Diastereoselective additions of H-phosphinates to alkenyl ketones under phase-transfer conditions

Abstract

Additions of bulky H-phosphinates to β,β-disubstituted alkenyl ketones under the action of a phase transfer-catalyst led to a new class of carbon quaternary center-containing phosphinates with unprecedented control of the relative configuration of the adjacent carbon and phosphorous chiral centers with high diastereoselectivity in most cases.

Compounds bearing the phosphinic acid group play a pivotal role as pseudo-peptides in a variety of therapeutic applications,¹ including metalloenzyme substrate mimics as antitumor agents.² These phosphinic acids are often generated from a precursor phosphinate ester whose stereochemical configuration has been ignored since it is lost upon conversion to the acid.³ However, we and others⁴ have lately become interested in exploring the biological potential of these phosphinate esters themselves. We thus required a convenient method for their preparation with control of the relative and absolute configurations of phosphorous and nearby carbon centers. Asymmetric Michael additions with a variety of phosphorous nucleophiles such as symmetrical phosphine oxides⁵ and phosphonates⁶ have been studied. Additionally, there are several reports of Michael additions involving phosphinates under the action of microwaves,⁷ base,⁸ or phosphine catalysts,⁹ though these reactions are not diastereoselective. To our knowledge, the only diastereoselective additions achieved to date have been with chiral amino-phosphinates; these reactions enable the creation of a carbon-based stereo-center in a subsequent alkylation step (Fig. 1a).¹⁰

Fig. 1.

Phospha-Michael additions leading to phosphinates.

To realize a diastereoselective addition to β-substituted Michael acceptor, we reasoned that a phase transfer catalyst (PTC) would offer a tunable parameter in the transfer of chirality from the phosphorous center in A to an electrophilic alkenyl carbon during the attack (Fig. 1b) resulting in diastereoselective product formation. Montchamp describes phosphinates existing as prototropic tautomeric forms P(v) and P(iii) (Fig. 1b).¹¹ The latter form has potential as a nucleophile and is probably more acidic.¹² We considered that carbonate salts, rendered more basic by the action of a PTC, would lead to nucleophilic intermediate A capable of diastereoselective additions. In the present communication, we describe our initial efforts to realize this phase-transfer catalyzed phospha-Michael reaction with alkenyl ketones leading to diastereoselective phosphinate product formation; to our knowledge this approach has not been previously reported.¹³

We began by examining a variety of bases for their potential to bring about the addition of phosphinate 1a to ketone 2a under the catalytic action of a PTC. Towards this end, several organic and inorganic bases, additives, and solvents were varied to optimize the addition reaction (Table 1). Our initial attempts revolved around hydroxide bases, However, we quickly realized that these bases led to product 3a without added PTC. By contrast, weaker bases such as potassium carbonate were unable to bring about phosphinate product 3a even after extended times (entry 1). However, in the presence of tetrabutylammonium bromide (TBAB), complete conversion to product was observed after reacting overnight though in a modest 3 : 1 diastereomeric ratio (dr). The use of 18-crown-6 (18C6), known for its high affinity towards the potassium cation, resulted in the successful formation of 3a. The crown ether catalyst led to an improved 5 : 1 dr (entry 3). The crystal structure of the major diastereomer of 3a revealed a syn orientation of the phosphinate oxygen and the phenyl substituent on the adjacent carbon (Fig. 2). Control experiments using crown ether itself resulted in no product formation. Further optimization studies using the excess equivalents of base and additives resulted in no significant improvement in diastereomeric ratio (dr) except for improved conversions in shorter reaction times (entry 5). While MTBE led to similar outcomes as toluene, we chose to proceed with the latter for convenience since it was less likely to evaporate over longer reaction times. Other phase transfer agents such as dicyclohexyl-18-crown-6 and 2-aminomethyl-18-crown-6 also led to product 3a in a 5 : 1 dr (data not shown).

Table 1.

Optimization of reaction conditions^a

Entry	Base equiv.	Additive	Solvent	Conv.b (%)	drb
1	10	—	Toluene	NR	—
2c	2	TBAB	Toluene	100	3 : 1
3c	2	18C6	Toluene	100	5 : 1
4	2	18C6	Toluene	80	5 : 1
5	10	18C6	Toluene	100	5 : 1
6	10	18C6	Toluene/H2O	88	5 : 1
7	10	18C6	CH2Cl2	25	—
8	10	18C6	Mesitylene	<10	—
9	10	18C6	MTBE	100	5 : 1

^aGeneral procedure: 1a (1.0 equiv.), 2a (1.2 equiv.), base, and additive (10 mol%) were stirred in solvent [0.2 M] at rt for 15 h.

^bDiastereomeric ratio (dr) and percent conversion were estimated using ³¹P NMR.

^cAdditive (50 mol%).

Fig. 2.

Crystal structure of 3a.

Optimized conditions were then used to bring about Michael additions to a variety of alkenyl ketones to understand the generality of this reaction (Table 2). We were pleased to observe that the inclusion of a methyl group at the β-position of substrate 2 (R¹ = Ph, R² = Me) led to addition product 3c with nearly complete diastereoselectivity. Phosphinates bearing an adjacent asymmetric quaternary carbon have not been previously reported. Not unexpectedly, the use of a bulky phosphinate 1a (R = Ad) or 1b (R = ^tBu) was necessary for high diastereoselectivities leading to product 3c and 3d. The smaller 1c (R = Et) led to product 3e with a significantly poorer diastereoselectivity (3 : 1). Other β,β-disubstituted substrates also afforded addition products, such as 3f–3i with impressive selectivities. Additions to an indanone substrate, though low yielding, also led to a highly diastereoselective outcome (see 3h). The reaction leading to product 3i suggests that the inclusion of an aryl group at the β-position is not essential for good diastereoselectivity, but it does appear to improve yield. Additions of 1a to an alkenyl sulphone and alkenyl cyanide led to modest yields and diastereoselectivities. Finally, taking advantage of (−)-menthyl phosphinate 4 previously reported as a non-racemic nucleophile,¹⁴ adduct 5 was produced in 64% yield and in 8 : 1 diastereoselectivity. A follow-up objective of the present study is to develop an asymmetric variation using an appropriate chiral PTC.¹⁵ Nevertheless, our synthesis of phosphinate 5 suggests that enantioenriched phosphinate products can be prepared in good diastereoselectivities using stoichiometric non-racemic reagents.

Table 2.

Substrate scope^a

^aConditions: 1 (1.0 equiv.), 2 (1.2 equiv.), base (10.0 equiv.), additive (20 mol%), solvent (0.2 M) at rt for 72–96 h. All yields shown are isolated yields for the diastereomeric mixture.

Diastereomeric ratios were determined using ³¹P NMR.

We are currently investigating the phosphinate addition mechanism to produce more selective product outcomes with β-monosubstituted alkenyl ketone substrates. Here the adamantyl-containing nucleophile 1a led to only a slightly higher diastereoselectivity relative to the much smaller ethyl group (compare 3a to 3n). The reaction appears to be sensitive to the degree of steric bulk at the β-position of Michael acceptor 2, though it is possible for a substituent to be too large (compare 3o and 3p). This steric bulk also led to a substantial difference between TBAB and 18C6 in terms of the diastereoselectivity of product 3o (see Table 3). With β-monosubstituted alkenyl ketones, phenyl substitution at the β-position was crucial for a stereoselective outcome. Alkenyl ketones bearing only an alkyl substituent at the β-position all led to diastereoselectivities less than 2 : 1 (data not shown). The electron density of the β-aryl of Michael acceptor seems to play only a small role in selectivity (compare 3a with 3q and 3r). Perhaps most surprisingly, the change of the ketone substituent from phenyl (3a) to naphthyl (3s) to methyl (3t) led to no difference in stereoselectivity.

Table 3.

Examination of steric and electronic factors^a

^aConditions: 1 (1.0 equiv.), 2 (1.2 equiv.), base (10 equiv.), additive (10 mol%), solvent (0.2 M) at rt for 2–3 d. All yields shown are isolated yields for the diastereomeric mixture.

Major diastereomer shown with ratios determined using ³¹P NMR.

^bPTC (0.1 equiv.).

In terms of the mechanism for the present reaction, we examined the possibility of the phosphinate addition being under thermodynamic control. To this end, the major diastereomer of 3a (produced as a 5 : 1 diastereomeric ratio) was isolated in pure form and subjected to the same reaction conditions used to produce it. The reaction was monitored at regular time intervals over the course of three days using ³¹P NMR showing no change and no indication that the minor diastereomer had formed. These results suggest that the phosphinate addition reaction is under kinetic control. Turning to transition state arguments, we noted that simple phenyl alkenyl ketone favors the S-cis conformation¹⁶ and that β-substitution increases the prevalence of this conformation.¹⁷ Assuming a closed transition state model, attack of the S-cis conformation of an (E)-β-aryl alkenyl ketone would lead to TS1 or TS2 (Fig. 3). There would be a clear preference for TS1 since this would lead to a stabilizing π–π interaction. This transition state leads to a product in which the phosphinate oxygen and the aryl group on the adjacent carbon are in a syn-orientation with respect to each other. Indeed, this is the relative stereochemistry revealed in the crystal structure of 3a (Ar = Ph) (Fig. 2). Importantly, the attack of a substrate in the S-cis conformation does not involve the ketone substituent, which is consistent with our finding involving products 3s and 3t. The model also explains why the adamantyl group on the phosphinate nucleophile leads to improved stereoselectivity since its size would further destabilize TS2. However, if the phosphinate nucleophile encounters the less likely S-trans conformer, both faces of the nucleophile can attack with similar facility since there is a stabilizing π–π and destabilizing steric repulsion interaction in each case (compare TS3 and TS4). This preference for an S-cis alkenyl ketone may also explain the complete lack of selectivity observed in product 3j (Table 2). In this case, the cyclopentenone substrate leading to the product is locked in an S-trans conformation. Based on this model, it appears that the exceptionally high diastereoselectivity with β,β-disubstituted substrates is largely due to their propensity to adopt the S-cis conformation, which presents a more favorable face for nucleophilic addition (i.e., TS1).

Fig. 3.

Proposed mechanism.

As mentioned previously, phosphinic esters are relatively unexplored as bioactive compounds; this is especially true in phosphacyclic systems.¹⁸ Seeking to make some headway towards this deficiency, we sought to prepare a cyclic phosphinic ester. To this end, compound 3c was reduced to give alcohol 6 in 80% yield and modest diastereoselectivity (Scheme 1). Subsequent treatment of the major diastereomer of 6 with camphor sulfonic acid (CSA) led to cyclic phosphinic ester 7 (a 1,2-oxaphosphorinane) as a single diastereomer. This cyclization reaction appears to retain the configuration of the phosphorous center present in 6. The same synthesis route was repeated with menthol-derived compound 5 leading to non-racemic product 7 in a relatively low unoptimized yield (16%). We note that both 3c and non-racemic 5 led to product 7, in each case having the same relative configuration. Except for a patent disclosure,¹⁹ 1,2-oxaphosphorinanes are unknown, though the synthesis of related oxaphosphorines (bearing a carbon–carbon double bond in the ring) via ring closing metathesis has been disclosed.²⁰

Scheme 1.

Stereoselective synthesis of a novel heterocycle.

In conclusion, we have developed a mild phase transfer-catalyzed diastereoselective phospha-Michael reaction involving phosphinate nucleophiles. With β,β-disubstituted alkenyl ketones, these reactions afforded high diastereoselectivities, leading to a new class of carbon quaternary center-containing phosphinates with control of the relative configuration of both carbon and phosphorous chiral centers. To our knowledge, a diastereoselective reaction of this type leading to phosphinates is unprecedented. Studies with β-substituted alkenyl ketones also revealed synthetically useful diastereoselectivities. The reactivity trends in this system seem most consistent with a closed transition state in which the alkenyl ketone is attacked in the S-cis configuration. This new reaction made possible, for the first time, the production of a substituted 1,2-oxaphosphorinane in good diastereoselectivity. Efforts are currently underway to develop a phase transfer catalyzed asymmetric variation of the present diastereoselective reaction.