Phyllotactic pattern and stem cell fate are determined by the Arabidopsis homeobox gene BELLRINGER

The plant shoot is derived from a group of undifferentiated stem cells within the shoot apical meristem (SAM). Lateral organs such as leaves arise on the flanks of the SAM by the coordinated repression of stem cell fate,determination of founder cells, and elaboration of the incipient primordium. Early events specify the spatial and temporal positioning of leaf and flower primordia and ultimately establish shoot phyllotaxy (Greek, `leaf order'). Later events define organ patterning along proximodistal, dorsoventral and mediolateral axes (reviewed by Kidner et al., 2002).

Initiation of successive lateral organs on the flanks of the SAM proceeds in predictable patterns generating a phyllotaxy. Spiral phyllotaxy observed in the vast majority of plants derives from the regular initiation of successive lateral organ primordia at a constant divergence angle approximating 137.5°. This pattern can be described in terms of two sets of opposing intersecting spirals or `contact parastichies' that connect adjacent primordia(Richards, 1951; Steeves and Sussex, 1989). The number of clockwise and counterclockwise spirals in each set is characteristic of a species, and corresponds to successive numbers in the Fibonacci series(3+5, 5+8, 8+13 etc.) (reviewed by Richards, 1951). Inter-specific and developmental variation in the number of contact parastichies is thought to reflect relative rates of primordia initiation and growth (Richards, 1951).

Genetic pathways that establish and maintain phyllotaxy are yet to be identified. Potentially the site of primordium initiation is established by molecular inhibitory signals from preexisting primordia or from biophysical stresses within the shoot apex (reviewed in Green, 1999). Consistent with both of these models alterations in phyllotaxy are often associated with changes in the dimensions and organization of the SAM. For example, in the Arabidopsis mutants fasciata1 (fas1) and fasciata2 (fas2) meristem shape and patterning is irregular and phyllotaxy is disrupted (Kaya et al.,2001; Leyser and Furner,1992). The pattern of organ initiation is also irregular in the serrate (se) mutant where the meristem is enlarged compared with wild type (Clarke et al.,1999; Ori et al.,2000; Prigge and Wagner,2001). Reduced meristem size is associated with phyllotaxy defects in the DNA topoisomerase mutant top1α(Takahashi et al., 2002). Alternatively, the abphyl1 mutant in maize has a phyllotaxy defect associated with an enlarged meristem. Additional leaves are formed in a decussate pattern rather than the normal alternate, distichous arrangement(Jackson and Hake, 1999). In contrast to these examples, terminal ear (te1) mutations in maize alter the rate of leaf initiation and phyllotaxy, but have relatively minor effects on the geometry of the shoot apex(Veit et al., 1998). te1 encodes a protein related to the RNA binding protein Mei2 from Schizosaccharomyces pombe. te1 is expressed in the presumptive internode in a region around the meristem excluding the site of organ initiation. This expression pattern suggests te1 may inhibit lateral organ differentiation.

One genetic pathway required for SAM function involves class 1 KNOX homeobox transcription factors. Recessive mutations in the Arabidopsis KNOX gene SHOOT MERISTEMLESS (STM) and in the closely related maize gene knotted1 have defects in meristem function indicating a requirement for SAM initiation and/or maintenance(Barton and Poethig, 1993; Long et al., 1996; Vollbrecht et al., 2000; Vollbrecht et al., 1991). Consistent with a role in SAM function both STM and kn1 are expressed in vegetative and reproductive SAMs but are down-regulated in founder cells and lateral organ primordia(Jackson et al., 1994; Long et al., 1996). STM maintains stem cell fate by negative regulation of the myb domain transcription factor ASYMMETRIC LEAVES1 (AS1) and a member of the LOB-like transcription factor family ASYMMETRIC LEAVES2(AS2) (Byrne et al.,2000; Byrne et al.,2002; Iwakawa et al.,2002; Shuai et al.,2002). AS1 is related to rough sheath2 in maize and PHANTASTICA in Antirrhinum. All three genes are expressed in lateral organ primordia where they function as negative regulators of KNOX genes (Byrne et al.,2000; Ori et al.,2000; Semiarti et al.,2001; Timmermans et al.,1999; Tsiantis et al.,1999).

Arabidopsis has three additional class I KNOX genes, BREVIPEDICELLUS (BP, formerly KNAT1), KNAT2 and KNAT6. Like STM, these genes are expressed in SAMs and downregulated in lateral organs, although the pattern and timing of expression differs from that of STM(Byrne et al., 2002; Dockx et al., 1995; Lincoln et al., 1994; Pautot et al., 2001; Semiarti et al., 2001). Mutations in BP alone do not cause meristem defects(Byrne et al., 2002; Douglas et al., 2002; Venglat et al., 2002). BP is, however, redundant with STM and has a role in SAM function in as1 stm and as2 stm mutant backgrounds(Byrne et al., 2002). Disruption of KNAT2 gene expression has no phenotypic effect,probably because of redundancy with the duplicate gene KNAT6(Byrne et al., 2002).

We have isolated several insertion alleles of BELLRINGER (BLR), a BELL1-like homeobox gene. Prominent defects in blr mutants include an increase in the number of leaves and disruption to the normal spiral pattern of primordia initiation. Genetic interactions also demonstrate that BLR is required for stem cell maintenance. Previously we reported that BP plays a role in meristem function in the absence of AS1 and STM (Byrne et al., 2002). BLR is also necessary for meristem function in the absence of AS1 and STM. BLR probably interacts directly with STM and BP in meristem function.

Mutant alleles, blr-1 (GT7797) and blr-2 (ET6411), were obtained from gene trap and enhancer trap lines generated as previously described (Springer et al.,1995; Sundaresan et al.,1995). The blr-3 allele was obtained from the The Salk Institute Genomic Analysis Laboratory (SIGnAL) collection. bel1-1, wus-1,clv1-1 and clv3-2 were obtained from the ArabidopsisBiological Resource Center (ABRC). Other mutants (as1-1, stm-1, stm-11,stm-2, and bp-2) were obtained as described previously(Byrne et al., 2002). All mutants other than blr-3 were in a Landsberg erectabackground. Plants were grown either on soil or on MS medium supplemented with sucrose, with a minimum day length of 16 hours.

Tests for allelism were carried out by crossing plants homozygous for blr-2 and blr-3 to plants homozygous for blr-1. F₁ plants from both crosses displayed the blr mutant phenotype. All other genetic interactions and phenotypic studies were carried out with the blr-1 allele in a Landsberg erecta background. To generate double mutants, plants homozygous for bp, as1, bel1, clv1or clv3 were crossed to plants homozygous for blr. All double mutants segregated in the F₂ progeny in a 1:15 ratio. Double blr wus, blr stm-11 and blr stm-2 mutants were generated by crossing plants homozygous for blr to plants heterozygous for wus, stm-11 or stm-2. Double blr wus mutants segregated 1:3 in F₃ seed from blr mutant plants. In lines carrying either stm-11 or stm-2 F₃ seed from homozygous blr plants segregated 1:3 stm mutants.

Triple blr as1 stm-11 mutants were generated by crossing double homozygous blr as1 plants to as1/as1 stm-11/+ plants. F₂ seed from blr/blr as1/as1 stm11/+plants segregated a meristemless phenotype in a 1:3 ratio. Triple bel1 as1 stm-11 mutants were generated by crossing as1/as1 stm-11/+ plants with bel1/bel1. F₃ seed from plants homozygous for as1 and heterozygous for stm11 and bel1 segregated only wild type, as1 stm11 and as1 bel1 phenotypes in a 9:4:3 ratio.

DNA extraction and manipulation were carried out using standard protocols(Sambrook et al., 1989). The Ds element copy number in lines carrying blr-1 and blr-2 was determined using Southern gel blot analysis as described previously(Vongs et al., 1993). The Ds-specific hybridization probe was obtained by PCR amplification of the Ds element using the primers agcccatgtaagaaatacctagcg and tgctgtactgctaagtgctgtgag. Identification of the tagged gene in blr-1and blr-2 was carried out by thermal asymmetric interlaced-PCR(Liu et al., 1995). To confirm the Ds element insertion sites in blr-1 and blr-2, PCR products were generated using Ds and gene-specific primers. Ds primers were acccgaccggatcgtatcggt and acggtcgggaaactagctctac. blr-1 primers were ctgctggtcaaagacatggat and tgcatgcttaattagcaagaaat. blr-2 primers were atcgtgcttcaaaaagacacc and gcagagaagaatcatcgtcgt. PCR products were sequenced using dye terminator cycle sequencing (Applied Biosystems).

Total RNA for northern and RT-PCR analysis was purified using Trizol reagent (Life Technologies). For northern hybridization, 20 μg of RNA was separated on a 1.4% gyloxal/DMSO denaturing gel. RNA was transferred to a membrane and hybridized using Ultrahyb buffer (Ambion). The BLR probe for northern analysis was a PCR product synthesized with the primers taatgtgggtcgtgggattta and aggagcatgatgatcaggaaa. RT-PCR was carried out as previously described (Byrne et al.,2002). Following DNase treatment and synthesis of complementary DNA with M-MuLV reverse transcriptase (New England Biolab) PCR reactions were performed with genespecific primers. BLR primers were as above. RBC primers were gaacaatggcttcctctatgc and cacaaggaatccacttgttgc. PCR products were subject to Southern hybridization using gene-specific probes.

The BLR::GUS construct was derived as follows. A 3.9 kb genomic fragment containing the BLR promoter was amplified from Landsberg erecta using the primers ttggcacgattctgaaacacg and ctcgccggctttgttgaaga. The product was cloned into pRITA, which contains the GUS reporter gene and NOS terminator sequences (a gift from John Bowman). The resulting plasmid, pRIP3, contains an in frame fusion of the start codon of BLR with GUS. This gene fusion fragment was cloned into a binary vector and introduced into plants using Agrobacteriumtransformation.

Inflorescences were prepared for sectioning by fixation in glutaraldehyde(2.5% in 0.1 M sodium phosphate buffer pH 7.0), dehydration through an ethanol series and infiltration with Histoclear prior to embedding in paraffin wax. All sections were 8 μm thick. Sections were cleared of paraffin wax using Histoclear, rehydrated to 50% ethanol, stained for 20 minutes in 0.1% safranin in 50% ethanol, rinsed in 70% ethanol, then stained for 3 minutes in 0.1% Fast Green in 95% ethanol. The sections were then dehydrated in 100% ethanol, and moved to Histoclear. For meristem size comparisons measurement were taken from longitudinal sections of 8 wild-type and 9 blr plants that were 23 days old.

GUS staining was carried out as previously described(Gu et al., 1998) using a substrate solution containing 100 mM sodium phosphate pH 7, 10 mM EDTA, 0.1%Triton X-100, 0.5 mg/ml 5-bromo-4-chloro-3-indolyl β-D glucuronic acid(X-Gluc), 100 μg/ml chloramphenicol, and either 2 mM or 5 mM each of potassium ferricyanide and potassium ferrocyanide. After 24-48 hours at 37°C samples were cleared in 70% ethanol at room temperature. Samples were viewed with a Leica MZ8 microscope and images captured with a Spot RT digital camera (Diagnostic instruments).

For scanning electron microscopy (SEM) inflorescences from wild-type and blr plants were first fixed overnight in 2.5% glutaraldehyde, then rinsed in 0.1 M sodium phosphate buffer and dehydrated through an ethanol series (30%, 50%, 70%, 80%, 95%, 100%) prior to critical point drying using a Tousimis Auotsamdri-815. Samples were subsequently mounted on silver tape and sputter coated with gold (Emitech K550) before viewing with an Hitachi S-3500N SEM under high vacuum and with a beam accelerating voltage of 3-5 kV. Measurements of meristem radius and organ divergence angle were derived from SEM images of 16 wild-type and 16 blr mutant that were 18 days old. Plastochron ratios were measured from 7 wild-type and 7 blrplants.

In situ hybridizations were performed using digoxigenin-labeled probes(Long and Barton, 1998; Long et al., 1996). Antisense and control sense BLR transcripts were synthesized from the plasmid pBlue028, which carries a 531 bp fragment, encompassing the 3′ end of BLR, in pBluescript. Antisense STM transcripts were synthesized from the plasmid Meri HB1 (a gift from Kathy Barton). All sections were 8 μm thick.

Full-length BLR cDNA was amplified with the primers ggtcgacgggctgatgcatacgagcct and agcggccgcatttcaattcccccatatc. The PCR product was digested with SalI and NotI cloned into the GAL4 transcriptional activation domain (TA) vector pBI-771(Kohalmi et al., 1997) forming the plasmid TA-BLR. STM cDNA was amplified with the primers acgcgtcgacgtatggagagtggttccaac and ataagaatgcggccgcccaagtataccgagaacc. The PCR product was digested with SalI and NotI and cloned into the GAL4 DNA-binding domain (DB) vector pBI-770. Other constructs, DB-BP, DB-KNAT4 and TABEL 1 were kindly provided by George Haughn and are as previously described(Bellaoui et al., 2001). All plasmids were transformed into the yeast strain pJ69A using a lithium acetate/polyethylene glycol protocol(Schiestl et al., 1993).

In a screen for mutants affecting shoot architecture, we recovered gene and enhancer trap insertions in a gene we named BELLRINGER. A third allele carrying a T-DNA insertion was also identified. All three alleles produced comparable phenotypes. Compared to wild type, blr mutants are reduced in stature and have a bushy appearance due to precocious growth of axillary branch meristems (Fig. 1A,B). There is an increase in the number of leaves and flowers(Table 1) although time to flowering is not significantly delayed (average of 16 days in wild type compared with 18 days in blr-1). In addition phyllotaxy is disrupted in these mutants such that the regular spiral arrangement of flowers on the inflorescence shoot is not strictly maintained(Fig. 1C,D). This is reflected in the relative displacement of flowers along both the radial and longitudinal axis of the stem. In the radial dimension, flowers can occur both closer together and further apart than in wild type. Inflorescence internodes are variable so that flowers occur at irregular intervals along the stem. blr mutants also have shorter siliques and ovary septum fusion defects (data not shown). The severity of these phenotypes is dependent on the genetic background and growth conditions. Principally the reduction in stature and loss of apical dominance are less severe under low light conditions and in a Columbia background.

Since phyllotaxy is often associated with changes in meristem size or shape, we compared meristems in wild type and blr mutants. In longitudinal section the size and shape of blr inflorescence meristems is comparable to that of wild type(Fig. 2A,B). Furthermore, the radius of blr mutant inflorescence meristems, estimated from SEM images of 18-day-old plants, was not significantly different from that of wild type. The average radius of wild-type meristems was 24.8 μm (range 22.6-28.1 μm) while that of blr mutants was 23.6 μm (range 20.5-27.0 μm).

Spiral phyllotaxy can be described by two parameters, the plastochron ratio and the divergence angle (Richards,1951). The plastochron ratio can be used to predict the number of contact parastiches and reflects alterations in the packing of lateral organs. This parameter is derived by calculating the ratio of the distance of two successive primordia from the meristem, where distances are measured from the center of the meristem to the center of each primordium. To compare initiation patterns in wild type and mutant we measured the plastochron ratio for the youngest 4-6 primordia on inflorescence apices from 18-day-old plants. Average plastochron ratios between initiating floral primordia on blr mutant inflorescence apices were not significantly different from wild type (1.1481 versus 1.1465). In plants with spiral phyllotaxy the divergence angle between successive primordia, measured from the center of the meristem to the middle of two successive organs, is approximately 137.5°. The average divergence angle between successive primordia in wild-type inflorescence apices was 136.72° (range 121.49°-152.31°), which was comparable with most blr mutant inflorescence apices at 137.52° (range 127.54°-154.91°). However, 2 out of 16 blr mutants had significantly reduced divergence angles between successive primordia (range 79.35°-112.13°) indicating abnormal sites of floral meristem initiation (Fig. 2C,D). This was consistent with the irregular phyllotaxy observed in mature plants(Fig. 2E-H). In agreement with these quantitative parameters, the number of clockwise and counterclockwise contact parastichies was unchanged. However, the angle between spiral parastichies was increased, accommodating the extra organs in a tighter spiral than normal (Fig. 2G,H).

The recessive blr-1 mutation was found to cosegregate with a single Ds element inserted into a homeobox gene, At5g02030, located on chromosome 5(http://cshl.genetrap.org)(The Arabidopsis Genome Initiative,2000). This gene encodes a protein most closely related to members of the BELL1 (BEL1) subclass of homeodomain transcription factors, of which there are 12 members in Arabidopsis(Becker et al., 2002). BLR comprises 4 exons and 3 introns(Fig. 3A). The Ds insertion in blr-1 is located in the second intron whereas blr-2 has a Ds insertion 2 bp upstream of the BLR ATG initiation codon. In both alleles the insertion does not activate the GUS reporter on the Ds element. The third allele, blr-3, carries a T-DNA insertion in the first intron of the gene. For all three alleles full-length BLR transcripts were not detected in homozygous mutant plants(Fig. 3B).

The expression pattern of BLR in wild-type plants was examined by northern hybridization. Low levels of BLR transcript were detected in 8-day-old seedlings and in leaf tissue(Fig. 3C). Higher transcript levels were detected in inflorescence shoots, including the inflorescence meristem and flower buds.

BLR expression was further examined by driving GUS reporter gene expression from the BLR promoter. In the embryo, BLRexpression was confined to the SAM (Fig. 4A) but weak expression could also be detected in the root tip(data not shown). In young seedlings, BLR was highly expressed in the SAM and could be detected in cotyledon and leaf vasculature(Fig. 4B). In the inflorescence, BLR was expressed in the inflorescence meristem, stem,flower pedicel and in developing flowers (data not shown).

In situ hybridization was carried out to confirm the authenticity of the BLR::GUS expression pattern. BLR expression in late heart stage embryos was confined to the SAM region between the two developing cotyledons (Fig. 4C). The expression pattern is similar to that of STM(Fig. 4D)(Long et al., 1996). In addition, BLR is expressed in the inflorescence SAM in a region similar to that of STM (Fig. 4E,F). BLR expression is initially downregulated in incipient floral primordia (Fig. 4G-I). Strong expression is subsequently detected in the central region of young floral primordia and the inner whorl of developing flowers(Fig. 4E,G-I). A low level of expression is also detected at the base of the developing flowers, in the pedicel and in subepidermal and core cells of the inflorescence stem. A BLR sense probe used as a negative control did not show a signal in embryo or inflorescence tissue (data not shown).

BELL-like proteins belong to a class of homeodomain transcription factors that can interact directly with KNOX class homeodomain transcription factors(Bellaoui et al., 2001; Müller et al., 2001; Smith et al., 2002). Since protein-protein interactions between heterologous homeodomain transcription factors are required in animals (Mann and Chan, 1996) it is likely that such interactions are functionally significant in plants. We therefore investigated genetic interactions between BLR and the class 1 KNOX genes STM, KNAT1 and KNAT2.

Embryos homozygous for strong stm alleles, including stm-1 and stm-11 (Fig. 5A), lack a SAM and develop cotyledons that are fused at their base (Barton and Poethig, 1993; Clark et al., 1996; Long et al., 1996). Double blr stm-11 mutants are similar to stm-11 mutants(Fig. 5B). However, blr enhances the phenotype of the weak allele stm-2. Single stm-2 mutants germinate with slight fusion at the base of the cotyledons (Clark et al.,1996; Endrizzi et al.,1996). After a brief delay, vegetative shoot development is initiated (Fig. 5C). In contrast, blr stm-2 double mutants do not form any vegetative shoot and resemble mutants of strong alleles of stm(Fig. 5D). This genetic interaction indicates BLR is required for SAM function when there is reduced STM activity.

The blr mutant phenotype shares some characteristics with bp mutants including reduced stature and precocious outgrowth of axillary meristems (Fig. 6A). In addition, in bp mutants differentiation of subepidermal chlorenchyma extending from the abaxial side of the pedicel into the adjacent stem is disrupted (Douglas et al.,2002; Venglat et al.,2002). We analyzed blr bp double mutants to determine if there is any interaction between these two genes. Plants homozygous for both blr and bp were much reduced in stature compared with either single mutant (Fig. 6B). In other respects the double mutant has features of both blr and bp single mutants. The number of rosette and cauline leaves in blr bp mutants is comparable to that of blr (data not shown)while flower pedicels and siliques are reduced in length and generally point downward as in bp (Fig. 6A,B), and subepidermal chlorenchyma below the flower is pale relative to surrounding tissue. The phenotype of the blr bp double mutant therefore appears additive. blr knat2 double mutants are indistinguishable from blr mutants indicating that the blrphenotype is not influenced by loss of KNAT2 expression (data not shown).

To further explore the possibility of genetic interactions between BLR and BP we took advantage of the conditional role of BP in SAM function. BP maintains the SAM in as1 stmdouble mutants (Byrne et al.,2002). Vegetative development in these double mutants resembles that of as1, whereas bp as1 stm mutants lack a SAM, fail to form a vegetative shoot, and are indistinguishable from stm single mutants (Byrne et al., 2000; Byrne et al., 2002). The phenotype of blr as1 double mutants is additive in all respects(Fig. 6C). However, blr as1 stm triple mutants have severely reduced SAM function. Following germination only one or two leaves are produced(Fig. 6D). Occasionally development is resumed to form a disorganized vegetative shoot(Fig. 6E,F). BLR is therefore required for SAM function in the absence of STM and AS1.

In contrast to blr mutants where phenotypic effects are evident in the shoot, mutations in the related gene bel1 only affect ovule development (Modrusan et al.,1994; Reiser et al.,1995; Robinson-Beers et al.,1992). blr bel1 double mutants display a blrshoot phenotype and are sterile as is bel1(Fig. 6G,H). This demonstrates that BEL1 is not redundant with BLR in shoot development. Since BEL1 protein directly interacts with BP we also investigated whether BEL1 is required for SAM function in an as1 stm background. In contrast to blr as1 stm mutants, the triple bel1 as1 stmmutants form a vegetative shoot similar to as1 stm double mutants(data not shown). Furthermore no novel phenotype is detected in progeny of blr as1 plants also segregating bel1 and stm. Therefore BEL1 is not required for SAM function in these contexts.

SAM function is also regulated by WUSCHEL (WUS) and the CLAVATA (CLV) genes, acting independently of KNOX genes. WUS is a homeodomain protein expressed in inner central zone stem cells of the SAM. Mutations in WUS result in loss of meristem function. CLV1 and CLV2 are transmembrane receptors and CLV3 encodes a secreted peptide. Mutations in all three CLVgenes result in a much enlarged meristem. CLV genes maintain meristem homeostasis by limiting WUS function (reviewed by Brand et al., 2001; Clark, 2001; Fletcher, 2002). We found blr wus, blr clv1 and blr clv3 double mutants are additive(not shown). Therefore, BLR appears to affect meristem function via a KNOX genespecific pathway.

Genetic analysis demonstrated that blr enhances a weak allele of stm and is required for SAM function in the as1 stmbackground. Since BELL class proteins are known to interact directly with KNOX proteins (Bellaoui et al.,2001; Müller et al.,2001; Smith et al.,2002) one possibility is that BLR directly interacts with and affects the activity of STM and BP. To test this possibility a yeast two-hybrid assay was carried out (Fig. 7). Yeast strains carrying the plasmids TA-BLR and DBBP or DB-STM were viable in the absence of histidine. In contrast, yeast carrying TA-BRL and DB-KNAT4 failed to grow in the absence of histidine. Therefore in this system BLR interacts with the class1 KNOX proteins STM and BP but not with the class 2 KNOX protein KNAT4. The negative control yeast strain carrying the plasmid TA-BLR in the presence of the DB vector showed no growth in the absence of histidine.

BLR is a BELL class gene which, like KNOX genes, are members of the TALE homeobox transcription factor family. TALE homeodomain proteins have an atypical homeodomain, characterized by three additional amino acids between the first and second helix. Several features distinguish BELL and KNOX proteins (Becker et al., 2002; Bharathan et al., 1999; Bürglin, 1997; Reiser et al., 2000). The 12 members of the BELL class share 54% amino acid identity in the homeodomain. Likewise, the 8 KNOX proteins share 54% amino acid identity in the homeodomain. However, between these two classes amino acid conservation in the homeodomain is 32%. In addition, BELL proteins have conserved SKY and BELL domains upstream of the homeodomain, whereas KNOX proteins are defined by conserved MEINOX, ELK and GSE domains upstream of the homeodomain(Bellaoui et al., 2001; Bürglin, 1997; Kerstetter et al., 1994).

BLR is strongly expressed in the embryonic SAM, and has a conditional role in SAM function as revealed by genetic interactions. Whereas as1 stm mutants form a vegetative shoot indistinguishable from that of as1 single mutants, shoot development is greatly reduced in the triple blr as1 stm mutant. This triple mutant strongly resembles pinhead/zwille mutants where a solitary vegetative organ appears to consume the meristem (Lynn et al.,1999; Moussian et al.,1998). BLR is therefore required to maintain the SAM in the as1 stm background. Meristem function in as1 stm mutants is dependent on BP (Byrne et al.,2002), so that it is probable that BP activity is compromised in blr. We tested this possibility by making bp blr double mutants, which were additive in all respects. Thus BPis still functional in a blr mutant, meaning that loss of bpfunction cannot completely explain the blr phenotype.

Strong alleles of stm are formally epistatic to blr,although the blr phenotype is difficult to observe in strong stm. However, a weak allele of stm is greatly enhanced by blr. This effect is far stronger than that of bp, which also enhances weak stm-2 (Byrne et al.,2002). Thus it is possible that BLR is required for STM function, consistent with its expression pattern in the embryo. However, BLR must also have an STM-independent role,revealed by the strong phenotype of blr as1 stm triple mutants. A likely explanation is that BLR is required for both BP and STM function. Consistent with this possibility, yeast two-hybrid assay demonstrates the BLR protein interacts directly with both STM and BP. The requirement for BLR must be only partial, as blr has a much milder phenotype than either bp or stm, and, as noted above, BP is still functional in a blr mutant. One explanation is that BLR is itself partially redundant. A strong candidate would be the BELL class gene (At2g27990) most closely related to BLR. Conservation between these two genes is 90% within the homeodomain and 48% overall.

BLR is more distantly related to the BELL class gene BEL1. The blr bel1 double mutant phenotype indicates a lack of functional overlap between these two genes. The phenotype of bel1mutants is restricted to the ovule where the morphology of the outer integument is abnormal while the inner integument is completely absent(Modrusan et al., 1994; Robinson-Beers et al., 1992). Consistent with this phenotype BEL1 expression in the ovule is restricted to the region where integuments initiate(Reiser et al., 1995). However, BEL1 is also expressed in vegetative tissues and roots. The lack of other plant phenotypes suggests BEL1 shares genetic redundancy. Similarly, BEL1 interacts physically with BP, yet the bel1 phenotype indicates that it is not required for BP function in the inflorescence. We have demonstrated that BEL1 is not required for SAM function in as1 stm, indicating it has no effect on BP activity in the embryonic or vegetative SAM. Again the apparent lack of BEL1genetic interactions with BP is possibly the result of redundancy. Candidates for such redundancy are two genes, BLH2 and BLH4,most closely related to BEL1(Becker et al., 2002).

Mutations in BLR result in phyllotaxy defects including both an increase in the number of lateral organs and displacement of organs along the stem. Although mechanisms governing phyllotactic patterning are still to be elucidated, early surgical experiments have shown that leaf primordia are positioned in response to preexisting primordia(Snow and Snow, 1931). This influence may be mediated by production of a diffusible inhibitor or may be biophysical in nature. A model where biophysical forces regulate phyllotaxy is supported by studies demonstrating induction of leaf formation by local concentration of the cell wall protein expansin(Fleming et al., 1997; Pien et al., 2001; Reinhardt et al., 1998). However, organ positioning can also be influenced by auxin. Chemical inhibition of polar auxin transport and mutations affecting auxin transport,including pinformed and pinoid in Arabidopsis,result in failure to initiate lateral organ outgrowth(Benjamins et al., 2001; Christensen et al., 2000; Vernoux et al., 2000). Primordial outgrowth in such cases can be induced at sites of localized auxin concentration increase (Reinhardt et al.,2000; Vernoux et al.,2000). Organs can be induced at any position around the circumference of the meristem, and also within a restricted region along the apical-basal axis of the SAM.

The phenotypic defects in blr, including aberrant organ initiation patterns may in part due to disruption in auxin signalling. Consistent with this proposal, blr mutants display reduced stature and loss of apical dominance typical of reduced auxin signalling(Lincoln et al., 1990). Interestingly, phyllotaxy defects are also observed in plants expressing a constitutively active form of a RHO GTPase that may be involved in mediating plant responses to hormones such as auxin(Li et al., 2001). The inflorescence phenotype in this case resembles that of bpmutants.

The increase in organ number and organ displacement indicates blrmutants are no longer fully responsive to inhibitory signals from preexisting organs such that distances between organs are not maintained. Aberrant initiation along the apical-basal axis of the SAM potentially contributes to variable internode lengths. Alternatively, regular partitioning of cells between organs and internodes is affected. In this respect blrresembles te1 in maize, which has also been interpreted as a phyllotaxy mutant (Veit et al.,1998).

Despite the phyllotactic defects in blr there is no appreciable difference in the size of the SAM compared with wild type. This may be coincident with more peripheral zone cells being specified as organ founder cells. Alternatively, in blr mutants an increase in recruitment of cells into lateral organs is offset by an increase in the number of stem cells, together maintaining SAM size. In this case BLR normally delays differentiation of stem cells in the SAM and slows their propagation. In each case, the function of BLR in delaying specification of lateral organs is consistent with BLR expression in peripheral cells of the inflorescence meristem, but not in initiating primordia.

Stem cell lineages expand according to the Fibonacci series when daughter cells are delayed from acquiring stem cell fate, raising the possibility that stem cells are responsible for phyllotactic patterns(Klar, 2002). In this respect, BLR fulfills a postulated stem cell function required for Fibonacci progression (Klar, 2002), in that BLR dictates how long daughter cells require to differentiate in the stem cell lineage. However, this model remains controversial and is yet to be substantiated (Fleming,2002). For example, stem cell lineages are multicellular in higher plants, such that extensive coordination in the meristem would be required for stem cell lineages to regulate organ initiation in this way. The effects of the blr mutation on phyllotactic pattern are intriguing in this context and will be examined further.

We thank George Haughn for providing yeast two-hybrid constructs, John Bowman for the GUS expression vector, Julie Thomas for advice on yeast transformation, Marja Timmermans and members of Rob Martienssen's lab for helpful discussion. We also thank Tim Mulligan for plant care, and Rulan Shen,Uma Umamaheswari, Amy Tang-Qui and the CSHL gene trap team for helping with plants and generating the blr line. This work was supported by USDA-NRI grant 2003-00967 to M.E.B. and R.A.M., and by NIH Postdoctoral Fellowship GM19974 and USDA-NRI 35301-10878 to A.T.G.