Synthesis and Biological Evaluation of Spirocyclic Chromane Derivatives as a Potential Treatment of Prostate Cancer

As a significant co-activator involved in cell cycle and cell growth, differentiation and development, p300/CBP has shown extraordinary potential target in cancer therapy. Herein we designed new compounds from the lead compound A-485 based on molecular dynamic simulations. A series of new spirocyclic chroman derivatives was prepared, characterized and proven to be a potential treatment of prostate cancer. The most potent compound B16 inhibited the proliferation of enzalutamide-resistant 22Rv1 cells with an IC50 value of 96 nM. Furthermore, compounds B16–P2 displayed favorable overall pharmacokinetic profiles, and better tumor growth inhibition than A-485 in an in vivo xenograft model.


Introduction
The acetylation and deacetylation of histones is a kind of reversible post-translational modification (PTM), which plays an important role in the regulation of gene expression in eukaryotic cells and is the arena of epigenetics [1][2][3]. The modification of the highly flexible and modifiable N-terminus of histones changes the state of histones, leading to genetic information transmission such as chromatin formation, transcription and DNA replication [4].
Histone acetyltransferase (HAT) and histone deacetylase (HDAC) are capable of affecting acetylation and deacetylation of histones, whose recruitment and normal function is the key regulatory step for gene expression and the cell cycle [3]. Functional deficiency of these enzymes may result in numerous diseases including tumors [5][6][7]. E1A binding protein of 300 kDa (p300) and CREB binding protein (CBP) are two main members of the HAT family. p300/CBP is involved in cell cycle progression and cell growth, differentiation and development, and is a significant co-activator [8][9][10][11][12][13]. By promoting histone acetylation to form an open chromatin structure, the transcription structure can be easily combined with chromatin to improve transcription activity [1,14]. Research showed that p300/CBP is highly expressed and activated in many different diseases, especially malignant tumor [5,9,15,16]. As a result, it is of great importance to develop p300/CBP inhibitors with high selectivity and potency [14][15][16].
Early exploration of HAT inhibitors included bisubstrate analogs and natural products, such as Lys-CoA (1) and anacardic acid (2) (Figure 1), which showed weak potency or poor selectivity [17,18]. Several inhibitors were discovered over time and C646 (3) became a commonly used tool molecule [19,20]. In 2017, Abbvie scientists discovered a potent and selective HAT inhibitor, A-485 (4), using high-throughput screening in silico and experimental optimization [21]. It competed with acetyl-CoA for the active site of p300, and selectively inhibited the proliferation of haematological and solid cancer cell lines, such as castration-resistant androgen receptor-positive prostate cancer [22,23]. In addition, A-485 also showed a certain positive effect on liver damage and metabolic diseases [24,25]. On the basis of A-485, Zhou et al. identified a preclinical candidate B026 (5) with excellent pharmacological properties with the assistance of artificial intelligence and further optimization [26]. In the past decade, several other HAT inhibitors 6-8 with excellent potency have been reported [27][28][29][30][31]. A-485 represents a significant breakthrough, but the cell potency still requires improvement. To better understand the binding mode of A485 and HAT domain, molecular dynamics studies were performed. To evaluate the quantitative contribution energy of the residues in the complex, the most practical MM-PBSA (GBSA) methods are employed [32][33][34]. The RMSD value of the protein backbone was calculated in a trajectory from 0 to 100 ns. As shown in Figure 2A, dynamic equilibrium was attained after approximately 40 ns of simulation. The protein-ligand complex was converged to approximately 1.8 Å. Based on the relatively stable conformation, the two N atoms of methylurea group formed strong hydrogen bonds with Gln1455 and His1451 ( Figure 2B,C). In contrast, the hydrogen bond between the carbonyl of spirocyclic lactam and Ser1400 became weaker after dynamics simulation (3.15 Å versus 2.94 Å). The 1,1,1-trifluoroisopropyl moiety was located in the lipophilic pocket formed by crucial Trp1466, together with Thr1411 and Pro1458. A large hydrophobic pocket formed by Val1401, His1451 and Pro1440 was also observed near the indenyl group. These two pockets offer sufficient imaginable space for molecular design. Therefore, we designed several series of new compounds ( Figure 3). First, phenyl or cyclopropyl groups were installed onto the indenyl to match the space and strengthen hydrophobicity. Second, a bioisosterism strategy was utilized in the spirocyclic core structure to acquire intellectual property, namely turning the carbamate into a lactam and the benzene into a pyridine. Finally, in consideration of the important hydrophobicity of Trp1466, we hypothesized that cyclization of the 1,1,1-trifluoroisopropyl and/or with the fluorobenzyl group based on conformational restriction strategy might benefit the affinity. The side chain of R group was expected to regulate the physicochemical properties.

2.2.
In Vitro Antiproliferative Activities: Structure-Activity Relationship Study 22Rv1 cells was chosen to evaluate the bioactivity in vitro, due to its AR-dependence but proliferation in the absence of androgens, and expression of a high level of AR-V7 variants. It was also representative of enzalutamide-resistant castration-resistant prostate cancer and unmet medical need furthermore [35,36]. As shown in Table 1, introducing a phenyl on the indenyl ring proved to be unsatisfactory, given the loss of antiproliferation activity in A1, indicating that this cavity is not large enough to accommodate the benzene ring. Attachment of smaller rings such as cyclopropyl (compound A2) was tolerated, showing comparable antiproliferation activity, but modification of the side chain led to decreased potency (compound A3). We also sought to construct an oxa-containing six-membered ring between the methylpyrazole and the benzene ring to restrict conformation (compound A4), but potency was lost. It was noteworthy that the dihedral angle between the indenyl and the urea moiety of A-486 was obviously bigger after dynamics simulation (54.9 • versus 27.6 • ), consistent with its lack of activity. Finally, changing the carbamate into a lactam (A5, A6) or benzene to pyridine (A7, A8) led to sharp decreases in potency. This was likely due to two subtle interactions, namely edge to face π-π stacking betwen the indenyl and His1451, and hydrogen bond between the carbamate carbonyl and Ser1400, respectively.  To optimize these spirocyclic chromane derivatives further, we turned our attention toward the third series of designs. As shown in Table 2, three substituents R 1 , R 2 , and R 3 were systematically investigated to further explore the structure-activity relationships. Changing fluorophenyl to 2,2-difluorobenzo[d] [1,3] dioxole (B1) led to 10-fold less potency than A-485. A two bridged ring was also incompatible (compounds B2, B3), indicating that the size and rigidity of the ring had a significant effect on the potency. When the 1,1,1trifluoroisopropyl group was displaced by an indenyl, several side chains were tolerated (compounds B4-B7), which confirmed our hypothesis of their regulatory effect on the physicochemical properties. Then an acetamide pyrazolyl was added to explore the displacement of 1,1,1-trifluoroisopropyl and 4-F-benzyl (B8-B16). An indenyl 5-position fluorine, 4-position pyridylation, and F-pyridylation of R 3 , all led to a 5-fold increased potency (cf. B11, B12 versus B10). The activity was getting stronger along the trend of 4,5,6,7-tetrahydro-1H-indazole, 1,2,3,4-tetrahydronaphthalene and chromane (compounds B14-B16). B16 exhibited 2-fold more potency than A-485. Combining R 2 and R 3 into one group seemed to be another way to get a new series of molecules, so B17 and B18 were designed, but the bioactivity was reduced dramatically, likely due to mismatch of the hydrophobic pocket.

Pharmacokinetic Evaluation of Compounds B8 and B16
In order to evaluate the druggability of these newly synthesized chromane derivatives, a pharmacokinetic test was performed on several potent compounds (Table 3). After administering 10 mg/kg orally to mice the half-life of B8 reached 23.9 h. The lower plasma concentration of B8 was probably attributable to the much larger volume of distribution (V ss = 13.2 h·ng/mL). Compound B16 showed an modest clearance rate (Cl = 1.8 L/h/kg), volume of distribution (Vss = 1.0 h·ng/mL) and good oral bioavailability (F = 33.2%). Due to the fact p300/CBP are ubiquitously expressed and involved in modulating immune responses [37], the plasma exposure of epigenetic regulator would be moderately controlled. Thus the pharmacokinetic profile of B16 met the criteria of druggability. Table 3. Pharmacokinetic profiles of selected compounds a .  Considering B16 was a racemic mixuture, we carried out phsical resolution by chiral HPLC and further investigated by 22Rv1 cell antiproliferation and pharmacokinetics profiles (Table 4). However, both B16-P1 and B16-P2 were identical and comparable to the racemate B16. Based on their higher maximal inhibitory efficiency (TOP value, Emax) of cell proliferation and availability, B16-P2 was chosen for further study. a Values are means of two independent experiments (cell antiproliferation, n = 2) and three independent experiments (PK, n = 3), * means multiplication sign.

Anti-Tumor Efficacy of Compound B16-P2 In Vivo
The anti-tumor activity of B16-P2 was then evaluated in vivo, using a 22Rv1 xenograft model ( Figure 4). All compounds were administered orally once daily for 15 days, with A-485 as a reference. A statistically significant tumor growth delay was observed in the B16-P2 group (p < 0.05). At the same dosage of 80 mg/kg, B16-P2 showed stronger efficacy than A-485 (tumor growth inhibition rate = 42% versus 35%), thus confirming our aforementioned conclusion that the plasma exposure was not a decisive factor. During the study period, B16-P2 was well tolerated, and no death or significant loss of body weight was observed.

Molecular Docking Study of Compound B16
To explore the possible binding mode and rationalize the observed potency of spirocyclic chromane derivative B16, molecular docking with p300 HAT domain was performed using a single isomer of spiro chirality ( Figure 5). These two structures overlapped well with structural maintenance of two key hydrogen bonds: carbonyl of the oxazolidinedione and Ser1400, C-H . . . π between 4-F-phenyl and Leu1398. In the side chain region, N-methyl-2-(1H-pyrazol-1-yl)acetamide of B16 formed NH . . . O hydrogen bond with Gln1455 and C-H . . . π bond with His1451, while two equivalent hydrogen bonds formed between methylurea of A-485 and the carbonyl of Gln1455. Importantly, an additional hydrogen bond was observed between the chromane oxygen of B16 and Arg1462, which might contribute to the improved activity. Thus this series of spirocyclic chromane derivatives exemplified by B16 might act as p300 HAT inhibitors.

Chemistry
Generally, unless otherwise specified, the starting materials were purchased commercially and used directly without further purification. All reactions were monitored by thin layer chromatography (TLC) on silica gel plates (HSGF254), and the components were visualized using ultraviolet light or phosphomolybdic acid.

Pharmacokinetics Procedures
Pharmacokinetic experiments of test compounds were performed in male Balb/C mice similarly to our previous work [38]. The mice were randomly assigned to two groups and were administrated the test compound orally and intravenously, respectively. The test compounds were prepared into 0.5 mg/mL oral solution with 10%PG, 10% ethanol, 10% solutol and 70% normal saline, or 0.1 mg/mL injection solution with 2% PG, 2% ethanol, 2% solutol and 94% normal saline. After intravenous or oral administration, blood was collected from the orbital venous plexus into heparinized EP tube (0.6 mL) at 5, 15, 30 min, 1, 2, 6, 10, and 24 h, temporarily placed on crushed ice. Certain processing was performed on blood samples and testing samples were sent to LC-MS/MS for analysis.

In Vivo Tumor Xenograft Model
A well-established tumorigenesis assay was used to evaluate the antitumor effect of B16-P2 in male NOD-SCID mice model. All mice were raised in standard specificpathogen-free (SPF) environment. Mice were randomly allocated to three groups (6 mice in each group) by an independent person in the laboratory. No statistical method was used to predetermine sample size. 5 × 10 6 22Rv1 cells (purchased from Nanjing KeyGEN Biotech Co. Ltd.) were injected subcutaneously into the NOD-SCID male mice at 5-to-6week-old (purchased from Shanghai lingchang animal Co. Ltd.). All compounds were prepared into solution using 5% ethanol, 30% PG, 25% PEG400, 10% solutol, and 30% pure water successively and were administrated by oral gavage. Mice were examined thrice a week for the development of tumors by Vernier caliper and tumor volumes were calculated using the formula V = 0.5 × length × width 2 . The investigators were not blinded to allocation during experiments and outcome assessment. The antitumor effects of the compounds were assessed by tumor growth inhibition (TGI) or relative tumor proliferation rate (T/C): TGI (%) = [1− (V t1 − V t0 )/(V c1 − V c0 )]×100%, where V c1 and V t1 are the mean volumes of control and treated groups at time of tumor extraction, while V c0 and V t0 are the same groups at the start of dosages; T/C (%) = T RTV /C RTV × 100%, where T RTV is the relative tumor volume (RTV) of treated groups, while C RTV is the RTV of control groups. (RTV = V t /V 0 , V t is the mean volumes of treated groups at time of tumor extraction, V 0 is the mean volumes of the same groups at the start of dosages).

Molecular Dynamic Simulation and Docking
Molecular dynamic simulation was performed with Desmond in Schrodinger Maestro 2019 using OPLS_2005 as force field (PDB code: 5kj2). The systems were solvated in TIP3P water molecules in a truncated octahedron periodic box, then neutralized by adding Na+ cations. After energy minimization, a sum of production simulation was performed for 100 ns using the NPT ensemble under a constant temperature of 300 K and pressure of 1 atm. Other parameters were maintained at the default configuration. Finally, the binding free energies for the complexes were calculated by Prime/MM-GBSA module. Molecular docking was performed with Glide module in Schrodinger Maestro 2019, with OPLS_2005 as force field and Extra Precision (XP) as algorithm.

Conclusions
In summary, twenty-six new compounds based on the 2-(2 ,4 -dioxo-2,3-dihydrospiro [indene-1,5 -oxazolidine]-3 -yl)acetamide scaffold were designed based on a bioisosterism and conformational restriction strategy. Their antiproliferative activities against enzalutamide resistant prostate cancer 22Rv1 cell were evaluated. A comprehensive SAR study was concluded, leading to the strongest inhibitor B16. Molecular docking predicted the possible binding mode with p300 HAT domain. The additional hydrogen bond between chromane oxygen of B16 and HAT protein was of critical importance for the observed stronger activity. Furthermore, compound B16 exhibited suitable PK properties. The in vivo 22Rv1 xenograft model revealed that compound B16-P2 inhibited tumor growth stronger than A-485 at the same dosage. In general, our results suggest that these spirocyclic chromane derivatives were a class of promissing therapeutic agents of prostate cancer for further optimization.