Interaction between insulin and androgen signalling in decidualization, cell migration and trophoblast invasion in vitro

Abstract Finely tuned decidualization of endometrial stromal fibroblasts into decidual cells is crucial for successful implantation and a healthy pregnancy. Both insulin and androgens are known to modulate decidualization, however, their complex effect on this process has not been fully elucidated. As hyperinsulinemia and hyperandrogenism are associated in clinical conditions, we aimed to investigate the interaction between insulin and androgens on decidualization. Primary human endometrial stromal cells were decidualized in vitro in the presence of insulin and/or androgens (dihydrotestosterone (DHT), testosterone). Gene or protein expressions of decidualization markers were measured, and cells size characteristics were determined. Migration of decidualizing endometrial stromal cells and invasion of HTR‐8/SVneo trophoblast spheroids were assessed. We found that insulin and androgens in combination enhanced the upregulation of several decidualization markers including prolactin, tissue factor, tissue inhibitor of matrix metalloproteinase 3 and connexin‐43, and also interacted in modulating cell size characteristics resulting in enlarged decidualizing cells. However, insulin and DHT together restricted the migration of decidualizing cells and invasion of trophoblast spheroids. Our findings suggest that insulin and androgens interact to potentiate the process of decidualization. On the other hand, inhibited cell migration and trophoblast invasion might negatively impact the function of decidualizing endometrial stromal cells.


| INTRODUC TI ON
Decidualization denotes the morphological and biochemical differentiation of endometrial stromal cells in the secretory phase of the menstrual cycle. During decidualization, spindle-shaped fibroblastlike endometrial stromal cells enlarge and acquire epithelioid characteristics, which is associated with enlarged nucleus, increased number of nucleoli, accumulation of glycogen and lipid droplets, dilatation of the rough endoplasmic reticulum and the Golgi systems and increased number of gap junctions between the neighbouring cells. 1,2 This transformation is of importance for embryo implantation and the formation of a functional feto-maternal interface as it controls trophoblast proliferation, migration and invasion. 3 A variety of factors are secreted from decidualized cells and have been proposed to serve as putative markers of decidualization, including prolactin (PRL), insulin-like growth factor binding protein 1 (IGFBP1), tissue factor (TF), tissue inhibitor of matrix metalloproteinase 3 (TIMP3), prokineticin 1 (PROK1) and the main gap junction protein connexin-43 (CX43). 4 Impaired decidualization is considered to be associated with reproductive disorders, including decreased implantation, recurrent miscarriage and placenta-related disorders. 5 The role of insulin in decidualization has been investigated in several studies, however, its effect on decidualization markers has shown varying results. Thus, it has been demonstrated that insulin downregulates IGFBP1 in decidualizing human endometrial stromal cells, [6][7][8] whereas PRL production is stimulated. 9 We have furthermore reported that insulin downregulates a number of decidualization markers via the transcriptional inactivation of Forkhead box protein 1 (Foxo1), a crucial transcription factor in decidualization, 8 whereas PROK1 is highly enhanced by insulin in decidualizing human endometrial stromal cells. 10 However, it was recently shown that adequate insulin signalling via insulin receptor substrate-2 supports the decidualization process. 7 Androgens may also play a role for decidualization. [11][12][13] In the endometrium, the androgen receptor (AR) expression is confined to the stroma and it fluctuates during the menstrual cycle with a gradual decrease from the early proliferative to the mid-secretory phase. 14 Furthermore, endometrial expression levels of enzymes that play a role in the biosynthesis and conversion of androgens are higher in the secretory phase of the menstrual cycle, suggesting a role of locally synthesized androgens in decidualization. 14,15 It was recently demonstrated that androgens (testosterone and dihydrotestosterone (DHT)) enhance the expression of decidualization markers such as PRL and IGFBP1 and promote the morphological and ultrastructural changes associated with the decidualization process. 1,15,16 Hyperinsulinemia and hyperandrogenism are common clinical conditions in polycystic ovary syndrome (PCOS) and obesity.
These conditions are related to implantation failure, increased risk of miscarriage and adverse pregnancy outcomes. [17][18][19][20] However, the mechanisms by which hyperinsulinemia and/or hyperandrogenism weaken the chance for a successful implantation and an uncomplicated pregnancy are still not elucidated. The purpose of the present study was to investigate the in vitro interaction between insulin and androgens on the decidualization process, focusing on the functional changes in decidualizing endometrial stromal cells.

| Subjects
Regularly cycling, healthy volunteers (n = 9) underwent collection of endometrial biopsy under local anaesthesia using a suction curette (Pipet Curet, CooperSurgical) in the proliferative phase of the menstrual cycle at cycle day 5-9. All women were between 18 and 35 years and had a body mass index between 19 and 28. Exclusion criteria were hormonal medication within 3 months prior to biopsy sampling, smoking, endocrine disorder, current chronic disease or continuous medication. They gave their written informed consent and the Regional Ethical Committee in Stockholm approved the study (DNR 2018/2199-31).

| Isolation of human endometrial stromal cells and culture conditions
Isolation of human endometrial stromal cells was carried out as described previously. 8 Endometrial stromal cells were seeded to 6-well Costar plates (Sigma-Aldrich) at a density of 10 5 /well and cultured in DMEM/F12-Glutamax medium (Thermo Fischer Scientific) supplemented with 10% heat-inactivated foetal bovine serum (HI-FBS; Sigma-Aldrich) and 0.2% penicillin-streptomycin (Sigma-Aldrich) until 80%-90% confluency. In vitro decidualization of endometrial stromal cells was induced in phenol red-free DMEM/F12 (Thermo In the experiments of flow cytometry, wound-healing assay and spheroid co-culture invasion assay, we chose to treat the cells with insulin and/or DHT, but not with testosterone, since DHT has the strongest affinity to the androgen receptor 21 and does not convert to other hormones.

| RNA isolation, cDNA synthesis and RT-PCR
Total RNA was extracted using Quick-RNA Miniprep Kit (Zymo Research) and subjected to cDNA synthesis using SuperScript VILO cDNA Synthesis Kit (Thermo Fischer Scientific). Gene expression levels of PRL and IGFBP1 were determined with TaqMan method. TF and TIMP3 were measured using the SybrGreen method. Ribosomal protein L13A (RPL13A) was used as a housekeeping gene to normalize gene expression levels. Gene expression levels were analysed with the ΔΔC t method. The employed TaqMan assays (Thermo Fischer Scientific) and oligonucleotides (Sigma-Aldrich) are listed in Tables S1 and S2. All determinations were performed in triplicate. We also determined forward scatter (FSC-reflecting cell volume), side scatter (SSC-reflecting the granularity or the internal complexity of the cells) and pulse-width (reflecting cell diameter) parameters in order to approximate cell size.

| Cell size determination using light microscopy
After 5 days of decidualization, cells were detached from the culture dishes using TrypLE Express (Thermo Fischer Scientific) and microphotographs were taken with 100x magnification using a Leica DFC420 C digital camera on a Nikon Eclipse TS 100 inverted microscope. At least 100 cells were analysed in each condition of each healthy donor. Photographs were analysed using ImageJ software.

| Wound-healing assay
A wound-healing assay was used to study the migratory potential of decidualized cells after treatment with insulin, DHT or the combination of both. Endometrial stromal cells were seeded in 24-well Costar plates (Sigma-Aldrich) and cultured in DMEM/F12-Glutamax supplemented with 10% HI-FBS and 0.2% penicillin-streptomycin until confluency. The cells were decidualized as stated above in the presence or absence of 100 nM insulin, 1 μM DHT or their combination for 6 days.
The cells were scratched with a 200 μl pipette tip and washed twice with PBS. 700 μl phenol red-free DMEM/F12 media supplemented with 2% charcoal-stripped foetal bovine serum and 0.2% penicillinstreptomycin was added to each well. No decidualization agents, insulin or DHT were added during the 24 h of the wound-healing assay.
The migration of the cells was followed by an IncuCyte S3 Live-Cell Analysis System (Sartorius) using a 4x objective (whole well, phase contrast imaging) for 24 h. Photographs were analysed using ImageJ software. The experiments were performed with cells from five healthy volunteers. Each experiment was performed in duplicate.

| Formation of trophoblast spheroids and coculture invasion assay
We applied a co-culture invasion assay to investigate the invasion of decidualized cells by HTR-8/SVneo spheroids in the presence of insulin, DHT or their combination. Spheroids consisting of 3*10 3 HTR-8/SVneo, a first trimester derived immortalized trophoblast cell line, were formed as described previously. 10 The co-culture experiments were performed in 24-well Costar plates. Primary endometrial stromal cells were seeded at a density of 10 5 /well and cultured until 90% confluency. Cells were either decidualized in the presence or absence of 100 nM insulin, 1 μM DHT or their combination as described above for 6 days or left untreated. Then, media was changed to 700 μl phenol red-free DMEM/F12 supplemented with 2% charcoal-stripped FBS and 0.2% penicillin-streptomycin. One HTR-8/SVneo spheroid was carefully transferred to each well onto the confluent stromal/decidual cells using 1 ml pipette tip previously cut with sterile blade in order to widen it. No decidualization agents, insulin or DHT were added during the 16 h of the co-culture invasion assay. The invasion of HTR-8/SVneo spheroids was followed by an IncuCyte S3 Live-Cell Analysis System using a 4x objective (whole well, phase contrast imaging). The invasion areas of spheroids were measured using ImageJ. The experiments were performed with cells from five healthy volunteers.

| Statistical analysis
Statistical analysis was performed with GraphPad Prism 9.0. Two-way ANOVA for repeated measurements was used with the within group factors insulin (yes/no), DHT (yes/no) or testosterone (yes/no) and the interaction insulin*DHT or insulin*testosterone.
Considering the limited sample size and the high risk of type II error (false-negative results), we performed simple main effect tests if interactions corresponded to p < 0.15. All the data were log-transformed because of skewness. p < 0.05 was considered statistically significant.

| Gene expression levels of decidualization markers in response to insulin, androgens and combined treatment
In vitro decidualization increased all studied markers (Figure S1A-D).

| PRL
The interaction between insulin and DHT was significant, and the post hoc test showed higher PRL expression by insulin in combination with DHT than for insulin (p = 0.003) or DHT alone (p = 0.002; Figure 1 and Table 1). Furthermore, insulin increased PRL expression compared with no treatment (p = 0.038). There was no interaction between insulin and testosterone, but a significant main effect for both insulin and testosterone (p = 0.030 and p = 0.000, respectively ( Figure 1 and Table 1).

| IGFBP1
There was no interaction or significant main effects of insulin or DHT on IGFBP1 gene expression (Table 1 and Figure S2). Furthermore, there was no significant interaction between insulin and testosterone. However, testosterone increased IGFBP1 (p = 0.005) but not insulin (Table 1 and Figure S2).

| TF
There was an interaction between insulin and DHT, and the post hoc test showed higher TF expression by the combined insulin and DHT treatment compared with DHT alone (p = 0.005). Furthermore, both insulin and DHT increased TF expression compared with no treatment (p = 0.000 and p = 0.004, respectively; Table 1 and Figure S2).
There was no interaction between insulin and testosterone, but both insulin and testosterone increased TF (p = 0.000 and p = 0.006, respectively; Table 1 and Figure S2).

| TIMP3
There was no significant interaction between insulin and DHT, however, both increased the gene expression of TIMP3 independently of each other (p = 0.001 and p = 0.016, respectively; Table 1 and Figure S2). The effect of testosterone exposure was similar to that of DHT treatment (Table 1 and Figure S2).

| Protein expression levels of the gap junction protein CX43 in response to insulin, DHT and their combined treatment
In vitro decidualization did not significantly affect CX43 protein expression (not shown). There was an interaction between insulin and DHT, and the post hoc test showed a significantly higher protein expression of CX43 when cells were treated with insulin and DHT in combination compared with DHT alone (p = 0.002). Insulin alone, also increased CX43 compared with no treatment (p = 0.007; Table 1 and Figure S3).

| Cell size determination
The interaction between insulin and DHT was significant for most parameters of cell characteristics ( Table 2). Insulin in combination with DHT increased FSC-A (reflecting cell volume) compared with insulin alone (p = 0.001), but not insulin or DHT alone (Table 2 and Figure S4). Furthermore, the combined treatment increased SSC-A (reflecting cell complexity) compared with DHT alone (p = 0.000), and insulin alone increased SSC-A compared with no treatment (p = 0.005; Table 2 and Figure S4). Insulin in combination with DHT also increased pulse-width (reflecting cell diameter) compared with insulin and DHT alone (p = 0.0058 and p = 0.0067, respectively; Table 2 and Figure S4). Similarly, insulin in combination with DHT increased cell size compared with insulin and DHT alone (p = 0.002 and p = 0.020, respectively; Table 2 and Figure S4).    Table S3).  Figure 3A-N, Videos S1-S4 and Table S3).

| DISCUSS ION
This is the first study demonstrating evidence of interaction be- productions, whereas IGFBP1 is downregulated in decidualizing endometrial stromal cells. [6][7][8][9][10] Here, we found that both insulin and androgens enhanced PRL, and testosterone also increased IGFBP1, whereas insulin had no significant effect on IGFBP1. In addition, we have shown that the decidualization markers TF and TIMP3 were upregulated by insulin and androgens.
In these experiments, we tested both DHT and testosterone, however, since there were no major differences in the actions of these two androgens on decidualization markers, we chose to use only DHT in the following experiments. This was because DHT has a 2-fold higher affinity to the androgen receptor than testosterone. 21 Furthermore, testosterone can exert diverse effects due to its conversion to both oestrogen and DHT in decidualizing endometrial stromal cells, via the action of aromatase and 5α-reductase type 1, respectively. 22 The main findings in our study are the interactions between in-  Several putative factors are involved in the regulation of endometrial stromal cell migration and trophoblast invasion including hormones, growth factors, chemokines and inflammatory mediators. 33 Wongwananuruk and collaborators studied the role of androgen signalling and demonstrated that DHT alone had no significant effect on trophoblast invasion in a co-culture system using HTR-8/ SVneo trophoblast cells and decidualizing endometrial stromal cells. 31 We found that combined treatment with insulin and DHT in-

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings are available from the corresponding author upon reasonable request.