From iridoids to dyes: a theoretical study on genipin reactivity

Abstract

A mechanistic theoretical study of the reaction between methylamine and genipin, an iridoid obtained from some natural sources, is here presented. Starting from the opening of the genipin six-membered ring experimentally hypothesized and proved in our previous theoretical work, methylamine reaction with the three different forms of genipin in equilibrium has been studied in terms of activation enthalpies and product stabilization. The kinetically favored reaction pathways, initiated by the attack of the amine on the carbonyl groups of the dialdehydic open-ring form of genipin, have been then studied until the formation of the first orange/red product. The catalytic role of a few water molecules has been clearly shown for many reactions in the process and a fundamental redox autocatalytic cycle has been proposed for the formation of the orange/red product. This mechanism explains well the experimental evidence previously reported in literature. ¹H-NMR and UV-visible spectra have been reproduced for each stable intermediate of the process.

---

1. Introduction

Aimed at designing new and low-impact chemical systems, great efforts have been made to mimic natural structures and functions. In particular, naturally colored molecules are increasingly studied to develop new technologies and obtain nontoxic and effective final products.^1,2 Genipin (1, Fig. 1) is a colorless iridoid of particular interest as a precursor for the synthesis of several dyes.³ It is produced from the enzymatic hydrolysis of geniposide, one of the primary active principles contained in the fruit of Gardenia jasminoides, well known as a source of colorants used in food chemistry.^3–5 The interest of genipin lies in its ability to react with primary and secondary amines. This reaction is of relevance, for instance, in polymer chemistry and in many biological applications where genipin, due to its biodegradability and low cytotoxicity, is investigated as a cross-linking agent.^6–8 It is also well known in the field of coloring research that the reaction between genipin and amine groups (either primary amines or amino acids) yields blue pigments.^9,10

Fig. 1 Topological formula and view of the optimized structure of genipin in its SRR, more stable isomer. More informations about stereochemistry of the others isomers considered for genipin closed ring form can be found in our previous work.¹² Structures of selected intermediates playing a major role in the reaction of methylamine with genipin (2, 2′, 7 and 8) are also represented.

Despite the industrial interest of genipin and its wide range of possible applications, its full characterization and reactivity are far from being cleared up. For instance, the structure of genipin in solid state has just recently been characterized.¹¹ In particular, two different unit cells and molecular packings have been found and the predominant stereoisomer in the crystal structure has been identified as the SRR one (see Fig. 1).¹¹ Some hints are also available in literature about the behavior of genipin in solution. In particular, it has been noted that, at basic pH, genipin is unstable and undergoes self-polymerization⁸ whereas in milder pH conditions, a ring-opening reaction has been proposed.^3,13

In a very recent study,¹² we analyzed the behavior of genipin in solution using a combined experimental and theoretical approach. In particular, accurate ¹H-NMR data supported by density functional theory (DFT) computations, indicate that the major form of genipin in solution, among all the possible isomers identified, is the SRR one (or its enantiomer RSS). Moreover, on the basis of theoretical evidences, we suggested that in solution, an equilibrium takes place between the SRR (closed-ring) isomer genipin and two open-ring tautomers (2 and 2′ in Fig. 1), and that ring-opening and tautomerization reactions are catalyzed by one or two water molecules, directly involved as ampholyte species.¹² This behavior of water is well known and, for instance, was previously suggested to explain the reactivity of some other organic molecules such as formamide¹⁴ and alanine and phenylalanine.¹⁵

Genipin open forms, 2 and 2′, could play a major role in the reaction with primary amines. As above mentioned, this reaction is relevant for dyeing and it has been experimentally studied, generally using methylamine as a model of more complex amine systems. In particular, two different mechanisms have been proposed for the nucleophilic attack of the amine on genipin. Park³ suggested that methylamine reacts with one of aldehydic moieties of the dialdehydic open-ring tautomer of genipin, 2. On the other hand, Butler⁶ and Touyama¹⁶ proposed that the methylamine reacts with carbon atom in position 3 of the closed-ring isomer of genipin (1, Fig. 1). The following reaction steps were hypothesized by Touyama et al.,^16,17 based upon isolation and characterization (via UV-visible, NMR and IR spectra and mass spectrometry) of some intermediates. In particular, a drastic color change of the solution, from colorless to yellow to red, was observed with reaction progress. These experimental investigations allowed for the identification of two species, namely a yellow intermediate, (7) whose hypothesized structure is reported in Fig. 1, and the fully characterized product, 2-methyl-4-methoxycarbonyl-8-methyl-2-pyridine (8, Fig. 1). The latter gives a reddish color to the solution and is supposed to be the starting point for the production of blue oligomers.¹⁶

In this scientific context, characterized by a lack of precise information and contrasting mechanistic hypotheses, molecular modeling can be a powerful tool to elucidate the reaction mechanism of genipin with methylamine leading to the experimentally observed intermediate and product (7 and 8). To this end, we have carried out a detailed study on the reaction of methylamine with genipin, based upon the evidences from our previous work on the structure of genipin in solution.¹² In particular, following the suggestions reported in literature, the reaction has been investigated by taking into account the possible role of the solvents (water and ethanol) and of the other species in solution (e.g. potential oxidizing or reducing agents). Theoretical results on the characterization of reaction intermediates (via the simulation of UV-visible absorption and ¹H-NMR spectra) will be compared with experimental data.^16,17

2. Computational details

The hybrid PBE0 exchange–correlation functional,^18,19 mixing 1/4 of Hartree–Fock exact exchange with the PBE functional,²⁰ and the 6-31+G(d) basis set were used to optimize structures and for subsequent frequency calculations to characterize stationary points as minima or first-order saddle points. The accuracy of this level of theory for the treatment of the thermochemistry of genipin reactivity has been validated in our previous work with respect to refined CBS-QB3 calculations.¹² In all the considered cases the discrepancy between PBE0 and CBS-QB3 barriers was less than 1 kcal mol.

Intrinsic reaction coordinate (IRC) calculations were also performed to verify that the identified products and the reactants were correctly connected on the reaction path.²¹ All the enthalpy values were obtained as a sum of electronic and thermal correction to electronic total energies.

Following the (few) available experimental data, ¹H and ¹³C-NMR and UV-visible spectra were simulated for the reaction intermediates 7 and 8. In particular, the first ten vertical excitations were computed at Time Dependent-DFT (TD-DFT) level²² using the approach already considered for optimization and frequency calculations (i.e. PBE0/6-31+G(d)).^23,24 NMR calculations were carried out within the GIAO²⁵ formalism, using the hybrid B3LYP functional²⁶ with the IGLO-III basis set.²⁷ This last level of theory provides accurate chemical shifts for genipin in solution.¹²

Finally, solvent effects were modeled using both implicit and explicit approaches. In particular, explicit water molecules, having a direct catalytic effect on the reactions, were taken into account at the same theoretical level of solute, while bulk solvent effects were introduced by the polarized continuum model in its integral equation formalism (IEF-PCM).²⁸ All the structures have been fully optimized in solution without any symmetry and structural constraints. All the calculations were performed using Gaussian 09 program.²⁹

3. Results and discussion

As mentioned in the introduction, both experimental and theoretical evidences suggest that the most stable isomer of genipin is the SRR one, whose optimized structure is shown in Fig. 1. In solution, a water assisted opening of the dihydropyran ring occurs resulting in an equilibrium between closed and open-ring stable forms of genipin.¹² These findings are the starting point of the present investigation. Furthermore preliminary computations on the reaction in solution suggested that bulk solvent effects are negligible (see Table S1 in ESI† for details). Accordingly, they will not be considered in the following. This is not the case, of course, for few water molecules strongly interacting with the solute and playing a relevant catalytic role (see infra).

Genipin shows a complex structure, where three chiral centers are present and different isomers possible. Our previous investigation¹² showed that the most stable isomer is the SRR one and, among all the isomers and diastereoisomers possible, only the intermediates arising from this reactant have been considered.

Finally, the relative enthalpies reported in the following text and tables, unless otherwise specified, are referred to a single reaction step and are computed as differences between reactants and products (ΔH) or reactants and transition state (ΔH^≠). The corresponding Gibbs free-energies (ΔG and ΔG^≠), reported in the ESI,† show a similar behavior and they will be not discussed in details.

3.1 Open-ring and tautomerization reactions vs. direct amine attack

The isomer SRR, in its closed-ring form (1), can react in two ways with methylamine (A). The first one involves direct attack of A on carbon atom in position 3 (path A, Fig. 2), as suggested by Butler⁶ and Touyama.¹⁶ The second involves nucleophilic attack of the amine on the carbonyl groups of the two open-ring tautomers of genipin, 2′ (path B) and 2 (paths C and D in Fig. 2 – see ESI† for the structure of the transition states). In this case, the first steps are, of course, the ring-opening reaction leading from 1 to 2′ and the tautomerization of this latter into 2. These reactions are both directly assisted and catalyzed by one or two water molecules.¹²

Fig. 2 Sketch of the equilibrium between the closed-ring form of genipin and the two open-ring tautomers and of the four nucleophilic attacks of methylamine considered.

The corresponding enthalpies, collected in Table 1, clearly show that tautomerization is kinetically favored with respect to the direct amine attack (ΔH^≠ = 16.8 and 45.0 kcal mol⁻¹, respectively), whereas the latter is under thermodynamic control (ΔH = 5.6 and −3.6 kcal mol⁻¹, respectively). Product 2′ could react with A (path B) or isomerize to 2: this last path shows the lowest barrier (26.6 vs. 19.2 kcal mol⁻¹, respectively) if the catalytic role of water is considered. Please note that path B is also thermodynamically favored, because of an exothermicity of −6.0 kcal mol⁻¹.

Table 1 Activation enthalpies (ΔH^≠, kcal mol⁻¹) and products stabilization (ΔH, kcal mol⁻¹) calculated for the reactions identified between genipin and methylamine. For each compound involved in the conformational equilibrium of genipin, ΔH^≠ and ΔH of the competing reaction (namely the water assisted open-ring reaction and tautomerization¹²) are also reported

		ΔH^≠	ΔH
1	A	45.0	−3.6
	1 → 2′ (H2O assisted)	16.8	5.6
2′	B	26.6	−6.0
	2′ → 2 (H2O assisted)	19.2	4.1
2	C	18.2	−13.0
	D	18.5	−10.3

---

Finally two competing pathways starting from 2 may occur, namely paths C and D, corresponding to the reaction of A with available carbonyl groups (see Fig. 2). The values of activation barriers calculated for these reactions are similar (about 18 kcal mol⁻¹ for both, Table 1) as well as product stabilization (−13 and −10 kcal mol⁻¹ for reactions C and D, respectively).

Globally, the ring-opening reaction of 1 leading to 2′ is kinetically favored compared to the direct reaction of the closed-ring form of genipin with A. Product 2′, in turn, preferentially tautomerizes to form the 2 dialdehydic open-ring species, whose carbonyls react with amine following two paths C or D which, at this point, cannot be clearly discriminated. The rate determining step of this first part is represented by the 2′ → 2 tautomerization, with an activation barrier of 19.2 kcal mol⁻¹, and the whole reaction is exothermic (about 3 kcal mol⁻¹, when considering, for instance, the process from 1 to the product of reaction path C).

3.2 A Maillard type reaction

The products obtained by reactions C and D, namely 2A and 2A′, react following a Maillard reaction-scheme.³⁰ This pathway can be decomposed in four elementary steps, as summarized in Fig. 3:

Fig. 3 Sketch of the Maillard type mechanism studied for the reaction of methylamine with genipin.

• dehydration of 2A (or 2A′) leading to imine 3 (or 3′)

• isomerization of the imine, giving 4 (or 4′)

• closure of six-membered nitrogen heterocycle, producing 5 (or 5′).

Focusing, for instance, upon the reaction path C, the reaction of A with the aldehydic moiety in position 3 of the genipin open-ring form 2 is exothermic by 13.0 kcal mol⁻¹ and has an activation barrier of about 18 kcal mol⁻¹, as indicated above. The two next steps of the process, notably the dehydration of 2A producing imine 3 and its subsequent isomerization into 4 through intramolecular proton transfer, are instead both endothermic and show prohibitive activation barriers (58.6 and 76.2 kcal mol⁻¹, respectively, Table 2). As for the equilibrium between closed and open-ring forms of genipin, water could act as catalyst. Indeed the presence of one or two water molecules lowers the activation barrier for more than 30 kcal mol⁻¹ (31.2 and 34.2 kcal mol⁻¹ with one or two water molecules, respectively) for the dehydration of 2A, the final estimation of barrier is equal to 24.4 kcal mol⁻¹ (with two water molecules). An even larger effect (about 50 kcal mol⁻¹) is found for the intramolecular proton transfer reaction of imine 3 isomerization into 4 (Table 2).

Table 2 Activation enthalpies (ΔH^≠, kcal mol⁻¹) and products stabilization (ΔH, kcal mol⁻¹) calculated in the gas phase for each step of the reaction pathway producing 7 and starting from nucleophilic attack of methylamine on genipin in the open-ring form 2 dubbed C in Fig. 2. Activation barriers and products stabilizations are also reported for the same reactions catalyzed by one or two water molecules

Reactions	ΔH^≠	ΔH
C	18.2	−13.0
2A → 3 + H2O	58.6	1.6
+ H2O	27.4	4.2
+ 2H2O	24.4	6.5
3 → 4	76.2	4.4
+ H2O	28.0	0.3
+ 2H2O	24.4	2.7
4 → 5cis	29.3	−25.6
+ H2O	1.9	−16.5
+ 2H2O	0.6	−14.1
5cis → 7 + H2O + H2O	43.1	7.0

---

These large variations in activation barriers upon water complexation are essentially due to reduction of ring strains of the pseudo-cycle formed in the transition state structure by the atoms directly involved in the reactions. For instance, in the transition state of imine 3 isomerization, the atoms mostly involved in the proton transfer reaction are arranged in a four-membered cycle, while a more stable six-membered structure is obtained with the involvement of one water molecule (see Fig. S1†). Of course, the larger is the cycle the lower are the ring strain and the activation barrier.

Activation barriers and product stabilizations calculated for pathway D (Table 3) are quite similar to those already discussed for pathway C. Indeed, the dehydration reaction that converts 2A′ into 3′ presents an activation enthalpy of 22.0 kcal mol⁻¹ (vs. about 24 kcal mol⁻¹ for 2A dehydration). Moreover isomerization of imine 3′ into 4′ shows an activation barrier of about 27 kcal mol⁻¹ (vs. 24.4 kcal mol⁻¹ for isomerization of 3). Both reactions are slightly endothermic, for 1.5 and 6.1 kcal mol⁻¹ respectively instead of 6.5 and 2.7 kcal mol⁻¹ calculated for the two correspondent reactions of path C.

Table 3 Activation enthalpies (ΔH^≠, kcal mol⁻¹) and products stabilization (ΔH, kcal mol⁻¹) calculated in the gas phase for each step of the reaction pathway producing 7 and starting from the nucleophilic attack of methylamine on genipin in the open-ring form 2 dubbed D in Fig. 2. Activation barriers and products stabilizations are also reported for the same reactions catalyzed by one or two water molecules

Reactions	ΔH^≠	ΔH
D	18.5	−10.3
2A′ → 3′ + H2O	50.0	0.7
+ 2H2O	22.0	1.5
3′ → 4′	56.5	3.3
+ 2H2O	27.1	6.1
4′ → 5′cis	19.6	−13.6
+ H2O	1.7	−16.2
+ 2H2O	1.1	−13.7
5′cis → 7 + H2O + H2O	26.3	6.5

---

The next step of both pathways is the nucleophilic attack of the amine group onto the carbonyl group of 4 (or 4′), producing the intermediate 5 (or 5′). For this last product, cis and trans isomers are possible, depending on the mutual position of the OH group on carbon 1 or 3 and the H atom in position 7a or 4 (see Fig. 4). However, only the cis isomer can undergo concerted dehydration to produce the partially aromatic species 7. Therefore only the reaction leading to cis-5 (or cis-5′) intermediate from 4 (or 4′) has been considered. This reaction is an intramolecular proton transfer followed by the closure of the six-membered nitrogen heterocycle and is characterized by an activation barrier of about 30 kcal mol⁻¹ (Table 2) and it is exothermic by −25.6 kcal mol⁻¹. Moreover water and its cooperative effect also allow in this case the activation enthalpy to be significantly lowered: one water molecule, bridging the two moieties of 4 involved in the reaction (see Fig. S2†), decreases the barrier to about 2 kcal mol⁻¹ and the process is barrierless (0.6 kcal mol⁻¹) if two water molecules are added. In both cases the reaction is still exothermic by −16.5 and −14.1 kcal mol⁻¹, respectively. The same energetic trend has been obtained for the reactions giving cis-5′ in the parallel pathway (see Table 3).

Fig. 4 Topological formula of the four isomers of the intermediate 5 that can be produced starting from 4 and 4′.

The production of intermediate 7, the first colored (yellow) species characterized in literature,^16,17 is obtained from 5 by the loss of a water molecule (Fig. 3).

This is a concerted and intramolecular reaction involving, in the case of cis-5, the hydroxyl group linked to the carbon atom in position 1 and a proton in position 7a. The transition state of the reaction without water catalysis has not been characterized whereas a large barrier of 43.1 kcal mol⁻¹ (Table 2) has been found for the dehydration assisted by one water molecule. Furthermore, product 7 is less stable than the reactant for 7.0 kcal mol⁻¹, so that the process is both kinetically and thermodynamically disfavored. In contrast, the dehydration of cis-5′ is still endothermic (6.5 kcal mol⁻¹, Table 3) but the activation barrier is found as 26.3 kcal mol⁻¹, more than 15 kcal mol⁻¹ lower than that of cis-5. Looking at the structure of the two cis isomers, it seems evident that the only difference between the two species is the position of the proton undergoing abstraction during the concerted dehydration, the OH group being in both cases close to nitrogen atom. In particular, in the case of cis-5, the proton is linked to one of the carbon atoms bridging the two cycles of the molecule while in the case of cis-5′ it is linked in position 4 to the carbon atom bound to ester function. The values obtained for the two barriers suggest that the ester function activates the C–H bond, thus favoring the dehydration of 5′cis.

The energetic of the Maillard reaction starting from genipin in the open-ring dialdehydic form 2 and yielding the first experimentally-characterized species 7, is summarized in Fig. 5. The two reaction pathways have very similar energy profiles until intermediate 5. Indeed, small enthalpy differences (ΔΔH = ΔH_C − ΔH_D) ranging between 2.7 (calculated between 2a and 2A′ complexes) and −2.3 kcal mol⁻¹ (between imines 3 and 3′) are observed. Both pathways are globally exothermic, respectively for 10.9 kcal mol⁻¹ for the path opened by reaction dubbed C and 9.9 kcal mol⁻¹ for the D one. As already discussed above, only the difference between activation barriers of dehydration reactions producing 7 is really notable (almost 15 kcal mol⁻¹). The rate determining step for the lowest energy path (D) is imine isomerization, characterized by an activation barrier of about 20 kcal mol⁻¹.

Fig. 5 Potential energy surfaces (ΔH in kcal mol⁻¹) of the two parallel pathways depicted in Fig. 3 opened by the two possible reactions of methylamine with the dialdehydic species dubbed 2. All the enthalpy values are relative to the intermediate 2.

It should be noted that our calculations suggest that all proton transfer reactions considered insofar are concerted, either for direct and water-assisted mechanism. The possibility of a stepwise mechanism in the water-assisted reaction has been also investigated for both gas-phase and solution (ethanol) reaction. In both cases it was not possible to identify intermediates (minima or transition states) characteristic of a stepwise mechanism, in agreement with our previous study on other water-assisted reactions, such as keto–enol isomerization in formamide.¹⁴

3.3 Spectroscopic characterization of the first intermediate 7

The intermediate 7 has been proposed as precursor of all the colored species produced in the following reactions, but its experimental characterization is far from being satisfactory. Indeed, the ¹H-NMR spectrum of this product has been collected and it has been defined as being yellow dyestuff, but its λ_max is not reported in literature.¹⁶

The results obtained from the computations of ¹H-NMR spectrum of 7 are collected in Tables 4 and 5 in terms of chemical shifts and coupling constants. At the best of our knowledge, experimental ¹³C-NMR spectrum of 7 is not available in literature. The agreement between theoretical data and experiments is very satisfactory. Looking for instance at the chemical shift of the proton linked to carbon atom in position 1 (H 1), the theoretical value is 5.64 ppm to be compared with the experimental value of 6 ppm. Moreover a constant of 1.3 Hz has been obtained for the coupling of H 1 with the proton in position 3 whereas a value of 1.2 Hz was experimentally found. More generally, the gap between experimental and theoretical data is never higher than 0.6 ppm for chemical shifts and than 1.4 Hz for coupling constants.

Table 4 Experimental¹⁶ and computed (B3LYP/IGLO-III) protons chemical shifts (δ, ppm) of intermediate 7 with respect to tetramethylsilane (TMS) absorption

σ	δ	Proton	δexp
24.59	7.09	H 3	7.1
25.28	6.40	H 6	5.8
26.04	5.64	H 1	6
27.05	4.63	H 8	4.4
27.06	4.62	H 8′
27.86	3.82	CH3 10	3.7
27.96	3.72
28.34	3.34
28.14	3.54	H 4a	3.3
28.32	3.36	H 5	3
29.47	2.21	H 5′	2.25
28.54	3.14	CH3 11	3.1
28.57	3.11
28.85	2.83
31.22	0.46	OH 8	—

---

Table 5 Comparison between experimental ¹⁶ and computed (B3LYP/IGLO-III) coupling constants of 7

Proton	Jexp (Hz)	Jtheo (Hz)
H 3	1.2	1.3 (H 1)
H 1	1.2	1.3 (H 3)
H 5	16.8	17.1 (H 5′)
	8.3	7.4 (H 4a)
H 5′	16.8	17.1 (H 5)
	9	7.6 (H 4a)

---

The theoretical UV-visible spectrum is reported in Fig. 8 and clearly confirms that the intermediate 7 in its optimized structure is a yellow precursor absorbing in the blue region with a λ_max calculated at 369 nm. The good agreement between theoretical and experimental data strongly supports the structure of intermediate 7 as that reported in Fig. 1.

3.4 A redox self-catalytic cycle

For the following of the process, Touyama et al.^16,17 suggested, starting from intermediate 7, the production of 8 (Fig. 6), a colored compound the cycles of which are both aromatic (see Fig. 1 for the optimized structure). The authors also proposed that 8 is produced via an intermediate 8Me, having an activated methylene in its structure.

Fig. 6 Sketch of redox mechanism studied for the production of intermediate 8 starting from 7 reaction by reaction (a) and in self-catalytical arrangement (b).

Our calculations indicate that the mechanism allowing 8 to be produced from 7 consists in a dehydration reaction interposed between two redox steps, as depicted in Fig. 6a. The first step involves oxidation of 7 by the loss of two electrons and a proton, giving the species 7ox, positively charged. The latter undergoes dehydration producing 8Me, still charged and having an activated methylene. The last step is the reduction of 8Me to yield 8.

We have made and explored various hypotheses on the nature of the oxidizing and reducing agents involved in the first and in the last step of the mechanism, respectively. Even though it is difficult to unequivocally determine the species responsible of the first oxidation of 7, the assumption of a mutual redox action of the compounds 7 and 8Me seems reasonable. The mechanism has been studied as a self-catalytic cycle (Fig. 6b). Enthalpy values calculated for reactions involved in this redox self-catalytic cycle are reported on the potential energy profile of Fig. 7.

Fig. 7 Potential energy surface (ΔH in kcal mol⁻¹) of the redox mechanism studied for the production of 8.

The first step, exothermic by more than 15 kcal mol⁻¹, is the formation of a reactant complex between compounds 7 and 8Me. From this adduct, the intermolecular transfer of the hydrogen atom is easy, requiring only 3.7 kcal mol⁻¹. The products formed in such redox reaction are the positively charged species 7ox from 7 through the loss of two electrons and the aromatic species 8 through a proton transfer to 8Me. As illustrated in Fig. 7, these products are more stable than initial reactants for about 24 kcal mol⁻¹. The following step, i.e. the concerted dehydration of oxidized species 7ox to produce 8Me, is the rate determining step of the process, with an activation barrier of 12.6 kcal mol⁻¹. Here the leaving water molecule comes from the hydroxyl group linked to carbon 8 and one of the protons in position 5 (see the optimized structure of TS in ESI†). This step is endothermic by about 12 kcal mol⁻¹. This final step yields product 8 and the whole redox cycle is globally exothermic (of 11.8 kcal mol⁻¹).

3.5 Spectroscopic characterization of the final product 8

Product 8 has been isolated and well characterized in literature as a red-brownish dyestuff using proton and carbon NMR, IR and UV-visible spectroscopy.^16,17 Table 6 shows data obtained from the simulation of ¹H and ¹³C-NMR spectra in terms of chemical shifts, while the proton coupling constants are collected in Table 7. The agreement between experimental and theoretical results is very satisfactory, with differences ranging 0.01 to 0.53 ppm (i.e. 14% of δ value) in absolute value for chemical shifts of proton (Table 6) and from 0.67 to 24.64 ppm (i.e. 6% of δ value). For what concerns proton coupling constants the difference between experimental and calculated values goes from 0.1 to 0.5 Hz (Table 7).

Table 6 Experimental¹⁶ and computed (B3LYP/IGLO-III) proton and carbon chemical shifts (δ, ppm) of 8 with respect to TMS absorption

σ	δ	Proton	δexp
23.94	7.74	H 3	7.73
24.08	7.60	H 1	7.76
24.16	7.52	H 5, H 6	6.84 and 7.14
27.53	4.15
27.66	4.02	CH3 10 and CH3 11	3.97 and 3.95
27.86	3.82
28.04	3.64
28.10	3.58
29.12	2.56	CH3 8	2.47

---

σ	δ	Carbon	δexp
2.97	176.97	C 9	167.7
31.96	147.98	C 6	133
39.70	140.24	C 7a	115.6
45.62	134.32	C 4a	128
46.88	133.06	C 1	129.4
49.54	130.40	C 3	126.8
56.57	123.37	C 7	122.7
56.72	123.22	C 4	113.7
64.30	115.64	C 5	104.5
125.36	54.58	C 10	51.8
132.82	47.12	C 11	44.6
165.72	14.22	C 8	12

---

Table 7 Comparison between experimental¹⁶ and computed (B3LYP/IGLO-III) coupling constants of 8

Proton	Jexp (Hz)	Jtheo (Hz)
H 3	1.5	1.6 (H 1)
H 5	2.9	3.0 (H 6)
	0.5	0.8 (H 1)
H 6	2.9	3.0 (H 5)
	1	1.5 (CH3 8)
H 8	1	1.5 (H 6)

---

Regarding optical properties, UV-visible analysis reported in literature indicates that product 8 has three absorption peaks at 282, 333 and 483 nm. The correspondent values issued from TD-DFT calculations are reported on the simulated spectrum in Fig. 8. The first three values obtained for electronic vertical transition are 278, 329 and 469 nm, respectively which means that the difference with respect to experimental data ranges between 4 and 14 nm. The color obtained for this dyestuff is, as shown on the UV-visible spectrum, a dark orange and the vertical excitation at 469 nm correspond to a π → π* HOMO → LUMO transition (ESI†) and both orbitals are fully delocalized on the whole molecule skeleton.

Fig. 8 Simulated UV-vis spectrum and vertical transition of 7 (above) and 8 (below). The optimized structure of the species and the expected emission color are also reported.

Also in this case the good agreement between experimental and theoretical data allows us to unambiguously assign the structure reported in Fig. 1 to product 8.

4. Conclusions

In this paper, we have reported a mechanistic theoretical study on the reaction between genipin and methylamine. Starting from the evidences from our previous work on the major isomer of genipin as well as on the opening of the six-membered ring in aqueous solution, the reaction of methylamine with the three forms (open and closed-ring) of genipin in equilibrium has been studied. Results obtained suggest that amine reacts preferentially on the molecule in its open-ring dialdehydic form (2) because of a lower activation enthalpy of the nucleophilic attacks on carbonyl groups (about 18 kcal mol⁻¹) than on the double bond of the closed ring form (1, ΔH^≠ = 45 kcal mol⁻¹). For the following steps of the process, a Maillard type mechanism has been studied and characterized. The catalytic role of water has been highlighted in many steps of the process, until the production of the first yellow intermediate 7. The next step involves conversion of the yellow intermediate into the first red dye 8: a redox self-catalytical cycle has been characterized. This step leads to the production of 8Me, a positively charged intermediate with an activated methylene group, together with 8. 8Me is generally considered being the precursor of all the red and blue dyes produced by the reaction of methylamine with genipin.

From a thermodynamic viewpoint, this cycle shows very low activation barriers (from about 4 to 12.6 kcal mol⁻¹) for all the reactions involved and the endothermicity of the reaction producing 8Me and 8 is not dramatic since both reaction products are quickly consumed in the self-catalytical cycle (8Me) and, as suggested in literature, in the production of further colored species (both of them). Nevertheless, the species inducing the first oxidation of 7 has not been unequivocally determined and the presence of an oxidant is probably necessary to trigger the cycle.

In terms of spectroscopical characterization, the optical and ¹H-NMR properties calculated for the characterized reaction intermediates (namely 7 and 8) are in highly satisfactory agreement with experimental data reported in literature.

Acknowledgements

S. D. T. thanks Dr Ilaria Ciofini and Dr Éric Brémond (Chimie ParisTech) for helpful discussions.

References

1. D. Frick, Rev. Prog. Color. Relat. Top., 2003, 33, 15.
2. O. J. X. Morel and R. M. Christie, Chem. Rev., 2011, 111, 2537.
3. J.-E. Park, J.-Y. Lee, H.-G. Kim, T.-R. Hahn and Y.-S. Paik, J. Agric. Food Chem., 2002, 50, 6511.
4. P. Pfister, A. Meyer and H. Pfander, J. Agric. Food Chem., 1996, 44, 2612.
5. H.-J. Choi, Y.-S. Park, M.-G. Kim, T.-K. Kim, N.-S. Yoon and Y.-J. Lim, Dyes Pigm., 2001, 49, 15.
6. M. F. Butler, Y. F. Ng and P. D. A. Pudney, J. Polym. Sci., Part A: Polym. Chem., 2003, 41, 3941.
7. H.-W. Sung, R.-N. Huang, L. L. H. Huang and C.-C. Tsai, J. Biomater. Sci., Polym. Ed., 1999, 10, 63.
8. F. L. Mi, S. S. Shyu and C. K. Peng, J. Polym. Sci., Part A: Polym. Chem., 2005, 43, 1985.
9. S. W. Lee, J.-M. Lim, S.-H. Bhoo, Y.-S. Paik and T.-R. Hahn, Anal. Chim. Acta, 2003, 480, 267.
10. S. Fujikawa, Tetrahedron Lett., 1987, 28, 4699.
11. S. L. Trevor, M. F. Butler, S. Adams, P. R. Laity, J. C. Burley and R. E. Cameron, Cryst. Growth Des., 2008, 8, 1748.
12. S. Di Tommaso, P. David, M. Gabant, H. David, J.-L. Morançais, J. Gomar and C. Adamo, RSC Adv., 2013, 3, 13764.
13. P. Slusarewicz, K. Zhu and T. Hedman, Nat. Prod. Commun., 2010, 5, 1853.
14. C. Adamo, M. Cossi and V. Barone, J. Comput. Chem., 1997, 18, 1993.
15. F. Duarte, E. Vöhringer-Martinez and A. Toro-Labbe, Phys. Chem. Chem. Phys., 2011, 13, 7773.
16. R. Touyama, K. Inoue, Y. Takeda, M. Yatsuzuka, T. Ikumoto, N. Moritome, T. Shingu, T. Yokoi and H. Inouye, Chem. Pharm. Bull., 1994, 42, 1571.
17. R. Touyama, Y. Takeda, K. Inoue, I. Kawamura, M. Yatsuzuka, T. Ikumoto, T. Shingu, T. Yokoi and H. Inouye, Chem. Pharm. Bull., 1994, 42, 668.
18. C. Adamo and V. Barone, J. Chem. Phys., 1999, 110, 6158.
19. M. Ernzerhof and G. E. Scuseria, J. Chem. Phys., 1999, 110, 5029.
20. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865.
21. K. Fukui, Acc. Chem. Res., 1981, 14, 363.
22. D. Jacquemin, E. A. Perpete, I. Ciofini and C. Adamo, Acc. Chem. Res., 2009, 42, 326.
23. C. Adamo and V. Barone, Chem. Phys. Lett., 2000, 330, 152.
24. D. Jacquemin, V. Wathelet, E. A. Perpete, I. Ciofini and C. Adamo, J. Chem. Theory Comput., 2009, 5, 2420.
25. K. Wolinski, J. F. Hilton and P. Pulay, J. Am. Chem. Soc., 1990, 112, 8251.
26. A. D. Becke, J. Chem. Phys., 1993, 98, 1372.
27. W. Kutzelnigg, U. Fleischer and M. Schindler, NMR Basic Principles and Progress, Springer Verlag, Berlin/Heidelberg, 1991, vol. 213, p. 165.
28. J. Tomasi, B. Mennucci and R. Cammi, Chem. Rev., 2005, 105, 2999.
29. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr, J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski and D. J. Fox, Gaussian 09, Revision A.02, Gaussian, Inc.,Wallingford CT, 2009.
30. G. P. Ellis, Adv. Carbohydr. Chem., 1959, 14, 63.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c3ra47159d

---