Ovatodiolide suppresses colon tumorigenesis and prevents polarization of M2 tumor-associated macrophages through YAP oncogenic pathways

We explored the public databases to show that YAP1 network is upregulated in clinical colon cancer samples and cell lines. First, in Sabates-Bellver colon database [16], composing of 64 clinical samples from different regions, YAP1 messenger RNA (mRNA) was significantly elevated in both colon and rectal adenoma samples (Fig. 1a). In Kaiser colon database analysis (Fig. 1b), comparing normal colon tissues and colon adenocarcinoma, we found that YAP1 mRNA was approximately 2.2-fold higher in the adenocarcinoma samples [17]. Consistently, an elevated YAP1 immunostaining was detected in the paraffin-fixed sections from a small cohort of colorectal cancer patients (Right panel, Fig. 1c, i). In normal colon tissue sections, YAP1 staining appears to be low to medium in the epithelial and glandular cells and low in the nerve/peripheral ganglion (Left panel, Fig. 1c, i) (images obtained from ProteinAtlas database, CAB009370). Our tissue microarray analysis supported that YAP1 expression was found in normal colon crypts and a significantly increased signal was identified in the rectal adenocarcinoma section (Fig. 1c, ii). Consistently, using the online bioinformatics tool SurvExpress, with the default setting, a higher YAP1 expression was associated with a significantly shorter time of recurrence (Fig. 1d) [18]. In addition, the IL-6/YAP signaling has been shown to be associated with increased stemness in breast cancer [19]. In the same database, patients with high risk of recurrence showed a significantly higher IL-6 and YAP1 mRNA level (upper panel, Fig. 1e); in a larger database (Colon Metabase, SurvExpress), a higher expression of IL-6 and YAP1 predicts a lower survival rate (lower panel, Fig. 1e). In support, using a colon tissue microarray, we identified a strong YAP1 expression in the late stages (stages IIIB, IIIC, IVA, and IVB) colon cancer samples (Fig. 2). Furthermore, a recent clinical study demonstrated that YAP1 activation was associated with the poor prognosis and certuximab resistance in colon cancer patients [20]. Collectively, these findings support our hypothesis that YAP1 represents a key target for therapeutic development in colon cancer patients.

Tumor-associated macrophages (TAMs) are implicated in the development and progression of colorectal cancer [21]. Here, we investigate the role of TAMs in generating drug-resistant CRC cells. Both DLD-1 and HCT116 cells were co-cultured with TAMs and subjected to flow cytometric analysis. We found that in the presence of TAMs, the percentage of CD133+ (Fig. 3a) and side-population cells (Fig. 3b) in both DLD-1 and HCT116 were significantly increased. TAM co-cultured CRC cells showed markedly increased colony-forming (Fig. 3c), migratory (Fig. 3d), and self-renewal (Fig. 3e) abilities as compared to their naïve counterparts. The increased tumorigenic phenotypes by TAM co-culture was associated with a higher expression in YAP1, ERK, Akt/mTOR (oncogenic makers), and β-catenin (both oncogenic and stemness marker) (Fig. 3g). More importantly, TAM co-culture resulted in increased resistance toward 5-FU treatment (Fig. 3f). For instance, in the presence of TAM, the IC50 value of 5-FU of DLD-1 was approximately 11-fold higher than its naïve counterpart (85.6 versus 7.73).

TAM-enhanced tumor-initiating ability was tested in vivo. Tumor spheres generated from naïve DLD-1 and from TAM co-cultured DLD-1 (termed + TAM) were injected subcutaneously into NOD/SCID mice to evaluate their tumor-initating potential. We found that the+TAM group initiated tumor growth significantly faster as compared to their naïve counterparts (Fig. 4a). Fold change in tumor size was measured semi-quantitatively based on the bioluminenesce signals; a significantly heavier tumor weight was found in the+TAM group, reflecting the increased tumorigenesis (Fig. 4b). In agreement with our in vitro data, +TAM tumor sample showed an increased level of YAP1, Kras, Akt/mTOR, NF-kB (oncogenic markers), and β-catenin (both oncogenic and stemness marker, Fig. 4c).

We intend to identify compounds which not only target cancerous cells but also TAMs, in contrast to the conventional chemotherapeutic agents which only aim to eliminate cancerous cells. Previously, studies from us and others showed ovatodiolide (OV) as an anti-inflammatory and anti-cancer agent in different cell models [10, 22, 23]. Here, we first demonstrated that OV effectively inhibited the cell viability of HCT116 and DLD-1 colon cancer cells (Fig. 5a). In addition, OV treatment significantly inhibited the colony formation (Fig. 5b) and migration (Fig. 5c). We subsequently showed that OV treatment suppressed the self-renewal ability of CRC cells under serum-deprived condition as reflected by the significantly lower number of colon spheres formed (Fig. 5d). OV-mediated anti-CRC effects were found in association with the decreased expression of YAP1, Kras/MEK/ERK, NF-kB, mTOR (oncogenic markers), β-catenin, and Notch1 (stemness markers) while an increased in Bax expression (pro-apoptosis marker).

After establishing ovatodiolide’s anti-CRC effects, we examined if it could also prevent the generation of M2-polarized tumor-associated macrophages (M2-TAMs), known to promote tumor progression in many cancer types including colon [24]. Human monocytic THP-1 cells were used as the macrophage precursor for our study. Ovatodiolide treatment prevented M2 polarization in THP-1 cells (Fig. 6a), as reflected by the significantly decreased M2 markers (ARG1 and CD23) while no significant difference in M1 markers (CD64 and CCR7). We then demonstrated that ovatodiolide treatment led to the decreased expression of YAP1, β-catenin, Atk, and NF-kB in the M2 TAMs as compared to their un-treated counterparts (Fig. 6b). One of the major cytokines secreted by M2-TAM is IL-6, and IL-6’s oncogenic roles have been well established in different cancer types including colon cancer [25, 26]. Notably, M2 TAMs express an elevated level of IL-6, and ovatodiolide treatment significantly decreased IL-6 expression (Fig. 6b).

To explore if ovatodiolide treatment can suppress IL-6-induced colon carcinogenesis, exogenous IL-6 was introduced to the culture medium. We showed that the addition of IL-6 significantly increased the expression of multiple oncogenic markers including Kras-MEK, YAP1, NF-kB, vimentin (EMT marker), β-catenin (a marker for both oncogenesis and stemness), and ABCG2 (drug resistance) in both HCT116 and DLD-1 cells. Ovatodiolide treatment significantly suppressed these IL-6-induced pathways (Fig. 6c). For instance, major pro-M2 TAM polarization cytokines, IL-4 and IL-13 expression, were markedly suppressed by the OV treatment even in the presence of IL-6 (Fig. 6c).

IL-6 treatment significantly induced the generation of colon spheres (Fig. 6d), and the addition of OV markedly reduced this effect (Fig. 6d). IL-6-enhanced tumor sphere-forming ability was disrupted significantly as the size and the number of formed spheres was significantly less comparing to IL-6-induced counterparts (Fig. 6d). In support, exogenous IL-6 resulted in an increase in CD133+ HCT116 and DLD-1 cells. The addition of ovatodiolide (in the presence of IL-6) significantly reduced the IL-6-induced effects (Fig. 6e). Furthermore, we tested the possibility of ovatodiolide working in synergy with 5-FU to suppress the viability of colon cancer cells. Different concentration combinations of ovatodiolide and 5-FU were assayed to generate the isobolograms. We found that several combinations of 5-FU and ovatodiolide inhibited both HCT116 and DLD-1 cell viability synergistically (CI index <1, in bold, Fig. 6f).

Subsequently, YAP1 expression was downregulated using siRNA technique in both HCT116 and DLD-1 cells (Fig. 7a); stemness markers (YAP1, β-catenin), oncogenic markers (Kras, Akt/mTOR), drug resistance marker (ABCG2), pro-M2 polarization cytokines (IL-4 and IL-13), and mesenchymal marker (vimentin) were downregulated in the wake of YAP1 silencing. These observations were comparable to those found in OV treatment (Fig. 6). The tumorigenic phenotypes including colony formation (Fig. 7b), migration (Fig. 7c), sphere formation (Fig. 7d), and resistance against 5-FU (Fig. 7e) were significantly suppressed in YAP1-silenced cells. In addition, the percentage of CD133+ cells were also markedly reduced in the YAP1-silenced cells; more importantly, YAP1-silencing led to a less IL-6-responsive cells as indicated by the lower percentage of CD133+ cells under the exogenous IL-6 treatment (Fig. 7f). When co-cultured with YAP1-silenced DLD-1 cancer cells, THP-1 monocytes differentiated into macrophages expressing a significantly reduced M2 marker. For instance, M2 markers such as ARG1 and CD23 were significantly lower than the THP-1 macrophages differentiated under YAP1 wild-type DLD-1 cells (Fig. 7g). Interestingly, the M1 markers such as CD64 and CCR7 did not appear to be significantly affected. In support, THP-1 macrophages co-cultured with YAP1-overexpressing DLD-1 or HCT116 cells resulted in an increased expression of M2 markers (Additional file 1: Figure S2B).

Equally important, the number of CD133+ cells were significantly increased in both YAP1-overexpressing DLD-1 and HCT116 cells (Additional file 2: Figure S1A); the tumor sphere-forming ability was also markedly enhanced in YAP1-overexpressing cells (Additional file 2: Figure S1B). In consistent with our proposed role for YAP1, when YAP1 expression is increased so is the level of β-catenin, NF-kB, vimentin, and Kras (Additional file 2: Figure S1C).

The role of YAP1 in M2 polarization was also tested in THP-1 cells by silencing YAP1 (Additional file 1: Figure S2A). We found that the resultant macrophages differentiated from YAP1-silenced THP-1 showed a significantly reduced level of IL-4, TGFB1, and Ym2 (M2 markers) while increased iNOs (M1 marker).

The potential colon cancer suppressive effect of ovatodiolide was investigated in vivo using immune competent mouse xenograft model with CT26 mouse colon cancer cells. OV treatment (5 mg/kg, five times/week) appeared to be as effectively in suppressing tumor growth as standard chemotherapeutic agent 5-FU (30 mg/kg, twice a week) while the combined treatment yielded the most significant tumor inhibitory effect (Fig. 8a, b). Immunohistochemical analysis of the tumor samples revealed that OV alone and OV + 5FU exhibited less number of M2 TAM, as reflected by the lower number of CD206+ cells as compared to those in control tumor counterparts (Fig. 8c); the tumor growth inhibitory effect of OV and OV + 5-FU may be attributed to the reduced staining of YAP1, mTOR, and β-catenin.