Dynamics of MONOPTEROS and PIN-FORMED1 expression during leaf vein pattern formation in Arabidopsis thaliana

Authors


(fax +1 604 291 3496; e-mail mattsson@sfu.ca).

Summary

Genetic evidence links the Arabidopsis MONOPTEROS (MP) and PIN-FORMED1 (PIN1) genes to the patterning of leaf veins. To elucidate their potential functions and interactions in this process, we have assessed the dynamics of MP and PIN1 expression during vascular patterning in Arabidopsis leaf primordia. Both genes undergo a dynamic process of gradual refinement of expression into files one to two cells wide before overt vascular differentiation. The subcellular distribution of PIN1 is also gradually refined from a non-polar distribution in isodiametric cells to strongly polarized in elongated procambial cells and provides an indication of overall directions of auxin flow. We found evidence that MP expression can be activated by auxin exposure and that PIN1 as well as DR5::GUS expression is defective in mp mutant leaves. Taken together the results suggest a feedback regulatory loop that involves auxin, MP and PIN1 and provide novel experimental support for the canalization-of-auxin-flow hypothesis.

Introduction

Vascular tissues differ from most other plant tissues in that cells need to be precisely connected in order for the tissue to carry out its functions. Vascular organization is particularly conspicuous in the highly ordered pattern of veins in leaves. Arabidopsis thaliana leaves develop a hierarchical reticulate venation pattern first with the formation of a central primary vein, followed by successive basipetal addition of secondary veins and finally higher-order veins (Kang and Dengler, 2002; Mattsson et al., 1999).

Numerous experiments suggest that auxin has an essential role in vascular patterning. Local application of auxin induces the formation of vascular strands (Sachs, 1981) and high levels have been detected in pre-procambial cells (Mattsson et al., 2003; Uggla et al., 1996), suggesting that auxin may act as a positional signal controlling vascular development. Pharmacological inhibition of auxin transport has a dramatic effect on vascular pattern formation (Mattsson et al., 1999; Sieburth, 1999), and mutations in auxin signaling-related MONOPTEROS (MP), BODENLOS and AUXIN-RESISTANT6 disrupt vascular continuity (Hamann et al., 1999; Hobbie et al., 2000; Przemeck et al., 1996). The incomplete vascular continuity in the mp mutant and expression of the auxin response factor MP in vascular tissues suggests an early function of MP in the establishment of vascular patterns (Hardtke and Berleth, 1998). Mutants in MP are defective in the auxin induction of several genes that may affect procambial development (Mattsson et al., 2003).

Auxin is actively transported within the plant (reviewed in Aloni, 1995). AUXIN1 (AUX1) and similar proteins are auxin influx carriers and the PIN-FORMED (PIN) family are auxin efflux carrier proteins that become localized at the poles within cells and facilitate directional auxin flow (Bennett et al., 1996; Bouttéet al., 2006; Friml, 2003; Petrášek et al., 2006; Wiśniewska et al., 2006). Several ABC proteins are also involved in auxin transport, possibly acting independently of PINs (Murphy et al., 2002; Noh et al., 2001; Petrášek et al., 2006). Polar localization of PIN carrier complexes is mediated by vesicle trafficking (Steinmann et al., 1999) and by polarity and cell fate determinants (Friml et al., 2004; Treml et al., 2005; Xu et al., 2006). Auxin seems to regulate PIN transcription, cellular trafficking and localization (Blilou et al., 2005; Leyser, 2005; Paciorek et al., 2005; Vieten et al., 2005).

Organ and vascular formation in plants may rely on auxin carrier proteins and other upstream factors driving formation of local auxin maxima (Benkova et al., 2003; Friml et al., 2003; Xu et al., 2006). In the shoot apical meristem, the subcellular localization of AUX1 and PIN1 proteins suggests that auxin flows in the epidermis towards a point where an auxin maximum is formed, followed by an internalization of flow, triggering the formation of leaf primordia and the midvein (Benkova et al., 2003; Reinhardt et al., 2003). PIN-FORMED 1 is further implicated in vascular formation since it becomes localized at the poles in vascular cells to transport auxin from the stem towards the root (Galweiler et al., 1998; Vieten et al., 2005), and pin1 mutants show increased vascularization similar to the effects seen with polar auxin transport inhibition (Mattsson et al., 1999).

Two models have been proposed for describing the selection of cells for vascular strand formation. The auxin canalization hypothesis proposes a positive-feedback mechanism whereby an initial broad region of auxin is actively channeled into a file of cells that eventually undergo vascular differentiation (Sachs, 1981). Support for this model comes from experiments showing that chemical or genetic inhibition of auxin transport can lead to thicker veins, presumably since auxin cannot drain properly and thus forms broader canals of cells with high auxin concentrations (Mattsson et al., 1999; Sieburth, 1999). A reaction–diffusion model based on the interaction of two or more diffusing substances has also been used to describe vascular patterning (Koch and Meinhardt, 1994). Local fluxes of an activator such as auxin may trigger a positive feedback loop leading to cells with a high activator concentration undergoing vascular differentiation. Koizumi et al. (2000) suggest that the van mutants, which show discontinuous but generally normal vein patterning, support the reaction–diffusion model since the canalization model assumes a continuous generation of strands. However, two studies using computer modeling (Rolland-Lagan and Prusinkiewicz, 2005) or showing localization of PIN1 in the van3 mutant (Scarpella et al., 2006) provide support for the existence of a canalization mechanism in mutants with discontinuous vein strands.

Both MP and PIN1 are implicated in leaf vein formation since the corresponding mp and pin1 mutants have vein defects. In this study we explore the roles of MP and PIN1 in Arabidopsis leaf vein patterning by assessing their expression during this process. In the final stages of preparation of this manuscript a study was published by Scarpella et al. (2006) describing PIN1 localization in developing Arabidopsis leaf primordia that overlaps in part with this study. These results will be further discussed in context with our own results (see below).

Results

In this study, we are assessing the spatio-temporal expression patterns of two components involved in leaf vascular development in A. thalianaMP and PIN1 and their potential interaction with auxin. Hereafter, we refer to ‘MP expression’ as the level of MP mRNA transcript indicated by in situ RNA hybridization, and ‘PIN1–GFP expression or localization’ as indicated by the fusion protein PIN1–GFP in pPIN1::PIN1–GFP transformed lines. Vascular terminology is as described in Mattsson et al. (2003) and Hickey (1979), and as also shown in Figure S1. Terminology regarding positions within organs and individual cells is as defined in Friml et al. (2006). First rosette leaf primordia stages are given in days after germination (DAG).

Dynamics of MONOPTEROS (MP) expression

In this study, we show how MP expression is progressively restricted from broad regions to pre-procambial cells. In 2 DAG leaf primordia, MP is initially expressed internally throughout the lamina excluding the margin in first leaf primordia (Figure 1a,b) or at elevated levels in the incipient primary vein (Figure 1c). By 3 DAG, MP expression indicates the incipient secondary veins along the margin (arrows in Figure 1d), which eventually sharpen to form secondary veins that are one to two cells wide (Figure 1e,f). For all subsequent basipetally formed secondary veins, MP expression is first low in a large region between the margin and the existing secondary pre-procambial strand (arrows in Figure 1g,i,j), before a distinct strand appears (Figure 1h,k). The same process occurs in the proximity of a leaf serration (arrows in Figure 1j,k). MONOPTEROS expression is also gradually restricted during the formation of tertiary veins (arrowheads in Figure 1i,j).

Figure 1.

 Expression of MP in leaf primordia.
Whole-mount in situ hybridizations of MP mRNA transcript in developing primordia for the first two (a–k) or later-forming (l–p) rosette leaves. All are planar median views of the leaf primordia, except (a, l, m) which are lateral median views. Scale bars are 20 μm (a–f, l, m) or 40 μm (g–k, n–p). Arrows point at emerging veins (see text).

In the subsequent leaves, the larger final size is accompanied by a more complex venation and MP expression pattern. MONOPTEROS is expressed at the site of incipient leaf primordia becoming confined to an internal domain of expression (arrows in Figure 1l,m). Expression of MP is initially strong and widespread (Figure 1n), and can remain high and diffuse simultaneously in the first two secondary pre-procambial veins (Figure 1o). Secondary veins in later leaves can appear in a two-step process with strong and focused MP expression in pre-procambial veins connecting the midvein with the margin prior to connecting with the more distal pre-procambial veins (arrow in Figure 1o). Figure 1(p) shows how the level of MP expression is highest in pre-procambial veins without overt cellular differentiation (arrow), intermediate in procambial veins with elongated cells (arrowhead) and low or absent in fully differentiated veins (asterisk). In summary, MP expression appears to go through a process of gradual refinement from regions of low levels of expression in many cells into single files of cells with strong expression before procambial differentiation occurs.

Dynamics of PIN1–GFP expression

Expression of PIN1–GFP overlaps considerably with MP expression and also becomes restricted to pre-procambial strands. In the first leaves, by 2 DAG a two to three cell wide region of PIN1–GFP-expressing cells depicts the primary vein connecting to the hypocotyl vasculature (Figure 2a), with expression narrowing to a one to three cell wide primary vein in older leaf primordia (Figure 2b–d). A similar progressive delimitation of the primary vein strand is observed for later-forming leaves. In all rosette leaves, the first indication of secondary pre-procambial veins based on PIN1–GFP expression appears as a cluster of PIN1–GFP-expressing cells in the basal adaxial region (asterisks in Figure 2b,f,k and arrows in 2e). This region of PIN1–GFP-expressing cells is gradually restricted to form a continuous secondary strand that is one to two cells wide (Figure 2f–h,k–m). Secondary vein formation often occurs in two stages, with the strand forming between the epidermis and midvein prior to connecting to a more distal region of the midvein (Figure 2b–d) or more distal secondary strand (Figure 2g,h). This biphasic connection always occurs for the second to fourth secondary veins in later-formed rosette leaves (Figure 2i–k), but is usually more simultaneous for the fourth and fifth pairs of secondary veins in the first leaves and the fifth and subsequently formed secondary veins in later rosette leaves (data not shown). Expression of PIN1–GFP in developing higher-order veins appears as either outgrowth from existing veins or connection of isolated or clustered PIN1–GFP-expressing cells to a lower-order vein (Figure S2a,b). Expression of PIN1–GFP in all vein types is progressively lost in a basipetal direction as the more distal veins differentiate (e.g. Figure 3e).

Figure 2.

 Expression of PIN1–GFP in leaf primordia.
First (a–h) or later-formed (i–m) rosette leaves in pPIN1::PIN1–GFP transformed plants. Images show formation of the primary (a, b) and first (c, d) or second (e–h) secondary veins in the first leaves, and the second or third secondary veins in later rosette leaves (i–m). All are planar median views except (a, b) which are lateral median views. Scale bars are 20 μm. Arrows point at various aspects of PIN1–GFP expression (see text).

Figure 3.

 Expression of PIN1–GFP in epidermal tissues of primordia.
First two (a–d) or later forming (e) rosette leaves. All show images taken through the median section (lower) or adaxial epidermal surface (upper) of the same primordia. Arrows point at the subcellular localization (a, b) or high-expression region (c) of PIN1–GFP. Scale bars are 20 μm.

PIN1–GFP expression extends from the epidermis into ground tissue

Expression of PIN1–GFP suggests a role for the epidermis in directing auxin into developing vascular regions. In slightly bulging primordia, PIN1–GFP fusion proteins are localized to the apical end of all epidermal cells leading towards the primordia apex, which overlies the developing primary vein (Figures 3a,b and 4s). About the time of emergence of the first pair of secondary veins, PIN1–GFP is predominantly lost from the abaxial epidermis and becomes progressively more restricted to basal adaxial and marginal epidermal cells (Figure 3c,d and Figure S2c). Thereafter, PIN1–GFP is always localized to the basal adaxial and marginal epidermal cells that are adjacent to the newly forming secondary veins (Figures 2e, 3d,e and Figure S2c), connecting the epidermis and developing veins (e.g. Figure 4t and Figure S2e) until the procambial strand is clearly delimited (Figure 2g,h,l,m). In later-formed rosette leaves, this epidermal region includes areas just proximal and distal to a serration (Figure 3e), respectively involved in the formation of secondary (large arrows in Figure 2k–m) and tertiary marginal (Figure S2b) veins. Some of the later-formed rosette leaves (<5%) also maintain abaxial PIN1–GFP expression along much of the length of the primordia up until the formation of the fourth or fifth secondary veins (Figure S2d).

Figure 4.

 Subcellular localization of PIN1–GFP.
All are planar median views except (o–q) which show epidermal (upper) and median (lower) sections. Arrows point at the subcellular localization of PIN1–GFP and also indicate the presumed direction of auxin transport except for (d) and (l), which indicate the location of pre-procambial strands. The large arrowhead in (r) shows the ‘hotspot’ of non-polarized PIN1–GFP in the epidermis adjacent to the next site of pre-procambial strand development. Scale bars are 10 μm.

PIN1–GFP polarization during procambial development

All vein classes show polar PIN1–GFP localization once the narrow procambial strands are clearly formed. Our results, however, indicate that PIN1–GFP proteins are initially diffusely distributed throughout the cytoplasm and/or on the plasma membrane. In young primordia that show PIN1–GFP polarity on the apical ends of epidermal cells, the procambial midvein cells often appear to have PIN1–GFP expression throughout the cytoplasm (Figure 4a), which later localizes in the basal end of cells (Figure 4b,c). Developing secondary pre-procambial cells often show diffuse PIN1–GFP cytoplasmic expression (e.g. arrows in Figure 4d), which later becomes polarized (Figure 4e). The most obvious examples of early non-polarized PIN1–GFP expression are observed in later-forming rosette leaves that have broad regions of cells with diffuse PIN1–GFP expression (Figure 4j,o). In these regions, PIN1–GFP gradually becomes more localized in fewer cells, eventually forming a secondary procambial strand with highly polarized PIN1–GFP localization (Figure 4k–n). Epidermal cells adjacent to developing secondary pre-procambial cells are often observed with PIN1–GFP expression throughout their cytoplasm, which is eventually more localized to the membrane as the secondary procambial strands develop or as a leaf serration expands (Figure 4o–r,t).

Based on localization of PIN1–GFP protein we can predict the direction of auxin flow. Auxin flow in the primary vein is predominantly in the basal direction (Figure 4b,c). Auxin flow is predominantly bidirectional along secondary strands, with the distal cells having apical (Figure 4f) or basal PIN1–GFP polarity, with the more basal cells always having basal polarity where they connect to the midvein (Figure 4h), with one or more cells within the strand having bidirectional PIN1–GFP localization (Figure 4g). Less frequently, PIN1–GFP occurs only on the basal ends of procambial cells, resulting in a basal direction of auxin flow along a given secondary strand (Figure 4i). In young leaf primordia PIN1–GFP becomes localized to the apical ends of epidermal cells, presumably resulting in auxin flow to the apex before it is internalized through the primary procambial strand region (Figures 3a,b and 4s). The PIN1–GFP can also become localized at the poles in epidermal cells adjacent to developing secondary strands, indicating epidermal auxin flow towards a serration apex and then internally into the developing secondary strand (Figure 4q,r,t).

MP expression is induced by auxin and similar to auxin - response patterns

Previous experiments to assess auxin induction of transcription of MP and other auxin response factors (ARFs) have been negative (Ulmasov et al., 1999). Instead, the auxin response mediated by ARFs is believed to come from auxin-induced degradation of Aux/IAA proteins that otherwise dimerize with and inhibit the DNA binding of ARFs (Leyser, 2002). As shown in Figure 1, the expression of MP is highly regulated. If it is not regulated, at least in part, by auxin, one has to consider mechanisms other than canalization of auxin flow for the patterning of MP expression. To assess whether MP expression can be induced by auxin at a local level, we exposed 3- or 4-day-old seedlings grown in liquid medium to the synthetic auxin 2,4-dichlorophenoxyacetic acid (2,4-d) for 16 h followed by whole-mount in situ hybridization. We used the strong auxin-response marker DR5::GUS (Ulmasov et al., 1997) as a control to identify suitable conditions for auxin exposure. In 3-day-old seedlings, exposure to 1 μm 2,4-d resulted in a patchy and weak activation of DR5::GUS expression, whereas exposure to 10 μm 2,4-d resulted in a strong response throughout the lamina (Figure 5b,c). Exposure to 1 μm 2,4-d enhanced pre-procambial MP expression and also resulted in ectopic expression in ground cells between pre-procambial veins predominantly in the basal part of leaf primordia regions where new secondary veins are forming (Figure 5e compared with d). Exposure to 10 μm 2,4-d, resulted in strong MP expression throughout the lamina except for a region two to three cells wide proximal to the margin (Figure 5f). Cells in this region appear to lack not only the competence to express MP in response to auxin, but also to respond to accumulation of endogenous auxin by vascular differentiation (Mattsson et al., 1999, 2003). That cellular competence may play a role in the response of MP to auxin is further supported by exposure of 4-day-old leaf primordia to auxin (Figure S5), where the response is limited almost entirely to the actively growing basal part of the primordia. The response of MP in primordia of this age to exogenous auxin is similar to the auxin response reported for pPIN1::PIN1–GFP to topical application of auxin (Scarpella et al., 2006). We can reproduce a similar induction of PIN1 mRNA transcript by auxin exposure in liquid medium and detection by in situ hybridization (Figure S5), providing further evidence of a similar response of MP and PIN1 to auxin.

Figure 5.

 Auxin induction of MP expression.
Expression of DR5::GUS (a–c) and in situ hybridizations to MP mRNA transcript (d–f) in leaf primordia after 3 DAG seedlings had been exposed to 0 (a, d), 1 (b, e) or 10 (c, f) μm of exogenous 2,4-d. Scale bars are 20 μm.

To assess whether MP expression responds to changes in the distribution of endogenous auxin, we exposed seedlings to auxin transport inhibitors rather than auxin. This kind of treatment is known to cause major alterations in auxin distribution (Mattsson et al., 2003) and correlated changes in venation patterns (Mattsson et al., 1999; Sieburth, 1999). In young leaf primordia of seedlings germinated in the presence of 10 μmN-(1-naphthyl)phthalamic acid (NPA), MP is expressed primarily in a region two to three cells proximal and parallel to the leaf margin, with low levels of MP expression throughout the central lamina (Figure 6g,h). In more advanced primordia, there can also be pre-procambial strands of MP-expressing cells extending from the margin towards the petiole (Figure 6i). This pattern of expression in response to NPA treatment is almost identical to that observed with the auxin-sensitive DR5::GUS reporter (Mattsson et al., 2003; Figure 8g–i).

Figure 6.

MP expression in primordia from mp mutant and NPA-grown plants.
Whole mount in situ hybridizations showing MP mRNA transcript in wild-type (a–c) compared with similar stages in the mp mutant (d–f) and in response to inhibition of auxin transport with 10 μm NPA (g–i). All are planar median views. Scale bars are 50 μm.

Figure 8.

 Auxin response patterns in primordia of mp mutant and NPA-grown plants.
Expression of DR5::GUS in the wild type (a–c) compared with mp (d–f), and in response to inhibition of polar auxin transport with 10 μm NPA (g–i). All are planar median views. Scale bars are 50 μm.

Since MP encodes an auxin response factor, it is in theory possible that it regulates its own expression. To assess this possibility, we analyzed the expression of MP in primordia of mp mutants that produce a defective MP transcript. We found MP expression in young primordia indicative of pre-procambial primary veins and the first two pairs of secondary veins, which is comparable to the expression in wild-type primordia of equal size (compare Figure 6a,b with d,e). Thus it appears that the MP gene product does not feed back significantly on its own expression during formation of pre-procambial strands. The expression of MP in mp mutants is much reduced in later stages (Figure 6f), possibly due to the lack of pre-procambial cells at this stage, and a limited number of cells undergoing procambial differentiation. In summary, we have found experimental support that MP expression can be altered by manipulation of auxin distribution, and that early MP expression does not depend on activity of the MP protein.

PIN1–GFP expression is regulated by MP and altered by inhibition of auxin transport

Since the MP gene encodes a transcriptional regulator, it is possible that PIN1 depends on MP for normal expression. To assess this possibility we introduced the pPIN1::PIN1–GFP marker into the mp mutant background. In contrast to the initially normal MP expression in mp mutants, PIN1–GFP expression in mp mutants is much reduced, both quantitatively and qualitatively (Figure 7d–f). The expression of PIN1–GFP in mp mutant primordia is evident in developing primary veins and incomplete secondary veins that terminate in a diffuse PIN1–GFP-expressing region (compare Figure 7d–f and Figure S3f–i with Figure 7a–c). In mp mutants, PIN1–GFP appears to be localized in the apical ends of epidermal cells in young emerging leaf primordia (Figure S3e). In older mp primordia, PIN1–GFP expression becomes predominantly restricted to basal adaxial regions (Figure S3f) with little or no epidermal PIN1–GFP polarity (Figure S3g–i). Similar to MP, in NPA-treated tissue PIN1–GFP has diffuse lamina expression two to three cells proximal to the margin in young primordia and marginal expression with strands leading towards the petiole in older primordia, although MP expression is higher near the margin of young primordia (compare Figures 7g–i and 6g–i). Treatment with NPA predominantly inhibits the development of subcellular PIN1–GFP localization in most internal cells (Figure S4) and in the epidermis (Figure S3a–d), although PIN1–GFP expression does become restricted to the basal adaxial and marginal epidermis in older primordia (Figure S3b).

Figure 7.

 Expression of PIN1–GFP in primordia from mp mutant and NPA-grown plants.
Control plants (a–c) compared with similar developmental stages in the mp mutant (d–f) and in response to inhibition of auxin transport with 1 μm NPA (g–k). All are planar median views. Scale bars are 20 μm.

mp mutants are defective in the organization of auxin maxima

The hypothesis of canalization of auxin flow predicts a feedback regulatory loop where auxin transport alters the pattern of auxin distribution (Sachs, 1981). If MP is part of such a regulatory loop, one may expect alterations in the auxin response pattern in mp mutant leaves. The auxin response pattern of DR5::GUS expression in wild-type primordia can be briefly summarized as having maxima at the apex and lateral serrations, in addition to pre-procambial and procambial expression (Mattsson et al., 2003; Figure 8a–c). Most mp primordia have an abnormally large and diffuse apical maximum, usually no lateral maxima, diffuse pre-procambial and rarely occurring procambial DR5::GUS expression (Figure 8d–f). In wild-type primordia, NPA treatment predominantly shifts DR5::GUS expression to the margin of the primordia (Mattsson et al., 2003; Figure 8g–i). A similar distribution of DR5::GUS expression is seen in older primordia of mp and pin1 mutants, although the latter can often also form two or more apical maxima of DR5::GUS activity (Figure S6). In summary, mp mutants appear to fail to delimit and organize vascular and marginal auxin maxima, while pin1 mutants can organize extra auxin maxima.

Discussion

The expression of MP and PIN1 during venation patterning of leaves is of particular interest since mutants in both genes have vascular defects and the gene products are involved in auxin signaling and transport. Here we propose a feedback model that involves MP, PIN1 and auxin in the formation of leaf veins. A part of our study (PIN1–GFP expression in developing leaf primordia) overlaps with a recent publication (Scarpella et al., 2006). Points where we provide additional information will be discussed below.

A model for canalization of auxin flow that involves MP and PIN1 activity

Currently, two models are favored for the explanation of the patterning of veins in leaves; the canalization of auxin flow hypothesis (Sachs, 1981) and the diffusion–reaction hypothesis (reviewed by Koch and Meinhardt, 1994). Our assessment is potentially biased towards the first hypothesis in that we have actively sought to understand the role of auxin-related components in vein formation. This hypothesis predicts a positive-feedback mechanism whereby cells respond to auxin by expressing auxin efflux carriers, which eventually become polarly localized to produce discrete files of auxin-transporting cells that undergo vein differentiation (Sachs, 1981).

Although the gradual refinement of expression that we observe for both MP and PIN1–GFP is consistent with both models, our observations support the following positive-feedback cellular response pathway driving auxin canalization, as summarized in Figure 9(a). (i) In response to increased auxin levels, the MP gene becomes transcriptionally activated (Figure 5 and Figure S5). To our knowledge, this is the first experimental support for auxin-induced expression of a gene encoding an auxin response factor, although we cannot rule out indirect induction at this point. (ii) The activity of the MP protein is modulated by inhibitory Aux/IAA proteins, which can be targeted for auxin-induced degradation (Leyser, 2002; Ulmasov et al., 1999). (iii) The MP and PIN1–GFP expression patterns in developing leaf primordia overlap considerably and undergo a similar gradual restriction of expression (Figures 1 and 2). We have shown that PIN1–GFP protein is reduced quantitatively and qualitatively in the mp mutant (Figure 7), suggesting that the expression of the PIN1 gene is partially regulated by the MP protein, possibly by direct interaction of MP protein with the PIN1 promoter. There is evidence that PIN1 is at least in part activated by auxin (Paciorek et al., 2005; Scarpella et al., 2006; Figure S5) and this response may be mediated by MP. (iv) We have shown that the subcellular localization of PIN1–GFP develops gradually from no obvious polarity in isodiametric cells to strong polarity that indicates auxin flow along a source–sink vector in elongated and interconnected procambial cells (Figure 4). The polarity of PIN1 is presumably regulated by mechanisms that involve vesicle trafficking and regulatory proteins such as PINOID (e.g. Friml et al., 2004; Steinmann et al., 1999). (v) The resulting gradual concentration of auxin into certain cells would lead to stronger activation of MP, possibly both at transcriptional and post-transcriptional levels, and as a consequence a stronger expression of the PIN1 gene. (vi) Cells that are drained of auxin would react by decreasing expression of MP and PIN1. While the above scenario provides a cellular framework for a positive-feedback loop, the relatively minor venation defects in mature leaves of both pin1 and mp mutants suggest that other regulatory and carrier proteins are probably involved.

Figure 9.

 Proposed regulatory loop and module for major vein formation.
Models on (a) a cellular feedback regulatory loop that leads to vein formation and (b) a module for major vein patterning based on overlapping and gradually refined patterns of MP (blue) and PIN1 (green) expression. Arrows indicate the direction of auxin transport based on subcellular localization of PIN1 proteins. Expression of PIN1 at the margins indicates expression in adaxial and marginal cells.

A module for major vein formation

Based on the patterns of MP mRNA transcript and PIN1–GFP fusion protein expression and subcellular localization, we propose a module for secondary vein formation as an extension of what has been proposed for leaf primordia and midvein formation based on polar auxin transport (Reinhardt et al., 2003). A similar model has been proposed by Scarpella et al. (2006) based on PIN1–GFP expression, and the main difference between the models is that we include the positive-feedback regulatory loop described above. The module includes several steps beginning with the expression of PIN1 in the adaxial and marginal epidermis and MP and PIN1 expression in the basal adaxial ground tissues of leaf primordia. Epidermal expression of PIN1 may not be driven by auxin alone as there is no corresponding MP and DR5::GUS expression in the epidermis. Growth initially occurs throughout the primordia and gradually becomes restricted to basal, actively growing regions in older primordia (Kang and Dengler, 2002), which are competent to respond to auxin with respect to PIN1 and MP regulation (Figure 5e and Figure S5). The auxin produced by cells in actively growing ground tissue and epidermis triggers the mechanism of canalization of auxin flow, beginning with wide regions of cells with low levels of MP and non-polarized PIN1 proteins and ending in narrow files of cells with high levels of MP and highly polarized PIN1 proteins. The source–sink relationship between actively growing regions (sources) and pre-existing veins (sinks) triggers first the formation of the midvein, then in turn the progressive formation of secondary veins in the growing basal regions of the primordia. Secondary veins often appear to emanate from a primary source at the leaf margin where PIN1 is particularly strongly expressed. From there, elongated clouds of MP- and PIN1-expressing cells connect to the midvein (sink) and a pre-existing secondary vein (sink). These clouds of expression are gradually resolved into strands in parallel with the gradual polar localization of PIN1. Figure 9(b) summarizes this module for reiterative secondary vein formation based on MP and PIN1 expression.

Based on our results for subcellular PIN1–GFP localization, an entry point of auxin appears in conjunction with the formation of each secondary vein, through which auxin moves from the epidermis into the ground tissues. The mechanism behind the formation of these entry points is unknown. We have noticed though that in the mp mutant background, the level of PIN1–GFP expression and degree of epidermal polarization is much reduced. This is surprising as MP is not expressed in the epidermis, and one has to consider indirect effects of MP activity to explain this apparent contradiction. The strong ground tissue expression of PIN1–GFP normally seen near these points is also weak or absent in the mp mutants. One possibility is that the internal canalization driven by the MP/PIN1 feedback loop is required to form a highly focused entry point in the epidermis. This scenario can be likened to pulling the plug in a bath tub (sink), resulting in a strong and focused vortex on the surface (epidermis) bringing surface particles (auxin) to the vortex, before they are engulfed. In mp mutants, a weakened vortex forms that results in partial secondary veins close to the sink (midvein), but is not focused enough at the surface to result in the formation of distal parts of secondary veins, a typical phenotype of young mp mutant leaves. Thus MP and PIN1 appear to be important components in auxin canalization affecting leaf vein formation. Additional components are likely to be added with time.

Experimental procedures

Plant material and growth

Arabidopsis thaliana Columbia (Col)-0 containing a pPIN1::PIN1–GFP fusion construct (stock no. CS9362) was obtained from the Arabidopsis Biological Resource Centre, Ohio State University, Columbus, OH, USA. The Col-0 plants containing the auxin-responsive promoter DR5::GUS construct were obtained from Tom Guilfoyle (University of Missouri, Columbia, MO, USA; Ulmasov et al., 1997). Sterilized seeds were plated in 9 cm Petri dishes containing solid A. thaliana salts medium (ATS; Lincoln et al., 1990). For the auxin transport inhibitor treatments, seedlings were plated on media containing 0.1, 1 or 10 μm NPA (TCI, Tokyo, Japan). Seeds were stratified at 4°C for at least 2 days, and grown in a short-day chamber (8-h light/16-h dark) at approximately 20°C and 50 μE light intensity.

For the auxin induction treatment, approximately 50 DR5::GUS seeds were grown in 9 cm Petri dishes containing 10 ml of sterile ATS medium on a rotary shaker for 3 or 4 days at 20°C. At these time points 0.1% (w/v) pluronic-F68 surfactant (Sigma, St. Louis, MO, USA) and 2,4-dichlorophenoxyacetic acid (Sigma) to final concentrations of 0, 1 or 10 μm were added to the medium and seedlings were grown for about 16 h. The seedlings were then GUS stained or prepared for in situ hybridization to detect DR5::GUS or MP expression. Observations were made on at least 10 primordia for each treatment.

PIN1–GFP and DR5::GUS localization

Expression was observed in leaf primordia extracted on sequential days. Results are based on observations of at least 20 leaves taken from each developmental stage. Individual leaves or leaf primordia were extracted from seedlings and mounted in a 30% aqueous glycerol solution. In order to avoid disruption of inherent PIN1–GFP localization, we mostly base polarity observations on tissues that were examined within about 15 min of mounting in water. In some cases determination of the polarization of PIN1–GFP localization in procambial cells was facilitated with induction of plasmolysis by mounting the extracted leaves in either a 2 m NaCl or 100 μm aqueous propidium iodide (Sigma) solution. A Zeiss LSM 410 confocal laser scanning microscope (Jena, Germany) was used to image PIN1–GFP localization. Green fluorescent protein was imaged using a 488-nm excitation filter and 500–530-nm emission filter combination. Background red autofluorescence was detected using a 568-nm excitation filter and an LP 580 emission filter set. DR5::GUS assays were performed as described in Mattsson et al. (2003). Differential interference contrast images were taken on a Nikon Eclipse E600 microscope (Tokyo, Japan) using a digital camera. Adobe Photoshop software was used to merge images with slightly different focal planes.

Whole-mount in situ hybridization

Whole mount in situ hybridization was carried out as described in Zachgo et al. (2000) with several changes. In our hands, a fixation time of least 4 h, and generally overnight, had to be used to obtain the expected expression patterns of several published genes, including MP. In addition, the background signal was decreased by agitation in a fresh solution containing 0.1 m triethanolamine (pH 8) and 0.5% (v.v) acetic anhydride for 15 min, followed by two washes in 1x PBT solution prior to hybridization for 2 days at 60°C. The MP sense and antisense probes were prepared as described in Hardtke and Berleth (1998).

Acknowledgements

We thank Afsaneh Haghighi-Kia for technical assistance, Steven Chatfield, Jiri Friml and the Arabidopsis Biological Resource Centre (Ohio State University, Columbia, OH, USA) for providing the pPIN1::PIN1–GFP line. The work was supported by a National Science and Engineering Research Council of Canada grant to J.M.

Ancillary