2. Results and Discussion
In our first experiments, 6-hydroxyquinoline (
1) was reacted with 1-naphthaldehyde (
2) and
N-benzylmethylamine (
3) under neat conditions (
Scheme 1). Of the tested temperatures (60 °C, 80 °C and 100 °C), 60 °C was found to be optimal, which was provided by classical heating (an oil bath). After 8 h, in the course of the workup of the reaction mixture, the crude product was subjected to column chromatography. Subsequent crystallisation from methanol gave a mixture with hydrate
5 as the main component, which is the stabilised form of
ortho-quinone methide
4 as disclosed by combined NMR methods. An analogous product was isolated in our previous work when 2-naphthol was attempted to be aminoalkylated with
N-benzylmethylamine in the presence of 1-naphthaldehyde [
12]. In order to evaluate the role of the amines, morpholine (
6) as a secondary cyclic amine was reacted with 6-hydroxyquinoline and 1-naphthaldehyde under conditions that were previously optimised (
Scheme 1). Thin-layer chromatography (TLC) confirmed the formation of a single product, which was then isolated by crystallisation. The NMR spectra provided evidence that the use of morpholine as an amine component led to the formation of the classical Mannich product
7. The differences between the reaction pathways can be explained in terms of the different steric demands and the basicity of the applied amines (pK
b = 4.25 and 5.64 for
3 and
6, respectively). Since
3 is bulkier and at least tenfold more basic than
6, it can be more prone to activate water molecules rather than to attack as a nucleophile on the potential electrophilic species present in the reaction mixture.
In order to explore further structure–reactivity relationships, 3-hydoxyisoquinoline (
8), a
N-containing electron-rich aromatic substrate, was selected for the subsequent studies. Accordingly, 3-hydroxyisoquinoline (
8), 1-naphthaldehyde (
2) and
N-benzylmethylamine (
3) were reacted under neat conditions. After 30 minutes of reaction time at 60 °C, the prepared TLC was multi-spotted, and no well-defined component could be isolated from the reaction mixture. Thereafter,
N-benzylmethylamine was switched to morpholine. A prolonged reaction time (6 h) at 80 °C was necessary for the appearance of a few new spots on the TLC. By means of column chromatography, using EtOAc:MeOH (20:1) as an eluent, we isolated methoxy-substituted lactam
10 as a distinct compound. In the
1H NMR spectrum of
10, the typical singlet of 3H intensity at 3.35 ppm and the amide-type NH doublet at 9.25 ppm along with
1H-
13C-HMBC connectivity between signal pairs OC
H3/C-1, H-1/O
CH
3, H-1/C-3, NH/C-3 and NH/C-4 unambiguously proved its structure. The formation of this product can be explained via the formation of the
ortho-quinone methide (
9), which reacted at the most electrophilic
N-acylimine site with methanol present in the solvent mixture used in the chromatographic purification (
Scheme 2).
To investigate the influence of the arylidene moiety on the apparently unexpected nucleophilic attack of MeOH at position 1, 3-hydroxyisoquinoline (8) and methanol were stirred at 80 °C. Since, even after a prolonged reaction time (20 h), TLC indicated only the presence of the starting compounds, the reaction was repeated at 100 °C and 120 °C. Because mixing 8 with MeOH at a higher temperature resulted again in the presence of the unchanged starting compounds, we focused our attention on the evaluation of the effects of stronger nucleophiles, such as N-benzylmethylamine and morpholine, on the outcome of the reactions. First, N-benzylmethylamine (3) and 3-hydroxyisoquinoline (8) were reacted at 80 °C under neat conditions. Surprisingly, after 12 h of reaction time, TLC showed the formation of a multicomponent mixture.
After purification by column chromatography, two components with characteristic compositions (
11a and
11b:
Scheme 3) could be isolated and identified by combined NMR methods, including the use of highly diagnostic
1H-
13C-HMBC. The apparently intriguing transformations can be rationalised by the equilibrium formation of tautomer
12, the key intermediate of the subsequent transformations. On one hand, the nucleophilic attack of tautomer
8 on the highly-electrophilic C-1 centre of this activated
N-acyl-imine results in the formation of the racemic mixture of the chiral dimer
11a. On the other hand, upon nucleophilic attack of
3, tautomer
12 is converted into adduct
13. The latter undergoes a synchronous imine- and dihydrogen-forming fragmentation, leading to Schiff base
14 with the simultaneous regeneration of
12. Hydrolysis of
14 gives benzaldehyde
15 and methylamine. Upon condensation with
15, saturated aminal adduct
13 is transformed into
16. This enone-containing aminal intermediate then undergoes sequential addition and elimination of
3, eventually affording
11b, the other isolated aromatic product. Finally, either of the two equilibrating tautomers of the heterocyclic precursor (
8 or
12) might also react with
15 to construct enone
17. The
Aza-Michael addition of the latter with
3 also leads to
11b.
Since
N-benzylmethylamine was expected to undergo uncontrolled decomposition, leading to a wide range of side products, morpholine was applied in our subsequent experiments. This stable secondary cyclic amine was stirred with 3-hydroxyisoquinoline at 80 °C under neat conditions. After 2 h, the spot indicating 3-hydroxyisoquinoline disappeared on a TLC; therefore, the mixture was worked up by neutral column chromatography to produce aminal
18 as a single isolated product (
Scheme 4). In its
1H-NMR spectrum, the separated AB signals (
J = 20 Hz) of the diastereotopic H-4 protons discernible at 3.37 ppm and 3.64 ppm indicated its lactame-type skeletal structure with a stereogenic centre at position 1. This structure was also supported by the chemical shift of C-1 (171.4 ppm) and by HMBC cross peaks revealing correlations between
1H/
13C coupled pairs H-4A/C-1, H-4B/C-1, H-1/C-1 and N
H/C-1 (
Scheme 4).
In order to unambiguously clarify the role of the amine and the aldehyde in this reaction and to avoid the rather unpredictable reactivity and decomposition of
N-benzylmethylamine, 3-hydroxyisoquinoline (
8) was reacted with morpholine (
6) in the presence of different aromatic aldehydes under neat conditions. Since the incorporation of the methoxy group in the case of
10 was also undesirable, the composition of the crude reaction mixtures was analysed using
1H NMR without any further purification. In accord with our previous findings with morpholine as the amine nucleophile, in addition to
18, we detected two types of products (
23A and
23B) in a time-dependent manner. Thus, the crude products were analysed at five different reaction times (1 h, 2 h, 4 h, 8 h and 16 h). In general, the first-appearing NMR signals can unambiguously be assigned to the protons of
18. In the progress of the experiments, the ratio of
18 was gradually decreased, while the ratios of the other products were increased. As a representative example, the transformation of
23d is demonstrated in
Figure 2.
The ratio of the regioisomers was obtained as the relative intensity of the diagnostic proton signals of particular components. Namely, a singlet in the range of 5.3–5.5 ppm is characteristic for Mannich adducts type
23A, and a doublet in the range of 5.0–5.1 ppm is diagnostic for enone aminals of type
23B (
Table 1). The relative amount of
18 could also be calculated by the intensity of the characteristic proton signals at 5.03 ppm and 8.43 ppm.
The
1H NMR analyses revealed the parallel formation of the classical Mannich product (
23b–
dA) and the corresponding arylidene derivative (
23b–
dB) when benzaldehyde (
15), 2-naphthaldehyde (
19) and 4-methoxybenzaldehyde (
20) were applied as the carbonyl component (
Figure 3). However, independently of the reaction time, when 1-naphthaldehyde (
2) was used as a coupling partner, the reactions led to the formation of the arylidene derivative
23aB, which could be isolated as a single product.
It must be pointed out that enones
23b–
dB were slowly transformed into the appropriate Mannich-type product
23b–
dA as highlighted by the
1H NMR spectra of the reaction mixtures that were registered at different times (
Figure 3).
In order to rationalise all the aforementioned aryl-group-dependent experimental findings as well as the exceptional reactivity of 4-nitrobenzaldehyde
26 manifested in its analogue transformation to attempt to obtain the Mannich-type product (discussed later), we undertook comparative DFT modelling studies on the aryl-group-dependent progress of the formation and possible interconversion of
18 and compound types
23A and
23B. We hypothesised that these transformations might take place along the pathways outlined on
Scheme 5. The feasibility of this assumption was assessed by the relative energetics of particular elementary steps and by the analysis of the acceptor orbitals (LUMO and LUMO + 1) of the possible electrophilic species involved in these Mannich-type and related conversions (
Figure 4).
First, the classical Mannich products were expected to be formed through an ion pair incorporating iminium cations
21a–
d and a heterocyclic anion
22 generated by the nucleophilic attack of
6 on the aldehyde component followed by the 3-hydroxyisoquinoline-promoted elimination of water. However, on the basis of the assumed relatively small amount of the ion pair
21b–
d/22 [cf., the changes in Gibbs free energy (ΔG) accompanying their formation by implication of either tautomers
8 or
12, as presented on
Scheme 6] and the dominance of
18 in the reaction mixtures obtained after 1 h, the share of these pathways was negligible in the construction of compounds type
21A. On the other hand, the rapid and dominant primary formation of
18 could be ascribed to the pronounced electrophilicity of
12, mainly ascribed to its low-energy LUMO, with a significant share on the C-1 atom. Accordingly, except for 4-nitrobenzaldehyde
26, the other arylaldehyde components had markedly higher LUMO energy relative to that of
12. This oxo-tautomer could be regarded as the main precursor of
18, even though the calculated ΔG value suggested that its tautomer
8 must be present in the reaction mixture in a substantially higher concentration. In the subsequent steps,
18 could have undergone condensation with the appropriate aldehyde to construct arylidenes
23B, or, with the involvement of
6 generating iminium ions
21a–
d, it could have been converted into
bis-morpholine intermediates type
24. Morpholine elimination from the aminal residue of
24 gave the corresponding Mannich product (
23A). It is of note that due to the thermodynamically unfavoured formation of
21a–
d under slightly acidic conditions (cf., data presented in
Scheme 6), this condensation–elimination sequence can be considered as a less feasible pathway. Since the relative energetic data calculated for all investigated isomer pairs
23A/
23B [(ΔG(
23B-
23A), see
Scheme 5] suggested that
23a–
dA are formed under thermodynamic control, the isomerisation of
23B into
23A can also be taken into account as a realistic process taking place under the applied conditions. In this regard, the calculated energetic data [(ΔG(
25+
6-
23B)] practically rule out the elimination–addition sequence
23B→
25→
23A as a realistic pathway because the competitive addition–elimination sequence proceeding via
bis-morpholine intermediates
24 seems to be more feasible as indicated again by the relative energetics calculated for the critical addition step [(ΔG(
24-
6-
23B)].
In this context, the relatively increased reluctance of 23aB to undergo isomerisation into 23aA seems to be the consequence of the pronounced endoergic addition step 23aB + 6→24a, which might be associated not only with its enhanced electron-donating character, but also with the steric bulk of the 1-naphthyl group.
The reaction of
8 and
6 with
p-nitrobenzaldehyde (
26, [
14,
15,
16]) was an exceptional case, because neither the expected Mannich product
23eA nor the appropriate benzylidene product
23eB could be isolated. Instead, 4,4′-((4-nitrophenyl)methylene)dimorpholine (
27) was formed as the main product (
Scheme 6). This unique conversion of
26, proceeding without the implication of equilibrating heterocyclic tautomers (
8 and
12), can be interpreted by its LUMO energy being markedly lower than that of
12 (
Figure 4). Furthermore, the relatively low-energy LUMO+1, also available to nucleophilic attack with significant share on the carbonyl group, further attenuates the electrophilicity of this reactive aldehyde. On the other hand, the tendency in the calculated energetic data—in accord with the general qualitative expectations—clearly shows that 3-hydroxyisoquinoline-promoted generation of the ion pair
21e/
22 is a less favoured process, than the formation of the ion pairs
21a–
d/22 containing a positively charged nitrogen atom without a strong electron-withdrawing substituent on the aryl group. It must be pointed out here that the readiness of iminium generation is strongly dependent on the proton source as shown by the energetic data obtained by replacing
8 with a hydroxonium ion in the calculations. The extreme difference between the two series of energetic data, otherwise both demonstrating the same tendency in the function of the electronic character of the aryl groups, can be considered as a strong indication of the dramatic dependence of the thermochemistry of ion-pair formation on a number of factors that might influence the stability of the charged particles, including, e.g., coulombic interaction, solvation and other intermolecular contacts. Avoiding unrealistically demanding computations, except for the estimated polarity of the reaction mixture (cf., description of the computational model in the “Materials and Methods” section), the aforementioned factors were neglected in the course of our DFT modelling studies.
Finally, we proposed a mechanism for the formation of
27, accounting for taking place without the implication of
8. Thus, according to our assumption, reductive elimination of the elements of the water molecule from the selectively generated adduct
28e gives carbene
29e stabilised by the 4-nitrophenyl group as represented by resonance hybrids
29e/I and
29e/II, which then reacts with
6 in an oxidative addition step affording
27 as the single isolable product (
Scheme 6). In this sequence, carbene formation is the critical step with indispensable assistance of the nitrophenyl group. This view was supported by the comparison of the relative energetics of water elimination from adducts
28a–
e (
Scheme 6). The calculated data [(ΔG(
29+H
2O-
28)] unambiguously indicated that the process
28e→
29e+H
2O is significantly less endoergic, and, consequently, it is more feasible than the analogous reaction steps
28a–
d→
29a–
d+H
2O. It must be emphasised again, that it is the tendency of the relative energetics that must be regarded to have diagnostic value in the assessment of the substituent effect, as solvation of the involved species along with other stabilizing and destabilizing interactions—which were not modelled in the calculations—are also expected to determine the real thermochemistry of the studied conversions. It is also of note that carbene
29e reacts with
6 rather than with
8. This substrate selectivity might be associated with the enhanced nucleophilicity of
6 compared to the neutral
8 of which the deprotonation-generating ion pair
21e/
22 with the highly reactive anion
22, a real competitor of
6, is suppressed by the nitrophenyl group as discussed above.