Consumer demand for fresh potato in the U.S. has been steadily decreasing since the 1990s, and this has directly impacted growers’ profitability and has been threatening the economic sustainability of the U.S. potato industry. Although the reasons for declining potato consumption are complex and have no single cause, negative comments from some nutritionists about potato in the diet, such as its association with weight gain, has been detrimental to potato perception by consumers and, therefore, to potato sales. In a 2010 survey conducted by the U.S. Potato Board, 55% of respondents reported various health-related statements as the worst things about potatoes (e.g. “Any good ways to prepare them are unhealthy, or at least have no nutritional value.”). In addition, a survey by the Oregon Potato Commission determined that the three factors that most influence consumers’ decisions to purchase potatoes are appearance, flavor and nutritional value. These surveys reveal that health is the biggest challenge for potatoes.
In an effort to remedy negative publicity about potato in the diet, there has been an increased demand from the stakeholders to develop cultivar-specific nutrition profiles with an aim toward advertising positive nutritional attributes and to focus breeding efforts toward enhancing the nutritional value of fresh potatoes in addition to other traditionally targeted traits such as yield potential, biotic and abiotic stress resistance, processing qualities, bruising and shrinkage resistance, storability, and appearance. The Oregon Potato Breeding and Variety Development program plays a key role in the Tri-State (Idaho, Oregon and Washington) potato variety development program, with research focus on the aforementioned traits.
Vitamin B9 deficiency is one of the most widespread nutritional deficiencies worldwide and is associated with the increased risk of birth defects (e.g., spina bifida, anencephaly), strokes, cardiovascular diseases, anemia, some types of cancers, and impairment of cognitive performance. Good nutrition is the basis of a healthy and productive life, and potatoes should have a prominent role to play since it is part of our diet. Vitamin B9-enriched potatoes will help sustain the economic viability of potato farms and will enhance the basal intake level of nutrients by consumers.
Thus, the Graduate student’s project focused on (1) exploring the genetic diversity of potato for folate content, and (2) identifying genetic factors that control folate accumulation.
First, 250 individual plants from 77 accessions and 10 Solanum species were screened for their folate content using a tri-enzyme extraction and microbial assay. The screening focused on species that had been previously shown to have individuals with high folate content. These species were Solanum tuberosum subsp. andigenum, Solanum vernei and Solanum boliviense. Other species that had never been analyzed for folate content before, S. stipuloideum, S. chacoense subsp. chacoense, S. candolleanum, S. acaule, S. demissum, S. microdontum, and S. okadae, were also evaluated. There was a 10-fold range of folate concentrations among individuals tested. Certain individuals within the species Solanum tuberosum subsp. andigenum, Solanum vernei and Solanum boliviense have the potential to produce more than double the folate concentrations of commercial cultivars such as Russet Burbank. These results show that exploring the genetic diversity of potato is a promising approach to increase the folate content of this important crop.
Second, to better understand the regulatory mechanisms that control folate accumulation in potato tubers, the expression of genes involved in folate metabolism was determined in high and low folate tuber samples using RNA-sequencing (RNA-Seq) and Real Time Quantitative RT-PCR (qPCR) analyses. RNA-Seq analysis showed that among folate biosynthesis and salvage genes, γ-glutamyl hydrolase 1 (GGH1) was consistently expressed at higher levels in high folate compared to low folate segregants of a Solanum boliviense accession. qPCR analysis was used to determine GGH1 expression in eight additional pairs of folate segregants. Results showed that GGH1 transcripts levels were higher in high folate compared to low folate segregants for seven out of eight pairs of folate segregants analyzed. These results suggest that GGH1 gene expression may be a determinant of folate content in potato tubers and may be considered as a target for folate engineering.
Third, an F2 population of 94 individuals from a cross between a high and a low folate genotype was evaluated for folate content and genotyped for Single Nucleotide Polymorphism (SNP). More than 3,000 high quality SNPs were used to assemble maps for each of the 12 potato chromosomes. SNP-trait association analysis and QTL single marker analysis was performed in order to find SNPs and genomic regions that associate with high folate content. SNPs associated with folate content were located on chromosomes 3, 6, and 7. Future research should focus on validating these SNP markers.
Potato consumption in the U.S. has been steadily declining since the 1990s. This has had a considerable impact on the economic sustainability of the potato industry, making the business of growing and selling potatoes challenging and less profitable. One of the key consumer trends which are attributed to the downturn in consumer demand for potatoes is the fact that consumers are becoming increasingly interested in making healthier food choices, and they are often dragged down by outdated beliefs that potatoes are unhealthy. Anti-potato advocates include Walter C. Willett at the Harvard School of Public Health who views potatoes as a starch, not as a vegetable, and who recommends to use potatoes sparingly. They are ignoring that the potato is packed full of vitamins, minerals, and nutrients. For instance, a medium-size potato provides 45% of the daily value (DV) for vitamin C, 20% of the DV for potassium, and12% of the DV for dietary fiber. The potato was also found to be the third most important source of vitamin B9, also known as folate (Figure 1), in the Dutch diet and provided 9-12% of the total vitamin B9 in a Norwegian study. Therefore, in order to secure increased potato use, promotional campaigns which dispel the myths of potatoes as unhealthy, “empty carbs” should continue to remind consumers that fresh potatoes are ‘vegetables’ and that they are a healthy choice to be included in their diets.
A complementary strategy which aims at increasing potato use, and thereby sustaining the economic viability of potato farms, is to breed potato for increased amounts of micronutrients essential for human health, especially white-fleshed varieties which are the most consumed form in the United States. This is particularly relevant because micronutrient deficiency is responsible for millions of deaths every year, affecting half of the world’s population, especially children, women, and the elderly of not only poor populations but also industrialized countries like the United States. Vitamin B9 deficiency is a common health issue worldwide and is most often caused by a dietary insufficiency. Vitamin B9 deficiency is linked to various serious diseases such as neural tube defects (NTDs), cardiovascular diseases, certain types of cancers, stroke, anemia, and impairment in cognitive performance. NTDs are among the most common congenital and devastating birth defects. The defects occur between the 21st and 27th days after conception, a time when many women do not realize they are pregnant. In the United States, despites the implementation of folic acid food fortification programs in staple foods such as rice and pasta since 1998, vitamin B9 intake remains suboptimal. For instance, the population of women aged 15–44 years who consume more than the Recommended Daily Allowance of 400 μg/day of folate has not yet reached the FDA’s 50% target. Folate-rich diets have been associated with decreased risk of cardiovascular disease, with reported 55% lower risk of an acute coronary event in men who consumed the most dietary folate.
The purpose of this research was to further explore folates’ natural diversity within potato germplasm, better understand the regulation of folate levels, and to begin the development of molecular tools to assist breeding efforts that aim at increasing folate content in commercial potato cultivars.
Objective #1: Exploring folate diversity in wild and primitive potatoes for modern crop improvement
The objective was to identify sources of high folate germplasm for introgression into modern cultivars. Two hundred and fifty individual plants from 77 accessions and 10 species, as well as a Russet Burbank standard, were planted in the greenhouses on OSU’s main campus in May 2014. Samples were harvested in early November of 2014.
Objective #2: Transcriptome analysis in low- versus high-vitamin B9 genotypes
In this objective, gene expression between low and high folate genotypes was compared to identify differentially expressed genes that may control folate content.
Objective #3: Single Nucleotide Polymorphism markers associated with high folate content from wild potato species
The objective was to identify SNPs and regions of the genome that may contribute to high folate levels. A selection from Solanum boliviense PI 597736 that showed high tuber folate content (referred as Fol 1.6) was crossed with the low/medium folate recombinant inbred clone USW4self#3 (labeled USW4s#3) to generate an F1 progeny. Twelve of the resulting F1 seedlings were intermated to produce an F2 population named BRR3. One hundred and fifty of these F2 seeds were planted, but only 94 of the individuals produced tubers for further analysis. These individuals were used for linkage group mapping, SNP genotyping, and QTL single marker anaylsis.
Objective #4: Extension activities.
Hydromania summer camp in the summer of 2015. Poster presentation at Potato Field Day at the HAREC station 2015. Farmtastic extension center day camp for kids 8-12 years old in the summer of 2015. Seminar given to the professional mentoring committee (growers from the area) during the summer 2015. Seminar given in the graduate student competition at the Potato Association of America annual meeting in Portland, ME 2015.
Folate screening for wild and primitive cultivated species included 250 individual plants from 77 accessions and 10 species (S. stipuloideum, S. chacoense subsp. chacoense, S. candolleanum, S. acaule, S. demissum, S. microdontum, S. okadae, S. tuberosum subsp. andigenum, S. boliviense, S. vernei) (Table 1). The species S. tuberosum subsp. andigenum, S. boliviense and S. vernei were selected based on previous data, which showed that they could contain accessions with high folate. The species S. stipuloideum, S. candolleanum, S. acaule, S. demissum, S. microdontum and S. okadae were selected because no or very few accessions within these species had been previously evaluated. S. chacoense subsp. chacoense was evaluated because it is one of the most widely distributed wild potato species. Russet Burbank, a commercial variety largely grown in North America, was used as the standard. Seeds of wild and primitive cultivated species were obtained from the U.S. potato gene bank (USDA Agricultural Research Service Germplasm Resource Information Network (GRIN), www.ars-grin.gov). Seeds were soaked in GA3 at 1000 mg/L overnight before planting to Metro-mix in June 2014. When plantlets reached about 8 cm high, they were transplanted in 8 cm square pots containing Sunshine® LA4 P. All-purpose fertilizer 20-20-20 was applied at 200 mg/L once a week until senescence. Plants were watered twice a week until senescence. Vines were killed on October 31, 2014, and tubers were harvested on November 11. Greenhouse temperature was set at 21 °C day time and 15 °C night time. Supplemental light was provided for 14 h per day from a mixture of 400-Watt high pressure sodium and 1000-Watt metal halide lamps. One to four individual plants per accession, with a minimum of three in most instances, were grown. One individual plant is a plant from one botanical seed. A representative set of tubers from one individual plant was pooled and processed together as follows. Tubers were left with skin intact, washed with cold water in a strainer, weighed and then flash-frozen with liquid nitrogen before storage at −80 °C. A few tubers from each genotype were stored at 4 °C as back-up for re-planting. Frozen samples were then lyophilized in a freeze-dryer (VirTis Benchtop K) (vacuum pressure <100 mTorr) for two to three days. Dried samples were weighed, and the initial moisture content was calculated by the weight difference before and after freeze-drying potato samples. Removal of water from tuber samples allows for a more consistent comparison of vitamin content among samples, because moisture content varies greatly in these materials (68% to 82%). Samples (i.e., one sample is made of several tubers from one individual plant) were then ground to a fine powder with a Waring blender and transferred to scintillation vials for long-term storage at −80 °C.
For gene expression and SNP analysis, a selection from Solanum boliviense PI 597736 that showed high tuber folate content (referred as fol 1.6) was crossed with the low/medium folate recombinant inbred clone USW4self#3 (labeled USW4s#3) to generate an F1 progeny. Twelve of the resulting F1 seedlings were intermated to produce an F2 population named BRR3. In addition, two high (named fol1.6 and fol1.3) and 2 low (named fol1.5 and fol1.11) folate clones were re-propagated from stolon shoots in early May 2012, then tubers were harvested in November 2012 and evaluated for folate. True potato seeds from the BRR3 F2 population were soaked in GA3 at 1000 mg/L overnight before germination in June 2014. When plantlets reached approximately 8-cm high, they were transplanted in 8-cm square individual pots containing Sunshine Mix LA4P. All-purpose fertilizer 20-20-20 was applied at 200 mg/L once a week until senescence. Plants were watered twice a week until senescence. Vines were killed on October 31, 2014, and tubers were harvested on November 11. Greenhouse temperature was set at 21ºC day time and 15ºC night time. Supplemental light was provided for 14 hours per day from a mixture of 400 Watt high pressure sodium and 1000 Watt metal halide lamps. While 150 seedlings were grown, only 94 produced tubers and were used in the folate analysis.
Folates were extracted by using a tri-enzyme extraction method, as previously published. Potato samples (100 mg) were homogenized in 15-mL Falcon tubes containing 10 mL of extraction buffer consisting of 50 mM HEPES/50 mM CHES, pH 7.85, 2% (w/v) sodium ascorbate and 10 mM β-mercaptoethanol and deoxygenated by flushing with nitrogen. Once homogenized, samples were boiled for 10 min and cooled immediately on ice in a covered cooler. The homogenate was then treated with protease (≥14 units) and incubated for 2 h at 37 °C, boiled again for 5 min and cooled immediately in a covered cooler of ice. The samples were then treated with α-amylase (≥800 units) and rat plasma conjugase in large excess (0.5 mL/sample), incubated for 3 h at 37 °C, boiled again for 5 min and cooled immediately in a covered cooler of ice. After centrifugation at 3000 g for 10 min, the supernatant was transferred to a new tube. The residue was re-suspended and homogenized in 5 mL of extraction buffer, re-centrifuged for 10 min, and the supernatant was recovered. Supernatants were then combined and the samples’ volume adjusted to 20 mL with extraction buffer. Aliquots of each sample were transferred to 1.5-mL microcentrifuge tubes, flushed with nitrogen and stored at −80°C until analysis by the microbiological assay. Controls containing all reagents, but potato samples, were used to determine the amount of any residual folates in the reagents. There were no detectable folates in any of the reagents used.
Folate concentrations were measured by microbiological assay using Lactobacillus rhamnosus. L. rhamnosus (ATCC 7469) cultures were obtained from the American Type Culture Collection (Manassas, VA, USA). Glycerol cryoprotected cells of L. rhamnosus were prepared as described previously. Assays were performed in 96-well plates (Falcon microtiter plates). Wells contained growth medium supplemented with folate standards or potato extracts, each plated in triplicate. Bacterial growth was measured at 630 nm after 18 h, 21 h and 24 h of incubation at 37 °C. The 24-h reading was usually used for analysis unless saturation was reached, in which case, the 21-h reading was used. All measurements were made with a BioTek Instrument EL 311 SX microplate auto-reader (BioTekInstrument, Winooski, VT, USA), analyzed with the KCJr EIA application software (BioTekInstrument, Winooski, VT, USA) and compiled in Microsoft Excel. Final results were calculated by reference to a standard curve using 5-formyl-THF and expressed as nanograms of folate per gram of dry sample.
A large batch of dried potato powder was prepared from tubers of Solanum pinnatisectum PI 275233 and was used as the reference material. Each batch of extractions contained 18 samples plus the reference material. Values obtained for samples were normalized to values obtained for the reference material. The average folate concentration of the reference material across all of the extractions was 1105 ± 76 ng·g−1 DW. All calculations were performed with standard function settings in Microsoft Excel.
RNA was extracted using a modified hot phenol method as described previously. One hundred milligrams of freeze dried tuber powder (for qPCR analysis) or 1-2 g fresh tuber tissue (for RNA-Seq analysis) were added to a mixture of 4 ml pre-warmed phenol (pH 4.3) and 4 ml extraction buffer consisting of 100 mM LiCl, 100 mM Tris pH 8.5, 10 mM ethylenediaminetetraacetic acid, 1% sodium dodecyl sulfate, and 15 mM dithiothreitol. Samples were vortexed and incubated at 60°C for 20 to 30 minutes. Four ml of chloroform:isoamyl alcohol (24:1) were added to the solution, and the sample was vortexed and centrifuged at 9000 rpm for 10 min at 4°C. The aqueous phase was transferred into a new tube containing 4 ml phenol:chloroform:isoamyl alcohol (25:24:1), vortexed, and centrifuged at 9000 rpm for 10 min at 4°C. The previous step was repeated twice with phenol:chloroform:isoamyl alcohol (25:24:1) and twice with chloroform:isoamyl alcohol (24:1). RNAs were precipitated with one volume of 4 M LiCl, washed with 70% ethanol, and re-suspended in 50 µl diethylpyrocarbonate-treated water. Genomic DNA was removed by DNase treatment using the DNA-Free kitTM (Ambion, Austin, TX). RNAs were quantified and normalized to 200 ng/µl using a Nanodrop (Thermo Scientific, Wilmington, DE). For each genotype, two technical replicates were extracted. One seed provided one plant. Tubers from each individual plant were bulked, freeze dried, and ground together. Two RNA isolations were performed on freeze dried material from each individual plant.
Two repetitions of each clone fol 1.3, fol 1.5, fol 1.6, and fol 1.11 that were harvested in November 2012 were used for RNA extraction. One repetition is a bulk of tubers from 3 to 4 plants. RNA samples (duplicate of each of the clones fol 1.3, fol 1.5, fol 1.6 and fol 1.11) were bar coded, pooled, processed together, and sequenced in one Illumina HiSeq2000 lanes (51-cycle v3 Single End). Illumina library preparation was done at the Center for Genome Research and Biocomputing at Oregon State University using TruSeq RNA. Illumina libraries were quantified by qPCR for optimal cluster density. Mapping of the RNA-Seq reads to the DM potato reference genome, transcript assembly, and determination of differences in expression levels were performed using JEANS, a modified version of GENE-counter, in combination with NBPSeq. NBPSeq has an inbuilt function for count normalization. Pseudo counts associated with folate genes were expressed relative to pseudo counts for the β-tubulin gene.
One to 2 µg of RNA were converted to cDNA using New England BioLab’s M-MuLV reverse transcriptase (New England BioLabs, Ipswich, MA) and Oligo-dT18 primer (Thermo Scientific, Wilmington, DE). RNA template (5-10 µl) was mixed with 1 µl Oligo-dT18 and nuclease-free water to a final volume of 12 µl. This solution was placed in a 70°C water bath for 5 minutes and then cooled on ice. Eight microliters of reverse transcriptase (RT) master mix (composed of 2 µl 10X MuLV buffer, 2 µl 10 mM dNTPs, 0.25 µl M-MuLV reverse transcriptase, and 3.75 µl nuclease-free water) were then added to each sample. RT reactions were carried out on a Bio-Rad C1000 thermocycler (Bio-Rad Laboratories, Hercules, CA). The RT cycle was 25°C for 5 min, 42°C for 1 hr, 65°C for 20 min. Samples were then stored in a -20°C freezer until analysis. cDNA templates were diluted four times prior to use in qPCR reactions.
Real-time quantitative RT-PCR (qPCR):
All qPCR reactions were run on an Agilent Stratagene Mx3005P (Agilent Technologies, Santa Clara, CA) using Taqman environmental Mastermix II (Thermo Scientific, Wilmington, DE). The PCR cycle was: 95°C for 10 min followed by 40 cycles with the following steps: 95°C for 30 seconds, 60°C for 30 seconds and 72°C for 30 seconds. All threshold values were set within the Mx3005 analysis software. Primers for both elongation factor one-alpha (EF1-α) and γ-glutamyl hydrolase I (GGH1) were developed based on the DM potato reference genome and compared to sequences of commercial cultivars in order to design primers within conserved regions of the genes (Appendix A. 1.).
Genomic DNA isolation:
Approximately 15 mg of freeze dried tuber sample were homogenized in 600 µl CTAB extraction buffer (2% cetyltrimethyl ammonium bromide, 1.4 M NaCl, 20 mM EDTA pH 8.0, 100 mM Tris-Cl pH 8.0, 0.2% β-Mercaptoethanol) in a 1.7 ml microcentrifuge tube and incubated at 65°C for 1 hour with gentle mixing every 15 minutes. Cold chloroform (300 ul) was then added and the solution was vortexed briefly to form an emulsion. After centrifugation at 12000 rpm for 5 minutes, the aqueous phase was transferred to a new 1.7 ml microcentrifuge tube. An equal volume of cold isopropanol was then added, the tubes were inverted several times to mix and placed on ice for 10-15 minutes. The resulting mix is centrifuged at 12000 rpm for 10 minutes and the supernatant was removed and the pellet was washed with 300 µl of cold 70% ethanol. Samples were again centrifuged for 2 minutes at 3500 rpm. The washing step with ethanol was repeated 2 more times, then the pellet was re-suspended in 100 µl deionized water. Genomic DNA extracts were treated with 1 µl RNase A for 1 hour at 37°C with gentle mixing every 15 minutes. RNase A was then inactivated by incubating samples in a water bath at 65° C for 5-10 minutes. Samples were then centrifuged quickly to remove bubbles and placed on ice. Samples were further cleaned by treatment with phenol and chloroform. Phenol:chloroform:isoamyl alcohol 25:24:1, pH 8.0 (100 µl) was added to the extracts in a chemical fume hood. Samples were vortexed briefly to form an emulsion. After centrifugation at 13000 rpm for 10-15 minutes, the aqueous phase (~90-100 µl) was transferred to a new microcentrifuge tube. Cold chloroform (100 µl) was then added and samples were vortexed for 10 seconds. After centrifugation at 13000 rpm for 15 minutes the aqueous phase (80-85 µl) was transferred to a new microcentrifuge tube. DNA was precipitated by addition of Na-acetate pH 5.3 (1/10 of the sample volume) and 95% ethanol (2.5 volumes) at -20°C for 2 hours or overnight. Samples were then centrifuged for 30 minutes at 13000 rpm at 4°C. The supernatant was discarded and the pellet was washed three times with 300 µL of cold 70% ethanol as described above. Pellets were re-suspended in 30 µl deionized water.
At least 400 ng of genomic DNA from 96 samples (94 progeny and two parents) were loaded onto a 96-well microplate and desiccated in an Eppendorf Vacufuge Plus (Eppendorf North America, Hauppauge, NY) in 30 min intervals at 25°C until all samples were completely dried. Samples were then sent to GeneSeek (© Neogen Corporation, Lincoln, NE) for custom SNP Profiling using the Illumina platform Infinium SolCAP 12K array.
SNP quality and filtering:
SNP data were imported to the Illumina GenomeStudio software (Illumina, San Diego, CA) where it was analyzed and allele calls were assigned. A three cluster, or diploid model was used to genotype the samples. For calling SNPs using the diploid model on the v2 SolCAP 12K array, auto-clustering was run in GenomeStudio using standard settings and followed by importing the 3 cluster calling files for the v1 SolCAP 8303 SNP array. The genotypes were then exported for further data filtering. After exporting the SNP calls, the data were filtered to remove “BAD” and “QUESTIONABLE” SNPs based on quality comments from the GeneSeek’s data summary file. SNPs with multiple hits to the potato genome pseudomolecule sequence DM v 4.03 were then removed. After this initial filtering, there were 10120 SNPs remaining. Further filtering for 10 % missing values resulted in 9590 SNPs that were used for SNP-trait association. Monomorphic SNPs and SNPs with no-calls for more than three individuals were removed leaving 3556 SNPs for linkage mapping.
Linkage mapping and association analysis:
Before being imported into JoinMap (JoinMap ® version 4.1, Kyazma B.V., Wageningen, Netherlands), the 3556 SNPs were coded for a cross-pollinated (CP) mapping population type. Loci heterozygous in the first parent (fol 1.6) were coded as lm x ll type, loci heterozygous in the second parent (USW4s#3) were coded as nn x np type, while loci that were heterozygous for both parents were coded as hk x hk type. Coded SNPs were imported into JoinMap, checked for coding errors, and a population node was created. SNPs with chi squared (χ2) values greater than 5 were removed immediately and linkage groups were chosen based on the highest logarithm of odds (LOD) score for 12 distinct groups representing more than 1000 of the SNPs imported (usually LOD of 5 or greater).
Initial mapping was run using the regression mapping algorithm (Kosambi’s function) using linkages with recombination frequencies less than 0.4. Further curation of linkage group maps was done manually, removing SNPs that contributed to map distortion (χ2 > 3) or those that were missing more than one individual call. Linkage group maps were generated for both parents individually (lm x ll and nn x np), as well as a consensus map with all markers integrated (lm x ll, nn x np, and hk x hk) (Appendix A. 5.).
SNP trait association and QTL:
SNP-trait association and QTL single marker analysis was carried out using JMP genomics 7 (JMP, A Business Unit of SAS Cary, NC). In brief, the folate data was log transformed to fit a more continuous distribution. The chromosome locations provided from the SNP array was used for both analysis types. The consensus map that was generated was used to determine how close the mapping order is to the physical genome and the SNPs that were used in this mapping (as well as additional SNPs from the genotyping dataset) were further used for SNP-trait association and QTL single marker analysis. Datasets were placed in SAS format and uploaded into JMP Genomics. The SNP-trait association function was run for a continuous trait distribution, with non-delimited genotypes and with Benjamini and Hochberg correction (FDR). This analysis treats genotypes as categorical variables and uses an ANOVA function for genotypes. For the “trend” test, quantitative variables are created based on the number of each allele that makes up the individual’s genotype and a regression is performed. QTL single marker analysis was run using standard settings within JMP Genomics. In JMP Genomics, this function performs a simple regression for each marker with trait values, and outputs the probability of QTL evidence for each marker. This analysis is not as detailed as interval or multiple-interval QTL-mapping but can identify genomic regions to perform interval or multiple-interval mapping on. Both types of analysis were run using alpha values (p-values) of 0.005 in order to narrow the results to those that are most associated with folate content.
One-way analysis of variance (ANOVA) was performed to compare normalized mean values of folate content in all species. The only species that was significantly different from all other species at a p-value ≤ 0.001 was S. vernei. All statistical analysis was performed with R in R-Studio with the “stats” package linear regression and ANOVA functions.
Objective #1: Exploring folate diversity in wild and primitive potatoes for modern crop improvement
Two hundred and fifty individual plants from 77 accessions and 10 Solanum species were screened for their folate content using a tri-enzyme extraction and microbial assay. There was a 10-fold range of folate concentrations among individuals. Certain individuals within the species Solanum tuberosum subsp. andigenum, Solanum vernei and Solanum boliviense have the potential to produce more than double the folate concentrations of commercial cultivars, such as Russet Burbank (Table 2). The majority of individuals (55% of all individuals tested) had folate concentrations between 500 and 1000 ng·g−1 dry weight, including the modern variety Russet Burbank (Figure 3). About 40% of individuals had folate concentrations below 500 or between 1000 and 1500 ng·g−1 dry weight. The remaining 10% had folate concentrations above 1500 ng·g−1 dry weight, with thirteen individuals between 1500 and 2000 ng·g−1 dry weight and three above 2000 ng·g−1 dry weight. In most cases, a minimum of three individual plants per accession were evaluated for folate.
Each accession showed different levels of variability between individuals, with some accessions displaying a low level of variability (e.g., PI 197760), while other accessions had individuals with up to a five-fold folate concentration range (e.g., PI 320293). Two individuals with folate concentrations above 2000 ng·g−1 dry weight were from the accessions PI 225710 and PI 320377, both from the species S. tuberosum subsp. andigenum. Overall, about 25% of all individuals from the species S. tuberosum subsp. andigenum had folate concentrations above 1000 ng·g−1 dry weight (Table 2 and Figure 4). For S. boliviense, one individual within the accession PI 597736 contained folate concentrations of 1947 ng·g−1 dry weight. Only 7.5% of all individuals from the species S. boliviense had folate concentrations above 1000 ng·g−1 dry weight (Table 2 and Figure 4). For S. vernei, one individual from the accession PI 558149 had folate concentrations above 2000 ng·g−1 dry weight, and seven individuals from six different accessions (PI 320332, PI 458371, PI 458372, PI 473306, PI 500066 and PI 558149) contained folate concentrations greater than 1700 ng·g−1 dry weight. Over 66% of all individuals from this species had folate concentrations above 1000 ng·g−1 dry weight. Amongst the species S. stipuloideum, S. candolleanum, S. acaule, S. demissum, S. microdontum, S. okadae and S. chacoense subsp. chacoense, no individual had folate concentrations above 1500 ng·g−1 dry weight. S. demissum had the lowest maximum folate concentration (760 ng·g−1 dry weight), while S. candolleanum had the highest (1367 ng·g−1 dry weight). As previously demonstrated, the peel contains a substantially higher amount of folates than the potato flesh. Because tubers from wild and primitive species are usually small, the relatively higher contribution of the peel could be responsible for at least part of the high folate concentrations observed. However, when the tuber length of five to six representative tubers for each individual tested in this study was plotted against folate concentrations, the coefficient of correlation r was 0.03 (Figure 5). These results indicate that tubers of a similar size can have very different folate content and show that a tuber with a relatively higher amount of peel does not necessarily contain a higher amount of folates.
These results show that there is an enormous amount of genetic diversity within the germplasm that was evaluated for folate content. It also shows that S. tuberosum subsp. andigenum, S. boliviense and S. vernei all contain individuals that have the ability to produce and accumulate significantly higher concentrations of folate (over two-fold) in their tubers than a modern commercial variety, such as Russet Burbank. These individuals are promising materials for breeding potato with high folate content. S. tuberosum subsp. andigenum, S. vernei and S. boliviense were selected for evaluation based on a previous study with fewer individuals that showed that high folate concentrations could be found within these species. Of particular interest is the accession PI 225710 from the species S. tuberosum subsp. andigenum from which we found the highest folate concentrations (>2000 ng·g−1 dry weight). Another individual (clone named RN018.03) from the same accession that contained folate concentrations above 2000 ng·g−1 dry weight had previously been identified. Therefore, this accession may be a good source of high folate individuals. However, more individuals will need to be evaluated to confirm this hypothesis. One individual from the accession PI 320377 from the species S. tuberosum subsp. andigenum also had folate concentrations above 2000 ng·g−1 dry weight. Other individuals from this accession were previously reported in the low to mid-range folate levels. The highest folate concentration found within the species S. boliviense was in an individual from the accession PI 597736 (1947 ng·g−1 dry weight). This accession had previously provided individuals with average folate concentrations above 3000 ng·g−1 dry weight and may also be a good source of high folate individuals. It should be noted that the concentration of 3000 ng·g−1 dry weight was found in tubers that were stored at a cold temperature for six months. We have found that cold storage could significantly increase folate concentrations. Re-evaluation of one of these individuals has shown more modest folate concentrations (1500 to 2000 ng·g−1 dry weight) in freshly-harvested tubers. Five individuals from the species S. vernei had folate concentrations above 1900 ng·g−1 dry weight, including one individual from the accession PI 230468. We had previously found individuals from this accession with average folate concentrations above 1500 ng·g−1 dry weight. The accession PI 558149 had one individual with folate concentrations above 2200 ng·g−1 dry weight and an average folate concentration for four individuals higher than 1600 ng·g−1 dry weight. Therefore, these accessions may be other good sources of high folate individuals. The species S. vernei also had a large number of individuals with folate concentrations above 1000 ng·g−1 dry weight (49 out of 74 individuals). By comparison, S. boliviense only had seven out of 93 individuals with folate concentrations above 1000 ng·g−1 dry weight. S. vernei may therefore be a good species to further evaluate. No high folate (>1500 ng·g−1 dry weight) individuals were identified in the other species evaluated in this study (S. stipuloideum, S. candolleanum, S. acaule, S. demissum, S. microdontum, S. okadae and S. chacoense subsp. chacoense). Although one cannot preclude that high folate individuals could still be identified by extending the screening, our results also indicate that these species may not be the best genetic pool to screen for high folate individuals.
Although our data indicate that some accessions and some species may be better sources of high folate individuals than others, our results also illustrate the high degree of variability within accessions and species. Thus, until a larger number of individuals are being evaluated within each accession and species, it is currently difficult to pinpoint with high confidence a specific accession and/or species for high folate content. It should be emphasized that screening wild or primitive potato species for folate is very tedious. First, folate analyses are very time consuming. Second, tubers from wild and primitive species are difficult to produce; they are very small, most often between the size of a marble and a golf ball, and most species do not tuberize in the field, so they have to be grown in winter greenhouses or crossed with adapted cultivated forms. It would therefore be very helpful to identify predictors of high folate tubers. To this end, we are currently genotyping for single nucleotide polymorphisms (SNPs) a segregating population from a cross between a high folate S. boliviense individual (accession PI 597736) with a diploid S. tuberosum clone. We are also examining the possibility of correlation between leaf, seed and tuber folate content and other tuber characteristics, such as pH, which all could decrease the time and effort needed to screen a large number of individuals.
Once identified, high folate individuals should be used to introgress the high folate trait(s) into S. tuberosum tetraploid cultivars adapted for commercial production. The species evaluated in this study have different ploidy levels (2×, 4× or 6×) and belong to different crossability groups. S. tuberosum subsp. andigenum is cultivated like S. tuberosum cultivars and very easy to introgress. We have obtained hybrids from a cross between a high folate individual from the accession PI 225710 and a diploid S. tuberosum clone. These hybrids were grown in the field and are currently being evaluated for folate. Both S. boliviense and S. vernei, which had high folate individuals, should be very easy to move into the cultivar genepool by 2n gametes or by making 4× versions of the wild species.
If the high folate trait(s) are successfully introgressed into modern potatoes, such as Russet Burbank, new commercial potato cultivars could contain double the amount of folate compared to currently-grown cultivars. Based on the current per capita consumption of 50 kg per year, or 137 g per day in the United States, such a potato would provide around 11% of the recommended daily need of 400 μg, assuming 20% dry matter and 80% retention during cooking. The highest folate concentrations measured in this study (e.g., >2000 ng/g dry weight or >400 ng/g fresh weight, assuming 20% dry matter) were higher than those found in lettuce, snap beans and oranges (~300 to 380 ng/g fresh weight, according to the USDA Nutrient Database), for instance, but still much lower than high folate sources, such as beans, lentils and spinach (~2000 to 6000 ng/g fresh weight). The increase obtained by genetic means could be further increased by optimizing the time of harvest, as young tubers contain up to two-fold the amount found in mature tubers, the storage conditions, as folates accumulate up to two-fold in tubers stored at cold temperature, and the cooking method, as studies show the retention rate fluctuating between 50% and 110% depending on the cooking method and the cultivar. The reported folate retention rate for processed potatoes (i.e., French fries) is usually high (>75%) compared to boiled potatoes, for instance. This may be due to the shorter cooking time and the insolubility of folates in cooking oil during processing. In addition, potatoes destined for processing are often stored for several months at a cold temperature. With the increasing consumption of processed potatoes in the United States, introgression into a processing cultivar may have the most impact on the U.S. population’s folate intake.
Objective #2: Transcriptome analysis in low- versus high-vitamin B9 genotypes
Gene expression in high (fol 1.3 and fol 1.6) and low (fol 1.5 and fol 1.11) folate genotypes was determined by RNA-Seq (Appendix A. 2.). Fourteen genes that are known to be involved in folate metabolism were examined. Only one of these genes, GGH1, showed consistently and statistically different expression patterns between the high and low fol genotypes (i.e. log2 fold-change greater than 2 in 3 out of 4 comparisons, and greater than 1.6 in the remaining comparison). Other genes involved in the folate biosynthesis pathway showed some significant differences in at least one comparison, but results were not as consistent as GGH1. FPGS showed log2 fold-change greater than 2 in 2 of 4 comparisons, ADCL showed log2 fold-change greater than 2 in 3 of 4 comparisons but pseudocounts were too low (<4) for reliable determination in two comparisons, GTPCHI showed log2 fold-change greater than 2 in 1 of 4 comparisons, DHNAs showed log2 fold-change greater than 2 in 1 of 4 comparisons, DHFR showed log2 fold-change greater than 2 in 2 of 4 comparisons, UDP-glucose–pABA glucosyltransferase showed log2 fold-change greater than 2 in 1 of 4 comparisons, and DHNTP-PPase showed log2 fold-change greater than 2 in 2 of 4 comparisons. The greatest number of genes showing significantly differential expression patterns was found in the comparison of fol 1.3 and fol 1.11, with 8 of 15 genes showing log2 fold-change greater than 2.
To investigate whether the differential GGH1 gene expression observed in low and high folate fol lines was a consistent pattern between low and high folate genotypes, folate levels (Table 3) and GGH1 gene expression was determined in eight high and low folate individuals from the segregating populations BRR1 and BRR3, and from the species S. tuberosum subsp. angidenum and S. vernei by real-time quantitative RT-PCR (Table 4). Pairwise comparison between high and low folate samples within the fol populations showed significant differences in mean GGH1 expression, with fol 1.6/fol 1.11 and fol 1.6/fol 1.5 showing a 15-fold and 88-fold difference, respectively, and fol 1.3/fol 1.11 and fol 1.3/fol 1.5 showing a 24-fold and 140-fold difference. In 7 out of 8 comparisons GGH1 expression was higher in high folate versus low folate genotypes, with fold change ranging from 2 to 481 (Table 4). Only one pair of genotypes, BRR3 90 and BRR3 56, showed the inverse trend, with a 10-fold higher GGH1 expression in the low folate genotype (BRR3 56) compared to the high folate genotype (BRR3 90). High folate versus low folate genotypes from the species S. vernei showed the greatest difference in GGH1 expression (481-fold difference).
Understanding the regulatory mechanism of folate biosynthesis, salvage, and accumulation is critical for improvement of vitamin B9 content in staple crops such as potato. RNA-Seq and real-time quantitative RT-PCR analyses showed that, except in one case, GGH1 gene expression was higher in high folate versus low folate genotypes. These results indicate that GGH1 should be considered as a “gene of interest” in the regulation of folate content in potato tuber. Further confirmation of these results in a greater number of species and populations that segregate for folate content is warranted. It remains unclear how higher expression of GGH1 may lead to increased folate content in tubers. In Arabidopsis, GGH1 is a vacuolar enzyme that cleaves glutamate residues from polyglutamylated folate molecules that are stored in the vacuole. Overexpression of GGH in vacuoles caused 40-45% reduction in total folate, while knocking down GGH activity increased total folate content by 34%. Based on these results, one would expect that potato tubers that have higher GGH activity would have lower folate content. However, GGHs can also cleave the glutamate of pABA-Glu, a product of folate degradation, to free pABA that can re-enter the biosynthesis pathway. Therefore, it is possible that higher GGH1 activity increases salvage of pABA-Glu and subsequently folate biosynthesis. Establishing the subcellular localization and biochemical activities of GGH proteins in potato tubers will be necessary to further confirm this hypothesis. In addition, although we focused our study on the expression of GGH1 gene based on our initial RNA-Seq data, FPGS gene expression showed a similar trend to that of GGH1 (although it was not significant across all samples), with higher FPGS gene expression in high versus low folate potato tubers. With respect to FPGS, these results need to be confirmed across a greater number of samples and diverse germplasm. Because folate polyglutamylation depends on both GGH and FPGS activities, and the polyglutamylation level seems to determine folate homeostasis, at least in Arabidopsis, future studies should focus on the gene expression and enzymatic activity of both GGH and FPGS.
Objective #3: Single Nucleotide Polymorphism markers associated with high folate content from wild potato species
Folate determination in the BRR3 population. Folate concentrations in the F2 BRR3 population ranged from 304 ± 16 to 2952 ± 276 ng·g−1 DW, representing 11-fold difference between the lowest and highest folate concentrations. The majority of individuals tested (52% of all progeny) had folate concentrations between 500 and 1000 ng·g−1 DW. Approximately 40% of individuals showed folate concentrations below 500 or between 1000 and 1500 ng·g−1 DW. The remaining ~8% had folate concentrations above 1500 ng·g−1 DW, with four individuals between 1500 and 2000 ng·g−1 DW and three above 2000 ng·g−1 DW. The parents Fol 1.6 and USW4s#3 of BRR3 progeny had folate concentrations of 950 ± 130 and 639 ± 71, respectively. It should be noted that Fol 1.6 was previously tested in 2011 and 2012 for its folate content and was chosen as a parent in this cross because it showed consistent tuber folate levels of above 1600 ng·g−1 DW. Values presented here represent Fol 1.6 clones grown in 2015 which showed lower levels of folate.
SNP mapping and association analysis. Three individuals within the BRR3 population that were used in the SNP genotyping (BRR3-13, BRR3-23, and BRR3-118) had high no-call rates, but were kept in the study as the majority of SNPs with no-calls for these samples were filtered out. Calculations for percent heterozygosity (% Het) revealed that the S. boliviense parent (fol 1.6) was only 5.3% heterozygous, while the inbred parent (USW4s#3) was 34.8% heterozygous (Appendix A. 4). Only one individual within the progeny, BRR3-80, showed more than 50% heterozygosity. Average heterozygosity across the population is 19.9% and the heterozygosity ranged from 5.6% to 54.8%. Linkage maps were generated separately for SNPs segregating from Fol 1.6 and USW4s#3 (Appendix A. 6. and A. 7.) as well as the consensus linkage maps (Appendix A. 5.). The consensus map showed a total length of 1431 cM, with 530 nn x np markers, 62 lm x ll markers, and 52 hk x hk markers, corresponding to a coverage density of a marker per 2.22 cM (Table 5). Most large genetic mapping studies try to associate genetic markers, such as SNPs, to phenotypic traits and map those markers to their physical map location. Because fine-scale mapping requires sophisticated analysis and large data sets, tools like SNP-trait association in JMP Genomics can provide a simpler genome-wide scan for associations of markers to the trait of interest. SNP-trait association analysis identified 109 of 9590 SNPs (p-value 0.005) associated with folate content (Appendix A. 8.). These 9590 were selected from the total 12808 SNPs included in the solCAP array after basic filtering stages for quality of hybridization to the array. Significant SNPs were distributed unevenly across chromosomes 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, and to unanchored regions (named chromosome 0 afterwards). The vast majority of these SNPs (86% or 94 SNPs) were associated with chromosome 3, 6, and 7. These SNPs were closely associated with the regions of 34 Mb to 54 Mb, 36 Mb to 51 Mb, and 48 Mb to 55 Mb, respectively (Figure 6). We further examined the SNPs used in SNP-trait association for evidence of a QTL at each marker. QTL single marker analysis identified 80 of 9590 SNPs (p-value 0.005) that associate with folate content as potential QTLs (Appendix A. 9.). Significant SNPs were distributed unevenly across chromosomes 0, 3, 4, 6, 7, 8, and 9. The vast majority of these SNPs (94% or 75 SNPs) were concentrated on chromosome 3 and 7. These SNPs were closely associated with the regions previously identified by SNP-trait association analysis (34 Mb to 54 Mb of Chr. 3 and 48 Mb to 55 Mb of Chr. 7) (Figure 7).
We used the Illumina Infinium SolCAP 12K SNP array platform to genotype 94 progeny of an interspecific cross in potato in order to associate SNPs to high folate content in potato tubers. After initial filtering stages to remove bad and questionable SNPs from the data set, SNPs with more than 10 no-calls for individuals were removed leaving 9590 SNPs. These SNPs were used for SNP-trait association and QTL single marker analysis, and enabled the identification of 109 and 80 SNPs, respectively. Genes related to folate biosynthesis in plants are well known and have been mapped to physical locations within the potato genome (Appendix A. 10.). SNP-trait association identified regions on chromosomes 3, 6, and 7 that are within reasonable proximity (less than 2 Mb) to 5-formyltetrahydrofolate cycloligase, dihydrofolate (DHF) synthase, and γ-glutamyl hydrolase 1, respectively. QTL single marker analysis did not identify any major QTLs on chromosome 6, but confirmed the SNP-trait association results for chromosomes 3 and 7.
SNP genotyping revealed that the clone USW4s#3 is moderately heterozygous. This can explain the irregular heterozygosity and segregation distortion among the F2 population. Even with this high level of heterozygosity in the mapping population, it appears that the majority of the SNPs used for linkage group mapping are in reasonable order and mapped to the correct linkage group. The mapped linkage group SNPs that do differ from physical SNP locations may be due to the increased segregation distortion or may be due to inconsistent levels of heterozygosity across the population. Because these differences are not significant across multiple parts of the genome when compared to the map order of the physical genome, these genome positions can still be used for genome mining and validation of SNP markers/potential QTLs.
Although approximately 150 individuals from the mapping population were planted, only 94 produced tubers. Wild potatoes and potatoes from interspecific crosses grow much differently than commercial potato cultivars and often will not tuberize under long-day conditions. In this case, the F2 population was grown throughout the summer and fall in a greenhouse, but many of the potatoes did not tuberize, leading to a greatly reduced number of individuals for folate analysis and SNP genotyping.
Ninety-four individuals are considered a small population size in mapping and association studies which can be problematic depending on the type of analysis that is applied. A study from Schon et al. 1998 compared the ability to detect QTLs between populations of different sizes. In progenies of N = 344 vs. N = 107 it was found that at a LOD of 2.5, the number of QTLs detected for all traits was three times as great for the N = 344 population. This suggests that major QTLs can be detected in small mapping populations, but that the detection of minor QTLs is easier in larger populations and allows for more rigorous analysis. In another study, Gardiner et al. 2014 compared genome-wide association analysis and QTL detection with SNPs in a population of 94 individuals. This study was not successful in identifying SNPs associated with a large effect QTLs and suggests using association analysis with dense SNP arrays (20000 markers) as an alternative approach to QTL analysis in full-sib populations. Even when significant associations between the complex quantitative traits like high folate levels in tubers and molecular markers are identified, the genomic regions that are identified are often large (as in this study) which makes the validation of such markers all the more important. The top 10 SNPs in common from both analyses were compiled and are in the process of being validated (Table 6).
Educational & Outreach Activities
- Robinson, B.R.; Sathuvalli, V.R.; Bamberg, J.B.; Goyer, A. Exploring folate diversity in wild and primitive potatoes for modern crop improvement. Genes 2015, 6, 1300-1314. Robinson et al 2015
Expression levels of the γ–glutamyl hydrolase I gene correlate with vitamin B9 content in potato tubers (in preparation)
- Single nucleotide polymorphism makers associated with high folate content from wile potato tubers (in preparation)
- Vitamin B9 and potatoes: breeding for enhanced nutrition in modern potatoes (in preparation)
Robinson, BR (Oral) Breeding for Nutritional Enhancement in Potato: Exploring Vitamin B9 diversity in Wild and Cultivated Potatoes. Oregon State University Departmental Seminar Series for Crop and Soil Sciences, Oregon State University, OR. Departmental Seminar 2014
Robinson BR, Sathuvalli V, Goyer A (Poster) Nutritional Enhancement of Vitamin B9 in Potatoes: Benefits for the Industry and the Consumer. Presented at Potato Field Day at the Hermiston Agricultural Research and Extension Center. June 24, 2015. Potato Field Day Poster 2015
Robinson BR, Sathuvalli V, Bamberg J, and Goyer A (Oral) Exploring Vitamin B9 Diversity in Poataoes: Biofortification of Modern Cultivars. Presented in the Frank L. Haynes Graduate Student Research Competition. Annual Meeting of the Potato Association of America, Portland, ME, July 19-23, 2015. Potato Association Presentation
Farmtastic: July 17, 2015. Twenty spots were available and filled.
Description: Spend the day learning about agriculture and nature in eastern Oregon! Topics included science of farming, nutrition, stream and grassland ecology, and invertebrate biology. This was be an interactive program taught by OSU graduate students. Activities included lessons on selected topics, games, crafts, and a tour of the OSU Hermiston Agricultural Research and Extension Center. Lunch was provided by Hermiston Parks & Recreation.
Hydromania: Summer 2015
Every other week throughout the summer break. Fourth and fifth grade students came to the station for about 2 hours for a tour/Q&A session about agriculture in the area. https://www.umatillaelectric.com/programs/youth-programs/hydromania/
Short term impact of results:
- Foundation data for breeding efforts toward nutritional enhancement of potato: this project has identified high vitamin B9 germplasm and genetic markers associated with folate accumulation.
- Awareness of potato as a healthy ingredient in the diet: the extension activities conducted under this project have contributed to the education of consumers (kids, adults, growers, stakeholders, county agents) about positive attributes of potato in the diet.
Medium term impact of results:
- Development and release of a vitamin B9-enriched potato variety: the graduate student efforts were the first steps toward the development of a vitamin B9-enriched potato variety which also satisfies other important traits for the potato industry (e.g., disease resistance, yield). Such potato variety will open the possibility to advertise the nutritional benefits of potato beyond the already well-known facts about vitamin C and potassium.
- Increase sales and new economic opportunities: the graduate student educational efforts and the continuing educational efforts by HAREC staff that will ensue beyond this project have contributed to give a good perception of potato in consumers’ minds. As health is a major criterion in food purchase decisions, one can expect increased sales and new economic opportunities for the potato industry.
Ultimate impacts of results:
- Sustainability of the potato industry: the economic viability of the potato industry can be guaranteed only if consumers purchase potatoes. Continuing campaigns about the positive nutrition characteristics of potatoes, and the development of nutrient-enriched potato varieties by potato researchers, will contribute to the economic sustainability of the potato industry.
- Better nutrition and health for society: Our data show that increasing vitamin B9 levels in potato tuber can be achieved by pre-breeding and breeding for high vitamin B9 from wild and primitive cultivated potatoes.
Although we have not conducted an economic analysis, a more nutritious potato variety will contribute to the economic sustainability of the potato industry, as mentioned above.
Our ultimate goal is to develop a potato variety that is more nutritious but also answers all the traditional criteria of the potato industry (e.g. yield, disease resistance, storage quality etc.). Therefore, we expect such a potato to be easily adopted by farmers.
Areas needing additional study
In the future, screening of additional individuals within promising accessions and species should help further identify more sources of high folate germplasm. Future research should also investigate the stability of the high folate genotypes across environments and the heritability of the high folate trait(s). Finally, the development of fast and easy-to-use predictors for high folate, such as molecular markers, is essential to accelerate the screening of potato genotypes and to assess the full potential of the potato genetic diversity.
Further research into this project should also focus on validation of those SNPs and potential QTLs identified as being associated with high folate and testing of markers in high and low folate genotypes. Along with the validation of these markers, further genotyping and mapping/association studies should be done in a larger, more predictable segregating population in order to detect associations that could have been missed in this study. If markers highly associated with high folate levels can be developed, it could be a valuable tool for selecting high folate individuals in breeding populations without having to test each one individually for their folate content. This could help breeders more efficiently develop cultivars that deliver enhanced nutrition to populations that need it.