Introduction

Microtubules, filamentous cytoskeleton protein polymers composed of α- and β-tubulin heterodimers, are vital components of all cells and play diverse roles in a variety of essential cellular processes including maintenance of cell structure, protein trafficking, chromosomal segregation, and mitosis1,2,3. Microtubule targeting agents are known to interact with tubulin through at least four binding sites: the laulimalide site, paclitaxel site, vinblastine site, and colchicine site4. As one of the tubulin-targeting agents, colchicine binding site inhibitors exert their biological effects by inhibiting tubulin assembly and suppressing microtubule formation5,6,7,8,9. A water-soluble phosphate prodrug CA-4P targeting colchicine binding site has FDA-designated orphan drug status for the treatment of anaplastic thyroid cancer and ovarian cancer10. However, neural and cardiovascular toxicities of CA-4P currently represent the main obstacle to broad its clinical application in different cancers11. Therefore, there is still a need to develop new inhibitors of tubulin polymerization by targeting the colchicine binding site for cancer therapy.

β-Lactam skeleton has attracted much attention from medicinal chemists for many years because of their numerous biological activities12,13, especially their antitumor activity14,15,16. Apart from their pharmacological use, β-lactams have been used as synthons in the preparation of various heterocyclic compounds of biological significance17. For example, suitably substituted hydroxyl β-lactam has been used in the semisynthesis of a side chain bridged paclitaxel18. Importantly, some β-lactam analogues were shown to cause apoptosis in cancer cells through induction of microtubule disorganization and mitotic catastrophe19,20. Hence, in this study, β-lactam was chosen as a basic skeleton to develop new antitumor agents.

In addition, azide moiety has been widely used as a scaffold to design new chemical entities for anti-proliferation21,22,23,24,25,26 and arylazide derivatives have been shown to be potent antitumor agents. For example, trans-2,3-dimethoxycinnamoyl azide derivative 1 (Fig. 1A) enhanced the in vitro and in vivo antitumor effect of romidepsin on bladder cancer cells27. The Combretastatin A-4 aryl azide analogue 2 (Fig. 1A) displayed a potent anti-tubulin activity with an IC50 value of 5.2 μM28. 3′-(4-Azidophenyl)-3′-dephenylpaclitaxel 3 (Fig. 1A) was developed as a novel paclitaxel photoaffinity probe and shown to be as active as paclitaxel in tubulin assembly and cytotoxicity assays29.

Figure 1
figure 1

(A) Azide derivatives as anticancer agents. (B) Five key regions (I-V) to explore detailed structure activity relationships of β-lactam-azides.

These intriguing findings and our continuous quest to identify more potent antitumor candidates30,31,32 led us to design novel β-lactam and azide hybrids. In this work, a series of β-lactam-azide derivatives as tubulin polymerization inhibitors were synthesized and evaluated their antitumor activity in vitro and in vivo. In addition, the detailed structure activity relationships in five regions of β-lactam-azides were explored (Fig. 1B) to provide further insight for developing more efficient tubulin targeting and antiproliferative agents for cancer therapy.

Results and Discussion

Chemistry

The synthetic routes of the desired novel β-lactam-azide derivatives were outlined in Fig. 2. Synthesis of β-lactams 12–28 was carried out using Staudinger reaction with in situ generation of a ketene and subsequent reaction with the appropriately substituted imines33. The trans stereochemistry was observed for azetidin-2-one derivatives 12–28 with aromatic rings directly attached to positions 3 and 4 of the β-lactam scaffold, as evidenced by the coupling constants, J 3,4  ≈ 2.4 Hz. No cis isomers (J 3,4  ≈ 5 Hz) were detected in this series, possibly due to steric hindrance between the 3- and 4-positions of the β-lactam ring (Fig. 2A,B). To explore the effect of a large group or a long chain on the phenyl ring at the C-4 position of the β-lactam, 1,2,3-triazole analogues 29–32 in Fig. 2C as ring-closing products were synthesized through a Huisgen 1,3-dipolar cycloaddition34. β-Lactams-triflones 35–36 were prepared via a Staudinger [2 + 2] cycloaddition of imines with aryl trifly ketene generated in situ from 2-diazo-1-phenyl-2-(trifluoromethylsulfonyl) ethanone 34 by a Wolff rearrangement in satisfactory to good yields from the reported procedure35. An X-ray crystallography study of the β-lactam products was undertaken to confirm the stereochemical assignments and explore possible important structural features for potent activity. ORTEP diagram for compound 28 (CCDC number: 1526687) was presented (see Supplementary Fig. S1).

Figure 2
figure 2

Reagents and conditions: (a) NaNO2, 2 M H2SO4, 0 °C, 30 min, NaN3, rt, 2 h; (b) PCC, CH2Cl2, rt, 2 h; (c) EtOH, reflux; (d) substituted phenylacetyl chloride, triethylamine, anhydrous CH2Cl2, reflux, 3~8 h; (e) triethylamine, imines 8–9 or 11, anhydrous CH2Cl2, MgSO4, reflux, 3~12 h; (f) CuSO4.5H2O, VcNa, THF-H2O, rt, 12 h; (g) CF3NaO2S, DMAC, N2, 50 °C, 14 h; (h) TfN3, CH3CN, pyridine, 0 °C, 16 h; (i) Substituted imine, toluene, N2, 100 °C, 2 h.

Biology

Structure Activity Relationships

All synthesized β-lactam-azide derivatives were evaluated for their antiproliferative activity against three cancer cell lines (MGC-803, MCF-7, A549) using CCK-8 proliferation assay. The well-known anticancer drug CA-4P was used as a control36,37. The results were summarized in Table 1.

Table 1 IC50 values (μM) of synthesized compounds.

In a series of analogues 1225, we mainly investigated the effects of substitution on the phenyl ring at the C-3 position of the β-lactam (Region I), the importance of 3,4,5-trimethoxyphenyl group as the β-lactam N-1 substituent (Region II). During the SAR studies, we found that the substitution on the phenyl ring at the C-3 position of the β-lactam was important for the activity showing over 8-fold activity loss against the growth of MGC-803 cells, when the hydrogen atom (18) was replaced with the methoxy group (19). β-Lactam-azide derivatives 12–16 and 21–22 without 3,4,5-trimethoxyphenyl group at the N-1 position of the β-lactam displayed relatively lower antiproliferative activity (IC50 > 20 μM) toward three cancer cell lines, indicating that the 3,4,5-trimethoxyphenyl group at the N-1 position was crucial for their antiproliferative activity.

To investigate whether the heterocycle displayed an effect on the antiproliferative activity (Region I), compounds with a thiofuran ring (26–28) at the C-3 position of the β-lactam were synthesized. Replacing the phenyl scaffold at the C-3 position of the β-lactam with a thiofuran ring led to an increased activity (compound 25 vs. 28), indicating the importance of thiofuran at the C-3 position of β-lactam for their antiproliferative activity. We also explored the relationship between the location of azide group and their antiproliferative activity (Region III). Compound 26, which contained -N3 at the para-position, gave the IC50 values of 0.154~4.203 μM toward three cancer cell lines. Interestingly, when the -N3 was moved to the meta-position, the obtained compound 28 displayed the better activity (IC50 values of 0.106–0.507 μM). Thus, the location of -N3 group on the phenyl ring at the C-4 position of the β-lactam displayed a significantly improved antiproliferative activity to the cancer cell lines.

Furthermore, β-lactam-1,2,3-triazoles 2932 were synthesized to evaluate the importance of azide group and the effect of a large group or a long chain on the phenyl ring at the C-4 position of the β-lactam (Region IV). When an azide group was replaced by a large group (1,2,3-triazole) or a long chain (1,2,3-triazole-dithiocarbamate), the inhibitory activity of analogues 2932 was completely lost. Those results suggested that a large group or a long chain on the phenyl ring at the C-4 position of the β-lactam was unfavorable for antiproliferative activity.

Because trifluoromethanesulfonyl moiety was an active group in some anticancer agents, it was often used as a promising scaffold for drug discovery38,39,40,41. To complete the SAR study, the effects of hydrogen atom and a large group (eg: −SO2CF3) as the β-lactam C-3 substituent were investigated (Region V). Changing the hydrogen atom at the C-3 position of the β-lactam to the trifluoromethanesulfonyl group resulted in inactive compounds 35–36, indicating that the hydrogen atom was critical for antiproliferative activity. The detailed structure activity relationships of all the synthesized β-lactam-azide derivatives were summarized as Supplementary Fig. S2.

Compound 28 induces G2/M arrest in cell cycle progression

Due to the most potent antiproliferative activity against all selected tumor cells, compound 28 was chosen to further investigate its underlying biological mechanisms42. As shown in Fig. 3, 28 induced cell cycle arrest at the G2/M phase in a concentration and time dependent manner. 28 treatment of MGC-803 cells at concentrations of 0, 0.1, 0.2 and 0.3 μM for 24 h resulted in 3.58%, 14.13%, 23.42% and 44.37% of G2/M populations, respectively (Fig. 3A). when MGC-803 cells were exposed to 0.1 μM 28 for 0, 12, 24, and 36 h, the percentages of MGC-803 cells at the G2/M phase were 5.77%, 12.63%, 23.02% and 44.59%, respectively (Fig. 3C).

Figure 3
figure 3

(A,B) MGC-803 Cells were treated with 28 at 0 μM, 0.1 μM, 0.2 μM and 0.3 μM for 24 h. (C,D) MGC-803 cells were treated with 28 at the indicated concentration (0.1 μM) for 0, 12, 24, and 36 h.

Compound 28 induces cell apoptosis by increasing the expression of BAX and decreasisng the expression of Bcl-2 leading to activation of the caspase cascade

We next evaluated whether 28 induces apoptosis in MGC-803 cells by flow cytometry analysis of propidium iodide (PI) and Annexin V stained cells43. As shown in Fig. 4, 28 caused cell apoptosis in a concentration-dependent manner. When MGC-803 cells were incubated with 28 at 0.1, 0.2, and 0.3 μM for 24 h, the percentages of early apoptotic cells were 12.4%, 40.5%, and 49.7%, respectively and those of late apoptotic cells 2.1%, 10.1%, and 9.4%, respectively (Fig. 4A). Bcl-2 family proteins were crucial components of mitochondrial stress-induced cellular apoptosis44. Thus, the expression of apoptosis-related proteins was also determined. Western blotting analysis of 28 treated MGC-803 cells further revealed an increased protein expression of BAX and a decreased expression of Bcl-2, which was accompanied by increased expression of cleaved caspase-3, caspase-9 and PARP in a concentration-dependent manner (Fig. 4C).

Figure 4
figure 4

(A,B) The apoptotic effects of 28 on MGC-803 cells at 0.1, 0.2 and 0.3 μM concentrations. (C) Western blotting analysis of apoptosis-related proteins in 28 and CA-4P (0.01 μM) treated MGC-803 cells.

Compound 28 induces tubulin destabilization targeting the colchicine site

As the microtubule system plays a vital role in the maintenance of cell shape and basic cellular functions, an immunofluorescence staining assay was performed to study whether compound 28 could disrupt the microtubule dynamics in living cells45. 0.075 μM 28 moderately depolymerized interphase microtubules, whereas the depolymerization effect of 0.3 μM 28 is much stronger in MGC-803 cells (Fig. 5A).

Figure 5
figure 5

(A) Immunofluorescence Staining of Tubulin. MGC-803 cells were plated in culture dishes and incubated with 28 at the indicated concentrations (0, 0.075, 0.15, 0.2 and 0.3 μM), 0.004 μM CA-4P, 0.030 μM Colchicine and 0.004 μM Paclitaxel for 24 h. (B) Inhibition of Tubulin Polymerization assay. (C) EBI competition assay on MGC-803 cells. (D) Molecular modeling study, superimposition of the compound 28 (Red) and DAMA-colchicine (Blue) within the colchicine-binding site (PDB code 1SA0). (E) ligand-protein interactions of 28.

The in vitro tubulin polymerization inhibitory activity of 28 was then evaluated. Purified and unpolymerized tubulin was incubated with 28 at indicated concentrations, and tubulin polymerization was measured by the method originally described by D. Bone et al.46. Derivative 28 inhibited in vitro polymerization of a concentration dependent manner (Fig. 5B), with an IC50 value of 2.262 ± 0.017 μM.

In order to evaluate whether 28 directly binds to tubulin at the colchicine binding site, we carried out a competition assay with N,N′-ethylene-bis(iodoacetamide) (EBI) in MGC-803 cells as described in a previously published paper47. EBI was an alkylating agent that cross-links the Cys239 and the Cys354 residues of β-tubulin involved in the colchicine-binding site, forming a EBI: β-tubulin adduct48. The adduct was easily detectable by Western blot as a second immunoreactive β-tubulin band that migrated faster than β-tubulin itself49. Preincubation of 28 (0.25 and 5 μM) dose-dependently prevented the formation of the EBI: β-tubulin adduct, resulting in the disappearance of the adduct band, which was consistent with the effect of colchicine (10 μM). Thus, the assay (Fig. 5C) indicated that 28 may directly bind to the cochicine-binding site of β-tubulin.

In continuation with our efforts to rationalize our experimental findings and investigate the potential binding site of the target compound with tubulin-microtubule system, molecular modeling studies were performed as described previously50. Docking studies in Fig. 5D and E showed that 28 occupied the colchicine binding site of tubulin in agreement with the X-ray structure of tubulin cocrystallized with a colchicines derivative, N-deacetyl-N-(2-mercaptoacetyl)colchicine (DAMA-colchicine, PDB entry 1SA0)51. Compound 28 formed hydrophobic interactions with the residues of Val318, Val315, Leu248, Leu255, Ile378, Ala316, Ala354. The azide group of phenyl ring at the C-4 position of 28 formed a hydrogen bond with the residue Val238. Importantly, the p-methoxy group of phenyl ring at the N-1 position of the β-lactam 28 formed a hydrogen bond with the residue Ala250, which could explain the importance of trimethoxyphenyl ring for its potent antiproliferative activity.

Compound 28 inhibits migration of MGC-803 cells by up-regulation of E-cadherin and ZO-1 expression and down-regulation of N-cadherin expression

The epithelial- mesenchymal transition (EMT) was an unique process for the phenotypic changes of tumor cells characterized by a transition from polarized rigid epithelial cells to migrant mesenchymal cells, thus conferring the ability of tumor invasion and metastasis52. EMT could suppress tubulin tyrosine ligaseand promote microtubule stability53, resulting in tubulin detyrosination and the formation of microtentacles for supporting endothelial cell attachment54. In this study, we found that 28 could reverse the EMT progress (Fig. 6).

Figure 6
figure 6

(A) Scarification test of 28 on MGC-803 cells. (B,C) Transwell test of 28 on MGC-803 cells. The data were presented as the mean ± SEM *P < 0.05, **P < 0.01. (D) MGC-803 cells were harvested and lysed for the detection of EMT-related markers after treated by different concentrations of 28.

In a scarification test, compared to control, the distances of scratches after MGC-803 cells were treated with 28 obviously increased in a time-dependent and concentration-dependent manner (Fig. 6A). In a transwell test, the average numbers of migrated cells in fields of control, 0.1 and 0.2 μM 28 treated MGC-803 cells were 29.4, 22.4 and 16.6, respectively (Fig. 6B and C). Based on the results from these tests, we next examined the protein expression of EMT-related makers such as E-cadherenin, N-cadherenin, ZO-1 and Vementin55. The results showed that the expression of E-cadherin and ZO-1 was up-regulated and that N-cadherence and vementin were down-regulated by 28 (Fig. 6D). This result suggests that 28 could inhibit the EMT process in tumor cells.

The in vivo antitumor effect of 28 in a xenograft model

To evaluate the potential antitumor effects of 28 in vivo, a MGC-803 xenograft model were established in nude mice by subcutaneously injecting MGC-803 cells at its logarithmic phase into the right flank of mice56. Tumor bearing mice were then randomly assigned to five groups (control, 100 mg/kg CA-4P, 25, 50, 100 mg/kg 28) with 10 mice per group. Then, the mice were gavaged with saline (control), CA-4P and different concentrations of 28 in saline solution daily. The results in Fig. 7 showed that 100 mg/kg 28 caused a considerable suppression of tumor growth. At the end of the observation period, mean tumor volumes of control, CA-4P, 25, 50, and 100 mg/kg 28 groups were 1665.98 ± 568.36 mm3, 642.61 ± 449.92 mm3, 788.18 ± 435.92 mm3, 1125.93 ± 668.25 mm3 and 1273.88 ± 513.69 mm3, respectively. The average tumor weights of control, CA-4P, 25, 50, and 100 mg/kg 28 groups were 1.23 ± 0.28 g, 0.45 ± 0.22 g (inhibitory rate: 63.27%), 0.53 ± 0.20 g (inhibitory rate: 59.34%), 0.85 ± 0.32 g (inhibitory rate: 30.78%), and 1.09 ± 0.31 g (inhibitory rate: 10.99%), respectively. The antitumor activity of 28 in vivo was similar to that of CA-4P. Importantly, the in vivo antitumor efficacy of 28 was achieved without causing any obvious loss of body weight (Fig. 7D). This result suggests that 28 has low toxicity toward mice.

Figure 7
figure 7

The in vivo antitumor activity of 28. After administered with control (saline), CA-4P, 25, 50, and 100 mg/kg 28 for 21 days, the mice were sacrificed, and the tumors were weighed. (A,B) The images of euthanized mice and excised tumors. (C) Tumor volumes of mice in each group. (D) Body weights of mice from each group at the end of the observation period. (E) The weights of excised tumors from each group. The data were presented as the mean ± SEM *P < 0.05, **P < 0.01, significantly different compared with the control by test.

Conclusion

A series of β-lactam-azide derivatives were designed, synthesized and evaluated for their antitumor activities. Among them, compound 28 possessed the most potent antiproliferative ability with an IC50 value of 0.106 μM against MGC-803 cells.

The first SAR for this β-lactam-azide scaffold was explored and highlighted as follows: (1) 3,4,5-trimethoxy phenyl ring at the N-1 position of the β-lactam was essential; (2) a hydrogen atom at the C-3 position of the β-lactam was required for their potent antiproliferative activity, the large group (e.g.: –SO2CF3) at the same position diminished the inhibitory activity; and (3) the large group 1,2,3-triazole and long chain 1,2,3-triazole-dithiocarbamate on the phenyl ring at the C-4 position of the β-lactam completely diminished its antiproliferative activity.

Preliminary mode of action studies demonstrated that 28 halted cell cycle progression at the G2/M phase and induced apoptosis in MGC-803 cells via increased expression of BAX and decreased expression of Bcl-2. Immunofluorescence staining, in vitro tubulin polymerization inhibition and EBI competition assays, as well as molecular modeling study identified that compound 28 was a novel tubulin polymerization inhibitor probably by binding to the colchicine site of tubulin. For the first time, we reported that 28 inhibited cell migration by inhibiting the EMT process in gastric cancer cells. Importantly, 28 inhibited in vivo tumor growth in a xenograft model without apparent toxicity. The antitumor efficacy of 28 in a xenograft model of MGC-803 cells is close to that of a FDA approved anti-tubulin drug, CA-4P. Taken together, compound 28 could be a lead candidate for its further development in treatment of gastric cancer.

Experiment Section

Chemistry section

(The detailed information is in Supplementary Information)

Biological section

Cell-based cytotoxicity screening assay

MGC-803 cell line was obtained from the Chinese Academy of Sciences (Shanghai, China). MCF-7 and A549 were from the National Cell Center, China. MGC-803 cells were cultured in DMEM culture medium (DMEM, Biological Industries, Kibbutz Beit Haemek, Israel). A549 and MCF-7 cells were cultured in 1640 medium (1640, Biological Industries, Kibbutz Beit Haemek, Israel). All medium were supplemented with 10% fetal bovine serum, 100 U/ml penicillin (North China Pharmaceutical group Co.), and 100 µg/ml streptomycin (North China Pharmaceutical group Co.). All cells were cultured at 37 °C in a humidified incubator containing 5% CO2.

Synthesized analogues used in the study were dissolved in 100% cell culture grade DMSO and the final concentration of DMSO as <0.1% for testing on cancer cell lines. Cytotoxicity assays were performed on the human gastric carcinoma cell line MGC-803, the human mammary carcinoma cell line MCF-7, the human lung carcinoma cell line A549. Cells (3500–5000/well) were seeded into 96-well plates in 100 µL of culture medium. The cells were treated in triplicate with a gradient concentration of testing compounds and incubated at 37 °C, 5% CO2 for 48 h. For all cell lines, CCK-8 assay was performed to measure cytotoxic effects. The absorbance was measured using a microplate reader (BioTek Instrument, Inc. Vermont, USA) with a test wavelength of 450 nm. The absorbance levels were corrected against untreated control absorbance values. All experiments were performed in triplicate and SPSS17.0 was used for data analysis to obtain IC50 values.

Cell cycle analysis by flow cytometry

MGC-803 cells were incubated in the absence and presence of 28. Subsequently, the cells were fixed with 70% alcohol in PBS. The fixed cells were incubated with RNase (1 µg/mL) propidium iodide (50 µg/mL) for 2 h. Flow cytometry analysis was performed using BD FACS (Becton Dickinson, San Jose, CA, USA).

Analysis of apoptosis

An Annexin V-FITC/PI kit (KeyGEN BioTECH, Nanjing, China) was used to detect apoptosis. Cells were seeded in 6-well plates and treated with 0, 0.1, 0.2 and 0.3 µM of 28 for 24 h. Then MGC-803 cells were collected and suspended in binding buffer containing Annexin V-FITC (0.5 mg/mL) and PI (0.5 mg/mL) and incubated in dark for 20 min and analyzed by flow cytometry (Becton, Dickinson and Company, NJ). The apoptosis related proteins of western blot analysis were conducted according to our previously reported method31,32.

Immunostaining and microscopy

Cell climbing slices were sterilized and placed on the bottom of a 24-well plate for 24 h before MGC-803 cells were incubated in DMSO, different concentrations of 28 or CA-4P solutions were added. The next day the cell climbing slices were fixed by 4% paraformaldehyde for 15 min after washed by PBS for 3 times. 0.5% Triton-X-100 was added and shaked for 20 min. 0.1% BSA was used to block for 30 min and then removed. The cell climbing slices were added α-tubulin antibody (1:100) and incubated overnight at 4 °C in humid box. On the third day the humid box was taken out and balanced at 37 °C for 30 min. The cell climbing slices were washed by PBST for 3 times for each of 3 minutes and coated with FITC antibody (1:500) in a dark place for 1 h at room temperature. DAPI was used to re-stained for 5 min and then removed. The cell climbing slices were sealing by an anti-fluorescence quenching agent and images collected by Laser scanning confocal microscope (Nikon, Japan).

In vitro tubulin polymerization assay

An amount of 5.6 mg/ml tubulin was resuspended in PEM buffer [80 mM PIPES (pH 6.9), 1 mM EGTA, 0.5 mM MgCl2, 1 mM ATP, 10.2% (v/v) glycerol] and then was preincubated with compound 28 or vehicle DMSO on ice. The reaction was monitored by a spectrophotometer in absorbance at 340 nm at 37 °C every 5 min. The final concentrations of 28 were listed as follows: 0, 1, 2, and 4 µM.

EBI competition assay

Six-well plates were seeded with MGC-803 cells at 5 × 105 cells per well. Cells were first incubated with compound 28 (0.25 and 5 µM), or colchicine (10 µM) for 2 h and afterward treated with EBI (100 µM). After 2 h, the cells were harvested and cell extracts were prepared for Western blot analysis. 20 μg of proteins was subjected to gel electrophoresis using 10% polyacrylamide gels. The proteins were transferred onto PVDF membranes, then blocked by 5% nonfatmilk for 1 h, and subsequently incubated with anti-β-tubulin antibody for 16 h at 4 °C. Next, the membranes were washed extensively and immunoreactive proteins were finally detected by chemiluminescence.

Molecular modeling Studies

We investigated the binding modes of the target compound by molecular docking study. For the receptor preparation, the PDB entry 1SA0 was downloaded from the Protein Data Bank (PDB). The 3D structures of the ligand 28 were generated using Chembio3D Ultra 11.0 followed by energy minimization. AutoDock 4.0 program equipped with ADT was used to perform the automated molecular docking50. A total of 60 possible binding conformations were generated and grouped into clusters based on a 1.0 Å cluster tolerance. The docking models were analyzed and represented using ADT.

Cell scarification assay

MGC-803 cells were seeded in 6-well plate until cells grew to confluence. Tips were used to make a scratch on cells. Control and 28 contained culture media without fetal bovine serum were added subsequently after 3 times of washing by PBS. Then the cells were cultured at 37 °C in a humidified incubator containing 5% CO2 and photos taken at 0, 16, 24 and 48 h, respectively.

Transwell cell migration assay

The test was performed in a transwell plate (Corning). MGC-803 cells were added to the upper chamber of a transwell plate. Control and different concentrations of 28 solutions were added to both upper and bottom chambers. Then the transwell plate was cultured at 37 °C in a humidified incubator containing 5% CO2 for 24 h. The next day, all contents in the upper chamber were removed and wiped by cotton buds. Alcohol was used to fix, and the crystal violet was used to stain the transwell for 30 min. After washed by PBS the cell chambers were observed and photos taken. The cell number of the migrated cells through the transwell was calculated by counting 5 visual fields of each group (P < 0.05)55,56.

In vivo anti-tumor activity

Animals were treated according to protocols established by the ethics committee of Zhengzhou University and the in vivo experiments were carried out in accordance with the approved guidelines and approved by the ethics committee of Zhengzhou University. BALB/c nude mice (18 g, aged 4–5 weeks) were purchased from Human SJA Laboratory Animal Co. Ltd. (Hunan, China). Mice were subcutaneously implanted with MGC-803 cells (5 × 106 cells per mouse) on the right flank of nude mice. Once tumor volumes reached to approximately 100 mm3, the mice were randomly divided into corresponding saline, CA-4P (100 mg/kg), 28 (100 mg/kg), 28 (50 mg/kg) and 28 (25 mg/kg) treatment groups (n = 10 mice for each group). The treatment groups received intragastric administration of 28 and CA-4P per day for a period of 21 days. Then, the mice were euthanized and tumors isolated and weighed. Their body weights were measured and tumor sizes determined by vernier caliper measurement every other day.

Statistical evaluation

Data were presented as means ± SD. Statistical analyses were performed by the analysis of variance (ANOVA). All statistical analyses were performed by SPSS 17.0.