The restriction of raw materials and resources led the organic chemists to change their mindset and design chemical processes based on the “Sustainable Development” concept. Inspired by Nature, since the late 1990s, scientists have laid the groundwork for asymmetric organocatalysis (Ahrendt et al., 2000; List et al., 2000), allowing a green and direct access to highly functionalised chiral products, including important key intermediates in the total syntheses of bioactive compounds (Xiang and Tan, 2020; Han et al., 2021).

At the dawn of organocatalysis, (S)-Proline (Pro) was identified as “the simplest enzyme” because of its ability to promote enantioselectivity in different reactions (List et al., 2000; 2001; Movassaghi and Jacobsen, 2002). Since then, plenty of analogues were designed to bypass the Pro limitations as an organocatalyst, such as its poor solubility in organic solvents (Obregón-Zúñiga et al., 2017). Moreover, the introduction of sterically hindered groups allowed the formation of more rigid transition states, leading to better stereo-induction within the studied reaction (Seebach et al., 1985; Seebach et al., 2013; Liu and Wang, 2017). As an example, MacMillan (Ahrendt et al., 2000) and Hayashi–Jorgensen (Hayashi et al., 2007; Reyes et al., 2007) catalysts are at present commercially available compounds for routine enantioselective syntheses. Due to the increasing demand of chiral compounds, the research of new organocatalysts continues to be a hot topic of research.

Recently, our research group reported on the synthesis of non-natural ß-amino acids (β-AA) with a constrained heterocyclic core (Oliva et al., 2019), mostly focussing on morpholine ß-amino acids (β-Morph-AAs) for different applications (Penso et al., 2012; Bucci et al., 2019; Bucci et al., 2020; Bucci et al., 2021; Vaghi et al., 2020), i.e., from the synthesis of photoluminescent nucleopeptides (Bucci et al., 2020) to their use as inducers of the polyproline helix when inserted in the model’s peptides (Bucci et al., 2019; Bucci et al., 2021; Vaghi et al., 2020).

Being inspired by the use of nitrogen-containing heterocycles in asymmetric synthesis via enamine, here, we studied the use of new ß-Morph-AAs as very challenging and stimulating organocatalysts. It has already been reported that in comparison to enamines with a pyrrolidine and piperidine core, the morpholine cores are orders of magnitude less reactive. Pyrrolidine enamines are the most reactive due to the higher p-character of the nitrogen lone pair in their five-membered ring, indicating higher nucleophilicity compared to the six-membered piperidine ring. The presence of oxygen in morpholine–enamines further increases the ionisation potential and consequently reduces nucleophilicity compared to piperidine cores (Kempf et al., 2003). Moreover, the most pronounced pyramidalisation of morpholine–enamines, resulting in poor reactivity, should be another limitation of the proposed catalysts (Brown et al., 1978; Schnitzer et al., 2020b).

To test our catalysts, we focussed on a model Michael addition reaction between aldehydes and nitrostyrenes, which is usually promoted by pyrrolidine-based organocatalysts. As the main drawback, except for few examples (Lombardo et al., 2009; Borges-González et al., 2019; Schnitzer et al., 2020a; Schnitzer et al., 2020b), their use requires a 10–20 mol% of the catalyst and an excess of the carbonyl compound (List et al., 2001; Sakthivel et al., 2001; Hayashi et al., 2005; Choudary et al., 2007; Ni et al., 2007; Llopis et al., 2018).

It has to be underlined that some base research on morpholine–enamine to understand the topological rule for C,C-bond-forming processes between prochiral centres was conducted by Seebach and Goliński (1981). On the other hand, the use of morpholine catalysts is, in general, very limited, mostly in terms of reagent conversion. Moreover, to the best of our knowledge, only few examples of chiral morpholine organocatalysts were tested for this reaction, yielding poor diastereo- and enantioselectivity (Mossé et al., 2006; Laars et al., 2009; Laars et al., 2010).

Starting from inexpensive commercially available a-AAs and chiral epichlorohydrin, we designed a straightforward enantioselective synthesis of ß-Morph-AAs I–IV (Figure 1) with a different sterically hindered group at C-5, which is derived from the a-AA side chain, and a carboxylic function at C-2, which is crucial for the success of this reaction. By playing with different stereochemistries of the two starting materials, it is possible to modulate the formation of cis or trans isomers of the two substituents and their absolute configuration that will reflect on the stereochemistry of the final compound.

The efficacy of the hindered group, together with the best steric relationship between C-2 and C-5 substituents of morpholine ring, was investigated. Both experimental and computational data confirmed that catalyst I has a remarkable ability to control the diastereo- and enantioselectivity of the 1,4-addition reaction between aldehydes and nitroolefins. Despite the already explained limitations of morpholine catalysts with enamine mode of action, different from the majority of the reported organocatalysts, only 1 mol% of I and 1.1 eq. of aldehyde are required to reach a quantitative conversion of the reagents. We also proved the crucial role of the carboxylic group; i.e., in the presence of I capped as methyl ester under standard conditions, no condensation products were observed after 48 h. Furthermore, excellent diastereoselection was detected (90%–99% d. e.), along with the enantioselection ranging from 70% to 99% e. e., depending on the reagents.

5-Substituted ß-Morph-AAs were synthesised from commercially available (R) or (S) a-AAs and (R)-epichlorohyidrin (Scheme 1). AAs 1 were treated with NaBH₄ (2.5 eq.) and I₂ (1 eq.) in refluxing THF, yielding the corresponding amino alcohol 2 (55%–72%). Subsequent reductive amination of 2 with benzaldehyde (1.3 eq.) and NaBH₄ (3 eq.) in MeOH at r. t. afforded compounds 3 (56%–85%). According to a one-pot procedure reported in the work of Breuning et al. (2007), Morph-derivatives 4 (66%–75%) were obtained by the treatment of 3, first with R-epichlorohydrin (1.3 eq.) in the presence of LiClO₄ (1.3 eq.) in toluene (60°C) and then with MeONa in MeOH. Using H₂ and Pd/C (10% loading) in the presence of Boc₂O (1.05 eq.) in THF, 4 was transformed into the Boc-protected amino alcohol 5 (76%–87%). Oxidation with TEMPO (0.2 eq.) and BIAB (2 eq.) in CH₂Cl₂/H₂O (2:1) provided the desired ß-Morph-AAs 6 (65%–71%) and then deprotected yielding catalysts I–IV, as CF₃CO₂H salts, characterised by different stereochemistry and substitution patterns.

In order to demonstrate the potential of our Morph-catalysts, the reaction between butyraldehyde (7a, 1 eq.) and trans-β-nitrostyrene (8a, 1.5 eq.) was chosen as the model. In principle, it can provide two diastereoisomers as a couple of enantiomers, i.e., (2R*,3S*)- and (2S*,3S*)-2-ethyl-4-nitro-3-phenylbutanals (9). To control the diastereo- and enantioselection, several reaction conditions (solvent, temperature, and reaction time) were tested using 1% of catalysts I–IV (Table 1) in the presence of N-methylmorpholine (NMM, 1 mol%) as the base, in order to obtain the free amino group of the catalyst.

Catalyst I was selected as the first catalyst to predominantly yield the syn (2R,3S)-adduct 9. Different solvents were screened (40°C, 12 h; entries 1–5, Table 1). Mixtures with fluorinated alcohols were also tested for their peculiar features, being known as efficient additives in Michael additions with proline as the catalyst (Pellissier, 2021).

In general, an excellent conversion of the reagents was achieved, revealing the use of alcoholic solvents or co-solvent beneficial for diastereoselection. These solvents were, thus, selected, and additional studies were performed, decreasing the temperature (0°C, 12h, entries 6–8; −10°C, 24 h, entries 9–11; Table 1). iPrOH was found to be the best solvent at −10°C (entry 11, Table 1), giving the expected product 9 with quantitative conversion and higher diastereo- (96% d. e.) and enantioselectivity (90% e. e.). This result agrees with the computational analysis that predicted the importance of protic solvents for the stabilisation of the transition state (see the following sections).

Catalyst II, having the opposite configuration at C-5, was then tested (entries 12–15, Table 1), which resulted to be less efficient; i.e., the reaction reached a quantitative conversion but only at 40°C, giving lower d. e. and e. e. (entries 12–14, Table 1). Operating in iPrOH at −10°C, 85% d. e., and 86% e. e. was indeed found but with a modest conversion (entry 15, Table 1). Interestingly, the inversion of the configuration at C-5 of II with respect to I allows the obtainment of syn (2S,3R)-adduct as the main enantiomer (entries 12–15, Table 1).

Since the (2S,5R)-stereochemistry of I resulted to be the most efficient, we chose to evaluate the effect of the group at C-5 by selecting phenyl and CH₂-2-naphtyl groups, considering their limited freedom and increased bulkiness, respectively. The phenyl group in III at 40°C caused a loss in the conversion rate and stereocontrol (entries 16–18, Table 1). Moreover, operating in iPrOH at −10°C, only traces of the desired compound were detected (entry 19, Table 1). On the other hand, operating at −10°C for 24 h and using the more hindered catalyst IV (entries 20–21, Table 1), 60% conversion, high level of diastereoselection (99% d. e.), and satisfactory enantioselection (73% e. e.) were reached. By increasing the reaction time, the conversion was slightly increased but with the loss of enantioselectivity (entry 21, Table 1).

In summary, the C-5 benzyl group of I, cis with respect to the C-2 carboxylic function, induces an excellent diastereo- and enantioselectivity operating in iPrOH at −10°C. Thus, with only 1% of the catalyst and only 1:1.5 ratio of 7a:8a, excellent conversion was observed. By using the aforementioned best conditions and catalysts, a series of aldehydes 7b–g and nitroolefins 8b–d were screened to expand the scope of the reaction (Table 2). To reach the right balance between conversion, d. e. and e. e., different attempts were carried out, using different amounts of catalyst, reaction times, and temperatures. The best results are summarised in Table 2 (additional results are described in SI, including two examples with aliphatic nitroolefins that gave poor results in terms of d. e. and e. e.).

Elongating the alkyl chain of the aldehyde, a decrease in conversion was observed, while a growth of e. e. was obtained with excellent d. e. (up to 99%, entries 1–4, Table 2).

On the other hand, compound 13 (entry 5, Table 2) was obtained with an e. e. slightly lower than 90% probably because of the multi-degree of freedom arising from the alkyl chain of the hexanal. To obtain compound 14 from phenylpropionaldehyde (7f) with satisfactory conversion, the amount of the catalyst was increased to 5% (comparing entries 6 and 7, Table 2). With the hindered cyclopentylaldheyde (7g), a similar amount of catalysts was needed, but operating at 40°C. A low conversion along with a moderate enantioselectivity was observed (entry 8, Table 2). Finally, we focussed on the reactivity of nitrostyrenes 8 containing an electron-rich (i.e., 4-MeOPh, thiophenyl; entries 9–11, Table 2) or electron-poor (i.e., 4-ClPh; entries 12–13, Table 2) aromatic moiety that were matched with 7b and 7c. With an exception (entry 10, Table 2), the reaction works with 1% of the catalyst. As a general trend, the more electron rich the aryl substituent in compounds 8 is, the higher is the conversion (comparing entries 9 with 12 and entries 10 with 13, Table 2), and in all cases, high d. e. and satisfactory e. e. were detected (d.e. > 89% and e. e. 67%–80%).

The mechanism for the addition of aldehydes to nitroalkenes through enamine catalysis was previously analysed both theoretically and experimentally (Sahoo et al., 2012; Földes et al., 2017). Two main hypotheses were carried out for the addition: the first implies the formation of a zwitterion (Int1-ZW-like, Figure 2) derived from the addition of the enamine to the ß-carbon of the nitroalkene. The second one suggests cycloaddition, leading to dihydrooxazine oxide (Int1-OX-like, Figure 2) in equilibrium with a cyclobutane species (Int1-CB-like, Figure 2). The dihydrooxazine oxide intermediate can be protonated at the a-carbon with respect to the nitro group and evolve to the imino intermediate (Int2-like, Figure 2) that is hydrolysed to the final product.

To elucidate the possible role of our catalyst, we decided to model the equilibrium steps described in Figure 2 by using consolidated density functional theory (DFT) methods (Gassa et al., 2010; Grimme et al., 2010; Sahoo et al., 2012; Giofrè et al., 2021).

We initially performed a conformational search on the hypothesised zwitterion Int1-ZW at the molecular mechanic level, considering both diastereoisomers. All conformations within 3.0 kcal/mol were optimised at the DFT level, but all simulations either failed or converged into the Int1-OX intermediate (Figure 2) for both (4R,5S) and (4S,5R) stereoisomers. These results, together with the lack of experimental evidence of the zwitterion intermediate, suggested that the reaction can directly proceed with a cycloaddition-like mechanism.

Consequently, we modelled both the Int1-OX and Int1-CB intermediates, considering both the 4S,5R/4R,5S or 2S,3R/2R,3S stereoisomers, respectively (Figure 2), that were subjected to a conformation search using molecular mechanics. Considering that a transient stereocentre is formed at the C-bearing morpholine group, both configurations were evaluated, and the most stable stereoisomer was further considered. All conformations within a range of 3 kcal/mol were optimised by DFT, and energy was calculated by considering iPrOH solvent effects and empirical correction for dispersive interactions (Grimme et al., 2010).

For both stereoisomers, the most stable conformation of the Int1-OX intermediate was used to model the transition state (TS1) for the attack of the enamine E-7a to the nitrostyrene 8a. The lowest-energy transition states (TSs) are shown in Figure 3.

The geometries of the two TS1 stereoisomers indicate that the reaction formally occurs as a [4 + 2] cycloaddition directly leading to the Int1-OX intermediates. This has been confirmed by intrinsic reaction coordinate (IRC) calculations that showed TS1 connecting the activated complex between E-7a and nitrostyrene 8a to Int1-OX (Supplementary Figures S1, S2, SI). A TS directly leading to Int1-CB was not located, even if experimental evidence shows that Int1-CB is often the dominant species (Burés et al., 2011; Sahoo et al., 2012). We hypothesised that the equilibrium between Int1-OX and Int1-CB occurs through a concerted ring opening/closure, according to Scheme 2. Despite several attempts, a unique TS for this equilibrium reaction was not found. However, since Int1-CB can be considered a dead end in the reaction mechanism (Sahoo et al., 2012), we focussed on identifying TSs that are relevant to step 3 (Figure 2).

Thus, TS2 was modelled by starting from the lowest-energy structure of Int1-OX. It was shown that the protonation step triggering the ring opening in step 3 was the reaction rate determining step (rds) (Sahoo et al., 2012). We, thus, hypothesised that the carboxylic group of catalyst I might play a key role in controlling the enantioselectivity observed in this study, as confirmed by experimental data (mentioned previously). We also hypothesised that iPrOH, used as a solvent, might bridge the H-transfer between the carboxylic group of the morpholine moiety and Cα-NO₂. Thus, one molecule of iPrOH was explicitly considered in TS2 geometries for both stereoisomers. Several conformations were evaluated for each TS2, and both the E and Z configurations were considered for the imino group of Int2. IRC calculations were conducted to confirm that TS2 connects Int1 to Int2 (Supplementary Figures S1, S2, SI). To compute that activation barriers and reaction energies are comparable with those reported in previous computational studies (Sahoo et al., 2012), all the selected stationary points were reoptimised using the ωB97X-D functional and 6-311G (d,p) basis set, including the solvent effect for iPrOH. More accurate single-point energies were then computed at the same level of theory by using 6-311++G (3df,3pd), as suggested by Sahoo et al. (2012). The obtained free energies were then used to draw the reaction path represented in Figure 4. The most stable geometries for (2R ,3S)- and (2S,3R)-TS2 are shown in Figure 5. We can observe that the energy path relative to the formation of the (2S,3R)-9 enantiomer is characterised by higher activation free energy barriers (ΔΔG^‡) for both TS1 and TS2, compared to the favoured (2R,3S)-9 enantiomer. Among the two energy paths, the greatest difference in ΔΔG^‡ is observed for TS2, where the protonation step occurs concerted to the ring opening (Figures 4, 5). This step is then confirmed as the r.d.s and the path leading to the isolated (2R,3S)-9 as the kinetically favoured path. Interestingly, the Int1 intermediate results as the global minimum on both the free energy (Figure 4) and enthalpy path (Supplementary Figure S3, SI), suggesting that the hydrolysis of the Int2 intermediate is the non-equilibrium step that drives the reaction towards the final product.

TS2 geometries (Figure 5) evidenced the role of the ß-Morph carboxylic acid in self-catalysing the proton transfer from iPrOH to Cα-NO₂. This role was also confirmed experimentally; i.e., the reaction between 7a and 8a catalysed by the methyl ester of I was performed under standard conditions, but no condensation products were observed after 48 h.

Interestingly, both TS2 structures share similar geometrical parameters related to the H-transfer (see distances in Figure 5). However, the most relevant difference that could be associated to the greater stability of (2R,3S)-TS2, as compared to (2S,3R)-TS2, is the configuration at the imino group of the forming Int2 product (Supplementary Figure S1, SI). Indeed, the opening of the dihydrooxazine oxide ring led to the E-configuration for (2R,3S)-TS2, while the more hindered Z-configuration is obtained from (2S,3R)-TS2 (Schnitzer et al., 2020a). To confirm this finding, the corresponding Z and E-TSs were also located for (2R,3S)-TS2 and (2S,3R)-TS2, but higher energies were obtained in both cases (Supplementary Figure S4, SI).

In conclusion, we presented here a new class of ß-AAs with a morpholine core prepared by a straightforward synthesis, from commercially available a-AAs and chiral epichlorohydrin, that can control the configuration of substituent patterns. Despite the known limitations of the morpholine ring during enamine catalysis, our results provided the first evidence on the actual effectiveness of this chiral catalyst that works efficiently in the selected 1,4-addition model reaction between aldehydes and nitroolefins, thanks to the presence of a carboxylic moiety in position ß to the amine. Experimental data have proven that the reaction goes with excellent conversion and diastereoselection and satisfactory to excellent enantioselection, depending on the substitution pattern of the two reagents. It is to be noted that only 1% of the catalyst and a 1/1.5 ratio of 7/8 is needed, iPrOH being the key solvent. These results were supported by a theoretical study on the best catalyst, evidencing the role of the ß-Morph carboxylic group in self-catalysing the protonation of the dihydrooxazine oxide intermediate, which is generally considered the rate limiting step of this class of reactions. Experiments confirmed this observation, laying the groundwork for further optimisation of the catalyst and opening its general use for a plethora of asymmetric syntheses.