Discovery of a new class of highly potent inhibitors of acid ceramidase: synthesis and structure-activity relationship (SAR).

: Acid ceramidase (AC) is an intracellular cysteine amidase that catalyzes the hydrolysis of the lipid messenger ceramide. By regulating ceramide levels in cells, AC may contribute to the regulation of cancer cell proliferation and senescence and to the response to cancer therapy. We recently identi ﬁ ed the antitumoral agent carmofur ( 4a ) as the ﬁ rst nanomolar inhibitor of intracellular AC activity (rat AC, IC 50 = 0.029 μ M). In the present work, we expanded our initial structure − activity relationship (SAR) studies around 4a by synthesizing and testing a series of 2,4-dioxopyrimidine-1-carboxamides. Our investigations provided a ﬁ rst elucidation of the structural features of uracil derivatives that are critical for AC inhibition and led us to identify the ﬁ rst single-digit nanomolar inhibitors of this enzyme. The present results con ﬁ rm that substituted 2,4-dioxopyrimidine-1-carboxamides are a novel class of potent inhibitors of AC. Selected compounds of this class may represent useful substituents. 21 This last part was carried out through the NBO 3.1 software included in the Gaussian package. All the structures were prepared using Gaussview 5.0. 26 In the alkyl side chain, methylene units beyond the ﬁ rst were omitted.


■ INTRODUCTION
Sphingosine-containing lipids are essential structural components of cell membranes that also serve important signaling functions in cell migration, cell recognition, and inflammation, as well as in the control of cell growth and differentiation. 1 Among the various groups of sphingolipids, the ceramides, amides of sphingosine with long-chain fatty acids, have attracted attention for their roles in the replication and differentiation of normal and neoplastic cells. 2 Stress-related signals such as tumor necrosis factor α (TNF-α) stimulate the production of ceramides in cells, which can cause in turn replicative senescence and apoptosis. 3 Thus, enzymes involved in the biological deactivation of ceramides have emerged as potential targets for therapeutic manipulation by small molecules. 4 Acid ceramidase (AC) is a ubiquitous cysteine amidase that operates within the acidic milieu of the lysosome to catalyze the hydrolysis of ceramides into sphingosine and fatty acid. By regulating the intracellular levels of ceramides, AC may influence the survival, growth, and death of tumor cells. 5 Consistent with this view, cells that express abnormally high levels of AC are more resistant than normal cells to pharmacological induction of programmed cell death. 6 Indeed, several types of cancer (including prostate, head and neck, colon and melanoma) overexpress AC, which is suggestive of a role for this enzyme in the resistance to radiotherapy and chemotherapy. 7 Although several AC inhibitors have been described in the literature, potent small-molecule compounds capable of inhibiting this enzyme in vivo are still needed. 4,7b,8 The majority of compounds reported so far, including oleoylethanolamide (1, OEA), B-13 (2), D-MAPP (3) (Figure 1), 7d,9 and their derivatives, are structural analogues of ceramide and therefore suffer from several limitations, including low potency and insufficient druglikeness. For example, pharmacological inhibition of AC with 1 has been related to sensitization of hepatoma cells to daunorubicin-induced cell death, but the low potency (K i ≈ 500 μM) and rapid metabolic degradation of this compound preclude its therapeutic use. 10 Compound 2 inhibits AC activity with a median inhibitory concentration (IC 50 ) of approximately 10 μM, causes apoptosis in several human cancer cell lines, and sensitizes prostate tumors to the effects of radiation. 11 Compound 3, which is structurally related to 2, inhibits both AC and neutral ceramidase activities and triggers apoptosis in squamous carcinoma cells. 12 These results suggest that AC blockade may promote or facilitate cancer cell death, especially when combined with the use of other chemotherapeutic agents. The findings also highlight the need to discover more potent and druglike compounds, which might be used to investigate further the usefulness of AC as a target for chemosensitizing drugs.
We recently reported that the 5-fluorouracil (5-FU) releasing drug carmofur (4a, Figure 2) inhibits AC activity with nanomolar potency (IC 50 for rat AC of 0.029 μM) and increases intracellular ceramide levels both in vitro and in vivo. 13 These effects are independent of carmofur's ability to generate 5-FU and are accompanied by a marked enhancement in the antiproliferative actions of other antitumoral agents. Preliminary structure−activity relationship (SAR) studies revealed that replacement of the fluorine atom of carmofur with selected chemical groups with different stereoelectronic properties leads to molecules that retain the ability to inhibit AC with high potency (e.g., 4e, IC 50 = 0.012 μM). 13 In the present report, we expanded our SAR studies around carmofur with the objective of identifying other potent smallmolecule inhibitors for this enzyme. We synthesized a series of 2,4-dioxopyrimidine-1-carboxamides to explore the role of    13) different substituents at carbons C5 and C6 and at nitrogen N3 of the uracil ring, as well as the effect of the alkyl side chain on AC inhibition (Figure 2). In addition, we conducted a quantum-mechanical study on the new molecules to explore Scheme 1. Synthesis of 5-Substituted Uracils 5i−l

■ CHEMISTRY
We explored and optimized various methods reported in the literature for the synthesis of ureas to access differently substituted 2,4-dioxopyrimidine-1-carboxamides. 14 The preparation of most 2,4-dioxopyrimidine-1-carboxamide derivatives was accomplished by coupling differently substituted uracils with alkyl isocyanates. Urea formation occurred regioselectively at position N1 under all conditions employed. Coupling reactions were conducted in pyridine at room temperature with a stoichiometric amount of DMAP (method A), in DMSO at 50°C (method B), or in pyridine under microwave irradiation at 100°C (method C). Dimethylaminopyridine was necessary for acylation reactions that proceeded incompletely or did not occur in pyridine or DMSO alone. 14b We first focused on the synthesis of carboxamide derivatives differently substituted at position 5. Table 1 shows the reaction conditions adopted for the coupling of 5-substituted uracils with n-hexyl isocyanate to afford the compound series 4f−o. Non-commercially available uracils such as 5-(benzyl(methyl)amino)uracil (5i), 5-(morpholin-4-yl)uracil (5j), 5-(4-methylpiperazin-1-yl)uracil (5k), and 5-phenyluracil (5l) were prepared as described in Scheme 1. Compounds 5i−k were obtained by nucleophilic substitution reaction of the appropriate amine on 5-bromouracil (5n) under microwave irradiation at 120°C. 15 The synthesis of 5-phenyl-1Hpyrimidine (5l) was accomplished in three steps starting from 5-iodouracil (5m). This compound was transformed into the N1-benzhydryl derivative (6m) by treatment with benzhydryl bromide in the presence of bis(trimethylsilyl)acetamide (BSA) and a catalytic amount of iodine (I 2 ). 16 Subsequent reaction with phenylboronic acid under standard Suzuki coupling conditions, followed by removal of the protecting group with 10% triflic acid solution in TFA, afforded the desired 5phenyluracil (5l) in good yield.
N-Hexyl-5-methylamino-2,4-dioxopyrimidine-1-carboxamide (4p, Table 2, Scheme S1) and N-hexyl-2,4-dioxohexahydropyrimidine-1-carboxamide (7, Table 2, Scheme S2) were obtained by catalytic hydrogenation from 4i and 4b, respectively. The N3-substituted-1-carboxamide derivatives 8a,b, 9a−10a were prepared following a three-step sequence starting from N1-Boc uracils 11a,b (Scheme 2). 17 These compounds were first reacted with the appropriate alkyl iodides in the presence of cesium carbonate to give N3-substituted 2,4-dioxopyrimidine-1carboxylates. The tert-butyloxycarbonyl group was then removed with potassium carbonate in methanol to afford 3alkyluracils 12a,b, 13a−14a which were then converted into the corresponding carboxamides 8a,b, 9a−10a following method A. Syntheses of N3-substituted uracils 15a−19a were accomplished with minor modifications of literature procedures, as reported in Scheme 3, by direct N3-acylation of N1carboxamide derivatives. No decomposition of the ureidic function occurred in agreement with what observed with N1alkoxycarbonyluracils. 18 Compound 15a was prepared starting from carmofur (4a) by reaction with isobutyryl chloride in the presence of triethylamine. The carbamoyl derivatives 16a−17a were obtained by treatment of 4a with the appropriate chloroformate in the presence of pyridine as base. The syntheses of 3-isobutyl carboxylate 18a and 3-benzoyluracil derivative 19a were carried out by treatment of 4a with isobutyl chloroformate and benzoyl chloride in pyridine as solvent and base.
The small series of compounds (20b−24b, 23a, and 23e) bearing N1-carboxamide alkyl chains of different length was prepared as shown in Scheme 4. Differently substituted uracils (5a,b, 5e) were coupled with various alkyl isocyanates following method A to afford the desired carboxamides in good yields.
An alternative synthetic procedure to access 1-carbamoyluracils was adopted for compounds 26 and 27b (Scheme 5). The 5,6-dimethyl substituted derivative 26 was obtained by reaction of 5,6-dimethyluracil (28) with triphosgene followed by treatment in situ with hexylamine. The synthesis of the Nmethylurea 27b was accomplished with a similar procedure starting from the sodium salt 29b of the uracil 5b (Scheme 5).

■ RESULTS AND DISCUSSION
The objective of the present study was to investigate substituted 2,4-dioxopyrimidine-1-carboxamides as a novel class of AC inhibitors. A preliminary SAR study on carmofur analogues (4b−e, Table 2) bearing substituents with different stereoelectronic properties at position 5 had shown that an electron-withdrawing group is crucial to achieve potent inhibition of AC activity. Accordingly, replacement of fluorine with electron-withdrawing substituents, such as chlorine and trifluoromethyl, led to the identification of two novel doubledigit nanomolar AC inhibitors (4d,e). The results obtained with this small set of compounds suggested that 2,4dioxopyrimidine-1-carboxamide might provide a promising scaffold for the discovery of potent AC inhibitors. Encouraged by these results, we decided to expand this series of compounds to elucidate the structural features that may be critical for AC inhibition.
The new substituted 2,4-dioxopyrimidine-1-carboxamides were tested for their ability to inhibit the hydrolysis of Nlauroylceramide by recombinant rat AC (r-AC). IC 50 values are reported in Tables 2−4.
The limited stability of some 2,4-dioxopyrimidine-1-carboxamide derivatives in DMSO-d 6 prompted us to ask first whether the compounds were sufficiently stable under the conditions of our assay to obtain meaningful IC 50 values. A set of molecules representative of the substitution pattern explored in our SAR analysis was selected and tested for stability in buffer over the 30 min period of the assay (Table S2). 20 Most compounds tested were stable enough to be included in the study except 4o, which was extensively hydrolyzed (22% of the initial amount remained after 30 min) and was not further investigated. The 6-substituted derivatives 26 and 30 were also excluded because of their chemical instability.
We initially focused our attention on substitutions at position 5 to expand the initial set of derivatives 4a−e (Table 2). In particular, the potency of 4e (IC 50 = 0.012 μM) prompted us to synthesize the 5-iodo, 5-bromo, and 5-cyano derivatives (4m− o) in order to further investigate the effect of electronwithdrawing groups at that position. The introduction of iodine led to a 5-fold decrease in activity (IC 50 = 0.147 μM) compared to carmofur (4a), while the 5-bromo derivative was equipotent to the 5-chloro-substituted carboxamide 4d, with an IC 50 of 0.051 μM. Next, we introduced a set of electron-donating groups at position 5 of the uracil ring. The compounds obtained (4f−h, 4p) were 14-to 100-fold less potent than carmofur (4a). Replacement of fluorine with an alkyl group was detrimental for activity. The 5-methyl derivative 4c (IC 50 = 1.46 μM) showed a 50-fold decrease in potency compared to carmofur, whereas a less pronounced decrease was observed with the 5-ethyl analogue 4f (IC 50 = 0.733 μM) and the 5hydroxymethyl derivative 4g (IC 50 = 0.417 μM). The introduction of substituents containing a heteroatom directly bound to the uracil ring led to a considerable loss in potency. The 5-methoxy derivative 4h showed low micromolar AC inhibition (IC 50 = 1.1 μM), as did the N-methylamino analogue 4p (IC 50 = 3.4 μM). Replacing the N-methylamino group with an N-methylbenzylamino moiety led to a further decrease in potency (4i, IC 50 = 5.5 μM), probably due to increased steric hindrance. We also introduced heterocyclic and aromatic rings at position 5. Consistent with the data obtained with analogues bearing a heteroatom at position 5, introduction of morpholine or N-methylpiperazine caused a decrease in potency, albeit less pronounced than that observed with the N-methylamino derivative 4p, as compounds 4j and 4k yielded IC 50 values of 0.875 and 1.4 μM, respectively. Surprisingly, introduction of a phenyl ring led to the most potent compound within this small series (4l, IC 50 = 0.177 μM), indicating that a planar group is well accepted even if it is sterically demanding.
To complete the SAR study around positions 5 and 6, we examined the role of the C5−C6 double bond. Compound 7, the dihydro derivative of 4b, showed no inhibitory activity toward AC, highlighting the functional importance of a fully conjugated 2,4-dioxopyrimidine-1-carboxamide system.
As the next step in our SAR study, we investigated the effect of substituents at position N3 of the uracil ring introducing alkyl, acyl, and carbamoyl groups at this position (Table 3) Methylation was first explored on 5-fluoro and unsubstituted 2,4-dioxopyrimidine-1-carboxamides (4a, 4b) and was found to be highly beneficial for the activity, resulting in the two doubledigit nanomolar inhibitors 8a (IC 50 = 0.013 μM) and 8b (IC 50 = 0.053 μM). Notably, an 8-fold increase in potency was observed for compound 8b vs 4b, and the 5-fluoro derivative 8a turned out to be equipotent to 4e. Considering that the 5unsubstituted derivative 8b was 4-fold less potent than 8a, we turned our attention to 5-fluoro substituted compounds. The introduction of progressively bulkier alkyl groups led only to a slight decrease in inhibitory activity, as the N3-ethyl substituted derivative 9a had an IC 50 of 0.018 μM, and compound 10a, bearing a methylcyclopropyl moiety, had an IC 50 of 0.061 μM. Comparable high potencies were obtained with acyl substituents: 15a, bearing a 2-methylpropanoyl group, and the benzoyl-substituted 19a had an IC 50 value of 0.02 μM. The effect of the N3-substitution was further explored by introducing a carbamoyl functionality, as in compounds 16a− 18a. Of particular interest was the result obtained with the methylcarbamoyl derivative 16a, which showed an IC 50 of 0.007 μM. High inhibitory potencies were retained with the more sterically demanding carbamoyl derivatives 17a and 18a, which displayed IC 50 values of 0.012 and 0.016 μM, respectively.
Next, we examined the role of the alkyl side chain of the N1carboxamide function, focusing on the influence of chain length. We chose the moderately potent 5-unsubstituted derivative 4b as reference compound to better appreciate possible changes in potency by variations in the chain length. Unsubstituted uracil derivatives 20b−24b, where the length of the aliphatic side chain was progressively increased from four to nine methylene units, were tested for their inhibitory activity; results are reported in Table 4. With respect to compound 4b (IC 50 = 0.426 μM), which bears a six-carbon alkyl chain, a decrease in chain length was associated with a drop in potency, as N-butyl-2,4-dioxopyrimidine-1-carboxamide 20b (IC 50 = 8.5 μM) was 20-fold less potent than 4b. Increasing the length of the alkyl chain from four to eight methylene units, we observed progressively higher potencies, with the N-octyl derivative 23b showing the highest activity (IC 50 = 0.283 μM) within this set of compounds. Further extension of the chain length to nine methylene units resulted, however, in a loss in potency (compound 24b, IC 50 = 0.464 μM).
The role of the 1-carboxamide NH group was investigated by preparing the N-methyl derivative 27b (Table 4). This compound showed no inhibitory activity toward AC.
Finally, we prepared and tested the 2,4-dioxopyrimidine-1carboxamides 23a, 23e, and 25a bearing at positions 5, N1, and N3 the best substituents identified in the previous SAR ( Table  4). The combination of the electron-withdrawing fluorine or trifluoromethyl group at position 5 with the n-octyl chain at N1 led to potent AC inhibitors. The 5-fluoro derivative 23a (IC 50 = 0.046 μM) turned out to be 6-fold more potent than the corresponding unsubstituted analogue 23b, while the 5trifluoromethyl compound 23e (IC 50 = 0.007 μM) showed a 40-fold improvement in potency compared to 23b. Notably, in line with our previous observations, carbamoylation of compound 23a at position N3 delivered the single-digit nanomolar inhibitor 25a, which displayed an IC 50 of 0.004 μM.
The critical importance of a fully conjugated 2,4-dioxopyrimidine-1-carboxamide system for displaying AC inhibitory activity, and the opposite effect on potency of electronwithdrawing and electron-donating substituents at position 5, prompted us to undertake a quantum mechanical study on all the reported compounds to test whether any correlation could be established between the potency of the compounds and electronic descriptors of the substituted 2,4-dioxopyrimidine-1carboxamide derivatives. The SAR exploration around position 5 found a computational counterpart in the second order perturbation theory analysis based on the natural bond orbital (NBO) approach. 21 Such technique was useful to understand how the orbital deviation from the ideal Lewis structure was modulated by different substituents and its relationship with activity (a complete list of calculated NBO energies is reported   Table S3). 22 As illustrated in Figure 3, we observed that the stabilization energy of the antibonding (non-Lewis) exocyclic N1-CO orbital linearly correlated with inhibitory potency (R 2 = 0.87). In other words, a progressively more stable and populated antibonding N1-CO orbital designates a compound that is a more potent AC inhibitor. These results also explained the importance of the uracil ring conjugated system: in the case of the saturated derivative 7, the stabilization energy (41.36 kcal/mol) was considerably lower than the energy values associated with pharmacologically active compounds. Through a partial least squares (PLS) regression technique (reported in detail in the Supporting Information), it was possible to confirm that the potency (pIC 50 ) of the 2,4-dioxopyrimidine-1carboxamide 5-substituted derivatives 4a−n and 4p and the orbital energies of the 39 valence NBOs of the common scaffold are highly correlated. The insights gathered from this analysis were in line with the stabilization energy data, assigning the most relevant contributions to the orbitals in close proximity to N1.

Journal of Medicinal Chemistry
We next extended the analysis to N3-substituted derivatives (8a,b, 9a−10a, 15a−19a; see Figure S2). With the only exception of compound 19a, the activity of which was lower than predicted by the orbital stabilization energy, a satisfactory degree of correlation was maintained between stabilization energy of the exocyclic N1-CO antibonding orbitals and inhibitory potency. The squared correlation coefficients between the stabilization energy of the N1-CO antibonding orbitals and the potency of derivatives are R 2 = 0.52 and R 2 = 0.73, including and excluding 19a, respectively (see Figure S2). It is reasonable to hypothesize that because of the bulky nature of the benzoyl substituent, the potency of 19a is reduced by a possible steric clash at the binding site of AC. A more comprehensive PLS model confirmed that when introducing N3-substituted derivatives, the overall potency variation is still mainly accounted for by contributions of the orbitals in close proximity to N1. However, it is interesting to note that in this case the orbitals surrounding N3 also play an important role.
As observed with potency, the stability of 5-substitued compounds is linearly related to the stabilization energy of the antibonding N1-CO orbital (see Figure S5). For the derivatives substituted at position 6 or N3, other electronic properties of the uracil scaffold, besides the stabilization energy, are likely to be taken into account to explain their stability. Nonetheless, it appeared that energy values below 45 kcal/mol guarantee chemical stability. In this light, the most active compounds in the series could be regarded as the result of an optimal trade-off between potency and stability.
Compounds 4l, 23b, and 8a, which showed good AC inhibition and stability, have been tested for their effect on the proliferation of the human colon adenocarcinoma cell line SW430. Cell viability was determined after a 72 h treatment schedule (see Supporting Information). Compounds 4l, 23b, and 8a decreased SW403 cell viability with EC 50 values of 20, 24, and 22 μM, respectively (Table 5). 23

■ CONCLUSIONS
We have identified and characterized 2,4-dioxopyrimidine-1carboxamides as a novel class of potent small-molecule AC inhibitors. A systematic SAR study around the uracil scaffold allowed a first elucidation of critical structural features associated with AC inhibition. Our study confirmed the importance of an electron-withdrawing group at position 5 of the uracil ring to access potent inhibitors, as previously observed. Additionally, we found that substitution at position N3 is beneficial for activity, leading to highly potent compounds, and that a 1-carboxamide alkyl chain of six to eight methylene units is important to achieve potent AC inhibition. Finally, our SAR study revealed that the C5−C6 double bond in the uracil ring and unsubstituted nitrogen in the 1-carboxamide moiety are mandatory structural features for activity. Computational studies provided a theoretical explanation for the potency of 2,4-dioxopyrimidine-1-carboxamides as AC inhibitors in terms of the electronic properties of the substituted uracil ring.
A notable result of this study is the identification of the first single-digit nanomolar inhibitors (16a, 23e, and 25a) of AC activity. Selected derivatives in this series may represent useful probes to characterize the functional roles of AC and assess the therapeutic potential of AC inhibitors in cancer and other  ■ EXPERIMENTAL SECTION a. Chemicals, Materials, and Methods. All the commercial available reagents and solvents were used as purchased from vendors without further purification. Dry solvents (pyridine, CH 2 Cl 2 , DMSO) were purchased from Sigma-Aldrich. Automated column chromatography purifications were done using a Teledyne ISCO apparatus (CombiFlash Rf) with prepacked silica gel columns of different sizes (from 4 to 40 g). Mixtures of increasing polarity of cyclohexane and ethyl acetate (EtOAc) were used as eluents. Flash column chromatography was performed manually on prepacked silica cartridges (2 or 5 g) from Biotage or on glass columns using Merck silica gel 60 (230−400 mesh) as stationary phase. Purifications by preparative HPLC−MS were run on a Waters Autopurification system consisting of a model 3100 single quadrupole mass spectrometer equipped with an electrospray ionization interface and a 2998 photodiode array detector. HPLC system included a model 2747 sample manager, model 2545 binary gradient module, system fluidic organizer, and model 515 HPLC pump. The PDA range was 210−400 nm. Purifications were performed on a XBridge Prep C 18 OBD column (100 mm × 19 mm i.d., particle size 5 μm) with a XBridge Prep C 18 (10 mm × 19 mm i.d., particle size 5 μm) Guard cartridge. Mobile phase was 10 mM NH 4 OAc in MeCN−H 2 O (95:5) at pH 5. Electrospray ionization in positive and negative modes was used.
c. Computational Methods. Electronic structure calculations were performed with Gaussian 09 software. 26 All the calculations adopted the B3LYP hybrid density functional, and a 6-311+g(d) basis set was employed. 27 All the optimizations were carried out in polarizable conductor calculation model (CPCM), and the chosen solvent was water. 28 The structures were optimized with tight convergence parameters and ultrafine grid. Finally, natural bond orbital (NBO) analysis and second-order perturbation theory analysis of Fock matrix in NBO basis were carried out to better understand the effect of the local changes in the electronic structure induced by different substituents. 21 This last part was carried out through the NBO 3.1 software included in the Gaussian package. All the structures were prepared using Gaussview 5.0. 26 In the alkyl side chain, methylene units beyond the first were omitted. Author Contributions ‡ D.P. and C.P. contributed equally to design and perform research.