$5.00 + .Kl 0360-30 16192 Copyright 0 1992 Pergamon Press Ltd.
Inr. J Rodiulron Oncology Bid. Phy~y.. Vol. 23. pp. W-897 Pnnted I” the U.S.A. All nghts reserved.
??Rapid Communication
ACTIVATION OF THE HEAT SHOCK TRANSCRIPTION FACTOR BY HYPOXIA IN NORMAL AND TUMOR CELL LINES IN VZVO AND IN VZTRO AMATO J. GIACCIA, Ph.D. ‘, ELIZABETH A. AUGER, Ph.D.‘, ALBERT KOONG, B.S. l, DAVID J. TERRIS, M.D.*, ANDREW I. MINCHINTON, Ph.D.‘, GEORGE M. HAHN, Ph.D.’ AND J. MARTIN BROWN, D.Phi1.l ‘Departmentof Radiation Oncology, 2Departmentof Surgery,StanfordUniversity,Stanford,CA 94305 Cells exposed to hypoxia increase their synthesis of a specific set of proteins called oxygen regulated proteins. Recently, three of these proteins have been identified as hemoxygenase, Glucose Regulated Protein 78 kilodaltons and Glucose Regulated Protein 94 kilodaltons. In contrast, reoxygenation from hypoxic conditions increases the synthesis of the heat shock proteins. Although the molecular signals required for regulation of both sets of proteins by hypoxia and reoxygenation are still under investigation, it is known that their expression is regulated at the transcriptional level. This finding suggests that these stresses work either singularly or together to control the activation of nuclear transcription factors which bind distinct regulatory sequences in the promoter region of these genes. One possible nuclear transcription factor which could act as a transcriptional regulator for both hypoxia and reoxygenation gene transcription is the heat shock transcription factor. In this report, we focused on the kinetics of HSF activation by hypoxia in normal and tumor cell lines of murine and human origins. In cell culture, both the normal diploid cell line AG1522 and the tumor cell line JSQ-3 possess the same kinetics of HSF activation (binding to the heat shock element) by hypoxia, with maximal induction at or after 3 hr. We have also shown that the activation of HSF occurs in the SCCVII tumor in vim without clamping, but not in SCCVII cells grown in monolayers. When SCCVII tumors are dissociated and allowed to reoxygenate in cell culture, HSF binding decreased in 5 hr, and was undetectable after 18 hr. Furthermore, one human tumor biopsy tested for the presence of hypoxia by both the ~02 histograph (Eppendorf, Germany) and HSF binding showed good agreement for both techniques. These results suggest that HSF binding may be a useful marker for monitoring the tumor hypoxia. Hypoxia, Heat shock transcription factor, Gel retardation assay.
such as metabolism of the nitroimidazoles and tumor/ cellular microenvironment (4). A more direct approach to determine tumor hypoxia is the measurement of tumor oxygenation by pOZ histography (3). By using microelectrodes, the oxygen content of tumors can be determined. Besides being technically demanding, this approach may disturb the tumor microenvironment. Also, p02 histography cannot differentiate between viable and necrotic tissue. For these reasons, we have sought to develop a biochemical assay to measure cellular hypoxia by monitoring the production or activation of a specific protein. This approach is advantageous over pre-existing ones in that it requires no administration of chemicals or probes to the patient and would also allow us to dissect the molec-
INTRODUCIION Hypoxia is important both clinically in the management of solid tumors by radiotherapy and, from the basic science perspective, in the regulation of gene expression. Hypoxia has been well documented in rodent tumors of diverse histological origins, the principal means of quantitation being using radiobiological assays (5). An alternate approach for determining tumor hypoxia both in rodent and in human tumors is to inject radioactively or fluorescently labelled nitroimidazole compounds (e.g., misonizadole), reductive metabolites of which bind to the macromolecules of hypoxic cells (7). The degree of binding reflects the amount of hypoxia. However, the binding ability of these adducts may be influenced by other factors
Presentedin part at the 33rd Annual Scientific Meeting of
Work supported by grant No. CA 1520 1 fromthe US National CancerInstitute,DHHS (JMB) and grantNo. CA44665(GMH)
the American Society for Therapeutic Radiology November, 199 1. Accepted for publication 30 April 1992.
and an Institutional ACS grant (AJG). Reprint requests to: Amato J. Giaccia, Department of Radiation Oncology, Division of Radiation Biology, CBRL, GK 115, Stanford University School of Medicine, Stanford, CA 94305-5468. 891
and Oncology,
892
I. J.
Radiation
Oncology
0 Biology
0 Physics
ular signals that hypoxia transmits to the cell for protection against this form of stress. In the last few years, Sciandra et al. (11) and Heacock ef al. (2) have found a series of proteins whose synthesis is dependent upon the level of oxygen in the microenvironment, designated oxygen regulated proteins (ORPs). Although it would be a relatively simple task to develop radioimmunoassays to detect and quantify these proteins in a tumor biopsy, this assay would not distinguish between basal levels and induced levels of these proteins. In addition, not all of these proteins may be present in every tumor type and may also be subject to regulation by factors that are independent of hypoxia. Therefore, we have focused on characterizing the kinetics of activation and decay of the heat shock transcription factor (HSF) by hypoxia in cell lines and in tumors. The precedent for studying HSF activation by hypoxia comes from the work of Benjamin et al. (1) who have shown that HSF was activated within 2 hr of hypoxic treatment in mouse myogenic cells. The heat shock transcription factor is also activated by other stresses in addition to heat and hypoxia; this may complicate interpretation of in viva results. We demonstrate however, that the activation of HSF by hypoxia is much slower (hr) than by heat (min) and its decay is similarly slow, thus making it a good marker for hypoxia and reoxygenation.
METHODS
AND
MATERIALS
Cell culture The human line AGl522 was obtained from the American Type Culture Collection* and was routinely cultured at low passage in alpha MEM supplemented with 10% fetal calf serum. JSQ-3 was generously supplied by Dr. Jeffrey Schwartz, Argonne National Labs, and was maintained in 75% Dulbecco’s Modification of Eagle’s Medium and 25% Ham’s Nutrient Mixture F12 containing 20% fetal calf serum, 0.4 pg/mL hydrocortisone, 100 U/mL penicillin and 100 yg/mL streptomycin. SCCVII cells are derived from a spontaneous mouse squamous cell carcinoma of a C3H mouse and generously provided by Dr. Herman Suit, Massachusetts General Hospital. One human biopsy was obtained from a parotid mucoepidermoid carcinoma and was handled according to the guidelines established by the Stanford Human Subjects Committee.
Hypoxic treatment of’cells The day of the experiment, exponentially growing cells were refed with 2 mLs of fresh media and allowed to incubate for 4-6 hr in 60 mm glass petri dishes with notched sides before being exposed to hypoxia for various time intervals. These petri dishes were placed into specially designed aluminum chambers that were prewarmed at 37°C overnight. The aluminum chambers containing the cells were subjected to 5 rounds of gassing with 95% N2
* Rockville, Md.
Volume
23. Number
4. 1992
followed by evacuation while being slowly agitated on a reciprocating shaker with a 10 min interval between the last round of evacuation and degassing. Oxygen concentration was monitored using an oxygen sensor** to validate that the final level of oxygen in the media was 0.0002%. Each evacuation with Nz reduces the amount of oxygen by 1 log, e.g., 20% to 2%, 2% to 0.2%, etc. The pH of the media during N2 gassing is controled by the CO2 in the gassing mixture and the sodium bicarbonate in the media. In 10 min, the oxygen levels in the chambers goes from 20% to 0.002% and the aluminum chambers containing the cells are agitated for an additional 10 min before final evacuation to reduce the level of oxygen to 0.0002%. These conditions do not cause any cellular toxicity. After the final round of gassing and evacuation (pOz levels = 0.0002%), the chambers were sealed and incubated at 37°C on a reciprocating shaker for the desired time interval. After the hypoxic treatment, the seals of the aluminum jigs were broken, the cells were trypsinized from the plates, centrifuged, washed once with phosphate buffered saline at 5”C, and then recentrifuged to collect the pellet which was then frozen in liquid nitrogen and kept at -70°C until assay. Hypoxic conditions were also produced by placing 10’ cells in 200 ~1 of media in a 1 mL tuberculin syringe at 37°C and allowing the cells to respire the O? in the media. These cells were subsequently treated the same as those treated in jigs and were kept as frozen pellets before assay.
Determination yf‘HSF binding by gel retardation assay Nuclear extracts were isolated using the procedure of Zimarino and Wu with slight modifications (15). Briefly, exponentially growing cells were resuspended in 2 volumes ofextraction buffer (10 mM HEPES pH 7.9,0.4 M NaCl, 0.1 mM EGTA, 0.5 mM DTT, 5% glycerol, and 0.5 mM phenylmethylsulphonyl fluoride) and lysed by three rounds of freeze thawing. To ensure complete lysis, the cells/extracts were subjected to disruption in a 1 mL Dounce tissue homogenizer. These extracts were then centrifuged at 33,000 X g for 10 min. After centrifugation, the supernatants were frozen at -70°C until further use. Protein determinations were performed using the bicinchoninic acid technique ( 12). The gel shift assay was that of Zimarino and Wu ( 15) except that electrophoresis was performed on a 4.5% native polyacrylamide gel (4.94 acrylamide/0.06 bis) in 0.5 X TBE at 140 V at room temperature for 2.5 hr. The sequence of the duplex oligonucleotide used in these studies was 5’GTCGACGGATCCGAGCGCCTCGAATGTTCTAGAAAAGG-3’(noncoding strand) with the coding strand lacking the last 12 nucleotides from the 3’end. This oligonucleotide was labeled by Klenow fill-in reaction using 200 ng of oligonucleotide, 2 mM dATP, 2 mM dGTP, 2 mM dTTP, IOX Klenow buffer, 8 U Klenow, 50 PC1 32P-dCTP (spe-
** Controls
Katharobic,
Edmonton,
Canada.
Activation
of the heat shock transcription factor 0 A. J. GIACCIA et al.
cific activity > 6000 Ci/mmol) and water to a volume of 20 ~1. The gel was fixed by soaking in 7% acetic acid for 5 min, and then rinsed in distilled water before being dried down on Whatman 3 MM Chromatography paper. The gel was then autoradiographed by exposing Kodak XAR-2 film to the gels at -70°C with intensifying screens. All binding experiments were performed at least twice. Quantitation was performed on dried gels using a proportional gas flow scanner.***
893
TIME (hours) 0
1
2
3
4
5
6
6’
BINDING
RESULTS Schematic of protocol
The protocol for analysis of HSF is depicted in Figure 1. Briefly, cells were made hypoxic by either gassing and evacuation in jigs or by respiring away their oxygen. After hypoxic treatment, cells were trypsinized, counted and frozen at -70°C until use. Heat Shock transcription factor binding was assayed by incubating extracts of cells that were exposed to hypoxia for varying durations with a 32Plabeled DNA oligonucleotide (see nucleotide sequence above) derived from the heat shock element. The heat shock element is a portion of DNA from the promotor region of the heat shock genes to which HSF binds. After 15 min of incubation, the extract and DNA mixture were electrophoresed on a non-denaturing polyacrylamide gel. If HSF is activated, it will bind to the radioactively labeled oligonucleotide and retard its migration through the gel (lanes 2-6). If no activation occurs, then only the constitutive (non-specific and non-inducible) form of binding CELLS MADE HYPOXIC
TRYPSINIZED,
COUNTED, CELL PELLETS FROZEN
FREEZE THAW, HOMOGENIZE CELL PELLET, CENTRIFUGE & MIX HYPOXIC CELL EXTRACT (40 pg) WlTH 0.5 nglml 32P - HSE
NON-SPECIFIC CONSTITUTIVE
FREE HSE
Fig. 2. A representative autoradiograph depicting the induction of HSF in the normal human diploid fibroblast cell line AG 1522 as a function of time (hrs) under hypoxia. Constitutive binding and hypoxia induced binding of HSF are designated with arrows. Lanes l-6 designate time points after hypoxia treatment for O6 hrs. Lane 6’represents nuclear extracts that were mixed with a 200-fold molar excess of unlabeled HSE to ensure the specificity of binding.
will be seen. The key hypothesis of this assay is that the degree of activation (binding) of HSF reflects the degree of cellular hypoxia. Hypoxia activates HSF in normal and tumor cells
incubate 15 min at RT J Time in hrs Inducible
Load extracts on native Gel g 1 2 3 4 5 6 N.E.
I
-.a.
I
Constitutive
t-
Fig. 1. Protocol for using the gel shift assay to determine of hypoxia in normal and tumor cell lines.
*** AMBIS Systems,
San Diego, CA.
kinetics
Using the gel shift assay described above, we analyzed the activation of HSF in a normal human diploid fibroblast cell line AG 1522 (Fig. 2). For this cell line, at least 2 hr of radiobiologic hypoxia (~0~ < 5 mm Hg) are required for HSF activation. Maximal activation was found after 3 hr of exposure to hypoxia; a slow decline in HSF activation was detected in cells after longer hypoxic exposures. To demonstrate the specificity of binding, we added unlabeled HSE oligonucleotide at a 200-fold molar excess to compete with the 32P-labeled HSE for the activated HSF. Specificity is clearly seen in lanes 6 and 6’of Figure 2 which respectively represent HSF binding activity without and with the addition of unlabeled competitor.
894
I. J. Radiation Oncology 0 Biology 0 Physics
The addition of unlabeled competitor dilutes the radioactivity in the bound material so that it cannot be detected. In contrast to AG1522 cells where no activated HSF binding was detected in untreated controls, the squamous cell carcinoma cell line JSQ-3 possessed strong HSF binding in its untreated controls (Fig. 3). The reason for this binding is unclear, but the enhanced intrinsic thermoresistance of this cell line suggests that this activated HSF is functional because the surviving fraction of JSQ-3 cells to a 30 min exposure to heat at 45°C is 68% compared to 6% for another human tumor cell line, HT1080 (data not shown). Furthermore, the HSF that is bound in the untreated cells may be a different isoform of HSF or a different conformational form of HSF than that which is activated by hypoxia, since HSF binding in JSQ-3 cells as determined by gel scanning initially decreases before steadily increasing. Although JSQ-3 possessed activated HSF in its untreated controls, after the initial decrease, it exhibited similar kinetics of activation as AG 1522 cells. This is illustrated in Figure 4 which graphically demonstrates the kinetics of HSF activation in both cell lines as a function of time exposed to hypoxia. The activation of HSF is apparent in both normal and tumor cell lines by 2 hr, peaking at 3 hr in the normal cell line, and continuing at a slow increase in JSQ-3. For both cell lines, the kinetics TIME 1
SPECIFIC BINDING NONSPECIFIC CONSTANT
1.5
(hours) 2
2.5
3
45°C
_
Volume 23, Number 4, 1992
202
0
4
2 TIME
6 (HRS)
Fig. 4. Kinetics of HSF induction in the normal human diploid fibroblast cell line AC I522 and the human squamous cell carcinoma cell line JSQ-3 as a function of time (hr) under hypoxia. Although HSF binding differs between AC1522 and JSQ-3 in the unstressed state (time = 0), both cell lines still exhibit similar kinetics of induction by hypoxia. Error bars represent + S.E. of the data.
of HSF activation by hypoxia are appreciably slower than those for HSF activation by heat; activation of HSF by hypoxia requires hours, while HSF activation by heat occurs on the order of minutes (data not shown). We further tested whether the kinetics of HSF activation would be altered if the cells were made hypoxic by respiring their oxygen away instead of by gassing with Nl in aluminum chambers. As seen in Figure 5, this method of
-
TIME
(minutes)
7 0
NON-SPECIFIC CONSTlTUTiVE
-
5
5’ 15 15’ 30 30’ 60 60’ 90 90’ 45’ 45”
NO EXTRACT
BINDING
NON-SPECIFIC CONSTITUTWE
FREE HSE
FREEHSE
-
Fig. 3. A representative autoradiograph depicting the induction of HSF in the human squamous cell carcinoma cell line JSQ-3 as a function oftime (hrs) under hypoxia (lanes l-3). Constitutive and hypoxia induced binding of HSF are designated by arrows. Cells were exposed to 45°C for 45 min to serve as a positive control for HSF induction (lane 45°C). This cell line possesses an activated form of HSF in the unstressed state.
Fig. 5. A representative autoradiograph depicting the kinetics of HSF activation in JSQ-3 cells that were made hypoxic by respiring away their 02 as a function of time (min). Lanes are designated by the time of exposure to this treatment (0, 5 mitt, 15 min, etc.,). To determine whether this binding was specific for HSF, we added a 200-fold molar excess of unlabeled HSE to compete with the “P-labeled HSE (lanes 5’, 15’, etc.). An extract of cells treated for 45 min at 45°C is also included as a positive control (45’, 45’).
Activation of the heat shock transcription factor 0 A. J. GIACCIA
producing hypoxia results in maximal HSF activation in half the time seen in the experiments using the N2 atmosphere. The activation of HSF in this manner more closely resembles that seen by heat. However, these data are more difficult to interpret as the cells become not only hypoxic but also acidic, and intracellular pH may activate HSF independent of or in conjunction with hypoxia. This method of producing hypoxia more closely mimics the in viva microenvironment of the tumor, since the tumor’s microenvironment is subjected to both forms of stress.
el
al.
895
TIME AFTER TUMOR DISSOCIATION 0
0’
5
18
(hours)
EXTRACT
BINDING
HSF activation in vivo
Although HSF activation has been well documented in cell culture systems for a variety of different stresses, no one has yet studied the activation of HSF in animal tumors. Our initial experiments have focused on the murine SCCVII cell line which can be maintained as a cell line in culture or grown as a tumor. Figure 6 shows an in viva/ in vitro comparison of a SCCVII tumor assayed for the
NON-SPECIFIC CONSTITUTIVE
see-VII in vivo
in vitro FREE HSE
BINDING.
NON-SPECIFIC CONSTITUTIVE
FREE HSE
Fig. 6. A representative autoradiograph depicting the in viva/in vitro comparison of HSF activation in the murine tumor SCCVII without stress. Lanes are appropriately designated.
Fig. 7. A representative autoradiograph depicting the kinetics of HSF decay after 15 min of clamping, and reoxygenation for 5 and 18 hrs (lanes 0, 5, 18). Lane 0’contained a 200 fold molar excess of unlabeled HSE to compete with “P-labeled HSE for binding of HSF. The lane designated extract represents the binding reaction done without nuclear extracts.
presence of HSF. The first two lanes show the presence of HSF from an SCCVII tumor that was excised and immediately frozen in liquid nitrogen before assay. The third lane shows that SCCVII cells grown in culture do not normally possess an activated form of HSF. This implies that for this system, the activation of HSF is caused by the tumor microenvironment. To test this system further, we physically clamped the tumor for 15 min to create a hypoxic environment, removed and dissociated the tumor, and reoxygenated the resulting cell suspension in cell culture. Figure 7 shows a very strong activation for HSF after clamping (lane l), and slow kinetics of decay during reoxygenation (lanes 3-6). Therefore, the molecular signals which activate HSF in vitro, also seem to activate HSF in vivo. The demonstrated ability of this assay to detect tumor hypoxia is essentially worthless as a predicitive assay unless it can be used to quantify hypoxia. We compared the results of HSF induction with another means of measuring
hypoxia, the clinically used pOZ histograph. Figure 8 is an autoradiograph comparing the kinetics of HSF activation between AG 1522 cells (lanes O-6’) and a biopsy
896
1. J. Radiation
Biology 0 Physics
Oncology 0
1
2
3
4
5
6
6’
BX
i
BINDING
NON-SPECIFIC CONSTITUTIVE
FREE tiSE
Fig. 8. A representative autoradiograph depicting the comparison of HSF binding in a tumor biopsy (lane BX) with HSF binding in AG1522 cells as a function of time under hypoxia (lanes O6’). The biopsy displays little HSF binding.
from a large human mucoepidermoid tumor (lane BX). By comparing the small amount of HSF binding in the tumor to that found in AG 1522 by gel shift analysis, we concluded that this tumor was well-oxygenated. We compared these results with p0~ histography readings from three perpendicular diameters of this same tumor (Fig. 9); the pOZ histograph also showed it to be a well-oxygenated tumor. Hence, at least for this tumor, the two techniques are in good agreement.
M.A.: Parotid Mucoe idermoid Room A Pr L
25 _ ? e
20 _
25
50
75
100
125
4, 1992
DISCUSSION
TIME (hours) 0
Volume 23, Number
150
175
200
Tissue pO2 (mmW Fig. 9. Frequency histogram of 105 readings with Eppendorf microelectrodes for the same tumor (mean pOz = 76.4 mm Hg). These readings suggest that this is a well oxygenated tumor.
We have shown that the kinetics of HSF induction by hypoxia are similar for a normal human diploid fibroblast cell line, AG 1522, and the human tumor cell line JSQ-3. When JSQ-3 cells were made hypoxic by putting 10’ cells into 200 ~1 of medium and allowed to respire away their 02, we found that HSF activation was more rapid than when HSF was generated by hypoxic conditions in the jigs. However, the kinetics of HSF activation for both methods of producing hypoxia are on the order of hours compared to the activation of HSF by heat, which is on the order of minutes. Previously, it has been demonstrated that the activation of HSF by hypoxia or heat does not require de ~OVO protein synthesis, implying that posttranslational modification of the protein was responsible for activation ( 15). Although the post-translational modifications required for activation are still under investigation, phosphorylation has been strongly implicated in the activation process (14). Conformational changes in HSF going from a monomer to a multimer have also been proposed to account for its activation (13). However, it is still hard to envision how hypoxia is able to physically induce a conformational change in the HSF protein. A more reasonable hypothesis is that HSF is activated through a signal transduction pathway, and the difference in kinetics of activation by heat and hypoxia reflect their differential ability to induce this pathway. We have also demonstrated that the squamous cell carcinoma cell line JSQ-3 possesses activated HSF in an unstressed state. JSQ-3 is also intrinsically thermal resistant compared to other tumor cells, suggesting that its HSF activation is biologically relevant. Why should a cell have a constitutively activated HSF under conditions of less stress? It has previously been reported that two forms of HSF exist in mammalian cells (6, 8, 9). In mouse cells, the two HSF isoforms differ in the conditions under which they will bind to the DNA. The first form, mHSF1, is activated by heat shock and is unable to bind DNA under normal conditions. The second isoform, mHSF2, binds DNA under normal conditions, but its binding ability is lost upon heat shock. Both proteins are constitutively expressed by cells, although the levels of these two HSF isoforms in the cell are unknown. If these characteristics of the HSF isoforms are also true for the human HSFs, we can then hypothesize the following: In normal cells, a small amount of the constitutively binding HSF is made, but the majority of the HSF present is the inducible binding isoform. In the JSQ-3 cell line, a promoter mutation exists in the constitutively binding HSF gene, thus increasing the level of this HSF. The binding that is seen in the controls initially decreases before increasing, suggesting that a different HSF or conformation of HSF is required for binding under hypoxia. An alternate, and equally plausible possibility, is that the mechanism that activates the inducible HSF in viva is deregulated and thus keeps HSF in its activated state. The second expla-
Activation of the heat shock transcription factor 0 A. J. GIACCIAef al.
nation would imply that other cellular systems involved in this pathway would also be constitutively activated. Indeed, this is the case as JSQ-3 is also intrinsically radioresistant (10). Ultimately, we would like to use HSF activation to monitor tumor hypoxia. This information may be important in determining when to deliver radiation and hyperthermia in order to optimize their combined effects. For one human tumor biopsy, we have found a correlation
897
between the p02 histograph data and HSF activation as determined by gel-shift analysis. We are presently attempting to use human xenographs that possess different amounts of hypoxia in scid mice and correlate these data with our ~01 histograph data. Further analysis of HSF activation in tumors and normal tissue will allow us to gain new insights on the regulation of HSF in a multicellular system that is influenced by cell-cell contact and local microenvironments.
REFERENCES 1. Benjamin, I. J.; Kroger, B.; Williams, S. Activation of the heat shock transcription factor by hypoxia in mammalian cells. Proc. Natl. Acad. Sci. USA 87:6263-6267;1990. 2. Heacock, C. S.; Sutherland, R. M. Enhanced synthesis of stress proteins caused by hypoxia and relation to altered cell growth and metabolism. Br. J. Cancer 62:2 17-225; 1990. 3. Kallinowski, F.; Zander, R.; Hoeckel, M.; Vaupel, P. Tumor tissue oxygenation as evaluated by computerized-p02-histography. Int. J. Radiat. Oncol. Biol. Phys. 19:953-96 1;1990. 4. Ling, L. L.; Sutherland, R. M. Dependence of misonidazole binding on factors associated with hypoxic metabolism. Br. J. Cancer 56:389-393;1987. 5. Moulder, J. E.; Rockwell, S. Hypoxic fractions of solid tumors: Experimental techniques, methods of analysis, and a survey of existing data. Int. J. Radiat. Oncol. Biol. Phys. 10: 695-712;1984. 6. Rabindran, S.; Giorgi, G.; Clos, J.; Wu, C. Molecular cloning and expression of a human heat shock factor, HSF 1. Proc. Natl. Acad. Sci. USA 88:6906-69 10; 199 1. 7. Raleigh, J. A.; Miller, G. G.; Franko, A. J.; Koch, C. J.; Fuciarelli, A. F.; Kelly, D. A. Ruorescence immunohistochemical detection of hyoxic cells in spheroids and tumors. Br. J. Cancer 56:395-400;1987. 8. Sarge, K. D.; Zimarino, V.; Holm, K.; Wu, C.; Morimoto, R. I. Cloning and characterization of two mouse heat shock factors with distinct inducible and constitutive DNA binding ability. Genes. Dev. 5: 1902- 19 11; 199 1.
9. Schuetz, T. J.; Gallo, G. J.; Sheldon, L.; Tempst, P.; Kingston, R. E. Isolation of a cDNA for HSFZ: Evidence for two heat shock factor genes in humans. Proc. Natl. Acad. Sci. USA 88:6911-6915;1991. 10. Schwartz, J. L.; Mustafi, R.; Beckett, M. A.; Weichselbaum, R. R. Prediction of the radiation sensitivity of human squamous cell carcinoma cells using DNA filter elution. Radiat. Res. 123:1-6;1990. 11. Sciandra, J. J.; Subjeck, J. R.; Hughes, C. S. Induction of glucose-regulated proteins during anaerobic exposure and of heat-shock proteins after reoxygenation. Proc. Natl. Acad. Sci. USA 81:4843-4847;1984. 12. Smith, P. K.; Krohn, R. I.; Hermanson, G. T.; Mallia, A. K.; Gartner, F. H.; Provenzano, M. D.; Fujimoto, E. K.; Goeke, N. M.; Olson, B. J.; Klenk, D. C. Measurement of protein using bicinchoninic acid. Anal. Biochem. 150: 76-85;1985. 13. Sorger, P. K.; Nelson, H. C. M. Trimerization of a yeast transcriptional activator via a coiled-coil motif. Cell 59:807813;1989. 14. Sorger, P. K.; Pelham, H. R. B. Yeast heat shock transcription factor is an essential DNA binding protein that exhibits temperature-dependent phosphorylation. Cell 54:855864;1988. 15. Zimarino, V.; Wu, C. Induction of sequence-specific binding of Drosophila heat shock activator protein without protein synthesis. Nature 327:727-730;1987.
Inr. J Rodiulron Oncology Bid. Phy~y.. Vol. 23. pp. W-897 Pnnted I” the U.S.A. All nghts reserved.
??Rapid Communication
ACTIVATION OF THE HEAT SHOCK TRANSCRIPTION FACTOR BY HYPOXIA IN NORMAL AND TUMOR CELL LINES IN VZVO AND IN VZTRO AMATO J. GIACCIA, Ph.D. ‘, ELIZABETH A. AUGER, Ph.D.‘, ALBERT KOONG, B.S. l, DAVID J. TERRIS, M.D.*, ANDREW I. MINCHINTON, Ph.D.‘, GEORGE M. HAHN, Ph.D.’ AND J. MARTIN BROWN, D.Phi1.l ‘Departmentof Radiation Oncology, 2Departmentof Surgery,StanfordUniversity,Stanford,CA 94305 Cells exposed to hypoxia increase their synthesis of a specific set of proteins called oxygen regulated proteins. Recently, three of these proteins have been identified as hemoxygenase, Glucose Regulated Protein 78 kilodaltons and Glucose Regulated Protein 94 kilodaltons. In contrast, reoxygenation from hypoxic conditions increases the synthesis of the heat shock proteins. Although the molecular signals required for regulation of both sets of proteins by hypoxia and reoxygenation are still under investigation, it is known that their expression is regulated at the transcriptional level. This finding suggests that these stresses work either singularly or together to control the activation of nuclear transcription factors which bind distinct regulatory sequences in the promoter region of these genes. One possible nuclear transcription factor which could act as a transcriptional regulator for both hypoxia and reoxygenation gene transcription is the heat shock transcription factor. In this report, we focused on the kinetics of HSF activation by hypoxia in normal and tumor cell lines of murine and human origins. In cell culture, both the normal diploid cell line AG1522 and the tumor cell line JSQ-3 possess the same kinetics of HSF activation (binding to the heat shock element) by hypoxia, with maximal induction at or after 3 hr. We have also shown that the activation of HSF occurs in the SCCVII tumor in vim without clamping, but not in SCCVII cells grown in monolayers. When SCCVII tumors are dissociated and allowed to reoxygenate in cell culture, HSF binding decreased in 5 hr, and was undetectable after 18 hr. Furthermore, one human tumor biopsy tested for the presence of hypoxia by both the ~02 histograph (Eppendorf, Germany) and HSF binding showed good agreement for both techniques. These results suggest that HSF binding may be a useful marker for monitoring the tumor hypoxia. Hypoxia, Heat shock transcription factor, Gel retardation assay.
such as metabolism of the nitroimidazoles and tumor/ cellular microenvironment (4). A more direct approach to determine tumor hypoxia is the measurement of tumor oxygenation by pOZ histography (3). By using microelectrodes, the oxygen content of tumors can be determined. Besides being technically demanding, this approach may disturb the tumor microenvironment. Also, p02 histography cannot differentiate between viable and necrotic tissue. For these reasons, we have sought to develop a biochemical assay to measure cellular hypoxia by monitoring the production or activation of a specific protein. This approach is advantageous over pre-existing ones in that it requires no administration of chemicals or probes to the patient and would also allow us to dissect the molec-
INTRODUCIION Hypoxia is important both clinically in the management of solid tumors by radiotherapy and, from the basic science perspective, in the regulation of gene expression. Hypoxia has been well documented in rodent tumors of diverse histological origins, the principal means of quantitation being using radiobiological assays (5). An alternate approach for determining tumor hypoxia both in rodent and in human tumors is to inject radioactively or fluorescently labelled nitroimidazole compounds (e.g., misonizadole), reductive metabolites of which bind to the macromolecules of hypoxic cells (7). The degree of binding reflects the amount of hypoxia. However, the binding ability of these adducts may be influenced by other factors
Presentedin part at the 33rd Annual Scientific Meeting of
Work supported by grant No. CA 1520 1 fromthe US National CancerInstitute,DHHS (JMB) and grantNo. CA44665(GMH)
the American Society for Therapeutic Radiology November, 199 1. Accepted for publication 30 April 1992.
and an Institutional ACS grant (AJG). Reprint requests to: Amato J. Giaccia, Department of Radiation Oncology, Division of Radiation Biology, CBRL, GK 115, Stanford University School of Medicine, Stanford, CA 94305-5468. 891
and Oncology,
892
I. J.
Radiation
Oncology
0 Biology
0 Physics
ular signals that hypoxia transmits to the cell for protection against this form of stress. In the last few years, Sciandra et al. (11) and Heacock ef al. (2) have found a series of proteins whose synthesis is dependent upon the level of oxygen in the microenvironment, designated oxygen regulated proteins (ORPs). Although it would be a relatively simple task to develop radioimmunoassays to detect and quantify these proteins in a tumor biopsy, this assay would not distinguish between basal levels and induced levels of these proteins. In addition, not all of these proteins may be present in every tumor type and may also be subject to regulation by factors that are independent of hypoxia. Therefore, we have focused on characterizing the kinetics of activation and decay of the heat shock transcription factor (HSF) by hypoxia in cell lines and in tumors. The precedent for studying HSF activation by hypoxia comes from the work of Benjamin et al. (1) who have shown that HSF was activated within 2 hr of hypoxic treatment in mouse myogenic cells. The heat shock transcription factor is also activated by other stresses in addition to heat and hypoxia; this may complicate interpretation of in viva results. We demonstrate however, that the activation of HSF by hypoxia is much slower (hr) than by heat (min) and its decay is similarly slow, thus making it a good marker for hypoxia and reoxygenation.
METHODS
AND
MATERIALS
Cell culture The human line AGl522 was obtained from the American Type Culture Collection* and was routinely cultured at low passage in alpha MEM supplemented with 10% fetal calf serum. JSQ-3 was generously supplied by Dr. Jeffrey Schwartz, Argonne National Labs, and was maintained in 75% Dulbecco’s Modification of Eagle’s Medium and 25% Ham’s Nutrient Mixture F12 containing 20% fetal calf serum, 0.4 pg/mL hydrocortisone, 100 U/mL penicillin and 100 yg/mL streptomycin. SCCVII cells are derived from a spontaneous mouse squamous cell carcinoma of a C3H mouse and generously provided by Dr. Herman Suit, Massachusetts General Hospital. One human biopsy was obtained from a parotid mucoepidermoid carcinoma and was handled according to the guidelines established by the Stanford Human Subjects Committee.
Hypoxic treatment of’cells The day of the experiment, exponentially growing cells were refed with 2 mLs of fresh media and allowed to incubate for 4-6 hr in 60 mm glass petri dishes with notched sides before being exposed to hypoxia for various time intervals. These petri dishes were placed into specially designed aluminum chambers that were prewarmed at 37°C overnight. The aluminum chambers containing the cells were subjected to 5 rounds of gassing with 95% N2
* Rockville, Md.
Volume
23. Number
4. 1992
followed by evacuation while being slowly agitated on a reciprocating shaker with a 10 min interval between the last round of evacuation and degassing. Oxygen concentration was monitored using an oxygen sensor** to validate that the final level of oxygen in the media was 0.0002%. Each evacuation with Nz reduces the amount of oxygen by 1 log, e.g., 20% to 2%, 2% to 0.2%, etc. The pH of the media during N2 gassing is controled by the CO2 in the gassing mixture and the sodium bicarbonate in the media. In 10 min, the oxygen levels in the chambers goes from 20% to 0.002% and the aluminum chambers containing the cells are agitated for an additional 10 min before final evacuation to reduce the level of oxygen to 0.0002%. These conditions do not cause any cellular toxicity. After the final round of gassing and evacuation (pOz levels = 0.0002%), the chambers were sealed and incubated at 37°C on a reciprocating shaker for the desired time interval. After the hypoxic treatment, the seals of the aluminum jigs were broken, the cells were trypsinized from the plates, centrifuged, washed once with phosphate buffered saline at 5”C, and then recentrifuged to collect the pellet which was then frozen in liquid nitrogen and kept at -70°C until assay. Hypoxic conditions were also produced by placing 10’ cells in 200 ~1 of media in a 1 mL tuberculin syringe at 37°C and allowing the cells to respire the O? in the media. These cells were subsequently treated the same as those treated in jigs and were kept as frozen pellets before assay.
Determination yf‘HSF binding by gel retardation assay Nuclear extracts were isolated using the procedure of Zimarino and Wu with slight modifications (15). Briefly, exponentially growing cells were resuspended in 2 volumes ofextraction buffer (10 mM HEPES pH 7.9,0.4 M NaCl, 0.1 mM EGTA, 0.5 mM DTT, 5% glycerol, and 0.5 mM phenylmethylsulphonyl fluoride) and lysed by three rounds of freeze thawing. To ensure complete lysis, the cells/extracts were subjected to disruption in a 1 mL Dounce tissue homogenizer. These extracts were then centrifuged at 33,000 X g for 10 min. After centrifugation, the supernatants were frozen at -70°C until further use. Protein determinations were performed using the bicinchoninic acid technique ( 12). The gel shift assay was that of Zimarino and Wu ( 15) except that electrophoresis was performed on a 4.5% native polyacrylamide gel (4.94 acrylamide/0.06 bis) in 0.5 X TBE at 140 V at room temperature for 2.5 hr. The sequence of the duplex oligonucleotide used in these studies was 5’GTCGACGGATCCGAGCGCCTCGAATGTTCTAGAAAAGG-3’(noncoding strand) with the coding strand lacking the last 12 nucleotides from the 3’end. This oligonucleotide was labeled by Klenow fill-in reaction using 200 ng of oligonucleotide, 2 mM dATP, 2 mM dGTP, 2 mM dTTP, IOX Klenow buffer, 8 U Klenow, 50 PC1 32P-dCTP (spe-
** Controls
Katharobic,
Edmonton,
Canada.
Activation
of the heat shock transcription factor 0 A. J. GIACCIA et al.
cific activity > 6000 Ci/mmol) and water to a volume of 20 ~1. The gel was fixed by soaking in 7% acetic acid for 5 min, and then rinsed in distilled water before being dried down on Whatman 3 MM Chromatography paper. The gel was then autoradiographed by exposing Kodak XAR-2 film to the gels at -70°C with intensifying screens. All binding experiments were performed at least twice. Quantitation was performed on dried gels using a proportional gas flow scanner.***
893
TIME (hours) 0
1
2
3
4
5
6
6’
BINDING
RESULTS Schematic of protocol
The protocol for analysis of HSF is depicted in Figure 1. Briefly, cells were made hypoxic by either gassing and evacuation in jigs or by respiring away their oxygen. After hypoxic treatment, cells were trypsinized, counted and frozen at -70°C until use. Heat Shock transcription factor binding was assayed by incubating extracts of cells that were exposed to hypoxia for varying durations with a 32Plabeled DNA oligonucleotide (see nucleotide sequence above) derived from the heat shock element. The heat shock element is a portion of DNA from the promotor region of the heat shock genes to which HSF binds. After 15 min of incubation, the extract and DNA mixture were electrophoresed on a non-denaturing polyacrylamide gel. If HSF is activated, it will bind to the radioactively labeled oligonucleotide and retard its migration through the gel (lanes 2-6). If no activation occurs, then only the constitutive (non-specific and non-inducible) form of binding CELLS MADE HYPOXIC
TRYPSINIZED,
COUNTED, CELL PELLETS FROZEN
FREEZE THAW, HOMOGENIZE CELL PELLET, CENTRIFUGE & MIX HYPOXIC CELL EXTRACT (40 pg) WlTH 0.5 nglml 32P - HSE
NON-SPECIFIC CONSTITUTIVE
FREE HSE
Fig. 2. A representative autoradiograph depicting the induction of HSF in the normal human diploid fibroblast cell line AG 1522 as a function of time (hrs) under hypoxia. Constitutive binding and hypoxia induced binding of HSF are designated with arrows. Lanes l-6 designate time points after hypoxia treatment for O6 hrs. Lane 6’represents nuclear extracts that were mixed with a 200-fold molar excess of unlabeled HSE to ensure the specificity of binding.
will be seen. The key hypothesis of this assay is that the degree of activation (binding) of HSF reflects the degree of cellular hypoxia. Hypoxia activates HSF in normal and tumor cells
incubate 15 min at RT J Time in hrs Inducible
Load extracts on native Gel g 1 2 3 4 5 6 N.E.
I
-.a.
I
Constitutive
t-
Fig. 1. Protocol for using the gel shift assay to determine of hypoxia in normal and tumor cell lines.
*** AMBIS Systems,
San Diego, CA.
kinetics
Using the gel shift assay described above, we analyzed the activation of HSF in a normal human diploid fibroblast cell line AG 1522 (Fig. 2). For this cell line, at least 2 hr of radiobiologic hypoxia (~0~ < 5 mm Hg) are required for HSF activation. Maximal activation was found after 3 hr of exposure to hypoxia; a slow decline in HSF activation was detected in cells after longer hypoxic exposures. To demonstrate the specificity of binding, we added unlabeled HSE oligonucleotide at a 200-fold molar excess to compete with the 32P-labeled HSE for the activated HSF. Specificity is clearly seen in lanes 6 and 6’of Figure 2 which respectively represent HSF binding activity without and with the addition of unlabeled competitor.
894
I. J. Radiation Oncology 0 Biology 0 Physics
The addition of unlabeled competitor dilutes the radioactivity in the bound material so that it cannot be detected. In contrast to AG1522 cells where no activated HSF binding was detected in untreated controls, the squamous cell carcinoma cell line JSQ-3 possessed strong HSF binding in its untreated controls (Fig. 3). The reason for this binding is unclear, but the enhanced intrinsic thermoresistance of this cell line suggests that this activated HSF is functional because the surviving fraction of JSQ-3 cells to a 30 min exposure to heat at 45°C is 68% compared to 6% for another human tumor cell line, HT1080 (data not shown). Furthermore, the HSF that is bound in the untreated cells may be a different isoform of HSF or a different conformational form of HSF than that which is activated by hypoxia, since HSF binding in JSQ-3 cells as determined by gel scanning initially decreases before steadily increasing. Although JSQ-3 possessed activated HSF in its untreated controls, after the initial decrease, it exhibited similar kinetics of activation as AG 1522 cells. This is illustrated in Figure 4 which graphically demonstrates the kinetics of HSF activation in both cell lines as a function of time exposed to hypoxia. The activation of HSF is apparent in both normal and tumor cell lines by 2 hr, peaking at 3 hr in the normal cell line, and continuing at a slow increase in JSQ-3. For both cell lines, the kinetics TIME 1
SPECIFIC BINDING NONSPECIFIC CONSTANT
1.5
(hours) 2
2.5
3
45°C
_
Volume 23, Number 4, 1992
202
0
4
2 TIME
6 (HRS)
Fig. 4. Kinetics of HSF induction in the normal human diploid fibroblast cell line AC I522 and the human squamous cell carcinoma cell line JSQ-3 as a function of time (hr) under hypoxia. Although HSF binding differs between AC1522 and JSQ-3 in the unstressed state (time = 0), both cell lines still exhibit similar kinetics of induction by hypoxia. Error bars represent + S.E. of the data.
of HSF activation by hypoxia are appreciably slower than those for HSF activation by heat; activation of HSF by hypoxia requires hours, while HSF activation by heat occurs on the order of minutes (data not shown). We further tested whether the kinetics of HSF activation would be altered if the cells were made hypoxic by respiring their oxygen away instead of by gassing with Nl in aluminum chambers. As seen in Figure 5, this method of
-
TIME
(minutes)
7 0
NON-SPECIFIC CONSTlTUTiVE
-
5
5’ 15 15’ 30 30’ 60 60’ 90 90’ 45’ 45”
NO EXTRACT
BINDING
NON-SPECIFIC CONSTITUTWE
FREE HSE
FREEHSE
-
Fig. 3. A representative autoradiograph depicting the induction of HSF in the human squamous cell carcinoma cell line JSQ-3 as a function oftime (hrs) under hypoxia (lanes l-3). Constitutive and hypoxia induced binding of HSF are designated by arrows. Cells were exposed to 45°C for 45 min to serve as a positive control for HSF induction (lane 45°C). This cell line possesses an activated form of HSF in the unstressed state.
Fig. 5. A representative autoradiograph depicting the kinetics of HSF activation in JSQ-3 cells that were made hypoxic by respiring away their 02 as a function of time (min). Lanes are designated by the time of exposure to this treatment (0, 5 mitt, 15 min, etc.,). To determine whether this binding was specific for HSF, we added a 200-fold molar excess of unlabeled HSE to compete with the “P-labeled HSE (lanes 5’, 15’, etc.). An extract of cells treated for 45 min at 45°C is also included as a positive control (45’, 45’).
Activation of the heat shock transcription factor 0 A. J. GIACCIA
producing hypoxia results in maximal HSF activation in half the time seen in the experiments using the N2 atmosphere. The activation of HSF in this manner more closely resembles that seen by heat. However, these data are more difficult to interpret as the cells become not only hypoxic but also acidic, and intracellular pH may activate HSF independent of or in conjunction with hypoxia. This method of producing hypoxia more closely mimics the in viva microenvironment of the tumor, since the tumor’s microenvironment is subjected to both forms of stress.
el
al.
895
TIME AFTER TUMOR DISSOCIATION 0
0’
5
18
(hours)
EXTRACT
BINDING
HSF activation in vivo
Although HSF activation has been well documented in cell culture systems for a variety of different stresses, no one has yet studied the activation of HSF in animal tumors. Our initial experiments have focused on the murine SCCVII cell line which can be maintained as a cell line in culture or grown as a tumor. Figure 6 shows an in viva/ in vitro comparison of a SCCVII tumor assayed for the
NON-SPECIFIC CONSTITUTIVE
see-VII in vivo
in vitro FREE HSE
BINDING.
NON-SPECIFIC CONSTITUTIVE
FREE HSE
Fig. 6. A representative autoradiograph depicting the in viva/in vitro comparison of HSF activation in the murine tumor SCCVII without stress. Lanes are appropriately designated.
Fig. 7. A representative autoradiograph depicting the kinetics of HSF decay after 15 min of clamping, and reoxygenation for 5 and 18 hrs (lanes 0, 5, 18). Lane 0’contained a 200 fold molar excess of unlabeled HSE to compete with “P-labeled HSE for binding of HSF. The lane designated extract represents the binding reaction done without nuclear extracts.
presence of HSF. The first two lanes show the presence of HSF from an SCCVII tumor that was excised and immediately frozen in liquid nitrogen before assay. The third lane shows that SCCVII cells grown in culture do not normally possess an activated form of HSF. This implies that for this system, the activation of HSF is caused by the tumor microenvironment. To test this system further, we physically clamped the tumor for 15 min to create a hypoxic environment, removed and dissociated the tumor, and reoxygenated the resulting cell suspension in cell culture. Figure 7 shows a very strong activation for HSF after clamping (lane l), and slow kinetics of decay during reoxygenation (lanes 3-6). Therefore, the molecular signals which activate HSF in vitro, also seem to activate HSF in vivo. The demonstrated ability of this assay to detect tumor hypoxia is essentially worthless as a predicitive assay unless it can be used to quantify hypoxia. We compared the results of HSF induction with another means of measuring
hypoxia, the clinically used pOZ histograph. Figure 8 is an autoradiograph comparing the kinetics of HSF activation between AG 1522 cells (lanes O-6’) and a biopsy
896
1. J. Radiation
Biology 0 Physics
Oncology 0
1
2
3
4
5
6
6’
BX
i
BINDING
NON-SPECIFIC CONSTITUTIVE
FREE tiSE
Fig. 8. A representative autoradiograph depicting the comparison of HSF binding in a tumor biopsy (lane BX) with HSF binding in AG1522 cells as a function of time under hypoxia (lanes O6’). The biopsy displays little HSF binding.
from a large human mucoepidermoid tumor (lane BX). By comparing the small amount of HSF binding in the tumor to that found in AG 1522 by gel shift analysis, we concluded that this tumor was well-oxygenated. We compared these results with p0~ histography readings from three perpendicular diameters of this same tumor (Fig. 9); the pOZ histograph also showed it to be a well-oxygenated tumor. Hence, at least for this tumor, the two techniques are in good agreement.
M.A.: Parotid Mucoe idermoid Room A Pr L
25 _ ? e
20 _
25
50
75
100
125
4, 1992
DISCUSSION
TIME (hours) 0
Volume 23, Number
150
175
200
Tissue pO2 (mmW Fig. 9. Frequency histogram of 105 readings with Eppendorf microelectrodes for the same tumor (mean pOz = 76.4 mm Hg). These readings suggest that this is a well oxygenated tumor.
We have shown that the kinetics of HSF induction by hypoxia are similar for a normal human diploid fibroblast cell line, AG 1522, and the human tumor cell line JSQ-3. When JSQ-3 cells were made hypoxic by putting 10’ cells into 200 ~1 of medium and allowed to respire away their 02, we found that HSF activation was more rapid than when HSF was generated by hypoxic conditions in the jigs. However, the kinetics of HSF activation for both methods of producing hypoxia are on the order of hours compared to the activation of HSF by heat, which is on the order of minutes. Previously, it has been demonstrated that the activation of HSF by hypoxia or heat does not require de ~OVO protein synthesis, implying that posttranslational modification of the protein was responsible for activation ( 15). Although the post-translational modifications required for activation are still under investigation, phosphorylation has been strongly implicated in the activation process (14). Conformational changes in HSF going from a monomer to a multimer have also been proposed to account for its activation (13). However, it is still hard to envision how hypoxia is able to physically induce a conformational change in the HSF protein. A more reasonable hypothesis is that HSF is activated through a signal transduction pathway, and the difference in kinetics of activation by heat and hypoxia reflect their differential ability to induce this pathway. We have also demonstrated that the squamous cell carcinoma cell line JSQ-3 possesses activated HSF in an unstressed state. JSQ-3 is also intrinsically thermal resistant compared to other tumor cells, suggesting that its HSF activation is biologically relevant. Why should a cell have a constitutively activated HSF under conditions of less stress? It has previously been reported that two forms of HSF exist in mammalian cells (6, 8, 9). In mouse cells, the two HSF isoforms differ in the conditions under which they will bind to the DNA. The first form, mHSF1, is activated by heat shock and is unable to bind DNA under normal conditions. The second isoform, mHSF2, binds DNA under normal conditions, but its binding ability is lost upon heat shock. Both proteins are constitutively expressed by cells, although the levels of these two HSF isoforms in the cell are unknown. If these characteristics of the HSF isoforms are also true for the human HSFs, we can then hypothesize the following: In normal cells, a small amount of the constitutively binding HSF is made, but the majority of the HSF present is the inducible binding isoform. In the JSQ-3 cell line, a promoter mutation exists in the constitutively binding HSF gene, thus increasing the level of this HSF. The binding that is seen in the controls initially decreases before increasing, suggesting that a different HSF or conformation of HSF is required for binding under hypoxia. An alternate, and equally plausible possibility, is that the mechanism that activates the inducible HSF in viva is deregulated and thus keeps HSF in its activated state. The second expla-
Activation of the heat shock transcription factor 0 A. J. GIACCIAef al.
nation would imply that other cellular systems involved in this pathway would also be constitutively activated. Indeed, this is the case as JSQ-3 is also intrinsically radioresistant (10). Ultimately, we would like to use HSF activation to monitor tumor hypoxia. This information may be important in determining when to deliver radiation and hyperthermia in order to optimize their combined effects. For one human tumor biopsy, we have found a correlation
897
between the p02 histograph data and HSF activation as determined by gel-shift analysis. We are presently attempting to use human xenographs that possess different amounts of hypoxia in scid mice and correlate these data with our ~01 histograph data. Further analysis of HSF activation in tumors and normal tissue will allow us to gain new insights on the regulation of HSF in a multicellular system that is influenced by cell-cell contact and local microenvironments.
REFERENCES 1. Benjamin, I. J.; Kroger, B.; Williams, S. Activation of the heat shock transcription factor by hypoxia in mammalian cells. Proc. Natl. Acad. Sci. USA 87:6263-6267;1990. 2. Heacock, C. S.; Sutherland, R. M. Enhanced synthesis of stress proteins caused by hypoxia and relation to altered cell growth and metabolism. Br. J. Cancer 62:2 17-225; 1990. 3. Kallinowski, F.; Zander, R.; Hoeckel, M.; Vaupel, P. Tumor tissue oxygenation as evaluated by computerized-p02-histography. Int. J. Radiat. Oncol. Biol. Phys. 19:953-96 1;1990. 4. Ling, L. L.; Sutherland, R. M. Dependence of misonidazole binding on factors associated with hypoxic metabolism. Br. J. Cancer 56:389-393;1987. 5. Moulder, J. E.; Rockwell, S. Hypoxic fractions of solid tumors: Experimental techniques, methods of analysis, and a survey of existing data. Int. J. Radiat. Oncol. Biol. Phys. 10: 695-712;1984. 6. Rabindran, S.; Giorgi, G.; Clos, J.; Wu, C. Molecular cloning and expression of a human heat shock factor, HSF 1. Proc. Natl. Acad. Sci. USA 88:6906-69 10; 199 1. 7. Raleigh, J. A.; Miller, G. G.; Franko, A. J.; Koch, C. J.; Fuciarelli, A. F.; Kelly, D. A. Ruorescence immunohistochemical detection of hyoxic cells in spheroids and tumors. Br. J. Cancer 56:395-400;1987. 8. Sarge, K. D.; Zimarino, V.; Holm, K.; Wu, C.; Morimoto, R. I. Cloning and characterization of two mouse heat shock factors with distinct inducible and constitutive DNA binding ability. Genes. Dev. 5: 1902- 19 11; 199 1.
9. Schuetz, T. J.; Gallo, G. J.; Sheldon, L.; Tempst, P.; Kingston, R. E. Isolation of a cDNA for HSFZ: Evidence for two heat shock factor genes in humans. Proc. Natl. Acad. Sci. USA 88:6911-6915;1991. 10. Schwartz, J. L.; Mustafi, R.; Beckett, M. A.; Weichselbaum, R. R. Prediction of the radiation sensitivity of human squamous cell carcinoma cells using DNA filter elution. Radiat. Res. 123:1-6;1990. 11. Sciandra, J. J.; Subjeck, J. R.; Hughes, C. S. Induction of glucose-regulated proteins during anaerobic exposure and of heat-shock proteins after reoxygenation. Proc. Natl. Acad. Sci. USA 81:4843-4847;1984. 12. Smith, P. K.; Krohn, R. I.; Hermanson, G. T.; Mallia, A. K.; Gartner, F. H.; Provenzano, M. D.; Fujimoto, E. K.; Goeke, N. M.; Olson, B. J.; Klenk, D. C. Measurement of protein using bicinchoninic acid. Anal. Biochem. 150: 76-85;1985. 13. Sorger, P. K.; Nelson, H. C. M. Trimerization of a yeast transcriptional activator via a coiled-coil motif. Cell 59:807813;1989. 14. Sorger, P. K.; Pelham, H. R. B. Yeast heat shock transcription factor is an essential DNA binding protein that exhibits temperature-dependent phosphorylation. Cell 54:855864;1988. 15. Zimarino, V.; Wu, C. Induction of sequence-specific binding of Drosophila heat shock activator protein without protein synthesis. Nature 327:727-730;1987.
