Design, synthesis, and in vitro evaluation of BP-1-102 analogs with modified hydrophobic fragments for STAT3 inhibition

Abstract Twelve novel analogs of STAT3 inhibitor BP-1-102 were designed and synthesised with the aim to modify hydrophobic fragments of the molecules that are important for interaction with the STAT3 SH2 domain. The cytotoxic activity of the reference and novel compounds was evaluated using several human and two mouse cancer cell lines. BP-1-102 and its two analogs emerged as effective cytotoxic agents and were further tested in additional six human and two murine cancer cell lines, in all of which they manifested the cytotoxic effect in a micromolar range. Reference compound S3I-201.1066 was found ineffective in all tested cell lines, in contrast to formerly published data. The ability of selected BP-1-102 analogs to induce apoptosis and inhibition of STAT3 receptor-mediated phosphorylation was confirmed. The structure–activity relationship confirmed a demand for two hydrophobic substituents, i.e. the pentafluorophenyl moiety and another spatially bulky moiety, for effective cytotoxic activity and STAT3 inhibition.


Introduction
Signal transducers and activators of transcription (STAT) signalling pathways belong to signalling modules that are relatively simple and straightforward: upon appropriate stimuli (binding of a polypeptide-based ligand), a plasma membrane receptor-associated kinase phosphorylates, and thus activates (without use of a second messenger) transcription factors of the STAT family to execute a specific biological function by regulating gene expression. Human STAT transcription factors (STAT1, STAT2, STAT3, STAT4, STAT5a, STAT5b, and STAT6) are involved in the response to autocrine or paracrine stimuli in a multitude of cellular functions such as regulation of cell proliferation, apoptosis, differentiation, and stress response 1-3 .
The STAT3 protein structure is characterised by four domains involved in oligomerisation, binding to DNA, and transactivation activity, and a Src homology 2 (SH2) domain mediating phosphorylation-dependent dimerisation. The SH2 domain contains tyrosine 705 (Y705), which is phosphorylated in response to extracellular signalling by kinases of the Jak and Src families. STAT3 proteins with phosphorylated Y705 (pY705) form dimers via reciprocal binding of SH2 domains and they are further translocated into the nucleus to bind to specific DNA response elements. STAT3 are both activators and repressors of hundreds of genes, including cell cycle regulators MYC and CCND1 and anti-apoptotic BCL-2 family genes BCL2, BCL2L2 (BCL-W), BCL2L1 (BCL-XL), MCL1, and BIRC5 (survivin; reviewed in refs. 4,5 ).
The signalling pathway executed by STAT3 (hence the STAT3 signalling pathway) attracts specific attention of pharmaceutical research due to its involvement in human inflammatory and malignant diseases. Under normal conditions, activation of the STAT3 signalling pathway is transient due to a number of negative regulators. However, STAT3 signalling was found to be constitutively activated in several solid and haematological malignancies (e.g. refs. 6,7 ). In effect, STAT3 signalling contributes to all hallmarks of cancer as defined by Hanahan and Weinberg 8 , including aberrant cell proliferation, regulation of cell death, tumour-promoting inflammation, invasion and metastasis, immunosuppression, genome instability, tumour metabolism, and treatment resistance 4,5,9,10 . Constitutive STAT3 activation is associated with worse prognosis, for instance, of patients with gliomas 11,12 . Due to this functional role in malignant progression, the inhibition of activity of STAT3 signalling is considered a potential target for cancer treatment 13,14 .
Current approaches to inhibiting the STAT3 signalling pathway focus on several levels of the signalling module from inhibition of upstream activating kinase Jak1, inhibition of SH2 domain phosphorylation and dimerisation, to inhibition of the transcription-activating domain and DNA-binding activity [15][16][17][18][19][20][21][22] . However, despite intensive efforts, currently there is no STAT3 inhibitor in clinical practice. As summarised by Huang et al., the reasons for that are multiple. The most unfavourable for development of inhibitors targeting the SH2 domain is the fact that the pY705peptide binding site within the STAT3 SH2 domain spreads over large, relatively flat surfaces and lacks well-defined deep binding pockets. It makes the specific binding of the small molecular inhibitors difficult. In addition, several direct STAT3 inhibitors are also characterised by low aqueous solubility, low cell permeability and poor oral bioavailability, which may hinder the drug development and clinical studies 23 .
Among the small molecules, BP-1-102 was formerly found to be a direct STAT3 inhibitor with reasonable in vivo tumour-inhibiting activity and commercial availability. It was synthesised as an analogue of S3I-201.1066 and it is proposed to bind specifically to the STAT3 SH2 domain 24 .
In this study, novel analogs of commercially available STAT3 inhibitor BP-1-102 (1 ; Table 1) were designed, synthesised, and their cytotoxic activity plus receptor-mediated STAT3 phosphorylation inhibitory ability were determined.

Results and discussion
2.1. Molecular design and prediction of physical-chemical properties BP-1-102 (1) was found to be an effective inhibitor of STAT3 both in vitro and in vivo 24 , but with limited solubility in water/buffer compartments. Therefore, it is possible to modify the hydrophobic substituents by other moieties to plausibly decrease the hydrophobicity of the molecule and enhance the water/buffer solubility, while at the same time evaluating the antitumor efficacy (Table 1). First, the cyclohexylbenzyl moiety can be omitted or replaced by less bulky hydrophobic fragments (e.g. t-butyl, methoxy, and trifluoromethyl). Additionally, the pentafluorophenyl moiety can be replaced by a tolyl fragment (similarly as in the effective STAT3 inhibitor S3I-201.1066 25 , 2) or by mono-, di-, or trifluorophenyl with very similar steric/spatial properties.
The physical-chemical properties (logP, logD 7.4 , and logS 7.4 ) of the designed molecules were predicted to depict differences from the parent molecule 1 (Table 1). Using the proposed changes in hydrophobic substituents, the logP values were usually decreased compared to standards 1 or 2, reaching the optimal range for barrier penetrability ($5) 25 with the exception of molecules 5, 6, 10-12, and 14. In case of logD, the values were decreased when compared to standards, again with the exception of molecules 5, 6, 10-12, and 14, suggesting lower lipid binding. Concerning the solubility prediction (logS 7.4 ), the majority of the proposed molecules, except for molecules 10 and 14, should be at least slightly more soluble than the standard 1.

Synthesis of novel compounds
The synthetic strategy towards preparation of standards 1-2 and designed p-aminosalicylic acid sulphonamides 3-14 was based on the methodology reported by Page et al. 26 The convergent synthesis involved two main routes. Initial benzylation of p-aminosalicylic acid with BnBr and t-BuOK in DMF gives 59% yield of desired dibenzylated derivative 15a, with tribenzylated 17c and mono-Obenzylated 15b byproducts (Scheme 1). Dibenzylated amine 15a was used for direct condensation with selected p-substituted benzaldehydes 16a-e with subsequent reductive amination. These conditions allowed preparation of a series of substituted secondary amines 17a-e, which represent the first coupling partners for amide reaction. Substituted aldehyde 16a was prepared according to formerly reported procedures 27 , other aldehydes 16b-e were commercially purchased.
Sarcosine t-butyl ester hydrochloride was used as the starting material for preparation of the carboxylic coupling partner (Scheme 2). Sarcosine ester in reaction with selected benzenesulfonyl chlorides 18a-f provided corresponding sulphonamides 19a-f in good to excellent yields. The following acid hydrolysis promoted with TFA in CH 2 Cl 2 allowed preparation of desired substituted sulphonamide carboxylic acids 20a-f in excellent yields. The substituted carboxylic acids 20a-f were the second partners for the amide coupling reaction.
The microwave-assisted amide coupling 28 between substituted secondary amines 17a-e and substituted carboxylic acids 20a-f was promoted with Ph 3 PCl 2 and represents the key reaction step 29,30 . Although yields of the coupling reactions were highly variable (Table 1), provided amides 21a-n were the direct precursors of the final products. Thus, final debenzylation catalysed with 5% Pd/C in CH 3 OH/THF (1:1) yielded target p-aminosalicylic acid analogs 1-14 (Scheme 3). The structure of the final products was characterised by NMR (Figures S1-S14) and HRMS analysis. The purity was estimated using the HPLC method and was found to be >95% for all final products.
To complement the resazurin assay, which estimates cell metabolic activity as an indirect measure of the compound cytotoxicity, cell death induction was confirmed by annexin V and propidium iodide (PI) staining for apoptosis/necrosis and analysis by   Figure S15C). Thus, the compound 6 showed comparable inhibitory activity as the parent molecule BP-1-102 (1). Note that the IC 50 of BP-1-102 agreed with the previously estimated values 24 . Further, the cytotoxicity and possible cell type-dependent differences of the effective BP-1-102 analogs were tested in human cancer cells of various origin. Cell lines derived from several human solid malignancies, namely, human breast adenocarcinoma MCF-7, cervix carcinoma HeLa, prostate carcinoma DU-145, and osteosarcoma U2-OS, were treated with the most potent compound 6 at concentrations 0, 5, 10, 20, and 50 mM for 24 h, and cytotoxicity was estimated by the resazurin assay. As shown in Figure 2(A), all tested cell lines were sensitive to BP-1-102 (1) to a comparable extent; nevertheless, HeLa and U2-OS cells were relatively more sensitive to BP-1-102 (1) compared to cells with constitutive STAT3 activity DU-145 and MCF-7. Compound 6 in the DU-145, HeLa, MCF-7, and U2-OS cells manifested a similar or slightly lower cytotoxic effect as the parent molecule BP-1-102 (1). Furthermore, the cytotoxicity of compound 6 was compared to BP-1-102 (1) in human glioblastoma cell lines U-87 and T98 in concentration range 0-80 mM for 24 h using the crystal violet assay. T98 cells, one of the most chemoresistant human glioblastoma cell lines, showed remarkable resistance to both BP-1-102 (1) and 6 when compared to U-87 cells (Figure 2(B)). Altogether, the novel analogue 6 manifested similar cytotoxic activity for human cancer cells as parental compound BP-1-102.
The less potent compound 14 was tested in four human cancer cell lines DU-145, HeLa, MCF-7, and U2-OS. Although the cytotoxic effect on HeLa, MCF-7, and U2-OS was comparable to BP-1-102 (1) and compound 6 at 50 mM, it was markedly decreased at lower concentrations, and the prostate carcinoma DU-145 cell line exhibited a weaker response to compound 14 ( Figure 2(A)).
To uncover the possible differences between mouse vs. human cells, the whole set of compounds was screened in mouse lung TC-1 and prostate TRAMP-C2 cancer cell lines using the MTT assay ( Figure S16). Similarly to human cells, compound 2 (S3I-201.1066) did not demonstrate any cytotoxic effect in both murine cell lines, whereas compounds 6 and 14 displayed cytotoxic activity comparable to BP-1-102 (1) (Figure 3) with IC 50 summarised in Table  2. Additionally and in contrast to human cells, compound 8 manifested a mild cytotoxic effect. Further, the cytotoxicity of the selected compounds 1, 6, and 14 in murine cells was also confirmed by annexin V and PI staining for cell apoptosis/necrosis and by FACS analysis (Figures S17-S18).

Inhibitory effect on STAT3 phosphorylation
The inhibitory effect of the selected compounds on receptor (JAK)-mediated phosphorylation of STAT3 were demonstrated in MDA-MB-231 cells with constitutive STAT3 Y705 phosphorylation (1) and newly synthesised compound 6 was tested on human glioblastoma U-87 and T98 cell lines using crystal violet staining. Data were normalised to control/ untreated samples and plotted as mean ± SD (n ¼ 3).
(pY705). The cells were exposed to BP-1-102 (1), compounds 6 and 14 in a micromolar concentration range for 4 h, and then the levels of STAT3 pY705 and total STAT3 were probed by immunoblotting with specific antibodies. As shown in Figure 4(A), both the reference and two novel compounds inhibited constitutive STAT3 phosphorylation. Similar effects of all three compounds were also observed using mouse cell lines TC-1 and TRAMP-C2 (Figure 4(B); cytotoxic concentrations of individual drugs used in these experiments were derived from annexin V/PI apoptosis analysis; see Figures S17-S18). It can be concluded that both novel compounds 6 and 14 manifested inhibitory effects on receptormediated phosphorylation of Y705 of STAT3 in both human and mouse cell lines.

Structure-activity relationship
The structure-activity relationship was based on chemical modification made in both hydrophobic regions and its correlation to cytotoxic effects of the studied molecules. First, the cyclohexyl moiety omission led to a decrease of cytotoxic activity (1>4), i.e. increased IC 50 values in the selected cell lines (Table 2; Figure 5). The replacement of cyclohexyl by t-butyl or trifluoromethyl resulted in higher cytotoxicity (decreased IC 50 ) for all tested cell lines, and spatially bulkier t-butyl was found twofold more potent than trifluoromethyl, but still less effective than the original cyclohexyl moiety (1>6>14).
Additionally, replacement of the pentafluorophenyl moiety by mono-, di-, or trifluorophenyl led to complete loss of cytotoxic activity irrespective of changes of the R 1 substituent in compounds 2-3, 5, 7, 9, and 10-13. Only one compound with the intact pentafluorophenyl group (8) showed no cytotoxic activity in human MDA-MB-231 and U373 cells, while mild cytotoxicity in murine cells ( Figure S16), indicating that the pentafluorophenyl group highly contributes to the cytotoxic effect of the BP-1-102derived scaffold. Interestingly, our data obtained with compound S3I-201.1066 (2) differ from the previously published studies. S3I-201.1066 (2) was reported to affect viability of human breast cancer MDA-MB-231 and pancreatic cancer Panc-1 cell lines 31 . Although our findings do not support that, as compound 2 did not show any cytotoxic effect, they are still in line with the importance of the pentafluorophenyl moiety for cytotoxic activity and STAT3 inhibition.
Concerning prediction of the physical-chemical parameters, the molecules (1, 6, 14) with increased logP, logD 7.4 and decreased logS 7.4 were found to have better cytotoxic activity, presuming these compounds to be well penetrable across the membranes, but less soluble in the water/buffer environment. Concerning the cytotoxic activity overall, hydrophobic substituents with an optimal pentafluorophenyl moiety in the sulphonamidic part of the molecule and a spatially bulky moiety in the benzyl part of the molecule are important prerequisites for the biological effect of the compound.    23.1 ± 10.5 a The inhibitory effect of the compounds on the proliferation of four human cell lines was determined by the resazurin assay: SDstandard deviation; data are the mean ± SD from at least two independent experiments. b The inhibitory effect of the compounds on the proliferation of two mouse cell lines was determined by the MTT assay: SDstandard deviation; data are the mean ± SD from at least two independent experiments.

Conclusions
Twelve novel analogs of STAT3 inhibitor BP-1-102 were designed with the aim to modify hydrophobic fragments of the molecule, and these compounds were successfully synthesised. The cytotoxic activity of reference compounds BP-1-102 (1) or S3I-201.1066 (2) and the novel compounds was screened using two human and two mouse cancer cell lines, and two potent inhibitors (6,14) were selected. The selected compounds were further tested in detail in other six human and two murine cancer cell lines, in all of which they manifested the cytotoxic effect in a micromolar range. Their ability to induce apoptosis and inhibition of STAT3 receptor-mediated phosphorylation was confirmed. Notably, compound S3I-201.1066 (2) emerged as ineffective for the growth inhibition of the used cell lines. Importantly, it was found that glioblastoma cell line T98, one of the most chemoresistant glioblastoma cell lines, is also resistant to BP-1-102 (1) and its most effective analogue in this study, compound 6. The structure-activity relationship confirmed the demand for two hydrophobic substituents, i.e. the pentafluorophenyl moiety and the second spatially bulky moiety for effective cytotoxic activity and STAT3 inhibition.

General synthetic methods
All used commercial reagents and solvents were purchased in the highest available purity from supplier Sigma-Aldrich (St. Louis, MO). Prepared compounds were purified by column chromatography on silica gel Kieselgel 60 (0.040-0.063 mm, 230-400 mesh, Merck, Kenilworth, NJ). Thin layer chromatography was performed on Merck silica gel 60 F 254 analytical plates, using a mobile phase corresponding to the mobile phase used for the glass column chromatography. Structures on TLC were detected using either UV light (254 nm) or spraying with the detection reagent (10% solution of phosphomolybdic acid in MeOH) with subsequent heating.  Microwave-assisted reactions were accomplished in a focussed Discover Microwave System (CEM Corporation, Stallings, NC), and the contents of the vessel were cooled down rapidly by a stream of compressed air. Melting points of solid compounds were recorded in a Melting Point Apparatus-B€ uchi M-565 and were uncorrected. Electrosprayionisation mass spectrometry (ESI-MS) was evaluated in an Agilent 6470 Triple Quadrupole mass spectrometer (Agilent Technologies, Palo Alto, CA) in a positive or negative mode. High resolution mass spectrometry (HRMS) was determined by a Q Exactive Plus hybrid quadrupole-orbitrap spectrometer (ThermoFisher, Bremen, Germany). NMR spectra were recorded in CD 3 OD-d 4

General procedure for benzylation (15a-b)
t-BuOK (2.418 g; 21.55 mmol) was added to p-aminosalicylic acid (3000 g; 19.59 mmol) dissolved under N 2 atmosphere in dry DMF (103 ml; 5.26 ml/mmol) cooled to 0 C. After 15 min, BnBr (2.56 ml; 21.55 mmol) was added dropwise, and the mixture was stirred for another 15 min at 0 C. The mixture was then allowed to warm to RT and stirred for 5.5 h. Further, the reaction mixture was again cooled down to 0 C and a second portion of t-BuOK (2.418 g; 21.55 mmol) was added. After 15 min, a second portion of BnBr (2.56 ml; 21.55 mmol) was added dropwise to the reaction. The mixture was stirred for another 15 min at 0 C then allowed to warm to RT and stirred for 17 h. The reaction mixture was cooled to 0 C, and then saturated aqueous solution of NaHCO 3 (50 ml) and H 2 O (75 ml) were added. The mixture was transferred to a separatory funnel, and the organic product was extracted to ethyl acetate (EA) (3 Â 20 ml). Organic layers were combined, washed with H 2 O (75 ml), then with saturated aqueous NaCl (40 ml), and dried with Na 2 SO 4 and filtered. The organic solvent was removed under reduced pressure, and the residue was purified by column chromatography on silica gel in the mobile phase petrolether (PE):EA:Et 3 N (4:1:2%) to produce the desired product 15a with tribenzylated 17c and monobenzylated 15b as byproducts.

General procedure for preparation of substituted secondary amine (17a-e)
To a stirred solution of corresponding aldehyde 16a-e (1.70 mmol) in anhydrous MeOH (13.3 ml; 13.3 ml/mmol) under N 2 atmosphere, 4 Å molecular sieves were added followed by acetic acid (0.10 ml; 1.80 mmol). The mixture was stirred for 1 h at RT, the dibenzylated amine 15a (1 mmol) was added, and the mixture was heated to 50 C for 3.5 h. The reaction mixture was cooled to RT, NaBH 3 CN (157 mg; 2.50 mmol) was added, and the mixture was stirred for 40 h. The solvent was removed under reduced pressure, the residue was dissolved in CH 2 Cl 2 , filtered through a glass frit, and concentrated under reduced pressure. The organic phase was transferred to a separatory funnel, then saturated aqueous NaHCO 3 (8 ml) and H 2 O (8 ml) were added for extraction. The aqueous layer was extracted with two portions of CH 2 Cl 2 (both 8 ml). Combined organic layers were washed with H 2 O (12 ml) and brine (12 ml). After drying with Na 2 SO 4 , filtration and evaporation, the residue was purified by glass column chromatography on silica gel in mobile phase heptane:EA:Et 3 N (5:1:2%) to give corresponding secondary amine 17a-e.

General procedure for preparation of substituted t-butyl sulphonamide acetates (19a-f)
The solution of sarcosine t-butyl ester hydrochloride (182 mg; 1.00 mmol) was dissolved in anhydrous CH 3 CN (3.59 ml; 3.59 ml/ mmol) under N 2 atmosphere and the mixture was cooled to 0 C. Further, DIPEA (0.44 ml; 2.50 mmol) was added and the mixture was stirred at 0 C for 10 min. Corresponding substituted sulphonyl chloride 18a-f (1.50 mmol) was added dropwise and the mixture was stirred for 30 min at 0 C, then it was allowed to warm to RT and stirred for 30 min. The reaction was quenched with H 2 O (5 ml) and the product was extracted to EA (5 ml). The aqueous layer was extracted with two portions of EA (both 5 ml). Organic layers were combined, washed with saturated aqueous NaHCO 3 (8 ml), then with brine (8 ml). The organic layer was dried with Na 2 SO 4 , filtered and evaporated. The crude product was purified by glass column chromatography on silica gel in mobile phase PE:EA (20:1) to produce desired t-butyl esters 19a-f.

4.2.4.
General procedure for preparation of substituted carboxylic acid (20a-f) To a solution of substituted t-butyl ester sulphonamide acetates 19a-f (1 mmol) in CH 2 Cl 2 (6.3 ml; 6.3 ml/mmol), TFA (6.3 ml; 6.3 ml/ mmol) was added and the mixture was stirred at RT for 3.5 h. Solvent and TFA were then evaporated under reduced pressure, to give pure desired substituted acids 20a-f.