Dual inhibition of Janus and Src family kinases by novel indirubin derivative blocks constitutively‐activated Stat3 signaling associated with apoptosis of human pancreatic cancer cells

Abstract

Constitutively‐activated JAK/Stat3 or Src/Stat3 signaling plays a crucial role in tumor cell survival, proliferation, angiogenesis and immune suppression. Activated JAK/Stat3 or Src/Stat3 has been validated as a promising molecular target for cancer therapy. However, prolonged inhibition of Src family kinases (SFKs) leads to reactivation of signal transducer and activator of transcript 3 (Stat3) and tumor cell survival through altered JAK/Stat3 interaction. This compensatory feedback suggests that dual inhibition of Janus kinases (JAKs) and SFKs might be a promising strategy for targeting downstream Stat3 signaling in the clinic. In this study, we identify that the natural product derivative E738 is a novel dual inhibitor of JAKs and SFKs. The IC50 values of E738 against recombinant JAKs and SFKs in vitro are in the ranges of 0.7–74.1 nM and 10.7–263.9 nM, respectively. We observed that phosphorylation of both Jak2 and Src was substantially inhibited in the submicromolar range by E738 in cultured human pancreatic tumor cells, followed by blockade of downstream Stat3 activation. E738 down‐regulated expression of the Stat3 target proteins Mcl‐1 and survivin, associated with induction of apoptosis. Computational models and molecular dynamics simulations of E738/Tyk2 or E738/Src in silico suggest that E738 inhibits both tyrosine kinase 2 (Tyk2) and Src as an ATP‐competitive ligand. Moreover, the planar E738 molecule demonstrates a strong binding affinity in the compact ATP‐binding site of Tyk2. In sum, E738 is the first dual inhibitor of JAKs and SFKs, followed by inhibition of Stat3 signaling. Thus, according to in vitro experiments, E738 is a promising new therapeutic agent for human pancreatic cancer treatment by blocking both oncogenic pathways simultaneously.

Highlights

• ► We identify that indirubin analog E738 is the first dual inhibitor of JAKs and SFKs.
• ► Blockade of JAK/Stat3 or Src/Stat3 is associated with induction of apoptosis.
• ► These findings suggest a pharmacological mechanism of action of E738 in cancer cells.
• ► E738 is a promising new therapeutic agent for human pancreatic cancer treatment.

Abbreviations

IRD

indirubin derivative

STAT

signal transducer and activator of transcription

JAKs

Janus kinases

SFKs

Src family kinases

Tyk2

tyrosine kinase 2

AAD

All-Around Docking

MD

molecular dynamics

1. Introduction

JAK family members have similar structures and functions shared among each of the family members (O'Shea et al., 2002; Rane and Reddy, 2000; Yu et al., 2009). Association of cytokine receptors, interferons and growth hormones leads to activation of JAKs, thereby mediating tumor cell proliferation, survival and invasion (O'Shea et al., 2002; Rane and Reddy, 2000; Yu et al., 2009). Constitutively‐activated JAKs phosphorylate critical cellular substrates on tyrosyl residues such as STAT family members, including Stat3, are associated with oncogenic signal transductions (Darnell, 2002; Yu et al., 2009). Constitutively‐activated JAK/STAT signaling has been extensively validated as a new molecular therapeutic target pathway for human cancer treatment (Levine and Gilliland, 2008; Luo and Laaja, 2004; O'Shea et al., 2004; Yu et al., 2009). Recently, the discovery of the Jak2 mutation, which leads to constitutive activation of Jak2, accelerated development of small‐molecule therapeutic agents targeting inhibition of Jak2 kinase activity (Baxter et al., 2005; Hedvat et al., 2009; Kralovics et al., 2005).

Levels of expression and activity of SFKs are elevated in a variety of human cancer cell lines and patient tumors (Kim et al., 2009; Parsons and Parsons, 2004; Yeatman, 2004). SFKs activate oncogenic downstream proteins through phosphorylation of crucial cellular substrates on tyrosyl residues (Parsons and Parsons, 2004; Yu and Jove, 2004). Constitutively‐activated Stat3, one of Src substrates, modulates cell survival, proliferation, angiogenesis and immune evasion of tumor cells (Yu and Jove, 2004; Yu et al., 2009). Many studies demonstrate that Src/Stat3 signaling is an attractive molecular target for human cancer treatment (Kim et al., 2009; Nam et al., 2005; Yu and Jove, 2004). Here, we identify the indirubin derivative (IRD) E738 as a novel dual inhibitor of JAKs and SFKs, which leads to inhibition of Stat3 signaling. These findings suggest a molecular mechanism of action of E738 that has important implications for pancreatic tumor treatment. This is the first dual JAKs and SFKs inhibitor to be reported. Thus, E738 is a promising new therapeutic agent for human pancreatic cancer treatment.

2. Materials and methods

2.1. Cell lines and reagents

Human Panc‐1, MIA‐PaCa2, BxPC3 and AsPC1 pancreatic cancer cells were obtained from the American Type Culture Collection (Manassas, VA). Panc‐1 and MIA‐PaCa2 cells were cultured in DMEM media containing 10% (v/v) fetal bovine serum (FBS). BxPC3 and AsPC1 cells were grown in RPMI 1640 media supplemented with 10% FBS. MEF‐Stat3‐YFP cells were prepared and cultured in DMEM media containing 10% FBS as described previously (Herrmann et al., 2004). Polyclonal antibodies to p‐Jak2 (Tyr1007/1008), Jak2, p‐Src family (Tyr419), Src, p‐Stat3 (Tyr705) and Stat3 were from Cell Signaling Technologies (Cambridge, MA). Polyclonal antibody to Mcl‐1 was from Santa Cruz Biotechnology (Santa Cruz, CA). Polyclonal antibody to survivin was obtained from Alpha Diagnostic International (San Antonio, TX). Monoclonal antibody to β‐Actin was obtained from Sigma (St. Louis, MO). Polyclonal antibody to poly (ADP‐ribose) polymerase (PARP) was purchased from Roche Molecular Biochemicals (Indianapolis, IN). IRD E738 was prepared as described previously (Hoessel et al., 1999; Nam et al., 2005). Structure and purity of E738 were analyzed using ¹³C‐ and ¹H‐NMR spectroscopy and elemental analyses. E738 displayed over 98% purity in NMR analysis.

2.2. Kinase assays in vitro

The kinase assays were performed with recombinant Jak1, Jak2, Tyk2, c‐Src, Lyn and Hck, proteins using the HotSpot protocol (Reaction Biology Corp, Malvern, PA). Briefly, proteins, freshly prepared substrates and ³³P‐ATP (specific activity 0.01 μCi/μl final) were mixed in reaction buffer (20 mM HEPES pH 7.5, 10 mM MgCl₂, 1 mM EGTA, 0.02% Brij35, 0.02 mg/ml BSA, 0.1 mM Na₃VO₄, 2 mM DTT) in the presence of DMSO as control or E738. The mixtures were reacted for 120 min at room temperature. Samples were transferred onto P81 ion exchange paper and filters were extensively washed with 0.75% phosphoric acid. The radioactivities were monitored.

2.3. Western blot analyses

Western blot analyses were performed as described previously with minor modification (Nam et al., 2005). Briefly, human pancreatic cancer cells were treated with E738. Whole‐cell lysates (40 μg) were resolved by SDS‐PAGE and immunoblotted with specific antibodies. Primary phospho‐specific antibodies were incubated in TBS (pH 7.5) with 0.1% (v/v) Tween‐20 and 5% (w/v) BSA with gentle agitation overnight at 4 °C. Other specific antibodies were diluted in PBS (pH 7.5) with 5% (w/v) nonfat milk and 0.1% (v/v) Tween‐20 overnight at 4 °C. Horseradish peroxidase‐conjugated secondary antibodies were incubated in TBS (pH 7.5) with 5% (w/v) nonfat milk and 0.1% (v/v) Tween‐20 at a 1:2000 dilution for 1 h at room temperature. Positive immuno‐reactive proteins were detected using the ECL system (Pierce, Rockford, IL).

2.4. Nuclear translocation of Stat3‐YFP

The experiment was performed as described previously (Herrmann et al., 2004). MEF‐Stat3‐YFP cells were treated with E738 for 4 h and stimulated with OSM [25 ng/ml] (R&D Systems, Minneapolis, MN) for 30 min or left unstimulated. Cells were fixed with 2% formaldehyde, incubated with Hoechst33342 (Invitrogen, Grand Island, NY) and then mounted with Mowiol. Confocal imaging was carried out on an LSM 510 Meta confocal microscope (Zeiss, Thornwood, NY). YFP emissions were detected with BP 535–590 nm Hoechst33342 was visualized with a two‐photon laser exciting at 435–485 nm.

2.5. Viability and apoptosis assays

MTS assays were performed for cell viability as described by the supplier (Promega, Madison, WI). Human Panc‐1, MIA‐PaCa2, BxPC3 and AsPC1 pancreatic cancer cells were seeded in 96‐well plates (5000/well), incubated overnight at 37 °C in 5% (v/v) CO₂ and exposed to E738 in a dose‐dependent manner for 48 h. DMSO was used as the vehicle control. Cell viability was determined by tetrazolium conversion to its formazan dye and absorbance was measured at 490 nm using an automated ELISA plate reader.

Apoptosis assays of human MIA‐PaCa2 cells based on loss of membrane integrity were carried out using Annexin V‐FITC as described by the supplier (BD Biosciences Pharmingen, San Diego, CA). Cells were analyzed using a FACScan flow cytometer to quantify fluorescence. Apoptotic cells were defined as Annexin V‐FITC positive.

2.6. Computational modeling

Initial docking models of E738 or E804 binding to Tyk2 (PDB 3nz0) (Tsui et al., 2011) or Src (PDB 2h8h) (Hennequin et al., 2006) were established using the All‐Around Docking (AAD) method we developed. Briefly, the AAD automatically and systematically determined the best docking site on the whole protein surface and inner cavities by locally implementing Glide (Friesner et al., 2004) or other docking methods. Glide XP docking scores (Friesner et al., 2006) were used to evaluate the docking poses of AAD. After adding 15 layers of explicit water molecules, 0.5 M sodium chloride, minimization, pre‐heating and equilibration of the initial models, molecular dynamics simulations were performed utilizing NAMD (Phillips et al., 2005). It took about 2 weeks for a 30 ns MD simulation running on 32 CPUs of Dell HPC clusters. For the trajectory analysis, NAMD was used for pair interaction calculation and VMD (Humphrey et al., 1996) was used for hydrogen bond analysis. The water‐bridge analysis was carried out based on our in‐house developing methodology, which analyzes fixed water molecules with low energies during MD simulations. All pictures were generated by PyMOL (Schrödinger LLC 2010).

2.7. Statistical analysis

The statistical significance of group differences was analyzed using a Student's t‐test with the two‐tailed distribution. P values less than 0.01 were considered statistically significant.

3. Results

3.1. E738 inhibits SFKs and JAKs activities in vitro

Previously, we reported that IRD E804 directly inhibits Src kinase activity in vitro, associated with inhibition of downstream Stat3 signaling (Nam et al., 2005). Most IRDs are poorly water‐soluble and display limited bioavailability (Cheng et al., 2010). E804 carrying a dihydroxybuyl 3′‐oxime ether substituent (Nam et al., 2005), shows improved, yet still has low water‐solubility and bioavailability (2 mg/mL). Nevertheless, it demonstrates effective biological and cellular activities in cells (Nam et al., 2005). E738, which contains dihydroxypropyl 3′‐oxime ether and 5‐methoxy substituents (Figure 1A), surprisingly exhibited 25‐fold increased water‐solubility (50 mg/mL) compared to E804 (Marko et al., 2001). IRDs are kinase inhibitors with high affinity to the ATP‐binding pocket of kinases (Hoessel et al., 1999). Moreover, IRDs block Stat3 signaling in human cancer cells (Nam et al., 2005).

Figure 1.

A. Structure of E738. B. E738 inhibits activities of JAKs in vitro. C. E738 inhibits activities of SFKs in vitro. The kinase assays were performed as described in Materials and methods. Recombinant proteins of Jak1, Jak2, Tyk2, c‐Src, Lyn and Hck, freshly prepared substrates and 33P‐ATP were mixed in reaction buffer in the presence of DMSO as control or E738. The mixtures were reacted for 120 min at room temperature. The radioactivities were monitored.

To determine whether E738 inhibits JAKs activities of Stat3 upstream, an in vitro kinase assay was performed with recombinant human JAKs proteins. As shown in Figure 1B, E738 reduced Jak1, Jak2 and Tyk2 activities in vitro with IC₅₀ values of 10.4 nM, 74.1 nM and 0.7 nM, respectively. Since the structure of E738 is closely related to E804, which is a potent inhibitor of Src kinase activity in vitro (IC₅₀ = 0.43 μM) (Nam et al., 2005), we tested whether E738 likewise reduces Src kinase activity in vitro (Figure 1C). In an in vitro kinase assay with recombinant human SFK proteins, E738 produced strong inhibitory activities against Src, Lyn and Hck with IC₅₀ values of 10.7 nM, 29.8 nM and 263.9 nM, respectively (Figure 1C). Furthermore, in a recombinant human kinase profiling assay, E738 showed potent inhibitory activity against a broad spectrum of serine/threonine and tyrosine kinases with strong effects against Aurora A, CDK2/cyclin A, GSK3β and VEGFR2 (Supplementary Figure 1). Together, these results demonstrate that E738 not only inhibits a broad spectrum of kinases aberrantly expressed in human cancer, but also behaves as a dual inhibitor of JAKs and SFKs.

3.2. Effects of E738 on pancreatic cancer cells

Pancreatic cancer is the fourth leading cause of cancer death among both men and women in most countries and approximately 227,000 deaths/annum are estimated worldwide (Raimondi et al., 2009; Vincent et al., 2011). Five year survival rate after pancreatic cancer diagnosis is less than 5% due to locally‐occurring diseases and broad metastasis (Vincent et al., 2011). Anti‐cancer drugs are poorly effective to pancreatic tumors (Vincent et al., 2011). Novel small molecule agents with high inhibitory activity against signal transduction pathways relevant to pancreatic cancer may carry promise for clinical therapy. To examine whether E738 displays antitumor activity against human pancreatic cancer cells, Panc‐1, MIA‐PaCa2, BxPC3 and AsPC1 cell lines were selected for effects of E738 on cell viability using an MTS assay. Panc‐1, MIA‐PaCa2 and BxPC3 cell lines were isolated from primary tumors, whereas AsPC1 was derived from metastatic ascites (Deer et al., 2010; Nagaraj et al., 2010). Most pancreatic tumors have mutant K‐ras genes, followed by mutations of tumor suppressors such as p53 and Smad4 genes (Almoguera et al., 1988; Rozenblum et al., 1997). Panc‐1, MIA‐PaCa2 and AsPC cell lines include K‐ras mutations and wild‐type Smad4 genes (Deer et al., 2010; Nagaraj et al., 2010).

E738 substantially reduced cell viabilities of these cells with IC₅₀ values = 0.76 μM, 0.68 μM, 2.2 μM and 1.3 μM against Panc‐1, MIA‐PaCa2, BxPC3 and AsPC1 cells, respectively (Figure 2A). Thus, K‐ras mutations appear to render pancreatic tumor cells more sensitive to E738. In addition, Panc‐1 and MIA‐PaCa2 cells have mutated p53 genes, whereas AsPC1 cells include wild‐type p53 ones (Deer et al., 2010; Nagaraj et al., 2010). p53 mutated pancreatic tumor cells displayed slightly enhanced susceptibility to loss of cell viability mediated by E738 (Figure 2A).

Figure 2.

Effects of E738 on human pancreatic cancer cells. A. MTS assays were conducted for cell viability as described in Materials and methods. Panc‐1, MIA‐PaCa2, BxPC3 and AsPC1 pancreatic cancer cells were exposed to E738 in a dose‐dependent manner for 48 h. DMSO was used as the vehicle control. Each experiment was performed in quadruplicate. Data are mean ± SD. B. Panc‐1, MIA‐PaCa2, BxPC3 and AsPC1 cells were treated with 5 μM of E738 for 4 h. Whole‐cell lysates were resolved by SDS‐PAGE and immunoblotted with specific antibodies to p‐Jak2 (Tyr1007/1008), Jak2, p‐Src (Tyr419), Src, p‐Stat3 (Tyr705) and Stat3.

3.3. Effects of E738 on Jak2/Stat3 or Src/Stat3 signaling

E738 exhibited potent inhibitory activities against JAKs and SFKs in vitro (Figure 1). To assess whether E738 blocks tyrosyl phosphorylation of Jak2, Src and Stat3 in Panc‐1, MIA‐PaCa2, BxPC3 and AsPC1 cell lines, Western blot analysis was carried out with specific antibodies to p‐Jak2 (Tyr1007/1008), p‐Src (Tyr419) and p‐Stat3 (Tyr705). E738 treatment resulted in strong reduction of tyrosyl p‐levels of Jak2, Src and Stat3 in cell lines tested, indicating E738 impairs JAK/Stat3 or Src/Stat3 signaling (Figure 2B). These biological and cellular effects indicate the promise of E738 as potential therapeutic agent for human aggressive malignant pancreatic tumor treatment with dual inhibitory activities of JAKs and SFKs.

JAKs or SFKs are involved in malignant tumor progress in various cancer types, activating crucial cellular substrates, including Stat3 (Darnell, 2002; Kim et al., 2009; Yu and Jove, 2004; Yu et al., 2009). JAK/Stat3 or Src/Stat3 is emerging as a promising molecular target for cancer therapy (Hedvat et al., 2009; Kim et al., 2009; Nam et al., 2005; O'Shea et al., 2004; Yu and Jove, 2004; Yu et al., 2009). To further examine the molecular mechanism mediating biological effects of E738, Panc‐1 and MIA‐PaCa2 cells, which include constitutively‐activated JAK/Stat3 or Src/Stat3 signaling, were treated with E738 in a dose‐dependent manner for 4 h. Western blot analysis revealed that the levels of p‐Jak2, p‐Src and p‐Stat3 were substantially reduced in the range of 0.5 μM–1 μM, whereas total protein levels of Jak2, Src and Stat3 were not changed (Figure 3A). This reduction of tyrosyl phosphorylation is correlated well with loss of cell viability. Next, to further investigate whether E738 inhibits Stat3 nuclear translocation, murine embryonic fibroblast (MEF)‐Stat3‐YFP were treated with E738 and stimulated with oncostatin M (OSM). As expected, consistent with inhibition of tyrosyl phosphorylation of Stat3, Stat3 translocation to nucleus was abrogated as shown in confocal imaging (Figure 3B). These results indicate that E738 inhibits phosphorylation of Jak2 and Src, followed by blockade of downstream Stat3 signaling.

Figure 3.

Effects of E738 on Jak2/Stat3 or Src/Stat3 signaling. A. Panc‐1 and MIA‐PaCa2 cells were treated with E738 in a dose‐dependent manner for 4 h. Western blot analysis was carried out as described in Materials and methods. Whole‐cell lysates were resolved by SDS‐PAGE and immunoblotted with specific antibodies to p‐Jak2 (Tyr1007/1008), Jak2, p‐Src (Tyr419), Src, p‐Stat3 and Stat3. B. Effects of E738 on Stat3 nuclear translocation. MEF‐Stat3‐YFP cells were treated with E738 in a dose‐dependent manner for 4 h and stimulated with OSM for 30 min or left unstimulated. Confocal imaging was carried out on an LSM 510 Meta confocal microscope. Blue and yellow colors mark the nucleus and Stat3 localization, respectively. Red arrows indicate inhibition of Stat3 nuclear translocation on cells.

3.4. E738 down‐regulates anti‐apoptotic proteins Mcl‐1 and survivin, associated with induction of apoptosis

STAT signaling contributes to survival of malignant cancer cells. Activated Stat3 induces expression of anti‐apoptotic proteins such as Mcl‐1, Bcl‐x_L and survivin (Yu and Jove, 2004; Yu et al., 2009). In contrast, inhibition of Stat3 signaling down‐regulates expression of these Stat3 downstream anti‐apoptotic proteins, followed by induction of apoptosis (Yu and Jove, 2004; Yu et al., 2009). To determine whether E738 down‐regulates cell survival proteins Mcl‐1 and survivin, Western blot analysis was conducted using specific antibodies to Mcl‐1 and survivin proteins. As shown in Figure 4A, expression of anti‐apoptotic proteins Mcl‐1 and survivin was decreased by E738. This observation correlates well with inhibition of Jak2/Stat3 or Src/Stat3 signaling.

Figure 4.

E738 down‐regulates anti‐apoptotic proteins Mcl‐1 and survivin, associated with induction of apoptosis. A. Panc‐1 and MIA‐PaCa2 cells were treated with E738 in a dose‐dependent manner for 24 h. Whole‐cell lysates were resolved by SDS‐PAGE and immunoblotted with specific antibodies to Mcl‐1, survivin. and Actin. B. For detection of PARP cleavage as an apoptosis indicator, Panc‐1 and MIA‐PaCa2 cells were treated with E738 in a dose‐dependent manner for 24 h. Whole‐cell lysates were resolved by SDS‐PAGE and immunoblotted with specific antibodies to poly (ADP‐ribose) polymerase (PARP) and Actin. C. E738 induces apoptosis. MIA‐PaCa2 cells were treated with E738 in a dose‐dependent manner for 48 h. Cells were labeled with Annexin V‐PE. Cells were analyzed using a FACScan flow cytometer to quantify fluorescence. Apoptotic cells were defined as Annexin V‐FITC positive. Each experiment was performed in quadruplicate. Data are mean ± SD. *P < 0.01; **P < 0.001.

Next, to assess whether the effects of E738 on down‐regulation of Mcl‐1 and survivin proteins are associated with induction of apoptosis on pancreatic cancer cells, Western blot analysis using whole cell lysates from cells treated with E738 for 24 h was carried out to detect PARP cleavage as an apoptosis indicator. Indeed, the PARP cleavage was revealed in a dose‐dependent manner 24 h after treatment (Figure 4B). These results are consistent with down‐regulation of anti‐apoptotic proteins Mcl‐1 and survivin. To further investigate the biological effect of E738 on induction of apoptosis, a flow cytometry analysis was performed using Annexin V as an early marker of apoptosis. Consistent with reduction of cell viability as displayed in Figure 2A, 0.5 μM of E738 significantly induced apoptosis on these cancer cells (Figure 4C). Thus, inhibition of constitutively‐activated JAK/Stat3 or Src/Stat3 signaling, associated with down‐regulation of anti‐apoptotic proteins such as Mcl‐1 and survivin, suggests a pharmacological mechanism of action of E738 by induction of apoptosis on pancreatic carcinoma cells.

3.5. Computational modeling of E738 or E804 binding to Tyk2 or Src

Models of E738 or E804 binding to Tyk2 or Src in silico were established using the All‐Around Docking (AAD) method we developed by utilizing Glide (Friesner et al., 2006). The AAD method determined the best binding pocket of E738 to Src though moving E738 all‐around on the protein surface (Figure 5A), thereby discovering the site that has the highest docking score. It was designed to determine the best binding site and docking pose of a ligand automatically and systematically. We observed that E738 binds to the ATP‐binding pockets of both Tyk2 (PDB 3nz0) (Tsui et al., 2011) and Src (PDB 2h8h) (Hennequin et al., 2006) with the DFG‐in conformations (Liu and Gray, 2006) (Figure 5B and C). The high Glide XP docking scores (Friesner et al., 2006) of E738 binding to Tyk2 (−12.9 kcal/mol) and to Src (−11.1 kcal/mol) implicate E738 as a potent dual inhibitor of Tyk2 and Src. The dual inhibition of E738 could be mainly caused by several hydrogen bonds formed in the complex of E738/Tyk2 or E738/Src. In addition, the initial docking models were further examined by molecular dynamics (MD) simulations. Thirty ns MD simulations were pursued with the complex of E738/Src or E738/Tyk2 using NAMD (Phillips et al., 2005) (Figure 5B and C).

Figure 5.

Computational modeling of E738/Src or E738/Tyk2 complex. A. AAD method determines the best binding site of E738 to Src. E738 moves on all‐around Src protein surface. The best docking site was determined as the ATP‐binding site, which is circled by red dashed lines. B. Model of E738 binding to the ATP‐binding site of Src. Residues D404 and M341, which are painted with pink and blue colors on the Src surface, respectively, have the highest pair‐interaction energy with E738 during the 30 ns MD simulation. C. Model of E738 binding to the ATP‐binding site of Tyk2. Residue D1041 (pink color) has the highest pair‐interaction energy with E738. D. Hydrogen bonds formed in the complex of E738 and Tyk2. Four hydrogen bonds were formed with residues E905, E979, V981 and R1027 in one snapshot of 15 ns MD simulation. Hydrogen bonds were determined by hydrogen donor‐acceptor distance ≤3.2 Å and donor‐hydrogen‐acceptor angle ≤125°. A water‐bridge between H907 and E738 was also observed during 80% of the MD simulation time. The water‐bridge was determined by the sum of distances between two polar atoms and a water molecule ≤7 Å.

The model of E804 binding to Src was also established as shown in Figure 6A. The Glide XP docking score was −10.86 kcal/mol, which is lower than that of E738 (−11.13 kcal/mol). Interestingly, the dihydroxybutyl 3′‐oxime ether of E804 (Figure 6A) adopted a different orientation from the dihydroxypropyl 3′‐oxime ether of E738 (Figure 6B), while the indirubin substructure retained the same pose.

Figure 6.

Model of E738/Src or E804/Src complex. A. Model of E804 binding to the ATP‐binding site of Src. E804 formed hydrogen bonds with residues MET341, ALA390 and SER345. B. Model of E738 binding to the ATP‐binding site of Src. E738 formed hydrogen bonds with residues MET341, ALA390 and ASP404.

4. Discussion

Inhibition of persistent STAT signaling reduces tumor cell survival in cell culture and in vivo studies (Darnell, 2002; Yu et al., 2009). The identification of direct Stat3‐targeting modulators might be a promising approach for development of new cancer therapies (Yu and Jove, 2004; Yu et al., 2009). Attractive modulators comprise small interfering RNAs, dominant–negative protein expression vectors and decoy oligonucleotides of Stat3 (Yu et al., 2009). However, these therapeutic approaches show low efficiencies in vivo due to their delivery challenges. Therefore, direct or indirect small‐molecule inhibitors of Stat3 signaling, including upstream tyrosine kinases such as JAKs and Src, might be more practical for clinical development (Yu and Jove, 2004; Yu et al., 2009). In this respect, the natural herbal product derivative E738 is a promising therapeutic agent, inhibiting tyrosyl phosphorylation of Stat3 through inactivation of upstream JAKs and SFKs in pancreatic tumor cells.

Recently, it was reported that prolonged inhibition of Src causes reactivation of Stat3 and tumor cell survival through altered JAK/Stat3 interaction (Johnson et al., 2007; Sen et al., 2009). Previous studies demonstrated that JAKs and SFKs cooperate to constitutively activate Stat3 signaling (Garcia et al., 2001; Nam et al., 2005; Zhang et al., 2000). However, identification of E738 as a novel dual inhibitor of JAKs and SFKs suggests pharmacological benefits of E738, which can overcome this compensatory drawback through blockade of both JAKs and SFKs activities.

The high Glide XP docking scores (Friesner et al., 2006) of E738 binding to Tyk2 and Src were −12.9 kcal/mol and −11.1 kcal/mol, respectively (Figure 5B and C). E738 prefers binding to Tyk2 with −1.8 kcal/mol higher binding affinity compared to Src. This observation is consistent with the 15‐fold increase in the experimental IC₅₀ values of E738 against Tyk2 and Src in vitro (Figure 1b and C). The ATP‐binding site of Src is more open than that of Tyk2 (Figure 5B and C). E738 is basically a planar molecule that fits the narrow ATP‐binding pocket of Tyk2. Thus, interaction of the E738 molecule with Src is expected to be weaker than with Tyk2. In addition, the MD simulation study supports that E738 is a dual inhibitor of Tyk2 and Src, since it can form several hydrogen bonds associated with high pair‐interaction energies to both proteins in the simulations.

The MD simulation study suggests some important points. The average ligand–protein pair interaction between E738 and Tyk2 was −85.8 kcal/mol. This interaction energy includes the electrostatic and Van der Waals pair interaction energy, but not the ligand/solvent interaction. The electrostatic interaction energy was −35.5 kcal/mol. On the other hand, the interaction energy between E738 and Src was −76.7 kcal/mol. The electrostatic interaction energy was −39.1 kcal/mol. However, the binding free energy of E738 to Tyk2 could be lower than that to Src. These results provide evidence that the inhibitory activity of E738 against Tyk2 in vitro is more potent than against Src.

Amino acid residue ASP1041 of Tyk2 (Figure 5C) demonstrated about −8.2 kcal/mol pair interaction energy, which is ranked as the highest pair interaction between E738 and Tyk2, whereas the highest energy of the corresponding residue ASP404 of Src (Figure 5B) was −16.0 kcal/mol. This observation suggests the residue in the DFG‐motif plays a critical role to enhance E738 binding to Tyk2 and Src. Other important residues, which contribute to binding of Tyk2, are VAL981 (−7.9 kcal/mol), LEU903 (−6.8 kcal/mol), TYR980 (−6.3 kcal/mol) and LEU1030 (−5.1 kcal/mol). Substantial residues of Src involved in drug binding are MET341 (−9.6 kcal/mol), TYR340 (−6.7 kcal/mol) and LEU273 (−5.6 kcal/mol). The pair interaction energy also confirms the hydrogen bonds formed between E738 and proteins. For example, of the −7.9 kcal/mol interaction energy of VAL981 in the complex of E738/Tyk2, the electrostatic energy was −5.3 kcal/mol, while Van der Waals interaction was −2.6 kcal/mol. This result implicates that the hydrogen bond between VAL981 to E738 might be very strong. On the other hand, for the −6.8 kcal/mol interaction energy from LEU903 to E738, Van der Waals interaction was −5.3 kcal/mol. Taken together, the interaction between LEU903 and E738 could be mainly hydrophobic interactions.

In the MD simulation, the average number of hydrogen bonds formed in the complex of E738/Tyk2 is 2.9, while the complex of E738/Src has 3.0. As shown in Figure 5D, four hydrogen bonds in the complex of E738/Tyk2 were determined in one snapshot of the 15 ns MD simulation. This simulation result suggests E738 can bind to both Tyk2 and Src effectively. Substantial hydrogen bonds in the complex of E738 and Tyk2 are formed at residues VAL981_N (backbone N‐atom) at 97.8% of simulation time, GLU979_O (88.9%) and GLU905_O (51.1%). On the other hand, critical hydrogen bonds in the complex E738 and Src are formed at residues MET341_N (97.2%), ASP404_OD1 (77.1%) and ALA390_O (76.1%). In comparison of average hydrogen bond numbers of the two complexes, E738/Tyk2 has slightly less hydrogen bonds than Src; however, even though E738 and ASP1041 of Tyk2 show strong electrostatic interaction, this interaction was not counted as a hydrogen bond since the average distance of ASP1041_OD1 to E738 is 3.8Å. The observed electrostatic interaction might raise the inhibitory activity of E738 against Tyk2 in vitro through the increased binding affinity of E738 to Tyk2. Besides direct hydrogen bonds, a water‐bridge was also observed between HIS907 and E738 that might enhance the ligand binding in Tyk2, thereby increasing the potency of E738 against Tyk2.

With respect to in vitro inhibition of Src kinase activity, E738 displayed over 40‐fold increase in inhibitory activity compared to E804 (Figure 1C) (Nam et al., 2005). The modeling study of E738/Src or E804/Src complex suggests another possible explanation for the enhancement of inhibitory activities. The dihydroxypropyl 3′‐oxime ether substituent of E738 elevates the binding to Src in two aspects. This 3′‐oxime ether substituent interacts with nearby residue LEU273 with Van der Waals interactions. The substituent blocks the 5‐methoxy group of E738 to adopt the conformation of E804 (Figure 6B). Consequently, it hinders free rotation of the 5‐methoxy group, thereby lowering the entropy of E738. Next, the 3′‐oxime ether substituent of E738 can gain hydrophobic interaction with nearby residues, so the binding free energy of E738 to Src is more favorable.

In summary, we report that the natural product derivative E738 is the first dual inhibitor of JAKs and SFKs, which in turn inhibits Jak2/Stat3 and Src/Stat3 signaling, associated with induction of apoptosis in malignant pancreatic cancer cells. Computational modeling indicates that E738 strongly binds to the ATP‐binding site of Tyk2 or Src as an ATP‐competitive inhibitor. In addition, these findings suggest a pharmacological mechanism of action of E738 in cancer cells that may have important implications for pancreatic tumor treatment. Thus, E738 has potential for development of new therapeutic agents for human pancreatic cancer treatment.

Funding

This study was supported by NIH grant R01 CA115674‐05 to RJ.

Supporting information

The following is the supplementary data related to this article:

Supplementary caption

Supplementary data 1.

1.1.

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.molonc.2012.10.013.

Nam Sangkil, Wen Wei, Schroeder Anne, Herrmann Andreas, Yu Hua, Cheng Xinlai, Merz Karl-Heinz, Eisenbrand Gerhard, Li Hongzhi, Yuan Yate-Ching, Jove Richard, (2013), Dual inhibition of Janus and Src family kinases by novel indirubin derivative blocks constitutively‐activated Stat3 signaling associated with apoptosis of human pancreatic cancer cells, Molecular Oncology, 7, doi: 10.1016/j.molonc.2012.10.013.