|Home | About | Journals | Submit | Contact Us | Français|
Structure-activity relationship studies of a 1,2,4-triazolo-[3,4-b]thiadiazine scaffold, identified in an HTS campaign for selective STAT3 pathway inhibitors, determined that a pyrazole group and specific aryl substitution on the thiadiazine were necessary for activity. Improvements in potency and metabolic stability were accomplished by the introduction of an α-methyl group on the thiadiazine. Optimized compounds exhibited anti-proliferative activity, reduction of phosphorylated STAT3 levels and effects on STAT3 target genes. These compounds represent a starting point for further drug discovery efforts targeting the STAT3 pathway.
While advancements in detection and treatment have aided in the longstanding campaign against cancer, new annual U.S. cancer cases and related deaths still exceed 1.5 million and 580,000, respectively.1 Recently, new therapies have targeted signaling pathways that are aberrantly activated in transformed cells. These pathways promote cancer cell survival and proliferation, but they play a less important role in normal cell survival.2 As a part of our interest in the development of mechanism-based anticancer agents,3,4 we have been pursuing novel small molecule inhibitors of the signal transducer and activation of transcription 3 (STAT3) pathway.5,6,7 STAT3 is a transcription factor that influences many of the acquired capabilities of cancer tumorigenesis, thereby making it an attractive target for the development of oncolytics.8–12
There is mounting evidence for its role in cancer such as increased levels of activated STAT3 (pSTAT3-Y705) observed in many cancers including head and neck squamous cell carcinomas (HNSCC).4,5 We previously reported the use of a high content phenotypic screen to identify selective inhibitors of the STAT3 activation pathway compared to STAT1 which served as an important selectivity control since the latter is a tumor-suppressive transcription factor.6,7 We identified several scaffolds that met this criteria. Herein, we describe the optimization and structure activity relationship for a series of pyrazole-containing 1,2,4-triazolo-[3,4-b]thiadiazines with selective STAT3 pathway inhibition.
The high content phenotypic screen, which utilized an interleukin-6 (IL-6)-induced STAT3 activation assay in Cal33 head and neck tumor cells, identified several triazolo-thiadiazines as selective STAT3 pathway inhibitors (e.g., 1a and 2b, Table 1).7 The biological activities of these structurally similar analogs were confirmed through resynthesis and re-assay (vide infra). These HTS/HCS hits had no effect on interferon-γ (IFN-γ)-induced STAT1 pathway activation at concentrations up to 50 µM. This selectivity was not observed for many other STAT3 pathway inhibitors reported in the literature including the pan-Janus kinase (JAK) inhibitor, pyridone 6.6 Furthermore, this series exhibited acceptable drug-like properties: low molecular weight (<400), clogP values between 3 and 4,13 and anti-proliferative activities with several HNSCC cell lines (GI50 14–45 µM with 686LN, Cal33, FaDu, and OSC19).
Triazolothiadiazines exhibit an array of pharmacological effects including anti-proliferative activities.14 However, the HTS library included a number of inactive analogs of 1a and 2b where the pyrazole was replaced with an alkyl, aryl, or alternative heterocyclic substituents (Figure 1). This suggested that we were not observing broadly promiscuous effects with this scaffold. These observations together with the desirable biological selectivity profile and favorable drug-like physical properties encouraged us to pursue a medicinal chemistry optimization effort for this series.
The synthesis of 1a and 2b required key amino-triazole intermediates 3a and 3b that were readily assembled according to literature procedures.15–19 Alkylations with the appropriate α-halo ketone and microwave-assisted cyclodehydrations afforded the original hits 1a and 2b (Scheme 1).
In addition to 1a and 2b, two sub-libraries of triazolothiadiazines that maintained either the fused cyclopentyl-pyrazole group (“a”) or the pendant phenyl-pyrazole group (“b”) with diversified R-group modifications on the triazolothiadiazine were synthesized. These compounds were prepared according to Scheme 1 with yields ranging from 24–90% by using a variety of α-halo ketones in the final microwave-assisted cyclodehydration reaction. A protected aldehyde derivative was used to prepare 9a and 9b.
We selected substituents with diverse steric, polarity, and electronic characteristics. Table 2 illustrates a representative subset of these analogs and their activities in STAT3 and STAT1 assays. The R-substitution on the thiadiazine influenced activities by a factor of greater than 10 with many modifications leading to a loss of STAT3 potency. The importance of the chlorine substitution on the arene groups was evident by the significant drop in activity seen with the removal of the halogen regardless of the pyrazole scaffold (5a vs. 1a; 5b vs. 2b). As a general trend, hydrogen (9), aliphatic (10, 11), and heterocyclic (7) R-groups were inactive regardless of the pyrazole substructure. One notable exception was the chlorothiophene analog 8a that maintained comparable potency to the initial hit. However, when the chlorothiophene was combined with the phenyl-pyrazole (“b”) scaffold, the loss of activity of analog 8b was consistent with other heterocyclic derivatives.
A few SAR trends diverged between the two pyrazole scaffolds. For example, exchanging the R-substituents of 1a and 2b provided analogs 1b and 2a that were 3–4 times less active. The reduced STAT3 potency of compounds 2a (26 µM) and 4a (21 µM) established the importance of the ortho- and para-chlorine atoms on the phenyl ring in 1a. Interestingly, the phenyl-pyrazole with the same meta-chlorophenyl R-substituent (4b) retained, or perhaps, improved the STAT3 potency compared to the para-chlorophenyl 2a, suggesting that either lipophilicity or electronic effects were important. However, the potent activity of the para-methoxyphenyl analog 6b did not support a purely electronic contribution. The selective inhibition of STAT3 over STAT1 activation was maintained within this entire subseries.
To evaluate the effect of thiadiazine modifications, hydrazone 6a was converted to dihydrothiadiazines 12 and 13 (Scheme 2). Both analogs were inactive in the STAT3 assay, which established the requirement for an unsaturated ring system.
The modified pyrazole analogs shown in Figure 2 were prepared according to Scheme 1.15 The para-chloro and para-fluoro substitutions on the 3-arylpyrazoles were well tolerated. The para-fluoro analog 14 exhibited an IC50 of 7.5 µM, which was a ~4-fold improvement in STAT3 potency over the unsubstituted analog 1b.20 In the fused pyrazole “a” series, the cyclohexyl homolog 16 was ~2-fold less active than the cyclopentyl analog 1a.
Despite the favorable biological profiles for some of these compounds, they exhibited poor metabolic stability [1a: t1/2=14 min in human liver microsomes (HLM); 4 min in mouse liver microsomes (MLM)]. Predictions of metabolic sites using SMARTCyp21 pointed toward the thioether as the most susceptible site of a cytochrome P450-mediated oxidation (see Supporting Information). Subsequent analysis determined that oxidized thioethers were inactive (vide infra).
Analysis of the HTS SAR data suggested that carbon and oxygen heterocyclic analogs of the triazolothiadiazines would be inactive despite their potential to be more metabolically stable. Therefore, in an effort to address this possible metabolic liability, we designed, prepared and tested a series of analogs that were modified at the position alpha to the sulfur atom with the aim of sterically blocking oxidation of the thioether (Scheme 3 and Table 3). These compounds were prepared by microwave-mediated cyclodehydration with the corresponding α-halo ketones and amino-triazoles. All of the final products maintained desirable physical properties such as MW and clogP (Table 3 & Supporting Information). The structure of 17 (UPCDC10263) was confirmed by X-ray analysis (Figure 3).
It was apparent from this series that alkyl substitution alpha to the sulfur atom consistently led to potent inhibition of STAT3 activation while maintaining high selectivity over STAT1. The size of the alkyl groups did not seem to influence potency since large groups, such as i-Bu (25 and 26) and i-Pr (23 and 24), were as potent or perhaps slightly more potent than the methyl analogs (17–22). The benzyl analog 27 was less active, but only one example was prepared and we cannot speculate whether this was a general trend. In contrast, alkyl groups with polar substituents (28 and 29) and gem-dimethyl substitution (30) were either significantly less potent or inactive in the STAT3 assay. Compounds 28 and 29 exhibited potent anti-proliferative activity in the absence of STAT3 inhibition that is presumed to be due to off-target effects. We prepared the sulfoxide analog of compound 21 (not shown) as the putative oxidative metabolite; however, the resultant compound was inactive.
In some cases, combining the α-methyl substitution with the 3-arylpyrazole scaffold (19, 20, and 21) yielded a 2–3 fold improvement in STAT3 potency compared to the original hit 2b (Table 3). Chiral separation (Supporting Information) provided individual enantiomers of 22 (ent-1 and ent-2), and a 3-fold difference in STAT3 potency was observed between these enantiomers. While these differences were not dramatic, the trend suggested the existence of a eutomer [22 (ent-1), configuration not assigned] and a distomer within this series. Interestingly, the eudismic ratio was even larger when considering the anti-proliferative activity, which may reflect additional differences in metabolism or transport between the enantiomers (Table 3). Importantly, effects of α-methyl substitutions on microsomal stability were also evident. By incorporating this substitution, we were indeed able to increase the metabolic stability of compound 19 in HLM and MLM (t1/2 = 27 min and 10 min, respectively), thereby doubling the stability compared to our initial hits. Some analogs containing alkyl groups larger than methyl (e.g. 23–26) exhibited low micromolar potency but suffered from reduced compound solubility (data not shown) which excluded them from further characterization. Anti-proliferative activity in Cal33 cells was also determined. In most cases, compounds that inhibited STAT3 also inhibited the growth of HNSCC cell lines (Table 4). However, direct correlations between the two activities are likely complicated by differences between assay conditions, distinct genetic alterations among the HNSCC cell lines, and/or differences between a compound’s metabolic stability, residence time, and transport.22
With these encouraging cell-based results in well-characterized HNSCC models, we evaluated the effects of compounds 19 (UPCDC10131) and 22 (UPCDC10205) on STAT3 expression levels by Western blot analysis (Figures 4 and and5).5). Consistent with the pSTAT3 HCS assay, pronounced decreases in pSTAT3 levels were observed, and both compounds exhibited significant inhibition of the downstream STAT3 target gene, Cyclin D1.23
In summary, we identified pyrazole-linked 1,2,4-triazolo-[3,4-b]thiadiazines as a new class of potent and selective STAT3 pathway inhibitors. Optimized compounds exhibited anti-proliferative activity, reduction of phosphorylated STAT3 levels, and reduction of downstream effects on STAT3 gene expression targets. Structure-activity relationships established that a pyrazole group and specific aryl substitution on the thiadiazine were required for activity. Significant improvements in potency and metabolic stability were realized through α-substitution of the thiadiazine heterocycle. Further studies of the mechanism of action of these compounds will be reported in due course.
The authors would like to thank Dr. Steven J. Geib (University of Pittsburgh) for the X-ray analysis of 17, and Mr. Peter G. Chambers (University of Pittsburgh) for LCMS/ELS analyses. We are grateful to Christina Kraml (Lotus Separations) for kindly performing chiral separation of 22. We would also like to thank Lynn O. Resnick for her contributions especially during the early hit-triage and design phase. The authors also appreciate efforts and thoughtful discussions by project team members including Shelby M. Anderson (UPCDC), Erin Skoda (UPCDC), Courtney Vowell (UPCDC), Gabriela Mustata Wilson (UPCDC), Joanne I. Yeh (Univ. Pittsburgh), Xiang-Qun Xie (UPCDC), Kyaw Z. Myint (UPCDC), Qi Xu (UPCDC), Lin Fu (UPCDC), Lirong Wang (UPCDC), Seia Comsa (UPDDI), Albert H. Gough (UPDDI), Tongying Shun (UPDDI), David Zaidins (UPDDI), Neal Green (Leidos Biomedical Research), William J. Moore (Leidos), Shizuko Sei (Leidos), Beverly A. Teicher (NCI), Dianne L. Newton (NCI), Myrtle Davis-Millin (NCI), Joel Morris (NCI), Andy Stephen (NCI). This project was funded in part with federal funds from the National Cancer Institute, National Institutes of Health, under Chemical Biology Consortium Contract No. HSN261200800001E.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.