Tertiary Alkylamines as Nucleophiles in Substitution Reactions at Heteroaromatic Halide During the Synthesis of the Highly Potent Pirinixic Acid Derivative 2-(4-Chloro-6-(2,3-dimethylphenylamino)pyrimidin-2-ylthio)octanoic Acid (YS-121)

YS-121 [2-(4-chloro-6-(2,3-dimethylphenylamino)pyrimidin-2-ylthio)octanoic acid] is the result of target-oriented structural derivatization of pirinixic acid. It is a potent dual PPARα/γ-agonist, as well as a potent dual 5-LO/mPGES-1-inhibitor. Additionally, recent studies showed an anti-inflammatory efficacy in vivo. Because of its interference with many targets, YS-121 is a promising drug candidate for the treatment of inflammatory diseases. Ongoing preclinical studies will thus necessitate huge amounts of YS-121. To cope with those requirements, we have optimized the synthesis of YS-121. Surprisingly, we isolated and characterized byproducts during the resulting from nucleophilic aromatic substitution reactions by different tertiary alkylamines at a heteroaromatic halide. These amines should actually serve as assisting bases, because of their low nucleophilicity. This astonishing fact was not described in former publications concerning that type of reaction and, therefore, might be useful for further reaction improvement in general. Furthermore, we could develop a proposal for the mechanism of that byproduct formation.

Based on those findings, more detailed investigations concerning the mechanism of mPGES-1 inhibition, the selectivity profile and the in vivo activity of YS-121 have been conducted [9]. The results of those experimental studies demonstrated that YS-121 inhibits mPGES-1 in a selective, concentration-dependent, reversible and noncompetitive manner. Moreover, the above-mentioned lack of interference with both isoforms of cyclooxygenase still remains, even in assays performed in human whole blood. Additionally, an anti-inflammatory efficacy of YS-121 was shown in vivo [9].
In summary, YS-121 is a promising drug candidate, because of its interference with many targets involved in inflammatory diseases, but there is still a huge lack of in vivo data concerning dosing, pharmacokinetics and effectiveness in humans.
Ongoing preclinical studies will thus necessitate huge amounts of YS-121. In medicinal chemistry the amount of compound needed for biochemical assays to get first in vitro data during structural optimization steps is very low. For most assays 3-4 mg are sufficient and the syntheses are planned along that required low amount of compound. But the demand for compound rises up to gram scale by conducting animal experiments to get first in vivo data of an in vitro hit. To cope with those requirements, we have optimized the synthesis of YS-121 resolving the problem of byproduct formation during that synthesis conducted in a larger scale. Furthermore, we isolated and characterized some byproducts enabling us to develop a proposal for the mechanism of that byproduct formation.

Results and Discussion
YS-121 was synthesized in a four-step reaction originally published by d'Atri et al. [10] and already modified by Koeberle et al. [8]. Here, we modified and optimized that synthesis again to make it convenient to synthesize YS-121 in amounts of a few grams. The specific conditions utilized, the problem of byproduct formation in step (iii a) and the optimization step to circumvent that byproduct formation (iii b) are illustrated in Scheme 1.
During the first step, 2-mercaptopyrimidine-4,6-diol (3) reacted by a nucleophilic substitution with ethyl 2-bromooctanoate in DMF in the presence of triethylamine to form the thioether derivative ethyl 2-(4,6-dihydroxypyrimidin-2-ylthio)octanoate (4). Chlorination with POCl 3 and N,N-diethylaniline gave the chlorinated pyrimidine derivative ethyl 2-(4,6-dichloropyrimidin-2-ylthio)octanoate (5). Through a nucleophilic aromatic substitution at the pyrimidine core of 5, one chloro group was substituted by 2,3-dimethylaniline with triethylamine in EtOH to form the monoaminated derivative ethyl 2-(4-chloro-6-(2,3-dimethylphenylamino)pyrimidin-2-ylthio)octanoate (6, iii a). Additionally, the disubstituted byproduct ethyl 2-(4-(diethylamino)-6-(2,3-dimethylphenylamino)pyrimidin-2-ylthio)octanoate (7) was formed during that step. Changing the assisting base to sodium carbonate in step (iii b), lead to a circumvention of that byproduct formation. The last step to get to the desired compound YS-121 (2) consisted of a saponification of the ethyl ester derivative 6 with LiOH in EtOH to give the carboxylic acid derivative 2-(4-chloro-6-(2,3-dimethylphenylamino)-pyrimidin-2-ylthio)octanoic acid (2). Because 6 and 7 could not be separated during the purification procedure of step (iii a), 7 was also hydrolyzed during step (iv) to give the byproduct (2-(3-(diethylamino)-5-(2,3-dimethylphenylamino)phenylthio)octanoic acid (8), which could be separated from 2 and its structure elucidated afterwards. In comparison to Koeberle et al. [8], several changes have been made to optimize the reaction in general. In step (i) the relative amounts of ethyl 2-bromooctanoate and triethylamine as well as the reaction temperature were reduced in order to minimize byproduct formation initially caused by a dehydrohalogenation of ethyl 2-bromooctanoate leading to a Michael acceptor system that is able to react with nucleophiles as well. The dehydrohalogenation is a β-elimination. With regard to the desired nucleophilic substitution by 3 on ethyl 2-bromooctanoate, high temperatures and bulky bases raise the probability of β-elimination [16]. Because of the lower reaction temperature, longer reaction times were required until TLC-control showed total conversion of 3. In step (ii) we reduced the reaction temperature to guard the product from decomposition reactions. Because of the lower reaction temperature, longer reaction times were required again until TLC-control showed total conversion of 4. In steps (iii) and (iv) only the reaction times were expanded. Perhaps it would have been possible to reduce the reaction times of all steps, but our aim was first and foremost to get the highest yield possible during each respective step. As described above, during structural optimization steps in medicinal chemistry, syntheses are planned to get the test compound on mg scale, which is sufficient for first in vitro data, but we conducted the reaction steps on a larger scale to cope with the rising demand for test compound on gram scale for further in vivo testing. Those higher absolute amounts of compounds during synthesis should also have expanded the reaction times.
The challenge of optimizing the synthesis of YS-121 consisted of the low yields of the different steps that were primarily based on the high level of byproduct formation. That necessitated huge reaction batches, followed by extensive purification procedures and finally led in a large part to those low yields. One of those byproducts was formed in high amounts in step (iii a). It was not described in former publications concerning that type of reaction [6,8,[10][11][12][13][14], also with different aromatic amines as nucleophiles, and this forced us to modify the reaction procedure as well as the way of purification. In step (iii a) the chlorinated pyrimidine derivative 5 should be monoaminated to 6 by a nucleophilic aromatic substitution with 2,3-dimethylaniline. The reaction was conducted with triethylamine in EtOH, and the amount of the nucleophile 2,3-dimethylaniline was 1.05 equiv. with regard to 5. After the reaction was finished, TLC-control showed the formation of at least four byproducts next to 6. Two purification steps by column chromatography with silica gel only led to a rough purification in which three of the four byproducts could be separated from the desired product 6 in that step.
Further conversion of the mixture of step (iii a) in step (iv) again led to the isolation of two different compounds. In comparison with the ESI data of step (iii a), mass spectrometry indicated that both compounds were converted into carboxylic acids. One of them was the desired final compound 2. We tried to separate them by column chromatography, but we were not successful. Afterwards, we switched to RP-18 silica gel for column chromatography, which finally allowed a satisfactory separation. Furthermore, after termination of the complete synthesis, we found out that purification by RP-18 silica gel for column chromatography would have been convenient for the purification in step (iii a) as well.
During the purification procedure of step (iv), it was possible to isolate the byproduct. The amount of the byproduct was 1:1, with regard to the amount of 2. Characterization by the whole lineup of structure determination methods (including 1 H-NMR, 13   As far as the formation of byproduct 8 is concerned, a disubstitution would have only been possible in step (iii a), if diethylamine or diethylamine-related reactive intermediates had been present next to 2,3-dimethylaniline as nucleophiles as well. Either that byproduct formation is the result of a diethylamine impurity of the used commercially available reagents triethylamine, 2,3-dimethylaniline or EtOH in step (iii a) or it is a further indication that triethylamine could react as a nucleophile during nucleophilic aromatic substitution reactions at heteroaromatic halides as previously described by Matsumoto et al. [15].
To get deeper insights into the mechanism leading to the formation of byproduct 8, we repeated step (iii a) with triethylamine as assisting base and could reproduce that byproduct formation in two reaction batches reacted independently from each other. Afterwards, we tried to isolate other byproducts of that reaction that could tell us something about the mechanism leading to 8. In fact, we were successful, because we could isolate the stable intermediate 6- Figure 4), the precursor of compound 7. Compound 9 is a quaternary ammonium cation. Furthermore, both chlorines were already substituted, so the formation of 9 probably lead in the first stage to a Meisenheimer complex intermediate 10 ( Figure 5) that was formed by a nucleophilic attack of triethylamine leading in second stage to an elimination of a chloride ion. Additionally, other byproducts were formed in step (iii a), but we were not able to separate them satisfactorily. Because the insights into the mechanism of byproduct formation were still not sufficient, we repeated step (iii a) again, now with N-methylpyrrolidine as assisting base as a member of cyclic tertiary alkylamines. Again, we tried to isolate byproducts giving us new insights into the mechanism of byproduct formation and again, we were successful. We could isolate two byproducts telling us a lot about the mechanism. First, we could isolate the byproduct ethyl 2-(4-ethoxy-6-(pyrrolidin-1yl)pyrimidin-2-ylthio)octanoate (11, Figure 6). In comparison to compound 5, several changes took place at the pyrimidine core. Both chlorines were substituted, the first by N-methylpyrrolidine and the second by the solvent ethanol. During byproduct formation, N-methylpyrrolidine lost its methyl group. But that byproduct as well as byproduct 8 could not tell us how in both cases the quaternary ammonium cation had lost its fourth carbon residue leading to a N,N-dialkylaminyl substituent at the pyrimidine core. In summary, during formation of byproduct 11, one equivalent of methyl chloride and one equivalent of hydrochloric acid were eliminated.
During formation of byproduct 12 the pyrrolidine ring was opened but no carbon residue was eliminated. In summary, only one equivalent of hydrochloric acid was eliminated through a solvolytic reaction with ethanol, but the terminally chlorinated n-butyl residue as well as the methyl residue of the N,N-dialkylaminyl substituent at the pyrimidine core provided evidence for the ring opening mechanism. That ring opening probably had happened either through a nucleophilic attack of a chloride ion during the state of a Meisenheimer complex intermediate 13 as a concerted process together with the elimination of the chloride ion from the pyrimidine core or during the state of a quaternary ammonium cation such as 1-(6-ethoxy-2-(1-ethoxy-1-oxooctan-2-ylthio)pyrimidin-4-yl)-1methylpyrrolidinium (14, Figure 8) as a non-concerted process.  We could not isolate either of the two intermediates, but because we could isolate byproduct 9, so the non-concerted process of a nucleophilic attack of a chloride ion after elimination from the pyrimidine core seems to be more likely. The solvolytic reaction at the pyrimidine core also stresses that last presumption. Only during the stage of the quaternary ammonium cation, a nucleophilic attack of the solvent is possible and catalyzed, because N-methylpyrrolidine is a good leaving group in that case. The potential of tertiary amines to catalyze nucleophilic substitution reactions at aromatic heterocycles was already described in the literature [17]. The fact that we could not isolate solvolytic reaction byproducts from the reaction with triethylamine could be due to the steric hindrance at the tertiary ammonium cation 9 preventing a nucleophilic attack of the solvent.
Another interesting fact is that during reaction with N-methylpyrrolidine we could not isolate any amount of desired compound 6 at all, although we had not changed the relative amounts of educts. That suggests a higher nucleophilic potential of N-methylpyrrolidine as well as a higher potential of N-methylpyrrolidine to catalyze solvolytic reactions at the chlorinated pyrimidine core in comparison to triethylamine.
To sum it up, the following Scheme 2 depicts our proposal for the mechanism of byproduct formation based on the isolated and characterized byproducts 8, 9, 11 and 12 during step (iii a) with two different tertiary alkylamines as assisting bases: Scheme 2. Postulated mechanism of byproduct formation.
(i) A nucleophilic attack of a tertiary alkylamine at a 4,6-di-chlorinated pyrimidine core 15 leads to a Meisenheimer complex intermediate 16; (ii) afterwards, a chloride ion is eliminated leading through rearomatization to a quaternary ammonium cation intermediate stage 17; (iii) finally, a nucleophilic attack by a chloride ion at any carbon residues of the quaternary N,N,N-trialkylaminyl substituent of 17 leads to a monochlorinated pyrimidine core with a N,N-dialkylaminyl substituent 18.
With the isolation of byproduct 9 and the support from isolated byproducts 11 and 12 we can state that formation of byproduct 8 during step (iii a) was due to a direct nucleophilic attack of the tertiary alkylamine triethylamine at the chlorinated pyrimidine core of 5. Because triethylamine and other tertiary alkylamines are broadly used as assisting bases during nucleophilic aromatic substitution reactions due to their low nucleophilicity, our results put their use into question during that type of reactions.
As stated above, the formation of that sort of byproducts was not described in former publications about that type of reaction showing the synthesis of pirinixic acid derivatives. One cause of the fact that byproduct 8 is apparently formed exclusively during reaction batches on a larger scale could probably be the higher concentration of triethylamine during step (iii a) that has raised the probability of a nucleophilic attack of triethylamine at the chlorinated pyrimidine core. Therefore, we think that it could be useful to lower the concentration of triethylamine in step (iii a) or to change the assisting base triethylamine to sodium carbonate as depicted in step (iii b) and as previously described by Rau et al. [6] and Popescu et al. [11] to avoid the formation of byproduct 8.
We checked that last proposal for effectiveness during step (iii b) and indeed, byproducts were formed to a lesser extent. That made the purification procedure less complex. Therefore, we succeeded in optimizing reaction step (iii a) by changing the assisting base triethylamine to sodium carbonate as depicted in step (iii b). So, all forms of reactions utilizing inorganic assisting bases during nucleophilic aromatic substitution reactions at heteroaromatic halides seem to be more suitable than reactions using tertiary alkylamines as assisting bases.

General
Compounds and Chemistry. All commercial chemicals and solvents were of reagent grade and were used without further purification. For step (i), commercially available triethylamine was distilled after refluxing over KOH. All reactions were carried out in an argon atmosphere and compounds 2, 4, 5, 6, 8, 9, 11 and 12 were permanently stored under argon after their purification. The structures of compounds 2, 4, 5, 6, 8, 9, 11 and 12 were confirmed by 1 H-NMR spectroscopy and mass spectrometry (ESI). The structures of final compound 2 and byproducts, 8, 9, 11, and 12 were additionally confirmed by 13 C-NMR spectroscopy and in part by high-resolution mass spectrometry (HRMS). Furthermore, the purity of compound 2 was determined by combustion analysis and high pressure liquid chromatography (HPLC) analysis and was 99%. The mobile phase gradient used consisted of 2 phases (MeOH = phase A, Milli-Q distillate 0.1% formic acid = phase B) starting with 10% phase A up to 100% phase A within 45 min. The used flow rate was 1 mL/min and the separation temperature was set at 25 °C. Reactions were monitored by thin-layer chromatography that was carried out on Merck TLC silica gel plates 60 F 254 and RP-18 F 254 s (Merck, Darmstadt, Germany) using UV light as visualizing agent. Macherey Nagel silica gel 60, particle size 0.040-0.063 mm/230-400 mesh ASTM (Macherey Nagel, Düren, Germany) and Sigma-Aldrich octadecyl-functionalized silica gel, particle size 200-400 mesh, extent of loading: 20-22% (Sigma-Aldrich, St. Louis, MO, USA) were used for column chromatography. The synthesis of YS-121 followed the route originally published by d'Atri et al. [10] and already modified by Koeberle et al. [8]. That synthesis was modified and optimized again in some cases to be convenient to resolve the problem of byproduct formation during that synthesis conducted in a larger scale.

Ethyl 2-(4-Chloro-6-(2,3-dimethylphenylamino)pyrimidin-2-ylthio)octanoate (6) and 2-(4-Chloro-6-(2,3-dimethylphenylamino)pyrimidin-2-ylthio)octanoic Acid (2)
Step (iii a). A solution of the chlorinated pyrimidine derivative 5 (18.7 g, 53.2 mmol, 1.00 equiv.) obtained from step (ii), triethylamine (16.2 g, 160 mmol, 3.00 equiv.) and 2,3-dimethylaniline (6.78 g, 55.9 mmol, 1.05 equiv.) in EtOH (62.5 mL) was refluxed under stirring for 65 h (TLC control). Then, EtOH and excess triethylamine were evaporated under reduced pressure. The residue was dissolved in EtOAc, and the organic layer was extracted with diluted HCl, saturated NaHCO 3 solution and brine. Then, the organic layer was dried over MgSO 4 , filtered and evaporated under reduced pressure. The crude product was roughly purified by column chromatography (silica gel, n-hexane-EtOAc 10:1) to give a mixture of the desired compound 6 together with one byproduct as a yellow oil (14. After completion of step (iv), further purification of 500 mg of that mixture by column chromatography (RP-18 silica gel, acetonitrile-water 4:1) gave the desired compound 6 (210 mg) as a yellow oil. Therefore, purification by RP-18 silica gel is already useful and applicable at that stage of synthesis. By extrapolating the result of RP-18 column chromatography of 500 mg up to 14.0 g, a yield of 5.88 g, 25% of 6 could have been achievable. R f = 0.2 (silica gel, n-hexane-EtOAc 10:1); 1 H-NMR (250. 13  Step (iv). The mixture of 6 and one byproduct (13.5 g) obtained in step (iii a) was dissolved in EtOH (420 mL) at room temperature. An approximate excess of LiOH (2.61 g, 109 mmol) was added, and the resulting mixture was stirred at room temperature for 120 h (TLC control). Then, EtOH was removed under reduced pressure, and the obtained residue was dissolved in water under heating, while low amounts of MeOH were added. The solution was acidified with diluted HCl, and the precipitate was filtered, washed to neutrality with water and then with n-hexane. A solution of the precipitate in EtOAc was dried over MgSO 4 , filtered and evaporated under reduced pressure. The crude product was purified by column chromatography (RP-18 silica gel, acetonitrile-water 4:1). Finally a solution of the purified product in EtOAc was extracted with diluted HCl, washed to neutrality with water and was dried over MgSO 4 . After filtration, the solvent was evaporated under reduced pressure and the product was precipitated from EtOAc/n-hexane. After two weeks at 26 °C, the solid product was filtered and was washed with n-hexane. In the end 2.00 g of the carboxylic acid derivative 2 were obtained as a pale yellow solid. By including the extrapolated content of 6 in the mixture used in that step, a yield of 38% was achieved.

Conclusions
Here, we have demonstrated that syntheses conducted in a larger scale can generate problems and byproducts that did not appear at these syntheses at the original small scale, but we also show that it is possible to resolve these problems by a change in the purification procedure or reagents used as well as in the way of synthesis and thus obtained clean compounds. By conducting the described synthesis, it was possible to generate up to 2.00 g of YS-121 (2) as a pure compound for use in ongoing preclinical studies. By a change of the assisting base triethylamine to sodium carbonate in step (iii b), we succeeded in optimizing that step. Byproducts were formed to a lesser extent and the purification procedure was dramatically simplified.
The reaction times are probably reducible, but our aim was first and foremost to get the highest yield possible during each respective step. After having conducted several reactions with two different assisting bases in step (iii a) and after having isolated and characterized different byproducts formed during those reactions, we could develop a proposal for the mechanism of byproduct formation. Therefore, we could clearly identify triethylamine as the reason for the formation of byproduct 7 during step (iii a) that was isolated later, in step (iv), after saponification, as 8.
Additionally, we could show that also other tertiary alkylamines could act as nucleophiles during a nucleophilic aromatic substitution reaction at a heteroaromatic halide. Such a formation of byproducts resulting from an initial nucleophilic attack of a tertiary alkylamine and a following elimination of one alkyl residue leading to a N,N-dialkylaminyl substituent at the heteroaromatic core was not described before during the used conditions. That unexpected fact can be very useful for reaction improvement of other nucleophilic aromatic substitution reactions at heteroaromatic halides in general.