Total synthesis and cytotoxicity of the marine natural product malevamide D and a photoreactive analog.

The marine natural product malevamide D from the cyanobacterium Symploca hydnoides was synthesized for the first time. The final peptide coupling linked the dolaisoleuine and dolaproine subunits. The phenyl group of malevamide D was also functionalized with a photoreactive diazirine moiety, which was carried through seven reaction steps. Comprehensive assessment of the cytotoxicity in a panel of 42 human cancer cell lines revealed a geomean IC70 value of 1.5 nM (IC50 0.7 nM) for malevamide D, whereas the photoreactive derivative proved to be less active by a factor of at least 200. COMPARE analysis indicated tubulin interaction as likely mode of action of malevamide D.

Introduction

The pan class="Chemical">depsipeptide pan class="Chemical">malevamide D (1, Figure 1) belongs to the pan class="Chemical">dolastatin class of marine natural products and has been isolated from the cyanobacterium Symploca hydnoides by Scheuer and co-workers in 2002 (7.5 mg, 0.014% of the dry weight) [1]. Malevamide D (1) was reported to exhibit in vitro cytotoxicity in the subnanomolar range (IC50 values about 0.7 nM). Closely related to malevamide D (1) is isodolastatin H (2), which is also active in vivo, but slightly weaker than dolastatin 10 (3) [2]. Although there has been no study regarding the mode of action, it can be assumed that malevamide D (1) acts similarly to dolastatin 10 (3) by inhibiting the formation of microtubuli from tubulin and, thereby, cell division. Dolastatin-type natural products continue to be tested in clinical trials for cancer therapy [3]. This includes an antibody conjugate of monomethylauristatin E (MMAE), which has been approved for the treatment of Hodgkin lymphoma (Brentuximab vedotin) [4].

Figure 1

Structures of the strongly cytotoxic marine natural products malevamide D (1), isodolastatin H (2), and dolastatin 10 (3).

Structures of the strongly cytotoxic marine natural products pan class="Chemical">malevamide D (1), pan class="Chemical">isodolastatin H (2), and dolastatin 10 (3).

In this paper, we report the first total synthesis of pan class="Chemical">malevamide D (1) and an assessment of its in vitro n>an class="Disease">cytotoxicity in a panel of 42 cell lines. We also synthesized a diazirinylated analog of 1, which, by photocrosslinking, might contribute to the understanding of the binding of 1 to tubulin.

Results and Discussion

Total synthesis of malevamide D

The natural products pan class="Chemical">malevamide D (1), pan class="Chemical">isodolastatin H (2), and dolastatin 10 (3) share two β-methoxy-γ-amino acid building blocks. The closest structural analog of malevamide D (1) is isodolastatin H (2) with an identical ester section composed of dolaproine and (2S)-3-phenylpropan-1,2-diol. On the N-terminal side, the isopropyl and two sec-butyl side chains of isodolastatin H (2) are replaced by one sec-butyl and two isopropyl side chains, respectively, in the case of malevamide D (1).

The total synthesis of 1 adopts the strategy by Yamada and co-workers in their total synthesis of 2, which sets the first retro cut at the tertiary pan class="Chemical">amide bond between pan class="Chemical">dolaisoleuine and dolaproine [2]. The tert-butyl ester 7 of Cbz-protected 3-methoxy-5-methyl-4-(methylamino)hexanoic acid (MMMAH) was obtained in multigram quantities within five steps from Cbz-valine (8), adapting a convenient sequence developed for dolaisoleuine by Pettit and co-workers [5]. Cbz-protected N-methylvalinol (4) [6] was oxidized to the aldehyde 5 under Parikh–Doering conditions, followed by aldol addition of the lithium enolate of t-BuOAc (Scheme 1). β-Hydroxyhexanoate 6 was obtained as a mixture of diastereomers (5:4), which was separated by column chromatography (silica) on a multigram scale. Treatment of (3R,4S)-6 with Meerwein's salt in presence of proton sponge afforded Cbz-protected MMMAH tert-butyl ester 7 in 87% yield. There are also other, enantioselective syntheses of 7 [7] and Boc-protected versions of MMMAH [8-9], but we intended to also have the (3S)-epimer of 7 at our disposal.

Scheme 1

Total synthesis of malevamide D (1). a) DMSO (16 equiv), NEt3 (5 equiv), pyridine·SO3 (5 equiv), 0 °C, 1 h, 91%. b) LDA (2.5 equiv), t-BuOAc (1.3 equiv), THF, −78 °C, 1 h; separation of diastereomers, (3R,4S)-6: 47%. c) 1,8-bis(dimethylamino)naphthalene (2.5 equiv), Me3OBF4 (2.6 equiv), 4 Å MS, 1,2-DCE, 0 °C, 2 h, rt, 16 h, 87%. d) H2, Pd/C (5%), MeOH, rt, 1 h. e) 8 (3 equiv), NEt3 (4 equiv), DEPCl (2 equiv), DCM, 0 °C, 2 h, rt, 16 h, 80% from 7. f) H2, Pd/C (5%), MeOH, rt, 90 min. g) 10 (3 equiv), NEt3 (4 equiv), DEPC (1.5 equiv), DCM, 0 °C, 2 h, rt, 16 h, 83% from 9. h) TFA, DCM, 0 °C, 90 min, quant. i) 13 (1 equiv), NEt3 (6 equiv), DEPC (1.3 equiv), DCM, 0 °C to rt, 16 h, 68%. j) TBAF·3H2O (5 equiv), THF/H2O 19:1, rt, 3 h, 71%. DEPCl: diethyl phosphorochloridate; DEPC: diethyl phosphorocyanidate.

Total synthesis of pan class="Chemical">malevamide D (1). a) pan class="Chemical">DMSO (16 equiv), pan class="Gene">NEt3 (5 equiv), pyridine·SO3 (5 equiv), 0 °C, 1 h, 91%. b) LDA (2.5 equiv), t-BuOAc (1.3 equiv), THF, −78 °C, 1 h; separation of diastereomers, (3R,4S)-6: 47%. c) 1,8-bis(dimethylamino)naphthalene (2.5 equiv), Me3OBF4 (2.6 equiv), 4 Å MS, 1,2-DCE, 0 °C, 2 h, rt, 16 h, 87%. d) H2, Pd/C (5%), MeOH, rt, 1 h. e) 8 (3 equiv), NEt3 (4 equiv), DEPCl (2 equiv), DCM, 0 °C, 2 h, rt, 16 h, 80% from 7. f) H2, Pd/C (5%), MeOH, rt, 90 min. g) 10 (3 equiv), NEt3 (4 equiv), DEPC (1.5 equiv), DCM, 0 °C, 2 h, rt, 16 h, 83% from 9. h) TFA, DCM, 0 °C, 90 min, quant. i) 13 (1 equiv), NEt3 (6 equiv), DEPC (1.3 equiv), DCM, 0 °C to rt, 16 h, 68%. j) TBAF·3H2O (5 equiv), THF/H2O 19:1, rt, 3 h, 71%. DEPCl: diethyl phosphorochloridate; DEPC: diethyl phosphorocyanidate.

After hydrogenation of 7, the resulting secondary pan class="Chemical">amine was coupled with n>an class="Chemical">Cbz-valine (8) affording dipeptide 9 (pan class="Chemical">DEPCl, diethyl phosphorochloridate), 80% over two steps, Scheme 1). N,N-Dimethylisoleucine (10), obtained by reductive dimethylation of isoleucine with aqueous formaldehyde and H2 on Pd/C [10-11], was appended at the N-terminus (DEPC, diethyl phosphorocyanidate) affording tripeptide tert-butyl ester 11. Treatment with TFA liberated the free acid 12, which was coupled with the trifluoroacetate of dolaproine ester 13, liberated from the N-Boc-protected analog [2].

With the subsequent coupling step affording pan class="Chemical">TBDPS-protected pan class="Chemical">malevamide D (14), we initially had difficulties. It proved to be crucial to remove excess of TFA from both coupling partners 12 and 13 completely, by repeated evaporation of DCM solutions. Finally, a coupling yield of 68% was reached following the DEPC protocol. Desilylation (TBAF in THF/H2O) afforded 1 (HRESIMS found 755.49257, calcd. for C40H68NaN4O8 755.49294) in 12% yield after 10 steps from valinol 4 (longest linear sequence). The optical rotation of the synthesized material ([α]D28 −37, c 0.09, MeOH) is a little smaller than reported for the isolated natural product ([α]D26 −55, c 0.10, MeOH) [1]. The 1H and 13C NMR spectra of 1 shows two sets of signals in a ratio of ca. 1:1. Clearly separated in the 13C NMR spectrum are, for instance, the ester carbonyl signals of the conformers (174.0 and 173.3 ppm), the carbonyl signals of the pyrrolidine amide (170.5 and 170.6 ppm), and all phenyl carbon signals. Of the eastern section, only three of the four pyrrolidine carbon signals are broad. Most of the carbon signals of the western tripeptide of malevamide D are broadened (see Supporting Information File 1). The 1H NMR spectrum shows many broad and overlapping signals, the assignment of which was only possible by careful interpretation of the 2D NMR data set. The only isolated signals belonging to different signal sets are located at δH 3.84 (one of the diastereotopic OCH2 signals) and 5.21 ppm (acylated carbinol). The two sets of signals are probably caused by the presence of both conformers of the acyl pyrrolidine. An assignment of the data sets to either of the two conformers was not possible. The occurrence of conformers has been mentioned by Scheuer and co-workers and is also known for dolastatin 10 [12] and symplostatin 1 [13].

As an alternative route to the N-terminal pan class="Chemical">tripeptide 11, we also investigated the coupling of the N-terminal pan class="Chemical">dipeptide 15 with pan class="Chemical">MMMAH tert-butyl ester (16, Scheme 2). However, after treatment of dipeptide 15 with secondary amine 16 (0.43 equiv) and DEPC (17, 2.5 equiv), the only isolable product was oxazolylphosphate 18 (35%) without any amide formation occurring. The structure elucidation of 18 required C–P coupling constant analysis, which showed doublet 13C NMR signals of oxazole C-5 (10 Hz) and C-4 (6.2 Hz). To the best of our knowledge, this is only the second time that a peptide-derived oxazol-5-ylphosphate has been characterized. Earlier, Boyd and co-workers had converted an oxazolone to an oxazolyl phosphate by treatment with excess diphenyl chlorophosphate [14]. In our case, a similar pathway is to be assumed with the oxazolone being formed first. The secondary amine 16 appears to react too slowly to be able to compete with oxazolone formation.

Scheme 2

Formation of oxazolylphosphate 18 on attempted DEPC-mediated coupling of dipeptide 15.

Formation of pan class="Chemical">oxazolylphosphate 18 on attempted n>an class="Chemical">DEPC-mediated coupling of dipeptide 15.

Diazirinyl-substituted malevamide D

Encouraged by the strong pan class="Disease">cytotoxicity of 1, we addressed the synthesis of a pan class="Chemical">photoactivatable analog. The question was whether such a derivative would still be cytotoxic. The phenyl ring of 1 was chosen as location of a trifluoromethyldiazirinyl substituent. Pettit and co-workers have shown that introduction of a pan class="Chemical">phosphate moiety at the p-position of the phenyl ring of auristatin PE did not lead to loss of cytotoxicity [3]. Recently, we have synthesized a hemiasterlin derivative with a diazirinylated indole moiety by incorporation of a thermally stable L-phototryptophan carrying a 1-azi-2,2,2-trifluoroethyl unit [15].

Racemic pan class="Chemical">photo pan class="Chemical">phenylpropanediol 25 was obtained starting from pan class="Chemical">aryl(trifluoromethyl)ketone 19 (Scheme 3), which itself was synthesized via Grignard monoallylation of 1,4-dibromobenzene, followed by dihydroxylation, dioxolane protection and trifluoroacetylation of the remaining brominated position [16]. Refluxing of ketone 19 with NH2OH·HCl in pyridine led to nearly quantitative hydrolysis of the dioxolane affording diol oxime 20 as 95:5 mixture of E- and Z-isomers. However, if solid NaOH was added to the reaction mixture and heated for several hours, the dioxolane survived, affording a 1:1 mixture of E/Z-oximes, which were tosylated after short work-up to (E)- and (Z)-22.

Scheme 3

Synthesis of tosyloximes (Z)-22 and (E)-22, X-ray structure of (E)-22. a) NH2OH·HCl (1.5 equiv), pyridine, reflux, 2 h, 80%. b) 2,2-dimethoxypropane (12.4 equiv), p-TsOH·H2O (5 mol %), DMF, rt, 1 h, 95%. c) p-TsCl (1.05 equiv), NEt3 (1.3 equiv), DCM, 0 °C to rt, 18 h, 83%. d) (i) NH2OH·HCl, NaOH, EtOH, pyridine, reflux (for details see Supporting Information File 1); (ii) p-TsCl (1.05 equiv), NEt3 (1.3 equiv), DCM, 0 °C to rt, 18 h, 74%.19F NMR chemical shifts (ppm) in CDCl3.

Synthesis of pan class="Chemical">tosyloximes (Z)-22 and pan class="Gene">(E)-22, X-ray structure of pan class="Gene">(E)-22. a) NH2OH·HCl (1.5 equiv), pyridine, reflux, 2 h, 80%. b) 2,2-dimethoxypropane (12.4 equiv), p-TsOH·H2O (5 mol %), DMF, rt, 1 h, 95%. c) p-TsCl (1.05 equiv), NEt3 (1.3 equiv), DCM, 0 °C to rt, 18 h, 83%. d) (i) NH2OH·HCl, NaOH, EtOH, pyridine, reflux (for details see Supporting Information File 1); (ii) p-TsCl (1.05 equiv), NEt3 (1.3 equiv), DCM, 0 °C to rt, 18 h, 74%.19F NMR chemical shifts (ppm) in CDCl3.

Reprotection of the pan class="Chemical">diol unit of 20 and tosylation of 21 afforded class="Chemical">n>an class="Chemical">tosyloxime (E)-22, which crystallized overnight from the neat mixture and was analyzed by X-ray crystallography [17]. Thus, it became possible to assign the 19F NMR chemical shifts of both isomers ((E)-22: δF −66.9; (Z)-22: δF −61.9).

Conversion of pan class="Gene">(E)-22 to pan class="Chemical">diaziridine 23 (colorless solid, δF −75.96, −75.93) was performed quantitatively (liquid pan class="Chemical">NH3 in t-BuOMe, Scheme 4). As expected, the NMR spectra of the colorless solid 23 showed two sets of signals. Oxidation of 23 with I2/NEt3 to diazirine 24 (δF −65.7) and acidic deprotection afforded diazirine diol 25. Figure 2 shows the DSC (differential scanning calorimetry) profile of 25, which melts at 54 °C and starts to decompose above 80 °C, indicating sufficient thermal stability for biological studies. Monosilylation of the primary hydroxy group of 25 afforded building block 26. Coupling with Boc-protected dolaproine (27, DCC/DMAP/CSA) yielded ester 28 as a mixture of two diastereomers. After deprotection of the pyrrolidine, coupling with tripeptide 12 afforded O-silylated diazirinyl-substituted malevamide D (29, 68%). Desilylation (TBAF, THF/H2O 19:1) led to a product mixture, the separation of which required semipreparative HPLC (RP-18, MeOH/H2O 9:1). Diazirinyl-substituted malevamide 30 was obtained in 33% isolated yield as a mixture of two epimers differing at the carbinol center (HRESIMS found 863.48644, calcd. for C42H67NaF3N6O8, 863.48647). In the NMR spectra, four sets of signals were observed because of additional cis/trans isomerism of the acyl pyrrolidine moiety. As in the case of 1, the tripeptidic western half showed mostly broad signals, whereas the signals of the phenyl carbons are resolved better, even if not fully separated.

Scheme 4

Synthesis of photo malevamide D 30. a) NH3(l), t-BuOMe, −40 °C, 2 h, rt, 16 h, quant. b) I2 (1.2 equiv), NEt3 (2.2 equiv), Et2O, 0 °C, 15 min, rt, 1 h, 99%. c) THF, 2 N HCl, rt, 3 h, 97%. d) TBDPSCl (1.1 equiv), imidazole (2.2 equiv), DMF, 0 °C, 1 h, 80%. e) 27 (0.97 equiv), DMAP (0.41 equiv), DCC (1 equiv), CSA (0.23 equiv), 0 °C to rt, 16 h, 69%. f) TFA, DCM, 0 °C, 90 min. g) 12 (1 equiv), NEt3 (6 equiv), DEPC (2.14 equiv), DCM, 0 °C, 150 min, 68%. h) TBAF·3H2O (5 equiv), THF/H2O 19:1, 0 °C, 2 h, 33%.

Figure 2

DSC curve of diazirine 25, heating rate 5 °C/min.

Synthesis of pan class="Chemical">photo pan class="Chemical">malevamide D 30. a) pan class="Chemical">NH3(l), t-BuOMe, −40 °C, 2 h, rt, 16 h, quant. b) I2 (1.2 equiv), NEt3 (2.2 equiv), Et2O, 0 °C, 15 min, rt, 1 h, 99%. c) THF, 2 N HCl, rt, 3 h, 97%. d) TBDPSCl (1.1 equiv), imidazole (2.2 equiv), DMF, 0 °C, 1 h, 80%. e) 27 (0.97 equiv), DMAP (0.41 equiv), DCC (1 equiv), CSA (0.23 equiv), 0 °C to rt, 16 h, 69%. f) TFA, DCM, 0 °C, 90 min. g) 12 (1 equiv), NEt3 (6 equiv), DEPC (2.14 equiv), DCM, 0 °C, 150 min, 68%. h) TBAF·3H2O (5 equiv), THF/H2O 19:1, 0 °C, 2 h, 33%.

DSC curve of pan class="Chemical">diazirine 25, heating rate 5 °C/min.

Cytotoxicity studies of malevamide D and diazirinyl-substituted malevamide D

The synthesized 1 proved to be potently cytotoxic with a geomean IC70 value of 1.5 nM (IC50 0.7 nM) in a panel of 42 pan class="Species">human n>an class="Disease">cancer cell lines (see Table 1 and Supporting Information File 1). The cytotoxicity was about 10-fold higher than that of paclitaxel. The breast cancer cell lines MAXF-401 (IC70 0.2 nM) and MCF-7 (0.5 nM), the colon cancer cell lines HT-29 (0.3 nM) and RKO (0.5 nM), the lung cancer cell line H460 (0.7 nM), the liver cancer cell line LIXF-575 (0.6 nM), and the renal cancer cell line RXF-393 (0.4 nM) all proved to be very sensitive. Against 36 of the 42 cell lines, IC70 values of below 10 nM were determined. The most resistant cell lines were shown to be RXF-1781 (kidney), CXF-269 (colon), MEXF-276 (melanoma, IC70 about 50 nM, each), and in particular LXFA-289 (lung, adenocarcinoma), PAXF-546L (pancreas), and PXF-1118 (pleuramesothelioma, IC70 > 100 nM, IC50 about 50 nM, each). Diazirinyl-substituted malevamide D 30 was about 200 times less cytotoxic than 1.

Table 1

Cytotoxicities (IC70, IC50, nM) of malevamide D (1) against selected human cancer cell lines.a

histotype	cell line	IC70	IC50
Colon	HT-29	0.3	0.2
Stomach	GXF-251	5.0	0.6
Lung, adeno	LXFA-629	0.7	0.4
Lung, large cell	LXFL-529	0.4	0.3
Breast	MAXF-401	0.2	0.2
Melanoma	MEXF-462	0.6	0.4
Ovary	OVXF-899	1.0	0.7
Pancreas	PAXF-1657	4.0	1.0
Prostate	PRXF-22Rv1	0.8	0.6
Mesothelioma	PXF-1752	0.4	0.3
Kidney	RXF-486L	0.7	0.5
Uterus	UXF-1138L	0.5	0.4

aFor complete results in the panel of 42 human cancer cell lines and a COMPARE analysis, see Supporting Information File 1.

pan class="Disease">Cytotoxicities (IC70, IC50, nM) of pan class="Chemical">malevamide D (1) against selected pan class="Species">human cancer cell lines.a

aFor complete results in the panel of 42 pan class="Species">human n>an class="Disease">cancer cell lines and a COMPARE analysis, see Supporting Information File 1.

To obtain clues on the likely mode of action of 1 and 30, the in vitro activity data were compared with those of 177 reference compounds by COMPARE analysis. Individual IC50 values of a test compound were correlated with those of 177 reference compounds as determined for Oncotest's 42 cell line panel and expressed quantitatively as Spearman correlation coefficients (see Supporting Information File 1). Furthermore, 1 and 30 were correlated to each other. For compound 30, COMPARE analysis revealed the highest similarities towards tubulin interacting compounds like the pan class="Chemical">vinca alkaloids n>an class="Chemical">vinfluine, vincristine, vindesine and the taxoid compound docetaxel. Furthermore, good correlations were detected towards PLK-1 inhibitors. Similar results were obtained for 1. Determining a Spearman coefficient of 0.742 between 1 and 30 suggested that these compounds share the same mode of action, most probably antimitotic activity based on tubulin interaction.

Conclusion

Our pan class="Disease">cytotoxicity test results for 1 confirm those obtained by Scheuer and co-workers who determined IC50 values of about 0.7 nM against four cell lines (pan class="Species">mouse pan class="Disease">lymphoma P-388, human lung carcinoma A-549, human colon carcinoma HT-29, human melanoma MEL-28) [1]. We had hoped that introduction of a Brunner-type diazirine unit in the para-position of the phenyl ring of 1 could lead to an analog that was still active, because it had been shown in several SAR studies that limited variation of the amine or alcohol component at the C-terminal amide or ester moieties of the dolastatin family members can be tolerated without much loss of cytotoxicity. This includes epimerization of the thiazole-carrying stereocenter of 3 [18], replacement of the phenyl by a methyl group, or removal of the thiazole ring [7]. Auristatins, of which some undergo clinical trials, are characterized by lacking the thiazole ring of 3 [19]. Pettit and co-workers prepared auristatin TP, which carries a phosphate unit in the para-position of the phenyl ring, and also two aminoquinoline derivatives that were as cytotoxic as dolastatin 10 [3]. Antibody conjugates of monomethylauristatin E functionalized on the C-terminal side have also shown promising pharmacological profiles [20-21]. However, compound 30 is at least 200 times less active than the natural product 1. Lower cytotoxicity may be associated with solubility problems. We will now search for more water-soluble, photoactivatable analogs of 1. Nunes and coworkers had synthesized a benzophenone-labelled derivative of the analog HTI-286, which competes with 3, and were able to identify α-tubulin as binding partner [22].

Procedures of synthesis and biotest, X-ray data and pan class="Chemical">1H, class="Chemical">n>an class="Chemical">13C NMR spectra of selected compounds.