Improved Methods for Thermal Rearrangement of Alicyclic α-Hydroxyimines to α-Aminoketones: Synthesis of Ketamine Analogues as Antisepsis Candidates

Ketamine is an analgesic/anesthetic drug, which, in combination with other drugs, has been used as anesthetic for over 40 years. Ketamine induces its analgesic activities by blocking the N-methyl-D-aspartate (NMDA) receptor in the central nervous system (CNS). We have reported that low doses of ketamine administrated to patients before incision significantly reduced post-operative inflammation as reflected by reduced interleukin-6 (IL-6) sera-levels. Our data demonstrated in a rat model of Gram-negative bacterial-sepsis that if we inject a low dose of ketamine following bacterial inoculation we reduce mortality from approximately 75% to 25%. Similar to what we have observed in operated patients, the levels of TNF-α and IL-6 in ketamine-treated rats were significantly lower than in septic animals not treated with ketamine. On the base of these results, we have designed and synthesized series of new analogues of ketamine applying a thermal rearrangement of alicyclic α-hydroxyimines to α-aminoketones in parallel arrays. One of the analogues (compound 6e) displayed high activity in down-regulating the levels of IL-6 and TNF-α in vivo as compared to ketamine.


Introduction
Uncontrolled inflammation and immune response lie at the heart of bacterial sepsis. Sepsis is a major complication among post-operative and trauma patients which leads to multi-organ dysfunction, multisystem failure and eventually to death in up to 60% of septic patients in chirurgical intensive care units [1]. The septic response is usually triggered when microorganisms spread from the gastrointestinal tract, skin or lung into contiguous tissues and blood. Animals recognize certain microbial molecules such as lipopolysaccharide (LPS) and rapidly produce various inflammatory mediators including cytokines, such as tumor necrosis factor (TNF) and interleukin-6 (IL-6), that amplify the LPS signal and transmit it to other cells and tissues [2,3]. Inflammatory cytokines amplify and diversify the response. These proteins can exert endocrine, paracrine, and autocrine effects. Blood levels of TNF and IL-6 are high in most patients with severe sepsis. Moreover, intravenous infusion of TNF can elicit many of the characteristic abnormalities of sepsis [4].
Current anti-inflammatory pharmacological products such as non-steroidal anti-inflammatory drugs, corticosteroids, antibiotics and anti-cytokines agents have limited efficacy and/or an inadequate safety profiles. Therefore, an intensive effort is currently being made by the scientific community to develop new anti-inflammatory drugs. Ketamine ( Figure 1) is an analgesic/anesthetic drug, which, in combination with other drugs, has been used as anesthetic for over 40 years. Ketamine induces its analgesic activities by blocking the N-methyl-D-aspartate (NMDA) receptor in the central nervous system (CNS) [5]. Studies involving the administration of ketamine to patients and animal have demonstrated an anti-inflammatory activity at a plasma concentration of 0.1-1 µg/mL [6,7]. We have previously shown that ketamine at this concentration had no direct effect on LPS-induced cytokine secretion from leukocytes. Despite these findings, various in vitro studies have examined the anti-inflammatory activity of ketamine at concentrations 10-to 1000-fold larger [8][9][10]. We found that at these elevated concentrations ketamine exerts nonspecific cytostatic effects, arrest of cell proliferation, and blockade of cytokine production [11]. Thus, direct inhibitory effect on toll like receptor (TLR) -4 which mediate the effect of LPS is excluded since we found the anti-inflammatory effect of ketamine only "in-vivo". Adenosine mediates the anti-inflammatory effect of ketamine in vivo. We previously found that ketamine induced a significant increase in plasma and peritoneal concentrations of the nucleoside adenosine, which occurs within 20-30 min after ketamine administration [12]. Adenosine is an important modulator of inflammation. Its levels rise during ischemia, hypoxia, inflammation and trauma and it exhibits anti-inflammatory effects through A 2A receptor (A 2A R) [13]. We have suggested that the anti-inflammatory effect of ketamine were mediated by adenosine since we found that its effects were blocked by an A 2A R specific antagonist and were mimicked by an A 2A adenosine receptor agonist [10][11][12]. Since ketamine in normal pharmacological doses has no activity in vitro we were obligated to use an animal based assay to test the new compounds we have developed.
Ketamine, a common anesthetic compound used in surgery, is synthesized using a tedious multistep procedure of six consecutives reactions [14][15][16][17], including two thermal rearrangements. Obviously, this classical procedure is not adapted to the synthesis of libraries of ketamine analogues. Herein we propose methods for the fast generation of ketamine analogue libraries in moderate yields.

Chemistry
Our approach is described in Scheme 1. Intermediate mono-acetal 1a can be easily obtained after partial conversion of one ketone function in 1, 2-dioxocyclohexane into an acetal using ethylene glycol. Further reaction with appropriate Grignard reagents R 1 MgBr generate alcohols 2 and 2, which upon acetal hydrolysis are converted into hydroxyketones 3 and 3. The key-steps in this approach are, firstly, the generation of imines (4a-g, 5a-g) and secondly, their rearrangement to mixtures of six-member ring αaminoketones (6a-g, 7a-g) directly related to ketamine, and five-member ring αaminoketones (6a-g, 7a-g), of potential interest as ketamine isomers bearing the same pharmacophores, but with different structural constrains. A methodology has been developed for obtaining products 6[a-g], 7[a-g], 6'[a-g] and 7'[a-g] in a one pot reaction using both classical or microwave heating of mixtures of the corresponding primary amines R 2 -NH 2 and the different hydroxyalcohols 3a,b. In prior works, this type of process needed high temperatures (200 °C) and long reaction times (10-24 h) that were also employed here.

6[a-g] 7[a-g]
This type of thermal rearrangement has been studied for almost half century first by Stevens [18][19][20][21][22][23][24][25][26] and later on by Compain [27][28][29][30], however the conditions needed for the reactions were not improved in those works. Interestingly, our starting compounds 4[a-g], 5[a-g] for the thermal rearrangement were hypothesized (but not isolated) by Stevens (see Figures 2,4 and 5) as being the intermediates between the starting five member ring A and the product six member ring C isomers in his studies. Compain, who also started with six member ring imines (see a in Figure 3), obtained only six membered ring products after rearrangement of the exo-propargylic (a in Figure 3) or allylic (b in Figure 3) systems, with no trace of the endo-migration that generates five member ring products previously observed by Stevens (see A in Figure 2).
Our results demonstrate experimentally for the first time that the imine intermediates 4-5 ( Figure 2) hypothesized (but not observed or isolated) by Stevens as being the intermediate for interconverting A into C and D, are indeed the intermediates between the six and five member rings in the thermal rearrangement.
Two parallel libraries of analogues were synthesized in a Radley's parallel reactor using classical heating. The first library with R1 = phenyl 6[a-g] for the six member ring analogues and 6'[a-g] for the five member ring analogues, and the second R 1 = methyl 7[a-g] for the six member ring analogues and 7'[a-g] for the five member ring analogues, were generated using a panel of aliphatic and aromatic primary amines R 2 NH 2 ( Table 1).
Overall yields of the product mixtures ranged from low for aromatic amines to good for aliphatic amines. In general, yields were higher for R 1 = phenyl than for the R 1 = methyl. The ratio of isomers was favorable to the 6-membered ring adduct for most of the cases where R 1 = phenyl, however, for R 1 = methyl the ratio was rather favorable to the 5-membered ring adducts. Isomers were easily isolated using preparative HPLC and fully characterized using a panel of NMR methods and HR-MS and the isolated quantities were large enough for performing substantial biological assays. Further experiments aimed at improving the reaction conditions (time, temperature and yields) were conducted using microwave assisted reactions. A reduced model was chosen using R 2 = phenyl and a panel of four different R 1 -NH 2 for which the highest or lowest yields were obtained using the classical thermal conditions (Table 2). To our surprise we could obtain separable mixtures of compounds 6 and 6' after short reactions. The products were easily separated by preparative HPLC. Interestingly, yields varied for the different amines, while using classical heating method an overall yield of 48% was observed for 6a/6'a mixtures, the use of microwave improved the yield to 90% with a different distribution of isomers. On the other hand, hexyl-(6b, 6'b) and propyl-(6c, 6'c) amine that gave relatively high yields using classical heating, led to reduced yields with microwave energy. Finally, while reacting aniline (6f and 6'f), yields were similarly low for both methods, but in contrast to classical heating that resulted in 1:1 isomer mixtures, the microwave reaction led exclusively to isomer 6'f. Table 3 summarizes the comparative yields obtained by the thermal vs. microwave heating methods. A peculiar side compound 8 of synthetic interest has been isolated from the microwave assisted reaction using benzylamine in 8% yield (see Table 3, [6a-6'a]). This compound appears to be the result of aromatization of the cyclohexane ring followed by the thermal migration (rearrangement) of the N-benzyl group to the para position in the generated aromatic system in a similar way to the previously reported photo- [31,32] and microwave- [33] assisted rearrangement of N-alkyl anilines (Scheme 2). Compound 8 was unambiguously characterized by NMR and HR-MS. o-p-Disubstituted anilines similar to 8 are interesting scaffolds in medicinal chemistry, thus we are currently developing a synthetic methodology for the microwave mediated thermal rearrangement of o-substituted N-alkyl anilines for generating o-p-di substituted anilines.
We conclude that the use of microwaves for the presented thermal rearrangement is advantageous in terms of reaction time (minutes instead of overnight) but not always positive in terms of yield. Strikingly, when using aniline as precursor, classical thermal heating generates a mixture of 6f/6'f (1:1) while the microwave assisted reaction generated exclusively isomer 6'f. On the base of the works of Stevens and Compain, we speculate that the 5-membered ring is the kinetic product, while the 6-membered ring is the thermodynamic product: when using the classical heating method, equilibrium between both isomers is achieved, however, when performing the reaction for a short time, this equilibrium is not always reached (note that for all the reactions in Table 3 the 6:5 membered ring ratio diminishes favoring the 5-membered ring products when using microwave conditions). Aniline seems to be an extreme case where only the "kinetic" five member ring product is observed. Other mechanistic considerations of this rearrangement will be discussed elsewhere. Finally when using benzylamine as precursor (Table 3, [6a-6'a]), a reduced yield of the products is acknowledged accompanied by the presence of side product 8 in 8% yield generated by a tandem eliminationoxidative aromatization of the cyclohexane ring followed by another type of thermal rearrangement of the N-benzyl group from the nitrogen to the p-position of the generated aromatic ring.

Biological Assays: Inhibitory Effect on IL-6 and TNF-Secretion
All the compounds were screened for their inhibitory effect on IL-6 and TNF-α secretion in lavage as compared to saline controls and ketamine in vivo. We report here results obtained for the active compound 6e (see Table 4). Treatment with ketamine significantly suppressed the production of IL-6 (p = 0.0450) and TNF-α (p = 0.0158) in lavage after bacterial challenge as compared to saline group. Table 4. Inhibitory effect of compound 6e on IL-6 and TNF-α secretion in an animal model of sepsis. Compound was injected at 10 mg/kg prior to sepsis (See experimental section for details). Compound 6e showed a significant inhibitory effect on IL-6 secretion in lavage as compared to saline group (p = 0.0046). Similarly to the IL-6 results observed in lavage, treatment with compound 6e significantly down-regulated TNF-α levels as compared to saline mice (p = 0.0069). Overall compound 6e with inhibitory effect higher than ketamine for IL-6 and TNF-α secretion appears to be the best candidate for further studies. In addition to cytokines levels, mice behavior was visually monitored following E. coli and compound injection. Sixteen hours after bacterial inoculation mice showed clinical signs of illness such as hunched posture, ruffled fur and reduced activity. Mice treated with ketamine were significantly stronger and more active as compared to mice treated with saline. In concordance to the reduced levels of IL-6 and TNF-α in lavage of 6e compound-injected mice, these mice showed a similar physical response to the bacterial challenge as the ketamine mice. Overall, we have found that compound 6e (p = 0.005) has improved inhibitory effect on secretion of IL-6 and TNF-α as compared to ketamine. Moreover, animals treated with this compound were significantly stronger and more active as compared to animals treated with saline.

General Methods
Reagents, unless otherwise mentioned, were purchased from Aldrich and used without further purification. 1,2-Cyclohexanedione was purchased from Acros, Tetrahydrofuran (THF) and diethyl ether (Et 2 O) were distilled from sodium/benzophenone under argon at atmospheric pressure immediately prior to use. Toluene was distilled from calcium hydride under an inert atmosphere and stored over sodium. All other solvents were analytically pure grade and were used without further purification. Analytical and preparative HPLC were performed on a Waters HPLC system equipped with a 717-Plus autosampler, a 600-controller pump, a 996-photodiode array detector and a Gilson 202 fraction collector; the system was operated with the Millennium software (Waters). Microwave irradiations were carried out with a professional "Initiator" microwave from Biotage at pre-fixed temperature. The intensity differed from 0 to 300W at frequency of 2.456 GHz. 1 H and 13 C-NMR spectra were recorded on Bruker DPX-300 and Avance DMX-600 spectrometers. 1 H-NMR data was obtained by 2D, COSY and HOHAHA methods (t = 40 ms). 13 C-NMR data was obtained by 2D techniques, NOESY, HMQC and HMBC methods with a delay of 3.45 60 ms respectively in the reverse mode. Chemical shifts are in ppm relative to TMS internal standard or relative to the residual solvent resonance.
Mass spectra analyses were recorded on an AUTOSPEC-FISSONS VG (Micromass) high-resolution mass spectrometer under DCI (desorption chemical ionization) conditions (CH 4 ) and by ESI (electron spray ionization) mass spectrometry on a Q-TOF (quadropole time of flight) low-resolution micromass spectrometer (Micromass-Waters, Corp.). Microwave reactions were performed using a mono-mode Initiator station from Biotage. Purity of compounds was over 95% as assessed by two different HPLC conditions, QTOF-MS-MS was used for assessing purity of the molecular peaks.

Synthesis of Intermediates
Preparation of 1,4-dioxaspiro [4.5]decan-6-one (1): A solution of 1,2-cyclohexanedione (10 g, 0.089 mol) in toluene (300 mL), an equimolar amount of ethylene glycol (5 mL, 0.089 mol) and p-toluenesulfonic acid (100 mg) were heated at reflux for 8 h in a Dean-Stark apparatus (water was continuously separated). The resulting solution was washed twice with 1 N NaOH solution, dried over MgSO 4 and concentrated under reduced pressure to afford 9.61 g (0.04 mol, 46% yield) of the crude yellow oily product. The crude product containing a certain amount of the diacetal 1 (2:1 for the monoketone, according to NMR), was used without further purification in the next step. 1

General Procedure for the Parallel Synthesis of Library Compounds 6a-g, 6a-g
To seven pressure-tubes each containing 2-hydroxy-2-phenylcyclohexanone (3, 190 mg, 1 mmol) was added the appropriate primary amine (benzylamine, hexylamine, propylamine, isopropylamine, butylamine, aniline, 4-chloroaniline, 1.15 mmol). The tubes were flushed with argon, sealed and heated at 200 °C overnight in a Radley's combinatorial station with stirring. The residues were diluted with ether and extracted with 0.1 M HCl. The acidic phase washed with ether and concentrated and the combined ether phase was dried over MgSO 4 and concentrated. Preparative HPLC conditions were selected among methods D-F, according to the elution times obtained by analytical HPLC using method A.

Characterization Data for Compounds 6a-g, 6a-g
For the NMR data numbering system of compounds 6a-g, 6a-g please refer to Table 5.  Synthesis of 6a-6a: benzylamine was used as amine component in the reaction. The mixture was purified by preparative HPLC using method D to give the desired products.  Synthesis of 6b-6b: Hexylamine was used as amine component in the reaction. The mixture was purified by preparative HPLC using method D to give the desired products.  Synthesis of [6c-6′c]: n-Propylamine was used as amine component in the reaction. The mixture was purified by preparative HPLC using method D to give the desired products.   Synthesis of [6e-6′e]: n-Butylamine was used as amine component in the reaction. The mixture was purified by preparative HPLC using method D to give the desired products.   Synthesis of [6g-6′g]: 4-Chloroaniline was used as amine component in the reaction. The mixture was purified by preparative HPLC using method F to give the desired products.

General Procedure for the Parallel Synthesis of Library Compounds 7a-g, 7a-g
To seven pressure-tubes each containing 2-hydroxy-2-methylcyclohexanone (3, 128 mg, 1 mmol) were added the appropriate primary amine (benzylamine, hexylamine, propylamine, isopropylamine, butylamine, aniline, 4-chloroaniline, 1.15 mmol). The tubes were flushed with argon, sealed and heated at 200 °C overnight in a Radley's combinatorial station with stirring. The aliphatic residues were concentrated and purified by silica gel column chromatography (100% CH 2 Cl 2 , CH 2 Cl 2 -MeOH 15:0.1 and then CH 2 Cl 2 -MeOH 15:1) to give the desired products. All the cleaned products were treated with 0.1M HCl and washed with ether to give their corresponding hydrochloride salts.
The aromatic residues were diluted with ether and extracted with 0.1 M HCl. The acidic phase washed with ether several times and the combined ether phases were dried over MgSO 4 and concentrated. Preparative HPLC were performed with method D, according to the elution times obtained by analytical HPLC using method A.

Characterization Data for Compounds 7a-g, 7a-g
For the NMR data numbering system of compounds 7a-g, 7a-g please refer to Table 6. Table 6. Numbering system for NMR chemical shifts attribution for library 7a-g, 7a-g.

Microwave-Assisted Syntheses
Synthesis of [6a-6′a]+ compound 8: To a microwave flask containing 95 mg (0.5 mmol) of 2-hydroxy-2-phenylcyclohexanone 3, was added 0.575 mmol of benzylamine and NMP (1.5 mL). The flask was flushed with argon and sealed. The reaction mixture was heated at 230 °C for 10 min (until starting reagents disappeared according to analytical HPLC method B). The residue was diluted with ten folds of water and TFA was added until clearness. The obtained solution was filtered and purified by preparative reverse-HPLC (method E).  Synthesis of [6c-6′c]: To a microwave flask containing 95 mg (0.5 mmol) 2-hydroxy-2phenylcyclohexanone 3, was added 0.575 mmol of propylamine and NMP (1.5 mL). The flask was flushed with argon and sealed. The reaction mixture was heated at 230 °C for 30 min (until starting reagents disappeared according to analytical HPLC method B). The residue was diluted with ten folds of water and TFA was added to clearness. The resulting solution was filtered and purified by preparative reverse-HPLC (method E).