Abstract
In plants, directional transport of the hormone auxin creates concentration maxima and paths of transport that provide positional, polarity, and growth regulatory cues throughout development. In Arabidopsis thaliana, the polar-localized auxin efflux protein PIN-FORMED1 (AtPIN1) is required to coordinate development during flowering. However, Arabidopsis has a derived PIN family structure; the majority of flowering plants have retained a clade of PIN proteins phylogenetically sister to PIN1, the Sister-of-PIN1 clade (SoPIN1), which has been lost in the Brassicaceae, including Arabidopsis. Based on PIN localization in the grasses Brachypodium distachyon and Zea mays, which have both SoPIN1 and PIN1 clades, we previously proposed that the organ initiation and vein patterning roles attributed to AtPIN1 were shared between the SoPIN1 and PIN1 clades in grasses. Here we show that sopin1 and pin1b mutants have distinct phenotypes in Brachypodium. sopin1 mutants have severe organ initiation defects similar to Arabidopsis atpin1 mutants, while pin1b mutants initiate organs normally but have increased stem elongation. Heterologous expression of Brachypodium PIN1b and SoPIN1 in Arabidopsis atpin1 mutants provides further evidence for functional distinction between the two clades. SoPIN1 but not PIN1b can complement null atpin1 mutants, while both PINs can complement an atpin1 missense allele with a single amino acid change. The different localization behaviors of SoPIN1 and PIN1b when heterologously expressed in Arabidopsis provide insight into how PIN accumulation at the plasma membrane, tissue-level protein accumulation, transport activity, and interaction, all contribute to the polarization dynamics that distinguish PIN family members. Combined, these results suggest that the PIN polarization and trafficking behaviors required for organ initiation differ from those required for other PIN functions in the shoot, and that in most flowering plants these functions are split between two PIN clades.
Introduction
The plant hormone auxin is an essential mobile signal controlling growth and patterning throughout plant development (Leyser, 2010). Auxin can passively enter cells, triggering a vast array of downstream signaling events (Wang and Estelle, 2014), but it cannot easily exit the cell without active transport (Raven, 1975; Rubery and Sheldrake, 1974). As a result, directional efflux mediated by the polar-localized PIN-FORMED (PIN) efflux carriers can organize auxin accumulation patterns, creating concentration maxima and paths of transport that regulate growth, position organs, and pattern tissues (Adamowski and Friml, 2015). Because auxin itself feeds back to regulate PIN-mediated transport both transcriptionally and post-transcriptionally (Leyser, 2006), the transport system shows remarkable robustness and plasticity. For example, compensatory changes in PIN abundance between PIN family members can mitigate PIN loss-of-function mutant phenotypes (Blilou et al., 2005; Paponov et al., 2005; Vieten et al., 2005), environmental inputs can trigger tissue-level changes in PIN abundance and polarity leading to altered plant growth (Habets and Offringa, 2014), and auxin transport paths can be reorganized in response to injury (Xu et al., 2006), or spontaneously in tissue culture (Gordon et al., 2007). The self-organizing properties of the auxin transport system thus gives this patterning mechanism extraordinary versatility, and allows it to coordinate local and long range communication in the plant.
The correct initiation and positioning of organs (leaves, flowers, stems) in the growing tip, or shoot apical meristem, of Arabidopsis thaliana (Arabidopsis) plants requires the action of the PIN-FORMED1 (AtPIN1) auxin efflux carrier (Okada et al., 1991). AtPIN1 is targeted to the plasma membrane and polarized in cells (Gälweiler et al., 1998). In the meristem epidermis, polarization of AtPIN1 in neighboring cells converges around the initiation sites of new organs, suggesting that polarized AtPIN1 concentrates auxin into local maxima causing organ initiation (Benková et al., 2003; Heisler et al., 2005; Reinhardt et al., 2003). Accordingly, in atpin1 loss-of-function mutants, or if auxin transport is pharmacologically inhibited, organ initiation is aborted, but it can be rescued with local auxin application to the meristem flank (Reinhardt et al., 2003; Reinhardt et al., 2000). Organ initiation in atpin1 mutants can also be rescued with epidermal-specific AtPIN1 expression (Bilsborough et al., 2011) and reducing AtPIN1 function specifically in the epidermis compromises organ positioning and initiation (Kierzkowski et al., 2013), demonstrating the importance of convergent AtPIN1 polarization in the epidermis during organ formation.
The recurrent formation of AtPIN1 convergence points surrounding auxin maxima in the meristem epidermis has been the focus of several computational models that attempt to explain how auxin feeds back on its own transport via AtPIN1 to concentrate auxin and control organ spacing (Abley et al., 2016; Bayer et al., 2009; Bhatia et al., 2016; Heisler et al., 2010; Jönsson et al., 2006; Smith et al., 2006; Stoma et al., 2008). However, AtPIN1 is also expressed during the patterning of the vascular strands formed coincident with organ positioning, and in these sub-epidermal cells AtPIN1 is polarized basally, away from the presumed auxin maxima, suggesting that the control of AtPIN1 polarity with respect to auxin is not consistent across tissues (Bayer et al., 2009).
Indeed, AtPIN1 has several functions post-initiation that are not necessarily associated with convergent polarization patterns (Gälweiler et al., 1998; Scarpella et al., 2006). During the vegetative phase, AtPIN1 is not required for organ initiation, but the organs that do form are misplaced and have severe morphological and vascular defects similar to those observed upon pharmacological inhibition of auxin transport, suggesting an important role for AtPIN1 in post-initiation morphogenesis and vein patterning in leaves (Guenot et al., 2012; Sawchuk et al., 2013; Verna et al., 2015). Furthermore, in mature tissues, AtPIN1 is polarized basally (root-ward) in vascular-associated cells and is required for efficient long distance transport of auxin down the shoot in the polar auxin transport stream, and this has been proposed to play an important role in the regulation of shoot branching (Bennett et al., 2016; Bennett et al., 2006; Gälweiler et al., 1998; Shinohara et al., 2013). Mutations in other PIN family members in combination with atpin1 mutants suggest further functions in embryo development, root development and during plant growth responses to light and gravity (Leyser, 2005). Unfortunately, the myriad roles for AtPIN1 during plant development are genetically obscured by the severity of atpin1 organ initiation defects.
We previously showed that all sampled flowering plants outside of the Brassicacea family have a clade of PIN proteins sister to the PIN1 clade (The Sister-of-PIN1 or SoPIN1 clade), while Arabidopsis and other Brassicacea species have lost this clade (O’Connor et al., 2014). During organ initiation in the grass Brachypodium distachyon (Brachypodium) SoPIN1 is highly expressed in the epidermis, polarizes towards presumed auxin maxima, and forms convergent polarization patterns during the formation of new organs, suggesting a role in creating the auxin maxima required for organ initiation. In contrast, the PIN1 clade members in Brachypodium, PIN1a and PIN1b, are not highly expressed in the epidermis, orient away from presumed auxin maxima, and are primarily expressed during patterning in the sub-epidermal tissues. Thus, the combined expression domains and polarization behaviors of SoPIN1, PIN1a, and PIN1b in Brachypodium largely recapitulate those observed for AtPIN1 in Arabidopsis.
The localization and polarization of the Brachypodium SoPIN1 and PIN1 clades can be modeled with two different polarization modes with respect to auxin; SoPIN1 polarizes “up-the-gradient”, towards the neighboring cell with the highest auxin concentration, while PIN1a and PIN1b polarize “with-the-flux”, accumulating in the membrane with the highest auxin flux (O’Connor et al., 2014). Both polarization modes were previously applied to AtPIN1 in order to capture the switch in polarity observed during organ initiation and vein patterning, first orienting toward auxin maxima during convergence point formation, then orienting away from maxima during vein patterning below the epidermis (Bayer et al., 2009). These localization and modeling results suggest that in most angiosperm species the organ placement and vascular patterning functions attributed to AtPIN1 in Arabidopsis are split between the PIN1 and SoPIN1 clades, and that these two clades have different polarization properties with respect to auxin.
Here we present the functional analysis of both SoPIN1 and PIN1 protein clade members in Brachypodium, a species with the canonical two-clade family structure. We show that SoPIN1 and PIN1b have different functions during Brachypodium development, with SoPIN1 being required for organ initiation during the flowering phase, and PIN1b regulating stem elongation. Using heterologous expression in Arabidopsis, we show that the two proteins have different accumulation, polarization and transport behaviors that result in different functional properties independent of transcriptional context. In addition to elucidating several ways in which PIN family members can be functionally distinct, these results suggest that the Arabidopsis AtPIN1 protein represents an example of an evolutionary phenomenon the opposite of subfunctionalisation in which protein functions are amalgamated into a single protein rather than diversified amongst paralogs. AtPIN1 has a repertoire of roles, and associated polarization behaviors that are distributed among several clades of PIN proteins in most flowering plants.
Results
SoPIN1 and PIN1b have different functions in Brachypodium
During organ formation in the Brachypodium shoot, both SoPIN1 and PIN1b expression precede PIN1a, which only accumulates significantly at the site of vein formation after the organ begins to grow. In the earliest stages of initiation, prior to the periclinal cell divisions that are the hallmark of morphogenesis, SoPIN1 forms convergent polarization patterns around the presumed auxin maxima in the meristem epidermis, while PIN1b is expressed internally and orients away from the maxima (O’Connor et al., 2014). Because of their early expression and opposing polarization patterns, we focused on characterizing SoPIN1 and PIN1b as representatives of the SoPIN1 and PIN1 clades.
We targeted Brachypodium SoPIN1 and PIN1b with CRISPR and recovered loss-of-function mutants in both genes (see methods). Both sopin1-1 and pin1b-1 mutants have single base-pair lesions that result in frame-shifts and premature stop codons (Figure 1A). sopin1-1 mutants show severe organ initiation defects in the inflorescence remarkably similar to loss-of-function atpin1 mutants in Arabidopsis (Figure 1B-C, Figure 1 - supplement 1)(Okada et al., 1991). The indeterminate flowering shoots of the Brachypodium inflorescence, called spikelets (Figure 1B, inset), often fail to initiate in sopin1-1 despite sometimes clear definition of node vs internode tissue (n, and i, in Figure 1C). When spikelets do form, the spikelet meristems are often devoid of new organs (Arrows in Figure 1C inset).
In wildtype spikelet meristems, SoPIN1 convergence point formation is coincident with a decrease in the nuclear auxin response reporter protein DII-Venus (Brunoud et al., 2012) (DII) (Figure 1E), which functions in Brachypodium and is degraded in the presence of auxin in spikelet meristems (Figure 1 - supplement 2). In sopin1-1 meristems DII accumulation is uniformly high for long stretches of the epidermis, and the patterned reduction of DII both in the meristem epidermis and internally fails to occur, suggesting a failure to organize auxin maxima (Figure 1F arrow).
In contrast to the severe defects of sopin1-1, organ initiation in pin1b-1 mutants is unaffected (Figure 1D). However, pin1b-1 mutants show minor shoot twisting and increased internode (stem tissue) length, especially in the basal few internodes (Figure 1G, 1H). The longer internodes of pin1b-1 lead to an overall increase in plant height (Figure 1H). The internode defects of pin1-b are consistent with the abundant PIN1b accumulation observed in wildtype stem tissue (Figure 1I). The difference in phenotypes between sopin1-1 and pin1b-1 Brachypodium mutants suggests a functional distinction between the SoPIN1 and PIN1 clades, and indicates that while PIN1b is expendable for organ initiation, it is involved in the regulation of internode growth.
SoPIN1 and PIN1b accumulate differently in Arabidopsis
The difference between the sopin1-1 and pin1b-1 phenotypes in Brachypodium may be due to their different expression patterns and not necessarily to differences in polarization behavior as previously hypothesized (O’Connor et al., 2014). In order to determine the functional differences that exist between the proteins themselves, we expressed both Brachypodium proteins in wildtype (Columbia, Col-0) Arabidopsis under the control of a 3.5kb Arabidopsis PIN1 promoter fragment known to drive PIN1 expression sufficient to complement pin1 mutants (proAtPIN1) (Heisler et al., 2005). Remarkably, despite the loss of SoPIN1 from Arabidopsis, Brachypodium SoPIN1 created clear convergent polarization patterns around the sites of organ initiation in Arabidopsis inflorescence meristems (Figure 2I asterisk, 25 of 27 meristems from 4 independent transgenic events). SoPIN1 protein abundance was highest in the meristem epidermis and SoPIN1 convergence points were most clearly observed surrounding I2 and I1 primordia (Figure 2A). Immediately below the apex, SoPIN1 accumulated in an ill-defined ring shape within which the vascular bundles will form (Figure 2J, 15 of 23 meristems from 4 independent transgenic events).
In contrast, significant PIN1b accumulation was absent from the meristem epidermis in 19 of 29 meristems from 7 independent transgenic events. In the few meristems where PIN1b was significantly expressed in the epidermis, it did not show clear convergent polarization patterns, and its polarity was often unclear (Figure 2B). Within initiating organs PIN1b often localized to punctate vesicular bodies inside cells, not the cell membrane (Figure 2B arrow). The PIN1b expression level remained low just below the meristem apex, but in contrast to SoPIN1, PIN1b formed defined domains around the presumptive developing vascular bundles (Figure 2L). The lack of PIN1b protein in the meristem epidermis was not due to silencing of the transgene in these lines because we observed abundant PIN1b protein in the developing vasculature below the apex, even in plants where the meristem had no epidermal expression (Figure 2D) (8 samples from 4 events). In the same tissues SoPIN1 accumulated in both the vasculature and the epidermis (Figure 2C) (5 samples from 2 events).
In order to determine whether there were similar tissue-level differences in protein accumulation in mature tissues, we imaged SoPIN1 and PIN1b in the basal internode. Here, AtPIN1 normally accumulates in a highly polar manner in the root-ward (basal) plasma membranes of cambium (c) and xylem parenchyma (xp) vascular associated tissues (Bennett et al., 2016; Gälweiler et al., 1998). PIN1b accumulated in a similar pattern to AtPIN1 (Figure 2F, 2H. 10 samples from 5 events). In contrast, in addition to accumulating in the cambium and xylem parenchyma, SoPIN1 accumulated in the central pith tissue (p) (Figure 2E, 2G. 15 samples from 4 events). AtPIN1 is not normally observed in the pith (Bennett et al., 2016; Gälweiler et al., 1998). In the basal internode, both proteins showed the characteristic AtPIN1 root-ward polarization pattern regardless of tissue-level abundance (Figure 2K, 2M).
Taken together, these results show that even under the same transcriptional control SoPIN1 and PIN1b show distinct tissue-level accumulation patterns in Arabidopsis. While the overall behavior of the two Brachypodium proteins is similar to AtPIN1 in many tissues, there are behaviors unique to each. PIN1b fails to accumulate in the epidermal tissues where AtPIN1 and SoPIN1 remain high, whereas SoPIN1 accumulates in the pith tissue where AtPIN1 and PIN1b do not. The convergent polarization patterns of SoPIN1 and the vascular accumulation of PIN1b in Arabidopsis are remarkably similar to their native behaviors in Brachypodium (O’Connor et al., 2014), suggesting conserved mechanisms might control tissue level abundance between the two species.
SoPIN1 but not PIN1b can restore organ initiation and bulk auxin transport in AtPIN1 null mutants
To determine whether the observed differences in SoPIN1 and PIN1b polarization and accumulation have functional consequences in Arabidopsis, we used the proAtPIN1 driven SoPIN1 and PIN1b constructs to complement the Arabidopsis pin1-613 mutant (also known as pin1-7). The pin1-613 allele is a putative null T-DNA insertion mutant with severe organ initiation defects in the inflorescence (Bennett et al., 2006; Smith et al., 2006; Zourelidou et al., 2014). Given that epidermal PIN1 function is important for organ initiation (Bilsborough et al., 2011; Kierzkowski et al., 2013), as expected only SoPIN1 was able to complement the pin1-613 mutation and mediate organ initiation (Figure 3A) (3 out of 6 independent transgenic events showed complementation). However, phenotypic complementation of pin1-613 by SoPIN1 was incomplete, and mature plants showed a variety of phenotypic defects (Figure 3A, Figure 3 - supplement 1). Most notably, each flower produced more sepals and petals than wild-type, but almost no stamens (Figure 3B, 3C, Figure 3 - supplement 2). SoPIN1 complemented pin1-613 plants were thus sterile. We wondered if these phenotypes could be explained if SoPIN1 had poor auxin transport function in Arabidopsis. However, SoPIN1 restored wild-type levels of bulk auxin transport to pin1-613 basal internodes (Figure 3D). Thus SoPIN1 is at least in part functionally capable of initiating organs and mediating root-ward auxin transport in the stem, but it is not functionally identical to AtPIN1 under the same promoter.
In SoPIN1 complemented pin1-613 mutants, SoPIN1 accumulation increased in the meristem epidermis relative to wild-type or heterozygous plants, but the pronounced convergent polarization patterns observed in the WT background were less clear (Figure 4A, Figure 4 - supplement 1) (16 of 16 meristems). SoPIN1 complemented meristems showed a variety of phyllotactic defects and had highly variable morphologies (Figure 4 - supplement 1) (16 of 16 meristems). Similar to the pattern observed in the wild-type background, sub-epidermal SoPIN1 in pin1-613 mutants accumulated in a loosely defined ring within which individual vein traces were difficult to discern (Figure 4I) (13 of 16 meristems). In the mature tissues, SoPIN1 accumulated in the epidermis, vasculature, and mature pith tissues similar to the wild-type background (Figure 4C, 4E, 4G).
In contrast to SoPIN1, PIN1b-expressing pin1-613 plants had pin-formed inflorescences that were indistinguishable from pin1-613 alone (Figure 3A) (all 7 events failed to complement). The lack of complementation mediated by PIN1b was not caused by silencing or low expression level because abundant PIN1b signal was observed in pin1-613 meristems (23 of 26 pin1-613 meristems from 7 events). In most PIN1b expressing pin1-613 samples, expression increased in the epidermis relative to wildtype, forming a ring-shaped domain around the meristem apex (Figure 4B, 4D arrow, Figure 4 - supplement 2) (14 of 19 meristems from 6 events). Unlike in the wildtype background, PIN1b in the epidermis of pin1-613 meristems was more consistently targeted to the membrane and polar (Figure 4K). However, even with this elevated polar expression in the meristem epidermis, PIN1b was unable to mediate organ initiation in pin1-613. Below the apex, PIN1b was polarized root-ward in pin1-613 meristems (Figure 4J), forming defined traces associated with the vasculature (Figure 4F, 4L). In the basal stem of pin1-613 mutants PIN1b accumulated in a pattern similar to wild-type, although the arrangement of vascular bundles was irregular (Figure 4H). Remarkably, despite clear polar PIN1b expression in pin1-613 mutant stems (Figure 4M), PIN1b was unable to rescue bulk auxin transport in this tissue (Figure 3D).
Although PIN1b was incapable of supporting organ formation or mediating bulk transport in pin1-613, when an auxin maximum was created artificially by addition of lanolin paste infused with IAA, PIN1b epidermal accumulation increased during the initiation of the resultant primordia (Figure 4 - supplement 3) (4 of 6 samples from 2 independent transgenic events). Thus, in the absence of AtPIN1, PIN1b accumulation in the epidermis is still auxin responsive and capable polar localization in pin1-613 meristem tissue, but it is not able to mediate organ initiation itself. These results demonstrate that when expressed in Arabidopsis, there is a clear functional separation between SoPIN1 and PIN1b independent of transcriptional control.
SoPIN1 and PIN1b show different behaviors when expressed in the meristem epidermis
Epidermal-specific AtPIN1 expression is sufficient to rescue organ initiation in atpin1 mutants (Bilsborough et al., 2011). In order to drive increased PIN1b expression in the epidermis, and to help reduce transgene position-effect variation of expression level, we utilized a two-component expression system in the Landsberg erecta (Ler) background to drive SoPIN1 and PIN1b under the control of the epidermis-enriched Arabidopsis ML1 promoter (Hereafter designated proAtML1>>) (Lenhard, 2003; Sessions et al., 2002). Under the control of proAtML1 we achieved consistently high epidermal accumulation of both SoPIN1 and PIN1b, but similar to the proAtPIN1 driven localization described above, only SoPIN1 showed clear convergent polarization patterns around the sites of organ initiation (Figure 5A–5D, Figure 5 supplement 1 and 2) (11 of 11 meristems). Despite consistently high epidermal expression with this system, PIN1b polarity remained difficult to determine, and in many cells the abundance of protein on the membrane remained low (Figure 5D) (13 of 13 meristems). Instead, PIN1b accumulated in intracellular bodies, especially in the cells of the apical dome and the central domain of initiating organs (Figure 5B, 5D arrow). PIN1b abundance and polarity was highest at the boundaries of lateral organs (Figure 5 - supplement 2). Thus SoPIN1 and PIN1b show consistent behaviors in the meristem epidermis when expressed under either proAtPIN1 or proAtML1. However, despite increased PIN1b expression under proAtML1, and a resulting increase in protein accumulation in the apex, PIN1b was still unable to form convergent polarization patterns in wildtype plants.
Both SoPIN1 and PIN1b can rescue the Arabidopsis pin1-4 mutation when expressed in the meristem epidermis
In order to determine whether the increased PIN1b abundance in the meristem epidermis achieved by the proAtML1 two-component system had functional consequences, we crossed these transgenes into pin1-4 mutant. The pin1-4 allele is in the Landsberg erecta background and has a single P579 to L amino acid change in the second-to-last transmembrane domain of AtPIN1 (Bennett et al., 1995), but the phenotype is similarly severe to pin1-613 (Figure 6A). Remarkably, both SoPIN1 and PIN1b driven by proAtML1 were able to rescue the organ formation defects of pin1-4 (Figure 6A). In contrast to the SoPIN1-mediated complementation of pin1-613 described above, both SoPIN1 and PIN1b-complemented pin1-4 plants made WT flowers that produced seed (Figure 6 - supplement 1). In addition, both proAtML1 SoPIN1 and PIN1b expressing pin1-4 lines were able to rescue bulk auxin transport in the basal internode, although PIN1b was less effective than SoPIN1 (Figure 6B). Compared to wildtype and SoPIN1-complemented plants, PIN1b-complemented pin1-4 plants showed a significant increase in stem diameter (Figure 6C).
SoPIN1-complemented pin1-4 meristems were slightly smaller than wildtype, and in rare cases showed defects in phyllotaxy (Figure 5 - supplement 1), but the protein localization was similar to the pattern observed in the WT background, with clear convergent polarization around initiating organs (Figure 5E, 5G) (10 of 10 meristems). In contrast, compared to the WT background, PIN1b localization in pin1-4 was dramatically altered (Compare Figure 5B with Figure 5F). Most obvious was an increase in membrane targeted PIN1b and a corresponding reduction in intracellular PIN1b (Figure 5H). PIN1b polarity in the pin1-4 background was more apparent than in wildtype, and convergent polarization patterns clearly marked incipient organs (Figure 5H) (10 of 10 meristems). PIN1b-complemented meristems accumulated less PIN protein in the apical dome compared to SoPIN1-complemented meristems, and the meristems were larger (Figure 5 - supplement 2).
In the basal internode, both PINs had similar accumulation patterns in the outer few cell layers (Figure 5I-J arrows), and both showed basal polarization in the epidermis (Figure 5K-L arrows). Despite this expression domain being drastically different than the wildtype vascular-associated pattern of AtPIN1 (Bennett et al., 2006; Gälweiler et al., 1998), expression in these few cortex layers and epidermis was apparently sufficient to drive wildtype levels of rootward bulk auxin transport in pin1-4 (Figure 6B). Thus while both proteins can complement the pin1-4 organ initiation phenotype, the SoPIN1 and PIN1b complemented lines have differing localization patterns, slightly different auxin transport properties, and minor differences in meristem and mature plant morphologies, suggesting once again that SoPIN1 and PIN1b are not functionally identical.
Discussion
SoPIN1 and PIN1b have different functions in Brachypodium
During spikelet development in Brachypodium SoPIN1 forms convergent polarization patterns surrounding the sites of organ initiation and strong expression of the auxin response reporter DR5 (O’Connor et al., 2014). We provide additional evidence here that SoPIN1 polarizes towards sites of high auxin concentration by showing that a DII minimum occurs at SoPIN1 convergence points. In sopin1 mutants the reduction of DII does not occur, suggesting that SoPIN1 functions to concentrate auxin at epidermal maxima, and similar to Arabidopsis, this is required for organ initiation in the inflorescence. The specificity of SoPIN1 for the outer tissues in Brachypodium provides further support for the idea that maxima formation is necessary for organ initiation, and that this is primarily mediated by convergent PIN in the meristem epidermis (Bhatia et al., 2016; Jönsson et al., 2006; Kierzkowski et al., 2013; Smith et al., 2006).
SoPIN1 clade mutants have been reported in the legume Medicago truncatula and in tomato (Solanum lycopersicum), and these mutants show pleiotropic phenotypes involving phyllotaxy, organ initiation, inflorescence branching, leaf serrations, and leaf compounding, but they do not form barren pin meristems (Martinez et al., 2016; Zhou et al., 2011). These wider morphogenetic events also involve epidermal PIN convergence points and associated auxin maxima, suggesting a general role for SoPIN1 clade members in generating such maxima (Barkoulas et al., 2008; Bilsborough et al., 2011). But the lack of barren pin-formed meristems in these mutants may suggest that different species are variably dependent on SoPIN1-generated maxima for organ initiation.
In contrast to sopin1, loss of PIN1b function has no clear organ initiation defects, despite being expressed in developing organs (O’Connor et al., 2014). Given that auxin drainage is thought necessary for proper organ size and placement (Bhatia et al., 2016; Deb et al., 2015), the lack of an organ initiation phenotype in pin1b is surprising. However, it is possible that PIN1a performs some of this function, and double mutants are needed to address the redundancy of PIN1a and PIN1b during organ initiation.
The increased internode elongation in pin1b mutants provides new genetic tractability to address how PINs regulate tissue growth in the shoot independent of organ initiation. Grasses contain intercalary meristems, bands of indeterminate tissue separated from the apical meristem that are responsible for internode growth after organ initiation. PIN1b expression around this meristematic tissue (Figure 1I) suggests that PIN1b may regulate growth by influencing auxin distribution in this meristem. This is consistent with evidence that loss of the ABCB1 auxin exporter in maize results in dwarfism associated with reduced activity of intercalary meristems (Knöller et al., 2010). How PIN1b alters auxin dynamics to control internode growth will be an important direction for future research.
The properties that define PIN behavior and function
Membrane accumulation. Because of their differing phenotypes in Brachypodium, we used heterologous expression of SoPIN1 and PIN1b in Arabidopsis to explore the ways in which different PIN family members may have different properties posttranscription (Summarized in Figure 7). When expressed in the meristem epidermis in wild-type Arabidopsis, SoPIN1 is localized to the membrane in most cells while PIN1b often accumulates internally (Compare Figure 5C and D). Thus, with the same transcriptional control different PINs can vary in the degree to which, after protein production, they accumulate at the plasma membrane. The differential membrane targeting of PIN1b and SoPIN1 is a tissue-specific phenomenon however, because unlike in the epidermis, in the basal internode both PINs accumulate at the plasma membrane (Figure 2K, 2M). The regulation of PIN plasma membrane polar targeting and endocytic recycling has been an important avenue for understanding PIN function and general membrane protein biology (Luschnig and Vert, 2014). Our results provide further evidence that at least some of the signals governing membrane accumulation are inherent in, and vary between different PIN family members.
Tissue accumulation. Under the same transcriptional control SoPIN1 and PIN1b show different tissue-level accumulation patterns in Arabidopsis. In wildtype plants proAtPIN1-expressed PIN1b shows less overall accumulation in the epidermis compared to SoPIN1 (Compare Figure 2C and 2D). The punctate PIN1b signal in the meristem epidermis may be PIN1b protein being actively targeted for degradation as has been shown for PIN2 in the root (Abas et al., 2006). In contrast, SoPIN1 is abundant in the epidermis and accumulates in an expanded expression domain in the sub-epidermal meristem tissues (Compare Figure 2J and 2L). Also, in the basal internode SoPIN1 accumulates in the pith tissue where AtPIN1 and PIN1b do not (Figure 2E, 2G). The presence of SoPIN1 in the pith is not easy to explain because AtPIN1 shows no protein accumulation in this tissue. It is possible however that the sub-epidermal SoPIN1 protein accumulation in the meristem persists into the mature internode.
In Arabidopsis, endogenous PIN family members show a degree of cross-regulation where loss-of-function mutations in one PIN family member result in ectopic accumulation of a different PIN in a compensatory pattern (Blilou et al., 2005; Paponov et al., 2005; Vieten et al., 2005). We observed similar behavior in the pin1-613 null background where SoPIN1 and PIN1b accumulation in the meristem epidermis was increased in the absence of AtPIN1 (Figure 4 - supplements 1 and 2). However, we did not observe the same cross-regulation in the pin1-4 background where SoPIN1 and PIN1b tissue-level accumulation seemed similar between pin1-4 mutant and wild-type meristems (Figure 5 - supplements 1 and 2). These variable tissue-level abundances, and PIN cross-regulation behaviors highlight the overall redundancy of some PIN behaviors, and further demonstrate the importance of PIN post-transcriptional regulation for controlling PIN abundance.
Transport activity. In Arabidopsis, phosphorylation of PINs by several different families of protein kinases is necessary for efficient auxin transport (Barbosa et al., 2014; Jia et al., 2016; Willige et al., 2013; Zourelidou et al., 2014). The necessity for PIN activation by phosphorylation may explain the inability of PIN1b to mediate bulk auxin transport in the basal internode of pin1-613 plants despite being expressed, accumulating at the membrane, and being polarized root-ward in this tissue (Figure 4M). It is possible that in the proAtPIN1 domain PIN1b does not interact with the appropriate activating kinase, and it is thus inactive. Indeed, a partially un-phosphorylatable form of AtPIN1 fails to complement fully the bulk auxin transport defect of pin1-613 mutants in the basal internode (Zourelidou et al., 2014). However, when expressed using proAtML1, PIN1b expression in the outer tissue layers of the basal internode appears sufficient to mediate bulk auxin transport in pin1-4 (Figure 6B), suggesting that PIN1b activity may be tissue dependent, perhaps because of the differing expression domains of activating kinases (Zourelidou et al., 2014). Indeed, Arabidopsis PIN4 and PIN7 are present in the proAtML1 domain (Bennett et al., 2016), making it conceivable that these PINs are the normal targets of activating kinases in this tissue. Regardless, the behavior of PIN1b in pin1-613 Arabidopsis provides a clear indication that even once a PIN has accumulated at the cell membrane in a tissue it may not be active.
Interaction. A particularly striking result is the ability of PIN1b to form convergent polarization patterns and mediate organ initiation in the pin1-4 missense mutant background when it is unable to do so in the null pin1-613 background. The strong influence of pin1-4 on PIN1b membrane targeting and polarity in the meristem epidermis (Compare Figures 6D and 6H) suggests that PIN1b may be cooperating with a partially functional pin1-4 protein and together they recapitulate the functions of wildtype AtPIN1. The presence of some pin1-4 function is supported by the result that SoPIN1 complementation of the null pin1-613 allele is partial, and because of flower defects the plants are sterile (Figure 3B, 3C), while complementation of pin1-4 is complete and flowers are phenotypically normal and set seed (Figure 6 - supplement 1). Consistent with these different complementation phenotypes, SoPIN1 convergent patterns are more evident in the presence of pin1-4 than they are in the null pin1-613 background (Compare 4A and 5E), further evidence for partial pin1-4 function. If PIN1b is indeed inactive in null pin1-613 mutants as we hypothesized above, then it is possible pin1-4 facilitates the interaction of PIN1b with the appropriate activating kinase, and this allows PIN1b to perform organ initiation. Alternatively, pin1-4 may provide polarity information that PIN1b lacks, and even though pin1-4 is non-functional, it is able to target or stabilize PIN1b on the appropriate membrane to mediate convergent polarization patterns and organ initiation. However, pin1-4 interaction with PIN1b cannot explain the ability of PIN1b to rescue bulk transport in the basal internodes of pin1-4 mutants, because the two proteins presumably do not overlap in this tissue. In this case the necessary interaction between PIN1b and pin1-4 may be set up early during organ initiation, and protein modifications propagated to the basal internode. Direct PIN interaction has so far never been shown, but if one PIN type can convey polarity or activity information to another through direct interaction this may be important for understanding auxin transport in tissues where multiple PINs overlap, such as in the root meristem (Blilou et al., 2005), or in Brachypodium where multiple PINs are present during organ initiation in the shoot (O’Connor et al., 2014).
Polarity. We previously showed that the polarization dynamics of SoPIN1, PIN1a, and PIN1b in Brachypodium could be modeled by assigning two different polarization modes to the SoPIN1 and PIN1 clades (O’Connor et al., 2014). In the model, SoPIN1 orients toward the adjacent cell with the highest auxin concentration, thus transporting auxin up the concentration gradient and providing a positive feedback to concentrate auxin into local maxima. In contrast, in the model PIN1a and PIN1b proteins are allocated in proportion to auxin flux, thus providing a positive feedback where flux through the tissue is amplified by the allocation of PIN1a/b in the direction of that flux. The assignment of two different polarization modes was previously used to describe the behavior of AtPIN1 during organ placement and vein patterning utilizing an auxin-concentration based switching mechanism between the up-the-gradient (UTG) and with-the-flux (WTF) polarization modes (Bayer et al., 2009). However, it has also been suggested that a flux-based mechanism alone can account for both convergence points and vein patterning (Abley et al., 2016; Stoma et al., 2008).
Despite evidence that PIN polarization is dependent on localized auxin signaling (Bhatia et al., 2016), there are still no known mechanisms for direct sensing of intercellular auxin gradients or flux across membranes. However, the sopin1 and pin1b phenotypes in Brachypodium are consistent with different polarization modes. SoPIN1 is required for organ initiation and the formation of auxin maxima in Brachypodium, which is primarily modeled using UTG polarization (Bayer et al., 2009; Jönsson et al., 2006; Smith et al., 2006). On the other hand, pin1b mutant plants do not show organ initiation defects, but rather only have internode elongation defects, a tissue where WTF models have been used to explain PIN dynamics and measured auxin transport kinetics during vein patterning and the regulation of branch outgrowth (Bayer et al., 2009; Bennett et al., 2016; Mitchison, 1980; Mitchison et al., 1981; Prusinkiewicz et al., 2009).
In wild-type Brachypodium the SoPIN1 and PIN1b expression domains are almost entirely mutually exclusive (O’Connor et al., 2014), making it possible that the observed polarization differences between the two clades are due to expression context and not functional differences between the proteins themselves. More specifically, perhaps an UTG mechanism dominates the epidermis while a WTF mechanism is utilized in the internal tissues, and different PINs interact equally with these context-dependent mechanisms. Our heterologous expression studies do not exclusively support context-dependent or protein-dependent mechanisms for SoPIN1 and PIN1b polarity. It is clear that alone only SoPIN1 and AtPIN1 show the convergent polarization patterns associated with UTG polarization, and alone only SoPIN1 and AtPIN1 are thus able to mediate organ initiation, while PIN1b cannot. On the other hand, all three PINs are capable of root-ward polarization in the basal internode tissue. The results presented here do not demonstrate whether within a single cell PIN1b and SoPIN1 would orient differently with respect to auxin as might be expected for the dual polarization model (O’Connor et al., 2014). However such context-independent polarization behavior was previously observed for PIN1 and PIN2 in the root where both PINs can polarize in opposing directions within a single cell type when expressed in the PIN2 domain (Kleine-Vehn et al., 2008; Wisniewska et al., 2006).
Outlook
In total, our Brachypodium mutant phenotypes and heterologous expression results point to multiple levels at which PIN family members can be functionally distinct. Differential membrane targeting, tissue level accumulation, transport activity, indirect or direct interaction, and the resultant polarity may all contribute to the dynamics of PIN action during plant development. In most flowering plants two PIN clades, SoPIN1 and PIN1, with differing properties post-transcription mediate auxin transport in the shoot, but these properties are seemingly combined into AtPIN1 in Arabidopsis and other Brassicaceae species. Because PIN1b is unable to mediate organ initiation while AtPIN1 can, and these two PINs are both members of the same clade, AtPIN1 may have gained the ability to form convergent polarization patterns and mediate organ initiation after, or coincident with, the loss of the SoPIN1 clade. Indeed, when comparing Brassicaceae PIN1 proteins against a broad sampling of other angiosperm PIN1 proteins, the Brassicaceae PIN1 proteins have several divergent protein domains (Figure 7 - supplement 1), suggesting possible neofunctionalization within the Brassicacea family. Alternatively, an expansion of the PIN3,4,7 clade is also characteristic of Brassicacea species (Bennett et al., 2014; O’Connor et al., 2014), making it possible duplicated members of this clade buffered the loss of SoPIN1. However, there is no indication that PIN3,4,7 have a role in organ initiation in the inflorescence (Guenot et al., 2012). Regardless, we believe the combination of SoPIN1 and PIN1 characteristics into AtPIN1 coincident with the loss of the SoPIN1 clade represents a form of reverse-subfunctionalization, the combination of functions originally split between homologs into a single protein after gene loss. It is not surprising that PINs may be particularly amenable to this kind of functional evolution because, as described above, there are several post-transcriptional regulatory steps that ultimately combine to control PIN function in plants. The output of auxin transport is the sum of a vast network of post-transcriptional interactions that all act to regulate auxin transport itself, and this gives the system plasticity during development, and perhaps also over evolutionary time.
Materials and Methods
sopin1-1 and pin1b-1 creation with CRISPR
SoPIN1 (Bradi4g26300) and PIN1b (Bradi3g59520) were targeted with CRISPR using vectors developed for rice (Miao et al., 2013). CRISPR constructs were transformed into Brachypodium inbred line Bd21-3 using previously published methods (Bragg et al., 2015).
sopin1-1 CRISPR
The SoPIN1 guide was AGGCTGTCGTACGAGGAGT. This guide was shorter than the typical 20bp in an effort to provide greater target specificity for SoPIN1 (Fu et al., 2014). In the T0 regenerated plants, 5 out of 9 independent transgenic events showed severe organ initiation defects, and all 5 contained lesions in the SoPIN1 CRIPSR target site. Unfortunately, only one of the events with a T0 phenotype set seed. In the T1 progeny of this event only those individuals that contained the CRISPR transgene showed lesions in the SoPIN1 CRISPR target site, and these plants showed the sopin1 phenotype and thus failed to set seed, suggesting active editing by the SoPIN1 CRISPR transgene in this event.
Not all events showed such efficient editing however, and we identified an independent T1 family where a C insertion in the SoPIN1 CRISPR target site co-segregated with the barren inflorescence phenotype. We designated this allele, which causes a premature stop codon before the end of the third exon codon 739 base pairs downstream from the target site, sopin1-1. We backcrossed a heterozygous sopin1-1 plant to the Bd21-3 parental line and all F1 progeny (N=4) were wildtype. In the F2 generation, the sopin1-1 lesion co-segregated with the barren inflorescence phenotype (N=60: 32 het, 18 homo,10 wt). Amongst these plants, 16 did not have the Cas9 transgene, and the barren inflorescence phenotype still co-segregated with the sopin1-1 lesion (N=16: 8 het, 3 homo, 5 wt). We crossed the T1 sopin1-1 heterozygous plant with a line homozygous for the SoPIN1-Citrine genomic reporter line (O’Connor et al., 2014). In the F2 we identified families homozygous for sopin1-1 but segregating for the SoPIN1-Citrine transgene. Only individuals that lacked the SoPIN1-Citrine transgene showed a sopin1-1 phenotype, while those that contained the SoPIN1-Citrine transgene made spikelets and set seed. This complementation was independent of the presence of Cas9.
pin1b-1 CRISPR
The PIN1b guide was AGGGCAAGTACCAGATCC. We identified a single plant from the regenerating T0 PIN1b CRISPR population that had longer basal internodes and twisted leaves. This plant was homozygous for an A deletion in the PIN1b CRISPR target site causing a premature stop in the second exon 502 base pairs downstream, here designated pin1b-1. All T1 progeny showed the pin1b phenotype and were homozygous for the pin1b-1 lesion. We backcrossed these T1 plants to Bd21-3 and all F1 progeny had a wild-type phenotype (N=11). In the F2, the pin1b phenotype co-segregated with the pin1b-1 lesion (N= 215, 91 het, 39 homo, 26 wt). Amongst these plants, 24 did not have the Cas9 transgene, and the pin1b phenotype co-segregated perfectly with the pin1b-1 lesion (N=24: 10 het, 6 homo, 8 wt).
Brachypodium Reporter Constructs
All constructs were cloned using Multi-site Gateway (Invitrogen) and were transformed into Brachypodium Bd21-3 using previously published methods (Bragg et al., 2015). For pZmUbi::DII-Venus, we first cloned the maize ubiquitin promoter into pDONR P4-P1R (Primer IDs 1-2 Table 1) and this was subsequently recombined with pDONR 221 containing Arabidopsis DII and pDONR P2R-P3 containing VENUS-N7 (Brunoud et al., 2012) into the Multi-site Gateway binary vector pH7m34GW (http://gateway.psb.ugent.be/). In the T3 generation, degradation of DII-Venus in the presence of auxin was validated by treating excised Brachypodium spikelet meristems with 1 μM 1-naphthaleneacetic acid (NAA) or mock treatment in 70% ethanol, and imaging every 30 min (Figure 1 - figure supplement 2).
For SoPIN1-Cerulean, the promoter plus 5’ coding pDONR-P4-P1R and 3’ coding plus downstream pDONR-P2R-P3 fragments from (O’Connor et al., 2014) were used. Maize codon-optimized Cerulean florescent protein, courtesy of David Jackson, was amplified with 5x Ala linkers and cloned into pENTR/D-TOPO. These three fragments were then recombined into pH7m34GW.
Arabidopsis Reporter Constructs
All constructs were cloned using Multi-site Gateway (Invitrogen) and transformed using standard floral dip. For proAtPIN1 complementation, a 3.5kb Arabidopsis PIN1 promoter region was amplified from a genomic clone previously reported to complement the pin1 (Heisler et al., 2005) and cloned into Gateway vector pDONR P4-P1R (Primer IDs 3-4 Table 1). For each Brachypodium PIN-Citrine fusion construct, the entire PIN coding region, including the Citrine insertion, was amplified from the previously published reporter constructs (O’Connor et al., 2014) and cloned into pENTR/D-TOPO (Primer IDs 5-8 Table 1). The proAtPIN1 pDONR P4-P1R and PIN coding region pENTR/D-TOPO vectors were then recombined into Gateway binary vector pH7m24GW (http://gateway.psb.ugent.be/) and transformed by floral dip into both Col-0 and plants heterozygous for pin1-613 (also known as pin1-7, SALK_047613) (Bennett et al., 2006; Smith et al., 2006). Complementation was assessed in the T3 generation, and all plants were genotyped for both the pin1-613 mutation (Primer IDs 9-11 Table 1) and for presence of the PIN transgene (Primer IDs 12-14 Table 1).
For the proAtML1 lines the PIN coding regions with Citrine insertion pENTR/D-TOPO Gateway vectors were recombined downstream of the two-component OP promoter in vector pMoA34-OP (Moore et al., 1998) and then transformed into the proAtML1 driver line in the Landsberg erecta background (Lenhard, 2003). Lines homozygous for both the proAtML1 driver and OP::PIN were crossed to het pin1-4 and complementation was assessed in the F2 and F3 generations. All complemented plants were genotyped for pin1-4 (Primer IDs 15-16 Table 1 with AciI digestion), the Brachypodium PINs (Primer IDs 12-14 Table 1), and the presence of the ML1 driver transgene (Primer IDs 17-18 Table 1).
Confocal Imaging
All confocal images were captured on a Zeiss 780 laser scanning confocal using 514nm excitation and a W Plan-Apochromat 20x magnification 1.0 numerical aperture objective. Detection wavelengths: 517-570nm for Citrine tagged PINs, 631-717nm for Propidium Iodide, and 646-726 for chlorophyll A auto-fluorescence. The pinhole was set to 1 airy unit for all meristem stacks and details of sub-epidermal polarization, but was open to the maximum setting for tiled longitudinal and cross sections of the basal internode (Figures 2C-H, 4C-H and 5I-J). Detection gain and laser power were varied according to signal strength unless direct comparisons between genotypes were made as indicated in figure legends.
Auxin Transport Assays
Auxin transport assays were carried out as described in (Crawford et al., 2010). Briefly, 17 mm long basal internodes were excised and the apical end submerged in 30 μl Arabidopsis salts (ATS) without sucrose (pH = 5.6) containing 1 μM 14C-IAA (American Radiolabeled Chemicals). After 6 hours incubation, the basal 5 mm segment was excised, cut in half, and shaken overnight at 400 RPM in 200 μl scintillation liquid prior to scintillation counting. 10μM N-1-Naphthylphthalamic Acid (NPA), an auxin transport inhibitor, was added prior to incubation for negative controls.
Acknowledgments
Thanks to Fabrizio Ticchiarelli for genotyping help, Martin van Rongen for assistance with transport assays and pin1-613 oligos, Tom Bennett for proAtPIN1 oligos, Marcus Heisler for AtPIN1-GFP construct and pin1-4 genotyping assistance, Teva Vernoux for DII plasmids, David Jackson for maize codon-optimized Cerulean, and to all the members of the Leyser lab. Thanks also to Graeme Mitchison, Katie Abley and Pau Formosa-Jordan for helpful comments on the manuscript.