Expression of maize KNOTTED1 related homeobox genes in the shoot apical meristem predicts patterns of morphogenesis in the vegetative shoot

Development of the shoot system in plants proceeds through the elaboration of organs initiated on the flanks of the shoot meristem(s), which are situated at the apex of the shoot above the most recently formed leaf. How the components of the shoot, such as leaves, stem internode and axillary buds become organized from initially indeterminate cells in the meristem is unclear. This process is, however, a major determinant of plant form, since, for example, it establishes the phyllotaxy, or spatial arrangement, of lateral organs around the apex (Richards, 1951).

The vegetative shoot meristem has been extensively described histologically (reviewed by Steeves and Sussex, 1989; Medford, 1992). A zonation model describes the meristem by cytological characteristics (Ledin, 1954; Wardlaw, 1957), although the presence and extent of different cytological zones differs between species. In general, cells at the apex (‘central zone’) of the meristem are larger and thought of as initials, having some properties in common with animal stem cells (Sussex, 1989). Daughter cells of the initials replenish the subjacent organogenic region (‘peripheral zone’) which is characterized by cytoplasmically dense cells that are metabolically more active. Below this is the rib zone, composed of regular files of cells, which give rise to the stem. Meristems have also been described in terms of layering, since the periphery of the meristem consists of regular layer(s) of cells (Hanstein, 1868), called tunica layers, which surround the central corpus (Schmidt, 1924). Each tunica layer is clonally derived since it is replenished predominantly by anticlinal cell divisions (Satina et al., 1940). Though the number of tunica layers differs widely between plant species (e.g. Kaplan, 1970), in maize and other grasses there is generally only one, the protoderm (Esau, 1965).

Vegetative shoot development is a repetitive process whereby leaf primordia are successively initiated at the meristem. In maize, leaves are initiated singly on the flank of the meristem opposite to the previous leaf, resulting in a distichous phyllotaxy. A periclinal cell division in the tunica on the flank of the meristem precedes the initiation of a leaf primordium, and is closely followed by a reorientation of divisions in the corpus (Ledin, 1954; Esau, 1965). Clonal analysis has indicated that approximately 250 leaf founder cells are present at the time of leaf initiation (Poethig, 1984). As the new primordium enlarges, a wave of cell division spreads around the meristem circumference so that the base of the leaf encircles the meristem, and the disc of leaf insertion is formed (Sharman, 1942; see Fig. 1). Cells at the base of the disc of leaf insertion give rise to the internode (stem) and axillary bud (Sharman, 1942). Thus the vegetative shoot can be viewed as consisting of segments, each comprising of a leaf, leaf node, internode and axillary bud. These segments have also been termed phytomers (Galinat, 1959). Consistent with the histological observations, clonal analysis has shown that in maize the leaf is clonally related to the internode and axillary bud below it (Johri and Coe, 1982; McDaniel and Poethig, 1988), though this is not the case in all plants (e.g. Furner and Pumfrey, 1992; Irish and Sussex, 1992). During shoot ontogeny, the time interval between the initiation of leaf primordia is termed a plastochron (Askenasy, 1880). In addition, the plastochron index describes the position of leaf primordia in relation to the meristem (Lamoreaux et al., 1978). The youngest primordium is thus referred to as the plastochron 1 or P₁ leaf, with successive older leaves being P₂, P₃ etc. (Fig. 1). The position of the incipient leaf is referred to as plastochron zero or P₀.

Despite the many studies of vegetative meristems, there is a paucity of molecular markers with which one could probe their structure (e.g. Fernandez et al., 1991; Medford et al., 1991; Pri Hadash et al., 1992; Fleming et al., 1993). An insight into the organization of meristem domains in maize has come from studies on the KNOTTED1 gene. KN1 is characterized by a series of dominant mutations which affect the development of the leaf blade; mutant leaves show localized zones of extra cell division in all cell layers, resulting in outgrowths or knots (Bryan and Sass, 1941). KN1 encodes a homeodomain protein that is, however, expressed at only very low levels in leaves of wild-type plants (Vollbrecht et al., 1991; Smith et al., 1992). It is expressed strongly in the meristem and in ground tissue of the unexpanded stem, but is absent from the predicted position of the incipient or P₀ leaf in the meristem. Ectopic expression of KN1 in leaf veins of mutant plants correlates with the presence of the knotted phenotype. Based on these observations it was proposed that the KN1 gene product is involved in the maintenance of the indeterminate state of plant meristems, and that its down regulation leads to the initiation of determinate lateral organs such as leaves or floral organs (Smith et al., 1992).

We were interested in the possibility that other maize homeobox genes in addition to KN1 might be useful markers for meristem organization. The homeobox encodes the homeodomain, which is a DNA binding domain (Pabo and Sauer, 1984; Desplan et al., 1988), and it has been shown in many organisms that related homeobox genes have similar functions and specify important determination events in development (Gehring, 1987; Ingham, 1988; McGinnis and Krumlauf, 1992). A family of KN1 related genes was therefore isolated from maize (Vollbrecht et al., 1991, 1993), and has been named KNOX, for KNOTTED related homeobox (Kerstetter et al., unpublished data). The expression patterns of KN1 and three related homeobox genes are described here. One of these genes has been shown to correspond to the dominant leaf mutation ROUGH SHEATH1 (Freeling, 1992; Becraft, Schneeberger, Hake and Freeling, unpublished observation). The results presented here for RS1 expression have also been confirmed elsewhere (R. Schneeberger, personal communication). We show that, like KN1, the three other maize homeobox genes are expressed in shoot meristems and in the developing stem, but not in determinate lateral organs. This collection of probes defines regions in the vegetative shoot apical meristem whose positions predict the site of leaf initiation and the basal limit of the vegetative segmentation unit, and should be useful tools to explore morphogenetic events in the meristem.

The probes were prepared by in vitro transcription with digoxigeninlabeled UTP (Boehringer Mannheim) according to manufacturers instructions, except the ratio of labeled to unlabeled UTP was 1:1. Probes were purified and hydrolyzed as previously described (Jackson, 1991). The clones used were complete or partial maize cDNAs, in some cases subcloned as small fragments to verify that the same expression pattern was obtained when using conserved and non-conserved parts of the cDNA as a probe.

1. KN1, 3 subclones (0.4-0.65 kb, the conserved homeobox region was contained within one of the probes) of the 1.6 kb cDNA clone were transcribed separately, and used alone or as a mixture.

2. KNOX8 (previously called cR18), a 0.6 kb cDNA clone.

3. RS1 (previously called ECB), a 0.8 kb cDNA clone.

4. KNOX3 (previously called ECA), two cDNA subclones of the cDNA (0.6 kb AatII-NaeI fragment and a 0.3 kb NaeI-XhoI fragment, from the middle (including the homeobox) and 3′ part of the cDNA, respectively).

5. ubiquitin, a 0.65 kb PstI-SacI subclone of a maize cDNA clone (Christensen and Quail, 1989).

Other aspects of the molecular characterization of the homeobox clones will be published elsewhere (Kerstetter, unpublished data).

All plants used were from the inbred line B73 (Pioneer Hi-bred International), except for those carrying the Kn1 mutations, as described in the text; wild-type siblings from these lines were used as controls. Seedlings were grown for two to three weeks in a mist room then transferred to larger pots and grown in the greenhouse. Roots were taken from seeds germinated on damp tissues as well as soil grown seedlings. Tissue was fixed and embedded and used as described (Jackson, 1991), with the following modifications: (1) vegetative shoot apices were fixed in FAA (5% formalin, 5% glacial acetic acid, 45% ethanol) overnight at 4°C instead of in formaldehyde, (2) The high stringency post-hybridization washes were in 0.2 × standard saline-citrate buffer (1 × is 150 mM sodium chloride, 15 mM sodium citrate) at 55°C, (3) digoxygenin-labeled probes were used at a concentration of 0.5 ng/μl/kb probe complexity, and detected as described by Coen et al. (1990). Some sections were stained in basic fuchsin, 0.05% in water, and slides were dehydrated through an ethanol series, rinsed twice in histoclear (National Diagnostics), and mounted in Merckoglas (EM Science).

A large number of cDNA clones was isolated from seedling and immature ear cDNA libraries using the KN1 homeobox as a hybridization probe (Vollbrecht et al., 1991, 1993; Kerstetter et al., unpublished data). The expression of four homeobox genes, KN1, KNOX8, RS1 and KNOX3 is shown in median longitudinal sections through the shoot apex of 2–3 week old seedlings in Fig. 2. At this stage the meristem has initiated approximately ten to twelve leaves. The expression of all four genes is limited to the meristem and developing tissues in the stem; there is no detectable message in young leaf primordia or in older leaves. Control sense strand probes showed no hybridization signal (not shown) except for some faint staining over the oldest leaves, which was also seen with antisense probes and was interpreted as being non-specific. The localization of KN1 mRNA (Fig. 2A) is similar to that described for the KN1 protein (Smith et al., 1992); specifically the meristem and developing stem are labeled with the exception of certain provascular cells associated with developing lateral vein traces in the stem (arrowed). KNOX8 is expressed in a similar pattern to KN1, throughout the meristem and developing stem, though the expression is weaker than KN1 and there is no differential labeling of cells associated with provascular traces (Fig. 2B).

The genes RS1 and KNOX3 show a more localized expression pattern in the vegetative shoot apex. Small groups of cells expressing these genes are observed near the base of each leaf (Fig. 2C,D). In the case of RS1, discrete spots of expression are also seen in the meristem, whereas KNOX3 usually shows diffuse staining throughout the meristem (see later). In transverse sections we found that the ‘spots’ are actually rings of expression which encircle the stem at each leaf insertion point (Fig. 2F,G). Since the maize leaf is attached all around the stem, these rings of expression correspond to the base of the disc of insertion of each leaf (Sharman, 1942). This is the region from which the clonally related internode and axillary bud develop (Sharman, 1942; Johri and Coe, 1982; McDaniel and Poethig, 1988). Indeed, near the base of older leaves we observed expression of RS1 and KNOX3 in both the developing internode and axillary bud (Fig. 2H). RS1 also labels provascular cells in the stem (arrowed, Fig. 2C); in adjacent sections we observed that these include the cells of the provascular lateral traces which are not labeled by KN1 (not shown).

Detailed localization of KN1, RS1 and KNOX3 expression in the shoot apical meristem is shown in Fig. 3. KN1 mRNA is expressed throughout the corpus of the meristem with the exception of two groups of cells, one on either side of the meristem, above the P₁ leaf. We interpret these two groups of cells as belonging to a doughnut shaped ring of cells corresponding to the disc of insertion of the P₀ preprimordium. In support of this interpretation, periclinal cell divisions, the first indication of the new primordium, are visible in the tunica opposite and above the P₁ leaf. (arrowed). KN1 mRNA is not detectable in the tunica layer of the meristem.

RS1 and KNOX3 expression was localized as either two spots or a stripe of expression just above the site of insertion of the P₁ leaf (Fig. 3B,C). As in the stem, we conclude that the cells expressing these genes are arranged in a ring which encircles the meristem; depending on the plane of the section this is visualized either as two spots, i.e. if the section is median, or as a stripe if the section is slightly off median and cuts through the edge of the ring of expression, as is the case in the example shown. Whereas all meristems probed with the RS1 probe showed this pattern of spots or a stripe, the majority of meristems probed with KNOX3 showed lower, diffuse staining (e.g. see Fig. 2D).

By probing adjacent sections with KN1, RS1 and KNOX3 we were able to align more precisely their regions of expression. We found that RS1 and KNOX3 are expressed in the same cells at the base of the disc of leaf insertion and that these are within the domain of KN1 expression (Fig. 2E-G, also seen in longitudinal sections (not shown)). This is also the case for the rings of RS1- and KNOX3-expressing cells in the meristem (not shown). Thus in the meristem, the ring of cells expressing RS1 and KNOX3 is just below the ring of cells which do not express KN1.

Although there are characteristic differences between the expression patterns of RS1 and KNOX3, we sought to rule out the possibility of probe cross hybridization. This was accomplished by competition experiments in which a ten-fold excess of unlabeled antisense RNA was included in the hybridization along with the labeled antisense probe. In each case the unlabeled RNA competed out only the signal from the same gene probe (not shown). In addition, for KN1 and KNOX3 we observed the same patterns of expression using non conserved parts of the cDNA as with full length cDNA probes.

We found an intriguing difference between the expression of KN1 protein and mRNA. Whereas the protein is detected in all cell layers of the meristem (Fig. 4B), we have never detected KN1 mRNA in the tunica layer of vegetative, inflorescence or floral meristems (Figs 3A, 4A). Control sections probed for ubiquitin mRNA show staining in all cell layers (Fig. 4C), indicating retention and accessibility of mRNA is not impaired in the tunica.

After the vegetative apical meristem has initiated around 16–22 leaves (depending on the cultivar) it elongates and forms the terminal inflorescence (Russell and Stuber, 1983; Irish and Nelson, 1991). This will later develop into the tassel, bearing male florets, whereas axillary inflorescences will develop into ears which bear female florets. Inflorescence meristems initiate branch primordia; these bifurcate to form spikelet meristems which later produce floral meristems (Postlethwaite and Nelson, 1964). Initial development of tassel and ear inflorescences is similar, but at relatively late stages they can be easily differentiated by differences in branching as well as by selective abortion of floral organs (Cheng et al., 1983).

In serial sections through tassel (Fig. 5A-D) and ear (not shown) inflorescences we observed that all four homeobox genes are expressed evenly throughout the corpus of the inflorescence meristem. In addition, the four genes are expressed in spikelet and floral meristems, but are not expressed in lateral organs such as glume primordia. The discrete rings of expression of RS1 and KNOX3 observed in the vegetative shoot apical meristem are not observed in inflorescence or floral meristems. In addition, KN1 is highly expressed in certain developing vascular strands, and more weakly in the ground tissue of the inflorescence shoot. RS1 and KNOX3 are expressed in the ground tissue of the inflorescence stem, though not expressed in those provascular cells labeled by KN1 (Fig. 5A,C,D, arrowhead). Since these genes are expressed in all apical meristems of the shoot system, we were interested to see if they might also be expressed in meristems of the root system. However, none of the homeobox genes is expressed in primary (Fig. 5G-J) or secondary (not shown) root apices.

The mutant phenotype of the knotted mutation Kn1-N2 is caused by ectopic expression of KN1 in developing lateral veins of the leaf blade; KN1 is immunohistochemically undetectable in wild-type leaves (Smith et al., 1992). We reasoned that if RS1 or KNOX3, which show expression in a subset of KN1-expressing cells, or KNOX8, which shows a similar pattern of expression to KN1, are directly regulated by KN1, then they might also be ectopically expressed in knotted leaves. However, in plants carrying two different dominant KN1 alleles, Kn1-N2 (Fig 5K-O) and Kn1-Z3 (not shown) (Freeling and Hake, 1985), KN1 mRNA is ectopically expressed in leaf vascular strands, as predicted, but no expression of the other 3 homeobox genes was detected in adjacent sections.

In summary, the expression of the four homeobox genes differs in the vegetative apex and meristem, whereas in inflorescence and floral meristems their expression is more similar. Although KNOX8, RS1 and KNOX3 are expressed in some of the same cells which express KN1, we have found no evidence for direct interactions controlling their expression.

Plant development differs fundamentally from animal development, in that only a basic body plan is laid down during plant embryogenesis, and new organs are initiated throughout the plant life cycle. We show here, however, a common feature in plants and animals is that pre-patterns of homeobox gene expression predict morphogenetic events.

Morphological studies (Sharman, 1942) and clonal analysis (Johri and Coe, 1982; McDaniel and Poethig, 1988) have shown a lineage relationship in maize between the leaf and the internode and axillary bud below it, and together these make up a segmentation unit of the vegetative shoot or ‘phytomer’. Using the different homeobox probes we can visualize domains within the organogenic zone of the meristem which will give rise to different parts of the segmentation unit. Absence of KN1 expression marks the presumed position of the incipient (P₀) leaf, while RS1 and KNOX3 are expressed in a ring of cells just below the P₀ leaf (summarized in Fig. 6). In the unexpanded shoot it is difficult to recognise the structures labeled by RS1 and KNOX3. However the position of their expression in relation to the P₀ preprimordium and at the base of the disc of insertion of young leaves leads us to believe that this expression corresponds to the incipient internode and axillary bud (Sharman, 1942). This is supported by the visualization of RS1 and KNOX3 expression in the developing internode and axillary bud near the base of older leaves; they may also be expressed in the leaf node which is morphologically indistinct in the unexpanded shoot. We conclude that RS1 and KNOX3 expression in the meristem corresponds to incipient internode and axillary bud, which make up the basal limit of the segmentation unit of the shoot. Why only the periphery of the developing internode, and not the internal ground tissue, is labeled by RS1 and KNOX3 is unclear, though this internal tissue is derived from a different part of the meristem, the rib zone.

It is interesting to speculate whether the earliest expression of KN1, RS1 and KNOX3 precedes the determination of the position of a new primordium. This determination event has been estimated in Lupinus albus to occur about half a plastochron before the emergence of the primordium (Snow and Snow, 1933). In the examples shown here, one or two periclinal divisions, which have been described as the first indication of the new primordium (Ledin, 1954; Esau, 1965), were visible in the tunica layer at the predicted position of the P₀. We have also occasionally observed two rings of cells that do not express KN1 and two rings of RS1 and KNOX3 expression above the P₁ leaf, suggesting that an even earlier differentiation event is marked by these homeobox genes. This suggests that these changes in gene expression patterns precede or at least are concomitant with the determination of the position of primordia in the meristem. Although a striped pattern of expression within the meristem is consistently observed for RS1, KNOX3 is often expressed diffusely throughout the meristem, though always in discrete stripes in the stem. It is possible that the stripe of KNOX3 expression in the meristem arises only at a specific stage during the plastochron.

How these patterns of homeobox gene expression arise in the meristem is unclear, though two major hypotheses have been forwarded to explain the positioning of primordia. The first suggests that inhibitory substances are responsible for positioning the new primordium at a given spacing relative to preexisting primordia (e.g. Wardlaw, 1949). The second hypothesis suggests that reorientation of cell wall cellulose microfibrils allows the change in growth polarity required for outgrowth of the primordium (e.g. Green, 1985). If the genes discussed here are indeed involved in determining the position of primordia in the meristem, then presumably their expression is in response to one or both of these factors.

In contrast to the differing expression patterns in the vegetative shoot meristem, these homeobox genes are all expressed uniformly throughout inflorescence and floral meristems. We view the uniform expression in the inflorescence meristem as consistent with our hypothesis about the expression patterns in the vegetative meristem. The products of the inflorescence meristem, branch primordia, are also indeterminate. Therefore all cells in the inflorescence meristem and branch primordia, being indeterminate, are labeled by KN1, and since cells of the inflorescence meristem will form lateral indeterminate buds or inflorescence shoot internode, they are also all labeled by RS1 and KNOX3. However, it is unclear why the patterns of RS1 and KNOX3 expression do not revert to stripes in floral meristems. This could be due to the compressed nature of the floral meristem, or may reflect differences in the state of determination of floral and vegetative meristems. None of the homeobox genes is expressed in the root apical meristem, suggesting that they are involved in functions specific to the shoot. This may reflect basic differences in the properties of the root and shoot apices, one of which is that lateral organs are initiated from the shoot meristem, but not from the root meristem.

Since the expression domains for these genes are similar, we were interested in the possibility that their expression might be controlled by direct regulatory interactions, as has been documented for homeobox genes controlling Drosophila development (e.g. Ingham, 1988). We therefore looked to see if KNOX8, RS1 or KNOX3 are expressed in knotted mutant leaves, where KN1 protein is ectopically expressed (Smith et al., 1992). Whereas ectopic expression of KN1 mRNA was seen in knotted leaves, no ectopic expression of the other genes is observed. A negative result does not disprove an interaction, however, since other factors that may be absent from leaves are probably required for the expression of these genes in the meristem.

KN1 mRNA is not detected in the tunica layer of meristems, although KN1 protein is present. We do not believe that these observations are artefactual, since ubiquitin mRNA is detected in the epidermis, and we have detected KN1 mRNA in the epidermis of tobacco plants overexpressing the KN1 cDNA under the control of the cauliflower mosaic virus 35S promoter (D. J., R. Williams and N. Sinha, unpublished data). Additionally we do not believe that this observation could be explained by cross reaction of the polyclonal antibody with another homeodomain protein, since two different anti-KN1 monoclonal antibodies, each of which recognize a part of KN1 outside of the conserved ELK-homeodomain part of the protein (Vollbrecht et al., 1993; D. J. and L. Smith, unpublished data), also label the nuclei of cells in the tunica. Whilst one explanation for the discrepancy could be cell-specific differences in stability of KN1 mRNA, another hypothesis is relevant, since a signal has been proposed to emanate from KN1-expressing cells. Clonal analysis showed that for knot formation to occur in leaves, the KN1 product is required only in the vascular/inner mesophyll layer, although changes in cell division and differentiation occur in all cell layers (Gelinas et al., 1969; Hake and Freeling, 1986; Sinha and Hake, 1990). It was proposed that a signal moves from the vascular/inner mesophyll layer to instruct the epidermis and other layers to divide. Assuming the observations described here are not artefactual, we hypothesize that KN1 protein itself could be the signal that moves between cells, though this movement would have to be short range or selective, since the protein does not move into the P₀ or young leaf primordia, and in knotted leaves does not move to adjacent veins. Whilst movement of transcription factors between cells has not been previously described, such a mechanism was also hypothesized for the establishment of stripes of hairy gene expression in the cellularized growth zone of the short germ band Tribolium embryo (Sommer and Tautz, 1993).

The expression patterns presented here suggest a common function of homeobox genes in the animal and plant kingdoms. In animals, homeobox genes are required for setting anterior-posterior coordinates in both segmented and non-segmented systems (e.g. Kenyon and Wang, 1991; Dolle et al., 1989). Even though plants and animals diverged at the unicellular stage, there may be some property of homeobox genes which made them well suited for the role of defining cell fate and for defining boundaries in morphogenetic fields during the evolution of multicellular organisms. Three dominant leaf mutations in maize, Kn1, Rs1 and Lg3 have now been found to be caused by lesions in homeobox genes (Freeling, 1992). The results presented here and elsewhere (Smith et al., 1992, Becraft, Schneeberger, Hake and Freeling, unpublished data) indicate that several members of this class of homeobox genes are not normally expressed in leaves, though the dominant leaf phenotypes of mutations in these genes and their restricted expression patterns in the meristem suggest roles in the control of cell fate.

We would like to thank Richard Schneeberger and Michael Freeling for discussing unpublished data, Toshi Foster for the SEM in Fig. 1, Julie Mathern for Fig. 6, Peter Quail for the gift of the ubiquitin cDNA clone, Laurie Smith for Fig. 4B, Steve Ruzin of the NSF Plant Center for advice and use of his microscope and members of the Hake laboratory for comments on the manuscript. D. J. is supported by a SERC (UK) - NATO postdoctoral fellowship. B. V. and S. H. are supported by the USDA and NSF.