The shoot apical meristem comprises undifferentiated stem cells and their derivatives, which include founder cells for lateral organs such as leaves. Meristem maintenance and lateral organ specification are regulated in part by negative interactions between the myb domain transcription factor ASYMMETRIC LEAVES1, which is expressed in lateral organ primordia, and homeobox transcription factors which are expressed in the shoot apical meristem (knox genes). The knox gene SHOOT MERISTEMLESS (STM) negatively regulates ASYMMETRIC LEAVES1 (AS1) which, in turn, negatively regulates other knox genes including KNAT1 and KNAT2, and positively regulates the novel gene LATERAL ORGAN BOUNDARIES (LOB). Genetic interactions with a second gene, ASYMMETRIC LEAVES2 (AS2), indicate it acts at the same position in this hierarchy as AS1. We have used a second-site suppressor screen to isolate mutations in KNAT1 and we show that KNAT1 is partially redundant with STM in regulating stem cell function. Mutations in KNAT2 show no such interaction. We discuss the regulation and evolution of redundancy among knox genes.
The shoot apical meristem (SAM) of higher plants is divided histologically into a number of zones, which are also defined by gene expression patterns. The central zone contains undifferentiated, slowly dividing cells that give rise to daughter cells in the peripheral zone. Groups of cells in the peripheral zone (founder cells) are recruited into initiating lateral organs, and undergo rapid cell divisions, expansion and terminal differentiation. Meristem homeostasis is achieved by a balance between slow cell divisions in the central zone and displacement of cells into lateral organ primordia. Mutations leading to loss of meristem homeostasis have defined a number of genetic pathways for maintaining a balance between stem cells and their derivatives (reviewed by Bowman and Eshed, 2000; Clark, 2001).
One pathway involved in meristem initiation and maintenance involves a highly conserved class of homeodomain transcription factors encoded by knox genes. knox genes are defined by homology to the maize knotted1 (kn1) gene and are separated into two classes based on sequence identity and conserved intron location (Bharathan et al., 1999; Kerstetter et al., 1994; Reiser et al., 2000). The Arabidopsis genome sequence has revealed 8 knox genes (The Arabidopsis Genome Initiative, 2000). Class I genes comprise STM, KNAT1, KNAT2 and KNAT6 (Lincoln et al., 1994; Long et al., 1996; Semiarti et al., 2001). Loss-of-function mutations in STM result in embryos that lack a SAM and so fail to develop any postembryonic vegetative tissue (Barton and Poethig, 1993; Clark et al., 1996; Long et al., 1996). STM is thus required to maintain proliferation of cells in the SAM and/or prevent their differentiation. Recessive mutations in the kn1 gene of maize also condition defects in meristem maintenance (Kerstetter et al., 1997; Vollbrecht et al., 2000). Both STM and kn1 are expressed throughout the SAM but are down-regulated in founder cells that are recruited to form lateral organs (Jackson et al., 1994; Long et al., 1996; Smith et al., 1992). Down regulation of knox genes in lateral organ primordia is a critical event in organ patterning as ectopic expression of knox genes disrupts normal leaf development (Byrne et al., 2001; Chuck et al., 1996; Reiser et al., 2000). In Arabidopsis, KNAT1 and KNAT2 are also expressed within the SAM and are down-regulated in lateral organ primordia, but so far there is no genetically defined role for these class I knox genes.
Mutations in AS1 result in plants that have abnormal leaves, with marginal outgrowths or lobes (Byrne et al., 2000; Ori et al., 2000; Semiarti et al., 2001; Tsukaya and Uchimiya, 1997). AS1 is a myb domain transcription factor related to ROUGH SHEATH2 (RS2) in maize and PHANTASTICA (PHAN) in Antirrhinum (Byrne et al., 2000). All three genes are expressed in lateral organ primordia and act 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; Waites et al., 1998). Unexpectedly, as1 suppresses the stm mutant phenotype, so that double mutants have an as1 vegetative shoot. Further, in stm mutant embryos, AS1 expression spreads throughout the apical region. This genetic interaction indicates that STM prevents AS1 expression in stem cells of the SAM and so maintains their undifferentiated state (Byrne et al., 2000). STM has additional roles in the inflorescence, since as1 stm mutants lack normal flowers. We previously proposed that other knox genes might replace STM in vegetative but not in floral meristems, accounting for the phenotype of as1 stm-1 plants (Byrne et al., 2000).
The mutant asymmetric leaves2 (as2) has a leaf phenotype comparable to as1, and knox genes are also mis-expressed (Ori et al., 2000; Semiarti et al., 2001). We show that AS2 is also negatively regulated by STM and likely interacts with AS1. We used second-site suppressor mutagenesis to identify meristem factors that replace STM in as1 stm double mutants. In this screen we isolated mutations in the KNAT1 gene, which corresponds to the classical locus BREVIPEDICELLUS (BP) (Douglas et al., 2002; Venglat et al., 2002). Thus KNAT1 and STM are redundant in embryo and vegetative development in the absence of AS1. Gene trap and enhancer trap lines were used to show that KNAT2 and the novel gene LATERAL ORGAN BOUNDARIES (LOB) are also regulated by AS1 but do not contribute significantly to the as1 phenotype. Interactions between leaves and meristems were first proposed to have a role in leaf patterning on the basis of surgical experiments (Sussex, 1954; Sussex, 1955). Our studies provide a molecular framework for some of these interactions.
MATERIALS AND METHODS
Plant material and growth conditions
Mutant alleles of as1-1, as2-2, stm-1 and stm-2 were obtained from the Arabidopsis Biological Resource Center (ABRC). as2-2, originally in the Er background, was backcrossed twice to Landsberg erecta prior to double mutant analysis. Kathy Barton kindly provided the stm-11 allele. bp-2 was kindly provided by Dan Riggs. Gene trap and enhancer trap lines were generated as previously described (Martienssen, 1998; Sundaresan et al., 1995). Plants were grown either on soil or on MS medium, supplemented with sucrose, with a minimum day length of 16 hours. Ethyl methanesulphonate (EMS) mutagenesis was carried out by treatment of seed from as1/as1 stm-1/+ plants with 0.5% EMS for 8 hours. Approximately 80 F2 seeds from each of 1200 fertile individuals, of the genotype as1/as1 stm-1/+ or as1/as1 +/+, were screened on soil for a shoot meristemless phenotype.
To generate as2 stm double mutants homozygous as2 plants were crossed to plants heterozygous for stm. AS2 and STM are linked on chromosome 1 and in F2 populations a novel phenotype segregated at a low frequency. F3 plants from individuals of the genotype as2 stm-1/as2 + segregated 1:3 for the double mutant phenotype [as2 159 (72.3%), as2 stm-1 61 (27.7%)]. To construct as1 bp and as2 bp double mutants, plants homozygous for as1 or as2 were crossed to plants homozygous for bp. Double as1 bp and as2 bp mutants segregated in the F2 progeny in the expected 1:15 ratio. The number of plants in each phenotypic class segregating as1 and bp were; wild type 182 (60.3%), as1 44 (14.5%), bp 58 (19.2%), as1 bp 18 (6.0%). The number of plants in each phenotypic class segregating as2 and bp were; wild type 123 (57.2%), as2 40 (18.6%), bp 40 (18.6%), as2 bp 12 (5.6%). Double stm-11 bp and stm-2 bp mutants were generated by crossing plants homozygous for bp to plants heterozygous for stm-11 or stm-2. Only stm and bp phenotypes segregated in the F2 generation. F3 seed from homozygous bp plants segregated 1:3 stm mutants. Segregation values for lines homozygous for bp and segregating stm-11 were; bp 215 (72.6%), double bp stm-11 59 (27.4%). Segregation values for bp mutant lines segregating stm-2 were; bp 229 (67.3%), double bp stm-2 75 (32.7%).
Triple as1 stm-1 bp mutants were generated by crossing plants homozygous for as1 and heterozygous for stm-1 to plants homozygous for bp. The F3 generation from selfed as1/as1 bp/bp stm-1/+ individuals segregated 1 in 4 shoot meristemless individuals [as1 bp 113 (75.3%) and as1 stm-1 bp 37 (24.7%)]. Plants homozygous for the Ds insertion allele of KNAT2 were crossed with homozygous mutants in the case of as1, as2 and bp, and with heterozygous plants in the case of stm-11 to generate double mutants. In F3 lines homozygous for as1, as2 or bp and segregating for knat2 and in lines homozygous for the knat2 allele and segregating for stm, no new phenotypes were observed. Triple as1 stm-1 kt2 mutants were generated by crossing as1/as1 stm-1/+ plants with kt2/kt2 plants. The F3 progeny from selfed as1/as1 stm-1/+ kt2/kt2 individuals segregated plants that were phenotypically as1 and plants with an as1 stm-1 phenotype in the ration 1:3.
DNA extraction and manipulation were carried out using standard protocols (Sambrook et al., 1989). To sequence EMS-induced mutations, DNA from mutant plants was amplified with primer pairs encompassing the exon regions of KNAT1. PCR products were sequenced with internal primers, using dye terminator cycle sequencing (Applied Biosystems). For RT-PCR, total RNA was purified using Trizol reagent (GibcoBRL). Following treatment with DNase (Boehringer Mannheim) complementary DNA was synthesized using 100 Units of M-MuLV reverse transcriptase (New England Biolab) in 50 mM Tris-HCl (pH 8.3), 30 mM KCl, 8 mM MgCl2, 10 mM DTT, 1 mM each of dATP, dCTP, dGTP, dTTP, 1 μM oligo(dT), 50 Units RNasin and 0.1 μg BSA. RT-PCR reactions were performed with gene-specific primers. KNAT2 primers (ACCACCGGAGACAATCAAAG and TCCGCTGCTATGTCATCATC) span the exon 3/exon 4 junction. PCR products were subject to Southern hybridization using gene-specific probes. ClustalW analysis of class I knox genes was performed using MacVector6.5.1 (Oxford Molecular Group).
Histology and microscopy
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, 2 mM each of potassium ferricyanide and potassium ferrocyanide. Seedlings were mounted in 50% glycerol before viewing. Inflorescences from plants carrying a DsG element in KNAT2 were first stained for GUS expression before fixing in FAA (50% ethanol, 5% glacial acetic acid, 3.7% formaldehyde), dehydrating through an ethanol series, embedding in paraffin and sectioning. Eight-day old seedlings were fixed in glutaraldehyde and dehydrated through an ethanol series prior to embedding in paraffin. 10 mm sections were cut and stained with Toluidine Blue. For scanning electron microscopy fresh material was mounted on silver tape (Electron Microscope Sciences) and viewed with an Hitachi S-3500N SEM using a beam voltage of 5 kV.
as1 and as2 interact similarly with stm
Rosette leaves of wild-type plants are elongate, entire and spatulate in shape (Fig. 1A) whereas rosette leaves of as1 are smaller and rounder (Fig. 1B) with the margins rolled downwards and lobed (Byrne et al., 2000; Ori et al., 2000; Semiarti et al., 2001). Lobing is variable and background dependent, but is most prominent in late rosette and cauline leaves. as2 mutants have similar defects in leaf patterning (Fig. 1C), except rosette leaves and petioles are more elongate than as1 (Ori et al., 2000; Semiarti et al., 2001).
AS1 is expressed in lateral organ primordia and negatively regulates KNAT1 and KNAT2, which are mainly expressed in peripheral regions of the SAM. To identify additional targets in the shoot apex, as1 plants were crossed with 10 gene trap and enhancer trap GUS reporter gene insertions (Springer et al., 1995; Sundaresan et al., 1995) that are expressed in this region (P. S. Springer, Q. Gu and R. A. M., unpublished). The only GUS reporter gene expression pattern that was altered in an as1 background was ET22. ET22 disrupts the LATERAL ORGAN BOUNDARIES (LOB) gene, and is expressed in the shoot apex, the hypocotyl and the roots (Shuai et al., 2002). In the shoot apex, GUS localization is restricted to a band of cells at the boundary between developing organ primordia and the SAM (Fig. 1D). LOB expression in this region is found in vegetative, inflorescence and floral stages of growth and persists throughout development (Shuai et al., 2002). In as1 mutants, expression of LOB in the vegetative shoot apex is absent in young seedlings (Fig. 1E) and reduced to two small patches at the outer margin of the leaf in older seedlings. LOB expression is also much reduced in the vegetative apex of as2 (Fig. 1F), although weak GUS staining is observed at the boundary between organ primordia and the SAM in older seedlings. In contrast, LOB expression in the hypocotyl, the root and the inflorescence is unaltered in either mutant (data not shown). Thus, as1 and as2 affect LOB expression in the same manner, suggesting that both AS1 and AS2 positively regulate LOB within the shoot apex. Seedlings homozygous for the insertion allele of LOB have a wild-type phenotype (Shuai et al., 2002) and there is no effect on either as1 or as2.
Given that as1 and as2 have similar phenotypes and are both required for normal expression of knox genes and LOB, we carried out double mutant analysis to determine if as2 also interacts with stm. Embryos homozygous for strong stm alleles, including stm-1 and stm-11, completely lack a SAM and develop cotyledons that are fused at their base (Barton and Poethig, 1993; Clark et al., 1996; Long and Barton, 1998). Weak stm mutants, such as stm-2, also germinate with fused cotyledons, but subsequently form a SAM and initiate leaves (Clark et al., 1996; Endrizzi et al., 1996). In as1 stm-1 double mutants vegetative shoots and leaves are indistinguishable from those of as1 single mutants. In reproductive development as1 stm-1 double mutants generate additional lateral shoots in the place of flowers. The phyllotaxy of lateral shoots in the inflorescence is also somewhat irregular compared with as1 single mutants (Fig. 2A) (Byrne et al., 2000). Mutants homozygous for as1 and the weaker stm-2 allele are similar to as1 stm-1 double mutants except that they form fewer lateral shoots and more flowers, most of which remain incomplete (Fig. 2C). Typically flowers have a normal number of sepals, a reduced number of petals and stamens and only occasionally form a central abnormal unfused carpel. Terminal flowers are often fusions of more than one flower (Fig. 2E).
Double mutants of as2 with either strong or weak alleles of stm are comparable in phenotype to as1 stm-1 and as1 stm-2 mutants, respectively (Fig. 2B,D). Vegetative shoots of as2 stm-1 and as2 stm-2 are indistinguishable from as2 alone. However, inflorescence development in double mutants is disrupted with only occasional and abnormal flowers produced in as2 stm-1 double mutants (Fig. 2B) and more often in as2 stm-2 (Fig. 2D) double mutants. As with as1 stm, the flowers in as2 stm double mutants lack reproductive organs and show some homeotic conversions (Fig. 2F). This result demonstrates that, like as1, as2 suppresses stm. Because STM expression is unaffected in as2 (Semiarti et al., 2001) this epistatic interaction indicates that AS2 is negatively regulated by STM.
Screening for suppressors of as1 stm-1
One function of STM is to prevent AS1 expression in stem cells of the SAM (Byrne et al., 2000). However, STM may have additional roles in meristem maintenance that are assumed by other factors redundant with STM that are only revealed in as1 stm-1 double mutants. Likely candidates are the other class I knox genes expressed in the SAM, namely KNAT1, KNAT2 and KNAT6 (Lincoln et al., 1994; Long et al., 1996; Semiarti et al., 2001). In pairwise comparisons (Fig. 3) STM is most closely related to KNAT1, sharing 44% identity over all and 70% identity within the homeodomain. However, KNAT2 is most closely related to KNAT6 sharing overall 70% amino acid identity and 89% identity in the homeodomain.
In order to identify factors redundant with STM we carried out a screen for mutants that suppressed the as1 stm-1 phenotype. Seed from plants of the genotype as1/as1 stm-1/+ were mutagenized with EMS, since the double homozygous mutant is sterile. Progeny from 1200 F1 individuals, two-thirds of which were heterozygous for stm-1, were screened for a shoot meristemless phenotype. In one line, EMS202, approximately 1 in 16 seedlings lacked a shoot meristem and were indistinguishable from those carrying strong alleles of stm. Upon flowering EMS202 also segregated as1 stm-1 double mutants as expected, as well as plants with reduced pedicels and downward-hanging flowers resembling the previously described mutant brevipedicellus (Koornneef et al., 1983). A likely candidate for mutation in EMS202 was KNAT1 since brevipedicellus (bp) has recently been shown to coincide with the KNAT1 locus (Douglas et al., 2002; Venglat et al., 2002). Sequence analysis revealed that EMS202 carries a single base change creating a stop codon in the first exon of KNAT1 (Fig. 3). This allele is designated bp-6. Two additional lines carried bp mutants with nucleotide disruptions in KNAT1 (Fig. 3). In one case (bp-7) a single nucleotide change creates a stop codon in the second exon. In the other (bp-8), multiple changes include two additional ACC repeats and two single nucleotide changes, which result in amino acid insertions, an amino acid substitution and a premature stop (Fig. 3).
To confirm that the shoot meristemless seedlings were derived from triple as1 stm-1 bp homozygotes, we constructed triple mutants between as1 stm-1 and an independently derived deletion allele of KNAT1 (Douglas et al., 2002). Progeny from as1/as1 bp/bp stm-1/+ mutants segregated 1 in 4 shoot meristemless plants, as expected. Like stm-1, these as1 stm-1 bp mutants have cotyledons fused at the base and no vegetative shoot (Fig. 4B,C), although rarely some leaves are formed. At 8 days after germination, the wild-type SAM is visible in sections as a dome of densely staining cells at the base of the cotyledons (Fig. 4D). In contrast, sections through the apex of as1 stm-1 bp mutant seedlings have no SAM (Fig. 4F), and are comparable to stm-1 single mutants (Fig. 4E). Thus KNAT1 is required for SAM maintenance in the absence of AS1 and STM.
Genetic interactions between as1, as2 and knox genes
The genetic interaction between STM, AS1 and KNAT1 was originally proposed based on molecular characterization of single and double mutants of as1 and stm (Byrne et al., 2000). To provide further support for this genetic pathway we also examined the interaction between STM and KNAT1. bp stm-11 double mutants were indistinguishable from stm alone demonstrating that stm is epistatic to bp. On the other hand, the weak allele stm-2 is enhanced in plants that are also homozygous for bp. Compared with stm-2 mutants, bp stm-2 double mutants have a much reduced vegetative shoot with many plants only giving rise to a few abnormal leaves and no flowers (Fig. 5). These interactions are consistent with KNAT1 being downstream of STM and AS1 (Byrne et al., 2000).
We also examined the genetic interaction of bp with as1 and as2 (Fig. 6). as1, as2 and bp mutants are smaller than wild type and bp plants show a slight loss of apical dominance (Fig. 6A-C,E). There is no apparent affect of bp on leaf development. as1 bp and as2 bp double mutants are smaller than either single mutant alone (Fig. 6D,F). Leaves of as1 bp and as2 bp double mutants show no significant difference from that of as1 and as2 single mutants, respectively. In wild-type flowers, the sepals enclose the flower until just before anthesis (Fig. 6G,M). In contrast, as1 sepals are reduced in size and do not enclose inner floral organs from an early stage of flower development, while petals do not elongate, and remain shorter than in wild type (Fig. 6I,O). The principal floral defect of bp is a reduction in the length of the pedicel (Fig. 6H,N). In as1 bp double mutants, floral organs are exposed from early in development, and both petals and pedicels are reduced in length (Fig. 6J,P). Sepals in as2 are also reduced, such that developing floral organs are exposed (Fig. 6K,Q). as2 bp double mutants also show aspects of both single mutants (Fig. 6L,R). Thus both the as1 bp and as2 bp double mutant phenotypes are additive in all respects indicating that, although KNAT1 is ectopically expressed in as1 and as2 mutants, it is not required for either phenotype.
Another knox gene, KNAT2, is misexpressed in as1 mutants (Byrne et al., 2000; Ori et al., 2000; Semiarti et al., 2001). A Ds transposon gene trap insertion allele of KNAT2 (GT7953) was identified by systematic sequencing of gene and enhancer trap GUS reporter gene insertions (http://www.cshl.org/genetrap). The gene trap reporter is inserted in the third and largest intron of KNAT2 in the sense orientation (Fig. 3) where it is expected to result in a GUS fusion protein (Springer et al., 1995). GUS activity is detected in the SAM region of embryos and vegetative plants (Fig. 7A,B). The expression pattern is broader in the reproductive shoot extending throughout the inflorescence and floral meristems. In flowers, GUS is initially detected in all organs, but is later confined to the carpels. In addition GUS is expressed in the inflorescence stem and in the pedicel of young flowers (Fig. 7C,D). This pattern closely parallels that reported for KNAT2 mRNA by in situ hybridization and for plants carrying a KNAT2 promoter-GUS transgene (Dockx et al., 1995; Pautot et al., 2001), except that gene trap expression in the inflorescence extends into the meristem. This slight difference may result from protein translocation, from disruption of a regulatory sequence, or from differences in the sensitivity of these experiments.
Full length KNAT2 transcripts are undetectable in plants homozygous for the knat2 gene trap insertion allele (Fig. 7G) but this has no phenotypic effect, either alone or in combination with as1 or as2. This indicates that misexpression of KNAT2 is not required for these phenotypes. Nonetheless, in both as1 and as2, the KNAT2 gene trap reporter expression is expanded somewhat into the base of the leaves (Fig. 7E,F). These results are consistent with misexpression of a KNAT2::GUS transgene reported previously (Ori et al., 2000). However, the transgene was misexpressed in as1 sepals, while gene trap reporter expression is unaltered (data not shown). The gene trap reporter is expressed normally in bp knat2 double mutants which have a bp phenotype. However, stm-11 knat2 double mutants have a stm phenotype and have no GUS expression (data not shown). The absence of KNAT2 expression in stm mutants is consistent with a genetic hierarchy whereby STM negatively regulates AS1, which in turn negatively regulates KNAT2. In addition, the KNAT2 gene trap is expressed in as1 stm double mutants (data not shown). However, the knat2 gene trap insertion had no effect on as1 stm double mutants indicating that, unlike KNAT1, KNAT2 could not functionally substitute for STM.
AS1 and AS2 are in a common genetic hierarchy
as1 and as2 have comparable defects in leaf development that are accompanied by misexpression of class I knox genes (Byrne et al., 2000; Ori et al., 2000; Semiarti et al., 2001). In addition, AS1 and AS2 positively regulate the gene LOB, which encodes a novel cysteine-rich protein and is a member of a large family of related genes (Shuai et al., 2002). LOB is normally expressed at the boundary between meristems and organ primordia but is absent in as1 and greatly reduced in as2. KNAT1 can also function as a positive regulator of LOB (P. S. Springer and R. A. M., unpublished) (Shuai et al., 2002), indicating that KNAT1 and AS1 are both required for LOB expression (Fig. 8). This could account for its localized expression at the boundary of AS1 and KNAT1 expression domains. Negative interactions between transcription factors in adjacent territories is a common mechanism for establishment of boundaries in animal systems and may be employed here (Byrne et al., 2001).
Both AS1 and AS2 interact with STM in a similar manner. Like as1, as2 suppresses the stm mutant phenotype leading to vegetative and inflorescence development but little floral shoot development. This genetic interaction suggests that AS1 and AS2 function in a common genetic pathway, both being negatively regulated by STM (Fig. 8). Double mutants of either as1 or as2 with a weak allele of stm produce more flowers and correspondingly less lateral shoots than combinations with strong alleles, suggesting either direct or indirect dosage-dependent interactions. It has been previously reported that as2 is epistatic to as1 (Ori et al., 2000; Serrano-Cartagena et al., 1999). However, AS1 transcripts can be detected at normal levels in as2 mutants indicating AS2 is not a negative regulator of AS1 (data not shown). As both genes are regulated by STM, but neither regulates the other, a strong possibility is that AS1 and AS2 directly interact to repress KNAT1. The subtle difference in mutant phenotype might then be accounted for by additional non-overlapping functions.
Previously we have shown that the leaf phenotype in as1 stm double mutants is unaltered compared with as1, indicating that STM is not required for the as1 phenotype (Byrne et al., 2000). Likewise, the as2 leaf phenotype is unaltered in double mutants with stm. Surprisingly, mutations in KNAT1 and KNAT2 also have no effect on as1 and as2 phenotypes. One explanation is that misexpression of any one knox gene is sufficient for the phenotype, requiring a triple knox mutant to suppress as1. Alternatively, misexpression of other factors may contribute to as1 and as2.
Redundancy of knox genes
The Arabidopsis genome sequence has revealed large-scale gene duplications that may reflect significant redundancy (The Arabidopsis Genome Initiative, 2000; Martienssen and Irish, 1999). For example, several closely related members of a large family of novel transcription factors, the KANADI genes, as well as members of the YABBY gene family play redundant roles in specification of organ polarity (Eshed et al., 2001; Siegfried et al., 1999). In flower development several groups of closely related MADS box transcription factor genes appear to be fully or partially redundant. Mutations in SHATTERPROOF1 and SHATTERPROOF2 have little phenotypic effect, but in combination they disrupt normal fruit development (Liljegren et al., 2000). Similarly, the three SEPALLATA genes have redundant roles, in that floral organs are replaced by sepals in the triple mutant but not in any other combination (Pelaz et al., 2000). A third group of closely related MADS box genes, APETALA1 (AP1), CAULIFLOWER (CAL) and FRUITFULL (FUL), have partially redundant functions in floral meristem identity (Ferrandiz et al., 2000; Gu et al., 1998; Mandel and Yanofsky, 1995).
In contrast to MADS box genes, Class I knox genes constitute a small family of only four genes in Arabidopsis. KNAT2 and KNAT6 share a high degree of amino acid sequence identity and, like SHATTERPROOF and SEPALLATA, they are located within segmental chromosomal duplications (The Arabidopsis Genome Initiative, 2000). Thus, redundancy probably accounts for the lack of phenotype we observed when a Ds transposon was inserted into KNAT2. KNAT1 and STM are also closely related, but these genes are not part of a segmental duplication and were probably duplicated earlier than KNAT2 and KNAT6. In the inflorescence, STM expression is found in all SAMs while KNAT1 expression is restricted to subepidermal cells of the stem and pedicel (Lincoln et al., 1994; Long et al., 1996). The stem and pedicel are affected in bp mutants, consistent with this expression pattern (Douglas et al., 2002; Venglat et al., 2002). In the vegetative apex, both genes are down-regulated in leaf founder cells, but KNAT1 expression is mainly in the peripheral zone while STM is expressed throughout the SAM (Lincoln et al., 1994; Long et al., 1996). Nonetheless, we have shown that KNAT1 assumes a redundant role with STM in the vegetative SAM in the absence of AS1. The lack of flowers in as1 stm double mutants shows that KNAT1 cannot substitute for STM in floral meristems, consistent with the lack of KNAT1 expression in these cells. This situation resembles the partial redundancy and overlapping expression patterns exhibited by the MADS box genes AP1, CAL and FUL.
Evolutionary implications of knox gene duplications
Phylogenetic analysis of knox genes in plants suggests a monophyletic origin, but the ancestral gene expression pattern remains unresolved (Bharathan et al., 1999; Reiser et al., 2000). One possibility is that STM and KNAT1 represent the ancient duplication of a gene involved in meristem maintenance that repressed AS1, a function that KNAT1 has subsequently lost. Alternatively, STM has acquired a new function to negatively regulate AS1. We favor the former possibility since repression of AS1 is critical to meristem maintenance. Following duplication, the differences between STM and KNAT1 will have favored evolutionary stabilization of both genes (Cooke et al., 1997).
In general, screens for patterning mutants in the vegetative phase have typically recovered negative regulatory genes such as AS1, CURLY LEAF, SERRATE and PICKLE (Byrne et al., 2000; Goodrich et al., 1997; Ogas et al., 1999; Prigge and Wagner, 2001) rather than loss-of-function mutations in individual homeotic genes. One explanation is that genes controlling organogenesis in the vegetative apex have been duplicated over evolutionary time. If one copy of each of these duplicate pairs acquired additional functions in the flower, but still retained its vegetative role, then mutants in floral development would be readily obtained, but leaf mutants would be masked by redundancy (Martienssen and Dolan, 1998). Only genes that regulate this redundancy, such as AS1, could lose function with phenotypic effect. Of course, dominant and haplo-insufficient alleles of homeotic genes could still be recovered (McConnell et al., 2001).
We thank Dan Riggs and Kathy Barton for providing alleles of bp and stm, respectively, and also Dan Riggs and Raju Datla for sharing unpublished results. Thanks to Marja Timmermans and Erik Vollbrecht for helpful discussion and critical reading of the manuscript. We also thank Tim Mulligan for plant care, and Manisha Lotlikar and Anupa Mandava for lab assistance. This work was supported by grants from National Science Foundation, Department of Energy and United States Department of Agriculture.