Root gravity response module guides differential growth determining both root bending and apical hook formation in Arabidopsis

Multicellular organisms have evolved mechanisms for the protection of cells that are essential for their survival. In dicot plants, the apical hook, which results from differential cell growth on two sides of the hypocotyl, has evolved to protect the delicate shoot meristem as the seedling approaches the soil surface (Darwin and Darwin, 1881; Raz and Ecker, 1999). By gradual bending of the apical part of the hypocotyl, the hook is formed soon after germination; it is maintained in a closed state while the hypocotyl continues to penetrate through the soil, and rapidly opens when exposed to the light in proximity of the soil surface (Vandenbussche et al., 2010; Žádníková et al., 2010).

Plant hormones, including auxin, play an important role in the regulation of apical hook development (Stepanova et al., 2008; Vandenbussche et al., 2010; Žádníková et al., 2010; Mazzella et al., 2014). Defects in auxin metabolism, transport and signaling dramatically affect all phases of apical hook growth (Gallego-Bartolomé et al., 2011; Abbas et al., 2013). In particular, asymmetric auxin distribution is linked with differential cell growth and proper apical hook development (Kuhn and Galston, 1992; Vandenbussche et al., 2010; Žádníková et al., 2010, 2016). Local auxin accumulation is indispensable for hook bending, and chemical or genetic inhibition of polar auxin transport severely interferes with apical hook development (Friml et al., 2002; Forner and Binder, 2007). Dynamic auxin distribution is tightly controlled by polar auxin transport, involving mainly AUXIN/AUXIN-LIKE (AUX/LAX) influx proteins and PIN-FORMED (PIN) efflux carriers (Petrasek and Friml, 2009; Vandenbussche et al., 2010; Žádníková et al., 2010; Cho and Cho, 2012; Swarup and Péret, 2012). Analysis of expression and membrane localization of auxin carriers suggests that auxin from the central cylinder is redirected through the endodermis towards the cortex and epidermis in the upper part of the hypocotyl, and afterwards, in the epidermis layer, redistributed to the inner side of the hook. Several PIN proteins, including PIN3, PIN4 and PIN7, are involved in this process. Mutations in these PIN genes led to specific defects in apical hook development. It has been proposed that the higher abundance of PIN3 and PIN4 at the cell membrane on the convex (outer) side compared with the concave (inner) side of the apical hook might enhance the draining of auxin from the outer cortex and epidermal layers and the formation of an auxin maximum at the concave side (Fig. S1A; Žádníková et al., 2010). Indeed, mathematical modeling of auxin transport dynamics, based on the observed PIN expression pattern, supports this model (Žádníková et al., 2016).

Although much progress has been made in apical hook development research, several key questions, such as what is the mechanism that determines the apical hook formation, remain unanswered. As no sign of asymmetric auxin distribution was reported prior to hook bending, the question arises of how asymmetry of auxin distribution is established in the early phases of seedling growth. Here, we show that bending of the root in response to gravity, which is driven by auxin accumulating at the gravi-stimulated side, is the initial cue coordinating formation of the apical hook. Interference with the gravity response of the root, either by genetic or mechanical means, affects hook formation. Accordingly, after germination is initiated, the polar auxin transport machinery that mediates the root gravity response is gradually established to align root growth with the gravi-stimulus. Importantly, during these early phases of germination, expression of the auxin efflux carrier PIN2 extends beyond the root border and, thus, at the gravi-stimulated side, the auxin maximum extends into hypocotyl. Such an enlarged auxin maximum might promote differential growth of both root and hypocotyl cells and trigger initial bending of the hypocotyl. We propose that this primary differential cell growth at the base of the hypocotyl might contribute to the establishment of a regulatory feedback loop that reinforces an axial asymmetry of polar auxin transport that governs hypocotyl bending and formation of the hook. Our study demonstrates that apical hook formation is the result of tight root-hypocotyl communication when the initial stimulus originating from root bending is transmitted and perceived by the hypocotyl. We show that a tight balance between abscisic acid (ABA) and gibberellins (GA), two principal hormonal regulators of seed maturation and germination (Nambara et al., 1995; Delmas et al., 2013; Liu and Hou, 2018), is crucial to preset hypocotyls for proper root-hypocotyl communication, which is required for apical hook formation.

The origin of the hypocotyl asymmetry that leads to apical hook formation is still unknown. Prior to germination, the Arabidopsis mature embryo is curled up in typical ‘U’ shape with the two cotyledons bent over as a result of seed coat mechanical constraints (Fig. S1B). We questioned whether this asymmetry is transmitted to apical hook formation in course of seedling germination. To explore whether embryonic shape acts as a regulatory cue for apical hook formation, mature embryos were excised from their seed coats and positioned on media in an orientated manner either root-downwards (embryo in a ‘Ո’ shape with both root and cotyledon poles facing downwards), or root-upwards (embryo in a ‘U’ shape with both root and cotyledon poles facing upwards; Fig. S1B, Fig. 1A,B), and apical hook formation during germination was examined. If embryo shape is decisive for hook curvature, then hook curvature should match the original embryonic bending regardless of orientation on the medium. Monitoring apical hook formation in real time revealed that embryos in the root-downwards position formed the apical hook that followed the embryo shape (Fig. 1A, Movie 1). However, germination of seedlings in the root-upwards position led to apical hook formation opposite to the original ‘U’ structure of the embryo (Fig. 1B, Movie 2). This result suggests that the shape of the mature embryo is not decisive for apical hook bending. However, observation of seedlings germinating from embryos in the root-upwards position indicated that growth and bending of the root in response to gravity might coordinate formation of the apical hook.

Germination of seedlings starts by outgrowth of the embryonic root and its proper alignment with the gravity vector. To explore the role of the embryonic root in apical hook development, we tracked hook curvature formation during seedling germination after dissection of the embryonic root. Monitoring of root-less seedling development in real time revealed severe defects in the formation of the apical hook. Frequently, these seedlings were either unable to form fully closed apical hooks, or apical hooks were formed with significant delays and were maintained for a shorter time compared with control seedlings with intact roots (Fig. S1C,D). To corroborate the role of root gravitropism in the coordination of apical hook formation, we searched for molecular factors that specifically control the root gravity response but are not otherwise involved in growth and developmental processes in etiolated seedlings. The PIN2 auxin efflux carrier, a key component of the root gravitropic response expression of which is restricted to the root in young seedlings (Müller et al., 1998), was considered one of the most suitable candidates to test the potential contribution of the root gravitropic response to hook formation. Mature embryos of pin2 mutants were dissected from seed coats and positioned either root-downwards or root-upwards to follow development of seedlings in real time. Hypocotyls of pin2 mutants that germinated from the root-downwards orientation typically curved in alignment with the embryonic root and were mostly able to form an apical hook, although the maintenance phase of these hooks was shorter compared with wild-type seedlings (Col-0) (Fig. 2A,B, Movie 3). The root-upwards orientation severely interfered with hook curvature and the seedlings either randomly developed an apical hook, with significant delays, or in some individuals no apical hook formed (Fig. 2C,D, Movie 4). This difference in ability of pin2 mutants to form an apical hook might correlate with penetrance of the root agravitropic phenotype, which we noticed was affected by initial orientation of mature embryos. In pin2 mutants that started to germinate from a root-upwards orientation, root growth was more erratic compared with root-downwards positioned seedlings.

Recently, NEGATIVE GRAVITROPIC RESPONSE OF ROOTS (NGR)/LAZY genes were implicated in the direction of root gravitropism. Unlike the agravitropic pin2 mutant, in which a root growth defect is caused by insufficient basipetal (shoot-ward) auxin transport, roots of the ngr1,2,3 triple mutant exhibit a negative gravitropic response. In the ngr1,2,3 mutant, auxin accumulates at the non-stimulated side of root and, accordingly, it enhances root bending away from the gravity stimulus (Ge and Chen, 2016). Hence, the ngr1,2,3 mutant was a suitable candidate for testing the correlation between root bending and apical hook formation. Monitoring mutant seedlings as they developed from the root-down or root-up orientated embryos revealed that the apical hook is formed in strict alignment with root bending (Fig. 2E-G). Remarkably, when the root of ngr1,2,3 mutants is facing upwards, the hook is formed in a unique upside-down direction, opposed to the gravity stimulus, similar to observations of root growth (Fig. 2F). These results indicate that root bending is an essential cue that coordinates formation of the apical hook.

Interestingly, root meristemless1 (rml1) mutants, despite a severe malfunction of the root apical meristem, as manifested by root growth arrest (Cheng et al., 1995), were still able to form apical hooks (Movie 5). Careful examination of the early phases of rml1 seedling germination revealed that the embryonic root is able to grow and respond to the gravity stimulus, and that root growth arrest occurs only in the later phases of germination (Movie 5). Similarly, inhibition of the root gravity response as a consequence of lateral root cap-specific accumulation of AXR3, an auxin signaling repressor (Ouellet et al., 2001; Swarup et al., 2005), did not affect apical hook formation in either the root-down (Movie 6) or root-up (Movie 7) position. Detailed examination of J0951, an activator used specifically to stimulate AXR3 in root tissues, revealed that its expression started only later, after the embryonic root outgrowth (Fig. S1E). Hence, in the early phase of germination, the embryonic root grew and responded to the gravity stimulus comparably to the wild-type control and loss of root gravitropism triggered by accumulation of AXR3 during later phases no longer interfered with formation of the apical hook.

The impact of early root gravity response deficiency on apical hook formation in pin2 and ngr1,2,3 mutants compared with rml1 and axr3 mutants indicates that there might be a short developmental window during which gravity-driven root bending acts as an important cue to coordinate formation of the apical hook. To assess the robustness and duration of such a developmental window, we turned seedlings 180° at different time points during germination and monitored the impact of the changed gravity vector on formation of the apical hook.

When embryos were turned shortly before or up to ∼6 h after germination started by root outgrowth, the apical hook always formed in alignment with the root bend (Fig. S2A,B, upper panel). However, changing the gravity vector at later time points (∼7-8 h after germination) by turning seedlings with emerged roots (longer than 0.7 mm) did not interfere with apical hook formation. Despite realignment of root growth due to gravistimulation, formation of the apical hook proceeded according to the original orientation determined by the root bend prior to turning (Fig. S2A,B, lower panel).

SCARECROW is an important regulator of endodermis formation, a layer that is essential for shoot gravitropism (Fukaki et al., 1998). To explore whether hypocotyl gravity response contributes to apical hook formation in addition to gravity-driven root bending, we examined scr-3 mutants. We found that regardless of embryo orientation, scr-3 seedlings were able to form the apical hook (Fig. S2C,D), suggesting that the hypocotyl gravity response might not be a limiting factor during apical hook formation.

Altogether, our results indicate that there is a short developmental window (less than 8 h after germination starts) during early phases of germination when the gravity-driven root bending acts as an important cue to trigger apical hook formation.

It is accepted that the apical hook in dicotyledonous plants is generated when seedlings germinate in darkness, whereas light stimulates rapid opening of the apical hook (MacDonald et al., 1983; Raz and Ecker, 1999; Raz and Koornneef, 2001; Vandenbussche et al., 2010; Žádníková et al., 2010). Identification of the root gravity response as a light-independent cue that orchestrates the early phase of apical hook formation prompted us to hypothesize that the apical hook might be formed irrespective of light conditions. To test this hypothesis, we germinated seedlings under constant illumination. Real-time monitoring revealed that early phases of seedling development in light resemble those occurring in darkness and light does not prevent formation of the apical hook curvature. As seedlings start to grow, the embryonic root rapidly expands, bends downwards with the gravity vector and drives formation of the apical hook (Fig. 3A, Movies 8,9). However, unlike etiolated seedlings, the apical hook formed in constant light exhibits a very short maintenance phase and tends to rapidly open (Fig. 3B). In summary, these results support the hypothesis that gravity-stimulated root bending is a crucial factor that coordinates apical hook formation regardless of light conditions.

Apical hook formation is driven by auxin, local accumulation of which in epidermal cells at one side of hypocotyl defines the concave side of the hook curvature (Raz and Ecker, 1999). Although functional polar auxin transport has been implicated in establishment of the auxin maximum, the mechanisms that govern the asymmetry of auxin distribution are unknown. To explore how and when the auxin maximum that drives formation of the apical hook is established during germination, mature embryos dissected from their seed coats were grown on the Murashige and Skoog (MS) medium and expression of DR5-derived auxin response reporters was carefully monitored during germination. As expected, mature embryos exhibited auxin response maxima in cotyledons and root columella cells; however, no DR5 activity in embryonic hypocotyls prior germination could be detected (Fig. 4A, Fig. S3A,B). Typically, germination of seedlings starts by expansion of the embryonic root and rapid adjustment of its growth direction with the gravity vector. The gravitropic response of the root is coordinated by asymmetric auxin distribution (Ottenschlager et al., 2003; Kleine-Vehn et al., 2010) and, accordingly, we observed a higher auxin signal at the lower (gravity-stimulated) side of the growing root (Fig. 4A, Fig. S3A,B). Over time, a transient auxin response maximum appeared at the root-hypocotyl junction and this local accumulation of hormone correlated with a local bending, which can be considered as the start of apical hook formation (Fig. 4A, Fig. S3A,B). At around 36 h, the local auxin maximum at the root-hypocotyl junction fades, but as the hypocotyl continues to grow, the auxin response at the concave side of the hook region becomes gradually stronger and the hook continues to bend until it is fully closed (Fig. 4A, Fig. S3B). A similar pattern of auxin response maximum establishment was observed using DR5::RFP reporter in seedlings germinated in light (Fig. S3C).

Hence, during seedling germination the earliest detectable asymmetry in auxin distribution is linked with the root response to gravity stimulus and afterwards a local auxin maximum that correlates with the formation of hook curvature is observed.

Root gravitropic bending as well as apical hook formation are driven by asymmetric auxin distribution mediated by polar auxin transport machinery with PIN2, PIN3, PIN4 and PIN7 being major efflux carriers involved in both these processes (Kleine-Vehn et al., 2008; Abbas et al., 2013; Žádníková et al., 2016). Typically, PIN2 expression is restricted to the root meristem where its localization at the apical membranes of epidermal cells mediates the basipetal (shoot-ward) auxin transport required for the proper root gravity response (Müller et al., 1998; Kleine-Vehn et al., 2008). Interestingly, during the early phases of germination the boundary between the root and the hypocotyl seems to be not strictly defined and PIN2 was detected in hypocotyls where it localizes to the apical membranes of epidermal cells (Fig. 4B, Fig. S5A). As the hypocotyl and root expanded, PIN2 expression in hypocotyls ceased, and remained restricted to the root meristem, as previously reported (Müller et al., 1998). The lack of a precisely confined border of PIN2 expression at the root-hypocotyl junction indicates that, at early phases of seedling germination, PIN2 might transport auxin across the root zone towards the hypocotyl. Also, early expression of PIN3 and PIN7 was detected in the root columella cells where they can contribute to the perception of gravity stimulus (Figs S4A,B, S5B-D; Friml, 2003; Kleine-Vehn et al., 2010). In embryonic hypocotyls, the expression of PIN genes is generally low, and as seedlings germinate and the apical hook gradually forms, it progressively enhances (Figs S4A,B, S5B-D; Žádníková et al., 2010). Similar patterns of PIN gene expression were detected in germinating seedlings exposed to light (Fig. S6A-D). Altogether, the spatiotemporal pattern of PIN gene expression indicates that the transport machinery, which mediates auxin redistribution to coordinate the root gravitropic response, is established at very early phases of germination, and subsequently the transport system for auxin redistribution through the hypocotyl is formed.

Our results indicate that root gravitropic bending acts as an important regulatory cue for apical hook formation and that the apical hook can be formed under both dark and light conditions. Hence, we hypothesized that, rather than light conditions, a predetermination of the skoto-morphogenic developmental program in seedlings germinating from mature non-green embryos might be decisive for apical hook formation. To test this hypothesis, we examined seedlings developing from mature embryos prior to de-greening (normally around 12-14 days after fertilization; Delmas et al., 2013). When monitored in real time, we found that seedlings developing from green embryos initiate germination by outgrowth of the embryonic root and its proper alignment with the gravity vector; however, hypocotyls of such seedlings are not able to form an apical hook (Fig. 5A, Movies 10,11). Importantly, seedlings developing from green embryos are fully viable and after 10 days largely comparable to those grown from non-green embryos under the light (Fig. S7). Hence, although seedling roots from germinating green embryos respond properly to gravity, this early stimulus is not transmitted into the differential hypocotyl growth required for apical hook formation. Because asymmetric redistribution of auxin directs both root gravitropic bending and apical hook formation, we monitored the auxin response in seedlings germinating from green embryos. Similarly to non-green embryo seedlings, those germinating from green embryos exhibited asymmetric auxin distribution that correlates with the root gravity response. However, no auxin asymmetry and formation of local maxima could be detected in hypocotyls of such seedlings (Fig. 5B). Together, these results indicate that apical hook formation might be dependent on two mechanisms, root gravity perception and hypocotyl-specific regulatory machineries, that control auxin distribution in a coordinated way.

A complete lack of the auxin response maximum and hook formation in seedlings germinated from green embryos, despite their normal root gravity response, provided a valuable model to further explore requirements for establishment of the polar auxin transport that guides bending of the hook. Detailed analysis of the spatiotemporal expression patterns in early phases of development revealed that PIN2 and PIN4 in seedlings germinating from green embryos are, apart from roots, also expressed in hypocotyls (Fig. S8A,B). Both PIN3 and PIN7 were expressed in seedlings germinated from green embryos from 24 h onwards. Their expression was confined to the central vasculature of hypocotyls and a weaker signal in outer tissues, including cortex and epidermis, could be detected (Fig. S8A,B). Quantitative RT-qPCR analysis largely corroborated these results (Fig. S8E). Surprisingly, despite detectable transcription of PIN genes, no PIN proteins were found in hypocotyls of seedlings developing from green embryos, either by observation of PIN-GFP lines using confocal microscopy (Fig. 5C), or by western blot analysis applying antibodies specific to GFP (Fig. S8F). In roots, regardless of whether grown from green or non-green embryos, membrane localized PINs were detected exhibiting expression patterns as previously described (Fig. 5C, Müller et al., 1998; Friml, 2003; Blilou et al., 2005). Together, these results indicate a notable difference in establishment of the auxin transport system in seedlings germinated from non-green versus green embryos. Whereas seedlings developing from non-green embryos establish auxin transport in both roots and hypocotyls, PIN-mediated transport in hypocotyls of seedlings originating from green embryos is largely attenuated.

The lack of PINs in hypocotyls grown from green embryos prompted us to test whether enhancement of PIN expression might be sufficient to recover the auxin transport required for apical hook formation. Constitutive 35S promoter-driven expression of PIN1-GFP was unable to promote hypocotyl growth and formation of the apical hook in seedlings germinated from green embryos (Fig. S9A). Detailed observation of subcellular localization revealed that weak or no membrane-localized PIN1-GFP signal could be detected in hypocotyls grown from green embryos; instead, a high accumulation in intracellular bodies was found. As expected in hypocotyls developing from non-green embryos, membrane-localized PIN1-GFP was observed with no (or very small) intracellular bodies. Those effects were clearer at 72 h, but could be seen at times as early as 24 h after germination (Fig. 6). In roots, PIN1-GFP was located in the plasma membrane regardless of whether grown from green or non-green embryos (Fig. S9B). Similarly to PIN1, the aquaporin PLASMA MEMBRANE LOCALIZED PROTEIN2 (PIP2; Dhonukshe et al., 2007) accumulated within hypocotyl cells with weak or no signal in the plasma membranes (Fig. 6), although in roots its plasma membrane localization was clearly detected (Fig. S9B). This suggests that subcellular trafficking of membrane proteins in hypocotyls of seedlings developing from green embryos dramatically differs from that occurring in hypocotyl cells of seedlings originating from non-green embryos.

Maturation of green embryos and their de-greening have been shown to be under negative control of the plant hormone abscisic acid (ABA). Typically, green embryos exhibit a peak of ABA concentration that decreases during the de-greening process (Nambara et al., 1995; Delmas et al., 2013). Consistent with this, we detected a higher expression of genes involved in ABA synthesis and signaling, as well as a downregulation of genes involved in ABA inactivation, in seedlings developing from green versus non-green embryos (Fig. S10A). To test whether a higher level of ABA might underlie the differences in hypocotyl growth and the lack of apical hook in seedlings developing from green compared with non-green embryos, we examined seedlings grown from non-green embryos in the presence of ABA. We found that ABA treatment severely restricted hypocotyl growth and the formation of the apical hook, but it did not prevent embryonic root outgrowth and its gravity response, thus strongly mimicking the phenotype of seedlings germinated from green embryos (Fig. 7A compared with 5A). Also, the effect of ABA on PIN gene expression largely resembled the pattern observed in seedlings developed from green embryos (Fig. S8C-E). Similar to observations in hypocotyls developed from green embryos, ABA enhanced expression of PIN2 and PIN4, whereas PIN3 and PIN7 expression was largely restricted to the central cylinder (Fig. S8C). Moreover, the inhibitory effect of ABA on hypocotyl growth and PIN gene expression could be eliminated by washing out the hormone (Fig. S8D). Also, as in seedlings developing from green embryos, an auxin maxima could not be detected in hypocotyls of ABA-treated non-green seedlings (Fig. S10B). Next, we tested whether ABA interferes with trafficking of plasma membrane proteins as observed in hypocotyls germinated from green embryos. Whereas membrane-localized PIN3 and PIN7 were observed in untreated hypocotyls, in the presence of ABA (5 h) no membrane-localized PINs in hypocotyl cells could be detected, although expression in this organ was still detectable (Fig. 7B, Fig. S10C). These results suggest that ABA might interfere with the proper trafficking of PINs to the plasma membrane of hypocotyl cells. To examine further the ABA effect on membrane protein trafficking, seedlings constitutively expressing PIN1 under the 35S promoter were treated with ABA. ABA enhanced intracellular accumulation of PIN1 in epidermal cells of hypocotyls compared with untreated control (Fig. 7C,D). Hence, similarly to green embryos, ABA interfered with the proper establishment of the auxin transport system required for apical hook formation in hypocotyls.

High ABA activity was identified as a potential factor that might interfere with establishment of the auxin transport machinery required for apical hook formation in hypocotyls of seedlings germinated from green embryos. To test whether reduction of ABA levels during germination of seedlings from green embryos might lead to a recovery of apical hook formation, we applied chemical and genetic tools. Reduction of ABA levels either chemically, using a synthetic inhibitor of ABA biosynthesis abamine (Abm; Han et al., 2004), or genetically, in a mutant defective in ABA biosynthesis, Arabidopsis thaliana aba deficient 2 (aba2; Nambara et al., 1998) (ABA2 is highly expressed in green embryos; Fig. S10A), did not lead to recovery of apical hook formation. Similar to untreated seedlings germinating from green embryos, the growth and gravity response of Abm-treated wild-type Col-0 and aba2 roots were largely unaffected, whereas the hypocotyls of both Abm-treated Col-0 as well as aba2 seedlings were unable to undergo elongation and form apical hooks (Fig. 8A,C). Moreover, additional reduction of ABA levels by applying Abm to the aba2 mutant also did not result in recovery of apical hook formation (Fig. 8D, Table S1). This indicated that reduction of ABA alone is not sufficient to promote apical hook formation and another regulatory factor might be involved. Seed maturation and germination are two tightly linked processes antagonistically regulated by ABA and GAs (Stamm et al., 2017; Liu and Hou, 2018). However, application of GA3 was not sufficient to promote hypocotyl elongation growth and hook formation in seedlings germinating from green embryos in the presence of Abm (Fig. 8B). Only when levels of ABA were reduced by applying Abm to aba2 and GA3 levels were increased simultaneously were both the hypocotyl growth and the apical hook formation recovered (Fig. 8E, Table S1). Together, these results suggest that, during embryo maturation and the transition to germination, a tight ABA-GA balance is required to stimulate hypocotyl growth and apical hook formation.

Apical hook formation is a result of differential growth, with auxin being a key player to guide asymmetry of cell elongation at the concave versus convex side of the hypocotyl (Raz and Ecker, 1999). Intriguingly, no asymmetry in auxin distribution and expression of PINs could be detected in embryonic hypocotyls prior to seedlings starting to develop, thus raising a question about the regulatory factors that determine the local increase of auxin that drives formation of the apical hook.

Thorough monitoring of auxin responses identified the gravi-stimulation of embryonic roots as one of the earliest triggers of asymmetric auxin distribution after seedlings start to germinate. This growth response, underpinned by redistribution of auxin towards the gravi-stimulated side of the root (Ottenschlager et al., 2003; Kleine-Vehn et al., 2010), is particularly important at early phases of germination when emerging roots adjust their growth direction with the gravity vector. Our work suggests that this gravity-driven root bending acts simultaneously as an important regulatory cue for apical hook formation. Real-time monitoring of seedlings germinating from embryos positioned either in a root downwards or upwards orientation showed that apical hooks are formed in a strict alignment with root bending. In contrast, the curved shape of mature embryos enclosed in the seed coat does not necessarily pre-determine a hook bend and can be overridden by gravity-driven root bending. Lack of the root gravity response in either the pin2 mutant or after mechanical destruction of the root tip interferes with apical hook formation. Observation of the ngr1,2,3 mutant further supports a contribution of the root bending as an initial cue in apical hook formation. Despite the fact that this negatively gravitropic mutant has roots that bend away from the gravity vector, the apical hook still aligns with the curvature of the root. Importantly, as in wild type, root bending is driven by auxin. Although in ngr1,2,3 mutants its increase at the non-gravi-stimulated side promotes bending away from the gravity vector (Ge and Chen, 2016; Yoshihara and Spalding, 2017). The ability to form a hook is maintained in seedlings germinating under constant light, an established trigger of hook opening, thus further supporting a dominating role of root bending in initial phases of apical hook formation. On the other hand, interference with root growth and gravity response during later germination phases does not affect formation of the apical hook, suggesting that there is a short developmental window after germination starts when root bending coordinates formation of the hook curvature. Perturbation of the gravity vector during development by rotating seedlings indicates that the developmental window critical for determination of the apical hook is ∼8 h from the start of germination. Altogether, these results support a model in which, during early phases of germination, a gravity-stimulated asymmetry in auxin distribution coordinates root bending and, consequently, the formation of the apical hook. However, scr-3 mutants affected in hypocotyl gravitropism did not exhibit any dramatic defects in the formation of the apical hook, indicating that hypocotyl response to gravity might not be limiting for formation of the hook.

It is worth noting that studies examining plant growth in microgravity conditions reported that apical hooks may be formed even under zero gravity (Ueda et al., 2000; Miyamoto et al., 2014). In agreement with these reports, our results (in particular the observations in ngr1,2,3 mutants) suggest that root bending, rather than gravi-stimulus per se, directs formation of the hook. Hence, in zero gravity, any bending of the root, which could be potentially triggered by other stimuli (e.g. negative root phototropism), might lead to the apical hook formation.

Over recent years, the mechanism controlling auxin redistribution in gravi-stimulated roots has been uncovered (Bennett et al., 1996; Galweiler et al., 1998; Luschnig et al., 1998; Müller et al., 1998; Friml et al., 2002; Swarup et al., 2005; Harrison and Masson, 2008). In response to gravity, PIN3 and PIN7 rapidly relocate from apolar to polar membrane localization with maxima at the bottom of root columella cells, which leads to an increase of auxin transported via PIN2 at the lower side of the gravi-stimulated root (Friml et al., 2002; Kleine-Vehn et al., 2010). Higher auxin limits cell expansion and ultimately promotes root bending (Ottenschlager et al., 2003; Brunoud et al., 2012). In line with a proposed role of the root gravity-stimulated bending as an initial factor coordinating apical hook formation, early onset of PIN3, PIN7 and PIN2 expression in root columella and epidermal cells was observed.

However, the question remains of how this initial stimulus related to the auxin-driven root bending is transmitted to coordinate formation of the apical hook. Detailed monitoring of PIN2 indicates that during early phases of germination, expression of this efflux carrier is not restricted to the root, but it extends towards a hypocotyl. Thus, PIN2-mediated polar auxin transport might contribute to the initial accumulation of auxin at (above) a root-hypocotyl junction and thereby stimulate differential growth of cells in this zone. In the hypocotyls, this initial auxin-driven growth asymmetry might be further reinforced by gradually established auxin transport machinery. Accordingly, we found that prior to germination the expression of PIN homologs, including PIN3, PIN4 and PIN7, in embryonic hypocotyls is low and that progressively, as seedlings grow, it gradually increases, adopting the pattern reported previously (Žádníková et al., 2010, 2016).

Another intriguing aspect is the establishment of axial asymmetry of PINs that reinforces an auxin response maximum formed initially because of root response to gravity mediated by PIN2. We propose that the initial auxin-driven differential growth at the root-hypocotyl junction might lead to the formation of a regulatory feedback loop, which reinforces an asymmetry in PIN expression and subcellular dynamics in cells at the outer versus inner side of bending hypocotyls. An important part of this regulatory feedback loop might be growth-related differences in mechanical forces in cells at the outer and inner side of the bending hypocotyl. Changes in mechanical strains, such as modifications of turgor pressure or the application of external force, have been found to affect subcellular trafficking and membrane localization of PINs. In tissue under strain, a higher proportion of PIN localized to the plasma membrane has been detected (Nakayama et al., 2012). Additionally, a differential distribution of auxin needs to be considered as an important element of this regulatory feedback loop. Auxin has been found to promote expression of PIN3 and PIN4, to attenuate PIN7 transcription, and to inhibit endocytosis of PINs. Thus, auxin might be an important factor in the establishment and maintenance of axial asymmetry of PINs in the hypocotyl (Paciorek et al., 2005; Vieten et al., 2005; Uyttewaal et al., 2010, 2012; Peaucelle et al., 2011).

The phenotype of seedlings germinating from green embryos hinted at additional aspects of mechanisms underlying the apical hook formation. First, the importance of proper root-hypocotyl communication via two tightly coordinated auxin transport machineries and, second, the role of the ABA pathway to preset mature embryos for the establishment of the polar transport machinery anew. We found that, unlike seedlings germinated from mature non-green embryos, hypocotyls of seedlings developing from mature, but still green embryos were not able to perceive and transmit the root-bending stimulus towards apical hook formation. Detailed observations of auxin and polar auxin transport machinery revealed that whereas in seedlings originating from non-green embryos the establishment of the polar auxin transport machinery in roots and hypocotyls was tightly co-regulated, in hypocotyls developing from green embryos PIN-mediated transport was significantly suppressed. These data indicate that formation of the apical hook is dependent on the coordinated establishment of polar auxin transport machineries in both roots and hypocotyls.

Mature embryos prior to de-greening are under strong control of ABA (Frey et al., 2004; Finch-Savage and Leubner-Metzger, 2006), as corroborated by the expression analysis of several ABA regulatory network components. Interestingly, seedlings developing from non-green embryos in the presence of ABA resemble those germinated from green embryos. Similar to seedlings developing from green embryos, hypocotyl growth and formation of the apical hook were severely suppressed by ABA applied to non-green embryos. This deficiency in hypocotyl growth and apical hook formation correlate with attenuated expression and trafficking of PINs to the plasma membrane. It is worth noting that similar effects of ABA on PIN expression have been reported in roots (Shkolnik-Inbar and Bar-Zvi, 2010; Yang et al., 2014; Rowe et al., 2016), although our results suggest that embryonic hypocotyls exhibit higher sensitivity to ABA compared with roots. Hence, high ABA in green embryos might be part of the regulatory pathway that at the end of embryogenesis helps to restrain former polar auxin transport machinery and thus enables to create a ‘tabula rasa’ on which a novel auxin transport system can be established as seedlings start to germinate. Recovery of the apical hook formation in seedlings developing from green embryos after simultaneous reduction of ABA and increase of GAs further suggests that a tight balance between both hormones is required to guide hypocotyl growth and apical hook formation.

Based on our observations, we propose a model for the establishment of auxin transport that guides apical hook formation. High levels of ABA during maturation of embryos promote pathways that suppress expression of polar auxin transport components and erase remnants of PINs from the membrane. As a result, mature embryos, after de-greening, are deprived of former auxin distribution machineries, which enables them to build the auxin transport system anew. During germination initiated by outgrowth of roots and their alignment with the gravity vector, PIN3, PIN7 in columella and PIN2 in root epidermal cell are expressed first. Extended expression of PIN2 towards the root-hypocotyl junction enables formation of a local auxin maximum at the transition zone, thereby triggering differential growth. The differential cell expansion at the root-hypocotyl junction driven by auxin might, together with GAs, lead to the establishment of a regulatory feedback loop reinforcing the differential expression and subcellular dynamics of PINs at the outer and inner sides of bending hypocotyls that are required for the auxin maxima and apical hook formation.

Arabidopsis thaliana (L.) Heynh (Arabidopsis) plants were grown in a growth chamber at 21°C under white light, which was provided by blue and red LEDs (70-100 µmol m⁻² s⁻¹ of photosynthetically active radiation). The transgenic lines used have been described elsewhere: DR5::GUS (Sabatini et al., 1999); PIN1::GUS, PIN2::GUS, PIN3::GUS, PIN4::GUS, PIN1::PIN1-GFP and PIN7::PIN7-GUS (Benková et al., 2003); PIN7::GUS and PIN4::PIN4-GFP (Vieten et al., 2005); PIN3::PIN3-GFP and PIN7::PIN7-GFP (Žádníková et al., 2010); pin2 mutant and rml1 mutant (Cheng et al., 1995); J0951≫UAS-AXR3 (Swarup et al., 2005); 35S::PIN1-GFP (Marhavý et al., 2011); PIN2::PIN2-GFP (Xu and Scheres, 2005); 35S::PIP2-GFP (Cutler et al., 2000); ngr1,2,3 (Ge and Chen, 2016); and scr-3 and aba2-1 from NASC.

Seeds were sterilized in 5% bleach for 10 min and rinsed with sterile water before plating on half-strength MS medium (Duchefa) with 1% sucrose, 1% agar (pH 5.7). Seeds were stratified for 3-4 days at 4°C, exposed to light for 2-4 h at 21°C, and cultivated in the growth chamber under appropriate light conditions at 21°C (wrapped in aluminum foil and placed in a cardboard box for dark growth, or into light in the same chamber). For non-green embryos, seeds were dissected after stratification and imbibition in water. For green embryos, flowers were marked by a marker line on the stem or by tying small thread on it, and siliques around 12-14 days after fertilization were used for the dissection as described elsewhere (Delmas et al., 2013). Seedlings were processed at the indicated times after germination or used for real-time phenotype analysis. Real-time analysis and statistics of apical hook development of seedlings were recorded at 1 h intervals for 5 days at 21°C with an infrared light source (880 nm LED; Velleman, Belgium) or constant light by a spectrum-enhanced camera (EOS035 Canon Rebel Xti, 400DH) with built-in clear wideband-multicoated filter and standard accessories (Canon) and operated by EOS utility software as previously described (Zhu et al., 2017). Angles between the hypocotyl axis and cotyledons were measured using ImageJ (NIH; http://rsb.info.nih.gov/ij) as described previously (Žádníková et al., 2010). Fifteen to twenty seedlings were processed.

Indole-3-acetic acid, 1-naphthaleneacetic acid (NAA) and abscisic acid (ABA) were from Sigma-Aldrich and were used as 10 mM in ethanol stocks. Gibberellic acid (GA3) (Sigma-Aldrich) was dissolved at 100 mM in ethanol. Abm was synthesized and purified as detailed in supplemental Materials and Methods and was dissolved at 10 mM in DMSO stock.

Histochemical β-glucuronidase (GUS) staining was performed as described (Jefferson et al., 1987) with minor modifications. In brief, the GUS reaction was carried out in reaction buffer containing 0.1 M sodium phosphate buffer (pH 7), 1 mM ferricyanide, 1 mM ferrocyanide, 0.1% Triton X-100 and 1 mgml⁻¹ X-Gluc for overnight incubations at 37°C. The GUS reaction was stopped by washing in 70% ethanol, and clearing performed by incubation in a solution containing 4% HCl and 20% methanol for 15 min at 57°C, followed by 15 min incubation in 7% NaOH/60% ethanol at room temperature. Next, seedlings were rehydrated by successive incubations in 40%, 20% and 10% ethanol for 5 min, followed by incubation in a chloral hydrate (Fluka) buffer. Finally, seedlings were mounted in chloral hydrate and imaged by differential interference contrast microscopy with a BX51 microscope (Olympus, with a DP26 Olympus camera). Images were processed in ImageJ.

Confocal laser-scanning micrographs were obtained with a Zeiss LSM700 with a 488-nm argon laser line for excitation of GFP fluorescence. Emissions were detected between 505 and 580 nm. Using a 20× air objective, confocal scans were performed with the pinhole at 1 Airy unit. Localization was examined by confocal z-sectioning. Each image represents either a single focal plane or a projection of individual images taken as a z-series. z-stacking was performed by collecting images through the cortex and epidermal layers. Full z-stack confocal images were 3D-projected using ImageJ software. At least ten seedlings were analyzed per treatment. Images were processed in ImageJ.

Embryos from stratified seeds or green siliques were peeled in a sterile hood using a stereomicroscope and incubated in MS plates with the indicated treatment in the dark. Around 150 embryos per sample were harvested after 24 days or 48 h of growth and frozen in liquid nitrogen. Tissue was ground using a ball mill (model MM400; Retsch) with 4-mm diameter balls in a 2-ml Eppendorf tube. Total RNA was isolated from embryos using an RNAeasy Plant Mini kit (Qiagen). During the extraction process, samples were treated with DNase according to the manufacturer's protocol (Qiagen). cDNA was prepared from 1 µg of total RNA with the iScript cDNA Synthesis Kit (Bio-Rad), dilutions of 1/10 were prepared and 1 µl was used in a 5-µl PCR reaction on a LightCycler 480 (Roche Diagnostics) with the SYBR Green I Master kit (Roche Diagnostics) according to the manufacturer's instructions. As control, non-RT-treated samples were included to test the purity of the cDNA. All experiments were performed with three technical replicates and three biological samples. The PP2A gene (At1g69960) was used as a control for normalization. Primer sequences can be found in Table S2.

Embryos from stratified seeds or green siliques were peeled in a sterile hood using a stereomicroscope and incubated in MS plates in the dark for 48 h. Then, around 250 embryos per sample were harvested and frozen in liquid nitrogen. Frozen tissue was ground with stainless steel 4-mm diameter balls in a 2-ml Eppendorf tube using a ball mill (model MM400; Retsch). Extraction buffer [100 mM Tris-HCl (pH 7.5), 25% (w/w, 0.81 M) sucrose, 5% (v/v) glycerol, 10 mM ethylenediaminetetraacetic acid (EDTA, pH 8.0), 10 mM ethyleneglycoltetraacetic acid (EGTA pH 8.0), 5 mM KCl and 1 mM 1,4-dithiothreitol (DTT)] was added to the frozen tissue. After Bradford quantification, 20 µg of protein were diluted in 20 µl of buffer and prepared for electrophoresis by adding 5 µl of loading buffer (5× SDS) and incubating at 45°C for 5 min. The electrophoresis was performed in a commercial 10% Mini-PROTEAN TGX Precast Protein Gels (Bio-Rad) at 35 mA. Transference was carried out by a semidry system with a Trans-blot turbo transfer pack PVDF (Bio-Rad). The blot was washed in TBST buffer (Tris buffered saline and 0.05% Tween 20) + 5% milk powder, and blocked overnight in the same buffer at 4°C. Hybridization was carried out using an anti-GFP antibody (mouse monoclonal, Sigma-Aldrich, G6539, 1:7000) in TBST for 2 h at room temperature. The blot was washed three times for 10 min each wash with TBST+5% milk powder and afterwards hybridized with peroxidase-linked anti-mouse antibody (anti-mouse-HRP from GE Healthcare, NA9310, 1:15,000) in TBST for 1 h. Finally, three washes in TBST and one in water were performed, prior to visualization using the SuperSignal West Femto Maximum Sensitivity Substrate kit (Thermo Scientific) and analysis using an Amersham Imager 600 (GE Healthcare).

Sequence data from this article can be found in the Arabidopsis Genome Initiative or GenBank/EMBL databases under the following accession numbers: At1g73590 (PIN1), At5g57090 (PIN2), At1g70940 (PIN3), At2g01420 (PIN4), At1g23080 (PIN7), At1g52340 (ABA2), At4g37580 (HLS1), At4g23100 (RML1), At3g53420 (PIP2).

We thank Jiri Friml and Phillip Brewer for inspiring discussion and for help in preparing the manuscript. This research was supported by the Scientific Service Units (SSU) of IST-Austria through resources provided by the Bioimaging Facility (BIF), the Life Science Facility (LSF).