Total Synthesis of (+)-Hinckdentine A: Harnessing Noncovalent Interactions for Organocatalytic Bromination.

Hinckdentine A, a marine-sponge-derived tribrominated indole alkaloid bearing a unique indolo[1,2-c]quinazoline skeleton, was completed in 12 steps featuring the construction of the Nα-quaternary carbon center by asymmetric azo-ene cyclization. A novel organocatalyst was developed to promote high-yielding tribromination, which represents a challenging process encountered in previous syntheses. Density functional theory calculations scrutinized viable substrates and deciphered the origin of the enhancement of C8 electrophilic bromination with a bifunctional organocatalyst. Moreover, the application of organocatalyst-enabled bromination on various substrates was demonstrated to highlight future late functionalizations of biologically intriguing targets.

Halogenated natural products have been of utmost interest to the synthetic community because of their fascinating biological activities.[1,2] Halogens have been found to control metabolism and improve pharmacological properties, such as bonding affinity, lipophilicity, and permeability.[3,4] One-third of the drugs currently in clinical application are halogenated, such as antibiotic vancomycin, antimicrotubule cryptophycin A, and antitumor rebeccamycin. Cognizant of the crucial role of halogens in medicine, numerous efforts have been dedicated to the development of effective halogenation reactions.[5−7] Among them, imitating environmentally benign biohalogenation has been an enduring objective for decades. Besides the proximity-guided regioselectivity to overcome the intrinsic reactivities of specific substrates in enzymatic catalysis, notably electrophilic halonium ions have been generated through leveraging their reactivity with noncovalent interactions.[8−10] For instance, in the mechanistic understanding of tryptophan 7-halogenase (RebH) (Scheme ),[11−13] the protonated amine group of lysine plays a significant role to activate hypohalous acid (HOX), a generally weak electrophilic reagent, for halogenation at the requisite position by hydrogen bonding interactions. Such alignment is perfectly settled through the confined architecture of enzymes. However, probing biohalogenation by enabling catalysts remains a formidable challenge.

Scheme 1

Flavin-Dependent Halogenation by RebH (A) and Representative Late-Stage Chemical Brominations (B)

Hinckdentine A was isolated by Blackman and co-workers from marine bryozoan Hincksinoflustra denticulata as a unique screw-shaped and tribrominated indolo[1,2-c]quinazoline.[14−17] Although construction of such unprecedented skeleton remains to be biosynthetically elucidated, synthetic endeavors have been devoted to devise novel strategies to access the pentacyclic architecture.[18] McWhorter and Liu suggested a late-stage enzymatic bromination of 2 to mimic the hypothetical biosynthesis of hinckdentine A (1).[19] However, such a tribromination event turned out to be extremely challenging because the introduction of the third bromine at C8 with various bromine reagents was unsuccessful. Kawasaki and co-workers disclosed an alternative approach to define a reduced form, N5-Boc intermediate 3, as a suitable substrate for tribromination under mild conditions.[20] The succeeding contributions by Kitamura and Fukuyama,[21] Zhu,[22] Xu,[23] and Cheon[24] decorated three bromines on different stages through development of novel synthetic strategies.

To expedite the substrate-defining process for the challenging bromination, we envisioned that a computational calculation could be used to rationalize valid intermediates, including those that closely resemble the precursor in the proposed biogenesis, that can participate in the synthetic design of complex natural products.[25] The distinct reactivities of advanced intermediates 2 and 3 upon bromination prompted us to revisit the origin of selectivity, both computationally and experimentally. To simplify the mechanistic study, we defined the tangled bromination at C8 as the reaction of interest. Therefore, dibrominated derivatives 4 and 5 (vide infra) were examined under Kawasaki’s condition with N-bromosuccinimide (NBS). The reaction mechanism involves a simplified sigma or condensed complex. As shown in the energy profiles (Figure ), for 8-debromohinckdentine A (4) in McWhorter’s synthesis,[19] C8-bromination occurred via an unusual concerted mechanism, and the energy barrier for C8-bromination was 40.8 kcal/mol, which indicated that it was extremely difficult to brominate this position. In contrast to 4, C8-bromination of 5, a proposed dibromine compound derived from Kawasaki intermediate 3,[20] proceeded via a conventional stepwise mechanism. Although the energy barrier for the rate-determining step of the C–Br bond formation is significantly lower than that of 4, the relatively high Gibbs energy (ΔG = 28.2 kcal/mol) indicates a dilemma in forming the Wheland intermediate (σ-complex) 5-C8-IM1, which is suspicious for the facile C8-bromination in Kawasaki’s actual synthesis.[20]

Figure 1

Energy profiles of noncatalytic C8-bromination of compound 2-derived dibromine 4 (3-derived 5) with NBS. The corresponding tribrominated products are hinckdentin A (1) and 6. Calculations were performed at the B3LYP-D3(BJ)/6-31G(d,p)(SMD,CH2Cl2) level of theory. The relative Gibbs free energies (ΔG) are given in kcal/mol. B-5-C8 (blue) represents the bromination of substrate 5 at C8; TS, transition state; IM, intermediate.

Despite that bromination at different stages with various halogenases[26,27] or even a reversed process (debromination) is also possibly adapted in nature,[28] the uncertainty in bromination of 2 and 3 and those circumvented approaches in the previous syntheses[24] prompted us to revisit the key tribromination at the late stage of synthesis. Such bioinspired execution by late-stage functionalization would offer a unique opportunity to explore reactivity and method development as well as future study on biological profile of derivatives.[29] As outlined in Scheme , the seven-membered lactam (caprolactam) in 3 can be derived from enone 7 via a regioselective Beckmann or Schmidt rearrangement. The essential Nα-quaternary center spotlighted in many alkaloid syntheses[30] can be established via azo-ene cyclization[31−34] of the key azo species 9, which should be readily prepared from tetrasubstituted alkene 10. Accordingly, two sequential palladium-catalyzed coupling reactions are designed to merge three readily available fragments (vida infra).

Scheme 2

Synthesis Plan of Hinchdentine A

The synthesis commenced with the α-arylation of cyclohexanone 11 with 2-chloroiodobenzene in the presence of Pd2(dba)3 and xantphos on a decagram scale (Scheme ). The excess amount of cyclohexanone 11 was readily recovered by treatment with NaHSO3 during the workup stage. After transformation of the corresponding ketone 12 into triflate, subsequent Suzuki coupling of boronate 13 proceeded smoothly to deliver tetrasubstituted alkene 10 in excellent yield (86%). In the Nα-quaternary stereogenic center, a racemic form was first realized by application of Leblanc’s method.[32] To this end, through treatment with triphosgene and immediate addition of hydrazine, the resulting urea-derived hydrazine 14 was effectively converted to the requisite Nα-quaternary center via oxidative azo-ene cyclization with NBS/pyridine.[35] Cleavage of the N–N bond by zinc powder in acetic acid and simultaneous deprotection of the ketal smoothly released enone 7 in 83% yield.

Scheme 3

Preparation of Advanced Intermediates 2 and 3

With a sufficient amount of enone 7 in hand, we first examined the Beckmann rearrangement to furnish the caprolactam ring.[36,37] To our surprise, after treatment with hydroxylamine, subsequent sulfonylation (such as Ts and Ms) failed to deliver the rearranged product under various conditions. Alternatively, an acid-promoted Schmidt rearrangement proved to be beneficial when methanesulfonic acid (MsOH) was used, and an excellent yield (98%) of the requisite lactam 16 was achieved on the gram scale. This high-yielding regioselective Schmidt rearrangement of the enone circumvented a six-step detour in previous synthesis.[21] To complete the pentacyclic skeleton, tremendous efforts to screen phosphine ligands had been explored, and the precatalyst arylpalladium(II) complex 17(38) proved to be essential to enable intramolecular Buchwald–Hartwig amination. The structure of the corresponding pentacyclic urea 18 was further verified by long-range 1H–15N heteronuclear shift correlation in NMR experiments.[39] The subsequent hydrogenation of the C4–C5 double bond generated saturated lactam 19 with excellent stereoselectivity (dr > 20/1). Complete reduction of the carbonyl group of the urea into a methylene group with DIBAL-H provided 20 in excellent yield.

To construct a Nα-quaternary stereogenic center in an asymmetric manner, the key objective is defining a chiral auxiliary for azo-ene cyclization. The pioneering work on asymmetric azo-ene reactions by Brimble and co-workers[40,41] forged the potential inception of this strategy. However, functional tolerance in the complex structures and subsequent cleavage of the N–N linker remain elusive due to the holistic adjustment of chiral auxiliaries for diastereomeric induction, oxidative formation of azo compounds, and reaction feasibility toward the following chiral auxiliary removal. Camphsultam 14a was proven to be a compromise to above scenarios and yielded 8a with good diastereoselectivity (dr 5/1) (Scheme ; also see Table S1). The direct cleavage of the N–N bond was also found to be problematic using previously known methods.[42] Alternatively, removal of the chiral auxiliary by hydrolysis was executed to deliver hydrazine 8b in good yield. Subsequently, after screening extensive protocols, deamination by nitrous acid[43] was found to be effective. These two steps were further merged in one pot to deliver the requisite (S)-enone 7 in good yield (67% for 8a to 7). Following the aforementioned synthetic route in Scheme , pentacyclic compound (+)-20 was readily constructed in a four-step sequence.

Scheme 4

Asymmetric Azo-Ene Cyclization to Access (S)-20

During the removal of the benzyl group by hydrogenolysis of 20, the yield of amide 21 was varied due to contamination of compound 2. To suppress the dehydrogenation, HCO2NH4 was defined as a surrogate for an effective hydrogen source, and 21 was isolated in 85% yield. Interestingly, using AcOH as the solvent, the addition of HCl (aqueous, 1.2 equiv) for the debenzylation process delivered compound 2 with a 95% yield, implicating that the protonation of aniline might facilitate the dehydrogenation at C6 to form an imidazolidine motif. An alternative route toward 21 was also devised by the facile reduction of 2 by NaBH4. Following protection with the Boc group, bis-Boc product 22 was first obtained, and immediate treatment with Mg(ClO4)2[44] afforded mono-Boc-protected compound 3 in 98% yield.

The advanced intermediates 2 and 3 were then subjected to verify the bromination under previous conditions.[19,20] Compound 2 was then resumed using various equivalents of NBS (Scheme ). The bromines at C10 and C2 were readily installed (i.e., 4), and bromination at C8 was extremely difficult. The requisite tribrominated product remained undetectable, even though excess NBS (∼8 equiv) or strong brominated reagents (i.e., Br2) were utilized. These unsuccessful entries implicated that C8-bromination of the hypothetical biogenesis via 2 may require stepwise[26,27] or proximity-controlled halogenation to overcome the inherent reactivity,[11] posing an intriguing problem for future biosynthetic studies. On the other hand, the availability of compound 3 in our unified route encouraged us to revisit another tribromination event. Surprisingly, under the known conditions with 4 equiv of NBS in THF,[20,23] the major product was C10-monobromo derivative 23 (82%), along with a small amount of C2,C10-dibrominated derivative 5 (9%). No appreciable amount of tribromide 6 was detected (entry 1, Table 1 in Scheme ). Further extending the reaction time, increasing the usage of brominating reagent, and enhancing the reaction temperature resulted in a low isolated yield (14%) of 6, even when mono and dibromo derivatives were consumed.[45] The relatively high Gibbs free energy in the transition state for the C8-bromination of 5 with NBS (Figure ) concurred with our experimental results.

Scheme 5

Bromination of 2 and 3

The yields in Table 1 were determined by 1H NMR (400 MHz, CDCl3) with internal standard CH2Br2.

The poor reproducibility of the original conditions[20] urged us to explore alternative approaches to late-stage bromination. Given the Lewis-base-activated bromination,[46−54] we examined several Lewis bases, such as DMSO[52] and Et2S[47] (Table 1 in Scheme ).[45] Following Gustafson’s intriguing discovery,[55] tributylphosphine sulfide was examined and proved to be effective in producing a 50% yield of tribromide 6. Further optimization defined triphenylphosphine sulfide (TPS) to be superior for C8-bromination (58% yield) within a shorter reaction time. During the bromination, we noticed a severe incapacitation of the catalyst by hydrolysis of the phosphine sulfide–NBS complex to phosphine oxide. To remedy this problem, we used 40 mol % of the phosphine sulfide catalyst (relative to the substrate). For the TPS-promoted C8-bromination of C2/C10 dibromine derivative 5, density functional theory (DFT) calculations demonstrated that the relative Gibbs energy of transition state 5-C8-TS1 (ΔG = 18.5 kcal/mol) (Figure S1)[45] in the C–Br bond formation step was significantly lower than that for the condition using NBS alone (B-5-C8-TS1, ΔG = 28.2 kcal/mol, Figure ). Consequently, hydrogen transfer becomes the rate-determining step (ΔG = 19.0 kcal/mol). In the presence of a stronger bromination reagent, that is, 1,3-dibromo-5,5-dimethylhydantoin (DBDMH), tribromide 6 was obtained in 80% yield.

The success of bromination of substrate 3 with DBDMH encouraged us to evaluate the feasibility of structurally resembled bis-Boc substrate 22. This notion supported the fact that the optimal combination of DBDMH and TPS (40 mol %, relative to the substrate) resulted in 87% isolated yield of tribromide 24 (entry 1, Table 2 in Scheme B). Further conversion revealed an intriguing dual function of cerium ammonium nitrate, namely, deprotection of the Boc group and concurrent dehydrogenation in a single step to complete the total synthesis of hinckdentine A (1) (Scheme A). To achieve a high turnover of the catalyst, 5 mol % of TPS was used, and the yield of given tribromide 24 deteriorated (32% in entry 2, Table 2 in Scheme ) due to the incapacitation of the catalyst by hydrolysis of the phosphine sulfide–DBDMH complex to phosphine oxide. Interestingly, N,N′-3,5-bistrifluoromethylphenyl thiourea (F-thiourea 25)[56] was also capable of promoting dibromination to deliver 28,[57−59] although introducing bromine at C8 remained incomplete due to the decomposition of the catalyst itself (entry 3), indicating that the third bromination of 22 remains the required step for catalyst development.

Scheme 6

(A) Completion of (+)-Hinckdentine A (1), (B) Optimization of Organocatalytic C8-Bromination of 22, and (C) Optimized Geometries for Transition States in the C8-Bromination of 28 with Different Catalysts (Bond Lengths in Å and Relative Gibbs Free Energies (ΔG) in kcal/mol)

Compound 22 (20 mg) in 1.5 mL of CH2Cl2 (0.026 M) at 0 °C, 1 h; DBDMH (4 equiv) was used as the bromination reagent.

All yields, except denoted, were determined by 1H NMR (400 MHz, CDCl3) with internal standard CH2Br2; C10-monobromo derivative 27 and C2,10-dibromo derivative 28 were isolated; see the Supporting Information for details.

Isolated yield.

NBS (4 equiv) was used as the bromination reagent.

The reaction was run on the 0.04 M and 1 mmol scale.

In order to further improve the efficacy of the organocatalyst in tribromination, we envisioned that merging phosphine sulfide and thiourea may further activate the bromination reagent, which in turn decreases the energy barrier to enhance catalyst turnover. The use of mixed catalysts indeed exhibited a significant improvement of tribromination (entry 4). A new catalyst 26a bearing two adjacent activating sites was thus prepared and subjected to the given transformation (Table 2 in Scheme ). It was gratifying that a catalyst loading level as low as 5 mol % in the presence of DBDMH was effective in achieving an excellent yield (86%) for the targeted tribromide 24, which is superior to the mixture use of Ph3P = S and 25 (entry 5 vs 4). Other catalysts 26b/c with weaker hydrogen bond donors (Figure S4)[45] resulted in a deteriorated yield (entries 6 and 7). Using the thiourea catalyst, bromination with NBS was also enhanced to 76% (entry 8). The reaction was readily scaled up, even with 2.5 mol % (entry 9). Theoretically, the N5-Boc group can be leveraged as an acceptor to perform dual hydrogen bonding with urea-like donor 26a in preorganized intermediate 28-C8-COM-b (Figure S4).[45] For the TPS-promoted bromination of dibromo derivative 28, in the presence of DBDMH, the theoretical calculation suggested facile formation of the Wheland intermediate via transition state 28-C8-TS1-a, with an energy barrier of only 4.4 kcal/mol (TS1-a shown in Scheme C; for details, see Figure S2).[45] The structural analysis of transition state 28-C8-TS1-b indicated that noncovalent interactions, including a hydrogen bond between the bromonium ion and the thiourea moiety as well as additional weak attraction, orient the substrate and strengthen the electrophilicity of the bromonium ion, leading to a compact transition state with lower relative Gibbs free energy (ΔG = 3.1 kcal/mol vs 4.4 kcal/mol of the parent TPS) (Scheme C and Figure S2).[45]

To demonstrate the application of organocatalyst 26a in halogenation of biologically intriguing substrates, aromatic compounds bearing various functional groups were evaluated with comparison to TPS-catalyzed halogenations[55] and background reaction by NBS or NCS only (“none” refers no catalyst) (section A, Scheme ). Organocatalyst 26a generally exhibited superior reactivity to TPS in terms of the conversion and isolated yield. For substrate 29m, a para-preference (rr 13/1; see the Supporting Information for details) was found with 26a and TPS, while ortho-bromination was obtained (rr 10/1) with NBS only. The difference is also profound when selected pharmaceuticals and natural products were examined in parallel. Notably, a wide range of functionality are well-tolerated, including acids, amines, ketone, furan, imidazole, and lactones. The brominated analogues of diclofenac, clotrimazole, estrone methyl ether, gemfibrozil, metaxalone, xanthotoxin, and naproxen were each accessed in excellent yield with this method (section B, Scheme ). For the chlorination, the reactivity of 26a toward gemfibrozil and naproxen was inferior to that of TPS, whereas it remains a better catalyst for xanthotoxin (98% yield brsm). In terms of the facile transformation of bromo compounds,[60] this new organocatalyst clearly provides an enabling approach to late-stage modification and intermediate for further derivatization.

Scheme 7

Catalytic Halogenations for Various Substrates with 26a and TPS

Reactions were conducted with 0.1 mmol scale of substrate and NBS or NCS (1.2 equiv) in CDCl3 (2 mL), and the yield was determined by 1H NMR (400 MHz) using CH2Br2 as an internal standard. The reactions were generally repeated two or three times for reproducibility. The catalyst loading (26a or TPS) was given in parentheses.

The yield for 30a–e and g with applying TPS as the catalyst was adapted from ref (54).

The reaction temperature is 0 °C.

DBDMH (1.2 equiv) was used as bromination reagent.

NBS or NCS (2.2 equiv) was used as halogenation reagent.

The isolated yield was given when xanthotoxin was applied in gram scale for bromination.

In summary, we streamlined the synthesis of hinckdentine A (12 steps) from simple starting materials, featuring the first asymmetric azo-ene cyclization to furnish the N-containing quaternary carbon center and highly regioselective Schmidt rearrangement and Buchwald–Hartwig amination to construct the full carbon framework of indolo[1,2-c]quinazoline. DFT calculations scrutinized the feasibility of tribromination in previous syntheses and defined viable substrates for successful C8-bromination. The novel bifunctional phosphine sulfide catalyst bearing a thiourea moiety as a hydrogen bonding donor effectively realized the requisite bromination. Moreover, halogenation of various substrates reinforced this superior method to provide intriguing derivatives for future modification and biological evaluation. Further exploration to design a new catalyst system for bromination of the proposed biosynthetic intermediate (i.e., 2) as well as harnessing synthetic capabilities on site-selective functionalization[61−63] via noncovalent interactions in the context of natural product synthesis is currently underway in our laboratory.