|Home | About | Journals | Submit | Contact Us | Français|
The reaction of dodec-11-ene-1,6-diynes or their heteroatom congeners with a hydrosilane catalyzed by Rh(acac)(CO)2 at ambient temperature and pressure of CO gives the corresponding fused 5-7-5 tricylclic products, 5-oxo-1,3a,4,5,7,9-hexahydro-3H-cyclopenta[e]azulenes or their heteroatom congeners, in excellent yields through a unique silicon-initiated cascade carbonylative carbotricyclization (CO-SiCaT) process. It has also been found that the 5-7-5 fused tricyclic products can be obtained from the same type of enediynes and CO through a novel intramolecular [2+2+2+1] cycloaddition process. The characteristics of these two tricyclization processes and fundamental differences in their reaction mechanisms are discussed. This novel higher-order cycloaddition reaction has also been successfully applied to the tricyclization of undeca-5,10-diyn-1-als, affording the corresponding 5-7-5 fused ring products bearing a 7-membered lactone moiety. Related [2+2+2] tricyclizations of enediyne and diynal substrates are also discussed. These newly discovered reactions can construct multiple bonds all at once, converting linear starting materials to polycyclic compounds in single step. Thus, these new processes provide innovative routes to functionalized polycyclic compounds that are useful for the syntheses of natural and unnatural products.
It would be ideal if the synthesis of complex target molecules can be achieved quickly, quantitatively and selectively by simple operation from readily available starting materials.1 The transition metal-catalyzed carbocyclization2 and cycloaddition3–4 reactions are among the synthetically most useful processes for rapidly increasing molecular complexity. Many of these processes are symmetry forbidden and impossible or difficult to realize in the absence of proper catalysts. Among various metals, nickel, ruthenium, cobalt, rhodium and palladium catalysts have been commonly used to promote these two processes.5–7 Since many bioactive compounds8 include fused ring systems in their structures, these polycyclization reactions warrant extensive studies. In fact, various polycyclization processes have been employed for the construction of natural and unnatural fused ring systems that can be further elaborated into specific targets. For instance, in the last two decades, considerable advances have been made in the development of higher-order cycloaddition reactions such as [4+3]9, [5+2]10, [6+2]11, [4+2+2]12 and [5+2+1]13 processes. Thus, transition metal–catalyzed carbocyclization and higher order cycloaddition reactions14 provide powerful methods for the construction of complex polycyclic systems.2c, 15–16
In the course of our investigation into the development of silicon-initiated cyclization processes, intramolecular silylformylation,7a silylcyclocarbonylation (SiCCa),17–18 silylcarbocyclization (SiCaC),18–19 silylcarbobicyclization (SiCaB)18d,e and silylcarbotricyclization (SiCaT)18c,d have been discovered. In 1996, we reported a stereospecific cascade SiCaC reaction20 of dodec-6-ene-1,11-diynes (Scheme 1). In this reaction, although intramolecular carbometalation of the vinylsilane moiety of intermediate A1 with the [Rh]-vinyl moiety to form the third ring was conceptually possible, the third cyclization did not take place. Thus, a simple reductive elimination preferentially occurred to give the corresponding bis(cyclopentyl) 2.
In the case of alkatriynes 3, however, silylcarbotricyclization (SiCaT) took place to give the fused tricyclic compounds 4a and 4b in good to excellent yields (Scheme 2)21 Reactions catalyzed by rhodium carbonyl clusters showed high selectivity to 4a.
As a part of our studies on the silicon-initiated cascade carbometalations, we looked at the reaction of dodec-11-ene-1,6-diynes 5. The reaction of the enediynes 5 with PhMe2SiH in the presence of a rhodium catalyst under ambient pressure of CO afforded, to our surprise, fused 5-7-5 ring products 6 incorporating a CO molecule and our preliminary study was communicated.22 This novel silicon-initiated cascade carbonylative carbotricyclization (CO-SiCaT) of enediynes promoted by rhodium catalysts provides a powerful method for the rapid construction of functionalized fused 5-7-5 ring systems (Scheme 3).
During our studies on the scope and limitation of the CO-SiCaT reaction, we serendipitously found that the reaction of 1-substituted dodec-11-ene-1,6-diyne 7 catalyzed by [Rh(COD)Cl]2 in the absence of a hydrosilane gave carbonylative tricyclization product 8 in good to excellent yields (Scheme 4). This observation turned out to be the first example of [2+2+2+1] cycloaddition reaction and preliminary results were communicated.23
Although these two processes, i.e., CO-SiCaT and [2+2+2+1] cycloaddition, lead to the formation of the same type of products, but these reactions proceed through fundamentary different mechanisms.22–23 We describe here a full account of our study on the novel Rh-catalyzed CO-SiCaT and intramolecular [2+2+2+1] cycloaddition processes of enediynes, involving CO as the single carbon component.
The scope of the CO-SiCaT reaction has been investigated systematically by looking at the key elements and reaction variables involved in this process. The enediyne 5a was used as the substrate for this study (Scheme 5). Results are summarized in Table 1. Reaction of enediyne 5a with PhMe2SiH (2.0 equiv.) catalyzed by Rh(acac)(CO)2 (1 mol %) in toluene under ambient pressure of CO at 70 °C for 1 h gave cyclopenta[e]azulene 6a and bis(cyclopentylidene) 9 as major products and aldehyde 10 as a minor product in 70% overall isolated yield (6a:9:10 = 36:43:21) (entry 1). The reaction under higher CO pressure (10 atm) disfavor the formation of 9 (entry 2). Addition of P(OPh)3 completely suppresses the formation of 10 (entry 3), while the use of PPh3 as the additive substantially favors the formation of 10 and reduces the yield of 6a (entry 4). The use of THF as the solvent at lower temperature (22 °C) gives better product selectivity for 6a (entry 5). Then, the reaction at lower concentration (0.015 M) dramatically increased the product selectivity, yielding 6a as the sole product (entry 6).
The observed remarkable dilution effects on the product selectivity appears to be, in part, attributed to the relative concentration of CO in solution to those of catalyst and hydrosilane, i.e., the relative concentration of CO is higher in dilute solution if the amounts of the catalyst and hydrosilane are kept constant. Also, this observation provides crucial information for the mechanism of this unique reaction. Although the mechanism of this novel carbocyclization processes is proposed and discussed later in the text (see Scheme 6), the formations of 9 and 10 should involve a bimolecular process at the final reductive elimination step, while that of 6a does not, i.e., the formation of 6a is a unimolecular process. Accordingly, the formation of 6a should be favored at higher dilution, while those of 9 and 10 should be disfavored. In addition, a stoichiometric amount of a hydrosilane should not be necessary for the formation of 6a since this product does not incorporate the hydrosilane component. This indicates that only sub-stoichiometric amount of hydrosilane would be sufficient for this reaction to proceed and such a condition should suppress the formations of 9 and 10. Thus, we carried out the reaction with a sub-stoichiometric amount of hydrosilane, and indeed our analysis and prediction was proven correct. Thus, the reaction with 0.5 equiv. of PhMe2SiH at high dilution conditions (0.015 M ) gives 6a exclusively in high overall yield (entries 7, 10, and 11). The reaction with 0.1 equiv. of the hydrosilane was 100% selective to the formation of 6a, but does not give complete conversion in 96 h (entry 8). In the absence of PhMe2SiH, no reaction takes place after 96 h, recovering enediyne 5a (entry 9). Thus, PhMe2SiH is proven necessary for this reaction to occur although no silyl group is incorporated into product 6a. The best result is achieved by using 0.5 equiv of PhMe2SiH at 0.015 M concentration, giving 6a as the single product in 92% isolated yield (entry 7).
As for the catalyst, Rh(acac)(CO)2, Rh4(CO)12, and [Rh(CO)2Cl]2 show similar efficacy. For hydrosilanes in this process, the reactions with Ph2MeSiH and (EtO)3SiH give 6a exclusively although the reactions using (EtO)3SiH are slower than those with PhMe2SiH. Trialkylsilanes behave differently, that is, the reaction with Et3SiH gives a 9:1 mixture of 6a and 9-TES in low conversion, while that with t-BuMe2SiH does not give any conversion after 96 h. Accordingly, PhMe2SiH is the best hydrosilane, so far, for this process.
As described above, the optimal reaction conditions for the CO-SiCaT reaction include the use of Rh(acac)(CO)2, 0.5 equiv. of PhMe2SiH at 0.015 M concentration in THF at room temperature and ambient pressure of CO. In order to examine the scope and limitations of this reaction for functional group tolerance, a variety of substrates containing ether, ester, hydroxyl, and sulfonamide groups have been subjected to the optimal conditions for the CO-SiCaT process. Results are summarized in Table 2.
As Table 2 shows, these functional groups and heteroatoms are well tolerated in this reaction to give the corresponding fused 5-7-5 tricyclic compounds 6 in good to excellent isolated yields. The reaction of 5d gives 6d that contains two fused pyrrolidine moieties in 85% isolated yield (entry 4), while the reaction of 5b that contains dipropargyl ether as well as propargyl allyl ether moiety is sluggish to give 6b in 50% yield after 48 h (entry 2). However, the reactions of unsymmetrical substrates bearing a malonate group as well as a dipropargyl ether or a sulfonamide moiety give the corresponding 5-7-5 products (6c,e,f) in excellent yields (entries 3, 5 and 6). In the case of 5g, bearing a t-butyl carbamate group, the reaction requires 40 °C to give 6g in high yield (entry 7). The substrates 5g-l derived from 5a bearing ethers, hydroxyl groups and ketals, are also well tolerated in this reaction, which give the corresponding 6g-l in good to high yields (entries 8–12).
The most plausible mechanism for this novel carbonylative carbotricyclization, which can accommodate all observed results, is illustrated in Scheme 6. The reaction begins with the insertion of the terminal alkyne moiety into the Si-[Rh] bond of the hydrosilane-Rh oxidative adduct.22 Carbocyclization then occurs to give (Z)-dienyl[Rh](H) intermediate A6. Because of the steric hindrance between the vinylsilane and the vinyl-[Rh] moieties, A6 isomerizes to C6 via B6 through the “Ojima-Crabtree mechanism”.7 Subsequent carbocyclization to D6, followed by reductive elimination, should give 9. The CO insertion to D6 gives acyl-[Rh](H) intermediate E6. Reductive elimination of E6 should yield aldehyde 10. Carbocyclization of E6 gives tricyclic intermediate F6 that has the silicon and the [Rh] moieties in syn positions. Subsequent -silyl elimination18e,21 should take place to afford the fused 5-7-5 tricyclic product 6 and regenerate the active catalyst species, R3Si[Rh](H). The proposed mechanism can nicely accommodate the observed three products under non-optimized conditions, dilution effects, and the fact that only substoichiometric hydrosilane is needed to yield 6.24
Next, we investigated the reaction of enediyne 11 in which the alkene group is a part of a cyclic system (Scheme 7). If the reaction proceeds via CO-SiCaT, fused 5-7-6-5 tetracyclic product 12 should be formed. However, the reaction of 11 under the standard CO-SiCaT conditions did not give 12, but afforded tricyclic product 13 in 63% isolated yield. As Scheme 7 illustrates, the result indicates that after the second carbocyclization, forming intermediate A7, -hydride elimination took place exclusively in intermediate B7 over CO insertion to form the corresponding acyl-[Rh] intermediate. This type of facile -hydride elimination in the cyclohexyl-containing enyne 15 was previously observed in the SiCaC of a propargyl(cyclohexyl)malonate, wherein the -hydride elimination was much faster than reductive elimination to give 16 (Scheme 8).19 The yield of 13 was improved to 79% when two equivalents of PhMe2SiH to 11 were used. The reactions under different conditions, i.e., (i) at higher temperatures (up to 70 °C), (ii) higher CO pressures (10–30 atm), (iii) use of Rh4(CO)12 as catalyst, (iv) use of other hydrosilanes (Ph2MeSiH, Et3SiH) or (v) addition of P(OPh)3, did not improve the yield of 13, but rather decreased the yield or slowed down the reaction (under higher CO pressures). It should be noted that 13 was formed as single stereoisomer (the stereochemistry was unambiguously determined based on NOE, i.e., 14 was not formed, which is consistent with the proposed mechanism in Scheme 6). This result shows a limitation of the CO-SiCaT process in the substrate type. However, the same result strongly suggests that the formation of fused 5-7-6-5 tetracyclic product could be achieved by simply blocking the facile -hydride elimination with appropriate introduction of a hetero-atom or gem-disubstitution in the cyclohexyl moiety of enediyne 11. Accordingly, further studies along this line are actively underway in these laboratories.
We also investigated the reaction of an enediyne 7a, which bears a 2-butynyl group instead of a propargyl group at one terminal to examine the effect of this terminal methyl group on the reactivity and mode of the reaction. To our surprise, the attempted CO-SiCaT reaction of 7a resulted in the formation of SiCaT products 17a and 18a without CO insertion (Scheme 9). Results are listed in Table 3. The reaction of 7a under the standard CO-SiCaT conditions for 24 h did not give any product and enediyne 7a was recovered (entry 1). However, when the reaction using 0.3 equivalents of PhMe2SiH was run at 70 °C in toluene for 3 h, SiCaT21 products 17a was obtained in 96% yield (entry 3). The reaction using 0.1 equivalents of the hydrosilane for 24 h gave a 1:1 mixture of 17a and 18a in 90% yield (entry 4). The use of higher CO pressure (25 atm) suppressed the reaction (entry 5). No reaction took place even after 72 h in the absence of the hydrosilane (entry 6), which clearly indicates that the hydrosilane is necessary for the reaction to occur. The reaction using Rh4(CO)12 as catalyst gave a 2:1 mixture of 17a and 18a in only 50% yield (entry 2).
The proposed mechanism for the formation of SiCaT products 17a and 18a is shown in Scheme 10. The reaction begins in the same manner as that of CO-SiCaT (Scheme 6) to form intermediate A10 after the first silylcarbocyclization. However, because of the methyl substituent, the Z –> E isomerization does not occur, which makes a sharp contrast with the CO-SiCaT mechanism (Scheme 6). Since the silyl group and the [M](H) moiety (M = Rh) is very close, a s-bond metathesis takes place to form metalacycle intermediate B10, liberating a hydrosilane. There are two possible ways that the olefin moiety inserts into the [M]-C bonds of B10, generating 5-7-5 tricyclic metallacycle C10 or 5-7-6 tricyclic metallacycle D10. Reductive elimination of C10 or D10 yields fused 5-6-5 tricyclic diene-[M] complex E10. When there is sufficient hydrosilane in the system, E10 reacts with the hydrosilane to form 17a and regenerate the active catalyst species R3Si[Rh](H), while aromatization via dehydrogenation competes to give 18a, generating dihydro-[M] species. Then, the [M](H)2 species reacts with a hydrosilane to liberate molecular hydrogen and regenerate the active species R3Si[Rh](H).
The proposed mechanism accommodates the results listed in Table 4 very well, and explains why CO-SiCaT did not take place for enediyne 7a. However, it looks possible that an apparent CO-SiCaT product can be obtained from intermediate C10 via CO insertion, while intermediate D10 would undergo facile reductive elimination. The fact that no CO incorporation took place in this reaction may suggest that this reaction proceeds through D10 rather than C10.
A possible intermediacy of metallacycle C10 encouraged us to explore hitherto unknown intramolecular [2+2+2+1] cycloaddition route to fused 5-7-5 tricyclic systems. The formation of metallacycles like B10 from 1,6-diynes is a rather common reaction pathway in carbocyclizations.2 Accordingly, we set out to search for reaction conditions and variables, which might promote the novel intramolecular [2+2+2+1] cycloaddition of enediyne 7 with carbon monoxide. Gratifyingly, we quickly found that the desired fused 5-7-5 tricyclic product 8 was formed through the novel [2+2+2+1] cycloaddition in the absence of hydrosilane although the reaction was accompanied by intramolecular [2+2+2] cycloaddition product 17 (Scheme 11). Thus, we immediately carried out optimization of this new higher order cycloaddition process. Results are summarized in Table 4.
As Table 4 shows, attempted CO-SiCaT reaction of 7 in the presence of PhMe2SiH catalyzed by Rh(acac)(CO)2 or [Rh(COD)Cl]2 (entries 1 and 3) failed to give 8, but yielded the 5-6-5 fused ring product 17 exclusively in 70% and 96% isolated yields, respectively. Under the same conditions but without hydrosilane (entries 2, 4), 8 was formed in 49% yield using Rh(acac)(CO)2 as catalyst and 77% yield when [Rh(COD)Cl]2 was used as catalyst. These sharply contrasting results in the presence and absence of hydrosilane clearly indicate that the CO-SiCaT and intramolecular [2+2+2+1] cycloaddition processes are fundamentally different although the same type of products is formed. Since [Rh(COD)Cl]2 was a better catalyst than Rh(acac)(CO)2, we used this catalyst for further optimization. We examined [Rh(COD)Cl]2 as potentially efficient catalyst for this novel [2+2+2+1] cycloaddition with CO because it has been shown that [Rh(COD)Cl]2 is an efficient catalyst for the catalytic Pauson-Khand reactions of enynes.24 We looked at possible effects of concentration by employing 0.015 M, 0.1 M, and 0.4 M concentrations, and found that 0.1 M was the best substrate concentration and either lower or higher concentrations decreased the yield and product selectivity of 8. As for the solvent for this reaction, DMF and EtOH (entries 7 and 8) yielded only 17. The use of dioxane and chloroform gave a mixture of 8 and 17 (entries 5 and 6). Other solvents such as THF, dimethoxyethane and dibutyl ether were also examined at different temperatures, but the product selectivity for 8 was more or less in the same range. Accordingly, toluene (entry 4) and dichloroethane (entry 9) gave so far the best yield and selectivity for 8 (92% yield and 77% selectivity). However, we found that the bubbling of CO gas into the reaction mixture dramatically increased the selectivity for the formation of 8, i.e., the reaction gave 8 in 88% isolated yield with less than 4% of the side product 17 (entry 10). This protocol was also effective for the reaction in toluene, giving 8 in 80% isolated yield accompanied by a small amount of 17 (9%). Interestingly, the reactions at higher CO pressures disfavored the formation of carbonylative product 8 (entries 12 and 13). Next, we screened other catalysts for this reaction. The reaction using Co2(CO)8 in the presence of molecular sieves (entry 22) or Rh4(CO)12 (entry 24) as catalyst yielded 17 exclusively. Also, the reaction catalyzed by Co2(CO)8-P(OPh)3 or [Ir(COD)Cl]2 gave only 17. Other catalysts, [Rh(dppp)(CO)Cl]2, [Cp*RhCl2]2, [Rh(COD)2]SbF6 and Rh(PPh3)3Cl/AgSbF6, strongly favored the formation of 17 accompanied by a minute amount of 8. The use of [Rh(CO)2Cl]2 (entry 14) and Rh(PPh3)3Cl (entry 16) gave a mixture of 8 and 17, but [Rh(CO)2Cl]2 was a much better catalyst than Rh(PPh3)3Cl although the efficacy was considerably lower than that of [Rh(COD)Cl]2.
Next, a variety of 1-substituted dodec-11-ene-1,6-diyne were employed to examine the scope of this reaction, especially for its functional group tolerance. Various enediynes 7 containing ether, sulfonamide, carbamate, ester, and ketal groups were examined. In addition, enediynes 7 bearing a methyl, phenyl or TMS as the substituent of the terminal acetylene moiety were employed as well. Results are summarized in Table 5. These functional groups and heteroatoms are generally well tolerated in this process, affording the corresponding 8b-n in good to excellent yields. For the substrates 7d (entry 4), 7h (entry 8), 7i-l (entries 9–12), and 7n (entry 14), [2+2+2+1] adducts 8d, 8h, 8i-l, 8n are the exclusive products, respectively. The reactions of 7e-g having a p-toluenesulfonamide moiety gave 8e-g as predominant products, but accompanied by 20e-g as minor products (entries 5–7), while the reaction of 7h bearing a t-Boc-amino moiety afforded 8h exclusively (entry 8). The reaction of 8m (R = Ph) gave a ca. 2:1 mixture of 8m (major) and 20m (minor), while that of 7n (R = TMS) yielded 8n as the sole product. Although Cl(CH2)2Cl is the preferred solvent in most cases, the use of toluene gave better selectivity in some cases. For example, the reaction of 7j in Cl(CH2)2Cl gave 8j in 75% isolated yield, but the yield of 8j was improved to 91% (86% isolated yield) when the solvent was changed to toluene (entry 10). More drastic solvent effect was observed in the reaction of 7m. The reaction in Cl(CH2)2Cl gave 8m in only 27% yield (20m: 61% yield), while the same reaction in toluene reversed the product selectivity, affording 8m and 20m in 56% and 28% isolated yields, respectively (entry 13). Apparently the use of toluene in these cases favored the CO insertion process.
When a 3-pyridinyl group was introduced to the 1 position, no reaction took place and the starting material 7o was recovered intact. No reaction occurred with substrate 7q, either. Free hydroxyl group in the substrate 7p was not tolerated in this reaction, resulting in very messy mixtures without a trace of 8p or 17p. Accordingly, these enediyne substrates set a limitation in the scope of the Rh-catalyzed [2+2+2+1] cycloaddition process. However, it should be noted that 7o did undergo the SiCaT reaction (see Scheme 10) to give 17o in 88% isolated yield, as shown in Scheme 12.
A proposed mechanism of the novel intramolecular [2+2+2+1] cycloaddition reaction is illustrated in Scheme 13. This reaction is believed to proceed through a series of metallacycles. The proposed mechanism includes (i) selective coordination of the diyne moiety of enediyne 7 to the active Rh-catalyst species, forming metallacycle A13([2+2+M]), (ii) insertion of the olefin moiety of 7 into the Rh-C bond to form 5-7-5 fused tricyclic rhodacycle B13 ([2+2+2+M]), (iii) coordination of CO to the [Rh] metal followed by migratory insertion of CO into the Rh-C bond to form 5-8-5 rhodacycle C13 or D13 ([2+2+2+1+M]), and (iv) reductive elimination to form [2+2+2+1] cycloadduct 8 and regenerate the active Rh-catalyst species. Reductive elimination from the 5-7-5 rhodacycle B13, prior to the CO insertion, gives [2+2+2] cycloadduct 17.
As discussed above, we have discovered two different reactions that give fused 5-7-5 tricyclic products, incorporating carbon monoxide. Although these two reactions yield the same type of products, the mechanisms of these reactions are fundamentally different, i.e., the CO-SiCaT reaction proceed through sequential carbocyclizations (Scheme 6), while the [2+2+2+1] cycloaddition reaction involves sequential metallacycle formations (Scheme 13). We have shown above a number of compelling cases in which two reaction conditions gave markedly different outcomes. However, those differences were observed for the reactions of enediynes 7 bearing a substituent at the terminal acetylene moiety, wherein the CO-SiCaT reaction did not take place and in most cases SiCaT reaction occurred to afford the corresponding fused 5-6-5 tricyclic products 17 and/or 18. Accordingly, a natural question arises, i.e., what will happen when enediynes that do not have a substituent at the terminal acetylene moiety for which the CO-SiCaT has taken place? To answer this question and confirm the fundamental difference between the CO-SiCaT and [2+2+2+1] reactions, we looked at the reactions of enediyne 5a under the optimized reaction conditions of these two processes (Scheme 14). Results are summarized in Table 6.
As Table 6 shows, the CO-SiCaT reaction of 5a was carried out with PhMe2SiH (0.5 equiv.) at ambient temperature and pressure of CO in THF or toluene in high dilution (0.015 M) for 24 h using Rh(acac)(CO)2 or [Rh(COD)Cl]2 as catalyst (entries 1 and 3), while the [2+2+2+1] reaction of 5a was performed at 50 °C and ambient pressure of CO in toluene or dichloroethane at 0.1 M concentration for 24 h using [Rh(COD)Cl]2 as catalyst (entries 6 and 7). The CO-SiCaT reaction of 5a gave 6a as exclusive product without any trace of 17a(H) (entries 1 and 3), and Rh(acac)(CO)2 in THF is a better catalyst system than [Rh(COD)Cl]2 in toluene (toluene is the best solvent for this catalyst). In the absence of PhMe2SiH otherwise under the same conditions, no reaction took place at all and 5a was recovered intact (entries 2 and 4). Thus, it is very clear that the CO-SiCaT reaction requires the hydrosilane. Also, it should be pointed out that the CO-SiCaT reaction proceeds smoothly at ambient temperature. In sharp contract, the [2+2+2+1] cycloaddition of 5a requires much higher concentration (0.1 M) and 50 °C to proceed. Also, the reaction gives a mixture of 6a (major) and 17a(H) (minor) (entries 6 and 7). A marked difference in the catalytic activity and selectivity was observed between [Rh(COD)Cl]2 and Rh(acac)(CO)2, i.e., the reaction catalyzed by Rh(acac)(CO)2 needed 90 °C to proceed and resulted in the exclusive formation of 17a(H) with no trace of 6a (entry 5).
Scheme 15 summarizes the difference and complementary nature of the CO-SiCaT and [2+2+2+1] cycloaddition reactions. It is confirmed that the mechanisms of these two reactions are fundamentally different. The CO-SiCaT reaction takes place only with 1-terminal free enediynes 5 though silicon-initiative cascade carbocyclization processes. On the other hand, the [2+2+2+1] cycloaddition reaction proceeds with 1-substituted and unsubstituted enediynes 7a and 5 through sequential metallacycle formations.
The fact that only fairly good product selectivity was observed for this enediyne substrate under the standard conditions for the [2+2+2+1] cycloaddition reaction suggests that the enediynes bearing a free acetylene moiety are not good substrates for this reaction since the reactions with 1-methyl enediyne 7a gave much better result (see Table 6, entry 1). In fact, the reactions of two other 1-terminus free enediynes 5b and 5d showed the same tendency (Scheme 17). The reaction of 5b under the standard [2+2+2+1] cycloaddition conditions proceeded slowly to give 6b in 51% isolated yield after 3 days at 50 °C. The CO-SiCaC reaction of 5b gave 6b in 91% isolated yield after 24 h at 22 °C (see Table 2, entry 2). In a similar manner, the [2+2+2+1] cycloaddition of 5d afforded 6d in 52% isolated yield after 24 h at 50 °C, while the CO-SiCaT reaction of 5d gave 6d in 84% isolated yield after 24 h at 22 °C (see Table 2, entry 4).
We also investigated the extension of the CO-SiCaT and [2+2+2+1] cycloaddition reactions ofenediynes to related substrates, i.e., allenediyne 19 and diynals 21a,b. The attempted [2+2+2+1] cycloaddition of allenediyne 19 catalyzed by [Rh(COD)Cl]2 under the standard conditions afforded the fused 5-6-6 tricyclic product 20 exclusively in 97% isolated yield (>98% yield by 1H NMR) through [2+2+2] cycloaddition (Scheme 17). To the best of our knowledge, this presents the first example of intramolecular [2+2+2] cycloaddition of allenediyne, and further studies on this novel reaction is actively underway in these laboratories.
The reaction of undeca-5,10-diyn-1-al (21a) catalyzed by Rh(acac)(CO)2 (1 mol%) under the CO-SiCaT conditions with PhMe2SiH (1 equiv.) at high dilution (0.015 M) in THF at 50°C for 24 h failed to give any CO-SiCaT or SiCaT product, but gave the corresponding bicyclization product 22 in 45% isolated yield. The formation of 22, as the sole product, was improved to 81% isolated yield at higher concentration (0.1 M) in toluene (Scheme 19). When the same reaction was run in toluene otherwise under the same conditions, a mixture of 22 (36%) and deoxygenated bicyclization product 23 (48%) was isolated (Scheme 18). The use of higher CO pressure (20 atm) only slowed down the reaction, forming 22, in toluene.
Since the diynal 21a did not seem to be suitable for the CO-SiCaT reaction, we examined the [2+2+2+1] cycloaddition conditions. The metal-catalyzed cycloaddition of an aldehyde to diynes has not been extensively studied. Thus, only a few successful examples have been reported for the intermolecular [2+2+2] cycloadditions of 1,7-diynes (no free terminal acetylene moieties) with aldehydes catalyzed by Ni(COD)2-PR3 (R = Bu or Cy). However, no intramolecular version has been reported to date.25 The reaction of dodeca-5,10-diyn-1-al (21b) was carried out with [Rh(COD)Cl]2 (5 mol%) at 50 °C and ambient pressure of CO in toluene (0.1 M) for 24 h. The reaction gave the corresponding [2+2+2+1] cycloaddition products 24b and 25b (65% combined yield) as well as an interesting acetylbis(cyclopentenyl) 26b (33%) in nearly quantitative overall yield (Scheme 20). On monitoring the reaction, it was found that fused 5-7-5 tricyclic lactone 24b was the kinetic product. Thus, only 24b and 26b were formed in the early stage of the reaction. The clean and complete isomerization of 24b to 25b was observed by leaving a mixture of 24b and 25b in CDCl3 in an NMR tube at ambient temperature for two weeks. The formation of acetylbis(cyclopentenyl) 26b can be readily explained by assuming the intermediacy of [2+2+2] cycloadduct A20, which underwent an electrocyclic ring opening of the central pyrane moiety to form the bicyclic dienone structure of 26b.
The preferential formations of 25b and 26b indicate that the two endo-double bond system is substantially more favorable than the two exo-double bond system. In fact, the tricyclization reaction of 21a, which does not have a methyl substituent at the terminal acetylene moiety, at 60 °C for 68 h gave fused 5-7-5 tricyclic lactone 25a (major) and 26a (minor), i.e., the kinetic product 24a was not isolated at all in this reaction (Scheme 21).
The overall results on the reactions of diynals 21 show that the intramolecular tricyclization involving a diyne and an aldehyde components is ready to occur by the catalysis of [Rh(COD)Cl]2, and the novel [2+2+2+1] cycloaddition of diynal with CO was achieved for the first time. However, the [2+2+2+1] cycloaddition and [2+2+2] cycloaddition processes are competing. Thus, further optimization of the [2+2+2+1] cycloaddition process is necessary for this reaction to be synthetically useful. Nevertheless, it is worth mentioning that unique fused 5-7-5 tricyclic lactones 25 can be obtained in fairly good isolated yields even at this discovery stage. Further studies on this novel higher order cycloaddition reaction are actively underway in these laboratories.
In summary, we reported here the full account of the novel silicon-initiated carbonylative carbotricyclization (CO-SiCaT) as well as higher order [2+2+2+1] cycloaddition reactions of enediynes with CO catalyzed by rhodium complexes. The most plausible mechanisms for these two reactions, which are fundamentally different from each other, were proposed and discussed. Another mechanism for the formation of fused 5-6-5 tricyclic products from 1-substituted enediynes under CO-SiCaT conditions was proposed. The scope and limitation of these unique processes was described. These newly discovered reactions allow us to construct multiple bonds all at once, converting linear starting materials to polycyclic compounds in one single step. Thus, these new processes have high synthetic potential to provide innovative routes to appropriately functionalized polycyclic intermediates for the syntheses of natural and unnatural products of medicinal and/or biological interests. We also reported the first examples of intramolecular tricyclization of diynals, including the formation of fused 5-7-5 tricyclic lactones through [2+2+2+1] cycloaddition with CO as one carbon unit. In addition, the first [2+2+2] cycloaddition of an allenediyne, yielding the corresponding fused 5-6-6 tricyclic product was reported. These novel polycyclization processes clearly warrant further investigations.
A typical procedure is described for the reaction of 4,4,9,9-tetra(carbethoxy)dodec-11-ene-1,6-diyne (5a). A reaction vessel equipped with a stirring bar and a CO inlet, was charged with Rh(acac)(CO)2 (0.6 mg, 1 mol%). After purging the reaction vessel with CO, THF (14.6 mL) was added to dissolve the catalyst. Dimethylphenylsilane (38 mL, 0.5 equiv.) was then added and the mixture was stirred at 22 °C. After 1 min, a solution of substrate 5a (112 mg, 0.25 mmol) in THF (2 mL) was added and the reaction mixture stirred under CO (1 atm) for 30 h. All volatiles were removed under reduced pressure. The crude products were purified by column chromatography on silica gel using hexanes:EtOAc (5:1) as eluent to give 2,2,8,8-tetra(carbethoxy)-5-oxo-1,3a,4,5,7,9-hexahydro-3H-cyclopenta[e]azulene (6a) as a colorless oil (109 mg, 92%): TLC (SiO2, hexanes:EtOAc = 5:1, Rf = 0.17); 1H NMR (300 MHz, CDCl3) δ1.24 (t, 6 H, J = 7.2 Hz), 1.25 (t, 6 H, J = 7.2 Hz), 1.96 (dd, 1 H, J = 11.7 Hz/J = 12.9 Hz), 2.51 (bd, 1 H, J = 12.9 Hz), 2.72 (m, 3 H), 3.01(m, 1 H), 3.13 (m, 4 H), 3.25 (dd, 1 H, J = 2.1 Hz/J = 17.4 Hz), 4.19 (q, 4 H, J = 7.2 Hz), 4.20 (q, 4 H, J = 7.2 Hz), 6.03 (s, 1 H); 13C NMR (75 MHz, CDCl3) δ13.90 (2C), 13.9 (2C), 37.7, 39.5, 40.5, 41.0, 43.6, 44.7, 57.5, 58.8, 61.7, 61.8, 61.8 (2C), 124.0, 129.4, 150.1, 154.1, 170.4, 170.6, 170.7 (2C), 197.5; IR (neat, cm−1) 2981 (m), 2938 (w), 1731 (s), 1646 (s), 1603 (w), 1465 (w), 1447 (m), 1366 (m), 1288 (s), 1252 (s), 1193 (s), 1159 (s), 1099 (m), 1070 (s), 1018 (m), 861 (w). HRMS (FAB) m/z calcd for C25H33O9 (M + 1)+ 477.2125, found 477.2124 (Δ= − 0.2 ppm).
A typical procedure is described for the reaction of enediyne 7a. All other reactions are run following the same procedure unless otherwise noted. Enediyne 7a (100 mg, 0.216 mmol) was introduced to a Schlenck tube, followed by Cl(CH2)2Cl (2.16 mL) under argon, and then CO was bubbled into the solution at room temperature. After 15 min, [Rh(COD)Cl]2 (5.3 mg, 0.0108 mmol, 5 mol%) was added under CO and the resulting mixture was stirred at room temperature for an additional 5 min. Then, the reaction mixture was heated to 50 °C with stirring and kept for 16 h under CO (ambient pressure, bubbled into the solution). All volatiles were removed from the reaction mixture under reduced pressure and the crude product purified by flash chromatography on silica gel (EtOAc:hexanes = 1:9 to 3:7) affording 8a as a light yellow oil (92.5% yield by 1H-NMR ; 93 mg, 88 % isolated yield) and 17a as a colorless oil (5.5% yield by 1H-NMR ; < 4 mg, <4 % isolated yield).
Colorless oil; TLC (SiO2, EtOAc:hexanes = 1:4, Rf = 0.30); 1H NMR (400 MHz, CDCl3) δ 1.25 (m, 12 H), 1.90 (s, 3 H), 1.96 (m, 1 H), 2.61–3.20 (m, 10 H), 4.22 (m, 8 H); 13C NMR (100 MHz, CDCl3) δ 14.0, 15.8, 37.4, 39.9, 40.3, 40.9, 42.4, 45.7, 56.9, 59.1, 61.7, 129.7, 130.8, 147.3, 149.5, 170.7, 170.8, 170.9, 171.0, 197.3; IR (neat, cm−1) 2910 (s), 1731 (s), 1644 (s), 1367 (m), 1289 (m), 1068 (m). HRMS (EI) m/z calcd for C26H34O9 (M+) 490.2203, found 490.2205 (Δ= 0.4 ppm).
Colorless oil; TLC (SiO2, EtOAc:hexanes = 1:4, Rf = 0.56); 1H NMR (400 MHz, CDCl3) δ 1.22 (m, 12 H), 1.70 (s, 3 H), 1.82 (m, 1 H), 1.84 (m, 1 H), 2.13 (m, 1 H), 2.60–3.07 (m, 8 H), 4.18 (m, 8 H); 13C NMR (100 MHz, CDCl3) δ 13.9, 14.0, 19.4, 35.9, 36.1, 36.7, 36.8, 39.7, 41.3, 59.4, 59.7, 61.4, 123.5, 128.2, 128.9, 131.1, 171.6, 171.7, 171.9. HRMS (EI) m/z calcd for C25H35O8 (M+H)+ 463.2332, found 463.2325 (Δ = −1.5 ppm).
This research was supported by grants from the National Institutes of Health (NIGMS) and the National Science Foundation. Generous support from support from Mitsubishi Chemical Corp. is gratefully acknowledged.
For other reactions and their products, see Supporting Information.