- © 1998 American Society of Plant Physiologists
The regular geometries of plants have inspired generations of botanists, philosophers, and mathematicians. Perhaps the most basic of plant patterns, the arrangement of leaves along a stem, the phyllotaxy, inspired the statement “from the contemplation of Plants, men might first be invited to Mathematical Enquirys” (see Thompson, 1942). Over the last century, many have accepted this invitation, providing mathematical descriptions of phyllotaxy in terms of both the actual arrangement of leaves and the mechanisms that may provide for their genesis (for reviews, see Williams, 1975; Harrison, 1987; Jean, 1994; Douady and Couder, 1996).
These reasonably straightforward representations have suggested to many that the mechanisms that underlie specific phyllotactic patterns may be relatively simple and are perhaps knowable in molecular terms. Given that phyllotaxies to a large degree reflect the timing and position of leaf initiation events, attempts to understand their basis have largely focused on the shoot apex. Despite these efforts, however, a satisfying explanation of how the position of leaves on a stem becomes determined has proven elusive. On pages 1427–1437 of this issue, Kuhlemeier and co-workers (Reinhardt et al., 1998) continue their novel approach to this question, providing evidence that the regulated loosening of cell walls may play a determining role in the leaf initiation process.
The significance of this finding is best appreciated in the context of our current knowledge of the leaf initiation process. Leaf primordia come into being in the dynamic environment of the shoot apical meristem (Steeves and Sussex, 1989). The growth of the apex can be viewed in terms of three cellular parameters. The first is manifested as regional polarities that can be distinguished by the preferred planes of cell division documented in different cell layers in the shoot apical meristem. In superficial layers that make up the shell-like tunica, cell divisions are predominantly anticlinal (i.e., new cell walls form perpendicular to the surface of the apex). By contrast, the more internal cells that make up the corpus undergo divisions in a variety of planes. Localized constraints on cell division are also apparent in the behavior of periclinal chimeras (Szymkowiak and Sussex, 1996). The stability of these chimeras implies constrained division polarities that are sufficient to maintain discrete, coherent populations of apical initials as well as their derivatives, the L1, L2, and L3 layers.
A second cellular parameter, which is superimposed on these regional cell division polarities, is a gradient of mitotic activity across the shoot apex. The rate of cell division is lowest in the so-called central zone at the summit of the apex but increases toward its periphery (reviewed in Steeves and Sussex, 1989). A third cellular parameter can be described in terms of a gradient of differentiation and determination in which clonal derivatives of the apical initials become more restricted in their fate as they are displaced away from the shoot apex.
From this parade of cells, leaf primordia are organized at periodic time intervals termed plastochrons. Analyses of clonal sectors indicate that each leaf becomes organized from a few dozen to a hundred or so cells, depending on the plant species and the size of the meristem. This group of founder cells is derived from both the superficial L1 layer and one or more internal cell lineages (Poethig, 1984, 1997). These estimates of founder cell numbers are broadly consistent with surgical studies, which point to a subapical position on the flank of the shoot apex as the site of leaf determination.
The position at which each primordium arises can be readily described in terms of its angular and longitudinal displacement relative to the position occupied by the previously initiated leaf. The magnitude and consistency of these displacement values lead to the elaboration of phyllotaxies that are characteristic of the plant species and/or its developmental state. More generally, these phyllotaxies can be described in terms of opposing spirals of leaves along a stem, parastichies whose numbers conform to a Fibonacci series (Williams, 1975).
How groups of cells become organized into leaves at predictable sites remains unclear. Experiments demonstrating that shoot apices physically isolated from the subjacent vasculature continue to initiate leaves in a relatively normal manner bring this question into sharp focus because they suggest that interactions within the shoot apex itself are sufficient to direct this process (Wardlaw, 1950; Ball, 1952). Not surprisingly, models seeking to explain phyllotaxy have tended to relate leaf initiation events to preexisting patterns within the shoot apex. Early suggestions that leaf initiation events occur in positions that maximize their separation from existing primordia (Hofmeister, 1868) are broadly compatible with observed phyllotactic patterns.
Models proposing that spatial constraints dictate where leaf primordia will form have been tested through surgical studies in which incisions to the shoot apex are seen to modify the positions of leaf initiation. Such changes in the position of primordia have been interpreted by some authors to indicate that primordia form at sites that most efficiently use the “available space” (Snow and Snow, 1952). However, comparable results with ferns (Wardlaw, 1949), in which space for lateral organ initiation is less limited, have been interpreted to indicate that each initiation site forms in relation to the positions of previously initiated primordia. These existing primordia are thought to inhibit the fixation of new initiation sites in a proximity-dependent manner.
The mechanism by which preexisting primordia may be able to inhibit nearby leaf initiation events has been hotly debated. One school of thought holds that previously initiated leaf primordia inhibit the initiation of new leaves in their immediate vicinity through the production of a diffusible substance (Richards, 1948; Wardlaw, 1949). Changes in phyllotaxy that result from surgical procedures could be viewed as a consequence of disrupting gradients of this hypothetical inhibitor. Turing (1992) modeled this form of chemically mediated pattern formation in terms of a “reaction–diffusion mechanism,” a model that has been refined by subsequent workers (Meinhardt, 1984; Harrison, 1987).
Auxin, a rich source of which is available in young leaves, has been suggested to play the role of the diffusing inhibitor of morphogenesis. This hypothesis has received support from studies in which application of auxin or its inhibitors perturbs phyllotaxy (Schwabe, 1971; Meicenheimer, 1981). However, it is unclear from these studies whether or not changes in auxin levels directly affect phyllotaxy. Similarly, the interpretation of changes in phyllotaxy that follow surgical procedures could be complicated by the influence of undefined wound responses.
A contrasting model for the determination of leaf initiation sites focuses on biophysical forces within the meristem. One early version of this model supposes that polarized arrays of cytoskeletal or cell wall elements confine growth to particular axes that correspond to sites of leaf initiation (Green, 1987). This axial growth would induce further changes in the polarity of cells higher on the shoot apex, leading to the establishment of new growth axes. An alternative biophysical model proposes that leaf initiation sites are a consequence of predictable patterns of buckling that are known to occur in surfaces under compression (Green, 1994).
One approach to discriminating among such models is to consider how well they explain changes in phyllotaxy that are observed in various contexts. Many of these changes can be associated with altered apex geometry. Indeed, in many dicot species, changes in phyllotaxy correlate with the changes in meristem geometry that occur normally during development or that result from surgical manipulations (Steeves and Sussex, 1989). Changes in phyllotaxy can also be a feature of mutants in which the size of the shoot apex is altered (Greyson et al., 1978; Clark, 1997; Laufs et al., 1998).
Models based on diffusible inhibitors would clearly accommodate some of these phyllotactic changes, at least to the extent that the presumed inhibitor's distribution would be expected to change with an altered geometry. Recent efforts to model leaf initiation patterns in terms of minimizing free energy states successfully explain why certain phyllotactic patterns are more commonly observed, and they emphasize the influence of apex geometry on the process (Douady and Couder, 1996). However, these models do not in their present form provide a strong basis for discriminating between mechanisms that may depend on chemical or biophysical cues.
As the resolving power of molecular techniques continues to improve, the basis for evaluating proposed mechanisms will become increasingly focused on whether candidate components display predicted structure–function relationships. Molecularly grounded models require a means to communicate positional information through inductive and/or morphogen-mediated signal transduction mechanisms that establish the position of leaf initiation sites and also provide a means to elaborate position-dependent determination states.
One recently cloned gene that may play a role in determination of leaf initiation sites is terminal ear 1 (te1), defined by a recessive mutant of maize that exhibits an irregular phyllotaxy (Veit et al., 1998). The te1 mutant initiates leaves more frequently than does the wild type, and many of them are abnormally positioned. In wild-type plants, te1 transcripts accumulate in a series of transverse semicircular rings that bracket sites of leaf initiation. This pattern, together with changes in the pattern of leaf initiation in the mutant, suggests that te1 normally constrains the timing and position of leaf initiation events. That te1 may function through an RNA binding activity is suggested by the presence of conserved RNA binding motifs in the protein. This observation raises the tantalizing possibility that it will be feasible to identify te1 target RNAs, an advance that should significantly contribute to our understanding of the early stages in the leaf initiation process.
Another gene that may function during this early period is introduced by Itoh et al. on pages 1511–1521 of this issue, with their description of the rice mutant plastochron1. The association between a substantially decreased plastochron and a delayed onset of reproductive development in this mutant suggests an intriguing link between mechanisms that regulate leaf initiation and the vegetative to floral transition.
Molecular studies have also provided some insight into the process by which the identities of founder cells become established. Homeobox encoding genes, such as knotted1 (kn1) of maize (Vollbrecht et al., 1991; Smith et al., 1992; Jackson et al., 1994) and its presumed Arabidopsis ortholog SHOOT MERISTEMLESS (Long et al., 1996), are expressed throughout the shoot apex except in leaf primordia. The downregulation of these genes in primordia takes place very early in the initiation process, occurring in a group of otherwise undifferentiated cells whose number and position are consistent with a founder cell identity.
Although the precise function of kn1-like genes is unclear, genetic studies suggest that their expression may help to maintain a state of indeterminacy in the cells in which they are expressed (Sinha et al., 1993; Chuck et al., 1996; Endrizzi et al., 1996). This view is reinforced by the behavior of the kn1-like genes of tomato, the expression patterns of which show no clear evidence for downregulation in the shoot apex but which are associated instead with indeterminate patterns of growth in leaves (Hareven et al., 1996; Chen et al., 1997). Thus, the downregulation of kn1-like genes in the incipient leaf may reflect the acquisition of a state of determinacy rather than a role in establishing a leaf founder cell population.
Despite some uncertainties regarding what downregulation of kn1-like genes achieves, as a marker of leaf founder cell identity this downregulation provides some intriguing clues as to how this population of cells first becomes defined (Jackson and Hake, 1997). In maize, kn1 downregulation apparently occurs in a progressive manner, beginning first in the anlagen of the midrib domain before expanding laterally into the marginal domains. In mutants such as narrow sheath, in which lateral domains are deleted, this spread of the kn1-downregulated state is blocked (Scanlon et al., 1996). These observations are consistent with the view that midrib anlagen may act as organizers for lateral regions of the leaf; however, other explanations cannot be excluded.
A second interesting aspect of kn1 expression is the lack of concordance between the locations at which kn1 mRNA and KN1 protein accumulate. Immunolocalization studies place the KN1 protein in all layers of the shoot apex (Smith et al., 1992). However, kn1 mRNA, although abundant in L2 tissues, is undetectable in the adjacent L1 layer (Jackson et al., 1994), suggesting either that the protein moves intercellularly from L2 to L1 tissues or that a relatively rapid turnover of kn1 transcripts occurs in the L1 layer. These observations have set the stage for more recent studies demonstrating that both the KN1 protein and the kn1 mRNA can be trafficked via a plasmadesmon-mediated system (Lucas et al., 1995). Thus, the absence of kn1 gene products in leaf founder cells might reflect export of the gene product rather than its downregulation at the transcriptional or translational level.
Although it seems that for some plants, the downregulation of a kn1-like factor is an early marker of leaf cell identity, the mechanism by which the fate of these cells becomes fixed is still largely unknown. Similarly, it is unclear how new growth axes that provide for the formation of the leaf become established. Although periclinal cell divisions in subepidermal layers provide the first obvious sign of leaf initiation (Esau, 1965), it is unclear whether divisions of this type are the driving force behind morphogenic events or an ancillary or derivative process (Kaplan, 1992; Barlow, 1994). Several observations favoring the latter possibility have been cited. For example, observed patterns of asymmetric growth associated with leaf development ordinarily continue when cell division is blocked by irradiation (Haber and Foard, 1963; Foard, 1971), suggesting that cell division per se is not essential to sustain anisotropic growth. Moreover, clonal analyses demonstrate that functionally equivalent derivatives of the apical meristem can be derived from very dissimilar patterns of cell division (Poethig, 1997), and mutations that substantially alter patterns of cell division in the developing leaf do not significantly alter its final shape (Hemerly et al., 1995; Smith et al., 1996). These examples of position-dependent determination are more readily accommodated by mechanisms that are defined by interactions that occur on a supracellular level but that express themselves locally in cellularly based terms (Meyerowitz, 1996, 1997).
To address the manner in which the new growth axes of the leaf become established, Kuhlemeier and his colleagues have taken a disarmingly direct approach to test a simple hypothesis. If a unique set of biophysically defined parameters marks the site of leaf initiation, perturbing any one of them might be expected to alter the phyllotaxy. As a means to cause localized changes in the biophysical state of the meristem, microbeads impregnated with the cucumber protein expansin were applied to tomato shoot apices at sites where cells still would be expected to be in an undetermined state (i.e., the I2 position).
This approach was suggested by in vitro studies demonstrating that expansins, a family of cell wall proteins, could drive cell wall extension. Biochemical studies suggest that this extension is mediated by expansin-catalyzed loosening of hydrogen bonds between cell wall polysaccharide microfibrils (see Cosgrove, 1997, for a review). Fourteen days after expansin application, leaflike structures were clearly visible at sites where the expansin-coated beads had been applied. These structures were judged leaflike by several criteria, including their dorsoventrality, the presence of trichomes on their surface, the expression of the rbcS gene, and finally, by their ability to influence sites of future leaf initiation, as shown by a reversal of phyllotaxy (Fleming et al., 1997).
The ability of expansin to induce the formation of leaflike structures at the site of its application is consistent with its possible role in the normal leaf initiation process. This interpretation is strengthened substantially by the in situ localization studies performed by Reinhardt et al. (1998), which are highlighted on the cover of this issue. These studies demonstrate that transcripts of one member of the tomato expansin family, LeExp18, accumulate in the shoot apex at predicted leaf initiation sites as early as one plastochron before primordia begin overt development. This early expression does not appear to be accompanied by any localized increase in cell division rates. These results, in conjunction with the induction of leaflike structures that develop after application of expansin to tomato shoot apical meristems (Fleming et al., 1997), suggest that the initiation of leaves may normally be determined by changes that are most easily defined at the biophysical level.
If leaflike bulges are the immediate consequence of localized, expansin-catalyzed wall softening, how are these physical changes propagated to fix a stable leaflike determination state? The answer to this question must await a more detailed description of the determination process; however, it is reasonable to suppose that one immediate outcome of expansin-mediated deformation might be a change in cellular polarity as cytoskeletal elements become aligned with respect to the new growth axis. Such changes could serve as a basis for changes in the polarities of cell division as well as the polarities that describe physiological relationships based on intercellular connections (McLean et al., 1997).
In conclusion, it appears that Reinhardt et al.'s data support at least some aspects of previously proposed biophysical explanations for leaf initiation (e.g., Green, 1987). Their success will no doubt reawaken interest in many of the classical approaches to understanding leaf initiation and subsequent determination events. With molecular genetic methods providing an increasingly more detailed list of players in this process, it may be especially instructive to consider the behavior of these genes and the proteins they encode in the context provided by such physically based approaches. Clearly, the continued integration of this biophysical perspective with an increasingly detailed understanding of morphogen- and induction-mediated processes can only yield a more complete appreciation for the complex process of pattern formation at the shoot apex.