A History of Endometriosis Is Associated With Decreased Peripheral NK Cytotoxicity and Increased Infiltration of Uterine CD68+ Macrophages

Women with endometriosis may have a defective immune system. However, evidence of the immune responses of endometriosis patients with a history of endometriosis surgery is lacking, and the association between the location of endometriosis lesions and immune responses is unclear. This retrospective study included 117 females with reproductive failure and a history of endometriosis and 200 females with reproductive failure but without endometriosis to analyze their endometrial and peripheral immune responses. The results show that endometriosis was associated with decreased peripheral natural killer (NK) cytotoxicity and increased uterine macrophages. Peripheral NK cytotoxicity at effector-to-target ratios of 25:1 and 50:1 was significantly reduced in women with a history of endometriosis from that of the control group (26.6% versus 33.3% and 36.1% versus 43.3%, respectively, both P < 0.001). Furthermore, after further division of patients into three subgroups according to the location of endometriosis lesions, we observed that NK cytotoxicity in the endometriosis subgroups, especially the mixed endometriosis group, was strongly decreased from that of the controls (P = 0.001). The endometrial CD68+ macrophage proportion in the mixed endometriosis subgroup was higher than that in the control group (2.8% versus 2.1%, P = 0.043). In addition, the baseline estradiol (E2) level was weakly correlated with the percentage of endometrial macrophages (r = 0.251, P = 0.009), indicating a potential association among the endocrine system, endometrial immune environment, and endometriosis. This study indicated that peripheral NK cytotoxicity and endometrial immune cell profiles could be useful for diagnosing and treating endometriosis and endometriosis-related reproductive diseases.


INTRODUCTION
Endometriosis, defined as the presence of endometrial-like tissue outside the uterus, is a common estrogen-dependent gynecological disease with an overall prevalence of 10% in women of reproductive age (1). Endometriosis is a risk factor for many obstetrical and gynecological disorders, and leads to infertility in approximately 30% of affected women (2). Moreover, endometriosis has a high recurrence rate (3). However, the pathogenesis of endometriosis is still unclear, which poses considerable obstacles to its diagnosis and treatment.
According to the location of endometriotic lesions, endometriosis is divided into deep infiltrating endometriosis, ovarian endometriosis (chocolate cysts), and peritoneal endometriosis (pelvic endometriosis) (4). To date, the molecular mechanisms governing the initiation and development of ectopic endometriotic lesions remain controversial. Along with retrograde menstruation, endometrial stem cell implantation, Müllerian remnant abnormalities, and coelomic metaplasia, the pathogenesis of endometriosis may also involve the dysregulation of estrogen and progesterone, angiogenesis or vasculogenesis, immune system, cytokines, and chemokines (5). Most importantly, systemic and local immune responses may play a primary role in the pathophysiology and symptomatology of endometriosis (6,7). It is speculated that women with endometriosis may have a defective immune system, with endometrial fragments themselves acquiring the ability to evade immune surveillance and clearance (6).
Previous studies have reported changes in the composition or functions of immune cells, including macrophages (8), natural killer (NK) cells (9), dendritic cells (DC) (10)(11)(12), and T (13)(14)(15) and B cells (15,16) in both the eutopic or ectopic endometrium or the peripheral blood of women with endometriosis. However, the composition or functions of those immune cells remain controversial. Additionally, evidence regarding the immune responses of endometriosis patients with a history of surgery for endometriosis is lacking, and the association between the location of endometriosis lesions and patient immune responses is unclear.
Therefore, in order to verify the endometrial and peripheral immune responses in patients with a history of endometriosis, this study retrospectively analyzes various immune parameters in 117 females with reproductive failure and a history of endometriosis and 200 females with reproductive failure but without endometriosis. The former group are further divided into three subgroups, including ovarian, pelvic, and mixed endometriosis, to determine the effect of the specific location of endometriosis on immune responses. Evaluating the immune response is important for revealing the pathogenesis of endometriosis and endometriosis-related reproductive failure. It can also facilitate the development of innovative diagnostics and therapeutics for associated diseases.

Study Population
This retrospective study was evaluated and approved by the ethics committee of Shenzhen Zhongshan Urology Hospital (Approval number: SZZSECHU-F-2019040). Initially, this study included 682 females who visited the Reproductive Immunology Department of Shenzhen Zhongshan Urology Hospital from January 2014 to October 2019, who had a history of repeated implantation failure (RIF), recurrent miscarriage (RM), or other cases of reproductive failure. RIF is defined as women who experienced two or more retrieval cycles and transferred more than 10 high-quality embryos (17) without pregnancy (18). RM is defined as patients who experienced two or more miscarriages before week 20 of gestation. Other cases of reproductive failure refer to patients who failed to achieve a pregnancy or experienced miscarriage but did not meet the standard of RIF or RM. Patients with hormonal or metabolic disorders, known clinical autoimmune disease, chronic endometritis, uterine, tubal or ovary abnormality, adenomyosis (n = 265), or lacking baseline data (n = 100) were excluded from this study. Finally, the analysis included 317 patients, including 117 women with a history of endometriosis and 200 women without endometriosis as the control group. Patients in the endometriosis group were diagnosed with endometriosis through laparoscopy, B-scan ultrasonography, and observations of the symptoms of endometriosis such as pelvic pain, infertility, pain after sex, and heavy menstrual periods. All patients in the endometriosis group were without endometriosis on the date of sampling according to an absence of the symptoms of endometriosis and B-scan ultrasonography analysis, with or without laparoscopy. On the other hand, patients in the control group were defined by their absence of endometriosis symptoms and B-scan ultrasonography, with or without laparoscopy. According to lesion locations, the endometriosis group was further divided into three subgroups, including ovarian (n = 44), pelvic (n = 57), and mixed endometriosis (n = 16, patients diagnosed with both ovarian and pelvic endometriosis).

Endometrial Preparation Program
Hormone replacement therapy cycles or natural cycles were used to collect the endometrium from the patients. During natural cycles, progesterone (Dydrogesterone Tablet, Solvay Pharmaceuticals, Netherlands) was administered at a dose of 20 mg on day 18 of the cycle, then continued until the mid-luteal phase of the menstrual cycle. During hormone replacement therapy cycles, estrogen and progesterone were used consecutively to mimic the endocrine conditions of the endometrium during the natural cycle. Estradiol valerate (Delpharm Lille SAS, Bayer AG, Berlin, Germany) was administered during the cycle as follows: 4 mg on day 3-7, 6 mg on day 8-12, 8 mg on day 13 to the day of endometrial sampling. Progesterone (Progesterone Injection, Zhejiang Xianju Pharmaceutical Co. Ltd. Zhejiang, China) was administered at a dose of 60 mg on day 18 of the cycle, then continued until the mid-luteal phase of the menstrual cycle.

Measurement of Serum Hormone Levels
Serum hormones were all collected on day three of the menstrual cycle. The concentrations of estradiol (E2), progesterone (P), follicle-stimulating hormone (FSH), luteinizing hormone (LH), prolactin (PRL), and testosterone (T) in the sera of the patients were measured with Elecsys Estradiol III, progesterone II, FSH, LH, Prolactin II, and Testosterone II kits, respectively (Elecsys, Olathe, USA), and analyzed in an immunology analyzer (Roche Diagnostics, Basel, Switzerland).

Endometrial Immunological Parameters
Endometrial tissue samples were collected during the mid-luteal phase (LH day 7-9) of the menstrual cycle. Samples were fixed and embedded in paraffin wax for immunohistochemical staining. Samples were stained in an auto-immunostainer (Leica Bond III system, Germany) using a Bond polymer refine detection kit (Leica Microsystems, Germany). All staining batches included appropriate controls and employed commercially available monoclonal antibodies for CD56, CD68, CD1a, CD83, CD57, CD8, and FoxP3, and the immunohistochemical protocols outlined in Supplementary  Table S1.
The samples were first scanned at low magnification. Then, images of 30 random fields (10.68 mm 2 ) per section were captured in high-power fields (HPF, magnification 200×). All immune cell populations from each sample were characterized and quantified using the cell segmentation and phenotype cell tool of InForm ® Cell Analysis ™ software (Perkin Elmer, Waltham, MA, USA) under the supervision of two pathologists. The ratio of each immune cell population was assessed as a percentage on all endometrial cells in 30 randomly selected highpower fields (HPF, magnification 200×).

Peripheral Immunological Parameters
Peripheral blood samples were collected during the mid-luteal phase (LH day 7-9) of the menstrual cycle for analyses of the peripheral immune responses, including lymphocytes subsets, NK cytotoxicity, NK cell surface receptors and intracellular mediators, and intracellular cytokines of peripheral CD4 + T cells.
NK cell cytotoxicity was evaluated by K562 target-killing assay with peripheral blood mononuclear cells (PBMC) isolated from the patients. K562 cells purchased from the China Center for Type Culture Collection were washed in PBS and dyed with 3,3-dioctadecyloxacarbocyx acarbocyanine perchlorate-FITC (Sigma-Aldrich, St. Louis, MO, USA). PBMCs were isolated by density gradient centrifugation using Ficoll-Paque (Amersham Bioscience, Shanghai, China). Subsequently, PBMCs and K562 cells were cocultured at effector-to-target ratios of 25:1 (ET25) and 50:1 (ET50) at 37°C in a 5% CO 2 humidified incubator. After 4 h incubation, propidium iodide-PE (Sigma-Aldrich, St. Louis, MO, USA) was added to determine the proportion of dead K562 cells by flow cytometry.
To evaluate the NK cell surface receptors and intracellular mediators, isolated PBMCs were stained with CD3-PerCP and CD56-PE-Cy7 to characterize the NK cells. NK cell surface receptors were stained with NKG2D-FITC, NKp30-PE, NKp46-APC, CD158a-FITC, and CD158b-PE for 30 min at room temperature in the dark, then detected by flow cytometry. Intracellular proteins were stained with granzyme B-PE, perforin-Alexa Fluor 647, and granulysin-Alexa Fluor 647 for 30 min at room temperature in the dark, then detected by flow cytometry.
For the intracellular cytokine staining of peripheral CD4 + T cells, fresh blood (500 mL) was stimulated with 10 ng/mL phorbol myristate acetate (PMA; Sigma-Aldrich, USA) and 1.25 mg/mL ion-omycin (Sigma) in the presence of 10 mg/mL brefeldin A (BFA) for 4 h at 37°C in 5% CO 2 . One hundred microliters of stained blood was aliquoted into a new flow tube. After removing red blood cells, samples were permeabilized with FACS Permeabilizing Solution (BD Bioscience), then stained for CD3-PerCP, CD8-APC-Cy7, TNF-a-APC, IFN-g-FITC, or matched isotype controls (BD Bioscience) in the dark for 30 min.

mRNA Expression of Estrogen Receptors and CD68 via EndometDB
EndometDB is an interactive web-based user interface for browsing the gene expression database of collected samples without the need for computational skills. EndometDB incorporates the expression data from 115 endometriosis patients and 53 controls, with over 24,000 genes and clinical features. The mRNA expression of estrogen receptor 1 (ESR1), estrogen receptor 2 (ESR2), and CD68 in the endometrium and peritoneum of the control group, and in the eutopic and ectopic endometria and peritoneum of patients with endometriosis were analyzed in EndometDB (https://endometdb.utu.fi/gene_ analysis/) (19). Specifically, all clinical and sample data in the database were selected, and the patients were separated by age. Box plots (parameter settings: combine lesions, use log2-scale, show legend, and display sample counts) were used to show the data. Pearson correlation was used to show the correlation matrix between ESR1, ESR2, and CD68 via EndometDB.

Statistical Analysis
Statistical analyses were performed using SPSS for Windows (version 25.0, SPSS Inc., Chicago, IL). The distribution of continuous data was checked for normality using the Kolmogorov-Smirnov test. Continuous variables with a normal distribution were analyzed by the student's t-test or one-way ANOVA, and shown as means ± standard deviations.
Continuous variables without a normal distribution were examined by the Mann-Whitney U test or Kruskal-Wallis test and presented as median values (interquartile range). Categorical variables were analyzed by the Chi-square test and shown as percentages of participants. The correlation was analyzed by Pearson correlation. In all comparisons, a two-tailed P-value < 0.05 was considered statistically significant.

Baseline Characteristics of the Study Population
Compared with the control group, females with a history of endometriosis had higher baseline serum E2 levels (  Figure S1. When the endometriosis population was further divided into subgroups according to the location of ectopic lesions ( Table 3, Supplementary Figure S2), NK cytotoxicity in the subgroups of ovarian, pelvic, and mixed endometriosis was still significantly lower than that in the control group (all P < 0.05). Interestingly, NK cytotoxicity in women with a history of mixed endometriosis was slightly less than that for other subgroups. Furthermore, peripheral NK cytotoxicity was significantly decreased in the endometriosis group from that of the control group in the "Other" population (Supplementary Tables S3, 4). The same tendency was observed in the RIF and RM population, but without statistical significance (Supplementary Tables S5-8).
Regarding the mechanisms of low PBMC cytotoxicity in patients with endometriosis, the toxic granules (granzyme B, perforin, and granulysin) and surface receptors (NKG2D, NKp30, NKp46, CD158a, and CD158b) of peripheral NK cells were analyzed. Nevertheless, all parameters were comparable between the control and endometriosis groups (all P > 0.05, Supplementary Figure S3). Negative results were also observed when analyzing the productions of IFN-g and TNF-a by CD4 + T cells in the endometriosis group compared with the control group (Supplementary Figure S4). More research should be conducted to determine the underlying mechanisms. The percentages of peripheral lymphocytes subsets were   Figure S1, P > 0.05).

Endometrial Immune Cell Profiles of Reproductive Failure Patients
There was no statistically significant difference between women with a history of endometriosis and the control group regarding endometrial immunological parameters ( Table 2, P > 0.05). Representative images of the uterine immune cells of patients are shown in Supplementary Figure S5. However, when the endometriosis group was further divided into subgroups ( Table 3, Supplementary Figure S6), the proportion of endometrial CD68 + macrophages in women with mixed endometriosis was significantly higher than that in the control group (2.8% versus 2.1%, P = 0.043), indicating that uterine macrophages are associated with mixed endometriosis. The remaining uterine parameters were still comparable among the subgroups ( Table 3, P > 0.05). Intriguingly, there was a weakly positive correlation between the proportion of endometrial CD68 + macrophages and baseline serum E2 levels, indicating a potential association among E2, local immune profiles, and endometriosis ( Figure 1, Supplementary Table S9, r = 0.251, P = 0.009). EndometDB data also showed that CD68 was slightly positively correlated with ESR2, but not with ESR1 (Supplementary Figure S7).

DISCUSSION
The hormone levels, BMI, peripheral NK cytotoxicity, and endometrial immune cell subtypes of the endometriosis population were significantly different from those of the control group. In this study, women with a history of endometriosis, especially those exhibiting overlap of multiple diseases (mixed endometriosis), were associated with decreased NK cytotoxicity and changed local immune cell profiles. At present, the most commonly used disease staging system (the R-ASRM classification) is based on the number and depth of lesions, endometrioma, and adhesions within the pelvic cavity (20). However, only a marginal relationship was demonstrated among the number of lesions, lesion symptoms, the severity of the disease, and the overall impact on quality of life (except for an association between pelvic pain and deep infiltrating endometriosis sites) (3). Thus, overlap of multiple diseases might provide a new disease staging system for endometriosis. Additionally, a slightly positive correlation between baseline E2 levels and endometrial CD68 + macrophages in women with a history of endometriosis indicated a potential association among hormones, immune cell responses, and endometriosis. Patients with a history of endometriosis had a higher prevalence of RIF, whose main etiology is low endometrial receptivity. Regarding the hormone effects, continual high expression of estrogen in women with a history of endometriosis likely interferes with the transition from proliferative to secretory phases and other important reproductive events (21). According to an analysis of immune responses, significantly increased endometrial CD68 + macrophages were observed in women with a history of mixed endometriosis compared to the control group. Macrophage-secreted proinflammatory cytokines, such as tumor necrosis factor-a, IL-15, and macrophage inflammatory protein 1B, could influence preparation of the endometrium for implantation (22). Macrophages also regulate the vascular network via the production of cytokines (such as IL-6, IL-10, IL-12, and TGF-b1) (23,24), which further regulates endometrial receptivity. Furthermore, endometriotic lesions secreting both inflammatory cytokines and reactive oxygen species could lead to the production of DNA and cytoplasmic fragmentation in embryos (25). This would also result in RIF.
There was a significant relationship between a history of endometriosis and decreased peripheral NK cytotoxicity in this study. Some studies also suggested that NK cytotoxicity in peripheral blood and peritoneal fluid was decreased in endometriosis patients compared to controls (26)(27)(28). A review by Sciezynska et al. (29) showed that factors including IL-6, IL-10, IL-12, IL-15, and TGF-b were involved in the inhibition of NK cell cytotoxicity during the course of endometriosis. For example, the interaction between macrophages and ectopic endometrial stromal cells could downregulate the expression of NKG2D, perforin, and IFN-g, as well as the cytotoxicity of NK cells by the secretion of IL-10 and TGF-b (30). IL-6 or peritoneal fluid from endometriosis patients reduces the cytolytic activity of NK cells, concomitantly with the down-regulation of granzyme B, perforin, and the killer activating receptor (NKp46), while increasing the killer inhibitory receptor (CD158b) (31). Moreover, platelet-derived TGF-b1 suppresses the expression of killer activating receptor (NKG2D) on peritoneal NK cells, resulting in reduced cytotoxicity in women with endometriosis; however, neutralization of TGF-b1 reversed this reduction (32). Increased expression of killer inhibitory receptor (CD158a) by circulating NK cells has also been observed in patients with endometriosis (33). However, differential expression of the toxic granules (granzyme B, perforin, and granulysin) and surface receptors (NKG2D, NKp30, NKp46, CD158a, and CD158b) of peripheral NK cells between females with or without endometriosis was not observed in our study (Supplementary Figure S3). Circulating NK cytotoxicity might be regulated by other factors, such as other surface receptors (NKp44, NKG2A, ITAM-KIR, etc.), the production of IFN-g and TNF-a, NK cell maturation and development, proangiogenic features of NK cells, or the interaction with other immune cells such as regulatory T cells (34)(35)(36)(37)(38). It is worth noting that, although PBMC cytotoxicity is used to represent NK cytotoxicity in many studies, peripheral T cells might also exhibit cytotoxicity via the production of cytokines, such as IFN-g, TNF-a, and the human cytotoxic factor (39,40). However, our data showed that the production of IFN-g and TNF-a by CD4 + T cells in the endometriosis group were comparable with that in the control group (Supplementary Figure S4). Therefore, further study is required, such as single-cell sequencing of peripheral lymphocytes, to verify the underlying mechanism and detect the function of each immune cell type. Nevertheless, it should be stressed that some studies could not confirm abrogated circulating NK cytotoxicity against K562 (41,42). For example, Ho et al. (42) found that women with endometriosis showed a decrease in peritoneal NK cytotoxicity against K562; however, the results of circulating NK cytotoxicities were comparable between endometriosis and non-endometriosis groups. Furthermore, a study by Vigano et al. (41) showed decreased cytotoxicity of PBMC against endometrial stromal cells but not K562. Some of these negative results might be explained by the low number of patients and controls and population differences.
Endometriosis was correlated with local macrophages. There is also some evidence that the macrophage proportion is enhanced in the peritoneal fluid and endometrium of women with endometriosis (43,44). In the endometriotic microenvironment, macrophages are a major source of elevated proinflammatory and chemotactic cytokines and angiogenic mediators (45), promoting the proliferation and invasion of endometrial stromal cells. Macrophages also produce anti-inflammatory cytokines, and play a critical role in accelerating the growth, adhesion, invasion, epithelialmesenchymal transition, fibrogenesis, and angiogenesis of ectopic lesions, as well as the impairment of immune surveillance (24). Additionally, macrophages may acquire memory-like characteristics to regulate endometriosis in low inflammatory microenvironments (46). Enhanced macrophages in the endometriotic microenvironment could promote the proliferation and invasion of endometrial stromal cells, and enhance the growth, adhesion, and invasion of ectopic lesions, which might lead to the appearance of ectopic lesions in multiple locations. On the other hand, the endometriotic microenvironment could recruit macrophages (47,48). Cause and effect are hard to distinguish, indicating the need for further research to determine the upstream and downstream mechanisms of macrophages in endometriosis. However, Braun et al. (49) showed a decrease of macrophages in the same endometrial phase in patients with endometriosis, which might be due to the small sample size or different population. Furthermore, it is worth noting that the P-value (0.043) of increased macrophages in the mixed endometriosis group was very close to 0.05; thus, this should be analyzed further with a larger or validation dataset. The proportions of CD56 + NK cells did not change in this study, which may be due to the fact that their functions in endometriosis depend on their cytotoxicity and not on cell numbers.
A history of endometriosis was correlated with decreased BMI. A consistent inverse association has also been observed between BMI and endometriosis in other studies (50,51), which may also relate to hormonal differences between heavier and lighter women. Moreover, obesity is associated with lower E2 levels (52,53). In this study, a history of endometriosis was also associated with increased baseline E2 levels, with a weak correlation between baseline E2 levels and uterine macrophages. Previous work has shown that that macrophage infiltration of endometriosis increases in response to E2 (47). Estrogenstimulated uterine growth factors such as the colonystimulating factor-1 or granulocyte-macrophage colonystimulating factor promote macrophage migration and infiltration (48). These partly explain why E2 and macrophages were associated with endometriosis. Due to the difference of sampling dates, this study further analyzed the correlation between CD68 and estrogen receptors in EndometDB. The data in EndometDB showed that CD68 was slightly positively correlated with ESR2, but not ESR1 (Supplementary Figure  S7). Previous research has shown that ESR2 upregulates CCL2 via NF-kB signaling in endometriotic stromal cells and recruits macrophages to promote the pathogenesis of endometriosis (54). On the other hand, M1 and M2 macrophage infiltration is comparable between ESR1 knockout mice and wild type mice (55), which suggests the potential effect of ESR2 in the recruitment of macrophages. This, therefore, suggests the possibility of developing appropriate antagonists of macrophage estrogen signaling as novel therapeutic agents in endometriosis.
Nowadays, the gold standard for diagnosis of endometriosis is still the invasive laparoscopic visualization of lesions (56). The main treatments for this disease include hormone therapies, with limited effectiveness and considerable side effects, and invasive surgery (3). The recurrence rate after surgery is high, and repeat surgery increases the risk of premature ovarian failure, adhesion, and organ injury (57). Therefore, the findings of this study regarding immune responses, and the interaction between endocrine and immune responses in women with endometriosis, provide new insights for the diagnosis and treatment of endometriosis. Although this study demonstrated an association between immune responses and a history of endometriosis, it still has limitations. Firstly, it is a retrospective study; thus, a welldesigned prospective study should be conducted to verify the above results and determine more potential mechanisms. The expression of surface receptors, the secretion of cytotoxic granules and cytokines of peripheral NK, and the cytokine secretion of peripheral T cells should also be collected from the entire study population. Subgroups of macrophages, NK cells, T cells, and dendritic cells should also be analyzed in future research. Secondly, basic research should be performed to verify the underlying mechanisms of low peripheral lymphocyte cytotoxicity, and to study the immune cell profiles of ectopic and eutopic endometria, as well as the functions and mechanisms of each immune cell type. Finally, not all patients underwent laparoscopic observations to confirm the absence of endometriosis in the control and endometriosis group at the time of endometrial and peripheral blood sampling. Moreover, there were no totally healthy control patients in this study. Therefore, future work should pay close attention to the inclusion and exclusion criteria in order to limit the unknown bias.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by The ethics committee of Shenzhen Zhongshan Urology Hospital. The patients/participants provided their written informed consent to participate in this study.