A Five-Component Biginelli-Diels-Alder Cascade Reaction

A new multi-component condensation was discovered during the reaction of a urea, β-keto ester, and formaldehyde. In the presence of catalytic indium bromide, a Biginelli dihydropyrimidinone intermediate was further converted to a five-component condensation product through a formal hetero Diels-Alder reaction. The product structure was confirmed by NMR and NOE analysis, and the proposed stepwise mechanism was supported by the reaction of the Biginelli intermediate with ethyl 2-methylene-3-oxobutanoate.

The DHPM structure is the product of a reaction between a urea, β-keto ester, and an aldehyde, commonly termed the Biginelli reaction. The creation of a structurally diverse library of pharmacologically active compounds is an attractive possibility with this three-component condensation (Wan and Pan, 2012). As part of our studies on the preparation of bioactive DHPMs, in particular heat shock protein 70 (Hsp70) antagonists and agonists such as MAL3-101 and MAL1-271 (Figure 1), we explored the scope of Biginelli reactions in solution, fluorous media, and on solid support, and developed analogs with potent anticancer, antiviral, and neuroprotective properties (Wipf and Cunningham, 1995;Studer et al., 1997;Huryn et al., 2011;Manos-Turvey et al., 2016).
Standard Biginelli reactions frequently use catalytic amounts of Brønsted acids, such as hydrochloric acid (HCl) (Nagarajaiah et al., 2016). Various Lewis acid catalysts, including InCl 3 (Ranu et al., 2000), FeCl 3 (Lu and Bai, 2002), BF 3 •OEt 2 (Hu et al., 1998), ZnCl 2 (Sun et al., 2004), GaCl 3 (Yuan et al., 2017), and InBr 3 have also been used successfully (Phucho et al., 2009). While working to advance the scope of our own DHPM library, we noted the use of 10% InBr 3 as a catalyst at reflux in ethanol for 7 h to produce DHPMs (Fu et al., 2002;Martins et al., 2004). We decided to explore these reaction conditions further as they gave good yields even with formaldehyde as a reaction component. Initially, we hoped to be able to introduce substituents at the 4-position of the DHPM ring through an oxidative photochemical reaction related to the previously reported Friedel-Crafts amidoalkylation with the visible light catalyst, Ru(bpy) 3 Cl 2 (Dai et al., 2012) (Figure 2). However, we found that InBr 3 conditions promoted a new five-component condensation with formaldehyde, and decided to first investigate the scope of this process.

RESULTS AND DISCUSSION
The originally reported InBr 3 reaction conditions utilized an excess of urea and equal equivalents of β-keto ester and aldehyde. For our initial test reactions, we chose N,N ′ -dimethylurea as the limiting reagent. The reaction of N,N ′ -dimethylurea, 1.8 equivalents of ethyl acetoacetate as the β-keto ester, and 3 equivalents of formaldehyde in 95% ethanol at reflux for 7 h yielded not only the expected DHPM product 1A but also a higher molecular weight product that we did not anticipate (Figure 3). Through NMR and NOE analyses, we were able to assign its fused bicyclic structure as 1B. We hypothesized that this secondary product occurred through a five-component condensation reaction, and an in situ formal hetero Diels-Alder reaction of the expected DHPM Biginelli intermediate was in agreement with the product structure.
Prior reports of the product of a Biginelli reaction being utilized as a dienophile in a hetero Diels-Alder reaction are rare. Sharma and coworkers demonstrated the reaction of an isolated Biginelli heterocycle with N-arylidine-N ′ -methylformamidines and N-arylidine guanidine in THF (Sharma et al., 2005). Our present synthesis represents the first one pot reaction to form these highly functionalized bicyclic products from readily available precursors.
We subsequently explored a range of reaction conditions in an attempt to improve overall yield and selectivity (Figure 3). For these optimizations, we used our original reaction components N,N ′ -dimethylurea, ethyl acetoacetate, and formaldehyde ( Table 1). The traditional Brønsted acid catalyst, HCl, gave no conversion with formaldehyde (entry 2). We experimented with other Lewis acids (entries 3-6) to catalyze the reaction but saw no conversion with FeCl 3 or ZnCl 2 . Conversion was modest with both InCl 3 and AlCl 3 . Increasing the proportions of the β-keto ester and the aldehyde to 2.5 and 5 equivalents, respectively, improved the yield of both products 1A and 1B (entry 7). However, increasing or decreasing the reactant concentration (entries 8 and 9) did not improve the overall yield. Doubling the reaction time to 14.5 h also did not increase the conversion to product 1B (entry 10 vs. entry 7). After a shortened reaction time of 2.0 h, only product 1A was isolated in 28% yield (entry 11). Accordingly, entry 7 represented our optimized conditions: 1 equivalent of urea, 2.5 equivalents of β-keto ester, 5.0 equivalents of aldehyde, and 0.1 equivalents of InBr 3 at reflux conditions in ethanol (0.2 M) for 7 h provided 45% of DHPM 1A and 48% of pyranopyrimidinone 1B.
We then used these optimized conditions to further explore the scope of the reaction ( Table 2). Formaldehyde was used as the aldehyde component in all reactions since preliminary trials with other aldehydes and ketones were not successful in producing any fused bicycles B. We also used symmetrical ureas exclusively to avoid possible regioisomers (exploratory reactions confirmed a lack of regioselectivity with unsymmetrical ureas; for example, while 1-methylurea and 1-(4-methoxyphenyl)-3methylurea provided both the Biginelli DHPM and the Diels-Alder products, we were unable to separate the regioisomers formed in an approximately 1:1 ratio). Curiously, the reaction with the N,N ′ -dimethylthiourea stopped at the Biginelli product 2A, which was isolated in 61% yield (entry 2). Methyl acetoacetate with thiourea 2 (entry 3) also gave similar results, but product 3A was formed in lower yield, likely due to trans-esterification with ethanol. Utilizing benzyl acetoacetate, we were able to isolate both Biginelli (4A, 5A) and five-component condensation products (4B, 5B) in good overall yields with the urea and thiourea derivatives (entries 4 and 5). However, the yield of the five-component condensation Diels-Alder product 4B was considerably higher with the urea derivative than the pyranopyrimidinethione 5B resulting from the thiourea. With allyl acetoacetate and N,N ′ -dimethylurea, we were able to isolate the five-component condensation product 6B, but not the corresponding Biginelli intermediate, as it proved unstable. We also attempted to replace the traditional β-keto ester with 5,5-dimethyl-1,3-cyclohexanedione (dimedone) to probe the feasibility of adding an additional ring in the hetero Diels-Alder reaction (Figure 4). Unfortunately, with both N,N ′ -dimethylurea and its thiourea equivalent, we were only able to isolate Biginelli products 7A and 8A, in significantly lower yields than for the β-keto ester conversions.   Table 1.
To confirm our hypothesis that the five-component condensation product formed through a hetero Diels-Alder reaction with the DHPM (Biginelli) intermediate, we reacted Biginelli product 1A with ethyl 2-methylene-3-oxobutanoate ( Figure 5). After stirring in sulfolane at room temperature for 18 h, we isolated the corresponding hetero Diels-Alder Product, 1B, in 11% yield. We were also able to repeat this conversion with the Biginelli intermediate 5A to form the hetero Diels-Alder Product 9B.
Interestingly, when the five-component condensation product 1B was subjected to Krapcho dealkylation conditions with LiCl in DMSO, we isolated the DHPM product, 1A, in 87% yield, presumable through a retro Diels-Alder reaction (Figure 6) (Krapcho et al., 1967). Figure 7 illustrates our proposed mechanism for the fivecomponent condensation reaction. In the presence of a Lewis acid, condensation of the urea with the aldehyde and subsequent loss of water generates a sufficiently reactive electrophile for attack by the β-keto ester, which then undergoes cyclization to give 1A. The excess β-keto ester and formaldehyde react to form a methylene group at the α-position, and this electrophile then undergoes a hetero Diels-Alder reaction with the intermediate DHPM 1A to provide the fused heterocyclic 1B as a single stereoisomer. We determined the configuration of the methyl group to be cis to the ester at the two quaternary ring fusion atoms by converting the ester to the primary alcohol 10 with LiBH 4 , and then using a NOE analysis to confirm that the methyl group protons showed a >10% percent double resonance enhancement with the newly formed CH 2 protons.

CONCLUSIONS
We have discovered and optimized experimental conditions for a novel one pot, five-component condensation reaction with a β-keto ester, urea, and formaldehyde. The reaction appears to proceed through an intermediate DHPM (Biginelli) product. The DHPM then reacts with the condensation product of the β-keto ester and formaldehyde through a formal hetero Diels-Alder reaction. While the scope of this new process is still limited to formaldehyde and symmetrical N,N ′ -dialkylated ureas, it provides easy access to bicyclic ring systems that were previously inaccessible through a single transformation.

Diels-Alder Reaction of DHPM Intermediate to Form 5-Component Condensation Product:
Ethyl 2-methyl-3-oxo-2-(phenylthio)butanoate. A stirred solution of N-chlorosuccinimide (6.21 g, 45.1 mmol) in CH 2 Cl 2 (66 mL) was treated with thiophenol (0.40 mL, 3.8 mmol) and heated to reflux. After the reaction mixture changed color from yellow to orange, indicating the initiation of the reaction, additional thiophenol (4.24 mL, 41.2 mmol) was added dropwise to maintain a gentle reflux. The mixture was then allowed to cool to RT. After 1 h, the mixture was further cooled to 4 • C on an ice bath, and ethyl 2-methylacetoacetate (6.81 mL, 46.7 mmol) was added over a 30 min period. After warming to RT and stirring for an additional 30 min, HCl was removed by bubbling nitrogen through the product solution and the solvent was removed under reduced pressure. The residue was suspended in petroleum ether (44 mL) and filtered. The resulting solid filtrate was washed with 5 portions (44 mL each) of petroleum ether. The combined organic fractions were combined and concentrated to give ethyl 2-methyl-3-oxo-2-(phenylthio)butanoate (11.6 g, 46.1 mmol) as a dark yellow liquid in quantitative yield with a small impurity. This product was used for the next step without further purification: 1 H NMR (400 MHz, CDCl 3 ) δ 7. 51-7.30 (m, 5 H), 4.28-4.19 (m, 2 H), 2.37 (s, 3 H), 1.49 (s, 3 H), 1.32-1.26 (m, 3 H).

AUTHOR CONTRIBUTIONS
TM, MF, and MR worked on the presented work under the guidance of PW. The manuscript was written by TM and PW with input from all authors.