Naphthylphthalamic acid and the mechanism of polar auxin transport

Abstract

Our current understanding of how plants move auxin through their tissues is largely built on the use of polar auxin transporter inhibitors. Although the most important proteins that mediate auxin transport and its regulation have probably all been identified and the mapping of their interactions is well underway, mechanistically we are still surprisingly far away from understanding how auxin is transported. Such an understanding will only emerge after new data are placed in the context of the wealth of physiological data on which they are founded. This review will look back over the use of a key inhibitor called naphthylphthalamic acid (NPA) and outline its contribution to our understanding of the molecular mechanisms of polar auxin transport, before proceeding to speculate on how its use is likely still to be informative.

ABCB, auxin, flavonol, indole-3-acetic acid, naphthylphthalamic acid, NPA, PIN, polar auxin transport, TIBA

Introduction

In 1957, a new approach to the functional characterization of polar auxin transport was proposed: the application of synthetic inhibitors. This approach was fueled by a frustration that, although 25 years had passed since Went and van der Weij first characterized the polar transport of auxin, its mechanism remained unknown (Niedergang-Kamien and Leopold, 1957). Since this time, several diverse inhibitors have been characterized and used to affect polar transport rates (Klima et al., 2016). Undoubtedly, significant progress has been made in the subsequent 60 years, especially into understanding the diverse developmental processes that are coordinated by polar auxin transport. Nevertheless a convincing argument can be made that a proper understanding of the underlying mechanisms of auxin transport remains elusive. In this review we will focus on naphthylphthalamic acid (NPA), one of the most popular auxin transport inhibitors, and highlight some of the most pressing open questions that have arisen from its use. These questions are still surprisingly fundamental: for example, previously characterized binding sites have not unequivocally been assigned to specific proteins, the proximity of NPA binding to the transporter is unclear, characterization of the regulatory networks targeted by NPA is incomplete, and the extent to which endogenous inhibitors and NPA can be considered functionally equivalent is unknown.

A brief history of NPA

By 1958, it was already known that phthalamic acid derivatives have severe effects on plant development. Indeed, NPA was singled out as a compound that caused unique and severe morphological effects as early as 1949 (Hoffmann and Smith, 1949). The ability of NPA to disrupt tropic growth prompted its identification as an inhibitor of polar auxin transport, but not of growth (Mentzer et al., 1950; Ching et al., 1956; Morgan and Soding, 1958). Later, the effect of NPA and 2,3,5-triiodobenzoic acid (TIBA) were characterized in maize coleoptile segments (Hertel and Leopold, 1962), but it was the demonstration that NPA could inhibit the efflux and thus stimulate net uptake of radiolabeled indole-3-acetic acid (IAA) into vesicle preparations that firmly placed NPA in the plant physiologist’s toolkit (Hertel et al., 1983). NPA is classified as a phytotropin, a wider class of chemicals that also inhibit tropic responses in stems and have a benzoic acid moiety substituted at the ortho position by a bridging group connected to a second aryl group; other important polar auxin transport inhibitors such as TIBA and morphactin are not considered to be phytotropins (Rubery, 1990). Phytotropins are all thought to bind to the same receptors, through which they may elicit their physiological responses (Brown et al., 1973; Geissler et al., 1975; Katekar and Geissler, 1977; Katekar et al., 1987). It is important to note that one way in which different auxins can be functionally distinguished from one other is that NPA does not inhibit their transport to the same degree. For example, the accumulation of [³H]2,4-dichlorophenoxyacetic acid (2,4-D) in tobacco suspension cells is not increased by NPA, whereas the accumulation of [³H]naphthylacetic acid (NAA) is more strongly stimulated than that of [³H]indole acetic acid (IAA), for which the IC₅₀ of NPA was estimated to be 3 µM (Delbarre et al., 1995).

Subsequently, diverse physiological effects were attributed to NPA-sensitive polar auxin transport, such as embryo development (Schiavone and Cooke, 1987), lateral root development (Casimiro et al., 2001), leaf vein patterning (Mattsson et al., 1999), apical dominance (Ruegger et al., 1997), phyllotaxy and shoot apical meristem integrity (Okada et al., 1991; Reinhardt et al., 2000), and tillering and adventitious root formation in monocotyledons (Scanlon, 2003; Zhou et al., 2003; Xu et al., 2005). Most significantly, it was Okada et al. (1991) who provided a new opening for the study of polar auxin transport by describing a genetic component: a mutant allele they termed pinformed1 (pin1) which was closely associated with the NPA-treated phenotype in Arabidopsis. Here, not only did plants have severely reduced rates of polar auxin transport through stem sections but all plants grown on NPA phenocopied pin1. This was in marked contrast to all other polar auxin transport inhibitors that were tested. The second most effective was morphactin, 9-hydroxyfluorene-9-carboxylic acid, also sometimes known as HCFA, with just over half of plants phenocopying pin1. Although TIBA has been used subsequently to induce the pin1 phenotype, it does so relatively inefficiently, with about 10% of plants phenocopying pin1 (Okada et al., 1991). Seven years later, the PIN1 gene was cloned and identified as a membrane transporter of the major facilitator superfamily (Gälweiler et al., 1998).

The striking polar localization of PIN1 (Gälweiler et al., 1998) had been previously predicted using a range of theoretical approaches (Goldsmith, 1977). However, theoretical work had also started to undermine the case for a strongly polarized auxin efflux carrier. An early model of auxin transport predicted that observed rates and characteristics of auxin transport could be simulated with only a 1–10% difference in efflux rate from either end of the cell. In this case, strong directional transport would be supported by the cumulative action over a file of many consecutive cells (Leopold and Hall, 1966). This work raised the possibility that PIN1 itself need not be the sole transporter but instead could provide directionality to auxin flow by increasing the underlying membrane permeability to auxin. Indeed, it is still not known whether PIN1 transports auxin directly in the plant or acts as a positive regulatory subunit of a larger auxin efflux protein complex. If the latter were to be the case, one would not be able to safely assume that NPA acted directly on PIN1 at all. Indeed, it has even been proposed that NPA may have effects that are independent of auxin transport altogether (Hossel et al., 2005). Thus the stage was set for a new phase of auxin transport research that over the last 20 years has seen the focus of NPA action move away from the PINs and towards other proteins.

Where does NPA bind?

Significant effort has been invested in the biochemical characterization of the NPA binding site. However, the findings of these studies have made it hard to form a coherent picture of the NPA binding landscape. Although TIBA, morphactin, and NPA all inhibit polar auxin transport at similar concentrations, with typical working concentrations at around 10 µM, they do not bind to isolated plant membranes with comparable affinity (Sussman and Goldsmith, 1981). Specific binding can be saturated using far lower concentrations when using NPA, suggesting that the site with the highest affinity for NPA does not influence polar auxin transport. Morphactins also bind to the NPA site (Thomson and Leopold, 1974), which is on the plasma membrane (Ray, 1977), either as an integral membrane protein (Bernasconi et al., 1996) or a peripherally associated protein (Cox and Muday, 1994), which may interact with the actin cytoskeleton (Butler et al., 1998). There may also be a single NPA binding site on the plasma membrane (Muday et al., 1993) or more than one site (Michalke et al., 1992). NPA may either bind directly to the auxin efflux carrier (Rubery, 1979; Sussman and Goldsmith, 1981) or to a distinct regulatory interacting protein (Cox and Muday, 1994; Bailly et al., 2008). Surprisingly for such an important research tool, many of these contradictions remain unresolved. Indeed, such contradictory data are in themselves consistent with a complicated NPA-binding environment. Here follows a summary of important NPA-binding targets, both experimentally verified and hypothetical.

ATP-binding cassette (ABC) transporters

In 2001, three proteins with NPA binding activity were isolated by affinity chromatography from solubilized Arabidopsis membranes. All three were ABC-transporters: a large family of active membrane proteins with a broad substrate range for transport activity (Noh et al., 2001). Although the corresponding loss-of-function phenotypes did not resemble NPA-treated plants, double knock-out plants were dwarfed and displayed a large reduction in apical dominance, they immediately pointed to a role for ABC-transporters in the transport of auxin. This study was quickly corroborated by several more reports, all of which firmly placed ABC transporters as important components of the auxin efflux machinery (Lin and Wang, 2005). Most significant was a demonstration that ABCB1 is able to mediate the efflux of auxin from yeast cells. However, inconsistencies and pressing questions remain. If ABC transporters provided NPA sensitivity to auxin transport machinery, why does NPA application give a pin1 and not an abcb19 phenotype? Also, the localization of ABCB auxin transporters is not unambiguously polar. For example, although ABCB19 (MDR1/PGP19) has been shown to be strongly polar after immunostaining, it shows a non-polar localization after the visualization of a GFP-fusion in Arabidopsis hypocotyls (Geisler et al., 2005; Mravec et al., 2008). Similarly in the root, the localization of ABCB19 has either been described as polar (Blakeslee et al., 2007; Titapiwatanakun et al., 2009) or non-polar (Wu et al., 2007) in root stele cells. In contrast, PIN localization correlates very well with observed fluxes of auxin transport and hence can be used successfully to model development in silico (Blilou et al., 2005; Heisler et al., 2005; Jonsson et al., 2006; Grieneisen et al., 2007). Can the same be said for auxin-transporting ABCB proteins?

At first glance, it seems that research on ABCB transporters and PINs has been making their relationship ever more opaque. In different reports it has been concluded that PINs and ABCB19 form largely independent (Mravec et al., 2008) or strongly interdependent polar auxin transport mechanisms (Bandyopadhyay et al., 2007). However, as both PIN1 and ABCB19 have been shown to each account for over 50% of basipetal auxin flux in Arabidopsis stems, some degree of interdependence must be present (Okada et al., 1991; Noh et al., 2001). The independent transport of auxin by both protein families, PINs (Petrasek et al., 2006; Rojas-Pierce et al., 2007; Mravec et al., 2008) and ABCBs (Blakeslee et al., 2007), has been described as being sensitive to NPA leaving open the possibility that both protein groups are directly regulated by NPA.

The first link between ABCB transporters and polar auxin transport was made after researchers searched for genes that were differentially expressed after treatment with a blocker of anion channel activity: another benzoic acid derivative, 5-nitro-2-(3-phenylpropylamino)-benzoic acid (NPPB) (Noh et al., 2001). The energy for polar auxin transport had long been considered to come from that stored in the pH difference across the plasma membrane and not from the direct hydrolysis of ATP by the transporter (Rubery and Sheldrake, 1974; Raven, 1975) - a model which is more easily reconcilable with PINs than the ATP-hydrolysing ABC transporters. An excellent overview of two theories of polar auxin transport, the chemiosmotic model and the polar secretion model, has already been published (Goldsmith, 1977). The chemiosmotic theory of polar auxin transport (Rubery and Sheldrake, 1974; Raven, 1975), unlike the polar secretion model, requires no direct hydrolysis of ATP and despite occasional counter-arguments (Hasenstein and Rayle, 1984; Benning, 1986) has come to be widely accepted. To date there has been no mechanistic explanation as to how ABCB19, as an energy-dependent transporter (Geisler and Murphy, 2006), could be reconciled with the chemiosmotic hypothesis, especially after NPA has been proposed to exert its inhibitory effect by competing directly with ATP for ABCB binding (Di Pietro et al., 2002). However, two reports present data that could explain how ABCB transporters may be working within a framework dictated by the chemiosmotic hypothesis. Firstly, in a vesicle-based in vitro assay, auxin acts to lower the K_m of membrane proton transport in a vanadate-sensitive manner (Gabathuler and Cleland, 1985). Could the activity of ABCB19 be behind this response? Secondly, the question of whether ABCB19 itself possesses channel activity was addressed through a series of patch-clamp experiments on mammalian cells expressing ABCB19 (Cho et al., 2014). Here, ABCB19 was clearly shown to function as an ion channel; there is a single precedent for ABC transporters behaving in this way (Liu et al., 2017). NPPB but not NPA inhibited this activity, suggesting NPA does not block directly ABCB19 activity. It will be informative to learn whether other auxin transport inhibitors are active in this context.

Twisted dwarf 1

TWISTED DWARF1 (TWD1) was originally identified in a plant stature screen (Kamphausen et al., 2002) and encodes an FKBP-like immunophilin with a transmembrane domain but without any measurable enzymatic cis-trans-peptidylprolyl isomerase activity. Using a yeast two-hybrid screen for interaction partners, TWD1 was found to associate with ABCB1 and ABCB19 to stimulate cellular auxin efflux in plant and mammalian cells, though the interaction is inhibitory in yeast (Geisler et al., 2003; Bouchard et al., 2006; Bailly et al., 2008), as well as two vacuolar ABC transporters, MRP1 and MRP2 to affect their capacity to transport their substrates across the tonoplast (Geisler et al., 2004).

TWD1 is predominantly localized to the ER, though it can also be seen on the plasma membrane (Wang et al., 2013) and has been suggested to be necessary for the proper localization of ABCB19 (Wu et al., 2010). TWD1 binds NPA in yeast, though only when ABCB1 is not present and vice versa; it is also able to compete with NPA for ABCB19 binding (Bailly et al., 2008). Thus, a picture has been steadily compiled of a TWD1-ABCB1/ABCB19-containing auxin-transport-competent protein complex that can be disrupted by NPA binding to both TWD1, ABCB1, and ABCB19. However, certain natural NPA-like analogs that have no inhibitory effect on polar auxin transport, for example quercetin O-glucoside, were able to completely inhibit the interaction between TWD1 and ABCB19 at 10 µM (Bailly et al., 2008). It therefore remains to be shown conclusively that TWD1-mediated polar auxin transport inhibition occurs due to its binding NPA.

Additional clues as to the relationship between TWD1 and NPA come from the characteristic twisting phenotype of twd1 plants. It has been hypothesized that organ twisting is due to the uneven lateral distribution of auxin response maxima observed in twd1 plants, which could move radially as a function of time, though this has not yet been demonstrated (Wu et al., 2010). This imbalance is thought to be suppressed by the application of NPA, resulting in the twisting phenotype being less pronounced. How may this observation be reconciled with a TWD1-ABCB response module for the action of NPA? It has been persuasively argued that this phenomenon can only be explained by the independent action of NPA on a TWD1-independent mechanism that regulates auxin transport (Wu et al., 2010). PIN proteins must be considered prominently in this case as their localization is unaffected in twd1 plants (Wu et al., 2010). Another pressing and as yet completely open question is how a functional interaction can be integrated with the different localizations of TWD1, ABCBs, and PINs at the ER and plasma membrane.

Aminopeptidase 1 (APM1)

An example of how the application of inhibitors can open new perspectives on the mechanism of polar auxin transport was seen with a report that identified bestatin as a polar auxin transport inhibitor (Murphy et al., 2000). Bestatin has been known to be an inhibitor of aminopeptidases since the mid-1970s and its inhibitory effect on polar auxin transport implied that aminopeptidase activity is a positive regulator of polar auxin transport. An interaction between an aminopeptidase, hereafter called APM1) and NPA was discovered after cross-linked NPA was used for the affinity purification of proteins from Arabidopsis (Murphy et al., 2002). The similarity shared between APM1 and the mammalian insulin-regulated aminopeptidase (IRAP) was noticed and suggested a mechanism for the regulation of auxin transport (Muday and Murphy, 2002). In mammals, glucose transporter type 4 (GLUT4) imports glucose primarily into muscle and adipose cells; in the cell, its plasma membrane localization is regulated by insulin, just as the subcellular localization of PIN1 has been shown to be affected by certain auxin analogs (Paciorek et al., 2005). However, it seems clear that at NPA concentrations that are sufficient to severely affect auxin efflux, the localization of important PIN proteins is unchanged, although the organization of the actin cytoskeleton may be disrupted (Rahman et al., 2007). This makes NPA-mediated control of polar auxin transport by the regulation of subcellular PIN localization via APM1 unlikely. However, this conclusion seems to be undermined by a new report that shows an effect of NPA on both actin bundling and PIN localization, effects that have been attributed to the activity of TWD1 (Zhu et al., 2016). However, again, discrepancies exist in the literature as another report claims NPA has no effect on actin dynamics (Dhonukshe et al., 2008). Nevertheless, apm1 mutants display lower rates of polar auxin transport with phenotypes characteristic of defects in polar auxin transport and mislocalized PINs and ABCB19 (Peer et al., 2009). In addition, inhibition of APM1 with PAQ22 caused the internalization of PIN3 when its expression was driven by a root-hair-specific promoter (Lee and Cho, 2012). The different effects of NPA and lack of APM1 on PIN localization make it unlikely that NPA acts exclusively through APM1, though a phenotypic similarity between apm1 and wild-type plants grown at high NPA concentrations of >30 µM supports APM1 making a regulatory contribution (Peer et al., 2009). Based on this series of results it may be justified to hypothesize that APM1 binds to NPA in a manner that is independent from its phytotropin activity.

PINs

At first glance, the direct inhibition of PIN1 by NPA is the most economical explanation for much of the observed data, not least the shared phenotype of NPA-treated and pin1 loss-of-function plants (Okada et al., 1991; Reinhardt et al., 2000; Xu et al., 2005). However, evidence for such a mode of action is conspicuous by its absence; indeed pin1 plants have been reported to have normal sensitivity to NPA (Garbers et al., 1996), though the Arabidopsis genome encodes three further PIN proteins which may account for this observation (Blilou et al., 2005; Paponov et al., 2005). In auxin transport assays that use heterologous expression systems, NPA can have an inhibitory effect on auxin transport in vector only controls. This background activity has been attributed to the non-specific inhibition of endogenous ABC transporters (Yang and Murphy, 2009). Although there have been sporadic reports of NPA having a direct effect on PIN activity (Petrasek et al., 2006; Rojas-Pierce et al., 2007; Yang and Murphy, 2009), these reports have failed to coalesce into a coherent argument for PIN proteins as a site of NPA action. Much of this may be attributed to the difficulty in establishing reliable efflux assays for PIN proteins. In cases when assays have been established, often no experiments using NPA were reported (Petrasek et al., 2006; Zourelidou et al., 2014) or NPA had no effect on PIN-dependent auxin efflux (Kim et al., 2010).

What seems clear is that if PINs bind NPA at all, they do so with a relatively low affinity when compared with ABCB19, which is itself a PIN-interacting protein (Blakeslee et al., 2007). Plant microsomes have, however, been measured as binding NPA with more than one distinct K_m, and it is the weaker affinity binding events which match well with the concentrations of NPA that inhibit auxin transport and could potentially be ascribed to PIN binding activity (Michalke et al., 1992). It should, however, be noted that NPA binds to microsomes prepared from pin1 Arabidopsis plants with only slightly less affinity than to those prepared from wild-type plants, though endogenous NPA-like compounds were present in this assay, as were other plasma membrane-localised PINs (Rojas-Pierce et al., 2007). In contrast, microsomes prepared from abcb19 plants bound only 50% of the NPA bound by wild-type plants (Rojas-Pierce et al., 2007).

What is the functional relationship between NPA and TIBA binding?

TIBA was first used as a tool to investigate the effects of photoperiod on flowering time (Zimmerman and Hitchcock, 1942), and it was quickly deduced that TIBA inhibited competitively the effects of IAA in soybean (Galston, 1947). As it is a weak aromatic acid, protonated TIBA is thought to diffuse into cells in a similar way to IAA, though with 100-fold greater permeability under physiological conditions (Depta et al., 1983). Unlike NPA, TIBA is thought to be polarly transported in a similar manner to IAA (Thomson et al., 1973) through a common efflux carrier (Depta et al., 1983). This conclusion was reached after TIBA and 2,4-D were shown to stimulate each other’s accumulation in segments of Cucurbita pepo hypocotyl, though with notable differences in accumulation among different species. Whether these differences are able to reconcile the experiments with the observation that 2,4-D is not a substrate for the efflux carrier is unclear (Delbarre et al., 1996). Several pieces of evidence point to TIBA being a straightforward competitor for the site of IAA transport activity: these include i) the mutual inhibition of efflux between cellular auxin and TIBA, ii) TIBA’s relatively weak activity when compared with NPA, iii) TIBA’s monophasic dose-response curve (Katekar and Geissler, 1980), iv) TIBA’s ability to inhibit the binding of IAA to membranes in vitro (Thomson et al., 1973), and vi) TIBA’s ability, again in contrast to NPA, to displace NAA from its membrane-localised binding site (Ray et al., 1977).

However, a single-site model for polar IAA and TIBA transport has been credibly dismissed after a series of experiments that showed that in a membrane vesicle auxin accumulation assay, no mutual inhibition of IAA and TIBA transport exists (Depta et al., 1983). In fact, the truth is likely to be a good deal more complicated than IAA and TIBA competing for a single binding site on a single carrier. A two site model for polar IAA and TIBA transport was subsequently proposed along with suggested compromises necessary to reconcile it with the binding data of Ray et al. (Ray et al., 1977; Depta and Rubery, 1984). In it, TIBA and IAA bind to separate, but functionally related sites on the efflux transporter, and the transport of either is inhibited when both are bound simultaneously. But how do the TIBA and auxin transport sites relate to the discrete NPA binding site? A measure of light was shed on this issue when, instead of the active 2,3,5-TIBA, NPA-bound membrane preparations were treated with the inactive 3,4,5-TIBA analog. In this case, 3,4,5-TIBA was able not only to displace NPA from its binding site, but also to recover auxin transport activity (Depta et al., 1983). When taken in the context of TIBA/IAA binding kinetics, this observation argues strongly that inhibitory TIBA and NPA binding occurs on the same protein molecule, and by extension, that this molecule is the efflux carrier.

Are NPA and the flavonols functionally equivalent?

This section does not aim to give a broad overview of the role of flavonols in the regulation of auxin transport. For this, many good summaries are already available (Rubery, 1990; Peer and Murphy, 2007; Buer et al., 2010). Instead, we aim to highlight some pressing questions raised by studies of the endogenous regulation of polar auxin transport.

Why should plants have evolved multiple protein structures able to bind, sometimes with high affinity, a synthetic chemical such as NPA to which they have never been exposed? This question was answered by a study in the late 1980s which demonstrated that the ability of different flavonols to inhibit vesicular auxin efflux correlates with the ability of those compounds to displace NPA from the same membrane preparations (Jacobs and Rubery, 1988). Flavonols form a class of compounds that belong to the wider family of flavonoids, which also includes flavanols, flavones, and anthocyanins. Though structurally similar, there is sufficient opportunity for modification to the basic backbone of three benzene rings to give the more than 4000 flavonoids that have been characterized. Flavonols can be distinguished from the other major classes of flavonoid compound by their distinct patterns of hydroxylation and oxidation of carbons on all three of their benzene rings (Taylor and Grotewold, 2005).

The series of experiments that confirmed the flavonols as endogenous regulators of auxin transport were conducted on a genotype known as transparent testa4 (tt4). This plant lacks chalcone synthase: the first enzyme that is dedicated to flavonoid biosynthesis (Shirley et al., 1995). tt4 plants display increased rates of auxin flux through inflorescence stem segments, these faster rates can be abrogated by the external application of naringenin, the precursor to flavonol biosynthesis (Brown et al., 2001).

Endogenous flavonols not only inhibit the rootward transport of auxin, but also affect the PIN2-dependent shootward reflux of auxin in the outside cell layers of the root meristematic zone (Buer and Muday, 2004). It is, however, unclear whether the regulation of shootward auxin flux is via the effect of endogenous flavonols on the efficiency of PIN mediated auxin transport (Rojas-Pierce et al., 2007) or through changing the localization of PIN2 as seen in rol1-2, a genotype in which the flavonols are particularly potent (Kuhn et al., 2017). The contribution of ABC transporters to the gravitropic response has been investigated with an unrelated inhibitor called gravacin which binds ABCB19 and competes effectively for the ABCB19-localised NPA binding site, though is not considered to interfere with PIN-dependent auxin transport. ABCB19 was identified in a mutant screen as the locus that confers sensitivity to gravacin (Rojas-Pierce et al., 2007). Here we come across a counterintuitive aspect of polar auxin transport regulation: inhibiting auxin flux can lead to an increase in sensitivity to gravity, whereas increasing flux causes a decrease (Noh et al., 2003; Buer and Muday, 2004). This has been explained by different treatments or genotypes having relatively different effects on the strength of shootward and rootward auxin transport systems. Here, an attenuation of the gravitropic response would be caused by relatively strong rootward flux in the absence of endogenous flavonols. This model predicts a relatively insensitive PIN2-dependent shootward auxin transport with respect to regulation by both flavonols, despite relocalisation of PIN2 in rol1-2, and ABCB19 (Noh et al., 2003). ABCB1 and ABCB19 have been shown to differentially regulate shootward (ABCB19) and rootward (ABCB1) flux lending this hypothesis some experimental support, though in this case no effect on the gravitropic response was measured after genetic attenuation of rootward auxin flux (Lewis et al., 2007).

The endogenous function of flavonols with respect to the regulation of auxin transport is likely to be more nuanced than is the exogenous application of NPA, with different flavonols and flavonol glycones being localized to different parts of the plant and playing different physiological and developmental roles (Kuhn et al., 2011; Yin et al., 2014; Kuhn et al., 2016). This view is supported by a series of experiments on tt flavonol-deficient mutants that showed a range of different abilities to transport auxin and displayed associated phenotypes such as gravitropic and lateral rooting defects (Buer et al., 2013). Though the ability to compete for NPA binding and the ability to inhibit auxin transport were generally correlated among a range of flavonols, there were notable outliers, for example morin, which associated readily with NPA binding sites but did not inhibit auxin transport (Jacobs and Rubery, 1988). PIN proteins are also likely to have different sensitivities to any given flavonol. For example, in pin2 plants, the PIN1 expression domain expands into the epidermis where it mediates shootward auxin flux (Santelia et al., 2008). As PIN1 is more sensitive to inhibition by flavonols than PIN2 (Peer et al., 2004), nanomolar concentrations of quercetin are able to partially restore the gravitropic response (Santelia et al., 2008).

Transport and environmental regulation of flavonol availability undoubtedly occur, but their overall significance still remains unclear (Thompson et al., 2010; Zeng et al., 2010; Lewis et al., 2011). For example, a transient increase in flavonoid concentration which occurs two hours after a gravity stimulus in wild-type roots plays a role in the gravitropic response, however it is unclear to what extent this response is tuned to a specific flavonol-PIN interaction (Buer and Muday, 2004).

Does NPA act through the regulation of protein phosphorylation?

Although protein phosphorylation stimulates the activity of auxin efflux carriers (Delbarre et al., 1998; Henrichs et al., 2012; Zourelidou et al., 2014), the use of a tyrosine kinase inhibitor has shown that it is also needed for the inhibition of auxin efflux by NPA (Bernasconi, 1996). However, assigning the activity of specific kinases to either the stimulation or inhibition of NPA-sensitive auxin efflux has not been straightforward. This is probably because various kinases participate in regulatory circuits that both enhance and repress auxin efflux, indicating regulation is multi-level and complicated (Bernasconi, 1996; Delbarre et al., 1998).

In some cases, kinase, phosphatase, and protein targets are known. For example, two independent lines of enquiry, each beginning with the characterization of phenotypes which resembled either NPA-treated (Bennett et al., 1995) or NPA-insensitive (Garbers et al., 1996) plants, later converged on the antagonistic phosphorylation and dephosphorylation of PIN1 at positions which control its subcellular localization (Michniewicz et al., 2007; Huang et al., 2010). ROOTS CURL IN NPA1 (RCN1) encodes the regulatory subunit of a PP2A phosphatase. It was identified after a screen that measured the sensitivity of mutant plants in a root growth assay; as roots grow through agar, after touching the bottom of a horizontal petri dish, they tend to grow in an inward spiral. NPA at a concentration of 5 µM abolishes this behavior causing wild-type roots to grow straight whereas rcn1 plants continue their curly growth. This growth response is specific: rcn1 plants respond as wild-type plants to other polar auxin transport inhibitors such as TIBA (Garbers et al., 1996). So is RCN1 another direct target of NPA? This is unlikely as shootward auxin transport at the root tip is inhibited by NPA to the same extent in rcn1 and wild-type plants (Rashotte et al., 2001). However, of great interest is a complementary experiment that demonstrated that NPA does not inhibit transport in the opposite direction in rcn1, especially as a phosphatase inhibitor had the same effects (Rashotte et al., 2001). These data suggest different mechanisms regulate shootward and rootward auxin flux in the root apical meristem and that RCN1 plays a different role in each.

Conversely, PINOID encodes a PDK1-activated AGC kinase that, when absent, results in the basal localization of PIN1 in cells of the inflorescence meristem and pin1-like defects in phyllotaxy (Christensen et al., 2000; Zegzouti et al., 2006). Other AGC kinases have also been shown directly to phosphorylate PIN1 to promote polar auxin transport, for example WAG1 and WAG2 that cause a reduction in sensitivity to NPA (Dhonukshe et al., 2010)) and the D6 protein kinases. Here, the phenotypes of overexpressor lines can largely be abrogated by application of 5 µM NPA (Barbosa et al., 2014).

ABCB19 is also phosphorylated by AGC kinases, for example by PHOT1 during the phototropic response (Christie et al., 2011). There is some evidence that ABCB19 is also phosphorylated by PINOID, effecting another level of control over polar auxin transport (Henrichs et al., 2012). This association was found after immunoprecipitation experiments identified a direct interaction between PINOID and TWD1 (Henrichs et al., 2012). However, the relationship among PID, ABCB1, TWD1, and the activation or inhibition of auxin transport appears to be rather complex. PID is able to either stimulate or inhibit ABCB1-dependent cellular auxin efflux depending on whether or not TWD1 is present (Henrichs et al., 2012). Another interesting aspect is that here, NPA may not be considered as simply a synthetic analog of the flavonols. Not only did quercetin prove to be a more effective inhibitor of PID activity than NPA, as measured by levels of PID autophosphorylation, but it showed an increased, rather than a decreased affinity for ABCB1 when a phosphomimetic glutamic acid was substituted at the PID substrate site (Henrichs et al., 2012). This work represents an exciting direction of enquiry that could uncover the molecular basis for different regulatory effects between NPA and the flavonols on polar auxin transport.

A careful genetic dissection of the activity of flavonols in rol1-2 also revealed a role for protein phosphorylation in flavonol-mediated polar auxin transport inhibition (Kuhn et al., 2017). rol1-2 is deficient in rhamnose synthase which crucially for this work results in changes to the plants’ flavonol profile, which has rather complex effects on the regulation of auxin transport (Kuhn et al., 2011). These effects require the activity of RCN1 as the rcn1 allele suppresses the rol1-2 phenotype without affecting the plant’s flavonol composition (Kuhn et al., 2017). Here, data are consistent with a hypothesis that the observed apical localisation of PIN2 in root cortex cells of rol1-2 plants is caused by the presence of unusual flavonols. Furthermore, the results from a gravitropic response assay have been used to make a compelling argument that PID kinase activity is decreased by the activity of flavonols (Kuhn et al., 2017). Further investigations of the key players behind these phenotypic relationships promise new insights into the mechanisms of polar auxin transport regulation.

Conclusion

NPA has given us rare glimpses into the inner workings of the auxin efflux carrier complex and its regulatory proteins but now stands at a crossroads. Although it has brought our understanding of polar auxin transport forward a long way, at the heart of this research and commentary there remains contradiction and controversy. For example, the chemiosmotic hypothesis, the accepted model that provides the foundation for our understanding of polar auxin transport, requires transporters energized by a pH gradient across the plasma membrane. However, the proteins that bind the most prominent transport inhibitors, and therefore must be placed at the centre of any transport model, belong to a protein family that depends on the hydrolysis of ATP for substrate efflux. The most convincing biochemical data point clearly to ABCBs and TWD1 as NPA targets; however, genetic data points to another: PIN1. The usefulness of NPA depends on its ability to precisely dissect either the auxin transport mechanism directly, or the signaling processes which control it. At best, our clearest understanding points to a simultaneous influence over transport and control. However, are exisiting inhibitors (Fig. 1) sufficient to separate these mechanisms in order to access them experimentally or do we need new ones? Wherever NPA takes us in the future, it is clear that the journey will need standardized, robust in vitro auxin efflux assays if results are to be compared effectively, and existing physiological data will need to be put into this molecular context as it continues to emerge.

Fig 1.

Auxin and some inhibitors of its polar transport, with their relevant analogs. Pink shading and a dotted line indicate compounds that do not occur naturally in plants. Blue shading and a solid line indicate auxin transport inhibitors.

Abbreviations:

Abbreviations:
 
• 2,4-D

  2,4-dichlorophenoxyacetic acid

 
• ABC

  ATP-binding cassette

 
• APM1

  aminopeptidase 1

 
• IAA

  indole-3-acetic acid

 
• NPA

  naphthylphthalamic acid

 
• PIN

  pinformed

 
• TIBA

  2,3,5-triiodobenzoic acid

 
• TWD1

  TWISTED DWARF1.

Acknowledgements

The authors gratefully acknowledge support by Bundesministerium für Bildung und Forschung (BMBF Microsystems), the Excellence Initiative of the German Federal and State Governments (EXC 294), SFB746 and DLR (50WB1522). Thanks also go to Rainer Hertel for valuable discussions. The authors declare that no competing interests exist.