Heterochronic effects of glossy15 mutations on epidermal cell identity in maize

During development, higher organisms pass through a series of discrete phases. The transition between these phases can be sudden, as is the case for many insects and amphibians, or gradual. The proper timing of each of these phases is necessary for normal growth and development of an individual. At the same time, changes in the relative timing of different developmental events can be an important source of novel traits, and indeed appear to have played an important role in both plant and animal evolution (Lord and Hill, 1987; Gould, 1982; Raff and Kaufman, 1983). Resolving the mechanisms of temporal regulation during development is therefore an important problem in biology.

Postembryonic shoot development in plants can be divided into three phases: juvenile vegetative, adult vegetative, and reproductive (Poethig, 1990). The polar nature of growth of the shoot apical meristem separates these phases spatially as well as temporally. Structures produced during an early phase of development are located at the base of the plant and structures produced later are located at a higher position. The timing of these phases specifies the morphology of the plant because the onset or duration of a particular phase determines the spatial pattern of expression of certain traits (reviewed by Conway and Poethig, 1993). However, it is rarely clear whether spatial or temporal cues are responsible for the onset and termination of each phase, since the expression of a developmental phase in time and space occurs simultaneously.

Maize is an excellent model system for studying phase change because the differences between juvenile and adult development have been well characterized and are superimposed on a simple repeating segmental unit called the phytomer (Poethig, 1990; Galinat, 1959). The juvenile, or basal, part of the shoot possesses axillary roots, tillers, and leaves that have a number of distinctive epidermal traits, the most obvious of which are the production of visible epicuticular wax and absence of epidermal hairs. The adult part of the shoot lacks prop roots, possesses ears in place of tillers, and has pubescent leaves that lack visible epicuticular wax. During the switch from juvenile to adult development, there is a ‘transition zone’ in which the shoot produces phytomers that express both juvenile and adult characteristics, usually in discrete domains of the phytomer.

The existence of visible, cellular traits that distinguish juvenile and adult phases of shoot development facilitates the analysis of mutations that affect the expression of these phases. Four dominant, gain-of-function mutations — Teopod1 (Tp1), Tp2, Tp3, and Corngrass (Cg) — that prolong the juvenile phase of development in maize have previously been described (Poethig, 1988a; Galinat, 1966, 1954a,b; Singleton, 1951; Whaley and Leech, 1950; Weatherwax, 1929; Lindstrom, 1925). These mutations cause phytomers in adult positions to express juvenile traits, and transform reproductive structures into vegetative ones. Because these mutations affect the expression of all known juvenile traits, it is likely that they play a regulatory role in phase change. The nature of this function cannot be accurately defined, however, until the loss-of-function phenotype of these and other related genes is determined.

To identify loss-of-function mutations in genes that regulate the juvenile phase, we screened for mutations that truncate the expression of juvenile traits. In practice, the easiest way to do this is to search for mutations that affect epicuticular wax because this trait is readily visible in young seedlings. Leaves with epicuticular wax have a dull grey-green appearance, whereas those without wax have a ‘glossy’ appearance like that of adult leaves. We began by examining the numerous existing glossy mutations to determine if any of them had pleiotropic effects on other phase-specific traits. Alleles of one gene, Glossy15 (Gl15), are unique in only affecting epicuticular wax on late juvenile leaves (Coe et al., 1988). Scanning electron microscopy reveals that the first two leaves of gl15 seedlings have normal wax extrusions whereas later leaves have reduced amounts like other glossy mutants (Bianchi and Marchesi, 1960). The nature of the defect in epicuticular wax production in gl15 plants has been investigated on the assumption that Gl15 controls an enzymatic step in wax biosynthesis (Bianchi et al., 1979, 1985). Here we demonstrate that Gl15 actually plays a more general role in leaf development. Our results suggest that Gl15 is required for the expression of juvenile epidermal traits in late juvenile leaves and either directly or indirectly suppresses the expression of adult epidermal traits. The interaction of gl15 with Tp1 and Tp2 demonstrates that Gl15 acts downstream of these genes and is required for their effect on epidermal development.

All of the mutant alleles of gl15 used in this study originated spontaneously. Stocks homozygous for gl15-2 and segregating for waxy1 (wx1) were provided by G. F. Sprague; stocks of gl15-1 wx1 were acquired from the Maize Genetics Stock Center; and stocks of gl151106 were kindly provided by Julie Vogel (University of California, Berkeley). wx1 is an endosperm mutation that maps 10 cM from gl15. Prior to phenotypic analysis, mutations were crossed three or four times to the W23 or A632Ht inbreds and self-crossed to produce families segregating for gl15 mutants. Families segregating gl15 and Tp1 or Tp2 were generated by crossing the original stocks of gl15-1 or gl15-2 to Tp/+ (W23) plants and then backcrossing the Tp plants in the F₁generation, as pollen or ear parents, to the respective gl15 parental stock. Families segregating gl15 and teosinte branched1 (tb1) were generated by crossing the original stock of gl15-2 wx1 by tb1 and then self-crossing the F₁progeny to produce an F₂family segregating for both mutations. Families segregating gl15 and Ragged leaves1 (Rg1) were generated by crossing the original stock of gl152 as females by Rg1/+ plants and backcrossing the Rg1 plants of the F₁ as females by the original gl15-2 stock.

Unless otherwise noted, plants were grown in the field in the summer of 1992 or 1993. Mutants were always compared to their wild-type siblings. Plants were scored visually for the presence of wax and hairs on the leaf blade and measured a few days after they had fully expanded, as scored by the emergence of the ligule; this was important because basal leaves senesce early in shoot development. A leaf was scored as having wax if any wax was present on the leaf blade. Similarly, a leaf was scored as having hairs even if only one macrohair was visible on the blade, excluding the hairs along the margin of the leaf blade. Prickle hairs and microhairs, which are too small to observe macroscopically, were not included in this study. All axillary buds were numbered according to the position of the subtending leaf, and prop roots were numbered according to the position of the leaf at the next node. Leaf dimensions were measured in plants of a family segregating gl15-2 in an A632 background that were grown in sandy soil in 15 cm pots in the greenhouses of the University of Pennsylvania under long day conditions. Leaf width was measured at the widest point of the blade, and leaf length was measured from the ligule to the tip of the blade. In tb1 plants, a phytomer was scored as having a tiller if the axillary shoot extended above the ligule, but shoots above phytomer 10 were excluded because ears normally developed at these positions in wild-type plants.

In families in which Tp plants had leaves in the tassel, the first tassel phytomer was considered to be the first phytomer with a spikelet in the axil of the leaf, or the first phytomer with more than one leaf originating from the node. The following classification scheme was used to quantify the severity of Tp tassel phenotypes: Teopod tassels were given a score of one if the tassel was branched; two if the tassel was unbranched but otherwise wild-type in appearance; three if the glumes were elongated; four if one quarter of the tassel was leafy; five if half of the tassel was leafy; six if three quarters of the tassel was leafy; and seven if the tassel was completely leafy.

One to two cm-long samples encompassing the midrib to the margin of the leaf blade were taken at the midpoint of the leaves of gl15 plants and their wild-type siblings grown in 15 cm pots in sandy soil under long day conditions in the greenhouses of the University of Pennsylvania. The abaxial surface of the samples was abraded with silica powder to facilitate staining, and then samples were fixed at room temperature in 3 parts ethanol to 1 part glacial acetic acid and stored in this solution for several weeks. Samples were stained overnight at room temperature in 1 part 0.05% Toluidine Blue (0.01 M sodium acetate, pH 4.4) to 15 parts H₂O. Photographs were taken of the adaxial surface midway between the midrib and the margin with Kodak Ektar color print film using an 80A color filter and brightfield optics on an Olympus BH-2 microscope.

Leaf samples were taken from the leaf blade midway along its length and midway between the midrib and the margin from the same plants used for the toluidine blue staining. Samples were fixed at room temperature in a solution of 3% glutaraldehyde, 1.5% acrolein and 1.6% paraformaldehyde (0.5 M sodium phosphate buffer, pH 6.8). Samples were post-fixed in 1% osmium tetroxide (0.5 M sodium phosphate buffer, pH 6.8). Samples were then dehydrated through an ethanol series and embedded in Spurr’s resin. Transverse 1 μm sections were stained with 0.2% toluidine blue (2.5% sodium bicarbonate, pH 11). Epidermal cell shape and cuticle thickness were measured by tracing photographs of each cross section on a digitizing pad using the MacMeasure program (anonymous ftp from alw.nih.gov in the directory/pub/image) running on a Macintosh computer. The cell shape factor is calculated as 4π(area)/(perimeter)². Only abaxial epidermal cells were measured to reduce the variability caused by the presence of bulliform cells on the adaxial surface of the leaf; variability was also minimized by excluding guard cells and the four surrounding cells from these measurements. Cell shape measurements were taken from 10-15 cells per leaf for four plants of each genotype except for leaves 8 and 9 for which only two plants were used for each genotype. Cuticle thickness was measured over the center of abaxial epidermal cells; five measurements were made per leaf on the same sections that were used for the cell shape measurements. Embryos from wild-type and wx1 kernels of the cross Gl15 Wx1/gl15-1 wx1; W23 X gl15-1 wx1/gl15-1 wx1; W23 were sectioned longitudinally to determine the effect of gl15 on leaf number. Seeds were imbibed in water overnight and embryos were excised from the kernel. Embryos were then fixed and sectioned as described above.

Seeds from a family segregating phenotypically wild-type (Gl15/gl15-2) and gl15 plants (gl15-2/gl15-2) in a 1:1 ratio were grown in the plant growth facilities of the Scottish Agricultural College, Edinburgh in the spring of 1991. Plants were grown in a peat/vermiculite mixture in 5 cm peat pots and later transplanted into 18 cm plastic pots. Fertilizer and pesticides were applied as needed. Long day (LD) conditions (16 hours light : 8 hours dark) were provided at a constant temperature of 20°C and a photon flux density of 132 μEinsteins m⁻²s⁻¹. Short day (SD) conditions (10 hours light:14 hours dark) were provided in growth cabinets at a constant temperature of 22°C and a photon flux density of 112 μEinsteins m⁻²s⁻¹. The number of leaves whose tip or ligule was visible was recorded before and after each treatment. The number of leaves initiated by the shoot was determined before and after each treatment by dissecting five or six plants that were grown under the same conditions as the experimental plants. After 35 days, all plants were moved to the greenhouse and grown to maturity under LD conditions with supplementary lighting.

B-A translocations were used to generate deficiencies for the long arm of chromosome 9. B-A chromosomes frequently undergo nondisjunction at the second pollen mitosis, producing sperm duplicated and deficient for the B-A chromosome (Roman, 1947; Beckett, 1978; Birchler and Alfenito, 1993). If the deficient sperm fertilizes the egg cell, the resulting plant will be hypoploid for the chromosome arm translocated to the B chromosome. gl15 ears were initially crossed by plants heterozygous for TB-9Lc. F₁ plants carrying the translocation were then used as male parents in a second cross to gl15 (i.e., gl15/gl15 × gl15/TB-9Lc) to generate families segregating homozygous and hypoploid gl15 plants. The presence of TB-9Lc in the F₁plants was confirmed by crossing these plants as males to plants heterozygous for defective kernel13, an endosperm mutation that is uncovered by TB-9Lc. Families segregating homozygous and hypoploid gl15 plants were grown for phenotypic analysis under long day conditions in sandy soil in 15 cm pots in the greenhouses of the University of Pennsylvania. Mutant plants with thin leaves and 50% pollen sterility were assumed to be gl15/−; mutant plants with broad leaves and completely normal pollen were assumed to be gl15/gl15.

Two spontaneous, recessive alleles of Glossy15 were examined in detail in this study. gl15-1 is the original allele of this locus, and gl15-2 is an allele obtained from Dr George Sprague. Because the origin of gl15-2 was unclear, we examined the genotype of these two mutations at RFLP loci located 1.8 cM proximal and 0.9 cM distal to Gl15. gl15-1 and gl15-2 are associated with different polymorphisms at both loci, strongly suggesting that these mutations arose independently (data not shown). This conclusion is supported by the observation that these mutations have slightly different phenotypes in every genetic background examined, the phenotype of gl15-1 being more severe than that of gl15-2 (Table 1). gl15-1106 is phenotypically similar to gl15-1 and gl15-2; however, because this allele was not available in an inbred background, a careful analysis of its phenotype was not undertaken.

Wild-type juvenile phase leaves possess visible epicuticular wax and epidermal cells that are circular in cross-section. These cells have weakly crenulated lateral walls and a thin outer wall/cuticle, which both stain purple with toluidine blue. In contrast, adult phase leaves lack epicuticular wax crystals, and have epidermal cells that are rectangular in cross-section. These cells have highly crenulated lateral walls and a thick outer wall/cuticle that both stain turquoise with toluidine blue. Adult phase leaves also produce epidermal hairs, which are completely absent from juvenile leaves. The macrohairs, which are the largest and most visible hairs, arise within rows of bulliform cells; bulliform cells are specialized cells that stain purple with toluidine blue in both juvenile and adult leaves.

gl15-1 and gl15-2 were originally identified because of their effect on the production of epicuticular wax. gl15-1 and gl152 block epicuticular wax production starting at leaf 2, 3, or 4, depending on the genetic background. To determine whether the precocious loss of epicuticular wax was caused by a temporal shift in the transition from juvenile to adult development, we examined the expression of several phase-specific traits in gl15-1 and gl15-2 mutants. Both mutations affect every juvenile and adult epidermal trait that we examined, with the exception of cuticle thickness (see below). In each case, the juvenile form of a trait is replaced prematurely by the adult form. Because the expression of epicuticular wax and epidermal hairs are tightly correlated with the expression of other juvenile and adult traits, we used these two readily visible traits to characterize the effects of gl15 alleles in different genetic backgrounds.

Table 1 shows the average position for the end of expression of epicuticular wax and the onset of expression of epidermal hairs in gl15 plants and their wild-type siblings in two different backgrounds. The phenotype of gl15 alleles is best illustrated by gl15-1. In a W23 inbred background, gl15-1 seedlings only produce epicuticular wax on the first two leaves of the shoot, whereas their wild-type siblings have epicuticular wax on the first five leaves. In gl15-1 plants, hairs are present on leaves 3 and above but do not appear in their wild-type siblings until leaf 4 or 5 (Table 1 and Fig. 1). Partially in leaf 2 and completely in leaves 3 and above, the epidermal anatomy of gl15-1 leaves becomes strikingly similar to that of adult leaves. However, leaf 1 is anatomically indistinguishable from wild type.

The phenotype of gl15-1 is even more striking in an A632 background because of the relatively prolonged juvenile phase of this inbred line. Wild-type A632 seedlings have nine leaves with at least some epicuticular wax and begin producing epidermal hairs starting with leaf 5. In contrast, gl15-1 (A632) seedlings only have three leaves with epicuticular wax and begin producing epidermal hairs starting with leaf 3. gl15-1 has a more severe effect on epicuticular wax than gl15-2 in both backgrounds tested; e. g., gl15-1 (A632) plants have three leaves with wax whereas gl15-2 (A632) plants have four or five leaves with wax. Despite this difference in their effect on epicuticular wax, gl15-1 and gl15-2 have the same effect on epidermal hairs, causing them to appear at the same position, usually by leaf 3 in both A632 and W23 (Table 1).

gl15 mutations affect the expression of adult traits less than they affect the expression of juvenile traits. In a W23 background, for example, gl15-1 reduces the number of leaves with epicuticular wax by three (from 5 to 2), but only accelerates the appearance of epidermal hairs by one leaf (from 4 to 3). This disparity is even more pronounced in an A632 back-ground where gl15-1 reduces wax production by six leaves and advances hair production by only one or two leaves. This unequal effect reduces the overlap between the expression of juvenile and adult traits, producing plants with a shorter transition zone.

Fig. 1 shows the effect of gl15-1 on the histochemistry and shape of epidermal cells in the W23 inbred line. In this inbred line, leaf 5 is the first leaf with epidermal cells that exhibit the tourquoise color and highly crenulated lateral walls characteristic of adult leaves; epidermal cells of the first four leaves stain a uniform purple color and have the weakly crenulated lateral walls characteristic of juvenile leaves (Fig. 1A-D). In gl15-1 (W23) plants, the first leaf to exhibit the staining pattern and cell shape typical of adult leaves is leaf 3 (Fig. 1G). Some evidence of these adult traits can be seen as early as leaf 2 (Fig. 1F), although most of the epidermal tissue of leaf 2 is juvenile in nature. Leaf 1 is the only leaf in gl15-1 (W23) plants that appears to be completely juvenile. gl15-2 has a similar effect on the staining pattern and crenulation of epidermal cells, but does not accelerate the change in expression of these traits as much as gl15-1 (data not shown).

Histological analysis of mutant and wild-type leaves reveals that gl15 mutations also affect the shape of epidermal cells in cross section (Fig. 2). In wild-type W23 plants, the epidermal cells of the first four leaves are round in cross section whereas the epidermal cells of leaf 5 are rectangular (Fig. 2A-D). Epidermal cells of gl15-1 plants become rectangular as early as leaf 3 (Fig. 2G). Epidermal cells of leaf 2 of gl15-1 plants appear to have a shape intermediate between juvenile and adult cells (Fig. 2F). The differences in cell shape can be quantified using the formula 4π(area)/(perimeter)², which produces a value of one for circular objects and an increasingly smaller number as the shape of an object approaches a line. Analyzing the shape of epidermal cells in this way demonstrates that gl15-1 plants have a completely adult epidermal cell shape by leaf 3 and have a partial adult shape by leaf 2 (Fig. 3A). In 0.9 contrast, the transition from a thin, juvenile cuticle to the thicker adult cuticle occurs at the same position in gl15-1 and wild-type plants (Fig. 3B). gl15-2 also affects the shape of leaf epidermal cells without affecting the thickness of the outer wall/cuticle (data not shown).

To determine if gl15 mutations have a general effect on the vegetative phase of the shoot, we examined the expression of a variety of morphological traits in families segregating mutant and wild-type plants. We specifically focussed on traits that are affected by the Teopod mutations, since these loci appear to be major regulators of the juvenile vegetative developmental program. gl15-1 had no obvious effect on the expression of any of the gross morphological traits affected by the Teopod mutations. In particular, it did not affect the number of phytomers producing prop roots (a juvenile trait), total leaf number, the position of the ear on the main stalk, or ear and tassel morphology (Table 2). The same is true for gl15-2 (data not shown). Because changes in leaf shape are associated with vegetative phase change in many species, such as English ivy (Poethig, 1990; Brink, 1962), and because the Teopod mutations affect leaf shape (Poethig, 1988a), we also measured the length and width of leaves in gl15-2 plants and their wild-type siblings in an A632 background. Leaves of gl15-2 and wild-type plants have identical dimensions at all positions (Fig. 4).

Tillers are elongated lateral branches that terminate in a tassel and are associated with juvenile phytomers in genetic backgrounds that are capable of forming tillers. Adult phytomers either fail to develop visible axillary shoots or form ears. teosinte branched1 (tb1) is a recessive mutation that induces juvenile phytomers to form tillers and transforms the terminal inflorescence of ear shoots into tassels. Double mutants between tb1 and Tp1 or Tp2 produce a very large number of tillers (Poethig, 1989), suggesting that production of tillers may be controlled by the juvenile vegetative program and might therefore be affected by gl15. To determine whether gl15 affected the production of tillers, we constructed double mutants between gl15-2 and tb1. tb1 and gl15-2; tb1 plants both produced tillers from approximately six or seven phytomers starting with the second phytomer (Table 3), demonstrating that gl15 has no effect on tiller initiation.

gl15-2 does have a significant effect on the character of tiller leaves, however. The vegetative character of a tiller reflects the vegetative phase of the node from which it arises. A tiller that arises from a node that has juvenile epidermal traits produces juvenile phytomers before switching to an adult phase, whereas a tiller that arises from a node that has adult epidermal traits produce only adult phytomers. For example, in the tb1 mutants described above, the primary axis had five leaves with wax and produced epidermal hairs starting with leaf 4 or 5 (Table 3). Tillers from the axil of leaf 2 switched from the juvenile to the adult phase after producing three leaves, and tillers from the axil of leaf 3 underwent this transition after producing two leaves. The gl15-2; tb1 plants in this experiment produced only two leaves with wax and began to produce epidermal hairs starting with leaf 3. All of the tillers on these double mutants produced leaves without epicuticular wax and with epidermal hairs, reflecting the character of the node from which they arose. Thus, gl15 affects the identity of axillary branches in addition to affecting the identity of leaves on the main stem.

Because there are no obvious phase-specific anatomical traits in the internal tissue of the leaf (Bongard-Pierce and Poethig, unpublished observations), it was necessary to find another type of marker in order to study the effect of gl15 on this tissue. A screen of several mutations reputed to be expressed late in shoot development revealed that the Ragged leaves1 (Rg1) mutation is appropriate for this purpose. Rg1 is a dominant mutation that conditions necrotic lesions on the leaves of maize plants (Brink and Senn, 1931). Rg1 causes apparently random cytolysis of the mesophyll cells early in leaf ontogeny, and indirectly affects the epidermis as a result of this effect on mesophyll cells (Mericle, 1950). The appearance of these lesions is correlated with the appearance of adult traits. Completely juvenile leaves do not express a Rg phenotype, whereas transition leaves usually develop only small chlorotic lesions. Later leaves develop large lesions that cause tears through the leaf. In Rg1; Tp double mutants the appearance of the Rg1 phenotype is delayed, as are other adult traits (unpublished observations). For these reasons, we believe the expression of Rg1 can be used as a marker for adult vegetative development in the internal tissue of the leaf.

To determine the effect of gl15 on the expression of Rg1, families segregating Rg1 and gl15-2; Rg1 double mutants were constructed. The Rg1 phenotype was first visible on leaf 7 in both Rg1 and Rg1; gl15-2 plants, despite the fact that Rg1 plants had nine leaves with wax and Rg1; gl15-2 double mutant plants had only three leaves with wax (Table 4). Although the expression of Rg1 is an indirect measure of the phase of the internal tissue, the fact that gl15 has no effect on Rg1 expression strongly suggests that this tissue is not affected by gl15. We conclude that gl15 accelerates the transition from juvenile to adult development in the epidermis but has no effect on internal cell layers.

In woody species the developmental phase of a plant is usually defined by its reproductive behavior, i.e. by the production of flowers or by the plant’s ability to flower in response to an environmental stimulus (Hackett, 1985; Poethig, 1990). Because the reproductive phase of the shoot is correlated with its vegetative morphology, these aspects of shoot development are rarely considered separately. Given this correlation, it was important to examine the effect of gl15 on reproductive development in some detail.

In the field, gl15 had no visible effect on reproductive development. The ear and tassel develop at the same position in wild-type and gl15 siblings (Table 2). Anthesis, an absolute measure of reproductive maturity, occurs at the same absolute time in gl15-2 and wild-type plants — after 65 days in an A632 background in the summer of 1992. These observations suggest that gl15 does not affect the timing of the differentiation of reproductive structures.

Another measure of the onset of the reproductive phase of development is the ability of the plant to respond to a photoinductive stimulus (Hackett, 1985; Poethig, 1990). In many species, plants enter a reproductively competent state long before they initiate reproductive structures (Bernier, 1988; Bernier et al., 1981). Although temperate varieties of maize will flower without a photoinductive stimulus, tassel initiation is advanced by short day (SD) conditions after the shoot reaches a certain stage of development (Tollenaar and Hunter, 1983; Bassiri et al., 1992). When induced, the maize meristem stops producing leaves and initiates the male inflorescence (tassel). Thus, in addition to undergoing anthesis earlier, photoperiodically induced plants have fewer leaves than non-induced plants.

To test the effect of gl15 on the sensitivity of the shoot to a flower inducing stimulus, mutant and wild-type siblings were exposed to SD conditions at various stages of shoot development (Table 5). Neither wild-type nor gl15-2 plants responded to SD conditions during plastochrons 8-14 (5-29 days after planting; DAP). However, plants of both genotypes subjected to SD conditions during plastochron 14-15 (29-35 DAP) produced approximately two fewer leaves than plants grown under long day (LD) conditions (significant at P<0.01). This 6-day SD treatment achieved only partial induction since the reduction in leaf number was not as great as when SD conditions were applied continuously from plastochron 8 to tassel initiation (5-35 DAP). This may have been caused by a significant difference in the total amount of radiation received by the plants in these two treatments. Because several parameters, including total irradiance and temperature, differed between the SD and LD conditions of this experiment, the exact nature of the environmental stimulus responsible for the flowering response in these plants is uncertain. It is significant, however, that gl15 has no effect on the timing of the period in which plants are sensitive to this floral stimulus, either in absolute time (DAP) or in developmental time (plastochron and leaf number at start of SD treatment).

In some genetic backgrounds the number of leaf primordia present in a mature embryo, generally five (Hubbard, 1951), correlates well with the number of juvenile leaves. The expression of juvenile traits may therefore be regulated by genes that act during seed development. If this is the case, then the reduction in the number of juvenile leaves in gl15 plants could be caused by a reduction in the number of leaves initiated in the embryo. To test this hypothesis we sectioned mature gl15-1 embryos (n=5) and their wild-type siblings (n=5), using the waxy1 mutation as a marker for gl15-1 seeds. All of these embryos had five leaf primordia (data not shown). Consequently, the reduction of the juvenile phase of development in gl15 is not caused by, or correlated with, a reduction in the number of embryonic leaves.

An alternative possibility is that gl15 acts by reducing the rate of leaf initiation, causing the shoot to initiate fewer leaves while it was in the juvenile phase. This is not the case, however, because mutant seeds initiate the same number of leaf primordia as wild-type seeds and appear to mature at the same time as wild-type seeds. Furthermore, the fact that gl15 does not affect total leaf number (Table 2) and the time to anthesis implies that this mutation also has no effect on the rate of leaf initiation after germination. A direct measurement of the rate of leaf initiation in mutant and wild-type plants confirmed this prediction (data not shown).

In order to determine the function of the normal Gl15 gene it is important to know its null phenotype. If either gl15-1 or gl15-2 is a null mutation, then Gl15 is not absolutely required for juvenile development since these mutations do not completely eliminate the expression of juvenile traits. On the other hand, it is possible that Gl15 is required for the expression of the juvenile phase of development and the short juvenile phase in gl15-1 and gl15-2 plants is a consequence of the hypomorphic nature of these mutations.

To distinguish between these possibilities, families segregating homozygous gl15/gl15 plants and hemizygous gl15/ – plants were generated by the following cross: gl15/gl15 × TB9Lc/gl15. The 9L breakpoint of TB-9Lc is located proximal to gl15, at cytological position 9L.10. Homozygous and hemizygous plants were expected to have identical phenotypes if the allele involved in this cross was a null mutation; in the case of a hypomorphic mutation, homozygous plants were expected to have a less severe phenotype than hemizygous plants.

However, for both gl15-1 and gl15-2, hemizygous plants had a lesssevere phenotype than homozygous gl15 plants (Table 6). Since the hypoploid plants produced by TB-9Lc are hemizygous for almost the entire long arm of chromosome 9, it is likely that this result is due to the presence of modifiers of juvenile gene expression within this deficiency rather than a direct effect of the dose of gl15-1 and gl15-2. Consequently, no conclusion can be made about the nature of these mutations.

Tp1 and Tp2 are semi-dominant, gain-of-function mutations that prolong the expression of a large number of juvenile traits, including traits affected by gl15 mutations. To determine if Gl15 is part of the same genetic pathway, families segregating gl15, Tp1 (or Tp2), wild type, and Tp1 (or Tp2); gl15 double mutants were constructed. gl15-2 is almost completely epistatic to Tp2 (Table 7) and Tp1 (data not shown) for those traits that gl15 affects. The results are the same for gl15-1 (data not shown). gl15; Tp2 plants had only slightly more (<1) glabrous leaves, and only slightly more (<1) leaves with epicuticular wax (Table 7) and a juvenile staining pattern (data not shown) than gl15 plants. In contrast, in a Gl15/gl15 genotype, Tp2 extended the expression of these traits by as many as seven leaves. Traits that do not differ between gl15 and wild-type plants are not affected by gl15 in either a Tp1 or Tp2 background. In particular, tassel morphology, leaf number, the number of nodes with prop roots, and ear position are identical in Tp/+ and Tp/+; gl15-2/gl15 -2 (Table 7) and Tp/+; gl15-1 plants (data not shown). These results suggest that Gl15 acts downstream of the Tp loci and is required for the effect of the Tp loci on epidermal cell identity.

The duration of particular phases of shoot development has profound effects on the morphology of a plant, because the character of the structures produced during each phase is significantly different. In maize, several genes that regulate the juvenile vegetative phase of shoot development are defined by dominant, gain-of-function mutations (Teopod1, Teopod2, Teopod3, and Corngrass) that prolong the expression of a juvenile vegetative program and affect virtually every aspect of shoot morphology (Poethig, 1988a). Previous studies have suggested that these mutations define major regulatory genes that act more- or-less independently of the genes that regulate other aspects of shoot growth, such as the timing of reproductive development (Bassiri et al., 1992). However, an accurate understanding of the mechanism of vegetative phase change requires learning how the shoot develops in the absence of the various gene products that regulate its development.

Recessive mutations of Gl15 have a phenotype opposite that of the Tp mutations. Instead of prolonging the expression of juvenile traits, gl15 mutations truncate the expression of juvenile traits and accelerate the appearance of adult traits. Although this is the phenotype expected of a loss-of-function mutation in a Teopod-like gene, the results presented here suggest that Gl15 has a more limited function. The phase-specific traits examined in this study included epidermal traits (epicuticular wax, cell shape, cuticle thickness, histochemistry and the presence of differentiated cell types), a trait originating in internal tissue of the leaf blade (Rg1), stem traits (prop roots and tiller formation) and reproductive traits (time-toflowering, photoperiodic sensitivity, tassel morphology, and ear position). Of these, only epidermal traits are affected by gl15. Apparently, Gl15 is only required to promote juvenile vegetative development and suppress adult vegetative development in the epidermis. Of the phase-specific epidermal traits that we examined, only cuticle thickness is not regulated by Gl15. Cuticle thickness may be regulated by internal tissue layers, or at the level of the entire organ. Double mutant studies show that gl15 is epistatic to Tp1 and Tp2, but only for epidermal traits; gl15 has no effect on any other aspect of the Tp phenotype. These observations suggest that Gl15 acts downstream of the Tp genes, which act non-cell-autonomously (Dudley and Poethig, 1993; Poethig, 1988b), and functions specifically to regulate the expression of phase-specific epidermal traits. This interpretation is consistent with the observation that glossy15 acts cell-autonomously (Coe et al., 1988), as might be expected of a tissue-specific regulatory factor. The phenotype of gl15; Tp double mutants also demonstrates that an ectopic increase in Gl15 function is not responsible for all aspects of the mutant phenotype of Teopod plants, such as the transformation of ears and tassels to vegetative structures.

Because gl15 affects the timing of both juvenile and adult epidermal traits, regulation of these two phases must share a common mechanism. This observation has several possible interpretations. One possibility is that juvenile traits are regulated combinatorially by Gl15 and one or more adult regulatory genes. In this scenario, adult identity is the ground state of vegetative leaves, and juvenile identity results from modification of this ground state by Gl15. In the absence of Gl15, the identity of the epidermis is determined by adult phase genes. A second possibility is that juvenile and adult epidermal development are two mutually exclusive states and Gl15 is a switch gene that prevents the expression of adult genes and promotes the expression of juvenile genes. This hypothesis implies that epidermal cells can only exist in either of two states, juvenile or adult. A third possibility is that Gl15 regulates the expression of both the juvenile and the adult developmental programs, but the sensitivity of the adult program to changes in Gl15 activity is less than that of the juvenile program. In this scenario Gl15 could suppress adult development in the epidermis either directly, or indirectly via the activity of downstream genes.

The first model predicts that all leaves possess the ability to express adult traits (like epidermal hairs) but this expression is masked by juvenile regulatory genes. This prediction conflicts with the observation that in gl15-1 (W23) plants the absence of epicuticular wax on leaf 2 is not accompanied by the presence of epidermal hairs, implying that leaf 2 does not have the ability to express adult epidermal traits, even in the absence of Gl15 function. The second model, in which epidermal cells can only exist in either of two states, is inconsistent with the existence of intermediate cellular phenotypes. In the inbred A632, for example, leaf 5 is completely waxy and yet produces epidermal hairs (Table 1); examination of these leaves with a dissecting microscope confirms that these hairs arise from areas of the leaf that appear to be completely covered with epicuticular wax. Other phase-specific anatomical traits, such as cell shape and cuticle thickness are expressed at intermediate levels in single cells in transition leaves, and the degree to which the epidermis expresses one or another trait varies continuously in different transition leaves (Bongard-Pierce and Poethig unpublished results). The inbred W23 has a rather abrupt transition from juvenile to adult development, so this intermediacy is not evident in Fig. 2; in backgrounds in which this transition is more gradual these intermediate phenotypes are more obvious. Intermediate cells are also observed in Tp mutants, where the expression of juvenile traits is dramatically prolonged. These observations suggest that cells can express juvenile and adult genes simultaneously, and it is the relative level of juvenile and adult regulatory factors that specify the identity of the leaf. This is the situation predicted by the third model, which we favor.

The observation that loss of juvenile epidermal traits is not necessarily accompanied by the expression of adult ones in gl15 plants has at least two interpretations. One possibility is that the shoot does not become competent to express adult traits until later in development because activators of the adult phase (e.g. an ‘adult hormone’) are not expressed until that time. Another possibility is that the expression of juvenile epidermal traits is regulated by several distinct factors with different domains of expression along the length of the shoot. If this is the case, the expression of adult traits might require a decrease in the expression of both Gl15 and one or more of these other independent regulatory factors. The existence of another juvenile regulatory program is suggested by the fact that all three gl15 alleles that we have examined fail to affect the development of the first one or two leaves of the shoot. It is possible, of course, that this is simply because these mutations are hypomorphic and retain enough activity to specify juvenile development in leaves 1 and 2. However, if this is the null phenotype, the implication is that the juvenile identity of early leaves is not specified by Gl15. Instead, the identity of these leaves might be regulated by a separate early juvenile program, or Gl15 might be required to maintain rather than to initiate the juvenile phase of development.

To determine if gl15-1 and gl15-2 are hypomorphic or null alleles we compared the phenotype of plants homozygous for these mutations with hypoploid gl15 siblings generated by the non-disjunction of TB-9Lc. Hemizygous gl15 plants are expected to have the same phenotype as gl15/gl15 plants in the case of a null allele, and a more severe phenotype if the allele is hypomorphic. Unexpectedly, plants that were hemizygous for gl15-1 or gl15-2 had a less severe phenotype than their homozygous siblings. Because both alleles responded in the same way and were in different backgrounds, we believe that this result is an indirect consequence of hypoploidy for 9L rather than a reflection of the nature of these alleles. A similar phenomenon was observed in the case of ADH1, and was later shown to be due to the presence of negative regulators of ADH1 present on the same chromosome arm (Birchler, 1979, 1981). This result makes it impossible to conclude anything about the nature of gl15-1 and gl15-2, since we cannot ascribe the suppression of the gl15 phenotype solely to a change in the dose of this locus. Second site suppressors of the gl15-1 and gl15-2 phenotypes could act by increasing the expression of these mutations (if they are hypomorphic), up-regulating downstream genes regulated by Gl15, or prolonging the hypothetical early juvenile developmental program mentioned above.

Despite the inability to determine directly whether or not gl15-1 or gl15-2 represents the null phenotype of this locus we favor the hypothesis that Glossy15 is not required for juvenile development in the first one or two leaves. Instead, we believe that juvenile identity in these leaves is regulated through a separate pathway because none of the known mutations in Gl15 affect these leaves and because the first two leaves are morphologically and anatomically quite different from other juvenile leaves (Bongard-Pierce and Poethig, unpublished observations). Since these leaves are initiated during seed development, their identity might be regulated by factors expressed at that time; alternatively, their phenotype might be a default program expressed in the absence of juvenile and adult stimuli.

The traits that distinguish juvenile and adult phases of vegetative growth can be quite diverse and frequently include features that have no obvious morphological or functional relationship. Although the expression of these traits generally changes in a coordinated fashion, it has long been known that their expression can be dissociated by various experimental treatments (Hackett and Murray, 1993; Wallerstein and Hackett, 1989; Borchert, 1976). This observation suggests that the developmental phase of a shoot is probably regulated by a large number of discrete developmental programs whose expression is coordinated by one or a few major regulatory factors. The phenotype of gl15 mutations supports this hypothesis in that it defines a group of phase-specific epidermal traits that are regulated independently of other juvenile and adult traits. More importantly, Gl15 provides a link between factors that act globally to regulate phase change (e.g. the Teopod loci) and a single, relatively simple aspect of this phenomenon. A molecular analysis of the function of this gene and how its expression is regulated is likely to yield significant new insights into the mechanism of phase change.

We would like to thank Kathy Barton, Laura Conway, Abby Telfer, and Maja Bucan for helpful comments on this manuscript. We also would like to thank George F. Sprague and Julie Vogel for generously providing stocks of gl15 seeds. Deverie Bongard-Pierce provided the organizational structure that made this work possible. We are also grateful to Tracy Byford and the staff of the Penn greenhouses. This work was supported by NSF grants DCB-9012315 and DCB9205279; M. M. S. E was supported by an NIH training grant T32 HD07067.