Syntheses of cyclotriveratrylene analogues and their long elusive triketone congeners.

Although interest in cyclotriveratrylene and its analogues has been significant, limitations in the ability to adjust its structure fully have hampered studies into their complete range of properties. A unique strategy to synthesize a previously unobtainable cyclotriveratrylene analogue and a procedure which adjusts the inner methylene bridges of that material to a triketone is reported. A second triketone synthesis and computational studies indicate the parameters needed for success.

Since its correct structural assignment in 1965,[1,2] cyclotriveratrylene (1, Figure 1) has received significant attention due its unique, three-dimensional shape. Indeed, with a rigid bowl-like architecture as shown in the additional depiction, one also referred to as a “crown conformation”,[3] this compound and many of its close analogues (such as 2) have been studied in an array of host–guest complexes, several of which have led to exciting applications.[4] Among these are size-selective buckminsterfullerene hosting,[5] anion and cation binding,[6] as well as the generation of self-assembled organic frameworks.[7] When these units are explicitly combined into a dimeric framework, as with cryptophane 3,[8] entirely new opportunities for binding result. Compound 3 and its close analogues, for instance, show a particularly high affinity for xenon, a property that is being exploited for improved MRI-based diagnostic techniques, including the detection of matrix metalloproteinase activity.[9]

Figure 1

Structure of cyclotriveratrylene (1) and analogues (2 and 3).

Yet, despite the increasing number of applications for this class of molecules, the ability to access a full array of structural congeners to maximize properties remains an unattained goal. At present, elaborating the aromatic rings, or “outer rim” of cyclotriveratrylene and its analogues, has proven achievable. For instance, aryl derivatives of 2 have been accessed that allow for the controlled appendage of additional groups to obtain extended-cavity cyclotriveratrylenes with greater affinity for certain guest molecules.[10] By contrast, synthetic manipulation of the methylene bridges, or “inner rim” of these molecules, to other cyclotriveratrylene congeners has proven far more challenging.

For instance, Cookson and co-workers reported in 1968 that the oxidation of cyclotriveratrylene (obtained in a single step from alcohol 4 as shown in Scheme 1) with chromic acid could afford its corresponding monoketone (5) and a small amount of symmetric triketone 6.[11] However, the Baldwin group refuted this claim later that same year, proving that spirocycle 8, not 6, was the obtained fully oxidized product; that material likely resulted through the indicated transannular rearrangement of 6 by way of intermediate 7.[12] Indeed, a similar result was observed in 1997 upon oxidation of the parent unfunctionalized [1,1,1]orthocyclophane 9,[13] with mono- and diketone products 10 and 11 obtained alongside an analogous spirocycle (12).[14] In fact, no report of successfully accessing the triketone of cyclotriveratrylene or any related analogue has ever been issued, with some authors suggesting that such an outcome is unachieveable.[12,13,15] Herein, we disclose a unique route to synthesize a novel cyclotriveratrylene-like analogue, one that may prove extendable to other synthetically challenging congeners, and a procedure that successfully oxidized that material to a triketone. In addition, we show that a second cyclotriveratrylene-like substrate is capable of forming a similar triketone, establishing in the process what we believe are the structural guidelines for when triketones can be formed with electron-rich substrates in preference to the spirocycles observed in past studies.

Scheme 1

Synthesis of Cyclotriveratrylene (1) and Its (a) Oxidation to Spirocycle 8, Presumably through Triketone 6, and (b) Similar Reactivity Observed with 9

Our interest in this unique class of compounds derived from our efforts[16] to synthesize 9-membered rings within the resveratrol natural products family, of which α-viniferin (13)[17] is representative (Scheme 2a). In one design, we viewed triketone 14 as a key intermediate, hoping that simultaneous elaboration of all three of its carbonyl moieties would lead to the rapid construction of the natural product. Our first attempted route to that triketone invoked cyclotriveratrylene analogue 15 as a precursor, hoping we could find conditions capable of effecting its oxidation without rearrangement to a spirocycle. We began our efforts by attempting to trimerize 16, in line with the precedents discussed earlier (cf. Scheme 1). Unfortunately, we did not observe any 9-membered ring formation.

Scheme 2

(a) Inspiration for 9-Membered Ring Exploration and (b) Synthesis and Attempted Oxidation of Cyclotriveratrylene Derivative 15

This outcome was not unexpected since all successful instances of one-step trimerizations involve substrates with an electron-donating substituent para to the benzylic alcohol, a group that likely assists in cation generation and thus increases the reaction rate toward desired processes. As a result, we embarked on an alternate, stepwise strategy to achieve the desired intermediate as shown in Scheme 2b. Here, site-specific monobromination and protection of benzylic alcohol 16 gave coupling precursor 17 in 94% overall yield. This material, upon lithiation, was then added into aldehyde 18 to give bis-benzylic alcohol 19. Subsequent reductive removal of the hydroxyl group then furnished the methylene bridge of 20, with repetition of the same general sequence ultimately affording cyclization precursors 21/22. As previously reported,[16,18] we found that treatment of 21 with BCl3 accomplished the desired cyclization to 15 in 60% yield through a Friedel–Crafts cyclization,[19] with X-ray crystallographic analysis revealing that 15 adopts a crown conformation much like cyclotriveratrylene.

Overall, this result was gratifying in that not only had the core carbon framework of our targeted natural product been formed, but we had also synthesized an entirely new cyclotriveratrylene analogue. Furthermore, the facility of the sequence and possibilities for interchanging the three aromatic coupling partners could potentially serve as a roadmap for the construction of other cyclotriveratrylene analogues unobtainable by the conventional, one-step acid-catalyzed trimerization method. Unfortunately, attempted oxidation of this material (15) with chromic acid afforded minimal oxidation of any type with no 14 observed. Attempts to deploy other conventional benzylic oxidations (such as Pd(OAc)2/t-BuOOH[20] and DDQ) were likewise met with poor conversion. Given the similar electronic environment of 15 and cyclotriveratrylene (1), we reasoned that the positioning of a methoxy group ortho to the methylenes within 15 might be responsible, with its steric bulk effectively blocking any incoming oxidant.

Given this failure, we next sought to preinstall the needed carbonyls before 9-membered ring formation so as to obviate the need for any benzylic oxidation postcyclization. Starting from the previously synthesized TIPS-protected bromide 19 (Scheme 3), protection of its free hydroxyl with a PMB group, followed by lithiation and addition of the final aromatic ring, smoothly furnished triaryl 23 in 74% yield. Next, TIPS removal using TBAF, oxidation of the two free alcohol groups with Dess–Martin periodinane, and PMB excision followed by in situ alcohol oxidation smoothly delivered diketoaldehyde 24, poised to attempt closure to a 9-membered ring.[21]

Scheme 3

Attempted Cyclization of Preoxidized Intermediate 24

Our explorations focused on treating diketoaldehyde 24 with various Lewis and Brønsted acids. Regrettably, these attempts consistently afforded the same product after subsequent benzylic oxidation with MnO2, namely spirocyclic lactone 25 as confirmed by X-ray crystallographic analysis of a demethylated analogue. Unclear is whether this spirocycle formed directly or if 9-membered ring formation occurred first and was followed by a transannular rearrangement of the type detailed in Scheme 1. In any event, resigned to the unattainability of the desired triketone 14 in line with past reports, we engaged other strategies toward the synthesis of 9-membered ring-containing natural products, ultimately synthesizing a related natural product through a unique Friedel–Crafts cyclization.[16]

Quite some time after our initial pursuit of the triketone in the context of Scheme 2, a publication appeared which reported new conditions for the oxidation of cyclotriveratrylene (1) using KMnO4 and MnO2 in refluxing pyridine.[22] Although no triketone was obtained, the authors did report a significantly higher yield of the mono- and diketones than any previous approach. With some 15 still in hand, we attempted its oxidation using these same conditions. Much to our surprise, we obtained a small amount of what appeared to be a fully oxidized product distinct from the previously obtained spirocycle 25. Optimization studies led to a slightly modified protocol which allowed us to isolate that material in higher yield (Scheme 4), and with the support of NMR, mass spectrometry, and a preliminary X-ray crystal structure (see the Supporting Information), we were able to identify this product as the elusive triketone (14).

Scheme 4

(a) Successful Oxidation of 15 to 14 and (b) Application to a Two-step Synthesis of Triketone Analogue 29

As indicated, diketone 26 is the favored product of this process, and interestingly, despite recovery of a small amount of starting material 15, no trace of the monoketone was observed.[23] Significantly, resubmission of diketone 26 to the same conditions gave additional triketone 14 in 20% yield along with 30% recovered 26. On the basis of these results, the first and third oxidations would appear to occur slowly, while the second oxidation to the diketone is more rapid.[24,25] X-ray crystallography revealed that diketone 26 and triketone 14 possess a saddle-like conformation (shown explicitly for 14).

On initial reflection, it may seem curious that we did not see rearrangement to the spirocyclic lactone 25 (cf. Scheme 3) from triketone 14 as observed previously with very similar substrates; we can stipulate that no observable trace of such a product was obtained having already synthesized 25 by other means with full knowledge of its spectral and chromatographic properties. We originally postulated that the steric penalties of the requisite intermediates needed in such a rearrangement, particularly those of type 7 (cf. Scheme 1), might be prohibitively high when methoxy groups are ortho- to the methylene positions, as compared to those previous cases where no such substitution was present. To test that theory, a series of DFT calculations (B3LYP/6-31g*) were performed considering the energies of the hypothetical triketone 6, triketone 14, and the corresponding intermediates following transannular attack of both of those triketones under acidic conditions. In the case of 6 to 7, that process is downhill energetically by ∼3 kcal/mol; by contrast, the same process with 14 is uphill by 0.7 kcal/mol. The difference appears not to be due to sterics, but rather stabilization of the protonated triketone by the neighboring o-methoxy groups in 14 that causes substantial twisting, forcing the attacking aryl group to be orthogonal to the protonated ketone (see the Supporting Information for structures), thereby preventing the transannular rearrangement. To support this overall theory for successful triketone formation from materials possessing a methoxy group ortho to a methylene, and assuming the relative energies are similar under the basic conditions of our developed procedure, we submitted cyclotriveratrylene analogue 29, obtained in one step from 28,[26] to the same oxidation protocol. As shown, our prediction for oxidation without rearrangement proved correct, with triketone 30 formed in 8% yield, just a mere two steps from commercially available alcohol 28.[27]

In conclusion, we have developed a stepwise synthetic strategy toward cycloveratrylene analogue 15, a compound unavailable through the conventional single step approach previously developed for cyclotriveratrylene and other related derivatives. We believe this strategy may be generalized for other cyclotriveratrylene analogues that do not possess the necessary substitution pattern for a one-step, acid-catalyzed trimerization. Furthermore, we have accomplished the first reported examples of synthesizing cyclotriveratrylene-like triketones. Not only does this key finding address a longstanding challenge in the field, it also illustrates the conditions and structural requirements needed for the formation of such materials, opening the door to further elaboration and study of these previously unexplored “inner rim” cycloveratrylene derivatives for new applications.