Fused Systems Based on 2-Aminopyrimidines: Synthesis Combining Deprotolithiation-in situ Zincation with N -Arylation Reactions and Biological Properties

,


Introduction
Aromatic ketones halogenated next to the carbonyl function are important key intermediates in organic synthesis. They can be used to access various scaffolds for applications in the fields of medicinal [1] and material [2] chemistry. Combining C-N bond formation [3] with condensation reactions [4] can provide an access from such ketones to elaborated fused systems based on 2aminopyrimidines, potentially endowed with various biological properties. [5] However, in spite of their interest, few efficient methods presently exist to selectively introduce a halogen onto an aroyl group next to the carbonyl. Unlike electrophilic halogenation, that in general requires the presence of additional directing substituents, [1b,1c,2] transition-metal-catalyzed C-H bond halogenation using N-halo succinimides as oxidants can regioselectively provide aromatic ketones halogenated next to the carbonyl. Indeed, rhodium(III)-catalyzed bromination or iodination can be performed in the presence of catalytic silver hexafluoroantimonate(V), and stoichiometric pivalic acid or copper acetate, [6] whereas palladium-catalyzed chlorination can be achieved in the presence of triflic acid and potassium persulfate. [7] The former has not been applied to diaryl ketones [6] and the latter showed a coordination-driven regioselectivity. [7] We developed an alternative way based on polar-reagentmediated deprotonative metalation in order to access various iodinated aromatic ketones.
We recently communicated our preliminary studies centered on the functionalization of pyridyl ketones by deprotolithiation-in situ zincation. [8] We here detail our investigations in order to better understand the behavior of the ketone function as directing group. Because CH acidity concept helps understand deprotometalation outcomes, pKa values in THF (THF = tetrahydrofuran) solution were calculated by means of quantum chemistry within the DFT framework using an approach elaborated earlier. [8,9]

Results and Discussion
Deprotolithiation has been largely used to activate aromatic compounds and make them react with a large range of electrophiles. [10] Whereas heteroatom-containing groups can help when connected to the aromatic ring, either by acidification (electron-withdrawing effect), [10e] or by metal coordination, [11] they can also modify the reaction outcome by reacting with the base or the generated aromatic lithium species as is the case for ketone functions. [12] Because of their low compatibility with organolithiums, ketone-containing substrates have only been sporadically deprotometalated. Chromone and N-methyl quinolone could be deprotozincated using TMP-based zinc amides (TMP = 2,2,6,6tetramethylpiperidino) either next to the carbonyl (3 position) or to the other heteroatom (2 position), respectively in the absence or presence of MgCl2, resulting of the Lewis acid coordination by the ketone. [13] Pyronones [13a,14] were functionalized at C3 by using TMPZnCl·LiCl whereas thiochromone [13a] was attacked at C2 by the same reagent. Tropolones were deprotometalated next to the function upon treatment by TMPZnCl·LiCl. [15] Fivemembered heteroaromatic aldehydes [16] were similarly functionalized using TMP-based zinc amides. In contrast, very few diaryl ketones were concerned: benzoyl behaved as tolerated group in the TMPMgCl·LiCl-mediated deprotonation of a polysubstituted diethyl 1,3-phenylenedicarboxylate, [17] whereas it only acted as directing group in the deprotometalation of benzophenone using a lithium-cadmium base. [18] Ketones are compatible to some extent with hindered lithium amides such as LiTMP, as shown for example in the generation of lithium enolates. In contrast, these functions are attacked by organolithiums. [19] Thus, to tackle the compatibility issue between organolithiums and ketones, we planned LiTMPmediated deprotometalation in the presence of a zinc-based in situ trap. Studies have shown that it is possible to employ species such as Zn(TMP)2 [20] and ZnCl2·2LiCl [21] to intercept polar arylmetals. Within this study, we chose ZnCl2·TMEDA [22] (TMEDA = N,N,N',N'-tetramethylethylenediamine) as in situ trap to attempt deprotolithiation followed by 'trans-metal trapping' [23] and iodolysis from various diaryl ketones (Table 1).
Xanthone dimethyl acetal can be dimetalated next to the pyran oxygen (4 and 5 positions) in tetrahydropyran at -13 °C upon treatment by butyllithium (3 equiv.) in the presence of potassium tert-butoxide (3 equiv.), a result evidenced by subsequent trapping with iodomethane. [24] Treatment of xanthone (1a) in THF containing ZnCl2·TMEDA (1 equiv.) with LiTMP (1.5 equiv.) at -55 °C for 15 min and then iodine (1.5 equiv.) provided the 1-iodo derivative 2a in 72% yield (entry 1). Thioxanthone (1b) was similarly converted to afford 2b (entry 2). The calculated pKa values of 1a,b can help rationalize this regioselectivity change and explain the directing group properties of the ketone. Compared with those next to the ketone function (1 and 8 positions), CH acidities next to the pyran or thiopyran heteroatom (4 and 5 positions) are higher. The observed deprotonation at the 1 position (beyond thermodynamic acidity) [11] could result from (i) ketone coordination to lithium, with induced acidity increase, (ii) favored approach of the base, and/or (iii) stabilization. Indeed, when compared with esters and N,N-dialkylcarboxamides, the coordination ability of the ketone carbonyl group is known to be higher. [7] [c] 20% [c] 7 [b] 1g 2g, 50% [d] 8 [b,e]  From xanthone (1a) to fluorenone (1c), changes were noticed. Thus, whereas the iodide 2a was isolated in 72% yield (together with recovered starting material), the 1-iodo derivative 2c was obtained in 52% yield due to the competitive formation of the keto-alcohol 2c' (entry 3). The latter had previously been isolated in 63% yield by reacting fluorenone (1c) with the base prepared in situ from ZnCl2·TMEDA and LiTMP in a 1:3 ratio, and supposed to be LiTMP-Zn(TMP)2, [20] with Zn(TMP)2 acting as in situ trap. [25] Because arylzincs hardly react with ketones, [26] 2c' would rather result from an addition of the generated 1lithiofluorenone onto starting 1c. At first sight, this is surprising since fluorenone (1c) benefits from a more favorable CH acidity at its 1 position than xanthone (1a). That such a competitive reaction does not take place from 1a is in favor of an impact of the ketone carbonyl orientation on the stabilization of the lithio compound before its interception by 'trans-metal trapping'. This is also in accordance with data previously recorded with N,Ndialkylcarboxamides as deprotolithiation directing groups. Indeed, from studies performed on benzamides, Beak and coworkers suggest a correlation between the reaction efficiency and the distance from the directing group oxygen to the ortho hydrogen, as well as an optimal distance to accommodate the base effectively in the transition structure. [27] By switching from symmetrical xanthone (1a) to the dissymmetrical azaxanthone 1d, one moves from two different potential deprotometalation sites to four. Nevertheless, from the two CH sites next to the carbonyl group, the pyridine 4 position is clearly the more acidified, and it is logically attacked to give the 4-iodo derivative 2d in 60% yield (entry 4). The corresponding thio analog 2e was obtained in 60% yield by reacting the azathioxanthone 1e under the same conditions (entry 5). In the case of 1d, the diiodide 2d' resulting from a twofold deprotonation at the 4 and 9 positions was also isolated (entry 4); that such a diiodo derivative is not formed from 1e might result from a lower propensity of sulfur to coordinate lithium, when compared with oxygen.
Due to its strong ability to coordinate lithium, the pyridine nitrogen is a good candidate to compete with the ketone oxygen in coordinating metals. Thus, it is not surprising that the azafluorenone 1f is also attacked at its 9 position (to afford 2f' in 20% yield) in addition to the 4 one giving 2f in 33% yield (entry 6). The second functionalization at the 9 position might happen after a first deprotolithiation-'trans-metal trapping' at the 4 position, as previously observed from other heterocycles. [9a,28] Because of their π-deficiency, benzoylpyridines are more sensitive toward nucleophiles than simpler aryl ketones; not only the function, but also the pyridine ring (and in particular when the ketone function activates its 2 and 4 positions toward nucleophilic attack) can be subjected to the addition of organolithiums. From 2-benzoylpyridine (1g), optimization of the reaction conditions was carried out in THF containing ZnCl2·TMEDA (1 equiv.) by using four different reaction temperatures, between -70 and -10 °C, and four different amounts of LiTMP, between 1 and 3 equiv. The best results were observed when 1.5 equiv. of LiTMP were employed at -30 °C for 15 min before interception with iodine to furnish the 3iodo derivative 2g in 50% yield (entry 7). The reaction from 4benzoylpyridine (1h) was performed as for its 2-isomer, but using 2 equiv. of LiTMP; under these conditions, the 3-iodo derivative 2h was isolated in 45% yield (entry 8).
Among the three different benzoylpyridines used 1g-i, 3benzoylpyridine (1i) is by far the more sensitive due to its free 4 position next to the ketone function. [29] To limit side nucleophilic attacks, the reaction has to be conducted at lower temperatures. Under the conditions used to make 1a-f react (reaction at -50 °C), the product 2i resulting from a metalation at the 4 position was isolated from a complex mixture in 30% yield. At -70 °C, the result was slightly improved, with 2i obtained in 37% yield (entry 9). Diiodides were suspected as side products in the course of the reactions coming from 1g-i; unambiguously, such a derivative (2h') was obtained in 10% yield from 1h (entry 8).
The pyridine regioselective deprotonation of 1g-i is in accordance with higher calculated pKa values for the phenyl ring. If the pyridine pKa values might help understand regioselective deprotometalation next to the ketone in the case of 1h and 1i, things are different for 1g. Indeed, for the latter, the most acidic sites are rather the 4 and 5 positions whereas the reaction takes place at the 3 position flanked by the function (entries 7-9). It is thus of importance to take into account possible coordination of the present heteroatoms onto metals in order to have a better idea of the pKa values within premetalation complexes or transition structures. To this purpose, we calculated the CH acidities of the 1:1 complexes between 1g-i and LiTMP ( Figure  1). As it was proved earlier for methoxypyridines, [28e] coordination by nitrogen was predicted to be slightly more effective than by oxygen (with the corresponding isomeric complexes being by 0.9 kcal mol -1 more stable for 1h and 1.1 kcal mol -1 for 1i; in the case of 1g, both atoms take part in the complexation). The metal coordination by the ring nitrogen lone pair drastically increases the pyridine CH acidity. Thus, in the case of 1g·LiTMP, metal chelation strongly increases the acidity at the 3 position, adjacent to the ketone function. Similar coordination by the ketone oxygen can also operate to increase acidity, favor the approach of the base and stabilize the arylmetal species. To identify which metal, between lithium and zinc, is involved in such coordinations, we first compared spectral data of 3benzoylpyridine (1i), ZnCl2·TMEDA, TMEDA and 3benzoylpyridine (1i) in the presence of ZnCl2·TMEDA. The IR, 1 H and 13 C NMR spectra recorded in d 8 -THF showed no significant difference between 3-benzoylpyridine (1i) with and without ZnCl2·TMEDA on the one hand, and between ZnCl2·TMEDA in the presence or not of 1i on the other hand (e.g. ~1 cm -1 for the C=O IR absorption band; ≤ 0.1 ppm for 13 C NMR chemical shifts). Because the chelating TMEDA is not displaced by addition of 3-benzoylpyridine (1i), and TMPZnCl·LiCl not capable of deprotonation at such low temperatures, [16b,16c,30] we can suggest that ZnCl2·TMEDA only operates after deprotolithiation and intercepts the generated aryllithium.
On this basis, the scope of our approach will be determined by the efficiency of the in situ traps employed, and by the stability of the transient lithiated arylketones involved. When present at pyridine 2 position, [31] halogens are known to acidify the 4 position, and thus to stabilize a 4-lithio derivative. This long-range effect, well-known for bromine, [32] also exists for chlorine [33] and fluorine. [28e,34] As a consequence, the deprotolithiation-transmetalation-iodolysis sequence carried out on the 2-halogenated 3-benzoylpyridines 1j-m offered more satisfactory yields (63-78%, entries 10-13) than from reference 3-benzoylpyridine (1i) (30%, entry 9). Nevertheless, in spite of the presence of a stabilizing chloro group, the reaction from 1n proved more complex, only affording 2n in a modest 27% yield (entry 14). Unlike halogens, methoxy is not a suitable group to acidify long-range positions. [28e] If the yield to convert 1o into 2o was found higher (88% against 30% from 1i under similar reaction conditions, entry 15), it might rather be related to the higher propensity of methoxy to make the pyridine ring less prone to nucleophilic attacks. [8] In all the above examples, functionalization takes place at a position adjacent to the ketone function. It was thus of interest to attempt the reaction on a substrate benefiting from a more activated site. To this purpose, we chose 2-benzoylthiophene (1p) for which the most acidic site is next to sulfur. When submitted to LiTMP in the presence of ZnCl2·TMEDA as before, the iodide 2p logically resulting from proton abstraction at the thiophene 5 position was isolated in 80% yield (entry 16).
Until now, we used iodine to evidence the formation of deprotometalated species, the resulting iodoketones being of high relevance to perform further functionalization. It is also of interest to combine deprotometalation with Negishi-type crosscoupling [35] in order to directly connect aryl groups. In the presence of catalytic amounts of palladium(II) chloride and 1,1'diphenylphosphinoferrocene (dppf), and under THF reflux, the deprotozincated product coming from xanthone (1a) was coupled with 2-chloropyridine to afford the derivative 3a in 76% yield (Scheme 1). The products 2d, 2f, 2g, 2h, 2h', 2i, 2j, 2k, 2l, 2m, 2o, 2p and 3a were identified unambiguously by X-ray diffraction (see Supporting information).
Pyridopyrimidine is a core present in compounds of medicinal importance for their anticancer, CNS, fungicidal, antiviral, anti-inflammatory, antimicrobial and antibacterial properties. [36] To access such compounds, efficient synthetic methods are required. The two-fold functionalization of the bifunctional substrates 2 combining C-N bond formation [3] from the iodide with imine formation [4] from the ketone appears as an attracting way to reach original fused systems based on pyrido[2,3-e]pyrimidines.
In order to access the 2-aminopyrido[4,3-d]pyrimidines 4i and 4o, for which the pyridine nitrogen is at the 6 position, we respectively involved in the reaction the 3-benzoyl-4iodopyridines 2i and 2o. In spite of the known higher reactivity of 4-iodopyridines (higher partial positive charge on the carbon bearing the halogen) over 3-iodopyridines in copper-catalyzed C-N bond formation, [40] similar yields of 68% and 65% were remarked from 2i and 2o (entries 3 and 4, Figure 3).
Whereas none of the derivatives was shown to inhibit significantly CDK5/p25, CDK9/Cyclin T, GSK-3 and RIPK3, positive results were noticed with CDK2/Cyclin A, PIM1, LmCK1, Haspin and Aurora B (Table 7). These first results were then verified by testing the dose-dependent effect of the small chemical compounds against a selected panel of kinases ( Table  8).
The inhibition reported for 4b on PIM1 is interesting. Indeed, Horiuchi et al. showed that PIM1 kinase inhibition should be explored for developing targeted therapy against triple-negative breast tumors with elevated MYC expression. [45] Moreover, despite the structural similarities between the three CDKs tested, 4b was shown to be selective for CDK2/Cyclin A.

Conclusions
We showed that the carbonyl group of aromatic ketones can be efficiently employed to induce deprotolithiation at an adjacent site. The reaction can be carried out by using a hindered nonnucleophilic lithium amide in the presence of a zinc salt that acts as in situ trap. The usefulness of aromatic ketones halogenated next to the carbonyl function was demonstrated by the synthesis of promising fused systems based on 2-aminopyrimidines.

Experimental Section
All the reactions were performed under an argon atmosphere. THF was distilled over sodium/benzophenone. Column chromatography separations were achieved on silica gel (40-63 μm). Melting points were measured on a Kofler apparatus. IR spectra were taken on a Perkin-Elmer Spectrum 100 spectrometer. 1 H and 13 C Nuclear Magnetic Resonance (NMR) spectra were recorded on a Bruker Avance III spectrometer at 300 MHz and 75 MHz, respectively. 1 H chemical shifts (δ) are given in ppm relative to the solvent residual peak, 13 C chemical shifts are relative to the central peak of the solvent signal, [46] and coupling constants (J) are given in Hz.

General procedure 1 for the synthesis of the aryl iodides 2.
To a stirred mixture of the required ketone (1.0 mmol) and ZnCl2·TMEDA (0.26 g, 1.0 mmol) in THF (3 mL) at -30 °C was added dropwise a solution of LiTMP (prepared by adding BuLi (about 1.6 M hexanes solution, 1.5 mmol) to a stirred, cooled (0 °C) solution of 2,2,6,6-tetramethylpiperidine (0.25 mL, 1.5 mmol) in THF (3 mL) and stirring for 5 min) cooled at -30 °C. After 15 min at -30 °C, a solution of I2 (0.38 g, 1.5 mmol) in THF (5 mL) was introduced, and the mixture was stirred overnight before addition of an aqueous saturated solution of Na2S2O3 (5 mL) and extraction with AcOEt (3 x 20 mL). The combined organic layers were dried over MgSO4, filtered and concentrated under reduced pressure. The crude product was purified by chromatography over silica gel (the eluent is given in the product description).

General procedure 2 for the synthesis of the aryl iodides 2.
To a stirred mixture of the required ketone (1.0 mmol) and ZnCl2·TMEDA (0.26 g, 1.0 mmol) in THF (3 mL) at -55 °C was added dropwise a solution of LiTMP (prepared by adding BuLi (about 1.6 M hexanes solution, 1.5 mmol) to a stirred, cooled (0 °C) solution of 2,2,6,6-tetramethylpiperidine (0.25 mL, 1.5 mmol) in THF (3 mL) and stirring for 5 min) cooled at -55 °C. After 15 min at -55 °C, a solution of I2 (0.38 g, 1.5 mmol) in THF (5 mL) was introduced, and the mixture was stirred overnight before addition of an aqueous saturated solution of Na2S2O3 (5 mL) and extraction with AcOEt (3 x 20 mL). The combined organic layers were dried over MgSO4, filtered and concentrated under reduced pressure. The crude product was purified by chromatography over silica gel (the eluent is given in the product description).

General procedure 3 for the synthesis of the aryl iodides 2.
To a stirred mixture of the required ketone (1.0 mmol) and ZnCl2·TMEDA (0.26 g, 1.0 mmol) in THF (3 mL) at -30 °C was added dropwise a solution of LiTMP (prepared by adding BuLi (about 1.6 M hexanes solution, 2.0 mmol) to a stirred, cooled (0 °C) solution of 2,2,6,6-tetramethylpiperidine (0.33 mL, 2.0 mmol) in THF (3 mL) and stirring for 5 min) cooled at -30 °C. After 15 min at -30 °C, a solution of I2 (0.51 g, 2.0 mmol) in THF (5 mL), and the mixture was stirred overnight before addition of an aqueous saturated solution of Na2S2O3 (5 mL) and extraction with AcOEt (3 x 20 mL). The combined organic layers were dried over MgSO4, filtered and concentrated under reduced pressure. The crude product was purified by chromatography over silica gel (the eluent is given in the product description).

General procedure 4 for the synthesis of the fused systems based on pyrido[2,3-e]pyrimidines 3g.
A mixture of CuI (20 mg, 0.10 mmol), the required amine (1.2 mmol), K3PO4 (0.44 g, 2.0 mmol), 2-benzoyl-3iodopyridine (2g, 0.31 g, 1.0 mmol) and DMSO (1.4 mL) was degased and heated under argon and stirring at 110 °C for 16 h. After filtration over celite (washing using AcOEt) and removal of the solvents, the crude product is purified by chromatography over silica gel (the eluent is given in the product description).

General procedure 5 for the synthesis of the fused systems based on 2-aminopyrimidines 4.
A mixture of CuI (20 mg, 0.10 mmol), guanidine hydrochloride (0.19 g, 2.0 mmol), K3PO4 (0.88 g, 4.0 mmol), the required halogenopyridine (1.0 mmol) and DMSO (0.5 mL) was degased and heated under argon at 110 °C for 24 h. After filtration over celite (washing using AcOEt) and removal of the solvents, the crude product is purified by chromatography over silica gel (the eluent is given in the product description).