Synthesis of modified miuraenamides – the Ugi approach

Miuraenamides, a group of marine cyclodepsipeptides, closely related to jasplakinolide and geodiamolide are found to accelerate nucleation and polymerization of actin


Introduction
Myxobacteria, highly interesting Gram-negative δ-proteobacteria, were initially considered as terrestrial microorganisms 1,2 but in the meanwhile a lot of marine representatives have been discovered during the last years. 3,4Their rich secondary metabolism makes them one of the best natural product producers, 5,6 and many of their metabolites are interesting candidates for drug development. 7n 2006, Lizuka et al. isolated a novel lightly halophilic myxobacterium, Paraliomyxa miuraensis, from a soil sample at the coast of Japan. 8Investigating the secondary metabolites produced by this species resulted in the isolation of miuraenamides A and B 8 and two years later also of the derivatives C-F (Figure 1). 9 The miuraenamides are structurally closely related to a range of other cyclodepsipetides, such as the chondramides, [10][11][12] also produced by myxobacteria or the geodiamolides, 13 seragamides 14 or jasplakinolide. 15,16ll these natural products have been isolated from sponges, but probably they are not produced by them but might also be secondary metabolites from bacteria living in them. 9Like the other cyclodepsipeptides the miuraenamides can be divided into two major building blocks: a tripeptide containing a halogenated tyrosine in the central position and a polyketide fragment.The derivatives mainly differ in the halogenation pattern of the tyrosine and the C-terminal unusual amino acid containing a methoxyacrylate (E or Z) motif.In miuraenamide F an additional hydroxylated polyketide is incorporated.Differences in this hydroxycarboxylic acid are also found, e.g. in the chondramides or geodiamolides, not only in the position of the double bond and the substitution pattern, but also the chain length.
By far most investigations concerning the biological activity and the mode of action were carried out with jasplakinolide (jaspamide), which shows high cytotoxicity towards a range of leukemia, breast and prostate cancer cell lines. 17Jasplakinolide was found to initiate actin polymerization and stabilizes already formed actin microfilaments, which causes significant morphological changes in the cell. 18,193][24] While the other natural products contain six to seven stereogenic centres, the miuraenamides A-C contain only three asymmetric Catoms (and one stereoisomeric double bond).Therefore, the miuraenamides might be ideal candidates for the development of potential anti-tumor drugs.

Scheme 1. Miuraenamides via late stage modifications.
In the miuraenamides the methyl-substituted double bond is at another position compared to all other cyclodepsipeptides (Figure 1).Therefore, we wanted to see what happens if we incorporate an "inverted" chondramide-type hydroxyacid with the same chain length.9][40][41][42][43][44][45] Although, these reactions should give mixtures of diastereomers, this is not a serious issue in this case, because one might expect that the different diastereomers should be separable, at least in the cyclized form.And in principle, all stereoisomers can be interesting candidates for SAR studies.
For the synthesis of the "inverted" polyketide fragment we started from the same building block 1, which was previously used during the synthesis of the miuraenamides.In this case 1 was converted into 2, which was coupled to the tripeptide and later on oxidized to the carboxylic acid which was subjected to cyclization (Scheme 2).Now, cleavage of the silyl protecting group, Swern oxidation and subsequent methyl Grignard addition gave rise to racemic secondary alcohol 4, which was subjected to enzymatic kinetic resolution.Both, the (S)-alcohol as well as the (R)-acetate 5 were obtained as pure enantiomer.Because we were only interested in (S)-4, (R)-5 was subjected to saponification, Mitsunobu reaction to (S)-5 and a second saponification provided finally additional (S)-4.Finally the methyl ester was converted into the corresponding allyl ester (S)-6 which could later on be cleaved without affecting the secondary ester.
The desired tripeptides 8 were obtained by Ugi reaction (Scheme 3).Besides phenylacetaldehyde (providing phenylalanine derivatives) also a range of aromatic aldehydes were used generating substituted aryl glycines (Table 1).Since arylglycines are known to be relatively sensitive to epimerization, we expected that the analogues cyclic depsipeptides might undergo "deracemization" towards the thermodynamically most stable stereoisomer, at least in some cases and the presence of base.This approach does not only allow the synthesis of N-methylated central amino acids, but may be used also for all kind on N-substituents.Exemplarily also an allyl substituent was incorporated.Using allyl amine as one component provided the desired product in high yield, although the reaction took two days to go to completion (entry 1).To speed up the process we carried out the further Ugi reactions at higher temperature in the microwave (µW), and most reactions were finished in 30-90 min.In most cases the yields obtained were only moderate, especially in case of the phenylacetaldehydes (entry 6), but the whole peptide 7 was obtained in only one step.All products were formed as almost 1:1 diastereomeric mixtures.Some of the derivatives could be separated by flash chromatography but the question was if this makes sense, because the next step was the saponification of the terminal methyl ester.As expected, the free tripeptide acids 8 were formed as a 1:1 mixture, independent if the ester was used in enantiomeric pure form or as stereoisomeric mixture, a clear indication for the configurational lability of the aryl glycine.Therefore, later on all esters 7 were saponified as diastereomeric mixtures.
Scheme 2. Synthesis of polyketide fragment (S)-6.Scheme 3. Synthesis of tripeptides via Ugi reaction.The tripeptide acids 8 were subsequently coupled with polyketide (S)-6 to 9 using the Steglich protocol (Scheme 4, Table 2). 46To be able to compare the new miuraenamides with the inverted polyketide directly with the natural products, we also coupled (S)-6 with the original tripeptide containing 4-allyloxy-3-bromotyrosine (8g).After cleavage of the allyl ester 10 with Pd(PPh3)4/morpholine the free acid 11 was activated as pentafluorophenyl ester according to Schmidt et.al. 47,48 The Boc-protecting group was removed with trifluoroacetic acid and the salt solution was added dropwise to a vigorously stirred suspension of CHCl3 and sat.NaHCO3 solution.Stirring overnight gave rise to the desired macrocyclic products 11.An indeed, our expectations became fulfilled, at least in part.While during the whole sequence the diastereomeric ratios of the tripeptides 7 and 8 were in all cases around 1:1, 11b was obtained as a 9:1 diastereomeric mixture.Surprisingly the ratio for the corresponding Nallyl derivative 11a was unchanged.Obviously, in case of 11b an equilibration occurred either during cyclization or afterwards.This effect was also observed with several other derivatives, although not in such a high ratio.In case of the halogenated allyl ethers 10c, 10d and 10g, both allyl protecting groups were removed simultaneously, and probably the free phenolic group caused problems in the cyclization step, especially for the bromo derivatives.In case of 10g the yield of the macrocycle was rather low, and the reaction of 10c resulted in complete decomposition of the substrate, while the corresponding chloro derivative 10d caused no significant problems.

AUTHOR(S)
Exemplarily, some of the tripeptides 8 were also coupled with the original polyketide precursor (S)-2 to figure out the influence of the side chain on the activity.Besides an unpolar naphthyl sidechain, also a protected chlorophenol was incorporated, which should be close to the natural substituent in miuraenamide C, missing only the benzylic carbon.In this case the allyl-protected phenol was subjected to cyclization, which occurred in excellent 71%, a further indication, that the free OH-group might be the critical functionality.Finally, the phenolic allyl protecting group of 14d was cleaved using CpRu(MeCN)3PF6 as a catalyst (Scheme 5). 49heme 5. Deprotected of allyl-protected cyclopeptide 14d.
The cytotoxicity of all new cyclic miuraenamide derivatives was investigated using two human cancer cell lines, HCT-116 (colon cancer) and U-2 OS (osteosarcoma).Some representative results are summarized in Table 3.For comparison, also the data for the simplified, highly active derivatives 16a/b (Figure 2, entries 6 and 7) are shown.Interestingly, all new derivatives showed a comparable cytotoxicity in the mid µM range, far away from the activity of the simplified miuraenamides 16.Although the bromo derivative 16a is significantly more active than the chloro analogue 16b, even this compound is around 1000-fold more active than the best new derivative (entry 1).Obviously modification at the central position of the tripeptide as well as in the polyketide fragment are not tolerated, independent if polar or unpolar groups are introduced.Even derivative 11g with the correct tripeptide was as inactive as all the others.Even if the same substitution pattern is used as in the miuraenamides, removal of the benzylic CH2-groups is also not accepted (entry 6 vs. 8).

Conclusions
In conclusion, we could show that Ugi reactions provide easily tripeptides which can directly be incorporated into new miuraenamide derivatives.Although the yields are moderate in certain cases, application of microwave heating speeds up the reaction significantly providing acceptable yields.Detailed SAR studies indicate that modifications at the C-terminus are well accepted, but not at the central position.Replacing the hydroxy acid by a geodiamolide-or chondramide-type polyketide fragment (inverted polyketide) is also not tolerated.All new derivatives synthesized showed comparable cytotoxicity in the mid µM range, which is at least 1000-fold lower than comparable simplified miuraenamide derivatives.Further investigations, especially regarding the binding mode, are in progress.

Experimental Section
General.All air-or moisture-sensitive reactions were carried out in dried glassware (>100 °C) under an atmosphere of nitrogen.Dried solvents were distilled before use: Dichloromethane was purchased from Sigma-Aldrich.The products were purified by flash chromatography on silica gel (0.063−0.(15  mL) was added in such a rate that a temperature of -68 °C was not passed over.5 Minutes after complete addition the alcohol 3 was added in dichloromethane (22 mL) and stirring was continued fur further 30 min.NEt3 (20 mL, 144 mmol) was added and the mixture was allowed to warm to room temperature overnight.The solution was hydrolyzed and the aqueous layer was extracted trice with dichloromethane and ethyl acetate (20 mL each).The organic layer was dried (Na2SO4) and after evaporation of the solvent in vacuo and flash chromatography (silica, hexanes:ethyl acetate 7:3) the corresponding aldehyde (5.33 g, 26.9 mmol, 93%) was obtained as pale yellow oil and directly used in the subsequent Grignard reaction.