- Email: [email protected]
Hypoxic fractions measured in murine tumors and normal tissues using the comet assay
Int. J. Radiation
Oncology
Pergamon
??Manipulation
HYPOXIC
Biol. Phys., Vol. 29. No. 3. pp. 487-491. 1994 Copyright 0 1994 Else&r Science Ltd Printed in the USA. All rights reserved 0360-3016/94 $6.00 + .OO
of Tumor Oxygen and Physiology
FRACTIONS MEASURED IN MURINE TISSUES USING THE COMET
TUMORS ASSAY
AND NORMAL
PEGGY L. OLIVE, PH.D., CHARLENE M. VIKSE, A.A. AND R.~LPH E. DURAND, PH.D. BritishColumbiaCancerResearchCentre,Vancouver,BC, Canada Purpose: To apply the alkaline comet assay to the detection of radiobiologically hypoxic cells in solid tumors and xtissues of mice, and to examine the influence of strand break repair on the oxygen enhancement ratio measured using the alkaline comet assay. Methods and Materials: In previous studies, we found that hypoxic fraction in squamous cell carcinomas growing in C3H mice could be reliably and easily measured using the alkaline comet assay. The comet assay applies fluorescence microscopy and image analysis to examine patterns of migration of deoxyribonucleic acid from individual cells embedded in agarose and exposed to an electric field. This method has sufficient resolution to detect subpopulations of hypoxic cells which show about 3 X fewer strand breaks than aerobic cells after irradiation. Results: Fast rejoining kinetics in vitro are comparable to those measured in Go, and rejoining of strand breaks Ghypoxic tumor cells occurs at a similar rate as rejoining in aerobic cells. Little residual damage was detectable using the comet assay in tumors 4-24 h following 15 Gy, allowing repeat measurements to be performed. Bone marrow and testis, but not liver, spleen, or jejunum contained a small fraction of hypoxic cells when mice breathed 10% oxygen during irradiation. Conclusion: The comet assay confirms that some normal tissues may border on hypoxia. Rejoining of strand breaks occurs rapidly in both oxic and hypoxic cells so that the oxygen enhancement ratio remains relatively constant with time after irradiation. Interestingly, a smaller oxygen enhancement ratio was observed in tumors than was expected, probably as a result of the presence of acutely hypoxic cells. Hypoxia, normal tissue, Hypoxia, tumor, DNA strand breaks, Transient hypoxia.
the comet assay compares favorably with hypoxic fraction measured using the paired-survival curve method (14). Recently this method has been used to detect hypoxic cells in human tumor fine needle aspirates removed from patients immediately after irradiation ( 1.5).The ability to identify individual hypoxic cells provides a new opportunity to examine questions relevant to tumor and tissue hypoxia. In this manuscript, repair rates are reported for both aerobic and hypoxic cells in vivo and compared to rejoining rates in vitro. In addition, the oxygenation status of normal tissues in C3H mice was examined since it has been suggested that some normal tissues may border on hypoxia (7, lo- 12), an observation which would be relevant to the successful application of hypoxic cell radiosensitizers or hypoxic cell cytotoxins.
INTRODUCTION
We recently described a method to quantify hypoxic fraction in cells from solid tumors ( 14, 15). The comet assay, based on a method originally described by Ostling and Johanson (I 7), uses fluorescence microscopy and image analysis to examine patterns of migration of deoxyribonucleic acid (DNA) from individual tumor cells embedded in agarose and exposed to an electric field. Hypoxic tumor cells are approximately 3 X less sensitive than aerobic cells to induction of DNA strand breaks by ionizing radiation (2) and can therefore be detected by the assay ( 14). The relation between oxygen concentration and production of strand breaks by radiation has been shown to be identical to the relation between oxygen concentration and cell killing by radiation (2), thus ensuring that cells determined to be hypoxic using the clinically relevant endpoint, cell survival, are also hypoxic when measured using the comet assay. In a transplanted squamous cell carcinoma growing in C3H mice (SCCVII), hypoxic fraction measured using
METHODS
AND
MATERIALS
Squamous cell carcinoma cells were transplanted subcutaneously over the sacral region of inbred male C3H/
Reprint requests to: Dr. Peggy L. Olive, Medical Biophysics Department, British Columbia Cancer Research Centre, 60 1 W. 10th Ave., Vancouver, BCVSZ 1L3 Canada. Acknowledgements-This work was supported by USPHS grant
number CA-37879 awarded from the National DHHS. Accepted for publication 8 October 1993.
487
Cancer Institute,
488
I. J. Radiation Oncology 0 Biology0 Physics
He mice, approximately 30 g in weight. Tumors were used for experimentation approximately 2 weeks later when they had reached a weight of 450-600 mg. Tumors were irradiated by placing unanesthetized mice in leadlined jigs and irradiating the tumor with 250 kVp X rays using parallel opposed fields at a dose rate of 4.5 Gy/min. Following irradiation, mice were sacrificed, tumors rapidly excised and placed in ice-cold phosphate buffer. A single cell suspension was prepared from the entire tumor by mincing the tissue and filtering the suspension through 50 ym nylon mesh. Cells were diluted to a concentration of 2-4 X lo4 cells/ml for the comet assay. For examination of damage to normal tissues, mice received whole body irradiation at a dose rate of 3.3 Gy/ min. For both tumor and normal tissues, approximately 30 s elapsed between the end of irradiation and removal of the tissue to ice-cold buffer. Large samples of tissue were used to obtain a single-cell suspension, thus ensuring that the sample was representative of the entire tissue. Tissues were minced in ice-cold phosphate buffer containing 20 mM ethylenediaminetetraacetic acid (EDTA)’ and 2% dimethylsulphoxide (DMSO),’ filtered through 50 pm nylon mesh, centrifuged and resuspended in phosphate-buffered saline. Bone marrow cells were removed from one tibia using a syringe. Jejunum (approx. 100 mg), spleen, and testis were removed and washed in ice-cold phosphate-buffered saline prior to mincing. With the exception of liver, all tissues were mechanically disaggregated at ice temperature. Liver (500 mg) was minced and incubated with trypsin,’ collagenase’ and deoxyribonuclease’ for 15 min at 37°C. Cells were embedded in type VII low gelling-temperature agarose,’ lysed in alkali and salt, and exposed to a low voltage to separate damaged from undamaged DNA. Comets were visualized by staining with propidium iodide,’ and patterns of DNA migration were quantified with a fluorescence microscope attached to an imaging system. Images of up to 600 cometsper-sample were digitized, and tail moment, a measurement of tail length and amount of DNA in the tail, was determined. Hypoxic fraction was calculated using a curve fitting algorithm applied to histograms describing number of comets vs. tail moment ( 14). Analysis of strand breaks in SCCVII tumor cells in vitro was performed using a cell line established from the tumor and maintained in exponential growth in a 5% oxygen atmosphere for 1 week. For anoxic irradiations, SCCVII cells were incubated in water-jacketed spinner culture flasks in Hepes buffered medium at 37°C with continuous gassing with oxygen-free nitrogen. After 1 h, cells were centrifuged and sealed in glass ampules at a density of 2 X 10’cells/ml followed by 15 min at 37°C to deplete any residual oxygen (5). Ampules were then cooled on ice and irradiated.
‘Sigma Chemical Co., St. Louis, MO.
Volume 29, Number 3, 1994
Time after 15 Gy (hr)
Fig. 1. Strand break rejoining kinetics in SCCVII tumors or cells irradiated and allowed to repair either in vivo or in vitro.(a) The
rejoining kinetics for tumor cells irradiated in vivo with 15 Gy. Tumors were excised at various times after irradiation. Data are extrapolated to the expected “zero” time point measured for cells (85% aerobic), irradiated in vitro on ice since cells from tumors could not be recovered in less than 5 min. (b) A comparison between early time points for tumor cells irradiated and allowed to repair in vitro (V) or tumors irradiated and allowed to repair in vivo (0). The “background” damage (mean f SD) is indicated by separate points. (c) The displacement between the aerobic and hypoxic peaks on the histograms (see Fig. 2) is shown for tumors (3-6-per-group, mean + SD) irradiated and allowed to repair in vivo.
RESULTS
The initial rate of single-strand break rejoining appears to be similar whether tumor cells are irradiated in the animal and allowed to rejoin in the animal or irradiated on ice and allowed to rejoin in culture at 37°C (Fig. 1). However, after lo- 15 min of repr lr, there is a discrepancy between in vivo and in vitro data which may be best explained by technical rather than biological factors. Absolute comparisons between these two data sets are limited by the inability to obtain samples from the mouse prior to significant repair of damage; the displacement between early time points for the two curves in Figure l(b) is a result of the repair that occurs in the tumor prior to excision. Approximately 50% of the single-strand breaks produced by 15 Gy are rejoined before removal of the tumor and preparation of a single-cell suspension. The repair half-time, measured in vitro for SCCVII tumor cells from data in Figure l(b), is 6.6 min (95% conf. limits 5.97.8) which compares well with strand break rejoining times for other cell types using the comet assay (13). A factor which has considerable influence on later repair times is the large variability in “background” damage between cells from different animals, see dotted lines in Figure 1(a), no doubt reflecting differences in damage induced by the tumor disaggregation procedure. Normalization procedures which subtract background damage prior to measurement of rejoining kinetics are therefore inappropriate for the in vivo data. Single-strand breaks are completely rejoined by 2 h after 15 Gy, Figure l(a), and there is little damage detectable
Hypoxic
fraction
489
using the comet assay 0 P. L. OLIVE e!a/
Air-breathing
Clamped
13
’ ”
1.30 Gy +
40
4.
I
’q
II’
15
min
.30 Gy + 30
min
K. 30 Gy + 4 hr
Dose (Gy)
Tail Moment Fig. 2. Histograms showing distributions of tail moments for comets of individual SCCVII tumors. (A-C) Response of tumors clamped prior to and following irradiation. (D) Mixture of cells from tumors of mice asphyxiated prior to irradiation with 15 Gy or 45 Gy. (E-K) Response of air-breathing mice to 15 or 30 Gy followed by a repair interval in viva (L) Mean tail moment of SCCVII tumor cells irradiated under aerobic or anoxic conditions. The oxygen enhancement ratio is 3:3.
even 24 h later when cell degradation may be expected, Figure l(a), Figure 2(h). For the comet assay, heavily damaged cells will most likely be lost during tumor disaggregation and during the lysis procedure, thus eliminating their contribution to overall repair kinetics. Apoptotic cells, which can be easily detected using the comet assay (16), do not influence rejoining rate estimates with this technique. Figure 2 verifies that tumor cells made hypoxic at the time of irradiation show less DNA damage than aerobic cells, and only one population of cells is present in clamped tumors “0,” 10 and 30 min after irradiation, Figure 2(a-c). For tumor cells from air-breathing animals, the displacement between the peaks for aerobic and hypoxic tumor cells remains relatively constant for cells allowed to repair damage in viva, Figure l(c), Figures 2(f, g, i, j), suggesting that hypoxic tumor cells rejoin breaks as rapidly as aerobic tumor cells. However, the displacement between the peaks is smaller than expected based on measurement of the oxygen enhancement ratio (OER) in vitro, even when background damage is taken into ac-
count. In Figure 2(d), a mixture of tumor cells from two mice asphyxiated prior to 15 Gy or 45 Gy gave a displacement between the peaks in the comet assay of 2.5. In comparison, tumors from air-breathing mice revealed a displacement of only 1.9 + 0.32 (19 tumors) between the aerobic and hypoxic populations. In addition, the mixed cell population gave much better resolution for the two peaks than observed for the tumor cells in Figure 2(e). An important advantage of the comet assay is the ability to detect DNA damage with excellent sensitivity without the necessity of radiolabelling cells. The number of strand breaks present immediately after irradiation in normal tissues of mice was consistently higher than observed in the SCCVII tumor (despite the lower dose rate), a confirmation that rejoining kinetics for strand breaks are generally slower in differentiated normal tissues than in tumor (12, 20). Tumor cells exposed to 15 Gy in viva showed a modal tail moment of about 20, bone marrow showed average values of about 25, and spleen and jejunum had values of 30 or higher.
490
1. J. Radiation Oncology 0 Biology 0 Physics
~~[-Jj.j-J 0
1530
Volume 29, Number 3, 1994
rjj--J:;~:/~:~~ 0
1530
0
15 30
0
1530
0
15 30
0
15 30
DNA Content Fig. 3. Histograms showing DNA damage measured using the comet assay for different normal tissues and SCCVII tumors from air-breathing C3H mice (row I), mice breathing 10% oxygen for 15 min before and during irradiation with 15 Gy (row 2), or mice asphyxiated 5 minutes prior to 15 Gy irradiation (row 3). The content of DNA, also measured using the comet assay, is shown in row 4. Damage to unirradiated normal cells resulted in mean tail moments of 1.5-3.5. Dashed vertical lines distinguish roughly between aerobic and hypoxic populations.
Measurable fractions of hypoxic cells in air-breathing mice were not observed for any normal tissue other than testis (Fig. 3, row 1). However, fewer strand breaks were detected after irradiation of testis as a result of the high degree of DNA condensation which occurs during late stages of spermatogenesis (8). Since the total fluorescence intensity of the DNA binding fluorochrome can be measured using the comet assay, the response of testis cells with lN, 2N and 4N DNA content can be distinguished (Fig. 3, row 4). We did not observe more “background” damage in cells from testis as might be expected from results of Singh et al. (18) who measured damage to sperm cells using a similar assay; a possible explanation is that our NaOH concentration during lysis is IO-fold lower. When mice were allowed to breathe 10% oxygen for 15 min before and then during irradiation, cells with small tail moments, which we interpret to be hypoxic, were detected in bone marrow and testes, and the proportion of hypoxic cells in tumors was significantly increased (Fig. 3, row 2). DISCUSSION
There are several advantages to quantifying hypoxia using the comet assay: the radiobiologic hypoxic fraction is measured independent of cell type or DNA content, cellular hypoxia (not hypoxia associated with necrosis) is observed, and the small sample size necessary for analysis permits the use of fine needle aspirates. Rapid rejoining of strand breaks means that there is not likely to be a
problem in applying the comet assay repeatedly, providing adequate time is allowed for repair (2 h or more for low doses). However, on the negative side, a sample of tumor cells must be obtained immediately after a dose preferably in excess of 3.5 Gy to minimize repair prior to analysis, and the assay is, of course, unable to distinguish between clonogenic and nonclonogenic hypoxic cells. Since the average tail moment for tumor cells removed immediately after 15 Gy is only about 60% of the tail moment of cells irradiated on ice, considerable repair must occur prior to biopsy. Less repair appears to occur in normal tissues as evidenced by the higher tail moment measured immediately after irradiation. Meyn et al. (10, 11) used the alkali filter elution method to detect DNA damage in mouse tissues following irradiation. They found significant variability in the numbers of DNA strand breaks induced in different tissues; differences in strand break rejoining rates, sulphydryl content, and especially oxygenation were suggested as possible explanations. Some normal tissues may border on hypoxia (7) an observation that should be of considerable interest in the application of bioreductive agents. Early work established that hypoxia protected mice against whole body irradiation (4). While mice breathing 10% oxygen showed a 36% decrease in oxygen tension in the spleen (6) our results indicate that this reduction does not result in a significant increase in radiobiologic hypoxia in this organ (Fig. 3, row 2). Our data indicate that radiobiologically hypoxic cells do not appear in the bone marrow cells of mice
Hypoxic fraction using the comet assay 0 P. L. OLIVE ef al.
breathing air, however our samples did not include cells from compact bone which have been shown to be preferentially sensitive to killing by misonidazole (1). Nonetheless, even readily accessible bone marrow cells may border on hypoxia as indicated by the presence of a population of resistant cells in the bone marrow of mice breathing 10% oxygen (Fig. 3, row 2). Although extensive misonidazole binding in liver has suggested the presence of cells bordering on hypoxia in this tissue (19), our data do not indicate significant hypoxia even in livers of mice breathing 10% oxygen. Cells of the testis and epididymus showed less overall DNA damage than expected, partly due to the presence of late spermatids; in these cells, histone proteins are replaced with protamines so that fewer than half the number of single-strand breaks are produced for a given dose of radiation (8). An observation which requires closer examination is the displacement between the aerobic and hypoxic cell populations in comet histograms from tumors of airbreathing animals. The ratio of the slopes of the lines in Figure 2(l) indicate an OER of about 3.0. Note that the displacement between the peaks on the comet histograms (analogous to OER) is smaller than 3.0, in part because “background” damage has not been subtracted. With background damage present, we expected to observe a peak displacement of about 2.5, however, a value of only 1.9 was consistently observed for the SCCVII tumor. In
491
addition, the displacement between the peaks remained relatively constant for at least 30 min after irradiation. It is possible that aerobic tumor cells rejoin single-strand breaks more rapidly than energy-deficient hypoxic cells for the first 5 min of repair thus explaining the lower than expected displacement between the peaks. Since tumor cells rejoin breaks during irradiation, it was not possible to address this hypothesis directly. It should be noted, however, that oxygen levels must be extremely low to inhibit strand break rejoining in vitro (9). Moreover, the presence of hypoxic cells in air-breathing or clamped tumors does not appear to reduce rejoining rate compared to cells allowed to repair under well-oxygenated conditions in vitro, Figures l(b) and 2(a-c). A more likely explanation for the lower than expected displacement between the peaks is that cells which change their oxygenation status during irradiation (i.e., “acutely” hypoxic cells) will appear as cells intermediate in oxygenation in this assay, thereby reducing the apparent OER and decreasing the displacement between the aerobic and hypoxic peaks, (compare Figure l(d) and l(e)). Since acutely hypoxic cells are known to contribute significantly to hypoxia in the SCCVII tumor (3), this result is not unexpected. Interestingly, a sample of six human breast tumors (15) exhibited the expected displacement of 2.5 which could be taken as an indication that transient hypoxia plays a less important role in these human tumors.
REFERENCES 1 Allalunis-Turner,
M. J.; Chapman, J. D; Turner, A. R. Identification of a hypoxic population of bone marrow cells. lnt. J. Radiat. Oncol. Biol. Phys. 9:227-232; 1983. 2. Chapman, J. D.; Dugle, D. L.; Reuvers, A. P.; Meeker, B. E; Borsa, J. Studies on the radiosensitizing effect of oxygen in Chinese hamster cells. Int. J. Radiat. Biol. 26:383-389; 1974. 3. Chaplin, D. J.; Olive, P. L; Durand, R. E. Intermittent blood flow in a murine tumour: Radiobiological effects. Cancer Res. 47597-601; 1987. 4. Dowdy, A. H.; Bennett, L. R; Chastian, S. M. Protective action of anoxic anoxia against total body roentgen irradia5.
tion of mammals. Radiology 55:879-885; 1950. Hall, E. J.; Lehnert, S.; Roizin-Towle, L. Split dose-experiments
with hypoxic
cells. Radiology 112:425-430, 1974. H. D. Studies on spleen oxygen tension and radioprotection in mice with hypoxia, serotonin and p-aminopropiophenone. Radiat. Res. 3 1:389-399; 1967. Hendry, J. H. Quantitation of the radiotherapeutic importance of naturally hypoxic tissues from collated experiments with rodents using single doses. Int. J. Radiat. Oncol. Biol. Phys. 5:97 l-976; 1979. Joshi, D. S.; Yick, J.; Murray, D.; Meistrich, M. L. Stagedependent variation in the radiosensitivity of DNA in developing male germ cells. Radiat. Res. 12 1:274-28 1; 1990. Koch, C. J.; Painter, R. B. The effect of extreme hypoxia on the repair of DNA single-strand breaks in mammalian cells. Radiat. Res. 64:256-264; 1975. Meyn, R. E.; Jenkins, W. T; Murray, D. Radiation damage to DNA in various animal tissues: A comparison of yields and repair in vivo and in vitro. In: Simic, M. G., Grossman, L., Upton, A. C., eds. Mechanisms of DNA damage and repair. New York. NY: Plenum Press; 1986: 15 1-l 58.
6. Hasegawa, A. T.; Landahl,
I.
8.
9.
10.
1 I. Meyn, R. E.; Jenkins, W. T. Variation in normal and tumor tissue sensitivity of mice to ionizing radiation-induced DNA strand breaks in vivo. Cancer Res. 43:5668-5673; 1983. 12. Murray, D.; Meyn, R. E. Differential repair of gamma-rayinduced DNA strand breaks by various cellular subpopulations in mouse jejunal epithelium and bone marrow in vivo. Radiat. Res. 109:153-164; 1987. 13. Olive, P. L.; Bar&h, J. P. Induction and rejoining of radiation induced DNA single-strand breaks: “Tail moment” as a function of position in the cell cycle. Mutat. Res. (DNA Repair) 294:275-283; 1993. 14. Olive, P. L.; Durand, R. E. Detecting hypoxic cells in a murine tumor using the comet assay. JNCI 85:707-711; 1992. 15. Olive, P. L.; Durand, R. E.; IX Riche, J.; Olivotta, I.; Jackson, S. M. Gel electrophoresis of individual cells to quantify hypoxic fraction in human breast cancers. Cancer Res. 53: 133-736; 1993. 16. Olive, P. L.; Frazer, G.; Bar&h, J. P. Radiation-induced apoptosis measured in TK6 human B lymphoblast cells using the comet assay. Radiat. Res. 136: 130- 136; 1993. 17. Ostling, 0.; Johanson, K. J. Microelectrophoretic study of radiation-induced DNA damages in individual mammalian cells. Biochem. Biophys. Res. Commun. 123:291-298; 1984. 18. Singh, N. P.; Danner, D. B.; Tice, R. R.; McCoy, M. T.; Collins, G. D.; Schneider, E. L. Abundant alkali-sensitive sites in DNA of human and mouse sperm. Expt. Cell Res. 184:461-479; 1984. 19. Van Os-Corby, D. J.; Chapman, J. D. Z?z vitro binding of “C-misonidazo1e to hepatocytes and hepatoma cells. Int. J. Radiat. Oncol. Biol. Phys. 12: 125 l- 1254; 1986. 20. Wang, T. S.; Wheeler, K. T. Repair of x-ray-induced DNA damage in rat cerebellar neurons and brain tumor cells. Radiat. Res. 73:464-475, 1978.
Oncology
Pergamon
??Manipulation
HYPOXIC
Biol. Phys., Vol. 29. No. 3. pp. 487-491. 1994 Copyright 0 1994 Else&r Science Ltd Printed in the USA. All rights reserved 0360-3016/94 $6.00 + .OO
of Tumor Oxygen and Physiology
FRACTIONS MEASURED IN MURINE TISSUES USING THE COMET
TUMORS ASSAY
AND NORMAL
PEGGY L. OLIVE, PH.D., CHARLENE M. VIKSE, A.A. AND R.~LPH E. DURAND, PH.D. BritishColumbiaCancerResearchCentre,Vancouver,BC, Canada Purpose: To apply the alkaline comet assay to the detection of radiobiologically hypoxic cells in solid tumors and xtissues of mice, and to examine the influence of strand break repair on the oxygen enhancement ratio measured using the alkaline comet assay. Methods and Materials: In previous studies, we found that hypoxic fraction in squamous cell carcinomas growing in C3H mice could be reliably and easily measured using the alkaline comet assay. The comet assay applies fluorescence microscopy and image analysis to examine patterns of migration of deoxyribonucleic acid from individual cells embedded in agarose and exposed to an electric field. This method has sufficient resolution to detect subpopulations of hypoxic cells which show about 3 X fewer strand breaks than aerobic cells after irradiation. Results: Fast rejoining kinetics in vitro are comparable to those measured in Go, and rejoining of strand breaks Ghypoxic tumor cells occurs at a similar rate as rejoining in aerobic cells. Little residual damage was detectable using the comet assay in tumors 4-24 h following 15 Gy, allowing repeat measurements to be performed. Bone marrow and testis, but not liver, spleen, or jejunum contained a small fraction of hypoxic cells when mice breathed 10% oxygen during irradiation. Conclusion: The comet assay confirms that some normal tissues may border on hypoxia. Rejoining of strand breaks occurs rapidly in both oxic and hypoxic cells so that the oxygen enhancement ratio remains relatively constant with time after irradiation. Interestingly, a smaller oxygen enhancement ratio was observed in tumors than was expected, probably as a result of the presence of acutely hypoxic cells. Hypoxia, normal tissue, Hypoxia, tumor, DNA strand breaks, Transient hypoxia.
the comet assay compares favorably with hypoxic fraction measured using the paired-survival curve method (14). Recently this method has been used to detect hypoxic cells in human tumor fine needle aspirates removed from patients immediately after irradiation ( 1.5).The ability to identify individual hypoxic cells provides a new opportunity to examine questions relevant to tumor and tissue hypoxia. In this manuscript, repair rates are reported for both aerobic and hypoxic cells in vivo and compared to rejoining rates in vitro. In addition, the oxygenation status of normal tissues in C3H mice was examined since it has been suggested that some normal tissues may border on hypoxia (7, lo- 12), an observation which would be relevant to the successful application of hypoxic cell radiosensitizers or hypoxic cell cytotoxins.
INTRODUCTION
We recently described a method to quantify hypoxic fraction in cells from solid tumors ( 14, 15). The comet assay, based on a method originally described by Ostling and Johanson (I 7), uses fluorescence microscopy and image analysis to examine patterns of migration of deoxyribonucleic acid (DNA) from individual tumor cells embedded in agarose and exposed to an electric field. Hypoxic tumor cells are approximately 3 X less sensitive than aerobic cells to induction of DNA strand breaks by ionizing radiation (2) and can therefore be detected by the assay ( 14). The relation between oxygen concentration and production of strand breaks by radiation has been shown to be identical to the relation between oxygen concentration and cell killing by radiation (2), thus ensuring that cells determined to be hypoxic using the clinically relevant endpoint, cell survival, are also hypoxic when measured using the comet assay. In a transplanted squamous cell carcinoma growing in C3H mice (SCCVII), hypoxic fraction measured using
METHODS
AND
MATERIALS
Squamous cell carcinoma cells were transplanted subcutaneously over the sacral region of inbred male C3H/
Reprint requests to: Dr. Peggy L. Olive, Medical Biophysics Department, British Columbia Cancer Research Centre, 60 1 W. 10th Ave., Vancouver, BCVSZ 1L3 Canada. Acknowledgements-This work was supported by USPHS grant
number CA-37879 awarded from the National DHHS. Accepted for publication 8 October 1993.
487
Cancer Institute,
488
I. J. Radiation Oncology 0 Biology0 Physics
He mice, approximately 30 g in weight. Tumors were used for experimentation approximately 2 weeks later when they had reached a weight of 450-600 mg. Tumors were irradiated by placing unanesthetized mice in leadlined jigs and irradiating the tumor with 250 kVp X rays using parallel opposed fields at a dose rate of 4.5 Gy/min. Following irradiation, mice were sacrificed, tumors rapidly excised and placed in ice-cold phosphate buffer. A single cell suspension was prepared from the entire tumor by mincing the tissue and filtering the suspension through 50 ym nylon mesh. Cells were diluted to a concentration of 2-4 X lo4 cells/ml for the comet assay. For examination of damage to normal tissues, mice received whole body irradiation at a dose rate of 3.3 Gy/ min. For both tumor and normal tissues, approximately 30 s elapsed between the end of irradiation and removal of the tissue to ice-cold buffer. Large samples of tissue were used to obtain a single-cell suspension, thus ensuring that the sample was representative of the entire tissue. Tissues were minced in ice-cold phosphate buffer containing 20 mM ethylenediaminetetraacetic acid (EDTA)’ and 2% dimethylsulphoxide (DMSO),’ filtered through 50 pm nylon mesh, centrifuged and resuspended in phosphate-buffered saline. Bone marrow cells were removed from one tibia using a syringe. Jejunum (approx. 100 mg), spleen, and testis were removed and washed in ice-cold phosphate-buffered saline prior to mincing. With the exception of liver, all tissues were mechanically disaggregated at ice temperature. Liver (500 mg) was minced and incubated with trypsin,’ collagenase’ and deoxyribonuclease’ for 15 min at 37°C. Cells were embedded in type VII low gelling-temperature agarose,’ lysed in alkali and salt, and exposed to a low voltage to separate damaged from undamaged DNA. Comets were visualized by staining with propidium iodide,’ and patterns of DNA migration were quantified with a fluorescence microscope attached to an imaging system. Images of up to 600 cometsper-sample were digitized, and tail moment, a measurement of tail length and amount of DNA in the tail, was determined. Hypoxic fraction was calculated using a curve fitting algorithm applied to histograms describing number of comets vs. tail moment ( 14). Analysis of strand breaks in SCCVII tumor cells in vitro was performed using a cell line established from the tumor and maintained in exponential growth in a 5% oxygen atmosphere for 1 week. For anoxic irradiations, SCCVII cells were incubated in water-jacketed spinner culture flasks in Hepes buffered medium at 37°C with continuous gassing with oxygen-free nitrogen. After 1 h, cells were centrifuged and sealed in glass ampules at a density of 2 X 10’cells/ml followed by 15 min at 37°C to deplete any residual oxygen (5). Ampules were then cooled on ice and irradiated.
‘Sigma Chemical Co., St. Louis, MO.
Volume 29, Number 3, 1994
Time after 15 Gy (hr)
Fig. 1. Strand break rejoining kinetics in SCCVII tumors or cells irradiated and allowed to repair either in vivo or in vitro.(a) The
rejoining kinetics for tumor cells irradiated in vivo with 15 Gy. Tumors were excised at various times after irradiation. Data are extrapolated to the expected “zero” time point measured for cells (85% aerobic), irradiated in vitro on ice since cells from tumors could not be recovered in less than 5 min. (b) A comparison between early time points for tumor cells irradiated and allowed to repair in vitro (V) or tumors irradiated and allowed to repair in vivo (0). The “background” damage (mean f SD) is indicated by separate points. (c) The displacement between the aerobic and hypoxic peaks on the histograms (see Fig. 2) is shown for tumors (3-6-per-group, mean + SD) irradiated and allowed to repair in vivo.
RESULTS
The initial rate of single-strand break rejoining appears to be similar whether tumor cells are irradiated in the animal and allowed to rejoin in the animal or irradiated on ice and allowed to rejoin in culture at 37°C (Fig. 1). However, after lo- 15 min of repr lr, there is a discrepancy between in vivo and in vitro data which may be best explained by technical rather than biological factors. Absolute comparisons between these two data sets are limited by the inability to obtain samples from the mouse prior to significant repair of damage; the displacement between early time points for the two curves in Figure l(b) is a result of the repair that occurs in the tumor prior to excision. Approximately 50% of the single-strand breaks produced by 15 Gy are rejoined before removal of the tumor and preparation of a single-cell suspension. The repair half-time, measured in vitro for SCCVII tumor cells from data in Figure l(b), is 6.6 min (95% conf. limits 5.97.8) which compares well with strand break rejoining times for other cell types using the comet assay (13). A factor which has considerable influence on later repair times is the large variability in “background” damage between cells from different animals, see dotted lines in Figure 1(a), no doubt reflecting differences in damage induced by the tumor disaggregation procedure. Normalization procedures which subtract background damage prior to measurement of rejoining kinetics are therefore inappropriate for the in vivo data. Single-strand breaks are completely rejoined by 2 h after 15 Gy, Figure l(a), and there is little damage detectable
Hypoxic
fraction
489
using the comet assay 0 P. L. OLIVE e!a/
Air-breathing
Clamped
13
’ ”
1.30 Gy +
40
4.
I
’q
II’
15
min
.30 Gy + 30
min
K. 30 Gy + 4 hr
Dose (Gy)
Tail Moment Fig. 2. Histograms showing distributions of tail moments for comets of individual SCCVII tumors. (A-C) Response of tumors clamped prior to and following irradiation. (D) Mixture of cells from tumors of mice asphyxiated prior to irradiation with 15 Gy or 45 Gy. (E-K) Response of air-breathing mice to 15 or 30 Gy followed by a repair interval in viva (L) Mean tail moment of SCCVII tumor cells irradiated under aerobic or anoxic conditions. The oxygen enhancement ratio is 3:3.
even 24 h later when cell degradation may be expected, Figure l(a), Figure 2(h). For the comet assay, heavily damaged cells will most likely be lost during tumor disaggregation and during the lysis procedure, thus eliminating their contribution to overall repair kinetics. Apoptotic cells, which can be easily detected using the comet assay (16), do not influence rejoining rate estimates with this technique. Figure 2 verifies that tumor cells made hypoxic at the time of irradiation show less DNA damage than aerobic cells, and only one population of cells is present in clamped tumors “0,” 10 and 30 min after irradiation, Figure 2(a-c). For tumor cells from air-breathing animals, the displacement between the peaks for aerobic and hypoxic tumor cells remains relatively constant for cells allowed to repair damage in viva, Figure l(c), Figures 2(f, g, i, j), suggesting that hypoxic tumor cells rejoin breaks as rapidly as aerobic tumor cells. However, the displacement between the peaks is smaller than expected based on measurement of the oxygen enhancement ratio (OER) in vitro, even when background damage is taken into ac-
count. In Figure 2(d), a mixture of tumor cells from two mice asphyxiated prior to 15 Gy or 45 Gy gave a displacement between the peaks in the comet assay of 2.5. In comparison, tumors from air-breathing mice revealed a displacement of only 1.9 + 0.32 (19 tumors) between the aerobic and hypoxic populations. In addition, the mixed cell population gave much better resolution for the two peaks than observed for the tumor cells in Figure 2(e). An important advantage of the comet assay is the ability to detect DNA damage with excellent sensitivity without the necessity of radiolabelling cells. The number of strand breaks present immediately after irradiation in normal tissues of mice was consistently higher than observed in the SCCVII tumor (despite the lower dose rate), a confirmation that rejoining kinetics for strand breaks are generally slower in differentiated normal tissues than in tumor (12, 20). Tumor cells exposed to 15 Gy in viva showed a modal tail moment of about 20, bone marrow showed average values of about 25, and spleen and jejunum had values of 30 or higher.
490
1. J. Radiation Oncology 0 Biology 0 Physics
~~[-Jj.j-J 0
1530
Volume 29, Number 3, 1994
rjj--J:;~:/~:~~ 0
1530
0
15 30
0
1530
0
15 30
0
15 30
DNA Content Fig. 3. Histograms showing DNA damage measured using the comet assay for different normal tissues and SCCVII tumors from air-breathing C3H mice (row I), mice breathing 10% oxygen for 15 min before and during irradiation with 15 Gy (row 2), or mice asphyxiated 5 minutes prior to 15 Gy irradiation (row 3). The content of DNA, also measured using the comet assay, is shown in row 4. Damage to unirradiated normal cells resulted in mean tail moments of 1.5-3.5. Dashed vertical lines distinguish roughly between aerobic and hypoxic populations.
Measurable fractions of hypoxic cells in air-breathing mice were not observed for any normal tissue other than testis (Fig. 3, row 1). However, fewer strand breaks were detected after irradiation of testis as a result of the high degree of DNA condensation which occurs during late stages of spermatogenesis (8). Since the total fluorescence intensity of the DNA binding fluorochrome can be measured using the comet assay, the response of testis cells with lN, 2N and 4N DNA content can be distinguished (Fig. 3, row 4). We did not observe more “background” damage in cells from testis as might be expected from results of Singh et al. (18) who measured damage to sperm cells using a similar assay; a possible explanation is that our NaOH concentration during lysis is IO-fold lower. When mice were allowed to breathe 10% oxygen for 15 min before and then during irradiation, cells with small tail moments, which we interpret to be hypoxic, were detected in bone marrow and testes, and the proportion of hypoxic cells in tumors was significantly increased (Fig. 3, row 2). DISCUSSION
There are several advantages to quantifying hypoxia using the comet assay: the radiobiologic hypoxic fraction is measured independent of cell type or DNA content, cellular hypoxia (not hypoxia associated with necrosis) is observed, and the small sample size necessary for analysis permits the use of fine needle aspirates. Rapid rejoining of strand breaks means that there is not likely to be a
problem in applying the comet assay repeatedly, providing adequate time is allowed for repair (2 h or more for low doses). However, on the negative side, a sample of tumor cells must be obtained immediately after a dose preferably in excess of 3.5 Gy to minimize repair prior to analysis, and the assay is, of course, unable to distinguish between clonogenic and nonclonogenic hypoxic cells. Since the average tail moment for tumor cells removed immediately after 15 Gy is only about 60% of the tail moment of cells irradiated on ice, considerable repair must occur prior to biopsy. Less repair appears to occur in normal tissues as evidenced by the higher tail moment measured immediately after irradiation. Meyn et al. (10, 11) used the alkali filter elution method to detect DNA damage in mouse tissues following irradiation. They found significant variability in the numbers of DNA strand breaks induced in different tissues; differences in strand break rejoining rates, sulphydryl content, and especially oxygenation were suggested as possible explanations. Some normal tissues may border on hypoxia (7) an observation that should be of considerable interest in the application of bioreductive agents. Early work established that hypoxia protected mice against whole body irradiation (4). While mice breathing 10% oxygen showed a 36% decrease in oxygen tension in the spleen (6) our results indicate that this reduction does not result in a significant increase in radiobiologic hypoxia in this organ (Fig. 3, row 2). Our data indicate that radiobiologically hypoxic cells do not appear in the bone marrow cells of mice
Hypoxic fraction using the comet assay 0 P. L. OLIVE ef al.
breathing air, however our samples did not include cells from compact bone which have been shown to be preferentially sensitive to killing by misonidazole (1). Nonetheless, even readily accessible bone marrow cells may border on hypoxia as indicated by the presence of a population of resistant cells in the bone marrow of mice breathing 10% oxygen (Fig. 3, row 2). Although extensive misonidazole binding in liver has suggested the presence of cells bordering on hypoxia in this tissue (19), our data do not indicate significant hypoxia even in livers of mice breathing 10% oxygen. Cells of the testis and epididymus showed less overall DNA damage than expected, partly due to the presence of late spermatids; in these cells, histone proteins are replaced with protamines so that fewer than half the number of single-strand breaks are produced for a given dose of radiation (8). An observation which requires closer examination is the displacement between the aerobic and hypoxic cell populations in comet histograms from tumors of airbreathing animals. The ratio of the slopes of the lines in Figure 2(l) indicate an OER of about 3.0. Note that the displacement between the peaks on the comet histograms (analogous to OER) is smaller than 3.0, in part because “background” damage has not been subtracted. With background damage present, we expected to observe a peak displacement of about 2.5, however, a value of only 1.9 was consistently observed for the SCCVII tumor. In
491
addition, the displacement between the peaks remained relatively constant for at least 30 min after irradiation. It is possible that aerobic tumor cells rejoin single-strand breaks more rapidly than energy-deficient hypoxic cells for the first 5 min of repair thus explaining the lower than expected displacement between the peaks. Since tumor cells rejoin breaks during irradiation, it was not possible to address this hypothesis directly. It should be noted, however, that oxygen levels must be extremely low to inhibit strand break rejoining in vitro (9). Moreover, the presence of hypoxic cells in air-breathing or clamped tumors does not appear to reduce rejoining rate compared to cells allowed to repair under well-oxygenated conditions in vitro, Figures l(b) and 2(a-c). A more likely explanation for the lower than expected displacement between the peaks is that cells which change their oxygenation status during irradiation (i.e., “acutely” hypoxic cells) will appear as cells intermediate in oxygenation in this assay, thereby reducing the apparent OER and decreasing the displacement between the aerobic and hypoxic peaks, (compare Figure l(d) and l(e)). Since acutely hypoxic cells are known to contribute significantly to hypoxia in the SCCVII tumor (3), this result is not unexpected. Interestingly, a sample of six human breast tumors (15) exhibited the expected displacement of 2.5 which could be taken as an indication that transient hypoxia plays a less important role in these human tumors.
REFERENCES 1 Allalunis-Turner,
M. J.; Chapman, J. D; Turner, A. R. Identification of a hypoxic population of bone marrow cells. lnt. J. Radiat. Oncol. Biol. Phys. 9:227-232; 1983. 2. Chapman, J. D.; Dugle, D. L.; Reuvers, A. P.; Meeker, B. E; Borsa, J. Studies on the radiosensitizing effect of oxygen in Chinese hamster cells. Int. J. Radiat. Biol. 26:383-389; 1974. 3. Chaplin, D. J.; Olive, P. L; Durand, R. E. Intermittent blood flow in a murine tumour: Radiobiological effects. Cancer Res. 47597-601; 1987. 4. Dowdy, A. H.; Bennett, L. R; Chastian, S. M. Protective action of anoxic anoxia against total body roentgen irradia5.
tion of mammals. Radiology 55:879-885; 1950. Hall, E. J.; Lehnert, S.; Roizin-Towle, L. Split dose-experiments
with hypoxic
cells. Radiology 112:425-430, 1974. H. D. Studies on spleen oxygen tension and radioprotection in mice with hypoxia, serotonin and p-aminopropiophenone. Radiat. Res. 3 1:389-399; 1967. Hendry, J. H. Quantitation of the radiotherapeutic importance of naturally hypoxic tissues from collated experiments with rodents using single doses. Int. J. Radiat. Oncol. Biol. Phys. 5:97 l-976; 1979. Joshi, D. S.; Yick, J.; Murray, D.; Meistrich, M. L. Stagedependent variation in the radiosensitivity of DNA in developing male germ cells. Radiat. Res. 12 1:274-28 1; 1990. Koch, C. J.; Painter, R. B. The effect of extreme hypoxia on the repair of DNA single-strand breaks in mammalian cells. Radiat. Res. 64:256-264; 1975. Meyn, R. E.; Jenkins, W. T; Murray, D. Radiation damage to DNA in various animal tissues: A comparison of yields and repair in vivo and in vitro. In: Simic, M. G., Grossman, L., Upton, A. C., eds. Mechanisms of DNA damage and repair. New York. NY: Plenum Press; 1986: 15 1-l 58.
6. Hasegawa, A. T.; Landahl,
I.
8.
9.
10.
1 I. Meyn, R. E.; Jenkins, W. T. Variation in normal and tumor tissue sensitivity of mice to ionizing radiation-induced DNA strand breaks in vivo. Cancer Res. 43:5668-5673; 1983. 12. Murray, D.; Meyn, R. E. Differential repair of gamma-rayinduced DNA strand breaks by various cellular subpopulations in mouse jejunal epithelium and bone marrow in vivo. Radiat. Res. 109:153-164; 1987. 13. Olive, P. L.; Bar&h, J. P. Induction and rejoining of radiation induced DNA single-strand breaks: “Tail moment” as a function of position in the cell cycle. Mutat. Res. (DNA Repair) 294:275-283; 1993. 14. Olive, P. L.; Durand, R. E. Detecting hypoxic cells in a murine tumor using the comet assay. JNCI 85:707-711; 1992. 15. Olive, P. L.; Durand, R. E.; IX Riche, J.; Olivotta, I.; Jackson, S. M. Gel electrophoresis of individual cells to quantify hypoxic fraction in human breast cancers. Cancer Res. 53: 133-736; 1993. 16. Olive, P. L.; Frazer, G.; Bar&h, J. P. Radiation-induced apoptosis measured in TK6 human B lymphoblast cells using the comet assay. Radiat. Res. 136: 130- 136; 1993. 17. Ostling, 0.; Johanson, K. J. Microelectrophoretic study of radiation-induced DNA damages in individual mammalian cells. Biochem. Biophys. Res. Commun. 123:291-298; 1984. 18. Singh, N. P.; Danner, D. B.; Tice, R. R.; McCoy, M. T.; Collins, G. D.; Schneider, E. L. Abundant alkali-sensitive sites in DNA of human and mouse sperm. Expt. Cell Res. 184:461-479; 1984. 19. Van Os-Corby, D. J.; Chapman, J. D. Z?z vitro binding of “C-misonidazo1e to hepatocytes and hepatoma cells. Int. J. Radiat. Oncol. Biol. Phys. 12: 125 l- 1254; 1986. 20. Wang, T. S.; Wheeler, K. T. Repair of x-ray-induced DNA damage in rat cerebellar neurons and brain tumor cells. Radiat. Res. 73:464-475, 1978.