Exploration of tricyclic heterocycles as core structures for RIOK2 inhibitors

Abstract

Right open reading frame kinase 2 (RIOK2) is an atypical kinase and has been proved to be involved in multiple human cancers including non-small cell lung cancer (NSCLC), acute myeloid leukemia (AML), glioblastoma and anemia. Although tremendous efforts have been devoted to the studies of RIOK2, its biological functions remain poorly understood. It is highly important to develop potent and selective RIOK2 inhibitors as potential research tools to elucidate its functions and as drug candidates for further therapies. We have previously identified a highly potent and selective RIOK2 inhibitor (CQ211). To confirm the importance of the “V-shaped” structure of CQ211 for binding with RIOK2, a variety of tricyclic compounds with different core structures instead of the [1,2,3]triazolo[4,5-c]quinolin-4-one core of CQ211 were designed, synthesized, and the binding affinities of these tricyclic heterocycles with RIOK2 were also evaluated.

By replacing the [1,2,3]triazolo[4,5-c]quinolin-4-one core of previously identified RIOK2 selective inhibitor CQ211, a variety of “V-shaped” tricyclic compounds were designed, synthesized and evaluated.

Introduction

Right open reading frame kinase 2 (RIOK2) is a member of the evolutionarily conserved RIO kinase family.^1,2 In the family, four members have been identified to date, including RIOK1-3 and RIOKB. RIOK1 and RIOK2 are the most common ones and they widely exist in many organisms including all eukaryotes and humans, whereas RIOK3 and RIOKB are found only in multicellular eukaryotes and certain archaea and eubacteria respectively.³ RIOK2 and its associated RIOK family member contain the kinase motif, but lack suitable motifs for substrate binding, which is different from typical protein kinases. This difference made RIOK2 work as an ATPase to hydrolyze ATP rather than a kinase.^4,5

The biological functions of RIOK2 have been studied over the past several years. However, the mechanisms by which RIOK2 functions still remain poorly understood. It has been proved that the absence of RIOK2 prevents maturation of 40S ribosomes, and in yeast RIOK2 is indispensable for the cytoplasmic maturation process which initiates from 20S pre-rRNA.⁶ Further studies also suggest that RIOK2 plays important roles in the mitotic progression, migration and human blood cell differentiation process through the PLK1, mTOR/AKT and GATA1 signal pathway.^7–11

Clinical reports also demonstrated that high RIOK2 expression was involved in a lot of human diseases including non-small cell lung cancer (NSCLC), acute myeloid leukemia (AML), glioblastoma, colorectal carcinoma, melanoma and anemia.¹² RIOK2 overexpression in cancers is associated with increased mortality, poor outcomes and metastasis. Knockdown of RIOK2 using shRNA could decrease both viability and migration in glioblastoma and non-small cell lung cancer (NSCLC) cells.¹³ Consequently, the inhibition of RIOK2 might offer an efficient method to suppress tumor growth and migration.

Small molecules with high binding affinity to RIOK2 may be utilized as useful research tools to elucidate the functions as well as valuable candidates for cancer therapies.¹⁴ However, compared with the intensive investigations on the biological function of RIOK2 in living organisms, the identification of potent and selective RIOK2 inhibitors has been less reported. In 2011, Zarrinkar and co-workers screened a panel of reported kinase inhibitors for the discovery of RIOK2 inhibitors.¹⁵ Since these inhibitors are originally developed for other kinases, it is understandable that compounds 1–3 are not selective for RIOK2 and displayed only weak binding affinity with RIOK2 with K_d values less than 0.2 μM (Fig. 1, compounds 1–3). Lilly scientists discovered a class of naphthyl–pyridine-based compounds as selective RIOK2 inhibitors and compound 4 showed high binding affinity to RIOK2 with a K_d of 160 nM.¹⁶ The crystal structure of compound 4 bound to hRIOK2 was also resolved, which provided structural basis for the development of selective small molecule inhibitors.¹⁷ However, compound 4 showed poor cellular activity. Modification of compound 4 by Axtman and co-workers led to the discovery of compound 5, which maintained the high binding affinity to RIOK2 and displayed weak cellular potency (IC₅₀ = 6600 nM) in the cellular-based assay known as NanoBRET.¹⁸ Compound 6 was discovered as a RIOK2 specific inhibitor by Mohamed et al. in 2018.¹⁹ It showed strong binding affinity with RIOK2 (K_d = 200 nM) and can effectively inhibit EGR-positive cancer in vitro and in vivo.

Fig. 1. Reported RIOK2 inhibitors.

Very recently, we have identified a class of [1,2,3]triazolo[4,5-c]quinolin-4-one compounds as novel RIOK2 inhibitors.^20,21 The representative compound 7 (CQ211, Fig. 1) displayed high binding affinity (K_d = 6.1 nM) and showed excellent selectivity to RIOK2 in both enzymic and cellular studies. It also demonstrated potent antiproliferative activity against several cancer cell lines and promising efficacy in vivo. However, the compound showed poor solubility in water (<10 mg L⁻¹ in H₂O) and organic solvents, and the physicochemical properties may need further improvement.

The structure of CQ211 is characterized with the tricyclic [1,2,3]triazolo[4,5-c]quinolin-4-one scaffold and two branched motifs: the 4-methoxy-pyridine and the 1-(4-(piperazin-1-yl)-3-(trifluoromethyl) phenyl motifs. The crystal structure of the RIOK2–CQ211 complex was resolved, which revealed that CQ211 is embedded in the ATP-binding pocket of RIOK2 (Fig. 2). The fused tricyclic motif played an important role in fixing the configuration of the molecule by positioning under the P-loop. The 4-methoxy-pyridine moiety was buried into a narrow cavity and the 1-(4-(piperazin-1-yl))-3-(trifluoromethyl)phenyl moiety was located in the hydrophobic binding pocket formed by M101, I111, L190, N192, and Y194. The unique V-shaped structure of CQ211 is highly compatible with the ATP-binding pocket of RIOK2, which ensured its high potency and excellent selectivity to RIOK2. In the previous study, we explored the structure-and-activity relationship of the two branched motifs of CQ211 but not the tricyclic scaffold. In this work, to confirm the importance of the V-shaped fused heterocyclic structure and to explore the possibility of replacing the [1,2,3]triazolo[4,5-c]quinolin-4-one core structure in CQ211 with other tricyclic scaffolds, compounds 8–15 (Fig. 3) were designed, synthesized and their binding affinities with RIOK2 were evaluated. Here we'd like to report the details of this research.

Fig. 2. Crystal structure of RIOK2 in complex with CQ211 (PDBcode: 7VBT). (a) Dimeric structure of the RIOK2 kinase domain in complex with CQ211. (b) “V-shape” structure of CQ211. (c) Close-up view of CQ211 in the binding pocket of RIOK2.

Fig. 3. Designed compounds 8–15 with novel scaffolds as RIOK2 inhibitors.

Results and discussion

Chemistry

The syntheses of target compounds 8–15 are represented in Schemes 1–6.

Scheme 1. Synthesis of compound 8: (a) CuI, DMSO, 90 °C; (b) 2-methoxy-5-pyridine boronic acid (19), Pd(PPh₃)₄ Cs₂CO₃, DMF/H₂O, 80 °C, 50% yield for two steps.

Scheme 2. Synthesis of compound 9: a) (i) DMF-DMA, PhMe, reflux, 4 h; (ii) (4-chloro-3-(trifluoromethyl)phenyl) hydrazine hydrochloride (21), EtOH, reflux; (iii) Fe, NH₄Cl, EtOH/H₂O, 80 °C; (iv) HCl, MeOH, 70 °C; b) (i) 2-methoxy-5-pyridine boronic acid (19), Pd(PPh₃)₄ Cs₂CO₃, DMF/H₂O, 80 °C; (ii) piperazine, Pd₂(dba)₃, RuPhos, tBuOK, dioxane, 120 °C.

Scheme 3. yntheses of compounds 10 and 11: a). (i) POCl₃, 110 °C; (ii) 1-(4-(4-amino-2-(trifluoromethyl)phenyl)piperazin-1-yl)ethan-1-one (24), AcOH, rt; (iii) Fe, NH₄Cl, EtOH/H₂O, 80 °C; b) (i) CDI, THF, 60 °C; (ii) MeI, NaH, DMF, 25 °C; (c) (i) m-CPBA, DCM, 0 °C; (ii) Ac₂O, 140 °C; (iii) 30% hydrochloric acid, 120 °C; (d) 2-methoxy-5-pyridine boronic acid (19), Pd(PPh₃)₄, Cs₂CO₃, DMF/H₂O, 80 °C; (e) triethyl orthoformate, 150 °C.

Scheme 4. Synthesis of compounds 12: a) (i) 1H-imidazole-2-carboxylic acid (30), HOBt, EDCI; (ii) NaH, DMA, 80 °C; b) 2-methoxy-5-pyridine boronic acid (19), Pd(PPh₃)₄, Cs₂CO₃, DMF/H₂O, 80 °C; c) NBS, DMF, 80 °C; d) (i) PdCl₂(PPh₃)₂, Cs₂CO₃, DMF/H₂O, 80 °C; (ii) TFA, DCM, 25 °C.

Scheme 5. Syntheses of compounds 13 and 14: a) n-BuLi, B(OEt)₃; b) PdCl₂(PPh₃)₂, Cs₂CO₃, DMF/H₂O, 80 °C; c) 2-fluoro-4-bromonitrobenzene (39), Cs₂CO₃, DMSO; d) (i) Fe, NH₄Cl, EtOH/H₂O, 80 °C; (ii) CDI, o-dichlorobenzene, reflux; e) (i) 2-methoxy-5-pyridine boronic acid (19), Pd(PPh₃)₄, Cs₂CO₃, DMF/H₂O, 80 °C; (ii) TFA, DCM, 25 °C.

Scheme 6. Synthesis of compound 15: a) AcOH; b) (i) TFA/TfOH = 1 : 1; (ii) 2-methoxy-5-pyridine boronic acid (19), Pd(PPh₃)₄, Cs₂CO₃, DMF/H₂O, 80 °C.

Compound 8 was designed as the ring-breaking structure of CQ211. As shown in Scheme 1, it was synthesized through a CuAAC reaction followed by a Pd-catalyzed Suzuki coupling.

Compounds 9–14 maintained the fused 6-6-5 tricyclic core structures but with differences to CQ211 at the five-membered ring motif, while compound 15 bears a 6-7-5 tricyclic 4H-benzo[b][1,2,4]triazolo[4,3-d][1,4]diazepin-5(6H)-one core structure. In these structures, the [1,2,3]triazolo[4,5-c]quinoxali-4-one core structure of CQ211 was replaced with varied tricyclic motifs, such as 1,5-dihydro-pyrazolo[4,3-c]quinoxali-4-one (9), 1H-imidazo[4,5-c]quinoline-2,4-dione (10), 1,5-dihydro-4H-imidazo[4,5-c]quinoxali-4-one (11), imidazo[1,2-a]quinoxaline-4(5H)-one (12), imidazo[1,5-a]quinoxaline-4(5H)-one (13), pyrrolo[1,2-a]quinoxalin-4-one core (14), and 4H-benzo[b][1,2,4]triazolo[4,3-d][1,4]diazepin-5(6H)-one (15). The syntheses of these compounds are shown in Schemes 2–6.

Solubility

In our previous study, it has been found that compound 7 (CQ211) displayed poor solubility in H₂O (2.3 mg L⁻¹) and other organic solvents, which may cause problems in biological evaluations in cells and animal models. Thus, with the new structures in hand, we then tested their solubility in different solvents. As shown in Table 1, the solubilities of compounds 8–15 in organic solvents such as DMSO, MeOH and CH₂Cl₂ have been significantly increased in comparison with compound 7. Their solubilities in H₂O are also relatively better than compound 7, except for compound 14 (7.9 mg L⁻¹ H₂O).

The solubility of compounds 7–16 in different solvents.

Comp.	Solubility in the solvent (mg L−1)
	DMSO	MeOH	CH2Cl2	H2O
7	4900	12	16	2.3
8	281 000	1960	7800	400
9	97 000	1260	740	108
10	64 000	810	600	78
11	53 000	700	570	59
12	46 000	740	520	42
13	52 000	610	600	33
14	23 000	340	370	7.9
15	171 000	2710	6480	1210

RIOK2 kinase binding affinity

The binding affinities of compounds 8–15 with RIOK2 were evaluated using KdELECT at DiscoverX and the results are shown in Table 2. Compound 7 (CQ211) was used as the positive control, and the K_d between RIOK2 and compound 7 is 7.3 ± 0.7 nM, similar to our previous report.²¹ Compound 8 was designed as a negative control. Without the “V-shaped” tricyclic structure, the linear compound 8 totally lost its binding affinity to RIOK2 (K_d > 1000 nM), which confirmed the importance of the fused tricyclic core structure for their RIOK2 activity. Compounds 9–14 kept the “V-shaped” tricyclic structure, but with differences to compound 7 at the five-membered ring motif. All these compounds displayed strong to moderate binding affinities with RIOK2. Among them, the most potent one is compound 9, with a K_d = 19 ± 1 nM, while compound 14 displayed relatively weak binding affinity with a K_d = 165 ± 5 nM. The seven-membered ring structure in compound 15 made it more flexible and bulkier in space; it totally lost its binding affinity with RIOK2 (K_d > 1000 nM). This result indicates that the size and planarity may be very important to design the core structure for RIOK2 inhibitors.

The binding affinities of compounds 8–15 with RIOK2.

Compound	RIOK2 binding affinity Kd,a [nM, AV ± SD]
7	7.3 ± 0.7
8	>1000
9	19 ± 1
10	48 ± 8
11	62 ± 2
12	53 ± 4
13	56 ± 13
14	165 ± 5
15	>1000

^aBinding constant values (K_d) were determined from DiscoverX KINOMEscan. The data are means from two independent experiments.

We have also evaluated the kinase inhibitory rates of these compounds in several kinases on hand such as ABL, FLT3, RET, FGFR1 and FGFR2 at a dose of 1 μM. The results showed that all compounds displayed no effects on ABL, RET and FGFR1, and weaker activities against FLT3 and FGFR2 with inhibitory rates of <30% at 1 μM.²²

In vitro antiproliferative activity

Compounds 7–15 were further evaluated in the proliferation assay in human glioblastoma cells (U87MG), human gastric cancer cells (MKN-1) and leukemia cells (MOLT4).²³ As shown in Table 3, compounds 10–14 showed inferior anti-proliferative activities in comparison with compound 7 in all the three cell lines, while compound 9 showed slightly better activities than compound 7. However, compound 8, with no binding activity with RIOK2, displayed moderate anti-proliferative activity to the three cancer cell lines, which may be due to the action on other unidentified targets. Compound 15, without binding with RIOK2, displayed much weaker antiproliferative activities than other compounds in all the three cancer cell lines. However, it is noteworthy that the no distinctive differences were observed for the cellular activities of compounds 7–14, which indicated that the actions of these compounds may be complexed in cells and may involve other targets rather than only RIOK2.

The anti-proliferative activities of compounds in cancer cells.

Compd.	IC50 (μM) AV ± SD
	U87MG	MKN-1	MOLT4
7	1.65 ± 0.53	1.69 ± 0.25	2.12 ± 0.10
8	4.22 ± 2.91	3.02 ± 1.41	5.34 ± 2.15
9	1.25 ± 0.08	1.05 ± 0.32	1.57 ± 0.47
10	4.35 ± 2.41	2.37 ± 0.94	1.75 ± 0.57
11	4.45 ± 1.25	3.34 ± 1.51	3.93 ± 0.69
12	3.88 ± 0.02	5.17 ± 4.69	2.76 ± 1.02
13	4.07 ± 0.80	4.67 ± 2.15	3.81 ± 0.51
14	3.85 ± 0.057	3.99 ± 2.34	2.64 ± 0.33
15	>10	>10	>10

Conclusions

In summary, a series of tricyclic structures were synthesized and evaluated as RIOK2 inhibitors. The structure modification and evaluation of binding affinity with RIOK2 confirmed the importance of the V-shaped fused 6-6-5 tricyclic structure for RIOK2 activity. Although the binding affinities are inferior to our previously reported compound CQ211, compounds 9–14 still showed strong binding affinity with RIOK2. This indicates that the V-shaped tricyclic structures may be used to replace the [1,2,3]triazolo[4,5-c]quinolin-4-one core structure in CQ211. The improvement of solubility in organic solvents and H₂O in these structures may bring advantages for in vitro and in vivo studies in the future.

Conflicts of interest

There are no conflicts to declare.