Isolation of BNI compound from DCM-wash
To collect hydrophobic compounds from the root surface of maize, “DCM-wash” was prepared based on a previous study of sorghum (Subbarao et al. 2013a, b). The total BNI activity of DCM-wash (220 mg) of maize roots was estimated at 7700 ATU based on the bioassay. To isolate the BNI compounds while reducing loss of sample, we applied a specific activity strategy (biological activity per unit weight of the compounds or fractions) to a bioassay-guided fractionation method. This method enabled us to focus on the target fraction by analysis of the bioassay result of each fraction.
Initially, we separated the DCM-wash concentrate (220 mg, 7700 ATU) by Sep-Pak C18 cartridge to obtain the most specific active BNI in 50% and 70% MeOH aqueous fractions (162 mg, 5000 ATU). Other fractions (H2O, MeOH, EtOH, and 50% DCM/MeOH) exhibited relatively much weaker inhibition (2800 ATU by four fractions). The 50–70% MeOH aqueous fraction was then separated by prep. HPLC into five fractions DCMw-1–DCMw-5 (Fig. 1).
Through purification of the most specific active fraction DCMw-3 by prep. HPLC, the target compound 1 (0.1 mg, 2233 ATU) was isolated (27.5 mg, 2300 ATU; Fig. 2).
Structural elucidation of compound 1
The molecular formula of compound 1 was elucidated as C12H10O4 with eight degrees of unsaturation by HR-ESI FT-ICR MS at m/z 219.0658 [M + H]+ (calculated 219.0652 for C12H11O4; Fig. S4a). The UV absorptions at 266 (π–π* transition), 293 (π–π* transition) nm, and the long tail band at 358 nm reaching far into the visible band for the pale yellow color of compound 1 suggested an oxygenated 1,4-naphthoquinone system, such as 2-methoxy-1,4-naphthoquinone (Little et al. 1948; Fig. S4b). The 1H-NMR spectrum of compound 1 showed six signals assignable to a dimethoxylated 1,4-naphthoquinone, two methoxy singlets at δH 3.92 and 3.87, a quinone proton at δH 6.08 (s), three aromatic protons at δH 8.00 (d, J = 8.6 Hz), 7.54 (d, J = 2.6 Hz), and 7.19 (dd, J = 8.6 and 2.6 Hz), to which their corresponding six 13C signals (δC 128.5, 120.8, 110.0, 109.8, 56.3, and 55.9) were attributed based on Heteronuclear Single Quantum Correlation (HSQC) correlations (Fig. 3, Fig. S5a, and Table 1). The remaining six carbon signals in the 13C NMR spectrum of compound 1 were assigned as two carbonyl carbons on 1,4-benzoquinone (δC 184.2 and 180.3), two methoxy groups attached carbons (δC 163.7 and 160.2), and two ring-condensation positional carbons (δC 132.9 and 125.4), respectively. The planar structure of compound 1 was elucidated as 2,7-dimethoxy-1,4-naphthoquinone by Heteronuclear Multiple Bond Correlation (HMBC) correlations of 2-OCH3/C-2; 7-OCH3/C-7; H-3/C-2, C-4, C-4a; H-5/C-4, C-7; H-6/C-4a; and H-8/C-1, C-8a (Fig. S5b and Fig. S5c). The 1H-NMR spectral data of compound 1 were in accordance with those of reported data for a synthetic compound; however, 13C-NMR spectral data and 2D-NMR data were not reported for that compound (Table 1; Guay and Brassard 1986). In the present study, we completely assigned the spectral data of compound 1 based on 2D-NMR correlations (HSQC and HMBC). Because this was the first isolation of naturally occurring 2,7-dimethoxy-1,4-naphthoquinone, we named this compound “zeanone” (Fig. 4a).
Zeanone (1) exhibited BNI activity with ED50 value of 2 μM and ED80 value of 8 μM (Fig. 4b), which was higher than those of earlier reported compounds, methyl linoleate (ED80 = 27 μM), sorgoleone (ED80 = 13 μM), and brachialactone (ED80 = 10.6 μM) (Subbarao et al., 2008, 2009, 2013a).
Isolation of BNI compound from DCM extract
Specific activity-based fractionation led to the isolation of zeanone (1) (0.1 mg, 2233 ATU) from DCM-wash (220 mg), but at low yield. Considering the remaining nearly 2700 ATU in the residue fractions DCMw-1, DCMw-2, DCMw-4, and DCMw-5, we inferred that high-content BNI compounds might exist within those fractions. With this hypothesis, we re-analyzed the HPLC data for DCM-wash (a stock sample before separation) and DCM extract. As a result, the dominant peak 2 was observed in both extracts (Fig. S6a and Fig. S6b). Furthermore, peak 2 had been fractionated in the fraction DCMw-2 (120 mg, 2300 ATU) from DCM-wash (Fig. 1); therefore, candidate compound 2 was expected to be a major BNI compound.
We then attempted to isolate compound 2 from fraction DCMw-2 (Fig. 1). However, the dominant peak 2 was undetected in DCMw-2, while a new peak 5 was strongly observed. Compound 5, suspected to be a degradation product of compound 2, was tentatively isolated by PTLC, and then established as a known benzoxazole MBOA (6-methoxy-2(3H)-benzoxazolone) by comparing the TLC and LC/MS results with those of commercially available MBOA (FUJIFILM Wako Pure Chemical Corp., Osaka, Japan) (Fig. S7 and Fig. S8). A previous report described that labile benzoxazinoids chemically decompose to give MBOA (5) and formic acid in vitro (Atkinson et al. 1991; Kosemura et al. 1994). Additionally, the UV absorption of compound 2 at 262 and 290 nm showed the characteristic pattern for benzoxazinoids (Fig. S9a). These data strongly suggested that compound 2 might be a degradable benzoxazinoid.
Next, we developed the isolation method for unstable compound 2 using a part of the DCM extract, which showed a similar HPLC chromatogram to that of DCM-wash before purification (Fig. S6). A part of the DCM extract (30 mg/395 mg) was fractionated using RP-HPLC to obtain a fraction containing pure compound 2. Then, the aqueous solution of 2 was directly extracted by EtOAc followed immediately by concentration of the organic layer to provide purified compound 2 (10 mg). Comprehensive analyses of NMR and MS spectra determined the structure of compound 2 as a known benzoxazinoid, 2-hydroxy-4,7-dimethoxy-2H-1,4-benzoxazin-3(4H)-one (HDMBOA) (Fig. 5a, Fig. S9b, and Table S1) (Maresh et al. 2006).
HDMBOA (2) was assayed to determine the BNI activity, which resulted in ED50 = 13 μM and ED80 = 70 μM (Fig. 5b). The natural benzoxazinoids, which are mainly distributed in Poaceae families including maize and wheat, showed a wide range of biological activities (allelopathy, regulating immune immunity, antimicrobial activity, etc.) (Ahmad et al. 2011; de Bruijn et al. 2018; Neal et al. 2012; Niemeyer 2009; Rice et al. 2012; Zhou et al. 2018). The biosynthetic pathway of benzoxazinoids has been extensively established and shares the same precursor indole-3-glycerol phosphate with primary metabolism in the biosynthesis of an essential amino acid tryptophan (Fig. S10) (Frey et al. 2009; Jonczyk et al. 2008; von Rad et al. 2001; Wright et al. 1992).
Following that, we confirmed the degradation of HDMBOA (2) into MBOA (5) in MeOH by HPLC analysis (Fig. S11). This result was consistent with a previous observation, in which compound 5 was the only aromatic compound detected in a MeOH solution of compound 2 after 12-h incubation (Escobar et al. 1997). Therefore, compound 2 in DCM-wash was confirmed to undergo degradation into compound 5 during the separation procedures by a Sep-Pak C18 column and prep. HPLC, followed by concentration.
The total weight of HDMBOA (2) accounted for 50 wt% of the original DCM-wash sample (110 mg, 2541 ATU) by quantification experiments (Fig. S6c). This result confirmed that compound 2 was a major constituent of maize roots in accordance with a previous report (Zhang et al. 2000).
Isolation of BNI compounds from MeOH extract
Next, we confirmed BNI activity of the MeOH extract and attempted to isolate responsible compounds based on the bioassay-guided fractionation method. The MeOH extract was partitioned between EtOAc and 30% MeOH aqueous solution by liquid–liquid distribution. The BNI active compounds were then concentrated in the EtOAc layer, so the EtOAc portion was further partitioned between n-hexane and 10% MeOH aqueous solution. Because the BNI activity was condensed into the 10% MeOH fraction (200 mg), this fraction was separated by prep. HPLC to yield two BNI compounds 3 (3.0 mg) and 4 (20 mg). Their structures were established as known benzoxazinoids, 2-hydroxy-7-methoxy-2H-1,4-benzoxazin-3(4H)-one (HMBOA, 3) (Fig. 6a, Fig. S12, and Table S2) and 2-hydroxy-4,7-dimethoxy-2H-1,4-benzoxazin-3(4H)-one-β-glucoside (HDMBOA-β-glucoside, 4) (Fig. 6c, Fig. S13, and Table S3), respectively. Both compounds were structural analogs of HDMBOA (2), namely, compound 3 carries an amine instead of the N-OMe group of 2 and compound 4 is the glucoside form of 2.
Compounds 3 and 4 showed BNI activities of ED50 = 91 μM, ED80 > 100 μM and ED50 = 94 μM, ED80 > 200 μM, respectively (Fig. 6b and d).
Quantification of BNI compounds in DCM-wash, DCM extract, and MeOH extract
We quantified four BNI compounds 1–4 in three extracts: DCM-wash, DCM extract, and MeOH extract (Fig. 7). The most specific BNI active zeanone (1) isolated from DCM-wash (220 mg) was also detected in DCM extract (395 mg) by RP-HPLC analysis, which was quantified as 0.05 mg (Fig. 7a, b, and d). We eventually isolated 0.05 mg of zeanone (1) from DCM extract by prep. HPLC. The amount of HDMBOA (2) in DCM-wash (220 mg) and DCM extract (395 mg) was quantified as 110 mg (50 wt%) and 132 mg (33 wt%), respectively. HMBOA (3) and HDMBOA-β-glucoside (4) identified in MeOH extract (10 g) were undetected in DCM-wash and DCM extract (Fig. 7c and d).