Key words
cryptoconcatone H - total synthesis - hidden symmetry - Mitsunobu reaction - Prins
cyclization - ring-closing metathesis
The cryptoconcatones are a family of structurally related natural products isolated
by Luo, Kong, and coworkers from the leaves and branches of Cryptocarya concinna, a monsoon evergreen found in subtropical mainland China.[1] Although the majority of cryptoconcatones contain terminal styryl and dihydropyranone
groups tethered by a linear trisubstituted, six-carbon chain, three members (cryptoconcatones
H, K, and L) possess a central 2,4,6-tetrahydropyranol ring (Figure [1]). Cryptoconcatone H was originally assigned to be 2,4,6-cis-tetrahydropyranol 1; however, combined computational and synthetic studies by Pilli and coworkers led
to structural revision of the putative structure to diastereomer 2, which differs in configurations at C2′ and C4′ (cryptoconcatone numbering).[2] The assignments of cryptoconcatones K (3) and L (4) have also been called into question by similar computational analysis, which, for
example, identified 5, possessing the opposite relative inter-ring stereochemistry as 3, as the most plausible structure of cryptoconcatone K.[2b]
Figure 1 Tetrahydropyranol cryptoconcatones from C. concinna
Given the structural ambiguity associated with the tetrahydropyranol cryptoconcatones,
we aimed to develop a stereochemically versatile synthetic approach to this class
of compounds. Herein, we report an efficient synthesis of (±)-1 and its C6 diastereomer that we believe may hold potential in facilitating the assignment
of cryptoconcatone K.
To date, synthetic studies on the tetrahydropyranol cryptoconcatones have resulted
the synthesis of ent-2
[2a] as well as two syntheses of 1.[2b]
[3] NMR data of synthetic materials has confirmed the reassignment of 2 as the correct structure of naturally occurring cryptoconcatone H. Strategically,
the reported approaches to 1 are related (Scheme [1]). In both, stereoselective tetrahydropyranol formation (Pd-catalyzed cyclization
of diol 6 by Pilli and workers and tandem deprotection/oxa-Michael addition of acetonide 7 by Csókás and Bates) preceded diastereoselective allylation to establish the C6 stereocenter
and ring-closing metathesis (RCM) to construct the dihydropyranone ring. The Pilli
and Bates syntheses of 1 were also similar in step count – requiring ten and eleven steps, respectively, from
known compounds. We envisioned an alternate approach to 1 that takes advantage of the latent symmetry about the core methylene-ring junction
that could significantly streamline the synthesis.[4] Specifically, we recognized that the tetrahydropyranol and dihydropyranone rings
could both be derived from homoallylic alcohols, by Prins cyclization and RCM, which
revealed symmetric diols syn- or anti-8, in which the inter-ring, C2′,C6-stereochemical relationship is fixed, as potential
starting points.
Scheme 1 Synthetic approaches to 1
Diols syn- and anti-8 were readily accessible from 1,1,3,3-tetramethoxypropane (9) as a chromatographically separable 1:1 mixture following the procedure of Samoshin
and coworkers (Scheme [2]).[5] The first step in our synthetic plan was a symmetry-breaking acylation of either
syn- or anti-8 in order to set up the RCM and ‘protect’ the C6 hydroxyl during the Prins cyclization.
Monoacylation of syn-8, which possesses the requisite C2′,C6-relative stereochemistry, with acryloyl chloride
led to only poor isolated yields of acrylate 10. Fortunately, synthetic equivalent 11 could be prepared in 67% yield from anti-8 by invertive Mitsunobu esterification with crotonic acid as reported previously by
Walleser and Brückner.[6] The Mitsunobu reaction of anti-8 failed to produce acrylate 10 if acrylic acid was used.
Scheme 2 Synthesis and monoacylation of syn- and anti-8
The key step in our synthesis was the Prins cyclization of homoallylic alcohol 11 with cinnamaldehyde to provide the 2,4,6-cis-tetrahydropyranol ring present in 1 (Table [1]). A limited number of Prins promotors have been reported that result in direct incorporation
of a hydroxyl substituent at the 4-position.[7] In our hands, three of these (Montmorillonite,[7a] Amberlyst-15,[7b] Bi(OTf)3
[7c]) led to no discernable formation of 12 from 11. However, we were pleased to find that use of either 40 mol% or 10 mol% phosphomolybdic
acid (PMA)[7d] in water successfully produced 12 as a single diastereomer,[8] albeit in low isolated yields. Conducting the reaction with 10 mol% PMA in CH2Cl2 led to significantly shorter reaction time and modestly increased yield, but a quantifiable
amount of the C4′ epimer, which could not be separated from 12, was also detected in the 1H NMR spectrum. Catalysis by Re2O7
[7e] proved to be much more effective, leading to the formation of a 10:1 mixture of
12 and its C4′ epimer in 69% yield in just 4 h when CH2Cl2 was used as solvent. In their initial study on Re(VII) catalysis of Prins cyclizations,
Tadpetch and Rychnovsky found that yields and stereoselectivities were strongly influenced
by solvent in reactions employing O3ReOSiPh3, and we made similar observations using Re2O7.[7e]
[9] Changing the solvent to CHCl3 had little effect on the outcome of the reaction; however, use of hexanes resulted
in a slower and lower-yielding reaction, but with excellent equatorial selectivity
for hydroxyl incorporation. We found that optimal isolated yield and selectivity (64%,
>20:1) could be achieved by using a 9:1 mixture of hexanes and CH2Cl2.[10] The rapid reaction times and high levels of diastereoselectivity observed in the
formation of 12 using commercially available Re2O7 as a catalyst are particularly notable.[11]
The stereochemical outcome of the Prins cyclization can be rationalized by considering
the mechanism outlined in Scheme [3]. Condensation of homoallylic alcohol 11 and cinnamaldehyde in the presence of Re2O7 leads to the formation of activated perrhenate ester 13, which readily ionizes to produce oxonium ion 14. Cyclization of 14 via a chairlike transition state in which both the C2′ and C6′ substituents adopt
pseudoequatorial orientations establishes the cis-stereochemical relationship between them that is typical of Prins reactions and results
in the formation of carbocation 15. Equatorial attack is sterically more accessible and overwhelmingly observed for
Prins reactions involving oxygen nucleophiles.[7] In our case, this leads to perrhenate ester 16, which undergoes reaction with perrhenic acid (produced during the formation of 13) to generate the desired 2,4,6-cis-tetrahydropyranol 12 and to regenerate the Re2O7 catalyst.
Table 1 Prins Cyclization of Homoallylic Alcohol 11 and Cinnamaldehyde
|
|
Scheme 3 Stereochemical rationale for the formation of tetrahydropyranol 12
Completion of the synthesis required only RCM of crotonate 12 to construct the dihydropyranone ring. Initial attempts using the first-generation
Grubbs catalyst in either the absence or presence of Ti(Oi-Pr)4
[12] led to the formation of cross-metathesis dimer 17 as the major product even at elevated temperatures (Scheme [4]).[13] Gratifyingly, use of the second-generation Grubbs catalyst in CH2Cl2 gave conversion of 12 into 1 as a viscous pale yellow oil in 74% yield.[14] Dimer 17 was observed by TLC in reactions using the second-generation Grubbs catalyst at room
temperature but it was cleanly converted into 1 upon heating to 40 ℃. NMR data of 1 produced in this manner is consistent with the extensive analysis presented for these
compounds by the Pilli[2b] and Bates[3] groups and is clearly different from naturally occurring cryptoconcatone H.
Scheme 4 RCM of crotonate 12
Using the same sequence of steps, C6 diastereomer 18 was prepared from syn-8 as shown in Scheme [5]. Yields for the individual steps were comparable to those in the conversion of anti-8 into 1. The 1H and 13C NMR spectra of 18 were identical to those reported for this compound by Bates.[3] With efficient access to both 18 and 1, efforts are currently underway to convert them into 3 and 5 (Figure [1]) in order to aid in the unambiguous assignment of cryptoconcatone K.
Scheme 5 Synthesis of 18 from syn-8
In summary, we have completed syntheses of the originally proposed structure of cryptoconcatone
H (1) and its C6 epimer 18 in just three steps and 32% and 36% yields, respectively, from (±)- and meso-1,8-nonadiene-4,6-diol. Our approach is distinct from those previously reported,
requires no protecting group or redox manipulations, highlights the unique utility
of Re2O7 as a Prins cyclization catalyst, and provides further confirmation of the reassignment
of cryptoconcatone H. Moreover, this work serves as an instructive illustration of
the step economy and efficiency created by the recognition of hidden symmetry in retrosynthetic
planning and the streamlining effect it can provide in organic synthesis.
