Leaves depend on highly developed venation systems to collect fixed carbon for transport and to distribute water. We hypothesized that local regulation of auxin biosynthesis plays a role in vein development. To this effect, we assessed the role of the SHORT INTERNODES/STYLISH (SHI/STY) gene family, zinc-finger transcription factors linked to regulation of auxin biosynthesis, in Arabidopsis thaliana leaf vein development.
Gene functions were assessed by a combination of high-resolution spatio-temporal expression analysis of promoter-marker lines and phenotypic analysis of plants homozygous for single and multiple mutant combinations.
The SHI/STY genes showed expression patterns with variations on a common theme of activity in incipient and developing cotyledon and leaf primordia, narrowing to apices and hydathode regions. Mutant analysis of single to quintuple mutant combinations revealed dose-dependent defects in vein patterning affecting multiple vein traits, most notably in cotyledons.
Here we demonstrate that local regulation of auxin biosynthesis is an important aspect of leaf vein development. Our findings also support a model in which auxin synthesized at the periphery of primordia affects vein development.
Plant leaves require highly developed venation systems to collect products of photosynthesis for transport and also to distribute water throughout the leaf blade. In the model plant Arabidopsis thaliana (hereafter referred to as Arabidopsis), a procambial midvein appears as soon as a leaf primordium emerges (Mansfield & Briarty, 1991; Mattsson et al., 1999) followed by several pairs of secondary veins and a highly variable pattern of tertiary and quaternary veinlets as the primordia grow (Goldberg et al., 1994; Donelly et al., 1999; Mattsson et al., 1999).
Auxin serves as an inducer of vascular cell fate and differentiation (reviewed in Fukuda, 2004). As such, auxin and active polar auxin transport (PAT) appear to regulate the patterns and extent of vascular bundles and veins in plants. Based on numerous experiments testing the interactions between applied auxin, endogenous auxin and tissue polarity on the pattern of trans-differentiation of stem parenchyma into vessel elements, Sachs (1981, 1989) proposed the ‘canalization of signal flow’ hypothesis as a mechanism for vascular patterning. This hypothesis states that initially diffuse distributions of auxin are drained by pre-existing vascular tissues, establishing preferred canals of flow that are gradually narrowed by a positive feedback mechanism. This creates specialized files of cells that transport auxin, drain surrounding cells of inductive auxin, and ultimately differentiate into vascular strands. There are now multiple levels of support for this model, especially in the development of leaf veins in Arabidopsis. For example, pharmacological inhibition of PAT in developing seedlings leads to enhanced leaf venation and defective connections between vessel elements, in line with reduced canalization of auxin flow (Mattsson et al., 1999; Sieburth, 1999). The changes in vein distribution are preceded by similar changes in expression patterns of the auxin response marker DR5::GUS (DR5), providing visual evidence for correlations between auxin distribution and responses with vascular differentiation (Mattsson et al., 2003). In addition, gradual refinement of both the expression and subcellular protein localization of the auxin efflux carrier PIN-FORMED1 (PIN1) into sites of vein formation provides visual support of the canalization of signal flow hypothesis (Reinhardt et al., 2003; Scarpella et al., 2006; Wenzel et al., 2007) while DR5 expression has been observed to mark dynamic sites of auxin maxima at the developing leaf margins (Aloni et al., 2003; Mattsson et al., 2003; Hay et al., 2006; Bilsborough et al., 2011).
Local auxin biosynthesis may also contribute to leaf vein development. The partially functionally redundant YUCCA (YUC) gene family of 11 members encodes flavin monooxygenases that carry out the final rate-limiting step of Trp-dependent biosynthesis of the main auxin, indole-3-acetic acid (IAA; Zhao et al., 2001; Cheng et al., 2006, 2007; Mashiguchi et al., 2011; Won et al., 2011). Three YUC members have unique expression patterns in young leaves and combinations of multiple yuc mutants result in plants with severe leaf venation defects and reduced auxin levels (Cheng et al., 2006, 2007).
STYLISH1 (STY1) and SHORT INTERNODES (SHI), of the SHORT INTERNODES/STYLISH (SHI/STY) protein family, have been found to positively regulate the transcription of YUC4 (Sohlberg et al., 2006; Eklund et al., 2010; E. Sundberg et al., unpublished data), and shi/sty mutants are hypersensitive to PAT inhibition just like the yuc4 mutant (Ståldal et al., 2008). The SHI/STY family, with its conserved 43-amino acid RING-like zinc finger domain and a more C-terminal, unique IGGH domain, appear to contain highly redundant members acting during organ development in a dose-dependent manner (Kuusk et al., 2002, 2006; phylogeny described in Kuusk et al., 2006 and Eklund et al., 2010). In addition, significantly reduced levels of auxin precursors and conjugates could only be detected in a mutant line carrying T-DNA insertions in no less than five of the SHI/STY genes (Sohlberg et al., 2006). It is therefore likely that at least the five most well-studied family members are direct activators of auxin-related developmental processes.
The roles of the SHI/STY family members as transcription factors and regulators of auxin biosynthesis imply that these members may influence several aspects of plant patterning and growth. In this paper, we seek to identify the functional roles of SHI/STY genes in leaf vascular development.
Materials and Methods
Plant material and growth
Arabidopsis thaliana (L.) Heynh. seed lines homozygous for the promoter::GUS fusions STY1::GUS, STY2::GUS, SHI::GUS, SRS5::GUS, sty1-1, sty2-1, shi-3 and srs5-1 single mutants, and double, triple and quintuple homozygous mutants were generated previously (Fridborg et al., 2001; Kuusk et al., 2002, 2006). All promoter::GUS fusions are in the Columbia (Col) background except for SHI::GUS (Landsberg erecta; Ler). sty1-1 and sty2-1 are in the Col ecotype background; shi-3 is in the Wassilewskija (Ws) background; lrp1 is in the Nossen background, and; srs5-1 is in the Ler background. Seeds were sterilized in gas chambers (Clough & Bent, 1998), plated on plant agar media (Mattsson et al., 2003) at a density of < 20 per Petri dish, stratified at 5°C for 2 d and germinated at 22°C under white fluorescent light (170–180 μmol m−2 s−1), for 16 h light and 8 h darkness at 50% RH. NPA and 2,4-D was added to medium as described in Mattsson et al. (2003). For experiments exploring the effects of exogenous auxin on gene expression, seedlings at 5 d after germination (DAG) were transferred to liquid media (Mattsson et al., 2003) containing 2,4-D and placed on an illuminated shaker for 16 h at RT before GUS analysis.
Gene expression patterns were observed in the first or second leaf once per day from emergence of the primordia to leaf maturity. Leaf primordia for the first, second and later-formed rosette leaves were dissected from representative seedlings in sterile water each day from the moment they were visible (3 DAG for the first leaf) until the later-formed leaves were mature (14 DAG). For the study of mature leaf venation patterns, wild-type and mutant cotyledons were dissected at 14 DAG and first/second and third leaves at 21 DAG, all in sterile water.
GUS assay and analysis
Leaf primordia and embryos were incubated in GUS substrate solution as described (Mattsson et al., 2003). Air was evacuated from leaf air spaces, followed by incubation at 37°C for 1–24 h, depending on colour development. Younger leaf primordia experienced a longer staining period in order to observe the faint initiation of expression. The samples were then cleared and mounted (Mattsson et al., 2003) and observed using differential interference contrast settings on a Nikon Eclipse E600 microscope. Photographs were taken using a Canon EOS 5D Mark II digital camera (Tokyo, Japan), then assembled using Adobe Photoshop.
Venation pattern analysis
First/second leaves, third leaves and cotyledons were fixed at 4°C overnight in 100% ethanol : acetic acid (6 : 1, v/v). They were then washed once in 100% ethanol and again in 70% (v/v) ethanol, followed by clearing in 50% (w/v) chloral hydrate at 65°C for 1 h. After being rinsed twice in water, all leaves were cleared again in 85% (w/v) lactic acid at room temperature for at least 3 d. All leaves were mounted in lactic acid and observed using dark field microscopy. All leaf sizes were measured and all venation features were counted using ImageJ (Schneider et al., 2012). Mean values and SEM were calculated and compared by multiple Student's t-tests using JMP v8.0.2 (SAS Institute Inc., Cary, North Carolina, USA).
SHI/STY genes are expressed in incipient cotyledon primordia and apices of developing cotyledons
We mapped the expression patterns of the four SHI/STY family members SHI, STY1, STY2 and SHI-RELATED SEQUENCE5 (SRS5) by the use of previously described promoter::GUS constructs (Fridborg et al., 2001; Kuusk et al., 2002, 2006). Embryos were first dissected from ovules then assayed for GUS expression.
We found that SHI is expressed already in the globular stage, primarily in the apical and central tiers of the embryo proper (Fig. 1a,b). In the triangular embryo, SHI is expressed primarily in the apical tier, at sites of cotyledon primordia (Fig. 1c), which is the exclusive site of expression in heart-stage embryos (Fig. 1d,e). As the cotyledons develop, SHI expression is gradually confined to the apex of cotyledons (Fig. 1f,g). STY1 expression emerges slightly later than SHI at the late globular stage (Fig. 1i); no STY1::GUS expression could be detected after > 24 h of staining at the early globular stage (Fig. 1h). At the onset, STY1 expression also appears to be more restricted than SHI, being limited to cells at the sites of incipient cotyledon primordia (Fig. 1i). Like SHI, STY1 expression is gradually focused on cotyledon apices (Fig. 1k–n) but also shows some expression along the cotyledon margin (Fig. 1m,n). STY2 expression appears only at the triangular stage, again localized to sites of emerging cotyledon primordia (Fig. 1q). From heart-stage embryo and onwards, STY2 expression is confined to the apices of cotyledon primordia (Fig. 1r–u). SRS5 expression does not begin until very late in embryo development when the cotyledons are almost mature. We observed SRS5 expression at a few cells at the apical end of the cotyledon (Fig. 1aa). In summary, SHI, STY1 and STY2 are all expressed in areas of incipient cotyledon formation (black arrows, Fig. 1a–c,i,j,q), and are later restricted to various degrees to the apices of cotyledon primordia as they develop. The SHI/STY expression in cotyledon apices of torpedo- and later-stage embryos overlaps with the activity of DR5::GUS (Fig. 1af,ag), a commonly used marker for detection of auxin response maxima (Ulmasov et al., 1997). Earlier work, using the more sensitive DR5rev::GFP marker, showed that the overlapping activity between auxin response and SHI/STY expression starts early in incipient cotyledon primordia of triangular-stage embryos (Benkova et al., 2003). Still, the apical expression of SHI and STY1 appears to precede that of DR5 (Fig. 1a,b,i). However, the DR5::GUS expression in the incipient root apical meristem (Fig. 1ac,ad,ae) and in procambial tissues (Fig. 1ag) does not overlap with SHI/STY acitivity indicating differences in their transcriptional regulation.
SHI/STY genes are also expressed in incipient leaf primordia, leaf apices and incipient hydathodes
We also mapped the expression of the same four SHI/STY reporter lines in developing rosette leaves. SHI expression was detected at the apical portion of the first leaf primordium, but only in the earliest stages observed, 3–5 d after germination (DAG; Fig. 2a–c). Expression was also detected later in forming hydathodes (Fig. 2e,f). STY1 expression was observed at the apices of leaf primordia from 3 to 7 DAG (Fig. 2g–k). Differently from SHI, STY1 expression was also detected along the margins of 5 DAG primordia (Fig. 2i), and later also along the lower margins and at the base of primordia (Fig. 2j–l). STY2 displayed apical expression throughout leaf primordia development (Fig. 2m–r). Interestingly, STY2 also appears to be expressed internally at the site of the incipient midvein from 3 to 5 DAG (Fig. 2m–o). STY2 expression was also detected in the hydathodes and trichomes near leaf maturity (Fig. 2r). SRS5 expression was not observed until c. 6 DAG through to leaf maturity, and in the apical portion and hydathodes of the leaf blade (Figs 2w,x, 4j). We also observed similar expression patterns for these genes during the development of the third leaf (Supporting Information Fig. S1).
In summary, we observed similar expression patterns in cotyledon and leaf development, with expression in incipient primordia followed by expression localized to the apices of primordia, and the forming hydathodes of rosette leaves. In addition, STY1 also had unique expression at the base of leaf primordia (Figs 1m, 2j–l).
Chemical auxin transport inhibition results in unfocused expression of SHI/STY genes in rosette leaves
Based on evidence for positive feedback mechanisms in auxin signalling (see 'Introduction'), we thought it possible that SHI/STY gene expression might be influenced by auxin distribution and levels. It has been shown that the distribution of DR5 expression is altered in leaf primordia developing in the presence of chemical auxin transport inhibitors, and that it is ectopically expressed in leaf primordia upon auxin treatment (Mattsson et al., 2003). We therefore used DR5::GUS as a positive control when carrying out similar treatments of seeds harbouring SHI/STY promoter::GUS fusions. First, we mapped the expression patterns of DR5 and our SHI/STY reporter constructs in the first pair of rosette leaves after seedlings were grown on media containing the auxin transport inhibitor, 1-N-naphthylphtalamic acid (NPA). Young leaves (5 DAG) of SHI::GUS, STY2::GUS, and SRS5::GUS show broadened domains of expression at the apical leaf margin on 1 and 10 μM NPA (0.05 and 0.1 μM NPA for SHI due to an apparent NPA hypersensitivity of this reporter construct in the Ler ecotype background) compared to those grown on control media (DMSO; Fig. 3d–l). SHI even shows ectopic expression at the basal leaf margins. Similar results for these reporter lines are also seen in more mature leaves (10 DAG; Fig. 4d–l), suggesting that the relationships between local auxin levels and SHI/STY family genes are not limited to an early time window in leaf development. Altered expression of SHI, STY2 and SRS5 (Figs 3d–l, 4d–l) coincides with broadened DR5 expression in the same position of the leaf with auxin transport inhibition (Figs 3a–c, 4a–c; Mattsson et al., 2003).
Unlike the other SHI/STY members studied, the expression of STY1 at the leaf apical tip at 5 DAG diminishes at 10 μM NPA (Fig. 3m–o). Instead, at 1 and 10 μM NPA, increased STY1 expression is observed at the basal leaf margins (Fig. 3n,o) as well as at 10 DAG (Fig. 4n,o). This is interesting because the expression of STY1 does not overlap largely with that of DR5 under control conditions, except for the apical tips of young leaf primordia (Fig. 3a,m) and the hydathodes of mature leaves (Fig. 4a,m), but STY1 nevertheless displays broadened regions of expression upon NPA treatment. Most STY1 expression in leaf primordia approaching maturity is at the basal margins (Fig. 2k,l). Our results suggest that auxin transport inhibition results in a relative increase of local free auxin at the apical end of the leaf, as well as a relative increase of local STY1 expression, with very little overlap of the separate domains.
In summary, we have found evidence that alteration of endogenous auxin distribution patterns also alters the distribution of SHI/STY gene expression patterns, and with exception of STY1, the changes are similar to those of the DR5 auxin response marker.
Exogenous auxin application lead to ectopic expression of SHI/STY genes in leaf primordia
Since altered auxin distribution also alters the distribution of SHI, STY1, STY2 and SRS5 expression as shown above, we expected the application of exogenous auxin to whole primordia to result in enlarged domains of SHI, STY1, STY2 and SRS5 promoter–reporter expression. Indeed, the domain of SHI expression expanded with 2,4-D treatment as with NPA treatment (Fig. 3d–f), from the apical end of the leaf primordium down along the apical and basal margins (Fig. 5c,d). STY1 expression with 2,4-D treatment also appeared somewhat similar to that of primordia treated with NPA (Fig. 3m–o), with increased expression at the basal leaf margins (Fig. 5e,f). But unlike treatment with NPA, expression persisted at the apical end of the primordia of STY1 seedlings, and low levels of ectopic expression could also be seen in the rest of the primordia (Fig. 5e,f). With 2,4-D treatment, STY2 was expressed ectopically in domains not previously observed under other growth conditions, at the basal margins (Fig. 5g,h). These results support the notion that increased local auxin can upregulate local SHI/STY expression, but also suggest that only certain domains of cells within the leaf primordia are competent to respond to this local upregulation by auxin.
shi/sty cotyledons have defective venation patterns
Because the expression of the four SHI/STY genes in developing cotyledon primordia overlaps with auxin response maxima implicated in venation patterning (Aloni et al., 2003; Mattsson et al., 2003; Hay et al., 2006; Bilsborough et al., 2011), we assessed mutants in these genes (Fridborg et al., 2001; Kuusk et al., 2002, 2006) for defective cotyledon venation patterns. The vascular architecture of wild-type cotyledons consists of one midvein (1°) and four secondary (2°) vein loops (Fig. 6a, Table S1). Secondary vein loops typically connect the midvein to another 2° loop without interruption. However, the basal secondary veins frequently lack more or less of their basal ends (Figs 6a, S2a), a phenomenon we have called lower-loop domain (LLD) deletions in accordance with developmental nomenclature by Scarpella et al. (2006). This type of defect correlates with a larger arch of the corresponding apical secondary vein (Fig. 6a).
We began characterizing shi/sty mutants by counting the number of secondary veins in cotyledons, regardless whether they were complete or not. All single shi/sty mutant and sty1-1 shi-3 double mutant cotyledons have the same number of 2° vein loops as their respective ecotype wild-types. However, sty1-1 sty2-1 double mutant, sty1-1 sty2-1 shi-3 triple mutant, and sty1-1 sty2-1 shi-3 lrp1 srs5-1 quintuple mutant cotyledons (LRP1, the SHI/STY member LATERAL ROOT PRIMORDIUM1; Smith & Fedoroff, 1995; Kuusk et al., 2006) have significantly fewer 2° veins loops compared to all ecotype wild-types (Figs 6k,l, 7a). As a consequence, the number of areoles – areas of the leaf blade completely surrounded by vascular tissues – are also reduced (Table S1). There is evidence that each secondary vein is formed as two fragments: one upper-loop domain (ULD) and one LLD. The patterning of each fragment is directed from an auxin convergence point at the primordium margin (Scarpella et al., 2006), resulting in the approximate vein domains indicated in Fig. 6(b). While partial or complete LLD deletions are not uncommon (Fig. S2a), ULD deletions, partial or complete, are rare in wild-type cotyledons (Fig. 7b). In the shi/sty mutants, the frequency of ULD deletions is elevated, reaching significantly higher levels in sty1-1 sty2-1 double, triple, and quintuple mutant combinations (Fig. 7b; see Fig. 6k,l e.g.). This is indicative of a dose-response dependence of ULD formation on SHI/STY activity.
We observed similar dose-response dependence for two other defect types in shi/sty mutants. The midvein runs fairly straight in wild-type cotyledons, ending close to the cotyledon apex (Fig. 6a–c). In shi/sty mutants and mutant combinations, the midvein frequently appears to have bifurcated or skewed to one side some distance before the apex, and it is no longer clear what is the apical end of the midvein (Fig. 6i–l). Bifurcated/skewed midveins occur in double or higher mutant cotyledons at frequencies significantly different from the wild-types (Fig. 7c).
shi/sty mutant leaves frequently have an ectopic vein fragment at the cotyledon apex, which is not necessarily an extension of the midvein. We refer to this vein fragment as the distal peg (Figs 6e,f, i–l, 7d, 8). It is rare in wild-type cotyledons, appearing in 0–2% of samples depending on the ecotype, but it occurs frequently in the cotyledons of sty1-1 and sty2-1 (36% and 22%, respectively) single mutants, and are found at the highest frequency, 84%, in sty1-1 shi-3 double mutants (Fig. 7d). We wanted to determine whether this defect was caused by either a displacement of secondary vein connections to the midvein towards the leaf base (distance (x) in Fig. 8a), or an extension of the midvein closer to the cotyledon apex (distance (y) in Fig. 8a). We measured these distances in genotypes with the most consistent distal peg phenotypes (sty1-1 shi-3, sty1-1 sty2-1, triple mutant; Fig. 7d) and their corresponding ecotype wild-types (Col, Ws). In the absence of a distal peg, such as in wild-type cotyledons, both distances (x) and (y) were equal. We found that distance (x) is the same for both ecotype wildtypes as well as for sty1-1 shi-3, but significantly larger for sty1-1 sty2-1 and the triple mutant (Fig. 8b), suggesting that secondary vein connections are basally displaced. Distance (y) is the same for both ecotypes and sty1-1 sty2-1 (Fig. 8c). This distance is significantly larger for the triple mutant indicating that, even with the distal peg phenotype, the vascular system of this phenotype does not sufficiently cover the apical portion of the cotyledon blade (see representative image in Fig. 8a). However, distance (y) is significantly smaller for sty1-1 shi-3 than in wildtype cotyledons (Fig. 8c). In other words, this specific mutant combination results in an extension of the vascular system towards the apical end of the leaf (Fig. 8a). In summary, we found that the two options (x) and (y) were not mutually exclusive, and the tendency was that the apical ends of secondary veins as well as apical end of the distal peg were displaced towards the basal end of leaves with increasing shi/sty loss of function.
We also scored shi/sty mutant cotyledons for additional venation criteria as described by Steynen & Schultz (2003). The scores show that cotyledons of some shi/sty single and multiple mutants display elevated frequencies of free vein ends and vascular islands (short satellite fragments of differentiated vascular tissue), which are markers of increased venation disconnect (Table S1).
shi/sty mutants also have defects in rosette leaf venation patterns
The venation of the first pair of true rosette leaves is more complex than that of cotyledons, consisting of a midvein with three 2° vein loops per half of the leaf blade in the wild-type (Figs 9a–c, 10a). Rosette leaves also contain small tertiary (3°) and quaternary (4°) veinlets, which form various closed or open connections (Fig. 9a). In addition to the above criteria for cotyledons, we also scored for marginal free ends in the first leaves, which are those found nearest the leaf blade edge. Marginal free end counts do not include distal pegs, if they are present (see later).
Unlike mutant cotyledons (Fig. 6a), shi/sty rosette leaves do not show a significant reduction in the number of 2° vein loops (Fig. 10a), There are, however, abnormal numbers of other rosette leaf venation markers in some shi/sty mutant combinations, indicating an impact of these genes also on true leaf vein patterns. The triple and quintuple mutants show significantly higher numbers of marginal free ends (Table S2); the triple mutants also show a significantly higher number of total free ends. ULD deletions occur in the first leaves of all wild-type and shi/sty mutants, and although there are differences between mutants, there are no significant differences between mutants and corresponding wild-type ecotype backgrounds (Fig. 10b). Bifurcated/skewed midveins are also seen in mutant rosette leaves, but again not at a frequency significantly different from the wild-types (Fig. 10c). While rare in wild-type first leaves, the distal peg is found frequently in sty1-1, sty2-1 and all multiple mutants (Figs 9, 10d). In summary, the first rosette leaves of shi/sty single and multiple mutant combinations displayed several of the phenotypic defects observed in cotyledons, but at reduced penetrance.
Statistical analyses were not performed for higher order leaves, but our observations suggest that triple and quintuple shi/sty mutant third leaves may also have more marginal free ends compared to corresponding wild-type leaves and that distal pegs and bifurcated/skewed midveins are not uncommon in higher order leaves as well (Fig. S3).
Shi/sty mutant vessel gaps are usually companioned by a corresponding gap in other vascular tissues
The dark-field images of shi/sty mutant cotyledons above visualize defects in xylem vessel continuity. To address whether or not defects are reflected also at the level of phloem or even procambium continuity, we assessed ULD gaps in the most apical pair of secondary veins in cotyledons of the shi/sty mutant combinations using differential interference contrast (DIC) optics. Fig. 11(a) shows a gap in ULD vessel continuity in which we could not detect any underlying elongated and interconnected cells typical of phloem or procambium extending from vein ends. Out of 47 assessed gaps, 44 had no underlying elongated cells spanning the gap (Fig. S4). Of these, two gaps had short extensions of elongated cells, but they did not span the gaps (illustrated in Fig. 11b). Three out of the 47 vessel gaps had underlying elongated cells spanning the gap (two shown in Fig. 11c,d). These were also the three shortest gaps recorded (Fig. S4). The absence of underlying phloem or procambium cells in the majority of gaps indicates that procambium never formed in these regions during vein patterning. At a lower frequency and in shorter gaps, the interruptions appear to be due to defective differentiation of vessel elements. In addition to the frequency of ULD gaps, there is also a tendency of increasing length of gaps with increasing number of combined shi/sty mutants, with the short gaps found in single and double mutants, and the longest gaps found in triple and quintuple mutants (Fig. S4), indicating a SHI/STY dosage dependence also for this trait.
SHI/STY gene expression relative to the sites of vein formation
A comparison of the sites of SHI/STY gene expression and the vein defects observed in corresponding mutants indicate that there may be a limited overlap between them. While the first anatomical evidence of veins appear in the form of elongated and interconnected cells, the selection of preprocambial cells occurs earlier from a population of isodiametric ground cells. This process is perhaps best visualized by the expression and gradual subcellular polarization of the PIN1 auxin efflux carrier, which has been done at high resolution in embryos (Benkova et al., 2003) and in the primordia of the first leaf (Scarpella et al., 2006; Wenzel et al., 2007). The most conspicuous venation defects in the shi/sty mutant combinations were those of cotyledons, that is the bifurcation of midvein, the distal peg of the midvein, the reduced number of secondary veins and the ULD deletions. Based on PIN1 localization, selection of preprocambial cells for the future midvein is initiated in the triangular stage embryo, proceeding into the heart-stage embryo (Benkova et al., 2003). At the triangular and heart stages, the expression of STY2 is high in a limited few protodermal cells at the apices of cotyledon primordia, but is low or undetectable in the regions of the future midveins (Fig. 12a,b). Cotyledon lamina formation begins in the early torpedo-stage embryo, companioned by formation of the first pair of secondary veins. At this stage, we see a partial overlap of STY2 expression with the domain of secondary vein formation but no overlap with the region in which the formation of another pair of secondary veins takes place (Fig. 12c). STY1 expression typically shows a similar pattern of no or limited overlap with the domains of preprocambial selection (Fig. 12d,e), although there is considerable variation between embryos (compare to Fig. 1k,l). SHI expression in the triangular embryo overlaps largely if not entirely with the region of preprocambial midvein selection, but less so with preprocambial secondary vein selection (Fig. 12f,g).
In the primordia of the first true leaf, cells for the first pair of preprocambial secondary veins are selected at a stage just before the lamina acquires a narrower base or a waist based on PIN expression (Scarpella et al., 2006; Wenzel et al., 2007). At this stage, the expression of STY2 overlaps in part with the domains of the first pair of preprocambial secondary veins (Fig. 12h). STY1, with its expression along the margin, and SHI with expression at the apex only do not appear to overlap with the sites of preprocambial secondary vein formation (Fig. 12h–j).
Because the visualized GUS expression can vary depending on the length of the histochemical assay and the leaching of the blue product, it is in most cases not possible to make any absolute statements whether or not SHI/STY gene expression overlaps with sites of vein formation. A couple of important points can be made though. First SHI/STY genes are expressed in protodermal or epidermal cells (Fig. 12). These cells do not normally develop into vascular cells. Second, the peak of expression is usually away from the sites of vein formation. Taken together, these points strongly suggest that SHI/STY genes, at least in part, act on vein formation noncell-autonomously and indirectly in the process of vein formation (see the 'Discussion' section).
The SHI/STY gene family members have been previously implicated primarily in the development of the flower gynoecium (Kuusk et al., 2002, 2006; Sohlberg et al., 2006). Here we show evidence, based on expression patterns and mutant phenotypes, that these genes also have functions in the embryo, cotyledon and leaf.
One would expect that genes affecting leaf vein development are expressed at sites that coincide with sites of vein formation. For example, PIN1 and the auxin response factor (ARF) MONOPTEROS (MP) as well as VASCULAR NETWORK3/SCARFACE (VAN3/SFC) are expressed early in leaf development, overlapping spatially with future vein locations while mutations in these genes result in abnormal leaf venation patterns (Hardtke & Berleth, 1998; Mattsson et al., 1999; Deyholos et al., 2000; Koizumi et al., 2000; Scarpella et al., 2006; Wenzel et al., 2007; Naramoto et al., 2009). However, SHI/STY genes are expressed at the leaf margins (Figs 1, 2), yet the shi/sty mutants display leaf venation anomalies (Figs 6, 9). Interestingly, recent evidence suggests that YUC family genes of auxin biosynthesis (Zhao et al., 2001) are also expressed primarily at the leaf margins (Wang et al., 2011) while combinations of multiple yuc mutants show reduced leaf venation (Cheng et al., 2006, 2007). Thus, both SHI/STY expression patterns and mutant phenotypes overlap with that of YUC genes and mutants. Given also that STY1 directly activates YUC4 and YUC8 transcription (Sohlberg et al., 2006; Eklund et al., 2010; Ståldal et al., 2012) and that shi/sty multiple mutants have reduced auxin levels and induced expression of STY1 results in elevated auxin biosynthesis rates (Sohlberg et al., 2006; Ståldal et al., 2008), our results suggest potential interactions between several SHI/STY and YUC gene family members in domains of overlapping expression in developing leaf primordia. For example, SHI, STY1 and STY2 expression at the apical ends of leaf primordia (Figs 1, 2) overlap with YUC4 expression in the same domains (Wang et al., 2011). A similar case can be made for STY1 and YUC1, which are both expressed at basal regions of the leaf margins (Figs 1, 2; Wang et al., 2011).
SHI/STY family genes themselves may also be regulated by auxin, as suggested by unfocused or ectopic expression after alteration of endogenous and exogenous auxin levels (Figs 3–5). Each SHI/STY family gene also contains one or more auxin response elements (AuxREs; Ulmasov et al., 1997) in their promoter sequences (I. Cierlik, et al., unpublished data), supporting potential contribution of local auxin to regulation of these genes. In addition, previous studies have reported that STY1 activation is mediated by DORNRÖSCHEN (DRN), DRN-LIKE (DRNL) and PUCHI (Eklund et al., 2011) of the APETALA2/ETHYLENE RESPONSE FACTOR (AP2/ERF) family of transcription factors (Nakano et al., 2006). Both DRN and DRNL contribute to embryonic development in a synergistic manner and are expressed at sites of incipient cotyledon formation (Chandler et al., 2011) similar to what we observe for several SHI/STY genes. Evidence also suggests that the activities of DRN are regulated by auxin signalling via direct targeting by MP (Cole et al., 2009).
There are several general conclusions that can be drawn from the series of defects in shi/sty cotyledon and leaf venation patterns (Figs 6, 9). First, defects can be found in both the pattern (i.e. the overall distribution of veins) and in the continuity of veins (e.g. secondary vein deletions), indicating that SHI/STY genes act on both the formation of veins as well as the subsequent differentiation of veins. Second, as both cotyledon and rosette leaf venation patterns are defective, the assessed SHI/STY genes act during both embryonic and post-embryonic development, in line with their expression patterns. Third, the frequency of leaf venation defects typically increases when multiple mutants are combined. In light of overlapping expression patterns, the observed dosage effect suggests that many of these genes have at least partly overlapping functions. As there are four additional active members in the SHI/STY gene family (Fridborg et al., 2001; Kuusk et al., 2002, 2006), it is also possible that different or more extensive mutant combinations would result in stronger phenotypes than the ones we have observed on account of functional redundancy among SHI/STY members. Fourth, although mutant phenotypes provide evidence for genetic redundancy, the expression patterns indicate that individual genes may have partially unique functions. For example, we see a temporal gradient for the onset of gene expression during embryo development and, to limited extent, also during development of true leaves. Fifth, because gaps in vessels are typically companioned by corresponding gaps in other vascular tissues (Fig. 11), it stands to reason that formation of procambium never occurred in these regions, and that SHI/STY genes influence the selection of preprocambial cells. This notion is further supported by the aberrant course of veins also found in shi/sty mutant cotyledons and leaves. Smaller interruptions in vessel continuity were bridged by phloem tissues, indicating that the role of the SHI/STY genes in vein formation is not limited to procambium formation but also extends throughout vein differentiation. Finally, as the expression of the assessed SHI/STY genes is primarily in the apices and margins of cotyledon and leaf primordia, and the phenotypic defects occur in vascular tissues (Fig. 12) it appears that the SHI/STY genes somehow act remotely in the process of leaf vein patterning and differentiation, as we propose below.
It is known that some transcription factors can act noncell-autonomously via transport through plasmodesmata (reviewed in Xu & Jackson, 2012) and therefore it is possible that the SHI/STY transcription factors act in a similar fashion. However, known instances of noncell-autonomous action of transcription factors, such as KNOTTED1 and SHORTROOT, occur only over short distances of a few cells (reviewed in Xu & Jackson, 2012). While formation of mid and secondary veins takes place in minute primordia containing few cells, the formation of minor venation such as the marginal tertiary vein loops occur some distance from SHI/STY maxima of expression, arguing against transport of SHI/STY proteins via plasmodesmata. It is also possible that the SHI/STY promoter::GUS fusions are missing regulatory elements that confer expression in regions of vein formation. The STY1 reporter expression domain shows a perfect overlap with RNA in situ hybridization domains during early embryo development (Kuusk et al., 2002), which argues against this option. Because auxin is known to act noncell-autonomously in several plant developmental processes (reviewed in Zhao, 2010), we present instead a model in which SHI/STY proteins act cell-autonomously in the activation of YUC genes at the leaf margins, triggering local auxin production. Synthesized auxin, in turn, acts noncell-autonomously in the formation of internal veins. The expression of the SHI/STY genes overlaps in part with the expression conferred by the auxin response marker DR5 during cotyledon (Benkova et al., 2003; Fig. 1) and leaf development (Aloni et al., 2003; Mattsson et al., 2003; Hay et al., 2006; Bilsborough et al., 2011), consistent with auxin accumulation at sites of SHI/STY expression. We also show here that SHI/STY gene expression is altered upon auxin transport inhibitor induced alterations of auxin distribution (Figs 3, 4), and that SHI/STY gene expression domains expand and become stronger upon auxin treatment of leaf primordia (Fig. 5), indicating that SHI/STY expression in itself may be influenced by auxin distribution. There is evidence that PIN1, an auxin efflux carrier implicated in generating auxin maxima at sites that overlap largely with the expression of SHI/STY genes is part of a positive feedback mechanism generating auxin maxima by local PAT (Benkova et al., 2003; Reinhardt et al., 2003; Heisler et al., 2005; Scarpella et al., 2006; Wenzel et al., 2007). We hypothesize that SHI/STY genes may be activated at these maxima, and in turn activate YUC genes, resulting in the subsequent formation of relatively stable auxin biosynthesis maxima at the cotyledon and leaf apices and hydathodes.
We offer our sincere gratitude to Roni Aloni of Tel Aviv University for advice on leaf clearing techniques and our reviewers for their constructive criticism. This research was funded by a National Science and Engineering Research Council of Canada discovery grant to J.M. and a Swedish Research Council Formas grant to E.S.