The wall surrounding plant cells provides protection from abiotic and biotic stresses, and support through the action of turgor pressure. However, the presence of this strong elastic wall also prevents cell movement and resists cell growth. This growth can be likened to extending a house from the inside, using extremely high pressures to push out the walls. Plants must increase cell volume in order to explore their environment, acquire nutrients and reproduce. Cell wall material must stretch and flow in a controlled manner and, concomitantly, new cell wall material must be deposited at the correct rate and site to prevent wall and cell rupture. In this review, we examine biomechanics, cell wall structure and growth regulatory networks to provide a ‘big picture’ of plant cell growth.
Plants use plastic growth to construct cells that vary enormously in size and shape (Fig. 1). Cell volumes can vary 10 000-fold within a species, increasing in size from meristemoids to xylem vessels (Cosgrove, 2005). Many plant cells are cylindrical or tubular, but cells may also be spherical, for example epidermal bladder cells in Mesembryanthemum crystallinum, or even stellate, as in the aerenchyma of Cyperus javanicus. Complex plant cell shapes are produced by tight regulation of growth, as in the interdigitation of Arabidopsis thaliana leaf epidermal cells, and may be maintained by structural reinforcement after the cessation of growth, as in xylem vessels. How plant cells are shaped has been discussed elsewhere (Martin et al., 2001) and, in this review, we focus on the mechanisms of cell growth and how these mechanisms are regulated. Understanding growth at the cellular level is vital because plant shape is dictated by two factors: cell number and cell size. Plant development includes numerous indeterminate processes, and responds drastically to environmental stimuli in order to maximize reproductive potential within the local environment. For example, Cryptomeria japonica can be both a diminutive bonsai and the ancient Jomon-sugi, which has a volume of c. 300 m3. Plant cell growth is produced and controlled at numerous different levels, which we will highlight in turn. Ultimately, plant cell growth is a biomechanical process governed by physical laws, which mathematics and modelling attempt to describe. The forces and material properties present in this process are determined by the osmolality of plant cells and the composition of their walls. Furthermore, these structural and biochemical features of plant cells are controlled by complex regulatory networks, which integrate diverse developmental and environmental inputs. A full understanding of plant cell growth requires a broad perspective of these different hierarchical levels, and the scientific fields that describe them.
II. Modes of growth
Plant cell growth is predominantly anisotropic; the rate and direction of growth vary across the cell. This allows the plant to form cell shapes other than spheres, which are produced by isotropic growth. Anisotropic growth includes diffuse and tip growth. Plant cells grow mainly by diffuse growth, in which expansion is dispersed over the cell surface, for example root cortex cells elongating in the longitudinal direction as they leave the root meristem. Some specialized cells produce structures by tip growth, such as root hairs and pollen tubes, in which expansion occurs at a single site. Diffuse growth is unique to plant cells, whereas some animal and fungal cells also undergo tip growth. There are numerous important differences between these growth forms. Wall material is deposited perpendicular to the direction of maximal growth in diffuse growth, whereas deposition is parallel with maximal growth in tip growing cells. Diffuse growth is equivalent to stretching a wall lengthwise whilst reinforcing it to prevent rupture; tip growth is like trying to burrow through the wall without breaking its surface. In addition to this mechanical distinction, the cytological properties of each mode of growth differ.
III. Biomechanics and modelling of plant cell growth
Plant cell growth occurs through the controlled expansion of the cell wall, which results from the interplay between turgor pressure and cell wall elasticity and extensibility (Ray et al., 1972). Turgor pressure is produced by the influx of water from the extracellular space, as a result of the lower osmotic potential of the cytoplasm. Turgor pressures in the range 0.3–1.2 MPa (similar to that within a car tyre) have been measured in A. thaliana cells (Forouzesh et al., 2013). Within a turgid cell at equilibrium, this outwards pressure is counterbalanced by stress within the cell wall, and produces a water potential difference of zero. Wall stress acts to oppose changes in wall shape, and therefore increases in cell volume. The negligible compressibility of water at physiologically relevant pressures ensures that any increase in cell volume (i.e. growth) requires lower water potential within the cell than outside it. The Lockhart equation (Lockhart, 1965) was the first to describe diffuse growth by linking turgor pressure and wall extensibility to volume increase:
(, rate of volume increase; Φ, irreversible wall extensibility; P, pressure; Y, yield threshold).
This models the cell wall as a viscoelastic substance that undergoes plastic deformation when turgor pressure exceeds the yield threshold. The yield threshold is the point above which pressure produces growth, and irreversible wall extensibility governs how much growth this excess (i.e. above yield threshold) pressure produces. The yield threshold describes how much energy the wall components can store elastically before being permanently deformed or moved relative to each other. Irreversible wall extensibility describes the stiffness of the cell wall. Both of these values are determined by the composition of the cell wall and the interactions between the cell wall components, although these relationships are complex and not currently well understood.
If a turgid cell is at equilibrium ( = 0, P > 0), negative water potential must be generated for growth to begin. This can occur, mathematically, by either increasing the concentration of solutes within the cell or reducing stress within the wall and thereby decreasing turgor pressure (Ray et al., 1972). The Lockhart equation has been modified repeatedly, for example to include factors such as elastic wall extensibility and transpiration rate (reviewed in Geitmann & Ortega, 2009). Recent approaches have also modelled Φ as variable through space and time (Pietruszka, 2011). Although the Lockhart equation is simplistic, it still encapsulates the key concepts thought to govern cell growth. It remains a point of debate whether osmolality or wall stress is altered first, to remove a cell from equilibrium and begin its growth. Put simply, is plant cell growth ‘wall first’ or ‘water first’? The classic Lockhart model has been challenged by loss-of-stability (LOS) theory (Wei & Lintilhac, 2003). LOS theory attributes growth to structural failure of the wall as a result of increasing turgor rather than active modification of wall properties (Wei & Lintilhac, 2007; Schopfer, 2008; for an opposing view). A weakness of LOS theory is that it cannot predict wall behaviour after the initiation of growth (Geitmann & Ortega, 2009).
The principles behind the modelling of diffuse growth have also been adapted to try and recreate the dynamics of tip growth, with alterations to account for the unique geometric and biological features of tip growth. Recent models include the hydrodynamic model and a calcium-cell-wall model. The hydrodynamic model suggests that water flux within pollen tubes can produce pressure gradients, and that this increased pressure at the tip drives growth (Zonia & Munnik, 2011). This model has been challenged on physical grounds, and because a pressure gradient is yet to be recorded within plant cells (Winship et al., 2011). An alternative calcium-cell-wall model has also been proposed, and is based on stretch-activated calcium channels in the pollen protoplast. As the cell wall thins with growth, wall stress increases (a general physical principle). Above a certain level of stress, this opens the calcium channels, stimulating vesicle fusion and wall material deposition at the tip (Kroeger et al., 2011). The evidence for models (of both diffuse and tip growth) that rely on wall modification to initiate growth is currently stronger than that for ‘water first’ models, such as LOS, and the hydrodynamic model. It remains unknown whether cells ever actively increase turgor pressure in order to grow, and ‘… in not a single example of cell growth has an active increase in turgor pressure been documented’ (Winship et al., 2011; Kierzkowski et al., 2012; Kutschera & Niklas, 2013).
Regardless of whether modification of wall properties or turgor pressure initiates growth, it is clear that wall modification is required to direct growth within a system governed by hydrostatic pressure. Turgor pressure is a scalar physical force and, although it is possible to exert active control over its magnitude, through the active transport of ions and solutes, its direction cannot be controlled. Consequently, turgor pressure would produce a sphere if the cell wall had a homogeneous capacity for expansion. Plants cells must therefore spatially regulate construction and/or alteration of their walls to produce non-spherical cells. Counter-intuitively, although plant cells require turgor pressure for growth, in the widespread phenomenon of anisotropic growth, plant cells are working actively against the largest pressure forces to produce growth that takes the cell shape further away from spherical. In a growing cylindrical cell, the hoop stress within the wall is double the longitudinal axial stress, but most expansion occurs in the longitudinal direction. Stress patterns for more complex cell shapes do not seem to have been formally investigated. Unless notable hydrodynamic pressure can be produced by the directional flow of water through plant cells, it is universal that turgor pressure is a requisite for growth, but when and where it occurs is controlled by wall modification.
IV. Plant cell wall structure
We briefly summarize the structure of the plant cell wall before discussing how its modification and composition can control growth. The plant cell wall is a complex and heterogeneous matrix of polysaccharides, glycoproteins, solutes and enzymes, which varies in composition within and between individual cells, tissues and species. We focus on the major wall polysaccharides, but there are numerous other wall components, such as glycoproteins and phenolics. Many excellent and recent reviews on the plant cell wall cover general structure (Cosgrove, 2005; Cosgrove & Jarvis, 2012) and specific components (Bunzel, 2010; Lamport et al., 2011). Primary cell walls are thin, lack lignification and are able to undergo growth, so are referred to in this review.
Cellulose, linear β-1,4-glucan, is the key structural component of cell walls. Crystalline cellulose microfibrils are inelastic and produced by multimeric complexes of cellulose synthases (CESAs) at the plasma membrane, which contain up to 36 CESA proteins. CESA proteins catalyse the extension of β-1,4-glucan polymer chains, which then spontaneously cocrystallize into microfibrils with other chains produced by the same complex (Somerville, 2006). In A. thaliana, CESA complexes require at least three different CESA subunits. The primary cell wall is produced by CESA1, CESA3 and, typically, CESA6 (Persson et al., 2007).
The remaining matrix polysaccharides are synthesized in Golgi stacks and released into the wall by exocytosis. Pectins are an extremely diverse group of complex high-molecular-weight polysaccharides that are soluble in aqueous and acidic conditions. Structural groups of pectins include homogalacturonan (HG), rhamnogalacturonan I and II, and xylogalacturonan. These components form complex hydrated gels, cross-linked by covalent and ionic bonds between different pectins and other wall polysaccharides (Marcus et al., 2008; Caffall & Mohnen, 2009). Pectins contribute strength and flexibility to the wall, separate microfibrils and link adjacent cells in the middle lamellae (Caffall & Mohnen, 2009).
Hemicelluloses are typically defined as polysaccharides that are not cellulose or pectins, which primarily increase wall strength by cross-linking cellulose microfibrils (reviewed in Scheller and Ulvskov, 2010). Xylans, xyloglucans (XGs), glucomannans and mannans are hemicelluloses thought to be present in all land plants. XG is the primary hemicellulose in dicotyledonous plant cell walls, and interacts with pectins and cellulose microfibrils, with a small proportion entrapped within microfibrils (Dick-Pérez et al., 2011).
The structure of the cell wall has been conceptualized previously as a ‘tethered network’ of cellulose cross-linked by XG, which is present within a relatively separate pectin matrix (Cosgrove, 2005). However, recent three-dimensional solid state nuclear magnetic resonance (NMR) of A. thaliana cell walls suggests that pectin is an intrinsic part of the load-bearing network of the wall. Pectins were found to interact extensively with both XG and cellulose, whereas XG–cellulose interactions were less common than previously thought (Dick-Pérez et al., 2011). In addition, the use of various β-1,4-endo-glucanases demonstrated that XG digestion alone was not sufficient to produce polymer creep (Park & Cosgrove, 2012b), which would be expected if the tethered network model represents the load-bearing structure of the cell wall. The new cell wall structure suggested by these data is that of a single complex polysaccharide network featuring interactions between each of cellulose, XG and pectin (Fig. 2; Dick-Pérez et al., 2011). These data leave open the question of whether microfibrils are linked together by XG, pectin or both, whether alone or in combination (all possibilities shown in Fig. 2), and further work should aim to determine the network topography.
V. Cell wall modification proteins
Plants can control wall structure both through alteration of the wall materials deposited outside the plasma membrane and by dynamic modification of the composition and interaction of wall components after their deposition. Post-depositional modification of cell wall polysaccharides allows plant cells to alter the physical properties of their walls in response to novel stimuli. This process is conceptually similar to the post-translational modification of proteins, which allows rapid modification of the structure, interactions and stability of proteins already present as a result of translation. Plant cells release numerous classes of proteins, such as glycosyltransferases, that alter cell wall properties and cell growth through the modification of different wall components (summarized in Table 1). We focus on two well-studied examples directly linked to cell growth: expansins, the classic example of a wall-modifying protein, and pectin methylesterases (PMEs), which have been shown recently to play an important role in the growth of pollen tubes, shoot apical meristem (SAM) primordia and hypocotyls (Tian et al., 2006; Pelletier et al., 2010; Peaucelle et al., 2011).
|β-Galactosidases||AtBGAL10||Xyloglucans (XG)||Cell wall||Figueiredo et al. (2011); Sampedro et al. (2012)|
|Expansins||AtEXP7||Cellulose microfibrils, XG||Cell wall||Cosgrove (2000)|
|Glycosyltransferases||AtXEG113||Extensins, HRGPsb||Golgi||Velasquez et al. (2011)|
|Pectin methylesterases||AtPME5||Pectins (e.g. homogalacturonan)||Cell wall||Pelloux et al. (2007); Peaucelle et al. (2011)|
|Prolyl 4-hydroxylases||AtP4H2||Extensins, HRGPs||Golgi, endoplasmic reticulum||Velasquez et al. (2011)|
|Transglycosylases||Unknown||XG, hemicelluloses||Cell wall||Johnston et al. (2013); Mohler et al. (2013)|
|UDP-D-Glucose 4-epimerase||AtRHD1||XG, pectins||Golgi||Rösti et al. (2007)|
|Xyloglucan endotransglucosylase/hydrolases||AtXTH18||XG||Cell wall||Van Sandt et al. (2007); Eklöf & Brumer (2010)|
Acid growth theory (Rayle, 1973) states that the influx of hydrogen ions to the wall causes loosening, allowing increased expansion through polymer creep. Expansins, the proteins generally accepted to be responsible for acid growth, were discovered in the early 1990s. Heat-inactivated walls had pH-dependent expansion rescued by the addition of proteins extracted from cucumber seedlings, which could also induce extension in other species (McQueen-Mason et al., 1992). Another family of expansins was later identified as a group of grass pollen allergens that loosen cell walls (Cosgrove et al., 1997). These two families of expansins, named α-expansin or EXPA and β-expansin or EXPB, respectively, are involved in a wide variety of cell wall modification processes, including fruit ripening, abscission, and the penetration of the style by pollen (Cosgrove, 2005). Expansins are present throughout land plant phylogeny, their expression corresponds with growth (Cosgrove, 2000) and expansin antisense treatment reduces cell size in vivo and cell wall extensibility in vitro (Cho & Cosgrove, 2002; Goh et al., 2012).
Expansins are small extracellular proteins with no apparent lytic activity that enhance polymer creep, and thereby wall extensibility, by disrupting non-covalent interactions between wall polysaccharides, in particular between cellulose microfibrils. Expansins weaken paper, which is a network of microfibrils with few other polysaccharides, and enhance cellulase activity, which is limited by the accessibility of glucan (Cosgrove, 2005). Targeted mutagenesis using bacterial EXPANSIN-LIKE X1 (EXLX1) showed that mutations that decreased hydrophobic cellulose binding reduced in vitro wall-loosening activity, whereas mutations that decreased electrostatic binding to pectins and hemicelluloses in fact increased wall-loosening activity (Georgelis et al., 2011). These data raise the intriguing possibility that the modification of non-cellulosic polysaccharides to alter their charge could modulate expansin activity. These key residues correspond to similar residues in plant expansins, but EXLX1 has lower wall-loosening activity and its activity is not pH dependent, suggesting possible differences in mechanism.
Even if the mechanism of expansin action only involves microfibrillar interactions, other wall polysaccharides affect expansin activity indirectly. For example, the A. thaliana XG synthesis double mutant xylosyltransferase1 (xxt1)/xxt2, which contains no known XGs, has walls with altered biomechanical properties. Cell wall responses to α-expansin activity were reduced in xxt1/xxt2 mutants (Park & Cosgrove, 2012b). This decrease in activity may be caused by XG absence if expansins act on them directly, or if XGs promote cellulose structures on which expansins act. However, pectin and arabinoxylan play a greater structural role in XG-deficient walls (Park & Cosgrove, 2012a), which may, in turn, be responsible for altered expansin activity. It appears that functional divergence of the expansin superfamily has occurred, as the application of purified maize pollen β-expansin was found to release glucuronoarabinoxylan and HG from the cell walls of grasses, but not other plants, probably through a non-enzymatic mechanism (Tabuchi et al., 2011).
2. Pectin methylesterases
PMEs catalyse the removal of methyl esters from the (1→4) α-d-GalA backbone of HG, the most abundant cell wall pectin. HG is deposited at the cell wall in a 70–80% methyl-esterified form. PME action produces a negatively charged carbon on the HG backbone, allowing the formation of calcium cross-links between chains of unmethyl-esterified GalA residues. The egg-box model describes this cross-linking (Liners et al., 1989): two negatively charged pectin chains chelate positive calcium ions between them. This modification also increases HG susceptibility to lysis catalysed by polygalacturonanases (PGases) and pectate lyases within the wall (Caffall & Mohnen, 2009). Calcium cross-linking increases the stiffness of cell walls, but pectin degradation increases wall extensibility. This seeming contradiction may be resolved by the pattern of demethyl-esterification. PMEs can remove methyl esters from the same HG backbone in a processive fashion, producing a string of negatively charged carbons, or randomly from multiple backbones. It is thought that ‘blocks’ of negative charge are most important for forming gels, whereas random methyl ester removal may promote degradation by PGases without causing a large increase in wall stiffness (Pelloux et al., 2007). PME activity, appropriately, has been found to be responsible for both the promotion and inhibition of cell wall growth.
Recent work has linked PME activity to cell growth in a number of developmental processes. In the A. thaliana SAM, lateral organs are produced from lateral outgrowths called primordia. Subepidermal cells in lateral organ primordia are more elastic and grow more quickly than surrounding cells (Peaucelle et al., 2011). PME activity is likely to be at least partly responsible for this change in wall properties and cell growth rate. Antibody labelling showed that methyl-esterified HG dominated the centre of the SAM dome, whereas demethyl-esterified HG was present in incipient and emerging primordia, suggesting the involvement of PMEs in primordium formation. Ethanol-inducible expression of the endogenous PME inhibitor PMEI3 prevented lateral organ outgrowth, and citrus PME loaded onto sepharose beads induced primordium initiation, showing that PME activity is necessary and sufficient for increased cell growth in the SAM during lateral organ outgrowth (Peaucelle et al., 2008). By contrast, tip growth in pollen tubes uses PME activity to stiffen cell walls in the tube shank: highly methyl-esterified HG is deposited at the growing tip, where the wall is softer (reviewed in Chebli et al., 2012). PME activity is restricted by tip-localized PME inhibitors, such as AtPMEI2 (Röckel et al., 2008).
There are a large number of putative PME isoforms in all sequenced plant genomes. For example, Populus trichocarpa has 89, A. thaliana has 66 and Oryza sativa has 35 (Pelloux et al., 2007). This diversity reflects their wide range of functions. PMEs are involved in seed germination, root tip elongation, lignin production (Pelloux et al., 2007) and pathogen responses (Lionetti et al., 2012). Isoforms differ in optimum pH, salt dependence and action mechanism (Jolie et al., 2010). The charaphacean alga Chara corollina utilizes a Ca+-pectate mechanism to control growth, and it has been suggested that this mechanism is maintained in land plants, with the addition of the cellulose–expansin system to complement it (Boyer, 2009; Peaucelle et al., 2012). Recent studies have shown PMEs, and other enzymes with pectin substrates, to be important in both starting and stopping plant cell growth (Peaucelle et al., 2012), and their ubiquity and diversity suggest that they may provide an integral mechanism of cell growth.
VI. Direction and duration of cell growth
Cell growth must be tightly controlled to construct a complex multicellular organism. Therefore, the control of the direction of cell growth and the duration of this growth is essential. We examine how diffuse growth and tip growth are guided, and discuss how the cell cycle may be involved in the control of the duration of cell growth.
1. Directing diffuse growth
Diffuse growth is essential for the growth of plant roots and shoots. Diffuse growth may occur over whole planes of the cell, as in the longitudinal expansion of hypocotyl cells, or on a portion of their surface, as in the interdigitation of epidermal leaf pavement cells. Diffuse growth requires the deposition of wall material over a wide area of the cell surface, and the wall must be constructed anisotropically. This allows the wall to respond differently to the stress produced by turgor pressure depending on its direction, which controls the direction of growth.
The orientation of cellulose microfibril arrays is a major determinant of the direction of diffuse growth; they are typically perpendicular to the axis of maximum elongation. The alignment hypothesis (Heath, 1974) suggested that cellulose deposition could be oriented by the cortical microtubule (MT) array. This was confirmed by observation of a fluorescent CESA6 construct (Paredez et al., 2006). CESA proteins are deposited preferentially at sites on the wall beside cortical MTs (Gutierrez et al., 2009), and MTs confer direction to CESA complex movement (Paredez et al., 2006). The association of complexes with individual MTs in cells treated with the MT-destabilizing drug oryzalin (Paredez et al., 2006) suggests that complexes are tethered to individual MTs, rather than constrained passively between MT arrays, as suggested previously (Giddings & Staehelin, 1991).
A crucial factor for CESA–MT association has been identified recently. CELLULOSE SYNTHASE INTERACTIVE PROTEIN1/POM-POM2 (CSI1/POM2) is required for microfibrils of normal length and orientation (Gu et al., 2010), binds MTs in vitro and is required for the association of CESA complexes and MTs in vivo (Bringmann et al., 2012; Li S et al., 2012). The guidance of CESA complexes by MTs provides a mechanism for plants to control the anisotropy of their walls. MT positioning was found to align with predicted principal stresses in the SAM, and to realign appropriately when stress patterns were altered through laser ablation (Hamant et al., 2008). This suggests that mechanical stress at the cellular and tissue level regulates MT placement and may therefore regulate microfibril arrangement in individual cells.
Ordered local cellulose deposition does not require an ordered cortical MT array, suggesting the existence of a separate mechanism that operates redundantly to produce transverse parallel arrays of microfibrils. Arabidopsis thaliana root epidermal cell MTs deviate from a transverse orientation roughly coincident with a deceleration in growth rate, whereas microfibril orientation remains transverse until long after cells cease elongation (Sugimoto et al., 2000). Furthermore, disruption of transverse cortical MTs using a temperature-sensitive mutant or oryzalin does not alter microfibril orientation significantly (Sugimoto et al., 2003). These results could be explained if CESA complex movement can be guided by the microfibril structure of the wall interior, that is, if CESA movement parallel to the microfibril array is energetically favoured (Baskin, 2001). However, this kind of mechanism was challenged with the observation that microfibril deposition can return to a parallel transverse array, without MTs, even after cell wall microfibril orientation is randomized (Himmelspach et al., 2003). The status of pectin within the wall may influence cellulose deposition. Cobtorin disrupts the deposition of transverse cellulose microfibrils, but leaves parallel cortical MT arrays intact (Yoneda et al., 2007). Overexpression of a PME or a PGase can supress this disruption and return cellulose deposition to a direction parallel to cortical MTs (Yoneda et al., 2010).
2. Directing tip growth
Tip growth is a specialized form of growth that is spatially regulated in the extreme. This growth form therefore provides a useful system for understanding how plant cells convert a scalar force, pressure, into growth, a vector. Land plants construct root hairs, pollen tubes and bryophyte rhizoids using tip growth. There are numerous differences between tip growth in different cell types and phylogenetic groups, but general features are as follows: an apical region enriched in vesicles; enrichment of Golgi stacks and mitochondria, for secretion; longitudinal arrangement of mictotubules and actin bundles along the tube shank, and cytoplasmic streaming (Rounds & Bezanilla, 2013).
The first stage of tip growth is the specification of the growth site. In A. thaliana root hair cells, there is a local auxin gradient with the highest concentration at the basal end, where the root hair forms; abolishing this gradient leads to hair formation at the apical or basal end of root hair cells (Fischer et al., 2006). The establishment of polarity is followed by the action of a RhoGTPase GDP dissociation inhibitor, which restricts ROOT HAIR DEFECTIVE2 (RHD2) to the growth site (Carol et al., 2005; Takeda et al., 2008). RHD2 produces reactive oxygen species (ROS), which trigger Ca2+ influx, which, in turn, stimulate RHD2 to increase ROS production (Takeda et al., 2008). This positive feedback may maintain polarity in the growing tip. ROS production may also be involved in pollen tube initiation (Speranza et al., 2012). The yield threshold of the wall at the growth site must be lowered to allow growth at this site rather than all over the cell. Wall pH drops at the site of hair initiation, and blocking the pH drop with buffer reversibly halts hair initiation and hair growth (Bibikova et al., 1998). This pH change may increase expansin activity at the growth site, and two expansins are expressed specifically in root hair cells immediately before the onset of root hair growth (Cho & Cosgrove, 2002).
Once a growing site has been specified and polarity established, cell wall material must be deposited at the site of growth. It is likely that the growing tip of cells contains one or more specialized membrane domains that regulate the targeted deposition of vesicles, as both sterol and phosphoinositide signalling are required for normal root hair initiation and growth (Ovečka et al., 2010; Boss & Im, 2012). In pollen tubes, the direction of the growing tip largely depends on a highly dynamic network of F-actin mediating exo- and endocytosis for the deposition of plasma membrane and cell wall components and for the recycling of other components (Lee & Yang, 2008). The rate of deposition must be regulated to prevent wall rupture. As mentioned above, a recent model postulates that the stretch-activated calcium channels at pollen tube tips regulate wall material deposition.
3. Duration of growth and endocycling
In numerous plant tissues, a rapid cell volume increase as a result of diffuse and tip growth is correlated with endocycling, also known as endoreduplication or endoreplication (Breuer et al., 2010; De Veylder et al., 2011). Endocycling is DNA replication in the absence of cytokinesis, leading to an increase in cell ploidy. Although various studies have revealed a positive correlation between cell size and ploidy level, such as in the A. thaliana leaf epidermis (Melaragno et al., 1993), this concept is challenged by other plant studies, strongly suggesting the existence of ploidy-independent growth mechnisms (Beemster et al., 2002; Schnittger et al., 2003). The trihelix protein GTL1 was found to repress cell growth during a late stage of trichome growth, through transcriptional repression of CELL CYCLE SWITCH PROTEIN 52 A1 (CCS52A1), an anaphase-promoting complex/cyclosome (APC/C) activator (Breuer et al., 2012). This suggests that the inhibition of endocycle progression is involved in halting cell growth in trichomes. Interestingly, the same study shows genetic evidence that cell growth regulation by GTL1 also involves a ploidy-independent pathway, but the molecular mechanism still requires further elucidation. In addition, a cell cycle regulator that promotes endocycling, KIP-RELATED PROTEIN5 (KRP5), has been linked to wall modification (Jégu et al., 2013). Further work should aim to demonstrate more functional links between endocycle progression and cell growth.
VII. Regulation of plant cell growth
The astonishing ability of plants to shape themselves to their environment requires complex regulatory systems that integrate endogenous and exogenous factors dynamically to control cell growth appropriately. This sensitive regulation is necessary because plants are trapped within their local environment. Plant development is so plastic that it may be useful to view it as the response of a complex network to endogenous and exogenous inputs of comparable importance, rather than the result of environmental perturbations to an underlying developmental programme. Recent genetic studies have begun to uncover some of the regulators involved in these upstream signalling cascades, but how they actually modify cell growth is far from clear. The network that controls growth will probably do so by modulating gene transcription and altering enzymatic activity and protein stability. Potential targets for this network include cytoskeletal components, enzymes that synthesize and modify cell wall components, such as CESAs and PMEs, and enzymes that modify cell wall properties, such as expansins. In this section, we summarize our current understanding on how diffuse growth and tip growth are regulated by various environmental cues.
1. Regulation of diffuse growth
We examine inputs, signal integration and outputs in a growing A. thaliana hypocotyl, which serves as an excellent model system to study how various environmental stimuli modulate diffuse growth (Fig. 3), as hypocotyl growth is predominantly based on cell elongation, not proliferation (Gendreau et al., 1997). Understanding of this model system has advanced rapidly over recent years, especially in the search for signalling hubs at which diverse inputs converge.
The shade avoidance response is a crucial feature of hypocotyl growth, and ensures that energy is not expended on cotyledon expansion until they reach a favourable light environment. Hypocotyls in favourable light conditions become de-etiolated, hypocotyl growth is inhibited and the cotyledons expand. In poor light conditions, rapid hypocotyl growth is promoted, cotyledons remain closed and an apical hook is present. This response is known as skotomorphogenesis or etiolation. Perception of the light environment occurs via multiple mechanisms, including phytochromes (Casal, 2013). A low red : far-red (R : FR) ratio indicates shade and produces low levels of active phytochromes, the far-red-absorbing Pfr. Active Pfr proteins from the PhyB clade of phytochromes are known to interact with PHYTOCHROME INTERACTING FACTORs (PIFs), resulting in phosphorylation and degradation of PIFs (reviewed in Chen & Chory, 2011). PIFs are basic helix–loop–helix (bHLH) transcription factors that promote cell elongation by indirectly (stimulating auxin production) and directly modulating growth-related genes. PIFs have a large overlap of target loci, and partial redundancy is common (Zhang et al., 2013). In high R : FR, such as in white light, phytochromes undergo a conformational change from the red-light-absorbing inactive Pr form into the active Pfr form, and repress PIF activity, resulting in decreased cell elongation and photomorphogenesis.
The circadian clock also regulates hypocotyl growth. Proteins involved in the maintenance of periodicity are known to form a complex, involving EARLY FLOWERING3 (ELF3), ELF4 and LUX ARRHYTHMO (LUX), which regulates the expression of PIF4 and PIF5 in the early evening (Nusinow et al., 2011). The hypocotyl growth rate varies through the day, peaking just before dawn; normal variation requires both light receptors and a functional circadian clock (Breton & Kay, 2007; Nozue et al., 2007). This system is known as the double coincidence mechanism, and is integrated with temperature and hormone signalling (Nomoto et al., 2012a,b). Metabolic status is also likely to regulate growth as it determines how much growth the plant can afford and what type of growth will be beneficial. Recently, sucrose has been found to promote hypocotyl growth and auxin levels in a partially PIF-dependent fashion (Stewart et al., 2011; Lilley et al., 2012).
Moderately high temperatures (c. 29°C) also promote hypocotyl growth (Gray et al., 1998). Recently, it has been discovered that temperature changes can be sensed by a histone variant. H2A.Z-containing nucleosomes bind DNA more tightly than typical H2A-containing nucleosomes, but H2A.Z nucleosome occupancy declines with increasing temperature (Kumar & Wigge, 2010). This increases the accessibility of DNA to transcription factors and the transcriptional machinery at higher temperatures. This process provides a direct functional link between temperature change and a transcriptional response. Interestingly, the pif4 mutation abolishes the temperature-dependent hypocotyl growth response. PIF4 expression is induced by high temperatures (Koini et al., 2009) and enhanced in arp6 mutants that lack H2A.Z incorporation (Kumar et al., 2012). This reflects direct or indirect control of PIF4 expression by H2A.Z temperature sensing. PIF4 activity is also regulated by the temperature-dependent availability of target loci located near H2A.Z nucleosomes (Kumar et al., 2012).
We have seen that the activity of PIF4 is modulated by the circadian clock, metabolic status, temperature and gibberellin (GA) levels in addition to light (reviewed in Proveniers & van Zanten, 2013). Recent research has suggested that PIF proteins form a regulatory module with hormone response pathways, thereby allowing cell growth to respond to and balance an even wider set of inputs.
DELLA proteins are nuclear-localized growth repressors (DELLA and GA signalling reviewed in Achard & Genschik, 2009). DELLAs are known to bind and sequester PIFs, preventing them from binding DNA. GA promotes growth by promoting DELLA degradation, and its levels are affected by environmental conditions, including temperature, light and salt stress (Achard & Genschik, 2009). In addition to binding PIFs, DELLAs also bind a key component of the brassinosteroid (BR) signalling pathway, the transcription factor BRASSINOZALE-RESISTANT1 (BZR1). The DELLA protein REPRESSOR OF GA1 (RGA1) physically interacts with BZR1 and thereby prevents its binding of transcriptional targets (Bai et al., 2012b). GA-stimulated hypocotyl elongation requires both PIFs and BR signalling, and this interaction is supported by the observation that 92% of genes differentially expressed in both a BR-insensitive and GA-deficient mutant were affected in the same way (Bai et al., 2012b).
A multiple PIF mutant, pifq, was found to show reduced hypocotyl elongation in response to BR, and PIF4 overexpression partially rescued this phenotype. However, the null mutation of pif4 showed a similar BR response to the wild-type, suggesting redundancy amongst PIFs in their interaction with BR signalling. BZR1 was found to interact with PIF4 in vitro and in vivo, and chromatin immunoprecipitation (ChIP) studies showed that target genes co-occupied the promoters of transcriptional targets in vivo. Chip-seq of BZR1 and PIF4 revealed a large overlap in targets, and that 80% of common targets were also regulated by light, whereas RNA-seq showed that BZR1 and PIF4 regulated a large number of target genes interdependently (Oh et al., 2012). Together, these data strongly suggest a central DELLA–BZR1–PIF4 module that integrates a diverse range of environmental inputs to modulate hypocotyl cell elongation appropriately (Wang et al., 2012; Casal, 2013). It will be fascinating to see how interaction affects the DNA binding and regulatory activity of BZR1 and PIF4. A candidate may be GATA2, which is bound directly by BZR1 and regulated by BR, but binding is much higher in dark-grown (when PIF4 levels are highest) than in light-grown seedlings (Luo et al., 2010). A ChIP of PIF4 in wild-type and bzr1 mutant plants would reveal how this interaction affects PIF4 DNA binding globally.
The power of signal integration on cell growth responses in this system can be demonstrated through the addition of ethylene. This hormone suppresses hypocotyl elongation in darkness and promotes it in the light, demonstrating how one stimulus may alter and even reverse the effects of another stimulus on growth. The ethylene-responsive transcription factor, ETHYLENE-INSENSITIVE3 (EIN3), activates PIF3 expression directly, as well as inducing the expression of ETHYLENE RESPONSE FACTOR1 (ERF1). PIF3 is required for ethylene-induced hypocotyl growth, whereas ERF1 represses it (Zhong et al., 2012). ERF1 is more stable in the light than in the dark, and PIF3 is more stable in the dark than in the light (as a result of phytochromes). Zhong et al. (2012) predicted that the induction of the less stable protein will have a larger effect on growth. For instance, during the day, ERF1 levels are high and PIF3 levels are low. The induction of both by ethylene has a growth-promoting effect because of a large relative increase in PIF3 levels. Another possibility is that the activity of an interaction factor for PIF3 and/or ERF1 varies depending on light conditions.
PIF4 plays a crucial role in the regulation of hypocotyl cell elongation, partly as a result of its promotion of auxin biosynthesis. PIF4 binds the promoters of several auxin biosynthesis genes directly, including YUCCA8 (YUC8) (Sun et al., 2012). Genetic evidence supports this interaction, as YUC8 expression is stimulated by temperature increase, but this increase is absent in pif4 mutants. PIF7 also binds and activates YUC8 and YUC9 promoters directly (Li L et al., 2012).
The gain-of-function mutant indole-3-acetic acid19/massugu2 (iaa19/msg2) shows reduced auxin sensitivity, and does not display enhanced hypocotyl elongation in response to high temperature (Maharjan & Choe, 2011). Furthermore, the gain-of-function short hypocotyl2 shy2/iaa3 mutant partially suppresses the hypocotyl elongation caused by PIF4 overexpression (Sun et al., 2012). This suggests that auxin response is required for temperature-dependent hypocotyl elongation. PIF4 is required for temperature-dependent induction of the SMALL AUXIN UP RNA genes SAUR19–24, and SAUR19 overexpression can complement temperature-dependent hypocotyl elongation in pif4 mutants (Franklin et al., 2011). In addition to the indirect promotion of elongation through an alteration in hormone levels, PIFs may also modulate growth by direct alteration of the transcription of genes involved in cell growth. A gene ontology (GO) analysis of putative PIF-regulated genes obtained by combining ChIP-seq data from PIF4 with microarrays for multiple pif mutants shows significant enrichment for transcriptional regulation and regulation of cell size (Oh et al., 2012).
The PACLOBUTRAZOL-RESISTANT (PRE) family of HLH transcription factors is a key output of the DELLA–BZR1–PIF gene regulatory network (GRN). PRE1, PRE5 and PRE6 are direct targets of BZR1 and PIF4, and an artificial microRNA (amiRNA) multiple knockout of PRE1, PRE2, PRE5 and PRE6 results in dwarfism, hypersensitivity to light and reduced sensitivity to BR, GA and high temperatures (Bai et al., 2012b; Oh et al., 2012). PREs are atypical HLH proteins that lack a known DNA binding domain, and promote cell elongation through antagonistic protein–protein interactions with growth-repressing proteins, such as ILI1 BINDING BHLH1 (IBH1) (Zhang et al., 2009). IBH1 represses cell elongation by binding and inhibiting the action of transcription factors that promote hypocotyl elongation, such as ACTIVATOR FOR CELL ELONGATION1 (ACE1) and HOMOLOG OF BEE2 INTERACTING WITH IBH1 (HBI) (Bai et al., 2012a; Ikeda et al., 2012). There is further evidence that PREs also bind and disrupt the activity of ATBS-INTERACTING FACTORS (AIFs) and LONG HYPOCOTYL IN FAR-RED1 (HFR1), which also repress cell elongation (Hyun & Lee, 2006; Wang et al., 2009).
CONSTITUTIVE PHOTOMORPHOGENIC1 (COP1) is a RING E3 ubiquitin ligase which represses photomorphogenesis. COP1 is essential for shade avoidance and is involved in a wide variety of light responses (reviewed in Lau & Deng, 2012). COP1 activity is supressed by activated phytochromes. COP1 therefore has high activity in the dark and poor light. Photomorphogenesis-promoting proteins, and phytochromes, are ubiquitinated by COP1 (as part of a multi-subunit complex), which triggers their degradation by the proteasome (Jang et al., 2010). COP1 also targets light signalling transcription factors that promote photomorphogenesis, including ELONGATED HYPOCOTYL5 (HY5), HFR1 and HY5-HOMOLOG (HYH).
PICKLE (PKL/EPP1), an ATP-dependent chromatin remodelling factor, represses photomorphogenesis and promotes hypocotyl cell elongation. PKL antagonizes the repressive activity of HY5 by reducing H3K27me3 repressive marks at multiple loci in a light-dependent manner (Jing et al., 2013). PKL expression is repressed by red, far-red and blue light by unknown mechanisms involving PhyB, PhyA and cryptochromes, respectively (Jing et al., 2013). pkl mutants have shorter hypocotyls and more expanded cotyledons than the wild-type. Single and double pkl and hy5 mutants were found to be antagonistic; target genes repressed in pkl were upregulated in hy5, and the pkl hy5 mutant had intermediate expression levels and phenotype (Jing et al., 2013). Two target loci mutants, exp2 and dwf4, also show short hypocotyls when grown in light (Jing et al., 2013).
2. Regulation of tip growth
We examine both A. thaliana root hairs and pollen tubes as examples of how tip growth is regulated by complex networks, which integrate multiple diverse inputs. Arabidopsis thaliana has type III root hair patterning, in which hair and non-hair cells are present in separate cell files. The underlying cortex provides positional information, releasing an unknown signal that causes epidermis cells overlapping two cortex cells to become hair cells. Epidermal cells not present over cortex cell junctions become non-hair cells. The GRN that converts the positional cue into differentiated cell files has been studied extensively (Grebe, 2012; Ryu et al., 2013).
Hair-forming cells express ROOT HAIR DEFECTIVE6 (RHD6) and RHD6-LIKE1 (RSL1), bHLH transcription factors that promote root hair outgrowth in a partially redundant manner. Root hair growth is almost abolished in rhd6 mutants. Patterning genes, such as CAPRICE (CPC), WEREWOLF (WER), TRANSPARENT TESTA GLABRA (TTG) and GLABRA2 (GL2), are epistatic to RHD6 function, and CPC promotes RHD6 expression, whereas WER, TTG and GL2 repress it (Menand et al., 2007). This suggests that the patterning network controls the expression domain of RHD6. Interestingly, the two genes most similar to RHD6 in the moss Physcomitrella patens are required for the production of analogous rooting structures in the haploid gametophyte (Menand et al., 2007), demonstrating that regulatory networks controlling cell growth may be conserved and co-opted across land plant phylogeny. The bHLH transcription factor RSL4, a direct target of RHD6, is necessary and sufficient for root hair growth. A microarray analysis identified c. 80 genes up-regulated by RSL4, 20% of which were exclusively expressed in root hair cells. Five of these genes have a demonstrated role in root hair growth, including a phosphatidylinositol transporter and extensins (Yi et al., 2010; Velasquez et al., 2011). Extensins are structural cell wall glycoproteins that self-assemble and may guide pectin positioning (Lamport et al., 2011).
Root hair growth is altered by a wide variety of nutrients (Fig. 4), such as phosphate, iron, zinc and manganese (Péret et al., 2011). Phosphate has a large effect on A. thaliana root hair development; root hair length is inversely correlated with phosphate concentration (Ma et al., 2001). A recently discovered class of plant hormones, strigolactones, promotes root hair growth in low phosphate conditions. This strigolactone-mediated response appears to be partially dependent on both auxin and ethylene signalling (Kapulnik et al., 2011a,b). RSL4 integrates nutritional and hormone responses as its transcription is activated by both low phosphate availability and auxin, leading to the promotion of root hair elongation (Yi et al., 2010). This induction appears to be independent of RHD6 and RSL1, suggesting that these environmental and developmental signals converse at the level of RSL4 transcription (Masucci & Schiefelbein, 1996).
Ethylene promotes root hair elongation (Pitts et al., 1998; Cho & Cosgrove, 2002) in response to both potassium and iron deficiency (Romera & Alcántara, 2004; Jung et al., 2009). Ethylene-stimulated root hair elongation requires auxin transport and response genes (Strader et al., 2010), suggesting that ethylene acts in a common pathway with auxin. Cytokinin (CK) inhibits hair growth in potassium-limited plants, and CK levels are reduced in low potassium conditions (Nam et al., 2012). However, CK has been reported to increase root hair length when applied exogenously (An et al., 2012).
CK and ethylene signalling is integrated via the action of ZINC-FINGER PROTEIN5 (ZFP5), which directly promotes CPC expression and enhances root hair elongation (An et al., 2012). CPC overexpression alone has not been observed to increase root hair length (Wada et al., 1997), suggesting the requirement of additional factors to enhance root hair elongation. zfp5 mutants have fewer and shorter root hairs as a result of a lower root hair growth rate. Exogenous application of an ethylene precursor and a synthetic CK could increase root hair length in wild-type but not zfp5 plants (An et al., 2012). CK treatment induces ZFP5 transcription, whereas ethylene increases ZFP5 protein levels through protein stabilization (An et al., 2012), demonstrating that CK and ethylene act synergistically to promote ZFP5 activity.
Pollen germination is followed by the formation and maintenance of an apical growth domain with highly polarized growth. Pollen tube growth must be tightly regulated for successful delivery of the sperm cells for egg cell fertilization and, to do so, the male gametophyte must communicate with multiple female tissues. On germination at the stigma, pollen tube growth is successively guided to the female gametophyte in a multi-step process. In A. thaliana, initial growth from the stigma to the style is mainly mediated by chemo-attractants (Higashiyama & Hamamura, 2008). Thereafter, pollen tube growth is mechanically guided through the transmitting tract to reach the ovary (Crawford et al., 2007), although a recent study has suggested that cysteine-rich peptides are also involved (Chae et al., 2009). The last step of pollen tube growth involves passage through the septum, along the funiculus and towards the micropylar end of the ovule. Here, the path of the pollen tube is guided by chemo-attractants from the female gametophyte, such as peptides and amino acids, which trigger receptor-mediated signalling cascades, ion- and cyclic nucleotide-gated channels at the growth tip of the pollen tube (reviewed by Takeuchi & Higashiyama, 2011). Despite these recent advances in understanding the nature of signalling molecules and the signalling responses involved, it remains unclear how directional changes in chemo-attractant gradients cause the rearrangement of structural components, vesicular flows and cytoskeleton dynamics to shift the apical growth domain and alter the direction of pollen tube growth.
Plant cells convert turgor pressure into growth using tightly regulated deposition and modification of cell wall components. Complex regulatory networks integrate myriad endogenous and exogenous inputs to produce appropriate responses, which vary across cell types and species. The understanding of these regulatory systems has accelerated rapidly in the past decade, and numerous cell growth models have now been characterized in impressive detail. The links between these systems and the process of growth remain unclear, but growth regulators are regularly linked to large lists of potential targets using high-throughput techniques. Ultimately, these growth regulators must modify cell wall properties to modulate cell growth. To further our understanding, it is crucial that more functional links are made between regulators of growth and effectors of growth.
Understanding the process of growth is difficult as a result of the complexity of plant cell walls, and the incredible ability of plants to adapt to environmental insults (including mutations) through gene redundancy and compensation mechanisms. It is practically ludicrous that removing a polysaccharide that makes up around one-third of the wall by weight (Cosgrove & Jarvis, 2012) does not produce a major phenotype. A range of new technologies and techniques should help to overcome these difficulties. Redundancy can be combatted with chemical genetics, sensitized screens and targeted knockdowns of related genes using amiRNAs (reviewed in Sablok et al., 2011; Vidaurre & Bonetta, 2012), and lethality circumvented with inducible constructs. Visualization techniques for wall polysaccharides are improving, with increasingly specific dyes and polysaccharide labelling using click chemistry now available (reviewed in Gonneau et al., 2012; Wallace & Anderson, 2012). In addition, microscopy is developing rapidly, with variable angle epifluorescence microscopy (VAEM) able to obtain high-quality images of cytoskeletal dynamics and vesicle deposition at the plasma membrane (Konopka & Bednarek, 2008), and light-sheet-based fluorescence microscopy (LSFM), which will allow cellular growth to be tracked in three dimensions over long time periods at the tissue level (Maizel et al., 2011).
The combination of modelling, structural and mechanical cell wall studies, and genetic approaches, will be required to generate a full understanding of plant cell growth. Modelling approaches have effectively utilized terms representing structures and processes suggested by experimental approaches (Kroeger et al., 2011; Dyson et al., 2012), illustrating that communication between these disciplines can be extremely beneficial. In summary, we are beginning to understand growth regulatory systems, or ‘why’ plant cells grow, but ‘how’ plant cells grow remains something of a mystery, as challenging and exciting as it was to Lockhart, 50 yr ago.
This work was supported by a grant from the Ministry of Education, Culture, Sports, Science and Technology (MEXT; No. 22119010) to K.S. We thank members of the Sugimoto laboratory for helpful discussions, and Bart Ryman, Momoko Ikeuchi and Sarah Ball for comments on the manuscript.