Use of Magnetotactic Bacteria as an MRI Contrast Agent for In Vivo Tracking of Adoptively Transferred Immune Cells

In vivo immune cell tracking using MRI can be a valuable tool for studying the mechanisms underlying successful cancer therapies. Current cell labeling methods using superparamagnetic iron oxide (SPIO) lack the persistence to track the fate and location of transplanted cells long-term. Magnetospirillum magneticum is a commercially available, iron-producing bacterium that can be taken up by and live harmoniously within mammalian cells as magneto-endosymbionts (MEs). MEs have shown promise as labeling agents for in vivo stem and cancer cell tracking but have yet to be evaluated in immune cells. This pilot study examined ME labeling in myeloid-derived suppressor cells (MDSCs), cytotoxic T lymphocytes (CTLs), and dendritic cells (DCs) and its effects on cell purity, function, and MRI contrast. MDSCs, CTLs, and DCs were incubated with MEs at various ME labeling ratios (MLR), and various biological metrics and iron uptake were assessed. For in vivo imaging, MDSCs were labeled overnight with either MEs or SPIO (Molday ION Rhodamine B) and injected into C3 tumor-bearing mice via tail vein injection 24 days post-implant and scanned daily with MRI for 1 week to assess cellular quantification. Following incubations, MDSCs contained > 0.6 pg Fe/cell. CTLs achieved Fe loading of < 0.5 pg/cell, and DCs achieved Fe loading of ~ 1.4 pg/cell. The suppressive functionality of MDSCs at 1000 MLR was not affected by ME labeling but was affected at 2000 MLR. Markers of CTL dysfunction were not markedly affected by ME labeling nor were DC markers. In vivo data demonstrated that the MDSCs labeled with MEs generated sufficient contrast to be detectable using TurboSPI, similar to SPIO-labeled cells. Cells can be labeled with sufficient numbers of MEs to be detectable with MRI without compromising cell viability. Care must be taken at higher concentrations of MEs, which may affect some cell types’ functional activity and/or morphology. Immune cells with minimal phagocytic behavior have much lower iron content per cell after incubation with MEs vs SPIO; however, MEs can successfully be used as a contrast agent for phagocytic immune cells.


Introduction
With the rapid advancement of personalized therapies in cancer, including adoptive cell transfer (ACT), peptide vaccines, and dendritic cell-based vaccines, the need to track therapy biodistribution and cell fate in vivo has grown considerably [1,2].However, a fundamental requirement for therapeutic success is that cells migrate effectively to and remain in the correct target organ or tissue [3].
One approach to assess the distribution of adoptively transferred cells in the body is to label cells with superparamagnetic iron oxide (SPIO) nanoparticles ex vivo and then track their migration with magnetic resonance imaging (MRI) following injection [4][5][6][7][8][9][10][11][12][13][14].Ex vivo cell labeling is achieved by incubating cells for a short time (often 24 h or less) in culture media containing SPIO particles.Cells internalize the particles, achieving intracellular iron concentrations in the nanogram (ng) to picogram (pg) range, depending on cell type and duration of incubation, without significant detriment to cell function [15][16][17].MRI cell tracking has even achieved single-cell resolution in vivo [18] with higher intracellular iron levels.However, the long-term fidelity of the SPIO signal in vivo is limited by the potential for signal dilution in dividing cell types and the generation of false-positive signals through the uptake of SPIO by resident macrophages following the release of the particles from dead cells [19].
Recently, magnetotactic bacteria have drawn interest for their ability to synthesize highly pure, membraneenveloped chains of magnetite (Fe 3 O 4 ) crystalsparticles that are often referred to as magnetosomesthat demonstrate effects on T 1 , T 2 , and T 2 * relaxation time [20,21], with the strongest effects on T 2 /T 2 *.Through endosymbiosis, non-pathogenic magnetotactic bacteria can be taken up by and live harmoniously within host mammalian cells, thereby becoming magnetoendosymbionts (MEs).Importantly, when the host cell dies, the magnetite chains should rapidly lose their structure, reducing the likelihood of false positives [22].The degradation is likely due to lysosomal digestion via macrophages and clearances via the reticuloendothelial system.This demonstrates the potential value of MEs as a live-cell-specific cell tracking agent in longitudinal studies.
For magnetotactic bacteria and/or their magnetosomes to be used to effectively track migrating cells in vivo, they must be sufficiently internalized within the cells of interest and persist long enough to allow subsequent imaging.A select number of studies have examined the cell tracking capabilities of MEs in various cell types following the internalization of either the isolated magnetosomes or entire bacteria [23][24][25].Notably, compared to SPIO (Molday ION) labeling of cells in a cardiac model, the ME signal correlated with bioluminescence.In contrast, the Molday signal persisted well after the extinction of the bioluminescent signal, suggesting the likelihood of false-positive signal detection over time when using synthetic magnetic particles [22].Another study [5] found that transverse relaxivity (R 2 ; equivalent to the inverse of T 2 ) values of MEs were significantly higher than those of Molday particles, demonstrating the innate strength of MEs as potential MRI contrast agents.These studies demonstrate the potential value of MEs as reliable livecell-specific contrast agents for cell tracking with MRI.
As mentioned above, whole magnetotactic bacteria as intracellular contrast agents may allow replication of the bacteria, potentially limiting attenuation of the signal over time and passing down of signal to daughter cells following meiosis [26].While a complete understanding of the fate of intracellular MEs over time has yet to be achieved, manipulation of MEs, either chemically or genetically, may allow MEs to persist for more extended periods within living cells, increasing their value in longitudinal imaging studies [26].As with isolated magnetosomes, internalization of MEs did not appear to affect cell viability or function in mice receiving transplants of ME-labeled cells [5,22,26] nor did they appear immunogenic.This evidence of biocompatibility, at both the cellular and organismal levels, along with the strong MR signal generated by ME-labeled cells in vivo, suggests MEs are a candidate as a contrast agent for preclinical cell tracking with MRI, with translatability to the clinical imaging milieu.
More work is required to assess the compatibility, reliability, and strength of MEs as contrast agents for studies of immune cell tracking [23].Currently, no examples of ME labeling with in vivo cell tracking in immune cells exist.In the present study, we compared MEs to SPIO as a cell tracking contrast agent for three immune cell types, isolated from mice, that are known to play critical roles in anti-tumor immunity: cytotoxic T lymphocytes (CTLs), dendritic cells (DCs), and myeloidderived suppressor cells (MDSCs).Iron assays were used to measure the uptake of MEs and intracellular iron in each cell type.Flow cytometry, immunohistochemistry (IHC), and live-cell assays were used to assess cell phenotype and functionality before and after ME labeling.The strength of the MRI contrast of labeled cells was assessed in vitro using traditional gradient-echo and balanced steady-state free precession pulse sequences.Finally, we used MRI, flow cytometry, and IHC to compare the migration and persistence of i.v.-injected homologous MDSCs labeled with either a traditional SPIO-Rhodamine B label or MEs in a murine model of cervical cancer.

Animals
Female C57BL/6-Tg(UbC-GFP)30Scha/J mice (n = 6) obtained from Jackson Laboratories (Bar Harbour, MN, USA) were used as donor mice for cell isolations.Female C57BL/6 mice (4-6 weeks old) were obtained from Charles River Laboratory (St.Constant, QC, Canada) and were bred in-house.For in vivo experiments, 6 GFP + (i.e., mice expressed green fluorescent protein on immune cells) mice were used as donor mice, and n = 13 C57BL/6 mice were used as recipient mice for labeled MDSC injections and subsequent imaging.All mice had access to food and water ad libitum and were housed under filter-top conditions.All animal experiments were carried out in accordance with protocols approved by the University Committee on Laboratory Animals at Dalhousie University, Halifax, NS, Canada.

Tumor Monitoring
Tumors were measured with calipers weekly beginning 6 days post-C3 implant.The formula [(shortest measurement) 2 × (longest measurement)/2] was used to calculate tumor volume.Once tumors reached 1000 mm 3 , measurements were performed twice weekly until the end of the study.Mice were terminated if tumor volumes exceeded 2000 mm 3 .

Cell Isolations
Mice used for cell isolations were euthanized by CO 2 inhalation at 20 days following C3 implantation.

Bone Marrow Isolation and Differentiation (DCs and MDCSs)
Tibias and fibulas were isolated from mice and sterilized in 70% ethanol.The marrow was flushed out with PBS using a 25-gauge needle and syringe.Red blood cells (RBCs) were lysed with 1 × RBC lysis buffer (Tonbo Biosciences) in PBS for 3 min.Cells were washed in PBS and resuspended at 3 × 10 5 cells/mL in cRPMI with antibiotics supplemented with either granulocyte-macrophage colony-stimulating factor (GM-CSF; 20 ng/mL, Peprotech) alone or with IL-6 (40 ng/mL, Peprotech) to drive differentiation of either DCs or MDSCs, respectively.Each cell type was cultured separately.Cells were cultured in 10 cm Petri dishes (3 × 10 6 cells in 10 mL of media/ dish) for 10 days.Fresh media (10 mL) with the same GM-CSF/IL-6 cytokine concentrations as day 0 were added on day 3 of culture.On day 6, 10 mL of cells was collected via gentle washing centrifuged at 300 × g for 5 min, resuspended in fresh media with the same cytokine concentrations as before and returned to the dish.All cells were collected on day 10.Cell collection from DC plates was done gently to not disturb adherent cells (collected only cells in suspension), whereas MDSCs were collected using a cell scraper (collected only adherent 1 3 cells).Cells were washed in PBS and resuspended in cRPMI without antibiotics in preparation for labeling.

Cell Labeling with SPIO and MEs
All cells were passaged into cRPMI without antibiotics at least one day prior to labeling with SPIO or MEs.For SPIO labeling, DCs and MDSCs were incubated at 2 × 10 6 cells/ mL with 0.03 mg/mL of Molday ION Rhodamine B (Biopal, Worchester, MA, USA) in cRPMI without antibiotics for 21 h.For ME labeling, MEs (trade name Magnelles, obtained from Bell Biosystems Inc., San Francisco, CA 5 ) were prepared as per supplier instructions.Frozen vials containing MEs were thawed at 37 °C, transferred to a conical tube, and centrifuged at 3000 × g for 15 min.The supernatant was removed, and MEs were resuspended in antibiotic-free cRPMI.Cells and MEs were combined in a 12-well plate at the volumes and concentrations shown in Supplementary Table 1 and cultured for 17-30 h.ME labeling ratios (ratio of the total number of magnetotactic bacteria to immune cells; MLRs) of either or both 1000 and 2000 were evaluated.DCs were tested at 1000 MLR, whereas CTLs and MDSCs were tested at 1000 and 2000 MLR.Following incubation, cells were washed twice in PBS and resuspended in cRPMI with standard 100 U/mL penicillin/streptomycin to clear away extracellular bacteria.

Prussian Blue Assay for the Quantification of Intracellular Iron Following SPIO Labeling
Intracellular iron in cells labeled with SPIO was quantified using a colorimetric Prussian Blue assay.A sample of 1-2 million cells was first lysed in 100 µL of 1 M hydrochloric acid (HCL, MilliporeSigma) overnight at 37 °C.The sample was centrifuged at 300 × g for 5 min, and the supernatant was transferred to a fresh tube.One hundred microliters of 1N potassium ferrocyanate (K 4 Fe(CN) 6 , Alfa Aesar, Ward Hill, MA, USA) was then added to the sample, resulting in blue color in the presence of iron.Absorbance was measured using ultraviolet/visible light spectrophotometry (l = 620 nm).Quantification was achieved by comparison to a standard curve of known iron concentrations ranging from 0.01 to 0.1 mg/mL.Iron per cell was derived by dividing total iron by the number of cells in the sample.

Ferrozine Assay for the Quantification of Intracellular Iron Following ME Labeling
Intracellular iron in cells labeled with MEs was quantified using a specialized colorimetric ferrozine assay (Bell Biosystems Inc.) as per supplier instructions.Briefly, 5 × 10 5 cells were lysed by heating the sample to 75 °C in 1.5 M HCL for 2 h.The sample was centrifuged at 300 × g for 5 min, and the supernatant was transferred to a new tube.The ferrozine iron probe was added to the sample, producing a purple color in the presence of iron.Absorbance was measured using ultraviolet/visible light spectrophotometry (l = 570 nm).

Cell Phenotyping with Flow Cytometry
Cell surface markers of cultured CTLs, MDSCs, and DCs were assessed using immunocytochemistry (ICC) and flow cytometry.ICC labeling was performed by first washing samples of 2-3 million cells three times in cold PBS.Following the third wash, cells were resuspended in 50 µL of PBS with 1 µL of Fc block (eBioscience, San Diego, CA, USA) and incubated at room temperature for 10 min.Following Fc blocking, 50 µL of PBS containing pre-conjugated antibody (Ab) was added to the tube, and the sample was incubated for 30 min at 4 °C.Following Ab incubation, cells were washed three times in cold PBS and resuspended in 300 µL of 4% paraformaldehyde (PFA; MilliporeSigma) in PBS for 10 min.Cells were then washed and resuspended in PBS and either immediately analyzed by flow cytometry or kept at 4 °C until analysis.

ME-labeled MDSCs, SPIO-labeled MDSCs, and unlabeled
MDSCs were prepared at increasing concentrations (0.5, 1, 1.5, and 2 × 10 6 cells for each type) in a small volume of PBS (100 µL) and then suspended in 2 mL of warm 8% Knox gelatine (ED Smith Foods, Winona, ON, CA).The gelatine-cell mixtures were quickly transferred to 5 mm NMR tubes and set on ice until imaging.

In Vitro MRI
Data was acquired on a 3 T preclinical magnet.The magnet was equipped with a 21-cm i.d.gradient coil (200 mT/m; Magnex Scientific, Oxford, UK) interfaced with a Varian DD Console (Varian Inc., Palo Alto, California, USA).A 30 mm i.d.quadrature transmit/receive RF coil (Doty Scientific, Columbia, SC, USA) was used for acquisition.Since spatially resolved T 2 * values were not necessary, bulk T 2 * values for the entire cell sample (in the NMR tube) were obtained by measuring the linewidth of individual samples.This was done using a non-spatially resolved hard pulse with a 45° flip angle and 10 μs pulse width and a 500 ms TR.We used the relationship of R 2 * = π * linewidth [28].T 2 * was calculated as 1/R 2 *.

MDSC Suppression Assay
T cell proliferation in the presence of MDSCs was measured to assess the presence of a suppressive phenotype.T cells were acutely isolated from LNs as previously described 10 .Following RBC lysis, the resulting isolate was used as responder T cells (Tresps).Tresps were washed twice in PBS to remove any FBS and then resuspended in PBS at 1 × 10 7 cells/mL.Cells were combined 1:1 with a warm 1 mM solution of e670 proliferation dye (eBioscience) in PBS and incubated for 10 min in the dark at 37 °C.Labeling was halted by adding 4-5 volumes of cold cRPMI and incubating on ice for 5 min.Cells were washed thrice and resuspended at 1 × 10 6 cells/mL in cRPMI.MDSCs -either naïve, SPIO-labeled, or ME-labeled -were collected, washed, resuspended at 1 × 10 7 cells/mL in cRPMI, and stored on ice until use.Cells were combined in a round-bottom 96-well plate in the amounts shown in Supplementary Table 2.All samples were plated in triplicate or more, avoiding outer wells.Outer wells were filled with water to reduce evaporation.Anti-CD3/Anti-CD28 coated dynabeads (Gibco) were added to appropriate wells to activate Tresps.Following 72 h in culture, samples were pooled, labeled by anti-TCRb-PerCP-Cy5.5 (1:100 dilution in PBS, clone H57-597, Invitrogen, Waltham, MA, USA) and anti-CD4-PE (1:100 dilution in PBS, clone GK1.5, eBioscience), and analyzed for e670 peak distribution (more peaks indicates proliferation) by flow cytometry to demonstrate suppression (or lack thereof) of Tresps.

Immunocytochemical Labeling
Cells were fixed by resuspending in 4% PFA in PBS for 10 min at room temperature and were then washed twice in PBS.Cells were permeabilized by incubating in 0.1% Triton-X (MilliporeSigma) with 1% bovine serum albumin (BSA; MilliporeSigma) in PBS for 3 min at room temperature, washed twice in PBS, and incubated in Alexa 488 phalloidin (Invitrogen) for 20 min at room temperature (2 × 10 6 cells in 200 mL of PBS with 5 mL of 6 mM Alexa 488 phalloidin stock solution).Cells were then washed twice in PBS.MElabeled cells were further labeled with rabbit anti-ME 594 (Bell Biosystems Inc., 1:1000 in 1% BSA in PBS) for 1 h at room temperature and then washed twice in PBS.After labeling, cells were resuspended in 8 µL of Fluoromount-G with DAPI (eBioscience), transferred to a slide, and coverslipped and sealed with clear nail polish.

Confocal Imaging
Slides were imaged using a Zeiss LSM 710 laser scanning confocal microscope (Carl Zeiss SBE, LLC, Thornwood, NY, USA) equipped with an XBO 50Wlamp for 4ʹ,6diamidino-2-phenylindole fluorescence (lex365, lem420 nm), an Argon laser (lex488 nm) with a 515 to 565 band-pass filter, and a HeNe laser for 548-and 633-nm excitation with a low-pass filter at 590 nm.Z-stacks were acquired for all images with a step size of 1 µm, and the upper and lower limits set to ensure all cells in the field of view were captured in focus.Imaging parameters included two averages per pixel at 1 AU, a pixel dwell time of 6 ms, and 1024 × 1024 resolution.

MDSC Culture and Labeling
Six days prior to cell injections, femurs and tibias were harvested from disease-matched donor mice.Bone marrow was flushed from femurs and tibias with a 25-gauge needle using PBS, and RBCs were lysed with 1 × RBC lysis buffer.The remaining cells were cultured for 5 days in cRPMI supplemented with GM-CSF (20 ng/mL) and IL-6 (40 ng/mL); see in vitro methods for further details on cell isolation and culture.Penicillin/streptomycin was omitted in media intended for use with MEs or ME-labeled cells.MDSCs were then incubated at a concentration of 2 × 10 6 cells/mL with either 0.03 mg/mL SPIO-Rhodamine B for 21 h or with MEs at an MLR of 1000 for 16 h.

MDSC Injections
MDSCs, labeled with either SPIO-Rhodamine B or MEs, were washed and resuspended in HBSS with 20 mM HEPES buffer at 15 × 10 6 cells/mL.Recipient mice were injected with 3 × 10 6 MDSCs (SPIO-labeled MDSCs, n = 8; ME-labeled MDSCs, n = 5) via the tail vein at 23 days post-C3 implant.The remaining cells were used to quantify iron loading using the same assays as in vitro studies.

Termination and Tissue Collection
One mouse from each group was terminated on days 24, 26, and 28 post-C3 implant for tissue collection.The remaining mice were terminated following the last imaging day (8 days post-cell injection).Spleens, tumors, and inguinal lymph nodes were harvested and either processed for flow cytometry or flash frozen in a 2:1 mixture of optimal cutting temperature compound (OCT; Fisher Healthcare, Houston, TX, USA) and 20% sucrose in PBS for immunohistochemistry.

Immunohistochemistry of Frozen Tissues
Cryosectioning of tumors was performed by the Pathology Department at the IWK Health Centre (Halifax, NS, Canada).Slides were fixed with 200 µL of 4% PFA for 15 min and then permeabilized with 200 µL of 0.5% Triton-X in PBS for 15 min.Samples were washed with PBS and blocked in 150 µL of 2% BSA in PBS for 1 h at room temperature.Samples were washed twice with PBS and incubated with the appropriate antibodies.Tissues of mice injected SPIO-labeled MDSC injections were incubated with anti-Gr-1 eFluor 660 (1:250; clone RBC-8C5, eBioscience) in 1% BSA in PBS overnight at 4 °C.Tissues were washed three times with PBS and mounted with Fluoromount-G with DAPI.Tissue slides from mice injected with ME-labeled MDSCs were incubated with rabbit anti-ME (Bell Biosystems Inc.) antibody diluted 1:1000 in PBS and anti-Gr-1 eFluor 660 (1:250; clone RBC-8C5) in 1% BSA in PBS overnight at 4 °C.Slides were washed three times with PBS and then incubated with goat anti-rabbit Alexa 594 (2 drops/mL PBS IgG H + L Ready Probes™, Invitrogen) for 1 h at room Slides were rewashed three times with PBS and mounted with Fluoromount-G with DAPI.Confocal imaging was done using the same microscope used for in vitro studies.

MRI Acquisition
The MRI and RF coil used for in vitro studies were used to image tumors and inguinal lymph nodes.Mice were anesthetized using ~ 2% isoflrane during MRI scans.Anatomical images were obtained using a 3D balanced steady-state free precession [29,30]

MRI Image Analysis
All images (anatomical, FSE, and R 2 * maps) were converted into NiFTI files and imported into Vivoquant (InVicro, Boston, MA, USA).Regions of interest (ROIs, tumors, and inguinal lymph nodes) were hand-drawn in VivoQuant and verified by a reviewer.The tumor and lymph node ROIs encompassed the entire tissue structure.MDSC recruitment was quantified by extracting frequency histograms of the R 2 * values within each 3D ROI.The histograms were converted from R 2 * values per voxel to cells per mm 3 using the calibration curves for SPIO-labeled MDSCs and ME-labeled MDSCs.All voxels in the ROI were summed.

Statistical Analysis
Analyses were done using GraphPad Prism 8 (San Diego, CA, USA).Linear regression was done to generate a line of best fit for R 2 * vs cellular concentration.A one-way ANOVA and student t-tests were used to evaluate whether there were any significant differences between ME-labeled MDSCs and SPIO-labeled MDSCs.

In Vitro Characterization of ME-Labeled Cells
Three different cell types were labeled with MEs using two potential loading ratios (MLRs) and evaluated iron (Fe) loading per cell (see Table 1).Increasing the MLR had a significant effect on Fe loading in MDSCs, causing a more than threefold increase in pg/cell when going from an MLR of 1000 to 2000.The Fe loading achieved with 2000 MLR was comparable to that achieved with SPIO.However, for CD8 cells, doubling MLR and increasing loading time did not increase Fe loading past 0.3 pg/cell, which is only a tenth of that seen in SPIO.For DCs, an MLR of 1000 led to a loading of 1.4 pg Fe per cell -significantly less than that  1 and 2).Clusters of MEs were visible inside DCs at MLRs of 1000 and 2000 (Fig. 1), and SPIO is also readily visible inside cells.MEs were also readily visible inside MDSCs (Fig. 2).ICC control images can be seen in Supplementary Fig. 1.
The MDSC suppression assay (Fig. 2B) demonstrated that MDSCs labeled with either SPIO (3rd column) or MEs using an MLR of 1000 (2nd column) retained comparable suppressive ability to unlabeled MDSCs, even at a wide range of Tresp:MDSC ratios.However, MDSCs labeled at an MLR of 2000 had reduced suppressive capacity, with T cell proliferation observed with all Treps:MDSC ratios.
Further characterization of ME-labeled cells focused on MDSCs with an MLR of 1000 due to their strong labeling and high purity and robustness in culture.The magnetic characteristics/contrast of SPIO-and ME-labeled MDSCs were evaluated using in vitro labeled MDSCs in tubes (Fig. 3).The R 2 * was plotted against cellular concentration in the tubes, and linear regression was done in Prism to generate a line of best fit.There was a significant non-zero slope (** indicates p < 0.05) for both SPIO-labeled MDSCs and ME-labeled MDSCs.SPIO-labeled MDSCs demonstrated a higher slope than ME-labeled MDSCs (0.03482 mm 3 /cells*s vs 0.01859 mm 3 /cells*s), which is consistent with the higher Fe loading observed in SPIO-labeled MDSCs as measured in our iron assays.

In Vivo Tracking of ME-Labeled MDSCs
MDSCs were isolated from the bone marrow of tumorbearing mice approximately 21 days post-C3 implant and cultured in vitro for 7 days (6 days for growth, last day for iron labeling).A subset of MDSCs was phenotyped using flow cytometry (Supplementary Fig. 2) prior to labeling with iron due to the potential for the rhodamine tag on SPIO to interfere with FACS labels.MDSCs were 100% CD11b + and were a mix of Ly6c int /Ly6g int and Ly6c hi / Ly6g lo , suggesting the presence of MDSC subpopulations.Bottom row: SPIO labeling of BM-DCs after a 20-h incubation with 0.6 mg/mL SPIO in media.Labels for bottom row: SPIO-rhodamine (red), phalloidin (green), DAPI (blue).Note that phalloidin stains the plasma membrane, while DAPI is a nuclear stain.All scales are 5 μm Cultured MDSCs were labeled with either SPIO or MEs and then injected i.v.into tumor-bearing mice approximately 23 days post-implant.Mice were imaged 24, 48, 72, 96, 120, and 192 h post-injection.Representative anatomical MR images with the R 2 * map overlaid are shown in Fig. 4 for both label types and three time points.Sample raw TurboSPI images pre-and postlabeled cell injections are shown in Supplementary Fig. 3.Both labeled cell types were present in tumors for up to 8 days postinjection.Labeled MDSCs were more densely recruited to the periphery of tumors, similar to what has been seen in previous studies 10 .Recruitment patterns were comparable between SPIOlabeled MDSCs and ME-labeled MDSCs.It is important to note that while the R 2 * scale is equivalent between the two labeled cell types, because of the different iron loading in each type, the Turbo SPI maps were used to estimate the cell recruitment based on known iron uptake in each cell type.The number of SPIO-labeled MDSCs appeared stable over the 8 days (Fig. 5A), consistent with SPIO rates seen anecdotally in our lab and previous studies 9,10 .Interestingly, the number of MElabeled MDSCs was also stable over time and consistently higher than the number of SPIO-labeled MDSCs present.While the number of ME-vs.SPIO-labeled cells was not significantly different on any given day, overall, using a mixed model analysis, the cell label had a significant effect (p = 0.0042) on the number of recruited cells over the whole data set.ME-labeled MDSC numbers were more variable between mice than those of SPIO-labeled MDSCs, particularly at the latest time points.
To confirm that labeled cells on MR images were, in fact, MDSCs, tumors were removed, sectioned, and analyzed with IHC.Both SPIO and ME-labeled cells were found to be co-localized with the Gr-1 (a composite of Ly6c/Ly6g antigens; Fig. 5B).

Discussion
The use of SPIO will always remain a limiting factor in MRI cell tracking during longitudinal studies due to the eventual degradation of the primary contrast agents.MEs have the potential to proliferate in vivo, avoiding this limitation (see further discussion below of issues affecting ME proliferation).The objective of this study was to assess the use of MEs as a potential contrast agent for MR imaging of immune cells.MEs have previously been studied in both cancer and stem cells [22,26,33], and isolated magnetosomes have been used for labeling DCs.However, to the best of our knowledge, no one has used MEs to label immune cells directly.Given that tracking of immune-based cell therapies is one of the primary goals for many MRI labs, it is essential to assess MEs' effects on immune cells.
Immune cells differ from other cells and between their own subtypes, in morphology and function, which can result in variable uptake of contrast agents.Therefore, this project began with three different immune cell types and Fig. 5 In vivo results of cell tracking of labeled MDSCs using MRI and IHC.A Quantitative results of cell densities in tumors as measured by TurboSPI.Error bars represent standard error.A mixed model ANOVA indicated that there was a significant effect due to the label used (p = 0.0042); however none of the individual days had significant differences due to labeling.B IHC results from tumor tissue taken post the 120-h time point.Donor cells were GFP + (green) and were labeled with Gr-1 (far red) and DAPI (blue).SPIO was conjugated to rhodamine (red), and MEs were labeled with Alexa 594 (red).White arrows indicate positively labeled Gr-1 + donor cells.Images were acquired at 100 × magnification, and scale bar is shown on the rightmost panel for each label type two different ME loading ratios (MLRs) at the in vitro level before choosing the most promising cell type and conditions to progress to in vivo experiments.
The smallest immune cells (5-7 μm), CD8 + T cells, which also have lower phagocytic activity, had much lower iron loading when labeled with MEs (0.22-0.3 pg/cell, both MLRs) compared to SPIO (3 pg).These results correspond well with other literature indicating average cell loading of 3-5 pg iron/cell when using SPIO [9,31].This level of iron loading in the MEs is typically on the low end of detectability in MRI cell tracking (ideally > 0.5 pg/cell), indicating that MEs may not be an optimal choice for this cell type, or that if they are used, more invasive methods of labeling may be necessary, such as transporation and transfection agents.
Both MDSCs and DCs were found to have much better loading from MEs, with MDSCs ranging from 0.63 to 2.2 pg iron/cell and DCs having approximately 1.7 pg/ cell.Notably, MDSCs, a normally suppressive cell type, demonstrated impaired suppression of T cell growth when labeled with a higher number of MEs (2000MLR).This may be due to pathogen-associated molecular pattern (PAMP) signaling induced by the MEs (which are bacterial) and can cause differentiation and/or maturation of MDSCs and DCs [34][35][36].Due to observed dysfunction at the higher MLR, it was decided to use an MLR of 1000 for the remaining experiments.Although both MDSCs and DCs had acceptable iron labeling (considered at a minimum, > 0.5 pg/cell), only the MDSCs were used for this in vivo study.Future studies will explore other cell types.
Previous studies using cardiomyocytes (approximately 100-150 μm × 20-25 μm in dimension), neural stem cells (< 12 μm diameter), and cancer cells (15-25 μm in diameter) indicated that it was possible to use much higher MLRs for these cell types [5,22,26,33].Given the range of sizes of these cells, it may not be purely a case of larger cells being able to hold more MEs.Also, when present in larger numbers, MEs may impact function in some cell types while minimally affecting others.Therefore, studies looking to use MEs should perform in vitro cell type-specific characterizations of cellular function with different MLRs before proceeding to in vivo studies.
It is important to note that this study does not assess the longevity of the MEs themselves.Due to the tumor model used, we were limited to an 8-10-day observation period before humane endpoint issues from tumor growth could occur.It is known from recent work [26] that, unfortunately, if unmodified, MEs will eventually end up in the phagocytic pathway of cell digestion.Research to modify the ME coating to eliminate or delay these phagocytic effects, thereby improving the longevity of MEs as a contrast agent, is ongoing by other groups [26] but is outside of the scope of this work.Interestingly, previous work has demonstrated that even when MEs themselves are undergoing degradation, the magnetosomes remain intact in live cells for considerably longer, with Lee et al. [26] demonstrating their presence in cancer cells at least 48 h post-injection (no data was acquired past that point) and McGinley et al. [33] demonstrating MR contrast in tissue 14-28 days post-injection, indicating the presence of iron crystals, if not intact magnetosomes.Given the MEs in this project are unmodified, it is possible that the MEs were undergoing degradation throughout the in vivo study, but that the magnetosomes remained intact, resulting in the MR contrast.
For in vivo studies, both ME-and SPIO-labeled MDSCs were detectable using TurboSPI.Both labeled cell types were generally clustered around the periphery of the tumor, which is common for immune cells in excluded tumor environments (as seen previously in this model [9,37]).While we have tested this method of quantifying cellular density before, it is very difficult to validate this in vivo, and therefore, these cellular densities are an estimation.We are working on testing potential validation methods for future studies.In our analysis technique, we do take the baseline R 2 * of these tumors into account (looking at the change of R 2 * only).Additionally, the TurboSPI sequence is specifically sensitive to encapsulated iron, as free iron will result in a large R 2 , which will not be refocused in the echo train used in TurboSPI.Cells with compartmentalized iron have R 2 * much bigger than R 2 , so there is remaining signal from these cells.
R 2 * values post-cell injection were similar for both tumors, but given the differences in loading between the two cell types, this resulted in differences (although not significant) between SPIO-and ME-labeled cells.Our method calculated higher cellular densities for ME-labeled cells at all time points compared to SPIO-labeled cells.This may be due to either biological effects or changes in the magnetic properties of cells.If the presence of MEs in MDSCs affected immune signaling in a way that we did not account for (our assessment was limited to suppression of T cells), it is possible that this could result in altered signaling/ recruitment for other MDSCs and/or immune cells in the tumor microenvironment.
There could also be changes in the amount or structure of iron, particularly over time, that we are not accounting for in our quantitative analysis.Our quantification relies on assumptions of iron in cells measured in vitro, but it is not clear how this may change in vivo.Accurate quantification of iron via standard laboratory techniques such as optical measurements or mass spectrometry is difficult due to the small cell numbers present in the tumors.We are currently working on studies attempting to better validate iron content in cells in vivo using other tumor models and inductively coupled plasma optical emission spectroscopy.

Table 1
Iron labeling results for immune cells achieved with SPIO.Fe loading of DCs at an MLR of 2000 was unable to be assessed due to sample volume limitations.ICC of ME-labeled cells demonstrated that MEs were present and localized within cells (Figs.