Peripheral blood samples and Leukocyte Reduction System cones (NHS Blood and Transfusion Service, NHS, UK) were obtained from healthy human volunteers. Samples were collected with informed consent and following a favourable ethical opinion from The Animal Welfare and Ethical Review Body (AWERB), Newcastle University.

Clinical-grade PLGA nanoparticles with entrapped perfluorocarbon in combination with gadoteridol (Prohance™; Bracco Imaging Europe) and a fluorescent dye, IC-Green (Akorn Inc., IL, USA) were prepared by Cenya Imaging (the Netherlands) using a mini-emulsion formulation method (20). All non-sterile solutions were filtered through an 0.22 mm filter. PLGA (100 mg, Resomer RG 502 H, lactide:glycolide molar ratio 48:52 to 52:48; Evonic Industry, Germany) was dissolved in dichloromethane (3 ml) and mixed rapidly with PFCE (900 μl, Exfluor Inc., TX, USA), after which IC-Green (1 mg) and Prohance™ (1780 µl) were added. This mixture was sonicated for 8 seconds at an amplitude of 40% at ≤21°C using a digital sonicator from Branson Ultrasonics (Connecticut, USA). The mixture was then added to an aqueous solution of poly(vinyl alcohol) (2.5% m/v) and emulsified for 3 minutes under sonication at 90% amplitude at ≤21°C. The solvent was evaporated overnight at room temperature under stirring in open class I glass flacons. Particles were collected by centrifugation (35 min, 13304 RCF, 4 °C) and washed 3 times with cooled water-for-injection (WFI). After these washing steps, the particle pellets were homogenized on a programmable rotator. Subsequently, 3 washing steps with cooled WFI were performed before the particle pellets were homogenized again, pooled and resuspended in WFI and lyophilized, generating a powder with a light green tinge. ¹⁹F content ranges from 5.8 × 10¹⁸ to 1.3 × 10^{19 19}F atoms/mg (determined with trifluoroacetic acid as internal reference). Gadoteridol content was measured using inductively coupled plasma mass spectrometry (ICP MS); the particles typically contain 0.5 μg/mg.

Peripheral blood mononuclear cells (PBMC) were isolated by gradient density centrifugation using Lymphoprep (Alere). Monocytes were further separated from PBMC by positive selection using CD14⁺ magnetic microbeads and a MACS LS column (Miltenyi) according to manufacturer’s instructions. Monocytes were seeded into 24-well culture plates (Corning) at 0.5 x 10⁶/ml (total of 1 ml per well) in GMP DC medium (CellGenix) containing IL-4 and GM-CSF (both at 50 ng/ml; Immunotools) and cultured for 7 days at 37°C, 5% CO₂. On day 3, IL-4 and GM-CSF were refreshed at 50 ng/ml final concentration, and half of the medium was replaced with fresh GMP DC medium. To generate tolDC, dexamethasone (Sigma) was added at 10^-8 M on days 3 and 6 of culture, and 10^-10 M 1,25-dihydroxyvitamin D3 (either Calcitriol from Tocris or Decostriol from Mibo) and 1 µg/ml of the TLR4 agonist monophosphoryl A (MPLA; Avanti) on day 6. To generate mature DC (matDC), cells were treated with 1 µg/ml MPLA (Avanti) on day 6 of culture only. In addition, on day 6 tolDC were labelled with ¹⁹F-NP (1.1, 2.2 or 4.4 mg/ml) or left unloaded as a control. DC were harvested on day 7, resuspended in HBSS + 1% FCS in a 30 ml universal tube (Starlab) and placed on ice for further processing (see sections below). Viability and cell numbers were determined by trypan blue (Sigma).

DC were cryopreserved in pre-cooled GMP DC medium (CellGenix) supplemented with 40% Human Serum Albumin (HSA)(Biotest) and 10% DMSO (Sigma) in a cryovial (Corning). Cryovials were immediately placed in a cool cell in a -80°C freezer for 24 hours before transfer into a storage box. Cryovials were stored short-term (less than one month) in a -80°C freezer or in liquid nitrogen long-term (more than one month). Cells were thawed in a 37°C water bath until ice crystals dissolved then transferred to an appropriate centrifuge tube where at least 5 ml of pre-warmed HBSS (Lonza) with 10% FCS (Biosera) was slowly added. The tubes were then topped up with HBSS with 10% FCS and washed in HBSS with 10% FCS by centrifugation twice at 400 xg for 8 minutes at room temperature before being counted and viability assessed by trypan blue (Sigma).

As part of the formal release criteria of therapeutic tolDC, we set up a flow cytometry panel to define the identity, contamination and viability of the cells. The following mAbs were incubated with cells, on ice in the dark for 30 minutes: CD11c (Biolegend), CD83 (Becton Dickinson), CD86 (Becton Dickinson), HLA-DR (Biolegend), TLR2 (Thermo Fisher), CD3 (BD), CD19 (BD), CD41 (Thermo Fisher), CD56 (BD) and CD66b (Biolegend), as well as relevant isotype-matched controls (see Table 1 in Supplementary Data for full details on mAbs; CE marked mAbs were used where possible). Zombie Aqua (Biolegend) was added to cells and incubated in the dark for 15 minutes to assess cell viability. Cells were fixed in 1% formaldehyde (TAAB Labs) and stored in the dark at 4°C for a maximum of 7 days. Cells were analysed on a Canto II or Fortessa flow cytometer (Becton Dickinson). Compensation was set up using compensation beads (for mAbs; Becton Dickinson) or using a mixture of stained and unstained cells (for Zombie Aqua) or unloaded and cells loaded with ¹⁹F-NP. Results were analysed by FCS Express RUO software (Denovo Software) or FlowJo software (FlowJo). Cells were gated by excluding debris and selecting the single cells. Cell viability and expression of surface markers (median fluorescent intensity (MFI) and percentage positive cells) was then calculated from this population.

Phantoms for MRI analysis were produced by heating 1% low melting point agarose in a water bath or microwave until all agarose had dissolved. Agarose was then briefly allowed to cool before layering 750 µl into a 1.5 ml microcentrifuge tube on ice. The agarose in the tube was allowed to set before adding a layer of DC resuspended in 200µl HBSS + 1% FCS. 100 µl of agarose was added on top of this and an additional 300 µl of agarose added again once the previous layer had set. Once set the tubes were stored at 4°C, protected from light until MRI was performed. MR imaging was performed on a Philips Achieva 3.0 Tesla (T) scanner (Philips Healthcare, Best, NL), equipped with a 20 cm diameter ¹⁹F-tuned surface coil (PulseTeq Ltd., UK) placed level with the scanner bed and loaded with 2 x 500 mL bags of 0.9% saline solution. The agar phantom was positioned upright, with DC suspension fixed 3 cm above the coil centre. ¹⁹F images were acquired with a 2D steady state free precession (SSFP) sequence with the following acquisition parameters: field of view = 160 x 160 mm², voxel size = 10 x 10 mm², slice thickness = 50 mm, echo time/repetition time = 2.5/5.3 ms, flip angle = 60°, bandwidth = 370 Hz/pixel, signal averages = 16,384, scan duration = 23 minutes. Analysis was performed in Matlab (MathWorks, Natick MA, USA) using the complex image data. Mean ¹⁹F signal ( S ¯ ) was measured from a manually drawn region of interest (ROI) around the signal locus. Noise mean ( N ¯ ) and standard deviation (σN ) were calculated from a 5 x 5 cm² ROI placed in a location devoid of ¹⁹F signal. SNR was calculated as ( S ¯   – N ¯ )/σN

As part of the formal release criteria of therapeutic tolDC, we optimised a ‘potency assay’ as an in-vitro surrogate for in-vivo activity to confirm that tolDC lacked the ability to induce IFNγ production by allogeneic cells. Allogenic PBMC from 4 different donors (10⁵ cells/well for each donor) were thawed (see above) and co-cultured with each of the DC populations (mature DC and/or tolDC; 10⁴ cells/well each) in a 96 well plate (Corning) in 200 µl of GMP DC medium (CellGenix) supplemented with 4% HSA (Biotest) in duplicate. After 4 days the supernatants from duplicates were combined, divided into aliquots and stored at -80°C for assessment of IFNγ by ELISA.

Day 7 DC culture supernatants were assessed for IL-10 and IL-12p70 and potency assay supernatants were assessed for IFNγ cytokine levels by ELISA. EIA plates (Corning) were coated with capture antibody (Becton Dickinson) and left overnight at 4°C. The next day plates were washed (PBS + 0.1% Tween-20) and block (PBS + 1% BSA) was added for at least 1 hour at room temperature. During the 1-hour incubation, samples and cytokine standards (Becton Dickinson) were thawed and diluted. The plates were then washed and samples and cytokine standards added to the plate and left to incubate overnight at 4°C. On the third day, plates were washed 4 times and detection antibody (Becton Dickinson) added. Plates were incubated at room temperature for 1 hour then washed and ExtrAvidin (Sigma) added and left for 30 minutes at room temperature. This was followed by 5 washes and the addition of OPD (Sigma) to develop colour and 3 M H₂SO₄ after at least 20 minutes to stop the reaction. Plates were read at 490 nm by a plate reader (Labtech). Microsoft Excel (Microsoft Corporation) was used to plot a concentration curve and calculate cytokine concentrations of the samples.

Day 7 DC were washed twice in glutaraldehyde and resuspended in glutaraldehyde for at least 1 hour to fix. Cells were then enrobed in 4% agar before further fixation and dehydration followed by processing to resin. A Hitachi TEM was used to image cells.

13 mm round glass slips (VWR) were prepared in 24 well plates by sterilising in 70% ethanol before monocyte seeding. PBMC separated monocytes were seeded in these 24 well plates and cultured for 7 days as described above. On day 7 DC on the glass cover slips (VWR) were washed, stained with DAPI (Thermo Scientific) and mounted with antifade mounting medium (Invitrogen) on a glass slide (Corning). Slides were left to dry for at least 24 hours in a dark area before transfer to a fridge. Imaging was performed on Zeiss LSM800 AiryScan confocal microscope using APC (ex651-em660) and DAPI (ex353-em465) channels with Plan-Apochromat 40x/1.30 Oil M27 objective. Image analysis was performed on Zen 2 software.

Statistical software in GraphPad prism 9 was used to produce and analyse graphs (GraphPad). Microsoft Excel was used to input, organise and calculate averages for flow cytometry, ELISA and Potency Assay data. For experiments where paired sample data were produced, Wilcoxon test was used to assess statistical significance. In the case of unpaired samples Kruskal-Wallis or Friedman test was used. Spearman’s Rank Correlation Coefficient was used to determine the correlation and significance of ¹⁹F-NP concentration and viability. A one-tailed binomial test was used to determine IL-12 ELISA detection significance between tolDC and matDC or ¹⁹F-tolDC and matDC.

The optimal concentration of ¹⁹F-NP nanoparticles for tolDC labelling was determined by assessing uptake and detection of these particles, as well as the effect of the labelling process on cell viability. ¹⁹F-NP nanoparticles are dually loaded with perfluorcarbon and ICG, allowing for detection by both MRI and fluorescence-based detection technologies. Initial flow cytometric analyses showed that labelling of tolDC with ¹⁹F-NP consistently resulted in nearly 100% of the cells taking up the nanoparticles ( Figure 1A ), irrespective of the ¹⁹F-NP concentration. The normalised median fluorescence intensity (MFI) values suggested some variability in the level of uptake, but again this was independent of the ¹⁹F-NP concentration used, with all labelling conditions showing a significantly higher normalised MFI compared to unlabelled tolDC, but not to each other ( Figure 1B ). Figure 1C shows an agar phantom containing ¹⁹F-tolDC for measurement of detectability by ¹⁹F-MRI. The field of view for a typical MRI scan ( Figure 1D ) demonstrates the intensity of signal-to-noise ratio (SNR) between increasing concentrations of ¹⁹F-NP detected. Higher concentration of ¹⁹F-NP in the tolDC medium resulted in higher ¹⁹F -MR signal. A statistically significant difference in ¹⁹F SNR is seen at a ¹⁹F-NP concentration of 2.2mg/ml compared to the 1.1 mg/ml concentration ( Figure 1E ). Viability of ¹⁹F-tolDC was variable but suggested a significantly negative correlation with the dose (p<0.05) ( Figure 1F ). Although the yield of the tolDC product was significantly reduced by labelling with ¹⁹F-NP ( Figure 1G ), the reduction was not prohibitive for the production of sufficient numbers of therapeutic tolDC and was therefore deemed acceptable. Uptake of ¹⁹F-NP by tolDC was confirmed by confocal microscopy ( Figures 1H–J ) and electron microscopy ( Figures 1K, L ), showing the presence of the ICG label or the merged ¹⁹F-NP PLGA membranes within the cytoplasm of the cells, respectively. Taken together, the data show that tolDC efficiently ingest ¹⁹F-NP and that the labelled cells are detectable by MRI. The 2.2 mg/ml concentration was taken forward for further work as it allowed for optimal detectability by MRI with acceptable tolDC viability and yield.

¹⁹F-tolDC surface marker expression was determined with a set of validated QC markers, used for our previous (12) and imminent tolDC clinical trial in RA patients. For formal QC purposes only the percentage of cells positive will be taken into consideration, but for completeness we have also presented MFI data ( Figure 2 ). CD11c and HLA-DR were used for identification of cells as myeloid antigen-presenting cells. Labelling of tolDC with ¹⁹F-NP did not change the number of cells expressing these markers, although the expression levels of both markers were reduced. Importantly, however, the markers used to define the ‘tolerogenic characteristics’ of tolDC (maturation marker CD83 and the co-stimulatory molecule CD86) were not increased in ¹⁹F-tolDC. In fact, the percentage of CD86 positive cells was even lower in ¹⁹F-tolDC as compared to unlabelled tolDC. TLR2 expression was determined because it is a known dexamethasone-induced marker which is expressed by tolDC (7). Although it is not part of the formal QC release criteria of our previous clinical-grade tolDC product, it has been included as a ‘for information only’ QC marker. Although expression levels of TLR2 were reduced in ¹⁹F-tolDC ( Figure 2 ), they remained higher than levels expressed by matDC for each individual product (data not shown). Not surprisingly, contamination of the tolDC product with other cell types was not affected by labelling with ¹⁹F-NP. The data demonstrate that, with regard to these established QC release markers, the impact of labelling with ¹⁹F-NP on tolDC is marginal and acceptable, particularly because the particles did not reverse the Dex/VitD3-induced inhibition of DC maturation.

We assessed whether labelling of tolDC with ¹⁹F-NP affected their cytokine secretion profile. We have previously shown that tolDC are characterised by production of anti-inflammatory IL-10 and low or undetectable levels of pro-inflammatory IL-12p70 (5–7), and both cytokines are part of the tolDC QC release criteria. ¹⁹F-tolDC were found to secrete significantly more IL-10 than unlabelled tolDC ( Figure 3A ) whilst, importantly, production of IL-12p70 was not enhanced ( Figure 3B ). These data support the notion that labelling of tolDC with ¹⁹F-NP does not negatively affect their tolerogenic features.

Current QC testing of Advanced Therapy Medicinal Products needs to include a quantitative measurement of a relevant biological activity, also referred to as a ‘potency assay’. Because a functional hallmark of tolDC is their reduced ability to induce IFN-γ in allogeneic T cells (5–7), we standardised a Good Manufacturing Practice (GMP)-compliant, short in vitro assay in which the ability of tolDC and matDC to induce IFN-γ production in allogeneic PBMC is compared. As shown in Figure 4A , both unlabelled and ¹⁹F-NP-labelled tolDC had similarly reduced ability to induce IFN-γ production in this assay when compared to matDC. In addition, further co-culture experiments showed that ¹⁹F-NP-labelled tolDC did not enhance the ability of matDC to induce IFN-γ ( Figure 4B ). Together with the phenotypic and cytokine secretion data, these functional experiments show that ¹⁹F-NP is compatible with tolDC.