Nicotianamine-chelated iron positively affects iron status, intestinal morphology and microbial populations in vivo (Gallus gallus)

Wheat flour iron (Fe) fortification is mandatory in 75 countries worldwide yet many Fe fortificants, such as Fe-ethylenediaminetetraacetate (EDTA), result in unwanted sensory properties and/or gastrointestinal dysfunction and dysbiosis. Nicotianamine (NA) is a natural chelator of Fe, zinc (Zn) and other metals in higher plants and NA-chelated Fe is highly bioavailable in vitro. In graminaceous plants NA serves as the biosynthetic precursor to 2′ -deoxymugineic acid (DMA), a related Fe chelator and enhancer of Fe bioavailability, and increased NA/DMA biosynthesis has proved an effective Fe biofortification strategy in several cereal crops. Here we utilized the chicken (Gallus gallus) model to investigate impacts of NA-chelated Fe on Fe status and gastrointestinal health when delivered to chickens through intraamniotic administration (short-term exposure) or over a period of six weeks as part of a biofortified wheat diet containing increased NA, Fe, Zn and DMA (long-term exposure). Striking similarities in host Fe status, intestinal functionality and gut microbiome were observed between the short-term and long-term treatments, suggesting that the effects were largely if not entirely due to consumption of NA-chelated Fe. These results provide strong support for wheat with increased NA-chelated Fe as an effective biofortification strategy and uncover novel impacts of NA-chelated Fe on gastrointestinal health and functionality.

. Biomarkers of Fe and Zn status following intraamniotic administration. Fe and Zn concentration (µg/g) in chicken (a,b) blood serum, respectively; and (c,d) liver, respectively. Bars represent mean ± SEM of at least three biological replicates. (e,f) Transcript quantification of genes in chicken duodenal and heart tissue, respectively. Values (expression ratio relative to 18S) represent mean ± SEM of at least three biological replicates, each with two technical replicates of quantitative RT-PCR. Different letters indicate significantly different values between treatment groups as analyzed by one-way ANOVA with Tukey post-hoc test (p < 0.05). NI: non-injected, C WF: control white flour extract, B WF: biofortified white flour extract.

Discussion
Both NA and DMA form high affinity 1:1 complexes with Fe 3+ (formation constants of 10 20 and 10 18 , respectively) and NA complexes Fe 2+ with a formation constant of 10 13 31,32 . By contrast, EDTA forms a pentagonal bipyramidal complex surrounding a single Fe 3+ atom with a formation constant of 10 25 , and likely provides Fe 3+ ions to the small intestine that require reduction by DCytB before absorption 31,45,46 . Intraamniotic administration of 'Fe EDTA' and 'Fe NA' significantly increased blood serum Fe and significantly downregulated the expression of intestinal DMT1 (a major Fe transporter) relative to administration of unchelated 'Fe' (Fig. 1A,E) suggesting that both EDTA-chelated and NA-chelated Fe are readily absorbed into the small intestine before export into the blood stream 3,47 . Decreased expression of DCytB (which catalyzes the reduction of Fe 3+ to Fe 2+ ) in 'Fe NA' relative to unchelated 'Fe' is evidence that NA delivers relatively more Fe 2+ ions, and that administration of an unchelated Fe solution delivers relatively more oxidized Fe 3+ ions to the intestine 48 . The expression of Ferroportin (the only known intestinal Fe exporter) was similar between treatment groups, and determining whether these NA-chelated Fe 2+ ions would be preferentially absorbed into intestinal enterocytes or transferred paracellularly into the blood stream requires further investigation 47 . Given that low expression of duodenal DMT1/DCytB relative to Ferroportin is linked with a positive gut microbiome 49 and that 'Fe NA' administration resulted in proliferation of probiotic Bifidobacterium in the ceca relative to Escherichia and Clostridium (Fig. 2D), we hypothesize that NA-chelated Fe is readily absorbed by the host and not available to Fe-responsive pathogenic bacteria 20,21,[50][51][52] . Increased cecal Escherichia abundance in 'Fe EDTA' instead suggests that EDTA-chelated Fe persists in the intestinal lumen and contributes to the proliferation of non-beneficial bacteria (Fig. 2D). Within the intestine, goblet cells are responsible for the synthesis and secretion of mucus, a polysaccharide/protein rich layer that physically protects epithelial cells, provides microbial habitat and facilitates nutrient exchange [53][54][55] . We hypothesize that highly bioavailable NA-chelated Fe is readily absorbed by the intestinal epithelia, leading to significantly increased goblet cell number ( Fig. 2A) and a mucosal habitat that supports probiotic Bifidobacterium 56 . By contrast, reduced intestinal goblet cells (and mucus production) coupled with increased villi surface area in 'Fe EDTA' may amplify the risk of bacterial infection due an increased proportion of potentially pathogenic Escherichia and Clostridium relative to probiotic Bifidobacterium 55 . Together these results suggest NA-chelated Fe is highly bioavailable to the host and improves intestinal functionality without causing dysbiosis and proliferation of pathogenic bacteria as commonly seen in traditional Fe supplements and fortificants 20,22 .
Biofortification is a cost-effective strategy to combat human micronutrient deficiencies by improving the density and/or bioavailability of micronutrients in staple crops through agronomic practices, conventional breeding, or modern biotechnology 57,58 . Biofortification efforts in pearl millet (Pennisetum glaucum L.) and common bean (Phaseolus vulgaris L.) have increased seed Fe concentration (up to 3.9-fold and 1.7-fold, respectively), and www.nature.com/scientificreports www.nature.com/scientificreports/ consuming these crops improves both Fe status and cognitive performance [59][60][61][62] . Traditional biomarkers of Fe status including blood serum Fe concentration and Fe homeostasis gene expression 47 were unchanged between chickens that received 'Biofortified' and 'Control' diets ( Fig. 3A,F). Caco-2 cell ferritin formation, a commonly used biomarker for measuring Fe bioavailability in vitro 37 , increased in biofortified white flour relative to control white flour but did not differ between digests of 'Biofortified' and 'Control' diets ( Fig. S2). Instead we observed a trend of increasing blood Hb and HME from week 4 onwards (Table 2), and significantly increased liver Fe concentration and heart COX gene expression 63 at week 6 ( Fig. 3D,F), indicating that 'Biofortified' chickens had improved Fe status relative to 'Control' chickens and demonstrating the importance of a holistic approach in evaluating host Fe status 37 . Interestingly we observed significantly decreased blood Hb, total body Hb and HME in 'Biofortified' chickens relative to 'Control' chickens at week 2 ( Table 2). The 'Control' chicken blood Hb and HME values at week 2 were the highest values obtained throughout the study and may be evidence of a carryover effect from consuming a nutrient-rich commercial diet prior to commencement of the study. We hypothesize that 'Control' chickens utilized greater amounts of Fe (and likely other nutrients) from the commercial diet as part of an adaptive response to the poor nutrient concentrations within the 'Control' diet (Table 1), and that 'Biofortified' chickens did not utilize the additional Fe within this commercial diet to the same extent. Together these results highlight the importance of conducting long-term feeding studies to accurately evaluate biofortified diets and more comprehensive investigation of this hypothetical nutrient utilization mechanism is warranted. Short-term exposure to extracts of biofortified white flour was insufficient to alter liver Fe storage in 'B WF' (Fig. 1C) and further highlights the importance of long-term exposure when evaluating biofortified diets. Given the large difference in feed consumption (Table 2) and relatively small (3 ppm) difference in dietary Fe concentration, 'Biofortified' chickens had lower Fe intake than 'Control' chickens over the course of the study (31.6 mg compared to 34.5 mg Fe). The increased liver Fe concentration to 'Biofortified' chickens relative to 'Control' chickens must therefore be the result of improved Fe bioavailability in the biofortified diet, likely due to increased concentration of NA-and/or DMA-chelated Fe given that both NA and DMA enhance Fe bioavailability in vitro 26,33,34 . Separating the effect of NA and DMA on dietary Fe bioavailability requires a follow-up study evaluating diets fortified with NA-chelated or DMA-chelated Fe and together these results reinforce the importance of the chelated form of Fe rather than target levels as a consideration for future Fe fortification and biofortification programs.
Here we show for the first time that the benefits of consuming a biofortified diet include altered intestinal functionality, enteric microbiota and feed energy conversion. Biofortified wheat consumption increased the abundance of Bifidobacterium and Lactobacillus in 'Biofortified' ceca relative to Clostridales (comprising Coprococcus Ruminococcus, Faecalibacterium and family Lachnospiraceae) and Escherchia (Fig. 4E) which is strikingly similar to the results obtaining following intraamniotic administration of NA-chelated Fe (Fig. 2D) and provides further evidence that NA-and/or DMA-chelated Fe is highly bioavailable and does not persist in the intestinal lumen where it can contribute to the proliferation of pathogenic bacteria 51,52 . The major phyla observed in this study: Firmicutes, Actinobacteria and Proteobacteria are shared between humans and chickens 36,64 . Typically Firmicutes are the most abundant (70-80%) and Actinobacteria least abundant (~5%) phyla in human and poultry, suggesting the atypical microbial composition of both 'Control' and 'Biofortified' (~20% and 38% Actinobacteria, respectively) is due to nutritional insufficiencies in both diets [65][66][67] . Bifidobacterium and Lactobacillus are major probiotic genera within Actinobacteria and Firmicutes respectively, and both genera symbiotically harvest additional nutrients and energy from the diet for the host 68 www.nature.com/scientificreports www.nature.com/scientificreports/ neutral and provide mucin with an appropriate chemical composition to support these populations 70 . We hypothesize that additional Bifidobacterium and Lactobacilli in the mucosal layer upregulate glycolysis/gluconeogenesis enzymes and increase the production of acetic, propionic and valeric SCFAs (Figs. 4C, 5C), leading to improved host Fe absorption and carbohydrate metabolism in 'Biofortified' chickens relative to 'Control' 66,69,71 . Improved metabolic capacity in 'Biofortified' chickens manifested as reduced cumulative FCR (consuming ~20% less for the same weight gain) and increased glycogen storage in both liver and pectoral tissues relative to 'Control' ( Table 2, Fig. 3E). Improved food energy conversion due to increased Bifidobacterium/Lactobacillus relative to Escherchia was observed following prebiotic supplementation in broiler chickens 72 , suggesting these effects may be due to NA and/or DMA acting as prebiotics in the biofortified diet (Table 1). Administering extracts of biofortified white flour (containing NA and DMA) increased intestinal goblet cell number and villi surface in 'B WF' relative to 'C WF' (Figs. 1D,E, 2), suggesting that even short-term exposure to biofortified wheat positively affects intestinal morphology. www.nature.com/scientificreports www.nature.com/scientificreports/ Traditional biomarkers of Zn status such as ZIP4 and ZnT1 gene expression and Zn concentration in blood serum, nails, and feathers 39,73 were unchanged in 'Biofortified' chickens relative to 'Control' , suggesting that Zn status was also unchanged (Figs. 3B,F, S3). Given the small differences in dietary Zn concentration (<3 ppm), 'Biofortified' chickens had lower Zn consumption than 'Control' chickens over the course of the study (21.0 mg compared to 22.1 mg Zn, respectively). Together these results suggest that 'Biofortified' chickens had improved Zn bioavailability likely due to consumption of increased dietary NA and/or DMA, although whether NA and/or www.nature.com/scientificreports www.nature.com/scientificreports/ DMA increase Zn bioavailability requires further investigation. We observed significantly decreased LA:DGLA at week 2 and a trend of decreased LA:DGLA from week 4 onwards in 'Biofortified' relative to 'Control' (Fig. 3C). As the LA:DGLA is a sensitive novel biomarker for evaluating Zn status 74 , these results suggest that longer-term (6 months) exposure to 'Biofortified' diet may demonstrate clearer improvements to Zn status and is warranted. Zinc deficiency in chickens is known to negatively alter the gut microbiome, and improved microbial composition in 'Biofortified' chickens may be an additional symptom of improved Zn status 75 . Given NA and DMA enhance Fe bioavailability we have previously argued these natural metal chelators function as phytonutrients in cereal foods 25,26,76 . It is well established that NA exhibits anti-hypertensive effects in vivo 34,77 although we did not detect differences in heart angiotensin-converting enzyme (ACE) and angiotensin II receptor type 1 (AT1R) gene expression throughout our study (Figs. 1F, 3F). We suspect similar heart ACE and AT1R expression between treatment groups in both the intraamniotic administration and feeding trial experiments is due to the relatively short exposure time to Fe solutions (4 days) or experimental diets (6 weeks) and it is worth investigating whether longer-term (6 months) exposure to increased dietary NA reduces hypertension. Nevertheless, the improved Fe status, gastrointestinal health and microbial composition in chickens following short and long-term exposure to NA-chelated Fe reinforces the idea of NA as an important phytonutrient in plant foods. Furthermore, utilization of NA-chelated Fe in food fortification and crop biofortification programs shows great potential to improve human health.

Materials and Methods
Plant material and white flour production. Vector construction, plant transformation and the initial selection of biofortified wheat material is described in 25 . In brief, the full-length coding sequence of OsNAS2 (LOC_Os03g19420) was PCR amplified from rice (Orzya sativa L.) cv. Nipponbare and recombined into a modified pMDC32 vector under transcriptional control of the maize (Zea mays L.) ubiquitin 1 (UBI-1) promoter with a hygromycin phosphotransferase plant-selectable marker (Fig. S1). Bombardment of the construct into immature wheat (Triticum aestivum L.) cv. Bobwhite embryos was performed at the University of Adelaide (Adelaide, Australia). One double-insert event and corresponding null segregant (termed 'Biofortified' and 'Control' , respectively) were grown were grown for two seasons for use in intra-amniotic administration (2016 field season) and feeding trial (2017 field season) in New Genes for New Environment facilities located in Merredin, Western Australia (Fig. S1, Table S1). Whole grain samples from Merredin were conditioned to 15% moisture content and milled (70-75% extraction) using a Quadrumat Junior laboratory mill (Brabender, Duisburg, Germany) for intraamniotic administration or a Buhler MLU-202 laboratory mill at The Commonwealth Scientific and Industrial Research Organisation (CSIRO, ACT, Australia) for the feeding trial. All break and reduction fractions of 'Biofortified' or 'Control' grain were combined to form either 'Biofortified' white flour or 'Control' white flour (Table 1). www.nature.com/scientificreports www.nature.com/scientificreports/ Preparation of extracts, solutions and diets. Wheat extracts were generated as described in 78 . In brief, 'Biofortified' white flour or 'Control' white flour was mixed in dH 2 O (50 g/L), filtered (600 µm) and centrifuged, and the resulting supernatant was dialyzed (MWCO 12-14 kDa, Medicell International Ltd., London, UK) exhaustively against dH 2 O (48 hrs.). The dialysate was lyophilized, and the resulting powder dissolved in 18MΩ H 2 O (0.05 g/mL) forming the white wheat flour extracts for intra-amniotic administration. Iron solutions were prepared by combining an Fe standard (1000 μg/mL, 2% HCl, High-Purity Standards, Charleston, SC, USA) with either 18MΩ H 2 O ('Fe'), or 1.6 mM NA (Toronto Research Chemicals  eggs were subsequently incubated for four days until hatch as described in 79,80 . Chicks were euthanized by CO 2 exposure after hatching and all tissues were collected. The remaining hatchlings (n = 30) were allocated based on body weight into two treatment groups: (1) 80% 'Control' white flour diet ('Control') and (2) 80% 'Biofortified' white flour diet ('Biofortified') as described in 38 . All chickens received a commercial diet (Nutrena ® Chick Starter Grower 18% Crumble, Cargill Inc, Wayzata, MN, USA) for one week prior to consumption of 'Control' and 'Biofortified' diets for six weeks. 'Control and Biofortified' diet formulations met the Nutrient Requirements for Poultry (NRC Poultry reference) excluding Fe and Zn. Chickens (n = 3) were housed in cages (1 m 2 ) and provided ad libitum access to food and H 2 O. Feed intakes were measured daily, and body weight and blood samples were obtained weekly. Feed conversion ratio (FCR) represents weekly feed intake (g) proportional to the weekly increase in body weight (g). Chickens were euthanized by CO 2 exposure seven weeks post-hatch and tissues collected.

Gene expression analysis (Tissue harvesting, RNA isolation, cDNA synthesis, primer design).
Total RNA extraction from duodenal and heart tissue (30 mg) using Qiagen RNeasy Mini Kit (RNeasy Mini Kit, Qiagen Inc., Valencia, CA, USA), cDNA synthesis and real time-polymerase chain reaction (RT-PCR) analysis were performed as previously described 38 www.nature.com/scientificreports www.nature.com/scientificreports/ was quantified using a seven-point standard curve in duplicate. Gene expression was obtained relative to 18 S (Cp), primer pair efficiency, and control treatments: 'NI' for intraamniotic administration and 'Control' for feeding trial 82 . Alkaline phosphatase (AKP) and sucrase isomaltase (SI) acted as intestinal reference genes following intraamniotic administration (Fig. 1E). All primers used for gene expression analysis are provided in Table S3. ferritin and glycogen analysis. Liver ferritin was determined as previously described 78 . In brief, samples (1 g) were homogenized in 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer (50 mM) and heat treated (75 °C, 10 min) before centrifugation. Native polyacrylamide gel electrophoresis (PAGE) gels were stained with Coomassie blue G-250 stain or potassium ferricyanide [K 3 Fe(CN) 6 ] and quantified using the Quantity-One 1-D analysis program (Bio-Rad, Hercules, CA). Liver and pectoral glycogen was determined colorimetrically as described in 83 with minor adjustments. After centrifugation and mixing with petroleum ether, homogenized tissue was mixed with color reagent (300 µL) and total glycogen determined on an ELISA plate reader (450 nm) according to a standard curve.
Intestinal functionality and short-chain fatty acid (SCFA) analysis. Duodenal samples were fixed in fresh 4% (v/v) buffered formaldehyde, dehydrated, and embedded in paraffin as previously described 38 . Serial sections (5 µm) were deparaffinized in xylene and stained with hematoxylin and eosin before goblet cell number and villi surface area examination under light microscopy using EPIX XCAP software (Standard version, Olympus, Waltham, MA, USA). Cecal samples were homogenized in HCl (2 ml, 3%, 1 M), centrifuged and combined with ethyl acetate (100 µL) and acetic acid-d4 (1 µg/mL) before collecting the organic phase to determine short chain fatty acid (SCFA) composition. Samples were quantified via GC-MS using a TRACE ™ 1310 gas chromatograph (Thermo Fisher Scientific, Waltham, MA, USA) and a TraceGOLD ™ TG-WaxMS A column (Thermo Fisher Scientific, Waltham, MA, USA).

Microbial population analysis.
Lactobacillus, Bifidobacterium, Escherichia, and Clostridium density in intraamniotic administration treatment groups was determined as previously described 79 . In brief, cecal contents were homogenized with phosphate-buffered saline (PBS, 9 ml), centrifuged and the pellet resuspended in ethylenediaminetetraacetic acid (EDTA, 50 mM) and treated (37 °C, 45 min) with lysozyme (10 mg/mL, Sigma Aldrich CO., St. Louis, MO, USA). Bacterial genomic DNA was isolated according to manufacturer's instructions (Wizard ® Genomic DNA Purification Kit, Promega Corp., Madison, WI, USA) and bacterial genera are presented in relative proportions. All primers used for microbial population analysis are provided in Table S4. 16S rRNA gene sequencing and analysis. Microbial genomic DNA extraction from 'Control' and 'Biofortified' cecal samples, gene sequencing and analysis was conducted as previously described 38 . In brief, 16S rRNA gene sequences were amplified from the V4 hypervariable region of microbial genomic DNA (Powersoil DNA isolation kit, MoBio Laboratories Ltd., Carlsbad, CA, USA, purified (AMPure, Beckman Coulter, Atlanta, GA, USA) and quantified according to manufacturer's instructions (Quant-iT ™ PicoGreen ™ dsDNA Assay Kit, Invitrogen, Carlsbad, CA, USA). Samples were sequenced at Bar Ilan University (Safed, Israel) using an Illumina MiSeq Sequencer (Illumina, Inc., Madison, WI, USA). Amplicon reads were analyzed using 'Divisive Amplicon Denoising Algorithm' (DADA2) and 'quantitative insights into microbial ecology' (QIIME) software before taxonomic classification using Greengenes database (http://greengenes.lbl.gov) [84][85][86] . Faith's phylogenetic diversity (PD) was used to assess α-diversity and principal component (PC) analysis of weighted UniFrac distances was used to assess β-diversity 87,88 . Relative abundance was determined using linear discriminant analysis effect size (LEfSe) and metabolic capacity was determined using 'phylogenetic investigation of communities by reconstruction of unobserved states' (PICRUSt) software compared to known pathways in the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (https://www.genome.jp/kegg) 89,90 . Statistical analyses. A mixed linear model was utilized to normalize body weight, Hb, total body Hb, HME, FCR and feed intake relative to baseline total body Hb as presented in Table 2 using MiniTab software (v 18.0, MiniTab). Significant differences between intraamniotic administration treatment groups was determined by ANOVA with a Tukey post-hoc test (p < 0.05) with additional significant differences between 'Fe' , 'Fe EDTA' and 'Fe NA' as well as between 'C WF' and 'B WF' determined by Student's t-test (p < 0.05). Significant differences in physiological measurements between 'Control' and 'Biofortified' samples were determined by Student's t-test (p < 0.05). Significant differences in Faith's PD and weighted UniFrac distances was determined by Kruskal-Wallis and permutational multivariate analysis of variance (PERMANOVA) tests, respectively and LEfSe significant differences were corrected for false discovery rate (FDR).

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.