Reversibility Effects on the Stereoselectivity of Pt(II)-Mediated Cascade Poly-ene Cyclizations

Abstract

Cyclization of 1,5-dienes bearing nucleophilic traps with electrophilic trisphosphine pincer ligated Pt(II) complexes results in the formation of a polycyclic Pt-alkyl via an Pt(η²-alkene) intermediate. With electron rich triphosphine ligands an equilibrium between the Pt(η²-alkene) and Pt-alkyl was observed. The position of the equilibrium was sensitive to ligand basicity, conjugate acid strength, solvent polarity, and ring size. In cases where the ligand was electron poor and did not promote retro-cyclization, then kinetic products adhering to the Stork-Eschenmoser postulate were observed (E-alkenes give trans-ring junctions). When retrocyclization was rapid, then alternative thermodynamic products resulting from multi-step rearrangements were observed (cis [6,5]-bicycles). Under both kinetic and thermodynamic conditions remote methyl substituents led to highly diastereoselective reactions. In the case of trienol substrates, long-range asymmetric induction from a C-ring substituent was considerably attenuated and only modest diastereoselectivity was observed (~2:1). The data suggests that for a tricyclization, the long-range stereocontrol results from diastereo-selecting interactions that develop during the organization of the nascent rings. In contrast, the bicyclization diastereoselectivities result from reversible cascade cyclization.

Introduction

The stereoselective cascade cyclization of polyprenoids into multicyclic steroid-like structures stands as one of the crowning achievements in organic chemistry.ⁱ The state of the art synthetic methods reflect the sum of enormous efforts aimed at understanding the rules for controlling the efficiency and selectivity of the non-enzymatic cation-olefin cascade responsible for ring construction. The Stork-Eschenmoser postulate (SEP) was one of the key early findings that enabled the ring junction stereochemistry to be predicted from the starting olefin geometry; E-alkenes give trans ring junctions and Z-alkenes give cis ring junctions.^ii,iii The postulate is a manifestation of a favorable anti addition of electrophile and nucleophile across the original alkene (e.g. Scheme 1). Ultimately, however, the stereochemical outcome of such cascade cyclizations is explainable by evaluating the role that neighboring group participation plays on the kinetics of the cation-olefin reaction;ⁱⁱⁱ the key being that coordination of a stabilizing group (Lewis base, alkene, etc) to a carbenium ion simultaneously advances the cyclization reaction coordinate.^iv Interaction of an electron rich neighboring group on a developing carbenium ion (i.e. stabilizing it) lowers the punitive enthalpic cost of localizing a full positive charge on the center in question at the cost of heightened entropic organization, while simultaneously preserving the stereochemical information (i.e. E vs Z) stored in the alkene.

Scheme 1.

The ability of the neighboring group to engage the developing carbenium ion also effects the efficiency and stereospecificity of cyclization reactions.^v When a good participant is available, the reactions tend to be more favorable and selective since they avoid free carbenium ions and stepwise pathways that can erode stereospecificity.

Another factor influencing the step-wise versus concerted cyclization manifolds is the stability of the initiating carbenium ion or reacting group (oxo-carbenium ion, allyl cation, etc). When the initiator is highly reactive, it does not need the stabilizing influence of a neighboring group to react with the internal alkene, which leads to free carbenium ions and less selective, possibly non-stereospecific step-wise reactions.^vi On the other hand, when the initiating groups are less reactive, then the beneficial effect of a stabilizing group (alkene or terminating protic group) on ΔG^‡ can be significant and this favors concerted, stereospecific reactions.

The debate over step-wise versus concerted reactions is also germane to the discussion of long-range stereocontrol. Numerous examples demonstrate that stereocenters remote from the point of initiation can influence the cascade selectivity (e.g. eq 1).^vii These results were most easily rationalized by invoking the influence of the substituent on the ordered conformation of the polyene in a concerted cyclization (i.e. pseudo-equatorial vs. pseudo-axial substituent in a nascent polycycle). Alternatively, a reversible cascade cyclization could account for the selectivity as the retrocyclization provides a means for error correction and access to thermodynamic products. Bartlett in his authoritative 1982 review concluded, primarily using free energy arguments, that long range stereocontrol in sterol cyclizations could not come from a reversible cyclization scenario and that the only logical means for achieving these remarkable effects was through the preorganization that accompanies a concerted cyclization.^1c,viii

(1)

Carbophilic transition metal complexes are also known to activate terminal alkenes for cascading cation-olefin reactions.^ix In most of these cases the Stork-Eschenmoser postulate holds and polyprenoids provide polycyclic structures through chair-like transition structures. Recent advances include oxidative polycyclizations using PdCl₂ catalysts in combination with benzoquinone,^x,xi (pyridyl-bisphosphine)Pd^II dications,^xii (trisphosphine)Pt^II dications,^xii and Hg(II) mediated polyprenoid cyclizations.^xiii In addition to transition metals, several important developments have occurred in the area of chiral electrophilic halonium ions^xiv and chiral Brønsted-Lewis acid catalysts^xv for promoting enantioselective cation-olefin polycyclizations. Unlike H⁺, Hg(II), and X⁺ (X= I, Br, Cl), the carbophilic Pd(II) and Pt(II) Lewis acids have a strong preference for coordinating and activating the least substituted alkene, which provides a strong chemodirecting effect in initiating the cation-olefin cyclization of polyene substrates.^xvi

Recent efforts in our laboratory have shown that pincer ligated Pd(II)- and Pt(II)-dications serve as excellent initiators of cation-olefin multicyclizations when the terminus is monosubstituted.^xvii The pincer ligands efficiently block migratory deinsertion pathways resulting in stable polycyclic organometallic products (e.g. eq 2). In the course of evaluating the effect of ligand electronics on the cyclization reaction, it was discovered that ligand basicity not only affected the kinetics of cyclization, but also the thermodynamics. With electron rich ligands, an equilibrium between cyclized and acyclic (η²-alkene) structures was discovered and the position of this equilibrium was responsive to ligand electronics.

(2)

Given the potential role of reversibility on the stereoselectivity of biomimetic cation-olefin cascade reactions, a study was initiated to establish its role on the diastereoselectivity of Pt-mediated cyclization reactions. We report herein, the effect that solvent, ligand basicity, ring strain, and ring size have on the position of the cyclized/acyclic equilibrium. These data were then utilized to establish conditions wherein retrocyclization was operative and could be utilized to measure the effect of remote substituents on the kinetic and thermodynamic cyclization diastereoselectivity.

Results

Activation Conditions

The Pt(II)-dications for this investigation were generated by one of two means. For complexes bearing more electron donating tridentate ligands (EtPPPEt, tBuPPPEt and PPPEt, see Chart 1) the dication was preferably generated via protonolysis of the Pt-Me precursor using [Ph₂NH₂][BF₄].^xix A typical activation procedure involved the addition of 10 equivalents of [Ph₂NH₂][BF₄] to a solution of [(RPPPR′)Pt-Me][BF₄] and dienyl substrate in the appropriate solvent; at the point of initiation, the reaction thus contained 9 equiv of [Ph₂NH₂][BF₄] and one equiv of Ph₂NH. For complexes bearing the more electron poor PPP and EtPPP ligands (Chart 1), the preferred mode of activation was with AgBF₄ and the diiodide [(RPPPR′)Pt-I][I] precursor, followed by the addition of one equiv of base (Ph₂NH or Ph₂NMe). Activation via halide abstraction was possible with the more electron rich Pt complexes, however, this approach required the addition of two equivalents of acetone to act as a weak labile ligand.^xviii

Chart 1.

Cyclization Reactions

Previous work on the conversion of 1,6-dienes into bicyclopropanes by (PPP)Pt(II) dications indirectly revealed that Pt-mediated C-C bond forming carbenium ion generation was both rapid and reversible.^xvii This observation also suggested a similar possibility in cascade cyclization processes. The first evidence for this behavior surfaced while monitoring the cyclization of 10 equiv of a 2:1 E:Z mixture of 1,5-dienyl sulfonamide (E:Z)-1 by (EtPPPEt)Pt²⁺ in CH₂Cl₂ (see Chart 1). Instead of the expected smooth conversion to a Pt-alkyl, an equilibrium between the starting η²-alkene adduct and expected product was observed (eq 3).^xix The identity of these complexes was readily apparent from the characteristic J_Pt-P of their central phosphorus^xx (Pt(η²-alkene), A, J_Pt-P = 2762 Hz; Pt-alkyl, B, J_Pt-P = 1281 Hz).^xxi The presence of uncyclized η²-alkene adduct was unexpected because all previous cyclizations of dienyl substrates with intramolecular traps had been rapid and proceeded to completion. After one hour, an A:B ratio of ~1:2 was observed by ³¹P NMR,^xxii and although decomposition occurred over extended times, an equilibrium constant could be measured at early times (~60). ^xxiii

(3)

Solvent and Ligand Effects on K_eq

From this starting point, an investigation into the factors controlling the equilibrium position was undertaken.^xxiv With [(EtPPPEt)Pt][BF₄]₂ and (E:Z)-1 as the probe, K_eq decreases with increasing polarity, presumably by the favorable solvation of the dicationic Pt(η²-alkene) complex over the monocationic Pt-alkyl (Table 1), though a solvent effect on the acid strength may also contribute. The poor solubility of [Ph₂NH₂][BF₄] limited the accessible solvents. Nitromethane (K_eq = 0.68, Figure 1) was selected for further analysis since Pt-olefin complex A was favored (Figure 1).

Table 1.

Solvent effects on the cyclization of (E:Z)-1 with (EtPPPEt)Pt^{2+ a}

Entry	Solvent	Keq (A:B)b	ΔG (kcal/mol)
1	CH2Cl2c	60 (1:2)	−2
2	ClCH2CH2Clc	110 (1:3)	−2.8
3	EtNO2	3.2 (4:1)	−0.69
4	MeNO2	0.68 (14:1)	0.23

^aConditions: [Pt] = 0.027 M, [Ph₂NH₂][BF₄] = 0.27 M, [1] = 0.27 M, 25(1)° C (see Supplementary Material).

^bRelative concentrations determined by ³¹P NMR. Average of three measurements.

^c[Pt] = 0.012 M.

Figure 1.

³¹P NMR of the cyclization of 1 with (EtPPPEt)Pt²⁺ in MeNO₂.

Since previous reports with the more electron poor PPP ligands had not indicated the potential for either reversibility, or stepwise cyclization of polyenes, the effect of ligand basicity on the cyclization equilibrium was explored.^xxv Table 2 shows the results of modifying the donor properties of the phosphine substituents and the consequent metal electrophilicity.^xxvi Not surprisingly, more electrophilic complexes shifted the cyclization equilibrium towards the monocationic Pt-alkyl species, a result of increased activation of the η²-alkene form and the favorable pairing of the less basic ligand with the more donating alkyl. While no Pt(η²-alkene) species (<5%) was detected for the most electrophilic case ((PPP)Pt²⁺), the detectable limit of reversibility was observed using EtPPP as the tridentate ligand (K_eq = 1100; Figure S3).

Table 2.

Ligand effects on the cyclization of (E:Z)-1 with (RPPPR′)Pt²⁺ (see Chart 1)^a

Entry	Ligand	Keq (A:B)b	ΔG (kcal/mol)
1	EtPPPEt	0.68 (14:1)	0.23
2	tBuPPPEt	7.2 (2:1)	−1.2
3	PPPEt	14 (1.2:1)	−1.6
4	EtPPP	1100 (1:10)	−4.1
5	PPP	>4200 (1:>20)	< −4.9

^aReaction conditions: [Pt] = 0.027 M, [Ph₂NH₂][BF₄] = 0.27 M, [1] = 0.27 M, 25(1)°C (see Supplementary Material).

^bRelative concentrations determined by ³¹P NMR. Average of three measurements.

Based on the cyclization reaction, it was anticipated that stronger bases would drive the equilibrium towards the cyclized product and that stronger conjugate acids would promote retrocyclization to the Pt(η²-alkene) complex. Experiments to confirm this were limited to bulky diaryl ammonium acids since the conjugate bases of smaller, more electron rich ammonium acids ligate the metal and poison the complex. With (EtPPPEt)Pt²⁺, the K_eq shifts from 0.68 (A:B = 14:1) using [Ph₂NH₂][BF₄] (Table 1), to 11 (A:B = 1.4:1) using [Ph₂NMeH][BF₄], consistent with the notion that a stronger base such as Ph₂NMe helps promote the cyclization.^xxvii Adding base or acid to a preequilibrated system also shifted the ratios of A and B in the expected fashion. Adding 5 equivalents of Ph₂NH to a mixture where A predominated (entry 4, Table 1) decreased the A:B ratio from 14:1 to almost 2:1. Likewise, 5 equivalents of [Ph₂NH₂][BF₄] shifted the equilibrium to the left; 1:10 to 1:4 (A:B).^xxviii

It was anticipated that ring strain in the cyclized organic moiety would also effect the equilibrium in a predictable manner. To investigate this factor, the dienyl sulfonamide E-2, which should produce a less strained 6,6 ring structure, was subjected to reversible conditions using (EtPPPEt)Pt²⁺. The less strained aza-decalin structure did indeed shift the equilibrium towards the cyclized complex 3 (K_eq = 1.0; eq 4),^xxix however the magnitude of the shift was less than expected based on generic ring strain measures (6.3 kcal/mol for cis-hydrindan versus -1.9 kcal/mol for trans-decalin).^xxx

(4)

Stereocontrol in Reversible Polycyclizations

As discussed above, the PPP complex smoothly reacts with 1 equivalent of (E:Z)-1 to generate alkyl complex 4 with no trace of η²-alkene (eq 5). This complex was isolated as its BF₄⁻ salt and characterized by X-ray crystallography. The ORTEP in Figure 2 clearly shows a cis ring juncture for the bicyclic fragment,^xxxi which contrasts the trans bicyclic products typically obtained from catalytic and stoichiometric cyclizations of E-alkenes.^xxxii To rule out the possibility of a selective cyclization of the minor Z isomer present in the 10 equivalents of (E:Z)-1 (which directly leads to the observed product via the SEP), E-1 was synthesized and cyclized (Scheme 2). By ³¹P NMR the Pt-alkyl obtained from this reaction was identical to 4, indicating that both stereoisomers of 1 converge to a single product. This result clearly conflicts with the notion of a concerted cyclization that follows the Stork-Eschenmoser postulate, both because only one product is obtained and because it requires an intermediate(s) capable of converging the E and Z isomers of 1 into a single stereoisomer of 4 (vide infra).

Figure 2.

ORTEP representation of 4. Hydrogen atoms and BF₄⁻ counter ion removed for clarity. Selected bond lengths (Å): Pt-P₁ = 2.269(4), Pt-P₂ = 2.247(3), Pt-P₃ = 2.308(4), Pt-C₄₅ = 2.150(1). Selected bond angles (deg): P₁-Pt-P₂ = 84.51(16), P₂-Pt=P₃ = 84.63(16), P₁-Pt-C₄₅ = 92.2(4), P₃-Pt-C₄₅ = 100.6(4), C₃-P₂-C₂ = 117.0(6).

Scheme 2.

(5)

The primary alcohol version of E-1, E-5, was also examined as a substrate using the (PPP)Pt²⁺ complex obtained by HNTf₂ activation of [(PPP)Pt-Me][BF₄]. In the presence of 1 equivalent of Ph₂NMe, compound E-5 cyclized smoothly to a single alkyl product. Cleavage of the bicycle using NaBH₄ generated the cis-fused octahydrobenzofuran 6, a product corresponding to a Pt-alkyl that again cannot be the primary cyclization product of E-5 since the SEP predicts a trans ring junction (eq 6). Similar to the cyclization of E-1, the isolated product is the thermodynamically more stable cis-[6,5] ring junction.

(6)

To investigate how existing stereocenters, especially those on remote positions of the polyene would influence the diastereoselectivity of the cyclization reactions, several readily available dienyl and trienyl alcohols were synthesized.^xxxiii The 1,5-dienyl alcohol E-7, bearing a methyl group β to the trapping –OH, was synthesized as shown in Scheme 3. It underwent rapid cyclization (<20 min) with (PPP)Pt²⁺ (Scheme 4) to generate one cationic (PPP)Pt-alkyl product (8) by ³¹P NMR (89.6 ppm, J_Pt-P = 1316 Hz), with no evidence for the intermediacy of other compounds.^xxxiv As expected from the previous results with (PPP)Pt²⁺ and 1, no Pt(η ²-alkene) was observed in the ³¹P NMR. In contrast to the above cases, however, crystallographic (X-ray) analysis revealed that 8 had a trans ring juncture and a pseudo equatorially disposed methyl group on the furan ring (Figure 3). The product thus appears to be that predicted from a concerted, chair-like transition state that positions the methyl group in a pseudoequatorial position (AA, Scheme 4). The observed 3,5 trans relationship^xxxv indicates efficient 1,6-stereoinduction, and since one would expect no diastereofacial preference in the alkene coordination step, it also indicates that the BB diastereomer either does not cyclize or it rapidly reverts back to starting material upon doing so.^xxxvi

Scheme 3.

Scheme 4.

Figure 3.

ORTEP representation of 8. Hydrogen atoms and BF₄⁻ counter ion omitted for clarity. Selected bond lengths (Å): Pt-P₁ = 2.262(3), Pt-P₂ = 2.290(3), Pt-P₃ = 2.290(3), Pt-C₁ = 2.158 (11). Selected bond angles (deg): P₁-Pt-P₂ = 84.80(10), P₂-Pt-P₃ = 83.15(10), P₁-Pt-C₁ = 90.6(4), P₃-Pt-C₁= 101.7(4), C₂₇-P₂-C₃₃ = 112.2(5).

An illuminating observation was made when E-7 was reacted with [(EtPPP)Pt][BF₄]₂ under the standard AgBF₄ activation conditions. Instead of a smooth conversion to a single alkyl product (or an equilibrium mixture of η²-alkene and alkyl), the ³¹P NMR spectrum indicated, after 3 hours at RT, a mixture of Pt(η²-alkene) at 100.4 ppm (J_Pt-P = 2904 Hz) and two Pt-alkyl species at 90.9 (J_Pt-P = 1241 Hz) and 90.2 ppm (J_Pt-P = 1239 Hz) in a 1:1 ratio. Over time, this mixture converged to a single Pt-alkyl (9) at 90.9 ppm (Figure 4) with no trace of the η²-alkene.

Figure 4.

³¹P NMR stack plot of the cyclization of E-7 with (EtPPP)Pt²⁺.

Upon completion of this reaction, 9 was isolated and crystallographically characterized. As shown in Figure 5, the bicyclic fragment has the cis ring junction, and is thus similar to 4 and 6. As in the case of 4, 9 has Pt in an equatorial position on the cyclohexyl A-ring and the heteroatom in a 1,4-trans relationship (see Figure 2 for comparison).^xxxvii Additionally similar to 4 and 6, the stereochemical outcome cannot be explained by a simple rearrangement (vide infra). When this reaction was stopped after three hours and treated with NaBH₄, it was determined (by GC) that the early forming product was the same trans isomer obtained by cleavage of the cycloalkyl in 8 (see Supporting Information).

Figure 5.

ORTEP representation of 9 with hydrogen atoms and BF₄⁻ counter ion omitted for clarity. Selected bond lengths (Å): Pt-P₁ = 2.3032(8), Pt-P₂ = 2.3016(8), Pt-P₃ = 2.2722(8), Pt-C₁ = 2.134(3). Selected bond angles (deg): 101.07(9), P₃-Pt-C₁ = 90.04(9), C₂₆-P₂-C₂₉ = 110.85(18).

To ascertain if the stereoselective reactivity would extend to a tricyclization, the trienol E,E-10, bearing a –Me substituent β to the trapping –OH on the nascent C-ring, was examined (Scheme 5). The trienol could be efficiently obtained by a sequence of dihydrofuran metallation/alkylation followed by Ni-catalyzed ring opening.^xxxviii A priori, one expects a larger driving force for the complete cyclization of a tricycle as compared to E-7. The ΔH_cyclization for E,E-10 is relatively straightforward (~20 kcal/mol more negative than E-7), and although ΔS_cyclization will certainly be more negative, it is difficult to estimate by how much.^id In any case, one can reasonably predict a more negative free energy for a tricyclization than a bicyclization, which would then tend to favor alkyl products over the Pt(η²-alkene) provided that the reaction is kinetically feasible.

Scheme 5.

Scheme 6 shows the reaction of E,E-10 with the (EtPPPEt)Pt²⁺, (EtPPP)Pt²⁺, and (PPP)Pt²⁺ cyclization initiators. ³¹P NMR analysis of the combination of (EtPPPEt)Pt²⁺ and E,E-10 under standard reaction conditions showed that although the Pt(η²-alkene) complex forms normally, it does not cyclize. Ten additional equivalents of Ph₂NMe pushes the reaction to ~90% Pt-alkyl and 10% Pt(η²-alkene), however, additional amine did not complete the reaction. Adding 10 equivalents of [Ph₂NHMe][BF₄] to this solution did not change the ratio of Pt-alkyl to Pt(η²-alkene), suggesting a poorly behaved, but static mixture (c.f. E-7).^xxxix In contrast, the cyclization of E,E-10 occurred smoothly when the more electron withdrawing EtPPP and PPP ligands were used. Since the ³¹P NMR for the (PPP)Pt-alkyls were broad, the cyclization diastereoselectivity was determined by first cleaving the tricycle with NaBH₄ and then analyzing by GC. This protocol yielded a 2:1 ratio of diastereomers that were inseparable by column chromatography, but whose structures could be ascertained by careful NMR analysis. These data are most consistent with the major diastereomer having the trans-anti-trans core with the C-ring methyl substituent adopting a pseudo-equatorial orientation, while the minor was most consistent with the trans-anti-trans tricycle having 3 axial methyl groups (Scheme 7).^xl

Scheme 6.

Scheme 7.

Discussion

Reversible Polycyclizations Using a 1,5-Dienyl Sulfonamide

The reversible cyclization of 1 proved to be a useful tool for establishing the effect of reaction conditions and metal complex electronics on the thermodynamics of cation-olefin cyclization. Taken together, the data paint an internally consistent picture wherein acid/base strengths, ring strain, ligand basicity, Pt-electrophilicity and solvent effects all have a predictable effect on the driving force for cascade cyclization. Not surprisingly, cyclization was favored by less polar solvents, more electrophilic metals, stronger bases, and a less strained bicyclic product. The data quantitatively describe the magnitude of these effects while guiding the choice of conditions to establish a reversible cyclization reaction for examining the effect of existing stereogenic centers on cation-olefin polycyclization diastereoselectivities.

Since (E:Z)-1 is a 2:1 mixture of E and Z isomers, generating two diastereomeric Pt-alkyls, one with a trans ring juncture and one with a cis ring juncture, should be possible (Scheme 8). Only one Pt-alkyl was observed by ³¹P NMR, suggesting that either Z-1 cyclized directly to 4 (Scheme 8) or that E-1 first cyclizes to a trans product and then subsequently isomerizes to the observed product, or both. The smooth conversion of E-1 to 4 (Scheme 2) suggests that the former case is not dominant. Since Z-1 does not build up in the ¹H NMR of the E-1 reaction, this eliminates the possibility of a reversible precyclization E/Z isomerization. It is important to note that isomerization of the trans product in Scheme 8 to 4 is not as simple as acid-promoted ionization of the C-N bond and recoordination to the opposite face, as this scenario provides an unobserved diastereomer where the [Pt] and N have a 1,4-cisoid arrangement instead of 1,4-transoid.

Scheme 8.

Mechanistically revealing was the set of experiments carried out on the methylated compound E-7. Under conditions that minimized retro-cyclization (electron poor PPP ligand), a single product with a trans ring junction was observed. We favor the hypothesis wherein this is the “kinetic” product, and that it proceeds through the most favorable chair-pseudo-chair transition state I, which places the B-ring methyl group in a pseudo equatorial orientation (8, Scheme 9). We qualify the term “kinetic” in this case because under irreversible conditions one should observe some of the less favored axial methyl product (from BB, Scheme 4) since diastereofacial control on η²-alkene formation should be low. If this product does form, it reverts to starting material and eventually traps at 8; nevertheless, we consider this chair-chair product the kinetic product and will refer to it as such (Scheme 9). We suspect that this chair-chair arrangement leads to the kinetic product in each of the cases studied herein, though as outlined below it may convert to a more stable isomer under certain conditions.

Scheme 9.

When the triphosphine is more electron rich than PPP, a scenario conducive to retrocyclization, the kinetic product converts to the more stable cis product 9. Since 9 does not result from a simple ionization of the C-O bond and recoordination to the opposite face of the carbenium ion (leading to III), a more complex sequence of steps is needed to invert the C3 stereocenter. One potential mechanism is outlined in Scheme 9, the keys steps being a retrocyclization of 8 and re-cyclization via the boat transition state IV to V, which then ionizes and recoordinates to form 9.^xli By invoking a boat transition state followed by C-O fragmentation and reformation it is possible to access a cis ring junction from an E-alkene, and thus avoid the constraints of the SEP. The potential for boat-like transition states was previously hinted at in the cyclization of a diene-phenol by (PPP)Pt²⁺ (eq 2). The diastereoselectivity in this reaction was 96:4, and >99:1 after cleavage of the bicycle with NaBH₄, a result that was interpreted as being due to competing chair-chair and boat-chair transition structures.^{xii,xlii,xliii} The importance of a B-ring boat transition state is well established in sterol biosynthesis.ⁱ In the present case IV is higher in energy than I and is normally non-competitive when retrocyclization is slow.

A similar sequence of steps was likely responsible for the formation of 4 and the Pt-alkyl that resulted from the cyclization of E-5. Less obvious was why E-5 converted to the cis product using (PPP)Pt²⁺ while E-7 did not. It is possible that the single methyl group on the THF ring reduced the rate of ring opening (an attenuated Thorpe-Ingold effect), with a concomitant lowering of the Pt(η²-alkene) concentration such that the kinetic product is temporally stabilized. This question will require more experimentation.

While a good case can be made for reversibility affecting the diastereoselectivity of bicyclization reactions, the same was not true for the tricyclization of E,E-10. Interestingly, under standard reaction conditions, the more electron rich ligand EtPPPEt was incapable of initiating the cyclization, though it could with 10 equivalents of the stronger base Ph₂NMe. With more electron poor ligands, however, a smooth conversion to alkyl products ensued and the diastereoselectivity could be ascertained by cleavage and GC analysis (2:1 dr). Careful NMR analysis of the diastereomer mixture pointed to products of the chair-anti-chair conformation, with the mixture occurring at the beta carbon of the THF ring (pseudo-axial and -equatorial). Although modest, this 2:1 selectivity represents 1,10-stereoinduction,^viii and since the reaction is apparently irreversible,^xliv it requires a different rate of cyclizing the two diastereomeric acyclic activated complexes (CC and DD), consistent with Bartlett’s assertion that long-range stereocontrol in sterol cyclizations comes from developing intra-ring interactions in the organizing assembly.^id Additionally consistent with a concerted cyclization was the formation of trans-anti-trans products, as reversibility in the C-O bond forming step would access the more stable [6.6.5]-trans-anti-cis products.^xlv

The marked difference in reactivity between the phosphine ligands in bi- vs. tricyclization was intriguing. In the former case, the ligands can each catalyze the cyclization, and the observed effect of ligand basicity is on the position of the equilibrium. In the tricyclization case, however, the electron rich EtPPPEt cannot initiate cyclization (standard conditions), while the less basic EtPPP and PPP ligands do so smoothly. This difference can be understood by noting that the bicyclization is not a true cation-olefin reaction since no developing carbenium ion reacts with another alkene, though the activated terminal alkene certainly behaves like one. The developing positive charge at C₄ is stabilized by the hydroxyl center, either with or without the aid of base (EE).^xxvii Since an alcohol will be a good neighboring group for stabilizing such a species, the reaction coordinate may proceed forward without needing to build up a large positive charge. On the other hand, B-ring formation in the tricyclization requires a genuine cation-olefin reaction, and since an alkene is a poorer ligand for the developing carbenium ion (FF), the degree of stabilization imparted by the neighboring group will be significantly lessened.^xlvi Moving the reaction coordinate forward in a tricyclization thus requires the buildup of more positive charge on C₄ to engage the alkene, which concomitantly requires a more electrophilic complex.

Systematically modifying the electrophilicity of the initiating Pt(η²-alkene) complex in the cation-olefin cyclization provided the means for examining how the cyclization thermodynamics responded. Not surprisingly, the more electrophilic the initiating group the more favorable the cyclization. When the cyclization was reversible high diastereoselectivities were observed, suggesting that this mechanism is indeed effective at error correcting and promoting long-range transfer of stereochemical information.^xlvii Moreover, when the cyclizations are reversible, alternative reaction pathways are explored that may access more stable products not otherwise accessible under the kinetic control of a typical cascade cyclization that is constrained by the SEP. The ability of pincer-ligated platinum(II) complexes to achieve this enables the formation of cis-ring fused bicyclic complexes that are not predicted by the SEP. Tricyclizations, however, were too favorable to find conditions where the η²-alkene form was stable enough to enable retrocyclization and electrophilic enough to initiate the forward cascade cyclization.

Supplementary Material

1File001. Supporting Information Available.

Tables and figures for equilibrium studies including sample ³¹P NMR spectra, full experimental procedures, and X-ray structural data for 4, 8 and 9. This material is free of charge via the Internet at http://pubs.acs.org.