INTRODUCTION

Upon exposure to a variety of harmful environmental conditions, cells respond by initiating the cellular stress response. Also termed the heat shock response due to the conditions under which it was first characterized, this phenomenon results in the increased expression of a family of proteins called heat shock proteins (hsps), which act to protect the cell from the harmful effects of the adverse environmental conditions. hsps act by binding to malfolded proteins caused by the stressful conditions and aiding in their folding back to the native state (1-3).

In addition to the heat-induced expression of hsps, previous studies have also examined the induction of the stress response by cold treatment or hypothermia. It has been shown that Drosophila larvae (4), cultured human diploid lung fibroblast cells (5), and Leuconostoc esenteroides (6) respond to cold exposure by synthesizing heat shock proteins during a posthypothermic recovery period. However, the mechanism by which cold treatment causes the induction of hsp expression is not well characterized. In order to explore the mechanism for cold induction of hsp expression, we performed a detailed analysis of the major parameters of the stress response in vivo in mouse tissues following hypothermic treatment. We found that the stress response was induced in all tissues examined at the levels of HSF¹ DNA binding and hsp mRNA synthesis. This is in contrast to results published previously (7), which suggest that cold induction of the stress response only occurs in brown adipose tissue of mice. We also observed that hsp expression is increased by hypothermia alone but that recovery is necessary for maximal stimulation of hsp expression. Furthermore, we show that cold induction of the stress response is mediated by HSF1 but that the HSF1 protein activated by cold treatment does not exhibit the characteristic hyperphosphorylation observed for HSF1 protein activated by heat treatment (8), suggesting that distinct mechanisms may be involved in mediating cold- versus heat-induced cellular stress responses. The biological and medical implications of these findings are discussed.

EXPERIMENTAL PROCEDURES

Adult male C3H/HeNCr MTV mice were obtained from Charles River Laboratories and maintained under a controlled light cycle (14 h of light and 10 h of darkness) at 22 ± 1 °C. Cold shock was performed by incubating single mice in standard plastic tubs in a 2-3 °C incubator for 8 h with food and water readily available. These treatment conditions decreased the rectal core temperatures from 36.5 to 34.0 °C, giving an average decrease in temperature of 2.5 °C. Where indicated, recovery was performed by transferring mice to a plastic tub at 22 °C for 1 h. These studies were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Animals were euthanized by cervical dislocation, and tissues were harvested immediately and quick frozen on dry ice. Tissues were minced with a sharp razor blade, and proteins were extracted by the addition of 5 volumes of 20 mM HEPES (pH 7.9), 25% (v/v) glycerol, 0.42 M NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol, 7 µg/ml pepstatin A. The tissues were further disrupted by grinding against the microcentrifuge tube wall and gentle pipetting. Insoluble material was removed by centrifugation in a microcentrifuge at 4 °C. The protein content of the supernatant was determined by absorbance readings at 205 nm, and unused aliquots were quick frozen in a dry ice/ethanol slurry and stored at 80 °C.

Gel shift analysis was performed using equal quantities of extracts by incubation in the presence of a ³²P-labeled self-complementary HSE consensus oligonucleotide (5-CTAGAAGCTTCTAGAAGCTTCTAG-3) as described previously (8). For experiments involving antibodies, 1 µl of a 1:50 dilution of polyclonal antibodies specific for HSF1 or HSF2 was added to extract aliquots and incubated at 22 °C for 5 min prior to the gel shift analysis.

Western blot analysis of tissue extracts was performed as described previously using a polyclonal antibody specific for HSF1 (8).

RNA was prepared by homogenization of the tissues in 4 M guanidine thiocyanate, 50 mM Tris-Cl (pH 7.5), 10 mM EDTA, 0.5% N-lauroylsarcosine, 1% -mercaptoethanol using a Biospec Products, Inc. Tissue-Tearor. Aggregated material was removed by centrifugation for 5 min in a microcentrifuge. 1400 µl of homogenate was loaded onto a 500-µl 5.7 M CsCl cushion. Samples were centrifuged for 16 h at 22 °C at 50,000 rpm in a TLS55 swinging bucket rotor in a Beckman Optima-TL centrifuge. Supernatant fluids were removed with an aspirator, and the pellet was dehydrated by the addition of ice cold 70% EtOH. The pellets were transferred to clean microcentrifuge tubes and centrifuged in a microcentrifuge for 5 min at 4 °C. The supernatants were removed, and the pellets were resuspended in TE (10 mM Tris-Cl, 1 mµ EDTA), phenol:chloroform-extracted, ethanol-precipitated, and resuspended in H₂O. RNA was quantified by measuring absorbance readings at 260 nm.

Reverse transcription was performed by heating 4 µg of total RNA at 65 °C for 3 min and then incubating in 1 × RT buffer (Boehringer Mannheim Biochemical Products), 0.5 mM dNTP, 18 units of RNAguard (Pharmacia Biotech Inc.), 0.5 mg of random hexamer primer (Pharmacia), 12.5 units of avian myeloblastosis virus reverse transcriptase (Boehringer Mannheim Biochemical Products) in a total volume of 20 µl at 22 °C for 10 min, followed by 42 °C for 90 min.

Amplification of cDNA was performed by PCR as follows: 10 mM Tris-Cl (pH 8.3), 50 mM KCl, 1.5 mM MgCl₂, 0.001% gelatin, 2 µl of reverse transcription reaction products, 100 ng each of oligonucleotide primers that bind up- and downstream, respectively, within the ribosomal protein S16 sequences (5-TCCAAGGGTCCGCTGCAGTC-3 and 5-CGTTCACCTTGATGAGCCCATT-3), 100 ng each of primers that bind up- and downstream, respectively, within the heat-inducible hsp72 sequences (5-ATCACCATCACCAACGACAAGG-3 and 5-TGCCCAAGCAGCTATCAAGTGC-3), 0.3 µl of 3000 Ci/mmol [-³²P]dCTP (DuPont NEN), and 1.5 units of Taq polymerase (Perkin-Elmer Corp.) in a total volume of 50 µl that was overlaid with mineral oil. PCR parameters were 95 °C for 1 min, 65 °C for 1 min, 72 °C for 2 min for 21-26 cycles. Measures were taken to ensure PCR protocol was in the linear range of amplification. PCR products were separated on a 5% polyacrylamide gel and quantified using a Molecular Dynamics PhosphorImager.

RESULTS

In order to characterize tissue-dependent differences in hypothermia-induced cellular stress response, we first examined the levels of HSF DNA binding activity that are induced in various mouse tissues following hypothermia. Tissues were taken from mice subjected to hypothermic treatment for 8 h, with or without recovery at room temperature (22-24 °C) for 1 h, and then extracts of these tissues were subjected to gel shift analysis using an HSE-containing oligonucleotide. As shown in Fig. 1, the level of DNA binding activity was low in extracts of tissues from control animals (lanes 1, 4, 7, and 10). However, in all tissues examined, an increase in DNA binding activity was observed following hypothermic treatment without recovery (lanes 2, 5, 8, and 11). Fig. 1 also shows that for the kidney, liver, and lung, allowing the mice to recover at room temperature for 1 h after the 8 h of hypothermia did not significantly alter the levels of HSF DNA binding from that observed for these tissues in mice subjected to hypothermia alone (compare lanes 5 and 6, 8 and 9, and 11 and 12). However, heart consistently showed a significant increase in HSF DNA binding following recovery at room temperature after the hypothermia treatment, relative to the levels induced by hypothermia alone (compare lanes 2 and 3). In some experiments it was observed that the DNA binding activity of lung extracts was also further stimulated following a recovery period.

Two heat shock transcription factors have been identified in mouse (9). HSF1 mediates induction of the stress response following exposure to stressful conditions (8), while HSF2 appears to be involved in the developmental regulation of hsp expression (10-12). We next sought to determine which HSF was responsible for the cold induced HSE binding activity by performing gel shift analysis in conjunction with polyclonal antibodies specific for HSF1 and HSF2. As shown in Fig. 2, hypothermia-induced HSE binding was shown to be due to HSF1 by the elimination of the HSF·HSE complex following preincubation of the extracts from hypothermic/recovered tissues with polyclonal antibody specific to HSF1 in heart, kidney, and lung (Fig. 2). Interestingly, the DNA binding activity in liver was decreased by pre-incubation in the presence of antibodies specific to both HSF1 and HSF2, suggesting that the induction of the cellular stress response in this tissue may occur by more than one mechanism. DNA binding activity of extracts prepared from non-recovered tissues showed a similar pattern (data not shown).

Previous studies have shown that the activation of HSF1 DNA binding in heat treated cells is coupled with an increase in phosphorylation of the HSF1 protein in these cells, which results in a decreased mobility of this protein on SDS-polyacrylamide gel electrophoresis gels (8, 13). Therefore, in order to determine whether HSF1 undergoes similar phosphorylation events in response to cold treatment, tissue extracts of mice subjected to hypothermic treatment, with or without recovery for 1 h, were subjected to Western blot analysis using polyclonal antibodies specific for mouse HSF1. The HSF1-specific antibody recognized a doublet of HSF1 polypeptides, consisting of the previously described HSF1 - and -isoforms (14), in each of the tissues from control, hypothermic, and hypothermic/recovered mice (Fig. 3). However, this analysis revealed no apparent change in mobility of HSF1 due to any phosphorylation events in response to hypothermic treatment, in contrast to that observed for the HSF1 protein in heat-treated NIH3T3 cells (Fig. 3, lane 14). These results suggest that while hypothermic treatment is able to induce HSF1 DNA binding, it is not able to cause inducible phosphorylation of the HSF1 protein as is observed under heat shock conditions.

We next wanted to determine whether the DNA binding activity of HSF1 induced in tissues by hypothermic treatment was able to mediate a productive stress response. Therefore, we examined the levels of heat-inducible hsp72 mRNA present in tissues of control mice and mice subjected to hypothermic treatment, with or without recovery. For this analysis, we employed RT-PCR (reverse transcription coupled with the polymerase chain reaction) using primer pairs specific for hsp72 mRNA (15) as well as the mRNA for the S16 ribosomal protein (16) as an internal control for efficiencies of reverse transcription and PCR amplification between samples. Fig. 4 shows that while the levels of S16 internal control were not significantly altered by hypothermic treatment, the levels of hsp72 mRNA were increased by cold treatment in all tissues examined. In order to determine the magnitude of the increase in hsp72 mRNA levels in each tissue caused by hypothermia, the results of the RT-PCR analysis were quantified using a PhosphorImager. The ratios of the values above background for hsp72/S16 products in each lane were compared, and the -fold induction of hsp72 mRNA levels following hypothermic or hypothermic/recovered treatment with respect to the control mice were plotted. The results of this analysis, shown in Fig. 4B, demonstrate that there are tissue-dependent differences in induction of hsp72 mRNA levels following hypothermia, ranging from approximately 2-fold induction in kidney and lung to 4-fold in heart and liver. In addition, in all tissues tested, recovery of the mice for 1 h at room temperature after hypothermia caused a further increase in hsp72 mRNA levels, resulting in overall increases in hsp mRNA levels that ranged from approximately 4-fold in kidney to nearly 30-fold in liver.

Analyses of stress response parameters were also performed on tissues from animals that had been cold treated for different lengths of times ranging from 1 to 48 h. Maximal levels of HSF1 activation with respect to DNA binding were observed at 8 and 48 h, with a decrease in activity observed during the intervening time points (data not shown). In no instance was phosphorylation of HSF1 observed. RT-PCR analysis of hsp72 mRNA levels was performed using RNA prepared from the lungs of 48-h cold-treated and cold-treated/recovered mice (Fig. 4A, lanes 13 and 14). The levels of hsp72 mRNA in the lungs of the 48-h cold-treated mice were similar to those of 8-h cold-treated mice, with higher levels of induction observed following a 1-h recovery period (Fig. 4B).

We next sought to examine the stress response pathway in the brain following hypothermia because of this tissue's high susceptibility to damage following insults such as ischemia and its ability to be protected from neurological damage by moderate hypothermia (17-19). Gel mobility shift analysis performed using whole brain tissue extracts demonstrated constitutive HSE binding activity with no additional induction following hypothermia or hypothermia followed by recovery (Fig. 5A). Although the brain initiates the stress response following as little as 5 min of ischemia (20, 21), it is not likely that this is the explanation for the high levels of HSF1 DNA binding even in the cage control as we removed the brains in less than 45 s following sacrifice. Interestingly, this HSE binding activity is due to both HSF1 and HSF2, as preincubation of the extracts with antibodies directed against both HSFs diminished the HSF·HSE complex signal (Fig. 5B, lanes 1-3). This mixed HSF1/HSF2 HSE binding activity was also characteristic of the cold-treated/recovered extracts (Fig. 5B, lanes 4-6). The contribution of HSF2 to the HSE binding activity was not surprising in light of previous studies, which have shown that HSF2 mRNA levels are higher in the brain than in the kidney, liver, and lung (22). Western blot analysis showed no change in the phosphorylation state of HSF1 in the brain as a result of hypothermia (data not shown). However, the levels of hsp72 mRNA are increased 2- and 5-fold in response to hypothermia and hypothermia/recovery, respectively (Fig. 5C). This suggests that a novel mechanism must be responsible for cold induced hsp72 expression within the brain, as it appears that HSF1 is already capable of binding to DNA. Perhaps this tissue requires a more immediate response to stressful conditions, and maintaining constitutive HSF1 DNA binding activity allows it to be primed for the induction of hsp72 expression.

DISCUSSION

We have shown that whole body hypothermia is capable of inducing the stress response in cells of a large number of mouse tissues. HSF DNA binding activity was induced in the heart, kidney, liver, and lung following cold treatment of mice. This activity was shown to be due to HSF1, which also mediates heat-induced expression of hsps. Interestingly, hypothermia treatment did not result in phosphorylation of HSF1, which has been shown to occur in heat-treated cells. Cold incubation increased levels of hsp72 mRNA at least 2-4-fold in all tissues tested. Furthermore, following recovery for 1 h at normal body temperature, a large increase in hsp72 mRNA synthesis occurred ranging from 4- to 29-fold higher than in control mice. Though HSF DNA binding activity was constitutively present in the brain, hypothermic treatment induced hsp72 expression in this tissue as well.

Our findings support a two-step mechanism for the activation of the stress response, which suggests that induction of HSF1 DNA binding activity precedes inducible phosphorylation of HSF1 and that these two steps are not necessarily coupled. Our data show that hypothermia did not induce phosphorylation though HSF1 clearly was capable of DNA binding. This was true regardless of the length of time the animals were exposed to hypothermia (data not shown). Thus, HSF1 DNA binding and phosphorylation are not tightly coupled. Furthermore, analysis of stress response parameters in the brain tissue demonstrated that HSF DNA binding activity alone is not sufficient for the induction of hsp expression, as this tissue did not exhibit an increased level of DNA binding following hypothermic treatment, but hsp levels were increased up to 7-fold.

Another group has examined the effect of hypothermia on the heat shock response in mice in vivo (7). They observed an induction of hsps in brown adipose tissue following cold exposure. However, they did not detect induction of the stress response in many other tissues including the brain, heart, and lung following hypothermia. In contrast, we detected induction of the heat-inducible hsp72 mRNA in the brain, heart, and lung following cold exposure. Furthermore, their study indicated that recovery was not required to achieve induction in brown adipose tissue, but we observed a major additional induction following a recovery period in brain, heart, kidney, liver, and lung. In support of our findings, Liu et al. (5) have demonstrated that cold treatment induces hsp gene expression at the transcriptional level in HeLa and human diploid lung fibroblast cells in vitro. The reason for the discrepancy between the data presented in this paper and that published by Matz et al. (7) is unknown. Perhaps the levels of hsp mRNA they detected would have been higher in the heart and lung if they had permitted a recovery period. One difference is that we utilized a more sensitive means of quantifying hsp mRNA levels; their study used Northern blot analysis while our study employed RT-PCR analysis. In agreement with their studies, we also observed a steady increase in the efficacy of the stress response for up to 8 h of cold exposure, followed by a decline, and another increase at 48 h (data not shown).

The data presented here demonstrate that a similar pathway for induction of the cellular stress response is utilized under both cold stress and heat stress conditions and that at least one major protein that is heat inducible is also cold inducible (hsp72). In addition, it is interesting to note that in vivo the degree of induction of hsp mRNA levels caused by cold treatment followed by recovery is similar to, and for some tissues even greater than, that observed by heat treatment.² This occurs even though the phosphorylation of HSF1 following heat shock, which has been suggested to be necessary for maximal transcription stimulation, does not occur following cold treatment and recovery. Thus, something else yet unidentified must also contribute to achieve the final result of increased hsp72 expression. In addition, something must be occurring during the recovery period that allows for a significant increase in the synthesis of hsp72 mRNA. One possibility is that the ability of the activated HSF1 trimer to be transported into the nucleus is inhibited by cold temperatures and is recovered after the animal returns to normal core body temperatures. Thus, DNA binding activity would be observed in cellular extracts, but the transcription factor would not be able to reach its destination at the promoter of heat shock genes until nuclear transport was restored. Other cellular processes besides nuclear transport are also inhibited by cold temperatures. This brings up the possibility that transcription itself is inhibited by the cold until recovery is permitted. However, this cannot explain the significant induction observed in the recovered samples, since the S16 ribosomal mRNA levels of the non-recovered samples are maintained at levels approximately equal to that of the cage control samples. One might argue that this is simply due to a high stability of the S16 RNA, and thus you might not expect changes in mRNA levels during the 8 h of cold treatment. However, there was no change in the S16 mRNA levels even after 48 h, while there were major increases in the hsp72 mRNA levels (see Fig. 4, A (lanes 13 and 14) and B).

The need for a protective response following cold stress is logical. First, cold treatment itself causes protein denaturation, and hsps would be required to assist in refolding these proteins. Second, metabolism in general slows during cold exposure, but as the animal recovers its cells will start producing new proteins, and thus additional hsps would be required. The cytoprotective effect of hypothermia is currently used to improve the outcome of many medical procedures such as organ transplantation, treatment of heart disease, and aneurysm repair, but the mechanism by which hypothermia protects the body's tissues is not well understood. Our results suggest a mechanism that may explain these beneficial effects of hypothermia. Several pharmacological agents have been identified that induce the heat shock response (23-27), and several reagents have been identified which lower body temperature, such as lipopolysaccharide (28), cyclooxygenase inhibitors (28), and 8-hydroxy-2-(di-n-propyl)aminotetralin, a serotonin 1A receptor agonist (29). Based on our findings, these types of pharmacological agents may provide useful tools for medical applications.