Total Synthesis of Diazonamide A

Abstract

Ever since its discovery in 1991, diazonamide A (1) has been eyed by synthetic chemists as a potential target because of its puzzling molecular architecture and potent biological activity. The race to synthesize this intriguing natural product was further complicated at the end of last year when a synthesis was completed only to prove that the originally proposed structure was in error. The new structural assignment (2) required retooling of synthetic strategies toward the new target. A total synthesis of diazonamide A has now been achieved confirming its newly proposed structure. In 1991, Fenical and co-workers reported structure 1 for diazonamide A, a marine natural product which had been isolated from the colonial ascidian diazona angulata1 .1 A decade later, Harran and his group synthesized the proposed structure 1 only to prove that it was in error, and advanced structure 2 for diazonamide A instead.2 In the intervening time, numerous efforts directed at the total synthesis of 1 had been reported;3–13 no research activities related to the newly proposed structure 2 have yet been disclosed. The appeal of diazonamide A (2) stems both from its biological activity (cytotoxicity against several tumor cell lines with IC50 values <15 ng mL−1) and its highly strained and unprecedented structure. Because of its scarcity from natural sources and its potential as a lead for the development of new chemotherapeutic agents, 2 presents a unique opportunity for chemical synthesis. In this communication, we report a total synthesis of diazonamide A (2) which required the development of new synthetic technologies and strategies and which provides a solution to this formidable molecular puzzle. The revised structure of diazonamide A (2) represented a greater synthetic challenge than the one posed by the originally proposed structure (1) in that it contains an additional ring (ring H) and a nitrogen atom instead of the oxygen substituent in ring F which leads to a structurally unique aminal moiety. In addition to the rigidity associated with this EFGH tetracyclic system, its quaternary center (C10) preeminently stood as the central cornerstone of any synthetic strategy. The two 12-membered rings of the structure were also expected to pose considerable recalcitrance to formation owing to their severely constrained environments. In all, structure 2, with its ten exquisitely arrayed rings and unique elements of stereochemistry, amounted to a rare challenge to modern synthetic chemistry. Scheme 1 presents the blueprint of our strategy towards this target molecule in retrosynthetic format. Thus, excision of the side chain and the chlorine atoms, and rupture of the macrolactam, aminal, oxazole, and bis-aryl ring systems as shown, led to fragments 3–7. In the synthetic direction, the challenge was thus reduced to the following milestone events: a) construction of the required building blocks; b) assembly of the quaternary center with its substituents; c) formation of the 12-membered macrolactam ring system; d) introduction of the indole–oxazole domain; e) closure of the 12-membered polycyclic framework and chlorination; f) fusion of the aminal bicyclic network; and g) attachment of the isovaleric acid side chain (the operations to be executed in that order). The proposed late introduction of the side chain would allow for expedient construction of analogues, including the synthesis of the natural product's epimer at C37 for purposes of structural and biological comparison. General retrosynthetic analysis of diazonamide A (2). Cbz=benzyloxycarbonyl, Boc=tert-butoxycarbonyl, MOM=methoxy methyl, TFA=trifluoroacetic acid. While all of the defined building blocks (3–7) were readily generated from commercially available starting materials according to literature procedures, the construction of the quaternary center (C10) with its peripheral structural motifs required considerable process development. To this end, the model study depicted in Scheme 2 was designed and executed. Thus, 7-bromoisatin (8)14 was treated with excess PhMgBr to afford tertiary alcohol 9 in 96 % yield. Upon mixing compound 9 with tyrosine derivative 10 in the presence of TfOH, and following acetylation of the resultant primary amine (as TfOH cleaved the Cbz group), phenolic lactam 11 emerged in 84 % yield as a mixture of diastereomers (ca. 1:1).15 Exposure of 11 to excess BH3⋅SMe2 resulted in the formation of the bicyclic aminal system 14 (40 % overall yield, unoptimized) containing some of diazonamide A's most challenging structural features, presumably via intermediates 12 and 13 as suggested in Scheme 2. Model studies exploring the generation of the C10 quaternary center: a) PhMgBr (3.0 equiv), THF, 0 °C, 96 %; b) 10 (1.1 equiv), TfOH (10 equiv), CH2Cl2, 0 °C, 20 min; then Ac2O (10 equiv), CH2Cl2, 25 °C, 5 min, 84 %; c) BH3⋅SMe2 (3.0 equiv), THF, 25 °C, 4 h, 40 %. TfOH=trifluoromethanesulfonic acid. Note: compounds 11–14 are mixtures of C10 epimers (ca. 1:1). Prompted by this encouraging set of preliminary results, we then proceeded to apply the developed chemistry to the target molecule 2 as shown in Scheme 3. Having failed to successfully utilize the known oxazole ester derivative 1516 in the intended coupling reaction, we resorted to its reduction (LiBH4) and selective protection (60 % overall yield)17 to afford benzyl ether 6 as an alternative substrate for this process. Reaction of 6 with 2.0 equiv of nBuLi in THF at −78 °C resulted in the formation of dianion 16 which was treated with 1.1 equiv of isatin derivative 718 at −78 °C to afford, upon standard workup, tertiary alcohol 17 in 73 % yield. The crucial step of establishing the quaternary center assembly with its structural entourage was then accomplished by refluxing 17 with tyrosine derivative 5 (4.0 equiv) in 1,2-dichloroethane in the presence of pTsOH to furnish compound 18, which had lost both the MOM and the Boc protective groups, in 33 % yield. It is important to note that the triflic acid conditions utilized successfully in the model study (Scheme 2) failed in this instance; only decomposition products were obtained, which most likely arose from the oxazole moiety. Reprotection of the primary amine in 18 with (Boc)2O led to two C10 epimers whose chromatographic properties allowed their convenient separation to yield pure 19 and its epimer (not shown). Since the configuration of the quaternary center in these intermediates could not be determined at this stage, both epimers were separately carried forward to the corresponding macrolactams (22 and its C10 epimer) whose spectroscopic data were suggestive of their structures (see below 22: Rf=0.41 (silica gel, ethyl acetate); IR (film): max=3392, 1715, 1649, 1614, 1508, 1455, 1390, 1261, 1220, 1126, 1090, 1020, 750, 597 cm−1; 1H NMR (600 MHz, CD3CN): δ=8.66 (s, 1 H), 7.47 (d, J=7.9 Hz, 1 H), 7.37 (m, 4 H), 7.34 (m, 1 H), 7.27 (br d, 1 H), 7.22 (s, 1 H), 7.07 (d, J=7.4 Hz, 1 H), 7.02 (d, J=7.4 Hz, 1 H), 6.90 (t, J=7.9 Hz, 1 H), 6.70 (d, J=7.9 Hz, 1 H), 6.23 (s, 1 H), 6.04 (br d, 1 H), 5.07 (AB, J=12.3 Hz, AB=21.9 Hz, 2 H), 4.45 (t, J=7.4 Hz, 1 H), 4.11 (t, J=9.2 Hz, 1 H), 3.63 (qd, J=9.2, 5.7 Hz, 2 H), 3.06 (t, J=12.7 Hz, 1 H), 2.77 (br s, 1 H), 2.62 (d, J=12.7 Hz, 1 H), 2.04 (m, 1 H), 1.01 (d, J=7.4 Hz, 3 H), 0.91 ppm (d, J=7.4 Hz, 3 H); 13C NMR (150 MHz, CD3CN): δ=175.5, 173.4, 164.6, 155.9, 153.0, 144.7, 142.4, 139.0, 138.0, 133.6, 133.0, 130.6, 130.0, 129.3, 128.8, 128.7, 128.6, 127.7, 125.4, 124.2, 116.2, 102.9, 67.0, 57.6, 57.0, 55.8, 38.4, 30.7, 19.6, 19.3 ppm; HR-MS (matrix-assisted laser desorption/ionization, MALDI) for C33H31BrN4O7Na+ [M+Na+]: calcd: 697.1268, found: 697.1266 26: Rf=0.40 (silica gel, ethyl acetate/hexane, 1:1); IR (film): max=3366, 2923, 2849, 1701, 1660, 1490, 1449, 1396, 1376, 1343, 1255, 1202, 1120, 750, 597 cm−1; 1H NMR (600 MHz, CD3CN): δ=10.41 (s, 1 H), 7.51 (d, J=7.0 Hz, 1 H), 7.40 (m, 5 H), 7.31 (dd, J=7.0, 6.6 Hz, 1 H), 7.30 (s, 1 H), 7.20 (d, J=7.0 Hz, 1 H), 7.19 (d, J=8.3 Hz, 1 H), 6.98 (d, J=6.6 Hz, 1 H), 6.88 (d, J=7.0 Hz, 1 H), 6.87 (d, J=7.5 Hz, 1 H), 6.78 (d, J=7.8 Hz, 1 H), 6.63 (dd, J=7.8, 7.5 Hz, 1 H), 6.38 (d, J=4.4 Hz, 1 H), 6.17 (d, J=8.0 Hz, 1 H), 5.88 (d, J=4.4 Hz, 1 H), 5.09 (s, 2 H), 4.85 (dd, J=8.3, 5.7 Hz, 1 H), 4.17 (m, 1 H) 3.28, (dd, J=12.3, 12.0 Hz, 1 H), 2.80 (dd, J=12.3, 3.0 Hz, 1 H), 2.11 (m, 1 H), 0.98 (d, J=7.2 Hz, 3 H), 0.90 ppm (d, J=7.2 Hz, 3 H); HR-MS (MALDI) for C43H32Cl2N6O6Na+ [M+Na+]: calcd: 821.1653, found: 821.1669 ). The following step involved protection of the phenolic hydroxy group and the oxoindole amide nitrogen atom, an objective which was accomplished by exposing 19 to excess MOMCl and K2CO3 to afford compound 20 (94 % yield). This precursor (20) was then transformed to the coveted macrolactam 21 by first liberating the carboxylic acid (LiOH) and the amino groups (TFA), in that order, and then cyclizing the resultant amino acid through the action of HATU and collidine in DMF/CH2Cl2 in 36 % overall yield. This challenging formation of the 12-membered ring required high-dilution conditions to minimize formation of dimers and higher oligomers. It should also be noted that MOM protection of the precursor considerably improved the yield of this ring closure as compared to that obtained with unprotected material (i.e., 19). Furthermore, of the two C10 epimers, only one cyclized smoothly whereas the other led to preferential dimerization even under extremely high dilution conditions. The next step in the sequence required removal of the benzyl group in the presence of the Cbz and the aryl bromine moieties, a challenging task that was accomplished by the use of BCl3 in CH2Cl2 at −78 °C followed by treatment with NaOH. In this reaction, the two MOM groups in 21 were concomitantly cleaved to yield compound 22 in 61 % overall yield.19 At this stage, the 1H NMR spectroscopic data of the more accessible macrolactam epimer (see Table 1) were reminiscent of those of diazonamide A (2). Therefore, we assumed the structure of this epimer to be that of 22 (that is to say, the one corresponding to the targeted natural product). Synthesis of advanced macrolactam 22: a) LiBH4 (2.0 M in THF, 3.0 equiv), EtOH (10 equiv), Et2O/THF (2:1), 0→25 °C, 3 h, 83 %; b) NaHMDS (1.0 M in THF, 2.0 equiv), THF, −78 °C, 10 min, then TBAI (0.1 equiv), BnBr (10 equiv), −78→25 °C, 24 h, 72 %; c) nBuLi (2.0 equiv), THF, −78 °C, 20 min, then 7 (1.1 equiv), 10 min, 73 %; d) 5 (4.0 equiv), pTsOH (4.0 equiv), ClCH2CH2Cl, 83 °C, 15 min, 33 %; e) (Boc)2O (1.1 equiv), aq. NaHCO3/dioxane (1:2), 25 °C, 2 h, 65 %; f) MOMCl (30 equiv), K2CO3 (40 equiv), acetone, 0→25 °C, 2 h, 94 %; g) LiOH (20 equiv), MeOH/THF/H2O (2:10:1), reflux, 10 min, 98 %; h) TFA, 10 min, 98 %; i) HATU (2.2 equiv), collidine (6.6 equiv), DMF/CH2Cl2 (1:2, 3.0×10−4 M), 25 °C, 12 h, 36 %; j) BCl3 (1.0 M in hexanes, 20 equiv), CH2Cl2, −78 °C, 2 h; then THF, sat. aq. NaHCO3, 10 % aq. NaOH, 25 °C, 30 min, 61 %. NaHMDS=sodium salt of 1,1,1,3,3,3-hexamethyldisilazane, TBAI=tetra-n-butylammonium iodide, pTsOH=p-toluenesulfonic acid, HATU=2-(1H-9-azabenzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate. Having chosen epimer 22 as the most likely candidate to yield diazonamide A (2), its further elaboration proceeded as shown in Scheme 4. Thus, selective protection of the phenolic hydroxy group of 22 (Boc2O), followed by sequential oxidation first with IBX and then with NaClO2 , furnished the expected carboxylic acid whose coupling with indole ammonium salt 4 in the presence of EDC and HOBt led to 23 (41 % overall yield from 22). This keto amide was then converted to the corresponding oxazole system through a Gabriel–Robinson cyclodehydration effected by a unique set of conditions (POCl3/pyridine)20 which set the stage for the critical cyclization event required to form the second 12-membered ring of diazonamide A. Most gratifyingly, radical cyclization of this precursor with Ph3SnH/AIBN induced the desired ring closure to 24, albeit in low yield (ca. 10 %). The efficiency of this ring closure was improved to 30 % by resorting to a Witkop-type reaction21a based on a literature procedure,2a a process that was proven effective in previous instances of macrocycle construction.21b,21c Significantly, and in contrast to the Ph3SnH-induced reaction (in which the majority of the substrate was reductively debrominated), the bulk of the remaining material from this process remained unchanged which allowed for productive recycling of recovered starting material. As such, this observation lends further support that this particular macrocyclization reaction proceeds through a mechanism based on intramolecular photo-induced electron transfer from the D-ring indole chromophore to the adjacent benzenoid E-ring,2a rather than some exclusively radical-based event in which case the halogen in the starting material would most likely be expelled. Final stages and completion of the total synthesis of diazonamide A (2): a) (Boc)2O (30 equiv), sat. aq. NaHCO3/dioxane (1:2), 25 °C, 24 h, 91 % based on recovered starting material; b) IBX (3.0 equiv), DMSO, 25 °C, 2 h; c) NaClO2 (5.0 equiv), NaH2PO4 (5.0 equiv), resorcinol (5.0 equiv), DMSO/H2O (10:1), 0→25 °C, 2 h, 69 % over two steps; d) 4 (3.0 equiv), EDC (5.0 equiv), HOBt (5.0 equiv), NaHCO3 (15 equiv), DMF, 25 °C, 12 h, 65 %; e) POCl3/pyridine (1:4), 25 °C, 2 h, 52 %; f) hν (200 nm), epichlorohydrin (3.0 equiv), LiOAc (2.0 equiv), MeCN/H2O (3:1), 25 °C, 15 min, 30 %—or Ph3SnH (2.0 equiv), AIBN (0.1 equiv), C6H6, 80 °C, 3 h, 10 %; g) NCS (4.0 equiv), CCl4/THF (1:1), 60 °C, 2 h, 53 %; h) TFA, 25 °C, 10 min, 98 %; i) DIBAL-H (1.0 M in toluene, 100 equiv, added portionwise), THF, −78→25 °C, 3 h, 56 %; j) H2 (2.0 atm), Pd(OH)2/C (20 wt %, catalytic), EtOH, 25 °C, 2 h; k) (S)-3 (5.0 equiv), EDC (5.0 equiv), HOBt (5.0 equiv), NaHCO3 (15 equiv), DMF, 25 °C, 12 h, 82 % over two steps. IBX=1-hydroxy-1,2-benziodoxol-3(1H)-one, py=pyridine, AIBN=2,2′-azobisisobutyronitrile, NCS=N-chlorosuccinimide, DIBAL-H=diisobutylaluminum hydride. With the establishment of both macrocyclic scaffolds of diazonamide A through the successful generation of 24, completion of the synthesis required only a few finishing touches (see Scheme 4). First, the two, still missing, chlorine residues were installed at their proper positions by electrophilic aromatic substitution (NCS, 53 % yield). Second, the resulting product was exposed to TFA to effect removal of the Boc group and furnish phenolic intermediate 25 (98 % yield). Chemoselective reduction of the five-membered ring lactam of 25 with DIBAL-H initiated ring closure with participation of the adjacent phenolic hydroxy group to afford the desired cyclic aminal 26 in 55 % overall yield. Finally, hydrogenolysis of the Cbz group with H2 and Pearlman's catalyst followed by installation of the remaining peptide side chain with isovaleric acid ((S)-3) in the presence of EDC/HOBt furnished structure 2 in 82 % overall yield. The 1H NMR signals of synthetic 2 in [D4]MeOH matched those reported for natural diazonamide A precisely, while those of the C37 epimer obtained by employing (R)-3 in the final step were slightly, but diagnostically, different.22 Based on this data, we confirm structure 2 for diazonamide A. In conclusion, a path has now been chartered for the total synthesis of the long-sought diazonamide A which confirmed its structure as 2, and through which this valuable, but scarce, natural substance is now rendered available for biological investigations.23 Further synthetic work is expected to facilitate chemical biology studies and establish a clear structure–activity profile for this highly unusual molecular architecture.