Synthesis of Hemigossypol and its Derivatives

Abstract

Hemigossypol (3), a sesquiterpene natural product, was previously isolated from Gossypium barbadense and was shown to display improved anti-fungal activity compared to gossypol (1), the disesquiterpene dimer of hemigossypol (3). Gossypol exhibits multiple biological activities. In order to study whether hemigossypol and it derivatives retain the various bioactivities of gossypol, we developed a short and convenient synthetic scheme to synthesize hemigossypol. This is the first de novo synthesis of this natural product. In addition derivatives of hemigossypol with various 2,5-alkyl substituents were synthesized. Modification of the synthetic scheme also afforded the natural product hemigossylic lactone (4) and its 2,5-substituted derivatives.

Gossypol (1), a natural product from cottonseed, is a disesquiterpene exhibiting atropisomerism owing to restricted rotation around the binaphthyl bond (Figure 1). Gossypol exhibits multiple biological properties, including spermicidal,¹ antiparasitic,² anticancer,³ and antiviral activities.⁴ Gossypol displays potent inhibition against anti-apoptotic Bcl-2 family proteins where it functions as a BH3 mimic.⁵ (−)-Gossypol is currently in phase II clinical trails, displaying single-agent antitumor activity in patients with advanced malignancies.^6,7 Multiple bioactivities of gossypol have stimulated wide interest in the development of gossypol derivatives to explore structure-activity relationships.^8–13 We have previously reported that gossylic lactone (2)⁹ (Figure 1), in which the aldehyde groups of gossypol that contribute to toxicity have been modified, retains antimalarial activity and exhibits enhanced inhibitory activity compared to gossypol against aldose reductase, an enzyme implicated in the etiology of diabetic complications.^9,14

Figure 1.

The question of interest is whether hemigossypol (3) and hemigossylic lactone (4) (Figure 1), the monomers of gossypol and gossylic lactone, respectively, still retain the various bioactivities of gossypol. Hemigossypol, a natural product isolated from the verticillium-infected stele tissue of Gossypium barbadense,¹⁵ is at least three-fold more toxic to the cotton seedling disease pathogen, Rhizoctonia solani, than gossypol.¹⁶ Furthermore, it was reported that methylated hemigossypol displayed comparable spermicidal activity as gossypol.¹⁷ Hemigossylic lactone with a methoxy group on position six is a natural product isolated from the root bark of B. malabarium;¹⁸ however, bioactivity has not been reported. In order to explore SAR of hemigossypol and hemigossylic lactone, we developed a short and convenient synthetic scheme to synthesize hemigossypol and its 2,5-alkyl substituted derivatives. Modification of the synthetic scheme also afforded hemigossylic lactone and its 2,5-alkyl substituted derivatives.

Synthetic Scheme 1 features the incorporation of the carbon atoms for the second ring of the naphthalene system in one step by the reaction of the Grignard reagent formed from 1-bromo-2-isopropyl-3,4-dimethoxybenzene (5a) and 1-bromo-3,4-dimethoxy-2-propylbenzene (5c) with ethyl 3-methyl-4-oxobutanoate (6a) and ethyl 2,3-dimethyl-4-oxobutanoate (6b) to form lactones 7a, 7b and 7c in 63–70% yield.¹⁹ The precursor bromides are readily prepared from commercially available starting materials using procedures described in the literature.^{20, 21, 22} In some preparations of the bromides using the Edward's procedure²² small amounts (5 to 10%) of the isomeric bromides 5-bromo-3-alkyl-1,2-dimethoxybenzene were formed.²³ An alternate highly regioselective silica gel catalyzed bromination, using N-bromosuccinimide proved more effective.²⁴ The precursors ethyl 3-methyl-4-oxobutanoate (6a) and ethyl 2,3-dimethyl-4-oxobutanoate (6b) were prepared by a Horner-Emmons-Wadsworth reaction followed by hydrogenation of the alkene bond and hydrolysis of the resulting acetal as previously described.^{25, 26} In our hands hydrogenation using palladium on charcoal resulted in loss of a methoxy group of the acetal but use of palladium on calcium carbonate gave the desired acetal, which was then hydrolyzed.

Scheme 1.

Synthesis of hemigossypol: Reagents and conditions: (a) Mg (3 equiv), BrCH₂CH₂Br (0.1 equiv), THF, reflux, 1.0 h then 6 (1.2 equiv), 0°C to rt, 12 h, H₂O/HCl, rt, 5 h, 70% of 7a, 66% of 7b, 63% of 7c. (b) concentrated sulfuric acid, rt, 60°C, 2.0 h, 5%. (c) BBr₃ (5 equiv), CH₂Cl₂, 2.0 h, −78°C, 24 h, rt, 90% of 9a, 70% of 9b, 81% 9c. (d) TiCl₄ (5 equiv), Et₂O/CH₂Cl₂, 0°C, 15 min, then Cl₂CHOCH₃ (2 equiv), 0°C, 3 h, rt, 18 h, 16% of 3a, 10% of 3b, 20% of 3c, and 4% of 10a, 34% of 10b, 25% of 10c; or 9b, hexamethylenetetramine (0.5 equiv), HAc, 100°C, 1.0 h, then H₂SO₄/H₂O, 60°C, 2 h, 25% 10b (e) CH₃CN/H₂O, 6M HCl, rt, 18 h, 90% of 3b, 100% of 3c. (f) BBr₃ (7 equiv), CH₂Cl₂, 0°C, then reflux 3 h, 0°C, CH(OC₂H₅)₃ (6 equiv), reflux, 1.0 h, rt, 24 h, 10%. of 10b from 7b, 10% of 10c from 7c. (g) BBr₃ (5 equiv), CH₂Cl₂, 0°C, reflux, 3 h, then CH(OC₂H₅)₃ (6 equiv), 0°C, reflux, 1 h, rt, 24 h, 25%. (h) (CH₃O)₂SO₂ (6.5 equiv), K₂CO₃ (6.5 equiv), (CH₃)₂C=O, reflux, 24 h, 90%.

In the synthesis of gossypol performed by Venuti²⁷ compound 5a was transformed to naphthalenol compound 8 in six steps. In one of the steps Venuti proposed compound 7a as an intermediate, but it was never isolated. We envisioned that lactone 7a could be converted to naphthalenol 8 in one step using a strong Lewis acid which would cause the opening of the lactone ring followed by Friedel Crafts cyclization of the resulting acid, dehydration and aromatization.²⁸ Attempts using dilute hydrochloric acid, phosphorous trichloride or titanium tetrachloride failed and concentrated sulfuric acid afforded only 5% of product. However lactones 7a–7c were successfully converted to naphthalenols 9a–9c using boron tribromide in 70–90% yield.²⁹ Compound 9a is slightly unstable and degrades during chromatography. The aldehyde group was introduced by formylation of naphthalenols 9a–9c with titanium tetrachloride and dichloromethyl methyl ether^30,31 followed by hydrolysis, to successfully afford the natural product hemigossypol 3a and its derivatives 3b and 3c in moderate yield.³² The proton NMR spectrum of compound 3a generally agreed with the literature¹⁵. Anhydrohemigossypol 10a has been observed in the mass spectrum of hemigossypol¹⁵ and anhydrogossypol, the dimer of compound 10a, is well documented in the literature.^{33, 34} The proton NMR spectrum of anhydrohemigossypol 10a and its derivatives 10b and 10c is similar to that of anhydrogossypol.³⁵ The anhydro compounds 10a–10c could be converted back to aldehydes 3a–3c using dilute hydrochloric acid in acetonitrile.

Alternate synthetic methods for the formation of compounds 3 from naphthalenols 9a and 9b were investigated in hopes of increasing the yield of product. Vilsmeier and N,N-diphenylformamidine methods did not afford product.^{36, 37, 38} The Duff reaction, which involves the use of hexamethylenetetramine in the presence of acetic acid, on compound 9b gave the dehydrated compound 10b in a 25% yield.³⁹ These disappointing results led to the one pot reaction of compound 7a with boron tribromide followed by triethylorthoformate to give compound 11 in 25% yield.⁴⁰ Unfortunately, compound 11⁴¹ is the 2-formyl regioisomer of compound 3 and its structure was verified by methylation to give the known compound 12.¹⁷ Using the same conditions on compound 7b, which has a methyl group in position two, 10% of compound 10b was obtained.

Since formylation of naphthalenols 9 resulted in poor yields of product, compounds 9a and 9b were methylated using dimethylsulfate to give compounds 13a and 13b.⁴² (Scheme 2). Formylation reactions on the the dimeric (5,5'-diisopropyl-1,1',6,6',7,7'-hexamethoxy-3,3'-dimethyl-2,2'-binaphthalene) and the monomeric form of compound 13a have been published by Meltzer.³⁷ Meltzer found that bisformylation and concomitant loss of both isopropyl groups occurred on reaction of the dimeric form of compound 13a using titanium tetrachloride and dichloromethyl methyl ether. Identical treatment of the monomeric compound, 13a, resulted in formylation in the 8 position, peri to the isopropyl group. Using these same conditions we reacted compound 13b and obtained a mixture of products with the major product lacking an isopropyl group. Fortunately, formylation of compound 13a using t-butyllithium followed by the electrophile N-methylformanilide and hydrolysis gave a 49% yield of compound 14 (Scheme 2) whereas reaction of 13b using these conditions failed to give the desired product.^{17, 43} Demethylation of compound 14 with boron tribromide¹⁰ afforded an 80% yield of compound 3a.

Scheme 2.

Synthesis of hemigossylic lactone and derivatives: Reagents and conditions: (a) (CH₃O)₂SO₂ (13 equiv), K₂CO₃ (13 equiv), (CH₃)₂C=O, reflux, 85% of 13a and 13b. (b) t-BuLi (8 equiv), C₆H₁₂, 0°C, 18 h then HCON(CH₃)C₆H₅ (2.5 equiv), 8 h, 49%. (c) BBr₃ (3 equiv), CH₂Cl₂, −78°C, 1.0 h, 0°C, 1.0 h, rt, 1.0 h, 80%. (d) t-BuLi (7 equiv), C₆H₁₂, 0°C, then 18 h, rt, then ClCOOCH₃ (19 equiv), −10°C, 8 h, 60% of 15a. (e) BBr₃ (4 equiv), CH₂Cl₂, −78 C, 1.0 h, 0°C, 1.0 h, rt, 3 h, 6M HCl, reflux, 3 h, 80% of 4a and 71% 4c. (f) Br₂ (1 equiv), CHCl₃, 0°C, 2 h, 88% of 15c.

Synthesis of hemigossylic lactone 4a and its derivative, 4c, is similar to the synthesis of compound 14 except for the electrophile. Reaction of compound 13a with t-butyllithium followed by methylchloroformate resulted in ester 15a.⁴⁴ Bromination of compound 15a gave bromide 15c.^{43, 45} The methyl ether groups in compounds 15 were readily removed using boron tribromide and the resulting mixtures were gently refluxed in dilute hydrochloric acid to undergo intramolecular lactonization and afford lactones 4a and 4c in 71–80% yield.⁴⁶

A one step transformation from the readily available intermediate 7b to naphtholactone 16 is an attractive route to explore. Since lactones 7a–7c have been successfully converted to naphalenols 8 and 9a–9c in one step using Lewis acids (Scheme 1), we further envisioned that chlorosulfonyl isocyanate (CSI),⁴⁷ a strong Lewis acid containing a reactive electrophilic isocyanate group, could convert lactone 7b into naphtholactone 16 in one step. Treatment of compound 7b with excess CSI followed by an acidic hydrolysis (Scheme 3) successfully afforded naphtholactone 16 in a 10% yield along with the unexpected amide compound 17 in 30% yield.⁴⁸ We propose that initially the CSI acts as a Lewis acid and readily converts the lactone 7b to the 2-methylsubstituted naphthalenol adduct of compound 8. This naphthalenol adduct further reacts with the electrophilic carbonyl carbon in the CSI and hydrolysis using dilute acid causes intramolecular attack of the carbonyl by the peri hydroxyl group on position one resulting in formation of lactone 16. Amide 17 was formed by the reaction of CSI with the substituted naphthalenol at position four followed by acidic hydrolysis. Because of the brevity of this synthetic sequence and the promising results, reaction conditions are being investigated to improve the yield.

Scheme 3.

Synthesis of dimethylhemigossylic lactones: Reagents and conditions: (a) ClSO₂NCO (9 equiv), CH₂Cl₂, rt, 24 h, then HCl/H₂O, EtOH, reflux, 1.0 h, 10% of 16 and 30% of 17.

In conclusion, a short and convenient synthetic scheme was developed to afford the bioactive natural product hemigossypol 3a and its 2, 5-substituted derivatives 3b and 3c. The modification of the synthetic scheme afforded hemigossylic lactones 4a and 4c as derivatives of the natural product 3-methylhemigossylic lactone. Bioactivities of hemigossypol, hemigossylic lactones and their derivatives are currently being investigated.