Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
•Temperature has a direct effect at the cellular level on an organism. For instance, in the case of biomembranes, cooling causes lipids to lose entropy and pack closely together. Reducing temperature should, in the absence of other factors, increase the viscosity of a lipid membrane. We have investigated the effect of temperature variation on plasma membrane (PM) viscosity.
•We used dispersion tracking of photoactivated green fluorescent protein (GFP) and fluorescence recovery after photobleaching in wild-type and desaturase mutant Arabidopsis thaliana plants along with membrane lipid saturation analysis to monitor the effect of temperature and membrane lipid composition on PM viscosity.
•Plasma membrane viscosity in A. thaliana is negatively correlated with ambient temperature only under constant-temperature conditions. In the more natural environment of temperature cycles, plants actively manage PM viscosity to counteract the direct effects of temperature.
•Plasma membrane viscosity is regulated by altering the proportion of desaturated fatty acids. In cold conditions, cell membranes accumulate desaturated fatty acids, which decreases membrane viscosity and vice versa. Moreover, we show that control of fatty acid desaturase 2 (FAD2)-dependent lipid desaturation is essential for this homeostasis of membrane viscosity. Finally, a lack of FAD2 function results in aberrant temperature responses.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
According to the fluid mosaic model of biomembranes, molecules such as lipids and proteins are constantly in motion and diffuse within the plane of the membrane. The fluidity of the membrane is an intrinsic propriety of the lipid bilayer and protein mobility is crucial for many cellular mechanisms, for example ligand–receptor interactions (Singer & Nicolson, 1972). Two main factors strongly influence membrane fluidity: the temperature and the relative amount of lipid saturation.
Interestingly, the lipid saturation in membranes is adjusted by the plant depending on environmental conditions. Membranes accumulate more desaturated fatty acids in cold than in warm growing conditions (Falcone et al., 2004). Desaturation of lipids makes them less able to tightly pack and, therefore, membranes with relatively higher amounts of fatty acid desaturation should be less viscous. The removal of hydrogen atoms that results in desaturation is done by members of the fatty acid desaturase (FAD) gene family, and the activity of these enzymes is probably inhibited in warm conditions (Burgos et al., 2011). Studies on RNA concentrations of FAD have shown little variation in response to either cold or warm treatment. Recently, however, O’Quin et al. (2010) have described temperature-sensitive degradation of FAD3 which might explain FAD-protein regulation in response to fluctuating temperatures.
In addition to the effect of the degree of saturation of fatty acids, temperature variation affects membrane fluidity directly. As temperature increases, membranes become more fluid (less viscous) and, conversely, a decrease in temperature results in increased viscosity. Because membranes react to temperature changes directly, they are presumed to be involved in temperature sensing. In Bacillus subtilis, the protein DesK is responsible for temperature sensing. DesK, a histidine kinase, has a role in sensing plasma membrane (PM) thickness variation induced by temperature changes. Autophosphorylation of DesK initiates transduction cascades which modify the gene expression profile in response to changing temperature (Cybulski et al., 2010). Other evidence highlights the role of lipid membranes in temperature sensing. Manipulation of lipid composition in yeast (Saccharomyces cerevisiae) and other organisms led to alterations in protein expression similar to those produced by temperature changes (reviewed in Vigh et al., 2005). In plants, study of a variety of model systems has also suggested that lipid membranes of cells may play a role in temperature sensing. For instance, in the plant Brassica napus, hot- or cold-activated protein kinases were up-regulated at ambient temperatures following the application of membrane fluidifying or rigidifying agents, respectively (Sangwan et al., 2002). These examples show that membrane fluidity could be a sensor input for cells.
The relationship between temperature, membrane composition and membrane viscosity has not been investigated in a plant system. Here we have demonstrated that fluorescence dispersal after photoactivation of green fluorescent protein (GFP) is an effective method for monitoring membrane viscosity in vivo. Surprisingly, we have found that when temperature fluctuates in a manner similar to natural conditions, PM viscosity does not decrease as temperature increases – the membrane actually became more viscous in warm conditions than in cold. By studying lipid saturation, we have shown that Arabidopsis compensates for changing temperature by adjusting the amount of fatty acid desaturation. Moreover, impairment in the fatty acid desaturation pathway abolishes the plant’s ability to adjust membrane viscosity and inhibits some physiological responses to temperature changes.
Materials and Methods
Plant materials and growth conditions
All experiments were carried out using the Columbia (Col0) accession of Arabidopsis thaliana (L.) Heynh or homozygous fad2.2 mutants in the Col0 background (described in Larkindale et al., 2005 and obtained from Professor M.R. Knight, Durham University, UK). Seedlings were grown on plates containing 1× Murashige and Skoog (MS) salts (Sigma) with 1% agar (Duchefa biochemie, Haarlem, Netherlands). Media was corrected to pH 5.8. Seeds were surface-sterilized before being stratified in the dark at 4°C for 72 h before transfer to a growth chamber.
Entrainment protocols consisted of warm : cold temperature cycles (WC, 12 h at 27°C followed by 12 h at 12°C) in constant white light (50 μmol m−2 s−1) or light : dark cycles (LD, 12 h of white light followed by 12 h of darkness) at constant 20°C. Temperature cycles were provided by a Percival E30-B chamber (< 5 min ramp time between temperatures), and light cycles and constant light regimes by Sanyo MLR-350 growth chambers.
Production of transformed A. thaliana lines
For photoactivation of a PM marker, photoactivatable GFP (paGFP) was fused to the intrinsic PM protein LTI6b (AtRCI2b, At3g05890). 35s::paGFP-LTI6b was cloned by replacing the DNA sequence of enhanced GFP with that of paGFP between BamH1 and EcoR1 restriction sites in the vector pBIB 35s::EGFP-LTI6b (Kurup et al., 2005). This was accomplished using forward primer 5′-GCT GGA TCC GGT ATG GTG AGC AAG GGC GAG GAG-3′ to add a BamH1 restriction site and reverse primer 5′-AGC GAA TTC TCT CAT CTT GTA CAG CTC GTC CAT GCC-3′ to remove the stop codon and add an EcoR1 restriction site to the paGFP PCR fragment. Lines of Col0 and fad2.2 homozygous for 35s::paGFP-LTI6b were produced by floral dipping (Zhang et al., 2006) followed by selection on antibiotics.
For fluorescence recovery after photobleaching (FRAP) experiments, seedlings expressing 35s::GFP-LTI6b were used (Kurup et al., 2005).
Two biological replicates of the experiment were performed as described previously (Edwards et al., 2005), with each replicate containing eight to 16 plants of each genotype. Following imbibition and stratification, seedlings were grown in LD 12 : 12 h at 22°C before commencement of imaging. Leaf movement rhythms were recorded in constant light (50 μmol m−2 s−1) at the appropriate temperature (12 or 27°C), starting at subjective dawn on day 11 postgermination. Circadian rhythms of leaf movement were analyzed using fast Fourier transform-nonlinear least squares (FFT-NLLS) (Johnson & Frasier, 1985; Straume et al., 1991; Plautz et al., 1997) via the BRASS interface (Brown, 2004). For period estimates, the first 24 h of data were discarded to ensure only the free-running period was measured.
Columbia and fad2.2 seeds were sown on square Petri plates (120 cm) and placed vertically in a growth chamber running the appropriate environmental regime for 7 d. In every case, foil-wrapped plates were included as dark controls. Plates were photographed and hypocotyl lengths measured using ImageJ software. Results are means (± SEM) of two biological replicates, each containing 40 seedlings.
Flowering time experiments were conducted in walk-in constant-environment rooms set to a long-day photoperiod of LD 16 : 8 h or a short-day photoperiod of LD 8 : 16 h at the appropriate temperature (22 or 27°C). Light intensity at the level of the plant was 50–60 μmol m−2 s−1. Following germination on 1× MS plates, 2-wk-old Col0 and fad2.2 seedlings were transferred to sterilized soil. Plants were checked daily until the floral meristem was clearly visible. At this point, plants were dissected to count the total number of rosette leaves. Results are means (± SEM) of 16–20 plants.
Columbia seedlings were grown for 8 d postgermination on 1× MS plates in LD 12 : 12 h or WC 27 : 12°C cycles. Samples were collected at 3 h intervals from dawn (time of lights on (LD) or temperature up (WC); that is, zeitgeber time (ZT) 0) on day 9 and immediately snap-frozen in liquid nitrogen. Analysis of total lipid composition was performed using gas chromatography, as described in Larson & Graham (2001). Results are means (± SEM) of five biological replicates. The degree of saturation of each 18-carbon fatty acid is indicated by the number of double bonds it contains (0, fully saturated; 3, fully desaturated).
In order to compare effects of light and temperature, data collected during the ‘day’ (ZT 3 h, 6 h and 9 h) were pooled and analyzed separately from those collected at ‘night’ (ZT 15 h, 18 h and 21 h). ZT 0 and 12 time points were not included in this analysis as we wished to avoid the transition points between light and dark or warm and cold. The ratio of fully desaturated : saturated fatty acid was calculated by dividing the total amount of fully desaturated fatty acids by the sum total amount of saturated and partially saturated fatty acids.
We recorded the diffusion rate of the intrinsic PM protein LTI6b (AtRCI2b, At3g05890). For this purpose, we fused the protein to paGFP. Quantification of the paGFP-LTI6b diffusion rate was by fluorescence dispersion after photoactivation. Experiments were carried out using a Zeiss LSM 510 META confocal microscope system. Col0 and fad2.2 seedlings were entrained to either LD or WC cycles as described earlier. For photoactivation experiments, sampling was done at 3 h intervals over 24 h. Seven-day-old seedlings expressing 35s::paGFP-LTI6b were used.
Seedlings were immobilized to prevent focus-shift during scanning by mounting them in 1% low-melting-point agarose at room temperature (20°C) and sealing the coverslip with VALAP (Vaseline : lanolin : paraffin wax). At each time point, paGFP-LTI6b was photoactivated in three hypocotyl cells per seedling in four individual seedlings (data shown are mean ± SEM); imaging of the 12 cells at each time point took c. 45 min. This procedure was repeated for each genotype in each of the WC or LD conditions. ‘Day’ (ZT 3 h, 6 h and 9 h) and ‘night’ (ZT 15 h, 18 h, 21 h) time points were again analyzed separately.
Photoactivation of paGFP-LTI6b
For each cell, we focused on the cell’s outer surface so that a circular sheet of membrane became fluorescent once photoactivated. A 63× 1.4 numerical aperture (NA) oil immersion objective was used at a digital zoom setting of 4. Preactivation and postactivation imaging of paGFP was done using a 488 nm argon-ion laser set at 50% output and 4% transmission. Three pre-scan images were made to establish the preactivation intensity of paGFP (generally very low to no fluorescence) and then a circular region of interest, 25 μm2 in area (5.6 μm diameter), was activated by 10 iterations of a 405 nm diode laser set at 100% transmission. Fluorescence dispersion was recorded during the 70 s following photoactivation with a delay of 1.5 s between frames. Images were 256 × 256 pixels and were made with a scan speed of 0.46 s per frame. We confirmed that the energy of the 488 nm laser used to record postactivation data had no bleaching effect on activated paGFP by recording control activated regions for 3 min.
Postactivation fluorescence intensity was measured within targeted regions of interest using the ‘Measure Stack’ function of ImageJ software (http://rsb.info.nih.gov/ij/). These data were normalized to convert values to percentage scales for comparison (see later for specific formulae), and nonlinear regression was used to fit the data so that half-time (t1/2) and curve plateau values were derivable. For photoactivation, curves start at 100% fluorescence intensity and decay exponentially to a plateau that represents the immobile fraction of paGFP-LTI6b. Photoactivation t1/2 is the amount of time required for fluorescence intensity to decay to halfway between 100% and the plateau. For analysis of the photoactivation of paGFP-LTI6b, we followed a previously published protocol (Runions et al., 2006). Data were normalized with the following equation:
where In is the normalized intensity, It is the intensity at any time t, Imin is the mean pre-photoactivation intensity and Imax is the brightest post-photoactivation intensity. Nonlinear regression was used to fit normalized intensity data for the post-photoactivation time period with GraphPad Prism 5 software using the exponentially decreasing equation:
where Y(t) is normalized intensity, A, B, and k are parameters of the curve, and t is time.
Each data set was first fitted independently with this equation to ensure goodness of fit (generally r2 was > 0.95) before a single curve was fitted to all data points of a given treatment by combining data sets and sharing curve parameter values. Half-time (t1/2) from this curve fitting was calculated as t1/2 = 0.69/k.
Finally, the t1/2 value was used for calculating the protein diffusion rate (Axelrod et al., 1976) as D = (0.88 R2)/(4 t1/2), where D is the diffusion rate in μm2 s−1 and R is the radius of the photoactivation spot.
Values of D were compared using either two-tailed t-tests or ANOVA followed by Tukey HSD using Microsoft Excel or SPSS software (SPSS software, IBM corporation, NY, USA).
Fluorescence recovery after photobleaching of EGFP-LTI6b and fluorescein isothiocyanate (FITC)
Fluorescence recovery after photobleaching experiments on EGFP-LTI6b were done using the same magnification and scan settings as described earlier for photoactivation. The radii of photobleached regions were 2.21, 3.48, and 4.43 μm in successive experiments. Pre-bleach and post-bleach scans were made using the 488 nm line of an argon-ion laser set to 50% output and 1% transmission. Bleaching was done using five iterations of the 488 nm line set to 100% transmission. Ten pre-bleach scans were made and fluorescence recovery was recorded during 143 s post-bleaching. For control FRAP experiments on FITC diluted in 100% glycerol, pre-bleach and post-bleach images were made with the 488 nm line of the argon-ion laser set to 50% output and 0.5% transmission. Bleaching of FITC was done within circular regions of radii 2.04, 3.49 and 5.79 μm in successive experiments using a 405 nm diode laser set at 100% transmission for 30 iterations and fluorescence recovery was recorded for 35 s with no time delay between scans. Image resolution was 128 × 128 pixels and scan speed was 0.123 s per frame.
For analysis of the FRAP data for GFP-LTI6b and FITC, the data were normalized using the equation:
where In is the normalized intensity, It is the intensity at any time t, Imin is the minimum post-photobleaching intensity and Imax is the mean pre-photobleaching intensity. Nonlinear regression was used to model the normalized FRAP data. In this case, a two-phase exponential association equation was used:
where Y(t) is normalized intensity, A, B, C, k1 and k2 are parameters of the curve, and t is time. t1/2 was calculated for each individual data set by interpolation at the intensity change midpoint using the curve-fitting equation.
Measurement of membrane viscosity
Direct measurements of membrane viscosity in living tissues are difficult to obtain. To overcome this problem, we relied on the propensity of membrane proteins to diffuse within the lipid bilayer. The speed of protein diffusion within the bilayer is inversely correlated with membrane viscosity. LTI6b, a small intrinsic PM protein, was fused to GFP and stably expressed in A. thaliana (GFP-LTI6b). The fusion protein was localized to the PM as previously reported (Cutler et al., 2000). To test the mobility of this protein in the PM, we used FRAP to estimate the fluorescence intensity recovery half-time and the percentage of recovery.
First, we tested whether GFP-LTI6b diffuses within the PM. In first-order diffusion kinetics (i.e. when other factors such as protein interaction which might limit diffusion are absent), there should exist a linear relation between the half-time of fluorescence recovery and the size of the bleached circular region (Sprague & McNally, 2005). To test this assertion, we did a series of FRAP experiments with different-sized bleaching spots in both a saturated solution of the fluorescent stain FITC in 100% glycerol, a purely diffusive system (Fig. 1a–c), and on GFP-LTI6b-expressing ‘plant cells’ (Fig. 1d–f). In both cases, a linear relation was observed when the half-time of fluorescence recovery was plotted against the area of the bleached spot (Fig. 1g). This means that GFP-LTI6b diffuses without any interactions which might limit or increase the rate expected in first-order diffusion and is, thus, a good candidate for monitoring membrane viscosity.
Second, to ascertain that fluorescence recovery is the result of laterally diffusing protein within the lipid bilayer as opposed to insertion into the membrane within the bleached region via exocytosis, we produced kymograms, that is graphical representations of changing fluorescence intensity along the line marked in Fig. 1(h) during the time course of FRAP experiments (Fig. 1h,i). A triangular shape is clearly formed during the recovery phase (Fig. 1i) demonstrating that recovery of GFP-LTI6b is the result of lateral diffusion from the nonbleached area.
Finally, to optimize measurement of diffusion, we compared two methods of visualizing protein movement: photobleaching and photoactivation. For photoactivation, a photoactivatable form of GFP (paGFP) was fused to LTI6b. Both dispersion of paGFP-LTI6b fluorescence after photoactivation and recovery of GFP-LTI6b fluorescence after photobleaching were recorded (cf. Fig. 1f,j) and analyzed to extrapolate relative diffusion coefficients (Fig. 1k). The standard deviation of the diffusion rate obtained for photodispersion of paGFP-LTI6b was smaller than that for photobleaching GFP-LTI6b (Dphotodispersion = 0.20 ± 0.02, Dphotobleaching = 0.18 ± 0.04). Consequently, we used photoactivation measurements to estimate membrane viscosity. Since LTI6b is free to diffuse, variation in the diffusion rate is the result of a change in membrane viscosity. Consequently, in these experiments, a higher diffusion rate reflects a decrease in PM viscosity, and lower diffusion reflects increased PM viscosity.
Membrane viscosity does not necessarily reflect the external temperature
In nature, plants rarely experience constant conditions of temperature or light. Thus we initially measured PM viscosity in A. thaliana (Col0 accession) seedlings grown under warm : cold cycles in constant light (WC, 27 : 12°C) or under light : dark cycles (LD, 12 : 12 h) at a constant 20°C. Averaged over 24h, the PM was more viscous (i.e. lower paGFP-LTI6b diffusion rate) in WC cycles than in LD cycles (P <0.05; Fig. 2a). Therefore, we compared PM viscosity between light and dark under LD conditions at a constant 20°C, and between warm and cold under WC conditions at constant light. Diffusion during the warm portion of the WC cycles differs from that during the cold portion of the cycle, while no difference was recorded between light and dark conditions (two-tailed t-test, P < 0.01; Figs 2b,c, S1). Surprisingly, however, we found the PM was more viscous during the warm portion of a temperature cycle (Fig. 2b). As this result contradicts earlier reports of decreased viscosity at higher temperatures (e.g. Vaultier et al., 2006), we repeated the experiment using plants grown under constant light and temperature conditions. In this case PM viscosity was indeed lower in plants grown at 27°C than at 12°C (two-tailed t-test, P <0.01; Fig. 2d). Thus PM viscosity reflects ambient temperature only under constant-temperature conditions. Taken together, these results demonstrate that membrane viscosity is not a direct reflection of external temperature but is subject to homeostatic regulation by an unknown mechanism.
The proportion of desaturated fatty acid increases during the cold phase of a temperature cycle
Membrane viscosity changes can be achieved by adjusting the proportion of desaturated fatty acids – a relatively greater amount of desaturated fatty acids equates to decreased viscosity. The degree of polydesaturation varies with growth temperature; as a general rule, it decreases as temperature increases (Marr & Ingraham, 1962; Falcone et al., 2004). Previous studies have been principally concerned with responses to constant temperature: the effects of regular temperature cycles upon lipid composition have not been investigated. Therefore we analyzed fatty acid composition of plants grown under cyclic conditions (Fig. S2).
There was a significant decrease in the proportion of polydesaturated (i.e. 16 : 3 and 18 : 3) fatty acids in plants grown in WC cycles relative to those in LD cycles (Fig. 3a–c), consistent with the greater PM viscosity observed in such plants (Fig. 2a). Closer examination revealed a significant decrease in the proportion of polydesaturated fatty acids during the warm portion of WC cycles (t-test, warm vs cold, P < 0.05; Fig. 3b). This matches the more viscous PM observed at this time (Fig. 2b). By contrast, the overall proportions of saturated and desaturated fatty acids did not differ over a LD cycle (t-test, light vs dark, P =0.06; Fig. 3c), in accordance with more stable PM viscosity in these conditions (Fig. 2c).
Membrane viscosity homeostasis is altered in fad2.2 plants
The observation of a higher proportion of desaturated fatty acids at the time of low PM viscosity (Figs 2b, 3b) suggests a causal connection between the two. To test this, we examined the effect of temperature on PM viscosity in the linolenic acid-deficient fad2 desaturase ethyl methanesulfonate (EMS) mutant (Lemieux et al., 1990). The endoplasmic reticulum (ER)-associated desaturase FAD2 catalyzes the desaturation of 18 : 1 fatty acids to 18 : 2 (Okuley et al., 1994), thus fad2.2 plants have a higher proportion of saturated and partially desaturated fatty acids than wild-type plants (Lemieux et al., 1990; Miquel et al., 1993). We predicted this would produce a more viscous PM.
Consistent with this prediction, the rate of lateral diffusion of paGFP-LTI6b within the fad2.2 PM was equivalent at 27 and 12°C (Fig. 4a) and matched the rate of diffusion in wild-type plants at 12°C. Moreover, PM viscosity of fad2.2 did not differ across a temperature cycle (t-test: P = 0.65; Figs 4b, S3a), indicating it did not adjust membrane properties in response to temperature changes. Thus high PM viscosity was associated with an increased proportion of saturated fatty acids, whether this was achieved by manipulating temperature or genetic background. Similar to wild-type plants, fad2.2 PM viscosity remained constant during LD cycles (t-test, P =0.43; Figs 4c, S3b). Therefore, FAD2-dependent lipid desaturation is essential for the diurnal homeostasis of membrane viscosity under temperature cycles.
fad2.2 plants have a partial temperature insensitivity phenotype
fad2.2 plants are impaired in their PM ‘homeoviscosity’. They cannot compensate for the effect of temperature on membrane viscosity (Fig. 4b). As the PM has been suggested to play a role in temperature sensing in different organisms (Sangwan et al., 2002; Vigh et al., 2005; Cybulski et al., 2010), we wondered if this mutant is correctly responding to temperature. We examined three physiological outputs known, in wild-type plants, to respond predictably to increased temperature.
Elevated temperature resulted in earlier flowering with a reduced number of leaves in wild-type plants growing in LD photoperiods (16 : 8 h at 22 vs 27°C). fad2.2 flowers earlier than Col0 at 22°C under long-day photoperiods (Fig. 5a) and short-day photoperiods (Fig. S4) but showed no earlier flowering response when grown at 27°C under LD conditions. Hypocotyl growth is also clearly related to temperature (Gray et al., 1998). As expected, Col0 had longer hypocotyls when grown at 27°C than when grown at 20°C in constant light, but a similar result was observed for fad2.2. In fact, the relative increase in hypocotyl elongation was slightly more for fad2.2 than for Col0 (1.6 times for Col0 and 2.4 times for fad2.2). Finally, we measured the free-running circadian period (FRP) of leaf movement rhythms, a robust measure of clock behavior. The FRP of the plant clock shortens slightly as ambient temperature increases (Edwards et al., 2005), although circadian clocks are temperature-compensated (Pittendrigh, 1960) across the ambient range. As expected, the FRP of wild-type plants was shorter at high temperature (22.4 h at 27°C vs 26.5 h at 12°C; t-test, P <0.01; Figs 5b, S4a). However, the FRP of fad2.2 was the same at both temperatures (25.8 h at 27°C vs 25.9 h at 12°C; Figs 5b, S4b).
For plants, temperature is, with light, one of the major abiotic stimuli. Indeed, seasonal fluctuation in temperature acts as an input for key physiological processes such as seed germination and floral induction. Understanding how plants sense and respond to temperature is crucial given current trends of climate change and the challenges these pose to agriculture and food security (Battisti & Naylor, 2009). In this study, we have shown a clear correlation among temperature, membrane viscosity and fatty acid desaturation. Our conclusion is that this represents a mechanism whereby plants stabilize membrane viscosity during temperature variation. This homeostasis of membrane viscosity is FAD2-dependent.
LTI6b mobility as a sensor of membrane viscosity
We measured dispersion of paGFP-LTI6b within the PM to monitor membrane viscosity. This small intrinsic protein had high mobility (Fig. 1k) which means that the majority of it is free to diffuse within the PM. To verify that the rate of LTI6b diffusion was not influenced by other cellular factors, we demonstrated that the measured half-time of its recovery after photobleaching is correlated with the size of the bleaching spot (Fig. 1g) and showed that its recovery is centripetal from the margins of the bleached region (Fig. 1i). Consequently, the speed of LTI6b diffusion is inversely related to the viscosity of the membrane (Goodwin et al., 2005). Although this technique is an indirect measurement of viscosity, it allows us to work directly with a living multicellular organism and abolishes artifacts induced by membrane preparation.
Arabidopsis thaliana compensates temperature‘s effect on PM viscosity by altering amounts of fatty acid saturation
Although there are diurnal changes in lipid composition in some plant species, including cotton (Rikin et al., 1993), in Arabidopsis the proportion of fatty acids at any given saturation differs very little between light and dark or warm and cold parts of a cycle (Fig. 3a) (Ekman et al., 2007), suggesting only a minor effect of the diurnal cycle upon lipid composition. Our striking observation that the PM had lower viscosity during the cooler portion of a temperature cycle makes it clear that alterations in viscosity and lipid composition over 24 h do not occur passively in response to changes in external temperature. Our data suggest that plants counteract the direct effects of temperature cycles on membrane viscosity by altering the relative proportion of desaturated fatty acids within the membrane. That this phenomenon results specifically from changes in external temperature, not from a rhythmic environment per se, is evident from the constant state of lipid composition and PM viscosity in plants maintained in LD cycles (Figs 2b, 3b).
fad2.2 plants are not able to regulate PM viscosity in either temperature cycles or under constant temperature (Fig. 4a,b). Consequently, the fatty acid desaturation mechanism, and especially synthesis of highly desaturated fatty acids (18 : 2 and 18 : 3), seems to be important for the regulation of membrane viscosity. fad2.2 mutants are particularly sensitive to low temperature, suffering chronic cell damage and death when held at 6°C, a temperature wild-type plants easily withstand (Miquel et al., 1993). fad2.2 germination is also impaired at low temperature but is indistinguishable from wild-type plants at 20°C (Miquel & Browse, 1994). Mutations in chloroplast-localized desaturases alter thermal tolerance (Kunst et al., 1988; Kodama et al., 1995; Zhang et al., 2005; and see Larkindale et al., 2005) and thus the plastid may be partially responsible for maintaining membrane fluidity (Penfield, 2008). Interestingly, it has recently been shown that the rate of turnover of the FAD3 protein is temperature-sensitive (O’Quin et al., 2010). The half-life of the protein is much longer in cold conditions, probably leading to an accumulation of polydesaturated fatty acids and consequently a less viscous membrane to compensate for the temperature effect. In conclusion, the fatty acid desaturation pathway seems to be important for maintaining a homeostasis of PM fluidity during daily temperature cycles.
Is there a link between membrane viscosity, the FAD pathway and temperature sensing?
Fatty acid synthesis mutants of Neurospora crassa have a long-period circadian phenotype and/or defective temperature compensation (Lakin-Thomas & Brody, 2000; Ruoff & Slewa, 2002). In Arabidopsis, mutation of FAD2 means that the free-running period of leaf movement does not shorten as it does in wild-type plants when grown at 27°C (Fig. 5c). The invariant FRP of fad2.2 grown at different temperatures suggests some impairment of the temperature response mechanism in these plants. This result implies a genetic link between fad2.2 and circadian rhythm. Alternately, fad2.2 flowering time and hypocotyl growth data do not show clear impairment of the temperature response mechanism (Fig. 5a,b). Consequently, it is difficult to conclude a clear temperature insensitivity of fad2.2. It is most likely that several mechanisms interact to ensure accurate temperature sensing by plants. For instance, occupancy of the histone variant H2A.Z in nucleosomes is temperature-sensitive; it drives gene expression and is involved in ambient temperature sensing (Kumar & Wigge, 2010). Further work will be required to determine whether the change in membrane viscosity could be an input for temperature sensing or if it is only a temperature output via the ‘homeoviscosity’ of the PM. In any case, it seems sensible to conclude that homeostasis of PM viscosity is important for all of the membrane processes, such as signaling and secretion, that require association of proteins.
We are grateful to Marc Knight for providing seed for the fad2.2 mutants and Andrew Millar and the IMPS (University of Edinburgh) for access to the leaf imaging system. We would like to thank Andrew Smith, Liam Dolan and Marc Knight for helpful discussions on the manuscript and Pauline White for expert technical assistance. J.R. and A.M. are funded by BBSRC grant number BB/F01407/1. SP and HGM are Royal Society University Research Fellows.