Pavement cells: a model system for non-transcriptional auxin signalling and crosstalks

Highlight

Investigating the mechanisms underlying the formation of the puzzle-piece cell shape of the Arabidopsis pavement cells reveals novel mechanisms for auxin signalling and crosstalks with cytokinin and mechanical stress.

Abstract

Auxin (indole acetic acid) is a multifunctional phytohormone controlling various developmental patterns, morphogenetic processes, and growth behaviours in plants. The transcription-based pathway activated by the nuclear TRANSPORT INHIBITOR RESISTANT 1/auxin-related F-box auxin receptors is well established, but the long-sought molecular mechanisms of non-transcriptional auxin signalling remained enigmatic until very recently. Along with the establishment of the Arabidopsis leaf epidermal pavement cell (PC) as an exciting and amenable model system in the past decade, we began to gain insight into non-transcriptional auxin signalling. The puzzle-piece shape of PCs forms from intercalated or interdigitated cell growth, requiring local intra- and inter-cellular coordination of lobe and indent formation. Precise coordination of this interdigitated pattern requires auxin and an extracellular auxin sensing system that activates plasma membrane-associated Rho GTPases from plants and subsequent downstream events regulating cytoskeletal reorganization and PIN polarization. Apart from auxin, mechanical stress and cytokinin have been shown to affect PC interdigitation, possibly by interacting with auxin signals. This review focuses upon signalling mechanisms for cell polarity formation in PCs, with an emphasis on non-transcriptional auxin signalling in polarized cell expansion and pattern formation and how different auxin pathways interplay with each other and with other signals.

Introduction

Multicellular organisms are typically composed of diverse cell types, each with particular functions that often depend on the acquisition of specific cell shapes. A poorly understood question is what are the developmental signals that determine morphogenesis of the various specialized cell types. Auxin is involved in an extraordinarily broad variety of basic cellular processes, such as endocytosis, cell cycle control, expansion, differentiation, and polarity formation (Mockaitis and Estelle, 2008; Chapman and Estelle, 2009; Sauer et al., 2013). At the developmental level, auxin is responsible for embryogenesis, morphogenesis, tissue patterning, de novo formation of organs, and light and gravity responses (Leyser 2006; Benjamins and Scheres, 2008; Vanneste and Friml, 2009). In roots and embryos, another important hormone, cytokinin, has been proposed to antagonize auxin action by activating the expression of AUX/IAA, transcriptional repressors and targets of the TIR1/AFB-mediated protein degradation for auxin responses (Dello et al., 2008; Müller and Sheen, 2008; Moubayidin et al., 2010).

In plants, the coordination of cell polarity along the plane of organ surface, known as planar cell polarity (PCP), is essential for development and pattern formation, e.g. auxin transporters are polarized to the same end of a cell in the entire file or field of cells, which allows auxin flow and gradient formation (Boutté et al., 2007; Vieten et al., 2007; Gao et al., 2008; Kleine-Vehn and Friml, 2008; Petrásek and Friml, 2009; Santos et al., 2010; Geisler et al., 2014). Morphogenesis of planar cells depends on extensive intercellular communication, and thus PCP establishment requires a mechanism for localized cell–cell coordination of cell polarity on the basis of cytoskeleton dynamics/reorganization and vesicular trafficking (Yang, 2008). How polarity is initiated along the plane of organ surface and how it is maintained remain fascinating questions.

In the last decade, important advances in the auxin–ROP signalling network for PCP and morphogenesis have been made by taking advantage of Arabidopsis thaliana leaf and cotyledon pavement cells (PCs), an interesting cell type that coordinately develops multiple polarities and undergoes intercalated growth to produce the puzzle-piece shapes with interdigitated lobes and indentations (Fig. 1A). PC interdigitation bears interesting resemblance to and shares underlying mechanisms with PCP-dependent processes, particularly cellular intercalation in convergent extension during embryogenesis in animals (Adler, 2002; Wallingford et al., 2002; Singh and Mlodzik, 2012; Sebbagh and Borg, 2014; Uriu et al., 2014). Several recent studies have uncovered an elaborate Rho GTPase-based signalling network underlying PC interdigitation, which includes intra- and intercellular interplays between two mutually exclusive Rho pathways and a positive feedback loop, which are all activated by auxin (Fu et al., 2005; Xu et al., 2010, 2014) (Fig. 1). These studies also reveal the first signalling pathways known to control the cytoskeleton in plants and a novel auxin-signalling mechanism downstream of the auxin binding protein 1 (ABP1) cell-surface auxin receptor and its functional partners transmembrane receptor-like kinases (Fu et al., 2005; Xu et al., 2010, 2014) (Fig. 1). Furthermore, the local auxin signal appears to be amplified by a positive feedback loop in which ROP2 promotes polarization of the PIN-FORMED 1 (PIN1) auxin exporter to the lobe tips (Xu et al., 2010; Nagawa et al., 2012) (Fig. 1). Here, we provide an overview of recent major progress towards an understanding of hormonal coordination and crosstalk, and molecular mechanisms underpinning planar morphogenesis and interdigitated patterning.

Fig. 1.

A working model for auxin control of the interdigitated pattern of plant PCs. (A) PCs highlighted to show lobes and indentations. The additional white arrows indicate lobes. (B) Schematic view of the coordination of two mutually exclusive ROP signalling pathways by both auxin and cytokinin signalling on the basis of known components, including ABP1, TMK, ROP2, ROP6, PIN1, RIC4, RIC1, KTN1, CRE1/AHK3, and ARR20. (C) A model for auxin- and cytokinin-based integrated control of interdigitated pattern through both inter- and intracellular coordination of the ROP2 and ROP6 pathways. On one hand, interdigitated growth of PCs is regulated by an auxin-dependent self-organizing mechanism. Auxin activates the ROP2 pathway via the cell surface ABP1–TMK auxin receptor complex and subsequent downstream RopGEF. Both the ROP2–RIC4 pathway and PID/PP2A-mediated phosphorylation switch polarizes PIN1 to nascent lobe tips, where it exports auxin to locally amplify extracellular auxin through the ROP2–PIN1 positive feedback loop. PIN1-exported auxin at the lobe sites can diffuse across the cell wall to coordinately activate ROP6–RIC1 at the complementary indenting site through another ABP1–TMK auxin receptor complex. Mutual inhibition of the ROP2 and ROP6 pathways maintains ROP polarity and promotes lobe protrusion and neck indentation. On the other hand, ARR20, the positive regulator of cytokinin signalling, acts upstream of ROP2 to suppress the formation of the interdigitated pattern.

The pavement cell model: a paradigm for planar cell polarity

Cell shape formation coordinated by cellular differentiation and behaviour between and within cell layers is important for proper tissue and organ development and morphogenesis in multicellular organisms. The acquisition of cell morphology depends on the initiation and coordination of specific signalling modulators that lead to cellular and intracellular asymmetries, which are defined as cell polarity. The establishment of polarity is initiated by a spatial cue marking the site of polarity, which in turn directs the concerted actions of polarity proteins. Typically, conserved Rho-family GTPases respond to extrinsic or intrinsic polarity cues, and consequently induce downstream signalling to trigger asymmetric organization of the cytoskeleton (Yang, 2008; Bloch and Yalovsky, 2013).

Coordinate establishment of cell polarity along the surface plane of a tissue layer, referred to as PCP, is required for pattern formation. Typically, PCP provides a directional mechanism that coordinates the uniform cellular organization and integration within the tissue plane (Goodrich and Strutt, 2011). It is important for dynamic cellular rearrangements and tissue formation throughout the life cycle of an organism (Sebbagh and Borg, 2014). PCP is intensively studied in animals, where it was originally found to regulate wing bristle orientation and eye development and was later recognized to have a broad impact on development and morphogenesis, such as convergent extension wherein cells become intercalated to change the shape of early embryos during gastrulation and neurulation (Adler, 2002; Yang, 2008; Singh and Mlodzik, 2012; Sebbagh and Borg, 2014). In humans, failure in convergent extension causes a common developmental disorder—neural tube defect (Lei et al., 2010). As such, accumulating evidence shows that defects in human core PCP members are closely associated with various developmental malformations, such as spina bifida (Kibar et al., 2007) and cleft palate (Yang et al., 2014), or heterotaxia and cystic kidney diseases (Otto et al., 2003). In plants, coordination of cell polarity along the plane of the plant surface is fundamental to development and pattern formation in most epidermal cells, such as trichoblasts in roots and PCs in leaves. The Arabidopsis root epidermis exhibits a uniform polarization of trichoblasts and root hairs along the apical–basal root axis within the plane of the tissue layer, from which the polarizing cue is provided by the auxin gradient in the root tip (Fischer et al., 2006, 2007; Petersson et al., 2009; Brunoud et al., 2012). The formation of auxin gradients and auxin flow, which are critical for pattern formation and morphogenesis in plants, requires the coordinate polarization of PIN auxin carriers (reviewed in Petrásek and Friml, 2009; Yang, 2008).

Leaf epidermal PCs serve as an excellent model to dissect the mechanism for localized cell–cell communication during planar polarity establishment (Smith, 2003; Fu et al., 2002, 2005, 2009; Settleman, 2005; Xu et al., 2010, 2014; Zhang C et al., 2011a ; Nagawa et al., 2012; Lin et al., 2013; Sampathkumar et al., 2014b ). Planar morphogenesis of PCs gives rise to a pattern with alternations in lobes and indentations between neighbouring cells, providing the layer of PCs with an interdigitated jigsaw-puzzle appearance (Fig. 1A). The development of PCs is initiated from nearly square cells elongating preferentially along the long axis of the leaf, which leads to the formation of nearly rectangular cells. These polygonal cells further laterally expand via diffuse isotropic growth, and produce multiple shallow lobes alternating with indentations or necks. As early lobes expand, reiterative outgrowing and indenting continue, resulting in fully extended lobes and highly wavy cell outlines (Fu et al., 2002, 2005). Such precise fitting of interdigitation requires intricate and dynamic polarity formation for cell-to-cell signalling during the spatiotemporal coordination of lobe outgrowth from one cell with local growth inhibition of the neighbouring cell, the pattern and underlying mechanisms of which bear similarity with the convergent extension requiring cell-to-cell intercalation for animal PCP (Goto et al., 2005; Klein and Mlodzik, 2005; Settleman, 2005; Price et al., 2006; Xu et al., 2010, 2014; Sebbagh and Borg, 2014). Comprehensive exploration of PC polarity formation is thereby taken as a physiological and mechanistic paradigm of PCP in other systems.

Increasing evidence indicates that the polarization of the regular undulating pattern of PCs is tightly correlated with the cytoskeleton through the dynamic coordination between microtubules and microfilaments (Fu et al., 2002, 2005, 2009; Smith, 2003; Settleman, 2005; Yang 2008; Xu et al., 2010; Zhang C et al., 2011a; Ivakov and Persson, 2013, Lin et al., 2013, 2014). In diffusely growing PCs, well-ordered cortical microtubules are organized into parallel bundles, which are transverse to the elongation axis. These highly ordered bundles of microtubules restrict expansion of PCs and then generate an indentation, presumably through controlling the deposition of cellulose microfibrils (Fu et al., 2005, 2009; Ehrhardt and Shaw, 2006; Lin et al., 2013; Lindeboom et al., 2013; Zhang C et al., 2011a ; Zhang et al., 2013). In contrast, ARP2/3-independent cortical fine microfilament are associated with the initiating lobe and outgrowth sites of PCs where lacking well-ordered cortical microtubules (Frank and Smith, 2002; Fu et al., 2002, 2005). The mode of action for this form of actin is believed to promote local growth, possibly through the stimulation of vesicular trafficking and the preferential targeting of growth-promoting materials to new plasma membrane (PM) regions. Overall, these two cytoskeleton elements are thought to be essential to establish multipolar growth for the pattern formation of PCs. Because the initiation of intercalary growth in PCs is spatio-temporally regulated during leaf development, the picture of the polarizing cue that triggers cytoskeleton-based interdigitated growth and the mechanism behind it is emerging, but far from clear.

Auxin: a small molecule initiating interdigitated growth of leaf pavement cells

An instructive role in the coordination of vascular tissue and of planar epidermal polarity has classically been attributed to the small molecule phytohormone auxin (Grebe, 2004; Fischer et al., 2006, 2007; Sauer et al., 2006; Ikeda et al., 2009; Nakamura et al., 2012; Xu et al., 2010, 2014); for example, in the root epidermis of the model plant A. thaliana, auxin can orchestrate planar polarization by providing vectorial information for the polar positioning of hairs close to one end of the cell (Fischer et al., 2006). It is proposed that the outgrowth of a lobe in one cell must be coordinated with the ingrowth of indentation in the complementary cell. Such precise coordination hints at an intriguing existence of intercellular signalling between adjacent PCs. Auxin is again an ideal candidate by behaving as a mobile morphogenetic signal.

Xu and colleagues recently revealed that auxin indeed acts as the polarizing signal modulating the intercalary growth of leaf PCs; the application of exogenous auxin notably promotes lobing of PCs in a dose-dependent manner, whereas mutating auxin biosynthesis genes reduces interdigitated growth by decreasing the lobe numbers—this defect can be rescued by supplying external auxin. These findings provide the first indication that auxin is capable of inducing the interdigitated pattern that possibly accounts for its involvement in the establishment of polarity in PCs (Xu et al., 2010, 2011). However, what the event is that induces the breaking of the symmetry between adjacent cells and whether auxin is the original polarizing cue remain unanswered. Given that the role for auxin as a polarizing signal is tightly correlated with the progression of the interdigitated growth, an exciting question that arises is how auxin, a diffusible small molecule, locally controls signalling to establish the cytoskeleton-dependent cell cortical regions that define lobe- or indentation-forming sites, which in turn initiates the interdigitation pattern.

Non-transcriptional ABP1–TMK auxin signalling locally controls interdigitated cell expansion by activating ROP signalling

Because auxin is essential to coordinate a great number of growth and developmental processes at different levels (Leyser 2006; Benjamins and Scheres, 2008; Mockaitis and Estelle, 2008; Vanneste and Friml, 2009), the mechanisms underpinning auxin perception and signalling transduction might be diverse. In the past two decades, one refined model on auxin signalling and action has emerged; it focuses on the paradigm that auxin regulates the expression of subsets of genes by acting as a molecular glue that stabilizes the interaction between TIR1/AFB proteins and the AUX/IAA transcriptional repressors (Dharmasiri et al., 2005; Kepinski and Leyser, 2005; Parry and Estelle, 2006; Mockaitis and Estelle, 2008). Different combinations of TIR1 and AUX/IAA proteins form co-receptor complexes with a wide range of auxin-binding affinities that might account for diverse and specific response outputs (Calderón Villalobos et al., 2012; Wang and Estelle, 2014). Although the TIR1/AFB pathway elicits a remarkable number of auxin-mediated cellular and developmental responses, some rapid cellular responses to auxin occurring within minutes or seconds, such as cell expansion and protein secretion, seem unlikely to be transcriptionally regulated (Yamagami et al., 2004; Paciorek et al. 2005; Badescu and Napier, 2006; Vanneste and Friml, 2009; Schenck et al., 2010; Shi and Yang, 2011). In this scenario, non-transcriptional signalling involving the plant-specific protein ABP1 is likely to be central to the regulation of auxin-mediated interdigitated cell expansion of PCs because ABP1 has been proposed to be another auxin receptor for some fast auxin responses associated with cell expansion, endocytosis, and the activity of PM ion channels (Jones et al., 1998; Chen et al., 2001; Badescu and Napier, 2006; Robert et al., 2010; Tromas et al., 2010).

ABP1 is a 22-kDa glycoprotein that was initially detected in Zea mays (maize) coleoptiles decades ago (Hertel et al., 1972) and was soon considered a candidate auxin receptor involved in a wide range of auxin-mediated processes, such as cell division and elongation and early auxin response on the PM (Tromas et al., 2010; Shi and Yang, 2011). ABP1, despite carrying a KDEL-endoplasmic reticulum (ER) retention motif, is partially secreted to the outer surface of the PM upon binding to a glycosylphosphatidylinisotol-anchored protein (Shimomura, 2006). Although most ABP1 molecules localize to the ER, the low amount of cell surface-associated ABP1 is proposed to sense extracellular auxin, which is required for the PM or cytoplasmic responses at a non-transcriptional level for the following reasons: first, ABP1-auxin binding occurs at the optimal pH of 5.5 and becomes minimal at neutral pH (Shimomura et al., 1986; Lobler et al., 1985). Second, the extracellular application of a C-terminal ABP1 peptide or ABP1 recombinant protein induces auxin-related hyperpolarization of tobacco protoplasts (Leblanc et al., 1999b ). Thirdly, the extracellular application of different ABP1 antibodies either blocks auxin-induced response or induces auxin-like responses (Leblanc et al., 1999a ; Venis et al., 1992). However, these early studies did not provide definite proof that ABP1 can act as a cell surface auxin receptor, because the cell surface ABP1 partner (or docking protein) and its downstream signalling events were unknown.

Recently, the molecular function of ABP1 and the existence of its associated players have been studied in leaf epidermal PCs by Xu and colleagues (Xu et al., 2010), who nicely addressed the question of how auxin locally controls the formation of the interdigitation pattern, and provided new insights into the mechanisms for PCP control and developmental patterning in general (Fig. 1). On the cell surface of the PCs, ABP1 was found to be required for the activation of PM-delimited ROP GTPases, the plant counterparts of the animal and fungal Rho-family GTPases that serve as the conserved and essential signal transducers during a large number of cellular events, including cytoskeletal organization, cell morphogenesis, and vesicle trafficking of polar auxin transport proteins (Yang 2008; Nagawa et al., 2010; Craddock et al., 2012; Chen and Friml, 2014). The rapid ABP1-dependant activation of ROP2 and ROP6 in the nanomole range of the NAA synthetic auxin that induces PC interdigitation, which increases more than 1-fold within 0.5min, further implies that the response is far too rapid to be regulated by transcription or translation and is distinct from the auxin activation of NtRac2/ROP that occurs at 5min and requires micromoles of auxin (Tao et al., 2002, 2005). Importantly, increasing studies show that the PM-associated ROPs are directly activated by the receptor-like kinase (RLK) family cell surface receptors through the direct interaction between the kinase domain of RLKs and ROP guanine nucleotide exchange factors (RopGEFs) that activate ROPs (reviewed in Miyawaki and Yang, 2014).

Binding of auxin to ABP1 may directly activate transmembrane signalling, which further activates ROPs to control interdigitated cell expansion given that this non-transcriptional event is associated with the outer surface of the PM. Because ABP1 lacks a transmembrane domain, a transmembrane docking protein such as RLK was thus expected to transmit auxin signals from ABP1 to cytoplasmic ROP signalling. Xu and colleagues further addressed this question by identifying the transmembrane kinase (TMK) subfamily of receptor-like kinases as the missing link (Xu et al., 2014; Chen and Yang, 2014) (Fig. 1). TMKs are good candidates for the co-receptors of ABP1 because Arabidopsis tmk mutants display auxin-related developmental defects similar to abp1 mutants in abnormally shaped leaf PCs, a defect that was not rescued by auxin application, suggesting that the genetic interaction between ABP1 and TMKs is involved in auxin-mediated PC interdigitation and other processes (Dai et al., 2013; Xu et al., 2014). Further immunoprecipitation (IP) studies confirmed that ABP1 is physically associated with membrane-localized TMK1 in an auxin dosage-dependent manner, which indicates that ABP1 associates with the TMK1 protein complex upon perceiving extracellular auxin (Xu et al., 2014). The abp1-5 mutation, which harbours a point mutation (H94Y) in the auxin-binding pocket of the ABP1 protein (Xu et al., 2010), remarkably impaired the association of ABP1 with TMK1 induced by auxin, implying the pivotal role of ABP1 in sensing auxin for the formation of the complex (Xu et al., 2014).

More interestingly, transient expression of TMK1 extracellular domain (TMK1-Ex) and ABP1 or abp1-5 followed by co-IP assay in Nicotiana benthamiana further proves that the interaction between ABP1 and TMK proteins is likely to be direct because both ABP1 and TMK1-Ex were overexpressed. Importantly, this co-IP assay also demonstrated that ABP1 and TMK interactions are independent, thus most likely upstream, of the activated TMK signalling pathway(s), because the extracellular domain of TMK1 was used for co-IP. The abp1-5 mutant protein failed to associate with TMK1-Ex in an auxin-dependent manner, further confirming the requirement of auxin binding to ABP1 in inducing the formation of the ABP1–TMK complex.

The ABP1–TMK interaction is also pH-dependent, occurring only at low pH (5.7, as in W5 buffer used in protoplast preparation before co-IP), but not at neutral pH (unpublished data, Tongda Xu). This is consistent with the acid environment of the apoplast and not the ER, and further supports the cell surface-based ABP1–TMK signalling. It is also interesting to note that ABP1–auxin binding is also pH-dependent as mentioned above. It is yet to be determined whether neutral pH disrupts ABP1–TMK interaction directly or indirectly, by lowering ABP1–auxin affinity.

Moreover, auxin-dependent activation of ROP2 and ROP6 in PCs was impaired in the tmk quadruple mutant and TMK1 dominant negative mutant, as it was in the abp1-5 mutant, suggesting that TMKs, similarly to ABP1, are necessary for the activation of the ROP2 and ROP6 pathways (Xu et al., 2014). ROP2 localizes to the lobe tips of PCs, where the ROP2 effector Rop-interacting CRIB-containing protein 4 (RIC4) induces the accumulation of cortical F-actin required for outgrowth (Fu et al., 2002, 2005) (Fig. 1). On the opposing side, ROP6 preferentially localizes to the PM of the indenting zone and orchestrates the parallel alignment of transversal cortical microtubules through its effector RIC1 (Fu et al., 2005, 2009), which exerts its function through the microtubule severing protein katanin (KTN1) to promote detachment of branched microtubules (Lin et al., 2013) (Fig. 1). Interestingly, ABP1 is also involved in the reorientation of cortical microtubules induced by auxin that inhibits cell elongation in roots (Chen et al., 2014).

Thus, well-ordered microtubule bundles aligning between indentations inhibit isotropic growth to the lobe regions, where cortical F-actin promotes further outgrowth (Fu et al., 2005). The mechanisms underlying such precise fitting of interdigitations between adjacent PCs have also been studied. The PM-localized active ROP2 in the lobe regions inhibits RIC1 association with microtubules, causing well-ordered microtubule bundles to be excluded from outgrowing lobe tips (Fig. 1). At the same time, upon activation by ROP6, RIC1-mediated microtubule organization not only inhibits outgrowth, but also suppresses ROP2 activity in the indenting regions (Fu et al., 2005, 2009) (Fig. 1). The countersignalling of ROP2–RIC4 and ROP6–RIC1 pathways may lay the molecular foundations of the pattern of interlocking PCs.

Xu and colleagues found that the localization of GFP–RIC4 at the PM and the association of YFP–RIC1 with cortical microtubules, the in vivo indicators of active ROP2 and active ROP6, were abolished in the tmk quadruple mutant just as in the abp1-5 mutant, further linking TMKs and ABP1 to the activation of ROP2 and ROP6 in Arabidopsis PCs (Xu et al., 2010, 2014). Moreover, the visualization of TMK1 and ABP1 distribution confirms that both of them are targeted to the cell surface, where ROP2 and ROP6 are activated (Xu et al., 2014). Taken together, the ABP1–TMK auxin sensing system provides an intriguing and non-transcriptional auxin signalling mechanism for the differential activation of the opposing ROP2 and ROP6 pathways in interdigitated PC expansion.

Results described in a recent study by Gao et al. using two null abp1 alleles (Gao et al., 2015) appear to contradict all previous results from 40 years’ studies using a wide range of loss- or reduction-of-function and gain-of-function approaches (reviewed in Tromas et al., 2010; Shi and Yang, 2011; Sauer et al., 2013; Chen and Yang. 2014; Grones and Friml, 2015). Using a CRISPER-induced abp1 mutant (abp1-1c) and a T-DNA insertional mutant (abp1-TD), neither having a detectable ABP1 protein on analysis by western blotting using an anti-ABP1 serum, Gao et al. found that these mutants do not have obvious growth and developmental phenotypes (Gao et al., 2015). This is a sharp contrast to embryo lethality in previously reported T-DNA insertional mutants (Chen et al., 2001; Tzafrir et al. 2004; Meinke et al., 2008; Sassi et al. 2014) and a series of morphological, cell biological, and molecular phenotypes observed in other abp1 lines, such as abp1-5 and inducible anti-sense and antibody lines (Braun et al. 2008; Tromas et al., 2009; Robert et al., 2010; Xu et al., 2010; Chen et al., 2012; Thomas et al., 2013; Pague et al., 2014; Chen et al., 2014). In particular, Gao et al. did not observe obvious changes in frequency of lobing in 7 days after germination (DAG) cotyledons of abp1-1c mutant (Gao et al., 2015), whereas reduced lobing in 3–5 DAG cotyledons was found in the abp1-5 tilling allele, whose lobing defect was complemented by 35S::ABP1, as well as in both knockdown lines with inducible expression of ABP1 anti-sense or antibody lines, respectively (Xu et al., 2010). These differences need to be explained and resolved by further investigation. First of all, careful and thorough analyses of subtle morphological phenotypes under various growth conditions and biochemical and cell biological phenotypes (e.g. ROP activation, endocytosis, and microtubule organization) in the abp1 knockout lines from Gao et al. (2015) should be carried out to assess whether ABP1 is indeed not required for auxin signalling (Gao et al., 2015). Second, it remains possible though unlikely that second site mutations, which suppress the effect of the null abp1 mutations, exist in these lines. Genome re-sequencing of these lines may help to rule out this possibility. Third, it is also possible that the abp1-1c and abp1-TD lines described by Gao et al. have up-regulation of other genes that compensate for the loss of ABP1, and RNA-sequencing analyses of these lines at different developmental stages would be useful for assessing this possibility. It is not uncommon for knocking out a gene to induce up-regulation or ectopic expression of other genes functionally related or redundant to the knocked-out gene (Blilou et al., 2005; Vieten et al., 2005; Nimchuk et al., 2015). For instance, PIN1 is up-regulated and PIN2 is ectopically expressed in the pin3 pin4 pin7 knockout mutant (Blilou et al., 2005; Vieten et al., 2005). Additionally and/or alternatively, the original T-DNA insertional lines might contain second site mutations contributing to the embryo lethality phenotype (Chen et al., 2001; Tzafrir et al., 2004; Meinke et al., 2008; Sassi et al., 2014). Genome re-sequencing should be carried out to determine whether these lines contain other mutations. Furthermore, the abp1-5 allele might be a dominant negative mutation because the H59Y mutation occurs in the auxin-binding pocket and could presumably interfere with the wild-type copy of ABP1 by forming a poisoning dimer with wild-type ABP1 (Woo et al., 2002). Alternatively, this tilling allele could have a sensitizing background mutation. Both scenarios could explain why the abp1-5 PC phenotype was complemented by 35S::ABP1 (Xu et al., 2010). A pure abp1-5 line lacking any background mutations should be used for further analysis. Last but not least, the ABP1 antisense RNA and/or antibody might have off targets, which may or may not be functionally redundant to ABP1. Given all of these possibilities, the most likely explanation that can reconcile the phenotype differences between the alleles of Gao and colleagues and those from earlier studies is the existence of compensatory auxin receptors or pathways, which could be ABP1-like proteins that belong to the cupin family (Uberto and Moomaw, 2013) or other proteins that are unrelated to ABP1 in the primary amino acid sequence. In fact, some cupin member proteins have been shown to bind auxin (Ohmiya et al., 1998; Yin et al. 2009). The existence of compensatory systems is supported by the gain-of-function studies of ABP1 (Robert et al., 2010; Chen et al., 2012; Čovanová et al., 2013) and by the observation that the extracellular domain of TMK1 interacts with wild-type ABP1 protein in an auxin-dependent manner but not with abp1-5 mutant protein, while the tmk1 tmk4 double or tmk1 tmk2 tmk3 tmk4 quadruple mutants exhibit severe defects in various auxin responses (Dai et al., 2013; Xu et al., 2014). Again PCs will likely serve as a useful model system for uncovering such mechanisms that may compensate for ABP1.

ROP signalling- and phosphorylation-based PIN polarization also modulates pavement cell interdigitation

In multicellular plants, auxin exerts its functions through spatial distribution within tissues (Vanneste and Friml, 2009). Because auxin is proposed to coordinately activate ROP2 and ROP6 at the opposing sides of the cell wall of PCs, localized extracellular auxin is expected to activate these two mutually exclusive ROP signalling pathways at lobes and indentations, respectively. This hints at the existence of mechanisms for the generation and maintenance of the local accumulation of extracellular auxin. The concentration gradients of auxin are created by both local auxin biosynthesis, and by directional and intercellular auxin transport (Petrásek and Friml, 2009). Polar auxin transport largely depends on the asymmetrical distribution of PIN PM auxin efflux carriers. In the PC system, Xu and colleagues discovered that the PIN1 protein preferentially localizes to lobes rather than indentations (Xu et al., 2010, 2014). Further studies by Nagawa and colleagues showed that PIN1 endocytosis is inhibited in the lobe region (Nagawa et al., 2012). Interestingly, inhibition of endocytosis was largely abolished by the expression of DN-ROP2, or in the rop4 rop2i mutant background, but rescued by CA-ROP2 expression in rop4 rop2i. DN-ROP2-induced PIN1 endocytosis was also blocked by inhibitors of clathrin-dependent endocytosis, such as ANTH(C) and Ikarugamycin (Ika), indicating that the ROP2-activated signalling pathway inhibits PIN1 removal from the PM by inhibiting clathrin-dependent endocytosis. Further biochemical, pharmacological, and genetic data showed that active ROP2 modulated RIC4, which stabilizes F-actin at the lobes of PCs, and inhibits clathrin-dependent endocytosis of PIN1. Considering that extracellular auxin activates ROP2 through the ABP1–TMK signalling pathway, the above results suggest a self-organizing positive feedback loop of lobe-specific PIN1 localization, auxin transportation, ABP1–TMK-dependent ROP2 activation, and RIC4-dependent F-actin accumulation and inhibition of clathrin-dependent endocytosis.

This positive feedback and self-organizing system appears to be universal because it has also been demonstrated in the root system, where the polarity of PIN proteins was self-maintained by auxin-induced inhibition of clathrin-mediated endocytosis (Paciorek et al., 2005; Robert et al., 2010; Lin et al., 2012; Chen et al., 2012). Robert et al. showed ABP1 acts as a positive regulator of PIN1 endocytosis, by facilitating the recruitment of clathrin to the PM, but auxin negatively regulates this recruitment via its binding with ABP1. Overexpression of ABP1 lacking its ER-retention sequence significantly promoted PIN protein endocytosis, which was inhibited by the application of auxin. However, overexpression of abp1-5 (ABP1 with H94Y point mutation that impairs auxin-binding affinity) was resistant to auxin-induced inhibition of endocytosis. Further studies showed that ROP6 and RIC1 act downstream of ABP1 to inhibit PIN1 and PIN2 endocytosis (Chen et al., 2012). Interestingly, the auxin activation of the ROP6–RIC1 pathway inhibited PIN2 internalization also by promoting F-actin accumulation (Lin et al., 2012). Taken together, these findings strongly support the hypothesis that auxin signalling at the PM, through the localized activation of ROP signalling to the F-actin cytoskeleton, locally inhibits clathrin-dependent endocytosis of PIN proteins, leading to their local retention at the PM and consequently forming a positive feedback loop to polarize PIN proteins.

What is unclear and needs further studies is whether or not the initial targeting of newly synthesized PIN proteins is polarized. In a paper published in 2008, Dhonukshe and colleagues initially proposed a model for PIN1 polarization: the delivery of newly synthesized PIN1 to the PM is non-polar, and the differential endocytosis of PIN proteins and recycling establishes their polarity. Thus, in this model, interference with endocytosis by auxin or manipulation of the Rab5 GTPase pathway negatively regulates PIN polarity. However, this research article was recently retracted, and the authors stand by all the results except to state that the results on the non-polar delivery of nascent PIN proteins to the PM cannot be trusted because the cells appeared to be not viable (Dhonukshe et al., 2014). The investigation of ICR1, a ROP interactor and polarity scaffold protein, supports polarized recycling of PIN proteins (Lavy et al., 2007; Hazak et al., 2010). ICR1 shows polarized localization to the PM and is required for PIN protein polar localization. ICR1 may promote exocytosis, because it was shown to interact with SEC3, a conserved component of exocyst. Evidence suggests that ICR1 is critical for PIN1 recycling (Lavy et al., 2007; Hazak et al., 2010). Therefore, it is very tempting to propose that the positive feedback network of auxin-mediated ROP signalling may also regulate polar PIN recycling in addition to local inhibition of PIN endocytosis.

The similarity of the mechanisms for PIN polarization between PCs and other cell systems is also reflected in the phosphorylation-mediated PIN polarization. The Ser/Thr protein kinase PINOID (PID) and the protein phosphatase PP2A, which regulate the phosphorylation status of PIN proteins, are implicated in the regulation of the polar trafficking of PIN in the root and shoot apex (Friml et al., 2004; Michniewicz et al., 2007; Kleine-Vehn et al., 2009; Dhonukshe et al., 2010; Huang et al., 2010; Zhang et al., 2010). In wild-type roots, PIN proteins localize to the same ends of bipolar cells to trigger the formation of directional auxin flow and auxin gradient (Wisniewska et al., 2006). Such polar targeting strongly relies on the PIN phosphorylation status, where phosphorylation by a PINOID kinase promotes apical targeting, and de-phosphorylation by a PP2A phosphatase promotes basal localization in bipolar roots and shoot apices (Friml et al., 2004; Michniewicz et al., 2007; Kleine-Vehn et al., 2009; Huang et al., 2010). In the multipolar PCs, PIN polarization mediated by its phosphorylation switch is also conserved for the modulation of the interdigitated pattern of PIN1 localization and cell expansion. PID- and PP2A-dependent phosphorylation status determines the fate of PIN1 polarization between lobes and indentations (Fig. 1), as the action of the apical–basal switch in bipolar cells. Either knockout of the FYPP1 gene encoding a PP2A phosphatase or hyper-phosphorylation of PIN1 through the overexpression of PINOID resulted in the suppression of PC interdigitation and the localization of PIN1 to indentations instead of lobes (Li et al., 2011). Moreover, high levels of PIN1 promote lobing, as auxin treatment did, whereas lobe formation is compromised in the pin1 mutant (Xu et al., 2010; Li et al., 2011). This PID/PP2A-mediated switching of PIN1 localization displays a pattern similar to that observed in the process of the apical–basal switch of PIN polarization in other cells. For instance, overexpression of PINOID or the deletion mutant of PP2A regulatory subunit PP2AA led to PIN redistribution from the basal to the apical end of root cells as well, resulting in the loss of auxin gradients in seedling roots (Friml et al., 2004; Michniewicz et al., 2007). These findings provide a unifying mechanism underlying the PIN polarization mediated by its phosphorylation status in the manipulation of auxin-dependent plant pattern formation and architecture. This also raises an interesting question of whether ROP signalling and PID/PP2A-modulated PIN phosphorylation work in the same regulatory network or interplay with each other to regulate PIN polarization in bipolar cells or multipolar interdigitated PCs.

ABP1/TMKs may interplay with TIR1/AFB auxin signalling to regulate pavement cell morphogenesis

Since the breakthrough discovery of auxin signalling involving TIR1/AFB proteins as nuclear auxin receptors (Dharmasiri et al., 2005; Kepinski and Leyser, 2005; Parry and Estelle, 2006; Tan et al., 2007; Mockaitis and Estelle, 2008; Chapman and Estelle, 2009; Wang and Estelle, 2014), possible mechanisms by which ABP1 acts in concert with the TIR1/AFB receptor system to coordinately regulate various cellular auxin responses during cell expansion, division, and patterning have been proposed. It is well documented that TIR1/AFB- and its substrate AUX/IAA transcriptional repressor-based auxin signalling is crucial for auxin-related growth, development, and pattern formation and that they rely on changes in gene expression. TIR1/AFB proteins target the AUX/IAAs for degradation. After diffusing into the nucleus and being perceived by TIR1/AFBs, auxin acts as a molecular glue to stabilize the interaction between TIR1/AFBs and their substrate AUX/IAA proteins. AUX/IAA is thereby targeted for ubiquitination and subsequent degradation by the 26S proteasome, which in turn releases ARF transcription factors that are repressed in the presence of AUX/IAA proteins, and then activates diverse auxin-regulated promoters (Dharmasiri et al., 2005; Kepinski and Leyser, 2005; Parry and Estelle, 2006; Mockaitis and Estelle, 2008). The function of ABP1 has been proposed to regulate non-transcriptional rapid auxin responses at the PM or in the cytoplasm during cell division, expansion, vesicle trafficking, and microtubule organization (Jones and Herman, 1993; Oliver et al., 1995; Jones et al., 1998; Chen et al., 2001; Badescu and Napier, 2006; Tromas et al., 2009, 2010; Robert et al., 2010; Xu et al., 2010; Shi and Yang, 2011; Chen et al., 2014). However, the effect of conditional inactivation of ABP1 on auxin-regulated gene expression demonstrated that ABP1 is also implicated in the regulation of early auxin response genes, such as a subset of AUX/IAA in roots, at the transcriptional level (Braun et al., 2008; Tromas et al., 2009), and transcriptional regulation of early auxin-regulated genes was less sensitive to auxin in the heterozygous abp1/ABP1 mutant (Effendi et al., 2011). Additionally, ABP1 also affects the expression of a wide range of cell wall-related genes, particularly cell wall remodelling genes, primarily through the TIR1/AFB-dependent pathway during cell expansion (Paque et al., 2014). Recently, Tromas and colleagues proposed that ABP1 acts upstream of TIR1/AFBs to promote the stabilization of AUX/IAA repressors and thus maintain or restore transcriptional repression via negatively regulating the TIR1/AFB pathway, and that the effect of ABP1 on AUX/IAA stability is independent of its mode of action on clathrin-dependent endocytosis (Tromas et al., 2013). These findings suggest a crosstalk between ABP1 and TIR1/AFB pathways. It will be necessary to determine whether ABP1 acts at the PM or in the ER to regulate auxin-regulated gene expression. In either case, an ABP1-dependent auxin signal would be transmitted from one of these membrane domains to the nucleus for the regulation of gene expression, where TIR1/AFBs are thought to function (Shi and Yang, 2011).

Although the non-transcriptional ABP1–TMK receptor transmits the auxin signal to downstream cytoplasmic ROP pathways that facilitate the interdigitated patterning of PCs (Chen and Yang, 2014; Xu et al., 2014), it does not exclude the involvement of ABP1 in the regulation of TIR1/AFB-dependent auxin responses that might promote PC morphogenesis. No physical association of ABP1 with TIR1/AFBs or AUX/IAAs can reasonably be proposed because they are physically separated: ABP1 sits at the PM and in the ER, whereas TIR1/AFBs and AUX/IAA substrates are mainly located in the nucleus. Hence, ABP1 should act indirectly through a signalling mechanism to affect TIR1/AFB-dependent gene expression. One candidate adaptor for auxin-based membrane-to-nucleus events is the ROP/RAC GTPases, which have been shown to participate in the nuclear processes in addition to their cytoplasmic function; constitutively active or dominant negative forms of the tobacco NtRac1 ROP affect auxin-responsive gene expression in both tobacco and Arabidopsis (Tao et al., 2002). Moreover, Rac GTPases stimulate the formation of nuclear protein bodies that contain components of the TIR1–AUX/IAA ubiquitination pathway (Tao et al., 2005). There are multiple possible candidates to be the missing link between ABP1 and the TIR1/AFB-dependent auxin pathway that may modulate ABP1 action on downstream events, e.g. PC patterning. These include the dual specificity phosphatase IBR5 (Monroe-Augustus et al., 2003; Strader et al., 2008; Jayaweera et al., 2014), MAP kinases (Mockaitis and Howell, 2000; Lee JS et al., 2009), and phospholipases (Scherer, 2002; Scherer et al., 2007, 2012; Labusch et al., 2013), which have been reported to be involved in auxin signalling distinct from the TIR1/AFB pathway. Interestingly, the ibr5 mutant is also deficient in the interdigitation of leaf PCs (Jayaweera et al., 2014). Thus, how these regulators coordinate with each other to link ABP1 with TIR1/AFB signalling during PC morphogenesis remains to be determined.

Cytokinin signalling regulates pavement cell morphogenesis, possibly through antagonistic interaction with auxin signalling

It is known that auxin promotes the degradation of the AUX/IAA transcriptional repressors, causing a diverse array of auxin responses (Tan et al., 2007; Mockaitis and Estelle, 2008; Wang and Estelle, 2014). Conversely, cytokinin, another major phytohormone, is proposed to activate the expression of the auxin-signalling inhibitor AUX/IAA and consequently influence auxin distribution in roots (Dello et al., 2008). Since its identification as an adenine derivative promoting shoot growth of tobacco cells in the 1950s (Skoof and Miller, 1957), cytokinin has been implicated in modulating various aspects of developmental processes, including embryogenesis, seed germination, root architecture, vascular patterning, and leaf senescence (Aloni et al., 2006; Riefler et al., 2006; To and Kieber, 2008; Bishopp et al., 2011a ; Kushwah et al., 2011). The synergistic crosstalk between cytokinin and auxin coordinately regulates the function of the shoot apical meristem. Auxin plays a role in maintaining high cytokinin levels in the meristem (Zhao et al., 2010), whereas either cytokinin treatment or an A-type ARR cytokinin signalling regulator could induce PIN1 expression and consequently target auxin localization in incipient leaf primordia in the shoot apical meristem during organ initiation and outgrowth (Lee BH et al., 2009; Pernisová et al., 2009). However, more advances have widely dissected the antagonistic interaction between these two hormones, particularly in a developmental context, in which their crosstalk is also achieved preferentially through shared components (Moubayidin et al., 2009; Bishopp et al., 2011a ; Bielach et al., 2012; Vanstraelen and Benková, 2012; El-Showk et al., 2013). For instance, high cytokinin levels induce expression of cytokinin response transcription factors ARR1 and ARR12 genes, promote the transcription of SHY2/IAA3, and negatively regulate PIN expression, which in turn represses auxin signalling at the root tip beneath the quiescent centre (Dello et al., 2008; Moubayidin et al., 2010), while auxin activates certain cytokinin signalling repressors in a tissue-specific context, such as in the maintenance of the root apical meristem (Müller and Sheen, 2008; Bishopp et al., 2011b ). The antagonistic effects of cytokinin and auxin might be primarily attributed to the negative control of cytokinin on auxin signalling and transport, especially on the PIN-dependent auxin distribution. Such opposite roles have also been revealed in several developmental processes, including lateral root formation (Laplaze et al., 2007), de novo auxin-induced organogenesis (Pernisová et al., 2009), and leaf position determination (Lee BH et al., 2009; Shimizu-Sato et al., 2009). Hence, their antagonistic interaction occurs at multiple levels. An important point to be clarified is whether this mechanistic mode is also involved in determining the boundary of other cellular patterns.

Recent findings from Li and colleagues revealed that cytokinin signalling participates in the regulation of PC interdigitation pattern, possibly in a way that is also opposite to the action of auxin (Li et al., 2013) (Fig. 1). More specifically, cytokinin acts antagonistically to auxin and suppresses interdigitation. Mutants with suppressed cytokinin signalling and cytokinin receptor mutants, such as the ahk3 cre1 cytokinin receptor double mutant, exhibit a high level of interdigitation; over-activation of ARR20, a positive regulator of cytokinin signalling, delays or abolishes lobe formation. The mechanism underlying cytokinin signalling-induced negative control of interdigitation is proposed to again involve ROP signalling, because genetic and biochemical analyses imply that cytokinin signalling acts upstream of ROPs and counteracts the auxin-induced activation of ROP2 (Ivakov and Persson, 2013; Li et al., 2013) (Fig. 1). These findings uncover novel roles of cytokinin signalling in PCP and PC morphogenesis and of cytokinin–auxin crosstalk in the spatial regulation of fundamental cellular processes. It also raises the intriguing question of how cytokinin signalling antagonizes the action of auxin in PC morphogenesis. Li and colleagues found that exogenous auxin could rescue PC interdigitation defects induced by cytokinin in a dosage-dependent manner (Li et al., 2013). This suggests that the establishment and maintenance of the interdigitated pattern of PCs might require cellular homeostasis between auxin and cytokinin, although detailed mechanisms for outputs of the auxin–cytokinin balance may differ in different processes (Aloni et al., 2006; Müller and Sheen, 2008; Moubayidin et al., 2009; Ruzicka et al., 2009; Jones et al., 2010; Bishopp et al., 2011a ; Marhavý et al., 2011). It is likely that cytokinin suppresses PC interdigitation by counteracting auxin-induced activation of ROPs or by directly down-regulating ROPs activities, given that ROP2 activity is inhibited by ARR20, which is the positive regulator of cytokinin signalling (Li et al., 2013) (Fig. 1). Cytokinin is well known to affect the mRNA level of PIN by mediating the transcription of the genes involved in the TIR1/AFB-based auxin signalling (Dello et al., 2008; Müller and Sheen, 2008; Moubayidin et al., 2010); it is also thus possible that cytokinin-mediated PC morphogenesis involves the global control of auxin gradients in leaf through the regulation of cytokinin signalling in PIN expression and/or TIR1/AFB auxin signalling at the transcriptional level. In addition, cytokinin has also been revealed to regulate PIN polar trafficking directly in root morphogenesis (Marhavý et al., 2011; Zhang W et al., 2011b ). Most recently, Marhavý et al. found that by acting as a polarizing cue, cytokinin preferentially targets dephosphorylated PINs for degradation and consequently operates in developmental programmes involving rapid auxin transport redirection, such as lateral root organogenesis (Bishopp and Bennett, 2014; Marhavý et al., 2014). Because an auxin–ROP2-based positive feedback loop leads to polar PIN1 distribution to the lobe tips during PC interdigitation (Xu et al., 2010; Nagawa et al., 2012), we do not exclude the possibility that similar post-transcriptional control of PIN1 distribution and polarity in PCs might also account for cytokinin-mediated PC morphogenesis.

Mechanical stress feedback regulates pavement cell morphogenesis: a stressful link to auxin?

Increasing numbers of studies point to a role for mechanical stress in feedback regulation of cytoskeletal organization and PIN protein localization in the control of pattern and morphogenesis at both tissue and cellular levels (Sampathkumar et al. 2014a ; reviewed in Sampathkumar et al. 2014b ). An intriguing recent study that integrated computational modelling and experimental analyses provides strong evidence that this mechanical stress-mediated feedback regulation is important for PC morphogenesis (Sampathkumar et al. 2014a ). It was found that indenting regions of PCs bear the maximum tensile stress, which acts as an instructive mechanical cue to promote the alignment of cortical microtubules in the same regions. Cortical microtubules in these regions are thought to guide the deposition of cellulose, which will reinforce the tensile stress in the indenting regions. This stress–microtubule feedback loop is obviously important for the maintenance of the indentations in PCs. However, these findings also raise a number of important and interesting questions, particularly with regard to the connection between mechanical and auxin signals. First, how is the mechanical cue sensed and transduced to regulate the organization of cortical microtubules? The arrangement of cortical microtubules in the indenting regions is regulated by the ROP6–RIC1 pathway that is activated by a high level of auxin (Fu et al., 2002, 2005, 2009; Xu et al., 2010). The mechanical stress-activated microtubule organization requires katanin (Sampathkumar et al. 2014a ), which was shown to be directly activated by RIC1 downstream of ROP6 (Lin et al., 2013). Hence,it would be interesting to test whether the mechanical stress signal is linked to the auxin–ROP6–RIC1–microtubule pathway, and how this pathway feedback regulates to enhance mechanical stress.

Another intriguing question is whether the mechanical signalling is also important for the formation of the PC lobing pattern. As discussed above, evidence suggests that the lobing pattern is controlled by auxin and the PIN1 feedback loop. In the shoot apical meristem, the combined effect of two types of mechanical stress modulates the polarization of PIN1 proteins and auxin flow towards the forming leaf primordia that generate the essential auxin maximum (Sampathkumar et al., 2014b ). Because the complementary lobing and indenting regions share the same cell wall, it is conceivable that the indenting side may exert a pulling force on the lobing side, which might affect PIN polarization to the lobing side. In agreement with this possibility, the three-way junction of the PC walls tends to be associated with a lobe. Nonetheless, future studies should address the possibly much broader function of mechanical signalling in PC morphogenesis.

Future perspectives

Recent important advances made in the PC system shed light on the mechanisms by which auxin and cytokinin signalling controls the local coordination of cellular intercalation as well as providing a framework for our understanding of PCP control and developmental patterning in general. Although the ABP1–TMK auxin-sensing complex and auxin activation of downstream ROPs have been documented (Xu et al., 2010, 2014; Chen and Yang, 2014), many important questions remain unanswered in this signalling system. The controversy raised by the recent Gao et al. (2015) report remains to be addressed by further experimentation involving the new abp1 knockout alleles and previously described abp1 genetic materials. This report also raises the question whether ABP1 is the only extracellular sensor, and thus there is an urgent need to discover other possible extracellular auxin sensors that act redundantly with ABP1. Furthermore, there is still a missing link connecting the steps between perception of the auxin signal and activation of ROPs, which will be worthwhile to investigate in the future. Recent examples of the mechanism of pollen tube growth and ROS-mediated polarized root hair development ties different RLKs with RopGEFs (guanine nucleotide exchange factors that activate ROPs) and/or with downstream ROPs (Kaothien et al., 2005; Zhang and McCormick, 2007; Duan et al., 2010; Humphries et al., 2011; Chang et al., 2013). Hence, we anticipate that the auxin activation of the ABP1–TMK complex possibly induces the auto-phosphorylation of TMKs, and TMK phosphorylation towards plant-specific RopGEFs would be a prerequisite for the activation of the downstream ROP signalling that relies on the RopGEF activity (Fig. 1). Further identification of specific RopGEFs regulated by different TMKs in various tissues will be one of the exciting challenges of the coming years. In addition, considering the host of Rho GTPase regulators and effectors, such as ICR1, and a growing body of polarity proteins, it will be important to determine other components involved in the processes of cell morphogenesis and PIN polarization in connection to ABP1/ROP-based auxin signalling network than those currently known to act during interdigitated cell expansion, and to extend the analogy of such elegant mechanisms to various spatial controls of cellular activities and further various plant or animal systems.

Similar to the action of auxin in plants, the WNT ligand acts as a universal developmental and polarizing signal driving coordinated cell polarization in vertebrates. The canonical WTN pathway is known to control transcription-based pattern formation, and the non-canonical WTN pathway controls Rho GTPase-dependent PCP (Chen et al., 2006; von Maltzahn et al., 2012; Yang, 2012; Mayor and Theveneau, 2014; Wang et al., 2014); analogously, TIR1/AFB-based auxin signalling controls development through gene transcription, and non-transcriptional ABP1/TMK-based auxin signalling controls cell polarity in plants. Nevertheless, transcriptional and post-transcriptional auxin crosstalk is still poorly characterized. Therefore, it will be interesting to investigate downstream components of ABP1 and elucidate how the ABP1 auxin perception/signalling mechanisms intersect with the TIR1/AFB nuclear pathway to control transcriptional regulation in response to auxin, and whether such crosstalk between the ABP1 and the TIR1/AFB pathways is a common design principle for the coordination of cell polarization processes by the highly conserved Rho family of small GTPases.

Cytokinin and auxin regulate a diverse array of plant processes, including PC interdigitation, often acting antagonistically to control developmental patterning, although the molecular mechanisms regarding this antagonism have only recently begun to be unravelled. Similarly, a possible connection between auxin signalling and mechanical stress will be an exciting topic in the puzzle-piece PC system. Further elucidating the mechanistic connection between these signalling pathways that achieve an integrated response to PC morphogenesis will likely be an exciting and fertile area of research in cell and developmental biology at least over the next decade.