Identification and epistasis analysis of quantitative trait loci for zeaxanthin concentration in maize kernel across different generations and environments

Abstract

Zeaxanthin, a natural fat-soluble pigment, not only increases plant resistance, but also has vital significance for human health. However, quantitative trait loci (QTL) and the epistatic effects of zeaxanthin concentration in maize kernel have not been well studied. To identify QTLs and analyse the epistatic effects of zeaxanthin concentration in maize kernel, two sets of segregating generations derived from the cross between HuangC (a high zeaxanthin concentration inbred line) and Rezi1 (a low zeaxanthin concentration inbred line) were evaluated in three different environments. One major-effect QTL, qZea6a, explains 41.4–71.4% of the phenotypic variation and two QTLs, qZea4a and qZea3a, show LOD > 3 for zeaxanthin concentration detected over two generations and three different environments. Four of the ten QTL pairs show epistatic effects, explaining 7.34–14.3% of the phenotypic variance. Furthermore, additivity was the major allelic action at zeaxanthin concentration QTLs located in F₂ and F_2:3 populations and plants with homozygous HuangC alleles have a strong genetic ability in enhancing zeaxanthin concentration in maize kernel. These results will contribute to understanding these complex loci better and provide awareness about zeaxanthin concentration to maize breeders and scientists involved in maize research.

Introduction

Maize is one of the most important crops in the world, and is rich in protein, fat, starch, carotene, nicotinic acid, and vitamins B1, B2, and E. Orange maize is an essential staple cereal crop that naturally accumulates carotenoids in the edible seed endosperm, particularly zeaxanthin, accounting for about 0.1–9.0 mg/kg maize kernel (Sajilata et al. 2008). Zeaxanthin is a natural fat-soluble pigment. As an important part of the xanthophyll cycle, zeaxanthin has strong antioxidant, drought resistant, and high-salt tolerant abilities (Wu et al. 2015). Additionally, zeaxanthin can prevent the oxidation of lipids, vitamins and nutrients in foods, as well as maintain the flavours, and extend the shelf life of foods (Sajilata et al. 2008). Zeaxanthin is distributed in the eyes, kidneys, and liver of humans and animals. It plays an important role in preventing age-related maculopathy, and cataract, helps reduce the risk of cancer, and boosts immunity (Loane et al. 2008, Moeller et al. 2000, Nilsson et al. 2003, Snodderly 1995). However, since humans and animals cannot synthesise zeaxanthin de novo, they must obtain it through food (Sengun et al. 2014). Orange maize kernel has a high content of zeaxanthin while its concentration is extremely low in rice or green vegetables (Humphries and Khachik 2003, Perry et al. 2009). O’Hare et al. (2015) reported that the content of zeaxanthin in orange maize was significantly higher than that in yellow maize. Therefore, it is important to breed the commercial maize varieties with high zeaxanthin concentration, due to its benefit for human health.

Studies related to zeaxanthin have also been reported in Arabidopsis, tobacco and maize. β-carotene hydroxylase (chyb) is the key rate-limiting enzyme in the biosynthesis of zeaxanthin (Kato et al. 2004, Vallabhaneni et al. 2009). Davison et al. (2002) constitutively over-expressed the β-carotene hydroxylase gene in Arabidopsis, resulting in a 2-fold increase in zeaxanthin content in the lutein cycle. Moreover, overexpression of the β-carotene hydroxylase (chyb) gene significantly enhanced zeaxanthin in production tobacco, which dramatically increased the flux of the xanthophyll cycle (Wu et al. 2015). Vallabhaneni et al. (2009) showed that zeaxanthin content positively correlated with the amount of Hydroxylase3 (HYD3) transcript present, and also found that only HYD3 was significantly associated with the β-carotene gene when compared to all carotene hydroxylase genes following analyses of their transcriptional levels. Yan et al. (2010) found that the gene ZmcrtRB1 was mainly associated with zeaxanthin in the maize endosperm.

To date, there have been only a few reports on QTLs of zeaxanthin in maize. Wong et al. (2004) identified some QTLs of zeaxanthin by using two sets of segregating generations—a set of F_2:3 lines derived from a cross betweenW64a and A632, and their testcross progeny with AE335. The results showed that three QTLs related to zeaxanthin were detected on chromosomes 6, 7, and 8, respectively and explained 6.5–19.8% of the phenotypic variation. Chander et al. (2008) selected 233 Recombinant Inbred Line (RIL) populations derived from the cross between By804 and B73 to construct genetic linkage maps, and the QTLs associated with zeaxanthin were identified on chromosomes 1, 6, 8, and 10, with 3.8–16.6% phenotypic contribution. These results showed many differences in QTL regions and contribution rates. Moreover, epistases between the QTLs were not analysed. Over-reliance on a single major QTL may lead to regardlessness important biosynthetic steps for which gene expression is controlled at interaction effect between different loci (Burt et al. 2011). Moreover, the possibility of combining epistatic effects with additive allelic effects from biparental populations for the creation of superior phenotypes in crops has been suggested (Latta et al. 2010). In the present study, four datasets from the F₂ and F_2:3 populations as well as three different environments were used. The objectives were (i) to detect the QTLs of zeaxanthin concentration in maize kernel and (ii) to analyse epistatic effects between the QTLs. We expect to identify the consistent QTLs among different generations and environments and elucidate the epistatic effects between them. This information will provide better understanding of the genetics behind zeaxanthin production and help the maize breeders and scientists involved in maize research to establish better commercial candidates.

Materials and Methods

Population construction

Based on the results of our previous study (Deng et al. 2015), a high zeaxanthin concentration inbred line HuangC (1.91 mg*kg^-1) and a low zeaxanthin concentration inbred line Rezi1 (0.10 mg*kg^-1) were selected as parents for generating the cross (Fig. 1), HuangC × Rezi1. All the inbred lines were provided by the Maize Research Institute of Southwest University (China). The original cross, HuangC × Rezi1, was produced at Beibei (BB), Chongqing (China) in the spring of 2015. Ten randomly selected F₁ ears were planted in the autumn of 2015 at Yuanjiang (YJ), Yunnan (China) to produce F₂ seeds. During the spring of 2016, F₂ ears were planted at Beibei, Chongqing and were self-pollinated to produce F_2:3 seeds.

Fig. 1.

The kernel phenotypes of HuangC and Rezi1 in maize. a: HuangC; b: Rezi1.

Field experiment

The F₂ generation and the two parental lines were planted at BB (29.45°N, 106.23°E, 249 m above sea level, loam soil) in the spring of 2016. The F_2:3 generations and the parental lines were planted at YJ (23.59°N, 102°E, 394 m above sea level, dry-red soil) in the autumn of 2016 while in BB and Hechuan (HC) (30.02°N, 106.15°E, 240 m above sea level, sand-loam soil), they were planted in the spring of 2017. There were a total of 286 plants in the F₂ generation and 242 families in the F_2:3 generation. The experimental design for the F_2:3 generation included a random complete block design in triplicates and 20 plants in each plot. All plants were self-pollinated to produce F_2:3 kernels.

Zeaxanthin concentration determination

Dry kernels from the F₂ and F_2:3 populations were ground into a powder for each ear and sieved with an 80-mesh sieve to obtain test samples. Concentration of zeaxanthin was determined using a spectrophotometric method with slight modifications (Deng et al. 2015, Xiu-Hong et al. 2013). Briefly, this included the following steps: (1) 5 mL of n-hexane: acetone (3:2, V/V) was added to a test tube containing 0.2000 ± 0.0001 g maize powder and was mixed well in the dark. (2) Samples were then incubated for 16 h at 37°C and left undisturbed for 1 h. (3) The supernatant was directly injected into a new tube, followed by the addition of 1 mL of 20% KOH-methanol solution for saponification reaction and then incubated for 1 h. (4) N-hexane (3 mL) and 10% sulphate solution (1 mL) were added to the tube for layering. (5) The supernatant was centrifuged (6) The absorbance of the centrifuged liquid was determined at 445 nm using an Ultraviolet-Visible Spectrophotometer (Shanghai Tianmei Scientific Instrument). The entire process was conducted in a dark environment. The zeaxanthin concentration was determined using the following formula: Zeaxanthin (mg*kg^–1) = OD₄₄₅ × V/(W × N × f); OD₄₄₅ is the absorption value of the sample at the wave length of 445 nm, V represents the total volume of the sample used, W is the mass of the sample used, N is the dilution factor of the sample, and f is the specific extinction coefficient, which is 2,589 for zeaxanthin in n-hexane.

Statistical analyses

Significance tests of zeaxanthin concentration between the parents were performed using Student’s t test. Variation analysis, skewness test and kurtosis test of zeaxanthin concentration distribution were conducted using SPSS Statistics 19.0 (IBM, 2010). Variances were estimated based on the general linear model: χ_ijk = μ + B_jk + G_i + E_j + (GE)_ij + ε_ijk, where B_jk, G_i, E_j, (GE)_ij, and ε_ijk are the block effect (within the environment), genotype effect, environment effect, genotype and environment interaction effect and experimental error, respectively. All these effects are random. All calculations were performed using SPSS Statistics 19.0 (IBM, 2010) and Microsoft Office Excel 2016. The broad heritability of zeaxanthin concentration in F_2:3 was calculated using the formula: h² = σ²_G/(σ²_G + σ²_GE/n + σ²_e/nr), where σ²_G, σ²_GE, and σ²_e are estimates of genotype variance, interaction variance between the genotype and the environment, and experimental error variance, respectively. n and r represents the number of environments and replicates, respectively (Knapp et al. 1985).

Linkage map construction

DNA was extracted from the leaves of F₂ plants and the parental plants using the cetyltrimethyl ammonium bromide (CTAB) method (Causse et al. 1994, Murray and Thompson 1980). Simple sequence repeats (SSR) were used for molecular marker mapping (Senior et al. 1996). Amplified products were separated on a 6% polyacrylamide gel electrophoresis (PAGE) and visualized by silver-staining (Xu et al. 2002). Two hundred and forty-three polymorphism markers out of 1,331 SSR markers covering the entire maize genome (http://www.maizegdb.org) were applied to construct a linkage map using Join map version 4.1 (Ooijen 2011). The recombinant frequency between the linked loci was converted into genetic distance (centi-Morgan, cM) using Kosambi’s function (Kosambi 1943).

QTL analysis

The procedure of interval mapping was used to identify QTLs and estimate their effects on zeaxanthin concentration (Li et al. 2007, 2008, Wang 2009). QTL mapping was performed using the integrated software Map QTL 6.0 (https://www.kyazma.nl/). QTLs with similar marker intervals, identified from different environments for zeaxanthin concentration (overlapping-LOD confidence intervals provided by the software), were assumed to be the same. Parameters for forward regression analysis were set with a window size of 10 cM and a walk speed of 1 cM. Significance threshold for QTL detection was calculated by 1,000 random permutations of the phenotypic data at 5% level. Gene action mode (d/a) of each QTL was estimated according to the ratio (|d/a|) of the additive (a) and the dominant (d) effects. The data were classified as additive (A; 0–0.20), partially dominant (PD; 0.21–0.80), dominant (D; 0.81–1.20), and over-dominant (OD > 1.20) (Stuber et al. 1987, Tuberosa et al. 1998). The nomenclature of the QTLs were as follows: in qZea3a-F₂, ‘q’ indicates QTL, ‘Zea’ is an abbreviation of the trait, namely zeaxanthin, the number ‘3’ represents the chromosome number, ‘a’ represents the order in which the QTL was detected, and ‘F₂’ indicates the F₂ generation; for qZea4a-YJ, ‘YJ’ represents the environment in which the QTL was identified. According to the constructed genetic linkage map and the analysed QTL results, Map Chart version 2.2 (Voorrips 2002) software was used to label the QTL linkage map.

Allelic and epistatic effect analyses

The segregation ratio among nine genotypes was examined using χ² analysis (a = 0.05). Multiple comparisons for zeaxanthin concentration in kernels among genotypes were conducted using SPSS Statistics 19.0 (IBM, 2010). QTL IciMapping, which was developed based on the Inclusive Composite Interval Mapping (ICIM) method and provided intuitive statistics for evaluating epistatic effects between the pairs of QTLs (Meng et al. 2015), was used to analyse the epistatic effects between the QTLs in the F_2:3 population.

Results

Phenotypic variation of zeaxanthin concentration

The phenotypic results for zeaxanthin concentration are shown in Table 1. The two parents, HuangC and Rezi1, showed significant differences in zeaxanthin concentration among the two generations and the three different sites (P < 0.01), which are indicative of the genetic differences between the two parents. A wide range of variation in zeaxanthin concentration is indicative of transgressive segregation and polygenic mode of in heritance.

Table 1. Phenotypic values of zeaxanthin content (mg*kg^–1 seed dry weight) of parents, F₂ and F_2:3 populations in three environments

Population	Trait	Sites	Parent line		Items
			P1	P2	Mean	Range	Std error of mean	Skewness	Std error of skewness	Kurtosis	Std error of Kurtosis
F2	Zeaxanthin	BB	2.1 ± 0.034**	0.11 ± 0.025	1.06	0.02–3.23	0.04	0.38	0.15	–0.03	0.29
F2:3		YJ	1.93 ± 0.031**	0.37 ± 0.015	1.25	0.02–3.11	0.05	0.26	0.16	–0.78	0.32
		BB	1.92 ± 0.013**	0.13 ± 0.013	0.92	0.02–3.36	0.04	0.77	0.16	0.39	0.31
		HC	1.7 ± 0.023**	0.13 ± 0.035	1.20	0.14–3.6	0.04	0.44	0.16	–0.25	0.31

** means significant at the 0.01 probability level.

Heredity of zeaxanthin concentration in the F_2:3 population

From the analysis of genetic variance for zeaxanthin in the F_2:3 population (Table 2), the values of σ²_G, σ²_GE, and σ²_e were 0.1597, 0.2134, and 0.1018, respectively. The broad heritability of zeaxanthin was 65.96%, indicating that zeaxanthin was hereditable.

Table 2. Variance components and heritability for zeaxanthin in the F_2:3 population

Items	Zeaxanthin
Genotype (σ2G)	0.1597
G × E (σ2GE)	0.2134
Error (σ2e)	0.1018
Heritability (h2)	65.96%

σ²_G, σ²_GE and σ²_e were estimates of genotype variance, interaction variance between genotype and environment, experimental error variance, respectively.

Genetic linkage map of the F₂ population

A total of 243 SSR markers were used for constructing a genetic linkage map, and all the SSR loci were mapped onto ten linkage groups covering 2609.8 cM of the entire genome with an average interval of 10.73 cM between the adjacent markers (Fig. 2). The markers integrated on each chromosome were 18 to 31 pairs, and the mean value was 24.3. Most of the positions of the SSR markers were similar to the positions presented in the IBM 2008 Neighbours Frame 6 from Maize GDB. Therefore, the linkage map can be used for QTL analysis.

Fig. 2.

The distribution of QTLs controlling zeaxanthin concentration mapped on maize chromosomes. QTLs detected in F₂ and F_2:3 populations derived from the cross between HuangC and Rezi1. Chr1–chr10: Chromosome1–10. QTL with red, blue, green, brown and black colour represent QTL detected in F₂ population, the F_2:3 population at YJ, the F_2:3 population at BB, the F_2:3 population at HC and joint data of three sites in F_2:3 population, respectively.

QTL mapping in the F₂ population

The results of QTL analysis for zeaxanthin in the F₂ population are shown in Fig. 2 and Table 3. A total of five QTLs were detected on chromosome 3, 4, 5, and 6. They explained 3.3–64.1% of the phenotypic variation and the LOD values of these QTLs were in the range of 2.07–62.75. qZea6a-F₂, explaining 64.1% of the phenotypic variation, was located in the marker interval, y1–bnlg1422 (bin 6.01–6.02), on chromosome 6. On the chromosome 3, qZea3a-F₂, explaining 6.1% of the phenotypic variation, was located in the interval bnlg1452–umc2263. In the interval umc2176–umc1963 on chromosome 4, qZea4a-F₂ was detected with a LOD of 3.13. qZea5a-F₂ (bin 5.07) on chromosome 5 accounted for 3.3% of the phenotypic variation for zeaxanthin concentration. Furthermore, the HuangC allele at these QTLs on chromosome 3, 5, and 6 increased zeaxanthin concentration, whereas the HuangC allele for QTLs on chromosome 4 decreased zeaxanthin concentration. The modes of gene action were over-dominance, additive and partial dominance, of which 60% of the QTLs favourable alleles originated from HuangC.

Table 3. QTLs for zeaxanthin in F₂ and F_2:3 populations

Population	QTL	Bins	Interval markers	Position	Size	LOD	R2 (%)	Add	Dom	GA
F2	qZea3a-F2	3.04	bnlg1452–umc2263	167.36	3.48	2.58	6.1	0.03	–0.28	OD
	qZea4a-F2	4.03–4.04	umc2176–umc1963	69.24	10.55	3.13	6.5	–0.05	0.3	OD
	qZea4b-F2	4.05	bnlg1217–umc1953	83.96	4.52	2.26	3.6	–0.05	0.26	OD
	qZea5a-F2	5.07	bmc2305	231.9	/	2.07	3.3	0.16	0.1	PD
	qZea6a-F2	6.01–6.02	y1ssr–bnlg1422	24.02	1.2	62.75	64.1	0.75	0.14	A
F2:3	qZea4a-YJ	4.03	umc2176	59.24	/	2.32	4.5	0.08	0.3	OD
	qZea6a-YJ	6.01–6.02	y1ssr–bnlg1422	24.02	1.2	26.72	41.4	0.64	0.09	A
	qZea1a-BB	1.03–1.04	umc1452–dupssr26	47.76	26.96	2	3.7	0.2	–0.12	A
	qZea3a-BB	3.04	bnlg1452–umc2263	167.36	3.48	3.68	7.8	0.15	–0.27	OD
	qZea3b-BB	3.09	umc2008–bnlg1754	317.26	8.76	2.69	5	0.16	–0.23	OD
	qZea3c-BB	3.09	bnlg1754–umc1052	331.01	12.94	2.79	5.2	0.16	–0.29	OD
	qZea6a-BB	6.01–6.02	y1ssr–bnlg1422	24.02	1.2	45.24	57.7	0.69	–0.02	A
	qZea1a-HC	1.06–1.07	umc2151–umc1833	107.17	29.79	2.07	3.9	0.22	0.04	A
	qZea1b-HC	1.07	umc1833–umc1147	141.97	47.42	2.11	3.9	0.25	–0.15	A
	qZea3a-HC	3.04	bnlg1452–umc2263	168.36	3.48	2.68	6	0.11	–0.26	OD
	qZea4a-HC	4.03–4.04	umc2176–umc1963	63.24	10.55	2.42	5.5	0.01	0.32	OD
	qZea4b-HC	4.04	bnlg490	72.97	/	2.41	4.5	–0.04	0.28	OD
	qZea6a-HC	6.01–6.02	y1ssr–bnlg1422	24.02	1.2	65.76	71.4	0.78	–0.08	A

Size: Size of marker interval. A: Additive effect; Negative values indicate that the alleles for increasing trait values are contributed by Rezi1, and positive values indicate that the alleles for increasing trait values are contributed by HuangC; D: Dominant effect; PD: Partial dominant effect; OD: Over dominant effect. LOD: Log of odds; R² (%): Percentage of phenotypic variance explained by the QTL; GA: Gene action.

QTL analysis at different sites in the F_2:3 population

In the F_2:3 population, a total of 13 QTLs for zeaxanthin concentration were detected on chromosomes 1, 3, 4, and 6 at three different sites, of which three QTLs were major effect QTLs (Table 3, Fig. 2). qZea6a-YJ, qZea6a-BB, and qZea6a-HC were located in the common interval y1–bnlg1422 (bin 6.01–6.02), which explained 41.4–71.4% of the phenotypic variation, and LOD ranged between 26.72 and 65.76. The genes showed an additive effect and the favourable alleles of these QTLs originated from HuangC.

Other QTLs were located on chromosome 1, 3, and 4 (Table 3, Fig. 2), which explained 3.7–7.8% of the phenotypic variation. On chromosome 1, three QTLs (qZea1a-BB, qZea1a-HC, and qZea1b-HC), explaining 3.7–3.9% of the phenotypic variation, were detected at BB and HC sites. These genes also showed an additive effect and the favourable alleles of these QTLs originated from HuangC. On chromosome 3, four QTLs (qZea3a-BB, qZea3b-BB, qZea3c-BB, and qZea3a-HC), explaining 5.0–7.8% of the phenotypic variation, were detected at BB and HC sites and LOD values ranged between 2.68 and 3.68. On chromosome 4, three QTLs (qZea4a-YJ, qZea4a-HC, and qZea4b- HC), explaining 4.5–5.5% of the phenotypic variation, were detected at YJ and HC sites and these gene actions showed over-dominance. In addition, two QTLs (qZea3a-BB and qZea3a-HC) were located in the common region flanked by markers bnlg1452–umc2263 (bin 3.04) and explained 6.0–7.8% of the phenotypic variation. Moreover, the HuangC allele at all QTLs increased zeaxanthin concentration, except for qZea4b-HC.

QTL analysis for joint data of three sites in the F_2:3 population

For joint data of the three sites in the F_2:3 population, five QTLs for zeaxanthin were detected (Table 4, Fig. 2). These showed additive effects and the favourable alleles of the QTLs originated from HuangC. One of the QTLs, qZea6a-F_2:3, located in the interval y1ssr–bnlg1422 on chromosome 6, was a major effect QTL, explaining 70.9% of the phenotypic variation. Four other QTLs (qZea1a-F_2:3, qZea3a-F_2:3, qZea4a-F_2:3, and qZea7a-F_2:3) explained 3.8–7.2% of the phenotypic variation and were located in the intervals of umc1833 and umc1147 on chromosome 1, bnlg1452 and umc2263 on chromosome 3, umc2176 and umc1963 on chromosome 4 and umc1407 on chromosome 7, respectively.

Table 4. QTLs for zeaxanthin in joint analysis across F_2:3 population

QTL	Bins	Interval markers	Position	Size	LOD	R2 (%)	Add	Dom	GA
qZea1a-F2:3	1.07	umc1833–umc1147	142.97	47.42	2.05	3.8	0.22	–0.13	PD
qZea3a-F2:3	3.04	bnlg1452–umc2263	167.36	3.48	3.13	7.2	0.11	–0.24	OD
qZea4a-F2:3	4.03–4.04	umc2176–umc1963	63.24	10.55	2.71	5.5	0.02	0.30	OD
qZea6a-F2:3	6.01–6.02	y1ssr–bnlg1422	25.02	1.2	64.87	70.9	0.7	–0.04	A
qZea7a-F2:3	7.05	umc1407	273.09	/	2.12	5.9	0.03	0.24	OD

Size: Size of marker interval. A: Additive effect; Negative values indicate that the alleles for increasing trait values are contributed by Rezi1, and positive values indicate that the alleles for increasing trait values are contributed by HuangC; D: Dominant effect; PD: Partial dominant effect; OD: Over dominant effect. LOD: Log of odds; R² (%): Percentage of phenotypic variance explained by the QTL; GA: Gene action.

Allelic effect analysis of marker genotypes in the F_2:3 population

In the F_2:3 population, the number of individuals and the distribution of phenotypic values for these QTLs (qZea1a-F_2:3, qZea3a-F_2:3, qZea4a-F_2:3, qZea6a-F_2:3, and qZea7a-F_2:3) at the nearest marker are shown in Fig. 3. The 242 plants were divided into three genotypes (homozygous HuangC, and heterozygous and homozygous Rezi1) based on their genotypes at five markers. For each of the five markers, the genotype matched the phenotype. The zeaxanthin concentration for the homozygous HuangC genotype was 0.99–1.82 mg*kg^–1. Comparing among genotypes, a significant difference (P < 0.01) of zeaxanthin concentration between the two homozygous genotypes indicated that the y1ssr marker was related to zeaxanthin concentration and that the homozygous HuangC alleles had a strong genetic effect on improving zeaxanthin concentration in maize kernel (Fig. 3d).

Fig. 3.

The zeaxanthin concentration distributions and mean values for marker genotypes in F_2:3 population. a–d: F_2:3 population were genotyped at markers umc1833, umc2263, umc1963, y1ssr, and umc1407, respectively. The zeaxanthin concentration distributions and mean values are shown for homozygous HuangC, heterozygous, and homozygous Rezi1 genotypes. * means significant at the 0.05 probability level; ** means significant at the 0.01 probability level.

Genotypic and phenotypic analyses at each combination of multiple marker genotypes (umc1833/umc1963, umc2263/umc1963, umc1963/y1ssr, and umc1407/y1ssr) are shown in Table 5. According to the law of independent assortment, a χ² test showed that the nine genotypes tested in this study fitted the Mendelian separation ratio (χ² = 6.73–9.23, χ²_0.05,8 = 15.51). In addition, the zeaxanthin concentrations in kernels were highly significantly different among genotypes. When one allelic genotype was homozygous dominant (AA), the differences in zeaxanthin concentration between AABB and AABb, AABB and AAbb, and AABb and AAbb were not significant, highly significant and highly significant, respectively, for the umc1833/umc1963 and umc1963/y1ssr combinations, and all highly significant for the umc2263/umc1963 and umc1407/y1ssr combinations. When one allelic genotype was heterozygous dominant (Aa), the differences in zeaxanthin concentration between AaBB and AaBb, AaBB and Aabb, and AaBb and Aabb were not significant, highly significant and highly significant, respectively, for the umc1833/umc1963 and umc2263/umc1963 combinations, all highly significant for the umc1963/y1ssr combination, and not significant, not significant, and significant, respectively, for umc1407/y1ssr combination. When one allelic genotype was homozygous recessive (aa), the differences in zeaxanthin concentration between aaBB and aaBb, aaBB and aabb, and aaBb and aabb were not significant, highly significant and highly significant, respectively, for the umc1833/umc1963 and umc2263/umc1963 combinations, all highly significant for the umc1963/y1ssr combination, but all not significant for the umc1407/y1ssr combination. Hence, it can be concluded that the epistatic effects among these combinations play a key role in improving zeaxanthin concentration in maize kernels.

Table 5. Genotypic and phenotypic analyses at each of combinations of multiple marker genotypes in F_2:3 population

Maker genotype	qZea1a × qZea4a (umc1833/umc1963)		qZea3a × qZea4a (umc2263/umc1963)		qZea4a × qZea6a (umc1963/y1ssr)		qZea7a × qZea6a (umc1407/y1ssr)
	Number of individuals	Mean of zeaxanthin concentration (mg*kg–1)	Number of individuals	Mean of zeaxanthin concentration (mg*kg–1)	Number of individuals	Mean of zeaxanthin concentration (mg*kg–1)	Number of individuals	Mean of zeaxanthin concentration (mg*kg–1)
AABB	12	1.46a/A†	13	1.27b/B	13	1.80a/A	13	1.65b/B
AABb	31	1.40ab/A	39	1.49a/A	31	1.85a/A	37	1.94a/A
AAbb	12	0.92f/E	14	1.19c/C	13	0.88de/D	14	0.98d/D
AaBB	29	1.19cd/BC	39	1.29b/B	40	1.84a/A	35	1.20ce/CD
AaBb	55	1.30bc/AB	54	1.27b/B	54	1.52b/B	54	1.24c/C
Aabb	33	1.00ef/DE	29	1.01d/D	27	0.94cd/CD	26	1.00ed/CD
aaBB	23	1.07de/CDE	11	1.05d/D	20	1.03c/C	12	0.63f/EF
aaBb	34	1.08de/CDE	28	1.00d/D	27	0.80e/D	25	0.59f/F
aabb	9	0.62g/F	10	0.70e/E	15	0.38f/E	21	0.43f/F

^† Lowercase letters indicate significance at the 0.05 level; Uppercase letters indicate significance at the 0.01 level.

Epistatic effects analysis in the F_2:3 population

Four of the ten QTL pairs epistatically affected the zeaxanthin concentration, explaining 7.34–14.3% of the phenotypic variance in the F_2:3 population (Table 6). The interactions between qZea1a and qZea4a, and qZea3a and qZea4a led to an increase in percentage explanation of phenotypic variation. Therefore, these minor QTLs (qZea1a, qZea4a and qZea3a) are also important for improving the zeaxanthin concentration in maize kernel.

Table 6. Epistasis analysis of QTLs across F_2:3 population

QTL × QTL	LOD	R2 (%)	Add by Add	Add by Dom	Dom by Add	Dom by Dom
qZea1a × qZea4a	5.2	7.34	0.15	–0.63	–0.23	–1.02
qZea3a × qZea4a	5.4	9.8	0.22	0.2	0.11	0.21
qZea4a × qZea6a	8.2	14.3	0.36	–0.19	–0.37	–1.42
qZea7a × qZea6a	5.1	9.3	0.13	0.14	–0.82	–0.63

LOD: Log of odds; R² (%): Percentage of phenotypic variance explained by the QTL; Add by Add: Additive by additive effect at the two scanning positions; Add by Dom: Additive by dominance effect at the two scanning positions; Dom by Add: Dominance by additive effect at the two scanning positions; Dom by Dom: Dominance by dominance effect at the two scanning positions.

Discussion

In maize, zeaxanthin accumulates predominantly in the endosperm (Blessin et al. 1963, Steenbock and Coward 1925). Generally, yellow maize kernel contains higher amounts of zeaxanthin than white maize (O’Hare et al. 2015, Scott and Eldridge 2005). Quantitative traits, controlled by minor-effect polygenes, are highly susceptible to environmental effects (Flint and Mackay 2009, Holland 2007).

In previous studies, the QTLs for zeaxanthin concentration were mainly concentrated to the regions 6.01–6.02, 7.00–7.02, 8.03–8.05, and 10.05–10.06 (Chander et al. 2008, Wong et al. 2004). Chander et al. (2008) identified a major effect QTL for zeaxanthin in the interval y1ssr–umc1595 on chromosome 6 (bin 6.01–6.02) by using a RIL population. Wong et al. (2004) detected the marker interval y1ssr–bnlg249 (bin 6.01–6.02) associated with a QTL for zeaxanthin by using F_2:3 populations. In our study, the QTL qZea6a was located in the region of y1ssr–bnlg1422 (bin 6.01–6.02) within a distance of 1.2 cM in the F₂ generation, in all the three sites investigated, and in the joint data of the F_2:3 population. Comparing with the previous results, the region of y1ssr–bnlg249 associated with qZea6a is narrower than y1ssr–umc1595 on the IBM2008 neighbours Frame 6. This makes it possible to narrow down the QTL to a small genetic interval and segregate the QTL as a single Mendelian gene. Therefore, qZea6a shows a major effect on zeaxanthin concentration and stable expression across environments and genetic backgrounds, and is most important for marker-assisted selection (MAS) (Ribaut and Hoisington 1998).

It was previously reported that the QTL for zeaxanthin on chromosome 6 (bin 6.01–6.02) accounts for 13.9–16.3% of the phenotypic variation (Chander et al. 2008, Wong et al. 2004). In this study, the QTL qZea6a on chromosome 6 had a higher contribution to zeaxanthin production, ranging from 41.4% to 71.4%. This might be due to the high genetic difference in zeaxanthin genes between the two parents. Furthermore, the alleles producing higher increasing zeaxanthin concentration were the progeny of parents with high zeaxanthin production ability. Therefore, the parent with high zeaxanthin concentration is very important for the improvement of zeaxanthin in maize breeding programs.

To date, bin regions 6.01–6.02, 7.00–7.02, 8.03–8.05, and 10.05–10.06 were reported to be associated with maize zeaxanthin concentration (Chander et al. 2008, Wong et al. 2004). In the present study, two other regions, bin 3.04 and bin 4.03–4.04, were detected to harbour QTLs for maize zeaxanthin concentration. Although these QTLs only contribute 5.0–7.8% to the phenotypic variation, they are worthy of further investigation since they are stable across different generations and environments. Several QTLs at different positions detected in different studies indicates that zeaxanthin concentration in maize is a complex quantitative trait.

The combination of epistatic effects and additive allelic effects from the two parents at different QTLs is related to the origin of transgressive segregation that produces superior phenotypes (Latta et al. 2010). In this study, the allelic effects of the QTLs varied among the F₂ and F_2:3 populations, and additivity was the major allelic action at zeaxanthin concentration QTLs detected in the F₂ population, indicating that dominance was not the most important genetic basis for improving zeaxanthin concentration in the F_2:3 population. Moreover, epistasis causes hidden quantitative genetic variation in natural populations and could be responsible for the small additive effects (Mackay 2014). The LOD values and contribution rates (R²) were different between a single QTL and a pair of epistatic QTLs. The four pairs of epistatic QTLs explained 7.34–14.3% of the phenotypic variance in the F_2:3 population, indicating that epistatic interactions for quantitative traits result in a change of magnitude of allelic effects at one locus. HuangC contributed to the favourable alleles involved in the epistatic interaction that increased the flux through the pathway.

Three of the twelve constitutive QTLs, qZea3a, qZea4a and qZea6a, were observed to be constitutively expressed across the two generations and joint data of the three sites in the F_2:3 population. These QTLs may possibly perhaps have higher values of MAS for improving quality in maize. Furthermore, the closely linked molecular markers that we developed for qZea6a will be useful in maize breeding programs for improving zeaxanthin concentration by applying MAS, and our results will also lay the foundation for the isolation of the candidate gene of maize zeaxanthin concentration.

Author Contribution Statement

EFD performed the fieldwork, collections of materials, data analysis, and manuscript drafting. YB and LPQ conducted the phenotypic measurement and DNA isolation. QYL and CXL were involved in SSR genotyping and PCR amplification. EFD and YLC was involved in experimental design and manuscript preparation. All authors read and approved the final manuscript.

Acknowledgments

The authors would like to appreciate funding from the Chongqing Major Research Project (cstc2016shms-ztzc80016, cstc2016shms-ztzx80017 and cstc2016shms-ztzx80013) and Science and Technology Research Key Project of Colleges and Universities in Hebei Province (ZD2017037).

Literature Cited

•  Blessin,  C.W.,  J.D.  Brecher and  R.J.  Dimler (1963) Carotenoids of corn and sorghum. 5. Distribution of xanthophylls and carotenes in hand-dissected and dry-milled fractions of yellow dent corn. Cereal Chem. 40: 582–586.
•  Burt,  A.J.,  C.M.  Grainger,  B.J.  Shelp and  E.A.  Lee (2011) Heterosis for carotenoid concentration and profile in maize hybrids. Genome 54: 993–1004.
•  Causse,  M.A.,  T.M.  Fulton,  Y.G.  Cho,  S.N.  Ahn,  J.  Chunwongse,  K.  Wu,  J.  Xiao,  Z.  Yu,  P.C.  Ronald and  S.E.  Harrington (1994) Saturated molecular map of the rice genome based on an interspecific backcross population. Genetics 138: 1251–1274.
•  Chander,  S.,  Y.Q.  Guo,  X.H.  Yang,  J.  Zhang,  X.Q.  Lu,  J.B.  Yan,  T.M.  Song,  T.R.  Rocheford and  J.S.  Li (2008) Using molecular markers to identify two major loci controlling carotenoid contents in maize grain. Theor. Appl. Genet. 116: 223–233.
•  Davison,  P.A.,  C.N.  Hunter and  P.  Horton (2002) Overexpression of beta-carotene hydroxylase enhances stress tolerance in Arabidopsis. Nature 418: 203–206.
•  Deng,  X.,  X.  Mei,  X.U.  De,  X.  Chen,  X.  Liu,  C.  Liu and  Y.  Cai (2015) Niacin and zeaxanthin contents in different maize kernels. Food Sci. 36: 119–124.
•  Flint,  J. and  T.F.  Mackay (2009) Genetic architecture of quantitative traits in mice, flies, and humans. Genome Res. 19: 723–733.
•  Holland,  J.B. (2007) Genetic architecture of complex traits in plants. Curr. Opin. Plant Biol. 10: 156–161.
•  Humphries,  J.M. and  F.  Khachik (2003) Distribution of lutein, zeaxanthin, and related geometrical isomers in fruit, vegetables, wheat, and pasta products. J. Agric. Food Chem. 51: 1322–1327.
•  Kato,  M.,  Y.  Ikoma,  H.  Matsumoto,  M.  Sugiura,  H.  Hyodo and  M.  Yano (2004) Accumulation of carotenoids and expression of carotenoid biosynthetic genes during maturation in citrus fruit. Plant Physiol. 134: 824–837.
•  Knapp,  E.W.,  S.F.  Fischer,  W.  Zinth,  M.  Sander,  W.  Kaiser,  J.  Deisenhofer and  H.  Michel (1985) Analysis of optical spectra from single crystals of Rhodopseudomonas viridis reaction centers. Proc. Natl. Acad. Sci. USA 82: 8463–8467.
•  Kosambi,  D.D. (1943) The estimation of map distance from recombination values. Ann. Hum. Genet. 12: 172–175.
•  Latta,  R.G.,  K.M.  Gardner and  D.A.  Staples (2010) Quantitative trait locus mapping of genes under selection across multiple years and sites in Avena barbata: epistasis, pleiotropy, and genotype-by-environment interactions. Genetics 185: 375–385.
•  Li,  H.,  G.  Ye and  J.  Wang (2007) A modified algorithm for the improvement of composite interval mapping. Genetics 175: 361–374.
•  Li,  H.,  J.M.  Ribaut,  Z.  Li and  J.  Wang (2008) Inclusive composite interval mapping (ICIM) for digenic epistasis of quantitative traits in biparental populations. Theor. Appl. Genet. 116: 243–260.
•  Loane,  E.,  J.M.  Nolan,  O.  O’Donovan,  P.  Bhosale,  P.S.  Bernstein and  S.  Beatty (2008) Transport and retinal capture of lutein and zeaxanthin with reference to age-related macular degeneration. Surv. Ophthalmol. 53: 68–81.
•  Mackay,  T.F. (2014) Epistasis and quantitative traits: using model organisms to study gene-gene interactions. Nat. Rev. Genet. 15: 22–33.
•  Meng,  L.,  H.  Li,  L.  Zhang and  J.  Wang (2015) QTL IciMapping: Integrated software for genetic linkage map construction and quantitative trait locus mapping in biparental populations. Crop J. 3: 269–283.
•  Moeller,  S.M.,  P.F.  Jacques and  J.B.  Blumberg (2000) The potential role of dietary xanthophylls in cataract and age-related macular degeneration. J. Am. Coll. Nutr. 19: 522S–527S.
•  Murray,  M.G. and  W.F.  Thompson (1980) Rapid isolation of high molecular weight plant DNA. Nucleic Acids Res. 8: 4321–4326.
•  Nilsson,  S.E.G.,  S.P.  Sundelin,  U.  Wihlmark and  U.T.  Brunk (2003) Aging of cultured retinal pigment epithelial cells: oxidative reactions, lipofuscin formation and blue light damage. Doc. Ophthalmol. 106: 13–16.
•  O’Hare,  T.J.,  K.J.  Fanning and  I.F.  Martin (2015) Zeaxanthin biofortification of sweet-corn and factors affecting zeaxanthin accumulation and colour change. Arch. Biochem. Biophys. 572: 184–187.
•  Ooijen,  J.W.V. (2011) Multipoint maximum likelihood mapping in a full-sib family of an outbreeding species. Genet. Res. (Camb) 93: 343–349.
•  Perry,  A.,  H.  Rasmussen and  E.J.  Johnson (2009) Xanthophyll (lutein, zeaxanthin) content in fruits, vegetables and corn and egg products. J. Food Compost. Anal. 22: 9–15.
•  Ribaut,  J.M. and  D.  Hoisington (1998) Marker-assisted selection: new tools and strategies. Trends Plant Sci. 3: 236–239.
•  Sajilata,  M.G.,  R.S.  Singhal and  M.Y.  Kamat (2008) The carotenoid pigment zeaxanthin—a review. Compr. Rev. Food Sci. Food Saf. 7: 29–49.
•  Scott,  C.E. and  A.L.  Eldridge (2005) Comparison of carotenoid content in fresh, frozen and canned corn. J. Food Compost. Anal. 18: 551–559.
•  Sengun,  I.Y.,  G.  Yildiz Turp,  F.  Icier,  P.  Kendirci and  G.  Kor (2014) Effects of ohmic heating for pre-cooking of meatballs on some quality and safety attributes. LWT-Food Sci. Technol. 55: 232–239.
•  Senior,  M.L.,  E.C.L.  Chin,  M.  Lee,  J.S.C.  Smith and  C.W.  Stuber (1996) Simple sequence repeat markers developed from maize sequences found in the GENBANK database: Map construction. Crop Sci. 36: 1676–1683.
•  Snodderly,  D.M. (1995) Evidence for protection against age-related macular degeneration by carotenoids and antioxidant vitamins. Am. J. Clin. Nutr. 62: 1448S–1461S.
•  Steenbock,  H. and  K.H.  Coward (1925) The quantitative determination of vitamin A. J. Biol. Chem. 47: 1639–1646.
•  Stuber,  C.W.,  M.D.  Edwards and  J.F.  Wendel (1987) Molecular marker-facilitated investigations of quantitative trait loci in maize. II. Factors influencing yield and its component traits. Crop Sci. 27: 639–648.
•  Tuberosa,  R.,  M.C.  Sanguineti,  P.  Landi,  S.  Salvi,  E.  Casarini and  S.  Conti (1998) RFLP mapping of quantitative trait loci controlling abscisic acid concentration in leaves of drought-stressed maize (Zea mays L.). Theor. Appl. Genet. 97: 744–755.
•  Vallabhaneni,  R.,  C.E.  Gallagher,  N.  Licciardello,  A.J.  Cuttriss,  R.F.  Quinlan and  E.T.  Wurtzel (2009) Metabolite sorting of a germplasm collection reveals the hydroxylase3 locus as a new target for maize provitamin A biofortification. Plant Physiol. 151: 1635–1645.
•  Voorrips,  R.E. (2002) MapChart: Software for the graphical presentation of linkage maps and QTLs. J. Hered. 93: 77–78.
•  Wang,  J.K. (2009) Inclusive composite interval mapping of quantitative trait genes. Acta Agronomica Sinica 35: 239–245.
•  Wong,  J.C.,  R.J.  Lambert,  E.T.  Wurtzel and  T.R.  Rocheford (2004) QTL and candidate genes phytoene synthase and zeta-carotene desaturase associated with the accumulation of carotenoids in maize. Theor. Appl. Genet. 108: 349–359.
•  Wu,  J.,  J.  Ji,  G.  Wang,  G.  Wu,  J.  Diao,  Z.  Li,  X.  Chen,  Y.  Chen and  L.  Luo (2015) Ectopic expression of the Lycium barbarum β-carotene hydroxylase gene (chyb) enhances drought and salt stress resistance by increasing xanthophyll cycle pool in tobacco. Plant Cell Tissue Organ Cult. 121: 559–569.
•  Xiu-Hong,  X.,  L.  Gui-Hua,  J.J.  Chen,  H.B.  Lu and  G.J.  Guo (2013) Accumulated dynamic change of zeaxanthin content in fresh corn grains during its filling stage. Acta Agriculturae Zhejiangensis 25: 685–688.
•  Xu,  S.B.,  Y.F.  Tao,  Z.Q.  Yang and  J.Y.  Chu (2002) A simple and rapid methods used for silver staining and gel preservation. Hereditas 24: 335–336.
•  Yan,  J.,  C.B.  Kandianis,  C.E.  Harjes,  L.  Bai,  E.H.  Kim,  X.  Yang,  D.J.  Skinner,  Z.  Fu,  S.  Mitchell,  Q.  Li et al. (2010) Rare genetic variation at Zea mays crtRB1 increases β-carotene in maize grain. Nat. Genet. 42: 322–327.