Heat stress

The Bacterial Heat Stress Response refers to the mechanism by which bacteria adapt to a sudden increase in the ambient temperature of growth. The precise components of the signal–transduction system that senses and responds to heat stress vary among bacteria. Even within a single bacterial species, several different mechanisms are used to respond to heat stress. However, the logic of the response is universal. During the induction phase, the cell senses increased temperature and produces a signal that activates a transcription factor to increase the transcription of a group of genes. Accumulation of unfolded proteins contributes to signal generation, and a majority of the proteins produced during this stress response help to restore the normal folding state of the cell. During the adaptation phase, as the folding state of the cell returns to normal, the signal is damped down and transcription of the heat shock genes declines. Thus, in general, the heat stress response is self-limiting. Only when organisms are switched to lethal temperatures does the response continue unabated for as long as the cells are able to synthesize protein. In the present article, we first consider the general inputs and outputs to emphasize the universal logic of the bacterial heat shock response. We then describe several specific responses in detail, emphasizing how different components are used to execute this logic. 

I. INPUTS TO THE HEAT STRESS RESPONSE 

What is the thermometer that allows the cell to sense even small changes in temperature? Unfolded or partially folded proteins are, by far, the best characterized inducers of the heat shock response, and they must comprise part of the cellular thermometer. Normally, the low levels of unfolded or partially folded proteins result either from newly synthesized proteins or those maintained in a partially folded state prior to transport across membranes. Upon heat shock, some fully folded proteins partially or completely denature, increasing the pool of unfolded proteins and the need for the cellular proteins that maintain folding state. This is a matter of crucial concern for the cell. Partially folded proteins may not simply lose their activity; they may, in fact, cause toxic outcomes for the cell through a variety of mechanisms. It is, therefore, a top cellular priority to decrease the extent of protein unfolding whenever it occurs. Additional stresses, such as alcohol, also increase protein unfolding, thereby activating the same response. Other consequences of heat stress, besides the general increase in level of protein unfolding, are sensed by the cell. However, most of these inputs are currently either unknown or uncharacterized, with one exception. As we shall see later, in E. coli, the cell directly senses the folding state of a critical RNA molecule, and this provides a secondary thermometer to sense temperature. How is the cellular thermometer constructed? Surprisingly, the nature of the primary thermometer sensing heat stress is not known in detail for any system. However, a variety of circumstantial evidence, described later, suggests that titration of the outputs of the response, the heat shock proteins themselves, may provide a thermometer of sufficient calibration to explain the response (Fig. 49.1). Some (or many) of the heat shock proteins have dual roles in the cell: they interact with unfolded protein substrates to promote folding and they interact with the heat stress transcription factor to regulate its activity. During the induction phase of the heat stress response, an increase in the cellular concentration of unfolded proteins increases the ratio of unfolded substrates to heat shock proteins. This could titrate them away from their “homeostatic” regulatory role vis-à-vis the heat stress transcription factor. As a consequence, the amount or activity of transcription factor will increase, resulting in an increased concentration of heat shock proteins. During the adaptation phase of the heat stress response, the ratio of unfolded protein substrates to heat shock proteins normalizes, the heat shock proteins resume negative regulation of the transcription factor, and the response is damped down (Fig. 49.1). 

II. OUTPUTS OF THE HEAT STRESS RESPONSE 

A. Chaperones Chaperones, or proteins that help other proteins fold, are a major class of the proteins produced after heat stress. In the test-tube, under very dilute conditions, small proteins fold by themselves, demonstrating that the polypeptide chain itself encodes the information necessary for proper folding. However, in the cell, proteins are present at very high concentrations, and the nascent, unfolded protein has a very high potential to aggregate with other nascent chains via hydrophobic interactions, rather than proceeding on its folding pathway. By interacting with nascent and partially folded proteins, the molecular chaperones successfully thwart the tendency toward aggregation. There are several different families of chaperones, most of which are highly conserved throughout evolution. One of the most prominent molecular chaperones is the Hsp70 family, called DnaK in bacteria, which work together with a co-chaperone Hsp40, called DnaJ in bacteria. The DnaKJ chaperone family is conserved in almost all organisms and members of each family are found in every compartment of the eukaryotic cell. Interestingly, in bacteria or in eukaryotic organelles, a third protein, GrpE, is necessary for this chaperone machine to function. GroEL, which cooperates with GroES, is a second major chaperone in bacterial cells and eukaryotic organelles, but is not found in the eukaryotic cytosol. The other major chaperones, Hsp90 and the small heat shock proteins, are conserved among most organisms; however, their role in the bacterial cell has not been completely elucidated. Some chaperones are very large protein machines, consisting of multiple subunits of one or more proteins. For example, the GroELS chaperone machine has a central cavity consisting of two seven-membered rings of GroEL subunits. One or both ends is “capped” by a single seven-membered ring of the smaller GroES subunits. Proteins first bind to an exposed hydrophobic cavity in GroEL. Then, a combination of GroES binding and ATP hydrolysis drives conformational changes, which first expose hydrophilic residues and then drive release of the protein from the cavity. This allows at least the first critical steps of protein folding to take place within the GroEL cavity, in total isolation from other unfolded molecules in the bacterial cytoplasm. 

By contrast, the DnaK chaperone machine consists of a single molecule of DnaK, which transiently interacts with its co-chaperone, a dimer of DnaJ molecules. For DnaK, its binding site for unfolded proteins is a small cleft located in the C-terminus of the molecule, which interacts preferentially with hydrophobic stretches in protein chains. The peptide-binding domain of DnaK is connected via a linker region to the N-terminal ATPase domain of the molecule. Interestingly, as was the case for the GroES co-chaperone, DnaJ both alters the rate of ATP hydrolysis and substrate release. GrpE works as a nucleotide release factor, promoting dissociation of whatever nucleotide is bound to DnaK. Despite its different construction, for both the DnaK and GroEL chaperone machines, an ATP-driven cycle of binding to and release from the chaperone underlies the action of the chaperone. B. Proteases Proteases, or proteins that degrade other proteins, are the second major class of proteins produced following heat stress. When proteins cannot be refolded by chaperone machines, the quality control system, composed of a variety of proteases, degrades the damaged proteins. This process serves two purposes. It releases the amino acids for reuse in making new proteins and eliminates proteins that cannot be repaired. Proteases use a variety of recognition systems, including exposure of hydrophobic C-terminal tails, internal hydrophobic stretches, and N-terminal signals. Interestingly, many proteases are also large machines, composed of one of more types of subunit. The active sites of these proteases are contained within the central cavity, thus confining protein degradation to this cavity, where it is insulated from the cytoplasm. Such proteases also have either separate subunits or separate domains that carry out protein recognition. These portions of proteases can serve as chaperones when removed from the cleavage machinery of the protease. Many, but not all, proteases require ATP, which, most likely, drives the unfolding process. C. Other heat shock proteins The two slowest steps in protein folding are isomerization around the cis-trans bond of proline and the making and breaking of disulfide bonds. Protein folding catalysts are proteins that speed up these slow steps in protein folding. Some of the peptidyl prolyl isomerases (catalyzing proline isomerization) and the disulfide bond proteins are heat shock proteins. More systematic study of the various proteins induced by heat has indicated that a number of heat shock proteins do not fall into these simple categories. The function of many such proteins is still being elucidated. 

III. THE ESCHERICHIA COLI CYTOPLASMIC HEAT SHOCK RESPONSE: REGULATION BY _32 

A. Description of the E. coli heat shock response Escherichia. coli cells can rapidly sense both temperature upshifts and temperature downshifts. The induction phase begins within one minute of shift from 30_C to the higher growth temperature of 42 _C, and the peak rate of synthesis of the 30 or more heat shock proteins is attained by 5 to 10 min after temperature upshift. At this point, the adaptation phase begins, and synthesis of the heat shock proteins declines until the new steady-state rate of synthesis is attained. At higher growth temperatures, both the maximal and steady-state rates of synthesis of the heat shock proteins are higher and the induction phase is prolonged. At temperatures so high that they are lethal to cells, the heat shock response is maintained at its maximal point for as long as the cells can synthesize protein. Indeed, at these temperatures, synthesis of heat shock proteins constitutes the major protein synthetic activity of the cell. In the converse temperature shift from the high growth temperature of 42 _C to a lower growth temperature of 30 _C, the synthesis of heat shock proteins begins to decline within 1 to 2 min, reaching a point of minimal synthesis at about 10 min after temperature downshift. Synthesis of heat shock proteins resumes slowly, reaching the rate normal for that temperature within 50 to 100 min after downshift. Note that this response on shift to lower growth temperature is distinct from the cold shock response. The cold shock response ensues when cells are shifted to very low temperatures (20 _C or below) and involves the induction of a distinct set of proteins. The cold shock response will not be discussed here. B. A transcriptional factor regulates the E. coli heat shock response In bacterial cells, the sigma subunit of RNA polymerase directs the transcriptase to the promoter region of the gene. Most bacteria contain multiple sigmas. In addition to one or more housekeeping sigmas, responsible for the bulk of the transcription, cells have several alternative sigma factors, which allow them to respond to environmental or developmental signals. Each sigma factor recognizes a distinctive set of promoters in the cell. RNA polymerase is directed to the promoters of the heat shock genes by one such alternative sigma, called _32. During the heat shock response, the rate of transcription of the heat shock genes and, consequently, the rate of synthesis of the heat shock proteins, changes in response to alterations in the amount and/or activity of _32. Upon temperature upshift, the amount of _32 increases rapidly, reaching its peak just prior to the peak rate of synthesis of the heat shock proteins, accounting for the induction phase of the response (Fig. 49.2). 

The amount of _32 increases both because this normally very unstable molecule (usually degraded with a half-time of only 1 min) is transiently stabilized and because the rate of translation of _32 transiently increases. During the adaptation phase, _32 becomes unstable again and its rate of translation decreases, resulting in a decline in the amount of _32 in the cell. Eventually, _32 reaches a new steady-state level characteristic of the particular growth temperature. The adaptation phase may be sharpened by some control over the activity of _32. Thus, the response of the cell to temperature upshift is primarily governed by changes in the amount of _32 (Fig. 49.2). In contrast, there is little change in the amount of _32 upon temperature downshift. In this case, a dramatic decrease in the activity of _32 accounts for the dramatic shutoff of transcription of the heat shock genes and the consequent decrease in heat shock protein synthesis. C. Two (or more) thermometers control the E. coli heat shock response At least two different thermometers control expression of the heat shock proteins. One thermometer controls the translation of _32 mRNA in a positive way: increased physical (environmental) stress leads to increased translation (Fig. 49.3). This pathway is induced by exposure to heat but not by accumulation of unfolded proteins. Asecond thermometer regulates the stability of _32 itself (Fig. 49.3). This pathway is induced by accumulation of unfolded proteins, as well as by exposure to heat. Elements of this thermometer may also regulate the activity of _32. The nature of the thermometer that controls translation of _32 during the induction phase has now been defined (Morita, 1999). The only player is the _32 message itself. At low temperatures, the translation start site is occluded by base-pairing with other regions of the _32 mRNA. Raising the temperature destabilizes this base-pairing, allowing increased translation of _32. Both mutational studies (which changed the basepairing in the critical region) and chemical probing  (which assessed base-pairing as a function of temperature) are consistent with this idea. 

Most importantly, by using an in vitro assay that determines accessibility of mRNA to translation by the ribosome, it has been possible to show that no components other than _32 mRNA are necessary for this regulatory system (Fig. 49.3). During the adaptation phase, translation of _32 mRNA declines. The DnaK chaperone machine has been involved in this translational repression, but the mechanism by which this chaperone operates is still unclear. The nature of the thermometer that regulates the stability and possibly the activity of _32 is more complex and less defined (Fig. 49.3). Since increased production of unfolded proteins in the cell stabilizes _32, unfolded proteins are at least part of the signal that calibrates this thermometer. We also know that the DnaK, DnaJ, and GrpE heat shock proteins are required for the instability of _32 and its inactivation. In other words, the DnaK chaperone machine is negatively regulating both the stability and activity of _32. A homeostatic mechanism coupling the occupancy of the DnaK chaperone machine with unfolded proteins to the amount and activity of _32 has been proposed. The key concept is that the thermometer is set by competition between _32 and all other unfolded or misfolded proteins that bind to the DnaK chaperone machine. During the induction phase, the increased amounts of unfolded or misfolded proteins titrate the DnaK chaperone machine away from _32, relieving their negative regulatory effects on _32 stability. As a consequence, _32 is stabilized and its amount will rise. This response is self-limiting. During the adaptation phase, overproduction of DnaK, DnaJ, and GrpE will restore the free pool of these chaperones to an appropriate level. This notion can be extended to account for inactivation of _32 during temperature downshift. Here, a decreased amount of unfolded or misfolded proteins will allow _32 to compete better for the DnaK, DnaJ, GrpE chaperone machine, with a consequence of inactivation of _32. 

In this model, the amount of free DnaK, DnaJ, and GrpE is a “cellular thermometer” that measures the “folding state” of the cell (Fig. 49.3). A key prediction of this model, that the amount of the DnaK chaperone machinery in the cell directly correlates with the activity of _32, has recently been verified. However, at present, this model is far from established, as many of the critical experiments to test the model have not been carried out. D. Additional mechanisms for the cytoplasmic heat shock response in E. coli Although _32 controls expression of many of the genes that respond to heat stress, induction of the _32 regulon is not synonymous with the cellular response to temperature upshift. Another alternative sigma factor, _s, primarily responsible for transition into stationary phase, is also induced by heat. Induction of the _s regulon after exposure to high temperature is slower than induction of the _32 regulon. The signal transduction system has not been studied in detail. However, like _32, the activity and amount of _s is controlled on multiple levels, including stability, translation, and, possibly, activity. In addition, one operon (psp) is controlled by a dedicated activator protein that promotes transcription by yet another sigma factor, _54, after shift to very high temperature. Finally, yet another sigma factor, _E, is responsive to heat stress generated in the periplasm. That response will be described in detail in Section V. E. Conservation and divergence of the _32 paradigm for the cytoplasmic heat shock response Homologues of _32 have been identified in diverse gram-negative bacterial species. Including E. coli, ten gamma proteobacterial species contain _32. In addition, _32 has been found in seven alpha and one beta proteobacteria. With one exception, these bacterial species have only one gene encoding _32. The exceptional species, Bradyrhizobium japonicum, has three genes encoding _32 with distinct functional and regulatory features. Where it has been examined, the amount of _32 increases with temperature, although the mechanism for accomplishing this is variable. The E. coli paradigm for regulating _32 is most closely conserved among gamma bacteria. Here, when the temperature increases, generally, the stability and translation of _32 increase as well. For some alpha bacteria, transcription of _32 appears to increase with temperature. Regardless of the mechanisms for accomplishing this, the increased cellular concentration of _32 gives the increased transcription of heat shock genes characteristic of the heat shock response. No homologues of _32 have been identified in gram-positive bacteria. 

IV. THE BACILLUS SUBTILIS HEAT SHOCK RESPONSE: THE CIRCE/HRCA CONTROLLED REGULON 

A. The heat shock response The dnaK and groE operons of B. subtilis exhibit a heat shock response similar to the one exhibited by these operons in E. coli. Upon shift to high temperature, transcription of the dnaK and groE operons rapidly increases, leading to a rapid increase in the rate of synthesis of the seven heat shock proteins encoded by the B. subtilis dnaK operon (DnaK, DnaJ, GrpE, the HrcA repressor, and three other proteins) and the two heat shock proteins encoded by the B. subtilis groE operon (GroEL and GroES). This response reaches its peak about 5 to 10 min after shift to high temperature. Transcription of these operons then declines, reaching a new steady-state rate by 30 to 60 min after the high temperature shift. However, the components generating this response are completely different from those responsible for the heat shock in E. coli. The dnaK and groE operons of B. subtilis are transcribed by its housekeeping sigma (SigA). The heat shock response of these operons is controlled by a negative regulatory system, which is composed of an operator site, called CIRCE, and its cognate repressor, called HrcA (Fig. 49.4). CIRCE is a nine base inverted repeat sequence separated by nine bases (TTAGCACTC- N9-GAGTGCTAA), which is located close to or overlapping the dnaK and groE promoters. 

The HrcA repressor (encoded in the dnaK operon) binds to the CIRCE element to shut off transcription of these operons. Upon temperature upshift, repression is temporarily relieved and synthesis of the CIRCEcontrolled heat shock proteins increases. The major elements of this regulatory system have been verified by genetic analysis and some in vitro studies. Mutating conserved bases in CIRCE results in constitutive expression of the operon downstream of the mutated element, as expected for an operator. Likewise, mutating hrcA results in constitutive expression of both the groE and dnaK operons. Direct binding studies, demonstrating that HrcA binds to CIRCE, have been performed with variable success. HrcA is very prone to aggregation in vitro, limiting the ability to characterize this reaction extensively. As will be seen in the next section, the propensity of HrcA to aggregate is likely to be for its function. B. The CIRCE/HrcA thermometer Heat shock regulation in this system requires that the HrcA repressor be transiently removed from its CIRCE operator upon shift to high temperature. The mechanism for accomplishing this constitutes the cellular thermometer. A priori, two solutions seem possible. First, HrcA itself may undergo a temperature- mediated transition. Second, achieving the native state of HrcA may require the intervention of some other molecule in the cell, which is itself responsive to temperature. A variety of evidence suggests that the latter solution has been chosen. The current idea is that HrcA requires constant interaction with a chaperone to maintain its native state. The GroEL/S chaperone is considered to play this role and, as for the E. coli heat shock response, a homeostatic mechanism has been proposed (Fig. 49.4). The occupancy of the B. subtilis GroEL/S chaperone machine with unfolded proteins is suggested to be coupled to the fraction of HrcA in the native state. During the induction phase, the increased amounts of unfolded or misfolded proteins titrate the GroEL/S chaperone machine away from HrcA, allowing it to aggregate or otherwise misfold in the cell. As a consequence, it will be unable to bind to CIRCE and repress transcription. This response is self-limiting. During the adaptation phase, overproduction of GroEL/S will restore the free pool of these chaperones to an appropriate level, HrcA will be maintained in its native state and repression from the CIRCE operator will be resumed. Two key predictions of this model have been verified qualitatively. First, this model implies that repression by HrcA will be responsive to the amount of GroEL/S in the cell and that is true. When GroEL/S is depleted from cells, the heat shock response is induced at low temperature. 

Conversely, overexpressing GroEL/S represses basal synthesis at low temperature and decreases the absolute amount of induction at high temperature. Second, this model implies that overexpression of known GroEL/S substrates, but not other proteins, will induce the heat shock response. This is also true. Finally, accumulating evidence suggests that GroEL/S and HrcA interact as required by this model. GroEL/S decreases aggregation of HrcA in vitro and promotes binding of HrcA to the CIRCE operator, which suggests that the two proteins interact directly, rather than through intermediate molecules. However, a great deal more work remains to be done to understand precisely how this thermometer works. Two observations make it unlikely that HrcA itself senses temperature. First, in GroEL/S depleted cells, HrcA is inactive even at low temperature, indicating that the protein is not intrinsically native at low temperature. Second, when HrcA from a thermophilic organism (B. stereothermophilis) is used to reconstitute the heat shock response in B. subtilis, heat shock occurs at the low temperature characteristic of B. subtilis, rather than the high temperature characteristic of B. stereothermophilis. Thus, it is the cellular milieu, rather than the HrcA protein per se, that senses temperature. C. Distribution of CIRCE/HrcA regulation The CIRCE/HrcA regulatory system is very widespread. It occurs in more than 40 different eubacteria, including gram-positive organisms, gram-negative organisms, cyanobacteria, and more distantly related eubacteria, such as chlamydia and spirochaeta. The particular genes regulated by the system vary, depending upon the organism. In gram-positive organisms with low G_C content examined thus far, CIRCE/HrcA regulates the dnaK and groE operons, but in those with high G_C content, and in _-proteobacteria, only groE is regulated. In addition, in some bacteria, several other genes have been reported to be regulated by this system. To add to the regulatory diversity, preliminary experiments suggest that in at least one gram-positive organism (Lactococcus lactis), DnaK regulates HrcA. Moreover, several cases where regulation by HrcA and _32 coexist have been documented. 

D. Additional mechanisms for response to heat stress in B. subtilis Although understanding of the CIRCE/HrcA regulatory system is most advanced mechanistically, this system controls only nine of the many genes induced after shift to high temperature. A great many others are controlled by _B, an alternative sigma factor that carries out the general stress response and the starvation response in B. subtilis. Heat is one of the many stimuli that induce this system. _B is controlled by a phosphorylation cascade that affects the ability of an anti-sigma to bind to _B. This system is currently under intense study but, at present, there are few details about how heat stress alters this regulatory cascade. Additional heat shock proteins in B. subtilis are controlled by unknown mechanisms. However, recent studies in other organisms have identified several repressors in addition to HrcA that are responsive to heat stress (OrfY, HspR) and such repressors may be involved in regulating additional heat stress proteins in B. subtilis as well. V. THE ESCHERICHIA COLI EXTRACYTOPLASMIC HEAT SHOCK RESPONSE: REGULATION BY _E A. Description of the E. coli extracytoplasmic heat shock response The hallmark of the gram-negative bacterial cell is the existence of two membrane layers, the inner or cytoplasmic membrane and the outer membrane, which, in turn, form the boundaries of two aqueous subcellular compartments, the cytoplasm and the periplasm. The conditions within each of these compartments differ markedly. The cytoplasm is an energy-rich, highly regulated reducing environment, in which basic cellular processes, such as transcription, DNA replication, and translation, are carried out. In contrast, the extracytoplasmic compartment is a relatively energy poor, oxidizing environment in which conditions vary with those of the external environment, due to the existence of pores in the outer membrane which allow the free exchange of small molecules and some specific substrates. Given the disparity in conditions between the two compartments, it is not surprising that the heat stress response in E. coli is compartmentalized. 

Exposure to high temperature activates both the cytoplasmic and extracytoplasmic responses, whereas conditions that specifically perturb protein folding in the periplasmic compartment activate only the extracytoplasmic response. B. The thermometer controlling the extracytoplasmic heat shock response The extracytoplasmic heat shock response is controlled by the alternative sigma factor _E, which is responsible for directing transcription of the _10 genes that comprise its regulon. Only some of these genes have been identified. These include the periplasmic protease DegP and the periplasmic peptidyl prolyl cis/trans isomerase, FkpA, which are involved, respectively, in protein degradation and folding. Interestingly, transcription of the heat shock factor _32 is also under the control of _E, suggesting an interconnection between the two stress regulons. The induction phase of the _E heat stress response is slow, with maximum synthesis occurring about 20 min after shift to high temperature. The adaptation phase of this response has not been studied. Studies on this response have advanced to the point where many of the players are known; however, the signaling mechanisms coupling the activity of _E to temperature are currently unknown. The central problem facing this regulatory system is to transduce a signal generated in the periplasm to the cytoplasm, where activity of _E is regulated. That problem has been solved by utilizing a protein chain consisting of two negative regulators. RseA, the major negative regulator of _E, is a membrane spanning protein whose cytoplasmic face acts as an anti-sigma factor. During steady-state growth conditions, RseA binds to _E, thereby preventing this sigma factor from binding to RNA polymerase. The periplasmic face of RseA binds the second negative regulator, RseB. RseB is located completely in the periplasm and is a weak negative regulator of _E. The activity of _E is controlled by regulated proteolysis. Upon shift to high temperature, RseA is destabilized, relieving negative regulation of _E. At least one function of RseB is to enhance the stability of RseA. Various agents that specifically enhance unfolded proteins in the periplasmic compartment (lack of folding agents, overexpression of outer membrane proteins, interruption of lipopolysaccharide biosynthesis) all induce _E. How generation of unfolded proteins is coupled to destabilization of RseA is currently unknown. Moreover, additional yet to be discovered mechanisms regulate _E. 

VI. SUMMARY AND PROSPECTS 

Study of the heat stress response has led us to realize the extraordinary importance of controlling the state of protein folding in the cell. Organisms have evolved multiple, graded responses to cope with this problem. The thermometers that calibrate the response to the level of stress are composed of different materials. However, almost all of them are induced, at least in part, by unfolded proteins and respond by overexpressing a set of universally conserved heat shock proteins. Currently, many, but not all, of the types of responses to heat stress have been identified. However, on the most basic level, the way in which the thermometer controls the gradual response to temperature is not really understood for any system. Moreover, the interaction of the various heat responsive systems with each other has not yet been studied. Clearly, considerably more will be learned about this response in the next few years.