WOX2 and STIMPY-LIKE/WOX8 promote cotyledon boundary formation in Arabidopsis


(e-mail: xuelinwu@usc.edu).


One of the key events in dicot plant embryogenesis is the emergence of the two cotyledon primordia, which marks the transition from radial symmetry to bilateral symmetry. In Arabidopsis thaliana, the three CUP-SHAPED COTYLEDON (CUC) genes are responsible for determining the boundary region between the cotyledons. However, the mechanisms controlling their transcription activation are not well understood. Previous studies found that several WOX family homeobox transcription factors are involved in embryo apical patterning and cotyledon development. Here we show that WOX2 and STIMPY-LIKE (STPL/WOX8) act redundantly to differentially regulate the expression of the CUC genes in promoting the establishment of the cotyledon boundary, without affecting the primary shoot meristem. Loss of both WOX2 and STPL results in reduced CUC2 and CUC3 expression in one side of the embryo, but an expansion of the CUC1 domain. Furthermore, we found that STPL is expressed in the embryo proper, and its activation is enhanced by the removal of WOX2, providing an explanation for the functional redundancy between WOX2 and STPL. Additional evidence also showed that WOX2 and STPL function independently in regulating different aspects of local auxin gradient formation during early embryogenesis.


In higher plants, embryogenesis fulfills the most basic needs for generating a plant body that is functional for post-embryonic development: setting up the apical–basal and radial patterns, and setting aside the stem cells in the primary shoot and root meristems, which are the sources of post-embryonic organ formation. In dicots, part of the apical patterning process is the transition from radial symmetry to bilateral symmetry, which can be first observed with the formation of the two cotyledon primordia. The successful development of the two cotyledons requires the coordination of two major events: tissue outgrowth from the cotyledon primordia and boundary formation, which prevents growth between the two cotyledons. In addition to physically separating the two cotyledons, the cotyledon boundary is essential for the formation of the primary shoot meristem (Souer et al., 1996; Aida et al., 1997, 1999).

In the past two decades, molecular and genetic studies have identified a group of NAC domain transcription factors as essential regulators of organ boundary formation in higher plants (Olsen et al., 2005). In Arabidopis, the three CUP-SHAPED COTYLEDON genes, CUC1, CUC2 and CUC3, are expressed in overlapping stripes between the two cotyledon primordia, and act redundantly to specify the cotyledon boundary (Aida et al., 1997, 1999; Takada et al., 2001; Vroemen et al., 2003). Genetically, the CUC genes interact with the auxin pathway, and disruptions in auxin transport and signaling result in changes in CUC1 and CUC2 expression patterns and enhance the cotyledon fusion phenotype of the cuc mutants (Aida et al., 2002; Furutani et al., 2004; Kajiwara et al., 2004).

Several genes have been found to regulate the transcription of the CUC genes. During embryogenesis, SHOOTMERISTEMLESS (STM) is required for the specification of the primary shoot meristem and cotyledon separation (Barton and Poethig, 1993; Long et al., 1996), a phenotype that can be partially explained by the mutual dependence between STM and CUC1, 2 (Aida et al., 1999; Kwon et al., 2006). More recently, it has been demonstrated that STM directly activates the transcription of CUC1 (Spinelli et al., 2011). Other positive regulators of CUC gene transcription include the two closely related ATPases in the SWI/SNF chromatin-remodeling complex, AtBRM and SYD, although they target different members of the CUC family (Kwon et al., 2006). In addition, the CUC gene activities are attenuated by the miR164 family microRNA through transcript degradation (Laufs et al., 2004; Mallory et al., 2004).

In Arabidopsis, the homeodomain transcription factor WUSCHEL (WUS; Laux et al., 1996) and several of the WUSHEL-RELATED HOMEOBOX (WOX; Haecker et al., 2004) genes are involved in different aspects of embryonic development. For example, WUS is essential for the establishment of the stem cell cluster in the shoot meristem (Laux et al., 1996; Mayer et al., 1998), and WOX5 has been shown to be important for the establishment of the quiescent center near the root meristem (Gonzali et al., 2005; Scheres, 2005). In addition, WOX2, STIMPY (STIP/WOX9; Wu et al., 2005) and STIMPY-LIKE (STPL/WOX8; Wu et al., 2007) regulate embryonic cell division (Haecker et al., 2004; Wu et al., 2007; Breuninger et al., 2008). Among these genes, WOX2 appears to function redundantly with a number of other WOX genes, including STPL, in cotyledon development. Although the loss of WOX2 leads to a very mild cotyledon defect, 20–30% of the wox2 wox8 double mutants display different degrees of cotyledon fusion phenotype (Figure 1; Wu et al., 2007; Breuninger et al., 2008). However, the cause of this cotyledon separation defect remains unknown. Furthermore, inconsistent with its function in the embryo proper, it has been reported that STPL expression is limited to the suspensor region (Haecker et al., 2004). Here, we further investigate the mechanisms behind the roles of WOX2 and STPL in cotyledon development. We show that they promote CUC2 and CUC3 expression and the establishment of the cotyledon boundary. In addition, we present evidence that STPL expression pattern is altered by the loss of WOX2 activities, which explains the observed functional redundancy.

Figure 1.

wox2 stpl double mutant seedlings display a range of cotyledon phenotypes.
Six-day-old wox2 stpl seedlings can have two normal cotyledons, similar to the wild type (a), two cotyledons of unequal size (b), two cotyledons positioned at an angle (c), heart-shaped cotyledons (d) and single (e) cotyledons. Other than the cotyledons and the first pair of true leaves, the plants in (b–e) develop normally.


Loss of WOX2 and STPL lead to the development of partially fused cotyledons

As reported previously, a small number of the wox2 mutant embryos display defects in cotyledon initiation, a phenotype that is enhanced by the loss of other WOX genes, including STPL (Table 1; Wu et al., 2007; Breuninger et al., 2008). A closer examination revealed that four classes of cotyledon phenotypes can be observed in a population of wox2 stpl seedlings. Nearly 80% of the seedlings appear wild-type (Figure 1a), and the remaining 20% have a range of cotyledon defects: one of the cotyledons significantly reduced (Figure 1b); two cotyledons were positioned at an angle, instead of opposite each other (Figure 1c); or, in most cases, only a single cotyledon was produced (Figure 1d,e). It is worth noting that in all cases the function of the shoot apical meristem was not affected.

Table 1. The frequency of cotyledon defects observed in different genotypes
Genotype n Class IClass IIClass IIIClass IVClass V
  1. Class I, wild-type looking seedlings; class II, seedlings with two cotyledons of unequal size; class III, seedlings that have cotyledons positioned at an angle, instead of opposite one another; class IV, seedlings with either a heart-shaped or a single cotyledon, representing the partial fusion of the two cotyledon primordia; class V, seedlings with a cup-shaped cotyledon.

stpl 289100.0%0.0%0.0%0.0%0.0%
wox2 11794.0%0.9%0.9%4.3%0.0%
stpl wox2 35679.5%1.4%3.4%15.7%0.0%
cuc1-1 28494.0%4.9%0.0%1.1%0.0%
cuc1-1 stpl 17996.6%2.2%1.1%0.0%0.0%
cuc1-1 wox2 22476.3%3.6%5.8%14.3%0.0%
cuc1-1 wox2 stpl (Col-like)31339.3%1.0%10.9%48.2%0.6%
cuc1-1 wox2 stpl (Ler-like)27958.4%1.8%8.6%31.2%0.0%

As the cotyledon fusion phenotype of the wox2 stpl mutants is not fully penetrant, we carried out complementation experiments by reintroducing the wild-type copy of the WOX2 and STPL genomic regions into the wox2 stpl double mutants. Of the 21 transgenic lines that carried the wild-type WOX2 genomic region, 18 were able to fully abolish cotyledon fusion, and three others showed reductions in the percentage of seedlings with fused cotyledons. In comparison, the wild-type copy of STPL resulted in a wider range of effect. A total of 25 transgenic lines were analyzed. Five were able to either completely rescue the cotyledon defect or reduce the level of fusion to that of the wox2 single mutants. An additional 11 lines showed partial rescue of the double mutant phenotype. These results confirmed that the observed cotyledon defects were caused by the loss of WOX2 and STPL, and that their different abilities to rescue the wox2 stpl double mutants may be attributed to the lesser role of STPL in the WOX2-STPL redundancy.

To further understand the cause of the observed cotyledon defects, we compared the seedling cotyledon vascular patterns of the wild type, wox2 and stpl single mutants, and the wox2 stpl double mutants. In wild-type seedlings, the cotyledon vasculature forms a typical pattern of between two and four closed loops around a single main vein (Figure 2a). Although the majority of the wox2 mutants display no cotyledon defect, the vasculature is often discontinuous, with multiple gaps in the side loops (Figure 2b), a phenotype shared by some of the stpl seedlings (Figure 2d). Of the wox2 seedlings with a single cotyledon, the majority of them have two main veins within the cotyledon (Figure 2c), indicating that it resulted from a partial fusion of the two cotyledons. The same two types of vascular defects are also observed in the wox2 stpl double mutants (Figure 2e,f), suggesting that the partially fused cotyledons seen in wox2 stpl are not caused by the failure to initiate two cotyledon primordia, but instead result from the partial fusion of the two cotyledons during development.

Figure 2.

WOX2 and STPL regulate cotyledon vasculature development.
(a) In 6-day-old Col-0 seedlings, the cotyledon vasculature contains between two and four closed side loops connected to one main vein. The vasculature of wox2 (b), stpl (d) and wox2 stpl (e) mutants often appear to be discontinuous. Two main veins (arrowheads) can be found in wox2 (c) and wox2 stpl (f) seedlings with a single cotyledon, indicating that it is the result of the partial fusion of the two cotyledon primordia. Scale bar: 100 μm.

wox2 stpl double mutants have cotyledon boundary defect

The discontinuous cotyledon vasculature seen in the wox2 and stpl mutants indicate a possible interruption in the auxin pathway during embryonic development (Sachs, 1981, 1991; Przemeck et al., 1996; Cheng et al., 2006; Scarpella et al., 2006). As localized auxin gradients, which result from both local biosynthesis and polar transport, play a prominent role in cotyledon primordia initiation and growth (Hardtke and Berleth, 1998; Steinmann et al., 1999; Friml et al., 2003; Geldner et al., 2003; Cheng et al., 2007b, 2008; Ikeda et al., 2009), this raised the possibility that the cotyledon phenotypes described above are caused by defects in auxin localization. In addition, although the vascular phenotype in Figure 2 indicates that two cotyledon primordia fuse to give rise to the single cotyledon in wox2 and wox2 stpl mutants, it does not exclude the possibility that this phenotype is caused by a failure to initiate one of the cotyledons (Nodine et al., 2007; Nodine and Tax, 2008). The position of the cotyledon primordia can be identified based on a pair of auxin response maxima using the DR5rev::GFP reporter (Friml et al., 2003). We therefore asked whether auxin localization is disrupted in embryos lacking WOX2, STPL or both.

Starting at the early heart stage, wild-type embryos display three auxin response maxima, as visualized by DR5rev::GFP: one at the base of the embryo and the top of the suspensor, and two at the tips of the emerging cotyledon primordia (Figure 3a,e; Friml et al., 2003). Although stpl embryos and seedlings are phenotypically normal, the majority of the stpl embryos show strong GFP fluorescence in the suspensor, in addition to the wild-type DR5 pattern (Figure 3b). This suspensor signal persists at least through the late heart to torpedo stages, indicating a defect in properly establishing the auxin apical–basal gradient (Friml et al., 2003). However, no disruption was found in the DR5::GFP patterns in the embryo proper throughout the development of the stpl embryos. The wox2 embryos, on the other hand, showed a normal DR5:GFP pattern (Figure 3c). Two types of disruptions in DR5 expression were observed in a minority of the wox2 stpl embryos: persistent GFP signal in the suspensor (Figure 3f) and additional GFP foci at the apical end of the embryo (Figure 3d). However, there is no correlation between these two phenotypes, suggesting that they are the result of disruptions in different processes. At the heart stage, the majority of the wox2 stpl embryos, including the ones showing a clear fusion between the two cotyledon primordia, have three correctly positioned auxin response maxima (Figure 3g), indicating that the two cotyledon primordia are correctly initiated in these wox2 stpl mutants. This finding strongly supports the conclusion that the cotyledon fusion phenotype of the wox2 stpl seedlings is not caused by auxin mis-localization, but is the result of a failure to prevent tissue growth at one side of the cotyledon boundary.

Figure 3.

 DR5::GFP expression in wox2 and stpl single and double mutants.
In wild-type embryos at the heart stage (a) or later (e), three DR5rev::GFP foci mark the two emerging cotyledon primordia and the future root meristem region. In addition to the wild-type expression, stpl mutants show strong GFP signal in the suspensor cells (b). Most of the wox2 mutants display a normal GFP pattern (c). Although a small number of wox2 stpl embryos show ectopic DR5 activities in the embryo (d), the majority of them have the normal DR5 expression pattern in the embryo proper (f, g). Note that the embryo in (g) is at the torpedo stage, with cotyledon fusion. Scale bar: 20 μm.

WOX2 and STPL promotes symmetry in CUC2 and CUC3 expression domains

The loss of a proper cotyledon boundary in wox2 stpl mutants raises the possibility that the expression of the CUC genes may be compromised in these embryos. However, when we compared the phenotype of the wox2 stpl mutant with any of the cuc single mutants, it is clear that the observed cotyledon partial fusions in wox2 stpl cannot be accounted for by the loss of a single CUC gene due to the high degrees of redundancies among the three CUC genes in setting the cotyledon boundary (Table 1; Aida et al., 1997; Takada et al., 2001; Vroemen et al., 2003). Therefore, we examined the embryonic expression patterns of all three CUC genes, starting with CUC2 and CUC3, because their single mutants show a low penetrance of cotyledon fusion phenotypes.

In early globular stage wild-type embryos, CUC2 mRNA is mostly found in the upper inner cells (Figure 4a). This pattern is further refined by the mid-globular stage, after the inner cells complete their longitudinal division. At this time point, the CUC2 signals are concentrated into each of the inner cell daughter cells that flank the central axis of the embryo, in the future shoot meristem region (Figure 4b). By the late globular stage, CUC2 is expressed in a narrow strip of cells just above the O’ line, spanning the shoot meristem region and between the two future cotyledon primordia (Figure 4c). This expression pattern persists till much later in embryogenesis (Figure 4d; Aida et al., 1999). It is important to note that, regardless of the developmental stage, CUC2 expression is consistently symmetric on both side of the central axis of a wild-type embryo. Two classes of CUC2 patterns can be found in wox2 stpl embryos. Whereas some of them show wild-type CUC2 expression, in the other embryos CUC2 is found only on one side of the central axis, sometimes in both daughter cells of a single upper inner cell at the mid-globular stage (Figure 4e). As these embryos further develop, CUC2 expression remains at one side of the embryo (Figure 4f,g), suggesting that WOX2 and STPL2 maintain the symmetry in CUC2 expression domain.

Figure 4.

CUC2 and CUC3 mRNA patterns in Col-0 and wox2 stpl embryos.
(a–d) CUC2 expression in wild-type embryos: (a) frontal section of an early globular embryo; (b) frontal section of a mid-globular embryo; (c) sagittal section of a late globular embryo; and (d) sagittal section of a heart-stage embryo.
(e–g) CUC2 expression in sagittal sections of: (e) mid-globular stage; (f) late-globular stage; and (g) heart-stage wox2 stpl mutant embryos. Compared with (c, d), all three embryos show CUC2 expression limited to one side of the embryonic central axis. In all three cases, consecutive sections of the same embryo were examined to confirm that the asymmetry in CUC2 expression is not a sectioning artifact. Arrowheads mark the central midline of the embryos.
(h–i) CUC3 expression in an eight-cell (a) and a late-globular stage (i) wild-type embryo.
(j) CUC3 expression in a sagittal section of a heart-stage wox2 stpl embryo, showing the asymmetric expression pattern. The arrowhead marks the central midline of the embryo. Scale bar: 20 μm.

Compared with CUC2, CUC3 expression is activated earlier in a slightly different pattern (Vroemen et al., 2003). In eight-cell wild-type embryos, CUC3 can be detected in the apical cells (Figure 4h). As the embryos further develop, this expression becomes concentrated in the protodermal cells, in a stripe between the two future cotyledon promordia (Figure 4i). In wox2 stpl embryos, CUC3 expression is also detected in the protoderm. However, in many cases, it is again only detected in one side of the embryo (Figure 4j), indicating that WOX2 and STPL play similar roles in promoting CUC3 expression as for CUC2. The loss of symmetry in CUC2 and CUC3 domains is consistent with the hypothesis that the wox2 stpl seedling phenotype is caused by defects in cotyledon boundary formation.

CUC1 embryonic expression domain expands in wox2 stpl mutants

Unlike CUC2 and CUC3, CUC1 expression is not reduced in response to the loss of WOX2 and STPL. In wild-type embryos, CUC1 can be first detected at the globular stage in two domains below the protoderm layer, near the top center of the embryo (Figure 5a,b). By heart stage, it is expressed between the two emerging cotyledon primordia, still mostly excluded from the protodermal cells (Figure 5c; Takada et al., 2001). In wox2 stpl embryos, we observed no significant reduction in CUC1 expression levels or domain size. On the contrary, approximately 20% of the samples showed an expansion of CUC1 expression into the protodermal layer (Figure 5d), indicating that WOX2 and STPL play a role in restricting the CUC1 to the inner cell layers.

Figure 5.

CUC1 expression expands in wox2 stpl embryos.
(a–b) CUC1 mRNA patterns in: (a) a frontal section and (b) a sagittal section of late-globular stage Col-0 embryos. (c) CUC1 mRNA patterns in a frontal section of a triangle-stage Col-0 embryo. In all three cases, no signal is detected in the protoderm.
(d) Frontal section of a wox2 stpl embryo, showing CUC1 expression in the protodermal cells.
(e) The loss of CUC1 significantly increases the frequency of partially fused cotyledons in both wox2 and wox2 stpl mutants. Scale bar: 20 μm.

Considering the largely protodermal expression of CUC3 (Figure 4h,i) and the redundant roles it plays with CUC1 in specifying the cotyledon boundary (Vroemen et al., 2003), we postulated that the expansion of CUC1 expression into the protoderm may compensate for the partial loss of CUC3 and account for the relatively low penetrance of the cotyledon fusion phenotype in the wox2 stpl mutants. To test this hypothesis, we generated mutant combinations among cuc1, wox2 and stpl plants, to examine their cotyledon morphology (Table 1). Consistent with our previous findings (Wu et al., 2007), stpl single mutants do not show any developmental defects. Less than 5% of wox2 seedlings show cotyledon partial fusion, and an additional 2% develop milder cotyledon size and position defects. Among the few cuc1-1 seedlings that display cotyledon defects, the majority has one cotyledon significantly smaller than the other, and only 1%cuc1-1 single mutants have cotyledon partial fusion. Although no phenotypic enhancement was found between cuc1-1 and stpl, the loss of CUC1 enhances the wox2 phenotype so that one-quarter of the cuc1-1 wox2 seedlings showed cotyledon defects. Two classes of phenotypes, both significantly stronger than either cuc1-1 wox2 or wox2 stpl, were observed in the cuc1-1 wox2 stpl triple mutant seedlings; however, the severity of the phenotype depends on whether the erecta (er) mutation from the Ler ecotype was present (Figure 5e; Table 1). Without the er mutation, nearly 50% of the cuc1-1 wox2 stpl seedlings developed partially fused cotyledons, and another 11% have the two cotyledons positioned at an angle instead of opposite one another. In rare cases, seedlings with cup-shaped cotyledons were also observed. The presence of er caused a reduction in the number of seedlings with partially fused cotyledons to approximately 30%, indicating a genetic interaction between ER and the three genes in this study. This interaction may be caused by changes in auxin signaling, as some of the YUCCA family auxin biosynthesis genes are expressed in the domains overlapping that of the CUC genes during the global stage, and the overexpression of YUC5 can suppress the er mutant phenotype during post-embryonic development (Woodward et al., 2005; Cheng et al., 2007a).

STPL has a dynamic expression pattern in the embryo proper

We found that STPL and WOX2 redundantly regulate the expression of the CUC genes at the cotyledon boundary. However, it has been reported that although WOX2 mRNA was detected in the apical region of the embryo, STPL expression was only found in the suspensor cells and the endosperm cells surrounding the globular and heart-stage embryos (Haecker et al., 2004). This raised the question of how the STPL protein made in the suspensor cells could function in the apical region of the embryo. Based on the reported expression patterns, one could imagine that STPL protein acts non-cell autonomously, as previously proposed by Breuninger et al. (2008), by either trafficking to the top of the embryo or activating a mobile signal that is sent from the suspensor to the cotyledon boundary region. The alternative explanation is that STPL is actually expressed in the embryo proper either at very low levels or only when WOX2 activity is lacking. To distinguish between these two scenarios, we re-examined the expression patterns of STPL and WOX2 in the wild type, and in stpl and wox2 single mutant embryos.

Consistent with the earlier report, we found WOX2 mRNA most concentrated above the O’ line in wild-type embryos, and no change was detected in the stpl embryo. However, a broader expression pattern was detected for STPL in the wild-type embryos. Starting from the one- to two-cell stage to the globular stage, weak STPL expression can be detected throughout the developing embryo, although a stronger STPL signal is found in the suspensor cells (Figure 6a–c). In globular stage embryos, STPL expression begins to be activated in the endosperm cells surrounding the embryos (Figure 6c). By the heart stage, the highest STPL expression is found in the endosperm cells and some in the suspensor cells, whereas the embryo proper has little or no STPL expression. Curiously, we consistently detected an STPL signal on the surface of the embryos (Figure 6d). In addition to its normal wild-type expression pattern, STPL is also activated in the L1 layer in late globular to early heart stage wox2 mutant embryos (Figure 6e,f), suggesting that WOX2 normally prevents STPL activation in these cells. Both the timing and the position of this STPL expression are consistent with its redundant role in promoting cotyledon boundary establishment with WOX2.

Figure 6.

STPL expression in Col-0 and wox2 embryos.
(a–d) STPL mRNA in (a) one- to two-cell, (b) eight-cell, (c) globular and (d) heart-stage Col-0 embryos. Endosperm expression is activated during the globular stage (c).
(e, f) STPL mRNA in globular (e) and triangle (f) stage wox2 mutant embryos. Both show STPL expression in the protoderm layer, in addition to its normal expression domains. Scale bar: 20 μm.


For dicot embryos, the emergence of the two cotyledon primordia marks the critical transition from radial to bilateral symmetry. Two interacting pathways have been shown to play the major roles in this process. While local auxin concentration gradients and signaling determine the position and out-growth of the cotyledon primordia (Hardtke and Berleth, 1998; Steinmann et al., 1999; Friml et al., 2003; Cheng et al., 2007b, 2008), the three CUC genes prevent tissue growth at the cotyledon boundary and allow the formation of the primary shoot meristem. We show that WOX2 and STPL regulate both pathways, however, at different stages of embryonic development. The cotyledon fusions caused by the loss of WOX2 and STPL mainly resulted from the disruptions in CUC2 and CUC3 expression.

WOX2 and STPL differentially regulate CUC gene expression at the cotyledon boundary

The three CUC genes expressed between the cotyledon primordia play redundant but essential roles, not only in the formation of the cotyledon boundary, but also in ensuring the proper formation of the primary shoot meristem. Existing evidence suggests that their expression is controlled by different mechanisms, and few transcriptional regulators have been identified (Aida et al., 1999, 2002; Laufs et al., 2004; Mallory et al., 2004; Kwon et al., 2006; Nikovics et al., 2006; Spinelli et al., 2011). Our study showed that WOX2 and STPL positively regulate the transcription of CUC2 and CUC3 (Figure 4). However, the much broader WOX2 and STPL expression domains and the localized changes in CUC2 and CUC3 expression patterns in wox2 stpl embryos argue against a simple activation model. Two plausible scenarios could be used to explain the roles of WOX2 and STPL in activating CUC2 and CUC3 expression: either they serve as co-factors to another activator that is localized at the cotyledon boundary to ensure the robustness of CUC2 and CUC3 expression, or CUC2 and CUC3 expression may be activated in a broader region but are restricted to the cotyledon boundary region by repressors in the emerging cotyledon primordia. Although the gradual refinement of the early CUC2 and CUC3 patterns (Figure 4a–d,h,i) argues for the latter hypothesis, the putative repressors remain to be identified. In the case of CUC1, we found that WOX2 and STPL have a role in preventing CUC1 activation in the prodermal cells at the cotyledon boundary (Figure 5), although we cannot rule out the possibility that its expansion into the L1 layer is caused by the loss of CUC3 in the same region instead of repression by WOX2 and STPL. It is also worth noting that we have observed normal CUC expression patterns in wox2 stpl embryos with clear division plane defects, indicating that the changes in CUC gene expression patterns are not likely to result from the early aberrant cell division patterns caused by the loss of WOX2 (Haecker et al., 2004). Further studies are needed to uncover the precise mechanisms by which WOX2 and STPL modulate CUC gene expression during embryogenesis.

WOX2 and STPL regulate different aspects of auxin gradient formation in embryonic development

As shown in Table 1, the functional redundancy between WOX2 and STPL in cotyledon boundary formation is unequal, with the role of WOX2 being more prominent. A different picture emerges in the case of auxin transport during early embryogenesis. Previous studies have indicated that STPL acts redundantly with WOX2 and STIP in regulating auxin polar transport, possibly by modulating the expression of the auxin efflux transporter (Wu et al., 2007; Breuninger et al., 2008). In our current study, we found that the loss of STPL alone results in strong ectopic DR5 activities in the suspensor (Figure 3b). This phenotype is consistent with the high levels of STPL expression detected in the suspensor cells. As ectopic DR5 signal in the suspensor may also result from treatment with an auxin eflux inhibitor (Friml et al., 2003), this finding implies that STPL is responsible for preventing auxin accumulation in the lower suspensor cells during embryonic development. Although STPL has been shown to regulate PIN1 expression along with STIP (Breuninger et al., 2008), the lack of any detectable changes in DR5 expression in the stpl embryo proper argues against the model that PIN1 is the primary target of STPL in the suspensor. Instead, considering the exclusive suspensor expression of PIN7 (Friml et al., 2003), it is far more likely that STPL positively regulates the expression or activity of PIN7, which in turn transports auxin out of the lower cells in the suspensor.

In comparison, we found normal DR5 expression in wox2 single mutant embryos (Figure 3c), although mutation in wox2 enhances the embryonic phenotype of mp (Breuninger et al., 2008). The only indication of possible disruptions in the auxin pathway in wox2 embryos came as gaps in cotyledon vasculature (Figure 2b), indicating that the role of WOX2 in the establishment of local auxin gradients in the embryo appears to be either rather minor or highly redundant with other genes. Furthermore, the largely combinatorial effect of WOX2 and STPL on both DR5::GFP distribution and the cotyledon vasculature patterns (Figures 2 and 3) suggests that they regulate different aspects of auxin localization during embryogenesis, with little overlap.

STPL is expressed in the embryo proper

During early embryogenesis, STPL functions redundantly not only with WOX2 in regulating cotyledon boundary formation, but also with STIP in promoting cell division in the embryo (Wu et al., 2007). Previous reports found that STPL is only expressed in the suspensor and the endosperm cells, which led to the hypothesis that it acts non-cell autonomously in regulating the development of the embryo proper (Haecker et al., 2004; Breuninger et al., 2008). Our finding of weak STPL expression in the embryo proper (Figure 6a–c) argues against the non-autonomous model. This discrepancy with the previous report may be the result of improved fixation and tissue treatment processes over the last decade. Consistent with our findings, CLE8, the newly described cell automous activator of STPL, is also expressed in both the endosperm and young embryos (Fiume and Fletcher, 2012). In addition, STPL expression is specifically activated in the protodermal cells from the globular to the heart stages of wox2 mutant embryos (Figure 6e,f). This expression provides an explanation for the functional redundancy between WOX2 and STPL in regulating the cotyledon boundary, and indicates that WOX2 acts as a repressor of STPL expression in the L1 layer at this stage. This result, taken together with the earlier finding that WOX2 acts downstream of STPL and STIP in specifying the apical fate during the first embryonic cell division (Breuninger et al., 2008), paints a complex picture of the genetic relationship between WOX2 and STPL.

The most curious observation here, however, is that STPL mRNA was detected on the surface of heart-stage embryos (Figure 6d,f). Although it is formally possible, we do not believe that this signal represents noise from the in situ process, because probes for all three CUC genes and WOX2 did not give rise to similar patterns in our study. The endosperm cells encasing the developing embryo, which are called either embryo surrounding region (Bandurski et al., 1992; Opsahl-Ferstad et al., 1997) or micropylar endosperm (MCE), are morphologically different from the rest of the endosperm. It is found in a wide range of plant species and has been shown to regulate seed size and embryo cuticle formation, possibly by transporting nutrients and signaling molecules, including peptides, from the endosperm to the embryo (reviewed by Cosségal et al., 2007; Yang et al., 2008). Therefore, considering the high levels of STPL expression found in the endosperm cells surrounding the globular and heart-stage embryos (Figure 6d), the most likely source of this surface STPL signal would be from endosperm expression. This, in turn, raises the question of the origin of STPL signal in the L1 layer of the wox2 embryos. Although the most common mechanism would be local activation after the removal of the WOX2 repressor, we cannot exclude the possibility that it is transferred into the protoderm from the surrounding endosperm cells. Further studies are required to distinguish these two modes of STPL activation in the L1 layer of wox2 mutant embryos.

Experimental procedures

Plant materials and growth conditions

Plants were grown under long days (16-h light/8-h dark cycles) under ∼120 μE m−2 per sec light at 22°C. To observe seedling phenotypes, seeds were germinated on half-strength Murashige Minimal Organics Medium (MS; PhytoTechnology Laboratories, http://www.phytotechlab.com) with 0.6% agar after 2 days of stratification at 4°C.

The T-DNA insertion alleles of STPL and WOX2 are in the Columbia ecotype and have been described previously (Wu et al., 2007). cuc1-1 (Aida et al., 1997) was obtained from the Arabidopsis stock center, and is in the Landsberg erecta ecotype. For phenotype analysis, all crosses were carried to F4 and F5, and seedlings were imaged using a Leica M165C stereomicroscope with a DFC295 camera (Leica, http://www.leica.com).

Plasmid construction

To express the wild-type copy of WOX2, a 7-kb genomic fragment including the 1.3-kb WOX2 coding region, 4.7-kb 5′ region, and 1-kb 3′ region, was cloned into the binary vector JHA212G (Yoo et al., 2005). For STPL genomic rescue, a 4.1-kb genomic fragment, including the 2-kb STPL coding region, 1.8-kb 5′ region and 300-bp 3′ region, was cloned into the binary vector JHA212B (Yoo et al., 2005). Both transgenes were transformed into the wox2 stpl double mutants, and their phenotypes were analyzed in the T2 generation.

In situ hybridization and morphological analyses

For in situ hybridization on tissue sections, siliques were fixed in 4% paraformaldehyde and 0.1% Tween-20 in PBS for 5 h after 1 h of vacuum infiltration, then infiltrated with paraffin using an H2850 Microwave Processor (Energy Beam Sciences, http://www.ebsciences.com), as described by Schichnes et al. (2005). In situ hybridization was performed as previously described (Schichnes et al., 1998), with the modification that tissue sections were treated with 10 μg ml−1 proteinase K (#P2308; Sigma-Aldrich, http://www.sigmaaldrich.com) at 37°C for 15 min pre-hybridization and 10 μg ml−1 RNase A at 37°C for 30 min post-hybridization. The Dig-labeled antisense probes of STPL, WOX2, CUC1, CUC2 and CUC3 were generated using in vitro transcription of respective full-length cDNAs cloned into pBluescript (Agilent Genomics, http://www.genomics.agilent.com).

To observe the cotyledon vascular patterns, 8-day-old seedlings were cleared in 30% glycerol containing 2.5 g ml−1 chloral hydrate. DR5::GFP patterns were imaged using embryos dissected from the developing seeds. All samples were photographed on a Zeiss Axio Imager equipped with an AxioCam HRc camera (Zeiss, http://www.zeiss.com).


We thank Jirl Friml for gifts of material, the Salk Institute Genomic Analysis Lab and the Arabidopsis stock center for supplying T-DNA knock-out lines and mutants, Phoebe Luong for technical assistance, and Anna Skylar and Dana Lynn for their critical reading of the article. CK was partially funded by the Genomics Research Experience for Undergraduates (GREU) program at USC. The authors declare no conflict of interest.