Benchtop 19F Nuclear Magnetic Resonance (NMR) Spectroscopy Provides Mechanistic Insight into the Biginelli Condensation toward the Chemical Synthesis of Novel Trifluorinated Dihydro- and Tetrahydropyrimidinones as Antiproliferative Agents

Benchtop nuclear magnetic resonance (NMR) spectroscopy has enabled the monitoring and optimization of chemical transformations while simultaneously providing kinetic, mechanistic, and structural insight into reaction pathways with quantitative precision. Moreover, benchtop NMR proton lock capabilities further allow for rapid and convenient monitoring of various organic reactions in real time, as the use of deuterated solvents is not required. The complementary role of 19F NMR-based kinetic monitoring in the fluorination of bioactive compounds has many benefits in the drug discovery process since fluorinated motifs additionally improve drug pharmacology. In this study, 19F NMR spectroscopy was utilized to monitor the synthesis of novel trifluorinated analogs of monastrol, a small molecule dihydropyrimidinone kinesin-Eg5 inhibitor, and to probe the mechanism of the Biginelli cyclocondensation, a multicomponent reaction used to synthesize dihydropyrimidinone and tetrahydropyrimidinones through a Bronsted- or Lewis-acid catalyzed cyclocondensation between ethyl acetoacetate, thiourea, and an aryl aldehyde. In the present study, a trifluorinated ketoester serves a dual purpose as being the source of the trifluoromethyl group in our fluorinated dihydropyrimidinones and as a spectroscopic handle for real-time reaction monitoring and tracking of reactive intermediates by 19F NMR. Further, upon extending this workflow to a diverse array of 3- and 4-substituted aryl aldehydes, we were able to derive Hammett linear free energy relationships (LFER) to determine stereoelectronic effects of para- and meta-substituted aryl aldehydes to corresponding reaction rates and mechanistic routes. In addition, we used density functional theory (DFT) calculations to corroborate our experimental results through the thermodynamic values of key intermediates in each mechanism. Finally, these studies culminate in the synthesis of a novel trifluorinated analog of monastrol and its subsequent biological evaluation in vitro. More broadly, we show an application of benchtop 19F NMR spectroscopy as an analytical tool in the real-time investigation of a mechanistically and chemically complex multicomponent reaction mixture.


■ INTRODUCTION
Multicomponent reactions (MCRs) have enabled the rapid chemical synthesis of complex and diverse molecular scaffolds, and this has, in turn, accelerated the discovery of small molecules with therapeutic potential. 1,2 The Biginelli cyclocondensation is a three-component acid-catalyzed MCR between an aldehyde, β-ketoester, and urea, which has been used to produce dihydropyrimidinone and tetrahydropyrimidinone heterocycles. 3−5 One such dihydropyrimidinone, monastrol (1), was found to be a potent inhibitor of kinesin Eg5, a motor protein necessary for the assembly of mitotic spindle fibers, thereby inducing apoptosis by preventing the development of spindle bipolarity and arresting cells in mitosis. 6−8 This discovery has inspired the synthesis of other antiproliferative dihydropyrimidinone scaffolds, including LaSOM-63 (2), which, unlike monastrol, induces apoptosis of glioma cells through the inhibition of ecto-5′nucleotidase/ CD73 activity. 9 The broad biological applicability of these scaffolds accessed through the Biginelli cyclocondensation highlights the utility of this MCR as a synthetic platform for the discovery of bioactive small molecules.
The Biginelli cyclocondensation proceeds through three possible mechanistic pathways: the imine route, involving initial Schiff base condensation between the aldehyde and urea; the enamine route, involving initial condensation between the urea and β-ketoester; and the Knoevenagel route, involving initial aldol condensation between the aldehyde and βketoester. 4 The mechanistic investigation of the Biginelli condensation has been complicated by short-lived reactive intermediates, which are difficult, if not impossible, to isolate, as well as the possibility for multiple competing mechanistic pathways. Previously, studies by Ramos and co-workers using electrospray mass spectrometry have suggested that the imine pathway is preferred when the reaction is performed with conventional β-ketoesters. 10 We decided to synthesize the trifluorinated counterparts of monastrol and LaSOM-63 because of the bioorthogonal nature of fluorine and the increased metabolic stability offered by the presence of carbon−fluorine bonds. 11−19 Of note, while it has been reported that the substitution of a β-ketoester with a 4,4,4-trifluorinated ketoester can be used to prepare trifluorinated versions of these dihydropyrimidinones and tetrahydropyrimidinones through the Biginelli reaction, the operative mechanism through which the Biginelli proceeds with 4,4,4trifluoroketoesters has not been investigated. Recent advances in benchtop nuclear magnetic resonance (NMR) spectroscopy have allowed for noncanonical applications of NMR spectros-copy as an analytical tool directly in a synthetic laboratory setting, including real-time reaction monitoring. 20−23 Here, we utilize benchtop 19 F NMR spectroscopy to probe the mechanism of the Biginelli cyclocondensation of trifluorinated ketoesters by tracking the production of various aryl-substituted trifluorinated tetrahydropyrimidinones and their reactive intermediates in real time. In this study, a library of various 3-and 4-substituted aromatic aldehydes not only gave access to a diverse library of 6-aryl trifluorinated tetrahydroprymidinones but also provided mechanistic insight into the Biginelli through meta-and para-Hammett linear free energy relationships (LFER) derived from 19 F NMR spectra of crude reaction mixtures. 24,25 While quantitative measurement of chemical species present in a reaction mixture might be possible with other means, such as high-performance liquid chromatography (HPLC), LC-MS, or GC-MS; the quantification of such species by 19 F NMR does not require any reaction workup or further sample processing and can be done within seconds or minutes of set time points, in a manner that is not invasive to the reaction mixture and therefore allows for direct measurement of species of interest in an unadulterated setting. Moreover, the ability to track a reaction by NMR is readily adapted for flow monitoring conditions. 26,27 Additionally, we employed computational modeling by density functional theory (DFT) calculations to further probe the thermodynamics of the Biginelli reaction involving trifluorinated ketoesters. 28,29 Finally, the antiproliferative efficacy of a selection of our compounds was explored in cancer cells. On a broader scope, we describe a workflow that may be applied to gain mechanistic understanding or to optimize the synthesis of other fluorinated compounds with potential medicinal significance.

■ RESULTS AND DISCUSSION
Our interest in trifluorinated tetrahydropyrimidinones and dihydropyrimidinones was initially motivated out of a desire to prepare trifluorinated analogs of monastrol and other dihydropyrimidinones and tetrahydropyrimidinones (Scheme 1a). Given the modularity of the Biginelli reaction, we employed an ethyl 4,4,4-trifluoroacetoacetate as the origin of the trifluoromethyl group (Scheme 1b). From the onset, we found that the trifluorinated ketoester was a convenient spectroscopic handle on benchtop 19 F nuclear magnetic resonance (NMR) spectroscopy for real-time, quantitative analysis of the rates of formation and absolute concentrations of reaction intermediates and product formation as a function of time. Since these instruments do not require deuterated solvents with proton lock capabilities, crude reaction mixtures or reaction aliquots could be directly analyzed in the instrument without further processing, and on an analytical scale, reactions could themselves be performed in standard 5 mm NMR tubes for rapid, high throughput reaction condition screening and optimization (Scheme 1c), and indeed we previously reported the use of 19 F NMR spectroscopy for reaction condition optimization for a scalable synthesis of other fluorinated small molecules. 30 To make this workflow quantitative, reaction mixtures were spiked with an internal standard, and for the purposes of this study, α,α,αtrifluorotoluene was selected over hexafluorobenzene as an internal standard. An initial screen of a variety of Lewis acid catalysts revealed that 8 mol % ytterbium(III) triflate in acetonitrile most efficiently catalyzed the Biginelli condensation between ethyl 4,4,4,-trifluoroacetoacetate, thiourea, and 3hydroxybenzaldehyde to the corresponding tetrahydropyrimidinone (Table 1). 31−33 As expected and in concordance with previous literature reports, the Biginelli cyclocondensation involving trifluorinated keto esters failed to undergo final dehydration into the dihydropyrimidinone, instead terminating at the tetrahydropyrimidinone, and this was ubiquitously observed across a select series of tested Lewis acid catalysts. From here, we followed a procedure utilizing excess p-toluenesulfonic acid (pTsOH, Scheme 2) in toluene established by Agbaje et al. 34 to dehydrate the tetrahydropyrimidinone (compound 4). Notably, a dehydrated trifluoromonastrol compound has not been previously reported in the literature. 35−37 Other dehydration conditions screened ( Table 2) did not give the desired product.
The synthesis of the tetrahydropyrimidinones can proceed through three possible mechanisms depending on which two reagents react first: the iminium mechanism, the enamine mechanism, and the Knoevenagel mechanism (Scheme 3). In prior studies, the synthesis of monastrol has been observed to occur through the iminium mechanism, wherein thiourea condenses with the benzaldehyde carbonyl, resulting in a Schiff base, which undergoes subsequent addition with the ketoester, which then undergoes a final 6-exo-trig cyclization to form the final tetrahydropyrimidinone. Using 19 F NMR, we were able to observe both the tetrahydropyrimidinone product (δ = −82.73   Figure 2c). Additionally, in the absence of the ytterbium catalyst, the Knoevenagel condensation product K-2 was the only observed intermediate, but this ultimately did not lead to any discernible production of the cyclized tetrahydropyrimidinone product even under extended reaction times. The necessity of the ytterbium(III) triflate catalyst in driving the operative mechanistic pathway might be rationalized by coordination of the metal cation with the 1,3dicarbonyl of the fluorinated ketoester, thereby activating the α,α,α-trifluoroketone to nucleophilic attack by thiourea.
After completing the synthesis of trifluoromonastrol, we then pursued the preparation of an analogous trifluorinated dihydropyrimidinone compound with a 4-N,N-dimethylamino aryl fragment in efforts to prepare a trifluorinated analog of LaSOM-63. During this synthetic process, we observed a drastically different rate of formation of intermediates and products in comparison to the synthesis of trifluoromonastrol, as observed by time course 19 F NMR of the crude reaction mixture. Intrigued by the difference an aryl substituent had on the reaction, we synthesized a library of trifluorinated tetrahydropyrimidinone with differing meta-and para-aryl substituents, including pyrrolidine, dimethyl, methoxy, hydroxy, methyl, bromo, chloro, fluoro, cyano, and nitro groups to investigate the role of aryl stereoelectronics on the reaction pathway (Scheme 4).
Reaction rates determined by aliquot 19 F qNMR were then used to derive Hammett linear free energy plots (Figure 3). Upon analysis of the effect of the aryl aldehyde Hammett values on the initial rate of product formation as a function as determined by quantitative 19 F NMR of timecourse reaction aliquots, we found a consistent negative Hammett linear free Scheme 2. Synthesis of the Trifluorinated Tetrahydropyrimidinones through the Biginelli Cyclocondensation, a One-Pot Multicomponent Reaction That Reacts Ethyl Acetoacetate, an Aryl Aldehyde, and Thiourea to Form a Dihydropyrimidinone (Tetrahydropyrimidinone When Using Trifluoroethylacetoacetate) a a Subsequent dehydration of tetrahydropyrimidinone analog of trifluoromonastrol proceeds with excess p-toluenesulfonic acid in reflux with 36% yield.  energy relationship in both 3-and 4-substituted aryl aldehyde series. This suggests that the rate-determining step results in an increase in positive charge in the transition state at the benzylic reaction center position, which we postulated might be the E1cb β-hydroxy elimination from E-5 to E-6, which would proceed through partial carbocation character at the benzylic carbon.
We further probed the thermodynamics of the three possible mechanistic pathways by performing density functional theory (DFT) free energy calculations on various reactive inter-  mediates en route to the tetrahydropyrimidinone in each of the mechanistic possibilities (Figure 4). We observed that, regardless of the identity of the aryl aldehyde modeled, the initial thiourea addition to the α,α,α-trifluoroketone to form tetrahydropyrimidinone E-1 was the most thermodynamically favorable (ΔG = −1.4 kcal/mol) pathway, which is consistent with our mechanistic hypothesis that the trifluorinated keto ester outcompetes the aryl aldehyde as the best electrophile, thereby committing the Biginelli cyclocondensation to the enamine route. This result, as shown by computer modeling, is consistent with our observation of the initial accumulation of the intermediate identified as the enamine (δ = −78.50 ppm in CH 3 CN), regardless of the identity of the aryl aldehyde. Moreover, we observed that the hydroxyl elimination prior to the final cyclization was highly thermodynamically unfavorable; this is consistent with the observed Hammett LFER, which pointed to this elimination as the rate-determining step. The rate of the final cyclization is dependent on the concentration of the dehydrated iminoester E-6, whose rate is, in turn, dependent on the equilibrium of the preceding β-hydroxy elimination step.
Moreover, we used density functional theory to probe the relative thermodynamics of the final dehydration step between monastrol and trifluoromonastrol. The ΔG of dehydration of  The LFER plots for the para-substituted tetrahydropyrimidinones. A negative line of best fit between the para-Hammett values (σ p ) and the reaction rate also indicates a buildup of positive charge/loss of negative charge at the benzylic center in the transition state of the rate-determining step. Rates of relative product formation were normalized with respect to the observed rate of formation with the unsubstituted benzaldehyde. All kinetics experiments were performed in triplicate with two operators to ensure reproducibility. the tetrahydropyrimidinone to synthesize monastrol was only 2.716 kcal/mol, while the change in energy for the dehydration of the tetrahydropyrimidinone to form trifluoromonastrol was 14.640 kcal/mol, accounting for the differences observed in synthesizing the dehydrated product.
Of the four possible diastereomers of the final tetrahydropyrimidinone, the relative stereochemistry was determined by both J-coupling constants of the two methine protons in the 1 H NMR as well as by density functional theory calculations. The lowest energy diastereomer (compound 3c), with a DFTcalculated dihedral angle of 172.8°between the two methine protons, had a predicted J value of 11.58 Hz, and this was consistent with our observed NMR spectra in both DMSO-d 6 and methanol-d 4 (see Supporting Information Figure S4). Moreover, this stereochemical assignment was consistent with previous literature reports. 34 We measured the antiproliferative activity of monastrol, trifluoromonastrol, and its hydrated tetrahydropyrimidinone against HCT-116 cells and found that monastrol had an EC 50 value of 0.0834 mM, and trifluoromonastrol had an EC 50 value of 0.291 mM. Trifluoromonastrol had trivially lower antiproliferative activity in comparison to monastrol, suggesting that the addition of the trifluoromethyl group results in only a modest loss in biological potency. Future studies on trifluoromonastrol's antiproliferative activity in other cell lines, and exploration of its ADMET properties, are currently underway.

■ CONCLUSIONS
In summary, en route to the preparation of a novel trifluorinated analog of monastrol, an antiproliferative dihydropyrimidinone small molecule, we determined that the Biginelli multicomponent reaction proceeds through an enamine mechanism preferentially over the imine mechanism when a 4,4,4-trifluoro-β-ketoester is used in lieu of conventional, nonfluorinated β-ketoesters. The mechanistic investigation of this reaction was enabled by benchtop 19 F NMR spectroscopy, with which the rates of formation and reaction of fluorinated intermediates and products could be quantified. With this workflow, we show that ytterbium(III) triflate effectively catalyzes the Biginelli cyclocondensation of 4,4,4trifluorinated β-ketoesters, aryl aldehydes, and thiourea into tetrahydropyrimidinones and that this proceeds through initial condensation of thiourea and the trifluorinated ketoester. This stands in contrast to earlier reports of the Biginelli cyclocondensation preferring the imine mechanism pathway, and this might be rationalized by the inductive activation by the neighboring fluorine atoms on the ketoester such that this outcompetes the aryl aldehyde as a preferred electrophile for initial condensation with thiourea. Also enabled by this workflow, the relative rates of this reaction observed by timecourse 19 F qNMR show a distinct negative Hammett linear free energy correlation, corresponding to a decrease in positive charge at the benzylic carbon in the transition state of the rate-determining step. Both of these observations were consistent with DFT calculations of reaction pathways.
Finally, this workflow has led to the rapid synthesis of a library of 19 tetrahydropyrimidinones and a novel trifluorinated analog of monastrol, which was subsequently evaluated for antiproliferative activity against nonfluorinated monastrol. On a broader level, we demonstrate the versatility of benchtop heteronuclear NMR spectroscopy as a quantitative analytical tool for the mechanistic investigation of complex reaction mixtures, and we envision this workflow might be applied to the investigation of other mechanistically complex chemical systems or to flow chemistry for real-time reaction monitoring on scale, ultimately providing access to rapid reaction optimization for medicinally relevant small molecule scaffolds. Such studies are currently underway in our laboratory.

■ MATERIALS AND METHODS
General Biginelli Synthesis Procedure (on Scale). To a reaction vial charged with a Teflon stir bar under a positive pressure nitrogen atmosphere was added ethyl 4,4,4-trifluoroacetoacetate (AK Scientific, 1.32 mmol, 193 μL), Yb(OTf) 3 (AK Scientific, 0.2 mmol, 124 mg), thiourea (Himedia, 1 mmol, 76.1 mg), and the substituted benzaldehyde stock solution (1 mmol, N/A). The reaction was stirred at 60°C under nitrogen and monitored by 19 F NMR (1−5 h). Upon completion, a majority of the acetonitrile was removed under reduced pressure, and the resulting crude reaction mixture was extracted in ethyl acetate (2 × 25 mL) over brine. The organic layers were combined and dried over anhydrous magnesium sulfate.
Purification Protocol A. The filtered product was directly concentrated in vacuo and recrystallized in cold acetonitrile to yield a white crystalline solid.
Purification Protocol B. The 4-pyrrolidine and 4dimethylamino substituted compounds did not recrystallize in acetonitrile, and so the filtered product was purified via column chromatography with a hexanes and ethyl acetate eluent (100% hexanes to 3:2 EtOAc/hexanes) to afford pure products 3j and 3k.
General Biginelli Synthesis Procedure (Micro Scale). To an NMR tube was sequentially added a solution of Yb(OTf) 3  , and a solution of trifluorotoluene in acetonitrile (AK Scientific, 4M, 0.05 mL). The NMR tube headspace was flushed with nitrogen, and an initial 19 F NMR using 16 scans and a 1.0 s scan delay was taken. The reaction mixture was placed into a 60°C water bath, and 19 F NMR spectra were taken in 30-minute intervals over a 3 hour period.
Kinetics Parameters. Kinetics were run in a water bath heated to a temperature of 60°C on a heat plate in duplicate. NMR spectra were taken with 200 ppm spectral width and a 1.0 s scan delay.