PGF2α induces a pro-labour phenotypical switch in human myometrial cells that can be inhibited with PGF2α receptor antagonists

Preterm birth is the leading cause of infant morbidity and mortality. There has been an interest in developing prostaglandin F2α (PGF2α) antagonists as a new treatment for preterm birth, although much of the rationale for their use is based on studies in rodents where PGF2α initiates labour by regressing the corpus luteum and reducing systemic progesterone concentrations. How PGF2α antagonism would act in humans who do not have a fall in systemic progesterone remains unclear. One possibility, in addition to an acute stimulation of contractions, is a direct alteration of the myometrial smooth muscle cell state towards a pro-labour phenotype. In this study, we developed an immortalised myometrial cell line, MYLA, derived from myometrial tissue obtained from a pregnant, non-labouring patient, as well as a novel class of PGF2α receptor (FP) antagonist. We verified the functionality of the cell line by stimulation with PGF2α, resulting in Gαq-specific coupling and Ca2+ release, which were inhibited by FP antagonism. Compared to four published FP receptor antagonists, the novel FP antagonist N582707 was the most potent compound [Fmax 7.67 ± 0.63 (IC50 21.26 nM), AUC 7.30 ± 0.32 (IC50 50.43 nM), and frequency of Ca2+ oscillations 7.66 ± 0.41 (IC50 22.15 nM)]. RNA-sequencing of the MYLA cell line at 1, 3, 6, 12, 24, and 48 h post PGF2α treatment revealed a transforming phenotype from a fibroblastic to smooth muscle mRNA profile. PGF2α treatment increased the expression of MYLK, CALD1, and CNN1 as well as the pro-labour genes OXTR, IL6, and IL11, which were inhibited by FP antagonism. Concomitant with the inhibition of a smooth muscle, pro-labour transition, FP antagonism increased the expression of the fibroblast marker genes DCN, FBLN1, and PDGFRA. Our findings suggest that in addition to the well-described acute contractile effect, PGF2α transforms myometrial smooth muscle cells from a myofibroblast to a smooth muscle, pro-labour–like state and that the novel compound N582707 has the potential for prophylactic use in preterm labour management beyond its use as an acute tocolytic drug.


Introduction
Every year, 14.9 million babies are born preterm (<37 weeks gestation), representing 11.1% of live births (Chawanpaiboon et al., 2019).Preterm birth (PTB) is the leading global cause of infant morbidity and mortality, accounting for 18% of all deaths in children aged under 5 years old (Goldenberg et al., 2008;Mangham et al., 2009;Walani, 2020).The current treatment for treating lifethreatening PTB is tocolytics, which include the calcium channel blocker, nifedipine, and the oxytocin receptor antagonist, atosiban (Vogel et al., 2014;Lamont and Jørgensen, 2019).Whilst these tocolytics can delay labour for ≤48 h, they do not prevent PTB and have shown limited improvement in both short-term and long-term neonatal outcomes (Schwarz and Page, 2003;Vogel et al., 2014;Lamont and Jørgensen, 2019).
The binding of PGF 2α to the FP receptor activates phospholipase C via Gα q , which converts the membrane-bound PIP 2 to IP 3 and DAG.The released IP 3 binds to IP 3 receptors on the sarcoplasmic reticulum, opening Ca 2+ channels and increasing cytoplasmic concentrations of Ca 2+ (Davis et al., 1987;Silvia and Homanics, 1988;Haddock and Hill, 2002).Ca 2+ binds to calmodulin, which subsequently phosphorylates myosin lightchain kinase, leading to force generation and contraction (Kamm and Stull, 1985;Horowitz et al., 1996;Berridge et al., 2003;Wray and Prendergast, 2019).
In rodents, PGF 2α initiates labour by regressing the corpus luteum and reducing systemic progesterone concentrations (Gross et al., 1998).The precise role of PGF 2α in active human labour remains undetermined as humans do not experience reductions in systemic progesterone but instead may experience a "functional" progesterone withdrawal (Csapo and Pinto-Dantas, 1965;Merlino et al., 2007).However, maternal plasma levels of PGF 2α increase in the third trimester prior to the onset of labour and increase further as labour progresses (reviewed by Wood et al., 2021).Therefore, while PGF 2α can stimulate myometrial contractility, it may also play a role in uterine activation and the initiation of parturition.
In this study, we used our newly derived myometrial cell line, MYLA, to assess the novel FP antagonist, N582707 (Figure 1), in comparison to four compounds from the literature (Table 1).OBE002 is the parent compound of prodrug OBE022 (ebopiprant), which has been used for the treatment of preterm labour and is currently involved in Phase II clinical trials (Pohl et al., 2019).Compound 46/47 and compound 39/40 were investigated as potential treatments for idiopathic pulmonary fibrosis, with BAY-6672 (derived from compound 46) exhibiting anti-inflammatory and antifibrotic effects in induced pulmonary fibrosis mouse models (Beck et al., 2020).Similarly, compound 68 was designed for the treatment of inflammation and proved to be potent in FLIPR functional assays in transfected HEK-293 cell lines (Martos et al., 2016).We demonstrate that both Ca 2+ release and Gα q coupling are inhibited by FP antagonists, with the novel FP antagonist, N582707, being more potent when compared to the previously published FP antagonists.Furthermore, through RNA-sequencing (RNA-Seq), we determined that PGF 2α causes significant time-dependent changes in mRNA transcription, stimulating a phenotypic switch from a fibroblastic-like phenotype to a smooth muscle-like phenotype with an increase in the number of expressed pro-labour mRNAs.This effect was reversed by FP antagonism.This supports the hypothesis that PGF 2α is not only important during labour to stimulate uterine contractions but also plays a significant role in transforming the myometrium from a quiescent to an activated state.

Results
To test the basic physiological response of our novel immortalised myometrial cell line (MYLA), we characterised FP receptor signalling by measuring intracellular Ca 2+ concentrations post PGF 2α stimulation.Increasing

FP antagonist Chemical name Medical application References
Compound   (Phillippe et al., 1997).
PGF 2α stimulates FP receptor coupling to Gα q/11 in MYLA cells It has previously been demonstrated that the FP receptor couples specifically to the Gα q subunit in mice, rats, and Chinese hamster ovary cells (Davis et al., 1987;Ito et al., 1994;Engstrøm et al., 2000;Le Gouill et al., 2010).To further characterise the MYLA cells, we investigated the coupling of the FP receptor to Gα-subunits by utilising the [35S]-GTPγS immunoprecipitation assay.PGF 2α significantly increased [35S]-GTPγS binding to Gα q/11 only (Figure 3).However, PGF 2α is promiscuous, binding not only to the FP receptor but also to other prostanoid receptors with relatively high affinity (Abramovitz et al.,  Equation = log (agonist) vs. response-variable slope (four parameters) + constrain bottom to 0. Outliers were identified using Grubb's (alpha = 0.05).Data are mean ± SD, N = 3.

2000)
. To demonstrate FP specificity, coupling to Gα q/11 was reversed by administration of the FP antagonist compound 39/40.
The novel FP antagonist, N582707, is more potent than comparator FP antagonists Using our MYLA cell line, we sought to assess the novel FP antagonist, N582707 (Figure 1), in comparison to four compounds from the literature (Table 1) on PGF 2α -induced Ca 2+ oscillations.MYLA cells were incubated with 1 µM PGF 2α (to achieve ~99% FP receptor occupancy and to reflect physiological concentrations) in the presence of 10 half-log incremental concentrations: 300 pM to 10 µM of FP antagonists.The results demonstrated that the FP antagonists inhibited PGF 2α -stimulated Ca 2+ release as determined by F max , AUC, and frequency (Figure 4; Table 3).Of the five tested FP antagonists, N582707 was the most potent for all tested criteria, with pIC 50 values in a nanomolar range [F max 7.67 ± 0.63 (IC 50 21.26nM), AUC 7.30 ± 0.32 (IC 50 50.43nM) and frequency of Ca 2+ oscillations 7.66 ± 0.41 (IC 50 22.15 nM)].
PGF 2α stimulates the development of a prolabour phenotype and induces a phenotypical switch from a myofibroblastic to a smooth muscle phenotype in MYLA cells It has previously been reported that PGF 2α is not only involved in the contractile phase of labour but also in the activation of parturition (Xu et al., 2013;Xu et al., 2015).To assess if PGF 2α stimulates the development of a pro-labour phenotype, MYLA cells were treated with either 1 µM PGF 2α or ethanol vehicle control and then harvested at seven time points after treatment.Principal component analysis (PCA) plot PC1 (Figure 5A) demonstrates a difference in grouping by time, defined as 'early' (1 h, 3 h, 6 h, 9 h, and 12 h) or 'late' (12 h and 24 h).PC2 captures a difference in grouping by treatment (control vs. PGF 2α ).Together, PC1 and PC2 demonstrate that the data were clean and clustered well, describing 50% of transcriptomic variation.PCA plot PC3 (Figure 5B) captured a second grouping of mRNA changes by time, whereby 'early' (1 h and 3 h) and 'late' (24 h and 48 h) time points are grouped, and 'mid' (6 h, 9 h, and 12 h) time points are grouped.Differential gene expression analysis showed that with an increase in time, there was an increase in the number of differentially expressed genes (Figure 6).
To investigate further changes in the mRNA levels, we generated Z-scores of the transcripts per million (TPM) values for the >2,500 differentially expressed genes.A heat map of the top 25 most significant differentially expressed genes as compared to their 1-h time point is depicted in Figure 7.As suggested by the PCA, there was a clear separation between PGF 2α treatment and vehicle control.Several key genes involved in labour, such as OXTR, were upregulated in the PGF 2α treatment group as well as upregulated over time.Also prominent was an upregulation of pro-labour genes that are associated with leucocyte infiltration and inflammation such as IL6 and IL11.Conversely, several genes such as CXCL12, ALDH1A3, and CPA4 decreased in expression with time and treatment.
The RNA-Seq data were then cross-referenced with a publicly available data set (WikiPathways 2021 human) to determine the top 10 enrichment terms (Table 4).The most enriched term was 'myometrial relaxation and contraction,' suggesting that PGF 2α alone was stimulating the development of genes associated with a pro-labour phenotype as well as a smooth muscle phenotype (e.g., CALD1 and ACTA2).The RNA-Seq data were then cross-referenced with the data set generated by Chan et al. (2014) that measured the transcriptome differences in human myometrial samples prior to and after the onset of spontaneous labour (Table 5).Comparatively, over time, the number of differentially expressed genes per time point in both data sets increased from 26 to 141 common genes and smooth muscle markers from 2 to 18 common genes, implicating again that PGF 2α alone was stimulating not only a pro-labour phenotype but also the MYLA cells to differentiate into a smooth muscle phenotype.

FP antagonism prevents PGF 2α -stimulated activation of the myometrium
To determine that the development of a pro-labour phenotype and a smooth muscle phenotype is specifically FP receptormediated, additional time series data were conducted using MYLA cells treated with 1 µM PGF 2α or 1 µM PGF 2α + 1 µM N582707 or 1 µM PGF 2α + 1 µM compound 39/40.Four fibroblast marker genes (DCN, LOXL1, FBLN1, and PDGFRA-Figure 8) and 10 smooth muscle marker genes (ACTA2, CNN1, COL4A1, COL4A2, MYOCD, TAGLN, TGFB2, TGFB3-Figure 9; CALD1 and MYLK-Figure 10) were analysed.The expression of all 10 smooth muscle markers was lower in both the control and FP antagonist-treated groups, while the expression of the four fibroblastic markers was higher in the control and FP antagonist-treated groups, as summarised in Figure 11.This indicates that PGF 2α initiates a phenotypical switch in the MYLA cells, whereby they undergo differentiation from a myofibroblastic to a smooth muscle phenotype via FP receptor signalling.
The pro-labour gene, OXTR, and two smooth muscle markers, CALD1 and MYLK, were used to validate these results using RT-qPCR.This determined that both N582707 and compound 39/40 reduced the expression of OXTR, CALD1, and MYLK (Figure 10).When specifically looking at the difference between gene expression at 48 h, there was a significant decrease in OXTR, CALD1, and MYLK expression when treated with either N582707 or compound 39/40 (Figure 12).Overall, this demonstrates that PGF 2α via the FP receptor could initiate activation of the myometrium prior to labour and implies a potential role for FP antagonism in the prophylactic management of PTB in addition to the known effects on uterine contractility.(1 h, 3 h, 6 h, 9 h, and 12 h) or 'late' (24 h and 48 h).PC2 captures a difference in grouping by treatment (control vs. PGF 2α ).In total, PC1 and PC2 capture 50% of transcriptomic variation.(B) PCA plot PC3 captures a second grouping by time, whereby 'early' (1 h and 3 h) and 'late' (24 h and 48 h) time points are grouped, and 'mid' time points (6 h, 9 h, and 12 h) are grouped.

FIGURE 6
Number of differentially expressed genes when comparing PGF 2α vs. control data.RNA-Seq data was collected from MYLA cells treated with either 1 µM PGF 2α or vehicle equivalent at time 0 followed by harvesting cells at 1 h, 3 h, 6 h, 9 h, 12 h, 24 h, and 48 h.The number of differentially expressed genes increased with time in two phases, 1 h to 6 h and then from 24 h to 48 h.

FIGURE 7
Differential gene expression in MYLA transcriptomes.RNA-Seq data was collected from MYLA cells treated with either 1 µM PGF 2α or vehicle equivalent at time 0 followed by harvesting cells at 1 h, 3 h, 6 h, 9 h, 12 h, 24 h, and 48 h.Heatmap of the top 25 differentially expressed genes in the control and PGF 2α -treated cells as compared to their 1 h time point, identified using DESeq2.

Discussion
It is commonly accepted that PGF 2α is involved in several events during human parturition, such as stimulating cervical ripening, rupturing the foetal membranes, and playing an important role during the final stages of parturition by regulating uterine contractility (Maclennan and Green, 1979;Casey and MacDonald, 1988;Senior et al., 1993;Lee et al., 2009).While evidence exists that PGF 2α can upregulate the expression of uterine activation proteins, an in-depth analysis of the effect of PGF 2α upon the myometrium is yet to be elucidated (Xu et al., 2013;Xu et al., 2015).Furthermore, the FP receptor is expressed in a limited number of human tissues (eye and myometrium) (Matsumoto et al., 1997;Mukhopadhyay et al., 2001) and is involved in pulmonary fibrosis (Oga et al., 2009).This makes the FP receptor an attractive target for the development of novel therapies for PTB.
This study was a first look at several comparator FP antagonists, of which only OBE002 has been investigated for the treatment of PTB, in comparison to a novel FP antagonist, N582707.We demonstrated that our MYLA cell line derived from myometrial tissue obtained from a pregnant, non-labouring woman was a suitable human myometrial model.Stimulation with PGF 2α activated Gα q -specific G-protein coupling with no effect on non-FP receptor-specific Gα s and Gα i coupling in addition to a concentration-dependent increase in intracellular Ca 2+ .Both Gα q coupling and Ca 2+ release were inhibited by FP antagonists, with the novel FP antagonist N582707 being the most potent when compared to several known FP antagonists.This highlighted the potential of N582707 to be used as a tocolytic treatment for PTB management.
Prior to the onset of labour, the myometrium must undergo a process of activation, whereby the muscle becomes more electrically excitable and susceptible to pro-contractile hormones (Blanks and Brosens, 2012).This is mediated by an increased expression of contraction-associated genes such as OXTR, PTGS2, and CX43 (Blanks and Brosens, 2012;Xu et al., 2013).Therefore, in addition to a tocolytic effect, we sought to determine if PGF 2α could initiate activation of the myometrium and, furthermore, if treatment with N582707 could inhibit this activation.Using RNA sequencing, we determined that PGF 2α stimulates a time-dependent phenotypic transformation of MYLA cells from a myofibroblast-like phenotype to a smooth muscle-like phenotype and increases the number of pro-labour genes.When cross-referencing our transcriptomics data to fresh human samples from term not in labour and spontaneous labour from data collected by Chan et al. (2014), we found a significant overlap in the mRNA signature seen in our PGF 2α -stimulated MYLA time course.FP antagonism inhibited this phenotypical switch, supporting the hypothesis that PGF 2α not only is important during labour to stimulate contractions but also plays a critical role in genetically transforming the myometrium from a quiescent to an activated state.Therefore, these data demonstrate that in addition to using FP antagonists as tocolytics, FP antagonism could be used prophylactically to prevent the maturation of the myometrial smooth muscle to a pro-labour phenotype.
While this time series provided insight into the changes elicited by PGF 2α in MYLA cells, we did not include an analysis on the mRNA transcriptome in the presence of progesterone.In a study by Madsen et al. (2004), it was determined that prostaglandins such as PGF 2α have the potential to induce functional progesterone withdrawal by modulating progesterone receptor isoform expression.Therefore, it would be beneficial to develop this labour model by the addition of other hormones such as progesterone to more closely simulate the changes that occur during pregnancy.Furthermore, in future studies, it will be important to demonstrate functional changes in the myometrial cell phenotype that are reversible with FP antagonism.
We hypothesise that in addition to the importance of PGF 2α in the progression of labour, PGF 2α may also be important in the activation of the myometrium by inducing a phenotypical switch in myometrial cells, causing the development of a smooth muscle phenotype and upregulating pro-labour genes.This suggests not only a therapeutic potential for FP receptor antagonists as a tocolytic treatment for PTB but also the potential of using FP antagonists prophylactically to prevent premature activation of the myometrium.

Cell culture
Primary myometrial cells (MYLA) were established from a myometrial sample obtained with informed consent from a pregnant woman at 38 weeks gestation undergoing elective caesarean section for breech presentation at term not in labour.Following delivery of the baby and prior to delivery of the placenta, a full-thickness myometrial biopsy was taken, prior to Syntocinon bolus, from the upper lip of the lower uterine segment incision in the midline.The sample was placed in modified Krebs-Henseleit solution (mM): NaCl, 133; KCl, 4.7; glucose, 11.1; MgSO 4 , 1.2; KH 2 PO 4 , 1.2; CaCl, 2.5; 2-[[1,3-dihydroxy-2-(hydroxymethyl)propan-2-yl]amino] ethanesulfonic acid, 10; pH, 7.4).Primary myocytes were isolated by digestion in 2 mg/mL collagenase (Type IV, Fisher Scientific, Loughborough, United Kingdom) in Dulbecco's modified Eagle's medium for up to 1 h at 37 °C.The cells were released by trituration through fire-polished glass pipettes.Freshly isolated myocytes were plated, prior to transformation and selection, in Dulbecco's modified Eagle's medium supplemented with 10% foetal calf serum, penicillin (100 IU/mL), and streptomycin (100 μg/mL).
Transformation: TEFLYA cells producing retroviruses either hTERT or a temperature-sensitive mutation of SV40 U19tsA58 Δ89-97 were cultured in DMEM supplemented with 10% FBS and antibiotics (penicillin/streptomycin plus 100 ug/ mL hygromycin for TEFLYA hTERT and 1.5 mg/mL G418 for TEFLYA SV40 U19tsA58 Δ89-97) at 37 °C and 5% CO 2 (O'Hare et al., 2001).Virus-containing supernatants were harvested after growing near confluent cultures in a T75 flask for 12 h in a 10 mL growth medium without antibiotic selection.Supernatants from both virus-producing lines were filtered through a 0.45-µm filter and mixed 1:1, and this virus stock was used immediately for the transduction of primary cells.
The primary myometrial cells were cultured in Dulbecco's modified Eagle medium (DMEM)/F12 supplemented with 10% foetal calf serum, penicillin (100 IU/mL), and streptomycin (100 μg/mL) in a T75 flask.For virus transduction, the medium was removed and replaced with 5 mL of virus stock plus 5 mL of fresh growth medium and 16 µL of 5 mg/mL polybrene (8 μg/mL final concentration).The medium was replaced the next day with 14 mL of fresh growth medium and allowed to grow for 3 days, passaging into a new flask as required.Four days after transduction, the cells were seeded at low density into 14-cm Petri dishes, and 0.25 μg/mL G418 and 30 μg/mL hygromycin B were added for selection of stable transformants.Three days after the start of the selection, the cells were cultured at a permissive temperature of 33 °C to activate SV40 large T antigen.Individual colonies were picked using cloning discs and transferred into a 96-well plate after 2-4 weeks and expanded into 24-well and 6-well plates.Conditionally immortalised MYLA cells were maintained for rapid proliferation at 33 °C and analysed at 37 °C when the large T antigen was inactive.The cells were authenticated with ASN-002 short tandem repeat (STR) profiling by Eurofins Genomics Europe Applied Genomics GmbH (Supplementary Materials S3) to provide a reference for future maintenance of cell purity and genomic integrity.The cells were subcultured at 90% confluency by lifting with 0.05% trypsin and not used beyond passage 12.

Literature FP antagonists
Literature FP antagonists were synthesised using the described literature methodologies or purchased from a commercial supplier.N582707 was synthesised by Ferring Research Institute Inc. (San Diego, CA, United States) using the described methodologies.All compounds were solvated in DMSO to 10 mM and then diluted prior to use in biological assays.The chemical names and literature references (Martos et al., 2016;Pohl et al., 2018;Beck et al., 2020) are listed in Table 1.

Synthesis of novel FP antagonist N582707
A complete schema for the synthesis of N582707 is shown in Figure 1.All reactions were carried out in an oven-dried roundbottomed flask under an inert nitrogen atmosphere with stirring.Solvents, reagents, and chemicals were purchased from various sources and used as received unless otherwise noted.Spectra for 1 H were recorded at room temperature with a Bruker PA BBO 400S1 BBF-H-D-05 Z SP (400 MHz) spectrometer or a Varian ASW probe (400 MHz) unless otherwise noted.Chemical shifts are reported in δ (ppm) relative units to residual solvent peaks CDCl3 (7.26 ppm for 1 H) and DMSO-d6 (2.50 ppm for 1 H and 39.5 ppm for 13 C).Splitting patterns are assigned as s (singlet), d (doublet), t (triplet), multiplet (m), and dd (doublet of doublet).Mass spectrometry measurements were recorded using the Vanquish Horizon UHPLC System connected to Thermo Orbitrap Q Exactive Plus in high-resolution positive mode.The predicted masses were extracted to ±5 ppm.
Tetrahydrothiophen-3-ol (B): To a mixture of A (20 g, 195.78 mmol, 16.67 mL, 1 eq) in THF (200 mL) was added BH 3 -THF (1 M, 195.78 mL, 1 eq) in one portion at 0 °C under N 2 .The mixture was stirred at 20 °C for 12 h.TLC (petroleum ether:ethyl acetate = 3:1, Rf = 0.40) showed that the reaction was complete.The residue was poured into MeOH (10 mL) and stirred for 30 min.The aqueous phase was extracted with EtOAc (3 × 10 mL).The combined organic phase was washed with brine (3 × 10 mL), dried with Na 2 SO 4 , filtered, and concentrated under vacuum.The residue was purified by silica flash column chromatography to afford 18.0 g (88% yield) of B as a white solid.
Tetrahydrothiophene-3-carbonitrile (D): To a mixture of C (29.5 g, 161.86 mmol, 1 eq) in DMF (300 mL) was added sodium cyanide (39.66 g, 809.28 mmol, 5 eq) in one portion at 70 °C under N 2 .The mixture was stirred at 70 °C for 12 h.TLC (petroleum ether: ethyl acetate = 5:1, Rf = 0.35) showed that the reaction was complete.The residue was poured into water (10 mL) and stirred for 3 min.The aqueous phase was extracted with MTBE (3 × 20 mL).The combined organic phase was washed with brine (3 × 20 mL), dried with Na 2 SO 4 , filtered, and concentrated under vacuum.The residue was purified by silica flash column chromatography to afford 10.8 g (59% yield) of D as a white solid.
N-methoxy-N-methyltetrahydrothiophene-3-carboxamide (E): A solution of D (5.25 g, 46.39 mmol, 1 eq) in EtOH (10 mL) was added to a solution of NaOH (19.30 g,482.42 mmol,10.4 eq) in H 2 O (223.1 mL) and EtOH (112.9 mL).The mixture was stirred at 100 °C for 3 h.TLC (petroleum ether: ethyl acetate = 5:1, Rf = 0) showed that the reaction was complete.The reaction was quenched with 1 M of hydrochloric acid to adjust pH to 3. The mixture was concentrated under vacuum.The mixture was dissolved in DCM (20 mL), dried with MgSO 4 , filtered, and concentrated under vacuum.The crude compound was used as is for the next step.
Fluorescence was captured at a rate of one frame per second for 3 min, with agonist/antagonist injections occurring after a 30-s basal period.Then, 1 μM PGF 2α or 100 nM oxytocin was added as a positive control, and 10 µM ionomycin (Sigma Aldrich, Poole, United Kingdom) provided a GPCR-independent positive control of Ca 2+ release and determined the F max for analysis calculations.
Data analysis: Videos were visualised, and data were analysed using ImageJ, the image analysis software.Changes in fluorescence and area under the curve (AUC) of Ca 2+ oscillations were calculated using the equation: ((raw data − background) − minimum background)/averaged ionomycin F max , and then percentage-corrected to stimulated (1 µM PGF 2α ).The frequency of Ca 2+ oscillations was calculated as the average number of oscillations over time in seconds (Hz).All data were visualised using the GraphPad Prism 9 software.
Data analysis: Specific binding was determined by CPM values and expressed as a % increase over basal (unstimulated), and plotted using the GraphPad Prism 9 software.
RNA was extracted from pelleted cells using the GenElute TM Total RNA Purification Kit (Sigma Aldrich, Poole, United Kingdom) according to the manufacturer's instructions.RNA concentration and purity were determined using a NanoDrop 1000 Spectrophotometer (Thermo Fisher Scientific, Loughborough, United Kingdom).All RNA samples were stored at −80 °C until use.
cDNA generation and RT-qPCR cDNA was generated using the VILO cDNA Synthesis Kit (Invitrogen, Paisley, United Kingdom) according to the manufacturer's instructions and using the Veriti TM 96-Well Fast Thermal Cycler (Thermo Fisher, Loughborough, United Kingdom).
RT-qPCR samples were technical triplicates of biological triplicates in 364-well optical plates.Amplification was performed in 10 μL reactions containing 5 μL of 2X EXPRESS qPCR SuperMix with premixed ROX reference dye (Thermo Fisher, Loughborough, United Kingdom), 0.5 μL of each specific TaqMan primer pairprobe (listed in Supplementary Materials S9), and 1 μL of cDNA or water control.qRT-PCR was performed using the Applied Biosystems QuantStudio 5 Real-Time PCR System (qPCR) with an initial denaturation for 10 min at 95 °C, primer annealing at 50 °C for 2 min, followed by 40 cycles of 15 s at 95 °C and 1 min at 60 °C.
Data analysis: The relative expression of the target genes was calculated using the delta-CT method as described by Pfaffl (2001), normalised to the geometric mean of three housekeeping genes (GAPDH, ACTB, and RPL19), and then plotted using the GraphPad Prism 9 software.
RNA quality assessment, library preparation, and sequencing RNA quality was verified using a Bioanalyzer according to the manufacturer's instructions (Eukaryote Total RNA Nano, Agilent).Illumina TruSeq RNA libraries were prepared and sequenced using a NextSeq 500 with a high-output 75 bp cycle cartridge (Illumina, Cambridge, United Kingdom) by the University of Warwick Genomic Facility.
RNA-sequencing bioinformatics analysis bcl2fastq v2.20.0.422 was used for demultiplexing each sample.FastQC v0.11.9 was used to quality control check the demultiplexed FastQC files, and MultiQC v1.9 consolidated the QC reports.Reads were aligned to the human reference genome build GRCh38 release 86 using STAR version 2.7.9a.The number of reads mapped to each genomic feature was calculated with featureCounts v2.0.1.The counts were imported into R Studio and analysed with the DESeq2 package.
Data analysis: TPM data were processed and analysed using R Studio and then plotted using the GraphPad 9 software.

Statistical analysis
Agonist and antagonist Ca 2+ assay Experiments were repeated on MYLA cells, where N represents the number of biological repeats.As provided in Figure 2 and Supplementary Materials S2, the data were analysed by one-way ANOVA with post hoc Dunnett's multiple comparison test comparing agonist treatment to basal or antagonist treatment to 1 µM PGF 2α -stimulated cells.The data calculated as p < 0.05 were considered statistically significant and are graphically represented as ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.In Figure 4, data were analysed by two-way ANOVA with post hoc Dunnett's multiple comparison test comparing antagonist treatment to vehicle-stimulated cells.Data calculated as p < 0.05 were considered statistically significant and are represented as ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.As provided in Table 2 and Table 3, EC 50 , pEC 50 , IC 50 , pIC 50 , E max , and Hill slope values were determined by removing outliers using Grubb's (alpha = 0.05) and then plotted using [log (agonist/inhibitor) vs. response-variable response (four parameters)] and constraining the bottom to 0 using the GraphPad Prism 9 software.

[ 35 S]-guanosine 5′-O-[gamma-thio] triphosphate binding
The experiments were repeated on MYLA cell membranes, where N represents the number of biological repeats.The data in Figure 3 were analysed by one-tailed, unpaired t-tests comparing vehicle + 1 µM PGF 2α to basal and vehicle + 1 µM PGF 2α to 1 µM compound 39/40 + 1 µM PGF 2α.Data calculated as p < 0.05 were considered significant and are graphically represented as **p < 0.01.Not significant results are depicted as ns.

RNA sequencing
The experiments were technical triplicates of biological triplicate samples taken from seven time points.The data in Figures 8-10 were analysed by two-way ANOVA and post hoc Dunnett's multiple comparison test comparing treatments to the ethanol vehicle.Data calculated as p < 0.05 were considered statistically significant and are graphically represented as ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.

RT-qPCR
The experiments were technical triplicates of biological triplicate samples taken from seven time points.The data in Figure 12 were analysed by one-way ANOVA and post hoc Dunnett's multiple comparison test comparing treatments to 1 µM PGF 2α.Data calculated as p < 0.05 were considered statistically significant and are graphically represented as ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.
FIGURE 2 Effect of the FP agonist, PGF 2α, upon calcium transients in MYLA cells.(A) Representative traces showing calcium transients in MYLA cells in response to increasing concentrations of PGF 2α (300 p.m.-10 µM) plus a positive control of 100 nM oxytocin.(B) PGF 2α (black circles) concentration-response curves and unstimulated/basal (red circles).Data were analysed to assess the peak responses (F max ), the area under the curve (AUC), and peak frequency (Hz) during the 3 min of PGF 2α stimulation.Data calculated as [(raw data − background)minimum background]/averaged ionomycin F max .Data are mean ± SD, N = 3 (data were analysed by one-way ANOVA and Dunnett's multiple comparison test comparing PGF 2α treatment to basal, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001).EC 50 , pEC 50, CI, E max , and Hill slope values obtained are shown in Table 2. Representative traces for the effect of the PGF 2α vehicle (DMSO) are shown in Supplementary Materials S1.Statistical values are shown in Supplementary Materials S2.

FIGURE 4
FIGURE 4 Effect of five FP antagonists upon PGF 2α -stimulated Ca 2+ transients in MYLA cells.FP antagonist + 1 µM PGF 2α concentration-response curves.Data were analysed to assess (A) maximal fluorescence response, (B) the area under the curve (AUC), and (C) the frequency of Ca 2+ oscillation during the 3 min of stimulation.Data was calculated as [(raw data − background) − minimum background]/averaged ionomycin F max and then percentage corrected to stimulated (1 µM PGF 2α ).Data are mean ± SD, N = 3.All IC 50 , pIC 50 , CI, and HillSlope values obtained are shown in Table 3. Data were analysed by two-way ANOVA and post hoc Dunnett's multiple comparison test comparing FP antagonist treatment to vehicle, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

FIGURE 5
FIGURE 5PCA plots of RNA-Seq data to capture the largest sources of variability in MYLA transcriptomes.RNA-Seq data was collected from MYLA cells treated with either 1 µM PGF 2α or vehicle equivalent at time 0 followed by cells harvested at 1 h, 3 h, 6 h, 9 h, 12 h, 24 h, and 48 h.Colours indicate different time points, and shapes indicate different treatments; control (circles) or PGF 2α (triangles).(A) Principal component analysis (PCA) plot PC1 captures a difference in grouping by time defined as 'early' (1 h, 3 h, 6 h, 9 h, and 12 h) or 'late' (24 h and 48 h).PC2 captures a difference in grouping by treatment (control vs. PGF 2α ).In total, PC1 and PC2 capture 50% of transcriptomic variation.(B) PCA plot PC3 captures a second grouping by time, whereby 'early' (1 h and 3 h) and 'late' (24 h and 48 h) time points are grouped, and 'mid' time points (6 h, 9 h, and 12 h) are grouped.

FIGURE 10
FIGURE 10Transcriptional changes in FP agonist and FP antagonist-treated MYLA cells during a 48-h period as determined by RNA-Seq and RT-qPCR.MYLA cells were stimulated for 1, 3, 6, 9, 12, 24, and 48 h with 1 µM PGF 2α (blue), EtOH vehicle (red), or 1 µM PGF 2α and 1 µM N582707 (green).(A) CALD1 and (B) MYLK were used as representative smooth muscle cell markers, and (C) OXTR as a pro-labour marker.Expressions of the three genes were analysed over time as determined by RNA-Seq and expressed as transcripts per million (TPM).(D) CALD1, (E) MYLK, and (F) OXTR expressions over time as determined by RT-qPCR and expressed as fold change (ΔΔCt) using the geometric mean of RPL19, GAPDH, and ACTB housekeeping genes and as compared to equivalent 1 h time points.In a second time series, MYLA cells were stimulated for 1, 3, 6, 9, 12, 24, and 48 h with 1 µM DMSO + 1 µM PGF 2α (orange) or 1 µM PGF 2α and 1 µM compound 39/40 (purple).(G) CALD1, (H) MYLK, and (I) OXTR expressions over time as determined by RT-qPCR expressed as fold change (ΔΔCt) using the geometric mean of RPL19, GAPDH, and ACTB housekeeping genes and as compared to equivalent 1 h time points.Each time point was performed in triplicate and is represented as mean ± SD.Data were analysed by two-way ANOVA and Dunnett's multiple comparison test comparing treatments to EtOH vehicle control, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

TABLE 1
List of FP antagonist chemical names and medical applications.

TABLE 2
Summary statistics for the effect of the FP agonist, PGF 2α, upon MYLA cells, as depicted in Figure1.

TABLE 3
Summary statistics for the effect of the five FP antagonists upon MYLA cells, as depicted in Figure4.

TABLE 4
Top 10 most significant enrichment terms and genes expressed within those terms as defined by WikiPathways 2021 human.Number of common differentially expressed genes per time point in our RNA-Seq data and the Chan et al. (2014) data set.