Background
High grade serous ovarian cancer (OC) remains a devastating gynecological malignancy hallmarked by a high degree of heterogeneity [
1]. Conventional classification scheme based on histopathological evaluation failed to achieve the primary objective of directing clinical intervention and stratifying patient prognosis [
2]. Inspiringly, tumor molecular profiling has launched a revolution aiming to better discriminate cancer subtypes and decipher the underlying driving molecular events of each category [
3]. Up to now, the most successful attempt in this field regarding epithelial OC was conducted by the Australia Ovarian Cancer Study (AOCS) group and The Cancer Genome Atlas (TCGA) network, which unanimously revealed that the subtype with a reactivated tumor stroma presented the worst prognosis [
4,
5]. The emerged evolutional conceptions based on molecular profiling are advancing our comprehension of OC heterogeneity and implying the importance of the host stroma components in determining the destiny of OC patients.
Reactivated tumor stroma is closely related with tumor desmoplasia and is characterized by the accretion of reactive stromal cells, extracellular matrix (ECM) and the resultant increased matrix stiffness [
6]. The role of desmoplasia in solid tumor initiation and progression is complicated and yet insufficiently characterized [
7]. Fibroblasts represent the majority of tumor stromal cells and primarily contribute to the reactivated stroma and tumor desmoplasia [
8,
9]. It is generally thought that the transition from normal fibroblasts to an activated state known as cancer-associated fibroblasts (CAFs), under the recruitment and transformation of tumor cells, is fundamental for cancer progression [
10,
11]. Even though the reciprocal interactions between cancer cells and fibroblasts has been extensively described, it is still currently unclear how quiescent fibroblasts are reprogrammed into CAFs [
12,
13]. In this scenario, a thorough understanding of the regulatory basis underlying fibroblast activation could aid in the development of stromal fibroblast-oriented strategies against the fibrotic subtype of OC.
The deregulation of transcription factors (TFs) is frequently observed in numerous cancer types. Aberrantly activated pathways in tumor cells eventually converge on the activation of sets of TFs to perturb target genes that coordinate tumor progression [
14]. To date, TF dysfunction research in cancer has focused on their capacity to initiate the transdifferentiation of epithelial cells, but few studies have dissected the pro-tumorigenic effects of TF expression in the stromal compartment [
10]. Recently, TF aberrations were found not merely restricted to tumor cells but also in stromal fibroblasts and lymphocytes [
10,
15,
16]. Among them, the importance of SNAI1 expression perturbation in the tumor stroma has been put forward in numerous studies [
17]. It was demonstrated to modulate fibroblast activation and ECM remodeling in the tumor stroma and predict poor patient survival in breast and colon cancer [
18,
19]. In parallel, Twist1 overexpression was mainly restricted to the tumor stroma and was required for fibroblast activation and matrix stiffness in colorectal cancer [
15]. In OC, CAF heterogeneity and the dominant intrinsic TF aberrations are only beginning to be interpreted. Therefore, we are determined to explore the potent transcriptional regulation that is involved in stromal fibroblast activation and the assignment of desmoplasia subtype of OC.
In this study, we demonstrated that SNAI2 was advantageously expressed in stromal fibroblasts and correlated with tumor stromal activation and a worse OC patient prognosis. Mechanistically, SNAI2 defined a signature in the desmoplasia OC subtype, which further contributed to stroma stiffness and subsequent fibroblast activation. SNAI2 transforms normal fibroblasts to an aggressive form and boost their tumor–supporting role in 3D organotypic culture and OC murine xenograft model. Our results address the role of SNAI2 in reprogramming OC stromal fibroblasts and the facilitation of tumor progression. Thus, SNAI2 and its regulated mesenchymal signature could be an attractive target for stroma-oriented therapy against the desmoplasia subtype of OC.
Methods
Cell culture
Epithelial OC cell line SKOV3 was purchased from ATCC (Rockville, MD, USA). Fibroblast cell line MRC-5 was obtained from the cell bank of the Chinese Academy of Sciences. All the cell lines were authenticated by their source organizations prior to purchase, routinely checked for mycoplasma contamination and used within 4 months after frozen aliquot recovery. Primary normal ovarian fibroblasts (NFs), ovarian CAFs were purified from EOC patient tumor tissues following procedures as previously described [
20]. SKOV3 was maintained in McCoy’s 5A medium and MRC-5, primary NFs and CAFs were cultured in DMEM/F-12 medium with 10% FBS and 1% penicillin/streptomycin (Thermo Scientific), at 37 °C in a 5% CO
2 and 80% humidity incubator. SKOV3 cells had been stably transduced with CMV-Fluc-IRES-RFP lentiviral particles (GeneChem, shanghai, China) and designated as SKOV3-Luc previously, which were further used in subsequent animal experiments.
Public database analysis
Gene expression data (GSE40595, GSE38666, GSE14548, GSE9890, GSE14548, GSE35602, GSE9014, GSE39396 profiling data) were downloaded as raw signals from Gene Expression Omnibus (
http://www.ncbi.nlm.nih.gov/geo), interpreted, normalized and log2 scaled using the online analysis tool GCBI website (
https://www.gcbi.com.cn). The Oncomine expression analysis tool (
www.oncomine.org) was used to examine SNAI2 expression in microdissected breast profile GSE9014. Statistical significance is based upon Student’s
t-test. Gene expression of SNAI2, ACTA2, FAP and PDGFRB in various cancer types was obtained from published gene expression profile included in the TCGA dataset via the cBioPortal tools (
http://cbioportal.org). Pearson’s correlation analysis was performed between SNAI2 and classical CAF markers.
Gene set enrichment analysis (GSEA) and single sample GSEA (ssGSEA)
To determine the enrichment score of specific signatures in the genesets positively correlated with SNAI2 expression in stromal profile GSE40595, GSEA was performed using the publicly available desktop application from the Broad Institute. Gene signatures associated with myofibroblast activation features such as “Stromal_Stimulation_Up”, “Cytoskeleton” and “Smooth_Muscle_Contraction”, and stromal prognostic signatures as “Up_In_Stroma_When_Mets”, “Stromal_Response_Signature” and “Stromal_Signature (Resistance To Chemo)” selected from the MsigDB and published profiles (Additional file
1: Table S1). To compare the activation degree of 141-stroma signature and SNAI2 mesenchymal signature in samples included in OC subtypes, we used ssGSEA to generate geneset activation score as described previously [
21].
Immunohistochemistry and Masson’s trichrome staining
EOC tissues were obtained from patients who had undergone surgery at the Department of Gynecological oncology of Tongji Hospital (Wuhan, China). Informed consent was obtained from all patients. Tumor tissues were from patients diagnosed with advanced (stages III and IV) serous adenocarcinoma. Immunohistochemical staining for SNAI2, was conducted as described in our prior study [
22]. Immunostaining scoring of SNAI2 in cellular nucleus of carcinoma cells and fibroblasts of each tissue was assessed using an scoring system encompassing staining intensity (0 = none, 1 = weak, 2 = moderate, 3 = strong) and the proportion of expressing cells (0 = 0%, 1 = 1~25% 2 = 26~50%, 3 = 51~75%, 4 = 76~100%); the sum of the scores produced the final score ranging from 0 to 12. (Score 0 to 4 was defined as low expression and score 5 to 12 was defined as high expression). Associations between SNAI2 expression and clinicopathological characteristics of the patients were analysed with χ2 test (Fisher’s exact test). The prognostic significance of SNAI2 expression and other clinicopathological variables were first assessed by univariate Cox regression analysis using overall survival time in a cohort of 160 EOC patients. Then, multivariate Cox regression analysis was performed to identify the prognostic effect of those factors. Myofibroblast count in xenograft tissues was assessed by the amount of aSMA-positive cells in the medium field of a view. Masson’s trichrome (Sigma, HT15) staining was performed as described previously on paraffin embedded sections of xenograft tumors [
23]. Quantification of the matrix content in xenograft tumors was carried out using color thresholding in ImageJ software, including data from 5 to 10 crypt units in 5–10 sections per mouse.
Western blot analysis and immunofluorescence
Cells were lysed with RIPA lysis buffer (Beyotime, Shanghai, China) supplemented with a protease inhibitor cocktail (Roche). Protein concentration was measured with a bicinchoninic acid (BCA) assay (Thermo Scientific), and 40 μg of total lysate for each sample was subjected to SDS-PAGE followed by blotting with the indicated primary antibodies. Antibodies for SNAI2 (#9585), SNAI1 (#3879) and YAP1 (#14074) were purchased from Cell Signaling Technology (Beverly, MA, USA). Antibodies against ACTA2 (ab5694), FAP (ab28244), PDGFRB (ab32570) and GAPDH (ab9485) were obtained from Abcam Biotechnology (Abcam, CA, USA). Then incubated with HRP-linked secondary antibody (Abcam) and finally detected using an enhanced ECL system (Pierce). Immunofluorescence of the cytoskeleton was performed as conducted previously [
24]. F-actin was visualized in cells by incubation with rhodamine-conjugated phalloidin (Thermo Scientific). Fluorescence images were taken using an Olympus BX53 microscope (Olympus, Tokyo, Japan). Quantification of fibroblast cellular area was assayed as total cell area divided by cell amounts and total area in a medium field performed by ImageJ software in each treatment group.
Survival analysis
To evaluate SNAI2 expression influence of EOC patient outcome, we performed a meta-analysis of 2970 EOC patient expression profiles and generated forest plot using the ‘curatedOvarianData’ Bioconductor package [
25]. To generate survival plot for SNAI2 in clinical samples and SNAI2 mesenchymal signature in public datasets, we performed this analysis using the “best cutoff” value of SNAI2 score and signature ssGSEA score (all percentiles are computed and the best performing threshold is automatically chosen as the cutoff). Patients were divided into a high-score and a low-score group using the selected cutoff value. Overall survival curves were calculated using the Kaplan–Meier method, and statistical signifcance was assessed using the log-rank test. The analyses were conducted with the R Bioconductor ‘survival’ package.
Collagen gel contraction assay
A total of 6 × 105 fibroblasts in each group were suspended in a collagen gel mixture composed of 100 μL of collagen mix (68.75 μl DMEM/F-12 medium, 0.72 μl 1 N NaOH, 31.25 μl Rat Tail Collagen, Type 1) (Thermo) and 100 μL DMEM/F-12 medium per well in a 24-well ultra-low attachment plate (Corning Life Sciences, Corning, NY). Gels were photographed 12 h after fibroblasts planting, and the area was measured using ImageJ software and expressed as a percentage of the original well area. All contraction assays were performed in triple.
3D organotypic culture
Three-dimentional organotypic culture was performed as described previously by Ernst Lengyel with some modification [
26]. 1 × 10
6 MRC5 or primary NF cells transfected with either Ade-ctrl or Ade-SNAI2 were added in the bottom layer, and 1 × 10
6 tumor cells were added to the top layer 3 days after fibroblast implanting. After 10 days of tumor cells implantation, organotypic gels were harvested and embedded in paraffin. Hematoxylin and Eosin (HE) was performed in paraffin sections and images were taken under an Olympus BX53 microscope (Olympus). Quantification of the proliferation assays was performed as described previously using ImageProPlus software [
27].
Transfection of siRNA, adenovirus and lentivirus
MRC5 or primary CAFs were transfected with Lipofectamine 2000 reagent (Invitrogen) using either negative control (si-NC), SNAI1 specific (si-SNAI1), SNAI2 specific (si-SNAI2) or YAP1 specific (si-YAP1) specific siRNAs (Santa Cruz Biotechnology, CA, USA) according to the manufacturer’s protocol. Recombinant adenovirus expressing SNAI2 was obtained from Vigene Biosciences for transient overexpression of SNAI2 in fibroblasts. Lentiviral shRNAs targeting SNAI2 was obtained from Genechem (Shanghai, China) for knocking down of SNAI2 in primary CAFs. RFP-based flow cytometry sorting was performed to select cells stably transduced with lentivirus targeting SNAI2 (sh- SNAI2) or control vector (sh- NC).
Web-based Gene Set Analysis Toolkit (WebGestalt), STRING and PRECOG analysis
The analysis of gene set enrichment among SNAI2 mesenchymal signature was performed using Web-based Gene Set Analysis Toolkit (WebGestalt) (
http://www.webgestalt.org/). STRING (
http://stitch.embl.de/) was used to annotate functional interaction of proteins. Survival z-scores for SNAI2 in various cancer types were obtained from the PREdiction of Clinical Outcomes from Genomic Profiles (PRECOG) portal (
https://precog.stanford.edu/). Expression of the SNAI2 mesenchymal signature genes mapped on the transcriptome of individual murine hematopoietic and stromal cell types in the ImmGene project (
immgen.com). The plot was generated using MyGeneset tool (
http://rstats.immgen.org/MyGeneSet/).
In vitro model of mechanically tunable COL1-coated polyacrylamide (PAA) gel and atomic force microscopy (AFM) measurement
An in vitro system of mechanically tunable COL1-coated polyacrylamide gel was established as previously described [
28]. Polyacrylamide gels with different mechanical stiffness levels were prepared by mixing 10% acrylamide and 0.01–0.5% bis-acrylamide (Sigma, USA) in a HEPES-buffered solution (pH = 8) supplemented with 10% ammonium persulfate (1/100 volume) and TEMED (1/100 volume). The following two groups of PA gel substrates were prepared: soft (10% acrylamide/0.05% bis-acrylamide), and stiff (10% acrylamide/0.3% bis-acrylamide). The Young’s moduli of the two gels were measured via indentation experiments using an atomic force microscope (AFM) and found to be 3.42 ± 0.21 kPa (soft), and 26.24 ± 0.42 kPa (stiff), respectively. The formed gels were further crosslinked and coated with 0.1 mg/ml of COL-1solution (BD) to facilitate cell adhesion. Fresh tissue samples preparation for AFM measurements were cut into 1~3 mm sections. Two-component epoxy glue (5 min hardening time) was utilized to be put under the tissue. Then place the sample-mounted round glass onto the sample stage and perform the measurements. SPM9700 scanning probe microscope was used to measure tissue stiffness. The Hertz model was used to determine the elastic properties of the tissue. Tissue samples were assumed to be incompressible and a Poisson’s ratio of 0.5 was used in the calculation of the Young’s elastic modulus.
Animal assay
The animal study was performed with the approval of the Committee on the Ethics of Animal Experiments in the Hubei province. Four- to Six-week-old female NOD/SCID mice were housed and maintained in laminar flow cabinets under specific pathogen-free condition. SKOV3-Luc (2 × 106) cells were inoculated subcutaneously in right back of the mice, either alone or co-injected with 2 × 106 Ade-NC or Ade-SNAI2 transfected CAFs (SNAI2low). In parallel experiment, SKOV3-Luc (2 × 106) cells were inoculated either alone or co-injected with 2 × 106 sh-NC or sh-SNAI2 stably transduced CAFs (SNAI2high) (n = 8 per group). Approximately 4 weeks later, mice were anaesthetized with 1% pentobarbital sodium, and imaged with the IVIS SPECTRUM system (Caliper, Xenogen, USA). Total flux (photons/s) of xenografts was analyzed using Living Image version 4.3.1 software.
Statistics
Results are expressed as the means ± standard deviation (s.d.) from at least three independent experiments. Single comparisons between two groups were determined by Student’s t-test. Comparisons between multiple groups were determined by one-way ANOVA followed by Tukey post-test. All statistical analyses were performed in R 3.0.0. P values <0.05 were considered significant.
Discussion
A majority of studies on SNAI2 have focused on its role in tumor cells. In this study, we extended the study of SNAI2 from tumor cells to stromal fibroblasts and discovered that SNAI2 and its associated mesenchymal signature were enriched in the desmoplasia subtype of OC. Moreover, SNAI2 and its defined signature were strongly correlated with tumor stroma activation and OC patient survival. Furthermore, the increased ECM stiffness exerted by SNAI2 in turn able to elevate fibroblast SNAI2 expression and sustain fibroblast activation. Together, our work establishes a role for stromal SNAI2 exertion on fibroblast activation and the facilitation of OC advancement, suggesting that SNAI2 could be a target of choice in aiming the stromal constituent of OC.
The importance of tumor stroma has long been appreciated and it is now widely accepted that CAFs exhibit multiple functions that fuel tumor metastasis [
34]. Molecular profiling has described a subset of the OC samples characterized by an extensive stromal desmoplasia and associated with the poorest survival. Accumulating evidence indicates that CAFs activity perturbs not only the biochemical but also the biomechanical homeostasis of the tumor microenvironment, which ultimately affects tumor cell behavior and promotes cancer advancement [
35]. These data highlighted that targeting stromal CAFs could alleviate stroma desmoplasia and the resultant disease exacerbation. In this setting, our analysis showed that SNAI2 was elevated specifically in the tumor stroma along OC progression and was capable of transforming normal fibroblasts into CAFs. Intriguingly, SNAI2 was most highly expressed in the desmoplastic subtype of OC and was correlated with stromal activation. Moreover, as a classic TF regulating the EMT program in various tumor cells [
36], SNAI2 was supposed to control the cascade of downstream effector gene activation. Thus, SNAI2 was highly accordant with the requirements for being a potent target of choice in aiming for the cooperative stroma.
SNAI2 encodes a zinc-finger protein of the SNAI family of TFs [
37]. It is expressed in the dorsal neural tube and drives EMT, which leads to neuralcrest cell migration during embryonic development [
38]. SNAI2 amplification or interaction with specific oncogenes has been demonstrated in a wide spectrum of human cancers [
39]. It was recently demonstrated that SNAI2 is involved in determining the stem cell state of breast and colon cancer [
40,
41]. Accumulating evidence revealed that TF expression and function are likely to be context-dependent, and different cell or tumor types respond distinctly to TF regulation [
42]. In this regard, both the functional roles in the tumor and the stroma compartment should be taken into consideration in determining the integrative effect of a particular TF. For instance, another SNAI family member, SNAI1, initiates the EMT program and regulates cancer metastasis in tumor cells. In stromal fibroblasts, SNAI1 regulates paracrine signaling and cell transformation [
17]. More recently, Snail1-expressing fibroblasts display mechanical properties that support metastasis in breast cancer [
18]. In parallel, Twist1 expression was correlated with fibroblast activation in gastric cancer. In OC, SNAI2 was previously reported to induced EMT and was correlated with tumor metastasis and angiogenesis [
15]. In addition, our finding first revealed that SNAI2 particularly exists in fibroblast cells and influences stroma activation. Meanwhile, the distinct expression pattern of SNAI1 and SNAI2 in cancer cells and fibroblasts coincides with the context-dependent role of TFs, for which the expression shift implied a unique role for SNAI1 and SNAI2 in the tumor and stromal compartment, respectively. Our subsequent finding that SNAI2, rather than SNAI1, correlated with disease outcome further supports the potential role of SNAI2 in OC progression. The exact mechanism involved in SNAI2 regulation of fibroblast activation in OC remains to be elucidated by future studies.
To better elucidate the SNAI2 regulating downstream cascade network in stromal fibroblasts of OC, we first defined a 77-gene SNAI2 mesenchymal signature through an integrative analysis of Tothill’s C1 cluster and the OC TCGA mesenchymal subtype, for which signature was enriched in functional categories concerning collagen deposition and ECM remodeling. Our defined mesenchymal signature emphasized the contribution of SNAI2 to stromal stiffness and was further demonstrated to be correlated with poor patient survival in OC patients. Our finding resembled a recently discovered collagen-remodeling signature regulated by TGF-β signaling that was associated with metastasis and poor survival in OC and an identified reactive stroma signature associated with primary chemoresistance and predicted OC outcome [
43,
44], which unanimously implied the crucial role of stromal activation in determining the fate of OC patients. Comparatively, our defined mesenchymal signature could better stratify tumor stroma from their normal counterpart in ovarian, breast, and colon microdissected profiles. These preliminary findings concerning SNAI2 in OC tumor stroma is completing our recognition of the role of SNAI2 in OC.
CAF-mediated changes in the physical properties of ECM facilitate tumor cell invasion and metastasis [
45], while the function of these mechanical signals on CAFs is only just being evaluated. Here, we showed that SNAI2 protein levels increase and accumulate in the nucleus of ovarian fibroblasts following exposure to a stiff ECM in culture and further sustain the continuous activation of fibroblasts. YAP was recently reported to be required for CAFs to promote matrix stiffening and was indispensible for matrix remodeling of fibroblast activation in breast cancer [
33]. In addition, SNAI1 was also found to mediate the mechanical signals that regulate CAF activity through the influence of the YAP level [
19]. In addition to the participation of YAP, SNAI2 upregulation was demonstrated to be involved in the regulation of stromal stiffness on fibroblast activation in OC. This observation was in accordance with the notion of the context-dependent regulation of TFs and implied a crucial role for SNAI2 positive feedback in amplifying the role of stroma stiffness on the tumor and stromal cells.
Acknowledgements
The authors give special thanks to Dr. Oneile Isau from the Department of Obstetrics and Gynaecology, Princess Marina Hospital of Gaborone in Botswana for revising the manuscript.