Is it true aleurone in the thick aleurone rice mutant?

Background Diet-related non communicable diseases (NCDs) impose a heavy burden on human health worldwide. Rice is a good target for diet-related disease prevention strategies because it is widely consumed. Liu et al. demonstrated that increasing the number of cell layers and thickness of putative aleurone in ta2-1 (thick aleurone 2-1) mutant rice enhances simultaneously the content of multiple micronutrients. However, the increases of aleurone-associated nutrients were not proportional to the increases in the aleurone thickness. Results In this study, rst, cytohistological analyses and transmission electron microscopy demonstrated that the multilayer in ta2-1 exhibited aleurone cell structural features. Second, we detected an increase in insoluble bre and insoluble bound-phenolic compounds, a shift in aleurone-specic NNSP (neutral non-starch polysaccharide) prole, enhancement of phytate and minerals such as iron, zinc, potassium, magnesium, sulphur, and manganese, enrichment of triacylglycerol and phosphatidylcholine but slight reduction in free fatty acid, and an increase in oleic fatty acid composition. Conclusion These ndings support our hypothesis that the expanded aleurone-like layers in ta2-1 maintained the distinctive aleurone features and composition associated with true aleurone. We provide perspectives to achieve even greater lling of this expanded micronutrient sink to alleviate the burden of NCDs.


Statistical methods and assumptions
Tukey's Honestly Signi cant Difference (Tukey's HSD) test was used to test the hypothesis that the nutrient content of ta2-1 mutant rice was different from wild-type with the null hypothesis as no difference (Kirk, 1968;Tukey, 1953). Statistical analyses were conducted using R and all the results reported as signi cant were those with signi cance level of P<0.05 (R Core Team, 2013). The exact pvalue of all the statistical analyses in this study can be found in the le of Fig. 5-12_RawData.xlsx in the supplementary information.
Cultivation of the wild-type and ta2-1 mutant rice for analysis Rice mutant ta2-1 was backcrossed three times (BC3) to remove unwanted ethyl methanesulfonate mutations, following by two rounds of self-pollination and selectionto identify lines homozygous and stable for the thick-aleurone phenotype. The rice grains of BC3F3 homozygous ta2-1 and wild-type ZH11 (Zhonghua 11) were sown side by side in the eld for the grain increase in Haidian, Beijing, Peoples Republic of China (39.95 o N,116.38 o E). In this study, 50 individual plants from ta2-1 and wild-type ZH11 were harvested and analysed. The hull (palea and lemma) of the rice caryopsis was removed. No polishing was done so that the bran layer (including the rice aleurone layers) was maintained. In order to represent the mean of all the grains harvested and to mimic the industrial crop product processing, rice caryopses from 50 individual plants from ta2-1 and wild-type were milled to wholegrain rice our by Laboratory Mill 3100 (LM3100, Perten) in the same day and under similar temperature and humidity. This milled wholegrain rice our sample was used for all the nutritional and biochemical analyses. All the replicates thus represented technical replicates from a single biological repeat of bulk wholegrain rice our sample. BC3F4 rice seed of ta2-1 and ZH11 lines has been analysed in Liu et al. (Liu et al., 2018) and BC3F5 seed was used for the current study.
Sudan red staining of aleurone tissue Stain solution was prepared by dissolving 1g of Sudan red IV in 50ml of polyethylene glycol solution (202398, Sigma-Aldrich), incubated at 90°C for one hour, and mixed with equal volume of 90% glycerol. After removing the fruit coat (palea and lemma) of each grain, mature rice grains were incubated in distilled water for ve hours and then sectioned transversely or longitudinally using a razor blade.
Sections were stained in Sudan red solution at room temperature for 24 to 72 hours, followed by counterstaining with Lugol's staining solution (32922, Sigma-Aldrich) at room temperature for 20min and observed under stereomicroscope (M80, Leica Microsystems) (Sreenivasulu et al., 2010).
In Calco uor white staining, sections were stained in 0.01% Calco uor white solution (18909, Sigma-Aldrich) at room temperature for 2min and examined by light microscopy.
In PAS and CBB staining, the xed sections on slides were incubated in preheated 0.4% periodic acid (375810, Sigma-Aldrich) at 57°C for 30min, followed by rinsing in distilled water for three times. Schiff reagent (3952016, Sigma-Aldrich) was applied and the slides were incubated at room temperature for 15min, following by rinsing three times in distilled water. The sections were then incubated in 1% Coomassie blue R-250 (20278, ThermoScienti c) at room temperature for 2 min and rinsed three times with distilled water. Sections were dehydrated in a series of alcohol solutions of 30%, 50%, 60%, 75%, 85%, 95%, and 100% ethanol for 2min each, followed by clearing of each slide in 50% xylene and 100% xylene solution (534056, Sigma-Aldrich) for 2min each. The sections were then mounted on coverslips with Eukitt® quick hardening mounting medium (03989, Fluka) and observed under a light microscope.

Transmission electron microscopic observation
After removing the fruit coat (palea and lemma) of each grain, mature rice grains were transversely sectioned by a razor blade. Sectioned rice grain was xed in 2.5% glutaraldehyde in a 0.1M sodium phosphate buffer (pH 7.0) for one hour, washed in the same buffer, and incubated at room temperature for 30min for three times. The xed rice grain was then undergone post-xation in 1% osmium tetraoxide at 4°C for 16 hours (overnight). After xation, samples were washed in sodium phosphate buffer (pH 7.0) for 30min for three times, followed by dehydration in a series of alcohol solutions of 30%, 50%, 60%, 75%, 85%, 95%, and 100% ethanol for 10min each. The dehydrated samples were then in ltrated with acetone and Spurr® solution (14300, Electron Microscopy Sciences) in a series of solution with acetone to Spurr ratio of 1:0, 2:1, 1:1, 1:2, and 0:1 for four hours each, and embedded at 60°C for 16 to 24 hours.

Total mineral content estimation and minerals composition measurement
The total mineral content of samples was measured by ash assay according to AOAC Method 923.03 (AOAC International, 2016). In ash assay, about 2g of desiccated rice our was heated at 540°C for 15 hours and the mass of ash residue was then weighed.
Minerals composition was determined by ICP-OES (inductively coupled plasma optical emission spectrometry) (Zarcinas & Cartwright, 1983;Zarcinas et al., 1987). In minerals composition assay, about 0.5g of rice our was digested using tube block digestion with 8M nitric acid at 140°C for eight hours. Zinc, iron, potassium, magnesium, phosphorus and sulphur content were then analysed using ICP-OES at CSIRO, Urrbrae, Adelaide, South Australia, at Waite Analytical Service (University of Adelaide, Waite, South Australia, Australia).

Total phytic acid content
Determination of the phytate content of the our samples was conducted based on the method of Harland and Oberleas (1986), as described in AOAC Method 986.11 (AOAC International, 2016;Harland & Oberleas, 1986). Brie y, 0.5 g cereal our sample was weighed, extracted with 2.4% hydrochloric acid for one hour, and centrifuged. The supernatant was diluted and subjected to anion exchange column (500mg, 59822065, Agilent Technologies) to remove the non-phytate elements. Phytate bound to the column was then eluted with 2M hydrochloric acid. Phosphorous levels were determined by spectrophotometer using the molybdate and sulphonic acid colouring method with absorbance readings at 640nm. Phytate was calculated using the following formula: Phytate (mg/g) = P conc*V1*V2/(1000*sample weight*0.282) where P concentration is the concentration of phosphorous (μg/ml), as determined by spectrophotometry, Samples of 1g of rice our were undergone sequential enzymatic digestions of heat stable α-amylase (300 U/ml, E-BLAAM, Megazyme) and amyloglucosidase (3300 U/ml, E-AMGDF, Megazyme) to hydrolyse the starch content, followed by protease (350 tyrosine units/ml, E-BSPRT, Megazyme) to depolymerise the protein content. After the enzymatic digestions, different extraction and puri cation methods have applied to yield the total-, soluble-and insoluble-bre fractions. Direct precipitation of sample after enzymatic digestions yielded the total-bre; ltration of the enzymatic digested sample through Celite (61790-53-2, Sigma-Aldrich)-embedded fritted glass crucible, following by ethanol precipitation of the ltrate yielded the soluble-bre while the residue after ltration represented the insoluble-bre. After the enzymatic digestions and bre extraction, the total-, soluble-and insoluble-dietary bres were determined by the gravimetric measurements of the residue dry mass of the total-precipitated-residue, the ltrateprecipitated-residue and the residue after digestion respectively, with the correction from the parallel measurement of ash and protein content of the triplicate of each sample.
Total NNSP NNSP was measured by the gas chromatographic procedure according to AOAC Method 994.13, as detailed by Theander et al. with slight modi cation (AOAC International, 2016;Theander et al., 1995). Brie y, the total-dietary bres extracted from total dietary bre assay were hydrolysed with sulfuric acid.
The neutral sugars released were then reduced by potassium borohydride solution and acetylated by acetic acids to alditol acetates, which are then quanti ed by gas chromatography. Gas chromatography conditions: BPX70 column (30m x 320µm x 0.25µm); injection 0.5μl; inlet at 240°C; 20:1 split; 1.62ml/min constant ow; oven at 180°C for 2min, 12 o C/min to 230°C and hold for 8min; ame ionization detector at 250°C; quanti cation against internal allose standard.

Total phenolic compounds
Total phenols assay by FCR is an electron transfer-based assay, which estimates the amount of antioxidant based on the reducing power (willingness to donate electrons) of the antioxidants against oxidant probe molybdates, quanti es the reducing power with gallic acid standards, and presents as the value of gallic acid equivalents. Combined with the differential extractions of the free-, conjugated-, bound-, and total-fractions, FCR assay is good at determining the antioxidant capacity of different groups of phenolic compounds.
Total phenolic compounds content, as well as phenolic compounds in the free-, conjugated-, and boundstates were determined according to the method described by Li et al. with minor modi cations (Li et al., 2008). Brie y, these four types of phenolic compounds were extracted from 100mg samples with different extraction methods. Total-phenolic compounds were determined using 100mg of samples, adding 200µl 80% methanol, followed by alkaline hydrolysis with (2M) sodium hydroxide; free-phenolic compounds were represented by the 80% methanol extraction of the 100mg samples; conjugatedphenolic compounds were the alkaline hydrolysed products from the 80% methanol extract; boundphenolic compounds were the alkaline hydrolysed products from the residues followed by the methanolic extraction of the free-phenolic compounds.
The amount of phenolic compounds in the treated/extracted samples was measured using FCR assay with reference to standard curve of known gallic acid concentrations. 1ml of standards were added to 4ml glass tubes. For test samples, 100μl aliquots of thoroughly mixed samples were added to 900μl water in 4ml glass tubes. 100ml of FCR (F9252, Sigma-Aldrich) was then added to each tube and vortexed immediately. 700μl sodium bicarbonate solution (1M) was added after 2min and mixed by vertexing. Each solution was incubated at room temperature in the dark for one hour. Absorbance was read at 765nm. Results were expressed in μg gallic acid equivalents/g sample.

Folate content
Total folate (Vitamin B9) was measured by commercially available VitaFast Folic acid kit (Folic Acid AOAC-RI, 100903, R-Biopharm) according to AOAC method 2004.05 as described by DeVries et al. (DeVries et al., 2005). The method incorporated the in vitro enzymes digestion and the growth response of the Lactobacillus to folate concentration in culture medium. First, pancreatin digestion was used for the release of the food matrix-bound folic acid. About 1g of rice our and 20mg pancreatin were added and lled up to 40ml with phosphate buffer (0.05M, 0.1% ascorbate, pH7.2), and the sample was incubated at 37°C for two hours in dark. These allowed the digestion of food matrix by protease and amylase in pancreatin and the hydrolyzation of polyglutamates to diglutamates by pancreatic conjugase to release food-bound folate which can be metabolized by Lactobacillus. After the incubation, the 1.5ml reaction was centrifuged at 8000 x g for 5min, diluted to 10ml with sterile distilled water, heated at 95°C for 5min, and ltered through a 0.2µm lter to get rid of the insoluble or undigested suspensions in the sample. The ltrate was then diluted in several dilutions, added into the wells of a microtiter plate coated with Lactobacillus casei subspecies rhamnosus, and incubated at 37°C for 44 to 48 hours. After the incubation, the OD at 610-630nm of the sample was measured and compared with the folate calibration standard. As the basal culture medium is lack of folate, the growth of the coated Lactobacillus, represented by the increase in turbidity of bacterial culture, is positively correlated with the extracted folate content in the sample. The folate concentration of the sample can be measured by comparing difference between the absorbance of the test sample with the calibration standards at OD 610-630nm.

Total lipid content
Total lipid was measured according to AOAC Method 983.23 (AOAC International, 2016). Sample of 5g rice our was incubated with 1% Clarase (MC23.31, Southern Biological) in 0.5M sodium acetate solution at 45°C for one hour. Lipids were extracted from the sample into chloroform/methanol by multiple extractions. The samples were then subjected to homogenization with multiple additions of chloroform/methanol. After centrifugation of the mixture into separate phases, the chloroform/methanol fraction was removed and dried at 101°C for 30min to recover the lipid. The total lipid in the sample was represented by the mass of residue after drying.

TAG, FFA, and PC fractionation and quanti cation
According to the methodologies described by Liu et al., the extraction of total lipid, the fractionation of neutral lipid (mainly TAG), free fatty acid, and polar lipid (mainly PC), and the quanti cation of lipid were carried out (Liu et al., 2017). Thin layer chromatography (TLC) was rst carried out to separate the neutral lipid and free fatty acid in solvent matrix of hexane:diethyl ether:acetic acid in 70:30:1 volume ratio. Then, another TLC was conducted in solvent matrix of chloroform:methanol:acetic acid:distilled water in 90:15:10:3 to separate the polar lipid PC. After TLC, samples were collected and lipid content in the fraction was extracted. The lipid fractions were undergone fatty acid methyl esters (FAME) and gas chromatography (GC) analyses, and quanti ed according to the methods described by Vanhercke et al. (Vanhercke et al., 2014).

Total β-glucan
Rice β-glucan was measured with reference to the methods in AOAC Method 995.16 (AOAC International, 2016).
Brie y, 20mg of rice our were subjected to sequential enzymatic digestions of Lichenase and βglucosidase, followed by quanti cation of the released glucose with standard glucose oxidase/peroxidase (GOPOD) system. Firstly, 200µl 50% ethanol was added into 20mg of rice our, then 1ml sodium phosphate buffer (20mM, pH6.5) was added and incubated at 100°C boiling water for 3min. Secondly, the reaction was brie y cooled and diluted. Thirdly, 10µl Lichenase (1U/µl) was added and incubated at 40°C for 1hour. The sample was then diluted by adding 3.8ml distilled water followed by centrifugation. Fourthly, the β-glucosidase reaction was conducted by adding 10µl of β-glucosidase (2U/ml) into 10µl sample supernatant collected in last step and incubated at 40°C for 15min.
Quanti cation of the glucose through GOPOD was conducted after the enzymatic reactions. Firstly, in 20µl sample, 150µl of glucose oxidase/peroxidase (GOPOD) reagent was added and incubated at 40°C for 20min. Secondly, the absorbance was measured at 510nm for each sample (E A ) and reagent blank (E BLANK ). The β-glucan content was measured using the following formula:

Results
The time course of aleurone development in ta2-1 To compare the aleurone development in ta2-1 and wild type, caryopses of 6, 8, 10, 12, 15, 18, 21, 24, and 30DAA (days after anthesis) were examined. These time points represent important developmental phases such as aleurone cell fate differentiation, aleurone nutrient accumulation, and aleurone maturation. These caryopses were sectioned and stained with Sudan red plus Lugol's iodine. The aleurone cells in ta2-1 become prominent between 6 to 8DAA, being positively stained red by Sudan red (Fig. 1) indicating the dominance of lipid and almost complete absence of starch. Wild type had a thick aleurone resembling ta2-1 up to 10DAA. From 12DAA onwards, ta2-1 showed a thicker Sudan red staining area, although a slight decline of red staining area at 24 and 30DAA. Although there is no indication that starchy endosperm cells can become aleurone, it seems possible that some cell layers that initially show the compositional attributes of aleurone can become starchy endosperm in the later stages of development. In maize, dek1 mutation triggered the transdifferentiation of the aleurone to starchy endosperm cells in the late stage of development (Becraft & Yi, 2011). Furthermore, this switch is more pronounced in wild type caryopses than in ta2-1, leaving ta2-1 with a thicker aleurone at maturity.
General cell structure of ta2-1 To assess the cellular and subcellular identities of aleurone cells in rice ta2-1, light microscopy coupled with two different staining methods were adopted.
Firstly, using periodic acid-Schiff and Coomassie brilliant blue staining (PAS-CBB), multiple layers of putative aleurone cells appeared distinctly different from starchy endosperm cells at the periphery of the mature rice caryopsis in ta2-1. The multiple layers of putative ta2-1 aleurone cells had thicker cell walls (stained pink by PAS) than the inner starchy endosperm cells ( Fig. 2b and d, black arrow), prominent protein bodies (stained blue by CBB) (Fig. 2e, asterisk), and much lower starch granule density ( Fig. 2f and g, black closed triangle). The putative aleurone cells in ta2-1 were more irregular in shape than wild type aleurone cells, and with fewer but larger protein bodies. Moreover, their cell distribution patterns were different. In wild type, there was one aleurone layer on the ventral side (Fig. 2a) and up to ve aleurone layers on the dorsal side of the rice caryopsis (Fig. 2c). In ta2-1, the aleurone layer was about three layers thick on the ventral side (Fig. 2b) and more than eight layers (Fig. 2d) on the dorsal side.
Secondly, Calco uor White staining was applied with epi uorescence microscopy to study the cell wall structure speci cally. Calco uor White binds to cellulose and other β-glucans in cell wall (Hughes & McCully, 1975). In both wild type and ta2-1, the outermost compressed cells of the testa and the aleurone were actively stained while the innermost starchy endosperm cells were not ( Fig. 3a and b). The putative extra aleurone layers of ta2-1 resembled authentic aleurone in this distinctive cell wall trait.
The accumulation of aleurone grains in ta2-1 Transmission electron microscopy was used to examine the subcellular details of the putative aleurone cells in ta2-1. In both wild-type and ta2-1 aleurone cells, thick cell walls were evident ( Fig. 4a and b, between two arrows). Also, protein bodies (sometimes referred to as aleurone grains (AG)) were abundant in both genotypes. As in the light microscopy study, AG in ta2-1 were generally larger than the wild type with similar subcellular content. The ta2-1 AG had other more subtle differences compared to wild type AG. Surrounded by tonoplast membrane, AG accumulate protein presumably as a source of amino acids for germination; they consist of three morphologically distinct features called the crystalloid, matrix, and globoid. These three features were present with similar electron density in the aleurone grains of both the wild type aleurone and the expanded ta2-1 cells ( Fig. 4c and e, arrows), however, the ta2-1 AG were more variable in size with many being larger.
Together, the results from light microscopy and electron microscopy supported the conclusion that the putative extra aleurone cells in ta2-1 retained the aleurone cell features of thick cell wall, low content of starch granules and presence of protein-rich AG. Distinctive AG features of crystalloid, matrix, and globoid were conserved in ta2-1. However, the ta2-1 aleurone cells had a greater variability in cell size and shape of AG.
To measure the spatial distribution of aleurone cells, light microscopy coupled with PAS was adopted.
From the outer to inner layers of the ta2-1 aleurone, there was a change in structure and subcellular content of the AG. In wild type, only one type of AG was observed. The wild-type AG had a small internal cavity and abundant protein matrix that was stained deep blue by CBB. Moreover, individual aleurone cells were smaller in size (Fig. 2a and c) to those in ta2-1 ( Fig. 2b and d). The subcellular content and structure of the wild-type AG was uniform and constant. However, in ta2-1, two types of AG could be distinguished, i.e. AG with wild type morphology ( Fig. 2e and f, single asterisk) and larger aleurone grains (LAG) (Fig. 2g, double asterisks). AG and LAG had similar deep-blue protein matrix feature, however, LAG had a larger internal globoid cavity with sometimes more limited protein matrix than AG. Moreover, the distribution patterns of LAG changed from outer to inner layers of the ta2-1 aleurone. In the outer layers of ta2-1 aleurone, the dominant storage compartment is AG while in inner layers, the dominant storage vacuoles resemble the LAG structure.
The contents of dietary bre, phenolic compounds, and antioxidants The aleurone cell wall in brown rice is an abundant source of dietary bres. We previously showed that total dietary bre content was increased in ta2-1 (Liu et al., 2018). In this study, we compared the insoluble-and soluble-dietary bre in the total bre content of ta2-1. Similar to our previous ndings, rice mutant ta2-1 showed 66% increase in total dietary bre. This was largely due to a 55% increase in insoluble-dietary bre, while no signi cant change (P=0.90>0.05) was observed in the content of solubledietary bre (Fig. 5). Since it is known that the bre associated with the aleurone cell wall is insoluble in nature (Collins et al., 2010), the observed increase in total and insoluble bre in ta2-1 is consistent with the thickened aleurone being the source of the extra bre.
Dietary bre is composed of neutral non-starch polysaccharide (NNSP), nondigestible oligosaccharide, resistant starch, lignin, and uronic acid containing polysaccharides (Theander et al., 1995). Among these components, NNSP is the major contributor of the total dietary bre content in rice caryopsis (Collins et al., 2010). The total NNSP content in the wholegrain our of ta2-1 was 61% higher than that in the wildtype (Fig. 6a). Moreover, being consistent with the earlier studies in different wholegrain (brown) rice varieties (Collins et al., 2010;Demirbas, 2005), the β-glucan content in the wild type and ta2-1 was very low at 0.63g and 0.59g per 100g respectively (Fig. 5). Aleurone is rich in arabinoxylan and low in cellulose and β-glucans, but starchy endosperm has an even distribution of these three bre types (Fincher & Stone, 2004). The arabinose content increased from 18% in the wild type to 24% in ta2-1, and the xylose content from 17% to 20%; while glucose content dropped from 50% in the wild type to 40% in ta2-1 (Fig. 6b), leading to changed NNSP composition pro le in ta2-1 (Fig. 6c). The basic monosaccharide units of arabinoxylan are arabinose and xylose, and the basic monosaccharide unit of cellulose and β-glucans is D-glucose. The increase in arabinose and xylose content at the expense of D-glucose in NNSP suggested that the expanded aleurone layers in ta2-1 maintained the aleurone cell identity of arabinoxylan-rich cell wall and accounted for the composition change in total NNSP.
Antioxidants of rice include a wide spectrum of biomolecules such as phenolic compounds, uronic acids, tocopherols, carotenoids, ascorbic acid, and gamma-oryzanol which stabilize multiple oxidant sources and free radicals by their electron-transferring or hydrogen-transferring ability (Prior et al., 2005). Some antioxidants can be esteri ed and bound to the aleurone cell wall. Our previous studies focused on the electron-transferring ability of wholegrain rice our through ORAC (oxygen radical absorbance capacity) and the hydrogen-transferring ability through FCR (Folin-Ciocalteu reagent) (Liu et al., 2018). Here we tested their association with the aleurone cell wall, by extracting antioxidant fractions with different solubility and measuring their hydrogen-transferring abilities.
Total-phenolic compounds in plants may exist in soluble free, soluble conjugated (esteri ed), and insoluble bound forms (Robbins, 2003). The antioxidants were rst extracted into the four fractions of total-, bound-, conjugated-, and free-phenolic compounds from the wholegrain rice our samples. Their anti-oxidation abilities were then measured by the FCR assay. In both wild-type and ta2-1, about 70% of total phenolic compounds in rice caryopsis were insoluble bound-phenolic compounds, followed by 16% soluble free-and 10% soluble conjugated-phenolic compounds. Compared with the wild-type, ta2-1 showed 43% increase in total-, 31% increase in insoluble bound-, 26% increase in soluble free-(not statistically signi cant) and 99% increase in soluble conjugated-phenolic compounds (Fig. 7). This pattern of increases is consistent with the increased aleurone thickness being responsible for the increased antioxidant capacity.

The contents of minerals and phytate
Analyses of phytate and phosphorus content revealed increases of 18% and 22% respectively in ta2-1 compared to wild type (Fig. 9). Assuming the molecular mass of phytate is 660.04gmol -1 and that of phosphorus is 30.97gmol -1 and each phytate molecular consists of six phosphorus molecules, 75.10% and 72.47% of total phosphorus are bound to phytate in wild type and ta2-1 respectively (Supplementary Table 1). These measurements resemble the nding of others in rice caryopsis that 73.7% of total phosphorus content is bound to phytate (Ravindran et al., 1994). The similarity of percentage increase between phytate (18%) and phosphorus (22%) and the similar proportion of phosphorus bound to phytate suggested the increase of phosphorus in ta2-1 is fully explained by the increase in phytate. Moreover, in addition to the increases in iron, zinc, and magnesium as observed by Liu et al. (2018), increases in manganese, potassium, and sulphur but not calcium content was observed (Fig. 10). Calculation of phosphorus content is based on the equations as follows: 1 The molar mass of phosphorus is 30.97gmol -1 , therefore, mole of phosphorus = mass of phosphorus / molar mass. 2 The molecular mass of phytate is 660.04 gmol -1 , therefore, mole of phytate = mass of phytate / molecular mass. 3 As the molecular formula of phytate is C 6 H 18 O 24 P 6 , one mole of phytate contains six moles of phosphorus.

The compositions and contents of lipid
It has been shown before that ta2-1 mutant had higher total lipid content (Liu et al., 2018). The total oil content and composition in rice aleurone (rice bran) is different from starchy endosperm (Choudhury & Juliano, 1980a, 1980b. These authors showed that the neutral lipids (largely triacylglycerol (TAG)) were concentrated in the bran (embryo and aleurone), while the phospholipids were equally distributed in the bran and starchy endosperm fractions. In our studies, lipid components were separated using thin-layer chromatography and fatty acid content quanti ed by gas chromatography. In both ta2-1 and wild type, TAG was the dominant type of lipid, followed by FFA (free fatty acid), and PC (phosphatidylcholine). In comparison with wild type, ta2-1 had a 79% increase in TAG (from 1.89% to 3.32%), 97% increase in PC (from 0.02390% to 0.04715%), and a 7% decrease in FFA (from 0.2171% to 0.2004%; not statistically signi cant) in wholegrain rice our samples (Fig. 11). The large increase in TAG is expected if the thick aleurone is responsible for the increase in total lipid. There were changes in fatty acid pro le in ta2-1 (Fig.  12a), including 32% increase in oleic acid content (from 32% to 42% of the total fatty acid), 22% decrease in linoleic acid content (from 35% to 28%), and 5% decrease in palmitic acid content (from 17% to 16%) ( Fig. 12b).

Discussion
General aleurone identity of the ta2-1 thick aleurone The aleurone cells in ta2-1 maintained the aleurone features of thick cell wall, with few starch grains, an abundance of AG and lipid bodies, and as a consequence increases in dietary bre, phenolic compounds, lipid composition, and changes to fatty acid pro le matching what would be expected if the increases come from normal aleurone cells.
Cytohistology with Calco uor white staining, cytohistology with PAS-CBB staining, and transmission electron microscopy all demonstrated a dramatic expansion of thick cell walls in the entire ta2-1 aleurone. The AG in ta2-1 aleurone cells had the same distinctive structures of crystalloid, matrix, and globoid as wild-type AG, however, they were more variable in size with many being larger than in wild-type aleurone due mainly to larger globoid.
The increase in dietary bre content in brown rice ours of ta2-1 was mainly due to insoluble-dietary bre, the type of bre that forms aleurone cell walls (Collins et al., 2010). The increase in arabinose and xylose content in ta2-1 our and decrease in D-glucose was also consistent with the predominance of arabinoxylan in aleurone cell walls (Fincher & Stone, 2004).
The enhanced thick cell wall in ta2-1 also changes the phenolic compound composition. Associated with the cell wall are the phenolic compounds such as ferulic acid and ρ-coumaric acid. They are mainly bound to NNSP (Goufo & Trindade, 2014). In ta2-1, the increase in total phenolic compounds was mainly attributable to an increase in insoluble bound-phenolic compounds.
Sudan red staining con rmed the enrichment of lipid bodies in the thickened ta2-1 aleurone. The increase in TAG and PC in ta2-1 our but slight decrease in FFA, was consistent with a shift to aleurone-(bran-) speci c lipid composition. Lipids in rice can be classi ed into non-starch lipids and starch-associated lipids (Zhou et al., 2003). Rice aleurone (bran) contributes to about 40% (39-41%) of total non-starch lipids that mainly consist of TAG and PC, while starchy endosperm stores about 60% (48-71%) of total starch-associated lipids with FFA as one of the major components (Choudhury & Juliano, 1980a, 1980b. In ta2-1 our, the 79% increase in TAG, 97% increase in PC, and the 7% decrease in FFA is consistent with the thick aleurone maintaining normal aleurone cell function and composition. On the other hand, as the starchy endosperm content is reduced in ta2-1, the content of starch lipid FFA decreases.
The enrichment of lipid bodies in ta2-1 aleurone also modi es the fatty acid pro le. The fatty acid composition pro les of aleurone (bran) and starchy endosperm are different. Rice aleurone is rich in oleic acid (36% of total fatty acid) and linoleic (37%) but low in palmitic acid (23%) content. In starchy endosperm, linoleic and palmitic acid content are higher (41% and 33% respectively) but the oleic acid content is lower (20%) (Choudhury & Juliano, 1980a, 1980b. The 32% increase in oleic, 22% decrease in linoleic, and 5% decrease in palmitic acid proportions again indicates that the multilayer aleurone in ta2-1 maintains the aleurone-speci c fatty acid pro le. Aleurone cells are rich in mitochondria (Jones, 1969). In this study, no direct measurement of mitochondria was conducted, however, the folate content may be indicative of mitochondria abundance.
Folate is enriched in aleurone-enriched bran fraction (Houston & Kohler, 1970). In plant, folate is synthesized in three subcellular compartments in which the nal ve steps are conducted in mitochondria (Gorelova et al., 2017). About 30 -50% of folate in cells is stored in mitochondria to help maintain the mitochondrial DNA stability (Depeint et al., 2006). Therefore, the 32% increase in folate may indirectly indicate that the multilayer aleurone in ta2-1 maintains the high mitochondrial content of true aleurone.
Together, the cell structural features, the increase in insoluble bre and insoluble bound-phenolic compounds, shift in aleurone-speci c NNSP pro le, enrichment of TAG and PC but slightly reduction in FFA, and increase in oleic fatty acid composition collectively support the hypothesis that the additional aleurone-like layers in ta2-1 maintain the distinctive features and composition of true aleurone cells. It's the expansion of the aleurone layer that results in an increase of the nutrients that are associated with true aleurone.
The improvement of multiple nutritional factors in ta2-1 can help prevention of NCDs. For example, dietary bre can maintain gastrointestinal health by increasing fecal bulk, decreasing transit time, decreasing the gastrointestinal contact of foodborne carcinogenic compounds, binding to mutagens, and lowering colonic pH (Glitsø et al., 1998;Le Gall et al., 2009). The insoluble antioxidants bound to cell wall materials can provide an antioxidant environment contributing to protection of the colon tissues from cancer (Sengupta et al., 2001). Phytate has potential anti-neoplastic and antioxidant functions (Fox & Eberl, 2002;Norhaizan et al., 2011). The multilayer aleurone rice with true aleurone composition can coordinately improve multiple nutritional factors to achieve diverse protective outcomes. In countries with rice-dominant diets, the consumption of wholegrain rice is very low (Cleveland et al., 2000;Harnack et al., 2003), but there is a growing trend from white rice to brown rice (with aleurone) consumption due to improved health consciousness and education (Selvam et al., 2017). By improving the nutrient composition of brown rice that is more readily accepted by the public, the multilayer aleurone rice ta2-1 could therefore deliver substantial public health advantage without remarkably changing the dietary habits. Furthermore, light polishing of thick aleurone rice retains more aleurone than wild type rice and therefore also retains more of the nutrients.
The commensurate increase in minerals and phytate content Wholegrain our of ta2-1 had an 18% increase in phytate content. Using ICP-OES, we con rmed and extended our previous ndings (Liu et al., 2018), showing ta2-1 also had 14 to 23% increased content of various minerals.
Aleurone is a concentrated source of many essential minerals. In synchrotron X-ray uorescence microscopy imaging study of rice, iron, zinc, manganese, and copper were highly concentrated in the aleurone layer (Hansen et al., 2012). Most of the minerals in rice are associated with phytate (Hansen et al., 2012;Mills et al., 2005;Simic et al., 2009). Wholegrain rice and wheat have similar phosphorus content (337mg vs 323mg/100gDW) (USDA, 2019), however, rice has a higher proportion of phytate phosphorus than wheat, accounting for 77% of the total phosphorus content (Ravindran et al., 1994). Phytate has a strong a nity for chelating minerals such as Zn, Fe, and Mg, which limits their absorption in the small intestine (Bohn et al., 2007;Raboy, 2009). This raises concern that phytate may impair small intestinal mineral absorption and compromise mineral status.
Wholegrain wheat our has higher phytate and iron content than processed white our, and in an in vitro Caco-2 cells test, wholegrain wheat our led to a lower ferritin response than processed white our, suggesting that the iron content in white our is more biologically available for cellular absorption and assimilation to ferritin (Eagling et al., 2014). Stevenson et al. summarized the ndings in wheat and suggested that the consumption of wholegrain wheat or wheat bran decreased the calcium and zinc bioavailability (Stevenson et al., 2012).
However, the Caco-2 test may not reveal the full picture concerning bioavailability. Diets high in whole grains do not adversely affect mineral nutrition but have favorable health outcomes. Different recommendations of daily wholegrain intake have been proposed in the U.S., Canada, and Australia, and no unfavourable health outcome has been reported regarding the high consumption of whole grains (Health Canada, 2011;HHS & USDA, 2015;National Health and Medical Research Council, 2013).
Recently, a meta-analysis provide further evidence that higher wholegrain consumption is associated with reduced risk of digestive tract cancers (Zhang et al., 2020). In both short-term (4 weeks) and long-term (2 years) studies in young and older women, the diet supplemented with phytate-rich wheat bran had no signi cant effect on different osteoporosis markers (Chen et al., 2004;Zittermann et al., 2007). Likewise, zinc absorption in young children was not negatively impacted by added phytate (Miller et al., 2015). Moreover, there is no consensus on the effect of phytate on iron bioavailability (Stevenson et al., 2012).
The digestion of phytate can happen in the human large intestine to release the chelated minerals for absorption. In pig studies, nearly complete (more than 97%) phytate digestion was observed in faecal samples following a normal diet with low intrinsic feed phytase (Schlemmer et al., 2001). In these studies, the highest phytate degradation occurred in the large intestine rather than the stomach or small intestine.
In human faecal studies, the dietary phytate degradation rate varied between 50 and 90% (Joung et al., 2007;McCance & Widdowson, 1935;Walker et al., 1948). In a human trial with both young and elderly women, it was also reported that the diet with high phytate content could enhance phytate degradation (Joung et al., 2007). In the large intestine, the microbial phytases, foodborne phytases in plant food sources, and endogenous phytases can degrade cereal-grain phytate to release the minerals bound in the mineral-phytate inclusion for human absorption (Sandberg & Andlid, 2002). Moreover, the fermentation of the NNSP and different dietary bres in rice can potentially acidify the luminal environment of the large bowel (Koh et al., 2016). This can improve the mineral bioavailability. Therefore, the commensurate increments of minerals and phytate in ta2-1 may not necessarily decrease the mineral availability; on the contrary, the increased phytate content in ta2-1 may stimulate the microbial phytate degradation in large intestine, releasing bioavailable minerals and digested phytate for human absorption and gut microbiome development. However, more studies regarding the minerals' bioavailability should be conducted to further test this hypothesis.
Why are the nutrient increases not greater?
The increases in multiple nutrients in ta2-1 our are signi cant but not in proportion to the increased layers or volume of aleurone cells. We brie y explore here two hypotheses, which are not mutually exclusive, that might explain this "nutrient gap".
Hypothesis 1: The new larger aleurone may have a lower capacity to synthesize and/or store nutrients compared to wild-type aleurone.
While we have shown the expanded aleurone in ta2-1 has the general features and composition of normal aleurone, there may be more subtle differences in biosynthetic or storage capacity. In rice, both wild-type and ta2-1 aleurone cells have prominent AG. However, the AG in ta2-1 aleurone cells are more variable in morphology and distribution pattern from outer to inner layers.
The rice ta2-1 had two types of AG, morphologically normal AG and LAG. While both types of aleurone grains had prominent globoid, crystalloid, and matrix structures, LAGs in ta2-1 had enlarged globoid cavities. The crystalloid is the storage compartment for the integral membrane proteins, whereas the matrix contains soluble protein (Jiang et al., 2002;Jiang et al., 2000). The globoid is the mineral storage compartment consisting of mineral-phytate inclusion crystals. Studies from energy-dispersive X-ray analysis in rice and synchrotron soft X-ray microscopy in wheat suggested that minerals and phosphorus are co-localized with the globoid structures in the AG (Ogawa et al., 1979;Regvar et al., 2011). However, during the sample preparation for transmission electron microscopy, the mineral-phytate crystals in the globoid can be easily dissolved out, thus leaving an electron-transparent internal globoid cavity in most of the electron microscopic analyses (Jacobsen et al., 1971). Despite these differences, some of the LAG in ta2-1 accumulate protein that is stained by CBB, suggesting they are still functioning as storage compartments.
PAS and CBB staining also suggested that the distribution patterns of AG change from outer to inner layers of the ta2-1 aleurone, with AG resembling wild-type structure in the outer layers, but LAG with empty or limited protein matrix predominant in the inner layers. The increased size of ta2-1 aleurone cells and subcellular aleurone grains may signal a potential decrease in protein and mineral storage capacity from the outer to inner layers ta2-1 aleurone.
Hypothesis 2: The aleurone development does not synchronize with the nutrient accumulation during grain development.
From 3-5DAA is the endosperm cellularization stage, during which cells at the periphery rst take on distinguishable aleurone features. From 6-9DAA, the aleurone and starchy endosperm cells continue to divide and expand during the endosperm differentiation stage (Olsen et al., 2008;Wu et al., 2016). The proportion of aleurone tissue to starchy endosperm tissue is high in these early stages as compared with the mid and late grain developmental stages with limited cell division. This may explain why thickaleurone was observed in both wild type and ta2-1 up to 10DAA. Our histological analyses suggested that aleurone development in ta2-1 occurs simultaneously with aleurone in the wild type throughout all phases of grain development, so it is likely biosynthetic activity and nutrient accumulation will be synchronized with the normal activities during grain development.
Of these hypotheses, the rst appears to have relevance regarding nutrients synthesized in the aleurone, and requires further investigation.

Future Perspectives
The embryo is high in vitamin B1, vitamin E, and lipid, (Juliano, 1993), so that any changes in embryo composition would also affect the overall wholegrain nutritional pro le. Future studies should separate and study the composition of aleurone-, starchy endosperm-, and embryo-enriched fractions of ta2-1. It may also be possible to use synchrotron analysis to explore compositional differences between the outer and inner layers of the ta2-1 thick aleurone, especially in mineral accumulation.
Future studies should also focus on assessing the bioavailabilities of minerals in ta2-1. Various digestion models such as in vitro Caco-2 cell, in vivo animal feeding trial, and human intervention studies should be applied to measure their absorption e ciencies along the gastrointestinal tract.