Increasing Lysine Content of Waxy Maize through Introgression of Opaque-2 and Opaque-16 Genes Using Molecular Assisted and Biochemical Development

The low lysine content of waxy maize cannot meet the nutritional requirements of humans, livestock, or poultry. In the present study, the high-lysine genes o2 and o16 were backcrossed into wx lines using the maize high-lysine inbreds TAIXI19 (o2o2) and QCL3021 (o16o16) as donors and the waxy maize inbred line QCL5019 (wxwx) as a receptor. In the triple-cross F1, backcross, and inbred generations, the SSR markers phi027 and phi112 within the wx and o2 genes and the SSR marker umc1121 linked to the o16 gene were used for foreground selection. Background selection of the whole-genome SSR markers was performed for the selected individuals. The grain lysine content was determined using the dye-binding lysine method. The waxiness of the grain was determined with the I2-KI staining and dual-wavelength spectrophotometric analysis. The BC2F2 generation included 7 plants of genotype wxwxo2o2O16_, 19 plants of genotype wxwxo16o16O2_, and 3 plants of genotype wxwxo2o2o16o16. In these seeds, the average amylopectin content was 96.67%, 96.87%, and 96.62%, respectively, which is similar to that of QCL5019. The average lysine content was 0.555%, 0.380%, and 0.616%, respectively, representing increases of 75.1%, 19.9%, 94.3%, respectively, over QCL5019. The average genetic background recovery rate of the BC2F3 families was 95.3%, 94.3%, 94.2%, respectively. Among these 3 wxwxo2o2O16O16 families, 4 wxwxo2o2O16o16 families, and 3 wxwxo2o2o16o16 families, the longest imported parent donor fragment was 113.35 cM and the shortest fragment was 11.75 cM. No significant differences in lysine content were found between the BC2F4 seeds and the BC2F3 seeds in these 10 families. This allowed us to increase the lysine content of waxy corn and produce seeds with excellent nutritional characteristics suitable for human consumption, animal feed, and food processing. This may be of significance in the breeding of high-quality corn and in improvement of the nutrition of humans, livestock, and poultry.


Introduction
Waxy maize (Zea mays L. sinensis Kulesh), also known as sticky maize, is one of nine sub-types of maize, first found in China and later found in other regions in Asia [1,2]. In 1909, Collins published an accurate description of waxy maize [3]. The endosperm of the dried grain is opaque with a dull, waxy appearance. In 1922, Weatherwax found the waxy corn starch to be completely composed of branched, small-molecular-weight amylopectin [4]. In 1935, Emersonk and colleagues mapped the wx gene in the long arm of chromosome 9, i.e., the 59 locus close to the centromere [5]. In 1943, Sprague discovered that the maize wx mutant lacks amylose [6]. The major mutations in waxy maize are insertion mutation, deletion mutation, and EMS mutagenesis [7][8][9][10]. These mutations cause splicing errors and translation errors in pre-mRNA so that the Wx gene is not normally expressed. The Wx gene encodes granule-bound starch synthase I (GBSS-I), which determines the amylose synthesis in maize endosperm and pollen [11]. Starch in the grains of normal corn (WxWx) was found to be composed of amylose (25%) and amylopectin (75%). The GBSS-I activity of the wx mutant decreased by 5% to 95%, resulting in lower amylose content in grain and waxy corns with various levels of amylose. Meng argued that the amylose content was less than 5% in waxy maize carrying the wx-a gene [12]. Zhang and colleagues suggested that the presence of the wx gene indicated that the amylose content would be between 0 and 5%, that the du gene indicated that amylose content would be between 5% and 15%, and that the ae gene indicated the amylose content would exceed 15% [13]. Sun and colleagues suggested that Wx was incompletely dominant to wx and that a dose effect was present between the amylopectin content and the endosperm wx gene [14]. Liu and Li indicated that it was difficult to achieve nearly 100% of amylopectin in waxy corn [15].
The Wx gene was first cloned and sequenced in 1986 [16]. This gene has a single copy in the maize genome with a 3.8 kb coding sequence of 14 exons and 13 introns [17]. The start codon is located in exon 2 and the stop codon is located in exon 14. These data laid the foundation for the research and application of the Wx gene, including the development of molecular markers within the gene in marker-assisted selection (MAS). The MaizeGDB website has published three SSR markers for the detection of the Wx loci: Phi022, phi027 and phi061.
MAS can shorten the recessive gene transfer from generation to generation, accurately identify target genes, and be not subject to the influence of identification conditions and heterofertilization of the seed endosperm [18]. In recent years, MAS has been used successfully in the selection of crops resistant to insect pests and drought and in the improvement of crop quality using single gene selection, polymerization of multiple genes resistant to the same disease, polymerization of multiple genes resistant to different diseases, and polymerization of resistance genes and other genes [19][20][21][22][23][24][25][26][27].
The level and types of amino acids found in maize grain, especially essential amino acids, is an important indicator of nutritional quality [28]. Generally, the humans should take in 51 mg lysine per gram of protein [29]. This requires the lysine content be more than 0.5% in maize grain. Livestock and poultry feed must be 0.6-0.8% lysine [30]. Waxy maize has excellent taste, texture, and other culinary qualities, but its nutritional value is relatively low. A survey of 93 samples of waxy corn grown in China's Yunnan Province found them to have a lysine content of 0.24-0.34%. A survey of 40 temperate waxy corns, showed the lysine content to be 0.14-0.39% [31]. The current opaque-2 (o2) maize grain contains circa 0.4% lysine, which does not meet standards for either food or fodder. However, the gene pyramiding of the opaque-16 (o16) and o2 genes has been found to significantly increase lysine content [32,33].
The main purpose of this study was to improve the nutritional quality of waxy corn by backcrossing the two high-lysine genes o2 and o16 into wx maize line using the multi-gene MAS combined with biochemical techniques, to produce waxy seeds with high lysine content, and to promote high-quality corn breeding and development of relevant industries.

Parent Materials and Population Construction
TAIXI19 is an inbred line of o2 maize, the seed lysine content of which is about 0.43%. QCL3021 and QCL5019 are inbred lines of o16 maize and waxy maize, the seed lysine content of which are 0.32% and 0.28%, respectively. The methods used for analysis of lysine content are as follows.
The 350 kernels seeds of three F 1 hybrid combinations were generated using TAIXI19 as the female parent and QCL3021 as the male parent in the field. The 400 kernels seeds of two triple hybrid populations were generated using the F 1 hybrid as the female parent and QCL5019 as the male parent in the field. In the triple hybrid F 1 generation, the 400 kernels seeds were sown in the field and 375 plants emerged; the 83 target individual plants,  double heterozygous at the o2 and o16 loci, were selected using foreground selection and used for backcross with recurrent parent QCL5019; 72 plants of them were harvested, and the 240 kernels seeds from the two plants, G31 and G167, were selected. In the BC 1 F 1 , the 240 kernels seeds were sowed in the field and 237 plants emerged; the 20 target plants with genotype of wxwxO2o2O16o16 were selected using foreground selection and used for backcross with recurrent parent QCL5019; 14 plants of them were harvested, and the 220 kernels seeds from two plants, G31-101 and G167-181, were selected after background selection and quality analysis. In the BC 2 F 1 , the 220 kernels seeds were sowed in the field, and were reserved after quality analysis. In the BC 2 F 3 generation, these 10 families were grown by row in the field and their genotypes were verified using molecular markers; background analysis was performed for the whole genome; and all of families were inbred to produce the BC 2 F 4 seeds.

DNA Extraction, PCR Amplification, and Electrophoresis
Young, seedling-stage leaves were collected for extraction of genomic DNA of individual plants of parents and each generation using the CTAB method for corn MAS [34]. PCR amplification and electrophoresis detection of amplification products was performed as reported previously [32,35]. PCR amplification was performed using a 2720 Thermal Cycler (Applied Biosystems, Foster City, CA, USA) and a DNA Engine Peltier Thermal Cycler (Bio-Rad, Hercules, CA, USA). Amplification products were separated using a Sequi-GenH GT DNA electrophoresis system (Bio-Rad).

Seed Lysine and Starch Content Measurement
Seed lysine content was measured using acid orange-12 dyebinding lysine colorimetry (DBL) [35]. Each sample was measured 2 or 3 times and the measurements were averaged. Seed waxiness was qualitatively and quantitatively determined using I 2 -KI staining and dual wavelength spectrophotometry (DWLS), respectively [36,37]. For quantitative determination, the absorption spectra of amylose and amylopectin were scanned using SPECORD 40 (Analytik Jena AG, Jena, Germany). Three repeated measurements were performed and averaged.

Foreground Selection and Background Selection
Foreground selection (FS). FS refers to selection of the target genes o2, o16, and wx. The o2 gene was detected using the SSR markers phil12, umc1066, and phi057 within the gene [32]. The o16 gene was detected using the linked SSR markers umc1141 and umc1121 [32]. The wx gene was detected using the SSR markers phi022, phi027 and phi061 [38]. Background selection (BS). BS refers to selection of the genetic background of the FS-selected individuals. Parental polymorphic SSR markers from genome-wide screening were used for BS. Polymorphic markers in the BS were divided into two categories. The first was a class of markers found to be polymorphic among the three parents. The second was a class of markers found to be polymorphic between the recurrent parent and the other donor parents but not between the two donor parents.
The PCR amplification primer sequences for the SSR markers in FS and BS were adopted from the Maizegdb website (http:// www.maizegdb.org) and synthesized by Shanghai Generay Biotech Co., Ltd (Shanghai, China).

Statistical Analysis
Electrophoresis band patterns A, B, H, and U of the SSR markers were used to establish the database. In the same migration position, a band pattern consistent with the recurrent parent was recorded as A, while a band pattern consistent with the donor parent was recorded as B. A heterozygous band pattern was recorded as H and an unidentified band pattern was recorded as U. Based on the statistical analysis of genetic background recovery rate of molecular markers, the formula G (g) = [L+X (g)]/(2L) was used to calculate the background recovery rate of the FS-selected individuals after BS. Here, G (g) indicates the genetic background recovery rate in the backcross g-generation, X (g) the number of molecular markers with the band pattern of receptor parent in the backcross g-generation, and L the number of molecular markers included in the analysis [39][40][41]. The theoretical genetic background recovery rate was calculated using the formula E [G (g)] = 12(1/2) g+1 , where g refers to the number of backcross generations.
Analysis of variance and calculation of standard deviation were performed using SPSS13.0 software. The absorption spectra of amylose and amylopectin were plotted using Origin7.5 software.  Graphical genotypes were analyzed and illustrated using GGT32 software with reference to an IBM2 2008 Neighbors Map.

Polymorphism of SSR Markers at Target Loci and Whole Genome between the Parents
As shown in Figure 1, among 3 markers (umc1066, phi057, and phil12) of the o2 gene, two marker loci (phi057, and phil12) were found to be polymorphic between TAIXI19 and the other two parents. Among the two markers in the o16 gene, the umc1121 locus showed polymorphism between QCL3021 with the other two parents. All 3 markers (phi027, phi061, and phi022) of the wx gene showed polymorphism between QCL5019 and the other two parents. These polymorphic markers were codominant, which rendered them usable for the MAS of the corresponding target genes. In the present study, the markers phil12, umc1121, and phi027 were selected for FS.
Two hundred and sixty-six SSR markers distributed on the 10 chromosomes of the maize genome were selected for the screening of polymorphisms between the three parents. Of these markers, 49 were found to be polymorphic between QCL5019 and the other two parents, and 33 markers were found to be polymorphic among the three parents. A total of 82 markers were used for BS, with an overall polymorphism ratio of 30.8% (Table 1).

Foreground Selection of the Target Genes in Various Segregating Generations
Because phil12 and phi027 served as markers within the target gene and QCL5019 was the recurrent parent, in every segregating generation of the triple-cross F 1 , BC 1 F 1 , and BC 2 F 1 , the wx locus of every individual was detected first, followed by the o2 locus of wx-selected individuals and the o16 locus of individuals selected from the wx and o2 loci. There were 83, 20, and 41 FS-selected individuals in the triple-cross F 1 , BC 1 F 1 , and BC 2 F 1 generations, and 72, 14 and 30 plants were harvested from each group ( Table 2).
In the BC 2 F 2 generation, seeds from 6 outstanding BC 2 F 1 plants were selected for planting, and FS was performed for the three target loci. Among 285 individuals, 12 were selected from the wxwxo2o2O16_ genotype, 35 from the wxwxo16o16O2_genotype, and 6 from the wxwxo2o2o16o16 genotype. As shown in Table 3, seven, nineteen, and three plants were harvested from these three groups. In the BC 2 F 3 generation, the double-recessive and triplerecessive gene pyramiding families obtained from the last generation were planted continuously. One row was planted for each family, and molecular markers at the wx, o2, and o16 gene loci were detected and verified. Finally, in the BC 2 F 4 generation, 3 wxwxo2o2O16O16 families, 4 wxwxo2o2O16o16 families, and 3 wxwxo2o2o16o16 families were produced.

Selection of Genetic Background Molecular Markers in Various Segregating Generations
In the BC 1 F 1 generation, the genetic background recovery rate of selected individuals was 73.8-86.6% with an average of 81.8%. This was 6.8% higher than the theoretical value. Two individuals with a recovery rate of 84.8% were selected for backcrossing. In the BC 2 F 1 generation, the genetic background recovery rate of selected individuals was 85.9-92.7%, with an average of 90.42%. This was 2.92% higher than the theoretical value. The genetic  background recovery rate of all the six plants selected from the BC 2 F 1 generation was higher than 87.5%. The genetic background selection was not conducted in the BC 2 F 2 , and all the seven wxwxo2o2O16_ plants and the three wxwxo2o2o16o16 plants were chosen. The genetic background recovery rate of the 10 preferred families in the BC 2 F 3 generation ranged from 93.4% to 96.3% (Table 4).
The amount of donor parent genome in the 10 families of the BC 2 F 3 generation was between 1.8% and 4.8%. The amount of heterozygote genome was 0-2.4%, and the amount of unidentified genome was 0-1.2%. Among the 10 families, families 67 and 120 had the highest genetic background recovery rate, and family 122 had the lowest recovery rate (Table 4). Family 122 also carried the longest donor fragment, which was 535.4 cM in length, and family  67 carried the shortest donor fragment, which was 200.75 cM in length. Among all families, the longest fragment imported from the donor parents was 113.35 cM, and it was located on chromosome 3. The shortest fragment imported from the donor parents was 11.75 cM, and it was located on chromosome 1.

Graphical Genotype Analysis of the Chromosome of the Target Gene
The genetic background recovery rate on chromosome 7 in 10 families of the BC 2 F 3 generation was 87.5-93.8%. The relative amount of donor parent fragment was 0-6.25%, and the proportion of heterozygous fragment was 0-12.5%. The genetic background recovery rate on chromosome 8 was 90.9-95.5%. The proportion of the donor parent fragment was 0-4.5%, and the proportion of heterozygous fragment was 0-9.1%. The genetic background recovery rate on chromosome 9 was 95.0-100%. Except the individuals of 3 families (122, 240, and 261) had an unidentified fragment, the recovery rate of all the other individuals approached 100% (Figure 2).
Family 240 had the shortest foreign fragment imported on chromosome 7, families 120, 134, and 135 shared the shortest donor parent fragment, and family 240 had no imported heterozygous fragment. Family 100 had the shortest foreign fragment imported on chromosome 8, families 67, 100, 120, 134, and 135 had the shortest donor parent fragment, and families 62, 100, 122, 240, and 261 had no imported heterozygous fragment. Any imported foreign fragment was not found on chromosome 9, only 3 families contained an unidentified fragment (Figure 2).

Analysis of the Donor Allele in Ten Preferred Families
The imported donor parent genomes of 10 preferred families were divided into three types: B1 -consistent with the alleles of donor parent TAIXI19; B2 -consistent with the alleles of donor parent QCL3021; and B3 -consistent with the alleles of both donor parents. B1 made up 0.6-2.4% of the total, B2 0-1.8%, and B3 0-1.2% (Table 5).

Lysine Content in Various Generations
The lysine content was 0.261-0.337% in 72 seeds of the BC 1 F 1 generation, 0.278-0.362% in 14 seeds of the BC 2 F 1 generation, and 0.346-0.549% in 30 seeds of the BC 2 F 2 generation. In the BC 2 F 3 generation, 171 seeds were harvested and 169 seeds were measured for lysine content (see discussion section for usage of the other 2 seeds). Analysis of variance showed that the lysine content was significantly different between different genotypes (P,0.01). These were, from highest to lowest, wxwxo2o2o16o16. wxwxo2o2O16_.wxwxO2_o16o16. wxwxO2_O16_. They had an average lysine content of 0.616%, 0.555%, 0.323%, and 0.282% respectively. These values were 94.3%, 75.1%, and 19.9% higher, respectively, than those of wxwx parent (Table 6). This indicates that the triple-recessive gene pyramiding families and doublerecessive gene pyramiding families have significant positive interaction effects and that the regulatory role of the o2 gene is greater than that of the o16 gene. The lysine content ranged from 0.505% to 0.639% in 3 wxwxo16o16o2o2 families and 7 wxwxO16_o2o2 families; 59.3-101.6% higher than the recurrent parent; 8.1-36.8% higher than the high-value parent o2 line; and 38.7-75.5% higher than the low-value parent o16 line (Table 7). No significant difference in lysine content was found between BC 2 F 4 seeds and BC 2 F 3 seeds (P.0.05) ( Table 7 and 8), suggesting that the lysine content tends to stabilize.

Qualitative and Quantitative Determination of Starch Content in Various Generations
The BC 1 F 1 , BC 2 F 1 , and BC 2 F 2 seeds selected by FS from the triple-cross F 1 , BC 1 F 1 , and BC 2 F 1 generations were qualitatively identified using I 2 -KI staining. Seeds whose endosperms were stained umber were selected. The DWLS method was used to quantitatively determine the levels of amylose and amylopectin in seeds of the 10 selected families. Amylopectin made up 96.26-97.06% of the total starch content. This is similar to the 96.84% observed in QCL5019. The total starch content was 53.94-56.17%, which was lower than 67.62% in QCL5019 (P,0.01) ( Table 9).

Discussion
In the present study, MAS technology was used to produce 3 wxwxo2o2o16o16 families. In these families, average lysine content was found to reach 0.616% and the amylopectin content was found to reach 96.62%. The lysine content in these seeds, which tend to be waxy, has met the needs of people, livestock, and poultry. These seeds are of some importance in the genetic improvement and breeding of special types of corn.
Recessive genes have their own specific genetic effects [42]. The interactions within the double-recessive or triple-recessive mutations formed by gene pyramiding can affect the quantity and quality of starch, sugar, and protein in the endosperm [43]. This can affect seed emergence, seedling growth, and flowering time agreement, causing a low level of FS selection and a reduced harvest after pollination. This increases the difficulty of selection and necessitates a large population for selection. In the present study, 340 kernels of seeds from 6 families were planted in the BC 2 F 2 generation. Of these seeds, 285 emerged. Then 6 wxwxo2o2o16o16 plants, 12 wxwxo2o2O16_ plants, and 35 wxwxo16o16O2_ plants were selected using FS. Finally, and 3, 7, and 19 plants were harvested from each of the three genotypic families.
Li and Liu argued that the double-recessive mutations formed by o2 and wx were not associated with significant changes in the total starch content of the grain [44]. However, in the present study, the total starch content of the seeds containing the o2 gene with double-recessive (wxwxO16_o2o2) and triple-recessive (wxwxo16o16o2o2) mutations were 55.22% and 54.33%, respectively, significantly less than the recurrent parent QCL5019 (67.62%, P,0.01). This may be related to the differences in genetic background or to the hybrid model.
In the present study, the BC 2 F 3 seeds of BC 2 F 2 plants with genotypes wxwxO16_ O2_ and wxwxo16o16O2_ were plump and smooth, but the BC 2 F 3 seeds of BC 2 F 2 plants with genotypes wxwxO16_o2o2 and wxwxo16o16o2o2 were depressed and wrinkled. Two BC 2 F 3 seeds from BC 2 F 2 plants with the genotype wxwxo16o16O2o2 were selected from 19 BC 2 F 3 seeds from the BC 2 F 2 generation with the genotype wxwxo16o16O2_ (No. G75 and G16, Figure 3). Of these, 240 seeds were smooth and 70 seeds were wrinkled, with an indoor germination emergence of 205 and 45 plants, respectively. Using phi112 detection, the wxwxo16o16O2o2 and wxwxo16o16O2O2 genotypes accounted for 97.6% of all smooth seeds, and 100% of shrunken seeds had the wxwxo16o16o2o2 genotype. The same detection process was applied to two BC 2 F 4 seeds from BC 2 F 3 plants with the genotype wxwxo16o16O2o2. Results showed the phenotype and genotype concordance rates to be 97.1% and 100%, respectively (Table 10). These findings indicate that the interactions between the o2 and wx genes in the endosperm cause the grain endosperm to shrink. The BC 2 F 1 seeds were obtained through a backcross with plants from the BC 1 F 1 generation of the genotype wxwxO16o16O2o2. Pale yellow seeds with high lysine content (G89) were phenotypically selected for three seasons of continuous self-breeding and the pyramiding yellow grain was harvested. The genotypes of these yellow grains were the same as those of the above listed white grains. In this way, the goal of selection was reached through marker-assisted selection of the early generations combined with phenotypic selection of the subsequent generations. This may also reduce the cost of experiments.
In the 10 preferred families, except that 4 wxwxo2o2O16o16 families (G100, G134, G240 and G263) need to be purified at O16 locus, 3 wxwxo2o2O16O16 families (G62, G122 and G261) and 3 wxwxo2o2o16o16 families (G67, G120 and G135) can be directly used in breeding programs because their recurrent parent QCL5019 has good combining ability, and used for pyramiding more other good traits.