Development of sub-tropically adapted diverse provitamin-A rich maize inbreds through marker-assisted pedigree selection, their characterization and utilization in hybrid breeding

Malnutrition has emerged as one of the major health problems worldwide. Traditional yellow maize has low provitamin-A (proA) content and its genetic base in proA biofortification breeding program of subtropics is extremely narrow. To diversify the proA rich germplasm, 10 elite low proA inbreds were crossed with a proA rich donor (HP702-22) having mutant crtRB1 gene. The F2 populations derived from these crosses were genotyped using InDel marker specific to crtRB1. Severe marker segregation distortion was observed. Seventeen crtRB1 inbreds developed through marker-assisted pedigree breeding and seven inbreds generated using marker-assisted backcross breeding were characterized using 77 SSRs. Wide variation in gene diversity (0.08 to 0.79) and dissimilarity coefficient (0.28 to 0.84) was observed. The inbreds were grouped into three major clusters depicting the existing genetic diversity. The crtRB1-based inbreds possessed high β-carotene (BC: 8.72μg/g), β-cryptoxanthin (BCX: 4.58μg/g) and proA (11.01μg/g), while it was 2.35μg/g, 1.24μg/g and 2.97μg/g in checks, respectively. Based on their genetic relationships, 15 newly developed crtRB1-based inbreds were crossed with five testers (having crtRB1 gene) using line × tester mating design. 75 experimental hybrids with crtRB1 gene were evaluated over three locations. These experimental hybrids possessed higher BC (8.02μg/g), BCX (4.69μg/g), proA (10.37μg/g) compared to traditional hybrids used as check (BC: 2.36 μg/g, BCX: 1.53μg/g, proA: 3.13μg/g). Environment and genotypes × environment interaction had minor effects on proA content. Both additive and dominance gene action were significant for proA. The mean proportion of proA to total carotenoids (TC) was 44% among crtRB1-based hybrids, while 11% in traditional hybrids. BC was found to be positively correlated with BCX (r = 0.68) and proA (r = 0.98). However, no correlation was observed between proA and grain yield. Several hybrids with >10.0 t/ha grain yield with proA content >10.0 μg/g were identified. This is the first comprehensive study on development of diverse proA rich maize hybrids through marker-assisted pedigree breeding approach. The findings provides sustainable and cost-effective solution to alleviate vitamin-A deficiency.

Introduction Malnutrition due to consumption of unbalanced diet affects two billion people worldwide [1]. Deficiency of micronutrients in the long run causes severe socio-economic problems. Among micronutrients, vitamin-A deficiency (VAD) is one of the major health problems found to subsist in human population [2]. Vitamin-A is essential for vision, immunity and various metabolisms [3]. Night blindness and complete loss of vision are the hallmarks of the VAD in humans [4]. The deficiency also induces higher risk to severe infections such as measles, diarrhoea and weakened immunity among children and pregnant women [5]. While 30% of preschool-age children and more than 19 million pregnant women in developing countries are vitamin-A deficient, 5.2 million of the same age preschool-age groups and 9.7 million pregnant women suffer from clinically night blindness (www.harvestplus.org).
Various avenues namely, food-fortification, medical-supplementation and food-diversification are implemented to alleviate VAD [6,7]. However, sustainability of these approaches is limited by the weak distribution system, low purchasing power of the rural people and crop seasonality [8]. 'Crop-biofortification' where micronutrient density is enhanced in edible parts of food through plant breeding, has now emerged as the most popular choice to address malnutrition through cost-efficient and sustainable approach [9]. Biofortified staple crops when consumed regularly have been found to improve the human health [10,11].
Maize (Zea mays L.) is an important cereal crop grown in almost all parts of the world and cultivated across diverse climatic spectrum [12]. It is a source of food to billions of people and also used as feed for poultry and livestock [13]. Traditional yellow kernel maize possesses high kernel carotenoids, but composed predominantly of non-provitamin-A (non-proA) fractions [lutein (LUT) and zeaxanthin (ZEA)] and very less provitamin-A (proA) fraction [α-carotene and β-carotene (BC), β-cryptoxanthin (BCX)] [14,15]. ProA content in traditional tropical maize is quite low (0.25-2.50 μg/g) and far-off from the targeted concentration of 15 μg/g as set by HarvestPlus [1]. CrtRB1 gene that codes for β-carotene hydroxylase is associated with higher accumulation of proA especially BC and BCX in maize. Rare natural variation in crtRB1 gene limits the hydroxylation of BC and BCX [16]. The wild type allele possesses a transposable element (TE) in 3'UTR region of the crtRB1, while the TE is absent in mutant version [17].
Diverse proA rich inbreds have been developed in the tropics [17][18][19]. However, the genetic base of proA rich inbreds is extremely low in the sub-tropical regions and thus the frequency of favourable crtRB1 allele is extremely low in Indian maize germplasm [2,20]. Therefore, strengthening the breeding programme by broadening the genetic base of proA germplasm assumes great significance. The present study was, therefore undertaken to (i) develop subtropically adapted diverse crtRB1-inbreds through marker-assisted pedigree breeding, (ii) characterize the newly developed crtRB1-based inbreds using microsatellite markers, (iii) PMI-PV-2 are parental inbreds of India's first proA rich maize hybrid (Pusa Vivek QPM9 Improved). PMI-PV-5, PMI-PV-6, PMI-PV-7, PMI-PV-8 and PMI-PV-9 are the parents of MABB-derived proA rich hybrids (APQH-1, APQH-4, Pusa HQPM5 Improved, Pusa HQPM7 Improved and APQH-8). Further, HP465-41, a CIMMYT-derived proA rich inbred derived through marker-assisted pedigree breeding was also included for characterization. Two low proA elite inbreds (PMI-Q2 and PMI-Q3, possessing the unfavourable allele of crtRB1) were also included as control for quality analysis. These 26 inbreds were planted in randomized complete block design (RCBD) at ICAR-IARI, New Delhi in rainy season of 2018. Each inbred was planted with two replications, in rows of 3 m with a plant-to-plant distance of 20 cm. The rows were spaced 75 cm apart. Recommended cultural practices were followed to raise a good experimental crop. To avoid contamination by foreign pollens, 2-3 plants in each row were selfed for estimation of carotenoids.
DNA extraction and PCR. Selfed seeds of 24 inbreds with favourable allele of crtRB1 were used for molecular analysis using simple sequence repeats (SSRs) markers. DNA was isolated from seeds using standard sodium dodecyl sulphate (SDS) extraction protocol [23]. A total of 77 SSRs distributed across the genome were used for characterization. The information of SSR markers with bin locations and nature of SSR repeats at each locus are provided in Table 2. Primer sequence information of maize SSRs was retrieved from public domain (Mai-zeGDB; http://www.maizegdb.org). PCR was carried out as per Choudhary et al. (2016) [24]. The PCR amplified products for each SSR were resolved on 4% agarose gel stained with 0.4 mg/ml ethidium bromide using horizontal electrophoresis system at 120 V for 3-4 h.
Genetic diversity analysis. For each allele, presence of a band in a genotype was indicated by 1 and absence of the band as 0. Five parameters, viz., gene diversity, major allele frequency, total number of alleles detected, heterozygosity and polymorphism information content (PIC) were estimated using PowerMarker v3.0 [25]. An allele appearing only in one genotype was scored as unique allele, while an allele with a frequency of �0.05 was considered as a rare allele. Genetic dissimilarity analysis using Jaccard's coefficient was calculated and tree was constructed using Neighbour-Joining (NJ) pattern in DARwin-6.0 [26]. Principal coordinate analysis (PCoA) was also carried out to complement the clustering pattern [27].
Estimation of carotenoids from inbreds. Carotenoids from the selfed seeds of 26 inbreds (24 with favourable allele of crtRB1 and two inbreds with unfavourable allele of crtRB1) were extracted from maize endosperm through protocol of Kurilich and Juvik (1999) [28] with modifications. Carotenoids were quantified using Dionex Ultimate 3000 UHPLC System (Ultra  [15]. Total carotenoid (TC) was calculated by adding the value of BC, BCX, LUT and ZEA [29].
Heterosis for grain yield. Grain yield (YLD) per plot was converted to t/ha as per the standard procedure. Magnitude of heterosis in hybrids over five commercial checks was estimated following Singh and Choudhary (1985) [31].
Statistical analysis. The statistical analyses on ANOVA, correlation coefficients and combining ability were computed using Windostat 8.0.

Selection of crtRB1-based segregants in F 2 populations
The recipient parents produced an amplicon of 296 bp, while the donor, a 543 bp amplicon. The true F 1 s had both 296 and 543 bp amplicons. The 10 F 2 populations were genotyped using crtRB1-specific 3'TE-InDel-based marker (Table 1). A representative gel depicting the segregation of crtRB1 gene in F 2 s is presented in Fig 1. Of the total 1043 segregants genotyped, only 201 were homozygous for favourable allele ofcrtRB1. The heterozygotes and homozygotes (wild-type allele) were 407 and 435, respectively. Thus, crtRB1 showed severe marker segregation distortion both cumulatively as well as in individual populations. Out of 201 favourable homozygotes, 75 segregants were selected based on ear-and grain-characteristics, and advanced to generate F 3 progenies. Finally, 15 locally adapted F 4 progenies (S2 Table) representing all 10 crosses were selected for further characterization.
Among crtRB1-based inbreds, BC and BCX contributed 28% and 15% of TC, while LUT and ZEA contributed 39% and 18%, respectively. The contribution of BC and BCX to TC was only 6% and 3% in check inbreds, while the same for LUT and ZEA was 58% and 33%, respectively.

Combining ability analysis for carotenoids
ANOVA for line × tester. Pooled ANOVA revealed that environment was highly significant for all the carotenoids except for ZEA (Table 6). Line effect was significant only for ZEA,  while tester effect was significant for BC, BCX, proA, LUT, non-proA and TC. The interactions of line × tester and environment × crosses were significant for all the carotenoids as well. Environment × line × tester interaction was also found to be significant for all characters except for LUT. On the other hand, environment × tester interaction was significant only for BCX, while environment × line interaction was non-significant for all the characters. Combining ability estimates. The proportion of additive and dominance variance, and the contribution of lines, testers and line × testers for pooled dataset are presented in Table 7. Variance due to specific combining ability (SCA) was higher than variance due to general combining ability (GCA) for all the characters. Though, dominance variance was predominant for BC, BCX, proA and TC, additive variance was found to be important as well. Additive variance was more for ZEA and non-proA, while both additive and dominance variance was of similar magnitude for LUT. When the contribution of lines, testers and line × tester were compared, line × tester interaction was found to be contributing more than line and testers for all the characters [36.55% (ZEA) to 67.15% (BC)].

Grain yield among experimental hybrids
Heterosis of the 75 experimental hybrids was estimated over commercial low-proA and high proA checks. CMH08-292 and DHM-121 were medium maturing hybrids with low-proA. 'Pusa Vivek QPM-9 Improved' is an early maturing proA rich hybrid and was released as country's first proA rich maize hybrid during 2017. 'Pusa HQPM-5 Improved' and 'Pusa HQPM-7 Improved' are the medium maturing proA rich hybrids released in 2020. All the experimental hybrids matured in 92-100 days, thus had medium maturity. The mean grain yield among experimental hybrids was 8.9 t/ha, with the highest yield of 11.4 t/ha (S4 and S5 Tables). Ten experimental hybrids possessed yield between 10 to 11 t/ha, while 25 hybrids had grain yield 9 to 10 t/ha. CoMH08-292 was the best low proA check with 10.4t/ha. A set of 20 experimental hybrids were either better or at par with CoMH-08-292. Among high proA checks, 'Pusa HQPM-5 Improved' emerged as the best check with 8.6 t/ha. A total of 26 experimental hybrids were significantly better than 'Pusa HQPM-5 Improved' for grain yield. These hybrids showed heterosis of 8.04% to 31.63% over the best high proA check.

Discussion
Plant based-food is the major source of nutrition especially in developing world. Traditional yellow maize lacks the required level of proA [17]. The mutant version of crtRB1 significantly enhances proA in maize kernel [16]. Diverse proA germplasm has been developed in the tropics [18,19]. However, the genetic base of proA rich inbreds is quite narrow in entire sub-tropics. The frequency of favourable allele of crtRB1 in the Indian maize germplasm is quite low (3.38%) [32]. So far, 9-10 MABB-derived crtRB1 inbreds have been developed in India. Thus, targeted breeding approach for selection of crtRB1 is essential for broadening the genetic base of proA rich maize germplasm [33].

Marker-assisted selection for crtRB1
Genotyping of F 2 populations indicated that crtRB1 did not segregate as per the expected 1  [17]. The segregation distortion could be due to activity of various gametophytic factors, defective kernel mutants, male sterility and embryo-specific mutation [35]. Since, the frequency of homozygotes for favourable allele was less, raising large backcross populations becomes a necessity for selection of desirable number of positive segregants. Here, crtRB1 gene could be precisely selected due to the reliable gene-based marker. In case of linked marker, there is always chance of selection of false positive individuals due to crossing over between the gene and marker [36]. The present study developed a set of diverse crtRB1-based inbreds using marker-assisted pedigree breeding. Earlier  [33] have introgressed crtRB1 into the elite inbreds using MABB approach. In case of MABB, the improved lines are genetically similar to the recurrent parents except for gene under introgression [36]. In the present study, since crtRB1-based inbreds were developed from F 2 populations, the genetic makeup remains novel leading to development of new and diverse crtRB1-based inbreds. The MAS-based selection of crtRB1 is quite cost-effective, as selection of genotypes for high proA using UHPLC involves US$30-35 per sample. On the contrary, PCR marker-based selection of crtRB1 employed here costs only US$0.5-1.0 per sample. Molecular breeding is now a preferred choice among the maize breeders to develop proA rich germplasm [2].

Molecular characterization of crtRB1-based inbreds
Knowledge of the genetic relationship among inbreds is essential for efficient exploitation in breeding programme. Molecular markers have been successfully employed to derive the genetic distance in maize [40]. High dissimilarity coefficient indicated the presence of high level of genetic diverseness among the inbreds. Lower major allele frequency also reflects diverse nature of the locus. In the present study, about one-third of total SSR loci had major allele frequency �0.5, which indicated large genetic dissimilarity among the inbreds. The genetic information obtained from cluster analysis was highly consistent with pedigree information. PCoA also supported the results of cluster diagram elucidating the diverse nature of inbreds. Inbreds derived from same population source were in general found to be in same cluster [39]. The study also identified few unique alleles and rare alleles among the inbred panel. The identified unique alleles could be useful to distinguish inbreds unambiguously from one another [24,41]. Low mean heterozygosity observed among the SSR loci indicated that inbreds reached appreciable level of homozygosity. Possible reason for heterozygosity observed may be due to tendency of some loci to segregate even after repeated inbreeding [42,43]. Inbreds developed conventionally exhibited higher degree of heterozygosity due to natural selection against homozygotes, when compared with doubled haploid (DH)-based inbreds [44].
In the carotenoid biosynthesis pathway, phytoene synthase 1 (psy1) or Yellow 1 (Y1) gene condenses two geranyl-geranyl pyrophosphate molecules into one molecule of phytoene [46]. The mutant/recessive y1 allele is unable to catalyze the reactions and white grains are formed due to no synthesis of carotenoids. However, when the Y1 is functional, crtRB1 causes hydroxylation of BC and BCX into ZEA. CrtRB1 located on chromosome 10 codes for β-carotene hydroxylase. The mutant version of crtRB1 drastically slows down the conversion, leading to more accumulation of proA carotenoids [16]. Though all the 15 inbreds possessed same crtRB1 allele, a large variation in proA was observed. This could be due to variation in the activity of other key genes such as psy1, lycopene β-cyclase (lcyB), lycopene ε-cyclase (lcyE), phytoene desaturase (pds) and z-carotene desaturase (zds) catalyzing the carotenoid biosynthesis [37]. Allelic variation for lcyE present on chromosome 8 has been observed [47]. Zunjare et al. (2017) [22] reported that the presence of lcyE along with crtRB1 is beneficial in enhancing proA in maize. Besides, several modifier loci/QTLs alone or in combination with other pathway genes could also influence the accumulation of proA in maize [45,48].

Proportion of carotenoids and their relationships
Among carotenoids, proportion of LUT was higher in both crtRB1-based and check-genotypes, thereby, suggesting greater flux of lycopene towards α-branch than the β-branch of pathway [29]. LUT serves as the precursor for various pathways, thus, is required in larger amount. Among check genotypes, ZEA was the second highest carotenoid after LUT. This is due to higher conversion of BC and BCX to ZEA as it serves as the precursor for synthesis of abscisic acid [17]. However, in crtRB1-based genotypes, the conversion of BC to ZEA is partially blocked, leading to higher proportion of BC next to LUT. ProA content among the crtRB1-based genotypes constituted 43-44% of TC as against 9-11% in the check genotypes. The non-proA component was less (56-57%) among crtRB1-based genotypes compared to 89-91% in checks [20,49]. In some of the genotypes, proA with >50% of TC was also observed. This is possibly due to presence of other favourable loci that act synergistically with crtRB1 [29].
BC, BCX and proA showed strong positive correlations, as BCX is produced from BC, while both contribute to proA [49]. Since, BC and BCX are converted to ZEA, negative correlation is expected [50]. Further, LUT in α-branch is produced at the cost of flux of lycopene towards β-branch, where BC and BCX are formed. This mechanism could be responsible for negative relationships among proA components with LUT. ProA carotenoids showed no association with grain yield. This suggested the possibility of developing high yielding maize hybrids with high proA. So far >40 proA rich hybrids and open-pollinated varieties (OPVs) with high grain yield have been developed and commercialized worldwide [2,51].

Genetics of kernel carotenoids
The current study revealed that environments had minor effect on BC, BCX and proA. Muthusamy et al. (2015b) [41] also reported low effects of environment on carotenoids through analysis of 95 maize lines over the environments. Minor effect of G × E interaction on kernel carotenoids has been reported by Muthusamy et al. (2016) [49] and Goswami et al. (2019a) [29]. The minor effect of environment on kernel carotenoids thus enables identification of potential experimental hybrids adapted over diverse locations [18]. Combining ability is an important area of research in hybrid programmes [52]. It provides useful information in understanding the genetic nature of a trait and aids in selection of suitable parents for superior cross combinations. The result of genetic analysis brought out the importance of both nonadditive and additive gene action. In the present study, though dominance variance was predominant for proA, additive variance was important as well. However, earlier studies have reported the predominance of additive gene action on carotenoid accumulation in maize [49,53,54]. This minor variation is possibly due to different germplasm used in the study. This indicated that parental inbreds with high proA may further lead to higher accumulation of proA in hybrids [37]. Various authors have reported that genotype homozygous for favourable allele of crtRB1 possesses much higher proA compared to heterozygote [16,17,22]. Considering this, both lines and testers were made homozygous for harnessing the benefits of crtRB1 in all hybrids. Since, all the hybrids were homozygous for favourable allele of crtRB1, other modifier loci could be the reason for dominance effects for further influencing the proA. Several lines including testers were identified as the best general combiners for proA. MGUH-57 (14.90 μg/ g), MGUH-52 (14.60 μg/g), MGUH-27 (14.36 μg/g) and MGUH-32 (14.26 μg/g) possessed high proA, and had both the parents being high in GCA effects as well. Besides, MGUH-1, MGUH-14, MGUH-15, MGUH-18, MGUH-19, MGUH-28, MGUH-31, MGUH-48, MGUH-50 and MGUH-75 had one of the parents (used as line) having high GCA effects for proA. The inbreds with high GCA for proA thus serve as a promising inbreds in the future breeding programme [49].
Further, parents of these proA rich hybrids also produce high yield from seed production perspective. The average grain yield among parental inbreds (used as lines) developed under this programme was 3.1t/ha with a range of 2.6-3.5t/ha. The same in crtRB1-donor was only 1.4t/ha, thereby suggesting its poor adaptability in subtropical conditions. The high grain yield of new crtRB1-based inbreds depicts better adaptability. The high yielding proA rich hybrids identified here would thus provide more productivity and profit to the farmers, and offer higher vitamin-A to the consumers [2]. Further, chickens accumulate more proA in egg yolk when fed with proA rich biofortified maize grains [55]. The proA rich maize used directly as food and indirectly through eggs would provide sufficient vitamin-A required for proper growth and development in humans. These proA rich hybrids thus, assume great significance in food and nutritional security, and would play important role in alleviating VAD in the country.

Conclusion
Diverse crtRB1-based inbreds have been developed in the study through marker-assisted pedigree breeding. These new inbreds possessed significantly higher proA than the checks. Molecular characterization of the inbreds depicted their diverse genetic nature. Genetic analysis revealed that both additive and non-additive variances were important. The crtRB1-based inbreds were successfully used in development of proA rich hybrids. Promising high yielding hybrids with very high concentration of proA have been identified. These proA rich hybrids are higher yielding than the exiting proA checks and at par with the normal checks in grain yield but with higher proA. The present study demonstrated the successful application of markers-assisted pedigree breeding in broadening the genetic base, and developing promising hybrids with higher grain yield as well as improved nutritional quality.