Synthesis of natural product hybrids by the Ugi reaction in complex media containing plant extracts | Scientific Reports – Nature.com

Designing a chemical engineering scheme based on the racemic Ugi reaction

Chemical hybridization of plant products was realized by chemical engineering of plant extracts using the racemic Ugi four-component reaction (Ugi-4CR), wherein four different types of functional group components, namely, an aldehyde or ketone, a primary amine, a carboxylic acid, and an isocyanide, assemble into an α-acylamino amide (Fig. 1b)^17,18. An isocyanide possessing both nucleophilic and electrophilic properties serves as a coupling reagent for the plant products. Typically, a one-pot procedure with a stoichiometric mixture of substrate components yields an Ugi adduct quantitatively, with water as the sole by-product¹⁹. The high step- and atom-economy of the Ugi-4CR would be suitable for executing the chemical engineering of natural product extracts without causing unnecessary complication to the reaction mixture^20,21,22,23. Additionally, the natural abundance of aldehydes, ketones, and carboxylic acids in plants^4,24,25 motivated us to employ plant extracts as substrates of the Ugi-4CR.

First, using model compounds, the substrate scope of the Ugi-4CR was briefly investigated (Supplementary Fig. S1). The reaction time was fixed for 7 days for all entries. Importantly, the reaction temperature was set at room temperature and four substrates were mixed stoichiometrically, in order to avoid undesired side reactions and/or decompositions of constituents in extracts. In summary, this mild reaction conditions exhibited broad scope and afforded Ugi adducts in moderate to high yields (11 examples, 45–93% yields), and the Ugi-4CR was thus found to be suitable for chemical engineering of plant extracts. While some of the substrates required only a couple of days to complete the reaction, other substrates exhibited slow conversions (Supplementary Fig. S1, entries 1a, 1b in TFE, 1f, and 1h). Thus, the reaction time for the following Ugi reaction using plant extracts was set to 7 days.

Synthesis of an Ugi adduct derived from an extract of Zanthoxylum piperitum

Under the optimized reaction conditions, the first chemical engineering of a plant extract was performed using the methanol extract of the commercially available pericarp of Zanthoxylum piperitum (ZP1) as the carbonyl component of the Ugi-4CR (Fig. 2). The content of the carbonyl compounds in the ZP1 extract was roughly estimated to be 0.156 ± 0.002 mg menthone per mg extract by colorimetric quantification using 2,4-dinitrophenylhydrazonate.²⁶ The ZP1 extract (1.32 g) was mixed with excess benzylamine, acetic acid, and p-toluenesulfonylmethyl isocyanide (TosMIC) in methanol. After stirring for 7 days, the crude mixture was repeatedly separated and purified by silica gel column chromatography and preparative thin layer chromatography (TLC) to isolate the novel Ugi adduct 2 (5.28 mg), which was found to derive from citronellal (3) in ZP1²⁷. The structure of 2 was well-characterized by a combination of ¹H NMR, ¹H–¹H COSY, ¹³C NMR, DEPT, HMQC, HMBC, IR, and MS analyses. During the separation and purification stages, the characteristic ¹H NMR signals derived from the tosyl and benzyl groups in 2 and the UV absorbance were effectively used to identify the product. The ease of handling of this process is in sharp contrast to the tediousness of isolating 3 from the ZP1 extract and the subsequent chemical manipulation processes because of the low content of 3 in the ZP1 extract and the absence of UV absorption at 254 nm.

Figure 2

Synthesis of 2 by the Ugi-4CR of the methanol extract of Z. piperitum (ZP1) in the presence of benzylamine, acetic acid, and TosMIC. The inset shows the carbonyl content of the ZP1 extract.

Simultaneous synthesis of Ugi adducts derived from extracts of the genus Curcuma

Then, chemical engineering of plant extracts was performed using the commercially available dried rhizome powder of Curcuma zedoaria from China (CZ1) as the carbonyl component of the Ugi-4CR. The carbonyl content of the methanol extract of CZ1 was estimated to be 0.162 ± 0.003 mg menthone per mg extract (Fig. 3a), and a stoichiometric mixture of the CZ1 extract, benzylamine, chloroacetic acid, and cyclohexyl isocyanide was subjected to the Ugi-4CR conditions. Based on the results presented in Supplementary Fig. S1, the most reactive acid, chloroacetic acid, was used as the acid component. The resulting engineered mixture and the untreated CZ1 extract were then roughly divided into six fractions by silica gel column chromatography, and their chemical composition was analyzed by TLC (Supplementary Fig. S2). The result showed that most spots (constituents) in the CZ1 extract remained unchanged under the engineering conditions, and clearly highlighted a disappeared spot (substrate) as well as newly appeared spots (products to be isolated). Repeated separation and purification of the engineered mixture resulted in the successful identification of two novel Ugi adducts 4 and 5, both of which were found to derive from curcumenone (6) in CZ1 (Fig. 3b)²⁸. The acetyl component in 5 was assumed to derive from CZ1. Both products had a newly constructed tetra-substituted stereocenter and existed as a 1:1 mixture of diastereomers. The conversion efficiency in this engineering procedure was roughly estimated to be 58% based on the content of 6 in the starting CZ1 extract, which was determined through tedious separation and purification processes (Table 1, entry 1). This conversion efficiency was quite satisfactory considering that the present reaction was performed in the presence of many unidentified chemicals, indicating the high chemoselectivity of the Ugi-4CR-based engineering process. The same process was also applied to the methanol extract of the dried rhizome powder of C. zedoaria from Japan (CZ2) to provide 4 in a slightly reduced yield (28%, entry 2, Supplementary Fig. S3). By contrast, no Ugi adduct was obtained from the reaction with C. phaeocaulis (CP1) because this extract did not contain curcumenone or other reactive aldehydes or ketones (entry 3). As expected, the chemical composition of the engineered CP1 mixture apparently remained unchanged under the present Ugi-4CR conditions (Supplementary Fig. S4). Reaction of the methanol extract of C. longa (CL1) containing curcumenone also afforded the Ugi adduct 4 with a conversion efficiency of 49% (entry 4, Supplementary Fig. S5), in good agreement with the engineering results for CZ1 (entry 1). The similar conversion efficiencies, regardless of plant origin, demonstrated the reproducibility of the present engineering method (Supplementary Fig. S6). The extracts containing curcumenone, CZ1, CZ2, and CL1, had different chemical composition, as shown by the corresponding TLC analyses (Fig. 3c,d); however, all engineering processes similarly provided the Ugi adducts, demonstrating the robustness of the proposed method. Of note, the isolated curucumenone (6) was found to be unstable and gradually decomposed. Thus, the present engineering method enabled the direct chemical modification of plant products that would otherwise be inaccessible.

Figure 3

(a) Carbonyl content of the extracts. (b) Structures of Ugi adducts 4 and 5, and curcumenone (6). (c) TLC of fraction No. 2 of the untreated extracts. The TLC was developed with hexane/ethyl acetate (4:1) and then visualized by UV light at 254 nm (i) or by staining with an acidic solution of p-anisaldehyde (ii). (d) TLC of fractions No. 2–4 of the engineered mixtures. The TLC was developed with 1,2-dichloroethane/ethyl acetate (19:6) and then visualized by UV light at 254 nm (i) or by staining with an acidic solution of p-anisaldehyde (ii). The dashed lines in (c) and (d) indicate the solvent front.

Table 1 Ugi-4CR using the extracts of the genus Curcuma.

Fluorescence-guided isolation strategy of an Ugi adduct derived from a plant extract

Instead of using benzylamine, a similar engineering of the CZ1 extract was performed in the presence of the fluorescent 1-aminopyrene to provide the expected Ugi adduct 7, where the benzyl group of 6 was replaced with the 1-pyrenyl motif (Fig. 4a). Upon excitation at 342 nm, product 7 exhibited fluorescence, where the maximum emission wavelength (({uplambda }_{max}^{em})) in DMSO was significantly blue-shifted from that of 1-aminopyrene (({uplambda }_{max}^{em}) = 441 nm) (Fig. 4b)^29,30. This blue-shift of ({uplambda }_{max}^{em}) associated with 1-pyrenyl amide formation enabled facile identification of the Ugi adduct 7 from the crude reaction mixture by irradiation with a handy UV lamp (365 nm), although the fluorescence intensity of 7 was lower than that of 1-aminopyrene. This engineering example showed that the fluorescence-guided isolation strategy³¹ can facilitate identification of plant products that may have otherwise been missed, and that the simple replacement of amine and isocyanide would provide rapid access to a series of analogues.

Figure 4

(a) Structure of 7. (b) Fluorescence spectra of 1-aminopyrene and 7 (1 μM) in DMSO with 342 nm excitation.

Chemical composition of castor oil fatty acids

Another engineering was achieved using castor oil fatty acids (CO-FA), prepared from the seed oil of Ricinus communis, as the carboxylic acid component of the Ugi-4CR. Compared to other plant oils, castor oil possesses a high content of ricinoleic acid and is thus highly soluble in alcohols. These unique properties of CO-FA prompted us to use it in the Ugi-4CR-based engineering process. The acid content of CO-FA was deduced from the given neutralization number (183.6 mg KOH g⁻¹), and a stoichiometric mixture (0.5 mmol) of CO-FA, (±)-citronellal, benzylamine, and cyclohexyl isocyanide was subjected to the Ugi-4CR conditions. The engineering proceeded smoothly and the subsequent two steps of separation and purification of the engineered mixture by silica gel column chromatography and preparative TLC revealed four different types of Ugi adducts 8a–8d, which were found to be derived from (R)-ricinoleic acid, dimeric (R)-ricinoleic acid³², oleic acid, and linoleic acid in CO-FA, respectively (Table 2)³³. The mixture also gave Ugi adduct 9, which was likely to be derived from (R)-ricinoleic acid and benzaldehyde in CO-FA. Taken together, CO-FA engineering using the Ugi-4CR afforded five novel Ugi adducts in a single synthetic operation. All the Ugi adducts were easily isolated based on their UV-absorbing properties and structurally well-characterized by a set of NMR analyses. The production ratio (wt%) of Ugi adducts 8a, 9, 8c, and 8d approximately corresponded to the reported content of (R)-ricinoleic acid, oleic acid, and linoleic acid in CO-FA, respectively (Table 2)³³. These results indicated that the present engineering method could be used to determine the composition of acids in CO-FA samples.

Table 2 Ugi adducts 8a–8d and 9 derived from CO-FA.

Synthesis of plant product hybrids

Our goal of synthesizing plant product hybrids was finally achieved by the Ugi-4CR using a mixture of the methanol extract of C. zedoaria (CZ1) and castor oil fatty acids (CO-FA) as substrates in the presence of benzylamine and cyclohexyl isocyanide (Fig. 5). A stoichiometric mixture of the four substrate components in methanol was stirred at room temperature for 7 days to give the engineered mixture. Separation and purification of the engineered mixture identified the novel Ugi adduct 10 (53.8 mg, 57%) as a 1:1 mixture of diastereomers, which was assigned to be a hybrid-type molecule of curcumenone (6) in CZ1 and ricinoleic acid in CO-FA. ¹³C NMR analysis of 10 confirmed the hybridization of curucumenone in CZ1 and ricinoleic acid in CO-FA together with incorporation of benzylamine and cyclohexyl isocyanide (Fig. 6): the signal of the carbonyl ketone of curcumenone (210 ppm) disappeared, and correspondingly, a peak for the newly constructed tetra-substituted carbon of 10 was observed at 65 ppm. In addition, both the acid and isonitrile carbons were transformed into amide carbonyl carbons with peaks at approximately 175 ppm. The signals of the other structural motifs of four substrate components were completely retained in the hybrid-type product 10. Additionally, further separation and purification of the newly appeared spots (compounds) led to the isolation of two plant product-like Ugi adducts 5 (1.26 mg, 2.0%) and 9 (12.3 mg), and three hybridized Ugi adducts 11 (2.97 mg, 2.3%), 12 and 13 (0.73 mg, 0.8%) (Fig. 7). Notably, TLC analysis showed that most components in the CZ1 extract remained unchanged under the present engineering conditions (Fig. 7a,b, and Supplementary Fig. S7). High chemoselectivity ensured facile detection and purification of the products, even from complex mixtures. By contrast, no product was detected by simply stirring a mixture of CZ1 and CO-FA in methanol at room temperature for 7 days (Supplementary Figs. S8d,d′, S9d,d′). This control experiment supported that the isolated products herein did not originate from the extracts themselves, but were indisputably synthesized by the present Ugi-4CR-based engineering process. The conversion efficiency based on the curcumenone content in the CZ1 extract was determined to be 62%, which was comparable to the conversion efficiency of the CZ1 extract (58%, Table 1). This incredibly high conversion efficiency demonstrated the robustness of the proposed engineering strategy based on the Ugi-4CR. Further detailed analysis coupled with high-resolution liquid chromatography of the engineered extracts may identify unexplored products, including those derived from minor components of the plant extracts. To the best of our knowledge, this is the first report on the simultaneous synthesis of natural product hybrids by chemical engineering of natural product extracts.

Figure 5

Synthetic of 10 by the Ugi-4CR using a mixture of CZ1 and CO-FA in the presence of benzylamine and cyclohexyl isocyanide. The inserts show the carbonyl content of the CZ1 extract and the neutralization number of CO-FA.

Figure 6

¹³C NMR spectra (151 MHz) of (a) curcumenone, (b) benzylamine, (c) ricinoleic acid, (d) cyclohexyl isocyanide, and (e) hybridized Ugi adduct 10 in CDCl₃. Units of chemical shift (δ) are ppm relative to residual chloroform (77.16 ppm) as an internal standard.

Figure 7

(a) TLC of the fractionated starting extract of C. zedoaria (CZ1). (b) TLC of the fractionated engineered mixture. Both TLCs were developed with CHCl₃/ethyl acetate (19:6) and then visualized by staining with an acidic solution of p-anisaldehyde. The upper dashed line indicates the solvent front. (c) Structures of the hybridized Ugi adducts 11–13.

Protease inhibition of Ugi adducts derived from plant products

The prepared Ugi adducts have a diamide motif and were therefore subjected to a protease inhibition assay. Protease inhibitors have received increasing attention for their therapeutic potential against respiratory virus infections including severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)^34,35. The inhibitory activity of the Ugi adducts was determined in the presence of 6 nM α-chymotrypsin and 100 μM succinyl-Ala-Ala-Pro-Phe-p-nitroanilide³⁶. Among those tested, the Ugi adducts 7, 8a, 8b, 9, and 10 exhibited more than 50% inhibitory activity at a concentration of 10 μM, and the corresponding IC₅₀s were determined to be 3.1–13.4 μM (Table 3), which were considerably higher (weaker) than that of the reference inhibitor chymostatin A–C (18.5 ± 1.1 nM). By contrast, all the starting materials, such as citronellal, curcumenone, and ricinoleic acid, were inactive against α-chymotrypsin inhibition (Supplementary Table S1). Thus, the proposed engineering method, although preliminary, can be used to create novel candidates for α-chymotrypsin inhibitors. Further structural optimization, such as removal of the benzyl group on the amide nitrogen, construction of a dipeptide motif, and installation of an aromatic amino acid residue at the C-terminal of dipeptides³⁷ will improve the inhibitory activities of the Ugi adducts and will be reported in due course. The molecular weights of the synthesized plant product-like molecules were in the range of 492.70–1011.57 g mol⁻¹, which corresponds to the category of beyond the rule of five (bRo5)³⁸, or middle-molecules. Therefore, the present chemical hybridization method could be used to construct a small library of middle-molecules mimicking natural products, which is useful for the discovery of peptidic modulators of protein–protein interactions³⁹. Ongoing biological evaluations of the present molecules will unveil their specific pharmaceutical potential.

Table 3 α-Chymotrypsin inhibitory activity of Ugi adducts derived from plant products.