Therapeutic regimens such as radiation or chemotherapy attempt to exploit the physiological differences between normal and malignant tissue. Tissue redox status and pO2 are two factors that are hypothesized to be different in normal and malignant tissues. Methods that can detect subtle differences in the above physiological parameters would greatly aid in devising appropriate treatment strategies. We have previously used in vivo electron paramagnetic resonance (EPR) spectroscopy and imaging techniques and shown that tumor tissues are highly reducing and hypoxic compared with normal tissues (P. Kuppusamy et al., Cancer Res., 58: 1562–1568, 1998). The purpose of the present study was to obtain spatially resolved redox data from normal and tumor tissues of radiation-induced fibrosarcoma (RIF-1) tumor-bearing mice and to examine the role of intracellular glutathione (GSH) on the tissue redox status. Experiments were performed using low-frequency (1.3 GHz) in vivo EPR spectroscopy and imaging techniques with a nitroxide redox probe. l-buthionine-S,R-sulfoximine (BSO), an inhibitor of GSH synthesis, was used to deplete tissue GSH levels. The results show the existence of significant heterogeneity of redox status in the tumor tissue compared with normal tissue. The tumor tissues show at least 4-fold higher concentrations of GSH levels compared with normal tissues in the tumor-bearing mice. Also BSO treatment showed a differential depletion of GSH and reducing equivalents in the tumor tissue. Thus, it appears that there is significant heterogeneity of tumor redox status and that manipulation of the tumor redox status may be important in tumor growth and therapy.
The redox environment (1) within the tumor cell is an important parameter that may determine the response of a tumor to certain chemotherapeutic agents, radiation, and bioreductive hypoxic cell cytotoxins (2, 3, 4, 5). A variety of intracellular molecules may contribute to the overall redox status in tissues including GSH,3 thioredoxins, NADPH, flavins, ascorbate, and others (1). Collectively, the reducing species may have an impact on how a particular tissue responds to oxidative stress induced by a particular drug or treatment modality, and, therefore, a noninvasive means of determining the redox status of a tissue should be a useful adjunct for clinical oncology.
Low-molecular-weight, stable nitroxide free radicals have been used as biophysical and biochemical probes in a variety of biological experiments (6, 7, 8), in which EPR spectroscopy is conveniently used for detection. The nitroxides exist in biological systems as a redox pair (Fig. 1), namely the nitroxide free radical form and the diamagnetic hydroxylamine, which is the one-electron reduction product of the nitroxide free radical (9, 10, 11). Prior studies indicated that nitroxides are reduced to the corresponding hydroxylamine in cellular incubations as well as in vivo and that the rate of nitroxide reduction is significantly enhanced under hypoxic conditions (12, 13, 14, 15, 16). Presumably, the nitroxide accepts cellular reducing equivalents that would have gone to the reduction of molecular oxygen. Hence, the pharmacology of nitroxides in an EPR experiment should report on differences that exist in tissues having different redox and oxygen status. Previously we have estimated the in vivo levels of the unreduced nitroxide from excised tumor and found it to be significantly lower in tumor as compared with bone marrow (17). We further performed noninvasive in vivo EPR spectroscopic measurements of nitroxide reduction profiles in RIF-1 tumor-bearing mice and reported that the tumor tissues were significantly hypoxic and highly reducing compared with the normal tissues (18). Thus, measurement and mapping of nitroxide reduction profiles in tumor and normal tissues in vivo should provide important physiological information on tissue redox and oxygen status.
Thiols such as intracellular GSH and protein thiols are key to maintaining the redox status of tissues. Chemically, nitroxides do not react with thiols. However, in microsomal preparations, thiol-containing biomolecules have been shown to play a significant role in the bioreduction of nitroxides (19). Because thiol/disulfide state generally reflects the redox status of the tissue, altering the concentration of the thiol or changing the ratio of redox pairs should have an impact on the clearance of the nitroxide and, thus, provide a means to study the importance of this class of compounds as a function of bioreduction. For example, depleting the GSH levels by using specific inhibitors of its synthesis such as BSO and monitoring the pharmacokinetics of redox-sensitive nitroxide probes in both normal as well as tumor tissue should delineate the role of thiols in modulating tissue redox status.
The purpose of this study was: (a) to develop a noninvasive method to measure and image tissue redox status; (b) to obtain redox images in vivo from normal and tumor tissues of RIF-1 tumor-bearing mice; and (c) to examine the role of intracellular GSH on the tissue redox status. Experiments were performed using low-frequency (1.3 GHz) in vivo EPR spectroscopy and imaging techniques with a nitroxide redox probe. The results show the existence of significant heterogeneity in the redox status of the tumor tissue, and that there is differential depletion of the reducing equivalents in the tumor compared with normal tissue in mice treated with inhibitors of GSH synthesis.
MATERIALS AND METHODS
The nitroxide probe 3-CP was purchased from Aldrich (Milwaukee, WI). Solutions of the nitroxide were freshly prepared at a stock concentration of 300 mm in saline. GSH (l-γ-glutamyl-l-cysteinyl-glycine) and BSO were purchased from Sigma Chemical Co.
Female C3H mice were supplied through the Frederick Cancer Research Center Animal Production, Frederick, MD. The animals were received at 6 weeks of age and housed five per cage in climate-controlled rooms and were allowed food and acidified water ad libitum. The animals were on average 50 days old at the time of experimentation and weighed 25 ± 3 g. Experiments were conducted according to the principles outlined in the Guide for the Care and Use of Laboratory Animals prepared by the Institute of Laboratory Animal Resources, National Research Council.
RIF-1 Tumor Growth.
RIF-1 tumor cells were grown in monolayered culture until injection in mice. The mice were injected s.c. in their right hind leg with a single cell suspension of tumor cells (106 cells in 0.1 ml) in PBS. The animals were observed closely, and the tumors became palpable ∼5 days after injection. Tumors were allowed to grow to a size of about 8–10 mm in the greatest dimension.
Tissue GSH Assay.
Mice were injected (i.p.) with 2.25 mmol/kg (bodyweight) of BSO in 100 μl of saline. Six h later, the animals were killed, and about 0.2–0.5 g of tumor tissue on the right leg and normal muscle tissue on the left leg were carefully excised and weighed. The tissues were placed in glass dounce homogenizer (5 ml) and homogenized with 1 ml of 200 mm methanesulfonic acid containing 5 mm diethylenetriaminepentaacetic acid (DTPA; pH <2.0). This homogenate was centrifuged for 30 min at 14,000 rpm at 4°C. The resultant supernatant was diluted 1:1 with mobile phase and stored frozen at −70°C for GSH assay.
The GSH assays were performed by high-performance liquid chromatography-electrochemical method (20). Samples were separated on a Polaris 5 μm, 0.4 × 20 cm C-18 column eluted with a mobile phase of 50 mm NaH2PO4, 0.05 mm octanesulfonic acid, and 2% acetonitrile adjusted to pH 2.7 with phosphoric acid and a flow rate of 1 ml/min. An ESA CoulArray detector was used with the guard cell set at +950 mV, electrode 1 at +400 mV, and electrode 2 at +880 mV. Full-scale output was set at 100 μA and peak areas were analyzed using the CoulArray software (ESA, Chelmsford, MA). A standard curve was obtained using a 10 μm-to-1 mm solution of GSH, from which concentrations of GSH were determined. The GSH values were expressed as μmol/g of tissue wet weight.
Imaging measurements were performed using the EPRI instrumentation consisting of an L-band EPR spectrometer, three sets of water-cooled gradient coils and a personal computer-based data acquisition system (18, 21, 22). EPR spectra were recorded using a custom-built surface resonator (18, 23). The resonator consisted of two concentric quartz tubes, 10 and 12 mm in diameter, with the outer surfaces covered with silver foil strips forming a bridged loop-gap structure. The resonator structure was housed inside a metallic tubular shield with one open end, 10 mm diameter, functioning as the active surface. The resonator was capable of sampling a cylindrical volume measuring 10 mm in diameter and 5 mm deep. The open structure of this resonator was ideal for localized measurements on large objects and thus was not limited by the size of the object.
Mice were anesthetized by breathing air containing 1% isoflurane delivered through a nose cone. The tail vein was cannulated with a heparin-filled 30-gauge catheter for infusion of the nitroxide probes. Either the tumor in the right leg or the normal tissue (muscle) on the left leg was prepared for measurements. The hair on the observation spot of the skin was shaved and the animal was placed on a bedplate with a circular slot (20 mm diameter) in such a way that the observation spot was centered at the slot. The animal was secured to the plate with adhesive tape and placed on top of the resonator such that the tumor or the normal muscle was in direct contact with the active surface of the resonator. An IR lamp was used to maintain normal body temperature, which was measured using a rectal thermistor probe.
Projection Acquisition and Image Reconstruction.
Projection data were acquired using angular sampling method. The projections were acquired as single scans (1024 points/projection) using constant sweep time. The measured projections were corrected for removal of hyperfine-based artifacts and deconvoluted with the corresponding zero-gradient projection (24). The deconvoluted projections were then convoluted with a Shepp-Logan filter and subsampled to 128 points for backprojection. A single-stage, filtered backprojection reconstruction algorithm was used to recover the image. Projection data acquisition and subsequent image reconstruction were performed using an Intel Pentium 600 MHz Personal Computer equipped with an IEEE-488 GPIB board (Capital Equipment Corporation, Burlington, MA).
All of the data were expressed as mean ± SE. Comparisons among groups were performed by student’s t test or one-way ANOVA, with Student-Newman-Keuls tests. The significance level was defined as P < 0.05.
Pharmacokinetics of nitroxide uptake and clearance from normal and tumor tissues of RIF-1 tumor-bearing mice were measured in vivo using EPR spectroscopy. 3-CP was used as a probe at a dose of 185 mg/kg. Mice were infused (bolus dose), via tail vein catheter, with a saline solution of 3-CP, and EPR spectra were measured continuously from the tumor on right leg or normal muscle tissue on the left leg. A few spectra, representative of the time course measurements from a RIF-1 tumor, are shown in Fig. 2. The field-swept EPR spectra are characterized by a triplet, arising because of hyperfine splitting from the 14N nucleus, with coupling constants 15.78 G and 16.30 G. The unequal coupling constants, which are usually observed at low-frequency measurements, are caused by breakdown of the high-field approximation, known as the “Breit-Rabi” effect. Each triplet of the spectrum is also characterized by a unique peak-to-peak width and amplitude. Whereas the coupling constant and width of the spectra remain more or less constant, the amplitude varies with time, reflecting changes in the nitroxide concentration in the tissue. Thus, either the area under the peak, measured as double integral of the spectrum, or peak-to-peak amplitude can be used as a quantitative measure of the nitroxide concentration, in arbitrary units. Fig. 3 shows a semilogarithmic plot of time course of the signal intensity of 3-CP in the normal and tumor tissues. The plot shows the data measured after the nitroxide intensity peaked, usually in about 3–4 min. It is observed that the nitroxide concentrations (in logarithmic units) in the tumor as well as normal tissues decrease linearly with time, suggesting that the clearance pharmacokinetics can be modeled by a pseudo-first order rate equation. The solid lines through the data points are linear fit to the respective data set suggesting compliance with a pseudo first-order rate law. The bar graph in Fig. 3 B shows the measured pseudo first-order rate constants of nitroxide clearance in the tissues. The data represent mean ± SE of measurements on 3–5 mice per group.
To investigate the effect of GSH depletion on the rate of clearance of tissue nitroxide, the above pharmacokinetics experiments were also performed on tumor-bearing animals treated with BSO. Mice were injected (i.p.) with 2.25 mmol/kg BSO in saline, and the EPR measurements were made 6 h later. The dose and duration of BSO treatment were chosen to produce ∼40–50% of GSH depletion from RIF tumors (4).
The nitroxide pharmacokinetic data indicate the following: (a) the nitroxide is reduced more rapidly in the tumor tissue compared with normal muscle tissue in the tumor-bearing mice; (b) the nitroxide reduction rate constant in the tumor of BSO-treated mice is decreased; and (c) the nitroxide reduction rate constant in the normal muscle tissue of BSO-treated tumor-bearing mice is not significantly different from that of untreated mice (data not shown). Taken together, the results suggest that BSO treatment provides a differential clearance of nitroxide in tumor tissues.
The pharmacokinetic data presented above represents global measurements, i.e., the data provide only an intensity-weighted mean clearance of the probe in the tissues. Spatially resolved changes in nitroxide intensity within the tumor volume are not obtained from the spectroscopic measurements. Because tumor tissues are characterized by significant heterogeneities in terms of redox status and oxygenation, it is desirable to obtain spatially resolved images of nitroxide distribution and clearance simultaneously within the tumor volume. This can be obtained using EPRI techniques that we have developed and used previously (18). Fig. 4 shows a few selected images from a series of two-dimensional spatial maps of nitroxide content in normal and BSO-treated tumor tissues obtained as a function of time after 3-CP infusion. It should be noted that these images were collected simultaneously on the same animals used for the measurement of pharmacokinetic data shown in Fig. 3. The data acquisition software was programmed to collect a series of alternating pharmacokinetic spectral (5 s) and image (∼90 s) data on the same tissue for up to 30 min, immediately after infusion of the probe. Each time course image within the series, as shown in Fig. 4, was normalized with respect to the maximum intensity obtained within that series, usually in the images collected in about 4–6 min after infusion. The nitroxide reduction in the tumor of BSO-treated animals was slower than that in the tumor of untreated mice.
The images in Fig. 4 are 64 × 64 pixel images cropped from 128 × 128 pixels of original data obtained from a field of view of 20 × 20 mm2. Thus, each pixel in the image represents a theoretical volume of 0.2 × 0.2 × 5.0 mm3, although the actual resolution based on the number of projections and gradient field may limit the plane resolution to 1 mm and hence the volume to 1 × 1 × 5 mm3. The series of image data can be used to follow time-dependent changes in each voxel within the image to obtain spatially resolved decay curves and hence rate constants. The rate constants for the clearance of the probe in each pixel were computed and displayed in the form of a color-coded image in Fig. 5. A frequency plot of the rate constants is also shown for each tissue in Fig. 5. The rate constant map and the frequency plot show the presence of a range of rate constants, which is a measure of heterogeneity in the redox status within each tissue.
The redox data show that the BSO treatment significantly decreases the rate constant of nitroxide reduction in the tumor tissue. It is known that BSO inhibits the γ-glutamylcysteine synthetase enzyme that is responsible for GSH synthesis (25). Studies have shown that tumor tissue levels of GSH can be depleted to very low levels without any toxicity (4). The effect of BSO treatment on the redox constants obtained in the present experiments, thus, can be attributed to alterations in the tissue GSH levels. To investigate the correlation between the observed rate constants and tissue GSH, the tissue levels of GSH were measured in the tissues of normal (control) and BSO-treated (same dose as for the EPR measurements) mice. The GSH assay data are shown in Fig. 6. The BSO-treated tissue data show a significant reduction in the GSH levels in the tumor tissue compared with untreated control. On the other hand, no significant differences were observed between treated and untreated normal tissues. The estimated GSH level (Fig. 6) correlates well with the nitroxide reduction rate constants (Fig. 3) in different tissues. Thus, it is evident that BSO causes a differential depletion among normal and tumor tissues.
The treatment of malignant tumors by nonsurgical modalities like radiation or chemotherapy may be influenced both by the genetic properties of the tumor cells and the epigenetic factors of the tumor micromilieu (26, 27). While extensive research is presently ongoing to develop gene therapy strategies targeted to tumor cells, the development of therapeutic regimens to exploit the physiological differences between normal and malignant tissue may also provide options for improved treatment. Some of the physiological differences between normal tissue and tumors include oxygen status, redox status, and intracellular pH (28, 29). These differences, at least in part, can result from the physical architecture of the tumors with compromised blood supply (30). Methods that can detect subtle differences in the above physiological parameters would greatly aid in devising appropriate treatment strategies. Techniques that enable rapid and noninvasive measurements of these two parameters in individual patients, therefore, would extend our knowledge and clinical acumen.
The present data obtained using in vivo EPR techniques and a nitroxide spin probe in RIF-1 tumor implanted in a C3H mouse model shows the differential physiology between normal and tumor tissue. After infusion, the reduction of the nitroxide, 3-CP, is enhanced in tumors compared with the normal tissue. The in vivo stability of nitroxides depends primarily on the ease of the oxidation of the nitroxide to the oxoammonium cation (31) or the reduction to the hydroxylamine (32). Consistent with this hypothesis are observations that the in vivo stability of the nitroxides is inversely related to the ease of reduction of the nitroxide. Furthermore, in vivo reduction of nitroxides is dependent on the oxygen content as well as on the levels of endogenous reducing agents such as GSH. The enhanced reduction of the nitroxide in tumor compared with normal tissue, thus, may be attributable to lower oxygen concentration as well as to the significantly enhanced GSH levels in the tumor.
In the pharmacokinetic experiments, we have followed changes in the nitroxide signal intensity in the tumor in vivo as a function of the postinfusion period. The measured rate of decrease of nitroxide signal intensity at the tumor site may depend on the following dynamic processes: (a) infusion of the probe into the tumor; (b) changes in the probe concentration caused by bioreduction in the tumor tissue; and (c) elimination of the probe out of the tumor volume by blood flow. Thus, the decrease in the EPR signal intensity of the nitroxide cannot be attributed to bioreduction processes alone. However, it has been reported by Gallez et al. (33) that bioreduction at the local site is the major factor that contributes to the overall decrease. This is also evident from the present work because different rate constants of clearance are observed between the tumor tissue and muscle on the nontumor-bearing leg. If perfusion is the major contributor, then any local measurement should reflect on the systemic levels of the probe, and, thus, the rates constants would have been identical between the two tissues. Thus, the clearance rate constant observed by this localized measurement can be used as a parameter of tumor tissue redox status, which is correlated to the bioreduction of the probe.
Several studies have shown that tissue oxygenation is an important factor that will affect the tissue redox status and hence the reduction rate of the nitroxide. Particularly, tissue hypoxia, which is known to be present in tumors, has been shown to enhance the bioreduction of nitroxides (18). Although this is evident from the enhanced rate of reduction from the tumor as compared with muscle tissue, which is relatively well oxygenated, the involvement of other factors, such as altered redox state and pH, may also be important.
GSH plays a central role in the maintenance of enzyme sulfhydral groups in the proper redox state (34). Because of this, alterations of the GSH redox state can have profound effects on cellular metabolism. GSH is present in significantly large quantities inside cells and it is a versatile protector. Some of the protective roles of the GSH include radical scavenging, restoration of damaged molecules by hydrogen donation, reduction of peroxides, and maintenance of protein thiols in the reduced state. Of these roles, hydrogen atom donation to DNA radicals is probably the most important. Because competing reactions are very rapid, particularly with oxygen in well-oxygenated normal tissues, high concentrations of GSH are required for protection. Thus, moderate depletion of GSH in normal cells has no effect on the radiosensitivity; whereas under hypoxic conditions, such as those that occur in tumor tissues, GSH may play a dominant protective role (35, 36, 37). Conversely, the depletion of GSH in the hypoxic tumor tissue may have beneficial effects on the treatment of tumors (38). The safety and efficacy of BSO-based metabolic alteration in cancer treatment have been extensively studied (4, 38). The results show that considerable variations exist in the effectiveness of GSH depletion depending on the tumor model and origin. Nevertheless, the method has the potential to offer therapeutic enhancements in such cases in which it is effective and may augment other procedures such as ARCON for radio- or chemotherapy (39, 40, 41, 42, 43).
In conclusion, noninvasive measurements and mapping of tumor redox status in tumor-bearing mice were performed using low-frequency EPR spectroscopy/imaging techniques. Implanted RIF-1 tumor in the right leg was studied, and the normal tissue on the left leg was used as control. Nitroxide pharmacokinetics showed overall higher magnitude and significant heterogeneity of reducing equivalents in the tumor tissue compared with normal tissue. Mice treated with BSO, a GSH-depleting agent, showed differential reduction in the tissue GSH levels and reducing equivalents. This differential reduction of reducing equivalents between the normal and malignant tissue may be useful for selective tumor killing.
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.
Supported by National Cancer Institute (NIH) Grants CA 78886 and RR 12190.
The abbreviations used are: GSH, glutathione; 3-CP, 3-carbamoyl-2,2,5,5-tetramethylpyrrolidine-N-oxyl, or 3-carbamoyl-proxyl; RIF, radiation-induced fibrosarcoma; EPR, electron paramagnetic resonance; EPRI, EPR imaging; BSO, l-buthionine-S,R-sulfoximine; ARCON, accelerated radiosensitization by carbogen and nicotinamide.