Pattern formation in the vascular system of monocot and dicot plant species

Authors

  • Enrico Scarpella,

    Corresponding author
    1. Department of Botany, University of Toronto, 25 Willcocks Street, Toronto ON, Canada M5S 3B2,
    2. Department of Biological Sciences, University of Alberta, CW405 Biological Sciences Building, Edmonton AB, Canada T6G 2E9,
    Search for more papers by this author
  • Annemarie H. Meijer

    1. Insitute of Biology, Leiden University, Clusius Laboratory, Wassenaarseweg 64, 2333 AL Leiden, The Netherlands
    Search for more papers by this author

Author for correspondence:
Enrico Scarpella Tel: +1-416-9467441/+1-780-4922869 Fax: +14169785878/+1-780-4929234 Email: enricos@botany.utoronto.ca/enricos@ualberta.ca

Abstract

Contents

  •  Summary

  • I. Introduction
  • II. The plant vascular system
  • III. Ontogeny of the vascular tissues
  • IV. Procambium development
  • V. The organisation of the vascular tissues
  • VI. The regulation of longitudinal vascular pattern formation
  • VII. The regulation of radial vascular pattern formation
  • VIII. Genetic screens for vascular development mutants
  • IX. Genes involved in vascular development identified through reverse genetics approaches 
  • X. Conclusions and perspectives
  •  Note added at the revision stage

  •  Acknowledgements

  •  References

Summary

Plant vascular tissues are organised in continuous strands, the longitudinal and radial patterns of which are intimately linked to the signals that direct plant architecture as a whole. Therefore, understanding the mechanisms underlying vascular tissue patterning is expected to shed light on patterning events beyond those that organise the vascular system, and thus represents a central issue in plant developmental biology. A number of recent advances, reviewed here, are leading to a more precise definition of the signals that control the formation of vascular tissues and their integration into a larger organismal context.

I. Introduction

The colonisation of land by plants occurred more than 400 million years ago and represents one of the most important events in the history of the biological world. To accomplish the transition from aquatic to terrestrial life successfully, several structural and functional changes had to occur in plants. For example, adaptation was necessary for protection against water loss, for absorption of nutrients and water from the soil, for efficient transport of assimilates throughout the plant body, and for mechanical support. A fundamental role in this process of adaptation has been played by the evolutionary development of vascular tissues, which, by solving the problem of long-distance transport of water and nutrients and by providing rigidity to the plant body, allowed the gradual colonisation of land by the first vascular plants. Primitive vascular plants had vascular tissues organised in a very simple fashion, but with the evolution of novel structures, such as the leaf, and the colonisation of various terrestrial habitats by different plant species, vascular tissues diversified to a variety of organisations.

Previous excellent reviews have discussed anatomical, physiological, cell biological, and developmental aspects of vascular patterning and differentiation, mostly in dicot plant species (Northcote, 1995; Fukuda, 1996, 1997, 2000, 2004; McCann, 1997; Nelson & Dengler, 1997; Sjölund, 1997; Berleth et al., 2000a,b; Roberts & McCann, 2000; Sachs, 2000; Aloni, 2001; Dengler & Kang, 2001; Dengler, 2001; Kuriyama & Fukuda, 2001; Turner & Sieburth, 2002; van Bel et al., 2002; Ye, 2002). The present review will first provide a brief description of the structure of the plant vascular system and will deal with its ontogeny and levels of organisation in both monocot and dicot species. Subsequently, it will discuss the regulation of vascular tissue patterning and its relationship to the signals that direct the organisation of the whole plant organism. Finally, it will summarise recent genetic approaches to identify genes involved in different aspects of vascular development in monocots and dicots.

II. The plant vascular system

The vascular system of contemporary seed plants is composed of a coherent and continuous network of strands called vascular bundles (Esau, 1965a). These structures extend through each organ and throughout the entire plant, functionally connecting every part of the shoot with the root system. Vascular bundles typically represent the organisation of two vascular tissues, the phloem and the xylem. The phloem is the route for dissolved carbohydrates, which are transported from tissues that are net producers of photoassimilates to tissues that are net users. Additionally, the phloem provides the path for the translocation of peptides, proteins, and mRNAs involved in plant growth and development and in defence against pathogens (Citovsky & Zambyski, 2000; Oparka & Santa Cruz, 2000; Ruiz-Medrano et al., 2001; van Bel et al., 2002). The xylem is the main conduit for water and mineral nutrients that travel from the root to the sites of evapo-transpiration in the shoot system. Furthermore, the xylem transports hormones such as abscisic acid and cytokinin (Haberer & Kieber, 2002; Hartung et al., 2002). Both phloem and xylem typically comprise a number of specialised vascular cell types, including conducting elements (tracheary elements in the xylem and sieve elements in the phloem), parenchyma, and sclerenchyma cells.

III. Ontogeny of the vascular tissues

In spite of the large structural and functional diversification displayed by the different cell types of the phloem and xylem, all these highly specialised vascular cell types differentiate, during the course of normal development, from a single primary meristematic tissue: the procambium, or provascular tissue (Esau, 1965a). In plants that undergo radial thickening (gymnosperms and dicots), secondary vascular tissues are formed from a lateral meristem called (vascular) cambium, the initials of which originate in part from the procambium within the vascular bundles, and in part from the parenchyma between the vascular bundles. Recently, a number of studies have advanced our understanding of the environmental and hormonal signals and the genetic cues that regulate the formation and the activity of the cambium. Because these subjects do not strictly fall into the realm of vascular pattern formation, the reader is referred to more specific, excellent reviews (e.g. Lachaud et al., 1999; Savidge et al., 2000; Han, 2001; Savidge, 2001; Bhalerao et al., 2003; Helariutta & Bhalerao, 2003). Monocot species usually lack secondary growth from a cambium, but may develop substantial stems (e.g. palms) by a thickening growth resulting from division and enlargement of parenchyma cells of the ground tissue (Esau, 1965b). Secondary growth by means of a special kind of cambium does occur in certain other monocots (e.g. Agave, Aloe, and Yucca). However, in these woody monocots, the cambium arises from the parenchyma outside the primary vascular bundles and produces new parenchyma and secondary vascular bundles. Finally, vascular elements can differentiate from parenchyma cells, in order to regenerate a vascular connection interrupted as a consequence of wounding or grafting in many dicot species, or to generate de novo a connection between the existing vasculature and a developing adventitious organ in both dicots and monocots (Sachs, 1981).

IV. Procambium development

There is a vast body of literature that describes the anatomy of vascular tissue development (reviewed in Nelson & Dengler, 1997). These classical investigations have provided a strictly structural definition of procambium. According to this definition, procambium becomes recognisable during organ development as continuous strands of cytoplasmically dense and narrow cells oriented with their major axes parallel to that of the strand itself (Fig. 1a). Procambial cells and their initials (preprocambial cells) preferentially divide parallel to the axis of the developing vascular strand. Because the axis of the strand itself lies almost invariably parallel to the direction of local organ growth (pre)procambial cells also divide parallel to the direction of growth. This orientation of the division plane represents a characteristic of procambial cells, in that normally the plane of cell division in the surrounding tissues is perpendicular to the direction of growth (Smith, 2001).

Figure 1.

Procambium development. (a) Detail of a longitudinal section through a mature rice embryo showing a mature procambial strand. Procambial cells are elongated and have a dense cytoplasm. Parenchyma cells at the left side of the procambial strand are also elongated, but they are more vacuolated. Like procambial cells, cells of the epidermis are cytoplasmically dense, but they are isodiametric. Parenchyma cells between the epidermis and the procambial strand are isodiametric and have a vacuolated cytoplasm. (b) Detail of a transverse section through the shoot apical region of a mature carrot plant showing a ring of ‘provascular tissue’/′residual meristem′. (c) Detail of a developing Arabidopsis leaf showing the onset of GUS reporter gene expression in the enhancer-trap ET1335 (Sundaresan et al., 1995) marking procambial cells (outlined in red, as opposed to the surrounding cells of the ground meristem, which are outlined in yellow). (d) Detail of a developing Arabidopsis leaf showing the onset of Athb8-GUS reporter gene expression (Baima et al., 1995) marking preprocambial cells (outlined in red, as opposed to the surrounding cells of the ground meristem, which are outlined in yellow). e, epidermis; p, pith; pc, procambium; pv, ‘provascular tissue’/’residual meristem’. Photograph in (b) reproduced with permission from Xia & Steeves (1999). Photographs in (c) and (d) reproduced with permission from Scarpella et al. (2004).

In general, preprocambial cells are polygonal and isodiametric and do not differ from the surrounding cells. However, in the axis immediately below the shoot apical meristem of some dicots and gymnosperms, it has been possible to distinguish a faintly delimited ring of meristematic tissue from which procambial cells will later emerge (Esau, 1965a) (Fig. 1b). There are two different interpretations of this meristematic tissue, which are reflected in the terminology used to describe it. In one interpretation, these ‘prodesmogen’, ‘provascular’, or ‘prestelar’ cells are considered to be the first stage of vascular differentiation occurring independently of the leaves, but in relation to the activity of the shoot apical meristem. In some species, the presence of carboxylesterase activity in the ‘provascular tissue’, but not in the cells of the shoot apical meristem that are continuous with this tissue, and/or the developmental fate of the ‘provascular tissue’ in the absence of leaf primordia seem to support this view (Wardlaw, 1950; McArthur & Steeves, 1972; Xia & Steeves, 1999, 2000). However, similar studies performed in other species seem to support the alternative interpretation, which is that the tissue in question does not represent the first step in the differentiation towards vascular tissues, but rather is a continuation of the shoot apical meristem (Helm, 1932; Ball, 1952; Young, 1954; Gahan & Bellani, 1984; Müller, 1995). Therefore, the authors refer to this tissue as ‘meristem ring’, ‘residual meristem’, or ‘restmeristem’, which they consider anatomically recognisable only because of the precocious vacuolation and expansion of the cells of the surrounding tissues. According to this interpretation, this tissue may or may not become procambium depending on its relation to the developing leaf primordia. More recently, the use of the term ‘provascular tissue’, historically linked to vascular development stages in the stem, has been extended to refer to preprocambial cells in all plant organs. However, different opinions exist as to which cells should be regarded as ‘provascular’ in the same organs at comparable developmental stages (Clay & Nelson, 2002; Holding & Springer, 2002). Furthermore, as mentioned above, certain authors consider provascular tissue as a synonym of procambium (e.g. Esau, 1965a). Therefore, in order to avoid any confusion, throughout this review, only the term ‘procambial’ will be used to describe anatomically identifiable precursors of vascular cells, and the stages prior to procambial differentiation will be regarded as ‘preprocambial’.

The arrangement of vascular strands along the main axis of the plant is largely maintained through the elongation and elaboration of existing strand patterns, rather than being newly generated. By contrast, vascular patterns in the leaf are created de novo during the development of each leaf primordium. Therefore, the study of vascular development in the leaf represents an attractive and convenient system to study the dynamics underlying vascular pattern formation. However, at the same time, understanding the mechanistic basis of vascular tissue patterning is particularly challenging in leaves. In fact, the exact position in the developing leaf primordium at which procambial strands will be formed is largely unpredictable. Because of this variability inherent in leaf vascular pattern formation, its study requires the analysis of extremely large numbers of individual samples, a task that is incompatible with the necessary, but time-consuming and labour-intensive, histological or optical tissue sectioning. Thus, the identification of genes specifically expressed at early stages of vascular development (Baima et al., 1995; Hiwatashi & Fukuda, 2000; Scarpella et al., 2000; Clay & Nelson, 2002; Kang & Dengler, 2002, 2004; Ohashi-Ito et al., 2002; Groover et al., 2003; Ohashi-Ito & Fukuda, 2003; Scarpella et al., 2004), and the isolation of numerous lines displaying early vascular reporter gene expression from different enhancer- and gene-trap lines (Sundaresan et al., 1995; Malamy & Benfey, 1997; Clay & Nelson, 2002; Holding & Springer, 2002; Scarpella et al., 2004) have generated invaluable tools to uncover the anatomically unidentifiable, early stages of procambium formation. An Arabidopsis (Arabidopsis thaliana) enhancer-trap line in which the onset of reporter gene expression reproducibly marks procambial cell identity acquisition has recently been identified and characterised, allowing the study of procambium formation in large populations of samples (Scarpella et al., 2004) (Fig. 1c).

An additional source for the isolation of genes specifically expressed in the procambium is potentially represented by the in vitro system for xylogenesis developed in the dicot species Zinnia elegans (Fukuda, 1996, 1997, 2000; McCann, 1997; Roberts & McCann, 2000; Kuriyama & Fukuda, 2001; Fukuda, 2004). In this cell culture system, mesophyll cells from young leaves are induced to transdifferentiate into xylem tracheary elements in the presence of a suitable hormone combination, and a large number of genes selectively associated with different steps of tracheary element formation have been isolated and characterised by using this system. Expression studies have demonstrated the colinearity between the different steps of the transdifferentiation of mesophyll cells into tracheary elements that takes place in this in vitro system and the process of xylem differentiation from procambial cells that occurs in planta, simultaneously leading to the discovery of early markers of procambium development in Zinnia (Demura & Fukuda, 1994; Ohashi-Ito et al., 2002; Ohashi-Ito & Fukuda, 2003). Unfortunately, promoter-reporter gene constructs for most of these genes are not available yet, and thus these potentially useful markers cannot be directly tested in other plant species. However, available evidence suggests that at least some of the genes specifically expressed in the early steps of mesophyll cell transdifferentiation could be employed in numerous species as markers of different stages of procambium development (Igarashi et al., 1998; Demura et al., 2002; Milioni et al., 2002; Ohashi-Ito et al., 2002; Ohashi-Ito & Fukuda, 2003).

A further difficulty intrinsic to the study of the dynamics underlying vascular pattern formation in leaves is that, in these organs, preprocambial cells are recruited from the apparently homogenous subepidermal ground meristem tissue, which will later give rise to both procambium and mesophyll. Therefore, in the leaf (unlike in the stem) preprocambial cells do not differ in any respect from the surrounding cells with regard to cell shape and extent of vacuolation. Because of this lack of anatomical distinction, preprocambial cells can only be distinguished from the surrounding ground meristem on the basis of differential gene expression (Scarpella et al., 2004). The onset of expression of the Arabidopsis thaliana homeobox gene8 (Athb8; Baima et al., 1995), which encodes a member of the homeodomain-leucine zipper (HD-Zip) III family of transcription factors (Sessa et al., 1994), marks the earliest stages of preprocambial development (Kang & Dengler, 2004; Scarpella et al., 2004) (Fig. 1d). The dynamics of Athb8 expression suggest that in Arabidopsis leaves, preprocambial strands develop in continuity with the pre-existing vasculature and extend progressively away form their point of origin. Once the Athb8 preprocambial domain has reached its maximal extension, procambial cells differentiate nearly simultaneously along the length of the entire preprocambial strand (Scarpella et al., 2004).

V. The organisation of the vascular tissues

Two levels of organisation can be distinguished within the vascular system (Esau, 1965b; Dengler & Kang, 2001): the longitudinal pattern, which is the array of vascular bundles within a certain organ, and the radial (or transverse, or transectional) pattern, which is the spatial arrangement of the phloem and xylem, and of their different cell types, within each vascular bundle. Both levels of organisation are organ-specific. In fact, developing vascular tissues must maintain their functionality while being integrated into the context of expanding organs with radically different morphology and anatomy, such as leaves, stems, roots, and flowers. This eventually gives rise to the typical organ-specific position and internal organisation of vascular bundles, and the reproducible differentiation of nonvascular cell types in fixed spatial relationships to vascular bundles (Fig. 2). The existence of organ-specific vascular patterns suggests that genetic instructions likely constrain the variety of possible organisations in order to integrate vascular and nonvascular tissue patterns functionally into the context of a determined organ. Furthermore, species-specific cues are also likely to be involved in the coherent organisation of the vascular tissues into a specific organ context, as shown, for example, by the successful use of species-specific leaf vascular patterns as taxonomic diagnostic features (e.g. Klucking, 1995). Dicot and monocot leaves, in fact, display highly divergent vascular patterns (Nelson & Dengler, 1997). Most dicot leaves show a reticulate pattern, in which veins of different size orders form a highly branched network, whereas most monocot leaves show a typical striate venation pattern, in which major longitudinal veins lie parallel along the proximo-distal axis of the leaf and are connected transversally by minor commissural veins (Fig. 3).

Figure 2.

Organ-specific vascular tissue organisation in the monocot rice and the dicot Arabidopsis. (a) Transverse section through a mature rice root. Vascular tissues are organised in a central vascular cylinder surrounded by the endodermis and embedded in the cortex. (b) Transverse section through a mature rice stem internode. Vascular tissues are organised in small and large bundles, which are arranged in two concentric circles. The small vascular bundles of the outer circle are located between two rings of sclerenchyma and embedded in a photosynthetic cortex. The large vascular bundles of the inner circle are embedded in a parenchymatous pith. (c) Detail of a transverse section through a mature rice leaf. Vascular tissues are organised in large and small longitudinal bundles and transverse commissural bundles, which are not visible in this section (see also Fig. 3). Longitudinal bundles are surrounded by a parenchymatous bundle sheath and embedded in the mesophyll. (d) Detail of the vascular cylinder in (a). The xylem forms a solid central core with ridgelike projections extending towards the periphery of the vascular cylinder. The phloem is organised in separate strands alternating with the xylem ridges. The number of xylem ridges varies in different species. Typically, monocot roots have more than four, and rice roots have six (Scarpella et al., 2003). (e) Detail of the internode wall in (b). Vascular tissues are organised in collateral bundles, where the phloem pole at the outer side of the bundle is juxtaposed with the xylem pole at the inner side of the bundle. (f) Detail of a large vascular bundle in (c). Vascular tissues are organised in collateral bundles, where the phloem pole at the abaxial (lower) side of the bundle is juxtaposed with the xylem pole at the adaxial (upper) side of the bundle. (g) Transverse section through a mature Arabidopsis root. Vascular tissues are organised in a central vascular cylinder surrounded by the endodermis and embedded in the cortex. (h) Transverse section through a mature Arabidopsis stem internode. Vascular tissues are organised in bundles that are arranged in a circle and connected by interfascicular fibres. The vascular bundles and the interfascicular fibres form a hollow cylinder that surrounds a central parenchymatous pith and is embedded in a photosynthetic cortex. (i) Detail of a transverse section through a mature Arabidopsis leaf. Vascular tissues are organised in bundles of different sizes that are surrounded by a parenchymatous bundle sheath and embedded in the mesophyll (see also Fig. 3). (j) Detail of the vascular cylinder in (g). The xylem forms a solid central core with ridgelike projections extending towards the periphery of the vascular cylinder. The phloem is organised in separate strands alternating with the xylem ridges. The number of xylem ridges varies in different species. Typically, dicot roots have up to four, and Arabidopsis roots have two (Dolan et al., 1993). (k) Detail of a transverse section through a mature Arabidopsis stem internode. Vascular tissues are organised in collateral bundles, where the phloem pole at the outer side of the bundle is juxtaposed with the xylem pole at the inner side of the bundle. (l) Detail of a vascular bundle in (i). Vascular tissues are organised in collateral bundles, where the phloem pole at the abaxial (lower) side of the bundle is juxtaposed with the xylem pole at the adaxial (upper) side of the bundle. 1, Primary vein (midvein); 2, secondary vein; 3, tertiary vein; b, bundle sheath; c, cortex; en, endodermis; if, interfascicular fibres; lv, large vascular bundle; m, mesophyll; p, phloem; pi, pith; s, sclerenchyma; sv, small vascular bundle; v, vascular cylinder; vb, vascular bundle; x, xylem. Slides in g,h,j,k courtesy of N. Dengler. Slides in i,l courtesy of J. Kang.

Figure 3.

Species-specific vascular tissue organisation in the leaf. (a) Whole-mount clearing of a mature Arabidopsis leaf viewed with dark-field illumination and showing a reticulate vascular pattern. The midvein, or primary vein, of the Arabidopsis leaf is a single vein that contains more vascular cells than the lower orders of veins because of a cell division activity that persists even after all other veins have completely differentiated (Donnelly et al., 1999; Kang & Dengler, 2002). (b) Detail of a whole-mount clearing of a mature rice leaf viewed with dark-field illumination and showing a striate vascular pattern. The midvein of the rice leaf is a complex structure consisting of one large and seven small veins (Scarpella et al., 2003). The midvein is usually separated from the first large vein by two or three small veins. Large veins are separated by an average of five small veins, except at the margins of the leaf, where one or two small veins separate the two most external large veins. 1, Primary vein (midvein); 2, secondary vein; 3, tertiary vein; 4, quaternary vein; cv, commissural vein; lv, large vein; mv, midvein; sv, small vein.

Although the organ- and species-specific vascular tissue organisations are highly reproducible, the final result is not a static and completely predictable picture. In fact, vascular patterns are simultaneously both consistent in their integration into the local tissue context and, with the possible exception of the midvein in the leaf, unpredictable in the precise course and arrangement of the vascular strands. Furthermore, the vascular system, at least in dicot species, retains a high level of plasticity and flexibility, in that fully expanded leaves and emerging adventitious organs can develop additional connections with the existing vasculature, and older parts of the root, stem, and leaf can still form new vascular bridges to circumvent a wound (Sachs, 1981 and references therein; Sachs, 1989). The apparent contradiction lying in the simultaneous existence of a high level of reproducibility and variability in the organisation of the vascular tissues suggests the involvement of directional signals that, in combination with the self-organising capacities of vascular tissues, generate continuous and perfectly aligned strands within variable and unpredictable vascular networks. The vascular system therefore represents a reproducible, yet flexible, entity where the overall pattern and arrangement of cell types is functional, but not completely identical from individual to individual.

VI. The regulation of longitudinal vascular pattern formation

1. Polar auxin transport and vascular development

At present, the molecular mechanisms underlying the different aspects of vascular tissue pattern formation are not known. However, various plant hormones have been reported to promote vascular differentiation in different species and experimental conditions (e.g. Aloni, 1987; Fukuda, 1997). Nevertheless, the role of the phytohormone auxin is unique. In fact, auxin not only triggers vascular differentiation per se, but also induces the transdifferentiation of a slender strip of parenchyma cells into a continuous vascular strand that will extend towards the basal pole of the plant in wounded stems and roots of various dicot species (Sachs, 1981) (Fig. 4a). However, a source of auxin alone does not seem to induce procambium formation in stems where all leaf primordia are excised (Young, 1954; McArthur & Steeves, 1972), although it can promote differentiation of procambium into xylem (Jost, 1939; Wangermann, 1967). Furthermore, vascular differentiation in response to auxin application does not readily occur in all species. For example, monocots are known for their recalcitrance (Aloni & Plotkin, 1985; our unpublished observations). These findings suggest that further factors acting in concert with auxin are probably required. However, the capacity of a simple signal to trigger a complex and oriented cellular response suggests that the signalling mechanism recruits polar cues already present in the organism. These directional signals integrating cell polarity and aligned differentiation could be provided by the polar auxin transport (PAT) that normally occurs in plants (Goldsmith, 1977; Lomax et al., 1995; Muday & DeLong, 2001; Friml & Palme, 2002; Muday & Murphy, 2002).

Figure 4.

Polar auxin transport and vascular development. (a) Local application of auxin (black blob) induces the transdifferentiation of parenchymatic cells to form vascular strands (black wavy lines) in hypocotyl segments (Sachs, 1981). The vascular differentiation response is characterised by specific properties. First, the response is local, as the newly formed vascular strands extend from the precise site of auxin application. Second, the response is polar, as auxin induces vascular strands only towards the morphologically basal pole of the organ. Third, the response is continuous, as the responding cells differentiate in continuity with one another, eventually connecting the auxin source with pre-existing vasculature (grey). Fourth, the differentiation zone is restricted in the radial dimension, as differentiation occurs only within a narrow strip of cells. Fifth, the patterned vascular differentiation response requires the use of polarly transported auxins, and is obstructed in the presence of auxin transport inhibitors. (b) Left: routes of polar auxin transport in the plant. In the shoot, auxin is transported mainly through the vascular tissues in a single direction, from the shoot apex to the base. The directionality of polar auxin transport within leaf vascular networks is not clear, and therefore has not been represented. In the root, movement of auxin from the shoot is from the base of the root towards the root apex through cells of the central vascular cylinder. Lower right: additionally, in the root, auxin is transported from the root tip towards the base through cells of the cortex and/or epidermis. Upper right: hypothetical cellular mechanism for polar auxin transport (redrawn from Muday & DeLong, 2001). Basal localisation of the putative auxin efflux carrier PIN1 protein (black oval) depends on vesicle transport, and polar auxin transport requires an intact actin cytoskeleton. Cortical actin filaments (black lines) might thus serve as tracks for vesicle delivery cycling of the PIN1 protein between the plasma membrane and undefined endosomal membrane structures (dotted) (Geldner et al., 2001, 2003). A high-affinity NPA-binding protein (white circle), known to interact with actin, may connect the transport vesicles (grey) with the actin tracks.

Auxin is synthesised predominantly in young apical regions, such as leaf primordia and floral buds (Sheldrake, 1973; Goodwin, 1978; Ljung et al., 2001). In the stem, auxin moves unidirectionally through the vascular tissues from the apex to the base (Goldsmith, 1977; Lomax et al., 1995) (Fig. 4b). In the root, two distinct polarities of PAT exist. Auxin moves acropetally through the central vascular cylinder and, after being redistributed at the root tip, proceeds basipetally towards the elongation zone through the cells of the epidermis and/or outer cortical layers of the root (Goldsmith, 1977; Lomax et al., 1995) (Fig. 4b). At the cellular level, auxin is thought to be translocated through the action of specific membrane-localised influx and efflux carriers. Auxin enters plant cells both by diffusion and through the facilitating action of an auxin influx carrier (Goldsmith, 1977; Lomax et al., 1995) that is thought to be encoded by AUX1 (Marchant et al., 1999) and possibly by related genes (Parry et al., 2001). Auxin cannot diffuse out of plant cells, and can thus exit only through an efflux carrier apparatus that requires the activity of at least two polypeptides (Morris, 2000; Muday & DeLong, 2001; Muday & Murphy, 2002) (Fig. 4b). The first of them is an integral membrane transporter thought to be encoded by members of the PIN FORMED (PIN) gene family (Palme & Gälweiler, 1999). The second component of the auxin efflux carrier apparatus performs a regulatory function, and represents a high-affinity binding site for PAT inhibitors (PATIs), such as 1-N-naphthylphthalamic acid (NPA) (Rubery, 1990). Several studies indicate that this NPA-binding protein (NBP) is a peripheral membrane protein associated with the cytosolic face of the plasma membrane, and that this protein interacts with the actin cytoskeleton (Morris, 2000; Muday, 2000). Interestingly, both the AUX1 and PIN proteins show an asymmetric localisation in the plasma membrane that is consistent with a role in controlling the polarity of auxin movement (Gälweiler et al., 1998; Müller et al., 1998; Swarup et al., 2001; Friml et al., 2002, 2003; Benkováet al., 2003).

Although the molecular details remain hypothetical, PAT itself is experimentally well-documented, and its characteristics could account for the geometrical properties of vascular strand formation. A number of different hypotheses have been postulated to explain the link between auxin physiology and the different aspects of vascular development (Mitchison, 1980; Sachs, 1981; Meinhardt, 1982; Nelson & Dengler, 1997; Aloni, 2001). However, ultimate experimental evidence has been difficult to obtain. With the currently available genetic and experimental tools, the only possibility to test the hypothesis that PAT and auxin signalling have a role in vascular development is by manipulating auxin flow and response. This has been performed in a variety of ways, and the results are consistent with a specific role of PAT and auxin signalling in vascular development, as detailed in the following paragraphs.

The application of several classes of PATIs to Arabidopsis seedlings has been used in two studies as a means to investigate the development of the vascular system under conditions of reduced PAT (Mattsson et al., 1999; Sieburth, 1999). Both studies report similar vascular tissue responses in several organs. Vascular cells were less aligned with each other, and generally more vascular tissues were formed. These findings seem to support a role for PAT in the concerted regulation of oriented cell differentiation and the restriction of vascular differentiation to narrow zones. Chemical interference with PAT also significantly affected leaf venation pattern (Mattsson et al., 1999; Sieburth, 1999). Increasing inhibition of PAT progressively restricted vascular differentiation towards the margins of the leaf, suggesting that this region harbours major auxin sources critical for vein formation. This pattern shift was already observed at low concentrations of PATIs that did not significantly alter leaf morphology. Furthermore, different orders of veins in the leaf became unresponsive to PAT inhibition at the time of their emergence as procambial strands, suggesting that these cells express auxin efflux carriers that are insensitive to the applied inhibitors. Consistent with this hypothesis, increased levels of expression of the Oryza sativa homeobox1 (Oshox1) HD-Zip II gene, a positive regulator of procambial development in rice (Oryza sativa), reduced the sensitivity of the PAT machinery towards inhibition (Scarpella et al., 2000, 2002). An approach similar to that undertaken in Arabidopsis has also been employed to study the possible, and historically more controversial, relationship between PAT and vascular development in two monocot species (Tsiantis et al., 1999a; Scarpella et al., 2002). In both maize (Zea mays) and rice, PATI application resulted in thickening of the vascular bundles of the leaf, suggesting a role for PAT in constraining the regions of vascular differentiation. Furthermore, rice leaves formed under conditions of PAT inhibition displayed reduced distance between longitudinal veins and increased distance between transverse veins, providing experimental evidence for a role of PAT in vascular patterning in a monocot species (Scarpella et al., 2002).

In addition to the results obtained through the experimental manipulation of PAT with chemical inhibitors, the molecular features of the PIN1, or AtPIN1, gene in Arabidopsis support a role for PAT in vascular development. The PIN1 gene encodes a putative auxin efflux carrier protein that is localised to the basal membrane of procambial and xylem parenchyma cells (Gälweiler et al., 1998; Steinmann et al., 1999) (Table 1). Mutations at the PIN1 locus result in reduced PAT in the stem, pin-shaped inflorescence morphology (a feature also observed upon chemical inhibition of PAT), and aberrant vascular patterning (Okada et al., 1991; Gälweiler et al., 1998; Mattsson et al., 1999). Excess vascular tissue formation is observed at sites of leaf insertions in mutant stems and at the margins of mutant leaves. All these abnormalities are similar to those found in Arabidopsis plants treated with PATIs (Mattsson et al., 1999; Sieburth, 1999). However, vascular defects in pin1 mutants resemble those evoked by treating wild-type plants with low concentrations of PATIs. This suggests that multiple redundantly acting genes are involved in PAT. Inhibition of PAT with high concentrations of PATIs probably mimics the appearance of a mutant simultaneously defective in all these genes. Such a mutant would be expected to have additional severe embryonic defects, since application of PATIs at early embryo stages results in the loss of embryonic axis formation and the consequent generation of ball-shaped embryos (Hadfi et al., 1998; Friml et al., 2003). However, these defects were not observed in the studies mentioned above, since PATIs were applied at germination. The phenotype of the gnom/emb30 (gn/emb30) mutant of Arabidopsis matches that predicted for a mutant strongly impaired in PAT. Mutant embryos and seedlings have no detectable apical-basal polarity, and the vascular system is limited to a centrally located series of vascular cells that are entirely disconnected and randomly oriented (Mayer et al., 1993) (Table 1). The GN/EMB30 gene encodes a guanine exchange factor required for vesicle transport-mediated polar localisation of PIN1 and possibly other auxin efflux membrane proteins (Steinmann et al., 1999). Interestingly, seedlings simultaneously carrying mutations in multiple PIN genes display phenotypes reminiscent of gn/emb30 mutants, supporting a redundant role of these genes in PAT (Friml et al., 2003). Furthermore, weaker gn/emb30 alleles exhibit leaf vascular pattern defects similar to those induced by postembryonic PAT inhibition (Koizumi et al., 2000; Geldner et al., 2004). Therefore, all available data are consistent with the view that both organismal apical-basal polarity and positioning of procambial strands depend on PAT.

Table 1.  Genes involved in longitudinal vascular pattern formation and organismal apical-basal polarity establishment
GeneOrganismMutant phenotypeOverexpressor phenotype*Molecular identityReferences
Vascular anatomyAuxin physiologyVascular anatomyAuxin physiology
  • *

    Unless otherwise indicated, overexpression is achieved through expression driven by the Cauliflower Mosaic Virus 35S promoter.

(At)PIN1ArabidopsisIncreased and discontinuous vascularisation in the stem of the homozygous mutant. Increased and marginalised vascularisation in leavesReduced auxin transportUnknownUnknownPutative auxin efflux carrierOkada et al. (1991);Gälweiler et al. (1998); Mattsson et al. (1999)
AXR6/AtCUL1ArabidopsisReduced, discontinuous, and centralised vascularisation in cotyledons of the homozygous mutantReduced auxin sensitivity (dominant trait)UnknownUnknownMember of the cullin/CDC53 family of proteinsHobbie et al. (2000); Hellmann et al. (2003)
BDL/IAA12ArabidopsisReduced and discontinuous vascularisation in cotyledons and leaves of the homozygous mutantReduced auxin sensitivity (dominant trait)UnknownUnknownMember of the Aux/IAA family of short-lived nuclear proteinsHamann et al. (1999), (2002)
EMB30/GN/VAN7ArabidopsisDisconnected, randomly oriented vascular cells in homozygous mutant seedlingsUnknownUnknownUnknownGuanine exchange factor Mayer et al. (1993); Shevell et al. (1994); Busch et al. (1996); Steinmann et al. (1999); Koizumi et al. (2000)
MP/ARF5ArabidopsisReduced, misaligned, and discontinuous vascularisation in the stem of the homozygous mutant. Reduced, misaligned, discontinuous, and centralised vascularisation in cotyledons and leaves. Discontinuous vascularisation in adventitious rootsReduced auxin sensitivity and transportNormal at the seedling stageUnknownTranscription factor of the ARF familyBerleth & Jürgens (1993); Przemeck et al. (1996); Hardtke & Berleth (1998); Mattsson et al. (2003)
RAL1RiceReduced, discontinuous and misaligned vascularisation in scutella and leaves of the homozygous mutant. Altered timing of vascular development in leaves. Reduced vascularisation in stem and adventitious rootsReduced auxin sensitivityUnknownUnknownUnknownScarpella et al. (2003)

The observation that changes in the expression of genes encoding proposed regulators of vascular development display alterations in PAT capacity (Carland & McHale, 1996; Przemeck et al., 1996; Tsiantis et al., 1999a; Scarpella et al., 2000; Scanlon et al., 2002) has been tentatively interpreted as independent evidence confirming the role of PAT in vascular tissue development. However, in these studies, the vascular tissue organisation was altered to different extents at the developmental stages at which the organs were analysed for their PAT properties. This made it difficult to eventually resolve the causal relationship between the vascular phenotypes and the associated PAT alterations. However, attempts to assess PAT capacity at developmental stages at which the vascular defects could not yet be anatomically identified, or in organs that showed wild-type anatomy, seem to support the view that alterations in PAT capacity represent the direct cause of aberrant vascular phenotypes (Zhong & Ye, 2001; Scarpella et al., 2002).

In summary, vascular strand formation is strictly dependent on apical-basal organismal polarity and PAT. This correlation suggests that apical-basal axis establishment and vascular strand formation have a common origin in the orientation and distribution of PAT. However, while obviously necessary, PAT is probably not sufficient to ensure the precision of aligned vascular differentiation, which is likely to be further promoted by other, unidentified mechanisms.

2. Auxin response and vascular development

Although PAT seems to account for the geometric properties of the vascular differentiation response, proper auxin response should nevertheless be essential for the relay of auxin signals in this process. In leaves, vascular differentiation indeed occurs at sites of maximum auxin response (Mattsson et al., 2003; Scarpella et al., 2003) (Fig. 5a,b), and proper positioning of these auxin response maxima requires PAT (Mattsson et al., 2003). Furthermore, vascular defects have been reported for auxin response mutants. For example, the auxin-resistant1 (axr1) mutant of Arabidopsis displays smaller vascular bundles in the stem (Lincoln et al., 1990). However, severe auxin insensitivity might have more dramatic effects resulting in embryo or seedling lethality, and therefore many mutants in this class might still be unidentified. This is suggested by the fact that mutations in the MONOPTEROS (MP) and AUXIN-RESISTANT6 (AXR6) genes of Arabidopsis lead to seedling lethality associated with a complex phenotype characterised by impaired auxin response, severely reduced vascular system, defective embryo axis formation, and consequent failure to produce an embryonic root (Berleth & Jürgens, 1993; Przemeck et al., 1996; Hobbie et al., 2000; Mattsson et al., 2003).

Figure 5.

Auxin response and vascular development. (a,b) Spatial patterns of auxin response in leaves of Arabidopsis (a) (Mattsson et al., 2003) and rice (b) (Scarpella et al., 2003), visualised through β-glucuronidase reporter gene expression driven by the DR5 synthetic auxin-inducible promoter (Ulmasov et al., 1997b). (a) Whole-mount preparation of a developing Arabidopsis first leaf. (b) Transverse section through the shoot apex of a rice seedling. (c) Hypothetical model of auxin regulation of gene expression in vascular development (for excellent, more comprehensive reviews on auxin regulation of gene expression, the reader is referred to Dharmasiri & Estelle, 2002; Hellmann & Estelle, 2002; Kepinski & Leyser, 2002; Leyser, 2002). In the presence of low cellular levels of auxin, MP/ARF5 (green) activity is repressed through the interaction with Aux/IAA proteins, such as BDL/IAA12 (red). In the presence of high cellular concentrations of auxin, the levels of Aux/IAA proteins are depleted by proteolysis through the 26S proteosome. Auxin-mediated Aux/IAA protein degradation is regulated through the E3 ubiquitin-ligase function of SCFTIR1 (light blue), which is composed of four subunits (RBX1, AXR6/CUL1, ASK1/2, and TIR1). Auxin promotes the interaction between the F-box substrate recognition subunit TIR1 and the targeted Aux/IAA protein (Dharmasiri et al., 2003; Tian et al., 2003). Upon auxin-mediated Aux/IAA proteolysis, MP/ARF5 can activate transcription of targets, which include genes expressed at early stages of vascular development, such as Athb8, and early auxin-inducible genes, such as Aux/IAAs (Mattsson et al., 2003). Therefore, auxin acts antagonistically by simultaneously promoting Aux/IAA protein degradation and Aux/IAA gene transcription, thus making Aux/IAA and ARF proteins central components of a potential regulatory feedback loop. This mechanism allows the rapid restoration of Aux/IAA protein abundance when auxin levels diminish, therefore providing a means to quickly and efficiently respond to auxin-mediated endogenous inputs or external signals. Photographs in (a) and (b) reproduced with permission from Mattsson et al. (2003) and Scarpella et al. (2003), respectively. 4, 5, 6, Fourth, fifth, and sixth leaf primordium, respectively; E1 (orange), ubiquitin-activating enzyme, E2 (yellow), ubiquitin-conjugating enzyme, white, ubiquitin.

Seedling lethality in mp mutants can be bypassed by generating adventitious roots in tissue culture, enabling studies of postembryonic stages (Berleth & Jürgens, 1993). Throughout mutant development, vascular tissues are incompletely differentiated (Przemeck et al., 1996) (Table 1). In leaves, the vascular system is reduced to its most central part: the midvein and a few secondary veins. Mutant stems show reduced PAT, but, because of the vascular defects, it is not clear whether these alterations in auxin flow represent the cause or the consequence of the vascular defects.

Homozygous axr6 mutant seedlings display defects very similar to those of mp mutants, but putative postembryonic functions of the AXR6 gene have not been studied because of the seedling lethal phenotype of the mutant (Hobbie et al., 2000) (Table 1). However, one important difference from mp, whose defects are recessive, is that the auxin sensitivity defects caused by the axr6 mutation are dominant. Heterozygous axr6 mutants are bushy, form fewer lateral roots than normal, and show auxin-resistant root elongation and gravitropism. Since vascular patterns have not been investigated in heterozygous axr6 mutants, it remains to be established whether the function of the mutated AXR6 gene also acts in a dominant fashion in vascular development. However, it is important to realise that inferring the wild-type function of a gene from a dominant phenotype can be difficult because the change caused by the mutation may give rise to novel functions not normally present in wild-type plants.

Mutation of the BODENLOS (BDL) gene of Arabidopsis results in a phenotype similar to those of the mp and axr6 mutants, but the defects in bdl are weaker and do not result in seedling lethality (Hamann et al., 1999) (Table 1). Development of the vascular system is reduced, and a hypocotyl of variable length ends in a basal peg rather than in a primary root meristem. Homozygous bdl mutant seedlings display reduced auxin responses, such as reduced hypocotyl swelling and callus formation. Similar to the axr6 mutation, the auxin sensitivity defects in the bdl mutant are dominant (Hamann et al., 2002).

A recessive mutation in the RADICLELESS1 (RAL1) gene of rice is characterised by defects very similar to those displayed by the above Arabidopsis mutants (Scarpella et al., 2003) (Table 1). Mutant ral1 embryos fail to form a primary root, and the embryonic procambial system is reduced to an interrupted and prematurely aborted primary strand with misaligned procambial cells. Like bdl, ral1 seedlings spontaneously produce adventitious roots, allowing the study of the postembryonic function of the RAL1 gene. In ral1 leaves, longitudinal and commissural (transverse) veins display altered spacing, and the commissural veins show interruptions in their continuity and display abnormal branching. All these vascular patterning aberrations originate from defects in the procambium, the formation of which is delayed in ral1 leaf primordia. Finally, the ral1 mutant displays a defective response to auxin and an enhanced sensitivity to cytokinin.

The similarities between the phenotypes of the mp, axr6, and bdl mutants of Arabidopsis and the ral1 mutant of rice suggest that they have related primary defects in the molecular machinery underlying the alignment of cell differentiation with the axis of auxin flow. Strong support for this interpretation comes from the fact that MP is a member of the auxin response factor (ARF) family of transcriptional regulators (Hardtke & Berleth, 1998). The MP DNA-binding domain can bind to auxin response elements, which are short conserved sequences essential for the rapid auxin regulation of certain classes of auxin inducible genes (Ulmasov et al., 1997a). ARF proteins have two conserved C-terminal domains in common with the related class of short-lived, nuclear-localised Aux/IAA proteins (Abel et al., 1994). Unlike ARF genes, Aux/IAA genes are rapidly induced by auxin, and their abundance seems to reflect the strength of the auxin response. The conserved C-terminal domains have been shown to mediate homo- and heterodimerisation within each class and between members of the ARF and Aux/IAA classes (Kim et al., 1997).

It has been suggested that the specificity of the auxin response could be conferred by nuclear complex combinations of ARF and Aux/IAA proteins, some of which could specifically promote the differentiation of vascular cells or other cells aligned with the axis of PAT (Berleth et al., 2000a). Consistent with this interpretation is the recently determined molecular identity of the BDL and AXR6 genes. The BDL gene encodes a member of the Aux/IAA family, IAA12, which is capable of interacting with the MP protein in yeast (Hamann et al., 2002). The bdl mutation is similar to mutations reported for other Aux/IAA genes which have been shown to result in the stabilisation of the mutant protein. Interpreting such a phenotype is somewhat problematic because the stabilised IAA12 protein in bdl mutant plants might be present at sufficiently high concentrations to interact with ARF proteins that it would not normally contact. If this is true, then the phenotype of the bdl mutant may reflect the expression patterns of the mutated gene, rather than its ‘real’ function. Recent genetic evidence, however, suggests that, although the BDL protein does not display any preferential interaction with MP/ARF5 in transient expression systems, a highly specific antagonistic interaction of both proteins in a variety of developmental contexts occurs in planta (Hardtke et al., 2004).

The AXR6 gene encodes AtCUL1 (Shen et al., 2002; Hellmann et al., 2003), a member of the cullin/CDC53 family of proteins, and is therefore one of the different subunits of the SCF (for SKP1, Cullin/CDC53, F-box protein) ubiquitin ligase (Gray et al., 1999). The SCFTIR1 ubiquitin ligase complex, which contains the TRANSPORT INHIBITOR RESISTANT1 (TIR1) F-box protein and the Arabidopsis SKP1-like protein, has been implicated in auxin-dependent degradation of Aux/IAA proteins (Gray et al., 1999, 2001).

Taken together, these data suggest a simplistic, but experimentally testable model for the auxin-mediated regulation of vascular pattern formation (Fig. 5c). In the absence of an auxin signal, Aux/IAA proteins such as BDL/IAA12 would act as inhibitors of MP transcriptional activator function. In the presence of auxin, Aux/IAA proteins would be targeted for SCF-mediated degradation, releasing MP from the inhibitory interaction. MP would then be free to activate the transcription of genes involved in vascular development, as well as that of Aux/IAA genes. The activation of Aux/IAA transcription by MP, with consequent inhibition of MP function, would be necessary to limit the duration of the auxin response, thus ensuring its prompt modulation upon changes in auxin signals. This model has found further experimental support in the recent discovery that enhanced levels of MP, in the absence of recognisable phenotypes, result in increased transcript abundance of genes expressed early in vascular development, such as the HD-Zip genes Athb8 and Athb20, as well as in the enhanced auxin inducibility of these and some Aux/IAA genes (Mattsson et al., 2003).

VII. The regulation of radial vascular pattern formation

Primary morphogenesis of the leaf temporally coincides with the appearance of the procambial strands that will give rise to its major veins (Nelson & Dengler, 1997; Dengler & Kang, 2001). Numerous mutants have been identified from screens for aberrant leaf shape in many species, but vascular pattern has been characterised only for a few of those (Tables 2 and 3). The most conspicuous vascular pattern defects observed in these leaf shape mutants are abnormalities in the number and placement of the major veins (i.e. those that are formed early at the morphogenetic stage of leaf development) (Tables 2 and 3). Similarly, enhanced and ectopic expression or antisense silencing of various genes results in leaf aberrations that are accompanied by vascular development effects (Tables 2 and 3).

Table 2.  Genes involved in radial vascular pattern formation and organismal adaxial-abaxial polarity establishment
GeneOrganismMutant phenotypeOver/misexpressor phenotype*Molecular identityReferences
Vascular anatomyAuxin physiologyVascular anatomyAuxin physiology
  • *

    Unless otherwise indicated, overexpression is achieved through expression driven by the Cauliflower Mosaic Virus 35S promoter.

AGO1 and ZLL/PNHArabidopsisAmphicribral vascular bundles in the cylindrical leaves of the ago1 homozygous mutant in the zll/pnh heterozygous backgroundUnknownUnknownUnknownPiwi/PAZ-domain proteinsLynn et al. (1999); Cerutti et al. (2000)
AVB1ArabidopsisAmphivasal vascular bundles in stem and leaves of homozygous and heterozygous mutantsNormal auxin transportUnknownUnknownTranscription factor of the HD-Zip classZhong et al. (1999); Ye (2002)
KAN1, 2 and 3ArabidopsisAmphivasal bundles in the stem of the triple mutantUnknownAbsence of vasculature in the cylindrical leaves of 35S-KAN1, AS1pro- KAN1 and AS1pro-KAN2UnknownTranscripton factors of the GARP classEshed et al. (2001); Kerstetter et al. (2001); Emery et al. (2003)
PHB/Athb14 and PHV/Athb9ArabidopsisAmphivasal vascular bundles in funnel-shaped leaves of the phb heterozygous dominant mutant. Absence of phloem or of both phloem and xylem in cylindrical leaves of the phb heterozygous dominant mutant. Amphicribral bundles in the cylindrical cotyledons of the phab phav rev triple loss-of-function mutantUnknownUnknownUnknownPHB/Athb14, PHV/Athb9, and REV/IFL1 are transcription factors of the HD-Zip III familyMcConnell & Barton (1998); McConnell et al. (2001); Emery et al. (2003)
PHAN and HBAnthirrinumIn the phan homozygous mutant, cylindrical leaves, which develop at the upper part of the stem, contain amphicribral vascular bundles. In the phan hb double mutant, all leaves are cylindrical and contain amphicribral vascular bundlesUknownUnknownUnknownPHAN is a transcription factor of the MYB class. The molecular identity of HB is unknownWaites & Hudson (1995); Waites et al. (1998); Waites & Hudson (2001)
REV/IFL1ArabidopsisAbsence of interfascicular fibres and reduced secondary xylem in the stem of the homozygous recessive mutant. Amphivasal bundles in the stem of the heterozygous and homozygous dominant mutantNormal auxin sensitivity. Reduced auxin transportUnknownUnknownTranscription factor of the HD-Zip III familyZhong et al. (1997); Zhong & Ye (1999), (2001); Ratcliffe et al. (2000); Otsuga et al. (2001); Emery et al. (2003)
RS1MaizeDisordered vascular anastomosis at the convoluted and sheath-like blade–sheath boundary in heterozygous and homozygous mutants. Increased and ectopic xylem production. Occasional amphivasal vascular bundles in the leafUnknownUnknownUnknownTranscription factor of the KNOX classBecraft & Freeling (1994); Schneeberger et al. (1995)
Table 3.  Genes involved in vascular development but with no clear effects on organismal polarity establishment
GeneOrganismMutant phenotypeOver/misexpressor phenotype*Molecular identityReferences
Vascular anatomyAuxin physiologyVascular anatomyAuxin physiology
  • *

    Unless otherwise indicated, overexpression is achieved through expression driven by the Cauliflower Mosaic Virus 35S promoter.

AGO1ArabidopsisReduced number or absence of higher-order veins in the narrow leaves of the homozygous mutant. Absence of vascularisation in the cylindrical leavesUnknownUnknownUnknownPiwi/PAZ-domain proteinBohmert et al. (1998); Cerutti et al. (2000)
AMP1/COP2/HPT/PTArabidopsisSimplified venation resembling that of cotyledons in the cotyledon-shaped leaves of the homozygous mutantUnknownUnknownUnknownGlutamate carboxypeptidaseJürgens et al. (1991); Chaudhury et al. (1993); Hou et al. (1993); Conway & Poethig (1997); Telfer et al. (1997); Mordhorst et al. (1998); Helliwell et al. (2001)
API7ArabidopsisPrimary vein often splits and rejoins in the pointed leaves with slightly incised margins of the homozygous mutantUnknownUnknownUnknownUnknownBerna et al. (1999); Candela et al. (2001)
APLArabidopsisAbsence of phloem in the homozygous mutant. Formation of ectopic xylem at positions where normally phloem differentiatesUnknownAbsence of xylem in the root of WOLpro-APLUnknownMYB-related transcription factorBonke et al. (2003)
AS1ArabidopsisReduced vascularisation in the small cotyledons and in the small and lobed leaves of the homozygous mutant. Secondary veins join the primary vein more proximally than in wild typeUnknownUnknownUnknownTranscription factor of the MYB classByrne et al. (2000); Sun et al. (2002)
AS2ArabidopsisReduced size or absence of recognisable primary vein in the asymmetric and lobed leaves of the homozygous mutant. Secondary veins run parallel and separately from the primary vein. Reduced number of higher-order veins. Discontinuous vascularisation. Delayed xylem differentiationUnknownAmphivasal vascular bundles in the cylindrical leavesUnknownLOB-domain proteinSemiarti et al. (2001); Iwakawa et al. (2002 ); Shuai et al. (2002); Lin et al. (2003)
Athb8ArabidopsisNormalUnknownIncreased and precocious vascularisation in root, stem, and leavesUnknownTranscription factor of the HD-Zip III familyBaima et al. (1995; 2001)
BOP1ArabidopsisMultiple primary veins associated with ectopic outgrowths in the petiole of the homozygous mutantUnknownUnknownUnknownUnknownHa et al. (2003)
COV1ArabidopsisIncreased vascularisation in the stem of the homozygous mutantNormal auxin sensitivityUnknownUnknownPutative integral membrane proteinParker et al. (2003)
CPD/DWF3/CBB3ArabidopsisIncreased phloem to xylem ratio in the vascular bundles of the stem of the homozygous mutantUnknownUnknownUnknownCytochrome P450Szeckeres et al. (1996); Choe et al. (1999)
CVP1/SMT2ArabidopsisReduced, misaligned, and discontinuous vascularisation in cotyledons of the homozygous mutant. Increased vascularisation in the stemNormal auxin content, sensitivity and transportUnknownUnknownSterol Methyltransferase2Carland et al. (1999; 2002)
CVP2ArabidopsisReduced and discontinuous vascularisation in cotyledons and leaves of the homozygous mutantNormal auxin content, sensitivity and transportUnknownUnknownUnknownCarland et al. (1999)
DC2.15CarrotDecreased phloem to xylem ratio in the veins of the twisted leaves of 35S-DC2.15 antisense. Twisting of one of the vascular bundles of the midveinUnknownUnknownUnknownPutative cell wall-plasmamembrane linker proteinHolk et al. (2002)
DLRiceAbsence of recognisable midrib in the narrow and ribbon-like leaves of the homozygous mutantUnknownMultiple midribs in the leaves of ACT1pro-DLUnknownTranscription factor of the YABBY classIwata & Omura (1971); Nagasawa et al. (2003); Yamaguchi et al. (2004)
DWF7/STE1/BUL1ArabidopsisDelayed procambium initiation in the small leaves of the homozygous mutant. Increased phloem to xylem ratio in the irregularly spaced vascular bundles of the stemUnknownUnknownUnknownΔ7 sterol C-5 desaturaseGashotte et al. (1995; 1996); Choe et al. (1999); Catterou et al. (2001)
ELI1ArabidopsisDiscontinuous and misaligned xylem strands in the root of the homozygous mutant. Increased xylem in the stemUnknownUnknownUnknownUnknownCaño-Delgado et al. (2000)
FIL/YAB1 and YAB3ArabidopsisReduced number of higher-order veins and discontinuous vascularisation in the petal-shaped and occasionally bifurcated leaves of the fil/yab1 yab3 double mutantUnknownAbsence of vasculature in cylindrical leavesUnknownTranscription factors of the YABBY classSawa et al. (1999); Siegfried et al. (1999)
FK/HYD2/ELL1ArabidopsisIncreased phloem to xylem ratio in the supernumerary/ectopic vascular bundles of the hypocotyl of the homozygous mutant. Reduced and discontinuous vascularisation in the small and abnormally shaped cotyledons and leavesIncreased auxin sensitivityUnknownUnknownSterol C-14 reductaseMayer et al. (1991); Jang et al. (2000); Schrick et al. (2000); Souter et al. (2002)
FKD1 and 2ArabidopsisSecondary veins fail to connect distally to pre-existing vasculature in cotyledons and leaves of single and double homozygous mutants. Increased number of freely ending veins. Discontinuous vascularisation. Delayed vascular development in fkd1Reduced auxin response in fkd1UnknownUnknownUnknownSteynen & Schultz (2003)
FZYPetuniaAbsence of secondary veins in the wide, curled-up leaves of the homozygous mutantNormal auxin contentUnknownIncreased auxin contentFlavin mono-oxygenase-like proteinTobeña-Santamaria et al. (2002)
HVEArabidopsisReduced vascularisation in cotyledons and leaves of the homozygous mutantNormal auxin sensitivityUnknownUnknownUnknownCandela et al. (1999; 2001)
HYD1ArabidopsisReduced and discontinuous vascularisation in the small and irregularly shaped cotyledons and leaves of the homozygous mutant. Reduced phloem to xylem ratio in the supernumerary/ectopic vascular bundles of the hypocotylIncreased auxin sensitivityUnknownUnknownSterol Δ8-Δ7 isomeraseTopping et al. (1997); Souter et al. (2002)
KN1MaizeIncreased distance between longitudinal veins and vein zig-zagging in the wide, occasionally twisted and convoluted leaves of the homozygous and heterozygous mutantsUnknownUnknownUnknownTranscription factor of the KNOX classFreeling and Hake (1985); Vollbrecht et al. (1991); Kerstetter et al. (1994)
KNAT1/BPArabidopsisPremature and enhanced xylem and sclerenchyma differentiation in the stem of the homozygous mutantUnknownDecreased size of the primary vein in the small leaves of PETEpro-KNAT1 lettuce plants. Increased size and number of veins at the leaf margins. Delayed xylem and sclerenchyma differentiation, and ectopic lignification of cortical and epidermal cells in stems of 35S-KNAT1 Arabidopsis plantsUnknownTranscription factor of the KNOX classLincoln et al. (1994); Frugis et al. (2001); Douglas et al. (2002); Mele et al. (2003)
LEM7ArabidopsisSecondary veins do not join the primary vein and run parallel in the narrow and asymmetric leaves of the homozygous mutant. Reduced number of higher-order veinsUnknownUnknownUnknownUnknownMeisel et al. (1996)
LEPArabidopsisUnknownUnknownIncreased xylem in the midvein of the leaves, which display blade growth on both sides of the petioleUnknownTranscription factor of the AP2/EREBP classvan der Graaff et al. (2000; 2002)
LOP1/TRN1ArabidopsisBifurcated and twisted primary vein and reduced vascularisation in the narrow, asymmetric, and irregularly shaped leaves with twisted petiole of the homozygous mutantReduced auxin transportUnknownUnknownUnknownCarland & McHale (1996); Cnops et al. (2000)
LSN1MaizeOccasional absence of recognisable midvein in the narrow and wavy leaves with curved and rolled margins of the homozygous mutant. Reduced distance between longitudinal veins. Altered reciprocal positioning of xylem and phloem in the vascular bundles of the leavesUnknownUnknownUnknownUnknownLandoni et al. (2000)
ISR and SR2MaizeReduced distance between longitudinal veins at the margins of the narrow leaves of the isr sr2 double mutantUnknownUnknownUnknownISR is a chloroplast protein similar to bacterial phosphatases. The molecular identity of SR2 is unknownPark et al. (2000)
MBLMilletAbsence of recognisable midvein in the narrow and ribbon-like leaves of the homozygous mutant. Reduced number of large veins. Increased number of small veins between large veinsUnknownUnknownUnknownUnknownFladung et al. (1991); Fladung (1994)
MRL1 and 2MilletAbsence of recognisable midvein in the narrow and ribbon-like leaves of the single homozygous mutants. Reduced number of longitudinal veinsUnknownUnknownUnknownUnknownRao et al. (1988; 1989)
Oshox1RiceUnknownUnknownPrecocious vascular differentiation in stem and rootIncreased auxin sensitivity and transportTranscription factor of the HD-Zip II familyScarpella et al. (2000; 2002)
OsKn2 and 3RiceUnknownUnknownReduced number and increased size of longitudinal veins in the narrow, irregularly shaped, and convoluted leaves. Irregular cellular arrangment in the xylemNormal auxin transportTranscription factors of the KNOX classPostma-Haarsma et al. (2002)
OsPNHRiceReduced number of longitudinal veins in the narrow, asymmetric, fused, and twisted leaves of ACT1pro-OsPNH antisense. Absence of vascularisation in the small cylindrical leavesUnknownUnknownUnknownPiwi/PAZ-domain proteinNishimura et al. (2002)
OV1BarleyAbsence of recognisable midvein in the narrow and ribbon-like leaves of the homozygous mutantUnknownUnknownUnknownUnknownSeip & Tsuchiya (1979)
PLSArabidopsisReduced vascularisation in the leaves of the homozygous mutantReduced auxin sensitivityIncreased leaf vascularisationNormal auxin sensitivityShort polypeptide with no significant homology with known proteinsCasson et al. (2002)
RON1ArabidopsisIncreased number of freely ending veins near the margins of the broad and rounded leaves of the homozygous mutantUnknownUnknownUnknownUnknownBerna et al. (1999); Candela et al. (2001)
RS2MaizeMultiple midveins and/or increased size of midvein in the narrow and twisted leaves of the homozygous mutant. Increased size and number of longitudinal veins and increased distance between them in the wide leavesReduced auxin transportUnknownUnknownTranscription factor of the MYB classSchneeberger et al. (1998); Tsiantis et al. (1999a,b); Timmermans et al. (1999)
SEM1MaizeReduced number and size of transverse veins in the leaf sheath of the homozygous mutant. Reticulated vascularisation in sheath marginal outgrowths developing at the boundary with the bladeReduced auxin transportUnknownUnknownUnknownScanlon et al. (2002)
SFCArabidopsisReduced and discontinuous vascularisation in cotyledons and leaves of the homozygous mutantIncreased auxin sensitivityUnknownUnknownUnknownDeyholos et al. (2000)
SMT1/CPH/ORCArabidopsisReduced and discontinuous vascularisation in the small and lobed cotyledons of the homozygous mutantReduced auxin transportUnknownUnknownSterol Methyltransferase1Scheres et al. (1996); Diener et al. (2000); Schrick et al. (2002); Willemsen et al. (2003)
TAN1MaizeThe position of the veins relative to each other and the surrounding tissues is highly irregular/disorganised in the leaves of the homozygous mutantUnknownUnknownUnknownMicrotubule binding proteinSmith et al. (1996); Smith (2001)
VAN1–4 and VAN6ArabidopsisReduced and discontinuous vascularisation in cotyledons and leaves of the homozygous mutants. Altered reciprocal positioning of xylem and phloem in the hypocotyl of van1, 2, 4 and 6UnknownUnknownUnknownUnknownKoizumi et al. (2000)
VAN5ArabidopsisIncreased size of the veins in the cotyledons and in the narrow and irregularly shaped leaves of the homozygous mutant. Reduced number or absence of higher-order veins. Discontinuous and misaligned vascularisationUnknownUnknownUnknownUnknownKoizumi et al. (2000)
VASArabidopsisNormalUnknownIncreased phloem in hypocotyls, stems, and leaf petiolesUnknownWeak similarity to non-specific lipid transfer proteinsvan der Graaff et al. (2002)
VCSArabidopsisSlightly reduced vascularisation in the pointy leaves of the homozygous mutant grown at lower temperatures. Highly reduced vascularisation in the narrow, asymmetric, epinastic, and chlorotic leaves at higher temperaturesNormalUnknownUnknownWD-domain proteinDeyholos et al. (2003)
VEP1/AWI31ArabidopsisReduced and discontinuous vascularisation in the leaves of VEP1/AWI31 antisense. Reduced vascularisation in cotyledons, stem, and rootUnknownUnknownUnknownPolypeptide similar to animal proteins involved in apoptosis (death-domain proteins)Jun et al. (2002); Yang et al. (1997)
VH1ArabidopsisReduced phloem transport at early stages of leaf development in the homozygous mutantUnknownReduced vascularisation in juvenile leavesUnknownLRR receptor kinaseClay & Nelson (2002)
WI1MaizeDelayed or absent metaxylem element differentiation in the stem of the homozygous mutantUnknownUnknownUnknownUnknownPostlethwait & Nelson (1957)
WOL/CRE1/AHK4ArabidopsisAbsence of phloem in the root and basal part of the hypocotyl of the homozygous mutantUnknownUnknownUnknownCytokinin receptor two-component histidine kinaseScheres et al. (1995); Mähönen et al. (2000); Inoue et al. (2001); Ueguchi et al. (2001a,b)
XTC1 and 2ArabidopsisSimplified venation resembling that of cotyledons in the cotyledon-shaped leaves of the homozygous mutantsUnknownUnknownUnknownUnknownConway & Poethig (1997)

The most striking leaf shape mutations that also alter vascular pattern are those in genes thought to be responsible for the maintenance of leaf dorsoventral (or adaxial-abaxial) polarity (Table 2). The formation of flat lateral organs involves the specification of adaxial and abaxial cell identities. Leaf anatomy and gene expression studies suggest that incipient leaf primordia are initially uniform and become polarized in the adaxial-abaxial dimension as soon as the primordium emerges from the flanks of the shoot apical meristem (Hudson & Waites, 1998; Bowman, 2000). A growing number of regulatory genes have been implicated in adaxial-abaxial patterning (Sessions & Yanofsky, 1999; Bowman et al., 2002). The adaxial-abaxial polarity of a leaf is reflected in the polar organisation of leaf veins, with xylem located towards the adaxial side of the leaf and phloem facing the abaxial side, referred to as collateral bundle organisation (Fig. 6). An integrated regulation of leaf and vascular strand polarity is suggested by the phenotypes resulting from mutations in various genes. Both adaxialised and abaxialised leaves do not form blades and thus develop as cylindrical organs with a single central vascular strand. A polarity shift of the surrounding organ is reflected in a corresponding shift towards amphivasal (i.e. a xylem ring surrounds a central phloem cylinder) or amphicribral (i.e. the phloem is organised as a sheath around a xylem cylinder) organisation (Fig. 6). Examples of leaves that develop as cylindrical organs with correspondingly altered internal vein organisation have been observed in Arabidopsis argonaute1 (ago1) mutant plants heterozygous for zwille/pinhead (zll/pnh) (Lynn et al., 1999), in the phantastica (phan) single mutant and the phan handlebar (hb) double mutant of Anthirrinum (Waites & Hudson, 1995; Waites et al., 1998; Waites & Hudson, 2001), in the phabulosa (phb) (McConnell & Barton, 1998; McConnell et al., 2001) and the amphivasal bundle1 (avb1) dominant mutants of Arabidopsis (Zhong et al., 1999), and in the Rough sheath1 (Rs1) dominant mutant of maize (Becraft & Freeling, 1994). In phb, avb1, and Rs1, the vascular strands are amphivasal, whereas in phan, phan hb, and the ago1 mutant in the zll/pnh heterozygous background, the single vein is amphicribral. Furthermore, ectopic overexpression of genes involved in leaf abaxial polarity acquisition, such as the KANADI (KAN) genes (Eshed et al., 2001; Kerstetter et al., 2001) and the FILAMENTOUS FLOWER/YABBY1 gene (Sawa et al., 1999), results in abaxially radialized leaves that do not develop any vasculature. Vascular bundles are also missing from most leaves of severely adaxialised phb plants (McConnell & Barton, 1998). As both extreme phenotypes are incompatible with vascular tissue formation, it is possible that intermediate positional values or adaxial-abaxial polarity per se may be essential for vascular tissue formation.

Figure 6.

Radial vascular tissue patterns in vascular bundles of the shoot. (a) Transverse section through a mature Ranunculus stem showing a collateral vascular bundle organisation. The phloem is located towards the abaxial side in flat organs, and towards the outside in cylindrical organs. The xylem is located towards the adaxial side in flat organs, and towards the inside in cylindrical organs. (b) Transverse section through a mature Polypodium rhizome showing an amphicribral vascular bundle organisation. The phloem surrounds the xylem. (c) Transverse section through a mature Cordyline (Acorus calamus) rhizome showing an amphivasal vascular bundle organisation. The xylem encloses the phloem. (d) Transverse section through a mature Cucurbita stem showing a bicollateral vascular bundle organisation. The phloem is located towards both the abaxial and the adaxial sides in flat organs, and towards both the inside and the outside in cylindrical organs. The xylem is located between the phloem poles. p, phloem; x, xylem. Slides courtesy of N. Dengler.

Adaxial-abaxial polarity may be derived from ancestral central-peripheral polarity in cylindrical organs (Tasaka, 2001), and this seems to be reflected in the organization of vascular bundles in stems and in the central-peripheral organization of hypocotyls, which in cases of extreme adaxial-abaxial shifts may be devoid of any vascular tissues (Kerstetter et al., 2001). In further support of this view, both leaves and stems of the avb1 mutant display amphivasal bundles (Zhong et al., 1999). A dominant mutation in the REVOLUTA/INTERFASCICULAR FIBERLESS1 (REV/IFL1) gene, which encodes a member of the same family of HD-Zip III transcription factors as PHB/Athb14 and PHAVOLUTA (PHV)/Athb9 (Zhong & Ye, 1999; Ratcliffe et al., 2000; McConnell et al., 2001; Otsuga et al., 2001), results in amphivasal bundles in the stem (Emery et al., 2003). Similarly, stems of kan1 kan2 kan3 triple mutants also show amphivasal bundles (Emery et al., 2003). Conversely, rev phb phv triple loss-of-function mutants, which fail to develop postembryonic organs, display amphicribral vascular bundles in their radialised cotyledons (Emery et al., 2003). While radialised vascular bundles in leaves of the phb dominant mutant could be viewed as a consequence of the radialisation of the leaves, the altered vascular patterning in the stems of rev dominant mutants and kan1 kan2 kan3 plants seems to suggest a direct role for KANADI and HD-Zip III genes in vascular patterning.

Recently, microRNAs (miRNAs) were identified in Arabidopsis that are complementary to a specific region of the PHB, PHV, and REV/IFL1 transcripts where the dominant mutations in these genes occur (Reinhart et al., 2002; Rhoades et al., 2002; Emery et al., 2003). This discovery suggests that the phenotypes induced by these HD-Zip III gene mutations may be due to the disruption of miRNA regulation (Reinhart et al., 2002; Tang et al., 2003). Accordingly, a modified REV/IFL1 sequence with reduced complementarity to these miRNAs, but with no change in amino acid sequence, when introduced into wild-type plants, gave rise to phenotypes similar to those induced by the dominant rev mutation (Emery et al., 2003). The AGO1 and ZLL/PNH genes, which encode members of a family of key components in RNA-mediated gene-silencing pathways that use short interfering RNAs or miRNAs to select their targets in a sequence-dependent manner (Cerutti et al., 2000; Carrington & Ambros, 2003), are expressed in a pattern that overlaps with that of PHB, PHV, and REV/IFL1 (Moussian et al., 1998; Lynn et al., 1999; Zhong & Ye, 1999; Otsuga et al., 2001; Emery et al., 2003). Recent evidence suggests that the miRNA-dependent regulation of at least some of these HD-Zip genes may be mediated by AGO1, and possibly by ZLL/PNH (Kidner & Martienssen, 2004). The maize gene ROLLED LEAF1 is the homolog of the Arabidopsis REV/IFL1 gene, and the semidominant Rld1-O mutation results in adaxialisation or partial reversal of adaxial-abaxial polarity in the leaf (Nelson et al., 2002; Juarez et al., 2004). However, it remains to be established whether the changes in adaxial-abaxial polarity observed in Rld1-O leaves have any effect on the internal organisation of vascular bundles.

In summary, there is accumulating evidence for an integrated regulation of adaxial-abaxial patterns within vascular strands and surrounding tissues in the majority of plant organs, and this control seems to be sufficient to account for the internal organization of vascular bundles and for their functional integration into the organ context. It will be a challenge of the near future to determine how the regulation of radial and longitudinal vascular patterning are integrated at the molecular level. The involvement of auxin in the regulation of radial vascular pattern formation is indeed not self-evident, in that transcription of the HD-Zip III genes that have been implicated in this process is neither responsive to changes in auxin levels nor dependent upon auxin signal transduction (Mattsson et al., 2003).

VIII. Genetic screens for vascular development mutants

Direct screens for mutants displaying aberrant vascular patterns within relatively normally shaped organs are expected to identify pathways specific for vascular tissue development. Furthermore, such screens are not biased by associated embryo or auxin perception defects and might thus identify novel signalling mechanisms in vascular development. Mutants with specific vascular defects recovered from such screens are discussed below. These mutants and the corresponding genes are also listed in Table 3, together with genes that have been identified through other means (discussed elsewhere in this review), and with genes with less clear roles in vascular development.

1. Foliar organ screens

Cotyledon screens  Cotyledon vascular patterns, which are simple, reproducible, and readily visible after germination, are well-suited for large-scale screens for vascular mutants. Mutations in two Arabidopsis genes, COTYLEDON VASCULAR PATTERN1 and 2 (CVP1 and CVP2), affect vascular tissue organisation in cotyledons without affecting their overall morphology (Carland et al., 1999). Both the CVP1 and 2 genes seem to be required at early procambial stages, but otherwise have clearly distinct functions. Mutations in CVP1 result in thickened vascular strands composed of supernumerary, insufficiently elongated, and misaligned vascular cells. This has been interpreted as a defect in the signalling process that integrates oriented cell differentiation and elongation along the strand axis. By contrast, mutations in CVP2 result in supernumerary strands that often terminate prematurely, suggesting that CVP2 promotes vascular differentiation, and its action is thus required for the full completion of strand maturation. Unlike cotyledons, the vascular pattern of cvp1 leaves is indistinguishable from that of wild type. On the other hand, cvp2 leaves display disconnected xylem strands similar to those seen in cotyledons. As a result of this defect, the vascular pattern of cvp2 leaves appears less reticulate when compared to wild type. Auxin content, transport, and perception do not appear to be compromised in cvp mutants, suggesting that the vascular defects are not further manifestations of impaired auxin flow or response. Mutations in cvp1 and cvp2 could therefore identify auxin-independent mechanisms important for vascular strand formation.

The CVP1 gene has now been cloned and found to encode the STEROL METHYLTRANSFERASE2 (SMT2) enzyme (Carland et al., 2002). The phenotypes of mutants impaired in sterol and brassinosteroid biosynthesis or perception suggest that these classes of chemically related compounds regulate diverse embryonic and postembryonic processes in plants, including vascular development (Szekeres et al., 1996; Choe et al., 1999; Jang et al., 2000; Souter et al., 2002). However, vascular defects in these classes of mutants are associated with severe organ aberrations, pointing to a more general role for sterols and brassinosteroids in concerted cell division and expansion. The discovery that CVP1 is SMT2 suggests that sterol signalling may also play a specific role in vascular patterning and/or that sterols are involved in an aspect of membrane organisation that is essential for continuous vascular strand formation (Carland et al., 2002).

In order to generate a functional vascular system, mature phloem and xylem conduits need to be continuous over long distances. This requires that procambial cells are properly aligned with each other during strand formation. Because of this strict geometric requirement, vascular cells may represent one of the most sensitive indicators of general perturbations in patterns of coordinated cell division and differential expansion. The CVP1 gene is expressed in a pattern that extends outside the vasculature (Carland et al., 2002), suggesting possible functions of this gene in other tissues, which might have been obscured in the cvp1 mutant by the presence of functionally redundant SMT3 activity (Carland et al., 2002). It is therefore possible that sterols may be required for cell polarity as a whole, rather than for specific aspects of vascular development. In support of this view, the smt1orc mutant of Arabidopsis displays mislocalisation of different members of the PIN protein family, reduced auxin transport, and vascular discontinuities, but these alterations are embedded in defects in the establishment of polarity in a variety of cell types (Willemsen et al., 2003). Furthermore, indirect evidence supports a role for sterols in the regulation of patterning events beyond those that organise the vascular system. A highly conserved sterol/lipid-binding domain has been identified in the HD-Zip III family of transcription factors (Ponting & Aravind, 1999), and mutations within this domain in the PHB, PHV, and REV/IFL1 genes result in dramatic radial organ patterning and shoot meristem phenotypes (McConnell et al., 2001; Otsuga et al., 2001).

The high reproducibility of the vascular patterns of Arabidopsis cotyledons has also been employed in another screen that has identified six genes, VASCULAR NETWORK (VAN) 1–6, putatively involved in vascular pattern formation (Koizumi et al., 2000). All van mutants display interruption or partial loss of minor veins of cotyledons and vegetative leaves, without any clearly detectable effect on major vein formation. Furthermore, all the van mutant vascular defects, except those of the van5 mutant, occur in leaves that do not display strong deviations from the wild-type shape. In the case of the van3 mutant, the vascular defects were traced back to interruption of secondary procambial strands in late-embryo cotyledons. Mutations in any of these six genes interfere with the viability of the mutant plants at various stages of vegetative development. This is unlikely to result from the leaf network pattern defect alone, as other Arabidopsis mutants with dramatically reduced leaf vasculature grow to adult stages. The lethality might instead be caused by vascular disorganisation along the plant axis, in that reduced phloem was observed in cross-sections through the hypocotyl of several van mutants (Koizumi et al., 2000). From these phenotypes, at least some of the VAN gene products might turn out to be regulators of internal patterning within vascular strands.

Leaf screens  Although the vascular pattern of the first Arabidopsis leaf is less reproducible than that of the cotyledon, it has been successfully employed to identify further genes involved in vascular pattern formation. A survey of leaf venation patterns in more than 250 Arabidopsis ecotypes identified numerous abnormal leaf shape variants with associated aberrant venation patterns, but only one ecotype with altered venation within otherwise normal leaves (Candela et al., 1999). The reduced venation density in the ecotype Eifel-5 has been attributed to the single locus HEMIVENATA (HVE). Double mutant analysis has recently shown that the HVE gene does not genetically interact with the PIN1, AXR1, and MP genes, suggesting that auxin transport or sensitivity defects are not the primary cause for the vascular defects in the hve mutant (Candela et al., 2001). In further support of this view, the hve mutant displays normal sensitivity to exogenous auxin and PATI application (Candela et al., 2001).

In the scarface (sfc) mutant, higher-order vascular strands are primarily affected, and interruptions in lower-order veins in the leaf or in vascular strands along the plant axis are rare or absent (Deyholos et al., 2000). In fragmented veins, stretches of fully differentiated vascular strands are separated by regions apparently devoid of any kind of vascular tissues. Closer inspection confirmed that, as for the van3 mutant, strands were already interrupted at procambial stages, and therefore neither phloem nor xylem differentiated continuously. These observations suggest that aligned differentiation of vascular cells within a short segment does not depend on tissue continuity in a larger context, a conclusion with possible implications for the underlying patterning mechanism. Unlike mutations in the MP, AXR6, and BDL genes, mutations in the SFC gene are associated with enhanced, rather than reduced, responses to external auxin. Therefore, SFC has been suggested to represent a negative regulator of auxin response. However, mp sfc double mutant analysis suggests that SFC has partially overlapping functions with MP in vascular patterning. It seems difficult to reconcile the proposed role for SFC as a negative regulator of auxin response with its redundancy with MP, which mediates positive auxin signals, but this issue might be resolved at the molecular level, once the identity of the SFC gene has been determined.

A dominant mutant line with abnormal leaf venation pattern has been isolated in the frame of a broader project of random antisense mutagenesis in Arabidopsis (Jun et al., 2002). The mutant phenotype is due to the antisense suppression of the VEIN PATTERNING1/ARABIDOPSIS WOUND INDUCIBLE31 (VEP1/AWI31) gene. The vep1/awi31 mutant shows a reduction in the number of minor veins, and their connection with major veins is incomplete. The mutation also reduces the complexity of venation patterns in cotyledons. Furthermore, the more slender mutant stems contains fewer and smaller vascular bundles. Finally, the vep1/awi31 mutant shows reduced secondary vascular tissue formation in both stem and root. The VEP1/AWI31 gene encodes a predicted peptide that displays some similarity to animal proteins involved in apoptosis (Yang et al., 1997).

The forked1 and 2 (fkd1 and 2) mutants of Arabidopsis have been recovered because, in their cotyledons and leaves, secondary veins fail to join the pre-existing vasculature distally (Steynen & Schultz, 2003). Detailed morphometric analysis of the vascular defects in these mutants has shown that the overall number of freely ending veins is increased, and that isolated patches of xylem occur frequently in mature mutant leaves. This suggests that the FKD1 and 2 genes may be required for vein network formation. In fkd1, the reduced vascular reticulation is associated with delayed vascular development and reduced auxin-responsive reporter gene expression in incipient veins. Because PATI-induced auxin accumulation at the leaf margin normalises both the vascular and auxin-response mutant defects, it has been proposed that FKD1 is necessary for vascular development in response to auxin signals.

A significant number of Arabidopsis leaf vascular pattern mutants are characterised by failure of secondary veins to connect to the midvein at the leaf tip, and by reduced vascular pattern complexity (Przemeck et al., 1996; Carland et al., 1999; Koizumi et al., 2000; Steynen & Schultz, 2003). A recent study suggests that these vascular phenotypes can arise not only because of specific defects in vascular development, but also because of premature differentiation of the nonvascular tissues of the leaf (Scarpella et al., 2004). Therefore, the final overall vascular pattern of an Arabidopsis leaf appears to be the result of the interaction between a vascular patterning process and the differentiation of the nonvascular tissues. This conclusion has important consequences, in that it introduces the differentiation of nonvascular tissues as an additional influence that has to be taken into account when studying both wild-type and mutant leaf vascular patterns.

Some attempts have also been made to identify vascular pattern mutants in genetic screens of leaves of monocot plants. Several potential pattern mutants were identified after chemical mutagenesis in millet, including variants in the spacing of veins, supernumerary small veins, aberrant bundle sheaths, and the absence of midrib (Fladung, 1994). However, the heavy level of mutagenesis used to produce these mutants raises the possibility that the observed phenotypes may be attributable to the combined effects of mutations at more than one locus. This hypothesis could not be tested because of the low fertility of several of the isolated variants and the extreme difficulty of propagating millet by artificial pollination.

The vascular strands in the leaves of the tangled1 mutant of maize seem disorganised and irregularly spaced (Smith et al., 1996; Jankovsky et al., 2001). However, these defects derive from prolonged and abnormally oriented divisions of all cells of the leaf primordium, and not only of those of the vascular lineage (Smith et al., 1996; Jankovsky et al., 2001). More recent attempts to isolate maize mutants specifically defective in leaf vascular pattern formation have not been successful (T. Nelson, personal communication). Therefore, the ral1 mutant of rice, discussed above in relation to auxin response, is thus far the only mutant in monocots with specific vascular pattern defects.

The observation that the procambium can differentiate discontinuously with the pre-existing vasculature in several leaf vascular pattern mutants suggests that continuity may not be a prerequisite for vascular strand formation. However, differentiating procambium is not necessarily a requirement, but rather the result of a preceding patterning mechanism, regardless of whether this mechanism involves a continuous signal. Interestingly, even in discontinuous vein mutants, the interrupted procambium is still formed along the same paths where it would normally differentiate continuously in wild type, suggesting that the isolated stretches of procambium are visible segments within a larger invisible pattern. A limitation of the genetic analysis of venation patterning could therefore lie in the inability to identify, visualise, and manipulate implicated molecular cues.

2. Stem screens

The examination of cross-sections of inflorescence stems represents another type of genetic screen with the potential to identify mutants with abnormal tissue composition and vascular bundle organisation. The most obvious defect in ifl1 recessive mutants is the absence of fibres between stem vascular bundles (Zhong et al., 1997). Furthermore, in mutant vascular bundles, secondary xylem is reduced or absent. IFL1 is the same gene as REV, a gene implicated in shoot apical meristem development (Talbert et al., 1995), and encodes a member of the HD-Zip III family of transcription factors (Zhong & Ye, 1999; Ratcliffe et al., 2000; Otsuga et al., 2001). It has been shown that the anatomically recognisable vascular defects in ifl1 mutants are preceded by a reduction in PAT and that the defects are associated with lower expression of putative auxin efflux carrier genes (Zhong & Ye, 2001). This suggests that the primary defect in ifl1 mutants may lie in a defective PAT capacity, which, in turn, would result in defective secondary xylem and fibre differentiation. However, the relationship between the reduction in PAT detected in ifl1 allele and the defects in meristem development reported for the rev mutants remains to be elucidated.

The same type of screen that lead to the discovery of ifl1 has identified the avb1 mutant of Arabidopsis. In both leaves and stems of this mutant, the normal collateral placement of xylem and phloem is disrupted, leading to the formation of amphivasal vascular bundles with xylem surrounding phloem (Zhong et al., 1999). Furthermore, the ring-like organisation of vascular bundles typical of wild-type stems is replaced in avb1 by a disordered arrangement of branched bundles that are scattered into the central pith. In spite of the severely distorted vascular organisation, mutant stems do not display alterations in PAT (Zhong et al., 1999), suggesting that the AVB1 gene might identify a PAT-independent regulatory pathway involved in vascular tissue development. Like REV/ILF1, AVB1 has recently been reported to encode a member of the HD-Zip class of transcription factors (Ye, 2002).

A more recent screen for altered vascular pattern in the Arabidopsis stem has yielded the continuous vascular ring1 (cov1) mutant (Parker et al., 2003). Mutations in the COV1 gene result in seemingly laterally expanded and radially flattened vascular bundles in the stem, with abundant phloem differentiation and associated reduction of the interfascicular fibre region. Anatomical analysis of vascular bundle development revealed that immediately below the shoot apical meristem of cov1 mutants, phloem differentiates ectopically in the regions adjacent to and between normally patterned vascular bundles. cov1 mutants display normal vascular pattern in other organs and are not affected in auxin sensitivity. Furthermore, direct measurement of endogenous auxin levels indicated no difference between cov1 and wild type in the upper stem, where the mutant vascular defects were first detected. A small difference in internal auxin content was detected at the base of the stem, but this may be a consequence, rather than the cause, of the altered vascular tissue development in cov1 plants. The COV1 gene encodes a predicted membrane protein of presently unknown function.

3. Other screens

As expected from the various roles the vascular system plays in plants, mutants with specific defects in vascular tissue development have also been isolated through screens that were not originally designed for that specific purpose. For example, the wooden leg (wol) mutant was identified in a screen aimed to isolate genes involved in the control of radial tissue organisation in the Arabidopsis root (Scheres et al., 1995). In the wol mutant, all procambial cells of the root and basal part of the hypocotyl differentiate into protoxylem, a defect that has been associated with reduced division of procambial cells in those organs (Scheres et al., 1995; Mähönen et al., 2000). The cloning of WOL has shown it to be allelic to the CYTOKININ RESPONSE1 (CRE1) gene, which encodes a cytokinin receptor (Mähönen et al., 2000; Inoue et al., 2001), thus providing novel evidence for a role of cytokinin in vascular development. Consistent with the mutant phenotype, WOL/CRE1 is expressed in the procambium of the embryonic axis (Mähönen et al., 2000). Recent evidence suggests, however, that WOL/CRE1-mediated cytokinin signal transduction and procambial cell proliferation may operate through at least partially independent regulatory circuits (de Leon et al., 2004).

Because of defective phloem-specific cell division and differentiation programs, the altered phloem development (apl) seedling lethal mutant of Arabidopsis does not develop phloem, and ectopic xylem is formed at positions normally occupied by phloem (Bonke et al., 2003). The APL gene encodes a nuclear-localised MYB-related protein and is expressed in a phloem-specific fashion throughout plant development. Ectopic APL expression throughout the procambium of the root results in the absence of protoxylem and delayed metaxylem differentiation, and suppresses the all-protoxylem vascular pattern of wol roots, resulting in an array of phloem–xylem patterns. These results, together with the differentiation of xylem elements in positions normally occupied by phloem in the apl mutant, suggest that in addition to promoting phloem differentiation, APL may also be required for inhibiting xylem differentiation in phloem poles during vascular development.

The ectopic lignification1 (eli1) mutant of Arabidopsis was isolated because of its altered pattern of staining with a lignin-specific dye (Caño-Delgado et al., 2000). In association with the patchy mosaic of intense lignin staining in nonvascular cells, eli1 roots exhibited interruptions in the xylem strands. It remains to be established whether such discontinuities originate from defects in procambium formation or in its subsequent differentiation into xylem elements.

The floozy (fzy) mutant of petunia has been identified because the initiation of the floral organ primordia of the three outermost whorls is blocked at an early stage (Tobeña-Santamaria et al., 2002). The leaves of the fzy mutant are slightly wider than wild-type leaves and frequently appear curled-up, but otherwise do not show any obvious deviation from the wild-type leaf shape. However, fzy leaves specifically lack secondary veins, while the midvein and the network of tertiary and quaternary veins seem to be present. The FZY gene encodes a flavin mono-oxygenase-like protein, and its ectopic overexpression results in increased auxin levels and in phenotypes resembling those of auxin-overproducing plants. Arabidopsis homologs of the FZY protein have been shown in vitro to be able to catalyse a step in a putative auxin biosynthesis pathway (Zhao et al., 2001). However, because of the difficulty in judging the significance of this activity in vivo, indirect effects of FZY on auxin synthesis are still conceivable.

The POLARIS (PLS) gene of Arabidopsis was identified as a promoter-trap transgenic line showing reporter gene expression in the embryonic and seedling root tip and, at a lower level, in the vascular tissues of the aerial parts of the seedling (Topping et al., 1994; Topping & Lindsey, 1997; Casson et al., 2002). Mutation in the PLS gene results in fewer higher-order veins in the leaf, reduced response to auxin, and enhanced sensitivity to cytokinin (Casson et al., 2002). The PLS gene has recently been cloned and found to encode a small peptide required for correct auxin-cytokinin homeostasis (Casson et al., 2002).

Similar to PLS, the VASCULAR HIGHWAY1 (VH1) gene of Arabidopsis was isolated from a screen of an enhancer-trap collection (Clay & Nelson, 2002). In the originally isolated line, the reporter gene is expressed in all subepidermal cells of young leaf primordia, but becomes restricted to procambial cells at later stages of leaf organogenesis. The VH1 gene encodes a leucine-rich repeat receptor kinase and is expressed in all subepidermal tissues of young lateral organs and globular and heart-stage embryos. At subsequent stages of development, VH1 expression becomes localised to the cells of the procambium, and the expression persists during vascular differentiation. Mutations in VH1 give rise to wild-type-looking plants. However, vh1 leaves exhibit aberrations in chloroplast development and undergo premature and uniform senescence. These defects have been tentatively attributed to the reduced phloem transport that has been visualised in the anatomically normal veins of the mutant at early stages of leaf development, which could possibly be an indication of delayed phloem differentiation. Overexpression of VH1 gives rise to seedlings with severe defects in the coordinated patterning of the vascular and nonvascular tissues of juvenile leaves, and these aberrations have been shown to be associated with reduced proliferation and premature differentiation of nonvascular cell types.

The dominant lettuce mutant of Arabidopsis was recovered from an activator T-DNA tagging screen because blade growth was observed on both sides of the leaf petiole (van der Graaff et al., 2000). The leaf phenotype was shown to be due solely to the overexpression of the APETALA2-like LEAFY PETIOLE (LEP) gene, although another gene, mac12.6, was shown to be up-regulated in the let mutant. Subsequent analysis has revealed that the size of the vascular bundles in the aerial organs of the let mutant is increased because of an increased number of both phloem and xylem cells (van der Graaff et al., 2002). Overexpression of the LEP gene and mac12.6, renamed VASCULAR TISSUE SIZE (VAS), and encoding a non-specific lipid transfer protein each accounted for a separate component of the let vascular phenotype. In fact, LEP overexpression could recapitulate both the leafy petiole phenotype and the increase in xylem cell number, whereas VAS overexpression resulted in an increased phloem cell number without any overt effect on organ morphology.

The wilted1 (wi1) recessive mutant of maize was originally isolated because of its pronounced and constitutive phenotypical signs of water deficiency, even when the mutant plants were grown in soil nearly saturated with water or in water culture (Postlethwait & Nelson, 1957). However, the primary defect in wi1 plants appears to be the delayed maturation of late metaxylem elements in the vascular bundles of the stem. Preliminary results suggest that a similar anatomical defect could be the reason for the wilted phenotype displayed by the Wilty2, Wilty3 and Wilty-2445 dominant mutants of maize (Rock & Ng, 1999).

IX. Genes involved in vascular development identified through reverse genetics approaches

Although the involvement of transcription factors of other classes is by no means excluded, it is noteworthy that several members of the HD-Zip class seem to play a role in different aspects of vascular tissue development. As discussed above, HD-Zip genes, such as REV/IFL1 and PHB, have been identified by forward genetics screens because of their mutant phenotypes associated with vascular tissue aberrations (Zhong et al., 1997; McConnell & Barton, 1998; Zhong & Ye, 1999; Ratcliffe et al., 2000; McConnell et al., 2001; Otsuga et al., 2001). In addition, members of the same class of genes were found to be specifically expressed in different cell types and/or during different stages of vascular tissue development in different plant species, and the functions of some of these genes were subsequently addressed in a reverse genetics approach (e.g. Athb8: Baima et al., 1995; Baima et al., 2001; Kang & Dengler, 2002, 2004; Scarpella et al., 2004; Athb15: Ohashi-Ito & Fukuda, 2003; Athb20Mattsson et al., 2003; CHB6: Hiwatashi & Fukuda, 2000; Oshox1: Scarpella et al., 2000, 2002; Vahox1: Tornero et al., 1996; ZeHB3: Nishitani et al., 2001, 2002; ZeHB-10, −11, −12, and −13: Ohashi-Ito et al., 2002; Ohashi-Ito & Fukuda, 2003). Members of the HD-Zip class have been grouped in different families based on sequence homologies (Sessa et al., 1994). The HD-Zip family III is a small group that includes genes, such as Athb8, Athb15, IFL1/REV, PHB, PHV, and ZeHB-10, −11, and −12, which have been implicated in leaf dorsoventral polarity acquisition and/or vascular development because of their mutant phenotypes and/or expression profiles. Other HD-Zip genes related to vascular development on the basis of functional and/or expression analyses are members of the larger and more closely related families I (Athb20, CHB6, Vahox1, ZeHB3) and II (Oshox1), both of which also include genes that have been linked to stress responses (e.g. Söderman et al., 1996; Deng et al., 2002).

The Athb8 gene of Arabidopsis does not seem to affect leaf polarity like other HD-Zip III genes, but appears to have a more specific role in vascular development. Athb8 is expressed early in procambial development and is inducible by external auxin application (Baima et al., 1995; Kang & Dengler, 2002, 2004; Mattsson et al., 2003; Scarpella et al., 2004). Although Athb8 is not essential for vascular tissue development, in that loss-of-function mutants do not display any phenotype, the proposed function of Athb8 as a positive regulator of vascular development is supported by the fact that ectopic overexpression of the gene leads to increased and premature formation of xylem tissue during primary and secondary vascular development (Baima et al., 2001).

The onset of the expression of the rice HD-Zip II gene Oshox1 marks a stage in procambium development at which cell fate has been specified but not stably determined towards vascular differentiation (Scarpella et al., 2000). Ectopic overexpression of Oshox1 induces vascular tissues to differentiate closer to the apical meristems and reduces the sensitivity of PAT to the PATI NPA (Scarpella et al., 2000). Furthermore, in the absence of any overt phenotypic change, Oshox1 overexpression specifically reduces the affinity of the NBP towards NPA, and enhances PAT and its sensitivity towards auxin (Scarpella et al., 2002). These results are consistent with the hypothesis that Oshox1 promotes fate commitment in procambial cells by increasing their auxin conductivity properties and stabilising this state against modulations of PAT by an endogenous NPA-like molecule.

X. Conclusions and perspectives

The amazingly complex anatomy and physiology of vascular tissues has long intrigued both descriptive and experimental biologists, and the mechanisms underlying their beautifully diversified organisation in different organs of different plant species has been, and will still continue to be, a challenge for developmental and molecular biologists. Although vascular pattern and its ontogeny have been the focus of many excellent classical studies in the past, more recent work has increasingly pointed out the need for a better understanding of the anatomically elusive earliest stages of procambial cell development, including their definition at the cellular and molecular level, the spatial signals that induce and maintain them, the earliest responses to those signals, and the stability of these states. A pivotal role in the control of these events has long been suspected for auxin. More recent work suggests that this plant hormone may genuinely function as an intercellular signal mediating the integrated polarisation of plant cells and the differentiation of vascular strands along routes of elevated auxin concentrations. This concept does not imply that auxin is sufficient for these processes, nor does it explain how the positions of auxin sources and sinks are themselves regulated and how the highly reproducible vascular patterns of monocot leaves are ensured. However, it reconciles plasticity and continuity of vascular tissues in the variable vascular patterns of dicot leaves, and may direct the routes of vascular strands outside of leaves in both dicots and monocots. At present, most conclusions are obviously still based on indirect evidence, largely because auxin remains basically invisible at cellular resolution. However, the molecular details of its action are now becoming experimentally tractable because of rapid advances in various fields. The molecular identity of recently discovered genes, their correlated mutant phenotypes, and new experimental approaches have together generated a conceptual framework for auxin action in vascular patterning, in which model predictions can now be tested at the molecular level.

As internal structures with inherent variability, vascular tissues have not been particularly amenable to mutant surveys. Nevertheless, the study of mutants affecting embryo and postembryonic organ development has generated an emerging picture in which vascular tissues are patterned by apical-basal and adaxial-abaxial/central-peripheral organismal polarity cues. Obvious subsequent steps will include determining the molecular and cellular details of both these organismal polarities, and how their pathways are integrated in the regulation of vascular tissue patterning. However, certain levels of vascular tissue organisation, such as reproducible leaf vascular patterns, cannot be easily integrated in concepts of organismal polarities.

Historically, the formation of the intricate network of vascular strands in leaves has been a focus of interest in vascular research. Previously established mechanisms integrating aligned cell differentiation during vascular strand formation were probably not reinvented in the evolution of leaves, but more likely recruited and revised by leaf specific controls. As this presumably reflects some selective pressure on precise vein positioning, leaf vascular patterns are not merely interpretations of organismal polarities, but are probably subject to additional, perhaps even entirely unknown, controls. Although the model dicot plant Arabidopsis is undoubtedly a powerful genetic system to investigate the molecular mechanisms regulating different aspects of vascular tissue patterning, the highly reproducible features of vascular organisation in monocot leaves should make these species suitable candidates for the genetic dissection of leaf vascular pattern formation.

The body pattern of a plant is clearly more complex than the combination of two axes. However, three-dimensional patterning activities could use the axial architecture of the vascular system as a positional reference. It has long been known that numerous local cell patterning events are correlated with the positions of vascular strands, suggesting that these strands serve as reference systems in local cell patterning (Nelson & Dengler, 1997). Details have already become apparent of the molecular mechanisms through which vascular tissues polarly supply positional information in ground tissue patterning and root meristem initiation (reviewed in Berleth & Chatfield, 2002; Berleth & Scarpella, 2004). Therefore, also in other organs, vascular strands could represent an anchoring scaffold for adjacent cell patterning events that would ensure the functional integration of vascular tissues into the surrounding nonvascular cell layers.

The preliminary nature of the interpretations presented herein should be emphasised. Stronger molecular groundwork is needed before these interpretations can be considered more than constructive working hypotheses. Novel molecular models are expected for the near future and should be evaluated by their ability to explain the outlined developmental correlations. The worldwide commitments to whole-genome sequencing in many plant species, together with large-scale gene-trap enterprises, genome-size expression profiling, and the generation of tagged mutant collections and misexpression resources will in the near future greatly expand our knowledge of vascular development at the molecular level in species other than Arabidopsis. In an optimistic scenario, combinations of all these new tools could soon provide a detailed concept of how local signals can induce continuous vascular differentiation, which could later become integrated into a molecular understanding of tissue patterning in each organ of different plant species.

Note added at the revision stage

After this manuscript was submitted, the molecular identity of the CVP2 gene, and the nature of the avb1 lesion were reported (Carland & Nelson, 2004; Zhong & Ye, 2004, respectively).

Acknowledgements

The authors would like to thank Nancy Dengler and Julie Kang for kindly providing microscopy slides, Taylor Steeves for kindly providing pictures, Thomas Berleth for invaluable discussion during the preparation of this manuscript, Timothy Nelson for kindly allowing the citation of a personal communication, Nancy Dengler, Julie Kang, and Wenzi Ckurshumova for helpful suggestions on the manuscript, and Naden Krogan for precious, critical help during the preparation and revision of this manuscript. We would also like to thank Oxford University Press, The Company of Biologists, and the American Society of Plant Biologists for kindly allowing reproduction of published images.

Ancillary