Synthesis of the reported structures for kealiinines B and C

Abstract

Synthesis of the reported structures of kealiinines B and C have been executed. An intermolecular electrophile-induced cyclization of a pendant arene on an ene-guanidine affords the tetracyclic, oxidized naphthimidazole cores.

Marine sponges of the class Calcarea provide a large number of small molecule alkaloids containing a 2-aminoimidazole core with interesting biological activities (Figure 1).^1,2 For example, naamine D has been shown to be a competetive inhibitor of iNOS.³ Naamidine A has shown antimitogenic effects by inhibiting EGF-stimulated DNA synthesis in the growth of squamous tumor cells as well as cell cycle arrest in the G1 phase.^4,5 As a result, interest has emerged to incorporate this heterocyclic scaffold as a pharmacophore in biomedical research.

Figure 1.

Representative Calcarea alkaloids

Within this class of natural products are several oxidized variants, including 14-hydroxynaamidines A and G⁶ which are hydroxylated at C14 and naamidine F⁷ which is isolated as the A-ring quinone. 2-Deoxy-2 aminokealiiquinone⁸ is oxidized both in the A-ring and at C14.

In 2004, a chemical investigation of the marine sponge Leucetta chagosensis by Proksch and co-workers afforded the new imidazole alkaloids kealiinines A-C.⁹ Related to 2-deoxy-2-aminokealiiquinone, these natural products are oxidized at C14 by virtue of a new C-C bond, but remain at the hydroquinone oxidation state in the A-ring. Although their biosynthesis has not been studied, it is reasonable to suggest that the kealiinines are intermediates between kealiiquinones and naamine derivatives by ring closure and aromatization.^6,10 For example, naamines may be hydroxylated at C14 (Scheme 1, A). These intermediates might be subject to ionization to promote a Friedel-Crafts reaction (pathway 1). Oxidation would then provide the kealiinine skeleton. Alternatively, the hydroxylated intermediate may be subject to elimination (pathway 2) to give an intermediate which can undergo 6-π electrocyclization and oxidation to give the kealiinines. This biosynthetic strategy has also been explored in the laboratory by both Ohta and Lovely to access these naphthimidazole cores.¹¹

Scheme 1.

Plausible biomimetic strategy.

Our strategy for synthesizing kealiinines B and C was also inspired by these hypothetical reactive intermediates that can be generated from the oxidized naamine (Scheme 1, B). If a cyclic ene-guanidine were reactive toward an electrophile, the intermediate formed would be reactive with a pendant nucleophile leading to Friedel-Crafts type alkylation or elimination / 6-π electrocyclization. Access to these ene-guanidines was envisaged to be straightforward with the development of 5-exo-dig selective propargylguanidine hydroaminations recently reported by both our group and Van der Eycken’s.^12,13

Synthesis of the propargylguanidine precursors was initiated with a Cu(I)-catalyzed 3-component iminiumacetylide addition (A3-coupling)¹⁴ of 4-ethynylanisole, N-methylallylamine, and the aryl acetaldehydes 1 or 2 (Scheme 2).¹⁵ Although this reaction required elevated heating for extended reaction times due to the formation of stable enamine intermediates, 3 and 4 could be obtained in good yields on multi-gram scale. Deallylation of these intermediates with Pd(0) gave the secondary propargylamines 5 and 6.¹⁶ Guanylation of these intermediates with S-Me-N,N′-di-Boc-pseudothiourea and Hg(II) provided the propargylguanidines 7 and 8 in good yield.¹⁷

Scheme 2.

Substrate preparation.

We initiated our cyclization studies with the trimethoxy substituted arene for kealiinine C as it is symmetric (Scheme 2). To our surprise, cyclization of the propargylguanidine 7 gave 9 as a single regioisomer in good yield. Previous studies had shown that electron rich alkynes (e.g. pMeOPh-) typically gave poor 5-exo / 6-endo selectivity. Angle compression provided by substitution at the propargylic position presumably overrides this electronic selectivity through a Thorpe-Ingold effect. We next anticipated that activation of the ene-guanidine with NBS should generate the bromonium ion A. A key to this synthetic plan was that this bromonium ion can be opened via pathway a and not cyclization through pathway b to give intermediate B. While at first one might expect the formal products of a 5-exo cyclization (pathway b) to outcompete a 6-endo cyclization, there are several examples where the related electrophile-induced cyclizations are successful.¹⁸ Further, donation from the p-methoxy substituent might make this a formal 6-exo cyclization.¹⁹ In our experience the ene-guanidines are not very nucleophilic, suggesting that donation of the neighboring nitrogen lone pair is hampered by steric compression. This gave us hope that pathway a might be competitive.

To our delight treatment of 9 with NBS in acetonitrile provided the cyclized, fully aromatic and mono-Boc protected intermediate 10 in 56% isolated yield. The only other compound isolated from this reaction was brominated at C8 (circa 5%). While it does appear that the p-methoxyphenyl substituent on the ene-guanidine directs cyclization to give the 6-membered ring, we cannot rule out a 5-exo pathway followed by a pinacol rearrangement or that an elimination 6-π elctrocyclization is operative. Either way, any potential intermediates must funnel to intermediate C, which is capable of losing a Boc group and undergoing a 2 electron oxidation to give 10. Deprotection of this intermediate gave kealiinine C in good yield, and the structure was ultimately confirmed by X-ray crystallography.²⁰ In the isolation paper only ¹H-NMR data in CD₃OD and DMSO-d₆ were reported for both kealiinines B and C. Comparison of our spectra to that reported in DMSO-d₆ shared very little agreement. However the TFA salt of synthetic kealiinine C did match the natural product well in CD₃OD when referenced to the observed proton resonances for the natural product and not the solvent.

We next targeted kealiinine B to examine the regioselectivity of the oxidative cyclization relative to the A ring. Treatment of the propargyl guanidine 8 with AgOAc gave the ene-guanidine 11, also as a single regioisomer (Scheme 4). Exposure to NBS again delivered a single cyclized, fully aromatic and mono-Boc protected intermediate 12. This was presumed to be the correct isomer for kealiinine B as H8 was a singlet. Presumably attack of the bromonium ion is preferred by rotamer A as rotamer B would encounter more demanding steric interactions with the neighboring aromatic ring. Deprotection of 12 gave kealiinine B in 82% yield.

Scheme 4.

Synthesis of Kealiinine B

Again, the ¹H NMR data for synthetic kealiinine B did not match that reported for natural compound in DMSO-d₆. The free base did not match either, but it was a crystalline solid which permitted us to unambiguously assign the structure of our synthetic material via X-ray crystallography.²¹ Having now realized that kealiinine C only matched the natural material in CD₃OD, we examined the TFA salt of kealiinine B in CD₃OD and when aligned to the reported peaks there was excellent agreement for all proton resonances except for that at H6 which differed by 0.27 ppm.

Still concerned about the spectroscopic data, the Proksch laboratory was generous enough to compare our synthetic sample with a small amount of natural material by HPLC. This revealed two things: 1) that the natural sample was actually an inseperable mixture of kealiinines B and C and 2) the compounds did co-elute, suggesting they were indeed identical. We were also able to examine the original spectrum in CD₃OD for the natural products (a mixture of B and C) which revealed that H6 in kealiinine B is actually at 7.74 ppm not 7.90 ppm as originally reported.²² This is in good agreement with our synthetic material, and gave us confidence that the structures of the natural products are correct and agree nicely with our synthetic materials.

The fact that the spectra were still not in agreement in DMSO-d₆ was puzzling (most protons differing by 0.1 – 0.2 ppm). We hypothesized that this might be due to 1) a concentration effect, however dilution experiments did not reveal a change in the chemical shifts of the protons or 2) that the presence of both natural products in the sample might give rise to unique intermolecular interactions which change the chemical shifts, however titration of kealiinine C into a solution of B also showed no obvious change in chemical shifts. While we are confident, from the LC and ¹H NMR data in CD₃OD, that these are indeed the structures corresponding to the natural products, we must concede the possibility that alternate structures might exist that fit the DMSO-d₆ data with better agreement.

In the original isolation of kealiinines A-C only biological activity for kealiinine A was reported. This was reported in a brine shrimp lethality assay with an IC₅₀ = 20 μg / mL.⁹ The activity of kealiinines B and C were not reported, likely due to the small amounts of isolated material and the fact that they were isolated as an inseparable mixture. We evaluated our synthetic kealiinines B and C against several breast cancer cell lines, including a normal breast cell line (MCF-10A), a luminal ER+ cell line (MCF-7), an aggressive luminal ER+ line (T47D) and a “basal-like” triple negative cell line (MDA-MB-231). Kealiinine B showed antiproliferative activity against all of these cell lines with IC₅₀’s ranging from ~ 10–13 μM (Figure 2). Kealiinine C, however, did not appreciably inhibit proliferation at concentrations below 100 μM.

Figure 2.

Antiproliferative activities.

The antiproliferative effects of members of this family have most extensively been studied for naamidine A, which is known to cause apoptosis through a caspase-3 dependent cascade after activation of the MAPK pathway.²³ We examined the similarity of kealiinines B and C’s activity in this regard, and it appears that the antiproliferative effects arise from a caspase independent pathway (Figure 3). Treatment of MCF-10A or T47D cells, which are sensitive to caspase activation,²⁴ with 30 μM concentrations of kealiinine B or C for 48 hrs showed no evidence of caspase-3 cleavage or PARP cleavage, known hallmarks of caspase activation (Lanes 2 and 3). Due do limited amounts of natural naamidine A, staurosporine was used as a positive control,²⁵ and shows that indeed both cell lines are capable of undergoing caspase-3 dependent apoptosis (Lane 4).

Figure 3.

Interrogating Caspase 3 activity.

In conclusion, we have demonstrated that ene-guanidines, derived from the hydroamination of propargylguanidines, are useful intermediates for the synthesis of the reported structures for kealiinines B and C via an electrophilic activation / cyclization sequence. The ability to prepare these compounds individually has allowed us to provide independent and comprehensive characterization for these purported natural products, as well as an independent evaluation of their antiproliferative activity. Current efforts are directed to understand the mechanism by which kealiinine B inhibits cell proliferation.

Scheme 3.

Synthesis of Kealiinine C