Metabolome analysis using multiple data mining approaches suggests luteolin biosynthesis in Physcomitrella patens

Abstract

The model land plant Physcomitrella patens synthesizes flavonoids which may act as protectant of ultraviolet-B radiation. We aimed to uncover its flavonoid profile, for which metabolome analysis using liquid chromatography coupled with Ion trap/Orbitrap mass spectrometry was performed. From the 80% methanol extracts, 661 valid peaks were detected. Prediction of the elemental compositions within a mass accuracy of 2 ppm indicated that 217 peaks had single elemental composition. A compound database search revealed 47 peaks to be annotated as secondary metabolites based on the compound database search. Comprehensive substituent search by ShiftedIonsFinder showed there were 13 peaks of potential flavonoid derivatives. Interestingly, a peak having m/z 287.0551, corresponding to that of luteolin, was detected, even though flavone synthase has never been identified in P. patens. Using P. patens labeled with stable isotopes (¹³C-, ¹⁵N-, ¹⁸O-, and ³⁴S), we confirmed the elemental composition of the peak as C₁₅H₁₀O₆. By a comparison of MS/MS spectra with that of authentic standard, the peak was identified as luteolin or related flavone isomers. This is the first report of luteolin or related flavones synthesis and the possibility of the existence of an unknown enzyme with flavone synthase activity in P. patens.

Bryophytes are the earliest land plants (embryophytes) that diverged from early water-dwelling algae in the Ordovician period, about 470 million years ago (Karol et al. 2001; Lewis and McCourt 2004; Ruhfel et al. 2014; Wodniok et al. 2011). The moss Physcomitrella patens is studied as a model plant because its evolutionary position is at the base of land plants and its genome has been sequenced (Lang et al. 2018; Rensing et al. 2008). For successful life on land, bryophytes such as P. patens had to develop ultraviolet (UV)-B radiation-protective mechanisms in the process of evolution because of the increased exposure to UV-B radiation (280–320 nm), which damages DNA, RNA, and proteins directly or indirectly via the production of free radicals (Foyer et al. 1994; Jin et al. 2000; Li et al. 1993; Ries et al. 2000). Plants synthesize a wide variety of metabolites, with the total number estimated at 1,060,000 (Afendi et al. 2012). Of these, organic compounds called secondary metabolites constitute the majority. These metabolites have evolved to acclimate organisms to changing environments. Flavonoids are the major class of plant secondary metabolites and are known to act as UV filters in plants (Stafford 1991; Tohge et al. 2016; Winkel-Shirley 2001). In response to UV-B radiation in P. patens, the accumulation of a quercetin derivative and an unknown metabolite has been reported (Wolf et al. 2010). However, flavonoid profiling of P. patens has been limited. Here, we performed metabolome analysis of P. patens using liquid chromatography coupled with Ion trap/Orbitrap mass spectrometry (LC-Orbitrap-MS) (Thermo Fisher Scientific Inc., Waltham, Massachusetts, USA).

Gametophores of P. patens ssp. patens: strain Gransden Wood were grown on solid (0.5% phytagel) half-strength Murashige–Skoog medium containing 0.5% glucose at 22°C under 18 h light/6 h dark long-day conditions using a sealed-glass bottle system (Kera et al. 2018) for 2 months (Figure 1). Gametophores were harvested in three biological replicates, frozen in liquid nitrogen, and lyophilized. After samples were ground by a pestle, metabolites were extracted with 80% methanol and injected in an LC-Orbitrap-MS [LC, Agilent 1100 system (Agilent, Santa Clara, California, USA); Orbitrap-MS, LTQ-Orbitrap (Thermo Fisher Scientific Inc.)]. The analysis conditions were as described previously (Kera et al. 2018). In brief, metabolites were separated by a C₁₈ reversed-phase column (TSKgel ODS-100 V; 4.6×250 mm, 5 µm; TOSOH, Syunan, Yamaguchi, Japan) under the following gradient: water +0.1% formic acid/acetonitrile +0.1% formic acid, 97/3 to 3/97% for 45 min, 3/97% for 5 min, and 3/97 to 97/3% for 7 min. The full mass scans were operated in electrospray ionization positive mode, and the MS/MS of the five most intense ions from each full mass scan were recorded by ion trap, covering a mass range from m/z 100 to 1500. Details are also available in the database Metabolonote (http://metabolonote.kazusa.or.jp/SE44/) (Ara et al. 2015).

Figure 1. The labeling system of Physcomitrella patens. Gametophores of P. patens were grown by the sealed-glass bottle system.

All raw data files were analyzed by MSGet, PowerFT, and PowerGet (Sakurai et al. 2014), and 661 valid ion peaks from the three biological replicates were extracted. After annotation by MFSearcher (Sakurai et al. 2013) within a mass accuracy of 2 ppm, the elemental compositions of 217 peaks were estimated as single-element and three peaks as multiple-element composition (Table 1). The 217 peaks were further annotated by searching in compound databases, such as KEGG (https://www.genome.jp/kegg/), KNApSAcK (http://www.knapsackfamily.com/KNApSAcK/), HMDB (https://hmdb.ca/), and LIPIDMAPS (https://www.lipidmaps.org/), and 86 and 47 peaks were annotated as the primary and secondary metabolites, respectively (Table 2). Furthermore, 11/47 peaks were classified as flavonoids (Table 2).

Table 1. Elemental composition within 2 ppm.

Status	Number (peaks)	Rate (%)
Single	217	32.8
Multiple	3	0.5
No hits	441	66.7
Total	661	100.0

Table 2. Category-wise results of database search.

Category		Number (peaks)	Rate (%)
Aminocarboxylic acids	Primary metabolite	32	23.5
Sugars	Primary metabolite	30	22.0
Nucleosides	Primary metabolite	2	1.4
Lipids	Primary metabolite	14	10.2
Organic acids	Primary metabolite	8	5.8
Flavonoids	Secondary metabolite	11	8.0
Alkaloids	Secondary metabolite	1	0.7
Iridoids	Secondary metabolite	2	1.4
Steroids	Secondary metabolite	3	2.2
Phenolics	Secondary metabolite	19	13.9
Terpenoids	Secondary metabolite	3	2.2
Others	Secondary metabolite	8	5.8

Flavonoid derivatives were further searched by ShiftedIonsFinder, which predicts the modification variation of several flavonoid aglycons regardless of compound database (Supplementary data S1) (Kera et al. 2014). Xylosylation (Xyl)-(C₅H₈O₄, m/z 132.04226), glucosylation (Glc)-(C₆H₁₀O₅, m/z 162.05282), rhamnosylation (Rha)-(C₆H₁₀O₄, m/z 146.05791), glucuronidation (GlcUA)-(C₆H₈O₆, m/z 176.03209), cinnamoylation (cinnamoyl)-(C₉H₆O₁, m/z 130.04186), coumaroylation (coumaroyl)-(C₉H₆O₂, m/z 146.03678), caffeoylation (caffeoyl)-(C₉H₆O₃, m/z 162.03169), feruloylation (feruloyl)-(C₁₀H₈O₃, m/z 176.04734), and malonylation (malonyl)-(C₃H₂O₃, m/z 89.0003939) were selected as modification groups, and the parameters were the same as previously described (Kera et al. 2018). As a result, 3 additional peaks (total 14 peaks) were selected as flavonoid derivatives. For effective estimation of elemental composition, ¹³C-, ¹⁵N-, ¹⁸O-, and ³⁴S-labeled samples were prepared using a sealed bottle system as described previously (Kera et al. 2018), and data curation was performed manually, which confirmed the elemental compositions of 13/14 peaks as single (Table 3). Although flavones have never been identified from P. patens, peak 2 (m/z 287.0551, C₁₅H₁₀O₆, 25.4 min) and peak 9 (m/z 287.0551, C₁₅H₁₀O₆, 29.2 min), corresponding to luteolin, were detected (Figure 2A). Since the elution time of authentic luteolin was 29.2 min (Figure 2A) and the MS/MS fragment pattern of peak 9 was the same as that of authentic luteolin (Figure 2B), peak 9 was identified as luteolin. Moreover, according to a rare incorporation of an ¹⁸O-atom into oxygen atoms of A- and C-ring of apigenin in Medicago truncatula (Kera et al. 2018), peak 9 contained two ¹⁸O-atoms in the structure, suggesting the labeling of 3′- and 4′-hydroxy groups of luteolin (Figure 2C). Flavone biosynthesis is catalyzed by flavone synthase (FNS), which is further divided into 2-oxoglutarate-dependent dioxygenase-type FNSI and cytochrome P450-dependent monooxygenase-type FNSII (Martens and Mithofer 2005). Since the corresponding genes have not been found in the P. patens genome, P. patens is considered unable to produce flavones. Apigenin glycosides and luteolin glycosides have been found in the mosses Bryum algens (Webby et al. 1996) and Leptostomum macrocarpon (Brinkmeier et al. 1998). However, this is the first report of flavone synthesis in P. patens and the first indication of the presence of FNS with no amino acid similarity with FNSI or FNSII (Koduri et al. 2010). Eight biflavonoids (dimers of flavonoid moieties linked by a C–C or C–O–C bond) were annotated (Table 3). The elemental composition of peaks 1, 3, 10, and 11 (C₃₀H₁₈O₁₂); peak 4 (C₃₀H₂₂O₁₄); and peaks 6, 7, and 12 (C₃₀H₂₀O₁₂) corresponded to those of the dimers of C₁₅H₁₀O₆, C₁₅H₁₂O₇, and C₁₅H₁₁O₆, respectively. Furthermore, peaks 5 and 8 (C₃₀H₂₀O₁₃) could be the biflavonoid reported in Hypnum cupressiforme (Sievers et al. 1992). Since biflavonoids accumulated in the moss Ceratodon purpureus have antioxidant and UV protection functions (Waterman et al. 2017), P. patens may also accumulate some biflavonoids with UV-protective properties.

Table 3. List of predicted flavonoids.

No	Retention time (min)	Detected m/z	Ionization	Elemental composition	Annotation
1	24.7	571.0872	[M+H]+	C30H18O12	Biflavonoid (C15H10O6 × 2)
2	25.4	287.0551	[M+H]+	C15H10O6	Flavonoid
3	26.9	571.0871	[M+H]+	C30H18O12	Biflavonoid (C15H10O6 × 2)
4	27.1	607.1082	[M+H]+	C30H22O14	Biflavonoid (C15H12O7 × 2)
5	27.1	589.0977	[M+H]+	C30H20O13	No DB hits
6	27.2	573.1027	[M+H]+	C30H20O12	Biflavonoid (C15H11O6 × 2)
7	27.7	573.1027	[M+H]+	C30H20O12	Biflavonoid (C15H11O6 × 2)
8	28.9	589.0977	[M+H]+	C30H20O13	No DB hits
9	29.2	287.0551	[M+H]+	C15H10O6	Luteolin
10	30.1	571.0870	[M+H]+	C30H18O12	Biflavonoid (C15H10O6 × 2)
11	30.6	571.0871	[M+H]+	C30H18O12	Biflavonoid (C15H10O6 × 2)
12	30.8	573.1026	[M+H]+	C30H20O12	Biflavonoid (C15H11O6 × 2)
13	36.7	489.2273	[M+H]+	C30H32O6	Flavonoid

Figure 2. LC-MS analysis of Physcomitrella patens extraction. A. Ion chromatograms by Orbitrap mass spectrometry. Top, The full mass scan of an unlabeled sample; middle, chromatogram at m/z 287.0551±10 ppm from an unlabeled sample; bottom, chromatogram at m/z 287.0551±10 ppm from luteolin. B. MS/MS analysis of m/z 287 from the unlabeled sample (above) and luteolin (bottom) by ion trap mass spectrometry. C. The full mass scan of unlabeled (above) and ¹⁸O-labeled samples (bottom) by Orbitrap mass spectrometry.

Abbreviations

FNS

flavone synthase

LC-MS

liquid chromatography-mass spectrometry

MS

mass spectrometry

UV

ultraviolet

Conflict of interest

The authors declare no conflict of interest.

Data deposition

The raw data sets were deposited into the MassBase metabolome database (http://webs2.kazusa.or.jp/massbase/) as accession numbers MDLC1_36678–36686, 40772–40779, 40782–40785 and 40787–40790, and are available for free downloading.

Research funding

This study was supported by the Kazusa DNA Research Institute Foundation and by the Hirata Corporation Foundation, the Ministry of Education, Science, Sports and Culture; Grants-in-Aid for Scientific Research on Innovative Areas, no. 20200062, 2008–2010 (to H.S), no. 23108528, 2011–2012 (to H.S), and no. 25108727 and 2013–2014 (to H.S); Grants-in-Aid for Scientific Research (C) no. 23510272, 2011–2013 (to H.S), no. 26350967, and 2014–2016 (to H.S); and Metabolomics US-Japan Joint Program for a Low Carbon Society funded projects, 2012–2014 (to H.S).