The PINOID protein kinase regulates organ development in Arabidopsis by enhancing polar auxin transport

The plant hormone auxin plays a crucial role in development throughout the life cycle of a plant by directing basic developmental processes such as cell division, cell elongation and differentiation. Experiments with radio-labelled derivatives of indole-3-acetic acid (IAA), the predominant naturally occurring auxin, indicated that IAA is transported downward from its main site of synthesis, the shoot tip, to the root. Treatment of seedlings or plant tissues with inhibitors of this polar auxin transport (PAT), such as naphthylphtalamic acid (NPA) or 2,3,5,-triiodo-benzoic acid (TIBA), showed that PAT provides directional and positional information for developmental processes such as vascular differentiation, apical dominance, organ development and tropic growth. To describe PAT, a chemiosmotic model was proposed in which auxin enters the cell in its protonated form through diffusion or through the action of a saturable auxin import carrier. Auxin is deprotonated at the higher pH of the cytoplasm and can only exit the cell through active export by auxin efflux carriers (AECs). The specific location of AECs at the basal side of the cell was hypothesised to be the driving force of PAT (Lomax et al., 1995).

Recent studies using molecular genetic approaches in Arabidopsis thaliana have shed new light on the molecular mechanisms behind PAT and auxin action. Important new insights were obtained through molecular characterisation of Arabidopsis pin-formed or pin1 mutants (Okada et al., 1991). These mutants develop a pin-like inflorescence, which is characteristic of wild-type plants grown in the presence of PAT inhibitors. Occasionally, flowers are produced on the inflorescence of pin1 mutant plants that have less sepals, more petals, no stamens and abnormal carpels. Some of these flowers consist only of carpelloid structures (Okada et al., 1991). Moreover, pin1 mutant embryos show defects in cotyledon number and position, a phenotype that can be mimicked by culturing plant embryos with PAT inhibitors (Liu et al., 1993). The AtPIN1 gene was cloned through transposon tagging and appeared to encode a transmembrane protein with similarity to bacterial-type transporters (Gälweiler et al., 1998). This suggested that the AtPIN1 protein represented the elusive AEC. In agreement with its proposed function as an AEC, the AtPIN1 protein was found to be localised to the basal end of xylem parenchyma and cambial cell files in the Arabidopsis inflorescence axis (Gälweiler et al., 1998).

AtPIN1 was found to be part of a multigene family in Arabidopsis comprising 8 members (Friml, 2000). Allelic loss-of-function mutants in another member of this gene family, AtPIN2, were independently isolated based on root agravitropism or ethylene resistance phenotypes (Chen et al., 1998; Luschnig et al., 1998; Müller et al., 1998; Utsuno et al., 1998). Immunolocalisation showed that the AtPIN2 protein is present at the anti- and periclinal sides of cortical and epidermal cells in the root tip (Müller et al., 1998). The distinct expression and cellular localisation of AtPIN1 and AtPIN2 combined with the phenotypes of the respective loss-of-function mutants suggested that the different members of the AtPIN family each direct distinct processes in plant development, which involve PAT.

The recent finding that lateral organs can be induced on pin1 inflorescences by exogenous application of IAA (Reinhardt et al., 2000) indicated that the IAA content in the pin1 inflorescence apex is sub-optimal for organ formation. This result suggested that continuous supply of IAA is essential for proper positioning and development of organs from the inflorescence meristem. The expression of AtPIN1 in floral organ primordia (Christensen et al., 2000), even at very young stages (Vernoux et al., 2000), suggested that polar transport of IAA is required to guarantee this supply.

Two Arabidopsis mutants that share several of the phenotypic characteristics of the pin1 mutant are monopteros (mp) and pinoid (pid) (Berleth and Jürgens, 1993; Bennett et al., 1995). mp mutants differ from pin1 and pid in that their vascular strands are disconnected and the root meristem is not formed in the embryo. However, like pin1 and pinoid, cotyledon positioning in the mp embryo is aberrant and the inflorescence carries few flowers and terminates prematurely (Przemeck et al., 1996). Flowers have fewer outer whorl organs and have abnormal carpels. In one mp allele PAT was found to be reduced, possibly through absence of a continuous vascular strand (Przemeck et al., 1996). The MONOPTEROS gene encodes ARF5, a protein with homology to the auxin responsive element binding factor ARF1 (Hardtke and Berleth, 1998). MP/ARF5 was found to be a positive regulator of auxin induced gene expression (Ulmasov et al., 1999) and likely has a function in cell axialisation and vascular development during plant development in response to auxin gradients. The phenotype of pid mutants closely resembles that of pin1 mutant plants, but is less severe. A cross between pid and the auxin resistant mutant axr1 suggested that AXR1 and PID have overlapping functions and that PID plays some role in an auxin-related process (Bennett et al., 1995). The PID gene was recently cloned and found to encode a protein-serine/threonine kinase (Christensen et al., 2000). Based on the phenotypes induced by constitutive expression of PID, Christensen and co-workers (Christensen et al., 2000) concluded that the protein kinase is a negative regulator of auxin signalling.

Here we show that PID is an auxin-responsive gene and that the main site of PID expression is the vascular tissue in young developing organs. Based on these results, and on detailed analyses of 35S::PID overexpression phenotypes, we propose that PID functions as a positive regulator of polar auxin transport.

Plant transformations were performed by floral dip (Clough and Bent, 1998). Arabidopsis seeds were surface sterilised and plated on solid M-A medium (Masson and Paszkowski, 1992), where needed containing 25 mg/l kanamycin, 15 mg/l phosphinothricine or 20 mg/l hygromycin for selection, and germinated at 21°C under a 16-hour photoperiod with 3000-4000 lux. Two- to three-week-old plants were transferred to soil and grown at 21°C under a 16-hour photoperiod and 60% relative humidity.

Poly(A) enriched RNA was isolated from root tips using the Quick Prep Micro mRNA Purification Kit (Pharmacia). Total RNA isolation and RNA blot analysis were performed as described previously (Memelink et al., 1994) and signal detection was performed by Phosphor-Image analysis (Molecular Dynamics). Whole-mount in situ localisation of PID mRNA was performed as described previously (de Almeida Engler, 1998; Friml, 2000), using a 337 bp 5′ fragment of the PID cDNA and T3- and T7-polymerase (Promega) for digoxigenin-labelled sense and anti-sense probe synthesis, respectively.

The pid::En197 and pid::En310 mutants were identified from a collection of En-1 transposon mutagenized lines by a PCR-based screen using the En-1 specific primers En205 and En8130 (Wisman et al., 1998) and PINOID specific primers PKV (5′-TCCTTTCTCTCAAACCTCACCGATCC-3′) and PKIII (5′-CGTAGAGAAACACTCCAAAGGCCCAC-3′). The presence of the pid-2 allele was detected by amplification of the PID locus using primers 1D.3 (5′-CATGCATTGACTCTGTTCAC-3′) and 1D.4 (5′-TAACATTATCTATCGTTACAGTG-3′) and digestion of the PCR product with DdeI.

General cloning and molecular biology procedures were performed as described previously (Sambrook et al., 1989) using E. coli strain DH5α. For plant transformation, binary vectors were transferred to Agrobacterium strain LBA1115 by tri-parental mating (Ditta et al., 1980) or through electroporation (Den Dulk-Ras and Hooykaas, 1995).

The PID cDNA was isolated from a cDNA library of auxin-treated root cultures of Arabidopsis ecotype C24 (Neuteboom et al., 1999). DNA sequencing was performed by Eurogentec (Belgium).

The fusion between PID and gusA was created by cloning the SphI-MspAI genomic fragment containing 3.6 kb of 5′ untranslated region and the complete PID gene, excluding the last six codons, in-frame with the gusA gene in pCAMBIA1381Xb (McElroy et al., 1995). For the sense overexpression constructs, the PINOID cDNA was cloned into the expression cassette of pART7, which was subsequently introduced as a NotI fragment onto binary vector pART27 (Gleave, 1992).

The pACT and pEF constructs containing the mGAL4:VP16 gene and UAS promoter, respectively, will be described in detail elsewhere (D. W., J. Haseloff, E. van Ryn, P. H. and R. O., unpublished). The DR5::GUS reporter was obtained by cloning a synthetic fragment containing 7 copies of the CCTTTTGTCTC sequence (Ulmasov et al., 1997) upstream of the –47 35S promoter and fusing the resulting promoter to the GFP::GUSA reporter gene (Quaedvlieg et al., 1998).

Starch granule staining was performed as described previously (Sabatini et al., 1999). To detect gusA expression, plant tissues were fixed in 90% acetone for 1 hour at –20°C, washed three times in 10 mM EDTA, 50 mM sodium phosphate (pH 7.0), 2 mM K₃Fe(CN)₆, and subsequently stained for up to 16 hours in 10 mM EDTA, 50 mM sodium phosphate (pH 7.0), 1 mM K₃Fe(CN)₆, 1 mM K₄Fe(CN)₆ containing 1 mg/ml 5-bromo-4-chloro-3-indolyl-D-glucuronide (AG Biosynth). Tissue was cleared using chloral hydrate after fixation in ethanol:acetate (3:1). GUS-stained tissues were embedded in Technovit 7100 (Heraeus, Germany) and 5 μm sections were stained with Safranin (0.05% in water) for 10 seconds, washed with excess water and mounted in Epon. GUS expression and starch staining were visualised using a Zeiss Axioplan2 imaging microscope with DIC optics. For confocal laser scanning microscopy (CLSM) roots were stained for 10 minutes in 10 mg/l propidium iodide and visualised using a Zeiss Axioplan microscope equipped with a BioRad MRC 1024 confocal laser. Microscopic images were recorded using a Sony DKC 5000 3CCD digital camera and Adobe PhotoShop software. Angles of hypocotyls and root tips towards the horizontal axis were determined using Adobe PhotoShop. Root and hypocotyl lengths were measured using NIH Image.

As part of our studies on auxin regulated gene expression, we analysed the expression and function of the putative protein-serine/threonine kinase gene PINOID (PID). This gene had previously been defined through loss-of-function mutants that form a pin-like inflorescence (Bennett et al., 1995), and was recently cloned (Christensen et al., 2000). The PID protein kinase shows significant similarity in the catalytic domain to members of the flowering plant-specific AGC group VIII of protein-serine/threonine kinases (Hanks and Hunter, 1995) (Fig. 1A). The closest member of this group in Arabidopsis shares less than 50% overall similarity with PID, indicating that PID is a single copy gene in Arabidopsis.

Northern blot hybridisation of poly(A) enriched RNA from Arabidopsis roots showed that the PID protein kinase gene is upregulated 4 hours after auxin treatment (Fig. 1B). Furthermore, induction of expression by cycloheximide and decline of expression after 20 hours of incubation with auxin (Fig. 1B) suggested that the gene belongs to the group of primary auxin response genes. Because of the auxin inducibility of the gene we looked for previously characterised auxin responsive elements in the promoter region of the gene. A single auxin responsive TGTCTC element (Ulmasov et al., 1997) was found within 500 bp of the transcription start (at position –33).

By screening En-1 transposon mutagenized lines (Wisman et al., 1998) for insertion mutations in the protein kinase gene, we obtained two new pid alleles in which a transposon disrupted the region encoding the conserved catalytic domain of the protein kinase. We named the mutants pid::En197 and pid::En310 according to the codon that was disrupted by the En-1 insertion (Fig. 1A). Mutant plants develop an inflorescence that ends in a pin-like structure and carries only a few aberrant flowers. Flowers generally contain few or no sepals and stamens, more petals, have a trumpet shaped pistil and produce no or only few seeds (Fig. 1C,F). Occasionally flowers develop with only carpelloid structures (Fig. 1D). Approximately 50% of the mutant seedlings showed abnormal cotyledons, with three cotyledons being the most common phenotype (Fig. 1E). The penetrance of the abnormal cotyledon phenotype indicated that the pid::En mutants represent strong loss-of-function alleles (Bennett et al., 1995; Christensen et al., 2000).

PID mRNA is most abundant in young flower buds (not shown). Expression in both roots and shoots of seedlings is low but can be induced by auxin and cycloheximide treatment (Fig. 1B and not shown). To obtain a reliable impression of the spatial and temporal distribution of PID gene expression and the cellular localisation of the PID protein, we fused the complete gene, including the 3.8 kb 5′ untranslated region, but excluding the last six codons, in-frame to the gusA reporter gene on pCAMBIA1381Xb (PID:GUS). Multiple lines were generated in ecotype Columbia (Col) that expressed the 4 kb PID:GUS transcript and showed the same β-glucuronidase (GUS) expression pattern. Transformation of the empty pCAMBIA1381Xb vector did not result in a detectable GUS expression. One representative line, PID:GUS-18, was selected for further analysis.

The pid-2 mutant (Bennett et al., 1995) was crossed with the PID:GUS-18 line and F₂ progeny were tested for complementation of the intermediate pid-2 allele by the PID:GUS gene. Among 301 F₂ seedlings only 5 developed abnormal cotyledons, a phenotype that showed 20% penetrance in the pid-2 mutant. The data were significant for goodness of fit (χ²=0.4, P>0.5) with the 1:80 ratio expected for complementation of the pid-2 mutation by the PID:GUS transgene. After transfer to soil, F₂ individuals were checked by PCR for the presence of the pid-2 allele and the PID:GUS construct. Plants homozygous for the pid allele developed the typical pid inflorescence, whereas plants that were homozygous for the pid allele but also contained the PID:GUS construct developed a wild-type inflorescence (not shown). These results proved that the PID:GUS fusion protein restores normal growth to pid mutant plants and therefore has wild-type PINOID function. More importantly, the complementation of pid-2 by PID:GUS showed that the GUS activity in line PID:GUS-18 is likely to reveal the spatial and temporal expression and cellular localisation of the endogenous PID protein kinase.

GUS expression in PID:GUS-18 seedlings was mainly localised to the vascular tissue and was strongest in regions of vascular differentiation proximal to the meristems and lateral root primordia (Fig. 2A,C,E). A cross section of the hypocotyl just below the shoot apical meristem showed that GUS activity is present in the xylem parenchyma cells and in the endodermis around the vasculature (Fig. 2G). Moreover, closer examination of the sub-cellular localisation of the GUS signal in untreated and auxin-treated seedlings suggested that PINOID does not accumulate in the nucleus (Fig. 2H and data not shown). Treatment of seedlings with 5 μM IAA induced a significant increase of expression in vascular tissue and leaf primordia (Fig. 2B,D,F). In the inflorescence, expression was detected in anther primordia, in the vasculature of the growing flower stalk, of young pedicels and bracts (Fig. 2I,J,K,L) and of developing sepals, but not in petals (Fig. 2M). In pistils, PID was transiently expressed in the vasculature of the style and the septum (Fig. 2M), in the integuments and funiculus of the developing ovule (Fig. 2N) and in the cotyledon primordia of embryos (Fig. 2O).

PID:GUS expression in flowers and embryos corroborates the in situ localisation of PID mRNA in these tissues by Christensen and colleagues (Christensen et al., 2000). This expression pattern correlates with the phenotypic changes observed in the loss-of-function mutants, i.e. altered cotyledon and floral organ numbers and altered pistil morphology (Fig. 1C,D,E,F), and confirms the importance of PID for the development of these organs. The localisation of PID expression in the vasculature and in leaf primordia was confirmed by whole-mount mRNA in situ hybridisation on seedlings (Fig. 3A,D), although expression of PID, as detected by in situ hybridisation in the shoot, is less pronounced compared to the expression of the PID:GUS fusion gene.

The phenotype of the pid loss-of-function mutants (Bennett et al., 1995) and the auxin responsive expression pattern of PID imply a role for the PID protein kinase in regulating auxin action. To further define this role, the PID cDNA was cloned in sense orientation behind the strong Cauliflower Mosaic Virus 35S promoter (35S::PID) and introduced into Arabidopsis thaliana ecotypes Col and C24. Of each ecotype, three lines with different levels of PID overexpression were selected for further analysis (Fig. 4A,B). One of these lines, C24-6, was exceptional in that it did not express PID at a detectable level (Fig. 4). Instead of a clear overexpression phenotype, adult plants developed inflorescences typical for pid mutants and 8% of the seedlings of this line developed abnormal cotyledons. The pid mutant phenotypes combined with the non-detectable overexpression indicated that this line represented a partial loss-of-function mutant due to silencing of the endogenous PID gene by the multiple transgene locus present in this line.

Seedlings of 35S::PID lines showed agravitropy and reduced elongation growth of roots and hypocotyls (Fig. 4C). The severity of the phenotype corresponded to the level of overexpression in each case (Fig. 4A,B) and was reproducible in subsequent generations. Lateral root formation was delayed in 35S::PID seedlings and interestingly, we observed collapse of the main root meristem within a few days after germination. The remaining root tip consisted of only a few layers of large elongated cells, several of which showed epidermal identity as evidenced by the development of root hair structures (Fig. 5A,B). Staining for starch granules indicated the absence of cells with columella identity (Fig. 5B,C). Lateral roots emerged following the disintegration of the primary root meristem (Fig. 5D). These lateral roots were again agravitropic, but their meristems showed normal patterning (Fig. 5E) and collapse of these meristems was not observed.

Flowering plants of lines with high levels of 35S::PID expression showed reduced apical dominance, developing a primary inflorescence with two or three lateral inflorescences (Fig. 5F,G). Also, the emergence and development of axillary inflorescences was significantly enhanced in these plants.

The phenotype of pid loss-of-function mutants most closely resembles that of the pin1 mutants, which are proposed to be blocked in PAT owing to the absence of a functional AEC (Gälweiler et al., 1998). This led us to hypothesise that the PID protein kinase acts by regulating PAT. To further examine this possible role for PID, we studied the effect of PAT inhibitors on root meristem collapse. Seeds of wild-type Arabidopsis and the 35S::PID lines Col-10 and Col-21 were germinated on medium containing 0.1 or 0.3 μM naphthylphtalamic acid (NPA). These specific concentrations were used because such treatments exerted only mild effects on wild-type root development, as observed by root growth, lateral root initiation (Fig. 6A,B) and patterning of the root meristem (Fig. 6E,F). Moreover, the expression of an auxin responsive DR5::GUS reporter indicated that only minor changes occurred with regard to auxin distribution, or sensitivity, in roots treated with these NPA concentrations (Fig. 6I,J). Growth of 35S::PID seedlings on NPA significantly increased root elongation (Fig. 6C,D) and prevented collapse of the primary root meristem (Fig. 6G,H). Similar results were obtained when seedlings were grown on TIBA, which belongs to a different class of PAT inhibitors. Clearly, suppression of PAT by inhibitor concentrations that only mildly interfere with the development of wild-type seedlings was sufficient to rescue the organisation of the primary root meristem in 35S::PID lines.

We introduced the auxin responsive DR5::GUS reporter gene into the 35S::PID back ground and confirmed the observation by Christensen and co-workers (Christensen et al., 2000) that GUS expression in root tips of young 35S::PID seedlings is significantly reduced (Fig. 6K). In view of the auxin efflux enhancing activity of PID, this result can be explained by lower auxin levels in root tips of 35S::PID seedlings. Interestingly, reduction of auxin efflux by NPA treatment partially restored the level of DR5::GUS expression in root tips of these seedlings (Fig. 6L).

More detailed analysis showed that root meristem collapse in the strong overexpression line 35S::PID Col-21 was observed by 3 days after germination in some seedlings, and after 7 days in most (Fig. 6M). Meristem disintegration also occurred in line 35S::PID Col-10, which expresses an intermediate level of the 35S::PID transgene, but in only up to 50% of the seedlings and with significantly delayed timing. Primary root meristems of the majority of the 35S::PID Col-21 seedlings were rescued with 0.3 μM NPA. This concentration was critical, since at a slightly lower concentration (0.1 μM) the rescue was only partial for high expressing line 35S::PID Col-21, whereas rescue was almost complete for the intermediate expressing line 35S::PID Col-10.

The p35S-driven PID overexpression typically inhibits lateral root development in young seedlings, but the inhibition is overcome by collapse of the primary root meristem. Since root meristem disintegration occurs earlier and more frequently in the high expressing 35S::PID lines, seedlings of these lines generally develop a more vigorous root system. In the presence of NPA however, the meristem is rescued and lateral root formation remains suppressed. To quantify this prolonged suppression by NPA, we grew wild-type and 35S::PID Col-10 and Col-21 seedlings for 14 days on 0.3 μM NPA and determined the length and the number of lateral roots per primary root. NPA-rescued primary roots of both 35S::PID lines were similar in length, but significantly shorter than those of wild-type seedlings (P<0.001; Fig. 6N). Although the inhibition of lateral root formation was maintained for both 35S::PID lines in the presence of NPA, it was apparent that Col-21 seedlings developed significantly more lateral roots than those of line Col-10 (P<0.001; Fig. 6N). Apparently, high, as opposed to intermediate, levels of p35S-driven PID expression promote branching of primary roots. These findings clearly do not fit with a role of PID as a negative regulator of auxin signalling (Christensen et al., 2000), which implies a negative correlation between root branching and the level of 35S::PID expression. Instead, our observations are more readily explained by general enhancement of auxin efflux due to ectopic PID expression, which results in dynamic changes in auxin distribution during development of the 35S::PID seedlings.

Participation of PID in the regulation of PAT implies that PID does not only act locally in tissues where it is expressed, but also exerts its effect over a distance. To test this, we analysed the effects of tissue-specific PID expression using a GAL4-based transactivation-reporter system (D. W., J. Haseloff, E. van Ryn, P. H. and R. O., unpublished). This system enabled us to study the effects of the ectopic expression of PID and to simultaneously localise PID expression through the observation of GFP:GUS reporter transactivation. An activator plant line ACT-LTP1 containing the epidermis-specific promoter of the Arabidopsis thaliana Lipid Transfer Protein 1 (LTP1) gene fused to the GAL4:VP16 gene was crossed with an effector line (EF-PID) harbouring both the PID cDNA and the GFP:GUS reporter under control of the GAL4-dependent UAS promoter. F₂ seedlings were germinated without selection and 11- to 12-day-old seedlings were stained for GUS expression. Segregating at the expected 9:7 ratio for GUS-positive and GUS-negative seedlings, the F₂ population clearly displayed significantly enhanced lateral root formation by LTP1 promoter-driven PID expression (Table 1, exp. 1). A striking observation, especially since LTP1 promoter activity is confined to the aerial parts of seedlings (Fig. 7A), except for expression in the young epidermis of lateral roots. Since pLTP1 activity in lateral roots is not detectable before a stage VI primordium (Malamy and Benfey, 1997) (Fig. 7B,C), it is unlikely that this local pLTP1-driven PID expression is the cause of the enhanced lateral root formation.

Lateral root induction is known to rely on PAT from the shoot into the root (Reed et al., 1998; Casimiro et al., 2001). We therefore hypothesised that the pLTP1 driven PID expression would result in enhanced auxin efflux from the epidermis in the shoot, thereby inducing an increased transfer of auxin from the shoot into the root. Local application of the auxin transport inhibitor NPA to the transition zone of the seedlings reduced the difference between GUS+ and GUS– seedlings to a non-significant level, confirming that the increase in lateral roots is caused by enhanced efflux of auxin from the shoot into the root (Table 1, exp. 2). These data support a function of the protein kinase in enhancing polar auxin transport.

The PID locus in Arabidopsis thaliana was initially defined by a series of 8 allelic mutants exhibiting the inflorescence phenotype of few aberrant flowers and a terminal pin-like structure. Depending on the severity of the mutant allele, up to 50% of mutant seedlings show aberrations in cotyledon number and/or separation (Bennett et al., 1995). Recently, Christensen et al. (Christensen et al., 2000) reported that the PID gene encodes a protein-serine/threonine kinase that regulates auxin response. We show that PID is an auxin-responsive gene, we provide important new insights into the expression of PID throughout Arabidopsis development and we present data that further our understanding of the role of the PID protein kinase as a mediator of auxin action in plant development.

PID expression was observed in the cotyledon primordia of embryos, in young leaves and in young floral organs (Christensen et al., 2000; our observations). This expression pattern corroborates the function of PID in the regulation of aerial organ development and cotyledon positioning and separation. PID expression in vascular tissues initially suggested that PID might direct vascular development in young developing organs. However, mild vascular defects were only observed in the aberrant flowers and not in other organs of pid loss-of-function mutants (Christensen et al., 2000) (our observations), indicating that the regulation of vascular development is not the primary task of PID. More likely, based on the interpretation of the loss-of-function mutant phenotype and gene expression, PID is probably involved in determining the position and outgrowth of cotyledon, leaf, flower and floral organ primordia. The importance of auxin and the involvement of PIN1 in positioning and outgrowth of lateral aerial organs was demonstrated recently (Reinhardt et al., 2000; Vernoux et al., 2000).

In contrast to the clear developmental defects observed in aerial organs, the effects of pid loss-of-function mutations on root development were not very obvious. However, we did observe an irregular root waving pattern in pid loss-of-function mutants (not shown), while 35S::PID overexpressors had strong alterations in root gravitropism and development. This suggests that some PID functions do indeed influence auxin-mediated processes in roots.

Based on two phenotypes of the 35S::PID lines, decreased expression of the auxin responsive DR5::GUS reporter and reduced lateral root initiation even upon exogenous application of auxin, Christensen et al. (Christensen et al., 2000) concluded that PID acts as a negative regulator of auxin signalling. However, the new observations that we have made argue against such a function for PID. DR5::GUS expression is reduced in young roots of the 35S::PID lines, but we found that expression of the reporter was clearly present in the root vasculature after collapse of the primary root meristem or in roots of older plants (not shown), suggesting that auxin signalling to this promoter is not impaired, but rather, prevented in young primary roots. Another auxin-dependent process that is initially perturbed in 35S::PID lines is lateral root formation. However, 14-day-old seedlings of the high expressing 35S::PID line, Col-21, developed more lateral roots than those of the intermediate expressing line, Col-10, indicating that auxin signalling leading to lateral root induction is not repressed by PID (over)expression. Moreover, roots of 35S:PID lines are clearly as sensitive to exogenously applied auxins as wild-type roots (Christensen et al., 2000) (our observations). Reduced sensitivity of root elongation to exogenously applied auxin has been one of the major criteria for distinguishing mutations in components or regulators of auxin signalling (e.g. axr1). In conclusion, none of the observations provides sufficient evidence that auxin signalling in 35S::PID roots is perturbed and, although we do not exclude an involvement of the PID protein kinase in auxin signalling, we consider the available data indicating a role for the protein kinase as a positive regulator of PAT.

The pin-shaped inflorescences and the aberrant cotyledons and flowers of the pid mutants closely resemble those of pin-formed mutants. The cotyledon and inflorescence phenotypes can be mimicked by treatment with auxin transport inhibitors of globular stage embryos or mature plants, respectively (Okada et al., 1991; Liu et al., 1993; Hadfi et al., 1998). Recently it was shown that some aspects of the aberrant flower development in pid and pin1 mutants can be mimicked by spraying flowers with the same inhibitors (Nemhauser et al., 2000). Moreover, expression studies show that PID co-localises with AtPIN1 in the xylem parenchyma cells of the vascular tissue (our observations) (Gälweiler et al., 1998). All these data strongly suggest that PID acts in concert with AtPIN1 in the vasculature, or with other AtPINs in other tissues, to positively regulate PAT. If so, loss-of-function pid mutants would be expected to show reduced levels of PAT and PID function would be sensitive to PAT inhibitors.

Indeed, auxin transport seemed reduced in the inflorescence stems of loss-of-function pid alleles (Bennet et al., 1995) (our observations). However, PAT levels in inflorescence stems are determined by many indirect factors, such as the presence and number of developing aerial organs (Bennett et al., 1995; Oka et al., 1998) and proper development of the vascular tissue (Carland and McHale, 1996; Przemeck et al., 1996), which indicates that conclusions from direct transport measurements should also be confirmed by other functional tests. Two important observations support the role of PID as a regulator of PAT: (i) the PAT inhibitor sensitivity of PID action and (ii) the fact that, spatially, PID expression and the resulting effects on plant development do not necessarily overlap.

The NPA/TIBA-sensitivity of PID action was first demonstrated by the rescue of root growth of 35S::PID seedlings by low doses of these PAT inhibitors. We propose a model to explain the 35S::PID phenotype (Fig. 8), in which the 35S promoter-mediated PID expression preferentially enhances downward-directed PAT through the axis of the seedlings. This canalisation of auxin depletes the more peripheral tissue layers of auxin, which leads to reduced elongation and agravitropy of the hypocotyl and delays lateral root formation. Indeed, increasing endogenous auxin levels by growing the seedlings at 28°C (Gray et al., 1998) resulted in enhanced hypocotyl elongation of 35S:PID seedlings (not shown), thereby confirming that IAA concentrations are sub-optimal for hypocotyl elongation. In our model, ectopic PID expression results in enhanced efflux of auxin from the root meristem (Fig. 8), thereby resolving the proposed auxin maximum as organiser of the root meristem (Sabatini et al., 1999) and thus leading to collapse of the meristem. The sub-optimal IAA levels in the root tip explain the reduced pDR5::GUS expression in 35S::PID background roots and the slow and agravitropic growth of these roots. The collapse of the primary root meristem in 35S::PID seedlings alleviates auxin depletion of the root and thereby releases the initial delay in lateral root formation. This is in line with previous findings that removal of the root meristem or the root cap enhances the formation of lateral root primordia and the emergence of lateral roots (Torrey, 1950; Reed et al., 1998; Tsugeki and Fedoroff, 1999). In the presence of NPA at sub-micromolar concentrations, elevated PAT is slowed down in roots of 35S::PID seedlings, allowing for maintenance of the organising auxin maximum (Sabatini et al., 1999), rescue of the root meristem and partial rescue of the gravitropy and growth of the root. The recent finding that PAT inhibitor treatments induce an accumulation of IAA in the root tip (Casimiro et al., 2001) is consistent with this model.

Both the PAT inhibitor-sensitivity of PID action and the fact that PID expression does not necessarily have a local effect on development were demonstrated by the LTP1 promoter-mediated ectopic expression of PID in the aerial part of seedlings (Thoma et al., 1994) (Fig. 7). The resulting significant increase in lateral root formation could be reduced by application of NPA at the transition zone between shoot and root, indicating that pLTP1-mediated PID expression increases auxin transfer from the shoot into the root (Fig. 8).

Based on the data presented above we propose that the protein kinase PID is a positive regulator of PAT. The inducibility of the PID gene by auxin suggests that PAT is enhanced by accumulation of PID in the cell in response to an increase in local IAA levels. Indications for regulation of PAT by auxin are provided by the fact that PAT levels in inflorescence stems are significantly reduced when developing flower buds or siliques, the presumed source tissues of IAA, are removed or absent (Bennett et al., 1995; Oka et al., 1998), by the report that auxin efflux is enhanced in the auxin overproducing sur1 mutant (Delarue et al., 1999) and from the observed twofold increase of acropetal [³H]IAA transport through the root in the presence of cold IAA (Rashotte et al., 2000). The phenotypic characteristics of the pid loss-of-function mutants suggest that the PID function is essential for proper cotyledon positioning and development, for maintenance of the inflorescence meristem, for whorl definition during flower development and it is important for wild-type root growth. Enhancement of PAT may be necessary to prevent feed-back regulation of auxin biosynthesis due to accumulation of auxin, and may possibly guarantee a continuous source-to-sink transport of auxin that is essential for the organisation of, and differentiation in, these developing organs. One important step in elucidating the mechanism of the PID-mediated enhancement of PAT will be to investigate interactions between PID and the putative AECs.

The pid-2 mutant was kindly provided by the Arabidopsis Biological Rescource Center at Ohio State University. We thank Bert van der Zaal for helpful comments and for providing the root-specific cDNA library, Sacco de Vries for providing the LTP1 promoter, Andrew Gleave for sending the constructs pART7 and pART27, Elvira Baumann and Ellen Wisman for help and advice during screening of the collection of En-1 mutants (MPI, Cologne), Gerda Lamers and Celia Vogel for technical assistance, Peter Hock, Adri ’t Hooft and Peter van Mulken for artwork, Jir′i Friml and Enrico Scarpella for helpful discussions and for help with in situ mRNA hybridisations and transformation of pCAMBIA1381Xb respectively, and Keithanne Mockaitis, Jan Traas, Klaus Palme and Kim Boutilier for useful comments on the manuscript. D. W. was supported by the Research Council for Earth and Life Sciences (ALW) with financial aid from the Netherlands Organisation for Scientific Research (NWO).