Histidine kinase Hik33 is an important participant in cold-signal transduction in cyanobacteria


  • This paper is dedicated to Dr Marilyn Griffith, who passed away on February 19th, 2005, to recognize her outstanding contribution to research on cold stress in plants.

  • Edited by C. Guy

e-mail: murata@nibb.ac.jp


Acclimation of living organisms to cold stress begins with the perception and transduction of the cold signal. However, traditional methods failed to identify the sensors and transducers of cold stress. Therefore, we combined systematic mutagenesis of potential sensors and transducers with DNA microarray analysis in an attempt to identify these components in the cyanobacterium Synechocystis sp. PCC 6803. We identified histidine kinase Hik33 as a potential cold sensor and found that Hik33 participates in the regulation of the expression of more than 60% of the cold-inducible genes. Further study revealed that Hik33 is also involved in the perception of hyperosmotic stress and salt stress and transduction of the signals. Complexity of responses to cold and other environmental stresses is discussed.

Abbreviations – 

histidine kinase


response regulator.


Decreases in ambient temperature reduce enzymatic activities and ultimately depress various physiological activities. When the temperature changes suddenly and significantly, organisms often fail to survive. When such change is gradual, organisms can acclimate to their environment by sensing the shift in temperature and expressing large numbers of previously unexpressed genes, with resultant synthesis of specific proteins and metabolites that participate in protection against low temperature (Fig. 1). Acclimation begins with the perception of the shift in temperature and transduction of the resultant signal. Organisms and/or individual cells appear to be equipped with sensors and signal transducers that perceive and transduce cold signals. This review focuses on the initial events in cold-inducible gene expression, describing our analysis of potential sensors and transducers of cold stress in cyanobacteria (also reviews by Los and Murata 2002, 2004, and by Mikami and Murata 2003).

Figure 1.

A general scheme for the responses of a cyanobacterial cell to cold stress.

Unicellular cyanobacteria are particularly suitable for studies of stress responses at the molecular level. The general features of their plasma and thylakoid membranes resemble those of the chloroplasts of higher plants in terms of both lipid composition and assembly of membranes. Thus, cyanobacteria appear to provide powerful model systems for studies of the molecular mechanisms of acclimation to low temperature (Murata and Wada 1995).

Some strains of cyanobacteria, such as Synechocystis sp. PCC 6803 (hereafter, Synechocystis), Synechococcus sp. PCC 7942 and Synechococcus sp. PCC 7002, are naturally competent, incorporating foreign DNA that is efficiently integrated into their genomes by homologous recombination (Haselkorn 1991, Williams 1988). Therefore, many researchers have used cyanobacteria for the production of mutants with disrupted genes of interest (for review, see Vermaas 1998).

The Synechocystis genome was sequenced in 1996 (Kaneko et al. 1996) with subsequent publication of the genome sequences of other cyanobacteria (see Murata and Suzuki 2005 for a complete list with references). Genome sequences provide basic information that can be exploited for the genome-wide study of gene expression. A commercially available genome-wide DNA microarray for analysis of gene expression in Synechocystis (Takara Bio Co., Ohtu, Japan) covers 3079 of the 3165 (97%) genes on the chromosome of Synechocystis (99 genes for transposases are excluded from this calculation). There are also 397 genes on plasmids harboured by Synechocystis (Kaneko et al. 2003). These genes are not included in the above-mentioned DNA microarray.

Genome-wide analysis of cold-stress-inducible genes in Synechocystis

We used the DNA microarray from Takara to characterize the genome-wide expression of genes in response to environmental stresses, starting with cold stress (Suzuki et al. 2001). The expression of a large number of genes was enhanced in response to cold stress, while that of another large number of genes was repressed.

Some researchers have inferred that cold stress induces cellular dehydration that is essentially identical to that induced by hyperosmotic stress. To examine whether Synechocystis recognizes these kinds of stress similarly, we compared the effects of cold stress and hyperosmotic stress by DNA microarray analysis (Fig. 2; Mikami et al. 2002). These two kinds of stress enhanced, in common, the expression of genes for high light-inducible proteins (hliA, hliB, and hliC), for rare lipoprotein A (rlpA), for DNA mismatch repair protein (mutS), for a sigma factor (sigD), and for proteins with other functions. However, only cold stress enhanced the expression of the rbpA gene for an RNA-binding protein, the ndhD2 gene for NADH dehydrogenase subunit 4, the crhR gene for an ATP-dependent RNA helicase, the fus gene for translation elongation factor EF-G, the feoB gene for ferrous iron transport protein, the infB gene for translation initiation factor IF-2, and various genes for proteins of known and unknown function. By contrast, only hyperosmotic stress enhanced the expression of genes for heat-shock proteins (hspA, dnaK2, groEL2, and clpB1), for the synthesis of glucosylglycerol (ggpS and glpD), for the synthesis of lipids and lipoproteins (fabG and repA), and for various other proteins (htrA, spkH, sodB, htpG, and gloA). Thus, cold and hyperosmotic stress each induced expression of a number of specific genes, while both stresses induced expression of a relatively small number of common genes (Fig. 2), suggesting that Synechocystis recognizes cold stress and hyperosmotic stress as different signals via specific signal-transduction pathways.

Figure 2.

Some cold-stress-inducible genes and hyperosmotic stress-inducible genes are the same, and some are different. The diagram includes genes that are induced during incubation of Synechocystis for 20 min after a shift in growth temperature from 34° C to 22° C (cold stress) and after addition of sorbitol at 0.5 m (hyperosmotic stress). Adapted, with permission, from Mikami et al. (2002) with inclusion of recent results from authors' laboratory.

Two-component systems: positive regulation and negative regulation

The existence of two-component systems has been well-established in Escherichia coli and Bacillus subtilis (Aguilar et al. 2001, Stock et al. 2000). Each two-component system consists of a histidine kinase (Hik) and a cognate response regulator (Rre). The Hik perceives a change in the environment via its sensor domain, and then a conserved histidine residue within the Hik domain is autophosphorylated, with ATP providing phosphate group (Stock et al. 2000) that is transferred from the Hik to a conserved aspartate residue in the receiver domain of the cognate Rre. Upon phosphorylation, the conformation of the Rre changes, allowing the binding of the Rre to promoter regions of genes that are located downstream in the acclimation pathway (Koretke et al. 2000).

In E. coli and B. subtilis, the genes for each Hik and its cognate Rre are generally located close to one another on the chromosome. Moreover, these genes are often located near functionally related genes.

Two-component systems are found also in cyanobacteria (Mizuno et al. 1996). However, genes for Hiks and Rres in Synechocystis are, in most cases, distributed somewhat randomly on the chromosome. Among the 44 genes for Hiks, 14 are located near genes for potentially cognate Rres, whereas genes for the other 30 Hiks are located far from any rre genes. Thus, it is difficult to predict the pairs of Hiks and Rres that might function as individual two-component systems. Therefore, we systematically mutated the genes for all the potential sensors and transducers of environmental signals and examined gene expression under various conditions using DNA microarrays.

Synechocystis has 3661 putative genes, of which 47 encode Hiks and 45 encode Rres (http://www.kazusa.or.jp/cyanobase/Synechocystis/index. html). There are 44 genes for Hiks on the chromosome, which we named hik1 through hik44. We deduced that three putative Hiks, namely, Hik11, Hik17, and Hik37, might be inactive as Hiks, because they lack the conserved histidine residue in the Hik domain. Hik32 might also be inactive as a Hik, because the hik32 (sll1473) gene is part of a larger gene, namely, sll1473-sll1475, which is interrupted by a transposon (sll1474) in the strain that was used for genome sequencing and systematic mutagenesis (Okamoto et al. 1999). The gene for the Hik encoded by pSYSM, a plasmid harboured by Synechocystis, was designated hik45, and the genes for Hiks encoded by pSYSX, another plasmid, were designated hik46 and hik47. There are 42 chromosomal genes for Rres and two and one on plasmids pSYSX and pSYSM, respectively. We designated the 42 chromosomal genes for rre1 through rre42. The rre genes on pSYSM and pSYSX were designated rre43 and rre44 and 45, respectively.

We inactivated each putative hik gene in Synechocystis by inserting a spectinomycin-resistance gene cassette to create a gene-knockout library (Suzuki et al. 2000; CyanoMutants, http://www.kazusa.or.jp/cyano/Synechocystis/mutants/). This library has proved to be a powerful tool for the identification of sensors of various stimuli and the corresponding signal transducers in Synechocystis (Marin et al. 2003, Paithoonrangsarid et al. 2004, Shoumskaya et al. 2005, Suzuki et al. 2000, Suzuki et al. 2004, Yamaguchi et al. 2002). We examined the genome-wide expression of genes in wild-type cells and in each line of hik mutant cells with DNA microarrays in an attempt to identify the Hiks involved in the regulation of expression of stress-inducible genes.

Regulation of gene expression in response to stress can be positive or negative (for details, see Murata and Suzuki 2005). In positive regulation, a two-component system is inactive under non-stress conditions. Genes controlled by such systems are silent or expressed. When cells are exposed to the appropriate environmental stress, the two-component system is activated (by phosphorylation in response to the stress), and then the activated system enhances the expression of genes that are silent under non-stress conditions or represses the expression of genes that are expressed under non-stress conditions (Murata and Suzuki 2005). Most stress-inducible gene expression in Synechocystis is positively regulated.

In negative regulation, the two-component system is active under non-stress conditions, and the expression of the genes controlled by such a system is either enhanced or repressed. Under appropriate environmental stress, the two-component system becomes inactive. Genes that are expressed or repressed under non-stress conditions become silent or are released from repression, respectively. Levels of expression of genes fall in the former case and rise in the latter.

Changes in phenotype due to mutations in Hiks and Rres, which reflect the effects of such two-component systems on gene expression, differ between positive regulation and negative regulation. A knockout mutation in either the Hik or the Rre in a two-component system for negative regulation has marked effects on gene expression. The expression of genes controlled by a negatively regulating two-component system is either enhanced or repressed under non-stress conditions. Therefore, a specific signal-transduction pathway with a specific Hik and a specific Rre can be identified with relative ease in cases of negative regulation.

By contrast, a knockout mutation in a Hik or an Rre in a two-component system that operates via positive regulation does not have a significant effect on gene expression under non-stress conditions. In such a system, the identification of the Hik and the Rre in a specific signal-transduction pathway requires the screening of knockout libraries of hik and rre genes under individual types of stress.

Negative regulation of gene expression

Using DNA microarray, we analyzed the effects of mutation of each hik gene on gene expression in mutant cells that had been grown under normal conditions. No significant alterations in gene expression were evident in most of the mutants. However, three mutants, Δhik27,Δhik34, and Δhik20, each with a mutation in the indicated Hik, were unique, because, in these mutants, the expression of some genes was enhanced and that of some others was repressed, indicating that Hik27, Hik34, and Hik20 might be involved in signal transduction via the negative regulation of gene expression.

We compared gene expression in Δhik27 cells with that in wild-type cells under normal conditions (Yamaguchi et al. 2002). Marked changes, with induction factors higher than 10, due to mutation of the hik27 gene (slr0640) were recognized only in the expression of three genes, namely, mntC, mntA, and mntB, which constitute the mntCAB operon that encodes subunits of the ABC-type Mn2+ transporter (Bartsevich and Pakrasi 1995, 1996). Under normal growth conditions, Hik27 might transduce a signal that represses the expression of the mntCAB operon. Moreover, disappearance of this signal, due to inactivation of Hik27, might allow the expression of the mntCAB operon. This scenario corresponds to the scheme outlined for negative regulation. We examined the effects of mutation of various rre genes on gene expression in mutant cells. The mutation in Δrre16 cells enhanced the expression of only the mntC, mntA, and mntB genes. This phenomenon was very similar to the change in gene expression detected in the Δhik27 mutant. In its active form, Rre16 might have repressed expression of the mntCAB operon. Then, inactivation of Rre16 in Δrre16-mutant cells eliminated the repressive effect of Rre16, allowing the expression of the mntCAB operon. These findings suggested that Hik27 and Rre16 might constitute a two-component system, acting as the sensor and signal transducer of Mn2+ deficiency. Ogawa et al. (2002) identified this two-component system independently by ‘traditional’ methods.

Mutation of Hik34 altered the genome-wide expression of genes under normal conditions, namely, at a growth temperature of 34° C (Suzuki et al. 2005). In ΔHik34 cells, levels of transcripts of heat-shock genes, such as htpG, groES, and groEL1, were elevated, suggesting that Hik34 might act as a negative regulator of the expression of these genes during normal growth. Because Hik34 appeared to be a negative regulator of heat-shock-responsive genes, we postulated that its overexpression should result in repressed expression of these genes. We observed that the effect of overexpression of hik34 was the opposite of that of inactivation of hik34 on the expression of heat-shock genes at the normal growth temperature (Suzuki et al. 2005). This observation is consistent with the hypothesis that Hik34 is important in the regulation of expression of heat-shock genes.

Positive regulation by Hik33 in response to low temperature

Mutation of the 42 other hik genes did not induces significant changes in gene expression. Therefore, it is likely that the Hiks encoded by these 42 genes regulate gene expression in a positive manner. To confirm this hypothesis, we screened our library of hik mutants under specific kinds of stress.

To monitor the inducibility of the cold-inducible desB gene for the ω3 desaturase, we generated the pdesB::lux strain of Synechocystis, in which the promoter region of the desB gene was ligated to the coding region of the luxAB gene for a bacterial luciferase (Suzuki et al. 2000). Thus, we could use luciferase activity as an indicator of cold-inducible changes in the activity of the desB promoter. Then we inactivated separately the gene for each Hik in pdesB::lux cells by inserting a spectinomycin-resistance cartridge (Spr ), creating a gene-knockout library (Suzuki et al. 2000). We screened the members of this library for loss of cold inducibility of gene expression by monitoring luciferase activity at a low temperature. Only pdesB::lux/ΔHik33 and pdesB::lux/ΔHik19, with disruption of the genes for Hik33 and Hik19, respectively, exhibited reduced ability to activate luciferase at low temperature, suggesting that Hik33 and Hik19 might be involved in the perception and tranduction of cold signals.

DNA microarray analysis of Δhik33 cells indicated that Hik33 regulates the expression of 21 of 36 cold-inducible genes, with induction factors higher than 3.0 (Fig. 3; Suzuki et al. 2001). These 21 genes include ndhD2, hliA, hliB, hliC, feoB, crp, and genes for proteins of unknown function. By contrast, 15 of the 36 cold-inducible genes were not regulated by Hik33. Therefore, we can deduce that Synechocystis might have at least one other pathway for cold-signal transduction. DNA microarray analysis of the expression of genes in Δhik19 cells indicated that Hik19 is unlikely to be a cold sensor.

Figure 3.

Hypothetical schemes showing the two-component systems that are involved in the transduction of cold stress and hyperosmotic stress, as well as the genes that are under the control of the individual two-component systems. Primary signals are represented by open arrows. Hiks are indicated as ellipses, Rres are indicated as hexagons, and selectively regulated genes are shown in boxes. Uncharacterized mechanisms are represented by question marks. Genes with induction factors higher than 3.0 are included in these schemes. Minor discrepancies with respect to cold-inducible genes between this figure and Fig. 2 are due to the use of different versions of the DNA microarray in the respective experiments. (A) Cold stress; adapted originally from Suzuki et al. (2001) and Mikami et al. (2002), with inclusion of more recent results from authors' laboratory. (B) Hyperosmotic stress; adapted from Paithoonrangsarid et al. (2004) and Shoumskaya et al. (2005).

To identify the Rre that is located downstream of Hik33, we screened an Rre-knockout library by RNA slot-blot hybridization using some of the cold-inducible genes controlled by Hik33 as probes. We identified Rre26 as a candidate for the Rre that, with Hik33, constitutes a two-component system for cold-signal transduction (our unpublished results; Fig. 3A).

Tu et al. (2004) and Hsiao et al. (2004) postulated that Hik33 might negatively regulate the expression of a set of photosynthetic and high light-responsive genes. Their hypothesis was based on changes in the global expression of genes under normal conditions and the complementation of the Δhik33 mutation in Synechocystis by the homologous nblS gene from Synechococcus with respect to the light-induced expression of hli genes, as monitored by Northern blotting. However, because the complementation test did not examine the genome-wide expression of genes, it is possible that their ΔHik33 mutant cells had, in addition to the mutation in hik33, a further mutation that might have produced changes in gene expression under normal conditions.

To confirm our conclusion that Hik33 is a positive regulator of signal transduction, we replaced the entire open-reading frame of the hik33 gene with a spectinomycin-resistance gene cassette that contained the Ω sequence, which is a strong terminator of transcription and inhibits the read-through of inserted genes on both sides of the cassette (our unpublished work). The growth rate of cells with deletion of hik33 was similar to that of the wild-type and the insertion mutant of Hik33, contradicting the inferences made by Hsiao et al. (2004). DNA microarray analysis of the deletion mutant, comparing the pattern of gene expression with that of the insertion mutant under normal conditions, confirmed the absence of the major changes in gene expression reported by Hsiao et al. (2004). Our results suggest that the Hik33-dependent-signalling pathway involves the positive regulation of gene expression.

Responses to salt stress and hyperosmotic stress

We demonstrated previously that the perception of hyperosmotic stress involves Hik33 and other unknown components (Mikami et al. 2002). Later when we used the DNA microarray to investigate the impact of mutations in each Hik on changes in gene expression under salt stress (Marin et al. 2003), we found that the inducibility of gene expression by elevated levels of NaCl was significantly affected in ΔHik33-, ΔHik34-, ΔHik16-, and ΔHik41-mutant cells. In each of these mutants, several genes were no longer induced by salt, or their inducibility by salt was markedly reduced.

Because hyperosmotic stress and salt stress might affect some aspects of the physiology of cyanobacterial cells similarly, we examined whether these Hiks might also regulate gene expression under hyperosmotic stress. DNA microarray analysis demonstrated that they are indeed involved in hyperosmotic-signal transduction. Fig. 3B shows that they regulate the expression of 38 of 52 hyperosmotic stress-inducible genes (Paithoonrangsarid et al. 2004).

Screening our Rre-knockout library by RNA slot-blot hybridization and with DNA microarray, we identified four Rres, namely, Rre31, Rre1, Rre3, and Rre17, that are involved in hyperosmotic signal transduction. Further analysis with microarray showed that four Hik-Rre systems, namely, Hik33-Rre31, Hik34-Rre1, Hik10-Rre3, and Hik16-Hik41-Rre17, as well as another system that included Rre1 and possibly Hik2, appeared to be involved in the perception of hyperosmotic stress and transduction of the signal (Paithoonrangsarid et al. 2004). Fig. 3B shows a hypothetical model of the hyperosmotic signal-transducing systems that involve these Hiks and Rres, including the hyperosmotic stress-inducible genes that are controlled by the individual Hik-Rre systems.

The Hik33-Rre31 two-component system regulates the expression of 11 hyperosmotic stress-inducible genes. Inactivation of either Hik33 or Rre31 resulted in the elimination of or a marked reduction in the hyperosmotic stress-inducible expression of these genes, indicating that Hik33 and Rre31 are tightly coupled in the signal-transduction pathway. The Hik10-Rre3 two-component system regulates the hyperosmotic stress-inducible expression of only htrA, which encodes a serine protease.

The Hik16-Hik41-Rre17 system regulates the hyperosmotic stress-inducible expression of sll0939 and slr0967. Inactivation of Hik16, of Hik41, or of Rre17 eliminated the expression of these genes, suggesting that Hik16, Hik41, and Rre17 are all essential for the perception of hyperosmotic stress and for transduction of the corresponding signal. Hik41 probably acts downstream of Hik16, because Hik41 is a hybrid-type Hik that contains both a signal-receiver domain and a Hik domain, whereas Hik16 is a typical Hik with a Hik domain and potential sensory domain that, hypothetically, spans the membrane seven times. It is also possible that Hik16 and Hik41 might perceive hyperosmotic stress as a complex.

The Hik34-Rre1 system regulates the expression of 19 hyperosmotic stress-inducible genes (for heat-shock proteins and for proteins of unknown function). The Hik2-Rre1 system regulates the expression of five genes that include the sigB gene for a sigma factor. Rre1 might perceive hyperosmotic signals from both Hik34 and Hik2. However, the regulated genes are specific to either His34 or Hik2 (Fig. 3B; Paithoonrangsarid et al. 2004).

DNA microarray analysis revealed that expression of 14 of the 52 hyperosmotic stress-inducible genes was not controlled by any of the five Hiks and four Rres discussed above (Fig. 3B). The signals, due to hyperosmotic stress, that induce the expression of these genes are probably perceived by unknown mechanisms that differ from typical Hik-Rre two-component systems. Such signals might act directly to regulate either transcription or the stability of the transcripts of these inducible genes.

Using similar methods, we identified that five two-component systems are involved in the salt-stress signal-transduction pathway. To our surprise, they were identical to those involved in response to hyperosmotic stress. However, the genes controlled by the individual pathways are different (Shoumskaya et al. 2005), as discussed below.

His33 is a multifunctional regulator

As described above, Hik33 is involved in the perception and transduction of the cold signal and the hyperosmotic-stress signal (Fig. 3). However, the cold-responsive genes controlled by Hik33 are not identical to the hyperosmotic stress-responsive genes controlled by the same system. Fig. 4 shows that eight genes, including hliA, hliB, hliC, and sigD, are induced by both kinds of stress under the control of Hik33. However, Hik33 regulates the expression of eight other genes, including ndhD2, fus, crtP, and feoB, in response to cold stress specifically, and not to hyperosmotic stress, whereas it regulates the expression of three other genes, including fabG and gloA, in response to hyperosmotic stress specifically, and not to cold stress. Thus, cold stress, hyperosmotic stress, and salt stress are perceived as distinct signals by a sensory system that includes Hik33. In addition, recent studies (Hsiao et al. 2004, Tu et al. 2004) suggest that Hik33 might be involved in response to light stress. Moreover, a homolog of Hik33 in Synechococcus is involved in sensing nutritional deficits (van Waasbergen et al. 2002).

Figure 4.

A schematic representation of genes that are induced by cold stress and hyperosmotic stress under control of Hik33. The diagram includes genes that are induced during incubation for 20 min after a shift in growth temperature from 34° C to 22° C (cold stress) and after addition of sorbitol at 0.5 m (hyperosmotic stress).

Another level of compexity reflects the differential involvement of Rres in signal transduction and gene expression. Microarray analysis indicated that Rre26 is involved in cold-signal tranduction (Fig. 3A), whereas Rre31 is involved in hyperosmotic- and salt-signal transduction (Fig. 3B; Paithoonrangsarid et al. 2004, Shoumskaya et al. 2005). Moreover, eight genes respond, in common, to cold stress and hyperosmotic stress via Rre26 and Rre31, respectively. Our observations suggest that the mechanisms for perception of cold stress and hyperosmotic stress by Hik33 are complex and that as-yet-unidentified components exist that provide the sensory systems with their respective activities for induction of responses specific to each individual type of stress.

Hik33 perceives the cold signal via rigidification of membrane lipids

The Hik33 sensory kinase includes a type-P linker, a leucine zipper, and a PAS domain (Fig. 5A; Los and Murata 1999, 2002, 2004, Mikami and Murata 2003). The type-P linker contains two helical regions in tandem that are assumed to transduce stress signals via intramolecular structural changes that result from interactions between the two helical regions and lead to intermolecular dimerization of membrane proteins (Aravind et al. 2003, Williams and Stewart 1999). In Hik33, cold stress might promote a conformational change in the type-P linker, with subsequent activation of Hik33 via dimerization of the protein (Fig. 5B; Los and Murata 2000, 2004).

Figure 5.

A hypothetical scheme for the structure of Hik33. (A) Domain structure of Hik33. (B) A putative dimeric form of Hik33 and its relationship to the cell membrane. A decrease in temperature rigidifies the membrane, leading to compression of the lipid bilayer, which forces the membrane-spanning domains closer changes the linker conformation and finally causes dimerization and autophosphorylation of histidine kinase domains. TM1 and TM2, Transmembrane domains; LZ, leucine zipper; Type-P, a type-P linker domain; and PAS, PAS domain that contains amino acid motifs Per, Arnt, Sim, and phytochrome (Taylor and Zhulin, 1999); H in circles, histidine residues that can be phosphorylated in response to cold stress.

There are two transmembrane domains in the amino-terminal region of Hik33 (Los and Murata 2004, Mikami and Murata 2003). Because it has been suggested that changes in membrane fluidity might be involved in the sensing of temperature (Los and Murata 2000, Murata and Los 1997), it is likely that the transmembrane domains of Hik33 can recognize changes in membrane fluidity at low temperatures (Los and Murata 1999, 2000, 2004).

In 1996, we produced a series of mutants in which the extent of unsaturation of fatty acids is modified in a step-wise manner (Tasaka et al. 1996). In one such mutant, the desA and desD genes for the Δ12 and Δ6 fatty acid desaturases, respectively, are inactive as a result of targeted mutagenesis. The desA/desD-double mutant synthesize only a saturated C16 fatty acid and a mono-unsaturated C18 fatty acid, regardless of growth temperature, whereas wild-type cells synthesize di-unsaturated and tri-unsaturated C18 fatty acids in addition to the mono-unsaturated C18 fatty acid (Tasaka et al. 1996). FTIR spectrometry revealed that the double mutation of the desA and desD genes rigidified the plasma membrane of Synechocystis at physiological temperatures (Szalontai et al. 2000).

Using microarray, we examined the effects of the above-described membrane rigidification on the expression of genes at low temperature in Synechocystis (Inaba et al. 2003). We monitored changes in gene expression in wild-type and desA/desD cells after growth at 34° C and subsequent incubation at 22° C for 30 min. In wild-type cells, cold stress more than doubled the levels of expression of 168 genes. In desA/desD cells, in addition to the enhanced expression of these same 168 cold-inducible genes, we observed enhanced expression of 96 additional genes. Thus, rigidification of membrane lipids apparently enhanced the response of gene expression to low temperature in Synechocystis. By contrast, under isothermal conditions, the double mutation had no significant effect on gene expression.

We divided cold-inducible genes into three groups according to the effects of the double mutation (Inaba et al. 2003). The first group included genes that were not induced by low temperature in wild-type cells but were strongly induced by low temperature in desA/desD cells. The second group included genes whose low-temperature inducibility was moderately enhanced by the double mutation. The third group included genes whose inducibility by low temperature was unaffected by the double mutation. The response to low temperature of the expression of the genes in these three groups might be regulated by different mechanisms with respect to membrane rigidity. Induction of expression of the genes in the first group might require greater rigidification of membrane lipids than the low-temperature responses of genes in the second and third groups. The rigidification of membrane lipids did not enhance the cold inducibility of genes in the third group, perhaps because the rigidity of membranes in wild-type cells is sufficient at low temperatures for the maximum induction of expression of these genes.

To examine whether Hik33 might regulate the cold-responsive gene expression that depends on membrane rigidity, we examined genome-wide gene expression in desA/desD/hik33 cells, in which the hik33 gene had been mutated in addition to mutation of the desA and desD genes. Mutation of Hik33 abolished or significantly reduced the inducibility by low temperature of 10 of the 17 genes in the second group and of seven of 25 genes in the third group. By contrast, mutation of Hik33 had no significant effect on the low-temperature inducibility of genes in the first group. These results indicate that Hik33 regulates the expression of many genes in the second and third groups. They also suggest that the activity of Hik33 in the sensing of low temperature depends on membrane rigidity and that there are at least two other cold sensors, one of which depends on membrane rigidification, while the other functions independently of membrane rigidity (Inaba et al. 2003).

Cold sensors and cold-signal transducers in other organisms

Aguilar et al. (2001) identified DesK of Bacillus subtilis as a cold-sensing Hik and DesR as the cognate Rre that regulate the cold-inducible expression of the des gene for Δ5 desaturase. The desK and desR genes form an operon on the genome of B. subtilis.

DesK is a membrane-bound protein with four transmembrane domains and a Hik domain. However, in contrast to Hik33, DesK lacks the PAS and leucine zipper domains. DesK is a bifunctional enzyme with kinase and phosphatase activities. It has been suggested that DesK is involved in two signalling reactions: phosphorylation in response to membrane rigidification and dephosphorylation in response to membrane fluidization (Mansilla et al. 2004). In fact, the carboxy-terminal portion of DesK (DesKC) acts as an autokinase as well as a phosphatase; the phosphoryl group of phosphorylated DesKC is transferred to DesR. The resultant phosphorylated DesR can be dephosphorylated in the presence of DesKC in vitro. These findings suggest that DesK has the ability to modify DesR through both its kinase and its phosphatase activities, depending on the physical state of the membrane. It is likely but, as yet, unproved that transmembrane segments of DesK sense changes in membrane fluidity due to changes in temperature (Albanesi et al. 2004).

DesR binds specifically to the promoter region of the des for the Δ5 desaturase. Induction of expression of des in B. subtilis by the DesK-DesR two-component system is inhibited by exogenous unsaturated fatty acids or isoleucine (Aguilar et al. 2001; Cybulski et al. 2002), suggesting the presence of a feedback loop between the function of the sensor and the extent of fatty acid unsaturation. Cybulski et al. (2004) demonstrated that DesK, DesR, and the promoter region of the des gene interact directly with one another. The dephosphorylated form of DesR is unable to bind to a regulatory region of the des gene. DesK phosphorylates dimeric DesR, which becomes a tetramer, and binds upstream of the promoter of the des gene in a sequence-specific manner, with activation of des through recruitment of RNA polymerase to the promoter. Thus, the DesK-DesR two-component system regulates the expression of cold-inducible des, allowing the cell to optimize the fluidity of membrane phospholipids (Cybulski et al. 2004, Mansilla and de Mendoza 2005).

The DesK-DesR system regulates the cold-inducible expression of the des gene but of no other genes. By contrast, the cyanobacterial Hik33 sensor regulates the expression of more than 50 cold-inducible genes (Mikami et al. 2002, Suzuki et al. 2001).

In plants, the discovery of the cold-regulation pathway that involves CBF/DREB led to further progress in the characterization of cold-signal transduction (reviews by Guy 1999, Thomashow 1999, Yamaguchi-Shinozaki and Shinozaki 2005, Xiong et al. 2002). Analysis of the transcriptional control of two cold-inducible genes (rd29A and cor15a) in Arabidopsis thaliana led to the identification of a cold-responsive element, the CRT/DRE [(C-repeat)/(dehydration responsive element)], in their promoters (Shinozaki and Yamaguchi-Shinozaki 2000). Members of a family of AP2-domain-transcription factors, namely, DREB1 (DRE-binding protein) and CBF (CRT-binding factor), bind to the CRT/DRE element and activate transcription. Expression of genes for these transcription factors is rapidly induced on cold treatment of plants. Moreover, at normal temperatures, overexpression of CBF1, CBF2, or CBF3 in transgenic A. thaliana enhanced the expression of 41 genes, 30 of which had been identified as cold-inducible genes in wild-type plants (Fowler and Thomashow 2003). Thus, CBF transcription factors might regulate the expression not only of cold-inducible genes but also of genes whose expression is induced by other signals. By contrast, some cold-inducible genes do not appear to be controlled via the CBF pathway. Thus, other regulatory systems might exist for the cold-inducible regulation of approximately 60 genes (Fowler and Thomashow 2003).

Despite the accumulation of important results about the cold regulation of gene expression, little is known about the temperature sensors in plants. Sensors and transducers of cold signals in plants remain to be identified.


Genome-based systematic analysis is a powerful technique for the identification of Hiks and Rres that are involved in the perception and transduction of cold signals and other kinds of stress signals. Using this approach, we showed, with relative ease, that Hik33 regulates the expression of most of the cold-inducible genes in Synechocystis, that membrane rigidification is intimately involved in the sensing of low temperature, and that Hik33 is also involved in the perception and transduction of other types of stress. We must now determine how the same Hik can perceive and transduce more than one kind of environmental signal and regulate the expression of different sets of genes (Los and Murata 2002, 2004, Mikami et al. 2002, Shoumskaya et al. 2005). Our findings cannot be explained by the current model of two-component systems, in which a Hik perceives a specific signal and regulates the expression of a particular set of genes via the phosphorylation-dependent activation (or inactivation) of a cognate Rre. It is likely that as-yet-unidentified components are important in determining the specificity of the responses to individual types of stress. It is also possible that the sensors of environmental signals are highly organized protein complexes, in which Hiks, Rres, and various unidentified components are somehow associated. To identify these components, we shall have to develop new techniques, which, most probably, will also exploit the information encoded in the genomes of cyanobacteria and other organisms.

Acknowledgements –  This work was supported by a Grant-in-Aid for Scientific Research on Priority Areas (no. 14086207) from the Ministry of Education, Science, Sports and Culture of Japan to N.M. It was also supported by grants from the Russian Foundation for Basic Research (nos. 03-04-48581 and 05-04-50883) and by a grant from the ‘Molecular and Cell Biology Program’ of the Russian Academy of Sciences to D.A.L.