Magnetic resonance spectroscopic imaging has been used to follow glutathione metabolism and evaluate glutathione heterogeneity in intact tumor tissue. Stable isotope-labeled glutathione was detected in s.c. implanted fibrosarcoma tumors in anesthetized rats following infusion of [2-13C]glycine. Using 1H-decoupled 13C magnetic resonance spectroscopy, the appearance of [2-13C]glycine at 42.4 ppm and the subsequent incorporation of this isotope label into the glycyl residue of glutathione at 44.2 ppm can be detected. The identity and relative concentrations of labeled metabolites observed in the in vivo spectrum were confirmed in studies of tissue extracts. The high level of isotopic enrichment and the concentration of glutathione in tumor tissue allow for collection of spatially localized spectra using 13C chemical shift imaging methods. These data provide the first direct images of glutathione in intact tumor tissue and show metabolic heterogeneity. This method may lead to the ability to monitor changes in tumor tissue redox state that may ultimately affect diagnosis, monitoring, and treatment.
Glutathione is a tripeptide of glutamate, cysteine, and glycine and is found at millimolar concentration in most tissues. Recent studies have shown that proliferating cells in general, and cancer cells in particular, exist in a highly reduced state characterized by a high ratio of the concentration of reduced glutathione to its oxidized disulfide counterpart (1). Consistent with these observations are the results of clinical studies showing that tumor tissue is often higher in glutathione content than normal tissue (2). These high glutathione levels have been linked to therapeutic resistance (3) and reduced overall survival (4) although other studies question this relationship (5). One reason for these contradictory results may be that single tumor biopsy specimens are not representative of the entire tumor (6). Another explanation is that the rate of glutathione metabolism, in addition to its steady-state level, is an important factor when evaluating therapeutic response (7, 8). Clearly, an in vivo method that can noninvasively monitor glutathione metabolism would circumvent sampling problems and allow an assessment of metabolic rates. We have therefore extended our studies developed in cell culture (7, 8) to monitor glutathione metabolism in intact fibrosarcoma (FSA) tumors implanted s.c. in Fischer 344 rats. Glutathione biosynthesis from infused [2-13C]glycine was observed by in vivo 13C magnetic resonance spectroscopy and the heterogeneity of glutathione metabolism across the tumor tissue was detected using chemical shift imaging (CSI). This allows reconstruction of glutathione images in intact tumor tissue. We confirm the identity and tissue distribution of 13C resonances from a study of tissue extracts and ex vivo tumor sections.
Materials and Methods
[2-13C]Glycine was purchased from Cambridge Isotope Laboratories, Inc. (Andover, MA).
Animal protocols. The Animal Care and Use Committees of each institution approved all protocols. Fresh FSA tissue fragments from donor rats were implanted into the flank of female Fischer 344 rats and grew into ∼1.0 cm3 tumors within 3 weeks. A catheter was implanted in the exterior jugular vein of the rats and exteriorized between the scapulae. A fitted infusion harness and lines allowed free movement throughout the cage. Sterile solutions of [2-13C]glycine were prepared in saline and infused at a rate of 0.5 or 1.0 mmol/kg/h (0.5 mL/h).
The University of Florida Animal Care Services staff did rat blood cell counts and chemistry analyses.
Magnetic resonance experiments. Magnetic resonance data were acquired using an 11.1-T, 40-cm bore horizontal magnet (Magnex Scientific, Oxfordshire, United Kingdom) interfaced to a Bruker (Billerica, MA) spectrometer and console. A 1.2-cm three-turn 13C (118 MHz) surface coil was placed around the tumor, orthogonal to a separate 3-cm single turn 1H (470 MHz) surface coil for 1H imaging and decoupling. Rats were initially anesthetized with 5% isoflurane/oxygen, maintained with 2% isoflurane/oxygen, and placed on a heating pad to maintain body temperature.
One-dimensional 13C CSI scans were acquired in 16 phase encoding steps over a field of view of 3.2 cm, with a 200-μs excitation pulse without slice selection and WALTZ 1H decoupling during the phase encoding and acquisition periods. A variable k-space averaging pulse sequence was used with 1:128 averages per phase encoding step. The repetition time was 1.5 seconds and the total scan time was 30 minutes.
Two-dimensional 13C CSI scans were acquired in 8 × 8 phase encoding steps over a field of view of 2.4 × 2.4 cm. Using variable k-space averaging (1:128 averages) and a repetition time of 1.5 seconds, the total scan time was 70 minutes. The tip angle was ∼45 degrees for all scans. Data were reconstructed into a 16 × 16 matrix.
Tissue extraction. At the conclusion of magnetic resonance experiments, tumor tissue was surgically removed and flash frozen in liquid nitrogen. For glutathione analysis, tissue samples were homogenized in pH 7.4 phosphate buffer containing EDTA, acivicin, and monobromobimane. Neutralized perchloric acid extracts of tumor tissue were analyzed by high-performance liquid chromatography for glutathione (7) or glycine (9). Both analytes were normalized to tissue wet weight. The isotopic enrichment of glutathione was determined as previously described (10) using the 13C satellites flanking the glycinyl methylene proton resonances.
Optical microscopy. Frozen tumor tissue was sectioned in the same orientation as the magnetic resonance imaging plane. Serial sections were stained separately with Mercury Orange (11) and H&E.
The glutathione level determined in FSA tumor extracts from rats infused for 36 hours with 1.0 mmol/kg/h [2-13C]glycine was 2.18 ± 0.85 μmol/g tissue (n = 8), which was not significantly different from levels in rats not infused with glycine (2.06 ± 0.54 μmol/g tissue, n = 11). Glycine infusion at 1.0 mmol/kg/h for up to 60 hours had no effect on animal behavior. This infusion level was used in the initial magnetic resonance experiments.
In control rats not infused with [2-13C]glycine, the 13C spectrum of the FSA tumor is characterized by broad envelopes of resonances between 12 and 37 and 52 and 65 ppm (not shown). In all rats in which background spectra were recorded, no resonances were detectable between 35 and 50 ppm within the 10-minute acquisition period. Infusion of 1.0 mmol/kg/h [2-13C]glycine for 12 hours resulted in the appearance of resonances at 42.4 ppm from the infused [2-13C]glycine and at 44.2 ppm from the incorporation of this label into the glycyl-residue of glutathione (Fig. 1A). After a total of 36 hours of [2-13C]glycine, the rat was returned to the magnet and the spectrum in Fig. 1B was recorded. There are distinct increases in the intensities of the glutathione and glycine resonances at this later time point. Slight differences in signal-to-noise between spectra in Fig. 1A and B are due to differences in positioning of the rat and the coil on return to the magnet. A broad component between these resonances may be due to magnetic field inhomogeneity or to incorporation of [2-13C]glycine into proteins as the spectrum of the extract of this same tumor shows only the glycine and glutathione resonances (Fig. 1C).
Data showing glutathione metabolic heterogeneity were obtained from localized spectra from within the tumor using CSI methods. The 1H image of a FSA tumor after 12 hours of infusion of 1.0 mmol/kg/h [2-13C]glycine is shown in Fig. 2A. The arrows in Fig. 2 indicate the location of the 2-mm-thick slices giving rise to the localized 13C magnetic resonance spectra in the one-dimensional CSI data set (Fig. 2B). Compared with the nonlocalized spectra in Fig. 1, clearly defined resonances with baseline separation for glycine at 42.4 ppm and glutathione at 44.2 ppm are observed. After a total of 36 hours of infusion with [2-13C]glycine, the 1H image in Fig. 2C was acquired followed by the one-dimensional CSI data in Fig. 2D. Increased resonance intensities are evident across the tumor as additional glycine enters the tissue between 12 and 36 hours and is metabolized to glutathione. Furthermore, there are slices through the tumor showing little 13C-glutathione compared with 13C-glycine, suggesting metabolic heterogeneity. The 13C spectrum of the extract of this tumor is shown as the bottom spectrum in Fig. 2D and approximates an average of the localized spectra.
A two-dimensional CSI data set from another FSA tumor was acquired within 70 minutes (Fig. 3). The 1H image of this tumor and a superimposed matrix is shown in Fig. 3A. The localized 13C spectra obtained from a region of this matrix are shown in Fig. 3B. Using the spectral data shown in Fig. 3B, glutathione (Fig. 3C) and glycine (Fig. 3D) images of the tissue were generated.
The data shown in Figs. 1 to 3 show that good magnetic resonance data can be obtained by infusing [2-13C]glycine at 1.0 mmol/kg/h. To reduce the risk of potential physiologic effects of infusion, the infusate concentration was lowered to 0.5 mmol/kg/h [2-13C]glycine. Glycine levels in tumors from control animals with no infused glycine were 4.74 ± 0.19 μmol/g tissue (n = 4). For animals infused for 36 hours at 1.0 mmol/kg/h, tumor glycine levels increased to 9.49 ± 2.19 μmol/g tissue (n = 7). Infusion of [2-13C]glycine at 0.5 mmol/kg/h for 36 hours resulted in tumor glycine levels of 6.31 ± 0.60 μmol/g tissue (n = 3). To further investigate the effects of glycine infusion, complete blood cell counts and serum chemistry analyses were done after 36-hour infusion with 0.5 mmol/kg/h [2-13C]glycine. Blood cell analysis showed no significant differences and serum chemistry was unchanged except for a slight elevation of blood urea nitrogen in animals receiving glycine (23.3 ± 1.2 mg/dL) compared with controls (18.0 ± 1.0 mg/dL).
The 1H images in Fig. 4A are from a FSA tumor infused for 12 and 36 hours with 0.5 mmol/kg/h [2-13C]glycine. The 13C spectra from the central pixels of the two-dimensional data set after 12 and 36 hours of infusion are shown in Fig. 4B. Compared with data obtained in Figs. 1 to 3, the lower infusate concentration results in a glycine resonance of diminished intensity relative to the glutathione resonance. To detect heterogeneity in the metabolic data, the spatial distribution of the ratio of the peak areas of labeled glutathione to labeled glycine (effectively using glycine as an internal standard) was calculated after 12 and 36 hours (Fig. 4C). These images show a higher glutathione-to-glycine ratio at the tumor periphery at both time points. This same pattern of heterogeneity was detected in tumors infused at a higher dose of 1.0 mmol/kg/h (data not shown). As the data were acquired with a surface coil, maximum signal-to-noise was obtained from the region of the tumor nearest to the coil. Because of this limitation, the ratio image can be obtained only from about half the tumor volume nearest the coil.
At the completion of the magnetic resonance experiments, the tumor was flash frozen. Frozen tissue sections of this tumor were stained with H&E (Fig. 4D) and show very homogeneous tissue across the sample with proliferating cells along the tumor margins with only a few small, isolated areas of necrosis. An adjacent tumor section was stained with Mercury Orange (Fig. 4F), a fluorescent stain for thiols (mainly glutathione; ref. 11) in the tissue. Although the magnetic resonance ratio image cannot cover the entire tumor, the tumor regions showing increased glutathione-to-glycine ratio qualitatively correspond to regions of the tumor exhibiting higher Mercury Orange staining. The high-resolution 13C magnetic resonance spectrum of the extract from this tumor (Fig. 4E) shows similar resonance intensities to those observed in vivo.
From extract data, FSA tumors infused for 36 hours with [2-13C]glycine at 1.0 or 0.5 mmol/kg/h yielded fractional enrichments of glutathione of 0.359 ± 0.012 and 0.292 ± 0.024, respectively.
To our knowledge, these are the first images of glutathione obtained from intact tumor tissue. Other methods for imaging glutathione do not detect the metabolite directly but rely on the introduction of artificial fluorescent, spin label, or radionuclide probes. Acquisition of these images required infusion of [2-13C]glycine that may have as yet unknown effects on tumor or host biochemistry. The advantage of introducing stable isotopes is the ability to obtain metabolic flux data; labeled glycine is currently used for measuring glutathione metabolism in humans by mass spectrometry (12). Recent work in rat liver showed the feasibility of using magnetic resonance methods to monitor [2-13C]glycine incorporation into glutathione at a 4 mmol/kg/h dose (13). The 1.0 and 0.5 mmol/kg/h infusate levels used herein did not affect glutathione levels in the FSA tumor but did increase glycine levels. Glycine infusion at either dose level does not seem to affect rat well-being and, at the lower dose, does not perturb blood counts or electrolyte values. Lower doses may be preferential because rats fed a glycine-rich diet for 5 to 12 days show an inhibition of tumor angiogenesis (14). At this time, it is not known if the i.v. infusions for shorter time periods used in these experiments would similarly affect tumor angiogenesis. Even at the lower dose, both one-dimensional and two-dimensional CSI data can be collected after 12 hours of infusion.
The infused glycine does not seem to equilibrate fully with total tissue glycine as reduction in the infusate level by half results in only a 19% reduction in the fractional enrichment of glutathione.
Localized magnetic resonance data can detect labeled glycine in all regions of each of the tumors examined, suggesting this substrate is delivered adequately. However, some regions show little glutathione. Absolute quantification of metabolites in the 13C spectra is difficult; therefore, the ratio of the glutathione-to-glycine peak areas was used as an initial assessment of metabolic heterogeneity. These ratio images show enhanced glutathione relative to glycine at the periphery of the tumors distal from the body of the rat. This matches our histochemical staining of cellular thiols by Mercury Orange, which is mainly due to glutathione (11). The correspondence of the in vivo and histochemical staining suggests our ratio images are a valid method for assessing glutathione metabolic heterogeneity. H&E staining showed cellular proliferation near the periphery of the tumor, consistent with a recent study showing that oxygen levels in the FSA tumor are also higher in this region (15). These results, which suggest viable better-perfused tissue leads to increased glutathione metabolic activity, may be at odds with other studies in cervical tumors (16).
It should be noted that the resolution of the in vivo 13C spectra will not allow discrimination of the reduced and oxidized (disulfide) forms of glutathione as these resonances differ by only 0.1 ppm. However, our extract data show no significant oxidized glutathione in the tissue, suggesting levels must average <10% across the tumor samples studied. Even with overlap of the reduced and oxidized glutathione resonances in the in vivo spectrum, perturbations in the redox balance induced by therapy might still be detectable because studies have shown that changes in redox balance may result in formation of mixed glutathione-protein disulfides (17) or cellular extrusion of oxidized glutathione disulfide (18), both mechanisms potentially changing our magnetic resonance–detectable resonances. Studies showing the feasibility of monitoring these therapy-induced changes are currently under way.
These data present the first magnetic resonance generated images of glutathione metabolism in intact tumor tissue. Opstad et al. (19) used 1H magnetic resonance spectroscopy methods to observe higher steady-state glutathione levels in human meningiomas compared with normal brain. Their results showed the ability of magnetic resonance techniques to detect static levels of glutathione in intact tissue; however, imaging of heterogeneity was not done and dynamic measurements of metabolism were not possible. In addition to the images, our one-dimensional method shows that metabolic rates may be obtained relatively quickly and can be used to monitor changes in glutathione content and turnover following treatment. Although a 11.1-T magnet was used in these studies, we know from previous experience that the glycine and glutathione 13C resonances can be resolved at 4.7 T in vivo (13), and as magnetic resonance imaging technology advances, it is clear that the novel methods reported here may ultimately be applicable to human studies.
Grant support: NIH grants R21 CA097871 and P41 RR16105 and the National High Magnetic Field Laboratory (Tallahassee, FL).
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 Drs. Mark Dewhirst, Zeljko Vujaskovic, and O. Michael Colvin of Duke for their support and Dr. Rachel Richardson of Duke and Raquel Torres of the University of Florida for their assistance.