SiDREB2-based SNAP Marker-Assisted and Multi-Trait Selection in The Early Generation of Foxtail Millet (Setaria italica L. Beauv.)

Setaria italica L. or foxtail millet is known for its nutritious grains and adaptability to unfavorable environmental conditions. High productivity, early heading, medium stature, and tolerance to drought-or salinity stress are among the breeding objectives for foxtail millet. The objective of this study was to select F 3 families of foxtail millet from the cross of Botok-10xICERI-6 by weighted selection index and assisted by SiDREB2 -based SNAP marker. Genotyping of 178 F 3 families using the SiDREB2 -based SNAP marker resulted in 29 A/A genotypes, 121 A/G genotypes, and 28 G/G genotypes. Further evaluation was conducted on 48 F 3 families consisting of 27 A/A genotypes and 21 A/G genotypes in an augmented randomized complete block design together with their parental genotypes (Botok-10xICERI-6) and three check genotypes (Mauliru-2, NTB-1, and Toraja). Plant height and heading time had high broad-sense heritability, whereas grain weight per plant had a moderate broad-sense heritability. Ten potential F 3 families were selected based on a weighted selection index with 20% intensity, comprised of seven A/ G genotypes and three A/A genotypes with a weighted selection index ranging from 0.84 to 3.76. The F 3 family with pedigree numbers B10I6-15-136, B10I6-15-161, and B10I6-15-70 with A/A genotypes are considered putative transgressive segregants and could be continued to the next generation for further breeding process. Copyright


INTRODUCTION
Millet is a group of underutilized small-seeded cereals from the Panicoidae subfamily commonly cultivated in areas with water scarcity (Panchal et al. 2023).India, African countries, and China are the top three global millet producers (FAO 2021).One of the major millet produced globally is foxtail millet (Setaria italica L. Beauv) which ranks second after pearl millet (Pennisetum glaucum) (Panchal et al. 2023).The other members of millet include barnyard millet, finger millet, kodo millet, little millet, and proso millet (Saini et al. 2021).Foxtail millet is considered a functional food due to its nutritional benefits, including its low glycemic index and high contents of protein, dietary fiber, and antioxidant in its grain (Arora et al. 2023).Additionally, several health benefits have been reported for foxtail millet, including cancer (Zhang & Liu 2015) and cardiovascular disease (Jali et al. 2012) prevention.Broad adaptation of foxtail millet to unfavorable environmental conditions, including drought (Xiao et al. 2021) and salinity (Ardie et al. 2015;Han et al. 2022) has increased the importance of this species in marginal areas.Despite the remarkable benefits of this species, foxtail millet is not a popular food crop in Indonesia.
Recombination breeding through hybridization is a conventional yet useful strategy for generating a superior variety (Al-Khayri et al. 2019).However, the self-pollinated nature, floral morphology, tiny size of the flower, and anthesis behavior are the main challenges in the hybridization of foxtail millet (Moharil et al. 2019;Nagaraja et al. 2023) leading to no Indonesian superior variety of foxtail millet has been released to date.Nugroho (2020) induced male sterility in foxtail millet by warm water treatment to facilitate artificial hybridization of Indonesian local foxtail millet genotypes, namely Botok-10 and ICERI-6.Botok-10 is a local foxtail millet genotype from East Nusa Tenggara with relatively high potential productivity but with tall plants, and late heading time.Meanwhile, ICERI-6 is one of the foxtail millet collections in the Indonesian Cereals Research Institute (ICERI) with moderate plant height, and early heading time but low potential productivity (Ratnawati et al. 2024).Furthermore, molecular assessment using the SiDREB2-based SNAP marker categorized ICERI-6 as a tolerant genotype, while Botok-10 as a sensitive genotype to salinity or drought stress (Widyawan et al. 2018).A suitable selection strategy is necessary to identify progenies with high productivity, moderate plant height, early heading, and tolerance to drought or salinity stress from the recombination of Botok-10xICERI-6.
One of the selection methods commonly applied for multiple traits is weighted index selection, in which relative weights are used for traits of interest (Moeinizade et al. 2020).A weighted index selection based on productivity, heading time, and plant height was used to select superior F 2 individuals from the crosses of ICERI-5 x Botok-10 (Sintia et al. 2023).However, this method was not able to identify F 2 individuals potentially tolerant to drought or salinity stresses.Plant abiotic tolerance evaluation requires proper experimental design and replications for an accountable result (Negrão & Julkowska 2020).Therefore, phenotypic selection for abiotic stress tolerance is impractical to be performed in the early generation of a segregating population.Markerassisted selection (MAS) with proper molecular markers is expected to overcome such challenges in early-generation selection (Hasan et al. 2021).
The dehydration-responsive element binding (DREB) is a plant transcription factor involved in the complex regulatory tolerance mechanisms to drought and salinity stresses in many plants (Singh & Chandra 2021).A DREB2 homolog in foxtail millet, SiDREB2, was reported to possess single nucleotide polymorphism (SNP) at the 558 th nucleotide (an A/G substitution), and this SNP was further associated with drought tolerance in foxtail millet (Lata et al. 2011).A SiDREB2based single nucleotide amplified polymorphism (SNAP) marker was further developed by Widyawan et al. (2018) to estimate the tolerance level of foxtail millet to drought or salinity.As part of foxtail millet breeding through hybridization, the objective of this study was to select F 3 families from the cross of Botok-10xICERI-6 by a weighted selection index and assisted by SiDREB2-based SNAP marker.

Genetic materials
The parental genotypes Botok-10 (a local foxtail millet genotype from East Nusa Tenggara, Indonesia) and ICERI-6 (a collection of Indonesian Cereals Research Institute, ICERI), and 178 F 3 families from the cross of Botok-10 and ICERI-6 were used as genetic materials in this experiment.Seeds from each parental genotype and F 3 family were sown in two tray holes with ten seeds per hole in seedling trays containing compost and manure (1:1, v/v).The shoot parts of 14-day-old seedlings were harvested as a bulk sample (10-20 seedlings per F 3 family number) and were preserved in a 2 mL microtube containing 700 µL CTAB (Cetyl-Trimethyl Ammonium Bromide) buffer at -20 o C for further DNA isolation.
Total Genomic DNA isolation and DNA amplification.The CTAB method (Doyle & Doyle 1990) was used to extract total genomic DNA from the shoot parts of 14-day-old seedlings with slight modification namely, we exclude the use of 0.2% (v/v) 2-mercaptoethanol in the lysis buffer.The SiDREB2-based SNAP markers consisted of two forward primers and one reverse primer as listed in Table 1.The PCR reaction with a total volume of 10 μL consisted of genomic DNA (2.5 μL, 12 ng.μL - ), forward (SD2-558-SNP-A or SD2-558-SNP-G) and reverse primer (SD2-558-SNP-Rev) (2.5 μL, 10 pmol), and 5.0 μL of 2× PCR mix (KAPA2G Fast HotStart ReadyMix, Sigma-Aldrich, Germany).The PCR was performed using Esco's Swift Maxi Thermal Cycler (Esco Technologies, Singapore) following the PCR profile reported by Ratnawati et al. (2024).
Analysis of molecular data Successful amplification using forward primers (SD2-558-SNP-A or SD2 -558-SNP-G) and reverse primer (SD2-558-SNP-Rev) resulted in a 300 bp amplicon.Amplicons were analyzed by electrophoresis at 90 volts for 40 minutes in 1x TAE buffer on 1.5% (w/v) agarose gel.The agarose gels were immersed in ethidium bromide solution (0.5 µg.mL -1 ) prior to gel visualization using a UV transilluminator (AlphaImager® Mini).The SiDREB2-based SNAP marker-assisted selection was conducted by evaluating the presence of a 300 bp band for the A allele, G allele, or both A and G alleles in particular F 3 family (Figure 1).The band specific for the G allele appeared in the female parent genotype (Botok-10), while the band specific for the A allele appeared in the male parent (ICERI-6).Weighted index selection of Botok-10xICERI-6 derived F3 family Plant materials Fifty F 3 family numbers were further selected from the above 178 F 3 families for further field evaluation.These 50 F 3 families consisted of 29 A/A genotypes and 21 A/G genotypes based on SiDREB2-SNAP marker.However, two F 3 family numbers with A/A genotypes failed to grow in the field and further analyses were conducted on the remaining 48 F 3 family.Five check genotypes used include the parental genotypes (Botok-10 and ICERI-6) and three local genotypes (Toraja, NTB-1, and Mauliru-2).The Toraja genotype originated from Sulawesi, the NTB-1 genotype originated from West Nusa Tenggara (NTB-1), and the Mauliru-2 genotype originated from East Nusa Tenggara.

Procedure
The experiment was conducted from January to May 2023 in the Cikabayan Bawah Experimental Station of IPB University, Bogor, Indonesia (6°33'24.23"S,106°43'33.4"E).The agro-climates conditions during this experiment were recorded to be 21.43°Caverage temperature, 87.82% average humidity, and 1,100 mm per month average rainfall (BMKG 2023).This experiment was arranged in an augmented randomized complete block design with five replicates.Each replicate was a 40 m x 0.8 m size block consisting of ten F 3 family numbers and five check genotypes.Each F 3 family number and check genotype were planted in three rows, resulting in 45 planting rows per block.Each row consisted of eight plants, with plant spacing of 75 cm x 10 cm.Seeds were subjected to hot water treatment to reduce the risk of seed-borne fungi as described by Parlindo et al. (2022).Treated seeds were then directly sown in planting holes containing 3% Carbofuran.Fertilizers of SP-36 (150 kg.ha -1 ) and KCl (75 kg.ha -1 ) were applied two weeks after planting (WAP), while urea was applied two times at 2 and 6 WAP with the rate of 150 kg.ha -1 at each application.A plant net was installed at 2 WAP to prevent crop loss due to birds.
The observation was conducted for 11 characters according to the UPOV descriptor (UPOV 2013) on the following characters: plant height (cm), the length and width of flag leaf (cm), stem diameter (mm), heading time (DAP), harvest time (DAP), 100-grain weight (g), the length (cm) and weight (g) of main panicle, main panicle grain weight (g), and grain weight per plant (g).
Scatter plots were built using Microsoft Excel based on the selection index (Y-axis) and means of standard deviation of the three targeted traits (X-axis).The mean of the standard deviation of five check genotypes is indicated by the vertical dashed line, while the mean of SI calculated from 48 F 3 families is shown by the horizontal dashed line.Only 25 F 3 families with a minimum of 12 observable plants per family were mapped in the plot.

RESULTS AND DISCUSSION
SiDREB2-based SNAP marker-assisted selection of F3 family derived from Botok-10xICERI-6 cross Molecular markers have been widely used to improve abiotic stress tolerance in crops (Younis et al. 2020).Our study suggests that markerassisted selection using the SiDREB2-based SNAP marker is a simple method to select potentially drought-or salinity-tolerant lines in an early segregating population.Figure 1 shows the representative visualization of amplicons using a particular primer pair.The A/G genotype was indicated by 300 bp amplicons produced by both SD2-558-SNP-A/Rev and SD2-558-SNP-G/Rev primer pairs.The A/A genotype only showed the 300 bp amplicon produced by SD2-558-SNP-A/Rev primer pair, while the G/G genotype only showed the 300 bp amplicon produced by SD2-558-SNP-G/Rev primer pair.The 300 bp amplicons produced by the SD2-558-SNP-A primer indicate tolerant genotypes, while amplicons produced by the SD2-558-SNP-G primer indicate sensitive genotypes (Widyawan et al. 2018;Ratnawati et al. 2024).Drought-or salinitytolerance estimation using SiDREB2-based SNAP markers on 178 F 3 families derived from Botok-10xICERI-6 cross resulted in 29 A/A genotypes, 121 A/G genotypes, and 28 G/G genotypes.There were no significant differences in plant height, heading time, and grain weight per plant between the A/A, A/G, and G/G genotypes (Table 2), indicating that there were no associations between the 558 th base variation of the SiDREB2 gene and the observed phenotypic traits at the F 2 generation under non-stress conditions.The selection index calculated based on the three previously mentioned traits showed that the A/A genotype's selection index ranged between -3.69 to 12.67, while the A/G and G/G genotypes' selection index ranged from 5.79 to 15.39 and -5.69 to 11.13, respectively.Lata et al. (2011) reported that SiDREB2 gene expression increased under drought or salinity conditions, indicating that the effect of the SiDREB2 allele would be more pronounced under stress conditions.Therefore, the effect of the 558 th base variation of the SiDREB2 gene needs to be further evaluated under stress-and no-stress conditions in the later generation.Further field evaluations were then conducted on 50 F 3 family numbers having the A allele (A/A or A/G genotypes) with the highest selection index from the 178 F 3 family numbers evaluated above.

Weighted index selection of Botok-10xICERI-6 derived F3 family
Some of the major traits targeted in the foxtail millet breeding program are high yield, early heading time, medium stature, and tolerance to drought/ salinity stress.Shorter plants and earlier heading times in comparison to the female parent (Botok-10) were observed in the F 3 population (Table 3).Furthermore, the F 3 population showed higher grain weight per plant than the male parent (ICERI-6), indicating that there are potential segregants with higher yields in the F 3 population.The F 3 population showed higher average values of the length and width of flag leaf, and 100-grain weight than both parents, while the remaining traits showed average values between the two parents.
The F 3 population in this study showed a lower standard deviation for plant height than the two parents, while the standard deviation for heading time, and grain weight per plant were in between the two parents.This indicates that although the phenotypic variation for these target traits was lower than at least one of the parental genotypes, further selection is still necessary for more uniform performances in the next generation.
In order to better understand the extent of genetic variability in the F 3 population, the variance components, phenotypic and genotypic coefficient of variation, and broad-sense heritability were calculated and presented in Table 4 Table 3. Mean value and standard deviation of Botok-10, ICERI-6, and F 3 population from the cross of Botok-10 and ICERI-6.
Note: DAP: days after planting genetic variability, whereas traits with low GCV values demonstrate low levels of genetic variability.Meanwhile, the extent of the differences between the PCV and GCV implies the relative significance of genetic and environmental influences on a given trait, with large differences indicating a significant environmental influence and small differences indicating a significant genetic influence (Xu 2021).Therefore, successful selection for targeted traits can be expected from traits with high GCV and with minimum differences between PCV and GCV.Moderate GCV and a small difference between PCV and GCV (6.64) were recorded for plant height, while low GCV and a small difference between PCV and GCV (2.54) were recorded for heading time.Grain weight per plant showed high GCV and a relatively greater difference between PCV and GCV (20.47).These results indicate that environmental influence was more pronounced for grain weight per plant, while genetic influence was more dominant for plant height and heading time.Moreover, moderate to high GCV estimates indicate that selection can be performed based on plant height and grain weight per plant, while the heading time was relatively less varied between F 3 families.Sintia et al. (2023) also reported moderate to high GCV estimates for the plant height and grain weight per plant of an F 2 population derived from the ICERI-5xBotok-10 cross of foxtail millet.Heritability also needs to be considered in determining effective selection traits.A high value of broad-sense heritability on a particular trait indicates that the total variability of the trait is under genetic control, and selection based on this trait would be advantageous for trait improvement (Schmidt et al. 2019).As shown in Table 4, plant height and heading time have high broad-sense heritability, while grain weight per plant has moderate broad-sense heritability.A previous study of an F 2 population derived from the ICERI-5xBotok-10 cross of foxtail millet by Sintia et al. (2023) showed high broad-sense heritability for grain weight per plant and moderate broad-sense heritability for plant height and heading time.Meanwhile, Anuradha and Patro (2020) reported that flowering time, plant height, and grain yield had the highest heritability values based on their study on eight foxtail millet genotypes in India.These different heritability estimates might be due to different parental genotypes as well as different breeding generations.Altogether, the PCV, GCV, and broad-sense heritability values in this study indicate that a weighted selection index based on the three main target traits would be effective and can be performed accordingly.Note: σ 2 p : phenotypic variance, σ 2 e : environmental variance, σ 2 g : genetic variance, PCV: phenotypic coefficient of variation, GCV: genotypic coefficient of variation, h 2 bs : broad-sense heritability, DAP: days after planting A weighted index selection with an intensity of 20% of 48 F 3 families from the cross of Botok-10 x ICERI-6 resulted in ten F 3 families (Table 5).The top ten F 3 families comprised three A/A genotypes and seven A/G genotypes, with the selection index ranging from 0.84 to 3.76.The selection indexes of the top ten F 3 families were higher than the parental genotypes and all check genotypes, except the Mauliru-2 genotype.The Mauliru-2 genotype showed a considerably high selection index (3.03).Ratnawati et al. (2024) also identified Mauliru-2 as a potential high-yielding genotype compared to the other seven Indonesian foxtail millet genotypes.Given that Mauliru-2 has the G allele for the SiDREB2 gene, this genotype is potentially developed further as a superior foxtail millet variety through pure line selection for nonstressed areas.
The variability between F 3 individuals within a particular F 3 family can be seen from the mean standard deviation of the three target traits.The top ten F 3 families still showed a greatly varied mean standard deviation from 4.64 to 13.35.The F 3 family with a high selection index and a low mean standard deviation is desirable to be selected and can be classified as putative transgressive segregants.Considering the check genotypes were planted in five blocks, while each F 3 family number was planted in only one block, the mean standard deviation of the check genotypes could be used as a suitable comparison to identify putative transgressive segregants in the F 3 population derived from Botok-10xICERI-6 cross.The scatter plot in Figure 2 shows the distribution of 25 F 3 families based on their selection index and the mean standard deviation of the three target traits used to develop the selection index.Genotypes in quadrants (I) and (IV) are those with selection index values lower than the mean selection index of all F 3 families observed, thus they were considered not potential to be selected further.Meanwhile, genotypes in quadrants (II) and (III) are those with selection index values higher than the mean selection index of all F 3 families observed.The F 3 family with pedigree numbers B10I6-15-136, B10I6-15-161, and B10I6-15-70 with A/A genotypes are considered putative transgressive segregants since they are located in quadrant (I).These three F 3 families have a higher selection index than the mean selection index of all F 3 families and a lower mean standard deviation than the mean standard Note: DAP: days after planting; Mean SD = mean of the standard deviation of the three target traits deviation of the check genotypes.Moreover, these three F 3 families have A/A genotypes that indicate their potential tolerance to drought/ salinity stress.The Mauliru-2 genotype is also located in quadrant (II), confirming its potential as a superior foxtail millet variety.Although it is located in quadrant (III), the pedigree number B10I6-15-177 with A/G genotype showed the highest selection index.Therefore, F 4 families with A/A genotype generated from F 3 individuals in this pedigree also potential to be evaluated further.

CONCLUSIONS
Multiple-traits selection using SiDREB2-SNAP marker combined with weighted selection index on F 3 families of foxtail millet from the cross of Botok-10xICERI-6 identified 10 potential F 3 families with the highest selection index.Three F 3 families with A/A genotypes (pedigree numbers B10I6-15-136, B10I6-15-161, and B10I6-15-70) are considered putative transgressive segregants and are recommended to be continued to the next generation for further breeding process.

Figure 1 .
Figure 1.Representative gel electrophoresis result of A/G, A/A, and G/G genotypes using the SiDREB2-based SNAP marker.

Table 1 .
The SiDREB2-based SNAP marker used in this experiment.

Table 2 .
. Traits with high GCV values indicate high levels of The effect of A/A, A/G, and G/G genotypes on plant height, heading time, and grain weight per plant of F 2 generation derived from the Botok-10xICERI-6 cross.

Table 4 .
Variance component, phenotypic coefficient of variation, genotypic coefficient of variation, and broadsense heritability of F 3 population from the cross of Botok-10 and ICERI-6.

Table 5 .
Weighted index selection results in the F 3 population of Botok-10xICERI-6 cross.