Temperature is an omnipresent physical variable reflecting the rotational, vibrational and translational motion of matter, what Richard Feynman called the “jiggling” of atoms. Temperature varies across space and time, and this variation has dramatic effects on the physiology of living cells. It changes the rate and nature of chemical reactions, and it alters the configuration of the atoms that make up nucleic acids, proteins, lipids and other biomolecules, significantly affecting their activity. While life may have started in a “warm little pond”, as Charles Darwin mused, the organisms that surround us today have only made it this far by devising sophisticated systems for sensing and responding to variations in temperature, and by using these systems in ways that allow them to persist and thrive in the face of thermal fluctuation.
In this Primer, we describe some of the basic molecular strategies organisms use to sense temperature. While the activities of all biomolecules are altered as a function of temperature, the thermosensors we focus on here are molecules whose temperature sensitivity (Box 1) provides information about the thermal environment that is used to trigger an appropriate physiological or behavioral response. In this overview we have not sought to be comprehensive, but rather to focus on select examples that illustrate some of the emerging general principles. We do not discuss the heat shock chaperone system, as this is induced by the accumulation of misfolded proteins under a wide array of insults that trigger proteotoxic stress.
Box 1.
What Q10 means
Temperature sensing relies on temperature-dependent changes in biochemical activities such as kinase function and ion flow. Such changes are frequently quantified in terms of ‘Q10’. Q10 is a dimensionless quantity that represents the ratio of a property measured at two temperatures 10°C apart. For example, if the expression of a thermoTRP channel results in 1 µA of current at 27°C and 8 µA of current at 37°C, its Q10 would be 8. The thermal sensitivity of any quantifiable property can potentially be expressed using Q10, whether it be kinase activity, neuronal spiking, or the rate of flashing of lightning bugs.
An important aspect of Q10 is that the properties described scale exponentially rather than linearly with temperature. For example, while subjecting a thermoTRP with a Q10 of 8 to a 10°C temperature rise would lead to an 8-fold increase in current, a 20°C rise yields a 64-fold increase. The formal formula for Q10 reflects its exponential nature:
Q10 = (R2/R1)^(10)/(T2- T1)
In this formula, T2 and T1 are the two temperatures at which a property is measured, and R2 and R1 are the values of the property measured at T2 and T1, respectively.
Q10 values are widely used as quantitative expressions of temperature sensitivity because they are straightforward to calculate and interpret, and because they describe many biological phenomena quite well. As a rule of thumb, processes exhibiting Q10 values under ~4 are not considered particularly temperature sensitive, while processes with Q10 values above ~7 are considered quite thermosensitive. The Q10 values of many thermosensors are much higher: TRPM8 has a Q10 of ~24, while TRPV1 has a Q10 of ~40 and the Anopheles gambiae thermoTRP agTRPA1 has a Q10 of ~200.
Nucleic acid thermometers
Structural lability in response to temperature changes is a broadly acting mechanism that underlies the thermosensing functions of a wide range of molecules ranging from nucleic acids to proteins to lipids. The thermally labile hydrogen bonds that join complementary base pairs in nucleic acids provide a particularly fertile substrate for the creation of biological temperature sensors, and this property has been exploited by many bacteria which deploy DNA- and RNA-based environmental temperature sensors. The responses to temperature changes mediated by these thermosensors include immediate, fast-adapting responses, as well as longer-term, persistent responses that allow the organism to adapt to its new thermal environment. Both short- and long-term responses include changes in gene expression; for example, pathogenic bacteria rapidly induce expression of virulence genes upon a rise in temperature to 37°C, a temperature shift indicative of invasion of a warm-blooded host. Temperature appears to alter gene expression at both transcriptional and post-transcriptional steps via DNA and RNA thermosensors.
At the transcriptional level, temperature is read out in part via regulation of access of transcriptional machinery to DNA. This can be mediated by changes in DNA structure driven by primary DNA sequence. In Shigella, transcription of the virF virulence gene required for host invasion is induced at temperatures higher than 32°C. At lower temperatures, cooperative binding of H-NS repressor molecules to two sites in the virF promoter separated by a DNA bend occludes the transcriptional machinery; the bend melts at temperatures >32°C, relaxing the DNA and allowing binding of transcriptional activators (Figure 1A). Similar mechanisms of temperature-dependent gene expression utilizing DNA curvature as a temperature sensor may be a feature of thermosensation in other bacterial pathogens as well.
Figure 1. Examples of biological thermosensors.
(A) DNA thermosensor. DNA is bent at lower temperatures, allowing cooperative binding of repressor molecules and inhibition of transcription. Melting of the bend at higher temperatures promotes transcription. (B) RNA thermosensor. The stem-loop structure occludes the Shine-Dalgarno sequence and AUG initiation codon at low temperatures and prevents translation. At higher temperatures, the stem-loop unfolds and allows translation initiation. (C) Protein thermosensor. Dimerization of a repressor protein at low but not higher temperatures promotes DNA binding and inhibition of transcription. (D) Lipid–protein thermosensor. Altered membrane fluidity and thickness at different temperatures affects transmembrane protein–lipid interactions and regulates protein activity as a function of temperature. In this example, at lower temperatures, the ‘buoy’ region of DesK is buried in the membrane and results in increased kinase activity. Exposure of the buoy (or other protein conformation changes) upon increased membrane fluidity and decreased thickness at higher temperatures turns off protein signaling. (Drawing by David Doroquez.)
Temperature sensing via modulation of changes in secondary (or tertiary) structure is also employed by so-called ‘RNA thermometers’ to regulate translational efficiency of virulence, heat shock, and other genes in bacteria. Essentially, the mRNA adopts a thermolabile stem-loop structure in its 5’ UTR at low temperatures, effectively blocking access to the Shine-Dalgarno sequence and preventing translation. At higher temperatures, the hairpin structure opens and allows binding of ribosomes and translation (Figure 1B). A related (although structurally more complex) RNA ‘zipper-like’ thermosensory mechanism in part underlies the rapid and massive induction of the alternative RNA polymerase sigma factor σ32 upon heatshock of Escherichia coli. An alternative use of this mechanism is in the induction of the cspA RNA chaperone in E. coli and the cIII pro-lysogenic genes of phage λ at low temperatures. In contrast to the RNA zipper mechanism described above, both cspA and cIII transcripts instead adopt alternative secondary structures that promote or inhibit mRNA stability and translation at low and high temperatures, respectively.
These simple yet effective RNA thermometer mechanisms have been used to confer temperature sensitivity onto the expression of synthetic molecules in bacteria. Are related mechanisms employed in eukaryotes? Although less well-understood, the secondary structure at the 5’ UTRs of the mRNAs of hsp70 and hsp90 heat shock genes has been suggested to regulate translational efficiency at higher temperatures, and changes in temperature-dependent functions of the non-coding RNA hsr1 underlie activation of the HSF-1 heat shock transcription factor. A somewhat distinct post-transcriptional mechanism transduces temperature information to the circadian clock in Neurospora. Low temperatures promote translation initiation at nonconsensus Kozak sequences upstream of the frq circadian gene; efficient initiation at these sequences decreases FRQ protein abundance at low temperatures and vice versa. RNA thermometers have some advantages over DNA thermosensors as the former involve structural alterations to existing mRNAs resulting in immediate responses. Given the diversity of RNA-based thermosensory mechanisms deployed across organisms, it is likely that additional nucleic acid-based thermosensors remain to be identified.
Thermosensing via protein conformational change
As proteins are complex biomolecules comprising thousands of interacting atoms, they are also highly susceptible to experiencing significant structural alterations in response to temperature changes. Temperature-dependent changes in protein conformation have been exploited to regulate a host of cellular processes in response to ambient temperature. Protein thermosensors have been extensively studied in eubacteria, which have devised multiple solutions to the problem of coping with sudden environmental changes. This temperature detection is mediated not only by nucleic acid thermometers (described above) but also by protein thermosensors.
Environmental temperature changes can be read out by proteins implicated in the regulation of range of cellular processes, including transcription, protein stability and signal transduction. In Salmonella, the TlpA autoregulatory repressor protein is functional at low but not at high temperatures such as those encountered upon host infection. TlpA adopts a coiled-coil homodimeric configuration at low temperatures allowing it to bind DNA and repress its own transcription. However, at 37°C the dimers dissociate leading to loss of transcriptional repressor activity (Figure 1C). Similarly, the virulence regulator RovA in Yersinia and the RheA transcriptional regulator of heat shock genes in Streptomyces act as thermometers due to temperature-dependent alterations in their protein conformation and DNA-binding properties. In addition to altering structures of transcriptional regulators, temperature can also specifically alter the conformation of proteins involved in regulating protein quality control. An excellent example of this is provided by the DnaK Hsp70 chaperone system. In this process, temperature-dependent changes in the structure of the co-chaperone GrpE thermosensor protein alters the substrate affinity of DnaK by modulating the regulation of ADP/ATP exchange; this altered affinity in turn affects protein folding and aggregation at higher temperatures.
Conformational changes can also alter the intrinsic enzymatic properties of proteins such as kinases, leading to activation of a signaling cascade in a temperature-regulated manner. The two-component histidine kinase/response regulator signaling module is used widely in bacteria to sense and respond to a range of environmental stimuli, including chemicals, pH, osmolarity and as described below, temperature. This simple module generally consists of a membrane-bound sensor protein with an intracellular kinase domain containing a histidine residue. Upon encountering an environmental signal, the histidine is autophosphorylated. The phosphate group on the histidine is then transferred to an aspartate on an intracellular response protein which can regulate a number of cellular pathways, including transcription. In the plant pathogen Pseudomonas syringae, the CorS histidine kinase sensor autophosphorylates at 18°C and transfers the phosphate to the CorR transcriptional regulator to effect temperature-specific gene expression. Similarly, in Agrobacterium tumefaciens, virulence genes are activated only at temperatures below 32°C via the VirA histidine kinase/VirG response regulator. The DesK/DesR two-component system of Bacillus subtilis is probably the best studied example of a thermoresponsive sensor kinase system (Figure 1D; see below). However, although in all cases temperature is thought to directly alter the conformation of the sensor kinase, these proteins exhibit remarkably diverse structures, suggesting that temperature may regulate the properties of these kinases via multiple as yet unknown mechanisms.
A particularly intriguing protein-based thermosensory mechanism is employed by E. coli to navigate spatial thermal gradients. E. coli uses a biased random walk strategy to navigate both thermal and chemical gradients; flagellar motility is altered to increase tumbling when moving in an unfavorable direction and, conversely, tumbling rates are decreased and smooth swimming is enhanced when moving in a favorable direction. Interestingly, the Tar and Tsr transmembrane proteins are both the major chemoreceptors as well as thermoreceptors. Tar binds Asp, and Tsr binds Ser or Gly amino acids; both are also warmth sensors. The activation state of these proteins is read out by the intracellular CheA histidine kinase and CheY response regulator to alter flagellar motor function and bacterial motility. However, prolonged exposure to Ser/Gly or Asp results in methylation of Tsr and Tar, respectively, resulting in adaptation of the receptor and decreased signaling via CheA/CheY. Rather remarkably, the thermosensory functions of Tar and Tsr are altered by their chemical-induced methylation state. While in the unmethylated state both Tar and Tsr are warmth receptors; in the methylated state, Tsr is thermo-insensitive, whereas Tar switches to sensing cold. Thus, in essence, these proteins act as integrators of both chemicals and temperature and direct the appropriate response in an experience-dependent manner. The molecular mechanisms of temperature sensing by either unmethylated or methylated Tar or Tsr receptors are not fully understood.
Thermosensing via changes in membrane properties
In addition to altering protein conformations directly, temperature changes can affect protein activity as a secondary consequence of structural alterations in intimately associated molecules. In the case of the bacterial cold sensor DesK, for example, temperature-dependent changes in the lipid membrane appear to be the primary mediator of the protein’s thermal responsiveness.
DesK is a multi-pass transmembrane histidine kinase that acts as part of a two-component system in B. subtilis to maintain membrane fluidity at low temperature (Figure 1D). Upon cooling, DesK activates the response regulator DesR, which induces transcription of the fatty acid desaturase Des. By introducing a double bond into saturated lipids, Des induces a kink in the lipid tail that increases membrane disorder, offsetting the fluidity decrease that otherwise accompanies cooling. This DesK-dependent introduction of unsaturated fatty acids into the bacterial membrane enhances low temperature survival.
The activation of DesK upon cooling appears to be intimately related to changes in membrane structure. Among other effects, temperature changes alter membrane thickness, with the decrease in disorder that accompanies cooling causing an increase in membrane thickness. Changes in membrane thickness can alter the activity of membrane proteins by changing the orientation or conformation of transmembrane regions. Intriguingly, two hydrophilic residues near the amino terminus of DesK’s first transmembrane domain are critical for its cold-activation. This region has been dubbed the ‘buoy’, as the buoy’s hydrophilicity drives it toward the lipid/water interface, while the surrounding residues’ hydrophobicity anchors the buoy to the membrane and can potentially pull the buoy into the membrane interior. The ‘sunken-buoy’ model of thermosensing poses that as the membrane thickens upon cooling, the hydrophilic buoy is pulled into the hydrophobic membrane, an energetically unfavorable situation that elicits conformational changes within the DesK protein that increase the activity of its histidine kinase domain (Figure 1D).
Molecular tests to date support the sunken-buoy model. For example, moving a critical lysine residue within the buoy deeper inside the transmembrane domain, which should increase the residue’s burial at all temperatures, increases kinase activity. Conversely, increasing the length of the nonpolar region that lies carboxy-terminal to the buoy, a manipulation that should help unbury its hydrophilic residues, decreases kinase activity. Furthermore, when DesK is incorporated into bilayers of defined lipid content, the longer the fatty acid chains (the thicker the bilayers), the greater DesK’s kinase activity. Together, these studies suggest that DesK regulation is indeed governed by changes in membrane thickness that trigger buoy-dependent conformational changes in DesK.
While the precise structural changes within DesK’s transmembrane region remain uncertain, this model not only provides a compelling scenario for how DesK senses temperature, it links DesK activation to its homeostatic function in maintaining membrane fluidity. In this scenario, as membrane fluidity decreases, membranes thicken, activating DesK and inducing Des expression. Des activity then increases disorder, helping restore appropriate membrane fluidity and thickness. This, in turn, turns off DesK kinase activity and terminates the response. Such a regulatory loop is well-suited to maintaining levels of membrane fluidity within an optimal range. More generally, since many protein thermosensors contain transmembrane domains, such lipid-mediated thermosensing could prove a general theme.
Eukaryotic thermosensors
Among eukaryotes, multiple classes of multi-pass transmembrane ion channels have emerged as key mediators of thermal sensing. These include members of the Transient Receptor Potential (TRP) family of cation channels, the Anoctamin family of chloride channels and the TREK-1 potassium channel. In addition, Drosophila rhodopsin has been suggested as a possible thermosensor. However, it should be noted that while the loss of many molecules (including rhodopsin) can lead to thermosensory behavior defects, most do not act as temperature sensors themselves but rather support the behavior in some other capacity, such as by controlling the development or function of the thermosensory cell or circuitry.
Among eukaryotic thermal sensors, the thermal-sensing properties of TRP channels have been most extensively characterized. TRPs are a large ion channel family, with 27 human members, of which a subset, the ‘thermoTRPs’, exhibit robust thermosensitivity. In animals from flies to humans, thermoTRPs are expressed in a wide range of neuronal and non-neuronal cell types and mediate a host of thermosensory responses, exerting their effects by triggering membrane depolarization and increasing intracellular calcium. Responses mediated by thermoTRPs include the avoidance of excessively cool or warm environments as well as pain and inflammation in mammals. Chemical activation of thermoTRPs also underlies the familiar sensations of heating and cooling that result from the consumption of chili pepper and mint; the former contains capsaicin, which activates the heat-responsive TRPV1 channel, and the latter contains menthol, which activates the cool-responsive TRPM8 channel. ThermoTRPs have also attracted interest as potential mediators of the extreme thermal sensitivity of pit viper ‘heat-vision’ and of vampire bats. However, experiments testing whether these responses require thermoTRP function have not yet been reported.
The molecular basis of thermoTRP temperature responsiveness remains unknown. Importantly, both TRPV1 and TRPM8 function as temperature sensors when purified and incorporated into lipid bilayers, demonstrating that thermoTRPs can function as thermosensors without the need for additional protein co-factors. Biophysical changes to lipids have not been excluded as contributors to thermoTRP activation, but membrane fluidity manipulations to date have not dramatically altered thermoTRP activity. At the protein level, mutations to both cytoplasmic and transmembrane regions of thermoTRPs affect thermosensitivity. For example, a 77 amino-acid region of TRPV1’s cytoplasmic amino terminus confers robust thermosensitivity when transplanted into the channel’s less thermosensitive relatives, suggesting an important role in temperature responsiveness. However, the precise molecular mechanisms by which these various regions affect thermosensitivity remain to be determined.
An interesting possibility for thermoTRP activation, detailed in a recent theoretical examination of thermoTRP function, is that the activation process could resemble temperature-dependent protein unfolding. In this scenario, channel opening is associated with the exposure of previously buried hydrophobic residues, similar to the denaturation of a globular protein. TRPs are six transmembrane domain proteins that form tetrameric ion channels containing thousands of amino acid residues. Thermodynamic estimates suggest that exposure of just 10 to 20 hydrophobic amino acid side chains per thermoTRP subunit could create a sufficient increase in the heat capacity of open versus closed channel states to drive their exceptional thermosensitivity. It will be of interest to determine whether the regions implicated in thermosensitivity to date participate in such conformational changes.
Conclusion
As is evident from the above discussion, biological thermosensors are diverse, reflecting the widespread effects of temperature on all molecules. Some sensors such as RNA thermometers are optimized to rapidly and specifically affect a defined process such as altered translation rates of the relevant mRNA upon a temperature change. Conversely, thermosensors such as DesK and TRP channels alter cellular physiology more globally via alteration of membrane fluidity and excitability, respectively. While the molecular mechanisms by which nucleic acid sensors and some bacterial proteins sense temperature are now fairly well understood, it is clear that our knowledge of exactly how many other thermosensors work is quite poor and will be an important avenue for future investigations. Indeed, in an era of massive climate change, the ability to sense and adapt to environmental temperature perturbations may be a life-or-death challenge for many species and may increase selective pressure on the evolution of biological thermometers.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Contributor Information
Piali Sengupta, Email: sengupta@brandeis.edu.
Paul Garrity, Email: pgarrity@brandeis.edu.
Further reading
- Campbell AL, Naik RR, Sowards L, Stone MO. Biological infrared imaging and sensing. Micron. 2002;33:211–225. doi: 10.1016/s0968-4328(01)00010-5. [DOI] [PubMed] [Google Scholar]
- Cho H, Yang YD, Lee J, Lee B, Kim T, Jang Y, Back SK, Na HS, Harfe BD, Wang F, et al. The calcium-activated chloride channel anoctamin 1 acts as a heat sensor in nociceptive neurons. Nat. Neurosci. 2012;15:1015–1021. doi: 10.1038/nn.3111. [DOI] [PubMed] [Google Scholar]
- Clapham DE, Miller CA. A thermodynamic framework for understanding temperature sensing by transient receptor potential (TRP) channels. Proc. Natl. Acad. Sci. USA. 2011;108:19492–19497. doi: 10.1073/pnas.1117485108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cybulski LE, Martin M, Mansilla MC, Fernandez A, de Mendoza D. Membrane thickness cue for cold sensing in a bacterium. Curr. Biol. 2010;20:1539–1544. doi: 10.1016/j.cub.2010.06.074. [DOI] [PubMed] [Google Scholar]
- Dhaka A, Viswanath V, Patapoutian A. TRP ion channels and temperature sensation. Annu. Rev. Neurosci. 2006;29:135–161. doi: 10.1146/annurev.neuro.29.051605.112958. [DOI] [PubMed] [Google Scholar]
- Klinkert B, Narberhaus F. Microbial thermosensors. Cell. Mol. Life Sci. 2009;66:2661–2676. doi: 10.1007/s00018-009-0041-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kortmann J, Narberhaus F. Bacterial RNA thermometers: molecular zippers and switches. Nat. Rev. Microbiol. 2012;10:255–265. doi: 10.1038/nrmicro2730. [DOI] [PubMed] [Google Scholar]
- Salman H, Libchaber A. A concentration-dependent switch in the bacterial response to temperature. Nat. Cell. Biol. 2007;9:1098–1100. doi: 10.1038/ncb1632. [DOI] [PubMed] [Google Scholar]
- Schumann W. Thermosensor systems in eubacteria. Adv. Exp. Med. Biol. 2012;739:1–16. doi: 10.1007/978-1-4614-1704-0_1. [DOI] [PubMed] [Google Scholar]
- Shapiro RS, Cowen LE. Thermal control of microbial development and virulence: molecular mechanisms of microbial temperature sensing. MBio. 2012;3:e00238-12. doi: 10.1128/mBio.00238-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wu LJ, Sweet TB, Clapham DE. International Union of Basic and Clinical Pharmacology. LXXVI. Current progress in the mammalian TRP ion channel family. Pharmacol. Rev. 2010;62:381–404. doi: 10.1124/pr.110.002725. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yao J, Liu B, Qin F. Modular thermal sensors in temperature-gated transient receptor potential (TRP) channels. Proc. Natl. Acad. Sci. USA. 2011;108:11109–11114. doi: 10.1073/pnas.1105196108. [DOI] [PMC free article] [PubMed] [Google Scholar]

