Letter Cite This: Org. Lett. 2019, 21, 508−512
pubs.acs.org/OrgLett
Total Synthesis of Covalent Cysteine Protease Inhibitor N‑Desmethyl Thalassospiramide C and Crystallographic Evidence for Its Mode of Action Jeremy Fournier,† Karen Chen,† Artur K. Mailyan,† Jeffrey J. Jackson,† Brad O. Buckman,‡ Kumar Emayan,∥ Shendong Yuan,∥ Ravi Rajagopalan,‡ Shawn Misialek,‡ Marc Adler,§ Michael Blaesse,⊥ Andreas Griessner,⊥ and Armen Zakarian*,† †
Department of Chemistry and Biochemistry, University of California, Santa Barbara, California 93111, United States Blade Therapeutics, 442 Littlefield Avenue, South San Francisco, California 94080, United States § ChemPartner, 280 Utah Avenue Suite 100, South San Francisco, California 94080, United States ∥ Bayside Pharma, 2600 Hilltop Drive, Richmond, California 94806, United States ⊥ Proteros Biostructures GmbH, Bunsenstraße 7 a, 82152 Planegg-Martinsried, Germany
Org. Lett. 2019.21:508-512. Downloaded from pubs.acs.org by UNIV DE BARCELONA on 01/18/19. For personal use only.
‡
S Supporting Information *
ABSTRACT: A total synthesis of N-desmethyl thalassospiramide C, a unique strained macrocyclic proteobacterial depsipeptide, enabled a detailed crystallographic study of its covalent complex with cathepsin K, a member of a medicinally important family of cysteine proteases. The study provides support for the mechanism of action, and the insight gained can be used for structure-based drug design targeting these calpain proteases.
I
thalassospiramide C complex with cathepsin K. The results provide support for its mode of action as a covalent Michael acceptor with macrocyclic strain and provide high-resolution structural detail for the mode of binding. One of the main structural features of these molecules is a rigid 12-membered cyclic depsipeptide incorporating an unusual (4R)-4-amino-5-hydroxy-2-pentenoic acid. During the course of our studies, we misjudged the 12-membered macrocycle as a straightforward synthesis target: it does not immediately reveal the challenges associated with its synthesis likely attributed to significant macrocyclic ring strain. Our findings suggest that this notable ring strain results from unsaturation at C17−18. 2D NMR studies1 pointed to a low energy conformation of the 12-membered macrocycle which, counterintuitively, locks the C16 NH amide in the s-cis and the C11 NMe amide in the s-trans orientation (Scheme 1a, box). Our first synthesis target in this area became a N-desmethyl congener of thalassospiramide C, the most potent calpain 1 inhibitor reported in this series. Structural analysis and preliminary investigations led to the synthesis plan outlined in Scheme 1. After removal of the lipopeptidic side chain at N20, intermediate i had been identified as an intermediate macrocyclic target (Scheme 1a). The construction of the
n 2007, Fenical and co-workers reported the isolation and structural characterization of unique immunosuppressive depsipeptides from marine α-proteobacterium Thalassospira sp. CNJ-328, thalassospiramides A and B (Figure 1).1 In 2013, Qian, Moore and co-workers described the discovery and a biosynthesis study of an extended family of 14 new members from four Tristella and Thalassospira isolates of marine bacteria. The entire family of thalassospiramides showed significant inhibitory activity against human calpain 1 protease, with IC50 ranging between 3.4 nM, for the most potent thalassospiramide C, and 79 nM for thalassospiramide B1.2 Calpains and cathepsins are important biological targets for the treatment of multiple disorders, including fibrosis, osteoporosis, and neurodegeneration.3 A proposed mechanism of action supported by chemical studies suggested a conjugate addition of a cysteine residue to the α,β-unsaturated amide at C17- present in the macrocyclic unit, resulting in a covalent bonding to the calpain.4 Removal of the double bond by hydrogenation or opening of the macrocyclic ring by ester hydrolysis resulted in a total loss of activity.4 We became interested in developing a chemical synthesis of thalassospiramides with the purpose of reevaluating calpain and cathepsin activity of the natural products as well as the activity of the 12membered macrocyclic depsipeptide pharmacophore unit. Here, we report the first total synthesis in this family of natural products and an X-ray crystal structure of N-desmethyl © 2019 American Chemical Society
Received: November 29, 2018 Published: January 10, 2019 508
DOI: 10.1021/acs.orglett.8b03821 Org. Lett. 2019, 21, 508−512
Letter
Organic Letters
precursor v to diketopiperazine vi (Scheme 1b). Therefore, the disconnection along the C16 amide appeared to provide higher odds for success. To probe the effect of conformational constraints imposed by the C17−18 double bond experimentally, two precursors were targeted for the study: α,βunsaturated amide ii (path A) and β-phenylthio amide iv (path B). Olefination of serine-derived Garner’s aldehyde with methyl (triphenylphosphoranylidene)acetate followed by ester hydrolysis with lithium hydroxide readily afforded unsaturated carboxylic acid 2 in 85% yield as a single E isomer (Scheme 2a). Hydrolysis of the ester prior to the addition of thiophenol was required, because attempted hydrolysis of the βScheme 2
Figure 1. Structures of Thalassospiramide A−F. Sat. FA = saturated fatty acid at the N-terminus.
Scheme 1. Synthesis Design Factors
strained 12-membered ring could be approached from several alternative directions. Disconnection along the C1 ester was ruled out due to the high rigidity of the resultant acyclic secoacid precursor.5 Two amides and the C17−18 double bond would render an extended rigid conformation for the precursor, hampering ring closure. Preliminary studies also revealed that disconnection along the C16 amide (i → v, Scheme 1b) is problematic due to rapid fragmentation of 509
DOI: 10.1021/acs.orglett.8b03821 Org. Lett. 2019, 21, 508−512
Letter
Organic Letters
tyrosine TBS removal7 completed the synthesis, affording 20 mg of N-desmethyl thalassospiramide C (1). Other methods of amide formation resulted in partial epimerization at C22. Synthetic access to these quantities of 1 enabled a study of the molecular underpinning of its properties as a protease inhibitor. Analysis of biochemical potencies of 1 against human calpain and cathepsin isoforms showed potent nanomolar inhibition (Table 1). Significant selectivity was observed for
(phenylthio)ester resulted in elimination of the phenylthio group and recovery of 2. From 2, thiophenol addition occurred smoothly to afford a 4:1 mixture of the conjugate addition products in 91% yield giving 3 as the major diastereomer. Intermediates 2 and 3 were advanced in parallel to test the mode-of-action-inspired strain-relief macrocyclization as the central hypothesis of the synthesis plan (Scheme 2b). Appendage of L-ValOAll was achieved in excellent yields using HATU as the coupling reagent. Selective cleavage of the acetonide without removal of the N-Boc group was ultimately accomplished with bismuth trichloride, as the standard acid hydrolysis was consistently accompanied by concurrent fragmentation of the Boc group. Esterification of 5a/5b with N-Alloc(OTBS) tyrosine in the presence of EDC hydrochloride and DMAP gave 6a/6b in good yields, and reductive cleavage of both N-Alloc and O-Alloc groups was highly efficient in the presence of 5 mol % of Pd(PPh3)4, affording the zwitterionic substrates for macrolactamization 7a and 7b. As anticipated, the macrocyclization turned out to be exceptionally challenging, with 7a failing to provide even traces of the 12-membered cyclic depsipeptide under multiple macrocyclization strategies with any of the reagents tested (Scheme 2b).6 On the other hand, β-phenylthio congener 7b underwent slow macrocyclization with several tested reagents, revealing that epimerization at C12 of the valine unit is a problem. This problem was further magnified by the inability to separate the diastereomers. COMU and EDCI-HOAt, while affording lower yields, were able to suppress the epimerization, therefore proving to be the reagents of choice for the macrocyclizaton of 7b. Completion of the synthesis required reintroduction of the ring strain in the form of the C17−18 double bond and attachment of the lipopeptide at N20 (Scheme 3). Oxidation of the phenyl sulfide 8b to the sulfoxide followed by thermolysis in DMF succeeded in installing the requisite double bond. The Boc group was removed efficiently with CF3CO2H, indicating the stability of the strained macrocyclic ring system under acidic conditions. Finally, appendage of the side chain was achieved via hydroxy-succinate ester 12 and
Table 1. Biochemical Potencies of 1 against Proteasesa Protease inhibitor calpain 1 calpain 2 cathepsin cathepsin cathepsin cathepsin
K B S L
IC50, nM 175 210 3 65 46 1
a
Experimental details of the assay are reported in the Supporting Information.
cathepsins K and L, while potency against calpain 1 and 2 was approximately 60-fold less than that reported for thallasospiramide C, perhaps due the substitution of Tyr N-Me amide at C10 with NH amide.2 We were successful in crystallizing the adduct between 1 and cathepsin K, which allowed its detailed structural analysis by Xray crystallography. Figure 2a reveals that the macrocycle of 1 is anchored in the P1′ site of the enzyme. The active site Cys25 is covalently bound to C18. The torsional angle at the C17−18 double bond is 105°, indicating a considerable release of strain upon the addition of Cys25 to the double bond. The proximal C16 carbonyl accepts an H-bond from Gln19 (2.9 Å). This mimics the H-bonding found in the oxyanion hole of cathepsins. The phenolic OH of the endocyclic Tyr at C7 has a weak H-bond to the side chain of Asn159 (3.4 Å). The HN at C10 forms a hydrogen bond to the backbone carbonyl of Asn161 (2.7 Å). The C10 amide is N-methylated in thalassospiramide C and would not form this hydrogen bond. Modeling indicates that the added N-methyl would cause local shifts of ∼1 Å or more. However, the general features of the entire macrocyclic conformation of 1 bound to cathepsin are remarkably similar to the solution structure of thalassospiramide A elucidated by NMR spectroscopy.1 The fatty acid side chain (FA) extends from the macrocycle and binds to the P1−P3 subsites of cathepsin K. The interactions are typical of a peptidomimetic bound to cathepsins or to the structurally homologous family of calpains. The C21 amide forms H-bonds to the backbone carbonyl groups of Asn161 and the amide group of Gly66. The isopropyl group at C22 is buried in the hydrophobic pocket P2. These P1−P2 interactions represent a conserved motif for this protease family and have been observed in many crystallographic structures.9a,11 There are only minimal interactions in the P3 pocket and no apparent H-bonds. The electron density is weak for both the phenol that the FA side chain indicating some structural disorder. The conformation of the macrocycle helps explain the relative potency of 1 in different cathepsins. The P1′ pocket in cathepsin L is fairly homologous to cathepsin K. The cathepsin L crystal structure indicates that there would be no steric interference with the bound conformation of 1 (Figure 2b).8 By contrast, Ser138 → Arg substitution in cathepsin S may
Scheme 3. Completion of the Total Synthesis
510
DOI: 10.1021/acs.orglett.8b03821 Org. Lett. 2019, 21, 508−512
Letter
Organic Letters
Calpain 1 and 2 show nearly a 100-fold loss of potency in spite of the strong structural homology to the catalytic domains of the cathepsins. Figure 2e indicates that the P1′ binding site is partially blocked by residues 254−262 in human calpain 1.11 However, this loop does show conformational heterogeneity in different crystal forms of the rodent calpains.12,13 Differences in the chemical reactivity of the active site cysteine may also contribute to the reduced potency in calpains. In summary, the marine 12-membered cyclic depsipeptides thalassospiramides are characterized by a strained macrocyclic electrophile that conceals the challenges associated with its synthesis. Completing the first synthesis in this area was enabled by an approach inspired by the proposed mode of action of thalassospiramides that is based on releasing the macrocyclic ring strain. Access to synthetic N-desmethyl thalassospiramide C allowed for determination of calpain and cathepsin protease inhibition and crystallographic structure of its covalent adduct with cathepsin K. The cysteine proteases calpain and cathepsin are of intense current interest as therapeutic targets for neurodegenerative disorders, fibrosis, cardiovascular diseases, cancer, osteoporosis, and more.3,14 The structural and inhibition data provided in this study may provide additional insight for protease inhibitor drug design.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.orglett.8b03821. Structure of N-desmethyl thalassospiramide C complex with cathepsin K (PDB) Detailed experimental procedures, assay procedures, copies of 1H and 13 C NMR spectra, HPLC traces, and X-ray crystallographic data (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
Armen Zakarian: 0000-0002-9120-1232 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported in part by the NSF (CHE 1463819). We thank Dr. Hongjun Zhou (UCSB) for the assistance with NMR spectroscopy and Dr. Uchenik and the UCSB mass spectroscopy facility for the assistance with mass-spectroscopic analysis.
Figure 2. (a) Crystal structure of 1 bound to cathepsin K, dotted lines show intermolecular H-bonds (PDB entry 6hgy). The fatty acid chain (FA) extends toward viewer and is cropped to show the underlining interactions. The structure of cathepsin K and 1 (green) superimposed on the published structures of other homologous cysteine proteases: (b) cathepsin L (brown, PDB entry 2xu3); (c) cathepsin S (yellow, PDB entry 4p6g); (d) cathepsin B (gray, PDB entry 1gmy); (e) calpain 1 (cyan, PDB entry 1zcm). Residue labels show potential steric clashes.
■
REFERENCES
(1) Oh, D.-C.; Strangman, W. K.; Kauffman, C. A.; Jensen, P. R.; Fenical, W. Org. Lett. 2007, 9, 1525−1528. (2) (a) Ross, A. C.; Xu, Y.; Lu, L.; Kersten, R. D.; Shao, Z.; AlSuwailem, A. M.; Dorrestein, P. C.; Qian, P.-Y.; Moore, B. S. J. Am. Chem. Soc. 2013, 135, 1155−1162. (b) Lu, L.; Meehan, M. J.; Gu, S.; Chen, Z.; Zhang, W.; Liu, L.; Huang, X.; Dorrestein, P. C.; Xu, Y.; Moore, B. S.; Qian, P.-Y. Sci. Rep. 2015, 5, 8783. (3) Ono, Y.; Saido, T. C.; Sorimachi, H. Nat. Rev. Drug Discovery 2016, 15, 854−876.
impinge upon the C11−C12 region of the inhibitor,9b accounting for more than 10-fold loss in activity (Figure 2c). Cathepsin B10 has an alternative fold in the P1′ site that is unique in the cathepsin family. This conformational difference can account for the lower potency in cathepsin B (Figure 2d). 511
DOI: 10.1021/acs.orglett.8b03821 Org. Lett. 2019, 21, 508−512
Letter
Organic Letters (4) Lu, L.; Meehan, M. J.; Gu, S.; Chen, Z.; Zhang, W.; Liu, L.; Huang, X.; Dorrestein, P. C.; Xu, Y.; Moore, B. S.; Qian, P.-Y. Sci. Rep. 2015, 5, 8783. (5) This hypothesis was also borne out by preliminary studies. Examples of macrolactonization during cyclic depsipeptide synthesis are rare: (a) Hamada, Y.; Shioiri, T. Chem. Rev. 2005, 105, 4441− 4482. (b) Li, W.; Schlecker, A.; Ma, D. Chem. Commun. 2010, 46, 5403. (c) Davies, J. S. J. Pept. Sci. 2003, 9, 471−501. (d) Batiste, S. M.; Johnston, J. N. Proc. Natl. Acad. Sci. U. S. A. 2016, 113, 14893− 14897. (6) El-Faham, A.; Albericio, F. Chem. Rev. 2011, 111, 6557−6602. (7) The byproduct with free phenolic OH can be silylated in >95% yield and used productively in the next step of the synthesis. (8) Hardegger, L. A.; Kuhn, B.; Spinnler, B.; Anselm, L.; Ecabert, R.; Stihle, M.; Gsell, B.; Thoma, R.; Diez, J.; Benz, J.; Plancher, J. M.; Hartmann, G.; Banner, D. W.; Haap, W.; Diederich, F. Angew. Chem., Int. Ed. 2011, 50, 314−318. (9) (a) Boros, E. E.; Deaton, D. N.; Hassell, A. M.; McFauyden, R. B.; Miller, L. R.; Paulick, M. G.; Shewchuk, L. M.; Thompson, J. B.; Willard, D. H., Jr.; Wright, L. L. Bioorg. Med. Chem. Lett. 2004, 14, 3425−3429. (b) Jadhav, P. K.; Schiffler, M. A.; Gavardinas, K.; Kim, E. J.; Matthews, D. P.; Staszak, M. A.; Coffey, D. S.; Shaw, B. W.; Cassidy, K. C.; Brier, R. A.; Zhang, Y.; Christie, R. M.; Matter, W. F.; Qing, K.; Durbin, J. D.; Wang, Y.; Deng, G. G. ACS Med. Chem. Lett. 2014, 5, 1138−1142. (10) Greenspan, P. D.; Clark, K. L.; Tommasi, R. A.; Cowen, S. D.; Mcquire, L. W.; Farley, D. L.; Van Duzer, J. H.; Goldberg, R. L.; Zhou, H.; Du, Z.; Fitt, J. J.; Coppa, D. E.; Fang, Z.; Macchia, W.; Zhu, L.; Capparelli, M. P.; Goldstein, R.; Wigg, A. M.; Doughty, J. R.; S Bohacek, R.; Knap, A. K. J. Med. Chem. 2001, 44, 4524−4534. (11) Li, Q.; Hanzlik, R. P.; Weaver, R. F.; Schonbrunn, E. Biochemistry 2006, 45, 701−708. (12) Cuerrier, D.; Moldoveanu, T.; Inoue, J.; Davies, P. L.; Campbell, R. L. Biochemistry 2006, 45, 7446−7452. (13) Hanna, R. A.; Campbell, R. L.; Davies, P. L. Nature 2008, 456, 409−412. (14) Turk, V.; Stoka, V.; Vasiljeva, O.; Renko, M.; Sun, T.; Turk, B.; Turk, D. Biochim. Biophys. Acta, Proteins Proteomics 2012, 1824, 68− 88.
512
DOI: 10.1021/acs.orglett.8b03821 Org. Lett. 2019, 21, 508−512
