Synthesis and Biological Evaluation of Novel Hybrid Molecules Containing Purine, Coumarin and Isoxazoline or Isoxazole Moieties

2.1. Chemistry

Some characteristic syntheses and selected data are given below:

2.2. Biology

3.1. Chemistry

The reactions studied and the title new compounds received are depicted in Schemes (1-5). The nitrile oxide, generated from the oxime 2a [18] by chlorination with NCS in DMF solution followed by addition of Et₃N in an one-pot procedure, was treated with the allyloxycoumarin 3a [19] under stirring for 24 h and Ar atmosphere (Scheme 1) to give the isoxazoline 4a in 68% yield (Table 1, entry 1). The isoxazoline 4a has the expected regiochemistry [18] as indicated by HMBC experiments. There is correlation between the protons of NCH₂ group [5.15 ppm (s, 2H)] with the carbon of 4-CH₂ (isoxazoline) (37.5 ppm in ¹³C-NMR) and not with the 5-CH (isoxazoline) (78.6 ppm).

The analogous reaction of the 6-allyloxycoumarin 3a with the nitrile oxide formed in situ from the oxime 2b (prepared by treatment of aldehyde 1b [13] in ethanol-water with NH₂OH.HCl in the presence of anhydrous CH₃COONa at 80°C for 2.5 h) resulted in the isoxazoline 4b in 68% yield (Table 1, entry 2). The 1,3-dipolar cycloaddition reactions of 7-allyloxycoumarin 3b [37, 38] with nitrile oxides resulted from the oximes 2a,b and 2c (synthesized from the aldehyde 1c [13]) gave the isoxazolines 4c,d,e respectively (Table 1, entries 3-5). The one-pot reactions of 4-allyloxycoumarin (3c) [12, 20] with the nitrile oxides of oximes 2a,b led to the isoxazolines 4f,g in 73% and 70% yield respectively (Table 1, entries 6,7). The isoxazolines 4h-j isolated from the reactions of 7-butenyloxycoumarin 3d [12] with the nitrile oxides formed from oximes 2a-c respectively (Table 1, entries 8-10). The isoxazoline 4h has the same regiochemistry, like the others, as the protons of NCH₂ group [5.16 ppm (s, 2H)] in HMBC experiments are correlated with the carbon of 4-CH₂ (isoxazoline) (38.7 ppm in ¹³C-NMR) and not with the 5-CH (isoxazoline) (78.9 ppm). In all the above experiments none of the possible furoxans 5a-c was detected.

In the case of (coumarin-6-yl)acrylate 3e [12] the one-pot procedure with NCS, Et₃N and oxime 2a led to the isoxazoline 4k in only 34% yield along with the furoxan 5a (27%) [18]. When the reaction of acrylate 3e with the oxime 2a was performed with PIDA as oxidizing agent in the presence of catalytic amount of TFA in methanol under stirring for 4 h, the yield for the isoxazoline 4k increased to 71% (Table 1 entry 11). No furoxan 5a was detected. The reactions of acrylates 3e,f with the oximes 2b,c and 2a,b respectively in the presence of PIDA and TFA gave the isoxazolines 4l-o (Table 1, entries 12-15). No furoxans 5b,c were detected by this method.

Another way is presented in Scheme (2) for the synthesis of the hybrid compounds 11a-g using the 1,3-dipolar cycloaddition reactions of nitrile oxides, formed in situ from the (coumarinyloxy)acetaldehyde oximes 9a-d, with the 9-allylpurines 10a-d. The oximes 9a-d (84-88% yields) were prepared for first time from the corresponding substituted acetaldehydes 8a-d [21, 22] by treatment with NH₂OH.HCl in ethanol-water in the presence of anhydrous CH₃COONa at 80° C for 1.5 h. The acetaldehydes 8a-d were in turn prepared in 87-94% yields by refluxing a hydrochloric acid solution of the corresponding acetals 7a-d for 1 h. The acetals 7a-d were synthesized from the hydroxycoumarins 6a-d after heating with 2-bromo-1,1-diethoxyethane and K₂CO₃ in DMF at 90°C for 24 h.

The reaction of 9-allylpurine 10a with the nitrile oxide, resulted from the oxime 9a, was carried out by the above described one-pot procedure with NCS and Et₃N and led to the isoxazoline 11a in 70% yield (Table 2, entry 1). In the case of isoxazoline 11a the coumarin moiety is connected in the 3-position of the isoxazoline ring and the purine moiety in the 5-position, differentiated from the isoxazoline 4a. The regiochemistry of 11a is demonstrated by HMBC experiments as there is correlation of the protons of OCH₂ group with the carbon of 4-CH₂ (isoxazoline) (38.3 ppm in ¹³C-NMR) and not with the 5-CH (isoxazoline) (78.6 ppm).

The isoxazolines 11b,c were isolated in 67% and 70% yields respectively from the reactions of 9-allylpurines 10b,a with the nitrile oxides synthesized from the oximes 9a,b (Table 2, entries 2,3). The analogous reactions of the oxime 9c with the purines 10a-c in the presence of NCS and Et₃N gave the isoxazolines 11d-f (Table 2, entries 4-6) (Scheme 2). The isoxazoline 11g was obtained from the reaction of nitrile oxide, produced from the oxime 9d, with the purine 10a (Table 2, entry 7).

We examined next the reactions of nitrile oxides generated from the oximes 2a-c with the propargyloxycoumarins 12a-d in order to obtain the hybrids 13a-i with isoxazole ring (Scheme 3). The reaction of propargyloxycoumarin 12a [23] with the nitrile oxide, resulted from the oxime 2a under the above described one-pot procedure (NCS, Et₃N), gave the expected product 13a only in 28% yield (Table 3, entry 1). In order to increase the yield of this 1,3-dipolar cycloaddition reaction, we investigated the best reaction conditions using different oxidants and solvents under different temperatures (Table 3).

PIDA as oxidant of oxime 2a, for the formation of the corresponding nitrile oxide in MeOH under r.t. gave a little better yield for the product 13a (37%) (Table 3, entry 2). The method with PIDA and a catalytic [18] amount of TFA was the best with the yield for 13a to increase (64%). By changing the temperature to 0°C or 60°C the yields for 13a were decreased, while the amount of furoxan 5a was increased and the hydrolysis product, aldehyde 1a, was the major product at 60°C (Table 3, entries 4,5). When MeOH/water was used as solvent, the yield of 13a was a little lower (Table 3, entry 6), while the DCM or the THF led to furoxan 5a as the major product (Table 3, entries 7 or 8 respectively). The increase in the amount of TFA or the PIDA gave a larger amount of the aldehyde 1a (Table 3, entries 9 or 10 respectively). The PIFA, which was the oxidant of choice for analogous reactions without solvent [18], led to lower yield of isoxazole 13a (Table 3, entry 11).

After the examination of suitable reaction conditions, the propargyloxycoumarins were reacted with the nitrile oxides generated from the oximes 2a-c in the presence of PIDA and catalytic amount of TFA (Scheme 3, Table 4). The reactions of 4-methyl-6-propargyloxycoumarin (12a) with the oximes 2a,b led to the isoxazoles 13a,b respectively (Table 4, entries 1,2). The isoxazole 13a has the expected regiochemistry [18] as indicated by HMBC experiments. There is a correlation between the protons of NCH₂ group [5.48 ppm (s, 2H)] with the carbon of 4-CH (isoxazole) (103.4 ppm in ¹³C-NMR) and not with the 5-C (isoxazole) (168.8 ppm). The isoxazoles 13c,d were received from the reactions of 4-methyl-7-propargyloxycoumarin (12b) [24] with the oximes 2a,b (Table 4, entries 3,4). The analogous reactions of 7-propargyloxycoumarin (12c) [25] with the oximes 2a-c led to the isoxazoles 13e-g (Table 4, entries 5-7). The isoxazoles 13h,i were isolated from the reactions of 4-propargyloxycoumarin (12d) [26] with the oximes 2a,b (Table 4, entries 8,9).

We studied also the reactions of oxime 9d with the 9-propargylpurines 14a-c (Scheme 4) and the 9-vinylpurines 17a-c (Scheme 5) in the presence of PIDA and catalytic amount of TFA. From the reaction of purine 14a the isoxazole 15a (56%) was isolated along with the dimerization product, furoxan 16 (19%). The regiochemistry of 15a is demonstrated by HMBC experiments as there is correlation of the protons of OCH₂ group with the carbon of 4-CH (isoxazole) (103.1 ppm in ¹³C-NMR) and not with the 5-C (isoxazole) (166.8 ppm). The reactions of purines 14b,c led to the isoxazoles 15b (59%) and 15c (53%) respectively, while the furoxan 16 (19% and 17%) was also formed. The above resulted isoxazoles 15a-c were formed despite the possibility for isomerization of alkynes 14a-c to the corresponding allenes [27].

The 9-vinylpurine 17a [28] reacted with the oxime 9d to give the isoxazoline 18a (66%) (Scheme 5). No furoxan 16 was detected in the reaction mixture. The isoxazoline 18a has the same regiochemistry, like the others, as the protons of OCH₂ group [5.06 ppm (d, 1H)/5.12 ppm (d, 1H)] in HMBC experiments are correlated with the carbon of 4-CH₂ (isoxazoline) (46.9 ppm in ¹³C-NMR) and not with the 5-CH (isoxazoline) (84.4 ppm). The reactions of purines 17b,c gave the isoxazolines 18b (62%) and 18c (60%) respectively, along with the aldehyde 8d (15%).

3.2. Biological Evaluation

The formation of Reactive Oxygen Species (ROS) is a consequence of cell metabolism for aerobic organisms. Due to the extreme reactivity and tendency of ROS to initiate and participate in chain reactions, the role of antioxidants as a defense system is highly recognized. Epidemiological studies revealed the link between reactive oxygen species, inflammation, ischemia and stroke risk. A key strategy to prevent potential damage to cellular compounds such as DNA, proteins and lipids is to reduce the free radical load [29].

The compounds were studied for their antioxidant activity by the use of the stable 2,2-diphenyl-1-picrylhydrazyl radical (DPPH) at concentration 0.1mM after 20 min. A freshly prepared DPPH solution exhibits a deep purple colour with an absorption maximum at 517 nm. This purple colour generally disappears in the presence of an antioxidant. The reduction of absorbance is a measure of the free DPPH due to the action of the antioxidant. The antioxidant activity was expressed as the RA% (Reducing Activity). The RA(%) values for the tested compounds of groups 4, 11, 13, 15 and 18 at 100µM, is very low, if any with the exception of compound 4j presenting 44% (Table 5) in comparison to the reference drug NDGA. Within the group of derivatives 4 the presence of the morpholinyl or piperidinyl rings at position A as well as the carbonyl group at position B, hinder the interaction of the compounds to the free stable radical DPPH. It seems that compound 4j with the combination of pyrrolidinyl ring and (CH₂)₂ chain (instead of a carbonyl group), present the structural features which support the interaction as well as the reducing ability and it does not face any stereochemical hindrance.

Superoxide (O₂^-.) anion and hydroxyl radical (^.OH) are free radical species of potential importance. In the acidic conditions of ischemic brain, O₂^-. is probably protonated to give HO₂^-. species. Iron released from damaged brain cells is more likely to be readily available to catalyze the generation of OH radicals. Among the ROS, the hydroxyl (^•OH) free radical is possibly the most toxic, as it reacts with a number of biological important molecules. Polyunsaturated fatty acids are found in high concentrations in the CNS, and are particularly vulnerable by free radicals. Thus, we tried to test the ability of our compounds to scavenge hydroxyl radicals. The competition of compounds with DMSO for HO^., generated by the Fe³⁺/ascorbic acid system, expressed as percent inhibition of formaldehyde production, was used for the evaluation of their hydroxyl radical scavenging activity In this experiment, the 13f, 13g, 13e, 13h, 15c, 15a, 13b, 13a and 4l showed remarkable activity at 100 μM, with values higher than the well known antioxidant trolox (Table 5). A number of compounds like 4c, 4f, 4j, 4n, 4o, 11b, 11d, 13i,18a, 18b did not present any activity whereas 4d, 4k, 13d and 15d showed lower response. Within the compounds 4a-4o the most potent derivatives 4d and 4j contain a morpholinyl ring in their structure which seems to be correlated with their scavenging activity. All the derivatives of series 13 (except of 13i) present high antioxidant activity (80-100%) which is not able to be correlated with any specific structural characteristic since all contain a coumarin, a purine and an isoxazolyl moiery. This observation is not followed within 15a-c where both 15a and c are almost equipotent whereas 15b the morpholinyl analogue exhibits half of their activity (48%). It seems that the overall molar configuration influences the response. However, antioxidants with hydrophilic or lipophilic character are both needed to act as radical scavengers in the aqueous phase or as chain-breaking antioxidants in biological membranes.

Anti-lipid peroxidation activity. The water-soluble azo compound AAPH has been extensively used as a clean and controllable source of thermally produced alkyl peroxyl free radicals, through spontaneous thermal decomposition. The use of the free radical reactions initiator AAPH is recommended as more appropriate for measuring radical-scavenging activity in vitro, because the activity of the peroxyl radicals produced by the action of AAPH shows a greater similarity to cellular activities such as lipid peroxidation. In the AAPH assay, the highly reactive alkylperoxyl radicals are intercepted mainly by a hydrogen atom transfer (HAT) from the antioxidant. Compounds 13a, 13b, 13e, 15a presented high activity whereas 4a, 4c, 4e, 4f, 11e, 13i, 13h showed 62-86% inhibition of lipid peroxidation. The rest exhibited limited or no activity (Table 5).

LO is the key enzyme in leukotriene biosynthesis [30]. Leukotrienes derived from the biotransformation of arachidonic acid catalyzed by 5-lipoxygenase (5-LO), are important inflammatory mediators [31] implicated in several diseases. LOs play a role in membrane lipid peroxidation by forming hydroperoxides in the lipid bilayer. Inhibitors of LO have attracted attention initially as potential agents for the treatment of inflammatory diseases. Most of the LO inhibitors are antioxidants or free radical scavengers, since lipoxygenation occurs via a carbon-centered radical [32, 33]. The evaluation of the novel coumarin hybrids against soybean lipoxygenase LO was accomplished by the UV-based enzyme assay [34].

Study of LO inhibition values demonstrates that compound 4k provided the best activity (IC₅₀ = 10μM) followed by 4n (35μΜ), 11d (48μΜ), 4f (53μΜ), 13h and 15a (55μM), 4b (60mM), 4l (61mΜ), 13e and 13g (62.5μM), 13i (76μΜ), 18a (90μΜ), 15b and 18b (100μM). It seems that the ester 4k is more interesting and potent hybrid compared to the corresponding ether 4a. However, ester 4l is almost equipotent to ether 4b. Also the presence of a 4-methyl group enhances activity. Thus, 4k is more potent inhibitor compared to the 4n in which the methyl group is missing. The nature of A ring is a structural characteristic of importance. Thus, the piperidinyl derivative 4n is more potent (35μΜ) compared to 4o in which a morpholinyl group has replaced the piperidinyl group. The presence of pyrrolidinyl ring in compounds 4g (10%), 11g (no), 13g (62.5 μΜ) is correlated with no or low activity.

Within the compounds of 11a-g subgroup no inhibitory activities were observed. The only exception was compound 11d. Again a piperidinyl 6-substituted coumarin derivative was found to be more potent (IC₅₀ = 48μM).

Among the isoxazole coumarin derivatives 13a-i most interesting results were given by 13h (55 μM), 13e (62.5μM), 13g (62.5 μM), and 13i (76 μM) (13h>13g, 13e>13i). The most potent 13h was a piperidinyl substituted derivative, whereas the replacement by a morpholinyl ring (13i) led to a decrease. Thus, the nature of the ring was implicated in the biological response. The equimolar response of 13e and 13i support the idea that the nature of A ring did not affect the inhibition.

Changing the attachment position of the isoxazole ring the analogues 15a-c were taken, from which 15a was more potent (55 μM) followed by 15b (100 μM) and 15c (9%). These findings follow the previous one supporting the significant role of the piperidinyl group. Also the piperidinyl derivative 18a was found slightly more active than the 18b.

Herein, the antilipid peroxidation activity does not go in parallel to the anti-LO activity (Table 5). Hydroxyl scavenging activity also was not found to be correlated with the above responses.

Considering the interesting results shown in Table 5 that clearly confirm the antioxidant potential of some of the new hybrid derivatives, we found interesting to test them as ChE inhibitors. Using already described protocols for the determination of the AChE [35] inhibition we obtained the IC₅₀ values shown in Table 5. In Table 5 are only given the IC₅₀ values for AChE inhibition of the more active compounds as well as the % inhibition for BChE. For comparative purposes some reference molecules have been incorporated.

Regarding the AChE inhibition (Table 5), hybrids 4f>13e>4b>13b>13i>11b>11e>4i showed significant eeAChE inhibition activity and they all are ethers of coumarin. The 4-, 6, or 7- position of attachment does not influence the result. For esBChE, most of the hybrids were poor inhibitors, with the exception of 13e, one of the two most potent AChE inhibitors (4f and 13e). All tested hybrids were less potent compared to the reference compound galantamine, but they retain fair AChE inhibitory activities in the low micromolar range.

For the MAO-A and B inhibition, we choose a subgroup consisted of the active anti-AChE compounds (Table 6). Only compound 11b showed an interesting inhibition activity IC₅₀ = 9.5 μΜ against MAO-B acting as a selective agent. All the others present low (%) or not any activity at 10 μΜ. Hybrid 11b showed very lower inhibitory potency compared with the reference clorgyline, a well-established MAO-A selective inhibitor used in the treatment of depression. Indeed, the interesting MAO-B selectivity can be considered a good starting point for a structure-based refinement, in view of the potential of MAO-B selective inhibitors as neuroprotective agents in the therapy of neurodegenerative diseases [36].

Considering the MAO inhibition data, we conclude that hybrid 11b is a moderate, but selective inhibitor. Hybrids 4b, 4f, 4i, 11b, 11e, 13b, 13i are potent and selective inhibitors for AChE, whereas 13e is a dual AChE and BChE inhibitor.