Videos S3 and S4).
Together, these sets of experiments show that it was feasible to track ex vivo [ 89 Zr]Zr-PLGA-NH 2 NPs labeled THP-1 cells in two disease models.

Discussion
PET is currently the most sensitive whole-body-imaging modality for clinical studies that is ideal for in vivo tracking of small numbers of labeled cells. The long-lived positron emitter 89 Zr 4+ allows for imaging up to several days post-injection. This prompted us to explore the potential of [ 89 Zr]Zr-PLGA-NH2 NPs for cell labeling and in vivo tracking with PET.
We previously developed PLGA-NH2-based NPs that were able to intrinsically complex and retain [ 111 In]InCl3 for SPECT [31]. Here we demonstrated that these NPs also allow for intrinsic labeling with [ 89 Zr]ZrCl4 for PET. As expected, labeling with nonradioactive Zr slightly increased the NPs' size and zeta potential.
PLGA-NH2 NPs showed efficient labeling with [ 89 Zr]ZrCl4, compared to normal PLGA NPs without -NH2. In PBS and human serum, 89 Zr was retained for >80% by the NPs for up to 2 weeks. This indicates that the particles are able to retain the 89 Zr-label without the use of a chelator, such as desferrioxamine (DFO). However, when challenged with EDTA, 89 Zr was partly released from the particles, even at 0.1 mM (0.1 equivalents of EDTA) concentration. 89 Zr-release upon EDTA (1000 equivalents) challenge was also reported for DFO-conjugated trastuzumab, which showed a release of 25% and 50% in the first 24 h 7 days, respectively, which is slower than observed in our study [32]. From the literature, it was known that 89 Zr requires a strong Lewis base, such as OH − ions, and an 8-coordination for optimal binding and retention [33], which cannot be secured in the NPs, as chelation depends on free primary amine groups. However, for our application, the [ 89 Zr]Zr-PLGA-NH2 NPs mainly serve the purpose of ex vivo cell labeling, and the release, in the first instance, is mainly limited to the intracellular compartments of the labeled cells.

Discussion
PET is currently the most sensitive whole-body-imaging modality for clinical studies that is ideal for in vivo tracking of small numbers of labeled cells. The long-lived positron emitter 89 Zr 4+ allows for imaging up to several days post-injection. This prompted us to explore the potential of [ 89 Zr]Zr-PLGA-NH 2 NPs for cell labeling and in vivo tracking with PET.
We previously developed PLGA-NH 2 -based NPs that were able to intrinsically complex and retain [ 111 In]InCl 3 for SPECT [31]. Here we demonstrated that these NPs also allow for intrinsic labeling with [ 89 Zr]ZrCl 4 for PET. As expected, labeling with non-radioactive Zr slightly increased the NPs' size and zeta potential.
PLGA-NH 2 NPs showed efficient labeling with [ 89 Zr]ZrCl 4 , compared to normal PLGA NPs without -NH 2 . In PBS and human serum, 89 Zr was retained for >80% by the NPs for up to 2 weeks. This indicates that the particles are able to retain the 89 Zr-label without the use of a chelator, such as desferrioxamine (DFO). However, when challenged with EDTA, 89 Zr was partly released from the particles, even at 0.1 mM (0.1 equivalents of EDTA) concentration. 89 Zr-release upon EDTA (1000 equivalents) challenge was also reported for DFO-conjugated trastuzumab, which showed a release of 25% and 50% in the first 24 h 7 days, respectively, which is slower than observed in our study [32]. From the literature, it was known that 89 Zr requires a strong Lewis base, such as OH − ions, and an 8-coordination for optimal binding and retention [33], which cannot be secured in the NPs, as chelation depends on free primary amine groups. However, for our application, the [ 89 Zr]Zr-PLGA-NH 2 NPs mainly serve the purpose of ex vivo cell labeling, and the release, in the first instance, is mainly limited to the intracellular compartments of the labeled cells. However, in the course of time or upon cell death, 89 Zr can be released and redistributed within the body.
The biodistribution of the [ 89 Zr]Zr-PLGA-NH 2 NPs was in line with our previous observations with [ 111 In]In-PLGA-NH 2 NPs [34]. The signal at the tail was probably due to partial s.c. injection of the NPs. Interestingly, the accumulation in liver was half that of [ 111 In]In-PLGA-NH 2 NPs [31]. Furthermore, in spleen, activity at day 14 was only 50 %ID/g for [ 89 Zr]Zr-PLGA-NH 2 NPs, while it was >100 %ID/g for [ 111 In]In-PLGA-NH 2 NPs. Accumulation of 89 Zr was observed in the femur and knee at day 3, but this did not increase further at day 14. From the literature, it is known that free 89 Zr released from the targeting vehicle has the tendency to accumulate in bone tissue [29]. The radioactivity in femur and knee might be explained by (I) the <5% free 89 Zr present during injection of the NPs, (II) 89 Zr-release from the NPs after injection or (II) macrophages and monocytes that take up the NPs and are present in or migrate to bone marrow.
The labeling of the THP-1 cells with [ 89 Zr]Zr-PLGA-NH 2 NPs was not very efficient, as only ∼4% of the NPs was taken up by the cells. In general, cell labeling with [ 89 Zr]Zr-oxine is faster (15-30 min) and more efficient (10-40% labeling efficiency) when compared with NP-based cell labeling [35][36][37][38]. However, the specific activity of the NPs labeled cells was in range with the results from the literature, where human mesenchymal stem cells or chimeric antigen receptor (CAR) T cells were labeled for in vivo imaging with a broad range of specific activity of 0.009-0.370 MBq/10 6 cells, using desferrioxamine or oxine as carrier [21,37,39,40]. Moreover, higher specific activity per cell is not desired, as this could lead to radiotoxicity [37]. Furthermore, 89  Zr-oxine-labeled cells was also rapid for certain cell types (DCs and CAR T cells), i.e., >25% release after 2 days. These indicate that the NPs used in this study could play a role in cell labeling and in vivo tracking. However, future studies are needed to demonstrate feasibility of radiolabeling of other cell types, such as T cells. One strategy to enhance overall cellular uptake would be to modify the coating of NPs with, for example, cell-penetrating peptides or Lipofectamine [42][43][44]. Alternatively, to improve labeling of specific subsets of immune cells, NPs can be decorated with antibodies or peptides with the desired specificity [45,46].
In vivo studies showed that we were able to detect small numbers of labeled THP-1 cells, using PET. A clear signal was observed in mice which were transplanted s.c. with 10,000-100,000 [ 89 Zr]Zr-THP-1 cells (395-3950 Bq). Furthermore, minimal redistribution of radioactivity to other organs was observed, except for the femur and bone marrow, potentially caused by [ 89 Zr]Zr-THP-1 cells migrating to bone marrow or 89 Zr released from the cells. This indicates that 89 Zr is well retained inside cells.
Next, we injected [ 89 Zr]Zr-THP-1 cells i.v. and tracked their biodistribution in S. aureus inflammation model and a MDA-MB-231 tumor model. We detected a radioactive signal in the inflamed muscle and at the tumor site. However, it should be noted that the tumor accumulation was minimal, most likely because the tumor environment is less chemotactic compared with the S. aureus induced inflammation. Other studies have also developed techniques for PET-based cell tracking. For example, [ 89 Zr]Zr-oxine-based cell labeling has been evaluated in several studies with different type of cells and disease models. Recently, the potential of surface labeling with [ 89 Zr]Zr-DFO was shown by using human cardiopoietic stem cells for in vivo tracking in an ischemic-heart-failure mice model. Alternatively, a signal cell labeling and tracking was demonstrated with [ 68 Ga]Ga-mesoporous silica NPs, using PET [47]. The concept of single-cell tracking is highly challenging, as a high load of radioactivity per cell (>70 Bq) is required for accurate tracking. This could pose a problem in prolonged studies (24-72 h), since more radioactivity per cell would be required, as the half-life of 68 Ga is 67 min. Single-cell tracking would be interesting to study the behavior of that single cell; however, most effector mechanisms require cooperation with a multitude of other cells [48].

Conclusions
As PET is a highly sensitive imaging modality, in combination with novel cell-labeling approaches, it is ideally positioned for whole-body in vivo cell tracking. Here we expanded on our previous radiolabeling strategy and demonstrated for the first time that [ 89 Zr]Zr-PLGA-NH 2 NPs can be used as a tool for cell labeling and sensitive in vivo cell tracking, using PET. For future (clinical) applications, however, cell-labeling efficiency can be improved by coating the surface of the NPs with cell-specific antibodies, peptides, nanobodies or other targeting agents.

Data Availability Statement:
The data presented in this study are available in this article and supplementary material. The raw data are available upon request.