Establishment of Cereal Endosperm Expression Domains: Identification and Properties of a Maize Transfer Cell–Specific Transcription Factor, ZmMRP-1

Abstract

In maize, cells at the base of the endosperm are transformed into transfer cells that facilitate nutrient uptake by the developing seed. ZmMRP-1 is the first transfer cell–specific transcriptional activator to be identified. The protein it encodes contains nuclear localization signals and a MYB-related DNA binding domain. A single gene copy is present in maize, mapping to a locus on chromosome 8. ZmMRP-1 is first expressed soon after fertilization, when the endosperm is still a multinuclear coenocyte. The transcript accumulates in the basal nucleocytoplasmic domain that gives rise to transfer cells after cellularization. The transcript can be detected throughout transfer cell development, but it is not found in mature cells. ZmMRP-1 strongly transactivates the promoters of two unrelated transfer cell–specific genes. The properties of ZmMRP-1 are consistent with it being a determinant of transfer cell–specific expression. Possible roles for ZmMRP-1 in the regulation of endosperm and transfer cell differentiation are discussed.

INTRODUCTION

The endosperm is a triploid seed tissue in which the nutrients required for the early stages of seedling growth accumulate as storage proteins. The basal endosperm contains a layer of transfer cells that develop extensive cell wall ingrowths supporting an enlarged plasmalemma surface area that promotes solute transport (Pate and Gunning, 1972; Thompson et al., 2001). In addition to the most basal cell layer (the basal aleurone layer), two or three adjacent endosperm cell layers also possess cell wall ingrowths, decreasing in size toward the kernel center (Schel et al., 1984; Shannon et al., 1986; Davis et al., 1990). The absence of a properly formed transfer cell layer, as seen in endosperms with an imbalanced parental genome ratio, is correlated with reduced rates of grain filling and seed abortion (Brink and Cooper, 1947; Charlton et al., 1995).

Detailed cytological studies of endosperm development have been reported for maize, barley, and Arabidopsis (Kiesselbach, 1949; Berger, 1999; Brown et al., 1999; Olsen et al., 1999). Endosperm develops from the fertilized triploid central cell, which undergoes a period of nuclear divisions without cytokinesis to produce a coenocyte containing nuclei distributed evenly throughout the peripheral cytoplasm of the central cell. Once the nuclear division phase is complete, the multinucleate homogeneous cytoplasm reorganizes into nucleocytoplasmic domains and endosperm cellularization commences (∼100 hr after pollination in maize). Periclinal cell division of the first cell layer gives rise to two cell types: the outer aleurone and inner starchy endosperm initials. Further divisions yield a central mass of starchy endosperm surrounded by a single-cell aleurone layer. At the base of the endosperm, however, aleurone and endosperm initials differentiate into the basal endosperm transfer layer (BETL), possibly in response to signals from the maternal chalazal pad.

An interaction between maternal tissues and endosperm has been illustrated by studies with the miniature-1 maize mutant, in which a deficiency in endosperm transfer cell invertase activity, encoded by the Incw-2 gene, leads to alterations in maternal tissue (Cheng et al., 1996). Studies on transfer cells in the cotyledons of Vicia faba suggest the involvement of sugar uptake in the induction of transfer cell differentiation (Offler et al., 1997). Alternately, signals governing endosperm regional differentiation might be present already in the central cell.

The basal endosperm contains, in addition to transfer cells, a second cell type, the embryo-surrounding region cells, around the suspensor and basal half of the embryo in fully developed seed. This domain cellularizes earlier than the rest of the endosperm coenocyte, possibly in response to signals from the embryo. In maize, the identification of genes expressed exclusively in the embryo-surrounding region (Esr-1; Opsahl-Ferstad et al., 1997) and BETL cells (BETL-1, -2, -3, and -4; Hueros et al., 1995, 1999b) demonstrates that these are separate gene expression domains.

The BETL-specific genes isolated to date share some structural features that suggest that they play a role in defense against pathogen entry into the developing seed. They encode small, Cys-rich secreted peptides that accumulate only transiently during endosperm development. In one gene, BAP-2 (synonymous with BETL-2), the mature peptide is secreted and accumulates to form an antifungal barrier in the placentochalaza (Serna et al., 2001). BETL gene expression is detected as early as 5 days after pollination (DAP), at the start of the cellularization stage (Hueros et al., 1999b). The barley endosperm-specific transcript END-1 (Doan et al., 1996) was localized to cytoplasmic domains surrounding basal coenocyte nuclei. This indicates that factors determining transfer cell–specific expression must be present in the coenocyte before cellularization has occurred, although their identity is unknown at present.

The MYB domain is a DNA binding region of ∼52 amino acids. In metazoan MYB proteins, the domain is repeated three times, with a helix-helix-turn-helix structure in each repeat. In plants, the predominant MYB protein class contains two repeats, although members with one and three MYB domains are known (Martin and Paz-Ares, 1997; Jin and Martin, 1999; Kranz et al., 2000). The DNA recognition domain of the single–MYB domain proteins differs from those present in classic R1-R2-R3 and R2-R3 proteins and may bind DNA in a manner resembling homeodomain proteins to form heterodimers or homodimers (Nishikawa et al., 1998). Rose et al. (1999)identified a family of genes that exhibit a highly conserved amino acid motif, SHAQK(Y/F)F, within the single MYB-related DNA binding domain. Members of this family of MYB-related proteins are known to be DNA binding proteins (MybSt1; Baranowskij et al., 1994) and transcriptional regulators, such as LeMYB1 (Rose et al., 1999). They also include the circadian clock–associated LHY (Schaffer et al., 1998) and CCA1 (Wang et al., 1997).

This article reports the identification of a novel MYB-like gene (ZmMRP-1) belonging to the SHAQKYF family of single–MYB domain proteins. The gene is expressed specifically in endosperm transfer cells, accumulating at the basal pole of the endosperm coenocyte soon after fertilization. The ZmMRP-1–encoded protein is localized in nuclei and can transactivate the promoters of transfer cell–specific genes. A role for ZmMRP-1 in regulating transfer cell differentiation is proposed.

RESULTS

Isolation of the MYB-Related Transfer Cell–Specific Gene

ZmMRP-1 was isolated in the course of differential screening to identify new BETL-specific genes. The procedure used was based on that of Sokolov and Prockop (1994). Briefly, young maize seed (10 DAP) were dissected to separate the bottom (basal part of the kernel) and top (upper part) halves. mRNA was extracted and reverse transcribed into bottom and top cDNAs. These cDNAs were used as templates for polymerase chain reaction (PCR) using pairs of 10-mer oligonucleotides of arbitrary sequence as PCR primers (Williams et al., 1990). Comparison of the band patterns produced from either top or bottom cDNAs allowed the identification of bottom-specific PCR bands. These were cloned and sequenced. RNA gel blot analyses using these bands as probes confirmed that one was amplified from a basal kernel-specific transcript (data not shown). This clone, which had sequence similarities with the MYB family of transcriptional activators, was designated ZmMRP-1 (for Zea mays MYB-Related Protein-1).

To obtain the full-length cDNA for ZmMRP-1, an immature (10-DAP) endosperm cDNA library was screened (Hueros et al., 1995) and clones covering the 3′ end but with truncated 5′ ends were obtained. Rapid amplification of cDNA ends (RACE) (Marathon RACE kit; Clontech, Palo Alto, CA) then was used to amplify the 5′ end. The antisense primer for RACE (RACE1 in Figure 1A)

Figure 1.

ZmMRP-1 Contains a MYB-Related Domain.

(A) Nucleotide sequence of the ZmMRP-1 cDNA and deduced amino acid sequence of the protein. The nucleotide sequence obtained by RACE is shown in italic type. The putative MYB-like DNA binding domain is shaded. Arrows above the sequence mark the positions of different primers used during the characterization studies (see text). An arrowhead marks the position of the single intron of the gene. The stop codon is marked with an asterisk.

(B) Amino acid alignment of the putative DNA binding domain of ZmMRP-1 and those from MCB1, LeMYB1, MybSt1, LHY, CCA1, and the maize R2-R3 MYB gene C1. Amino acids identical in at least four sequences are shaded.

was positioned 71 bp from the 5′ end of the available cDNA sequence. Continuity between the RACE product and the cDNA clone was confirmed by reverse transcriptase–mediated (RT)-PCR (Figure 5)

Figure 5.

RT-PCR Analysis of the Expression of ZmMRP-1.

Total RNA samples (1 μg) from coleoptiles (C), leaves (L), roots (R), silk (S), tassels (T), unpollinated flowers (U), kernels at 3, 8, 17, and 29 DAP, no RNA (0), and genomic DNA (DNA).

(A)  ZmMRP-1–specific primers.

(B) Primers designed to amplify a ubiquitous, weakly expressed R2-R3 MYB gene.

In both cases, PCR primers were designed to span a genomic sequence containing an intron. The arrow in (A) indicates the ZmMRP-1 band derived from cDNA. Size markers are indicated at left.

using a forward primer (F in Figure 1A) derived from the RACE product and a reverse primer (R in Figure 1A) derived from the cDNA clone. The sequence obtained by merging the RACE product and the longest cDNA was 1041 bp long, in agreement with the estimated length of ZmMRP-1 from RNA gel blot analyses (∼1.1 kb; Figure 2)

Figure 2.

RNA Gel Blot Analysis of ZmMRP-1 Expression.

mRNA samples (2 μg) from different maize tissues were hybridized with a ZmMRP-1 probe (A) or a ubiquitin probe (B). T, upper part of 10-DAP developing kernels; B, lower part of 10-DAP developing kernels; U, unpollinated flowers; R, roots; C, coleoptiles; S, silk; Ts, tassels; G, germinating kernels (detached from coleoptiles and roots).

. Figure 1shows the complete sequence of ZmMRP-1 with the RACE-derived sequence in italics. The positions of the 10-mer primers E18 and T7, which were used to amplify the initial cDNA product, are indicated by arrows.

The full-length cDNA contains an open reading frame of 242 amino acids (Figure 1A), predicting a protein with a molecular weight of 27,592 and a pI of 8.91. The ZmMRP-1–encoded polypeptide contains a single centrally located MYB domain (shaded in Figure 1A) most similar in sequence (Figure 1B) to that present in the transcriptional activators LeMYB1 of tomato (52% identity; Rose et al., 1999), MybSt1 of potato (49% identity; Baranowskij et al., 1994), and the barley genes MCB1 and MCB2 (52% identity), transcription factors that bind to chlorophyll a/b binding protein promoters. Lower similarity was observed with the MYB-like proteins LHY (41% identity; Schaffer et al., 1998) and CCA-1 (33% identity; Wang et al., 1997) from Arabidopsis. The first MYB repeat of the transcriptional activator C1 (Paz-Ares et al., 1987) also is included in Figure 1Bto show the divergence between the ZmMRP-1–related MYB genes and the R2-R3 plant MYB genes.

Rose et al. (1999)reported the presence of a highly conserved domain, the SHAQK(Y/F)F motif, in the third putative α-helix of the DNA binding domain of a large group of proteins containing a single MYB domain. This motif, absent in the third α-helix of the canonical MYB repeats, is conserved perfectly in ZmMRP-1. More than 40 other genes containing a single–MYB-like DNA binding domain also were found in database searches. However, none were of maize origin. Most were anonymous sequences from rice (the best matches with ZmMRP-1 had up to 50% similarity in the MYB domain) and Arabidopsis genome projects. In all cases, homologies were restricted to the DNA binding domain; therefore, no ZmMRP-1 isolog was identified.

Genomic Organization and Mapping of the ZmMRP-1 Locus

The position of ZmMRP-1 in the maize genome was determined by restriction fragment length polymorphism analysis using a population of immortalized F2 lines (Tx303/CO159 IF2 1995-8) (Burr et al., 1994). Genomic DNA from the parental lines was digested with a collection of 16 restriction enzymes, of which only BglII produced a polymorphism, and used subsequently for mapping (data not shown). The gene mapped on chromosome 8 of maize, between hox1 and csu84. To our knowledge, no gene influencing seed phenotype has been mapped to this region.

Extensive genomic DNA gel blot analyses in the parental lines of the mapping population (Co159 and Tx303) (Burr et al., 1994), as well as in the inbred line A69Y (data not shown), demonstrated that ZmMRP-1 is a single-copy gene in maize. The genomic clone containing ZmMRP-1 was reconstructed by a combination of direct and inverse PCR. The genomic sequences included within the transcribed region were amplified using sense and antisense primers derived from the cDNA. Sequencing of the resulting clones revealed the presence of a single 226-bp intron at position 358 to 359 of the cDNA sequence (arrowhead in Figure 1A) flanked by GT and AG canonical intron border sequences.

The ZmMRP-1 upstream and downstream genomic sequences were obtained by inverse PCR. The coding region and upstream and downstream sequences were joined in a single contig of 2949 bp. Computer analyses of the available 770 bp of 5′ upstream sequence revealed the presence of a probable transcription start site 3 bp upstream of the putative RNA 5′ terminus and a putative TATA box (CTA-TAAATACCGGGC) 40 bp farther upstream. Both are consistent with the reported cDNA sequence being full length.

ZmMRP-1 Is Expressed Exclusively in the Basal Region of the Endosperm Coenocyte and in Endosperm Transfer Cells

The expression pattern of ZmMRP-1 was investigated first by analyzing mRNA from different maize tissues on RNA gel blots (Figure 2). The ZmMRP-1 transcript of ∼1100 nucleotides (Figure 2A) was present only in mRNA extracted from the basal part (Figure 2, lane B) of developing kernels at 10 DAP.

In a time course study of ZmMRP-1 mRNA accumulation (Figure 3)

Figure 3.

RNA Gel Blot Analysis of ZmMRP-1 Expression during Seed Development.

mRNA samples (2 μg) from maize kernels at different developmental stages (DAP indicated at top) were hybridized with a ZmMRP-1 probe (A) or a ubiquitin probe (B).

, the transcript was abundant already at the earliest time for which poly(A)⁺ RNA could be isolated (8 DAP) and decreased from 11 DAP onward. This finding is in contrast to reports for other BETL genes, which are first detectable by RNA gel blot analyses at 7 to 8 DAP and peak between 14 and 16 DAP (Hueros et al., 1995, 1999b). Thus, ZmMRP-1 expression precedes that of other BETL-specific genes.

In situ hybridization of ZmMRP-1 was performed on sagittal kernel sections at various developmental stages (Figure 4)

Figure 4.

In Situ Hybridization Analysis of the Expression of ZmMRP-1.

Sagittal sections of maize kernels at various developmental stages were hybridized with antisense ([A], [B], [C], [D], [E], [G], and [I]) or sense ([F] and [H]) RNA probes derived from the 3′ end of the ZmMRP-1 cDNA.

(A) and (B) Less than 3-DAP kernel. (B) shows the same section as (A) but at higher magnification and after counterstaining with DAPI to show the nuclei.

(C) and (D) 3 DAP.

(E) and (F) 5 DAP.

(G) and (H) 11 DAP.

(I) 16 DAP.

(J) Development of a maize kernel. The endosperm, as a coenocyte at 1 and 3 DAP or cellular after 5 DAP, is shown yellow. The embryo-surrounding region is shown green, and the embryo is shown blue. Drawings are not to scale. Red dots indicate nuclei where ZmMRP-1 is transcribed.

Em, embryo; En, endosperm; ESR, embryo-surrounding region; PC, placentochalaza; Ph, phloem terminals at the pedicel; TC, transfer cells. Unlabeled arrows point to positive hybridization signals (bright silver grains) at the base of the endosperms. Bars = 100 μm in (A) to (D) and 500 μm in (E), (G), and (I). (A) to (I) are dark-field photographs; sections were counterstained with calcofluor to visualize the cytoplasm and, in the case of (B) to (D), with DAPI to visualize nuclei.

. The sense probe (Figures 4F and 4Hfor 5 and 11 DAP, respectively) produced no signal in any tissue examined, whereas the antisense probe gave signal in transfer cells at <3 DAP (Figures 4A and 4B), 3 DAP (Figures 4C and 4D), 5 DAP (Figure 4E), 11 DAP (Figure 4G), and 16 DAP (Figure 4I). The intensity of the signal was very high already by 3 DAP, reached a maximum at 11 DAP, and did not decrease before 16 DAP. The decrease in ZmMRP-1 RNA at the whole-kernel level between 11 and 16 DAP could be attributable to an increase in the RNA fraction from the apical endosperm and embryo at these stages. The decrease in ZmMRP-1 mRNA accumulation observed by RNA gel blot analysis after 16 DAP was confirmed by the in situ results showing signal reduction to almost undetectable levels by 29 DAP (data not shown).

Interestingly, the ZmMRP-1 transcript was detected in the basal endosperm at 3 DAP, before cell wall formation (Figures 4C and 4D). Here, the signal was restricted to the cytoplasmic domains that surround the basal endosperm nuclei. The ZmMRP-1 transcript also was weakly detectable in the chalazal pole of the endosperm sac at 30 to 40 hr after pollination (Figures 4A and 4B), or ∼10 hr after fertilization (Kiesselbach, 1949). Figure 4Bshows a higher magnification of the section in Figure 4A, after counterstaining with 4′,6-diamidino-2-phenylindole (DAPI) to reveal the nuclei.

In addition to the basal endosperm transfer layer, maize has transfer cells in the loading terminals of the phloem and in certain secretory glands. Because of the relatively small size of these tissues compared with the organs in which they are located, the transcript might be overlooked by RNA gel blot analysis. Therefore, the presence of ZmMRP-1 mRNA in different plant tissues was examined by RT-PCR using primers spanning the intron (Figure 5). The expected cDNA fragment (Figure 5A, arrow) was amplified only from RNA samples of kernels at different developmental stages. No product was obtained using RNA from other tissues, although primers for a ubiquitous, weakly expressed R2-R3 MYB gene (G. Hueros, unpublished data) yielded a product from the same templates (Figure 5B). Therefore, ZmMRP-1 is an endosperm transfer cell layer–specific gene and does not play a general role in the translation of polar signals in transfer cells.

ZmMRP-1 Is a Nuclear Protein

The hypothesis that ZmMRP-1 is a transcription factor in maize predicts that it possesses nuclear localization signals. To test this possibility, a translational fusion between ZmMRP-1 and green fluorescent protein (GFP) was prepared and tested by transient expression in tobacco leaf protoplasts (Figure 6)

Figure 6.

Nuclear Localization of ZmMRP-1.

Tobacco protoplasts were transformed with the GFP coding sequence under the control of the 35S promoter (A), a translational fusion between ZmMRP-1 and GFP (B), and carrier DNA alone (C). Protoplasts were photographed under UV light. Green signal, GFP fluorescence; red signal, chloroplast autofluorescence. N, nuclei.

. Figure 6Ashows the vector-only control, in which green fluorescence was distributed evenly throughout the cell. A protoplast containing only a few chloroplasts was selected for this image to avoid interference between the GFP signal and the chloroplast-derived red autofluorescence. The ZmMRP-1::GFP fusion protein (Figure 6B) yielded fluorescence concentrated in the nucleus. This was observed in all of the fluorescing cells obtained (transformation efficiency was 1 to 5%). No signal was observed in untransformed tobacco protoplasts (Figure 6C). Thus, ZmMRP-1 contains signals that can promote the accumulation of GFP within the nuclei and very likely is a nuclear protein in maize. However, no putative localization signals were detected after extensive scrutiny of the protein motif databases with the ZmMRP-1 amino acid sequence. This is not surprising because there is no strict consensus sequence for the nuclear localization signals (Buchanan et al., 2000). Consequently, in many cases, functional analysis is the only approach that can be used to identify these motifs.

ZmMRP-1 Transactivates Transfer Cell–Specific Promoters in Tobacco Protoplast Cotransformation Experiments

To determine if ZmMRP-1 could function in maize as an activator of BETL promoters, cotransformation experiments were performed in tobacco leaf protoplasts. Promoter sequences from ZmMRP-1 (0.8 kb; isolated in this work) or from the previously isolated BETL-1 (1 kb), BETL-2 (1.6 kb), BETL-3 (0.8 kb), and BETL-4 (0.5 kb) genes (Hueros et al., 1999a, 1999b) were fused to the β-glucuronidase (GUS) coding sequence and the nopaline synthase (NOS) terminator, generating plasmids pZmMRP1-GUS, pBETL1-GUS, pBETL2-GUS, pBETL3-GUS, and pBETL4-GUS. The ZmMRP-1 protein was expressed from the ubiquitin promoter (pUBI-MRP) with the ubiquitin promoter fused to the NOS terminator (pUBI-NOS) as a negative control. The ubiquitin promoter fused to the luciferase coding sequence (pUBI-LUC) was used as an internal control for each transformation experiment. Finally, two positive control plasmids, pUbi-GUS and p35S-GUS, were used for comparison of the relative GUS activities.

Figure 7A  

Figure 7.

ZmMRP-1 Transactivates BETL-Specific Promoters in Tobacco Protoplasts.

(A) Transactivation of BETL promoters in tobacco protoplasts. Tobacco protoplasts were cotransfected with the promoter–GUS reporter plasmid indicated in the key at bottom, plus either a control plasmid containing the ubiquitin promoter and the NOS terminator (−) or the effector plasmid ubiquitin promoter ZmMRP-1 coding sequence NOS terminator (+). The scale on the y axis is logarithmic; the mean obtained for each experiment is shown above the corresponding column.

(B) Deletion analyses of the BETL-2 promoter. GUS activity measurements are the means of four independent experiments. The error bars show the average scatter from the four experiments. GUS activity is expressed as μmol 7-hydroxy-4-methylcoumarin mg⁻¹ protein min⁻¹. Green, REP construct; dark blue, full-length promoter; light blue, DEL construct; magenta, no DNA added. The effector plasmid used is indicated below each group of bars. O2, Opaque-2; O2(AD), activation domain from O2. Error bars indicate ±sd.

shows the results of the cotransformation experiments. Relative GUS/LUC activities are presented for experiments in which the effector plasmid was either the negative control pUBI-NOS (−) or pUBI-MRP (+). Columns represent the mean of at least three independent experiments (mean values are shown above the columns; note that the y axis scale is logarithmic). The construct pZmMRP1-GUS (light green columns) was not transactivated significantly by ZmMRP-1, indicating that ZmMRP-1 probably cannot bind to its own promoter. In contrast, pBETL1-GUS (dark red columns) was activated strongly by ZmMRP-1, showing an 86-fold increase in activity upon cotransformation with pUBI-MRP. Similar results were obtained with the BETL-2 promoter (dark green columns).

Remarkably, the interaction pUBI-MRP/pBETL1-GUS or pUBI-MRP/pBETL2-GUS gave relative GUS/LUC activities 3 and 10 times higher, respectively, than pUBI-GUS (blue column). These results strongly suggest that ZmMRP-1 is an activator of the expression of BETL-1 and BETL-2 in maize endosperm transfer cells. In contrast, both BETL-3 (dark blue columns) and BETL-4 (light red columns) promoters were active in protoplasts in the absence of ZmMRP-1 and did not respond to ZmMRP-1.

The interaction between ZmMRP-1 and the BETL-2 promoter was studied further by testing the ability of two promoter deletions to be activated by the ZmMRP-1 protein. In addition to the full-length (1.6-kb) BETL-2 promoter (Figure 7B, dark blue), a 165-bp fragment covering a simple sequence repeat (SSR) region (from −259 to −94, termed REP; green in Figure 7B) and the full-length promoter deleted for this region (DEL; light blue in Figure 7B) were tested. The promoter fragments REP and DEL were fused to a −46 minimal cauliflower mosaic virus 35S promoter–GUS construct. The three reporter constructs were cotransformed into tobacco protoplasts with a series of three effector constructs expressing ZmMRP-1, ZmMRP-1 fused to the activation domain of the transcription factor Opaque-2 (Schmitz et al., 1997), or the complete Opaque-2 protein. The results (Figure 7B) confirmed the specificity of the interaction between ZmMRP-1 and the BETL-2 promoter. Activation was limited to promoter fragments containing the SSR region, although, because the REP construct contains additional sequences, further deletion experiments should be performed before concluding that this represents the ZmMRP-1 binding site. ZmMRP-1 alone was capable of effecting activation, which increased twofold on fusion with the Opaque-2 activation domain. Opaque-2 alone was unable to activate transcription from any version of the BETL-2 promoter.

ZmMRP-1 Ectopically Transactivates a Transfer Cell–Specific Promoter in Transgenic Maize Plants

The cotransformation experiments in tobacco protoplasts indicated a possible in vivo interaction between ZmMRP-1 and the promoters of BETL genes. Because this is a heterologous system in which both effector and reporter plasmids are present in excess, ZmMRP-1 transactivation of a BETL promoter integrated into the maize genome was tested after particle gun bombardment of maize leaves. Transgenic maize plants containing the BETL-1 promoter fused to GUS express the reporter exclusively in the transfer cells of the developing endosperm, GUS expression being undetectable in leaves (Hueros et al., 1999a). The leaves of transgenic BETL-1/GUS plants were bombarded a minimum of 10 times per experiment with the ZmMRP-1 gene under the control of the ubiquitin promoter (Figure 8B)

Figure 8.

Transactivation of the BETL-1 Promoter by ZmMRP-1 in Transgenic Plants.

Transgenic maize leaves (the structures of the transgenes are shown above the leaves) were bombarded with a ubiquitin promoter–GUS construct (A), a ubiquitin promoter–ZmMRP-1 construct (B), or a ubiquitin promoter construct (C). In all cases, the NOS terminator (T) was used. After 24 hr of incubation, leaves were stained for GUS (indigo spots). PROM, promoter; Ubi, ubiquitin.

. Histochemical staining of the bombarded leaves revealed that the integrated BETL-1/GUS reporter was induced by the ZmMRP-1 protein. Bombardment with the ubiquitin promoter alone did not produce blue spots (Figure 8C). As a positive control, transgenic leaves were bombarded with the ubiquitin promoter fused to GUS (Figure 8A). These leaves consistently displayed more spots than did those bombarded with the ubiquitin promoter. The same result was obtained using three different BETL-1/GUS transgenic lines (Hueros et al., 1999a).

DISCUSSION

ZmMRP-1 Is a MYB-Related Gene Specific for Endosperm Transfer Cells

Using a PCR-based differential screening procedure, a new BETL-specific gene, ZmMRP-1, was isolated. The polypeptide encoded by ZmMRP-1 contains a 53–amino acid domain highly homologous (Figure 1B) with a MYB-related DNA binding domain identified in several DNA binding proteins (MybSt1 [Baranowskij et al., 1994]) and transcriptional regulators (LeMYB1 [Rose et al., 1999], LHY [Schaffer et al., 1998], and CCA1 [Wang et al., 1997]). MYB-related genes possess a single central MYB domain, in contrast to classic MYB genes, which have two or three N-terminal domains (Martin and Paz-Ares, 1997; Jin and Martin, 1999; Kranz et al., 2000). The ZmMRP-1 DNA binding domain contains the characteristic SHAQK(Y/F)F motif (Rose et al., 1999). Proteins that contain this motif fall into two classes, one that includes the circadian clock–associated genes CCA1 (Wang et al., 1997) and LHY (Schaffer et al., 1998) and a second group (to which ZmMRP-1 belongs) that includes the rbcS gene regulators LeMYB1 (Rose et al., 1999) and MybSt1 (Baranowskij et al., 1994).

ZmMRP-1 Expression Precedes the Formation of Transfer Cells and the Expression of Other BETL Genes

ZmMRP-1 transcripts appear earlier than those of previously identified BETL genes (Hueros et al., 1995, 1999b; Doan et al., 1996), paralleling the development of transfer cells (Figure 4). ZmMRP-1 expression was detected first ∼10 hr after pollination (Figures 4A and 4B), and it was highly expressed already by 3 DAP (Figures 4C and 4D), when the endosperm coenocyte is reorganizing into nuclear-cytoplasmic domains (Olsen et al., 1999). By 3 DAP, the basal nuclei start to differentiate as a separate domain within the endosperm, as indicated by the expression of other transfer cell markers (Doan et al., 1996; G. Hueros, unpublished data). Between 5 and 16 DAP (Figures 4E, 4G, and 4I), cellularization and the differentiation of transfer cells occurs, reflected in the progressive accumulation of cell wall ingrowths. Despite the pronounced polar location of signals in the coenocyte, the ZmMRP-1 transcript is later distributed evenly in the cytoplasm of individual cells. The mechanism responsible for the polar localization of ZmMRP-1 mRNA, therefore, is restricted temporally.

Two basic mechanisms might exist. (1) The active localization of mRNA synthesized by all coenocytic nuclei. The targeting of a plant mRNA to a specific cellular location through protein anchorage to the cytoskeleton has been reported (Choi et al., 2000; Sami-Subbu et al., 2001). (2) The synthesis of ZmMRP-1 RNA only by basal nuclei in the coenocyte, with apparent basal localization reflecting nonequilibrium diffusion from the site of synthesis. A combination of both processes is possible as well, the basic difference being the primary target for the polar signal. In mechanism 1, this would be ZmMRP-1 mRNA; in mechanism 2, it would be the ZmMRP-1 promoter itself. These alternatives could be tested with transgenic plants containing the ZmMRP-1 promoter fused to reporter genes.

Transfer cell morphology remains essentially unchanged from 16 DAP until maturity. During this period, the ZmMRP-1 mRNA concentration decreases, with only trace amounts remaining by 29 DAP (data not shown). ZmMRP-1 is not detected at any stage in the embryo-surrounding region, a distinct domain of gene expression in the basal endosperm (Opsahl-Ferstad et al., 1997).

ZmMRP-1 Has the Properties of a Regulator of Transfer Cell–Specific Expression

The response of a number of promoter sequences from transfer cell–specific genes to ZmMRP-1 has been investigated (Figure 7A). Cotransformation with a plasmid overexpressing ZmMRP-1 activated the promoters of the BETL-1 and BETL-2 genes. MybSt1 also is capable of autonomous activation of a target promoter in a heterologous system (Baranowskij et al., 1994). Both the isolated BETL-1 (Hueros et al., 1999a) and BETL-2 (R. Thompson, unpublished data) promoter regions have been shown to confer transfer cell–specific expression on transgenic maize, although they share little sequence similarity. The relative promoter activities obtained after transactivation with ZmMRP-1 (GUS/LUC mean values of 1.034 for pBETL1-GUS and 4.071 for pBETL2-GUS) correspond well with the relative amounts of BETL-1 and BETL-2 mRNAs in developing endosperm (Hueros et al., 1999b). In contrast, the promoter regions used from BETL-3 and BETL-4, and from ZmMRP-1 itself, were not transactivated significantly by ZmMRP-1, possibly because the required cis-acting motifs were missing.

The binding site for potato MybSt1 contains the core element GGATA (Baranowskij et al., 1994), whereas tomato LeMYB1 binds at GGATGAGATAAGA (Rose et al., 1999). Interestingly, both BETL-1 and BETL-2 promoters contain a microsatellite repeat near the putative TATA box, with the sequence (GATA)_n 5′ to 3′ in BETL-2 and the complementary sequence (TATC)_n 5′ to 3′ in BETL-1. This provokes the speculation that ZmMRP-1 could bind to these sequences. Transient expression studies demonstrated that a 166-bp fragment from −259 to −94 in BETL-2, which contains the repeat, mediates the transactivation by ZmMRP-1. Experiments are now under way to define more precisely the target for ZmMRP-1 in BETL-1 and BETL-2.

Further support for a specific interaction is provided by the observation that a transcription factor, Opaque-2, which regulates gene expression in apical endosperm cells (Schmidt et al., 1990; Lohmer et al., 1991), was unable to activate the BETL-2 promoter. Also supporting the specificity of the interaction between ZmMRP-1 and BETL-specific promoters, it is known that ZmMRP-1 fails to transactivate the promoter of a gene (AE3; Magnard et al., 2000) expressed specifically in the embryo-surrounding region of the kernel (P. Rogowsky, personal communication).

In maize lines containing the GUS gene under the control of the BETL-1 promoter, GUS activity is restricted to endosperm transfer cells (Hueros et al., 1999a). Leaves of these plants, however, expressed the BETL-1/GUS reporter after bombardment with a plasmid expressing ZmMRP-1, showing that the ZmMRP-1 protein also is capable of activating an integrated copy of the BETL-1 promoter in a differentiated heterologous cell type.

How Does ZmMRP-1 Expression Mediate Polar Domain Formation?

Analyses of the ZmMRP-1 expression pattern (Figures 3and 4) show a close association between the process of transfer cell development and the expression of this gene. Indeed, ZmMRP-1 can regulate the expression of two different transfer cell–specific genes (Figures 7and 8). These findings suggest a role for ZmMRP-1 early in transfer cell differentiation, its site of expression reflecting a polar signal diffusing from maternal cells. ZmMRP-1 expression, however, could reflect “intrinsic” generation of polarity in the developing endosperm, as seen with maize endosperms generated in vitro (Kranz et al., 1998). It would be interesting to determine whether these in vitro–produced endosperms express ZmMRP-1 and, if so, where.

Polarity in the distribution of transcriptional regulators is thought to be involved in endosperm production (Olsen et al., 1999) and plays a role in plant embryo development (Souter and Lindsey, 2000). Positional effects on endosperm and aleurone gene expression have been demonstrated elegantly by Becraft and Asuncion-Crabb (2000), who indicated the requirement for an active mechanism to maintain the differentiated state. This would be consistent with the presence of the ZmMRP-1 transcript long after cell divisions have ceased and transfer cell determination has occurred. The existence of diffusible morphogens acting in seed development has been postulated to explain developmental gradients in the seed (Soave and Salamini, 1984; Kowles and Phillips, 1988). Davis et al. (1990)suggested the involvement of a diffusible substance in determining transfer cell phenotype to explain the transition, over four cell layers, from cells with dense wall ingrowths to cells with smooth walls. A gradual transition also is observed in the cell phenotypes at the border between transfer cells and aleurone. We propose that ZmMRP-1 is a molecular transducer of a maternally derived signal that mediates transfer cell differentiation. To test this hypothesis, experiments are under way to analyze the effect of the elimination of ZmMRP-1 function in the developing endosperm.

METHODS

Plant Material

Maize (Zea mays) line A69Y shoots, leaves, immature seed, silk, tassels, roots, and germinating kernels were extracted either from greenhouse-grown plants or from seedlings germinated on moist filter paper in Petri dishes. DNA from the parental lines and recombinant inbred lines (15 individuals per line) was extracted from young leaves.

Isolation of RNA and DNA and Analysis by Filter Hybridization

DNA and RNA were manipulated according to Hueros et al. (1995)and Maniatis et al. (1982).

Inverse Polymerase Chain Reaction

To clone the genomic sequences immediately upstream and downstream of the coding region contained in the cDNA clone, a physical map of the genomic sequences containing the gene was constructed in the maize inbred line A69Y. Using this map, we selected a restriction enzyme that generated a genomic DNA fragment of <3 kb containing the coding sequence as well as the flanking regions. Genomic DNA (100 μg) was digested with NsiI for upstream promoter isolation and with DraI for downstream genomic sequence isolation. The size fraction containing the gene of interest with flanking sequences was purified by excision from an agarose gel. Recovered DNA was resuspended in water. One-tenth was used for a self-ligation reaction in 50 μL. Self-ligated DNA was precipitated and resuspended in 10 μL of water. One microliter was used for polymerase chain reaction (PCR) performed with forward and reverse oligonucleotides derived from the cDNA sequence, with a Tm (melting temperature) of ∼65°C. After 35 cycles of PCR, amplified bands were gel purified and cloned in pBluescript SK− for sequencing. Clones from three independent PCR procedures were sequenced to exclude PCR artifacts.

In Situ Hybridization

Maize seed were fixed in 4% paraformaldehyde and 0.1% glutaraldehyde in 0.1 M sodium phosphate buffer, pH 7.2, for 12 to 24 hr depending on the tissue volume. Samples were dehydrated and embedded in Fibrowax (Plano GmbH, Marburg, Germany). Eight- to 10-μm sections were hybridized essentially as described by Cox and Goldberg (1988). ³⁵S-labeled RNA was synthesized from a pBluescript SK− derivative template using an in vitro transcription kit (Roche Diagnostics, Mannheim, Germany). Slides were exposed for 2 to 4 days at 4°C using LM-1 silver grain emulsion (Amersham). After developing, sections were stained with calcofluor (0.01%) for 15 min and counterstained with water for 3 × 10 min each. In some cases, sections also were stained with 4′,6-diamidino-2-phenylindole (DAPI) (2 μg/μL) for 10 min to visualize the nuclei.

Construction of Plasmids for the Transactivation Experiments

Reporter plasmids for protoplast transformation were prepared by PCR amplification of the corresponding promoter (ZmMRP-1 promoter from this study, BETL-1, -2, -3, and -4 promoters from Hueros et al., 1999a, 1999b), which was fused transcriptionally to the uid gene. The downstream PCR primer was designed to introduce a NcoI restriction site containing the translation start codon of the corresponding gene. In this way, the ATG codon of β-glucuronidase (GUS) replaces that of the endogenous ATG. The effector plasmid pUBI-MRP was obtained by introducing the coding sequence of ZmMRP-1 (amplified with PCR primers containing added restriction sites) between the maize ubiquitin promoter, including the first intron of the gene (Christensen et al., 1992), and the nopaline synthase (NOS) terminator.

Construction of Plasmids for the Green Fluorescent Protein Localization Experiments

The plasmid pGFP-JS (kindly provided by Dr. J. Sheen, Massachusetts General Hospital, Boston), containing the green fluorescent protein (GFP) coding sequence under the control of a double 35S promoter, was used as a control to analyze the cellular localization of GFP. pZmMRP1-GFP was generated by inserting the complete ZmMRP-1 coding sequence in the vector described above, generating a ZmMRP-1 N-terminal translational fusion to GFP.

Tobacco Protoplast Transformation

Protoplasts from young plantlets (<2 weeks) of tobacco (Nicotiana tabacum cv SR1) were prepared and transformed according to Negrutiu et al. (1987). Aliquots (300 μL) containing 0.5 to 1 million cells were transformed with a mixture of plasmid DNA (5 μg of each construct) and carrier DNA (50 μg). After polyethylene glycol–mediated transformation, cells were incubated in the dark for 48 hr. Protoplasts were washed and divided into two aliquots that were stored at −80°C until assayed for GUS and luciferase activities, respectively. In the case of GFP experiments, cells were visualized directly by fluorescence microscopy.

Particle Bombardment

Transgenic maize leaves (2 cm wide) were surface sterilized, sectioned into 1- to 2-cm-long pieces, and maintained on the solid medium described below until bombardment with DNA-coated gold particles. Coating and bombardment were performed according to Knudsen and Müller (1991). Five micrograms of each plasmid for transformation was used to prepare a gold particle batch that was then used to bombard seven leaf fragments. After bombardment, tissue samples were incubated at 25°C for 24 hr in the dark on solid (0.5% agarose) Murashige and Skoog (1962)medium containing 100 mg/L myo-inositol, 2 g/L Asp, 2 g/L Gln, 30 g/L Suc, and Murashige and Skoog vitamins (Sigma).

GUS Assays

Histochemical detection of GUS expression was performed by staining according to Jefferson et al. (1987). Leaf pieces were stained for GUS in a medium containing 0.5 mg/mL X-glucuronide (Clontech, Palo Alto, CA), 0.5 mM K⁺-ferrocyanide, 0.5 mM K⁺-ferricyanide, 10 mM Na₂EDTA, 50 mM phosphate buffer, pH 7, 0.1% Triton X-100, and 20% (v/v) methanol.

For quantification, protoplasts were lysed in GUS extraction buffer. GUS activity was assayed fluorimetrically in the supernatant as described by Jefferson et al. (1987). As an internal control, aliquots from each experiment were lysed in luciferase extraction buffer (GeneLux kit; Wallac, PerkinElmer Life Sciences Inc., Finland) and light emission was measured in a Wallac Victor² 1420 instrument. Relative GUS/luciferase activities were measured for a minimum of three independent transformation experiments for each construct.

Accession Numbers

The accession numbers for the genes and proteins described in this article are AJ318518 (ZmMRP-1), AJ318519 (2949-bp contig of ZmMRP-1), GI:12406993 and GI:12406995 (MCB1 and MCB2), GI:6688529 (LeMYB1), GI:7705206 (MybSt1), GI:3281846 (LHY), GI:7484928 (CCA1), and GI:127585 (maize R2-R3 MYB gene C1).

ACKNOWLEDGMENTS

The immortalized maize F2 lines used for mapping the ZmMRP-1 locus were kindly provided by Dr. Ed H. Coe. This work was supported by European Union Grant No. BIO4-CT97-2158 and by Spanish Ministerio de Ciencia y Tecnologia Grant No. BIO2000-0848.

Author notes

1

To whom correspondence should be addressed. E-mail ; fax [phone]

Article, publication date, and citation information can be found at www.plantcell.org/cgi/doi/10.1105/tpc.010365.