Genes & Genetic Systems
Online ISSN : 1880-5779
Print ISSN : 1341-7568
ISSN-L : 1341-7568
Regulation of the heat shock response in Escherichia coli: history and perspectives
Takashi Yura
ジャーナル オープンアクセス HTML

2019 年 94 巻 3 号 p. 103-108


The heat shock response mediated by transcription factor σ32 is a major stress response to cope with heat and other stresses in Escherichia coli. Although much attention has been paid to the role of highly conserved heat shock proteins such as chaperones and proteases in sustaining cellular protein homeostasis under stress, relatively little is known about the dynamic nature of underlying regulatory mechanisms. When cells are suddenly exposed to high temperature, synthesis of σ32 is rapidly induced by activated translation of rpoH mRNA, which encodes σ32, through disruption of mRNA secondary structure. The increased synthesis of σ32 is accompanied by stabilization of σ32, which is normally very unstable and rapidly degraded by the membrane-localized FtsH protease. It was recently found that σ32 must be localized to the inner membrane by the SRP-dependent pathway to work properly for regulation, but the roles played by membrane and other components of the cell remained unknown. Random transposon mutagenesis of the strongly deregulated I54N-σ32 mutant has now started to unravel the complex regulatory circuit, involving membrane protein(s), other cellular components or σ32-interfering polypeptides, for dynamic fine-tuning of σ32 activity that could be of vital importance for cell survival.


The heat shock response (HSR) plays a major role to sustain protein homeostasis to cope with heat and other stresses in all organisms (Morimoto, 2011). Whereas the past decades’ studies on the structure and function of well-conserved heat shock proteins (HSPs) such as DnaK/DnaJ and GroEL/GroES chaperones and proteases have illuminated the central importance of this response (e.g., Yura et al., 1993; Hartl, 1996; Bukau and Horwich, 1998; Guisbert et al., 2004, 2008; Georgopoulos, 2006; Schumann, 2016), the actual regulatory mechanisms of the HSR, mediated by specific transcription factor(s) (σ32 in the case of Escherichia coli), received relatively little attention for many years.

When E. coli cells growing at 30 ℃ are suddenly shifted to 42 ℃, synthesis of HSP is induced very rapidly (induction phase) through increased synthesis and stabilization of σ32, which is extremely unstable (half life, ~1 min) and normally produced in very small amounts (~50 molecules per cell) (Grossman et al., 1984). Both in vivo and in vitro analyses revealed that the increased synthesis of σ32 occurs at the level of translation through disruption of the secondary structure of rpoH mRNA, encoding σ32, in which the 5’ region of this mRNA directly serves as a ‘thermometer’ to respond to high temperature (Nagai et al., 1991; Morita et al., 1999a, 1999b, 2000). This induction of HSPs (~3 to 5 min) is soon followed by a gradual decrease of σ32 level (adaptation phase) for 10 to 15 min (negative feedback control) to reach a new steady-state level. The regulatory circuit of the HSR in E. coli is schematically depicted in Fig. 1.

Fig. 1.

Regulation of the σ32-mediated heat shock response in E. coli. This figure is a modified version of the original cover illustration used in Genes and Genetic Systems, 93, 6 (2018), prepared by members of the Akiyama laboratory, Institute for Frontier Life and Medical Sciences, Kyoto University. For discussion of the possible roles of McrC/YdbA2 proteins in feedback regulation of the HSR, see Yura et al. (2018).

Although the network of chaperones and proteases has been shown to play key roles in the folding of nascent protein chains and in removing unfolded and mis-folded proteins (Gamer et al., 1992, 1996; Bukau and Horwich, 1998; Guisbert et al., 2004, 2008; Georgopoulos, 2006), how this network is regulated to cope with heat stress remains largely unknown. σ32 activity reflects the cellular protein folding state through a negative feedback loop mainly exerted by the DnaKJ/GrpE and GroEL/ES chaperone systems. Free chaperones can directly bind and inactivate σ32, but these chaperones are also titrated out by unfolded proteins that accumulate during the HSR. Depletion or overexpression of chaperone substrates increases or decreases σ32 activity, whereas overexpression of either chaperone system rapidly inhibits σ32 activity. However, the chaperone-resistant phenotype of a deregulated mutant of σ32 (I54N-σ32) cannot be recapitulated in vitro (Yura et al., 2007).

It was a pleasant surprise when transposon (Tn5) insertion upstream of the ftsY gene encoding SR, the receptor for the signal recognition particle (SRP), increased the σ32 level strikingly in the presence of excess chaperones. This eventually led to the finding that σ32 must be transported to the inner membrane by the SRP-SR-SecY pathway for heat shock regulation to work properly (Lim et al., 2013; Miyazaki et al., 2016). However, how the membrane localization of σ32 is linked with regulation of σ32 activity and level remained unknown. This regulatory network now seems to have manifested itself, as the result of recent suppressor studies of the strongly deregulated and hyperactive mutant (I54N-σ32), which is defective in the membrane localization of σ32 in E. coli (Yura et al., 2018). This review summarizes historical development of the field, and points out some directions for the future.


Our characterization, in the mid-1970s, of temperature-sensitive mutants with possible defects in RNA polymerase subunits initiated the work which led to the discovery of the heat shock response in E. coli. Extensive pulse-labeling experiments revealed the characteristic profile of HSP induction when cells growing at 30 ℃ were suddenly shifted to 42 ℃ (Yamamori et al., 1977). Subsequent studies with a temperature-sensitive mutant previously reported by Cooper and Ruettinger (1975) revealed that the HSP induction is genetically determined (Yamamori et al., 1978; Yamamori and Yura, 1980, 1982; Neidhardt and VanBogelen, 1981). A number of hin/htpR mutants defective in the key regulatory gene encoding σ32 were then characterized (Yura et al., 1984), and the first minor σ factor found in E. coli32) turned out to control the HSR (Grossman et al., 1984). The regulatory network controlling the σ32-mediated HSR in E. coli is schematically depicted in Fig. 1.

The above findings in E. coli overlapped with the time period when major HSPs such as DnaK, DnaJ, GrpE and GroEL, GroES in E. coli and their conserved homologs in yeast and other organisms were identified in various laboratories (see Craig and Gross, 1991; Liberek et al., 1992; Bukau and Horwich, 1998; Guisbert et al., 2004, 2008). Despite the apparent similarity between the HSR in E. coli and the ‘heat shock puff’ found much earlier in Drosophila salivary gland chromosomes (Ritossa, 1962), it took some time to realize that the observations in E. coli and the heat shock puff of Drosophila are indeed equivalent. This delay came largely from the fact that the Drosophila heat shock puff had long been believed to be unique to flies, the order Diptera.


Subsequent findings in a number of laboratories world-wide, as well as the evolutionarily conserved features of major HSPs such as HSP60 (GroEL), HSP70 (DnaK) and HSP90 (HtpG) (see Craig and Gross, 1991; Morimoto, 2011), led to the general belief that the HSR is ubiquitous in all organisms (bacteria, yeast, and animal or plant cells); this led to a series of heat shock meetings, starting at Cold Spring Harbor, New York in 1982. Molecular chaperones such as DnaK, DnaJ, GroEL and GroES proteins in the nomenclature of E. coli were directly shown to play critical roles in supporting growth at high temperature and became a subject of an exciting new field, but it has taken much longer to understand regulation of the HSR. Whereas a similar σ32-mediated HSR was found widely among Gram-negative bacteria (Nakahigashi et al., 1998), the mechanisms controlling the activity, amount and stability of σ32 remained poorly understood even in E. coli.


The first publication (Cooper and Ruettinger, 1975) that eventually led to the finding of the σ32-mediated HSR in E. coli dealt with a nonsense mutant defective in the synthesis of a major protein (later identified as GroEL chaperone) at high temperature. When the same mutant was analyzed by two-dimensional SDS-polyacrylamide gel electrophoresis, however, several HSPs were found to be affected, and the mutated gene was called hin or htpR (later renamed as rpoH), suggesting that the gene encodes a transcription factor that controls synthesis of a group of proteins, now called heat shock proteins (HSPs), that are found at higher concentrations when cells are exposed to higher temperatures (Neidhardt and VanBogelen, 1981; Yamamori and Yura, 1982; Neidhardt, 1984).

Further studies with wild type and several htpR/hin mutants defective in HSP synthesis (including htpR deletions), along with nucleotide sequencing analysis of the htpR/hin gene (Landick et al., 1984; Yura et al., 1984), established that this gene is essential for growth and survival at high temperatures (~40–42 ℃). However, it was found to be dispensable at low temperature, because the htpR deletion mutants can grow at or below 20 ℃ (Zhou et al., 1988). Therefore, increased requirements for σ32 at high temperature may come from the fact that higher levels of HSPs (including chaperones and proteases) are required for growth at higher temperatures. However, when excessive amounts of HSPs (e.g., DnaKJ, GroEL) accumulate, they start to inhibit their own synthesis very strongly (Tilly et al., 1983; Straus et al., 1987, 1990).


When log-phase cells of E. coli grown at 30 ℃ are suddenly exposed to 42 ℃, synthesis of a set of HSPs increases rapidly and transiently through increased synthesis and stabilization of σ32 (Grossman et al., 1984, 1987; Kamath-Loeb and Gross, 1991). Based on computer prediction of extensive secondary structure formed between the translation initiation region and an internal region of rpoH mRNA, multicopy plasmids carrying the rpoH-lacZ translational fusion were constructed and their expression was analyzed by measuring β-galactosidase to see the effects of particular base changes on expression (rpoH translation). Structural and functional analyses of these fusion constructs clearly demonstrated the interplay of two cis-acting mRNA segments in the translational control of σ32 (Nagai et al., 1991). Further comparison of RNA properties such as CD and translation efficiencies confirmed the model that the mRNA secondary structure is largely responsible for inhibiting translation at low temperature, thus serving as an RNA thermometer to protect cells from heat stress (Nagai et al., 1991, 1994; Yuzawa et al., 1993; Morita et al., 1999a, 1999b).

In spite of the potential importance of the RNA thermometer, possible roles of trans-acting factor(s) that might modulate the mRNA secondary structure nevertheless remained unknown. Contrary to the involvement of the RNA thermometer or factors that can modulate its sensitivity, the induction of HSPs caused by accumulation of ‘abnormal proteins’ (e.g., certain human proteins produced in E. coli) depends solely on stabilization of σ32 and not on increased synthesis of σ32 (Kanemori et al., 1994).


Intracellular stability of σ32 is primarily controlled by FtsH protease, an essential, membrane-bound AAA protease (Herman et al., 1995; Tomoyasu et al., 1995), although other proteases such as HslVU may also degrade σ32 significantly (Kanemori et al., 1999). Besides σ32, FtsH is known to control concentration of both sugar and lipid moieties of lipopolysaccharides and uncomplexed forms of SecY protein, which is essential for transport of membrane proteins (Ito and Akiyama, 2005). In view of the recent finding that membrane localization of σ32 is required for heat shock regulation to work (see below), it seems plausible that the membrane localization of σ32 somehow helps to control the amount and stability of σ32; however, the apparent complexity of structure and function of FtsH protease precludes our understanding of the exact mechanism controlling degradation of σ32 as well as the regulatory impact of σ32 stability control under normal or stress conditions (Bittner et al., 2017).


During extensive analyses of rpoH mutants producing hyperactive and stable σ32 protein (e.g., I54N-σ32), a clear discrepancy was observed between the phenotypes found in vivo (Horikoshi et al., 2004; Yura et al., 2007) and in vitro (Yura et al., 2007); namely, strong feedback resistance of the mutant σ32 to chaperone inhibition in vivo could not be recapitulated in vitro (Yura et al., 2007), suggesting involvement of some unknown factor(s). In trying to dissolve this discrepancy, transposon (Tn5) insertion was used to look for new target genes that might affect activity, synthesis or stability of σ32. Indeed, Tn5 insertion into the region upstream of the ftsY gene, which encodes SR (the receptor for SRP; Luirink et al., 1994), increases the activity and level of σ32. Since the SRP-SR-SecY pathway was known to be involved in localization of some membrane proteins in E. coli, this started a series of experiments which finally led to the discovery that the regulatory region of σ32 (region 2.1) directly recognizes and binds SRP, and that the SRP-dependent localization of σ32 to the membrane is required for feedback control of σ32 to work properly (Lim et al., 2013; Miyazaki et al., 2016).


In view of the unexpected development of HSR regulation described above, an extensive search for suppressors was carried out by random Tn5 mutagenesis of the hyperactive σ32 mutant that produces stable I54N-σ32 protein as the parent, looking for clones with reduced σ32 activity (Yura et al., 2018). Unexpectedly, two novel Tn5 suppressor insertions were found, one within the mcrC gene that encodes the DNA restriction subunit of McrBC restriction enzyme (Raleigh, 1992), and the other within the ydbA2 gene that encodes part of a putative auto-transporter.

Interestingly, these suppressor Tn5 insertions reduced the activity but not the amount or stability of I54N-σ32. Moreover, the same truncated form of McrC protein (fused with part of Tn5) expressed from the pUC18 plasmid strongly inhibited σ32 activity of cells producing I54N-σ32. The suppression was specific for I54N-σ32 and did not work with other σ32 mutants tested. Moreover, deletion of the mcrC or ydbA2 gene hardly affected the I54N-σ32 activity. These results, taken together, suggested that the truncated form of McrC protein, and presumably also of YdbA2 protein, resulting from the respective Tn5 insertion, could directly interact with I54N-σ32 and enhance the membrane mobilization and sequestration of I54N-σ32 (Yura et al., 2018).


Although much remains to be learned about the nature of the regulatory network that controls synthesis, activity, stability and localization of σ32, the recent development encourages further efforts, using various in vivo and in vitro strategies, to dissolve one of the long-lasting problems of biology today, namely the ubiquitous cellular response to heat and other stresses. Future directions may include reconstruction of negative feedback regulation (SRP-mediated membrane export and chaperone-dependent degradation) of σ32 in vitro using the SRP system, chaperones, membranes with SecYEG/FtsH and other factors, protein–protein interaction studies by in vivo photo-cross-linking experiments (PiXie; Miyazaki et al., 2018, 2019) to monitor dynamic behavior of σ32, and structural analysis of σ32 and its complexes with SRP, DnaK/J and/or FtsH. It may also be worth noting that the heat shock response in bacteria may be evolutionarily related to “unfolded protein responses” such as the ER (endoplasmic reticulum) stress response in eukaryotic organisms.

For further information on the HSR in E. coli and other bacteria, readers are referred to the various reviews and papers listed below.


The author gratefully acknowledges C. A. Gross (UCSF, USA), K. Ito (Kyoto Sangyo University), Y. Akiyama (Kyoto University) and S. Chiba (Kyoto Sangyo University) for their kind support and discussion of his work, which were essential to write this review since his official retirement in 1993 and 2000 from the Institute for Virus Research, Kyoto University and from the HSP Research Institute, Kyoto, Japan, respectively. He also thanks K. Ito, Y. Akiyama, H. Mori, Y. Hizukuri, R. Miyazaki and other former colleagues for their kind help in preparing this review.

© 2019 by The Genetics Society of Japan