https://orcid.org/0000-0003-1247-3844
Abstract
Low-grade and early-stage endometrial cancer usually has a favorable prognosis, whereas recurrent or metastatic disease is often difficult to cure. Thus, the molecular mechanisms underlying advanced pathophysiology remain to be elucidated. From the perspective of the origin of advanced endometrial cancer, the characterization of cancer stem-like cells (CSCs) will be the first step toward the development of clinical management. We established long-term culturable patient-derived cancer cells (PDCs) from patient endometrial tumors by spheroid cell culture, which is favorable for the enrichment of CSCs. PDC-derived xenograft tumors were generated in immunodeficient NOD/Shi-scid, IL-2RγKO Jic mice. Morphologically, PDCs derived from three distinct patient samples and their xenograft tumors recapitulated the corresponding original patient tumors. Of note, CSC-related genes including ALDH1A1 were upregulated in all of these PDCs, and the therapeutic potentiality of aldehyde dehydrogenase inhibitors was demonstrated. In addition, these PDCs and their patient-derived xenograft (PDX) models exhibited distinct characteristics on the basis of their hormone responsiveness and metastatic features. Interestingly, genes associated with inflammation and tumor immunity were upregulated by 17β-estradiol in PDC lines with high estrogen receptor expression and were also overexpressed in secondary PDCs obtained from metastatic tumor models. These results suggest that PDC and PDX models from endometrial cancer specimens would be useful to elucidate CSC traits and to develop alternative diagnostic and therapeutic options for advanced disease.
Endometrial cancer (EnC) is a common gynecological tumor that is derived from the uterine endometrium, the development and maintenance of which are regulated primarily by estrogen and progestin. The disease is the sixth-most commonly diagnosed cancer and 14th leading cause of cancer-related deaths in women worldwide, with 320,000 estimated new cases and 76,000 deaths in 2012 (1). Incidence rates of EnC are much higher in Western and developed countries than in developing countries, including Asia, Africa, and South America (2, 3). In terms of pathogenesis of EnC, several risk factors are assumed to contribute, such as nutritional lifestyle, reduced physical activities, obesity, diabetes, hypertension, and long-term exposure to estrogen (4).
The majority of EnCs are categorized as type I tumors, mostly endometrioid adenocarcinomas, which are generally associated with endometrial hyperplasia (5), mainly due to prolonged exposure to estrogen in the absence of sufficient progestin. In general, early-stage, well-differentiated endometrioid carcinoma usually expresses estrogen receptor α (ERα) and progesterone receptor (PGR). Progestin treatment has been administered to young patients with low-grade EnC for preserving fertility, although the clinical benefit of endocrine therapy in EnC has not been fully clarified in comparison with breast cancer (6). Nevertheless, during clinical evaluation, ERα and PGR are considered prognostic markers for EnC, though in a very limited manner (6). Type II tumors are usually less well differentiated and include serous, clear cell, or squamous carcinomas. The prognosis of type II tumors is poorer than that of type I tumors, as type II tumors are 10% to 20% of the cases of EnC but contribute to 40% of the deaths. Type II tumors generally develop from atrophic endometrial tissues in older women and are mostly estrogen independent.
Recently, cancer stem-like cells (CSCs) have gained particular attention in the field of cancer research, as they are assumed to contribute to cancer therapeutic resistance and recurrence (7, 8). CSCs are defined as small subpopulations of stem-like cells with self-renewal, differentiation, and tumorigenic properties. Tumor-derived spheroid culture was recently shown to facilitate the enrichment of CSC populations from primary patient-derived cancer cells (PDCs) (8). PDC spheroid cultures have been established from several solid tumors such as gliomas (9), breast cancer (10), ovarian cancer (11), and colon cancer (12). PDC spheroids and PDC-derived xenograft models are useful for analysis of the gene expression signature of cancer cells with CSC traits and for screening of antitumor drugs. This could be achieved because they can recapitulate their corresponding original tumors and can reflect the original characteristics of drug sensitivity.
On the basis of current trends in cancer research, the National Cancer Institute in the United States has developed a national repository of patient-derived models consisting of PDCs and patient-derived xenografts (PDXs) and has used this repository for drug screening instead of the National Cancer Institute-60 human cancer cell line screen (13). With respect to experimental models of EnC, PDX models from surgical specimens (14, 15) or cell line‒based xenograft models (16–20) have been reported; nevertheless, the CSC traits of EnC remain to be characterized. In addition, hormone-responsive EnC models, which can be applied to preclinical study, remain to be established. To our knowledge, the Ishikawa cell line is the only appropriate model for the study of hormone responsiveness, which is often diminished under conventional culture conditions.
In the current study, we aimed to establish several different types of EnC models directly from original patient tumors and to elucidate the pathophysiological characteristics of the models, particularly for defining CSC traits of EnC. We generated long-term culturable PDC spheroids from clinical EnC specimens and PDC-derived xenograft tumors in immunodeficient mice. Established PDC models have common CSC traits but distinct features: a hormone-responsive endometrioid carcinoma model with activation of cytokine signals, an endometrioid carcinoma model with mild hormone responsiveness, and a hormone-independent mixed adenocarcinoma (serous and clear cell carcinomas) model. Metastatic models were also obtained from the hormone-independent mixed adenocarcinoma model. According to common CSC traits, the therapeutic potentiality of aldehyde dehydrogenase (ALDH) inhibitors was demonstrated in each disease model. The PDC spheroid cultures and PDC-derived xenograft models will be particularly useful to identify novel therapeutic targets for advanced disease and to screen for antitumor drugs.
Materials and Methods
Clinical specimens, isolation of cancer cells, and spheroid cultures
Surgical specimens from patients with EnC were obtained from the Saitama Medical University International Medical Center. All procedures were performed under a protocol approved by the Saitama Medical University International Medical Center institutional review board (#12-096), and written informed consent was obtained from all patients. Details of patient characteristics are summarized in an online repository (21).
Tumor samples were minced and enzymatically dissociated with 1 mg/mL collagenase D (Roche, Mannheim, Germany) and 1 μg/mL DNase I (Roche) for 30 minutes at 37°C. Then, cells were filtered through a 100-μm cell strainer (Corning, Corning, NY) and centrifuged. The cells were then cultured on ultra-low-attachment culture dishes (Corning) and grown in StemPro hESC SFM‒Human Embryonic Stem Cell Culture Medium (Invitrogen, Carlsbad, CA) supplemented with 8.8 ng/mL of basic fibroblast growth factor (Invitrogen), 20 μmol/L of Rho-associated coiled-coil forming kinase inhibitor Y-27632 (Wako, Osaka, Japan), 50 U/mL of penicillin (Nacalai Tesque, Kyoto, Japan), and 50 μg/mL of streptomycin (Nacalai Tesque) at 37°C in a humidified atmosphere of 5% CO2 in air. Spheroid culture medium was changed every 2 or 3 days, and spheroid cells for serial passage were dispersed into single cells with Accumax (Innovative Cell Technologies, San Diego, CA). Patient-derived EnC cells were stably maintained in spheroid culture medium for >2 months.
Spheroid growth assay
Three thousand Accumax-dispersed single cells per well were seeded in ultra-low attachment 96-well plates (Corning). To examine the effect of ALDH inhibitors on spheroid growth, disulfiram (Sigma-Aldrich, St. Louis, MO), N,N-diethylaminobenzaldehyde (DEAB) (Sigma-Aldrich), or vehicle [dimethyl sulfoxide (Nacalai Tesque) for disulfiram or ethanol (Nacalai Tesque) for DEAB] was added to the medium at the indicated concentration (22). Alternatively, to assess the hormonal regulation of spheroid growth, the spheres were treated with 10 nM of 17β-estradiol (E) (Sigma-Aldrich), 10 nM of progesterone (P) (Sigma-Aldrich), or vehicle (ethanol) (Nacalai Tesque). Spheroid cell growth was evaluated using the CellTiter-Glo® 3D Cell Viability Assay kit (Promega, Madison, WI) using the TriStar2 S LB 942 Multimode Reader (Berthold Technologies, Bad Wildbad, German). Chemiluminescent values were normalized with the corresponding values at day 0.
Animal experiments in vivo
Established, stable cultures of patient-derived EnC cells were used for heterotransplantation in vivo. The spheroid cells were dispersed into single cells with Accumax, suspended in 150 μL of mixture containing 50% StemPro medium and 50% Matrigel (Corning), and subcutaneously injected into the flank of 10-week-old female NOD/Shi-scid, IL-2RγKO Jic (NOG) mice (In-Vivo Science, Tokyo, Japan) as a xenogenic host for testing tumorigenicity. Body weight and tumor size were measured once a week. Tumor volume was determined using the equation 0.5 × r1 × r2 × r3 (r1 < r2 < r3) by calipers. All animal experiments were approved by the Animal Care and Use Committee of Saitama Medical University and were conducted in accordance with the Guidelines and Regulations for the Care and Use of Experimental Animals by Saitama Medical University. NOG mice were kept in safety racks and maintained in a temperature-controlled room (23°C) with a 12-hour light/dark schedule and fed a standard diet (CE2; CLEA Japan, Tokyo, Japan), with free access to water.
Histological analysis
EnC specimens, PDCs solidified in iPGell (Genostaff, Tokyo, Japan), and PDX tumors were fixed in neutral buffered formalin and embedded in paraffin. The sections were deparaffinized, rehydrated, and stained with hematoxylin and eosin using a standard protocol.
Quantitative real-time PCR
Total RNA extracted from EnC specimens, PDCs, and PDXs was reverse-transcribed by SuperScript III (Invitrogen) to synthesize cDNA. Gene expression levels were examined by quantitative real-time PCR (qRT-PCR) using the cDNA with gene-specific primers (23). The sequences of PCR primers are described in an online repository (21). The comparison between the amounts of PCR product of the target gene relative to GAPDH control was carried out by the comparative cycle threshold method.
Microarray analysis and pathway analysis
Gene expression in the PDCs was examined using Affymetrix GeneChip (Human Clariom S Array), according to the manufacturer’s protocol. The microarray data have been deposited in the Gene Expression Omnibus under accession number GSE127238 (24). A global analysis of gene expression and differentially expressed genes in pathways and in clusters of functionally related genes was performed with the DAVID functional annotation clustering tool (https://david.ncifcrf.gov/) (25, 26).
Statistical analysis
Significance of differences between two groups was analyzed by the unpaired Student t test.
Results
Establishment of patient-derived EnC cells and their xenografts from clinical cancer specimens
Of 21 cases of EnC, PDCs were established from surgical specimens from three distinct patients, comprising two cases of endometrioid carcinoma (EnC-A and EnC-B) and one case of mixed carcinoma (serous and clear cell carcinomas, EnC-C) (21). The tumor specimens from subjects EnC-A and EnC-B exhibited characteristics of grade 3 tumor cancer (e.g., the cells were poorly differentiated), whereas EnC-A and EnC-B tumors showed characteristics corresponding to Federation Internationale de Gynecologie et d'Obstetrique or International Federation of Obstetrics and Gynaecology (FIGO) stages 1B (e.g., the tumor had spread to one-half or more of the myometrium) and 3A (e.g., the tumor had spread to the serosa of the uterus and/or to the tissue of the fallopian tubes and ovaries, but not to other parts of the body), respectively. The specimen from subject EnC-C had characteristics of a high-grade tumor and FIGO stage 3C1 (e.g., the cancer had spread to the regional pelvic lymph nodes).
EnC specimens were enzymatically digested and maintained in three-dimensional culture for more than 2 months, resulting in the establishment of long-term culturable EnC PDCs (Fig. 1A). Photographs of spheroid-forming EnC PDCs established from three distinct patients are shown in Fig. 1B. Next, the EnC PDCs were injected into the flanks of immunodeficient NOG mice to generate PDC-derived primary xenografts (Fig. 1A). Some of the primary xenografts were further implanted into NOG mice to generate secondary or tertiary xenografts. Thus, both primary xenografts and their derivative xenografts could be defined as “patient-derived xenograft (PDX)” models. Some of the secondary xenografts were again dissociated with enzymes and used to generate secondary PDCs.
Establishment of PDCs and PDXs in endometrial cancer. (A) Schematic representation of the strategy used for establishing PDCs and PDXs is shown. The samples, approved by the ethics committee at the Saitama Medical University International Medical Center, were used to establish PDCs and PDXs as described in the “Materials and Methods” section. (B) Representative images show the established endometrial cancer PDCs (EnC-A, EnC-B, and EnC-C). Scale bars, 100 μm.
Generation of EnC PDC-derived primary xenograft tumors and their secondary PDC spheroids
As PDC-derived PDX models, EnC PDCs established from spheroid culture were dissociated into 200,000 cells and subcutaneously injected into the flanks of 10-week-old NOG mice to develop xenograft tumors. The xenograft take rate of the implanted EnC-A PDCs into three distinct mice was 100% (three of three) (Fig. 2A‒2C). The volume of EnC-A PDC-derived primary xenograft tumors, or passage 1 (P1) of the EnC-A PDX model reached >500 mm3 ∼125 to 180 days after implantation (Fig. 2A). The P1 xenograft tumors were excised from the mice (Fig. 2C) and further used to generate secondary PDC spheroids (Fig. 2D). Passage 2 (P2) and passage 3 (P3) generations of PDX models in NOG mice could also be established by subcutaneous implantation of tumor excised from P1 and P2 xenografts with a size of 2 mm × 2 mm × 2 mm (21). PDC spheroids could also be established from P2 and P3 generations of EnC-A PDX models (21).
PDCs of EnC-A-developed xenograft tumors in severely immunodeficient NOG mice. (A) Tumor volumes generated by EnC-A PDCs in NOG mice are shown. The PDCs were injected subcutaneously into NOG mice at day 0, and their tumor volumes were measured once a week. P denotes the passage number of tumors in NOG mice. (B and C) Representative images show (B) tumor-bearing mice and (C) dissected tumors at the end point. (D) PDCs reestablished from PDXs of EnC-A in NOG mice are shown. Scale bars, 100 μm.
Compared with EnC-A xenograft tumors, EnC-B xenograft tumors exhibited faster growth in NOG mice, as the volume of EnC-B P1/P2 tumors reached >500 mm3 by ∼85 to 110 days after implantation (21). EnC-B P1/P2 xenograft tumors were excised from mice and further used to generate secondary PDC spheroids (21).
EnC-C xenograft tumors in NOG mice developed at the fastest rate compared with EnC-A and EnC-B tumors, and the volume of EnC-C P1‒P3 tumors reached >500 mm3 by ∼50 to 75 days after implantation (21). In the EnC-C #1 P2 model, liver and lung metastases developed in addition to the subcutaneous PDX tumor (21), and these metastatic tumor-derived PDCs could also be developed under spheroid culture conditions (21).
Pathologic features of original tumors, their corresponding PDCs, and PDC-derived PDX models
To characterize morphological features of the original EnC-A, EnC-B, and EnC-C tumors and their corresponding PDCs as well as the PDC-derived PDX models, hematoxylin and eosin staining was performed on paraffin-embedded, sliced pathologic sections (Fig. 3).
Histological analysis of PDCs and PDXs generated from endometrial cancers compared with their original tumors. Pathological characteristics of surgical specimens and their associated spheroids (PDCs) and PDC-originated xenografts (PDX tumors) were examined by hematoxylin and eosin staining of (A) EnC-A, (B) EnC-B, and (C) EnC-C tumors. Scale bars, 100 μm.
In general, PDC spheroids and PDX tumors morphologically resembled the original patient tumors. The original tumor from subject EnC-A featured characteristics of endometrioid carcinoma grade 3 tumor, such as poorly differentiated endometrioid carcinoma with small nonvillous papillae, including small intraluminal groups of cells with rather small pleomorphic nuclei and eosinophilic cytoplasm. EnC-A PDX tumors featured endometrioid carcinoma with a solid appearance and small nonvillous papillae. The original tumor from subject EnC-B featured endometrioid carcinoma grade 3 characteristics with poorly differentiated endometrioid carcinoma with a diffuse growth, including small cohesive polygonal cells and fibrovascular cores. EnC-B PDX tumors exhibited high-grade tumor features composed of polygonal epithelial-like cells with loosely cohesive and mitotic nuclei. The original tumor from subject EnC-C exhibited mixed epithelial carcinoma with histologic features of both serous and clear cell carcinomas, including a solid tumor or pseudoglands composed of atypical cells with either eosinophilic cytoplasm or clear cytoplasm. EnC-C PDX tumors exhibited mixed epithelial carcinoma properties with solid serous tumor, including focal clear cell components. Morphological features of the liver and lung metastatic sites were similar to those of the PDX tumors (Fig. 3C) (21).
Expression of CSC-related factors in EnCs
We previously showed that spheroid culture from primary cancers is favorable for the enrichment of CSCs (8). CSC-related markers such as CD44, CD133, and ALDH1A1 have been shown to be upregulated in PDC spheroids established from fresh ovarian cancer specimens. Thus, we examined the expression of CSC-related markers, including CD44, CD133, ALDH1A1, NANOG, SOX2, and OCT3/4, in original primary tumors, PDC spheroids, and PDC-derived xenograft tumors from EnC-A, EnC-B, and EnC-C (21). qRT-PCR analysis showed that elevation of CD44 levels was apparent in EnC-A/EnC-C PDC spheroids. CD133 expression was elevated in all the PDC spheroids. ALDH1A1 expression was also elevated in all the PDC spheroids, particularly in EnC-B PDC spheroids. The NANOG level in the EnC-B primary tumor was relatively high compared with NANOG levels in EnC-A/EnC-C primary tumors, whereas it was apparently elevated in EnC-A/EnC-C PDC spheroids. According to the expression study, these PDC spheroids exhibited CSC-like features.
Inhibition of ALDH activity suppressed PDC spheroid growth
Because ALDH1A1 levels were upregulated in the PDC spheroids from three distinct primary tumors and ALDH activity is required for the proliferation of cancer spheroid cells such as those from ovarian cancers (8), we next examined whether the inhibition of ALDH activity suppressed the growth of EnC PDCs (Fig. 4). ALDH inhibitor disulfiram completely blocked the proliferation of all the PDC spheroids at a concentration of 100 μM. EnC-A PDCs were particularly sensitive to disulfiram treatment compared with EnC-B/EnC-C PDCs, as 1 μM of disulfiram was sufficient to block growth. Another ALDH inhibitor, DEAB, completely blocked the proliferation of EnC-A/EnC-B PDC spheroids at a concentration of 500 μM. EnC-C PDCs were relatively resistant to DEAB treatment compared with other PDCs, as more than half of the activity remained at 100-μM DEAB concentration at day 7 after drug treatment.
Inhibition of PDC growth by ALDH inhibitors. PDCs were cultured in spheroid culture medium with the ALDH inhibitors disulfiram and DEAB at the indicated dosage for days 3 and 7. Dimethyl sulfoxide and ethanol were used as vehicle for disulfiram and DEAB, respectively. PDC growth was estimated using a CellTiter-Glo 3D Cell Viability Assay kit. Data were normalized with the corresponding values at day 0 and are presented as mean ± SEM (n = 5). *P < 0.05; **P < 0.01 using Student t test.
Expression of hormone receptors and hormone responsiveness in EnC tumors and PDCs
It is known that ∼70% to 80% of EnCs are defined as endometrioid carcinomas, which characteristically express ERα. Expression levels of the ERα-encoding gene ESR1 and its downstream target PGR were evaluated by qRT-PCR (21). All the original patient tumors significantly expressed both ESR1 and PGR mRNA to a similar extent. ESR1 mRNA was abundantly expressed in EnC-A PDC spheroids compared with the corresponding primary tumor, whereas the ESR1 mRNA level in EnC-B PDCs was lower than that in EnC-A PDCs. In contrast, ESR1 expression was almost abolished in EnC-C PDC spheroids and xenograft tumors, suggesting that ERα-negative EnC cells were predominantly selected in the population of EnC-C PDC spheroids. Because the spheroid culture medium is serum free, PGR mRNA levels were basically low in all the PDC spheroids. The PGR mRNA level was particularly elevated in the EnC-A xenograft tumor compared with the original primary tumor, indicating that ERα-enriched EnC-A PDCs could well respond to estrogen signaling in the PDX model.
With respect to hormone responsiveness, EnC PDC spheroids were treated with vehicle (V), E, or P at 10 nM or with both 17β-estradiol and progesterone (EP) for 48 hours (Fig. 5A and 5B). Of note, ESR1 levels were high and PGR expression was significantly induced by E or EP in EnC-A PDC spheroids. Although the ESR1 level in EnC-B PDC spheroids was lower than that in EnC-A PDC spheroids, PGR expression in EnC-B PDCs was slightly increased by E treatment. In the case of EnC-C PDCs, ESR1 expression was very limited or lacking, whereas PGR expression was slightly increased by E treatment.
Hormone responsiveness of PDCs. (A and B) qRT-PCR analysis was performed to detect expression levels of (A) ERα and (B) PGR in PDCs of EnC-A, EnC-B, and EnC-C treated with E, P, EP, or V. (C) Time course of PDC growth in EnC-A, EnC-B, and EnC-C under spheroid culture conditions by E, P, or EP is shown. PDC growth was estimated using a CellTiter-Glo 3D Cell Viability Assay kit. Data were normalized with the corresponding values at day 0 and are presented as mean ± SEM (n = 5). *P < 0.05; **P < 0.01 using Student t test.
Consistent with hormone responsiveness in EnC-A PDCs, spheroid viability of EnC-A PDCs was significantly increased at day 7 after E or EP treatment (Fig. 5C). The viability of EnC-B PDCs was also increased at day 7 after EP treatment, whereas the viability of EnC-C PDCs was not altered by hormone treatment.
Hormone-dependent alteration of expression of inflammation- and immune-related genes in estrogen receptor‒positive EnC-A PDCs
We further investigated the pathways that may be activated or repressed by hormone treatment in estrogen-responsive EnC-A PDCs. Expression microarray analysis showed that genes activated in the immune system‒related pathways, such as neutrophil chemotaxis, immune response, and inflammatory response, were enriched more than twofold by E treatment (21). In addition, genes in these pathways that were enriched by E treatment in EnC-A PDCs were downregulated when treated with a combination of E and P (21), indicating inflammation-repressive action by P in the EnC-A model.
To verify the estrogen/progestin regulation of these pathway genes in EnC-A PDCs, we focused on several genes and performed qRT-PCR using mRNA obtained from EnC-A PDCs treated either with hormones or with V (Fig. 6A‒6G). Substantial upregulation of mRNA expression was observed for IL6, IL1B, IL1RAP, IL18, CXCL2, CXCL8, and CD274/PD-L1 in EnC-A PDCs treated with E vs V. Except for IL1RAP, E-induced upregulation of all these genes was impaired significantly by EP treatment. Considering the rather moderate effects of P alone on the repression of these genes (Fig. 6A‒6G), we assumed that estrogen-induced expression of PGR would be required for P treatment to be completely responsive.
Hormonal regulation of inflammation- and immune-related genes in EnC-A PDCs. EnC-A PDCs were treated with E, P, EP, or V. qRT-PCR analysis was performed to examine expression levels of inflammation- and immune-related genes: (A) IL6, (B) IL1B, (C) IL1RAP, (D) IL18, (E) CXCL2, (F) CXCL8, and (G) CD274/PD-L1. Data are presented as mean ± SEM (n = 3). *P < 0.05; **P < 0.01 using Student t test. (H) Model for regulation and function of inflammation- and immune-related genes in hormone-responsive cancer stem-like spheroid cells in endometrial cancer is shown. EnC-A PDCs are useful as a cancer model to understand hormone responsiveness of endometrial cancers.
Metastasis-associated regulation of inflammation- and immune-related genes in EnC PDCs
To analyze the metastasis-associated genes in EnC-C PDCs, RNA was extracted from the secondary PDCs established from liver and lung metastatic tumors and from primary EnC-C PDCs. qRT-PCR analysis showed that the expression levels of IL6, IL1B, and CD274/PD-L1, whose expressions were increased by E in EnC-A PDCs, were also upregulated in secondary PDCs from liver and lung metastatic tumors compared with the original PDCs (21).
Discussion
The hallmark of this study is the establishment of EnC PDC spheroids and their xenograft models with characteristics similar to CSC traits and distinct characteristics with different gene signatures such as hormone responsiveness in distinct models of EnC-A/EnC-B and EnC-C, respectively. Although EnC PDX models have been developed directly from patient tumor tissue pieces (27), PDC-derived xenograft models may have an advantage in terms of stable success rate after establishment of PDCs. From the viewpoint of the cancer microenvironment, tissue piece‒derived tumors have the advantage that they can form tumors including cells originated from the patient microenvironment, such as stromal cells and immune cells. Nevertheless, it is sometimes difficult to generate cultured cells from direct PDX tumors. In contrast, spheroid cell‒derived tumors have an advantage in that the conversion between spheroid cells and xenograft tumors is usually easy, so the system can be applied to both in vitro and in vivo studies. Therefore, compared with direct PDX models that need serial transplantation into immunodeficient mice, PDC-derived xenograft models are time- and space-saving when studying aggressive cancers with CSC traits.
Regarding CSC traits in EnC PDCs, ALDH1A1 expression level is particularly enriched in the models compared with their original patient tumors. ALDH1 isotypes are ALDH enzymes that catalyze the oxidation of endogenous and exogenous aldehyde substrates and are known markers of normal stem cells and CSCs. ALDH1A1 especially catalyzed the oxidation of retinaldehyde to retinoic acid, which functions as a ligand for retinoic acid receptors and retinoic X receptors, leading to the regulation of gene expression related to the development and differentiation of various organs. High expression of ALDH1A1 is often associated with poor prognosis in patients with cancer, and its increased activity drives the proliferation of various types of tumors including liver, lung, breast, pancreas, and gastrointestinal cancers (28). Considering the pivotal role of ALDH1A1 in CSC pathophysiology, it is notable that the ALDH inhibitors disulfiram and DEAB could inhibit the growth of spheroids of EnC PDCs established in the current study. In a study using ALDH-enriched cancer cell line subpopulations with high tumorigenic abilities, two widely studied small-molecular-weight compounds, CP-31398 and PRIMA-1, which are p53 modulators, could reduce the growth and sphere formation abilities of high ALDH-expressing EnC cells from the Ishikawa and AN3 CA cell lines (29), suggesting the possible interaction of p53 and ALDH signaling in CSCs.
With respect to other CSC markers, CD133 levels were substantially higher in PDC spheroids than in their original patient tumors or xenograft tumors, although the levels differed among the three cases. CD44 levels were higher in EnC-A and EnC-C PDCs than in their original patient tumors or xenograft tumors. NANOG levels were higher in EnC-A and EnC-C PDC-derived xenograft tumors than in their original patient tumors or PDCs, whereas the basal NANOG level in EnC-B patient tumors was relatively high compared with levels in other patient tumors. The ability of CD44-negative/CD44 low-expressing colon cancer PDCs to form CD44 high-expressing tumors in mice has been reported (12), suggesting the existence of dynamic transitions between CSC and non-CSC states that could be influenced by factors in tumor microenvironments, such as alterations of the cytoskeleton exerted by the ROCK inhibitor Y27632.
Despite similar CSC behavior exhibited by PDC spheroids, the pathophysiological diversity of each PDC model and its xenograft model was demonstrated. The EnC-A model was unique for its high hormone responsiveness with abundant expression of ERα and PGR as well as E-inducible expression of inflammation- and immune-related genes. The EnC-B model exhibited mild E/P responsiveness in the PDC spheroids. Considering that E plus P could substantially promote the proliferation of hormone-naive EnC-A and EnC-B, the supplementation of hormones may improve the take rates of PDC spheroids. The EnC-C model is derived from mixed adenocarcinoma, including serous and clear cell components, and EnC-C PDCs could develop lung and liver metastatic lesions.
Chronic inflammation is regulated by various inflammatory cytokines and is thought to stimulate initiation/progression of cancers. As in other cancers, a high expression level of IL6 is known to correlate with poor survival of patients with EnC (30). In EnC cells, IL6 promotes an autocrine regulatory loop to activate the ERK‒nuclear factor κB pathway (31). Interestingly, it is reported that IL6 stimulates sphere formation, self-renewal, and stem-like characteristics in several cancer cells (32). In addition, IL6 receptor subunits are increased in endometrial CSCs, which exhibit a high level of ALDH activity, and the blockade of IL6 receptor reduced cancer cell growth (33). Several lines of evidence also showed that IL-1 and its regulation play an important role in cancer-associated inflammation (34). In colon cancer CSCs, IL1B increases spheroid formation in serum-free medium and upregulates the expression of stemness-related genes and epithelial-to-mesenchymal transition (35). IL18 is a member of the IL1 family and functions as a proinflammatory cytokine (36). Because the immune-stimulating effect of IL18 is thought to exert an anticancer effect, IL18 has been proposed as a novel adjuvant treatment against cancer. However, serum IL18 levels are increased in patients with several types of cancer and is associated with disease progression and poor clinical survival, suggesting a pathophysiological role of IL18 in cancer (37).
In the current study, we demonstrated that E stimulated the growth of EnC-A PDCs with higher ERα expression along with the upregulation of inflammation- and immune-related genes, including IL6, IL1B, and IL18, suggesting a tumor-promoting mechanism of estrogen in estrogen-responsive EnC (Fig. 6H). Interestingly, the E-mediated upregulation of IL6, IL1B, and IL18 in EnC-A PDCs was repressed by P, suggesting the anti-inflammatory function of P in endometrial CSCs. It is noteworthy that progestin opposes estrogen-driven growth in the endometrium and that reduced progestin action is the risk of EnC (38). Interestingly, progestin is used in the treatment of EnC, mainly for fertility-preserving management. A positive feedback loop of estrogen signaling has been reported in EnC cells whereby estrogen-induced IL6 production activates aromatase expression in stromal cells, suggesting that hormone signaling could modulate the cancer microenvironment that is favorable for cancer survival (39).
Interestingly, these inflammatory cytokines were also upregulated in secondary EnC-C PDCs prepared from liver and lung metastatic lesions compared with the primary EnC-C PDCs. It is tempting to speculate that inflammatory cytokines may support metastasis of endometrial CSCs (21).
Meanwhile, the immunomodulating role of estrogen remains an unresolved paradox (40). Estrogen at high concentrations is assumed to suppress bone resorption and the production of inflammatory cytokines from immune cells, and it also induces regulatory T cells, which act to repress immune response and inflammation. Therefore, estrogen may suppress inflammation by inhibiting cellular immune reactions. On the other hand, estrogen can promote B cell activation or antibody production. In agreement with these findings, during pregnancy, when estrogen persists at high levels, multiple sclerosis and rheumatoid arthritis, which are presumed to be predominantly inflamed by cellular immunity, may be relieved. However, systemic lupus erythematosus, in which various autoantibodies play a role in the diseased state, may be exacerbated in pregnancy. In addition, Mohammad et al. (41) recently showed in a mouse model that ERα is involved in the development of autoimmune T cell responses. With respect to transcriptional regulation of IL6, in human osteoblastic cells, IL6 promoter is inhibited by estrogen through the downregulation of nuclear factor κB and C/EBP (42). Further study is required to understand the mechanism of estrogen-induced expression of IL6 in EnC PDCs.
CD274/PD-L1 is a transmembrane protein that is assumed to play a role in suppressing the immune system. In cancer cells, increased PD-L1 expression suppresses T cell activity, promoting immune escape and hence promoting tumor survival (43). However, the prognostic significance of PD-L1 remains controversial and may depend on cancer types (44). Tumor PD-L1 expression level is positively correlated with the clonality of infiltrated T cells, and a high PD-L1 and high clonal T cell expression phenotype provides favorable prognostic information in patients with EnC (45). It has been reported that PD-L1 expression correlates with microsatellite instability status and high grade in EnCs (46). In addition, immunohistochemical analysis of EnCs indicated that immunonegativity for PD-L1 was significantly higher in FIGO stage I than in the sum of the other stages (47). Regarding CSCs, increased expression of PD-L1 was found in multiple cancer types, including breast, colon, and lung cancers, and is thought to contribute to stemness, epithelial-to-mesenchymal transition, and chemoresistance (48, 49). In breast cancer, PD-L1 promotes OCT4 and NANOG expression and self-renewal capability of CSCs (50). Here, we demonstrated that PD-L1 expression was enhanced in EnC-A PDCs after E treatment as well as in secondary EnC-C PDCs from metastatic tumors compared with primary PDCs. Therefore, some endometrial CSCs may modulate the tumor immune microenvironment by hormones (Fig. 6H) or metastasis (21) in concert with inflammatory cytokines.
In conclusion, our PDC and PDX models established from EnCs would be useful in understanding the hormonal regulation and metastasis of CSCs and may help develop alternative diagnostic and therapeutic options for patients with advanced disease.
Acknowledgments
We thank W. Sato, N. Sasaki, T. Suzuki, A. Iwasa, S. Shimoyokkaichi, and Y. Ijima for their technical assistance.
Financial Support: This work was partially supported by a grant of the Support Project of the Strategic Research Center in Private Universities from the Ministry of Education, Culture, Sports, Science, and Technology, Japan (to S.I.); by the Practical Research for Innovative Cancer Control [JP18ck0106194 (to K.I.)] and the Project for Cancer Research and Therapeutic Evolution [P-CREATE, JP18cm0106144 (to S.I.)] from the Japan Agency for Medical Research and Development; by grants from the Japan Society for the Promotion of Science, Japan [16K09809 (to K.I.); 16K15496 (to K.H.-I.); 17H04205 (to K.H.-I.); 17K16170 (to S.S.)]; and by the Takeda Science Foundation (to S.I.).
Author Contributions: S.S. and S.I.: conception and design; S.S. and K.O.: development of methodology; T.S., D.S., and K.H.: acquisition of data; K.I. and K.H.-I.: analysis and interpretation of data; S.S., K.I., T.S., D.S., K.O., K.H.-I., K.H., and S.I.: writing and reviewing of the manuscript.
Additional Information
Disclosure Summary: The authors have nothing to disclose.
Data Availability: All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.
Abbreviations:
- ALDH
aldehyde dehydrogenase
- CSC
cancer stem-like cell
- DEAB
disulfiram
- N,N-diethylaminobenzaldehyde
E
- 17β-estradiol
EnC
- endometrial cancer
EP
- 17β-estradiol and progesterone
ERα
- estrogen receptor α
FIGO
- Federation Internationale de Gynecologie et d'Obstetrique or International Federation of Obstetrics and Gynaecology
NOG
- NOD/Shi-scid
IL-2RγKO Jic
- P
progesterone
- P1
passage 1
- P2
passage 2
- P3
passage 3
- PDC
patient-derived cancer cell
- PDX
patient-derived xenograft
- PGR
progesterone receptor
- qRT-PCR
quantitative real-time PCR
- V
vehicle
