*For correspondence: Institute of Cell and Molecular Biology, University of Edinburgh, The King’s Building (Botany), Mayfield Road, Edinburgh EH9 3JH, UK (fax +44 131 650 5392; e-mail firstname.lastname@example.org).
Cold elicits an immediate rise in the cytosolic free calcium concentration ([Ca2+]c) of plant cells. We have studied the concerted action of the three underlying mechanisms, namely sensing, sensitisation and desensitisation, which become important when plants in the field are subjected to changes in temperature. We applied different regimes of temperature changes with well-defined cooling rates to intact roots of Arabidopsis thaliana expressing the calcium-indicator, aequorin. Our results indicate that temperature sensing is mainly dependent on the cooling rate, dT/dt, whereas the absolute temperature T is of less importance. Arabidopsis roots were found to be sensitive to cooling rates of less than dT/dt = 0.01°C/s. However, at cooling rates below 0.003°C/s (i.e. cooling 10°C in 1 h) there is no detectable [Ca2+]cresponse at all. At low temperature, the sensitivity of the plant cold-detection system is increased. This in turn produces greater cooling-induced [Ca2+]celevations. Prolonged or repeated cold treatment attenuates the [Ca2+]cresponses to subsequent episodes of cooling.
However, the conditions encountered by plants in the field are quite different. Under natural conditions plants will mostly experience slow cooling rates (i.e. dT/dt < 1°C/s) and hence are likely to respond differently. Thus, it is of value to perform experiments which simulate more realistic situations. Electrophysiological studies have already revealed that ion transport differs depending on the cooling rate (dT/dt) ( Minorsky & Spanswick 1989). It has been hypothesised that plasmalemma associated Ca2+-permeable channels are involved in temperature sensing ( Minorsky 1989; Monroy & Dhindsa 1995; Pickard 1984).
Previous studies have shown that there is a correlation between the [Ca2+]c signal and subsequent cellular events, with [Ca2+]c transients controlling downstream processes. In particular, [Ca2+]c has been implicated in the control of cold-responsive gene expression ( Knight et al. 1996 ; Tähtiharju et al. 1997 ) and in the acquisition of freezing tolerance ( Monroy & Dhindsa 1995; Monroy et al. 1993 ). The expression of cold-responsive genes controlled by calcium can be increased in response to natural temperature changes occurring in the wild. Furthermore, the acquisition of freezing tolerance is mediated by long periods at low temperature (days), and is also affected by natural temperature changes encountered in the wild. Therefore, a role for calcium in mediating responses (gene expression and acclimation) to temperature changes occurring in the field is implied. Less is known about the possible role of Ca2+ during chilling, although there are preliminary data suggesting that Ca2+ may be involved in acclimation to chilling temperatures ( Kitigawa & Yoshizaki 1998)
To understand the role of each facet of these complex processes, it is necessary to first understand the stimulus perception and early signal transduction which eventually leads to the mentioned orchestrated alterations in plants. The sensor for temperature perception has not yet been found. Murata & Los (1997) emphasised the role of membrane fluidity. They speculate that the sensor is located in microdomains of the membrane and able to detect physical phase transitions which then lead to conformational changes and/or phosphorylation dephosphorylation cycles due to changes in temperature. Huner et al. (1998) mentioned the photosynthetic apparatus which might sense changes in temperature through increased energy imbalance and photoinhibition. Monroy & Dhindsa (1995) proposed a calcium permeable channel as primary sensor. Similarly, Minorsky (1989) proposed that the cold-induced [Ca2+]c response was the primary sensing event. Thus, we have measured [Ca2+]c in Arabidopsis roots in response to different temperature change regimes in order to gauge the effects of these regimes on the cold-sensing system in plant cells.
We designed a new experimental set-up to enable long-term measurements of the cytoplasmic free calcium concentration ([Ca2+]c) in whole intact tissues in parallel with temperature (T) measurements. [Ca2+]c was measured in intact root systems of transgenic Arabidopsis expressing the photoprotein aequorin which produces luminescence which is directly related to [Ca2+]c ( Knight et al. 1991 ; Knight et al. 1993 ; Knight & Knight 1995).
The most striking feature of the responses in Fig. 1 is the spike-like [Ca2+]c kinetics which suggests a ‘high-pass filter behaviour’ of the system ( Brook & Wynne 1988; Pearson 1992). Making this assumption implies that the system (i.e. the plant) responds like a differentiator, which means that the cooling rate dT/dt and not the absolute value of the temperature T is the important parameter which determines the response.
In order to give a quantitative basis for this statement, the dependence of the response (the [Ca2+]c-peak value) on the cooling rate is shown in Fig. 2. Here data from 13 different and comparable experiments, where only the response of the very first cooling event is taken into account (i.e. zero cold-history of the specimen) and where the cooling started in the range of 14°C < T < 18.5°C (i.e. similar sensitivity of the specimen), are summarised. The good correlation (r = –0.91) supports our view that the main factor which influences the [Ca2+]c response is dT/dt.
Another interesting feature of the cooling-induced [Ca2+]c response is shown in Fig. 3. In this experiment, temperature was decreased in a stepwise fashion. Again, as in Fig. 1, the spikes in [Ca2+]c demonstrate the role of dT/dt. Consequently, there is no obvious increasing of [Ca2+]c during the horizontal parts of the T-signal (when T is not decreasing but is maintained at a particular value). However, in this experiment we decreased the mean value of temperature. This caused a strong increase in sensitivity (as measured by increasing [Ca2+]c amplitudes with decreasing temperature) which even overrides the attenuation (as measured by decreasing [Ca2+]c amplitudes with time of cold exposure) seen in Fig. 1(a).
The experiment in Fig. 3 also shows that the [Ca2+]c response is almost suppressed when the plant has already had prolonged exposure to low temperatures. The [Ca2+]c responses to a second set of stepwise temperature drops in Fig. 3 confirm that the attenuation mechanism was operative in this experiment. However, a subsequent cooling at a higher cooling rate still elicited the typical [Ca2+]c spike, demonstrating that this attenuation could be overcome by increasing the magnitude of the primary signal (dT/dt).
If a much lower temperature than in Fig. 1 is maintained after initial cooling, the resulting [Ca2+]c response exhibits a quite different biphasic kinetic behaviour ( Fig. 4). The first part of the response is the already described fast [Ca2+]c spike responding to cooling ( Fig. 1 and 3). However, it is then followed by a more prolonged second transient [Ca2+]c elevation. Both phases show attenuation when the cooling procedure is repeated with the same plant ( Fig. 4). This biphasic behaviour can also be seen in Nicotiana plumbaginifolia (data not shown) and therefore is not unique to Arabidopsis.
In order to obtain a clearer picture of the influence of the cooling rate (dT/dt) and the absolute temperature (T) on the kinetics of the [Ca2+]c response, we conducted cooling experiments with a single cooling event (from T0= 18°C down to Tf= 4°C), but with different initial cooling rates ( Nagai & Nakaoka 1998). The experimental results ( Fig. 5a–d) show again that the cooling rate is the main parameter which determines the magnitude and form of the [Ca2+]c increase; with high cooling rates only one single [Ca2+]c peak is observed ( Fig. 5a). Lower cooling rates ( Fig. 5b,c) reveal the biphasic response already shown in Fig. 4. An interesting finding here is that at very low cooling rates the [Ca2+]c trace is lacking the first peak completely ( Fig. 5d) and only the second, prolonged slow response is observed. With extremely low cooling rates (dT/dt < 0.003°C/s, i.e. cooling by 10°C in more than 1 h) no [Ca2+]c response was observed at all ( Fig. 5d– lower trace).
However, we always found that the [Ca2+]c response started almost simultaneously with the onset of cooling. This means that even a drop of dT < 1°C is able to elicit a response as long as a sufficient cooling rate, dT/dt, is provided.
The fact that an increase in cooling sensitivity (i.e. sensitisation as measured by increasing [Ca2+]c amplitudes with decreasing temperature) and attenuation of the [Ca2+]c response (i.e. desensitisation as measured by decreasing [Ca2+]c amplitudes with time of cold exposure) have opposite effects is demonstrated by the experiments shown in Fig. 6. Here, a periodic cycling pattern of cold stimuli was given. From Fig. 1 it would be expected that the amplitude of responses would decrease with time due to attenuation. However, the superimposed steady decrease of the mean temperature T ( Fig. 6c) increases sensitivity and just compensates the effect of attenuation. This leads to permanent [Ca2+]c oscillations where the amplitude of the [Ca2+]c signal ( Fig. 6a) is correlated to the cooling rate amplitude ( Fig. 6b).
Finally, Fig. 7 presents an experimental protocol which clearly demonstrates the operation of the three mechanisms (i.e. sensing, sensitisation and desensitisation) revealed above. Again, a stepwise decrease in temperature was applied as in Fig. 3. However, now the cooling rate (dT/dt) of the steps was increased from step to step ( Fig. 7). This increase in steepness together with the sensitising effect of the decrease in absolute temperature overcompensates attenuation. Thus, the [Ca2+]c trace ( Fig. 7a) appears to be almost a mirror image of the cooling rate ( Fig. 7b) with increasing amplitudes. Concomitant with this increase in amplitude is an increase in the baseline of [Ca2+]c. This is most probably related to the mechanism of sensitisation as it is not observed when only attenuation is active ( Fig. 1a,c). After T has levelled off to a steady value (in Fig. 7 at t = 24 min), [Ca2+]c starts to return to the previous baseline value. This indicates that sensitisation depends on the absolute temperature but that the degree of attenuation increases with the length of time of exposure to cold.
The cooling rate-dependent [Ca2+]c influx
The important feature of the data presented here is the dependence of the [Ca2+]c increase on the cooling rate dT/dt rather than the absolute temperature T. This fact also substantiates the hypothesis of Minorsky (1989) who investigated this phenomenon in cucumber seedlings by means of electrophysiological measurements ( Minorsky & Spanswick 1989). It is also in line with a recent study showing the cooling response of [Ca2+]c in animal cells ( Nagai & Nakaoka 1998). Hence, the first conclusion from our present study is that the Ca2+ influx mediating channels are dependent on the cooling rate, dT/dt, and not on the absolute temperature, T.
Attenuation to cold is induced by prolonged or repeated cold exposure
Another interesting feature in Fig. 1(a) is that the amplitudes of the [Ca2+]c spikes decrease when the temperature stimuli are repeated ( Fig. 1a,c). The dependence on past low temperature episodes that the plant has previously encountered ( Fig. 1a,c) indicate that the plant [Ca2+]c response to cooling can be attenuated ( Knight et al. 1996 ). This attenuation, which can also be termed as ‘desensitisation’, is a property of the time of cold exposure given. This feature can also be seen in Fig. 3. Figure 1(b,d), however, shows that the attenuation of the [Ca2+]c responses can be (over)compensated by an increasing dT/dt in subsequent cold shocks. This finding is also in line with what is known about cold-induced action potentials in Cabomba australis and Cucurbita pepo ( Minorsky 1989).
The overall sensitivity to cold is increased at lower absolute temperatures
Figure 3 provides information about the sensitivity of the temperature-sensing system. With decreasing absolute temperature (T) the responses to cooling are increased although the intensity of the stimulation (i.e. the cooling rate, dT/dt) is the same. Thus, sensitivity is dependent on the absolute temperature, T. As it seems likely that the Ca2+ influx mediating channels are dependent on dT/dt ( Fig. 1 and 2) we conclude that sensitivity acts on another mechanism (i.e. the Ca2+-extrusion mechanism) which counteracts the Ca2+-influx. Ca2+ extrusion is achieved by Ca2+-ATPases ( Briskin 1990; Bush 1995) which have been shown to be strongly dependent on the absolute temperature, T ( Caldwell & Haug 1981). At low temperatures the power of pumps is reduced to a certain fraction of full capacity defined by the Q10 value of the pumps. Thus, the increased response at lower T can be explained in terms of cold-induced inhibition of pump activity.
The biphasic character of the cold [Ca2+]c response
Further information regarding the mechanisms of attenuation and sensitivity is shown in Fig. 4 and 5. On the basis of the findings in Fig. 1 and 3, the time course of the [Ca2+]c responses in Fig. 4 and 5 can be described as follows. The steep temperature slope at the beginning of the cold period causes the initial [Ca2+]c peak which would return to the baseline under the conditions of Fig. 1. However, in Fig. 4 and 5 the horizontal part of the temperature kinetic occurs at a lower temperature. This increases the sensitivity which causes a long-lasting increase in [Ca2+]c observed directly after the initial sharp peak. As the Ca2+-influx is dependent on dT/dt it cannot be active when dT/dt is almost zero and thus cannot be responsible for the second phase. Therefore, a possible explanation for the second phase in the [Ca2+]c kinetic is that the steady-state Ca2+-leak current remains unaffected whereas the pumps become unable to counteract at low temperatures. A new higher [Ca2+]c is thus established. The third mechanism, attenuation, comes into play thus bringing the [Ca2+]c back to the baseline, as demonstrated in Fig. 1 and 3.
At very low cooling rates which plants naturally undergo in the field, the attenuation mechanism is strong enough to suppress any [Ca2+]c increase completely during the cooling process ( Fig. 5d). We therefore assume that the attenuation process is related to an increase in the number of active pumps which may be induced by the elevated Ca2+-influx and which occurs in order to counteract the cold-induced inhibition of pump activity.
Thus, we suggest that attenuation and sensitisation both act on the Ca2+-extrusion mechanism which operates when the Ca2+-influx mechanism is inoperative. (This statement is drawn from Fig. 4 and 5 and is further supported by Fig. 6 and 7).
Cold-induced increases in [Ca2+]c are mainly a function of the cooling rate, dT/dt. Thus, under persistent low temperature, [Ca2+]c returns to normal resting levels.
Sensitivity increases (i.e. sensitisation occurs) with lower absolute temperature, T. Thus, increased [Ca2+]c responses to cooling occur at lower T even when the cooling rate, dT/dt, is the same.
Attenuation (i.e. desensitisation) occurs with increasing time of low T exposure. Thus, the [Ca2+]c responses to cooling decrease with time upon repeated exposure to low temperatures under identical regimes.
Sensitisation and desensitisation operate on the Ca2+-extrusion mechanism whereas the cooling rate operates on the Ca2+-influx mechanism.
All experiments were carried out in unbuffered standard medium (SM) containing KNO3, CaCl2 and MgCl2, 0.1 m m each.
Roots of transgenic Arabidopsis thaliana (biotype RLD1) expressing cytosolic apoaequorin were used for [Ca2+]c measurements ( Knight et al. 1997b ) instead of whole plants because of their higher transparency and lack of air-filled apoplast which would hinder fast and reproducible heat exchange. Plants were grown in 9 cm Petri dishes on vertical 1.2% agar plates with half-strength MS medium ( Gamborg et al. 1968 ; Murashige & Skoog 1962) at 21°C with 16 h photoperiod. After 2 weeks, plants were individually transferred to sterile 50 ml falcon tubes and cultivated hydroponically on 0.5 × MS for another 2–4 weeks under the same conditions.
Preparation of roots
Whole root systems were dissected from plants and used when 4- to 6-weeks-old. For [Ca2+]c measurements reconstitution of aequorin was performed in vivo essentially as described previously ( Knight et al. 1991 ). Briefly, roots were incubated in water containing 2.5 μm coelenterazine (Prolume Ltd, Pittsburgh, PA, USA) in the dark overnight at 21°C in order to allow formation of functional aequorin.
The experiments were performed by placing a root on wet filter paper fixed on a glass microscope slide. The slide with the root was fixed in front of a purpose-built heat exchange coil which could be placed in a 50 ml falcon tube. During the experiments the tube was permanently perfused with fresh SM. The SM was always aerated in order to avoid anoxic stress which can also cause changes in [Ca2+]c ( Sedbrook et al. 1996 ). The pH of the unbuffered aerated SM was always in the range between 5.4 and 5.7.
The tube with the root and heat exchanger was placed in purpose-built light tight sample housing in front of a digital chemiluminometer. Luminescence counts were integrated every 2 sec and calibrated in terms of [Ca2+]c-values as described previously ( Knight et al. 1997b ; Knight & Knight 1995). The temperature was measured in parallel by means of a NTC-resistor which was placed near the roots as described previously ( Plieth & Hansen 1992). The A/D converted thermosensor voltage was fed into a second computer which simultaneously triggered the photon counter. This experimental set-up enabled us to measure temperature (T) with a resolution of less than 0.01°C simultaneously with the calcium-dependent luminescence emitted from the Arabidopsis roots expressing aequorin, with a sample frequency of 0.5 Hz.
Cooling or heating of the roots was performed by drawing ice-cold or pre-warmed water through the heat exchange coil. The rate of temperature changes (dT/dt) was determined by the flux through the coil. Regression analysis was performed using MicroCal Origin vers. 2.8 (MicroCal, Northhampton, MA, USA).
We thank Mr M. Brix (Physics Department Workshop, University Kiel, Kiel, Germany) for manufacturing the luminometer sample housing. This study was supported by the Deutsche Forschungsgemeinschaft (C.P. was funded by PL253/1–1). M.R.K. is a Royal Society University Research Fellow.