In vivo effects of AZD4547, a novel fibroblast growth factor receptor inhibitor, in a mouse model of endometriosis

Abstract Endometriosis is a chronic disease, characterized by the growth of endometrial‐like cells outside the uterine cavity. Due to its complex pathophysiology, a totally resolving cure is yet to be found. The aim of this study was to compare the therapeutic efficacy of AZD4547, a novel fibroblast growth factor receptor inhibitor (FGFRI), with a well‐characterized progestin, etonogestrel (ENG) using a validated in vivo mouse model of endometriosis. Endometriosis was induced by transplanting uterine fragments from donor mice in proestrus into the peritoneal cavity of recipient mice, which then developed into cyst‐like lesions. AZD4547 and ENG were administered systemically either from the day of endometriosis induction or 2‐weeks post‐surgery. After 20 days of treatment, the lesions were harvested; their size and weight were measured and analyzed histologically or by qRT‐PCR. Stage of estrous cycle was monitored throughout. Compared to vehicle, AZD4547 (25 mg/kg) was most effective in counteracting lesion growth when treating from day of surgery and 2 weeks after; ENG (0.8 mg/kg) was similarly effective in reducing lesion growth but only when administered from day of surgery. Each downregulated FGFR gene expression (p < 0.05). AZD4547 at all doses and ENG (0.008 mg/kg) caused no disturbance to the estrous cycle. ENG at 0.08 and 0.8 mg/kg was associated with partial or complete estrous cycle disruption and hyperemia of the uteri. AZD4547 and ENG both attenuated endometriotic lesion size, but only AZD4547 did not disrupt the estrous cycle, suggesting that targeting of FGFR is worthy of further investigation as a novel treatment for endometriosis.


| Test system
Non-pregnant virgin female sexually mature C57/6 J mice (n = 124), aged 12-18 weeks, weighing 19-22 g (Envigo Ltd, UK), were housed in a room with constant controlled temperature (21 ± 2°C) and 45%-55% humidity with a 12/12-h light/dark cycle and access to food and water ad libitum. The stage of the estrous cycle was assessed by daily vaginal lavage. 14 Mice in proestrus were chosen as donors because the higher circulating E2 levels ensured the uteri were larger and more vascularised compared to other stages of the cycle. 15,16

| Experimental design
The animal work was based on a well-validated endometriosis mouse model 12 and performed under the UK Home Office Project Licence Number (70/8458). All experimental procedures and animal care were conducted in accordance to the Animal (Scientific Procedures) Act 1986 Amendment Regulations (SI 2012/3039) and the Animal Welfare Act 2006. Test system and experimental design were reported in adherence to ARRIVE guidelines. Donor mice (n = 28) were humanely killed by cervical dislocation under Schedule 1 (S1). After harvesting the uteri, a longitudinal incision along each of the two horns was performed to open the lumen exposing the endometrial layer. These uteri were then cut into approximately 2 mm 2 fragments, which were weighed. The recipient mice (n = 96) were anesthetized with 2% v/v oxygen isoflurane, their ventral area was shaved and cleaned with ethanol/betadine solution and 0.1 mg/ml of buprenorphine was administered. Body temperature and respiration were constantly monitored during the surgery, and Lacri-lube ocular lubricant was used to avoid optical dryness. An incision on the right side of the peritoneal cavity of the mice was performed, and three uterine fragments were stitched onto the parietal peritoneum using Mersilkbraided silk nonabsorbable sutures. Uterine fragments were attached unilaterally so that the endometrial layer was adjacent to the peritoneal wall of the recipient mice. Vicryl absorbable sutures were used to close the muscle layer underneath the incision and the skin, and betadine solution was applied to the wound. The mice were closely | 3 of 13 SANTORELLI ET AL. monitored after surgery, placed in an incubator with controlled temperature and easier access to food for an hour and subsequently housed in groups of six. Recipient mice were randomly divided into two main groups receiving once a day (QD) treatment with AZD4547 at the doses of 5 (n = 6), 12.5 (n = 6), 25 mg/kg (n = 6) and vehicle only (polyethylene glycol [PEG] 400) (n = 6) by oral gavage (p.o), or, ENG at the doses of 0.008 (n = 6), 0.08 (n = 6), and 0.8 mg/kg (n = 6) and vehicle only (sesame oil, n = 6) by subcutaneous injection (s.c.), starting from the day of endometriosis induction (S0, n = 48), or, 2 weeks after (S2, n = 48, when the lesions were already established), 12 for 20 days.
Dose ranges and length of treatments of AZD4547 10,17,18 and ENG 19 were chosen according to those published in the literature. AZD4547 was suitable for oral gavage due to its reported bioavailability, selectivity and anti-tumor efficacy by this route, whereas ENG was administered parenterally to avoid its rapid metabolism by the liver. This also reflected their clinical routes of administration.
AZD4547 was prepared in a 30% v/v solution of PEG 400 in deionized water adjusted to pH 4. ENG was prepared in 5% v/v absolute ethanol and 95% v/v sesame oil. Twenty days after the start of the treatment, mice were killed by cervical dislocation under S1 to evaluate the therapeutic efficacy of AZD4547 and ENG towards the establishment and development of the endometriotic lesions.
Lesions were weighed, and their volumes were measured using a caliper. Lesions were snap frozen or fixed overnight in 4% w/v paraformaldehyde (PFA) in phosphate-buffered saline (PBS), paraffin embedded, sectioned, and stained with hematoxylin and eosin (H&E) to examine their structural changes. Blind testing was only possible when lesions were harvested but could not be performed when administering the freshly prepared drugs as different routes of administration were required (same operator).

| RNA isolation and quantitative real-time PCR
Total RNA was extracted from snap frozen lesions using TRIzol reagent and mirVana kits according to the manufacturers' instructions.
F I G U R E 1 Progesterone (P4), estrogen (E2), and FGFR signaling pathways: P4 and E2 bind to their receptors PR and ERα. The binding to these receptors leads to the activation of Src kinases, which activates FGFR by phosphorylation (P). FGFR is a transmembrane receptor associated with heparan sulfate proteoglycan (HPSG), which stabilizes the FGF binding and prevents FGF from being degraded by protease enzymes. As consequence of the FGFR phosphorylation, fibroblast growth factor receptor substrate 2 (FRS2) activates the PI3K/ AKT, IκB kinase/nuclear factor kappa-light-chain-enhancer of activated B cells (IKK/NF-kB), and the MAPK signaling pathways, which can, in turn, activate ERα and PR receptors by phosphorylation. P4 and E2 can also activate transcriptional pathways, binding to specific DNA sequences (progesterone and estrogen responsive elements [PRE and ERE]) found on progesterone and estrogen responsive genes leading to activation of FGFR and lesion development in endometriosis 11 RNA quality was measured on a NanoDrop 2000 (Fisher Scientific, Loughborough, UK) before 1 µg was transcribed into cDNA using QuantiTect Reverse Transcription kits. Quantitative real-time (qRT)-PCR was performed on a Quantstudio 12 K flex real-time PCR system (Applied Biosystems, Paisley, UK) using SYBR Green Supermix and primer sequences designed by our laboratory (NCBI Primer Blast; Table S1). Samples were run in triplicate using 0.5 µl cDNA in a total reaction volume of 10 µl. Gene expression was determined using the ΔΔCT method, normalized against β-actin and GAPDH reference genes, and data were expressed as fold-change relative to vehicle.

| Data and statistical analysis
Data were analyzed using GraphPad Prism 8 and expressed as means ± standard error of the mean (SEM). Animal number per group n = 6, lesion number per group, n = 18, with the exception of the groups treated with the vehicle of AZD4547 and ENG at the dose of 0.8 mg/kg from 2 weeks after surgery, whose lesion numbers were respectively n = 15 and n = 17, due to the loss of the lesion fluid during the delicate process of excision of the cysts from the peritoneal wall and separation of the omentum. The group size was calculated using 80% power to detect 50% difference in lesion size at 5% significance level. Distribution of the data was analyzed using D'Agostino and Pearson Test. The significance between two dependent groups was evaluated using Wilcoxon test (nonparametric); the significance between independent groups was assessed using Student's t test p values <0.05 were considered statistically significant: *p < 0.05, **p < 0.01, ***p < 0.001. The data and statistical analysis comply with the recommendations on experimental design and analysis in pharmacology. 20

| AZD4547 administered from S0
Following surgical induction of endometriosis, lesions developed in all groups, having the appearance of large cysts filled with fluid, clumped together, most with omental adhesions (Table S2). Stage of estrous cycle of recipients at the time of surgery ( Figure S1) or sacrifice (Table S3; Figure S2) had no effect on final lesion size. In the AZD4547-treated groups lesions appeared smaller than the vehicle group and not always filled with fluid ( Figure 2A). Animals responded well to the treatment without any significant weight gain or loss, indicating tolerability of the drug ( Figure 2B). The increase in the weight of the lesions from the day of implantation to the day of harvest was statistically significant in the mice treated with the vehicle and AZD4547 at the doses of 5 and 12.5 mg/kg (***p < 0.001) but not in the 25 mg/kg group ( Figure 3A).
Lesion volume decreased for AZD4547-treated groups compared to vehicle groups; this corresponded to increasing dose ( Figure 3B). Compared to its vehicle, AZD4547 at 5 mg/kg was responsible for 63.3% volume reduction, 12.5 mg/kg for 62.5% and 25 mg/kg for 73%.
Lesion weight (expressed as a percentage increase in weight of the original fragment implanted, Figure 3C) was significantly higher in the AZD4547 vehicle group (**p < 0.01), in the groups treated with AZD4547 at 5 mg/kg (**p < 0.01) and 12.5 mg/kg (*p < 0.05) compared to the 25 mg/kg group. Compared to its vehicle, AZD4547 at the dose of 5 mg/kg was not responsible for lesion weight regression but for a 11.4% increase, whereas the doses of 12.5 and 25 mg/kg induced a regression of 12.2% and 83.3%, respectively. Percentage of lesions filled with fluid decreased passing from the vehicle (89%) to the drug-treated groups (78%).

| ENG administered from S0
Following surgical induction of endometriosis, large cysts filled with fluid, clumped together with omental adhesion developed in the vehicle group. In the groups treated with ENG at 0.008 and 0.08 mg/ kg, lesions appeared smaller than the vehicle group and not always filled with fluid, whereas in the 0.8 mg/kg group, no fluid-filled cysts or lesion development was observed ( Figure 4A). ENG was well tolerated, and no significant weight gain or loss was observed ( Figure 4B); however, uteri appeared hyperemic with vascular engorgement increasing with dosage of the drug compared to the vehicle group ( Figure 4C).
When ENG was administered from the day of surgery, the vehicle and the lowest dose (0.008 mg/kg) groups similarly showed no impairment of the 4-5 day estrous cycle; interruption followed by recovery of the cycle regularity was observed in the 0.08 mg/kg treatment group, whereas the dose of 0.8 mg/kg caused complete disruption of the cycle and establishment of anestrus, with observation of mucus and thickening in the vaginal smear ( Figure 4D). The lesions developed in the group treated with ENG at 0.008 mg/kg with similar histological appearance (glandular and luminal epithelium, stroma with glandular structures, a lumen filled with leukocytes) and size as seen in the vehicle group, whereas lumen reduction was shown in the 0.08 mg/kg group and no fluid-filled lumen was observed in the 0.8 mg/kg group ( Figure 4E).
The increase in weight of the lesions from the day of implantation to the day of harvest was statistically significant in the mice treated with vehicle (***p < 0.001) but not in any of the ENG treated groups ( Figure 5A).
Lesion volume decreased for ENG-treated groups compared to the vehicle; this corresponded to increasing dose (*p < 0.05 when ENG vehicle was compared to the 0.8 mg/kg group, Figure 5B).
Lesion weight (expressed as a percentage increase in weight of the original fragment implanted, Figure 5C) was significantly higher in AZD4547 Vehicle AZD4547 5mg/kg AZD4547 12.5mg/kg AZD4547 25mg/kg

Mouse weight
Week post surgery the ENG vehicle group compared to the 0.008, 0.08 (**p < 0.01), and

| AZD4547 administered from S2
The transplanted syngeneic uterine fragments were left to develop for 2 weeks post-surgery before treatment was administered. Lesions  Figure 7E).
The increase in the weight of the lesions from the day of implantation to the day of harvest was statistically significant in all the groups (***p < 0.001 in the groups treated with AZD4547 vehicle and 12.5 mg/kg, **p < 0.01 in the groups treated with 5 and 25 mg/kg, Figure 8A). Lesion volume decreased from vehicle group compared to the groups treated with AZD4547 at 12.5 (*p < 0.05) and 25 mg/ kg (**p < 0.01) ( Figure 8B). Compared to its vehicle, AZD4547 at the dose of 5 mg/kg was responsible for 44.3% volume reduction, the dose of 12.5 mg/kg for 55.6% and the dose of 25 mg/kg for 53.6%.

| ENG administered from S2
When ENG was administered from 2 weeks after endometriosis induction, the developed lesions showed no macroscopic difference in size or appearance between the groups. Development of blood filled cysts was observed in one out of the six mice treated with the dose of 0.08 mg/kg ( Figure 9A). As observed when the compound was administered from the day of surgery, no significant weight gain or loss was experienced by the animals, confirming the tolerability of the drug ( Figure 9B). However, as seen when the treatment with ENG was started from the day of endometriosis induction, uteri appeared increasingly hyperemic in the ENG treated groups compared to the vehicle group (the hyperemia seemed to increase with the increasing of the dose of the drug [ Figure 9C]).
When ENG was administered from 2 weeks after surgery, the vehicle and the lowest dose (0.008 mg/kg) groups similarly showed non impairment of the 4-5 day cycle; temporary interruption fol- smear (as observed when the drug was administered from the day of endometriosis induction, Figure 9D). No histologic difference was observed in the lesions between all groups (H&E staining confirmed the lesions well developed architecture in all groups, Figure 9E).
The increase in the weight of the lesions from the day of implantation to the day of harvest was statistically significant in all the groups (***p < 0.001, Figure 10A). Lesion volumes in the ENG treated groups were very similar and not considerably lower than the vehicle ( Figure 10B). Compared to its vehicle, ENG at the dose of 0.008 mg/kg reduced the volume only by 2%, the dose of 0.08 mg/ kg by 13.2% and the dose of 0.8 mg/kg by 6%.
Lesion weight (expressed as a percentage increase in weight of the original fragment implanted, Figure 10C) was not significantly higher in the ENG vehicle group compared to the 0.008 and 0.08 mg/ kg group, and very similar to the 0.8 mg/kg group. Compared to its vehicle, ENG at the dose of 0.008 mg/kg induced lesion weight regression of 47.6%, the dose of 0.08 mg/kg of 48.9%, and the dose of 0.8 mg/kg of 6.7% only (Table S5)

F I G U R E 6
Expression of ERα, PRA/B, FGFR1, FGFR2 and FRS2 transcripts in lesions from mice treated with vehicle (A) AZD4547 at 5, 12.5, and 25 mg/kg or (B) ENG at 0.008, 0.08, and 0.8 mg/ kg (n = 4/ group). AZD4547 and ENG were delivered QD by oral gavage or subcutaneous injection respectively for 20 days from the day of surgery. Total RNA was extracted using the TRIzol method, purified using a mirVana kit and quantified using qRT-PCR. Gene expression was normalized against β-actin and GAPDH and data were expressed as mean ± SEM relative to vehicle and analyzed using a Kruskal-Wallis test with Dunn's multiple comparisons post hoc adjustment; *p < 0.05 and **p < 0.01 compared to the vehicle fluid decreased from the vehicle (100%) to the drug-treated groups (78% in the 0.008 mg/kg group, 61% in the 0.08 mg/kg group, and 71% in the 0.8 mg/kg group).

| D ISCUSS I ON AND CON CLUS I ON
The results obtained from our animal model showed that the trans-   and platelet-derived growth factor receptor (PDGFR) signaling pathways. 5,6 H&E staining showed well-developed lesions in all the groups treated with both AZD4547 and ENG; however, the lumen size of the lesions in the mice treated 2 weeks after surgery with AZD4547 at 25 mg/kg appeared smaller compared to the groups treated with any dosage of ENG. ENG at 0.08 mg/kg was more effective than any dose of AZD4547 with respect to the reduction of percentage of fluid-filled cysts (Table S7); however, AZD4547 displayed higher percentage lesion weight reduction at the dose of 25 mg/kg and higher volume regression (shown as a percentage change from the vehicle) at every dose compared to ENG, confirming the highest dose of AZD4547 to be more effective than ENG also when administered 2 weeks after endometriosis induction (Table S5).
The inhibitory effect of AZD4547 on the establishment and progression of the endometriotic lesions, without impairing estrous cycle regularity, suggests clinical usefulness of this novel drug for short-or long-term prophylactic or management therapies. The administration of ENG, instead, proved to be effective only when administered from the day of surgery and not when the lesions were already developed. Moreover, ENG was associated with the unwanted impairment of fertility via ovarian atrophy and estrous cycle interruption ( Figure S3).
In conclusion, our results confirm that targeting FGFR signaling could form the basis of a new strategy for endometriosis therapy.
Based on the promising results obtained by measurement of lesion weight, volume, and regularity of the estrous cycle, AZD4547 is worth further investigation as a potential treatment for endometriosis, widening the spectrum of the therapies currently available.

ACK N OWLED G M ENTS
We are very grateful to AstraZeneca for funding the project and to Dr. Johanna Laru for her supervision and advice. We thank Prof.
Sarah Herrick for her help with the model and Dr. Mike Harte and Dr.
Brian Telfer for their help with the surgeries.

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 supports the findings of this study are available in the supplementary material of this article.

R E FE R E N C E S
F I G U R E 1 0 (A) Weight of the uterine fragments before being transplanted and after sacrifice following 20-day QD treatment with vehicle and ENG at the doses of 0.008. 0.08, and 0.8 mg/kg s.c. starting from 2 weeks after endometriosis induction. (B) Final lesion volume (cm 3 ). (C) Percentage increase in lesion weight. Data are expressed as mean ± SEM. Animal number per group n = 6, lesion number per group, n = 18, with the exception of the groups treated with ENG at the dose of 0.8 mg/kg, whose lesion number was n = 17