Synthesis of Avenanthramide C and Its Analogs Utilizing Classical Condensation, Saponification and Acidification

Abstract

This work presents a concise and practical synthetic route to the biologically multifunctional avenanthramide C (Avn C, 1a), one of the natural products derived from oats, utilizing three classical organic reactions. Condensation of diacetyl caffeoyl chloride (5) with methyl 5-hydroxyanthranilate (6) in tetrahydrofuran at 80 °C in the presence of pyridine, followed by saponification with 5.0 equiv of lithium hydroxide (3.0 M in H₂O) in a THF-H₂O (3:1 v/v) cosolvent system and subsequent acidification to approximately pH 2 using 6.0 M hydrochloric acid, successfully afforded 1a as a pale brown solid in >82% yield and 99% purity. Furthermore, 14 structural analogs of 1a were efficiently synthesized in 60–88% yields by applying the same synthetic strategy.

Introduction

Avenanthramide analogs (1a-i) are naturally occurring phenolic alkaloid derivatives isolated from oats in 1988 (Figure ). These phenolic alkaloids have a broad spectrum of biological applications, including antioxidant, anticancer, antiproliferative, anti-inflammatory, antiallergic, antithrombotic, and antipruritic effects. −

1.

Naturally occurring avenanthramide analogs and their biological activities.

Among these natural metabolites, it was reported that avenanthramide C (Avn C, 1a) showed significant potency against Alzheimer’s disease (AD), noise- and drug-induced hearing loss, and chemotherapy-induced peripheral neuropathies (CIPN). − The notable biological potency of 1a prompted us to explore the development of a scalable synthetic route to enable its preparation on a multigram scale, with the aim of facilitating further preclinical and potential clinical studies. To date, only a limited number of synthetic methodologies have been reported for 1a, underscoring the need for an efficient, practical, and scalable synthesis. Only a few chemical synthetic methods ,, and enzymatic synthesis have been reported for small-scale synthesis of 1a. Alternatively, natural Avn C is able to be extracted in microgram to milligram quantities, up to 80 mg per kilogram of oats by using a previously reported extraction method. Representative small-scale synthesis of 1a has been reported by Collins’ group (Figure , A). The condensation of diacetylated caffeoyl chloride (5) with 5-hydroxyanthranilic acid (3), followed by deprotection and purification with a Sephadex LH-20 resin (Figure , A) to give 1a in 50% yield. In 2011, Wise’s group prepared 1a in 40% yield by using the coupling reaction of diacetylated caffeic acid (4) with 3 in the presence of Bop [benzotriazol-1-yl-oxy-tris­(dimethylamino)-phosphonium hexafluorophosphate] reagent, followed by purification with a Sephadex resin (Figure , B). As an alternative synthetic approach, the Avn C analog was prepared using an aldol condensation reaction of 2-methylbenzoxazin-4-one with 4-acetoxy-3,5-dimethoxybenzaldehyde in the presence of catalytic p-toluenesulfonic acid (TsOH), followed by deprotection and acidification. Using the enzymatic approach, Avn A (1b) in Escherichia coli strain HA-Hpa has been converted into 1a on a microgram scale. Although these prior approaches are suitable for small-scale reactions, none are useful for synthesizing multigram quantities of 1a, owing to moderate yields, expensive reagents, and limitations associated with resin-based separation methods. Currently, the lack of an efficient muti-gram synthesis of 1a remains a significant obstacle to its advancement in preclinical studies. Therefore, we initiated the development of a practical synthetic approach for 1a to enable pharmacodynamic and pharmacokinetic studies in animal models of AD, noise-induced and drug-induced hearing loss, and CIPN.

2.

Synthetic approaches of Avn C (1a).

Herein, we initially present the results of coupling reaction of diacetyl caffeic acid (4) with 5-hydroxyanthanilic acid (3) and methyl 5-hydroxyanthranilate (6) using representative coupling reagents, as part of our effort to develop a synthetic method for 1a. Next, we report a convenient gram-scale synthetic approach for 1a (Figure , C), involving the condensation of diacetylated caffeoyl chloride (5) with 6, followed by saponification of the intermediate 8 and direct purification of 1a through pH-controlled salting-out. Our efficient synthetic approch enables large-scale preparation of Avn C without column chromatography, significantly improving yield over previous methods. Finally, this work presents the synthesis of several natural and novel avenanthramide derivatives using the new practical synthetic approach.

Results and Discussion

For the multigram-scale synthesis of Avn C (1a), initial efforts focused on the coupling reaction between diacetylated caffeic acid (4) and 5-hydroxyanthranilic acid (3), according to previously reported methods. The treatment of caffeic acid (2) with acetic anhydride (Ac₂O) in pyridine afforded diacetyl caffeic acid (4) in 97% yield as a white solid. As shown in Table S1 in the Supporting Information, the coupling reaction of 4 with 3 using the coupling reagents, namely N,N′-dicyclohexylcarbodiimide (DCC), 1-ethyl-3-(3-(dimethylamino)­propyl)­carbodiimide hydrochloride (EDC·HCl), (O-(benzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate (HBTU), and N,N,N′,N′-tetramethyl-O-(N-succinimidyl)­uronium tetrafluoroborate (TSTU), in the presence of diisopropylethylamine (DIEA) in dimethylforamide (DMF), did not observed the desired product 7 on thin-layer chromatography (TLC) (Table S1, entries 1–4). Additionally, when the same reaction was carried out with pyridine instead of DIEA/DMF, on both TLC and high-performance liquid chromatography (HPLC) analysis, trace amounts of compound 7 were consistently detected only in reactions using the DCC reagent, but not with the other reagents (Table S1, entries 5–8). Furthermore, compound 7 could not be perfectly separated from the reaction mixture by silica gel chromatography due to its high polarity and structural similarity relative to the byproducts. Therefore, compound 7 and the associated byproducts from the coupling reaction were monitored by HPLC [see HPLC chromatograms (a–e) corresponding to Table S1, entries 5–8 in Supporting Information].

As an alternative approach, 5-hydroxyanthranilic acid (3) was converted into its corresponding methyl ester derivative (6) to improve solubility in polar aprotic solvents and to increase hydrophobicity sufficiently for separation by silica gel column chromatography. Treatment of 3 with H₂SO₄ in MeOH under reflux condition successfully provided methyl 5-hydroxyanthranilate (6) in quantitative yield (Table , entry 5). However, alternative acidic conditions did not afford the desired product 6 (Table , entries 1–4).

1. Esterification of 5-Hydroxyanthranilic Acid (3).

entry	reaction condition	yield (%) of 6
1	3 M HCl in MeOH, rt, 24 h	-
2	3 M HCl in MeOH, reflux, 12 h	-
3	(COCl)2, MeOH, reflux, 12 h	-
4	H2SO4, MeOH, rt, 12 h	-
5	H2SO4, MeOH, reflux, 48 h	98

^aThe product was not observed on TLC.

In a subsequent experiment, compound 4 was coupled with 6 under the same conditions previously applied to the coupling of 4 with 3, as summarized in Table S2. Under DIEA conditions, the desired product 8 was not observed (Table S2, entries 1–4) on TLC. When excess pyridine was used as the base and solvent in place of DIEA, interestingly, the DDC and EDC reagents provided the desired product 8 along in low yield, along with a small amount of methyl N,O-biscaffeoyl anthranilate derivative 9, respectively [see HPLC chromatograms (a–e) corresponding to Table S2, entries 5 and 6]. The product 8 and byproduct 9 were partially separated in less than 5% yield from the crude mixture by silica gel column chromatography. Unfortunately, the other coupling reagents yielded only trace amounts of 8, as determinated by HPLC analysis (see HPLC chromatogram corresponding to Table S2, entries 7 and 8). The coupling results indicated that pyridine was slightly more effective than the combination of DIEA and DMF, serving as both a base and a cosolvent. However, based on these preliminary results, Wise’s approach for the multigram-scale synthesis of 1a using coupling reagents was found to be ultimately unsuitable and was therefore discontinued. It may be hypothesized that the formation of a crude mixture and the low yield of the desired product 8 resulted not only from steric hindrance in the activated intermediate formed between the caffeic acid derivative and the coupling reagent, but also from the inherently weak nucleophilicity of the aniline amino group. Furthermore, these observations suggest that the presence of a carboxylic acid group in 3 or a methyl ester group in 6, located adjacent to the amine, may reduce the nucleophilicity of the aniline moiety. This adverse effect is likely due to a combination of steric hindrance, electron-withdrawing effects, and potential intramolecular hydrogen bonding, arising from the proximity of a carboxylic acid or ester group to the amine nucleophile. − Therefore, to minimize steric hindrance around the electrophilic carbonyl group and facilitate amide bond formation between caffeic acid and anthranilic acid analogs, we considered a classical condensation reaction, similar to the approach described by Collins. Initially, compound 4 was converted to the corresponding caffeoyl chloride (5) as a highly activated acid intermediate, using oxalyl chloride [(COCl)₂] in the presence of catalytic DMF. After that, addition of a solution of 3 in excess pyridine to a solution of 5 in acetone at 0 °C, followed by refluxing at 100 °C for 20 min, afforded the desired product 7 in less than 14% yield (Table , entry 1). Although compound 7 was isolated from the reaction mixture by silica gel column chromatography, its purification remained challenging due to the presence of polar functional groups, including an amide (CONH), a carboxylic acid (CO₂H), and a hydroxyl group (OH). Under the same reaction condition, the condensation of methyl 5-hydroxyanthranilate (6) with 5 resulted in a dramatic improvement in the yield of product 8 (92%) without the formation of side products (Table , entry 2). On a 0.10 g scale of 4, product 8 was conveniently isolated from the reaction mixture by silica gel column chromatography. The excellent yield of 8 may be attributed to the role of pyridine, which functioned as a base, a nucleophile, and a cosolvent in the acyl chloride-mediated synthesis of carboxylic acid analogs. It is well-known that pyridine can react with acyl chlorides to form N-acylpyridinium ion intermediates, which are even more reactive toward nucleophiles than the parent acyl chlorides. In the next step, the coupling condition of 5 with 6 was optimized for a multigram scale synthesis by adjusting the solvent and temperature (Table , entries 3–8).

2. Results of the Condensation Reaction.

entry	starting material	reaction condition	product, yield (%)
1	3	acetone, pyridine, 100 °C, 20 min	7, 14
2	6	acetone, pyridine, 100 °C, 20 min	8, 92
3	6	CH2Cl2, pyridine, 50 °C, 20 min	8, 27
4	6	THF, pyridine, 40 °C, 20 min	8, 64
5	6	THF, pyridine, 60 °C, 20 min	8, 75
6	6	THF, pyridine, 80 °C, 20 min	8, >95
7	6	THF, pyridine, 100 °C, 20 min	8, >95
8	6	THF, pyridine, 100 °C, 40 min	8, >95

^aReaction temperature refers to silicon oil bath temperature.

^bAll reactions were performed on a 0.1 g scale of compound 4. Product 8 was isolated by a silica gel column chromatography as a single E-isomer.

^c N,O-Biscaffeoylanthramide 9 was obtained in 10–25% yields.

When the condensation of 5 with 6 was carried out under reflux in anhydrous dichloromethane instead of acetone, the yield of 8 significantly decreased to 27%, while byproduct 9 was formed in 25% yield (Table , entry 3). Unexpectedly, when the same reaction was performed in anhydrous tetrahydrofuran (THF) in an oil bath at 80 and 100 °C, compound 8 was isolated in over 95% yield without detectable formation of 9, regardless of the reaction time (Table , entries 6–8). However, upon lowering the reaction temperature in the same solvent, the yield of 8 gradually decreased to 75% at 60 °C (entry 5) and 64% at 40 °C (entry 4), accompanied by the formation of byproduct 9 in 10–16% yield. Therefore, refluxing a mixture of 5 and 6 in THF at 80 °C in the presence of pyridine afforded the desired product 8 in excellent yield, with minimal formation of byproduct 9, thereby identifying these as the optimal condition for the classical condensation reaction. Compound 8 was obtained exclusively as the E-isomer, as confirmed by the coupling constant (J = 15.5 Hz) in the ¹H NMR spectrum (see Supporting Information for the ¹H NMR data of 8) and by comparison with previously reported data. Considering the observed trend in the yield of 9, the ester bond formed between the carboxylic acid group of 5 and the 5-hydroxyl group of 6 may be cleaved under thermal conditions, possibly after partial formation. It is plausible that the phenolic ester was selectively cleaved under mild thermal and basic conditions in the presence of normal esters, due to its inherent lability.

Reagents and conditions: (a) (COCl)₂, CH₂Cl₂, cat. DMF, 0 °C then rt, 3 h; (b)­pyridine, THF, 0 °C, then 80 °C, 20 min; (c) (i) THF-H₂O (3:1 v/v), 3 M LiOH (aq), rt, 6 h; (ii) 6 M HCl (aq), 0 °C.

Subsequently, to investigate the scalability of the developed method, compound 8 was prepared on a gram scale using the optimized reaction condition, as summerized in Table . In the above optimization study, it was consistently observed that when the residue from the workup was dissolved in methanol for adsorption onto silica gel, the resulting solution gradually became turbid, indicating precipitation of the product. Fortunately, the condensation reaction of 6 with 5, which was prepared from 0.1 g of 4, followed by the same workup and precipitation in methanol at 0 °C afforded 8 in quantitative yield (Table , entry 1). On a practical scale (1.0 g of 4), the condensation reaction of 6 with 5 provided a dark-colored suspension, which was filtered to remove the insoluble salts, primarily pyridinium-hydrochloride. After evaporation of the filtrate, the residue was dissolved in toluene to remove excess pyridine, and then concentrated under reduced pressure. The resulting brown sticky oil was stirred with MeOH at 0 °C for 30 min, during which the desired product 8 spontaneously precipitated and was obtained in >95% yield (Table , entry 2). When the same procedure was applied to 5.0 and 10.0 g scales of 4, the solid product 8 was successfully provided in 88% and 85% yield, respectively, without the use of silica gel or resin-based column chromatography (Table , entries 3 and 4).

3. Results of Preparing 8 and Synthesis of Avn C (1a) on Gram Scales.

entry	amount of 4 (g)	yield (%) of 8	yield (%) of 1a
1	0.1	>97	94
2	1.0	>95	92
3	5.0	88	85
4	10.0	85	82

^aThe yield of 8 was calculated after solidification in MeOH.

^bThe overall yield of 1a was obtained from 4 over 4 steps.

Finally, as shown in Scheme , an efficient lithium hydroxide (LiOH)-mediated saponification of the methyl ester and acetyl groups of 8 was employed to complete the practical-scale synthesis of Avn C (1a) in high yield. Furthermore, to avoid the use of silica gel and resin column chromatography, a precipitation-based method exploiting the salting-out effect through pH adjustment of the aqueous solution was developed for the isolation of highly polar 1a. Treatment of 8 (0.10 g) with aqueous 5.0 equiv of LiOH (3.0 M in H₂O) led to the formation of the highly polar 1a as its corresponding lithium salt form. Following acidification of the basic reaction mixture to pH 4–5 with hydrochloric acid solution (6.0 M in H₂O), attempts to extract compound 1a into organic solvents were unsuccessful. This was likely due to its high hydrophilicity, attributed to the presence of a carboxylic acid (CO₂H) and three phenolic hydroxyl (OH) groups, which favored retention in the aqueous phase. The presence of these functional groups typically resulted in poor solubility of compound 1a in organic solvents, posing a significant challenge for purification by silica gel column chromatography. Ultimately, resin-based or reversed-phase C18 column chromatography proved to be an effective alternative. However, it was unexpectedly reported that caffeic acid analogs bearing a phenylpropenoic acid scaffold exhibited low solubility in aqueous solution at pH 2–3. Thus, the solubility of 1a was anticipated to be low in aqueous solution at approximately pH 2, owing to its molecular structure, which contains three hydroxyl groups and a carboxylic acid moiety, both characteristic of the caffeic acid scaffold. As expected, when the acidic solution (pH 4) was further acidified to pH 2 using 6.0 M HCl, compound 1a spontaneously precipitated as a pale brown solid in quantitative yield (Table , entry 1). The observed precipitation of 1a under acidic condition is believed to be driven by a salting-out effect, which is presumed to arise from the increased ionic strength upon addition of hydrochloric acid. Starting from 1.0 g of 4, a classical condensation of 5 with 6 in THF at 80 °C in the presence of pyridine, followed by saponification of intermediate 8 with 3.0 M LiOH (5 equiv) in a THF-H₂O cosolvent system, and subsequent acidification to pH 2 using 6.0 M HCl solution, successfully afforded 1a in 92% yield over four steps (Table , entry 2). When the same procedure was applied on a 5.0 g scale of 4, the yield of 1a was slightly reduced to 85% over four steps (Table , entry 3). As a final trial, transformation of 4 on a 10.0 g scale into its corresponding acid chloride 5, followed by classical condensation with 6, saponification, and acidification, afforded compound 1a in 82% yield with >99% purity as determined by HPLC (see HPLC chromatogram for purity of 1a in Supporting Information), without the use of reverse-phase or resin column chromatography (Table , entry 4 and Scheme ). Considering overall yield, the final two steps (saponification and acidification) provided a pale brown solid (1a) in >95% yield from intermediate 8. Currently, the prepared 1a is undergoing a non-GLP pharmacokinetic and toxicity study in a mouse model.

1. Synthesis of Avn C (1a).

As an extension of this work, several natural and novel Avn C analogs were conveniently synthesized using the efficient synthetic strategy developed for Avn C (1a). Structurally, these analogs are based on a 5-hydroxyanthranilic acid (3) scaffold, which serves as a core pharmacophore underlying their biological activity. Therefore, a series of Avn C analogs containing various cinnamic acid derivatives were synthesized using commercially available cinnamic acids, as depicted in Scheme . First, treatment of the hydroxyl groups in cinnamic acid analogs (10a–e) with acetic anhydride (Ac₂O) provided the corresponding O-acetylated cinnamic acid derivatives (11a–e), respectively. Subsequent reaction of 11a–e with (COCl)₂ in the presence of catalytic DMF in CH₂Cl₂ afforded the corresponding O-acyl cinnamyl chlorides (12a–e) in situ, which were used without further purification after concentrating the solution. Direct coupling reaction of intermediates 12a–e with 6 in the presence of pyridine in THF at 80 °C, followed by saponification and acidification conveniently afforded the desired products 1b,c and 13a–c in 77–88% yields, respectively. Additionally, the condensation of 6 with a bunch of cinnamoyl chlorides (12f–n), prepared by treating the corresponding cinnamic acid analogs (11f–n) with (COCl)₂, successfully afforded the desired Avn C derivatives. Subsequent treatment of the Avn C derivatives with 3.0 M LiOH (5 equiv), followed by adjustment of the pH to 2 using 6.0 M HCl, spontaneously afforded the corresponding Avn C analogs 13d-l, in yields ranging from 60% to 87%, respectively. The newly synthesized analogs are currently under investigation for their biological activity in models of AD, noise- and drug-induced hearing loss, and chemotherapy-induced peripheral neuropathy (CIPN).

2. Synthesis of Avn C Analogs (1b,c, 13a–l).

^a Intermediates 12a–n were prepared in situ from 11a–n, concentrated, and used directly in the subsequent step without purification.

Conclusions

In summary, an efficient synthetic route for Avn C (1a) was developed through classical condensation of diacetyl caffeoyl chloride (5) with methyl 5-hydroxyanthranilate (6), followed by saponification with 3.0 M LiOH in (5 equiv) and acidification to pH 2 with 6.0 M hydrochloric acid solution. This optimized method provided 1a in over 85% yield on gram scales starting from 6, as well as significantly simplified the purification process, notably avoiding the use of silica gel or resin column chromatography. Furthermore, applying the same synthetic procedure as for 1a, 14 Avn C analogs (1b,c and 13a–l) were also successfully prepared in 66–88% yields. This classical synthetic approach proves highly effective for amide bond formation between polar cinnamic acid analogs and sterically hindered anthranilic acid derivatives. The successful preparation of a variety of Avn C analogs (1b,c and 13a–l) demonstrates the broad substrate compatibility of this synthetic strategy and its effectiveness in overcoming the steric hindrance that often limits conventional coupling reactions. Avn C (1a) is currently being evaluated in ongoing non-GLP studies, and its novel analogs are being evaluated for biological activity in models of AD, noise- and drug-induced hearing loss, and CIPN. The outcomes of these investigations will be reported in due course.

Methods

General experimental procedures and the synthesis of Avn C and its analogs are reported in the Supporting Information section. All final compounds are ≥95% pure by HPLC and have their identities confirmed by ¹H-/¹³C NMR and high-resolution mass spectrometry (HRMS).

Glossary

Abbreviations

AD

Alzheimer’s disease

Avn C

avenanthramide C

CIPN

chemotherapy-induced peripheral neuropathies;

E. coli strain HA-Hpa

E. coli strain phenylacetic acid-hydroxyphenylacetic acid

Ac₂O

acetic anhydride

HBTU

(O-(benzotriazol-1-yl))-N,N,N′,N′-tetramethyluronium hexafluorophosphate

Bop

benzotriazol-1-yl-oxy-tris­(dimethylamino)-phosphonium hexafluorophosphate

DDC

N,N′-dicyclohexylcarbodiimide

EDC·HCl

1-ethyl-3-(3-(dimethylamino)­propyl)­carbodiimide hydrochloride

TSTU

N,N,N′,N′-tetramethyl-O-(N-succinimidyl)­uronium tetrafluoroborate

DIEA

diisopropylethylamine

Et₃N

triethylamine

HCl

hydrochloric acid

LiOH·H₂O

lithium hydroxide monohydrate

(COCl)₂

oxalyl chloride

TsOH

p-toluenesulfonic acid

CH₂Cl₂

dichloromethane

MeOH

methyl alcohol

DMF

N,N-dimethylformamide

THF

tetrahydrofuran

TMS

tetramethylsilane

HPLC

high performance liquid chromatography

HRMS

high resolution mass spectrometry

Mp

melting points

NMR

nuclear magnetic resonance

non-GLP

nongood laboratory practice

pH

power of hydrogen (potential of hydrogen)

TLC

thin layer chromatography

UV

ultraviolet.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c07816.

• Supporting Information (HPLC), experimental procedures for the synthesis and characterization of compounds (¹H- and ¹³C NMR spectra) and HRMS for new compounds (PDF)

†.

These authors (SMC and YJA) contributed equally. SMC designed synthetic methods and synthesized compounds. YEN, SJK and ERC assisted in synthesis of compounds. YJA synthesized compounds and reviewed and edited the draft. All authors have approved of the final version of the manuscript.

The authors declare no competing financial interest.