Synthesis and cytotoxic/antimicrobial screening of 2-alkenylimidazo[1,2-a]pyrimidines

Polyaza-heterocycles show a plethora of biological properties and represent a significant percentage of clinically used drugs. However, the imizado[1,2-a]pyrimidine ring system needs more attention in terms of pharmaceutical applications. Herein, we report a sequence involving an aldolic condensation/bromination/Hantzsch reaction to construct a series of 2-alkenylimidazo[1,2-a]pyrimidines for exploring their cytotoxic and antimicrobial properties. After performing a preliminary screening, two compounds displayed good cytotoxicity against prostate, breast, and colon cancer types; either bulky or an extra phenyl attached at the styryl moiety seems to be a requirement for good activity. With respect to the antimicrobial effect, some compounds showed considerable inhibition against multi-drug resistant K. pneumoniae, one of the most threating bacteria nowadays. Thus, the series may become the basis for the design of more active compounds.


Introduction
Undoubtedly, polyaza-heterocycles play a key role in modern scientific disciplines, being extremely relevant in catalysis and medicinal chemistry [1]. Nitrogen-containing compounds become indispensable substances in food and constitute pharmacologically relevant secondary metabolites of plants and microorganisms (i.e., alkaloids and peptides) [2]. Many factors contribute to the privileged status of polyaza-heterocycles such as the potential hydrogen bond formation -which may stabilize a stronger ligandreceptor complex-, the basic nature of sp 2 -hybridized nitrogen and their capability for improving the ADMET properties of candidates [3]. Thus, it is not surprising that about 60% of small drugs bear an aza-heterocycle in their structure [4].
In addition to the confirmed properties, the imidazo[1,2a]pyrimidine resembles the purine nitrogenous bases present in DNA and RNA [12] and its structure mimics the guanidine functionality. Several natural products (mainly marine alkaloids) own the bicycle -or the related 2-aminoimidazole-and possess antimicrobial or antibiofilm properties [13].
Motivated by the properties of the imidazo[1,2-a]pyrimidine scaffold, and to further expand their biological activity scope, we envisioned the synthesis of 2-alkenyl substituted derivatives (Compound 4, Fig. 1) to explore their biological activity against cancer cell lines and relevant bacteria from the ESKAPE group (an acronym of the most common nosocomial pathogens which includes Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter sp.) [14]. Previous publications demonstrated that the cytotoxic activity is attributed to pyruvate kinase M2 modulation [15] or tubulin inhibition [7,16], but no works concerning 2-stirylimidazo[1,2-a] pyrimidines have been described. We prepared the analogs as small molecules since this kind of candidates are pharmaceutically promising due to their facility to penetrate the cell membrane and their enhanced probability to be orally administered [17,18] Additionally, they can be optimized in terms of efficacy and toxicity without diminishing drug likeness in an easier way than bigger compounds [19].
The preparation of the compounds began with the formation of the double bond of the enone 9 (Table 1). Several aromatic aldehydes 8 were converted to the corresponding but-3-en-2-ones when reacted with acetone/water under a diluted solution of NaOH (0.025% w/v) ( Table 1). The transformation proceeded without complications and permitted the isolation of various aryl/heteroaryl ketones (9a-r) as the E isomer (all ketones revealed a coupling constant about 16 Hz, a typical value for E isomer). In the case of the biphenyl derivatives, a typical Suzuki-Miyaura cross-coupling (using Pd(PPh 3 ) 4 and K 2 CO 3 as catalyst and base, respectively) constructed the sp 2 -sp 2 C-C bond. However, when moving to the halogenation step, we ran into an obstacle. We tested several methodologies for the alpha bromination of methyl ketones (the chloro derivative 9h, R=chlorine in Table 1, was used in all the experiments of the optimization process) [22,23]. Typical halogenation with bromine afforded the dibromo ketone 11h (Table 1 [1][2][3]. Modification of the reaction conditions (solvent, quantity, and type of bromination agent) did not improve the yield of 10h (Table 1,  entries 4 and 5). Gratifyingly, when replacing the bromine source by copper bromide (II) [24], the yield of the desired mono-brominated compound improved, except when the transformation was conducted under mechanochemical synthesis (Table 1, entries 6-11). The best conditions consisted in stirring methyl ketone 10h (Table 1, R = Cl) with 1.5 equivalents of CuBr 2 in ethanol at room temperature, resulting in a 83% of the desired compound and only traces of the dibromo ketone (entry 8). Finally, a protocol that uses zinc bromide generated in situ (by reacting bromine and zinc dust in water) resulted in a futile experiment (Table 1, entry 9) [25].
Next, we proceeded to the final step of the synthesis. The Hantzsch reaction of the bromoketones 10 and 2-aminopyrimidine (12) occurred in several polar solvents like methanol, acetonitrile, and ethanol; regarding heating, both conventional and microwave irradiation afforded the cyclocondensation. Nonetheless, the best conditions for the construction of the desired products 13 (Table 2) resulted in heating the two reagents in ethanol for several hours (Table 2). After a few experiments, we concluded that the imidazo[1,2-a] pyrimidines can be obtained in moderate to good yields by reacting the unpurified bromoketone (only a filtration under a silica pad was necessary to remove the copper salts) and immediately submitting it to the Hantzsch cyclization.
The two-step protocol (starting from the butenones 9) allowed the synthesis of sixteen 2-alkenylimidazo[1,2-a] pyrimidines 13a-p in low to moderate yields (18-62%, Table 2); we attribute those yields to the unreactive 2-aminopyridine which difficulted the purification through column chromatography. Regarding reactivity, we did not find any correlation between the yield and the electronic nature of the substituents. The protocol enabled the synthesis of analogs with electron donating (13a-f), electron withdrawing (13h-j), bulky groups (13d and 13g) and heterocycles (13n and 13o). We also wanted to know the influence of the substitution in meta in terms of electronic properties (13k and 13l) and steric hindrance (13m). Although the last step gave moderated yields, the accessibility of the aldehydes required for the first step (which added several functionalities at the alkenyl moiety) and the rapidness of the methodology make it suitable for the synthesis of 2-alkenylimidazo[1,2-a]pyrimidines.

Biological evaluation of the imidazo[1,2-a] pyrimidines
Having in hand the bicycles 13a-p, we became interested in exploring the potential applications in pharmacology. According to the literature, one of the most common reported biological activities of imidazo[1,2-a]pyrimidines is related to oncology. Cancer, an abnormal and malignant proliferation of cells in an organism, leads as one of the major causes of death worldwide; furthermore, the costs for its control, morbidity and mortality are expeditiously growing [26,27]. Therefore, we tested the series against a panel of six human cancerous cell lines (some of them being the most frequently encountered cancer types) [28]: glioblastoma (U-251), prostatic adenocarcinoma (PC-3), chronic myelogenous leukemia (K562), colorectal adenocarcinoma (HCT-15), breast adenocarcinoma (MCF-7) and lung adenocarcinoma (SKLU-1). Table 3 shows the percentage of inhibition at 20 µM of the most affected cancer cell lines (prostate, leukemia, and colon), while table S1 (see supplementary material) resumes data of the remnant tumors, which were not significantly attacked by the series. Etoposide, a topoisomerase II inhibitor, was used as positive control since it affects the viability of the cell lines tested. Unsubstituted 13a and its vinylogue 13p lack of relevant activity (maximum inhibition of 19%). Neither the addition of electron donating groups (13e and 13f, with maximum activity of 23.7% against K562) nor electron withdrawing substituents (13h-j) improved the growth inhibition. Similar findings occurred when replacing the phenyl ring with the isosteric thiophene (13n, which was inactive against all the cell lines) but also with the meta regioisomers (13k and 13l). Interestingly, the replace of methoxy group in 13e to meta position (13k) slightly increased the cytotoxicity in PC-3 (from 4.4 to11.4%) and MCF-7 (from inactive to 12.2 %), but no differences were detected with the fluorine regioisomers (comparing 13i and 13l). However, when bulky groups were attached to the phenyl ring, the activity improved in almost all cell lines: aliphatic bulky groups such as t-Bu (13d, inhibition ranging from 38 to 55.3%) were active against colon and breast cancer, while a phenyl at para position (13g) showed the best activity in prostate (58%) and leukemia (76.4%) cancer. Interestingly, when moving the phenyl to meta position (example 13m), the activity in leukemia drastically dropped (from 76.4% in 13g to 30.0% in 13m) but the inhibition of prostate cancer remained (58.7%, Table 3). 13m also showed the best activity against breast cancer (52.9%) among all the series.
The IC 50 of the most active compounds resulted in 41.1 ± 2.2 μM for MCF-7 and 27.7 ± 1.0 μM for HCT-15 in the case of 13d, and 21.5 ± 1.9 μM for 13g in the prostate PC-3 cell line (Fig. 2). We omitted the IC 50 determination of 13m and 13o due to their significant toxicity in the healthy cell line COS-7 (20.9% and 21.0%, respectively (see Table  S1 in supplementary material). It is worth to note that both 13d and 13g compounds lacked toxicity against COS-7, and the activity of the biphenyl derivative 13g surpassed the inhibition observed for the control Etoposide (IC 50 for PC-3 = 31.5 ± 2.2 μM). Piperazine-containing imidazo[1,2-a] pyrimidines previously reported [15] also showed a preference for inhibiting breast cancer over colon and lung cancerous cell lines; in this case, the authors reported the inhibition of pyruvate kinase M2 -an enzyme involved in cancer invasion, division and metastasis-as a plausible anticancer mechanism. Other related imidazo[1,2-a]pyrimidines with homopiperazine linkage demonstrated good activity against the prostate cancer cell line DU-145 but lack activity in A-549 (lung cancer) [7].
In our series, the most active compounds resulted in those with an extra phenyl ring (13g and 13m). Therefore, to corroborate the influence of it we prepared two new analogs of 13g: 13q, bearing a 4-methylphenyl at para position and 13r, having a 3,5-difluorophenyl. The incorporation of two fluorine atoms at 3 and 5 position caused the total loss of the activity (13r, Table 4). In the case of 13q, the activity also significantly dropped in all the cell lines with respect to the unsubstituted analog 13g (maximum inhibition of 44.5 in MCF-7). Apparently, the cytotoxicity in cancerous cell lines is extremely susceptible to modifications at the extra phenyl ring so the synthesis of more analogs would explore the SAR of this moiety.
Even though the series herein reported was not as active as other reported imidazo[1,2-a]pyrimidines [7] nor our previous cytotoxic macrocycles [29,30], we consider that this collection may become a starting point for future works due to the feasibility for modifications on the bicycle ring and the aldehyde moiety.
Besides cytotoxicity, we evaluated the potential antimicrobial effects of the series. Recently, the WHO cataloged bacterial diseases as a serious health problem worldwide (https://www.who.int/news-room/fact-sheets/ The inhibition higher than 50% is colored in red NC Non-cytotoxic a Etoposide was tested at 50 μM detail/antibiotic-resistance), so humankind require novel treatments against multidrug-resistance organisms. Some of the most alarming pathogens belong to the ESKAPE bacteria [31], so we tested the compounds 13a-p against four of them. The growth inhibition at 50 μM is shown in Fig. 3 and Gentamicin (Gtm at 50 µM) was used as control.
Compounds were inactive against the pan-drug resistant (PDR) A. baumannii (data not shown) and only four compounds showed a reduction of Methicillin-resistant Staphylococcus aureus (MRSA) higher than 10% (those substituted with bulky groups (13c and 13d) or with fluorine atoms (13i). This Gram-positive pathogen is a common etiology of hospital-acquired infections [32] such as bacteremia, infective endocarditis, infections in skin, soft tissue, and respiratory system. The most interesting compounds against S. epidermidis -an opportunistic pathogen relevant in health care-associated infections, mainly in patients with medical devices- [33] were those unsubstituted or adorned with electron withdrawing groups (13b-e, inhibition ranging from 20 to 28%) and, similarly to the cytotoxicity test, those with an extra phenyl group (4g, 25%, and 4m, 44%) [34]. On the other hand, the best observable effect of the series occurred against the Enterobacteriaceae Klebsiella pneumoniae: with exception of the sulfured 4f (no inhibition), all compounds had a significant reduction and seven of them, surpassed 40% of inhibition (4e, 4g-j, 4k and 4n). The most active analog, 4k, showed a 63% of reduction; this outcome is significant if we consider that K. pneumoniae accumulates almost one third of hospital-acquired infections caused by Gram negative pathogens, which may complicate to endocarditis and septicemia [35]. Apparently, methoxy groups (4e and 4k with inhibition of 47.5 and 63.3%, respectively), and a phenyl (13g with 51% of inhibition) or a highly electronwithdrawing group (13j with 47%) seems to be required for the antimicrobial effect. Although the activity of the series did not surpassed the effect of gentamicin, the 2alkenylimidazo[1,2-a]pyrimidine appears as an scaffold with pertinent antimicrobial properties. We determined the IC 50 values of the compounds having an inhibition higher than 40 % against K. pneumoniae (Fig. 4). Among the series, the most active resulted in those with a methoxy group attached to the styryl moiety (43 ± 1.3 μM for 13k and 184 ± 1.5 μM for 13e) or a trifluoromethyl at para position (196 ± 1.3 μM for 13j), while the replacement of the phenyl by a 2-thiophenyl also had a beneficial effect (13n, IC 50 = 180 ± 1.7 μM). Although the biphenyl-containing analog 13g positioned as the second most active -according to the preliminary screening (Fig. 2)-its IC 50 value raised to 203 ± 1.7 μM, mainly due narrow-spectrum antimicrobials, since most of the compounds showed significant inhibition of K. pneumoniae while lower or no activity was observed against Gram positive pathogens and A. baumannii. Narrowed-spectrum antimicrobials are less probable of generating selectivity pressure against non-susceptible bacteria, becoming a current alternative for treating infectious diseases [36]. Finally, we predicted some physicochemical properties related with Lipinsky´s rule of five (Table 5) by using the SwisssADME web tool [37]. Those parameters are highly correlated with bioavailability, so they must be considered during drug discovery programs. As it was anticipated, none of the imidazo[1,2-a]pyrimidines violated any of the components (molar mass < 500 g/mol; number of hydrogenbond acceptors < 10; number of hydrogen-bond donators < 5; Log P < 5) [38], therefore being within the range for good oral adsorption. LogP values have a range of 2.57-4.55, molar mass of 222-414 and the quantity of H-bond donors/acceptors fits on the accepted value, which indicate the possibility of addition of further functionalities without reducing drug-likeness. In resume, the observed activity, as well as the feasibility for modifications and good physicochemical properties make the imidazo[1,2-a]pyrimidines 13a-r excellent candidates for further exploration.

Conclusions
Although the resemblance of the imidazo[1,2-a]pyrimidine with purine heterocycle and guanidine functionality (thus expecting numerous potential biological applications), this bicyclic system needs more exploration in terms of its biological scope. Since our recent scientific curiosity focus on this scaffold, we prepared derivatives bearing an alkenyl portion at 2-position through a three-step protocol which involved an aldolic condensation followed by an α-bromination/Hantzsch sequence.
According to our main goal, compounds were tested against six cancer cell lines and four resistant ESKAPE pathogens. We found that two compounds showed interesting cytotoxicity in cancerous cell lines without affecting COS-7 growth: 13d, against colon and breast cancer, and 13g, against prostate and leukemia. Regarding antimicrobial effect, compounds lacked effect in PDR A. baumannii and some compounds moderately affected the growth of MRSA and S. epidermidis. Conversely, the most gratifyingly results involved the inhibition of MDR K. pneumoniae, reaching 63% of inhibition for 13k. While it is true that compounds did not reach the potency of the controls, biological data showed a promissory potential of the series, especially if we consider that compounds possess different points for modification that may conduct to more interesting candidates -such as the styryl moiety and the positions 2,5 and 6 of the imidazo[1,2-a]pyrimidine core. We want to highlight the biaryl derivatives 13g and 13m, whose properties make them susceptible for optimization. Those efforts are ongoing in our lab.

General methods
Solvents and reagents were purchased from Merck-Aldrich. UV and sulfuric-vanillin were used as revealing agents. 1 H-NMR and 13 C-NMR spectra were recorded with a JEOL Eclipse-300 spectrometer and a Bruker Avance III-400 model spectrometer by using CDCl 3 or DMSO-d6 as solvents. NMR coupling constants are reported in Hertz (Hz), while chemical shifts (δ) are reported in ppm relative to the solvent signal. Signal splitting patterns are described as: singlet (s), doublet (d), triplet (t), quartet (q), quintet (quint), septet (sept), broad signal (br s), doublet of doublets (dd) or multiplet (m). Mass spectra were recorded with a JMST100LC AccuTOF LC with an ionSense DART SVP100 controller ionization source or a Jeol The MStation JMS-700.

Synthesis of butenones 9a-r
In a round-bottom flask, the corresponding benzaldehyde (1.98 mmol) was suspended in a mixture of Acetone-Water (1:1.25, 9 mL). Then, a solution of aqueous NaOH (0.05% w/v, 1.4 mL) was slowly added during 20 min. The mixture was heated to 50°C for 2 h. After completion of the reaction, the excess of acetone was removed under reduced pressure and the residue was extracted with AcOEt (2 × 10 mL). The organic layer was separated, dried with Na 2 SO 4 and purified through silica gel column chromatography in a suitable Hex-AcOEt mixture to afford the corresponding 3-but-2-enones (9a-r) in moderate to excellent yields.

Synthesis of imidazo[1,2-a]pyrimidines 13a-r
To a solution of the corresponding butenone 13a-r (0.5 mmol) in EtOH (9.8 mL), was added CuBr 2 (084 mmol) in one portion. The mixture was allowed to react at room temperature for 12-24 h. After consumption of the methylketone, the mixture was evaporated and passed through a short silica gel pad eluted with AcOEt. After remotion of the organic solvent, the residue was redissolved in EtOH (4.9 mL) and 2-aminopyrimidine (0.5 mmol) was added. The Hanztsch reaction was allowed to reflux for 6-8 h until the starting material disappeared. The presence of a very polar fluorescent compound (blue fluorescence at 365 nm), which correspond to the desired imidazo[1,2-a] pyrimidine, indicated the completion of the reaction. The ethanol was evaporated under reduced pressure and the residue was redissolved in water. The mixture was basified to pH = 8 with an aqueous solution of NaOH 1.0 M and then extracted with DCM (2 × 15 mL). The bicycles 4a-r were isolated after a flash column chromatography using AcOEt as eluent. (E)-2-(4-methylstyryl)imidazo [1,2-a]

Microdilution method for antimicrobial screening
Antibacterial assays were performed following the methodology described by the CLSI M07-A10 using a microdilution test [40]. Briefly, all tested microorganisms (Acinetobacter baumannii, Klebsiella pneumoniae, Pseudomonas aeruginosa, Staphylococcus aureus, multiresistant clinical isolates) were grown in Müeller-Hinton broth overnight. Subsequently, the bacterial suspension was adjusted to an optical density (OD600nm) of 0.08-0.13 equivalent to 1.5 × 108 UFC/mL, and then diluted 1/20 in the same media. Tested compounds 13a-p were dissolved in DMSO at 2.5 mM. Microdilution assays were carried out in a 96-well plate at a final volume of 100 μL. To each well, 10 μL of diluted bacterial suspension, 2 μL of molecules (13a-r) to a final concentration of 50 μM, and 88 μL of Müeller-Hinton broth were added, and the OD600nm was measured (t0). Afterward, the plates were incubated at 37°C for 24 h and measured again (t24, OD600nm). Data are reported as the percentage of inhibition. Gentamicin was used as positive control. The minimum inhibitory concentration (MIC) experiments were performed following the same methodology at final concentrations ranging from 0.0025 to 250 μM. MIC was calculated by non-linear regression using a modified Gompertz function [41] with GraphPad Prism (Eq. 1).
where is the lower asymptote of y (approximately zero), B is a slope parameter, C is the distance between the upper and lower asymptote (approximately 1) and M is the log concentration of the inflexion point.