Flax lines with high linolenic acid (Duguid et al., 2014), low linolenic acid (Green and Marshall, 1984; Rowland, 1991; Dribnenki et al., 1996), and low cyanogenic glucosides (Duguid, personal communications) are some of the major achievements in linseed genetics and breeding. Whereas, the first two attributes are associated with fatty acid content, the last one refers to flax seed content in linustatin and neolinustatin, two cyanogenic glucosides that are toxic to mammals (EFSA, 2007). The nutritional benefits of flaxseed linolenic acid in human and animal health are now well established, especially for the positive roles it plays in reducing incidence of cardiovascular diseases (Simopoulos, 2008). Flaxseed is also the richest source of lignans (Thompson et al., 2006; Pan et al., 2009; Touré and Xueming, 2010) for which a wide variety of purported health benefits have been reported (Webb and McCullough, 2005; Adolphe et al., 2010; Buck et al., 2011; Wang et al., 2015). In planta, flax lignans are usually found as secoisolariciresinol diglucosides (SDGs) ester-linked within an oligomeric matrix called the lignan macromolecule (Struijs et al., 2009; Kosińska et al., 2011). Its monomeric aglycone (SECO) and intermediate monoglucoside (SMG) forms do not naturally accumulate in planta.

Lignan metabolism in mammals has been the focus of many detailed studies (During et al., 2012; Udani et al., 2013; Setchell et al., 2014; Mukker et al., 2015). After consumption, the lignan macromolecule complex is hydrolyzed, SDG is deglucosylated into SECO and absorbed in the gut (Clavel et al., 2007). Non-absorbed SECO (50–72% of ingested SECO) is subsequently metabolized into the enterolignans, enterolactone (ENL), and enterodiol (END) mainly in the colon by the intestinal microflora (Clavel et al., 2006, 2007; Landete et al., 2015). Enterolignans have been reported to elicit modest estrogen-like activity by binding estrogen receptors (ERs), ERα or ERβ (Satake et al., 2015). Nonetheless, low concentrations of intact lignans have been detected in the sera of mammals fed lignan-rich diets, suggesting that non-metabolized lignans are taken up by the mammalian digestive system and manifest ER-independent activities in vivo and in vitro (During et al., 2012; Satake et al., 2015). As a β-glucoside however, SDG metabolism may vary between individuals due to differences in their gut microflora (Rowland et al., 2000; Clavel et al., 2005; Landete et al., 2015), thereby affecting its bioavailability in some individuals (Possemiers et al., 2007) similar to what has already been shown with isoflavone diazein, which is metabolized only by one third of humans into its bioactive S-equol form (Setchell and Cole, 2006; Landete et al., 2015). Recently, During et al. (2012) reported a linear increase in the uptake of lignan aglycone forms, including pinolariciresinol (PINO), SECO, and ENL, by human intestinal Caco-2 cells by simple diffusion or low affinity transporter. Less than 0.1% SDG absorption was observed, compared to 2% for SECO and PINO, and 7% for ENL, evidencing the effect of glucosylation on absorption and bioavailability. However, no SMG of flax lignan has been available so far for such testing. Due to its lower molecular weight of 524.6 g/mol and the lack of a second bulky and polar glucoside moiety, it is logical to conceive that SMG uptake by Caco-2 cells through diffusion would exceed that of SDG which has a molecular weight of 686.7 g/mol. Currently, SECO and SMG are only obtained after acid hydrolysis of SDG (Li et al., 2008), a process that is not environmentally friendly. Because SDG deglucosylation is a pre-requisite to any absorption or conversion into END and ENL in vivo, production of seeds with altered SDG glucosylation in vivo is of interest for designing a highly bioavailable functional flaxseed food. It would be advantageous to create flax lines with reduced lignan glucosylation in the forms of SECO or SMG in planta.

Glycosylation is a biological process that leads to chemical complexity and diversity of plant natural products (Osmani et al., 2009; Wang and Hou, 2009). This synthesis modification is catalyzed by enzymes of the glycosyltransferases (GT) superfamily in which family 1 GT, refers to uridine glycosyl transferases (UGTs) (Caputi et al., 2012) that are characterized by a PSPG box consisting of a 44 amino acid consensus signature motif (Gachon et al., 2005). UGTs transfer UDP-activated sugar moieties to specific acceptor molecules (Witte et al., 2009) and their glycosylation activities on acceptors can modulate their pharmacological activity (Osmani et al., 2009). Recently, UGT74S1 (JX011632), a family 1 GT, was identified in flax and characterized to sequentially glucosylate SECO into SMG and SDG (Ghose et al., 2014; Figure 1). UGT74S1 has also been reported to be a single-copy gene and the key player in controlling SDG formation in flax (Fofana et al., 2017). Furthermore, Trp³⁵⁵ and His³⁵² were found to be critical for the UGT74S1 glucosylation activity toward SECO whereas Gln³³⁷ and Ser³⁵⁷ appeared essential for SMG conversion into SDG in vitro (Ghose et al., 2015). Until now however, no study has assessed the direct effect of UGT74S1 mutations in planta. Moreover, despite growing interests in flax lignans, breeding for SDG per se has so far received little attention (Pavelek et al., 2015), with most studies being limited to observed variations among flax varieties and accessions within collections (Saastamoinen et al., 2013). So far, no flax lines with altered lignan glucosylation have been reported.

Flaxseed is a health-functional crop and the commercial production of genetically modified flax remains an impediment to conquering such markets (Fofana et al., 2010). Mutation breeding is a well-established method in flax improvement and has led to the development of cultivars with reduced linolenic acid (Green and Marshall, 1984; Rowland et al., 1995; Ntiamoah and Rowland, 1997). Targeting Induced Local Lesions in Genomes (TILLING) is a reverse genetic tool used in functional genomics and crop improvement (McCallum et al., 2000; Sikora et al., 2011). Traditional mutagenesis, including radiation or chemical mutagenesis, is followed by the detection of mutations in the causal gene(s) responsible for the phenotype(s) of interest (McCallum et al., 2000; Slade and Knauf, 2005; Uchida et al., 2011; Chantreau et al., 2013). Recently, for example, Chantreau et al. (2013) developed a flax EMS mutagenized population and characterized the roles of coumarate-3-hydroxylase (C3H) and cinnamyl alcohol dehydrogenase (CAD) in fiber biosynthesis in flax stems by TILLING via forward and reverse genetics. The chemical mutagen, ethyl methane sulfonate (EMS), generates a wide range of mutant alleles, primarily point mutations (Comai and Henikoff, 2006) and, current advances in next generation sequencing have greatly improved the efficiency of induced-mutation detection (Galindo-González et al., 2015). With the release of the flax genome sequence (Wang et al., 2012), the DNA sequence of almost all flax genes is known. However, the gene functions and their link to phenotypic traits remain largely incomplete or completely unknown.

The present study was undertaken to (1) characterize an EMS mutagenized flaxseed population through the identification of lines carrying mutations in UGT74S1 and displaying altered lignan glucosylation phenotypes, and (2) determine the extent of the role played by UGT74S1 in SDG formation through SECO glucosylation. Using a next generation sequencing platform for the amplicon re-sequencing of the UGT74S1 gene in an EMS mutagenized flax population, we identified stable mutant lines with altered SDG lignan profile and demonstrated that the loss-of-function of SECO glucosylation into SDG was attributable solely to UGT74S1.

Establishing phenotype to genotype relationship and assigning functions to variant alleles are of interest to plant breeders and biologists. Using chemically induced mutations and TILLING as a reverse genetics tool, gene and allele functions have been elucidated in several biosynthetic pathways in plants (Slade and Knauf, 2005; Till et al., 2007; Xin et al., 2008; Dahmani-Mardas et al., 2010; Uchida et al., 2011; Rawat et al., 2012; Chantreau et al., 2013; Dhaliwal et al., 2015; Galindo-González et al., 2015). In the current study, we mined an EMS mutagenized flaxseed population and used an NGS-based reverse genetics platform to identify flax lines carrying mutations in the lignan glucosylation gene UGT74S1 and to determine allele-specific mutational effects on SDG lignan production in planta. The study demonstrated that EMS-induced mutations in UGT74S1 resulted in alteration of SDG accumulation in seeds and, as such, it constitutes the first report of non-transgenic flax germplasm with altered lignan content.

Whereas, TILLING terminology is relatively new, generating mutant alleles in flax using EMS is not (Green and Marshall, 1984; Rowland et al., 1995). The novelty of the chemical mutagenesis in the plant reverse genetics field resides in the much larger screening throughput capacity that we can now achieve (Sikora et al., 2011; Vicente-Dólera et al., 2014). In the present report, 1996 M2 families were screened by reverse genetics and, as expected, only G to A or C to T transition mutations were observed (Sikora et al., 2011). The calculated mutation frequency of one mutation per 28 kb is close to the range recently reported rate of one mutation per 30–49 kb in flax treated with different doses of EMS (Chantreau et al., 2013). A high proportion of 91% of the SNP variation was located in exons, in agreement with their higher G/C content which are more prone to EMS alkylation causing mainly transition mutations (Sikora et al., 2011). In this study, 22.5% of exonic mutations were homozygous, with potential loss (Uchida et al., 2011) or gain-of-function effects (Bailly et al., 2014). Indeed, whereas DNA mutations can affect any of the amino acids, not all substitutions have the same mutational effects on gene functionality (Rogers et al., 2000). Moreover, depending on the amino acids, homozygous missense mutations showing low to medium mutational effects can be expected to be neutral or hypermorphic. In the latter, the mutation may cause an increased contribution of the mutated gene to the original function (Lu et al., 2012).

By genotyping three generations of plants, the identified EMS-induced mutations were heritable, stable, or segregating, as reported in other plants (Uauy et al., 2009). Through phenotyping we showed that nonsense and missense mutations in UGT74S1 coding regions altered SDG lignan and phenolic acid glucoside profiles in flax seed. In a previous study, we showed that site-directed mutagenesis of targeted amino acids within the PSPG (W³³⁴–Q³⁷⁴) motif of UGT74S1 abolished completely or altered the SECO glucosylation in vitro (Ghose et al., 2015). In agreement with these findings, the EMS treatment randomly mutagenized nine amino acids (A342F, S343N, W355Stop, S357L, E360K, A361T, L362F, G365R, and G375D) within the PSPG in planta and, as previously observed in vitro, homozygous nonsense EMS-induced mutation of W³⁵⁵ to stop codon completely abolished SDG formation in planta, but retained trace amounts of SMG. Homozygous nonsense mutation of W²⁰⁵ to a stop codon outside the PSGP as well as the homozygous missense G282E mutation also led to loss-of-function of UGT74S1 for SECO glucosylation into SDG, but again a small amount of SMG remained. In a recent study, we reported that UGT74S1 is the key player in controlling SECO glucosylation into SDG, while also showing that UGT74S3 and UGT74S4, a pair of duplicated genes most closely related to UGT74S1, were able to glucosylate SECO into SMG at low efficiency but unable to synthesize SDG from SMG (Fofana et al., 2017). These findings suggested that these two paralogs may play roles in other biological processes other than SDG formation from SECO (Fofana et al., 2017). This was corroborated herein where we showed that plants from four UGT74S1 EMS lines lacking mutations in UGT74S3 and UGT74S4 failed to produce SDG but could still produce trace amounts of SMG. This clearly demonstrates occurrence of low background activity of UGT74S3 and UGT74S4 catalyzing trace SMG formation despite UGT74S1 being knocked out, in agreement with our previous report (Fofana et al., 2017). Moreover, some residual glucosylation activity may also be associated with each amino acid substitution effect within the mutated UGT74S1, as reported by in vitro site-directed mutagenesis data (Ghose et al., 2015). Additionally, the pool of phenolic acid glucosides was found to be increased in all mutant lines, as compared to the lignan component of the polymer. Additional biochemical studies using the phenolic acids encountered in the lignan macromolecule as substrates are required to characterize the exact role(s) of these two UGT paralogs in planta.

Targeted gene panel sequencing by Ampliseq methodology confirmed the mutations detected in UGT74S1 by amplicon re-sequencing and KASP assays. It also identified the P471S mutation in UGT74S3 from UGT74S1 mutant lines G282E-Homo-2004-9 and G282E-Homo-2004-14, each showing small amounts of SMG, thereby demonstrating the residual activity from UGT74S4 and/or the residual glucosylation activity associated with the G282E substitution in the mutated UGT74S1. The M4 plants from lines 2004-9 and 2004-14 also carried mutations, albeit with low effects, in both DIR4 and DIR5 genes at positions Y35H, K38H, R40Q, and L15S, but still produced trace amounts of SMG. This observation demonstrates that SECO was produced in these lines and that DIR4 and DIR5 are not rate limiting in the pinoresinol formation. Considering that not all the flax genome was scanned in the current study, one may wonder whether some transcription factors controlling lignan biosynthetic genes may have also been affected by EMS. Whereas, this assumption cannot be totally ruled, it seems to be highly unlikely or these mutations, if any, may be of very low or neutral effect because the lignan phenotypes observed in the lines herein described are in agreement with that from the in vitro data (Ghose et al., 2015), in the absence of implication of any transcription factors.

In conclusion, this study found EMS mutagenesis to be successful in altering SECO glucosylation toward SDG formation in planta. The wide range of lignan profiles observed from the current investigation constitutes a useful non-GMO genetic resource for flax breeders in developing new cultivars with high SDG content or new cultivars with SMG as a new trait in flax. It also offers plant biologists and natural health product scientists an opportunity to better understand the behavior of plants carrying the new trait in field environments. Using Ampliseq sequencing of UGT74S1 along with 10 other genes potentially playing a role in the SDG production, we validated our SNP detection approach and provided further evidence that UGT74S3 and UGT74S4 are not critical for SDG production in planta and that UGT74S1 is unique in controlling SDG formation from SECO.

BF: Conception, coordination, design, experiments, data analysis, interpretation, and writing of the manuscript; KG: Performed experiments, data analysis, interpretation, drafting and revision of the manuscript; JM: UPLC and mass spectrometry data acquisition and analysis of lignan extracts, revision of manuscript; DM and AS: EMS population nursery; SC: coordination, administration, and revision of the manuscript; MD and GR: EMS mutagenesis and M2 population generation, revision and proof reading of the manuscript. All authors read, commented and approved the manuscript.