|Home | About | Journals | Submit | Contact Us | Français|
A series of dimeric macrocyclic compounds were prepared and evaluated as antagonists for inhibitor of apoptosis proteins. The most potent analogue 11, which binds to XIAP and c-IAP proteins with high affinity and induces caspase-3 activation and ultimately cell apoptosis, inhibits growth of human melanoma and colorectal cell lines at low nanomolar concentrations. Furthermore, compound 11 demonstrated significant antitumor activity in the A875 human melanoma xenograft model at doses as low as 2 mg/kg on a q3d schedule.
The ability of tumor cells to evade apoptosis, or programmed cell death, is a hallmark of human cancers.1 In addition to cancer cell survival, defects in apoptotic pathways (i.e., extrinsic and intrinsic) may also contribute to tumor progression and chemo-resistance. The inhibitor of apoptosis proteins (IAPs), which includes the X-linked inhibitor of apoptosis protein (XIAP) and two cellular inhibitor of apoptosis proteins (cIAP-1 and cIAP-2), are a family of pro-oncogenic proteins that can inhibit apoptosis by regulating initiator and effector caspases.2 XIAP directly binds to and inhibits caspase-3, -7, and -9 through its second and third baculovirus IAP repeat (BIR2 and BIR3) domains, resulting in inhibition of apoptosis.3 While not direct inhibitors of caspases, cIAPs are involved in NF-κB signaling and play important roles in regulating caspase-8 activation.4 Elevated expression of XIAP and cIAPs has been reported in many cancers and is associated with poor prognosis.5
The antiapoptotic activities of the IAP proteins can be neutralized by the mitochondrial protein Smac/Diablo (second mitochondrial activator of caspases/direct IAP binding protein with a low isoelectric point), which binds to the BIR domains of the IAP proteins through its N-terminal AVPI tetrapeptide binding motif.6 Importantly, it has been demonstrated that the AVPI peptide itself, or small molecule AVPI mimetics, can potently antagonize the IAP proteins and elicit antitumor activity in vivo.7−10 Furthermore, dimeric AVPI mimetics, which can bind to both the BIR2 and BIR3 domains and thus achieve much higher binding affinity for the IAP proteins, were shown to be significantly more potent at inducing apoptosis in multiple cancer cell lines. Several Smac mimetics that bind selectively to the BIR domains of XIAP and cIAPs have entered clinical trials and shown encouraging results.11
We recently reported the discovery of dimeric macrocyclic IAP antagonist 1 (Table 1) via the generation of “millamolecular” libraries.12,13 Macrocyclization has been exploited extensively as a means of conformationally restricting peptides and thereby improving their proteolytic stability, target affinity, and cell penetration.14−16 We have shown that compound 1 binds to both cIAP and XIAP proteins with high affinities and is active in cellular assays despite its peptidic nature and high molecular weight. One drawback of compound 1 is that it is about 10-fold less potent in cell-based proliferation assays than other nonmacrocyclic IAP antagonists we have recently disclosed.17,18 As a result, significantly higher doses are required to achieve similar levels of efficacy in mouse tumor xenograft models. In an effort to improve the in vitro and in vivo potencies of this millamolecular series with the ultimate goal of identifying a clinical candidate, we systematically designed and synthesized a series of analogues of 1. Herein, we report the synthesis, structure–activity relationship, and in vivo evaluation of a series of macrocyclic dimeric IAP antagonists with improved potency and pharmacokinetic properties.
Compound 1 is a potent antagonist of XIAP BIR2-3 protein (IC50 = 1.4 nM, Table 1) and inhibits the proliferation of A875 human melanoma cells with an IC50 of 73 nM. On the basis of our predicted binding model and previous SAR, we hypothesized that compound 1 occupies the same binding pocket as the AVPI peptide on the surface of the BIR2-3 protein (Figure (Figure1).1). In this model, the C-terminal carboxylic acids are solvent exposed and do not contribute significantly to binding potency. In contrast to this prediction, however, the mono- and bis-methyl esters analogues 2 and 3 are significantly less potent than 1 in both the XIAP BIR2-3 FRET binding assay19 and the A875 antiproliferation assay (IC50 = 310 and 690 nM, respectively, Table 1). Several additional analogues of 1, where the carboxylic acids were replaced with nonacidic primary or secondary amide groups, also gave poor biochemical and cellular activities (data not shown). These results lead us to postulate the acid moieties may be important for conformational reasons. Compounds with the acid moieties may be able to more easily adopt the conformation required for binding simutaneously to the BIR2 and BIR3 proteins. Consistent with this hypothesis, replacing one or both of the carboxylic acid groups with similarly acidic cyclopropyl acylsulfonamide or tetrazole moieties was well tolerated. The bis-cyclopropyl acylsulfonamide 4 is equipotent to 1 in both biochemical (XIAP BIR2-3 IC50 = 1.8 nM) and cellular antiproliferation assay (A875 IC50 = 79 nM), whereas monocyclopropyl acylsulfonamide analogue 5 gave similar biochemical potency but improved cellular potency (A875 IC50 = 39 nM).20
On the basis of these promising results, we decided to determine if the acid isosteres had improved pharmacokinetic (PK) properties. As shown in Table 2, following a 1 mg/kg IV bolus injection, bis-cyclopropyl acylsulfonamide 4 demonstrated reduced clearance and enhanced exposure (AUC0–7 = 2350 nM h) relative to compound 1. Monocyclopropyl acylsulfonamide 5 provided lower clearance, lower steady state volume of distribution, and higher exposure than both 1 and 4 (AUC0–7 = 5850 nM h) at the same dose. Thus, in addition to maintaining an optimal level of cellular potency, the acylsulfonamide acid isosteres also benefited from improved PK properties relative to the initial lead 1.
We next sought to evaluate the effect of the triazole linkers on cellular potency, using monoacylsulfonamide 5 as reference. Previous investigations of dimeric IAP antagonists revealed that the length, rigidity, and polarity of the linker can have a profound effect on biochemical binding and cell permeability.21,22 The relatively polar and rigid triazole linker was selected in the initial optimization to take advantage of the efficiency of constructing macrocycles using azide–alkyne click chemistry.23 However, our predicted binding model indicated that both triazole groups in 1 occupy hydrophobic pockets on the surface of the XIAP BIR2-3 proteins (Figure (Figure1),1), suggesting that a less polar and nonaromatic linker may be tolerated. We explored ring closing metathesis (RCM), which is compatible with acylsulfonamide-containing peptide substrates to provide access to analogues 7 and 8, where one or both of the triazoles were replaced with less polar but similarly rigid propenyl groups. Both 7 and 8 displayed comparable biochemical and cellular activities to the bis-triazole analogue 5, thus demonstrating the triazole linker was not required for binding or cell permeability. Macrocyclization was also shown to be essential for potent in vitro activity, as the macrocycle 8 was greater than 20-fold more potent than the corresponding uncyclized compound 9 in the biochemical binding and antiproliferation assays (see Table 3).
To remove the potentially labile allyl ether functionality, reduction of the alkene groups provided bis- or monopropyl-linked analogues 10 and 11. Binding data showed that despite increased conformational flexibility, both 10 and 11 bind to XIAP BIR2-3 proteins with IC50 values in the low single-digit nanomolar range. Both compounds also displayed approximately a 5-fold improvement in cellular potency relative to compound 1 (A875 IC50 = 15 and 19 nM, respectively).
Encouraged by the excellent cellular potency of compounds 10 and 11, we evaluated their physiochemical properties to select a compound for full in vitro and in vivo characterization. In particular, we aimed to identify a compound with sufficient aqueous solubility compatible with intravenous administration. We found that in this series, aqueous solubility correlates well with lipophilicity and overall charge of the peptide. Compounds that are more lipophilic and net neutral are in general less soluble. Accordingly, the most lipophilic compound 10, while among the most potent compounds tested in biochemical and cellular assays, has greatly diminished aqueous solubility (<0.01 mg/mL at pH 7.4) relative to compounds 1 and 11 (0.13 and 0.05 mg/mL, respectively, at pH 7.4). On the basis of its in vitro potency and aqueous solubility, compound 11 was selected for further characterization in additional biological assays.
We previously described a solid-phase synthesis of compound 1 using an on-resin cyclization promoted by a Cu-mediated azide–alkyne click reaction.13 Synthesis of compound 11 on resin, however, was significantly more challenging, especially on the scale required to provide sufficient material for additional profiling. Since the key RCM reaction could only be performed in solution and the alkene-containing amino acid building blocks required to assemble the linear peptide precursors were difficult to access, a solution based method was developed to prepare compound 11. Starting from commercially available 2-naphthyl alanine 12, coupling to the known 4-substituted prolines 13a and 13b furnished compounds 14a and 14b. Successive incorporation of tert-leucine and N-methylalanine building blocks using EDC-HOAt as the coupling reagents furnished 15a and 15b after hydrolysis. HATU-mediated amide bond coupling of 15a and 15b with substituted tyrosine derivatives 16 and 18 provided pentapeptides 17 and 19, respectively. Cycloaddition of 17 and 19 in the presence of CuSO4, followed by RCM and TFA deprotection, furnished compound 8. Compound 8 was then converted to 11 via a straightforward hydrogenation reaction. Multiple batches of compound 11 were prepared on the 400–500 mg scale using this approach (see Scheme 1).
Compound 11 was evaluated in additional FRET-based binding assays and a cell-free caspase-3 rescue assay24 (Table 4). FRET assays confirmed binding of compound 11 to the isolated BIR3 domain of the XIAP protein as well as the BIR2-3 domain of cIAP-1 protein with IC50 values in the low single-digit nanomolar range. Consistent with the mechanism of action of antagonizing XIAP to relieve inhibition of effector caspases, 11 was found to increase caspase-3 activity in a dose-dependent manner, with an IC50 of 9.4 nM.
Compound 11 was further profiled to establish its in vitro safety and ADME properties (Table 4). Despite the increased lipophilicity of the acylsulfonamide substituent and alkyl linker, compound 11 possesses a favorable inhibitory profile toward human cytochrome P450 (CYP) isoforms, has very low potential for CYP induction based on the human PXR transactivation assay, and showed almost no propensity for hERG inhibition. Plasma protein-binding of 11 was measured by equilibrium dialysis. No significant difference in plasma protein binding was observed in mouse, rat, dog, or human plasma. Additionally, compound 11 demonstrated excellent metabolic stability in liver microsomes (t1/2 = 84 min in rat and >120 min in mouse, dog, and human liver microsomes). Finally, PK properties of compound 11 were assessed in mouse, rat, and dog following a single IV dose. Consistent with the predicted hepatic clearance from liver microsome metabolic stability assays, compound 11 demonstrated short to moderate half-lives and very low clearance across species.
In vivo efficacy studies were conducted with 11 in A875 human tumor xenografts implanted in athymic mice. Gratifyingly, compound 11 demonstrated robust antitumor activity (Figure (Figure2).2). Intraperitoneal administration of 11 at 2 mg/kg every 3 days for six doses resulted in 67% tumor growth inhibition (TGI). Complete tumor stasis (>100% TGI) was observed throughout dosing at 5 mg/kg. Compound 11 was also efficacious when administered twice on a weekly schedule at 5 mg/kg (TGI = 80%). Importantly, compound 11 was well tolerated. No overt toxicity (weight loss or morbidity) was observed. These results compare favorably to those of 1, where a 50 mg/kg dose on a more frequent q3d schedule was required to achieve similar efficacy.
Compound 11 was further profiled against a panel of human colorectal cancer cell lines, including those that harbor the multidrug-resistant (MDR) phenotype. Significantly, compound 11 displayed robust activities in these tumor cell lines regardless of their MDR status. For example, compound 11 was highly potent at inhibiting cell growth in both the parental HCT116 cell line and its MDR positive variant HCT116/VM4625 (IC50 = 92 and 11 nM, respectively). This is in contrast to several previously reported nonmacrocyclic IAP antagonists reported by us and others, which were shown to be susceptible to the MDR phenotype.17,26
In summary, a potent dimeric macrocyclic IAP antagonist 11 with improved in vitro and in vivo activities was identified through systematic optimization of the carboxylic acid and linker regions of compound 1. Despite its high molecular weight and large number of lipophilic aromatic rings, compound 11 demonstrated desirable in vitro pharmacology, safety, and PK parameters. Compound 11 was found to be efficacious at a much lower dose compared to compound 1 when evaluated in the A875 xenograft model. Furthermore, given the large number of tumors that are susceptible to the MDR phenotype, the intriguing potency of this novel series of IAP antagonists in MDR-positive cell lines may provide a distinct advantage for this target class. Further profiling of compound 11 in additional MDR-positive cell lines both in vitro and in vivo is ongoing and will be reported in due course.
We thank Caroline Fanslau for generation of XIAP binding data and Sarah Traeger, Gerry Evelof, and Celia D’Arienzo for analytical support.
Experimental procedures for the synthesis of all new analogues and procedures for IAP binding assays, A875 proliferation assay, caspase rescue assay, pharmacokinetic experiments, and in vivo efficacy experiments. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.5b00091.
All authors have given approval to the final version of the manuscript.
The authors declare no competing financial interest.