Abstract

1. Introduction

The Osmundaceae family comprises several living genera. This family is classified as a basal group of leptosporangiate ferns, which have undergone little morphological and anatomical change since the Mesozoic time, and, therefore, some authors consider them as a “primitive” species [1,2,3]. Osmunda regalis L. (Figure 1), also known as royal fern, is a species with widespread occurrence in the temperate zones of the Northern and Southern hemispheres, widely distributed throughout Europe, Southern Africa, America, and New Zealand. It is deciduous or evergreen in warmer regions, a terrestrial fern, typically growing to a height of 1 to 2 m. The frond is bi-pinnate, meaning that each leaf is twice divided into pinnae and smaller pinnules. The fern is often found in moist woods, swamps, lakes, or stream banks [4]. In some countries such as Spain, Italy, and Poland, O. regalis is a threatened plant species with an ecological impact in grazing for wild animals [5,6].

Open in a separate window

Figure 1

Fern leaves of Osmunda regalis L. Photo: Heidi Lie Andersen.

The royal fern has been widely used for its medicinal properties in traditional medicine for several centuries. Hieronymus Brunschwig, a German surgeon, alchemist, and botanist, wrote about the plant’s potential use in treating cancer, fistulas, and fractures in his book Kleines Destillierbuch [7,8]. Moreover, in some regions of Northern Spain, such as Galicia, Asturias, and Cantabria, the plant has been traditionally used to treat bone fractures and alleviate muscular pain [9]. Even though O. regalis has been utilised as a medicinal plant for more than five centuries, only scarce information about its chemical constituents is available in the current literature. Only a few compounds have been identified, including sitosterol, which has been found in different parts of the plant [10,11], as well as the steroid Ecdysterone (20-hydroxyecdysone) [12]. Several relatively nonpolar compounds such as saturated fatty acids [13,14], alkanediols, ketoaldehydes, methyl palmitate, and other fatty acid esters [15,16] have been reported from this species. Anthocyanins based on the aglycones pelargonidin and cyanidin have also been detected, where the cyanidin derivative is indicated to be an acylated anthocyanin [17]. Further phenolic compounds, such as tannins and gallotannins [18], as well as a multitude of common phenolic acids, including caffeic, p-coumaric, p-hydroxybenzoic, cinnamic, and vanillic acids, have previously been identified from this species [19,20]. Only a few specific compounds of this species, including osmundalactone [21] and osmundalin, where the latter compound is the glucoside of osmundalactone [15], have hitherto been reported in the current literature.

This paper describes the characterisation of 17 natural products from the aerial parts of Osmunda regalis. Fifteen of these compounds, including six undescribed compounds, are identified in this species for the first time. As part of our ongoing research to identify new lead compounds for future anticancer drugs, the cytotoxicity of the isolated compounds towards acute monocyte leukaemia (MOLM13), rat cardiomyoblasts (H9c2), and kidney epithelial cells (NRK) were examined. The theoretical framework for this study is a fundament of the work presented in the doctoral thesis of Carpinteyro Diaz [22]

2. Results and Discussion

A methanolic extract of 2.1 kg of leaves from Osmunda regalis was concentrated under reduced pressure and subjected to a liquid–liquid extraction (LLE) with petroleum ether followed by ethyl acetate. The components of the aqueous and ethyl acetate extracts were further separated by gradient XAD-7 adsorption chromatography, Sephadex LH-20 gel filtration chromatography, and preparative HPLC.

The eleven known compounds chalconaringenin 2’-O-β-glucopyranoside (1), vanillic acid (4), p-hydroxy-benzoic acid (5), p-hydroxy-benzoic acid methyl ester (7), dihydrodehydrodiconiferyl alcohol 4-O-α-rhamnopyranoside (8), apigenin 7-(2’’-O-α-rhamopyranosyl-β-glucopyranoside) (9), epoxyconiferyl alcohol (10), 5-hydroxy-2-hexen-4-olide (11), blumenol C 9-O-β-glucopyranoside (14), n-hexyl-O-β-glucopyranoside (16), and 2-hexenoic acid (17) (Figure 2 and Tables S1–S9) were identified from the fern leaves of Osmunda regalis L. The identifications were based on a combination of several 1D and 2D NMR spectroscopic techniques. These compounds were detected in this plant species for the first time, except for vanillic (4) and p-hydroxy-benzoic acids (5).

Open in a separate window

Figure 2

Molecular structures of compounds 1–17 characterised from O. regalis (L).

The UV spectrum of compound 2 recorded online during HPLC analysis exhibited λ_max values at 346 and 266 nm (Figure S63), which matches a flavonol derivative [23]. The aromatic region of the 1D ¹H NMR spectrum of 2 (Figure S1) showed a 4H AA’XX’ system at δ 8.05 (‘d’ 8.9 Hz, H-2’,6’) and δ 6.93 (‘d’ 8.9 Hz, H-3’,5’), which is consistent with a p-substituted B-ring, and a 2H AX system at δ 6.42 (d 2.0 Hz, H-8) and δ 6.18 (d 2.0 Hz, H-6), which is consistent with a kaempferol derivative. The sugar regions of the 1D ¹H and 1D ¹³C CAPT spectra of 2 showed the presence of three sugar units. All ¹H and ¹³C resonances of these glycosyl substituents were wholly assigned by the 1D ¹H selective TOCSY spectra of each glycosyl substituent, in addition to the 2D ¹H-¹H COSY, the 2D ¹H-¹³C HSQC, the 2D ¹H-¹³C HSQC-TOCSY, and the 2D ¹H-¹³C H2BC spectra of 2. The sugar units were identified to be two β-glucopyranose units and an α-rhamnopyranose unit, respectively, by the observed ¹H coupling constants in the 1D ¹H and the 1D ¹H selective TOCSY spectra, in addition to the 17 ¹³C chemical shift values belonging to these sugar units observed in the 1D ¹³C CAPT spectrum of 2. The anomeric coupling constants revealed the β-configurations of the anomeric carbons of the glucosyl substituents and the α-configuration of the anomeric carbon of the rhamnosyl substituent (Table 1 and Table 2).

Assignments of the ¹³C resonances belonging to the aglycone and the inter-residual connections were determined by the 2D ¹H-¹³C HMBC experiment. The cross-peak at δ 5.71/132.9 (H-1’’/C-3) confirmed the linkage between the glucopyranosyl unit and the aglycone at the 3-hydroxyl. The downfield chemical shift of C-2’’ (δ 77.8) of this glucosyl unit indicated the presence of a sugar substituent at this position. The cross-peaks at δ 4.97/77.8 (H-1’’’/C-2’’) and δ 3.70/100.2 (H-2’’/C-1’’’) confirmed the linkage between the inner glucosyl substituent and the terminal glucosyl substituent to be at the 2’’-position. Cross-peaks at δ 5.05/77.0 (H-1’’’’/C-2’’’) and δ 3.23/100.3 (H-2’’’/C-1’’’’) proved the linkage between the terminal glucosyl unit and the rhamnosyl unit to be at the 2’’’-position. Thus, compound 2 was identified to be the previously undescribed compound kaempferol 3-O-(2’’-O-(2’’’-α-rhamnopyranosyl)-β-glucopyranosyl)-β-glucopyranoside. A sodium-added molecular ion [M+Na]⁺ at m/z 779.1999 corresponding to C₃₃H₄₀O₂₀Na⁺ (calculated: m/z 779.2005; Δ = −0.73 ppm) observed in the HRMS of compound 2 confirmed this identification (Figure 2 and Figure S53).

Compound 3 exhibited a UV λ_max absorption at 352 and 256 nm (Figure S64), which resembles a flavonol derivative [23]. The aromatic region of the 1D ¹H NMR spectrum of 3 (Figure S13) showed a 3H ABX system at δ 7.49 (d 2.3 Hz, H-2’,6’) and δ 6.89 (d 8.5 Hz, H-3’,5’), which is consistent with an m- and p-substituted B-ring, and a 2H AX system at δ 6.38 (d 2.1 Hz, H-8) and δ 6.17 (d 2.1 Hz, H-6), which accords with a quercetin aglycone. The 1D ¹H and 1D ¹³C CAPT spectrum of 3 indicated the existence of three sugar units in the sugar regions. The ¹H and ¹³C resonances of these glycosyl substituents were entirely determined by the 1D ¹H selective TOCSY spectra of each glycosyl substituent, in addition to the 2D ¹H-¹H COSY, the 2D ¹H-¹³C HSQC, the 2D ¹H-¹³C HSQC-TOCSY, and the 2D ¹H-¹³C H2BC spectra of 3. The sugar units were identified to be two β-glucopyranose units and an α-rhamnopyranose unit, respectively, by the observed ¹H coupling constants in the 1D ¹H and the 1D ¹H selective TOCSY spectra, in addition to the 17 ¹³C chemical shift values belonging to these sugar units observed in the 1D ¹³C CAPT spectrum of 3. The anomeric coupling constants revealed the β-configurations of the anomeric carbons of the glucosyl substituents and the α-configuration of the anomeric carbon of the rhamnosyl substituent (Table 1 and Table 2).

Assignments of the ¹³C resonances belonging to the aglycone and the inter-residual connections were determined by the 2D ¹H-¹³C HMBC experiment. The cross-peak at δ 5.70/133.2 (H-1’’/C-3) confirmed the linkage between the glucopyranosyl unit and the aglycone at the 3-hydroxyl. The downfield chemical shift of C-2’’ (δ 77.20) of this glucosyl unit indicated the presence of a sugar substituent at this position. Cross-peaks at δ 4.94/77.20 (H-1’’’/C-2’’) and δ 3.73/100.4 (H-2’’/C-1’’’) confirmed the linkage between the inner glucosyl substituent and the terminal glucosyl substituent to be at the 2’’-position. The cross-peaks at δ 5.06/76.9 (H-1’’’’/C-2’’’) and δ 3.23/100.3 (H-2’’’/C-1’’’’) ratified the linkage between the terminal glucosyl unit and the rhamnosyl unit to be at the 2’’’-position. Thus, compound 3 was identified to be the previously undescribed compound quercetin 3-O-(2’’-O-(2’’’-α-rhamnopyranosyl)-β-glucopyranosyl)-β-glucopyranoside. A sodium-added molecular ion [M+Na]⁺ at m/z 795.1951 corresponding to C₃₃H₄₀O₂₁Na⁺ (calculated: m/z 795.1954; Δ = −0.37 ppm) observed in the HRMS of compound 3 confirmed this identification (Figure 2 and Figure S54).

The UV spectrum of compound 6 recorded online during HPLC analysis exhibited UV-absorption maxima at 330 and 268 nm, with notably strong absorption at 330 nm, which matches a flavonol derivative acylated with a cinnamic acid (Figure S65) [23]. The aromatic region of the 1D ¹H NMR spectrum of 6 (Figure S23) showed a 4H AA’XX’ system at δ 7.98 (‘d’ 8.8 Hz, H-2’,6’) and δ 6.83 (‘d’ 8.9 Hz, H-3’,5’) in accord with a p-substituted B-ring and a 2H AX system at δ 6.37 (d 2.1 Hz, H-8) and δ 6.16 (d 2.1 Hz, H-6), which accords with a kaempferol derivative. The sugar regions of the 1D and 2D NMR spectra of 6 showed the presence of three sugar units (Figures S23–S28). All ¹H and ¹³C resonances of these glycosyl substituents were assigned by the 1D ¹H spectrum of 6, in addition to the combined information from the 2D ¹H-¹H COSY, the 2D ¹H-¹³C HMBC, the 2D ¹H-¹³C HSQC, the 2D ¹H-¹³C HSQC-TOCSY, and the 2D ¹H-¹³C H2BC spectra of 6. The sugar units were identified to be two β-glucopyranose units and an α-rhamnopyranose unit, respectively, by the observed ¹H coupling constants in the 1D ¹H NMR spectrum, in addition to the 17 ¹³C chemical shift values belonging to these sugar units observed in the 2D ¹H-¹³C HSQC and H2BC spectra of 6. The anomeric coupling constants revealed β-configurations of the anomeric carbons of the glucosyl substituents and α-configuration of the anomeric carbon of the rhamnosyl substituent (Table 1 and Table 2).

Assignments of the ¹³C resonances belonging to the aglycone and the inter-residual connections were determined by the 2D ¹H-¹³C HMBC experiment. At δ 5.71/133.2 (H-1’’/C-3), this cross-peak confirmed the linkage between the glucopyranosyl unit and the aglycone at the 3-hydroxyl. The downfield chemical shift of C-2’’ (δ 78.2) of this glucosyl unit indicated the presence of a sugar substituent at this position. Cross-peaks at δ 5.02/78.2 (H-1’’’/C-2’’) and δ 3.66/100.4 (H-2’’/C-1’’’) confirmed the linkage between the inner glucosyl substituent and the terminal glucosyl substituent to be at the 2’’-position. The cross-peaks at δ 5.05/77.1 (H-1’’’’/C-2’’’) and δ 3.69/100.4 (H-2’’’/C-1’’’’) established the linkage between the terminal glucosyl unit and the rhamnosyl unit to be at the 2’’’-position. The acyl moiety was identified as E-caffeic acid by the 3H AMX system cross-peaks at δ 6.95 d 2.2 Hz (H-2’’’’’), δ 6.84 dd 8.2; 2.2 Hz (H-6’’’’’), and δ 6.66 d 8.2 Hz (H-5’’’’’) and an 2H AX system at δ 7.36 (H-7’’’’’) and δ 6.16 (H-8’’’’’). A coupling constant of 15.8 Hz between H-7’’’’’ and H-8’’’’’ showed the E-configuration (Table 1 and Table 2). The downfield chemical shift of C-6’’’ (δ 63.4) of the glucose unit indicates the linkage between this sugar unit and the caffeoyl group. The cross-peaks at δ 4.30/166.5 (H-6A’’’/C-9’’’’’) and δ 4.19/166.5 (H-6B’’’/C-9’’’’’) confirmed the linkage between the terminal glucosyl substituent and the caffeoyl unit to be at the 6’’’-position. Thus, compound 6 was identified to be the previously undescribed compound kaempferol 3-O-(2’’-O-(2’’’-O-α-rhamnopyranosyl-6’’’-O-(E)-caffeoyl-)-β-glucopyranosyl)-β-glucopyranoside (Figure 2). A sodium-added molecular ion [M+Na]⁺ at m/z 941.2318 corresponding to C₄₂H₄₆O₂₃Na⁺ (calculated: m/z 941.2322; Δ = −0.46 ppm) observed in the HRMS of compound 6 (Figure S55) confirmed this identification.

The aromatic region of the 1D ¹H NMR spectrum of 8 showed the presence of a 3H ABX system at δ 6.99 d 2.1 Hz (H-2), 7.04 d 8.3 Hz (H-5), and 6.85 dd 8.3; 2.1 Hz (H-6) and a 2H AB system at δ 6.68 d 1.8 Hz (H-2’) and δ 6.67 dd 1.8, 0.8 Hz (H-6’), which is consistent with dihydrodehydrodiconiferyl alcohol, where the aromatic rings are connected through a C3 unit showing a 4H spin system at δ 5.46 d 6.5 Hz (H-7), δ 3.42 dddd 7.4, 6.5, 5.3, and 0.8 Hz (H-8), δ 3.71 dd 10.8, 5.3 Hz (H-9A), and δ 3.60 dd 10.8, 7.4 Hz (H-9B). A further 4H spin system of a C3 substituent belonging to dihydrodehydrodiconiferyl was observed at δ 2.52 (H-9’), δ 1.68 (H-8’), and δ 3.40 d 6.5 Hz (H-7’). Two 3H signals belonging to methoxy groups were observed at δ 3.73 (3-OCH₃) and 3.76 (3’-OCH₃). The 20 ¹³C resonances belonging to dihydrodehydrodiconiferyl alcohol aglycone observed in the 1D ¹³C CAPT spectrum of 8 were assigned by the 2D ¹H-¹³C HMBC spectrum and the 2D ¹H-¹³C HSQC spectrum of 8. The glycosyl unit of 8 was identified as rhamnose by the six ¹H signals observed in the 1D ¹H spectrum belonging to this unit and the corresponding six ¹³C signals belonging to this unit observed in the ¹³C CAPT spectrum of 8, which were assigned by the 2D ¹H-¹³C HSQC spectrum, the 2D ¹H-¹³C H2BC spectrum, and the 2D ¹H-¹H COSY spectrum of 8. A cross-peak at δ 5.23/144.9 (H-1’’/C-4) observed in the 2D ¹H-¹³C HMBC spectrum of 8 confirmed that the rhamnopyranosyl unit was attached to dihydrodehydrodiconiferyl alcohol aglycone at the 4-position. Thus, compound 8 was identified as dihydrodehydrodiconiferyl alcohol 4-O-α-rhamnopyranoside, which NMR data can be seen in Table 3. The stereochemistry of this compound was established by circular dichroism (CD) spectroscopy. The CD spectrum of 8 (Figure S59) showed a negative Cotton effect at [θ]₂₇₉, indicating that the configurations of C-7 and C-8 are R and S, respectively [24,25]. Hence, compound 8 was identified as (7R,8S)-dihydrodehydrodiconiferyl alcohol 4-O-α-rhamnopyranoside (Figure 2). Dihydrodehydrodiconiferyl alcohol 4-O-α-rhamnopyranoside has previously been reported from Pedicularis torta [26]; however, only incomplete NMR data exist for this compound in the current literature.

The UV spectrum of compound 12, which exhibited λ_max at 274 and 212 nm (Figure S66), combined with its short HPLC retention time, may be indicative of a polar lactone [27,28], a compound type frequently reported from ferns [29]. The 1D ¹H NMR spectrum (Figure S29) of 12 showed an 8H spin system at δ 4.22 dd 4.2, 1.3 Hz (H-4), δ 4.07 dt 6.8, 4.3 Hz (H-3), δ 3.74 dd 6.5, 4.2 Hz (H-5), δ 2.80 dd 18.2, 6.8 Hz (H-2A), δ 2.39 dd 18.2, 1.3 Hz (H-2B), and δ 1.12 d 6.5 Hz (H-6), in addition to a methoxy group at δ 3.23 (3-OCH₃). The 2D ¹H-¹³C HMBC spectrum (Figure S31) assigned the seven ¹³C signals observed in the 1D ¹³C CAPT spectrum 12, the 2D ¹H-¹³C HSQC spectrum, and the 2D ¹H-¹³C H2BC spectrum (Figure S30) 12. The 7H spin system observed in the 1D ¹H NMR spectrum of 12 were assigned by the 2D ¹H-¹³C HSQC spectrum, the 2D ¹H-¹³C HSQC-TOCSY spectrum, the 2D ¹H-¹³C H2BC spectrum, and the 2D ¹H-¹H COSY spectrum of 12 (Figures S32–S36). The cross-peaks at δ 2.80/76.0 (H-2A/C-3), δ 2.39/76.0 (H-2B/C-3), δ 4.07/35.0 (H-3/C-2), δ 4.07/88.1 (H-3/C-4), δ 4.22/76.0 (H-4/C-3), δ 4.22/65.6 (H-4/C-5), δ 3.74/88.1 (H-5/C-4), δ 3.74/19.0 (H-5/C-6), and δ 1.12/65.6 (H-6/C-5) observed in the 2D ¹H-¹³C H2BC spectrum of 12 were particularly useful for the assignments of these signals. The cross-peaks established the position of the lactone carbonyl carbon C-1 at δ 2.80/175.9 (H-2A/C-1), δ 2.39/175.9 (H-2B/C-1), δ 4.07/175.9 (H-3/C1), and δ 4.22/175.9 (H-4/C-1) observed in the 2D ¹H-¹³C HMBC spectrum of 12. A cross-peak at δ 3.23/76.0 (3-OCH₃/C-3) observed in the 2D ¹H-¹³C HMBC spectrum of 12 confirmed that the methoxy group was attached to the 3-position. Thus, 12 was identified as 3-methoxy-5-hydroxy-4-olide. The stereochemistry of 12 was established by circular dichroism (CD) spectroscopy. The CD spectrum of 12 (Figure S60) showed a positive Cotton effect at [θ]₂₅₀, which indicates R and S configurations of C-4 and C-5, respectively, whereas the positive Cotton effect at [θ]₂₀₈ indicated an S configuration of C-3 [15,21,30]. Thus, 12 was identified as the previously undescribed natural product (3S,5S,4R)-methoxy-hydroxy-olide (Figure 2 and Table 4). A molecular ion [M+H]⁺ at m/z 161.0809 corresponding to C₇H₁₂O₄H⁺ (calculated: m/z 161.0808; Δ = −0.1 ppm) observed in the HRMS (Figure S56) of compound 12 confirmed this identification (Figure 2).

The UV spectrum of 13 (Figure S67) showed a λ_max at 296 nm, 266 nm, 220 nm, and 198 nm that corresponds to a saturated γ,δ-dilactone [27]. The 1D ¹H NMR spectrum (Figure S37) of compound 13 showed the presence of two CH₃ groups at δ 1.29 d 6.3 Hz (5-CH₃) and δ 1.13 d 6.6 Hz (H-2’’), one CH₂ group at δ 2.82 dd 15.2, 11.5 Hz (H-2A) and δ 2.66 dd 15.2, 6.6 Hz (H-2B), in addition to seven CH groups at δ 2.91 dddd 11.5, 8.0, 6.6, and 1.4 Hz (H-3), δ 4.04 dd 8.7, 8.0 Hz (H-4), δ 4.39 dd 8.7, 6.3 Hz (H-5), δ 3.21 dd 5.3, 1.4 Hz (H-2’), δ 4.82 d 5.3 Hz (H-3’), δ 4.28 d 5.3 Hz (H-4’), and δ 3.78 dd 6.6, 3.7 Hz (H-1’’). The corresponding ¹³C signals were observed at δ 18.19 (5-CH₃), δ 19.02 (C-2’’), δ 32.73 (C-2), δ 38.23 (C-3), δ 80.22 (C-4), δ 73.19 (C-5), δ 51.87 (C-2’), δ 78.56 (C-3’), δ 86.97 (C-4’), and δ 65.51 (C-1’’) in the 1D ¹³CAPT NMR spectrum of compound 13 (Figure S38), in addition to two ester carbonyls at δ 171.25 (C-1) and δ 176.85 (C-1’), which describe the presence of two different lactone rings attached through a C-C bond (Figure 2). The 2D ¹H-¹³C HSQC-TOCSY spectrum of compound 13 revealed that the 15 hydrogens observed in the 1D ¹H NMR spectrum all belonged to the same spin system. These twelve ¹³C signals observed in the 1D ¹³C CAPT spectrum of 13 were assigned by the 2D ¹H-¹³C HMBC spectrum, the 2D ¹H-¹³C HSQC spectrum, and the 2D ¹H-¹³C H2BC spectrum of 13. The ¹H spin systems observed in the 1D ¹H NMR spectrum of 13 were assigned by the 2D ¹H-¹³C HSQC spectrum, the 2D ¹H-¹³C HSQC-TOCSY spectrum, the 2D ¹H-¹³C H2BC spectrum, and the 2D ¹H-¹H COSY spectrum (Figures S40–S42). The cross-peaks at δ 2.82/38.2 (H-2A/C-3), δ 2.66/38.2 (H-2B/C-3), δ 2.91/32.7 (H-3/C-2), δ 2.91/80.2 (H-3/C-4), δ 4.04/73.2 (H-4/C-5), δ 4.04/38.2 (H-4/C-3), δ 4.39/80.2 (H-5/C-4), δ 4.39/18.2 (H-5/C-6), and δ 1.29/73.2 (H-6/C-5) observed in the 2D ¹H-¹³C H2BC spectrum of 13 belonging to the δ-lactone ring and the cross-peaks at δ 3.21/78.6 (H-2’/C-3’), δ 4.82/51.9 (H-3’/C-2’), δ 4.28/65.5 (H-4’/C-1’’), δ 3.78/87.0 (H-1’’/C-4’), δ 3.78/19.0 (H-1’’/C-2’’), and δ 1.13/65.5 (H-2’’/C-1’’) observed in the 2D ¹H-¹³C H2BC spectrum of 13 belonging to the γ-lactone ring were particularly useful for complete assignment of the non-quaternary carbons of 13. The ester carbonyls of the γ-lactone ring and the δ-lactone ring were assigned by the observed cross-peaks in the 2D ¹H-¹³C HMBC spectrum (Figure S39) of 13. The cross-peaks at δ 4.04/51.9 (H-4/C-2’), δ 2.82/51.9 (H-2A/C-2’), δ 2.66/51.9 (H-2B/C-2’), δ 2.91/51.9 (H-3/C-2’), δ 2.91/176.9 (H-3/C-1’), δ 2.91/78.6 (H-3/C-3’), δ 3.21/32.7 (H-2’/C-2), δ 3.21/38.2 (H-2’/C-3), and δ 3.21/80.2 (H-2’/C-4) observed in the 2D ¹H-¹³C HMBC spectrum of 13 confirmed the linkage between the γ-lactone ring and the δ-lactone ring from C-3 to C-2’. This was further confirmed by the observation of the cross-peak at δ 2.91/3.21 (H-3/H-2’) observed in the 2D ¹H-¹H COSY spectrum of 13 (Figure S43). The cross-peaks at δ 1.13/87.0 (H-2’’/C-4’), δ 4.28/19.02 (H-4’/C-2’’), and δ 3.78/87.0 (H-1’’/C-4’) confirmed that the 1’’-hydroxyethyl substituent was attached to C-4’. Thus, compound 13 was identified as the previously undescribed bilactone 4-hydroxy-3-(3’-hydroxy-4’-(hydroxyethyl)-oxotetrafuranone-5-methyl tetrahydropyranone. Completely assigned chemical shift values of this previously undescribed bilactone are shown in Table 5. The CD spectrum of compound 13 (Figure S61) shows a negative Cotton effect ([θ]₂₂₀) and a positive Cotton effect ([θ]₂₅₅) in accordance with (4R,5S,3’S,1’’S,4’R) configuration [15,21,30,31]. Thus, compound 13 was identified as (4R,5S,3’S,1’’S,4’R)-4-hydroxy-3-(3’-hydroxy-4’-(1-hydroxyethyl)-oxotetrahydrofuran-5-methyltetrahydropyranone (Figure 2). A molecular ion [M+H]⁺ at m/z 275.1124 corresponding to C₁₂H₁₈O₇H⁺ (calculated: m/z 275.1125; Δ = 0.5 ppm) observed in the HRMS (Figure S57) of compound 13 confirmed this identification.

Compound 15 showed a UV spectrum of a saturated δ-lactone with λ_max at 296, 266, and 220 nm (Figure S68) [27]. The 1D ¹H NMR spectrum (Figure S45) of 15 showed a 9H spin system at δ 5.30 ddd 6.6, 3.4, 1.5 Hz (H-4), δ 6.87 dd 9.9, 3.4 Hz (H-3), δ 4.60 p 6.6 Hz (H-5), δ 6.11 dd 9.9, 1.5 Hz (H-2), and δ 1.30 d 6.6 Hz (H-6), acylated with a 5’-hydroxy-oxohexanoyl group displaying signals at δ 2.54 (H-2’), 2.86 (H-3’), δ 4.04 q 6.9 Hz (H-5’), and δ 1.16 d 6.9 Hz (H-6’), respectively. The 12 ¹³C signals observed in the 1D ¹³C CAPT spectrum 15 (Figure S46) were assigned by the 2D ¹H-¹³C HMBC spectrum, the 2D ¹H-¹³C HSQC spectrum, and the 2D ¹H-¹³C H2BC spectrum 15 (Fig. 47-50). The 9H spin system observed in the 1D ¹H NMR spectrum of 15 were assigned by the 2D ¹H-¹³C HSQC spectrum, the 2D ¹H-¹³C HSQC-TOCSY spectrum, the 2D ¹H-¹³C H2BC spectrum, and the 2D ¹H-¹H COSY spectrum (Figure S51) of 15. The cross-peaks at δ 6.11/67.3 (H-2/C-4), δ 6.87/122.3 (H-3/C-2), δ 6.11/143.8 (H-2/C-3), δ 5.30/76.0 (H-4/C-5), δ 4.60/67.3 (H-5/C-4), δ 4.60/17.9 (H-5/C-6), and δ 1.30/76.0 (H-6/C-5). The cross-peaks of the 5’-hydroxy-oxohexanoyl at δ 2.54/32.1 (H-2’/C-3’), δ 2.86/27.3 (H-3’/C-2’), δ 4.04/19.7 (H-5’/C-6’), and δ 4.04/19.7 (H-5’/C-6’) observed in the 2D ¹H-¹³C H2BC spectrum of 15 (Figure S50) were particularly useful for the assignments of these signals. The cross-peaks established the position of the lactone carbonyl carbon C-1 at δ 6.11/161.9 (H-2/C-1), δ 6.87/161.9 (H-3/C-1), and δ 4.60/161.9 (H-5/C-1) observed in the 2D ¹H-¹³C HMBC spectrum of 15. A cross-peak at δ 2.54/171.7 (H-2’/C-1’), δ 2.86/171.7 (H-3’/C-1’), and δ 5.30/171.7 (H-4/C-1’) observed in the 2D ¹H-¹³C HMBC spectrum of 15 confirmed that the hydroxy oxohexanoyl group is attached in the 4-position. Thus, 15 was identified as 4-O-(5-hydroxy-4-oxohexanoyl) osmundalactone. The complete chemical shift values of compound 15 are presented in Table 6 (Figures S45–S52). The CD spectrum of compound 15 (Figure S62) shows a negative band at ([θ]₂₁₈) and a positive band at ([θ]₂₆₀) that corresponds to a 4R, 5S configuration, respectively [15,21]. A further negative band was observed at ([θ]₂₉₅), which could be ascribed to the R-configuration of C-5’ [32]. Thus, compound 15 was identified to be the previously undescribed natural product 4R-O-(5S-hydroxy-4-(5’R-hydroxy4-oxohexanoyl) osmundalactone. The sodium-added molecular ion [M+Na]⁺ at m/z 279.0841 corresponding to C₁₂H₁₆O₆Na⁺ (calculated: m/z 279.0839; Δ = 0.63 ppm) observed in the HRMS (Figure S58) of compound 15 confirmed this identification. Compound 15 represents the first complete assignment of NMR signals of the osmundalactone core structure. The ¹H NMR chemical shift values are in good agreement with the previously published incompletely assigned ¹H NMR data of osmundalactone [21], with the exception of the chemical shift value of H-4, which is experiencing a considerable downfield shift because of the effect of acylation of the 4-hydroxyl with 5’R-hydroxy-4-oxohexanoyl [33].

The previously reported beneficial properties of Osmunda regalis as a traditional medicinal plant encouraged studies to determine the potential cytotoxic activity against leukaemia cells of natural products isolated from this plant source. The cytotoxicity of compounds 2, 3, 6, 8, 10, 11, 12, 13, 14, and 15, which include the six previously undescribed natural products, were tested towards the acute myeloid leukaemia cell line MOLM13 and compared with a DMSO control which was the solvent used to dilute the pure compounds. Only compounds 11 and 15, which are γ- and δ-lactones, respectively, exhibited significant cytotoxic activity against the MOLM13 cells, with EC₅₀ values of 46.2 ± 0.07 µM (11) and 111.9 ± 0.07 µM (15) after 72 hours of incubation. These two compounds were further tested towards normal cell lines of rat kidney epithelial (NRK) cells and rat cardiomyoblasts H9c2 cells to reveal if the cytotoxic activity was selective towards cancer cells. However, the cytotoxic activity of 5-hydroxy-2-hexen-4-olide lactone (11) did not differ between the cell lines tested, and the EC₅₀ values were in all instances similar: 46.2 μM for MOML13, 48.7 μM for H9c2, and 49.6 μM for NRK (Table 7). The 4-O-(5-hydroxy-4-oxohexanoyl) osmundalactone (15) was only cytotoxic towards the heart and AML cell lines, exhibiting similar EC₅₀ values for these cell lines (102.7 and 111.9 μM, respectively). We did not detect reduced metabolic activity in the NRK cells at the highest concentration tested (200 μM), indicating lower cytotoxicity of compound 15 towards this cell line. Monolactones such as γ- and δ-lactones have previously been reported to induce DNA damage in the blood and the central nervous systems of rats [34] and to exhibit pro-apoptotic activity against various cancer cells through inhibition of NF-κB [35,36]. The fact that γ-lactones have been reported to be more cytotoxic against cancer cells than δ-lactones [37] may explain the observed difference in cytotoxic potency between compound 15 and 11.

3. Materials and Methods

4. Conclusions

The identification of 17 natural products from the aerial parts of Osmunda regalis L., a “primitive” fern species, demonstrates that this plant is a valuable source of previously unknown natural products. During the identification process, the complete structure of three previously unidentified flavonoids; kaempferol 3-O-(2’’-O-(2’’’-α-rhamnopyranosyl)-β-glucopyranosyl)-β-glucopyranoside, quercetin 3-O-(2’’-O-(2’’’-α-rhamnopyranosyl)-β-glucopyranosyl)-β-glucopyranoside, and kaempferol 3-O-(2’’-O-(2’’’-α-rhamnopyranosyl-6’’’-O-(E)-caffeoyl-)-β-glucopyranosyl)-β-glucopyranoside; and three previously undescribed lactones; 3-methoxy-5-hydroxy-4-olide, 4-hydroxy-3-(3’-hydroxy-4’-(hydroxyethyl)-oxotetrafuranone-5-methyl tetrahydropyranone, and the 4-O-(5-hydroxy-4-oxohexanoyl) osmundalactone; were determined. The cytotoxic activity of the mono lactones and bilactones in this work confirms that γ-lactones are more cytotoxic against cancer cells than δ-lactones and even bilactones that did not present cytotoxicity against these cell lines correlate well with what could be expected for a unique species that represents one of the last intact survivors of a Jurassic ecosystem.

Acknowledgments

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/molecules29174247/s1.

Funding Statement

The study was partly supported financially by the Norwegian Society for Children’s Cancer (Grant nos. 180007 and 190004) for financial support. The Research Council of Norway partly supported this work through the Norwegian NMR Platform, NNP (226244/F50).

Author Contributions

Conceptualization, H.L.A., A.E.C.D. and T.F.; methodology, B.H., C.B., A.V., A.E.C.D. and T.F.; investigation, A.E.C.D. and T.F.; resources, L.H. and T.F.; data curation, L.H., A.E.C.D. and T.F.; writing—original draft preparation, B.H., C.B., H.L.A., L.H., A.E.C.D. and T.F.; writing—review and editing, L.H., A.E.C.D. and T.F.; visualization, A.E.C.D.; supervision, L.H. and T.F.; project administration, T.F.; funding acquisition, L.H. and T.F. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

Conflicts of Interest

The authors declare no conflicts of interest.

Footnotes

References