A Bifunctional Cyclopropeneimine with a Tunable Hydrogen-Bond Donating Group and Its Application in the Enantioselective Synthesis of α,β-Diamino Phosphonates

Abstract

Lambert and co-workers have developed several chiral bases using a cyclopropeneimine as the basic moiety. Typically, these catalysts have a pendant hydroxyl group which acts as a hydrogen-bond donor and activates the electrophile. In catalysts with a hydrogen-bond donor, prior work from the Sigman group has shown that the acidity of the donor plays an important role in imparting selectivity. However, in the case of cyclopropeneimines reported by Lambert and co-workers, the acidity of the pendant arm is not easily tunable. To address this, we have developed a cyclopropeneimine catalyst with cyclohexanediamine as the chiral scaffold. One of the amino groups on the cyclohexanediamine is part of the cyclopropeneimine moiety, while the other is functionalized with an acyl group. The acidity of the pendant NH group was varied by changing the acyl group. We probed the utility of this catalyst in an enantioselective synthesis of α,β-diamino phosphonates using N-carbamoyl imines and fluorenone-imine phosphonates as the substrates. By varying the acyl group on the catalyst, we were able to optimize the enantioselectivity of the reaction. Using this approach, an enantiomer ratio of 92.5:7.5 was obtained with a 1-naphthyl derived imine. In general, imines with an electron-deficient aromatic group gave higher selectivities. Additionally, the optimized catalyst has good long-term stability in solution.

Introduction

The deprotonation of weakly acidic CH bonds with strong anionic bases is a well-established tool for generating carbon-centered nucleophiles. Traditionally, this has been accomplished using stoichiometric amounts of strong bases like alkyl lithiums and lithium diisopropyl amide. On the other hand, more acidic CH bonds like those present in malonate esters and nitroalkanes can be readily deprotonated with much weaker neutral bases likes amines. Indeed, reversible deprotonation of such CH bonds by chiral amine-based catalysts has been used in a number of enantioselective reactions. − In an effort to improve the range of CH bonds that can be deprotonated, several groups have synthesized neutral organic bases that are stronger than trialkyl amines. Schwesinger and co-workers have reported on the synthesis of phosphazenes, which are significantly stronger than amines. Another structural motif that has been actively pursued in this context is guanidine. Several groups have reported on the synthesis of novel guanidines and utilized them in enantioselective catalysis. − Recently, Lambert and co-workers have introduced chiral cyclopropeneimines as catalysts for enantioselective reactions. In their initial report, catalyst 4 (Figure ) was used to deprotonate a glycine imine ester. Michael addition to an acrylate ester gave the product in good yields and ee (Figure a). The hydroxy group on the catalyst was shown to be critical for obtaining high enantioselectivities. A key limitation of the initial reports was the very short half-life of the catalyst. , This was attributed to an intramolecular deprotonation of the hydroxyl group followed by a nucleophilic attack of the alkoxide on the iminium carbon, which led to ring-opening of the cyclopropene. Subsequently, a structure activity study showed that a catalyst derived from amino indanol had a half-life greater than five years in solid state (Figure a). However, in toluene solution the catalyst had a half-life of 36 h. Since these initial reports, other catalysts have been reported based on the cyclopropeneimine core. Apart from this, cyclopropenium cations have found extensive applications in organocatalysis. − In spite of this, to the best of our knowledge, there have not been any reports where the acidity of the hydrogen-bond donating group has been systematically varied. This has been shown to have a significant impact on the enantioselectivity of a reaction. With this in mind, we decided to synthesize the catalyst shown in Figure b and evaluate the role of hydrogen-bond in selectivity. In our catalyst design, the cyclopropeneimine scaffold is appended with a monoacyl cyclohexane diamine. The acidity of the ‘-NH’ group can be easily tuned by changing the acyl group. This provides an easy handle to optimize reactivity and selectivity. An added advantage is that intramolecular deprotonation of the NH group would result in the formation of an amide anion in which the negative charge is delocalized (Figure ). Nucleophilic attack on the iminium carbon through the nitrogen would result in the formation of a strained trans-5,6-fused system, whereas attack through the oxygen would result in the formation of a seven-membered ring. Kinetically, both of these reactions are expected to be slow. Taken together, these factors should significantly reduce the possibility of catalyst decomposition.

1.

(a) Initial report on cyclopropeneimine catalysts. (b) Current work on a tunable cyclopropeneimine and its application in the synthesis of scalemic α,β-diamino phosphonate.

2.

(a) Degradation pathway of the initially disclosed cyclopropeneimine. Adapted with permission from [Bandar, J. S.; Lambert, T. H. Enantioselective Bro̷nsted Base Catalysis with Chiral Cyclopropenimines. J. Am. Chem. Soc. 2012, 134, 5552–5555.]. Copyright [2012] [American Chemical Society] (b) Expected rationale for stabilization of cyclopropeneimine with a pendant amide group.

Results and Discussion

We began our studies by synthesizing the monofunctionalized diamine from the dihydrochloride salt 8 as shown in Scheme . Treatment of the dihydrochloride salt with the acyl imidazole 7a–d led to the isolation of the monoacylated diamine 9a–d. Alternatively, the monohydrochloride was treated with BOC anhydride to give the mono-BOC protected diamine. After the reaction with acid chloride, the BOC group was removed to yield the corresponding monoprotected diamine. The cyclopropenimine core was prepared using the Lambert protocol as shown in Scheme . Lambert and co-workers utilized their cyclopropeneimines in the synthesis of α,β-diamino esters. We decided to test the principles of our catalyst design by utilizing it in the synthesis of chiral diamino phosphonates. This requires the use of imino phosphosphonate esters, which are less acidic than imino esters. Previously, Bernardi and co-workers have shown that the benzophenone imine of phosphoglycine 11 can be deprotonated using cinchona alkaloid-derived quarternary ammonium hydroxide under phase transfer conditions. Further reaction of the anions with imines, proceeded with excellent enantioselectivity for certain substrates (Scheme a). However, the substrate scope was limited. Since our organic bases are less basic than hydroxide, we decided to look for alternatives to 11. In this context, Kobayashi and co-workers have reported the use of fluorenone imines as more acidic alternatives to benzophenone imines. Specifically, they showed that fluorenone imine of phosphoglycine can be deprotonated using either lithium 4-methoxy phenoxide or potassium t-butoxide and treated with N-carbamoyl imines to yield diaminophosphonates with good diastereoselectivities (Scheme b).

1. Synthesis of Catalyst.

2. Previous Syntheses of α,β-Diamino Phosphonate by (a) Bernardi and Coworkers Adapted with Permission from [Momo, R. D.; Fini, F.; Bernardi, L.; Ricci. A. Asymmetric Synthesis of α,β-Diaminophosphonic Acid Derivatives with a Catalytic Enantioselective Mannich Reaction. Adv. Synth. Catal. 2009, 351, 2283–2287]. Copyright [2009] [John Wiley and Sons] (b) Kobayashi and Coworkers Adapted with Permission from [Kobayashi, S.; Yazaki, R.; Seki, K.; Yamashita, Y. The Fluorenone Imines of Glycine Esters and Their Phosphonic Acid Analogues. Angew. Chem., Int. Ed. 2008, 47, 5613–5615]. Copyright [2008] [John Wiley and Sons].

However, enantioselective variants of this reaction were not reported. Also, it was shown that weaker bases like triethylamine were ineffective for this reaction. We hypothesized that since cyclopropeneimines are significantly stronger bases, it would deprotonate the fluorenone imine of phosphoglycine.

In our initial attempt at an enantioselective version of this reaction, we used the ethyl phosphate 15a and the N-benzyloxy carbamoyl imine 18a as the substrates (Table ). With 10 mol % of 10a as the catalyst and toluene as the solvent, we performed the reaction at room temperature. The obtained product was hydrolyzed to remove the fluorenone group and the corresponding amine was treated with benzoyl chloride to obtain the benzamide. Analysis by HPLC showed that the enantiomers were obtained in a ratio of 54:46 (Table , entry 1). In an attempt to improve the enantiomer ratio (er), we performed the reaction at −40 °C. While this led to an improvement in the diastereomer ratio, the er did not improve (Table , entry 2).

1. Variation of Selectivity with Catalyst Structure,

entry	catalyst	catalyst-R	temperature	time	yield %	dr	er
1	10a	H	RT	0.5 h	64	10:1	54:46
2	10a	H	–40 °C	12 h	26	>20:1	54:46
3	10b	4-Br	–40 °C	12 h	39	>20:1	64:36
4	10c	4-NO2	–40 °C	12 h	72	>20:1	66:34
5	10d	3,5-bis CF3	–40 °C	12 h	75	>20:1	59.5:40.5
6	10b	4-Br	–50 °C	26 h	99	>20:1	66.5:33.5
7	10c	4-NO2	–50 °C	NR	NA	NA	NA
8	10d	3,5-bis CF3	–50 °C	4 h	80	>20:1	72:28
9	10d	3,5-bis CF3	–80 °C	54 h	69	>20:1	56.5:43.5

^aReaction conditions: (i) 0.28 mmol of 15a, 0.56 mmol of 18a, 0.028 mmol of catalyst, 1.5 mL of toluene, 300 mg of 3 Å molecular sieves. (ii) 0.4 mL of 1 M HCl, 8 mL of THF. (iii) 3 eq. of BzCl, sat. aq. NaHCO₃.

^bIn the absence of catalyst, there was no reaction.

^cCrude yield after hydrolysis.

^dCatalyst is insoluble below −50 °C.

Next, we varied the acyl group on the catalyst and studied its effect on the enantioselectivity of the reaction. Previously, Sigman and co-workers have reported on the effect of the catalyst acidity in a hydrogen-bond-promoted hetero Diels–Alder reaction. In their study, a chiral oxazoline with a pendant carboxamide group was used as the catalyst. The acyl group on the carboxamide was systematically varied and its effect on the selectivity was studied. A linear relationship was observed between the pK _a of the parent carboxylic acid and the logarithm of the enantiomer ratio. Based on this, we hypothesized that varying the acidity of the carboxamide group might have an impact on the selectivity. Using the method described above, we synthesized three catalysts with various substituents on the aromatic ring. Reactions were performed with each of these catalysts at −40 °C (Table , entries 3, 4, 5). Under these conditions, catalyst 10c with a nitro substituent gave the best performance with an er of 66:34 (Table , entry 4). To further improve the selectivity, reactions with each of these catalysts were performed at −50 °C. Somewhat surprisingly, with catalyst 10c, there was no reaction (Table , entry 7). However, catalyst 10d gave an improved er of 72:28 (Table , entry 8). When we lowered the reaction temperature further to −80 °C, the selectivity decreased (Table , entry 9). Since the highest er was obtained with catalyst 10d, we chose this catalyst for further optimization studies.

At this point, we sought to improve the selectivity by varying the phosphate substituent as shown in Table . Increasing the steric bulk from an ethyl group to an isopropyl group led to the shutting down of the reaction at −50 °C. However, when the reaction was allowed to slowly warm up to room temperature, we obtained the product with an er of 87.5:12.5. Further increase in the steric bulk led to a drastic reduction in reactivity as well as selectivity. Puzzled by the lower selectivities at −80 °C, we decided to carefully study the role of temperature and catalyst loading on selectivity. To do this, we carried out reactions with 1, 5, and 10 mol % of catalyst at three different temperatures. Two trends were clearly visible from these experiments. For each catalyst loading, the reaction at room temperature gave the highest er with a steady decrease in the er as the temperature was lowered. (Table ). At the same temperature, the lowest catalyst loading consistently gave the highest selectivity. Song and co-workers have observed similar trends in the desymmetrization of meso anhydrides with a cinchona alkaloid-based thiourea catalyst. They attributed this to the existence of a monomer–dimer equilibrium of the catalyst. In light of this, our results are also attributable to catalyst aggregation.

2. Variation of Enantioselectivity with Change in Phosphonate R Group .

entry	substrate	R	temperature	time	yield %	dr	er
1	15a	Et	–50 °C	4 h	80	20:1	72:28
2	15b	i-Pr	–50 °C to RT	100 h	89	8:1	87.5:12.5
3	15c	i-Bu	–20 °C	204 h	19	5:1	53:47

^aReaction conditions: (i) 0.28 mmol of 15a–c, 0.56 mmol of 18a, 0.028 mmol of catalyst, 1.5 mL of toluene, 300 mg of 3 Å molecular sieves. (ii) 0.4 mL of 1 M HCl, 8 mL of THF. (iii) 3 eq. of BzCl, sat. aq. NaHCO₃.

^bCrude yield after hydrolysis.

3. Variation of Selectivity with Temperature and Catalyst Loading .

entry	catalyst	catalyst mol %	temperature	time	% yield	er
1	10d	1	25 °C	<30 min	43	88:12
2	10d	5	25 °C	<30 min	82	84.5:15.5
3	10d	10	25 °C	<30 min	80	83:17
4	10d	1	0 °C	96 h	40	86:14
5	10d	5	0 °C	30 min	79	75.5:24.5
6	10d	10	0 °C	30 min	67	73:27
7	10d	1	–20 °C	NR
8	10d	5	–20 °C	30 min	84	62:38
9	10d	10	–20 °C	20 min	82	61.5:38.5
10	20	5	25 °C	39 h	64	86.5:13.5
11	20	5	0 °C	67 h	60	84:16
12	20	5	–20 °C	168 h	56	56:44

^aReaction conditions: (i) 0.28 mmol of 15b, 0.56 mmol of 18a, 0.0028 mmol of catalyst (10d, 20), 1.5 mL of toluene, 300 mg of 3 Å molecular sieves. (ii) 0.4 mL of 1 M HCl, 8 mL of THF. (iii) 3 eq of BzCl, sat. aq. NaHCO₃.

^bCrude yield of amine.

To understand whether the dialkyl substituent on the nitrogen plays a role in catalyst aggregation, we decided to make the diisopropyl substituted catalyst. The diisopropyl substituent is smaller in size compared to the dicyclohexyl substituent. Lambert and co-workers had shown that the smaller size planarizes the nitrogen. This enhances the lone pair conjugation and results in greater basicity. Importantly, we hypothesized that the smaller size of the alkyl substituent would likely lead to greater aggregation. To test this, we performed our standard reaction with substrate 15b using 5 mol % of catalyst 20 (Table , entries 10, 11, and 12). The reactions were slower than the reactions performed with catalyst 10d at corresponding temperatures. This is attributable to the smaller catalyst size leading to greater formation of a less reactive aggregate. Once again, the reaction performed at room temperature gave the product with the highest er (Table , entry 10).

When the reaction was carried out at 0 °C, the er was reduced to 84:16. Lowering the temperature to −20 °C led to a more significant reduction in the er (56:44) than in the reaction catalyzed by 10d (Table entries 2 and 8 and entries 10 and 12). This is consistent with the smaller size of the catalyst favoring greater aggregation. To further clarify this, we carried out ¹HNMR studies of catalyst 10d in C₆D₆. The concentration of the catalyst was varied from 2.6 mM to 42.7 mM. At the lowest concentration, the two hydrogens on the chiral center in the catalyst appeared together as a broad singlet (Figures S56 and S57). As the concentration was raised, the broad peak resolved into two multiplets. We then added DMSO to the solution that had the highest catalyst concentration and the multiplets reverted to a poorly resolved broad peak. This suggests that the catalyst likely forms aggregates in nonpolar solvents.

We then evaluated the substrate scope of the reaction using catalyst 10d. When the substrate carried a 4-methoxy substituent, the corresponding product was obtained with an er of 88.5:11.5 (Table ). Similarly, with the 4-methyl derivative as the substrate, the product was obtained with an er of 88.5:11.5, while the 3-thienyl derivative gave an er of 83.5:16.5. The highest er of 92.5:7.5 was obtained with 1-naphthyl-derived substrate. In general, substrates with an electron donating substituent gave good ers. However, when an electron-withdrawing substituent was used, a significant reduction in the er was observed (Table o-ClC₆H₄ derivative). This is in contrast to the results reported by Bernardi and co-workers. In their reaction, substrates with an electron-withdrawing substituent gave higher selectivities in comparison to substrates with an electron-donating substituent. Thus, the reactions performed with our catalysts provide a complementary set of conditions to the ones reported by Bernardi and co-workers. In order to determine the identity of the major diastereomer, we hydrolyzed 21b to obtain the diamino phosphoric acid. The NMR of the major isomer matched with the reported NMR for the syn isomer.

4. Substrate Scope .

entry	aryl imine	aryl imine-Ar	product	time	yield %	dr	er
1	18a	phenyl	21b	24 h	43	6.2:1	88:12
2	18b	3-thienyl	21d	48 h	61	2.2:1	83.5:16.5
3	18c	4-methoxy phenyl	21e	48 h	47	4.5:1	88.5:11.5
4	18d	2-chloro phenyl	21f	120 h	31	2.9:1	76.5:23.5
5	18e	1-naphthyl	21g	120 h	51	3.9:1	92.5:7.5
6	18f	4-methyl phenyl	21h	48 h	56	2.4:1	88.5:11.5
7	18g	3-methoxy phenyl	21i	18 h	55	1.5:1	84.5:15.5

^aReaction conditions: (i) 0.28 mmol of 15b, 0.56 mmol of 18a–g, 0.0028 mmol of catalyst, 1.5 mL of toluene, 300 mg of 3 Å molecular sieves. (ii) 0.4 mL of 1 M HCl, 8 mL of THF. ers were measured for the corresponding benzoyl derivative.

^ber for minor diastereomer (the major diastereomer was nearly racemic).

Finally, we tested the durability of catalyst 10d by storing it under ambient conditions for three months. The aged catalyst gave the product with same yield and selectivity as a freshly prepared batch of catalyst. This shows that the catalyst can be stored under ambient conditions without taking special precautions. Additionally, we used NMR spectroscopy to study the solution state stability of this catalyst. A solution of the catalyst in C₆D₆ did not show any deterioration over a period of 10 days.

Conclusions

In conclusion, we have developed a cyclopropeneimine-based catalyst with a pendant hydrogen-bond donor that can be tuned easily. The prepared catalyst is stable under ambient conditions in the solid state as well as in solution. An enantioselective synthesis of diaminophosphonate was explored with this catalyst. Utilizing the tunability of the hydrogen-bond-donating arm, we were able to optimize the selectivity of the reaction. Imines with electron-donating groups on the aromatic ring gave good enantiomer ratios, while those with electron-withdrawing groups gave more modest ers. In this regard, the method is complementary to the one disclosed by Bernardi and co-workers. More importantly, we have shown that the catalyst is both tunable and stable. The decrease in selectivity with lower temperatures and higher catalyst loading are indicative of catalyst aggregation and this is supported by ¹H NMR studies. Improvements to enantioselectivity should be possible with better structural understanding of this issue. Further exploration of this catalyst in other enantioselective reactions is also under progress.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c02727.

• Synthetic procedures, spectroscopic data, and HPLC data (PDF)

§.

Bio Sciences, Corporate Division, ITC Limited, ITC Life Sciences & Technology Centre, 1st Main, Phase 1, Bangalore 560058, India

The authors declare no competing financial interest.