Forward screening for seedling tolerance to Fe toxicity reveals a polymorphic mutation in ferric chelate reductase in rice

Rice contains the lowest grain Fe content among cereals. One biological limiting factor is the tolerance of rice to Fe toxicity. Reverse and forward genetic screenings were used to identify tolerance to Fe toxicity in 4,500 M4 lines irradiated by fast neutrons (FN). Fe-tolerant mutants were successfully isolated. In the forward screen, we selected five highly tolerant and four highly intolerant mutants based on the response of seedlings to 300 ppm Fe. Reverse screening based on the polymorphic coding sequence of seven Fe homeostatic genes detected by denaturing high performance liquid chromatography (dHPLC) revealed MuFRO1, a mutant for OsFRO1 (LOC_Os04g36720). The MuFRO1 mutant tolerated Fe toxicity in the vegetative stage and had 21-30% more grain Fe content than its wild type. All five highly Fe-tolerant mutants have the same haplotype as the MuFRO1, confirming the important role of OsFRO1 in Fe homeostasis in rice. FN radiation generated extreme Fe-tolerant mutants capable of tolerating different levels of Fe toxicity in the lowland rice environment. Mutants from both reverse and forward screens suggested a role for OsFRO1 in seedling tolerance to Fe toxicity. The MuFRO1 mutant could facilitate rice production in the high-Fe soil found in Southeast Asia.


Findings
Fe toxicity tolerance and grain Fe content Fe toxicity is a serious agricultural problem, particularly when plants are grown in acidic soils (Quinet et al. 2012). More than 100 million hectares of lowland rice production on low-pH soil in Southeast Asia is limited by iron toxicity (Becker and Asch 2005). Fe toxicity can occur in flooded soils with a pH below 5.8 under aerobic conditions, and at a pH below 6.5 under anaerobic conditions (Fageria et al. 2008). Plants grown under such conditions accumulated two-fold more Fe in their leaves (Bashir et al. 2014). In low pH paddy field, anaerobic condition leads to the reduction of Fe 3+ to Fe 2+ , resulting in excessive Fe availability and increased absorption (Quinet et al. 2012).
Genetic variation for tolerance to Fe toxicity exists in local landraces. However, most of their adaptive mechanism is associated with genetic variation in avoidance to Fe absorption and resulting in low grain Fe density. That association limits the chance for improving grain Fe density in acid soil, where high levels of Fe +2 are available for uptake and translocation to the grain. Therefore, it is important to understand natural genetic variation in enriching grain Fe density under Fe toxicity. One of the tolerance mechanisms that reduce excess Fe absorption is by reducing Fe +2 concentrations in rhizosphere by increasing the oxidative capability of roots (Ando 1983) or by excluding Fe from the rhizosphere (Tadano 1976). Another possible mechanism is increasing tissue tolerance to excessive levels of Fe +2 while increasing the rate of mobilization to grains. Such tolerance rice may link to Fe homeostasis that is not easily identified in existing germplasm. Therefore, one strategy is to find double mutation combining moderateto-high grain Fe content under neutral pH soil conditions while maintaining in the ability to withstand Fe-toxic conditions. These mutants may be likely to gain more Fe +2 to transport excessive Fe to grains.

FN mutant library
The fast neutron library was developed from Jao Hom Nin (JHN), a photoperiod non-sensitive, purple rice variety. By taking advantages of its distinctive color of leaves and grains, semi-dwarfism, early flowering and nutrient-rich grains, such mutant population is valuable for discovering mutation expressing useful genetic variation for both agronomic and nutritive characteristics. With its high combining ability, JHN mutants could be utilized as sources of new traits for marker-assisted selection in rice. Approximately 100,000 breeder seeds from JHN were mutagenized by using 33 Gy fast neutrons (FN). Successive generations from M 1 -M 4 , family history was traceable from individual M 1 plant. Due to abnormal mutation affecting seed set, several families were terminated leaving only 21,024 M 4 mutant families forming the base population for genetic screening (Rice Science Center, Kasetsart University, Thailand). For Fe toxicity screening, 4,500 lines were randomly chosen for forward screening while 500 pooled DNA libraries (representing 24 M 1 plants per pool) from the M 4 generation were used for reverse screening (Figure 1).

Forward screening identified Fe toxicity tolerant mutants
The 4,500 lines were screened in Fe-toxic (pH 3.0, FeEDTA 300 ppm) nutrient solution at the five-leaf seedling stage. The low pH nutrient solution released excessive ferrous Fe +2 , resulting in extensive leaf bronzing. The base population was assessed using a leaf bronzing index (LBI) (Arbeit 2003) in a time-course manner  following the death of the wild type JHN seedlings. In addition, the localization of iron in various parts of the leaf was visualized by PPB staining (Prom-U-thai et al. 2003). The days-to-seedling-death (DSD) of each variety was also recorded every other day. Phenotypic responses to Fe toxicity were categorized into tolerant, moderate and intolerant based on LBI and leaf PPB. Rice grains were also stained with PPB to reveal the embryonic Fe content, which is strongly associated with the grain Fe density. The first round screening of the 4,500 M 4 families yielded 95 tolerant and 57 intolerant mutants ( Figure 2A). After repeated screenings, only 9 tolerant and 32 intolerant lines remained ( Figure 2B). Selected mutants were scored for LBI every other day under the Fe toxic treatment (Figure 3). The result showed that the LBI of each mutant line increased rapidly for 20 days after Fe toxicity treatment, but the rate of increase can clearly be divided into tolerant and intolerant groups. By the 5th day, some intolerant mutant lines began to die, while the tolerant lines by the 11th day. The tolerant Mu783 prolonged seedling death to 19 days.

Designing the reverse screen
Mutant lines identified by forward screening can be used to develop functional markers for marker-assisted breeding. However, gene identification via forward mutant screen is complex, as FN-induced mutagenesis hits multiple targets. By reverse genetics, putative mutants carrying the candidate allele can be directly screened for the target phenotypic changes. This approach is called TIL-LING, Targeted Induced Local Lesion in Genome (Till et al. 2003). Using TILLING, seven candidate genes for iron homeostasis were selected, including ferric chelate reductase1 (   Gene-specific primers were designed for polymorphic coding sequences of the seven candidate genes for reverse screening. Potential mutable sequence variations were identified for each candidate gene and queried for potential SNV via public domains. Selected differential genotypes for grain Fe density, including Xua Bue Nuo (XBN), JHN, IR68144, RB#3, KDML105 (KD), Azucena (Azu) and Nipponbare, were genotyped for the SNVs, which may be associated with tolerance to Fe toxicity (Additional file 1: Table S1). Primer pairs for PCR amplification and denaturing conditions of each mutable site for dHPLC are listed (Additional file 2: Table S2). Before injection into the dHPLC column, PCR amplicons were denatured at 95°C for 5 min and annealed gradually from 95°C to 65°C over 30 min (Callery et al. 2006).

Reverse screen by heteroduplex
Reverse screen was initiated among 192 M 4 genomic DNA pools. Each pool represented either a set of 24 M 1 lines, or the total 4,608 lines. Amplified target fragments from the six candidate genes were screened for heteroduplexes using dHPLC in a 1:1 admixture with the control (wild type) amplicons. The results indicated that heteroduplex was only detected on OsFRO1 amplicons in DNA pool no.P024C12 ( Figure 4A). Individual members of the P024C12 pool were analyzed for potential mutants. Heteroduplex-forming amplicons were confirmed by sequencing ( Figure 4B). No mutation was found in the remaining candidate genes. However, we cannot rule out the possibility of mutations within other parts of the candidate genes, such as introns and promoters that were not included in the design.

Mutation on OsFRO1
The OsFRO1 mutant was purified and designated 'MuFRO1'. The OsFRO1 target gene was confirmed by gene sequencing. Sequence comparison between wild type and MuFRO1 identified four new single nucleotide polymorphisms (SNPs) and one indel in several introns. Two single amino acid changes (SAP) were identified in Exons 4 and 5 (Table 1). One SAP on Exon 4 exhibited a Valine (V) to Isoleucine (I) change in the same hydrophobic group. Because the SAP is located within the ferric chelate reductase domain, the amino acid change may affect the functioning of OsFRO1 (Marchler-Bauer et al. 2013). OsFRO1 also contained an AAA deletion, a SNP in intron 2 and three SNPs in intron 3, similar to the mutations found in KDML105, the landrace Fe toxicity tolerant strain, but unlike JHN and IR68144, the intolerant varieties (Table 1). Therefore, there are multiple FN-induced SNVs in the mutant OsFRO1. We cannot rule out the possibility of finding more mutation on other part of the genome. Ferric chelate reductase was first reported in Arabidopsis (Robinson et al. 1999) and later two FRO-like genes were identified in rice (Ishimaru et al. 2006). OsFRO1 was detected in leaves of Zn, Mn and Cu deficient rice whereas OsFRO2 transcript was found on Fedeficient leaves but not in roots under Fe deficiency. The result indicated that rice posses a unique Fe 2+ -uptake system via OsIRT1 and OsIRT2. A transgenic plant that fused refre1/372 from high pH tolerant yeast with the promoter of OsIRT1 showed strong increase in Fe 3+ chelate-reductase activity and Fe-uptake rate than control under Fe-deficient conditions. When grown under calcareous soil with high pH and low Fe availability, the transgenic rice yielded 7.9 times more productive. Recently, subcellular localization of the FRO families from Arabidopsis were identified (Jain et al. 2014). AtFRO7, found in chloroplast, may play important roles in Fe transport into chloroplast whereas AtFRO3 and AtFRO8, found in mitochondria, may involve in mitochondrial metal ion homeostasis (Jain et al. 2014). Cellular function of OsFRO2 was recently identified as a membrane   (Bashir et al. 2014). Furthermore, under Fe toxic condition, peroxidases, the enzyme known to cope with reactive oxygen species (ROS), was up-regulated in root (Quinet et al. 2012).
Comparison between rice varieties with high and low grain Fe density, OsFRO1 expression in grains show no difference (Das et al. 2013). However, only OsFRO1 transcript was found in root of the high grain Fe density variety. This finding leads to more investigation on the role of OSFRO1 in enriching grain Fe content for rice grown under excessive Fe.

Development of functional markers
Marker-assisted selection is most efficient when the functional marker for a target trait is utilized. For Fe toxicity tolerance, the two non-synonymous SAPs on Exons 4 and 5 of the OsFRO1 gene are used as functional markers. To develop agarose-based, co-dominant markers for Fe toxicity tolerance, bi-directional PCR was developed (Table 2), combining four primers in a single PCR amplification (Liu et al. 1997). Target amplicons were detected by 1.2% agarose gel electrophoresis ( Figure 5). These primer set and amplification protocols  can be used for marker-assisted selection to improve tolerance to Fe toxicity. Two haplotypes of the two SAPs, ' A-G' and 'G-C' , can differentiate the highly tolerant samples from the rest ( Table 3). Genotyping of the 41 selected mutants for forward screening ( Figure 2B) revealed four new highly Fe toxicity-tolerant mutants, Mu1463, Mu3130, Mu11183 and Mu783. Phenotypic screening confirmed mutants that were highly tolerant (5), moderate (1) and highly intolerant (4) to Fe toxicity. Such haplotypes can be directly applied to MAS for Fe toxicity tolerance.
In fast neutron treated population, it is not uncommon to identify multiple mutated genes from reverse screen. To ascertain if the multiple gene mutation was not false positive, we conducted a whole genome sequencing of the wild type JHN and extensive GBS of selected mutated genes using random lines core collection of the JHN mutant population to see if there have already existed in the JHN.

Target sequencing on MuFRO1
To further investigate the possibility of finding more candidate genes, targeted enrichment sequencing was conducted on 40 candidate genes that play roles in Fe transport from soil to seeds (Gross et al. 2003;Koike et al. 2004;Bashir et al. 2010;Masuda et al. 2012). The nucleotide sequence of 40 candidate genes (240 Kb) was collected for probe design (Additional file 3: Table S3). Targeted enrichment sequencing was conducted based on the Sure Select-XT Target enrichment system (Illumina paired end and multiplexed sequencing library by Agilent Technologies).
A new missense nucleotide variation was identified in OsYSL7, the metal-nicotinamide transporter protein (Table 4). The expression of OsYSL5, OsYSL6, OsYSL7, OsYSL14 and OsYSL17 were detected in epidermis, exodermis, cortex and stele of 3 week-old seedling root grown under Fe-deficient conditions for 2 weeks (Inoue et al. 2009). While, no expression was detected from maximum tillering to the flowering stages (Chandel et al. 2010).

Phenotypic evaluation of MuFRO1
Seeds of JHN and MuFRO1 harvested from two planting seasons were analyzed for Fe, Zn and Cu contents using ICP-OES at the Institute of Nutrition, Mahidol University,    (Table 5). MuFRO1 remained green with more biomass (data not shown) than wild type ( Figure 6). This result may suggest that MuFRO1 performed better Fe homeostasis by maintaining lower Fe content in the shoots. One such mechanism is simply by efficient partitioning into storage organelles like mitochondria and chloroplast. Therefore, OsFRO1 may play important roles in iron homeostasis and the maintenance of high biomass when grown under Fe toxicity conditions. JHN and MuFRO1 seeds were analyzed for Fe and Zn contents. MuFRO1 seeds contained 30% more grain iron than wild type JHN, but there was no difference in zinc content (Table 6). This opening a new opportunity to develop new rice varieties to withstand lowland Fe toxicity as well as enrichment of grain Fe density in the greater lowland rice growing area in the rice bowl of Asia.