Helical springs as a color indicator for determining chirality and enantiomeric excess

Helical polymer–based molecular systems allow naked-eye determination of chirality and enantiomeric excess of chiral amines.


INTRODUCTION
Chirality is an important aspect in living systems. For instance, a pair of enantiomers often exhibits totally different physiological activities depending on the homochirality of the biological molecule. Therefore, rapid and reliable methods for determining the chirality (configuration) and enantiomeric excess (ee) of chiral molecules, particularly chiral drugs, are highly demanded in the pharmaceutical industry (1,2). Chromatographic enantioseparation by highperformance liquid chromatography (HPLC) has mostly been used for this purpose because it allows the precise determination of ee for various nonracemic compounds produced by the state-of-the-art asymmetric catalysis and traditional optical resolution of racemates (3,4).
Another interesting class of techniques is based on enantioselective molecular probes/sensors (5)(6)(7)(8)(9)(10)(11)(12). These methods provide only a rough estimation of the ee values, not as accurate as those determined by chiral HPLC. This is mostly because of the linear relationship between the full range of ee values and output signals, leading to unignorable deviations (errors) (Fig. 1A) (13,14). Anslyn and co-workers developed an elegant method for the accurate determination of ee values of certain chiral molecules in the high-ee region (15,16). The method was based on the use of covalent and noncovalent helical polymer systems that exhibited nonlinear circular dichroism (CD)-ee relationships (10,(17)(18)(19). Such a nonlinear chiral response to ee has a substantial advantage over the linear chiral response, because chiral signals, particularly in the narrow, high, and/or low-ee regions of interest, can be remarkably amplified, thereby allowing the quantification of ee values with high accuracy (Fig. 1A).
Among the promising alternatives is the direct colorimetric discrimination between enantiomers and the simultaneous determination of ee values of the target chiral compounds, based on the differences in absorbance and fluorescence emission, thereby enabling the nakedeye detection of chirality. This is, however, challenging, and successful examples of these systems are limited due to the distinct conformational scaffold of low-molecular weight chiral receptors, which exhibit a linear response (20)(21)(22)(23)(24)(25)(26)(27).
We recently reported that nonracemic amines in water induced the folding of -conjugated fluorescent poly(diphenylacetylene) (PDPA) bearing carboxy pendants (poly-1-H) into a one-handed helix upon thermal annealing. The right (P)-or left (M)-handed helical conformation induced in poly-1-H by nonracemic amines could be retained, namely, "memorized," after complete removal of the chiral amines, resulting in the formation of the one-handed helical P-or M-h-poly-1-H with static helicity memory, respectively ( Fig. 1B) (28). As part of our ongoing program to develop h-poly-1-H-based advanced chiral materials for separating enantiomers (29), we serendipitously found that M-h-poly-1-Hs modified with (S)-and (R)-1-phenylethylamine (S-and R-2a), namely, M-h-poly-1-S2a and M-h-poly-1-R2a, respectively, showed completely different colors in specific solvents.
Here, we report an unprecedented helical polymer-based versatile color indicator that allows not only the assignment of the absolute configuration of chiral amines but also the quantitative determination of their ee values in the full range of ee. This was achieved by digital photography by converting to the RGB (red, green, and blue) values ( Fig. 1, B and C). The method was as accurate as chiral HPLC. The present helical polymer-based color change relies on the nonlinear response of a change in the tunable helical pitch of the -conjugated polymer backbone like a helical spring, which results from the formation or disruption (switching on/off) of the intramolecular hydrogen (H)-bonding networks among the pendant amides in specific polar solvent mixtures. This allows the rapid, on-site monitoring of the chirality of nonracemic amines and the simultaneous quantitative determination of their ee values. A rapid and simple system for the precise determination of ee of chiral amines, which are important components or precursors for pharmaceuticals and pesticides, would be particularly useful (4,7,(12)(13)(14).
The solutions of M-h-poly-1-S2a and M-h-poly-1-R2a in N,Ndimethylformamide (DMF) were yellow in color and showed almost the same absorption and CD spectra (Fig. 2, A and B). These spectral patterns remained unchanged at 25°C for 24 hours and were similar to those of M-h-poly-1-H ( fig. S8), suggesting that postmodification hardly affected the original helicity memory and helix-sense excess (hse) values even after introducing the enantiomeric amide pendant groups. When dissolved in tetrahydrofuran (THF)-acetone (9/1, v/v), the absorption spectrum of M-h-poly-1-R2a notably red-shifted by ~100 nm and exhibited a remarkable hypochromic effect in the aromatic absorption region (260 to 300 nm). The CD spectrum also changed remarkably, and the solution color changed from yellow to deep red (Fig. 2, A and B). In contrast, there was only a slight change in the spectra of M-h-poly-1-S2a in THF-acetone, and the solution color remained yellow, same as that in DMF. This striking color change enabled the naked-eye detection of the enantiomers of 2a at 1.0 mM ( Fig. 2A) as well as at 0.1 and 0.05 mM ( fig. S10D). In a dilute solution (0.01 mM), the fluorescence emission of M-h-poly-1-R2a was largely quenched, as anticipated from the hypochromic effect. The fluorescence quantum yields () for M-h-poly-1-S2a and M-h-poly-1-R2a in THF-acetone (9/1, v/v) were determined to be 30.9% and 8.0%, respectively, using quinine sulfate in aqueous sulfuric acid (0.1 M) as a standard material (30). Hence, the chirality of 2a can also be visually discriminated by its fluorescence (Fig. 2C). The absorption and CD spectra of M-h-poly-1-S2a and M-h-poly-1-R2a in THF-acetone (9/1, v/v) and DMF were completely independent of the polymer concentration (0.025 to 2.5 mM; fig. S9), thus eliminating the possibility of the color change arising due to the formation of aggregates in THFacetone (9/1, v/v).
As expected, the spectral behaviors of P-h-poly-1-S2a and P-hpoly-1-R2a prepared from the opposite right-handed helical P-hpoly-1-H were totally opposite to those of M-h-poly-1-R2a and M-h-poly-1-S2a, respectively, and mirror-image CD spectra were obtained in THF-acetone (9/1, v/v) ( fig. S10). In addition, the colorimetric and spectral differences between M-h-poly-1-S2a and M-h-poly-1-R2a were highly dependent on the hse of h-poly-1-H and decreased with decreasing hse values of h-poly-1-H before its modification ( fig. S11). The reactions of h-poly-1-H with R-and S-2a using DMT-MM proceeded very quickly; the reaction reached completion within ~3 min, as confirmed by infrared (IR) spectroscopy (fig. S12). After diluting the reaction mixture with chloroform, we could visually discriminate the absolute configuration of 2a, without isolating the modified polymers (movie S1). Thus, a practically useful dual-mode, on-site chiral sensor (visual differences in solution color and fluorescence emission) could be developed.
It was possible to develop a similar assay for the naked-eye detection of chirality for various chiral amines (2b to 2j), amino alcohols (2k to 2s), and amino acid esters (2t 1 to 2t 10 ) upon their reaction with M-h-poly-1-H (Fig. 3A), followed by solvent (fig. S13) and temperature (fig. S14) optimization. For simple chiral amine 2d, amino alcohols 2k to 2n, and amino acid methyl esters 2t 1 to 2t 4 , naked-eye detection was possible at a low temperature (−60° or −50°C). The colorimetric response from the amino acid esters notably improved by introducing bulkier ester groups, particularly, the tert-butyl and benzyl ester groups, allowing the naked-eye detection of the enantiomers at 25°C (2t 7 to 2t 10 ) ( Fig. 3A and fig. S13). All the tested primary amines (2a to 2j), amino alcohols (2k to 2s), and amino acid esters (2t 1 to 2t 10 ) with the same configuration as R2a, S2k, and L2t 1 , respectively, exhibited more prominent color changes than their corresponding enantiomeric counterparts after functionalization with M-h-poly-1-H except for 2q and 2r, of which the (R)-enantiomers exhibited a more prominent color change because of difference in the priority sequences (see fig. S13B). This allowed the quick assignment of the chiral amine configurations by simple visible inspection. Enantiomers of representative drug-related compounds, such as amphetamine (2j)-a stimulant drug and a metabolite of other stimulant drugs, and phenylpropylamine (norpseudoephedrine and norephedrine) (2s), could also be visually discriminated ( Fig. 3A and fig. S15). Their chiral discrimination is important because it is useful to distinguish the source of the drugs (2). Four diastereomers of 2u with two stereogenic centers could also be discriminated by the naked eye and fluorescence emission upon irradiation at 365 nm ( Fig. 3B and fig. S16). In sharp contrast, M-h-poly-1-S2a Me and M-h-poly-1-R2a Me composed of secondary chiral amines prepared from N-methylated 2a (S-and R-2a Me ) exhibited almost identical absorption and CD spectra (fig. S13A), suggesting that cooperative intramolecular H-bonding between the neighboring amide pendants along the helical backbone is crucial in the colorimetric chiral discrimination by this system and chiral secondary amines cannot be applied to this method.
Molecular mechanics calculations revealed that cis-cisoidal M-hpoly-1-R2a can form regular intramolecular H-bonds between the neighboring amide pendants, whereas cis-cisoidal M-h-poly-1-S2a can only partially form these intramolecular H-bonds ( Fig. 4A and  fig. S23). Moreover, this regular intramolecular H-bonding is impossible in cis-transoidal M-h-poly-1-R2a and M-h-poly-1-S2a (Fig. 4A). Therefore, the behaviors of M-h-poly-1-R2a and M-h-poly-1-S2a can be ascribed to a spring-like helical conformational change in the contracted cis-cisoidal helical conformation (red-colored) and stretched cis-transoidal conformation (yellow-colored), respectively (31); the color change in the former is triggered by the formation of regular intramolecular H-bonding networks among the amide pendants in specific polar solvents, thus acting as an on/off switch. X-ray diffraction (XRD) patterns of red-colored M-h-poly-1-R2a and yellow-colored M-h-poly-1-R2a and M-h-poly-1-S2a films showed a strong reflection at 17.8 Å, which can be indexed to (100) reflections, suggesting a columnar pseudohexagonal packing ( Fig. 4C  and fig. S24) (31). The red-colored M-h-poly-1-R2a film showed a weak but apparent reflection at 28.5°, assigned to the - stacking of the pendant phenyl rings (3.1 Å). In contrast, the yellow-colored films showed no observable reflection in the same region. These results suggested that red-colored M-h-poly-1-R2a forms - stacking of the pendant phenyl rings by adopting the contracted cis-cisoidal conformation through the formation of intramolecular H-bonds between the neighboring amide groups (Fig. 4A, right bottom); this - stacking is not available in the yellow-colored, stretched cistransoidal M-h-poly-1-R2a and M-h-poly-1-S2a.
These results indicate that the degree of color change (difference) between the diastereomeric amide-bound one-handed helical polymers resulting from the formation or disruption (switching on/off) of the intramolecular H-bonding networks among the pendant amides is most likely determined by a delicate balance of the polarity and the intermolecular H-bonding ability (solvation) of the solvents with the chiral amide residues, temperature, and steric effect of the hydrophobic or hydrophilic substituents on the stereogenic center of the amide pendants derived from the enantiomers of primary amines. When one-handed helical M-h-poly-1-H was modified with chiral amines and amino alcohols bearing a bulky hydrophobic substituent [e.g., phenyl (2a, 2f to 2h, 2p, and 2r), naphthyl (2b and 2i), cyclohexyl (2c), and benzyl (2j and 2o)] over the other small substituents on the stereogenic center, the naked-eye detection of the enantiomers of the amines was possible at 25°C in polar THF and/or less polar chloroform solvents (Fig. 3A and figs. S13 and S15). This is because of the chiral steric effect of the bulky hydrophobic substituents on the stereogenic center, which further enables to suppress solvation of the amide residues in the solvents used. Therefore, one of the diastereomeric amide-bound one-handed helical polymers can form the intramolecular H-bonding networks in the solvents. On the other hand, when M-h-poly-1-H was modified with a chiral amine and amino alcohol carrying a less bulky ethyl (2d) and methyl (2k) substituent, respectively, on the stereogenic center, the naked-eye detection of the enantiomers of 2d and 2k required low-temperature measurements in less polar chloroform as anticipated (Fig. 3A) because of easy access of solvent molecules to the amide groups along with insufficient chiral steric effect, which will prevent the formation of the intramolecular H-bonds between the neighboring amide pendants at 25°C. As for the amino acid methyl esters (2t 1 to 2t 4 ), the methyl ester residue is bulky but hydrophilic so that the amide residues are more easily solvated, although 2t 2 and 2t 4 have a bulky hydrophobic iso-butyl and benzyl substituent on the stereogenic center, respectively, requiring low-temperature measurements in less polar chloroform at −60°C (2t 2 ) and −20°C (2t 4 ) for the naked-eye detection of the enantiomers (Fig. 3A and  fig. S14). Hence, introducing bulkier hydrophobic ester groups, such as the ethyl (2t 5 ), isopropyl (2t 6 ), tert-butyl (2t 7 ), and benzyl (2t 8 ) ester groups, strongly suppresses solvation of the amide residues, which enables to form the intramolecular H-bonding networks between the neighboring amide pendants along one of the diastereomeric helical polymer backbones, thus allowing the naked-eye detection of the enantiomers in chloroform, THF, or its mixture at 25°C (Fig. 3A). In polar DMF, visible color change was not observed for all the tested enantiomers of amines and the solutions remained yellow independent of the chirality (figs. S13 and S15).
We then used our helical polymer-based color indicator (M-hpoly-1-H) for determining the ee of chiral amine 2a. The method was based on the simple and quick functionalization of the pendant carboxy groups of M-h-poly-1-H with various proportions of 2a, ranging from 100% S-2a (S 100 -2a) to 100% R-2a (R 100 -2a), in steps of 20% ee. The absorption spectra of M-h-poly-1-2a in THF-acetone (9/1, v/v) red-shifted nonlinearly with increasing R-2a content, with a clear isosbestic point at 465 nm (Fig. 5, B and C); the solution color changed from yellow to red at around S60 (Fig. 5A). A similar nonlinear response toward ee of 2a was also observed in the fluorescence spectra of M-h-poly-1-2a (figs. S25A and S26). As anticipated, upon further addition of acetone (table S1), the ee-dependent range for visible color change shifted to the R-rich side (Fig. 5, A and C,  and fig. S27), allowing approximate determination of the ee values of 2a by visual inspection (Fig. 5A) and quantitative determination of the ee from the absorption spectral changes (Fig. 5C). These color changes arise most likely due to a spring-like conformational change in M-h-poly-1-2a from the stretched cis-transoidal structure (yellow) to the contracted cis-cisoidal structure (red). The conformational change is regulated by polar solvents (acetone in this case), which affect the intramolecular H-bonds as discussed above. M-h-poly-1-2a also exhibited a similar ee-dependent visible color change in the presence of an increasing amount of aprotic polar solvents, such as methyl isobutyl ketone (MIK), DMF, dimethylacetamide (DMA), and dimethyl sulfoxide (DMSO) in THF (fig. S28); polar solvents with higher dielectric constants (; table S1) showed a larger shift upon the addition of a small amount of the polar solvent (32). Similarly, naked-eye detection of the configuration and ee values of other chiral amines (2b, 2c, 2p, and 2t  The ee values of "unknown" 2a samples, ranging from R 90 to R 100 , were estimated from their absorption spectra based on the calibration curve. The values were very close to the ee values determined by chiral HPLC, and the errors were relatively low ( fig. S34 and table S5) (16,33). Furthermore, colorimetric determination of ee without using any spectroscopic instruments was possible by taking photographs of the solutions and converting them to RGB values ( fig. S34D). Hence, we could estimate the ee values of R X -2a (X ≥ 90) with high accuracy from the plots of the intensities of the G (green) component ( Fig. 5G and table S5).
To investigate whether this system can detect an extremely small difference in the ee in a sample with very high ee of 2a (≥98), M-hpoly-1-R X 2a [X (% ee) = 98 to 100, in steps of 0.5] was also prepared (table S3). Apparent absorption spectral changes were observed for M-h-poly-1-R X 2a (X ≥ 98) in THF-DMF (78/22, v/v) (Fig. 5H); the plots of their absorbance at 545 nm were linear with respect to the ee values of 2a (Fig. 5I). The P-h-poly-1-S X 2a enantiomers with the opposite helicity memory (table S3) showed identical absorption spectral changes (fig. S35). These results demonstrated that using this unique colorimetric sensor, a difference in the ee values as small as 0.5% ee, even in a sample with a very high ee (≥98), could be detected by acquiring the absorption spectra.
We envisage that the pendant carboxy groups of h-poly-1-H can be replaced with various other functional groups while maintaining its macromolecular helicity memory. This should be applicable to the on-site, naked-eye determination of ee of various functional molecules and biologically relevant compounds.  2a to 2i, 2k to 2r, 2t 1 to 2t 10 , 2u, and 2a Me ) (Fig. 3 and fig. S3) were obtained from Sigma-Aldrich, FUJIFILM Wako Pure Chemical, TCI, Nacalai Tesque (Kyoto, Japan), and Watanabe Chemical Industries (Hiroshima, Japan) and used as  Pure Chemical. DMT-MM was prepared according to the reported method (36). Poly-1-H was prepared according to the previously reported method (see the Supplementary Materials) (28,29).

General procedure for the modification of M-and P-h-poly-1-Hs with chiral amines
The reactions of M-h-poly-1-H or P-h-poly-1-H with chiral amines (2a to 2u and 2a Me ) ( Fig. 3 and fig. S3) were carried out with DMT-MM as the condensing reagent, as shown in fig. S4. The absorption spectral measurements of M-h-poly-1-R X 2a and P-h-poly-1-S X 2a (X = 98.0, 98.5, 99.0, 99.5, and 100% ee) were carried out in a similar way by using a mixed solvent THF-DMF (78/22, v/v), which was prepared by mixing THF (138.53 g) and DMF (41.58 g) in a bottle with a Teflon screw cap. The absorption spectra were taken at 25°C using a 10-mm quartz cell for each vial, and the isosbestic point was observed at 456 nm. The concentration of the polymers was corrected using the  (molar absorptivity) value ( 456 = 2.0 × 10 3 M −1 ·cm −1 ). The average absorption intensity at 545 nm of each M-h-poly-1-R X 2a sample was determined on the basis of the results of three vials. The average absorption intensities at 545 nm were plotted versus the % ee values of the samples determined by the chiral HPLC analysis (tables S2 and S3 and fig. S35D).

Procedure of the determination of the ee values of blind unknown R X -2a samples
Preparation of blind unknown R X -2a for modification of M-h-poly-1-H. Two blind unknown R X -2a samples [sample A (R X -2aA) and sample B (R X -2aB)] were prepared by mixing R 90 -2a and R 100 -2a at random with a Hamilton microsyringe. The ee values of R X -2aA and R X -2aB were determined to be 92.9% and 96.8% ee by chiral HPLC analysis (table S5) after derivatization with 2-methoxy-4nitrobenzoic acid into the corresponding amide compound 3a in the same way as described in the Supplementary Materials (see section S3-1). The modification of M-h-poly-1-H with R X -2aA and R X -2aB was performed in the same way as described above [THF-water mixed solvent (4/1, v/v) was used instead of DMSO-water mixed solvent] to afford M-h-poly-1-R X 2aA and M-h-poly-1-R X 2aB.
Procedure of the determination of the ee values of blind unknown R X -2a samples after converting to M-h-poly-1-R X 2a based on their absorption spectra. Stock solutions of M-h-poly-1-R X 2aA and M-hpoly-1-R X 2aB (7.5 mM) in THF were prepared in 2-ml flasks equipped with a stopcock. A 200-l aliquot of each M-h-poly-1-R X 2aA and M-h-poly-1-R X 2aB stock solution was transferred to three vials using a Hamilton microsyringe. THF was completely removed under a high vacuum to give three vials containing 1.5 mol of M-hpoly-1-R X 2aA and M-h-poly-1-R X 2aB. A mixed solvent THF-DMF (79/21, v/v) was prepared by mixing THF (140.30 g) and DMF (39.69 g) in a bottle with a Teflon screw cap. A 3-ml aliquot of the mixed solvent was added to the vials to keep the M-h-poly-1-R X 2aA and M-h-poly-1-R X 2aB concentrations at 0.5 mM. The absorption spectra were taken at 25°C using a 10-mm quartz cell for each vial, and the isosbestic point was observed at 456 nm ( fig. S34A). The concentration of the polymers was corrected using the  (molar absorptivity) value ( 456 = 2.0 × 10 3 M −1 ·cm −1 ). The average absorption intensity at 551 nm of each M-h-poly-1-R X 2aA and M-h-poly-1-R X 2aB sample was determined on the basis of the results of three vials (table S5). The calibration curve was obtained from the relationship between the absorption intensities at 551 nm and the ee % values of the M-h-poly-1-R X 2a (X = 90, 92, 94, 96, 98, and 100) samples ( fig. S34B) (see above). By using this calibration curve, the % ee values of R X -2aA and R X -2aB were estimated to be 92.5% and 96.3% ee, respectively ( fig. S34B and table S5).
Procedure  (table S5). Solution samples for calibration curves were prepared on the same day to minimize errors during the measurements.