Wheat, along with rice and maize, is one of three major cereals cultivated worldwide. The adaptability to a wide range of conditions, the good nutritional profile, along with its unique dough visco-elastic properties are the main reasons for its success (Shewry, 2009).

Durum wheat (Triticum turgidum ssp. Durum), the second wheat species most cultivated worldwide, is a key raw material in a wide variety of traditional foods largely consumed in the Mediterranean basin as a part of a diet style recognized as one of the healthiest in absolute terms (Willett et al., 1995; Sofi et al., 2010).

Where diets rely on plant-derived foods, it is crucial to increase the amount of health-promoting compounds (i.e. fibres, proteins, vitamins, antioxidants, minerals) in durum wheat as a valuable strategy to maintain good health and to prevent important non-communicable diet-related diseases (obesity, diabetes, cardiovascular disorders, osteoporosis, cancer), along with those associated to “hidden hunger”. In this regard, hidden hunger indicates a particular form of undernutrition that occurs when the intake and absorption of micronutrients are not enough for the daily requirement. This pathology affects almost one-third of the population, triggering serious diseases such as anaemia, weak bones, fatigue and weakened immune system and a few hidden impacts on the general well-being, lowering the life quality and increasing the risk of new pathologies (Lockyer et al., 2018; Vitamin and Mineral Nutrition Information System (VMNIS), 2022).

Genetic biofortification of staple crops represents an effective and sustainable strategy to boost essential minerals (i.e. iron, magnesium, calcium, potassium and zinc) in the human diet as it does not require the addition of fertilizers during the cultivation or additives in food production giving, as result, a stable over generations crop enriched in the target compound (Vasconcelos et al., 2017; Roberts and Mattoo, 2019).

Myo-inositol-1,2,3,4,5,6-hexakisphosphate (InsP₆) is a ubiquitous component of eukaryotic cells that plays several regulatory roles (Shears, 2001). Also known as phytic acid (PA), it is the major phosphorus storage sink within the plant seeds and other plant tissues and organs such as pollen, roots, tubers and turions. The amount and distribution of PA within the kernel depend on the plant species. In barley, rice and wheat about 80% of PA is stored in the aleuronic layer and bran, whereas in maize and Arabidopsis mainly in the embryo and scutellum (O'Dell et al., 1972). Differently from cereals and Arabidopsis, about 95% of PA is accumulated in the cotyledons in legumes (Ariza-Nieto et al., 2007).

Due to its negative charges, PA binds important mineral cations such as zinc, iron, potassium, magnesium and calcium precipitating in the form of phytate salts poorly digested by monogastric animals, including humans, due to the lack of phytases inside the digestive tract. As it chelates ions the bioavailability of phosphorus and minerals is decreased and PA is considered an anti-nutritional compound. In addition, the excretion of undigested phosphate contributes to environmental pollution by accelerating the eutrophication of the soil (Raboy, 2009; Secco et al., 2017).

In plants, phytic acid biosynthesis is carried out through two metabolic pathways (Sparvoli and Cominelli, 2015). The lipid-dependent pathway acts in all plant tissues, while the lipid-independent pathway is mainly active in seeds. In the first step, glucose-6-phosphate is converted into myo-inositol-3-phosphate (Ins (3) P₁) by myo-inositol-3-phosphate synthase (MIPS). The subsequent steps involve sequential phosphorylation of the inositol ring through various enzymes (inositol phosphate kinase 2, IPK2; inositol 1,3,4-trisphosphate 5-/6 kinase, ITPK; inositol polyphosphate 2- kinase, IPK1). The synthesized phytic acid is accumulated into globoids, spherical inclusions found within protein bodies (Krishnan, 2008; Regvar et al., 2011), and stored inside the vacuoles where it is transported by specific PA protein transporters, such as the Multidrug-Resistance associated Proteins (MRPs) (Sparvoli and Cominelli, 2014; Colombo et al., 2020; Cominelli et al., 2020a).

Previous breeding programs focused on the reduction of PA to increase the mineral content of food and the sustainability of agricultural production in different crops (Raboy, 2020). Low phytic acid (lpa) genotypes have been produced in all major grain crops using different strategies. In this regard, lpa mutants can be divided into three classes according to the step where the mutations affect the PA biosynthetic pathway or transport: 1) the first biosynthetic step in which glucose-6-phosphate is converted into Ins(3)P₁ by MIPS; 2) the last biosynthetic step in which IPK1 phosphorylates InsP₅ in the 2-position to synthesize PA; 3) the transport and storage of phytic acid into the vacuole through the targeting of MRP transporters (Sparvoli and Cominelli, 2015; Colombo et al., 2020).

In bread wheat TaMRP3 homeoalleles are located on the long arm of the 5A, 4B and 4D chromosomes and encode multidrug resistance-associated proteins belonging to the ABCC cluster of plant ATP-binding cassette (ABC) transporters (Bhati et al., 2014). MRP proteins are characterized by a common structure consisting of two soluble nucleotide-binding domains (NBD1 and NBD2), two hydrophobic transmembrane domains (TMD1 and TMD2) and an additional hydrophobic N-terminal extension (TMD0) connected by a cytosolic loop to the rest of protein (Sparvoli and Cominelli, 2014; Colombo et al., 2020). Lpa mutants were produced by targeting MRP genes in Arabidopsis thaliana, rice, soybean and common beans (Colombo et al., 2020), demonstrating that MRP proteins are involved in the transport of PA within the vacuole (Shi et al., 2007; Nagy et al., 2009).

Interestingly, Bhati et al., 2016 previously reported a partial suppression of TaMRP3 genes by RNA interference (RNAi) and demonstrated its functional role in bread wheat. Due to its central role in PA transport, TdMRP3 is an attractive target for increasing the accumulation of minerals in durum wheat through a non-transgenic genetic approach focused on its inactivation.

In this paper, the genes encoding TdABCC13 (TdMRP3) were completely disrupted at DNA level through a Targeting Induced Local Lesions in Genomes (TILLING) strategy, a highly processive non-transgenic reverse genetics technique that combines chemical mutagenesis with a PCR-based screening for the identification of mutations in the gene of interest (McCallum et al., 2000).

We show that the effect of TdMRP3 silencing was a significant reduction in PA content, resulting in an improved capability to accumulate micronutrients (Fe, Zn, Mn) in wheat seeds. In addition, it was tested whether the suppression of TdMRP3 genes in durum wheat generates modifications in root system architecture with significant consequences on root ability to acquire water and nutrients.

To the best of our knowledge, this study represents the first example of a genetic approach that successfully reduced the accumulation of phytic acid and increased the bioavailability of essential minerals in durum wheat kernel.

Hidden hunger is a global health issue, that involves more than 3 billion people around the world, mostly in Africa, Asia, and Latin America. Diets poor in essential vitamins and minerals (micronutrients), typical of some countries of the above-cited continents, are the main driver of the hidden hunger. The development of biofortified staple crop varieties is an effective and low-cost strategy pursued by several major public and private sector organizations (such as Universities, INRA, CIMMYT and ICARDA) in Southern America, Asia and Africa. In this context, Harvest Plus is a big initiative focused on the development of bio-fortified wheat varieties, that involves partners from dozens of countries (Wani et al., 2022). Although numerous efforts have been focused on the realization of biofortified wheat genotypes, it remains still challenging.

In plants, PA is stored in the vacuole and functions as a Pi sink to aid plant growth upon seed germination. Due to its negative charge, PA chelates the cations forming poorly bioavailable phytate salts and limiting mineral ion bioavailability, thus promoting mineral deficiencies in the body (Grases et al., 2017; Samtiya et al., 2020). For this reason, boosting the bioavailability of minerals in plant food can be achieved by reducing PA. Lpa mutants can be produced by the impairment of the transport and storage of PA into the vacuole: in the cytosol PA is exposed to a dephosphorylation process carried out by cytosolic phosphatases, decreasing the final amount of phytates and increasing free Pi and cations (Sparvoli and Cominelli, 2015; Colombo et al., 2020). Lpa mutants were developed in wheat via ethyl methane sulfonate mutagenesis, and germplasm derived from these mutants was used for breeding purposes (Guttieri et al., 2004; Guttieri et al., 2006; Venegas et al., 2022).

Here, lpa mutants were generated in durum wheat targeting TdMRP3 genes by TILLING, thereby causing the impairment of PA into the vacuole. TILLING strategy (McCallum et al., 2000) is a reverse genetics approach widely used in functional studies and breeding programs in numerous species, including bread and durum wheat (Slade et al., 2005; Botticella et al., 2011; Hazard et al., 2014; Sestili et al., 2015; Sestili et al., 2019; Garcia Molina et al., 2021) as it offers the big advantage to expand the genetic variability and to produce nontransgenic plants, that can be used for commercial purposes.

Our in silico analysis showed that the protein structure of MRP transporters was highly conserved among the major cereals. In detail, all the analysed sequences had the typical domains of ABCC transporters in forward orientation (Colombo et al., 2020), with a high identity degree in according to previous studies (Bhati et al., 2016; Cominelli et al., 2020b). Preliminary analysis on Ensembl database highlighted that durum wheat MRP3 proteins from Svevo genome v1 were different in length with respect to bread wheat and other cereal sequences. These durum wheat proteins missed an N-terminal region that included the TMD0 domain and part of the TMD1. Differently, the two durum wheat MRP3 sequences isolated by means of the new Svevo Platinum sequence showed the same length of TaMRP3 proteins (1510 and 1505 aa) and a perfect amino acid identity, sharing high phylogenetic proximity with other cereal species.

TdMRP3-A1 and -B1 were clustered into the same subgroup and were closer to the MRP proteins of Triticinae and barley and more divergent from other species, including the major cereals (rice and maize). Our data are in agreement with previous comparative studies that showed a high level of synteny in several gene loci among wheat, barley and other members of the Triticeae (Linde-Laursen et al., 1997; Mayer et al., 2011). Although the MRP3 sequences are highly conserved among the A, B, and D genomes of wheat, small amino acidic differences were identified. Unexpectedly a higher synteny was observed between the homeologous sequences encoded by the B and D genomes compared to MRP3-A1. In this regard, previous studies demonstrated higher synteny levels between A and D genome homeologues compared to those of the B genome (Akhunov et al., 2003; Pont et al., 2013).

In TdMRP3 double null mutants, PA was reduced by roughly 85% compared to the control in line with what was observed in lpa genotypes produced in legumes and other cereals, like maize and rice that showed a reduction of 80 and 90%, respectively (Silva et al., 2021). Differently, partial suppression of TaMRP3 genes targeted by RNAi resulted in a modest reduction of PA content (in the range from -22 to -34%) in bread wheat. These differences can be explained by the different efficiency of silencing of the two approaches: full silencing in TILLING mutants, and reduction in the range of 40-72% of TaMRP3 transcripts in the case of RNAi plants (Bhati et al., 2016). Accordingly, the accumulation of Fe, Zn and Mn in the grain was significantly raised in the TdMRP3 null mutants than in the transgenic RNAi lines (about +186% vs +18% for Fe; about +36% vs 0% for Mn; +78% vs +13% for Zn).

Generally, the phytic acid mutants described in many plant species were presented as agronomically poor for the low yields and for the poor germination capacity of the seeds (Raboy, 2020). In bread wheat, previous studies conducted on EMS mutants showed a reduction in PA of 30 to 40% (Guttieri et al., 2004), and a more contained loss of yield (from 8 to 25%).

In our study, the reduction in yield was comparable to bread wheat, although the reduction in PA was much higher (>85%). Actually, Guttieri et al. (2006) showed the absence of consistent effects on yield and yield-related traits in wheat lpa mutants, depending on the different genetic backgrounds in which the trait was transferred. In particular, in the soft white spring genetic background, the kernel weight of the lpa mutants was greater than that of the control wild type, confirming our findings, in which the partial reduction in yield, due to the reduction in the number of spikelets and kernels for spike, was compensated by a greater kernel weight. This suggested that the negative agronomic effects of the lpa genotype can be mitigated by adopting an adequate genetic improvement strategy.

In maize the suppression of ZmMRP4 was associated with ungerminability and seed weight loss (Pilu et al., 2005; Cerino Badone et al., 2012). Pilu et al., 2005 found a decreased content of PA in maize mutant lines and showed that kernels with less than 20% of PA are unable to germinate. The same pleiotropic effects were reported in rice with the suppression of OsMRP5, orthologous of TdMRP3 and ZmMRP4 (Xu et al., 2009). In detail, the T-DNA knock line, in which OsMRP5 was disrupted, showed a strong reduction of PA (>90%) and an inability to germinate.

Non-lethal lpa mutants were obtained in soybean and common bean through the silencing of the orthologous genes of TdMRP3 (Oltmans et al., 2005; Cominelli et al., 2018). PA dropped between 75 and 90% in both the species but only in common bean no pleiotropic effect on yield traits and plant phenotype was observed.

The morphology of the root apparatus was markedly altered in our complete null TdMRP3 mutant line. In detail, the root length, volume and surface area and the number of tips, were decreased, while the root diameter was increased compared to the control plants. The morphological alteration of the root apparatus is a common adaptive strategy used by plants to cope with suboptimal nutrients (mainly nitrogen, sulfur and phosphate) and water availability (Aiken and Smucker, 1996; López-Bucio et al., 2003; Péret et al., 2011). The root diameter generally matches the elongation rate, which stops when the root reaches the minimal diameter (Pagès et al., 2020). Accordingly, shorter roots generally have a larger diameter.

It is well known that the plants face phosphate starvation by modifying the root system architecture through the inhibition of primary root growth, the increase of the lateral root formation and the growth and production of root hairs, thus allowing their root systems to efficiently utilize Pi from soils (Vance et al., 2003; Hinsinger et al., 2009; Péret et al., 2011). Besides soil P concentration, also plant Pi status can trigger morphological root responses (Vance et al., 2003; Lambers et al., 2006). Thus, the short-root phenotype showed by our mutants could be attributed to changes in Pi homeostasis induced by TdMRP3 silencing resulting in increased availability of free phosphate in the cells and lower Pi demand from the soil. Moreover, MRP3 transporters are known to be expressed in several plant tissues including roots. So, we cannot rule out the possibility that the silencing of MRP3 transporter may have a more direct effect on anatomical root differences known to impact tolerance to Pi deficiency.

From a future perspective, it could be interesting to develop novel approaches able to limit the silencing of the TdMRP3 expression to the seed to minimize the pleiotropic effects associated with lpa mutants. Bhati et al. (2016), using a different silencing technology, did not find a significant difference in the root length between the control and the RNAi transgenic plants. Moreover, they observed that the number of lateral roots was higher in the transgenic lines compared to the control plants. We could interpret these contrasting results by considering that the critical P supply and plant status for the activation of P-deficiency response, including root alterations, differ among plant species (Teng et al., 2013).

In conclusion, this study shed light on the functional role of TdMRP3 in durum wheat. In detail, it confirmed that the reduction of PA through the silencing of MRP transporters is a good strategy to obtain durum wheat genotypes biofortified in essential minerals and with acceptable agronomic performances. To date, to the best of our knowledge, this represents the first study in wheat allowing at the same time to increase the accumulation of essential minerals such as Zn, Fe and Mn.

Noteworthy, the plant lines can be used as starting material in breeding programs focused on mineral biofortification without any legal restriction associated with genetically modified organisms.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material .

AF, SP, SC, MF and PV: investigation. EB and FS: conceptualization. FS and MV: funding acquisition. AF, SP and FS: data curation. SA, FS, PV and SM: visualization and validation. AF and FS: writing-original draft preparation. PV, EB, SA, SM and SM: writing-review & editing. FS: project administration. All authors contributed to the article and approved the submitted version.