Introduction
The recent development of high-density technologies has enabled a dramatic extension of the exploration of living organisms. Such exploration typically consists in the parallel measurement of large numbers of compounds, at different developmental stages or following the application of contrasted environmental stimuli. Correlation analysis or more sophisticated statistical approaches are then used to identify clusters of co-varying compounds with the aim of reconstructing the underlying regulatory network governing system responses. Such approaches are now routinely used to identify predictive biomarkers (Steinfath et al., 2010; Riedelsheimer et al., 2012), including candidate genes (Carreno-Quintero et al., 2012).
Originally developed for unicellular organisms, these approaches are now applied to plants for the analysis of transcripts, proteins, metabolites (Gibon et al., 2006; Stitt et al., 2010; Liberman et al., 2012) and more recently enzymes activities (Gibon et al., 2004; Saito et al., 2008; Moreno-Risueno et al., 2010).
Most studies involve homogenization of specific tissues or even whole organs (e.g., fruit pericarp, leaf), without considering the subcellular localization of the measured compounds. Plant cells distinguish from other cells in possessing a large central vacuole, which size may vary dramatically between tissues, genotypes (species, cultivars) and developmental stages. Whereas young cells have small vacuoles, mature cells have large vacuoles that can encompass more than 95% of the cell volume.
Here, we will demonstrate that without taking into account the volumes of the cell compartments, the analyses of the dynamics of the compounds and the subsequent compound-compound correlation analyses might be biased, especially when the functional significance of the study is bound to concentrations, as it is the case for enzymes and metabolites. Possible correction strategies are discussed, with special emphasis to their pertinence and applicability to specific questions.
Results
Dilution by growth strongly affects metabolic concentrations
Let us consider for the sake of simplicity a cytosolic metabolite M. In many studies the metabolite concentration is expressed per volume (or, equivalently, per gram of fresh mass) of the whole tissue, as
[Mt] = M/Vt with Vt the volume of the tissue and M the metabolite quantity in this tissue. The metabolite being cytosolic, its concentration, to be physiologically meaningful, should be calculated as
[Mc] = M/Vc with Vc the volume of cytosol in the tissue.
It is clear that [Mt] underestimates [Mc] because Vt is often much bigger than Vc due to the presence of other subcellular compartments (e.g., vacuole and plastids). More importantly, the ratio [Mc]/[Mt] = Vt/Vc, may not be constant among genotypes or during the development of organs where the ratio Vt/Vc strongly increases as the cell vacuole enlarges. This may lead to an erroneous interpretation of changes in metabolite concentration when comparing different experiments. Let's make an example. Assuming that Vc is constant and Vt increases with time, Figure 1A shows that depending on metabolite dynamics, the measured metabolite concentration [Mt] can be strongly different from the real cytosolic concentration, both quantitatively and qualitatively. In the extreme case, a decrease of [Mt] can be measured instead of an increase when the growth of the volume is faster than the metabolite accumulation in the cytosol (Figure 1A, blue line). This means that the seasonal variations of [Mt] during development of organs may not have any biological significance, reflecting the evolution of cell growth rather than a specific regulatory strategy.
Figure 1
Note that the effect of dilution depends on the specific features of both the growth curve and metabolite time course, so that it is difficult to correctly estimate the changes in [Mc] in absence of a good knowledge of the subcellular compartmentalization and its evolution in terms of relative volume occupancy within the cell. Of course the effect of dilution by growth is expected to be limited, at least in the last developmental stages, as long as vacuolar compounds are considered but, as shown before, it can become dramatic when compounds are localized in smaller organelles or in the cytosol.
Metabolic correlation analysis are severely biased by dilution
The above considerations naturally extend to the analysis of correlations among concentrations of compounds, in a system-level perspective.
As an example, let's now consider 50 cytosolic metabolites, with linear or non-linear seasonal variation (Figure 1B). As shown in Figure 1B the pattern of correlation between the different metabolites can change dramatically due to dilution by growth. Indeed, most correlation values are altered and 47% of the metabolites display an opposite correlation when considering the total tissue in place of the cytosol (Figure 1C).
As a consequence, metabolites that could be meaningfully clustered together according to their cytosolic variations are now split into separate groups whereas non-correlated metabolites are incorrectly merged into a common cluster by dilution effect (Figures 1D,E). Overall, the effect of volume expansion is to increase positive correlation among cytosolic metabolites, damping variations at late developmental stages.
Discussion and prospects
Plant development is a dynamic process that involves a complex series of molecular and biochemical events associated to volume expansion. Understanding its regulation is one of the great challenges of modern biology and has been the goal for many omics studies over the last years.
It is striking that in many studies fresh weight is used as the reference to express concentrations of biomolecules, irrespective of their subcellular localization. We show here that considering subcellular compartments is of primary importance for the correct interpretation of experimental results, especially when dealing with a system level perspective. Note that in the example we used, the organ volume was assumed to change with time but our conclusions are valuable for any other factor inducing organ volume variation, such as carbon or water stress, temperature, or genotypic traits. As a consequence, the bias induced by growth can potentially affect not only studies on dynamic profiling (Lombardo et al., 2011; Osorio et al., 2011, 2012) but also comparison among genotypes or the investigation of system response to stress (Cross et al., 2006; Sulpice et al., 2010; Osorio et al., 2012). For instance, Steinhauser et al. (2010) found that most of tomato enzyme activities decrease during fruit development and interpreted this as being at least in part due to vacuole expansion (in their study, the fruit volume varied from almost 0 to 60 cm3).
In the last few years a number of studies have pointed out subcellular compartmentalization as a key feature of plant cell biology and as an unavoidable requirement for accurate experimental measurements (Sun et al., 2004; Kruger et al., 2007; Fernie and Stitt, 2012) and models (Grafahrend-Belau et al., 2009; Hay and Schwender, 2011). In spite of these warnings, however, the impact of compartmentalization and especially of its variability among genotypes, experimental conditions or developmental stages is still largely underestimated. A key challenge for the future is therefore to improve existing methods to evaluate concentrations in cellular compartments. Recently, a first compartmentalized map of the metabolome of Arabidopsis has been proposed (Krueger et al., 2011) and the quantification of metabolite concentrations has been undertaken for specific organelles (Oikawa et al., 2011; Tohge et al., 2011). However, although many technical solutions ranging from sub-cellular fractionation (Gerhardt and Heldt, 1984) to in situ imaging (Okumoto et al., 2012) have been developed, none of them is yet fully applicable in routine (Krueger et al., 2012).
In the meanwhile, there are other ways to express concentrations of biomolecules that might prove more pertinent than fresh weight basis. The choice of the appropriate normalization factor depends on the scientific question of interest, the experimental protocol and the available data.
For tissues undergoing vacuolar expansion (Li et al., 2012), for instance, protein content represents an interesting option, as a proxy for the cytoplasmic compartment, assuming that proteins are by far less abundant in the vacuole. In many plants, a strong correlation has been showed between ploidy, nuclear size, and the volume of the cytoplasm (Sugimoto-Shirasu and Roberts, 2003), so that nuclear size could also be used as a proxy for the cytoplasmic compartment. However, methods for routine measurement of nuclear size have to be developed.
In absence of molecular measurements, dry mass, can be used as a normalization factor in situations where the water content and thus fresh mass can undergo strong variations (e.g., under water stress). If in addition to water, the content in storage compounds (starch, sucrose, amino acids, etc.) is expected to change, the above strategy can be iterated by expressing the concentration with respect to the structural dry mass (SDM) of the tissue, as: where STO is storage compounds content of the tissue. In this case, the first component describes the dilution due to water, the second describes the dilution due to the accumulation of storage carbohydrate and the third describes the concentration of X, independently of the dilution processes.
Another standardization used in water stress studies consists in expressing data on fresh mass at full turgor (Hummel et al., 2010). This method provides the possibility to evaluate the contribution of accumulated metabolites to osmotic adjustment. It nevertheless involves an additional and work intensive experimental step, which restricts its use to small scale experiments.
In the case of studies on leaf tissues, normalization with respect to leaf surface (Krapp et al., 1993) is generally advantageous from a practical point of view (collection of leaf discs of known surface) and perhaps also conceptually. In particular, various studies indicate that under stresses or genetic manipulations affecting photosynthesis, photosynthetic activity appears less sensitive when expressed on a mass-basis than on an area-basis, the latter being the one that correlates best with relative growth rate (Poorter et al., 2009). Chlorophyll content has also been used for leaves (Holaday et al., 1992) but its stability has to be checked.
Another way to circumvent the problem bound to subcellular volumes is to express data in a “semi-quantitative” way, i.e., not an absolute amount per unit of biomass but a relative amount. In the simplest case a signal obtained for a given analyte is divided by the signal obtained by another analyte, either internal, or external to the system under study. In the case of mass spectrometry-based metabolomics (Katajamaa and Oresic, 2007), reference samples are often used and metabolite levels are expressed as “fold-changes.” Their use assumes that variations in the relative sizes of compartments do not affect the output. However, only a fraction of the metabolites are solely located within one cell compartment, and their distributions between compartments may vary. The use of isotope-labeled biological standards does actually not solve this problem. Metabolomics also use single or multiple “internal” (added before the extraction) or “external” standards (added after the extraction). Metabolites are then expressed relative to one or more standards, but subcellular compartmentation is again critical. In contrast to metabolomics, transcriptomics (Irizarry et al., 2003; Bullard et al., 2010) and proteomics (Clough et al., 2012) usually involve normalization procedures where the whole population of analytes of a sample, an experiment or a series of experiments represents the reference, thus reducing the bias induced by changes in compartments/organelles volumes. Although such normalization probably is the most suitable for correlation analyses, it is considered as less robust for metabolites (Katajamaa and Oresic, 2007). It is worth mentioning that “semi-quantitative” data are suitable for studying fold-changes or for looking for correlations between variables, but not necessary for building mechanistic models (Rohwer, 2012). Indeed, whereas a range of integrative approaches can cope with alternatives to fresh weight to express amounts and/or activities of biomolecules, in others, such as mechanistic modeling, it will be crucial to evaluate the concentrations and thus the volume of the compartments where the enzymes or the metabolites are present. In that perspective, a challenging issue for the histologists moving to systems biology would be to find new techniques allowing quick measurement of compartment volumes and to feed existing databases on organelles dynamics (Mano et al., 2008, 2011).
Statements
Acknowledgments
This work has been funded by the Eranet Erasysbio+ FRIM project and by INRA.
References
1
BullardJ. H.PurdomE.HansenK. D.DudoitS. (2010). Evaluation of statistical methods for normalization and differential expression in mRNA-Seq experiments. BMC Bioinform. 11:94. 10.1186/1471-2105-11-94
2
Carreno-QuinteroN.AcharjeeA.MaliepaardC.BachemC. W. B.MummR.BouwmeesterH.et al. (2012). Untargeted metabolic quantitative trait loci analyses reveal a relationship between primary metabolism and potato tuber quality. Plant Physiol. 158, 1306–1318. 10.1104/pp.111.188441
3
CloughT.ThaminyS.RaggS.AebersoldR.VitekO. (2012). Statistical protein quantification and significance analysis in label-free LC-MS experiments with complex designs. BMC Bioinform. 13:S6. 10.1186/1471-2105-13-S16-S6
4
CrossJ. M.Von KorffM.AltmannT.BartzetkoL.SulpiceR.GibonY.et al. (2006). Variation of enzyme activities and metabolite levels in 24 Arabidopsis accessions growing in carbon-limited conditions. Plant Physiol. 142, 1574–1588. 10.1104/pp.106.086629
5
FernieA. R.StittM. (2012). On the discordance of metabolomics with proteomics and transcriptomics: coping with increasing complexity in logic, chemistry, and network interactions scientific correspondence. Plant Physiol. 158, 1139–1145. 10.1104/pp.112.193235
6
GerhardtR.HeldtH. W. (1984). Measurement of subcellular metabolite levels in leaves by fractionation of freeze-stopped material in nonaqueous media. Plant Physiol. 75, 542–547. 10.1104/pp.75.3.542
7
GibonY.BlaesingO. E.HannemannJ.CarilloP.HöhneM.HendriksJ. H. M.et al. (2004). A Robot-based platform to measure multiple enzyme activities in Arabidopsis using a set of cycling assays: comparison of changes of enzyme activities and transcript levels during diurnal cycles and in prolonged darkness. Plant Cell16, 3304–3325. 10.1105/tpc.104.025973
8
GibonY.UsadelB.BlaesingO. E.KamlageB.HoehneM.TretheweyR.et al. (2006). Integration of metabolite with transcript and enzyme activity profiling during diurnal cycles in Arabidopsis rosettes. Genome Biol. 7:R76. 10.1186/gb-2006-7-8-r76
9
Grafahrend-BelauE.SchreiberF.KoschützkiD.JunkerB. (2009). Flux balance analysis of barley seeds: a computational approach to study systemic properties of central metabolism. Plant Physiol. 149, 585–598. 10.1104/pp.108.129635
10
HayJ.SchwenderJ. (2011). Metabolic network reconstruction and flux variability analysis of storage synthesis in developing oilseed rape (Brassica napus L.) embryos. Plant J. 67, 526–541. 10.1111/j.1365-313X.2011.04613.x
11
HoladayA. S.MartindaleW.AlredR.BrooksA. L.LeegoodR. C. (1992). Changes in activities of enzymes of carbon metabolism in leaves during exposure of plants to low temperature. Plant Physiol. 98, 1105–1114. 10.1104/pp.98.3.1105
12
HummelI.PantinF.SulpiceR.PiquesM.RollandG.DauzatM.et al. (2010). Arabidopsis plants acclimate to water deficit at low cost through changes of carbon usage: an integrated perspective using growth, metabolite, enzyme, and gene expression analysis. Plant Physiol. 154, 357–372. 10.1104/pp.110.157008
13
IrizarryR. A.HobbsB.CollinF.Beazer-BarclayY. D.AntonellisK. J.ScherfU.et al. (2003). Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics4, 249–264. 10.1093/biostatistics/4.2.249
14
KatajamaaM.OresicM. (2007). Data processing for mass spectrometry-based metabolomics. J. Chromatogr. A1158, 318–328. 10.1016/j.chroma.2007.04.021
15
KrappA.QuickP. W.StittM. (1993). Ribulose-1,5-bisphosphate carboxylase-oxygenase, other Calvin-cycle enzymes, and chlorophyll decrease when glucose is supplied to mature spinach leaves via the transpiration stream. Planta186, 58–69.
16
KrugerN. J.Le LayP.RatcliffeR. G. (2007). Vacuolar compartmentation complicates the steady-state analysis of glucose metabolism and forces reappraisal of sucrose cycling in plants. Phytochemistry68, 2189–2196. 10.1016/j.phytochem.2007.04.004
17
KruegerS.GiavaliscoP.KrallL.SteinhauserM.-C.BüssisD.UsadelB.et al. (2011). A topological map of the compartmentalized Arabidopsis thaliana leaf metabolome. PLoS ONE6:e17806. 10.1371/journal.pone.0017806
18
KruegerS.SteinhauserD.WillmitzerL.GiavaliscoP. (2012). High-resolution plant metabolomics: from mass spectral features to metabolites and from whole-cell analysis to subcellular metabolite distributions. Plant J. 70, 39–50. 10.1111/j.1365-313X.2012.04902.x
19
LiZ.PalmerW. M.MartinA. P.WangR.RainsfordF.JinY.et al. (2012). High invertase activity in tomato reproductive organs correlates with enhanced sucrose import into, and heat tolerance of, young fruit. J. Exp. Bot. 63, 1155–1166. 10.1093/jxb/err329
20
LibermanL. M.SozzaniR.BenfeyP. N. (2012). Integrative systems biology: an attempt to describe a simple weed. Curr. Opin. Plant Biol. 15, 162–167. 10.1016/j.pbi.2012.01.004
21
LombardoV.OsorioS.BorsaniJ.LauxmannM.BustamanteC.BuddeC. O.et al. (2011). Metabolic profiling during peach fruit development and ripening reveals the metabolic networks that underpin each developmental stage. Plant Physiol. 157, 1696–1710. 10.1104/pp.111.186064
22
ManoS.MiwaT.NishikawaS.MimuraT.NishimuraM. (2008). The plant organelles database (PODB): a collection of visualized plant organelles and protocols for plant organelle research. Nucleic Acids Res. 36, D929–D937. 10.1093/nar/gkm789
23
ManoS.MiwaT.NishikawaS.MimuraT.NishimuraM. (2011). The Plant Organelles Database 2 (PODB2): an updated resource containing movie data of plant organelle dynamics. Plant Cell Physiol. 52, 244–253. 10.1093/pcp/pcq184
24
Moreno-RisuenoM. A.BuschW.BenfeyP. N. (2010). Omics meet networks - using systems approaches to infer regulatory networks in plants. Curr. Opin. Plant Biol. 13, 126–131. 10.1016/j.pbi.2009.11.005
25
OikawaA.MatsudaF.KikuyamaM.MimuraT.SaitoK. (2011). Metabolomics of a single vacuole reveals metabolic dynamism in an alga Chara australis. Plant Physiol. 157, 544–551. 10.1104/pp.111.183772
26
OkumotoS.JonesA.FrommerW. B. (2012). Quantitative imaging with fluorescent biosensors. Annu. Rev. Plant Biol. 63, 663–706. 10.1146/annurev-arplant-042110-103745
27
OsorioS.AlbaR.DamascenoC. M. B.Lopez-CasadoG.LohseM.ZanorM. I.et al. (2011). Systems biology of tomato fruit development: combined transcript, protein, and metabolite analysis of tomato transcription factor (nor, rin) and ethylene receptor (Nr) mutants reveals novel regulatory interactions. Plant Physiol. 157, 405–425. 10.1104/pp.111.175463
28
OsorioS.AlbaR.NikoloskiZ.KochevenkoA.FernieA. R.GiovannoniJ. J. (2012). Integrative comparative analyses of transcript and metabolite profiles from pepper and tomato ripening and development stages uncovers species-specific patterns of network regulatory behaviour. Plant Physiol. 159, 1713–1729. 10.1104/pp.112.199711
29
PoorterH.NiinemetsÜ.PoorterL.WrightI. J.VillarR. (2009). Causes and consequences of variation in leaf mass per area (LMA): a meta-analysis. New Phytol. 182, 565–588. 10.1111/j.1469-8137.2009.02830.x
30
RiedelsheimerC.Czedik-EysenbergA.GriederC.LisecJ.TechnowF.SulpiceR.et al. (2012). Genomic and metabolic prediction of complex heterotic traits in hybrid maize. Nat. Genet. 44, 217–220. 10.1038/ng.1033
31
RohwerJ. M. (2012). Kinetic modelling of plant metabolic pathways. J. Exp. Bot. 63, 2275–2292. 10.1093/jxb/ers080
32
SaitoK.HiraiM. Y.Yonekura-SakakibaraK. (2008). Decoding genes with coexpression networks and metabolomics - majority report by precogs. Trends Plant Sci. 13, 36–43. 10.1016/j.tplants.2007.10.006
33
SteinfathM.StrehmelN.PetersR.SchauerN.GrothD.HummelJ.et al. (2010). Discovering plant metabolic biomarkers for phenotype prediction using an untargeted approach. Plant Biotechnol. J. 8, 900–911. 10.1111/j.1467-7652.2010.00516.x
34
SteinhauserM.-C.SteinhauserD.KoehlK.CarrariF.GibonY.FernieA. R.et al. (2010). Enzyme activity profiles during fruit development in tomato cultivars and Solanum pennellii. Plant Physiol. 153, 80–98. 10.1104/pp.110.154336
35
StittM.SulpiceR.KeurentjesJ. (2010). Metabolic networks: how to identify key components in the regulation of metabolism and growth. Plant Physiol. 152, 428–444. 10.1104/pp.109.150821
36
Sugimoto-ShirasuK.RobertsK. (2003). Big it up: endoreduplication and cell-size control in plants. Curr. Opin. Plant Biol. 6, 544–553. 10.1016/j.pbi.2003.09.009
37
SulpiceR.TrenkampS.SteinfathM.UsadelB.GibonY.Witucka-WallH.et al. (2010). Network analysis of enzyme activities and metabolite levels and their relationship to biomass in a large panel of Arabidopsis accessions. Plant Cell22, 2872–2893. 10.1105/tpc.110.076653
38
SunQ.EmanuelssonO.Van WijkK. J. (2004). Analysis of curated and predicted plastid subproteomes of Arabidopsis. Subcellular compartmentalization leads to distinctive proteome properties. Plant Physiol. 135, 723–734. 10.1104/pp.104.040717
39
TohgeT.RamosM. S.Nunes-NesiA.MutwilM.GiavaliscoP.SteinhauserD.et al. (2011). Toward the storage metabolome: profiling the barley vacuole. Plant Physiol. 157, 1469–1482. 10.1104/pp.111.185710
Summary
Keywords
concentration, correlation, dilution, growth, metabolic, subcellular
Citation
Génard M, Baldazzi V and Gibon Y (2014) Metabolic studies in plant organs: don't forget dilution by growth. Front. Plant Sci. 5:85. doi: 10.3389/fpls.2014.00085
Received
17 December 2013
Accepted
23 February 2014
Published
11 March 2014
Volume
5 - 2014
Edited by
Alisdair Fernie, Max Planck Institute for Plant Physiology, Germany
Reviewed by
Jocelyn K. C. Rose, Cornell University, USA; Sonia Osorio, Málaga University, Spain
Copyright
© 2014 Génard, Baldazzi and Gibon.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: michel.genard@avignon.inra.fr
This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science.
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.