Tricyclic thiazoles are a new class of angiogenesis inhibitors

Abstract

Tricyclic thiazoleamine derivatives that were identified as hits in a screen against human umbilica vein endothelial cell proliferation were subjected to a structure-activity relationship study. Two structurally superimposable scaffolds—4H-thiochromeno[4,3- d]thiazol-2-amine and 5,6-dihydro-4H-benzo[6,7]cyclohepta[1,2-d]thiazol-2-amine derivatives—yielded low-micromolar inhibitors, and two among them 37 and 43 also exhibited antiangiogenic activity in an endothelial tube formation assay Thus, 37 and 43 can serve as leads to develop a novel class of antiangiogenic agents.

Angiogenesis plays an essential role in tumor growth and metastasis. Inhibitors of angiogenesis are emerging as a new class of anticancer drugs.^1,2,3 In the clinic, it has been found that inclusion of an antiangiogenic drug like bevacizumab, sunitinib, or sorafenib in the combination chemotherapy produces significant survival benefits^4,5 and hence, antiangiogenic dru have become an integral part of front-line therapy in treating different types of cancers. Unfortunately, primary and acquired resistance to antiangiogenic therapy is becoming a real impediment and new agents with novel mechanisms of action are urgently needed to tackle this problem.⁶ Because proliferation of endothelial cells is an obligatory step for in vivo angiogenesis, direct growth inhibition of endothelial cells in culture has come to serve as a proxy for anti-angiogenesis screening. During a routine test of target compounds and intermediates synthesized in our laboratory, we discovered that four tricyclic thiazoles—3, 7, 9, and 11 (Table 1)—were moderately antiproliferative against human umbilical vein endothelial cells (HUVEC) in a [³H]-thymidine incorporation assay In fact, these tricyclic thiazoles had been synthesized in the course of our development of methionine aminopeptidase (MetAP) inhibitors as antimycobacterial agents.⁷ It has been established earlier using fumagillin that inhibition of human MetAP2 leads to the growth inhibition of HUVEC.^8,9 However, tricyclic thiazoles 3, 7, 9, and 11 did not inhibit (up to 20 μM) either isoforms of human MetAPs (hMetAP1 and hMetAP2), suggesting that HUVEC inhibition proceeded through a different mechanism. Herein, we disclose a structure-activity relationship (SAR) study of this novel class of endothelial cell inhibitors.

Table 1.

Inhibition of HUVEC proliferation by thiazoles.

Entry	structure	Substituents (if not specified, R# = H)	HUVECa (IC50, μM)
1		R2 = H	>10
2		R2 = COMe	>10
3		R2 = CO(CH2)4Me	3.0 ±1.0
4		R2 = H, R6 = OMe	>10
5		R2 = COMe, R6 = OMe	>10
6		R2 = H, R7 = OMe	>10
7		R2 = CH2(2-furanyl), R7 = OMe	4.5 ±1.2
8		R2 = COMe, R7 = OMe	>10
9		R2 = CO(CH2)4Me, R7 = OMe	3.7 ± 1.0
10		R2 = H, R8 = OMe	>10
11		R2 = COMe, R8 = OMe	4.2 ±1.1
12		R2 = H, R7, R8 = OMe	>10
13		R2 = COEt, R7, R8 = OMe	>10
14		R2 = H	>10
15		R2 = COMe	>10
16		R2 = H, R6 = OMe	>10
17		R2 = COMe, R6 = OMe	4.4 ±1.5
18		R2= H, R6= F	>10
19		R2= COEt, R6= F	>10
20		R2= H, R6, R7= OMe	>10
21		R2= COMe, R6, R7= OMe	>10
22		R2 = H, R7 = Cl	>10
23		R2= COMe, R7= Cl	>10
24		R2 = H	>10
25		R2 = COEt	>10
26		R2 = H, X = O	>10
27		R2 = COEt, X = O	>10
28		R2 = H, X = O, R8 = F	>10
29		R2 = COEt, X = O, R8 = F	>10
30		R2 = H, X = S	1.9 ±0.3
31		R2 = COEt, X = S	2.0 ±0.3
32		R1, R2 = H, X = CH2	>10
33		R1 = H, R2 = COEt, X = CH2	3.7 ±1.1
34		R1, R2 = COEt, X = CH2	1.5 ±0.3
35		R1, R2 = H, X = O, R8–R9 = OCH2O	5.1 ±1.1

^aIC₅₀ values are averages of three independent experiments and the ± standard deviation is given.

The tricyclic thiazoles were synthesized as per our earlier procedure⁷ using a variation of Hantzsch thiazole synthesis. Two typical examples are shown in Scheme 1. Briefly condensation of 6-chlorothiochroman-4-one (eq. 1) or 1-benzosuberone (eq. 2) with thiourea in the presence of iodine at 100 °C generated the tricyclic thiazoleamines 36 and 32, respectively, which upon neutralization served as starting materials for the subsequent steps. Thiazoleamine 37 (Scheme 1, eq. 1) was prepared by forming the Schiff base followed by reduction using sodium cyanoborohydride. Thiazoleamine 32 was treated with 2,6-difluorobenzoyl chloride in triethylamine containing dichloromethane to obtain the corresponding benzamide derivative 43 (Scheme 1, eq. 2). Thiazoleamine 45 (Table 2) was obtained by alkylating amine 30 with 6-azidohex-1-yl tosylate following the procedure of Salvatore et al.¹⁰ Benzamides 46 and 47 (Table 2) were synthesized by coupling thiazoleamine 30 and 32, respectively, with 4-propynyloxybenzoic acid (see Supplementary data).

Scheme 1.

A typical synthesis of tricyclic thiazole derivatives. Conditions: i. thiourea, iodine, EtOH, 100 °C, 3 h, then aqueous NaHCO₃; ii. furfural/MgSO4/MeOH and then NaBH3CN; iii. 2,6-difluorobenzoyl chloride, Et3N/DCM.

Table 2.

Two classes of thiazoles: Antiproliferative activities against four cell lines (IC₅₀, μM).

Entry	Structure	Substituents (if not specified, R# = H)	HUVECa	Jurkat-Ta	BT474a	HeLaa	Selectivity indexb
30		H	1.9 ±0.3	0.9 ±0.1	8.2 ±1.4	6.8 ±1.1	0.5/4.3/3.6
31		COEt	2.0 ±0.3	4.9 ±0.5	5.9 ±1.1	8.0 ±1.1	2.5/3.0/4.0
41		COPh(2,6-difluoro)	12.0 ±2.2	11.8 ±1.1	7.2 ±1.3	13.1 ±2.1	–
45		(CH2)6N3	5.9 ±1.1	ND	ND	ND	–
46		COPh(4-OCH2C≡CH)	2.9 ±0.8	4.6 ±0.9	0.3 ±0.2	5.0 ±1.1	1.6/0. 1/1.7
36		H	1.9 ±0.4	1.6 ±0.7	3.8 ±1.1	11.9 ±1.0	0.8/2.0/6.3
37		CH2(2-furanyl)	1.1 ±0.3	2.5 ±0.6	2.3 ±0.8	10.3 ±1.1	2.3/2.1/9.4
38		CO2Et	2.8 ±0.4	6.3 ±1.0	6.2 ±1.1	6.5 ±1.6	2.3/2.2/2.3
42		COPh(2,6-difluoro)	5.2 ±0.4	17.3 ±1.1	3.6 ±1.1	>20	3.3/0.7/>4
32		R1, R2 = H	19.1 ±1.2	>20	>20	>20	–
33		R2 = COEt	3.7 ±1.1	>20	>20	>20	>5/>5/>5
34		R1, R2 = COEt	1.5 ±0.3	2.8 ±0.4	>20	5.9 ±1.1	1.9/>13/4
39		R2 = CO2Et	9.2 ±1.2	>20	>20	>20	–
43		R2= COPh(2,6-difluoro)	1.5 ±0.4	>20	>20	15.2 ±2.2	>13/>13/10
47		R2 = COPh(4-OCH2C≡CH)	7.3 ±1.1	ND	ND	ND	–
40		H	14.8 ±1.2	>20	>20	16.2 ±2.3	–
44		COPh(2,6-difluoro)	13.5 ±2.1	>20	>20	>20	–

^aIC₅₀ values are averages of three independent experiments and the ± standard deviation is given.

^bSelectivity index: IC50 values of Jurkat-T/BT474/HeLa compared to that of HUVEC. ND: Not determined.

A collection of 35 tricyclic thiazole derivatives (Table 1) comprising of thiazoleamines and their amides were synthesized and screened for their antiproliferative activities in HUVEC culture. Among a series of 4H,5H-naphto[1,2-d]thiazoleamines containing different patterns of methoxy substitutions on the A-ring (see eq. 1 in Scheme 1 for ring designation), all the parent primary amines 1, 4, 6, 10, and 12 failed to register an IC₅₀ below 10 μM. Only a single furanyl substituted thiazoleamine 7 showed a moderate inhibition of HUVEC (4.5 μM). We had acetamide, propanamide, and hexanamide derivatives in this series where both the hexanamides 3 and 9 inhibited HUVEC proliferation moderately (3.0 and 3.7 μM, respectively), but none of the acetamides except 11 (4.2 μM) exhibited HUVEC inhibition. Compounds 14 and 15, comprising of an inversely fused tricyclic thiazole ring system, were ineffective. Next in our SAR effort, we produced and screened thiazoles embodying a contracted B-ring (16–19), a completely severed B-ring (20–23), and a totally eliminated A-ring (24 and 25). None of these compounds (16–25) except N-(6-methoxy-8H-indeno[1,2-d]thiazol-2-yl)acetamide (17, 4.4 μM) exhibited HUVEC inhibition. We then introduced an oxygen atom in the B-ring to obtain chromenothiazole derivatives 26 through 29, but they were all inactive. Finally, when we prepared and tested thiochromenothiazole derivatives 30 and 31, both the parent thiazoleamine and its propanamide were found to inhibit HUVEC proliferation with an IC₅₀ around 2 μM. Another variation we tried was to enlarge the B-ring to make seven-membered tricyclic thiazole derivatives 32 through 35 and in this set all of them exhibited IC₅₀ values below 10 μM, except thiazoleamine 32. Reading into the SAR data accumulated so far, we reasoned that the important feature contributing to the antiproliferative activity of the thiazoleamines against HUVEC is the tricyclic framework (chiefly based on the fact that compounds 20 through 25 were inactive). We also concluded that the amino group of tricyclic thiazoleamine is permissive to derivatization without signigicant loss in activity (IC₅₀'s of amine 30 and amides 3, 9, 31, 33, and 34 are below 4 μM). Interestingly, further analysis revealed that the two structural scaffolds with good HUVEC inhibitory activity, namely thiochromenothiazole and benzocycloheptathiazole are perfectly superimposable (Figure 1) due to the larger atomic radius of sulfur in 31 spanning the space occupied by two methylene groups in the seven-membered ring counterpart 33. Thus, we reckoned that further medicinal chemistry based upon thiochromenothiazole and benzocycloheptathiazole skeletons is likely to afford more potent inhibitors of HUVEC proliferation. We synthesized and evaluated 8-chloro substituted thiochromeno[1,2-d]thiazoleamine 36 which was found to be about as potent as the parent thiazoleamine 30 for inhibiting the growth of HUVEC (Table 2). Interestingly, when thiazoleamine 36 was alkylated with 2-furanylmethyl group, the resulting secondary amine 37 became the most potent inhibitor of HUVEC in the entire collection of compounds. The potency of HUVEC inhibition of ethyl carbamate 38 was close to that of propanamide 31. In the series of benzocycloheptathiazoles, imide 34 turned out to be more potent for HUVEC inhibition than the amide 33 (1.5 versus 3.7 μM). However, carbamate 39 was an inferior inhibitor. Remarkably, in the series of thiochromenothiazoles both amine (30, 36, and 37) and amide (31 and 38) derivatives were generally potent at inhibiting HUVEC proliferation, whereas in the case of benzocycloheptathiazoles—32 through 34, 39, and 40—strictly the amide 33 and imide 34 were good inhibitors, but the parent amines 32 and 40 were poor inhibitors. At this juncture it was deemed necessary to assess if the HUVEC inhibiting thiochromenothiazoles and benzocycloheptathiazoles were capable of exhibiting cell-type selectivity We assessed the effects of these compounds on the proliferation of three cancer cell lines—Jurkat-T (T cell leukemia), BT474 (breast cancer), and HeLa (cervical cancer). Thiochromenothiazoles 30, 31, and 36 through 38 were in general only two-fold more selective to HUVEC. On the other hand, of the two benzocycloheptathiazole amides that inhibited HUVEC potently, 33 showed an impressive selectivity for HUVEC (>5 fold), and 34 was reasonably selective.

Figure 1.

Structures of 31 (green) and 33 (gray) were minimized (MM2) and were overlayed using Chem3D.

Upon a closer scrutiny of the cell inhibition data, we became curious with the observation that compounds 30, 34, 36, and 37 manifested high inhibitory activities against Jurkat-T cells besides inhibiting HUVEC. A thorough literature search revealed that analogs of N-(4H,5H,6H-benzo[6,7]cyclohepta[2,1-d]thiazol-2-yl)2,6-difluorobenzamide and the corresponding inversely fused thiazole derivatives were claimed to be potential immunosuppressive agents.¹¹ Thus, we synthesized few more analogs consisting of 2,6-difluorobenzamide moiety—41 and 42 derived from thiochromenothiazoles, and 43 and 44 derived from benzocycloheptathiazoles. Among these benzamides, 43 turned out to be a potent (1.5 μM) and highly selective inhibitor of HUVEC, while inhibition by 42 was moderate. Surprisingly, benzocycloheptathiazoleamide 44 was nearly ten times less potent than 43 and this can only be attributed to the presence of a fluoro group on the A-ring.

Analyzing the SAR of thiochromenothiazole and benzocycloheptathiazole derivatives thus far (Table 2), we concluded that the A-ring and the 2-amino group on the thiazole are the two apparent loci amenable for modifications to seek potency improvement and perhaps attachment of affinity probes. Thus, we prepared two thiochromenothiazoleamine derivatives 45 and 46, and a benzocycloheptathiazoleamide 47, all carrying probe attachable functionalities (via azide-alkyne Huisgen cycloaddition). Not only did these derivatives retain HUVEC inhibitory activity, but also, amide 46 which is a 2.9 μM inhibitor, may serve as a useful tool in the target identification studies.

In order to evaluate the effectiveness of these two classes of compounds in blocking angiogenesis, we performed in vitro tube formation assay¹² with a potent thiochromenothiazoleamine 37 and a selectively potent benzocycloheptathiazole derivative 43 (Figure 2). The HUVEC tube formation was inhibited by both 37 and 43 in a dose-dependent manner. An aggregate of tube length, size, and number of junctions was quantified (Figure 2). Thiazoleamine 37 inhibited HUVEC tube formation very potently (≈35% of control at 5 μM) whereas 43 showed a moderate inhibition (≈75% of control at 5 μM).

Figure 2.

Representative images of tubes formed on Matrigel after seeding with HUVEC and treating with DMSO, 37 or 43 (2, 5, and 10 μM). Tubes were stained with Calcein-AM and observed under fluorescence microscope at 40× magnification. Tube formation was quantified using AngioQuant v1.33 software (The MathWorks, Natick, MA). The experiments were performed in triplicate.

In conclusion, we identified tricyclic thiazole derivatives as inhibitors of HUVEC proliferation with distinct structures and new mechanisms. We prepared, in total, 47 analogs by systematically manipulating the three rings of the tricyclic thiazoles. Two tricyclic systems, thiochromenothiazoles and benzocycloheptathiazoles emerged as the most potent inhibitors. Tw most potent HUVEC inhibitors 37 and 43 (1.1 and 1.5 μM respectively) representing the two tricyclic scaffolds stated above, also inhibited endothelial tube formation. Also, benzocycloheptathiazole 43 is a very selective inhibitor of HUVEC and hence these compounds serve as potential leads for the development of promising antiangiogenic agents. We will pursue synthesis of more analogs related to these two lead structures to seek further improvement in potency and we plan to undertake a study to elucidate the mechanism of HUVEC inhibition by thiochromenothiazoles and benzocycloheptathiazoles.