Tài liệu Báo cáo khoa học: A systems biology approach for the analysis of carbohydrate dynamics during acclimation to low temperature in Arabidopsis thaliana doc

carbon metabolism, with a shift in partitioning ofnewly fixed carbon into sucrose rather than starch [9,10], indicating cold-induced selective stimulation of sucrose synthesis, which coul

carbohydrate dynamics during acclimation to low

temperature in Arabidopsis thaliana

Thomas Na¨gele, Benjamin A Kandel*, Sabine Frana*, Meike Meißner and Arnd G Heyer

Biologisches Institut, Abteilung Pflanzenbiotechnologie, Universita¨t Stuttgart, Germany

Introduction

Low temperature is an important environmental factor

affecting plant growth, and constraining crop

produc-tivity and species distribution [1,2] Whereas many

tropical and subtropical species have only limited

capacities to cope with low temperature, plants from

temperate climates, such as Arabidopsis thaliana, grow

at low temperature and can increase their freezing

tol-erance when exposed to low but nonfreezing

tempera-tures, in a process termed cold acclimation [3] The

acclimation process is a very complex phenomenon comprising numerous changes in metabolism and affecting gene expression, membrane structure, and the composition of proteins and primary and secondary metabolites [4–7] In this context, many studies have shown a strong correlation between changes in the regulation of central carbohydrate metabolism and freezing tolerance [4,8] In Arabidopsis, the development

of leaves at low temperature causes reprogramming of

Keywords

acclimation dynamics; Arabidopsis;

carbohydrate metabolism; freezing

tolerance; mathematical modelling

Correspondence

T Na¨gele, Biologisches Institut, Abteilung

Pflanzenbiotechnologie, Universita¨t

Stuttgart, Pfaffenwaldring 57, D-70550

Stuttgart, Germany

Fax: +49 711 685 65096

Tel: +49 711 685 69141

E-mail: Thomas.Naegele@bio.uni-stuttgart.de

*These authors contributed equally to this

work

(Received 11 August 2010, revised 22

Sep-tember 2010, accepted 22 November 2010)

doi:10.1111/j.1742-4658.2010.07971.x

Low temperature is an important environmental factor affecting the perfor-mance and distribution of plants During the so-called process of cold acclimation, many plants are able to develop low-temperature tolerance, associated with the reprogramming of a large part of their metabolism In this study, we present a systems biology approach based on mathematical modelling to determine interactions between the reprogramming of central carbohydrate metabolism and the development of freezing tolerance in two accessions of Arabidopsis thaliana Different regulation strategies were observed for (a) photosynthesis, (b) soluble carbohydrate metabolism and (c) enzyme activities of central metabolite interconversions Metabolism of the storage compound starch was found to be independent of accession-specific reprogramming of soluble sugar metabolism in the cold Mathemati-cal modelling and simulation of cold-induced metabolic reprogramming indicated major differences in the rates of interconversion between the pools of hexoses and sucrose, as well as the rate of assimilate export to sink organs A comprehensive overview of interconversion rates is pre-sented, from which accession-specific regulation strategies during exposure

to low temperature can be derived We propose this concept as a tool for predicting metabolic engineering strategies to optimize plant freezing toler-ance We confirm that a significant improvement in freezing tolerance in plants involves multiple regulatory instances in sucrose metabolism, and provide evidence for a pivotal role of sucrose–hexose interconversion in increasing the cold acclimation output

Abbreviations

eInv, extracellular invertase; FrcK, fructokinase; FW, fresh weight; GlcK, glucokinase; LT 50 , 50% lethality temperature; nInv, neutral invertase; Rsch, Rschew; SD, standard deviation; SPS, sucrose phosphate synthase; vInv, vacuolar invertase.

carbon metabolism, with a shift in partitioning of

newly fixed carbon into sucrose rather than starch

[9,10], indicating cold-induced selective stimulation of

sucrose synthesis, which could be the reason for the

elevated cellular sucrose content that is found in many

plants upon cold exposure Sucrose may act directly as

a cryoprotectant, as has been shown in vitro with

artificial membrane systems [11], or serve as a

sub-strate for the synthesis of other cryoprotective

compounds, such as raffinose, the level of which has

been found to correlate with freezing tolerance in species

as diverse as A thaliana [12], grape vines [13] and woody

conifers [14]

As already outlined [12], direct correlation of a

mul-tigenic trait such as freezing tolerance with the

concen-tration of only one or a few metabolites may not be

what one would expect This was demonstrated by

work [15] showing that, despite the correlation of

freezing tolerance with raffinose levels in natural

acces-sions of Arabidopsis, varying raffinose concentrations

in accession Col-0 by overexpression of galactinol

syn-thase or knockout of raffinose synsyn-thase did not affect

freezing tolerance Considering the complexity of the

metabolic and regulatory networks, indicated by the

schematic and very simplified structure of primary

car-bohydrate metabolism in Fig 1, it becomes obvious

that, to investigate such nonintuitive networks, an

approach is needed that incorporates multiple and, in

part, circular metabolite interconversions and

regula-tion strategies This is provided by systems biology

techniques, which have rapidly become integrated into

metabolic research, driven by the need to study

com-plex interactions among components of biological

sys-tems [16] Basically, the intention of syssys-tems biology is

to resolve the relationship between individual entities, e.g molecules or genes, that are parts of highly inter-connected networks, in order to understand the result-ing system behaviour, e.g a phenotype of an organism To handle complex networks, formal repre-sentation by mathematical models is indispensable Integration of data on, for example, gene expression, protein abundance, metabolite concentration and other biological parameters with an iterative model, and exploration of model characteristics such as modular-ity, optimality and robustness, promise to advance our system-wide understanding of complex biological net-works [17]

In this work, we present a systems biology approach focused on the dynamic modelling of cold-induced reprogramming of central carbohydrate metabolism in

A thaliana Performing experiments with two acces-sions of different origin, i.e Rschew (Rsch), originat-ing from Russia, and C24, originatoriginat-ing from southern Europe, which show significantly different cold-accli-mation capacities, we prove that mathematical model-ling of metabolism and validation by experimental data offers an attractive possibility for the study of complex system–environment interactions

Results Freezing tolerance Changes in freezing tolerance of Rsch and C24 during

7 days of exposure to cold (4C) was analysed with the well-established electrolyte leakage method, as described in Experimental procedures, with measure-ments at days 0, 1, 3 and 7 (Fig 2) The 50% lethality

Fig 1 Schematic representation of central carbohydrate

metabo-lism in leaf cells of Arabidopsis thaliana Reaction rates (r) represent

central processes of carbon input, output and interconversion.

Fig 2 Freezing tolerance of Rsch (black, continuous line) and C24 (grey, dotted line) over time of exposure to 4 C Closed circles rep-resent means ± SD (n = 6) of LT50.

temperature (LT50) values of both accessions were

sig-nificantly different at all time points during

acclima-tion, confirming that Rsch is more tolerant to freezing

than C24, and has a higher acclimation capacity, as

previously outlined [6] The basic tolerance of C24

was ) 3.3 ± 0.07 C, whereas that of Rsch was

) 4.9 ± 0.09 C Rsch showed the strongest reduction

in LT50between 1 and 3 days, whereas the gain in

tol-erance was only minor during the first 24 h of cold

exposure and between days 3 and 7 In contrast, LT50

decreased almost continuously in C24 until day 3 and

did not change thereafter, resulting in final freezing

tolerances of ) 5.4 ± 0.12 C in C24 and ) 9.1 ±

0.16C in Rsch

Enzyme activites of central carbohydrate

interconversions

As enzyme activities represent crucial points of

regula-tion in metabolic networks, we analysed the maximum

activities (Vmax) of prominent enzymes in central

car-bohydrate metabolism with respect to different

dura-tions of exposure to 4C (Fig 3) The enzymes

analysed included vacuolar invertase (vInv), neutral

invertase (nInv), extracellular invertase (eInv), sucrose

phosphate synthase (SPS), fructokinase (FrcK) and

glucokinase (GlcK) Significant differences in Vmax

between Rsch and C24 were found for vInv (Fig 3A)

and SPS (Fig 3D) Whereas SPS activities were

consis-tently higher in Rsch, C24 showed significantly higher

activities of vInv at 0, 1 and 3 days of cold exposure

The activity of vInv in Rsch increased continuously

during cold exposure, and became significantly higher

than in C24 after 7 days at 4C As compared with

that of vInv, the activities of nInv and eInv were low,

and became noticeably higher only in Rsch after

7 days of cold exposure (Fig 3B,C) However, in both

accessions, values of Vmax for eInv increased

continu-ously from 0 to 3 days of cold exposure

Maximum activities of the hexose-phosphorylating

FrcK and GlcK showed similar patterns in both

acces-sions over the whole period of cold exposure

(Fig 3E,F) The Vmax of GlcK rose sharply in both

accessions by a factor of  1.5 during the first day of

cold exposure (Fig 3F)

Cold-induced changes in net carbon uptake and

sink export

To obtain a quantitative measure of how exposure to

4C influenced the process of photosynthesis, gas

exchange of plants was measured by infrared gas

anal-ysis Measurements were performed during the first 8 h

of the light phase, representing the time period of pho-tosynthetic activity until plants were harvested for analysis of metabolites (see below) The rate of net car-bon uptake was integrated and divided by the time period of measurement to obtain the mean uptake rate per hour (Fig 4A) Mean net carbon uptake was not significantly influenced by cold exposure in Rsch, but showing a slight decrease during the first day at 4C and stabilization over the following time period C24 showed slightly lower mean rates of carbon uptake before and during the first day of cold acclimation After 3 days of cold exposure, the mean rate of carbon uptake was significantly lower for C24 than for Rsch (P = 0.03), and this was followed by recovery until

7 days at 4C, when the mean uptake rate [21.5 ± 1.03 lmol C1Æh)1Æg)1 fresh weight (FW)] was almost the same as in Rsch (24.7 ± 1.8 lmol

C1Æh)1Æg)1FW)

Calculated means of uptake rates were fed into the mathematical model, and standard deviations (SDs) were set as boundaries in the estimation process for model parameters (Fig 4A) As described in Experi-mental procedures, the rate of assimilate export from photosynthetically active source organs to consuming sink organs or metabolic pathways other than carbo-hydrate pathways was calculated as the difference between net carbon uptake and changes in cellular car-bohydrate content The resulting surplus of carbon equivalents (Fig 4B) was regarded as being exported

to sink organs or other pathways The time courses of simulated export rate during the first day of exposure to

4C were very similar in both accessions, showing a slight decrease, which was also found for net carbon uptake (see above) During the following days of cold exposure, Rsch showed a noticeably faster regeneration

of sink export rate than did C24, although both acces-sions reached almost the same export rate after 7 days

of cold exposure Discontinuities in the calculated export rate after 1 day and 3 days result from the sharp increase in carbohydrate content (starch and soluble car-bohydrates) during that time period of cold exposure

Effect of cold exposure on levels of soluble carbohydrates and starch

Contents of leaf starch, sucrose, hexoses and raffinose were determined over the course of cold exposure (Fig 5) In both accessions, starch content was not altered at 1 day of cold exposure (Fig 5A), but showed a significant increase between 1 day and 3 days (PRsch< 0.0001; PC24< 0.0001), coinciding with the main increase in freezing tolerance (see Fig 2) The starch content of C24 decreased nonsignificantly until

7 days of cold exposure, reaching 16.2 ± 7.07 lmol

-C6Æg)1 FW, whereas Rsch had a starch level of

23 ± 7.4 lmol C6Æg)1 FW after the cold acclimation

period

Over the time course of acclimation, changes in con-centrations of soluble carbohydrates during cold expo-sure displayed some similarities with respect to dynamics, but differed greatly in absolute values

A

B

C

D

E

F

Fig 3 Maximum activities of enzymes in central carbohydrate metabolism during cold exposure (A–C) Vmaxvalues of three invertase iso-forms: vInv, nInv and eInv (D) Vmaxof SPS (E, F) Vmaxvalues of FrcK and GlcK Significant differences between the ecotypes Rsch (black) and C24 (grey) are indicated by asterisks (P < 0.05) Bars represent means ± SD (n = 7).

(Fig 5B–D) Sucrose content increased significantly

and reached peak values after 3 days of cold exposure:

7.1 ± 2.3 lmolÆg)1FW in Rsch and 3 ± 0.8 lmolÆg)1

FW in C24 (Fig 5B) This was followed by a slight but nonsignificant decrease until 7 days of cold expo-sure Concentrations of free hexoses, calculated as the

Fig 4 Rates of net photosynthesis (A) and simulated sink export (B) during cold exposure in Rsch (black) and C24 (grey) Open circles rep-resent means of measurements ± SD (n = 3) Continuous lines reprep-resent means of model simulations (n = 50) Dotted lines reprep-resent results of model simulations with lower and top values of kinetic parameters.

Fig 5 Cold-induced dynamics of central carbohydrates in Rsch (black) and C24 (grey) Open circles represent means of measure-ments ± SD (n = 5) Continuous lines represent means of model simulations (n = 50) In (B) (sucrose) and (C) (hexoses), dotted lines repre-sent the results of model simulations with lower and top values of kinetic parameters.

sum of fructose and glucose equivalents, were similar

in both accessions at the beginning of cold exposure

(Fig 5C) However, after 3 days of cold exposure,

Rsch (67.1 ± 9.3 lmol C6Æg)1 FW) accumulated

almost three times as much hexose as C24

(28.1 ± 2.8 lmol C6Æg)1 FW), and it maintained this

level until 7 days, whereas C24 showed a significant

decrease in hexose level to 15.1 ± 3.7 lmol C6Æg)1FW

(P < 0.001) The raffinose concentration increased

almost linearly with time of cold exposure in both

accessions In Rsch, the raffinose content increased

sig-nificantly from 0.13 ± 0.04 to 2.25 ± 0.6 lmolÆg)1

FW after 7 days of cold exposure (P < 0.01), and was

about twice as high as in C24, which showed an

increase from 0.09 ± 0.02 to 0.96 ± 0.39 lmolÆg)1

FW (P < 0.01; Fig 5D)

Simulation of metabolic levels and rates of

interconversion

Identification of parameters used to describe the

meta-bolic network as represented in Fig 1 was performed

by applying a constraint-based approach (for the

expli-cit model structure, see Experimental procedures)

Model constraints were set by experimental data on

net carbon uptake, metabolite levels and maximum

enzyme activities, which gave a provisional estimation

of the maximum flux capacity of the corresponding

pathway Experimental data on maximum enzyme

activities of SPS, GlcK, FrcK and invertase at 4C

were used as lower and upper bounds in the process of

parameter identification The resulting model

simula-tion using identified parameters was successful in

describing cold exposure-dependent changes in

carbo-hydrate levels (Fig 5A–D, continuous lines) To test

the statistical robustness of the identified model

parameters and to validate them with experimental

data, 50 independent identification processes with

vary-ing initial carbohydrate levels were performed, yieldvary-ing

means with corresponding SDs of estimated kinetic

parameters Identified values of Vmax matched the

val-ues from experiments, and comparison of identified Km

and Ki values agreed with values from the literature

(Table 1) Rate constants and corresponding rates of

sucrose synthesis were compared with Vmax values for

both hexokinase activity (GlcK and FrcK) and SPS

activity, as both enzymes contribute to hexose-based

sucrose synthesis (see also ‘Model documentation’ in

Doc S3) Simulations resulting from upper, lower and

mean values of parameter sets described metabolic

changes during cold exposure within the SDs of

experi-mental results (Fig 5A–D), thus proving reproducibility

of the obtained parameters and of simulation results

Mean values of accession-specific parameter sets were used to analyse low-temperature effects on inter-conversion rates during the 7-day cold acclimation per-iod Rates of sucrose–hexose interconversions showed significant differences between Rsch and C24 after

7 days of exposure to 4C (Fig 6A,B), but were the same for the first 3 days of cold exposure, except for a small peak in sucrose cleavage rate in Rsch on day 2 (Fig 6A) In order to obtain a comprehensive over-view of all simulated rates of metabolite interconver-sions, a three-dimensional surface plot was created (Fig 7A,B) that allowed (a) assessment of the trajec-tory of interconversion rates as a function of time of cold exposure, (b) comparison of the magnitudes of the various interconversion rates, and (c) lineup of the accessions with respect to their metabolic acclimation strategies Major differences in sucrose metabolism between the accessions were identified Whereas C24 showed a cold-induced reduction of carbon channelling into sucrose synthesis from the start until day 3 of exposure to 4C, the corresponding flux in Rsch was reduced only during the first 24 h of cold exposure (Fig 7A,B, CO2 to sucrose) A similar pattern was observed for rates of CO2 uptake and export of sucrose to sink organs, but not for starch synthesis As already illustrated in Fig 6, sucrose cleavage and hex-ose-based resynthesis were increased in Rsch, whereas C24 showed a significant reduction in sucrose cycling during cold exposure (Fig 7A,B, sucrose to hexoses, hexoses to sucrose)

In silico experiments

To estimate the metabolic impact of differences between Rsch and C24 concerning sucrose cycling, we performed in silico experiments, using the validated mathematical model in terms of predictive metabolic engineering [18] Replacing Vmax values and k values

in the C24 model with the identified values for Rsch resulted in simulations that were not successful in describing the whole experimental dataset on sucrose and hexoses (Fig S1) The sucrose content after 1 day

at 4C was predicted to be higher than the experimen-tal value, whereas the simulated hexose content was lowered Performance of a further in silico experiment

in which the Vmaxand k values of C24 were applied to the Rsch model confirmed that the main differences in reprogramming of carbohydrate metabolism occur dur-ing the first 3 days of exposure to low temperature (Fig S2) In particular, the sucrose content after 1 day

at 4 C was underestimated and, simultaneously, the hexose content showed a faster increase than in the corresponding experimental data

Km

Ki

(Vmax

1 Æg

Km

Ki

Vmax

Parameter estimation

Parameter estimation

Parameter estimation

Parameter estimation

Vmax

Km

Ki

Vmax

Km

Ki

Cold acclimation of plants involves a large number of

metabolic changes as well as readjustments in other

cellular processes Numerous studies have emphasized the importance of primary carbohydrate metabolism during cold acclimation, and have identified regulatory instances with significant influence [9,10,12,15,19,20]

A

B

Fig 7 Surface plot of simulated rates of

metabolite interconversion for accessions

C24 (A) and Rsch (B) For comparison, all

fluxes are represented in lmol C6Æh)1Æg)1

FW In addition to surface topography,

quantities of fluxes are indicated by colour

as defined in the colour bar.

Fig 6 Dynamics of rates of sucrose cleavage (A) and hexose-based sucrose synthesis (B) during exposure to 4 C Lines represent means

of simulation (n = 50) for Rsch (black) and C24 (grey) Dotted lines represent results of model simulations with lower and top values of kinetic parameters.

However, the complex interactions of metabolic

path-ways precludes the generation of a full picture of

cold acclimation through assembly of reaction details

In the present study, a systems biology approach with

dynamic modelling was developed and validated by

experimental data on two Arabidopsis accessions, Rsch

and C24, with different cold acclimation capacities

Dynamics were generated by varying the time periods

for which plants were exposed to 4C, thus capturing

different stages of metabolic adjustment to low

temper-ature As indicated by the LT50values, the freezing

tol-erances of the accessions differed not only in terms of

the absolute values but also in the progression of the

acclimation process This is an important outcome, as

it allows estimation of the impact of different

meta-bolic responses on the improvement in freezing

toler-ance Comparison of changes in metabolism between

1 day and 3 days of cold exposure revealed significant

differences in net carbon uptake and sink export rate

between Rsch and C24 Whereas net carbon uptake

and rate of sink export were constantly reduced in C24

over the entire exposure time, Rsch almost completely

compensated for the low-temperature effects at day 3

This coincides with the time point of maximal

toler-ance acquisition, thus proving the importtoler-ance of

pho-tosynthesis and long-distance transport for

acclimation The requirement for photosynthetic

activ-ity has also been demonstrated [21], and it was shown

that acclimation does not take place in total darkness

Strand et al [22] found that cold acclimation was

sig-nificantly enhanced in plants with increased SPS

activ-ity, leading to higher photosynthetic performance at

low temperatures Interestingly, model simulations for

C24 and Rsch revealed that synthesis of soluble sugar

was never limited by photosynthetic capacity Even

C24, which displayed a reduction in photosynthesis at

days 3 and 7, had the capacity to assimilate at least

about 3 lmol C6Æh)1Æg)1 FW, which would have been

sufficient to bring about much higher sugar levels than

those determined Therefore, we suggest that assimilate

allocation within the plant may become limiting in the

cold This was also demonstrated for cucumber, in

which the sucrose supply to sink organs rather than

source capacity correlated with low-temperature

toler-ance [23] It appears that the major difference between

Rsch and C24 is the capacity to re-establish

homeosta-sis in carbon allocation This is supported by the

observation that Arabidopsis plants with SPS

overex-pression, which show a significant increase in freezing

tolerance as compared with the wild type, not only

accumulate sucrose in their leaves, but also specifically

increase the expression of the high-affinity sucrose

transporter AtSUC1, which is highly homologous to

the phloem loading transporter AtSUC2 [20,24] How-ever, it has to be kept in mind that the sink export rate

in our model is composed of assimilate export to sink organs and flux into further pools of carbon-contain-ing metabolites and structural components, e.g amino acids and cell wall components Therefore, the real rates of export of carbohydrates to sink organs might

be smaller than predicted by our model

In contrast to soluble carbohydrates, the starch con-tent of plants did not show significant differences between the accessions over the whole period of cold exposure, even though net carbon uptake rates varied strongly This suggests that starch metabolism was directly correlated neither with photosynthesis nor with the cold acclimation process This may explain why

we, using C24 and Rsch, did not find a negative corre-lation of freezing tolerance with channelling of carbon into starch, whereas Klotke et al [12] reported such a correlation for C24 and Col-0, which has a freezing tolerance similar to that of [6] Given that starch plays

an important role as a major integrator in the regula-tion of plant growth [25], it is noticeable that, at least

in Rsch, the most significant changes in starch content occurred simultaneously with the largest increase in freezing tolerance Considering that rates of rosette biomass increase are negatively correlated with starch levels [25], our data confirm the observation that the acquisition of freezing tolerance is coupled to a meta-bolic state in which growth is suspended [26]

Major changes in pools of free hexoses and sucrose took place until the third day of cold exposure, but after this no further significant changes could be observed Therefore, we conclude that the process of cold acclimation is divisible into three consecutive stages: (a) immediate response to displacement of homeostasis; (b) reprogramming of central carbohy-drate metabolism; and (c) stabilization of a new state

of metabolic homeostasis with respect to carbohydrate metabolism Simulation of metabolite interconversion rates revealed a distinct difference in sucrose metabo-lism of Rsch and C24 In particular, rates of sucrose cleavage and hexose-based sucrose resynthesis showed significant differences with respect to both their abso-lute values and the time course From the simulations,

it appears that the ability to sustain the cycling of sucrose, which has been postulated to function in the stabilization of mesophyll sugar metabolism [27–29], positively correlates with low-temperature acclimation capacity Additional support for this hypothesis arises from experimental data on enzyme activities, which show that invertase activity is increased during cold exposure in Rsch, whereas acitivity is reduced in C24 after 7 days in the cold Regarding the question of

how to engineer plant metabolism in order to improve

freezing tolerance, one could suggest increasing the

maximum activities of enzymes participating in sucrose

cycling Assuming that Rsch is optimized for cold

acclimation, we suggest, on the basis of the results of

the in silico experiments (Figs S1 and S2), that the

metabolism of C24 has to be changed in a way that

leads to an increased sucrose content and a

simulta-nous reduction in hexose concentration, particularly

during the initial period of cold exposure Using RNA

interference-mediated inhibition of the dominating

vac-uolar invertase ATbFRUCT4 (At1g12240), we have

already demonstrated this [12] However, it was shown

that fully cold-acclimated transformants of C24 did

not differ from the wild type with regard to freezing

tolerance and, at the same time, differences in sucrose

concentration between the C24 genotypes were lost

Therefore, we suggest that inhibition of invertase must

be linked with overexpression of SPS, as described in

[22], to achieve sucrose accumulation, a decrease in

hexose content and, in consequence, a significant

increase in freezing tolerance

Conclusions

The present study elucidates differences in

cold-induced reprogramming of central carbohydrate

metabolism Mathematical modelling of metabolism

with respect to the dynamics of freezing tolerance

revealed a significant correlation of sucrose synthesis

and degradation with the process of cold acclimation

We conclude that acclimation to low temperature

rep-resents a dynamic process, the investigation of which

therefore requires approaches that take into account

metabolic dynamics and interdependencies rather than

simple steady-state concentrations We present a

method based on dynamic modelling that allows for

the quantification and visualization of cellular rates of

metabolite interconversion during an acclimation

pro-cess incorporating environmental changes

Further-more, we suggest that successful metabolic engineering

of freezing tolerance in plants should include such an

analysis of the dynamics of metabolism to gain

com-prehensive information about the effects of individual

overexpression or knockout events on the whole

accli-mation process

Experimental procedures

Plant material

A thaliana plants of the accessions Rsch and C24 were

grown in GS90 soil and vermiculite (1 : 1), with three

plants per 10-cm pot in a growth chamber at 8 h of light (50 lmolÆm)2Æs)1; 22C) ⁄ 16 h of dark (16 C) for 4 weeks, and then transferred to a growth chamber with a tempera-ture regime of 22C in the day (16 h) and 16 C at night (8 h) The light intensity was 50 lmolÆm)2Æs)1, and the rela-tive humidity was 70% Plants were watered daily, and fer-tilized every 2 weeks with standard NPK fertilizer After

42 days, plants were shifted to a 16-h⁄ 8-h light ⁄ dark regime

of 4⁄ 4 C and a light intensity of 50 lmolÆm)2Æs)1 Leaf samples consisting of two rosette leaves each were taken from nine individual plants grown in three different pots

in a random design before and after 1 day, 3 days and

7 days of exposure to 4C Samples were taken after a light period of 8 h At that stage, the aerial part of the plant is exclusively composed of rosette leaves, allowing

a direct comparison of metabolite with CO2 exchange data Leaf samples were weighed, immediately frozen in liquid nitrogen and stored at ) 80 C until metabolite extraction

Electrolyte leakage measurement Freezing tolerance was tested according to the electrolyte leakage method as previously described [30], with a few modifications The cooling rate was set to 4C ⁄ h, and sam-ples were taken at 2C intervals over a temperature range

of 0 to ) 18 C Conductivity was measured with an ino-Lab740 conductivity meter (WTW GmbH, Weilheim, Ger-many) and multilabpilot software The LT50values were calculated as the log EC50 values of sigmoidal dose– response curves, fitted to the measured leakage values with graphpad prism3 software

Gas exchange measurement Exchange rates of CO2were measured with an infrared gas analysis system (Uras 3 G; Hartmann & Braun AG, Frank-furt am Main, Germany) A whole-rosette cuvette design was used as described in [31] Gas exchange was measured

in the growth chamber shortly before plant harvesting Means of raw data for gas exchange were converted to flux rates per gram of FW obtained at the end of the exposure

by weighing complete rosettes

Carbohydrate analysis Frozen leaf samples were homogenized with a

Rets-ch MM20 ball mill (Retsch, Haan, Germany) The homogenate was extracted twice in 400 lL of 80% etha-nol at 80C Extracts were dried and dissolved in

500 lL of distilled water Contents of glucose, fructose, sucrose and raffinose were analysed by high-performance anion exchange chromatography (HPAEC) with a Carb-oPac PA-1 column on a Dionex (Sunnyvale, CA, USA)