Synthesis and Applications of iso-Hajos–Parrish Ketones

Abstract

Although numerous natural products possess ring systems and functionality for which “iso-Hajos–Parrish” ketones would be of value, such building blocks have not been exploited to the same degree as the more typical Hajos–Parrish hydrindane. Herein we outline an efficient three-step synthesis of such materials fueled by a simple method for the rapid preparation of highly functionalized cyclopentenones, several of which are new chemical entities that would be challenging to access through other approaches. We then show how one iso-Hajos–Parrish ketone can be converted into two distinct natural product analogs as well as one natural product. As one indication of the value of these new building blocks, that latter target was obtained in 10 steps, having previously been accessed in 18 steps using the Hajos–Parrish ketone.

The Hajos–Parrish ketone (1, Scheme 1)^[1] has long been a valuable starting material for accessing natural products given the wealth of structures containing similar 6,5-fused ring systems with a 1,2-relationship between the highlighted stereogenic center and ketone (or functional groups derived from it).^[2] By contrast, strategies to access “iso-Hajos–Parrish” ketones of general structure 6, materials that possess a 1,3-placement of these two key groups and a change in alkene location, are far less developed even though numerous compounds carry one or more such domains (including 7–9), sometimes in a ring-expanded format (as in 10). One solution for these materials begins with the Hajos–Parrish ketone (1) itself and uses a linear array of subsequent synthetic steps to transpose its functional groups into molecules such as sarcandralactone A (12).^[3] We envisioned, however, that a more expedient and general strategy for the synthesis of natural products like 7–10 and 12^[4] could exist if one could readily prepare an array of iso-Hajos–Parrish ketones (6) with diversity at the sites indicated by group R. For that task, an ideal approach might be to merge varied cyclopentenones (14) directly with the Danishefsky diene (13) in a Diels–Alder reaction followed by an in situ decarboxylation; however, while this idea is simply stated, such Diels–Alder reactions are rare, often needing multi-step or mechanistically distinct solutions such as double Michael additions.^[5] Perhaps more critically, methods to efficiently and reliably prepare the requisite functionalized cyclopentenones (14) are lacking. Indeed, while cyclopentenones can be prepared by a number of different pathways, including metallation chemistry,^[6] Michael additions/reoxidations,^[7] metal cyclizations/Conia ene chemistry,^[8] and Nazarov cyclizations,^[9] general, mild, and step-economic processes have not been demonstrated for many variants of the types desired as broadly defined by R within 14.

Scheme 1.

The value of the Hajos-Parrish ketone in total synthesis, and a proposal that iso-Hajos-Parrish ketones of structure 6 could be of equal value if the merger of 13 and 14 can be achieved in one-pot and if diverse cyclopentenones of type 14 can be prepared.

Herein, we detail a simple, two-step solution capable of affording structurally diverse cyclopentenones. We then show how these materials can be converted into complex polycycles, serve as masked forms of cyclopentadienone, and afford several iso-Hajos–Parrish ketones (6) in just 3 steps overall. Finally, we illustrate that one of these ketones can be advanced into two natural product analogs and natural product 12. As one measure of that new ketone’s value, the latter target was obtained in only 10 steps, having been previously made in 18 total steps using 1.

Our overall design for achieving a rapid and broadly applicable cyclopentenone synthesis is shown in Scheme 2. Following Wittig olefination of an anhydride (15) with a stabilized ylide to generate 16,^[10] we hoped that subsequent ring opening as facilitated by the Weinreb amine would afford a new material (17) that could, in the same pot, be enolized, treated with a nucleophile (here a methyl anion) to form a new ketone (18), and then undergo a Knoevenagel condensation and generate 14.^[11] Although all these steps individually are known operations, their combination in this cascade is, to the best of our knowledge, without precedent.

Scheme 2.

Proposed method for broad cyclopentenone synthesis.

Following significant reaction screening and optimization, particularly of the second of these two steps, this designed process could be achieved and a number of functionalized cyclopentenones were prepared in moderate to good yield as shown in Table 1. The initial Wittig reaction was conducted in toluene at 50 °C for 16 h, and afforded only a single alkene stereoisomer in all cases presented within the Table. The optimized process for step two involved initial formation of a Weinreb amide^[12] under standard conditions followed by treatment with 1.3 equivalents of NaH in THF at 0 °C for 30 min and then exposure to the desired nucleophile (1.1 equiv) at −78 °C for 1 h; once complete, the reaction mixture was then heated at 50 °C in the presence of excess MeOH (50 equiv) for 1 h to effect the terminating Knoevenagel condensation. As indicated with the entries listed for succinic anhydride (19), many different nucleophiles can be used, with lithiated species generally giving superior yields over Grignard reagents (entries 1–5);^[13] various groups labelled as X can be incorporated in the olefination step as well to ultimately give different exocyclic esters (entries 6–9), with control of reaction temperature in the final Knoevenagel condensation being key to preventing transesterification with the added MeOH. Critically, anhydrides with additional substituents, and thus increased steric bulk near the key reacting carbonyl groups, were also tolerated, affording the means to effect inaugural syntheses of cyclopentenones 29, 30, and 32 (entries 10–12). The reaction sequence could also be performed on gram scale (entry 10).

Table 1.

Initial exploration of the scope of cyclopentenone synthesis from acid anhydrides.^[a]

Entry	Anhydride	X	Nucleophile	Product	Yield [%]
1	19	Me	MeLi	20	68
2		Me	MeMgBr		54
3	19	Me	n-BuLi	21	65
4	19	Me	DIBAL-H	22	48
5	19	Me		23	46
6	19	Allyl	MeLi	24–27	58
7		Bn	MeLi		62
8		t-Bu	MeLi		71
9			MeLi		61
10	28	Me	MeLi	29–30	72[b]
11		t-Bu	MeLi		71
12	31	Me	MeLi	32	53

^[a]General conditions: Wittig phosphorylidine (1.0 equiv), anhydride (1.0 equiv), 50 °C, 10–16 h; N,O-dimethyl hydroxylamine•HCl (1.3 equiv), pyridine (5.0 equiv), CH2Cl2, 25 °C, 6 h, then THF, NaH (1.3 equiv), 0 °C, 30 min, then nucleophile (1.1 equiv), −78 °C, 1 h, then MeOH (50 equiv), 50 °C, 1 h.

^[b]Synthesized on gram scale.

Globally, while the overall yield of these two step sequences are in the 45–70% range, that outcome correlates to >80% yield per operation within the 4 part cascade sequence of step 2 for even the lowest yielding entry. Key to these yields is the initial formation of a Weinreb amide; indeed, although direct treatment of materials of type 16 (cf. Scheme 2) with the desired nucleophile could ultimately afford cyclopentenones of type 14, these processes sometimes failed and overall yields when they were successful were significantly diminished. Of note, several of the final compounds in Table 1, such as the acetal of 23, the allyl group of 24, and the cyclopropane of 29 and 30 are unlikely to result from and/or survive the conditions of available approaches, particularly Lewis acid-promoted Nazarov cyclizations.^[14]

Two more advanced examples of cyclopentenone synthesis are shown in Scheme 3. In the first case, a non-symmetric anhydride could be converted into 33 in 37% overall yield. Here, the initial Wittig reaction proceeded to afford a 1:1.2 ratio of regioisomers, indicating that the neighboring bulk of the methyl group did not dramatically deter the formation of the desired intermediate.^[15] The resultant compound (33) was then advanced in 2 additional and standard transformations via the intermediacy of an acid chloride into the natural product merrekentrone D (34).^[16,17] In the second example, a nucleophile was used bearing a pendant diene system with anhydride 19 (cf. Table 1) to afford a cyclopentenone (35) that could be directly converted into the more complex polycycle 36 in 32% yield overall through an intramolecular Diels–Alder reaction. This result is of significance both for the complexity of the final material and as an additional example of a Diels–Alder reaction occuring under simple thermal conditions with an α, β-unsaturated system bearing a β-substituent.^[5,18]

Scheme 3.

Selected examples of advanced cyclopentenones and their use in additional applications, including a total synthesis of merrekentrone D (34): a) p-TsOH•H2O (1.0 equiv), HCO2H, 90 °C, 4 h, 91%; b) oxalyl chloride (2.0 equiv), DMF (cat.), CH2Cl2, 0 °C, 30 min, then 25 °C, 4 h; concentrate; 3-LiFuran (1.1 equiv), THF, −116 °C, 1.5 h, 18%; c) toluene, 150 °C, 24 h, 32%. Ts = tosyl.

We next turned our attention to generating iso-Hajos–Parrish ketones of general structure 6. Scheme 4 presents two examples of such materials we have prepared. In the first, following the synthesis of cyclopentenone 37 in 62% yield using the standard procedures delineated above, subsequent exposure to a Danishefsky diene surrogate using LiHMDS followed by acid treatment delivered a formal equivalent of a Diels–Alder product in a double Michael addition process in 44% yield (53% based on recovered starting material). In a thermal Diels–Alder reaction between the Danishefsky diene (13) and 37, the desired product was obtained in only 15% yield following treatment with TFA. While that outcome is non-optimal, its occurrence in any yield is of note.^[5ab,19] In any event, we were then able to take the resultant product and extrude cyclopentadiene in a retro-Diels–Alder reaction in 91% yield upon heating at reflux in 1,2-dichlorobenzene. As such, this process highlights an example of a formal cyclopentadienone equivalent. In the second case, we were able to execute an effective thermally-promoted Diels–Alder reaction between 30 and 13, finding that the t-butyl ester group in the resultant product could be extruded in the same pot by heating in the presence of TFA at 75 °C to generate 39 in 95% yield. No double bond isomerization was noted.^[20]

Scheme 4.

Synthesis of 2 different iso-Hajos-Parrish ketones, the use of one of them for the generation of 2 different natural product frameworks (40 and 42) and 1 natural product (12), and a method to prepare enantiopure 29: a) dienophile (1.5 equiv), LiHMDS (1.8 equiv), THF, −45 °C, 45 min, then 37 (1.0 equiv), −78→25 °C, 3 h; b) p-TsOH•H2O (2.0 equiv), CHCl3, reflux, 3 h, 44% over two steps, 53% b.r.s.m.; c) 1,2-dichlorobenzene, reflux, 9 h, 91%; d) Danishefsky diene (13, 2.5 equiv), toluene, sealed tube, 140 °C, 16 h; concentrate; TFA, 75 °C, 15 h, 95%; e) aldehyde (1.3 equiv), NaH (1.3 equiv), THF, 0 °C, 2 h; TFA, toluene; concentrate; p-TsOH•H2O (1.14 equiv), toluene, 25 °C, 13 h, 32%; f) Me2CuLi2CN (20 equiv), TMSCl (4.0 equiv), DMPU, THF, −40→25 °C, 1 h, 83%; g) 1 M HCl, CH2Cl2, 25 °C, 95%; h) SmI2 (1.0 equiv), DMPU, THF, 25 °C, 30 min, 71%; i) NaH (1.3 equiv), THF, 0 °C, 30 min; methyl pyruvate (1.3 equiv), 0 °C, 30 min, repeat 6 times, 89%; j) p-TsOH•H2O (1.9 equiv), toluene, 100 °C, 16 h, 66%; k) H2, Pd/C (10%, 0.1 equiv), EtOAc, 25 °C, 2.5 h; filter and concentrate; Ac2O, p-TsOH (0.55 equiv), 25 °C, 16 h; l) DBU (10 equiv), THF, 25 °C, 16 h, 63% over two steps; m) 43 (3.0 equiv), NaHMDS (2.5 equiv), THF, −78→ −40 °C, 3 h, 81%; n) NaOH (5 equiv), THF, pyridine, H2O, 25 °C, 24 h, 55%; o) NaBH4 (6.1 equiv), MeOH, 0→25 °C, 4 h, 86%; p) SeO2 (4.5 equiv), 1,4-dioxane, 80 °C, 1 h, 85%; q) standard procedure cf. Table 1, 66%; r) MeOH, 4-DMAP, toluene, Δ, 64–83%. HMDS = hexamethyldisilazane.

As a final study, we sought to determine what types of frameworks could be accessed from iso-Hajos–Parrish ketone 39. As shown in the middle portion of Scheme 4, we were able to convert it into 4-desmethylpinguisone (40) in 3 steps by attaching a furan ring system onto the core and ruptering the cyclopropane into a β-methyl group through the reductive action of SmI2.^[21] We also were able, after the formation of intermediate 41, to convert the cyclopropane and its adjoining ring system into a functionalized 6-membered ring (42) reflective of the core of the eudesmanolides following treatment with p-TsOH in toluene at 100 °C.^[22] We could also convert 41 in 6 additional steps into sarcandrolide A (12).^[23,24] That sequence completed a 10 step route to this target overall from commercial materials, noting that as shown in Scheme 1 it had previously been prepared using the Hajos–Parrish ketone (1) in 18 total steps (16 steps from 1), with most of those operations involving the transposition of the core ketone group avoided here with the efficient preparation of 39 in just 3 steps. Lastly, although all the studies described thus far have afforded racemic materials, chiral 29 could be accessed for enantiospecific syntheses by implanting a menthol chiral auxiliary as part of the ester component in the initial Wittig coupling (generating 45 in 2.7:1 d.r.), recrystallizing 45 to >100:1 diastereopurity in a mixture of CH2Cl2 and hexanes, performing the cyclopentenone formation sequence, and then cleaving the auxiliary through methanolysis.^[25]

In summary, we have developed a two-step sequence capable of delivering a number of uniquely functionalized cyclopentenones from anhydrides, several of which are unlikely to arise from other available approaches. We have then shown how these materials can be used in a number of applications, most notably as precursors that undergo thermal Diels–Alder chemistry with the Danishefsky diene to deliver iso-Hajos–Parrish ketones. These materials, in turn, can afford structural diversity pertinent to several natural product classes as highlighted by syntheses of natural targets and analogs. Future work seeks to expand the number of these building blocks that can be made available for the synthetic chemists’ toolbox and to achieve a number of additional natural product total syntheses from them.