CORONA, PHABULOSA and PHAVOLUTA collaborate with BELL1 to confine WUSCHEL expression to the nucellus in Arabidopsis ovules

Ovules are the developmental precursors of seeds. In angiosperms, the ovule consists of three developmental domains; the nucellus, chalaza and funiculus (Fig. 1A) (Balasubramanian and Schneitz, 2000). The nucellus contains a megasporangium, where the female gametophyte develops. The inner and outer integuments form from the chalaza and enclose the nucellus. The funiculus is a stalk-like structure that connects the ovule to the ovary wall. For correct development of the ovules, it is essential to establish this proximal-distal patterning; hence, a shift in the boundary between domains results in aberrant ovule morphology (e.g. Balasubramanian and Schneitz, 2000; Gross-Hardt et al., 2002).

WUSCHEL (WUS) is a homeobox gene that characterizes the nucellus by its restricted expression in this domain, and this nucellar expression is necessary for initiation of the two integuments that develop from the chalaza (Gross-Hardt et al., 2002). WUS promotes the expression of an auxin efflux facilitator, PIN-FORMED 1 (PIN1), via transcriptional activation of SPOROCYTELESS [SPL; also known as NOZZLE (NZZ)], and this mechanism is necessary for nucellus formation (Bencivenga et al., 2012). Extension of the integuments is incomplete when WUS expression is driven in the chalzal domain by the AINTEGUMENTA promoter (pANT) (Gross-Hardt et al., 2002). In pANT≫WUS plants additional putative outer integuments also emerge just below the extended WUS expression area (Gross-Hardt et al., 2002; Sieber et al., 2004), as if a new boundary is established between the nucellus and chalaza. These results indicate that WUS regulation is key to defining the nucellus-chalaza boundary.

A recent study showed that externally applied cytokinin expands WUS expression into the chalaza (Bencivenga et al., 2012). WUS expression is also altered in bell1 (bel1) (Bencivenga et al., 2012; Brambilla et al., 2007) (see Fig. S2), but the change in the expression is not as profound, suggesting that other factors play roles in WUS regulation.

Class III homeodomain leucine zipper (HD-ZIPIII) genes establish the identity of the adaxial tissue of lateral organs (Emery et al., 2003; McConnell et al., 2001), as well as regulating WUS expression in shoot and floral apices in cooperation with CLAVATA3 or ERECTA (ER) homologs (Green et al., 2005; Landau et al., 2015; Lee and Clark, 2015; Mandel et al., 2014). In ovules, HD-ZIPIII genes are involved in the development of the integuments; thus, in most ovules of loss-of-function HD-ZIPIII gene mutants [i.e. corona phabulosa phavoluta (cna phb phv)] the integuments are absent, reduced or malformed (Kelley et al., 2009). The mechanism of these effects remains unclear.

Here, we show that CNA, PHB and PHV cooperatively repress WUS expression in the chalaza, and that this repression is important for ovule development. A combination of these mutations with bel1 leads to additional misexpression of WUS outside of the nucellus and a novel phenotype not seen in either class of mutant. Thus, we identify transcription factors necessary for boundary demarcation between the nucellus and chalaza and for patterning of WUS expression.

In cna phb phv carpels with ovules at stage 2 (for stages, see Schneitz et al., 1995), WUS expression was ∼1.8-fold higher than in wild-type (WT) carpels at the same stage (Fig. S1). Since gynoecium formation is almost completed in cna phb phv at this stage, we hypothesize that the increase results from elevated WUS expression in developing ovules.

In contrast to WT, where WUS expression was confined to the nucellus (Fig. S2), in most cna phb phv ovules at stage 2-II, WUS expression extended into the chalaza, including the integuments (Fig. 1B). Misexpression of WUS in the chalaza was still observed in ovules at stage 2-III, whereas expression in the nucellus decreased (Fig. 1C). A subset of ovules did, however, exhibit WT-like expression of WUS (Fig. 1D). In cna phb phv some ovules do develop normally, although most exhibit aberrant integument development (Kelley et al., 2009). Coexistence of ovules with normal and abnormal WUS expression is thus consistent with the ovule phenotypes.

In summary, CNA, PHB and PHV are required for preventing WUS expression in the chalaza. However, the sporadic WT-like phenotype in cna phb phv ovules implies that other gene(s) are also involved in the regulation of WUS.

Growth of the integuments was disturbed in pANT≫WUS plants (Gross-Hardt et al., 2002). Thus, the misexpression of WUS could account for the aberrant shape of the integuments in cna phb phv.

The ectopically expressed WUS in cna phb phv was knocked down using an artificial microRNA for WUS (amiRWUS). Since the CNA promoter (pCNA) drives gene transcription throughout the chalaza, including the two integuments, we utilized pCNA as the driver of amiRWUS expression (Fig. S1). In gynoecia of cna phb phv pCNA:amiRWUS at stage 2, amiRWUS reduced WUS expression to a level that averaged 60% of that observed in cna phb phv (Fig. 2A). Suppression of WUS expression was also evaluated using a gWUS-GFP₃ transgene (Tucker et al., 2008) (Fig. S2). This gene showed an expanded GFP signal in cna phb phv, but the expanded signal was absent from the chalaza in almost all ovules of cna phb phv pCNA:amiRWUS (compare Fig. 2C,E with 2D,F; Fig. S2).

Development of the integuments was restored in most ovules of cna phb phv pCNA:amiRWUS (compare Fig. 2G,I with 2H,J). The percentage of normal ovules per carpel significantly increased in cna phb phv pCNA:amiRWUS as compared with cna phb phv (Fig. 2B). Therefore, the misexpression of WUS partly accounts for the aberrant shape of the integuments in cna phb phv.

Similar to phenotypes of loss-of-function cna phb phv mutants, ovules in HD-ZIPIII gain-of-function mutants, such as phb-1d or phv-1d, have aberrant integuments (Kelley et al., 2009). However, in phb-1d, WUS expression did not differ from WT (Sieber et al., 2004), suggesting that PHB has effects on integument growth that are independent of WUS repression.

As in lateral organ primordia, CNA, PHB and PHV are expressed in the adaxial tissue of the inner integument. Thus, it is suggested that polarity establishment is also required for inner integument expansion (Kelley et al., 2009). The aberrant ovules in cna phb phv pCNA:amiRWUS suggest that CNA, PHB and PHV promote integument growth by other mechanisms, such as establishing adaxial-abaxial polarity, as well as by repressing WUS. Alternatively, these aberrant ovules might be attributed to variability in expression of the amiRWUS transgene.

Seeds are formed even when either CNA, PHB or PHV is constitutively expressed by the CaMV 35S promoter (Prigge et al., 2005), in contrast to the seedless phenotype of wus (Gross-Hardt et al., 2002), suggesting that HD-ZIPIII factors do not directly repress WUS expression. We drove expression of CNA under the control of the WUS promoter to corroborate these results. Since HD-ZIPIII transcripts are post-transcriptionally targeted for degradation by microRNA (miR) 165/166 (Emery et al., 2003), we used CNA-δmiRNA, in which the miRNA binding site is modified to be insensitive to miR165/166, in addition to WT CNA.

Both in pWUS:CNA-δmiRNA (Fig. 2L,O) and pWUS:CNA (Fig. 2M,P), ovules did not obviously differ from those of WT (Fig. 2K,N), suggesting that CNA requires other factors to regulate WUS in the chalaza.

WUS expression extends into the chalazal domain when cytokinin is externally applied to the gynoecium (Bencivenga et al., 2012). Therefore, misexpression of WUS in cna phb phv might result from cytokinin upregulation. We compared cytokinin responses between WT and cna phb phv ovules using a TCS:GFP marker (Müller and Sheen, 2008).

In WT ovules, cytokinin response is observed in the chalaza and funiculus at stage 2-III and later (Fig. 3A,B) (Bencivenga et al., 2012). The same pattern is observed in cna phb phv ovules with aberrant (Fig. 3C,D) or normal (Fig. 3E,F) integuments, but expression levels are reduced, compared with WT (Fig. 3G). These data indicate that CNA, PHB and PHV regulation of WUS expression is not by means of cytokinin upregulation. In addition, ovule morphology is not affected when cytokinin degradation is promoted by constitutive expression of Cytokinin oxidase/dehydrogenase genes (Werner et al., 2003), supporting the conclusion that the cna phb phv phenotype is independent of cytokinin regulation.

In bel1 an amorphous structure forms in place of the integuments and can convert to a carpelloid organ (e.g. Robinson-Beers et al., 1992; Ray et al., 1994). In bel1 ovules WUS expression is shifted downward into the boundary region between the nucellus and chalaza (Bencivenga et al., 2012) (Fig. S2). Thus, we crossed cna phb phv with bel1 to evaluate the interaction of the genes.

bel1 cna phb phv ovules form the amorphous structure seen in bel1, but the primary ovulate axis was observed to branch through the formation of ectopic ovule primordia (Fig. 3H,I). These extra primordia were formed directly on the funiculus below the chalaza, in contrast to the extra putative nucelli of bel1, which are born on the adaxial side of the amorphous organ (Robinson-Beers et al., 1992). The ectopic ovules form amorphous organs typical of bel1 mutants in place of integuments (Fig. 3I). Formation of ectopic ovules is preceded by nodes of ectopic WUS expression in the funiculus (Fig. 3J,K). WUS is expressed in the nucellus and chalaza of bel1 cna phb phv ovules (Fig. 3J), as in cna phb phv ovules (Fig. 1B,C). However, stronger expression was detected in the nucelli of the ectopic ovules at the same stage in their development (Fig. 3J). As the amorphous organ enlarged, WUS expression decreased in the nucellus and chalaza of the primary ovulate axis, whereas expression persisted in the ectopic ovules (Fig. 3K).

The ovulate axis does not branch in cna phb phv (Kelley et al., 2009) nor in bel1 (Robinson-Beers et al., 1992). Consistent with the novel phenotype of bel1 cna phb phv, the WUS expression pattern in this quadruple mutant is different from that observed in bel1 (Fig. S2) or in cna phb phv (Fig. 1B,C). These synergistic phenotypes suggest that CNA, PHB and PHV act in a different pathway from BEL1 to regulate WUS. The combination of the two classes of mutations appears to allow for WUS expression further down the ovule axis than is observed in either single class. The formation of ectopic ovules from the funiculus implies that cells in the funiculus manifest placenta-like properties in the quadruple mutant.

We show that CNA, PHB and PHV play a major role in the negative regulation of WUS in the chalaza. This regulation contributes to the establishment of the boundary between the nucellus and chalaza and promotes the proper development of two integuments. CNA, PHB and PHV repress WUS expression independently from BEL1 or cytokinin upregulation (Fig. 4A).

Integument growth is normal in cna, phb and phv single mutants (Kelley et al., 2009), suggesting that they redundantly repress WUS expression. In stage 2 ovules, expression of CNA, PHB and PHV does not completely overlap (Kelley et al., 2009; Sieber et al., 2004) (Fig. 4B), whereas WUS is misexpressed throughout the chalaza in cna phb phv, a much broader area than the sum of the CNA, PHB and PHV expression areas. This discrepancy would imply that CNA, PHB and PHV regulate the expression of WUS through their repression of a diffusing factor or an action of the factor (X in Fig. 4). Auxin is a possible candidate for this factor because it is predicted to flow through the chalaza (Bencivenga et al., 2012; Kelley et al., 2012), and alterations in the pattern of auxin response correlate with ectopic WUS expression seen in bel1 mutants (Bencivenga et al., 2012). Alternatively, CNA or PHV might have broader expression areas in stage I ovules, as actually reported for PHB (Sieber et al., 2004), and failure in suppression at this stage could affect the expression patterns of WUS at subsequent stages.

The striking synergistic branched ovule phenotype observed in bel1 cna phv phb mutants implies that the two classes of genes act in relative independence in their suppression of WUS activity outside of the nucellus. Angiosperm ovules are hypothesized to be homologous to the cupulate organs of some extinct gymnosperms, each of which were born on the apices of a branched axis (e.g. Doyle, 2006), but such branched structures are not observed in extant angiosperms. If this hypothesis is correct, then BEL1 and HD-ZIPIIIs could have played a role in the evolution of the unbranched ovule form seen in all extant angiosperms.

Arabidopsis plants were grown under continuous light at 23°C on soil. Seeds were sourced as described in the supplementary Materials and Methods.

The amiRWUS fragment was synthesized following Web MicroRNA Designer 3 (http://wmd3.weigelworld.org/cgi-bin/webapp.cgi). amiRWUS binds 607 to 627 nucleotides of WUS (Fig. S1). amiRWUS, as well as pCNA (−4046 to −319), were inserted into pMLBarT (Fig. S1) using the GeneArt Seamless PLUS Cloning and Assembly Kit (Life Technologies). bel1-6/+ cna-2 phb-13 phv-11 er-2 plants were transformed with the construct by the floral dip method (Clough and Bent, 1998).

To generate pWUS:CNA-δmiRNA or pWUS:CNA, the WUS promoter and CNA cDNA sequences were obtained from Col-0 genomic DNA or cDNA by PCR, respectively. CNA-δmiRNA sequence was synthesized by overlap PCR as previously described (Emery et al., 2003). These fragments were cloned into pMLBarT as described above.

Further details of constructs and genotyping are given in the supplementary Materials and Methods and Table S1.

Fixation, embedding of tissue and in situ hybridization were performed as previously described (Mayer et al., 1998). GUS staining of pWUS>>uidA plants is described in the supplementary Materials and Methods.

For scanning electron microscopy (SEM), ovules were fixed (McAbee et al., 2006) or epoxy molds were made (Williams et al., 1987). Fluorescence images of GFP were taken with excitation and emission wavelengths of 470/20 nm and 505-530 nm, respectively, using an Axio Scope 2 Plus (Carl Zeiss).

Total RNAs were extracted from gynoecia containing stage 2 ovules and contaminating DNA was digested with DNase. qRT-PCR analyses were performed using the One Step SYBR PrimeScript RT-PCR Kit (Takara). WUS and GFP expression levels were normalized to those of PP2AA3 (At1g13320) (Czechowski et al., 2005). For further details, see the supplementary Materials and Methods.

We thank John Bowman, Steven Clark, Shinobu Takada and Nottingham Arabidopsis Stock Centre (NASC) for seeds; and Debra Skinner and Marissa Simon for helpful comments.