key: cord-0831779-tr8p7v8t
authors: Mangel, Nathalie; Fudge, Jared B.; Gruissem, Wilhelm; Fitzpatrick, Teresa B.; Vanderschuren, Hervé
title: Natural Variation in Vitamin B(1) and Vitamin B(6) Contents in Rice Germplasm
date: 2022-04-04
journal: Front Plant Sci
DOI: 10.3389/fpls.2022.856880
sha: 53f9ff5530e0445da02e049790156b28ed518e17
doc_id: 831779
cord_uid: tr8p7v8t

Insufficient dietary intake of micronutrients contributes to the onset of deficiencies termed hidden hunger—a global health problem affecting approximately 2 billion people. Vitamin B(1) (thiamine) and vitamin B(6) (pyridoxine) are essential micronutrients because of their roles as enzymatic cofactors in all organisms. Metabolic engineering attempts to biofortify rice endosperm—a poor source of several micronutrients leading to deficiencies when consumed monotonously—have led to only minimal improvements in vitamin B(1) and B(6) contents. To determine if rice germplasm could be exploited for biofortification of rice endosperm, we screened 59 genetically diverse accessions under greenhouse conditions for variation in vitamin B(1) and vitamin B(6) contents across three tissue types (leaves, unpolished and polished grain). Accessions from low, intermediate and high vitamin categories that had similar vitamin levels in two greenhouse experiments were chosen for in-depth vitamer profiling and selected biosynthesis gene expression analyses. Vitamin B(1) and B(6) contents in polished seeds varied almost 4-fold. Genes encoding select vitamin B(1) and B(6) biosynthesis de novo enzymes (THIC for vitamin B(1), PDX1.3a–c and PDX2 for vitamin B(6)) were differentially expressed in leaves across accessions contrasting in their respective vitamin contents. These expression levels did not correlate with leaf and unpolished seed vitamin contents, except for THIC expression in leaves that was positively correlated with total vitamin B(1) contents in polished seeds. This study expands our knowledge of diversity in micronutrient traits in rice germplasm and provides insights into the expression of genes for vitamin B(1) and B(6) biosynthesis in rice.

Insufficient dietary intake of micronutrients contributes to the onset of deficiencies termed hidden hunger-a global health problem affecting approximately 2 billion people. Vitamin B 1 (thiamine) and vitamin B 6 (pyridoxine) are essential micronutrients because of their roles as enzymatic cofactors in all organisms. Metabolic engineering attempts to biofortify rice endosperm-a poor source of several micronutrients leading to deficiencies when consumed monotonously-have led to only minimal improvements in vitamin B 1 and B 6 contents. To determine if rice germplasm could be exploited for biofortification of rice endosperm, we screened 59 genetically diverse accessions under greenhouse conditions for variation in vitamin B 1 and vitamin B 6 contents across three tissue types (leaves, unpolished and polished grain). Accessions from low, intermediate and high vitamin categories that had similar vitamin levels in two greenhouse experiments were chosen for in-depth vitamer profiling and selected biosynthesis gene expression analyses. Vitamin B 1 and B 6 contents in polished seeds varied almost 4-fold. Genes encoding select vitamin B 1 and B 6 biosynthesis de novo enzymes (THIC for vitamin B 1 , PDX1.3a-c and PDX2 for vitamin B 6 ) were differentially expressed in leaves across accessions contrasting in their respective vitamin contents. These expression levels did not correlate with leaf and unpolished seed vitamin contents, except for THIC expression in leaves that was positively correlated with total vitamin B 1 contents in polished seeds. This study expands our knowledge of diversity in micronutrient traits in rice germplasm and provides insights into the expression of genes for vitamin B 1 and B 6 biosynthesis in rice.

Vitamin B 1 (thiamine) and vitamin B 6 (pyridoxine) are among the water-soluble vitamins that are essential micronutrients for humans and other animals. Several chemically related forms of vitamin B 1 and B 6 are found, termed vitamers, with thiamine diphosphate (TDP) and pyridoxal-5′-phosphate (PLP) the most well-known vitamers, respectively, owing to their known April 2022 | Volume 13 | Article 856880 Mangel et al. Vitamin B 1 and B 6 in Rice Germplasm coenzyme functions in primary metabolism across all kingdoms (Colinas and Fitzpatrick, 2015) . As humans obtain the bulk of vitamins from dietary sources, micronutrient deficiency disorders often arise when dietary intake is insufficient, either alone or in concert with congenital factors such as impaired vitamin metabolism and/or certain lifestyles (Bailey et al., 2015) . Micronutrient deficiency ("hidden hunger") is a global health problem, with over 2 billion people estimated to have one or more micronutrient deficiencies (Bailey et al., 2015) . Monotonous consumption of staple crops with low dietary levels of vitamin B 1 and B 6 contributes to non-attainment of the recommended dietary allowance (RDA) of vitamin B 1 (for adults: 1.1-1.3 mg/ day) and B 6 (for adults: 1.3-2.0 mg/day; Fitzpatrick et al., 2012; Titcomb and Tanumihardjo, 2019) . Additional factors may further contribute to the inability to reach micronutrient RDAs, such as post-harvest deterioration, certain food processing and preparation practices, or consumption of foodstuffs containing vitamin-degrading enzymes (Van Der Straeten et al., 2020) . Furthermore, the COVID-19 pandemic has reduced incomes in 63 low-and middle-income countries and increased the proportion of people unable to afford healthy diets, with implications for calorific and micronutrient deficiencies (Laborde et al., 2021) . Vitamin B 1 deficiency disorders (also known as thiamine deficiency disorders) are endemic in several low-and middleincome countries in India, Asia, and Africa, in particular where polished rice is a major source of calories and interventions to improve dietary micronutrient intake are absent (Johnson et al., 2019; Whitfield et al., 2021) . Deficiency in vitamin B 1 may lead to a broad spectrum of neurological disorders including beriberi and tropical ataxia neuropathy (Whitfield et al., 2021) , with Wernicke's encephalopathy and Wernicke-Korsakoff syndrome prevalent in alcoholics (Isenberg-Grzeda et al., 2012; Latt and Dore, 2014; Dhir et al., 2019) . Vitamin B 1 deficiency also occurs in high income countries and may arise through diverse pathophysiological mechanisms beyond alcoholism (Gomes et al., 2021; Onishi et al., 2021) . Congenital defects in vitamin B 1 metabolism are also known to result in deficiency disorders (Ortigoza-Escobar et al., 2016) . Although data at the population level is lacking for vitamin B 1 deficiency compared to other micronutrients, studies on Chinese cohorts showed 91.8% of children and 81.7% of adults over 60 years do not reach their dietary estimated daily intake of vitamin B 1 Liu et al., 2019) . Vitamin B 6 deficiency may manifest with seizures and other neurological events, because of the role of PLP in neurotransmitter biosynthesis (Wilson et al., 2019; Akiyama et al., 2020; Rojo-Sebastián et al., 2020) . Genetic defects in vitamin B 6 metabolism could also lead to vitamin B 6 -dependent epilepsy (Wilson et al., 2019) . Vitamin B 6 deficiencies have been associated with low socioeconomic status (Liu et al., 2019; Zhu et al., 2020) , certain medications (Lussana et al., 2003; Allen et al., 2015; Porter et al., 2019 ) and a range of diseases including inflammation (Paul et al., 2013) , diabetes (Nix et al., 2015; Porter et al., 2019) , cardiovascular disease (Dhalla et al., 2013; Wei and Ji, 2020) , and certain cancers (Gylling et al., 2017) . Lee (2021) reviews in detail vitamin B 6 deficiency and women's health issues. In countries where vitamin B 6 status has been studied at the population level, 24% of Americans are at risk or deficient in vitamin B 6 (Bird et al., 2017) , rising to one-third in South Korea (Kim and Cho, 2014) . Among a Chinese cohort, 95.1% of subjects aged 60 years and over were marginal or deficient for vitamin B 6 status (Liu et al., 2019) . Increased dietary intake of vitamin B 1 and B 6 is therefore likely to assist in mitigating respective deficiency disorders in a diverse range of populations (Titcomb and Tanumihardjo, 2019) .

In addition to the well-established roles as enzymatic cofactors and the micronutrient deficiency disorders that may arise in turn when scarce, B 1 and B 6 vitamers are implicated in environmental stress responses in plants (Fitzpatrick, 2011; Fitzpatrick and Chapman, 2020) . Both vitamins exhibit antioxidant activity in vitro (Gliszczyńska-Świgło, 2006; Czégény et al., 2019) and vitamin B 6 can quench singlet oxygen in plants (Denslow et al., 2005) and fungi (Bilski et al., 2000) . In cases where mutant alleles are non-lethal, downregulation or knock out of certain vitamin B 1 and B 6 metabolism genes typically results in stunting or increased susceptibility to various plant pathogens Fitzpatrick and Chapman, 2020) . Protective effects of both vitamin B 1 and B 6 have been observed under abiotic stress, together with an induction of biosynthesis de novo (Ribeiro et al., 2005; Denslow et al., 2007; Rapala-Kozik et al., 2008; Tunc-Ozdemir et al., 2009; Rapala-Kozik et al., 2012; Huang et al., 2013; Moccand et al., 2014; Dell' Aglio et al., 2017) . Exogenous vitamin B 1 provision could induce expression of defense-related genes and "prime" plants for enhanced resistance upon subsequent pathogen challenge (Ahn et al., 2005 (Ahn et al., , 2007 Boubakri et al., 2012; Huang et al., 2016) . Mutants impaired in vitamin B 6 metabolism are hypersensitive to abiotic stress and are susceptible to pathogen challenge (Vanderschuren et al., 2013; Zhang et al., 2015; Samsatly et al., 2020) . Despite this body of evidence, relatively little is known about the molecular dialog between stress signaling and vitamin metabolism. Hanson et al. (2016) hypothesized that protective effects from exogenous vitamin supplementation complement acute cofactor deficiencies in planta upon stress, as opposed to simply having direct antioxidant properties. Taken together, crop varieties with high vitamin B 1 and B 6 contents may confer some resistance to stress through being poised to readily supply TDP and PLP to apoenzymes when demands on primary metabolism increase.

Biosynthesis de novo of vitamin B 1 and B 6 in plants (Figure 1 ) has largely been unraveled in model species, chiefly Arabidopsis (Arabidopsis thaliana, At), maize (Zea mays, Zm), and tobacco (Nicotiana tabacum, Nt; Colinas and Fitzpatrick, 2015) . This has permitted rational design of metabolic engineering strategies toward biofortifying staple crops with enhanced levels of these vitamins in consumed tissues Goyer, 2017; Strobbe and Van Der Straeten, 2018) . However, several limitations have been reported, including marginal increases in vitamin contents through upregulation of biosynthetic pathways, with limited over-accumulation of vitamins in the target tissues and organs (Dong et al., 2015 (Dong et al., , 2016 Mangel et al., 2019; Strobbe et al., 2021a,b) . Further effort is therefore required to overcome Frontiers in Plant Science | www.frontiersin.org 3 April 2022 | Volume 13 | Article 856880 A B FIGURE 1 | Biosynthesis de novo of vitamin B 1 and B 6 in plants based on Arabidopsis. (A) Vitamin B 1 biosynthesis de novo is predominantly localized to the chloroplast and comprises formation of pyrimidine and thiazole heterocycles linked by a methylene bridge. The pyrimidine moiety of vitamin B 1 is supplied via conversion of 5-amino-imidazole ribotide (AIR) into 4-amino-5-hydroxymethyl-2-pyrimidine phosphate (HMP-P) by HMP-P synthase (encoded by THIC) in an ironsulfur cluster and SAM-dependent reaction, with the release of methionine and a 5′-deoxyadenosine radical. Next, HMP-P is phosphorylated to HMP-PP by the bifunctional HMP-P kinase/TMP synthase (encoded by THIAMIN REQUIRING 1, TH1). The thiazole moiety of vitamin B 1 is produced from nicotinamide adenine dinucleotide (NAD + ), glycine, and a sulfur atom donated from a conserved cysteine residue from the single-turnover enzyme ADT synthase/THI4 (encoded by THI1 in plants, THI4 in yeast) to produce adenosine diphosphate thiazole (ADT). ADT is converted to hydroxyethylthiazole phosphate (HET-P) by HET-P synthase activity (loci encoding this activity currently unconfirmed). HET-P and HMP-PP are condensed by TMP synthase activity of TH1 to form thiamine monophosphate (TMP). TMP is dephosphorylated by TMP phosphatases (TMPase) including TH2/PALE1 (THIAMIN REQUIRING 2/PALE GREEN 1) and HALOACID DEHALOGENASE (HAD), forming free thiamine. The cofactor vitamer thiamine diphosphate (TDP) is generated by diphosphorylation of free thiamine at the hydroxyl position by thiamine pyrophosphokinases (encoded by AtTPK1-2, OsTPK1-3). (B) Vitamin B 6 biosynthesis de novo is localized to the cytosol to form the cofactor vitamer pyridoxal 5′-phosphate (PLP). PDX2 is a glutaminase that yields glutamate and ammonia, the latter is used by PLP synthase (PDX1, encoded by PDX1.1-1.3 in Arabidopsis and PDX1.3a-c in rice) to form PLP with the addition of either ribose 5-phosphate (R5P) or glyceraldehyde 3-phosphate (G3P). B 6 vitamers are interconverted in a

Frontiers in Plant Science | www.frontiersin.org bottlenecks to biofortify vitamin B 1 and B 6 in rice endosperm at meaningful levels to approach respective RDAs. Germplasm of certain crops has proven a useful source to biofortify maize, cassava, and sweet potato with provitamin A (ß-carotene; Bouis and Saltzman, 2017; Bechoff et al., 2018; Govender et al., 2019; Foley et al., 2021) . Several crop species including cassava, potato, quinoa, rice, maize, wheat, and pulses have been studied for variation in B complex vitamins (Villareal and Juliano, 1989; Sotelo et al., 1990; Kennedy and Burlingame, 2003; Goyer and Navarre, 2007; Dong et al., 2011 Dong et al., , 2014 Goyer and Haynes, 2011; Goyer and Sweek, 2011; Shewry et al., 2011; Robinson et al., 2015; Nowak et al., 2016; Singh et al., 2016; Mangel et al., 2017; Zarei et al., 2017; Bali et al., 2018; Freitag et al., 2018; Granda et al., 2018; Li et al., 2018; Guo et al., 2019; Riaz et al., 2019; Robinson et al., 2019; Jha et al., 2020) . However, only some of these studies have sought to understand the genetic basis underpinning such variation and/or carried out detailed vitamer profiling and quantification by analytical techniques such as HPLC. Knowledge of constituent vitamer profiles and molecular determinants for high vitamin content accessions could aid the introgression of desirable alleles into farmer-preferred crop varieties to deliver bioavailable vitamers to consumers most in need. Such detailed knowledge of rice germplasm is limited with respect to vitamin B 1 (Villareal and Juliano, 1989; Sotelo et al., 1990; Kennedy and Burlingame, 2003) , while vitamin B 6 contents are known only for three rice varieties from the United States (Zarei et al., 2017) . No information on the molecular mechanisms behind such variation in rice have been published to our knowledge.

Rice is a staple crop for 50% of the world's population, including some of the world's poorest (Bhullar and Gruissem, 2013) . Cooked white, polished rice (lacking the seed coat and embryo) is a poor source of several micronutrients, including both vitamin B 1 and B 6 , when considered as the source of 80% of the daily calorific intake (Fitzpatrick et al., 2012; de Pee, 2014) . As much as 90-98% of grain vitamin B 1 contents are lost during polishing (Dong et al., 2016; Strobbe et al., 2021a) and up to 85% of vitamin B 6 (Mangel et al., 2019) . Therefore, rice endosperm is a rational target for improvement of vitamin B 1 and B 6 contents. Toward obtaining the upper ranges of the respective RDA (for lactating women) from a 233 g serving of cooked polished rice, 32-and 13.6-fold increases in vitamins B 1 and B 6 are required in polished seeds .

Here, toward determining if rice germplasm could be exploited for endosperm biofortification with vitamin B 1 and B 6 , we performed an in-depth characterization of a genetically diverse panel of 59 rice accessions grown under greenhouse conditions. The selected accessions spanning three subspecies were sourced from geographically diverse countries and vary in their resistance and susceptibility to major rice diseases.

Accessions with contrasting vitamin contents in three tissue types were selected for in-depth vitamer profiling by HPLC. RT-qPCR assays of selected vitamin biosynthesis de novo genes revealed differential expression across the accessions, which did not correlate with vitamin contents in leaves. Accessions profiled here with the highest polished seed vitamin B 1 and B 6 contents did not display sufficient levels of vitamin B 1 and B 6 to meet the respective RDAs. Future efforts ought to consider the use of substantially larger germplasm panels, alongside additional metabolic engineering strategies, toward combatting vitamin B 1 and B 6 deficiencies for rice consumers in greatest need. Table S1 ) were obtained as dry seeds from the International Rice Research Institute, Philippines, and selected using the International Rice GeneBank Collection Information System. 1 Dry seeds were dehusked and surface sterilized in 70% (v/v) ethanol for 30 s, followed by 30 min of agitation in 1.5% (v/v) sodium hypochlorite solution +0.01% (v/v) Tween-20. Seeds were then rinsed five times in sterile water and then transferred to sterile plastic jars containing full-strength MS media (Murashige and Skoog, 1962) supplemented with 3% (w/v) sucrose and 0.3% (w/v) Gelrite, at pH 5.8. Seeds were incubated in darkness at 28°C for 48 h, before transfer to a growth cabinet for 12 d under 16 h light and 8 h darkness at 28°C. Seedlings were then transferred to a greenhouse under controlled conditions (12 h artificial light at 30°C and 80% humidity, 12 h darkness at 22°C, and 60% humidity). Three seedlings of the same accession were planted in one pot (pot diameter 18 cm, 12 cm height).

For experiment 1, 49 accessions were propagated simultaneously in growth cabinets prior to transfer of the seedlings to a greenhouse (June-August 2013, Eschikon, Switzerland). In a second independent experiment (hereinafter called experiment 2), 21 selected accessions grown in experiment 1 were propagated for a second time under greenhouse conditions (August 2014-October 2014, Eschikon, Switzerland), together with 10 additional accessions from the Oryza SNP Sequencing Project (McNally et al., 2009) , resulting in 31 accessions grown in experiment 2 and a total of 59 accessions screened across both experiments. Sampling was consistently performed in the morning (10.30 am-12 pm). Leaves were sampled from three different tillers of 50 days old vegetative plants in experiment 1 and 40 days old plants in experiment 2. Leaf tissue was pooled, snap frozen in liquid nitrogen, and stored at −80°C 1 https://www.genebanks.org/genebanks/irri/ FIGURE 1 | salvage pathway with known components localized to the cytosol, mitochondria, and chloroplasts. Unknown phosphatases convert phosphorylated vitamers to non-phosphorylated vitamers. Pyridoxal kinase (encoded by SALT OVERLY SENSITIVE 4, SOS4) phosphorylates pyridoxal (PL), pyridoxine (PN), and pyridoxamine (PM) at the 5′ position. PL is converted to PN by pyridoxal reductase (encoded by PLR1). The PNP/PMP oxidase (encoded by PDX3) converts pyridoxine 5′-phosphate (PNP) and pyridoxamine 5′-phosphate (PMP) to PLP. Unknown glucosyltransferases and glucosylhydrolases convert PN to PN-Glu and vice versa. At: Arabidopsis thaliana. Os: Oryza sativa. Green boxes denote genes assayed by qRT-PCR in this study. until further use. Fully ripened panicles (maturing at different times owing to variation in heading dates) were harvested and dried for 5 d at 37°C. Dried, mature seeds were dehusked (referred to hereinafter as unpolished seeds) and stored at −80°C until further use, or polished for 2 min in a PEARLEST polisher (Kett) to remove the embryo, aleurone layer, and seed coat (polished seeds), and stored at −80°C until further use.

Vitamin B 1 For vitamin B 1 , a Saccharomyces cerevisiae thi4 mutant deficient in vitamin B 1 biosynthesis de novo was used for quantification of total vitamin B 1 from rice tissues from plants grown in experiments 1 and 2 (Raschke et al., 2007) . Fifty milligrams of frozen, ground rice leaves or seeds were used for extraction of vitamin B 1 in 500 μl of 20 mM sulfuric acid in darkness at room temperature for 30 min, before heating to 100°C for 1 h. The solution was adjusted to pH 5.7 with 3 M sodium acetate and centrifuged. To convert phosphorylated B 1 vitamers to free thiamine to permit uptake by yeast, supernatants were treated with acid phosphatase (0.2 U/10 μl per 50 μl of plant extract) overnight for 12-15 h at 37°C. Total vitamin B 1 was calculated from the linear range of a standard curve prepared with 5-100 ng of thiamine hydrochloride provided to the thi4 yeast mutant in thiamine-deficient media.

Samples from candidate accessions with contrasting total vitamin B 1 contents were selected for confirmation and vitamer profiling of thiochrome derivatives by HPLC using a method first described by Moulin et al. (2013) . Fifty milligrams of frozen, ground rice leaves or seeds were used for extraction of soluble vitamin B 1 in 100 μl in 1% (v/v) trichloroacetic acid by aggressive vortexing at room temperature for 30 min. Samples were centrifuged at full speed in a tabletop microcentrifuge for 10 min at room temperature. The clear supernatant was neutralized with 3 M sodium acetate to 10% of the final volume and oxidized to thiochrome derivatives using 15 μl of freshly prepared 30 mM potassium ferracyanide in 15% (w/v) NaOH, with 15 μl 1 M NaOH and 25 μl methanol according to Moulin et al. (2013) . Samples were injected into an Agilent Technologies 1260 HPLC to determine vitamer profiles by separation of thiochrome derivatives on a Cosmosil π-NAP column (150 × 4.6 mm, 3 μm pore size) using a methanol gradient at 1 ml min −1 detailed in Moulin et al. (2013) , with a 40 min run time. Peaks of fluorescence corresponding to the retention time of the commercial standards of B 1 vitamers TDP (Sigma), thiamine monophosphate (TMP; Fluka), and thiamine (Fluka) were integrated and extrapolated against a standard curve for each vitamer. Peak area was integrated only from non-saturated peaks and in cases of peak saturation, samples were reinjected in lower volumes. Injection volumes ranged from 10 to 40 μl.

Data for leaf samples were normalized to fresh weight (FW) and seed samples to dry weight (DW).

For vitamin B 6 , a Saccharomyces pastorianus American Type Culture Collection 9080 strain was used (Tambasco-Studart et al., 2005) as reported (Mangel et al., 2019) . Fifty milligrams of frozen, ground rice leaves or seeds were used for extraction of vitamin B 6 in 500 μl of 20 mM sulfuric acid in darkness at room temperature for 30 min, before heating to 100°C for 1 h. The solution was adjusted to pH 5.7 with 3 M sodium acetate and centrifuged. To convert phosphorylated and glucosylated B 6 vitamers to non-phosphorylated vitamers to permit uptake by yeast, supernatants were treated with acid phosphatase and β-glucosidase (0.2 U/10 μl of each enzyme per 50 μl of plant extract) overnight for 12-15 h at 37°C. Total vitamin B 6 was calculated from the linear range of a standard curve prepared with 0.15-2.4 ng of pyridoxine hydrochloride provided to the yeast mutant in pyridoxine-deficient media.

Samples from candidate rice accessions with contrasting total vitamin B 6 contents were selected for vitamer profiling by HPLC using an established protocol . Fifty milligrams of frozen, ground rice leaves or seeds were used for extraction of vitamin B 6 in 100 μl of 50 mM ammonium acetate pH 4.0 with aggressive vortexing for 10 min at room temperature. Samples were centrifuged at full speed in a tabletop microcentrifuge for 15 min at room temperature. The supernatant was incubated for 3 min at 99°C and again centrifuged for 15 min at room temperature before analysis. Extracts were injected into an Agilent Technologies 1200 HPLC to separate B 6 vitamers on a Sunfire C18 column (Waters), 4.6 × 150 mm, 3.5 μm particle diameter, with post-column derivatization in 0.7 M potassium phosphate buffer with 1 g L −1 sodium bisulfite added freshly, flow rate 0.3 ml min −1 . Samples were separated on an isocratic gradient of 50 mM ammonium acetate pH 4.0, flow rate 1 ml min −1 in a 40 min run time. Quantification was carried out using the linear range of a standard curve constructed with known amounts of standards (Colinas et al., 2016) , with vitamin B 6 glucoside (PN-Glu) determination calculated as PN equivalents (Mangel et al., 2019) . Standards were prepared and injected into the HPLC with every set of extractions. Peak area was integrated only from non-saturated peaks and in cases of peak saturation, samples were reinjected in lower volumes. Injection volumes were typically 10-30 μl.

Data for leaf samples were normalized to fresh weight (FW) and seed samples to dry weight (DW).

RNA was isolated from rice leaf samples using an established protocol (Chang et al., 1993) with Mangel et al. (2019) . Frozen, homogenized tissue was mixed with 1 ml of extraction buffer (2% w/v polyvinylpyrrolidone K-30, 100 mM Tris-HCl pH 8.0, 25 mM ethylenediaminetetraacetic acid, 2 M NaCl, and 0.5 g L −1 spermidine) and 2% (v/v) β-mercaptoethanol added freshly. Samples were incubated at 50°C for 15 min with agitation at 400-500 rpm, before being extracted twice with 1 volume of chloroform:isoamylalcohol (24:1, pH 7.5-8.0). Nucleic acids were recovered from the aqueous phase by absolute ethanol precipitation for 30 min at −80°C followed by centrifugation at top speed in a microcentrifuge for 30 min at 4°C. The pellet Frontiers in Plant Science | www.frontiersin.org was washed in 80% (v/v) ethanol and resuspended in diethyl pyrocarbonate (DEPC)-treated water. RNA was precipitated overnight at −20°C in 2 M lithium chloride and collected by centrifugation at top speed in a microcentrifuge for 30 min at 4°C. The RNA pellet was washed sequentially in 80% (v/v) and absolute ethanol, before vacuum drying and resuspension in DEPC-treated water. RNA was quantified spectrophotometrically (Nanodrop) and stored at −80°C until further use. Expression of selected genes encoding vitamin B 1 and B 6 biosynthesis de novo enzymes was assayed by RT-qPCR. Two micrograms of total RNA were converted to cDNA with random hexamer oligonucleotide primers using the RevertAid First Strand cDNA Synthesis kit (Thermo Fisher) in accordance with the manufacturer's instructions. Real-time qPCR reactions were prepared in 10 μl reaction volumes comprising 5 μl of 2x Fast SYBR Green master mix (Applied Biosystems), 2 μl of template cDNA diluted 5-fold in DEPCtreated water, 1 μl each of forward and reverse primer (1 μM working concentration), and 1 μl DEPC-treated water. A Roche LightCycler 480 II real-time qPCR machine was used with an initial denaturation at 95°C for 2 min, followed by 40 cycles of denaturation at 95°C for 10 s, annealing at 60°C for 20 s, and extension at 72°C for 30 s. Target gene expression was normalized to the OsUBQ5 reference gene (Jain et al., 2006) and relative expression calculated using the delta-delta C t method (Livak and Schmittgen, 2001) . Samples for RT-qPCR were pipetted in duplicate and amplicon identity confirmed by melt curve analysis and Sanger sequencing. Primers for RT-qPCR were developed using Nipponbare genome sequences obtained from Phytozome v7_JGI (Goodstein et al., 2012) . Sequences of oligonucleotide primers used for RT-qPCR are listed in Supplementary Table S2 . Conservation of oligonucleotide primer binding sites in sequenced Oryza sativa accessions could be confirmed for Nipponbare (japonica), IR64 (indica), and I-Kung-Pao (indica), by search of the rice Molecular Breeding Knowledgebase. 2 Multiple alignments were assembled for alleles belonging to these three accessions and the number of accessions from each subspecies belonging to a given GID group (i.e., bearing a given allele) are shown in Supplementary Figure S1 . qRT-PCR threshold cycle (C t ) values for the reference gene UBQ5 are shown in Supplementary Table S3 .

Data were analyzed in GraphPad Prism (version 9) and R (R Core Team, 2013). To determine the effect of accession on vitamin contents and leaf biosynthetic gene expression levels, one-way ANOVA tests were carried out with post-hoc Tukey tests ( a = 0.05) to correct for multiple comparisons. Statistically significant differences between accessions are denoted by different letters in respective figures and tables. For correlation analyses of total vitamin contents with leaf biosynthetic gene expression levels, Pearson's correlation coefficients were determined in GraphPad Prism with two-tailed value of p tests and 95% 2 www.mbkbase.org/rice/ confidence intervals. Standard deviations of ratios were estimated using the Taylor expansion formula.

To obtain insights into the diversity of vitamin B 1 and B 6 contents in rice germplasm, we selected a total of 59 accessions for analysis from geographically diverse countries (Supplementary Table S1 ). Selected accessions included landraces and modern breeding lines spanning O. sativa subspecies japonica, indica, and javanica, ten accessions characterized in the OryzaSNP project (McNally et al., 2009) , and with contrasting susceptibility and resistance to major rice diseases. These accessions were cultivated under greenhouse conditions across two independent plantings (experiment 1 and 2). To control for seasonal differences in greenhouse cultivation between the two independent plantings, plants were phenotyped in each experiment. The range of measured variation for accessions cultivated twice remained largely similar for plant height, number of panicles, and leaf dry weight (Supplementary Table S4 ). Total vitamin B 1 and B 6 contents in leaves, unpolished seeds, and polished seeds were first quantified using a yeast assay (Tambasco-Studart et al., 2005; Raschke et al., 2007 ; Figures 2, 3; Supplementary Tables S5, S6) . Statistically significant differences in vitamin contents between accessions were determined in all cases by one-way ANOVA with multiple comparisons and Tukey's post-hoc tests (α = 0.05). Accessions were placed into low, intermediate or high vitamin content groups based on their ranking below the 25th percentile, between the 25th and 75th percentile, or above the 75th percentile, respectively. Groups below the 25th percentile and above the 75th percentile were referred to as contrasting groups.

In experiment 1, 49 accessions were cultivated, with leaf tissues harvested from vegetative plants 50 days after germination (Supplementary Table S5A ). Leaf vitamin B 1 contents varied 3.32-fold, with Hanumanjata at 1.11 ng mg FW −1 as the lowest and Tapol with 3.68 ng mg FW −1 as the highest. Unpolished seed vitamin B 1 contents varied 3.9-fold, from 2.35 ng mg DW −1 for Vaid Butti to Phulpata at 9.15 (± 1.11) ng mg DW −1 (Supplementary Table S5B) . This was similar to previously reported ranges when converted to ng mg DW −1 (Villareal and Juliano, 1989; Sotelo et al., 1990; Kennedy and Burlingame, 2003) Table S5C ). Twenty-one accessions across the three percentile groups from experiment 1 were cultivated a second time under the same greenhouse conditions for experiment 2, together with 10 additional accessions from the Oryza SNP Project (McNally et al., 2009) . For vitamin B 1 contents in experiment 2, leaf samples varied 1.98-fold, ranging from 2.93 ng mg FW −1 in Aanga to 5.81 ng mg FW −1 in IR64 (Figure 2A) Twenty-one accessions cultivated in experiment 1 were re-sown, alongside 10 additional accessions from the Oryza SNP Project (experiment 2). The accessions with vitamin B 6 content below the 25th percentile of the distribution were considered as low vitamin B 6 accessions and those with vitamin content above the 75th percentile were considered as high vitamin B 6 accessions. Low, intermediate, and high vitamin B 6 accessions selected for HPLC analysis are bolded. Mean ± SD of 3 biological replicates, except Nipponbare (n = 6), IR64 (n = 6), and TP309 (n = 6) for the three tissues; IR 64-21 (n = 2) for leaves; Nipponbare (n = 2) and IR 64-21 (n = 2) for unpolished seeds; and Nipponbare (n = 2), IR 64-21 (n = 2) and Phulpata (n = 0) in polished seeds. The effect of accession on total vitamin B 6 contents in panels (A-C) was determined by one-way ANOVA (α = 0.05) with multiple comparisons and Tukey's post-hoc test. Statistically significant differences between accessions are denoted by different letters.

varied 2.23-fold, from 3.36 ng mg DW −1 for Nipponbare to Phulpata at 7.48 ng mg DW −1 (Figure 2B ). Polished seed vitamin B 1 contents varied 6.06-fold, from 0.43 ng mg DW −1 for Tainung 67 to FR 13 A at 2.58 ng mg DW −1 (Figure 2C ). Seed polishing led to a reduction in vitamin B 1 content ranging between 57.9 and 89.3% (Supplementary Table S6 ). Although vitamin B 1 contents showed some variation between experiment 1 and 2, the ranking of certain accessions based on vitamin B 1 contents was not significantly different between the two experiments (Figure 2; Supplementary Table S5 ). Based on vitamin B 1 contents, nine accessions from experiment 2 were selected for confirmation and in-depth vitamer profiling by HPLC and biosynthetic gene expression assays (see below). Variation in vitamin B 6 contents in rice germplasm was probed in an identical screening strategy as for vitamin B 1 . In experiment 1, vitamin B 6 in leaves varied 4.64-fold, from 0.88 ng mg FW −1 in Hanumanjata to 4.07 ng mg FW −1 in Indane (Supplementary Table S7A Table S7C ). In experiment 2, vitamin B 6 in leaves varied 3.77-fold, from 2.13 ng mg FW −1 in DNJ52 to 8.04 ng mg FW −1 in Daw Magawk (Figure 3A) , with higher ranges compared to experiment 1. Seed samples were more stable in terms of variation between experiments compared to leaves, with unpolished seed vitamin B 6 contents varying 4.0-fold, from 0.75 ng mg DW −1 for IR64 to Tapol at 3.12 ng mg DW −1 (Figure 3B) . Consistent with our previous report (Mangel et al., 2019) , polishing of rice seeds (that is, removal of the seed coat, embryo, and aleurone) results in losses of vitamin B 6 contents in seeds for dietary intake, with a reduction in vitamin B 6 content between unpolished and polished seeds ranging from 20% (Juchitan A74) to 76.5% (Aanga; Figure 3C ; Supplementary Table S8 ). Similar to the vitamin B 1 greenhouse screen results, vitamin B 6 contents and fold changes also showed some variation between experiments. Several accessions from contrasting groups maintained similar vitamin B 6 contents and eight were selected from experiment 2 for confirmation and vitamer profiling by HPLC and biosynthetic gene expression assays (see below).

To validate the yeast assay screen data for respective total vitamin contents in three rice tissue types from selected accessions with low, intermediate and high vitamin contents, a precise quantification of their soluble constituent vitamers was performed by HPLC against known amounts of external standards Szydlowski et al., 2013) . For B 1 vitamers, profiles were obtained for TDP, TMP, and thiamine. B 6 vitamers quantified were pyridoxal (PL), pyridoxamine (PM), pyridoxine (PN), and their phosphorylated esters. Since as much as 50% of vitamin B 6 is found as glucosides in certain plant samples (Gregory, 1998) , pyridoxine glucosides (PN-Glu) were determined as "PN equivalents" by treatment of samples with a ß-glucosidase and the increase in PN interpreted as PN-Glu (Mangel et al., 2019) . HPLC assays revealed that the most abundant B 1 vitamer in rice leaf tissues was TDP (94-97%), followed by minor but detectable pools of TMP and thiamine (Figures 4A,B;  Supplementary Table S9A ). This is consistent with previous observations in wild-type rice leaf samples (Dong et al., 2016; Hsieh et al., 2021) , maize , Arabidopsis (Mimura et al., 2016; Hofmann et al., 2020) , and cassava . Total vitamin B 1 (the sum of TDP, TMP, and thiamine) in the leaves of rice accessions assayed here showed statistically significant 2.07-fold variation, ranging from 1.04 in Hanumanjata to 2.13 ng mg FW −1 in I-Kung-Pao [F(6, 14) = 7.331, p = 0.0011, Figure 4B ; Supplementary Figure 4C ; Supplementary Table S9B ]. As the seed coat, embryo, and aleurone cell layer are the rice seed compartments that store the bulk of vitamin B 1 , polishing rice grain is known to deplete grain thiamine contents by as much as 90-98% (Dong et al., 2016; Strobbe et al., 2021a) . Here, total vitamin B 1 in polished seed samples decreased approximately 10-fold compared to unpolished seeds, with 3.94-fold variation between the lowest and highest accessions, ranging from 0. Table S10A ). Unphosphorylated B 6 vitamers similarly comprised the majority of the vitamin B 6 contents of Arabidopsis shoot material and is consistent with a previous observation for leaves of wild-type TP309 rice (Raschke et al., 2011; Mangel et al., 2019) . For unpolished seeds, the total vitamin B 6 contents varied 3.93-fold between the lowest and highest accessions, with 0.45 in IR64 to 1.78 ng mg DW −1 in Tapol [F(7, 16) = 65.63, p < 0.001; Figures 5A,C; Supplementary Table S10B ]. In general, unphosphorylated B 6 vitamers and PN-Glu comprised the bulk of vitamers in unpolished seeds (Figure 5C ; Supplementary Table S9B ). Similar to vitamin B 1 , polishing of rice grains depletes vitamin B 6 stores by as much as 85% (Mangel et al., 2019) . Polished seed vitamin B 6 contents varied 3.95-fold between the lowest and highest accessions, corresponding to TP309 at 0.1 and 0.41 ng mg DW −1 for DNJ52 [F(7, 16) = 3.918, p = 0.0113]. Unphosphorylated vitamers comprised the bulk of polished seed vitamin B 6 pools in all accessions, while phosphorylated vitamers (PLP, PMP, and PNP) and PN-Glu comprised relatively minor constituents or were below the detection limit (Figures 5A,D; Supplementary Table S10C ).

Previous studies observed correlations between the expression of vitamin biosynthesis genes and vitamin B 1 contents in cassava and vitamin B 9 in potatoes (Robinson et al., 2019) . To understand the molecular determinants underlying contrasting vitamin B 1 and B 6 contents of the rice accessions characterized here, we chose to quantify transcript levels of selected genes for biosynthesis enzymes (see Figure 1 for vitamin B 1 and B 6 metabolism schemes). RNA was isolated from leaf tissues sampled for vitamin analyses and converted to cDNA for RT-qPCR assays of THIC for vitamin B 1, and PDX1.3a-c and PDX2 for vitamin B 6 (Figures 6, 7) . A single copy of THIC is encoded in Nipponbare and IR64 genomes Qin et al., 2021) , with THIC expression quantified in Kitaake and Tainung 67 accessions (Dong et al., 2016; Hsieh et al., 2021) . Consistent with the model that vitamin B 1 biosynthesis de novo in plants ( Figure 1A) is predominantly localized to photosynthetic tissue Colinas and Fitzpatrick, 2015) , THIC expression was detected in leaves of all assayed rice accessions. THIC was differentially expressed across the accessions sampled [F(8, 18) = 18.94, p < 0.0001; Figure 6A ]. To determine if tissue total vitamin B 1 contents correlated with differential expression of THIC in leaves, total vitamin B 1 content means (as determined by HPLC analyses in Figures 4B-D) were plotted against expression level means of THIC for each accession. Pearson's correlation coefficient THIC expression in leaves did, however, show a statistically significant, positive correlation with polished seed total vitamin B 1 contents ( Figure 6D , R = 0.7682, R 2 = 0.5901, p = 0.0156).

The THIC RT-qPCR primers used here bind in the coding region of THIC and do not discriminate between splice variants in the THIC 3′ UTR which contains a TDP riboswitch (Bocobza et al., 2007; Wachter et al., 2007; Noordally et al., 2020) . Vitamin B 6 biosynthesis de novo in plants ( Figure 1B) is considered to take place ubiquitously throughout plant tissues (Titiz et al., 2006) . Three PDX1.3 orthologs (PDX1.3a-c) are expressed in rice accessions Nipponbare and TP309 (Dell' Aglio et al., 2017; Mangel et al., 2019) and the encoded enzymes are predicted to be catalytically active (Moccand et al., 2014) .

A single PDX2 locus is encoded in the Nipponbare and IR64 genomes and is expressed in TP309 (Mangel et al., 2019; Qin et al., 2021) . Expression levels of these rice PDX genes were assayed by RT-qPCR using RNA samples isolated from leaf samples also used for HPLC quantification of vitamin B 6 presented in Figure 5B . All four assayed PDX genes were expressed in leaves and are differentially expressed between the accessions PDX1.3a: [F(7, 16) = 17.29, p = 0.0301]; PDX1.3b: [F(7, 16) = 3.65, p = 0.0152]; PDX1.3c: [F(7, 16) = 3.946, p = 0.0109]; and PDX2: [F(7, 16) = 4.425, p = 0.0066; Figures 7A-D] . PDX1.3b was the most abundant PDX1.3 transcript in leaves under these conditions, reaching levels over 50-fold higher than PDX1.3c for DNJ52 (Figures 7B,C) . The mean of total vitamin B 6 content was plotted against PDX gene expression for each accession for Pearson's correlation coefficient tests (Figures 7E-H) . PDX1.3a-c and PDX2 expression levels showed no correlations with total vitamin B 6 contents in leaves of the rice accessions sampled here 

Previous studies reported natural variation for seed contents of total vitamin B 1 across 121 accessions in rice germplasm (Villareal and Juliano, 1989; Sotelo et al., 1990; Kennedy and Burlingame, 2003) , but such information is limited to three rice accessions for vitamin B 6 contents in rice germplasm (Zarei et al., 2017) . We profiled B 1 and B 6 vitamers using microbiological and HPLC assays in three tissue types (leaves, unpolished and polished grain) across a panel of 59 genetically diverse rice accessions, thereby expanding knowledge on these traits and providing insight into vitamer proportionality. The knowledge was complemented with an investigation of a select set of key genes involved in biosynthesis de novo of vitamin B 1 and B 6 . This resource has the potential to inform on strategies for biofortification purposes as well as providing insight into possible regulatory pathways for vitamin content. First for vitamin B 1 , seven accessions within our panel had been measured previously for unpolished and polished seed total vitamin B 1 contents (Villareal and Juliano, 1989; Sotelo et al., 1990; Kennedy and Burlingame, 2003) and served as benchmarks in our screens (Figures 2, 4C Figure 4B for qRT-PCR assay of THIC expression. Expression was normalized to UBQ5 and data are the mean ± standard deviation of n = 3 biological replicates. The effect of accession on gene expression was determined by one-way ANOVA (α = 0.05) with multiple comparisons and Tukey's post-hoc test. Statistically significant differences between accessions are denoted by different letters. Accessions are ordered according to leaf total vitamin B 1 content as determined in Figure 4B . Figure 5B for qRT-PCR assays of PDX1.3a (A), PDX1.3b (B), PDX1.3c (C), and PDX2 (D) expression. Expression was normalized to UBQ5 and data are the mean ± standard deviation of n = 3 biological replicates. Accessions are ordered according to leaf total vitamin B 6 content as determined in Figure 5B . The effect of accession on gene expression was determined by one-way ANOVA (α = 0.05) with multiple comparisons and Tukey's posthoc test. Statistically significant differences between accessions are denoted by different letters. Correlation of PDX1.3a (E), PDX1.3b (F), PDX1.3c (G), and PDX2 (H) gene expression from panels (A-D), respectively, with leaf total vitamin B 6 contents in Figure 3B by Pearson's correlation coefficient test (p < 0.05).

the different B 1 vitamers quantified here by HPLC across the three rice tissue types was consistent with previous studies (Dong et al., 2016; Strobbe et al., 2021a) . Our study expands the list of accessions profiled for vitamin B 1 contents and reinforces evidence that the genetic diversity of this trait in rice seeds is limited in the surveyed rice germplasm. Data collected so far indicate that immediate exploitation of rice germplasm is likely impractical for a breeding-based vitamin B 1 biofortification strategy. The molecular basis underpinning vitamin B 1 variation in staple crop germplasm has been investigated in cassava, wheat, and potato (Goyer and Haynes, 2011; Goyer and Sweek, 2011; Mangel et al., 2017; Li et al., 2018) . In certain cassava accessions bearing duplications of THI1 and THIC genes, leaf vitamin B 1 contents are negatively correlated with transcript levels of MeTHIC2 (containing a functional TDP riboswitch) and MeTHI1b . Based on this finding, and similar observations for vitamin B 9 in potato germplasm (Bali et al., 2018; Robinson et al., 2019) , we sought to determine if this was also the case in rice as a potential molecular determinant underpinning vitamin B 1 diversity (Figure 6 ). Clear differential expression of THIC transcripts in rice leaves was observed across the rice accessions contrasting in vitamin B 1 contents ( Figure 6A ). Leaves are the principal site of vitamin B 1 biosynthesis de novo (Fitzpatrick and Chapman, 2020) . The expression pattern of THIC did not correlate with leaf total vitamin B 1 contents under greenhouse conditions ( Figure 6B ). Yet, further correlation analyses with total vitamin B 1 contents revealed a statistically significant, positive correlation between THIC expression in leaves and total vitamin B 1 contents in polished seeds ( Figure 6D) . THIC encodes the chloroplast-localized HMP-P synthase and is the first enzyme committed to pyrimidine moiety biosynthesis de novo of vitamin B 1 in plants ( Figure 1A ; Raschke et al., 2007; Coquille et al., 2013; Palmer and Downs, 2013) . THIC is functional in rice (Dong et al., 2016; Strobbe et al., 2021a) . The promoter activity of Arabidopsis THIC is positively regulated by light (Raschke et al., 2007) and negatively regulated by the CIRCADIAN CLOCK ASSOCIATED 1 (CCA1) transcription factor that phases time of day circadian expression (Bocobza et al., 2013; Noordally et al., 2020) . A TDP riboswitch is conserved in the 3′ UTR of THIC in higher plants and in Arabidopsis is subject to alternative mRNA 3′ processing in response to high or low nuclear TDP concentration (Raschke et al., 2007; Noordally et al., 2020) , which produces stable or unstable transcripts through use of alternative polyadenylation sites (Bocobza et al., 2007; Wachter et al., 2007) . The TDP riboswitch regulation of THIC is thought to be conserved in the green lineage (Bocobza et al., 2007; Croft et al., 2007; Wachter et al., 2007) and has been experimentally validated in both cassava THIC genes . Arabidopsis THIC mRNA alternative 3′ processing is considered to finetune intracellular TDP supply in response to changes in demand for TDP-dependent enzymes throughout the day (Bocobza et al., 2013) , independently of the circadian control of the THIC promoter (Bocobza et al., 2013; Noordally et al., 2020) . A CCA1 binding site is present 28 bp upstream of the Nipponbare OsTHIC 5′ UTR (Supplementary Figure S3) , suggesting the OsTHIC promoter could be subject to circadian regulation by CCA1. How high THIC expression in leaves of rice varieties might correspond to higher endosperm thiamine contents in the presence of a presumably functional TDP riboswitch in rice requires further investigation. With additional functional evidence, OsTHIC relative expression levels in leaves might serve as a useful indicator of vitamin B 1 contents in rice endosperm. TDP riboswitch activity in OsTHIC needs to be confirmed by base editing, alongside analysis of thiC null mutant vitamer and developmental phenotypes. Based on Arabidopsis, rice, and maize models of vitamin B 1 metabolism in plants, future work could characterize expression patterns of other genes encoding enzymes for vitamin B 1 biosynthesis de novo in leaves (the main organ of biosynthesis de novo activity). For example, research efforts in vitamin B 1 biofortification could explore the expression of the TH1 gene responsible for condensation of the pyrimidine and thiazole heterocycles ( Figure 1A ) in rice endosperm, or candidate genes involved in transport or salvage pathways in tissues with minimal vitamin B 1 biosynthesis de novo activity, such as seeds (Yazdani et al., 2013; Guan et al., 2014; Zallot et al., 2014; Dong et al., 2016) . THI1, which encodes a single-turnover enzyme in thiazole moiety biosynthesis (Figure 1A) , is functional in rice (Wang et al., 2006) and is sufficient to over-accumulate HET and vitamin B 1 when ectopically expressed in transgenic rice (Dong et al., 2016; Strobbe et al., 2021a) . THI1 therefore represents an additional candidate gene and awaits characterization in diverse rice germplasm. Alternative sources of the thiazole moiety have been investigated (Sun et al., 2019) and could be exploited for biofortification purposes. It is clear nonetheless that an enhanced understanding of vitamin B 1 metabolism is required in crops, including in rice, to permit successful endosperm biofortification at useful levels either by metabolic engineering or through identification of higher vitamin B 1 content accessions other than those profiled here and previously (Villareal and Juliano, 1989; Sotelo et al., 1990; Kennedy and Burlingame, 2003; Dong et al., 2015; Hanson et al., 2018; Strobbe et al., 2021a) . Given the high prevalence of suboptimal vitamin B 1 status and deficiency disorders around the world, together with the wide consumption of vitamin B 1 -poor polished rice (Bhullar and Gruissem, 2013; Dhir et al., 2019; Johnson et al., 2019; Titcomb and Tanumihardjo, 2019; Whitfield et al., 2021) , vitamin B 1 biofortification of rice should be considered as a priority in assisting to combat micronutrient deficiency. In contrast to vitamin B 1 , natural variation in vitamin B 6 contents has not been extensively studied in rice germplasm with only three varieties quantified to date (Zarei et al., 2017) , but no information is published on the molecular basis of such variation in rice. Vitamin B 6 contents in germplasm of other crops have also received minimal attention, except for analyses in potatoes (Mooney et al., 2013; Goyer et al., 2019) , wheat (Shewry et al., 2011; Freitag et al., 2018; Granda et al., 2018) , and a small number of accessions of barley (Freitag et al., 2018; Granda et al., 2018) , field beans (Freitag et al., 2018) , and quinoa (Granda et al., 2018) . Furthermore, regulation of vitamin B 6 metabolism has not been widely investigated in monocot species compared to eudicots (Dell' Aglio et al., 2017; Yang et al., 2017; Mangel et al., 2019; Suzuki et al., 2020) . Rice PDX1 and PDX2 genes are active (Dell' Aglio et al., 2017; Mangel et al., 2019) and are differentially expressed in transgenic lines over-expressing MALATE DEHYDROGENASE 1, concomitant with alterations to B 6 vitamer profiles (Nan et al., 2020) . Maize PDX2 is functional (Yang et al., 2017; Suzuki et al., 2020) and is required for proper embryo development (Yang et al., 2017) . Contrasting results from vitamin B 6 biofortification efforts in model and crop plants that aimed at increasing vitamin B 6 in target tissues through metabolic engineering indicates further research is needed to understand the regulation of vitamin B 6 metabolism and sequestration, particularly in cereal endosperm (Chen and Xiong, 2009; Raschke et al., 2011; Li et al., 2015; Fudge et al., 2017; Mangel et al., 2019) . Our results show that moderate natural variation does exist for vitamin B 6 contents in rice germplasm (Figures 3, 5; Supplementary Tables S6, S9 ). Similar to vitamin B 1 , accessions profiled here with the highest vitamin B 6 contents in polished seeds fall well below a practical level to justify introgression of such a trait for biofortification purposes (Figures 3, 5; Supplementary Table S9 ). Although no statistically significant correlations were observed between the expression in leaves of genes encoding for biosynthesis de novo enzymes and vitamin B 6 contents under greenhouse conditions, future studies should expand such analyses to include analyses of vitamin B 6 salvage pathway genes in leaves or expression analyses in the endosperm. Given the prevalence of vitamin B 6 deficiency in certain populations, combined with the wide consumption of vitamin B 6 -poor polished rice, biofortification of rice with vitamin B 6 also remains a priority in combatting micronutrient deficiency (Kim and Cho, 2014; Bird et al., 2017; Liu et al., 2019; Zhu et al., 2020) .

A growing abundance of rice genetic resources such as the rice 3,000 genomes project (The 3000 rice genomes project, 2014; Wang et al., 2018) , single-nucleotide polymorphism databases (McNally et al., 2009; Chebotarov et al., 2016) , together with contemporary genome or base editing techniques (Zhu et al., 2017; Jin et al., 2019; Lu et al., 2020) , remain alternative avenues to explore in order to biofortify rice endosperm with enhanced micronutrient contents. Redesigning the energetically costly vitamin B 1 biosynthesis de novo pathway (Nelson et al., 2014) has also been proposed (Hanson et al., 2018; Sun et al., 2019) and could be drawn on for biofortification purposes.

In conclusion, our screen of diverse rice accessions under controlled greenhouse conditions advances our understanding of micronutrient trait diversity in rice germplasm by combining quantifications of total vitamin B 1 and B 6 levels as well as vitamer partition, over three tissue types of rice with insights into biosynthetic gene expression patterns. Further analyses using larger germplasm panels together with in silico genetic analyses such as genome-wide association studies (GWAS) might be useful for identifying accessions and loci to exploit for biofortification.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

NM propagated and sampled the rice accessions, conducted yeast assays and RT-qPCR experiments, performed data analysis, and drafted figures and prepared tables. JBF performed the HPLC measurements and data analysis, prepared figures and tables, and wrote the manuscript. WG, TBF, and HV conceived the study, obtained funding, contributed to the analysis of the data, and edited manuscript drafts for submission. NM, JBF, WG, TBF, and HV commented and agreed on the final version of the manuscript and contributed to the article and approved the submitted version.

Frontiers in Plant Science | www.frontiersin.org

Vitamin B 1 functions as an activator of plant disease resistance

Vitamin B 1 -induced priming is dependent on hydrogen peroxide and the NPR1 gene in Arabidopsis

Vitamin B6 in acute encephalopathy with biphasic seizures and late reduced diffusion

Antiretroviral therapy provided to HIV-infected Malawian women in a randomized trial diminishes the positive effects of lipid-based nutrient supplements on breast-milk B vitamins

The epidemiology of global micronutrient deficiencies

Single nucleotide polymorphism (SNP) markers associated with high folate content in wild potato species

Yellow is good for you': consumer perception and acceptability of fortified and biofortified cassava products

Nutritional enhancement of rice for human health: the contribution of biotechnology

Vitamin B 6 (pyridoxine) and its derivatives are efficient singlet oxygen quenchers and potential fungal antioxidants

Risk of deficiency in multiple concurrent micronutrients in children and adults in the United States

Riboswitch-dependent gene regulation and its evolution in the plant kingdom

Orchestration of thiamin biosynthesis and central metabolism by combined action of the thiamin pyrophosphate riboswitch and the circadian clock in Arabidopsis

Thiamine induced resistance to Plasmopara viticola in grapevine and elicited host-defense responses, including HR likecell death

Improving nutrition through biofortification: a review of evidence from HarvestPlus

A simple and efficient method for isolating RNA from pine trees

Rice SNP-seek database update: new SNPs, indels, and queries

Enhancement of vitamin B 6 levels in seeds through metabolic engineering

Balancing of B 6 vitamers is essential for plant development and metabolism in Arabidopsis

Natures balancing act: examining biosynthesis de novo, recycling and processing damaged vitamin B metabolites

High-resolution crystal structure of the eukaryotic HMP-P synthase (THIC) from Arabidopsis thaliana

Thiamine biosynthesis in algae is regulated by riboswitches

Multiple roles for vitamin B(6) in plant acclimation to

Proposing nutrients and nutrient levels for rice fortification

The pseudoenzyme PDX1.2 sustains vitamin B 6 biosynthesis as a function of heat stress

Regulation of the Arabidopsis thaliana vitamin B 6 biosynthesis genes by abiotic stress

Regulation of biosynthetic genes and antioxidant properties of vitamin B 6 vitamers during plant defense responses

Mechanisms of the beneficial effects of vitamin B6 and pyridoxal 5-phosphate on cardiac performance in ischemic heart disease

Neurological, psychiatric, and biochemical aspects of thiamine deficiency in children and adults

Determination of folate content in rice germplasm (Oryza sativa L.) using tri-enzyme extraction and microbiological assays

Identification of QTLs underlying folate content in milled rice

Enhancement of thiamin content in Arabidopsis thaliana by metabolic engineering

Overexpression of thiamin biosynthesis genes in rice increases leaf and unpolished grain thiamin content but not resistance to Xanthomonas oryzae pv

Vitamin B6 in plants: more than meets the eye

Vitamin deficiencies in humans: can plant science help?

The importance of thiamine (vitamin B 1 ) in plant health: From crop yield to biofortification

Scaling up delivery of biofortified staple food crops globally: paths to nourishing millions

Impact of conventional and integrated management systems on the water-soluble vitamin content in potatoes, field beans, and cereals

Rationalising vitamin B 6 biofortification in crop plants

Antioxidant activity of water soluble vitamins in the TEAC (trolox equivalent antioxidant capacity) and the FRAP (ferric reducing antioxidant power) assays

Thiamine deficiency unrelated to alcohol consumption in high-income countries: a literature review

Phytozome: a comparative platform for green plant genomics

Consumer perceptions and acceptability of traditional dishes prepared with provitamin A-biofortified maize and sweet potato

Thiamin biofortification of crops

Vitamin B1 content in potato: effect of genotype, tuber enlargement, and storage, and estimation of stability and broad-sense heritability

Determination of folate concentrations in diverse potato germplasm using a trienzyme extraction and a microbiological assay

Effect of low temperature storage on the content of folate, vitamin B 6 , ascorbic acid, chlorogenic acid, tyrosine, and phenylalanine in potatoes

Genetic diversity of thiamin and folate in primitive cultivated and wild potato (Solanum) species

Content of selected vitamins and antioxidants in colored and nonpigmented varieties of quinoa, barley, and wheat grains

Nutritional properties and significance of vitamin glycosides

Divisions of labor in the thiamin biosynthetic pathway among tissues of maize

Genetic mapping of folate QTLs using a segregated population in maize

Vitamin B-6 and colorectal cancer risk: a prospective populationbased study using 3 distinct plasma markers of vitamin B-6 status

Redesigning thiamin synthesis: prospects and potential payoffs

Does abiotic stress cause functional B vitamin deficiency in plants?

On the nature of thiamine triphosphate in Arabidopsis

The rice PALE1 homolog is involved in the biosynthesis of vitamin B1

Thiamine-induced priming against root-knot nematode infection in rice involves lignification and hydrogen peroxide generation

Effect of abiotic stress on the abundance of different vitamin B6 vitamers in tobacco plants

Wernicke-Korsakoff-syndrome: under-recognized and under-treated

Validation of housekeeping genes as internal control for studying gene expression in rice by quantitative real-time PCR

Folate profile diversity and associated SNPs using genome wide association study in pea

Cytosine, but not adenine, base editors induce genome-wide off-target mutations in rice

Thiamin deficiency in low-and middle-income countries: disorders, prevalences, previous interventions and current recommendations

Analysis of food composition data on rice from a plant genetic resoruces perspective

Evaluation of vitamin B6 intake and status of 20-to 64-year-old Koreans

COVID-19 pandemic leads to greater depth of unaffordability of healthy and nutrientadequate diets in low-and middle-income countries

Thiamine in the treatment of Wernicke encephalopathy in patients with alcohol use disorders

The role of vitamin B6 in women's health

Genomewide association mapping of vitamins B1 and B2 in common wheat

Increased bioavailable vitamin B 6 in field-grown transgenic cassava for dietary sufficiency

Dietary micronutrients intake status among Chinese elderly people living at home: data from CNNHS

Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) method

Targeted, efficient sequence insertion and replacement in rice

Blood levels of homocysteine, folate, vitamin B6 and B12 in women using oral contraceptives compared to non-users

Vitamin B 1 diversity and characterization of biosynthesis genes in cassava

Enhancement of vitamin B 6 levels in rice expressing Arabidopsis vitamin B 6 biosynthesis de novo genes

Genomewide SNP variation reveals relationships among landraces and modern varieties of rice

Arabidopsis TH2 encodes the orphan enzyme thiamin monophosphate phosphatase

The pseudoenzyme PDX1.2 boosts vitamin B6 biosynthesis under heat and oxidative stress in Arabidopsis. The journal of Bological

Genotype-specific changes in vitamin B 6 content and the PDX family in potato

Analysis of Chlamydomonas thiamin metabolism in vivo reveals riboswitch plasticity

A revised medium for rapid growth and bio assays with tobacco tissue cultures

Rice plastidial NAD-dependent malate dehydrogenase 1 negatively regulates salt stress response by reducing the vitamin B6 content

Proteins with high turnover rate in barley leaves estimated by proteome analysis combined with in planta isotope labeling

Vitamin B status in patients with type 2 diabetes mellitus with and without incipient nephropathy

The coenzyme thiamine diphosphate displays a daily rhythm in the Arabidopsis nucleus

Assessment of the nutritional composition of quinoa (Chenopodium quinoa Willd

High proportion of thiamine deficiency in referred cancer patients with delirium: a retrospective descriptive study

Treatment of genetic defects of thiamine transport and metabolism

The thiamine biosynthetic enzyme ThiC catalyzes multiple turnovers and is inhibited by S-adenosylmethionine (AdoMet) metabolites

Mechanistic perspective on the relationship between pyridoxal 5′-phosphate and inflammation

Hyperglycemia and metformin use are associated with B vitamin deficiency and cognitive dysfunction in older adults

Pangenome analysis of 33 genetically diverse rice accessions reveals hidden genomic variations

R: A Language and Environment for Statistical Computing

Modulation of thiamine metabolism in Zea mays seedlings under conditions of abiotic stress

The upregulation of thiamine (vitamin B 1 ) biosynthesis in Arabidopsis thaliana seedlings under salt and osmotic stress conditions is mediated by abscisic acid at the early stages of this stress response

Enhanced levels of vitamin B 6 increase aerial organ size and positively affect stress tolerance in Arabidopsis

Vitamin B 1 biosynthesis in plants requires the essential iron sulfur cluster protein

Folate content analysis of wheat cultivars developed in the North China plain

Functional characterization of the thi1 promoter region from Arabidopsis thaliana

Expression levels of the γ-glutamyl hydrolase I gene predict vitamin B9 content in potato tubers

Exploring folate diversity in wild and primitive potatoes for modern crop improvement

Vitamin B6 deficiency in patients with Parkinson disease treated with levodopa/carbidopa

Vitamin B6 is under a tight balance during disease development by rhizoctonia solani on different cultivars of potato and on Arabidopsis thaliana mutants

Genotype and environment effects on the contents of vitamins B1, B2, B3, and B6 in wheat grain

Genetic variability for vitamin B9 and total dietary fiber in lentil (Lens culinaris L.) cultivars

Chemical composition of different fractions of 12 Mexican varieties of rice obtained during milling

Toward eradication of B-vitamin deficiencies: considerations for crop biofortification

Metabolic engineering of rice endosperm towards higher vitamin B1 accumulation

Metabolic engineering provides insight into the regulation of thiamin biosynthesis in plants

Parts-prospecting for a high-efficiency thiamin thiazole biosynthesis pathway

Construction and applications of a B vitamin genetic resource for investigation of vitamin-dependent metabolism in maize

Recycling of pyridoxine (vitamin B 6 ) by PUP1 in Arabidopsis

Vitamin B 6 biosynthesis in higher plants

The 3,000 rice genomes project

Global concerns with B vitamin statuses: biofortification, fortification, hidden hunger, interactions, and toxicity

PDX1 is essential for vitamin B 6 biosynthesis, development and stress tolerance in Arabidopsis

Thiamin confers enhanced tolerance to oxidative stress in Arabidopsis

Multiplying the efficiency and impact of biofortification through metabolic engineering

Strategies for vitamin B 6 biofortification of plants: a dual role as a micronutrient and a stress protectant

Variability in contents of thiamine and riboflavin in brown rice, crude oil in brown rice and bran-polish, and silicon in hull of IR Rices

Riboswitch control of gene expression in plants by splicing and alternative 3′ end processing of mRNAs

Dual function of rice OsDR8 gene in disease resistance and thiamine accumulation

Genomic variation in 3,010 diverse accessions of Asian cultivated rice

Do Chinese children get enough micronutrients

Modification of vitamin B6 on the associations of blood lead levels and cardiovascular diseases in the US adults

Thiamine fortification strategies in low-and middle-income settings: a review

Disorders affecting vitamin B(6) metabolism

Small kernel2 encodes a glutaminase in vitamin B 6 biosynthesis essential for maize seed development

Identification of the thiamin salvage enzyme thiazole kinase in Arabidopsis and maize

Salvage of the thiamin pyrimidine moiety by plant TenA proteins lacking an active-site cysteine

Rice bran metabolome contains amino acids, vitamins & cofactors, and phytochemicals with medicinal and nutritional properties

Vitamin B6 contributes to disease resistance against Pseudomonas syringae pv. Tomato DC3000 and Botrytis cinerea in Arabidopsis thaliana

Characteristics of genome editing mutations in cereal crops

Vitamin status and diet in elderly with low and high socioeconomic status: the lifelines-MINUTHE study

The authors wish it to be known that NM and JBF are equal first authors and that WG, TBF, and HV are equal last and corresponding authors. For the purpose of their CVs, the respective authors can list their name as the first or as the last author. All authors contributed to the article and approved the submitted version.

We gratefully acknowledge financial support from the Swiss National Science Foundation (grants 31003A-141117/1 and 31003A-162555 to TBF; grant 31003A-140911 to WG, HV, and TBF), the VELUX Foundation (WG), the Université de Genève (TBF), and ETH Zurich (WG). WG is supported by a Yushan Scholarship of the Ministry of Education in Taiwan.

The authors thank Irene Zurkirchen (ETH Zurich) for support of the greenhouse work. JBF wishes to thank Michael Moulin (Université de Genève) for advice with HPLC analyses.

The Supplementary Material for this article can be found online at https://www.frontiersin.org/articles/10.3389/fpls.2022.856880/ full#supplementary-material Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.Publisher's Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is