Immunochemical analysis of iron transporters and M2 macrophages in ovarian endometrioma and clear cell adenocarcinoma

Endometriosis is a pathology of the female reproductive tract ( 1 ), with a morbidity rate of ~10% in women of reproductive age ( 2 ). Ovarian endometriomas (OE) are clinical manifestations of endometriosis and comprise a risk factor for the development of ovarian clear cell adenocarcinoma (CCC) and endometrioid carcinoma, regarded as endometriosis-associated ovarian cancers (EAOC) ( 3 , 4 ). The primary histological type of EAOC in Asia is CCC, accounting for 20% of all ovarian cancers ( 5 ). The prevalence of EAOC will likely increase along with that of endometriosis; however, the underlying mechanisms associated with carcinogenesis are controversial. Hence, it is important to elucidate the mechanisms underlying malignant transformation and pathogenesis of EAOC. Endometriosis is caused due to the reflux of menstrual blood containing endometrial tissues ( 6 ), and OE form when menstrual blood flows into the ovary ( 6 ), carrying endometrial epithelium and endometrial stroma into the ovary ( 7 ). Moreover, iron within the menstrual blood serves as an etiological factor for the development of ovarian cancer from OE ( 8-10 ). However, the precise mechanism through which iron ions flow in endometrial epithelial cells remains unknown. Iron ions are primarily absorbed by the epithelium of the intestinal tract ( 11 ), after which iron is transported by divalent metal transporter1 (DMT1) that imports and ferroportin (FPN) that exports iron ions ( 12 , 13 ). Furthermore, transferrin receptors (TfR) uptake iron ions in other organs ( 14 ). These transporters regulate intracellular iron concentrations ( 11 ). Meanwhile, iron transporters in OE and CCC have rarely been investigated, and the mechanism of iron flow through endometrial epithelial cells of OE or CCC tumor cells remains unclear. Macrophages within the endometrial stroma of OE accumulate around the epithelium to englobe iron ( 7 ). The M1 macrophage phenotype induces inflammatory cytokines and bactericidal activity ( 15 ), and M2 macrophages polarize toward tumor-associated macrophages (TAMs) after T cells and tumor cells release cytokines such as IL-4, IL-13, and macrophage colony-stimulating factor (M-CSF) ( 15 ). Thereafter, TAMs migrate to hypoxic regions, where they release growth factors and angiogenetic factors that promote proliferation and metastasis ( 16 ). Although M2 macrophages have been detected in OE and CCC ( 17 , 18 ), their effect on OE carcinogenesis has not been investigated. The present study aimed to immunohistochemically determine the expression of iron transporters in the endometrial epithelium of OE and tumor cells of CCC. We also evaluated the relation of macrophages to endometrial epithelial carcinogenesis in OE.

Table I lists the clinical features of the 18 CCC patients and 20 OE patients. The average age of the CCC (56 years) and OE groups (47 years) differed significantly (P=0.0105). Most CCC patients had a stage I disease, according to the International Federation of Gynecology and Obstetrics (FIGO) staging guidelines. Of the 18 CCC patients, seven had a medical history of OE, two had a histopathological diagnosis of OE, one had both medical history and histopathological diagnosis of OE, and the medical history of OE in eight patients was unknown. The maximal tumor diameter was significantly larger in CCC than that in OE (P<0.001). We initially investigated iron localization in OE and CCC by staining tissue sections with Prussian blue ( Fig. 1 ). Iron was deposited in the endometrial epithelium ( Fig. 1A arrow) of four OE samples and englobed macrophages in the stroma just below the epithelium ( Fig. 1A ) of all OE samples. Iron was deposited throughout the stroma in all OE specimens ( Table I ). Stroma located beneath the aggregated englobed macrophages (referred to herein as ‘deep layer’; Fig. 1B , region within the dotted line) also contained iron. Iron was deposited in CCC cancer cells and stroma in 13 specimens ( Fig. 1C and Table I ). Considering that iron deposition was observed in both OE and CCC, we investigated the expression of iron transport proteins DMT1, TfR, and FPN by immunostaining ( Fig. 1D ). All specimens of OE epithelium stained positive for these proteins. Specifically, DMT1 was scored 3 (intense staining) in 16 ( Fig. 1E ), TfR was scored 1 (light staining) in 12 ( Fig. 1F ), and FPN was scored 2 (moderate staining) in 11 ( Fig. 1G ) OE specimens. All three transporters were expressed at high levels in CCC. Comparing the expression of the iron transport proteins in OE and CCC specimens, we observed no remarkable difference in DMT1 expression ( Fig. 1E ), whereas expression of TfR and FPN was significantly higher (P<0.001) in CCC than in OE ( Fig. 1F and G ). Fig. 1 indicates that iron ions flowed into and drained from cells. Iron was deposited not only in epithelial cells but also in the stroma of OE. Although Prussian blue staining revealed iron deposition in macrophages located immediately below the epithelium, it was difficult to ascertain whether iron ions in the entire stroma were englobed by macrophages. Hence, we examined the distribution of macrophages in the stroma to determine if they englobed iron ions as well. We detected the pan-macrophage marker CD68 in the deep stroma layer, and the iron englobed macrophages below the epithelium ( Fig. 2A , CD68). Immunohistochemical staining for CD11c (M1 macrophage marker), CD163, and CD206 (M2 macrophage markers) markers revealed the presence of CD11c + , CD163 + , and CD206 + cells immediately just below the OE epithelium, and only M2 macrophages in the deep layer of the stroma ( Fig. 2A , region within the dotted line). Double staining for iron and CD163 revealed the presence of M2 macrophages englobed iron ions in the deep layer of the OE stroma ( Fig. 2A arrow). We located significantly more CD163 + and CD206 + cells, than CD11c + cells in the deep layer of OE ( Fig. 2C , P<0.001). CD68 + cells, along with conspicuous staining of CD163 + and CD206 + cells, were observed in the stroma of CCC tumors ( Fig. 2B ). We also observed macrophages with englobed iron ( Fig. 2B arrow) and far more abundant M2 than M1 macrophages in CCC specimens ( Fig. 2D , P<0.001). There were no statistically significant differences in the numbers of cells positive for CD11c, CD163, and CD206 between OE and CCC (data not shown). We postulated that the macrophages with englobed iron in the deep layer of the OE stroma resulted from endometrial epithelium infiltration of the OE stroma. We further hypothesized that M2 macrophages colonized the epithelium and invaded the stroma. The endometrial epithelium forms rows in typical OE ( Fig. 3A ). Fig. 2 demonstrates more abundant M2 than M1 macrophages in the deep layers of the stroma ( Fig. 3A ; CD68, CD11c, CD163, and CD206 are shown in Fig. 2A ). The positive rate of Ki67 (tumor growth marker) was low in the endometrial epithelium (4.6%±1.6, Fig. 4A ). Staining for the epithelial cell marker keratin revealed the regions where the endometrium epithelium had infiltrated the stroma and had formed lumens in one OE specimen ( Fig. 3B : Patient no. 1, region within the black square in the H&E staining figure) and in two CCC specimens with OE adjacent to CCC ( Fig. 3C-D : Patient no.2-3, region within the black square in the H&E staining figure). These regions contained more abundant CD163 + and CD206 + cells than CD11c + cells ( Figs. 3B-D and S1 , and Ki67 positive rate of 14.3-38.5%; Patient no.1: 38.5±6.0%, Patient no. 2: 36.1±3.7%, Patient no. 3: 14.3±2.4% was observed in the endometrium epithelium infiltrating the stroma ( Fig. 4B-D ). Endometrium, as well as ovarian follicles, stained positive for keratin. To avoid misinterpreting these findings as normal, we distinguished endometrial epithelium from ovarian follicles using CD10, a diagnostic marker of endometriosis ( 19 ). The endometrial stroma was positive, while the follicular epithelium was negative for CD10 ( Fig. S2 ).

In the present study, we detected iron transporters in OE and CCC using immunohistochemistry and identified M2 macrophages englobed with iron throughout the stroma of OE. The endometrial epithelium infiltrating the stroma was highly proliferative. Although the precise mechanisms associated with the carcinogenesis of EAOC remain under investigation, the co-occurrence of the AT-rich interaction domain 1A (ARID1A) and the phosphatidylinositol-4, 5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) mutation promotes CCC growth ( 20-23 ). Moreover, an iron-rich environment contributes to cell proliferation in OE, with oxidative stress and reactive oxygen species implicated in DNA damage ( 8 , 10 , 23 , 24 ). Thus, these findings suggest a relationship between iron and OE tumorigenesis. However, the mechanisms through which endometrial epithelium and tumor cells transport iron ions have remained unknown. Therefore, we aimed to demonstrate iron flow using immunohistochemistry. The transmembrane protein for ferric metal iron, DMT1, plays a major role in the intestinal uptake of iron ions ( 11 ). This protein is located not only on the cell surface but also in endosomes ( 12 ), with TfR also located on the cell surface. Because transferrin-binding trivalent iron molecules bind TfR, the complex becomes internalized in cells via endocytosis ( 12 ). Alternatively, FPN is the only transporter that exports iron ions in humans ( 25 ). The iron-responsive element (IRE) and iron regulatory proteins regulate DMT1, TfR, and FPN expression ( 26 , 27 ). Hence, iron metabolism is strictly controlled by these mechanisms. This control system collapses in cancer, and the iron overload associated with cancer is largely due to iron dysregulation ( 28 ). Specifically, DMT1 expression is upregulated in colorectal cancer as compared to healthy colons ( 29 ). Furthermore, the expression of TfR and FPN is high and low in high-grade serous ovarian carcinoma, respectively, implying storage of excess iron ions in cells within this carcinoma type ( 30 ). The present findings indicate the upregulation of DMT1 and downregulation of TfR and FPN expression in OE. Moreover, the levels of the iron transporters differed in CCC tissues. We also verified that iron ions flow in and out of the endometrial epithelium, indicating that the differential abundance of transporters is associated with intercellular iron concentrations, which participate in carcinogenesis. Excessive iron was detected not only in the endometrial epithelium but also in the stroma of OE, and M2 macrophages englobed iron ions in the stroma. Macrophages polarized toward the M2 phenotype generally produce anti-inflammatory cytokines and participate in tissue repair and angiogenesis ( 15 ), whereas M1 macrophages induce inflammatory cytokines and bactericidal activity. Therefore, M2 macrophages are regarded as TAMs ( 15 ). These TAMs have been detected within ascites from patients with ovarian cancer, where interaction between tumor cells and M2 macrophages induced tumor progression via signal transducer and activator of transcription 3 (STAT3) ( 31 ). In agreement with these findings, we detected more M2 than M1 macrophages in the stroma of OE and CCC. Furthermore, we identified M2 macrophages englobed with iron in the deep layer of OE stroma, which became a central focus of the study. Considering that ovarian cancer is an epithelial cancer, these M2 macrophages englobed with iron likely influence epithelial carcinogenesis by infiltrating endometrial epithelium in the stroma. The endometrial epithelium OE typically forms smooth rows of cells ( 7 , 32 ). However, we demonstrate that endometrial epithelium infiltrates and forms lumens within the stroma. Moreover, the infiltrating epithelium was highly positive for Ki67 compared with normal endometrial epithelium. This indicates that the endometrial epithelium has potent proliferative capacity, which is a state closer to cancer. The infiltrative endometrial epithelium adjacent to CCC is likely to be a precancerous lesion. However, infiltrative endometrial epithelium in OE also has potential carcinogenic capacity, as benign OE has loss-of-function mutations in ARID1A like in CCC ( 33 , 34 ). Considering that OE harbors ARID1A loss-of-function mutations, M2 macrophages in these regions might accelerate carcinogenesis of the endometrial epithelium infiltrating the stroma. We suggest that OE carcinogenesis occurs not only in endometrial cysts but also in epithelium infiltrating the stroma. Studies in a larger patient cohort are needed to verify these findings. In a preliminary study, we performed an experiment using an ovarian clear cell adenocarcinoma cell line and identified that TfR and DMT1 could be detected by western blotting. We thought that we should conduct additional experiments using ovarian endometrioma, ovarian clear cell adenocarcinoma, and normal ovary tissues as the study's primary aim was to compare between ovarian endometrioma and ovarian clear cell adenocarcinoma. Due to the lack of ovarian endometrioma and normal ovary cell lines, we intend further explore the role of iron transporters using an animal model, where we will be able to access more tissue samples and perform western blots. Nevertheless, our results provide insights into a potential underlying mechanism of carcinogenesis from OE to CCC while elucidating tumor growth features. Confirming these findings using pathological specimens, we could recognize the possibility of them being precancerous lesions. Moreover, careful follow-up is needed for the prevention and early detection of CCC. Further investigations of the mechanisms responsible for the infiltration of the endometrial epithelium into the stroma are warranted. In summary, the present study discovered iron transport proteins in the epithelium of OE and CCC tumor cells. We also revealed M2 macrophages englobed with iron in the stroma of OE and CCC. Epithelium invading the stroma of OE implies carcinogenesis under the influence of M2 macrophages englobed with iron. These findings offer a new perspective on the malignant transformation of OE into EAOC, which may serve as a precancerous index factor.

The study was approved by the Institutional Review Board of Tokyo Women's Medical University Hospital (Approval number 5371). We retrospectively analyzed 20 OE tissues without malignant lesions and 18 ovarian CCC that were surgically excised at Tokyo Women's Medical University Hospital between January 2014 and December 2019. Opt-out informed consent was obtained from all patients. Tissues were stained with Perls Prussian blue using potassium hexacyanoferrate (II) trihydrate (Wako Pure Chemical Industries, Osaka, Japan) and enhanced with Nuclear Fast Red (Cosmo Bio Co, Tokyo, Japan). Paraffin-embedded tissue sections (4 µm) were deparaffinized for 10 min. Then the sections were stained with H&E according to routine hospital procedures. For immunohistochemistry, antigen was retrieved by autoclaving for 20 min or microwave heating for 15 min in 10 mM sodium citrate buffer, pH 6.0 or 9.0. Nonspecific protein binding in tissue sections was blocked using Protein Block Serum-Free (Dako A/S, Glostrup, Denmark) at 25˚C, for 20 min. The sections were incubated overnight at 4˚C with primary antibodies, followed by secondary antibodies (Dako) for 30 min at 25˚C, and finally with 3,3'-diaminobenzidine (Nichirei Biosciences, Tokyo, Japan) to visualize antigen binding. Nuclei were stained with hematoxylin (Wako Pure Chemical Industries). Primary antibodies against the following were used: DMT1 (1:800), TfR (1:500), CD68 (1:400), CD11c (1:500), CD163 (1:500), and CD206 (1:5,000) (all from Abcam, Cambridge, United Kingdom), FPN (1:400; Novus Biologicals, Centennial, CO, USA), Ki67 (1:100) and cytokeratin (1:200) (both from Dako), and CD10 (1: 50; Leica Biosystems, Wetzlar, Germany). Samples prepared in the same manner but without incubation with the primary antibodies were used as negative controls (data not shown). The intensity of DMT1, TfR, and FPN staining was scored using an arbitrary scale as 0, absent; 1, light; 2, moderate; 3, intense. We used the score that was the most prevalent in 10 fields. Due to endometrial epithelial detachment, 10 fields could not be evaluated in three OE samples, and hence, these samples were excluded from the analysis. We evaluated CD11c, CD163, and CD206 expression as the average number of antibody-positive cells counted in 10 fields at x400 magnification. We evaluated ratios (%) of Ki67-positive nuclei and averaged the numbers of cells counted at x400 magnification in 10 fields. Two pathologists assessed the immunohistochemical scores and numbers of positive cells in all sections. Data were statistically analyzed using JMP ® 14 (SAS Institute Inc., Cary, NC, USA). Continuous variables are presented as mean ± standard deviation. Categorical variables (Nulliparous and Premenopause) are presented as numbers (percentage) and compared with χ 2 test or Fisher's exact test. Shapiro-Wilk test was performed to determine if the continuous variables were normally distributed. The levels of CA125 are expressed as the median with interquartile range (IQR) as they were not normally distributed. Comparison of two groups with normally distributed data (mean age and maximal cyst/tumor diameter) was performed using Welch's t-test; otherwise, comparison of continuous variables (median CA125) was performed using the non-parametric Wilcoxon rank sum test. Two groups and more than two groups were compared using the non-parametric Wilcoxon rank sum test and Steel-Dwass test (used after Kruskal-Wallis test), respectively, to confirm the significance of iron transport proteins and the number of positively immunostained cells. Statistical significance was set at P<0.05.