Department of Medical Research, National Taiwan University Hospital, Taipei, Taiwan
National Center of Excellence for Clinical Trial and Research, National Taiwan University Hospital, Taipei, Taiwan
Address reprint requests to: Kuen-Feng Chen, M.D., Ph.D., Department of Medical Research, National Taiwan University Hospital, No. 7, Chung-Shan S. Rd., Taipei 10055, Taiwan. E-mail: firstname.lastname@example.org; Tel: +886-2-23123456, ext: 63548; fax: +886-2-23225329.
Errata: Correction Volume 65, Issue 6, 2135, Article first published online: 28 April 2017
Potential conflict of interest: Nothing to report.
Sorafenib is the first approved targeted therapeutic reagent for hepatocellular carcinoma (HCC). Here, we report that Src homology region 2 (SH2) domain-containing phosphatase 1 (SHP-1) is a major target of sorafenib and generates a series of sorafenib derivatives to search for potent SHP-1 agonists that may act as better anti-HCC agents than sorafenib. Sorafenib increases SHP-1 activity by direct interaction and impairs the association between the N-SH2 domain and the catalytic protein tyrosine phosphatase domain of SHP-1. Deletion of the N-SH2 domain (dN1) or point mutation (D61A) of SHP-1 abolished the effect of sorafenib on SHP-1, phosphorylated signal transducer and activator of transcription 3 (p-STAT3), and apoptosis, suggesting that sorafenib may affect SHP-1 by triggering a conformational switch relieving its autoinhibition. Molecular docking of SHP-1/sorafenib complex confirmed our findings in HCC cells. Furthermore, novel sorafenib derivatives SC-43 and SC-40 displayed more potent anti-HCC activity than sorafenib, as measured by enhanced SHP-1 activity, inhibition of p-STAT3, and induction of apoptosis. SC-43 induced substantial apoptosis in sorafenib-resistant cells and showed better survival benefits than sorafenib in orthotopic HCC tumors. Conclusion: In this study, we identified SHP-1 as a major target of sorafenib. SC-43 and SC-40, potent SHP-1 agonists, showed better anti-HCC effects than sorafenib in vitro and in vivo. Further clinical investigation is warranted. (Hepatology 2014;58:190–201)
signal transducer and activator of transcription 3
vascular endothelial growth factor receptor
vascular endothelial growth factor receptor
Hepatocellular carcinoma (HCC) is currently the fifth most common solid tumor worldwide. Sorafenib (Nexavar), a tyrosine kinase inhibitor against the vascular endothelial growth factor receptor (VEGFR) family, has been approved in HCC[2, 3] since 2007. Since then, several large phase III studies of other VEGFR inhibitors, such as sunitinib and brivanib, have been conducted. However, the results from these studies were unsatisfactory.[4, 5] None of these kinase inhibitors more potent than sorafenib is effective in HCC. These clinical data suggest that the effect of sorafenib on patients with HCC might be more than the inhibition of VEGFR or kinases.
Previously, we have reported that signal transducer and activator of transcription 3 (STAT3) is a kinase-independent target of sorafenib in HCC.[6-8] Src homology region 2 (SH2) domain-containing phosphatase 1 (SHP-1), a negative regulator of phosphorylated STAT3 (p-STAT3), is involved in many hematopoietic signaling processes, and its role in solid tumors is still not very clear. SHP-1 belongs to a family of nonreceptor protein tyrosine phosphatases (PTPs) and consists of two SH2 domains that bind phosphotyrosine, a catalytic PTP domain, and a C-terminal tail. Although many reports have investigated SHP-1 in hematopoietic cells, comparatively few reports have looked at the biological importance of SHP-1 in solid tumors, even though early studies have shown that SHP-1 is a potential tumor suppressor modulated in cancer progression. The phosphatase activity of SHP-1 is highly dependent on its structural variability, as evidenced by its open- or closed-form chemical structure. The N-SH2 domain protrudes into the catalytic domain to directly block the entrance into the active site, and the highly mobile C-SH2 domain is thought to function as an antenna to search for the phosphopeptide activator.[10-12] In addition, the activity of the catalytic domain is determined by the flexibility of the WPD loop, which contains the active-site residue, Asp421.[10, 11, 13]
Here, we explored the molecular mechanism by which sorafenib increases SHP-1 activity. Then, through generating new sorafenib derivatives designed based on the premise that the effect of sorafenib is through increasing SHP-1 activity by a conformational switch that relieves its autoinhibition, we identified new drugs that show better anti-HCC effects than sorafenib. Targeting SHP-1/STAT3 might represent a promising strategy for treatment of HCC.
Materials and Methods
Sorafenib (Nexavar) was kindly provided by Bayer Pharmaceuticals (Pittsburgh, PA). For cell-based studies, sorafenib at various concentrations was dissolved in dimethyl sulfoxide and then added to the cells in fetal bovine serum-free Dulbecco's modified Eagle's medium. SHP-1 inhibitor (PTP III) was purchased from Calbiochem (San Diego, CA).
SHP-1 Phosphatase Activity
After treatment of sorafenib or SC derivatives, PLC5 protein extract were incubated with anti-SHP-1 antibody (Ab) in immunoprecipitation (IP) buffer (20 mM of Tris-HCl [pH 7.5], 150 mM of NaCl, 1 mM of ethylenediaminetetraacetic acid, 1% NP-40, and 1% sodium deoxycholate) overnight. Protein G-Sepharose 4 Fast flow (GE Healthcare Bio-Science, Township, NJ) was added to each sample, followed by incubation for 3 hours at 4°C with rotation. A RediPlate 96 EnzChek Tyrosine Phosphatase Assay Kit (R-22067) was used for SHP-1 activity assay (Molecular Probes, Invitrogen, Carlsbad, CA).
Xenograft Tumor Growth
For the subcutaneous (SC) model (n = 10), each mouse was inoculated SC in the dorsal flank with 1 × 106 PLC5 cells suspended in 0.1 mL of serum-free medium containing 50% Matrigel (BD Biosciences, Bedford, MA). When tumors reached 100-200 mm3, mice received sorafenib, SC-43, or SC-40 (10 mg/kg per oral, once-daily). Tumors were measured twice-weekly using calipers, and their volumes were calculated using the following standard formula: width × length × height × 0.523. For the orthotopic model (n = 6), mice were inoculated into the liver directly with luc2-expressed PLC5 cells. The treatment initiates when the luciferase activity of mice can be monitored. Mice were randomized into vehicle, sorafenib (10 mg/kg/day), and SC-43 (10 mg/kg/day). The survival curve was determined by the endpoint of treatment.
Other extensive methods were moved to a Supporting Information section.
Comparisons of mean values were performed using the independent samples t test in SPSS for Windows 11.5 software (SPSS, Inc., Chicago, IL).
Sorafenib Increases the Phosphatase Activity of SHP-1 by Direct Interaction
Sorafenib significantly enhanced the phosphatase activity of SHP-1 in a dose-dependent manner in all tested HCC cell lines (Fig. 1A). Sorafenib activated SHP-1 in SHP-1-containing IP extract at very low concentrations (nM), whereas the activity was not affected in SHP-1 catalytic dead mutant (C453S)-expressing cell extract (Fig. 1B). Incubation of sorafenib with cell-free SHP-1 proteins increased SHP-1 activity significantly at low concentrations (Fig. 1C), suggesting that sorafenib activates SHP-1 through direct interaction with SHP-1 proteins. Notably, sorafenib did not alter interactions between SHP-1 and STAT3 (Fig. 1D), although sorafenib down-regulated p-STAT3-related proteins in HCC cell lines in a dose-dependent manner (Fig. 1E).
Sorafenib-treatment in PLC5 with high levels of SHP-1 showed more inhibition of p-STAT3 and induced more apoptosis (Fig. 1F). Otherwise, sorafenib did not alter the activity of SHP-2 significantly either in HCC cell lines or purified SHP-2 proteins (Supporting Fig. 1). These data suggest that SHP-2 is not a major target of sorafenib.
Sorafenib Relieves Autoinhibition of SHP-1 by Interfering With the Inhibitory N-SH2 Domain
Next, we generated a series of domain-deletion mutants of SHP-1 and further assayed their phosphatase activity and susceptibility to dephosphorylation of STAT3 (Fig. 2A). Notably, the intramolecular inhibition of SHP-1 is protected by various biochemical associations between N1 and the PTP catalytic domain, such as Asp61 and Lys362 (salt bridge). The dN1 or D61A mutants demonstrated significantly increased SHP-1 activity, indicating that these two mutants mimic the open conformation and serve as constitutive activators (Fig. 2B). In contrast, deletion of the C-SH2 domain of SHP-1 (dN2) did not result in significant induction of SHP-1 activity. Even if dN2 slightly increased phosphatase activity in SK-Hep1 cells, it may be explained by its flexible orientation and unknown mechanism for searching of phosphotyrosine activators.
Accordingly, sorafenib-induced SHP-1 activity was significantly inhibited in recombinant dN1 and D61A mutants (Fig. 2C). These results suggest that sorafenib may bind to the N-terminal SH2 domain directly. Notably, mutation from Asp to Ala at residue 61 of SHP-1 protein significantly inhibited the effect of sorafenib on SHP-1, indicating that D61 of the inhibitory N-SH2 domain is crucial for up-regulation of SHP-1 activity by sorafenib. Sorafenib-induced down-regulation of p-STAT3 was found in PLC5 cells expressing vector, wild-type (WT), or dN2 mutants of SHP-1. But, ectopic expression of dN1 and D61A restored the expression of p-STAT3 (Fig. 2D). Consequently, dose-escalation studies of transfection of dN1 and D61A further supported this molecular event (Fig. 2E). Sorafenib treatment did not show significant changes in cells with the catalytic dead mutant (C453S). STAT3-related transcriptional activity was restricted in vector, wtSHP-1, and dN2-expressed cells, but not in dN1 or D61A mutants (Fig. 2F, left). Furthermore, sorafenib still increased SHP-1 activity in cells expressing wtSHP-1 or dN2, but could not increase activity significantly in dN1- or D61A-expressing SHP-1 mutants (Fig. 2F, middle). Sorafenib induced significantly less apoptosis in cells expressing dN1 and D61 mutants than in vector-transfected cells (Fig. 2F, right). Together, our data suggest that sorafenib may affect SHP-1 by switching the confirmation from autoinhibitory (closed) to active (open).
Sorafenib Directly Impairs the Interaction Between N1 and the Catalytic PTP Domain
PLC5 cells expressed either hemagglutinin antigen (HA)-tagged N1 or N2, in combination with Myc-tagged PTP, were assessed for stability of the N/C interaction after sorafenib treatment. Sorafenib abolished the interaction between N1 and the PTP domain directly, and the C-terminal SH2 domain (N2) could not interact with PTP, serving as a negative control for N/C interaction (Fig. 3A). The interaction-based results verify the role of sorafenib in regulating the conformational changes to elevate SHP-1 activity. Moreover, ectopic expression of the N1 domain strongly inhibited endogenous phosphatase activity of SHP-1 (Fig. 3B). In contrast, N2 did not affect endogenous SHP-1 activity. Sorafenib could further release the N1-induced inhibition of SHP-1 activity significantly up to 5-fold, in comparison with nontreated cells (Fig. 3C). The expression level of p-STAT3 was up-regulated in N1-expressing cells, but was inhibited again after sorafenib treatment. We confirmed that sorafenib could reactivate N1-induced SHP-1 activity inhibition in a dose-dependent manner (Fig. 3D). Together, these results confirmed that the N-terminal SH2 domain is a critical docking site of sorafenib.
We further assessed the role of SHP-1 in HCC formation. Ectopic expression of dN1 significantly reduced colony numbers, compared to vector control or wtSHP-1 (Fig. 3E). There was a 33% reduction in colony formation in PLC5 cells ectopically expressing dN1, in comparison with wtSHP-1. We also observed an almost 50% reduction in colony numbers in SK-Hep1 cells, ectopically expressing dN1. Ectopic expression of D61A also exhibited fewer colonies than the control. These results imply that activated SHP-1 protects against tumor cell proliferation. Next, an immunohistochemistry (IHC) study was conducted to examine the role of SHP-1 in tissues from patients with HCC. p-STAT3 was expressed in the majority of HCC tissue, but less SHP-1 was found expressed in the same tissues (Fig. 3F). Further investigation of the role of SHP-1 in HCC tumor progression is warranted.
Generation of Potent Novel Sorafenib Derivatives on the Basis That SHP-1 Is a Target of Sorafenib
Working upon the assumption that sorafenib relieves the autoinhibition of SHP-1, we generated a series of sorafenib derivatives to search for potent SHP-1 agonists that may act as better anti-HCC agents than sorafenib. Among the sorafenib analogs generated, we identified two promising new agents, SC-43 and SC-40, the structures of which are shown in Fig. 4A. Both SC-43 and SC-40 had potent effects on induction of SHP-1 activity in vitro and in vivo. SC-43 and SC-40 effectively up-regulated SHP-1 activity at lower concentrations than sorafenib, either in SHP-1-containing cell extract (Fig. 4B) or purified recombinant SHP-1 proteins (Fig. 4C). In addition, both SC-43 and SC-40 did not significantly alert SHP-2 activity in PLC5 and Hep3B cells. Furthermore, SHP-2 activity was not affected in SC-43- or SC-40-treated recombinant SHP-2 proteins (Supporting Fig. 2).
STAT3-related proteins Mcl-1, cyclin D1, and survivin were examined in SC-43- and SC-40-treated HCC cells (Fig. 4D,E). Both SC derivatives resulted in substantial apoptosis in HCC cells, as evidenced by sub-G1.
SC-43 and SC-40 Show a Significant Anti-HCC Effect and Overcome the Resistance of Sorafenib
SC-43 and SC-40 decreased the viability of HCC cells in a dose-dependent manner (Fig. 5A). Both SC-43 and SC-40 showed lower 50% inhibitory concentration, compared to sorafenib. In addition, SC-43 and SC-40 showed more potent inhibition of the p-STAT3-related signaling pathway (Fig. 5B). SC-43 revealed submicromolar inactivation of p-STAT3, relative to sorafenib (Fig. 5C). Furthermore, SC-43 and SC-40 resulted in significant apoptosis in sorafenib-resistant cells at submicromolar concentrations (Fig. 5D). The endogenous induction of p-STAT3 was observed in sorafenib-resistant cells, but not in parental Huh7 cells, which may explain why these cells showed resistance to sorafenib.
Molecular Models of the SHP-1/Sorafenib Complex
Our findings provide a molecular rationale for drug optimization on the basis of the crystal structure of SHP-1. We hypothesize that sorafenib binds to the N-SH2 domain and subsequently releases and activates the PTP domain (Fig. 5E). Sorafenib was docked into the pocket between the N-SH2 domain and formed a hydrogen bonding with R44 through the trifluoromethyl group. The interaction of sorafenib and the N-SH2 domain might lead to a release of the D61 catalytic site and activation of SHP-1 activity. SC-43 act as a potent SHP-1 enhancer and was also docked in the same site. The trifluoromethyl group of SC-43 formed a hydrogen bond with Q529. In addition, the length of the phenylcyanyl group in SC-43 is shorter than pyridine-mehtylamide of sorafenib, which reduces the steric-hindering effect in the N-SH2 domain. Moreover, the meta connection of the phenyl ring between the urea and phenylcyanyl moiety in SC-43 reduces total length and results in a better fit in the pocket of N-SH2. The discrepancy in potency between sorafenib and SC-43 was likely attributable to these two factors. We further modified SC-43 based on bioisosteric substitution. For example, SC-40, with the replacement of the urea and phenylcyanyl moiety in SC-43 by sulfonamide and nitroaniline, respectively, was able to activate SHP-1 activity. Also, SC-40 demonstrated that the sulfonamide moiety formed hydrogen bonds with R44 and Q529 in the docking model. Together, this discrepancy in binding ability may affect the potency of pharmacological effect among sorafenib, SC-40, and SC-43 (Fig. 5F).
SHP-1 Mediates the Effects of SC Derivatives on p-STAT3 and Apoptosis
Apoptosis was inhibited in myc-tagged STAT3-overexpressing HCC cells after exposure to SC derivatives for 24 h as evidenced by sub-G1 analysis (Fig. 6A). In addition, SHP-1 phosphatase-specific inhibitor (PTPIII) reversed SC-induced cell death and inhibition of p-STAT3 (Fig. 6B). Silencing SHP-1 markedly restored SC-43- and SC-40-induced apoptosis and inhibition of p-STAT3 (Fig. 6C). Conversely, overexpression of WT SHP-1 induced potent apoptosis and inhibited p-STAT3 as a result of SC-43 and 40 treatments in PLC5 cells (Fig. 6D). Titration of dN1 or D61A also gradually restored inhibition of p-STAT3 in SC-43-treated cells; and the apoptosis induced by SC-43 was abolished in dN1 and D61A-expressing PLC5 cells (Fig. 6E,F).
SC-43 and 40 Show More Potent Inhibition of Tumor Growth, Compared to Sorafenib in Orthotopic and SC HCC Models
We established an HCC orthotopic model using luc2-expressed PLC5 cells inoculated into liver of nude mice. Long-term monitoring showed that SC-43 treatment had an evident anti-HCC effect and significant survival benefit, compared with mice treated with vehicle or sorafenib (Fig. 7A). In addition, SC PLC5 tumor-bearing mice were treated daily with vehicle, sorafenib, SC-43, or SC-40 at the dosage of 10 mg/kg/day orally. Compared to sorafenib, SC treatment had an inhibitory effect on tumor growth and the average tumor sizes of animals were less than half that of control mice at the end of treatment (Fig. 7B). To further correlate the molecular mechanism with the anticancer effect in vivo, p-STAT3 and SHP-1 activity in tumor extract from vehicle- and SC-treated mice was analyzed by immunoblotting. Down-regulation of p-STAT3 and elevation of SHP-1 activity were noted in SC-43/40-treated tumor lysate (Fig. 7C,D). The pharmacokinetics of SC-43 was determined (Fig. 7E). SC-43 exhibited a longer period of stability in vivo than that reported for sorafenib in a previous study.
Taken together, these results confirm that the sorafenib derivatives had increased SHP-1 activity that repressed the p-STAT3 involved in tumor inhibition in PLC5 xenograft and were more potent SHP-1 enhancers than sorafenib.
SHP-1 was first identified in hematopoietic cells and is an important regulator of various biological processes in lymphocytes. However, the underlying molecular mechanism by which SHP-1 affects carcinogenesis is still poorly understood. In leukemia and lymphoma cell lines, SHP-1 is believed to function as a tumor suppressor, as evidenced by decreased levels of protein and messenger RNA.[13, 16] SHP-1 is also believed to play a suppressive role in other tumors, such as in estrogen-receptor-negative breast cell lines. However, in some solid tumors, such as prostate cancer,[17, 18] ovarian cancer, and breast cell lines,[20, 21] overexpression of SHP-1 accompanied aggressive tumor progression. Interestingly, SHP-2, which shares almost 70% sequence similarity with SHP-1, was recently reported to have a novel tumor-suppressor function in HCC. Notably, we did not find that sorafenib or SC compounds significantly induced the induction of SHP-2 activity in HCC cells and in purified SHP-2 proteins. In addition to D61, several critical sites of inhibitory N-SH2 domain, which are also involved in blocking the WPD loop to form autoinhibited structure, are different between SHP-1 and SHP-2, such as S59 and F62 in SHP-1 and T59 and Y62 in SHP-2. The further investigation of molecular discrimination between SHP-1 and SHP-2 will be helpful to improve sorafenib-related clinical studies.
The crystal structure of the ligand-free SHP-1 protein has an autoinhibition formation dependent on the inhibitory effect of the N-SH2 domain on the catalytic PTP domain.[10-12] Of note, the specific residue, D61, forms a critical salt bridge resulting in a “closed” catalytic PTP domain. Having clarified the molecular mechanism by which sorafenib affects SHP-1 in HCC, we then attempted to engineer novel sorafenib derivatives with elevated SHP-1 activity to analyze their anti-HCC effect. After screening, SC-43 and SC-40 were characterized to have better biological effects than sorafenib in HCC with more potent SHP-1 activity. We found significant induction of SHP-1 activity in purified SHP-1 protein after treatment with SC-43 or SC-40, compared to sorafenib. The molecular docking between SC derivatives and SHP-1 crystal structure proposed a binding to the N-SH2 domain and further releasing of the PTP domain in SHP-1, which provided a molecular foundation for the present structure-based optimization. SC-43 and SC-40 exerted better biological effects on HCC cells and were effective in HCC cells that were resistant to sorafenib.
In conclusion, in this study, we demonstrated that sorafenib affects SHP-1 activity by impairing the inhibitory N-SH2 domain to release SHP-1. New sorafenib derivatives have been developed as better anti-HCC agents by targeting SHP-1.