Enantiomeric Variability of Distaminolyne A. Refinement of ECD and NMR Methods for Determining Optical Purity of 1-Amino-2-Alkanols.

Sample configurations of distaminolyne A (1a); isolated from the ascidians Pseudodistoma opacum and P. cereum, and collected at different sites in New Zealand, were investigated by two methods: Exciton coupled electronic circular dichroism (EC ECD) of the corresponding N,O-dibenzoyl derivative 1b; and chiral reagent derivatization of 1a with (S)- and (R)-α-methoxyphenylacetic acid (MPA), followed by ¹H-NMR analysis. Configuration and optical purity of 1a (%ee) was found to vary depending on the geographic distribution of ascidian colonies. An improved method for preparing N,O-diarenoyl derivatives of 1a was optimized. The EC ECD method was found to be complementary to the MPA-NMR method at different ranges of %ee.

Unlike sphingosine, AA configurations are heterogeneous: The relative configuration (RC) and absolute configuration (AC) of 2-AAs vary from species to species. Configuration assignment to AAs is deceptively simple, and confounded by weak specific rotations that have been misinterpreted in the past leading to erroneous assignments [4], or no assignment, which defer the problem to total synthesis [5,6]. A single chirogenic center in 1-AAs is responsible for [α] D with magnitudes within a narrow range of~±1-3; for example, in S-distaminolyne A (1a, [α] D −1), a C 17 AA from the New Zealand ascidian Pseudodistoma opacum [7]. The configuration of 1a was assigned from interpretation of exciton coupling observed in the electronic circular dichroism ECD spectrum of the corresponding N,O-dibenzoyl derivative 1b [7] according to an extension of the exciton coupling circular dichroism (EC ECD) dibenzoate method pioneered by Nakanishi and Harada [8]. Prior to this report, the EC ECD method was applied successfully to 2b, the N,O-dibenzoyl derivative of an unnamed 1-AA natural product; 2a, from a didemnid ascidian collected on the Great Barrier Reef [9].
The configuration of (S)-1a was recently challenged by Sun et al. and 're-assigned' as R based on a total synthesis of (S)-and (R)-1a, and the comparison of specific rotations-albeit of low magnitudes-with the published value of the natural product [10]. Paradoxically, a subsequent synthesis of 1a by Dumpala et al. using different methodology, upheld the original 2S stereoassignment [11]. Notably, the validity of the EC ECD method used for the original assignments of 1b [7] or 2b [9] was not disputed-in fact, not even addressed-in either report, leaving a dilemma: which assignment is correct?
Upon revisiting the original assignment, we found the assignment of AC of (S)-1a was upheld by careful re-examination of the ECD data, and new comparisons to model 1-AAs, (S)-3a, and (R)-3b prepared by rational asymmetric synthesis, and converted to dibenzoyl derivatives 4a,b [12]: The ECD spectrum of (S)-1b was essentially the same as that of (S)-4a. An elementary lesson in chiroptical analysis, reprised here, is that reassignment of natural products based solely on comparisons of [α] D of weakly rotatory compounds is tenuous, at best, and unsupportable in light of stronger counter-evidence; for example, assignments based on the non-empirical interpretation of EC ECD [8].
In this report, we investigate chiroptical properties of additional acylates of 1-AAs, the curious chain-length dependence of diacylation reactions under standard conditions, and improved methods for preparing 1b and related acylates. We also found configurational variability of 1a isolated from samples of P. opacum and P. cereum collected at different sites from the North Island of New Zealand, and outline complementary methods for measuring AC and enantiomeric excess (%ee) of 1a by ECD of bis-arenoyl derivatives, and 1 H-NMR after conversion with a chiral derivatizing agent (CDA).

Results
Unlike acylation of 2-AAs, benzoylation of 1-AAs with BzCl-pyridine has delivered variable success, giving the desired dibenzoyl derivatives in low yields. For example, benzoylation of 1a (Figure 1b; Method A: BzCl, 4 equiv, DMAP, pyridine, 50 • C, 64 h), gave 1b in only 8%, but a larger amount (16%) of the mono-acylated N-benzamide, even with excess equivalencies of reagents [7]. Similar yields were obtained for model compounds (S)-3a and (R)-3b, giving (S)-4a (15%) and (R)-4b (25%; [12]). From earlier work, alternative conditions for benzoylation (N-benzoylimidazole, DBU, CH 3 CN, 60 • C [13]) applied to (2S,3R)-2-amin ododecan-3-ol, from Clavelina oblonga [14] or sphingolipids [15], were also less than satisfactory. Although it is certain that benzoylation of the amino group occurs as the first step, it was not clear why benzoylation of the secondary OH of the benzamide intermediate was so sluggish. One possibility is formation of micelles in the dipolar solvents pyridine, and CH 3 CN that retard the second benzoylation step (see below).  [7] according to an extension of the exciton coupling circular dichroism (EC ECD) dibenzoate method pioneered by Nakanishi and Harada [8]. Prior to this report, the EC ECD method was applied successfully to 2b, the N,O-dibenzoyl derivative of an unnamed 1-AA natural product; 2a, from a didemnid ascidian collected on the Great Barrier Reef [9]. The configuration of (S)-1a was recently challenged by Sun et al. and 're-assigned' as R based on a total synthesis of (S)-and (R)-1a, and the comparison of specific rotations-albeit of low magnitudes-with the published value of the natural product [10]. Paradoxically, a subsequent synthesis of 1a by Dumpala et al. using different methodology, upheld the original 2S stereoassignment [11]. Notably, the validity of the EC ECD method used for the original assignments of 1b [7] or 2b [9] was not disputed-in fact, not even addressed-in either report, leaving a dilemma: which assignment is correct?
Upon revisiting the original assignment, we found the assignment of AC of (S)-1a was upheld by careful re-examination of the ECD data, and new comparisons to model 1-AAs, (S)-3a, and (R)-3b prepared by rational asymmetric synthesis, and converted to dibenzoyl derivatives 4a,b [12]: The ECD spectrum of (S)-1b was essentially the same as that of (S)-4a. An elementary lesson in chiroptical analysis, reprised here, is that reassignment of natural products based solely on comparisons of [α]D of weakly rotatory compounds is tenuous, at best, and unsupportable in light of stronger counterevidence; for example, assignments based on the non-empirical interpretation of EC ECD [8].
In this report, we investigate chiroptical properties of additional acylates of 1-AAs, the curious chain-length dependence of diacylation reactions under standard conditions, and improved methods for preparing 1b and related acylates. We also found configurational variability of 1a isolated from samples of P. opacum and P. cereum collected at different sites from the North Island of New Zealand, and outline complementary methods for measuring AC and enantiomeric excess (%ee) of 1a by ECD of bis-arenoyl derivatives, and 1 H-NMR after conversion with a chiral derivatizing agent (CDA).

Acylation of Model Compounds
In order to investigate the chain-length dependence and byproduct, we explored the acylation of homologous 1-AAs: The C 10 compounds (S)-3a and (R)-3b (Figure 2), and the C 6 and C 4 homologs (R)-5b and (S)-6a, respectively. Naphthoylation of (R)-4b (Method A) gave mono N-naphthamide 7b (Figure 2b) as the major product (18%), and smaller amounts of oxazoline 8 [0.4%, [α] D +2.6 (c 0.5, CHCl 3 )], and the di-N,O-naphthoyl derivative (R)-9b (0.5%). Cyclodehydration of AA benzamides is known to occur with inversion of configuration at C-2 under certain conditions that activate the secondary OH group (e.g., Brønsted acid, SOCl 2 ) for intramolecular nucleophilic S N 2 displacement by the carboxamide oxygen [16,17]. Clearly, if a similar inversion was occurring at C-2 during formation of 1b under elevated temperature and prolonged reaction times of Method A, it would undermine the validity of the ECCD assignment method for all 1-AAs. Likewise, if oxazolines were the intermediates on the pathway to 4a,b, these model compounds would be invalid. In order to test this hypothesis, benzoylation of (R)-3b was carried out under the milder conditions (Method B; see below). Gratifyingly, the product (R)-4b showed an essentially identical specific rotation ([α] 23 D −24.1 (c 1.48, MeOH)) to that of (R)-4b produced by Method A ([α] 23 D −26.1 (c 1.78, MeOH); [12]) eliminating the possibility of inversion at C-2 under the latter conditions.

Acylation of Model Compounds
In order to investigate the chain-length dependence and byproduct, we explored the acylation of homologous 1-AAs: The C10 compounds (S)-3a and (R)-3b (Figure 2), and the C6 and C4 homologs (R)-5b and (S)-6a, respectively. Naphthoylation of (R)-4b (Method A) gave mono N-naphthamide 7b (Figure 2b) as the major product (18%), and smaller amounts of oxazoline 8 [0.4%, [α]D +2.6 (c 0.5, CHCl3)], and the di-N,O-naphthoyl derivative (R)-9b (0.5%). Cyclodehydration of AA benzamides is known to occur with inversion of configuration at C-2 under certain conditions that activate the secondary OH group (e.g., Brønsted acid, SOCl2) for intramolecular nucleophilic SN2 displacement by the carboxamide oxygen [16,17]. Clearly, if a similar inversion was occurring at C-2 during formation of 1b under elevated temperature and prolonged reaction times of Method A, it would undermine the validity of the ECCD assignment method for all 1-AAs. Likewise, if oxazolines were the intermediates on the pathway to 4a,b, these model compounds would be invalid. In order to test this hypothesis, benzoylation of (R)-3b was carried out under the milder conditions (Method B; see below). Gratifyingly, the product (R)-4b showed an essentially identical specific rotation ([α]  Benzoylation of the shortest C4-chain 1-AA (S)-6a by Method A gave no dibenzoyl compound, but instead the N,N,O-tribenzoyl derivative (S)-10b as the only identifiable product (5%). Imide (S)-10b was deemed to be unsuitable for ECCD assignments due to the difficulty of establishing the directionality of the charge-transfer electronic transition dipole vectors of the three chromophores.
Method B was also adapted to sub-µmole acylation of 1-AAs with modification of Nakanishi's approach employing melting point capillaries as reaction vessels [18,19]. Treatment of 5b (0.35 µmol; Method C) with a standard solution prepared from the reagents in DCE (2.8 equiv.; See Experimental) and heating (45 °C or 67 °C, 60 min; Method C) gave, after recovery and purification by preparative Benzoylation of the shortest C4-chain 1-AA (S)-6a by Method A gave no dibenzoyl compound, but instead the N,N,O-tribenzoyl derivative (S)-10b as the only identifiable product (5%). Imide (S)-10b was deemed to be unsuitable for ECCD assignments due to the difficulty of establishing the directionality of the charge-transfer electronic transition dipole vectors of the three chromophores.
Method B was also adapted to sub-µmole acylation of 1-AAs with modification of Nakanishi's approach employing melting point capillaries as reaction vessels [18,19]. Treatment of 5b (0.35 µmol; Method C) with a standard solution prepared from the reagents in DCE (2.8 equiv.; See Experimental) and heating (45 • C or 67 • C, 60 min; Method C) gave, after recovery and purification by preparative TLC, (R)-12b in yields of 31% and 40%, respectively (Calculated from measured absorbances at λ max 231 nm (TFE)).

Enantiomeric Purity of 1a-Geographic Variation
We examined the optical purity of 1a from various samples of P. opacum and one of P. cereum, collected respectively from Whangateau Harbor and Princes Island, North Island of New Zealand, by conversion to the N,O-dibenzoate 1b, and measurement of their ECD spectra and calculation of %ee by comparison with CE data observed for model compound 4b (Table 1). To our surprise, a recollection of P. opacum from the same location as reported previously [7] (Entry 4), which revealed that unlike the original collection, the newer sample of 1a was nearly racemic (Table 1, Entry 5; Figure  4a). It was noted that this new collection of P. opacum contained a mixture of two ascidian color- The ECD spectra of the (R)-N,O-bis-(2-naphthoyl) derivatives of C 6 and C 12 1-AAs displayed ECD spectra of the same form (negative split CE), with only slight differences in magnitudes (Figure 3a,b). Compared to the longer C 10 chain compound (R)-13b (A = 172, A is defined as the difference of ∆ε measured from trough to crest in the split-CE [8], MeOH [12], Figure 3b), the C 6 homolog (R)-9b was about 30% less intense when measured in 2,2,2-trifluoroethanol (TFE, A = 119; Figure 3a), but essentially within the expected range.

Enantiomeric Purity of 1a-Geographic Variation
We examined the optical purity of 1a from various samples of P. opacum and one of P. cereum, collected respectively from Whangateau Harbor and Princes Island, North Island of New Zealand, by conversion to the N,O-dibenzoate 1b, and measurement of their ECD spectra and calculation of %ee by comparison with CE data observed for model compound 4b (Table 1). To our surprise, a re-collection of P. opacum from the same location as reported previously [7] (Entry 4), which revealed that unlike the original collection, the newer sample of 1a was nearly racemic (Table 1, Entry 5; Figure 4a). It was noted that this new collection of P. opacum contained a mixture of two ascidian color-morphs; white and beige, which could possibly have some bearing on the configurational hetereogeneity of 1a.
Consequently, another collection of ascidian material was undertaken, taking care to separate the two color-morphs. Analysis of 1a from these two color-morphs (entries 1 and 2; Figure 4b) revealed that both samples were also nearly racemic with %ee within experimental error. It appeared that none of the collections of P. opacum from that same location in the same season contained 1a of high optical purity. Similar ECD analysis of the dibenzoyl derivative of 1a isolated from specimens of P. cereum collected at Princes Island (Entry 3) identified the natural product as a scalemic mixture (~49%ee) of predominantly (S)-configuration (Figure 4b).  5 Average of (S)-and (R)-MPA %ee values. 6 This work. 7 From P. cereum. 8 2a, from an unidentified didemnid ascidian collected on the Great Barrier Reef, was converted to 2b using a variant of Method A [9]. 9 Optical purity from (S)-1,2-epoxydecane (>97%ee), prepared from the Jacobsen's hydrolytic kinetic resolution (HKR) reaction, and used in the synthesis of (S)-4a.   (Table 1).
In order to avoid possible introduction of bias to the ratio of diastereomeric products through fractionation during column chromatography, the crude CDA derivatization products were evaluated directly by 1 (Table 1).
In order to avoid possible introduction of bias to the ratio of diastereomeric products through fractionation during column chromatography, the crude CDA derivatization products were evaluated directly by 1 Figure S21, see Supplementary Material), and application of the configurational model [20] confirmed the AC (Table 1; Entry 7). In the case of 1a sourced from P. cereum, 1 H-NMR spectra of both MPA-derivatized samples 1c and 1d displayed the presence of all four CαH resonances, showing that 1a ( Figure S22) was a mixture of enantiomers favoring an excess of S (42%ee) when quantified using 1c, or 26%ee for 1d (average of 34%ee). The optical purities of 1a samples obtained from other collections of P. opacum were evaluated in a similar manner (Entries 1, 2 and 5); in each case very low %ee values were observed (data not shown).

Variable Acylation Efficiency of Long-Chain 1-AAs
The unexpected chain-length dependence of benzoylations or naphthoylations of 1-AAs under conventional conditions (e.g., Method A) suggests subtle differences in solution phase that amplify to larger differences in rates of reaction for the second reaction step: The acylation of the secondary OH group. It should be stressed that diacetylation of 1-AAs does not suffer these dramatic changes: Diacetylation reactions of long-chain 1-AAs under standard conditions (neat 1:1 pyridine-acetic anhydride as solvent and reactant) were unimpeded and high-yielding [12] as expected with the more reactive acylation reagent and larger over-equivalencies. We presume the uniform success of acylation of long-chain 1-AAs under the conditions of Method B avoids micelle formation when carried out in solvents of lower dielectric constant (e.g., CH 2 Cl 2 or DCE) than pyridine.

The EC ECD Conformational Model is Upheld
Interpretation of the EC ECD spectra of 1b and 2b are based on Nagai's observations of the sign of the split CE and conformation analysis of (S)-14, the simplest chiral homolog of an N,O-dibenzoyl-1-AA ( Figure 5) [21]. Of the three dominant gauche conformers, i is disfavored by steric hindrance. Conformer ii is expected to show ∆ε~0 due to the antiperiplanar arrangement of the transition dipole moments of the two Bz chromophores, leaving iii as the favored conformer with the positive helicity inducing a negative split CE. The measured ECDs of the homologs 4b, 9b and 13a,b in this study all conform to this fundamental model; S enantiomers correlate with positive split-CE and R enantiomers correlate with a negative split CE. The strongest CEs in the series are observed by replacement of the Bz chromophore with 2-naphthoyl group (viz. 9a,b and 13a,b; [12]); consequently, introduction of the latter chromophore is preferred for stereoanalysis of natural 1-AAs.

The EC ECD Conformational Model is Upheld
Interpretation of the EC ECD spectra of 1b and 2b are based on Nagai's observations of the sign of the split CE and conformation analysis of (S)-14, the simplest chiral homolog of an N,O-dibenzoyl-1-AA ( Figure 5) [21]. Of the three dominant gauche conformers, i is disfavored by steric hindrance. Conformer ii is expected to show  ~0 due to the antiperiplanar arrangement of the transition dipole moments of the two Bz chromophores, leaving iii as the favored conformer with the positive helicity inducing a negative split CE. The measured ECDs of the homologs 4b, 9b and 13a,b in this study all conform to this fundamental model; S enantiomers correlate with positive split-CE and R enantiomers correlate with a negative split CE. The strongest CEs in the series are observed by replacement of the Bz chromophore with 2-naphthoyl group (viz. 9a,b and 13a,b; [12]); consequently, introduction of the latter chromophore is preferred for stereoanalysis of natural 1-AAs.

Semi-Quantitative Comparison of ECD and CD methods for Stereoassignment of 1-AAs
The value of %ee, defined by a ratio (Equation (1)) of the sum and difference of concentrations, C, of (+)-and (−)-enantiomers, is prone to errors of measurement in each. Comparisons of []D for sample and standard have long been used for determinations of %ee, but they become flawed when the magnitudes of the specific rotations are low, or when sample solutions of different concentrations induce non-linearity. The complementarity of the two methods used in this work-ECD and CDA derivatization with MPA-for establishing %ee and AC of 1-AAs, such as 1a, is apparent from a simple semi-quantitative analysis of the sources of error and observation of the discrepancies between the two methods ( Table 1). The ECD method is best suited for %ee of near-racemic samples  [21]. See also [8]; p. 159).
Finally, the naphthoyl-naphthimido derivatives, while well-served for 'fingerprinting' threo and erythro isomers of 2-amino-1,3-diols such as sphingosines and dihydrosphingosines [18,19], offer no advantage over a simple di-naphthoyl derivative (e. g., 9b or 13a,b) for assignments of 1-AAs; the latter derivatives offers simplicity, exceptional sensitivity combined with ease of preparation and interpretation of their ECD spectra.

Semi-Quantitative Comparison of ECD and CD methods for Stereoassignment of 1-AAs
The value of %ee, defined by a ratio (Equation (1)) of the sum and difference of concentrations, C, of (+)-and (−)-enantiomers, is prone to errors of measurement in each. Comparisons of [α] D for sample and standard have long been used for determinations of %ee, but they become flawed when the magnitudes of the specific rotations are low, or when sample solutions of different concentrations induce non-linearity. The complementarity of the two methods used in this work-ECD and CDA derivatization with MPA-for establishing %ee and AC of 1-AAs, such as 1a, is apparent from a simple semi-quantitative analysis of the sources of error and observation of the discrepancies between the two methods ( Table 1). The ECD method is best suited for %ee of near-racemic samples where the null point corresponds to a racemate (∆ε sample /∆ε standard = 0). The sign and magnitude of non-zero split-CE signal (~4%ee and higher) indicate configuration and non-racemic mixtures, respectively, with errors in the latter associated only with absolute A values and the signal-to-noise of the measurement.
EC ECD of vicinal N,O-di-arenoyl derivatives has two advantages over [α] D for determination of %ee: ∆ε is a molar quantity that conforms to the Beer-Lambert law and the absolute magnitude of the ∆ε signal (or A) is typically larger. Consequently, EC ECD is preferred for AC determination and quantitation of %ee, but there are limitations. Errors in %ee will be larger for samples near 100%ee where the ratio is obtained from ∆ε sample and ∆ε standard values that differ little in magnitude. MPA-derivatization-although requiring the preparation of separate Sand Rderivatives-is equally reliable for assigning AC of 1a. For optical purity, however, the MPA method is better suited for measurements within a different range of %ee values (closer to 50%ee) where cumulative bias in integral measurement (C (+) or C (−)) due to instrumental factors (e.g., absolute error of integrals, partial NMR signal saturation, poor S/N and non-linear error propagation) is lowest.

Geographic Variation of 1a
The finding of heterogeneous enantiomeric compositions of different samples of 1a is unusual, and suggestive of a biosynthesis from lipid and amino acid precursors that differs from conventional long-chain base biosynthesis at one or more later steps. The proposed biosyntheses of 1-AA natural products (Figure 6a) is modeled on sphinganine-mycotoxin biosynthesis [22] with parallels to mammalian sphingosine biosynthesis (Figure 6b) [1]: Condensation of the C 16 palmitoyl CoA ester with Ser, with concomitant decarboxylation, gives ketosphingosine (15) that undergoes stereospecific reduction of the keto group by NADPH and subsequent oxidative desaturation (Strictly, desaturation occurs on the corresponding ceramide, and free D-erythro-sphingosine is released through the action of a ceramidase ([1] p. 1622])) to give (2S,3R)-sphingosine.

Geographic Variation of 1a
The finding of heterogeneous enantiomeric compositions of different samples of 1a is unusual, and suggestive of a biosynthesis from lipid and amino acid precursors that differs from conventional long-chain base biosynthesis at one or more later steps. The proposed biosyntheses of 1-AA natural products (Figure 6a) is modeled on sphinganine-mycotoxin biosynthesis [22] with parallels to mammalian sphingosine biosynthesis (Figure 6b) [1]: Condensation of the C16 palmitoyl CoA ester with Ser, with concomitant decarboxylation, gives ketosphingosine (15) that undergoes stereospecific reduction of the keto group by NADPH and subsequent oxidative desaturation (Strictly, desaturation occurs on the corresponding ceramide, and free D-erythro-sphingosine is released through the action of a ceramidase ([1] p.1622])) to give (2S,3R)-sphingosine. The latter sequence is highly conserved in higher animals: All mammalian sphingosines, including variants with different chain-lengths, are of the D-configuration. By comparison, the proposed biosynthesis of 1a would appear to proceed by activation of a hypothetical long-chain diyne-ene C17 fatty acid CoA ester 16, followed by decarboxylative condensation with Gly instead of Ser, and reduction of the resultant ketone 17 by NADPH. Unlike 15, the proposed intermediate The latter sequence is highly conserved in higher animals: All mammalian sphingosines, including variants with different chain-lengths, are of the D-configuration. By comparison, the proposed biosynthesis of 1a would appear to proceed by activation of a hypothetical long-chain diyne-ene C 17 fatty acid CoA ester 16, followed by decarboxylative condensation with Gly instead of Ser, and reduction of the resultant ketone 17 by NADPH. Unlike 15, the proposed intermediate aminoketone 17 undergoes reduction with lower stereofidelity than reduction of 15. In contrast, all natural 2-AAs appear to be biosynthesized with hallmark high stereofidelity even when the relative and absolute configurations differ from species to species [23].

Broader Implications
The foregoing results should have broader applications to both stereochemical analysis of other aminoalkanols from Nature and synthesis. While reliable, prior analyses of 2-AAs exploiting ECD of bis-N,O-dibenzoyl derivatives (e.g., halisphingosines; [15]) suffered the same problem of lower-than desirable yields. The corresponding bis-N,O-di-(2-naphthoyl) derivatives, prepared by Method B, should ameliorate problems of compound yield and even permit sub-nanomole analyses (e.g., Method C) on vanishingly small sample sizes. In the asymmetric syntheses of 1-AAs and 2-AAs, the challenges of determination of %ee can better be met by ECD, or CDA derivatization and NMR; methods that nicely overcome the limitations imposed by sole reliance on comparisons of [α] D .

Materials and Methods
All UV-vis and chiroptical measurements were made with solutions prepared in spectroscopic grade solvents; MeOH, CHCl 3 (stabilized with amylenes) or CF 3  Isolation and purification. The freeze-dried ascidians were macerated and cold extracted with MeOH (4 × 200 mL), and the extract filtered, concentrated in vacuo affording a green solid, which was subjected to repeated C 18 reversed-phase column chromatography using a gradient of H 2 O (0.05% TFA) to 100% MeOH. Distaminolyne A·TFA salt eluted with 70% MeOH/H 2 O as a pale yellow gum.   the tube were removed with a narrow capillary and applied directly onto a glass backed TLC plate (6 × 6 cm, silica, 250 µm, prewashed by 2× development in 1:1 EtOAc/n-hexane) in replicate lanes (n = 6). After development of the TLC plate in 1:4 EtOAc/n-hexane, the fluorescent spots corresponding to the N,O-dinaphthoyl-1-AA product (for (R)-12b, R f 0.20) were scraped from the plate into a 6 mL vial containing a magnetic stir bar, and the mixture extracted with TFE (1.0 mL) by vigorous stirring for 30 min. The vial was centrifuged and measurements of the UV-vis and ECD spectra were carried out directly on the supernatant (30-40% yield based on absorbance at λ max 231 nm).