New acylated anthocyanin isolated from purple yam using molecular networking analysis

Abstract

Purple yam (Dioscorea alata L.) is a widely distributed tropical and subtropical food source, which is characterized by its color and viscosity. It contains stable acylated anthocyanins, which are used as naturally sourced colorants. We investigated the isolation of new acylated anthocyanins in purple yam using molecular networking analysis. Regardless of the degrees of acylation, all anthocyanins found in purple yam were clustered to the same network. The cluster of anthocyanins showed that purple yam contained other minor anthocyanins. We found a new minor acylated anthocyanin named alatanin H. Its structure was determined via spectroscopic analysis, including NMR and MS.

Introduction

Yams belong to the genus Dioscorea of the family Dioscoreaceae and are a staple food in various tropical countries. Approximately 600 species of yams have been identified globally (Falade et al., 2007). Yams are typically converted to flour via peeling, slicing, blanching, and sun-drying. Most cultivars of edible yam are white although some are purple. Purple yam (Dioscorea alata L.), which is known as “Ube” in the Philippines, is popularly processed into a powder and used to color confectionery products. We previously reported eight anthocyanins in purple yam, including alatanins A–G and cyanidin-3-gentiobioside acylated with a sinapoyl or feruloyl group (Fig. S1) (Moriya et al., 2015). Moreover, the HPLC analysis of a purple yam extract suggested the presence of other minor anthocyanins in addition to the isolatable eight compounds.

Acylated anthocyanins have also been isolated from other colored foods, such as purple sweet potato (Odake et al., 1992), purple potato (Fossen and Andersen, 2000), and red cabbage (Idaka et al., 1987). These compounds exhibit remarkable stabilities owing to their unique self-association and intramolecular sandwich-type stacking properties (Yoshida et al., 1991). Recently, the demand for naturally sourced food colorants has increased and is now higher than that for synthetic colorants.ⁱ⁾ To use acylated anthocyanins as naturally sourced food colorants, the diversity of acylated anthocyanins should be fully investigated. Moreover, the elucidation of the physicochemical properties of acylated anthocyanins will most likely lead to their efficacious use as natural colorants.

In this study, we further investigated the isolation of acylated anthocyanins in purple yam using molecular networking analysis. Molecular networking is an effective method for determining the metabolomes of microorganisms, plants, animals, and humans (Wang et al., 2016). This approach is based on the following hypothesis: compounds possessing chemical similarities will share similar fragmentation patterns and, thus, related neutral losses and/or fragments. In this method, compounds that belong to a metabolome are annotated according to their MS/MS features in clusters and their spectra are compared with those in spectral libraries to avoid the re-isolation of known compounds. Therefore, this method provides the chemical profiles of various samples easily and rapidly. In addition, this approach makes it possible to isolate the peaks thought to be new or bioactive compounds selectively (Tsugawa et al., 2021). We conducted the molecular networking analysis of purple yam, which revealed the presence of minor acylated anthocyanins. The new acylated anthocyanin was isolated using the molecular networking-guided method. Finally, its structure was determined.

Materials and Methods

General experimental procedures    DIAION HP-20 was purchased from Mitsubishi Chemical (Tokyo, Japan). Reversed-phase (RP)-HPLC separations were performed with a recycling system, which included a Jasco UV-970 detector (Tokyo, Japan), Jasco PU-2086 Plus Intelligent prep pump, and Capcell Pak ACR C18 column (5 µm, 10 × 250 mm) from Osaka Soda (Osaka, Japan), and Capcell Pak ACR C18 column (5 µm, 20 × 250 mm) from Osaka Soda. The analytical HPLC comprised MD-4017 photodiode array detector, a PU-4180 RHPLC pump, and AS-4050 HPLC autosampler from Jasco. Data were analyzed with ChromNAV software v.2 from Jasco. HPLC-grade solvents were used in all experiments. 1D and 2D NMR spectra were acquired on a Bruker BioSpin AVANCE-III (400 MHz) spectrometer (Rheinstetten, Germany). The chemical shifts were expressed in ppm. The NMR spectra were referenced to residual solvent (CD₃OD) peaks (¹H NMR 3.30 ppm and ¹³C NMR 49.0 ppm). HRESIMS spectra were acquired on a Q-Exactive HR-ESI-Orbitrap-MS from Thermo Fisher Scientific (Waltham, MA, USA).

Biological material    We used the same purple yam sample in a previous study (Moriya et al., 2015). Purple yam was purchased at a market in Manila, the Philippines in 2010. The sample was stored at −30 °C prior to extraction.

Data-dependent LC-ESI-HRMS/MS analysis    Data-dependent LC-ESI-HRMS/MS analysis was carried out by the previously reported method with a slight modification (Miyata et al., 2022). LC analyses were performed with an Ultimate 3000 liquid chromatographic system from Thermo Fisher Scientific. AQUITY UPLC BEH C18 column (1.7 µm, 2.1 × 100 mm) from Waters (Milford, MA, USA) was used. The mobile phase comprised H₂O–acetonitrile with 0.1% formic acid and the gradient was from 95:5 (0 min)–90:10 (15 min)–85:15 (25 min)–85:15 (35 min). The flow rate was 0.4 mL/min. The temperature of the column oven and injection volume was 40 °C and 1 µL, respectively.

HR-ESI-MS/MS spectra were acquired on a Q-Exactive HR-ESI-Orbitrap-MS from Thermo Fisher Scientific, which was equipped with an ESI dual source and operated in the positive-ion mode. The source parameters were set as follows: auxiliary gas flow rate, 10 units; sheath gas flow rate (N₂), 40 units; source voltage, 3.8 kV; S-lens RF level, 60; and capillary temperature, 320 °C. The data-dependent MS/MS events were acquired for the three most intense ions detected in full-scan MS (Top3 experiment). The stepped normalized collision energy was set at 15, 30, and 45 units. The MS/MS isolation window width was 1 Da. In the data-dependent MS/MS experiments, full and MS/MS scans were acquired at the resolutions of 70 000 and 17 500 fwhm, respectively. Both scans were performed with an automatically determined maximum injection time. Parent ions were placed in a dynamic exclusion list for 2.0 s after acquisition in a MS/MS scan.

Molecular networking analysis of purple yam    Molecular networking of purple yam was created using the previously reported parameters with a slight modification (Miyata et al., 2022). The MS data were converted from raw (Thermo) standard data format to abf data format using a Reifycs Abf Converter software.ⁱⁱ⁾ The abf data were analyzed using the MS-DIAL software package v.4.60 (Tsugawa et al., 2015). The parameters were adjusted as follows: retention time, 1−35 min; mass accuracy, MS1 and MS2 tolerances of 0.01 and 0.025 Da, respectively; maximum charge number, 2; mass range, 600−2 000 Da; mass slice width, 0.1 Da; minimum peak height, 3 × 10⁶ amplitude; sigma window value, 0.5; MS/MS abundance cut-off, 0 and adducts, [M + H]⁺. The analysis results were exported using a dedicated “Peak list export” function. The molecular networks were created using exported peak list data via the MetGem software package v.1.2.2 (Olivon et al. 2018). The parameters of MetGem were used as followed: minimum matched peaks, 4; m/z tolerance, 0.02 Da; minimal cosine score value, 0.65; maximum neighbor number, 10; and maximum connected component size, 1 000.

Extraction and isolation    Edible part of purple yam (4.0 kg) was extracted with 1% trifluoroacetic acid (TFA) in 50% MeOH (5 L) with stirring at room temperature for 24 h. The solids were then removed via filtration. The filtrates were concentrated at a reduced pressure to give a MeOH extract (134 g). An aliquot of the extract was separated using the DIAION HP-20 column chromatography (100 × 600 mm) with gradient elution using H₂O–MeOH containing 0.1% TFA (1:0, 12 L; 4:1, 6 L; 3:2, 10 L; 0:1, 4 L), yielding seven fractions (fr. 1, 1.9 g; fr. 2, 0.8 g; fr. 3, 0.8 g; fr. 4, 4.3 g; fr. 5, 1.5 g; fr. 6, 2.9 g; fr. 7, 1.9 g). After preparative RP-HPLC with H₂O–acetonitrile (87:13, 0.5% TFA) as eluent and detection at 280 nm, fraction 4 gave 9 (4.5 mg). HRMS/MS spectrum was acquired with all ion fragmentation mode using the same MS conditions described above. Conjugated sugars were determined using thin layer chromatography (TLC) analysis. Compound 9 was hydrolyzed with 2 M HCl at 100 °C for 2 h. The reaction mixtures were analyzed via TLC using CHCl3:MeOH:acetic acid:H₂O at 30:20:6:4 where galactose and glucose gave the retention factor (R_f) values of 0.30 and 0.34, respectively.

Alatanin H (9): purple amorphous solid; UV (0.01% HCl in MeOH) λ_max (log ε) 283 (3.99), 331 (3.91), and 520 nm (4.07); ¹H and ¹³C NMR (CD₃OD/CF₃COOD = 9:1), see Table 1; HRESIMS molecular ion (M⁺) peaks m/z 1141.3230 (calcd for C₅₀H₆₁O₃₀, 1141.3242), 979.2695 (calcd for C₄₄H₅₁O₂₅, 979.2714), 817.2166 (calcd for C₃₈H₄₁O₂, 817.2186), 611.1599 (calcd for C₂₇H₃₁O₁₆, 611.1607), 449.1073 (calcd for C₂₁H₂₁O₁₁, 449.1078), 287.0547 (calcd for C₁₅H₁₁O₆, 287.0550), and 207.0650 (calcd for C₁₁H₁₁O₄, 207.0652).

Table 1. Compounds annotated by LC-ESI-HRMS/MS (+) in the MeOH extracts of the purple yam sample.

	Rt (min)	Annotated compound	Chemical formula	m/z and ionization mode
1	20.71	alatanin D	C55H61O29	1185.3301 [M+]
2	18.77	alatanin A	C67H81O39	1347.3838 [M+]
3	19.40	alatanin B	C61H71O34	1509.4358 [M+]
4	16.85	alatanin G	C37H39O19	787.2074 [M+]
5	24.55	alatanin F	C39H43O20	831.2342 [M+]
6	14.46	alatanin E	C44H51O25	979.2717 [M+]
7	15.50	alatanin C	C38H41O20	817.2178 [M+]
8	1.96	cyanidin-3-gentiobioside	C27H31O16	611.1610 [M+]
9	10.42	alatanin H	C50H61O30	1141.3290 [M+]

Results and Discussion

Molecular networking analysis of purple yam    LC-ESI-HRMS/MS and molecular networking analysis were used to determine the chemical profiles of purple yam and revealed the presence of a cluster of anthocyanins (Figs. 1 and S3). The anthocyanins isolated previously from purple yam were annotated, as shown in Fig. 1 and Table 1. Regardless of the degrees of acylation, all anthocyanins found in purple yam were clustered to the same network. The cluster of anthocyanins showed that purple yam contained other minor anthocyanins. The detected peaks in the MeOH extracts of purple yam with the same m/z values were assigned to 5 and 6, which are structural isomers with different substituted positions of their sugar and acyl moieties. However, only one peak with the m/z value of 1141.3290 was found in the cluster. Because an acylated anthocyanin with this exact mass has not been reported, this peak was most likely from a new anthocyanin compound. Therefore, we isolated the compound in this peak from the MeOH extracts of purple yam. Its structure (9) was determined via spectroscopic analysis.

Fig. 1.

Cluster of anthocyanins in the created molecular network of purple yam.

Structure elucidation of new compound (9)    Compound 9 was obtained as a purple amorphous solid. Its determined molecular formula was C₅₀H₆₁O₃₀ based on HRESIMS, which gave a M⁺ ion peak with the m/z value of 1141.3230 (calcd for C₅₀H₆₁O₃₀, 1141.3242). Further MS/MS experiments showed a characteristic fragmentation pattern with the m/z values of 979.2695 (loss of a hexose moiety), 817.2166 (loss of two hexose moieties), 611.1599 (loss of a sinapoyl and two hexose moieties), 449.1073 (loss of a sinapoyl and three hexose moieties), and 287.0543 (loss of a sinapoyl and four hexose moieties). From the fragment ion peak with the m/z value of 287.0543, the aglycone moiety of 9 was thought to be cyanidin. The acid hydrolysis of 9 with 2 M HCl produced the free sugar moiety, which was identified as glucose using TLC (TLC R_f value of 0.34 in the hydrolyzed and authentic glucose samples). The ¹H NMR chemical shifts for 9 in CD₃OD/CF₃COOD (9:1) are listed in Table 2. The characteristic H-4 signal of anthocyanidin was observed at δ_H 8.69. The ¹H NMR spectrum also revealed the presence of H-2′, H-5′, and H-6′ protons in the B ring, which were located at the δ_H 8.28, 7.17, and 8.67, respectively. The aglycone was assigned to a cyanidin skeleton based on the HSQC and HMBC spectra. The remaining signals in the aromatic or olefinic proton region of the ¹H NMR spectrum at the δ_H 6.15, 6.23, and 7.37 indicated the presence of a sinapic acid. As described above, the MS/MS fragment ion peaks and integral values of the ¹H NMR spectrum of 9 suggested the presence of a sinapic acid unit. The geometric configuration of the sinapic group was determined as an E configuration using the coupling constant (J = 16 Hz) at the α/β positions. The signals that were assigned to the anomeric protons of four sugar moieties in 9 were observed at the δ_H 4.45, 5.08, 5.29, and 5.35. The large coupling constants in these anomeric protons (J = 7.3, 7.4, 7.5, and 7.9 Hz) indicated a β-glucosidic linkage in the four glucose units. The ¹H and ¹³C NMR signals of the three sugar units were successfully assigned using ¹H-1H COSY, HSQC, and HMBC experiments. The connectivity in different units of 9 was determined using HMBC analysis (Fig. 2). The presence of a cross peak from H-1 (δ_H 5.29 and 5.35) for the glucose moieties (G1 and G3) to C-3 (δ_C 145.9) and C-7 (δ_C 166.1) for the aglycone in the HMBC spectrum indicated the connection at the 3- and 7-positions of cyanidin to the glucosyl units G1 and G3, respectively. The position of glucose moiety G4 was from the HMBC correlation of H-1 (δ_H 5.08) to C-3′ (δ_C 146.7) of the aglycone. The diglucoside linkage was confirmed by the correlated signals belonging to H-6 (δ_H 3.91 and 4.04) of G1 to C-1 (δ_C 106.9) of G2, which indicated the connection between the C-3 position of the aglycone and gentiobioside moiety (6-O-β-d-glucopyranosyl-β-d-glucose). A correlation was also observed between the proton signals at the H-6 position (δ_H 4.08 and 5.34) of the glucose unit (G2) and carbonyl carbon signal (δ_C 168.2) of the sinapoyl group. Therefore, the structure of 9 was elucidated to be 3-O-(6-O-(6-O-(E)-sinapoyl-β-d-glucopyranosyl)-β-d-glucopyranosyl) -7-O-(β-d-glucopyranosyl)-3′-O-(β-d-glucopyranosyl) cyanidin, which is a new anthocyanin named alatanin H.

Table 2. ¹H (400 MHz) and ¹³C NMR (100 MHz) data for alatanin H (9) in CD₃OD/CF₃COOD (9:1).

	δC	type	δH (J in Hz)
Aglycone
2	162.9	C	—
3	145.9	C	—
4	133.7	CH	8.69 s
5	156.3	C	—
6	157.5	CH	6.83 d (1.4)
7	166.1	C	—
8	94.9	CH	7.17 d (1.4)
9	156.7	C	—
10	113.6	C	—
1′	120.6	C	—
2′	120.5	CH	8.28 d (2.2)
3′	146.7	C	—
4′	157.5	C	—
5′	118.5	CH	7.17 d (8.4)
6′	133.8	CH	8.67 dd (2.2, 8.4)
Glucose		Glucose (G1)
1	103.7	CH	5.29 d (7.9)
2	77.2	CH	3.69 t (7.9)
3	78.0	CH	3.83 m
4	72.2	CH	3.19 t (9.0)
5	77.7	CH	4.30 t (9.0)
6	73.1	CH2	3.91 m
			4.04 m
Glucose		Glucose (G2)
1	106.9	CH	4.45 d (7.4)
2	74.7	CH	3.48 m
3	77.9	CH	3.45 m
4	69.9	CH	3.68 m
5	74.4	CH	3.59 m
6	60.5	CH2	4.08 m
			5.34 m
Glucose		Glucose (G3)
1	100.7	CH	5.35 d (7.3)
2	74.8	CH	3.58 m
3	77.6	CH	3.59 m
4	74.4	CH	3.45 m
5	77.9	CH	3.80 m
6	62.5	CH2	3.5–4.0
			3.5–4.0
Glucose		Glucose (G4)
1	103.8	CH	5.08 d (7.5)
2	71.3	CH	3.41 m
3	77.5	CH	3.45 m
4	74.4	CH	3.60 m
5	78.0	CH	3.46 m
6	62.5	CH2	3.5–4.0
			3.5–4.0
Acyl	Sinapoyl
1	124.6	C	—
2	104.8	CH	6.23 s
3	147.3	C	—
4	138.4	C	—
5	147.3	C	—
6	104.8	CH	6.23 s
α	116.1	CH	6.15 d (16)
β	147.4	CH	7.37 d (16)
CO	168.2	C	—
OMe	56.0	CH3	3.43 s

Fig. 2.

Connectivity of the moieties in 9 (bold line) determined using ¹H-¹H COSY spectrum. The significant HMBC correlations (solid arrows) observed for 9 are shown.

The molecular networking analysis revealed the presence of minor anthocyanins in purple yam. The isolated 9 is a minor anthocyanin, and its the structure was determined using spectroscopic analysis. It is a new compound named alatanin H. The strategy used in this study may lead to the elucidation and discovery of other acylated anthocyanins.

Acknowledgements    We would like to thank Dr. Hiroko Nagai and Tito Genova Valiente (Ateneo de Manila University, the Philippines) for supplying purple yam. This research was supported by JSPS KAKENHI grant JP20J23632.

Conflict of interest    There are no conflicts of interest to declare.

Abbreviations

COSY

correlation spectroscopy

HMBC

heteronuclear multiple-bond correlation spectroscopy

HPLC

high performance liquid chromatography

HSQC

heteronuclear single-quantum correlation spectroscopy

NMR

nuclear magnetic resonance

LC-ESI-HRMS/MS

liquid chromatography-electrospray ionization-high resolution-tandem mass spectrometry

References

•  Falade,  K.O.,  Olurin,  T.O.,  Ike,  E.A., and  Aworh,  O.C. (2007). Effect of pretreatment and temperature on air-drying of Dioscorea alata and Dioscorea rotundata slices. J. Food Eng., 80, 1002-1010.
•  Fossen,  T. and  Andersen,  O.M. (2000). Anthocyanins from tubers and shoots of the purple potato, Solanum tuberosum. J. Hortic. Sci. Biotechnol., 75, 360-363.
•  Idaka,  E.,  Suzuki,  K.,  Yamakita,  H.,  Ogawa,  T.,  Kondo,  T., and  Goto,  T. (1987). Structure of monoacylated anthocyanins isolated from red cabbage, Brassica oleracea. Chem. Lett., 16, 145-148.
•  Miyata,  R.,  Hoshino,  S.,  Ahn,  M.-R., and  Kumazawa,  S. (2022). Chemical profiles of Korean bee pollens and their catechol-O-methyltransferase inhibitory activities. J. Agric. Food Chem., 70, 1174-1181.
•  Moriya,  C.,  Hosoya,  T.,  Agawa,  S.,  Sugiyama,  Y.,  Kozone,  I.,  Shin-Ya,  K.,  Terahara,  N., and  Kumazawa,  S. (2015). New acylated anthocyanins from purple yam and their antioxidant activity. Biosci. Biotechnol. Biochem., 79, 1484-1492.
•  Odake,  K.,  Terahara,  N.,  Saito,  N.,  Toki,  K., and  Toshio,  Honda. (1992). Chemical structures of two anthocyanins from purple sweet potato, Ipomoea batatas. Phytochemistry, 31, 2127-2130.
•  Olivon,  F.,  Elie,  N.,  Grelier,  G.,  Roussi,  F.,  Litaudon,  M., and  Touboul,  D. (2018). MetGem software for the generation of molecular networks based on the t-SNE algorithm. Anal. Chem., 90, 13900-13908.
•  Tsugawa,  H.,  Cajka,  T.,  Kind,  T.,  Ma,  Y.,  Higgins,  B.,  Ikeda,  K.,  Kanazawa,  M.,  VanderGheynst,  J.,  Fiehn,  O., and  Arita,  M. (2015). MS-DIAL: Data-independent MS/MS deconvolution for comprehensive metabolome analysis. Nat. Methods, 12, 523-526.
•  Tsugawa,  H.,  Rai,  A.,  Saito,  K., and  Nakabayashi,  R. (2021). Metabolomics and complementary techniques to investigate the plant phytochemical cosmos. Nat. Prod. Rep., 38, 1729-1759.
•  Wang,  M.,  Carver,  J.J.,  Phelan,  V.V.,  Sanchez,  L.M.,  Garg,  N.,  Peng,  Y.,  Nguyen,  D.D.,  Watrous,  J.,  Kapono,  C.A.,  Luzzatto-Knaan,  T.,  Porto,  C.,  Bouslimani,  A.,  Melnik,  A.V.,  Meehan,  M.J.,  Liu,  W.T.,  Crüsemann,  M.,  Boudreau,  P.D.,  Esquenazi,  E.,  Sandoval-Calderón,  M.,  Kersten,  R.D.,  Pace,  L.A.,  Quinn,  R.A.,  Duncan,  K.R.,  Hsu,  C.C.,  Floros,  D.J.,  Gavilan,  R.G.,  Kleigrewe,  K.,  Northen,  T.,  Dutton,  R.J.,  Parrot,  D.,  Carlson,  E.E.,  Aigle,  B.,  Michelsen,  C.F.,  Jelsbak,  L.,  Sohlenkamp,  C.,  Pevzner,  P.,  Edlund,  A.,  McLean,  J.,  Piel,  J.,  Murphy,  B.T.,  Gerwick,  L.,  Liaw,  C.C.,  Yang,  Y.L.,  Humpf,  H.U.,  Maansson,  M.,  Keyzers,  R.A.,  Sims,  A.C.,  Johnson,  A.R.,  Sidebottom,  A.M.,  Sedio,  B.E.,  Klitgaard,  A.,  Larson,  C.B.,  Boya,  C.A.P.,  Torres-Mendoza,  D.,  Gonzalez,  D.J.,  Silva,  D.B.,  Marques,  L.M.,  Demarque,  D.P.,  Pociute,  E.,  O'Neill,  E.C.,  Briand,  E.,  Helfrich,  E.J.N.,  Granatosky,  E.A.,  Glukhov,  E.,  Ryffel,  F.,  Houson,  H.,  Mohimani,  H.,  Kharbush,  J.J.,  Zeng,  Y.,  Vorholt,  J.A.,  Kurita,  K.L.,  Charusanti,  P.,  McPhail,  K.L.,  Nielsen,  K.F.,  Vuong,  L.,  Elfeki,  M.,  Traxler,  M.F.,  Engene,  N.,  Koyama,  N.,  Vining,  O.B.,  Baric,  R.,  Silva,  R.R.,  Mascuch,  S.J.,  Tomasi,  S.,  Jenkins,  S.,  Macherla,  V.,  Hoffman,  T.,  Agarwal,  V.,  Williams,  P.G.,  Dai,  J.,  Neupane,  R.,  Gurr,  J.,  Rodríguez,  A.M.C.,  Lamsa,  A.,  Zhang,  C.,  Dorrestein,  K.,  Duggan,  B.M.,  Almaliti,  J.,  Allard,  P.M.,  Phapale,  P.,  Nothias,  L.F.,  Alexandrov,  T.,  Litaudon,  M.,  Wolfender,  J.L.,  Kyle,  J.E.,  Metz,  T.O.,  Peryea,  T.,  Nguyen,  D.T.,  VanLeer,  D.,  Shinn,  P.,  Jadhav,  A.,  Müller,  R.,  Waters,  K.M.,  Shi,  W.,  Liu,  X.,  Zhang,  L.,  Knight,  R.,  Jensen,  P.R.,  Palsson,  B.,  Pogliano,  K.,  Linington,  R.G.,  Gutiérrez,  M.,  Lopes,  N.P.,  Gerwick,  W.H.,  Moore,  B.S.,  Dorrestein,  P.C., and  Bandeira,  N. (2016). Sharing and community curation of mass spectrometry data with Global Natural Products Social Molecular Networking. Nat. Biotechnol., 34, 828-837.
•  Yoshida,  K.,  Kondo,  T., and  Goto,  T. (1991). Unusually stable monoacylated anthocyanin from purple yam Dioscorea alata. Tetrahedron Lett., 32, 5579-5580.
• i) https://www.foodexecutive.com/en/ingredients/279-natural-colours-overtake-artificial-synthetic-colours-for-the-first-time (Jan. 31, 2022)
• i) https://www.reifycs.com/ AbfConverter/ (Jan. 31, 2022)