BRD-6929

Total Synthesis of Trapoxin A, a Fungal HDAC Inhibitor from Helicoma ambiens

It has been known for several decades that post-translational reversible histone acetylation at the ε-amino groups of conserved lysins plays a major role in the regulation of gene expression.1 Enzymatic modulations of these processes are histone acetyltransferases (HATs), also known as writers, and their counterparts, the histone deacetylases (HDACs), known as erasers.2 While these enzymes are responsible for altering the histones through covalent modifications, bromodomains act as readers of the acetylation state and complement the epigenetic toolboX.3 The HDACs are involved in cellular pathways controlling all shapes and differentiation. Therefore, inhibitors of these enzymes are interesting candidates for the treatment of cancer. Four relatively simple compounds, such as SAHA (vorinostat),4 have recently been approved by the FDA for this purpose.

So far, 11 human Zn2+-dependent isoforms (HDAC 1−11) have been discovered,6 which can be subdivided into HDACs class I, IIa, IIb, and IV. Class IIa enzymes show less deacetylase activity compared to class I enzymes due to changes in the otherwise highly conserved active site.7

In addition, seven general does not allow specific binding toward the different isoenzymes, simply because the contact area is too small for differentiation. Luckily, a wide range of macrocyclic peptides have also been identified as naturally occurring HDAC inhibitors.9 For example, trapoXin, a fungal metabolite from Helicoma ambiens RF-102310 was found to cause accumulation of highly acetylated histones in various mammalian cell lines by irreversible inhibition of the deacetylating enzymes.11 In containing no Zn2+ in the active site and being NAD+- dependent.8 These class III enzymes differ significantly from the others and are not affected by inhibitors such as SAHA, containing typical zinc-binding motifs. SAHA is an example of a synthetic HDAC inhibitor, showing three important parts of HDAC inhibitors (Figure 1): (a) a zinc-binding motif, in general, a hydroXamic acid, (b) a linker (spacer), which simulates the lysine side chain of the natural substrate, and (c) a cap region, which should interact with the surface of the three parts: (a) the epoXy ketone as a Zn2+-binding motif, (b) a spacer, and (c) the peptide ring as cap region. Reduction of the epoXy ketone results in a loss of activity, indicating that the epoXy ketone probably acts as an irreversible inhibitor by binding covalently to the HDAC.10,11 In contrast, recent studies by Porter and Christianson found the epoXide moiety to be intact in a trapoXin A-HDAC6 cocrystal. The reported results indicate that trapoXin A acts as a mimic of the natural substrate after addition of water to the carbonyl function.12

The epoXy ketone side chain of the (2S,9S)-2-amino-8-oXo- 9,10-epoXydecanoic acid (Aoe), acting as a warhead, is not unique to trapoXin, but is also the essential motif in several cyclopeptides such as chlamydocin,13 HC-toXin,14 and many others.9 The different natural products differ in the amino acid sequence and their configuration in the tetrapeptide ring. They show different selectivity toward the various HDACs, indicating that the substitution pattern of the peptide ring plays a significant role in surface recognition and selectivity. But all natural HDAC inhibitors of this family show some common features. They all contain at least one (R)-amino acid, a (hetero)aromatic, and a secondary amino acid, such as proline or pipecolinic acid. Most of these compounds also nanomolar range, which makes them interesting candidates for the development of antitumor drugs.15 While several syntheses have been accomplished toward chlamydocin16 and the other representatives,9 including the unusual amino acid Aoe, to the best of our knowledge, only one synthesis of trapoXin B was reported by Schreiber et al. in the mid 1990s.17

Our group is involved in the synthesis of unusual amino acids and peptidic natural products, preferably with anticancer activity. Recently, we synthesized and evaluated tubulin binders such as pretubulysin18 and some actin binding cytostatics.16f,20 Our aim was to develop a rather flexible protocol which allows us to introduce substituents and modifications at several positions, e.g., for SAR studies. Since the trapoXins are by far less investigated, compared to other cyclopeptide inhibitors, and nothing is reported about their cytotoXicity profile, we decided to develop exemplarily a synthesis of trapoXin A, based on a stereoselective peptide enolate modification approach.21 This concept goes back to the work of Seebach et al. developed in the mid 1980s22 when his group had shown that rather complex natural products, such as cyclosporin, could be alkylated regioselectively at sarcosine subunits.23 While cyclic peptides generally give good diastereoselectivities, the stereoselectivity of the alkylation step with linear peptides is a serious issue.24 This is mainly caused by an unknown conformation of the polydeprotonated peptide formed under the basic reaction conditions. We could solve this problem by adding multivalent metal salts, probably resulting in the formation of a metal−peptide enolate complex,25 in which one face of the peptide enolate is shielded by the side chain of the adjacent amino acid. Generally, an (S)- amino acid generates an (R)-amino acid and the other way round. This effect is observed with intramolecular reactions, such as Claisen rearrangements,26 but also intermolecular ones, e.g., transition-metal-catalyzed allylic alkylations of peptides.27 We decided to apply a combination of stereoselective chelate− enolate Claisen rearrangement and diastereoselective allylic alkylation for the synthesis of the (R)-Pip-(S)-Aoe fragment A (Scheme 1). For stereoselective peptide enolate allylations, a N-terminal primary amine functionality is required, allowing deprotonation and peptide−metal complex formation. There- fore, a suitable precursor for the synthesis of B should be C with a functionalized side chain allowing subsequent cyclization toward the desired pipecolic acid moiety. This side chain can easily be introduced via asymmetric chelate Claisen rearrangement of chiral glycine ester D.28

The required chiral building blocks 1 and 2 (Scheme 2) for this approach could both easily be obtained from tartaric acids.Steglich’s protocol31 afforded 4 in high yield (Scheme 3). Subsequent ester enolate Claisen rearrangement with LDA and ZnCl2 proceeded readily and with good transfer of the stereogenic information via a chairlike transition state. The obtained amino acid 5 was directly coupled to afford dipeptide 6 via a miXed anhydride. In principle, modifications on the side chain can be performed on this stage. But for the synthesis of
trapoXin, hydrogenation of the double bond and simultaneous cleavage of the Cbz group with Pd/C afforded the free amine, which was directly TFA protected (7). This protecting group has been chosen because previous studies had revealed that the combination of TFA-protected N-termini with C-terminal tBu esters give the best diastereoselectivities in the allylation step.27 Therefore, dipeptide 7 was subjected to a Pd-catalyzed allylic alkylation using allyl carbonate 3, providing the allylated dipeptide 8 with good yield and diastereoselectivity in favor of the required (S)-configured Aoe precursor. Cleavage of the benzyl ether proved to be challenging and was accomplished only at pressures as high as 30 bar in THF for several hours. Finally, the corresponding alcohol was obtained in almost quantitative yield and could subsequently be subjected to a Mitsunobu cyclization, giving rise to dipeptide 9.

Saponification of the TFA group in 9 under standard conditions seemed to proceed, but isolation of the free amine protocol developed by Weygand and Frauendorfer to reduce the TFA-group with NaBH4 in ethanolic solution.32 The crude product was then coupled to Cbz-(S)-Phe-OH using TBTU as a coupling reagent. Subsequent cleavage of the Cbz protecting group and coupling to another Cbz-(S)-Phe-OH using the same protocol afforded linear tetrapeptide 11 in high yield (Scheme 4).

Unfortunately, all attempts to cleave the tBu ester in 11 progressive decomposition starting with TIPS deprotection. Therefore, we replaced the TIPS-protecting group by a tosylate through cleavage with TBAF and consecutive treatment with TosCl in pyridine (12). Installing the tosylate at this point of the synthesis diminished the number of transformations with the cyclic peptide, since the tosylate is used to build up the epoXide on a later stage. To our delight, cleavage of the tBu ester of 12 with TMSOTf proceeded without decomposition, but surprisingly, the yield of this reaction never exceeded 50% and the reaction suffered from poor reproducibility. We therefore decided to synthesize 13 by removal of the ketal, subsequent cleavage of the tBu ester, and reinstallation of the ketal in three steps without intermediate purification. This protocol proved to be more robust and convenient with higher yields compared to the previous one step transformation. With linear tetrapeptide 13 in hand, activation as the pentafluoro- phenyl (Pfp) ester proceeded readily and with excellent yields. Subsequent cyclization of this active ester 14 was not a trivial issue. Liberating the N-terminus by hydrogenation of the Cbz
group was a severe problem and failed under a wide range of reaction conditions. Finally, performing the reaction at 85 °C and passing through a continuous stream of H2 yielded 28% of the cyclic tetrapeptide 15 as a single diastereomer. With this compound in hand, we were able to finish the total synthesis of trapoXin A using an approach following Schreiber’s synthesis17 of trapoXin B. Acidic cleavage of the ketal with aq HCl afforded the corresponding diol, which was then subjected to a base- mediated epoXidation using DBU. EpoXy alcohol 16 was finally oXidized with Dess−Martin periodinane to give trapoXin A. The NMR spectra of our synthetic compound were in accordance to previously reported data of the isolated natural product.10

In conclusion, we have developed a straightforward protocol to synthesize the natural HDAC inhibitor trapoXin A. We could show that chelate enolate Claisen rearrangements, in combination with palladium-catalyzed allylic alkylations, are powerful tools for the construction of complex, nonproteino- genic amino acids. The developed protocol not only allows the synthesis of the natural product but should also be suitable for the synthesis of various derivatives by altering the allyl substrates in the peptide modification steps. Further investigations concerning the influence of different side chains on the biological activity and the selectivity toward the different HDAC isoforms are currently BRD-6929 under investigation.