Int. J. Radiation Oncology Bid. Phys., Vol. in the U.S.A. All rights reserved.
0360-3016192 $5.00 + .M) Copyright B 1992 Pergamon Press plc
22, pp. 403-405
??Session A: Tissue Oxygenation Manipulation and Tumor Blood Flow
DEVELOPMENT OF AN ELISA FOR THE DETECTION OF 2-NITROIMIDAZOLE HYPOXIA MARKERS BOUND TO TUMOR TISSUE J. A. RALEIGH, PH.D.,’ E. M. ZEMAN, PH.D. ,l M. RATHMAN,’J. K. LADINE, B. SC., ’ J. M. CLINE, D. V. M.2 ANDD. E. THRALL, D. V. M.2 ‘Department of Radiation Oncology, University of North Carolina School of Medicine, Chapel Hill, NC 27599; and ‘Department of Anatomy, Physiological Sciences and Radiology, College of Veterinary Medicine, North Carolina State University, Raleigh, NC 27606 Canine and rodent tumors covaiently bind the fluorinated 2uitroimidaxole, CCI-103F, in a way that immunobistochemicai analysis shows is consistent with the location of tumor hypoxia. We have now developed a rapid, quantitative, and non-radioactive enzyme linked immunosorbent assay for the binding of CCI-103F in biopsy samples of spontaneous canine tumors. Issues of antigen stability during tissue processing, calibration of the ELBA, and the use of biopsy samples for measuring tumor hypoxia by the ELISA approach are addressed. Tumor hypoxia, Hypoxia markers, 2uitroimidazoles,
CCI-103F, ELISA, Canine tumors.
during tissue processing, a suitable calibration technique for the ELISA, and the variability in labeling in different regions of the same tumor. For these studies we have used the hexafluorinated 2_nitroimidazole, CCI-103F (Fig. l), which was originally prepared as a hypoxia marker for 19F MRS studies (5).
We have recently shown (1) by means of au immunohistochemical technique (6) that hypoxia marker binding occurs in vivo in a variety of spontaneous canine tumors in a way that is consistent with the expected location of hypoxic cells within the tumors. Morphometric analysis of multiple, immunoperoxidase stained sections from the ex-
cised tumors indicated hypoxic fractions ranging from 4 to 13%. The immunohistochemical analysis provided information about the number of cells that were labeled and the spatial distribution of these cells within the tumor. However, the analysis did not give the intensity of the labeling in terms of the number of marker molecules bound within hypoxic tumor tissue. From the point of view of developing non-invasive techniques based on hypoxia markers (e.g., 19F magnetic resonance spectroscopy (2-5) it is of interest to learn whether the overall intensity of tumor labeling is correlated with morphometrically determined hypoxic fractions. Radioactively-labeled hypoxia markers could be used for this purpose (7), but they are not convenient in a clinical setting. An alternative to radioactively-labeled markers is unlabeled markers in conjunction with an enzyme-linked immunosorbent assay (ELISA). Some of the issues involved in the development of an ELBA for quantifying the binding of hypoxia markers in tissue samples are addressed in this paper and include the questions of antigen stability
The hypoxia marker CCI-103F (Fig. l), the tritiated form of CCI-103F (specific activity 837 pCi/mg), the rabbit polyclonal antibodies to protein-bound CCI-103F, and the bovine serum albumin conjugated fonn of CCI-103F were prepared as described previously (1, 6). Alkaline phosphatase-labeled, goat antirabbit second antibody and its 4-nitrophenyl phosphate substrate were purchased. * Tritium content in tissue homogenates was measured with a scintillation counter.? Polystyrene microtiter plates were used in a standard ELISA technique. A kinetic plate reader$ was used to follow color development in the ELISA. The kinetic results are recorded as optical density units per minute. For the labeling of tumor and liver tissue, a 0.9% saline solution containing tritiated CCI-103F (specific activity reduced to 9.87 kCi/mg) was administered by way of the cephalic vein as a single, rapid infusion of 5-10 min duration to tumor-bearing dogs at a dose of 40 mg/kg body weight or approximately 118 p,M whole body concentra-
Reprint requests to: J. A. Raleigh, Ph.D., Department of Radiation Oncology, CB# 7512, UNC School of Medicine, Chapel Hill, NC 27599. Acknowledgements-The authors gratefully acknowledge financial support from the National Cancer Institute (CA-50995) and from the National Science Foundation (DI-9000801)
(M. Rathman). Accepted for publication 3 July 199 1. *Sigma Chemical Company. tBeckman Model LS5000 TA. $Molecular Devices V,,,. 403
I. J. Radiation Oncology 0 Biology 0 Physics
Volume 22, Number 3, 1992
Fig. 1. Structure of CCI-103F.
The dogs, which were scheduled for euthanasia, weighed 25.5 and 25.9 kg and carried a synovial sarcoma and a liposarcoma, respectively. Liver and tumor tissues were excised postmortem at 24 hr or approximately three plasma half lives of CCI-103F (1) after the administration of the tritiated marker. Tissues were frozen and stored for subsequent analysis. Weighed samples of tissue were minced and homogenized in four volumes of a pH 7.4 phosphate buffered saline solution containing Tween (PBS Tween) by means of a Dounce homogenizer. The protein content of the homogenate was determined by a BCA procedure with commercially available reagent and bovine serum albumin standard. 9 For ELISA and scintillation counting, 0.5 mL of the homogenate was incubated overnight at 37” with 0.5 mL of an ELISA compatible, pH 7.4 PBS Tween solution containing 1.O mg/mL of proteinase K. The insoluble material remaining after this treatment was pelleted by centrifugation and the clear supematant treated with 10 p.L of a 0.2 mM solution of phenylmethylsulfonyl fluoride to inactivate proteinase K. Aliquots of the digest supematant were analyzed for tritium content by scintillation counting. Serial dilutions of the digest were then analyzed by ELISA for the presence of the CCI-103F derived antigen. The insoluble pellets from the proteinase K digest were dissolved in 1 N KOH and their tritium content determined by scintillation counting. Thin layer chromatography of ethyl acetate extracts of the tumor tissue homogenates was performed on plastic backed silica gel plates with ethyl acetate as a solvent.
BOUND, rnicromob I kg
Fig. 2. Calibration curve for ELISA using canine liver tissue has been labeled in viva with tritium labeled CCI-103F. The cised tissue was digested as described in the text. The digest serially diluted and subjected to ELISA. The ELISA results expressed in terms of OD units/min. The data points represent results of two independent ELISA analyses.
that exwas are the
RESULTS The proteinase K digest of the homogenate from the synovial sarcoma left a relatively large insoluble pellet, but §Pierce Chemical Company.
neither this nor the smaller pellets of insoluble material from the liposarcoma and the liver homogenates possessed significant tritium counts, indicating that little, if any, of CCI-103F was bound to these unidentified tissue elements. The marker molecule, CCI-103F, is lipophilic (5) and conceivably could have been trapped in the liposarcoma without being covalently bound. Thin layer chromatography of ethyl acetate extracts of the liposarcoma digests showed, however, that no more than 1% of the radioactivity in the tumor was due to unbound CCI-103F. The ELISA of the tissues as described below shows that the antigen produced by the bioreductive binding of CCI-103F to tissue proteins is stable to the action of proteinase K. In addition, the antigen produced in vivo was found to be two orders of magnitude more efficient than the parent compound, CCI103F, as a competitive inhibitor (data not shown). Although the samples of tumor tissue were taken from widely spaced regions of the tumors in situ, the binding of CCI-103F was relatively constant within each tumor type with the liposarcoma binding in the range of 21.6 to 32.3 pmole/kg and the synovial sarcoma binding in the lower range of 5.6 to 13.9 p,mole/kg . As has been observed previously with rodent livers (4, 5), the dog livers bound CCI103F. The level of binding was 23.8 PM, which was calculated on the basis of tritium content and specific activity of CCI-103F. This level of binding is lower than that recorded for the binding of CCI-103F to rat livers (40-70 p,M) (4). The difference is most likely due to different exposure levels. The liver samples were used to calibrate the ELISA in terms of pmole CCI-103F bound per kg of tissue (Fig. 2). Serial dilutions of the tumor digests (Fig. 3) provided ELISA OD readings for the tumor samples which could be converted to p,mole CCI-103F bound in the tumors by reference to the calibration curve established with the liver sample. In Table 1, the ELISA values for binding
Detection of 2-nitroirnidazole hypoxia markers 0 J.
A. RALEIGHet al.
Table 1. Comparison of CCI-103F bound in tumor tissue as
measured by scintillation counting and by Elisa CCI-103F bound, @mole/kg Tumor sample*
doo4i ::;:_‘yi>e_e 1 0.001 SERIAL
DILUTION OF TISSUE
Fig. 3. ELISA analysis of the serially diluted Sl sample from the liposarcoma. ELISA OD/min units are plotted against the serial dilution factors.
are compared with those calculated from the tritium content of the same samples. The ELBA values are in reasonably good agreement with those calculated from tritium content when the comparison is made on the basis of the wet weight of the tissues. The ELBA also gave values in good agreement with the amount of tritium bound in terms of pmole CCI-103F bound per mg protein in the liposarcoma, but, for reasons that are not clear, the ELISA gave consistently lower values for this parameter in the case of the synovial sarcoma.
DISCUSSION The results show that ELBA is sensitive and is capable of giving binding results in reasonably good agreement with those obtained by scintillation counting of the radioactively labeled hypoxia marker. A calibrated ELISA will remove the need for radioactively-labeled CCI-103F in quantitative studies of the binding of the marker to tumor
Sl s2 s3 Ll L2 L3 L4 L5
32.3 21.6 29.5 7.7 8.2 5.6 13.9 6.5
38 31 29 7.7 8.2 3.5 16 4.3
(4) (7) (2) (0.9) (1) (0.6) (4) (0.3)
*The S series of samples are from the liposarcoma and the L series are from the synovial sarcoma. tMean and standard deviation (SD) of three measurements of CCI-103F bound in tumors from the serial dilution experiment using Figure 2 as a standard curve.
tissue. It should be useful in experimental studies of hypoxia marker binding and could be extended to other hypoxia markers. The more difficult question of whether ELISA on a limited series of biopsy samples would provide a reliable measure of tumor hypoxia in clinical specimens is not yet answered. It is perhaps interesting that, whereas the tumor samples in the present study varied in size from 210 to 600 mg, the range of CCI-103F bound was reasonably narrow for the synovial sarcoma (8.4 + 3.2, range 5.6 to 13.9 pmole/kg tissue) and for the liposarcoma (27.8 + 5.5, range 21.6 to 32.3 pmole/kg) with a clear distinction possible between the two types of tumor. It seems that tumor samples of this size and smaller (J. M. Cline, J. A. Raleigh, and D.E. Thrall, unpublished data, 1991) can be representative of the general degree of hypoxia marker binding in spontaneous tumor tissue.
REFERENCES Cline, J. M.; Thrall, D. E.; Page, R. L.; Franko, A. J.; Raleigh, J. A. Immunohistochemical detection of a hypoxia marker in spontaneous canine tumors. Br. J. Cancer 62:925931; 1990. Jin, G-Y.; Li, S-J.; Moulder, J. E.; Raleigh, J. A. Selective retention of fluorinated misonidazole (CCI-103F) in mouse tumors quantified by ‘I-I/i9F magnetic resonance spectroscopy. Int. J. Radiat. Biol. 58:1025-1034; 1990. Maxwell, R. J.; Workman, P.; Griffiths, J. R. Demonstration of tumor-selective retention of fluorinated nitroimidazole probes by 19F magnetic resonance spectroscopy in vivo. Int. J. Radiat. Oncol. Biol. Phys. 16:925-929; 1989. Raleigh, J. A.; Franko, A. J.; Kelly, D. A.; Trimble, L. A.; Allen, P. S. Development of an in viva i9F magnetic resonance method for measuring oxygen deficiency in tumors.
Magn. Res. Med. (In press) 1991. 5, Raleigh, J. A.; Franko, A. J.; Treiber, E. 0.; Lunt, J. A.; Allen, P. S. Covalent binding of a fluorinated 2-nitroimidazole to EMT/6 tumors in BALB/C mice. Detection by F-19 nuclear magnetic resonance at 2.3 T. Int. J. Radiat. Oncol. Biol. Phys. 12:1243-1245; 1986. Raleigh, J. A.; Miller, G. G.; Franko, A. J.; Koch, C. J.; Fuciarelli, A. F.; Kelly, D. A. Fluorescence immunohistochemical detection of hypoxic cells in spheroids and tumors. Br. J. Cancer 56:395400; 1987. Urtasun, R. C.; Chapman, J. D.; Raleigh, J. A.; Franko, A. J.; Koch, C. J. Binding of ‘H misonidazole to solid human tumors as a measure of tumor hypoxia. Int. J. Radiat. Oncol. Biol. Phys. 12:1263-1267; 1986.