LATERAL ORGAN FUSION1 and LATERAL ORGAN FUSION2function in lateral organ separation and axillary meristem formation in Arabidopsis

The shoot apical meristem (SAM) plays a crucial role in plant development as a continuous source of new cells for organogenesis. In contrast to animals,which form organs primarily during embryogenesis, plants produce organs throughout their life cycle. The SAM contains a population of self-renewing stem cells at its center, whereas cells at the periphery are recruited into organ primordia or into stem tissue during stem elongation(Bowman and Eshed, 2000; Clark, 1997; Fletcher, 2002; Gallois et al., 2002; Long et al., 1996). A balance between maintenance of the SAM and recruitment of cells into organ primordia is crucial for precise patterning. In Arabidopsis thaliana, class I KNOX family homeodomain transcription factors together with ASYMMETRIC LEAVES 1 (AS1) and ASYMMETRIC LEAVES 2 (AS2)have key roles in regulating the balance between SAM maintenance and organogenesis (Byrne et al.,2000; Byrne et al.,2002; Lin et al.,2003; Long and Barton,1998; Long et al.,1996; Ori et al.,2000; Semiarti et al.,2001). In the meristem, the KNOX gene SHOOT MERISTEMLESS (STM) represses AS1 and AS2 to prevent differentiation and maintain meristem fate, whereas KNOXgenes are repressed by AS1 and AS2 in lateral organs,allowing for their elaboration. Genes expressed at the boundary between the SAM and lateral organs may also play a role in regulating the balance between these two reciprocally repressing domains and might function in boundary differentiation.

The boundary region between organ primordia and the SAM is established early during organ initiation (Aida and Tasaka, 2006; Breuil-Broyer et al., 2004; Hussey,1971). Cells in the boundary have lower rates of cell division and are smaller than cells in either of the adjacent domains, morphological differences that are apparent soon after primordium initiation(Gaudin et al., 2000; Hussey, 1971). Boundary cells are thought to serve as a barrier to isolate populations of cells with distinct identities (Aida and Tasaka,2006). In the shoot, boundaries located in leaf axils have the potential to form axillary meristems, making them important regulators of plant architecture. Despite the importance of boundaries, a clear understanding of the molecular mechanisms that control their identity and function is lacking.

A number of genes that are expressed specifically in organ boundaries have been described. Members of the BLADE ON PETIOLE (BOP), CUP-SHAPED COTYLEDON (CUC), LATERAL ORGAN BOUNDARIES (LOB), LATERAL SUPPRESSOR (LAS) and REGULATORS OF AXILLARY MERISTEMS (RAX) families show boundary-specific expression patterns in Arabidopsis and other plant species (Aida et al., 1997; Aida and Tasaka, 2006; Borghi et al., 2007; Greb et al., 2003; Ha et al., 2003; Keller et al., 2006; Müller et al., 2006; Schmitz and Theres, 2005; Shuai et al., 2002). These genes, which encode several different classes of transcription factors, act largely redundantly to control boundary formation during embryogenesis and post-embryonic development. A number of boundary genes have also been implicated in the control of axillary meristem formation.

The combined activities of the CUC genes, which encode NAC-domain transcription factors, function to control embryonic meristem formation and cotyledon separation (Aida et al.,1997; Hibara et al.,2006; Souer et al.,1996; Vroemen et al.,2003; Weir et al.,2004). CUC genes have been proposed to function as inhibitors of cell growth, an activity required for organ separation at boundaries. CUC genes also control meristem initiation via regulation of STM (Aida et al.,1997; Aida et al.,1999; Takada et al.,2001). LAS and the RAX genes are required for axillary meristem formation, in part via regulation of STM expression(Greb et al., 2003; Keller et al., 2006). Double mutant analyses uncovered synergistic interactions demonstrating that CUC genes are also required during post-embryonic organ separation and axillary meristem formation, whereas LAS contributes to both organ separation and axillary meristem formation(Hibara et al., 2006). Mutations in STM cause defects in meristem maintenance and organ separation,providing further evidence of a connection between organ separation and SAM formation (Kanrar et al.,2006; Long et al.,1996). In addition, mutants that are compromised for the functions of PENNYWISE (PNY) and POUND-FOOLISH(PNF), which encode KNOX-interacting BELL homeodomain proteins,exhibit pedicel-stem fusions (Kanrar et al., 2006; Smith et al.,2004). Thus, organ separation may be closely connected to meristem identity and maintenance.

We report here the identification of two new components of organ boundary formation. LATERAL ORGAN FUSION 1 (LOF1) and LATERAL ORGAN FUSION 2 (LOF2) encode MYB transcription factors that function in both lateral organ separation and axillary meristem formation, in part through interaction with CUC2, CUC3 and the KNOX gene STM.

Enhancer- and gene-trap lines in the Landsberg erecta(Ler) ecotype are from the Cold Spring Harbor collection(Sundaresan et al., 1995) and were identified based either on their β-glucuronidase(GUS) expression pattern (P.S.S. and R. Martienssen, unpublished results), or on the reported insertion site near LOF1 and LOF2. To determine the site of Ds element integration,TAIL-PCR was performed as described(Rojas-Pierce and Springer,2003). The DsE element in ET4016 was integrated at position -5308 and the DsG element in GT12154 was integrated at position -1363 relative to the start codon of LOF1. The DsEelements in ET101 and ET5681 were integrated at positions -3377 and -413,respectively, relative to the start codon of LOF2.

T-DNA insertion lines were isolated from the Salk Institute Genomic Analysis Laboratory collection (Alonso et al., 2003) and are in the Columbia (Col) ecotype. Homozygous mutants of lof1-1 (Salk_025235) and lof2-1 (Salk_064076)were selected by genomic DNA gel blot analysis and PCR-based genotyping;primers sequences used are available upon request.

For plant transformation, binary plasmids were transformed into Agrobacterium tumefaciens strain GV3101(Nagel et al., 1990) and plants were transformed using the Agrobacterium-mediated floral dip method (Clough and Bent,1998). pLOB:GUS(Shuai et al., 2002), pSTM:GUS (Kirch et al.,2003), stm-10 (Kanrar et al., 2006), cuc1-13, cuc2-3 and cuc3-105(Hibara et al., 2006) were in the Col background.

The LOF1 coding sequence was amplified from a Col inflorescence cDNA sample. Primers contained introduced restriction sites for cloning and were designed to amplify the full coding sequence, predicted based on comparisons with related MYB-gene sequences. The resulting PCR product was sequenced to confirm the 1077 nt open reading frame, and subsequently subcloned into the binary vector pCL0011, containing the Cauliflower mosaic virus 35S promoter(Lin et al., 2003) to generate p35S:LOF1.

The LOF2 promoter region (-3795 from ATG to +53) was amplified from Col genomic DNA, sequenced to confirm its integrity, and cloned into the binary vector pCB308 (Xiang et al.,1999) using introduced SmaI sites to create an in-frame fusion to the GUS gene. Primer sequences are available upon request.

Plants were grown on soil at 23°C under either long-day (16 hours light/8 hours dark) or short-day (8 hours light/16 hours dark) conditions. To examine early paraclade development, plants were grown on soil in short days for 30 days, then transferred to long days to induce flowering. After 2, 4, 6 and 8 long days, tissue was harvested and embedded for sectioning. Examination of the number and length of accessory shoots and anatomical analyses were carried out on 30- or 40-day-old plants grown in soil under long-day conditions. All single- and double-mutant combinations containing lof1 and stm-10 or cuc mutations were grown under long-day conditions. To assess the potential for accessory shoot formation,primary or axillary inflorescence stems of 40-day-old plants were removed and elongated accessory shoots were measured 2 weeks later.

To examine the transcript levels of boundary genes in paraclade junctions,total RNA was isolated from a 0.5 cm region spanning the paraclade junction. cDNAs were synthesized as previously described(Lin et al., 2003) and PCR reactions were done using gene-specific primers (sequences are available upon request). Two biological replicates were conducted. RT-PCR products were detected by blotting and probing with gene-specific probes after 20 cycles of amplification. ACT2 was used as a control with 15 cycles of amplification.

For detection of GUS activity, samples were incubated in staining solution overnight (except for ET4016, which was stained for 6 hours), then transferred to 70% (v/v) followed by 100% (v/v) ethanol to clear chlorophyll(Geisler et al., 2002). Tissue was embedded using a modified in situ hybridization protocol(Lee et al., 2003; Long and Barton, 1998). Stained tissue was transferred to FAA (50% [v/v] ethanol, 5% [v/v] acetic acid, and 3.7% [v/v] formaldehyde) for 10 hours, then dehydrated and embedded. Images were captured under dark field or DIC optics using a Leica DMR microscope. For anatomical analyses, plants were fixed in FAA fixation solution for 10 hours and embedded for sectioning.

Enhancer-trap line ET4016 exhibited GUS reporter activity in the boundary regions between the SAM and lateral organs during vegetative development(Fig. 1A,B). GUS activity was first detected in 4-day-old ET4016 seedlings and persisted throughout leaf development in a band of cells on the adaxial side of rosette leaf bases(Fig. 1B,G,H). No GUSexpression was detected in roots. After flowering, GUS activity was detected at the base of floral organs, in the adaxial junction between the primary stem and pedicels, in the adaxial junction between the primary and axillary stems,in the junction between the axillary stems and cauline leaves(Fig. 1C-F) and in the junctions between the inflorescence meristem and flower primordia(Fig. 1I,J). In cauline leaf axils, GUS activity was detected both before and after axillary meristem initiation (Fig. 1L-O). In all cases, GUS activity was restricted to adaxial boundary regions and was detected in both epidermal and underlying cells(Fig. 1G-O).

The ET4016 transposant contained a single DsE element (data not shown) inserted 5308 bp upstream of the translation start codon of AtMYB117 (At1g26780). Based on the loss-of-function phenotype (see below) we named this gene LATERAL ORGAN FUSION 1(LOF1). Attempts to detect LOF1 transcripts using in situ hybridization were unsuccessful, perhaps due to low transcript abundance. A second transposant line, GT12154, containing a DsG element 1363 bp upstream of the translation start codon of LOF1 showed a GUSexpression pattern virtually identical to that of ET4016 (data not shown),suggesting that the enhancer trap line faithfully reports the expression of LOF1. In a survey of dissected tissues, LOF1 transcripts were only detected in RNAs isolated from nodes (pedicel and paraclade junctions; see Fig. 1P) and inflorescence apices (Fig. 1Q),consistent with the pattern of GUS expression in ET4016 and GT12154. Because both Ds insertions were between LOF1 and At1g26770, we also examined the developmental expression profile of At1g26770 reported in AtGenExpress(Schmid et al., 2005). At1g26770 transcripts were broadly detected and present in hypocotyls, cotyledons and leaves, a pattern inconsistent with the GUS expression patterns in ET4016 and GT12154, suggesting that these lines accurately represent the expression of At1g26780/LOF1.

cDNA sequencing confirmed that LOF1 produces a transcript encoding two predicted MYB DNA-binding domains and an FxDFL motif of unknown function that is common to seven Arabidopsis R2R3 MYB transcription factors in subgroup 21 (Stracke et al.,2001) (see Fig. S1 in the supplementary material).

We identified a LOF1 T-DNA insertion mutant, lof1-1 (see Fig. S2 in the supplementary material), that showed a significant reduction in transcript accumulation (Fig. 2J; and data not shown). When grown under our standard long-day growth conditions, lof1 plants were morphologically normal during vegetative development, but after flowering, the axillary stems produced in cauline leaf axils of the primary inflorescence (the paraclade junction; see Fig. 1P) were deflected downward compared with the wild type and the axillary stem and subtending cauline leaf were fused at their bases(Fig. 2A-D). lof1mutants also lacked accessory shoots, which wild-type plants produced in the junctions between the axillary stems and cauline leaves(Fig. 2E,F). lof1axillary branches eventually grew upward, perhaps in response to phototropism or gravitropism (Fig. 2F). When grown in short-day conditions, the lof1 fusion was more severe;cauline leaves were fused to stems along a greater extent and mutants produced clusters of leaves resembling aerial rosettes that were not present in wild-type plants (Fig. 2G-I).

LOF1 transcript levels were slightly reduced in lof1vegetative shoots and the inflorescence apex compared with wild type, but undetectable in both pedicel nodes and paraclade junctions. The dramatic reduction in transcript abundance in lof1 paraclade junctions correlates with the location of visible defects in lof1 mutants. The lack of apparent phenotypes in other organ junctions where LOF1 is expressed may be due to the presence of LOF1 transcripts in those locations, or alternatively, could be due to overlapping functions of LOF1 with redundant genes.

Following synchronous induction of flowering (see Materials and methods),we examined early axillary branch development by histological analyses. No significant differences between wild-type and lof1 plants were observed during the early stages of axillary meristem initiation, but soon after axillary stem elongation was observed, fusion between the axillary branch and cauline leaf was visible in lof1. At this stage, cells in the paraclade junction of wild-type plants were small, arranged in well-organized cell layers, and dividing cells were rarely observed(Fig. 2K,L). By contrast, lof1 cell files were disrupted and enlarged cells were observed,especially near the vascular bundles (Fig. 2M,N). No evidence of epidermal cell identity was apparent in the internal fused region. Thus, lof1 mutants exhibit defects in cell division and expansion at the boundary between cauline leaf and axillary branch, which result in failure of the two structures to properly separate.

To confirm that the lesion in LOF1 was responsible for the observed phenotypes in lof1, we introduced p35S:LOF1 into Col and lof1 backgrounds. In Col, 32% (44/139) of p35S:LOF1plants exhibited a range of abnormal phenotypes, including leaves that were asymmetric, serrated or curled upward at the margin (compare Fig. 2O with 2P). In general, transgenic plants had reduced organ size, which resulted in the formation of open flower buds (Fig. 2Q). In the lof1 mutant background, p35S:LOF1 resulted in the same range of phenotypes as in wild type and also rescued both the fusion and accessory shoot defects. Seventy-five percent (24/32) of lof1 plants transformed with the p35S:LOF1 construct lacked observable fusion between the axillary stem and cauline leaf and produced accessory shoots(compare Fig. 2R with 2S). The remaining 25% (8/32)of lof1 transgenic plants failed to form accessory shoots and exhibited some degree of fusion. The observed phenotypes in p35S:LOF1plants suggest a role in growth limitation, consistent with the loss-of-function lof1 phenotype. In addition, the phenotypic rescue conferred by p35S:LOF1 indicates that the mutation in LOF1is responsible for the observed fusion defects.

Phylogenetic analyses of the MYB proteins in subgroup 21(Stracke et al., 2001),revealed that LOF1 is most closely related to At1g69560/MYB105, which we call LOF2 (see Fig. S3 in the supplementary material). We therefore identified transposon and T-DNA insertions in LOF2 to examine possible functional overlap with LOF1. The enhancer-trap lines ET101 and ET5681 contained a DsE element inserted upstream of LOF2 (see Fig. S4 in the supplementary material). In both lines and in transgenic plants harboring a pLOF2:GUS construct, GUS expression was observed in organ boundaries (Fig. 3A-C; see Fig. S5 in the supplementary material). In early-stage flowers however, GUS activity was detected broadly throughout the primordia, including the adaxial boundaries (Fig. 3B), slightly overlapping the LOF1 domain (see Fig. S5 in the supplementary material). Later in development, GUS activity was restricted to the abaxial pedicel junctions, the base of floral organs, the abaxial base of cauline leaves, and the junction between cauline leaves and axillary stems(Fig. 3A-C). Thus LOF1and LOF2 expression had limited overlap early in development, and later expression of the two genes was distinct.

We identified a T-DNA mutant, lof2-1, which appears to be a null allele (data not shown). lof2 plants were phenotypically normal, but lof1 lof2 double mutants exhibited additional phenotypes compared with the lof1 single mutants. Fusion between the flower pedicel and the primary inflorescence stem was observed in 11% (11/100) of lof1 lof2 double mutants (compare Fig. 3D with 3E,F) and the cauline leaf was extended below the point of insertion and fused to the inflorescence stem forming a decurrent strand in 28% (28/100) of lof1 lof2 plants(compare Fig. 3G with 3H,I). These additional phenotypes implicate LOF2 in organ-boundary establishment, and suggest that LOF1 and LOF2 have functional overlap.

We further analyzed lof1 lof2 double mutants by sectioning. In the wild type, sections above the pedicel-stem junction revealed two well-separated structures with intact epidermal layers(Fig. 3J), whereas lof1 lof2 double mutants showed fusion between the pedicel and stem, with a continuous cortical cell layer (Fig. 3K,L). Cells within the fused junction were larger than cortical cells in other regions, and abnormalities in vascular pattern were observed. In wild-type plants, the vascular trace extending from the stem into the pedicel was well separated from the adjacent vascular bundle in the stem and showed clear asymmetry, with internal xylem and external phloem(Fig. 3J,M). In the lof1 lof2 double mutants, the pedicel vascular trace formed but was not well separated from the primary vascular bundle(Fig. 3N) and displayed an abnormal amphivasal (radialized) pattern with xylem surrounding phloem(Fig. 3N,O). No vasculature was visible in the decurrent strands, which exhibited continuity between cortical layers of stem tissues and cauline leaf mesophyll(Fig. 3Q,R). Radialized vasculature is a common hallmark of adaxial-abaxial polarity defects(Eshed et al., 2001; Kerstetter et al., 2001; McConnell and Barton, 1998);however, no other polarity defects were apparent in lof1 or lof1 lof2 plants.

The primary inflorescence stem of wild-type and lof2 plants contained an average of three axillary shoots with subtending cauline leaves,and on average two of three axils contained a visible accessory shoot(Fig. 4J). Accessory shoots were never observed in lof1 single mutants or lof1 lof2double mutants (Fig. 4J). In histological sections, accessory buds were clearly visible in the junctions between the axillary stem and cauline leaves of wild-type plants(Fig. 4A,D,G), whereas in lof1 and lof1 lof2 plants, enlarged cells were present and there was no evidence of meristem cells(Fig. 4B,C,E,F,H,I).

We also tested whether lof1 mutants had the potential to form accessory shoots when relieved of apical dominance. Primary or axillary inflorescence apices were removed from 40-day-old plants and accessory shoot formation was evaluated 2 weeks later. In wild-type plants, removal of axillary but not primary inflorescence apices resulted in a dramatic outgrowth of accessory shoots, which do not normally elongate significantly(Fig. 4K; see Fig. S6 in the supplementary material). Thus, apical dominance originating from the axillary inflorescence normally suppresses accessory shoot outgrowth. Accessory shoots were not induced in lof1 mutants by removal of either primary or axillary inflorescence apices, which is consistent with a complete lack of accessory shoot meristem formation in these mutants.

A number of mutations that affect the formation of axillary meristems have been described, including those in the LAS, RAX and CUCfamilies (Greb et al., 2003; Hibara et al., 2006; Keller et al., 2006; Müller et al., 2006). A common feature of these mutants is that expression of the meristem marker STM is reduced. To examine the relationship between STM and the accessory shoot defects in lof1, we generated lof1mutants expressing a GUS reporter driven by the STM promoter(Kirch et al., 2003). GUS activity was observed in paraclade junctions of pSTM:GUS plants before accessory shoots were visible and persisted in accessory shoot apices after their initiation (Fig. 5A,B). By contrast, no GUS activity was detected in the paraclade junctions of lof1 mutants (Fig. 5C), although GUS expression was unaltered in the SAM,inflorescence meristem and floral meristem (data not shown). These data indicate that cells with meristem identity were lacking in the paraclade junctions of lof1 mutants.

We examined the expression of boundary gene markers LOB and LOF2 in lof1 mutants using promoter:GUS fusion constructs. pLOB:GUS (Shuai et al., 2002) was expressed in wild-type organ boundaries in a pattern similar to that of LOF1(Fig. 5D). In the lof1background, GUS activity was detected in a normal pattern, although slightly expanded into the region of fusion (Fig. 5E). By contrast, pLOF2:GUS expression was significantly reduced in lof1 paraclade junctions compared with wild type(Fig. 5F,G), suggesting that LOF2 is downstream of LOF1, although this regulation may be indirect.

To further characterize gene expression in lof1 mutants,transcript levels of several boundary genes were examined in dissected wild-type and lof1 paraclade junctions by RT-PCR (see Fig. S7 in the supplementary material). Transcript levels of LAS, RAX1, LOF2, CUC1,CUC2 and CUC3 were reduced in lof1 paraclade junctions,whereas LOB and REV transcript levels were unaffected and BOP1 transcript levels appeared to be slightly elevated. These data indicate that the lof1 mutant has defects in boundary gene expression and place LOF1 upstream of RAX1, LAS and the CUCgenes.

Because of their demonstrated role in boundary specification, we investigated interactions between LOF1 and the CUC family genes (Hibara et al., 2006). We generated double mutants between lof1 and each of the cucmutants. None of the single or double mutants had visible phenotypes during vegetative development. After flowering, the lof1 cuc2 and lof1 cuc3 double mutants displayed severe phenotypes. Pedicels were fused to the primary stem (Fig. 6A-F,T;see Table S1 in the supplementary material) and the plants exhibited more dramatic fusion in paraclade junctions compared with lof1 plants(Fig. 6G-N,U; see Table S2 in the supplementary material). The double mutants exhibited additional fusion between primary and axillary stems and between cauline leaves and primary stems (decurrent strand) (Fig. 6J,K,M,N). These data indicate overlapping roles for LOF1,CUC2 and CUC3 to control organ separation during reproductive development. lof1 cuc2 and lof1 cuc3 double mutants also exhibited axillary meristem defects. In lof1 cuc3 double mutants, 19%of axillary meristems on the primary inflorescence stem were replaced by a solitary flower (Fig. 6O,P,V;see Table S3 in the supplementary material) and 85% of cauline leaf axils on secondary/axillary branches were either barren or contained a solitary flower(Fig. 6Q-S,W; see Table S4 in the supplementary material). Similar defects were observed at lower frequencies in cuc3 and lof1 cuc2 plants (and to a lesser degree the other single mutants). Thus, LOF1 and CUC3contribute to axillary meristem maintenance/initiation.

Having established a link between LOF1 and meristem formation, we examined possible interactions between LOF1 and STM by combining lof1 with the weak stm-10 allele. stm-10single mutants failed to maintain a SAM, stalling after formation of two or three rosette leaves (Fig. 7B; Table 1), but eventually producing additional leaves and ultimately bolting to produce abnormal flowers lacking a central carpel (Fig. 7H; and data not shown). As previously reported(Kanrar et al., 2006), some fusions were observed in mature stm-10 plants. lof1 stm-10double mutants had more severe meristem defects than stm-10 - 16% of double mutants produced no leaves and 9.8% arrested after forming a single rosette leaf (Fig. 7C; Table 1). Thus, 26% of lof1 stm-10 double mutants exhibited dramatic defects in meristem maintenance. The remainder of double mutants terminated earlier than the stm-10single mutant (Fig. 7H).

lof1 stm-10 double mutants also displayed organ fusion during vegetative development - 44% of 16-day-old lof1 stm-10 plants exhibited fusion between rosette leaf petioles(Fig. 7D; Table 1). This phenotype was never observed in lof1 or in 16-day-old stm-10 plants,although older (30-day-old) stm-10 plants occasionally produced some type of rosette leaf fusion (data not shown). Analyses using cleared whole mounts or sectioning revealed the presence of two distinct vascular bundles in the fused petioles, both in the fused basal region and in the unfused apical region (Fig. 7E-G), indicating that the organ-boundary defects did not affect vascular bundle formation. Double mutant plants that did not arrest bolted after about 2 months, but terminated prematurely compared with stm-10 single mutants(Fig. 7H). Some double mutants eventually produced cauline leaves and axillary inflorescence branches that exhibited severe fusions between the primary stem and pedicels and between the axillary stems and cauline leaves compared with stm-10 (data not shown). These data indicate that LOF1 functions in both meristem maintenance and organ separation.

LOF1 and LOF2 function redundantly to regulate boundary formation. The domains of LOF1 and LOF2 expression overlap in boundaries between the inflorescence meristem and young floral primordia,between the axillary stem and cauline leaves, and at the base of floral organs, but diverge later in development. LOF2 transcripts are reduced in lof1 mutants, indicating that LOF1 contributes to LOF2 expression. The lof1 mutant displays two phenotypes consistent with a function in organ boundaries: organ fusion and reduced axillary meristem activity. Fusion between the axillary stem and cauline leaf in lof1 mutants is a novel phenotype that, to our knowledge, has not been reported for other boundary mutants. Additional fusions - between stem and pedicels, primary and axillary stems, primary stem and cauline leaves, and rosette leaf petioles - that were observed in lof1 double mutant combinations indicate that LOF1 functions broadly in organ separation. In addition, LOF1 has a role in meristem initiation or maintenance, as lof1 mutants failed to produce accessory meristems and enhanced the meristem defects of cuc3 and stm-10mutants.

Mutations in a number of boundary-specific genes result in organ fusion. During embryogenesis, the CUC genes function redundantly to control organ separation, such that single mutations in CUC1, CUC2 or CUC3 cause only subtle defects, whereas any combination of two cuc mutations results in fusion along the cotyledon margins and a dramatic cup-shaped cotyledon phenotype(Aida et al., 1997). CUC2 and CUC3 also function to specify boundaries during post-embryonic development and have roles in axillary meristem formation(Hibara et al., 2006; Raman et al., 2008). The fusion phenotypes in the cuc mutants might result from both organ-organ and organ-meristem fusions(Aida and Tasaka, 2006). Organ-organ fusions may give rise to cup-shaped cotyledons, whereas organ-meristem fusions may generate cotyledon petiole fusions and could also affect meristem maintenance or initiation by alteration of KNOX gene expression (Aida et al.,1999).

lof1 enhanced the weak stm-10 mutant, a somewhat surprising finding given that GUS expression was not detected in embryos of either ET4016 or GT12154, which report LOF1 expression. Although we cannot exclude a low level of LOF1 expression in the embryo, a more likely explanation is that LOF1 has a function in post-embryonic SAM maintenance. Consistent with this idea,meristem-maintenance defects were observed in lof1 stm-10 plants later in development. In addition, whereas lof1 enhanced cuc2 and cuc3, the double mutants did not show cotyledon fusion or meristem arrest, which might be indicative of embryonic function. Similarly, las cuc double mutants did not show enhanced cotyledon fusion (Hibara et al., 2006; Raman et al., 2008). These data suggest that LOF1 and LAS do not significantly contribute to cotyledon separation during embryogenesis.

CUC2, CUC3 and LAS function redundantly in organ separation. Although no single mutant displays organ fusions during vegetative development, cuc2 cuc3 and cuc3 las double mutants produce rosette-leaf fusions (Hibara et al.,2006). By contrast, lof1 cuc double mutants did not show vegetative organ fusion. In lof1 stm-10 double mutants, petioles were fused on their adaxial faces rather than laterally and the meristem was apically shifted to the site of petiole separation. These phenotypes may indicate that the lof1 stm-10 fusions were between the meristem and forming leaves, which could result in an apical shift in meristem position. The observed meristem loss in some lof1 stm-10 plants is also consistent with fusion events involving the meristem. Together our data suggest that LOF1 probably does not play a major role in organ-organ separation during vegetative development, but does contribute to organ-meristem separation.

The CUC genes and LAS also function in organ separation during reproductive development. las single mutants occasionally exhibited fusion between primary and axillary stems under short-day growth conditions (Greb et al.,2003). Although the day-length dependence of these phenotypes is not understood, it is noteworthy that the lof1 phenotype was also more severe under short-day conditions, suggesting a day-length-dependent aspect to boundary formation. Combining any two of the mutants cuc2,cuc3 and las results in fusion between the primary and axillary stem and/or between the stem and pedicels, indicating that these three genes function redundantly to control organ separation(Hibara et al., 2006). LOF1 and LOF2 also function in organ separation during reproductive development - lof1 single mutants have a novel fusion between the axillary stem and cauline leaf, and additional fusions resulted when lof1 was combined with lof2, cuc2 or cuc3. By contrast, LOB appears to function in a pathway distinct from LOF1 and LOF2, despite the similar expression pattern. LOB expression was unaffected in the lof1 mutant background and genetic interactions were not observed in lof1 lob or lof2 lob double mutants (data not shown).

Why do the boundary mutants have defects in organ separation? Examination of Arabidopsis cuc1 cuc2 mutants and the petunia mutant nam,which both have fused cotyledon petioles, revealed that meristem cells were replaced by enlarged cells (Aida et al.,1997; Souer et al.,1996). Similar defects were observed in lof1 - boundary cells were enlarged and exhibited abnormal cell division patterns. The enlarged cells in lof1 boundaries may prevent or delay separation of the axillary branch and the subtending cauline leaf. Aberrant cell expansion also coincided with a downward deflection of axillary stems in lof1.Thus, changes in cellular organization are a common feature of mutants that exhibit fusion defects.

lof1 mutants fail to form accessory meristems, have no detectable expression of the meristem marker STM in the paraclade junction, and reduce transcript accumulation of many, but not all, boundary markers. The relationship between defects in axillary meristem formation and organ fusion is not entirely clear, however, as the two phenotypes are not always correlated. las and rax mutants produce very few axillary meristems, but do not exhibit fusion (Greb et al., 2003; Keller et al.,2006; Müller et al.,2006), whereas loss-of-function lob mutants exhibit fusion between cauline leaves and axillary stems, but produce normal accessory shoots (W. C. Lin and P.S.S., unpublished data). Thus, LOF1 function in axillary meristem initiation or maintenance may be independent of its function in organ separation. The ability of lof1 to enhance the weak stm-10 phenotype also supports a role for LOF1 in meristem initiation or maintenance. It is possible that LOF1 acts in the boundary to regulate the activity of meristem genes. Alternatively, the boundary may indirectly contribute to meristem function, perhaps by acting as a physical barrier.