Boronic acid treatment phenocopies monopteros by affecting PIN1 membrane stability and polar auxin transport in Arabidopsis thaliana embryos

The micronutrient boron (B) is essential for many organisms (Kliegel, 1980). Membrane-localised transporters in plants and animals control B homeostasis (Takano et al., 2002; Park et al., 2004; Tanaka et al., 2008; Durbak et al., 2014) and act to counteract the detrimental effects that B excess or depletion can have for plants and animals (Warington, 1923; Sommer and Sorokin, 1928; Lanoue et al., 1998; Rowe and Eckhert, 1999; Goldbach et al., 2001; Fort et al., 2002; Reid et al., 2004; Nielsen, 2008; Martin-Rejano et al., 2011; Abreu et al., 2014). Consequently, the concentration of B in soils, whether high or low, is of considerable importance for crop production (Dell and Huang, 1997; Shorrocks, 1997; Reid et al., 2004; Durbak et al., 2014).

The key to understanding the molecular functions of B, in the form of boric acid B(OH)₃ or borate anions B(OH)₄⁻, is its reactivity with (two) cis-diol pairs and its capability to form (ester) cross-links between neighbouring molecules through their cis-diol moieties (Kliegel, 1980; Brown et al., 2002; Bolaños et al., 2004). Experimental analysis has often explored the reaction of organisms under low B concentration (see aforementioned citations). This is challenging because B-free experimental conditions cannot be attained to completion because B is ubiquitous. As a simpler approach, the use of boronic acids, such as phenylboronic acid (PBA), has emerged as an alternative to imitate B depletion conditions (Bassil et al., 2004). Boronic acids and their boronates are specific B competitors with the same binding specificity as boric acid but that cannot serve to cross-link two molecules because they only have one hydroxyl pair (e.g. Torssell, 1963), a feature used in boronate-affinity chromatography (e.g. Wimmer et al., 2009).

In spite of this long history, the mechanistic role of B is understood in only few processes; for instance, as a component of the bacterial quorum sensor (Chen et al., 2002) and in plant cell wall stabilisation through cross-linking rhamnogalacturonan-II polysaccharides (Ishii and Matsunaga, 1996; Kobayashi et al., 1996; O'Neill et al., 1996). However, the fact that B is required in plants as well as in animals suggests roles beyond cell wall stabilisation. One hypothesis is that B might be required for the stability of plasma membrane (PM) proteins, which control key processes in physiology and morphogenesis, for instance by cross-linking glycoproteins and/or lipids (Parr and Loughman, 1983; Brown et al., 2002; Bassil et al., 2004; Goldbach and Wimmer, 2007). In fact, numerous B depletion- or boronic acid-induced phenomena affect morphogenesis and organogenesis processes, which are likely to depend on PM-relayed signal transduction pathways, such as root growth and root apical meristem (RAM) development (Sommer and Sorokin, 1928). Prominent in plants is the PBA-induced monocotyly in the dicot Eranthis hyemalis (Haccius, 1960). The reason for this remains unknown, most likely because this plant has a special mode of embryo development that takes place during germination and requires several months. Exploring a possible B-related morphogenetic process in plant embryos, we detected that boronic acids are capable of inducing phenocopies of the Arabidopsis thaliana rootless pattern mutant monopteros by interruption of polar auxin transport. The data show that PBA affects the membrane stability of proteins such as PIN1.

We detected that treatment of siliques of different A. thaliana accessions with PBA produced up to 98% (per silique) rootless seedlings, whereas controls gave only wild-type seedlings (Fig. 1, Table S1). The rootless seedlings were highly similar to mutants of the transcription factor MONOPTEROS (MP) based on morphological and histological criteria. Both displayed occasionally irregular cotyledons, severe defects of the vascular system, little hypocotyl with large vacuolated and few epidermal cells and complete absence of the RAM (Fig. 1A-D). Only a few seedlings were lethal or stunted. A few others displayed longer hypocotyls lacking RAMs or abnormal cotyledon number and size, suggesting late or incomplete contacts of embryos with PBA (<3%; Fig. S1). In order to exclude indirect effects of silique physiology (e.g. upon metabolising PBA), isolated ovules with embryos up to the heart stage were PBA treated and cultured according to Sauer and Friml (2004). Up to 44% of all embryos (n=35; water-treated controls, 0%) of late heart stage and beyond were identified as phenocopies (Fig. 1E,F). For convenience, we use the term ʻmonopteros (mp) phenocopies' hereafter.

Binding affinities of boronic acids can depend on various chemical parameters, in particular different substituents (Fig. 2A) (Yan et al., 2004). Testing 3-methoxyphenylboronic acid (3-MPBA), 4-methoxyphenylboronic acid (4-MPBA) and 3-nitrophenylboronic acid (3-NPBA) demonstrated their capability to induce mp phenocopies with variable frequencies: a maximum of 78%, 100% and 100% phenocopies per silique, respectively (Fig. 2B, Fig. S1, Table S1; the abbreviations are for convenience and do not necessarily conform with any chemical convention). In view of these results, we tested the structurally unrelated butylboronic acid (BBA), which has a non-phenolic substituent (Fig. 2A). In fact, BBA treatment of siliques produced 2% phenocopies at most (Fig. 2B, Table S1). By introducing a lateral opening in the siliques (see the Materials and Methods) only a maximum of 35% mp phenocopies per silique was achieved (Table S1, Fig. S1). The BBA frequency remained low in embryos from ovule culture (Fig. S2) in spite of using a tenfold higher concentration in comparison to PBA (8%, n=128; water-treated controls, 0%).

The known chemical competition of B (boric acid) versus boronic acids for cis-diols suggests that one should outcompete the other if present in excess. However, elevated concentrations of boric acid (120 mM, 150 mM) failed to suppress the effect of low (10 mM) PBA concentrations in several experiments (as compared with PBA alone; Table S2). At the same time, application of the same amounts of boric acid alone did not affect the seedling morphology (Table S2). From the silique treatments, ovule culture and competition experiments we therefore concluded that the structure of boric and boronic acids affects the capacity to penetrate tissues and to induce mp phenocopies.

Boronic acids share structural similarity to acetic and benzoic acid compounds (Fig. 2A). Acetic acids have been used to verify that the effects of boronic acids are specifically due to the disruption of B cross-links and not to other general effects (Bassil et al., 2004). After application of phenylacetic acid (PAA), 3-methoxyphenyl-, 4-methoxyphenyl- and 3-nitrophenylacetic acid most seedlings had a wild-type appearance, except in 14/265 siliques, where 1.8-35% cotyledon fusions were found (Fig. S3, Table S3). This is compatible with the fact that PAA is known to be a ʻmild auxin' in other plants (Simon and Petrásek, 2011). All treatments failed to produce mp phenocopies. Note that at least one of these compounds (3-nitrophenylacetic acid) is a much stronger acid than PBA, 4-MPBA or 3-NPBA (Bassil et al., 2004). In addition, experiments with benzoic acid, 3-methoxybenzoic, 4-methoxybenzoic and 3-nitrobenzoic acid produced predominantly wild-type progeny. No mp phenocopies were generated. Other seedlings germinated but were growth retarded, displayed malformed apices or short roots with regular cell files. Notably, defective seedlings occurred in all siliques, excluding an effect on a particular developmental stage (Table S4). The aforementioned results showed that every boronic acid tested is capable of inducing mp phenocopies. Together with the (negative) results obtained with acetic and benzoic acids, this shows that the boronic acid moiety alone is responsible for mp phenocopy induction.

mp phenocopy induction did not depend on the accession/line or whether siliques grew on primary or secondary stems. It rather occurred in siliques harbouring early embryo stages (Fig. 2B, Fig. S4, Table S1). This suggested that embryos up to early torpedo stage responded to the boronic acid treatment, since these stages are crucial for forming the RAM (Jürgens and Mayer, 1994). RAM formation starts with a horizontal division of the hypophysis cell in A. thaliana (Schlereth et al., 2010). With this in mind, all siliques of selected branches were treated with PBA and the embryos of every silique were separately evaluated after 24 h. Thus, for every silique, the frequency of embryos corresponding to a particular stage subdivided into the frequencies (%) of normal embryos and of embryos displaying abnormal cell divisions were scored (Fig. 3A, Table S5). For instance, the third silique of a treated Col-0 plant (in Fig. 3A) produced 22% and 7% abnormal early and mid heart stage embryos, respectively (white box, red squares). In addition, the same silique gave normal embryos of the late globular (7%), transition (4%), early heart (39%), mid heart (11%) and late heart (11%) stages (white box, black squares). The third silique of a comparable untreated plant produced only wild-type embryos (grey box, white circles). In this way, the effect of PBA could be traced back to the early (32-cell) globular stage as a deviation from the first formative division of the hypophysis resembling that observed in mp mutants (Fig. 3Ba-c,Ca-c). Vascular precursor cells could also display abnormal morphologies, being less elongated along the apical-basal axis of the embryo (Fig. 3B,C).

Next, we focussed on the two signals known to trigger the formative division of the hypophysis, namely the accumulation of IAA (basal auxin maximum) and the presence of the mobile TARGET OF MONOPTEROS 7 (TMO7) transcription factor (Benková et al., 2003; Schlereth et al., 2010). We considered that boronic acids could bind cis-diols of nucleic acids and thus interfere with gene expression processes. However, RT-PCR analyses revealed no significant differences in the expression of MP, the MP repressor BDL or ACTIN2 in mp phenocopies versus control seedlings (Fig. 4A, Fig. S5A). By contrast, boronic acid-induced mp phenocopies showed a severe reduction, if not a complete absence, of TMO7 transcripts, as in mp and bdl mutants. All controls displayed strong TMO7 expression (Fig. 4A). We concluded that boronic acid treatment does not generally affect transcription at this stage but specifically and indirectly that of TMO7.

Transcriptional differences of TARGET OF MONOPTEROS 5 (TMO5), which is required for vascular development (De Rybel et al., 2013), were less pronounced (Fig. 4A). In this case, analyses of reporter constructs, including TMO5p:TMO5-3XGFP, were more informative. We applied conventional epifluorescence in order to capture the complete fluorescence of the embryos. Upon PBA treatment, heart stage embryos carrying the IAA-responsive DR5::GFP reporter displayed significantly reduced (45%) or absent (43%) root auxin maxima while simultaneously retaining the cotyledon tip maxima after 24 h (Fig. 4B; n=65; signal present in all controls). In TMO7p:TMO7-3XGFP embryos the threefold GFP fusion blocks mobility and shows where TMO7 originates. Here, TMO7-3XGFP could be severely altered, reduced or completely absent in embryos at 24 h or 48 h after PBA treatment (Fig. 4C,D; for statistics see Fig. S5B,C). Analysis of PBA-treated embryos carrying TMO5p:TMO5-3XGFP, which is visible in vascular precursor cells of cotyledons and the hypocotyl, displayed a mild but detectable delay of TMO5-3XGFP signal emergence around the heart stage (Fig. 4E,F; for statistics see Fig. S5D,E). Note that after 48 h embryos can be morphologically identified as mp phenocopies (Fig. 4D,F). Ovule culture experiments revealed essentially the same results for embryos carrying DR5:GFP and TMO7p:TOM7-3XGFP reporters (GFP signal absent in n=5/18 and 12/39, respectively; signal present in all controls; Fig. S6).

In the embryo, the basal polarity of PIN1 in the central stele establishes the strong basal auxin maximum required for root development. PIN1 apical polarity in the epidermis, as determined by the activity of the PINOID kinase and the colocalised NPH3-like protein ENHANCER OF PINOID (ENP), establishes the weaker auxin maxima required for the development of cotyledon primordia (Friml et al., 2003; Benková et al., 2003; Treml et al., 2005; Furutani et al., 2014).

Within 30 min after PBA treatment of ovules, confocal laser scanning microscopy (CLSM) showed that in 88% of PIN1p:PIN1-GFP embryos (n=33) there was a complete delocalisation of PIN1 in all cells including central stele cells (Fig. 5A-D; in 21/22 controls PIN1 linked to stele precursors was recognisable). After 48 h and 72 h, internal PIN1 was completely missing in the residual basal region of 90% of mp phenocopies (n=20; PIN1 present in all controls). Residual PIN1 remained in the upper cotyledon tissue, showing partial polarity in the central (vascular) tissue and full restoration in the epidermal tissue (Fig. 5E-H). PBA also caused irregularly developed cotyledon primordia (Fig. 5E-H, Fig. S7).

We then tested the dynamics of these proteins by time-limited immersion of ovules in PBA solutions. We focussed on epidermal and near-epidermal cells because these layers allow easy monitoring of protein-GFP localisation by CLSM. PBA treatment of PIN1p:PIN1-GFP or 35Sp:EGFP-ENP embryos resulted in diffuse cytosolic internalisation of GFP signals (Fig. 5I,J). A similar effect was seen for ectopic EGFP-ENP in the root (Fig. 5K). In all cases internalisation could be significantly reversed after washing (Fig. 5I-L; for statistics see Table S6). Simultaneous staining with the PM marker FM4-64 showed only a few FM4-64 vesicles, excluding a general internalisation or perforation of the PM bi-layer under the conditions applied (Fig. 5M). Comparison with PIN2, a cortex- and epidermis-localised IAA efflux carrier that is not detectable in the embryo (Friml et al., 2003) but required in the post-embryonic root, did not show this marked internalisation (Fig. 5N; n=26). All applications were almost indistinguishable from each other in their effect, except a variable number of vesicles seen after 10 mM PBA (Fig. 5N,O), some of which merged with FM4-64 (Fig. 5O). Thus, PBA targets proteins involved in IAA transport differentially.

IAA-triggered degradation of MP repressors, the Aux/IAA proteins, is known to release MP, which activates PIN1. We tested whether PBA could interfere with the degradation step in this regulatory circuit using the IAA signalling sensor DII-VENUS. This construct harbours a fluorescent protein fused to IAA-interaction domain II, which mediates Aux/IAA protein degradation (Brunoud et al., 2012). Application of PBA (10 mM) and IAA (10 µM) mixtures to roots of DII-VENUS plants (n=30) showed suppression of DII-VENUS degradation as compared with IAA treatment alone (Fig. 6; n=30). Water and PBA controls (n=30) showed signals comparable to those of the PBA/IAA co-treatment. Similar results were obtained with DII-VENUS using 1 µM IAA, whereas a resistant mutant mDII was not degraded by IAA (Fig. S8). Thus, PBA is able to directly or indirectly interfere with the step leading to BDL degradation.

In order to exclude or confirm a possible specificity of PBA for IAA-related embryo proteins we examined the root localisation of the B transporter BOR1, a PM protein unrelated to IAA transport. BOR1 function is directly linked to B homeostasis and shows a conspicuous response to cellular B concentration (Fig. 7, Figs S9 and S10). High concentrations of B, as given in MS media, lead to its endocytic degradation, whereas low concentrations lead to its accumulation (Takano et al., 2010). Seedlings carrying the 35Sp:BOR1-GFP reporter were grown on 0.5MS for 10 days and then transferred for 2-3 days on H₂O-agar plates with low B concentration (ʻbefore', Fig. 7A). All showed high BOR1-GFP abundance in their root tips, as measured from the average fluorescence intensity. By contrast, subsequent incubation of the same seedlings in 0.5MS without sucrose (MS-S) and with 100 µM H₃BO₃ (MS-S+H₃BO₃) showed a reduced concentration of the fusion protein (Fig. 7B). Direct incubation in water or re-incubation of H₃BO₃-treated seedlings on water again displayed strong GFP signals (ʻ+H₂O', Fig. 7C, Fig. S9). Surprisingly, incubation of roots in 0.5MS-S+H₃BO₃+PBA (100 µM, 10 mM) behaved like those in water (Fig. 7D). The difference to the MS-S+H₃BO₃ incubation was significant (P<0.001, t-test; Fig. 7E, Fig. S9, Table S7). Note that the cells of these roots encountered the same concentration of B as those without PBA. Essentially, the same picture emerged from experiments using B-depleted 0.5MS medium instead of a water incubation to challenge BOR1 expression (Fig. S10). These experiments also showed the effect of 0.5MS+H₃BO₃+PBA versus 0.5MS(−H₃BO₃)+PBA. In both cases BOR1 was expressed (Fig. S10, Table S8). Thus, PBA not only impacted on the localisation of particular IAA-related proteins but also on that of other PM proteins. The effect ranged from depletion (PIN1, ENP) to neutral impact (PIN2) to enhanced accumulation (BOR1).

Bassil et al. (2004) demonstrated that boronic acids and B compete for the same binding sites. However, competition experiments showed only a limited capability of increasing boric acid concentrations to restore PBA-inhibited pollen germination (Bassil et al., 2004). Using the ʻsilique system' presented in this study, the induction of PBA-induced mp phenocopies could not be convincingly suppressed with increasing amounts of boric acid (up to 150 mM). There are several likely reasons for this observation. Compared with boronic acids, boric acid has a higher pKa, indicating weaker dissociation and therefore affecting the strength to form ester bonds. Consequently, instead of flooding the cells with boric acid, washout of PBA is more suitable to revert its effects, such as the internalisation of ENP and PIN1. In this context, it is worth mentioning that PBA suppresses a known B effect by stabilising instead of internalising a membrane protein (BOR1). Also, a successful competition by boric acid has only a narrow time window during hypophysis division. Furthermore, the comparison of BBA with the other boronic acids suggests that molecular structure impacts the capability of a compound to penetrate tissues and to access its molecular targets, and thus to induce mp phenocopies. Apparently, boronic acids with a phenolic group outperform BBA with a carbon tail. In turn, boric acid might be even less able to penetrate through or access cis-diols at membranes. This problem cannot be circumvented using A. thaliana embryo culture because isolated embryos are not viable. However, these experiments helped to clarify another important question: the effect of BBA, when considered in terms of its structure, excluded the phenolic ring as causative for mp phenocopy induction. This was corroborated by the results obtained with all boronic acid ʻanalogues', which excluded the possibility that similar molecular structures were sufficient to induce phenocopies. Together, this showed that the ability to induce mp phenocopies is a property of the boronic acid moiety.

Our data suggest a specific effect of PBA in the feedback signal transduction cascade leading to RAM development. All known A. thaliana mutations causing mp or mp-like phenotypes map along this pathway (De Rybel et al., 2013; Schlereth et al., 2010; Hamann et al., 2002; Cheng et al., 2007; Friml et al., 2003).

In the early embryo, apical cells produce IAA (Wabnik et al., 2013), which triggers the degradation of Aux/IAA proteins, such as BDL, the repressors of MP (Fig. 8A). Released MP activates TMO7, TMO5 and also PIN1 in corresponding tissues (Schlereth et al., 2010; Hamann et al., 2002; De Rybel et al., 2013) and also impacts ENP expression (Furutani et al., 2014). Together, this establishes a feedback system because MP is itself dependent on IAA transported by PIN1. The steady-state flux of IAA from the source requires a continuous directional transport enabled by the polar localisation of PIN1 and regionally supported by proteins such as ENP (Friml et al., 2003; Benková et al., 2003; Treml et al., 2005; Furutani et al., 2014). This generates two weak auxin maxima for the cotyledon primordia and a strong auxin maximum in the hypophysis cell. The strong auxin maximum and TMO7 (coming from the central cells) signal to induce the formative horizontal division of the hypophysis (Fig. 8A).

In this scenario, internalisation of PIN1 by PBA blocks IAA transport to and from central cells and causes the loss of the root auxin maximum and MP activity due to failure of IAA-mediated degradation of BDL (Fig. 8B). Consequently, the expression of TMO7 and PIN1 is also abolished, as in mp mutants (Fig. 8B). Regionally, this also impacts the presence of TMO5. In the apex, internalised PIN1 and ENP cause a disturbance of IAA transport, as manifested in abnormalities in cotyledon maxima and cotyledon growth. Note that endogenously synthesized IAA enables MP release and activity, which explains, at least partially, TMO5 expression in PBA-induced mp phenocopies, which have an intact MP gene, in contrast to mp mutants (Fig. 8B).

However, the response of the DII-VENUS construct to PBA/IAA is also compatible with an alternative model whereby PBA has a direct inhibitory role, suppressing IAA-mediated BDL degradation and ultimately inducing mp phenocopies (Fig. 8B). It remains to be determined whether this DII-VENUS response could also result from a rapid PBA-induced rearrangement of membrane transport systems, precluding effective penetration by IAA. The effect of PBA on another transport system (BOR1) rather supports a more general effect of PBA on PM proteins.

The restoration of (epidermal) PIN1 polarity after 48 h demonstrates the transient impact of the early PBA pulse, validating PBA as a tool to analyse the significance of timing in development. It shows that a transient interruption of RAM formative divisions is sufficient to prevent embryonic root development completely. It is not possible to draw such a conclusion from the study of mutants, which constantly lack MP function.

There are interesting parallels between the induced loss of an embryonic root and the halted post-embryonic root development in other plants, although both root forms also have significant differences (e.g. weak embryonic BOR1 expression in the former; GENEVESTIGATOR database and our unpublished data). Sunflower, field bean and A. thaliana respond with partial or complete reduction of RAM activity when grown on B-deficient soil or media (Parr and Loughman, 1983; Dell and Huang, 1997; Martin-Rejano et al., 2011; Abreu et al., 2014; Li et al., 2015). A. thaliana also shows a reduction of localised PIN1 in the stele (Li et al., 2015). However, these effects result from long-term growth, in contrast to the rapid responses to transient PBA application.

Excess or reduction of B, as well as the presence of boronic acids (which mimic B depletion), can cause various physiological effects in vitro and in vivo ranging from abnormal gene splicing and expression to variable plant growth and lethality (e.g. Sommer and Sorokin, 1928; Kliegel, 1980; Parr and Loughman, 1983; Shorrocks, 1997; Dell and Huang, 1997; Goldbach et al., 2001; Shomron and Ast, 2003; Reid et al., 2004; Bassil et al., 2004; Martin-Rejano et al., 2011; Abreu et al., 2014; Durbak et al., 2014). However, under the conditions applied, PBA demonstrated an intriguing specificity on a morphological, cellular and also molecular level as evidenced by its variable but specific effects on four different PM proteins (PIN1, ENP, PIN2 and BOR1). This might indicate that disturbance of the B/cis-diol cross-links themselves could be part of a cellular B sensing mechanism, explaining why, in the presence of PBA, the cells do not respond to high cellular B levels.

PIN1-driven IAA transport might also be affected in Eranthis hyemalis, where PBA-induced monocotyly (Haccius, 1960) is likely to be a reflection of its unusual mode of development. At present, the PBA targets in these processes are not known, although plants possess numerous potential glycosylated targets, some with known morphogenetic impact (Motose et al., 2004; Geshi et al., 2013; Voxeur and Fry, 2014). Plants might also have N- and O-linked glycosylated protein sorting codes, as known from animals (Mellman and Nelson, 2008). In either case, the observations presented in the current study are of prime importance in our understanding of membrane structure. The combination of all PBA-specific effects supports the long-standing hypothesis that B has a functional role in membrane architecture. This finding in plant embryos might point to a comparable role in animals, which display B-related embryo defects but lack a cell wall (Lanoue et al., 1998; Rowe and Eckhert, 1999; Fort et al., 2002).

As wild types, Columbia (Col-0) and Landsberg erecta (Ler-0) were used. Mutants were mp-B4149 (Schlereth et al., 2010) and bdl (without designation, courtesy of G. Jürgens, University of Tübingen, Germany). Routinely, all siliques of a stem/branch and (often) the first three flowers were thoroughly coated (∼10-15 times) with the corresponding solution using a brush. Treated flowers eventually produced seeds or became necrotic. Plants were treated only once, except in one series with a single treatment per day on five consecutive days. The seeds of untreated terminal flower buds of the same stem and/or a different stem, and seeds from siliques of separated plants treated with water were harvested as controls. For additional BBA treatments, siliques were opened laterally with microsurgical forceps (Fine Science Tools). Plants were grown under conditions as previously described (Treml et al., 2005) and siliques were separately harvested. For ovule culture, siliques were opened and fixed on surface-sterilised slides with double adhesive tape. Ovules were covered for 5 or 10 min with PBA (5 mM), BBA (50 mM in 0.1% DMSO) and water as control and then transferred to medium and cultured as previously described (Sauer and Friml, 2004). For short-term analysis (shown Fig. 5A-D), ovules were covered in 10 mM PBA for 30 min and then immediately analysed by CLSM.

Except for PBA (Merck), all compounds were obtained from Sigma-Aldrich. Unless stated otherwise, solutions were prepared in water. A series of experiments revealed satisfactory induction frequencies of mp phenocopies with 50 mM PBA (for instance, 10 mM produces fewer than 5% phenocopies). Where possible, 50 mM concentrations were adjusted for the correspondingly tested compounds. In the case of 3-nitrophenylboronic acid, 4-methoxyphenylacetic acid, 3-nitrophenylacetic acid, 3-methoxybenzoic acid, 4-methoxybenzoic acid and 3-nitrobenzoic acid, saturated solutions were used because 50 mM surpassed the corresponding solubilisation capacity.

Total RNA isolation, reverse transcription, PCR and purification of amplification products of material of wild types, induced mp phenocopies, boronic acid-treated seedlings of wild-type appearance, and mp or bdl mutants were performed according to conventional methods (Treml et al., 2005). Amplification products were sequenced through Eurofins/MWG services. Primers (5′-3′, forward and reverse) were: ACT2, TTGTTCCAGCCCTCGTTTGT and CCTGGACCTGCCTCATCATACT; TMO5, ATGTACGCAATGAAAGAAGA and ATTATAACATCGATTCACCA; TMO7, ATGTCGGGAAGAAGATCACG and TTGGGTAAGTAAGCTTCTGA.

Whole-mounts and semi-thin sections were prepared as previously described (Haberer et al., 2002; Treml et al., 2005). High-magnification images of seedlings were taken with a Hitachi TM3000 table-top microscope.

Epifluorescence of embryos carrying GFP constructs was analysed 24-72 h after treatment of siliques with 50 mM PBA using a Zeiss Axiophot 1 or an Olympus BX61 microscope.

The dynamics of reversible PM protein localisation was analysed by CLSM of embryos carrying PIN1-GFP or EGFP-ENP reporter. Initial experiments revealed that higher (mM) PBA concentrations led to necrosis and death of isolated embryos, whereas lower (µM) concentrations had no effect even after hours. Empirically, the best results were obtained when ovules of 10-15 siliques were collected in water purified through Millipore columns (mQ water) and then incubated for 15 min in 25 mM PBA (ENP) or for 10 min in 30 mM PBA (PIN1); PBA was dissolved in mQ water. A small fraction of the initial collection of ovules was kept in mQ water for the duration of the whole experiment as an experimental control. After incubation, an aliquot of embryos, released from the ovules by pressing the coverslip, was checked by CLSM. The remaining ovules were washed three times with mQ water and then left for washout in mQ water for 2 h. Then, this washout fraction and the experimental control embryos were analysed by CLSM. The quality of the embryos in the experimental control was decisive for evaluating the experiment. If the quality was not sufficient (e.g. as shown by poor retention of proteins at the PM, visible necrosis), the complete experiment was repeated. For embryo and root analyses, epidermal cells were best suited.

For the analysis of roots (ENP, PIN2), initially 24 seedlings were collected in 0.5MS without sucrose (MS-S); then, 16 seedlings were incubated in 10 mM PBA for 30 min in two independent experimental series and for 60 min in another experiment. The remaining 8 seedlings remained for the duration of the whole experiment in 0.5MS-S (experimental control). After incubation, 8 seedlings were counterstained with FM4-64 and analysed by CLSM. The remaining 8 seedlings were washed three times with 0.5MS-S and then left for washout in 0.5MS-S for 2 h. The washout fraction and the experimental control were counterstained with FM4-64 and analysed by CLSM.

For CLSM analysis, we used an Olympus LSM FV100 and associated FluoView software. Images were taken of epidermal cells (embryos) and of transverse optical sections (roots) of epidermal cells in the area where the lateral root cap ends. Excitation of GFP and VENUS probes at 488 nm and 515 nm, respectively, was with a multiline argon laser (500-550 nm and 500-600 nm slit width, respectively). For embryo analyses the laser intensity was kept at 10-15%, for root images at 1% (PIN2: HV, 550; gain, 1; offset, −3) or 0.5% (ENP: HV, 500; gain, 1; offset, –3) and the images were acquired with a high-sensitivity GaAsp detector unit (Olympus). One-way scan images (Kahlman frame) were obtained using an Olympus Plan APO 60× water objective. Excitation of FM4-64 was at 561 nm and the laser intensity was kept at 0.9% (HV, 538; gain, 1; offset, 5).

We considered that accumulation of the B transporter BOR1 is controlled by the B conditions (Takano et al., 2010, 2005; Kasai et al., 2011; Yoshinari et al., 2012).

Seedlings carrying 35Sp:BOR1-GFP were surface sterilised and grown for 10 days on sterile 0.5MS plates. The subsequent transfer onto H₂O-agar plates for 2-3 days led to significant enhancement of BOR1-GFP fluorescence (Fig. 7, Figs S9 and S10). For each experiment between 8 and 24 seedlings were prechecked by CLSM. Then 8 seedlings each (only 4 seedlings in the case of experiment no.2) were transferred for 2 h into different solutions. The basic 0.5MS used for these incubations was without sucrose and without H₃BO₃. If present in the medium, B was always adjusted with H₃BO₃ to 100 µM final and PBA was adjusted to 10 mM. Collectively, four different conditions were analysed: directly after growth on H₂O-agar plates (ʻbefore') and after incubation in 0.5MS-S/+H₃BO₃, water (mQ) or 0.5MS-S/+H₃BO₃/+PBA medium. Four independent experiments were performed (Fig. S9). After incubation, the GFP fluorescence of the seedlings was documented by CLSM (laser 488 nm; intensity, 1%; HV, 530; gain, 1; offset, –3). In three experiments we also tested the response of roots incubated in 0.5MS-S/+H₃BO₃ by recultivating them on H₂O-agar plates overnight. Most of these root tips displayed stronger fluorescence than with 0.5MS-S/+H₃BO₃ incubation alone (Fig. S9).

In an additional experimental series the water (mQ) treatment was replaced by incubation in 0.5MS without sucrose and H₃BO₃ (0.5MS-S/-H₃BO₃) and further incubations were in 0.5MS-S/+H₃BO₃, 0.5MS-S/+H₃BO₃/+PBA or 0.5MS-S/+PBA (Fig. S10). Seedlings with obvious root defects were discarded from further analysis. For fluorescence intensity quantification the images were cropped using GIMP software (www.gimp.org) and intensities were measured using Fiji software (Schindelin et al., 2012). The significance of intensity differences was assessed by t-test (see below; Tables S7 and S8). For better display, brightness was increased for all cropped images to the same extent (Figs S9 and S10).

The depletion of the PM from proteins (EGFP-ENP, PIN1-GFP) is not always complete, for instance because some cells of embryos and roots are injured during the isolation and transfer procedures; also, it is unlikely that all tissue parts contact the compound to the same extent, and the same holds true for the washing steps. Therefore, we assessed the distribution of GFP signals in a semi-quantitative way by estimating the extent of cytosolic versus PM fluorescence signal for epidermal cells by eye. For each reporter gene (EGFP-ENP, PIN1-GFP) and tissue (embryo, root) at least three independent experiments were performed. Four persons (the authors, two technicians) independently rated CLSM images of embryo and root epidermal cells (see Table S6). Depending on the quality of the image, 3-12 cells per image (equals one data point) were evaluated (Table S6). Therefore, single data points in Table S6 take fractions and not positive integers. We established four intensity classes or categories for cytosol (C0-C3) and the PM (P0-P3). Instead of separating 16 different classes, we established seven different groups: P3C0 (no.1; indicates complete localisation at the PM); the combined groups P3C1/P2C0 (no.2), P3C2/P2C1/P1C0 (no.3), P3C3/P2C2/P1C1/P0C0 (no.4), P2C3/P1C2/P0C1 (no.5), P1C3/P0C2 (no.6); and P0C3 (no.7; indicates exclusive localisation in the cytosol).

For statistical analyses we used Excel (Microsoft) or Prism (GraphPad) software. The data groups of the 0.5MS-S/+H₃BO₃ versus 0.5MS-S/+H₃BO₃/PBA experiments of 35S:BOR1::GFP plants could be directly analysed (Tables S7 and S8). The data of the cytosolic versus PM comparison of EGFP-ENP and PIN1-GFP in PBA-treated versus washed material required an additional step: here, prior to further processing, the mean of three independent assessments (see above, semi-quantitative analysis of fluorescence signals) was determined (Table S6).

The data groups to be subjected to (two-tailed) t-tests were first analysed for Gaussian distribution with a Kolmogorov–Smirnov (KS) test. All data groups passed a first KS test except that of the EGFP-ENP wash experiment. For this group, a KS test with Bonferroni correction for multiple tests showed that the assumption of a normal distribution was not rejected at a significance level of 5%. Thus, all experimental comparisons could be analysed with (two-tailed) t-tests to assess whether the differences in root and embryo experiments were significant (see Figs 5, 7, Figs S9, S10, Tables S6-S8).

We thank E. Grill, E. Isono, F. Assaad and C. Schwechheimer for manuscript reading (first version); R. Jonczyk, W. Frommer, N. von Wiren and anonymous reviewers for helpful advice; F. Assaad and E. Torres Ruiz for corrections; C.-C. Schön, N. Krämer and the Plant Breeding Team at the Wissenschaftszentrum Weihenstephan for help with statistics; O. Peis and H. Miller-Mommerskamp for technical assistance; J. Friml, G. Jürgens, C. Luschnig, D. Weijers and the Nottingham Arabidopsis Stock Centre for (transgenic) A. thaliana lines. We thank A. Gierl for consistent support. Dedicated to D. Engl, V. Hemleben, R. and G. Matthes.