Abstract
Molecular imaging methods have been used recently to investigate biologic events. To develop a molecular imaging method suitable for monitoring viable cancer cells, we made a dual-imaging reporter gene system and examined the correlation between cancer cell number and signals from 2 reporter genes, sodium iodide symporter (NIS) and luciferase. Methods: NIS and luciferase genes were linked with the internal ribosomal entry site and transfected into SK-HEP1 cells to generate SK-HEP1-NL cells. 125I uptake assays, luciferase assays, and scintigraphic and luminescence imaging were performed on SK-HEP1-NL cells. After treating with doxorubicin, cell counting, assays, and imaging were performed. SK-HEP1 and SK-HEP1-NL cells were inoculated subcutaneously into the flanks of nude mice. After incubation, scintigraphic and luminescence images were acquired and quantitated. Results: The results of radioiodide uptake, luciferase assay, and scintigraphic and luminescence imaging in vitro correlated well with viable cell numbers. Upon increasing the concentration of doxorubicin, cell numbers decreased, and this correlated with a decrease in radioactivity and luminescence intensity. The radioactivity from in vivo scintigraphic images and the intensity from luminescence images were also found to be proportional to the tumor weight. Conclusion: The developed dual-reporter imaging method using NIS and the luciferase gene reflected viable cancer cell numbers and could detect changes in cell number after doxorubicin treatment.
It is important to be able to determine viable cancer cell numbers and to monitor changes after therapeutic intervention in the fields of cancer research and clinical oncology. Though many investigations have been performed on cancer biology and on anticancer drug development, few methods are available for monitoring the therapeutic effect (e.g., 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyltetrazolium bromide (MTT) assay for metabolism, thymidine uptake assay for proliferation (1), and trypan-exclusion cell count for viable cell count). Moreover, these conventional assay tools can provide useful information only at the in vitro level.
Imaging methods based on optical systems, such as bioluminescence and fluorescence, can play a role in detecting viable cancer cells and in evaluating the kinetics of tumor-cell clearance after anticancer therapy in vitro and in vivo (2–5). The advantages of this approach are (a) the simplicity of quantifying the viable cell count, (b) the possibility of acquiring noninvasive and real-time data, (c) the possibility of making multiple determinations during the disease course in a single animal, (d) the possibility of optimizing therapy (choice of modality, combination therapy, route of administration) because external signals are proportional to the tumor-cell burden, and (e) the ability to detect small cell numbers (>2,500 tumor cells). Despite of these advantages, optical imaging methodologies are limited in terms of clinical application because their tissue-penetrating abilities are poor.
The sodium iodide symporter (NIS) is one nuclear imaging reporter gene that could be used to overcome the limitations of optical imaging strategies. NIS is a transmembrane protein, which actively transports iodide ions into thyroid cells (6). The accumulation of radioiodine in specific cells or tissues mediated by the NIS gene transfer shows its possibilities for molecular imaging (7–12). An imaging system based on the NIS gene can produce images with commonly used radionuclides, such as 123I, 125I, or 99mTc-pertechnetate, and simple γ-camera systems.
The present study was undertaken to develop a noninvasive method for monitoring viable cancer cell numbers using the dual-reporter gene and to investigate the correlation between cell number and reporter gene expression both in vitro and in vivo. We constructed a reporter vector that linked the NIS and luciferase using the internal ribosomal entry site (IRES). After producing a stable cell line by transferring the reporter vector, the signals of both reporters were evaluated using radionuclide-based and optically based methods in vitro and in vivo.
MATERIALS AND METHODS
Recombinant DNA Plasmid Construct
pIRES-NL, containing human NIS (hNIS) and luciferase, was made by inserting hNIS and luciferase into pIRES vector (Clontech). hNIS, cloned into pCDNA3 (Invitrogen), was kindly provided by Dr. Sissy Jhiang (Ohio State University). The coding sequence fragment of hNIS was cloned using the polymerase chain reaction with forward primer 5′-CGGCTCGAGATGGAGGCCGTGGAGACC-3′ and reverse primer 5′-GGCCACGCGTTCAGAGGTTTGTCTCCTGCT-3′. The fragment was inserted upstream of the IRES sequence in pIRES vector after digestion with Xho I and Mlu I (Intron). Luciferase was cloned from pGL2 vector (Promega) into downstream of the IRES sequence using the same method as that used for hNIS except for the use of forward primer 5′-CCGGGTCGACATGGAAGACGCCAAAAACAT-3′, reverse primer 5′-CCGGGCGGCCGCTTACAATTTGGACTTTCCGC-3′, restriction enzyme Sal I, and Not I (Intron).
Cell Culture and Stable Cell Line Generation
The human hepatocellular carcinoma cell line SK-HEP1 and transfectants were cultured in RPMI 1640 medium (WelGENE, Inc.) supplemented with 10% fetal bovine serum and 1% penicillin and streptomycin solution (GIBCO). After pIRES-NL had been transfected into SK-HEP1 cells using the liposome formulation Lipofectamine plus (Invitrogen), according to the manufacturer’s instructions, SK-HEP1-NL cells expressing NIS and luciferase stably were generated at under 1,000 μg/mL G418 (Invitrogen) selection.
Radioactive Iodide Uptake Assay
Radioactive iodide uptake studies were performed at 37°C using a modification of the method described by Weiss et al. (13). A dilution series of SK-HEP1 and SK-HEP1-NL cells were inoculated into 24-well plates over a range of 3 × 104 to 5 × 105 cells per plate. After a 12-h incubation, the 125I uptake level was examined. The iodide uptake level was determined by incubating the cells with 500 μL Hanks’ balanced salt solution (HBSS) containing 0.5% bovine serum albumin, 10 mmol/L 2-(4-[2-hydroxyethyl]-1-piperazinyl)ethanesulfonic acid-NaOH, pH 7.4, and 3.7 kBq (0.1 μCi) Na125I at 37°C for 20 min. After incubation, the cells were washed twice as quickly as possible (<15 s) with 2 mL of iodide-free ice-cold HBSS buffer. Cells were detached with 500 μL trypsin, and the radioactivity was measured using a γ-counter (Canberra Industries).
In Vitro γ-Camera Imaging
The uptake of 99mTc-pertechnetate, another radiotracer transported by NIS, was determined using the same procedure as that used for radioiodine uptake. After incubation with 370 kBq (10 μCi) 99mTc-pertechnetate, the cells were washed twice as quickly as possible with 2 mL of radiotracer-free ice-cold HBSS. Afterward, the 24-well plate was placed under a γ-camera (ON 410; Ohio Nuclear) equipped with a pinhole collimator and scanned for 5 min.
Luciferase Assay
For the in vitro luciferase assay, a dilution series of SK-HEP1 and SK-HEP1-NL cells were inoculated into 24-well plates over a range of 3 × 104 to 5 × 105 cells per plate. After a 12-h incubation, each well was washed with phosphate-buffered saline (PBS). The luciferase assay was performed using a Luciferase Assay Kit (Applied Biosystems). Lysis solution was added to each well, and cells were detached from the plate with a cell scraper. The cell lysate was then transferred to a microplate and bioluminescence was measured using a microplate luminometer (TR717; Applied Biosystems)
In Vitro Bioluminescent Imaging
For in vitro bioluminescent imaging, a dilution series of SK-HEP1 and SK-HEP1-NL cells were inoculated into 24-well plates over the range 3 × 104 to 5 × 105 cells per plate. After a 12-h incubation, each well was washed with PBS. Bioluminescent images were acquired using a Bright-Glo luciferase assay system (Promega) and LAS-3000 equipped with a cooled charge-coupled device (CCD) camera (Fuji Film). According to the manufacturer’s instructions, 100 μL of a mixture of luciferin and buffer solution was added on each well. After incubation for 10 min, the plate was placed in the dark box and the gray-scale image was acquired. Bioluminescence was collected for 5 min. Rainbow-scale pseudocolor images were converted and superimposed over the gray-scale-shape image and processed using graphic software (Photoshop; Adobe). Quantification analysis of the acquired bioluminescent images was performed using an image analysis program, Multi Gauge vs 2.02 (Fuji Film). To prevent distortion of data, quantification analysis was performed under no modification of the raw image. The optical density (OD) of the region of interest (ROI) on the gray-scale image was measured by the image analysis program. The same size of ROI was selected for comparison, and the OD of the background was subtracted on each well plate.
In Vitro Assays After Doxorubicin Treatment
Doxorubicin was used to evaluate the therapeutic effect. SK-HEP1 and SK-HEP1-NL cells were inoculated into a 24-well plate (5 × 105 cells per well). After 12 h, the medium was replaced with fresh medium containing 0, 0.5, 1, 2, 5, and 10 μg/mL of doxorubicin and cells were incubated in 5% CO2 for 24 h at 37°C. Viable cells were then counted using the trypan blue–exclusion method. The bioluminescent assay, bioluminescent image acquisition, radioiodide uptake assay, and nuclear image acquisition were performed according to the methods described with minor modifications. The correlation between cancer cell numbers and measured values was performed using Excel 2002 (Microsoft).
Animal Experiments
Animal experiments were performed with the approval of the Seoul National University Animal Research Committee. Male BALB/c nude mice, 3 wk old, were maintained at our facility under specific pathogen-free conditions. SK-HEP1 and SK-HEP1-NL cells were xenografted subcutaneously into the left fore flank (SK-HEP1; 5 × 106 cells), right fore flank (SK-HEP1-NL; 5 × 106 cells), left hind flank (SK-HEP1-NL; 1 × 107 cells), and right hind flank (SK-HEP1-NL; 5 × 106 cells) of each nude mouse. After allowing 2 wk for tumor growth, bioluminescent and γ-camera images were acquired. Anesthesia was performed by an intraperitoneal injection of 40 μL of ketamine and xylazine (4:1) solution. Luciferin was prepared by dissolving the luciferin potassium salt (Molecular Probes). An aqueous solution of luciferin was injected into the peritoneal cavity 10 min before imaging, and the animal was placed in the dark box of LAS-3000 film (Fuji Film). After gray-scale images under white light had been acquired, bioluminescent images were acquired by collecting and integrating the light for 10–15 min. Image processing and analysis were performed as described previously, and the γ-camera images were acquired. Thirty minutes after injecting 11.1 MBq (300 μCi) of 99mTc-pertechnetate into the tail vein, mice with tumors were placed in a spread-prone position and scanned with a γ-camera (ON 410; Ohio Nuclear) equipped with a pinhole collimator.
RESULTS
Correlation Between Radioactivity and Cell Number
The SK-HEP1-NL cell line, which stably expresses NIS and luciferase, was generated for this study. Figure 1 shows the result of radioiodide uptake and γ-camera images in a dilution series of SK-HEP1-NL and SK-HEP1 cells. As the numbers of SK-HEP1-NL were increased, radioiodide uptakes were also increased (Fig. 1A). A high correlation (r2 = 0.9987) was observed between radioiodide uptake and cell number (Fig. 1B). Increased radioactivity was also observed in γ-camera images. According to increases in the numbers of SK-HEP1-NL cells, the accumulation of 99mTc-pertechnetate also increased (Fig. 1C). An analysis of intensities in the ROI indicated a high correlation (r2 = 0.9775) between intensity and cell number (Fig. 1D).
In vitro radioactivity assay according to cancer cell number. After inoculating a dilution series of SK-HEP1-NL cells (NIS expressing) and SK-HEP1 cells (control), radioiodine uptake assay and in vitro γ-camera image acquisition were performed. (A) Result of 125I uptake shows similar accumulations regardless of SK-HEP1 cell number. In contrast, SK-HEP1-NL cells showed a significant increase of 125I uptake according to cell number increase. (B) Linear regression analysis indicated high correlation between cell number and 125I uptake. Correlation coefficient was r2 = 0.9987. (C) 99mTc uptake in a dilution series of SK-HEP1-NL cells was visualized using a γ-camera. Higher numbers of cells (right line) showed more intense 99mTc uptake. (D) Linear regression analysis indicated high correlation between cell numbers and intensity of ROI in image (r2 = 0.9775). OD of ROI was determined using image analysis program.
Correlation Between Bioluminescence and Cell Number
Figure 2 shows the result of bioluminescence and bioluminescent images produced using a luminometer and a highly sensitive cooled CCD camera, respectively. As the numbers of SK-HEP1-NL cells were increased, the bioluminescence increased (r2 = 0.9979); SK-HEP1 showed a very low level of luminescence in cells (Figs. 2A and 2B). The increase of bioluminescence on increasing cell number was also demonstrated by bioluminescence imaging (Fig. 2C). The analysis of intensities in the ROI indicated a high correlation (r2 = 0.9984) between intensity and cell number (Fig. 2D).
In vitro bioluminescence assay according to cancer cell number. After inoculating a dilution series of SK-HEP1-NL cells (NIS expressing) and SK-HEP1 cells (control), bioluminescence assay and in vitro bioluminescent image acquisition were performed. (A) Result of bioluminescence assay showed almost no luminescence change due to increased SK-HEP1 cell number. In contrast, SK-HEP1-NL showed significantly increased 125I uptake as cell numbers increased. (B) Linear regression analysis indicated high correlation between cell number and bioluminescence (r2 = 0.9979). (C) Bioluminescence of a dilution series of SK-HEP1-NL was visualized using a cooled CCD camera. Large number of cells (right line) showed intense luminescence. (D) Linear regression analysis indicated high correlation between cell number and intensity of ROI (r2 = 0.9984). OD of ROI was measured using image analysis program.
Monitoring of Therapeutic Intervention In Vitro
Treatment with doxorubicin killed both SK-HEP1 and SK-HEP1-NL cells in vitro and reduced radioactivity and bioluminescence signals. The results of monitoring therapeutic intervention in a doxorubicin dose series in vitro are presented in Figure 3. Higher doses of doxorubicin induced more SK-HEP1 and SK-HEP1-NL cell death (data not shown). After adriamycin treatment, the uptakes of radioiodide in SK-HEP1-NL cells were found to be proportional (r2 = 0.9821) to the viable cell numbers (Fig. 3A). Similar to the result of radioiodide uptake, bioluminescence signals from SK-HEP1-NL cells were proportional to the cell numbers, as determined by luminescence assay and image (Fig. 3B).
Correlation between cell counts and assay results after doxorubicin treatment. Various concentrations of doxorubicin (0, 0.5, 1, 2, 5, and 10 μg/mL) were administrated to 2 × 105 cells and incubated for 24 h. Viable cells were then counted by trypan blue exclusion. Viabilities of cells are presented as percentages. 125I uptake or bioluminescence assay was performed. (A) Linear regression analysis was performed between 125I uptake and viable cell count. Uptakes of radioiodide in SK-HEP1-NL were found to be proportional (r2 = 0.9821) to viable cancer cell count. (B) Linear regression between 125I uptake and viable cell count showed that bioluminescence signals were proportional (r2 = 0.9788) to viable cancer cell count.
In Vivo Image of Tumor Xenografted Mouse Model
For noninvasive quantitative monitoring of viable tumor burden, SK-HEP1-NL cells were inoculated. The injection of luciferin and the acquisition of bioluminescence image enabled the visualization of the viable tumor burden for SK-HEP1-NL, but the luminescence of SK-HEP1–induced tumors was not observed (Fig. 4A). A moderate correlation (r2 = 0.7756) was observed between bioluminescence signals and tumor weights (Fig. 4C). The injection of 99mTc-pertechnetate and the acquisition of a γ-camera image did enable the visualization of the SK-HEP1-NL viable tumor burden but did not allow this for SK-HEP1 tumors (Fig. 4B). A similar correlation was observed between the radioactivity, from the scintigraphic image, and the tumor weight (data not shown). Biodistribution data showed a higher correlation coefficient (r2 = 0.8661) between radioactivity and tumor weight than that between the scintigraphic image intensity and tumor weight (Fig. 4D). A moderate correlation (r2 = 0.7782) was observed between the OD from the bioluminescent image and the radioactivity from the tumor (Fig. 4E).
In vivo image of tumor xenografted mouse model. SK-HEP1 and SK-HEP1-NL cells were xenografted subcutaneously into left fore flank (SK-HEP1; 5 × 106 cells), right fore flank (SK-HEP1-NL; 5 × 106 cells), left hind flank (SK-HEP1-NL; 1 × 107 cells), and right hind flank (SK-HEP1-NL; 5 × 106 cells) of each mouse. After tumor growth, bioluminescent and γ-camera images were acquired, and biodistribution of radioactivity was investigated. (A) Injection of luciferin (126 mg/kg) and acquisition of bioluminescence image visualized viable tumor burden of SK-HEP1-NL cells, but not of SK-HEP1 cells. Left hind flank site was most highlighted, which was inoculated with largest number of cells. (B) Injection of 99mTc-pertechnetate and γ-camera imaging visualized viable tumor burden of SK-HEP1-NL cells, but not of SK-HEP1 cells. Using the same manner as that of the bioluminescent image, tumor on left hind flank site was most highlighted, which was inoculated with largest number of cells. (C) Correlation between bioluminescence signal and tumor weight (correlation coefficient, r2 = 0.7756). (D) After acquisition of scintigraphic image, mice were sacrificed and biodistribution of radioactivity was examined. Correlation between tumor weight and radioactivity of each tumor is shown (r2 = 0.8661). Correlation between radioactivity of biodistribution and tumor weight was higher than that between intensity of ROI from scintigraphic image and tumor weight. (E) Comparison of modality was estimated by plotting optical signal vs. radioactivity of tumors. Moderate correlation (r2 = 0.7782) was observed between OD from bioluminescent image and radioactivity from tumors.
DISCUSSION
The rapid development of molecular imaging techniques enables the visualization of various events, such as gene delivery (14), endogenous gene expression (15), cell trafficking (16), signal transduction (17), apoptosis (18), tumor angiogenesis (19), and protein–protein interaction (20) noninvasively and repetitively in living organisms. In the fields of cancer research and anticancer drug development, these molecular imaging techniques have been used to visualize tumorigenesis, tumor cell growth (21), tumor metastasis (5,22), and the effect of therapy (2,4) using bioluminescent and fluorescent imaging techniques.
There are several modalities for molecular imaging in living organisms, such as fluorescent, bioluminescent, and nuclear imaging and MRI. Each modality has its own advantages and disadvantages. Fluorescent images dominate at the cellular level and, more recently, at the small animal level. Many cellular events can be visualized using fluorescent proteins and dyes under the microscope (17,23–26). Bioluminescence images are well established in small animal models using a cooled CCD camera. But these 2 optical approaches have limitations in large animal studies, especially in human applications, because of poor tissue penetration, whereas nuclear imaging and MRI are useful in large animals and in humans. Recent research, using these imaging modalities, has achieved many advances in small animal models. However, nuclear imaging, especially PET and MRI, are at a disadvantage because of their high cost and low research accessibility.
Because of such inherent weaknesses, no single modality can undertake this role from research to clinical application. In recent studies, a combination of each modality has been reported (27–29). Such multimodality approaches can overcome these limitations by using a dual- or triple-imaging reporter. These multimodality imaging reporters can be used in wide-range application from the cellular level to large animals (including humans). Recently, Gambhir’s group reported a triple-fusion reporter that can be used at the cellular level (microscopic image, fluorescence-activated cell sorter) with a fluorescence probe, at the small living animal level with a bioluminescence probe, and at the large living animal level with a radionuclide probe (29). Making a fusion gene (27,29), linking by the IRES (30,31), dividing 2 reporter genes under a bidirectional promoter (32), placing both genes at separate locations in the same vector (33), and cotransferring both viral vectors at the same titer (34) can be used to connect a reporter gene. Because each approach has some disadvantages—such as alteration of the tertiary protein structure in a fusion protein or attenuation of the downstream gene in the case of the IRES approach—the choice of an optimal linkage method is necessarily dependent on each situation. We used the IRES method in this study because the location of the reporter genes differs (NIS is an intrinsic membrane protein and luc is a cytosolic protein). Recent studies reported that the efficiency of the IRES function depends on organs (35–37). Yanagiya et al. concluded that the tissue-specific activity of the IRES may be reflected in tissue tropism of a virus whose specific translation initiation is driven by the IRES (36). If one uses the current approach, the signal ratio of 2 genes that are linked by the IRES would be changed with different organs. After the bioluminescent imaging methods are established, immediate translation to clinical trials using well-established nuclear imaging in humans may be possible.
Recently, we found that the NIS gene has properties that make it suitable for use as an imaging reporter gene—the most valuable advantage being its ease of application in humans (38). It requires only inexpensive and widely used radioiodines or 99mTc-pertechnetate and a γ-camera, which are available in most general hospitals worldwide. The size of the gene is small enough to deliver, and the NIS protein is not immunogenic because of its human origin.
The present study demonstrates the possibility of using the dual-reporter approach to measure the viable tumor cell burden and therapeutic effects noninvasively, repetitively, in vitro and in vivo. The reporter signals from both the nuclear approach and the optically based approach were highly correlated with cancer cell numbers in vitro. Next, we applied this method to evaluate the therapeutic effect of the anticancer drug doxorubicin; reduced viable cancer cell numbers were well correlated with the results obtained. In the animal experiments, nuclear and optical images detected xenografted tumors. The measured signals from the scintigraphic and bioluminescent image obtained reflected tumor weight, though they were not as proportional to tumor weight as were the results of in vitro results.
The advantage of evaluating the therapeutic effect of cancer cells using molecular imaging is that the method enables the measurement of viable cells only. Conventional evaluation methods of tumor mass rely on direct measurement of size, which is based on tumor anatomic mass. However, tumor mass consists of not only viable tumor cells but also necrotic and inflammatory regions, whereas molecular imaging counts only viable tumor cells. In this study, the correlation between tumor weights and signals was relatively poor in the tumor xenografted animal model compared with the in vitro monolayered cell culture. This difference might be because the signals from the animal experiment reflected only viable cells. We speculate that this molecular imaging is more accurate than conventional measurements of tumor size or weight for detecting viable cancer cells. In addition, early and more accurate changes in viable tumor cells can be detected by molecular imaging because the anatomic changes of tumors are observed during the later stages of therapy. The bioluminescent image in vivo showed a much lower correlation than that of the nuclear image. The optical signal can be influenced by several factors, including hemoglobin, water, and oxygen saturation (39–42). Hemoglobin and the red blood cell itself attenuate the light signal by absorption and scattering. Some investigators attempt to reduce this attenuation by using near-infrared light, a blood substitute, and so forth (42). In this study, tumor growth can result in angiogenesis, and the increase in vascularity can cause attenuation of the light signal by hemoglobin and the red blood cell. This can lead to a lack of correlation between the number of cells and the bioluminescence signal.
The disadvantage of the strategy used in this study is that it is limited to ex vivo or stable cell lines. The prerequisite of this strategy is the expression of the reporter in all viable cells during the time of interest. Because in vivo gene delivery to specific cells is limited, the manufacture of cells expressing the reporter gene must be undertaken before inoculation. Images acquired after gene delivery in vivo visualize not viable cells but, rather, the distribution of the delivered reporter gene.
CONCLUSION
The described modality using a dual-imaging reporter would be useful to evaluate the effects of various therapeutic approaches, such as chemotherapy, radiotherapy, gene therapy, and immunotherapy. Because of the advantages offered by a combination of optical and nuclear images, it is hoped that this technique will move quickly from in vitro and animal level study to clinical trials.
Acknowledgments
This work was supported by grant 03-2002-013-0 from the Seoul National University Hospital research fund (2002), a research grant from the Cancer Research Institute, Seoul National University (2003), and the BK21 Project for Medicine, Dentistry, and Pharmacy (2003).
Footnotes
Received Mar. 19, 2004; revision accepted Jul. 30, 2004.
For correspondence or reprints contact: June-Key Chung, MD, PhD, Department of Nuclear Medicine, Seoul National University Hospital, 28, Yongon-dong, Chongno-gu, Seoul 110-744, Korea.
E-mail: jkchung{at}plaza.snu.ac.kr