Control of Shoot Cell Fate

Homeodomain transcription factors were first characterized in animals, in which they are key regulators of body plan development (Gehring, 1987). Indeed, the understanding of the manner in which homeobox-defined developmental pathways shape organismal form has been one of the triumphs of 20th century biology (Nusslein-Volhard, 1994).

In 1991, the KNOTTED1 (KN1) gene of maize was cloned and shown to encode a homeodomain protein (Vollbrecht et al., 1991). This demonstrated that homeobox genes are also key developmental regulators in plants. A series of dominant mutations in the KN1 locus condition a variety of developmental aberrations in the maize leaf. These include proximal (sheath) to distal (blade) tissue transformations and outgrowths of aberrantly differentiated tissue in association with the vasculature (Hake, 1992; Smith et al., 1992). KN1 defined a small family of genes (class1 KNOX genes) that share both sequence similarity and distinct functional features (Kerstetter et al., 1994). The KNOX subfamily of homeodomain proteins is distinct from other types of plant homeobox transcription factors that have a wide spectrum of functions in plant growth and development. KNOX genes are expressed in overlapping domains within the shoot apical meristem (SAM) and the unexpanded stem of higher plants (Jackson et al., 1994). However, KNOX proteins (and transcripts) are excluded from young leaf primordia. Indeed, exclusion of KNOX proteins from leaf founder cells (a group of cells in the meristem that is destined to form a leaf) is believed to be instrumental in the acquisition of leaf fate (Smith et al., 1992; Jackson et al., 1994). Ectopic expression of KNOX genes in leaves of different plants results in dramatic tissue transformations, including ectopic meristematic activity (Sinha et al., 1993; Lincoln et al., 1994; Schneeberger et al., 1995; Chuck et al., 1996; Sentoku et al., 2000). Detailed studies of these transformations in maize have suggested that ectopic KNOX expression may result in retardation of leaf developmental programs (Muehlbauer et al., 1997). In barley, the dominant Hooded mutation conditions ectopic expression of the HvKNOX3 gene (Muller et al., 1995). This results in formation of an ectopic flower on the infloresence, as does overexpression of the maize KN1 gene in barley (Williams-Carrier et al., 1998). These phenotypes are consistent with the observation that ectopic KNOX expression results in ectopic meristematic activity in several species (Sinha et al., 1993; Lincoln et al., 1994; Chuck et al., 1996; Sentoku et al., 2000). Analyses of the effects of ectopic KNOX expression have been complemented by the more recent characterization of loss-of-function mutations in KNOX genes. Such mutations in the SHOOTMERISTEMLESS (STM) gene of Arabidopsis and the KN1 gene of maize result in defects in shoot meristem formation and maintenance, clearly demonstrating that at least some KNOX proteins are essential for meristem function (Long et al., 1996; Kerstetter et al., 1997; Vollbrecht et al., 2000).

There has been considerable interest in uncovering additional components of KNOX-defined developmental pathways. ROUGH SHEATH2 (RS2) in maize and PHANTASTICA (PHAN) in Antirrhinum are negative regulators of KNOX gene expression in leaves. These genes are closely related in sequence and encode myb domain proteins. Thus, RS2 and PHAN function in similar pathways (conserved across monocots and eudicots) required to exclude KNOX genes from developing leaves (Schneeberger et al., 1998; Waites et al., 1998; Timmermans et al., 1999; Tsiantis et al., 1999). Loss-of-function mutations in RS2 condition phenotypes associated with ectopic KNOX expression in maize leaves, the most striking being proximal (P) to distal (D) transformations. phan mutants display a wide range of leaf phenotypes, including severe dorsoventral (D/V) transformations resulting in completely ventralized radial leaves (Waites and Hudson, 1995).

Two recent articles greatly increase our insight into the organization of KNOX pathways in plants. Both articles concern the genetic control of KNOX gene expression in Arabidopsis. Ori et al. (2000) describe the characterization of mutations in the ASYMMETRIC LEAVES1 (AS1) and AS2 loci and demonstrate that both AS1 and AS2 are required for the negative regulation of KNOX expression in leaves. Mutations in AS1 or AS2 resulted in the inappropriate expression of two KNOX genes (KNAT1 and KNAT2, for KNOTTED-like from Arabidopsis thaliana) in leaves, leading to alterations in leaf shape that resemble weak lobing phenotypes conditioned by overexpression of KNAT1 under the control of the Cauliflower Mosaic Virus 35S promoter. Ori et al. (2000) further show that mutations in the SERRATE (SE) locus (Clarke et al., 1999) condition a strong enhancement of the as1 and as2 phenotypes. Leaves of the corresponding double mutants dis-play deep leaf lobing and ectopic meristematic activity and therefore phenocopy 35S::KNAT1-overexpressing plants. Double mutants between as1 or as2 and the pickle mutation, which is known to enhance meristematic activity in carpels (Eshed et al., 1999), also result in pronounced meristematic activity on leaves but not an increase in leaf lobing. In both instances, enhancement occurs, with neither se nor pickle affecting KNOX expression. As a result, for the first time we have a glimpse into novel factors that may regulate cellular competence to respond to KNOX transcription factors. Both SE and PICKLE encode chromatin remodeling factors (Eshed et al., 1999; Prigge and Wagner, 2001). Thus, chromatin-based regulation emerges as a control point for KNOX-mediated developmental events in plants.

It has been known for several years that chromatin structure represents an important control point for animal homeobox gene expression. For example, the chromatin-modifying Polycomb group proteins act to repress homeobox genes during animal embryo development. The conceptual similarity of these processes to KNOX repression by PHAN/AS1/RS2 has been noted by several authors (Timmermans et al., 1999; Tsiantis et al., 1999). It also has been demonstrated that the Polycomb-like gene CURLY LEAF is required for the repression of floral homeotic genes in Arabidopsis leaves (Goodrich et al., 1997). However, as yet, no direct evidence for the regulation of KNOX expression via chromatin remodeling exists. It will be interesting to discover whether any of the many putative chromatin remodeling factors identified via the completion of the Arabidopsis genome (Arabidopsis Genome Initiative, 2000) have a role in this process. Ori and co-workers have shown that chromatin remodeling processes may be central to controlling KNOX-mediated developmental events without affecting KNOX expression as such. This led them to suggest that the se and pickle mutations affect the sensitivity of leaf tissue to the presence of ectopic KNOX protein conditioned by the as1 and as2 mutations. This could be the case if regulatory regions of the targets of KNOX proteins are made more accessible due to disruption in normal chromatin packaging in se and pickle. It will be exciting to determine whether genes such as SE or PICKLE can influence the function of KNOX proteins in their natural domain of expression in the meristem.

Byrne et al. (2000) were able to fine map the AS1 gene and demonstrate that it corresponds to AtPHAN, the previously isolated Arabidopsis ortholog of the PHAN/RS2 genes (Timmermans et al., 1999). They characterize two independently derived alleles of as1 and demonstrate that AS1 negatively regulates KNAT1 and KNAT2 in leaves. They also show that AS1 and STM are expressed in mutually exclusive domains at the shoot apex. Significantly, they were able to take the network of interactions controlling shoot patterning one step further. They demonstrate that as1 mutations can suppress the meristem maintenance defects of stm (presumably because KNAT1 and KNAT2 are derepressed). Furthermore, they present evidence that AS1 transcripts accumulate in a wider expression domain in stm mutants than in wild-type embryos. The authors propose that STM negatively regulates AS1 in the stem cell population of the meristem. In founder cells, STM is down regulated, thus allowing expression of AS1. AS1 in turn represses KNAT1 and KNAT2, thus promoting leaf cell fate. Therefore, Byrne et al. have uncovered a network of negative interactions in which the STM-expressing stem cells and the AS1-expressing founder cells are able to distinguish themselves within the shoot meristem. It will be exciting to test this model further by studying the expression of KNAT1 and KNAT2 in as1 stm double mutants. This should confirm whether the rescue of stm by as1 does indeed occur because of KNAT1 or KNAT2 being expressed in a wider domain. If this is the case the implication would be that KNAT1 and 2 are to some degree redundant to STM, and that their different developmental roles reflect differences in their expression patterns. This hypothesis could be further tested by performing promoter swap studies between KNAT1, KNAT2 and STM.

Both of these articles greatly improve our understanding of the mechanisms that control shoot development (see Figure 1 for a hypothetical model incorporating these recent advances on control of KNOX expression). They also raise further questions. AS1, PHAN, and RS2 are negative regulators of KNOX gene expression in leaves of their respective species. However, despite the equivalence of action at the molecular level, the phenotypic outcomes of the corresponding mutations are, to some extent, divergent (Table 1). phan mutants display radial leaves, whereas rs2 and as1 mutants do not. This is somewhat surprising at first. Several workers in the field have speculated about this discrepancy, and there are two views of the matter, which are not necessarily mutually exclusive.

Figure 1.

Hypothetical Model Depicting Mechanisms Controlling KNOX Expression in Higher Plants.

Continuous lines represent connections for which there is experimental evidence. Dotted lines represent hypotheses that await experimental testing. Question marks denote areas in which there is still some uncertainty. Note that it is possible that AS1/RS2/PHAN (ARP) proteins as well as chromatin remodeling factors control KNOX expression. If this is the case, ARP proteins could mediate repression by recruiting chromatin remodeling factors to KNOX gene regulatory regions (see Eshed et al. [1999] for models postulating repression of meristematic genes by the combined action of chromatin remodeling factors and specific transcription factors). AS2 (not included in the diagram) also represses KNOX gene expression in Arabidopsis leaves (Ori et al., 2000). However, the cloning of AS2 and the manner in which AS2 interacts with STM have not yet been reported. The diagram considers KNOX proteins as transcriptional activators; it is also possible that they act as repressors.

Table 1.

Consequences of Loss-of-Function Mutations in Arabidopsis AS1, Maize RS2, and Antirrhinum PHAN Genes^a

	Ectopic KNOX	P/D Aberrationsb	D/V Aberrationsb
as1	Yes	Yes	Maybe minor
rs2	Yes	Yes	No
phan	Yes	Yes	Yes

^aBased on phenotypic interpretations by Waites et al., 1995, 1998 (Antirrhinum); Chuck et al., 1996; Schneeberger et al., 1998; Timmermans et al., 1998; Tsiantis et al., 1999 (maize); Ori et al., 2000 (Arabidopsis); Byrne et al., 2001.

^bP/D, proximodistal; D/V, dorsoventral. Mediolateral axis is not considered.

According to one line of thought, the PHAN gene in Antirrhinum has an additional role (separate from repressing KNOX genes) in specifying the D/V axis. In maize, such a role is either absent or fulfilled in a redundant manner in conjunction with a hypothetical RS2 duplicate factor (Timmermans et al., 1999). Testing of this hypothesis will require identification of a loss-of-function mutation in such a factor and study of the double mutants with rs2. In any event, as1 mutants of Arabidopsis do not display radial leaves. This would suggest that PHAN/AS1/RS2-like genes do not have generalized D/V-specifying roles across the angiosperms. Of course, it is possible that this is the case in some species (e.g., Antirrhinum) but not in others. This could reflect divergence in downstream targets of the genes. Divergence of downstream targets of structurally related transcription factors has been reported between maize and eudicots and between eudicots for myb-transcriptional regulators of the anthocyanin biosynthetic pathway (Jin and Martin, 1999). To answer this question fully, it will be necessary to obtain null alleles of as1, because it is possible that differences between as1 and phan reflect differences in allelic severity. This is especially important given the suggestion by Ori et al. (2000) that some aspects of the as1 phenotype may be interpreted as minor perturbations across the D/V axis. Finally, because the radial leaves of Antirrhinum occur in higher nodes, it is possible that modifiers of PHAN action exist that enhance the phan phenotype in upper leaves. If this is the case, it will be interesting to determine whether such modifiers are associated with loci believed to control phase change transitions in plants (Telfer and Poethig, 1998).

An alternative view is that the apparent loss of D/V polarity in phan is an indirect effect of ectopic KNOX expression. This could reflect differences in the manner in which different species elaborate laminae. In Antirrhinum, KNOX-induced proximal to distal transformations could result in radialized leaves. This would occur if the distal part of the leaf lamina acquires ventral features of the more proximal petiole tissue. This would be equivalent to the blade (distal) to sheath (proximal) transformations that occur in rs2 (Tsiantis et al., 1999). Furthermore, the ectopic presence of KNOX genes within early leaf primordia may result in the aberrant reformation of a (presumed) morphogenetic boundary defined by the area where cells that do not express KNOX are contiguous with KNOX-expressing cells in the meristem. This boundary may be involved in specifying the D/V axis of the primordium; therefore, its reformation may result in ill-defined D/V axes in phan leaves. The possibility that KNOX expression affects both P/D and D/V axes also suggests that the formation of the two axes may be interdependent. Genetic crosses between loss-of-function KNOX mutants and phan mutants should reveal whether radiality in phan leaves is mediated by ectopic KNOX expression.

Given that Arabidopsis and Antirrhinum are both eudicots, it is surprising that their phenotypes are less reconcilable than those of rs2 and as1. Possible explanations are the divergence of downstream targets such that KNOX expression in Antirrhinum leaves interferes with aspects of D/V axis formation (e.g., expression of the YABBY [Siegfried et al., 1999] or PHABULOSA [McConnell and Barton, 1998] genes) and the presence of modifying loci. It is also possible that these differences highlight subtle differences between leaf development programs. For example, leaf primordia are dorsoventral from their inception (Sylvester et al., 1996). However, there may be slight temporal differences in the elaboration of the D/V axis or cell division dynamics. Thus, ectopic KNOX expression would have different phenotypic outcomes depending on when KNOX expression occurred (a radial outcome may be less likely if the D/V axis is elaborated earlier). More research on comparative morphology and knowledge of cell division patterns during early leaf development (Donnelly et al., 1999) should help resolve these issues.

Research on tomato has already highlighted the broad spectrum of phenotypes that can result from ectopic KNOX expression (Hareven et al., 1996; Parnis et al., 1997; Janssen et al., 1998). These differences are thought to relate to the precise time and place of expression. Of particular interest are the Mouse ear and Curl mutations that condition aberrant transcription of the tomato TKN2 gene and that result in distinct phenotypes (Chen et al., 1997; Parnis et al., 1997). A high proportion of upper leaves in Mouse ear plants are reduced to almost bladeless elongate lateral appendages. Similar phenotypes are obtained when TKN2 is overexpressed under the control of the 35S promoter (Parnis et al., 1997). This suggests that ectopic KNOX expression alone can be sufficient to condition severe inhibition of lateral growth in dicot leaves. Tomato also represents an exception with respect to the mutually exclusive expression patterns of KNOX and PHAN/RS2/AS1like genes as (unlike maize, Anthirrinum and Arabidopsis) tomato PHAN and KNOX transcripts are co-localized within the shoot apex (Koltai and Bird, 2000). The full significance of this result is not yet clear, but it may relate to the fact that tomato leaves are dissected to leaflets, and unlike species with simple leaves (e.g., maize, Anthirrinum and Arabidopsis), display KNOX expression within leaf primordia (Hareven et al., 1996; Janssen et al., 1998).

An important aspect of the control of KNOX function in early leaf development is still unclear. Although KNOX genes are misexpressed in rs2/phan/as1 mutant leaves, the initial down regulation of KNOX genes associated with founder cell recruitment is intact. Thus, it appears that separate pathways may exist to facilitate the initial down regulation of KNOX and the maintenance of homeoboxes in an “off” state later in development. Genetic systems other than Arabidopsis may help define the mechanisms required for the initial down regulation event. For example, maize has a larger SAM, which could make it easier to uncover subtle changes in SAM function and the corresponding consequences for leaf development. Both the leaf bladeless (lbl) mutant (Timmermans et al., 1998) and the duplicate factor narrow sheath mutant (ns1 ns2) (Scanlon et al., 1996) of maize show aberrant KNOX expression in small groups of cells within the founder cell population of the SAM. In fact, NARROW SHEATH1 has been shown to exert a localized signaling function to promote founder cell recruitment at the flanks of the meristem (Scanlon, 2000). Cloning of the lbl and ns loci should help to increase our understanding of the elusive process of founder cell recruitment.

The comparative consideration of the as1, phan, and rs2 mutant phenotypes has been invaluable in fully comprehending the functions of the respective genes. We are just beginning to understand how taxon-specific differences in the function of genes involved in developmental patterning relate to the different plant morphologies apparent in nature. No doubt, this will be one of the major challenges of postgenomics biology.