The herbal materials of HJD composed of Fructus gardeniae (90 g), Cortex phellodendri (60 g), Radix scutellariae (60 g), and Rhizoma coptidis (90 g). The herbs were all purchased from Ma'anshan Jingquan Traditional Chinese Medicine Decoction pieces, Co., Ltd. The authenticity of the plant species was confirmed by Dr Huang Yawei from Ma'anshan Jingquan Traditional Chinese Medicine Decoction pieces Co., Ltd. HJD was refluxed with water (1:10, w/v) for 2 h. Afterwards, the HJD extract was obtained by drying the concentrate with vacuum concentration. Powered HJD samples were sonicated with methanol for 1 h. The sample was finally filtrated through a 0.45 µM membrane filter prior to HPLC analysis. The separation was performed on a Waters Acquity C18 column (250×4.6 mm, 5 µm). Separation was carried out by acetonitrile (A) and 0.05% K₂HPO₄ and 0.05% triethanolamine in water (B). The linear gradient elution was 82% A, in 0–9 min; 82–73% A, in 9–27 min; 73–60% A, in 27–45 min; 60–82% A, in 45–60 min with a flow rate of 1.0 ml/min. The UV detection was operated at 254 nm, and the injection volume was 20 µl. The external standard method was applied for the identification by comparing their retention time and spectrum against known standards (Fig. 1 and Table I).

The NSCLC HCC827, A549, H460, and H1975 cell lines were obtained from the American Type Culture Collection (Manassas). Cell lines were maintained in DMEM medium supplemented with 10% fetal bovine serum (FBS) at 37°C with 5% CO₂. The acquired erlotinib-resistant HCC827 cell line (HCC827/ER) was established by exposing erlotinib-sensitive HCC827 cells to gradually increasing concentrations of erlotinib from 0.1 to 10 µM over a 6-month period at 37°C. Erlotinib was purchased from Selleck (cat. no. S7786). The caspase-3 inhibitor Z-DEVD-FMK was obtained from Gene Operation (cat. no. IAP1201). Stock solution was prepared in dimethyl sulfoxide at 10 mM and was stored at −20°C.

The Cell Counting Kit-8 (CCK-8) assay was used to determine cell proliferation. Cells (5×10⁴) were prepared and mixed evenly, and then 0.1 ml cell suspension was added to each of the 96-well plates. Cells were cultured in a 5% CO₂ incubator at 37°C for 72 h. Subsequently, 10 µl CCK-8 solution was added to each well and incubated for 4 h (in the dark). Cell proliferation was evaluated by using FLUOstar^® Omega microplate reader to measure the absorbance of cells at the wavelength of 450 nm.

Apoptosis was detected using the apoptosis detection kit (Biosea Biotechnology). The HCC827, H1975 and HCC827ER cells were separately inoculated on a 6-well plate and cultured in complete medium for 72 h, then treated with DMSO (control), HJD only, erlotinib only, and HJD+E, respectively. The cells were collected, washed twice in cold PBS, then stained with Annexin V and propyl iodide (PI). The stained cells were collected and analyzed by flow cytometry. FlowJo software was used to analyze the results.

Caspase-3/7 activity was measured by the Apo-One homogeneous caspase-3/7 Assay kit (Promega, no. G7790). Then, 1×10⁴ cells per well were inoculated, incubated for 24 h and then treated with HJD, erlotinib and HJD+E, respectively, for 72 h. The cells were then washed with PBS three times. The activity of caspase-3/7 was evaluated according to the manufacturer's instructions.

The cells were mixed with RIPA lysis buffer to extract total protein. The protein concentration was measured by bicinchoninic acid (BCA) method. Then protein samples (50 µg) were subjected to 10% SDS-PAGE gel electrophoresis and transferred to PVDF membranes. Subsequently, PVDF membranes were blocked in 5% skimmed milk for 1 h at room temperature and then incubated with primary antibody overnight at 4°C. Cleaved-caspase 3 (cat. no. ab49822, 1:1,000), p-STAT3 (cat. no. ab76315, 1:3,000), t-STAT3 (cat. no. ab68153, 1:2,000), Bcl-2 (cat. no. ab182858, 1:2,000), Bcl-XL (cat. no. ab32370, 1:1,000), GAPDH (cat. no. ab8245, 1:5,000), p-STAT1 (cat. no. ab109461, 1:3,000), t-STAT1 (cat. no. ab180814, 1:1,000), p-STAT2 (cat. no. ab53132, 1:3,000), t-STAT2 (cat. no. ab32367, 1:2,000), p-STAT5 (cat. no. ab32364, 1:2,000), t-STAT5 (cat. no. ab32364, 1:1,000) were purchased from Abcam. Next day, the membranes were incubated with horseradish peroxidase-conjugated goat anti-rabbit IgG H&L (cat. no. ab205718; 1:6,000, Abcam) or goat anti-rat IgG H&L (cat. no. ab97057; 1:5,000, Abcam) secondary antibodies at room temperature for 2 h. Finally, the protein bands were detected using the Tanon-5200 Multi chemiluminescent gel imaging system.

Six-week-old male nude mice were purchased from the Model Animal Reasearch Center of Nanjing University (Nanjing, China). Twenty-four mice (22 g) were maintained in a specific pathogen-free environment with a 12-h light/dark cycle and fed with standard food and water ad libitum. This study and the animal experimental protocols were approved by the Institutional Animal Care and Use Committee of Jiangsu Provincial Academy of Chinese Medicine.

Approximately 5×10⁶ HCC827/ER cells or H1975 cells were suspended in 0.2 ml DMEM medium and injected subcutaneously into the right flank region of nude mice, respectively. When the average tumor size reached ~75 mm³, the mice were randomized into four groups (n=6 each): i) Vehicle group, ii) HJD group (50 mg/kg), iii) erlotinib group (20 mg/kg), iv) combination of HJD (50 mg/kg) and erlotinib (20 mg/kg). All of the treatments were orally administrated once a day for 27 consecutive days. The formula, 1/2(length × width²), was used to estimate tumor sizes. The body weight and the tumor volume (V) were recorded every 3 days. At the end of the experiment, the mice were sacrificed by manual cervical dislocation, which resulted in euthanasia within approximately 10 sec. The tumors were then dissected and weighed to calculate the inhibitory rate.

The animal experiments were approved by the ethics committee of the Affiliated Hospital of Integrated Traditional Chinese and Western Medicine (Nanjing, China).

Tissues were collected and fixed in 10% formalin for 24 h at room temperature, then embedded in paraffin, and sectioned. The tissue sections (5 µm) were dewaxed, incubated with 3% methanol-hydrogen peroxide and washed with PBS three times. Sections were placed in citric acid buffer, heated and boiled, and washed with PBS. Then, the sections were blocked with goat serum (cat. no. 7481; Abcam) for 20 min, and incubated with STAT3 and Bcl-2 antibodies at 4°C overnight. Next day, the sections were incubated with goat anti-rabbit IgG H&L (horseradish peroxidase) antibodies (cat. no. ab205718; 1:5,000, Abcam). A 3,3′-diaminobenzidine (DAB) kit was used to observe the specific labeling (brownish yellow staining), and hemotaxylin was used for counterstain cell nuclei (blue staining). Images were obtained from each slide using a digital trinocular microscope (Zeiss Axio Observer A1).

The collected nude mouse tissues were fixed with formalin, paraffin-embedded, cut into 4 µm slices, and roasted in an oven for 3 h. Sections with xylene were first washed three times, then different gradient ethanol (75, 85, 95 and 100%) was used to rinse the tissues twice, prior to staining the sections with hematoxylin. Then, sections were immersed in hydrochloric acid-ethanol solution, dyed with ammonia water, and redyed with eosin. Finally, sections were immersed in different gradient of ethanol (75, 85, 95 and 100%) and sealed with neutral resin.

SPSS Statistics version 13.0 was used for all statistical analyses (SPSS Inc.). Statistical comparisons were performed using one-way ANOVA followed by Tukey's multiple comparison post hoc test and P<0.05 was considered statistically different. The results were presented as the mean ± SD.

To evaluate whether HJD could overcome the resistance of NSCLC cells to erlotinib, the inhibitory effect of co-treatments on HCC827, HCC827ER, H1975, H460 and A549 cell growth was assessed using a standard CCK-8 assay. The mutation profile of NSCLC cell lines is shown in Table II. Consistent with earlier studies, cells bearing mutated EGFR (HCC827) were more sensitive to erlotinib than those bearing wild-type EGFR (H460, A549). However, H1975 cells with both L858R and T790M mutations were relatively insensitive to erlotinib treatment. HCC827 erlotinib-resistant cells (HCC827ER) were established by a series of stepwise increase of the erlotinib concentration until HCC827ER was no longer responsive to 5 µM erlotinib. HCC827ER cells were more resistant to erlotinib than HCC827 cells (IC₅₀=31.26 vs. 0.11 µM) (Fig. 2A and B).

To determine whether HJD can inhibit the growth of NSCLC cells, cell viability was determined. HJD (0–100 µg/ml) did not produce significant toxicity in NSCLC cells, independent of the EGFR-mutated status (Fig. 2C). However, HJD acting on NSCLC cells showed synergistic effects with erlotinib. The combination of HJD (50 µg/ml) with erlotinib led to 57% inhibition on the ability of HCC827ER cells (Fig. 2D), although erlotinib (10 µM) alone exhibited a slight effect (~20% inhibition) on cell ability (Fig. 2E). In parallel, the H1975 cell line was also sensitive to the combined administration (Fig. 2F).

Apoptosis is involved in the anticancer effect of erlotinib (23). To measure whether the synergistic effect of HJD on erlotinib was associated with apoptosis, cells were stained with Annexin V-FITC/PI and analyzed using flow cytometry. As shown in Fig. 3A, erlotinib-stimulated apoptosis was significantly enhanced by HJD in H1975, HCC827 and HCC827ER cells (Fig. 3A). As demonstrated in ELISA assays, Caspase-3/7 activation was also significantly induced by combination treatment in these cell lines (P<0.05, Fig. 3B), suggesting that erlotinib-augmented activation of caspase was meaningfully promoted by HJD. Moreover, the caspase-3 inhibitor Z-DEVD-FMK could almost abrogate the activation of caspase 3 caused by HJD (Fig. 3C).

To better understand the drug resistance mechanism of HCC827ER cells, we carefully analyzed its characteristics. We found that the level of p-STAT3 in erlotinib-sensitive primary HCC827 cells was much lower than that in HCC827ER cells (Fig. 4A). Abnormal activation of STAT3 in tumor cells ensures the survival of malignant cells under stress conditions and also leads to multi-drug resistance (24). Erlotinib did not reduce the higher levels of STAT3 phosphorylation in HCC827ER cells (Fig. 4B). Therefore, we detected the effect of STAT3 on HJD-mediated erlotinib sensitization. Erlotinib therapy resulted in the feedback activation of STAT3 in H1975 cells (Fig. 4C), consistent with previous studies (13,25). However, HJD significantly inhibited the levels of p-STAT3 and its downstream targets, Bcl-2 and Bcl-XL proteins induced by erlotinib, suggesting that STAT3 may be related to HJD-mediated erlotinib sensitizing. In addition, we evaluated other members of the STAT family. The effect of Erlotinib and HJD was independent of the phosphorylation of STAT1, STAT2, and STAT5 (Fig. 4D). Furthermore, HJD did not potentiate the inhibitory effect of erlotinib on EGFR phosphorylation of HCC827ER and H1975 cells (Fig. S1).

To verify the effect of STAT3 on erlotinib resistance, we transfected HCC827, HCC827ER, and H1975 cells with STAT3-specific siRNAs. The effect of siSTAT3 on apoptosis was assessed. The apoptosis rate of HCC827ER cells induced by erlotinib alone was 24.1±4.4%, and the apoptosis rate increased to 49.1±5.2% after siSTAT3 transfection (P<0.01). A similar increment in erlotinib-induced apoptosis was also observed in both H1975 and HCC827 cells after siSTAT3 transfection (Fig. 5A). siSTAT3 significantly inhibited endogenous STAT3 and enhanced the inhibition of erlotinib on H1975 and HCC827ER cell viability (Fig. 5B and C). Furthermore, plasmids expressing constitutively activated STAT3 (STAT3-C) were employed. Overexpression of STAT3 in H1975 cells significantly inhibited HJD-mediated erlotinib sensitization (Fig. 5D).

Next, we evaluated whether combination therapy could alleviate erlotinib resistance in HCC827ER and H1975 ×enograft nude mice. The nude mice were divided into the control, HJD, erlotinib, and HJD+erlotinib groups. After erlotinib treatment of HCC827ER xenograft, tumor growth was retarded. Although the tumor volume treated with HJD was smaller than that of the control group, there were no substantial differences. The combination of HJD with erlotinib in particular elicited an extremely strong tumor regression (Fig. 6). At the end of the experiment (27 days of co-treatment), tumors were barely palpable. In summary, these results indicate that erlotinib and HJD contributed to substantial tumor shrinkage of established xenografts bearing EGFR T790M mutation.

Caspase 3-positive cells were greatly increased in the co-treatment group, whereas Ki67 expression was decreased in the co-treatment group, as shown in immunohistochemical assays (Figs. 7 and 8). The levels of STAT3 and its downstream signaling target, Bcl-2, were upregulated in the erlotinib group in mice-bearing HCC827ER tumors, compared to the saline-treated group (Fig. 7). As expected, the combination of HJD and erlotinib blunted the expression of STAT3 and Bcl-2 in mice-bearing H1975 ×enografts (Fig. 8). These results were also confirmed by western blot assays. For safety evaluation of the combinatorial treatment, the kidneys, liver, lungs and heart of nude mice were histologically analyzed. No obvious toxicity was observed in the main organs of mice-bearing xenografts subjected to HJD plus erlotinib treatment, as shown in Figs. S2 and S3. In vivo studies suggested that HJD could overcome TKI resistance by inhibiting the STAT3 signaling pathway. The STAT3/Bcl-2 pathway is involved in the in vivo antitumor effect of HJD combined with erlotinib.