Herman
Nikolayevskiy
a,
Maung Kyaw
Moe Tun
a,
Paul R.
Rablen
b,
Choukri
Ben Mamoun
c and
Seth B.
Herzon
*ad
aDepartment of Chemistry, Yale University, New Haven, CT 06520, USA. E-mail: seth.herzon@yale.edu
bDepartment of Chemistry and Biochemistry, Swarthmore College, Swarthmore, PA 19081, USA
cDepartment of Internal Medicine, Yale School of Medicine, New Haven, CT 06520, USA
dDepartment of Pharmacology, Yale School of Medicine, New Haven, CT 06520, USA
First published on 4th May 2017
Ocimicide A1 (1) and the semisynthetic derivative ocimicide A2 (2) are highly potent antimalarial agents efficacious against chloroquine-sensitive and -resistant Plasmodium falciparum strains with IC50 values in the nanomolar and picomolar range, respectively. Members of this family have demonstrated radical cure in rhesus monkeys, without detectable toxicity, but their structure–function relationships and mechanism of action are unknown. Herein we describe a twelve-step synthesis of an advanced N-acylated pentacyclic precursor to the proposed structure of 1 (11% overall yield). Instability and poor P. falciparum growth inhibition of the corresponding free donor–acceptor cyclopropylamine, and large discrepancies between reported and both experimental and DFT-calculated 13C chemical shifts and coupling constants, suggest that substantial revision of the proposed structures may be necessary.
Fig. 1 Structures of the alkaloids ocimicides A1 (1) and B1 (3), the semisynthetic derivatives ocimicides A2 (2) and B2 (4), and chloroquine (5). |
With resistance to the front-line antimalarial artemisinin increasing,3 there is a pressing need for the development of novel agents with unique modes of action.4 We initiated synthetic studies toward ocimicide A1 (1), with the goal of elucidating the structure–function relationships and mechanism of action of this new class of antimalarials.
Since E-ring substitution influences activity,1 we targeted late-stage construction of this ring from the pentacyclic intermediate 6 (Scheme 1). A tandem epoxide-opening–ring contraction of 8, followed by activation and invertive displacement of the alcohol 7, was used to install the aminocyclopropane5 and lactam substituents of 6. The epoxide 8 was simplified to the vinyl triflate 9 and the stannane 10. The vinyl triflate 9 was prepared in racemic form in four steps and 57% overall yield from 4-methoxypyridine by a sequence developed by Comins6 for closely-related substrates (see ESI†). The stannane 10 was prepared in one step and 70% yield by site-selective metalation7 of 2-cyano-6-methoxyquinoline8 with lithium tetramethylpiperidide, followed by the addition of trimethyltin chloride (see ESI†).
Stille coupling of the vinyl triflate 9 and the stannane 10 [tetrakis(triphenylphosphine)palladium, copper iodide, cesium fluoride]9 provided the coupling product 11 in >99% yield (Scheme 2A). Site-selective oxidative cleavage of the exocyclic alkene within 11 [osmium tetroxide, then bis(acetoxy)iodobenzene],10 followed by in situ oxidation under Pinnick–Lindgren conditions11 provided the acid 12. The carboxylic acid function was employed in a carefully choreographed multistep relay to control the relative stereochemistry of the target. Treatment of 12 with N-bromosuccinimide and 4-dimethylaminopyridine provided the bromolactone 13 as a single regioisomer. The bromolactone 13 could be isolated, but in practice was treated directly with potassium carbonate in methanol to induce a ring-opening–ring-closing sequence, to form the epoxy ester 8 (45% from 11). The remaining mass balance was attributed to formation of a cyclic imidate resulting from addition of the alkoxide intermediate to the nitrile.
Extensive experimentation was required to develop conditions to effect the epoxide-opening–ring contraction sequence (Scheme 2A).12 Ultimately, we found that treatment of 8 with lithium hexamethyldisilazide (1.10 equiv.) in toluene at 103 °C provided the cyclopropyl alcohol 7 in 44% yield (+10% unreacted 8). The temperature profile of this step was critical; although deprotonation occurred at 24 °C (as evidenced by deuterium incorporation experiments), heating to 103 °C for short amounts of time (2 h) was essential to obtain the product 7 in workable yields. Higher temperatures and longer reaction times led to degradation.
Density functional theory calculations [M062X/6-31+G(2df,p)] suggest that the enolate is required to adopt a boat-like conformation to effect invertive opening of the epoxide (see 14). This conformation orients the quinoline ring in a pseudoaxial position, leading to destabilizing non-bonded interactions in the transition state. The calculated activation energy of 27.9 kcal mol−1 for this step is in agreement with the experimentally-determined reaction temperature (103 °C).
The synthesis of 6 was completed by a high-yielding cascade sequence. Exposure of the cyclopropyl alcohol 7 to methanesulfonyl chloride and triethylamine provided the mesylate 15. Following concentration of the reaction mixture, the unpurified mesylate 15 was dissolved in anhydrous methanol and treated with sodium methoxide to effect 1,2-addition of methoxide to the nitrile and invertive displacement of the mesylate (15 → 16). Concentration of the reaction mixture, followed by dilution with sulfuric acid, resulted in smooth hydrolysis of the methyl imidate 16 to provide the key pentacycle 6 (96% from 7). The facile addition of methoxide to 15 is likely a reflection of electrophilic activation of the nitrile substituent by the quinoline ring. By this route, the racemic lactam 6 was obtained in only five steps, four isolated intermediates, and 19% yield from the Stille coupling partners 9 and 10 (twelve steps and 11% yield overall from commercial reagents).
The structural assignment of 6 was confirmed by X-ray analysis of the crystalline Weinreb amide 18,13 which was prepared by saponification, activation of the resulting acid 17 with methanesulfonyl chloride,14 and addition of N-methoxy-N-methylamine (Scheme 2B; 83% overall). While 6, 18, and related N-acylated intermediates were amenable to purification and handling, the corresponding free amines were unstable and underwent rapid decomposition. For example, attempted neutralization of the trifluoroacetate salt 19, formed by exposure of 6 to trifluoroacetic acid, led to extensive decomposition (Scheme 3). Although the complexity of the decomposition mixtures precluded characterization, we believe that the conjugation of the secondary lactam through the electron-deficient quinoline ring effectively renders the cyclopropane within 19 a donor–acceptor system.15 Decomposition may occur by amine-initiated ring-opening.
The instability of our synthetic intermediates prompted us to reexamine the original structural assignment using density functional theory.16 As spectroscopic data for 1 were not disclosed,1 we focused on the dimethyl ether derivative ocimicide A2 (2), for which tabulated spectroscopic shifts were presented. Following the protocol of Hoye and co-workers,17 32 structures (corresponding to all possible diastereomers at nitrogen 15 and carbons 12, 13, 14 and 17 of 2,18Fig. 2) were generated. These structures were imported into BOSS19 and each was separately subjected to a conformational search. Conformers within 5.02 kcal mol−1 of the lowest energy isomer (8–30 conformers for each diastereomer) were advanced to density functional theory geometry optimization [gas phase, B3LYP/6-31+G(d,p)]. Geometry-optimized conformers were confirmed as real local-minima by the absence of imaginary frequencies. The chemical shifts of the optimized conformers were then calculated using a modified WC04 functional20 and 6-31G(d) basis set in methanol (an initial screen of several functional and basis set combinations indicated that this level of theory provided an acceptable compromise of computational time and accuracy). Finally, the 13C chemical shifts were Boltzmann-averaged to generate the heat maps shown in Tables S1 and S2.† The geometry optimization and NMR calculation methods selected were benchmarked against 19-(E)-hunteracine (20),21 the structure of which has been unequivocally established by X-ray analysis.22 The mean absolute error (MAE) for 20 was 2.8 ppm [absolute error (AE) = 0.2–5.7 ppm], which was within the error typically obtained from similar levels of theory under optimized conditions.23
Fig. 2 Mean absolute error (MAE) and absolute error (AE) of calculated 13C chemical shifts for the reference compound 19-(E)-hunteracine (20), ocimicide A2 (2; numbered as in patent1), the trifluoroacetate salt 21, and all ocimicide diastereomers (varied configuration at carbons 12, 13, 14, 17, and nitrogen 15). Geometries were optimized using B3LYP/6-31+G(d,p). 13C NMR chemical shifts were calculated using modified WC04/6-31G(d). Data highlighted in blue corresponds to the pentacyclic core. |
Large deviations, particularly in proximity to the lactam (carbons 9 and 20; AE = 7.3–13.1 ppm) and pyrrolidine rings (carbons 13, 14, 22, and 23; AE = 8.3–20.9 ppm), were observed between the calculated and reported113C chemical shifts for ocimicide A2 (2) (MAE = 5.6–6.0 ppm). While 17 of the 30 diastereomers investigated achieved better agreement (MAE = 4.9–5.5 ppm), no diastereomer could, within acceptable error, replicate reported values at carbons 14, 22, or 23 (AE > 6.6 ppm).
For comparison, the trifluoroacetate salt 21 was synthesized (via treatment of 18 with trifluoroacetic acid) and computationally subjected to analogous geometry optimization and NMR calculation methods.24 A focused analysis of pentacyclic core MAE values for 2, 21, and alternate ocimicide diastereomers revealed that while 21 was comparable in error to the benchmark 20 (2.9 ppm vs. 2.8 ppm), ocimicide diastereomers (including 2) had, on average, higher MAEs (4.3 ppm for 17R; 3.8 ppm for 17S). Moreover, calculated AE values for carbons 9 and 20 were significantly lower for 21 than any ocimicide diastereomer (0.3–1.8 ppm vs. 5.1–13.8 ppm). These data suggest our computational methods achieved an acceptable level of accuracy in this system.
The rigidity of the azabicyclo[3.1.0] ring system present within 6 and 2 suggests that a reasonable comparison of coupling constants may be made between the methine proton of carbon 17 and the methylene protons of carbon 16. For this three-proton spin system, the J values observed in 6 (5.2 and 0 Hz) poorly match those reported for 2 (9.2 and 4.4 Hz). For comparison, coupling constants were calculated for all diastereomers with the reported 17R and the alternate 17S configuration. Interestingly, while J values for the reported diastereomers (3.3–4.1 Hz and 0.1–0.3 Hz) closely resemble the experimental J values of 6, the reported data for 2 is better approximated by the alternate diastereomer (7.6–8.1 Hz and 5.7–6.7 Hz). While not definitive, the computational studies described herein suggest that a structural revision of the ocimicides is required.
Finally, to probe the antimalarial activity of our synthetic intermediates, we examined the growth inhibitory potential of several compounds against the P. falciparum 3D7 isolate (Table 1). While the bromolactone 13, the epoxy ester 8, and the cyclopropyl alcohol 7 inhibited parasitic growth by 20–28% at 100 nM, more advanced synthetic intermediates were significantly less active. The pentacyclic lactam 6 completely failed to inhibit parasitic growth, while the amine 19 (Scheme 3) demonstrated only 15% inhibition at 500 nM. These data and in particular the low activity of 6 and 19 provide some circumstantial support for a structural revision of the metabolites.
Compound | % inhibition, 100 nM | % inhibition, 500 nM |
---|---|---|
13 | 26 | 40 |
8 | 20 | 30 |
7 | 28 | 41 |
6 | 0 | 0 |
19 | 8 | 15 |
Footnote |
† Electronic supplementary information (ESI) available: Supplementary schemes, figures, and tables, general experimental remarks, synthetic procedures, catalogues of experimental and calculated nuclear magnetic resonance spectra. CCDC 1536046. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c7sc01127j |
This journal is © The Royal Society of Chemistry 2017 |