LARGE-SCALE BIOLOGY ARTICLE

Linking Gene Expression and Membrane LipidComposition of ArabidopsisW OPEN

Jedrzej Szymanski,1 Yariv Brotman, Lothar Willmitzer, and Álvaro Cuadros-Inostroza

Max Planck Institute of Molecular Plant Physiology, 14476 Potsdam-Golm, Germany

Glycerolipid metabolism of plants responds dynamically to changes in light intensity and temperature, leading to the modification ofmembrane lipid composition to ensure optimal biochemical and physical properties in the new environment. Although multipleposttranscriptional regulatory mechanisms have been reported to be involved in the process, the contribution of transcriptionalregulation remains largely unknown. Here, we present an integrative analysis of transcriptomic and lipidomic data, revealing large-scale coordination between gene expression and changes in glycerolipid levels during the Arabidopsis thaliana response to light andtemperature stimuli. Using a multivariate regression technique called O2PLS, we show that the gene expression response is strictlycoordinated at the biochemical pathway level and occurs in parallel with changes of specific glycerolipid pools. Five interestingcandidate genes were chosen for further analysis from a larger set of candidates identified based on their close association withvarious groups of glycerolipids. Lipidomic analysis of knockout mutant lines of these five genes showed a significant relationshipbetween the coordination of transcripts and glycerolipid levels in a changing environment and the effects of single geneperturbations.

INTRODUCTION

Glycerolipid metabolism of plants is strongly involved in the responseand adaptation to changes in environmental conditions (Moelleringand Benning, 2011). Two common environmental parameters af-fecting glycerolipid metabolism are light intensity and temperature,and the effects of both have been intensively studied in plants overthe last four decades (Yoshida and Sakai, 1974; Browse et al., 1981).

One of the most well-known effects of prolonged chilling stress isincreased desaturation of glycerolipids, serving as a compensationmechanism for decreased membrane fluidity (Williams et al., 1988;Welti et al., 2002; Tasseva et al., 2004). In agreement with this, cold-tolerant species (Sakamoto et al., 2004; De Palma et al., 2008) andvarieties (Horvath et al., 1983) and cold-acclimated plants(Degenkolbe et al., 2012) have been shown to have relatively highlevels of desaturated glycerolipids. Changes in glycerolipid satura-tion levels are also accompanied by other more specific effects,such as a decrease of monogalactosyldiacylglycerols (MGDGs)(Li et al., 2008) and an increase of triacylglycerides (TAGs) con-nected with the outer chloroplast envelope remodeling mediated bygalactolipid:galactolipid galactosyltransferase (Moellering andBenning, 2011). Besides galactolipid:galactolipid galactosyl-transferase, a number of proteins have been shown to connectlipid metabolism with cold stress. Common examples include thephospholipases PLDa1 and PLDd (Welti et al., 2002; Rajashekar

et al., 2006; Li et al., 2008) and acyl-CoA binding proteins (Chenet al., 2006; Du et al., 2010a, 2010b). In contrast with the effect ofcold stress/treatment, an increased saturation of membrane lipidshas been observed as a heat stress–induced adaptationalmechanism (Larkindale and Huang, 2004). On the other hand,light intensity affects membrane lipid composition in a differentway. Desaturation of acyl chains is not light regulated (Browseet al., 1981), but darkness strongly represses the de novo syn-thesis of fatty acids (FAs) (Ohlrogge and Jaworski, 1997) and thusalso the influx of saturated FAs into glycerolipid biosyntheticpathways. High light has an opposite effect, leading to a surplusof NADPH (Stitt, 1986) and increased FA synthesis. Therefore, aswe discussed by Burgos et al. (2011), light affects lipid metabo-lism mainly by defining energy and carbon availability. The in-volvement of specific glycerolipids, such as phosphatidylglycerol(PG), in the oligomerization and stabilization of photosyntheticcomplexes (Frentzen, 2004) further suggests a more complexrelationship between light and membrane lipid composition.Whereas most of the described effects might be attributed to

specific posttranscriptional regulatory processes, there are hints thatregulation at the transcriptional level also contributes to remodelingof the plant membrane composition. The most important evidenceis the tight coexpression of genes involved in lipid metabolismand in its particular biochemical pathways in response to stress(Obayashi et al., 2007; Loraine, 2009; Avin-Wittenberg et al., 2011).Several transcriptional regulators of specific lipid metabolic path-ways have also been reported (Baud and Lepiniec, 2010; Bernardand Joubès, 2013), but the details of transcriptional regulation inmembrane glycerolipid remodeling are not well understood. There-fore, in this work, we ask three basic questions. (1) To what extent istranscriptional regulation involved in the remodeling of plant mem-brane lipid composition in response to changes in light and tem-perature? (2) Are there specific pathways and metabolic processes

1Address correspondence to szymanski@mpimp-golm.mpg.de.The author responsible for distribution of materials integral to the findingspresented in this article in accordance with the policy described inthe Instructions for Authors (www.plantcell.org) is: Jedrzej Szymanski(szymanski@mpimp-golm.mpg.de).W Online version contains Web-only data.OPENArticles can be viewed online without a subscription.www.plantcell.org/cgi/doi/10.1105/tpc.113.118919

The Plant Cell, Vol. 26: 915–928, March 2014, www.plantcell.org ã 2014 American Society of Plant Biologists. All rights reserved.

exhibiting coordination between the expression of the enzymaticgenes and the accumulation of certain lipid species? And (3) doesthis coordination remain in agreement with the effects of single geneperturbations and thus could be useful for gene function prediction?

To answer these questions, we revise the transcriptomic andlipidomic data for Arabidopsis thaliana, described separately in ourprevious studies (Burgos et al., 2011; Caldana et al., 2011), in a newintegrative analysis. Integrative transcriptomic–metabolomic studies,aiming at the identification of significant and biologically relevantcoordination between gene expression and metabolite levels(Redestig et al., 2011), proved to be useful in uncovering large-scaleorganization of metabolic regulation and in highlighting candidatesfor new enzymes and regulators of plant metabolism (Urbanczyk-Wochniak et al., 2003; Hirai and Saito, 2004; Hirai et al., 2007;Jozefczuk et al., 2010). Here, we use O2PLS, amultivariate regressionanalysis method developed by Trygg (2002) and Trygg and Wold(2003) and applied in multiple systems-scale studies (Bylesjö et al.,2007, 2009; Consonni et al., 2010; Zamboni et al., 2010; Kusanoet al., 2011). O2PLS is an extension of orthogonal projections tolatent structures (OPLS) (Trygg and Wold, 2003); whereas OPLSwas designed for analysis of a single data set, O2PLS assessessystematic trends across multiple data sets. In principle, O2PLS isa multivariate analysis method like, for example, principal compo-nent analysis (PCA), but it parses out the variation in large data setsdifferently. O2PLS focuses on predictive information, separating thevariance common for two data sets (correlated between the datasets) from the variance unique for only one data set (correlated withinone data set only) and from analytical noise (residual uncorrelatedvariance). This is especially important in the case of time-series ex-periments, where differences between the accumulation of lipids andgene expression changes arise from different dynamics of lipidmetabolism and gene regulation and where a large quantity ofplatform-specific analytical noise is expected (Burgos et al., 2011).

Our study is divided into three major steps. (1) By using O2PLSanalysis, we assess the fraction of the predictive variance in bothlipid and transcript data and, thus, the extent to which changes inlipid composition are reflected by changes in gene expression andvice versa. (2) We interpret the identified coordination in the contextof known regulatory mechanisms and pathway-level regulation ofgene expression. (3) Finally, we compare the identified interactionswith the effect of individual gene knockouts and evaluate to whatdegree the environmental transcript–lipid coordination reflectscausal transcript–lipid relationships.

The results indicate a significant coordination between geneexpression and glycerolipid accumulation in response to changingenvironmental conditions related to a concerted regulation ofmajor lipid biosynthetic pathways. Furthermore, a correspondencebetween transcript–lipid abundance coordination and the targetedgene perturbations is shown.

RESULTS

Experimental Setup

Applied treatments include four light regimes: high light (400 mE),control light (150 mE), low light (75 mE), and darkness; and threetemperature regimes: cold (4°C), control temperature (21°C), and

heat (32°C). These regimes were combined such that all light in-tensity treatments were performed in control temperature and thecold and heat stress experiments were performed under controllight and additionally in darkness, resulting in eight different com-binations (Figure 1). In the following sections, the conditions areindicated by the following letter codes: 4-L (cold/normal light), 4-D(cold/darkness), 21-HL (control temperature/high light), 21-L (con-trol temperature/normal light; control conditions), 21-LL (controltemperature/low light), 21-D (control temperature/darkness), 32-L(heat/normal light), and 32-D (heat/darkness). For each combina-tion, we obtained time-course data for the first 6 h of the plantresponse with sampling every 20 min. From strictly quality-filteredprobe sets, a subset representing the expression of 480 acyl-lipidmetabolism genes was selected based on the ARALIP databaseclassification (http://aralip.plantbiology.msu.edu; version from Jan-uary 2013) (Li-Beisson et al., 2013). All details on the experimentaland analytical procedures are described in Methods as well as byBurgos et al. (2011) and Caldana et al. (2011).

Statistical Model

The basic concepts of O2PLS are key to understanding the resultsof this study; therefore, first we will introduce the basic terminologyand idea of the method. O2PLS is a statistical method allowing theconvenient integration of two data sets collected from the samesamples and expected to be causally connected, such as tran-scripts and metabolites (Trygg, 2002; Trygg and Wold, 2003). In anO2PLS model, each data set is decomposed into three variancestructures describing predictive, unique, and residual variation of itsvariables (Figure 2). The predictive structures describe a multivariaterelationship between two data sets, allowing reciprocal predictionbetween them. In our experiment, it allows the estimation of mem-brane lipid levels from the gene expression and vice versa andmightbe related to a direct metabolic and regulatory link between geneexpression and levels of membrane lipids. The unique structuresrepresent patterns that are not useful to predict the other data set.These might be linked to platform-specific effects, such as baselinebias, or variation that is not reflected in the other data set, such as

Figure 1. Array of Applied Treatments.

916 The Plant Cell

gene expression regulation having no effect on lipid levels or met-abolic changes originating from other than transcriptional regulatorymechanisms. Finally, the residual variance structures representnoise and stochastic effects.

In O2PLS, each variance structure is composed of a certainnumber of latent variables and their weights (called loadings),describing the contribution of the latent variables in the varianceof each observable variable. Thus, changes of all observablevariables (e.g., all measured lipids) in the frame of a certain vari-ance structure (e.g., lipid predictive) are a linear combination ofjust a couple of latent variables mixed with different proportions.

Here, the O2PLS analysis was performed using a modifiedversion of the algorithm described by Trygg (2002) (see Methodsfor details). The number of latent variables for each variancestructure was identified by choosing the configuration thatminimized the generalization error, which was estimated byusing a group-balanced Monte Carlo cross-validation (MCCV)(Bylesjö et al., 2007). The optimal setup consisted of eight jointvariance components (Figure 3A shows two of them), fivetranscript-unique components, and six lipid-unique components(Supplemental Figure 1 and Supplemental Table 1). In the fol-lowing sections, latent variables of the respective model struc-tures will be described by letter codes: J-LV1 to J-LV8 for bothtranscript and lipid latent variables of the predictive structure,JT-LV1 to JT-LV8 for the transcript latent variables of the pre-dictive structure, JL-LV1 to JL-LV8 for the lipid latent variablesof the predictive structure, UT-LV1 to UT-LV5 for the latentvariables of the transcript-unique structure, and UL-LV1 toUL-LV6 for the latent variables of the lipid-unique structure. It isimportant to note that, unlike in PCA, the numbering of O2PLSlatent variables is arbitrary and does not relate to the amount ofvariance explained.

Identification of Predictive Transcriptome–LipidomeVariance

The total joint variance between the data sets reaches 61.6%for the transcript-predictive structure and 27.6% for the lipid-

predictive structure. Unique variance reaches 20.1% for thetranscript data and only 4.8% for the lipid data. The model showsthat the environmental stimuli result mostly in systematic andcoordinated changes of lipids and lipid metabolism gene expres-sion. A considerable amount of transcript-unique variance alsoindicates that a significant proportion of the gene expressionchanges are accompanied by changes in lipid levels in theexperimental time scale. On the other hand, the relatively lowcontribution of lipid-unique variance suggested that almost allsystematic effects observed in lipid data could be connected withsome changes in transcripts. To verify that the residual structuresdo not contain systematic variance, these were analyzed by PCA,which indicated no significant treatment- or time-specific effects.To determine if the lipid metabolic genes have higher predictive

power toward lipid profiles than any other set of genes, we per-formed a permutation test, where an O2PLS model of the samecomplexity (the same number of variance components for eachrespective variance structure) was fitted to a randomly chosenset of nonlipid genes of the same size as the lipid genes group(480 genes). After 1000 iterations, none of the random gene setsexceeded the ratio of the joint variance obtained for the lipid genes(P < 0.001). Accordingly, the size of the unique variance structurewas higher for the nonlipid genes, with P < 0.05 (SupplementalFigure 2). This result indicated that the changes in expression oflipid-related enzymatic genes were indeed related to the accu-mulation of glycerolipids and were more significant compared withchanges in any other random set of genes.In the next step, we performed an O2PLS discriminatory analysis

(O2PLS-DA) (Bylesjö et al., 2006). O2PLS-DA identifies latent var-iables discriminating preselected sets of observations within theO2PLS model. Here, O2PLS-DA was performed on predictivevariance structures (Figure 3A; see Supplemental Figure 3 for plotsof all latent variables of predictive variance structures), and its resultwas regressed on the gradient on light and temperature conditions(abbreviated as TEMP-LV and LIGHT-LV, respectively; Figure 3B).The new TEMP-LV and LIGHT-LV obtained from the O2PLS-DAmodel extract a portion of the joint variance, which is related toeither the light or temperature gradient. This operation allows rep-resenting the transcripts and lipids on a common 2D plane (Figure3C), where the dimensions describe the light and temperaturespecificity of the response and the proximity between transcriptsand lipids is related to the coordination of their changes.

Large-Scale Coordination of Lipid Biochemical Processes

Applied treatments led to moderate changes in glycerolipid classlevels. Darkness resulted in the accumulation of phosphatidylinositol(PI), PG, and phosphatidylethanolamine (PE), reflected by negativecorrelation of the sum of all molecular species of the respectiveclass with the light-specific latent variable (Figure 4; the generalizedchanges of whole lipid classes are represented by arrows). In thecase of PI, the effect is slightly stimulated by elevated temperature,whereas PG and PE accumulate more pronouncedly in dark/cold conditions. Sulfoquinovosyldiacylglycerol (SQDG), MGDG,and phosphatidylserine, on the other hand, exhibited temperature-specific effects, accumulating under 32°C independently of thelight conditions. Fluctuations at the level of whole glycerolipid poolsare rather minor; however, major differences occurred between

Figure 2. Overview on the O2PLS Model Structures Obtained for In-tegration of the Transcript and Lipid Data.

Each model structure shows the percentage of the total varianceexplained.

Integrative Analysis of Lipid Metabolism 917

molecular species belonging to the same glycerolipid class. Thisindicated that most of the significant effects occurred due to a shiftin saturation level or acyl chain length (or both) and not in the levelof head group chemistry. Indeed, the light effect (represented byLIGHT-LV) correlates significantly with the saturation level of multipleglycerolipids (Supplemental Figure 4B). This concerns in particularMGDGs, phosphatidylcholines (PCs), PEs, and SQDGs, exhibitinggradients of saturation from the desaturated lipid species accu-mulating in darkness to saturated species increasing in light andhighlight conditions. This effect was enhanced by heat and con-cerns multiple lipid classes. The most extreme effect is exhibited bya group of normally low-abundance species of PC, PE, PI, andSQDG containing fully desaturated 16-C acyl chains. The mostrepresentative of them are 34:6 SQDG, PC, and PE, 34:5 PI, and32:3 PI; the probable source of these species and their connectionwith temperature are debated by Burgos et al. (2011). The sum-marized changes of the molecular lipid species containing the samenumber of carbons in their acyl chains indicate that shorter 16-Cchains accumulate in darkness and heat (Supplemental Figure 4C).Heat treatment was positively correlated with molecular specieshaving two 16-C acyl chains, whereas for the control temperature indarkness the combination of 18-C and 16-C was dominant. On theother hand, light was not significantly correlated with any certainnumber of acyl chain carbons, and cold treatment promoted theaccumulation of lipid species with double 18-C acyl chains.

Parallel changes in gene expression indicate that remodelingof the membrane lipid composition coincides with a coordinatedregulation of multiple lipid metabolism pathways (Figures 4B to 4D).

The general trends of the pathway gene expression, obtained in thesame way as sum changes of the lipid classes, match the latentvariables describing major joint variance components (Figure 3C).There are three major transcriptional programs, light-specific/temperature-independent, temperature-specific/light-independent,and intermediate light/temperature-modulated, each specific toa certain subset of lipid biochemical pathways. The major one(intermediate light/temperature-modulated), including FA synthesisand trafficking, prokaryotic glycerolipid synthesis, FA elongationand wax biosynthesis, oxylipin metabolism together with cutin andsuberin biosynthetic pathways, follows the J-LV4 pattern, exhibitinga strong positive correlation with light regime and dominated bytemperature-modulated downregulation of the gene expression indarkness. FA degradation and both TAG degradation and bio-synthesis follow closely J-LV1 and J-LV2, being positively corre-lated with the temperature gradient. Finally, the last group ofpathways, including sphingolipid metabolism, phospholipid sig-naling, and eukaryotic phospholipid synthesis, follows J-LV3, whichcovers largely opposite responses to J-LV4 but without a strongmodulating effect of temperature.

Relationship between Environmental Coordination andGenetic Control

To estimate the degree to which the observed coordinationbetween transcript and lipid changes is related to the actualmetabolic functions of particular genes, we selected five genes,exhibiting various degrees and specificity of response to light

Figure 3. Characteristics of O2PLS Latent Variables.

(A) Exemplary plot of two out of eight identified O2PLS latent variables (LV). Samples are denoted by numbers representing the time points of eachexperiment (expressed as minutes upon treatment). The plot should be interpreted analogously to PCA. The variances captured by the LV1 (r2X = 0.21and r2Y = 0.6, where X and Y refer to the transcript and lipid data, respectively) and LV2 (r2X = 0.27 and r2Y = 0.12) separate samples are mostlyaccording to temperature conditions and the time progression of the experiment.(B) Light-specific latent variables obtained by partial least squares regression of the O2PLS-DA–predicted values (r2X = 0.28 and r2Y = 0.20) on the lightgradient and temperature-specific latent variables obtained by partial least squares regression of the O2PLS-DA–predicted values (r2X = 0.32 and r2Y =0.28) on the temperature gradient. As shown, the light- and temperature-specific latent variables capture the variance of the predictive variancestructure, being directly related to the applied temperature treatment and light intensity (note the clear temperature-related separation of samples alongthe x axis and the light-related separation along the y axis).(C) Correlations between O2PLS-DA latent variables and the original eight latent variables of the transcript O2PLS predictive model structure. Eachgreen arrow represents one latent variable (numbered next to the arrowhead). The thickness of the arrow represents the proportion of the variancedescribed by each particular latent variable. Additionally, correlations of individual transcripts and lipids are plotted, indicating that variance in both datasets covers a broad spectrum of combinations of light- and temperature-related effects.

918 The Plant Cell

Figure 4. Correlation Loading Plots of the Lipids and Transcripts in the O2PLS-DA Model.

Lipids are denoted as triangles and transcripts as squares. The location of each variable on the plot is described by the correlation of its joint varianceprojection with light-specific (x axis) and temperature-specific (y axis) latent variables of the O2PLS-DA model. The gradients are from darkness- to highlight–specific responses (left to right) on the x axis and from cold- to heat-specific responses (bottom to top) on the y axis. Each plot shows thedistribution of all analyzed variables (small gray symbols), but on each of them a different group is highlighted by magnified and color-coded symbols.(A) Glycerolipids. The total number of double bonds in glycerolipid acyl chains is represented by the color scale from red (six double bonds) to blue(no double bonds). Sum changes of the lipid classes are represented by green arrows. From all plotted lipids, only those with significant correlation(P < 0.01) with one of the latent variables have been plotted.(B) Transcripts of the FA biosynthesis pathway. Genes belonging to the FA biosynthesis pathway are highlighted in green. Sum changes of all lipidbiochemical pathways are plotted as gray arrows. The sum changes of the FA biosynthesis pathway are highlighted by a green arrow. The abbreviationsfor the pathways are as follows: Cutin, cutin biosynthesis; EuP, eukaryotic phospholipid biosynthesis; FA, FA biosynthesis; FA&TAG deg, FA and TAGdegradation; FAE&Wax, FA elongation and wax biosynthesis; LT, lipid trafficking; Oxylipin, oxylipin metabolism; P sig, phospholipid signaling; Pro GSP,prokaryotic galactolipid, sulfolipid, and phospholipid metabolism; Sph, sphingolipid biosynthesis; Suberin, suberin biosynthesis; TAG synth, TAGbiosynthesis.(C) Genes of prokaryotic and eukaryotic galactolipid, sulfolipid, and phospholipid metabolism pathways.(D) Genes involved in storage oil metabolism.

and temperature stimuli, and performed lipidomic analysis oftheir knockout mutant lines.

Mutant Phenotypes

The five genes selected for knockout mutant analysis were thoseencoding ketoacyl-ACP synthase II (FAB1/KASII); two enzymes ofthe prokaryotic glycerolipid biosynthesis pathway: digalactosyldia-cylglycerol synthase 2 (DGD2) and UDP-sulfoquinovose:DAG sul-foquinovosyltransferase (SQD2); and two enzymes of the eukaryoticglycerolipid biosynthesis pathway: long-chain acyl-CoA synthetase 4(LACS4) and putative CDP-DAG synthase (CDS2). LACS4 and KASIIrepresent dark-induced genes, SQD2 is light-specific, CDS2 exhibitsa heat-specific response, and DGD2 is cold-induced (Figure 5).

T-DNA insertion lines (Alonso and Stepanova, 2003) for thesegenes were obtained from the Nottingham Arabidopsis Stock Centrecollection (http://arabidopsis.info) and are listed in SupplementalTable 2. All selected mutant lines were different from those pub-lished previously for these genes, with the exception of LACS4lines, being identical to those described by Jessen et al. (2011). Inwild-type plants, transcripts of all five genes were found to ac-cumulate well within the range of the quantitative RT-PCR de-tection limit. In the mutant plants, no transcripts were detected infive biological replicates, with the exception of sqd2, where a 116.6-fold decrease in the SQD2 transcript level has been observed(Supplemental Table 3). Samples of the mutant plants, together withthe control wild type, were collected in standard nonstress con-ditions and were subjected to a lipidomic analysis using ultra-performance liquid chromatography coupled with Fourier transformmass spectrometry (Hummel et al., 2011). Obtained features(m/z at a certain retention time) were queried against an in-houselipid database, providing 119 annotated glycerolipid species(Supplemental Data Set 1). Because the mass spectrometry wasperformed only in positive ionization mode, the obtained dataset lacks phosphatidylserine and low-abundance molecular spe-cies of PI and PG. In total, 61 of the detected lipids overlappedwith those reported by Burgos et al. (2011). Samples from sixbiological replicates were measured for each mutant and controlwild-type line. The significance of the mutant lipidomic pheno-type was estimated by two-way ANOVA followed by the Tukeytest. The match between mutant lipidomic phenotype and stress-related changes in gene expression was estimated by Spearmancorrelation analysis (described in detail in Methods).

All of the mutants analyzed exhibited multiple significantchanges in their glycerolipid profiles, with independent lines ofLACS4 having very similar phenotypes (Figure 6; source dataare given in Supplemental Data Set 2). The strongest lipidomicphenotype was obtained in the cds2 mutant. Knockout of CDS2resulted in decreased general phospholipid:galactolipid ratio,drastic decreases in PI and PG, and depletion in certain PCsand PEs, including highly abundant 34:2 and 34:3 PC speciesand all molecular species of PE except the low-abundance 38:2and 38:3 PE. A general decrease in most molecular species ofSQDG was also observed, including a 2-fold change in the mostabundant 34:3 SQDG. At the same time, a general increase ingalactolipid level occurred, including the most abundant 34:6and 36:6 MGDG and 34:3 and 36:6 digalactosyldiacylglycerol.Taking into account that CDS2 is correlated with many genes

involved in storage oil biosynthesis and degradation (Figure 4D),the cds2mutant exhibits no effect on TAGs. To some extent, thekasII mutant exhibited an opposite effect: a severe depletionof almost all TAGs except the most abundant 54-C molecularspecies and a general decrease in galactolipids and sulfolipids,balanced by a slight accumulation of several phospholipids.Additionally, a general shift toward shorter acyl chains wasseen for galactolipids, marked by a significant decrease in 36-Cmolecular species of digalactosyldiacylglycerol and MGDG.Considering the relative abundance of measured glycerolipids,the most important effect of LACS4 knockout was the depletionof 36:6 DAG. This change reached a 2-fold decrease and hasbeen observed in both analyzed mutant lines; however, due tothe high deviation between replicates of the lacs4.2 line, thisresult is statistically significant only in lacs4.1. This change wasaccompanied by a drastic decrease in the content of low-abundance phospholipids and triacylglycerols: 32:3, 34:5, and 34:6PC, 32:3 PE, and 52:5, 52:6, 52:7, and 52:9 TAG, for which a com-mon structural element is the desaturated 16-C acyl chain. The effecton TAG is very similar to the one in kasII. The dgd2mutant exhibitedan interesting phenotype, showing an accumulation of several sat-urated PCs, including 32:1, 32:2, 34:1, and 36:1 molecular species,besides a significant accumulation of PE when taking all PEspecies into account. sqd2 exhibited an expected decrease inSQDG: the most abundant SQDG species decreased up to 2-fold;however, due to the high analytical noise, none of these changescrossed the 0.01 P value threshold. A parallel increase in PG wasobserved as well as a mixed effect on short-chain TAGs.

Match between Environmental Coordination andGenetic Control

The relationship between transcript–lipid correlation in changingenvironmental conditions and gene function was estimated bymatching the results of O2PLS analysis with lipidomic profiles of

Figure 5. Locations of the Selected Gene Candidates on the O2PLS-DACorrelation Loading Space.

The correlation loadings of lipids are denoted as triangles, and those oftranscripts are denoted as squares.

920 The Plant Cell

the selected gene knockout lines. For convenience, we call thetranscript–lipid coordination in the O2PLS model “environmentalcoordination” and the effect of the gene knockout on lipid levels“genetic control.”

The environmental coordination of a gene was calculated as thecorrelation between its transcript changes and the levels of all thelipids in the frame of the predictive variance structure (and thusthe correlation between the joint variance projection of the transcriptand the joint variance projections of all lipids). In this way, we esti-mated the relationship between changes of individual transcripts andlipids irrespective of the unique variation and residual noise. Thegenetic control was estimated as the effect of the gene knockout onthe lipid levels, represented as the mutant to wild-type log2 foldchange value. Because the knockout of a gene represents its radicaldownregulation, lipids depleted in themutant are treated as positivelydependent and those accumulated are treated as negatively de-pendent. Finally, the relationship between environmental coordinationand genetic control of the selected genes was scored by theSpearman correlation coefficient. A significant positive correlationwill result if the lipids that positively correlate with the gene of interestare also depleted in the respective mutant line or, analogously, thenegatively correlated lipids are accumulated in the mutant.

Among selected genes, KASII,DGD2, SQD2, andCDS2 exhibitedsignificant positive correlation between their environmental co-ordination and genetic control, whereas LACS4 showed significantnegative correlation (Figure 7). Although significant, the correlationcoefficients ranged from 0.24 to 0.4, indicating that the major por-tion of the variance is not explained by the single gene knockout (thesignificance of the Spearman correlation coefficient has been sup-ported by nonparametric bootstrap analysis; Supplemental Table 4).For KASII, exhibiting the highest correlation, the lipids mostly re-sponsible for the significant match are also those most significantlyaffected in the kasII mutant (i.e., KASII showed close coordinationwith the accumulation of highly desaturated phospholipids in dark-ness), which were conversely deficient in the knockout mutant. Theheat-related effect of CDS2 was positively related to the effect of itsknockout via the glycerolipid:phospholipid ratio. Inspection of thesum change of the lipid classes (Figure 4A) shows that the accu-mulation of MGDG and SQDG is associated with heat, whereas PG,PE, and PC accumulation is associated with cold. The same effectobserved upon the perturbation of CDS2 expression points to thegene as a strong candidate for acting as a regulator of temperature-related glycerolipid:phospholipid ratio changes. Additionally, mo-lecular species of PC and PE, identified as the most significantlyaffected lipids by the CDS2 knockout, exhibit the closest correlationwith the environmental coordination. A glycerolipid:phospholipidratio is also affected in DGD2 environmental and genetic phono-types. In this case, however, the effect of DGD2 knockout coincideswith the cold-related accumulation of phospholipids. Yet anotherscenario was seen for SQD2, which exhibited a significant matchbetween environmental coordination and genetic control for the bulkof lipids, with the exception of two specific lipids, 32:0 and 34:3 PG.Interestingly, LACS4 exhibited a significant but negative matchbetween environmental coordination and genetic control. This wasmainly related to the most affected phospholipids in lacs4, 34:6 and38:6 PC and 36:2 PE, but also occurs for the bulk of nonsignificantlyaffected lipids, indicating that even slight mutant-induced changesmay represent biologically relevant information.

Figure 6. Heat Map Representing Results of the Lipidomic Analysis ofSelected Knockout Mutant Lines.

The color represents the log2 value of the median mutant/wild-type foldchange. Asterisks in the heat map mark the significance of the change(***P < 0.001, 0.001 < **P < 0.01, 0.01 < *P < 0.1). The bar plot parallel tothe heat map represents the relative abundance of each lipid species inthe frame of the lipid class in wild-type plants (e.g., a value of 60% forMGDG 34:6 means that this molecular species of MGDG represents60% of the measured MGDG pool).

Integrative Analysis of Lipid Metabolism 921

DISCUSSION

We previously described the effects of the various temperature andlight treatments on lipid levels independently of the transcript data(Burgos et al., 2011). It is striking, therefore, that all of the describedeffects are encompassed in the lipid predictive structure of theO2PLSmodel, which includes only 27% of the original variance. Thereason for this is that the lipid predictive variance lacks the highplatform-specific technical noise observed in the lipid data set only,leaving only smooth changes clearly related to treatments and re-sponse development in time. Investigation of the lipid-unique andresidual variance (Supplemental Figures 5 and 6, respectively) in-dicates that, indeed, all of the nonpredictive variance accounts forchanges related neither to the treatment nor to the time dimensionof the response. A different situation is observed in the case oftranscripts. Here, two major effects discussed by Caldana et al.(2011), a dominance of gene expression changes induced by heat indark conditions and a halt of circadian changes in cold (describedalso in Espinoza et al., 2008), are largely encompassed in the uniquevariance structure. Similarly, the contrasting transcriptional changesin response to 32-L and 4-L conditions were identified as transcript-

unique. This demonstrates the advantage of the integrative ap-proach over analysis of the data blocks separately, as it (1) cleanedthe data from platform-specific analytical noise, allowing previouslyhindered biologically interpretable patterns to emerge, and (2) fo-cused the analysis on patterns shared between transcripts andlipids. Here, we discuss the coordination of lipid and gene expres-sion changes, with an emphasis on lipid metabolism gene expres-sion, which was not thoroughly covered by Caldana et al. (2011).

Parallel Changes of Lipids and Transcripts Reflect theReprogramming of Lipid Metabolism

The applied treatments mostly affected the saturation of acylchains and to a lesser extent the length of the acyl chains and thepools of the whole lipid classes. This is understandable, consid-ering the short time span of the experiment, which was probablyinsufficient to cover changes involving the displacement of largecarbon pools. As described by Burgos et al. (2011), changes in lightintensity, with the accumulation of desaturated lipid species andthe decrease of their less desaturated precursors in darkness (inparticular MGDG, PC, PE, and SQDG), might be directly related to

Figure 7. Scatterplots Showing the Match between Environmental Coordination and Genetic Control of the Analyzed Gene Candidates.

Environmental coordination is expressed as Spearman correlation values between joint variance projections of the gene expression and glycerolipids inall applied conditions. Genetic control is described by the log2 value of the median mutant/wild-type fold change. All lipid species are represented bytheir names plotted in gray; the significant ones, exceeding P > 0.1 in the mutant study, are plotted in black. Above each plot, the Spearman correlationcoefficient is provided (bootstrap statistics are given in Supplemental Table 4).

922 The Plant Cell

the inhibition of the de novo synthesis of FAs (Ohlrogge andJaworski, 1997) accompanied by the continuous activity of desa-turases (Browse et al., 1981). Temperature changes, on the otherhand, lead to effects that might be related to substrate specificityand shifts in substrate availability of the affected enzymes. Thisconcerns most importantly the desaturases FAD2, FAD3, and SSI2in heat and the phosphatidic acid binding protein TGD2 in coldtreatment. Finally, it is important to note that heat and cold do nottrigger the membrane fluidity compensation mechanism during thefirst 6 h of the stress response, which has been observed 3 d aftercold exposure of Arabidopsis rosettes (Welti et al., 2002) andBrassica napus cotyledons (Tasseva et al., 2004). This is an in-teresting observation considering that the dark-induced changeis rapid, showing that plants potentially can modify the satura-tion level of their membrane lipids within a few hours.

These lipid changes are accompanied by a reorganization of lipidmetabolism gene expression along three transcriptional programs,light-specific/temperature-independent, temperature-specific/light-independent, and intermediate light/temperature-modulated, eachspecific to a certain subset of lipid biochemical pathways (Figures4B to 4D). To highlight the implications of this observation, wediscuss the major pathways in detail.

FA Synthesis

Whereas FA synthesis is known to be light regulated at the level ofACCase by the redox mechanism (Kozaki and Sasaki, 1999), herewe observe that genes involved in FA synthesis are also coordinatelyregulated at the transcriptional level. This involves the dark-induceddownregulation of almost all key pathway enzymes, including ele-ments of the FA synthase complex: KASI, KASIII, hydroxyl-ACPdehydrase, and enoyl-ACP reductase ENR1; the biotin carboxylasesubunit of ACCase, malonyl-CoA:ACT malonyltransferase, the long-chain acyl-CoA synthetase LACS9, the acyl-ACP synthetase AAE15;and components of the pyruvate dehydrogenase complex: oneof the genes for b-PDH (At1g30120) and LPD2 (Figure 4B). In-terestingly, two acyl-ACP thioesterases, FatA and FatB, are an-tagonistically regulated by light. Whereas FatA, exhibiting higherspecificity to 18:0-ACP and 18:1-ACP substrates (Salas andOhlrogge, 2002), is strongly downregulated by dark treatment, FatB,exhibiting higher specificity toward shorter 16:0-ACP substrate, isupregulated in dark conditions. This coincides with the down-regulation of steaoryl-ACP desaturase FAB2, potentially further de-creasing the flux of ACP-FAs toward FatA. These concertedchanges might constitute a mechanism shifting glycerolipid bio-synthesis toward shorter acyl chains in response to carbon de-ficiency in dark conditions, as observed, for example, in seed oil inunfavorable nutrient and light conditions (Li et al., 2006; Ekman et al.,2007). The significant downregulation of the long-chain acyl-CoAsynthetase LACS9 suggests that, in addition to the shift towardshorter acyl chains, the FA efflux from the chloroplasts is inhibited,and thus prokaryotic glycerolipid synthesis is promoted. Due to thelack of positional data for the measured lipid molecular species, it isdifficult to estimate the change of ratio between lipids originatingfrom eukaryotic and prokaryotic pathways. On the other hand, darktreatment is positively correlated with the occurrence of 34-C lipidspecies (Supplemental Figure 4C), indicating a larger contribution ofshorter 16-C acyl chains.

Prokaryotic and Eukaryotic Glycerolipid Synthesis

Whereas the changes in FA elongation and desaturation sug-gest a shift toward prokaryotic glycerolipid synthesis, compari-son of the expression patterns between genes involved inprokaryotic and eukaryotic glycerolipid synthesis indicates theopposite (Figure 4C). Almost all genes of the prokaryotic path-way are strongly downregulated in response to darkness andupregulated in light conditions. Such coordination of the plas-tidial glycerolipid synthesis pathway has been observed pre-viously in plants harboring a mutated version of Escherichia coliGDPH gpsAFR insensitive to feedback inhibition (Shen et al.,2010). In these plants, exhibiting 3- to 4-fold higher levels of G3Pthan the wild type, a coordinated upregulation of the plastidialglycerolipid biosynthetic genes was observed. In darkness, how-ever, Gly-3-P has been shown to significantly decrease both in thenormal diurnal cycle (Gibon et al., 2006) and in prolonged nightconditions (Usadel et al., 2008). Whereas the correlation betweenGly-3-P accumulation and the expression of glycerolipid synthesisgenes in darkness gives a first indication of a possible causal re-lationship, the specific effect of feedback-insensitive GDPH mu-tants supports the argument that the regulatory mechanisms mightbe related to Gly-3-P.On the other hand, the eukaryotic pathway is in large part

dark-induced. This concerns, for example, genes involved ineukaryotic diacylglycerol synthesis (putative glycerol-3-phosphateacyltransferase, 1-acylglycerol-3-phosphate acyltransferase 4, andphosphatidate phosphatase 2) and four genes involved in acylediting: phospholipase A2 (LCAT-PLA) and three 1-acylglycerol-3-phosphocholine acyltransferases (LPLAT, LPLAT1, and LPEAT2).Acyl editing contributes largely to the incorporation of newlysynthesized FAs to PC in the sn-2 position and is a key step inthe synthesis of polyunsaturated TAG (Bates et al., 2007, 2009;Lu et al., 2009; Bates and Browse, 2012). Here, in conditions ofinhibited FA synthesis and in parallel with an increasing pool ofdesaturated glycerolipids, the activation of the acyl-editing mech-anism might play a role in the distribution of the increasingpool of polyunsaturated FAs, which, upon desaturation by FAD2and FAD3 in PC, return to the pool of acyl-CoA and enter thephospholipid synthesis pathway via 1,2-diacylglycerol-3-phosphate.This observation is supported by the dark-specific upregulationof LACS4 (the only LACS gene up-regulated in response todarkness).

Storage Lipid Metabolism

Although TAGs were not measured, it is possible that the ob-served coupling of light- and temperature-specific effects in thelipid data set originates from the parallel action of temperature-regulated storage lipid metabolism and mostly light-regulatedpathways of FA biosynthesis and glycerolipid metabolism.Genes of all the core enzymes of b-oxidation are specificallyupregulated in 21-D and 32-D conditions, including the TAGLlipase SDP1, the acyl-CoA oxidases ACX1, ACX2, and ACX4,and the main isoform of 3-ketoacyl-CoA thiolase (KAT2/PED1)(Germain et al., 2001; Carrie et al., 2007), with the exception ofthe multifunctional protein MFP2 gene, exhibiting an oppositebehavior. This is a surprising observation, taking into account

Integrative Analysis of Lipid Metabolism 923

that ACX1,2,3,4 and MFP2 have been shown to be coordinatelyupregulated during rapid oil breakdown in postgerminative growth(Eastmond and Graham, 2000; Rylott et al., 2001). The dark-relatedexpression of peroxisomal enoyl-CoA hydratase (ECH2) indicates anincreased degradation of unsaturated FAs (Goepfert et al., 2006).Similar to other pathways, the expression of genes involved in FAdegradation is boosted in heat stress and compromised by cold.Whereas the regulation of b-oxidation genes, including ACX, MFP2,and KAT2, has been described at the transcriptional and post-transcriptional levels during postgerminitive seedling growth (Hayashiet al., 1998; Richmond and Bleecker, 1999; Eastmond and Graham,2000; Eastmond et al., 2000; Rylott et al., 2003; Goepfert et al.,2005), our observations indicate that the regulation of the b-oxidationpathway genes is also coupled with light signaling. This is likely re-lated to the role of KAT2 in abscisic acid signaling (Jiang et al., 2011),since b-oxidation in darkness is a source of reactive oxygen species(Mittler, 2002). Coordinated changes in the expression of other genesinvolved in b-oxidation suggest that the transcriptional signal mightnot be limited to KAT2. Conversely, the key genes involved in TAGbiosynthesis exhibit minor light-related effects.

Identified Transcript–Lipid Relationships Are SignificantlyLinked to the Genetic Control of MembraneLipid Composition

Lipidomic analysis of the knockout mutant lines indicates a causalrelationship between observed transcript changes and changes inglycerolipid levels. Whereas in the case of LACS4 this relationshipis negative, indicating a more complex, context-dependent link,KASII, DGD2, SQD2, and CDS2 exhibit direct correlation betweentheir environmental coordination and genetic control.

KASII, responsible for a condensation reaction from palmitoyl-ACP to stearoyl-ACP (Huang et al., 1998), exhibits the mostsignificant match between its environmental coordination andgenetic control, coming mostly from the affected phospholipid:glycerolipid ratio. Two previously studied KASII knockout alleles,fab2-1 and fab2-2, exhibited a significant effect on FA abun-dance, decreasing the level of 18-C FAs and increasing the 16-C(Carlsson et al., 2002). This result was characterized in storageoil composition; however, no glycerolipid analysis has beenshown so far. The transcript data show that the upregulation ofKASII expression in darkness coincides with the promotion ofthe incorporation of 16-C acyl chains into glycerolipids, which isan opposite effect to the one expected from the data of Carlssonet al. (2002). This is an interesting observation, suggesting a newputative function for the transcriptional regulation of KASII inchanging light conditions. The match between the environ-mental and genetic phenotypes of DGD2 is mainly due to theaccumulation of specific saturated phospholipids, although, inagreement with Kelly et al. (2003), the phospholipid:glycerolipidratio is not affected in the dgd2 mutant. The accumulation ofsaturated PE and PC might indicate that temperature-dependentregulation of DGD2 expression is involved in the partial redirectionof the de novo–synthesized FAs to phospholipids. In the case ofSQD2, the knockout mutant phenotype (across all the SQDGspecies, although not highly significant) matches the decrease inSQDG content in dark conditions, where SQD2 gene expressionis downregulated. A previous study reported the complete absence

of sulfolipids in the sqd2mutant, indicating that our line containsa functional protein, although probably with decreased activity(Yu et al., 2002). In that case, SQD2 provides a perfect exampleof a clear lipidomic phenotype occurring both upon environmen-tally induced SQD2 gene expression changes and upon reductionof SQD2 expression/activity upon genetic perturbation. Finally, thephonotype of the cds2 mutant and its heat-specific response in-dicate that CDS2might be a key enzyme involved in the temperature-related change in phospholipid:glycerolipid ratio. CDS2 is one offive cytidine diphosphate diacylglycerol synthases (Beisson et al.,2003) and is responsible for transfer of the cytidyl group fromcytidine triphosphate (CTP) to phosphatidic acid and the forma-tion of CTP-DAG, a substrate for PI and PG biosynthesis.Although not perfect, the described match between environ-

mental coordination and genetic control is noteworthy, especiallytaking into account that (1) our study included only two environ-mental parameters; (2) gene knockouts are harsh genetic per-turbations and often lead to strong pleiotropic effects, making itdifficult to associate an observed phenotype with the gene func-tion; and (3) metabolic pathways exhibit a remarkable ability tocompensate the effects of single gene perturbations (Papp et al.,2004). It is important to note, however, that the selected genesrepresent only a limited data set. Thus, the knockout mutantanalysis should be regarded as a case study and an attempt toevaluate the usefulness of O2PLS analysis for predicting singlegene function and proposing biologically sound hypotheses.

Future Challenges

In contrast to other integrative studies focused on the identification ofpairwise metabolite–gene relationships (Nikiforova et al., 2005; Rischeret al., 2006; Hirai et al., 2007), here we described lipid–transcript as-sociations at the systems level and supported our findings by testinga set of gene knockout lines. Such system-scale integration pre-viously has been shown to be a successful strategy for investigatinghousekeeping metabolic processes and for considering various typesof data and different computational methods (Wentzell et al., 2007;Kerwin et al., 2011). We believe that our analysis is an important stepin the emerging field of systems biology of plant lipid metabolism, andwe expect that this work, complemented by other system-scalestudies, such as high-resolution genome-wide association mappingand large-scale functional genomics, will give an even deeper in-sight into the principles of plant lipid metabolism regulation.

METHODS

Environmental Treatment Experiment

For the details on plant growth conditions, applied treatments, and samplingprocedures, see Caldana et al. (2011). The details on microarray and lipidomicanalyses are given in Caldana et al. (2011) and Burgos et al. (2011), re-spectively. Genes involved in lipid metabolism were selected according to theARALIP database (http://aralip.plantbiology.msu.edu; version from January2013) (Li-Beisson et al., 2013).

Knockout Mutant Lines: Selection, Genotyping, andGrowth Conditions

Arabidopsis thaliana Columbia-0 ecotype (wild-type) plants were used ascontrols throughout this work. Arabidopsis T-DNA insertion lines, SALK

924 The Plant Cell

and GABI-KAT (Alonso and Stepanova, 2003; Kleinboelting et al., 2012),were obtained from the Nottingham Arabidopsis Stock Centre collection(http://arabidopsis.info). Plants were selected on plates supplementedwith appropriate antibiotics in order to obtain nonsegregating homozy-gous lines. See Supplemental Table 2 for the list of mutant lines andgenotyping and gene expression primers. Quantitative PCR analysis ofthe mutants lines was performed as described by Brotman et al. (2013)with gene-specific primers after the insertion. The mutant lines (six in-dependent biological replicates for each line) were grown in trays undera short-day regime (22°C, 8 h of light and 16 h of dark). Each traycontained in addition six wild-type plants, and all the plants were ran-domly placed. Four-week-old plants (rosette leaves) were collected andshock frozen in liquid nitrogen for subsequent lipidomic analysis. Allbatches (mutants and wild-type plants) were subsequently extracted andprocessed together.

Lipid Profiling of the Mutant Lines

Samples were processed using ultra-performance liquid chroma-tography with a C8 reverse-phase column coupled with an Exactivemass spectrometer (Thermo-Fisher; http://www.thermofisher.com) inpositive ionization mode. Processing of chromatograms, peak de-tection, and integration were performed using REFINER MS 7.5(GeneData; http://www.genedata.com). Processing of mass spec-trometry data included the removal of the fragmentation information,isotopic peaks, as well as chemical noise. The obtained features (m/zat a certain retention time) were queried against an in-house lipiddatabase for further annotation (details on compound annotation aregiven in Supplemental Data Set 3).

Data Integration

O2PLS analysis (Trygg, 2002; Trygg and Wold, 2003) of the transcript andlipid data was performed using the algorithm provided by Bylesjö et al.(2007). All the calculations were performed using R (http://www.R-project.org) (R Core Team, 2013) and the pcaMethods (Stacklies et al., 2007) andpls (Mevik and Wehrens, 2007) packages. In all steps of the O2PLS al-gorithm that required the extraction of principal components, the singularvalue decomposition method was used. Below we describe briefly themethod principle.

O2PLS is a bidirectional multivariate regression method that aims toseparate the covariance between two data sets (it was recently extendedto multiple data sets) (Löfstedt and Trygg, 2011; Löfstedt et al., 2012) fromthe systematic sources of variance being specific for each data setseparately. In principle, it identifies a set of components P being commonfor X and Y and, as such, it represents the joint variance between bothdata sets. By removing this joint variance, it is possible to identify or-thogonal components PX and PY specific for each respective data set. TheO2PLS model for X and Y data sets can be written as:

X ¼ TWT þ TY 2 ortho PTY 2ortho þ E ð1Þ

Y ¼ UCT þ UX2 ortho PTX2ortho þ F ð2Þ

U ¼ T þ H ð3Þwhere P and U are score matrices for X and Y, respectively, W and C arethe joint variance component matrices, and E and F are the residualmatrices. PY 2ortho and TY 2ortho are the Y-orthogonal loadings and scorematrices, respectively. Analogously, PX2ortho andUX2ortho are the X-orthogonalloadings and scorematrices, respectively. Equation 3 describes the inner relationbetween U and T, where H is a residual matrix. Predictive equations for X and Yare then:

X ¼ UWT     and    Y ¼ TCT ð4ÞIn our analysis, the transcript data set (480 genes3 152 samples) and the lipiddata set (92 lipids 3 152 average values of three biological replicates) werereferred to as the X and Y matrices, respectively. Prior to the O2PLS analysis,both lipid and transcript data sets were column-wise mean-centered andscaled to unit variance. Subsequently, in order to ensure that transcripts andlipids have equal weight, both data sets were scaled to give a total sum ofsquares of 1. In order to avoid overfitting of the model, the optimal number oflatent variables for each model structure was estimated using group-balancedMCCV. For each combination of the latent variable’s number (between 4 and 15for the predictive structures and 1 to 10 for the orthogonal structures), 10-foldMCCV was performed. From these 10 permutations, the average Q2 valueswere calculated as:

Q2 ¼ 12 err=Vartotal ð5Þwhere

err ¼ Varexpected 2Varestimated ð6ÞFor the top 10 models, 10-fold MCCV was performed 50 times and new av-erage Q2 values were calculated. From three best almost equally scoredmodels with the lowest generalization error, the one with the lowest number oflatent variables was chosen (Supplemental Table 1). The residual variancestructures were analyzed by PCA, showing no treatment-related effects, al-though several samples in the lipid data set could be described as technicaloutliers (Supplemental Figure 6).

TheO2PLS-DAanalysiswasperformedasdescribedbyBylesjö et al. (2007);briefly, the O2PLS predictive variation [TWT, UCT] was used for a subsequentO2PLS-DA analysis (additional O2PLS-DA analysis of transcript- and lipid-unique variations is shown in Supplemental Figure 7) . O2PLS-DA models withone to eight latent variables were tested, but no significant improvement, interms of multiple independent light- or temperature-specific effects, was ob-served for n > 1 (Supplemental Figure 8). In order to improve the visualization ofthe light- and temperature-specific effects, the O2PLS-DA results (projection ofthe predictive variation [TWT, UCT] on the first latent variable of the respectiveO2PLS-DA model) were regressed onto the light and temperature variablesusing partial least squares regression. Theweights for the light and temperaturevariableswerechosenequidistant,meaning that thedistancebetweendarknessand low light was set equal to the distance between low light and control lightand the distance between normal light and high light.

Generalized Lipid and Transcript Trends

The analytical method for lipid quantification allows extracting only relativeinformation about the lipid changes, and the comparison of abundances ofparticular lipid species is additionally possible only in the frame of a singleglycerolipid class. Therefore, when summing up the changes in the frame ofa single lipid class (as in Figure 4A), the same for a particular number of doublebonds or carbons in the acyl chains is an approximation (hence, the intensitiesof the lipid species of different classes are summed). This approximation isperformed by within-class normalization of the species average abundance.The averages of the species abundances across all the conditions have beendivided by their sum, such that they sum up to 1. Newly obtained values wereused to normalize the intensity values across the treatments. As a result, theweight of each class in the computation of general saturation and carbonnumber trends becomes equal.

Supplemental Data

The following materials are available in the online version of this article.

Supplemental Figure 1. Cross-Validation of the O2PLS Model.

Integrative Analysis of Lipid Metabolism 925

Supplemental Figure 2. Significance of the Transcript and LipidChanges in the Frame of Lipid Metabolism Pathways.

Supplemental Figure 3. Top Five Latent Variables of the O2PLSPredictive Variance Structure.

Supplemental Figure 4. O2PLS-DA Correlation Loading Plots.

Supplemental Figure 5. Top Four Latent Variables of the Lipid- andTranscript-Unique Variance Structures.

Supplemental Figure 6. PCA of the Lipid and Transcript ResidualVariance.

Supplemental Figure 7. O2PLS-DA of the Transcript- and Lipid-Unique Variance Structures.

Supplemental Figure 8. First Five Latent Variables of O2PLS-DAModels Calculated for Light and Temperature Groups.

Supplemental Table 1. Top Scoring O2PLS Models and TheirGeneralization Error for the Transcript and Lipid Data.

Supplemental Table 2. List of Primer Sequences for Mutant Geno-typing and Quantitative PCR Analyses.

Supplemental Table 3. Verification of Gene Knockout by ExpressionAnalysis Using Quantitative RT-PCR.

Supplemental Table 4. Bootstrap Statistics for Figure 7.

Supplemental Data Set 1. Lipidomic Analysis of Selected KnockoutMutant Lines.

Supplemental Data Set 2. Median Fold Change with Respect to theWild-Type Control (Source Data for Figure 6).

Supplemental Data Set 3. Metabolite Reporting Checklist andCompound Annotation.

ACKNOWLEDGMENTS

We thank Ke Xu, Izabela Sierzputowska, and Aenne Eckardt for technicalassistance, Asdrubal Burgos for discussions, and Elmien Heyneke forhelp with writing.

AUTHOR CONTRIBUTIONS

J.S., A.C.-I., and Y.B. performed the analysis. J.S. wrote the article. L.W.and A.C.-I. supervised the work.

Received September 22, 2013; revised January 27, 2014; acceptedFebruary 17, 2014; published March 18, 2014.

REFERENCES

Alonso, J.M., and Stepanova, A.N. (2003). T-DNA mutagenesis inArabidopsis. Methods Mol. Biol. 236: 177–188.

Avin-Wittenberg, T., Tzin, V., Less, H., Angelovici, R., and Galili, G.(2011). A friend in need is a friend indeed: Understanding stress-associated transcriptional networks of plant metabolism usingcliques of coordinately expressed genes. Plant Signal. Behav. 6:1294–1296.

Bates, P.D., and Browse, J. (2012). The significance of differentdiacylgycerol synthesis pathways on plant oil composition andbioengineering. Front Plant Sci 3: 147.

Bates, P.D., Durrett, T.P., Ohlrogge, J.B., and Pollard, M. (2009). Analysisof acyl fluxes through multiple pathways of triacylglycerol synthesis indeveloping soybean embryos. Plant Physiol. 150: 55–72.

Bates, P.D., Ohlrogge, J.B., and Pollard, M. (2007). Incorporation ofnewly synthesized fatty acids into cytosolic glycerolipids in pealeaves occurs via acyl editing. J. Biol. Chem. 282: 31206–31216.

Baud, S., and Lepiniec, L. (2010). Physiological and developmentalregulation of seed oil production. Prog. Lipid Res. 49: 235–249.

Beisson, F., et al. (2003). Arabidopsis genes involved in acyl lipidmetabolism: A 2003 census of the candidates, a study of thedistribution of expressed sequence tags in organs, and a web-based database. Plant Physiol. 132: 681–697.

Bernard, A., and Joubès, J. (2013). Arabidopsis cuticular waxes:Advances in synthesis, export and regulation. Prog. Lipid Res. 52:110–129.

Brotman, Y., Landau, U., Cuadros-Inostroza, A., Tohge, T., Fernie,A.R., Chet, I., Viterbo, A., and Willmitzer, L. (2013). Trichoderma-plant root colonization: escaping early plant defense responses andactivation of the antioxidant machinery for saline stress tolerance.PLoS Pathog. 9: e1003221.

Browse, J., Roughan, P.G., and Slack, C.R. (1981). Light control offatty acid synthesis and diurnal fluctuations of fatty acid composition inleaves. Biochem. J. 196: 347–354.

Burgos, A., Szymanski, J., Seiwert, B., Degenkolbe, T., Hannah,M.A., Giavalisco, P., and Willmitzer, L. (2011). Analysis of short-termchanges in the Arabidopsis thaliana glycerolipidome in response totemperature and light. Plant J. 66: 656–668.

Bylesjö, M., Eriksson, D., Kusano, M., Moritz, T., and Trygg, J.(2007). Data integration in plant biology: The O2PLS method forcombined modeling of transcript and metabolite data. Plant J. 52:1181–1191.

Bylesjö, M., Nilsson, R., Srivastava, V., Grönlund, A., Johansson,A.I., Jansson, S., Karlsson, J., Moritz, T., Wingsle, G., and Trygg, J.(2009). Integrated analysis of transcript, protein and metabolitedata to study lignin biosynthesis in hybrid aspen. J. Proteome Res.8: 199–210.

Bylesjö, M., Rantalainen, M., Cloarec, O., Nicholson, J.K., Holmes,E., and Trygg, J. (2006). OPLS discriminant analysis: Combiningthe strengths of PLS-DA and SIMCA classification. J. Chemometr.20: 341–351.

Caldana, C., Degenkolbe, T., Cuadros-Inostroza, A., Klie, S.,Sulpice, R., Leisse, A., Steinhauser, D., Fernie, A.R., Willmitzer,L., and Hannah, M.A. (2011). High-density kinetic analysis of themetabolomic and transcriptomic response of Arabidopsis to eightenvironmental conditions. Plant J. 67: 869–884.

Carlsson, A.S., LaBrie, S.T., Kinney, A.J., von Wettstein-Knowles,P., and Browse, J. (2002). A KAS2 cDNA complements thephenotypes of the Arabidopsis fab1 mutant that differs in a singleresidue bordering the substrate binding pocket. Plant J. 29:761–770.

Carrie, C., Murcha, M.W., Millar, A.H., Smith, S.M., and Whelan, J.(2007). Nine 3-ketoacyl-CoA thiolases (KATs) and acetoacetyl-CoAthiolases (ACATs) encoded by five genes in Arabidopsis thaliana aretargeted either to peroxisomes or cytosol but not to mitochondria.Plant Mol. Biol. 63: 97–108.

Chen, J., Burke, J.J., Xin, Z., Xu, C., and Velten, J. (2006).Characterization of the Arabidopsis thermosensitive mutant atts02reveals an important role for galactolipids in thermotolerance. PlantCell Environ. 29: 1437–1448.

Consonni, R., Cagliani, L.R., Stocchero, M., and Porretta, S. (2010).Evaluation of the production year in Italian and Chinese tomatopaste for geographical determination using O2PLS models. J. Agric.Food Chem. 58: 7520–7525.

926 The Plant Cell

Degenkolbe, T., Giavalisco, P., Zuther, E., Seiwert, B., Hincha, D.K.,and Willmitzer, L. (2012). Differential remodeling of the lipidome duringcold acclimation in natural accessions of Arabidopsis thaliana. Plant J.72: 972–982.

De Palma, M., Grillo, S., Massarelli, I., Costa, A., Balogh, G., Vigh, L.,and Leone, A. (2008). Regulation of desaturase gene expression,changes in membrane lipid composition and freezing tolerance in potatoplants. Mol. Breed. 21: 15–26.

Du, Z.Y., Xiao, S., Chen, Q.F., and Chye, M.L. (2010a). Depletion ofthe membrane-associated acyl-coenzyme A-binding protein ACBP1enhances the ability of cold acclimation in Arabidopsis. Plant Physiol.152: 1585–1597.

Du, Z.Y., Xiao, S., Chen, Q.F., and Chye, M.L. (2010b). Arabidopsisacyl-CoA-binding proteins ACBP1 and ACBP2 show different rolesin freezing stress. Plant Signal. Behav. 5: 607–609.

Eastmond, P.J., and Graham, I.A. (2000). The multifunctional proteinAtMFP2 is co-ordinately expressed with other genes of fatty acidbeta-oxidation during seed germination in Arabidopsis thaliana (L.)Heynh. Biochem. Soc. Trans. 28: 95–99.

Eastmond, P.J., Hooks, M.A., Williams, D., Lange, P., Bechtold, N.,Sarrobert, C., Nussaume, L., and Graham, I.A. (2000). Promotertrapping of a novel medium-chain acyl-CoA oxidase, which isinduced transcriptionally during Arabidopsis seed germination. J.Biol. Chem. 275: 34375–34381.

Ekman, A., Bülow, L., and Stymne, S. (2007). Elevated atmosphericCO2 concentration and diurnal cycle induce changes in lipidcomposition in Arabidopsis thaliana. New Phytol. 174: 591–599.

Espinoza, C., Bieniawska, Z., Hincha, D.K., and Hannah, M.A.(2008). Interactions between the circadian clock and cold-responsein Arabidopsis. Plant Signal. Behav. 3: 593–594.

Frentzen, M. (2004). Phosphatidylglycerol and sulfoquinovosyldiacylglycerol:Anionic membrane lipids and phosphate regulation. Curr. Opin. Plant Biol.7: 270–276.

Germain, V., Rylott, E.L., Larson, T.R., Sherson, S.M., Bechtold, N.,Carde, J.P., Bryce, J.H., Graham, I.A., and Smith, S.M. (2001).Requirement for 3-ketoacyl-CoA thiolase-2 in peroxisome development,fatty acid beta-oxidation and breakdown of triacylglycerol in lipid bodiesof Arabidopsis seedlings. Plant J. 28: 1–12.

Gibon, Y., Usadel, B., Blaesing, O.E., Kamlage, B., Hoehne, M.,Trethewey, R., and Stitt, M. (2006). Integration of metabolite withtranscript and enzyme activity profiling during diurnal cycles inArabidopsis rosettes. Genome Biol. 7: R76.

Goepfert, S., Hiltunen, J.K., and Poirier, Y. (2006). Identification andfunctional characterization of a monofunctional peroxisomal enoyl-CoA hydratase 2 that participates in the degradation of even cis-unsaturated fatty acids in Arabidopsis thaliana. J. Biol. Chem. 281:35894–35903.

Goepfert, S., Vidoudez, C., Rezzonico, E., Hiltunen, J.K., andPoirier, Y. (2005). Molecular identification and characterization ofthe Arabidopsis D3,5,D2,4-dienoyl-coenzyme A isomerase, a peroxisomalenzyme participating in the b-oxidation cycle of unsaturated fatty acids.Plant Physiol. 138: 1947–1956.

Hayashi, M., Toriyama, K., Kondo, M., and Nishimura, M. (1998).2,4-Dichlorophenoxybutyric acid–resistant mutants of Arabidopsis havedefects in glyoxysomal fatty acid b-oxidation. Plant Cell 10: 183–195.

Hirai, M.Y., and Saito, K. (2004). Post-genomics approaches for theelucidation of plant adaptive mechanisms to sulphur deficiency. J.Exp. Bot. 55: 1871–1879.

Hirai, M.Y., et al. (2007). Omics-based identification of ArabidopsisMyb transcription factors regulating aliphatic glucosinolate biosynthesis.Proc. Natl. Acad. Sci. USA 104: 6478–6483.

Hummel, J., Segu, S., Li, Y., Irgang, S., Jueppner, J., andGiavalisco, P. (2011). Ultra performance liquid chromatography and

high resolution mass spectrometry for the analysis of plant lipids.Front. Plant Sci. 2: 54.

Horvath, I., Vigh, L., Vanhasselt, P.R., Woltjes, J., and Kuiper, P.J.C.(1983). Lipid-composition in leaves of cucumber genotypes as affectedby different temperature regimes and grafting. Physiol. Plant. 57:532–536.

Huang, W., Jia, J., Edwards, P., Dehesh, K., Schneider, G., andLindqvist, Y. (1998). Crystal structure of beta-ketoacyl-acyl carrierprotein synthase II from E. coli reveals the molecular architecture ofcondensing enzymes. EMBO J. 17: 1183–1191.

Jessen, D., Olbrich, A., Knufer, J., Kruger, A., Hoppert, M., Polle,A., and Fulda, M. (2011). Combined activity of LACS1 and LACS4 isrequired for proper pollen coat formation in Arabidopsis. Plant J. 68:715–726.

Jiang, T., Zhang, X.F., Wang, X.F., and Zhang, D.P. (2011).Arabidopsis 3-ketoacyl-CoA thiolase-2 (KAT2), an enzyme of fattyacid b-oxidation, is involved in ABA signal transduction. Plant CellPhysiol. 52: 528–538.

Jozefczuk, S., Klie, S., Catchpole, G., Szymanski, J., Cuadros-Inostroza, A., Steinhauser, D., Selbig, J., and Willmitzer, L. (2010).Metabolomic and transcriptomic stress response of Escherichia coli.Mol. Syst. Biol. 6: 364.

Kelly, A.A., Froehlich, J.E., and Dormann, P. (2003). Disruption ofthe two digalactosyldiacylglycerol synthase genes DGD1 and DGD2in Arabidopsis reveals the existence of an additional enzyme ofgalactolipid synthesis. Plant Cell 15: 2694–2706.

Kerwin, R.E., Jimenez-Gomez, J.M., Fulop, D., Harmer, S.L.,Maloof, J.N., and Kliebenstein, D.J. (2011). Network quantitativetrait loci mapping of circadian clock outputs identifies metabolicpathway-to-clock linkages in Arabidopsis. Plant Cell 23: 471–485.

Kleinboelting, N., Huep, G., Kloetgen, A., Viehoever, P., andWeisshaar, B. (2012). GABI-Kat SimpleSearch: New features of theArabidopsis thaliana T-DNA mutant database. Nucleic Acids Res.40: D1211–D1215.

Kozaki, A., and Sasaki, Y. (1999). Light-dependent changes in redoxstatus of the plastidic acetyl-CoA carboxylase and its regulatorycomponent. Biochem. J. 339: 541–546.

Kusano, M., Jonsson, P., Fukushima, A., Gullberg, J., Sjöström, M.,Trygg, J., and Moritz, T. (2011). Metabolite signature during short-dayinduced growth cessation in Populus. Front. Plant Sci. 2: 29.

Larkindale, J., and Huang, B.R. (2004). Changes of lipid composition andsaturation level in leaves and roots for heat-stressed and heat-acclimatedcreeping bentgrass (Agrostis stolonifera). Environ. Exp. Bot. 51: 57–67.

Li, P., Sioson, A., Mane, S.P., Ulanov, A., Grothaus, G., Heath, L.S.,Murali, T.M., Bohnert, H.J., and Grene, R. (2006). Responsediversity of Arabidopsis thaliana ecotypes in elevated [CO2] in thefield. Plant Mol. Biol. 62: 593–609.

Li, W., Wang, R., Li, M., Li, L., Wang, C., Welti, R., and Wang, X.(2008). Differential degradation of extraplastidic and plastidic lipidsduring freezing and post-freezing recovery in Arabidopsis thaliana.J. Biol. Chem. 283: 461–468.

Li-Beisson, Y., et al. (2013). Acyl-lipid metabolism. The ArabidopsisBook 11: e0161, doi/10.1199/tab.0161.

Löfstedt, T., and Trygg, J. (2011). OnPLS—A novel multiblockmethod for the modelling of predictive and orthogonal variation. J.Chemometr. 25: 441–455.

Löfstedt, T., Hanafi, M., Mazerolles, G., and Trygg, J. (2012). OnPLSpath modelling. Chemom. Intell. Lab. Syst. 118: 139–149.

Loraine, A. (2009). Co-expression analysis of metabolic pathways inplants. Methods Mol. Biol. 553: 247–264.

Lu, C., Xin, Z., Ren, Z., Miquel, M., and Browse, J. (2009). An enzymeregulating triacylglycerol composition is encoded by the ROD1 gene ofArabidopsis. Proc. Natl. Acad. Sci. USA 106: 18837–18842.

Integrative Analysis of Lipid Metabolism 927

Mevik, B., and Wehrens, R. (2007). The pls package: Principal componentand partial least squares regression in R. J. Stat. Softw. 18: 1–24.

Mittler, R. (2002). Oxidative stress, antioxidants and stress tolerance.Trends Plant Sci. 7: 405–410.

Moellering, E.R., and Benning, C. (2011). Galactoglycerolipidmetabolism under stress: A time for remodeling. Trends Plant Sci.16: 98–107.

Nikiforova, V.J., Daub, C.O., Hesse, H., Willmitzer, L., andHoefgen, R. (2005). Integrative gene-metabolite network withimplemented causality deciphers informational fluxes of sulphurstress response. J. Exp. Bot. 56: 1887–1896.

Obayashi, T., Kinoshita, K., Nakai, K., Shibaoka, M., Hayashi, S.,Saeki, M., Shibata, D., Saito, K., and Ohta, H. (2007). ATTED-II: Adatabase of co-expressed genes and cis elements for identifyingco-regulated gene groups in Arabidopsis. Nucleic Acids Res. 35:D863–D869.

Ohlrogge, J.B., and Jaworski, J.G. (1997). Regulation of fatty acidsynthesis. Annu. Rev. Plant Physiol. Plant Mol. Biol. 48: 109–136.

Papp, B., Pál, C., and Hurst, L.D. (2004). Metabolic network analysisof the causes and evolution of enzyme dispensability in yeast.Nature 429: 661–664.

R Core Team (2013). R: A Language and Environment for StatisticalComputing (Vienna: R Foundation for Statistical Computing).

Rajashekar, C.B., Zhou, H.E., Zhang, Y., Li, W., and Wang, X.(2006). Suppression of phospholipase Dalpha1 induces freezingtolerance in Arabidopsis: Response of cold-responsive genes andosmolyte accumulation. J. Plant Physiol. 163: 916–926.

Redestig, H., Szymanski, J., Hirai, M.Y., Selbig, J., Willmitzer, L.,Nikoloski, Z., and Saito, K. (2011). Data integration, metabolicnetworks and systems biology. Annu. Plant Rev. 43: 261–316.

Richmond, T.A., and Bleecker, A.B. (1999). A defect in b-oxidationcauses abnormal inflorescence development in Arabidopsis. PlantCell 11: 1911–1924.

Rischer, H., Oresic, M., Seppänen-Laakso, T., Katajamaa, M.,Lammertyn, F., Ardiles-Diaz, W., Van Montagu, M.C., Inzé, D.,Oksman-Caldentey, K.M., and Goossens, A. (2006). Gene-to-metabolite networks for terpenoid indole alkaloid biosynthesis inCatharanthus roseus cells. Proc. Natl. Acad. Sci. USA 103: 5614–5619.

Rylott, E.L., Hooks, M.A., and Graham, I.A. (2001). Co-ordinateregulation of genes involved in storage lipid mobilization inArabidopsis thaliana. Biochem. Soc. Trans. 29: 283–287.

Rylott, E.L., Rogers, C.A., Gilday, A.D., Edgell, T., Larson, T.R., andGraham, I.A. (2003). Arabidopsis mutants in short- and medium-chain acyl-CoA oxidase activities accumulate acyl-CoAs and revealthat fatty acid beta-oxidation is essential for embryo development.J. Biol. Chem. 278: 21370–21377.

Sakamoto, A., Sulpice, R., Hou, C.X., Kinoshita, M., Higashi, S.I.,Kanaseki, T., Nonaka, H., Moon, B.Y., and Murata, N. (2004).Genetic modification of the fatty acid unsaturation of phosphatidylglycerolin chloroplasts alters the sensitivity of tobacco plants to cold stress. PlantCell Environ. 27: 99–105.

Salas, J.J., and Ohlrogge, J.B. (2002). Characterization of substratespecificity of plant FatA and FatB acyl-ACP thioesterases. Arch.Biochem. Biophys. 403: 25–34.

Shen, W., Li, J.Q., Dauk, M., Huang, Y., Periappuram, C., Wei, Y.,and Zou, J. (2010). Metabolic and transcriptional responses ofglycerolipid pathways to a perturbation of glycerol 3-phosphatemetabolism in Arabidopsis. J. Biol. Chem. 285: 22957–22965.

Stacklies, W., Redestig, H., Scholz, M., Walther, D., and Selbig, J.(2007). pcaMethods—A Bioconductor package providing PCAmethods for incomplete data. Bioinformatics 23: 1164–1167.

Stitt, M. (1986). Limitation of photosynthesis by carbon metabolism. I.Evidence for excess electron transport capacity in leaves carrying outphotosynthesis in saturating light and CO2. Plant Physiol. 81: 1115–1122.

Tasseva, G., de Virville, J.D., Cantrel, C., Moreau, F., andZachowski, A. (2004). Changes in the endoplasmic reticulum lipidproperties in response to low temperature in Brassica napus. PlantPhysiol. Biochem. 42: 811–822.

Trygg, J. (2002). O2-PLS for qualitative and quantitative analysis inmultivariate calibration. J. Chemometr. 16: 283–293.

Trygg, J., and Wold, S. (2003). O2-PLS, a two-block (X-Y) latentvariable regression (LVR) method with an integral OSC filter. J.Chemometr. 17: 53–64.

Urbanczyk-Wochniak, E., Luedemann, A., Kopka, J., Selbig, J.,Roessner-Tunali, U., Willmitzer, L., and Fernie, A.R. (2003).Parallel analysis of transcript and metabolic profiles: A newapproach in systems biology. EMBO Rep. 4: 989–993.

Usadel, B., Bläsing, O.E., Gibon, Y., Retzlaff, K., Höhne, M.,Günther, M., and Stitt, M. (2008). Global transcript levels respondto small changes of the carbon status during progressive exhaustionof carbohydrates in Arabidopsis rosettes. Plant Physiol. 146: 1834–1861.

Welti, R., Li, W., Li, M., Sang, Y., Biesiada, H., Zhou, H.E.,Rajashekar, C.B., Williams, T.D., and Wang, X. (2002). Profilingmembrane lipids in plant stress responses: Role of phospholipase Dalpha in freezing-induced lipid changes in Arabidopsis. J. Biol.Chem. 277: 31994–32002.

Wentzell, A.M., Rowe, H.C., Hansen, B.G., Ticconi, C., Halkier, B.A.,and Kliebenstein, D.J. (2007). Linking metabolic QTLs with networkand cis-eQTLs controlling biosynthetic pathways. PLoS Genet. 3:1687–1701.

Williams, J.P., Khan, M.U., Mitchell, K., and Johnson, G. (1988). Theeffect of temperature on the level and biosynthesis of unsaturatedfatty acids in diacylglycerols of Brassica napus leaves. PlantPhysiol. 87: 904–910.

Yoshida, S., and Sakai, A. (1974). Phospholipid degradation in frozenplant cells associated with freezing injury. Plant Physiol. 53: 509–511.

Yu, B., Xu, C., and Benning, C. (2002). Arabidopsis disrupted inSQD2 encoding sulfolipid synthase is impaired in phosphate-limitedgrowth. Proc. Natl. Acad. Sci. USA 99: 5732–5737.

Zamboni, A., et al. (2010). Identification of putative stage-specificgrapevine berry biomarkers and omics data integration into networks.Plant Physiol. 154: 1439–1459.

928 The Plant Cell

DOI 10.1105/tpc.113.118919; originally published online March 18, 2014; 2014;26;915-928Plant Cell

Jedrzej Szymanski, Yariv Brotman, Lothar Willmitzer and Álvaro Cuadros-InostrozaArabidopsisLinking Gene Expression and Membrane Lipid Composition of

 This information is current as of June 7, 2020

 

Supplemental Data /content/suppl/2014/03/10/tpc.113.118919.DC1.html

References /content/26/3/915.full.html#ref-list-1

This article cites 83 articles, 29 of which can be accessed free at:

Permissions https://www.copyright.com/ccc/openurl.do?sid=pd_hw1532298X&issn=1532298X&WT.mc_id=pd_hw1532298X

eTOCs http://www.plantcell.org/cgi/alerts/ctmain

Sign up for eTOCs at:

CiteTrack Alerts http://www.plantcell.org/cgi/alerts/ctmain

Sign up for CiteTrack Alerts at:

Subscription Information http://www.aspb.org/publications/subscriptions.cfm

is available at:Plant Physiology and The Plant CellSubscription Information for

ADVANCING THE SCIENCE OF PLANT BIOLOGY © American Society of Plant Biologists