Structure‐reactivity relationships between the fluorescent chromophores and antioxidant activity of grain and sweet sorghum seeds

Abstract Polyphenolic structures are the putative cause of a variety of seed functions including bird/insect resistance and antioxidant activity. Structure‐reactivity relationships are necessary to understand the influence of polyphenolic chromophore structures on the tannin content and free radical quenching ability determined by the traditional calorimetric methods. This study investigated the relationships between the structural attributes of fluorescent chromophore and the following seed characterization methods: procyanidin (by acid‐butanol assay) and flavonoid (by vanillin assay) contents, radical quenching (by DPPH assay), electron‐donating capacity (by FeIII reduction), and λ max (by UV/visible spectrophotometry). Distinctively different response was observed for different seed categories: U.S. grain sorghum hybrids, African grain sorghum, and sweet sorghum. The U.S. grain sorghum varieties (low‐tannin to maximize the livestock digestion) responded only to the DPPH assay. For sweet sorghum and African grain sorghum, linear correlation was observed between (1) the antioxidant activity (2) the amounts of procyanidins and flavonoids, and (2) the aromaticity of fingerprint fluorescent structures.


Introduction
Based on the hypothesis that phenol-containing structures (including procyanidin, flavonoids, and short oligomers of phenolic structures) are responsible for the bird/insect resistance of sorghum seeds, a number of standardized tannin methods have been established (Butler 1982). For example, 4,8 linkages of procyanidin are cleaved in the acid-butanol assay, while vanillin assay targets meta-substituted flavonoids present in the terminal units of procyanidin (Butler 1982). Modified tannin methods have been developed to estimate the degree of polymerization based on the ratio from vanillin and acid-butanol assays (Butler 1982). However, these standardized tannin methods are influenced by the experimental artifacts, including the solvent composition (Xu and Chang 2007), kinetically controlled reactions, overlapping spectra near the detection wavelength, low sensitivity, and the reactivity ORIGINAL RESEARCH Structure-reactivity relationships between the fluorescent chromophores and antioxidant activity of grain and sweet sorghum seeds Minori Uchimiya 1 , Xinzhi Ni 2 & Ming Li Wang 3 of nontarget structures (Butler 1982). Structure-reactivity relationships are necessary to understand the influence of polyphenolic chromophore structures on the tannin content and free radical quenching ability determined by the traditional calorimetric methods.
The objective of this study was to investigate the structure-reactivity relationships between fluorescent chromophores and procyanidin (by acid-butanol assay) and flavonoid (by vanillin assay) contents, radical quenching (by DPPH assay), electron-donating capacity (by Fe III reduction), and λ max (by UV/visible spectrophotometry). Fourteen U.S. grain sorghum hybrids, four African grain sorghum, and four sweet sorghum samples were selected, in order to investigate diverse seed categories.

Materials and Methods
Distilled, deionized water (DDW) with a resistivity of 18 MΩ cm (APS Water Services, Van Nuys, CA) was used in all procedures. All chemical reagents were obtained from Sigma-Aldrich (Milwaukee, WI) with the highest purity available.
Total Fe, K, Mg, P, Ca, Mn, Na, and Zn concentrations of selected seed samples (seed 3, seed 4, 4401, and 4437) were determined by duplicate extraction of 1.25 g seed in 25 mL of 0.2 mol L −1 ammonium oxalate (pH 3.5) by end-over-end rotation (70 rpm) for 24 h in the dark. Filtered (0.45 μm) extracts were acidified to 4 vol% nitric acid (trace metal grade) for the determination of dissolved P, K, Ca, Mg, Al, Fe, Mn, and Na concentrations using inductively coupled plasma atomic emission spectrometer (ICP-AES; Profile Plus, Teledyne/Leeman Labs, Hudson, NH). Blanks, blank spikes, and matrix spikes were included for the quality assurance and control for the ICP-AES analysis (USEPA, 2001). As reported in the literature (Pontieri et al. 2014), all four seed samples were enriched with Mg and Fe (Table S2).
After adding 0.2 mL of 2% (w/v) NH 4 Fe III (SO 4 ) 2 (in 2 mol L −1 HCl), the reactor was vortexed and then placed in boiling water bath for 50 min. Full spectrum (280-650 nm) was taken before and after boiling, and absorbance was recorded at 550 nm. Blank spectra were obtained for each extract before boiling. Calibration was obtained by repeating above-described procedures for 1-5 mg L −1 delphinidin chloride.
Vanillin assay for meta-substituted flavanoids was conducted following the literature protocol (Price et al. 1978). Briefly, 1 mL of methanol extracts (20 g L −1 seed in methanol) was reacted with 5 mL of working reagent (2.5 mL of 1% vanillin + 2.5 mL of 8% HCl) in 30°C water bath for 20 min, and the absorbance was recorded at 500 nm. Above-described procedure was repeated to obtain (1) blanks (1 mL of sample and 5 mL of 4% HCl) and (2) five-point calibration (1 mL of catechin standards + 5 mL of working reagent).

Fluorescence EEM spectrophotometry with PARAFAC
Fluorescence excitation-emission (EEM) spectra of methanol and acetone-water extracts for each seed sample were obtained without dilution using F-7000 spectrofluorometer (Hitachi, San Jose, CA) set to 220-500 nm excitation and 280-730 nm emission wavelengths in 5 nm intervals; 5 nm excitation and emission slits; 0.5 sec response time; and 2400 nm per min scan speed. As described in detail elsewhere (Stedmon and Bro 2008), parallel factor analysis (PARAFAC) models three-way data (samples, excitation wavelengths, and emission wavelengths) by minimizing the sum of squares of the residuals. The blank EEM for background solution (methanol and acetone-water) was obtained daily, and was subtracted from each sample to remove the lower intensity Raman scattering (Christensen et al. 2005). After the removal of additional regions dominated by Rayleigh and Raman peaks and the region without fluorescence, PARAFAC modeling (Stedmon and Bro 2008) was conducted with non-negativity constraint using MATLAB version 8.5.0.197613 (R2015a) with PLS toolbox version 8.0.1. For methanol extracts, raw EEM spectra were further preprocessed by normalization to the maximum intensity of each seed sample. On the basis of (1) residual/leverage analysis, (2) comparison with the raw EEM spectra, and (3) core consistency diagnostic scores of 2-7 component models, 3 component model was selected to interpret PARAFAC results of methanol and acetonewater extracts.

Results and Discussion
Characterization of seed samples by DPPH, acid-butanol, ferrozine, vanillin, and UV/vis spectra Table 1 presents the characteristics of 14 U.S. grain sorghum hybrids, four sweet sorghum, and four African grain sorghum seeds: DPPH radical quenching (in μg trolox g −1 seed), acid-butanol assay (in μg delphinidin g −1 seed), reduction of 50 μmol L −1 Fe(III), vanillin assay (in mg catehin g −1 seed), λ max , and absorbance at 210 nm. All values are given as mean ± S.D. of triplicate seed extractions. For the U.S. grain sorghum seeds, aphid/webworm/bird resistance responses information was available in the literature, and are provided as fair (F), good (G), and very good (VG) (Ni et al. 2014).
For the U.S. grain sorghum seed extracts, the values were below detection limit for acid-butanol, ferrozine, and vanillin assays. In addition, a single λ max was observed at low wavelength of 210 nm. These observations indicate low chromophore contents of the commercial U.S. grain sorghum hybrids. The U.S. grain forage sorghums are intentionally made low-tannin to maximize the digestion when fed to livestock (Price and Butler 1977). In Table 1, no clear trend was observable between the DPPH radical quenching (in μg trolox g −1 seed) and the resistance to aphid/webworm/bird (Ni et al. 2014). One-way ANOVA of DPPH results for the U.S. grain sorghum seeds in Table 1 indicated significant difference among hybrids (P = 8.98 × 10 −8 ). Multicomponent one-way ANOVA (Table S3) indicated significant (P ≤ 0.05) difference in DPPH values between the seed 4437 and eight other seeds (of 14 total hybrids), 4426 with five other seeds, and 4425 with four other seeds; seeds 4409, 4402, 4422, and 4433 were significantly different from seeds 4437, 4426, and 4425. For absorbance at 210 nm (Table 1), one-way ANOVA for triplicate methanol extracts indicated significant difference (P = 0.005) among 14 U.S. grain sorghum seeds; however, multicomponent one-way ANOVA did not show significant (P ≤ 0.05) difference, because of the large error in the absorbance unit. Two-way ANOVA on Table 1. Characteristics of U.S. grain sorghum hybrids (aphid, webworm, and bird resistance are given as fair (F), good (G), and very good (VG)), sweet sorghum, and African grain sorghum seeds: DPPH radical quenching (in μg trolox g −1 seed), acid-butanol assay (in μg delphinidin g −1 seed), reduction of 50   the trolox and A(210 nm) analyses showed a significant difference between seeds (P = 0.0005) as well as the analyses (trolox and A(210 nm), P = 0.0381), and indicated interactions between the seeds and analysis method (P = 0). In contrast to the U.S. grain sorghum, values were above detection limit for the following tannin-rich (based on the acid-butanol assay in Table 1) sweet sorghum and African grain sorghum seeds for all analytical methods: keller, M81E, seed 1 and seed 4. Unlike the U.S. grain sorghum, there were two λ max at 230 and 282 nm in these samples (seed 1, seed 4, M81E and Keller, see Fig.  S3 for the full spectra). In contrast, Dale and Theis showed low absorbance with one λ max near 210 nm, much like the U.S. grain sorghum (Table 1). Linear correlations were observed between trolox and delphinidin (radical quenching vs. proanthocyanidins; r 2 = 0.99 for sweet sorghum and r 2 = 0.81 for African grain sorghum, Fig. S2) as well as trolox and catechin (radical quenching vs. flavonoids; r 2 = 0.99 for 4 sweet sorghum varieties and r 2 = 0.97 for 4 African grain sorghum varieties, Fig. S2).

EEM-PARAFAC of acetone-water and methanol extracts
Figure 1A-C present 3 component EEM fingerprints obtained by PARAFAC analyses of methanol extracts. The total of 22 raw EEM spectra was normalized to the maximum intensity such that each sample will have equal impact on the PARAFAC model. This preprocessing procedure enabled us to focus on the variations among seeds, rather than the absolute magnitude of EEM intensity. Selected raw spectra of methanol extracts are provided in Figure S4. Component 2 (lowest Ex/Em wavelengths) of methanol extracts was the primary contributor to seed 2 and seed 3. A longer emission wavelength indicates more conjugated, aromatic, condensed, and higher MW structures (Lichtman and Conchello 2005). Component 1 had higher Ex/Em wavelengths than Component 2, and showed an opposite contribution trend of Component 2. For the U.S. grain sorghum, the contribution of Component 1 decreased from the left (4401) to right (4436), while the opposite trend was observed for Component 2 ( Figure 1D). Overall, Component 2 (lowest Ex/Em) was characterized by high contribution to seed 2, seed 3, and Dale. Component 1 behaved oppositely to Component 2: high contributions to seed 1 and seed 4, and decreasing contribution to U.S. grain sorghum from left (4401) to right (4436). Component 3 strongly contributed to seed 1, seed 4, Keller, and M81E.
For acetone-water extracts, Component 1 showed the lowest Ex/Em wavelengths ( Fig. 2A), and was the primary contributor to seed 2 and seed 3, and Dale (Fig. 2D), similarly to the Component 2 (lowest Ex/Em wavelengths) of methanol extracts. Oppositely, Component 3 has the highest Ex/Em wavelengths (Fig. 2C), and was the primary contributor to seeds 1 and 4 (Fig. 2D). In conclusion, low Ex/Em fingerprints (Component 2 of methanol and Component 1 of acetone-water) were the primary contributor to African (seed 2 and seed 3) and sweet (Dale) sorghum having low DPPH, acid-butanol, ferrozine, and vanillin responses in Table 1. Oppositely, high Ex/Em fingerprints (Component 1 of methanol and Component 3 of acetone-water) were the primary contributors to seed 1 and 4.
To visualize the relationships, linear correlation was obtained between DPPH radical quenching (Table 1 in in μg trolox g −1 seed) and % PARAFAC contributions (Figs. 1, 2). Component 3 of methanol extract (Fig. 1) correlated with DPPH radical quenching (r 2 = 0.91, Fig.  S5); linear correlation was observed for sweet sorghum seeds when M81E was removed (r 2 = 0.98). No correlation was observed for the U.S. grain sorghum. For acetone/ water, Component 2 correlated with trolox of seeds 1-4, and there was no correlation for sweet sorghum or U.S. grain sorghum. In conclusion, DPPH radical quenching correlated with flavonoids and proanthocyanidin contents of African grain and sweet sorghum seeds (Fig. S2); no