In vivo Imaging and Quantitation of Adoptively Transferred Human Antigen-Specific T Cells Transduced to Express a Human Norepinephrine Transporter Gene

Evaluations of biological therapies targeting human malignancies increasingly depend on thorough monitoring of the biodistribution and ultimate disposition of the therapeutic agents following their introduction into the organism and throughout the course of their treatment effects. Molecular gene imaging offers a convenient tool for noninvasive observation of the biological processes in live subjects and also in real time (1). However, effective clinical translation has been subject to several limitations (2), particularly the immunogenicity of the reporter gene product. Use of a protein of human origin as the reporter may circumvent this problem. Several human reporter genes have been developed, including human dopamine D2 receptor (hDDR2), human sodium iodine symporter (hNIS), and human somatostatin receptor subtype 2 (hSSTR2; refs. 3–7). Another is the human norepinephrine transporter (hNET; ref. 8), a protein involved in the transport of norepinephrine analogues into cells. Its properties have been used for heart imaging and for detection of malignancies of neural crest origin (e.g., neuroblastoma and pheochromocytoma; refs. 9, 10). [¹²³I]Metaiodobenzylguanidine ([¹²³I]MIBG) is a standard, clinically approved radiolabeled probe for imaging tissues expressing high levels of hNET (11, 12). Thus, hNET has several potential advantages as a reporter: (a) it is a human protein that should minimize its immunogenicity; (b) a radiolabeled probe for imaging hNET expression is currently used in the clinic; (c) this probe, MIBG, can be radiolabeled with ¹²³I or ¹³¹I for single-photon emission computed tomography (SPECT) and γ-camera imaging and with ¹²⁴I for positron emission tomography (PET) imaging.

In this study, we have examined the feasibility of using the hNET-based reporter system for noninvasive in vivo imaging of human antigen-specific CTLs in a murine model of adoptive immune cell therapy of human EBV-associated lymphoma with [¹²³I]MIBG and SPECT or [¹²⁴I]MIBG and PET. We also evaluated the potential of hNET, when used in combination with a second clinically applicable reporter, for dual-modality imaging to monitor two separate lymphocyte subpopulations in vivo.

Cloning cDNA for hNET and construction of retroviral vectors for directing constitutive expression of hNET. The hNET gene was cloned as described previously (8). Specific primers (GeneLink) were designed for the 3′ and 5′ ends of the cDNA sequences of hNET, described in the sequence database (NM_001043, M65105). hNET cDNAs were amplified by reverse transcription-PCR from the neuroblastoma cell line SK-N-SH (American Type Culture Collection; ref. 8). After separation in agarose gel, a band corresponding to the molecular weight of hNET was extracted, purified, and used for vector production according to techniques described previously (8). Briefly, cDNA fragments were cloned into the pGEM-TEasy vector (Promega) and checked using standard T7 and SP6 primers. The product was ligated with pQCXIX (Clontech). An hNET-IRES fragment was obtained from the resulting pQChNETIX plasmid using BamHI and XhoI endonucleases (New England Biolabs) and cloned into pLEGFP-N1 backbone using the BglII and XhoI insertion sites. The precise vector used for transduction, which is pLhNETiGFP, is produced by insertion of hNET-IRES fragment into pLEGFP-Nq plasmid (Clontech). The DNA obtained was used to transfect a packaging cell line, H29, for subsequent transduction of the retroviral producer cell line, PG-13 (Clontech).

Cell line transduction, selection, and testing. The technique used for transduction of the T cells is based on that described by Pollok et al. (13) as modified for transduction of T cells specifically sensitized with autologous EBV-transformed B cells (EBV-BLCL; refs. 14, 15). Briefly, aliquots of 2 mL/well of isolated T cells prestimulated with autologous irradiated EBV-BLCL for 8 days were cultured for 8 h at a concentration of 1 × 10⁶/mL in nontissue culture six-well plates (Costar) coated with 15 μg/cm² retronectin (recombinant human fibronectin fragment CH-296, Takara Bio, Inc.) together with a 35-fold concentrate of cell-free retroviral vector-containing supernatant from the PG-13 cells at a final concentration of 10⁷ infectious viral particles/mL (multiplicity of infection = 10; as determined by terminal dilution transfection of Jurkat cells) and 8 μg/mL polybrene (American Bioanalytical). After washing and culturing for 12 h, the transduction was repeated. Thereafter, the washed cells were incubated for 3 days in Yssel's medium supplemented with 5% heat-inactivated human serum and 10 units/mL of recombinant human interleukin-2 (IL-2; Chiron, Inc.) at a concentration of 0.6 × 10⁶ T cells/mL. The transduced T cells were then analyzed and sorted in a FACSCalibur sterile sorter (Becton Dickinson).

Separate aliquots of T cells from the same CTL line were simultaneously transduced with a previously described TK-eGFP vector following the same protocol for transduction and sorted by expression of green fluorescent protein (GFP). In separate experiments, subsets of CD4⁺ and CD8⁺ EBV-specific T lymphocytes were fractionated from the parental line by fluorescence-activated cell sorting (FACS) of cells stained with fluorescent anti-CD4 or anti-CD8 antibodies (Becton Dickinson), separately transduced with hNETiGFP and TK-eGFP vectors, respectively, and selected by FACS based on GFP expression.

Expression of hNET was assessed by FACS staining with mouse anti-human NET antibodies (Mab Technologies, Inc.) and secondary goat anti-mouse phycoerythrin (PE)-conjugated antibodies (Becton Dickinson). The level and the rate of transporter activity were evaluated using a standard in vitro uptake assay (2, 14) by measuring the accumulation of specific ([¹²³I]MIBG) and nonspecific [[¹²⁴I]2′-deoxy-2′-fluoro-β-d-arabinofuranosyl-5-iodouracil ([¹²⁴I]FIAU)] radiotracers in hNET-transduced cells (CTL-hNETiGFP) and compared with the uptake of these radiotracers in TK-eGFP–transduced T cells (CTL-TKGFP) isolated from the same EBV-CTL.

To assess the influence of the new reporter gene on transduced T cells, the proliferation and functional activity of CTL-hNETiGFP were compared with those of unmodified CTL or CTL-TKGFP of the same origin. Proliferation was assessed by monitoring the total number of viable cells in the CTL cultures at specified intervals. To assess the influence of specific radiolabels on the EBV-specific cytotoxicity of CTL-hNETiGFP, CTL-TG, and unmodified CTL, each cell type was preincubated for 3 h with 0.1 mCi/mL of [¹²³I]MIBG, [¹²⁴I]MIBG, [¹²⁴I]FIAU, or MIBG alone. EBV-specific cytotoxicity and HLA restriction were defined in standard ⁵¹Cr release assays against a panel of EBV⁺ and EBV⁻ targets sharing or not sharing specific HLA alleles with the T-cell donor as described previously (14) and compared with the responses of the same T cells incubated with medium alone.

Human EBV lymphoma xenograft model. For the in vivo imaging studies, we used a model of adoptive cell therapy previously developed in our laboratory (14, 16). Briefly, as required for specific experiments, groups of at least three and up to five nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice received one or four anatomically separated s.c. inoculations of different human EBV-BLCL (BLCL) or human leukemic cells, varying in expression of EBV and/or HLA alleles (3 × 10⁶ to 5 × 10⁶cells per injection). Sorted populations of CTL-hNETiGFP or CTL-TKGFP were given either intratumorally or i.v. when measurable tumors were detected at all sites. At specified times thereafter, the animals received i.v. infusions of the specific radiolabeled probes and were subsequently imaged by PET or SPECT.

PET imaging was performed on a MicroPET-IIR (Concorde) scanner using 15-min list-mode acquisition of a single static frame (energy range, 420–580 keV). Images were obtained weekly 4 or 24 h after administration of 200 μCi of the ¹²⁴I-labeled radiotracers. To compare MIBG imaging by PET and SPECT, a combination of 200 μCi [¹²⁴I]MIBG and 500 μCi [¹²³I]MIBG was given 4 h before sequential PET and SPECT imaging sessions, respectively. Each PET image was reconstructed using a filtered back projection algorithm. Coronal sections were used to draw regions of interest (ROI) and quantitatively evaluate the radiotracer accumulation.

SPECT images were acquired for 30 min, beginning 4.5 h after administration of 500 μCi [¹²³I]MIBG using an X-SPECT instrument (Gamma Medica) equipped with 2-mm pinpoint (day 1) or LEHR parallel hole collimator (days 1, 8, 21, and 28). SPECT detectors were calibrated for ¹²³I (range, 143–175 keV). The higher dose of [¹²³I]MIBG given and the longer exposure time used for image acquisition were selected to yield images comparable with those generated in PET scans evaluating accumulations of ¹²⁴I-labeled MIBG or FIAU. A structural image of the animal using CT scan was obtained with 128 frames immediately following the SPECT acquisition. Image reconstruction was performed using filtered back projections applying a Butterworth 50 filter 0.4. CT scans were reconstructed using Cobra software. Coregistration of CT and SPECT images was performed on X-SPECT with the vendor's software.

Statistical analysis. The coefficient of determination (R-squared) was used to assess the fit of the regression line to the data points. The R-squared statistic ranges between zero and one, where a value of one indicates a perfect fit of the regression line to the data and a value of zero represents that a linear relationship does not exist between the two variables under study. A P value was generated to provide a level of evidence that a linear relationship exists between each pair of variables. For paired comparisons of samples, the χ² and Fisher's exact tests were applied.

Characterization of EBV-CTL transduced to express hNET reporter gene system. Sequencing analysis showed identity of 5′ and 3′ ends of cloned hNET with the published National Center for Biotechnology Information sequence (M65105). Following pLhNET-iGFP retroviral vector transfection, >25% of primary EBV-CTL expressed hNET and GFP, which were then selected by GFP expression to 98% purity (Fig. 1A).

[¹²³I]MIBG accumulation in hNET-transduced primary human antigen-specific T lymphocytes in vitro. The influx constant K1 was 1.8 ± 0.5 mL/min/g and the steady-state accumulation level, K1/K2, was 58.6 ± 6.8 mL/g. The mean value for the accumulation observed at 60 min in three repeated tests was 58.5 ± 3.0 mL/g. This level of in vitro activity of the hNET reporter was slightly higher than that of HSV-TK, as measured by accumulation of [¹²⁴I]FIAU detected in CTL-TKGFP. We therefore hypothesized that in vivo MIBG uptake of CTL-hNETiGFP would be sufficient to yield images at least comparable with those achievable using [¹²⁴I]FIAU to image HSV-TK–transduced cells (Fig. 1B; ref. 14).

CTL-hNETiGFP proliferated in response to antigen-specific stimulation (Fig. 1D) at levels similar to those detected in unmodified CTL or CTL-TKGFP from the same EBV-specific T-cell line. They also exhibited comparable HLA-restricted EBV-specific cytotoxicity (data not shown). The level of EBV-specific cytotoxicity of CTL-hNETiGFP was not affected by preincubation of the T cells with unlabeled MIBG or MIBG labeled with either ¹²³I or ¹²⁴I when compared with CTL-TKGFP or nontransduced EBV-specific T cells from the same line (Fig. 1C).

Sensitivity of PET and SPECT for imaging hNET⁺ T cells in vivo. We initially wished to determine the number of CTL-hNETiGFP detectable in a targeted tumor in vivo. Accordingly, groups of mice, each bearing a 1 cm³ s.c. EBV-BLCL tumor xenograft, received direct intratumoral injections of doses of 10⁴, 10⁵, 10⁶, or 5 × 10⁶ of EBV-specific CTL-hNETiGFP per tumor (n = 3 mice/CTL dose) in 100 μL of saline followed by an i.v. infusion of a combined dose of 200 μCi [¹²⁴I]MIBG and 500 μCi [¹²³I]MIBG. Four hours thereafter, the mice were imaged by PET for 15 min and, subsequently, by SPECT for 30 min.

In these experiments, we also wished to gain an initial assessment of the capacity of PET to image EBV-specific CTL-hNETiGFP accumulating [¹²⁴I]MIBG when compared with EBV-specific CTL-TKGFP, which, as we previously reported (14), can be imaged based on their accumulation of [¹²⁴I]FIAU. Accordingly, we treated separate groups of NOD/SCID mice bearing EBV-BLCL tumor xenografts with intratumoral injections of EBV-specific CTL-TKGFP from the same parental T-cell line at doses of 10⁴, 10⁵, and 10⁶ T cells per tumor (n = 3 mice/T-cell dose) followed by i.v. infusion of 200 μCi [¹²⁴I]FIAU and, 4 h thereafter, imaging by PET.

As shown in Fig. 2A, the minimal dose of CTL-hNETiGFP detected by PET was 10⁴ cells. The signal intensity recorded in the tumors by PET and the well count activity detected in the tumors isolated after autopsy (Fig. 2B) correlated closely with the dose of T cells injected (P < 0.001). As shown in Fig. 2C, PET images of mice injected with graded doses of EBV-specific CTL-TKGFP were also able to distinguish tumors injected with as few as 10⁴ T cells (Fig. 2C). The dose of CTL-TKGFP inoculated was also significantly correlated with tumor uptake of [¹²⁴I]FIAU (Fig. 2D). However, the steeper slope of the plot in Fig. 2D suggests that PET imaging of CTL-TKGFP may be more sensitive for quantitations of T cells accumulated.

Of note, the whole-body PET images obtained within 4 h after i.v. infusion of each radiotracer also showed high level of radioactivity in the gut and urinary tract caused by excretion of the ¹²⁴I metabolites (Fig. 2A and C). These nonspecific accumulations were detected both in the animals that had received intratumoral injections of T cells and in control animals that had not received injections of T cells. Although these nonspecific accumulations can be avoided by obtaining PET images 24 h after infusion of the radiotracer (additional images not shown; ref. 14), images obtained within 4 h of inoculation of T cells and infusions of the radiotracer were used to estimate the numbers of T cells to minimize inaccuracies potentially caused by local diffusion or systemic redistribution of the intratumorally injected T cells through the circulation or by replication or apoptosis of the T cells after interaction with their targets.

Subsequent X-SPECT images of the injected tumors obtained 4.5 h after the coadministration of 500 μCi/mouse [¹²³I]MIBG and 200 μCi/mouse [¹²⁴I]MIBG indicated that accumulation of as few as 10⁴ hNETiGFP⁺ T cells could also be imaged by this modality (Fig. 3A). Application of dual-modality imaging with CT overlapping the SPECT image permitted exact localization of the increased radiotracer accumulation area to the tumor. Again, measurements of signal intensity showed a direct correlation between signal intensity and dose of T cells injected in each tumor (P < 0.001; Fig. 3B). As expected, and consistent with our prior studies (14), there was also a close correlation between the amount of radiotracer detected by the γ-spectrometer well counter in the tumors and the dose of hNET⁺ T cells previously injected into the tumor.

Measurements of accumulation of EBV-specific CTL-hNETiGFP in a target tumor over extended time intervals. In our prior studies, we had shown that accumulations of EBV-specific T cells transduced to express HSV-TK in targeted xenografts could be imaged in vivo over a 2-week period by sequential infusions of [¹²⁴I]FIAU followed by PET. To assess the utility of hNET as a reporter for long-term monitoring of EBV-specific T cells, EBV-specific CTL-hNETiGFP at doses of 12 × 10⁶ CTL-hNETiGFP per mouse was infused i.v. into five mice bearing a single s.c. tumor xenograft of autologous EBV-BLCL. At 1, 8, and 28 days after T-cell infusion, the animals were injected with 200 μCi/mouse [¹²⁴I]MIBG. Four hours thereafter, they were imaged by PET.

As shown in Fig. 4A, a progressive increase in the PET signal was detected in the tumors over 28 days. The number of T cells accumulated in the targeted tumors was estimated by calculation of the number of T cells in the tumors based on the radiotracer accumulation per dose of T cells plot in Fig. 2B and confirmed by quantitation of the radioactivity in the targeted tumors and by enumeration of the absolute number of GFP⁺ human T cells in the targeted tumor (Fig. 4B). The initial accumulation of T cells in the tumors, calculated from the PET signal intensity of [¹²⁴I]MIBG activity as described earlier (Fig. 2A), was ∼0.5 × 10⁶ T cells and increased to ∼6 × 10⁶ T cells by day 28. As shown in Fig. 4B, there was a direct and significant correlation between cell numbers in each tumor calculated based on well count activity of [¹²⁴I]MIBG in the targeted tumor and the signal accumulation recorded for each tumor by PET imaging (P < 0.001). The number of cells in the tumors calculated based on radioisotope accumulation was also significantly correlated with the number of GFP⁺ T cells detected in the tumor (R² = 0.73; P = 0.043).

PET and SPECT imaging of targeted accumulations of virus-specific CTL-hNETiGFP in EBV⁺ tumors over time. To evaluate the antigen-targeted accumulation of EBV-specific CTL-hNETiGFP over time, we treated a group of five mice, each bearing four distinct anatomically separated EBV⁺ and EBV⁻ tumor xenografts differing in their HLA genotypes, with i.v. infusions of 12 × 10⁶ EBV-specific CTL-hNETiGFP per mouse. On days 1, 8, 21, and 28 thereafter, the mice received a combined infusion of 500 μCi [¹²³I]MIBG and 200 μCi [¹²⁴I]MIBG (days 1, 8, and 28); due to limited availability of ¹²⁴I, only [¹²³I]MIBG was given on day 21. PET and/or SPECT images were obtained 4 and 4.5 h after infusion of the radiolabeled MIBG, respectively. In this experiment, the EBV-specific CTL-hNETiGFP used contained a predominance of CD8⁺CD4⁺ double-positive CD3⁺ T cells restricted by HLA-A0201 together with CD4⁺ T cells restricted by HLA-DRB₁0402. As shown in the SPECT and SPECT-CT fusion images in Fig. 5A as well as the PET images in Fig. 5B, radiolabeled MIBG was selectively and progressively accumulated in the HLA-A0201⁺EBV⁺ tumor (T1) from day 1 to day 28. In contrast, accumulations of [¹²³I]MIBG and [¹²⁴I]MIBG in the HLA-DRB₁0402⁺EBV⁺ tumor (T4) were not detected by either SPECT or PET until day 28 after infusion. Quantitations of radioactivity accumulated in the tumors and normal tissues obtained at autopsies on day 28 (Fig. 5C) showed significantly increased uptake of MIBG in the T1 tumor as well as a less striking increase of label in the T4 tumor. In contrast, labeled MIBG did not accumulate in the allogeneic EBV⁺ (T2) or HLA-A0201⁺DRB₁0402⁺EBV⁻ tumor (T3). As expected, uptake of MIBG was increased in the heart and adrenal glands relative to normal tissues. Indeed, the accumulation of labeled MIBG in the T4 tumor detected at day 28 was only moderately greater than that detected in the heart and adrenal glands. Taken together, these studies of EBV-specific CTL-hNETiGFP confirm and extend our prior studies showing selective accumulation of EBV-specific, HLA class I–restricted CD8⁺ T cells in tumors expressing EBV and the restricting HLA allele (14). Sequential imagings showed the ability of HLA class II–restricted CD4⁺ T cells to accumulate in tumors expressing EBV and their restricting allele (i.e., DRB₁0402) but also showed that limited population of antigen-specific T cells may not be imaged in targeted tissues until they have expanded to sufficient numbers to permit their detection.

Dual-modality imaging. Based on the comparable imaging of CTL-hNETiGFP and CTL-TKGFP, we explored whether these two reporters could be simultaneously used to evaluate the migration of two separate T-cell populations in the same animal.

As shown in Fig. 6A, after coinfusion of [¹²⁴I]FIAU and [¹²³I]MIBG, PET showed selective [¹²⁴I]FIAU accumulation by CD8⁺CTL-TKGFP in the EBV⁺HLA-A0201⁺ tumor (T1). No signal was detected in the T2 tumor. Thus, PET scan specifically registered signals from HSV-TK⁺CD8⁺ T cells accumulating [¹²⁴I]FIAU but not from CD4⁺ T cells accumulating [¹²³I]MIBG, reflecting the ability of PET to exclude the low-energy γ-rays emitted by [¹²³I]MIBG. In contrast, although SPECT imaging showed maximal signal in the HLA-DRB₁0701⁺EBV⁺ tumor (T2) targeted by the CD4⁺ EBV-specific CTL-hNETiGFP, reflecting their accumulation of [¹²³I]MIBG, low levels of radioactivity were also detected in the T1 tumor, caused by coregistration of the downscatter of the 511-keV photons, produced by [¹²⁴I]FIAU positron decay, into the SPECT imaging energy window used to measure uptake of ¹²³I.

To avoid this downscatter and to differentially image the two types of T cells, mice bearing xenografts of the same two EBV-BLCLs were coinfused with hNET/GFP⁺CD4⁺ and HSV-TK/GFP⁺CD8⁺ T cells derived from the same EBV-specific T-cell line. Thereafter, infusions of [¹²³I]MIBG and [¹²⁴I]FIAU and associated SPECT and PET imaging were performed sequentially. Initially, the mice received [¹²³I]MIBG and were imaged by SPECT 4 h later (Fig. 6B). This image detected selective accumulation of [¹²³I]MIBG in the EBV⁺DRB₁0701⁺ T2 tumor expected to be targeted by the hNET⁺CD4⁺ T cells. Thereafter, each mouse was given 200 μCi [¹²⁴I]FIAU and imaged by PET 2 h later. In these PET images, radiolabel was selectively detected in the EBV⁺HLA-A0201⁺ tumor (T1) that is targeted by the HLA-A0201–restricted CD8⁺ EBV-specific T cells (Fig. 6B). Selective accumulation of CD8⁺ or CD4⁺ T cells in T1 and T2 tumors, respectively, was confirmed by FACS analysis of T cells isolated from each tumor (Fig. 6C). Thus, this sequential imaging approach permitted clear differentiation of the selective accumulations of two separate T-cell populations transduced to express distinguishable reporter proteins in the same mouse within 6 h on the same day.

Currently, several reporter gene systems are being used in preclinical models to transduce normal or malignant cells so as to be able to image their migration, growth, and function in vivo (14, 17, 18). However, the clinical applicability of many of these reporters is limited because the proteins they encode are immunogenic in man. Indeed, immune responses to certain reporter gene products have developed early after adoptive transfer of transduced cells and compromised evaluations of their tissue distribution and function (19, 20).

To address this limitation, attention has been focused on identification of human reporter genes encoding enzymes, receptors, or transport proteins with restricted expression in normal tissues, which can mediate differential uptake of radiolabeled drugs or biologicals. Three such genes have been reported previously: (a) hDDR2, by which cells differentially accumulate 3-(2′-[¹⁸F]fluoroethyl)-spiperone (3, 6); (b) hSSTR2, expressed by carcinoid tumors, which differentially binds the somatostatin analogue [¹¹¹In]diethylenetriaminepentaacetic acid-octreotide (4); and (c) hNIS, which mediates differential cellular uptake of either radiolabeled iodide or pertechnetate (2, 5).

In this report, we evaluated the in vivo imaging of adoptively transferred lymphocytes transduced to express hNET, a human protein also unlikely to be immunogenic in man. The hNET protein is normally expressed on the plasma membrane of noradrenergic neurons at their synapses and acts to transport norepinephrine by endocytosis from the synaptic cleft back into the neuron. Although these transporters are normally expressed throughout the sympathetic system, cells expressing hNET are particularly concentrated in the adrenal glands and the heart (12).

In 1967, Morales et al. (21) first showed that [¹⁴C]norepinephrine, infused into dogs, selectively accumulated in the adrenal medulla. Subsequently, radiolabeled MIBG, a structural analogue of norepinephrine that can block its neurotransmitter activity, was also shown to accumulate in the adrenal gland and could be used to image this organ (11). In initial clinical trials, it was shown that [¹³¹I]MIBG could be used to detect pheochromocytomas (9, 10). It is now also extensively used for imaging and treatment of neuroblastoma (22). In these tumors, uptake of [¹³¹I]MIBG is based on their high level of expression of functional norepinephrine transporters.

For our study, we constructed a bicistronic retroviral vector directing the constitutive expression of hNET and GFP and transduced it into EBV-specific T cells of defined antigen specificity and HLA restriction. Following transduction, T cells expressing the vector-encoded GFP could be isolated in high purity by FACS. In vitro studies showed highly reproducible kinetics of radiotracer uptake in transduced T cells. Exposure of hNET-transduced EBV-specific T cells to MIBG or to MIBG labeled with either ¹²⁴I or ¹²³I did not alter their cytotoxic activity, confirming our previously published data that showed that the exposure of HSV-TK–transduced EBV-specific T cells to ¹²⁴I- or ¹³¹I-labeled FIAU did not alter their EBV-specific HLA-restricted cytotoxicity in vitro or tumoricidal effect in vivo (14).

In prior studies using this model, we were able to image EBV-specific HSV-TK/GFP–transduced T cells accumulating in targeted tumors by i.v. infusions of [¹²⁴I]FIAU followed by PET (14). To gain initial estimation of the relative capacity of PET to detect transduced T cells, defined doses of EBV-specific CTL-hNETiGFP and CTL-TKGFP were directly injected into EBV-BLCL tumor xenografts and then imaged after i.v. infusion of [¹²⁴I]FIAU or [¹²⁴I]MIBG. Signal intensities detected by PET and accumulations of [¹²⁴I]MIBG or [¹²⁴I]FIAU, counted in the tumor sites, were closely correlated with the number of CTL-hNETiGFP or CTL-TKGFP T cells injected (Fig. 2). Furthermore, doses of as few as 10⁴ CTL-hNETiGFP or CTL-TKGFP could be clearly imaged. When mice were given [¹²³I]MIBG i.v. and imaged by X-SPECT, signal intensity in the tumors was also closely correlated with dose of T cells down to the lowest dose tested, 10⁴ T cells per tumor (Fig. 3). Fusion of CT and SPECT images permitted improved precision in the localization of the radiotracer accumulation sites.

Our group first showed the potential of PET imaging for long-term monitoring of antigen-specific T cells in vivo in studies of the migration of human EBV-specific T cells transduced to express an HSV-TK/GFP reporter gene (14). When transferred i.v. into NOD/SCID mice bearing EBV⁺ lymphomas and EBV⁻ leukemia xenografts of specific HLA genotypes, EBV-specific HSV-TK/GFP T cells selectively accumulated in and induced regressions of EBV⁺ tumors coexpressing the restricting HLA allele. Sequential PET images obtained after weekly i.v. injections of [¹²⁴I]FIAU recorded these targeted accumulations of T cells over a period of 15 days after T-cell infusion.

The current study confirms these findings, showing, both by PET and by SPECT imaging, progressive accumulations of hNET⁺ EBV-specific T cells in targeted EBV⁺ tumors over a period of 28 days. The present study also provides evidence suggesting that image intensities recorded in vivo by PET and SPECT can be used to estimate the number of T cells accumulating in targeted tumors over time. This is shown by the significant correlations between the number of T cells actually accumulated in the targeted tumor, as measured by quantitation of GFP⁺ T cells and measurements of radiolabel accumulation in tumors isolated at autopsy, and the number of T cells in the tumor calculated based on the signal intensities measured in PET images of graded doses of T cells directly injected into targeted tumors in vivo. These results are consistent with the studies of Su et al. (23) who measured uptake of [¹⁸F]FHBG in mice injected intratumorally with nonspecifically activated murine T cells transduced to express HSV-TK. Based on both measured and calculated numbers of T cells accumulated, T cells increased in the targeted tumor by over 12-fold over the 28 days of observation. In the EBV-specific T-cell line used for this experiment, 14% of the T cells were found to secrete IFN-γ in response to autologous EBV-BLCL. If this represents the true proportion of functional EBV-specific T cells in the line, it can be estimated that 30% of the EBV-specific T cells infused i.v. accumulated in the targeted tumor by day 1 after infusion. The 12-fold increase in T cells detected by day 28 after infusion supports the hypothesis that this progressive accumulation is based largely on proliferation of these T cells at the tumor site.

Recently, other studies in murine tumor models have evaluated the short-term migratory patterns of unsensitized and antigen-specific T cells in vivo using PET to image T cells transduced to express HSV-TK (17, 24) or bioluminescence to track T cells transduced to express luciferase (25, 26). In these studies, T-cell accumulations in targeted tumors have initially been evaluated and detected 3 to 4 days after adoptive transfer into the tumor-bearing mice and have increased over a subsequent 1 to 2 weeks of observation (24). In studies using PET, quantitation of signal intensity emanating from the targeted tumors and other organs has been closely correlated with levels of radiolabel detected in each tissue. However, only one other study has directly correlated signal intensities over time with quantitation of transduced T cells detected in the tissue (24). In that study, Su et al. used sequential PET imaging over 15 days to follow naive and memory-type ovalbumin-specific CD8⁺ T cells derived from OT-1 mice and expressing a mutant sr39-HSV-TK and GFP, after adoptive transfer into mice, bearing transgenic tumors expressing ovalbumin. In that study, memory T cells rapidly accumulated in and eradicated antigen-positive tumors grown in RAG^−/− mice. Although naive T cells from these Tcr transgenic mice could also eradicate these tumors, their accumulation at tumor sites was delayed and quantitatively less than that of the memory T cells. The concentration of T cells required to induce tumor regression was estimated to be 2 × 10⁶ to 4 × 10⁶ CD8⁺ T cells per gram of tumor tissue. Strikingly, such expansions of OT-1 CD8⁺ T cells to numbers inducing tumor regression occurred in lymphopenic RAG-1^−/− but not in normal wild-type hosts.

The potential utility of in vivo imaging for distinguishing the biological activities of T-cell populations transferred was also highlighted in experiments presented in Fig. 5, which used a line of hNETiGFP⁺ EBV-specific T cells containing a predominance of HLA-A0201–restricted, EBV-specific double-positive T cells and a small subpopulations of HLA-DRB₁0402–restricted, EBV-specific CD4⁺ T cells. Prior studies have indicated that CD4⁺CD8⁺ double-positive human T cells, when detected in the blood, usually represent CD8⁺ T cells, which up-regulate expression of CD4 during activation in vivo (26). Recent studies have provided evidence that these CD4⁺CD8⁺ T cells are potent effector memory T cells that are HLA class I restricted and prominent in individuals chronically infected with HCV or EBV and are particularly concentrated in sites of chronic infection (27). Our findings are consistent with those characterizations, showing, both by PET and SPECT imaging, rapid accumulation of the CD4⁺CD8⁺ T cells in EBV⁺ xenografts coexpressing the restricting HLA-A0201 allele. In contrast, the late accumulation of T cells in the EBV⁺ xenograft sharing the class II–restricting allele HLA-DRB₁0701 likely reflects the small population of EBV-specific cytotoxic CD4⁺CD8⁻ T cells restricted by this allele detected in the T cells infused.

Because interactions between different types of lymphocytes contribute to the kinetics, intensity, antitumor activity, and persistence of adoptive immune responses, a priority for our studies has been the development of technology permitting contemporaneous imaging of such interactions in vivo. In prior studies, Beilhack et al. (28) used bioluminescence imaging to examine the distribution and proliferation of splenic T cells expressing an in vivo luciferase transgene following transplantation into lethally irradiated H-2 disparate allogeneic hosts. During the course of graft-versus-host disease (GVHD), targeted tissues identified by their accumulations of luciferase⁺ T cells were evaluated by immunohistochemistry for their content of different CD4⁺ and CD8⁺ T cells to distinguish the potential of alloresponsive naive and memory T cells to induce GVHD in vivo. However, this single reporter system did not permit continuous monitoring of each T-cell compartment in vivo. Alternative approaches using vectors encoding different bioluminescent reporters [e.g., firefly, Renilla reniformins, and click beetle luciferases (25)], distinguishable by the substrates used and/or by the spectra of light emitted, have been used to contemporaneously monitor more than one transduced cell population in vitro and have been proposed for contemporaneous imaging of two to three cell populations in vivo. Development of such approaches is highly desirable because imaging by bioluminescence can be performed repeatedly at low cost and the background in tissues not containing cells expressing luciferase reporters is extremely low. However, to date, no studies have been reported in which different bioluminescent reporters have been used to successfully discriminate distinct cell populations in vivo. Furthermore, because the transmission of different light spectra through tissues differs, the accuracy of estimates of cell populations expressing certain luciferases when they accumulate in sites distant from the body surface may be limited.

In this report, we show that two T-cell populations bearing hNET and HSV-TK reporter genes can be contemporaneously imaged in vivo over extended intervals using SPECT to detect differentially accumulated [¹²³I]MIBG and PET to detect [¹²⁴I]FIAU. Because PET is more sensitive than SPECT when equivalent doses of radiolabel are used, we intentionally increased the ratio of doses of ¹²³I to ¹²⁴I in favor of ¹²³I-labeled probes and also increased the acquisition time for SPECT images. This approach permitted achievement of a comparable sensitivity of detection in images of the same CTL-hNETiGFP with [¹²³I]MIBG and [¹²⁴I]MIBG (Figs. 2 and 3) or equivalent doses of CTL-hNETiGFP and CTL-TKGFP with [¹²³I]MIBG and [¹²⁴I]FIAU, respectively.As a result of this modification, in mice bearing EBV⁺ xenografts differing in their expression of class I and II HLA alleles, we could demonstrate, by SPECT, that the hNET⁺ EBV-specific CD4⁺ T cells retaining [¹²³I]MIBG selectively accumulated in EBV⁺ tumors coexpressing their restricting HLA-DRB₁0701 allele. In the same mice, HSV-TK⁺ EBV-specific CD8⁺ T cells, which selectively take up [¹²⁴I]FIAU by subsequent PET imaging, were shown to have accumulated in EBV⁺ tumors expressing HLA-A0201. This sequence of imaging permits clear differentiation and quantitation of the accumulated activity of i.v. given [¹²³I]MIBG and [¹²⁴I]FIAU in the hNET-transduced CD4⁺ and HSV-TK–transduced CD8⁺ T cells, respectively. Although the possibility of such dual imaging has been raised by several groups interested in in vivo imaging techniques (29, 30), this report is, to our knowledge, the first to show its feasibility. Furthermore, because our studies also indicate that the numbers of cells detectable by PET and SPECT are comparable, this approach may be of particular value for accurately measuring the accumulations of two different effector cell populations within a living subject. Both reporter systems used could also be applied clinically. However, additional preclinical trials beyond the scope of the current experiments as well as clinical trials will be required to ascertain the relative merits of each reporter and imaging modality for sequential long-term assessments of adoptively transferred cells in immunocompetent or immunoablated subjects.

Recent reports of leukemias developing in transduced cells in children with X-linked SCID late after treatment with autologous hematopoietic progenitor cells transduced with a retroviral vector encoding a normal IL-2 receptor (γ-chain) have raised concerns about the tumorigenic potential of cells transduced with such vectors (31, 32). Although long-term analyses of retrovirally transduced T cells following adoptive transfer into human patients and mice have thus far not detected either a dominance of integration sites or malignant transformation of transduced cells (33–35), such concerns may limit broad use of such methods for imaging T cells until further and more extended safety data are accrued. Alternatively, other viral vectors or more efficient nonviral strategies, which may reduce or minimize risk of transformation, may be used to introduce reporter genes into these cells (36, 37).

In conclusion, our studies indicate that the hNET vector described induces a level of expression of the hNET that permits effective imaging, with either [¹²³I]MIBG and SPECT or [¹²⁴I]MIBG and PET of as few as 10⁴ T cells accumulated in a 1 cm³ tumor nodule. In these studies, the sensitivity of detection was comparable with that achieved by in vivo imaging of T cells transduced to express HSV-TK with PET after i.v. infusion of [¹²⁴I]FIAU. We also show the feasibility of contemporaneously imaging the in vivo biodistribution and targeted accumulation of two effector cell populations in the same animal by separately transducing each effector population to express a distinct reporter protein mediating the uptake of a radiolabeled tracer differentially imaged by SPECT and PET. Lastly, the close correlation observed between quantitation of signal intensity by PET and SPECT and the numbers of T cells directly injected into tumors or detected in targeted tumors after accumulation over time indicates the potential of such imaging strategies for quantitatively estimating the effector cell populations required for effective adoptive immunotherapy.

Grant support: NIH grants CA23766, CA59350, and CA106450; In vivo Cellular and Molecular Imaging Center grant 2P50 CA086438-06; Claire L. Tow Chair in Pediatric Oncology Research; Ryan E. McGeough Charitable Gift Fund; Banbury Fund; Burton Abrams Charitable Trust; Larry H. Smead Fund (97-98); and Aubrey Fund For Pediatric Cancer Research.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Anna Bierwiazconek for excellent technical assistance, Dr. Ronald G. Blasberg for assistance in radiotracer kinetic modeling, and Glen Heller for assistance in statistical analysis.