Galectin-3 as a Candidate Marker of Endocrine-modulatorassociated Endometrial Remodeling During Breast Cancer Endocrine Therapy

Abstract

Background/Aim

Breast cancer endocrine therapies may induce endometrial alterations, but the endometrial effects of antiprogestins remain incompletely defined. This study evaluated the effects of tamoxifen, mifepristone, and ulipristal acetate on endometrial morphology and galectin-3 expression, with the aim of exploring galectin-3 as a candidate marker of endocrine-modulator-associated endometrial remodeling.

Materials and Methods

Endometrial effects were examined in a rat mammary carcinogenesis model treated with tamoxifen, mifepristone, ulipristal acetate, or vehicle control. Uterine tissues were assessed by histology and galectin-3 immunohistochemistry. Complementary in vitro experiments in Ishikawa endometrial adenocarcinoma cells evaluated agent- and dose-associated galectin-3 protein-expression patterns by western blotting. Histological outcomes and in vitro fold-change values were summarized descriptively.

Results

In the analyzed cohort, tamoxifen-treated animals showed the highest descriptive proportion of complex endometrial hyperplasia, whereas mifepristone and ulipristal acetate were associated predominantly with simple hyperplastic changes. Normal endometrial morphology was retained in selected mifepristone- and ulipristal acetate-treated animals and in all control animals. Galectin-3 immunoreactivity was low but detectable in morphologically normal endometrium, whereas hyperplastic lesions showed no detectable galectin-3 immunoreactivity under the staining conditions used. In Ishikawa cells, galectin-3 expression showed agent-specific fold-change patterns: tamoxifen and mifepristone were associated with reduced expression at higher concentrations, while ulipristal acetate showed a non-monotonic pattern with increased expression at low and intermediate concentrations and return toward baseline at the highest concentration.

Conclusion

In this exploratory model, endocrine-modulator-associated hyperplastic endometrial remodeling was accompanied by reduced or undetectable galectin-3 immunoreactivity, while low-level expression was retained in morphologically normal endometrium. These findings support further investigation of galectin-3 as a candidate marker of endocrine-associated endometrial remodeling, but validation in larger animal cohorts and human endometrial samples from treated patients is required.

Introduction

Hormone therapy plays an important role in the targeted management of hormone-dependent breast cancer, particularly in tumors that express estrogen (ER+) and/or progesterone receptors (PR+) (1). The aim of endocrine therapy is to block or modulate hormone-driven signaling pathways that induce the growth and survival of tumors (2). The main therapeutic options include selective estrogen receptor modulators (SERMs) such as tamoxifen (3), aromatase inhibitors (AIs) such as anastrozole, letrozole and exemestane, which reduce the natural estrogen production from the ovaries especially in postmenopausal women (4). The duration of treatment typically ranges from 5 to 10 years, depending on disease progression, risk of recurrence, and menopausal status (5). The choice of regimen is determined by the patient’s age, receptor profile and overall prognosis with the aim of minimizing the risk of recurrence and to improve survival (6).

Galectin-3 is a β-galactoside-binding lectin involved in many processes that contribute to endocrine-related carcinogenesis, including cell adhesion, extracellular matrix remodeling, apoptosis regulation, inflammation and epithelial-stromal communication. It is natively expressed in the endometrium and is implicated in endometrial hyperplasia, carcinogenesis and other gynecological conditions like endometriosis (7, 8). These observations suggest that galectin-3 could be tracking alterations in tissue architecture and cellular survival pathways in hormonally responsive tissues. Consequently, galectin-3 might represent a plausible biologic marker for monitoring how endocrine therapy might be influencing endometrial homeostasis and identifying early changes in the tissue that can lead to adverse endometrial outcomes.

While tamoxifen is predominantly an estrogen antagonist in breast tissue, it has partial agonist activity in the endometrium resulting in gynecological complications, especially in post-menopausal women (9). The risk of endometrial hyperplasia (10), polyps, uterine abnormalities and endometrial carcinoma is higher in those receiving tamoxifen long term, although it remains a low but clinically significant incidence (11). The risk of endometrial cancer remains low overall but becomes significant after more than five years of exposure (12). On the contrary, aromatase inhibitors cause endometrial atrophy and are not associated with endometrial neoplasia but are often complicated by musculoskeletal adverse effects (13). It is known that tamoxifen binds to estrogen receptors in the breast and in the endometrium, but the outcome of the binding is dependent on the cellular environment in which the interaction occurs. In breast tissue, the tamoxifen-estrogen receptor complex recruits co-repressor proteins (14), thereby suppressing transcription of estrogen-dependent genes and promoting an anti-estrogenic environment. In contrast, within the endometrium, the same receptor-tamoxifen complex preferentially mobilizes co-activator proteins, which stimulate the transcription of estrogen-responsive genes (15). This shift toward co-activator driven signaling results in a relative estrogenic milieu in the endometrium, contributing to proliferative effects and supporting the development of hyperplastic or neoplastic changes during prolonged exposure.

Antiprogestins such as mifepristone and ulipristal acetate have been evaluated as potential treatments for hormone-dependent breast cancer since they inhibit the action of progesterone in PR+ cells (16), which drives growth through proliferation and inhibition of apoptosis (especially in ER+/PR+ tumors) (17). Ulipristal acetate is a selective progesterone receptor modulator (SPRM) with unique tissue-dependent properties of agonism and antagonism, and it is used in the management of uterine fibroids. Based on animal studies, this compound may exhibit antiproliferative effects in the breast, but its breast cancer therapy use is still exploratory (18). Antiprogestins may therefore complement or, in selected settings, potentially replace standard endocrine regimens but data about the clinical safety of these treatments remain insufficient in the long term, and especially for the endometrium.

On this basis, the present study investigates the comparative effects of tamoxifen, mifepristone, and ulipristal acetate on endometrial morphology and galectin-3 expression in vivo and in vitro, with the aim of exploring galectin-3 as a candidate marker of endocrine-modulator-associated endometrial remodeling. Histological outcomes were interpreted descriptively, while complementary in vitro analyses were used to examine exploratory agent and dose associated changes in galectin-3 protein expression.

Materials and Methods

Animals and conditions. To determine the number of female Wistar rats required for the study, an a-priori power analysis was performed using the G*Power software (version 3.1.9.8; Heinrich Heine University Düsseldorf, Düsseldorf, North Rhine-Westphalia, Germany) for a fixed-effect one-way ANOVA comparing four experimental groups. The calculation was based on an alpha level of 0.05, a desired statistical power of 0.80, and an expected large effect size corresponding to approximately Cohen’s f=0.53. This effect size was considered appropriate for detecting biologically meaningful differences among the experimental groups. The power analysis indicated that a minimum total sample size of 43 animals was required. To compensate for an anticipated attrition rate of approximately 10%, mainly related to acclimatization difficulties, perioperative complications, or humane endpoint criteria, 48 animals were initially procured. During the acclimatization period, five animals were excluded because of persistent feeding difficulties and failure to establish stable food intake despite of close monitoring. These animals were therefore considered unsuitable for subsequent surgical procedures and experimental interventions. The parameters of the power analysis and the resulting sample-size estimation are summarized in Figure 1.

Figure 1.

Power analysis for the planned statistical model performed using G*Power software. The plot illustrates the null hypothesis distribution (H₀, red) and the alternative hypothesis distribution (H₁, blue), with the critical t value indicated by the green vertical line. The α error region (type I error) is shown to the right of the critical value, while the β error region (type II error) corresponds to the overlap of the alternative distribution with the null distribution. The table summarizes the input parameters used for the analysis (effect size f²=0.15, α=0.05, desired power=0.80) and the resulting estimated required sample size (n=43).

Based on the literature, the mean acceptable weight for safe surgical procedures in young adult female Wistar rats is ~170-200 grams (21, 22). From the initial 43 animals that entered the study, 41 met the predetermined weight criterion to undergo ovariectomy and tumor induction. All the animals (four weeks old, mean weight of 175±10g) were housed in controlled environmental conditions (temperature 22±2 °C, humidity 50±6%, 12-hour light/dark cycle), with access to standard chow and water ad libitum. Animals were assessed for general health, grooming, mobility, appetite, hydration, and evidence of pain or distress on a twice-daily basis. Tumor sites were regularly examined to assess the rate of tumor growth, ulceration, or impaired mobility.

Humane endpoints were predefined according to EU Directive 2010/63/EU (23) and included: loss of ≥15-20% of body weight, evidence of severe lethargy or diminished responsiveness, impaired ambulation, or inability to access food or water, tumor size exceeding 2 cm, ulceration or infection at the tumor site, or any persistent sign of distress unresponsive to supportive care. Following ovariectomy and tumor induction, 2 animals did not survive the immediate postoperative period during the first five days. These losses were associated with postoperative complications and included: difficulty recovering from anesthesia, postoperative respiratory depression, and ulceration at the surgical incision sites. No additional animals experienced unexplained or sudden death throughout the remainder of the experiment.

After recovery from ovariectomy and tumor induction, animals eligible for treatment were allocated to the experimental groups using simple random allocation. Group assignment was performed after confirmation of postoperative recovery and eligibility for continued participation. Two animals were terminated from the study due to the predefined humane endpoints. One animal developed ulceration at the tumor site and one developed a tumor diameter larger than 2 cm. In all cases, affected animals were provided supportive care; however, all affected animals ultimately met the criteria requiring humane euthanasia.

Treatment administration could not be blinded because of the nature of drug preparation and oral gavage procedures. However, histological evaluation, galectin-3 immunohistochemical scoring, and Western blot densitometric analysis were performed by investigators blinded to treatment group assignment. Slides and blot images were coded before analysis, and group identity was revealed only after completion of scoring and densitometric quantification.

Therefore, this is the sequential reduction in numbers as it occurred during the experiment so far: 48 animals were procured and 5 animals were removed from the study due to feeding difficulties during acclimation. 43 animals entered the study out of which 41 animals were eligible for surgery based on weight criteria since 2 animals died during the immediate postoperative recovery period. 2 animals were euthanized during the experimental course, leaving 37 animals completing the study (Table I).

Table I. Flow of animal inclusion, exclusion, randomization, and final histological analysis.

The table summarizes the sequential reduction in animal numbers throughout the experimental study.

Overall, the study was designed and reported in accordance with ARRIVE 2.0 recommendations (24). Inclusion and exclusion criteria were defined before the start of the experiment. Animals were eligible for surgery if they showed stable food intake during acclimatization, no evidence of illness or distress, and body weight within the predetermined range required for safe ovariectomy and tumor induction. Animals were excluded before surgery if they failed to establish stable food intake, showed signs of illness or distress, or did not meet the minimum body-weight criterion. After recovery from ovariectomy and tumor induction, animals eligible for treatment were randomly allocated to the experimental groups using a simple random allocation method. The individual animal was considered the experimental unit for all in vivo analyses. Histological evaluation, galectin-3 immunohistochemical scoring, and Western blot densitometric analysis were performed by investigators blinded to treatment group assignment. Slides and blot images were coded before analysis, and group identity was revealed only after completion of scoring and densitometric quantification. Animals were housed in standard individually ventilated polycarbonate cages, with 2 or 3 animals per cage, using autoclaved wood-chip bedding and environmental enrichment consisting of nesting material. Animals received standard laboratory chow supplied by the institutional animal facility and water ad libitum. Animals were obtained from National Hellenic Research Foundation, Athens, Greece.

Ethics statement. All animal procedures were conducted in accordance with the “Guide for the Care and Use of Laboratory Animals, 8th edition”, published by the National Research Council (19), in compliance with applicable Greek and European animal welfare legislation, including Directive 2010/63/EU on the protection of animals used for scientific purposes (20). All animal procedures were reviewed and approved by the Directorate of Veterinary Services of the Prefecture of Athens, Hellenic Ministry of Rural Development and Food of Greece, under Experimental License/Approval No. 284011, dated 12 April 2021. All experimental procedures were conducted at the Laboratory of Experimental Anatomy “Christeas”, National and Kapodistrian University of Athens, Greece, under the supervision of qualified veterinary staff. Animal care and experimental procedures complied with applicable Greek and European animal welfare legislation, including Law 2015/1992 and the European Convention for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes.

Mammary tumorigenesis. Mammary tumorigenesis was incorporated to reproduce the biological and therapeutic context in which endocrine modulators are administered for hormone-responsive breast cancer. The aim of this model was not to assess mammary tumor response or antitumor efficacy, but to evaluate endometrial remodeling as an off-target tissue effect occurring during endocrine-modulator exposure in a breast cancer-relevant experimental setting. Therefore, the mammary tumor-induction procedures were retained as part of the systemic disease model, whereas the predefined endpoints of the present study were restricted to uterine histology, galectin-3 immunohistochemistry, and complementary in vitro galectin-3 expression analyses.

The combined use of DMBA-induced mammary carcinogenesis and MCF-7 xenografting was selected to provide two complementary hormone-responsive breast cancer contexts. DMBA-induced tumors model chemically induced mammary carcinogenesis within the host animal, whereas MCF-7 xenografts provide a well-characterized estrogen receptor-positive human breast cancer component. This combined approach was intended to approximate endocrine-modulator exposure in a breast cancer-associated biological background. However, mammary tumor burden, tumor regression, and treatment-related antitumor response were not predefined outcomes of the present analysis and were therefore not used to support the conclusions of this study.

MCF-7 cells were chosen for the xenograft model because they represent a classic, well-validated model of hormone-dependent breast cancer and reliably form estrogen-responsive tumors in vivo (25-27). This makes them particularly suitable for evaluating antiprogestin activity, since progesterone signaling is tightly linked to estrogen-regulated pathways. MCF-7 cells are strongly ER-positive and express functional PR, reflecting intact hormonal responsiveness, while being HER2-negative, consistent with a luminal A-type phenotype. This receptor profile mirrors the clinical subtype in which antiprogestins are most biologically relevant, thereby providing an appropriate and physiologically meaningful model for investigating their effects in hormone-dependent breast cancer.

Mammary carcinogenesis was induced according to the validated protocol described by Kontogeorgi et al. (28). At eight weeks of age all rats underwent bilateral ovariectomy to induce estrogen deprivation. At ten weeks of age tumorigenesis was initiated by administering a single subcutaneous injection of 20 mg 7,12-dimethylbenz(a)anthracene (DMBA) diluted in 1 ml soybean oil into the right thoracic mammary gland, while on the left thoracic mammary gland a xenograft was established through the subcutaneous injection of 1 ml containing 7.2×10⁶ MCF-7 human breast adenocarcinoma cells suspended in Basement Membrane Extract at a 1:1 ratio. Tumors became palpable from week five at the xenograft site, whereas DMBA-induced tumors were confirmed histologically at necropsy.

Experimental process. Following ovariectomy and tumor induction and due to postoperative complications, only 39 animals were able to survive the immediate post-surgical recovery period of 5 days. These animals were assigned randomly to four different groups. The groups included a control group, one treated with tamoxifen, another treated with mifepristone, and a final group treated with ulipristal acetate. The control group consisted of 9 rats, while the other three groups each had 10 rats. The control group did not receive active pharmacological treatment. To control the differences related to handling and oral gavage, control animals received sterile saline alone by oral gavage at the same volume, frequency, and duration as the treatment groups. The tamoxifen group received tamoxifen at a rat-equivalent dose of 2.07 mg/kg/day. The mifepristone group received mifepristone at a rat-equivalent dose of 31.0 mg/kg/day. The ulipristal acetate group received ulipristal acetate at a rat-equivalent dose of 0.52 mg/kg/day. Doses were calculated by human-to-rat allometric conversion using the FDA Km factor method (29). Human reference doses used for the calculation were tamoxifen 20 mg/day, ulipristal acetate 5 mg/day, and mifepristone 300 mg/day. Based on a mean animal body weight of 0.21 kg, the approximate absolute daily doses were 0.43 mg for tamoxifen, 6.51 mg for mifepristone, and 0.11 mg for ulipristal acetate.

At the end of the experimental period, all animals were euthanized in accordance with Annex IV of Directive 2010/63/EU (30). Euthanasia was performed by anesthetic overdose through overexposure to sevoflurane vapor in an induction chamber until loss of reflexes and respiratory arrest. Death was subsequently confirmed by thoracotomy, in line with European regulatory requirements for confirmation of death. This procedure is recognized as a humane and acceptable method for rodent euthanasia within the European Union. Following euthanasia, the uteri were excised, and histology evaluation was performed (31).

Histology evaluation. Tissue sections were deparaffinized in xylene, rehydrated in descending ethanol series, and then stained in Mayer’s hematoxylin (catalogue no. MHS16; Sigma-Aldrich, Darmstadt, Germany) for 3 min following rinsing in 0.1% HCl (catalogue no. 1.09060; Merck KGaA, Darmstadt, Hesse, Germany) solution and tap water. Sections were stained in eosin stain (catalogue no. HT110116; Sigma-Aldrich, Inc., St. Louis, MO, USA) for 1-2 min and then rinsed in tap water. Sections were also dehydrated in ascending ethanol series, cleaned with xylene, and mounted.

Immunohistochemical staining. To minimize inter-assay variability, all slides were stained in the same staining run by the same researcher. Immunohistochemical staining was performed on 5-μm formalin-fixed, paraffin-embedded uterine tissue sections obtained using a standard microtome. Sections were deparaffinized in xylene and rehydrated through graded ethanol solutions to distilled water. Antigen retrieval was performed in citrate buffer, pH 6.0, for 20 min in a microwave oven, followed by cooling at room temperature for 20 min. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 10 min, followed by protein blocking with 5% normal goat serum for 20 min.

Sections were incubated overnight at 4°C with mouse monoclonal anti-galectin-3 antibody (catalogue no. 255M-1; Sigma-Aldrich Inc.), at a dilution of 1:100. After washing, sections were incubated with an HRP-conjugated polymer-based anti-mouse secondary detection system for 30 min at room temperature. Immunoreactivity was visualized using 3,3′-diaminobenzidine tetrahydrochloride, DAB, and sections were counterstained with hematoxylin, dehydrated, cleared, and mounted. Brown granular cytoplasmic, membranous, or nuclear staining in endometrial cells was considered positive.

Appropriate positive-control tissue known to express galectin-3 was included in the staining run, and negative controls were processed by omission of the primary antibody. Staining was considered interpretable only when the positive control showed the expected immunoreactivity and the negative control showed no specific signal.

Galectin-3 immunostaining was evaluated semi-quantitatively by two independent pathologists who were blinded to treatment-group assignment. For each slide, three representative high-power fields at 40× magnification were assessed. Staining intensity was scored as 0 = absent, 1 = weak, 2 = moderate, and 3 = strong. The percentage of positive endometrial cells was estimated in each field. A combined immunoreactivity score was calculated by multiplying staining intensity by the percentage of positive cells, generating a score ranging from 0 to 300. Final immunoreactivity scores were recorded for each animal and summarized by treatment group. Cases with discrepant scores were reviewed jointly, and a final consensus score was assigned.

Digital morphometric analysis. Digital morphometric analysis was performed on seven representative hematoxylin-and-eosin-stained sections showing endometrial hyperplasia. Images were analyzed using ImageJ software, version 1.53a (National Institutes of Health, Bethesda, MD, USA). The total endometrial tissue area was segmented into glandular and stromal compartments. Luminal regions were digitally dilated to include the surrounding epithelial rim, thereby providing an estimate of the complete glandular compartment rather than the luminal area alone. The glandular fraction was calculated as the glandular area divided by the total tissue area and expressed as a percentage. The stroma-to-gland ratio was calculated by dividing the stromal area by the glandular area and multiplying the resulting value by 100. Lower ratios indicated a greater relative glandular component and reduced stromal separation, consistent with glandular crowding and more complex hyperplastic architecture, whereas higher ratios indicated greater stromal separation and simpler hyperplastic patterns. The morphometric findings were analyzed descriptively.

In vitro experiments in Ishikawa cells. Ishikawa human endometrial adenocarcinoma cells were obtained from the European Collection of Authenticated Cell Cultures/Sigma-Aldrich (ECACC catalogue no. 99040201; Sigma-Aldrich Inc.). Ishikawa human endometrial adenocarcinoma cells were grown in DMEM/F12 (catalogue no. D8437; Sigma-Aldrich Inc.) with 10% fetal bovine serum (FBS) (catalogue no. F7524; Sigma-Aldrich Inc.) and 1% penicillin-streptomycin (catalogue no. P4333; Sigma-Aldrich Inc.) at 37°C in a humidified incubator with 5% CO₂. Cells were authenticated with short tandem repeat profiling and confirmed negative for mycoplasma prior to all experiments; passages 5-20 were used. With regards to hormone-modulator studies, cultures were pre-equilibrated for 48h in phenol-red-free DMEM/F12 (catalogue no. 21041025; Gibco, Life Technologies Corporation, Grand Island, NY, USA) supplemented with 10% charcoal/dextran-stripped FBS (catalogue no. F6765; Sigma-Aldrich Inc.) to decrease background steroids, as is typical for endocrine assays in Ishikawa cells. After pre-equilibrating, cells were detached with 0.25% trypsin-EDTA (catalogue no. T4049; Sigma-Aldrich Inc.), counted with a hemocytometer (or calculator counter) via trypan blue exclusion, and added at a density of 2×10⁵ cells per well in six-well plates with 2 ml media per well. Plates were arranged and numbered to randomize treatment assignment across the wells in each plate. Cells were given up to 24h to attach and were permitted to become approximately 70-80% confluent prior to the initial dosing (32).

Drug stocks were prepared fresh or obtained from single-use aliquots stored at −20 °C and protected from light. 4-Hydroxytamoxifen (4-OHT) (catalogue no. H6278; Sigma-Aldrich Inc.), mifepristone (RU-486) (catalogue no. M8046; Sigma-Aldrich Inc.), and ulipristal acetate (catalogue no. SML1469; Sigma-Aldrich Inc.) were dissolved in anhydrous DMSO (catalogue no. 276855; Sigma-Aldrich Inc.) to yield 10 mM stocks, except if there were any solubility concerns that demanded the use of ethanol; the final vehicle concentration was always maintained at ≤0.1% (v/v) in all wells, including the vehicle control. Nine treatment groups were created including three concentrations for each agent; the concentrations were selected based on commonly referenced dosages from the Ishikawa literature: 4-OHT 0.01 µM, 0.1 µM, and 1 µM (33); mifepristone 0.1 µM, 1 µM, and 10 µM (34); and ulipristal acetate 0.04 µM, 0.4 µM, and 4 µM (30). Optional parallel reference hormone conditions were included as benchmarks for receptor responsiveness and included 10 nM 17β-estradiol (catalogue no. E8875; Sigma-Aldrich Inc.) and 100 nM progesterone (catalogue no. P0130; Sigma-Aldrich Inc.). Treatments were applied for 48h in phenol-red-free stripped serum media (catalogue no. F6765; Sigma-Aldrich) and the medium was not changed during the exposure period. Each biological experiment consisted of cultures independently plated from different passages unless indicated otherwise. At the time of harvest, monolayers were rinsed twice in ice-cold PBS (catalogue no. D8537; Sigma-Aldrich Inc.), put on ice immediately, and subjected to protein extraction as described below. The complete experimental workflow, including cell plating, drug exposure, protein extraction, and Western blot analysis, was independently repeated on three separate occasions to confirm the consistency and reproducibility of the experimental procedure. These repetitions were performed as procedural validation runs and were not treated as independent biological replicates for statistical inference. For each condition within each run, duplicate technical lysates were prepared to assess technical consistency.

Western blot for galectin-3. Monolayers were lysed on ice in Pro-Prep protein extraction solution (catalogue no. 17081; iNtRON Biotechnology, Inc., Seongnam, Gyeonggi-do, Republic of Korea) supplemented with the manufacturer’s protease inhibitor cocktail. Lysis was allowed to proceed for 20 min with gentle trituration every 5 minutes. Lysates were clarified at 14,000×g for 20 min at 4°C and the supernatants transferred to pre-chilled tubes. Total protein was quantified by Bradford assay (Bio-Rad Laboratories, Inc., Hercules, CA, USA) against BSA standards prepared in the same buffer. Samples were adjusted to equal protein concentration and mixed 1:3 with 4× Laemmli sample buffer (catalogue no. S3401; Sigma-Aldrich) containing 5% β-mercaptoethanol (catalogue no. M3148; Sigma-Aldrich Inc.), heated at 95 °C for 5 min, and briefly centrifuged. Equal protein amounts (20 µg per lane unless otherwise specified) were resolved by SDS-PAGE on 10-12% polyacrylamide minigels at constant voltage and transferred to nitrocellulose membranes (Schleicher & Schuell BioScience GmbH, Dassel, Lower Saxony, Germany) using wet transfer at 100 V for 60-90 min at 4 °C. Transfer quality and uniformity were verified by Ponceau S staining (catalogue no. P7170; Sigma-Aldrich Inc.); membranes were then blocked for 1 hour at room temperature in 5% (w/v) non-fat dry milk in TBS-T (20 mM Tris-HCl, 150 mM NaCl, 0.1% Tween-20, pH 7.6).

Membranes were incubated overnight at 4 °C with rat anti-galectin-3 monoclonal antibody (clone TIB-166, ATCC) at 1:5000 in TBS-T containing 1% milk. After three 10-minute washes in TBS-T, membranes were incubated for 1 h at room temperature with HRP-conjugated anti-rat IgG (Santa Cruz Biotechnology, Inc., Dallas, TX, USA) at 1:5000. Signals were developed with enhanced chemiluminescence (WEST-one, iNtRON Biotechnology, Inc.) and captured on Agfa medical X-ray film with exposure times adjusted to remain within the linear detection range; if needed, shorter exposures were acquired to avoid saturation of the galectin-3 band. Membranes were stripped according to the ECL manufacturer’s instructions and reprobed with mouse anti-β-actin (Sigma; 1:5,000) followed by HRP-conjugated anti-mouse IgG to verify equal loading and transfer. The expected molecular weight for galectin-3 was ~29-30 kDa; a single dominant band at this position was required for acceptance of a blot (35).

For densitometry, 16-bit TIFF images of non-saturated films were analyzed in ImageJ or equivalent software. Identical rectangular regions of interest were drawn around each band, and background subtraction was performed using a local area adjacent to each lane. For each lane, the background-corrected integrated density of galectin-3 was divided by that of β-actin to yield a normalized value. Within each blot, the vehicle control (0.1% DMSO) mean was set to 1.0, and all other lanes were expressed as fold-change relative to this control. When multiple blots were required to accommodate all conditions, at least one shared control lane and one shared reference condition were included on each membrane to facilitate cross-blot scaling; fold-changes were then averaged across biological replicates. Quality criteria included absence of air-bubble artifacts, uniform transfer by Ponceau S, and a coefficient of variation for β-actin within a blot of <15%; if β-actin varied beyond this threshold, total-protein normalization was used as a prespecified alternative (36).

Statistical analyses. Descriptive statistics were used to summarize all experimental outcomes. Histological outcomes were categorical and were therefore summarized descriptively as counts and percentages for each experimental group. Because of the modest number of animals available for final histological assessment, between-group histological comparisons were interpreted descriptively rather than as definitive inferential comparisons. Percentages were calculated using the final number of animals analyzed in each group as the denominator.

Western blot densitometric quantification was performed using ImageJ [Version 1.53a, 2020, National Institutes of Health (NIH)]. Galectin-3 protein expression was normalized to β-actin and expressed as fold-change relative to vehicle control. Fold-change values were summarized descriptively for each compound and concentration. Because the comparative data were available as summarized fold-change estimates without replicate-level variance measures, formal inferential testing was not used to support interpretation of the in vitro findings. The in vitro results were therefore interpreted as descriptive and exploratory.

Results

Histological findings. Thirty-nine female Wistar rats were included in the experiment and allocated into four groups: tamoxifen (n=10), mifepristone (n=10), ulipristal acetate (n=10), and control (n=9). During the experimental period, two animals from the tamoxifen group met predefined humane endpoint criteria and were removed before final histological assessment. As a result, these animals were not included in the final uterine histological analysis, and the tamoxifen group had a smaller final sample size than the other treatment groups; therefore, 37 animals were included in the final analysis. In the tamoxifen group, 8 animals were evaluated histologically. Of these, 4 animals showed simple endometrial hyperplasia and 4 animals showed complex endometrial hyperplasia, corresponding to 50% simple hyperplasia and 50% complex hyperplasia among the analyzed animals. In the mifepristone group, all 10 animals survived and were analyzed; normal endometrial morphology was observed in 3 animals, while simple hyperplasia was observed in 7 animals, corresponding to 30% normal endometrium and 70% simple hyperplasia. No complex hyperplasia was observed in this group. In the ulipristal acetate group, all 10 animals survived and were analyzed; 2 animals showed normal endometrial morphology, 6 showed simple hyperplasia, and 2 showed complex hyperplasia, corresponding to 20% normal endometrium, 60% simple hyperplasia, and 20% complex hyperplasia. In the control group, all 9 animals survived and showed normal endometrial morphology. Overall, compared with controls, treatment groups showed increased endometrial alterations. Descriptively, tamoxifen showed the highest proportion of complex hyperplasia among the treatment groups, whereas mifepristone was associated only with simple hyperplasia and ulipristal acetate showed a mixed pattern of normal, simple hyperplastic, and complex hyperplastic findings.

Taken together, the histological and immunohistochemical findings indicate that endocrine-modulator exposure produced treatment-specific endometrial remodeling, with loss of detectable galectin-3 staining in hyperplastic lesions and preservation of weak galectin-3 expression in morphologically normal endometrium. These integrated in-vivo findings, together with the corresponding Ishikawa in-vitro galectin-3 expression patterns, are summarized schematically in Figure 2.

Figure 2.

Integrated summary of endometrial morphology, galectin-3 immunoreactivity, and in-vitro findings. (A) Representative hematoxylin-and-eosin-stained section showing normal endometrial morphology in a control animal. (B) Corresponding galectin-3 immunohistochemistry showing weak positive staining in morphologically normal endometrium. (C) Representative section showing complex endometrial hyperplasia in the tamoxifen-treated group. (D) Galectin-3 immunohistochemistry showing no detectable staining in the tamoxifen-associated hyperplastic lesion. (E) Representative section showing simple endometrial hyperplasia in the mifepristone-treated group. (F) Galectin-3 immunohistochemistry showing no detectable staining in the mifepristone-associated hyperplastic lesion. (G) Representative section showing preserved normal endometrial morphology in an ulipristal acetate-treated animal. (H) Galectin-3 immunohistochemistry showing retained positive staining in morphologically normal endometrium from the ulipristal acetate group. (I) Schematic representation of treatment-associated changes in Ishikawa-cell proliferation. (J) Schematic representation of treatment-associated changes in galectin-3 expression in Ishikawa cells. The schematic illustrates the relationship between endometrial morphology, galectin-3 immunoreactivity, and the corresponding in-vitro observations.

We performed a digital morphometric analysis of seven representative histological slides of endometrial hyperplasia using H&E staining. Digital morphometric analysis was performed using ImageJ (Version: 1.53a, 2020, NIH). Tissue areas were segmented into glandular (epithelium-inclusive) and stromal compartments. In this approach, luminal regions were dilated to incorporate the epithelial rim, providing a more accurate estimate of the glandular compartment.

Across the representative slides analyzed, the glandular fraction ranged from near-zero to 48% of total tissue area, with a mean of 18%, while the corresponding stroma-to-gland ratios ranged from 206 to 2319 (mean ~747). The stroma-to-gland ratio was calculated as stromal area divided by glandular area and expressed as a scaled index multiplied by 100. Therefore, values of 206 to 2319 correspond approximately to raw stromal-to-glandular area ratios of 2.06:1 to 23.19:1. Lower values indicated a greater relative glandular component and reduced stromal separation, consistent with glandular crowding and more complex hyperplastic architecture. Higher values reflected a larger stromal component relative to glands, consistent with preserved stromal separation and simpler hyperplastic patterns.

Immunohistochemistry of Galectin-3 (IHC). Galectin-3 immunoreactivity was low but detectable in morphologically normal endometrium from control animals and from ulipristal acetate-treated animals, with faint, predominantly cytoplasmic staining in scattered glandular epithelial cells. The corresponding semi-quantitative immunoreactivity scores were low but non-zero, with a median combined immunoreactivity score of 10 and a range of 5-20. In contrast, all evaluated simple and complex hyperplastic endometrial lesions from the tamoxifen, mifepristone, and ulipristal acetate groups showed no detectable Galectin-3 immunoreactivity and had a combined immunoreactivity score of 0. These findings were consistent across both simple and complex hyperplastic lesions, suggesting that loss of detectable Galectin-3 expression was associated with hyperplastic morphology rather than with a specific treatment group. Representative images are shown in Figure 3 and Figure 4.

Figure 3.

Galectin-3 immunohistochemical staining in morphologically normal rat endometrium. (A) Representative low-magnification view showing preserved endometrial architecture. Arrow 1 indicates an endometrial gland with faint cytoplasmic galectin-3 immunoreactivity in glandular epithelial cells, whereas arrow 2 indicates weak galectin-3 staining in the luminal surface epithelium. (B) Higher-magnification view of morphologically normal endometrium. Arrows 3 and 4 indicate endometrial glands with weak, predominantly cytoplasmic galectin-3 immunoreactivity in scattered glandular epithelial cells, and arrow 5 indicates weak staining in the luminal epithelium. Brown staining represents positive galectin-3 immunoreactivity, and hematoxylin was used as the counterstain. Original magnification: (A) 10×; (B) 40×.

Figure 4.

Effects of tamoxifen, mifepristone, and ulipristal on Galectin-3 protein expression in Ishikawa cells. A representative western blot is shown together with densitometric quantification of Galectin-3 expression. Galectin-3 band intensity was normalized to [name of loading control] and expressed relative to the untreated control. Bars represent mean±SD from three independent biological experiments (n=3). Plotted values for each individual group are as follows: Control=1.00±0.1349, Tamoxifen=0.65±0.4294, Mifepristone=0.55±0.2316, Ulipristal=0.70±0.1960.

Galectin-3 densitometry expression in Ishikawa endometrial cells. Galectin-3 protein levels in Ishikawa cells were quantified by Western blot densitometry using Galectin-3 densitometry expression in Ishikawa endometrial cells. Galectin-3 protein levels in Ishikawa cells were quantified by Western blot densitometry and expressed as fold-change relative to the vehicle control. The fold-change values showed agent-specific patterns across the tested concentrations. Tamoxifen produced a modest increase in Galectin-3 expression at the lowest concentration, followed by reduced expression at the intermediate and highest concentrations (Figure 5). Mifepristone showed an overall suppressive pattern, with the greatest reduction observed at the highest concentration (Figure 6). Ulipristal acetate showed a non-monotonic response, with increased Galectin-3 expression at low and intermediate concentrations followed by a return toward the vehicle-control level at the highest concentration (Figure 7). Because the presented values were derived from a representative Western blot analysis and the procedural repetitions were not combined for statistical analysis, no replicate-level variance estimates or formal inferential comparisons are reported. These findings were therefore interpreted descriptively and should be considered exploratory. The individual fold-change values and their corresponding direction of change relative to the vehicle control are summarized in Table II.

Figure 5.

Dose–response effect of tamoxifen on galectin-3 expression in Ishikawa cells. Galectin-3 protein expression was quantified by Western blot densitometry, normalized to β-actin, and expressed relative to the vehicle control, which was set to 1.0. Circles represent values obtained using ImageJ, squares represent values obtained using ImageLab, and crosses indicate the consensus values calculated from both densitometry methods. The solid green line represents the fitted four-parameter logistic dose–response curve. The dashed horizontal line indicates the vehicle-control level. The estimated model parameters were E_min=0.32, E_max=1.49, EC₅₀=0.0245 μM, and Hill coefficient h=1.48.

Figure 6.

Dose–response effect of mifepristone on galectin-3 expression in Ishikawa cells. Galectin-3 protein expression was quantified by Western blot densitometry, normalized to β-actin, and expressed relative to the vehicle control, which was set to 1.0. Circles represent values obtained using ImageJ, squares represent values obtained using ImageLab, and crosses represent the consensus values calculated from both densitometry methods. The solid green line represents the fitted four-parameter logistic dose–response curve, and the dashed horizontal line indicates the vehicle-control expression level. The estimated model parameters were E_min=0.00, E_max=0.91, EC₅₀=10.4 μ"M" , and Hill coefficient h=5.00.

Figure 7.

Dose-response effect of ulipristal acetate on galectin-3 expression in Ishikawa cells. Galectin-3 protein expression was quantified by Western blot densitometry, normalized to β-actin, and expressed relative to the vehicle control, which was set to 1.0. Circles represent values obtained using ImageJ, squares represent values obtained using ImageLab, and crosses represent the consensus values calculated from the two densitometry methods. The solid green line represents a quadratic fit based on the logarithm of the ulipristal acetate dose. The dashed horizontal line indicates the vehicle-control expression level. Galectin-3 expression showed a non-monotonic, bell-shaped response, with increased expression at low and intermediate concentrations and a return toward the control level at the highest concentration.

Table II. Summary of fold-change values of galectin-3 relative to control for each drug and condition.

Measurements were performed using ImageJ. The direction column reflects the overall trend in galectin-3 modulation, indicating increased (↑↑), decreased (↓↓), or unchanged (≈) expression. This comparative approach enables evaluation of consistency between analytical platforms and supports interpretation of dose-dependent or condition-specific effects on galectin-3 expression.

Discussion

Taken together, the present data provide an integrated exploratory evaluation of endometrial responses to tamoxifen, mifepristone and ulipristal acetate, combining in vivo morphological observations with in vitro analysis of galectin-3 expression. In the analyzed cohort, treatment-associated histological patterns differed descriptively among groups. Tamoxifen-treated animals showed the highest proportion of complex hyperplasia, whereas mifepristone- and ulipristal acetate-treated animals more frequently showed simple hyperplasia. Normal endometrial morphology was retained in selected ulipristal acetate-treated animals. Across the hyperplastic lesions examined, galectin-3 immunoreactivity was not detectable under the staining conditions used, whereas low basal expression was retained in morphologically preserved endometrium. Complementary in vitro experiments showed agent-specific modulation of galectin-3 expression. These findings suggest that galectin-3 may reflect endocrine-associated structural remodeling of the endometrium, but its value as a biomarker of hyperplastic transition requires validation in larger cohorts.

Galectin-3’s biological functions provide mechanistic support for these observations. As a lectin involved in epithelial adhesion, stromal communication, apoptosis regulation and extracellular matrix organization, galectin-3 contributes to the maintenance of epithelial architecture (37, 38). Reduced galectin-3 expression is expected to weaken β-galactoside-mediated adhesive interactions, destabilize anti-apoptotic pathways and predispose to epithelial disorganization (39, 40). These mechanistic features are consistent with the histological patterns observed in the present hyperplastic lesions, including glandular crowding and stromal attenuation. Preservation or induction of galectin-3 under physiological progesterone signaling or moderate SPRM exposure aligns with a more controlled epithelial phenotype. The in vitro densitometry data were consistent with agent-specific regulation of galectin-3, with suppression at higher tamoxifen and mifepristone concentrations and a non-monotonic response to ulipristal acetate. Given the modest number of biological replicates and limited dose levels, these molecular findings should be interpreted as exploratory and hypothesis-generating.

A review of the existing literature as can be seen in Table III signifies the relevance of these findings. Antiprogestins such as mifepristone and ulipristal acetate have been extensively studied in breast cancer research for their ability to suppress proliferation, induce apoptosis and modulate progesterone receptor-dependent pathways (16, 41-43). Evidence from cell-line models, patient-derived xenografts, BRCA1-associated breast cancer models and clinical pilot studies indicates that antiprogestins may benefit selected patient groups, particularly those with elevated PRA/PRB ratios or progesterone-responsive tumors (44-50) (Table III). However, despite their increasing use, the endometrial consequences of antiprogestins therapy have been insufficiently studied. Most investigations have focused exclusively on breast tissue endpoints, leaving uterine safety insufficiently characterized. This gap is particularly notable given the well-documented endometrial effects of tamoxifen, including its association with polyps, hyperplasia and carcinoma.

Table III. Literature review of anti-progestin administration as hormonal therapy to breast cancer.

The present findings contribute to this gap by showing that, in this experimental cohort, mifepristone and ulipristal acetate were associated mainly with simple hyperplastic changes, whereas tamoxifen-treated animals showed a higher descriptive proportion of complex hyperplasia. The absence of detectable galectin-3 immunoreactivity across hyperplastic samples suggests a shared molecular feature of endocrine-associated epithelial remodeling in this model. The preservation of low-level galectin-3 expression in morphologically normal ulipristal acetate-treated specimens may indicate partial maintenance of a more physiological epithelial phenotype, although this interpretation requires confirmation using larger groups and additional dose levels. These observations align with ulipristal’s established partial-agonist behavior at the progesterone receptor.

A limitation of the present study is that the final analyzed animal cohort was smaller than the minimum sample size estimated by the a-priori power analysis. Although 48 animals were initially procured to meet the calculated requirement of 43 animals and to compensate for anticipated attrition, exclusions during acclimatization, pre-surgical eligibility screening, postoperative mortality, and humane-endpoint removals reduced the final analyzed cohort to 37 animals. Consequently, the study may have had reduced statistical power for detecting smaller between-group differences, particularly for categorical histological outcomes and subgroup-level comparisons. The in vivo findings should therefore be interpreted as exploratory and hypothesis-generating rather than definitive. Nevertheless, the consistency between the histological observations, galectin-3 immunohistochemical pattern, and complementary Ishikawa-cell data supports further investigation of galectin-3 as a candidate biomarker in larger, adequately powered animal studies and human endometrial samples.

An additional limitation concerns the use of Ishikawa cells for the in vitro component of the study. Ishikawa cells are human endometrial adenocarcinoma-derived epithelial cells and, although widely used as a hormone-responsive endometrial model, they cannot fully reproduce the biology of normal endometrial epithelium. In particular, this monoculture system does not capture stromal-epithelial interactions, immune-cell contributions, extracellular-matrix remodeling, or the cyclical hormonal regulation that characterizes intact endometrial tissue. Therefore, the in vitro findings should be interpreted as exploratory evidence of direct drug-associated modulation of galectin-3 protein expression in an endometrial epithelial cancer-cell model, rather than as definitive evidence of normal endometrial physiology.

Existing studies on galectin-3 in endometrial pathology have reported divergent patterns, with some indicating progressive upregulation during neoplastic transformation and others showing reduced expression in invasive or poorly differentiated lesions (7). Such discrepancies highlight the context-dependent nature of galectin-3 and the influence of hormonal milieu, cellular compartmentalization and methodological variability. The context-dependent behavior of galectin-3 is also evident in non-gynecological tissues. In a 6-hydroxydopamine-induced Parkinsonian rat model, galectin-3 was upregulated in injured nigral dopaminergic neurons and colocalized with activating transcription factor 3, indicating that galectin-3 expression may vary according to cell type, tissue environment, and the nature of the cellular injury (52). Within the specific setting of hormonally driven hyperplasia, the present pattern characterized by downregulation of galectin-3 during early architectural disruption helps refine its biological interpretation and differentiates these alterations from malignant behavior.

Overall, the combined in vivo and in vitro findings support further investigation of galectin-3 as an exploratory candidate marker of endocrine-modulator-associated endometrial remodeling. Its apparent modulation in relation to hyperplastic morphology and endocrine exposure suggests biological relevance, but its diagnostic, predictive, or monitoring value cannot be established from the present study. Larger animal studies, additional dose-response experiments, and validation in human endometrial samples from treated patients will be required before galectin-3 can be considered for clinical risk stratification or surveillance.

The interpretation of immunohistochemical biomarkers should remain anchored to tissue morphology. In thyroid core-needle biopsy specimens, p62 immunostaining improved the distinction between true intranuclear inclusions and processing-related nuclear vacuoles, illustrating the diagnostic value of context-specific immunohistochemical markers as adjuncts to morphological assessment (53). Accordingly, galectin-3 staining in the present study should be regarded as an exploratory adjunct to endometrial morphology rather than as a stand-alone diagnostic marker.

Cross-species molecular comparisons may also strengthen the translational validation of the present animal model. Ishigure et al. demonstrated that disease-associated inflammatory gene co-expression networks in a choline-deficient diet-induced rat model of non-alcoholic fatty liver disease were preserved in human datasets, with LGALS3 identified as a hub gene within an inflammation-associated module (54). A comparable cross-species approach could help determine whether the galectin-3 patterns observed in the present rat model are also present in human endometrium during endocrine therapy.

Future studies incorporating stromal-epithelial co-culture systems, LGALS3 transcriptional analyses, functional assays of adhesion and apoptosis, and human endometrial samples under clinical endocrine therapy will be essential to determine whether galectin-3 can support future assessment of endometrial safety in patients treated with SERMs, SPRMs or antiprogestins. In addition, the concept of co-administration of tamoxifen with antiprogestins represents an intriguing potential therapeutic strategy. Such a combination may theoretically enhance tumor suppression in breast tissue while allowing a hypo-estrogenic environment to counteract the adverse uterotrophic effects associated with tamoxifen monotherapy. Although this possibility extends beyond the scope of the present study, it represents an important direction for future research aimed at optimizing endocrine regimens while minimizing endometrial risk.

Conflicts of Interest

The Authors declare that there are no conflicts of interest.

Authors’ Contributions

Conceptualization: AK, IB, GA, SNK, AM. Data Curation: AK, DK, GA, NK, EZ. Formal Analysis: AK, TSD, GA. Funding Acquisition: IB, SNK. Investigation: AK, TSD, GA, NK, EZ. Methodology: IB, TSD, GA, NK, EZ. Project Administration: IB, SNK. Resources: IB, NK, EZ. Software: AK, DK, NK. Supervision: SNK, AM. Validation: TSD, GA, EZ, SNK. Visualization: AK, GA. Writing – Original Draft: AK, TSD, GA, NK, EZ. Writing – Review & Editing: IB, SNK, AM.

Artificial Intelligence (AI) Disclosure

During the preparation of this manuscript, a large language model (ChatGPT 5.6, 2026, OpenAI) was used solely for language editing and stylistic improvements in select paragraphs. No sections involving the generation, analysis, or interpretation of research data were produced by generative AI. All scientific content was created and verified by the authors. Furthermore, no figures or visual data were generated or modified using generative AI or machine learning-based image enhancement tools.