Abstract
Low levels of the essential amino acids lysine (Lys) and methionine (Met) in a maize-based diet are a major cost to feed and food. Lys deficiency is due to the abundance of Lys-poor proteins in maize kernels. Although a maize mutant, opaque-2 (o2), has sufficient levels of Lys, its soft kernel renders it unfit for storage and transportation. Breeders overcame this problem by selecting quantitative trait loci (QTL) restoring kernel hardness in the presence of o2, a variety called Quality Protein Maize (QPM). Although at least one QTL acts by enhancing the expression of the γ-zein proteins, we could surprisingly achieve rebalancing of the Lys content and a vitreous kernel phenotype by targeting suppression of γ-zeins without the o2 mutant. Reduced levels of γ-zeins were achieved with RNA interference (RNAi). Another transgenic event, PE5 expresses the Escherichia coli enzyme 3′-phosphoadenosine-5′-phosphosulfate reductase involved in sulfate assimilation, specifically in leaves. The stacked transgenic events produce a vitreous endosperm, which has higher Lys level than the classical opaque W64Ao2 variant. Moreover, due to the increased sulfate reduction in the leaf, Met level is elevated in the seed. Such a combination of transgenes produces hybrid seeds superior to classical QPMs that would neither require a costly feed mix nor synthetic Met supplementation, potentially creating a novel and cost-effective means for improving maize nutritional quality.
IN many developing countries, maize serves as an important source of nutrition in human and animal diets. Cereals like maize are limiting in essential amino acids (EAAs) Lys, Met, and tryptophan (Trp), whereas legume crops like soybean are deficient in Met. Therefore, corn is usually supplemented with soybean and synthetic free Met to provide a balanced amino acid diet for animal feed. Although humans could live on a diet of beans and corn, they cannot consume synthetic free Met, which, as a racemic substance, would not be allowed for health reasons. Met as a limiting nutrient had been demonstrated in the effects of methionine fortification of a soy isolate-based formula in infant feeding. Normal infants fed soy isolate-based formula fortified with methionine had significant weight gain compared to infants fed not supplemented soy isolate-based formula (Fomon et al. 1979). Therefore, a lot of research efforts has been expended on genetic improvements to the amino acid balance of maize kernels, either through conventional breeding or the use of recombinant DNA technology.
The bulk of proteins in mature maize kernels, prolamins, called zeins in maize, possess an amino acid imbalance toward proline, glutamine, alanine, and leucine residues. Zeins are classified into four distinct classes of α- (19- and 22-kDa), β- (15-kDa), γ- (16-, 27-, and 50-kDa), and δ-zeins (10- and 18-kDa) and coalesce into discrete spherical structures called protein bodies (PBs) (Kim et al. 2002). Mature PBs have a shell of γ- and β-zeins surrounding the core of α-zeins. The 27-kDa γ-zein controls PB initiation, whereas the 50-kDa γ-zein plays a significant role in PB expansion, the bulk of which is driven by the 19-kDa α-zein. The 22-kDa α-zein provides structural support in packaging the 19-kDa α-zein (Guo et al. 2013). On the other hand, the nonzein protein fraction in the endosperm (glutelins, globulins, and albumins) is relatively balanced in their amino acid composition (Prasanna et al. 2001). This abundance of zeins effectively dilutes the contribution of other endosperm proteins to the kernel Lys and Trp contents. Therefore, alterations in the accumulation of zeins have led to the identification of mutants with altered nutritional quality. For instance, increased accumulation of the β-, γ-, and δ-zeins relative to the more abundant α-zeins (Supplemental Material, Table S1 in File S1) has raised the sulfur (S) amino acid contents of kernels and their nutritional value (Planta et al. 2017). Reduction of α-zeins either through transcriptional regulation (e.g., loss of O2 transcription factor), or a transgene that reduces their transcript accumulation through RNA interference (RNAi), result in more nutritionally balanced endosperm proteins.
One of the zein-reduction mutants, the recessive o2 mutant, has about a 50% reduction in zeins (Tsai et al. 1978), and approximately double the amount of Lys compared to normal genotypes (Mertz et al. 1964). It primarily affects the synthesis of α-zeins. The 22-kDa α-zeins are reduced to a very low level, whereas the 19-kDa α-zeins are also reduced, but to a lesser extent (Jones et al. 1977). The o2 mutant, aside from a reduced Lys-poor zein content, has a compensating increase in the levels of Lys- and Trp-rich nonzein proteins. However, its soft and starchy endosperm makes it susceptible to fungi and insect infestation, both in storage and in the field (NRC 1988). Identification of o2 modifiers (mo2s) function as suppressors that can restore the normal kernel phenotype in the presence of o2, resulting in a new type of maize germplasm, known as quality protein maize (QPM) (Prasanna et al. 2001).
Alternatively, transforming maize with an RNAi transgene targeting the α-zeins also enhances levels of Lys and Trp in maize kernels (Segal et al. 2003; Huang et al. 2004, 2006). The observed increase in Lys and Trp in the α-zein-reduced kernels was also due to the replacement of the Lys-poor zeins with the Lys-containing nonzein proteins (Huang et al. 2006). Doubling the Lys levels without changing the protein content in corn could add up to $480 million in annual gross value to US corn in the global feed market (Johnson et al. 2000).
Among the maize opaque kernel mutants that affect the accumulation of the zeins, only a few have been reported that alter the synthesis of the γ-zeins. The maize Mucronate mutation is a deletion in the 16-kDa γ-zein that produces an abnormal 16-kDa γ-zein, whereas opaque-15 reduces the 27-kDa γ-zein synthesis and appears to be a mutation of an o2 modifier gene (Dannenhoffer et al. 1995; Kim et al. 2006). Near-isogenic lines of several high-Lys opaque mutants in the W64A genetic background showed that o2 has the highest kernel Lys content among the opaque mutants (Hunter et al. 2002). Efforts to improve the protein quality of maize seeds have focused on o2 seeds as other opaque mutants offer no additional advantage over o2 in terms of Lys content and nutritional quality.
Given these results, we sought to generate a sole crop diet of corn that is enriched in both Lys and Met. We crossed the high-Met maize line PE5 (Planta et al. 2017) with several zein RNAi lines. The RNAi lines used target α-, β-, γ-, α-/γ-, or γ-/β-zeins. The opaque PE5;α-/γ- kernels have higher Met and Lys contents than the vitreous PE5;γ- kernels, and both kernel genotypes have higher Lys and Met contents than the o2 mutant. Moreover, PE5 with RNAi targeting of γ-zeins could potentially be a preferred QPM variety because of higher Lys and Met levels and their dominance for introgression into elite lines.
Materials and Methods
Genetic stocks
The α-, γ-, and β−RNAi transgenic plants were from our laboratory stocks, and have been described elsewhere (Wu and Messing 2010, 2011; Wu et al. 2012). Both the γRNAi and βRNAi are homozygous for A654-Dzs10, a nonfunctional allele of the 10-kDa δ-zein gene from the inbred A654. The βRNAi plant used for crosses is homozygous for the RNAi transgene as all kernel progenies tested positive for βRNAi genotyping, whereas the γRNAi is hemizygous, as about half of the tested kernel progenies from crosses with γRNAi had the RNAi transgene. αRNAi, on the other hand, is in a Hi-II A×B hybrid background, and is hemizygous for the RNAi transgene. The hybrid genotype Hi-II A×B was used for maize transformation.
The transgenic event PE5 was backcrossed twice to the high-Met inbred line B101 prior to being crossed with the RNAi lines. It expresses the Escherichia coli enzyme 3′-phosphoadenosine-5′-phosphosulfate reductase, designated as EcPAPR (Martin et al. 2005), driven by the PepC promoter (Sattarzadeh et al. 2010). EcPAPR is involved in assimilatory sulfate reduction, and maize plants expressing this enzyme shows increased kernel Met content when used as the maternal parent (Tarczynski et al. 2003; Martin et al. 2005).
Crosses between the maternal PE5 and the paternal RNAi lines were performed during the summer of 2014 and 2015. PE5; γ- plants generated during the summer of 2014 were crossed with αRNAi during the summer of 2015. The rest of the crosses of γ/βRNAi, βRNAi, and αRNAi with PE5 were made during the summer of 2014. These crosses gave five distinct ears that were used for analysis: (1) PE5;αRNAi, (2) PE5;β-, (3) PE5;γ-, (4) PE5;γ-/β-, and (5) PE5;γ-/α-.
Genotyping
Genomic DNA was isolated from maize leaves at the three- to four-leaf stage using a modified CTAB extraction method (Sawa et al. 1997). For extraction of genomic DNA from mature maize kernels, a portion of the kernel that is mostly endosperm with no embryo tissues were ground to a fine powder and subjected to DNA extraction with the Nucleospin Plant II kit (Takara Bio). Transgenic plants were screened for the presence of both RNAi and EcPAPR transgenes using the primer pairs 5′-ACAACCACTACCTGAGCAC-3′/5′-ATTAAGCTTTGCAGGTCACTGGATTTTGG-3′ (Wu and Messing 2010) and 5′-CTCCCCATCCCTATTTGAACCC-3′/5′-GGTAGGTTTCCGGGAACAAGTA-3′, respectively. PCR amplification for the RNAi transgenes produced amplicons of the sizes 365, 913, and 1096 bp corresponding to α-, β-, and γRNAi lines, respectively, whereas screening for PE5 yields a 696-bp product. Kernels from an ear segregating and nonsegregating for the RNAi transgenes were pooled separately and used for DNA extraction and genotyping.
To determine possible allelic variation between different maize genotypes, amplified DNA fragments of the 27-kDa γ-zein gene were digested with PstI to display restriction fragment length polymorphisms (Konieczny and Ausubel 1993). Primers 5′-CCACCTCCACGCATACAAG-3′ and 5′-ATGGACTGGAGGACCAAGC-3′ were used to amplify a 487-bp fragment of the 27-kDa γ-zein gene spanning positions 50–546 of the coding region (Das et al. 1991). Digestion with PstI would produce three DNA fragments (487, 292, and 195 bp), when the gene exists as a tandem copy, or only two fragments for a single-copy gene (292 and 195 bp).
Protein extraction, SDS-PAGE analysis, and western blotting
Total protein from pooled endosperm samples of mature maize kernels were extracted with an alkaline sodium borate extraction buffer (Wallace et al. 1990), whereas the alcohol-soluble zeins from the endosperm of mature maize kernels were fractionated and separated in SDS-PAGE as previously described (Wu and Messing 2010).
Mature maize kernels were imbibed in water for 2 days to facilitate easier separation of the embryo from the endosperm. Proteins were isolated from embryos macerated in an SDS sample buffer [10% (v/v) glycerol, 2.3% (w/v) SDS, 5% (v/v) β-mercaptoethanol, 62.5 mM Tris-Cl, pH 6.8] at a ratio of 50 mg tissue per milliliter of buffer and the extracts processed as described previously (Belanger and Kriz 1989; Puckett and Kriz 1991).
Total protein from three mature maize leaf discs was extracted following the procedure of Conlon and Salter (2007). Ten microgram of total protein was separated in a 12% Tris-glycine SDS-PAGE gel and immunoblotted with an antibody against EcPAPR kindly provided by Dr. Jens Schwen (Krone et al. 1991). For immunodetection, the secondary antibody is a goat anti-rabbit peroxidase conjugate (Sigma-Aldrich) and was used at a 1:60,000 dilution while the primary anti-EcPAPR antibody in a TBST buffer with 0.5% BSA was used at a 1:4000 dilution.
Protein identification
Protein bands that have differential accumulation in kernels segregating for the α-zein RNAi transgene were excised out of the SDS-polyacrylamide gel and analyzed by trypsin-nano-LC-MS using Q Exactive HF hybrid quadrupole-Orbitrap mass spectrometer (Thermo Scientific). Proteins from the samples were identified at the Biological Mass Spectrometry Facility at Rutgers University.
Amino acid composition analysis
Transgenic mature kernels were ground to fine powder and ∼10 mg were used for amino acid composition analysis conducted by the Proteomics and Mass Spectrometry Facility at the Donald Danforth Plant Science Center (St. Louis, MO). Samples were pretreated with performic acid prior to acid hydrolysis, yielding the acid stable forms cysteic acid and Met sulfone that can be measured by separation on a UPLC column. Trp was not detected following acid hydrolysis.
Data availability
The zein profiles of the RNAi lines and the different EcPAPR transgenic events are presented in Figure S1 and Figure S2 in File S1. Maize genetic stocks and reagents used in this study are available upon request.
Results
Rebalancing of kernel sulfur using transgene stacking
PE5 is a high-Met transgenic maize line that specifically expresses the PepC promoter-driven E. coli enzyme 3′-phosphoadenosine-5′-phosphosulfate reductase in the leaf (EcPAPR; Figure 1A). This transgenic PE5 line exhibits an increased kernel Met content when used as the maternal parent (Planta et al. 2017). PE5 plants were therefore crossed with the zein reduction lines (Wu and Messing 2010) (Figure S1 in File S1) as the pollen donor to manifest the increased seed Met phenotype in resulting ears. Five types of distinct ears were obtained from crosses of PE5 with different zein reduction lines: (1) PE5;α-, (2) PE5;γ-, (3) PE5;β-, (4) PE5;γ-/α-, and (5) PE5;γ-/β-. At least 14 kernels from each transgenic ear were analyzed individually by phenotyping with SDS-PAGE or genotyping for the RNAi transgenes. Kernels were then pooled depending on whether they were segregating or nonsegregating for the RNAi transgenes (Figure 1B).
Zein profiles of kernels from crosses of PE5 with RNAi lines targeting the α-, γ-, β-, α-/γ-, or γ-/β-zeins. (A) Western blotting for detection of the bacterial EcPAPR in leaf tissues of PE5 plants, which were used as the maternal parents for crossing with the zein reduction lines. NT, nontransgenic null segregant of PE5. (B) SDS-PAGE zein profiles (upper panel) of segregating populations of kernels from PE5 × zein RNAi crosses. Kernels nonsegregating (odd-numbered lanes) and segregating (even-numbered lanes) for the RNAi transgenes were pooled and used for analysis. The RNAi transgenes in these segregating populations were detected by electrophoresis in 1% (w/v) agarose gel (lower panel).
Loss of β- and γ-zeins promote a more pronounced increase in the accumulation of the Met-rich 10-kDa δ-zein compared to the reduction in the S-poor α-zeins (Figure 1B). This increase in the accumulation of the 10-kDa δ-zein was previously demonstrated in transgenic maize expressing the β- and γ-zein RNAi transgenes (Wu et al. 2012). Reduction in both β- and γ-zeins channels even more protein S to the 10-kDa δ-zein than the loss of either β- or γ-zein. Although enhanced assimilation of S in the source leaf tissues increases accumulation of S-containing zeins in the seed sink tissues, loss of the S-containing zeins reallocate the protein S to the remaining or available S-containing zeins (Planta et al. 2017).
PE5 influences kernel opacity depending on zein gene expression
As the loss of α- and γ-zeins by RNAi induces a full or partial opaque seed phenotype, respectively (Segal et al. 2003; Wu and Messing 2010), hybrid kernels were inspected whether the PE5 transgenic event affects endosperm modification. Opacity of the hybrid PE5;zein RNAi kernels were phenotyped with a light box (Figure 2). Loss of β-zeins did not change the phenotype of the vitreous kernel, as indicated by the thick outer layer of the vitreous endosperm in sliced kernels (Figure 2G). Reduction in γ-zeins produced a partial opaque phenotype. The regular pattern of endosperm modification extends from the crown midway to the base of the kernel (Figure 2D), whereas reduction in both β- and γ-zeins produced complete opaque kernels (Figure 2F). However, combining PE5 with the reduction in both β- and γ-zeins completely restored the vitreous kernel phenotype (Figure 2I). This reversion to a more vitreous phenotype was also observed in PE5;γ- kernels (Figure 2E). As βRNAi kernels are vitreous, combining it with PE5 did not change its kernel phenotype (Figure 2H). Stacking of PE5 with αRNAi (Figure 2B) or α/γRNAi (Figure 2C) did not alter the opacity of the hybrid kernels. The PE5 transgenic event in PE5;γ- or PE5;γ-/β- can influence endosperm modification depending on which RNAi event is used.
Endosperm phenotypes of transgenic zein reduction kernels (upper panels) sliced in half to reveal the degree of vitreous endosperm (lower panels). (A, D, F, and G) Zein RNAi and (B, C, E, H, and I) PE5;zein RNAi lines. γRNAi and βRNAi are in a hybrid Hi II A×B and A654 backgrounds, αRNAi is in a Hi II A×B background, and PE5 was backcrossed twice to B101 prior to being crossed with the RNAi lines. PE5;α- is in a Hi II A×B and B101 backgrounds, whereas the other PE5;zein RNAi lines are in a hybrid background of Hi II A×B, B101, and A654 (see Materials and Methods).
Kernels from a cross of PE5 with αRNAi showed increased level of the 27-kDa γ-zein (Figure 1B). This increase in the γ-zein was also observed in crosses of αRNAi with other transgenic events overexpressing EcPAPR (Figure S2 in File S1), and seems to be a response to the decreased levels of α-zeins. However, reduced levels of 27-kDa γ-zein in αRNAi lines alone was attributed to segregation of the 27-kDa γ-zein gene from the A and B lines used for transformation (Segal et al. 2003), and gene silencing of the 27-kDa γ-zein gene due to the use of the γ-zein promoter in the transgenic expression cassettes (Huang et al. 2004).
QPMs, which have reduced levels of the 22-kDa α-zein, also have two- to threefold increases in the 27-kDa γ-zein (Ortega and Bates 1983; Wallace et al. 1990). This increase in the 27-kDa protein correlates, and is necessary, if not sufficient, for endosperm modification and depends on the genetic background (Lopes and Larkins 1991; Wu et al. 2010). For instance, maize inbreds have one or two copies of the 27-kDa γ-zein gene that could influence its protein level (Das and Messing 1987; Das et al. 1991). Therefore, DNA was isolated from PE5 and αRNAi, their hybrid, and their parental lines to determine the nature of γ-zein alleles (Figure S3 in File S1). Two highly similar, tandemly duplicated genes of the 27-kDa γ-zein, “A” and “B,” in inbreds like W22 and A188, have different PstI recognition sites, which were utilized for the cleaved amplified polymorphic sequence assay (CAPS; Figure S3A in File S1). The inbred line, A188, has a single rearranged B gene (Rb) originating from homologous recombination at the highly conserved 5′ regions of the two repeats (Das and Messing 1987; Das et al. 1991).
The DNA gel profile in Figure S3B in File S1 shows that Hi-II A×B, B101, PE5, αRNAi, and kernels from a PE5×αRNAi cross (Figure S3, B and C in File S1) have a single copy of the γ-zein gene. There seems to be no copy number variation in the hybrids and inbreds related to PE5 and αRNAi, and thus, unlikely that modifiers of the opaque phenotype are associated with copy number variations of the γ-zein genes in our crosses. Indeed, γ-/β- is opaque (Figure 2F), but PE5;γ-/β- is vitreous (Figure 2I), exemplifying restoration of the normal phenotype in the absence of the γ-zeins, but in the presence of PE5. However, without PE5, duplication of the 27-kDa γ-zein gene in QPMs enhances its expression and promotes endosperm modification (Liu et al. 2016).
Protein accumulation patterns in PE5;zein RNAi kernels
To determine whether protein accumulation patterns are different between transgenic zein reduction and nontransgenic controls, kernels from an ear of a PE5×RNAi cross were pooled depending on the segregation of the RNAi transgenes. Total protein and the nonzein protein fractions were extracted from these kernels and separated in an SDS-PAGE gel (Figure 3). Aside from the reduction in zeins due to the RNAi transgenes, PE5;γ- and PE5;β- kernels had similar protein accumulation profiles compared to their PE5;non-RNAi controls (Figure 3A). Reduction in α-zeins, however, in kernels of PE5;α- and PE5;α-/γ- had different seed protein accumulation patterns than their corresponding PE5;non-RNAi kernels (Figure 3, A and B). The loss of α-zeins conditioned an increase in the accumulation of nonzein proteins in the kernel. Proteins in the range of 60–70 and 45–55 kDa were differentially upregulated in the α-zein-mutant kernels.
Total seed proteins (A) and nonzein proteins (B) from transgenic progeny kernels from crosses of PE5 with RNAi lines targeting the α-, β-, γ-, β-/γ-, or α-/γ-zeins separated in a 12% SDS-polyacrylamide gel. Kernels nonsegregating (odd-numbered lanes) and segregating (even-numbered lanes) for the RNAi transgenes were pooled and used for analysis. Arrows indicate proteins in the ∼65–70 kDa range that have increased accumulation in kernels segregating for the α- and α-/γ-zein RNAi transgenes. The identities of these protein bands (indicated by arrows) were determined by mass spectrophotometric analysis.
Two protein bands with sizes of ∼60- and 65-kDa (arrows in Figure 3A) were particularly increased in PE5;α- and PE5;α-/γ- kernels, with the latter having more accumulation of these upregulated proteins than the former. These mixtures of proteins were identified by mass spectrophotometric analysis. Spectral quantitation showed that the upper band had 52.1% (316 out of 606), and the lower band had 29.8% (909 out of 3048) of its identified peptides to be fragments of the GLB1 (globulin-1) protein. The abundance of the identified GLB-1 peptides makes it likely to be upregulated in the α-zein-reduced kernels. GLB1 has no known enzymatic function, and, just like zeins, is thought to function as a storage protein (Kriz 1989). Pulse-chase labeling and in vitro translation studies showed that the primary translation product of Glb1 undergoes at least three post- and/or cotranslational processing steps to produce the mature GLB1 protein (Kriz and Schwartz 1986). The lower band of ∼60 kDa would therefore represent the mature GLB1, whereas the upper band of ∼65 kDa would be the processing intermediate GLB1’. GLB1 is mostly an embryo-specific protein that is estimated to account for 10–20% of the total embryo protein along with GLB2 (Kriz 1989). As the deduced amino acid composition of GLB1 has 4.11% Lys (Belanger and Kriz 1989), an increased accumulation of this Lys-rich protein could contribute to an increase in the content of protein-bound Lys in the kernel. Although the 27-kDa γ-zein was increased in PE5;α- kernels, the increase in kernel Lys could not be attributed to overaccumulation of this protein as γ-zein is devoid of Lys (Table S1 in File S1).
Loss of α-zeins redistributes nitrogen, primarily stored in asparagine and glutamine in α-zeins (Table S1 in File S1), to the nonzein protein fraction by compensatory increases of proteins in this fraction. Puckett and Kriz (1991) and Hunter et al. (2002) showed that GLB1 is upregulated in o2 mutants, along with other Lys-rich proteins that contribute to an increase in kernel Lys content. Proteins that were upregulated in o2 kernels, such as GLB1 (Puckett and Kriz 1991; Hunter et al. 2002) and the Lys-rich (>8% Lys residues) glyceraldehyde-3-phosphate dehydrogenase (Damerval and Le Guilloux 1998), were also confirmed in the transgenic α-zein reduction lines (Frizzi et al. 2010).
Amino acid composition analysis of PE5;zein RNAi kernels
The o2 mutant has about twice the Lys content compared to normal phenotype (Mertz et al. 1964), and mutant o2 alleles in different backgrounds display variations in Lys contents and penetrance of the opaque phenotype (Balconi et al. 1998; Hunter et al. 2002). Of the crosses of PE5 with different RNAi lines, only PE5;α-, PE5;γ-, and PE5;α-/γ- kernels have statistically significant higher Lys content over nontransgenic A×B kernels (Figure 4A). The Lys contents in PE5;α- and PE5;α-/γ- were 60 and 128.2% higher, respectively, compared to their corresponding PE5;non-RNAi controls (Table 1). This genotype-specific variation of the Lys contents in respect to their controls suggests that hybrid genetic backgrounds have an impact on Lys accumulation. Compared to the Hi-II A×B kernels, PE5;α- kernels had 151.9% and PE5;α/γ- kernels had 83.8% more Lys. PE5;α- kernels have higher Lys content than α- kernels, suggesting that the high-Met PE5 maternal background also influences Lys accumulation in the seeds (Figure 4A). As maize proteins contain ∼4 times more Lys than Trp, either amino acid can be used as a single parameter for evaluation of protein quality (Hernandez and Bates 1969; Villegas et al. 1984).
Total seed Lys (A) and Met (B) contents. Values are means from three independent measurements of pooled samples and error bars indicate SD. The hybrid genotype Hi-II A×B was used for maize transformation while PE5 had been previously characterized (Planta et al. 2017). Statistical analysis was performed by using the one-way ANOVA with post hoc Tukey HSD test; significant differences between samples are indicated by different letters.
PE5;γ- kernels have significant increases in Met and Lys levels, and are vitreous
W64Ao2 has reduced levels of kernel Met due to reduced amounts of the 10-kDa δ- and 15-kDa β-zeins (Hunter et al. 2002). PE5;α-/γ- kernels have an increase of 26.4% more Met than its PE5;non-RNAi control (Table 1 and Table 2), though levels are not as high as in the parental PE5 introgressed into the B101 background (Figure 4B). Therefore, positive regulators of expression of the 10-kDa δ-zein in the B101 background might have been lost in the hybrid kernels. Still, PE5;α-/γ- kernels have higher levels of both Met and Lys compared to the A×B and PE5;non-RNAi controls (Table 1).
Both Met and Lys can be increased by the synergistic effect of a high-Met PE5 background and a combined α-/γ-zein reduction, with the hybrid genetic background of the kernel progenies affecting accumulation of these amino acids. Compared to other reports regarding increased Lys by transgenic zein reduction in maize (Table S2 in File S1), our combination of PE5 with reduction in the α-zeins seems to promote the highest increase in Lys content. Although the higher Met and Lys contents of PE5;α-/γ- would make it a more balanced animal feed alternative than the W64Ao2 kernels (Figure 4), its opaque phenotype (Figure 2) may preclude its general use. Therefore, the higher Met and Lys contents of the vitreous PE5;γ- would be a better alternative than the W64Ao2 kernels in terms of its nutritional quality and grain characteristics.
Knockdown of zeins by RNAi not only changes the accumulation profile of zeins (Figure 1) but also the accumulation of amino acids sequestered in the reduced zeins (Figure 5 and Table 2). PE5;α- has more changes in the kernel amino acid composition compared to PE5;γ- and PE5;β- (Figure 5, A–C). The changes in the amino acid composition of PE5;α- kernels were exacerbated by stacking it with γRNAi (Figure 5D and Table 3). Surprisingly, the loss of the γ- and β-zeins in PE5;γ-/β- kernels (Figure 5E and Table 3) did not produce a large change in amino acid composition compared to other PE5;zein RNAi kernels. The levels of seed amino acid composition in PE5;γ-/β- were like its PE5;non-RNAi control (Figure 5 and Table 3) and this might have contributed to the restoration of the vitreous phenotype in PE5;γ-/β- kernels.
Amino acid composition analysis of segregating kernels from a cross of the maternal PE5 with the zein RNAi lines. Kernels from an ear resulting from the cross and segregating or nonsegregating for the RNAi transgene(s) were pooled and used for analysis. Shown are changes in the amino acid composition of (A) PE5;α-, (B) PE5;γ-, (C) PE5;β-, (D) PE5;γ-/α-, and (E) PE5;γ-/β- relative to the corresponding controls of PE5;RNAi- kernels. Cya and MetS refer to cysteic acid and Met sulfone, respectively, the acid stable forms of cysteine and Met produced after performic acid treatment of the sample. Values are means from three independent measurements of pooled samples and error bars indicate SD. Boxes in gray denote that values for the PE5;RNAi+ kernels are statistically different from the corresponding control at P < 0.01.
Amino acid changes were the lowest for PE5;γ-/β- kernels compared to other genotypes (Figure 5 and Table 3). Of the 17 amino acids that were analyzed in PE5;α-/γ- kernels, only three amino acids did not significantly differ from the PE5;non-RNAi control, whereas PE5;β-/γ- kernels only had leucine as statistically different from normal kernels. In comparison to wheat grain, maize has a high leucine content, which contributes to its relatively poorer nutritional performance in human trials (Kies and Fox 1972). An excess of dietary l-leucine acts as an antimetabolite of isoleucine as rats fed an excess of l-leucine in a low-protein diet, or diets deficient in isoleucine, exhibited growth retardations (Harper et al. 1955). Only PE5;α- and PE5;α-/γ- kernels had significantly reduced leucine contents (Figure S3 in File S1 and Table 3). The αRNAi lines, as well as o2, also had reduced leucine contents (Segal et al. 2003; Huang et al. 2004, 2006). Zeins in general, particularly α-zeins, are exceptionally rich in leucine (Table S1 in File S1) and, therefore, reduction in α-zeins would also decrease seed leucine accumulation.
Discussion
The phenotypic plasticity of seed storage proteins in maize was exploited to generate maize kernels that have enhanced accumulation of the EAAs Lys and Met. To reduce supplementation requirements of a corn-based diet, transgenic zein reduction lines were crossed with the high-Met line PE5. As seed storage proteins mainly serve as the reservoir of nitrogen in the germinating maize seedling and not so much of specific amino acids, it has been proposed that seeds are functionally able to accommodate a wide range of variations in amino acid composition (Shotwell and Larkins 1991). These properties of the seed storage proteins would make it a good target for improving the nutritional quality of maize either through a direct manipulation of zein synthesis (e.g., by RNAi) or by increasing the amino acid supply to the kernels due to increased source strength or through a combination of both methods. Out of the five genotypes that were tested for enhanced accumulation of Met and Lys, only PE5;α-, PE5;γ-, and PE5;α-/γ- kernels had higher Lys and Met contents than the high-Lys W64Ao2. The latter, however, has Met levels that are lower than that of PE5 (Figure 4). PE5;α- and PE5;α-/γ- have opaque kernels, whereas PE5;γ- has vitreous kernels (Figure 2).
Amino acid composition of maize kernels varies widely across different genetic backgrounds. Balconi et al. (1998) and Hunter et al. (2002) reported that the opaque mutant o1 exhibits different Lys levels in different maize genetic backgrounds. The o1 mutant has a Lys content that approximates that of the normal level when the allele is in a W64A background (Hunter et al. 2002). However, it increased the Lys content of the inbred line A69Y as much as W64Ao2 (Balconi et al. 1998), indicating a differential response of these backgrounds to the same mutant allele. Met accumulation in PE5 also varies when this transgenic event is in different backgrounds. Introgression of PE5 to the high-Met inbred line B101 (in at least four generations of backcrosses) has higher Met content than when it is in an A×B hybrid background at the F3 generation (Planta et al. 2017). Results reported here refer to hybrid genetic backgrounds resulting from a combination of crosses of PE5 with the transgenic zein reduction lines (see Materials and Methods).
In QPMs, the RNAi-induced loss of the 27-kDa γ-zein abrogates the ability of the mo2s to restore kernel hardness, suggesting that it is necessary for endosperm modification (Wu et al. 2012). Although the 27-kDa gene does not require O2 for its expression (Schmidt et al. 1992), the 27-kDa γ-zein locus is linked to a QTL of mo2 in QPM (Holding et al. 2008). Gene duplication is implicated in the enhanced expression of the 27-kDa γ-zein for endosperm modification (Liu et al. 2016), whereas absence of the 27- and 50-kDa γ-zein genes in a QPM deletion mutant abolished vitreous endosperm formation (Yuan et al. 2014). This QPM deletion mutant also showed a happloinsufficient role for γ-zein in o2 endosperm modification. Alternatively, enhanced mRNA transcription or stability, rather than gene amplification, was hypothesized to be the reason for enhanced expression of the γ-zein as modified and nonmodified o2 genetic backgrounds that were studied possess one or two copies of the gene (Geetha et al. 1991). However, cleaved amplified polymorphisms exhibited no variation in 27-kDa γ-zein gene copy number in the genetic background that was used in this study (Figure S3 in File S1). Therefore, increased levels of γ-zeins in PE5;α- (Figure 1B) and αRNAi (Figure S4 in File S1) are probably due to post-transcriptional regulation of gene expression as previously described (Geetha et al. 1991).
Although the increase in 27-kDa γ-zein expression in the presence of αRNAi occurred independently of the o2 mutation, it can result in kernel modification in certain genetic backgrounds (Figure S4 in File S1). It appears that kernel modification in reduced levels of α-zeins requires at least two factors: increased accumulation of the 27-kDa γ-zein and genetic modifiers of the opaque phenotype. The modifiers in Mo17 are probably dominant as it was used as the paternal parent in a cross with the maternal αRNAi line. Different genotypes studied for inheritance of modified endosperm in o2 backgrounds exhibited a complex system of genetic control involving gene dosage effects in the triploid endosperm, cytoplasm effects, and unstable and incomplete penetrance of the modifier genes. These modifier genes have either dominant, semidominant, synergistic, or recessive modes of action (Belousov 1987; Lopes and Larkins 1991).
One explanation for the increased expression of the native 27-kDa γ-zein protein could be the selection pressure on the 27-kDa γ-zein promoter used for expression of the αRNAi cassette, in a phenomenon proposed as proxy selection (Bodnar et al. 2016). In proxy selection, the increased activity of a transgene under the control of a native promoter can enhance the protein levels of the native gene with the same promoter. One complication in the use of αRNAi lines is the use of sequences of the 27-kDa gene that drive the RNAi expression cassette, resulting in variations of γ-zein levels in kernels of the progenies of these RNAi lines. In transgenic α-zein reduction lines, gene silencing could occur or accumulation of variable levels of expression of the γ-zein due to segregation of its alleles (Segal et al. 2003; Huang et al. 2004).
The increased accumulation of nonzein proteins in both the endosperm and embryo of PE5;α- and PE5;α-/γ- kernels (Figure 3) is a response to α-zein reduction, as also evident in o2 (Hunter et al. 2002). Because some of these nonzein proteins contain some amount of Lys residues, the effective kernel Lys is increased (Kriz 2009). Because α-zeins make up >30% of the total seed proteins, their loss would also entail a major reduction in the levels of N-transport amino acids like glutamine. However, o2-converted lines showed only a minor decrease in the protein content compared with the analogous normal inbred lines (Gupta et al. 1974), implying a protein N redistribution from zein to nonzein proteins. We have found that GLB1 is likely to be upregulated in our transgenic α-zein reduction lines, similar to what was observed in o2 (Puckett and Kriz 1991).
Reduction in γ-zeins can also induce the opaque seed phenotype, albeit at a less severe degree than the loss of α-zeins. Opacity of the β/γRNAi kernels is caused by incomplete embedding of the starch granules in the outer, vitreous endosperm rather than a reduction in the vitreous area observed in αRNAi kernels (Wu and Messing 2010; Guo et al. 2013). Kernels with reduced levels of γ-zein in PE5;γ- and PE5;γ-/β-, but not in PE5;α-/γ- kernels, have the vitreous phenotype (Figure 2). This endosperm modification is probably an indirect effect of the PE5 transgenic event, where it can overcome the opaque phenotype mediated by γRNAi but not by αRNAi or α/γRNAi. Zeins confer the distinct shape to PBs and can form intra- and intercellular disulfide bonds with other proteins. α-zeins are postulated to serve as the “brick” and the γ-zeins the “mortar” in the seed during maturation and desiccation. During desiccation, the rough ER associated with the PBs breaks down, mixing the zeins with other components of the cytosol and associating directly with the starch granules. The peripheral 27-kDa γ-zein then serves as the “mortar” that bonds the starch granules in a proteinaceous matrix in the outer vitreous zone of the kernel, imparting the hard endosperm phenotype to the kernel (Chandrashekar and Mazhar 1999).
In tobacco leaves, the 10-kDa δ-zein can form novel PBs, which is unlikely the case in PE5;γ-/β- as the δ-zein has a strong interaction with the α-zeins (Bagga et al. 1997; Kim et al. 2002). It is more likely that the 15-kDa β-zein has a redundant function with the 27-kDa γ-zein in terms of embedding the starch granules in a proteinaceous matrix because of its cysteine content. We have previously shown that PE5 increases expression of cysteine-rich nonzein proteins (Planta et al. 2017). It is possible that one of these proteins is an accessory protein that associates with PBs and promotes its structural integrity.
Although QPMs have been proven to be effective, the complexity of introducing multiple, unlinked loci of mo2s into a defined o2 background has slowed the creation and widespread use of QPMs (Gibbon and Larkins 2005). Here we report an alternative strategy to generate QPM with an additional high-Met content without a reduction of α-zeins. PE5;γ- kernels could be generated in one generation of crossing, whereas QPMs entail generations of backcrossing the o2 mutant allele into a desirable germplasm and subsequent backcrosses of the o2-converted germplasm with the mo2s. It would take ∼17 generations to convert a desirable germplasm into a QPM (Wu and Messing 2011). Even if both the γRNAi and PE5 lines are introgressed into a desirable germplasm, the eight generations it would take to make an introgression line with both the PE5 and γRNAi transgenes are still about half the time it takes to generate a classical QPM.
Acknowledgments
This research was supported by the Selman A. Waksman Chair in Molecular Genetics of Rutgers University to J. M.
Footnotes
Supplemental material is available online at www.genetics.org/lookup/suppl/doi:10.1534/genetics.117.300288/-/DC1.
Communicating editor: J. Birchler
- Received September 13, 2017.
- Accepted October 17, 2017.
- Copyright © 2017 by the Genetics Society of America