CXCR4 or CXCR7 antagonists treat endometriosis by reducing bone marrow cell trafficking

Abstract Adult stem cells have a major role in endometrial physiology, including remodelling and repair. However, they also have a critical role in the development and progression of endometriosis. Bone marrow‐derived stem cells engraft eutopic endometrium and endometriotic lesions, differentiating to both stromal and epithelial cell fates. Using a mouse bone marrow transplantation model, we show that bone marrow‐derived cells engrafting endometriosis express CXCR4 and CXCR7. Targeting either receptor by the administration of small molecule receptor antagonists AMD3100 or CCX771, respectively, reduced BM‐derived stem cell recruitment into endometriosis implants. Endometriosis lesion size was decreased compared to vehicle controls after treatment with each antagonist in both an early growth and established lesion treatment model. Endometriosis lesion size was not effected when the local effects of CXCL12 were abrogated using uterine‐specific CXCL12 null mice, suggesting an effect primarily on bone marrow cell migration rather than a direct endometrial effect. Antagonist treatment also decreased hallmarks of endometriosis physiopathology such as pro‐inflammatory cytokine production and vascularization. CXCR4 and CXCR7 antagonists are potential novel, non‐hormonal therapies for endometriosis.

Bone marrow-derived stem cells (BMDSCs) are involved in the pathogenesis of the disease and contribute to the development of endometriotic lesions. 7 BM-derived stem cells participate in epithelial and stromal regeneration in endometrial tissue 8 and endometriotic lesions, 9,10 and they are also likely the principal source of extra-pelvic endometriosis. 7,10 Further, BM-derived endothelial progenitor cells (EPCs) also contribute to endometriosis vascularization. 11,12 We have recently demonstrated a functional role of bone marrow-derived stem cells in pregnancy. 13 In cancer models, signals from the tumour microenvironment systemically stimulate quiescent bone marrow compartments, resulting in the expansion, mobilization and recruitment of BMderived cells 14,15 with a clear role in tumour development. 14,16 C-X-C motif chemokine ligand 12 (CXCL12) is one of the best-characterized chemokines active in mobilization of BM-derived stem cells in cancer and inflammation. 15 In the endometriosis niche, CXCL12 participates in epithelial/glandular cell proliferation, 17,18 vasculogenesis and angiogenesis. 19 CXCR4, a G protein-coupled receptor (GPCR), is a receptor for CXCL12 expressed by bone marrow-derived mesenchymal stem cells. CXCL12-CXCR4 signalling is increased in women with endometriosis 20,21 and has been established as a crucial signal for BMDCs migration to endometriosis. 22 CXCR7, also a GPCR, was recently identified as another receptor for CXCL12. 23,24 CXCR7 is phylogenetically closely related to other chemokine receptors and binds CXCL12 with a higher affinity than CXCR4, but it fails to couple to G proteins to induce typical chemokine receptor-mediated cellular responses. 25 CXCR7 has been demonstrated to regulate cell migration and survival through several pathways that include ligand scavenging, direct signal transduction and direct interaction with CXCR45. [26][27][28] CXCR4 activation promotes tumour growth and metastasis in cancer, 29,30 and recent preclinical studies have shown similar effects for CXCR7. 28,31,32 Agents blocking chemokine signalling by binding to these receptors are currently being developed for cancer therapy 27,33 (eg ClinicalTrials. gov identifier NCT02179970, NCT02737072), supporting an evaluation of their efficacy in endometriosis. Given the central role of the CXCL12/CXCR4/CXCR7 axis on BMDCs trafficking and in the pathogenesis of endometriosis, we hypothesized that blocking CXCR4 or CXCR7 in endometriosis using a reproductive and immune-competent murine model would inhibit the growth of endometriosis. Both CXCR4 and CXCR7 may represent new targets for regulating BMDCs engraftment into endometriosis with potential therapeutic applications.

| Animals
C57BL/6J wild-type and ubiquitin-GFP mice were purchased from Charles River Laboratories and The Jackson Laboratory, respectively. Mice were housed and maintained (four to five per cage) in a room (21 ± 1C) with a 12-hour light/dark cycle (7:00 am to 7:00 pm) with ad libitum access to food (Purina Chow; Purina Mills) and water, in the Yale Animal Resources Center (YARC) at Yale School of Medicine. Oestrous cyclicity was evaluated by vaginal cytology before endometriosis induction in order to transplant all tissue in dioestrus phase. In addition, oestrous cyclicity was evaluated during injection of vehicle or treatment to investigate any potential effect on the oestrous cycle. All animal experiments were conducted in accordance with approved protocols from Institutional Animal Care and Use Committee (IACUC) of Yale University. IACUC guidelines were clearly followed for animal care, 5FU treatments, bone marrow cell injections, blood collection and anaesthesia.

| BM conditioning and transplantation
Six-week-old female C57BL/6J wild-type mice received 125 mg/kg of 5-FU by i.p injections 6 days and 1 day before bone marrow transplantation (BMT). In addition, stem cell factor (SCF, 50 mg/kg) was injected i.p twice before BMT, as we have previously described. 34 Transplantation of fresh BM cells was performed as described previously. 9 Briefly, bone marrow cells were obtained from 6-to 10-week-old C57BL/6J ubiquitin-GFP male donor mice by flushing the marrow from femurs and tibias into cold sterile PBS and filtered through 70-μm cell strainer (BD Biosciences, San Jose, CA, USA).
The yield and viability of BM cells were determined by trypan blue staining. Next, 20 × 10 6 unfractionated BM cells were iv injected to recipients 6 days after the beginning of BM conditioning. Lesions were stained for Ki-67 proliferation marker as described below.

| Induction of endometriosis in mice
Endometriosis in mice was surgically induced under aseptic conditions and anaesthesia using a modified method previously described. 10,35 Surgery was performed 30 days following BMT. Uterine horns were removed from wild-type female donor mice at dioestrus (low oestrogen stage), opened longitudinally, cut into fragments of 3-mm and transplanted onto the peritoneal wall of recipient mice by suturing. Three uterus fragments from wild-type mice as well as CXCL12 −/− were systematically transplanted into peritoneal wall of each mouse. After treatments, ectopic lesions were collected.
Ectopic lesion volume was calculated as a half ellipsoid that approximated lesion shape on the peritoneum, using formula V = (1/2) (4/3) πr1 2 r2 (r1 and r2 are radii, r1 < r2). 36 Sham mice were subjected to the surgery, but in place of suturing uterine tissue, peritoneal tissue from the ventral midline was used. Sham mice did not receive any treatment.

| Immunohistochemistry and immunofluorescence
Tissue from endometriotic lesions was fixed in 4% paraformaldehyde and embedded in paraffin. Five-micrometre tissue sections were mounted on slides followed by 10 minutes boiling in sodium citrate (pH 6) for antigen retrieval, and blocking using 10% serum (Vector Laboratories). Tissue sections on slides were incu- Laboratories.) Isotype controls were used to confirm the specificity of CXCR4, CXCR7 and CXCL12 by immunofluorescence studies with rabbit IgG antibody (Abcam, #37415) for CXCR4 and CXCL12 or mouse IgG1 (# MAB002, R & D Systems) shown in Figure S1.
Visualization of the slides was performed using a laser scanning confocal microscope (LSM 710; Zeiss) and the ZEN software (Carl Zeiss). Stained cells were quantified by monitoring the average numbers of positively stained cells relative to the total number of cells from six randomly chosen fields.

| Apoptosis assays
The TUNEL assay was performed to assess the apoptosis activity in ectopic lesions sections with the In situ Apoptosis Detection Kit (Abcam, catalog #ab206386) according to the manufacturer's instructions. The TUNEL reaction preferentially labels DNA strand breaks generated during apoptosis and allows discrimination of apoptosis from necrosis. In this assay, terminal deoxynucleotidyl transferase (TdT) binds to exposed 3′-OH ends of DNA and are de- Leica CTR 5000; Germany).

| Quantitative real-time polymerase chain reaction (qRT-PCR)
Total RNA was isolated from endometriotic lesions by TRIzol reagent (Life Technologies) followed by purification using Quiagen Gene expression was normalized to the expression of β-actin for each sample. Relative mRNA expression for each gene was calculated using the comparative cycle threshold (Ct) method, also known as the 2 −ΔΔC(T) method. 38 All experiments were carried out in triplicate and nuclease-free water was used as a negative control replacing the cDNA template.

| Statistical analysis
Data were analysed using GraphPad Prism 7.0 (GraphPad Software Inc). An unpaired Student's t test for percentage of labelled cells (PLC) or one-way ANOVA for cell counts and RT-PCR data was used to determine statistical significance. Data are expressed as means ± standard error (SE).

| BMDSCs engrafting endometriosis express CXCR4, CXCR7 and stem cell markers
To characterize BMDSCs engrafting endometriosis in vivo, we used a 5-FU-based submyeloablation mouse bone marrow transplantation model according to our previously published protocol. 34 This method results in efficient bone marrow donor chimerism (approximately 50% donor-derived cells) while preserving ovarian function and fertility. Previous studies on BMDCs trafficking in endometriosis have relied on sub-lethal irradiation as conditioning for BMT with resulting loss of ovarian function, oestrous cyclicity, 9,11,39,40 dysfunctional immune signalling 41 and blood flow. 42 Preserving gonadal activity is crucial because models of murine endometriosis using reproductive and immuno-competent mice show the same endocrine features of human disease, namely aberrant oestrogen signalling and progesterone resistance. 43,44 Moreover, gonadal steroids regulate the CXCL12-CXCR4 axis in endometrium and endometriosis. 45 Figure 1D). Remarkably, a small percentage of BMDCs engrafting endometriotic lesions expressed CXCL12 (0.1 ± 0.02% of total cells) ( Figure 1E). No GFP (bone marrow-derived cells) was found in the epithelium (Figure 2A). Also, we noticed that some engrafted bone marrow-derived cells were positive for CD45 and some of them were simultaneously positive for F4/80, a marker for macrophages ( Figure 2B).

| Pharmacological antagonism of CXCR4 or CXCR7 reduced the engrafting of BMDSCs into endometriosis
The high expression of CXCL12 in endometriosis and expression of CXCR4 and CXCR7 in BMDCs suggested that CXCL12 might serve to regulate BMDCs trafficking towards endometriosis. To test this hypothesis in our murine model, we administered the small molecule CCX771, a CXCR7 antagonist, 51 AMD3100 (Plerixafor), a CXCR4 antagonist, or vehicle control for 15 days (day 1-15 after graft) ( Figure 3A).
We showed that of total endometriosis cells, the majority of BMDCs engrafting endometriosis were CD45 − compared to CD45+ cells (leucocytes) as shown in Figure 3B. AMD3100 significantly reduced the engraftment of GFP+ cells in endometriosis including GFP + CD45 + leucocytes and GFP + CD45 − BMDSCs, consistent with a reduction of recruitment of total BMDCs in the lesions. Similarly, CCX771 reduced total GFP + cells significantly in endometriosis including GFP + CD45 + and GFP + CD45 − cells ( Figure 3C). Taken together, these findings demonstrate that targeting CXCR4 or CXCR7 reduced BMDCs engraftment, with a marked predominance for BM cells not expressing the pan-leucocyte protein CD45. Blocking CXCL12 activity on its receptor greatly reduced BMDCs engraftment in endometriosis.

| Pharmacological antagonism of CXCR4 or CXCR7 led to the regression of endometriosis
To investigate whether targeting CXCR4 or CXCR7 could have clinical application, we examined volumetric and molecular changes of endometriotic lesions following treatment. We first used an early growth model (treatment over days 1-15 after graft placement) to examine the effectiveness of AMD3100 and CCX771 in preventing ectopic lesion establishment. (Figure 4A). Either AMD3100 or CCX771 treatment prevented the development of endometriosis following the engraftment of donor uteri fragments in normal cycling mice. Treatment with AMD3100 or CCX771 resulted in more than a 50% reduction of the lesion volume after 15 days comparing either agent to vehicle treated controls. There were no significant differences between the two active treatments ( Figure 4B). IHC labelling confirmed reduction in Ki-67 expression in the stroma following both treatments whereas only CCX771 reduced epithelial Ki-67 expression ( Figure 4C). Both treatments reduced the vessel density by 40% in endometriosis determined by IHC staining ( Figure 4D).
Treatment was also associated with a decline in mRNA expression of several pro-inflammatory cytokines including IL-6, TNF-α, IL-1β and TGF-β, known to be highly expressed in endometriosis. Both treatments reduced the mRNA expression of vascular endothelial growth factor A (VEGF-A) and matrix metalloproteinase MMP2 and MMP9. CCX771 had no effect on IL-6 or COX-2 expression in this model ( Figure 4E).
In order to further test our hypothesis, treatment with AMD3100 and CCX771 was started after ectopic lesions were already established in recipient mice (treatment over days 15-30 after graft placement) ( Figure 5A). This more closely models the pre-existing lesions found in humans at the time of endometriosis diagnosis. Both treatments reduced the vessel density in endometriosis determined by IHC staining and VEGF-A mRNA expression ( Figure 5C). Consistent with the results presented above, both AMD3100 and CCX771 reduced endometriosis lesion volume by 60% ( Figure 5B) and cell proliferation by 70% in the stromal compartment as determined by ki-67 staining ( Figure 5D). Both treatments also increased apoptosis as determined by TUNEL assay ( Figure 5E). CCX771 treatment increased the apoptosis significantly both in stroma and epithelium while AMD3100 reduced only stromal apoptosis. Moreover, the administration of either drug (d15-d30 after graft) reduced the expression of pro-inflammatory cytokines such as IL-1β, IL-6 and TNF-α, COX-2 and invasion pathway proteins MMP2 and MMP9 as shown in F I G U R E 2 Immunostaining of CXCR4, CXCR7 and CXCL12 and BMDCs. A, Representative immunofluorescence images of CXCR4, CXCR7 and CXCL12 staining in the epithelial cells of murine endometriosis. In our model, no GFP (bone marrow-derived cells) was identified in the epithelium. Scale bar: 10 μm. B, Engrafted BMDCs express CD45 and F4/80. BMDCs: Engrafted bone marrow-derived immune cells (BMDCs) were stained with DAPI or immunofluorescence for CD45 and F4/80. Engrafted BMDCs shows the presence of non-immune (CD45 negative) and immune cells (CD45 positive), some of which co-express F4/80, a marker of macrophages. Included is a cell that is a bone marrow-derived (GFP+), CD45+, F4/80+ macrophage (arrow) Scale bar: 10 μm Figure 5F, confirming our hypothesis that CXCR4-CXCR7 may treat endometriosis and endometriosis-associated inflammation.
In addition, the oestrous cycle was evaluated during injection of vehicle, AMD3100 and CCX771, to investigate any potential effects on mice oestrous. Phase identification was performed by vaginal cytology. No significant changes in the length of each oestrous cycle stage were recorded in animals receiving either drug, supporting the absence of systemic endocrine effects and demonstrating that the effects of these agents worked thorough a hormone independent pathway targeting stem cell recruitment.

| No significant change in lesion size in endometriosis from uterine-specific CXCL12 null mice
To assure that the effects of treatment was due to blockade of BMDC recruitment and not local effects of endometrial cell-derived CXCL12, we repeated these experiments creating endometriosis derived from uterine-specific CXCL12 null mice. Uteri of PGR-Cre+/CXCL12 −/− mice demonstrated dramatically decreased CXCL12 mRNA relative to CXCL12 fl/fl controls (P = .037) ( Figure S3a). When these uteri were used in our endometriosis model, lesions still grew to sizes comparable to those of CXCL12 fl/fl controls (0.762 ± 0.104 mm 2 , n = 35 lesions vs 0.547 ± 0.118 mm 2 , n = 12 lesions, respectively, P = NS) ( Figure S3b) and exhibited comparable cycle-matched proliferation ( Figure S3c), suggesting that expression of CXCL12 by the endometriosis lesion is not essential for lesion development.

| In vitro studies on primary endometrial cells from endometriosis patients and AMD3100
To further evaluate the effect of AMD3100 directly on endometrial cells, primary endometrial cells stromal from women with endometriosis were treated with AMD3100. No significant changes in cell proliferation were observed when primary endometrial cells from endometriosis patients were treated with AMD3100 compared to untreated cells as shown in Figure S4. AMD3100 did not directly affect endometrial cells, further supporting the effect of AMD3100 specifically on BMDCs.

Interaction of CXCL12 with its G protein-coupled receptors CXCR4
and CXCR7 plays a crucial role during embryonic development, tissue repair as well as in many pathologic processes. Activation of the CXCL12-CXCR4-CXCR7 signalling axis in endometriosis is associated with dysfunctional chemotaxis, increased cell proliferation, angiogenesis and reduced autophagy. 18 Confirming this observation, our in vitro studies showed that there was no significant effect of AMD3100 on endometrial stromal cells from endometriosis patients. AMD3100 likely exerts its effects on endometriosis through BMDSCs. Similarly, blocking CXCR7 by CCX771 reduces angiogenesis in vivo interrupting the pro-angiogenic CXCL12-CXCR7 autocrine loop exerted by activated endothelial cells. 62 Additional studies using CXCR4 and CXCR7 knockout mouse donor tissue would clarify the role of specific cell populations in endometriosis regression following AMD3100 and CCX771 treatment.
In conclusion, in a preclinical model of endometriosis, we showed that targeting the CXCL12-CXCR4-CXCR7 axis blocks bone marrow-derived stem cell recruitment. Treatment leads to regression of both new and established lesions. Bone marrow-derived stem cell recruitment is likely necessary for endometriosis lesion maintenance, and blocking trafficking of these cells treats the disease. Both drugs exhibited similar efficacy. Clinical use will F I G U R E 5 Targeting CXCR4 or CXCR7 induced the regression of endometriosis. Treatment model: AMD3100 or CCX771 or vehicle was subcutaneously injected into three groups of mice separately (each group n = 6 mice) of mice starting on day 15 after grafting uterine fragment into the peritoneal cavity (day 0) of wild-type animals (A). Ectopic lesions were harvested after 15 days of treatment (day 30). Representative pictures of ectopic lesions isolated from wild-type mice with endometriosis subcutaneously treated with either vehicle or receptor antagonists (B). Lesion volume measurement after 15 days of treatment. IHC, and quantification of microvessel density of ectopic lesions of wild-type mice treated with ligands or vehicle using CD31 staining (C). Scale bar: 100 μm. IHC and quantitative analyses of the expression patterns of Ki-67 in epithelial or stromal cells of ectopic lesions of wild-type mice after treatment (D). Scale bar: 50 μm. Representative pictures and quantification of deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick end labelling (TUNEL) assays in epithelial and stroma cells (E) (the number of TUNEL-positive stained cells are shown in the bar graphs as percentage to the total cells). Scale bar: 50 μm. Expression of IL-1β, IL-6, TNF-α, COX-2 and MMP-2 and MMP-9 in ectopic lesions was analysed by quantitative reverse transcription PCR (RT-PCR) after 15 days of treatment (F). mRNA levels are expressed relative to transcript level in ectopic lesions of mice treated with vehicle, set at 1.0. Results show mean ± SE. *P < .05 vs vehicle likely depend on side effect profile; the effects of prolonged use are not well characterized. In addition, these therapies may have off-target effects on other tissues; the specificity of these treatments to endometriosis has not been evaluated in these studies.
CXCR4 and CXCR7 antagonists are promising novel, non-hormonal therapies for endometriosis.

ACK N OWLED G EM ENTS
This work was supported by the NIH (NIH U54 HD052668 and R01 HD076422 to H.T). We thank Ying Lu for help in FACS experiments, Reshef Tal for suggestions in experiments. We thank Tran Dang for technical help and proof reading manuscript.

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