SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Supercoiling
  5. Changes in RNA conformation
  6. Chaperone-mediated sigma factor binding and release: heat shock
  7. Temperature sensing by a protein
  8. More proteinaceous thermometers
  9. Conclusions
  10. References

Many bacterial gene regulatory circuits are controlled by temperature. Temperature-mediated regulation occurs at the level of transcription and translation. Supercoiling, changes in mRNA conformation and protein conformation are all implicated in thermosensing. Bacterial virulence functions are often temperature regulated and thus many an example of thermoregulation comes from pathogenic organisms. H-NS is at the crossroads of regulation in many such systems. mRNA melting has also been shown to act as a thermosensing mechanism in various contexts. Proteins can also act as temperature sensors as exemplified by the gene regulator TlpA in Salmonella typhimurium.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Supercoiling
  5. Changes in RNA conformation
  6. Chaperone-mediated sigma factor binding and release: heat shock
  7. Temperature sensing by a protein
  8. More proteinaceous thermometers
  9. Conclusions
  10. References

Temperature has a profound effect on many cellular processes, and therefore bacteria must possess molecular thermosensing devices in order to adjust to changes in temperature. For example, the bacterial thermotactic response is based on temperature sensing by methyl-accepting proteins that regulate the swimming behaviour via a phosphotransfer signalling cascade (Nishiyama et al., 1997). In other instances, a sensed temperature cue needs to be plugged into gene regulatory circuits before the elicited activity can take place. Regulation of bacterial virulence gene functions is critical to successful invasion and/or colonization of a host organism. The need to regulate these specialized functions is explained by the high energy cost of unnecessary expression of virulence genes. Furthermore, such regulation may also be needed to exert differential expression of virulence functions while at different stages of the host infection. Transfer from the environment (or a heterologous host) offers a number of cues that the bacteria can use to transmit a signal to turn on their virulence functions (Maurelli, 1989; Miller et al., 1989; Mekalanos, 1992). One such signal is the increase in temperature. The thermosensing moiety is still far from clear, but at least three events can be distinguished that are involved in thermoregulation: supercoiling, change in mRNA conformation and change in protein conformation. A specialized case of thermosensing is the general system operating in bacteria, the heat shock response, that protects from various stress conditions such as increased growth temperature. The heat shock response also needs to transform the thermosensing information into needed gene activation.

In this review, we outline bacterial thermosensing strategies in the context of gene regulation. We are heavily biased towards pathogenic organisms, mainly because virulence loci are often influenced by temperature and have been much studied. In most cases a clear picture is yet to emerge, but several central themes can be distinguished that point to possible ‘molecular thermometers’ (Fig. 1). We do not attempt to present all of the different examples in which temperature has been shown to have a regulatory effect on transcription or translation of certain genes (there are just too many examples of genes induced by temperature) but instead we have chosen a few illustrative examples from among the very many that have been described.

image

Figure 1. . Schematic representation of thermosensing mechanisms in gene regulation. To be converted into gene regulatory signals, temperature changes have to be sensed by the cellular components. This may occur via conformational changes in mRNA or protein, or through DNA supercoiling. It is not clear if DNA itself can serve as a thermosensor or if a sensor protein is involved.

Download figure to PowerPoint

Supercoiling

  1. Top of page
  2. Abstract
  3. Introduction
  4. Supercoiling
  5. Changes in RNA conformation
  6. Chaperone-mediated sigma factor binding and release: heat shock
  7. Temperature sensing by a protein
  8. More proteinaceous thermometers
  9. Conclusions
  10. References

Supercoiling is the superhelical tension adopted by chromosomal or plasmid DNA, a process regulated by the balancing activities of two enzymes, topoisomerase I and the DNA gyrase (topoisomerase II). In bacteria, changes in the cellular ATP/ADP balance, environmental osmolarity, temperature and anaerobiosis are among the factors known to affect DNA supercoiling and subsequently alter gene expression. Changes in these physiological and environmental conditions are also among those one might expect bacteria to encounter when moving from one niche to another. The activity of some bacterial promoters responding to temperature changes is known to correlate with the degree of DNA supercoiling, and therefore supercoiling is considered to be one link between environmental changes and gene expression. Supercoiling has emerged as a central actor in many temperature-regulated virulence regulons (Dorman, 1991).

Uropathogenic Escherichia coli produce Pap pili when grown at 37°C but not at 25°C, indicating that the pap locus is thermoregulated (Göransson and Uhlin, 1984). Pili production is controlled by the transcription activator PapI, whose transcript papI was shown to be limiting at lower temperatures and to accumulate at a thermal upshift to 37°C. Overproduction of papI mRNA leads to a loss in thermoregulation. Later, papI was shown to be derepressed in a Δhns strain, implicating the H-NS protein in the repression of pap genes at lower temperatures (Göransson et al., 1990). H-NS is a histone-like protein with the ability to affect DNA supercoiling and is involved in condensing the bacterial chromosome into a chromatin-like structure, influencing processes such as recombination, transposition and transcription (Higgins et al., 1990; Hulton et al., 1990). Thus, the authors suggested that temperature causes changes in the DNA topology around the pap locus and thereby influences the binding and repressing ability of H-NS, hence the temperature regulation (Göransson et al., 1990). More recent studies demonstrated that in vitro H-NS blocks methylation of pap DNA (methylation is essential to pap regulation) in a protein concentration-dependent manner, and that in vivo, below physiological temperature, the presence of H-NS leads to methylation protection that represses transcription (White-Ziegler et al., 1998). The authors speculated that temperature directly affects H-NS and compromises its ability to productively repress DNA (White-Ziegler et al., 1998). A situation parallel to the pap system exists in Shigella, in which the virulence loci are regulated by the VirF regulator that activates virB transcription leading, in turn, to activation of the target effector genes (Tobe et al., 1991). In experiments in which virF mRNA was overproduced at 30°C virB could not be activated, whereas in a Δhns strain virB was constitutively active, suggesting that H-NS (VirR in Shigella) negatively regulates this process (Dorman et al., 1990; Tobe et al., 1993). Also, in vitro footprinting and transcription assays showed that H-NS binds a region in the virB promoter, probably blocking the action of RNA polymerase and thus acts as a repressor, whereas VirF binds to an upstream sequence and activates transcription but only from supercoiled DNA (Tobe et al., 1993). Interestingly, experiments on overproduction of H-NS at 37°C show that it is able to act as repressor at that temperature in contrast to the wild-type situation (where less H-NS is produced) and led to a proposal that it is the amount of H-NS that is the key to regulation (Hromockyj et al., 1992). In this proposed scenario, the levels of H-NS dictate its state of oligomerization that determines the ability to act as a repressor (Hromockyj et al., 1992). Yet another pathogenic bacterium, Yersinia enterocolitica seems to use changes in DNA topology as a regulatory means. In Y. enterocolitica chromosomally encoded YmoA is a histone-like protein with properties similar to H-NS but no shared similarity at the amino acid sequence level (Mikulskis and Cornelis, 1994). YmoA seems to be a ‘global’ repressor of the yop virulence regulon in Yersinia, implicating supercoiling as one putative mechanism of control as in the above examples (Lambert de Rouvroit et al., 1992). One study proposes that elevated temperature, which has been shown to affect reporter plasmid supercoiling, would dislodge YmoA and allow the VirF activator in Y. enterocolitica (not to be confused with Shigella VirF) to activate genes (Rohde et al., 1994).

Given the above examples, how does supercoiling affect promoter activity and what is the temperature sensor in this process? There are several candidate scenarios, some of which may be acting in concert. It is quite plausible that the ability of an effector protein, such as H-NS, to dock on the nucleic acid target is influenced by a temperature-mediated change in the topology of that particular stretch of DNA. Alternatively, H-NS may itself respond to temperature and release bound DNA, which can then be altered with respect to supercoiling. In-depth structure–function analysis is needed to address the question of whether H-NS is alone able to sense temperature. H-NS sensing could proceed by ways such as direct temperature-induced conformational changes or varying the state of oligomerization, both leading to the altered ability to bind DNA. Perhaps both DNA topology and H-NS conformation (which would dictate the ability to bind) are both temperature sensitive. Also, topoisomerase I, DNA gyrase and proteins capable of binding to curved DNA (other than H-NS) may act as direct sensors (although there is no evidence to this effect) or via factors that regulate them, complicating the issue even further. In this scenario, DNA topology would be modulated according to the environmental conditions, and factors such as H-NS would be secondary mediators in this regulatory scheme. It is tempting to speculate that all paths would converge on a protein(s) that receives and transmits the thermal cue.

Changes in RNA conformation

  1. Top of page
  2. Abstract
  3. Introduction
  4. Supercoiling
  5. Changes in RNA conformation
  6. Chaperone-mediated sigma factor binding and release: heat shock
  7. Temperature sensing by a protein
  8. More proteinaceous thermometers
  9. Conclusions
  10. References

An intriguing temperature-sensing mechanism has been described in Yersinia pestis in which LcrF (homologous to VirF in Y. enterocolitica) is the transcription activator involved in virulence gene regulation. Unlike virF, the lcrF gene is transcribed at equal rates at 25°C and 37°C. The regulatory mechanism here relies on thermal effect exerted directly on mRNA. A loop structure in the Shine–Dalgarno region of lcrF mRNA is melted at higher temperature and, consequently, allows lcrF translation to proceed (Hoe and Goguen, 1993). This model would not necessitate the presence of any proteinaceous negative effectors at lower temperatures. Another example of thermoregulation at the translational level is illustrated by the E. coli heat shock sigma factor rpoH mRNA. In vegetative cells, this mRNA is translationally inactive and not abundant. Induction of the heat shock response (see below) also includes the translational activation of rpoH mRNA. Studies with gene fusions have identified cis-acting elements within rpoH involved in thermal translational regulation. This implies that the production of RpoH could be influenced by conformational changes in the heat shock sigma factor mRNA itself (Kamath-Loeb and Gross, 1991; Nagai et al., 1991). In bacteriophage lambda, there are three levels of regulation when gene activity is increased at lower temperature favouring the lysogenic response. One mechanism of regulation seems to be the temperature effect on cIII mRNA when structure B is favoured by a conformational switch at lower temperatures allowing translation to proceed (Altuvia et al., 1989). Careful analysis of mRNAs involved in temperature-regulated loci might reveal several more examples of this nature. There is also some evidence pointing to temperature effects on the translation machinery itself (VanBogelen and Neidhardt, 1990). However, the direct effect of temperature on ribosome function remains an issue to be resolved. Possible mechanisms include changes in conformation of ribosomal RNA or protein, or simply a temperature effect on kinetics of translation.

Chaperone-mediated sigma factor binding and release: heat shock

  1. Top of page
  2. Abstract
  3. Introduction
  4. Supercoiling
  5. Changes in RNA conformation
  6. Chaperone-mediated sigma factor binding and release: heat shock
  7. Temperature sensing by a protein
  8. More proteinaceous thermometers
  9. Conclusions
  10. References

Shifting the growth temperature of E. coli from 30°C to 42°C results in an induction of about 20 proteins, while most others remain unaffected (Lemaux et al., 1978; Yamamori et al., 1978;). The induced proteins constitute the heat shock regulon comprising several chaperones and a few proteases for assisting the repair, stabilization or degradation of misfolded polypeptides (Hartl, 1996). Hsp70 in eukaryotes and the homologous DnaK chaperone in prokaryotes, have been proposed to serve as cellular ‘thermometers’ (Craig and Gross, 1991). Temperature-mediated upregulation and downregulation of heat shock genes is controlled at the level of stability and synthesis of the alternative sigma factor, σ32, which recognizes heat shock promoters (Yura et al., 1993). The DnaK chaperone system, consisting of DnaK, DnaJ and GrpE, has been proposed as a central regulator in charge of σ32 stability (Straus et al., 1990). This system can stabilize or destabilize σ32 and also repress its translation (Gamer et al., 1996). The details of this model propose that cycling of DnaK and DnaJ with σ32 sequester it in an inactive state that can also lead to degradation by way of FtsH (Gamer et al., 1996). Upon heat shock, the DnaJ and DnaK bind misfolded polypeptides and release σ32 for binding RNA polymerase. As the amount of misfolded polypeptides declines, DnaK and DnaJ chaperones are free to strip the σ32 from the RNA polymerase and sequester it again. Thus, direct denaturation of cellular polypeptides serves as the indicator of the temperature increase and the chaperone interaction releases the appropriate sigma factors to initiate transcription.

The chaperone-mediated ‘mechanism’ of thermosensing is unique to the heat shock response, however it is exploited as a regulatory means by at least one bacterial species. In Vibrio cholerae the ToxR/ToxS regulatory system of virulence functions is influenced by the heat shock gene regulon. In V. cholerae the production of both the cholera toxin and the Tcp fimbriae (colonization factor) are thermoregulated, their expression being higher at 30°C than at 37°C. The promoter for toxRS and the heat shock gene htpG are overlapping but divergently transcribed (Parsot and Mekalanos, 1990). Experiments with lacZ operon fusions to the htpG and toxRS promoters indicate that, even at moderate temperatures (such as 37°C), the activity of the σ32-dependent divergently transcribed htpG promoter hinders the expression of the toxRS promoter. It has been proposed that this repression at higher temperature could be involved in silencing the operon at an inappropriate stage during the infection such as during the passage through the stomach and could be later turned on by way of other transcription factors (Skorupski and Taylor, 1997).

Temperature sensing by a protein

  1. Top of page
  2. Abstract
  3. Introduction
  4. Supercoiling
  5. Changes in RNA conformation
  6. Chaperone-mediated sigma factor binding and release: heat shock
  7. Temperature sensing by a protein
  8. More proteinaceous thermometers
  9. Conclusions
  10. References

TlpA is a first-documented case of a temperature-sensing gene regulator (Hurme et al., 1997). Importantly, it is an intriguing example of how a protein might ‘mechanistically’ achieve the task of thermal sensing at its simplest and then couple this information into gene regulation. TlpA is encoded by the plasmid gene tlpA and is conserved in Salmonella strains carrying the virulence plasmid. TlpA is an autoregulatory repressor, which uses its folding equilibrium to regulate the DNA binding activity. The mechanism is based on a long coiled-coil domain that spans more than two-thirds of its 371 amino acids (Hurme et al., 1996). Coiled coils are formed by two (or more) intertwining helices that wrap around each other (Cohen and Parry, 1990). The specialized sequence of the coiled coils is made up by the amino acid heptads (a–b–c–d–e–f–g), where a and d positions are occupied by hydrophobic residues that constitute the buried apolar stripe which is the driving force of this fibrous conformation (Cohen and Parry, 1990). Coiled coils can be described by a single two-state equilibrium between unfolded monomers and folded oligomers. This equilibrium is both temperature and concentration dependent. In TlpA this equilibrium seems to be adjusted such that it can act as a temperature sensor at temperature upshifts to physiological temperatures, all the way up to about 43°C. Transcription of tlpA is kept in check by the repressor TlpA, which in its dimeric and folded coiled-coil conformation is able to bind DNA and repress transcription. Temperature upshift leads to a shift in the equilibrium that favours the non-functional and unfolded monomeric form. The decrease in active repressor leads to increased transcription of tlpA and a subsequent increase in the subunit concentration. The increase in subunit concentration shifts the equilibrium to favour the folded and active form of TlpA. Temperatures exceeding the melting midpoint can lead to ‘runaway’ transcription as then TlpA, regardless of subunit concentration, remains monomeric and unfolded. However, the unfolding is reversible and a downshift in temperature leads again to the formation of stable repressors. The precise role of TlpA in Salmonella remains unresolved but it might act as an accessory repressor in the intricate gene regulons of this pathogen (Hurme et al., 1997).

More proteinaceous thermometers

  1. Top of page
  2. Abstract
  3. Introduction
  4. Supercoiling
  5. Changes in RNA conformation
  6. Chaperone-mediated sigma factor binding and release: heat shock
  7. Temperature sensing by a protein
  8. More proteinaceous thermometers
  9. Conclusions
  10. References

The coiled-coil motif is a versatile structure used in a number of proteins (Lupas, 1996). Thus, the TlpA sensor mechanism based on the coiled-coil structure comes as no surprise. In eukaryotes, the heat shock factor (HSF) is responsible for transcriptional activation of genes necessary for coping with temperature and other stresses. HSF also contains a coiled-coil element that trimerizes the protein in a temperature-dependent manner (Peteranderl and Nelson, 1992). Although HSF has an inherent ability to respond to temperature, there are other factors that regulate its activity, such as Hsp70 (Wu, 1995). The coiled-coil motif can be used for temperature sensing also when the protein is not a gene regulator. It is of interest to note that Tar, the aspartate chemoreceptor involved in thermo/chemotaxis, contains a cytoplasmic signalling domain flanked by two regions, MH1 and MH2, predicted to form coiled coils (Nishiyama et al., 1997; Surette and Stock, 1996). Although the exact mechanism through which Tar would sense temperature is not clear, the available information invites speculation that as in TlpA this would be based on the properties of the coiled-coil motif. M proteins of group A Streptococci are surface proteins forming coiled-coil dimers. As some classes of M protein appear as dimers at low temperatures, but as monomers at 37°C, it has been suggested that also the M protein could sense temperature via its coiled coil and possibly even transmit a signal through the bacterial cell wall (Cedervall et al., 1997).

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Supercoiling
  5. Changes in RNA conformation
  6. Chaperone-mediated sigma factor binding and release: heat shock
  7. Temperature sensing by a protein
  8. More proteinaceous thermometers
  9. Conclusions
  10. References

It is clear that temperature acts as a cue for gene regulation in numerous systems. Whether it is nucleic acid or protein, the sensing boils down to built-in molecular thermometers or thermostats whose structure is sensitive to a certain range of temperatures and is prone to conformational perturbations. Thermoregulated supercoiling is a highly complex system in which it is difficult to clearly point out the sensing moiety. However, it may be controlled by a temperature-sensing protein(s) that manipulates the DNA topology. The sensing mechanism (and at the same time the sensor itself) used by bacteria to propagate the ‘change in temperature’ signal is likely to be in many cases a thermally induced change in mRNA or protein conformation. The primary sensor can be denatured as occurs in heat shock when the unfolded polypeptides are ‘sensed’ as a measure of temperature, or ‘melted’ as in various mRNA stem–loop structures allowing translation to occur. As TlpA demonstrates a gene regulator protein can take direct orders from the environment, in this case using for this task its folding/unfolding equilibrium. TlpA is not irreversibly denatured and it serves as an active thermometer and gene regulator. It would not be surprising to find similar proteinaceous thermometers in other systems.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Supercoiling
  5. Changes in RNA conformation
  6. Chaperone-mediated sigma factor binding and release: heat shock
  7. Temperature sensing by a protein
  8. More proteinaceous thermometers
  9. Conclusions
  10. References