Absence of estrogen receptor beta leads to abnormal adipogenesis during early tendon healing by an up‐regulation of PPARγ signalling

Abstract Achilles tendon injury is one of the challenges of sports medicine, the aetiology of which remains unknown. For a long time, estrogen receptor β (ERβ) has been known as a regulating factor of the metabolism in many connective tissues, such as bone, muscle and cartilage, but little is known about its role in tendon. Recent studies have implicated ERβ as involved in the process of tendon healing. Tendon‐derived stem cells (TDSCs) are getting more and more attention in tendon physiological and pathological process. In this study, we investigated how ERβ played a role in Achilles tendon healing. Achilles tendon injury model was established to analyse how ERβ affected on healing process in vivo. Cell proliferation assay, Western blots, qRT‐PCR and immunocytochemistry were performed to investigate the effect of ERβ on TDSCs. Here, we showed that ERβ deletion in mice resulted in inferior gross appearance, histological scores and, most importantly, increased accumulation of adipocytes during the early tendon healing which involved activation of peroxisome proliferator‐activated receptor γ (PPARγ) signalling. Furthermore, in vitro results of ours confirmed that the abnormity might be the result of abnormal TDSC adipogenic differentiation which could be partially reversed by the treatment of ERβ agonist LY3201. These data revealed a role of ERβ in Achilles tendon healing for the first time, thereby providing a new target for clinical treatment of Achilles tendon injury.


| BACKG ROU N D
Injury owing to sports has risen with the sports participation being increasingly popular in recent years, among which the Achilles tendon injury has been one of the most common. Achilles tendon injuries typically occur in the low vascularized region, 2 to 6 cm to the insertional point. 1 In less than 10 years, the individual Achilles tendon injury rate increased from 2:100 000 to 12:100 000. 2 According to our clinical observations and statistical data, one of the possible intrinsic factors that predispose athletes to Achilles tendon rupture is the level of oestrogen. 3 The injury ratio of male to female ranges from 2:1 to 12:1. 4 It was also reported that women athletes were more likely to suffer from tendon injury when the oestrogen level was low. 5 And factors attributed to healing properties like proliferation rate and collagen I were lower in ovariectomized rats. 6 However, a systematic review on pre-and post-menopausal women showed that the effect of oestrogen supplementation on tendon was contradictory and inconsistent. But the review also suggested there was a need for further studies to understand the effects of female sex hormone supplementation on tendon tissue at a mechanical, morphological and molecular level. 7 It is controversial for us to draw a conclusion about the relationship between oestrogen and tendon injury, which is possibly due to different properties of different tendon structures. Tendon possesses the capability to heal by a repair process controlled by tendon cells and their surrounding extracellular matrix (ECM). 8 But damaged tendon heals slowly and rarely retains the structural integrity and mechanical strength of a healthy tendon. [9][10][11] Recent evidence suggested that modulation of the early stage following tendon repair was important for improving healing. 12 However, the precise mechanisms remain poorly understood. 13 Thus, finding a new target for improving the healing of tendon injuries is one of the most essential challenges in sports medicine field.
ERβ is one kind of estrogen receptors belonging to nuclear transcription factors that are involved in the regulation of many complex physiological processes. For a long time, ERβ has been considered to be an important factor in regulating metabolic pathways and adipose tissue function. 14 In recent days, modulation of ERβ has raised more attention, not only due to its promising role in more applications, 15 such as some types of cancer and neuropathies, but also because it has less side effects compared to activating estrogen receptor α (ERα). As for tendon healing, one study on rat Achilles tendon showed that oestrogen has a positive effect. 6 From a clinical perspective, ERβ-selective agonists might be therapeutically useful. LY3201 is one of the special agonists of ERβ which has been proved to effectively and specifically activate ERβ but not ERα. 16 Tendon possesses the capability to heal by a repair process controlled by tendon cells, especially TDSCs. TDSCs are multi-potent adult stem cells with broad differentiation plasticity that makes them of great importance in cell-based therapy for the repair of tendons. 17 So, LY3201 and TDSCs were studied in our research for promising clinical application.
In the present study, we found that ERβ played an important role in the healing process of tendon injury. We showed that ERβ deletion in mice led to abnormal adipocyte accumulation via up-regulation of expression of PPARγ signalling which might be related to TDSC adipogenesis that could be partially attenuated by LY3201.

| Animal model and surgical procedure
Male ERβ −/− mice and their WT littermates were used in this study.
The generation of the ERβ −/− mice and their primary phenotype were described by Krege et al in 1998, which were all on a C57BL/6J background. 18 Surgical procedures were performed as previously described by Palmes et al with 6-month-old mice that had reached maturity of skeletal system. 19 In brief, (a) anaesthesia, (b) exposing both Achilles tendons from their origins from the gastrocnemius muscle down to their insertion into the calcaneum, (c) resecting the tendon of the plantaris muscle, (d) transecting the Achilles tendon by a standardized procedure at the position about 5 mm above its insertion into the calcaneum and about 5 mm below the first muscle fibres of the gastrocnemius, (e) re-adapting the cut ends of the tendon with a Kirchmayr-Kessler suture (6-0 Dermalon) and further supported with a circular fine suture (10-0 Dermalon), (f) closing the skin, (g) restricting the movement of the ankle by an external fixation designed by us in order to avoid suture failure due to overstretching of the operated tendons for the first two days after operation.
After the operation, the mice were given eight days to repair corresponding to the early healing stage of injured tendons, after which the mice were killed. The whole hindlimb including the gastrocnemius-Achilles tendon-calcaneus complexes was retained for further processing.

| Cell isolation, culture and reagents
Tendon-derived stem cells were isolated as previously described by and 1% penicillin/streptomycin (Pen/Strep) (all from Invitrogen, Carlsbad, CA). TDSCs were grown at 37°C and 5% CO 2 and passaged when 70% confluent with the culture media changed every third day. Cells in passages 2-3 were used for experiments. TDSCs were seeded onto 24-well plates for cell staining and onto 6-well plates for protein and RNA extraction. The identification of TDSCs is shown in Figure S3 according to Bi et al. 21 The ERβ agonist LY3201 was a gift from Eli Lilly. 22 The PPARγ agonist rosiglitazone (ROSI) was purchased from Selleck Chemicals (Houston, TX). All the agonists were dissolved with dimethyl sulfoxide (DMSO) purchased from Santa Cruz Biotechnology, Inc (Santa Cruz).

| Histomorphometry and cellular morphometry
After fixing in 4% buffered formalin at 4°C for 24 hours followed by 30% sucrose at 4℃ for 24 hours, tendons were dehydrated and embedded in optimal cutting temperature compound (OCT) and processed for longitudinal sections (7 μm). Haematoxylin and eosin (HE) staining was used to examine the histology of Achilles tendon at defected zone and then graded by two blinded investigators to analyse the histological score modified by us based on histological scoring system of Stoll et al given in Table S1

| Quantitative reverse transcription polymerase chain reaction
Total RNA from tendon tissues and TDSCs was isolated with kit (Beyotime Institute of Biotechnology) and used for qRT-PCR. Briefly, PCRs were pipetted on ice and each well contained 2.5 μL cDNA, 2.5 μL primer and 5 μL iTaq TM Universal SYBR Green Supermix.
Plates were subsequently sealed and centrifuged down for 20 seconds at 450 g, and then incubated at 95℃ for 5 minutes followed by 30 cycles of a three-step temperature programme of 1 minutes at 95°C, 20 seconds at 65°C and 30 seconds at 72°C. The relative gene expression was quantified by densitometry and normalized to the amount of housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and presented as fold change to WT controls.
All PCR results were reproduced independently in five experiments.
The primer sequences used in these experiments are listed in Table   S2.

| Statistical analysis
Statistical analysis was performed with SPSS 25.0 software (SPSS Inc). Statistical differences between two groups were determined using two-tailed unpaired Student's t test, or two-tailed non-parametric Mann-Whitney test, or one-way analysis of variance followed by Fisher's protected least significant difference post hoc test among three or five groups. Sample size and experimental reproduction are indicated for each method. Results are presented as mean ± SD Differences reached statistical significance at *P < .05, **P < .01, ***P < .001.

| ERβ −/− tendon scars have inferior gross appearance, lower histological scores and cell density paralleled with more increased adipocytes and vessel accumulation
To analyse ERβ involvement during early tendon healing, we established a mouse model of full-thickness Achilles tendon injury

| ERβ −/− tendons have less cell proliferation, whilst more cell apoptosis
To test whether the reduction in cell numbers was due to a decreased proliferation or increased apoptosis, we carried out proliferative and apoptotic assays by immunohistochemistry of Ki67 and terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end labelling (TUNEL) staining (Figure 2D,E,G,H). TUNEL assays showed that ERβ −/− scars had a higher number of apoptotic cells ( Figure 2F).
Furthermore, Ki67 analysis confirmed a lower number of proliferating cells at the scar site of injured Achilles tendon in ERβ −/− mice than WT controls ( Figure 2I). The above data demonstrated that the absence of ERβ results in reduced cell proliferation and increased cell apoptosis of tendon scars during early healing process.

| ERβ −/− tendon scars have more adipocyte accumulation and up-regulation of PPARγ signalling pathway
In order to further investigate the effect of the absence of ERβ on adipogenesis during early tendon healing, we carried out Oil Red O staining ( Figure 3A,B), and the result showed that there was more adipocyte accumulation in the ERβ −/− mice compared with the WT controls ( Figure 3C). The same result could be further confirmed by

| Dose-dependent LY3201 activation of ERβ promotes the proliferative capacity and inhibits the adipogenic differentiation of TDSCs
In order to investigate whether the proliferative capacity of TDSCs was affected by activation of ERβ, cells were cultured

| Dose-dependent LY3201 activation of ERβ inhibits the expression of PPARγ and adipogenic differentiation of TDSCs
In order to further investigate whether PPARγ signalling pathway was involved in the process of LY3201 inhibiting adipogenic differentiation of TDSCs, we continued to carry out adipogenic differentiation culture to TDSCs with ERβ agonist LY3201 and PPARγ antagonist ROSI. The results showed that the inhibition by LY3201 of adipogenic differentiation of TDSCs was reversed with the treatment of 10μM ROSI (Figure 5A-C,G). Next, immunofluorescence of PPARγ showed a significant decrease by LY3201 and an increase again with the treatment of ROSI ( Figure 5D-F,H). Finally, we examined related genes of PPARγ signalling pathway, both PCR and WB results confirmed the involvement ( Figure 5I,J,K), and the results showed a significant activation of it.

| D ISCUSS I ON
Tendon injury has been a big challenge affecting patients by the failed healing and relapse, especially in athletes. 25 It is reported that oestrogen may be related to the occurrence of tendon injury, 5,26 and there have also been many studies investigating the effects of oestrogen on tendon injury in post-menstrual women or ovariectomized animals, 6,26,27 which had the limitations of interference of age (post-menstrual women) and inaccurate target (ovariectomized animals), respectively. In many other studies, investigations have been conducted especially with regard to possible treatments of diseases in tendons by the means of hormone replacement therapy. 28,29 However, only a few studies committed themselves to an approach at a molecular level, especially data by using a knockout model.
The natural healing process in tendon is considered to consist of three overlapping phases: inflammation, proliferation and matrix remodelling. 30 Early tendon healing process mainly contains proliferation of fibroblasts, inflammatory cells and stem cells such as bone marrow-derived stem cells (BMSCs) and TDSCs, which is of great importance creating conditions for next step of healing. 31,32 Therefore, in our study ERβ −/− mice were used to establish a full-thickness Achilles tendon injury for further exploration of the potential role of ERβ in the early healing process. Based on all the results, we demonstrated for the first time that there was more expression of ERβ in injured Achilles tendon, and the absence of ERβ in mouse Achilles tendon led to poor early healing process characterized by increased angiogenesis, reduced cell proliferation, increased cell apoptosis and adipocyte accumulation.
Angiogenesis during early inflammation stage is related to tendon healing. 33 General speaking, an acceleration of angiogenesis regulates inflammation reaction and healing process. 34 The importance of PPARγ in angiogenesis was demonstrated by the generation of knockout animals in 1999. 35 In our study, we found that there were significantly increased vessel numbers labelled by CD34 in and finally an increased p53 expression in ERβ knockout mice ( Figure   S6F). Thus, the absence of ERβ during early tendon healing induced apoptosis through the PPARγ/PTEN/AKT signalling pathway. At the same time, we also examined expression of phosphate-ERK and ERK, which has been demonstrated to be related to cell proliferation. 40,41 The results showed a significant decrease of phosphate-ERK in ERβ knockout mice tendon with no change of total ERK, which confirmed the involvement of ERK ( Figure S6A,G,H). And in vitro results confirmed enhanced proliferative capacity of TDSCs with the treatment of ERβ agonist LY3201. But the mechanism of how ERβ modulates this process was not researched in this study.
Perfect re-establishment of injured tendon is supposed to recover well-organized collagen structure; both abnormal ectopic endochondral ossification and adipocyte accumulation will interrupt natural healing process. Adipocyte accumulation is one of the common pathological changes that occur in ruptured tendons. 23,42 Our study found that the absence of ERβ resulted in adipocyte accumulation in the scar site of injured Achilles tendon. It was reported that PPARγ was down-regulated by the activation of estrogen receptors in hepatocellular carcinoma. 43 Another previous research has also shown that blockade of PPARγ signalling in adipose tissue of ER-β −/− mice resulted in a reversal of the metabolic phenotype corroborating the importance of adipose PPARγ in the high-fat diet model. 44 There are complex interactions between PPARγ and ERβ. [45][46][47] Interestingly, in our study, we found that ER-β −/− mice had more adipocyte accumulation revealed by HE staining, Oil Red O staining F I G U R E 6 Model of adipogenesis action of ERβ is mediated by a negative crosstalk with PPARγ. During early healing process of tendon injury, absence of ERβ will augment the PPARγ expression because of the negative crosstalk between the two so that PPARγ signalling is activated including upstream CD36 and target FABP4, the result is more TDSCs differentiating into adipocytes leads to a poor repair and immunofluorescence of Perilipin compared with WT controls.
At the same time, we found significant increase in CD36, PPARγ and FABP4 that were all involved in PPARγ signalling pathway regulating adipocyte differentiation both at RNA and at protein levels. All these results in vivo brought up a hint that the absence of ERβ leads to adipocyte accumulation through the activation of PPARγ signalling pathway.
In order to confirm this further, TDSCs were tested in vitro. Recent data showed that TDSCs showed higher clonogenicity, proliferation and multi-lineage differentiation potential compared with BMSCs in vitro. 21,48 LY 3201 is one of the specific agonists of ERβ that has been proved to effectively and characteristically activate ERβ but not α that renders it of great clinical application value. 16

| CON CLUS IONS
Collectively, our present findings reveal for the first time the absence of ERβ leads to an abnormal healing process mainly characterized by more adipocyte accumulation during early repair of mouse Achilles tendon which involves an augmented PPARγ signalling. Furthermore, in vitro results confirm that the abnormity might be due to abnormal TDSC adipogenic differentiation which could be partially reversed by the treatment of ERβ agonist ( Figure 6). Implications of these findings might involve a new target for clinical treatment of Achilles tendon injury.