Astagali Radix (AsR), Glycyrrhizae Radix (GlR), Cinnamomi Cortex (CC), Rehmanniae Radix (RR), Paeoniae Radix (PR), Cnidii Rhizoma (CR), Angelicae Radix (AnR), Ginseng Radix (GiR), Hoelen (Ho), and Atractylodis Lanceae Radix (ALR) were purchased from Uchida-Wakanyaku Co., Ltd. (Tokyo, Japan). In this study, Cinnamomi Cortex was added to an appropriate volume of distilled water (w/v, 1:10) and extracted at 100°C for 1 h. The extracted solution was filtered and then freeze dried to obtain dried powder. A voucher sample of this extract (INM 10000007, University of Toyama) was preserved in the Research Promotion Office, Institute of Natural Medicine, University of Toyama, Toyama, Japan. Six other Cinnamomi Cortex samples were collected from different regions of Vietnam and China, coded as CC-1 (Vietnam), CC-2 (Guang Xi, China), CC-3 (Guang Xi, China), CC-4 (Guang Dong, China), CC-5 (Guang Xi, China), CC-6 (Vietnam). Water extracts of the above-mentioned six CC samples were prepared and provided by the National Institute of Biomedical Innovation, Osaka, Japan, and a part of each extract was deposited at our institute (voucher specimen no. INM 10000001–10000006).

Human lung adenocarcinoma A549 cells were cultured in RPMI-1640 medium (Invitrogen, Carlsbad, CA, USA) with 10% fetal bovine serum (FBS; ICN Biomedicals, Aurora, OH, USA), 2 mM L-glutamine (Invitrogen), 100 U/ml penicillin and 100 μg/ml streptomycin in 5% CO₂ at 37°C. The cells were treated with recombinant human TGF-β (5 ng/ml) (Peprotech, London, UK) for various times as indicated, after pretreatment with TGF-β receptor I kinase inhibitor (10 μM) (Merck, Whitehouse, NJ, USA), each herbal component from Juzentaihoto, or Cinnamomi Cortex extracts (CC, or CC-1 to CC-6) for 30 min.

Whole cell lysates were collected in lysis buffer supplemented with some protease and phosphatase inhibitors as described previously (13). Equal amounts of protein were resolved by electrophoresis on acrylamide gels and transferred to PVDF membranes. Antibodies against phospho-specific Smad-2 (Ser 465/467), E-cadherin, and N-cadherin were purchased from Cell Signaling Technology (Beverly, MA, USA) and an antibody against PCNA was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA).

Cell morphology was determined by BZ-8000 (Keyence, Osaka, Japan) after staining with hematoxylin and eosin. For the migration assay, Transwell cell culture chambers were used as described previously (14). Briefly, the filters were precoated with 1.25 μg fibronectin on the lower surfaces. The cell suspension (3×10⁴ cells/100 μl) in serum-free medium was added to the upper compartment and incubated for 6 h. The migrated cells were stained with hematoxylin and eosin, and counted under the microscope in three predetermined fields at a magnification of ×400.

CC extract, 2.0 g, was subjected to reversed-phase silica gel (Cosmosil 75C₁₈-OPN; Nacalai Tesque Inc., Kyoto, Japan) using medium pressure liquid chromatography (MPLC; Buchi, Flawil, Switzerland) with a H₂O-CH₃CN gradient system (98:2→96:4→92:8→90:10→80:20→60:40→40:60 →10:90) to obtain eight fractions (fr. 1, 435 mg; fr. 2, 282 mg; fr. 3, 123 mg; fr. 4, 212 mg; fr. 5, 105 mg; fr. 6, 305 mg; fr. 7, 205 mg; fr. 8, 105 mg). The bioactive fraction 4 (fr. 4, 200 mg) was further subjected to preparative HPLC (Discovery C18 column; 10×250 mm i.d., 5 μm particle size; Supelco, PA, USA) with H₂O-CH₃CN (92:8) containing 0.01% trifluoroacetic acid (TFA) at a flow rate of 2 ml/min to yield a procyanidin trimer (1, 5.5 mg, t_R 23.3 min). The molecular formula of compound 1 was determined by HR-TOF-MS to be C₄₅H₃₈O₁₈ [m/z 865.2003 (M - H)⁺]. Its chemical structure was further identified to be epicatechin-(4β→8)-epicatechin-(4β→8)-epicatechin (procyanidin C1, Fig. 3C) by comparing the [¹H] nuclear magnetic resonance (NMR), [¹³C] NMR, and circular dichroism (CD) spectral data with those in the literature (15).

For chemical profiling of 7 CC extracts, liquid chromatography-mass spectrometry (LC-MS) analysis was performed with a Shimadzu LC-IT-TOF mass spectrometer (Kyoto, Japan) equipped with an ESI interface (Shimadzu). The ESI parameters were as follows: source voltage +4.5 kV, capillary temperature 200°C and nebulizer gas 1.5 l/min. The mass spectrometer was operated in positive ion mode scanning from m/z 200 to 2,000. A Waters Atlantis T3 column (2.1 mm i.d. × 150 mm, 3 m; Milford, MA, USA) was used and the column temperature was maintained at 40°C. The mobile phase was a binary eluent of (A) 5 mM ammonium acetate solution and (B) CH₃CN under the following gradient conditions: 0–30 min linear gradient from 10 to 100% B, 30–40 min isocratic at 100% B. The flow rate was 0.15 ml/min. Mass spectrometry data obtained from the extract were deposited in the MassBank Database and stored with pharmacological information on the extract in the Wakan-Yaku Database System, Institute of Natural Medicine, University of Toyama.

Our previous studies showed that Juzentaihoto, a Japanese Kampo medicine, prevents metastasis in mouse models (10–12). Because Juzentaihoto contains ten (‘Ju’ means ‘ten’ in Japanese) kinds of herbal ingredients, we thus investigated whether each component can suppress TGF-β-induced EMT. We firstly screened the expression of an epithelial marker, E-cadherin, in a human non-small-cell lung cancer cell line, A549 cells, treated with TGF-β, TGF-β receptor kinase inhibitor (TGFRi), or herbal medicines (Fig. 1A). Known as EMT, the reduction of E-cadherin expression in A549 cells was detected after TGF-β treatment in A549 cells. Among the ten herbal components, only Cinnamomi Cortex (CC) extract strikingly suppressed TGF-β-induced EMT, as indicated by the restoration of E-cadherin. We could not detect any inhibition of the cell viabilities by CC extract (data not shown).

In order to further determine the inhibition of TGF-β-induced EMT by CC extract, E-cadherin, N-cadherin, and phosphorylated Smad-2 were examined. Similarly to Fig. 1A, restoration of E-cadherin expression by CC extract was detected in a concentration/time-dependent manner (Fig. 1B and C). Interestingly, phosphorylation of Smad-2 by TGF-β, a downstream molecule of TGF-β signaling, was suppressed by CC extract after the early phase (10 and 20 min) as well as the late phase (24 and 48 h). We also detected the inhibition of TGF-β-induced EMT by CC extract at mRNA levels of snail, E-cadherin and fibronectin (data not shown). In addition to expression levels of EMT markers, we observed morphological changes with CC extract (Fig. 1D). A549 cells treated with TGF-β showed spindle-like shapes as compared with untreated cells. On the other hand, TGF-β-stimulated cells treated together with CC extract showed cobblestone-like shapes similar to the untreated cells or the cells with TGFRi and TGF-β. Moreover, consistent with cell morphology, TGF-β-induced cell migration was also suppressed with CC extract (Fig. 1E). These results support that CC extract suppresses TGF-β-induced EMT in A549 cells and consequently inhibits cell migration.

To gain insight into the variations of CC extracts in EMT-inhibitory activities, we examined six additional CC (CC-1 to -6) extracts, which differ according to the harvest location (Fig. 2A). CC-2, CC-3 and CC-5 extracts showed restoration of E-cadherin expression stronger than CC-1, CC-4, and CC-6 extracts; therefore, an active component(s) for the inhibition of EMT might be included in CC-2, CC-3 and CC-5 extracts more than in CC-1, CC-4 and CC-6.

To identify the active chemical compound(s) that is able to inhibit TGF-β-induced EMT, total ion chromatogram analysis was performed in CC and CC-1 to CC-6 extracts. Notably, the peaks of catechin-catechin-catechin trimer (catechin trimer) significantly overlapped with EMT inhibitory activities in CC-2, CC-3 and CC-5 extracts, but were not detected in CC-1, CC-4, and CC-6 extracts (Fig. 2B), suggesting that the catechin trimer may have crucial roles in suppressing TGF-β-induced EMT.

Next to assess whether catechin trimer fraction of CC extract has activity to inhibit EMT, we examined the protein expression levels with the catechin trimer fraction from CC (Fig. 3A). As predicted, the catechin trimer fraction inhibited E-cadherin reduction and N-cadherin induction. Similarly to protein levels, the change of cell morphology (Fig. 3B) also supported the inhibition of TGF-β-induced EMT.

We finally purified the catechin trimer from CC and identified its structure, which was an epicatechin-epicatechin-epicatechin trimer, procyanidin C1 (Fig. 3C). To check the inhibition of EMT by procyanidin C1, the protein expression and cell migration activities were examined (Fig. 3D and E). Strikingly, EMT markers represented the inhibition of EMT in western blotting, and TGF-β-induced cell migration was suppressed by procyanidin C1.