Expression patterns of genes encoding HD-ZipIV homeo domain proteins define specific domains in maize embryos and meristems


*For correspondence (fax +33 4 72 72 86 00; e-mail


A family of homeo box genes with cell layer-specific expression patterns defining subdomains of the embryo and certain meristems has been isolated from maize. These genes encode proteins from the class of plant specific homeo domain-leucine zipper (HD-Zip) transcription factors containing the previously described ZmOCL1 protein, and have been designated ZmOCL2, ZmOCL3, ZmOCL4 and ZmOCL5. ZmOCL3, ZmOCL4 and ZmOCL5, like ZmOCL1, showed essentially L1 or epidermis-specific expression. However, each gene was expressed in a distinct region of the embryonic protoderm during early development, with ZmOCL3 showing suspensor-specific expression, ZmOCL4 transcripts being localized to the adaxial face of the embryo proper and ZmOCL5 showing a more abaxial expression pattern. All three genes were also expressed in vegetative, inflorescence and floral apices, although ZmOCL3 transcripts were excluded from meristems and very young organ primordia. In contrast, ZmOCL2 expression was entirely meristem-specific and was excluded from the L1 layer, appearing instead to be largely restricted to a cell layer directly beneath the L1, especially in floral meristems. This expression pattern is unprecedented and may indicate that cell-layer organization in maize meristems is more complex than that suggested by the classical L1/L2 (outer cell layer/inner cell mass) model. These differing expression patterns indicate that the members of the HD-ZipIV family of maize may not only play roles in defining different regions of the epidermis during embryonic development, but could also be responsible for maintaining cell-layer identity in meristematic regions.


Compartmentation into periclinal cell layers is a universal feature of angiosperm meristems. Classically two main compartments have been distinguished: the outer layer or tunica, which contains one or more distinct cell layers proliferating by anticlinal cell division; and the inner region or corpus, which is composed of cells proliferating by both anticlinal and periclinal divisions (Van der Schoot & Rinne 1999). Although the functional importance of the layered structure of meristems has never been directly demonstrated, numerous observations demonstrating clonal and symplastic separation of cell layers and cell layer-specific gene expression of developmentally important genes indicate that the layered organization of meristems is probably integral to their normal function (Clark et al. 1997; Jackson et al. 1994; Rinne & van der Schoot 1998; Szymkowiak & Sussex 1996). In many commonly studied dicotyledonous plants, such as Arabidopsis and Antirrhinum, the tunica is divided into two cell layers designated L1 and L2, and the corpus is commonly designated L3 (Evans & Barton 1997). In monocotyledonous plants such as maize the situation is less clear, with only two cell layers being distinguished, the L1 or outer cell layer (tunica) and the L2 or inner cell mass (corpus) (Abbe et al. 1951; Steffensen 1968).

The L1 cell layer of plant meristems is derived from the protoderm (outer cell layer) of the developing embryo. The specification of ‘protoderm' identity, and the effective clonal isolation of the protodermal cell layer, are thought to be amongst the first cell fate specification events during the early embryogenesis of plants (Jürgens et al. 1991; Meinke 1991; Randolph 1936; Vroemen et al. 1996). The subsequent role of the protoderm in embryogenesis is unclear, but it is likely to play an important role in signalling to developing internal tissues, notably the subepidermal cell layer. Clonal analysis in dicotyledonous species has shown that many of the internal tissues of vegetative organs, and the germ line, arise from the meristematic L2 layer, whilst the meristematic L3 layer gives rise mainly to the vascular system (reviewed by Szymkowiak & Sussex 1996). During the embryogenesis of model dicotyledonous species, the separation of the protoderm, ground (or subepidermal) and provascular (or inner) cell layers occurs early and is clearly visible at a cytological level, although the exact origin of the meristematic L2 and L3 layers with respect to the embryonic ground and provascular layers is not clear. In comparison, monocotyledonous species such as maize have a relatively disorganized pattern of embryonic cell division. However, clonal analysis and cytological observations have shown that there is effective clonal isolation of the protodermal cell layer from subepidermal cells in maize at the early transition stage (6–7 days after pollination) (Poethig et al. 1986; Randolph 1936).

The molecules responsible for specifying the identity of individual regions or compartments in developing embryos were first isolated in Drosophila during the analysis of homeotic mutants in which specific regions of the fly undergo changes in identity. This led to the discovery that a large number of closely related transcription factors sharing a common DNA binding motif (the homeo domain) act as master proteins regulating regional specification events during fly development (Akam 1985; Lawrence & Morata 1994). The genes involved were found to be expressed in patterns correlating closely with their mutant phenotypes, and were also shown to act in a combinatorial fashion to specify cell identity. The same class of transcription factors (homeo domain proteins) implicated in cell fate specification in animals have also been shown to play roles in cell fate specification events in plants. For example the ‘meristematic' identity of cells in the apical meristem of Arabidopsis has been shown to depend on the activity of the SHOOT MERISTEMLESS (STM) homeo domain protein (Long et al. 1996). The KNOTTED-1 (KN-1) protein is thought to play a similar role in maize (Sinha et al. 1993; Vollbrecht et al. 1991). Another example is the GLABRA2 (GL2) homeo domain protein of Arabidopsis, which is involved in the specification of the outer cell layer in both the shoot (Rerie et al. 1994) and the root (Di Christina et al. 1996; Masucci et al. 1996). Together with TRANSPARENT TESTA GLABRA (TTG), GL2 is thought to promote trichome differentiation and to inhibit root hair formation (Hung et al. 1998).

The mechanisms behind specification and maintenance of the identity of different regions of the plant embryo and meristem remain largely unresolved. Although several mutants have been described showing defects in specific regions of the embryo axis of Arabidopsis and other model species, it is unclear what types of genes are implicated in crucial patterning events such as specification of the apical (embryo proper) and basal (suspensor) regions, or specification of cell-layer identity. Several genes have been described with expression patterns which could be considered to implicate them in such processes. The ATML1 gene of Arabidopsis belongs, like GL2, to the plant-specific class of HD-Zip homeo box genes, but is expressed after the first cell division of the developing zygote, specifically in the apical cell. ATML1 expression then becomes restricted to the protoderm cell layer at the dermatogen stage making this gene a candidate for specification both of ‘apical' and ‘protoderm' identity (Lu et al. 1996). We have previously described an ATML1-related gene in maize (ZmOCL1) which shows a similar, but not identical, expression pattern (Ingram et al. 1999). More recently a SHAGGY-related protein kinase (ASKη), also from Arabidopsis, has been shown to be expressed specifically in suspensor cells, indicating a role in specification of ‘basal', or at least ‘suspensor' cell fate (Dornelas et al. 1999).

Here we describe four new members of the HD-Zip class of homeo box genes to which GL2, ATML1 and ZmOCL1 belong. We have isolated cDNA clones for these genes (designated ZmOCL2, ZmOCL3, ZmOCL4 and ZmOCL5), and have analysed their expression patterns by in situ hybridization. This analysis has shown that three of the new genes, like GL2, ATML1 and ZmOCL1, show essentially L1/protoderm/epidermis-specific expression patterns, but that the expression pattern of each gene differs spatially, especially during early embryogenesis and endosperm development. In contrast, one of the four genes, ZmOCL2, shows an entirely novel meristem-specific expression pattern which, contrary to that of the other described members of this gene family, is restricted to subepidermal cell layers.


Isolation of ZmOCL2, ZmOCL3, ZmOCL4 and ZmOCL5

In order to determine whether relatives of ZmOCL1 existed in the maize genome, a genomic library was screened at medium stringency with a complete ZmOCL1 cDNA. Two clones showing medium-strength hybridization were partially sequenced, and revealed the presence of a gene strongly related to ZmOCL1 which was designated ZmOCL2. Since attempts to isolate a ZmOCL2 cDNA from a stage 1 embryo cDNA library were not successful, 5′ and 3′ RACE reactions were carried out in a variety of tissues. Young female inflorescences yielded the complete cDNA sequence downstream of the leucine zipper coding region and most of the cDNA upstream of this point, including the entire homeo domain-encoding region (in total 2660 bp of sequence containing 2179 bp of ORF). However, a 5′ region predicted to contain a short region of coding sequence, the ATG and the 5′ non-translated region was missing.

Although no ZmOCL2 clones were found in the cDNA library screen described above, a total of 25 positive clones had been isolated. Six clones were found to correspond to ZmOCL1. Eleven clones corresponding to a new gene designated ZmOCL3 were identified, the longest of which was 3580 bp in length and contained an ORF of 2589 bp encoding a predicted 863 amino acid protein of 91.5 kDa. Five clones corresponded to a gene subsequently designated ZmOCL4, the longest of which was 4001 bp in length and contained an ORF of 2526 bp encoding a predicted 842 amino acid protein of 90.7 kDa. Three clones encoded almost identical proteins (designated ZmOCL5α and ZmOCL5β), but showed significant divergence in DNA sequences, especially within the 5′ and 3′ untranslated regions, leading us to conclude that the corresponding genes probably arose from a recent gene duplication. As these two genes could not readily be distinguished by hybridization, they were treated as one gene (ZmOCL5) for the purpose of expression analysis. The sequence data provided for ZmOCL5 is that from ZmOCL5α. for which the longest cDNA was 3010 bp in length and contained a 2385 bp ORF encoding a predicted 796 amino acid protein of 85.4 kDa. The sequences of the putative gene products of these genes are shown in Fig. 1.

Figure 1.

Amino acid sequences deduced from ZmOCL cDNAs compared with other HDZipIV-class proteins.

Amino acids identical to the consensus are shaded in black. The homeo domain, the putative leucine zipper domain and the START domain are underlined with black, hatched and shaded bars, respectively.

Mapping of ZmOCL2, ZmOCL3, ZmOCL4 and ZmOCL5

The positions of ZmOCL2, ZmOCL3, ZmOCL4 and ZmOCL5 were mapped using RFLPs in several different populations of plants (Burr & Burr 1991; Maugenest et al. 1997). The results obtained are shown in Table 1. Only ZmOCL5 showed a profile in DNA gel blots suggestive of the presence of two genes within the genome, and the presence of two ZmOCL5 genes was confirmed by our mapping data. It is probable that ZmOCL2, ZmOCL3 and ZmOCL4 represent single-copy genes.

Table 1. . Mapping of ZmOCL genes
GeneChromosomeNearest markers
ZmOCL2Chr 10LiM350 and UMC44A
ZmOCL3Chr 7LiM209B and UMC110
ZmOCL4Chr 1npi453 and npi429
ZmOCL5Chr 4npi386 and npi95A
Chr 10UMC64 and UMC44A

ZmOCL2, ZmOCL3, ZmOCL4 and ZmOCL5 belong to three distinct subfamilies of HD-ZipIV proteins

Amino acid sequence comparisons were carried out between ZmOCL2, ZmOCL3, ZmOCL4 and ZmOCL5, and previously characterized HD-ZipIV proteins (GL2, ATML1 and Anthocyaninless2 (Anl2) from Arabidopsis, ZmOCL1 from maize and O39 from Phalaenopsis, Ingram et al. 1999; Kubo et al. 1999; Lu et al. 1996; Nadeau et al. 1996; Rerie et al. 1994). This analysis showed that these proteins fell into four main groups: ZmOCL2 and ZmOCL3 fell into the same group as ZmOCL1 and ANL2; ZmOCL5 was found to show most similarity to the AtML1 and O39 proteins; and both ZmOCL4 and GL2 were relatively divergent and did not fit into any group (Fig. 2).

Figure 2.

Phylogenetic tree of amino acid sequences deduced from ZmOCL genes and related HDZipIV-class proteins.

ZmOCL2 expression is meristem-specific and excluded from the L1 layer

Expression of ZmOCL2 in embryos was not strong enough to be detected by in situ hybridization until 12–15 DAP. Expression was never detected in the endosperm and, in embryos, was restricted to the apical meristem (Fig. 3a,b). Transcripts were not observed within the L1 cell layer of the embryonic apical meristem, and appeared strongest within the cells immediately below the L1 layer, although weaker expression was also detected in inner cell layers. Expression appeared to be downregulated in the region of the incipient leaf primordium (P0). The detected signal was weak and was most easily seen in transverse sections of the apical meristem. The pattern of ZmOCL2 transcripts in the apical meristem after germination was identical to that in embryonic meristems; no signal was detected in root tips (data not shown).

Figure 3.

Expression of ZmOCL2.

Transverse sections of embryonic apical meristems at 15 DAP (a, b), longitudinal sections of young female flowers before initiation of stamen primordia (c, d) and longitudinal sections of young male floral meristems (e, f) were probed with ZmOCL2 and photographed in bright field (a, c, e) or dark field (b, d, f). c, coleoptile; fm, floral meristems; gr, gynoecial ridge; P0, incipient leaf primordium; P2, second visible leaf primordium; P3, third visible leaf primordium; sc, scutellum; sf, secondary flower. The first leaf primordium P1 was not visible in the plane shown (a, b). Scale bar, 0.1 mm.

RACE experiments showed that ZmOCL2 was expressed in young female inflorescences (data not shown). This was confirmed by in situ analysis in young female and male inflorescences. Expression in female inflorescence meristems and young female floral primordia was largely, if not exclusively, restricted to a single cell layer directly below the L1 layer (Fig. 3c,d). Expression was observed at the youngest stages of floral development, before the start of floral organ initiation, and again later in floral development, just before initiation of the gynoecial ridge (precursor of the carpel wall primordium) in the gynoecial meristem. At this later stage, clear subepidermal expression (again restricted to only one cell layer) was detectable within restricted regions of the gynoecium. Expression in the youngest stages of male flower development was again very clearly restricted to one, immediately subepidermal, cell layer (Fig. 3f,g) but was not detected at later stages.

ZmOCL3 shows a ‘basal' expression pattern in young embryos and endosperm

Strong expression of ZmOCL3 was detected in the basal region of the embryo (suspensor region) at 5 DAP (Fig. 4a). In addition, expression was detected in the basal region of the endosperm or embryo surrounding region (ESR) which envelops the developing embryo at this stage. However, at 7 DAP, although ZmOCL3 expression remained strong in the suspensor, the transcript detected in the ESR was greatly diminished (Fig. 4d). No transcript was detected in the apical part of the embryo until 9 DAP when a small region of expression was observed at the tip of the scutellum (Fig. 4g). The expression of ZmOCL3 in the suspensor had ceased to be uniform at this stage and was restricted to the outer cell layers of the suspensor (Fig. 4j). Expression of ZmOCL3 at 12 and 15 DAP remained strong in the suspensor, with weaker expression observed in the tip and abaxial face of the scutellum.

Figure 4.

Expression of ZmOCL3, ZmOCL4 and ZmOCL5 during embryogenesis.

Embryos (longitudinal section within caryopsis) at 5 (a, b, c), 7 (d, e, f) and 9 DAP (g, h, i) probed with ZmOCL3 (a, d, g), ZmOCL4 (b, e, h) and ZmOCL5 (c, f, i). At 5 and 7 DAP the embryo proper and suspensor are indicated with an arrow and an arrowhead, respectively. The embryos are situated within the endosperm (end). Details of a transverse section of a suspensor (12 DAP) probed with ZmOCL3 (j) and endosperm (7 DAP) probed with ZmOCL4 (k) or ZmOCL5 (l) are also shown. Scale bar 0.1 mm.

ZmOCL3 expression is excluded from meristematic regions in developing shoots, roots and flowers

During vegetative development, weak expression of ZmOCL3 was detected in the epidermal cell layer of young leaf primordia (Fig. 5a). However, expression was excluded from meristematic regions and the youngest leaf primordia. Similarly, no expression was detected in root meristems, while very weak expression was occasionally seen in some parts of the root epidermis (Fig. 5d). During flower development ZmOCL3 transcripts were absent from inflorescence meristems and floral meristematic regions (Fig. 5g). Expression in young floral organs was relatively weak (Fig. 5j), except in developing stamens of both male and female flowers, which showed a strong L1-specific expression of ZmOCL3 until relatively late in development (Fig. 5m).

Figure 5.

Expression of ZmOCL3, ZmOCL4 and ZmOCL5 in vegetative and floral meristems.

Vegetative apical meristems (a, b, c), root tips (d, e, f), female inflorescence meristems (g, h, i), young female flowers (j, k, l) and older female flowers (m, n, o) were probed with ZmOCL3 (a, d, g, j, m), ZmOCL4 (b, e, h, k, n) or ZmOCL5 (c, f, i, l, o). a, anther (arrested); fm, floral meristem; g, gynoecium primordium; im, inflorescence meristem; ip, integument primordia; l, leaf primordia; m, shoot apical meristem; o, ovule; rc, root cap; rm, root meristem; s, silk; sf, secondary flower. Scale bar, 0.1 mm.

ZmOCL4 and ZmOCL5 are expressed in the embryo proper during embryogenesis

ZmOCL4 and ZmOCL5 both exhibited protoderm-specific expression within the apical part of the embryo during embryogenesis. At 5 DAP, expression of ZmOCL4 formed a ‘cap' covering the ‘top' of the embryo and descending into the adaxial face (Fig. 4b). ZmOCL5 transcripts also formed a cap covering the top of the embryo, but tended to spread into the abaxial region (Fig. 4c). ZmOCL5 expression in the endosperm was very weak or absent, whereas ZmOCL4 transcript was abundant in the developing endosperm, especially in outer cell layers.

ZmOCL4 expression at 7 DAP was similar to that at 5 DAP, with strong expression still detectable in the outer cell layers of the adaxial face of the embryo (Fig. 4e). Expression of ZmOCL5 was detectable in both the abaxial and adaxial regions of the embryo at 7 DAP, but adaxial expression was still weaker than that in abaxial regions (Fig. 4f). While ZmOCL4 remained strongly expressed in the outer cell layer of the endosperm (Fig. 4k), no ZmOCL5 expression was detected in the endosperm at this stage (Fig. 4l) or thereafter.

By 9 DAP the expression patterns of ZmOCL4 and ZmOCL5 were very similar, with expression throughout the whole of the outer cell layer of the embryo proper and no expression within the suspensor. However, ZmOCL4 expression in the abaxial face of the embryo remained weaker (Fig. 4h) than that of ZmOCL5 (Fig. 4i).

By 12–15 DAP, ZmOCL4 and ZmOCL5 transcripts were detected in the outer cell layers of all structures apart from the suspensor and the root apical meristem (data not shown), while ZmOCL4 expression in the endosperm started to decline after 15 DAP.

ZmOCL4 and ZmOCL5 show strong expression in vegetative, inflorescence and floral meristems

The expression patterns of ZmOCL4 and ZmOCL5 in vegetative, inflorescence and floral meristems were virtually indistinguishable, and in addition were very similar to that previously described for ZmOCL1. Expression was detected within the L1 layer of vegetative inflorescence meristems and in the epidermis of young leaf primordia and leaves (Fig. 5b,c) but was absent from root tips (Fig. 5e,f). Later, expression was detected throughout the L1 layer of the inflorescence apex, young floral meristems (Fig. 5h,i) and in developing floral organs (Fig. 5k,l) until relatively advanced stages (Fig. 5n,o). As for ZmOCL1, ZmOCL4 and ZmOCL5 did not appear to be expressed in mature ovules. Two slight differences were observed in comparison to ZmOCL1 expression. First, in the female flower ZmOCL4 and ZmOCL5 were both expressed in the epidermal layer of the developing stamens (which are programmed to abort), whereas ZmOCL1 expression appeared to be specifically excluded from these structures. Secondly, ZmOCL4 and ZmOCL5 both gave a signal which was tightly restricted to the extreme outer cell layer, whereas the signal detected for ZmOCL1 appeared less restricted, with transcripts often detectable in the cell layer underlying the epidermis or L1 layer. The expression patterns of all five ZmOCL genes are summarized schematically in Fig. 6.

Figure 6.

Summary of ZmOCL expression in maize.

Expression of ZmOCL1–ZmOCL5 in caryopses at 7 DAP, embryos at 9 DAP, vegetative meristems and young female flowers. Strong expression is shown in red, weak expression in yellow.

RT–PCR analysis of gene expression

To compare the patterns of expression of ZmOCL2, ZmOCL3, ZmOCL4 and ZmOCL5 with that of ZmOCL1 in tissues other than those analysed by in situ hybridization, RT–PCR analysis was carried out for each gene using intron-spanning primer pairs. The results of this analysis are shown in Fig. 7. Samples analysed were from mature leaves, mature stems, mature root mass, the meristematic region of germinating seedlings (including young leaf bases), root tips of germinating seedlings (lateral and primary roots), and endosperms dissected at 9 DAP. Samples were standardized using the constitutively expressed control gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Richert et al. 1996). This analysis largely confirmed the results obtained by in situ hybridization. For example, ZmOCL4 showed stronger expression in endosperm than the other genes, and all genes were, as expected, found to be expressed within ‘meristematic region' samples. Interestingly, all genes except ZmOCL5 were found also to show expression within root tips, although it was not strong enough to be detected by in situ hybridization (Fig. 5). In mature tissues no expression was observed for ZmOCL2 or ZmOCL3. ZmOCL1, as previously observed, was expressed in mature leaves, and weakly in the mature root mass. ZmOCL5 was strongly expressed in mature leaves and weakly in stems, whilst ZmOCL4 showed expression only in stems.

Figure 7.

RT–PCR analysis of ZmOCL expression in maize.

RT–PCR reactions were carried out on reverse transcribed mRNA samples from mature leaves (L), mature stems (S), mature root mass (R), the apical meristem region of seedlings (M), root tips (Rt), endosperm at 9 DAP (E9), and a control sample of non-reverse transcribed meristem mRNA (C).


At least six genes encoding HD-ZipIV-class homeo domain proteins are expressed during maize embryogenesis

We have characterized four new members of the HD-ZipIV family of plant specific homeo box genes from maize, at least one of which is duplicated, bringing the total number of such genes in maize to at least six. We have named this family of genes from maize OCL (Outer Cell Layer) as they present diverse expression patterns which are, nevertheless, all largely restricted to the outer cell layer(s) of embryos and meristems (Fig. 6). Two closely related genes from Arabidopsis, GL2 and the more recently characterized ANL2, present mutant phenotypes indicating that they may play a role in cell fate specification during plant development (Kubo et al. 1999; Masucci et al. 1996; Rerie et al. 1994). It is thus possible that the OCL genes from maize also play roles in specifying cell fate during plant development.

The presence of a putative leucine zipper domain immediately downstream of the homeo domain in OCL proteins, in combination with their overlapping RNA expression patterns, raises the possibility that they could interact in combination as homo- or hetero-dimers (Di Christina et al. 1996). Unfortunately, strong transcriptional activation activity shown by four of the six genes characterized made simple verification of this possibility using the yeast two-hybrid system unfeasible (unpublished results), and a more complex approach will be required. A recent study has shown that HD-ZipIV proteins also contain a ‘start' domain downstream of the putative leucine zipper domain, leading to speculation that they could regulate transcription in a steroid-dependent manner (Ponting & Aravind 1999). Many recent studies have shown that the plant-specific brassinolides/brassinosteroid class of molecules may act as endogenous plant hormones regulating many aspects of development (Creelman & Mullet 1997).

Gene expression in the ‘basal' region of the embryo

As in dicotyledonous plants, the apical/basal polarity of the maize embryo is established at the first cell division of the zygote, which gives a large highly vacuolarized basal cell thought to be the precursor of the suspensor, and a small highly cytoplasmic apical cell thought to give rise to the embryo proper. Two of the maize HD-ZipIV-class genes were expressed in the basal/suspensor region of the maize embryo at early developmental stages. ZmOCL1, as previously described, showed weak and largely uniform expression in the suspensor region during early embryogenesis in addition to expression in the embryo proper, but this basal expression was not observed at later stages. In the present study, ZmOCL3 was found to be strongly expressed specifically in the suspensor from an early stage of embryo development, with expression being excluded from apical regions until later. ZmOCL3 therefore provides a marker for suspensor tissue during early embryogenesis. Although at early stages (up to 7 DAP) expression of ZmOCL3 appeared uniform throughout the suspensor, at 9 DAP expression started to become restricted to the outer cells of the suspensor, in a manner reminiscent of the restriction of ZmOCL1 to the outer cells of the embryo proper. This spatial restriction within suspensor tissues is an important indication that, in contrast to the accepted view, the maize suspensor is an organ which undergoes cellular differentiation at a relatively early stage.

In addition to the large vacuolarized cells of the suspensor, ZmOCL3 was also expressed during early development of the basal region of the endosperm, which is destined to give rise to the small, active and highly cytoplasmic cells of the ESR. The expression of this gene during the early development of both these basal zones indicates that, if ZmOCL3 is implicated in cell fate specification, it is probably the specification of ‘basal' identity rather than ‘suspensor' or ‘ESR' identity. Expression of other suspensor-specific genes (such as ASKη from Arabidopsis) or ESR-specific genes (such as the Esr genes from maize) was not observed in both embryo and endosperm, indicating that such genes could act downstream of genes such as ZmOCL3, and in a more tissue-specific manner (Dornelas et al. 1999; Opsahl-Ferstad et al. 1997). Unfortunately, the delicate structure of maize embryos at the one- and two-cell stages made it impossible to pinpoint the exact timing of initiation of ZmOCL3 expression in the basal region.

Gene expression in the ‘apical' region of the embryo

Previous studies of the expression of the HD-ZipIV-class gene AtML1 from Arabidopsis led to the proposal that this gene could be implicated not only in protoderm specification, but also in the establishment of ‘apical' identity in the two-cell embryo due to its expression specifically within the apical cell at this stage (Lu et al. 1996). In maize, ZmOCL1 and two of the HD-ZipIV-class genes described in this paper (ZmOCL4 and ZmOCL5) showed expression in the apical part of the embryo during early embryogenesis. However, none of these three genes, including ZmOCL5 which is strikingly similar to AtML1 at the molecular level, showed an expression pattern exactly comparable to that of AtML1.

In contrast to AtML1 and ZmOCL1, transcripts of ZmOCL4 and ZmOCL5 were first detected at the tip of the embryo proper at 5 DAP. From 6 to 8 DAP expression spread down the adaxial and abaxial faces of the embryo, respectively. The cumulative expression pattern of these genes thus mirrors exactly the process of protoderm formation as described from cytological studies by Randolph (1936). The coincidence of the initiation of ZmOCL4 and ZmOCL5 expression with protoderm formation suggests that these genes are not implicated in the specification of protoderm per se. It could be that such genes act downstream of genes such as ZmOCL1 and act to maintain protoderm identity and/or to specify subsequent protoderm subcompartments. The observation that ZmOCL4 has an abaxial, and ZmOCL5 a more adaxial expression pattern during early embryogenesis tends to support the later proposition. During maize embryogenesis the specification of protoderm is rapidly followed by its subdivision into visibly different cell types. The epidermal cells on the abaxial face of the scutellum (the monocotyledonous equivalent of a cotyledon) are recognizably different in shape and division pattern to those on the adaxial (meristem-forming) face of the embryo from an early stage (Van Lammeren 1986). It is possible that ZmOCL4 and ZmOCL5 are among the genes responsible for this differentiation. It should be remembered that the expression pattern described for ZmOCL5 is the expression of two genes and that it is possible that the expression patterns of ZmOCL5α and ZmOCL5β differ.

Gene expression in meristematic regions during embryogenesis, vegetative and floral development

In both animal and plant systems it is common for the master proteins involved in early developmental decisions and specification of cell fate to play roles later in development, either in the maintenance of cell fate within certain groups of cells, or in later cell-specification events. It was therefore not surprising that the genes expressed during protoderm specification in the maize embryo (ZmOCL1, ZmOCL4 and ZmOCL5) continued to be expressed within the L1 cell layer of vegetative, inflorescence and floral meristems, and in young organ primordia, during the adult life of the plant. In our study all three genes showed strong expression in the outer cell layer of all meristematic regions tested. However, expression of ZmOCL1 was slightly different from that of ZmOCL4 and ZmOCL5 in that it was down regulated in the stamen primordia of female flowers and, in addition, it was less strictly excluded from underlying cell layers, with some expression being visible in the cell layer underlying the L1 or epidermis. It is probable that the continued expression of ZmOCL1, ZmOCL4 and ZmOCL5 in the meristematic region reflects roles for all three genes in the maintenance of cell layer identity. This function may be especially important in the vegetative phase of monocotyledonous species such as maize, where clonal analysis has demonstrated that L1-derived cells contribute to many of the cell layers of the mature leaf, and indeed that leaf primordia are initiated by periclinal cell divisions in the meristematic L1 layer (Abbe et al. 1951). In addition, in maize it has been shown that the clonal separation of the L1 layer from underlying cells within the vegetative meristem is not nearly so strict as in dicotyledonous species, and that interlayer invasions may well be relatively common (Dawe & Freeling 1990; McDaniel & Poethig 1988). Interestingly, ZmOCL4 and ZmOCL5 were expressed in mature stems and leaves, respectively. The expression of ZmOCL5 in leaves is particularly surprising as AtML1, its closest homologue from Arabidopsis, was reported to show no expression in mature tissues (Lu et al. 1996)

The ZmOCL2 gene showed a meristem-specific expression pattern. ZmOCL2 expression was unprecedented for an HD-ZIPIV-class gene as it appeared to be specifically excluded from the meristematic L1 layer. In addition, expression of ZmOCL2 appeared to be downregulated as soon as leaf primordia started to develop. Expression in vegetative meristems thus superficially resembled that reported for Kn1, although unlike Kn1, ZmOCL2 expression did not spread into the vascular tissue beneath the meristem, and appeared to be stronger in the cells immediately underlying the L1 layer than in the central cells of the meristem (Vollbrecht et al. 1991). The expression pattern of ZmOCL2 differed between vegetative and floral development, with expression in floral meristems being more tightly restricted to one layer of cells immediately underlying the L1 layer (i.e. in a cell layer corresponding to the L2 cell layer of Arabidopsis). To our knowledge, ZmOCL2 is the first gene isolated either from monocotyledonous or dicotyledonous species to show this type of expression pattern. This observation is particularly surprising as it suggests that three cell layers exist in maize, at least in floral/inflorescence meristems: an epidermal (L1) layer; a second layer (similar to the L2 layer of Arabidopsis); and a corpus. Clonal analysis in maize indicates that the tunica is composed of only one cell layer. However, recent observations in Iris (a monocotyledonous species considered to have a two-layered tunica) have shown that the symplastic connections linking the L2 to the L1 and the corpus become much less numerous after the vegetative–reproductive transition (Bergmans et al. 1997). Although similar studies have not been carried out in maize, it has been noted that the clonal isolation of the L1 layer if maize is much stricter in flowers (male) than in vegetative regions (Dawe & Freeling 1990). These observations, taken together with the expression pattern of ZmOCL2, may indicate that in maize, as in Iris, meristem organization undergoes important changes at the vegetative–reproductive transition. It may be that ZmOCL2 is involved in specification of ‘not L1' cell fate within meristems, and that this role is refined in floral meristems to give a specific function in an increasingly isolated L2-like subepidermal layer.

Only one of the genes analysed, ZmOCL3, was entirely excluded from meristematic regions at the level of RNA expression. Weak epidermis-specific expression was, however, observed in most floral organs and leaves shortly after their initiation, with particularly strong expression being observed in the stamen primordia in both male and female flowers. It is therefore possible that ZmOCL3 has a secondary role in developing organs. Unlike ZmOCL1, ZmOCL4 and ZmOCL5, ZmOCL3 was undetectable in mature tissues by RT–PCR.

Finally, unlike ZmOCL1, none of the four new genes described in this paper shows expression in in situ hybridizations of the developing primary root primordium. The recently characterized ANL2 gene of Arabidopsis, which is most closely related to ZmOCL1, appears to have an effect on primary root development, further supporting the proposition that ZmOCL1 may play a unique and important role in specification of the primary root protoderm (Kubo et al. 1999).

In conclusion, the HD-ZipIV gene family of maize is composed of at least six members which show a range of novel overlapping tissue- and cell layer-specific expression patterns. These expression patterns indicate that this gene family may be involved in several key processes during the development of plant embryos, meristems and organ primordia. These genes provide an important source of markers for various tissues during embryogenesis, and thus provide a valuable resource in the analysis of existing mutants affected in embryo development.

Experimental procedures

Plant material

The maize inbred line A188 (Gerdes & Tracy 1993) and maize hybrid DH5 × DH7 (Barloy et al. 1989) were grown in a growth chamber with a 16 h illumination period (700 μEm−2 sec−1) at 24/19°C (day/night) and 80% relative humidity. Plants were hand pollinated for the isolation of staged tissue.

Library constructions and screens

A total of 500 embryos of inbred line A188 were dissected at stage 1 (12 DAP) and used for the construction of a cDNA library in λZAP II (Stratagene). A genomic library of the hybrid DH5 × DH7 was prepared in λEMBL3 SP6/T7 (Clontech). To isolate ZmOCL2 genomic clones, the genomic library was screened at medium/high stringency (hybridization at 55°C, washes at 60°C, 0.5 × SSC and 0.1% SDS) with a full-length ZmOCL1 cDNA labelled using a random-primed DNA labelling kit (Boerhinger). One of the resulting clones was subcloned, and hybridizing subclones were partially sequenced to allow primers to be designed (see below).

To isolate cDNA clones of ZmOCL2 and related genes, the cDNA library was screened at medium stringency (hybridization at 55°C, washes at 55°C, 0.5 × SSC and 0.1% SDS) with a 600 bp ZmOCL2 probe obtained by amplification of genomic DNA with primer ZmOCL2-5 (5′-GTCAAGTTCTGGTTCCAGAACCG) situated in the ZmOCL2 homeo domain and primer ZMOCL2-4 (5′-CTGTTCCCGTTGAGCTGGATGTG) situated downstream of the leucine zipper.

Poly(A)+ RNA isolation, 5′ and 3′ RACE-PCR

mRNA was isolated from tissue samples using the STRAIGHT A'sTM mRNA Isolation System (Novagen). In order to isolate complete ZmOCL2 cDNA sequences, RACE samples were prepared from the mRNA of young inflorescences using the Marathon RACE kit (Clontech). 3′ RACE was carried out using ZmOCL2-5 and a nested primer, ZmOCL2-6 (5′-TGGTGTGCA- GCTGCTGTGGCGG) and 5′ RACE was carried out using ZmOCL2-3 (5′-TCAATGGTGACAATGCCGGATTCCC) with ZmOCL2-4 as a nested primer. Reactions were carried out in a Perkin Elmer 2400 cycler using the Advantage cDNA PCR kit (Clontech) following a touchdown PCR procedure: 5× (95°C/5 sec, 72°C/4 min), 5× (95°C/5 sec, 70°C/4 min), 30× (95°C/5 sec, 68°C/4 min). Products were visualized by agarose-gel blotting and hybridizing with existing genomic sequences.

Cloning and sequence analysis

RACE products were cloned into the pGEM-T Easy vector (Promega). All other cloning/subcloning was carried out in the vectors BSSK+ and BCSK+ (Stratagene). For nucleotide sequence analysis, template DNA was isolated with the QIAprep spin plasmid miniprep kit (Qiagen), sequenced with the dye terminator PRISM ready reaction AmpliTaq FS kit (Perkin Elmer) and run on an ABI 373A automated sequencer (Applied Biosystems). The DNA sequences were given the following accession numbers: AJ250984(ZmOCL2), AJ250985(ZmOCL3), AJ250986(ZmOCL4) and AJ250987(ZmOCL5). Phylogenetic trees were constructed using the Clustal method of the DNAstar module Megalign (Lasergene).


Using the method described above, mRNA was isolated from mature leaves, mature stems, mature root mass, the meristematic region of germinating seedlings (including young leaf bases), root tips of germinating seedlings (lateral and primary roots), and endosperms dissected at 9 DAP. Samples were reverse transcribed using the superscript II reverse transcriptase system (Gibco-BRL) in the presence of the primer 5′-GGGCTCGAGTTTTTTTTTTTTTTT. Samples were standardized using primers in the constitutively expressed gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH); GAPDH-3 and GAPDH-5 (Richert et al. 1996). PCRs were then carried out for 20 cycles (94°C/30 sec, 60°C/30 sec, 72°C/60 sec) in a Perkin Elmer 2400 cycler in the presence of pairs of gene-specific primers. Primer pairs used were as follows: ZmOCL1, ZmOCL1-3 and ZmOCL1-5 (expected product 410 bp in cDNA, 530 bp in genomic DNA) (Ingram et al. 1999); ZmOCL2, ZmOCL2-4 and ZmOCL2-6 (expected product 307 bp in cDNA and 425 bp in genomic DNA); ZmOCL3, ZmOCL3-3 (5′-TGGCCTGAGGGTTTCGGAATTCATCAC) and ZmOCL3-5 (5′-CAATCAGGCCCATACGACCCAAGAC) (expected product 607 bp in cDNA and 696 bp in genomic DNA); ZmOCL4, ZmOCL4-3 (5′-GCACCGTGGTTATCATGAACAGCAT- CAC) and ZmOCL4-5 (5′-TCCAAGATTGGCTCCCAGAAGCACTG) (expected product 656 bp in cDNA and 1630 bp in genomic DNA); ZmOCL5, ZmOCL5-3 (5′-TGGCAGGGAGCTACAATGGTGCATT- ACA) and ZmOCL5-5 (5′-CCAAACGCGAGGCCTGAATTGACCAA) (expected product 410 bp in cDNA and 520 bp in genomic DNA). Products were visualized by agarose gel blotting and hybridizing with the appropriate cDNA fragments under stringent conditions.

RFLP mapping

Radioactively labelled cDNA probes were used to identify restriction fragment length polymorphisms between the parents of various inbred mapping populations (Burr & Burr 1991; Maugenest et al. 1997), and the resulting polymorphisms were scored within the corresponding inbred populations. The data were analysed with the mapmaker program on the current Brookhaven National Laboratory map (Brookhaven National Laboratory, Upton, NY, USA).

In situ hybridization

The methods for digoxygenin labelling of RNA probes, tissue preparation and in situ hybridization were essentially as described by Coen et al. (1990) with the modifications described by Bradley et al. (1993). All probes used were transcribed from subclones of cDNAs from which the homeo box region and poly A tail had been removed.


We thank P. Perez and A. Murigneux (Biogemma SA, Clermont-Ferrand, France) as well as B. Burr (Brookhaven National Laboratory, USA) for use of their recombinant inbred lines and the analysis of our RFLP data. Excellent technical assistance was provided by Fabienne Deguerry, Monique Estienne, Hervé Leyral and Richard Blanc. G.I. was supported by an EMBO long-term fellowship and an INRA fellowship. The work was supported in part by contract BIO4-CT96-0210 of the European Commission.


  1. EMBL accession numbers AJ250984, AJ250985, AJ250986and AJ250987.