Development of a Clinically Relevant Reporter for Chimeric Antigen Receptor T-Cell Expansion, Trafficking, and Toxicity

Abstract

Although chimeric antigen receptor T (CART)-cell therapy has been successful in treating certain hematological malignancies, wider adoption of CART-cell therapy is limited due to minimal activity in solid tumors and development of life-threatening toxicities, including cytokine release syndrome (CRS). There is a lack of a robust, clinically relevant imaging platform to monitor in vivo expansion and trafficking to tumor sites. To address this, we utilized the sodium iodide symporter (NIS) as a platform to image and track CART-cells. We engineered CD19-directed and BCMA-directed CART-cells to express NIS (NIS⁺CART19 and NIS⁺BCMA-CART, respectively) and tested the sensitivity of ¹⁸F-TFB-PET to detect trafficking and expansion in systemic and localized tumor models and in a CART-cell toxicity model. NIS⁺CART19 and NIS⁺BCMA-CART-cells were generated through dual transduction with two vectors and demonstrated exclusive ¹²⁵I uptake in vitro. ¹⁸F-TFB-PET detected NIS⁺CART-cells in vivo to a sensitivity level of 40,000 cells. ¹⁸F-TFB-PET confirmed NIS⁺BCMA-CART-cell trafficking to the tumor sites in localized and systemic tumor models. In a xenograft model for CART-cell toxicity, ¹⁸F-TFB-PET revealed significant systemic uptake, correlating with CART-cell in vivo expansion, cytokine production, and development of CRS-associated clinical symptoms. NIS provides a sensitive, clinically applicable platform for CART-cell imaging with PET scan. ¹⁸F-TFB-PET detected CART-cell trafficking to tumor sites and in vivo expansion, correlating with the development of clinical and laboratory markers of CRS. These studies demonstrate a noninvasive, clinically relevant method to assess CART-cell functions in vivo.

Introduction

Clinical trials of chimeric antigen receptor T (CART)-cell therapy in patients with hematological malignancies, including B-cell acute lymphoblastic leukemia (ALL), B-cell non-Hodgkin lymphoma, B-cell chronic lymphoblastic leukemia, and multiple myeloma (MM) have shown unprecedented therapeutic efficacy (1-7). Most notably, the U.S. Food and Drug Administration has approved CD19-directed CART (CART19)-cell therapy for patients with relapsed/refractory B-ALL (8), diffuse large B-cell lymphoma (9,10), and mantle cell lymphoma (11). Following infusion, CART-cells undergo massive in vivo proliferation, which correlates with the development of toxicities after CART-cell therapy and often predicts a stronger response to CART-cell therapy (8,12,13). In solid tumors, the application of CART-cell therapy has been largely unsuccessful to date. Multiple mechanisms for the lack of CART-cell activity in solid tumors have been postulated, including poor T-cell expansion, T-cell exhaustion, poor trafficking to tumor sites, and inhibition by the tumor microenvironment (14-19).

To date, there is also a lack of a robust, clinically validated in vivo imaging platform to assess trafficking and expansion of CART-cells after infusion. Such a platform would allow 1) real-time, serial monitoring of CART-cell proliferation and trafficking to tumor sites, 2) the rapid implementation of appropriate strategies to enhance CART-cell activity, and 3) the potential to predict severe cytokine release syndrome (CRS) associated with massive T-cell expansion after CART-cell administration. Several technologies have been reported for in vivo CART-cell imaging, such as herpes simplex virus-thymidine kinase (HSV-1-TK), Escherichia coli dihydrofolate reductase enzyme (eDHFR), somatostatin receptor 2 (SSTR2), and prostate-specific membrane antigen (PSMA) (20-25). Although these studies demonstrate the ability to label and follow CART-cell trafficking in preclinical studies, clinical experience using any of these reporters has been limited so far (21,26).

Sodium iodide symporter (NIS) has been widely used in the clinic due to its ability to mediate uptake of a variety of radioisotopes by the thyroid gland. Our group has pioneered the use of NIS in gene therapy and demonstrated applications for oncolytic viral imaging in preclinical studies and clinical trials (27-29). Although incorporation of NIS into CART-cells has been reported, no platforms have been developed to track CART-cell expansion or to predict the development of CRS (30,31). Here, we report NIS as a modality both to track CART-cells by ¹⁸F-TFB-PET and to diagnose and potentially predict toxicities after CART-cell administration.

Materials and Methods

Cells lines and primary cells

The CD19⁺ ALL cell line Nalm6 and CD19^- and BCMA^- cell line K562 were purchased from ATCC (Manassas, VA, USA) in 2017, and the B-cell maturation antigen (BCMA)⁺ MM cell line OPM-2 was purchased from DSMZ (Braunschweig, Germany) in 2018. Nalm6 and OPM-2 were used as target cells. K562 served as a negative control. All cell lines were regularly tested for mycoplasma and phenotype was confirmed by flow cytometry. The number of passages was limited to ten. These cell lines were transduced with a luciferase-ZsGreen lentivirus (Addgene, Cambridge, MA, USA) and sorted with FACS Aria II (BD Biosciences, San Diego, CA, USA) instrument to 100% purity. Cell lines were maintained in R10 medium as previously described (32). Primary leukemia cells from 3 different donors were obtained from the Mayo Clinic Biobank for ALL patients under an Institutional Review Board approved protocol (IRB 17-008762). All the patients provided signed informed consents. The use of recombinant DNA was approved by the Institutional Biosafety Committee (IBC HIP00000252.16). All cell lines used regularly tested negative for mycoplasma contamination throughout the whole duration of this study.

CART-cell generation

Second-generation CD19 and BCMA CAR constructs were synthesized de novo [Integrated DNA Technologies, Inc. (IDT), Coralville, IA, USA]. These constructs consisted of a single-chain variable fragment against CD19 (clone FMC63) or BCMA (clone BCMA-02) cloned into a second-generation 4-1BB co-stimulated CAR in a third-generation lentivirus as previously described (32). The human NIS plasmids were obtained from Imanis Life Science (Rochester, MN, USA) through collaborative research and development and material transfer agreements. Primary cells were cultured in T-Cell Medium (TCM) made with X-Vivo 15 (Lonza, Walkersville, MD, USA) supplemented with 10% human serum albumin (Corning, NY, USA) and 1% Penicillin-Streptomycin-Glutamine (Gibco, Gaithersburg, MD, USA). Lentiviral particles were generated through the transient transfection of plasmids into 293T virus-producing cells (purchased from ATCC (Manassas, VA, USA)], in the presence of Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA), VSV-G, and packaging plasmids (Addgene, Cambridge, MA, USA).

The density gradient technique was used to isolate peripheral blood mononuclear cells (PBMCs). T-cell isolation from PBMCs (n=8) was performed via a negative selection magnetic bead kit (magnetic beads against CD15, CD14, CD 34, CD36, CD56, CD123, CD235a, CD19, and CD16). The purity of T-cell population after the isolation was more than 95%. T-cells isolated from normal donors were stimulated using Cell Therapy Systems Dynabeads CD3/CD28 (LifeTechnologies, Oslo, Norway) at a 1:3 cell:bead ratio. Stimulated T cells were transduced with lentiviral particles at a multiplicity of infection (MOI) of 3.0, 24 hours after stimulation. NIS⁺CART-cells were generated through dual transduction of both NIS and CAR virus. NIS⁺ cells were selected by adding 1 µg/mL of puromycin dihydrochloride (Millipore Sigma, Ontario, Canada) to the TCM on days 3, 4, 5, and 6. The expression of NIS and CAR were analyzed by flow cytometry on day 6. CARs were stained with goat anti-mouse IgG. NIS was stained with anti-human ETNL NIS followed by a PE-conjugated anti-rabbit secondary antibody (Supplementary Table S1). ETLN antibody recognizes the cytosolic C-terminus of NIS. Therefore, cells were permeabilized prior to staining. Beads were removed from T cells by using DynaMag™−50 (Invitrogen, Carlsbad, CA, USA) on day 6. On day 8, CART-cells were harvested and cryopreserved in freezing medium composed of 90% fetal bovine serum (FBS, Millipore Sigma, Ontario, Canada) and 10% dimethyl sulfoxide (DMSO, Millipore Sigma, Ontario, Canada) for planned experiments. CART-cells were thawed and rested in TCM 6-12 hours prior to the individual experiments.

In vitro ¹²⁵ iodide (I) uptake assay in NIS⁺ CART-cells

Untransduced T-cells (UTD) or NIS⁺ CART19 cells (300,000 cells) were washed once by Hanks’ Balanced Salt Solution (HBSS) modified with 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) HEPES (STEMCELL Technologies, Vancouver, Canada). Cells were then re-suspended in HEPES/HBSS or HEPES/HBSS with 1 mM KClO₄ and incubated at 37°C, as indicated in the specific experiment. Radioactive substrate solutions were prepared immediately prior to each assay. Na¹²⁵I in 0.1 M NaOH (Perkin Elmer, Waltham, MA) was diluted in uptake buffer and was added to each tube [500,000 counts per minute (CPM) of ¹²⁵I in each sample]. Cells were then incubated at 37°C for 60 minutes prior to the assay. After incubation, samples were centrifuged, and the supernatant was aspirated. Cells were washed with cold HEPES/HBSS and centrifuged. Cells were then re-suspended in NaOH for quantification on a 2470 Automatic Gamma Counter (Perkin Elmer, Waltham, MA, USA).

Xenograft mouse models

Male and female 8-12 week-old non-obese diabetic/severe combined immunodeficient bearing a targeted mutation in the interleukin (IL)2 receptor gamma chain gene (NSG) mice were purchased from the Jackson Laboratory (Bar Harbor, ME, USA, catalog #005557) and maintained within the Mayo Clinic Department of Comparative Medicine under an Institutional Animal Care and Use Committee approved protocol (IACUC A00001767 and A00003102).

A. Nalm6 xenograft models:

Subcutaneous Nalm6 xenografts were established through the subcutaneous injection of 1 x 10⁶ luciferase⁺ Nalm6 cells in PBS. Tumor burden was assessed by BLI using a Xenogen IVIS-200 Spectrum camera (PerkinElmer, Waltham, MA, USA) to confirm engraftment weekly after Nalm6 inoculation. Once the tumor volume reached 50 mm in diameter, mice were then randomized into groups receiving 1) UTD or 2) NIS⁺CART-cells (5 x10⁶ intravenously). Weekly Imaging was performed 10 minutes after the intraperitoneal (IP) injection of 10 µL/g D-luciferin (15 mg/mL, Gold Biotechnology, St.Louis, MO, USA). Bioluminescent images were analyzed using Living Image version 4.4 (PerkinElmer, Waltham, MA, USA). One week after T-cell injection, mice were imaged both with BLI and ¹⁸F-TFB-PET/CT as described below. Mice were euthanized after the PET imaging and the experiment was concluded.

B. OPM-2 xenograft models:

The BCMA⁺ luciferase⁺ MM OPM-2 cell line was used to establish localized or systemic MM xenografts. For the localized model, mice were implanted with 2.5 x 10⁵ luciferase⁺ OPM-2 cells subcutaneously in the flank. For the systemic model, 1 x 10⁶ cells were resuspended in PBS and injected via the tail vein. The tumor burden was assessed weekly by BLI using a Xenogen IVIS-200 Spectrum camera to confirm engraftment of OPM-2 cells. In the localized tumor model, engraftment was confirmed four weeks after OPM-2 injection. In the systemic model, engraftment was confirmed three after OPM-2 inoculation. Mice were then randomized based on tumor burden to receive 5 x 10⁶ cells of 1) UTD, 2) BCMA-CART-cells, or 3) NIS⁺BCMA-CART-cells. Mice were imaged with BLI and ¹⁸F-TFB-PET/CT as described below. Imaging was performed one week after T-cell injection in the localized model and two weeks after T-cell injection in the systemic model. Mice were euthanized if they develop paralysis, hunched posture, or abdominal distention. At the conclusion of this experiment, femurs were harvested and bone marrow cells were flushed. Bone marrow cells were stained with APC-Cy7 mouse CD45 (clone 30-F11, cat# 103116), brilliant violet (BV) 421 human CD45 (clone HI30, cat# 304032), and PE-Cy7 human BCMA (clone 19F2, cat# 357507).

C. K562 xenograft models:

The CD19^- and BCMA^- luciferase⁺ CML K562 cell line was used to establish localized or systemic CD19^-/BCMA^- tumor model. For the localized model, mice were implanted with 1 x 10⁶ luciferase⁺ K562 cells subcutaneously in the flank two weeks before T-cell infusion. For the systemic model, 1 x 10⁶ cells were resuspended in PBS and injected via the tail vein, one week before T-cell infusion. The tumor burden was assessed weekly by BLI using a Xenogen IVIS-200 Spectrum camera to confirm engraftment of K562 cells as well as visual inspection. For the localized model, mice were treated when the tumor size reached 50 mm in diameter. The mice were randomized according to the tumor burden assessed by IVIS to receive 5 x 10⁶ of 1) UTD, 2) NIS⁺CART19, or 3) NIS⁺BCMA-CART-cells. One week after T cell infusion mice were imaged with BLI and ¹⁸F-TFB-PET as described below. Mice were euthanized after the imaging with ¹⁸F-TFB-PET.

D. Patient-derived ALL xenografts for CRS:

A model for CART-cell toxicity was established using patient-derived xenografts as previously reported (32). Briefly, mice were first IP injected with 30 mg/kg busulfan (Selleckchem, Houston, TX, USA). The following day, mice received 1-3 x 10⁶ leukemic blasts derived from the peripheral blood of patients with relapsed ALL via tail vein injection. Mice were monitored for human CD19⁺ cell engraftment for 10-13 weeks by peripheral blood sampling via tail vein bleeding. Here, 70 µL peripheral blood was processed with red blood cell lysis using BD FACS Lyse buffer (BD Biosciences, San Diego, CA, USA). Cells were then stained with APC-Cy7 mouse CD45, BV421 human CD45, and PE human CD19 (clone SJ25C1, cat# 340720) (Supplementary Table S1). Absolute count of human CD19⁺ cells was determined using counting beads via flow cytometry. Mouse serum was isolated from the remaining peripheral blood volume to perform cytokine/chemokine assay as described below. When the peripheral blood human CD19⁺ cell count was ≥10 cells/µL at day −1, mice were randomized to receive NIS⁺ CART19 (5 x 10⁶ cells intravenously) or control PBS. Mice were weighed daily and monitored for motor weakness and well-being. The development of CRS was defined by the combination of weight loss, decline motor function, and elevated cytokines. Seven days after treatment, mice underwent ¹⁸F-PET/computed tomography (CT) imaging as described below. The peripheral blood was collected via tail vein sampling the following day after ¹⁸F-PET/CT imaging (day 8) to assess in vivo CART-cell expansion and cytokines. Here, 70 µL mouse peripheral blood was processed with red blood cell lysis using BD FACS Lyse buffer (BD Biosciences, San Diego, CA, USA) and then used for flow cytometric studies as described below. Mice were euthanized on day 8 to harvest the spleen and liver in order to confirm the trafficking of CAR-T cells via flow cytometry.

PET/CT Imaging

Imaging was performed in the Mayo Clinic Small Animal Imaging Core using an Inveon Multiple Modality PET/CT scanner (Siemens Medical Solutions, Knoxville, TN, USA). ¹⁸F-TFB for PET/CT imaging was produced as previously described (33). Forty-five minutes to one hour prior to PET imaging, 9.25 MBq of ¹⁸F-TFB was delivered to the mice via intravenous injection. CT image acquisition was performed in 5 minutes with 360-degree rotation and 180 projections at 500 µA, 80 keV and 200 ms exposure. PET Image acquisition began approximately 45 minutes following isotope injection with total acquisition time of 20 minutes.

Co-registered images were rendered and visualized using the PMOD software (PMOD Technologies Ltd., Switzerland). To calculate standardized uptake value (SUV), the volume of interest (VOI) was determined by the PMOD software. Then, SUV was calculated using the formula as below.

$$\text{SUV in VOI = Concentration of activity in VOI }\left(\text{MBq/mL}\right)\text{ / }\left(\text{administered dose }\left(\text{MBq}\right)\text{ / body weight }\left(\text{g}\right)\right)$$

The use and handling of radiotracers were approved by the Institutional Biosafety Committee (IBC HIP00000252).

In vivo sensitivity assay for NIS⁺ CART-cells

Male and female 8-to-12-week-old NSG mice were utilized in this study. NSG mice were purchased from the Jackson Laboratory (cat# 005557, Maine, CA, USA). NSG mice (naïve mice) were subcutaneously injected in the legs with different doses of NIS⁺CART19 cells (0.1 x 10⁵ - 1.25 x 10⁶) or PBS along with matrigel (Corning, Corning, NY, USA) (50 µL NIS⁺CART-cell + 50 µL of matrigel), as indicated in the specific experiment. 10-15 minutes after the injection of matrigel and NIS⁺CART-cell mixture, mice were intravenously injected with 9.25 MBq of ¹⁸F-TFB, and images were acquired as described above.

In vitro antigen-specific degranulation and cytokine production assays

Intracellular cytokine analysis and T-cell degranulation assays were performed following incubation of CART-cells and Nalm6, OPM-2, or K562 cells for 4 hours in the presence of FITC-conjugated CD107a, monensin (cat# 554724, BioLegend, San Diego, CA, USA), human CD49d (clone L25, cat# 340976, BD Biosciences, San Diego, CA, USA), and human CD28 (clone L293, cat# 348040, BD Biosciences, San Diego, CA, USA). After 4 hours, cells were harvested, and intracellular staining was performed after surface staining APC-H7 anti-human CD3 and LIVE/DEAD^™ Fixable Aqua Dead Cell Stain Kit followed by fixation and permeabilization with fixation medium A and B (cat# GAS001S100 and GAS002S100, Life Technologies, Oslo, Norway).

In vitro antigen-specific proliferation assay

For proliferation assays, CFSE (cat# C34554, Life Technologies, Oslo, Norway) labeled CART-cells and irradiated Nalm6, OPM-2, or K562 cells were co-cultured at 1:1 ratio. All cell lines were lethally irradiated (120 Gy) prior to plating. Cells were co-cultured for 5 days, as described in the specific experiments, and then cells were harvested and surface staining with APC-H7 anti-human CD3 and LIVE/DEAD^™ Fixable Aqua Dead Cell Stain Kit was performed (Supplementary Table S1). Phorbol myristate acetate (PMA) and ionomycin (Millipore Sigma, Ontario, Canada) were used as a positive non-specific stimulant of T-cells, at different concentrations as indicated in the specific experiments.

In vitro cytotoxicity assay

Luciferase⁺ Nalm6, OPM-2, and K562 cells were incubated at the indicated ratios with effector T cells for 24 hours. The killing was calculated by bioluminescence imaging (BLI) on a Xenogen IVIS-200 Spectrum camera (cat# 124262, PerkinElmer, Hopkinton, MA, USA) as a measure of residual live cells. Ten minutes prior to imaging, samples were treated with 1 µL D-luciferin (30 µg/mL, Gold Biotechnology, St. Louis, MO, USA) per 100 µL sample volume.

Multiplex analysis of cytokines and chemokines

Cytokine and chemokine profiles from the patient-derived ALL xenograft sera before (day −1) and after (day 8) the treatment of NIS⁺ CART19 cell were interrogated with the HCYTMAG-60K-PX38 Milliplex kit (Millipore Sigma, Ontario, Canada), following the procedure described in the manufacturer’s manual. Sera from the mice that did not receive CART-cells were used as controls. Sera were diluted 1:2 with human serum matrix (provided within the Milliplex kit) prior to plating. Data were collected using a Luminex instrument (cat# 40-012, Millipore Sigma, Ontario, Canada). The xPONENT® Software was used to analyze the data (cat# MAP0200, Invitrogen, Carlsbad, CA, USA).

Multi-parametric flow cytometry

Anti-human and anti-mouse antibodies were purchased from Biolegend, eBioscience, or BD Biosciences (San Diego, CA, USA) (Supplementary Table S1). BD FACS lyse buffer (BD Biosciences, San Diego, CA, USA) was used to lyse red blood cells of mouse (patient-derived ALL xenograft) peripheral blood samples (day −1 and 8 of NIS⁺ CART-cell infusion) prior to staining and flow cytometric analysis. Cells from in vitro culture (in vitro antigen-specific degranulation/cytokine production assays and antigen-specific proliferation assay), mouse peripheral blood, or mouse organs (spleen, liver, and bone marrow) were analyzed by flow cytometry. Prior to staining, cells were washed twice in a flow buffer [PBS supplemented with 1% FBS and 1% sodium azide (Ricca Chemical, Arlington, TX, USA)], and stained at room temperature. For cell number quantitation, Countbright beads (Invitrogen, Carlsbad, CA, USA) were used according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA, USA). In all analyses, the population of interest was gated based on forward vs. side scatter characteristics, followed by singlet gating, and live cells were gated following staining with LIVE/DEAD™ Fixable Aqua Dead Cell Stain Kit (Invitrogen, Carlsbad, CA, USA).

Surface expression of the CAR was detected by staining with a goat anti-mouse F(ab’)2 antibody. In brief, an aliquot of the CART-cells (e.g., 50,000 T-cells) was first washed and then resuspended in 50 µL of a flow buffer. Cells were then stained with 1 µL of goat anti-mouse antibody and 0.3 µL of LIVE/DEAD^™ Fixable Aqua Dead Cell Stain Kit for excluding dead cells and incubated in dark for 15 minutes at room temperature. After incubation, cells are washed by adding 150 µL of a flow butter and centrifuged at 650 x g for 3 minutes at 4°C. After surface staining, cells are fixed and permealized by adding 100 µL of a fixation medium and incubated for 15 minutes at room temperature. Cells are then washed with 100 µL of a flow buffer. Fixed/permeabilized cells were then resuspended in 50 µL of a permeabilizing buffer. NIS expression was detected by anti-human ETNL (NIS antibody). The NIS antibody recognizes the cytosolic C-terminus of NIS. To stain for NIS, 0.3 ng of anti-human NIS antibody was added along with 50 µL of a flow buffer and incubated for 1 hour at 4°C. Then 100 µL of flow buffer was added and cells were centrifuged at 650 x g for 3 minutes at 4°C.Then 2.5 µL of anti-rabbit secondary antibody along with 50 µL of flow buffer were added and cells were incubated for 30 minutes at 4°C. Cells were then washed and resuspended in 200 µL of a flow buffer and acquired on a flow cytometer. Flow cytometry was performed on a CytoFLEX (Beckman Coulter, Chaska, MN, USA) three-laser cytometer. Analyses were performed using FlowJo X10.0.7r2 software (Ashland, OR, USA). See Supplemental Table 1 for specific details of flow antibodies.

Statistics

All statistics were performed using GraphPad Prism version 8.05 for Windows (GraphPad Software, La Jolla CA, www.graphpad.com). Statistical tests are described in detail in the representative figure legends.

Results

Development of dual NIS⁺CART-cells

We used two CAR constructs that have been studied extensively in preclinical and clinical studies against B-cell malignancies and MM: CAR19 and anti-BCMA-CAR, respectively (Fig. 1A). We generated NIS⁺CART19 and NIS⁺BCMA-CART-cells through dual lentiviral transduction (NIS and CAR vectors, see Methods). On day 6 of expansion, cells were stained for the CAR and NIS and analyzed by flow cytometry. Representative flow plots of NIS⁺CART-cells are shown in Fig. 1B, and the gating strategy is shown in Supplemental Fig. S1A. The summary of NIS⁺CAR⁺ transduction efficiency is shown in Supplemental Fig. S1B. No T-cell phenotypic changes were seen after the incorporation of NIS into CART-cells (Supplemental Fig. S1C).

Figure 1. Development of dual NIS⁺CART cells.

A, Schematic representation of the lentiviral vectors used in this study. The CAR19 consisted of anti-CD19 single chain variable fragment (scFv) linked to a 4-1BB costimulatory domain and a CD3ζ signaling domain. EF1α, elongation factor-1α; H, hinge; TM, transmembrane. The BCMA CAR consisted of anti-BCMA scFv linked to 4-1BB costimulatory domain and CD3ζ signaling domain. The sodium iodide symporter (NIS) is under control of the EF1α promoter, and the puromycin (puro) resistance gene is linked via a P2A cleavage peptide. B, Representative flow plots of untransduced T-cells (UTD), CART19, BCMA-CART, NIS⁺CART19, and NIS⁺BCMA-CART-cells.

Incorporation of NIS into CART-cells does not impair their function

To confirm that incorporating NIS into CART-cells did not have a negative impact on effector functions, we first investigated the in vitro antigen-specific activity of NIS⁺CART-cells. Antigen-specific killing of CD19⁺ Nalm6 cells by NIS⁺CART19 or CART19 cells was comparable. Similarly, antigen-specific killing of BCMA⁺ OPM-2 cells by NIS⁺BCMA-CART or BCMA-CART-cells was comparable (Fig. 2A). In a 5-day antigen-specific proliferation assay, we observed a potent and equivalent expansion of both NIS⁺CART19 and CART19 cells upon stimulation in co-cultures with irradiated CD19⁺ Nalm6 cells (Fig. 2B, Supplemental Fig. S2). NIS⁺BCMA-CART-cells also showed a comparable antigen-specific proliferation to BCMA-CART-cells upon stimulation in co-cultures with BCMA⁺ OPM-2 cells (Supplemental Fig. S2, S3A). Similarly, no differences were observed in antigen-specific degranulation or intracellular cytokines between CART19 and NIS⁺CART19 cells (Fig. 2C, Supplemental Fig. S4A-B) or between BCMA-CART and NIS⁺BCMA-CART-cells. (Supplemental Fig. S3B, S4C). To confirm the antigen specificity of NIS⁺CART-cells, the CD19^-BCMA^- K562 cell line was used as a negative control. NIS⁺CART19 or NIS⁺BCMA-CART-cells did not exert any effector functions upon stimulation with antigen-negative cell lines in co-cultures with the K562 cells (Supplemental Fig. S3D, S5A-C). Overall, these results indicate that the incorporation of NIS in CART-cells did not inhibit effector functions in vitro.

Figure 2. Incorporation of NIS into CART-cells does not impair effector functions.

A, NIS⁺CART19 or NIS⁺BCMA-CART-cells were co-cultured at different effector to target (E:T) ratios with CD19⁺ luciferase⁺ Nalm6 or BCMA⁺ luciferase OPM-2⁺ cell lines, respectively. At 24 hours, cytotoxicity was assessed by bioluminescent imaging relative to controls. Data are plotted as mean ± SEM. ****p<0.0001, n.s. not significant, two-way ANOVA; n=3 biological replicates (independent experiments), n=2 technical replicates per biological replicate. B, CFSE-labeled UTD, CART19, or NIS⁺CART19 cells were co-cultured with medium alone, PMA, 5 ng/mL), and ionomycin (1 μg/mL) as a non-specific stimulant, or lethally irradiated (120 Gy) CD19⁺ JeKo-1 cell lines at a 1:1 ratio. On Day 5, absolute numbers of cells were counted by flow cytometry. ***p<0.001, n.s. not significant, one-way ANOVA; n=3 biological replicates (independent experiments), n=2 technical replicates per biological replicate. C, CART19 or NIS⁺CART19 cells were co-cultured with CD19⁺ JeKo-1 cells at a 1:5 E:T ratio for 4 hours, intracellularly stained, and analyzed via flow cytometry. Data are plotted as mean ± SEM. *p<0.05, **p<0.01, ***p<0.001, n.s. not significant, one-way ANOVA; n=3 biological replicates (independent experiments), 2 technical replicates per biological replicate.

NIS⁺CART-cells demonstrate specificity for radioiodine and high sensitivity in vivo

Next, we aimed to measure the specificity of dual transduced NIS⁺CART-cells for radioiodine uptake. NIS⁺CART19 cells or UTD were incubated in the presence of radioactive iodide molecules (¹²⁵I) with or without the NIS-specific inhibitor potassium perchlorate (KClO₄). ¹²⁵I uptake occurred exclusively in NIS⁺CART19 cells, and this uptake was appropriately inhibited by KClO₄ (Fig. 3A), suggesting that iodide uptake was mediated by NIS⁺CART-cells. We then performed a sensitivity experiment to determine the lowest number of NIS⁺CART-cells that can be detected by ¹⁸F-TFB-PET imaging in vivo. Here, NSG mice were subcutaneously injected with different doses of NIS⁺CART19 cells, and 10-15 minutes after, mice were intravenously injected with ¹⁸F-TFB. ¹⁸F-TFB-PET showed that NIS⁺CART-cells were detectable at numbers as low as 0.4 x 10⁵ cells (Fig. 3B-C). Physiological uptake of TFB by endogenous NIS was seen in the thyroid/salivary glands, stomach, and bladder, as expected.

Figure 3. Specificity and sensitivity of NIS⁺CART-cells.

A, ¹²⁵I was added to each sample as depicted on the x axis, incubated for one hour at 37°C, and quantified with a gamma counter. KClO₄ was used as an inhibitor of NIS. Data are plotted as mean ± SEM., **p<0.005, t-test; 2 independent experiments, 2 replicates per experiment. CPM: counts per minute. B, In vivo maximum intensity projection (MIP) views. Different numbers of NIS⁺CART-cells along with matrigel were subcutaneously injected in the legs of NSG mice 10 minutes before the administration of ¹⁸F-TFB. Control (Ctr), thyroid/salivary glands (T/S), stomach (St), and bladder (Bl) physiologic accumulation of TFB is shown. C, Six different doses of NIS⁺CART-cells, and PBS as a negative control, were tested in 4 mice, and 3 different experiments were performed. Data are plotted as mean ± SD. SUV: standardized uptake value.

PET imaging of NIS⁺CART-cells efficiently detects trafficking to tumor sites

Having demonstrated that NIS⁺CART-cells provided a sensitive method of detection by ¹⁸F-TFB-PET and did not have impaired effector functions, we utilized this platform to assess CART-cell trafficking to tumor sites first using a MM xenograft model. Here, NSG mice were engrafted with BCMA⁺ OPM-2 cells (see systemic model in Methods; Fig. 4A, Supplemental Fig. S6A). A K562 xenograft model was established as a negative control (Supplemental Fig. S6B). Three weeks after inoculation of OPM-2 cells, tumor burden was assessed by bioluminescence imaging (BLI), and mice were then randomized to receive UTD, BCMA-CART, or NIS⁺BCMA-CART-cells via tail vein injection. Mice were then serially imaged with BLI as a measure of disease burden and ¹⁸F-TFB-PET to assess CART-cell expansion and trafficking. ¹⁸F-TFB-PET suggested trafficking of NIS⁺BCMA-CART-cells to the bone marrow, corresponding to the MM tumor site based on BLI (Fig. 4B, Supplemental Video S1A, Supplemental Fig. S6A). In order to have a better assessment of NIS⁺CART-cell trafficking, NIS physiological uptake was removed from the PET images using PMOD software (Fig. 4B). Supplemental Fig. S6A-B depicts PET images that include the background NIS signal. Both BCMA-CART and NIS⁺BCMA-CART-cells exhibited similar anti-tumor activity and resulted in complete remission (Fig. 4C) and prolonged overall survival of the mice (*p=0.01, log-rank test) (Fig. 4D). ¹⁸F-TFB-PET imaging did not detect NIS⁺BCMA-CART-cells in K562 xenografts (Supplemental Fig. S6B). Quantitative assessment of ¹⁸F-TFB uptake in the tumor sites revealed significantly higher uptake in OPM-2 xenografts compared with K562 xenografts (Supplemental Fig. S6C). K562 xenografts had a significantly worse survival compared with OPM-2 xenografts (Supplemental Fig. S6D) following treatment with NIS⁺BCMA-CART-cells, indicating antigen-specific activation and anti-tumor activity of NIS⁺BCMA-CART-cells in vivo. Bone marrow samples were harvested and analyzed by flow cytometry. Analyses confirmed infiltration of OPM-2 cells in the bone marrow, which correlated with BLI findings (Fig. 4E).

Figure 4. ¹⁸F-TFB-PET imaging of NIS⁺CART-cells is an efficient platform to detect CART-cell trafficking.

A, Schema of the systemic OPM-2 xenograft model. 1 x 10⁶ BCMA⁺ luciferase⁺ OPM-2 cells were intravenously injected into NSG mice. Three weeks later, bioluminescence imaging (BLI) was performed, and mice were randomized to receive 5 x 10⁶ untransduced (UTD) T-cells, BCMA-CART-cells, or NIS⁺BCMA-CART-cells (n=2 independent experiments, 15 mice per experiment, 5 mice per group). B, In vivo maximum intensity projection (MIP) views. ¹⁸F-TFB-PET imaging of NIS⁺BCMA-CART-cells and BCMA-CART-cells to monitor trafficking to tumor sites in the systemic MM xenograft model. BLI of MM xenografts, 7 days after treatment is also shown. NIS physiological uptake was removed from the PET images using PMOD software. C, Tumor growth with and without treatment with BCMA-CART-cells or NIS⁺BCMA-CART-cells in MM xenografts. Data are plotted as mean ± SEM. ****p<0.0001, n.s. not significant, two-way ANOVA. D, Kaplan-Meier survival curves showing survival of mice from (C). **p<0.01, log-rank test. E, Flow cytometric analysis of bone marrow from mice treated with UTD.

Next, we intended to further demonstrate the efficiency of this technology to assess NIS⁺CART-cell trafficking to tumor sites using subcutaneous Nalm6 and OPM-2 xenograft models. K562 xenografts were used as negative controls. After engraftment was confirmed by BLI, Nalm6 xenografts were randomized to receive either CART19 or NIS⁺CART19 cells (5 x 10⁶ intravenously), whereas OPM-2 xenografts were randomized to receive BCMA-CART or NIS⁺BCMA-CART-cells (5 x 10⁶ intravenously). Mice underwent serial BLI to determine disease burden and serial ¹⁸F-TFB-PET imaging to track CART-cell trafficking. The administration of ¹⁸F-TFB was performed through tail vein injection 45 minutes prior to imaging. In order to visualize the NIS⁺CART-cell trafficking clearly in PET images, background signal from thyroid, salivary glands, stomach, and bladder were removed using PMOD software. In both models, ¹⁸F-TFB-PET demonstrated trafficking of NIS⁺CART19 and NIS⁺BCMA-CART-cells to the subcutaneous tumor sites (Fig. 5A and 5C, Supplemental Video S2) seven days after CART-cell injection, whereas NIS⁺CART19 cells or NIS⁺BCMA-CART-cells did not traffic to K562 tumors (Supplemental Fig. S7A-B). Quantitative analysis of ¹⁸F-TFB uptake in the tumor sites revealed that Nalm6 xenografts showed significantly higher NIS⁺CART19 cell accumulation compared with K562 xenografts (Fig. 5B). Similar to Nalm6 xenografts, OPM-2 xenografts revealed higher ¹⁸F-TFB uptake of NIS⁺BCMA-CART-cell compared to K562 xenografts (Fig. 5D).

Figure 5. NIS⁺CART-cells efficiently traffic to tumor sites.

A, Bioluminescence (BLI) and ¹⁸F-TFB-PET imaging of subcutaneous Nalm6 xenografts. 1 x 10⁶ CD19⁺ luciferase⁺ Nalm6 cells were subcutaneously injected into the right flanks of NSG mice. Four weeks later, mice were treated with 5 x 10⁶ NIS⁺CART19 cells intravenously. BLI and ¹⁸F-TFB-PET/CT were performed seven days after the administration of NIS⁺CART19 cells. As a negative tumor antigen control, K562 xenografts were used (see Methods, Supplemental Fig. S7). In vivo maximum intensity projection (MIP) views are shown. NIS physiological uptake was removed from the PET images using PMOD software. St, stomach. B, ¹⁸F-TFB uptake in the Nalm6 compared to K562 xenografts following treatment with NIS⁺CART19 cells. Data are plotted as mean ± SEM. *p<0.05, t-test. C, Bioluminescence (BLI) and ¹⁸F-TFB-PET imaging of subcutaneous OPM-2 xenografts. 0.25 x 10⁶ BCMA⁺ luciferase⁺ OPM-2 cells were subcutaneously injected into the right flanks of NSG mice. Four weeks later, mice were treated with 5 x 10⁶ NIS⁺BCMA-CART-cells intravenously. BLI and ¹⁸F-TFB-PET/CT were performed seven days after NIS⁺BCMA-CART-cell treatment. As a negative tumor antigen control, K562 xenografts were used (see Methods, Supplemental Fig. S7). NIS physiological uptake was removed from the PET images using PMOD software. D, ¹⁸F-TFB uptake in OPM-2 xenografts compared to K562 xenografts following treatment with NIS⁺BCMA-CART-cells. Data are plotted as mean ± SEM. ***p=0.0005, t-test. n=2 independent experiment, 12 mice per experiment.

¹⁸F-TFB-PET detects massive NIS⁺CART-cell expansion in a CRS xenograft model

Finally, we explored whether ¹⁸F-TFB-PET imaging could detect CART-cell expansion in vivo and whether this correlated with the development of CRS. We used a patient-derived xenograft model for CART-cell toxicity (see Methods and experimental schema in Fig. 6A). Once mice developed high leukemic burden (defined as human CD19⁺ cells ≥10 cell/µL, Fig 6B), they were randomized to receive either NIS⁺CART19 cells (5 x 10⁶ cells I.V.) or PBS. Within 5-7 days after NIS⁺CART19 cell treatment, the majority of treated mice developed muscle weakness, hunched bodies, and weight loss, as expected in this model (32) (Fig. 6C). However, two mice treated with NIS⁺CART19 cells did not develop any weakness or symptoms and were therefore used as internal controls. ¹⁸F-TFB-PET imaging was performed on day 6 post-treatment. In mice treated with vehicle control, ¹⁸F-TFB-PET imaging only demonstrated physiological ¹⁸F-TFB uptake in the thyroid and stomach, as expected (Supplemental Fig. S8A). In mice that developed CRS symptoms after NIS⁺CART19 cell treatment, ¹⁸F-TFB-PET imaging revealed extensive ¹⁸F-TFB uptake in the bone marrow, spleen, liver, and lungs (“high ¹⁸F-TFB uptake,” Fig 6D), whereas in NIS⁺CART19 cell-treated mice that did not develop CRS symptoms, ¹⁸F-TFB-PET imaging detected limited ¹⁸F-TFB uptake in the spleen (“low ¹⁸F-TFB uptake,” Fig. 6E). Quantitative analysis of ¹⁸F-TFB uptake was significantly higher in mice that developed CRS compared to mice that did not develop CRS (Fig. 6F). Multiplex analysis of serum cytokines and chemokines six days after CART-cell injection demonstrated a significant elevation of granulocyte-macrophage colony-stimulating factor (GM-CSF), tumor necrosis factor-α (TNFα), interferon-γ (IFNγ), monocyte chemoattractant protein (MCP)-1, macrophage inflammatory protein (MIP)-1α, and MIP-1β in mice that developed CRS symptoms, correlating with the high ¹⁸F-TFB uptake, but not in mice that did not develop CRS symptoms (Fig. 6G). Flow cytometric analysis of peripheral blood samples six days following treatment with NIS⁺CART19 cells demonstrated a significant expansion of T cells in the mice that developed CRS, but not in the mice with no CRS symptoms (Fig. 6H, Supplemental Fig. S8B). This experiment suggested a correlation between¹⁸F-TFB uptake with the development of CRS symptoms, T-cell expansion, and cytokine elevation in this CART-cell model.

Figure 6. ¹⁸F-TFB-PET imaging of NIS⁺CART-cells is an efficient platform to detect CART-cell trafficking, in vivo expansion, and associated toxicities.

A, Schema of the patient-derived xenograft toxicity model. NSG mice were inoculated with patient-derived acute lymphoblastic leukemia cells (5 x 10⁶ cells, intravenous). 2-4 weeks later, engraftment was confirmed through peripheral blood sampling. When leukemic blast count reached >10 CD19⁺ cells/µL, mice were treated with either PBS or 5 x 10⁶ NIS⁺CART19 cells. Six days after NIS⁺CART19 treatment, peripheral blood was collected, and flow cytometric analysis was performed. Serum was isolated and analyzed for cytokines by multiplex. Seven days after NIS⁺CART19, mice were imaged with ¹⁸F-TFB-PET/CT (T/S thyroid, St stomach, H heart), n= 9 mice per experiment. B, Flow cytometric analysis of mouse peripheral blood prior to treatment with NIS⁺ CART19 cells to assess engraftment. C, Weights of mice treated with NIS⁺CART19. Data are plotted as mean ± SD. *p<0.05, two-way ANOVA. D, ¹⁸F-TFB-PET imaging of systemic uptake in mice that developed CRS. In vivo maximum intensity projection (MIP) views are presented. E, ¹⁸F-TFB-PET imaging of uptake in the spleens of mice that did not develop CRS after treatment with NIS⁺CART19 cells. MIP views are presented. F, Quantitative measurement of ¹⁸F-TFB uptake in lungs of mice that developed CRS symptoms compared to the mice that did not develop CRS symptoms or to untreated xenografts. Data are plotted as mean ± SEM. *p<0.05, one-way ANOVA. G, Cytokine analysis of sera 6 days after treatment with NIS⁺CART19 cells in mice that did and did not developed CRS symptoms. Data are plotted as mean ± SEM. **p<0.01, ****p<0.0001, two-way ANOVA. H, Flow cytometric analysis of mouse peripheral blood for T-cell expansion 6 days after NIS⁺CART19 in mice that developed CRS symptoms compared to mice that did not develop CRS symptoms. *p<0.05, t-test.

Discussion

In this study, we report the development and characterization of NIS as a sensitive, specific, and clinically relevant reporter platform to image CART-cells by PET scan, to track their trafficking and in vivo expansion, and to potentially predict their toxicities. We show that the incorporation of the NIS transgene into CART-cells did not impair their function or anti-tumor activity in vitro or in vivo. ¹⁸F-TFB-PET imaging of NIS⁺CART-cells successfully detected trafficking to tumor sites in xenograft models. ¹⁸F-TFB-PET imaging of NIS⁺CART-cells also efficiently detected CART-cell in vivo expansion and correlated with the development of clinical and laboratory CRS markers in models of toxicity after CART-cell administration.

The two major challenges for the wider application of CART-cell therapy in the clinic are limited tumor responses and the development of life-threatening toxicities. Despite the significant overall response rates after CART-cell therapies in hematological malignancies, rates of durable responses are low, and most patients relapse within the first 1-2 years (7,12,34). CART-cell activity in solid tumors is extremely limited, and mechanisms for suboptimal activity are complex, including lack of ideal targets, inadequate T-cell expansion, poor trafficking to tumor sites, and inhibition by the tumor microenvironment (14-19,35).

Because CART-cell therapy is becoming widely used in the clinic, the associated toxicities, which are dissimilar from conventional cancer therapies, are being increasingly recognized as potentially life-threatening, challenging to predict and treat, and detrimental to the broader adoption of CART-cell therapy beyond select treatment centers. The most commonly observed toxicity of CART-cell therapies is CRS, which often requires close monitoring and treatment in the intensive-care unit (36). The pivotal clinical trials reported an incidence of Grade ≥3 CRS in 13-22% of patients after CART-cell therapy (8,12,37). In the clinic, the development of CRS is associated with increased proliferation of T cells and elevation of inflammatory and effector cytokines (8,12,13,38).

Development of an efficient system to image CART-cells in vivo can therefore represent a strategy to monitor both their trafficking to tumor sites and expansion. Several in vivo imaging strategies for adoptive cell therapy using reporter gene systems have been described. Constitutive expression of a transporter represents an attractive strategy because reporter expression is not affected by cell division (39,40). HSV-1-TK has been widely utilized to image adoptive cells in vivo (41-44). Najjar et al. have successfully imaged CART-cells in vivo using engineered T cells that express CD19-directed CAR and HSV-1-TK (45). This reporter system was also tested in IL-13Rα2-directed CART-cells in a clinical trial for patients with glioblastoma (21). Although this platform was able to efficiently detect trafficking of HSV-1-TK⁺CART-cells, there was non-specific uptake of the tracer, ¹⁸F-FHBG (fluoro-3-(hydroxymethyl) butylguanine), in the brain before infusion of CART-cells. The high background seen with HSV-1-TK transporter is a major limitation for the accurate assessment of cell trafficking or expansion. Another disadvantage of HSV-1-TK platform is its immunogenicity, which can lead to immune-mediated elimination of infused cells. This has been demonstrated in early clinical trials of adoptive immunotherapy of HSV-1-TK⁺ T cells (26,46).

The SSTR2 reporter system has also been investigated as an imaging platform for CART-cells. It has been demonstrated to have high sensitivity with the ability to detect as low as 50,000 target cells via PET scan (25). The main challenge of SSTR2 is its expression in endogenous T-cells, which can interfere with T-cell activation and potentially negatively impact anti-tumor efficacy (25). eDHFR has been introduced as another strategy for CART-cell imaging (47). Although eDHFR allows high sensitivity in imaging CART-cells in vivo, similar to HSV-1-TK, eDHFR has a significant disadvantage due to its immunogenicity, which could result in patient rejection of eDHFR⁺ cellular products.

PSMA-targeted PET imaging has also been described as a sensitive reporter to image CART-cells that are engineered to express PSMA (20). This system possesses high sensitivity, does not impair CART-cell functions, and efficiently tracks CART-cell trafficking. However, at least an hour is needed for the PSMA⁺CART-cells to fully uptake the ¹⁸F-DCFPyL radiotracer.

NIS is an alternative reporter system which has been shown to be a sensitive strategy for PET imaging in viral and cell therapy clinical trials (33,48-50). There are several advantages for this reporter platform over others. First, the endogenous expression of NIS is restricted to the thyroid, salivary glands, and the stomach at low levels (28); therefore, it does not interfere with imaging of the vast majority of other organs (50). Second, to detect positive cells, imaging of NIS⁺ cells depends on the ATP-driven cellular Na⁺/K⁺ gradient (28,51), resulting in an enhanced overall sensitivity for this platform (28). Third, unlike HSV-1-TK or eDHFR, NIS is a human protein and therefore does not carry any risk for immunogenicity. Lastly and most importantly, the NIS transgene is uniquely characterized by its rapid uptake of radiotracers. The majority of ¹⁸F-TFB uptake by NIS⁺ cells occurs over the first 10 minutes, with some slower uptake occurring up to 30 minutes after the administration of the tracer (33). In the other reporter systems, such as SSTR2 or HSV-1-TK, more than one hour is needed for the tracer to be taken up by the targeted cells. ¹⁸F-TFB, the most commonly used radiotracer for NIS, has a short half-life of 110 minutes which, when combined with its rapid clearance, provides a lower amount of radiation to the patient. The advantages and disadvantages of different reporter systems used in CART-cell imaging have been reviewed (52).

The incorporation of NIS in CART-cell imaging has been reported in two studies (30,53). Both studies validated our findings that NIS is a sensitive reporter platform for CART-cell imaging. Emami-Shahri and colleagues report a preclinical study using NIS for real-time trafficking of PSMA-directed CART-cells. NIS⁺CART-cells were able to eliminate tumor cells in vivo with a high sensitivity. SPECT/CT was used to visualize CART-cells, resulting in lower resolution of assessment of CART-cell trafficking, and no characterization of CART-cell expansion or anti-tumor activity was performed (30). Volpe and colleagues reported the PET imaging of NIS⁺CART-cells in breast cancer models. In our study, we chose to utilize mouse models of hematological malignancies to allow us to study both CART-cell trafficking and their in vivo expansion in toxicity models. We demonstrated that the incorporation of NIS in CART19 or BCMA-CART-cells did not impair their anti-tumor activity or other effector functions. We also showed that NIS is a sensitive and specific platform for CART-cell imaging to detect their trafficking to tumor sites and in vivo expansion, correlating with the development of CRS. The main limitation of the NIS-based imaging platform is its background uptake in some organs, such as the thyroid, stomach, and salivary glands due to their endogenous expression of NIS. This will pose a challenge to the visualization of T-cells trafficking to these organs but does not represent a major issue in patients with hematological malignancies or solid tumors.

In summary, the NIS transporter system provides a sensitive, clinically applicable platform for direct visualization of CART-cells by ¹⁸F-TFB-PET imaging. This platform can be used for dynamic in vivo monitoring of CART-cell expansion and trafficking, as well as for the assessment of CART-associated toxicities.

Synopsis.

Currently, a robust imaging platform to monitor CART-cells is lacking. Imaging using ¹⁸F-TFB-PET is demonstrated to be an efficient, non-invasive technique to monitor CART-cell expansion and trafficking in vivo in multiple tumor models.