Introduction

Chirality has been recognized as an essential entity for life. However, it still remains a big mystery how the homochirality in nature emerged in essential biomolecules, such as sugars and amino acids1,2. Certain achiral molecules are known to form crystals adopting a chiral geometry, which usually form as a conglomerate, i.e., a racemic mixture of enantiomorphic crystals. Post chemical transformation of such geometrically chiral single crystals may afford enantiomerically enriched products3,4,5. This is one of the typical approaches to absolute asymmetric synthesis6,7,8, which has attracted long-term attention since the time of Pasteur. If mirror symmetry breaking occurs in the crystallization of achiral compounds, a mixture of unequal numbers of enantiomorphic crystals forms. For example, Kondepudi et al. reported that stirred crystallization is often effective to induce mirror symmetry breaking, leading to a high enantiomeric excess (ee) of the resulting crystal9,10,11. Notably, Viedma et al. later demonstrated that continuous grinding of a slurry of a racemic mixture of enantiomorphic crystals could cause its deracemization to produce an enantiopure product (Viedma ripening)12,13,14,15,16,17,18. Especially, Blackmond and coworkers for the first time experimentally supported a hypothesis of the emergence of solid-phase homochirality from racemic conglomerates composed of rapidly racemizing enantiomers13,19,20. Tsogoeva and coworkers achieved a complete deracemization (100% ee) of an asymmetric reaction product by combining its reversible asymmetric reaction and physical grinding of crystals involved21,22. Nevertheless, an even more challenging subject to tackle is how one can amplify the enantiomeric bias generated as a consequence of mirror symmetry breaking. In 2008, Soai and coworkers employed such a mirror symmetry-broken crystalline mixture of cytosine as a chiral auxiliary for the enantioselective addition reaction between diisopropylzinc and pyrimidine-5-carbaldehyde23. Although the range of eligible substrates is highly limited, this reaction is well-known for amplifying the enantiomeric bias in an autocatalytic manner24.

Some covalent helical polymers25,26,27,28,29,30,31,32,33,34,35 and their noncovalent analogues36,37,38,39,40, prepared from certain chiral monomers, are known to serve as chiral ligands for asymmetric catalysis. Suginome and coworkers have developed a series of helical poly(quinoxaline-2,3-diyl) copolymers bearing metal-binding phosphine pendants as chiral ligands for several palladium-catalyzed asymmetric reactions29,30,31,32,33,34. Examples of noncovalent helical supramolecular assemblies include hydrogen-bonded helical stacks of chiral benzene-1,3,5-tricarboxamide (BTA) derivatives37,38,39,40. In our previous paper, we reported that an achiral BTA motif that was appended with an ethyl cinnamate group in each of its three side chains self-assembled into helical nanoribbons, where mirror symmetry breaking occurred under certain conditions to form a mixture of unequal numbers of right-handed (P) and left-handed (M) helical nanoribbons41,42. We envisioned that such mirror symmetry-broken supramolecular assemblies41,42,43,44,45,46,47,48,49,50,51,52,53,54 might mediate asymmetric reactions if they are properly designed to accommodate catalytically active transition metal ions.

Herein, we report the first example of asymmetric catalysis realized by using a mirror symmetry-broken helical supramolecular assembly of an achiral BTA motif as the ligand (Fig. 1). The achiral BTA derivative (BTABA, Fig. 1a)55 carries a benzoic acid group in each of its three arms for chelating a Cu2+ ion that can catalyze the Diels–Alder cycloaddition reaction between aza-chalcone and cyclopentadiene (Fig. 1c). The enantiomeric excess of the reaction reaches 46% (conversion: >99%, turnover number: ~90).

Fig. 1
figure 1

Mirror symmetry breaking of helical supramolecular nanoribbons for asymmetric catalysis. a, b Molecular structure of the achiral BTABA monomer and schematic representation of the preparation of helical PBTABA nanoribbons with either (P)- or (M)-dominant helical handedness by applying rotary stirring during the cooling process of a DMF/water solution of BTABA. (P)- and (M)-dominant PBTABA display negative ((−)-PBTABA) and positive ((+)-PBTABA) CD signs at 316 nm, respectively. c Diels–Alder cycloaddition reaction between aza-chalcone and cyclopentadiene catalyzed by Cu2+ in combination with (−)-PBTABA or (+)-PBTABA. The reaction affords the endo isomer preferentially (~90%). The (S)-endo and (R)-endo enantiomers were obtained preferentially when (−)-PBTABA and (+)-PBTABA were used, respectively

Full size image

Results

Mirror symmetry breaking of helical nanoribbons

BTABA was synthesized and unambiguously characterized (see Supplementary Methods, Supplementary Figs. 1−6). Analogous to previous examples41,42,55, BTABA self-assembled into PBTABA, a supramolecular helical nanoribbon (Fig. 1b). We found that this self-assembly event proceeds with mirror symmetry breaking to afford either (P)-dominant PBTABA ((−)-PBTABA) or (M)-dominant PBTABA ((+)-PBTABA) (Fig. 1b) under physical agitation generated by magnetic rotary stirring. The method for the preparation of mirror symmetry-broken PBTABA is slightly complicated as described below (Fig. 2) but quite analogous to that reported in our previous paper42: Water (0.3 mL) was added to a N,N-dimethylformamide (DMF, 0.7 mL) solution of BTABA (3 mg, 5.3 × 10−6 mol) in a 5-mL cylindrical glass vial (Fig. 2a), whereupon the mixture, when allowed to stand at 25 °C for 2 h, underwent physical gelation (Fig. 2b and Supplementary Tables 1 and 2). The gel was heated at 100 °C for 3 min, and the resultant homogeneous solution (Fig. 2c) was allowed to cool to 25 °C with magnetic rotary stirring (1200 rpm) for 40 min, yielding a white suspension (Fig. 2d, e), which turned out to be optically active (see below). This result was highly reproducible (Supplementary Fig. 7), except that the sign of the circular dichroism (CD) activity at 316 nm of the resultant suspension was either negative ((−)-PBTABA) or positive ((+)-PBTABA) stochastically (see below). Although the CD sign was independent of the direction of rotary stirring (Supplementary Fig. 7), a stable CD activity, once emerged, lasted without fading unless the suspension was heated to transform into a homogeneous solution. Note that, as demonstrated for the suspension of (−)-PBTABA in Supplementary Fig. 8, its CD spectrum was barely contaminated with a spectral artifact due to linear dichroism (LD). Next, an aliquot of the suspension was taken out from the glass vial and cast on a silicon wafer for air-drying, followed by evacuation under reduced pressure. Scanning electron microscopy (SEM) of the resulting specimens revealed that (−)-PBTABA consisted mostly of (P)-helical nanoribbons (Fig. 2d), whereas (M)-helical nanoribbons were dominant in (+)-PBTABA (Fig. 2e). The gel (Fig. 2b) was CD-silent (Supplementary Figs. 9 and 10). Consistently, its cross-linked 3D network appeared to be composed of equal numbers of (P)- and (M)-helical nanoribbons (Fig. 2b). The same held true when a clear, hot solution of BTABA (Fig. 2c) was allowed to cool to 25 °C without rotary stirring (Fig. 2f and Supplementary Fig. 11). Not only magnetic rotary stirring but also mechanical rotary stirring using an overhead stirrer, vortex mixing, and even sonication resulted in the formation of a CD-active suspension of (−)-PBTABA or (+)-PBTABA (Supplementary Fig. 12).

Fig. 2
figure 2

Procedures for the preparation of mirror symmetry-broken helical PBTABA nanoribbons. a Photograph of a DMF (0.7 mL) solution of BTABA (3 mg, 5.3 × 10−6 mol). b Photograph of a CD-silent gel of PBTABA ([BTABA] = 5.3 mM) formed in DMF/water (7/3, v/v; 1 mL) and SEM image of its air-dried specimen. The gel can be prepared via an antisolvent method through direct addition of water (poor solvent) into a DMF (good solvent) solution of BTABA without magnetic rotary stirring. c Photograph of a DMF/water (7/3, v/v; 1 mL) solution of BTABA (5.3 mM) after heating at 100 °C for 3 min. d, e Typical photographs of CD-active DMF/water (7/3, v/v; 1 mL) suspensions of PBTABA ([BTABA] = 5.3 mM) and SEM images of their air-dried specimens. The suspensions of (−)-PBTABA and (+)-PBTABA display mirror-image CD spectra of one another (Fig. 4b). They can be prepared by natural cooling of a clear, hot DMF/water (7/3, v/v; 1 mL) solution of BTABA (5.3 mM) with magnetic rotary stirring at 1200 rpm for 40 min. f Photograph of a CD-silent precipitate of PBTABA ([BTABA] = 5.3 mM) in DMF/water (7/3, v/v; 1 mL) and SEM image of its air-dried specimen. This precipitate can be prepared by natural cooling of a clear, hot DMF/water (7/3, v/v; 1 mL) solution of BTABA (5.3 mM) without magnetic rotary stirring

Full size image

Interestingly, an ethyl ester derivative of BTABA did not assemble into helical nanoribbons under otherwise identical conditions to those for BTABA (Supplementary Fig. 13), suggesting that the peripheral carboxylic acid groups of BTABA play a critical role in its helical assembly. Fourier transform infrared (FT-IR) spectroscopy of a xerogel of PBTABA, dried under reduced pressure, showed small vibrational bands at 2665 and 2543 cm−1 (Supplementary Fig. 14), which are characteristic of the OH stretching vibrations due to hydrogen-bonded dimers of carboxylic acids. Two bands at 3311 and 1654 cm−1, assignable to the stretching vibrations of hydrogen-bonded amide NH and C = O groups, respectively, were also observed. We found that BTABA is quite soluble in DMF alone ([BTABA] = 5.3 mM), where it showed a much sharper absorption band at 290 nm (Supplementary Fig. 15) than its self-assembled state ([BTABA] =5.3 mM) in the gel formed in DMF/water (7/3, v/v, Fig. 2b). Compared with the DMF solution of BTABA, the gel was much more luminescent (~300 times) at 447 nm upon excitation at 290 nm (Supplementary Fig. 16), suggesting that the BTABA units in the gel network are located in a highly congested environment. Further to note, a DMF/water suspension of PBTABA (Fig. 2d/e) was silent in polarized optical microscopy (Supplementary Fig. 17), and a wet sample of PBTABA isolated by centrifugation of its suspension showed no X-ray diffraction (Supplementary Fig. 18). These results allow us to conclude that PBTABA is a noncrystalline material.

When a clear, hot solution of BTABA (Fig. 2c) in DMF/water (7/3, v/v) was allowed to cool to 25 °C with continuous rotary stirring, the absorption dissymmetry factor (gabs) of the resulting suspension, evaluated from the CD intensity relative to the absorbance at 316 nm, gradually increased with the stirring time and then leveled off in 40 min at 2.2 × 10−2 (Supplementary Fig. 19). We noticed that faster stirring resulted in the emergence of a larger CD intensity at 316 nm (Supplementary Fig. 20). Rotary stirring of the CD-silent gel did not induce any CD band (Supplementary Fig. 21), indicating that physical agitation such as rotary stirring does not affect the helical handedness of the PBTABA nanoribbons once formed. From these observations, we proposed a mechanism for the formation of the (P)- or (M)-dominant helical nanoribbons from BTABA as follows: BTABA self-assembles into (P)- and (M)-helical short 1D aggregates (primary nucleation). Such aggregates grow up to longer (P)- and (M)-helical nanoribbons, respectively. By physical agitation, the long helical nanoribbons preferentially break up into numerous short aggregates with the same handedness (secondary nucleation). In the resulting nonequilibrated state, once the mirror symmetry of the PBTABA nanoribbon is broken as a consequence of stochastic fluctuation, a small chiral bias thus generated can be amplified through repeated cycles of the secondary nucleation, as proposed for the mechanism of chiral amplification via thermal cycling or attrition in conglomerate crystal systems9,10,11. Accordingly, the addition of (P)-dominant PBTABA as a seed into a DMF/water (7/3, v/v; 1 mL) solution of BTABA (5.3 mM) before rotary stirring always resulted in the formation of a suspension of (P)-dominant PBTABA with a negative CD band at 316 nm (Supplementary Fig. 22).

Asymmetric catalysis

Previously, groups of Engberts56,57 and Roelfes25,27 reported highly enantioselective Cu2+-catalyzed Diels–Alder reactions between aza-chalcone and cyclopentadiene using amino acids and DNA as the chiral mediators, respectively. Inspired by these successful examples, we chose the same reaction to investigate a possibility of mirror symmetry-broken PBTABA nanoribbons as a chiral ligand. Typically, Cu2+-coordinated (−)-PBTABA or (+)-PBTABA ([Cu2+]/[BTABA] = 1.0%), employed as a catalyst for the Diels–Alder reaction (Fig. 3a), was prepared by mixing an aqueous solution of Cu(NO3)2 (4.1 mM, 13 µL) and a DMF/water (7/3, v/v; 1 mL) suspension of (−)-PBTABA or (+)-PBTABA ([BTABA] = 5.3 mM, Fig. 2d, e) at 25 °C for 1 h. The luminescence of PBTABA was quenched upon being mixed with Cu(NO3)2 (Supplementary Fig. 23), suggesting that PBTABA accommodates Cu2+ presumably at the peripheral carboxylate units. As observed by SEM and CD spectroscopy, PBTABA maintained its helical nanoribbon structure even after being mixed with Cu2+ (Supplementary Fig. 24). For the Diels–Alder reaction, an acetonitrile (40 µL) solution of aza-chalcone (1 mg) and freshly distilled cyclopentadiene (40 µL) were successively added to a DMF/water (7/3, v/v) suspension of (−)-PBTABA/Cu2+ or (+)-PBTABA/Cu2+, and the mixture was stirred at 25 °C. After 36 h, the reaction mixture was extracted with ethyl acetate (0.5 mL), and an organic extract, after being evaporated to dryness, was subjected to chiral high performance liquid chromatography (HPLC), where the endo isomer was revealed to be the major product (∼90%, conversion: >99%, turnover number: ~90, Fig. 3b, c). Of particular interest, (−)-PBTABA/Cu2+ (Fig. 2d) and (+)-PBTABA/Cu2+ (Fig. 2e) preferentially afforded the enantiomers of (S)-endo (Fig. 3b) and (R)-endo (Fig. 3c), respectively. We conducted a total of 76 runs of the Diels–Alder reaction using 76 different batches of (−)-PBTABA/Cu2+ and (+)-PBTABA/Cu2+ under the conditions described above and confirmed that this stereochemical correlation between the catalyst and product was perfectly reproduced (Fig. 3d and Supplementary Table 3). However, there is no clear correlation between the CD intensities of the catalyst and enantiomeric excess values of the product (Supplementary Fig. 25). We confirmed that the nanoribbons of PBTABA after the reaction were shorter than those before the reaction (Supplementary Fig. 26). However, the ee value of the endo isomer did not substantially change with the reaction time (Supplementary Fig. 27). In sharp contrast, repeated usage of the catalyst, isolated each time via centrifugation from the reaction mixture, resulted in lowering of the enantiomeric excess of the product (Supplementary Table 4). Meanwhile, CD-silent PBTABA/Cu2+ nanoribbons (Fig. 2b) also catalyzed the Diels–Alder reaction. However, as expected, the reaction was not enantioselective (Supplementary Fig. 28), as in the case where PBTABA was not employed for the Cu2+-catalyzed Diels–Alder reaction (Supplementary Fig. 29).

Fig. 3
figure 3

Enantioselective Diels–Alder reaction catalyzed by Cu2+-coordinated PBTABA nanoribbons. a Diels–Alder reaction between aza-chalcone and cyclopentadiene catalyzed by (−)-PBTABA/Cu2+ or (+)-PBTABA/Cu2+, which preferentially affords the endo isomer (~90%). b, c Chiral HPLC traces of the reaction mixtures in run No. 1 (b) and 34 (c) in (d). Symbols ee (S) and ee (R) represent enantiomeric excess (ee) values of the endo isomers with the dominancy of their (S)-endo and (R)-endo enantiomers, respectively. d Statistical ee values of the endo isomers obtained from a total of 76 runs of the Diels–Alder reaction catalyzed at 25 °C by 76 different batches of (–)-PBTABA/Cu2+ and (+)-PBTABA/Cu2+ at [Cu2+]/[BTABA] = 1.0%. Blue and pink-colored bars represent the ee values of the endo isomers obtained when (−)-PBTABA/Cu2+ and (+)-PBTABA/Cu2+, respectively, are used as the catalysts. e Effects of [Cu2+]/[BTABA] on the ee values of the endo isomers obtained at 25 °C with (−)-PBTABA/Cu2+ or (+)-PBTABA/Cu2+. f Effects of [NaOH]/[BTABA] on the ee values of the endo isomers obtained at 25 °C with (−)-PBTABA/Cu2+ or (+)-PBTABA/Cu2+ at [Cu2+]/[BTABA] = 1.0%, where NaOH was added to the suspensions of (−)-PBTABA or (+)-PBTABA prior to the addition of Cu(NO3)2. Error bars represent the standard deviation. Source data are provided as a Source Data file

Full size image

We attempted to optimize the conditions for the Diels–Alder reaction. First, we varied [Cu2+] versus [BTABA] and found that the ee value of the endo isomer reached a maximum at [Cu2+]/[BTABA] = 1.0% (Fig. 3e). When [Cu2+]/[BTABA] exceeded 1.0%, the ee value of the endo isomer dropped, most likely because of a possible increase in the amount of Cu2+ unanchored to the helical PBTABA nanoribbons. A strong base such as NaOH is expected to deprotonate the benzoic acid groups of PBTABA to promote the anchoring of Cu2+ onto the helical PBTABA nanoribbons36. In fact, as shown in Fig. 3f, the addition of NaOH to the reaction system resulted in a considerable enhancement of the enantioselectivity of the reaction, where an average ee value of >40% was realized at [NaOH]/[BTABA] = 50% with the highest ee value of 46% (major/minor enantiomers = 73/27; Supplementary Fig. 30).

Chirality transfer via electrostatic interaction

We explored the stereochemical environment of the cation-binding sites in the helical PBTABA nanoribbon by using a cationic fluorescent dye, methylene blue (MB, Fig. 4a). When an aqueous solution of MB (21.2 mM, 50 µL) was added to a DMF/water (7/3, v/v; 1 mL) suspension of mirror symmetry-broken (−)-PBTABA ([BTABA] = 5.3 mM, [MB]/[BTABA] = 20%), a broad positive CD band at 570–700 nm emerged (Fig. 4d and Supplementary Fig. 31). When (+)-PBTABA ([BTABA] = 5.3 mM) was employed instead of (−)-PBTA for mixing with MB, the sign of the CD band at 570–700 nm was negative (Fig. 4d and Supplementary Fig. 31). Considering that MB has an absorption band at 570–700 nm (Supplementary Fig. 31), the CD bands in this region are assignable to MB, which is supposedly located in a chiral environment by being bound to the helical nanoribbon of PBTABA at its carboxylate units (Fig. 4a)58,59. Note that mirror symmetry-broken (−)-PBTABA/MB and (+)-PBTABA/MB displayed positive and negative circularly polarized luminescence (CPL) bands at 600–800 nm, respectively (λext = 550 nm, Fig. 4e and Supplementary Fig. 32b). These CPL bands are obviously different from those observed for (−)-PBTABA and (+)-PBTABA (400–600 nm, λext = 290 nm; Fig. 4c and Supplementary Fig. 32a) and therefore are assignable to MB. The luminescence dissymmetry factors (glum) of (−)-PBTABA/MB and (+)-PBTABA/MB, evaluated from their CPL intensities relative to the fluorescence intensities at 700 nm, were ~10−2, which are much larger than those commonly observed for chiral organic compounds (10−5–10−2)60. This result again indicates unambiguously that MB is localized in a chiral environment provided by (−)-PBTABA or (+)-PBTABA.

Fig. 4
figure 4

Chiral environment of the cation-binding sites in PBTABA nanoribbons. a Schematic representation of the binding of methylene blue (MB) onto PBTABA at its carboxylate units via an electrostatic interaction. b, c Circular dichroism (CD, b) and circularly polarized luminescence (CPL, λext = 290 nm, c) spectra at 25 °C of DMF/water (7/3, v/v; 1 mL) suspensions of (−)-PBTABA (blue) and (+)-PBTABA (pink) ([BTABA] = 5.3 mM, Fig. 2d, e). d, e CD (d) and CPL (λext = 550 nm, e) spectra at 25 °C of DMF/water (7/3, v/v; 1 mL) suspensions of (−)-PBTABA (blue) and (+)-PBTABA (pink) ([BTABA] = 5.3 mM) containing MB ([MB]/[BTABA] = 20%). Source data are provided as a Source Data file

Full size image

Discussion

All of these results allow us to conclude that a mirror symmetry-broken helical supramolecular nanoribbon composed of an achiral monomer can serve as a chiral ligand for mediating enantioselective catalysis. Although the noncrystalline nature of PBTABA (Supplementary Figs. 17 and 18) does not permit elucidation of its molecular packing, the successful CPL observation from MB in Fig. 4e suggests that the catalytic site having a Cu2+ ion is chiral. As for the modest enantioselectivity of the reaction, we assume that this is due to the helical nanoribbon structure, which is formed by rolling up of a rather planer, sheet structure. It is likely that the catalytic sites located on such a gently curved surface cannot efficiently bias the stereochemical course of the reaction. We also noted that the mirror symmetry-broken helical nanoribbons of BTABA display intense electronic CD (Δε = ~570 L mol−1 cm−1). This is very rare, but some helicene nanoribbons are known to show such an amplified chiroptical behavior61. The next grand challenge to tackle is how one can obtain a helical catalyst with a preferred handedness by mirror symmetry breaking. Together with our achievement, the design of mirror symmetry breaking may revolutionize the manufacturing process of chiral compounds having many potential applications.

Methods

Materials

Unless otherwise noted, reagents and solvents were used as received from commercial sources without further purification. Benzene-1,3,5-tricarbonyl trichloride was purchased from Alfa Aesar. Ethyl 4-aminobenzoate was purchased from Adamas. Copper (II) nitrate (Cu(NO3)2) was purchased from J&K Scientific. Methylene blue was purchased from Tokyo Chemical Industry (TCI). Dicyclopentadiene was purchased from Acros Organics. Cyclopentadiene was prepared by distillation of commercial dicyclopentadiene at 180 °C. (E)-3-Phenyl-1-(pyridin-2-yl)prop-2-en-1-one (aza-chalcone) was prepared according to a previously reported method56.

Characterization

1H and 13C nuclear magnetic resonance (NMR) spectra were recorded on a JEOL model ALPHA-500 spectrometer in which the chemical shifts for 1H NMR spectroscopy were determined with respect to a non-deuterated solvent residue, CHD2(CD3)SO (δ 2.50 ppm), and those for 13C NMR spectroscopy were determined with respect to DMSO (δ 39.52 ppm). Matrix-assisted laser desorption/ionization time-of-flight mass (MALDI-TOF MS) spectrometry was performed using α-cyano-4-hydroxycinnamic acid (CHCA) as a matrix using a Bruker Daltonics AutoflexTM Speed MALDI-TOF/TOF spectrometer. Electronic absorption spectra were recorded in sandwich-type quartz cuvettes with an optical path length of 0.1 mm using a JASCO UV-550 spectrometer. CD and LD spectra were recorded in sandwich-type quartz cuvettes with an optical path length of 0.1 mm using a JASCO J-810 spectrometer. Circularly polarized luminescence (CPL) spectra were recorded in sandwich-type quartz cuvettes with an optical path length of 0.1 mm using a JASCO CPL-200 spectrometer. Fluorescence spectra were recorded using a Hitachi F-4500 spectrophotometer. FT-IR spectra were recorded using a JASCO model FT/IR-6100Plus FT-IR spectrometer. SEM was performed using a Hitachi S-4800 FE-SEM with an accelerating voltage of 10 kV. X-ray diffraction (XRD) analysis was performed on a Rigaku D/Max-2500 X-ray diffractometer (Japan) with Cu Kα radiation (λ = 1.5406 Å), which was operated at a voltage of 40 kV and a current of 200 mA. Polarized optical microscopy (POM) was performed using a Leica DM2700 upright materials microscope. Chiral HPLC was performed using a Waters 1525 HPLC system with a Daicel Chiralcel® OD-H column.