γ-Zeins are essential for endosperm modification in quality protein maize

Abstract

Essential amino acids like lysine and tryptophan are deficient in corn meal because of the abundance of zein storage proteins that lack these amino acids. A natural mutant, opaque 2 (o2) causes reduction of zeins, an increase of nonzein proteins, and as a consequence, a doubling of lysine levels. However, o2’s soft inferior kernels precluded its commercial use. Breeders subsequently overcame kernel softness, selecting several quantitative loci (QTLs), called o2 modifiers, without losing the high-lysine trait. These maize lines are known as “quality protein maize” (QPM). One of the QTLs is linked to the 27-kDa γ-zein locus on chromosome 7S. Moreover, QPM lines have 2- to 3-fold higher levels of the 27-kDa γ-zein, but the physiological significance of this increase is not known. Because the 27- and 16-kDa γ-zein genes are highly conserved in DNA sequence, we introduced a dominant RNAi transgene into a QPM line (CM105Mo2) to eliminate expression of them both. Elimination of γ-zeins disrupts endosperm modification by o2 modifiers, indicating their hypostatic action to γ-zeins. Abnormalities in protein body structure and their interaction with starch granules in the F1 with Mo2/+; o2/o2; γRNAi/+ genotype suggests that γ-zeins are essential for restoring protein body density and starch grain interaction in QPM. To eliminate pleiotropic effects caused by o2, the 22-kDa α-zein, γ-zein, and β-zein RNAis were stacked, resulting in protein bodies forming as honeycomb-like structures. We are unique in presenting clear demonstration that γ-zeins play a mechanistic role in QPM, providing a previously unexplored rationale for molecular breeding.

Grain hardness is a key agronomic trait in maize (Zea mays L.) because it provides resistance to damage during harvesting and marketing, as well as to insect and fungal damage. Kernel texture is determined by the relative amounts of hard (vitreous) and soft (opaque) endosperm and there is a positive correlation between zein storage proteins and kernel vitreousness (1). Zeins are a heterogeneous mixture of alcohol-soluble proteins, falling into four classes based on their structure (α-, β-, γ-, and δ-zeins) (2). The zeins extracted with the Osborne method (3) are classified as z1 (19- and 22-kDa α-zeins) and the cross-linked z2 group (50-, 27-, and 16-kDa γ-zeins, 15-kDa β-zein, and 18- and 10-kDa δ-zeins) (4, 5). Zeins are deposited in rough endoplasmic reticulum-delimited protein bodies (PBs) in endosperm cells from around 10 d after pollination (DAP) (6, 7). Alpha- and δ-zeins are mainly stored in the center of PBs, and γ- and β-zeins are deposited in the peripheral region (8). The Cys-rich γ- and β-zeins have redundant function in the stabilization of PB morphology (9). The translucency (vitreousness) of the mature kernel is influenced by PB composition and the spatial organization of α-, β-, γ-, and δ-zeins (10–16).

Because zeins are essentially devoid of lysine and tryptophan, their high-level accumulation results in poor grain-protein quality (17, 18). The opaque2 (o2) mutant, identified by Singleton and Jones in the 1920s (19), initially showed great promise in monogastic animal feeding trials because it was shown to have twice the normal levels of lysine and tryptophan (15). The improved nutritional value resulted from much reduced α-zein accumulation, which paradoxically caused a soft endosperm texture that ultimately prevented the commercial success of o2. Quality protein maize (QPM) was developed by selecting modified o2 lines with restored vitreous endosperm that maintained the low-α-zein, high-lysine phenotype (20).

Genetic analysis of o2 modifiers identified several disperse quantitative trait loci (QTLs). Although their molecular identities have remained unknown, QTLs could be correlated with observed increases in 27-kDa γ-zein transcript and protein in QPM (21). Unlike the 22-kDa α-zein genes, the 27-kDa γ-zein gene is not under the transcriptional control of the O2 protein (22, 23). Two different QTLs, which are candidates for o2 modifier genes, affect 27-kDa γ-zein gene expression. The first of these is associated with increased expression (24) and the other is linked to o15, a mutation at a different chromosome 7 location (25), which causes decreased 27-kDa γ-zein expression. The first QTL could be a cis-acting mutation of the 27-kDa zein gene, and the latter a trans-acting factor. In the B73 genome, single-copy γ-zein genes encode the 50-, 27-, and 16-kDa proteins (6). The 27- and 16-kDa γ-zein genes originated from a common progenitor by allotetraploidization and share high DNA and protein-sequence similarity (5). Their proteins amount to about 20 to 25% of total zeins (26); the low abundance 50-kDa γ-zein gene has low similarity to the other two γ-zein genes (6). Although the 16-kDa γ-zein expression is not elevated like the 27-kDa γ-zein gene in QPM, probably because of diverged regulation, their protein products and β-zein have redundant and unique function in protein body stabilization (9). Furthermore, neither the 27-kDa γ-zein-null mutant nor the β-zein RNAi seeds showed any opaque phenotype (9, 27). When the 27- and 16-kDa γ-zeins were knocked-down by γRNAi, only partial opacity occurred (9). The opacity was strongly intensified when the γRNAi and βRNAi were combined. The opacity was not caused by reducing the thickness of the vitreous endosperm as in o2, but by an incomplete embedding of starch granules in the vitreous area (9). Because the expression of the β-zein gene is also regulated by O2 (28) and significantly reduced in QPM (29), the amount of γ-zeins would become critical to keep starch granules embedded in the vitreous area. To examine the role of γ-zeins in QPM, we used an RNAi construct designed from the inverted coding sequences of the 27-kDa γ-zein gene to knock down both 27- and 16-kDa γ-zeins by taking advantage of their DNA sequence conservation (9). Indeed, in the progeny with the genotype Mo2/+; o2/o2; γRNAi/+ from the cross of QPM with O2/o2; γRNAi/+, the modified phenotype was lost, indicating that the action of the o2 modifiers is preconditioned by the expression of γ-zeins, which is genetically described as hypostasis. In the resulting endosperm cells, although discrete PBs were still observed, larger, honeycomb-like masses of unseparated PBs were also observed. This indicated that both z1 and z2 proteins are necessary to maintain the density, integrity, and spatial separation of PBs. By stacking of the 22-kDa α-zein, γ-zein, and β-zein RNAis (designated as z1CRNAi, γRNAi, and βRNAi, respectively), which eliminated the possible pleiotropic effects caused by o2, the unseparated PB phenotype became more severe. This finding reinforces the fact that different zeins have evolved to play distinct roles in the development of the endosperm.

Results and Discussion

Segregation Analysis of Zein Accumulation Patterns of QPM by the γRNAi Line.

The knock-down lines or natural null mutants of α-, β-, γ-, and δ-zein genes and their resulting phenotypes have been characterized previously (9, 16, 30). To test the theory that the increased γ-zein in QPM is necessary for endosperm modification, the γRNAi knock-down line described above was introduced into a QPM line. As a validation of the starting material, we compared protein levels of both parental lines by SDS/PAGE. As shown previously, total zeins extracted from CM105+, CM105o2, and CM105Mo2 at 18 DAP revealed an enhanced level of the 27-kDa γ-zein in CM105Mo2 (Fig. 1A) (31). Besides the 22-kDa α-zeins, the 15-kDa β-zein, also regulated by O2 (28), was markedly reduced in CM105o2 and CM105Mo2. Interestingly, the accumulation of the 19-kDa α-zeins was also significantly—although not as drastically—reduced, consistent with previous observations (29, 31, 32). As expected in the stack of the γRNAi and βRNAi, both γ- and β-zeins were knocked down to an undetectable level, and thus exhibited a highly specific inhibitory effect on zein gene expression (Fig. 1B).

Fig. 1.

Zein accumulation in normal, o2, QPM, RNAi, and their stacked mutant seeds at 18 DAP detected by SDS/PAGE. (A) CM105+, CM105o2, and CM105Mo2. The slice of protein markers run in a different gel was composited with the samples, as indicated by a white line. (B) BA (nontransgenic hybrid seed of B and A lines as a control), βRNAi, γRNAi, and the stack of the two RNAis. (C) Zein accumulation patterns for 26 kernels dissected from the cross of CM105Mo2 × O2/o2; γRNAi/+. M, protein markers from top to bottom being 25, 20, 15, and 10 kDa. Total zein loaded in each lane was equal to 500 μg of fresh endosperm at 18 DAP. The size for each band is indicated by the numbers in the “kDa” columns.

Because genotypes for o2, o2 modifiers, and RNAi were recessive, semidominant, and dominant, respectively, the γRNAi was crossed with W64Ao2, and the F1 progeny were used to pollinate a QPM line, CM105Mo2. At 18 DAP, 26 kernels were dissected for extraction of total zeins. The embryo of each individual was saved to extract genomic DNA for genotyping (Materials and Methods). PCR results showed that kernels 4, 5, 7, 8, 10, 11, 13, 17, 18, 19, 21, 23, 24, and 25 inherited the γRNAi gene (14 of 26 kernels) (Materials and Methods). As expected, the phenotypes of the zein accumulation patterns exhibited Mendelian segregation. Kernels 2, 6, 9, 14, 16, and 20 represented the genotype Mo2/+; o2/O2, with enhanced accumulation of the 27-kDa γ-zein and normal levels of the other zeins. Kernels 1, 3, 12, 15, 22, and 26 were genotype Mo2/+; o2/o2, with the 27-kDa γ-zein increased, and α-zeins (22- and 19-kDa) and β-zein decreased. Kernels 8, 10, 11, 13, 17, 18, 19, and 21 were selected to represent the genotype Mo2/+; o2/o2; γRNAi/+, and showed similar phenotype to Mo2/+; o2/o2, except the knock-down of the 27- and 16-kDa γ-zeins. Kernels 4, 5, 7, 23, 24, and 25 were of Mo2/+; o2/o2; γRNAi/+ genotype, and showed decreased protein accumulation not only for α- and β-zeins, but also for the 27- and 16-kDa γ-zeins. All kernels with this genotype show that the γ-RNAi transgene eliminates γ-zein expression in the presence of Mo2, suggesting a hypostatic relationship between the two.

Transmission Electron Microscopy of Endosperm from γRNAi in QPM.

There are very specific differences in PBs between o2 and QPM. Consequently, suppression of the modifiers can also be monitored by observing the structure of PBs. Therefore, several 18-DAP kernels with the Mo2/+; o2/o2; γRNAi/+ genotype were analyzed with transmission electron microscopy. Kernels from CM105+, CM105o2, CM105Mo2, and γRNAi were used as references. As shown in Fig. 2, the normal background CM105+ exhibited round and discrete PBs, although CM105o2 developed PBs with dramatically reduced density and size (Fig. 2 A and C and Fig. S1 A and C) (33). In the QPM line, CM105Mo2, the average number and size of PBs were significantly larger than those in CM105o2 (Fig. 2 C and D and Fig. S1 C and D), consistent with a previous report (21). This result could be further confirmed under scanning electron microscopy observation (Fig. 3 B and C) (31). When γ-zeins were knocked down, the PBs were slightly irregular in size and morphology (Fig. 2B) (9).

Fig. 2.

Transmission electron micrographs of protein bodies in different genotypes. (A) CM105+. (B) γRNAi. (C) CM105o2. (D) CM105Mo2. (E) Mo2/+; o2/o2; γRNAi/+. (F) Mo2/+; o2/o2; γRNAi/+; βRNAi/+. (Scale bars, 500 nm.) PB, protein body; RER, rough endoplasmic reticulum.

Fig. 3.

Scanning electron micrographs of protein bodies in different genotypes. (A) CM105+. (B) CM105o2. (C) CM105Mo2. (D) Mo2/+; o2/o2; γRNAi/+. (Scale Bars, 10 μm.) PB, protein body; SG, starch granules.

A unique effect on PBs was observed when the γRNAi and the QPM properties were combined in the Mo2/+; o2/o2; γRNAi/+ genotype. PBs were no longer discrete, but appeared as multilobe irregular structures (Fig. 2E and Fig. S1E). This result has previously been observed in mutants where zeins were not properly processed, like floury2 (13, 34, 35). Interestingly, when a triple combination of γRNAi, βRNAi, and QPM was examined, these abnormal PB masses were larger and more frequently observed (Fig. 2F and Fig. S1F).

Scanning Electron Microscopy of Endosperm from γRNAi in QPM.

Further examination of PBs with scanning electron microscopy permits the analysis of the interaction between PBs and starch granules, which is thought to be important in determining kernel vitreousness (36). During endosperm development, starch granules and PBs are embedded in a proteinaceous cytoskeletal matrix (31, 37). In o2 endosperm, the proteinaceous matrix is almost totally absent, resulting in loose and noncompacted starch granules (Fig. 3B), whereas in QPM, a matrix is partially restored (Fig. 3C) (31). However, the partial matrix was abolished by knock-down of γ-zeins (Fig. 3D). This finding is consistent with its considerable reduction not only in α-zeins, but also γ- and β-zeins (Fig. 1C). In QPM, although PB size, number, and proteinaceous matrix were all reduced compared with wild-type endosperm, these parameters were all considerably larger than in o2 (Fig. 3C) (31). However, when the γRNAi was introduced into QPM, the resulting genotype Mo2/+; o2/o2; γRNAi/+ generated the same phenotype as the o2 mutant (Fig. 3 B and D), where the proteinaceous matrix is completely disrupted, indicating that endosperm modification is abolished.

The elevated accumulation of the 27-kDa γ-zein in QPM endosperm is associated with an increase in the number and size of PBs, and vitreous endosperm formation in the o2 mutant background. Whether the restoration of the proteinaceous network is related to the elevated expression of 27-kDa γ-zein remains to be seen. Interestingly, uncharacterized opaque mutants, such as opaque9, exhibit no reduction in zein accumulation (38), suggesting that they might only have changes in spatial deposition of zeins in PBs, like floury1 (10), or encode cytoplasm-resident proteins that are necessary for vitreous endosperm formation during kernel maturation. It is possible that such hypothetical proteins are increased in QPM. The modification effects of the dominantly acting o2 modifier genes on the soft o2 phenotype are completely suppressed by the γRNAi. This shows that γ-zeins are necessary for modification but does not prove that they are alone sufficient. If other genes act in concert with the γ-zeins and increase to allow modification, we propose that actions of these genes are hypostatic to the high expression of γ-zein genes.

Loss of QPM Kernel Phenotype by Knock-Down of γ-Zeins.

The breeding of QPM lines can easily be monitored by kernel phenotype because modifiers restore the translucency of the kernel that is lost in opaque mutants. In the progeny of CM105Mo2 × (W64Ao2 × γRNAi/+), kernels with the genotype Mo2/+; o2/o2; γRNAi/+ are not only sharply reduced in α-zein, but also in γ-zeins. The 27-kDa γ-zein locus is linked to one of the QTLs for endosperm vitreousness (24) and is associated with increased 27-kDa γ-zein expression (21). It is not known if a mutation in the 27-kDa γ-zein gene or promoter is the QTL itself or if its expression is affected by another gene, which is itself the QTL. In either case, the question is raised as to whether the restoration of vitreousness depends on a threshold level of γ-zeins. If enhanced γ-zein levels allows endosperm modification either acting independently or in concert with other o2 modifiers, its loss should result in reappearance of the soft o2 phenotype. Indeed, the resulting ear exhibited a 1:3 segregation of opaque and vitreous kernel phenotypes, instead of the 100% vitreous, as would be expected if 27-kDa γ-zein were not necessary for modification (Fig. 4). Progeny kernels carrying the dominant RNAi construct were totally opaque and contained negligible hard endosperm compared with normal and QPM kernels (Fig. 4), although normal kernels with only the γRNAi showed no or only slight opacity in the crown area (Fig. S2C) (9). When the kernels with the genotype Mo2/+; o2/o2; γRNAi/+ were backcrossed with CM105Mo2, progeny exhibited 1:1 segregation of opaque and vitreous kernel phenotypes (Fig. S3). In such cases, half of the progeny should be homozygous for the 27-kDa γ-zein locus, the major QTL for endosperm modification in QPM. If heterozygosity of this modifiers would be insufficient for a vitreous phenotype, one would expect only a 3:1 segregation because only a quarter of the progeny would be both γRNAi and homozygous for the modifier. These data confirm the link between the QPM phenotype and γ-zein levels in endosperm. There is a good correlation between transmission electron microscopy observations of PBs and kernel phenotypes.

Fig. 4.

Kernel phenotype for CM105+, CM105o2, CM105Mo2, and new mutant Mo2/+; o2/o2; γRNAi/+. Photographs for intact or decapped kernels were taken under incandescent light (Top and Bottom) or with transmitted light (Middle).

Triple Stack of z1CRNAi, γRNAi, and βRNAi Transgenes.

It has been shown that O2 regulates more genes than just the 22-kDa α- and β-zeins (39, 40). Furthermore, the QPM phenotype comprises several QTLs, which make quantitative contributions (24). It is therefore conceivable that other factors might be responsible for the disintegration of the subcellular structure of PBs and starch granules in addition to or instead of the regulation of γ- and α-zein gene expression. To exclude these potential pleiotropic effects, a triple stack of z1CRNAi, γRNAi, and βRNAi transgenes was created (Fig. 5) by reciprocal crosses between a homozygous z1CRNAi transgenic plant (16) and a plant heterozygous for γRNAi and βRNAi (9).

Fig. 5.

Protein accumulation analysis of the γ/β RNAi, z1CRNAi and their triple-stack (z1CRNAi used as pollen) and transmission electron micrographs of their protein bodies. (A) Eight kernels dissected from the cross of γ/β RNAi × z1CRNAi at 18 DAP were analyzed. C, nontransgenic hybrid seed of B and A lines as a control. M, protein markers from top to bottom being 25, 20, 15, and 10 kDa. The slice of protein markers and control run in a different gel was composited with the samples, as indicated by a white line. Total zein loaded in each lane was equal to 500 μg of fresh endosperm at 18 DAP. The size for each band is indicated in the “kDa” column. (B–E) Transmission electron micrographs of protein bodies in γ/β RNAi stack (B), z1CRNAi (C), and the stack of γ/β RNAi and z1CRNAi (D and E). (Scale bars, 500 nm.)

Both parental lines were used as references for SDS/PAGE and transmission electron microscopic analysis (Fig. 5 A–C and Fig. S4 A and B). At 18 DAP, eight immature kernels were dissected for extraction of total zeins. A nontransgenic Hi II hybrid seed of B × A lines was used as a control (Materials and Methods). As expected, all kernels exhibited reduced 22-kDa α-zein level. Other zeins were not affected, indicating the highly specific action of RNAi (Fig. 5A). Kernels 2, 3, 7, and 8 represented the genotype z1CRNAi/+, with only the 22-kDa α-zeins reduced; kernels 4 and 5 represented the genotype z1CRNAi/+; γRNAi/+; βRNAi/+, with not only the 22-kDa α-zein but also γ- and β-zeins reduced; kernels 1 and 6 represented the genotype z1CRNAi/+; γRNAi/+ and z1CRNAi/+; βRNAi/+, respectively, both besides substantially losing the 22-kDa α-zeins, the γ-zeins or β-zein being knocked down to barely detectable levels as well (Fig. 5A). As expected, the kernels with the triple stack of the z1CRNAi, γRNAi, and βRNAi showed a full opaque phenotype, similar to kernels with z1CRNAi alone (Fig. S2E) (9, 16, 41), indicating that a significant loss of α-zeins is sufficient to cause kernel opacity alone (Fig. S2E).

Most PBs in the triple stack of z1CRNAi, γRNAi, and βRNAi had an irregular shape similar to the double stack of βRNAi and γRNAi (Fig. 5 B, D, and E and Fig. S4 A–C) (9). Moreover, the mature PBs were no longer discrete but consisted of masses that appeared to comprise tens of unseparated PBs, with diameters considerably larger than the normal 1- to 2-μm size (Fig. 5 D and E and Fig. S4C). This finding was similar to the observations in the stack of QPM and γRNAi (Fig. 2 E and F and Fig. S1 E and F), only more severe. Scanning electron microscopy revealed that these PB masses could still interact with starch grains (Fig. S5). Because of the abnormal size and shape of PB masses and their unknown surface characteristics, it seems that interactions with starch grains are not able to contribute to the formation of a vitreous endosperm upon desiccation (Fig. 6).

Fig. 6.

Model for vitreous endosperm formation in which midmaturation stage (18 DAP) starchy endosperm cells are depicted. PBs are represented with gray spheres, starch grains with white spheres, and proteinaceous matrix with blue lines. In wild-type and QPM, compact stable matrices give rise to glass-like, vitreous endosperm at maturity. In opaque mutants and dominant RNAi low-zein lines, small, sparse, or lobed unseparated PBs produce loose, unstable matrices, which shatter during desiccation, producing an opaque texture at maturity.

Mechanism of PB Aggregation.

Undoubtedly, one effect of the γRNAi effect on QPM is to prevent accretion of zeins into normally shaped endoplasmic reticulum-derived PBs. Other mutants with similar clumped PBs always show the opaque endosperm phenotype (9, 11, 12, 14). What could cause this abnormal PB phenotype? Our use of an RNAi transgene permits us to exclude the pleiotropic effects of the conventional trans-acting factors, like O2. Lobing and failed separation of PBs occurs when both Cys-rich (γ- and β-) and Cys-poor (22-kDa α-) zeins were reduced simultaneously, suggesting that the phenotype is associated with a quantitative and qualitative loss of zeins, rather than a reduction only in either the 22-kDa α-zein or γ- and β-zeins (Fig. 5 B and C and Fig. S4 A and B). Protein levels alone cannot be the basis for the phenotype because, although both triple-stacked RNAi and o2 have dramatically reduced total zeins, o2 has discrete PBs. The difference is that the γ-zeins reach normal levels in the o2 mutant but are lost in the triple stack. Therefore, lobed and unseparated PB masses occur when both α- and γ-/β- zein classes are reduced to low levels. It is possible that the residual 19-kDa α-zeins in the QPM × γRNAi and the α-, β-, and γ-RNAi triple stack prevent PB separation because of the loss of encapsuling γ-zeins, and that their difference in relative severity relates to the amount of residual α-zeins. In comparison, o2 has normal discrete PBs because residual α-zeins are encapsulated in γ-zeins. This explanation for disrupted PB separation is summarized in Table 1.

Table 1.

Effect of zein disruptions on protein-body morphology

Genotype	Zeins present	Zein interactions	Phenotype
o2	Low levels of 19- and 22-α; normal γ	Residual α are encased in γ	No PB lobing
QPM γRNAi	Low levels of 19- and 22-α; low γ	Residual α are not encased in γ	Moderate PB lobing
Triple stack (22-α, -γ, and -β RNAi)	Moderate 19- and low 22-α; low γ	Lots of 19 α are not encased in γ	More severe PB lobing

Interestingly, the morphology and organization of PBs in this triple stack is reminiscent of fl2, where protein-body separation is also disrupted (34); fl2 is linked to a single 22-kDa zein gene with an amino acid change that prevents processing of the signal peptide, thereby slowing down the deposition of other zeins into PBs. During early PB formation, only γ- and β-zeins are deposited (8), and this explains the variable penetrance of the phenotype in relation to the PB size gradient between the subaleurone layer and the central portion of the endosperm (34). The fused PBs observed in our triple RNAi stack may arise by the same mechanism as in fl2, as both have reduced accumulation of α-zeins, as well as γ- and β-zeins. A common feature of fl2 and the triple stack may be their “out of context” accumulation of α-zeins, where the semidominant mutant fl2 α-zein or large amounts of the 19-kDa α-zeins lack the encapsulating 22-kDa α- and γ-zeins. An unknown mutation causes similarly aggregated PBs in a highly digestible sorghum cultivar (42, 43), inviting speculation that it is caused by a semidominant kafirin mutation analogous to floury 2.

It has also been shown that PBs in endosperm cells are not randomly distributed but are evenly distributed around starch granules embedded in a protein matrix rich in EF-1α and cytoskeletal elements, such as actin and microtubules (37). Indeed, seed architecture could have evolved to use the cytoskeletal network in the distribution of PBs. If this is the case, the interaction between PBs and the cytoskeleton could depend on an appropriate internal PB structure, which depends on the timely accumulation of the correct proportions of γ-, β-, and α-zeins. In o2 endosperm, this balance is upset both in terms of PB size, shape, and the surrounding protein matrix (Fig. 6). These parameters are only partially restored to wild-type in QPM but result in kernels that are as vitreous as wild-type. In this case, it is possible that the increased 27 kDa γ-zein allows the accumulation of small but numerous PBs that, along with the proteinaceous matrix, participate in a mature endosperm structure whose vitreousness perhaps develops in a somewhat distinct manner from wild-type (Fig. 6). Whether similar or distinct to wild-type endosperm maturation, we are unique in showing that an increase in γ-zein is essential for endosperm modification in QPM.

Materials and Methods

Genetic Stocks.

Maize inbred line CM105+, its o2 mutant CM105o2, and the modified o2 mutant CM105Mo2 have been reported previously (24). CM105 wild-type, o2 and Mo2 were grown in the field in Lincoln, Nebraska in summer, 2009. The z1CRNAi (for knockdown of the 22-kDa α-zeins), γRNAi (for knockdown of the 27- and 16-kDa γ-zeins), and βRNAi (for knockdown of the 15-kDa β-zein) were generated previously (9, 16).

Total Zein Extraction.

Immature kernels at 18 DAP were harvested. The embryos were saved for extraction of DNA individually. The genotype of each kernel with respect to the γRNAi gene was confirmed by PCR with primer pair 5′-ACAACCACTACCTGAGCAC-3′ and 5′- ATTAAGCTTTGCAGGTCACTGGATTTTGG-3′, which has been described elsewhere (9). Each endosperm was cut into two halves. One was fixed for transmission or scanning electron microscopy. The other half (≈50 mg) was used for zein extraction. The finely ground endosperm was mixed and vortexed with 400 μL of 70% ethanol/2% 2-mercaptoethanol (vol/vol), then kept on the bench at room temperature for 2 h; the mixture was centrifuged at 13,000 rpm (Eppendorf, Centrifuge 5417C) in a microfuge for 10 min, then 200 μL of the supernatant liquid was transferred to a new tube; 10 μL of 10% SDS was added to the extract, the mixture was dried by vacuum, and resuspended in 100 μL of distilled water. Next, 2 μL (equal to 500-μg endosperm powder) of each sample was analyzed by 15% SDS/PAGE gel.

Stacking of Zein RNAi Transgenes.

The QPM CM105Mo2 was pollinated by the F1 progeny of W64Ao2 × γRNAi or W64Ao2 × (γRNAi × βRNAi).

The triple stack of the z1CRNAi, γRNAi, and βRNAi was generated from reciprocal crosses between a homozygous z1CRNAi plant and a double stack of γRNAi and βRNAi.

Transmission and Scanning Electron Microscopy.

Previously published methods were used with some modifications (34, 44). Two 2-mm thick sections were sliced perpendicular to the aleurone layer to include the pericarp, aleurone, and 10 to 20 cell layers of the endosperm. All these slices were fixed in 5% glutaraldehyde in 0.1 M sodium cacodylate buffer, PH 7.4, containing 2% sucrose, in a 2-mL tube. Fixation was kept at 4 °C overnight and for another 3 h at room temperature. The tissues were rinsed for 2 to 3 h with several changes of 0.1 M sodium cacodylate buffer containing decreasing amounts of sucrose. They were then postfixed in buffered 1% osmium tetroxide at 4 °C overnight followed by dehydration in a graded series of acetone washings and embedded in epon resin.

For electron microscopy, 90-nm thin sections were cut on a Leica EM UC 6-μL tramicrotome. Sectioned grids were stained with saturated solution of uranyl acetate and lead citrate. Sections were analyzed at 80 Kv with a Philips CM 12 transmission electron microscope.

For scanning electron microscopy, the dehydrated 18-DAP samples were dried to a critical point, using CO₂ in a dryer (Balzers CPD 020); the dried samples were mounted on the surface of a brass disk using double-sided adhesive silver-tape, coated with gold/palladium by a sputter coating unit (Balzers CSD 004) and viewed on a scanning electron microscope (Amray 1830 I).