Synthesis of the diaryl ether cores common to chrysophaentins A, E and F

Abstract

The synthesis of the diaryl ether subunits of the marine natural products chrysophaentin A, E and F is described. These natural prodcuts feature tetrasubstituted benzene rings with complex substitution patterns. The central strategy involves an S_NAr reaction between a complex phenol and a polysubstituted fluoronitrobenzene. Subseqent attempts to construct the unusual E-chloroalkene linkage through several different approaches are also disclosed.

The chrysophaentins comprise a family of biosynthetically unique natural products with interesting antimicrobial activity. In 2010, Bewley and co-workers reported the isolation of eight related secondary metabolites from the marine chrysophyte (aka “golden algae”) Chrysophaeum taylori.¹ The structures of the chrysophaentins (Figure 1) were assigned by NMR spectroscopy and mass spectrometry. Each member of this family has two benzylic E-chloroalkene subunits that are asymmetrically macrocyclic (A–D) 1, symmetrically macrocyclic (E) 2, or acyclic (F–H) 3. Although compound 4 has been prepared synthetically, it has not been isolated from the natural source, even though it would appear to be a biogenic precursor of several of the chrysophaentins.² Moreover, there are no similar natural products known and no biosynthetic hypothesis has been advanced to explain the origin of these unusual compounds. The only previous report of secondary metabolites from C. taylori does not describe any related chemical structures.³

Figure 1.

Structures of chrysophaentins A, E and F and “hemi-chrysophaentin,” synthetic subunit (not a natural product).

The chrysophaentins exhibit antimicrobial activity that is consistent with inhibition of bacterial cell division.^1,4 Several members exhibited low micromolar minimum inhibitory concentrations (MICs) against several Gram-positive organisms, including methicillin-resistant Staphylococcus aureus (MRSA). In addition, chrysophaentin A perturbed the polymerization of FtsZ, the central cell division protein in prokaryotes, and inhibited the GTPase activity of this protein. Preliminary NMR spectroscopy experiments paired with docking studies are also consistent with a direct interaction between chrysophaentin A and FtsZ.

Our research group has endeavored to synthesize and study natural products and other small molecules that perturb the function of FtsZ.^5–8 This protein has emerged as a new potential target for antimicrobial chemotherapy designed to fight infections that are resistant to current drugs.^9–14 Although small molecule inhibitors of FtsZ have been studied for two decades, there are still fundamental gaps in the basic understanding of the basic mechanisms by which inhibition occurs. To date, only a single X-ray crystal structure of a small molecule bound to FtsZ has been reported.^15,16 The activity of certain inhibitors is often limited to a protein from a single strain of bacteria or has been difficult to reproduce outside of the laboratories from which the studies originate. Importantly, many inhibitors, including several for which docking models have been proposed, have been demonstrated to act unselectively through aggregation.¹⁷ Among the natural products that have been reported as FtsZ inhibitors, two have been demonstrated to act through aggregation, one has structural features make it a suspect for aggregation,¹⁸ and one has activity that can’t be reproduced (viriditoxin).^17,19 Aside from these phenolic natural products, the phenanthridine alkaloids berberine^20–24 and sanguinarine²⁵ have also been reported as FtsZ inhibitor, but there is little evidence of a direct and selective interaction between these cationic alkaloids and the putative target. Collectively, these results demonstrate the stark difference between FtsZ and its eukaryotic homolog tubulin, for which three structurally unrelated natural products (Vinca alkaloids, colchicine, taxol) have all been demonstrated crystallographically to bind in distinct sites on the protein. ^26–28

We have initiated a program to synthesize chrysophaentin A in order to better understand the origin of its activity. First, the structural novelty of the chrysophaentins paired with the complete lack of biosynthetic understanding suggests that independent assignment of their structures would be helpful. Although there is no reason to suspect that they have been misassigned, arguments about their molecular interactions with FtsZ would be best understood if the structures were unequivocally established by crystallography or total synthesis. Second, a synthetic route would establish methods for the preparation of photo-affinity reagents, analogs for structure-activity relationship (SAR) studies, and potentially enable access to larger quantities of the compound for co-crystallization. In this communication, we describe a synthetic route to the diaryl ether cores common to chrysophaentins A, E and F as well as preliminary findings on an approach to the unusual E-chloroalkene subunit of these natural products.

The diaryl ether subunits of the chrysophaentins present a significant challenge due to the positioning of substituents on the benzene rings. The two isomeric subunits required for chrysophaentins A, E and F (1–3) are highlighted in blue and red, respectively, in Figure 1. A para relationship between carbon and Oxygen 5 is required for the D-rings of 1 and 3 as well as the B/B’ rings of 2 (Figure 2). The isomeric ortho relationship 6 is required for the C-ring of 1. We envisioned that each of these cores could arrive from similar synthetic routes, specifically either a Chan-Lam coupling^29,30 (as adapted by Evans³¹ for aryl ethers) or from an S_NAr reaction. The Chan-Lam route could arise from two possible combinations of a phenol and a boronic acid. Although two analogous S_NAr disconnections are possible, the combination shown would minimize the subsequent manipulations required and would avoid use of an aromatic electrophile that would be deactivated by two ortho substituents, one of which would be highly electron donating.

Figure 2.

Retrosynthetic analysis of the diaryl ether fragments of chrysophaentin A, E and F.

We first explored the Chan-Lam route, which would enable the most direct access to the requisite diaryl ethers. Initial results with model substrates were disappointing. The reaction of 10 with 11 under a variety of conditions yielded no detectable product 12 formation. Control experiments indicated that the presence of three ortho substituents was deleterious to this process. o-Chlorophenyl boronic acid 10 underwent smooth conversion to the corresponding ether 14 when combined with p-methoxyphenol 13 under standard conditions. Similar efficiency was observed with 2,6-dimethoxyphenol (11) when combined with phenyl-boronic acid 15 (Scheme 1).

Scheme 1.

Attempted Chan-Lam coupling reactions.

Pursuit of the S_NAr route commenced with the synthesis of the two requisite phenol isomers with the general structure of 5 and 6 (Figure 2). The choice of carbon substituent was important for the synthetic route to the phenol and for the subsequent manipulations that would eventually install the E-chloroalkene subunits of the targets. We settled on an allyl group in order to enable access to a variety of potential coupling partners with an eye toward employing an olefination with concomitant installation of the chloride. 2,4-Dihydroxybenzaldehyde 17 was isopropylated³² in high yield and then converted to phenol 19 using magnesium monoperoxyphthalate (MMPP) to effect a Dakin oxidation.³³ This intermediate could be allylated³³ and then heated to induce Claisen rearrangement to phenol 21 (Scheme 2). This robust route was routinely executed on large (>25g) scale. The para-substituted isomer was prepared by a similar route starting with resorcinol 22. After isopropylation,³⁴ a sequence of ortho-metallation, borylation and oxidation installed the requisite central hydroxyl group of 24. Allylation and tandem Claisen and Cope rearrangements^35,36 proceeded (with similar efficiency as seen with 19) to yield phenol 26 in 81% yield over two steps (Scheme 2).

Scheme 2.

Synthesis of phenols 21 and 26.

The requisite electrophile for the S_NAr reaction required extensive exploration and optimization. The 1,2,4,5 substitution pattern defies many traditional sequences of benzene functionalization. We initially explored several routes that would install the carbon substituent in a regioselective manner, i.e. para to the chlorine. These included Claisen rearrangements and the lesser-known vicarious S_NAr substitution reaction,^37,38 which installs enolate-like nucleophiles ortho to nitro groups (not shown). The failure of these routes prompted us to explore substrates with carbon installed first, which ultimately lead to nitration of a suitable precursor. Although fluorine is mildly de- activating for electrophilic aromatic substitutions, this halogen is known to induce high levels of para selectivity.^39–43 Benzyl bromide 27 was treated with cyanide and hydrolyzed to phenylacetic acid 29 (Scheme 3). This intermediate could be nitrated in high yield with high para selectivity, provided fuming HNO₃ was employed. Reduction and protection as the TBS ether provided fluoronitrobenzene 31. This five-step sequence could be completed on multigram scale in 63% overall yield with a single chromatographic separation. Treatment of 31 with phenol 21 or 26 and potassium carbonate produced diaryl ethers 32 and 33 in 78% and 75% yields, respectively (Scheme 3).

Scheme 3.

Synthesis of S_NAr substrate 31 and reactions with 21 and 26.

Conversion of the nitro group to a protected hydroxyl group proved to be a significant challenge. Direct hydrolysis with base was unsuccessful. The nitro group of 32a was reduced to the corresponding aniline 32b in high yield under various conditions (Scheme 4). Although Sandmeyer reactions via the diazonium intermediate to either an ether or phenol 34a or 34b were also fruitless, conversion to the iodide 32c occurred in high yield. Under aqueous conditions, the TBS ether remained intact and under the anhydrous conditions of Knochel,⁴⁴ the TBS ether was cleaved. Many attempts were made at conversion of the iodide to the isopropoxy and methyl ethers under copper-mediated conditions. In the case of 32c, cyclization to benzodihydrofuran 35 was observed. Although this intermediate could not easily be converted to the required acyclic structure, we were prompted to oxidize 32c to the corresponding acid 36 to explore lactone formation. Unfortunately a variety of conditions failed to convert 36 into lactone 37 (Scheme 4).

Scheme 4.

Attempted nitro-to-hydroxyl conversion.

Ultimately the iodide intermediate was employed in a successful nitro-to-hydroxyl conversion sequence. Iodide 32c was re-silylated and then metallated using recently-reported conditions that employ the lithium chloride complex of ispropylmagnesium chloride, aka “turbo-Grignard.”^45,46 Trapping of the organomagnesium intermediate with tri-isopropylborate and subsequent oxidation yielded phenol 34a in high yield in this one-pot process (Scheme 5). Protection of the phenol as the isopropyl ether once again removed the silyl group, which was subsequently re-installed to give ether 34b. An analogous sequence converted 33a to 33d with similar efficiency. Although this route provides access to the required intermediates with high levels of efficiency for the individual steps, it suffers from significant repetitive changes in oxidation state.

Scheme 5.

Nitro-to-hydroxyl conversion to prepare fragments 34b and 33d.

With reliable routes to the diaryl ether fragments in hand, we turned our attention to the construction of the E-chloroalkene 38a. Although the synthesis of alkenes with similar substitution patterns have been reported,⁴⁷ E-configured chloroalkenes with two benzylic carbons are unknown in the literature outside of the chrysophaentins. In addition, the quasi-symmetry, i.e. similar steric demands of the two benzylic substituents, suggests that hydrometallation reactions employing boranes⁴⁸ or Schwartz’s reagent⁴⁹ would not be regioselective. Alkene cross-metathesis in the synthesis of chloroalkenes is limited to vinyl chloride.⁵⁰ With these limitations in mind, we considered several approaches involving olefination of an aldehyde precursor. The most straightforward approach would involve a chloro-substituted phosphorous ylide 38b, for which there was precedent.^51,52 This approach is complicated by the high enol content of phenyl acetaldehydes 38c, which makes these substrates particularly troublesome for olefination reactions (Figure 3). The combined basicity and steric bulk of 38c led to degradation of the aldehyde, presumably through aldol addition and condensation reactions. As such we investigated two possible alternatives to reduce steric bulk and basicity in the olefination component. First, we investigated a Wittig reaction,⁵³ which would use a stabilized anion 40 with reduced basicity. The product of this reaction could later be converted to the E-chloroalkene by the Hunsdiecker decarboxylation-halogenation reaction⁵⁴ or related processes reported by Barton.⁵⁵ Second, we explored Schlosser’s one-pot approach to trisubstituted alkenes by the SCOOPY sequence,⁵⁶ which employs an unsubstituted Wittig reagent 41.

Figure 3.

Retrosynthetic analysis of E-chloroalkene subunit.

The necessary aldehyde and phosphorous ylide components were prepared from 34b and 33d, respectively. Hydroboration/oxidation of 34b produced alcohol 42, which was carried through to the analogous tert-butyl ester 43 by a three-step oxidation and esterification sequence. Cleavage of the TBS ether and oxidation yielded aldehyde 44 (Scheme 6). A similar sequence was initially pursued for the ylide in which the ester would be halogenated and displaced with triphenylphosphine. The failure of the requisite α-halogenation led us to a route based on alkylation of a pre-formed ylide. The allyl group of 33d was isomerized with palladium(II) chloride, unexpectedly cleaving the silyl ether, which had to be re-installed. One-pot oxidative cleavage of alkene 45 to the aldehyde was followed up with reduction to produce benzylic alcohol 46. Conversion of this intermediate to the iodide proceeded smoothly. Alkylation of the iodide with the phosphonium ylide 47 yielded coupling partner 48, which was used without purification after a quick work-up to remove the base. Unfortunately, treatment of this ylide with aldehyde 44 produced none of the desired alkene product 49. As was the case with unstabilized ylides, complete destruction of the aldehyde was observed, presumably by the aforementioned self-aldol processes for which phenylacetaldehydes are notorious (Scheme 7). This sequence demonstrated that reducing the basicity of the ylide partner was not sufficient to suppress aldehyde decomposition.

Scheme 6.

Synthesis of aldehyde 44 for olefination.

Scheme 7.

Synthesis of ylide 48 for olefination.

Our final attempt to install the E-chloro alkene subunit involved exploration of Schlosser’s SCOOPY method.⁵⁶ This process relies on deprotonation of the betaine intermediate in a Wittig reaction and allowing this anion to react with an electrophile before the final elimination that results in alkene formation. In order to enable thorough exploration of this route, we prepared substrates to serve as model system. Chlorohydroquinone 50 was isopropylated and formylated to benzaldehyde 51.^57,58 Two-step homologation provided aldehyde 54.⁵⁹ This material proved to difficult to store and handle and was in later runs, reduced to the alcohol 53 and oxidized with DMP as needed (Scheme 8). The ylide partner was prepared from Claisen-Cope product 26 by protection as the ethoxymethyl (EOM) ether, oxidative cleavage, reduction and bromination. Subsequent displacement with triphenylphosphine provided the triphenylphosphonium salt 56 in modest yield. This intermediate was deprotonated with phenyllithium (PhLi), treated with aldehyde 54, then deprotonated again with PhLi and treated with electrophilic chlorine reagents. In no case was any of the desired haloalkene observed and, as with previous attempts at olefination, the aldehyde decomposed (Scheme 8). Although we had hoped that addition to the aldehyde at lower temperature by the ylide of 56 would be fast enough to avoid aldehyde decomposition, it appears that the enolization problem represents a fundamental flaw in our synthetic plan.

Scheme 8.

Attempted SCOOPY route to E-chloroalkene 57.

In summary, we have successfully assembled the diaryl ether subunits common to chrysophaentins A, E and F. The routes are high-yielding and easily scaled up to multigram quantities, in spite of a cumbersome nitro-to-hydroxyl conversion that relies on a long sequence of atom-transfer reactions. We have also thoroughly explored three olefination routes to the E-chloroalkene subunit of these unique natural products. In each case, the enolizability of the required phenylacetaldehyde prevented alkene formation. We are continuing to explore more efficient diaryl ether assembly sequences as well as alternate E-chloroalkene-forming processes that eschew the problematic olefination reactions of phenylacetaldehydes.