This Article Abstract Full Text (PDF) All Versions of this Article: 14/3/655    most recent tpc.010431v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via ISI Web of Science (19) Citing Articles via Google Scholar Google Scholar Articles by Kim, C. S. Articles by Larkins, B. A. Search for Related Content PubMed PubMed Citation Articles by Kim, C. S. Articles by Larkins, B. A. Agricola Articles by Kim, C. S. Articles by Larkins, B. A.

Zein Protein Interactions, Rather Than the Asymmetric Distribution of Zein mRNAs on Endoplasmic Reticulum Membranes, Influence Protein Body Formation in Maize Endosperm

Department of Plant Sciences, University of Arizona, Tucson, Arizona 85721

¹ To whom correspondence should be addressed. E-mail larkins{at}ag.arizona.edu; fax 520-621-3692

	Abstract

TOP Abstract INTRODUCTION RESULTS DISCUSSION METHODS References

 
Prolamin-containing protein bodies in maize endosperm are composed of four different polypeptides, the -, -, -, and -zeins. The spatial organization of zeins within the protein body, as well as interactions between them, suggests that the localized synthesis of -zeins could initiate and target protein body formation at specific regions of the rough endoplasmic reticulum. To investigate this possibility, we analyzed the distribution of mRNAs encoding the 22-kD -zein and the 27-kD -zein proteins on cisternal and protein body rough endoplasmic reticulum membranes. In situ hybridization revealed similar frequencies of the mRNAs in both regions of the endoplasmic reticulum, indicating that the transcripts are distributed more or less randomly. This finding implies that zein protein interactions determine protein body assembly. To address this question, we expressed cDNAs encoding -, -, -, and -zeins in the yeast two-hybrid system. We found strong interactions among the 50-, 27-, and 16-kD -zeins and the 15-kD -zein, consistent with their colocalization in developing protein bodies. Interactions between the 19- and 22-kD -zeins were relatively weak, although each of them interacted strongly with the 10-kD -zein. Strong interactions were detected between the - and -zeins and the 16-kD -zein and the 15-kD -zein; however, the 50- and 27-kD -zeins did not interact with the - and -zein proteins. We identified domains within the 22-kD -zein that bound preferentially the - and -zeins and the - and -zeins. Affinities between zeins generally were consistent with results from immunolocalization experiments, suggesting an important role for the 16-kD -zein and the 15-kD -zein in the binding and assembly of -zeins within the protein body.

	INTRODUCTION

TOP Abstract INTRODUCTION RESULTS DISCUSSION METHODS References

 
The mechanisms by which cells organize proteins in their cytoplasm and organelles have been studied actively in recent years. It is known that unique peptides (signal/transit/targeting) within proteins direct their movement either cotranslationally or post-translationally into the secretory system (endoplasmic reticulum, Golgi, vacuoles), double membrane organelles (plastids, mitochondria), and the nucleus (Bar-Peled et al., 1996; Keegstra and Cline, 1999; Vitale and Denecke, 1999). Another emerging paradigm is the role played by the cytoskeleton in mediating mRNA targeting to specific regions of the cell, where translation brings about protein localization. This is an efficient mechanism that allows both soluble and secreted proteins to be made proximal to where they are needed, and it also provides a mechanism to separate proteins that otherwise might interact inappropriately (Rings et al., 1994). Evidence of mRNA sorting was obtained first from studies of embryogenesis in Xenopus and Drosophila, in which the asymmetric distribution of mRNAs in the oocyte and early embryo, respectively, was shown to be essential for the development of polarity (St. Johnston, 1995).

The mechanisms of mRNA sorting have not been investigated widely in plants, although Okita and co-workers described evidence for this phenomenon in developing rice endosperm (Li et al., 1993a). Rice seed contain two types of storage proteins: prolamins (oryzins), which form accretions directly within the lumen of the rough endoplasmic reticulum (ER), and glutelins, which are synthesized on rough ER membranes but are transported to the vacuole, where they form protein bodies. Li et al. (1993a) reported that although both mRNAs are found in membrane-bound polysomes, oryzin mRNAs are localized preferentially to the ER surrounding prolamin-containing protein bodies and glutelin mRNA is associated predominantly with polysomes on the cisternal ER. The mechanism(s) responsible for the asymmetric distribution of these two types of mRNAs is unknown, but recent studies indicate a role for the mRNA 3' noncoding sequences (Choi et al., 2000).

Studies of storage protein synthesis in maize endosperm suggest that the cytoskeleton also plays a role in targeting maize prolamin (zein) mRNAs to the ER. Zein protein bodies form in the lumen of the ER throughout these large (100 to 200 µm diameter), nearly isodiametric cells (Lopes and Larkins, 1993). Consequently, there must be mechanisms that transport zein mRNAs from the nucleus to distant sites on the ER, organize zeins within the protein body, and limit protein body enlargement. Abe et al. (1991) and Stankovic et al. (1993) discovered associations between actin filaments, polysomes, and protein bodies after the homogenization of developing maize endosperm in a cytoskeleton-stabilizing buffer. Other studies examining cytoskeleton organization in developing endosperm showed that actin occurs as fine filaments in the cortex and cytoplasm as well as in clusters between starch grains (Clore et al., 1996). Not only does the distribution of actin closely match the location of protein bodies, but actin also surrounds protein bodies. Some microtubules in these cells are found in a multidirectional array between the starch grains in close apposition to protein bodies. It also was found that eEF1a is localized around protein bodies, where it is merged into a complex with actin. However, the function and mechanism of zein mRNA cytoskeleton associations remain to be elucidated.

The mechanisms by which zeins assemble into protein bodies are poorly understood. Zein protein bodies are composed of three structurally distinct types of proteins: -zein, -zein (which includes -zein), and -zein (Larkins et al., 1989). -Zeins, which typically are the most abundant, contain 40 N-terminal amino acids that precede a series of nine or 10 repeated peptides of 20 amino acids. These repeats are predicted to be -helical and wind the protein into a rod-shaped molecule (Argos et al., 1982; Garratt et al., 1993). -Zein, which is related to the -zeins (Woo et al., 2001), contains no repetitive peptides and appears to consist mostly of -sheet and turn conformation (Pedersen et al., 1986). Each of the -zeins has a unique N-terminal sequence (Prat et al., 1985, 1987; Woo et al., 2001). In the 50-kD -zein, this region is 136 amino acids in length and is very His rich (18%). The 27-kD -zein protein has a series of eight tandem hexapeptide repeats (PPPVHL) that occur 11 amino acids after the N terminus. This Pro-rich sequence appears to function as an ER-retention mechanism (Geli et al., 1994). The first eight amino acids of the 16-kD -zein protein are identical to those of the 27-kD -zein, but it has three degenerate versions of the Pro-rich repeat (PPPF/HH/YM/L). Approximately the last 140 amino acids of the - and -zeins are 85% identical. The -zeins, which are the most hydrophobic proteins of the group, contain no repetitive peptides and are exceptionally rich in Met (23%) and Cys (4%) (Kirihara et al., 1988; Chui and Falco, 1995).

The pattern of temporal and spatial accumulation of zeins in protein bodies implies that specific interactions between them are important for their association (Lending and Larkins, 1989; Esen and Stetler, 1992). The smallest protein bodies consist of aggregates of - and -zeins. The - and -zeins subsequently penetrate this network, filling the interior of the protein body and expanding it until it reaches a diameter of 1 to 2 µm. Immunogold labeling showed that the - and -zeins are concentrated toward the periphery of the protein body, although they also are detected in the interior. This sequential pattern of zein protein accumulation is consistent with the temporal and spatial distribution of their mRNAs in endosperm cells (Woo et al., 2001).

Experiments with transgenic plants also support the hypothesis that specific interactions between zeins influence protein body assembly. Studies in which genes encoding one or more types of zein proteins were expressed in developing endosperm (Coleman et al., 1996) or other tissues of transgenic tobacco plants (Bagga et al., 1997) provided evidence that - and -zeins interact with - and -zeins, promoting their retention in the ER and their incorporation into protein bodies. When synthesized individually, - and -zeins formed protein accretions that were retained within the ER and that appeared to be reasonably stable over prolonged periods of time. However, when - and -zeins were synthesized alone, they were not retained in the ER and appeared to become degraded. The PPPVHL repeats at the N terminus of the 27-kD -zein protein could provide the mechanism for its retention in the ER (Geli et al., 1994). It was suggested that this sequence can form an amphipathic helix that interacts with the surface of the ER (Rabanal et al., 1993). Perhaps this Pro-rich sequence nucleates protein body formation, leading to interactions between -zeins via their conserved C-terminal regions. These proteins then could bind and retain the - and -zeins, leading to their accumulation in the protein body.

Because zein protein bodies enlarge to form uniform spherical protein accretions in what appear to be localized regions of the ER, a hypothesis that explains the previous observations describing protein body formation in maize endosperm involves targeting of these mRNAs, or at least the -zein mRNAs, to distinct regions of the rough ER. To investigate the importance of mRNA targeting in zein protein body formation, we characterized the subcellular distribution of zein mRNAs in developing endosperm. On the basis of in situ hybridization, we did not find differences in the occurrence of the 27-kD -zein and the 22-kD -zein mRNAs on cisternal and protein body ER. Thus, although zein mRNA trafficking to the ER is possible, the mRNAs do not seem to be targeted asymmetrically to domains of the cisternal and protein body ER.

To examine the interactions between zein proteins and their role in protein body assembly, cDNAs encoding the various types of zeins were expressed in the yeast two-hybrid system. These experiments revealed strong affinities between - and -zeins and relatively weak interactions between -zeins. Surprisingly, there were stronger interactions between the 15-kD -zein and the 16-kD -zein and the 19- and 22-kD -zeins than between the -zeins themselves. By expressing deletion mutants of a 22-kD -zein in the yeast two-hybrid system, we were able to define domains within this protein that bind with other types of zein proteins. Protein bodies were formed in yeast cells by constitutively overproducing zeins as native proteins or green fluorescent protein fusions. Yeast cells synthesizing these proteins grew slower than did controls, but they appeared to accumulate zeins in the ER, where they formed small accretions of varying sizes. Because of the small amount of zeins in these accretions, we were unable to evaluate their structural organization.

	RESULTS

TOP Abstract INTRODUCTION RESULTS DISCUSSION METHODS References

 
In Situ Hybridization Reveals a Symmetric Distribution of 27-kD -Zein and 22-kD -Zein mRNAs on ER Membranes
To visualize the distributions of 27-kD -zein and 22-kD -zein mRNAs within maize endosperm cells, high-resolution in situ hybridization experiments were conducted. When sections were incubated with either 27-kD -zein or 22-kD -zein antisense probes, transcripts (seen as gold particles) were found on the ER surrounding protein bodies and on the cisternal ER connecting protein bodies (Figures 1A and 1B, circles). Typically, the use of antisense probes resulted in strands or clusters of gold particles. Such arrays have been reported in other high-resolution in situ hybridization studies (Singer et al., 1989; Pomeroy et al., 1991) and are thought to represent complexes of antibodies bound to probe-endogenous mRNA hybrids. In contrast, when sections were incubated with 27-kD -zein and 22-kD -zein sense probes (Figures 1C and 1D, respectively), gold particles (typically as singlets) were seen occasionally, usually in the cytoplasm. Therefore, the antisense signal but not the sense signal seemed to be the result of the specific binding of probes to endogenous mRNA molecules.

View larger version (150K): [in this window] [in a new window]   Figure 1. Localization of the 27-kD -Zein and the 22-kD -Zein mRNAs in Developing Maize Endosperm. (A) Electron micrograph of an endosperm section hybridized with the 27-kD -zein antisense probe. (B) Electron micrograph of an endosperm section hybridized with the 22-kD -zein antisense probe. (C) Electron micrograph of an endosperm section hybridized with the 27-kD -zein sense probe. (D) Electron micrograph of an endosperm section hybridized with the 22-kD -zein sense probe. Circles are around colloidal gold particles, which appear in clusters or strings in the antisense-treated samples and usually as singlet particles in the sense-treated samples. In samples hybridized with antisense probes, label was found primarily around protein bodies (PB) or on the cisternal (cis) ER, whereas in the sense-treated samples, label was found at random locations in the cell. These observations were confirmed based on the statistical analyses in Figure 2. Bars = 1.0 µm.

View larger version (34K): [in this window] [in a new window]   Figure 2. Distribution of Colloidal Gold Particles per Micrometer of Protein Body versus Cisternal ER on Endosperm Sections Treated with 22-kD -Zein and 27-kD -Zein Antisense and Sense Probes (as Illustrated in Figure 1). The method for counting gold particles is described in Methods; error bars indicate standard error. cis, cisternal; PB, protein body.

The mean density of gold particles on total ER membranes was approximately seven-fold higher for the 27-kD -zein antisense probes than for the sense probes and 17-fold higher for the 22-kD -zein antisense probes than for the sense probes (Figure 2). The distribution of label on the two different types of ER (protein body and cisternal) also was analyzed (Figure 2). The density (i.e., number per micrometer) of gold particles was 0.138 ± 0.014 on the protein body ER of samples treated with 27-kD -zein antisense probes and 0.194 ± 0.114 for samples treated with 22-kD -zein antisense probes. The density of gold particles was 0.110 ± 0.059 on the cisternal ER of samples treated with 27-kD -zein antisense probes and 0.210 ± 0.116 for samples treated with 22-kD -zein antisense probes. Therefore, taking into account standard error values, the extent of labeling on the protein body ER was similar to that on the cisternal ER for both 27-kD -zein and 22-kD -zein antisense probes. Most importantly, there was no asymmetry in the distribution of 27-kD -zein versus 22-kD -zein mRNAs, because both were found in approximately equivalent amounts on protein body as well as cisternal ER.

Analysis of Zein Protein Interactions with the Yeast Two-Hybrid System
Interactions between the different types of zein proteins appear to mediate their association into protein bodies (Coleman et al., 1996; Bagga et al., 1997). To investigate the nature of these interactions and the roles of specific zeins in protein body formation, we expressed zein coding sequences, after removal of their signal peptides, in yeast two-hybrid vectors (Bai and Elledge, 1996). The different types of zein genes were expressed pairwise in both activation domain (pACT2) and DNA binding domain (pAS2) plasmids, and the strength of the interaction was evaluated by the intensity of the X-Gal color reaction (Figure 3A) and, more sensitively, by growth in medium containing 3-aminotriazole (Figure 3B). The 50-kD -zein interacted strongly with itself and the 27- and 16-kD -zeins in both the activation and DNA binding domain plasmids. It showed very weak or no interaction with the 22- and 19-kD -zeins and the 15-kD -zein, although there was weak interaction with the 10-kD -zein when it was in the activation domain vector. The 27-kD -zein interacted more strongly with the 50- and 16-kD -zeins than it did with itself; these interactions were strongest when it was in the activation domain plasmid. The 27-kD -zein had a very weak interaction with the 19-kD -zein, and this was true as well of the reciprocal relationship.

View larger version (30K): [in this window] [in a new window]   Figure 3. Assay of Zein Interactions with the Yeast Two-Hybrid System. After removal of the signal peptide sequence, zein coding regions were subcloned into the pAS2 and pACT2 plasmids of the yeast two-hybrid system and transformed reciprocally into yeast cells. (A) Yeast cells inoculated on filters and grown with X-Gal overnight. (B) Yeast grown on plates containing 3-aminotriazole for 6 days. Strong (s) indicates that colonies expressing the two indicated zeins had a positive reaction in the X-Gal filter lift assay and scored positively for growth on medium containing 3-aminotriazole compared with negative controls. Weak (w) indicates that colonies expressing the indicated zeins showed a slight color reaction in the X-Gal assay but grew slowly on selection medium containing 3-aminotriazole compared with controls. (-) indicates that no X-Gal reaction was detected.

During protein body assembly, -zeins appear to penetrate a network of -zeins and then associate, along with the small amount of -zeins, and fill the interior of the protein body (Lending and Larkins, 1989; Esen and Stetler, 1992). Because little is known about the interactions between - and -zeins, we used the yeast two-hybrid system to examine peptide domains within a 22-kD -zein that interact with other types of zein proteins. For these experiments, a construct encoding a 22-kD -zein with a C-terminal GFP fusion was used so that the proteins could be visualized subsequently in transformed yeast cells. We expressed the 22-kD -zein::GFP protein in the pACT2 plasmid, and the other zein proteins were expressed in the pAS2 vector. As shown in Figure 4 (construct a), the 22-kD -zein::GFP fusion protein interacted strongly with the 22-kD -zein, 15-kD -zein, 10-kD -zein, and 16-kD -zein proteins. These results were similar to those obtained with the mature zein proteins (Figure 3) and demonstrated that the GFP fusion did not interfere with the interactions between zeins leading to an active GAL4 transcription factor. Furthermore, no -galactosidase activity was detected in yeast cells with a pACT2 construct expressing GFP alone and a pAS2 construct expressing zein proteins (Figures 4A and 4B, construct i).

View larger version (35K): [in this window] [in a new window]   Figure 4. Assay of Yeast Two-Hybrid Interactions between 22-kD -Zein Deletion Mutants and Other Types of Zein Proteins. (A) Construction of the 22-kD -zein deletion mutants in the pACT2 plasmid. White boxes correspond to the N-terminal leader following the signal peptide; dotted, hatched, and cross-hatched boxes (1 to 9) correspond to the 20–amino acid repeated peptides; the C-terminal 17 amino acids (17 aa) are indicated by solid gray boxes; and the C-terminal GFP fusion is shown by black boxes. (B) The nature of the yeast two-hybrid interaction between 22-kD -zein deletion mutants and zein proteins. Strong (s) indicates that colonies expressing the two indicated zeins showed a positive reaction in the X-Gal filter lift assay and scored positively for growth on medium containing 3-aminotriazole compared with negative controls. Weak (w) indicates that colonies expressing the indicated zeins showed a slight color reaction in the X-Gal assay but grew slowly on selective medium containing 3-aminotriazole compared with controls. (-) indicates that no X-Gal reaction was detected.

To further define the regions within the 22-kD -zein that bind with the other zein proteins, an additional series of deletion mutants was constructed with a variable number of repeated peptides with or without the N- and C-terminal leader sequences. Construct k in Figure 4 contains the N-terminal leader, the last three peptide repeats, and the C terminus. This protein interacted strongly with the 16-kD -zein and the 15-kD -zein but interacted weakly with the 22-kD -zein and the 10-kD -zein. If the N- and C-terminal peptides were removed from construct k (Figures 4A and 4B, construct l), there was no interaction with any of the zein proteins, again suggesting the importance of the N- and C-terminal peptide sequences. Because the last two repeated peptides plus the C-terminal sequence (Figures 4A and 4B, construct f) promoted a weak interaction with other zein proteins, an additional construct was made in which the last 18 amino acids of the C terminus were deleted (Figure 4, construct m). Like construct f, construct m also had a weak interaction with these proteins, suggesting that the first half of the C terminus is functionally important for zein interactions.

To examine the interactions of the N-terminal sequence of the 22-kD -zein, the last three peptide repeats plus the C terminus were eliminated (Figure 4, construct n), because this region appeared to be sufficient to promote a strong yeast two-hybrid interaction with multiple types of zein proteins. Surprisingly, this construct gave a strong -galactosidase reaction with each of the other zeins. Therefore, we made constructs consisting of only the N terminus plus two peptide repeats or just the first two peptide repeats alone (Figures 4A and 4B, constructs o and p). With construct o, there was a strong interaction with the 22-kD -zein and the 10-kD -zein and a weak interaction with 16-kD -zein and the 15-kD -zein. However, no interaction was apparent when only the first two repeated peptides were expressed, indicating that the N terminus must play an important role in these interactions.

On the basis of the analyses of the 22-kD -zein mutants, it appeared that amino acid sequences within the N- and C-terminal regions, as well as a small number of repeated peptides, promote interactions with the 22-kD -zein, 15-kD -zein, 16-kD -zein, and 10-kD -zein proteins. Because constructs containing the last two repeated peptides, all or part of the C terminus (Figures 4A and 4B, constructs f and m) and the N terminus, and the first two repeated peptides gave strong or at least weak interactions with this subset of zein proteins, two additional constructs were tested. The first contained the N terminus, the first two peptide repeats, the last two peptide repeats, and the first 17 amino acids of the C terminus (Figures 4A and 4B, construct q). This protein interacted strongly with these zein proteins, consistent with the other experiments showing the importance of the N- and C-terminal sequences plus a small number of peptide repeats for promoting interactions with other zein proteins. With a view to using this truncated version of the 22-kD -zein as a vehicle to produce fusion proteins in transgenic plants, we inserted GFP between the first two and last two repeated peptides of construct q to evaluate how the hybrid protein (Figures 4A and 4B, construct r) would interact with these proteins. As was true of construct q, this protein interacted very strongly with this group of zein proteins based on growth in 3-aminotriazole and the -galactosidase filter assay.

Because the -galactosidase filter assay does not provide a quantitative measure of the two-hybrid interaction between zein proteins, a second set of experiments was performed in which -galactosidase activity was assayed in lysates of yeast cells expressing combinations of zein proteins. Figure 5A shows the -galactosidase activity of yeast cells expressing the full-length 22-kD -zein in the pACT2 plasmid and the other types of zeins in the pAS2 plasmid. For this assay, a 300-µL aliquot of an overnight culture was inoculated in fresh medium, and the yeast cells were harvested 10 hr later when the cultures reached an OD₆₀₀ of 0.8. After cell lysis, -galactosidase was assayed spectrophotometrically using o-nitrophenyl -D-galactopyranoside (ONPG) as a substrate. Fundamentally, the results of this analysis were consistent with those obtained from the filter assay, although the measurement of -galactosidase activity was much more sensitive and quantitative. The 22-kD -zein interacted most strongly with the 16-kD -zein and the 15-kD -zein. It had a stronger interaction with the 10-kD -zein than it did with itself or the 19-kD -zein. Interactions with the 50- and 27-kD -zeins were barely detectable with this assay.

View larger version (57K): [in this window] [in a new window]   Figure 5. Assay of -Galactosidase Activity Resulting from Yeast Two-Hybrid Interactions between Selected 22-kD -Zein Deletion Mutants and Other Zein Proteins. Activation of the LacZ reporter gene, measured as -galactosidase activity with ONPG substrate, was assayed to measure the strength of the yeast two-hybrid interaction. The values shown are averages of three experiments. A diagram illustrating the structure of the 22-kD -zein deletion mutant (see Figure 4A) in the pACT2 plasmid is provided for each assay. (A) Interaction of the complete 22-kD -zein (construct pACT2-a) with other zeins. (B) Yeast coexpressing construct pACT2-k and the 22-kD -zein, 16-kD -zein, 15-kD -zein, and 10-kD -zein. (C) Yeast coexpressing construct pACT2-o and the 22-kD -zein, 16-kD -zein, 15-kD -zein, and 10-kD -zein. (D) Yeast coexpressing construct pACT2-q and the 22-kD -zein, 16-kD -zein, 15-kD -zein, and 10-kD -zein. (E) Yeast coexpressing construct pACT2-r and the 22-kD -zein, 16-kD -zein, 15-kD -zein, and 10-kD -zein. Error bars indicate ±SD.

Synthesis and Processing of Zein Proteins in Yeast Cells
We tested yeast cells as a heterologous system to synthesize zeins and form protein bodies. The coding sequences of genes encoding the 22-kD -zein, 15-kD -zein, and 27-kD -zein were subcloned into the yeast expression vectors pGPD414 and pGPD426 between the glyceraldehyde-3-phosphate dehydrogenase constitutive promoter and the cytochrome c oxidase terminator (Mumberg et al., 1995). Yeast cells synthesizing -, -, -, and -zeins, or GFP, were found to grow more slowly than empty vector controls. Typically, freshly inoculated cultures of the empty vector control strain reached an A₆₀₀ of 1.0 OD (midlog phase) in 12 to 15 hr, whereas cells expressing any one of the heterologous zein proteins reached this OD in 25 to 30 hr. Strains synthesizing GFP or the 22-kD -zein::GFP, 15-kD -zein::GFP, and 16-kD -zein::GFP fusions (see below) grew more rapidly than those producing native zein proteins; those expressing the 15-kD -zein and the 27- and 16-kD -zeins grew slowest. We observed different phenotypes for cultures producing individual and combinations of zein proteins. Cells synthesizing 22-kD -zein::GFP with the 15-kD -zein or 22-kD -zein::GFP with the 16-kD -zein tended to clump in solution and formed more friable colonies on plates.

Figure 6 shows immunoblots made after SDS-PAGE separation of the 22-kD -zein, 15-kD -zein, and 27-kD -zein synthesized in yeast cell cultures. The 22-kD -zein and the 27-kD -zein were produced in sufficient amounts that they were detected in 5 µL of cell lysate (Figure 6A, lanes 2 and 4), but the 15-kD -zein and the 16-kD -zein accumulated in smaller amounts (Figure 6B, lanes 2 and 4), and it was necessary to load four times more lysate to detect them easily. Although zeins from yeast extracts tended to form smeared bands during SDS-PAGE (Figures 6A and 6B, cf. lanes 1 and 2 with lanes 3 and 4), especially with overloaded samples, their apparent molecular masses were similar to those of native zein proteins. In underloaded gels, the proteins showed identical mobility with native zein proteins (data not shown). Thus, these proteins appeared to undergo normal signal peptide cleavage in yeast cells, a conclusion supported by the subcellular localization of the proteins within membrane vesicles, presumably the ER (see below).

View larger version (43K): [in this window] [in a new window]   Figure 6. Immunodetection of -, -, and -Zein Proteins Synthesized in Yeast. Protein extracts from yeast transformed with native zein-expressing plasmids were separated by 12% SDS-PAGE and immunoblotted with the zein antisera described below. The molecular masses of protein standards (kD) and the migration of the various zein proteins are indicated at left. (A) Immunoblot detecting the 22-kD -zein and the 27-kD -zein. Lanes 1 and 3, maize endosperm extract; lane 2, yeast expressing the 22-kD -zein; lane 4, yeast expressing the 27-kD -zein. (B) Immunoblot detecting the 16-kD -zein and the 15-kD -zein. Lane 1 and 3, maize endosperm extract; lane 2, yeast expressing the 16-kD -zein; lane 4, yeast expressing the 15-kD -zein.

View larger version (40K): [in this window] [in a new window]   Figure 7. Immunoblot Analysis of Yeast Cells Producing GFP and Zein-GFP Fusions. Yeast extracts were prepared from cells grown to an OD600 of 1.0, 17 µL of cell lysate was loaded in each lane, and the proteins were detected with a GFP monoclonal antibody. Lane 1, wild-type yeast; lane 2, yeast expressing GFP; lane 3, yeast expressing 15-kD -zein::GFP; lane 4, yeast expressing 16-kD -zein::GFP; lane 5, yeast expressing 22-kD -zein::GFP; lane 6, yeast coexpressing 22-kD -zein::GFP and the 15-kD -zein; lane 7, yeast coexpressing 22-kD -zein::GFP and the 16-kD -zein; lane 8, yeast expressing construct pGPD-r (see Figure 8A); lanes 9, 10, and 11, yeast coexpressing construct pGPD-r and the 15-kD -zein, 16-kD -zein, and 22-kD -zein, respectably. The molecular masses (kD) of protein standards are shown at left.

View larger version (23K): [in this window] [in a new window]   Figure 8. Confocal Laser Scanning Microscopy Images of Yeast Cells Producing Maize Zein Proteins as GFP Fusions. (A) Yeast cells synthesizing GFP alone. (B) Yeast cells synthesizing 22-kD -zein::GFP. (C) Yeast cells synthesizing 22-kD -zein::GFP and the 15-kD -zein. (D) Yeast cells synthesizing 22-kD -zein::GFP and the 16-kD -zein. (E) Yeast cells synthesizing construct pGPD-r. (F) Yeast cells synthesizing construct pGPD-r and the 15-kD -zein. Yeast cells expressing zein::GFP fusion proteins contained relatively small, fluorescent structures (arrows) and fluorescent clumps (arrowheads) of various shapes and sizes, whereas fluorescence was diffuse in the cytoplasm of yeast cells expressing GFP alone. The fluorescence was excluded from the large central vacuole (v). All images are at a same magnification; bar in (A) = 10 µm.

To examine the subcellular localization of the zein::GFP fusion proteins, yeast cultures were grown to an OD₆₀₀ of 1.0 and the cells were examined by laser scanning confocal microscopy. Figure 8A shows examples of cells expressing GFP alone. More than 90% of the yeast population showed a high level of green fluorescence that was diffuse throughout the cytoplasm, except in the region of the large central vacuole. In contrast, only 50 to 70% of the yeast cells expressing zein::GFP fusions showed green fluorescence. Cells synthesizing 22-kD -zein::GFP contained green fluorescent spherical structures that varied in shape and size and were found typically at the periphery of the cells. In some cases, these protein aggregates were observed as bright green fluorescent accretions or granules (Figure 8B). Similar results were obtained with yeast cells expressing 15-kD -zein::GFP and 16-kD -zein::GFP (data not shown). Yeast cells synthesizing 22-kD -zein::GFP with the 15-kD -zein (Figure 8C) or 22-kD -zein::GFP with the 16-kD -zein (Figure 8D) contained protein aggregates of variable sizes, similar to those observed for 22-kD -zein::GFP alone.

Detection of Protein Bodies in Yeast Cells by Immunogold Labeling
Because of the limited resolution of confocal fluorescence microscopy, the localization of zein::GFP fusion proteins in yeast cells was analyzed by immunocytochemistry using transmission electron microscopy. In cells producing 22-kD -zein::GFP, small protein accretions labeled with three to five gold particles and amorphous clumps that ranged from 0.2 to 1 µm in diameter with dense gold particle labeling were observed (Figure 9A, arrowheads). The latter may correspond to the amorphous fluorescent protein accretions observed by confocal fluorescence microscopy. The accretions of 22-kD -zein::GFP occasionally appeared to be surrounded by membranes continuous with ER cisternae; however, in general, it was difficult to resolve membranes or determine membrane continuity in sections prepared for immunocytochemistry. In cells producing 15-kD -zein::GFP (Figure 9B), only small accretions 0.2 µm in diameter were observed. The small size and infrequent observation of these protein accretions were consistent with the relatively low level of the 15-kD -zein detected in yeast cells by immunoblotting (Figure 6).

View larger version (78K): [in this window] [in a new window]   Figure 9. Immunolocalization of Zeins in Yeast Cells. (A) Yeast cells synthesizing 22-kD -zein::GFP. (B) Yeast cells synthesizing 15-kD -zein::GFP. (C) Yeast cells synthesizing 22-kD -zein::GFP and the 15-kD -zein. (D) Yeast cells synthesizing construct pGPD-r. Arrowheads indicate accretions decorated with multiple gold particles. The most prominent protein bodies found from another cell of the corresponding yeast population are shown in the insets. For the single immunolabeling shown in (A), (B), and (D), 10-nm gold particles were used. In (C), 22-kD -zein::GFP was labeled with 10-nm gold particles, and the 15-kD -zein was labeled with 5-nm gold particles. Bars = 200 nm.

Ultrastructural analysis of yeast cells expressing the protein encoded by construct r showed that it formed accretions similar to the other zeins. Laser scanning confocal microscopy of yeast cells synthesizing this protein (data not shown) or construct r plus the native 15-kD -zein (Figure 8F) revealed accretions distributed at the periphery of the cell. Immunocytochemical analysis of thin sections prepared for transmission electron microscopy showed protein accretions comparable in size to those consisting of 22-kD -zein::GFP and the 15-kD -zein (cf. Figures 9C and 9D). Thus, the insertion of GFP into the center of the 22-kD -zein had little effect on its ability to form protein bodies in yeast cells.

	DISCUSSION

TOP Abstract INTRODUCTION RESULTS DISCUSSION METHODS References

 
Zein mRNA Localization Does Not Appear to Influence the Site of Protein Body Formation
We conducted high-resolution in situ hybridization experiments to test the hypothesis that some zein mRNAs are inserted preferentially at regions of the ER surrounding protein bodies. The 27-kD -zein and the 22-kD -zein were selected based on evidence that the 27-kD -zein plays a role in initiating protein body formation (Geli et al., 1994; Coleman et al., 1996) and because the two proteins have different patterns of spatial and temporal accumulation in protein bodies (Lending and Larkins, 1989). In addition, the nucleotide sequences of these genes are very different (Marks et al., 1985; Woo et al., 2001), making it unlikely that they would cross-hybridize during in situ hybridization.

The major finding of these experiments, namely, that both 22-kD -zein and 27-kD -zein mRNAs are distributed symmetrically on cisternal ER as well as on the ER surrounding protein bodies in tissue prepared for electron microscopy, can be interpreted in several ways. First, the results could indicate that zein mRNAs are inserted at random sites along the length of the ER membrane. However, because the distributions of only two types of zein mRNAs were assayed, it is a formal possibility that those not tested, -zein and -zein mRNAs, are localized to certain regions of the ER, whereas -zein and -zein mRNAs are not. Because there was no evidence of cytoskeleton (filamentous actin or microtubules) in these samples, this structure may have been destroyed during fixation or processing. Attempts were made to preserve the cytoskeleton for electron microscopic observation using high-pressure freezing, but these were not successful. Therefore, it is possible that the zein mRNAs, which may have been anchored at certain locations in vivo by cytoskeletal elements, were redistributed artifactually during sample preparation.

The method we used for mRNA localization is similar to that described by Li et al. (1993a) to determine mRNA spatial distribution in rice endosperm. In the rice study, it was concluded that an asymmetric distribution of glutelin versus prolamin mRNAs on the ER occurred despite the absence of cytoskeletal structures. Therefore, the random distribution of prolamin mRNAs on the ER of maize endosperm, compared with their asymmetric distribution in rice endosperm, may indicate a true difference between the two cereals. It is possible that in rice, the segregation of prolamin and glutelin mRNAs is functionally important, because the prolamins are retained in the ER, whereas the glutelins are transported to protein storage vacuoles. The segregation of zein mRNAs may not be necessary, because once the proteins are inserted into the lumen of the ER, they may simply diffuse and coalesce based on their biochemical and structural properties (see below). Regardless of whether zein mRNAs are inserted exclusively at sites in which protein bodies are forming, the cytoskeleton may play a role in transporting them to the ER membrane.

If zein mRNAs are targeted more or less randomly to ER membranes, by what mechanism do the proteins associate with one another and assemble into a discrete protein body? There is evidence that ER chaperonins, such as Bip and PDI, play a role in zein protein folding (Fontes et al., 1991; Zhang and Boston, 1992; Li and Larkins, 1996), but whether they are components of a macromolecular complex that assembles zeins simultaneously into a protein body, as suggested for rice prolamins (Li et al., 1993b), is unclear. The observation that zeins appear to be secreted more or less randomly into the ER suggests the existence of some type of assembly complex or an autoassembly process based on the structural features of the zein proteins themselves.

Interactions between Zein Proteins Can Be Identified with the Yeast Two-Hybrid System
Expressing zeins as Gal4 fusions in the yeast two-hybrid system appears to be an effective approach to investigate interactions between these proteins. In interpreting the results of these experiments, we made several hypotheses. First, protein associations in the nucleus are comparable to those occurring within the lumen of the ER. This hypothesis is implicit in many other experiments in which interactions between nonnuclear proteins are examined with the yeast two-hybrid system (Shaywitz et al., 1997; Takatsu et al., 2001). On the basis of the authenticity of the results from such experiments, we believe that this is a valid hypothesis. Second, we assume that the intensity of the -galactosidase reaction, as measured by the hydrolysis of X-Gal (Figure 3) or ONPG, reflects the amount of functional -galactosidase enzyme produced and hence the affinity between two proteins. We also assume that the affinities between interacting zeins are such that differences in the transcription and nuclear import of the fusion proteins were negligible. Although we did not measure the level of transcription between yeast strains expressing the various zein constructs, based on the observation of consistent levels of -galactosidase expression in replicated experiments, we believe that this was the case. Each assay of a two-hybrid interaction was duplicated, and the experiments were replicated many times. Based on filter and 3-aminotriazole assays of -galactosidase activity, a functional transcription factor was formed regardless of which vector was used to express zein coding sequences. Cells expressing the zeins with the pACT2 and pAS2 plasmids showed comparable rates of growth and exhibited normal phenotypes, in contrast to the expression of native zeins in yeast (see below). The addition of a C-terminal GFP fusion protein to the mature 22-kD -zein or deletion mutants of this construct did not influence the yeast two-hybrid interaction significantly, and GFP was a valuable tool for detecting the proteins when they were made as secretory proteins.

The strong yeast two-hybrid interactions between the - and -zeins are consistent with the colocalization of these proteins in developing protein bodies (Lending and Larkins, 1989; Woo et al., 2001). Although we do not know the structure of the domains that interact, based on the unique features of the N termini of the 50- and 27-kD -zein proteins, we hypothesize that it is most likely the C-terminal 130 amino acids that are highly conserved among them. We created deletion mutants of this region with the 27-kD -zein, but nearly all of these modifications eliminated interactions in the yeast two-hybrid system. Consequently, it appears that this sequence creates a highly ordered secondary structure that is disrupted easily.

In view of previous experiments showing that the 27-kD -zein associates with the 22-kD -zein and promotes its retention in the ER (Coleman et al., 1996), we were surprised not to detect a stronger association between the 27-kD -zein and the 22- and 19-kD -zein proteins in the yeast two-hybrid system (Figures 3 and 5). Nevertheless, we did observe a strong association between the 16-kD -zein and the 15-kD -zein and between the 22-kD -zein and the 10-kD -zein. This finding suggests that the C-terminal 130 amino acids of the - and -zein proteins also interact with the - and -zeins. If the Pro-rich repeats of the 27-kD -zein are directed at the surface of the protein body, as suggested by Geli et al. (1994), then the C-terminal regions of the -zeins would be internal, and they could be in contact with the - and -zeins.

We expected that the -zeins, which are structurally closely related (Argos et al., 1982), would have a greater tendency for self-interaction than for binding with other zein proteins. However, the self-interaction of the 22-kD -zein in the yeast two-hybrid system was weaker than that of the 19-kD -zein (Figure 3), and there was a surprisingly weak interaction between the 22- and 19-kD -zeins. A C-terminal GFP fusion to the 22-kD -zein enhanced its interaction with itself and with the 15-kD -zein somewhat (as measured in the 3-aminotriazole assay; Figure 3B); however, this influence was not obvious in the X-Gal filter assay (Figure 3A). Because yeast cells expressing mature zeins as GFP fusions generally grew faster than those without them, this stronger interaction may be a consequence. The predicted structure of the -zein proteins suggests that they associate through hydrogen bonds and hydrophobic interactions (Argos et al., 1982; Garratt et al., 1993). It is possible that these are authentic interactions, but they are too weak to promote the formation of an effective GAL4 transcription factor. Although yeast cells grew faster when the zein proteins were made as GFP fusions, the GFP protein did not appear to alter the nature of the interactions between these proteins. The C-terminal GFP fusion appeared to enhance the 22-kD -zein association with itself, but it did not alter qualitatively the interactions with the native 22- and 19-kD -zeins. Neither did the addition of GFP to the 22-kD -zein appear to influence the interactions with other types of zein proteins in the ONPG assay, which is the most sensitive and quantitative measure of yeast two-hybrid interactions (Figure 5).

To identify regions in the 22-kD -zein that interact with the -zeins, we constructed a set of deletion mutants in which the N terminus, the C terminus, and variable numbers of the repeated peptides were removed. Because equivalent interactions were obtained regardless of which yeast two-hybrid vector expressed the native proteins, we tested only the mutants in the pACT2 plasmid (Figure 4). Yeast cells expressing the deletion mutants grew twice as fast as those expressing the full-length 22-kD -zein, but the nature of the protein interactions was qualitatively the same (cf. Figure 5A with Figures 5B to 5E). The results of these experiments suggest that interactions between the 22-kD -zein and other zein proteins involve more than just the repeated peptide domain. The 39 amino acids of the N terminus and the last 31 amino acids of the C terminus clearly influence the association of the 22-kD -zein with other zein proteins. Interestingly, the N terminus and the first two repeated peptides appear to have an affinity for the 22-kD -zein and the 10-kD -zein, whereas the last two repeats and the C terminus have a greater affinity for the 15-kD -zein and the 16-kD -zein (cf. Figures 5B and 5C). It may not be valid to compare the strength of the affinities determined by this assay; however, the data in Figures 5B to 5E were obtained with cultures grown simultaneously for the same length of time.

Unfortunately, little is known about the three-dimensional structure of zein proteins, and models describing the conformation of -zeins make no predictions regarding the N- and C-terminal regions (Argos et al., 1982; Garratt et al., 1993). Consequently, it is difficult to explain the basis of the affinities of the N- and C-terminal regions of the 22-kD -zein for the -/-zeins and -/-zeins, respectively. The hydropathy plots of these regions suggest that the N terminus and the first two repeats are more hydrophobic than the last two repeats and the C terminus (Figure 10). This might explain the greater affinity of the 10-kD -zein for the N terminus, because the -zein is by far the most hydrophobic protein of the group. The C terminus and the last two repeats contain alternating hydrophobic and hydrophilic sequences that reflect the character of the C-terminal regions of the - and -zeins (Figure 10). However, it would be surprising if the nature of these interactions was based strictly on hydrophobic/hydrophilic interactions.

View larger version (25K): [in this window] [in a new window]   Figure 10. Kyte-Doolittle Hydropathy Analysis of the Primary Amino Acid Sequences of Zein Proteins. (A) The 22-kD -zein. (B) The 16-kD -zein. (C) The 15-kD -zein. (D) The 10-kD -zein.