A role for the rice homeobox gene Oshox1 in provascular cell fate commitment

The vascular tissues of plants are organised in a network of interconnected cell files whose continuity throughout the plant body enables efficient long-distance transport and provides mechanical support (Howell, 1998). The vascular system is composed of two types of specialised conducting tissues: the phloem, through which photoassimilates are transported, and the xylem, which is the conduit for water and soil-derived nutrients. The vasculature originates from embryonic tissues, the vascular meristems, whose cells retain the ability to continually multiply (Aloni, 1987). New vascular strands develop in dynamic relationship to one another during a differentiation process that persists throughout plant life, ensuring flexibility to adapt to changes in the surrounding environmental pressure (Sachs, 1993; Aloni, 1995). Within the vascular meristems two stages are distinguished: the procambium, or provascular tissue, and the (vascular) cambium. The procambium is the apical meristem that gives rise to primary xylem and phloem in young parts of shoots and roots (Shininger, 1979). The cambium, rare in monocotyledons, is a lateral meristem that develops in older parts of the plant axis, where it produces secondary phloem and xylem. Finally, vascular elements can occasionally differentiate from parenchyma cells, to regenerate an interrupted vascular connection during adventitious root development, or as a response to wounding or grafting (Sachs, 1981).

Most studies of vascular development have been mainly restricted to an anatomical approach, which has led to a strictly structural definition of procambium, according to which this tissue becomes recognisable as continuous strands of dense narrow cells, elongated parallel to the longitudinal axis of the organ (Nelson and Dengler, 1997). In older parts of the plant they become more vacuolated, but retain their elongated shape. Until overt cytodifferentiation occurs, procambial cells can only be described by operative criteria such as cell shape and vacuolation, which are variable and not necessarily correlated with their real developmental state. Indeed, we currently do not know whether such cytologically identifiable provascular cells have actually progressed towards differentiation into xylem or phloem, or whether they are simply uncommitted meristematic precursors.

It is widely accepted in the animal field to divide the process of cell fate commitment into two stages that are operationally defined, specification and determination (Slack, 1983). A cell or tissue explant is said to be specified to become a particular structure if it will develop autonomously into that structure when cultured in a neutral (non-inductive) environment. Determination involves a further restriction of developmental potential that is maintained in explants irrespective of the range of conditions present in the organism, and that persists stably in the course of normal development. Nevertheless, this process is not necessarily irreversible, since alterations in the developmental state can occasionally occur even after differentiation is completed, wherever cells have retained their genetic totipotency (Maclean and Hall, 1987). The concepts of cell fate specification and determination are extendible to plants, even though it is often difficult to discriminate these stages experimentally. To date, no criteria have been set to distinguish different stages in the sequence of events occurring during provascular cell fate commitment. Furthermore, only a few reports have addressed the identification of molecular or biochemical markers to describe this process (Gahan, 1981; Demura and Fukuda, 1994), and little is known about the regulatory proteins involved.

Recent reports have indicated that certain transcription factors of the homeodomain leucine zipper (HD-Zip) class may fulfil control functions in provascular and vascular development. HD-Zip proteins are encoded by a class of homeobox genes apparently unique to plants (Ruberti et al., 1991). Characteristically, they have a leucine zipper dimerisation motif linked to their DNA-binding domain. The HD-Zip class of transcription factors has been grouped in families I to IV (Sessa et al., 1994). Whereas HD-Zip IV proteins appear to be involved in establishing epidermal cell fates (Rerie et al., 1994; Di Cristina et al., 1996; Lu et al., 1996), specific members of the other HD-Zip families have been associated with (pro)vascular development. The Arabidopsis IFL1/REV gene encodes an HD-Zip III protein that appears to be implicated in differentiation of interfascicular and xylary fibres, and of tracheary elements (Zhong and Ye, 1999; Ratcliffe et al., 2000). Furthermore, the Arabidopsis Athb8 gene, which encodes another HD-Zip III protein, is expressed exclusively in procambial cells and during revascularisation following wounding (Baima et al., 1995). Finally, the tomato HD-Zip I protein Vahox1 is encoded by a gene that is specifically expressed in the phloem during secondary growth (Tornero et al., 1996).

In the current report we provide experimental criteria to distinguish progressive stages of provascular cell fate commitment, and identify the rice HD-Zip II gene Oshox1 (Meijer et al., 1997) as the first marker of a specification phase in procambium ontogenesis. Furthermore, we present evidence suggesting that this gene acts as a regulatory switch during the reversible transition between a totipotent meristematic ground state and an intermediate developmental condition at which procambial cell fate is specified but not yet stably determined.

Oryza sativa (L.) Indica cultivar IR58 total DNA, partially digested with Sau3AI, was cloned in λGEM11 (Promega) BamHI arms. Plaques of the amplified genomic library were lifted and hybridised with a radiolabelled Oshox1 cDNA fragment (Memelink et al., 1994). From two purified positive clones, three overlapping fragments hybridising to different regions of the Oshox1 cDNA were isolated, subcloned in pBluescript II SK(+) (Stratagene) and sequenced by the Double Strand sequencing service of Eurogentec. This revealed that in addition to the entire Oshox1 open reading frame, the clones contained 1596 bp of upstream sequences. To allow fusion of the Oshox1 promoter-leader region to the β-glucuronidase reporter gene (gusA), the sequence around the start codon (CAATGG) was modified to an NcoI site (CCATGG) by standard PCR techniques. Subsequently, the 1596 bp Oshox1 promoter-leader sequence was fused to the gusA start codon in binary vector pCAMBIA1391z (CAMBIA; Roberts et al., 1997), to obtain the Oshox1-GUS reporter construct. The 35S-Oshox1-nos expression cassette was excised from pMOG-Oshox1 (Meijer et al., 1997) and cloned between the EcoRI and HindIII sites of binary vector pMOG22 (Zeneca Mogen), to obtain the 35S-Oshox1 construct.

Embryonic calli were induced on scutella from germinated seeds of Oryza sativa (L.) Japonica cultivar Taipei 309 (Rueb, 1994) and transformed with Agrobacterium tumefaciens strain LBA1119 (C58 pTiBo542 ΔT-DNA) harbouring the desired binary vector, essentially as described by Hiei et al. (1994). Regenerated transgenic plantlets were transferred to the greenhouse and grown in hydroponic culture with a regime of 12 hours light, 28°C, 85% relative humidity and 12 hours dark, 21°C, 60% relative humidity. If not otherwise indicated, seeds were surface sterilised (Rueb et al., 1994) and germinated in the dark at 28°C for 4 days on filter paper moistened with water (wild type) or with a 75 mg/l hygromycin solution (transgenic). Germinated seeds were grown in a 12 hours light:12 hours dark cycle at 28°C. 35S-Oshox1 transgenic rice lines were selected on the basis of their resistance to hygromycin and of their Oshox1 overexpression at the RNA level (not shown). Analyses were performed on 9 independent 35S-Oshox1 lines. Photographs of wild-type and transgenic plants were taken with an Olympus CAMEDIA Digital Camera C-1400 L and processed using Adobe Photoshop 5.0.

To exclude any influence of transgene position on promoter activity, analyses were performed on 10 independent Oshox1-GUS lines. To assay β-glucuronidase (GUS) activity, 100 μm vibratome sections (Leica VT1000S) or freshly dissected plant organs were vacuum infiltrated for 5 minutes and subsequently incubated in 100 mM phosphate buffer pH 7.7 containing 2 mM 5-bromo-4-chloro-3-indolyl-β-D-glucuronic acid (X-gluc; Biosynth AG), 0.5 mM potassium ferricyanide, 10 mM EDTA and 0.1% Triton X-100 at 37°C, for 1 hour to overnight depending on staining intensity. Samples were depleted of chlorophyll in 70% ethanol and stored at 4°C. As a control for reliability of the GUS assay, CaMV 35S-GUS transgenic plants (pCAMBIA1301 binary vector; CAMBIA; Roberts et al., 1997) showed reporter gene activity in virtually all cell types of different organs, thereby demonstrating that substrate accessibility was not limiting GUS detection. GUS activity was never observed in non-transgenic rice tissues. To verify that the GUS staining patterns were not influenced by tissue metabolism or by leakage of GUS enzyme or 5-bromo-4-chloro-3-indole intermediate, control experiments were performed in which tissues were prefixed and/or in which detection was performed in different buffer systems. Prefixation in 2% glutaraldehyde in 100 mM cacodylate buffer pH 7.2 for 2 minutes on ice prior to GUS detection (Stomp, 1992) resulted in identical staining patterns as in standard GUS assays, although staining intensity was much reduced. The same was true when pretreatment with MS medium containing 1 mM spermidine was used as an alternative to fixation (De Block and Van Lijsebettens, 1998). Finally, a Tris-based reaction buffer (De Block and Van Lijsebettens, 1998) gave the same results as the standard buffer.

Fresh tissues or rehydrated GUS-stained samples were fixed overnight in 2% glutaraldehyde in 100 mM cacodylate buffer pH 7.2. Samples were then dehydrated in a graded ethanol series, embedded in Historesin (Leica) according to manufacturer’s instructions, and sectioned with a LKB ultramicrotome. Sections (10 μm) were dried onto slides at 37°C and stained with 0.1% Toluidine Blue in 0.5 M acetate buffer pH 4.4 (root apices), or counterstained with 0.5% Safranin O in water (GUS-stained material), before mounting in epoxy resin for microscopic observation. Vibratome sections (100 μm) of 10-day-old dark-grown wild-type and 35S-Oshox1 seedlings were stained with Toluidine Blue as above, and mounted in 80% glycerol for microscopic observation. Samples were viewed using a Leica MZ12 stereomicroscope or a Leitz Diaplan microscope with bright-field optics settings. Images were acquired with a Sony 3CCD Digital Photo Camera DKC-5000 and processed using Adobe Photoshop 5.0.

Seedlings (1- and 2-week-old) were transferred to the dark at 28°C 16-24 hours before harvesting to reduce their starch content. Root apices (1 cm) and seedling culm basal region (0.5 cm), containing the shoot apex, were fixed overnight in FAA (50% ethanol, 5% acetic acid, 3.7% p-formaldehyde). Samples were dehydrated in a graded ethanol series, after which ethanol was replaced stepwise with xylene. Infiltration and embedding in Paraplast Plus (Sigma) were performed as described by Sylvester and Ruzin (1994). Inflorescences (18 cm, approx. 8 days before heading) were fixed in 4% p-formaldehyde, 10 mM DTT, 1× PBS buffer. Dehydration, infiltration and embedding in butyl-methyl methacrylate (BMM) were performed as described by Baskin et al. (1992). Samples were sectioned on a Leitz rotary microtome (8 μm, Paraplast Plus) or on a LKB ultramicrotome (6 μm, BMM) and sections transferred onto poly-L-lysine coated slides. Linearised plasmid templates were used to synthesise digoxigenin-labelled antisense and sense riboprobes according to manufacturer’s instructions (Boehringer Mannheim), except that 20 units of RNasin (Promega) were included in the reactions. In situ hybridisation with hydrolysed probes was performed as described by Langdale (1994) except that sections were treated with 1 mg/l of proteinase K for 30 minutes at 37°C and that RNasin (Promega) was added to the hybridisation mix (0.4 units/μl). After detection, sections were dehydrated through a graded ethanol series, transferred to xylene and mounted in Eukitt (O. Kindler GmbH & Co.). For whole-mount in situ hybridisation, mature embryos were prefixed as described by Engler et al. (1998), embedded in 5% agarose, bisected longitudinally through the median axis and processed according to the method of Engler et al. (1998). Identical hybridisation patterns were obtained with riboprobes of the full-length Oshox1 cDNA and of the 0.4 kb 5′ region lacking the conserved homeobox sequence. If not otherwise indicated, hybridisation signal was not observed with control sense RNA probes.

The tissue inbetween two longitudinal vascular bundles of fully expanded leaves from greenhouse-grown wild-type and Oshox1-GUS plants was wounded using a needle. Attention was paid not to damage the vasculature, as this could lead to leakage of GUS and misinterpretation of the results. Wounded leaves were either left attached to the plant, or removed and incubated in 100 mM sodium phosphate buffer pH 7.7 in the greenhouse, which gave identical results. At different time points the wounded leaves were sampled and immediately subjected to GUS analysis (Oshox1-GUS) or longitudinally hand-sectioned and processed for histochemical detection of lignin with acidic phloroglucinol (wild type; Schneider, 1981).

Surface-sterilised seeds of Oshox1-GUS plants were sown on filter paper moistened with the following solutions: 5 μM indole-3-acetic acid (IAA), 1 μM each of naphthalene-1-acetic acid (NAA), naphthalene-2-acetic acid (β-NAA) and 1-N-naphtylphthalamic acid (NPA), 0.25 μM 2,4-dichlorophenoxyacetic acid (2,4-D), 500 mM each of sucrose and mannitol, 0.01 and 2.5 mM p-chloromercuribenzenesulfonic acid (PCMBS), 10 μM N6-benzyladenine (BA), 4 μM 22(S),23(S)-homobrassinolide and 6.8 μM uniconazole. All solutions were prepared in water and filter-sterilised, except for brassinolide and uniconazole, which were dissolved in dimethyl sulfoxide (DMSO). Control seeds were sown on water or DMSO at the final concentration as for chemical treatments. Seedlings were sampled after 1 week and subjected to GUS analysis.

Root apices (0.2, 0.5 and 1 mm, including the root cap) were excised aseptically from 2-week-old wild-type, Oshox1-GUS and 35S-Oshox1 rice seedlings under a dissecting microscope fitted with a calibrated eyepiece micrometer. Excised tips were placed on filter paper moistened with modified MS medium (Philips and Dodds, 1977), containing 2% sucrose and no hormones. After 9 days of culture at 28°C in the dark, the length of the segments was recorded and GUS analysis performed. For root tip regeneration experiments, seedlings were returned to the dark at 28°C and GUS analysis was performed 1 week after surgery.

Transport of [³H]IAA (25 Ci/mmol, Amersham) was measured in the most distal 2 cm of seminal roots (region in which lateral root primordia are absent) of 5-day-old vertically grown wild-type and 35S-Oshox1 seedlings, and in the 2 cm of mesocotyl immediately below the coleoptilar node of 5- and 10-day-old wild-type and 35S-Oshox1 seedlings. Excised tissues were placed in agar blocks containing 10⁻⁷ M [³H]IAA, in the presence or absence of 10⁻⁶M NPA, and incubated for 2.5 (mesocotyls) or 3 (roots) hours in the dark at room temperature. To prevent drying, the tissues were overlaid with silicon oil. After incubation, the explants were cut in 3 segments: a basal part (5 mm), which was the one enclosed in the agar, a middle (5 mm) and a top (10 mm) part. Radioactivity in the segments was counted after overnight incubation in a scintillation liquid. As a control for simple diffusion of [³H]IAA in mesocotyls, transport in the acropetal direction was used. Since this could not be used for the roots where active transport occurs in both directions, a diffusion factor (DF) was calculated as the ratio between the radioactivity accumulated in the presence of NPA in the middle and basal segment. Subsequently, the amount of diffused [³H]IAA in the middle segment in the absence of NPA was calculated as the product of the DF by the radioactivity measured in the basal part in the absence of NPA. Finally, active transport was calculated as the difference between the radioactivity measured in the middle segment in the absence of NPA and the diffused [³H]IAA in the middle segment in the absence of NPA. The same calculations were done for the basipetal transport in the mesocotyl, resulting in an amount of auxin active transport comparable to that calculated using the acropetal transport as a measure of simple diffusion of [³H]IAA. Since 35S-Oshox1 tissues showed a reduced sensitivity to NPA, wild-type DFs were also used for the transgenics. Incubation times were optimised in order to prevent accumulation of [³H]IAA in the top segment of the explants, which would eventually result in the achievement of a steady-state equilibrium in the diffusion process. This expedient enables the diffusion component occurring in the direction opposite to that of the polar auxin transport to be ignored, thus simplifying considerably the calculations and avoiding the introduction of further sources of error. The data obtained were tested for significance of the difference between wild-type and 35S-Oshox1 populations by repeated-measures analysis of variance (two-factor ANOVA with replication).

Previous RNA blot analysis showed that the Oshox1 gene is expressed in all plant organs at different developmental stages (Meijer et al., 1997). To further characterise its expression pattern, a reporter gene fusion strategy was employed. To this aim, the Oshox1 genomic sequence was isolated (Accession No. AF211193), showing that this gene is composed of three exons interrupted by two introns of 88 and 95 bp. The 1.6 kb region upstream of the start codon was translationally fused to the β-glucuronidase reporter gene and expression of the chimeric gene (Oshox1-GUS) was analysed in transgenic rice by histochemical localisation of the GUS enzyme activity.

In roots, clear Oshox1 expression was detected in the provascular and vascular cylinder (Fig. 1A). Expression did not extend to the most distal region of the apex or to incipient lateral root primordia (Fig. 1B). Oshox1 expression was localised in all differentiating cell types of the central cylinder, pericycle included (Fig. 1C,D), and was still present after completion of vascular differentiation in all the cells of the stele that are not subjected to selective autolysis of their cell contents (sieve elements), or undergo programmed cell death (tracheary elements). Oshox1 expression was also detected in provascular and vascular strands of leaves (Fig. 1E,F), auricles, ligules and culm. Expression was localised in differentiating and differentiated xylem and phloem elements, and in outer and inner bundle sheath cells of all vascular bundles (Fig. 1G-J). In addition, expression was also detected in trichomes of stem and leaves (Fig. 1L), and in guard cells of stomata (Fig. 1K). During the reproductive phase, Oshox1 expression was again observed in all provascular and vascular strands of the whole panicle (Fig. 1M), and additionally detectable in bract hairs (Fig. 1N), pollen and trichomes of the spikelet. Finally, in mature embryos Oshox1 was found to be expressed in provascular bundles of scutellum and embryonic axis (Fig. 1O,P).

To determine the precise location and timing of the onset of Oshox1 expression during (pro)vascular development, a selection of organs from Oshox1-GUS plants was examined in further detail. A series of root transverse sections revealed that Oshox1 was not expressed in provascular cells in the most distal part of the root apex, although a prostele was already identifiable (Fig. 2A), but first became detectable at approximately 0.25 mm from the root tip in provascular cells of the mitotic region. Expression started in the precursor of the central late metaxylem element (Fig. 2B), which is the first cell that will undergo vascular differentiation, as shown in Fig. 2D and as described by others (Kawata et al., 1978). Subsequently, expression spread basipetally and radially to the other prostelar elements, pericycle included, following the order in which they commence vascular differentiation (Fig. 2C-F and Kawata et al., 1978). In all cell types Oshox1 expression started clearly before any cytological sign of differentiation could be recognised (Fig. 2B-E). During lateral root formation, Oshox1 was found to be downregulated in pericycle cells which, upon dedifferentiation, resume their meristematic activity to give rise to a lateral root primordium. Expression reappeared after a provascular strand was recognisable in the lateral root primordium (Fig. 2G). In rice, the vascular bundles interconnecting adventitious root primordia with the vasculature of the main axis differentiate from ground parenchyma cells (Kaufman, 1959). Oshox1 expression was detected in such parenchyma cells that transdifferentiate into vascular elements (Fig. 2H). In the rest of the aerial part of the plant, Oshox1 expression became detectable in provascular strands at an early stage of procambium development, before any sign of vascular differentiation could be observed (Fig. 2I). During spikelet development, Oshox1 expression could be first detected in procambial strands of the palea and lemma as soon as they enclosed the underlying floral organ primordia (Fig. 2O). In agreement with the observation of others (Takeoka et al., 1993), no cytological evidence of an actual progression towards vascular differentiation could be observed at this stage of spikelet development (Fig. 2P). Furthermore, the procambial cells in the palea and lemma of these spikelets showed no cytological differences when compared to provascular cells in the palea and lemma of slightly younger spikelets that do not yet express Oshox1 (Fig. 2M,N). Finally, although Oshox1 was found to be expressed in provascular bundles of the mature rice embryo (Fig. 2J,K), no expression could be detected in 7-day-old embryos despite the fact that conspicuous provascular bundles are evidently present in the scutellum and embryonic axis (Fig. 2L and Jones and Rost, 1989). Only in the scutellum of the mature embryo, could occasional, partial differentiation into immature tracheary elements with secondary wall thickenings be observed (not shown). However, the majority of the Oshox1-expressing vascular strands in the mature embryo were still primarily procambial (Fig. 2J,K), as has been demonstrated also in previous studies where in most of the embryonic provascular tissue no evidence of secondary wall thickening could be detected by light or electron microscopy analysis (Bechtel and Pomeranz, 1978).

Taken together, analysis of provascular tissue in different rice organs revealed that Oshox1 expression invariably starts in procambial cells that do not yet show any cytological progression towards vascular differentiation. This would suggest that such provascular cells are undergoing hidden changes in their developmental state, and that Oshox1 expression might play a role in some of the regulatory aspects that such developmental process would imply.

To confirm the spatial and temporal aspects of Oshox1 gene expression as revealed by the reporter gene analysis, in situ mRNA hybridisation was performed on tissue sections and whole-mount preparations. In the root, Oshox1 transcript was detected specifically in procambium derivatives of the central cylinder, beginning approximately 0.25 mm from the root apex (Fig. 3A-C). In the shoot apical region, Oshox1 expression was found to be associated with provascular and vascular traces of the stem (Fig. 3H). Oshox1 transcript was also detected in parenchyma cells that transdifferentiate into the vascular elements that will connect the adventitious root to the main axis (Fig. 3I). In developing leaves, Oshox1 expression was restricted to provascular and vascular strands at different developmental stages, where the signal was mainly localised in the central differentiating region of the bundle (Fig. 3D-G). In young immature spikelets, Oshox1 transcript was detected in provascular and vascular strands of all organs (Fig. 3J). Finally, in the mature rice embryo Oshox1 expression was limited to its provascular tissue (Fig. 3L). In situ mRNA localisation also revealed expression of Oshox1 in developing trichomes of palea and lemma (Fig. 3K), again confirming what was observed during Oshox1-GUS analysis. It still remains unclear whether Oshox1-GUS expression in stomatal guard cells and pollen reflects bona fide Oshox1 expression. In conclusion, in situ hybridisation largely confirmed that the Oshox1-GUS pattern mimics expression of the endogenous Oshox1 gene.

Mechanical wounding often induces transdifferentiation of parenchyma cells into vascular elements (Sachs, 1981). Although this effect is thought to be largely due to the interruption of auxin transport following severance of vascular bundles, the wounding event in itself might also be involved in induction of vascular elements differentiation (Church and Galston, 1988). The (pro)vascular-specific Oshox1 expression pattern therefore prompted us to analyse the capacity of Oshox1 to respond to mechanical injury of the plant. Upon wounding the tissue between two longitudinal vascular bundles of leaves from Oshox1-GUS plants, Oshox1 was very rapidly, within minutes, induced ectopically in mesophyll and epidermis (Fig. 4A). As in most other monocotyledonous tissues, vascular regeneration does not occur in rice leaves, but cells do respond to injury by lignin deposition. The first indication of this event was detected approximately 6 hours following wounding. A massive induction of Oshox1 was particularly evident in bundle sheath cells facing the wounded area. Interestingly, only these bundle sheath cells, and not those at the opposite side of the vascular bundles, later showed lignin deposition (Fig. 4B). In roots, no wound inducibility of Oshox1 expression could be detected.

The role of auxin as the main regulatory factor in vascular development is well established and has been demonstrated in different experimental systems (Aloni, 1995; Fukuda, 1996). Furthermore, sucrose, in combination with auxin, is required to induce a balanced phloem and xylem differentiation (Aloni, 1987; Warren Wilson et al., 1994). Therefore, we tested the effect of auxin and sucrose on Oshox1-GUS expression. Oshox1 was induced ectopically in cortex and epidermis cells located at the boundary between the mitotic and the transition zone of the root and in the root cap by IAA, 2,4-D and NAA (Fig. 5A), but not by β-NAA (an inactive auxin analogue) or NPA (an auxin transport inhibitor). The same induction pattern was observed in the presence of 500 mM sucrose (Fig. 5B), but not upon a 500 mM mannitol treatment, thus excluding an induction due to osmotic stress.

Brassinosteroids induce entry into the final stage of tracheary element differentiation in cultured Zinnia elegans cells (Iwasaki and Shibaoka, 1991). Accordingly, inhibition of brassinosteroid biosynthesis by uniconazole prevents tracheary element differentiation, without affecting the expression of genes induced during procambium development and early stages of vascular differentiation (Yamamoto et al., 1997). Brassinosteroids and uniconazole, alone or in combination, were tested for their effect on tracheary element differentiation in rice calli, confirming what was observed in the Zinnia elegans xylogenesis system (unpublished results). No effect of brassinosteroids and uniconazole on Oshox1-GUS expression could be detected.

The finding that Oshox1 is affected by auxin and sucrose, but not by brassinosteroids suggests a relation of Oshox1 to procambial cell fate commitment rather than to subsequent vascular differentiation events.

In the root, the column of procambium develops uninterruptedly as a single axial entity behind the growing apical meristem. The resulting correlation between cell position along the main axis of the root and the degree of cell-type differentiation makes it possible to accurately predict the developmental history and fate of each of the cells from their original location (van den Berg et al., 1998). Therefore, we decided to use roots from Oshox1-GUS rice plants to establish the developmental fate of the provascular cells expressing Oshox1.

Oshox1 expression in the root could first be detected approximately 0.25 mm from the root apex in provascular derivatives of the central prostele (Fig. 2B). Root apices of 0.2, 0.5 and 1 mm in length, showing no indication of mature tracheary elements, were excised from Oshox1-GUS rice seedlings and placed on a non-inductive, hormone-free medium. After 9 days of culture in this neutral environment, the root apices either had not further developed (0.2 mm apices) or had grown to a length of about 6 (0.5 mm apices) and 3 (1 mm apices) mm and contained mature tracheary elements (Table 1 and Fig. 6A). This would suggest that the developmental fate of procambial initials and their immediate most distal derivatives (0.2 mm apices) is not yet specified towards vascular differentiation, whereas provascular cells located at the boundary between the division and the transition zone of the root (0.5 mm apices) have become specified towards vascular cell identity. Since Oshox1 is already expressed in procambial cells at 0.5 mm from the root tip, but not in 0.2 mm apices, this suggests that Oshox1 is expressed at a stage of provascular development at which procambial cells are at least specified to become vascular elements. To understand whether provascular cells expressing Oshox1 were also stably determined to become vascular elements, 0.5 and 1 mm root tips were removed from the seedlings, which were thereafter cultured on water for root tip regeneration. When 1 mm root tips were excised, the only response was an enhanced lateral root production around the site of root tip excision detectable in 40% of the cases (Table 1 and Fig. 6B). Regeneration of the root tip could never be observed, suggesting that procambial cells entering the elongation zone of the root (1 mm from the tip) appear stably determined towards vascular differentiation. When 0.5 mm root tips were cut off, root tip regeneration was detected in about 40% of the cases, while in 60% enhanced lateral root formation was again observed (Table 1 and Fig. 6B). Oshox1 appears, therefore, to be expressed at a stage of procambium development at which provascular cells are already specified but not yet stably determined to become vascular elements, since their broader prospective potential enables these cells to promptly switch to a different developmental program, in order to achieve formation of a new root apical meristem. Unexpectedly, in the roots where the tip had regenerated, a massive ectopic induction of Oshox1 was detectable at the site of root tip excision, involving cells of protoderm, ground meristem and root cap, in addition to provascular cells (Fig. 6B). Since this pattern resembled the induction of Oshox1 expression observed in the presence of sucrose or auxin (Fig. 5A,B), the same root tip regeneration experiment was done culturing the seedlings, after root tip excision, in the presence of PCMBS or NPA, inhibitors of sucrose and auxin transport, respectively. Neither root tip regeneration, nor Oshox1 induction could ever be observed under these conditions, suggesting that auxin and sucrose are both required for proper root tip regeneration. Since Oshox1 expression is not wound-inducible in roots, its ectopic expression in this case seems related to the process of root tip regeneration and reprogramming of a specified developmental fate. Such ectopic expression was subsequently rapidly silenced in the regenerating root tip. Here Oshox1 expression became again confined to the prostele, marking the region where we showed that the developmental fate of procambial cells has been specified but not stably determined.

To gain insight into the biological role of Oshox1 during (pro)vascular development, we attempted to overcome the regulatory mechanisms controlling its expression. Since the CaMV 35S promoter drives strong reporter gene expression in all rice organs, including the most distal region of the root and shoot apex (Fig. 7), this promoter is suited to ectopically express Oshox1 in provascular cells that normally do not yet express this gene.

35S-Oshox1 transgenic rice plants were indistinguishable from the wild type with respect to overall size and morphology, and to growth of the aerial part and root system (Fig. 8A,D).

Flowering time was unaffected, and the structure of the flowers displayed no obvious morphological alterations (not shown). The size and shape of leaves of the transgenic plants were comparable to those of wild type, and their vascular pattern did not show any discontinuity (not shown). The cellular anatomy of the leaf and root was examined in transverse sections, revealing that the cells of both transgenic and wild-type plants were arranged similarly and were of comparable size and shape (Fig. 8B,C,E,F). However, a subtle but remarkable phenotype was observed when the anatomy of differentiating vascular tissues was examined in further detail. In the root apical region of the 35S-Oshox1 plants, we could detect abundant secretory vesicles contributing to the growth of secondary cell wall thickenings in the cytoplasm of the precursors of the central late metaxylem elements approximately 0.7 mm from the root tip (0.67±0.06 mm, n=10; Fig. 9B). At the same distance from the root apex of wild-type plants (n=10), these cells still show the dense cytoplasm and small vacuole typical of undifferentiated cells of the procambium or other meristematic tissues (Fig. 9A). The first signs of secretory vesicle formation marking the onset of vascular differentiation could be detected in wild-type roots at approximately 1.2 mm from the apex (1.24±0.08 mm, n=10; Fig. 9C). At this distance from the root tip in 35S-Oshox1 plants (n=10), the precursors of the late metaxylem elements have already started elongation and show advanced stages of cytoplasmic degradation and cell wall modification (Fig. 9D). The situation apppears to be analogous in the shoot, where at 1 mm below the apical meristem of wild-type plants (0.95±0.02 mm, n=10), there is no detectable presence of a peripheral cylinder of vascular bundles (PCVB; Fig. 10A), and where in the vascular bundles that are in the most advanced stage of differentiation the protophloem is hardly visible and the precursor of the second protoxylem element has just begun radial enlargement (Fig. 10C). At the same distance from the shoot apical meristem (SAM) in 35S-Oshox1 plants (n=10), a PCVB is already clearly recognisable (Fig. 10B), and in the vascular bundles the protophloem is easily identifiable, up to four protoxylem vessels have already differentiated and the future late metaxylem elements can be clearly distinguished (Fig. 10D). The same developmental state could be observed in wild-type plants at 2.2 mm below the SAM (2.20±0.03 mm; Fig. 10E,G), a distance at which the vascular bundles in 35S-Oshox1 plants (n=10) have reached complete maturity, with the recognisable presence of a metaphloem, the stretching of protoxylem vessels into a lacuna and the full differentiation of late metaxylem elements (Fig. 10F,H). Taken together, this would suggest that expressing Oshox1 ectopically in procambial cells that do not yet express this gene is sufficient to anticipate their entrance into a vascular differentiation pathway.

Since the onset of Oshox1 expression marks procambial cells whose developmental fate has already been specified but that do not yet manifest any obvious sign of vascular differentiation, it was of particular interest to determine whether ectopic expression of Oshox1 in the most distal region of the root alters the developmental state of provascular cells even in the absence of any evident cytological sign of vascular differentiation. To address this question, root tips of 0.5 mm in length were excised from wild-type and 35S-Oshox1 roots, and the seedlings then cultured for root tip regeneration. As in the previous experiments, wild-type roots regenerated their apices in 40% of the cases, while 60% displayed enhanced lateral root formation around the site of excision. Root tip regeneration was detected in only 10% of the 35S-Oshox1 roots, while 30% showed increased lateral root production and 60% did not react in any appreciable fashion to the treatment (Table 2). Hence, provascular cells situated 0.5 mm from the root tip of 35S-Oshox1 plants showed a response resembling that of determined procambial cells, which in wild-type roots are situated 1 mm from the root apex. Similarly, when 0.2 mm root tips were cultured for root tip growth on a neutral medium, the wild-type tips did not develop any further, exactly as expected, whereas the transgenic ones grew considerably (up to 20 mm after 9 days in culture, with an average length of 12 mm; Table 2), showing a characteristic response of specified provascular cells, which are located at 0.5 mm from the tip in wild-type plants. Taken together, these observations would suggest that ectopic expression of Oshox1 in procambial cells that normally do not yet express this gene results in anticipation of their cell fate commitment, ultimately leading to premature vascular differentiation.

As described above, Oshox1 gene expression was found to be inducible by auxin. The critical role of auxin in (pro)vascular development results particularly from its polar transport through plant tissues (Lomax et al., 1995). The pathways of such polar auxin transport (PAT) are complex and not fully understood, but it is clear that there is a major stream of auxin in the vascular tissues running from the shoot apex to its base where it joins the root. Into the root itself, auxin travels predominantly down through the central stele, then upon reaching the root tip is distributed back upwards along the root in the epidermis and subtending cortical cells, towards the elongation zone (Jones, 1998). To determine whether Oshox1 ectopic expression could result in alterations of PAT, we monitored the transport of ³H-labelled IAA through excised mesocotyls and root apices of wild-type and 35S-Oshox1 plants. Whereas basipetal auxin transport was not significantly different in wild-type and 35S-Oshox1 mesocotyls (Fig. 11A), in 35S-Oshox1 root tips the acropetal transport was reduced approximately 60% when compared with the wild type (Fig. 11B). An important difference between these tissues is that vascular differentiation is still in progress in root apices (Fig. 2), whereas mesocotyls contain two fully mature vascular bundles. Hence, cytological differences resulting from the precocious vascular differentiation in 35S-Oshox1 plants are found in root tips but not in mesocotyls, where PAT appears to be identical in wild type and 35S-Oshox1.

In addition to (pro)vascular development, the co-ordinated movement of auxin within the plant has been implicated in a variety of vectorial developmental processes, including root elongation, lateral and adventitious root formation, and gravitropic response (Aloni, 1995; Lomax, 1995). Because of the difference in PAT capacity between wild-type and 35S-Oshox1 roots, we investigated whether this could result in a difference in those developmental processes in which auxin is known to play a major role. As the data in Table 3 show, we could not detect any significant difference between wild-type and 35S-Oshox1 plants concerning these parameters.

While wild-type and 35S-Oshox1 mesocotyls showed identical PAT capacity, we noticed that 35S-Oshox1 mesocotyls were significantly less sensitive to NPA, accumulating approximately 50% less auxin than the wild type in basal segments of explants subjected to PAT assays in the presence of such PAT inhibitor (Fig. 11C). Moreover, in 35S-Oshox1 root apices, the NPA-induced auxin accumulation was reduced to a higher extent (Fig. 11D) than expected based only on the reduced auxin transport capacity observed in these tissues (Fig. 11B), indicating also in root tips the presence of a decreased NPA sensitivity. Therefore, we decided to examine whether this difference in sensitivity to NPA also resulted in an altered phenotypic response to exogenously supplied auxin or NPA. The data in Table 4 show that whereas root elongation and gravitropism in 35S-Oshox1 were indistinguishable from the wild type, 35S-Oshox1 plants showed increased lateral and adventitious root formation, compared to the wild type, when cultured in the presence of auxin or NPA. In conclusion, lateral and adventitious root development in 35S-Oshox1 plants appear to be more sensitive to auxin and less sensitive to the inhibitory effect of NPA.

To date no information has been available concerning the actual developmental state of procambial cells, and as to when their commitment to differentiation into the various vascular cell types takes place. The operational distinction between a specification and a determination stage during the process of procambial cell fate commitment is an important result of our efforts to put (pro)vascular ontogenesis in a broader developmental perspective. Using the rice root as experimental system, we have shown that the developmental potential of procambial initials and their immediate derivatives has not yet been restricted towards vascular differentiation, and therefore they must be regarded as uncommitted meristematic precursors. Instead, the developmental fate of procambial cells located at the boundary between the mitotic and transition zone of the root has become specified towards a vascular cell identity. Finally, we have shown that procambial cells entering the elongation zone of the root appear to be stably determined towards vascular differentiation.

In addition to defining the developmental stages of provascular cell fate commitment, we have identified a transcriptional regulator that is expressed during the specification events occurring during the restriction of developmental potential of procambial cells. The sharp onset of the Oshox1 gene expression during this particularly dynamic period of provascular development provides evidence for a discrete regulatory event occurring at the beginning of this phase. Indeed, this is when multiple vascular-lineage-specific effector genes, which might represent potential targets of Oshox1 regulatory function, are upregulated or begin to be expressed (Demura and Fukuda, 1994). Interestingly, the regulatory mechanisms in charge of procambial cell fate specification are likely to be strictly conserved amongst plant species, as we could observe the same precise timing of Oshox1 promoter-driven reporter gene expression during provascular development in rice and Arabidopsis (unpublished results).

Ectopic expression of Oshox1 in provascular cells where normally this gene is not yet expressed, is not only sufficient to anticipate vascular differentiation events, but also to alter the timing of procambial cell fate commitment. This observation, together with the precise onset of Oshox1 expression during provascular cell fate specification, suggests a relevance for this gene as a timely molecular switch in the transition to a developmentally restricted phase of procambial cell ontogenesis. The responsiveness of Oshox1 expression to signals known to affect (pro)vascular development, such as wounding, sucrose and auxin, provides further support for the involvement of this gene in procambial cell fate commitment.

Auxin has been shown to exert its vast influence on (pro)vascular development, principally as a consequence of its polarised flow through plant tissues. The directionality of its transport is thought to result from the polar distribution of specialised carrier molecules in the plasma membranes. According to the chemiosmotic hypothesis, auxin is assumed to leave transport competent cells only through the activity of specific carrier proteins that are localised at the basal (basipetal PAT in the shoot and acropetal PAT in the root) or apical (basipetal PAT in the root) extremity of these cells (Jones, 1998). A large body of biochemical and physiological evidence suggests that at least three polypeptides are essential elements of the auxin efflux machinery: a plasma membrane-localised carrier protein; an NPA-binding protein; and a labile, probably cytosolic, component (Bennett et al., 1998). We have found that 35S-Oshox1 root apices, where an anticipation of vascular differentiation could be cytologically detected, showed a reduced PAT when compared with the wild type, while a difference in PAT could not be detected in mesocotyls, where the vasculature of wild-type and 35S-Oshox1 plants is cytologically indistinguishable. This would strongly suggests that the premature vascular differentiation and the reduced PAT observed in 35S-Oshox1 root apices are two intrinsically correlated phenomena, representing different aspects of the same developmental event. Interestingly, the putative plasma membrane-associated component of the auxin efflux carrier has been shown to be localised in provascular and xylem parenchyma cells (Gälweiler et al., 1998; Steinmann et al., 1999). Progression through the differentiation process necessarily implies a reduction in the amount of procambial cells (all of which may transport auxin), in favour of an increase in that of mature vascular cells, of which only a limited number (the parenchymatic cells of the xylem) would be able to efficiently transport auxin.

The reduction in PAT observed in 35S-Oshox1 plants was not phenotypically associated with any difference in auxin-dependent developmental responses, other than (pro)vascular development. However, 35S-Oshox1 plants did respond differently from the wild type to exogenous application of auxin or NPA in lateral root formation, a process strongly dependent on the acropetal component of PAT in the root (Reed et al., 1998; Table 3, which shows that lateral root formation, dependent on auxin flow from the shoot, was completely suppressed ex planta, whereas gravitropism and root elongation were not). In contrast, 35S-Oshox1 and wild-type plants did not show any significant difference in their response to exogenously applied auxin or NPA with regard to root elongation and gravitropic response, which are dependent on the basipetal component of PAT in the root (Rashotte et al., 2000; Table 3). Lateral root formation is believed to be preceded by inhibition of the acropetal PAT immediately downstream of the future site of primordia emergence (Mathesius et al., 2000). This effect would result from the action of an endogenous NPA-like molecule, partially inhibiting the acropetal PAT. This would give rise to a local enhancement of auxin concentration required for pericycle cells to enter division. The different responses of wild-type and 35S-Oshox1 roots to exogenously applied auxin can be explained by the precocious vascular differentiation and associated PAT reduction in 35S-Oshox1 root tips. In 35S-Oshox1 plants, the less efficient drainage of auxin away from the pericycle cells would result in an increased number of lateral and adventitious roots than in the wild type, leading to an apparently enhanced sensitivity to auxin. Exogenous application of NPA would inhibit the efflux carriers in wild-type plants, resulting in a depletion of auxin flowing from the shoot into the root, which would consequently give rise to a reduced lateral root production. The precocious vascular differentiation, and its concurrent PAT reduction, does not explain the observation that NPA-treated 35S-Oshox1 plants produce more lateral roots than the wild type. However, this would be consistent with the presence of NPA-insensitive contra NPA-sensitive efflux carriers in 35S-Oshox1 plants, which would allow more auxin than in the wild type to be transported from the shoot into the root. This assumption is supported by the fact that a reduced, but not completely abolished, NPA sensitivity of 35S-Oshox1 was indeed observed during monitoring of PAT. Taken together, the differences in PAT and auxin/NPA responses between wild-type and 35S-Oshox1 root apices can be consistently explained assuming that precocious and enhanced Oshox1 expression leads to a reduction in the total amount of functional efflux carrier proteins, part of which have now become insensitive to the PAT inhibitor NPA. The putative NPA-insensitive auxin efflux carriers in 35S-Oshox1 would be expressed independently of the differentiation status of vascular tissues, as a decreased NPA sensitivity was measured in roots as well as mesocotyls. The existence of an NPA-insensitive efflux carrier was previously postulated by Mattsson et al. (1999), who tentatively explained the unresponsiveness of anatomically recognisable provascular strands to experimental manipulations of PAT with the possible expression of NPA-insensitive auxin carrier proteins in provascular strands. Furthermore, in sunflower hypocotyls a loss of functional integrity of the auxin efflux system relative to the influx carriers was shown to occur with advancing tissue differentiation, due to a net loss of functional binding sites, as well as a reduction in the apparent affinity of the remaining sites for NPA (Süttle, 1991). Changes in the total amount of functional efflux carriers and ratio of NPA-sensitive versus NPA-insensitive carriers are thus likely to occur during (pro)vascular development, and may represent aspects of procambial cell fate commitment. The altered PAT capacity and NPA sensitivity upon Oshox1 ectopic expression suggest that this gene would act upstream of these events during normal development.

While a role for other regulatory proteins is by no means excluded, our results and those of others suggest that transcription factors of the HD-Zip type are involved at all different levels of (pro)vascular ontogenesis, from procambial cell identity acquisition to vascular differentiation. While some HD-Zip genes appear to play a role in development of specific vascular cell types (Tornero et al., 1996; Zhong and Ye, 1999; Ratcliffe et al., 2000), the HD-Zip gene Athb8 (Baima et al., 1995) is expressed in procambium at an earlier stage than Oshox1, and thus active in uncommitted meristematic vascular precursors. We propose here that Oshox1 represents a paradigm for a class of selector genes, the expression of which leads to a process that narrows the developmental potential of the totipotent meristematic precursors to culminate in provascular cell fate specification. After the checkpoint of provascular cell fate commitment, no significant changes in the expression of Oshox1 could be observed. The fact that many plant cells are able to dedifferentiate to a meristematic state suggests that their identity, once established, must be actively maintained throughout plant life. Oshox1 permanent expression in vascular tissues might therefore suggest a possible vascular tissue identity function for this gene. An additional or alternative possibility is that the permanent vascular expression of Oshox1 points to a function in environmental adaptation. Such a role has been proposed for certain other members of the HD-Zip class (e.g., Frank et al., 1998; Söderman et al., 1999), and possible mechanisms by which these factors could regulate a response to environmental changes have been discussed (Steindler et al., 1999). Consistent with a similar function of Oshox1 is the importance of the vascular system in ensuring morphological adaptability to environmental pressure, and the observation that Oshox1 is also expressed in developing trichomes and stomatal guard cells, which play a role in water availability, temperature control and defence responses. Moreover, outer bundle sheath cells, which also express Oshox1, are more susceptible to death-inducing signals than are mesophyll cells during hypersensitive response in plant-pathogen interactions. Localised death of bundle sheath cells would prevent pathogens gaining entry to the vascular system and spreading systemically (Pennell and Lamb, 1997).

In future work, combined approaches integrating upstream regulators of Oshox1 expression and downstream targets of its function should throw more light on the transcription factor network that establishes and maintains vascular identity, as well as on the role played by Oshox1 in provascular cell fate specification.

We are grateful to Peter Wittich for embedded rice inflorescences. Uniconazole was a generous gift from the ‘China National Chemical Construction Jiangsu Company’. We thank Elly Schrijnemakers for plant care, Anke Taal for tissue culture assistance and Rolf de Kam for pictures of the transgenic plants; Dolf Weijers, René Benjamins and Hanna Weiss for invaluable advice and help; Herman Spaink and Remko Offringa for critical comments on the manuscript. E.S. was supported by a European Commission TMR Marie Curie Research Training Grant (ERBFMBICT972716) and A.H.M. by the Biotechnology Program of the European Commission (BIO4-CT96-0390).

On June 19 2000, while this manuscript was in press, one of the authors, Dr J. Harry C. Hoge, associate professor at the Institute of Molecular Plant Sciences, died unexpectedly. This article is dedicated to his memory.