Transcription factor cooperativity with heat shock factor 1

Abstract

The heat shock response has been characterized by the induction of major heat shock proteins that suppress protein aggregation by facilitating protein folding. Recently, we found that mammalian heat shock factor 1, a master regulator of HSP genes, regulates non-HSP genes that suppress protein aggregation by controlling protein degradation in cooperation with the transcription factor NFAT.

When cultured cells or whole organisms are exposed to elevated temperatures, synthesis of a small number of highly conserved proteins, named “heat shock proteins (HSPs),” that include HSP110, HSP90, HSP70, HSP40 and HSP27, is sharply and dramatically induced. This response is universal; it is observed in every organism from bacteria to humans and in almost all cell types in multicellular organisms.¹ Simultaneously, synthesis of many proteins is induced depending on the organism and cell type, which here we call “heat-inducible proteins,” are involved in protein degradation, nucleotide repair, detoxification and metabolism.^2–4 In general, adaptive induction of the heatinducible proteins, including HSPs, is recognized as the “heat shock response.”

Cellular protein homeostasis (proteostasis) is maintained by a network of pathways that influences protein synthesis, folding, translocation, assembly/disassembly and degradation.⁵ Loss of cellular proteostasis causes many systemic and neurodegenerative disorders, known as protein-misfolding diseases. An increase in temperature causes protein misfolding and aggregation, resulting in a proteostasis imbalance. To maintain proteostasis under heat shock conditions, bacterial cells induce not only the synthesis of HSPs that inhibit protein denaturation and refold denatured proteins as molecular chaperones, but also proteases, such as HslUV, Lon, Clip and DegP, that clear misfolded proteins by degrading them. In mammalian cells, misfolded proteins during heat shock are mainly degraded by the proteasome,⁶ but the synthesis of machinery components is barely induced. Therefore, it had been considered until recently that mammalian cells rely on HSPs to maintain proteostasis in heat shock conditions.⁴

The heat shock response is regulated mainly at the level of transcription by heat shock factors (HSFs) that bind to the heat shock response element (HSE), which is composed of three inverted repeats of a conserved sequence nGAAn.^7–9 A single HSF regulates this response in C. elegans, while four members of the HSF family (HSF1-4) have been characterized in vertebrates, and the expression of HSPs is regulated mainly by HSF1 in mammals.¹⁰ Remarkably, a gain of HSF1 function significantly ameliorates disease progression in C. elegans models of neurodegenerative disorders such as polyglutamine diseases and Alzheimer's disease,^11–13 whereas loss of HSF1 function accelerates it through the reduced expression of HSPs. In mammals, this HSF1-mediated induction of HSP expression is correlated with the reduced progression of neurodegenerative disorders,^14,15 suggesting that HSF1 maintains proteostasis by regulating the expression of HSPs.¹⁶

Non-HSP Proteostasis Pathways

By analyzing the roles of mammalian and chicken HSF family members in terms of adaptation to proteotoxic stresses that cause protein misfolding, we came to believe that the HSF family members regulate the expression of non-HSP genes involved in proteostasis.¹⁰ We found that HSF1 deficiency accelerated an aggregation-prone polyglutamine (polyQ) protein aggregation in MEF cells, which was associated with reduced expression of four newly identified target genes including NFATc2.¹⁷ NFATc2 (NFAT1) is a member of the NFAT transcription factor family,¹⁸ and NFATc2 deficiency also accelerated polyQ aggregation. Importantly, overexpression of active HSF1 only partially suppressed the elevated polyQ aggregation in NFATc2-null MEF cells, whereas overexpression of NFATc2 suppressed it efficiently in HSF1-null cells. Moreover, NFATc2 deficiency accelerated polyQ aggregation in the brain, similar to HSF1 deficiency in Huntington disease mice, which was associated with a shortening of lifespan. This was the first demonstration that HSF1 maintains proteostasis through the regulation of non-HSP gene expression.

How does the NFATc2 pathway maintain proteostasis? We found that NFATc2 directly binds to and activates the PDZK3 and CRYAB genes in cooperation with HSF1.¹⁷ PDZK3 has six PDZ domains, which could be a scaffold for polyQ protein or ubiquitylated cellular proteins for degradation, and αB-crystallin, the product of CRYAB, is a component of the E3 ubiquitin ligase complex. Consistently, we showed that NFATc2 controls proteostasis in part by regulating the degradation of polyQ protein. Our results demonstrate that mammalian HSF1 maintains proteostasis by regulating the expression of genes, which are involved not only in protein folding, but also in protein degradation.

Constitutive DNA Binding

We examined the regulation of the NFATc2 gene by HSF1, and would like to emphasize first that HSF1 regulated gene expression even in unstressed cells. Biochemical analyses established that Drosophila HSF or mammalian HSF1 stays as an inactive monomer in unstressed cells, and is converted to an active trimer that binds to the HSE with high affinity in response to heat shock.^7,19 Subsequently, disruption of the HSF1 gene was shown to result in altered expression of HSP and non-HSP genes in the tissues and to be required for development and maintenance,^20,21 suggesting that HSF1 regulates gene expression in unstressed cells. We found that HSF1 binds to the promoter of the NFATc2 gene in unstressed MEF cells and its expression is markedly reduced in HSF1-null cells.¹⁷ Re-expressed HSF1 bound to the promoters and restored the reduced expression in HSF1-null cells. In contrast, HSF1 mutants, which could not form a trimer, neither bound to the promoter nor restored the reduced expression when they were re-expressed in HSF1-null cells. Similarly, HSF1 that can form a trimer upregulated another HSF1-target gene, IL-6, in unstressed MEF cells and macrophages.²² These results demonstrate that a small amount of HSF1 trimer directly binds to and activates its target genes in unstressed cells.

HSF1 constitutively binds to genomic DNA composed of an ambiguous HSE sequence in vivo. HSE is described as at least three inverted repeats of the 5 bp sequence nGAAn.⁸ Recombinant mouse HSF1 and HSF2 bind preferentially to inverted repeats of the consensus sequence nGAAn in vitro (Fig. 1A).²³ In contrast, recombinant human HSF4 binds to inverted repeats of a consensus sequence, nGnnn, in vitro (Fig. 1A), and most of the HSF4 binding region contains none or only one GAA sequence in vivo in the lens.²⁴ Furthermore, a substantial number of HSF4 binding regions are co-occupied by HSF1 and HSF2 in the lens. In the promoter of mouse NFATc2, we found HSE located at −1,883 to −1,864 from a transcription start site is required for both constitutive and inducible expression of the NFATc2 gene in MEF cells.¹⁷ Consistent with the original consensus nGAAn sequence, the “G” nucleotide is conserved in all four pentanucleotide units of HSE, whereas the nucleotides “AA” are not conserved at all. Taken together, we think that HSF1 is able to bind constitutively to genomic DNA with inverted repeats of a consensus sequence nGnnn in vivo (Fig. 1B). HSF1 cannot bind to such an ambiguous HSE sequence in vitro, but its binding may be stabilized by other HSFs or other HSF-interacting factors on the genome in unstressed cells. Upon heat shock, it is apparent that abundant trimeric HSF1 preferentially binds, probably by itself, to the HSE containing inverted repeats of the consensus nGAAn sequence.^2,25

Figure 1.

Sequence properties of HSF binding sites in unstressed cells. (A) DNA consensus sequences for recombinant HSF binding in vitro were generated with WebLogo, by using sequences recognized by mouse HSF1 and HSF2,²³ and by human HSF4.²⁴ The height of each letter represents the relative frequency of nucleotides at different positions in the consensus. (B) Proposed consensus sequence recognized by HSFs in unstressed cells in vivo. The enriched HSF4 binding consensus sequence is at least three inverted repeats of nGnnn,²⁴ which is consistent with the consensus sequence for recombinant HSF binding in vitro. A substantial number of HSF4 binding regions (70%) is co-occupied by HSF1 and HSF2, suggesting the proposed consensus sequence. This sequence is ambiguous compared with that composed of inverted repeats of nGAAn recognized by an activated HSF1 in heat-shocked cells in vivo.²,²⁵

Transcription Factor Cooperativity

We showed that both HSF1 and NFATc2 regulate the expression of PDZK3 and CRYAB, which promote degradation of polyQ protein. Interestingly, the overexpression of HSF1 into wild-type MEF cells induced PDZK3 expression several-fold, whereas that into NFATc2-null cells hardly induced the expression. Furthermore, HSF1 overexpression markedly induced PDZK3 expression when NFATc2 was simultaneously overexpressed. The cooperative effect on the expression of CRYAB was also observed. ChIP assays revealed that HSF1 and NFATc2 directly bind to different regions of the PDZK3 promoter (Fig. 2A).¹⁷ Similarly, HSF1 and NFATc2 directly bind to different regions of the CRYAB promoter. Heat shock induced binding of not only HSF1, but also NFATc2, to both promoters. The binding of one factor to the promoters was not affected by the other, indicating that HSF1 and NFATc2 independently bind to the PDZK3 and CRYAB genes. Taken together, HSF1 induces the expression of another transcription factor, NFATc2, which controls expression of the degradation-related genes, to maintain proteostasis by protein degradation. Importantly, the cooperativity of HSF1 and NFATc2 is required for maximal induction of some genes, although the mechanism is still unclear.

Figure 2.

Cooperativity of HSF1 with transcription factors induced by HSF1. (A) To maintain protein homeostasis, HSF1 and NFATc2 independently bind to the promoters of PDZK3 and CRYAB, which promote degradation of polyQ protein and synergistically activate the genes, probably by stabilizing a complex containing histone modifying enzymes (X complex).¹⁷ (B) HSF1 binds to the promoters of IL-6, NOS2 and ICAM1 to partially open the chromatin structure in the absence of stimuli, in part by recruiting a chromatin remodeling complex (BRG1complex) and complexes containing histone modifying emzymes (Y complex).²²,²⁸ During the inflammatory response, ATF3 is induced by HSF1, binds to the nucleosome-free promoter and suppresses expression of inflammatory genes.

We also found that HSF1 cooperates with activating transcription factor 3 (ATF3) in inflammatory responses (Fig. 2B). ATF3 is a member of the ATF/CREB family of transcription factors,²⁶ and is induced in response to various types of stresses and inflammatory responses.²⁷ We showed that HSF1 induces the expression of ATF3 and the treatment of LPS and heat shock synergistically induce the expression in part through HSF1.²⁸ As a result, ATF3 suppresses the expression of many genes for inflammatory cytokines.²⁷ The inflammatory cytokines support the invasion of inflammatory cells and promote tissue repair, but at excessively high levels are detrimental. This negative regulatory loop may inhibit any excessive febrile response and injuries to tissues throughout the body.²⁸ In immune cells, the chromatin structure of the promoters in early inflammatory genes is partially opened and RNA polymerase II is stalled near the promoters before stimulation.^29,30 Interestingly, we found that HSF1 constitutively binds to the promoters of inflammatory genes such as IL-6, NOS2 and ICAM1 in unstressed MEF cells and macrophages, and partially opens the chromatin structure to facilitate binding of transcription factors such as ATF3 and NFκB.²² Thus, HSF1 cooperates with ATF3, a product of its target gene, to suppress inflammatory gene expression.

Involvement of HSF1 in regulating the chromatin structure of the HSP70 gene has been extensively studied especially in Drosophila cells.³¹ In unstressed cells, GAGA factor binds to the promoter of HSP70 and is setting up the pause of RNA polymerase II on the region residing 20–40 base pairs down stream of the transcription start site, which results in the loss of nucleosomes on the promoter proximal region of HSP70. In response to heat shock, Drosophila HSF quickly binds to the nucleosome-free region and the nucleosomes in the body of HSP70 are lost rapidly.³² HSF binding is followed by recruitment of RNA polymerase II, elongation factors such as P-TEFb and Spt6, and histone modifying enzymes such as Trithorax and CREB-binding protein (CBP).^33,34 In mammalian cells, no functional analogue of Drosophila GAGA factor has been identified yet, but HSF1 constitutively binds to HSP70, IL-6, PDZK3 and CRYAB genes.^17,22,28 Mammalian HSF1 physically interacts with BRG1, a component of SWI/SNF chromatin remodeling complex, histone acetyltransferases p300 and CBP, and an ASC-2 complex containing histone H3K4 methyltransferases.^35–37 Therefore, we could speculate that HSF1 and NFATc2 may cooperatively stabilize a complex containing histone modifying enzymes on PDZK3 and CRYAB genes (Fig. 2A),¹⁷ like the cooperativity of NFATc2 with Smad3.³⁸ On the inflammatory genes, HSF1 may recruit a chromatin remodeling complex through BRG1 and histone modifying enzymes including CBP to partially open chromatin structure (Fig. 2B).²⁸

HSF1 alone directly regulates the expression of many genes, including HSP. In addition, as we discussed here, it cooperates with other transcription factors induced by HSF1 to play diverse physiological roles through regulating non-HSP expression. In future, we would like to understand the cooperativity mechanisms in detail by analyzing factors that interact with HSF1.

Abbreviations

HSE

heat shock response element

HSF

heat shock factor

HSP

heat shock protein

MEF

mouse embryonic fibroblast

polQ

polyglutamine