19F-MRI and 19F-MRS can identify specific cell types after in-vitro or in-vivo 19F-labeling. Knowledge on the potential to track in-vitro 19F-labeled immune cells in tumor models by 19F-MRI/MRS is scarce.
To study 19F-based MR techniques for in-vivo tracking of adoptively transferred immune cells after in-vitro 19F-labeling, i.e. to detect and monitor their migration non-invasively in melanoma-bearing mice.
Splenocytes (SP) were labeled in-vitro with a perfluorocarbon (PFC) and IV-injected into non-tumor bearing mice. In-vitro PFC-labeled ovalbumin (OVA)-specific T cells from the T cell receptor-transgenic line OT-1, activated with anti-CD3 and anti-CD28 antibodies (Tact) or OVA-peptide pulsed antigen presenting cells (TOVA-act), were injected into B16 OVA melanoma-bearing mice. The distribution of the 19F-labelled donor cells was determined in-vivo by 19F-MRI/MRS. In-vivo 19F-MRI/MRS results were confirmed by ex-vivo 19F-NMR and flow cytometry.
SP, Tact, and TOVA-act were successfully PFC-labeled in-vitro yielding 3x1011-1.4x1012 19F-atoms/cell in the 3 groups. Adoptively transferred 19F-labeled SP, TOVA-act, and Tact were detected by coil-localized 19F-MRS in the chest, abdomen, and left flank in most animals (corresponding to lungs, livers, and spleens, respectively, with highest signal-to-noise for SP vs TOVA-act and Tact, p<0.009 for both). SP and Tact were successfully imaged by 19F-MRI (n = 3; liver). These in-vivo data were confirmed by ex-vivo high-resolution 19F-NMR-spectroscopy. By flow cytometric analysis, however, TOVA-act tended to be more abundant versus SP and Tact (liver: p = 0.1313; lungs: p = 0.1073; spleen: p = 0.109). Unlike 19F-MRI/MRS, flow cytometry also identified transferred immune cells (SP, Tact, and TOVA-act) in the tumors.
SP, Tact, and TOVA-act were successfully PFC-labeled in-vitro and detected in-vivo by non-invasive 19F-MRS/MRI in liver, lung, and spleen. The portion of 19F-labeled T cells in the adoptively transferred cell populations was insufficient for 19F-MRS/MRI detection in the tumor. While OVA-peptide-activated T cells (TOVA-act) showed highest infiltration into all organs, SP were detected more reliably by 19F-MRS/MRI, most likely explained by cell division of TOVA-act after injection, which dilutes the 19F content in the T cell-infiltrated organs. Non-dividing 19F-labeled cell species appear most promising to be tracked by 19F-MRS/MRI.
Citation: Gonzales C, Yoshihara HAI, Dilek N, Leignadier J, Irving M, Mieville P, et al. (2016) In-Vivo Detection and Tracking of T Cells in Various Organs in a Melanoma Tumor Model by 19F-Fluorine MRS/MRI. PLoS ONE 11(10): e0164557. doi:10.1371/journal.pone.0164557
Editor: Bernhard Ryffel, Centre National de la Recherche Scientifique, FRANCE
Received: August 23, 2016; Accepted: September 2, 2016; Published: October 13, 2016
Copyright: © 2016 Gonzales et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper.
Funding: This work is supported in part by grants from the Swiss National Science Foundation (JS: grant no: 310030_144077; LH: grant no: 200020_152953), the Swiss Heart Foundation (JS: grant 2013) and the Oncosuisse (OM: grant no: KFS-3180-02-2013).
Competing interests: The authors have declared that no competing interests exist.
Cell tracking by magnetic resonance imaging (MRI) is an emerging method to visualize and monitor labeled cells after transplantation non-invasively and without the use of ionizing radiation. Recently, 19F-fluorine-MRI has been used to detect and track well-defined cell populations [1–7]. Because of the effective absence of 19F background signal in the body, any19F signal detected after injection of a 19F compound is unequivocally produced by this injected compound. As the MR signal is directly proportional to the amount of 19F nuclei present in the tissue, it can be related to a reference of known 19F concentration, rendering this technique quantitative [3, 4]. Moreover, these compounds are not limited by signal decay over time and therefore the time window for their detection can last several days. Finally, the 19F signal can be merged with conventional 1H-MRI images to identify its exact anatomic location and to add information on structure, function, and tissue characteristics. Direct IV injection of emulsions containing 19F-based perfluorocarbons (PFC) has been performed in different rodent models for angiography  and to detect non-invasively inflammation in myocardial infarction [5, 9], cerebral ischemia , myocarditis , pneumonia , atherosclerosis , arthritis  and tumors infiltrated by macrophages . Distinctively, defined cell populations such as dendritic cells , T cells [3, 4, 14, 15], or mesenchymal stem cells  were tracked non-invasively in rodents by 19F-MRI or 19F-MR spectroscopy (19F-MRS) after their in-vitro 19F-labeling. Recently, clinical 19F-MRI cell detection using labeling by PFC has also been described in patients with colorectal adenocarcinoma in order to detect autologous immunotherapeutic dendritic cells . This technique could therefore be applied to detect tumor cells as well as to monitor adopted cell transfer cancer therapies.
In recent years adoptive cell transfer therapies using ex-vivo activated T cells have undergone intensive testing [17, 18], and various types of T cells have been used for adoptive immunotherapy. It is essential to know whether the administered T cells reach their target and this is currently assessed by biopsies, which are invasive and not practical for all patients . Also, with a biopsy-based approach the total amount of T cells in a tumor, their distribution, and the kinetics of cell fluxes are difficult to assess. Non-invasive visualization of the trafficking of administered T cells could potentially allow one to predict responsiveness to these therapies. Therefore, a reliable non-invasive imaging method to monitor anti-tumor cell traffic is highly desirable. Moreover, as T cells with specific anti-tumor properties can migrate to and infiltrate tumor tissue by recognizing tumor antigens , they could, in principle, be used as a probe to detect tumor cells at metastatic sites when labeled with PFCs.
In the present study the migratory behavior of 3 different cell populations was tracked by means of non-invasive 19F-MRS and 19F-MRI and compared with invasive flow cytometry analyses and high-resolution in-vitro 19F-NMR. Initially, splenocytes (SP) were labeled in-vitro by a PFC to test the feasibility of non-invasive in-vivo tracking by 19F-MRS and 19F-MRI in control mice. SP represents a heterogeneous cell population comprising not only T cells (both CD8+ and CD4+, naïve, effector, memory and regulatory cells), but also B cells and antigen presenting cells (including dendritic cells, monocytes, macrophages and myeloid cells). The activated T cell populations, TOVA-act and Tact, whereas, are mostly CD8+ and these cytotoxic lymphocytes express one unique T cell receptor (TCR) called OT-1. To distinguish how the T cells were activated and expanded in-vitro, we named “Tact” the T cells that were stimulated with anti-CD3 and anti-CD28 antibodies, and “TOVA-act” the cells derived from single-cell suspensions of dissociated spleens stimulated with the specific OVA257-264 peptide. The OVA257-264 antigen was used as a tumor-specific antigen in the current study, and TOVA-act and Tact were produced from OT-1 mice expressing only the TCR OT-1 specific for Kb-OVA257-264 which is expressed at the surface of B16-OVA tumor grafted on recipient mice.
Splenic-derived OT-1 CD8+ T cells, stimulated either by OVA-peptide (= TOVA-act) or by anti-CD3 and anti-CD28 antibodies (= Tact) will expand and differentiate into various states including central memory (TCM), effector memory (TEM) and terminally differentiated, short-lived effector T cells (TE). Importantly, the newly activated T cells will also maintain a high state of proliferation for several days. While TE cells are typically found in peripheral tissue and provide a critical first line of defense to foreign antigen, TCM cells migrate to areas of secondary lymphoid organs, and compared to naive TN cells have a higher sensitivity to antigen stimulation. TEM tend to home to inflamed tissues, and have a more rapid effector function as compared to TCM . Activated tumor-antigen specific TOVA-act and Tact were labeled in-vitro (by the same PFC as used for SP) to test for non-invasive in-vivo tracking by 19F-MRS and 19F-MRI in mice bearing a B16-OVA tumor. Accordingly, the aim of the study was to develop a reproducible protocol for the in-vitro 19F-labeling of the three cell groups, to determine the detection limits of 19F-MRS and 19F-MRI for the in-vivo detection of these cells, and to test this application in B16-OVA tumor bearing mice.
Materials and Methods
B16-OVA melanoma cells (Ludwig Branch for Cancer Research, Lausanne) were maintained in Dulbecco's modified Eagle's medium (GIBCO Invitrogen, Grand Island, NY) supplemented with 10% heat-inactivated fetal bovine serum (GIBCO Invitrogen) and penicillin-streptomycin (GIBCO Invitrogen).
Abs and reagents
Fluorescent antibodies against mouse CD3, CD4, CD8, CD19, CD11b, CD44, anti-CD62L, PD-1, CD127, CD45.2, CD107a and IFNɣ were purchased from eBioscience (San Diego, CA). Purified anti-mouse CD3 and anti-mouse CD28 were from Biolegend (San Diego, CA). The OVA peptide (SIINFEKL) was produced by the Protein and Peptide Chemistry Facility located in the Department of Biochemistry of the University of Lausanne. FITC-conjugated or unconjugated 19F-based perfluorocarbons (PFC) were purchased from Celsense (Pittsburgh, PA).
All animal procedures were approved by the animal ethics committee (SCAV: Service de la Consommation et des Affaires Vétérinaires, Epalinges, Switzerland). All MR examinations were performed under ketamine-medetomidine anesthesia, and all efforts were made to minimize suffering. Mice were maintained under specific pathogen-free conditions. Ovalbumin-specific TCR transgenic (OT-1) mice were used to produce SP (described below). OT-1 mice were on a RAG1-/- background. CD45.1 C57BL/6 mice were used as recipients for adoptive transfer (described below). Ten days prior to adoptive transfer, tumors were implanted subcutaneously and dorsolaterally with inoculations of 106 B16-F10-OVA melanoma cells in 50 μl saline in CD45.1 C57BL/6 mice . The in-vivo protocol is depicted in Fig 1.
Overview of the time scale of the different experimental procedures. At day 0 (D0) eight CD45.1 C57BL/6 mice received 106 B16-F10 melanoma cells by subcutaneous injection in order to induce a malignant melanoma. On the same day SP were prepared from OT-1 mice and two different protocols were applied to generate Tact or TOVA-act (as described in Materials and Methods, T cells isolation and activation section). At day 8 (D8), PFC was added in the cell culture medium for 18h in order to label SP, Tact and TOVA-act with 19F. Then, at day 9 (D9) the 19F-labeled cells were injected IV: 2 control mice (with no tumors) received 50 x 106 SP, 3 mice received 20 to 50 x 106 Tact and 5 mice received 20 to 40 x 106 TOVA-act. Finally, 9 mice were imaged at day 10 (D10; 1 Tact injected mouse was not imaged) and all mice were immediately sacrificed for subsequent analysis of the organs (liver, lungs, spleen and tumor) by flow cytometry (all mice) and high resolution in-vitro NMR spectroscopy (2 SP injected mice, 3 TOVA-act injected mice and 1 Tact injected mouse). The study protocol was performed in a total of 10 animals. In black: in-vivo part; in blue: cell preparation.
T cell isolation and activation
Spleens from CD45.2+ OT1 mice were removed aseptically and homogenized by passing through a cell strainer (40μm). Red blood cells were lysed by the addition of a buffered ammonium chloride solution. The nucleated remaining cells (SP) were resuspended in complete medium (RPMI-1640 medium with 10% FBS, 100 μg/ml each of streptomycin and penicillin, 10 mM HEPES and supplemented with 2-mercaptoethanol) and two different protocols were applied to produce either Tact or TOVA-act. The SP population was used immediately after isolation for in-vitro FITC-conjugated or un-conjugated 19F-labeling and dead cells were eliminated with Ficoll (GE Healthcare) prior to injection into CD45.1+ C57BL/6 mice. Tact were obtained by stimulation of the SP population with recombinant murine IL2 (20 ng/ml), anti-mouse CD3 (500 ng/ml) and anti-mouse CD28 (0.1 μg/ml) for 2 days. TOVA-act were obtained by stimulation of the SP with the OVA peptide (SIINFEKL; 2 μg/ml) for 2 days in the presence of recombinant murine IL2 (20 ng/ml). In both protocols, two days after stimulation, clusters were formed and harvested to form a single-cell suspension. Dead cells were eliminated with Ficoll and cells were seeded in complete medium containing recombinant human IL15 at 20 ng/ml. Medium was changed every 2 days for one week.
SP, Tact, and TOVA-act were labeled in-vitro with Cell Sense (CS-1000), a 19F-based MR imaging agent. Cell Sense is an aqueous colloidal suspension (= nanoemulsion) of a perfluoropolyether perfluorocarbon polymer (PFC), having total fluorine content of 145 mg/mL (Celsense Inc., Pittsburg, PA, USA). The average nanoemulsion droplet size is 180 nm. It is formulated with excipients that facilitate PFC uptake into all cell types, regardless of their ability to phagocytose. The PFC used in Cell Sense is stable at low pH . SP, Tact, and TOVA-act were also labeled in-vitro with FITC conjugated PFC. In all conditions, PFC was added to the cell culture medium at a concentration of 10 mg/mL and incubated with the SP, Tact, or TOVA-act for 18 hours at 37°C, 5% CO2 (SP n = 5; Tact n = 14; TOVA-act n = 10). After this incubation period, the cells were washed three times with PBS and counted.
High resolution in-vitro 19F-NMR spectroscopy of labeled cells
In order to measure the mean 19F content present in the cells after labeling, quantitative 19F NMR measurements were performed in lysed cell pellets. A known number of labeled cells (~3x106) were spun down, resuspended in 250 μl of 1% Triton X100 v/v in PBS to lyse the cells. The cell lysate was mixed with 250 μl of a calibrated 19F reference solution, trifluoroacetic acid (TFA) at 0.1% v/v in D2O, and placed in a 5 mm NMR borosilicate tube. The 19F NMR measurements were performed using a Bruker AVANCE III HD 400 MHz (9.4 T) NMR spectrometer (Bruker BioSpin AG, Fällanden (ZH), CH). The average 19F-fluorine content per cell was calculated from the ratio of the integrated areas of the TFA and PFC 19F spectra, normalized to the total cell number in the lysate. PFC 19F spectra, acquired with 256 scans and processed with a line broadening of 5 Hz, contain several peaks with a major one located at -93 ppm and the TFA peak at -75 ppm. These two peaks were used for quantitative calculations.
Functional assay of PFC-FITC labeled cells
One hundred thousand SP, Tact, and TOVA-act were cocultured with 50x105 B16-OVA tumor cells in the presence of anti-CD170a antibody and Golgi Stop reagent (BD Biosciences, San Jose, CA). After 5 hours of incubation at 37°C, cells were washed and stained with fluorescent anti-CD3 and anti-CD8 antibodies at 4°C for 20 minutes. Following fixation and permeabilization, the cells were stained with anti-IFNɣantibody and analyzed on a FACS LSRII (BD Biosciences) and BD FACS Diva software.
Cytotoxicity assay of PFC-FITC labeled cells
Fifty thousand SP, Tact, and TOVA-act were co-cultured with 25x105 B16-OVA tumor cells in the presence of Cytotox red reagent (Essen Bioscience, Ann Arbor, Michigan), according to the manufacturer’s instructions. Images were acquired every 2 hours with the Incucyte Zoom System (Essen Bioscience) and analyzed with its software.
Different dilutions of TFA (0.6M, 0.5M, 0.4M, 0.3M, 0.2M, 0.1M) in 0.3M NaCl were prepared in agarose gel for imaging. 1H images were acquired with a 9.4T spectrometer (Varian, Palo Alto, CA) using a gradient echo sequence (repetition time (TR) 13.2 ms, echo time (TE) 2.4 ms, signal averages 8, matrix 128×128, field of view 18×18 mm2, 3 slices with a slice thickness of 2 mm, total acquisition time 13.5 s). Next, for the 19F acquisitions at the same locations of the 1H images, a fast spin echo sequence was used with TR = 500 ms, TE = 3.7 ms; echo-train length 4, signal averages 960 scans, matrix 32x32, field of view 18×18mm2, slice thickness 2 mm, and a total acquisition time of 64 minutes.
Adoptive transfer of T cells
The study protocol is shown in Fig 1. PFC labeled SP-CD45.2+ (50 ×106) suspended in 250 μl NaCl solution were injected i.v. into two control (no tumor implanted) recipient CD45.1+ C57BL/6 mice. PFC labeled Tact or TOVA-act (CD45.2+; 20 to 50 ×106) suspended in 250 μl NaCl solution were injected i.v. into recipient CD45.1+ C57BL/6 mice that were implanted with B16-F10-OVA melanoma cells 9 days before. The injection of CD45.2+ immune cells into CD45.1+ C57BL/6 recipient mice allows discriminating injected immune cells from host cells.
1H and 19F-MRI
The day after adoptive transfer (24h to 36h post injection, Fig 1) mice were anesthetized with intraperitoneal injection of ketamine: medetomidine (75 mg/kg: 0.1 mg/kg). This anaesthetic combination was chosen to avoid any isoflurane 19F-MR background signal resulting from its accumulation in the fat pads . The body temperature was monitored with a rectal probe (SA Instruments, Stony Brook, NY) and kept constant at 37.0°C by using tubing with circulating warm water. The animals were placed under a custom-designed 18-mm diameter quadrature surface coil tunable to both the 1H and 19F frequencies (400.2 and 376.6 MHz, respectively). To acquire coil-localized spectra of 19F (128 scans) the coil was positioned at 4 different places, the chest, abdomen, left flank, and the right thigh to cover primarily the liver, the lungs, the spleen, and the tumor, respectively.
In 3 mice, the 19F spectroscopic signal was deemed sufficient for 19F-MRI (signal-to-noise ratio (SNR) >200). In these mice a stack of 6 axial 1H images of the liver was acquired with a gradient echo sequence (repetition time (TR) 29.7 ms, echo time (TE) 1.9 ms, signal averages 4, matrix 128×128, field of view 30×30 mm2, slice thickness 2 mm, total acquisition time 15.2 s). Next, a stack of axial 19F images was acquired at the identical position as the 1H images using a fast spin echo sequence (TR = 500 ms, TE = 3.7 ms; echo-train length 4, signal averages 960, matrix 16x16, field of view 30×30mm2, slice thickness 2 mm, total acquisition time 32 minutes).
Immediately after the MRI session the mice were euthanized by cervical dislocation to harvest the liver, lungs, spleen and tumor. A single-cell suspension was then prepared from the different organs using a cell strainer (70 μm) and RPMI medium. The different cell suspensions were washed once with RPMI and then split into two groups with 5% of the cell suspension used for flow cytometry analyses and 95% for high resolution ex-vivo spectroscopy 19F-NMR analyses.
High resolution ex-vivo 19F-NMR spectroscopy of excised organs
The 19F NMR measurements were performed on the cell suspensions prepared from the different organs described above. Ninety-five percent of the cell suspensions from liver, lungs, spleen and tumor were centrifuged and then resuspended into 250 μl of 1% Triton X100 v/v in PBS to lyse the cells. The cell lysates were then mixed with 250 μl of TFA 0.1% v/v in D2O (calibrated 19F reference solution), and placed in a 5 mm NMR borosilicate tube. The acquisition method used was described previously in the “High resolution in-vitro 19F-NMR spectroscopy of labeled cells” section.
In vitro labeling and function of immune cells
The 3 cell groups were in-vitro labeled or not with FITC conjugated or unconjugated 19F -PFC in order to assess cell labeling efficiency, and to compare cell viability, phenotype and T cell function. After 18 hours of incubation with PFC the cells were stained by Trypan blue exclusion assay to evaluate the potential cytotoxicity due to labeling. For the 3 cell groups (SP, Tact and TOVA-act), the amount of dead cells after PFC incubation was comparable to the untreated control condition (difference when compared to untreated cells of the same type: 0.5%, 7% and 3% for SP, Tact, and TOVA-act, respectively). This result shows that the PFC-based protocol safely labels these cells in-vitro. Moreover, PFC labeling does not affect the proportion of cell populations, no difference was observed after FITC-conjugated PFC staining (Fig 2). SP are composed of ~25% of CD3+ T cells (Fig 2A), ~55% of CD19+ non-T cells (Fig 2B), ~5% of CD11b+ non-T cells (Fig 2C), whereas Tact and TOVA-act are composed of only CD3+ T lymphocytes (Fig 2A). In CD3+ T cells, almost all cells are CD8+ T cells (Fig 2A) because they are derived from transgenic OT-1 mice.
(A) Percentage of CD3+ T cells in the total population and proportion of CD8+ T cells in CD3+ population. (B) and (C) Percentage of CD19+ and CD11b+ cells in the total population. (n = 3 for each group of cells)
Fig 3 depicts the 19F content (i.e. the number of 19F atoms per cell, for the SP, Tact, and TOVA-act cells) after 18h of incubation with the PFC agent quantified by high-resolution ex-vivo 19F NMR. The mean 19F content per cell was similar for the 3 cell groups (Fig 3, overall-p = 0.72).
19F content (number of 19F atoms per cell) in SP, Tact and TOVA-act after incubation with PFC for 18h. Cellular loading was determined in a known number of cells by using an internal reference (TFA) with a known number of 19F atoms.
In the SP group, the non-T cells take up the most PFC (44.93% ± 3.76% of CD3-PFC+ vs 3.70% ± 0.21% of CD3+PFC+, n = 3, Fig 4A). In the Tact and in TOVA-act groups the percentage of PFC+ cells are 13.70% ± 0.61% and 28.67% ± 7.30%, respectively, n = 3 (Fig 4A). Non-T cells were preferentially labeled by PFC and TOVA-act presented a 2-fold higher labeling than Tact. In the SP group, the majority of lymphocytes are naïve T cells (54,96% ± 5.91% of CD62L+CD44- cells in CD3+ cells, n = 3) with weak expression of PD-1, a major inhibitory receptor regulating T-cell exhaustion (Fig 4B). After stimulation, Tact and TOVA-act present the phenotype of central memory (CD44+CD62L+) and effector memory (CD44+CD62L-) cells (Fig 4B). The proportion of these 2 populations is different in Tact and TOVA-act. OVA-peptide stimulation preferentially induces an effector memory (62.80% ± 6.25%, n = 3) rather than a central memory phenotype (34.13% ± 5.85%, n = 3). Conversely, antibody stimulation leads to a higher central memory (50.73% ± 5.29, n = 3) than effector memory phenotype (43.33% ± 4.72, n = 3). Moreover OVA-peptide stimulation is stronger than antibody stimulation in our experimental conditions as it induces an overexpression of PD-1 (88.83% ± 0.82%, n = 3) compared to Tact cells (46.63% ± 3.62%, n = 3). All of these phenotypic differences may impact PFC uptake.
(A) PCF labelling of total splenocytes (SP), anti-CD3/anti-CD28 activated (Tact) and Ova-peptide activated (TOVA-act) splenocytes from OT-1 cells. (B) Phenotype of CD3+ T cells: CD3+CD8+ T cells were analyzed by flow cytometry for CD44, CD62L, CD127 and PD-1 expression. By this flow-cytometric analysis, the following phenotypes could be identified in the CD3+CD8+ T-cell population: CD62L+CD44- (naive T cells), CD62L+CD44+ (memory T cells) and CD62L-CD44+ (effector T cells). n = 3 for each group of stimulation. Data are representative of 3 independent experiments.
Finally to determine the impact of PFC on T cell function, we performed a series of experiments before and after PFC-labeling. As expected, SP showed a weak response due to the small proportion of CD3+ T lymphocytes (Fig 5). PFC-labeling induced a decrease of response in CD107a upregulation (Fig 5A) and IFNɣ (Fig 5B) secretion assays, but did not impact the cytotoxic capacity of T cells (Fig 5C). Hence, PFC-labeled T cells are able to recognize and kill their target.
CD107a upregulation (A) and IFNɣ production (B) by CD3+CD8+ T cells were analyzed after co-culture of corresponding cells with B16-OVA mouse melanoma cells, used as antigen presenting cells. (C) Cytotoxicity of corresponding cells in same conditions was measured by the incorporation of red reagent in dead cells. The acquisition was done by Incucyte microscope. (n = 3 for each group of stimulation)
Limit of detection of 19F-PFC-labeled immune cells by 19F-MRI
In order to determine the limit of detection of the method, a phantom experiment was performed using different TFA dilutions (Fig 6A and 6B). Both 1H and 19F images were acquired for each dilution (Fig 6C). Under these conditions the limit of detection for 19F-MRI was 1.5 x 1017 19F spins (at a SNR level of 3), which would correspond to 150’000 cells per voxel of 0.63 mm3 assuming a cell labeling of 1012 19F atoms/cell (Fig 6E) corresponding to 238’000 cells per μl.
(A) Concentration in M of TFA per well. (B) Equivalent number of 19F atoms per well. (C) Axial view of the tubes. Upper panels: 19F images acquired; lower panels: 19F and 1H overlaid images. Voxel volume: 0.63mm3. (D) Coronal view of the tubes, 1H images. (E) Mean SNR in the region of interest (ROI) with respect to 19F atoms/voxel. Each line corresponds to the acquisition of a slice as shown in panel D.
In-vivo detection of 19F-MRI signal
In order to follow the migration of the injected immune cells in-vivo, 19F-MRS was performed in different anatomic areas (chest, abdomen, left flank and right thigh) of the mice injected with 19F labeled SP, Tact, and TOVA-act cells. These acquisitions were performed 24h after adoptive cell transfer. Fig 7 shows examples of 19F coil-localized spectra and Fig 8A. No signal was detectable in the right thigh (corresponding to the tumor area) whereas 19F signal was detectable in the chest, abdomen, and left flank of most of the animals, corresponding primarily to the lungs, the liver, and the spleen, respectively. The 19F signal measured in the abdomen and in the left flank was significantly higher in the SP injected group compared to the other groups (one way ANOVA, p = 0.0083 and p = 0.0076 respectively, Fig 8).
Representative 19F spectra of coil-localized spectroscopy acquired in-vivo in 4 different regions (chest, abdomen, left flank, and right thigh) of the mice injected with SP, Tact and TOVA-act. Highest signals were measured in the chest and the abdomen while no reliable signal was detected in the right thigh. 19F-MRI of SP and Tact injected animals are depicted in Fig 9B (middle and lower panels).
(A) SNR (mean ±SD) of 19F coil-localized spectroscopy acquired in the four different regions of the mice. In SP injected mice higher 19F-signals were observed compared to the Tact and TOVA-act injected mice both in the abdomen and the left flank regions (p<0.05, one way ANOVA). (B) Ex-vivo 19F content (mean ±SD) in liver, lung, and spleen. No 19F signal was detected in the tumor. The internal reference (TFA) contained a known number of 19F atoms. There is a higher 19F load in the liver in SP injected compared to TOVA-act injected mice (p = 0.05: all other comparisons are non-significant). Note logarithmic scale of y-axis in B.
Fig 9 shows representative images of 19F and 1H-MRI overlays with a strong 19F-signal in the liver of 3 animals, i.e. in which the spectroscopic 19F signal was deemed sufficient for 19F-MRI (SNR >200). As anesthesia duration was limited, the abdomen was the only area imaged.
(A) Image showing the position of the surface coil that covers the abdomen (liver and part of the lungs) used for 19F-MRS, 19F-MRI, and 1H-MRI. (B) Representative axial MRI of mice that received either SP (upper and middle panels) or Tact (lower panels). The livers can clearly be identified. The 19F-MRS spectra of these animals (SP middle panels and Tact lower panels) are illustrated in Fig 4. No images were acquired for mice injected with TOVA-act. 1H-MRI in grey and 19F-MRI in orange. The total 19F content in the liver of the animals measured by ex-vivo 19F-MRS was 1.5 x 1018 and 1.2 x 1018 after SP injection, and 1.15 x 1018 after Tact injection (compare also Fig 8B).
Ex-vivo high resolution 19F-NMR spectroscopy
A post-mortem in-vitro quantitative analysis of 19F-NMR spectra of different organs (liver, lungs, spleen, tumor) was performed as a reference for 19F organ content after administration of 19F labeled cells. The 19F content of the different homogenized organs is depicted in Fig 8B. 19F signal was consistently measured in the liver (in 6 out of 7 animals) as well as in the lungs (4 of 7) and occasionally in the spleen (1 of 7). No 19F signal was observed in the tumors of these animals. Moreover, the 19F measured signal was higher in SP injected mice compared to TOVA-act injected mice (p<0.0001 in liver and not significant in lungs, ANOVA).
In order to confirm the label tracking results from in-vivo 19F-MRS, in-vivo 19F-MRI, and ex-vivo 19F-NMR, the distribution of the donor T cells in the different organs was determined by flow cytometry according to their expression of CD3, CD8 and CD45.2. There was massive infiltration and proliferation of donor cells in a variety of peripheral tissues, including liver, spleen, and lungs (Fig 10). A small population of these donor cells (CD3+ CD8+ CD45.2+) was also found in the tumors, with no significant difference between the 3 cell types. However, contrary to what was observed with in-vivo and ex-vivo 19F-analyses, the amount of adoptively transferred SP was very low in all the organs analyzed and the amount of adoptively transferred TOVA-act more abundant in the different organs compared to either SP or Tact, although not to a significant degree (liver: p = 0.1313; lungs: p = 0.1073; spleen: p = 0.109).
(A) Amount of (CD3+; CD8+; CD45.2+) cells in the liver, lungs, spleen, and tumor. Cell suspensions from the organs were stained with monoclonal antibodies for CD3 and CD8 together with monoclonal antibody for CD45.2 to identify adoptively transferred cells. Gating was performed on CD3+; CD8+ cells with subsequent analysis of CD45.2+ (to differentiate the host cells (CD45.1+) from transplanted cells (CD45.2+)). Data are expressed in CD3+; CD8+; CD45.2+ cells as percent of gated cells. (B) Data from one representative Tact injected mouse are shown for the different organs analyzed.
In this study 19F-PFC was used to in-vitro label SP and activated T cells and to follow their migration in-vivo in B16-OVA-melanoma bearing mice using 19F-MRS and 19F-MRI.
In-vitro labeling of immune cells and the detection threshold by 19F-MRI
SP, Tact, and TOVA-act were successfully labeled in-vitro, achieving similar 19F content per cell in the 3 populations ranging from 3 x 1011 to 1.4 x 1012 atoms/cell. Labeling of activated T cells is consistent with data published by Srinivas et al. reporting a 19F loading per cell of 1.7 ± 0.9 × 1012 19F/cell . In the present study there are, however, some differences compared to the study of Srinivas. We used a commercially available PFC, while Srinivas et al. used a perfluorinated polyether emulsion prepared in their own lab, which required 3 days of incubation to label cells, in contrast to the 18h-incubation in the present protocol. Importantly, the labeling procedure used in the current work did not affect cell viability.
With an in-vitro phantom experiment we determined the detection threshold for 19F-MRI for in-vitro 19F-PFC-labeled cells. Approximately 150’000 cells with an assumed 19F loading of 1012 atoms/cell are detectable in a minimal voxel volume of 0.63mm3 (measured over 34 minutes at 9.4T at an SNR of 3).
In-vivo and ex-vivo detection of in-vitro 19F-labeled immune cells by 19F-MRS and 19F-MRI
Initially, 19F-PFC labeled SP were tested in non-tumor bearing mice as a proof of concept. Using in-vivo 19F-MRS we detected the donor cells in the area of the abdomen, the chest, and the left flank corresponding mainly to the liver, the lungs, and the spleen, respectively (Fig 8A). These in-vivo data were confirmed post-mortem by in-vitro quantitative 19F-NMR, yielding a similar distribution of 19F signals as shown in Fig 8B. Then, in a next step, 19F-labeled Tact and TOVA-act immune cells were adoptively transferred into B16-OVA tumor-bearing mice. As with SP cells, donor Tact and donor TOVA-act were detected by in-vivo 19F-MRS in the areas of the liver, lungs, and spleen (Fig 8A) and these data were confirmed by ex-vivo quantitative 19F-NMR measurements (Fig 8B). As for in-vivo 19F-MRS, the post-mortem 19F-NMR detected highest 19F quantities in SP injected animals with most frequent positive findings in livers (6 of 7 animals) followed by lungs (4 of 7 animals) and the spleen (1 of 7 animals). Thus, in-vivo 19F-MRS allows for cell detection in agreement with the true 19F distribution in these organs. In addition, the same donor cells were successfully imaged in-vivo by 19F-MRI in the livers of 3 mice.
Our data show that following IV injection of in-vitro 19F-labeled cells, the majority of injected cells are trapped in the liver and the lungs (Fig 8B). It has previously been shown that a large fraction of IV-injected CD8+ T cells preferentially migrate into the interstitium of normal lungs . In fact, that study suggested that peripheral homing and retention of CD8+ T cells in the respiratory tract is a mechanism to ensure an adequate number of memory T cells being available at the site of potential future respiratory tract infections.
Using in-vivo 19F-MRS we detected the donor cells in the area of the abdomen (corresponding mainly to the liver), the chest (corresponding mainly to the lungs and a portion of the liver) and the left flank (corresponding to the spleen and a portion of the liver) (Fig 8B). While MRS is most sensitive for 19F signal detection, it yields only limited spatial information. However, if sufficient 19F is brought into the target tissue, MRI is able to detect and image 19F-labeled SP and Tact, as exemplified in Fig 9.
SP and T cell behavior after IV injection
While ex-vivo high-resolution 19F-NMR was used to determine the true 19F atom content of the organs, flow cytometry was employed to determine the donor cell distribution in the animals. With this technique, the adoptively transferred cells were identified in the liver, lungs, and the spleen as with 19F-MRS, but flow cytometry also detected the donor cells in the tumor. As SP with a percentage of 0.06% ± 0.05% of donor cells (CD3+; CD8+; CD45.2+) measured by flow cytometry were detected in the liver by 19F-MRI, one would expect to also image Tact and TOVA-act as they were observed in the organs at a considerably higher percentage than SP (i.e. 0.07% to 0.71%, for Tact 0.35% to 2.94% for TOVA-act in the various organs; Fig 10A). However, 19F was detected by 19F-MRI in only one Tact treated mouse. This low rate of detection could be explained by the fact that following both antibody (against CD3 and CD28) and peptide stimulation, both Tact and TOVA-act have the capacity to divide rapidly in-vivo. This proliferation of T cells in-vivo would then induce a subsequent dilution of the PFC content in the daughter cells. This dilution of 19F signal in the daughter cells can be verified in the present study by comparing the ex-vivo high resolution spectroscopy data to flow cytometry data.
The flow cytometry data also demonstrate that TOVA-act have a better capacity than Tact and SP to infiltrate the different organs and the tumor. Both, Tact and TOVA-act were produced from OT-1 mice and therefore, both T cells are expected to present predominantly the TCR that recognizes the OVA antigen. Nevertheless, flow cytometry in Fig 10A and 10B demonstrates a trend towards higher infiltration of tumors by TOVA-act. In line with the high presence of TOVA-act in the liver, lungs, spleen, and tumor, they were detected by 19F-MRS but not at a 19F signal level that allowed for imaging. This is in contrast to SP, which are detectable by imaging even with their low presence in these organs. The SP represent a heterogeneous population. Besides T and B cells they contain dendritic cells and macrophages, both of which are phagocytic and can accumulate 10 to 1000 times more 19F compared to T cell populations . Also, as these phagocyte populations are terminally differentiated cells, they do not divide, unlike activated T cells. 19F-MRI was performed 24h after adoptive cell transfer, A division cycle every 8 hours of the PCF-labeled activated T cells e.g. could already reduce the cellular 19F content by a factor of 8 at the time of imaging. Taken together, this could explain why despite the relatively high number of TOVA-act and Tact detected by flow cytometry in the target organs, a diminished 19F loading was present in these organs, which limited their detection by 19F-MRS and 19F-MRI.
From these data, it might be speculated that non-dividing cells would be best for this type of tracking, such as SP or highly-differentiated T cells e.g. killer cells. Another alternative would be to inject a higher number of 19F-labeled cells to compensate for dilution according the cell division rate.
Limitations of the study and strategies to improve the detection limit of 19F-labeled T cells in the tumor tissue
Possible ways to increase the sensitivity of the 19F-MRI method to detect PFC-labeled T cells in the tumor could be to: 1) increase the 19F-label content of the injected cells by further optimizing the in-vitro T cell PFC-labeling procedure, 2) enrich the proportion of 19F-labeled T cells in the injected cell populations (e.g. by sorting PFC-FITC labeled cells by flow cytometry), 3). use preferentially non-dividing T cells to minimize the label dilution effect caused by cell division (which was the most likely reason in our study preventing T cell detection in the tumors), 4) use modified PFCs with shorter T1 (e.g. by gadolinium-coupling), 5) exploit higher magnetic field strength, 6) develop high performance 19F coils, and 7) exploit emerging fast pulse sequences. For example, combining this 19F-MRI technique with compressed sensing could be advantageous with regard to shortening the acquisition time. Compressed sensing was already applied successfully for 19F-MRI by Zhong and co-workers, but this pulse sequence was not available on our 9.4T system .
A potential limitation of the study was the rather rigorous threshold of >200 SNR of the 19F-spectroscopic signal to proceed to 19F-MRI. As non-localized spectroscopy yields the entire signal of the volume within the coil, a spectroscopic signal below this threshold does not exclude the possibility for locally high 19F concentrations that would allow for 19F-MRI. This notion is supported by the liver 19F-images after PFC-labeled Tact injection, which demonstrate a non-uniform signal distribution in the liver (Fig 9B).
Immune cells, including total SP and activated T cells can be successfully labeled in-vitro by 19F-PFC and the T cells maintain the capacity to detect and kill the tumor cells after 19F-labeling. They are detectable after IV administration by in-vivo coil-localized 19F-MRS in liver, lungs, and spleen. IV-injected SP can also be imaged by in-vivo 19F-MRI while this is more difficult for T cells. In particular, the proportion and/or 19F content of the injected 19F-labeled T cells was too low to allow for tumor imaging. Flow cytometry of liver, lungs, spleen, and tumor demonstrates a higher number of TOVA-act and Tact than SP in these organs. The difficulty in reliably detecting 19F-labeled TOVA-act and Tact in flow cytometry-positive organs by 19F-MRS could be explained by TOVA-act and Tact proliferation, which would dilute the 19F signal in the daughter cells and thus, in the target organs. Non-dividing in-vitro 19F-labeled cell species appear most promising to be tracked by 19F-MRS and/or 19F-MRI.
We are grateful to Carola J. Romero, Corina M. Berset, Anne-Catherine Clerc and Caroline Poisson for their technical assistance.
- Conceptualization: CG HAIY ND JL MI PM LH OM JS.
- Formal analysis: CG HAIY ND.
- Funding acquisition: LH OM JS.
- Investigation: CG HAIY ND JL MI PM.
- Methodology: CG HAIY ND JL MI PM LH OM JS.
- Project administration: CG HAIY MI LH OM JS.
- Resources: MI LH OM JS.
- Software: HAIY.
- Supervision: MI LH OM JS.
- Validation: CG HAIY ND PM.
- Visualization: CG HAIY ND JS.
- Writing – original draft: CG HAIY ND MI JS.
- Writing – review & editing: CG HAIY ND JL MI PM LH OM JS.
- 1. Ahrens ET, Flores R, Xu H, Morel PA. In vivo imaging platform for tracking immunotherapeutic cells. Nature biotechnology. 2005;23(8):983–7. doi: 10.1038/nbt1121 pmid:16041364.
- 2. Schwitter J. Extending the frontiers of cardiac magnetic resonance. Circulation. 2008;118(2):109–12. doi: 10.1161/CIRCULATIONAHA.108.790139 pmid:18606926.
- 3. Kadayakkara DK, Beatty PL, Turner MS, Janjic JM, Ahrens ET, Finn OJ. Inflammation driven by overexpression of the hypoglycosylated abnormal mucin 1 (MUC1) links inflammatory bowel disease and pancreatitis. Pancreas. 2010;39(4):510–5. doi: 10.1097/MPA.0b013e3181bd6501 pmid:20084048; PubMed Central PMCID: PMC2859977.
- 4. Srinivas M, Turner MS, Janjic JM, Morel PA, Laidlaw DH, Ahrens ET. In vivo cytometry of antigen-specific t cells using 19F MRI. Magnetic resonance in medicine. 2009;62(3):747–53. doi: 10.1002/mrm.22063 pmid:19585593; PubMed Central PMCID: PMC2763624.
- 5. Flogel U, Ding Z, Hardung H, Jander S, Reichmann G, Jacoby C, et al. In vivo monitoring of inflammation after cardiac and cerebral ischemia by fluorine magnetic resonance imaging. Circulation. 2008;118(2):140–8. doi: 10.1161/CIRCULATIONAHA.107.737890 pmid:18574049; PubMed Central PMCID: PMC2735653.
- 6. van Heeswijk RB, De Blois J, Kania G, Gonzales C, Blyszczuk P, Stuber M, et al. Selective in vivo visualization of immune-cell infiltration in a mouse model of autoimmune myocarditis by fluorine-19 cardiac magnetic resonance. Circulation Cardiovascular imaging. 2013;6(2):277–84. doi: 10.1161/CIRCIMAGING.112.000125 pmid:23343515.
- 7. Ahrens ET, Helfer BM, O'Hanlon CF, Schirda C. Clinical cell therapy imaging using a perfluorocarbon tracer and fluorine-19 MRI. Magnetic resonance in medicine. 2014;72(6):1696–701. doi: 10.1002/mrm.25454 pmid:25241945; PubMed Central PMCID: PMC4253123.
- 8. van Heeswijk RB, Pilloud Y, Flogel U, Schwitter J, Stuber M. Fluorine-19 magnetic resonance angiography of the mouse. PloS one. 2012;7(7):e42236. doi: 10.1371/journal.pone.0042236 pmid:22848749; PubMed Central PMCID: PMC3407132.
- 9. Ye YX, Basse-Lusebrink TC, Arias-Loza PA, Kocoski V, Kampf T, Gan Q, et al. Monitoring of monocyte recruitment in reperfused myocardial infarction with intramyocardial hemorrhage and microvascular obstruction by combined fluorine 19 and proton cardiac magnetic resonance imaging. Circulation. 2013;128(17):1878–88. doi: 10.1161/CIRCULATIONAHA.113.000731 pmid:24025595.
- 10. Ebner B, Behm P, Jacoby C, Burghoff S, French BA, Schrader J, et al. Early assessment of pulmonary inflammation by 19F MRI in vivo. Circulation Cardiovascular imaging. 2010;3(2):202–10. doi: 10.1161/CIRCIMAGING.109.902312 pmid:20061515; PubMed Central PMCID: PMC3138443.
- 11. van Heeswijk RB, Pellegrin M, Flogel U, Gonzales C, Aubert JF, Mazzolai L, et al. Fluorine MR Imaging of Inflammation in Atherosclerotic Plaque in Vivo. Radiology. 2015;275(2):421–9. doi: 10.1148/radiol.14141371 pmid:25496216.
- 12. Flogel U, Burghoff S, van Lent PL, Temme S, Galbarz L, Ding Z, et al. Selective activation of adenosine A2A receptors on immune cells by a CD73-dependent prodrug suppresses joint inflammation in experimental rheumatoid arthritis. Science translational medicine. 2012;4(146):146ra08. doi: 10.1126/scitranslmed.3003717 pmid:22875828.
- 13. Balducci A, Wen Y, Zhang Y, Helfer BM, Hitchens TK, Meng WS, et al. A novel probe for the non-invasive detection of tumor-associated inflammation. Oncoimmunology. 2013;2(2):e23034. doi: 10.4161/onci.23034 pmid:23526711; PubMed Central PMCID: PMC3601170.
- 14. Janjic JM, Srinivas M, Kadayakkara DK, Ahrens ET. Self-delivering nanoemulsions for dual fluorine-19 MRI and fluorescence detection. Journal of the American Chemical Society. 2008;130(9):2832–41. doi: 10.1021/ja077388j pmid:18266363.
- 15. Srinivas M, Morel PA, Ernst LA, Laidlaw DH, Ahrens ET. Fluorine-19 MRI for visualization and quantification of cell migration in a diabetes model. Magnetic resonance in medicine. 2007;58(4):725–34. doi: 10.1002/mrm.21352 pmid:17899609.
- 16. Gaudet JM, Ribot EJ, Chen Y, Gilbert KM, Foster PJ. Tracking the fate of stem cell implants with fluorine-19 MRI. PloS one. 2015;10(3):e0118544. doi: 10.1371/journal.pone.0118544 pmid:25767871; PubMed Central PMCID: PMC4358825.
- 17. Dudley ME, Wunderlich JR, Robbins PF, Yang JC, Hwu P, Schwartzentruber DJ, et al. Cancer regression and autoimmunity in patients after clonal repopulation with antitumor lymphocytes. Science. 2002;298(5594):850–4. doi: 10.1126/science.1076514 pmid:12242449; PubMed Central PMCID: PMC1764179.
- 18. Robbins PF, Morgan RA, Feldman SA, Yang JC, Sherry RM, Dudley ME, et al. Tumor regression in patients with metastatic synovial cell sarcoma and melanoma using genetically engineered lymphocytes reactive with NY-ESO-1. Journal of clinical oncology: official journal of the American Society of Clinical Oncology. 2011;29(7):917–24. doi: 10.1200/JCO.2010.32.2537 pmid:21282551; PubMed Central PMCID: PMC3068063.
- 19. Manzo T, Heslop HE, Rooney CM. Antigen-Specific T Cell Therapies for Cancer. Human molecular genetics. 2015. doi: 10.1093/hmg/ddv270 pmid:26160910.
- 20. Arsenio J, Kakaradov B, Metz PJ, Kim SH, Yeo GW, Chang JT. Early specification of CD8+ T lymphocyte fates during adaptive immunity revealed by single-cell gene-expression analyses. Nature immunology. 2014;15(4):365–72. doi: 10.1038/ni.2842 pmid:24584088; PubMed Central PMCID: PMC3968536.
- 21. Overwijk WW, Restifo NP. B16 as a mouse model for human melanoma. Current protocols in immunology / edited by Coligan John E [et al]. 2001;Chapter 20:Unit 20 1. doi: 10.1002/0471142735.im2001s39 pmid:18432774; PubMed Central PMCID: PMC2763508.
- 22. Helfer BM, Balducci A, Nelson AD, Janjic JM, Gil RR, Kalinski P, et al. Functional assessment of human dendritic cells labeled for in vivo (19)F magnetic resonance imaging cell tracking. Cytotherapy. 2010;12(2):238–50. doi: 10.3109/14653240903446902 pmid:20053146.
- 23. Eger EI 2nd, Saidman LJ. Illustrations of inhaled anesthetic uptake, including intertissue diffusion to and from fat. Anesthesia and analgesia. 2005;100(4):1020–33. doi: 10.1213/01.ANE.0000146961.70058.A1 pmid:15781517.
- 24. Galkina E, Thatte J, Dabak V, Williams MB, Ley K, Braciale TJ. Preferential migration of effector CD8+ T cells into the interstitium of the normal lung. The Journal of clinical investigation. 2005;115(12):3473–83. doi: 10.1172/JCI24482 pmid:16308575; PubMed Central PMCID: PMC1288831.
- 25. Ahrens ET, Zhong J. In vivo MRI cell tracking using perfluorocarbon probes and fluorine-19 detection. NMR in biomedicine. 2013;26(7):860–71. doi: 10.1002/nbm.2948 pmid:23606473; PubMed Central PMCID: PMC3893103.
- 26. Zhong J, Mills PH, Hitchens TK, Ahrens ET. Accelerated fluorine-19 MRI cell tracking using compressed sensing. Magnetic resonance in medicine. 2013;69(6):1683–90. doi: 10.1002/mrm.24414 pmid:22837054; PubMed Central PMCID: PMC3494778.