PIN-FORMED1 and PINOID regulate boundary formation and cotyledon development in Arabidopsis embryogenesis

In dicotyledonous plants, the above-ground part of the seedling exhibits bilateral symmetry, as evidenced by two symmetrically located cotyledons and the shoot apical meristem (SAM) between them. This symmetry can be traced back to early embryogenesis, where two cotyledon primordia start to grow from the apical region of a radially symmetrical globular embryo(Long and Barton, 1998; Bowman and Eshed, 2000; Jürgens, 2001; Aida et al., 2002; Furutani et al., 2003). How plant embryos promote the outgrowth of cotyledon primordia with a bilaterally symmetric pattern has remained an important question and has been studied using physiological or genetic approaches.

Previous studies have indicated that auxin is involved in various patterning processes, including apical patterning during embryogenesis. Auxin displays asymmetric distribution that changes dynamically throughout early embryogenesis and polar auxin transport is important for this distribution(Sabatini et al., 1999; Friml et al., 2002; Benková et al., 2003; Friml et al., 2003). Treatment of embryos with exogenous auxin or polar transport inhibitors causes variable defects in the apical pattern formation, including abnormal positioning or fusion of cotyledons (Liu et al.,1993; Hadfi et al.,1998; Friml et al.,2003). These results suggest that proper auxin distribution is important for the symmetrical positioning of cotyledon primordia and the establishment of cotyledon boundaries.

Genetic studies support the role of auxin in the above-mentioned processes. In Arabidopsis, mutations in the PIN-FORMED1(PIN1), MONOPTEROS (MP) and PINOID(PID) genes disrupt the patterning of cotyledons(Okada et al., 1991; Berleth and Jürgens, 1993; Bennett et al., 1995). These mutants are also defective in lateral organ formation from the postembryonic shoot meristem, indicating their significant role in lateral shoot organ development. The PIN1 gene encodes a member of the putative auxin efflux regulator proteins that promote polar auxin transport(Gälweiler et al., 1998)and is suggested to promote organ formation by regulating auxin distribution(Benková et al., 2003; Reinhardt et al., 2003). The MP gene encodes a transcriptional activator that binds in vitro to an auxin-responsive cis-element and is suggested to promote primordium formation by mediating auxin-induced transcriptional activation(Ulmasov et al., 1997a; Hardtke and Berleth, 1998). The PID gene encodes a serine/threonine kinase, the transcription of which is induced by exogenous auxin(Christensen et al., 2000; Benjamins et al., 2001). Similar to the pin1 mutant, the pid mutant displays a reduction of polar auxin transport in the stem(Bennett et al., 1995). Moreover, plants overexpressing PID exhibit reduced root growth and this phenotype is suppressed by treatment with polar auxin transport inhibitors. These results suggest that PID functions as a positive regulator of polar auxin transport(Benjamins et al., 2001).

The expression of CUP-SHAPED COTYLEDON1 (CUC1) and its functionally redundant homolog, CUC2, has been analyzed in pin1 and mp mutant embryos(Aida et al., 2002). These genes encode transcription factors of the NAC family, promote cotyledon separation at the boundaries and cause cotyledon fusion when both of them are mutated. Although these two genes are normally expressed in a stripe between cotyledon primordia, CUC1 expression is expanded to the periphery of the apical region and CUC2 expression is restricted to the center. The effects of pin1 or mp mutations on CUC1 and CUC2 expression are well correlated with the fusion phenotype of pin1 or mp mutants as well as their double mutant combinations with cuc1 or cuc2. These results suggest that PIN1 and MP regulate boundary formation by regulating the CUC1 and CUC2 genes. The SHOOT MERISTEMLESS(STM) gene, a kn1-type homeobox gene required for SAM formation and maintenance (Barton and Poethig, 1993; Clark et al.,1996; Endrizzi et al.,1996; Long et al.,1996), also participates in promoting cotyledon separation and its contribution is particularly prominent in the pin1 mutant background(Aida et al., 2002).

To further investigate the molecular relationship between auxin and apical pattern formation in the Arabidopsis embryo, we examined the functions of PIN1 and PID genes in this process. We found that pin1 and pid mutations, when combined in the double mutant, lead to a striking seedling phenotype that is represented by a radially symmetric shape without any cotyledons. This phenotype is associated with the prolonged expansion of CUC1, CUC2 and STMexpression domains to the periphery of the embryo apex, and contrasts with the mild and transient changes in CUC1 and CUC2 expression observed in the pin1 or pid single mutant. Triple and quadruple mutant analysis indicates that the ectopic expression of CUC1,CUC2 and STM genes in the pin1 pid double mutant is mainly responsible for the growth inhibition of cotyledon primordia. Our results thus demonstrate that the overlapping function of PIN1 and PID is largely responsible for the establishment of bilateral symmetry and cotyledon outgrowth, and that the latter process involves the negative regulation of boundary-specific downstream effectors, the CUC and STM genes.

Arabidopsis thaliana ecotype Lansberg erecta(Ler) was used as the wild type. The following mutant alleles were used: cuc1-1 (Ler)(Takada et al., 2001), cuc2 (Ler) (Aida et al.,1997), pin1-3 (Ler)(Bennett et al., 1995), pin1-6 [Wassilewskija (WS)](Vernoux et al., 2000), pin1-201 [Columbia (Col)], pid-2 (Ler)(Christensen et al., 2000), pid-7.1.2.6 (WS), pid-3 (Col)(Bennett et al., 1995) and stm-1 (Ler) (Barton and Poethig, 1993). pin1-201 carries a T-DNA insertion at the third intron and is supposed to be null. This allele was obtained from the Arabidopsis Biological Resource Center (SALK_047613)(Alonso et al., 2003) and was backcrossed four times to Col wild type prior to any phenotypic analysis and construction of the pin1 pid double mutant. The pid-7.1.2.6allele contains a substitution that modifies codon 271, creating a stop (Q271 stop) that eliminates approximately one-third of the protein, suggesting that it is a null allele. Plants were grown on soil as previously described(Fukaki et al., 1996), and siliques were collected for analyses of embryo phenotypes and in situ hybridization. Stages of embryogenesis were as defined previously(Jürgens and Mayer,1994). For analysis of seedling phenotypes, seeds were surface sterilized and germinated on Murashige and Skoog plates, as previously described (Aida et al.,1997).

For the construction of the pin1-3 pid-2 double mutant, plants heterozygous for pin1-3 were crossed with pid-2 homozygotes. Among the F2 population, plants homozygous for pid-2 and heterozygous for pin1-3 were selected by using PCR primers that detected the mutations and self-fertilized. Among F3 populations, double mutants were selected by PCR-based genotyping or the presence of the novel specific phenotype (see Results). pin1-6 pid-7.1.2.6 and pin1-201 pid-3 double mutants also displayed the same phenotype. For the construction of the pin1 pid cuc1, pin1 pid cuc2 or pin1 pid stm triple mutants, plants heterozygous for pin1-3 and homozygous for pid-2 were crossed with cuc1-1 homozygous, cuc2 homozygous or stm-1 heterozygous plants, respectively. Among the F2 populations, plants homozygous for cuc1-1 or cuc2, or heterozygous for stm-1 were selected by PCR-based genotyping. These plants were further selected for the heterozygous pin1-3 and homozygous pid-2 mutation by PCR, and for the pin-shaped inflorescence phenotype (Bennett et al., 1995). Phenotypes of the triple mutants were examined in the F3 populations and their genotypes were confirmed by PCR. The STMand PIN1 loci were located on chromosome 1 and closely linked. The occurrence frequency of the novel phenotype was 2.4% (n=127), which was nearly identical to that of the pin1 stm double mutant, as previously described (Aida et al.,2002). For the construction of the pin1 pid cuc1 cuc2quadruple mutant, plants heterozygous for pin1-3 and homozygous for pid-2 and cuc1-1 were crossed with plants heterozygous for pin1-3 and homozygous for pid-2 and cuc2. Among the F2 populations, plants heterozygous for both pin1-3 and cuc1-1, and homozygous for cuc2 were selected. Seedling phenotypes of the quadruple mutants were examined in the F3 populations.

For visualization of vasculature, seedlings were cleared as previously described (Aida et al., 1997). Scanning electron microscopy images were obtained as described previously(Aida et al., 1999).

In situ hybridization was performed as previously described(Aida et al., 2002). Hybridization was performed at 45°C. Templates for transcription of a PID antisense probe were derived from a PCR-amplified 1122 bp fragment corresponding to a region that spanned amino acids 44-417. Probes for the following genes have been reported previously: ANT(Long and Barton, 1998), CUC1 (Takada et al.,2001), CUC2 (Aida et al., 1999), FIL (Sawa et al., 1999) and STM(Long et al., 1996). As controls, we confirmed the expression patterns of FIL, PID, CUC1 and CUC2 genes in wild type. For any of these probes, we detected aberrant expression patterns (expanded or reduced) in fewer than 5% of the embryos (three out of 88 for FIL; four of 104 for PID; six of 113 for CUC1; and six of 132 for CUC2).

To detect GUS activity, embryos were stained with a solution described previously (Takada et al.,2001) at 37°C for 45 minutes. Stained embryos were dehydrated in a graded ethanol series (30, 50, 70, 90 and 100%) for 15 minutes each. Rehydration in a graded ethanol series (90, 70, 50, and 30%) was performed for 15 minutes each before observation.

Plants were grown under constant white light exposure until several siliques started to develop (∼3 weeks). All developing siliques were cut off before auxin treatment. The plants were subsequently sprayed with a heavy mist of 10 μM 2,4-dichlorophenoxy-acetic acid with 0.01% Silwet L-77. Mock treatments were performed with distilled water containing 0.01% Silwet L-77. Auxin treatments were repeated once a week for 1 month. Seeds were collected and germinated on plates for phenotypic analysis. For DR5::GUSanalysis, embryos were dissected from siliques 7 days after the treatment and stained with GUS staining solution.

A wild-type Arabidopsis seedling had two separated cotyledons symmetrically arranged with equal size and shape and the SAM at the center of the apex (Fig. 1A). A fraction of the pin1 mutant seedlings displayed defects in cotyledon number,separation, position and size to various extents(Fig. 1B)(Aida et al., 2002). The frequency of cotyledon phenotype was almost equal among three strong alleles of pin1 (Table 1). A fraction of pid seedlings displayed cotyledon phenotypes, similar to that of pin1 seedlings (Fig. 1C). The pid mutant, however, tends to cause decrease or increase in number of cotyledons more frequently than fusion, in contrast to pin1 in which fusion is frequently observed (data not shown). A small fraction of strong alleles, pid-7.1.2.6 and pid-3, produced seedlings lacking cotyledons. The intermediate pid-2 allele displayed slightly milder cotyledon phenotypes than did the two other pidalleles (Table 1). Both pin1 and pid seedling phenotypes originated from embryogenesis, as evidenced by asymmetrical growth or abnormal positioning of cotyledon primordia (Fig. 1D-F)(Aida et al., 2002).

To examine in detail symmetry development as well as cotyledon primordium formation in pin1 and pid mutant embryos, we analyzed the expression of marker genes that displayed specific expression patterns in the apical region of the embryo. The AINTEGUMENTA (ANT) gene was expressed in a radially symmetric ring around the embryonic apex, including cotyledon primordia at the early heart stage in wild-type embryos(Long and Barton, 1998)(Fig. 1G). FILAMENTOUS FLOWER (FIL) was expressed in the abaxial side of presumptive cotyledon primordia at the early heart stage, exhibiting bilateral symmetry(Siegfried et al., 1999)(Fig. 1H).

We analyzed the expression of these marker genes in pin1-3/+ or pid-2/+ siliques because pin1-3 homozygous mutants were completely sterile and the fertility of pid-2 homozygous mutants was significantly low. ANT was expressed in a ring in all the embryos examined in pin1-3/+ and pid-2/+ siliques (data not shown),indicating that neither homozygous mutation affected the ANTexpression pattern. By contrast, FIL expression was disturbed in both pin1-3 and pid-2 embryos(Fig. 1I,J). In ∼26% of the embryos in pin1-3/+ siliques (13 of 49), FIL was expressed in a ring that surrounded the apex of the embryos(Fig. 1I) or in an incomplete ring with a breakage at one side. FIL expression in a ring was detected also at the late heart stage (data not shown). In ∼21% (13 out of 62) of the embryos in pid-2/+ siliques, FIL expression was asymmetric so that the strength and size of the signals differed between the two domains of expression (Fig. 1J). Taken together, the radial expression pattern of ANTis preserved, whereas the bilateral expression pattern of FIL is disturbed in pin1 and pid embryos. In particular, the pin1 mutation severely disrupts FIL expression, resulting in a radially symmetric pattern.

To examine the genetic interaction between PIN1 and PIDin cotyledon development, we constructed the pin1 pid double mutant. Seedlings with the most severe phenotype completely lacked cotyledons,displaying radial symmetry (Fig. 2A,D; Table 2). Seedlings with the mild phenotype developed small bulges that were most likely rudimentary cotyledons. The epidermal cells of these bulges were small and round compared with those of wild-type cotyledons(Fig. 2B,E; Table 2). Seedlings that displayed the weak phenotype produced small flat structures with a ridge along the margin of the adaxial side (Fig. 2C; Table 2). All pin1 pid seedlings developed a functional SAM that produced leaf primordia, although these primordia displayed abnormal phyllotaxis and were often fused with each other (Fig. 2D, arrowheads). Mutant SAMs continued to produce leaves and eventually developed pin-like inflorescences similar to those of pin1or pid single mutant. These phenotypes were observed in three different combinations of pin1 and pid alleles, including putative null mutants (Table 2;see Materials and methods). Because the observed genetic interaction was allele-nonspecific, we conclude that PIN1 and PIDredundantly promote cotyledon development, but are not essential for SAM formation and maintenance.

To investigate the development of cotyledon primordia and symmetry in pin1 pid embryos, we examined ANT and FILexpression in siliques of pin1-3/+ pid-2/pid-2 plants as the double homozygous mutant was sterile. At the early heart stage, the expression patterns of these marker genes were almost identical to those observed in the pin1 single mutant: ANT expression was normal (data not shown) and FIL expression formed a ring in ∼19% of the embryos(seven out of 37 in pin1-3/+ pid-2/pid-2 siliques; Fig. 1K). Similar to pin1,FIL continued to be expressed in a ring-like domain in the pin1 pid double mutants at the late heart stage (data not shown). Taken together, our results indicate that the pin1 pid double mutant fails to establish bilateral symmetry although it retains radial symmetry.

We next examined the expression of pattern of PID in the embryo. Although previous studies have reported that PID expression is detected mainly in developing cotyledon primordia(Christensen et al., 2000; Benjamins et al., 2001), a more detailed expression study is required to assess PID function during embryogenesis.

At the globular stage, PID mRNA expression was detectable in two domains at opposite sides, each encompassing approximately three-quarters of the embryo along the longitudinal axis(Fig. 3A,H). In early heart-stage embryos, PID mRNA expression was again detected in two opposite domains that mainly included the boundary between cotyledon primordia and the basal part of the primordia (Fig. 3B,C,H). No expression was detected in the presumptive SAM region or at the top part of cotyledon primordia. In the late heart to torpedo stages, PID mRNA expression was found in the adaxial side of cotyledon primordia (Fig. 3D,H)as well as in the boundary between cotyledon primordia(Fig. 3E,H). Typically, the signal in the cotyledon boundaries was stronger than those in the other regions. Our results thus demonstrate PID mRNA expression at the boundaries, which has not been described in previous studies(Christensen et al., 2000; Benjamins et al., 2001).

To determine whether PIN1 affects PID expression, we examined embryos developing in siliques of pin1-3/+ plants. At the early heart stage, we found embryos in which one of the two domains of PID expression was significantly reduced (12 out of 57). In addition,embryos in which both expression domains were greatly reduced were found at a low frequency (five out of 57; Fig. 3F). In the late heart to torpedo stages, PID expression was detected in cotyledon primordia as well as in their boundaries at a level comparable with that of the wild type (Fig. 3G). These results indicate that PID expression partially requires PIN1 activity at the early stage but not at the late stages.

In the wild type, the CUC1 and CUC2 genes, which promote cotyledon separation by preventing growth at the boundaries, were expressed in a stripe between cotyledon primordia (Fig. 4A,B). We previously have shown that, in pin1 embryos, CUC1 expression expands to almost the entire apex, whereas CUC2 expression is restricted to a central spot at the early heart stage (Aida et al., 2002). In this study, we further analyzed the expression of these genes at the late heart stage. At this stage, the expansion of CUC1 expression to the periphery was partial, as revealed by the restricted spots of signals in cotyledon primordia (Fig. 5A). At the same stage, CUC2 expression was mainly detected at the center of the apex and occasionally in part of the cotyledon primordia(Fig. 5B).

We next examined the effect of pid mutation alone, as well as that of pin1 pid double mutations, on CUC1 and CUC2expression. In embryos developing in pid-2/+ siliques, we found that the stripe of CUC1 and CUC2 expression became slightly wide and slanted in ∼10% of the transition-stage embryos (four out of 42, five out of 47 respectively; Fig. 4C,D). As cotyledon development proceeds, the expression patterns became essentially the same as that of the wild type, with a few exceptions,namely, CUC1 expression branched into three lines or CUC2expression was curved, branched or broken in the middle of the stripe (data not shown). We conclude that the pid single mutation only mildly affects the expression of CUC1 and CUC2.

To analyze CUC1 and CUC2 expression in the pin1 pid double mutant, we examined embryos developing in pin1-3/+pid-2/pid-2 siliques. At the early heart stage, CUC1 expression was expanded to include nearly the entire apex in ∼24% of the embryos (13 out of 55; Fig. 4E). This pattern was essentially the same as that in the pin1 single mutant. At the late heart stage, however, CUC1 expression remained in almost the entire apex of the pin1 pid double mutant(Fig. 5C), in contrast to that of the pin1 single mutant. CUC2 expression was detected in the central region and occasionally at part of the periphery in 19% of early heart stage embryos (eight out of 42; Fig. 4F). This expression pattern was maintained at the late heart stage (Fig. 5D).

Collectively, these results indicate that the pid mutation, when combined with pin1, causes a prolonged and complete expansion of CUC1 expression into the peripheral region of the embryonic apex. This is in contrast to the pin1 single mutant, in which the CUC1 expression is expanded only transiently and soon becomes restricted at the late heart stage because of the action of the PIDgene. The pin1 pid double mutations also cause a slight expansion of CUC2 expression, in contrast to the pin1 single mutation in which CUC2 expression is reduced.

The prolonged expansion of CUC1 expression as well as the slight expansion of CUC2 expression in cotyledon primordia may account for the loss of cotyledon formation in the pin1 pid double mutant. To test this possibility, we constructed pin1 pid cuc1 and pin1 pid cuc2 triple mutants, as well as pin1 pid cuc1 cuc2 quadruple mutant to genetically eliminate CUC1 and/or CUC2 function from the pin1 pid background.

The pin1 pid cuc1 triple mutations markedly recovered cotyledon development, resulting in the formation of cup-shaped fused cotyledons(Fig. 5H). The extent of cotyledon fusion varied among seedlings, ranging from a partial fusion at the base to a nearly complete fusion. By contrast, most of the pin1 pid cuc2 triple mutants were indistinguishable from the pin1 piddouble mutants, except for a few seedlings that produced partially fused cotyledons, the sizes of which were larger than those of the rudimentary cotyledons observed in the pin1 pid double mutants(Fig. 5I). These results indicate that ectopic CUC1 activity mainly prevents cotyledon formation in the pin1 pid double mutant and CUC2 partially contributes to this process. The fusion phenotype observed in each triple mutant combination may be due to the reduced activities of cotyledon separation caused by the cuc1 or cuc2 mutation at least in part.

In the pin1 pid cuc1 cuc2 quadruple mutant, all seedlings developed cotyledon with a fused cup shape(Fig. 5J). The extent of fusion was more pronounced and complete than that in the pin1 pid cuc1triple mutant. Thus, the ectopic activities of CUC1 and CUC2fully account for the repression of cotyledon growth in the pin1 piddouble mutant.

To examine symmetry in the quadruple mutant, we observed the vascular pattern of cotyledons, a suitable marker for seedling symmetry(Aida et al., 1997). Both wild-type (data not shown) and the cuc1 cuc2 double mutant(Fig. 5G) displayed bilaterally symmetrical arrangement of vascular strands(Fig. 5K,M). By contrast, the pin1 pid cuc1 cuc2 quadruple mutant displays a radially symmetrical arrangement (Fig. 5L,N),similar to the arrangement described for the pin1 cuc1 cuc2 triple mutant (Aida et al., 2002). These results are consistent with the loss of bilateral symmetry in the pin1 pid double mutant and indicate that the addition of cuc1 and cuc2 mutations does not rescue the symmetry defect.

Finally, we examined the effect of cuc1 or cuc2 mutation on pid mutation alone. In the pid cuc1 double mutant, the extent of cotyledon fusion was slightly enhanced compared with that in the pid single mutant, whereas the frequency of fusion was not changed(data not shown). By contrast, seedlings of the pid cuc2 double mutant were phenotypically indistinguishable from those of the pidsingle mutant (data not shown). These results show that neither cuc1nor cuc2 markedly affects the phenotype of the pid single mutant.

We next examined the effect of the pin1 pid double mutations on the expression of the STM gene, which is involved in SAM formation and cotyledon separation downstream of the CUC1 and CUC2genes (Aida et al., 1999; Takada et al., 2001). STM was expressed in a stripe between cotyledon primordia at the heart stage of the wild type (Fig. 5E), whereas it was expanded to include almost the entire apex in the pin1 pid double mutant (Fig. 5F).

To eliminate STM activity from the pin1 pid background,we constructed the pin1 pid stm triple mutant. Addition of the strong stm-1 allele (Fig. 5O)to the pin1 pid double mutant partially recovered cotyledon development, as evidenced by the formation of large cotyledons compared with those of pin1 pid seedlings with mild phenotypes(Fig. 5P). The recovery of cotyledon growth by stm, however, was much less pronounced compared with that by the cuc1 mutation. These results indicate that ectopic STM expression in the pin1 pid double mutant is also responsible for the growth inhibition of cotyledon primordia, although its contribution is partial.

If the observed effects of the pin1 and pid mutations on the CUC1 and CUC2 genes were caused by changes in auxin distribution, the exogenous application of auxin could also affect these genes, thereby perturbing normal cotyledon development. Consistently, previous studies have shown that auxin treatment causes various cotyledon defects including fusion, a phenocopy of the cuc1 cuc2 double mutant(Liu et al., 1993; Hadfi et al., 1998; Friml et al., 2003). To strengthen this hypothesis, we further tested the effect of auxin on cuc1 and cuc2 single mutant embryos (Materials and methods).

When wild-type embryos were treated with synthetic auxin,2,4-dichlorophenoxyacetic acid (2,4-D), a fraction of the embryos developed into seedlings with abnormal cotyledons: 3.0% of the seedlings displayed weak fusion and 6.4% displayed complete fusion of cotyledons(Fig. 6A; Table 3). The vascular pattern of the latter class was considerably irregular and not bilaterally symmetric(Fig. 6B). In 2,4-D-treated embryos, the activity of an auxin-responsive reporter gene, DR5::GUS(Ulmasov et al., 1997b; Sabatini et al., 1999; Friml et al., 2003), was detected in a broader region compared with that in mock-treated embryos(Fig. 6C,D). These results suggest that the application of auxin to the embryo changes the auxin distribution in the apical region and causes cotyledon fusion, possibly by reducing the activities of the CUC genes.

We next treated cuc1 and cuc2 single mutant embryos with 2,4-D. Interestingly, auxin treatment of the cuc1 single mutant resulted in a significantly higher frequency of seedlings with the strong fusion phenotype (Table 3) than that of the wild type. However, the effect of auxin on the cuc2single mutant was similar to that on the wild type(Table 3). Considering that neither cuc1 nor cuc2 mutation changes auxin sensitivity per se (Daimon et al., 2003) and that the CUC1 and CUC2 genes are functionally redundant(Aida et al., 1997), the observed higher frequency of the fusion phenotype in cuc1 embryos may suggest that the activity of CUC2 is reduced more effectively than that of CUC1 upon auxin application.

In this study, we demonstrated a synergistic interaction between mutations in the PIN1 and PID genes, the functions of which are implicated in auxin-mediated organ formation. Seedlings of the pin1and pid single mutants exhibited relatively mild defects in cotyledon development, whereas those of the double mutant completely lacked cotyledons in the most severe cases. The synergistic interaction between pin1and pid mutations was not specific to the combination of alleles,suggesting that the PIN1 and PID genes act redundantly in the same process. Our results illustrate two major aspects of PIN1and PID functions in embryogenesis: establishment of bilateral symmetry and promotion of cotyledon growth.

The apical region of the embryo can be divided into three subregions, each of which follows different developmental fates(Fig. 7A)(Aida et al., 1999). In the wild type, CUC1 and CUC2 are expressed in both the presumptive SAM (PS) and the boundary of cotyledon margins (BCM), whereas FIL is expressed in the cotyledon primordia (CP). The expression patterns of these three genes are bilaterally symmetric. By contrast, ANT is expressed in both CP and BCM, reflecting radial symmetry. The single and double mutants of pin1 and pid all develop a functional SAM, suggesting that PIN1 and PID are not essential for the establishment of PS. Consistent with this, none of the expressions of the above markers reveal any abnormalities in PS of the pin1 and pid mutants.

The expression patterns of the bilateral markers indicate that PIN1 and PID regulate patterning in the peripheral region consisting of CP and BCM. At the early heart stage, both CUC1 and FIL are expressed as a ring in the peripheral region of pin1, suggesting that this region has a mixed identity of CP and BCM(Fig. 7B) and that PIN1 is essential for the initial positioning and partitioning of CP and BCM within the peripheral region. By contrast, no or only partial overlap of CP and BCM seems to occur in pid embryos, as reflected by relatively mild change of CUC1, CUC2 and FIL expression patterns in pid-2 embryos. This finding may suggest that PIDis not involved in the peripheral patterning process at the early heart stage. Alternatively, PID may be required for this process but the effect of pid-2 mutation is not apparent, simply because of the residual PID activity in this intermediate allele.

As embryogenesis proceeds, the area of BCM becomes partially excluded from CP in the pin1 single mutant (Fig. 7C), as indicated by the partial exclusion of the ectopic CUC1 expression at the late heart stage. This exclusion is dependent on PID, as the ectopic CUC1 expression remains in the entire peripheral region in the pin1 pid background at the same stage(Fig. 7D). This finding indicates that the late activity of PID can partially compensate for the failure caused by the loss of PIN1 activity.

Analysis of the pin1and pid mutants in the inflorescence meristem has suggested a difference between the functions of these genes(Reinhardt et al., 2003). When a large amount of auxin is applied locally to the inflorescence meristem of pin1, a fused, color-like flower primordium is induced at a site close to the application. By contrast, the same amount of auxin applied to the pid meristem induces multiple primordia having a normal size but no fusion. The observed response of pin1 meristems is consistent with the idea that PIN1 is involved in organ partitioning in both the embryo and the inflorescence meristem. However, the response of pidinflorescence meristems does not reveal any involvement of the PIDgene in flower primordium partitioning, in contrast to its proposed function based on our analysis of embryogenesis. This difference may reflect different functions of the PID gene between embryo and flower development. Alternatively, pid inflorescence meristems may also display partitioning defects that can be detected only by molecular markers during the early stages of primordium formation.

The prolonged expansion of CUC1 and CUC2 expression in the apex of the pin1 pid embryos suggests that cotyledon growth is suppressed by the ectopic activities of these genes. The elimination of both CUC1 and CUC2 activities from pin1 pid (i.e. quadruple mutant) results in the complete recovery of cotyledon growth, as evidenced by a fused cup-shaped cotyledon that surrounds the seedling apex. These results indicate that PIN1 and PID function to repress CUC1 and CUC2 expression and/or activity in cotyledon primordia, thereby allowing the primordia to develop. In contrast to the pin1 pid double mutant, the pin1 single mutant does not display severe reduction in cotyledon growth, despite the expansion of CUC1 expression. This is probably because that the expansion of CUC1 occurs only transiently. The later exclusion of CUC1expression from the peripheral region may be sufficient for the primordia to develop fully.

Although the pin1 pid embryos display severe reduction or complete loss of cotyledon growth, they express both ANT and FIL,each of which promotes different aspects of shoot organ development(Long and Barton, 1998; Sawa et al., 1999; Siegfried et al., 1999). This observation suggests that the double mutant initiates developmental programs for cotyledon development at least in part. This notion is consistent with the recovery of cotyledon growth in the pin1 pid cuc triple and quadruple mutants, which indicates that the embryos are competent to form cotyledons even when both PIN1 and PID activities are reduced.

The expression patterns of marker genes have also been examined in the inflorescence meristem of the pin1 mutant(Vernoux et al., 2000). Similar to the embryo, the expression of primordium-specific genes such as LFY and ANT is still present, and their expression domains overlap with that of the boundary-specific CUC2 gene, again suggesting that pin1 maintains competence for primordium growth although the growth is suppressed by the ectopic expression of boundary specific factors. Taken together, these results indicate that PIN1and PID promote shoot organ growth by repressing negative factors for organ formation such as the CUC1, CUC2 and STM genes, rather than promoting positive factors, in both the embryo and the inflorescence meristem.

Recent studies have shown that an auxin gradient maximum is present in initiating organ primordia (Benkováet al., 2003; Friml et al.,2003). In the embryo, the maxima of auxin gradients are present at the tips of cotyledon and root primordia, and those at the cotyledon primordia are likely to be dependent mainly on PIN1, although other members of the PIN family are also suggested to have partially redundant functions(Benková et al., 2003)(Fig. 7E). Complementary distributions of the auxin gradient maxima and the domain of CUC1 and CUC2 expression suggest that auxin negatively regulates the expression of these genes (Fig. 7F). This idea is consistent with our auxin application experiment, showing that an increased concentration of auxin in the apical region induces the cotyledon fusion phenotype. The frequency of the phenotype is higher in the cuc1 mutant than in the cuc2 mutant. Because each mutation does not affect auxin response per se(Daimon et al., 2003), this result may suggest that CUC2 is more effectively repressed by auxin than is CUC1.

Our results demonstrate that PID has an overlapping function with PIN1 in patterning the periphery of the embryonic apex. PIDtranscripts accumulate mainly at the boundaries of cotyledon primordia and slightly in regions that surround the base of cotyledon primordia(Fig. 7E). We also found that PID expression at the early heart stage is dependent on PIN1. Considering that PID is an auxin-inducible gene(Benjamin et al., 2001), we speculate that the initial PID expression is induced in response to the auxin distribution established by PIN1. Although the precise cellular function of the PID protein is still unclear, previous studies have suggested its role in promoting auxin transport(Benjamin et al., 2001). As PID and CUC expression domains are overlapping in the boundary of cotyledon primordia, we suggest that PID, by promoting auxin transport, reduces the level of auxin at the boundary and increases it in the primordia (Fig. 7E) to limit CUC1 and CUC2 expression to the boundary(Fig. 7F). Detailed analysis of the effects of PID on auxin transport and distribution and identification of the cellular process in which PID functions are required for uncovering the mechanism by which PID regulates patterning in the apical region of the embryo.

We thank Miyo Terao Morita for critical reading of the manuscript. We also thank David R. Smyth for providing pid-2 and pid-3 seeds. This work was partially supported by Grants-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Science, and Culture, Sports,Science, and Technology of Japan (14036222 to M.T.) and Encouragement of Young Scientists (12740439 to M.A.). M.F. was supported by fellowships from the Japan Society for the Promotion of Science.