Aspirin was purchased from Sigma-Aldrich (St. Louis, MO, USA) and erlotinib was obtained from LC Laboratories (Woburn, MA, USA). SB-203580, a p38 inhibitor, was obtained from Selleck Chemicals (Houston, TX, USA).

Human lung carcinoma cell lines (NCI-H1299 and A549), ovarian carcinoma cell line (HO-8910), colon carcinoma cell line (HCT-116) and gastric carcinoma cell line (SGC-7901) were obtained from Shanghai institute of biochemistry and cell biology. A549 was maintained in Ham's F12 medium + 10% fetal bovine serum (FBS). SGC-7901 was maintained in RPMI-1640 medium + 10% FBS. HO-8910, NCI-H1299 and HCT-116 were maintained in DMEM + 10% FBS.

The cytotoxic activity was measured by the SRB method, as previously described (27).

Cancer cells (500–1,000 cells/dish) were plated into six-well plates, treated with drugs every 3–4 days for 2 weeks, and then stained by crystal violet (28).

Cancer cells (3×10⁵/well) were exposed to the drugs, harvested and washed with PBS. Then, propidium iodide (PI) staining was used to detect apoptosis, and the mitochondrial membrane depolarization was determined by 5,5′,6,6′tetrachloro-1,1′,3,3′-tetraethylbenzimidazol-carbocyanine iodide (JC-1) staining as described previously (27).

The western blotting was performed as described previously (29). The mouse antibodies used for western blotting were obtained from different resources: Anti-β-actin monoclonal antibody (Ab) (from BD Biosciences, Franklin Lakes, NJ, USA) and anti-XIAP monoclonal Ab (Santa Cruz Biotechnology, Dallas, TX, USA). The rabbit antibodies used for western blotting were purchased from different resources: anti-PARP polyclonal Ab, anti-procaspase-3 polyclonal Ab, anti-Mcl-1 polyclonal Ab and anti-p38 polyclonal Ab (Santa Cruz Biotechnology), anti-phospho-p38 (Thr-180/Tyr-182) polyclonal Ab and anti-E-cadherin polyclonal Ab (Cell Signaling Technology, Danvers, MA, USA).

Cells were seeded in 24-well plates and cultured until they reached confluence. Confluent monolayer cells were gently scratched with a sterile pipette tip and then washed three times with PBS to clear cell debris and suspended cells. Fresh serum-free medium was added, and the cells were allowed to close the wound for 48 h under normal conditions. Images of the wound in the same relative position were captured with a computer-assisted microscope (Olympus, Tokyo, Japan).

Cell invasion experiments were performed using 24-well modified Boyden chambers (Costar, NY, USA) containing a polycarbonate membrane with 8.0-µm pores according to the manufacturer's instructions. First, the membranes were coated with 25 µg Matrigel (BD Biosciences). Cells were seeded at a density of 1×10⁶ cells/well in the upper chamber with culture medium (200 µl) alone, while the bottom of the plate was filled with culture medium (500 µl) supplemented with 20% FBS. Cells that invaded the underside of the membrane were fixed in 1% methanol and stained by crystal violet.

Inhibition of angiogenesis was determined using the CAM assay. Fertilized chicken eggs were incubated at 37°C in a 50% humidified atmosphere. On day 7, the eggshell was cracked and gently opened into the plate to avoid any unnecessary physical stress. It was made sure that the yolk sac membrane remained intact and that the embryo was viable. Then, a sterile filter paper square saturated with aspirin, erlotinib or their combination was placed in areas between vessels, but not onto any large vessels. After a 48-h incubation, the membranes were examined by microscopy and photographic documentation. Angiogenesis was quantified by counting blood vessel branches; at least 10 viable embryos were tested for each treatment.

Cells (3×10⁴) were seeded in 6-well plates. E-cadherin (RG220731; Origene Technologies, Rockville, MD, USA) and empty vector plasmid were transfected into cells using Lipofectamine 3000 (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA) according to the manufacturers instructions.

Cells (5×10⁴) were seeded in 6-well plates. P38 siRNA and control siRNA (Genepharma, Shanghai, China) were transfected into cells using Oligofectamine reagent (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. Sense p38 siRNA sequence was 5′-GCAUAAUGGCCGAGCUGUUTT-3′.

A549 xenografts were performed as previously described (29). BALB/c (nu/nu) mice were maintained under sterile conditions using an individually ventilated cage system, randomized to 4 groups and then treated with vehicle, aspirin (100 mg/kg, i.g. administration) daily and/or erlotinib (20 mg/kg, i.p. administration) twice per week for 29 days (n=6). Finally, the mice were sacrificed at the end of the treatment. The liver were fixed in 10% buffered formalin and embedded in paraffin, 5-µm tissue sections were stained with hematoxylin and eosin (H&E). All animal handling was performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and approved by the Zhejiang University City College Animal Care and Use Committee (Hangzhou, Zhejiang, China).

TUNEL assay was done using TUNEL apoptosis assay kits (Beyotime Institute of Biotechnology, Shanghai, China) as recommended by the manufacturer.

One-way ANOVA followed by Tukey's post hoc test and Two-tailed student's t-tests were used to examine the significance of differences among groups. Data points in graphs represent the mean ± SD (*P<0.05; **P<0.01). For SRB assay, combination index (CI) values were calculated using Calcusyn (Biosoft, Great Shelford, Cambridge, UK) and the mean CI values were chosen for presentation. A CI value <0.9 indicated synergism; 0.9 to 1.10, additive; and >1.10, antagonism.

First, we assessed the anticancer activity of the combination of aspirin and erlotinib by SRB assay in 5 human carcinoma cell lines. The survival curves for aspirin and/or erlotinib are shown in Fig. 1A. We found that aspirin plus erlotinib significantly reduced the survival fraction in human cancer cells compared with each agent alone. To verify the synergistic anticancer effect of aspirin and erlotinib, CI values were calculated. Aspirin plus erlotinib exerted synergistic cytotoxic effects on 5 human carcinoma cell lines (CI values <0.7). In a long-term colony-forming assay, the combination of aspirin and erlotinib resulted in significant inhibition on the proliferation of A549 and NCI-H1299 cells, while monotherapy induced a moderate inhibition (P<0.01, ANOVA) (Fig. 1B). Taken together, these finding indicate that the combination of aspirin and erlotinib was more effective in limiting colony formation and cell growth of human cancer cells in vitro compared with either agent alone.

We first investigated whether the synergistic anticancer effects of aspirin plus erlotinib were related to the induction of apoptosis. The percentage of apoptotic A549 cells was 6.95% in the control group, 28.96% with aspirin treatment, 18.67% with erlotinib treatment, and 62.23% in the aspirin plus erlotinib group (Fig. 2A, top panel). Aspirin plus erlotinib significantly enhanced apoptosis in both A549 and SGC-7901 cells compared with either drug alone (P<0.01, ANOVA; Fig. 2B and C). Next, we investigated whether aspirin plus erlotinib affected the mitochondrial membrane potential. As shown in Fig. 2A (bottom panel), D and E, aspirin plus erlotinib increased the percentage of mitochondrial membrane depolarized carcinoma cells compared with either drug alone (P<0.01, ANOVA). Thus, our data suggested that the synergistic effects of aspirin plus erlotinib are mediated via the mitochondrial apoptotic pathway. In addition, aspirin plus erlotinib markedly induced PARP cleavage and XIAP suppression in two of the cancer cell lines (Fig. 3A). The induction of EMT and activation of p38 are associated with resistance to erlotinib, and blockade of p38 was found to be able to suppress EMT in erlotinib-resistant cancer cells (30). Furthermore, the upregulation of E-cadherin increased the sensitivity of several TKIs, such as gefitinib and erlotinib (31,32). Thus, we were interested in examining the involvement of p38 and E-cadherin in the aspirin and erlotinib combination treatment. Interestingly, as shown in Fig. 3A, the enhanced apoptosis induced by aspirin plus erlotinib was accompanied by p38 inhibition and overexpression of E-cadherin in A549 and SGC-7901 cells, indicating that aspirin may reverse erlotinib resistance via the p38/E-cadherin pathway. To further investigate the involvement of the p38/E-cadherin pathway in the synergistic effects of aspirin and erlotinib, we first performed E-cadherin overexpression experiments by transfecting cancer cells with an E-cadherin plasmid. As shown in Fig. 3B and C, overexpression of E-cadherin increased the apoptosis induced by aspirin plus erlotinib in A549 and SGC-7901 cells. Next, a p38 inhibitor (SB-203580) was found to increase the apoptosis induced by aspirin plus erlotinib in SGC-7901 cells (Fig. 3D and E). Moreover, p38 depletion by siRNA also enhanced aspirin plus erlotinib-induced apoptosis (Fig. 3F). Therefore, these data indicated that the p38/E-cadherin pathway may be involved in the enhanced apoptosis induced by aspirin plus erlotinib treatment.

E-cadherin inhibition is an important step in EMT and a hallmark of metastatic cells (33). Our data already demonstrated that aspirin plus erlotinib could increase the expression of E-cadherin; thus, we hypothesized that aspirin plus erlotinib may inhibit EMT and cancer metastasis. EMT enables cancer cell migration and invasion, and a scratch assay was conducted on A549 and SGC-7901 cells to assess cell migration, which is a defining feature of the mesenchymal phenotype. As shown in Fig. 4A (top panel), untreated A549 cells migrated within 24 h after wounding, whereas there was moderate inhibition of cell migration when the A549 cells were treated with erlotinib at 2.5 µM. However, the suppressive effect on migration was most prominent in A549 cells treated with aspirin plus erlotinib. Similar results were observed in SGC-7901 cells (Fig. 4A, bottom panel). In addition, the effects of aspirin plus erlotinib on cancer cell invasion were evaluated by Matrigel Transwell invasion assays. Our data demonstrated that aspirin or erlotinib achieved moderate inhibition of cell invasion in the A549 and SGC-7901 cell lines. However, aspirin plus erlotinib led to a significant reduction in the number of invading cancer cells compared with the effects of either aspirin or erlotinib alone (P<0.01, ANOVA) (Fig. 4B-D). Our results indicated that the combination of aspirin and erlotinib enhanced the inhibition of cancer cell migration and invasion compared with either drug alone.

Endothelial cells can invade the surrounding basement membrane, and then migrate into the stroma during tumor angiogenesis. Finally, endothelial cells organize to form new capillaries, which are crucial for tumor metastasis (34). To evaluate the anti-angiogenic function of the aspirin plus erlotinib combination, we first detected its effects on migration and invasion of human umbilical vein endothelial cells (HUVECs). There was no inhibition of cell migration when HUVECs were treated with aspirin or erlotinib alone. However, when the two drugs were combined, the inhibitory effect on HUVEC migration was enhanced in the scratch assay (Fig. 5A). Similarly, aspirin plus erlotinib treatment induced a substantial decrease in the number of invading HUVECs compared with aspirin or erlotinib alone (P<0.01; ANOVA) (Fig. 5B and C). These observations suggested that aspirin plus erlotinib may inhibit angiogenesis via reducing invasion and migration of endothelial cells. Next, the CAM assay was performed to further investigate the anti-angiogenic effect of aspirin plus erlotinib. As shown in Fig. 5D, aspirin plus erlotinib blocked angiogenesis in the CAM assay. Quantitative analysis revealed that aspirin, erlotinib, and aspirin plus erlotinib caused a 26.7, 42.2 and 68.9% reduction in the number of blood vessels, respectively (P<0.01; ANOVA) (Fig. 5E). Our data obtained from the three models were sufficient to confirm the synergistic anti-angiogenic activity of aspirin plus erlotinib.

An A549 xenograft model was constructed to verify the anticancer and anti-metastatic efficacy of aspirin plus erlotinib. As demonstrated in Fig. 6A and Table I, aspirin exerted no significant anticancer effect [mean relative tumor volume (RTV)_aspirin: 13.3 vs. mean RTV_control: 14.2; P>0.05, t-test]. Erlotinib exerted a moderate anticancer effect (mean RTV_erlotinib: 9.7 vs. mean RTV_control: 14.2, P<0.05, t-test). Aspirin plus erlotinib induced significant tumor growth inhibition (mean RTV_combination: 3.1 vs. mean RTV_control: 14.2, P<0.01, t-test), which was markedly higher compared with that of either aspirin or erlotinib alone (mean RTV_combination: 3.1 vs. mean RTV_aspirin: 13.3, P<0.05, t-test and mean RTV_combination: 3.1 vs. mean RTV_erlotinib: 9.7, P<0.05, t-test). In addition, the loss of body weight did not differ significantly on day 29 compared with day 0 in the aspirin plus erlotinib group (P>0.05; ANOVA) (Fig. 6B). The TUNEL assay was performed to detect the apoptosis induced by aspirin plus erlotinib in the A549 xenograft model. The number of TUNEL-positive cells increased in the tumor tissues of mice receiving the combination treatment (Fig. 6C). Furthermore, caspase activation and XIAP inhibition were observed in the tumor tissues of mice receiving combination treatment, highlighting that apoptosis was involved in the tumor growth inhibitory effects induced by aspirin and erlotinib in vivo (Fig. 6D). In addition, the number of liver metastases was significantly reduced in the combination-treated group compared with the aspirin or erlotinib alone group in the A549 xenograft model (P<0.01; ANOVA) (Fig. 6E). In conclusion, the synergistic antitumor effects of aspirin plus erlotinib were confirmed in vivo.