Diquinane (octahydropentalene) is the simplest bicyclic system among the various cyclopentanoids, and is present in numerous polycyclic natural products as a critical structural unit.[1] Compounds in which a carbocyclic ring (a three-, four-, five-, or six-membered ring, etc.) is fused to the ring junction of a diquinane moiety are called propellanes. Propellanes are highly strained systems and have useful applications in various fields of chemistry.[2]
Among various propellane systems, six-membered-ring-fused diquinanes have attracted a great deal of attention from the synthetic community, due to the challenges involved in their synthesis, and because they are found as core units in many natural products.[3] Also, natural products containing a fused 5/5/6-tricyclic system are known to exhibit a wide range of biological properties.[4]
Depending on the mode of fusion of the six-membered ring to the diquinane moiety, 5/5/6-tricyclics are classified into various types 1–6 (Figure [1]).[5] The six-membered ring can be fused to the diquinane in a 1,2- or a 1,3-fashion (Figure [1a]). The 1,2-fused 5/5/6-tricyclics are of three types: linear (1), angular (2), or propellane (3). Along similar lines, 1,3-fused 5/5/6-tricyclics are categorized into 1,3-fused (4), 1,3-bridged (5), or 1,3-propellane (6) types. Furthermore, the 1,2- and 1,3-fusions can exist in either cis and/or trans forms. Linear and angularly fused 5/5/6-tricyclic systems (1 and 2) exist in both cis forms (Figures [1b] and 1c; 1a and 2a) and trans-forms (1b and 2b), whereas propellanes (3) exist only in a cis form (3a; exo or endo) because of the stereochemistry of the cis ring junction (Figure [1d]). Along similar lines, the 1,3-fused type 4 exists in both cis (4a) and trans forms (4b) (Figure [1e]), whereas the 1,3-bridged (5) and 1,3-propellane (6) types exist in a cis form (5a and 6a; exo or endo) only (Figure [1f]).
Figure 1 Types of fused 5/5/6-tricyclic skeletons
All these skeletal types 1–6 (Figure [1]) are found in many natural products, such as alkaloids or terpenoids, and show useful biological properties.[6] Among these, the syntheses of skeletal types 1–5 have been well explored by several groups,[7] including our group. We have recently reported syntheses of skeletal types 1, 2, and 4 through metathesis approaches.[5]
[8] However, synthetic efforts toward 1,3-propellane-type skeletons 6 have been limited. The 1,3-propellane-type skeleton 6 is present in the sesquiterpenoid quadrone and its analogues (Figure [2]).[9] These are isolated from the fungus Aspergillus terreus and contain a complex structural unit with a propellane-type 5/5/6-tricyclic core 6 and they exhibit useful biological properties that include antitumor activity.[10] Also, they are prone to undergo skeletal rearrangements due to the presence of ring strain.[11] Hence, they have become attractive targets for the synthetic community to develop new synthetic strategies.
Figure 2 Representative examples of natural products containing 1,3-fused 5/5/6-tricyclic framework
There are a limited number of reports on the synthesis of these carbocycles, including their total synthesis.[9d]
[e]
[10e]
[12] Some selected methods are based on key reactions that include cyclization,[13] Claisen rearrangement,[14] acid-catalyzed rearrangements,[15] cationic rearrangement,[16] and skeletal synthesis.[9f]
[17] Moreover, the reported methods involve linear syntheses and large numbers of steps starting from commercially available materials or readily available building blocks. Additionally, we have not found any reports on the use of olefin metathesis for their synthesis. Therefore, we wish to report a rapid synthetic route to a propellane-type 5/5/6-carbocyclic framework 6 from readily available starting materials.
In view of our long-term interest in the use of C–C bond-formation reactions (e.g., olefin metathesis) to develop new synthetic strategies,[18] we envisioned a rapid synthetic approach to the propellane-type 5/5/6-carbocyclic framework 6 from a readily available building block, a 3β-vinyl tricyclic ketone that can be prepared from endo-dicyclopentadiene-1-one,[18f] by employing tandem metathesis as a key step. Also, we aimed to investigate the feasibility of metathesis between the olefin moieties present on the carbocyclic frameworks. Earlier reports on the feasibility of the metathesis approach are shown in Figure [3].[19] The trans- disposition of olefinic moieties at 1,3- or 1,2-positions seems to disfavor the ring-closing metathesis (RCM) sequence.
Figure 3 Previous reports on ring-closing metathesis of 1,2-, and 1,3-trans-disposed olefinic moieties
We aimed to study both an early-stage methylation sequence and an early-stage allylation sequence to construct the propellane-type 5/5/6-carbocyclic framework 6, which serves as a key intermediate for the synthesis of a core skeleton of quadranoids. An overview of the present work is shown in Figure [4].
Our retrosynthetic approach to target compound 8 is depicted in Figure [5]. Tricyclic compound 8 could be prepared by following a five-step synthetic sequence starting from the vinyl derivative 7. The target compound 8 could be synthesized through hydrogenation of the tricyclic ketone 9. The tricyclic ketone 9 could be assembled through a tandem metathesis of allyl derivative 10. The ring-junction allyl derivative 10 might be prepared from compound 7 through methylation followed by a bridgehead allylation sequence. The vinyl derivative 7, in turn, could be obtained from endo-dicyclopentadiene-1-one through a conjugate addition with vinylmagnesium bromide.
Figure 4 An overview of the present work
Figure 5 Retrosynthetic analysis to tricyclic ketone 7
Our journey began with the preparation of the key building block, the 3-vinyl tricyclic ketone 7, from commercially available endo-dicyclopentadiene by following a three-step synthetic sequence.[18f]
[20] Having prepared a substantial amount of the starting material 7, we subjected it to regioselective methylation to deliver the gem-dimethyl derivative 11 (91%; Scheme [1]). This compound was then treated with allyl bromide in the presence of sodium hexamethyldisilazide (NaHMDS) to furnish the corresponding ring-junction-allylated derivative 10 (89%).[18f] Next, this compound was subjected to tandem metathesis with the aid of the Grubbs II (G-II) catalyst (10 mol%) under an ethylene atmosphere in an attempt to obtain the 5/5/6-tricyclic compound 9. Instead, however, the tandem metathesis precursor 10 delivered the ring-opening metathesis product 12 (85%) instead of the desired tricyclic compound 9 (Scheme [1]).
Scheme 1 Attempted synthesis of tricyclic ketone 9 by a tandem metathesis route
Alternatively, compound 12 might be obtained by a two-step sequence involving ROM and allylation of the gem-dimethyl derivative 11 (Scheme [2]). To this end, the diquinane derivative 13 (88%) was obtained by exposing the norbornene derivative 11 to G-II catalyst (5 mol%) under an ethylene atmosphere. Next, compound 13 was allylated at the ring-junction carbon in the presence of allyl bromide and t-BuOK to obtain the corresponding allyldiquinane 12 (86%). However, when this compound was subjected to RCM with the aid of the G-II catalyst (10 mol%), the ring-closure product 9 was not obtained, even when the reaction was carried out under heated conditions for a prolonged reaction time, and the starting material was recovered.[18k]
Scheme 2 Attempted synthesis of tricyclic ketone 9 by a ROM/RCM sequence.
Because early-stage methylation (Figure [5]) failed to give the desired tricyclic system 9, we could not proceed further with a synthesis of the target compound 8. We therefore revised our retrosynthetic strategy to one involving a late-stage methylation to quadranoid skeleton 8 (Figure [6]). The target compound 8 might be synthesized from the tricyclic ketone 14. The key intermediate 14 might, in turn, be assembled by tandem metathesis of ring-junction-allylated derivative 15, which could be synthesized from vinyl derivative 7 by regio- and stereoselective allylation.
Figure 6 Revised retrosynthetic analysis toward the quadranoid skeleton 8
To realize this strategy, compound 7 was subjected to a regio- and stereoselective allylation at the α′-position (i.e., the ring-junction carbon) of the tricyclic system in the presence of a base to deliver the ring-junction-allylated derivative 15. For this purpose, we screened several reaction conditions by changing the base (mild → strong and small → bulky) and its loading at various temperatures (Scheme [3] and Table [1]). Initially, when we used potassium carbonate (K2CO3) as a base, the allylation did not occur at any position of compound 7, and the starting material was recovered. With 1.2 to 2.0 equivalents of potassium tert-butoxide (t-BuOK), the α-monoallyl product 16 was formed as a diastereomeric mixture (Table [1], entries 1 and 2). When the amount of this base was increased to 4.0 equivalents by adding it in two portions at 0.5 hour intervals, the gem-diallyl derivative 17 (major) was also formed along with compound 16 (minor), and 10% of the starting material was recovered (entry 3). At this stage, conventional column chromatography failed to separate these compounds. However, when lithium diisopropylamide (LDA), freshly prepared from BuLi and diisopropylamine (DIPA), was used as a base at a low temperature (–78 °C) and the reaction mixture was stirred for three hours, the reaction merely initiated, and no further progress was observed. We therefore screened several reaction conditions by changing the amount of base and the reaction temperature (entries 4–9).[18e]
Scheme 3 Synthesis of allyl Derivative 15 by regio- and stereoselective allylation of compound 7
Table 1 Screened Reaction Conditions for Monoallylation of 15
a
|
Entry
|
Base
|
Equiv
|
AllBrb (Equiv)
|
Temp (°C)
|
Time (h)
|
Convc (%)
|
Yieldd (%)
|
|
15
|
16
|
17
|
|
1
|
t-BuOK
|
1.5
|
1.2
|
rt
|
0.5
|
42
|
–
|
21
|
–e
|
|
2
|
t-BuOK
|
2.0
|
2.2
|
rt
|
1.0
|
63
|
0
|
38
|
–e
|
|
3
|
t-BuOK
|
4.0
|
3.2
|
rt
|
3.0
|
90
|
0
|
11
|
66e
|
|
4
|
BuLi/DIPA
|
1.1/1.2
|
1.4
|
–78
|
4
|
32
|
21
|
trace
|
–
|
|
–50
|
2
|
|
–30
|
2
|
|
5
|
BuLi/DIPA
|
1.1/1.2
|
1.4
|
–78
|
3
|
69
|
52 (15 + 16)
|
–
|
|
–40
|
2
|
|
–20
|
2
|
|
rt
|
12
|
|
6
|
BuLi/DIPA/HMPA
|
2.2/2.4/2.2
|
2.8
|
–78
|
2
|
40
|
28
|
trace
|
–
|
|
–50
|
2
|
|
–20
|
2
|
|
–10
|
2
|
|
7
|
BuLi/DIPA/HMPA
|
3.0/3.2/3.0
|
3.0
|
–78
|
2
|
52
|
36
|
trace
|
–
|
|
–45
|
2
|
|
–20
|
2
|
|
–10
|
2
|
|
8
|
BuLi/DIPA/HMPA
|
3.0/3.2/3.0
|
3.0
|
–78
|
2
|
63
|
42
|
trace
|
–
|
|
–10
|
3
|
|
9
|
BuLi/DIPA/HMPA
|
3.0/3.2/6.0
|
3.0
|
–78
|
2
|
78
|
32
|
12
|
16
|
|
–10
|
3
|
|
0
|
6
|
a All reactions were carried out in anhyd THF under N2 unless otherwise stated.
b Freshly distilled allyl bromide.
c Conversion of based on recovery of the starting material.
d Isolated yield.
e Toluene was used as the solvent.
When the reaction temperature was slowly increased to –30 °C by stirring the mixture for various times, monoallylation occurred at the desired α′-position, and the corresponding ring-junction-allylated product 15 (21%) was obtained exclusively, along with 68% recovery of the starting material (Table [1]; entry 4). Although the conversion improved to 69% on increasing the reaction temperature and reaction time, compound 15 and compound 16 were formed as a chromatographically inseparable mixture in 52% yield (entry 5). However, with the use of HMPA solvent as an additive, both the conversion and the yield of the desired allyl derivative 15 were improved to a certain extent (entries 6–9). We therefore carefully screened several reaction conditions by increasing the loading of the base in the presence of HMPA as an additive at various reaction temperatures and time intervals. Eventually, the desired compound 15 was obtained in a 42% yield under the optimized reaction conditions (entry 8). From these results, we concluded that the amount of base does not affect the regioselectivity whereas the reaction temperature does influence the regioselectivity to yield the requisite allyl product 15.[18e]
At this point, the formation of compounds 15 and 16 and their structures were initially confirmed by NMR analyses. The structure of 15 was confirmed, as it exhibited five CH2 and eight CH signals in the DEPT-135 NMR spectrum, whereas the structure of 16 was confirmed as it showed four CH2 and ten CH signals in the DEPT-135 NMR spectrum.
Later, the structures of 15 and 16 were also confirmed by various chemical transformations. When compound 15 was subjected to a methylation sequence, the gem-dimethyl derivative 10 was obtained. Furthermore, compound 15 also gave the triallyl derivative 18 upon allylation with allyl bromide in the presence of NaH. Here, the triallyl compound 18 was identical to the compound obtained from the vinyl derivative 7 (Scheme [4]).
Scheme 4 Structure-confirmation studies for compound 15
Along similar lines, when 16 was subjected to allylation sequence delivered the same compounds, i.e. the diallyl derivative 17 and the triallyl derivative 18 (Scheme [5]). When the ring-junction-allylated derivative 15 was exposed to the Grubbs I catalyst (G-I catalyst; 10 mol%) under an ethylene atmosphere overnight, only the ROM product 19 was obtained, along with 75% of the starting material (by NMR: 3:1 ratio, 69%). Unfortunately, compound 19 could not be isolated in a pure form at this stage by using conventional column chromatography. However, compound 15 furnished the rearranged product 14, along with 10% of the ROM product 19 (by NMR: 9:1 ratio, 86%) on treatment with the G-II catalyst (5 mol%) under an ethylene atmosphere overnight. A complete conversion was achieved by using 10 mol% of G-II catalyst in a comparatively short reaction time, giving the tricyclic ketone 14 in a good yield (88%).[21]
Scheme 5 Structure-confirmation studies for compound 16
Later, mixtures of 15 and 19 and of 14 and 19 were also converted into the ring-closure product 14 in yields of 66% and 82%, respectively, by using 10 mol% of G-II catalyst (Scheme [6]).
Scheme 6 Tandem metathesis route to a key intermediate, the tricyclic ketone 14
Next, the keto derivative 14 was subjected to methylation in the presence of MeI and t-BuOK in an attempt to prepare the gem-dimethyl derivative 20; however, this compound was not formed and, instead, a complex mixture was obtained. Alternatively, when compound 14 was subjected to an allylation sequence in the presence of KH and allyl bromide under reflux conditions, the rearranged product 22 was obtained instead of the gem-diallyl derivative 21 (Scheme [7]). As a result, we could not proceed further with a synthesis of compound 8 or its spiro analogue 23 by following this route. However, compounds 8 and 23 might be accessible from tricyclic compound 14 by hydrogenation followed by alkylation and/or an RCM sequence.[18e] These studies will be reported in due course.
Scheme 7 Attempted synthesis of dimethyl and diallyl derivatives 20 and 21
We have successfully assembled the propellane-type 5/5/6-carbocyclic framework 14 in a good yield by employing early-stage regio- and stereoselective allylation, followed by a tandem metathesis sequence. Compound 14 could act as a key intermediate to access target compounds 8 and 23. The present strategy involves commercially available inexpensive starting materials and operationally simple reactions. Consequently, this methodology might be useful in medicinal chemistry to design various drug-like molecules. Further investigations into the synthesis of quadrone natural products would be a useful exercise.
