Introduction

Climate variation is one of the main evolutionary forces that cause a particular phenotype within a population to have a better chance of surviving and reproducing. Due to their evolutionary and ecological importance, physiological mechanisms in various organisms that promote adaptation to specific environmental conditions, particularly elevated temperatures, have been intensively investigated1,2,3,4. Many studies examined the effects of elevated temperatures on the physiology and survival of organisms using acute or chronic exposure to a fixed elevated temperature. However, most organisms live in thermally fluctuating environments5,6, and temperature fluctuations are theoretically predicted to exert different effects on organisms than stable elevated temperatures7,8. Indeed, daily temperature fluctuations have been shown to enhance the rates of development and survival in insects at cold mean temperatures, but reduce these rates at warm mean temperatures9,10,11,12.

Temperature fluctuations also markedly affect the performance of vertebrate ectotherms, including fishes, amphibians, and reptiles, the body heat of which is derived exclusively from the external environment13,14. Previous studies on fishes, such as salmon and trout, indicated that there were no significant differences in growth rates between constant and fluctuating temperature treatments utilizing cold mean temperatures, whereas an increase in the magnitude of daily temperature fluctuations around warm mean temperatures had a negative impact on growth rates15,16,17,18. This growth retardation is associated with cellular and endocrine stress responses18. On the other hand, endotherms, including mammals and birds, generate heat and regulate their body temperatures around specific set points. Nevertheless, the body temperature of most endotherms is not completely uniform13,14; it is warmest at the core but may be markedly lower in the extremities such as the skin, the temperature of which varies with ambient temperatures. Moreover, the core body temperature is markedly reduced under low ambient temperatures in some mammal and bird species, suggesting daily fluctuation in the wild19,20,21. Therefore, it would be important to understand effects of daily temperature fluctuations as well as acute and chronic fixed temperatures on the survival of cells from the endotherms2,22.

The heat shock response (HSR) is a universal mechanism of cellular adaptation to elevated temperatures, and is characterized by the induction of heat shock proteins (HSPs) or chaperones that facilitate protein folding and prevent the formation of aggregates and amyloid fibrils within cells23,24,25. Classical experiments have shown that cells pretreated with moderate elevated temperatures acquire resistance to more extreme, short-term lethal temperatures26. The survival of these cells is assessed by their proliferative capacity or plating efficiency measured by their colony-forming ability. This phenomenon is called induced thermotolerance, and its acquisition correlates with the synthesis of HSPs27,28,29. Previous studies demonstrated that the blockade of protein synthesis suppressed the acquisition of induced thermotolerance under some elevated temperature conditions, but not in other conditions30,31,32,33. Furthermore, yeast cells lacking HSP70 or HSP90 proteins still acquired induced thermotolerance, while their proliferation was retarded under long-term sustained high temperatures34,35. Therefore, cellular resistance to elevated temperatures is mediated not only by the synthesis of HSPs, but also by other unknown mechanisms.

HSR is mainly regulated at the level of transcription by evolutionarily conserved heat shock transcription factors (HSFs) that bind to heat shock response elements (HSEs) in eukaryotes36,37. A single HSF gene has been identified in yeast, worm, and fruit fly, while multiple HSF-related genes are present in vertebrates38,39,40. Among the four HSF family members (HSF1–HSF4), HSF1 and HSF3 (hereinafter referred to as HSF1/3) both induced the HSP expression during heat shock in ectotherms, such as the brown anole lizard and Western clawed frog41. On the other hand, HSF3, but not HSF1, induced the HSP expression in chicken cells42, while only HSF1 induced its expression in mouse and human cells43. Furthermore, mouse HSF1 and chicken HSF3 were required for the acquisition of maximal induced thermotolerance42,44,45. The biochemical characterization of HSF1 and HSF3 revealed that they mostly remained as an inert monomer and inert dimer, respectively, and both were converted to active DNA-binding trimers that enhanced HSP transcription upon heat shock41,46. However, their functional diversification during vertebrate evolution and their involvement in adaptation to daily temperature fluctuations currently remain unclear45.

To identify the missing link in the evolution of vertebrate HSF1/3 genes, we herein isolated their complementary DNAs from monotremes, a group of egg-laying mammals, and some birds based on their whole genomic sequences, which were recently released47,48, and examined their roles in the HSR as well as adaptation to daily temperature fluctuations. Based on previous findings and the present results, the potential of HSF1/3 proteins to maximally induce the HSR appears to be related to the homeothermic capacity of vertebrates. Furthermore, HSF1 was required for the proliferation and survival of human cells under daily temperature fluctuations. The ectopic expression of vertebrate HSF1 proteins, but not HSF3 proteins, restored the resistance to daily temperature fluctuations regardless of the induction of HSPs.

Results

Potential of vertebrate HSF1/3 to maximally induce HSP70 expression is closely related to homeothermic capacity

We hypothesized that HSF1 and HSF3 both possess the potential to induce the HSP expression even in some endotherms, which have a weak ability to maintain their body temperatures within a specific range. Monotremes, including the platypus and echidna, are egg-laying mammals that are characterized by a low body temperature (~ 30 °C) and metabolic rate49. They become hypothermic under cold ambient temperatures19,50. We isolated cDNA clones of platypus and echidna HSF1/3, which have important functional domains including the DNA-binding domain, HR-A/B, and HR-C41,45 (Fig. 1A, Supplementary Fig. S1). Phylogenetic trees showed that platypus and echidna HSF1 proteins were closely related to those in other mammalian species, while platypus and echidna HSF3 proteins were distantly related to the mammalian orthologs (Fig. 1B). We overexpressed HSF proteins in HSF1-null mouse embryonic fibroblasts (MEFs) and estimated the levels of endogenous HSP70 (HSPA1A/HSPA1B) protein during heat shock41. We found that the expression of HSP70 was maximally induced during heat shock in cells expressing platypus HSF1 (plaHSF1) and plaHSF3, like in wild-type cells (Fig. 1C, Supplementary Fig. S5A). The same result was obtained for cells expressing echidna HSF1 (echHSF1) and echHSF3. Thus, monotreme HSF1 and HSF3 possess the potential to maximally induce the HSP70 expression, like those in lizards and frogs41 (++, > 70%, Table 1). Interestingly, the basal expression of HSP70 was high in cells expressing plaHSF3 and echHSF3, suggesting that the HSF3 activity is under weak negative regulation in unstressed conditions41.

Fig. 1
figure 1

Potential of vertebrate HSF1/3 to maximally induce HSP70 expression is closely related to homeothermic capacity. (A) The percent identity between cHSF1 or cHSF3 and its vertebrate orthologs was established. The number of amino acids of each HSF is shown. DBD DNA-binding domain, HR hydrophobic heptad repeat, DHR downstream of HR-C, Region X a region between the HR-A/B and HR-C domains, Region Y a C-terminal region downstream of the DHR domain45. The red box in mHSF3 indicates an HR-C-like domain. The transcriptional activation domains (ADs) of human HSF1 (amino acids 372–529), mouse HSF1(amino acids 395–525), chicken HSF1 (amino acids 313–491), and chicken HSF3 (amino acids 261–313) are indicated as red bars. We herein refer to the C-terminal regions containing the Y region and DHR in HSF1 proteins as “core ADs”. The C-terminal region of platypus HSF3 has the potential to activate GAL4-driven luciferase reporter gene (see Supplementary Fig. S4). (B) Phylogenetic trees of HSF1 and HSF3 proteins in vertebrates. The phylogenetic tree was generated in CLUSTAL W79. Gaps were excluded from all phylogenetic analyses. The unrooted tree was drawn using the program TREEVIEW80. Bootstraps of 1000 replicates were performed. The bar represents 0.1 substitutions per site. The amino acid sequences used in tree construction are described in the “Methods” section. HSF proteins in monotremes and aves are indicated in red and blue, respectively. (C) Potential of monotreme HSF1/3 to induce HSP70 expression. HSF1-null MEF cells or wild-type cells (WT, lanes 1 and 2) were infected with adenovirus expressing each HSF protein or adenovirus expressing GFP as the control. These cells were untreated (C) or treated with heat shock at 42 °C for 1 h and then allowed to recover for 3 h (HS). Cell extracts were prepared, and aliquots were subjected to Western blotting using anti-HSP70 (W27), anti-HSF1 (anti-cHSF1x plus anti-mHSF1n), anti-HSF3 (anti-XtHSF3-2 or anti-mHSF3-1), anti-GFP, or anti-β-actin antibody. Human HSF1 (hHSF1), but not mouse HSF3 (mHSF3), showed the strong potential to induce HSP70 expression41. Arrows indicate major bands of HSF1/3. Asterisks indicate non-specific bands. Original blots are presented in Supplementary Fig. S5A. (D) Potential of avian HSF1/3 to induce HSP70 expression. HSF1-null cells or wild-type cells (WT) were infected with each adenovirus, and were treated as described in (B). Cell extracts were prepared, and aliquots were subjected to Western blotting using anti-HSP70 (W27), anti-HSF1 (anti-cHSF1x plus anti-mHSF1n), or anti-HSF3 (anti-XtHSF3-2) antibody. Arrows indicate major bands of HSF1 and HSF3 proteins. Asterisks indicate non-specific bands. Original blots are presented in Supplementary Fig. S5B.

Full size image
Table 1 Potential of vertebrate HSF1 and HSF3 to induce HSP70 upon heat shock.
Full size table

In consideration of marked daily fluctuations in body temperatures caused by high behavioral activity requirements (e.g. flying) of birds with high body temperatures51, we also characterized mallard and zebra finch HSF1/3, which are distantly related to chicken HSF1/3 among those in all bird species47 (Fig. 1B). We found that the expression of HSP70 was maximally induced during heat shock in cells expressing mallard HSF1 (malHSF1), but the expression levels were low in cells expressing chicken HSF1 (cHSF1) and zebra finch HSF1 (zefHSF1) (Fig. 1D, Supplementary Fig. S5B). On the other hand, the expression of HSP70 was maximally induced in cells expressing cHSF3 and zefHSF3, and the expression levels were very low in cells expressing malHSF3. These results suggested that either HSF1 or HSF3 proteins possess the potential to induce the maximal HSP70 expression in birds, unlike in lizards, frogs, or egg-laying mammals with a low homeothermic capacity. Collectively, these results supported our hypothesis that the potential of vertebrate HSF1/3 to maximally induce the HSR is closely related to homeothermic capacity.

Human cells deficient in HSF1 are sensitive to daily temperature fluctuations

To investigate whether HSF1/3 proteins play a role in cellular adaptation to daily temperature fluctuations in endotherms, we used immortalized human OUMS-36T-3F fibroblasts possessing only HSF1. The core body temperature is generally accepted as 37 °C (36.1–37.2 °C) and never exceeds 41.0 °C in humans; however, body temperatures in the extremities increases or decreases with ambient temperatures52,53. Accordingly, the proliferation rates of OUMS-36T-3F cells were higher at approximately 37 °C (35–38 °C) and the critical maximal temperature was approximately 40 °C in wild-type cells (see Supplementary Fig. S2D). We examined the expression of HSPs when cells were periodically incubated at a cold temperature of 30 °C for 14 h (18:00–8:00) (Fig. 2A). We found that HSP110, HSP70, and HSP40 protein levels reduced at a cold temperature and mildly increased at 37 °C in the first and third cycles (Fig. 2B, Supplementary Figs. S2A, S5C). Oscillations in HSP protein levels were associated with changes in their mRNA levels (Fig. 2C). We generated HSF1-null (KO) cells (Supplementary Figs. S2B–D, S5C,D) and found that HSP110, HSP90, HSP70, HSP40, and HSP27 protein levels were markedly reduced in HSF1 KO cells under unstressed conditions and were not up-regulated by temperature fluctuations (Fig. 2D, Supplementary Fig. S5E). Thus, HSF1 was responsible not only for basal HSP expression, but also for oscillations in HSP expression under daily temperature fluctuations.

Fig. 2
figure 2

Human cells deficient for HSF1 are sensitive to daily temperature fluctuations. (A) Schematic representation of the experimental design. Human OUMS-36T-3F cells were incubated every day at 37 °C from 8:00 to 18:00 (10 h) and then at 30 °C from 18:00 to 8:00 (14 h). Cells were harvested to investigate the expression of HSPs at the indicated time points (arrowheads, numbers 1–6). (B,C) Expression of HSPs during daily temperature fluctuations. Extracts of cells incubated as shown in (A), as well as unstressed cells (Cont.), were prepared and subjected to Western blotting. Protein levels were quantified and their levels relative to those in control cells are shown (B). Levels of mRNAs were quantified by RT-qPCR, and the levels relative to those in control cells are shown (C). Data are shown as mean ± SD (n = 3). (D) Expression of HSPs in HSF1-null (KO) cells during daily temperature fluctuations. Extracts of wild-type and HSF1 KO (KO1 and KO2) cells incubated as shown in (A) were prepared and subjected to Western blotting (upper). Protein levels were quantified and normalized to those of actin loading controls. Their levels relative to those in unstressed wild-type cells (fold changes) are shown (lower). Original blots are presented in Supplementary Fig. S5C. (E,F) Cell proliferation under conditions of daily temperature fluctuations. Wild-type (WT) and HSF1 KO cells were grown for 14 days under conditions of daily temperature fluctuations between a cold (30 °C) and the indicated warm (37–40 °C) temperatures. The number of viable cells excluding trypan blue was counted, and fold changes are shown (E). Doubling time ratios of gene-disrupted cells to that of wild-type cells are shown (F). Data are shown as mean ± SD (n = 3). Asterisks indicate ***P < 0.001 by two-way ANOVA. (G) Colony formation of cells exposed to daily temperature fluctuations. Cells, which were grown for 14 days under the conditions as described in (E), were re-plated in 60 mm culture dishes for 10 days and then stained with Giemsa stain solution. Representative photos of colonies under daily temperature fluctuation (37–30°C and 39.5–30 °C) are shown (left). The numbers of colonies were counted, and the percentages of colony numbers originating from single cells (plating efficiency) are shown (right). Data are shown as mean ± SD (n = 3). Asterisks indicate ***P < 0.01 by a two-way ANOVA.

Full size image

We then examined the proliferation and survival of cells exposed for 14 days to daily temperature fluctuations between warm (37–39.5 °C) and cold (30 °C) temperatures. The proliferation rate of wild-type cells slightly decreased at a warm temperature of 39 °C, but was markedly higher than that of HSF1 KO cells (Fig. 2E,F). The survival rate of wild-type cells, estimated as plating efficiency measured by colony-forming ability, did not decrease under the same condition, whereas a slight reduction was noted in that of HSF1 KO cells (Fig. 2G). Furthermore, the survival of HSF1 KO cells decreased much more under warm temperatures of 39.5 °C and 40 °C. These results indicated the requirement of HSF1 for cell proliferation and survival under daily temperature fluctuations.

Since the proliferation of yeast cells with mutations in HSP70 or HSP90 was retarded at chronically elevated temperatures34,35, we tested whether proliferation of HSF1 KO cells expressing low levels of HSPs decreases. Indeed, HSF1 was required for the proliferation of OUMS-36T-3F cells under a chronically high temperature of 39.5 °C, but not 39 °C (Supplementary Fig. S2E). Moreover, consistent with previous findings from mouse cells43,44, HSF1 was also required for the acquisition of induced thermotolerance (Supplementary Fig. S2F).

Vertebrate HSF1 proteins confer resistance to daily temperature fluctuations

To investigate whether the sensitivity of human HSF1 KO cells to daily temperature fluctuations is rescued by the expression of any one of vertebrate HSF1 orthologs, we initially generated HSF1 KO cell lines stably expressing one of the orthologs (Fig. 3A, Supplementary Fig. S5F). We found that the basal and heat-induced expression levels of HSPs (HSP110, HSP90, HSP70, HSP40, and HSP27) were all restored in hHSF1-, malHSF1-, and plaHSF1-expressing cells (Fig. 3B, Supplementary Fig. S5G). In contrast, the basal and heat-induced expression levels of HSP110 and HSP40 were only slightly restored in cHSF1-expressing cells, while those of the other HSPs were not54. In either case, the proliferation rates of these cells were as high as that of wild-type cells under normal conditions (37 °C) and with daily temperature fluctuations (39.5–30 °C) (Fig. 3C–E). Furthermore, the survival of all these cells was similar to that of wild-type cells under the same conditions (Fig. 3F). Thus, vertebrate HSF1 proteins confer resistance to daily temperature fluctuations.

Fig. 3
figure 3

Vertebrate HSF1 proteins confer resistance to daily temperature fluctuations. (A) Generation of human HSF1 KO cells expressing vertebrate HSF1 proteins. Cell extracts were prepared from wild-type (WT) cells, HSF1 KO OUMS-36T-3F (KO1) cells, and HSF1 KO cells stably expressing hHSF1, cHSF1, malHSF1, or plaHSF1, and aliquots were subjected to Western blotting using anti-HSF1 (anti-mHSF1s) or anti-β-actin antibody. Original blots are presented in Supplementary Fig. S5D. (B) Expression of HSPs in HSF1 KO cells expressing vertebrate HSF1 proteins. Cell lines described in (A) were untreated (C) or treated with heat shock at 42 °C for 6 h (H). Extracts of these cells were prepared and subjected to Western blotting. Two isoforms of HSP110 are indicated by arrowheads. Original blots are presented in Supplementary Fig. S5D. (C) Expression of vertebrate HSF1 proteins restored the proliferation rate of HSF1 KO cells. Cell lines described in (A) were grown at 37 °C for 3 days. The number of viable cells excluding trypan blue was counted, and fold changes are shown. Data are shown as mean ± SD (n = 3). Asterisks indicate *P < 0.05 by a one-way ANOVA, followed by the Tukey–Kramer test. (D,E) Proliferation of the cell lines under daily temperature fluctuations. Cell lines described in (A) were grown for 14 days under daily temperature fluctuations between cold (30 °C) and warm (39.5 °C) temperatures. The number of viable cells was counted, and fold changes are shown (D). The ratios of doubling time of each cell line to that of wild-type cells are also shown (E). Data are shown as mean ± SD (n = 3). Asterisks indicate ***P < 0.001 by a two-way ANOVA (D) or by a one-way ANOVA, followed by the Tukey–Kramer test (E). (F) Colony formation of the cell lines exposed to daily temperature fluctuations. The cells, which were grown for 14 days under the conditions as described in (D), were re-plated in 60-mm culture dishes for 10 days. The number of colonies were counted, and the percentages of colonies originating from single cells (plating efficiency) is shown. Data are shown as mean ± SD (n = 3). Asterisks indicate ***P < 0.001 by one-way ANOVA, followed by the Tukey–Kramer test.

Full size image

Vertebrate HSF3 proteins play a partial role in resistance to daily temperature fluctuations

We then investigated the effects of the expression of vertebrate HSF3 orthologs, except malHSF3, in human HSF1 KO cells using the same approach (Fig. 4A, Supplementary Fig. S5H). The basal and heat-induced expression levels of HSPs were mostly restored in cells expressing any HSF3 protein41 (Fig. 4B, Supplementary Fig. S5I). Nevertheless, the proliferation rates of these cells were markedly lower than that of wild-type cells under daily temperature fluctuations (39.5–30 °C), and slightly higher than that of HSF1 KO cells (Fig. 4C–E). Moreover, the survival rates of these cells were as low as that of HSF1 KO cells under the same conditions (Fig. 4F). These results suggested that vertebrate HSF3 proteins and the induction of HSPs play a partial role in resistance to daily temperature fluctuations.

Fig. 4
figure 4

Vertebrate HSF3 proteins play a partial role in resistance to daily temperature fluctuations. (A) Generation of HSF1-null cells expressing vertebrate HSF3 proteins. Cell extracts were prepared from wild-type (WT) cells, HSF1 KO (KO1) cells, and HSF1 KO cells stably expressing cHSF3, zefHSF3, plaHSF3, or echHSF3, and aliquots were subjected to Western blotting using anti-HSF3 (anti-XtHSF3-2), anti-HSF1 (anti-mHSF1s), or anti-β-actin antibody. Original blots are presented in Supplementary Fig. S5E. (B) Expression of HSPs in HSF1-null cells expressing vertebrate HSF3 proteins. The cell lines described in (A) were untreated (C) or treated with heat shock at 42 °C for 6 h (H). Extracts of these cells were prepared and subjected to Western blotting. Original blots are presented in Supplementary Fig. S5F. (C) Expression of vertebrate HSF3 proteins did not restore the proliferation rate in HSF1-null cells. Cell lines described in (A) were grown at 37 °C for 3 days. The number of viable cells was counted, and fold changes are shown. Data are shown as mean ± SD (n = 3). A n.s. mark indicates not significant (P > 0.05). (D,E) Proliferation of cell lines under daily temperature fluctuations. Cell lines described in (A) were grown for 14 days under daily temperature fluctuations. The number of viable cells was counted, and fold changes are shown (D). The ratios of doubling time of each cell line to that of wild-type cells are shown (E). Data are shown as mean ± SD (n = 3). Asterisks indicate **P < 0.01 and ***P < 0.001 by a two-way ANOVA (D) or by a one-way ANOVA, followed by the Tukey–Kramer test (E). (F) Colony formation of the cell lines exposed to daily temperature fluctuations. Cell lines, which were grown for 14 days under daily temperature fluctuations, were re-plated in 60-mm culture dishes for 10 days. The numbers of colonies were counted, and the percentages of colonies originating from single cells (plating efficiency) are shown. Data are shown as mean ± SD (n = 3). Asterisks indicate ***P < 0.001 by a one-way ANOVA, followed by the Tukey–Kramer test. (G,H) Potential of C-terminal fragments in hHSF1 and plaHSF3 to activate a reporter gene. Schematic representation of C-terminal fragments fused to GAL4 DNA-binding domain (GAL4 DBD, amino acids 1–147) (G, left). Red bars indicate the ADs of hHSF155 and plaHSF3 (this study). HEK293 cells were transfected with the reporter plasmid ptk-galp3-luc and the expression plasmid for each HA-tagged fragment fused to GAL4 DBD. Cells were divided into two groups, and were subjected to luciferase assay (G, right) and Western blotting using HA antibody (H). Data are shown as mean ± SD (n = 3). Asterisks indicate *P < 0.05 by a one-way ANOVA, followed by the Tukey–Kramer test. Original blots are presented in Supplementary Fig. S5G. (I) Generation of HSF1 KO cells stably expressing hHSF1-pD-pY. Cell extracts were prepared from wild-type (WT) cells, HSF1 KO (KO1) cells, and HSF1 KO cells stably expressing hHSF1-pD-pY, and aliquots were subjected to Western blotting using anti-HA, anti-HSF1 (anti-mHSF1s), or anti-β-actin antibody (right). Original blots are presented in Supplementary Fig. S5H. (J) Expression of HSPs in HSF1 KO cells expressing hHSF1-pD-pY. Cell lines described in (I) were untreated (Cont.) or treated with heat shock at 42 °C for 3 h (HS). Extracts of these cells were prepared and subjected to Western blotting. Original blots are presented in Supplementary Fig. S5H. (K,L) Proliferation of the cell lines under daily temperature fluctuations. Cell lines described in (I) were grown for 14 days under daily temperature fluctuations between cold (30 °C) and warm (39.5 °C) temperatures. The number of viable cells was counted, and fold changes are shown (K). The ratios of doubling time of each cell line to that of wild-type cells are also shown (L). Data are shown as mean ± SD (n = 3). Asterisks indicate *P < 0.05, **P < 0.01, and ***P < 0.001 by a two-way ANOVA (K) or by a one-way ANOVA, followed by the Tukey–Kramer test (L). (M) Colony formation of the cell lines exposed to daily temperature fluctuations. The cells treated as described in (K) were re-plated in 60-mm culture dishes for 10 days. The number of colonies were counted, and the percentages of colonies originating from single cells (plating efficiency) is shown. Data are shown as mean ± SD (n = 3). Asterisks indicate ***P < 0.001 by one-way ANOVA, followed by the Tukey–Kramer test.

Full size image

We further tested whether the transcriptional activation domain (AD) of HSF1 proteins, which is required for activation of HSP genes, can be substituted with that of HSF3 proteins. Vertebrate HSF1 proteins possess ADs in their C-terminal regions that is necessary for activation of HSP genes55,56,57,58,59,60 (Fig. 1A). We herein refer to the region containing the DHR and Y region (denoted by D-Y) as “core AD”. Since the amino acids of the transcriptionally inactive C-terminal region in cHSF3 was not conserved in the corresponding region of other vertebrate HSF3 proteins including plaHSF358 (Supplementary Fig. S1), the potential of the plaHSF3 C-terminal region to activate the GAL4-driven luciferase reporter gene was examined. We found that plaHSF3 D-Y (pD-pY) possessed the potential to activate the reporter gene, like the hHSF1 core AD (hD-hY) (Fig. 4G,H, Supplementary Fig. S5G). The transcriptional activity of plaHSF3 Y (pY) was not affected by the addition of its DHR (pD-pY), but was enhanced by the addition of its HR-C (pCT3). Thus, the core ADs (D-Y regions) of hHSF1 and plaHSF3 both possessed the potential to activate the reporter gene. We then generated HSF1 KO cells stably expressing chimeras between hHSF1 and plaHSF3. We found that hHSF1 possessing the core AD of plaHSF3 (hHSF1-pD-pY) showed the strong potential to induce the expression of HSPs during heat shock (Fig. 4I,J, Supplementary Fig. S5H). Its expression completely restored the proliferation and survival rates of HSF1 KO cells (Fig. 4K–M). Thus, the core AD of plaHSF3 is responsible for the induction of HSP expression and does not negatively affecting the resistance to daily temperature fluctuations. Taken together, the ectopic expression of cHSF1 restored the resistance in HSF1-null cells without inducing HSPs but that of HSF3 proteins possessing the ability to induce HSPs did not restore it. These observations supported our proposal that HSF1-mediated resistance to daily temperature fluctuations is dependent on non-HSP pathways.

It is important to note that the ectopic expression of cHSF3 and plaHSF3 in HSF1 KO cells mostly restored their proliferation at a constant high temperature (39.5 °C), like that of vertebrate HSF1 proteins (Supplementary Fig. S3A). Moreover, the expression of cHSF3 and plaHSF3 as well as any HSF1 protein in HSF1 KO cells enhanced the acquisition of induced thermotolerance (Supplementary Fig. S3B).

Specific up-regulation of a group of genes by vertebrate HSF1 proteins

To search for candidate genes that contribute to the resistance to daily temperature fluctuations, we performed a microarray analysis of wild-type cells and HSF1 KO cells expressing cHSF1 (possessing limited potential to induce the HSR) or cHSF3 under unstressed conditions and compared the results obtained to those from HSF1 KO cells (Fig. 5A, Supplementary Fig. S5I). cHSF3 up-regulated 1976 gens, while cHSF1 and hHSF1 up-regulated 1197 and 1944 genes, respectively (fold change > 1.3) (Fig. 5B, Supplementary Table S1). hHSF1 and cHSF1 commonly up-regulated 192 genes (Fig. 5C,D). Among them, 62 genes were up-regulated by the two HSF1 proteins, but not by cHSF3, suggesting that they are candidate genes for the resistance. A gene set enrichment analysis (GSEA) showed that some gene sets of pathways were enriched in hHSF1- or cHSF1-up-regulated genes, but not in cHSF3-up-regulated genes (Supplementary Fig. S4A). A common gene set enriched in both hHSF1- and cHSF1-up-regulated genes was not identified. We then classified 62 genes according to gene ontology categories and found various categories, including transcriptional regulation (12 genes) and lipid metabolism (7 genes) (Fig. 5E, Supplementary Fig. S4B). Representative genes, such as G0S2, MYD88, ATP8A2, CRABP2, ZNF239, and GCSAML, were up-regulated by hHSF1 and cHSF1, but not by cHSF3, whereas HSP genes were mostly up-regulated by cHSF3 (Fig. 5F, Supplementary Fig. S4C).

Fig. 5
figure 5

Specific up-regulation of a group of genes by vertebrate HSF1 proteins. (A) Human HSF1 KO cells expressing cHSF1 or cHSF3. Cell extracts were prepared from wild-type (WT) cells, HSF1 KO OUMS-36T-3F (KO1) cells, and HSF1 KO cells stably expressing cHSF1 or cHSF3, and aliquots were subjected to Western blotting. Original blots are presented in Supplementary Fig. S5I. (B) Volcano plots of differentially expressed genes. Microarray analysis was performed using total RNA isolated from the cells described in (A). Genes up-regulated and down-regulated by hHSF1 (WT), cHSF1 (KO + cHSF1), or cHSF3 (KO + cHSF3) were identified and indicated as red and green dots, respectively (fold change >  + 1.3 and < − 1.3, P < 0.05; n = 3). (C) Venn diagram showing genes up-regulated by hHSF1, cHSF1, and cHSF3. A total of 130 genes up-regulated by three HSF proteins are indicated in blue, and 62 genes up-regulated by two HSF1 proteins, but not by cHSF3, are indicated in red. (D) Heat map showing 192 genes significantly up-regulated genes by two HSF1 proteins. Representative genes specifically up-regulated by two HSF1 proteins, but not by cHSF3, are indicated by red lines on the right. (E) Classification of 62 genes specifically up-regulated by two HSF1 proteins according to gene ontology categories. Numbers of genes in each category are indicated. (F) Expression levels of representative genes specifically up-regulated by two HSF1 proteins. Means of normalized mRNA signal values (Log 2) in wild-type (WT) cells, HSF1 KO cells, and HSF1 KO cells stably expressing cHSF1 or cHSF3 are shown (n = 3). Each value is indicated as a dot. (G) Validation of G0S2 expression in HSF1 KO cells expressing HSF1 or HSF3 proteins. G0S2 mRNA levels in various cell lines described in Fig. 3 were quantified by RT-qPCR, and their levels relative to that in wild-type (WT) cells are shown. Data are shown as mean ± SD (n = 3). Asterisks indicate *P < 0.05 and **P < 0.01 by a one-way ANOVA, followed by the Tukey–Kramer test. (H–J) Overexpression of G0S2 partially promotes the proliferation of HSF1 KO cells under daily temperature fluctuations. HSF1 KO cells stably overexpressing G0S2 tagged with HA (clone #1 and #2) were generated (H). These cells were grown for 14 days under daily temperature fluctuations (39.5–30 °C). The number of viable cells was counted, and fold changes are shown (I). Doubling time ratios of each cell line to that of wild-type cells are shown (J). Data are shown as mean ± SD (n = 3). Asterisks indicate *P < 0.05, **P < 0.01, and ***P < 0.001 by a two-way ANOVA (I) or by a one-way ANOVA, followed by the Tukey–Kramer test (J). Original blots are presented in Supplementary Fig. S5I.

Full size image

We focused on the lipid regulator G0S2 and confirmed that G0S2 mRNA levels were increased by hHSF1, cHSF1, malHSF1, and plaHSF1, but not by cHSF3 or plaHSF3, under normal conditions61,62 (Fig. 5G). hHSF1, cHSF1, and cHSF3 bound directly to the G0S2 promoter (Supplementary Fig. S4D), while only the HSF1 proteins up-regulated the expression of G0S2 (Supplementary Fig. S4E). Of note, the up-regulated expression of HSP70 at 39.5 °C was attenuated later in wild-type cells, but not in cells expressing cHSF3, suggesting that the stress status persisted in the latter cells. Furthermore, the overexpression of G0S2 partially restored the proliferation rate (doubling time ratio) of HSF1 KO cells under daily temperature fluctuations (Fig. 5H–J, Supplementary Fig. S5I), but it did not affect their survival (plating efficiency) (Supplementary Fig. S4F). These results suggested that a group of genes exhibiting similar expression patterns, including G0S2, may be responsible for resistance to daily temperature fluctuations.

Discussion

HSF1 is a master regulator of the HSR or the transcriptional activation of HSP genes in mammals40,45. Strangely enough, the HSR is mainly regulated by HSF3 in chicken cells and is cooperatively regulated by HSF1/3 in the cells of the lizard Gekko gecko41,42,63. To understand the universal rule regarding the functional diversification of HSF1/3 during vertebrate evolution, we herein focused on homeothermic capacity and examined the roles of HSF1/3 in the HSR in monotremes and some birds. We showed that the platypus and echidna HSF1/3 proteins equally possess a potential to maximally induce the HSR (Fig. 1C). Mallard HSF1 and zebra finch HSF3 also possessed the potential, whereas that of mallard HSF3 and zebra finch HSF1 was markedly weaker (Fig. 1D). Based on previous findings and these results, we concluded that HSF1 and HSF3 both have the potential to maximally induce the HSR in vertebrates with a low homeothermic capacity. This group of vertebrates include not only ectotherms (amphibians and reptiles), but also some endotherms (mammals) with low body temperatures and metabolic rates, such as the monotremes49. Among other endotherms, including some bird species with large body temperature changes20,21,51, at least one out of HSF1/3 may have retained the potential to maximally induce the HSR. Specifically, HSF3 is not required for the HSR and no longer remains intact in humans45.

We wondered whether the induction of HSPs is crucial for adaptation to daily temperature fluctuations around warm mean temperatures. We examined the long-term effects of these fluctuations in human HSF1 KO cells, and found marked reduction in their proliferation and survival rates (Fig. 2E–G). These rates were fully restored by the ectopic expression of vertebrate HSF1 proteins (Fig. 3), but were only slightly restored by that of HSF3 proteins (Fig. 4D–F). Thus, HSF3 proteins and the induction of HSPs play a partial role in resistance to daily temperature fluctuations. On the other hand, we also showed that the ectopic expression of vertebrate HSF3 proteins mostly restored reductions in induced thermotolerance and decreased proliferation rates at high temperatures in HSF1 KO cells (Supplementary Fig. S3A,B). HSF1 was previously shown to be required for the acquisition of induced thermotolerance and for proliferation under chronic heat shock at constant temperatures and the induction of HSPs was crucial for these functions23,37,43,44,54,64. For example, the HSF-mediated up-regulation of HSP90 was required for cell proliferation under chronically high temperature conditions63,65,66. Our observations demonstrated that HSF1, but not HSF3, is required for the resistance to daily temperature fluctuations, and this function is distinguished from the other two functions.

We would like to understand mechanisms by which cells adapt to detrimental daily temperature fluctuations. The physiological effects of elevated temperatures are complex, and cells require diverse mechanisms to cope with any associated detrimental effects23. HSF1 controls various kinds of target genes, which are involved in many cellular processes including mitosis and metabolism67,68,69,70, and protects cells from elevated temperatures by regulating not only HSPs, but also non-HSP genes37,39. For example, cHSF1 protects cells against chronic high temperatures without inducing HSPs54. In Caenorhabditis elegans, an HSF1 mutant lacking the C-terminal AD does not have the potential to induce HSPs, but confers the induced thermotolerance71. Screening and analyses of target genes regulated by both the wild-type and this mutant showed that the HSF1-mediated stabilization of actin filaments is important for induced thermotolerance. We herein took advantage of human HSF1 KO cells ectopically expressing cHSF3 and cHSF1, the latter of which confers resistance to daily temperature fluctuations without inducing HSPs. We searched for transcripts with increased abundance in cells expressing cHSF1, but not cHSF3, among endogenous HSF1 targets. The 62 genes that met these criteria belonged to various categories, including transcriptional regulation, lipid metabolism, cell adhesion, immune response, ion channel, protein transport, cell cycle, and cell proliferation (Fig. 5A–F). They should be the candidate genes that contribute to the resistance to daily temperature fluctuations. To test this hypothesis, we selected the lipid regulator gene G0S261,62, one of the genes showing a representative expression pattern, because temperature fluctuations induce changes in metabolism, which may be associated with changes in cell proliferation and survival72. Indeed, the overexpression of G0S2 partially restored the proliferation of HSF1 KO cells under temperature fluctuations (Fig. 5G–J). G0S2 has been reported to regulate cancer cell proliferation; overexpression of G0S2 suppresses proliferation of some cancer cell lines73, whereas knockdown of G0S2 also reduces proliferation of other cell lines74. Interestingly, the expression of G0S2 changes under acidic pH and endoplasmic stress conditions75,76. The appropriate expression of G0S2 may maintain lipid homeostasis, thereby supporting cell proliferation under stress conditions. It is also possible that G0S2 directly regulates cell cycle and apoptosis irrespectively of stress conditions77,78. We speculated that cellular networks consisting of the 62 gene products including G0S2 may contribute to HSF1-mediated proliferation and survival under daily temperature fluctuations.

In summary, we herein demonstrated that the resistance to daily temperature fluctuations was mediated by HSF1. This function was executed by not only by HSPs, but also probably by non-HSP genes belonging to various categories. Homeothermic capacity in thermally fluctuating environments appears to be associated with the functional diversification of HSF1/3 during vertebrate evolution. Our observations provide novel insights for advances in clinical applications, such as hyperthermia for cancer patients and therapeutic hypothermia.

Materials and methods

Cell culture and colony formation assay

Immortalized wild-type (HSF1+/+, stock #10) or HSF1-null (HSF1−/−, stock #4) MEFs and immortalized human fibroblast OUMS-36T-3F (JCRB1006.3F, Japanese Collection of Research Bioresources Cell Bank) and human HEK293 (ATCC, CRL-1573) cells were maintained at 37 °C in 5% CO2 with high relative humidity in Dulbecco’s modified Eagle’s medium (DMEM) (Sigma-Aldrich) containing 10% fetal bovine serum (FBS) (Sigma-Aldrich)41. To expose cells to heat shock for short periods (1–6 h), culture dishes were submerged in a water bath at 42 °C, and some cells were then allowed to recover in a CO2 incubator at 37 °C. To expose cells to high and low temperatures for long periods (more than 6 h), culture dishes were moved to CO2 incubators at the indicated temperatures. In the colony formation assay, cells exposed to daily changes in temperature were seeded on 60-mm culture dishes for 10 days. These cells were fixed with 4% paraformaldehyde/phosphate-buffered saline (PBS) at room temperature (RT) for 10 min, washed three times with PBS, stained with 2% Giemsa stain solution/PBS for 30 min, and then washed with water. The number of colonies was counted, and the plating efficiency was calculated.

Molecular cloning of vertebrate HSFs

We obtained a partial fragment of mallard (Anas platyrhynchos) HSF3 cDNA based on the database by reverse transcription-polymerase chain reaction (RT-PCR) with KOD One PCR Master Mix (KMM-201, TOYOBO Co., Japan) using total RNA isolated from the brains of the E15.5 embryos and primer sets. The 5′-end fragment of malHSF1 cDNA was amplified using SMARTer RACE 5′/3′ Kit (Cat. # 634858, 634859, TaKaRa Bio Inc., Kusatsu, Japan). Full-length malHSF3 cDNA (DDBJ accession number LC806256; hold date, December 1, 2024) was obtained using the primer sets listed in Supplementary Table S2, and cloned into the pShuttle-CMV vector at the KpnI/XhoI sites. We determined the sequences of three independent malHSF3 cDNA clones (DNA Sequencing Services, Eurofins Genomics K.K., Tokyo, Japan). malHSF1 cDNA was similarly obtained and cloned into the pShuttle-CMV vector at the KpnI/EcoRV sites. Zebra finch (Taeniopygia guttata) HSF1 and HSF3 cDNAs were obtained by RT-PCR using total RNA isolated from the brains of the E14 embryos and primer sets (Supplementary Table S2), and were cloned into the pShuttle-CMV vector. Mallard eggs and zebra finch eggs were purchased from an mallard agency (Kamono-Satokai, Nagahama, Japan) and pet shop company (PetLand Atsumare, Ube, Japan), respectively. cDNA libraries were generated from the brains of platypus (Ornithorhynchus anatinus) and echidna (Tachyglossus aculeatus) (collected under AEC permit S-49-2006)48 using the iScript cDNA Synthesis kit (Bio-Rad). plaHSF1, plaHSF3, echHSF1, and echHSF3 cDNAs were obtained as described above by RT-PCR using these cDNA libraries and primer sets (Supplementary Table S2), and were cloned into the pShuttle-CMV vector. All methods were carried out in accordance with the Committee for Ethics on Animal Experiments of Yamaguchi University Graduate School of Medicine (permit number 06-003), and are reported in accordance with ARRIVE guidelines (https://arriveguidelines.org). We determined their sequences and verified that they were exactly the same as those in the database. The predicted amino acid sequences of HSFs from various vertebrate species were compared using GENETYX-MAC software (GENETYX Co., Tokyo, Japan). The amino acid sequences used to construct phylogenetic trees were those of HSFs from alligator (Alligator mississippiensis, XP_014465302.1 and XP_014462741.1), turtle (Chelonia mydas, XP_037747729.1 and XP_043378861.1), chicken (Gallus gallus, NP_001292185.1 and NP_001291970.1), mallard (XP_027308641.1 and this study), zebra finch (XP_041570061.1and XP_030128356.1), lizard (Anolis sagrei, BAZ95785 and BAZ95787), echidna (Tachyglossus aculeatus, XP_038616553.1 and XP_038603906.1), platypus (Ornithorhynchus anatinus, XP_028919528.1 and XP_028923365.1), armadillo (Dasypus novemcinctus, XP_004472077.1 and XP_023448036.1), guinea pig (Cavia porcellus, XP_005000946.1 and XP_012999964.2), frog (Xenopus laevis, NP_001084036.1 and XP_041428545.1), mouse (Mus musculus, P_001297683.1), and human (Homo sapiens, NP_005517.1).

Adenoviral vectors and infection

To generate expression vectors for chimeras between hHSF1 and plaHSF3, cDNA fragments of chimeras tagged with hemagglutinin (HA) at the C-terminus, were generated by PCR-mediated site-directed mutagenesis and inserted into pShuttle-CMV vectors (Stratagene) at KpnI/XhoI sites81. Viral DNA containing the cDNA for each HSF1, HSF3, or chimera was generated in accordance with the manufacturer’s instructions for an AdEasy adenoviral vector system (Agilent Technologies). These viral DNAs were transfected into HEK293 cells, and virus particles were enriched by CsCl gradient centrifugation and stored at − 80 °C until use. To overexpress each HSF, HSF1−/− MEF cells (stock #4) were infected with an adenovirus expressing each HSF (5 × 107 pfu/ml) for 2 h, and maintained in normal medium for 46 h.

Western blotting

Cells were lysed with NP40 lysis buffer containing 1.0% NP40, 150 mM NaCl, 50 mM Tris–HCl (pH 8.0), and protease inhibitors (1 μg mL−1 leupeptin, 1 μg mL−1 pepstatin, and 1 mM phenylmethylsulfonyl fluoride). After centrifugation, aliquots of the supernatant were subjected to SDS–polyacrylamide gel electrophoresis (SDS-PAGE). After transferring to a nitrocellulose membrane using a Trans-Blot transfer cell (Bio-Rad), the membrane was blocked with 5% milk/PBS at RT for 1 h. Primary antibodies were diluted in 2% milk/PBS and incubated at RT for 1 h or at 4 °C overnight. The following antibodies were used: anti-HSF1 (anti-cHSF1x, Nakai lab, 1/1000; anti-mHSF1n, Nakai lab, 1/000; anti-mHSF1s, Nakai lab, 1/000), anti-HSF3 (anti-XtHSF3-2, Nakai lab, 1/1000)41,82, anti-β-actin (Sigma-Aldrich A5441, 1/1000), anti-GFP (Nacalai Tesque GF200, 1/1000), anti-HA (Roche 3F10, 1/1000), anti-mouse HSP110 (anti-mHSP110b, Nakai lab, 1/1000), anti-HSP70 (Santa Cruz W27, 1/1000), anti-human HSP40 (anti-hHSP40-1, Nakai lab, 1/1000), anti-human HSP27 (anti-hHSP27a, Nakai lab, 1/1000)41,82. The membrane was washed three times with PBS for 5 min each, followed by incubation with horseradish peroxidase-conjugated secondary antibodies (goat anti-rabbit IgG, MP Biomedicals, 55689, 1/2000; goat anti-mouse IgG, Jackson, 115-035-003, 1/1000; goat anti-rat IgG, Jackson, 112-035-003, 1/1000) in 2% milk/PBS at RT for 1 h. The membrane was washed three times with PBS containing 0.1% Tween 20, and chemiluminescent signals from Amersham ECL detection reagents (GE Healthcare) were captured on an X-ray film (Super RX, Fujifilm). Apparent molecular weights on SDS-PAGE gels and Western blots were estimated using Precision Plus Protein Dual Color Standards (Bio-Rad, #1610374).

Assessment of mRNA

Total RNA was extracted from human OUMS-36T-3F cells using Trizol (Invitrogen), and first-strand cDNA was synthesized using PrimeScript II reverse transcriptase and oligo (dT)20 in accordance with the manufacturer’s instructions (TAKARA). Real-time quantitative PCR (qPCR) for human HSP70 (HSPA1A/HSPA1B) mRNA was performed using StepOnePlus (Applied Biosystems) with EagleTaq Master Mix with ROX (Roche) in accordance with the manufacturer’s instructions, and relative quantities were normalized against GAPDH mRNA levels. The qPCR for human HSP110 and other mRNAs was performed with Power SYBR Green PCR Master Mix (Applied Biosystems) and relative quantities were normalized against β-actin mRNA levels. Primers used for RT-qPCR reactions are listed in Supplementary Tables S3 and S4. All reactions were performed in triplicate with samples derived from three experiments.

Gene disruption using CRISPR/Cas9 system

Guide RNAs (gRNAs) targeting the hHSF1 gene were designed using an online tool (http://crispr.mit.edu/). The DNA sequences including the gRNAs of the hHSF1 gene (Supplementary Table S5) were cloned into an pX330 vector (Addgene, catalog no. 42230) at the BbsI site83, and a plasmid vector pX330-hHSF1-270 was generated. Human OUMS-36T-3F cells were transfected with this vector and a pcDNA3.1-Neo vector (Invitrogen) by electroporation using a Nucleofector 2b device (Lonza, program A-023) with Amaxa MEF2 Nucleofector Kit (Lonza, VAPD-1005), and were then maintained in DMEM containing 10% FBS in 100 mm culture dishes. Twenty-four hours after transfection, these cells were treated with 1.0 mg/ml neomycin only for 3 days. Cell extracts were prepared from all colonies and subjected to Western blotting using anti-mHSF1n antibody. Three cell lines lacking HSF1 were isolated from 45 colonies. To re-express hHSF1, HSF1-null line 1–19 (KO1) was transfected with pcDNA3.1-hHSF1-HA by electroporation as described above, and stable cell lines were isolated in medium containing 1.0 mg/ml neomycin. hHSF1-re-expressed HSF1-null lines, H-8 and H-9, were identified by measuring HSF1 levels using Western blotting. Vertebrate HSF1 and HSF3 genes were expressed into the HSF1-null line in the same manner. HSF1-null lines stably overexpressing G0S2 were also generated by transfection with an expression vector pcDNA3.1-G0S2-HA.

Microarray analysis

Total RNA was prepared from wild-type OUMS-36T-3F cells, HSF1-null (KO) cells (#1-19 KO1), and HSF1 KO cells stably expressing cHSF1 or cHSF3 using RNeasy Mini Kit (Qiagen). Target cDNA was prepared from 500 ng total RNA with WT PLUS Reagent Kit (Thermo Fisher Scientific). Gene expression was analyzed using a Clariom S Array, human, containing more than 211,300 probes (supporting more than 20,800 genes) (Thermo Fisher Scientific). Hybridization to the microarrays, washing and staining, and scanning were performed using the Affymetrix GeneChip System, composed of the Scanner 3000 7G, Workstation, Fluidics Station 450, and Hybridization Oven 645 (Affymetrix). Scanned image data were processed using the Affymetrix Expression Console Version 1.1. Fold-changes in the expression of each gene were evaluated by a Gene Expression Analysis in the software Transcriptome Analysis Console 4.0 (Thermo Fisher Scientific). GSEA was performed using GSEA software (https://www.gsea-msigdb.org/gsea/index.jsp) and the Molecular Signatures Database (MSigDB) collection of hallmark gene sets84,85. A false discovery rate < 0.25 was considered as statistically significant.

ChIP assay

Chromatin immunoprecipitation (ChIP) assay was performed using a kit in accordance with the manufacturer’s instructions (EMD Millipore). The following antibodies were used: antibodies for hHSF1 (anti-mHSF1m), cHSF1 (anti-cHSF1c), and cHSF3 (anti-cHSF3γ)41,82. Real-time qPCR of ChIP-enriched DNAs in the loci of G0S2 and HSP70 (HSPA1A), and the intergenic region downstream of HSP70 was performed using the primers listed in Table S6. Percentage input was assessed by comparing the cycle threshold value of each sample to a standard curve generated from a 5-point serial dilution of genomic input, and then compensated by values obtained using normal IgG. IgG-negative control immunoprecipitation for all sites yielded < 0.05% input.

Statistical analysis

Data were analyzed using the Student’s t-test, a two-way analysis of variance (ANOVA), or a one-way ANOVA, followed by the Tukey–Kramer post hoc test (JMP Pro 16 software, SAS Institute Inc., Cary, NC, USA). Asterisks in figures indicate that differences were significant (P < 0.001, 0.001, or 0.05). Data are represented as the mean ± standard deviation (SD) for multiple independent experiments. To ensure the reproducibility of the results, each experiment was repeated at least three times with triplicate samples for each condition.