Synthesis and preliminary evaluation of 5,7-dimethyl-2-aryl-3H- pyrrolizin-3-ones as angiogenesis inhibitors

a b s t r a c t
Sunitinib (Sutent®) is a receptor tyrosine kinase (RTK) and angiogenesis inhibitor approved for the treat- ment of renal cell carcinomas, gastrointestinal stromal tumours and pancreatic neuroendocrine tumours. A key structural motif retained throughout medicinal chemistry efforts during sunitinib’s development was the indoline-2-one group. In the search for new anti-angiogenic scaffolds, we previously reported that non-indoline-2-one-based derivatives of semaxanib (SU5416, a structurally simpler sunitinib prede- cessor that underwent Phase III trials) are active as angiogenesis inhibitors, indicating that the group is not essential for activity. This Letter describes the synthesis and structure–activity relationships of another class of non-indoline-2-one angiogenesis inhibitors related to sunitinib/semaxanib; the 5,7- dimethyl-2-aryl-3H-pyrrolizin-3-ones. A focussed library of 19 analogues was prepared using a simple novel process, wherein commercially available substituted arylacetic acids activated with an amide cou- pling reagent (HBTU) were reacted with the potassium salt of 3,5-dimethyl-1H-pyrrole-2-carbaldehyde in one-pot. Screening of the library using a cell-based endothelial tube formation assay identified 6 com- pounds with anti-angiogenesis activity. Two of the compounds were advanced to the more physiologi- cally relevant rat aortic ring assay, where they showed similar inhibitory effects to semaxanib at 10 lg/mL, confirming that 5,7-dimethyl-2-aryl-3H-pyrrolizin-3-ones represent a new class of angiogen- esis inhibitors.

Sunitinib 1 (Sutent®, Pfizer, Fig. 1)1 is an indolin-2-one-based angiogenesis inhibitor approved for the treatment of vascularised renal cell carcinomas,2 gastrointestinal stromal tumours3 and pan- creatic neuroendocrine tumours.4 Its mechanism of action involves inhibition of at least eight different receptor tyrosine kinases (RTKs), including vascular endothelial growth factor receptors 1–3 (VEGFR1–VEGFR3), platelet-derived growth factor receptors(PDGFRs) a and b, stem cell factor receptor (Kit), Fms-like tyrosine kinase 3 (FLT-3) and colony-stimulating factor-1 receptor (CSF- 1R).5 The indolin-2-one (oxindole) portion was considered crucial for activity of sunitinib 1 and was retained throughout develop- ment. Indeed, the structurally simpler indolin-2-one predecessor semaxanib (SU5416) 2 underwent a Phase III clinical trial for advanced colorectal cancer.6 In recent work, we showed that in spite of its perceived importance the indolin-2-one moiety is not essential for anti-angiogenic activity in this class, demonstrating that ring-opened 3,5-dimethyl-1H-pyrrol-2-yl-2-arylacrylate 3 variants of semaxanib 2 inhibit angiogenesis in a rat aortic ring model.7 We now report the discovery of a second class of non- indolin-2-one-based angiogenesis inhibitors related to sunitinib 1/semaxanib 2; the 5,7-dimethyl-2-aryl-3H-pyrrolizin-3-ones 4(Fig. 1).biological activity.8–13 Complex structures incorporating a 3H-pyr- rolizin-3-one-type core have been reported, including tricyclic pyr- rolo-indolones, pyrrolo-isoindolones, tetracyclic isoindolo- indolones and other polycyclic heterocycles.

A number of these are marine natural products or their derivatives.16Divergent access to 5,7-dimethyl-2-aryl-3H-pyrrolizin-3-ones appeared achievable in one-pot from 3,5-dimethyl-1H-pyrrole-2- carbaldehyde 5 and the many commercially available substituted phenylacetic acids. We had previously shown that the K+ salt of 5 (generated using KH in THF) is smoothly N-acylated by methyl chloroformate,7 leading us to consider that the same salt might also undergo N-acylation reactions with phenylacetic acids pre- activated as acid halides or with amide coupling reagents. A sim- ple, divergent synthesis was envisaged wherein activated pheny- lacetic acids generated in one flask are combined with the K+ salt of 5, which had been freshly prepared in a separate flask. Mixing of the flasks would initiate pyrrole N-acylation/amide formation and ensuing benzylic proton abstraction under the basic reaction conditions (excess KH) would trigger an intramolecular Knoeve- nagel condensation to deliver 5,7-dimethyl-2-aryl-3H-pyrrolizin- 3-ones.Preliminary synthetic efforts surveying a variety of amide cou-pling reagents, procedures and rates and orders of addition in model reactions with phenylacetic acid and pyrrole aldehyde 5 yielded 5,7-dimethyl-2-phenyl-3H-pyrrolizin-3-one 6 as the major product in many instances, as observed by TLC analysis. The high- est yield of 6 (48%, Scheme 1) was obtained when 1.1 mol equiv of amide coupling reagent 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetram- ethyluronium hexafluorophosphate (HBTU) and 2.0 equiv of N,N- diisopropylethylamine (DIPEA) were stirred in CH2Cl2 at room temperature with 1.0 equiv of phenylacetic acid to form the acti- vated hydroxybenxzotriazoyl ester in one flask (Flask 1) and2.0 mol equiv of KH was stirred in THF at 0 °C with 1.0 equiv ofaldehyde 5 in a separate flask (Flask 2) to generate the K+ pyrrolate salt.

Pre-cooling Flask 1 to 0 °C and pouring its contents into Flask 2 at 0 °C in one portion completed the procedure (Scheme 1).Employing the method with a systematic series of mono-substi- tuted phenylacetic acid derivatives carrying F, Cl, Br, Me, OCH3 and NO2 groups at each of the 20 -, 30 – and 40 -positions similarly pro- vided analogues 7–24 (Scheme 1). While the yields were low to moderate (4–48%) the simple procedure allowed rapid access to useable quantities of pure analogues for the SAR study. 1H and 13C NMR data for all compounds were unambiguous and consistent with 2-aryl-3H-pyrrolizin-3-ones. A single crystal X-ray structure obtained for 20 -NO2 derivative 22 further confirmed its structure(Fig. 2). The X-ray structure revealed that the 20 -nitrophenyl ring was tilted 35.5° relative to the plane of the 3H-pyrrolizin-3-one ring system and positioned the electron-deficient NO2 nitrogen in close proximity to the electronegative carbonyl oxygen (2.8 Å), suggesting a favourable interaction between these two atoms.Compounds 6–24 were screened for in vitro angiogenesis inhi- bitory activity at 10 lM using the human umbilical vein endothe- lial cell (HUVEC)-based endothelial tube formation assay.17 Matrix- cultured HUVECs differentiate in response to growth factors, becoming elongated and motile and able to self-organise into cap- illary-like structures over a 2–12 h period. This process is disrupted in a dose-dependent manner by small molecule inhibitors of angio- genesis.18 Quantification of angiogenesis and the inhibitory effectsof compounds uses the method of Aranda.17 Briefly, images are captured and the cells tallied and categorised as ellipsoid (undiffer- entiated), sprouting (differentiating) or connected to other cells (Supporting information, Fig. S1). Total numbers of closed polyhe- dra are also counted, along with ‘complex meshes’ or polyhedral containing walls more than 2 cells thick. Input of the data into an equation (Supporting information, Eq. S1) generates an angio- genesis score, which is then expressed as a percentage relative to the angiogenesis score of the vehicle control (100% angiogenesis).

Reductions in scores from 100% in the presence of added com- pounds correspond to angiogenesis inhibition.Figure 3a shows that 20 -NO2 analogue 22 was the most potentinhibitor in the series, reducing angiogenesis to 74.0% ± 12.6 of vehicle at 1 lM and 29.7% ± 4.1 at 10 lM. Sunitinib 1 completely inhibited angiogenesis at 10 lM. The 30 -NO2 23 and 40 -NO2 24 ana-logues showed almost no activity. Indeed all 30 (i.e., 8, 11, 14, 17, 20) and 40 (i.e., 9, 12, 15, 18)-substituted compounds failed to show notable activity, confirming that these positions do not tolerate substitution. The unsubstituted derivative 6 showed modest activ- ity (68.4% ± 6.6 of vehicle) and similar activities were observed with the 20 -fluoro 7, 20 -bromo 13, 20 -methyl 16 and 20 -methoxy 19 analogues. 20 -Chloro derivative 10 showed no activity. Com-pound 22 was shown to be 4-fold less toxic than sunitinib 1 towards related SVEC4-10 murine endothelial cells (Supporting information, Fig. S2), consistent with general cytotoxicity not being responsible for the observed anti-angiogenic effects. We conclude that while the 30 – and 40 -positions are intolerant of substitution, 20 -substituents are tolerated to some extent and in some cases (e.g., NO2 group) they can increase activity. This is in agreementpounds 6 and 22 were advanced to a rat aortic ring angiogenesis assay to corroborate their anti-angiogenic properties in a more physiologically relevant context. Both compounds showed robust inhibitory effects at 10 lg/mL, confirming 5,7-dimethyl-2-aryl- 3H-pyrrolizin-3-ones as a new class of angiogenesis inhibitors. Fur- ther structure–activity studies aimed at identifying more potent inhibitors are warranted. Key targets to pursue would include suni- tinib-like analogues, where the basic N,N-diethylethylenediamineside chain amide of sunitinib is appended at the 6-position of 5,7-dimethyl-2-aryl-3H-pyrrolizin-3-ones. Profiling the class for activity against a panel of RTKs should provide insights into the mechanism of action of these compounds.