Aleurone Cell Identity Is Suppressed following Connation in Maize Kernels

Abstract

Expression of the cytokinin-synthesizing isopentenyl transferase enzyme under the control of the Arabidopsis (Arabidopsis thaliana) SAG12 senescence-inducible promoter reverses the normal abortion of the lower floret from a maize (Zea mays) spikelet. Following pollination, the upper and lower floret pistils fuse, producing a connated kernel with two genetically distinct embryos and the endosperms fused along their abgerminal face. Therefore, ectopic synthesis of cytokinin was used to position two independent endosperms within a connated kernel to determine how the fused endosperm would affect the development of the two aleurone layers along the fusion plane. Examination of the connated kernel revealed that aleurone cells were present for only a short distance along the fusion plane whereas starchy endosperm cells were present along most of the remainder of the fusion plane, suggesting that aleurone development is suppressed when positioned between independent starchy endosperms. Sporadic aleurone cells along the fusion plane were observed and may have arisen from late or imperfect fusion of the endosperms of the connated kernel, supporting the observation that a peripheral position at the surface of the endosperm and not proximity to maternal tissues such as the testa and pericarp are important for aleurone development. Aleurone mosaicism was observed in the crown region of nonconnated SAG12-isopentenyl transferase kernels, suggesting that cytokinin can also affect aleurone development.

The cereal endosperm, which serves as the major source of storage reserves for the embryo, is composed of the starchy endosperm, basal endosperm transfer cells, and the aleurone, as well as starchy endosperm cell subtypes such as the subaleurone and embryo-surrounding region (Becraft, 2001; Olsen, 2001, 2004a). The aleurone layer, composed of cuboidal cells that are cytoplasmically dense, serves as the epidermal layer of the endosperm. Aleurone cells synthesize little starch but contain significant quantities of protein as inclusion bodies in small vacuoles as well as oil and phytin. Subaleurone cells are protein rich and contain fewer and smaller starch granules than do cells of the central starchy endosperm. The basal endosperm transfer cells adjacent to the pedicel facilitate nutrient transport from the maternal tissue to the developing endosperm, whereas the embryo-surrounding region lines the endosperm cavity where the embryo develops and may be involved in promoting embryo growth.

Development of the endosperm initiates following double fertilization, in which one pollen sperm nucleus fertilizes the egg cell to form the zygote and the second nucleus fuses with the two polar nuclei in the central cell of the megagametophyte to form the triploid endosperm (Becraft, 2001; Olsen, 2001, 2004a). Several rounds of mitosis without cell wall formation generate the syncitium or endosperm coenocyte that encircles a large central vacuole. The coenocyte undergoes cellularization through the formation of cell walls between the cytoplasmic domains of the nuclei as defined by a radial microtubular system. Cell wall formation initiates at the periphery of the endosperm and progresses centripetally to form alveoli, or tube-like structures that contain each nucleus of the coenocyte. Each nucleus then divides periclinally and a cell plate forms between daughter cells resulting in a completely cellularized layer of daughter nuclei at the periphery that serve as aleurone initials and an internal layer of daughter nuclei remaining in the alveoli. Subsequent rounds of periclinal division and similar cell wall formation results in the centripetal formation of radial files of endosperm cells that fill the central cavity, which is then followed by random cell division that results in a loss of the radial cell pattern. The first periclinal division of the initial peripheral cell layer of the coenocyte has been suggested to be a formative division that specifies daughter cell fate (Brown et al., 1996). In maize (Zea mays), the aleurone layer is just one cell layer thick and the aleurone initials divide mostly anticlinally. Periclinal division of aleurone cells results in the redifferentiation of the internal daughter cell into a starchy endosperm cell, indicating that aleurone identity is position dependent and flexible throughout endosperm development and that aleurone and starchy endosperm cells are clonally related (Becraft and Asuncion-Crabb, 2000). Recent studies have suggested that development of the aleurone does not require contact with maternal tissue but rather requires a peripheral position in the endosperm (Olsen, 2004b).

CRINKLY4 (CR4), a receptor-like kinase, is important for aleurone cell fate decision (Becraft et al., 1996). cr4 mutants exhibit aleurone mosaicism where, in the absence of aleurone cell development, the peripheral cells develop as starchy endosperm (Becraft et al., 1996). defective kernel 1 (dek1) mutants are devoid of aleurone, and like cr4 mutants, the peripheral cells adopt a starchy endosperm fate, indicating that this is the default cell type (Becraft and Asuncion-Crabb, 2000; Becraft, 2001). Single-celled revertant sectors in dek1 kernels or respecification of aleurone to starchy endosperm following loss of Dek1 function late in aleurone development demonstrates that Dek1, which encodes a membrane-localized, calpain-like protease, is necessary to maintain aleurone cell identity, which remains flexible until cell division has nearly completed (Becraft and Asuncion-Crabb, 2000; Becraft et al., 2002; Lid et al., 2002; Wang et al., 2003). Several other genes, including bareback*, naked*, collapsed2-o12, mosaic1, paleface*, and supernumerary aleurone layers 1, have been identified as important for aleurone development and may act at later stages than Cr4 or Dek1 (Becraft and Asuncion-Crabb, 2000; Shen et al., 2003). In other mutants such as disorganized aleurone layer 1 and 2, perturbations to the mitotic division plane of aleurone cells results in cells affected in size and shape and the presence of multiple layers of cells that maintain aleurone cell identity (Lid et al., 2004). Studies with the mutant globby1-1 that affects cell fate acquisition in the endosperm demonstrated that the acquisition of aleurone cell fate was dependent on factors from the endosperm itself (Costa et al., 2003).

During growth of the maize spikelet, the basic repeating unit of the maize inflorescence, development of an upper and lower floret initiates but the lower floret aborts, leaving only one pistillate floret per spikelet to complete its development (Cheng et al., 1982; Calderon-Urrea and Dellaporta, 1999). Expression from the cytokinin-synthesizing isopentenyl transferase (IPT) gene when under the control of the Arabidopsis (Arabidopsis thaliana) senescence-inducible promoter from the Cys protease gene, SAG12, inhibited abortion of the lower floret pistil, resulting in a spikelet with two florets, each containing a fertile pistil (Young et al., 2004). A single connated kernel, representing a two-seeded fruit that contained two genetically distinct embryos enclosed under a common pericarp, was generated from such spikelets. Fusion occurred along the abgerminal face, resulting in the juxtaposition and internalization of the two aleurone cell layers without perturbing their clonal relationship with the respective internal endosperm cells. In this report, we examined the consequence that connation has on cell identity along the fusion plane of the two-seeded fruit in maize. Although cells with aleurone characteristics were confined to the fusion junction with starchy endosperm present along most of the interior endosperm fusion plane, the presence of sporadic aleurone cells along the plane demonstrated that aleurone cell development did not require proximity to maternal tissues. Establishment of aleurone cell identity in nonconnated SAG12-IPT kernels was also partially disrupted resulting in aleurone mosaicism, suggesting that cytokinin can affect aleurone development. Additionally, the normal sequence of cell types from the periphery was also disrupted, resulting in the presence of cells typical of the central starchy endosperm proximal to the aleurone layer. The results illustrate that aleurone development is suppressed when positioned between starchy endosperms of independent kernels, supporting the notion that aleurone cell identity requires peripheral proximity at the surface of the endosperm.

RESULTS

The Presence of the Aleurone Layer Is Suppressed along the Fusion Plane of a Connated Endosperm

Inhibition of lower floret abortion in T₂ SAG12-IPT maize is partially penetrant (Young et al., 2004), resulting in connated kernels when the lower floret is rescued and a single, nonconnated kernel when the rescue of the lower floret fails. As a consequence of fusion along the abgerminal face, no testa/pericarp separated the endosperm contributed by each developing kernel but instead bounded the two-seeded fruit in a continuous layer (Fig. 1, B and E). The composition of connated kernels was similar to wild-type kernels (Fig. 1A) in that a high level of protein (indicated by blue staining) and starch (indicated by magenta staining) was observed in the embryos and fused endosperm, respectively (Fig. 1, A and B). In some cases, abortion of the kernel from the lower floret resulted in a rudimentary caryopsis that remained unfused with the kernel from the upper floret (Fig. 1C).

Figure 1.

Aleurone cell identity is lost when repositioned into the endosperm in connated SAG12-IPT kernels. Whole-kernel sections of mature control (A) and connated SAG12-IPT (B) kernels stained with TBO counterstained with PAS reagent to reveal protein and starch, respectively. Whole-kernel section of a mature SAG12-IPT kernel (stained with ABB/Sudan Black and counterstained with PAS reagent) in which lower kernel abortion occurred leaving an empty caryopsis. D, E, and G, Median sections showing the endosperm junction of connated SAG12-IPT kernels. F, The isolated aleurone cell along the endosperm fusion plane at the bottom of the section in D was stained with ABB to reveal protein content. H, A high magnification of the two aleurone layers along the endosperm fusion plane. I, Invagination of the aleurone layer into the starchy endosperm. Sections were stained with PAS reagent (D), or ABB counterstained with PAS reagent (E–I). al, Aleurone; j, junction between fused endosperms; p, pericarp; se, starchy endosperm.

The fusion plane between the two endosperms in connated SAG12-IPT kernels was indicated by the junction of the aleurone layers from the respective endosperms, resulting in a double aleurone layer (Fig. 1, D, E, and G) which persisted for only a short distance along the endosperm fusion plane before the development of one aleurone layer ceased. The size of the aleurone cells increased along the fusion plane as a function of their distance from the junction point (Fig. 1, D and E). The typical cuboidal shape was lost as the aleurone cells increased in size, with some adopting a more starchy endosperm-like shape (Fig. 1E). Development of the remaining aleurone layer ceased further into the endosperm and was replaced by cells that exhibited characteristics of the central starchy endosperm on either side of a fusion plane in which no maternal tissue such as the testa or pericarp was evident (Fig. 1, D, E, G, and H). Isolated putative aleurone cells were present along the endosperm fusion plane further into the interior of the connated endosperm (Fig. 1, D and F). Such cells retained aleurone cell characteristics in that they contained the high protein content and lack of starch granules characteristic of this cell type (Fig. 1F). Rare instances of limited invagination of the aleurone layer into the starchy endosperm were also observed (Fig. 1I). These observations indicate a loss of aleurone cell identity as the cells were displaced from the periphery of the connated kernel.

Aleurone cells are characterized by a high level of protein, oil, inorganic calcium, and phytic acid but no starch granules (Olsen, 2001). Starchy endosperm cells of the central endosperm are substantially larger than aleurone cells with thin cell walls. They contain numerous large starch granules but little protein, inorganic calcium, phytic acid, or oil. Separating these two cells types is the subaleurone region, which lies immediately beneath the aleurone layer and is typically several cell layers thick. Subaleurone cells differ from central starchy endosperm cells in that they contain fewer and smaller starch granules. They also contain a high level of protein but little oil, inorganic calcium, or phytic acid. Subaleurone cells can arise from periclinal division of differentiated aleurone cells in which one of the daughter cells is internally displaced and undergoes respecification to a subaleurone cell identity. To determine whether any of the aleurone or starchy endosperm cells bordering the fusion plane exhibited characteristics of other cell types, cells at the endosperm fusion plane were examined for characteristics associated with aleurone, subaleurone, or starchy endosperm cells.

To the extent that aleurone layers remained along the endosperm fusion plane, the cells in the layers were protein-rich as indicated by staining with Toluidine Blue O (TBO; Fig. 2, A, F, K, P, and U) and lacked starch granules as indicated by staining with Periodic acid-Schiff's (PAS) reagent (Fig. 2, D, I, N, S, and X). A high level of lipids (in the form of oil) characteristic of aleurone cells was observed in these aleurone cells as indicated by staining with Sudan Black (Fig. 2, D, I, N, S, and X). In addition, the presence of phytic acid and inorganic calcium was observed as indicated following staining with Acriflavine (ACF; Fig. 2, B, G, L, Q, and V) or Alizarin Red S (ARS), respectively (Fig. 2, C, H, M, R, and W). These characteristics were observed in isolated aleurone cells as in contiguous cells of the layer, indicating that these cells had maintained aleurone cell identity. Isolated aleurone cells were sometimes organized linearly along the fusion plane (Fig. 2, U–Y) whereas in rare instances, they lacked organization (Fig. 2, P–T). The proximal association of protein-rich subaleurone cells beneath the aleurone layers present along the endosperm fusion plane was not observed as indicated by the lack of staining by TBO (Fig. 2, A, F, and K) or Aniline Blue Black (ABB; Fig. 2, D, I, and N). Further along the fusion plane when both aleurone layers were absent, those cells bordering the fusion plane, i.e. the outermost cell layer of each endosperm, lacked the protein, oil, inorganic calcium, and phytic acid characteristic of aleurone cells. Instead, cells proximal to the aleurone layers along the fusion plane or that bordered the fusion plane in the absence of the aleurone layers had the low protein content and numerous large starch granules characteristic of central starchy endosperm cells (Fig. 1, D–F). Those cells surrounding isolated aleurone cells also lacked those characteristics associated with subaleurone cells but instead were indistinguishable from central starchy endosperm cells (Fig. 2, P–Y).

Figure 2.

Suppression of aleurone cell identity and disruption of the normal sequence of cell types along the endosperm fusion plane of connated SAG12-IPT kernels. A to E, Median sections of connated SAG12-IPT kernels at the junction of the two endosperms. F to O, The presence of adjacent aleurone layers along the endosperm fusion plane. K to O, Disorganization of the cells within the aleurone layers. P to T, An isolated and disorganized group of aleurone cells. U to Y, An isolated, linear array of aleurone cells. The sections were stained with TBO to reveal protein (A, F, K, P, and U), ACF visualized by fluorescence microscopy to reveal phytic acid (B, G, L, Q, and V), ARS counterstained with LG to reveal inorganic calcium (C, H, M, R, and W), ABB/Sudan Black and counterstained with PAS reagent to reveal protein, lipids (i.e. oil), and starch, respectively (D, I, N, S, and X), and ACF visualized by light microscopy to reveal aleurone cell walls (E, J, O, T, and Y).

The aleurone cell wall differs in composition from that of subaleurone and central starchy endosperm cells which is revealed as a bright yellow in light micrographs in ACF-HCl-stained sections (Fig. 2E). In addition to aleurone cell walls, those of the pericarp also stain a bright yellow. Only cell wall staining of aleurone cells present along the endosperm fusion plane was observed with no indication of pericarp tissue (Fig. 2, E, J, O, T, and Y). No staining was observed of the starchy endosperm cells that bordered the fusion plane (Fig. 2E), indicating that the cells in this region of the fusion plane did not exhibit aleurone cell characteristics. These results indicate that aleurone cell identity was maintained for a short distance from the junction of the connated endosperm but was suppressed further into the interior of the fused endosperm to be replaced by central starchy endosperm cells that did not retain any intracellular or extracellular characteristics of aleurone or subaleurone cells. The absence of subaleurone cells proximal to those aleurone cells that were present along the fusion plane suggests disruption of the normal sequence of cell types in the connated kernels.

Maintenance of Aleurone Cell Identity at the Endosperm Periphery Is Disrupted in Connated and Nonconnated SAG12-IPT Kernels

Suppression of aleurone cell identity along the abgerminal endosperm fusion plane in connated SAG12-IPT kernels may be a result of the loss of positional cues if a peripheral position is necessary to maintain aleurone cell identity. The suppression of aleurone development might also be a consequence of SAG12-IPT expression. If so, aleurone cell identity should be disrupted in the endosperm periphery to the same extent that it is along the abgerminal fusion plane. To examine this possibility, the development of the aleurone layer and subaleurone in the peripheral regions of the endosperm of connated SAG12-IPT kernels was examined.

Normal development of the aleurone layer and subaleurone was observed for most of the endosperm periphery outside the fused abgerminal face. A lack of aleurone cells at the endosperm periphery, however, was observed in the crown region of connated kernels (Fig. 3, A and C) not seen for control kernels (Fig. 3B). The pattern of loss was phenotypically similar to the aleurone mosaicism described for cr4 mutants (Becraft et al., 1996). In some instances in the absence of the aleurone layer, the outermost cell layer differentiated into subaleurone cells as evidenced by staining for protein (Fig. 3, E, J, and M) but a lack of aleurone cell walls and phytic acid (Fig. 3F), calcium (Fig. 3J), or lipids (Fig. 3E). In other instances, central starchy endosperm cells were present at the outermost cell layer as evidenced by the presence of numerous large starch granules and lack of protein (Fig. 3E). No subaleurone or starchy endosperm cells were present at the periphery of control kernels (Fig. 3, B, D, H, I, K, L, N, O, Q, and R). The observation that aleurone cells were virtually absent along the abgerminal endosperm fusion plane in connated kernels but were only sporadically absent in the endosperm periphery and then only in the crown region suggests that positional information is essential for aleurone cell identity and that SAG12-IPT expression can affect the maintenance of aleurone cell identity. It is not known if IPT is expressed in aleurone cells specifically or is expressed in other cells types, e.g. those of the nucellus, testa, or pericarp, therefore the cytokinin produced may be generated in aleurone cells that may affect its development directly or in neighboring tissues that may affect signaling.

Figure 3.

Suppression of aleurone cell identity and disruption of the normal sequence of cell types at the endosperm periphery of connated SAG12-IPT kernels. Median sections within the crown region of wild type (B, I, L, O, and R), vector-only control (D, H, K, N, and Q), and connated SAG12-IPT (A, C, E, F, G, J, M, N, and P) kernels stained with TBO (A–D), Sudan Black counterstained with PAS reagent (E), ACF visualized by light microscopy (F–I), ARS counterstained with LG (J–L), ABB counterstained with PAS reagent (M–O), and ABB/Sudan Black and counterstained with PAS reagent (P–R). sa, Subaleurone.

In addition to the observed aleurone mosaicism, disruption of the normal sequence of cell types in the periphery of connated SAG12-IPT kernels was observed. In regions where aleurone cell identity was suppressed and the peripheral layer had adopted a subaleurone cell identity, starchy endosperm cells indistinguishable from the central starchy endosperm were observed sporadically at the periphery as indicated by the presence of large starch granules and little protein (Fig. 3, E and J) not observed for the controls (Fig. 3, K, L, O, and R). In regions in which aleurone cells were present at the periphery, starchy endosperm cells indistinguishable from the central starchy endosperm were observed immediately beneath the aleurone cells, surrounded by subaleurone cells (Fig. 3, E, M, and P).

To determine whether the observed abnormalities in maintaining aleurone cell identity and starchy endosperm cell specification was specific to the fusion event that resulted in connated kernels, the aleurone layer and subaleurone region in the endosperm periphery of nonconnated SAG12-IPT kernels were examined. These kernels develop from the upper floret whereas kernel development from the lower floret aborts prior to pistil fusion leaving an empty caryopsis which remains unfused with the upper kernel (Fig. 1C). Consequently, the upper floret kernel contains a single embryo and a nonfused endosperm as in wild-type kernels. Aleurone mosaicism was observed in nonconnated SAG12-IPT kernels but was limited to the crown region (Fig. 4). No suppression of aleurone cell identity was observed along the abgerminal side of the kernel, which is the region involved in fusion in connated kernels, or in basal endosperm transfer cells present over the pedicel, or in the embryo-surrounding region (data not shown). The suppression of aleurone cell identity at the periphery sometimes involved a large number of cells (Fig. 4, A–C, E, F, and H) or a few (Fig. 4, D and G).

Figure 4.

Suppression of aleurone cell identity at the endosperm periphery of nonconnated SAG12-IPT kernels. Median sections within the crown region of nonconnated SAG12-IPT kernels stained with TBO (A and B), ABB and counterstained with PAS reagent (C and I), Safranin O counterstained with LG (D), Sudan Black counterstained with PAS reagent (E), ABB/Sudan Black and counterstained with PAS reagent (F and J–N), and ACF visualized by fluorescence microscopy (G, O, and P) or by light microscopy (H).

Disruption of the normal sequence of cell types was observed in nonconnated SAG12-IPT kernels as it was in connated kernels. Suppression of aleurone cell identity resulted in the presence of subaleurone cells (Fig. 4, C, F, and L) or starchy endosperm cells (Fig. 4, I and K) at the periphery. Starchy endosperm cells indistinguishable from the central starchy endosperm were also observed immediately beneath the aleurone layer, surrounded by subaleurone cells (Fig. 4, J, N, and O). Additional aleurone cells were observed occasionally beneath the aleurone layer in the region of the subaleurone (Fig. 4, M and P). These cells exhibited morphological and biochemical characteristics typical of aleurone cells including high protein and lipid content, aleurone cell walls, and lack of starch granules (Fig. 4, M and P). Whether these represent periclinal division of cells in the aleurone layer or differentiation of subaleurone cells into aleurone cells is unknown. These observations suggest a loss of control in the normal sequence of cell types in the outermost region of the endosperm, including the presence of aleurone and central starchy endosperm cells in the subaleurone region.

DISCUSSION

Alternative models for aleurone specification require signaling from maternal tissues such as the testa/pericarp or degenerating nucellus or from the zygote itself (Olsen et al., 1999; Becraft and Asuncion-Crabb, 2000; Becraft et al., 2001; Gifford et al., 2003). In the first model, aleurone cell identity is established by virtue of its position as the cell layer closest to the testa/pericarp where the concentration of the morphogen would be highest. In the second model, aleurone cell identity is established by virtue of its position as the most peripheral cell layer of the endosperm without requiring proximity to maternal tissues, described recently as the surface rule (Olsen, 2004b). The fusion plane of a SAG12-IPT connated kernel in which aleurone layers were positioned between two independent endosperms represents the repositioning of the abgerminal aleurone cell layer from its normal pericarp-proximal position to a position next to the adjoining aleurone cell layer in the connated endosperm. Thus connation results in the internalization of the abgerminal aleurone layer into the center of a single, connated endosperm while maintaining its clonal relationship with the endosperm from which it developed. Aleurone specification along the fusion plane of a connated kernel was suppressed in steps in which two aleurone layers were observed at the endosperm juncture and for 10 to 20 cells into the connated endosperm. This maintenance of aleurone cell identity is in contrast to the respecification of aleurone daughter cells into starchy endosperm cells when displaced internally by just one cell layer following periclinal division (Becraft and Asuncion-Crabb, 2000) but is similar to the maintenance of aleurone cell layers in regions of endosperm invagination of mutants affected in endosperm development (Olsen, 2004b). Following suppression of one aleurone layer within 10 to 20 cells into the connated endosperm, suppression of the remaining aleurone layer occurred within an additional five to 10 cells along the fusion plane, at which point the presence of aleurone cells became sporadic resulting in a few isolated aleurone cells before becoming largely absent. Instances of isolated aleurone cells were observed further into the interior of the connated endosperm but always at the fusion plane, which may represent regions of late or incomplete fusion of the connated endosperm. These observations support the conclusion that peripheral position and not proximity to maternal tissues such as the testa and pericarp are important for aleurone development.

Disruption of the normal sequence of endosperm cell types was also observed. Subaleurone cells were present beneath the aleurone layer of connated kernels up to the point of connation. However, they were absent from their position beneath each aleurone layer along the fusion plane for as far as the aleurone layers persisted, and, instead, starchy endosperm cells were present. Following suppression of both aleurone cell layers along the fusion plane further into the interior of the connated endosperm, only central starchy endosperm cells were present at the fusion plane. These observations support the notion that positional cues are essential for aleurone development. The observation that aleurone cells are largely absent within 20 to 30 cells of separation from the surface of the connated endosperm is consistent with a signal originating at or beyond the endosperm periphery, e.g. from maternal tissues. The development of aleurone cells within 20 to 30 cells of separation from the testa/pericarp suggests that direct contact with maternal tissue is not required, as observed with mutants affected in endosperm development (Olsen, 2004b), but does not rule out a maternally derived signal that may diffuse for a distance into the junction of the connated endosperms. Receipt of signals of maternal origin important for the development of zygotic tissues has been suggested in petunia (Petunia hybrida), where suppression of MADS box genes normally expressed specifically in the seed coat resulted in aberrant development of the seed coat and degeneration of the endosperm (Colombo et al., 1997).

The presence of isolated aleurone cells within the interior of the connated endosperm but at the outermost layer of cells of the endosperm from which it developed suggests that even in the absence of any maternally derived signal, position-derived information with respect to the starchy endosperm may be sufficient as aleurone cells that are completely surrounded by starchy endosperm have not been observed in wild-type endosperm. Interestingly, the mutant globby1-1 permits the development of aleurone cells within the starchy endosperm (Costa et al., 2003). This observation supports the importance of position as the outermost layer of endosperm from which it developed and may involve a signal originating from the starchy endosperm.

The most notable observation resulting from connation was the suppression of aleurone cells along most of the fusion plane, suggesting that the presence of starchy endosperm on either side of an aleurone cell serves to suppress its development either by inhibiting aleurone cell differentiation or by causing aleurone cells to undergo respecification into starchy endosperm cells. Thus, signals originating from the starchy endosperm are likely to be involved in determining the development of aleurone cells at the periphery. It should be noted that such a starchy endosperm-derived signal is not incompatible with the involvement of a maternally derived signal (e.g. produced by the testa/pericarp) that may also contribute to aleurone development.

Cytokinin as a possible candidate for a zygotically derived signal is supported by several observations. Expression of CycD3, which promotes progression through the G1-S transition of the cell cycle, is induced by cytokinin, and overexpression of CycD3 in Arabidopsis eliminated the requirement for exogenous cytokinin during callus growth (Riou-Khamlichi et al., 1999). In many species, including cereals, an increase in the level of cytokinin in the endosperm coincides with the rapid cell division that occurs following fertilization and declines again following the period of most rapid cell division (Morris, 1997), supporting the notion that cytokinin may promote endosperm cell division by increasing CycD3 expression (Olsen, 2001). Thus, endosperm expansion coupled with a decline in the level of cytokinin may result in a gradient of the hormone that is high in the central endosperm where cell division continues for longer but decreases toward the periphery.

The observed defects in aleurone development of SAG12-IPT kernels also support a regulatory role for cytokinin during endosperm development. Sporadic suppression of aleurone cell development in the crown region of SAG12-IPT connated and nonconnated kernels resulted in aleurone mosaicism, consistent with other mutants affecting aleurone development (e.g. bareback*, naked*, collapsed2-o12, mosaic1, and paleface*). Whether any of these genes are regulated by cytokinin will be of interest to determine. Cytokinin may have been generated in SAG12-IPT kernels in degenerating maternal tissues or from the endosperm itself as it undergoes cell death (Young and Gallie, 2000), although the autoregulated nature of SAG12-IPT expression precluded measurement of IPT transcripts or cytokinin. That aleurone cells were present along the abgerminal side of nonconnated SAG12-IPT kernels supports the conclusion that the observed defects in aleurone development along the endosperm fusion plane of connated SAG12-IPT kernels was a result of the fusion event and not from cytokinin-mediated changes in aleurone development.

In conclusion, one consequence of the formation of a two-seeded fruit resulting from the connation of adjacent kernels is the suppression of aleurone cell identity along the fusion plane, despite the fact that the relationship between the outermost layer of cells of each endosperm along the fusion plane is maintained with respect to the endosperm beneath it. Thus, the observations made with connated kernels supports the surface rule that aleurone cell identity is established when the most peripheral cell layer is bounded on just one side by endosperm.

MATERIALS AND METHODS

Generation of Transgenic Lines

Embryogenic callus from HiII (derived from A188 x B73) was used for transformation by particle bombardment (Gordon-Kamm et al., 1990). The ubiquitin promoter-bar herbicide resistance plasmid, pAHC20 (Christensen and Quail, 1996), was codelivered with the SAG12-IPT construct (Gan and Amasino, 1995) in order to select for transformants on bialaphos (De Block et al., 1987). Plants were regenerated, transgene-containing plants crossed to B73, and the progeny selfed as described previously (Young et al., 2004). T₃ kernels were used for the analysis.

Microscopy

Tissue was fixed in formaldehyde-acetic acid (50% ethanol, 5% acetic acid, 3.7% formaldehyde) at 4°C, dehydrated through a graded ethanol series to 100%, and embedded in resin. Plastic kernel sections (3 μm) were stained with TBO or ABB for protein (Lillie, 1977), Sudan Black B (SBB) for lipid (Bayliss and Adams, 1972), ACF for phytic acid (Fulcher, 1982), ARS for inorganic calcium (McGee-Russell, 1958), Light Green (LG) for general staining, or PAS reagent for starch (Hotchkiss, 1948).

For SBB staining, section-containing slides were first soaked in freshly made 0.1% bromine in water (v/v) for overnight. The slides were rinsed in water the following day, stained with 3 mg/mL SBB in 70% ethanol for 30 min, and washed with 70% ethanol. Excess ethanol was removed and the slides air-dried overnight. For ABB, sections were stained with 1% ABB, pH 7.0, in phosphate-buffered saline for 30 min and rinsed well in water. Excess water was removed and the slides air-dried overnight. For PAS staining, sections were oxidized in 1% periodic acid for 5 min and continuously washed in water for 3 min. The slides were then placed in a Coplin jar with Schiff's reagent (Sigma) for 10 to 40 s until the appearance of magenta color and were then continuously washed in water for 10 min. Excess water was removed and the slides air-dried overnight. For staining with ACF, sections were stained with 0.5% ACF in water for 1 h, washed in water, blotted, and air-dried overnight. For staining with ARS, sections were stained with 2% ARS in water for 1 h, washed in water, blotted, and air-dried overnight. For staining with LG, sections were stained with 1% LG in water for 10 s, washed in water, blotted, and air-dried overnight. For staining with TBO, sections were stained with 1% TBO in phosphate-buffered saline, pH 7.0, for 10 s, washed in water, blotted, and air-dried overnight.

For SBB/ABB/PAS triple staining, sections were stained first with SBB as described above and following washing with 70% ethanol, the sections were rehydrated in water and stained in 1% ABB as described above and finally stained with PAS as described above. For ABB/PAS double staining, sections were stained first with ABB and counterstained with PAS. For ARS/LG double staining, sections were stained first with ARS and counterstained with LG. Images of the stained kernel sections were collected using a light microscope.