Coupling water and carbon limitations through the Lockhart model?
ABA signalling pathway as a hub to coordinate water and carbon relations
Leaf venation: just a two-way pipe network?
Leaf ontogeny orchestrates the actors involved in the control of leaf growth
The growing leaf in a changing world
Leaf growth is the central process facilitating energy capture and plant performance. This is also one of the most sensitive processes to a wide range of abiotic stresses. Because hydraulics and metabolics are two major determinants of expansive growth (volumetric increase) and structural growth (dry matter increase), we review the interaction nodes between water and carbon. We detail the crosstalks between water and carbon transports, including the dual role of stomata and aquaporins in regulating water and carbon fluxes, the coupling between phloem and xylem, the interactions between leaf water relations and photosynthetic capacity, the links between Lockhart's hydromechanical model and carbon metabolism, and the central regulatory role of abscisic acid. Then, we argue that during leaf ontogeny, these interactions change dramatically because of uncoupled modifications between several anatomical and physiological features of the leaf. We conclude that the control of leaf growth switches from a metabolic to a hydromechanical limitation during the course of leaf ontogeny. Finally, we illustrate how taking leaf ontogeny into account provides insights into the mechanisms underlying leaf growth responses to abiotic stresses that affect water and carbon relations, such as elevated CO2, low light, high temperature and drought.
I. Leaf growth: volume, structures, water and carbon
Leaf growth consists of two components: an increase in volume – the expansive growth – and an increase in dry matter – the structural growth. Leaf growth, at timescales of minutes to days is strongly related to temperature (Parent & Tardieu, 2012; Eqn 1 in Fig. 1) thus defining a potential growth that must be confronted by major constraints. Indeed, the irreversible increase in the size of leaf cells during their expansion requires a massive influx of water into the vacuoles. It also demands a substantial supply of carbon skeletons to build new structures (e.g. cellulose and hemicellulose in the new cell walls) and additional photosynthesis products to fuel the various energy-consuming mechanisms (e.g. cell cycle, protein synthesis, functioning of pumps and channels to upload solute in vacuoles and generate turgor). From this, the availability of water and carbon have often been considered the main limitations of leaf growth, classically referred to as the hydraulic and the metabolic control, respectively (Kriedemann, 1986; Dale, 1988; Walter et al., 2009). These views are validated by experimental evidence and models.
1. Water relations and wall rheology provide the biophysical basis of expansive growth
Cell expansion relies on a tight coupling between water fluxes and cell wall rheological modifications (Cosgrove, 1986), as described in the Lockhart model (Lockhart, 1965; Eqn 5 in Fig. 1). Three steps must be satisfied in a coordinated manner for growth to be sustained (Moulia & Fournier, 2009; Eqn 5 in Figs 1, 2): (1) cell turgor pressure transmits tensional stresses in the cell wall, which stretches irreversibly when a yield threshold pressure is exceeded; (2) this deformation is accompanied by a passive water flow governed by the water potential gradient between the cell and the water source; and (3) this gradient is actively maintained by adjustments of the cell osmotic potential, which in turn generates cell turgor pressure. Originally developed and validated on single giant algal cells, the Lockhart model is supported by experimental data in several plant species at the leaf scale (e.g. Bunce, 1977; Bouchabké et al., 2006; Ehlert et al., 2009, 2011; Zhang et al., 2011). It accounts for the reduction of leaf expansion rate under soil water deficit when water potential gradients towards growing cells are less steep. It also accounts for a lowered leaf growth during high transpiration periods, when expansive growth competes for water with transpiration, although only 1–2% of transpiration water is used for growth (Fricke, 2002). Accordingly, expansive growth in both dicots and monocots is classically reported as being lower during the day than during the night, when stomata are closed (Ben-Haj-Salah & Tardieu, 1996; Pantin et al., 2011).
2. Carbon metabolism supplies the material and the energy to achieve structural growth
Stomatal opening in the light gives access to atmospheric CO2 for photosynthesis whose products are essential for growth, providing both the building bricks (e.g. cell wall cellulose and hemicellulose) and the energy required for structural growth (Dale, 1985; Fig. 2). Consistent with this, accumulation of plant biomass is strongly related to intercepted radiation and cumulated photosynthesis, as originally formalized by Monteith (1977; Eqn 2a in Fig. 1). Moreover, in many instances, leaf growth rate or shoot biomass correlate well with either net photosynthesis, activity of carbon metabolism enzymes, or level of carbon metabolites (maize, Rocher et al., 1989; tobacco Fichtner et al., 1993; Arabidopsis, Cross et al., 2006, Sulpice et al., 2009, 2010). During nights, carbohydrate availability in leaves is buffered by transitory storage compounds, among which starch predominates in most species (Stitt & Zeeman, 2012). Starch turnover is also controlled by the circadian clock and it could represent a crucial node in the regulation of growth (Sulpice et al., 2009; Graf et al., 2010). Accordingly, mutations in the starch metabolism have drastic effects on leaf growth at night, while normal leaf expansion is maintained during the day (Wiese et al., 2007; Pantin et al., 2011). In addition to their roles as substrate for metabolism, carbohydrates and sugars also play intricate signalling roles over a broad range of developmental processes and stress responses (Rolland et al., 2006). This multiplicity of roles, as well as the difficulty of measuring local carbohydrate availability (as opposed to water status for example), explain why, beyond correlations (Freixes et al., 2002; Cross et al., 2006; Muller et al., 2011), there is no available mathematical formalism relating growth to carbon availability at a fine time-scale. Therefore, it is tempting to question if and how carbon metabolism may fit into already established formalisms, notably the Lockhart biophysical model.
II. Coupling water and carbon limitations through the Lockhart model?
The Lockhart model offers several bridges between cell hydromechanics and carbon metabolism. Although rarely exploited, these couplings include osmotic adjustment, wall properties and turgor sensing.
1. Osmotic adjustment relies on organic compounds
In the Lockhart model, osmotic potential plays a central role by generating turgor pressure, which ultimately stretches cell walls (Eqn 5 in Fig. 1). Osmotic potential arises from a passive or active accumulation of solutes within the cell, which creates a difference in solute concentration between the inside and the outside of the cell. Obeying thermodynamics, a passive water flow then occurs into the growing cell where osmotic potential is lower. This coupling is generally at the core of models simulating fruit growth, where sugars are the main contributors to osmotic potential (Fishman & Génard, 1998; Martre et al., 2011; Muller et al., 2011). In leaves, although inorganic osmotica such as nitrate and potassium are classically reported as being the main contributors to the baseline osmotic potential, osmotic adjustment in response to water status fluctuations may strongly rely on organic compounds (Turner et al., 1978). In Arabidopsis leaves, variations of organic acids, proline, other amino acids and sugars represent > 50% of the osmotic adjustment under severe water deficit (Hummel et al., 2010).
2. Carbon control over wall properties
Several lines of evidence suggest that the cell wall, which is at the core of the Lockhart analysis, should not be restricted to two fixed parameters but envisioned as a carbon-modulated compartment within a hydromechanical framework. Carbon availability could modulate wall properties at two levels: directly as a substrate responsible for the adjustment of wall thickness and indirectly by modulating the enzymes controlling wall mechanics. At first sight, the Lockhart model does not tell much about the deposition of wall material during cell expansion because it assumes that the rate of deposition matches the rate of expansion-induced thinning. Although wall deposition and expansion are uncoupled under several circumstances, such as during hypocotyl elongation (Cosgrove, 2005), this coupling assumption accounts for the growth of cell tips such as pollen tubes (Dumais et al., 2006). This assumption implicitly requires that carbon availability crosstalks with wall material deposition, for example to prevent excessive thinning of cell walls during periods of low carbon (Hanba et al., 2002). A coupling between expansion and cell wall deposition could be the result of a tight regulation of the molecular actors involved in cell wall relaxation and synthesis. Indeed, cell wall extensibility is controlled by several proteins and enzymes such as expansins, endoglucanases, peroxidases and xyloglucan endotransglycolases/hydrolases (Cosgrove, 2005). Their function has been incorporated in a modified version of the Lockhart model, where the activity of loosening enzymes appears as a parameter that represents the rate at which load-bearing hemicellulose molecules linking cellulose microfibrils are cut (Passioura & Fry, 1992). In Arabidopsis and maize leaves, the expression or activity of these proteins is tightly associated with growth changes (Cho & Cosgrove, 2000; Muller et al., 2007) is affected by water deficit (Muller et al., 2007; Harb et al., 2010), and also responds to elevated CO2 in different species (Ranasinghe & Taylor, 1996; Ferris et al., 2001; Ainsworth et al., 2006) suggesting that these environmental influences on growth could be represented by changes in extensibility parameters in the Lockhart model. Finally, cellulose synthases, subordinated to the availability of their carbon substrate, are possibly feedback-regulated by the activity of enzymes responsible for cell wall loosening (Somerville, 2006). This may provide a molecular convergence in the control of wall metabolism and wall loosening.
3. Making use of turgor to control wall metabolism and mechanics
Another candidate for the coupling between carbon availability, wall metabolism, wall mechanics and the hydraulic control of growth could be turgor itself. In Chara corallina cells, turgor acts as the mechanical force to facilitate incorporation of large polymers into the wall (Proseus & Boyer, 2005). In this alga, a turgor-sensitive step in the process of wall synthesis has been identified as the ‘pectate cycle’ (Proseus & Boyer, 2007). In the tightening wall, calcium crosslinks the pectate polymers in the matrix. A threshold of turgor pressure is necessary to distort the bonds between pectate and calcium. When supplied to the wall matrix, newly synthesized pectate chelates the calcium released from the distorted bond, allowing wall loosening and irreversible expansion. The pectate then integrates the wall by developing crosslinks with the existing molecules in the matrix, while the free calcium enters the wall and retightens the structure. The complete model thus reconciles wall expansion, wall deposition and wall mechanical status through turgor sensitivity (Proseus & Boyer, 2007). Furthermore, the combined action of the pectate cycle and turgor on deposition of new materials in the cell wall also accounts for the so-called ‘stored growth’, that is, the above-normal flush of growth observed when turgor recovers after a period of turgor loss and stopped growth (Proseus & Boyer, 2008). Such interactions between turgor and wall metabolism could inspire plant modellers because stored growth is also observed in leaves, especially during recovery from water deficit (Hsiao et al., 1970).
Finally, it has also been proposed that cell expansion and carbohydrate metabolism in Arabidopsis is under the control of a cell wall sensing machinery able to detect changes in turgor pressure, and to feedback on turgor through the regulation of solute metabolism. First, inhibition of cellulose biosynthesis affects the cell wall integrity and modifies the expression of genes involved in mechanoperception, possibly through changes in turgor pressure, thereby triggering the same cascade as described upon hypo-osmotic shock (Hamann et al., 2009). Second, inhibition of cellulose biosynthesis also causes transcriptional, enzymatic and metabolic changes in the central metabolism, which are relieved upon osmotic stress, suggesting that a turgor-sensitive mechanism tunes carbohydrate metabolism (Wormit et al., 2012). Third, the expression and activity of vacuolar invertase, an enzyme that contributes to osmotic adjustment, are dependent on the functioning of a wall-associated kinase involved in cell wall sensing and cell expansion (Kohorn et al., 2006).
Leaf rheology is thus at the crossroads of hydraulics and metabolics, and it is tempting to subordinate the Lockhart parameters to carbon metabolism. Interestingly, a conceptual model was drawn in line with this view > 25 yr ago (Hsiao et al., 1985). Nevertheless, formalisms describing this complex subordination are still lacking, partly because rheological properties respond to numerous other stimuli, including pH (Thompson, 2001), reactive oxygen species (Liszkay et al., 2003), or ABA (Cramer et al., 1998). Because ABA affects wall rheology, tunes the fluxes of water and carbon and mediates growth responses to various stresses, its central role in orchestrating hydraulics and metabolic interactions is analysed in the next section.
III. ABA signalling pathway as a hub to coordinate water and carbon relations
Abscisic acid accumulates in plants under water deficit and affects leaf growth in multiple, sometimes contradictory, direct and indirect manners (Tardieu et al., 2010). Part of this complexity arises from the involvement of ABA in both carbon and water fluxes. It regulates stomata and aquaporins, which in turn control both water loss from the leaf and incoming of carbon dioxide. In addition, ABA signalling is known to interfere with sugar sensing, with hexokinase (HXK) being the central pivot of this interaction (Rolland et al., 2006). Other kinases such as the family of sucrose nonfermenting-1-related protein kinases (SnRKs) also connect ABA signalling with primary metabolism through sugar phosphate intermediates (notably trehalose-6-phosphate) and transcription factors (Delatte et al., 2011; Ma et al., 2011), thus coordinating metabolic and stress signalling (Hey et al., 2010). Abscisic acid also regulates carbon metabolism at the enzymatic level, both transcriptionally and post-transcriptionally (Trouverie et al., 2004; Zhu et al., 2011). It inhibits cell (Wang et al., 1998) and plastid (Galpaz et al., 2008) division. Thus, ABA affects both expansive and structural growth.
1. Stomata integrate hydraulic and metabolic cues via ABA signalling
As a major determinant of stomatal closure, especially under water deficit, ABA supervises the trade-off between transpiration and assimilation (Eqn 3 in Figs 1, 2). Hence, at first sight, ABA promotes expansive growth by saving leaf water and reducing xylem tension, but likely alters structural growth by limiting CO2 entry. It also interferes with stomatal responses to atmospheric CO2 and byproducts of photosynthesis and could therefore play a more subtle role in optimizing the trade-off between water loss and carbon fixation. Experimental data show an interaction between ABA and CO2 in the control of stomatal conductance, for example, in cocklebur (Xanthium strumarium; Raschke, 1975), in Arabidopsis (Leymarie et al., 1998) or in various tree species (Aasamaa & Sõber, 2011). Furthermore, guard cell sensitivity to CO2 depends on air relative humidity and vice versa (Bunce, 1996; Talbott et al., 2003), while stomatal response to humidity is partly mediated by ABA, as shown in Arabidopsis (Xie et al., 2006; Okamoto et al., 2009). At the molecular level, ABA and CO2 interact to regulate stomatal conductance in a network that is just starting to be revealed. Both ABA and elevated CO2 would act at an undefined convergence point where Ca2+ sensitivity is primed to trigger stomatal closure (Kim et al., 2010). Although rarely cited, sugars could also interact with ABA to control stomata. Indeed, carbon metabolism is involved in the control of night-time stomatal conductance, for example, in Arabidopsis (Lascève et al., 1997) or in Vicia faba (Easlon & Richards, 2009). Furthermore, starch, sucrose, hexoses and malate are convenient sources of energy and osmotica required for stomatal movements (Vavasseur & Raghavendra, 2005; Penfield et al., 2012). Thus, connections can be hypothesized between products of photosynthesis and the HXK and SnRKs network. Overall, the complex scheme of ABA, CO2 and sugar signalling pathways promises exciting perspectives to elucidate the way by which stomata arbitrate between water and carbon fluxes.
2. Aquaporins: internal modulators of water and carbon fluxes
In addition to stomata, aquaporins are another major target for ABA to regulate both water and carbon fluxes. Abscisic acid partly mediates aquaporin regulation in response to environmental stresses (Maurel et al., 2008) by modulating their gene expression and their protein abundance or activity. This affects plant water relations in various organs, via, for example, an increase in root hydraulic conductivity in maize (Parent et al., 2009) or tomato (Thompson et al., 2007), a decrease in leaf hydraulic conductance (Shatil-Cohen et al., 2011) or an increase in leaf protoplast water permeability in Arabidopsis (Morillon & Chrispeels, 2001), although the last result was attributed to an indirect effect of ABA on transpiration rate. Not only are aquaporins water channels, but they also transport CO2 and some organic solutes (Maurel et al., 2008; Fig. 2). They have a significant role in photosynthesis through mesophyll conductance to CO2 in tobacco or Arabidopsis (Flexas et al., 2006; Heckwolf et al., 2011), as well as in carbohydrate routing in Arabidopsis (Ma et al., 2004). Furthermore, they may be involved in the interaction between water and sugar transport required for the fruit development as shown in tomato (Chen et al., 2001). Thus, any change in aquaporin regulation mediated by ABA (or any other stimuli) could alter both carbon and water fluxes.
IV. Leaf venation: just a two-way pipe network?
We have emphasized the role of ABA in orchestrating the influence of water and carbon on leaf growth. Abscisic acid is highly mobile in the plant: local synthesis intermixes with a flow from roots to leaves via the xylem and a backflow from source leaves to sink organs via the phloem (Hartung et al., 2002). However, the central role of vascular networks in the interaction between hydraulic and metabolic control of growth goes well beyond ABA redistribution only.
1. Xylem and phloem are hydraulically coupled
Xylem conduits bring water and nutrients from roots to leaves while phloem conduits bring photosynthates from source leaves to sink organs. Interestingly, xylem and phloem conduits are hydraulically connected (Hölttä et al., 2006; Fig. 2). Active loading of organic solutes in phloem conduits of source leaves decreases the phloem osmotic potential and attracts water from the surrounding tissues, including xylem as a major source of water. This ultimately gives rise to turgor pressure in sieve elements of source leaves, which represents the hydrostatic force that triggers phloem flow (Fig. 1, Eqn 4). Consequently, the phloem flow is sensitive to changes in water potential in the xylem of the source leaves and any increase in transpiration rate decreases the phloem assimilate export to sink organs through a purely hydraulic mechanism (Lacointe & Minchin, 2008). This central role of xylem and phloem in long-distance redistribution is likely to have driven leaf venation to organize in a network that optimizes the use of water and carbon resources.
2. A cost–benefit approach of leaf venation
Remarkably, across species, leaf area is negatively correlated with the density of major veins and positively correlated with major vein diameter (Sack et al., 2012), two central leaf traits that govern the leaf hydraulic efficiency (Sack & Frole, 2006). Venation properties and leaf size become of primary importance under adverse hydraulic conditions, where low xylem lumen diameter, high major vein density and small leaf area confer greater tolerance (Blackman et al., 2010; Scoffoni et al., 2011). Interestingly, drought-tolerant species are also shade-tolerant in dry habitats, because the safety strategy they develop (slow growth, low photosynthesis, low risk of xylem cavitation) makes them better competitors in low-resource environments (Markesteijn et al., 2011). Indeed, the dual role of stomata implies that high photosynthesis partly depends on high transpiration, which requires high water supply capacity to minimize the risk of high xylem tension. Accordingly, leaf hydraulic conductance is coordinated with its photosynthetic capacity across a wide range of species (Sack & Holbrook, 2006). Notably, the course of angiosperm diversification was marked by a surge in leaf vein density and photosynthetic capacity (Boyce et al., 2009; Beerling & Franks, 2010; Brodribb & Feild, 2010; Feild et al., 2011). More precisely, the ability of the vein network to fill leaf space governs the photosynthesis rate by modulating either the photosynthate export capacity (Raven, 1994) or the leaf hydraulic efficiency (Brodribb et al., 2007). However, space-filling represents a massive investment because lignified tissues are net carbon sinks that do not contribute directly to photosynthesis (Fig. 2). The ability of plants to manage these construction costs in relation to their benefits has been a determining factor in their evolution (Beerling & Franks, 2010; Pittermann et al., 2012). In leaves, hydraulic efficiency relative to construction cost is increased by vein hierarchy, tapering and density (McKown et al., 2010). A recent study in Nothofagus cunninghamii also shows that the development of veins and stomata is coordinated to optimize the carbon investment in leaf venation (Brodribb & Jordan, 2011); similarly, along grass blades, an acropetal increase in stomatal conductance is balanced by an acropetal decrease in the distance from vascular bundles to stomatal pores, allowing a developmental increase in photosynthesis without a dramatic drop in water potential between vessels and evaporation sites (Ocheltree et al., 2012). This stresses the need for an ontogenetic vision of the way hydraulics and metabolics interact in the control of leaf growth.
V. Leaf ontogeny orchestrates the actors involved in the control of leaf growth
Despite some species-specific and environment-dependent variations on the relative timing of each developmental event, a general, species-independent sequence of morphological, anatomical and physiological changes during leaf ontogeny emerges from the literature, as shown for dicots in Fig. 2. A similar pattern could certainly be depicted in monocots along a spatial gradient. Leaf area expands in a sigmoidal way, with an early exponential growth, a marked increase and a deceleration followed by a plateau (Fig. 3). The absolute expansion rate peaks at the inflection point of the curve, when the leaf reaches 50% final area; however, the relative expansion rate, that is, the area accumulated per unit area and per unit of time, is at its highest in the earliest stages and decreases afterwards (Fig. 3). This pattern holds especially in dicot leaves where the growth is not restricted to a basal zone of stable size for several days as in monocots, although a basipetal gradient of increasing growth is generally observed in dicots (Granier & Tardieu, 1998; Walter et al., 2009; Kuchen et al., 2012). Similarly, the relative cell division rate is at its highest in the early stages of leaf development, and decreases sharply before the relative leaf expansion rate starts declining, for example, before 10% final leaf area (Fig. 3), with a basipetal progression of the cell cycle arrest front (Granier & Tardieu, 2009; Kazama et al., 2010; Andriankaja et al., 2012). By contrast, the relative cell expansion rate peaks shortly after the relative cell division rate starts declining (Fig. 3). Thus, cell division precedes cell expansion but these processes largely overlap (Kriedemann, 1986; Granier & Tardieu, 2009). Leaf thickness increases in pace with leaf area, with a rapid thickening occurring at c. 20% of final leaf area and corresponding to the volumetric expansion of the palisade cells and the development of intercellular spaces in the spongy parenchyma (Tichá, 1985; Wuyts et al., 2010, 2012; Fig. 3). The increase in dry mass lags slightly after leaf area, so that leaf mass per area increases during leaf ontogeny after a short period of stability or decrease (Tichá et al., 1985; Gratani & Bonito, 2009; Fig. 3). This general trend provides the framework to obtain deeper insights into the developmental pattern of the metabolic and hydromechanical control of leaf growth.
Water and carbon relations of the expanding leaf were reviewed 25 yr ago (Dale, 1985; Šesták, 1985; Barlow, 1986; Kriedemann, 1986). Considerable advances have been made since then, but a synthesis of these works is currently lacking. Using a genetic and environmental approach on Arabidopsis, we have recently shown that the control of leaf expansion is predominantly metabolic in the early stages, while hydraulic influence establishes as the leaf develops (Pantin et al., 2011). Immediately after leaf emergence, drops in leaf expansion have been observed during night-time, likely resulting from limited carbon availability in the starch pool stored on the previous day. In line with this, these early drops were exacerbated in a set of mutants affected in starch metabolism. Later, drops in leaf expansion occurred during the day, when water movement into the growing cells is limited by transpiration. Consistent with this, these diurnal drops were amplified and occurred earlier during ontogeny in a set of mutants affected in the stomatal control of transpiration, as well as in plants exposed to low soil water potential or high atmospheric vapour pressure deficit. The day/night pattern of leaf turgor and starch content in leaves of contrasting ages further highlighted that the Arabidopsis leaf experiences a developmental switch from a metabolic to a hydraulic control of growth (Pantin et al., 2011). We illustrate thereafter how developmental changes in water and carbon relations could define an ontogeny-dependent control of leaf growth in different plant species.
1. A demonstration by example: insights from CAM species
Plants fitted with the crassulacean acid metabolism (CAM) magnify the interaction between ontogeny, hydraulics and metabolics on the control of leaf growth. Crassulacean acid metabolism emerged several times during evolution in response to hydraulic-selective pressures (Borland et al., 2011). This mode of photosynthesis is mainly characterized by a time lag between light harvested during the day and stomatal uptake of CO2 at night, when evaporative demand lowers. This mechanism enables a substantial water economy by avoiding excessive water loss by transpiration during the day. In some species, either environmental or developmental conditions are able to trigger a metabolism switch from C3 to CAM to optimize carbon and water balance with regard to plant growth (Walter et al., 2008; Borland et al., 2011). In facultative CAM species, leaves operate routinely in C3 mode and induce CAM under adverse environmental conditions, especially drought stress. In other species, CAM is induced by leaf age. Remarkably, this metabolic switch translates into a switch in nycthemeral leaf growth patterns. For example, when the switch from C3 to CAM is induced either by drought stress in Clusia minor or by ontogeny in Clusia alata, leaf growth switches from a higher leaf growth rate at night to a higher leaf growth rate during the day (Walter et al., 2008), following changes in water relations induced by movements of stomata. As the timing of stomata opening in CAM species is first an adaptation to arid environments, as in C. minor under water stress, it is tempting to speculate that the developmental induction of CAM in C. alata arises from the ontogenetic emergence of hydraulic constraints, while the C3 mode would be used when juvenile to favour efficient carbon gain and support structural growth. Interestingly, in leaves of Mesembryanthemum crystallinum, the day/night abundance pattern of some aquaporin isoforms coordinates with the developmentally triggered CAM, so that the cell water permeability balances the nycthemeral changes in osmotic potential induced by CAM (Vera-Estrella et al., 2012). These results obtained in facultative CAM are consistent with the conclusion of Pantin et al. (2011) in Arabidopsis that the control of leaf growth switches from metabolics to hydraulics during the course of its ontogeny.
2. Leaf carbon balance and the sink-to-source transition
That the importance of carbon in the control of leaf growth would decrease during leaf ontogeny makes sense in the light of the sink-to-source transition, that is, the moment when a growing leaf becomes a net carbon exporter. This positive carbon balance is the result of both an increase in carbon supply through photosynthesis and a decrease in carbon demand by growth and respiration (Turgeon, 1989). In monocots, the growing zone is virtually nonphotosynthetic and the source behaviour is acquired with maturation, resulting in a basipetal progression of the carbon balance from net export (tip) to net import (base) on the same leaf. This was recently depicted at the molecular level (Li et al., 2010; Majeran et al., 2010; Pick et al., 2011), showing that the basal growing zone is characterized by transcripts and proteins associated with the establishment of the photosynthetic machinery and structural growth such as photoreceptors, chlorophyll precursors and secondary cell wall biosynthetic enzymes, while in the distal region the cellular machinery is strongly devoted to starch metabolism and photosynthesis reactions. In dicots, the sink-to-source transition occurs when the leaf reaches 30–60% of its final size (Turgeon, 1989; Fig. 3). This transition is accompanied by changes in central metabolism, enzymatic machinery, phloem structure and other anatomical traits to favour CO2 net assimilation and carbohydrate export. Notably, according to the thorough monograph of Šesták (1985) encompassing a wide range of species, net photosynthesis per unit area or mass increases strongly during early expansion and peaks at 25–100% final leaf area, with a stronger and earlier decrease in annuals than in grasses or evergreens. Conversely, dark respiration decreases dramatically during the early stages of leaf development. Photorespiration follows roughly the trend of net photosynthesis while the CO2 compensation point decreases dramatically during early ontogeny. This general pattern is concomitant with a gradual increase in mesophyll conductance to CO2, an initial increase in stomatal conductance, as well as a later increase in chloroplast number and volume, and in pigments such as chlorophylls and carotenoids (Šesták, 1985; Fig. 3). Finally, during leaf expansion, the carboxylase activity of RuBisCO or phosphoenolpyruvate carboxylase (PEPc) as well as the amount of these enzymes parallel the ontogenetic changes in activities of photochemical reactions and in net photosynthesis (Šesták, 1985). Accordingly, transcripts related to photosynthesis increase at the end of the proliferating phase and remain high in mature leaves in Arabidopsis (Skirycz et al., 2010; Andriankaja et al., 2012). This ontogenetic sequence of events is well-coordinated so that the diffusional and biochemical limitations to photosynthesis share similar proportion throughout leaf expansion (Grassi & Magnani, 2005).
It is generally observed that carbohydrate supply affects cell division but not cell expansion, because the carbon requirement for leaf expansion is remarkably low compared with photosynthetic capacity of the leaf itself (Kriedemann, 1986). Moreover, under low carbon availability leaves are able to maintain expansive growth at the expense of dry matter deposition per unit area (Tardieu et al., 1999). It makes sense that exceptions to this rule are likely to be very young leaves, where metabolic needs are maximized compared with photosynthetic capacity (Muller et al., 2001; Pantin et al., 2011). Any environmental, genetic or developmental perturbation that lowers carbon availability in the leaf also makes leaf expansion tightly dependent on carbon supply, for example, very low light or starch mutants in Arabidopsis (Wiese et al., 2007; Pantin et al., 2011), or a competition for carbon with other sinks in soybean (Wenkert et al., 1978). Hence, the young leaf growth rate strongly depends on local carbohydrate availability, as in many other sink organs (Muller et al., 2011). It is interesting to consider that sinks partly control photosynthesis of source leaves: a decrease in sink demand for assimilates leads to carbohydrate accumulation in phloem-loading sites of source leaves, which triggers a downregulation of photosynthesis (Fig. 4). The sink regulation of photosynthesis in source leaves occurs at several levels. For example, synthesis of endproducts, such as sucrose and starch, exerts a short-term metabolic feedback on photosynthesis, while carbohydrate contents interfere with the expression of photosynthetic genes (Paul & Foyer, 2001). In some experiments where sinks are removed from source leaves, photosynthesis is downregulated before carbohydrates accumulate, suggesting that photosynthesis is regulated by changes in sugar and starch turnover rather than assimilate contents (Nebauer et al., 2011; in Citrus). An alternative hypothesis is that ABA accumulates in the source leaf following sink removal and triggers stomatal closure (Setter et al., 1981; in soybean). Indeed, source leaves continuously feed young leaves with ABA via the phloem (Cornish & Zeevaart, 1984; in cocklebur). This suggests that source and sink leaves cross-talk to regulate not only photosynthesis but also stomatal conductance and water relations (Fig. 4). In line with this view, the gradual transition from sink to source during leaf ontogeny is thought to occur concurrently with the establishment of a hydraulic limitation (Pantin et al., 2011). How hydraulics could establish during the course of leaf ontogeny is discussed in the next paragraphs.
3. Ontogenetic changes in leaf venation and aquaporins accompany leaf expansion
Water flux per leaf increases in pace with development of leaf area. Thus, as a first approximation, leaf venation should closely follow leaf expansion for water supply to match transpiration. Accordingly, vascular networks are subjected to ontogenetic changes (Fig. 3). In most dicots, venation is constructed according to a hierarchical order during which the 1° vein or midvein differentiates acropetally, followed by the 2° veins which expand towards leaf margins and define the boundaries of the network; the 3° and minor veins finally confer the reticulate pattern (reviewed in Sack et al., 2012). This pattern is tightly coordinated with cell division and mesophyll differentiation (Scarpella et al., 2004; Kang et al., 2007). As illustrated in Arabidopsis, the ontogenetic dynamics of the different vein orders overlap each other, with a peak in vein density as procambium forms followed by a dilution owing to leaf expansion, although the density of minor veins get stabilized during late expansion through a maintained initiation; meanwhile, each vein increases in diameter, especially in the two first orders (reviewed in Sack et al., 2012; Fig. 3). This general developmental pattern accounts for the global scaling relationships between leaf size and traits of major veins (Sack et al., 2012). By contrast, some species-specific variations in the formation of minor veins, for example, a decrease in minor vein density during late expansion (on five Solanaceous species, Gupta, 1961; on Trifolium repens, Denne, 1966), uncouple leaf size from the density of minor veins across species. If the ontogenetic pattern in vein density is well-described, knowledge about its functioning is far more sporadic. Leaf hydraulic conductance decreases during the expansion phase of the horse-chestnut leaf (Nardini et al., 2010) while it increases in poplar (Aasamaa et al., 2005), but data on the very early stages are lacking in both studies.
The temporal framework observed in dicots translates into a spatial framework in monocots. Thus, in tall fescue, axial hydraulic conductivity increases throughout the growing zone and decreases as soon as the leaf is exposed to light in relation to xylem maturation (Martre et al., 2001). Interestingly, the capacity of the extra-xylem pathway to transport water could coordinate with the xylem properties. In barley, the hydraulic conductivity of epidermis and mesophyll protoplasts is higher in growing than in non-growing tissues (Volkov et al., 2007). Concomitant changes in aquaporins could also be involved: in monocots, differential patterns in expression of several aquaporin isoforms are observed among developmental stages of a leaf, suggesting specific roles yet to be discovered for some isoforms during leaf ontogeny (Wei et al., 2007; Hachez et al., 2008; Besse et al., 2011; Fig. 4).
4. Adjustment of water efflux: stomata and cuticle
If water supply is actively adjusted during leaf ontogeny, water efflux from either stomata or cuticle also shows a dynamic developmental pattern. From a hydraulic viewpoint, transpiration competes with expansive growth for water (Figs 1, 2). However, possible benefits of transpiration for the growing leaf could be that high transpiration flux carries more nutrients (Cramer et al., 2008) and favours phloem assimilate import (Lacointe & Minchin, 2008; Fig. 4). Thus, the possible benefit for leaf growth of a given transpiration pattern should be interpreted carefully. The general ontogenetic trend for transpiration shows a pattern similar to net photosynthesis and follows the changes in stomatal conductance, with an early increase and a peak at 25–100% final leaf area (Constable & Rawson, 1980; Tichá et al., 1985; Fig. 3). Although generally well-coordinated, the peaks in net photosynthesis and stomatal conductance can be partly uncoupled (Čatský et al., 1985). Moreover, contrary to what is observed across species in mature leaves, there is no clear developmental relationship between stomatal conductance and stomatal density, which peaks before stomatal conductance at 10–60% of final leaf area, before a rapid dilution owing to leaf expansion (Tichá, 1985; Fig. 3). This suggests that stomatal functioning is acquired with leaf ontogeny. In line with this, the ontogenetic decrease in stomatal density is balanced by an increase in the size of stomatal complex and the area of stomatal pore (Tichá, 1985; Kheibarshekan Asl et al., 2011; Fig. 3), as well as an increase in K+ in guard cells (Pappas et al., 1988). Thus, mature leaves have generally a higher stomatal conductance than young leaves where the stomata are typically more dense but occluded or underdeveloped (Snider et al., 2009). Finally, the boundary layer conductance may also vary dramatically during the course of leaf ontogeny because of changes in size, exposure and surface roughness of the leaf, but the huge differences in these variables between species as well as the scarcity of available data prevent drawing conclusions about a general trend.
If plants actively regulate their water balance through stomatal movements, the leaf cuticle acts as the major barrier to passive water loss (Kerstiens, 2006). Dramatic changes in cuticle composition and thickness occur during leaf ontogeny. In monocots, where the growing zone is enclosed by sheaths of older leaves, cutin deposition parallels cell elongation, while the spatiotemporal pattern of cuticular wax deposition is regulated to anticipate the moment when epidermal cells are exposed to the atmosphere, as in barley (Richardson et al., 2007) or maize (Hachez et al., 2008). In dicots, where leaf expansion occurs while tissues are exposed to the atmosphere, the cuticle thickens throughout leaf development because synthesis and deposition of both cutin and waxes are well coordinated with cell expansion, while wax composition changes continually, as seen in a broad range of species (Hauke & Schreiber, 1998). In both monocots and dicots, cuticle thickening as well as accumulation of waxes during the course of leaf expansion coincides with a decrease in cuticular permeance (Hauke & Schreiber, 1998; Richardson et al., 2007; Fig. 3). In line with this, the response to osmotic stress of genes involved in cuticle biosynthesis is differentially modulated by leaf developmental stage in Arabidopsis (Skirycz et al., 2010). Thus, the dynamic behaviour of the cuticle throughout leaf ontogeny is likely to influence water relations of the growing leaf.
5. Coordination between leaf rheology and leaf developmental program
The ontogenetic changes in water balance result in ontogenetic changes in leaf turgor, as illustrated in maize (Bouchabké et al., 2006) or Arabidopsis (Pantin et al., 2011). However, a direct conclusion on the role of turgor in relation to growth is difficult because leaf rheology also changes as the leaf ages (Fig. 4). In Robinia pseudoacacia, turgor increases during the course of leaf expansion but extensibility decreases while yield threshold increases, leading to a developmental decrease in leaf growth rate (Zhang et al., 2011). At the molecular level, genes associated with wall properties are differentially expressed in Arabidopsis leaves of contrasting ages (Cho & Cosgrove, 2000; Skirycz et al., 2010). Similarly, in maize, developmental expression patterns of expansins correlate with growth rates of leaf segments of different ages whatever the experimental condition, suggesting that expansins are downstream, unspecific targets of a range of developmental, environmental and genetic sources of variation (Muller et al., 2007). In barley, plasma membrane H+–ATPase activity is enhanced in the elongating zone of the leaf, suggesting that the apoplast is locally acidified to facilitate turgor-induced cell wall deformation (Visnovitz et al., 2012). Moreover, genes and proteins associated with wall synthesis are differentially regulated along the developmental gradient of maize leaves (Li et al., 2010; Majeran et al., 2010), while in aspen leaves, metabolic markers of cell wall maturation relate to the sink-to-source transition (Jeong et al., 2004). Finally, the cuticle itself strengthens and stiffens during leaf ontogeny and may gradually constrain expansion (Takahashi et al., 2012). These results support the view that leaf rheology is tightly adjusted during the course of leaf ontogeny to coordinate its growth rate with its developmental programme, including the dynamics of its hydraulic and metabolic limitations.
6. Leaf ontogeny reprograms the circadian clock
The ontogenetic gradients in leaf growth patterns arising from these developmental changes in hydromechanical and metabolic limitations imply that several growth rhythms coexist in plants where leaves of several developmental stages are present. This brings into question the role of the circadian clock in the central regulation of such a system. Experiments on Arabidopsis hypocotyls (Nusinow et al., 2011), roots (Yazdanbakhsh et al., 2011) or leaves (Poiré et al., 2010) strongly support the view that the clock is a major actor in the control of growth rhythms. The clock tunes the transcription of genes involved in photosynthesis, carbon metabolism, water influx, wall mechanics (Harmer et al., 2000), ABA metabolism (Fukushima et al., 2009) and ABA response (Mizuno & Yamashino, 2008). We propose that the clock orchestrates the actors of leaf growth in a differential manner throughout leaf ontogeny by anticipating metabolic and hydraulic constraints (Fig. 4). In support of this hypothesis, genes related to the circadian photoperiod are differentially expressed along the developmental gradient of the maize leaf (Li et al., 2010). Stomatal conductance and photosynthesis (Arabidopsis, Dodd et al., 2005), leaf starch breakdown (Arabidopsis, Graf et al., 2010), leaf hydraulic conductivity (sunflower, Nardini et al., 2005), aquaporin expression, hydraulic conductivity and water distribution in roots (Lotus japonicus, Henzler et al., 1999; Arabidopsis, Takase et al., 2011), all have been shown to oscillate with the clock. Interestingly, the shoot clock synchronizes the root clock (a simplified version of the shoot clock) by a photosynthesis-related signal (James et al., 2008). Although very speculative, it is tempting to generalize this relationship to other sink organs such as young leaves. Because changes in carbon and water relations occur during leaf ontogeny (see sections V.2 'Leaf carbon balance and the sink-to-source transition' to V.4 'Adjustment of water efflux: stomata and cuticle') and because the clock itself is regulated by sugars (Dalchau et al., 2011) and ABA (Legnaioli et al., 2009), both sugars and ABA could modulate the influence of the clock on leaf growth throughout leaf development. Furthermore, both sugars and ABA allow communication between leaves of contrasting ages because both compounds move through the phloem (Fig. 4). Finally, the involvement of the mobile flowering regulator FT in circadian movements of stomata and its expression in vasculature (Hubbard & Webb, 2011) also make it a putative candidate able to mediate differential coordination of water and carbon fluxes by the clock during leaf ontogeny.
VI. The growing leaf in a changing world
The partially uncoupled changes in carbon and water relations experienced by the growing leaf as it develops make it differentially sensitive to environmental stresses throughout its ontogeny. Furthermore, young and mature leaves cross-talk to adjust their respective response to environmental stresses. For example, systemic signals from mature leaves exposed to elevated CO2, shading or low humidity trigger decreases in stomatal density of new, nonexposed leaves in Arabidopsis or poplar (Lake et al., 2001; Coupe et al., 2006; Miyazawa et al., 2006), in a mechanism putatively involving sugar signalling, ABA, and transpiration rate itself (Lake et al., 2002; Lake & Woodward, 2008). We show in this section how the effects on leaf growth of those environmental stresses challenging water or carbon status are conditional on leaf ontogeny.
1. Young leaves strongly rely on light and CO2: sink sensitivity to carbon starvation
Because structural growth of young leaves requires a net import of carbon before the sink-to-source transition (see section V.2 'Leaf carbon balance and the sink-to-source transition'), environmental inputs affecting carbon balance have stronger impacts at early stages of leaf ontogeny (Fig. 4). The sensitivity of young leaves to assimilate supply is illustrated in shading experiments where young leaves exhibit a marked reduction of growth, unlike older leaves (Granier & Tardieu, 1999; Tardieu et al., 1999; Muller et al., 2001; Cookson & Granier, 2006). Consistent with this, leaves of several species grown under high CO2 grow faster, especially during the early exponential growth phase (Kriedemann & Wong, 1984; Ferris et al., 2001) and effects are visible at the transcriptional level (Ainsworth et al., 2006; in soybean). In line with this, leaves acclimatize to elevated CO2 early in their ontogeny. For example, when sugarcane or grain sorghum plants are subjected to high CO2, photosynthesis and central metabolism are upregulated early in leaf ontogeny and decline to levels comparable to those in control conditions at later stages (Vu et al., 2006; Prasad et al., 2009). Photosynthesis of tobacco leaves exposed to elevated CO2 reaches a maximum while leaves are still expanding and declines with leaf ageing earlier than in control plants, accounting for the downregulation of photosynthesis generally observed in mature leaves after long-term exposure to high CO2 concentrations (Miller et al., 1997). The effect of high CO2 may also be dependent upon the capacity of the source leaf to export assimilates at higher rates. In Ricinus communis, the phloem loading system of source leaves is saturated during the day under either ambient or elevated CO2 (sink limitation), while the export rate at night is reduced in plants under ambient CO2 (source limitation) (Grimmer & Komor, 1999). The benefit of elevated CO2 to leaf growth may not be mediated by a sole substrate effect. Indeed, wall extensibility at elevated CO2 is increased in the young leaves but not in older leaves of poplar (Taylor et al., 2003). Finally, considering that high temperature can induce carbon starvation (Vasseur et al., 2011), elevated CO2 and supra-optimal temperature are expected to have opposite effects on the growth of young leaves. This point deserves particular interest in a world where both atmospheric CO2 and temperature are continuously rising.
2. A crosstalk between leaves favours drought tolerance in the young leaf
It is commonplace to observe that when exposed to severe water deficit, young leaves start to wilt after older leaves and keep the capacity to boost their expansion rate when water is resupplied (Rawson & Turner, 1982; Fig. 4). Young leaves also accumulate more ABA than older leaves because of high import from source leaves combined with a low catabolism but, paradoxically, their stomatal conductance remains high and their water potential therefore becomes very low (Jordan et al., 1975; Raschke & Zeevaart, 1976; Sivakumaran & Hall, 1978; Cornish & Zeevaart, 1984). That the leaf response to ABA would be impaired in young leaves is consistent with a preferential transcription in mature leaves of the genes related to ABA signalling, as shown in Arabidopsis under osmotic stress (Skirycz et al., 2010). This ontogenetic gradient in leaf response to ABA could be related to acclimatization mechanisms, as some events in leaf history, such as early exposure to stress or ABA, modify the current ability of growth and stomata to respond to changes in leaf water status (Atkinson et al., 1989; Pospíšilová, 1996; Fanourakis et al., 2011). How could turgor be maintained in young leaves under water stress, despite a high stomatal conductance and a low water potential? We propose that maintaining transpiration only in the young, small leaves would favour solute translocation and osmotic adjustment therein, while minimizing water loss from the whole plant. Beyond these hydraulic relationships between source and sink leaves, mechanisms facilitating stress recovery and preventing ABA-induced senescence in the young leaves (Sivakumaran & Hall, 1978; Cornish & Zeevaart, 1984; Atkinson et al., 1989; Lee et al., 2011) could also reconcile ABA insensitivity and drought tolerance. From the above rationale, we raise the hypothesis that these cross-talks between young and older leaves confer an adaptive advantage to young leaves under water stress.
Given the negative effect of water deficit on photosynthesis, the conclusion that young leaves are more drought-tolerant than older leaves may seem contradictory with the fact that young leaves are more sensitive to carbon starvation (see section VI.1 'Young leaves strongly rely on light and CO2: sink sensitivity to carbon starvation'). However, because drought affects growth long before photosynthesis, plants are generally not carbon-limited at the beginning of soil drying and growth becomes uncoupled from carbohydrate availability in sink organs (Muller et al., 2011). Sugars in excess can be diverted to contribute to osmotic adjustment at low cost (Hummel et al., 2010), with a priority to young leaves that may result, for example, from differential activity of invertases between source and sink leaves (Kim et al., 2000; Luquet et al., 2008). Accordingly, expanding leaves under water stress develop source features earlier in their ontogeny than leaves of well-watered plants (Schurr et al., 2000). This time-lag of sensitivity to drought between growth and photosynthesis releases young leaves from their strong dependence on carbon availability, solving the apparent paradox that young leaves are more drought-tolerant than adult leaves, although they are more susceptible to carbon supply.
Leaf growth can be envisioned as a dual increment in water volume and in dry matter, two processes tightly coupled with the expansive growth and the structural growth. Analyses of the determinisms of water relations, tissue rheology and carbon partitioning at various scales have shown several interaction nodes between each other. First, transport systems for water and carbon within the plant share some key components: stomata arbitrate the trade-off between carbon gain and water loss; the aquaporin family comprises isoforms able to facilitate the transport of water or CO2; and the xylem is hydraulically coupled with the phloem. Second, the leaf water transport capacity is coordinated with its photosynthetic capacity, a property that can be considered to be the result of an evolutionary driven trade-off between the space-filling property of leaf venation and the carbon construction costs of veins. This translates into a crosstalk in the development of veins and stomata. Third, the Lockhart hydromechanical model accounting for volumetric growth has potentially interesting connections with carbon metabolism, including osmotic adjustment and wall biochemistry. Fourth, the drought hormone ABA has a central position in regulating water and carbon fluxes, by acting on stomata and aquaporins, flowing in every part of the plant through both xylem and phloem, and interfering with sugar sensing.
Acknowledging this complexity sheds light on the controversial nature of the debate about the main limitation of leaf growth. Certainly related to this is the active debate on the relative contribution of hydraulic and metabolic limitations to tree survival under severe drought episodes (McDowell, 2011). Here, we argue that leaf ontogeny has to be taken into account to study the influence on leaf growth of hydraulics, mechanics and metabolics. Source–sink relationships provide a cornerstone to address ontogenetic changes in leaf carbon balance. Water relations are likely to be ontogenetically modulated through functional changes in xylem network, stomata and aquaporins. Leaf rheology is modified during the course of leaf ontogeny through changes in biochemical properties of the walls, driven by the activity of key enzymes and proteins. The circadian clock, which exerts control over carbon partitioning, water relations and wall mechanics, could display some asynchrony between the young, sink leaves and the older, source leaves. The regulation of long-distance transport of sugars and ABA is a central question to tackle because these compounds interplay with numerous elemental processes involved in either expansive or structural leaf growth. Overall, we defend the view that the control of leaf growth switches from a metabolic limitation to a hydromechanical limitation during the course of leaf ontogeny. Finally, we believe that considering leaf ontogeny will greatly improve our knowledge of the mechanisms underlying leaf growth responses to the abiotic stresses that affect water and carbon status, such as those forecasted for crops and wild plants in our changing world.
We acknowledge the continuous support from INRA, Environment & Agronomy Department, the financial support of the French ministry of research to F.P. during his PhD, the French Agence Nationale de la Recherche (grants ANR-05-PJCJC-0052-01 and ANR-06-BLAN-0122-01) and the EU-FP6 Integrated project ‘Agronomics’ (LSHG-CT-2006-037704). We also acknowledge fruitful interactions with Alain Vavasseur, Bernard Genty, Yves Gibon, Christophe Maurel, André Lacointe and our colleagues from the LEPSE, especially François Tardieu, Denis Vile and Christine Granier. Heather Maclean provided useful corrections of the English language. The constructive comments by Lawren Sack (and his suggestion to build Fig. 3) and by two other anonymous reviewers contributed to greatly improve this manuscript. We apologize to colleagues whose work could not be cited because of space limitations.