3-Aminopropylazetidines: facile synthesis and application for medicinal chemical purposes

The 3-(azetidin-1-yl)propan-1-amine moiety is present in various potentially pharmacologically-active molecules and can be of interest also for the design of metal-complexing agents. In the present study, a new, one-pot protocol using mild conditions has been developed for the straightforward synthesis of various drug-like N - aminopropyl scaffolds. The process combines azetidine dimerization with a subsequent functionalization such as alkylation or amide formation. Analyzing more in detail the first step, the conditions (concentration, catalyst, solvent, temperature) affecting azetidine ring opening and controlled dimerization were investigated.


Introduction
In the course of our ongoing studies on fluorescent dyes and fluorescent turn-on probes for various metal ions, 1- 4 we were interested in the synthesis of novel chemical tools with an aminoquinoline heterocycle as the metalcomplexing scaffold (Scheme 1).Out of the various amine side chains, the 3-(azetidin-1-yl)propan-1-amine moiety is used in a number of potentially pharmacologically-active molecules, described mainly in the medicinal chemistry patent literature.7][8][9][10][11][12][13] Scheme 1. Selected examples of aminoquinoline derivatives used for the synthesis of metal sensors. 4heme 2. Examples of the 3-(azetidin-1-yl)propan-1-amine moiety in medicinal chemistry.
Inspired by the interest of medicinal chemistry in carbo-and heterocycles decorated with this moiety, we planned to exploit a different, straightforward access to compound 7 and its derivatives, based on the dimerization of the more readily available azetidine (1 mmol typically costing 2-3.5 USD). 27On the one hand, a systematic study of the reaction conditions and the scope of the dimerization was carried out, backed up by theoretical considerations.On the other hand, we aimed to test a practical and scalable one-pot procedure for the sequential preparation and functionalization of compound 7, where the dimerization step is followed by a further derivatization to yield various heterocyclic scaffolds.

Results and Discussion
The ring opening of aziridines (N-C σ-bond cleavage), which have a highly strained three-membered ring, is a well-known reaction widely used in synthesis. 28Azetidines, with a four-membered ring, are less strained, and studies of their ring openings are less numerous (the calculated aziridine vs azetidine strain energy is 114.2 kJ/mol vs 105.4 kJ/mol). 29,30Aziridine and azetidine ring openings are of considerable interest for polymer synthesis, giving rise to various linear or branched polyamines via different mechanisms, with potential applications in several fields, e.g., CO2 adsorption and antimicrobial coatings. 31Particularly, but not exclusively, the dimerization of azetidine, i.e., formation of 7 has been described in polymerization studies.Causey et al. identified compound 7 as a side product formed during the hydrogenation of N-benzhydrylazetidine in various organic solvents and subsequent distillation of the azetidine product, besides the formation of additional azetidine oligomers (7 was not isolated). 32Studying the cationic polymerization of azetidine (neat or in MeOH, 80 °C, catalytic HClO4), Schacht and Goethals observed the formation of 7 as the first step in the polymerization proceeding from this dimer.In this case, 7 was isolated by preparative gas chromatography and characterized with MS and 1 H NMR. Regarding the mechanism of the dimerization, first a protonation of azetidine by the acid catalyst was suggested, followed by a nucleophilic attack on the protonated species by another azetidine molecule.From the protonated dimer form, proton transfer to a more basic azetidine molecule keeps the reaction proceeding to full conversion (Scheme 4).Rate constants with different acid-initiator concentrations were determined in MeOH at 70 °C by gas chromatography, and further polymerization from the dimer was studied. 333-(Azetidin-1-yl)propan-1-amine (7) formation from azetidine and subsequent ring-opening polymerization were addressed more recently by Sarazen and Jones as well, calculating the degree of polymerization using 1 H NMR (7 was not isolated). 34heme 4. Formation of 3-(azetidin-1-yl)propan-1-amine (7) via acid-catalyzed dimerization according to Schacht and Goethals (adapted from Sarazen and Jones). 33,34rmation of the dimer as an intermediary step of the thermally-induced (80 °C) polymerization of an azetidine-ZnCl2 complex and azetidine + catalytic HCl was observed by Cherchenko and Abubakirov. 35In the framework of their studies on catalytic alkyl-exchange reactions of secondary amines, Murahashi et al. described the quantitative formation of 7 from azetidine with palladium black catalysis (neat, 140 °C), presumably via a reactive azetine intermediate, and extended the method to the preparation of triamines using azetidine as a source of the 3-aminopropyl group.In this case 7 was isolated by distillation. 36The most detailed experimental and theoretical investigation of nucleophilic ring openings of azetidine, in comparison with aziridine, and oxygen heterocycles oxirane and oxetane, was disclosed by Sharikov et al. 37 The presence of proton donors was found to be essential for the process, and a study of the kinetic parameters showed a correlation between the reactivity of the amine nucleophiles used and their basicity.Ring openings were studied with various primary and secondary amines (azetidine, ammonia, ethylene diamine, piperazine, morpholine, piperidine, propylamine, diethylamine, tert-butylamine, hydrazine and ethanolamine), using GC-MS monitoring and structure elucidation based on mass-fragmentation patterns (7 was not isolated).For synthetic purposes, we were interested in the selective formation of the dimer 3-(azetidin-1-yl)propan-1amine (7) product, i.e., avoiding further polymerization.Therefore, we set out to do a brief study on the effect of the following factors: i) temperature, ii) reaction time, iii) solvent, iv) type and ratio of the initiator, v) concentration -using 1 H NMR spectroscopy to monitor the process.Assessing, first, solutions of azetidine in various deuterated solvents (DMSO-d6, D2O, CD3OD, DMSO-d6/D2O 9/1) without a proton donor present, compound 7 was formed typically only in trace amounts (4-6%) after stirring for 96 h at rt, whereas azetidine remained intact in CDCl3 (Supplementary material, Figures S8, S14, S17, S24).Mild heating (50°C) in the above solvents increased the product ratio (10-30% after 96 h) depending on the nature of the solvent, however, the reaction remained slow, as could be expected given the importance of the protonation step.As acid initiator for further studies, we opted for a strong organic acid: trifluoroacetic acid (TFA, used in 0.5 eq).Apart from CDCl3, where mainly precipitation of the protonated species occurred, and acetone-d6 among the other solvents studied, in the presence of TFA, a considerable amount of the dimer product was obtained already at rt, and reasonable reaction rates were observed upon heating (50 °C) (Figure 1.).These results were in good agreement with the tendencies observed from calculations.The possible effect of water was assessed by running the reaction both in anhydrous DMSO-d6 and in a 9/1 mixture of DMSO-d6/D2O.Of note, in the latter mixture as well as in CD3CN, besides the 3-(azetidin-1-yl)propan-1-amine (7) product formation, side reactions also occurred upon heating.DMSO-d6 provided a slightly slower, but cleaner reaction profile; therefore, it became our solvent of choice for further experiments and developing the one-pot synthetic protocol.However, for isolating 7, the more volatile CH3CN could be a reasonable alternative as well (Experimental section).Modifying the amount of the acid initiator, 1.0 eq TFA led to trace amounts of compound 7, whereas, in the presence of catalytic (0.1 eq) TFA, side reactions increased (Figure 1).Running the experiments at 0.5 -1.0 M azetidine concentration led mainly to compound 7 formation, whereas at 5.0 M, peaks belonging to further products became more prevalent in the 1 H NMR spectra, presumably due to polymerization (Supplementary material, Figure S36).Since it has been described in the literature that the thermal ring cleavage of azetidine proceeds in the presence of ZnCl2, we assessed the effect of a small set of Lewis acids in catalytic amounts (Table 1), finding similar outcomes as those observed with the TFA initiator.For operational reasons (reaction under non-inert conditions, homogenous reaction mixture), and a slightly cleaner outcome however, TFA initiator was used for the synthetic experiments.
As typically longer reaction times were needed for full conversion, we considered using microwave heating, as it is has been often found to lead to shortened reaction times due to a more efficient heating profile. 38oreover, DMSO as a high MW absorbing solvent is an ideal choice also for this direction.Increasing the temperature to 100°C led to decomposition besides product formation, whereas at 75°C, in addition to a faster reaction (77% product after 4 h), side products' formation was less pronounced (Supplementary material, Figure S47-49).We investigated whether the reaction could be extended to larger rings.According to literature observations, the ring opening (N-C σ bond cleavage) is feasible for the 3-and 4-membered rings, but not for larger, sterically less strained homologues (Supplementary material, Figures S50, S54, S57, S60, S64, S66). 37This reactivity pattern was confirmed also by our NMR monitoring experiments (50°C, 72 h -no dimerization observed for 5-8 membered cyclic amines).In the reaction of azetidine with other secondary amines (monocyclic -5-to 8-membered rings -or benzo-fused), however, an aminopropylation could be expected under mild conditions (TFA initiator, DMSO, 50°C).In each case studied, the aminopropylation (i.e., formation of C products) was accompanied by azetidine dimerization as well (Table 2, Supplementary material, Figures S51-69).The formation of 7 could not be ruled out, even using a higher excess (3.0 eq) of the amine partner.For synthetic purposes, a fractional distillation step could be integrated, if necessary, before further functionalization.In the present study, the aminopropylated derivatives listed in Table 2 were not isolated.Alternatively, to rule out the reaction of azetidine with itself, ring opening of an N-substituted derivative, 1-diphenylmethylazetidine, was tested in the presence of pyrrolidine, piperidine and their benzo-condensed analogues.This reaction, followed by the cleavage of the diphenylmethyl protecting group, could offer a selective alternative for the aminopropylation step.Product formation was observed only in the case of the latter entry, albeit affording a modest (15-20%) conversion following even longer reaction times (7 days) (Supplementary material, Figure S74).No further optimization of this reaction was done.
Finally, we have set out to test the synthetic use of in situ azetidine ring opening for the preparation of amine building blocks relevant for medicinal chemistry. 39,40After a first heating session to obtain compound 7, the reagents for the second step (e.g., alkylation, amide formation) were introduced directly into the DMSO solution of 3-(azetidin-1-yl)propan-1-amine (7) without work-up and isolation, affording the expected products in good yields after final chromatography (Scheme 5).As a proof of concept, the one-pot protocol was tested also for the azetidine-pyrrolidine dimerization, using, in this case, the unseparated mixture for further functionalization (Scheme 6).By chromatography analytical samples of the different products could be isolated, however with modest yields, as the second steps of the sequences were not optimized (a uniform 48 h stirring at ambient temperature was used, followed by preparative HPLC separation).Reaction conditions: 20 μL azetidine in 0.6 mL DMSO-d6, 0.5 eq TFA, indicated eq amine B, stirring for 72 h at 50°C; *Composition of the product mixture calculated from 1 H NMR Scheme 5. One-pot aminopropylation followed by a substitution or amide-coupling.Scheme 6. One-pot aminopropylation followed by sulfonamide or amide formation.
In order to explain the various reaction rates of the dimerization of azetidine (A), the reaction mechanism was explored by computational methods at M06-2X/6-31G(d,p) level of theory under different conditions.The dimerization of the base form of azetidine (A) resulted in a quite high activation barrier (H ‡ = 214.9kJ mol -1 , Table 4, Entry 1), which is high enough to block the reaction at the temperature applied (below 100 o C), although the reaction is strongly exothermic.When the attacking azetidine is protonated (A+H + ) while simultaneously acylating the base form of azetidine (A), the activation enthalpy dropped down to 29.4 kJ mol -1 computed in vacuo (Table 3, Entry 1).The greater the relative permittivity of the solvent, the higher the calculated reaction enthalpy of the reaction, which agrees with chemical intuitions.In the case of DMSO, applied in the experimental section, the activation gap exhibited 51.4 kJ mol -1 , which predicts a relatively fast reaction rate.Besides the proton as the strongest Lewis acid (LA), the effect of some weaker LAs (Li + , BF3, AlCl3) were also considered experimentally and theoretically (Table 4).The strongest LA character results in lower activation enthalpies, and AlCl3 is at the forefront in the series (Table 4, Entry 4) after BF3 (Table 4, Entry 3) and Li + (Table 4, Entry 2).These H ‡ values are in agreement with the experimental findings, as LiCl resulted in poor conversion (< 20% in 3 days), while AlCl3 and BF3 exhibited 80% and 70% conversion in 3 days, respectively.In the presence of oxonium ion (H3O + ), the calculated activation enthalpy is practically equal with the value obtained with the protonated form (Table 4, Entry 6).Water provided almost the same value as Entry 1 (Table 4).It is worth studying the same dimerization of different N-containing heterocycles, such as aziridine (Table 5, Entry 1), pyrrolidine (Table 5, Entry 3) and piperidine (Table 5, Entry 4), and comparing them to azetidine (Table 5, Entry 2).As expected, and in line with the computed H ‡ values presented in Table 5, aziridine exhibited relatively low activation enthalpy, while pyrrolidine and piperidine showed higher and higher activation gaps, with nearly negligible reaction enthalpies.According to the results of this study, only aziridine and azetidine can form dimers under mild reaction condition.When we attack the base form of pyrrolidine and piperidine with the protonated azetidine (Table 5, Entries 5 and 7), the activation enthalpies are close to the azetidine dimerization, predicting a smooth reaction.This is in contrast with the opposite situation in which the base form of azetidine is attacked by the protonated form of pyrrolidine and piperidine (Table 5, Entries 6 and 8), as the activation enthalpies represent much higher values.

Experimental Section
General.All reagents and solvents were purchased from commercial sources and were utilized without further purification.Microwave (MW) irradiation experiments were carried out in a monomode CEM-Discover MW reactor, using the standard configuration as delivered, including proprietary software.The experiments were executed in 10 mL MW process vials, with control of the temperature by infrared detection.After completion of the reaction, the vial was cooled to 50 °C by air jet cooling.The 1 H and 13 C NMR spectra were recorded at ambient temperature, in the solvent indicated, with a Varian Mercury Plus spectrometer (Agilent Technologies, Santa Clara, CA, USA) at a frequency of 400 ( 1 H) or 100 MHz ( 13 C) and are reported in parts per million (ppm).Chemical shifts are given on the δ-scale relative to the residual solvent signal as an internal reference.In reporting spectral data, the following abbreviations were used: s = singlet, d = doublet, t = triplet, q=quartet, qn=quintet, m = multiplet, dd = doublet of doublets, dm = doublet of multiplets, tm = triplet of multiplets, and br = broad.For structure elucidation, one-dimensional 1 H, 13 C, DEPT, two-dimensional 1 H-1 H-COSY, 1 H-13 C-HSQC, 1 H- 13 C-HMBC measurements were run.Reactions were monitored by a Shimadzu LC-MS 2020 system.Preparative HPLC was applied for purification in several cases using an Armen SPOT Prep II instrument with UV detector (200-600 nm scan) equipped with a Phenomenex Gemini C18, 250×50.00mm; 10 μm, 110A column.Gradient elution was employed using 0.08 g NH4HCO3 in 1 L water (A) and acetonitrile (B) or 2 mL TFA in 1 L water (A) and acetonitrile (B) as eluent systems, using the gradient method.
Theoretical calculations.Gaussian 16 program package (G16), 41 using default convergence criteria was used, respectively.Computations were carried out at M06-2X/6-31G(d,p) level of theory 42 .The method and basis sets were chosen for their reliability shown in earlier studies. 40The vibrational frequencies were computed at the same levels of theory as used for geometry optimization to properly confirm that all structures reside at minima on their potential energy hypersurfaces (PESs).Thermodynamic functions, such as energy (U), enthalpy (H), Gibbs free energy (G), and entropy (S) were computed for 398.15K, using the quantum chemical, rather than the conventional thermodynamic reference state.NMR monitoring experiments.In a typical experiment, 20 μL (0.30 mmol) azetidine was dissolved in 0.6 mL deuterated solvent.The respective Lewis or Bronsted acid catalyst was added as relevant, and the mixture was stirred at the indicated temperature.At specific time points (typically 0, 24, 48, 72 h), 1 H NMR spectra were recorded, whereas at the end of the experiment, one-dimensional 1 H, 13 C, DEPT, two-dimensional 1 H-1 H-COSY, 1 H-13 C-HSQC, 1 H- 13 C-HMBC measurements were run for structure elucidation.General procedure for the synthesis of 5-/7-amino-2-methylquinolines.333 mg (1.5 mmol, 1.25 eq) 5-or 7bromo-2-methylquinoline was dissolved in 5 mL toluene and 1.20 mmol (1 eq) of 3-(azetidin-1-yl)propan-1amine (137 mg) was added to the solution.The solution was transferred into a closed microwave reaction tube equipped with a stir bar and 216 mg (2.25 mmol, 1.9 eq) NaO t Bu, 47 mg (0.08 mmol, 0.07 eq) rac-BINAP and 28 mg Pddba2 (0.05 mmol, 0.04 eq) were also added to it.The tube was flushed with Argon and the reaction mixture was heated to 120 °C as fast as possible, then stirred at that temperature for 4 hours.After cooling to room temperature, the mixtures were diluted with 15 mL DCM and extracted with water (3 × 15 mL).The organic phase was washed with brine (15 mL) and dried over MgSO4, then concentrated under reduced pressure.The crude product was purified by flash chromatography (hexane/ethyl acetate gradient).Azetidine dimerization on synthetic scale (A).Azetidine (506 μL, 7.5 mmol, 1.0 eq) was dissolved in 15 mL DMSO-d6 (to allow NMR monitoring of the process), followed by the addition of trifluoroacetic acid (287 μL, 3.75 mmol, 0.5 eq).The reaction mixture was stirred at 50 °C for 72 h (3-(azetidin-1-yl)propan-1-amine > 90%, Supplementary material, Figure S11).The obtained DMSO-d6 solution was used directly for the synthetic experiments.

Figure 1 .
Figure 1.Azetidine dimerization: left: in the presence of TFA (0.5 eq.) in various solvents; right: in the presence of varying amounts of TFA initiator and varying temperature.

Table 1 .
The effect of Lewis acid catalysts on azetidine dimerization *Composition of the product in the reaction mixture determined by 1 H NMR

Table 3 .
Solvent effect on the dimerization of azetidine in the presence of a strong acid (TFA), computed at M06-2X/6-31G(d,p) level of theory using PCM solvent model.r = relative permittivity