Kinetically Controlled aza-Michael/Epimerization Cascade Enables a Scalable Total Synthesis of Putative (+)-Fumigaclavine F

Abstract

The first total synthesis of putative (+)-fumigaclavine F (seven pots, 16% overall yield, 400 mg scale) and its proposed biosynthetic precursor (seven pots, 21% overall yield, 400 mg scale) have been achieved. The stereochemistry of both were unambiguously verified via XRD. The synthesis features a decagram scale asymmetric intramolecular Mannich reaction, followed by an aza-Michael/epimerization cascade that constructs the cis-fused tetracyclic scaffold. The synthesis provides an excellent approach to probe the bioactivity of clavine alkaloids.

Graphical Abstract:

Clavine alkaloids are a subclass of ergot alkaloids, a group of indole-derived mycotoxins that are produced by sac fungi, some of which possess a tetracyclic scaffold.¹ Clavine alkaloids possess a wide array of biological activity, including anti-cancer and anti-inflammatory activities, which has prompted extensive biological studies.^2–13 In addition, some clavine alkaloids can mimic neurotransmitters, which enables them to bind to dopamine, serotonin, and adrenergic receptors.^13–16 A 2014 study searching for new antibiotics identified five new tetracyclic clavine alkaloids (fumigaclavine D-H, Figure 1) produced by endophytic Aspergillus fumigatus isolated from Cynodon dactylon.¹¹ These five alkaloids were screened against a panel of anaerobic microbes including P. anaerobius, B. diatasonis, V. parvula, A. israelii, B. vulgatus, and S. anaerobius. Of these alkaloids, fumigaclavine F (1a) displayed the most promising activity, exhibiting broad spectrum anti- microbial properties (16–64 μg/mL) against sthe entire panel.¹¹ Fumigaclavine F (1a) features a cis-fused tetracyclic ring system and substitution at C-2 and C-9, which have not been previously reported together in clavine alkaloids. Fumigaclavine F (1a) shares a carbon skeleton with the clavine alkaloid costaclavine (1b), which has been asymmetrically synthesized by Liu and Osanai.^{17, 18} However these approaches are long, low yielding, and not amenable towards synthesizing fumigaclavine F (1a).^{17, 18} Therefore, there is need to develop a new approach to access fumigaclavine F (1a). Herein, the development of a seven-pot total synthesis of putative (+)-fumigaclavine F (1a) and its proposed biosynthetic precursor (2) are reported. To probe the bioactivity of clavine alkaloids, our approach was designed to be concise, modular, scalable, and to enable access to both enantiomers of 1a (Scheme 1).

Figure 1.

Clavine alkaloids isolated from endophytic Aspergillus fumigatus.

Scheme 1.

Retrosynthetic analysis of fumigaclavine F (1a)

We envisioned that fumigaclavine F (1a) could be prepared in a biomimetic manner via prenylation of its proposed biosynthetic precursor (2).^{11, 19} The proposed biosynthetic precursor (2) could be prepared via reduction and deprotection of cis-tetracycle 3a. We proposed that cis-tetracycle 3a could be rapidly assembled via an asymmetric Mannich reaction²⁰ of enone 4, followed by an aza-Michael/epimerization cascade. Enone 4 could be accessed from commercially available 4-bromo-3-formylindole (5) in three pots.

The synthesis (Scheme 2) was initiated by attempting to homologate aldehyde 5 to vinyl ether 6 via a Wittig-Levine²¹ reaction. The initial attempts provided poor yields (25%) of 6 due to reactivity of the Boc group of 6. To mitigate this, the order of addition was inverted, and the ylide solution was added slowly to the aldehyde. This improved yields to 65%. When monitoring this reaction, the betaine intermediate was observed to persist for several hours. We speculated that this intermediate was stabilized by lithium salts from n-BuLi, which could be causing undesired side reactivity. To address this, NaHMDS was utilized instead, which produced 6 in 93% yield. This reaction could be run on a 100-gram scale and could be purified via a simple silica gel plug.

Scheme 2.

Total synthesis of putative (+)-fumigaclavine f (1a) and its proposed biosynthetic precursor (2)

Using 6, we attempted to access 1a and 2 through three approaches (see the preliminary studies section of the Supporting Information). We abandoned our third approach after encountering difficulties with the key asymmetric intramolecular Mannich reaction (Scheme S6). We exclusively observed products resulting from addition of the aza-enolate into the ketone. We proposed that we could overcome this issue by using α/β-unsaturated ketone 4 instead of a ketone, which would be more likely to undergo the desired asymmetric intramolecular Mannich reaction because the carbonyl carbon is less electrophilic. This is due to electron-donating effects from the neighboring sp² π electrons, which provide electron density to the carbonyl through resonance and hyperconjugation into C-O π*. To efficiently prepare enone 4 through vinyl ether 8, conditions to couple diene 7 onto vinyl ether 6 were needed (Scheme 2). A screen of coupling conditions was performed (Table S6), including conditions published by Kapur et al.²². The screen revealed several conditions that accomplished the desired reaction in good yields. Similar to Kapur’s results, the use of bulky, electron rich phosphine ligands provided the best yields. This may be due to the electron-rich nature of the indole substrate, which requires an electron-rich catalyst to undergo oxidative addition.²³ However, our results contrast with Kapur’s in that use of CsF is not necessary and use of 1,1′-bis(di-tert-butylphosphino)ferrocene (DTBPF) and Pd(OAc)₂ did not promote the desired reaction. Our best method utilizes Pd(dba)₂/DTBPF and Bu₃SnF. To the best of our knowledge, this is the first example of a siloxydiene being coupled onto a heteroaryl species.

To prepare sulfinamide 4, the hydrolysis of 8 was attempted with both THF/aq. HCl, which produced a complex mixture, and with CeCl₃·7H₂O/NaI, which mostly provided undesired aldol reaction products. Next, we investigated milder conditions that combine the hydrolysis and sulfinamide condensation into a one pot process. The hydrolysis was conducted in acetonitrile, so that the water and HCl could be removed azeotropically. We hydrolyzed 8 in 3:1 MeCN:H₂O with 0.2 M HCl at 50 °C. Under these conditions, the reaction was completed in under 5 minutes. By way of contrast, the earlier conditions utilizing 0.4 M HCl in 3:1 THF:H₂O required several hours at 70 °C to consume 8. We suspect that the high polarity of MeCN stabilizes the oxocarbenium intermediate that is generated by protonation of the vinyl ether, and that exposes it to attack by water, thereby accelerating the hydrolysis. Our optimized reaction conditions are to add catalytic amounts of 2 M aq. HCl to a solution of vinyl ether (8) in MeCN, followed by stirring for 15 min., then concentration on a rotary evaporator at room temperature. After drying under high vacuum, (R)-Ellman’s sulfinamide was condensed onto the resulting aldehyde, providing 4 in one-pot and in 43% yield from 6. We then applied this hydrolysis condition towards the synthesis of S1, an intermediate in Cao’s formal synthesis of (+)-cycloclavine (Scheme S7).^{24, 25} This reduces the amount of pots to access (+)-cycloclavine from twelve to nine and improved the overall yield from 6.8% to 10.3%.^{24, 25}

We then focused our efforts on the key asymmetric intramolecular Mannich reaction to access fumigaclavine F (1a) and 2. To promote the key asymmetric intramolecular Mannich reaction, a variety of approaches were surveyed, including strong bases (Table 1), weak bases, Lewis acids, and Lewis acids in combination with weak bases ((Table S7 and Scheme S8 contain additional conditions screened),). The use of either LiHMDS (Table 1, entry 1) or NaHMDS (Table 1, entry 2) produced complex mixtures; the use of KHMDS (Table 1, entry 4) provided 9a in 35% yield (>20:1 d.r.) and 9b in 30% yield (>20:1 d.r.), which is formed by condensation of the aza-enolate onto the ketone. Use of KOtBu (Table 1, entry 5) provided a similar mixture. Using weak bases or both Lewis acids and weak bases (Table S7, entry 5–12) provided either 9b selectively or a complex mixture. Upon further investigation, the use of NaHMDS was shown to produce 9a (albeit in trace yields) without producing 9b. We hypothesized that the observed selectivity for 9a over 9b was due to a counterion effect. We speculated that the softer potassium cation may be large enough to coordinate both the nitrogen and the oxygen of the sulfinamide, which could stabilize the aza-enolate, resulting in a mixture of 9a and 9b. In addition, aza-enolates are softer than enolates, which could promote formation of the potassium-aza-enolate ion pair; both of these factors could result in the observed mixture. In comparison, sodium is a smaller and harder cation and thus may not provide the same stabilizing effects, which prevents formation of 9b. Because of the similar reactivity observed between KHMDS and KOtBu, the Mannich reaction was attempted with NaOtBu (Table 1, entry 3), a weaker base than NaHMDS. We were delighted to find 9a was produced selectively in 66% yield (95% b.r.s.m.), >20:1 d.r. on a 10g scale. With a scalable route to 9a, our attention turned toward forming the cis-fused D-ring. To this end sulfinamide 9a was hydrolyzed via treatment with catalytic HCl in 1% water/acetonitrile (Scheme 2). Next, a variety of bases were explored to promote the desired aza-Michael/epimerization cascade (including triethylamine, pyridine, N-methyl-imidazole, DMAP, NaHCO₃, KH₂PO₄, and K₂HPO₄). All of these conditions (except K₂HPO₄) either did not promote the desired aza-Michael reaction or formed a complex mixture. Interestingly, treatment with K₂HPO₄ provided the desired cis-tetracycle 3a and trans-epimer 3b in 78% yield (4:1 d.r.). We proposed two potential reaction pathways to account for the observed stereochemistry using model substrates (10a and 10b, acetyl substituted analogues of actual substrates) (Scheme 3A). In the first potential reaction pathway (pathway a), trans-tricycle 10a undergoes an intramolecular aza-Michael addition to form trans-tetracycle 11a, which subsequently epimerizes to cis-tetracycle 11b. Alternatively, in pathway b, trans-tricycle 10a first epimerizes to cis-tricycle 10b, positioning the amino group closer to the β-carbon of the enone, thereby making the aza-Michael addition geometrically feasible and accelerating the formation of cis-tetracycle 11b. To assess the kinetic and thermodynamic feasibility of each pathway involving the intramolecular aza-Michael reaction, we conducted density functional theory (DFT) calculations on a model reaction in methanol (Scheme 3B). As shown in Scheme 3B, trans-tricycle 10a is more stable than cis-tricycle 10b, as expected, with a free energy difference of 4.4 kcal/mol, indicating that the epimerization equilibrium favors trans-tricycle 10a. However, the free energy barrier for the intramolecular aza-Michael addition of trans-tricycle 10a (28.2 kcal/mol) is significantly higher than that of cis-tricycle 10b (17.8 kcal/mol), with calculated rate constants at 25 ºC of 1.3×10⁻⁸ s⁻¹ and 0.55 s⁻¹, respectively. This translates to a 50% conversion of cis-tricycle 10b to the aza-Michael addition (11b) product within milliseconds at room temperature, whereas the reaction with trans-tricycle 10a would take over two days. This rapid conversion of cis-tricycle 10b to the aza-Michael addition product would shift the epimerization equilibrium toward cis-tricycle 11b, ultimately favoring pathway b and accounting for the observed formation of the desired cis-tetracycle 3a as the major product.

Table 1.

Development of Mannich Reaction Conditions

Entry	Base	Yield of 9a[a]	Yield of 9b[a]
1	LHMDS	trace	0
2	NaHMDS	trace	0
3	NaOtBu	66 (95 b.r.s.m.)	0
4	KHMDS	35	30
5	KOtBu	33	30

^[a]isolated yields

Scheme 3. DFT calculations of the intramolecular aza-Michael reaction of the model system at the B3LYP/6–31+(d,p) level of theory in methanol, employing the Conductor-like Polarizable Continuum Model (CPCM).

(A) Scheme representing the chemical structures of the starting materials, transition states (TS), and products in pathways a and b. N-Boc substitution is replaced with N-acetyl (N-Ac) substitution for simplification. PR_trans(O-H) and PR_cis(O-H) correspond to the enol forms of 11a and 11b, respectively. (B) Calculated energy profile for the aza-Michael reaction. The energy values of all the species are depicted relative to 10a (cis-isomer).

To complete the synthesis of the proposed biosynthetic precursor of fumigaclavine F (2) (Scheme 2), ketone 3a was reduced with L-Selectride, then the amine was methylated via reductive amination to obtain alcohol 12 in 81% yield and >20:1 d.r.. Finally, the Boc group of 12 was cleanly removed by treatment with hexafluoroisopropanol (HFIP), which completes the total synthesis of the proposed biosynthetic precursor to fumigaclavine F (2) in 7-pots and 21% overall yield. The total synthesis of putative fumigaclavine F (1a) was completed by a one-pot Boc deprotection, chlorination, aza-Claisen, and elimination sequence on 12, which installed a reverse prenyl moiety at the 2-position in 67% yield (Scheme 2).¹⁹

Unfortunately, the NMR spectra of 1a did not match the literature report (See page S75).¹¹ Despite several protons having a close chemical shift to the literature report, many discrepancies exist in the aliphatic region. This provides strong evidence that these are two different compounds.

To confirm that we synthesized the correct structure, X-ray quality crystals of 2 were grown to ensure the accuracy of our work. Single crystal X-ray diffraction of 2 provided both the absolute and relative stereochemistry of 2 and 1a, matching the proposed structure of fumigaclavine F (1a) (Scheme 2). In addition, analysis of the 2-D NMR spectra of 1a is consistent with the structure of putative fumigaclavine F. This suggests that the structure of fumigaclavine F (1a) reported in the literature is incorrect. This is a common, but unfortunate conclusion that many others have encountered in the past.^26–36

In summary, we successfully expanded the scope of our asymmetric intramolecular Mannich reaction and applied it to the first total synthesis of putative (+)-fumigaclavine F (1a) and its proposed biosynthetic precursor (2), both in seven pots and in 16% and 21% yield, respectively. In addition, we completed the synthesis of an intermediate in Cao’s formal synthesis of (+)-cycloclavine (S10).^{24, 25} This reduces the total number of pots to access (+)-cycloclavine (S10) from twelve to nine (Scheme S7). We improved the yield of S1 from 51% to 84%, which corresponds to an overall yield of (+)-cycloclavine (S10) of 10.3% compared to 6.8% reported by Cao.^{24, 25} This makes this route the same length, but in higher overall yield, as the current shortest synthesis of (+)-cycloclavine by Wipf.³⁷ The concise and modular asymmetric syntheses reported herein provide an excellent platform to probe the bioactivity of clavine alkaloids.

Supplementary Material

ASSOCIATED CONTENT

Supporting Information

The Supporting Information is available free of charge on the ACS Publications website.

Experimental procedures, Preliminary studies, CIFs, and copies of NMR spectra (PDF) Accession Codes CCDC 2328515 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif or by emailing data_request@ccdc.cam.ac.uk, or by contacting The Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: +44 1223 336033.

Data Availability Statement

The data underlying this study are available in the published article and its Supporting Information.