Upregulation of p53 through Induction of MDM2 Degradation: Amino Acid Prodrugs of Anthraquinone Analogs

Abstract

Previously, we reported a class of MDM2-MDM4 dimerization inhibitors that upregulate p53 and showed potent anticancer activity in animal models. However, water solubility hinders their further development. Herein we describe our effort to develop a prodrug approach that overcomes solubility problem. The prodrug of BW-AQ-238, a potent anthraquinone analog, was made by esterification of the hydroxyl group with various natural amino acids. Cytotoxicity of these compounds toward Hela and EU-1 cells, their aqueous solubility, and the release kinetics of these prodrugs in buffer and in the presence of hydrolytic enzymes were studied. The results demonstrate that the amino acid prodrug approach significantly improved the water solubility while maintaining the potency of the parent drug.

Since its discovery in 1979,^1–3 the tumor suppressor p53 has been firmly established as a critical transcription factor responsible for the regulation of cell cycle, apoptosis, DNA repair, and senescence.^{4, 5} Dysregulation of p53 functions takes on two forms: deletion or mutation of p53, and overexpression of its inhibitory proteins.^{6, 7} Approximately 50% of human cancer has p53 mutation, resulting in inactivation or loss of p53 protein functions.⁸ Even in cancer with wild-type p53, its function is ultimately inhibited, leading to accelerated cancer development and growth.^{6, 9} The latter scenario is the result of the overexpression of its negative regulator, the murine double minute 2 (MDM2) protein, which masks the p53 transactivation domain, impairing nuclear import of the p53 protein, and catalyzes the ubiquitination, leading to the proteasomal degradation of the p53 protein.^{10, 11} Activating p53 in cancer by either down-regulating MDM2 or intercepting MDM2/p53 interaction has proven to be a promising approach for cancer therapy. Therefore, many efforts have been devoted to targeting the MDM2-p53 interaction by small molecules^12–15 and peptides¹⁶ for the reactivation of p53 for the treatment of cancer that retains wild-type p53.

Along this line, our lab has synthesized a series of anthraquinone-based compounds capable of inducing MDM2 degradation, p53 upregulation, and thus apoptosis.^17–19 The efficacy of such compounds has been proven in treating acute lymphocytic leukemia (ALL) in a mouse xenograft model using our lead compound, BW-AQ-101 (Figure1).¹⁷ Complete remission was achieved after treatment with 20 mg/kg/day three doses per week for two weeks followed by 150 days of observation; whereas all control mice died within 45 days as expected. Preliminary results also showed minimal or no toxicity to normal cells at therapeutic concentration. Studies in the EU-1 ALL cells revealed that BW-AQ-101 inhibited MDM2 but not MDM4, induced activation of p53, and its target proteins p21 and PUMA in a dose- and time-dependent manner.

Figure 1.

Lead compound derived from Rhein.

The results thus far clearly demonstrated the emerging potential of this class of anthraquinones. However, the poor water solubility of the active anthraquinone analogs posed a serious developability problem. Thus, we were interested in examining the possibility of using a prodrug strategy to solve this solubility problem. Amino acids have been widely used as a prodrug strategy to improve the delivery of those with poor solubility and permeability.^20–22 Some amino acid prodrugs were also shown to enable transporter-mediated drug transport in intestinal epithelial cells.²³ Herein, we report the development of amino acid-based prodrugs for this class of anthraquinones. These prodrugs were found to have improved water solubility and release the parent drug under near physiologic conditions upon esterase-mediated hydrolysis (porcine liver esterase and porcine pancreas kallikrein). The cytotoxicity of these prodrugs was evaluated by the water-soluble tetrazolium salt (WST) assay in Hela and EU-1 cell lines. The valine prodrug was found to have comparable potency along with significantly increased aqueous solubility.

In order to prepare ester prodrugs of the parent drug, a hydroxyl handle is needed. BW-AQ-101 has two phenolic hydroxyl groups, which could conceivably be used for prodrug derivatization. However, phenol esters are generally not stable. Therefore we designed and synthesized a new analog by installing a 2-hydroxyl ethoxy group on each of the phenolic hydroxyl positions to afford BW-AQ-230. This design was based on our earlier findings that the phenolic hydroxyl groups can be modified with alkyls, alkyxoyalkyl groups without abolishing the activity of BW-AQ-101.¹⁹ The two hydroxyl groups are intended as sites for prodrug derivatization.

The synthesis of BW-AQ-238 started with the methyl ester of rhein (1),¹⁸ followed by the installation of the tert-butyldimethylsilyl (TBS) protected ethyl alcohol using potassium carbonate as a base and potassium iodide as a catalyst to obtain compound 2. Hydrolysis of the ester in compound 2 using lithium hydroxide in water afforded the carboxylic acid derivative (3) in about 65% yield. The resulting carboxylic acid group was converted to an azido group using a literature procedure to afford compound 4. Then the azido moiety was subjected to Curtius rearrangement to yield compound 5, which was converted to the corresponding amide (6) through reaction with chloroacetyl chloride. Deprotection of 6 with tetrabutylammonium fluoride (TBAF) yielded compound 7 (BW-AQ-238) as a bright yellowish solid (Scheme 1, see Supporting Information for detail).

Scheme 1.

Synthesis of BW-AQ-238. a) (TBS)OC₂H₄Cl, K₂CO₃, KI, DMF, 120 °C, 4 h; b) LiOH in H₂O, THF, r.t., 2 h; c) DPPA, TEA, DMF, r.t., 2 h; d) (i) dioxane, reflux, 2 h (ii) H₂O, 60 °C, 1 h; e) chloroacetyl chloride, dioxane, r.t., 5 mins; g) TBAF, THF, r.t., 4 h.

HeLa cell line and an in-house established wild-type p53 leukemia cell line (EU-1)¹⁷ were used for cytotoxicity assessment using the WST assay. The IC₅₀ values of BW-AQ-238 were determined to be 3.76 µM and 0.89 µM in HeLa and EU-1 leukemia cells, respectively (Table 3, see supporting information for detail). In comparison, IC₅₀ values of BW-AQ-101 were determined to be 2.21 µM and 0.83 µM in HeLa and EU-1 leukemia cells, respectively. Western blot studies also showed that BW-AQ-238 induced downregulation of MDM2 in a dose- and time-dependent manner and therefore upregulated p53 level in EU-1 cells (Figure 2). Specifically, at the 15-h time point, BW-AQ-238 at a dosage of 0.8 μM significantly down-regulated cytoplasmic MDM2 and increased the p53 level. Down-regulation of MDM2 could be detected after 4-h treatment with 1 μM BW-AQ-238 and proceeded to 16 hours; accordingly, the p53 level was elevated after 4 hours of treatment and maintained at a high level after 8 hours. Such results are consistent with that of the lead compound BW-AQ-101¹⁷ and the preliminary SAR result we reported. ¹⁹

Table 3.

Cytotoxicity of amino acid prodrugs.

Compound No.	Amino Acid	IC50 (μM) [a]
		HeLa cell	EU-1 ALL
BW-AQ-238	-	3.76 (3.39–4.49)	0.89 (0.86–0.92)
9a	Gly	3.66 (3.44–3.90)	1.10 (1.04–1.16)
9b	Ala	8.21 (7.44–9.04)	1.66 (1.51–1.83)
9c	Val	4.15 (3.27–5.27)	1.08 (0.96–1.20)
9d	Leu	2.80 (2.51–3.13)	1.17 (1.03–1.31)
9e	Ile	5.75 (5.48–6.02)	1.14 (1.03–1.27)
9f	Phe	5.64 (5.52–5.76)	1.23 (1.16–1.30)
9g	Tyr	5.40 (4.74–6.16)	1.27 (1.00–1.60)
9h	Lys	10.7 (9.60–11.8)	1.28 (1.48–1.90)
9i	Arg	13.0 (11.4–14.6)	2.07 (1.91–2.25)
9j	Gln	12.3 (9.70–15.5)	1.45 (1.29–1.63)
9k	Glu	9.11 (7.89–10.5)	3.21 (2.94–4.10)
9l	Met	7.07 (6.52–7.66)	1.24 (1.16–1.32)

^[a]Experiment was done in triplicate and 95% confidence intervals of IC₅₀ are shown in parentheses.

Figure 2.

Western blot showed the downregulation of MDM2 and upregulation p53 by BW-AQ-238 in dosage- (left panel) and time-dependent (right panel) manner in EU-1 leukemia cells. GAPDH was blotted as the loading control.

With BW-AQ-238 in hand, the preparation of water-soluble amino acid was achieved by coupling with twelve amino acids choosing from the simplest glycine to hydrophobic phenylalanine to arginine with an ionizable side chain (Scheme 2, Table 1). First, coupling of compound 7 with the corresponding N-protected amino acid using 1-ethyl-3-(3-dimethylaminopropyl)carbo-diimide (EDC) as a coupling reagent and 4-dimethylamino pyridine (DMAP) as a catalyst in acetonitrile afforded compounds 8a-l. Subsequent deprotection with 2 M hydrogen chloride in diethyl ether using dichloromethane (DCM) as a co-solvent led to the formation of corresponding hydrogen chloride salts of amino acid prodrugs, 9a-l, in quantitative yield.

Scheme 2.

Synthesis of 9 a-l. a) N-Boc protected amino acids, EDC, DMAP, ACN, 0–21°C, 4 – 6 h; b) 2M HCl in diethyl ether, DCM, 0°C-r.t.

Table 1.

Prodrug intermediates and the amino acid prodrugs^[a].

R1		R2
8a	Boc-Gly-	9a	Gly-
8b	Boc-Ala-	9b	Ala-
8c	Boc-Val-	9c	Val-
8d	Boc-Leu-	9d	Leu-
8e	Boc-Ile-	9e	Ile-
8f	Boc-Phe-	9f	Phe-
8g	Boc-Tyr(OtBu)-	9g	Tyr-
8h	Boc-Lys(Boc)-	9h	Lys-
8i	Boc-Arg(Boc)-	9i	Arg-
8j	Boc-Gln(Trt)-	9j	Gln-
8k	Boc-Glu(tBu)-	9k	Glu-
8l	Boc-Met-	9l	Met-

^[a]All amino acids are L-amino acids; ester bond was formed with carboxylic acid group of the amino acid; detailed structure is listed in supporting information (Table S1).

One requirement for the success of a prodrug strategy is the ability to release the parent drug under physiologic conditions. Therefore, we examined the chemical and enzyme-mediated hydrolysis of 9a-l in PBS at pH 7.4 and 37 °C using HPLC. Specifically, porcine liver esterase (PLE) and porcine pancreas kallikrein (PPK) were chosen as model enzymes for the parent drug release studies. PLE has been extensively used by others²⁴ and us^25–27 in the study of esterase-sensitive prodrugs. PPK was used in this study to resemble human tissue kallikrein (KLKs). KLKs is a class of serine protease with diverse roles in a variety of human tissues and is found to be upregulated in several cancers including prostate, ovarian, and breast. Several KLKs members have been identified as cancer markers including prostate and breast tissues.²⁸ Among them, human glandular kallikrein 2 has been used to develop a peptide-based prodrug targeting prostate cancer.²⁹ Since PPK has been found to be able to hydrolyze the amino acid ester,³⁰ it was included in this study to explore other specific metabolic release of the prodrug besides esterase. The chemical hydrolysis stability and enzyme specific release of the prodrugs were examined in PBS solution by monitoring the formation of BW-AQ-238 and the disappearance of the respective prodrug. The metabolic profiles are shown in Figure S1–S3 (supporting information), and the -life values are shown in Table 2. In PBS solution the ester bond of the prodrug can be chemically hydrolyzed. The release rate relatively slow except for 9a with the least steric hindrance with a glycine promoiety. This is consistent with reported ester prodrugs.³¹ The hydrolysis rate was boosted by either PLE or kallikrein, with half-lives ranging from 0.2 to 118 min. These results clearly showed that the aromatic amino acid promoiety such as Phe and Tyr are moret susceptible to PLE with half-lives within 2 minutes. PLE also showed its preference to other hydrophobic amino acids, with the order of decreasing stability being Val ≈ Ile > Gly > Ala > Met > Leu. Such results are consistent with literature reports in the hydrophobic substrate specificity of PLE.³² In the presence of kallikrein, due to its trypsin-like nature, hydrolysis showed a strong preference for basic amino acids with the Arg prodrug being hydrolyzed faster than the Lys counterpart. In addition, cleavage of Phe and Tyr prodrug was also found with the treatment, suggesting a trypsin-like and chymotrypsin-like dual substrate specificity.³³ This could be further verified by its preference to the methionine prodrug 9l.

Table 2.

Prodrug degradation half-life.

No.	Half-life (mins) [a]
	PBS (pH 7.4)	PLE	PPK
9a	31.5 (26.1–39.6)	10.1 (9.2–11.1)	35.9 (35.5–36.3)
9b	100.9 (93.8–109.1)	6.2 (5.9–6.5)	24.7 (24.1–25.2)
9c	147.7 (87.4–123.5)	25.3 (23.3–27.5)	126.3 (120.9–132.3)
9d	127.7 (73.4–111.5)	2.5 (2.4–2.7)	39.6 (34.6–46.2)
9e	154.5 (123.4–206.5)	24.3 (23.1–25.7)	48.4 (44.7–52.8)
9f	73.7 (64.8–85.5)	1.0 (0.9–1.2)	6.5 (6.1–6.8)
9g	70.6 (60.3–85.1)	1.7 (1.6–1.7)	6.5 (6.3–6.8)
9h	140.8 (132.8–149.8)	44.2 (42.9–45.6)	2.4 (2.3–2.5)
9i	187.1 (176.8–198.6)	63.0 (61.1–65.0)	0.2 (0.1–0.3)
9j	137.1 (127.7–148.2)	73.0 (71.5–74.5)	66.9 (63.8–70.4)
9k	268.6 (264.4–272.9)	117.9 (116.2–119.6)	110.0 (104.9–115.6)
9l	44.8 (41.2–48.9)	3.7 (3.3–4.2)	12.0 (11.3–12.8)

^[a]Experiment was done in triplicate and 95% confidence intervals are shown in parentheses.

The key reason for the preparation of these prodrugs was to improve solubility. For this, we assessed the water solubility of these prodrugs. It is important to note that all the amino acid prodrugs could be easily dissolved in pH 7.4 PBS at a concentration greater than 100 mg/ml, indicating a significant improvement in solubility compared with the lead compound BW-AQ-101 and the parent drug BW-AQ-238. In addition, according to the observations from several reported studies that the prodrug with phenylalanine promoiety resulted in the lowest solubility among other amino acid-based prodrugs,²³ we chose the phenylalanine prodrug 9f as a benchmark to test aqueous solubility and quantitatively compare with its parent drug. Determined by an HPLC method³⁴ (see supplementary materials for detail), the water solubility of phenylalanine prodrug was 703 mg/ml, which was over 11000-fold greater than the solubility of parent drug BW-AQ-238 (60.0 µg/ml). Comparing with the solubility result BW-AQ-101 (8.5 μg/ml), the improvement was even more striking (Table S2, supporting information). Therefore, it is reasonable to assume that the solubility of the prodrugs was greatly improved.

With success in demonstrating improved solubility and the ability to release the parent drug, we were interested in examining whether the prodrugs would be able to deliver a sufficient quantity of the parent drug to cells for the desired cytotoxicity. Therefore, the cytotoxicity of these prodrugs was tested in Hela and EU-1 leukemia cell lines using the WST-8 assay. The IC₅₀ values of these prodrugs ranged from 2.8 to 13 μM in Hela cells and 1.08 to 3.5 μM in EU-1 cells (Table 3). The IC₅₀ of 9d in Hela cells is marginally lower than that of the parent. However, the small difference makes it hard to interpret whether there is any mechanistic significance.

To further examine the prodrug activation in cell compounds 9a and 9k were selected to test the stability in the culture media. Their half-lives turned out to be 19.3 and 156.1 respectively (Figure S4), which are shorter than their chemical hydrolysis in PBS (31.5 min for 9a and 268.6 min for 9k). Therefore, can assume that prodrug activation is due to the combined effect of enzymatic and chemical hydrolysis, with enzymatic activation likely playing a dominant role.

In EU-1 leukemia cells, the IC₅₀ values are around 1 μM for most prodrugs, which are comparable with that of the parent drug. In order to study prodrug activation by cellular components, we used none-denatured cell lysate of EU-1 leukemia cells and the valine prodrug 9c as an example. It was found that the ester bond was readily cleaved within 15 min to give the parent drug BW-AQ-238 (Figure S5). This activation was much faster than chemical hydrolysis (t_1/2 of ~148 min) and PLE-mediated degradation (t_1/2 ~25.3 min). Such results suggest that the activation of prodrug 9c can be at least partially be attributed to the cytosolic enzymatic degradation.

In conclusion, to address the poor solubility of anthraquinone-based anti-leukemia agents, twelve amino acid prodrugs of BW-AQ-238 were synthesized with remarkable improvement in aqueous solubility. The chemical and enzymatic hydrolytic release of the parent drug from 9a-l were determined by HPLC. Efficient release of the parent drug was achieved with both PLE and PPK. For PLE, the aromatic amino acid promoieties such as Phe and Tyr showed the fastest release rate followed by the hydrophobic amino acid such as Leu and Met. On the other hand, PPK favored the basic amino acids such as Arg and Lys. The kallikrein metabolic profile also revealed that this peptidase could also hydrolyze an ester bond. Both valine prodrug 9c and leucine prodrug 9d demonstrated comparable cytotoxicity with the parent drug in HeLa and EU-1 cell lines. Given that valine has been successfully utilized as the promoiety of antiviral compound Zanamivir-L-Val³⁵ and a marketed drug Valacyclovir,³⁶ 9c and 9d should be considered for further in-vivo anti-cancer therapeutic studies.