Design, Synthesis, and Biological Evaluation of APN and AKT Dual-Target Inhibitors

Abstract

Herein a novel series of APN and AKT dual inhibitors were derived from the clinical AKT inhibitor AZD5363. It was demonstrated that most compounds exhibited remarkable APN inhibitory activities with the most potent compound 8b (IC₅₀ = 0.05 ± 0.01 μM) being over 70-fold more potent than the approved APN inhibitor bestatin (IC₅₀ = 3.64 ± 0.56 μM). The moderate AKT inhibitory potencies of target compounds were also confirmed, with 5f and 5h possessing AKT1 IC₅₀ values of 0.12 and 0.27 μM, respectively. More importantly, the APN IC₅₀ values of 5f and 5h were 0.96 and 0.21 μM, respectively, indicating their balanced APN and AKT dual inhibition. HUVEC tube formation assays confirmed the superior APN inhibitory activities of 5f and 5h relative to bestatin at the cellular level. Western blot analysis demonstrated that 5h could effectively inhibit the phosphorylation of GSK3β, the intracellular substrate of AKT.

Tumor biology is very complex and is characterized by relapse and drug resistance.^1,2 Increasing therapeutic targets have been disclosed with the in-depth investigation of cancer etiology.^3,4 Unfortunately, the therapeutic effects of single-target drugs in the clinic are usually compromised by resistance and relapse frequently arising from the tumor heterogeneity and adaptability.^5,6 It is worth noting that multitarget drugs intervening with numerous tumor-related targets or pathways simultaneously can generate improved anticancer efficacy, therefore being the arsenal of defeating cancer resistance and relapse.⁷

AKT, namely protein kinase B (PKB), is a serine/threonine kinase involved in regulating cell growth, proliferation, apoptosis, and glycogen metabolism.⁸ There are three AKT isoforms (AKT1, AKT2, and AKT3)^9,10 which are closely related to the tumorigenesis and development of prostate, ovarian, breast, and pancreatic cancers.¹¹⁻¹³ Therefore, the development of AKT inhibitors is regarded as a promising anticancer strategy. Currently, at least four AKT inhibitors (AZD5363, MK-2206, Perifosine, and GDC-0068) have entered clinical trials as anticancer agents.¹¹

Aminopeptidase N (APN, CD13) is a cell surface marker of malignant myeloid cells¹⁴ and is misregulated on the cell surface of many kinds of cancers.¹⁵⁻²¹ APN can degrade the extracellular matrix (ECM), the important network preventing tumor invasion and metastasis.²² Increasing studies show that APN is also a molecular target promoting angiogenesis.^23,24 Therefore, APN is regarded as an important antitumor target.²⁵ Bestatin, the first approved APN inhibitor and initially designed as an immunomodulator,²⁶ has been used for treating gastric cancer, leukemia, non-small-cell lung cancer (NSCLC), and cervical cancer.²⁷⁻²⁹ However, the anticancer effect of APN inhibitors is limited when used alone, so APN inhibitors are usually combined with other anticancer therapies in clinic. For example, the combination of bestatin and busulfan is beneficial to adult patients with acute nonlymphocytic leukemia.³⁰

In recent years, many studies revealed the crosstalk between APN and AKT inhibitors in some solid tumor cells. For example, the APN inhibitor bestatin can promote autophagic death of bladder cancer cells through inhibition of the AKT signaling pathway.²⁹ The combination of bestatin and radiotherapy can effectively inhibit the AKT signaling pathway in renal cancer cells.³¹ Bestatin can also block the activation of the CD13/EMP3/PI3K/AKT/NF-κB pathway to combat cisplatin resistance in gastric cancer cells.³² Notably, a combination of bestatin and AKT inhibitors showed improved anti-melanoma activity. The regulatory effect of APN inhibitors on AKT signaling pathway and the promising anticancer effects of APN and AKT inhibitors combination support the development of APN and AKT dual-target inhibitors as antitumor agents.

AZD5363 (Figure 1), a AKT inhibitor in phase II clinical trials, is currently being investigated for the treatment of breast cancer (NCT02077569, NCT02423603), gastric cancer (NCT02451956), prostate cancer (NCT02525068), and other cancers. The binding mode of AZD5363 in AKT1 (PDB Code: 4GV1, Figure 1) revealed that the pyrrolopyrimidine moiety of AZD5363 formed two hydrogen bonds with the hinge residues Ala230 and Glu228 of AKT1. In addition, the 4-aminopiperidine-4-carboxamide moiety of AZD5363 formed two hydrogen bonds with AKT1, with the primary amino group inserting into an acid hole, which was formed by Glu278 and Glu234.

Figure 1.

Design strategy of compounds 5a–5j.

Therefore, modification of the pyrrolopyrimidine and 4-aminopiperidine-4-carboxamide moieties of AZD5363 could not be well tolerated. Note the p-chlorophenyl group of AZD5363 mainly formed hydrophobic interactions with the P-loop of AKT1, suggesting that replacement of this group with other hydrophobic groups could be tolerated. The binding mode of bestatin in hAPN (PDB Code: 4FYR, Figure 1) revealed that the carboxylate group of bestatin bidentately coordinated the catalytic Zn²⁺ of APN, with the terminal isobutyl and phenyl groups occupying the S1’ and S1 pockets, respectively. Our previous structure–activity relationship (SAR) study showed that isobutyl was a privileged group for the S1’ pocket and the S1 pocket could accommodate various larger aromatic groups.³³

According to the above structure analysis, compound BA was designed by hybridization of the key pharmacophores of AZD5363 and bestatin (Figure 1). Considering that hydroxamic acid is a more potent Zn²⁺ binding group, the carboxylic acid group of hybrid BA was replaced with hydroxamic acid, leading to target compounds 5c and 5h. Structural derivatization focusing on the isobutyl group of 5c and 5h led to target compounds 5a–5e and 5f–5j, respectively.

Based on SAR studies of 5a–5j, compounds 8a–8c, 12a–12c, and 16a–16c were further designed and synthesized to investigate and confirm the effects of the middle 6-membered ring on APN and AKT inhibitory activities (Figure 2).

Figure 2.

Design strategy of compounds 8a–8c, 12a–12c, and 16a–16c.

Compounds 5a–5j and 8a–8c were synthesized according to the procedures in Scheme 1. The starting material 1 reacted with 4-(Boc-amino)piperidine-4-carboxylic acid and piperidine-4-carboxylic acid to get the key intermediates 2 and 6, which were condensed with various amino acid methyl ester hydrochlorides to get 3a–3j and 7a–7c respectively. Compounds 3a–3j then were treated with NH₂OK in methanol to get hydroxamic acids 4a–4j, which were deprotected under acidic conditions to afford target compounds 5a–5j. Compounds 7a–7c could be directly converted to the corresponding target compounds 8a–8c.

Scheme 1. Synthesis of Compounds 5a–5j and 8a–8c.

Reagents and conditions: (a) 4-(Boc-amino)piperidine-4-carboxylic acid or piperidine-4-carboxylic acid, NaHCO₃, CH₃CN/H₂O, reflux, 64–76%; (b) l-/d-amino acid methyl ester hydrochlorides, EDCI, HOBT, Et₃N, DMF, rt, 19-94%; (c) NH₂OK, CH₃OH, rt, 29–86%; (d) HCl/EtOAc, rt, 74–98%.

Compounds 12a–12c and 16a–16c were synthesized according to the procedures in Scheme 2. The starting material 1 reacted with tert-butyl piperazine-1-carboxylate and tert-butyl piperidin-4-ylcarbamate to get 9 and 13, which were deprotected under acidic conditions to obtain intermediates 10 and 14, respectively. Compounds 10 and 14 were condensed with various d-amino acid methyl ester hydrochlorides in the presence of triphosgene to get compounds 11a–11c and 15a–15c, which were converted to target hydroxamic acids 12a–12c and 16a–16c by NH₂OK, respectively.

Scheme 2. Synthesis of Compounds 12a–12c and 16a–16c.

Reagents and conditions: (a) tert-Butyl piperazine-1-carboxylate or tert-butyl piperidin-4-ylcarbamate, NaHCO₃, Et₃N, CH₃CN/H₂O/DMF, reflux, 92–96%; (b) 12 M HCl, CH₃OH, rt, 45–51%; (c) various d-amino acid methyl ester hydrochlorides, triphosgene, NaHCO₃, DCM, 0°C; Et₃N, DMF, rt, 70–96% ; (d) NH₂OK, CH₃OH, rt, 30–83%.

The results of in vitro APN inhibitory assays of compounds 5a–5j are shown in Table 1, which clearly demonstrated that the compounds with the R-configuration possessed much better APN inhibitory activities than their S-enantiomers (5g vs 5b, 5h vs 5c, 5i vs 5d, 5j vs 5e). Among these analogues, compound 5h was the most potent APN inhibitor (IC₅₀ = 0.21 ± 0.01 μM), which was over 150-fold more potent than its S-enantiomer 5c (IC₅₀ = 32.37 ± 16.82 μM). Remarkably, compounds 5a, 5f, 5g, 5h ,and 5j exhibited superior APN inhibitory activities compared to the approved APN inhibitor bestatin.

Table 1. APN Inhibitory Activity of Compounds 5a–5j^a.

^aAssays were performed in replicate (n ≥ 2); the IC₅₀ values are shown as mean ± SD.

To investigate the effects of the middle 6-membered ring on APN inhibitory activities, compounds 8a–8c, 12a–12c, and 16a–16c were designed and synthesized. Their R substituents were designated as benzyl, isobutyl, and phenyl groups of R-configuration based on the SAR of compounds 5a–5j. It has been found (Table 2) that all these compounds except 16c possessed APN inhibitory activities comparable to those of their parent compounds 5a–5j, indicating that the middle 6-membered ring has little effect on the APN inhibitory activities. Moreover, all these compounds except 16c exhibited more potent APN inhibitory activities than bestatin. Note that the most potent compound 8b (IC₅₀ = 0.05 ± 0.01 μM) was over 70-fold more potent than bestatin (IC₅₀ = 3.64 ± 0.56 μM).

Table 2. APN Inhibitory Activity of Compounds 8a–8c, 12a–12c, and 16a–16c^a.

^aAssays were performed in replicate (n ≥ 2); the IC₅₀ values are shown as mean ± SD.

The AKT inhibitory potency of target compounds was first evaluated by determining the AKT1 inhibitory rates at 1 μM and 200 nM with AZD5363 as the positive control. It was revealed that compounds 5a–5j could effectively inhibit AKT1 dose-dependently (Table 3). Generally, compounds of R-configuration (5f–5j) were more potent than their S-enantiomers (5a–5e), which was in line with their SAR of APN inhibition (Table 1). Three compounds 5a, 5f, and 5h exhibited >50% AKT1 inhibition at 1 μM. The AKT1 IC₅₀ values of 5f and 5h were further determined (Table 3). Although less potent than the clinical AKT inhibitor AZD5363, compounds 5f (APN IC₅₀ = 0.96 ± 0.04 μM, AKT1 IC₅₀ = 0.12 ± 0.02 μM) and 5h (APN IC₅₀ = 0.21 ± 0.01 μM, AKT1 IC₅₀ = 0.27 ± 0.04 μM) possessed balanced APN and AKT dual inhibitory activities.

Table 3. AKT1 Inhibitory Activity of Compounds 5a–5j^a.

^aAssays were performed in replicate (n ≥ 2).

^bIC₅₀ values are shown as mean ± SD in parentheses.

The AKT1 inhibitory activities of compounds 8a, 8b, 12a, 12b, 16a, and 16b are presented in Table 4, which reveals their decreased AKT1 inhibitory activities compared with their parent compounds, especially parent compound 5h. Note that no significant inhibition was observed for compounds 16a and 16b at 1 μM. These results confirmed the important contribution of the middle 4-aminopiperidine group to the AKT inhibition of AZD5363, as revealed in the crystal complex structure of AZD5363 and AKT1 (Figure 1).

Table 4. AKT1 Inhibitory Activity of Compounds 8a, 8b, 12a, 12b, 16a, and 16b^a.

^aAssays were performed in replicate (n ≥ 2).

Since hydroxamate-based compounds might show off-target effects against other Zn²⁺-dependent metalloproteases, such as histone deacetylases (HDACs), compounds 5f and 5h possessing potent APN and AKT1 dual inhibitory activities were evaluated in HDAC inhibitory assays, with SAHA (a pan-HDAC inhibitor) and PCI34051 (a HDAC8 selective inhibitor) as positive controls. The results in Table 5 indicate that 10 μM 5f and 5h showed marginal inhibition against HeLa cell nuclear extract (mainly contains HDAC1 and HDAC2) and HDAC6. Although 5f and 5h showed around 50% inhibition against HDAC8 at 10 μM, their HDAC8 inhibitory potencies were far less potent than their APN and AKT1 inhibitory activities shown in Tables 1 and 3.

Table 5. HDAC Inhibitory Activity of Compounds 5f and 5h^a.

^aAssays were performed in replicate (n ≥ 2).

^bNot determined.

Considering their balanced APN and AKT dual inhibitory activities, 5f and 5h were further tested in an in vitro antiproliferative assay against the human lung adenocarcinoma cell line NCI-H1975, human breast cancer cell line MCF-7, and human gastric adenocarcinoma cell line AGS. The clinical AKT inhibitor AZD5363 was used as the positive control. The antiproliferative activity of the approved APN inhibitor bestatin was also evaluated. According to the results in Table 6, our dual inhibitors 5f and 5h showed less potent antiproliferative activities than that of AZD5363 and more potent antiproliferative activities than that of bestatin. Generally, the antiproliferative activities of 5f, 5h, and AZD5363 were in line with their AKT1 inhibitory activities. It is worth noting that the APN inhibitor bestatin showed negligible activity with IC₅₀ values over 200 μM, indicating that APN inhibition did not contribute to cytotoxicity, which was consistent with the reported results.³⁴

Table 6. In Vitro Antiproliferative Activity of Representative Compounds^a.

	IC50 (μM)
compd	NCI-H1975	MCF-7	AGS
5f	20.3 ± 3.5	24.8 ± 4.5	30.7 ± 5.1
5h	27.4 ± 5.3	32.7 ± 4.3	52.4 ± 4.8
AZD5363	2.7 ± 0.3	4.2 ± 0.8	9.6 ± 1.7
bestatin	>200	>200	>200

^aAssays were performed in replicate (n ≥ 2). IC₅₀ values are shown as mean ± SD.

It has been demonstrated that APN plays an indispensible role in the process of tumor angiogenesis and APN inhibitors possess antiangiogenic potency.³⁵ Accordingly, we used the HUVEC tubular structure formation model to evaluate the antiangiogenic potency of compounds 5f and 5h. HUVECs were treated with compounds at 10 μM, which could avoid cytotoxicity according to our in vitro antiproliferation assay (Table 6). The results in Figure 3 show that, after a period of incubation, HUVECs in the blank group migrated and connected to form a tubular structure on Matrigel. This process can be partially inhibited by bestatin and completely inhibited by 5f and 5h, which was in line with the superior APN inhibitory activities of 5f and 5h compared with bestatin (Table 1). In contrast, the AKT inhibitor AZD5363 and the HDAC8 inhibitor PCI34051 showed no significant inhibition of HUVEC tubular structure formation. These results collectively demonstrated that the antiangiogenic activities of 5f and 5h are ascribed to their potent APN inhibitory activities.

Figure 3.

Representative images of the tube network of HUVECs treated with DMSO or compounds.

GSK3β is the intracellular substrate of AKT. Phosphorylated GSK3β resulting from AKT activation is an important regulator of cell metabolism and plays key roles in cell growth, development, tumorigenesis, and glucose homeostasis.³⁶ In order to confirm the intracellular AKT inhibitory potency of compound 5h, Western blot analysis was conducted to detect the levels of p-GSK3β in AGS cells. The results shown in Figure 4 demonstrated that compound 5h can inhibit the phosphorylation of GSK3β in a dose-dependent and time-dependent manner.

Figure 4.

(A) AGS cells were treated with DMSO or compounds (2 or 10 μM) for indicated time (2 or 6 h). Levels of p-GSK3β were determined by immunoblotting. β-Actin was used as a loading control. Data are representative of three independent experiments. (B) Quantitative analysis results of the Western blot. *p < 0.05 versus the control groups by Student’s two-tailed t test.

The binding modes of 5h in APN and AKT1 were proposed by SYBYL-X-2.0. As shown in Figure 5, both bestatin and compound 5h could chelate the catalytic Zn²⁺ and form a hydrogen bond with Tyr477 with their carboxylate group and hydroxamate group, respectively. In contrast with bestatin, the isobutyl group of 5h flipped to occupy the S1 pocket while the pyrrolo[2,3-d]pyrimidine group fit into the S1’ pocket forming a hydrogen bond with Asp439 (Figure 5B). Moreover, the hydroxamate group of 5h could form another hydrogen bond with Glu389. The presence of more hydrogen bonds between 5h and APN could rationalize its superior APN inhibitory activity relative to bestatin.

Figure 5.

(A) Crystal structure of bestatin and hAPN (PDB Code: 4FYR). (B) Proposed binding mode of compound 5h in hAPN.

The proposed binding mode of compound 5h in AKT1 is shown in Figure 6B, which was quite similar to that of AZD5363 (Figure 6A). To be specific, both AZD5363 and 5h could form two key hydrogen bonds with the Ala230 and Glu228 residues in the hinge region. The chlorobenzene group of AZD5363 and the isobutyl group of 5h formed hydrophobic interactions with the P-loop. Compared with AZD5363, although 5h could form a new hydrogen bond with Asp292, the two hydrogen bonds between AZD5365 and Glu234 and Glu278 were not observed in the case of 5h. This might be the reason why 5h showed less potent AKT1 inhibitory activity than AZD5363.

Figure 6.

(A) Crystal structure of AZD5363 and AKT1 (PDB Code: 4GV1). (B) Proposed binding mode of compound 5h in AKT1.

Collectively, a series of AZD5363-based APN and AKT dual inhibitors were developed based on the cross talk between APN and AKT inhibitors. Compound 5h showed balanced APN and AKT1 dual inhibitory potencies, which were further validated by HUVEC tubular structure formation assay and Western blot analysis. Considering their moderate antiproliferative activities, structural optimization of these compounds is warranted to develop novel AKT and APN dual inhibitors with improved antiproliferative activity and uncompromised antiangiogenic activity.

Glossary

Abbreviations Used

APN

aminopeptidase N

AKT/PKB

protein kinase B

ECM

extracellular matrix

HUVECs

human umbilical vein endothelial cells

NSCLC

non-small-cell lung cancer

SAR

structure–activity relationship

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.1c00504.

• Chemistry experimental procedures, biological materials and methods, molecular docking study, ¹H and ¹³C NMR spectra, dosage response curves for APN and AKT1 IC₅₀ determination(PDF)

Author Contributions

^† Q.L. and H.D. contributed equally to this work. The manuscript was written through contributions of all authors. Y.Z. designed the research; Q.L., H.D., W.Z., G.Z., S. L., and Q.X. performed the experimental research; Y.Z., Q.L., and H.D. analyzed data.

The authors declare no competing financial interest.