The maize floury1 gene encodes a novel endoplasmic reticulum protein involved in zein protein body formation

Abstract

The maize (Zea mays) floury1 (fl1) mutant was first reported almost 100 years ago, but its molecular identity has remained unknown. We report the cloning of Fl1, which encodes a novel zein protein body membrane protein with three predicted transmembrane domains and a C-terminal plant-specific domain of unknown function (DUF593). In wild-type endosperm, the FL1 protein accumulates at a high level during the period of zein synthesis and protein body development and declines to a low level at kernel maturity. Immunogold labeling showed that FL1 resides in the endoplasmic reticulum surrounding the protein body. Zein protein bodies in fl1 mutants are of normal size, shape, and abundance. However, mutant protein bodies ectopically accumulate 22-kD alpha-zeins in the gamma-zein-rich periphery and center of the core, rather than their normal discrete location in a ring at outer edge of the core. The 19-kD alpha-zein is uniformly distributed throughout the core in wild-type protein bodies, and this distribution is unaffected in fl1 mutants. Pairwise yeast two-hybrid experiments showed that FL1 DUF593 interacts with the 22-kD alpha-zein. Results of these studies suggest that FL1 participates in protein body formation by facilitating the localization of 22-kD alpha-zein and that this is essential for the formation of vitreous endosperm.

Figure 1.

Phenotype and Gene Structure of the fl1-Mu and fl1-ref Mutations. (A) Light transmission by mature kernels. Kernels were randomly selected from six ears of W64A+ (top), W64A fl1-Mu1 (mto222-1) (middle), and W64A fl1-ref (bottom) and viewed on a light box. (B) Kernel phenotypes of the wild type, fl1-Mu1 through fl1-Mu7 (1 to 7), and fl1-ref (ref). Kernel crowns were ground to reveal the thickness of the vitreous endosperm layer. (C) Diagram of the Fl1 gene. The solid box shows the coding region, and numbered large arrowheads indicate position of Mu insertion alleles. The gray box shows the relative region coding for DUF593. The small arrowhead shows relative position of the fl1-ref point mutation (ref). (D) FL1 amino acid sequence showing the predicted transmembrane regions (gray boxes) and the conserved DUF593 domain (underlined). The Ser-to-Gly amino acid substitution in fl1-ref is marked with an arrowhead.

Figure 2.

Kernel Phenotypes Resulting from Allelism Test Crosses between mto222-1 (Subsequently Referred to as fl1-Mu1) and fl1-ref Mutants. The presence of starchy or vitreous endosperm was shown after removal of kernel crowns by grinding.

Figure 3.

Phylogenetic Tree of DUF593 Domain Proteins. Maize FL1 and all currently recognized DUF593 domain proteins from Arabidopsis and rice were aligned by ClustalW; the aligned sequences were trimmed to ∼90–amino acid residues covering the DUF593 domain (see Supplemental Table 1 online). The tree was constructed using PHYLIP version 3.6 (see Methods). Distances were estimated with a neighbor-joining algorithm, and bootstrap support is indicated to the left of branches. The accession number for proteins is preceded by the species: Zm (Zea mays), At (Arabidopsis thaliana), or Os (Oryza sativa). In the case of the two rice sequences labeled with the gene name rather than accession number, protein sequences had been incorrectly predicted from the genomic sequence (public databases). In these cases, the protein sequences were curated from available maize cDNA sequences (BLAST searches against Pioneer database, and resolution was confirmed by BLAST comparisons between rice genomic sequence and maize cDNA sequence). The different general structural classes of DUF593 domain proteins are shown on the right and below the tree. FL1 belongs to a small branch with apparent single member orthologous sequences in rice and Arabidopsis (shown boxed) that contain two or three transmembrane sequences in the N-terminal half and the DUF593 domain in the C-terminal half of the protein. In the key, black and gray boxes indicate predicted highly probable and likely transmembrane sequences, respectively; boxed “sp” indicates a predicted ER signal sequence.

Figure 4.

Comparison of Zein and Bip Accumulation in fl1-Mu1 and fl1-ref Endosperm. (A) SDS-PAGE analysis of the zein fraction from 18 and 24 DAP and mature endosperm of W64A+, W64A fl1-Mu1, and W64A fl1-ref (lanes 1 to 3, respectively). To compare protein accumulation between fresh, developing, and mature dry kernels, each lane was loaded with 1/200 of the extract from one whole endosperm. (B) Immunoblot comparing Bip accumulation in ER and protein body fractions of W64A+, W64A fl1-Mu1, and W64A fl1-ref (lanes 1 to 3, respectively). Extract was loaded from ∼0.2 mg (fresh weight) of endosperm. Coomassie blue–stained gel compares sample loading (10-fold overloaded compared with immunoblot). ER, endoplasmic reticulum fraction; PB, protein body fraction.

Figure 5.

Fl1 Gene Expression and Protein Accumulation in Opaque Mutants. (A) Semiquantitative RT-PCR analysis of Fl1 transcript in W64A+, W64A fl1-Mu1, and W64Afl1-ref endosperm. Numbers refer to DAP. Bottom panel shows expression of the RRb1 gene as a control. (B) Immunoblot comparing accumulation of FL1 protein in endosperms of various opaque mutants. (C) Immunoblot comparing accumulation of FL1 protein in endosperms of W64A+, W64A fl1-Mu1, and W64A fl1-ref. The left panel shows a standard 2-min exposure following the chemiluminescent detection reaction, whereas the right panel shows the same blot overexposed for 30 min. In (B) and (C), each lane was loaded with non-zein protein extracts from 5 mg (fresh weight) of endosperm; arrowheads mark the predicted size of FL1 (32.6 kD).

Figure 6.

Spatial and Temporal Expression of the FL1 Protein. (A) Immunoblot showing the presence of FL1 in total protein extracts from various maize tissues (200 μg of protein per lane). (B) Immunoblot showing time course of FL1 protein accumulation. To compare FL1 levels between fresh, developing, and dry, mature kernels, each lane was loaded with 1/200 of the extract from one whole endosperm. (C) Immunoblots showing FL1 accumulation in endosperm subcellular fractions: soluble, ER membrane fraction (ER), and protein body membrane fraction (PB). Extract from 2 mg of fresh weight endosperm was loaded. In (A) to (C), the + and – refer to W64A+ and W64A fl1-Mu1, respectively, and arrowheads represent the predicted size of FL1 (32.6 kD).

Figure 7.

Localization of FL1 in the Second Subaleurone Layer Endosperm Cells (22 DAP). (A) Immunolocalization of FL1 in wild-type high-pressure frozen/freeze-substituted endosperm samples. (B) Double labeling of FL1 (15-nm gold particles) and the ER protein zmTIP3-4 (5-nm gold particles). The white arrowheads indicate FL1 gold labeling on the protein body ER membrane, whereas the black arrowheads point out the zmTIP3-4 labeling on both protein body ER and cisternal ER (cER). Note that no FL1 labeling was detected on cisternal ER membranes. Bars = 500 nm.

Figure 8.

Immunolocalization of 22-kD α-Zein and 27-kD γ-Zein in Wild-Type, fl1-Mu1, and fl1-ref Mutant Endosperm Cells (22 DAP). The double labeling of the 22-kD α- and 27-kD γ-zeins was performed using anti-22-kD α-zein polyclonal antibodies raised in rabbit and anti-27-kD γ-zein polyclonal antibodies raised in chicken. For gold particle quantification, the distance from the protein body surface to the center of the gold particle was measured using ImageJ. Arrows indicate labeling on or very close to the protein body membrane. Arrowheads in (C), (F), and (I) indicate the area of maximal 22-kD α-zein signal in wild-type protein bodies. 103 gold particles in 41 protein bodies were counted in the labeled wild-type samples, 104 gold particles in 30 protein bodies in the fl1-Mu1 samples, and 78 gold particles in 28 protein bodies in the fl1-ref samples. Bars = 200 nm.

Figure 9.

Immunolocalization of 19-kD α-Zein and 27-kD γ-Zein in Wild-Type, fl1-Mu1, and fl1-ref Mutant Endosperm Cells (18 DAP). The double labeling of the 19-kD α- and 27-kD γ-zeins was performed using anti-19-kD α-zein polyclonal antibodies raised in rabbit and anti-27-kD γ-zein polyclonal antibodies raised in chicken. For gold particle quantification, the distance from the protein body surface to the center of the gold particle was measured using ImageJ. 273 gold particles in 12 protein bodies were counted in the labeled wild-type samples, 263 gold particles in 24 protein bodies were counted in the fl1-Mu1 samples, and 260 gold particles in 20 protein bodies were counted in the fl1-ref samples. Bars = 200 nm.

Figure 10.

Yeast Two-Hybrid Interactions between FL1 and Different Types of Zeins. Six representative colonies of pGADT7-zein × pGBKT7-DUF593 diploid strains were streaked on -Leu/-Trp/-Ade/-His dropout media. Control two-hybrid matings were performed as suggested in the BD Matchmaker library construction and screening manual. Diploid colonies from the positive interaction control mating (+) (pGADT7-RecT × pGBKT7-53) and from the noninteraction control mating (-) (pGADT7-RecT × pGBKT7-Lam) were also streaked.