The human endometrial adenocarcinoma cell lines HEC-1A (type II EC, post-menopausal model; HTB112, ATCC, Manassas, VA, USA) and Ishikawa (type I EC, pre-menopausal model; 99040201, Sigma-Aldrich, Munich, Germany) were used for the in vitro experiments. Cells were grown in Eagle’s minimal essential medium (MEM; Sigma-Aldrich) supplemented with 5.0 (Ishikawa) or 10% (v/v) (HEC-1A) charcoal-stripped fetal bovine serum (hormone-reduced FBS; Gibco, Waltham, MA, USA), 1.0% (v/v) penicillin/streptomycin (Gibco) and 1.0 % (v/v) non-essential amino acids (Sigma-Aldrich). All cells were grown in an incubator at 37˚C and 5.0% CO₂ in a humidified environment.

During experiments, cells were maintained in a normoglycemic (NG) environment (5.5 mM glucose), representing physiological blood glucose levels of 100 mg/dL. For the experiments under hyperglycemic (HG) conditions, the medium was supplemented with glucose (Sigma-Aldrich) to achieve a final concentration of 17.0 mM glucose, equivalent to 306 mg/dL as seen in diabetic patients. A 100 mM stock solution of metformin (Sigma-Aldrich) in deionized water was freshly prepared on the day of administration and further diluted in normo- or hyperglycemic culture medium, respectively, in order to adjust the desired concentrations of the drug (0.01 - 5.0 mM) during experiments. Furthermore, cell culture media were supplemented with 10 nM E2 (Sigma-Aldrich) during treatments in order to mimic elevated estrogen levels. A 10 µM stock solution of E2 was prepared in ethanol (stored at −20˚C) and diluted (1:1000) in medium directly before the use in experiments (final concentration of ethanol: 0.1% (v/v)). Elevated estrogen levels represent a common risk factor for EC development and progression and are therefore essential for a realistic in vitro simulation of the environment in EC patients and women with increased risk for EC development [4] [5] [22].

In order to evaluate the effects of metformin on cellular viability, HEC-1A and Ishikawa cells were seeded into transparent (MTT) or opaque-walled, white (ATP) 96-well plates at a density of 5000 cells/well and incubated in a normal or high glucose, drug-free medium for 24 h. Afterwards, cells were treated with different concentrations of metformin (0.01 - 20 mM) for 7 d and the medium was changed every 2 - 3 d. Untreated cells (medium (NG or HG) supplemented with 10 nM E2) served as the reference control. For the MTT assay, 20 μL of a 2.5 mg/mL MTT solution (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; Alfa Aesar, Karlsruhe, Germany) was added per well after treatment and incubated at 37˚C for 4 h. Afterwards, the MTT-containing media was substituted by 50 μL DMSO (Carl Roth, Karlsruhe, Germany) for the extraction of insoluble formazan crystals. The absorbance was measured at λ = 570 nm (OD₅₇₀) with a microplate reader (Anthos Microsystems, Friesoythe, Germany) and cellular viability was calculated with the following Equation (1) (cells with DMSO served as a blank sample):

% cell   viability = OD 57 0 , treatment − OD 57 0 , blank OD 57 0 , control − OD 57 0 , blank × 1 00 . (1)

For the adenosine triphosphate (ATP) assay, the CellTiter-Glo luminescent assay kit (Promega, Fitchburg, WI, USA) was used. Briefly, 100 µL of the CellTiter-Glo reagent solution was added to each well after treatment and mixed on a microplate shaker for 2 min in order to lyse the cells. Afterwards, the mixture was incubated at room temperature for another 10 min without shaking and luminescence (integration time: 1 s) was detected with a microplate luminometer (Berthold Technologies, Bad Wildbad, Germany). Cellular viability was calculated with the following Equation (2) (medium without cells served as a blank sample):

% cellviability = Luminescence treatment − Luminescence blank Luminescence control − Luminescence blank × 1 00 . (2)

For both assays, SPSS version 24 (IBM, Armonk, NY, USA) was used for the establishment of concentration-response curves and the calculation of IC₅₀ and IC₉₀ values (inhibitory concentrations, where 50% and 90% of the measured effect, i.e. loss of cellular viability, was observed; values were set at the inflection points (IC₅₀) of the sigmoidal log(concentration)-T/C (treated over control) curves). All experiments were carried out in triplicates and repeated at least three times.

HEC-1A and Ishikawa cells were seeded into 24-well plates at a density of 5000 cells/well and incubated in a normal or high glucose, drug-free medium for 24 h. Afterwards, cells were treated with different concentrations of metformin (0.01 - 5.0 mM) for 7 d with renewed treatments by medium changes every 2 - 3 d. Untreated cells served as the reference control. The relative cell proliferation was measured by crystal violet (CV) staining according to an adapted protocol [23]. Briefly, cells were rinsed twice with phosphate-buffered saline (PBS; Sigma-Aldrich) and stained with 0.5% (w/v) CV (Carl Roth) in a mixture of 80% (v/v) deionized water and 20% (v/v) methanol (Carl Roth) for 20 min. After repeated washing with PBS, the CV stain was extracted by incubation with 33% (v/v) acetic acid (Carl Roth) in deionized water for 20 min. From each well, 100 mL of the extract were transferred into 96-well plates in duplicates and the absorbance was measured at λ = 570 nm (OD₅₇₀) with a microplate reader (Anthos Microsystems). The absorbance of the untreated control group was set to 100% and the effect of different treatments was compared to the control. Each experiment was performed in duplicates and repeated at least three times in independent experiments.

HEC-1A and Ishikawa cells were seeded into 6-well plates at a density of 5000 cells/well and incubated in a normal or high glucose, drug-free medium for 24 h. Afterwards, cells were treated with different concentrations of metformin (0.01 - 5.0 mM) for 7 d with regular medium changes every 2 - 3 d. Untreated cells served as the reference control. Subsequently, cells were stained with 0.5% (w/v) CV, washed three times with PBS and air-dried before image acquisition. The colony formation was quantified as % area with the ImageJ software [24] and the ColonyArea plugin [25]. All experiments were repeated at least three times.

Cell migration was assessed by quantifying the % wound closure in the wound healing assay. The cells were seeded into 6-well plates at a density of 50,000 cells/well in a normal or high glucose, substance-free medium for 24 h. Afterwards, cells were treated with selected concentrations of metformin (0.5 or 5.0 mM) for 7 d with medium changes every 2 - 3 d. After 7 d, the confluent monolayers were wounded using a sterile pipette tip and the medium was replaced by fresh, substance-containing normo- or hyperglycemic medium in order to remove cellular debris. Representative images were taken with an inverse light microscope (Leica, Munich, Germany) at 40× magnification directly and 24 h after wounding of the monolayer. The migration area A was measured using the ImageJ software [24] and the % wound closure was calculated with the following Equation (3):

( A t = 0 h − A t = 24 h ) / A t = 0 h × 100 , (3)

where A_{t=0 h} and A_{t=24 h} are the area of the wound immediately after scratching (0 h) and 24 h later, respectively. Measurements were taken after 24 h in order to limit the observations to migration rather than cellular proliferation [26]. HEC-1A and Ishikawa cells display doubling times of 27 - 29 h during routine culture, and thus the effect of proliferation in the wound healing assay was minimized. All experiments were repeated at least three times in duplicates.

Quantitative data are presented as mean ± standard error of the mean (SEM) of at least three independent experiments. SPSS version 24 (IBM, Armonk, NY, USA) was used to perform a generalized estimating equation test or a paired t-test, as appropriate. A p value ≤ 0.05 was considered statistically significant.

Firstly, the effects of metformin on cellular viability were analyzed after long-term treatment with metformin for 7 d with the MTT cell viability assay and IC₅₀ and IC₉₀ values were calculated (Figure 1(a) and Table 1). The IC₅₀ values for Ishikawa cells were 3.7 mM (IC₉₀: 14.9 mM) under normal glucose conditions and 3.0 mM (IC₉₀: 13.7 mM) in a high glucose environment. For HEC-1A cells, an IC₅₀ value of 0.7 mM (IC₉₀: 6.1 mM) was established under normal glucose conditions. No IC₅₀ and IC₉₀ values could be calculated for HEC-1A in high glucose medium due to repeated overgrowth of the substrate surface by the cells. At a concentration of ≤0.1 mM, cellular viability for both cell lines was ≥80% compared to an untreated control, irrespective of glucose levels in the media. Cellular viability dropped to ≤10% at concentrations higher than 5.0 (HEC-1A) and 10 mM (Ishikawa) under any tested metabolic condition in a concentration-dependent manner.

Additionally, cellular viability was determined based on ATP levels with the ATP cell viability assay and, again, IC₅₀ and IC₉₀ values were calculated (Figure1(b) and Table 1). The IC₅₀ values for Ishikawa cells were 2.1 mM (IC₉₀: 5.0 mM) under normal glucose conditions and 18.3 mM (IC₉₀: > 20.0 mM) in a high glucose environment. For HEC-1A cells, an IC₅₀ value of 2.0 mM (IC₉₀: 4.6 mM) was established under normal glucose conditions. No exact IC₅₀ and IC₉₀ values could be calculated for HEC-1A in high glucose medium due to the high remaining cellular activity at the greatest tested metformin concentration of 20.0 mM. At a concentration of ≤1.0 mM, cellular viability for both cell lines was ≥80% compared to an untreated control at normoglycemia, while higher concentrations of >5.0 (Ishikawa) and >10.0 mM (HEC-1A) were necessary to decrease cellular viability below 80% in a high glucose environment. Cellular viability dropped to ≤10% at concentrations >5.0 mM for both cell lines in a normoglycemic environment, while concentrations of 10 - 20 mM were necessary in the presence of hyperglycemic conditions. In general, the detected inhibitory concentrations for metformin were substantially higher in the ATP assay at high glucose levels. Supplementation with E2 alone did not affect cellular viability in the ATP assay (96.6% - 100.4%) (FigureS1(a)).

In order to evaluate the effects of metformin and the role of glucose on endometrial cancer proliferation during long-term treatment, the present study

examined changes in growth after treatment with low and moderate concentrations of metformin after 7 d (Figure 1(c)). Independent of the glucose concentration in the medium, the lowest metformin concentrations (0.01 and 0.1 mM) did not show any effect on the growth potential of HEC-1A or Ishikawa cells compared to untreated reference cells. Nevertheless, metformin concentrations of 0.5, 1.0 and 5.0 mM led to a decrease in proliferation of HEC-1A cells at normal glucose levels by 32% (p = 0.051), 38% (p = 0.026) and 55% (p = 0.003),

respectively. In Ishikawa cells, proliferation rates significantly dropped by 12% (p = 0.005), 25% (p = 0.042) and 55% (p = 0.027) compared to the control group, when cultured in a normoglycemic environment. However, in the presence of high glucose levels, only the highest metformin concentration (5.0 mM) was able to significantly decrease the proliferation rate of HEC-1A cells (40%; p = 0.004), equal to a 15% significantly lower effect (p = 0.047) compared to results at the same metformin concentration under normoglycemic conditions. Yet, this resistance to metformin was not observed in Ishikawa cells at high glucose concentrations, where even 0.5, 1.0 and 5.0 mM metformin decreased the proliferation rate of cells significantly by 18% (p = 0.004), 37% (p < 0.001) and 62% (p = 0.009). Therefore, Ishikawa cells did not show a glucose-dependent resistance in response to metformin treatment. Supplementation with E2 alone did not increase proliferation of HEC-1A cells (proliferation rate of 95.0 (NG) and 97.2% (HG)) (FigureS1(b)). However, pro-proliferative effects of estrogen supplementation were observed in Ishikawa cells under normo- (121.3%, p = 0.026) and particularly hyperglycemic conditions (138.1%, p = 0.015) compared to untreated control cells as well as cells treated with 0.1% (v/v) ethanol as a vehicle (p = 0.015 (NG) and p = 0.018 (HG)).

Considering the excellent indication of long-term tumor cell survival of the colony formation assay in vitro, the effects of metformin on clonogenicity of HEC-1A and Ishikawa cells were assessed when exposed in an environment with normal or high glucose levels (Figure 2). Metformin treatment caused concentration-dependent effects on the colony formation of EC cells and decreased the number and size of the colonies. In HEC-1A cells, 0.5 mM metformin reduced the colony formation by 44% under normoglycemic conditions (p = 0.021) compared to untreated reference cells, but only by 29% under hyperglycemic conditions (p = 0.041), followed by a further reduction with increasing metformin concentrations. Treatment at 1.0 and 5.0 mM metformin inhibited clonogenicity by 59% (p = 0.0096) and 80% (p = 0.004) at 5.5 mM glucose as well as 42%

(p = 0.021) and 81% (p = 0.002) at 17.0 mM glucose, respectively. Additionally, the colony formation remained significantly higher (17% difference) at 1.0 mM metformin when cultured under high glucose conditions compared to normoglycemia (p < 0.001). A similar glucose-mediated protecting effect was also observed at 0.5 mM metformin, but the 15% difference in colony formation was not significant. In Ishikawa cells, however, substantial decreases in colony formation under normo- and hyperglycemic conditions were only observed at 1.0 mM (36% and 38%, p = 0.012) and 5.0 mM metformin (86%, p = 0.001 and 82%, p < 0.001), respectively. A significant difference in the proliferation rate at different glucose concentrations was observed at 0.5 mM metformin (p = 0.046) also in Ishikawa cells. E2 supplementation alone did not generate significant changes in clonogenicity of HEC-1A (104.3 (NG) and 102.8% (HG)) and Ishikawa cells (102.7 (NG) and 100.1% (HG)) (FigureS1(c)).

Considering the doubling time of HEC-1A and Ishikawa cells of 27 - 29 h, motility of EC cells after long-term metformin treatment was evaluated by a wound healing assay after 24 h to minimize the influence of proliferation on the out

come of the assay (Figure 3). At 24 h after scratching, untreated HEC-1A controlcells covered the scratch up to 71% and 56% at normal and high glucose levels, respectively, while Ishikawa control cells migrated into the scratch wound by 16% and 22%. Metformin treatment at 0.5 mM significantly reduced the migration ability of HEC-1A cells by 18% (p = 0.031) in a normoglycemic environment. However, in a high glucose environment, the migration was reduced by only 11%. Under these conditions, Ishikawa cells showed a minor reduction of motility by only 6% - 7% with 0.5 mM metformin at both glucose concentrations, which was significant only under high glucose conditions (p = 0.025). The highest effect of metformin was detected at 5.0 mM metformin, where the treatment significantly inhibited the motility of HEC-1A cells by 53% (p = 0.004) and 35% (p = 0.009) at 5.5 and 17.0 mM glucose, respectively. In Ishikawa cells, a drop of the motility rate by 25% (p = 0.019) was observed at high glucose levels with 5.0 mM metformin, resulting in a complete inhibition of cellular motility. The effect of metformin at 5.0 mM on the motility of Ishikawa cells could not be evaluated under normoglycemic conditions due to the fact that cells were not able to build up the required monolayer over a period of 7 d. Nevertheless, this finding highlights the inhibitory effect of high metformin concentrations on the migration capacity of Ishikawa cells at physiological glucose levels. Additionally, it should be noted that a significantly decreased wound closure was observed between untreated control groups under different glucose conditions in the HEC-1A cell line (p = 0.043), indicating an inhibiting effect of high glucose levels on motility.