Serum concentration: effects on cycle and x-ray sensitivity of mammalian cells.

If the content of serum in the culture medium of exponentially growing Chinese hamster cells is below optimum (15 percent), the doubling time and the resistance to x-irradiation of the cells are increased. In synchronously dividing populations the increase in doubling time is primarily caused by increase in duration of the postmitotic (G(1)) phase of the cells; this phase is relatively radiation resistant. The response of the cells growing synchronously is related quantitatively to the response of the cells dividing randomly.

Mammalian stress proteins HSP70 and HSP28 coinduced by nicotine and either ethanol or heat.

Cells exposed to a variety of stresses such as heat or ethanol respond by increasing their rate of synthesis of a set of proteins termed heat shock proteins (HSP). These proteins then appear to offer protection against the stressor and many other insults. The HSP also play important roles in unstressed cells. They are involved in the regulation of the cell cycle and during specific stages in the development of organisms. Exposure to stress during development (e.g., in pupal stages of insects or during gestation in mammals) leads to birth defects that are specific to the timing of the stress. It has been hypothesized that the ill-timed induction of HSP is responsible for this phenomenon. Epidemiological studies in humans have related teratogenic events to maternal exposures to hyperthermia or ethanol during pregnancy. The rate of alcohol-induced birth defects is greatly enhanced by smoking, suggesting a role for nicotine. Nicotine by itself, however, is said not to induce HSP. We hypothesized that nicotine may act as a coinducer (or facilitator) of stress responses. This possibility we tested on three levels: protection against heat (thermotolerance), induction of specific HSP, and binding of the heat shock transcription factor to the heat shock element. Each of these tests showed clearly that nicotine does indeed play such a role. This places nicotine in a novel position; to date, no other coinducers of stress responses have been reported. Our results may offer an explanation for the epidemiological data cited earlier.

Biochemical analysis of heat-resistant mouse tumor cell strains: a new member of the HSP70 family.

A series of heat-resistant mutants selected from a murine tumor cell line, RIF-1, display a markedly increased and stable resistance to heat shock. The mutant cell lines were analyzed for differences that may explain their increased resistance. Membrane lipid analysis showed no change in cholesterol content but an increase in the proportion of saturated fatty acids in the phospholipid fraction. Two-dimensional gel analysis revealed a generally increased constitutive synthesis of several major heat shock proteins (HSP), including HSP90, 68, 60, and 28. In addition, a new protein in the 70-kilodalton region is present in the resistant lines. The new protein has a lower isoelectric point than the constitutive HSP70 does, is only weakly induced by heat shock, and is immunologically cross-reactive with other members of the HSP70 family. After heat shock, the mutants display increases in HSP similar to those seen in the wild-type cells and they develop further transient tolerance to heat. Analysis of these mutants may help in understanding the function of HSP, both in normal growth and after heat shock.

Cytotoxic effects of hyperthermia and adriamycin on Chinese hamster cells.

Chinese hamster cells, exposed to simultaneous hyperthermia (43 degrees C) and adriamycin (ADM), were intially sensitized to the cytotoxic activity of ADM. If the duration of the combined treatment exceeded about 30 minutes, however, the cells became refractory to additional killing by ADM. If heat was applied before ADM exposure, the cells also could be rendered insensitive; this state persisted so that 24 hours after a 50-minute exposure to 43 degrees C, the cells still showed considerable resistance to ADM. The most reasonable explanation for the findings was that cell membrane permeability to ADM was initially increased by hyperthermia, but prolonged (greater than 30 min) heat exposure reversed this situation and inhibited additional ADM from penetrating to sensitive sites. However, the data yield no hints as to the precise mechanisms involved nor are they sufficiently precise to exclude other explanations. The results do point to precautions that must be observed if the combination of ADM and hyperthermia is to be used clinically.

Mechanism of antitumor activity of tumor necrosis factor alpha with hyperthermia in a tumor necrosis factor alpha-resistant tumor.

Cells from a radiation-induced fibrosarcoma (RIF-1) are exceedingly resistant to tumor necrosis factor alpha (TNF-alpha) in vitro. We tested whether the addition of mild hyperthermia (42.5 degrees C, 30 minutes) could enhance TNF-alpha activity against RIF-1 tumors growing in syngeneic hosts (C3H mice). TNF-alpha was administered intratumorally. Tumor cell killing essentially was not measurable following TNF-alpha, hyperthermia, or a combination of the two. Single-modality treatments also had no effect on tumor growth delay or on the x-ray dose (given 24 hours after the primary treatment) required to sterilize 50% of the tumors. The combination of TNF-alpha and hyperthermia, however, resulted in a marked increase in tumor doubling time and a highly significant reduction in the x-ray dose required to sterilize the tumors. Syngeneic lymph nodal lymphocytes and blood leukocytes did not appear to mediate the action of TNF-alpha on RIF-1 cells in vitro. Necrosis and hemorrhage were the most prominent histopathological alterations in the treated tumors. Electron microscopic studies 6 hours after therapy showed increased damage to capillary endothelial cells and accumulation of neutrophils in the capillaries of tumors treated with TNF-alpha with or without heat, suggesting that neutrophils may mediate the endothelial cell injury. These observations indicate a greater than additive tumoricidal effect of TNF-alpha with hyperthermia. Furthermore, they support the concept that the interaction between the two agents damages the vasculature, compromising the microcirculation and ultimately causing ischemic tumor necrosis.

Comparison of the malignant potential of 10T 1/2 cells and transformants with their survival responses to hyperthermia and to amphotericin B1.

Several responses of 10T 1/2 cells and of cells from strains transformed by X-irradiation or by methylcholanthrene exposure were examined. Malignant potential, as determined by the ability of the cells to initiate tumors in syngeneic hosts, either normal or rendered immunodeficient, varied by a factor of 10(6). The growth rate of tumors both in normal and in irradiated animals was closely correlated to malignant potential. The life spans of the animals were inversely correlated with malignant potential. Serum requirements were high for 10T 1/2 and for one of the X-ray transformants; neither of these cell strains gave rise to tumors when cells were injected into hosts. All malignant strains had uniformly lower serum requirements. While 10T 1/2 cells tended to be somewhat more heat resistant than were any of their transformed counterparts, this depended upon cell density (or state of growth) during heating. There was no correlation between malignant potential and heat sensitivity among the transformed strains. In fact, the most resistant transformant was also the most malignant; furthermore, its heat sensitivity was not modified by growth in vivo. No relationship between malignant potential and response to the polyene antibiotic amphotericin B could be observed.

Potential for therapy of drugs and hyperthermia.

The interaction of hyperthermia (41--45 degrees C) and chemotherapeutic agents frequently results in increased cytotoxicity over that predicted for an additive effect, although to date only a very limited number of drugs have been examined for such a possible interaction. At 42 degrees C, the upper limit of temperature useful for whole-body hyperthermia, the most promising agents of those examined to date appear to be the nitrosoureas and cis-platinum. Insufficient data exist for cyclophosphamide, whose long plasma half-life makes it an attractive candidate. Localized heating seems optimum at higher temperatures (43--45 degrees C). At these temperatures, not only those drugs effective at 42 degrees C but particularly bleomycin and possibly amphotericin B become candidates. No data exist in the literature on possible "thermic sensitizers," i.e., drugs which are noncytotoxic at 37 degrees C but which become effective at elevated temperatures. Two special cases are Adriamycin and actinomycin D. These drugs may be contraindicated for clinical use, since not only synergism but also protection by hyperthermia have been demonstrated, depending upon the time-sequence relationships of the heat and drug treatments.

Combination therapy: lonidamine, hyperthermia, and chemotherapy against the RIF-1 tumor in vivo.

Lonidamine enhances the cytotoxicity in vitro of several conventional antitumor drugs as well as that of hyperthermia (HT). We have investigated the possibility that such enhancement can also be demonstrated in vivo against the RIF-1 tumor system. Two assays were used to examine antitumor activity: tumor growth delay and clonogenicity of cells obtained from tumors from treated animals. We used drug (and HT) doses that by themselves did not achieve significant cell killing. The drugs whose interaction with lonidamine was tested were: cis-diamminedichloroplatinum (CDDP), mitomycin C (MMC), bleomycin, 5-fluorouracil, and nitrosourea. Of these only CDDP and MMC yielded positive data. Both assays gave essentially the same results, showing that antitumor activity reflected direct cell killing. CDDP and MCC activity was also enhanced by HT. When we combined all three modalities, however, the results of the trimodality therapies were no better than that of individual bimodality treatments. These last results suggest that lonidamine and HT have similar mechanisms, most likely inhibition of repair of DNA damage. Our data do suggest that lonidamine may have a role in multidrug therapies that include either CDDP or MMC as a component of the treatments.

A proposed operational model of thermotolerance based on effects of nutrients and the initial treatment temperature.

The phenomenon of thermotolerance in mammalian cells has been extensively documented in the literature. Because of its potential clinical importance as well as the fundamental biological interest, we pursued additional studies investigating pH and nutritional effects. Split-dose experiments were performed using plateau phase Chinese hamster HA-1 cells. The effects of the nutritional environment during the initial 43 degrees incubation and the second 43 degree treatment on the induction of thermotolerance were studied by comparing survival in full medium at pH 7.4 or in Hanks' balanced salt solution at pH 6.7. In additional experiments, we examined the effect on thermotolerance of changing the temperature of the initial treatment to 41 degrees. Thermotolerance was induced independently of the nutrient conditions of the first treatment. However, survival at the time of maximum expression of thermotolerance depended primarily on the duration and temperature of the initial treatment. Temperatures of 43 degrees or higher inhibited the development of thermotolerance during the first heat exposure. In contrast, if the initial exposure was at 41 degrees, thermotolerance was almost fully expressed by the end of this initial treatment. Changing pH in Hanks' balanced salt solution from 6.7 to 7.4 did not affect survival. On the basis of these and other data, we suggest that thermotolerance can be divided into three complementary and sometimes competing processes: an initial event ("trigger"); the expression of resistance ("development"); and its disappearance ("decay"). Trigger is induced at all hyperthermic temperatures while development requires a permissive temperature, i.e., less than 43 degrees for HA-1 cells. This model can provide plausible interpretations of several currently puzzling aspects of the survival kinetics of heat-exposed mammalian cells.

Thermosensitization by sulfhydryl compounds of exponentially growing Chinese hamster cells.

The effect of various sulfhydryl compounds on the survival of exponentially growing monolayer cultures of Chinese hamster cells (HA1) heated to temperatures of 37-43 degrees was examined. Concentrations of cysteamine which were nontoxic or minimally toxic at room temperature or 37 degrees became increasingly toxic at elevated temperatures, greatly potentiating the killing produced by heat alone in the absence of cysteamine. This enhancement of hyperthermia-induced cell killing increased with increasing cysteamine concentration, increasing duration of cysteamine exposure, and increasing temperature. Studies with synchronized Chinese hamster cells heated at 43 degrees for 1 hr in the presence of 16 mM cysteamine demonstrated that the potentiation of heat killing occurred in all phases of the cell cycle. Similarly, enhancement of hyperthermia-induced cell killing was seen for asynchronous cells exposed to 2-amino-ethylisothiourium bromides and cysteine, but the magnitude of the effect differed for the various sulfhydryl compounds.

Enhanced cell killing by bleomycin and 43 degrees hyperthermia and the inhibition of recovery from potentially lethal damage.

The effect of hyperthermia on bleomycin (BLEO) toxicity and repair was studied in "unfed" monolayer cultures of Chinese hamster cells. Synergy of toxicities was observed with simultaneous exposure to BLEO and 43 degrees. For example, when cells were exposed for 1 hr to BLEO (40 mug/ml) at 43 degrees, survival was reduced to 4 X 10(-5); separately, hyperthermia and 37 degrees BLEO exposure each resulted in a survival of 20%. Heating at 43 degrees prior to drug exposure at 37 degrees also produced substantial sensitization, indicating that the primary sensitizing effect involved cell damage rather than an increased rate of drug action; 41 degrees produced only modest cell sensitization to BLEO and the effect was not retained in cells heated prior to drug exposure. No increase in [14C]BLEO uptake was observed at 43 degrees over than at 37 degrees, and thus the increased cytotoxicity was not correlated with a gross change in cell permeability to BLEO, although increased drug availability to particular sensitive targets could not be ruled out. Studies of the repair kinetics after different 43 degrees BLEO protocols demonstrated that most of the cells sustaining potentially lethal damage rapidly recovered. However, 43 degrees hyperthermia inhibited this recovery and, with increasing durations of 43 degrees exposure, the fixation of potentially lethal damage was enhanced. Because of the substantial repair of BLEO damage observed in vivo, the possible usefulness of hyperthermia as an adjunct to BLEO therapy is discussed.

Effect of pH and elevated temperatures on the cytotoxicity of some chemotherapeutic agents on Chinese hamster cells in vitro.

The cytotoxicity of some drugs is a function of extracellular pH. We show that this dependence is greatly enhanced at 43 degrees. At this temperature, we have examined the cell-killing ability of four drugs in the pH range of 6.5 to 8.5. For 1,3-bis(2-chloroethyl)-1-nitrosourea, cytotoxicity is minimum at alkaline pH and increases in monotonic fashion as the milieu becomes more acidic. There is little or no effect of pH on the cytotoxicity of methotrexate. Bleomycin is most effective at acidic pH (less than 7.5). Above that pH, its cytotoxicity remains unchanged. Amphotericin B is least cytotoxic at the pH of normal tissue, 7.4. At higher or lower values, its cell-killing efficiency increases symmetrically. These results may have some relevance in designing thermo-chemotherapeutic treatment protocols.

Kinetic responses of murine sarcoma cells to radiation and hyperthermia in vivo and in vitro.

Cells of a solid mouse mammary sarcoma that can be cultured in vitro and which, upon inoculation, grow in vivo into new tumors, were exposed either in vivo or in vitro to doses of 300 or 600 rads of X-rays and/or to a temperature of 43 degrees for 1 hr. DNA histograms obtained with flow cytofluorometry were sampled at regular time intervals after treatments in order to obtain information on the cells' postexposure kinetics. X-irradiation of exponentially growing cells induced the expected G2 block; heat exposure caused cells to accumulate in S and G2. The sequential treatment (300 rads followed by 1 hr of hyperthermia) resulted in a mitotic delay that was longer than the sum of the delays of the individual treatments. The proliferative behavior of cycling cells in the tumor treated with a dose of X-rays was qualitatively similar to that seen for exponentially growing cells in vitro; however, marked differences were seen after 43 degrees exposure. The heat treatment of tumors in vivo caused a significant decrease in the tumor cell density as compared to the X-ray treatment alone. Sequential X-ray and heat treatment induced a higher fraction of cycling cells than that found in control tumors. However, X-ray or heat treatment alone caused no significant recruitment of resting cells into cycle 1 day after treatment. A model that permits estimation of the fraction of resting cells in a tumor is described.

Cytotoxicity of lonidamine alone and in combination with other drugs against murine RIF-1 and human HT1080 cells in vitro.

Lonidamine is an agent that is reported to inhibit recovery from potentially lethal damage. By itself, it has only mild anticancer activity. We have examined the ability of lonidamine to enhance the cytotoxicity of several drugs against a mouse and a human fibrosarcoma cell line in vitro. By itself, lonidamine showed only a limited cytotoxic effect with drug exposure up to 100 micrograms/ml and 24-h duration. Lower concentrations and shorter term exposures were not toxic to either of these tumor cell lines. When tested against the mouse line, the cytotoxicity of 5-fluorouracil, methotrexate, and etoposide was enhanced by lonidamine if the latter drug was given either before or after the exposure of the cells to the cytotoxic agents. For cisplatinum, bleomycin, mitomycin C, doxorubicin, and Actinomycin D, cytotoxicity was also enhanced, but only if lonidamine followed the other agents. In contrast, potentiation of 1,3-bis(2-chloroethyl)-1-nitrosourea toxicity was maximum when lonidamine preceded the nitrosourea. The human cells were more resistant to lonidamine and to the combination treatments than were the mouse cells. Nevertheless, substantial enhancement was seen particularly for cisplatin and mitomycin C. We examined in more detail the enhancement of cisplatin. Maximum interaction was obtained when lonidamine was given immediately following (or in conjunction with) the platinum agent. Our results suggest that lonidamine enhances the effects of several other agents in a time- and concentration-dependent manner and indicate a potential usefulness for lonidamine in multidrug therapy.

Comparison between tumor pH and cell sensitivity to heat in RIF-1 tumors.

We have performed a direct comparison between tumor pH and cell survival in heated RIF-1 tumors growing intradermally in thighs of C3H mice. The pH of individual tumors was measured by inserting a microelectrode into the tumor center. After pH measurements, the tumor-bearing legs were heated in a water bath (44.6 degrees C, 30 min). We then excised a small piece of tumor tissue (20-30 mg) from the area where the tip of the microelectrode had been placed and cell survival was determined by in vitro cloning. In some animals, 3 or 5 g/kg of glucose were injected i.p. 1.5 h before heating, to decrease the tumor pH. The average pH before heating was 6.83, 6.68, and 6.51, respectively, for tumors in untreated animals, those given 3 g/kg glucose, or those given 5 g/kg glucose. After heating, the average surviving fractions were 9.91 x 10(-2), 4.72 x 10(-2), and 5.43 x 10(-3), respectively. The cell yield did not differ significantly among the three treatments. As the tumor pH decreased, the surviving fraction also decreased for each treatment group. The correlation coefficient between tumor pH and log surviving fraction was highly significant for heat plus 5 g/kg glucose and all the treatment groups. The slope of the regression line obtained by a least squares method was steepest for all the treatment groups, followed by the heat plus 5 g/kg glucose and heat plus 3 g/kg glucose groups. The smallest slope of the regression line was noted for tumors treated by heat alone. The study shows that the tumor sensitivity to heat is enhanced when the tumor pH is lower and that adaptation of cells to low pH conditions may play a role in determining the relationship between pH and cell kill in RIF-1 tumors.

Ultrasound enhanced drug toxicity on Chinese hamster ovary cells in vitro.

Chinese hamster ovary cells (HA1) were exposed to therapeutic ultrasound (F = 2.025 MHz) in the presence of various drugs at temperatures of 37-43 degrees C. The space averaged intensities used were 0.5-2 W/cm2. The survival of these cells was subsequently tested using the clonogenic assay. Marked enhancement by ultrasound of the cytotoxicity of Adriamycin and amphotericin B was observed. For Adriamycin, the potentiation was dependent upon the intensity of sonication (exposure duration being 30 min). At 0.5 W/cm2, there was enhancement of cytotoxicity above 41 degrees C. At 1 W/cm2, there was a 3-order increase in cytotoxicity at 37 degrees C. Thus an increase in intensity resulted in a decrease in "threshold" temperature. The effect with Adriamycin could be explained in part by an increase in net uptake of drug into the cells. Further, ultrasound was observed to increase the sensitivity of cells to Adriamycin. For amphotericin B, the enhancement was observed only at exposure durations greater than 30 min and at 43 degrees C. There was no enhancement observed for cisplatin and etoposide. From these results, it appears that ultrasound potentiates the cytotoxicity of drugs the mode of action of which (at least in part) involves the plasma membrane.

Isolation and initial characterization of thermoresistant RIF tumor cell strains.

Heat-resistant cell strains were obtained from RIF-1 mouse tumor cells by repeated heatings of cells derived from survivors of previous heating cycles (60 min; 45 degrees C). Twenty thermally resistant (TR) strains were derived from single cells that had survived 11 heating and regrowth cycles. These were then analyzed for appropriate characteristics in vitro and in vivo. In vitro we looked for: marked heat resistance; high plating efficiency; growth rate similar to that of RIF-1 cells; and no obvious morphological abnormalities. In syngeneic hosts, we looked for: ability of the cells to form tumors whose growth rates were similar to that of RIF-1 tumors; high cellular heat resistance; good plating efficiency of tumor-derived cells; and low immunogenicity. Five strains having these desired characteristics were analyzed for survival kinetics. The heat-resistant phenotype was found to be stable in vitro, although partial reversion in vivo was seen occasionally. The "break" in the Arrhenius plot was found to occur at 45 degrees C in TR strains versus 43 degrees C in RIF-1. All TR strains and the RIF-1 line developed similar levels of thermotolerance (as defined by slope ratios) when given isosurvival heat exposures. X-ray responses of TR and RIF-1 cells were indistinguishable both with respect to survival and to heat-induced radiosensitization. While the number of live cells required to give tumor takes in 50% of the recipients for TR strains was appreciably higher than that for RIF-1 cells, radiation-killed cells from none of the strains were able to immunize efficiently against subsequent challenges by live cells.

Dansyl lysine, a new probe for assaying heat-induced cell killing and thermotolerance in vitro and in vivo.

Heat induces an increase in the fraction of cells staining with the nontoxic fluorescent membrane dye dansyl lysine (DL). The fraction of cells excluding DL can, under certain circumstances, be closely correlated to the fraction of cells surviving a heat treatment. Dansyl lysine has previously been shown to select for cholesterol-free membrane domains. We now describe the use of DL to provide rapid estimates of the kinetics of thermotolerance development and decay in vitro and in vivo. Following a 45 degree C-10-min heat shock, Chinese hamster ovary cells develop resistance to the lethal effects of a second heat treatment. This thermotolerance, as measured by both clonogenicity and resistance to DL staining, is maximal at approximately 12 h and gradually decays with a t1/2 of 2 to 3 days. DL staining also has utility in predicting the heat survival response in vivo. Radiation-induced fibrosarcomas grown in C3H mice were heated to 43 degrees C for 30 min. From 1 to 3 days later the tumors were excised and a single cell suspension was prepared. Tumor cells were then heated in vitro and assayed for survival or scored microscopically for DL staining. The two assays again yielded similar results showing maximal resistance by 19 h which decreased toward control heat sensitivity by 49 h. The evaluation of intrinsic heat sensitivity and induced thermotolerance is critical to rational treatment design in clinical hyperthermia. DL staining is rapidly quantitated by flow cytometry and can be applied to biopsy samples to provide estimates of heat sensitivity within hours.

Tumor responses following multiple hyperthermia and X-ray treatments: role of thermotolerance at the cellular level.

We have investigated the effect of increasing numbers of hyperthermia fractions given at 7-day intervals, with or without fractionated radiotherapy, on tumor cure, tumor growth, and cell survival after in vivo or in vitro heat. The murine RIF tumor was treated by capacitive radiofrequency hyperthermia at 44.0 degrees C for 20 min for one to five treatments at weekly intervals (1-5 wk D1). Single treatments (1 wk D1) induced cure in 5% of tumors. Additional treatments (2-5 wk D1) induced similar rates of cure (0-16%, P greater than 0.05 for 1 wk versus 2, 3, 4 or 5 wk D1). 1 wk D1 resulted in marked growth delay compared to controls. Mean tumor diameter doubling times increased from 13.2 days to 27.5 days (P less than or equal to 0.01). 2-5 wk D1 induced little additional growth delay (doubling times, 27.8-32.3 days, P greater than 0.05 for 1 wk versus 2, 3, 4 or 5 wk D1). Fractionated radiotherapy of 3200 rads (400 rads given twice each week) significantly prolonged mean tumor doubling time to 26.2 days. The addition of one hyperthermia session to the fractionated radiotherapy (1 wk D1 + XRT) further increased doubling time to 34.2 days (P less than or equal to 0.01). Additional treatments (2-5 wk D1 + XRT) only modestly increased doubling times (36.0-39.5 days, P greater than 0.05 for 1 wk versus 2, 3, 4 or 5 wk D1). In vitro assay of cells dissociated from tumors 5, 10, or 15 days after 3 wk D1 showed increased survival to 44 degrees C compared to previously untreated controls, and this cellular thermoresistance proved to be transient and noninheritable (i.e., thermotolerance). These results indicate that tumors can develop a prolonged thermal resistance after multiple weekly treatments which significantly modifies the response to subsequent treatment and which is associated with cellular thermotolerance.

Expression of HSP-28 and three HSP-70 genes during the development and decay of thermotolerance in leukemic and nonleukemic human tumors.

Leukemic cells appear to develop less thermotolerance and then to lose their thermotolerance more rapidly than do other tumor cell lines. The reason for this phenomenon is not known. After heat shock (or other environmental stresses), mammalian cells preferentially synthesize a set of proteins known as heat shock proteins (HSPs). HSP-28 and the various isoforms of HSP-70 have been suggested as being responsible for the development of thermotolerance. In these studies, we have attempted to determine by their expression with HSPs positively correlate with the development and decay of thermotolerance and whether the expression of these genes could explain the differing thermotolerance response observed between leukemic and nonleukemic tumor cells. Polymerase chain reaction was used to detect the expression of HSP-28 and several HSP-70 genes. Our data indicate that the expression of all three heat-inducible HSP-70 genes, 70A (Hunt and Morimoto, Proc. Natl. Acad. Sci. USA, 82: 6455-6459, 1985), 70B (Voellmy et al., Proc. Natl. Acad. Sci. USA, 82: 4949-4953, 1985), and 70B' (Leung et al., Biochem J., 267: 125-132, 1990) correlate with the development and decay of thermotolerance in nonleukemic tumor cell lines after heat or arsenite treatment. HSP-28 (Hickey et al., Nucleic Acids Res., 4: 4127-4145, 1986) failed to correlate with thermotolerance development; it was not induced after 45 degrees C primary heat shock. In leukemic cells, however, none of the HSPs were induced for extended periods of time. The lack of coordinate expression of HSP genes in cells of myeloid origin may explain the poor induction and maintenance of thermotolerance that is observed in these cells.