Decreased PD-1+ NK and T Cell Populations in Peritoneal Fluid contribute to Immune Dysregulation in Endometriosis

T.T. and B.V. conceptualised the study and planned experiments. A.T., N.T., C.J., N.B., S.B. and T.T. collected samples and isolated cells. A.T., N.T. and E.B. performed flow cytometry experiments. A.T., E.B. and T.T. analysed data. A.T. and T.T. wrote the manuscript. All authors reviewed and approved the submission.

This study was approved by the Monash Health Human Research Ethics Committee (HREC). All samples from consenting women were collected under Human Research Ethics Application 01067B with 20‐0000‐882A or under 20‐0000‐159A.

This work was supported by Endometriosis Australia, RG2022.

All samples from consenting women were collected under Human Research Ethics Committee approvals 01067B with 20‐0000‐882A or under 20‐0000‐159A. The participants (Table  S1 ) were women between 18 and 45 years old (the average age of people in the study was 31.2 ± 2.28 years old) who were undergoing laparoscopy for suspected endometriosis or unrelated health conditions. After obtaining informed consent, PF and blood were collected at the time of surgery. Exclusion criteria were pregnancy, malignancy and menopause. Hospital surgical records were used to confirm the diagnosis of endometriosis and the disease stage, hormone state and menstrual cycle stage. Only samples derived from patients diagnosed with Stages I and II of the disease were included in the endometriosis group. The control and endometriosis cohort did not significantly differ in clinical characteristics. Characteristics CTRL ( n  = 8) Endometriosis ( n  = 6) p Statistical test Age (median) 33.50 30.00 0.43 Mann–Whitney U test Menstrual phase Menstrual 12.50% 33.33% 0.54 Fisher's exact Proliferative 12.50% 33.33% 0.54 Fisher's exact Secretory 75.00% 33.33% 0.28 Fisher's exact Hormonal contraception Yes 37.50% 33.33% > 0.99 Fisher's exact No 62.50% 66.67% > 0.99 Fisher's exact Comorbidities Present 37.50% 33.33% > 0.99 Fisher's exact Absent 62.50% 66.67% > 0.99 Fisher's exact Adenomyosis Present 33.33% 16.67% > 0.99 Fisher's exact Absent 66.67% 83.33% > 0.99 Fisher's exact Blood samples were processed as described earlier [ 36 ]. Briefly, the EDTA blood was centrifuged at 1800 g for 5 min at RT and the supernatant removed. The RBCs in the pellet were lysed using RBC lysis buffer (Cell Signalling, NEB) in three sequential lysis steps in separate falcon tubes, then combined, washed in PBS and counted. PF samples were collected in 50 mL Falcon tubes and processed between 1 and 6 h after collection. Briefly, PF was centrifuged at 1800 g for 5 min at RT for supernatant removal. If the PF showed RBC contamination, a RBC lysis step was performed; otherwise, the cell pellet was washed in PBS and centrifuged at 500 g for 5 min at RT before being resuspended in PBS for cell counting. The antibodies used are listed in Table  S2 . All antibodies were titrated for optimal staining. For measurements, 2 × 10 5 cells in stain buffer (2% FBS, 0.5% EDTA in 1X PBS) were labelled with single antibodies and 1 × 10 6 cells with the full panel according to the manufacturer's instructions. 2.5 μL of working solution ViaDye Red Fixable Viability Dye was added to the cell pellet. Tubes were centrifuged at 400 g for 5 min at RT, followed by supernatant decanting and blotting on a paper towel. After a 20‐min incubation step in the dark, samples were washed in 3 mL of Stain Buffer and centrifuged at 400 g for 5 min at RT followed by decanting and blotting on a paper towel. Each sample was resuspended in 150 μL of Stain Buffer. For intracellular staining, tubes were then fixed in 300 μL of 1% paraformaldehyde (PFA), permeabilised in 0.1 X Triton X‐100 in deionised H 2 O, and stained with anti‐FOXP3. Cells were washed in stain buffer and resuspended in 300 μL of stain buffer for acquisition. All samples were measured with the Cytek Aurora (configuration 5L UV/V/B/YG/R) using SpectroFlo software. Daily QC was performed and passed with SpectroFlo QC beads before sample acquisition. The same acquisition settings were used for each sample type. The gating scheme is included in Figure  S1 for myeloid cells and in Figure  S2 for lymphoid populations. Autofluorescence was compensated for using unstained samples for each method of preparation. Single‐stain controls were established on PBCs, PFCs and CompBeads for accurate unmixing. Cell populations were defined by markers as laid out in Table  S3 . An unstained sample, a single‐colour control and an isotype control were established for each marker in both blood and PF, to verify that all antibodies included in the study could discriminate between positive and negative populations. A representative example is provided in Figure  S7 . PF and blood samples were collected from biological replicates of endometriosis patients (PF: n  = 6, blood: n  = 5) and controls (PF: n  = 8, blood: n  = 7), with staining and analysis performed once per sample. All data and plots were analysed with GraphPad Prism using the non‐parametric Kruskal–Wallis test or the Mann Whitney U test. Asterisks indicate the statistical significance. * p  < 0.05. No Artificial Intelligence Generated Content (AIGC) tools such as ChatGPT and others based on large language models (LLMs) have been used in the preparation of this manuscript.

On initial inspection of the overall immune cell populations side‐by‐side (Figure  1A ), the tSNE maps showed the PF to be dominated by myeloid cells, the peripheral blood by lymphoid cells (all marker tSNE plots in Figure  S3 ). As expected, several populations seemed exclusively present in one or the other sample species owing to the tight regulation of the presence of cells in circulation and in the peritoneal cavity, and to the transition of cells between the two compartments. Interestingly, in endometriosis patients the macrophage population in PF as well as the CD4 + and CD8 + T cell populations in both blood and PF appeared to be diminished. When quantifying the relative percentages of immune cells in PF versus blood in both controls and endometriosis patients, we found a significantly decreased number of CD8 + T cells in PF in endometriosis (Figure  1B ), with other populations largely similar. The immune environment varies demonstrably between peritoneal fluid (PF) and blood and between endometriosis patients and controls. Blood and PF samples from endometriosis patients and controls were analysed by full‐spectrum flow cytometry, concatenated, displayed using the viSNE algorithm [ 33 ] and separated according to sample species (peritoneal fluid, n  = 6 endometriosis and n  = 8 control samples; blood, n  = 7 endometriosis and n  = 4 control samples). (A) The resulting tSNE maps of major immune cell subpopulations highlight different distributions of immune cells among endometriosis patients and controls. (B) Quantification of immune populations shown in the tSNE plots (gating strategy in Figures  S1 and S2 ). Data include mean ± SD, analysed using FlowJo. * p  < 0.05, Mann–Whitney U test. We were interested in the respective developmental trajectory of immune cell populations, i.e., the relative abundance of monocytes and macrophages, T and B cells, DCs and NK cells according to activation and maturation state, and thus visualised these different immune configurations using self‐organising maps, or SOM (‘FlowSOM’ when analysing flow cytometry data), automatically clustered into tree‐like maps of cell populations, where each node's diameter represents the abundance of population and the distances represent the degree of difference between markers expressed (x‐shift algorithm, Figure  2 ). This approach allows for a more nuanced view of immune cell populations than manual gating in x – y plots. Thus, we identified 14 immune cell populations with notable differences between endometriosis patients and controls. In PF, the endometriosis group exhibited an expansion of NK and B cells compared to controls (circled), while in blood T cells were most prominent. Monocyte and T cell subsets differ in endometriosis patients both in blood and peritoneal fluid. Blood and peritoneal fluid (PF) samples from endometriosis patients and controls were analysed by full‐spectrum flow cytometry and the differences in hierarchic population sizes displayed using the x‐shift and FlowSOM algorithms (A, peritoneal fluid cells, n  = 5 endometriosis and n  = 4 control samples; B, blood cells, n  = 4 endometriosis and n  = 3 control samples). Each node represents a cell cluster, its size the abundance of the population. The resulting minimum spanning tree graphs highlight the increase in the peritoneal myeloid compartment in endometriosis (A) and in the lymphoid compartment in the blood (B). Data were analysed using FlowJo. Interestingly, in blood, endometriosis samples were characterised by increased levels of CD8 + T cells, specifically in CD8 + T RM cells and CD68 + Mϕ (all populations in Figure  S4 ). However, detailing the differences between endometriosis patients and controls in blood samples was challenging due to our small sample size. To investigate the putative role of immune cells in endometriosis pathophysiology, we next analysed the activation status of the respective populations in PF and blood cells by measuring the differences in population size of cells expressing CD69, CD127, CD27, CD25, CD16 and CD40 between endometriosis and control samples (Figure  3 ). The early lymphocyte activation marker, CD69 appeared widely on PF cell populations, irrespective of disease state; it was expressed less widely on blood cells (Figure  3A ). CD25 was found on a large proportion of CD4 + T cells in PF irrespective of disease state; in blood, the expression of CD25 on CD4 + T cells was scarce but significantly higher in endometriosis compared to controls (Figure  3B ). Interestingly, B cells in blood expressed CD25 widely, again with a significant increase in endometriosis compared to controls, and CD25 was also expressed by myeloid cells like monocytes and DCs. The combination of CD25 and CD127 (the IL‐7 receptor) indicating the presence of memory cells was found widely expressed on T cells in PF irrespective of disease state, mainly driven by CD8 + T cells, while in blood, myeloid cells also showed CD127 expression (Figure  3C ). Interestingly, CD8 + T cells in blood showed significantly wider expression of CD127 in endometriosis compared to controls. CD16 expression did not vary within the monocytic or myeloid compartment as a whole but was significantly more widely present on T cells in endometriosis compared to controls (Figure  3D ); in PF, the overall expression was rare compared to blood, but in both sample species, CD8 + T cells expressed CD16 significantly more widely in endometriosis than in controls; with CD4 + T cells showing the same in blood only. Of note, B cell and T cell expression of CD16 seemed almost inverse in blood in comparison to PF. CD27 as marker of B cell memory was significantly more widely carried on B cells in blood from endometriosis patients than in controls; the difference in PF was not significant (Figure  3E ). Overall, T cells in blood showed the same increased expression of CD27, predominantly driven by CD8 + T cells; no significant differences were seen in PF cells. The interaction between CD40 and CD154 (CD40L) is paradigmatic in immunology, activating CD4 T cell help for B cell class switching and for macrophage activation. We found CD40 widely expressed on monocytes and neutrophil granulocytes in PF, and on lymphocytes and DCs in blood (Figure  3F ). In addition, CD40 was significantly more widely expressed on blood lymphocytes in endometriosis, again with CD8 + T cells the main population to contribute. The activation of lymphoid immune cells is enhanced in blood and peritoneal fluid of endometriosis patients. Heatmaps showing the expression of activation markers on nine cell populations in peritoneal fluid and blood of endometriosis patients ( n  = 6) and controls ( n  = 8). Cells were gated by FSC/SSC and CD45 expression (see Figures  S1 and S2 , and Table  S3 ) into: all CD45 + cells, macrophages/monocytes, CD11c + DCs, CD3 + (all) T cells, CD3 + CD4 + T cells, CD3 + CD8 + T cells, CD19 + B cells, CD3 − CD19 − NK cells, neutrophils. (A–F) Scale bar, % of population positive for marker; (G) expression of CD16 on peritoneal fluid (PF) CD8 + T cells (data from D); representative flow cytometry plots and quantification shown for controls and endometriosis; (H) expression of CD40 on CD8 + T cells in blood (data from F); representative flow cytometry plots and quantification shown. * p  < 0.05 ** p  < 0.01, Mann–Whitney U test. This wide expression of key activation markers across multiple immune cell populations suggests a generalised pattern of immune activation in peripheral blood, in line with the paradigm of endometriosis as a systemic disease. Interestingly, the expression of CD16 on CD8 + T cells within the PF (Figure  3G ) corresponded to an increase in CD40 + CD8 + T cells in the circulation (Figure  3H ). The highly activated T cell compartment led us to wonder whether an increased expression of immune regulatory mechanisms or immune checkpoints could explain the persistence of endometriosis lesions within the peritoneal cavity despite this activation. We thus investigated the expression of FOXP3 as an indicator of T REG presence as well as the expression of PD‐1 as the paradigmatic immune checkpoint in PF and blood cells. We were surprised to find a significant decrease in PD‐1 expression overall in the PF cells (Figure  4 ), and an inversion of this picture in blood, while no significant differences were seen with regards to FOXP3 expression (Figure  S5 ). On closer inspection of PD‐1 + cell subsets within the NK and T cell compartments of the samples, we saw that the decrease in PD‐1 expression was not uniform across all cells but apparently due to populations missing in the PF endometriosis samples. In NK cells, the difference is mainly due to the CD56 DIM subset (Figure  4A ), while we found very few CD56 HI NK cells. In the T cell compartment, PD‐1 expressing T cells were significantly decreased in PF in endometriosis (Figure  4B ). Stratifying the data by menstrual cycle phase did not yield further insight, and warrants further investigation in larger cohorts (Figure  S6 ). PD‐1 expression is diminished in peritoneal fluid NK and T cells in endometriosis. The overall expression of the immune checkpoint marker PD‐1 was measured on overall CD45 + cells and NK and T cell subpopulations in peritoneal fluid (PF) and blood from endometriosis patients ( n  = 6) and controls ( n  = 8). PD‐1 + NK cells (A) and CD4 + T cells (B) were highlighted to determine differential distribution of PD‐1 expressing subpopulations within these compartments. (A) PD‐1 + NK cell distribution in peritoneal fluid and blood shows a decrease in populations in endometriosis compared to controls in tSNE maps. Flow cytometry plots of CD56 HI and CD56 DIM subsets versus PD‐1 show this to be mainly the CD56 DIM subset (representative plots shown). Quantification of data in (C). (B) PD‐1 + T cell populations in PF are shown to be lacking in endometriosis compared to controls in tSNE maps. Representative flow cytometry plots of T cells versus PD‐1 expression shown. Quantification of data in C. Data are mean ± SD; * p  < 0.05, Mann–Whitney U test. Overall, our results indicate the absence of specific populations of PD‐1 + T cells and PD‐1 + NK cells in the PF of women with endometriosis, along with significant activation of lymphoid cells in the blood. Investigating these populations further with regard to clonal activation and selective proliferation might hold a clue to understanding where the immune system goes awry in endometriosis.

Here, we investigated the composition, activation status and immunoregulation of PF and blood in women with and without endometriosis. This followed an earlier study using CyTOF and a similar marker panel [ 29 ]; however, we had not included FOXP3 and PD‐1 on that occasion. While we had expected to see increased regulation and inhibition of PF T cells as a permissive mechanism in endometriosis, we were surprised to see the opposite pattern emerge, with PD‐1 expression significantly decreased on CD4 + T cells and NK cells. The observation that PD‐1 levels are decreased in this environment suggests an explanation for the increased activation. The activation patterns themselves were reproduced from our earlier study to a degree, with CD69 more widely expressed in PF cells than in blood, CD25 on CD4 + T cells in PF more than in blood, CD127 on CD8 + T cells in both, PF and blood, CD16 on B cells in PF more than in blood, CD27 widely expressed on T cells, especially CD8 + T cells in PF and blood, and a strong CD40 signal on monocytes in PF. Considering the different modalities of measurement—FSFC versus CyTOF—these findings are remarkable. Our findings suggest that T cells are primed to actively engage in immune responses, potentially clearing endometriotic lesions. Given the reduction of PD‐1 expression, we would expect T cells to remain uninhibited via the PD‐1/PD‐L1 pathway, thereby allowing for more robust immune activity. However, despite this activation, lesion clearance does not occur, indicating a disruption in the immune response. One possible explanation for this failure is that T cell function may be impaired by the chronic and persistent inflammation in the peritoneal cavity, which could hinder effective immune responses. Alternatively, endometriotic cells may fail to express sufficient antigens or may not be recognised as ectopic tissue by the immune system, allowing them to escape immune surveillance despite the absence of PD‐1 inhibition. These potential mechanisms highlight the complexity of immune dysregulation in endometriosis and point to the need for a deeper understanding of how immune cells interact with endometriotic lesions. Dysregulated T cell activation and lack of PD‐1 driven inhibition point towards an interesting angle of anti‐inflammatory therapy in endometriosis. When going beyond NSAIDs, current practice suggests large‐scale immune suppression, for example, through corticosteroids like prednisolone; however, severe side‐effects are not acceptable in endometriosis, despite the severe disease burden a ‘benign’ condition, and broad, non‐specific interventions in immune function risk adversely affecting uterine receptivity and embryo implantation [ 34 ]. In our study, we included participants aged 18–45 years, with exclusion criteria comprising pregnancy, malignancy and menopause. While the sample size was limited to 14 participants due to the 1‐year collection period, this serves as a valuable starting point. To achieve definitive results, larger cohorts and extended study durations will be necessary. We recognise that immunological profiles may vary with age [ 35 ]; however, we aimed to include women across the entire 18–45 age range to identify an endometriosis‐specific signature that remains consistent through different stages of life. We included patients with Stage I and II endometriosis as these early stages are more biologically active than advanced stages. Identifying a specific signature for early‐stage endometriosis is crucial to understanding the initial immunological mechanisms. This could also point towards specific interventions to prevent the disease progressing further. This study included participants with potential confounding factors that may have influenced the results. Specifically, some individuals in both the endometriosis and control groups presented with comorbidities such as adenomyosis, mature cystic teratoma and polycystic ovary syndrome (PCOS). Additionally, participants using hormonal contraception were included in the analysis. While these factors may have introduced variability, they reflect the real‐world clinical presentation of patients with endometriosis, who often exhibit comorbidities. Despite this, any immune alterations specifically attributable to endometriosis should remain discernible, given the significant immune dysregulation typically associated with the disease. Future studies should consider stratifying these variables to better delineate the immune signature of endometriosis. The question of how and why endometriosis lesions persist in the peritoneal cavity remains unanswered for now. It would be interesting to investigate whether the T cells lacking PD‐1 expression are a clonal population, and if so, what antigen they might recognise. The sample size of 14 PF and 12 blood samples is reflective of the challenges associated with collecting paired clinical specimens, and larger cohorts will be needed to fully validate these observations.

The dysregulation of the immune system in endometriosis is characterised by a lack of intrinsic inhibition through the immune checkpoint inhibitor, PD‐1. The question of whether this non‐inhibited, activated T cell compartment is clonal in nature warrants further investigation. These findings are of particular interest with a view to treat endometriosis‐derived pain and inflammation in the peritoneal cavity.

Endometriosis is a steroid hormone‐dependent [ 1 ] chronic inflammatory disorder characterised by chronic and acute pelvic pain, dysmenorrhea, pain at ovulation, dyspareunia, abnormal bleeding, pain with bowel movements, fatigue and infertility [ 2 ]. Treatment options are limited, with non‐steroidal anti‐inflammatory drugs (NSAIDs) the main medication to manage endometriosis‐associated pain; hormonal treatments like combined hormonal contraceptives, progestogens, gonadotropin releasing hormone (GnRH) agonists or antagonists used to shut down the menstrual cycle and thus alleviate endometriosis‐related symptoms [ 3 ], and surgery to try and remove endometriosis lesions [ 4 ]. However, even the latter option is not guaranteed to prevent the recurrence of lesions and pain [ 5 ]. The lesions grow within the peritoneal cavity, a unique microenvironment with residual peritoneal fluid (PF) as lubricant between organs. Immune cells in the PF directly interact with endometriotic lesions, and endometriosis‐related pain is thought to be triggered by extra‐uterine menstruation from these lesions [ 6 ]. A connection between endometriosis and the immune system has long been suspected [ 7 ], and a recent analysis of UK Biobank data found an association with autoimmune diseases especially apparent [ 8 ]. Endometrial debris would normally be cleared by the immune system but is thought to persist in endometriosis due to a dysregulated immune response [ 9 ]. The dysregulated immune response involves altered immune cell populations, cytokine imbalances and chronic inflammation and is thought to play a central role in the disease's development and progression [ 10 ]; however, the exact role of the innate and adaptive immune compartments in endometriosis pathophysiology remains unclear. Monocytes and macrophages normally initiate inflammation, clear debris and promote tissue repair [ 11 ], while the lymphoid compartment orchestrates adaptive immune responses, regulates inflammation and ensures long‐term immunity [ 10 ]. NK cells share features of both innate and adaptive immunity and are specialised in the recognition and elimination of abnormal or infected cells [ 12 ]. If functioning correctly, these components work together to protect against pathogens, control inflammation and maintain immune homeostasis, while dysregulation results in aberrant immune tolerance, impaired immune surveillance and chronic inflammation [ 10 ]. In endometriosis, however, dysregulation prevails; macrophages (Mϕ) can promote or prevent disease progression depending on their location or origin [ 13 ], with evidence of M1 macrophage predominance in eutopic endometrium and M2 macrophages more abundant in PF of endometriosis patients [ 14 , 15 ]. T cells are thought to orchestrate a dysregulated immune response in PF [ 16 ]. Chronic inflammation and cytokine production are further enhanced by overexpression of nuclear factor κB (NFκB) and activation of mitogen‐associated kinase (MAPK) signalling pathways in macrophages, with subsequent production of reactive oxygen species [ 10 , 17 ]. These attract further monocytes, eosinophils, and T lymphocytes and exacerbate the inflammatory response associated with endometriosis [ 18 ]. NK cell cytotoxicity has been found deficient in endometriosis patients [ 12 , 19 , 20 ]. Dendritic cells (DCs) have been shown to attenuate lesion development in mouse models through activation of T lymphocytes [ 21 ], while being indispensable for lesion formation at the same time [ 22 ]—a finding that illustrates the difficulty of modelling an immunological complex disease adequately in mice. Regulatory T cells (T REG ) have a role in endometriosis pathogenesis by promoting immune suppression and limiting the immunological response to endometriotic lesions [ 23 , 24 ]. CD4 + T H 2 and T H 17 cells are overrepresented in endometriosis [ 24 ], the latter induced through IL‐23 [ 25 ]. The apparent sheer complexity of the immune system's response to endometriosis has made any investigation of the overall immune status of patients exceedingly difficult, until the advent of high parametric flow cytometric methods like cytometry by time‐of‐flight (CyTOF) and full‐spectrum flow cytometry (FSFC) made these analyses feasible very recently [ 26 , 27 ]. While the myeloid compartment has previously been investigated using FSFC [ 28 ], an earlier study identified an increase in CD69 + T cells in the PF of endometriosis patients with reduced expression of markers associated with T cell function using CyTOF [ 29 ]. This raised the question of the involvement of immune checkpoint inhibition in T cell function in endometriosis. Recent reports point towards an increase in the expression of PD‐1/PD‐L1 in peripheral blood cells of endometriosis patients [ 30 , 31 ], with one study emphasising a putative role for γδ T cells [ 32 ] but the relevance for the ongoing immune reactions within the peritoneal cavity is unclear, as these studies leave out the comparison with PF cells. Consequently, we set out here to investigate the status of PD‐1 alongside other activation markers in both sample fluids to explain the diminished activation of T cells. We hypothesised that an inhibition of lymphoid cell populations (T cells, NK cells) through PD‐1/PD‐L1 suppressed the immune response against endometrial debris and lesions in the peritoneal cavity, favouring the persistence of lesions.

The authors declare no conflicts of interest.

Figure S1: Gating strategy for monocyte/myeloid leukocyte populations. Figure S2: Gating strategy for lymphoid leukocytes. Figure S3: tSNE markers separating the immune cell populations. Figure S4: Flow cytometry data (frequency of parent, raw data points and ranges) for all cell populations. Figure S5: Statistical analysis of NK and T cell populations for FOXP3 expression and expression of PD‐1. Figure S6: Flow cytometry statistics by menstrual cycle phase. Figure S7: CD69 and CD56 spectral unmixing controls. Representative plots showing the single‐colour controls (A), isotype control (B) and unstained sample (C) used for spectral unmixing. Table S1: Study participants. Table S2: Antibodies used in full‐spectrum flow cytometry. Table S3: Cell populations as defined by surface marker combinations.