Circular dichroism spectra, optical rotations and absolute configurations of cis-dihydrodiol metabolites of quinoline and derivatives: The role of the nitrogen atom

Authors


  • Contribution to the special thematic project “Advances in Chiroptical Methods”

Abstract

The absolute configurations of cis-dihydrodiol bacterial metabolites derived from quinoline substrates and their quinolinium salts have been determined by comparison of experimental and calculated chiroptical data (circular dichroism spectra and optical rotations). The role of intramolecular H-bonding and other interactions in the preferred conformations has been investigated, including the effect of protonation of several cis-dihydrodiols. The presence of a strong intramolecular OH···N bond in several cis-dihydrodiol metabolites suggests that such species have a residually chiral oxygen atom. Chirality 21:E37–E47, 2009. © 2009 Wiley-Liss, Inc.

INTRODUCTION

Quinoline, the simplest member of the polycyclic azaarene series, and its derivatives, are widely distributed in the environment through the partial combustion of fossil fuels, forests and tobacco, their presence in crude oil and creosote, their occurrence as plant alkaloids and also diverse anthropogenic origins. Polycyclic azarenes (APAHs, e.g. quinoline, acridine and phenanthridine) are present in the environment with concentrations of up to 10% of the more common polycyclic aromatic hydrocarbons (PAHs).1 These APAHs have however often shown increased carcinogenic and mutagenic activity compared with PAHs. Thus, in contrast to naphthalene, quinoline is a heptocarcinogen in mice and rats2 and a mutagen in bacteria.3–5

In order to determine the origin of their biological activity, earlier studies from our laboratories have focussed on the metabolism of quinoline and derivatives using both animal liver microsomes (a source of cytochrome P-450 monooxygenases),6 and soil bacterial strains (a source of dioxygenases).7–9 This program has also involved the chemical synthesis of a wide range of isolated (or potential) quinoline metabolites including racemic and enantiopure arene oxides, cis- and trans-dihydrodiols as well as diol epoxides, N-oxides, and quinolinium derivatives.6, 10–14 The relative mutagenic activities of more than twenty derivatives of this type (resulting mainly from quinoline oxidation) have been reported.15

Our more recent studies have focussed on the bacterial dioxygenase-catalysed cis-dihydroxylation of quinolines and other APAHs.9, 16–21 Azaarene cis-dihydrodiol-derived 2,2′-bipyridine ligands for asymmetric aminolysis of epoxides and asymmetric allylation of aldehydes, Boyd et al. (submitted). The structural and stereochemical characterization of the corresponding enantiopure cis-dihydrodiol metabolites of quinoline (1c, 2c),8, 13 Boyd et al. (submitted), 2-chloroquinoline (1a, 2a),9 2-methoxyquinoline (4),9 2-phenylquinoline (5),9 4-chloro-3-methylquinoline (6), 3-bromoquinoline (7),9 4-chloroquinoline (3),19 2-chloro-3-methylquinoline (1b, 2b),20 acridine (8),21 phenanthridine (9)18 and the quinoline alkaloid dictamnine (10, 11),17 have to date largely depended on NMR spectroscopy, X-ray crystallography and stereochemical correlation methods.

Using whole cells of Pseudomonas putida UV4 (a source of toluene dioxygenase, TDO) and quinoline as substrate, the isolated yields of cis-dihydrodiols 1c and 2c were relatively low (<10%). However when using this bacterial strain with 2-chloroquinoline as substrate, the yields of the corresponding cis-dihydrodiols 1a and 2a were higher (ca. 40%).9 Furthermore the presence of the chlorine atom protected the quinoline substrate against oxidation at the C-2 position and facilitated coupling reactions in the corresponding tetrahydrodiol derivatives of cis-dihydrodiols 1a or 1b.20 The synthetic potential of the 7,8-cis-dihydrodiol metabolite of 2-chloroquinoline (1a) has thus recently been realized by their application as chiral precursors of a new series of chiral 2,2′-bipyridines (12). These chiral ligands, 12, have been used in asymmetric allylic hydroxylations (→97% ee),20 alkene cyclopropanations (→95% ee)20 and aminolysis of meso-epoxides(→84% ee) and allylation of aldehydes. Boyd et al. (submitted). Similarly the cis-dihydrodiol metabolite (3) of 4-chloroquinoline has proved to be a key building block in the synthesis of chiral metal-organic frameworks (MOFs) obtained from the resulting 4,4′-bipyridine derivatives.19 The corresponding mono-and di-N-oxide derivatives of chiral 2,2′-bipyridines (12) synthesized from the 2-chloroquinoline 7,8-cis-dihydrodiols (1a and 1b) have also recently been used in asymmetric allylations of aldehydes (→86% ee).22

The cis-dihydrodiol metabolites 1b and 2b (from 2-chloro-3-methylquinoline), and 3 (from 4-chloroquinoline) could not be obtained using P. putida UV4 and were only obtained by using Sphingomonas yanoikuyae B8/36. The biphenyl dioxygenase enzyme present in the latter strain contains a larger active site than in TDO and was better able to accommodate the larger substituted quinolines (2-chloro-3-methylquinoline and 4-chloroquinoline).

Absolute configurations of compounds 1c and 2c were previously determined by 1H NMR analysis of the diMTPA esters of the corresponding cis-tetrahydrodiols. from quinoline substrates and by stereochemical correlation (compound 2c).7, 8, 13, 22 Formation and 1H NMR analysis of the corresponding cyclic boronate derivatives of compounds 1a and 2a, using (S) and (R)-2-(1-methoxyethyl)benzeneboronic acids (MEBBA), and comparison of their CD spectra with that of compound 2c were used to provide assignments of their absolute configurations.91H NMR analysis of the corresponding MEBBA esters from cis-dihydrodiols 1b,2b and 3 were again used to provide a tentative assignments of absolute configuration.19, 20 In the absence of rigorous X-ray crystallographic evidence for the absolute configurations of cis-dihydrodiols 1a-c, 2a-c, and 3, a reliable, generally applicable and direct method based on chiroptical data was thus required.

The present study indicates how both experimentally-based and calculated optical rotation (OR) and circular dichroism (CD) spectroscopy methods can be used to provide information on the preferred conformations and unequivocal assignments of absolute configurations to the cis-dihydrodiol derivatives of quinoline and substituted quinolines.

A comparison of experimental and calculated chiroptical data emerges now as a general and more convenient method for the determination of absolute structure of chiral molecules.23–25 Although there are numerous methods offered currently by theoretical chemistry for stereochemical investigations, those based on time-dependent density functional theory (TDDFT) are most frequently used. Among the many functionals available to date, the B3LYP hybrid functional finds the broadest applicability,26–28 although in the last years the development of new functional forms, and their validation against diverse databases, have yielded new powerful density functionals of broad applicability to many areas of chemistry.29–32 Among these, hybrid density functionals, including a nonlocal correlation effect, give promising results in the case of calculations of molecular properties in the ground as well as in the excited states.33 For ECD spectroscopy, calculations with the use of newly developed double hybrid density functionals, with or without perturbative correction, outperform other TDDFT approaches34 and therefore can be used as the method of choice for reliable calculations of chiroptical properties of chiral systems.35, 36

In this paper the experimental CD/UV data are analyzed using on the grounds of TDDFT computational results. We have chosen to use the newly developed functional B2LYP for computations as well as the most popular B3LYP and the “classic” mPW1PW91 functionals for comparison, in conjunction with the enhanced 6-311++G(2D,2P) basis set. For independent assignments of absolute configurations we performed additional calculations of optical rotations of investigated molecules at the B3LYP/6-311++G(2D,2P) level which in general gave support to ECD-based stereochemical results.37–42

We used dihydrodiols 1-3 (see Chart 1) as representative molecules for computations. In order to establish the factors determining the chiroptical properties of dihydroquinoline cis-diols we used the computed data for naphthalene cis-diol 1343 and 7,8- (14) and 5,6-dihydroquinoline (15) for comparison.

Chart 1.

MATERIALS AND METHODS

cis-Dihydrodiols 1a-c, 2a-c, and 3 were available as bacterial metabolites of the corresponding quinolines (quinoline, 2-chloroquinoline, 4-chloroquinoline and 2-chloro-3-methyl-quinoline) from earlier studies where full characterization data is reported.8, 9, 19, 20

ORs were determined using a Perkin–Elmer 214 precision polarimeter with the specified solvent, concentration and wavelength at ambient temperature. CD spectra were recorded on a JASCO 810 instrument using spectroscopic grade acetonitrile.

Computational Details

In our computations, all excited-state calculations have been performed, based upon the ground state geometries of single molecules, with the use of a Gaussian program package.44 Rotatory strengths were calculated using both length and velocity representations. In the present study, the differences between the length and velocity of calculated values of rotatory strengths were quite small, and for this reason only length representations were taken into account. The CD spectra were simulated by overlapping Gaussian functions for each transition according to the procedure described by Diedrich and Grimme.45

Our computational analyzes have been performed according to a classical methodology. This includes as the first step, the conformational analysis for the cis-dihydrodiol, then the geometry optimization of each low-energy conformer, calculation of the rotatory strength and the optical rotation for individual stable conformers, and as the last step, a comparison of the Boltzmann averaged calculated CD spectra as well as the optical rotations with the corresponding experimental data. This methodology was described in detail previously.43, 46–48

To test which combination of the functional/basis set performs better in the case of calculations of the CD spectra of substituted quinoline cis-dihydrodiols, the B2LYP hybrid functional has been employed in combination with different basis sets. The basis sets varied from relatively small to enhanced, with or without the diffuse functions. The best performing combination was obtained with B2LYP/6-311++G(2D,2P) and this was used for calculating the CD spectra. Additionally, the CD spectra were calculated with the use of B3LYP and mPW1PW91 hybrid functionals and 6-311++G(2D,2P) basis set. The computed oscillator strengths and rotatory strengths were converted to the UV and CD spectra by broadening to Gaussian shape absorption curves. The calculated spectra were overestimated by ca. 10–15 nm in relation to the experimental ones when the B2LYP/6-311++G(2D,2P) method was used. By contrast, the transition energies calculated with the use of mPW1PW91 and B3LYP hybrid functionals and the 6-311++G(2D,2P) basis set were underestimated by 10 and 5 nm, respectively.

Although each experimental spectrum was reproduced well by either B3LYP or mPW1PW91 hybrid functionals, in conjunction with the basis sets augmented with diffuse functions, still better overall agreement was obtained with the use of the B2LYP hybrid functional and 6-311++G(2D,2P) basis set. These data are reported in this article. For all compounds the 25 excited states were employed to simulate the ECD spectra. Because the last 10 excited states are lying in the higher energy region, they did not influence the shapes of calculated spectra.

ORs were calculated at the B3LYP/6-311++G(2D,2P) and B3LYP/Aug-CC-pVTZ levels for selected model cis-dihydrodiols. Due to similar values of the calculated ORs and because of a longer calculation time required for the B3LYP/Aug-CC-pVTZ computations, the ORs were calculated only at B3LYP/6-311++G(2D,2P) level for all other cis-dihydrodiol molecules. No correlation for the medium dielectric constant was implemented.

RESULTS AND DISCUSSION

Conformational Analysis of cis-dihydrodiols 1a-1c, 2a-2c, and 3

In order to obtain reliable results from CD and OR calculations, we first carried out calculations of low-energy conformers of cis-dihydrodiols by the previously described protocol.43, 46–48 The calculated structures of the low-energy conformers were optimized by the use of the B3LYP functional and the enhanced basis set 6-311++G(2D,2P). Up to four conformers within the 2.0 kcal mol−1 energy window were obtained for each cis-dihydrodiol. These conformers belong to two families, distinguished by the helicity (M or P) of the 7,8- or 5,6-dihydroquinoline chromophore. Calculated dihydroquinoline skew angles γ (C9-C10-C5-C6 or C10-C9-C8-C7) of the cis-dihydrodiol metabolites (see Chart 1 and Table 1) are all within the narrow ranges of 10.6 to 12.6° (for P) and −10.6 to −12.7° (for M).

Table 1. Calculated B3LYP/6-311++G(2D,2P) relative free energies (ΔG), populations and torsion angles α, β, γ for low-energy conformers of 1a–1c, 2a–2c, and 3
DiolConformerΔG (kcal mol−1)Population (%)Torsion angle (º)
αβγ
1aM21.67552.0157.6−10.6
P30.009598.872.812.6
1bM21.56652.7157.6−10.3
P30.009498.973.912.5
1cM22.0554.2157.9−10.4
P30.00100100.074.212.6
2aM10.8116163.2153.8−12.0
M21.37667.3162.1−11.1
P10.0064−152.5−157.810.8
P20.8614−158.1−60.511.5
2bM10.8315163.6154.0−12.0
M21.49566.1161.9−11.1
P10.0064−152.7−157.610.6
P20.8216−158.2−63.311.3
2cM10.7616163.5154.3−12.2
M21.31769.3162.1−11.2
P10.0060−153.1−157.510.8
P20.7217−158.7−66.011.6
3M10.1642165.5153.5−12.7
M20.005545.4158.6−11.6
M31.603−74.3−69.6−12.3

The presence of the vicinal cis-diol groups greatly affects the conformational equilibria due to preferred modes of hydrogen bonding. A detailed list of the types of thermally accessible conformers for each cis-dihydrodiol is given in Scheme 1.

Scheme 1.

Diastereomeric P and M conformers of cis-dihydrodiols 1a1c, 2a2c, and 3.

As previously noted for cis-cyclohexa-3,5-diene-1,2-diols46–48 all conformers feature structures stabilized by intramolecular OH···O hydrogen bonds. In the M1 and P1 conformers, torsion angles α (H[BOND][BOND]O[BOND]H) and β (H[BOND][BOND]O[BOND]H) are anti, whereas in the case of the M2 and P2 conformers one of these torsion angles is anti and the other is syn (Scheme 1). Calculated values of angles α and β (see Table A) are within the ranges of ± (153 to 165°) for anti and ± (45 to 100°) for syn. These angles are positive for M1 and M2 conformers and negative for P1 and P2 conformers. Worth noting is an axial arrangement of the allylic hydroxy group in P conformers and an equatorial position in M conformers.

In 7,8-dihydro-7,8-dihydroxyquinolines 1a-1c the nitrogen atom introduces an additional possibility for conformer stabilization through an OH···N hydrogen bond. In each of these conformers (P3) the (pseudo)axial hydroxy group at C7 is a donor for the (pseudo)equatorial hydroxyl group at C8, and this in turn is a hydrogen bond donor to the nitrogen atom (Scheme 1). In P3 conformers both torsion angles α, β are (+)-syn. The calculated OH···distance is the shortest for 1c (2.087 Å) and slightly elongated when chlorine atom is introduced into the aromatic ring (up to 2.123 Å for 1b).

The presence of the chlorine substituent at C4 in dihydrodiol derivative 3 influences the conformational equilibrium in an alternative manner. In this case there is no attractive interaction between the (pseudo)equatorial hydroxy group and the chlorine atom and in effect the P3 conformer is over 5 kcal mol−1 higher in energy than the lowest-energy conformer M2. Another M-type conformer (M3) is found by calculations, to have both α and β torsion angles (−)-syn. However its population is relatively small (3%, ΔG calculation). The calculated populations of the conformers and the values of torsion angles α, β and γ are given in Table 1.

In the case of dihydrodiols 1a-1c, the lowest energy conformer is of P3 type. The presence of a strong OH···N hydrogen bond in the cases of 1a-1c shifts the conformational equilibrium almost completely toward the P3 conformer (over 90% abundance).

In the case of dihydrodiols 2a-2c, the P3 type conformer is not accessible and the lowest-energy conformer in each case is of P1 type, in which the equatorial hydroxy group at C5 donates a hydrogen bond to the axial oxygen atom at C6 and the hydrogen atom of this hydroxy group engages in the formation of a OH···π hydrogen bond. The population of P1 conformers is at least 60% (ΔG calculation), regardless of the substituent(s) in the aromatic ring. An equatorial OH···axial OH···π hydrogen bond pattern is apparently a stabilizing factor in all conformers of M1, M2, P1, and P2 type.

The more stable conformers of dihydrodiol 3 are of M helicity, evidently reflecting stronger C6 axial OH···π bonding with the aromatic π-electron cloud (M1). The C[BOND]Cl bond at C4 apparently destabilizes any P conformation having an equatorial hydroxy substituent at C5.

Absolute Configurations of cis-dihydrodiols from the Study of Circular Dichroism Spectra and Optical Rotations

Dihydrodiols 1a-1c, 2a-2c and 3 contain the azastyrene chromophore embedded in the 7,8-(14) or 5,6-dihydroquinoline (15) framework. Spectroscopic properties of compounds of this type have received little attention in the past. The characteristic electronic π–π* transition involving the orbitals of both the pyridine and the ethylene moieties within the azastyrene chromophore is located in the electronic spectra at ca. 260 nm; for 7,8-dihydroquinoline (14) in cyclohexane solution this transition appears at 257 nm (ϵ = 11,100), where for 5,6-dihydroquinoline (15) in cyclohexane solution this transition appears at 255 nm (ϵ = 20,000).49

In this study we used the calculated structures of the low-energy conformers of cis-dihydrodiols (Table 1) to calculate their CD spectra by the TDDFT method with the use of hybrid functionals B3LYP, mPW1PW91 and the newly developed B2LYP functional, in conjunction with the enhanced 6-311++G(2D,2P) basis set. Although all of these methods worked well, the best agreement between the calculated and the measured spectroscopic data has been obtained for the B2LYP/6-311++G(2D,2P) method, thus we limited our discussion to this method only. The computed energies (λ), oscillator strengths (f) and rotatory strengths (R) for the two lowest-energy electronic transitions in the azastyrene chromophore, which may determine signs and strengths of the long-wavelength Cotton effects, are collected in Table 2.

Table 2. Calculated at B2LYP/6-311++G(2D,2P) level excitation energies (λ, in nm), oscillator strengths (f) and rotatory strengths (R, in 10−40 erg esu cm Gauss−1) for the first two low-energy electronic transitions of the low-energy conformers of 1a–1c, 2a–2c, 3, 5a–5b
DiolConformerFirst electronic transitionSecond electronic transition
λfRλfR
1aM22640.113−13.1832510.276−10.689
P32570.273−0.7692480.11714.104
1bM22620.163−17.4902530.192−8.984
P32590.317−9.5212490.03828.613
1cM22560.132−19.3512450.170−21.648
P32510.2446.2612410.05521.309
2aM12600.213−8.5302440.057−11.474
M22630.215−8.4192450.056−23.385
P12630.2040.0892370.03149.078
P22660.2110.4252370.02847.310
2bM12610.256−9.7632490.080−2.784
M22640.261−8.4522500.072−15.985
P12640.266−1.1052420.03839.101
P22670.273−0.7472410.03637.733
2cM12540.218−21.4382410.0568.083
M22560.221−24.4872420.054−10.142
P12560.22818.1812380.003−3.910
P22590.23417.5812380.002−5.360
3M12530.216−21.9152420.0403.715
M22540.226−19.7722430.0555.495
M32510.248−5.2672390.0180.829
14P2530.283−0.9372440.03233.704
15P2560.280−12.4532390.01736.028

To analyze the data we started with the simple model molecules 14 and 15, which are achiral (racemic) on a macroscopic scale. These dihydroquinolines are calculated to have a strong long-wavelength electronic transition 1 (HOMO-LUMO) above 250 nm having π–π* character and polarized approximately along the x-axis (Fig. 1). Calculated rotatory strengths for the P conformers for this transition are small and negative (Table 2), however much stronger positive rotatory strengths are calculated for a closely lying (ca 240 nm) transition 2, having low oscillator strength and involving mainly HOMO and LUMO+1 orbitals. This results in very different calculated CD spectra being obtained for 14(P) and 15(P); the first giving only a positive Cotton effect at 244 nm and the second showing a pair of negative/positive Cotton effects at 256 nm and 239 nm, respectively (Fig. 1).

Figure 1.

UV and CD spectra, oscillator and rotatory strengths (vertical bars) and directions of polarizations of main electronic transitions for model compounds 1c, 2c, 14, and 15 calculated at TDDFT/B2LYP/6-311++G(2D,2P) level of theory (a) orientation of calculated molecules in cartesian axes (b).

Thus, these data demonstrate that no simple helicity rule is applicable to the long-wavelength electronic transition of the chiral dihydroquinoline chromophore.

Insertion of the cis-diol grouping to 14 and 15 leads to metabolites 1c and 2c, respectively. Comparing the rotatory strengths of 14 and 15 with these calculated for the individual conformers of 1c (M2 and P3) and 2c (M1, M2, P1, P2) allows the following observations. Axial allylic hydroxy groups in P1, P2, P3 make a positive contribution to the rotatory strength of transition 1 and a negative contribution to the rotatory strength of transition 2. An allylic equatorial hydroxy group in M1 and M2 also contributes to the rotatory strength of both transitions, however its contribution is of opposite sign. In both cases the sign of contribution of the allylic hydroxy group to the longest-wavelength Cotton effect is in accord with the “allylic chirality rule”, i.e. with the helicity of the O[BOND]C[BOND]C[DOUBLE BOND]C bond system.

The effect of a chloro or methyl substituent on the calculated position and oscillator strength of the long-wavelength transition 1 is small. Transition 2 is of low oscillator strength, except for conformers M2 of dihydrodiols 1a-1c for which transition 2 is calculated to have a higher oscillator strength compared to transition 1.

More significant substituent effects are evident from the calculated rotatory strengths. For example, the calculated rotatory strengths for the lowest-energy electronic transition of the M-type conformers of 1 are in all cases negative and this is in contrast to the results obtained for P-type conformers of 1. In the case of unsubstituted quinoline cis-dihydrodiol 1c(P3) the calculated lowest-energy rotatory strength is positive, and the introduction of the chlorine atom in 1a(P3) decreases the calculated R almost to zero, while a negative R is calculated for 1b(P3). Since the population of the P conformer is dominating, Boltzmann averaged CD spectrum is essentially due to this type of conformer. Because of the stability of P3 conformers one may postulate the presence of a “chiral oxygen atom”, attached to C8, with four different substituents, C(8), H [bonded to O(8)], H [bonded to O(7)] and an electron pair (Scheme 1). Although direct evidence for such an atom arrangement is not presented, results of our CD calculations appear strongly supportive for the presence of such a species.

Similarly, in the case of the quinoline cis-dihydrodiols 2 the calculated lowest-energy transition rotatory strengths reached the maximum value for unsubstituted 2c and decreased dramatically for mono- and disubstituted 2a and 2b respectively, with the signs reversal observed for P1 and P2 conformers of 2b. In the case of 3 the calculated lowest-energy transition rotatory strengths are negative for all M conformers.

Detailed inspection of data collected in Table 2 leads to a conclusion that for the second electronic transition, the calculated rotatory strengths are less sensitive to substitution pattern in quinoline cis-dihydrodiols, with the exception of 2c.

Calculated Boltzmann-averaged and experimentally measured CD spectra of cis-dihydrodiols are shown in Figure 2. In general, good (although not perfect) agreement between the experimentally measured spectra, and those calculated for the assumed absolute configurations shown in Scheme 1, is observed for the longer-wavelength Cotton effects but for the shorter-wavelength CD bands the agreement is worse. However, in this study we focused only on the long-wavelength Cotton effects, whose analysis allows us to determine the absolute configurations of these molecules. Despite the very good agreement found between the calculated and experimentally measured UV spectra (Fig. 2), a less satisfactory correlation can be expected from the corresponding CD spectra, as the experimentally measured CD spectra can be highly dependent on the solvent effects. This is observed for cis-dihydrodiol 3 for which the calculated and the experimental CD spectra do not match well, probably due to underestimation of P-type conformers population by calculation. Nevertheless, except for compound 3, the absolute configurations of the other cis-dihydrodiols derived from quinolines can be unequivocally determined by a comparison of the experimental and calculated CD spectra.

Figure 2.

UV (left panels) and CD (right panels) spectra of cis-diols 1-3 measured in acetonitrile solutions (solid lines) and Boltzmann averaged calculated at B2LYP/6-311++G(2D,2P) level (dashed lines) Rotatory strengths R were calculated in dipole-velocity representation. All calculated spectra were wavelength corrected to match experimental long-wavelength UV λmax values (1c was not available for CD measurements).

Further evidence for the influence of the nitrogen atom on the CD spectra of cis-5,6-dihydroxy-5,6-dihydroquinolines was obtained by comparing the experimental CD spectra of compound 2c with the CD spectrum measured for naphthalene cis-diol 1343 (Fig. 3). Since the population of individual conformers is in both cases virtually the same, the CD spectrum depends on the nature of the chromophore only. Thus, the observed long-wavelength Cotton effect is negative for 2c (although very weak), due to the contribution of M-type conformers exhibiting calculated negative rotatory strengths while it is positive for 13, in which case the calculated rotatory strength for 13(P1)43 conformer is strongly positive.

Figure 3.

CD spectra of 2c (solid line) and 13 (dashed line) measured in acetonitrile solutions.

The presence of a nitrogen atom in cis-diols 1-3 makes experimental determination of the hydrogen acceptor properties of this type of compounds possible. Since the quinoline cis-dihydrodiols undergo rapid aromatization under acidic conditions, we were only able to demonstrate, in a qualitative way, the different behavior of protonated diols of type 1 and 2. The UV and CD spectra of diols 1a-1b, 2a-2b and 3, measured in acetonitrile solution immediately after acidification with methanesulfonic acid, are shown in Figure 4.

Figure 4.

UV (left panels) and CD (right panels) spectra of protonated cis-diols: 1a-1b, 2a-2b and 3, in acetonitrile solution acidified with excess methanesulfonic acid.

Whereas in the case of protonated quinoline cis-dihydrodiols 2 and 3 the UV and CD spectra are red shifted about 50 nm in relation to the parent compounds, and preserve most of the features of the spectra of their nonprotonated counterparts, the CD and UV spectra measured for protonated and nonprotonated 1 show significant differences. For example, new absorption maxima at around 260 nm appear in the cases of protonated 1a and 1b. The negative long-wavelength Cotton effect found for nonprotonatedcis-diols disappeared after protonation and a new positive Cotton effect formed at around 310 nm. Although the explanation of these phenomena is not available at the moment, we assume that the changes observed in the case of 1 may be due to the change of polarization direction of the electronic transition dipole moments after protonation. This change is apparently less significant in the cases of 2 and 3, leaving the shapes of the CD spectra little changed, save for the red shift of the CD maxima.

A further strong confirmation of the assigned absolute configurations of dihydrodiols 1a-1b, 2a-2c, and 3 was obtained from the calculation of their optical rotations. The calculations at the B3LYP/6-311++G(2D,2P) level were carried out for each low-energy conformer and for the Boltzmann-averaged (ΔG) mixtures of conformers. To achieve high reliability in these calculations we computed the optical rotation values at four different wavelengths (589, 578, 546, 436 nm) and compared the averaged data with the experimental measurements taken in methanol solution (Table 3).

Table 3. Calculated at the B3LYP/6-311++G(2D,2P) level and measured optical rotations for cis-dihydrodiols 1a–1c, 2a–2c and 3
DiolConformerCalculated opical rotationsMeasured optical rotationsa
589 nm578 nm546 nm436 nm589 nm578 nm546 nm436 nm
  • a

    In methanol solution

  • b

    In tetrahydrofuran solution (Ref.8)

1aM2+36+17+14−42    
P3+148+192+222+408    
ΔG Boltzmann averaged+142+183+211+385+198+208+241+457
1bM2+53+34+33−16    
P3+119+157+182+341    
ΔG Boltzmann averaged+115+150+173+320+164+172+199+370
1cP3+175+250+292+581+45n.a.n.a.n.a.
2aM1+132+161+184+315    
M2−7−50−65−205    
P1+184+260+303+606    
P2+232+321+374+737    
ΔG Boltzmann averaged+171+234+272+529+171+179+207+386
2bM1+159+197+225+394    
M2+39+12+7−67    
P1+192+263+306+599    
P2+240+322+374+724    
ΔG Boltzmann averaged+187+250+290+555+160+169+195+364
2cM1+107+120+135+209    
M2−65−140−171−433    
P1+329+463+541+1085    
P2+384+537+627+1254    
ΔG Boltzmann averaged+275+378+441+867+220bn.a.n.a.n.a.
3M1+112+130+147+240    
M2+7−25−34−130    
M3+92+102+116+186    
ΔG Boltzmann averaged+54+43+46+35+87n.a.n.a.n.a.

In general, the calculated and experimentally measured [α]D values were large, making the comparison highly reliable. We observed excellent agreement between the experimentally measured, and the calculated, optical rotation data, even though the former were obtained in methanol solution. It has been reported before that intramolecular hydrogen bonded structures can still be retained in methanol solution.50 In all cases measured and calculated OR values at the sodium D-line were positive. Therefore, in combination with the CD data discussed above, the absolute configurations of the cis-dihydrodiols could be assigned unequivocally.

A more detailed analysis of the calculated OR values for individual conformers of cis-dihydrodiols (Table 3) revealed that, as it was in the case of the CD spectra, the sign of optical rotation does not correlate directly with the helicity of the azastyrene chromophore. With the exception of M2 conformers of 2a and 2c, M1 and M2 conformers display [α]D values of the same (positive) sign as do the P1 and P2 conformers. A striking finding is that the sign of OR can be directly correlated with the P helicity of the allylic bond system, consisting of C5-C6-C7-O in 1a-1c or C8-C7-C6-O in 2a-2c and 3, again the exception being the M2 conformers. Additionally we notice that the calculated optical rotation for M-type conformers is significantly dependent on the direction of the OH bond at the benzylic position. Whereas OR of M1 conformers is always positive, for M2 conformers is less positive or just negative. This ilustrates the general principle that OR is an extremely sensitive parameter for small variations of molecular structure.

The correlation between the OR and the helicity of the allylic bond system can be directly used for the assignment of absolute configuration at the allylic position, that is, 7S in 1, 6S in 2 and 3, based on the analysis previously proposed by us for the analogous naphthalene cis-dihydrodiols.43

CONCLUSIONS

In this study we have shown that the computational simulation of circular dichroism and/or optical rotation data provide solid support for the assignment of absolute configurations of cis-dihydrodiol metabolites of quinoline derivatives. Most have previously been tentatively assigned via1H-NMR studies of their diastereoisomeric cyclic boronates (compounds 1a, 1b, 2a, 2b, and 3) or diMTPA esters of the derived cis-tetrahydrodiols (1c and 2c).

Throughout this work the role of the nitrogen atom in the azastyrene chromophore of the quinoline metabolites has been demonstrated by TDDFT calculations and correlation with the experimental spectra. Firstly, the electronic properties of the azastyrene chromophore present in these dihydroquinoline derivatives differ significantly from those of the parent styrene chromophore (in dihydronaphthalene derivatives) and in addition the CD spectra are dependent on the position of the nitrogen atom in the skeleton. Secondly, the nitrogen atom serves as a proton acceptor for the hydroxy group (equatorial) at C8 in cis-diols 1a-1c, with cooperative OH···O bonding to the axial hydroxy group at C7. This type of conformation is highly stable (over 94% population) and provides an unusual example of residual stereoisomerism due to a “chiral oxygen atom” at C8. Such residual stereoisomerism cannot be directly observed experimentally, however calculations of the very stable conformer P3, allied to the close similarity between the calculated and the experimental CD spectra give strong support to such a conclusion. We also confirm our previous notion that the presence and the conformation of the hydroxy groups is of primary importance in determining the chiroptical properties of cis-dihydrodiol metabolites. For example, M1 and M2 type conformers were calculated to have quite different values of optical rotation, although their structural differences are limited to the direction of one of the OH bonds. On the other hand, the helicity of the azastyrene chromophore (torsion angle γ) seems to be of only secondary importance with regard to the CD spectra since there is no direct relationship between the sign of γ and the sign of the long-wavelength Cotton effect, corresponding to the excitation of the whole π-electron system.

A general conclusion, embracing all of the above observations is that the safest way to determine the stereostructure (absolute configuration) of a complex heteroaromatic system is to correlate the experimental and the calculated CD spectra and/or OR data, using the best available calculation method.

Ancillary