Mass spectrometry imaging reveals alterations in protein and N-glycan molecular signatures in endometriosis tissues

We employed MALDI MS imaging techniques to perform intact protein, tryptic peptide, and N-glycan analysis of endometriosis tissues ( 31 , 40 ). Note that in this initial study, we focused on optimizing and analyzing our datasets for each class of analytes (intact proteins, N-glycans, or tryptic peptides) individually. Future studies are directed towards adopting a more integrated approach, for example, recent studies in literature have focused on analyzing several classes of analytes sequentially on the same tissue section and performing co-registration of multiple data sets ( 41 ). For intact protein analysis, we used a sample preparation protocol prior to MALDI-MS imaging of fresh-frozen tissue sections of eutopic endometrial tissue and ectopic endometriosis lesions from patients with confirmed endometriosis determined during laparoscopy ( Scheme S1 ) ( 31 ). Results including molecular images and representative MALDI mass spectra are shown in Fig. 1 . After MALDI-MS imaging, the H&E-stained tissue sections were analyzed by histopathology to identify and delineate endometriosis lesions, characterized by a minimum of two of three histological features including endometrial glands, endometrial stroma, and hemosiderin. We then correlated the tissue regions of endometriosis with the spatial distribution and intensity of protein ions detected. The most abundant protein ions detected from the tissues analyzed were neutrophil defensins 1, 2, and 3 (DEFA 1, DEFA2, and DEFA3), histone H4, and hemoglobin (Hb) α and β. In endometriosis tissues, high relative abundances of DEFA 1, DEFA2, and DEFA3 were detected in regions of endometrial glands and stroma. Endometrium tissues presented high relative abundance of histone H4 and a more homogenous distribution Hb proteins α and β ( Fig. 1A and 1B ). When comparing the peak area of the major proteins detected ( Fig. 1C ), a significantly higher peak area for DEFA1 and DEFA2 was observed in endometriosis tissues in comparison to endometrium (t-test; p = 0.02 for DEFA1; p = 0.01 for DEFA2), while a nonsignificant increase in peak area was obtained for DEFA3 (t-test; p = 0.1). A significantly higher average peak area was detected in endometrium tissues for proteins Hb α (t-test; p = 1.95×10 −6 ) and Hb β (t-test; p = 1.12×10 −6 ), while a nonsignificant increase in peak area was observed for histone H4 (t-test; p = 0.1). Yet, despite the use of the lysis buffer protocol, a low number of intact proteins were detected from the tissues, and thus we performed N-glycan and peptide imaging to improve molecular coverage. Scheme S2 depicts the sample preparation workflow adapted from literature to perform subsequent N-glycan and tryptic peptide MALDI MS imaging on the same tissue sections ( 40 ). Using this protocol, we detected various N-glycans in endometriosis lesions and eutopic endometrium tissues as seen in the representative mass spectra of each tissue type ( Fig. 2A – C ). A variety of complex N-glycans, which refer to N-glycans in which the α3- and α6-linked mannose sites of the pentasaccharide glycan core (Hex3HexNAc2) are substituted with N-acetylglucosamine (GlcNAc) moieties, were detected including m/z 1663.58 (Hex5HexNAc4), m/z 1809.64 (Hex5dHex1HexNAc4), and m/z 2012.72 (Hex5dHex1HexNAc5). Other major classes of N-glycans detected include high mannose N-glycans, which refer to those containing only unsubstituted terminal mannose sites on the glycan core, such as m/z 1257.42 (Hex5HexNAc2) and m/z 1419.47 (Hex6HexNAc2). Hybrid N-glycans, which refer to those containing both unsubstituted and GlcNAc substituted mannose residues, were also detected, including m/z 1460.50 (Hex5HexNAc3). To explore alterations in N-glycan molecular profiles in endometriosis tissues, SAM was employed to compare eutopic endometrium and endometriosis lesions and identify statistically significant changes in the relative abundance of m/z values between tissues. Statistical significance by SAM is determined through calculation of a “d-score”, which computes the average change in the normalized peak abundance for each m/z between eutopic endometrium and endometriosis sample sets. In this approach, numerous random permutations of the original dataset are used to simulate the distribution of the expected relative differences and estimate the false discovery rate (FDR), which represents the rate at which a molecular ion will be incorrectly identified as significant. Greater deviation of the observed data from the expected value is used to determine significance by SAM. Although originally applied to analyze gene expression microarrays, SAM has been applied to MS imaging datasets as it similarly involves analysis of high-dimensional datasets with hundreds to thousands of m/z values detected. As a permutation-based resampling approach that controls for multiple testing using an FDR, SAM is a robust statistical method that has strong utility in capturing meaningful differences in molecular profiles, especially in the multivariate and complex datasets collected my MS imaging ( 42 , 43 ). Using SAM, 19 N-glycans were found to have significantly higher abundance in endometriosis lesions when comparison to eutopic endometrium ( Table S2 ). Boxplots depicting the ion abundances of the five most significant SAM features detected significantly higher in endometriosis compared to endometrium are depicted in Fig. S1 . Of these, 16 N-glycans were classified as both complex and fucosylated N-glycans ( Fig. 3A ). In addition, increased levels of branching from biantennary (5 out of 19 features) to tri- and tetra-antennary structures were observed for seven of the significant ions, while 10 out of 19 significant ions were determined to have a sialic acid group. Representative ion images of the N-glycans detected as significantly higher in endometriosis lesions are depicted in Fig. 3B . N-glycans selected as significantly higher in endometriosis by SAM included m/z 771.26, m/z 2158.78, 2304.82, and m/z 2174.77, which were localized to regions of glands and stroma in endometriosis lesions, while a more homogenous distribution of N-glycans were observed in endometrium tissues. Interestingly, intact neutrophil defensin proteins 1, 2, and 3 at m/z 3442.52, m/z 3371.48, and m/z 3485.51 were also detected in the same experiment at a lower signal intensity in endometrium tissues in comparison to endometriosis lesions, as we had observed in the intact protein MALDI experiments ( Fig. S2 ). Note that for the intact protein dataset, a t-test was used to compare the peak areas of a limited number of identified proteins. Conversely, SAM was applied to the N-glycan dataset, which included hundreds of detected N-glycans, as it is more suitable for the analysis of high-dimensional data which ensures a robust identification of significant molecular changes in molecular profiles. We next evaluated the results from tryptic peptides experiments. Representative mass spectra of eutopic endometrium and endometriosis tissue lesions of various tryptic peptides are shown in Fig. 4A . Putative identifications for the most abundantly detected peptides are summarized in Table S3 and S4 . Introduction of the tryptic digest step improved the molecular depth and increased the number of proteins detected when compared to the intact protein analysis. Several of the most abundant tryptic peptides detected include fragments from large structural and connective proteins including collagen, tubulin, actin, and filamin proteins with intact masses exceeding 40 kDa which are challenging to ionize and detect in their intact form. Although thousands of tryptic peptides were detected ( Fig. 4A ), many were not co-localized with endometriosis lesions. Thus, instead of a global comparison of all detected m/z values using SAM, we conducted a spatial analysis which focused on the m/z values detected at a higher abundance within regions of endometriosis lesions compared to the surrounding tissue. We used Pearson’s correlation analysis to find m/z values co-localized to regions of endometriosis lesions and identified 29 m/z values above a correlation threshold of 0.4 ( Table S5 ). Putative identification of these ions is provided in Table S6 , and ion images of top co-localized features are depicted in Fig. 4B . These include ion images of m/z 1612.94, putatively identified as high mobility group nucleosome-binding domain-containing protein 3 (HMGN3) (Pearson’s correlation = 0.61), m/z 985.57, putatively identified as tubulin alpha-4a chain (Pearson’s correlation = 0.51), and m/z 929.55, putatively identified as collagen alpha-1 (XIV) chain (Pearson’s correlation = 0.45), and others. As observed in the images, the peptide ions were highly localized to endometrial glands and stroma in endometriosis lesions, while a more homogenous distribution of these ions were detected throughout eutopic endometrium tissue.

Acetonitrile, acetic acid, chloroform, ethanol, trifluoracetic acid, water, MS-grade trypsin protease, super-DHB (9:1 (w/w) 2,5-dihydroxybenzoic acid:2-hydroxy-5-methoxybenzoic acid), and α-cyano-4-hydroxycinnamic acid (CHCA) were purchased from Fisher Scientific (Waltham, MA). Peptide:N-glycosidase F (PNGase F) was purchased from Bulldog Bio. This study was performed under approved IRB protocols at UT Austin (protocol #2017080087) and Baylor College of Medicine (protocol #H-50480). Eligible patients under the care of Dr. Michael Breen at Ascension Seton Medical Center Hospital (Austin, TX) already scheduled to undergo surgery and that provided written consent to participate in our study were included. Endometriosis lesions were resected from a variety of locations within the abdominal cavity while eutopic endometrial tissues were obtained from a subset of patients undergoing full hysterectomies. A total of 34 tissues from 27 patients undergoing laparoscopic resection of endometriosis were prospectively collected and analyzed. Table S1 summarizes the number, types, and MS analyses performed on samples from each patient. For intact protein imaging experiments, n = 11 endometriosis lesions and n = 9 eutopic endometrium samples from n = 17 patients were analyzed. The N-glycan sample set contained n = 7 endometriosis lesions and n = 3 eutopic endometrium samples from n = 9 patients. The tryptic peptide sample set included n = 14 endometriosis lesions and n = 7 eutopic endometrium samples from n = 17 patients. Tissue specimens were stored in airtight containers at 4°C until they could be flash frozen in liquid nitrogen and stored in a −80°C freezer. Tissues were sectioned at 12 μm using a CryoStar NX50 cryostat (Thermo Scientific, Waltham, MA) and mounted onto indium tin oxide (ITO) slides (Delta Technologies). After sectioning, slides were then placed in a desiccator for 15 minutes to dry prior to further sample preparation procedures for MALDI imaging experiments. Hematoxylin and eosin (H&E) stained tissue sections were visualized on a NanoZoomer-SQ digital slide scanner at 40× magnification (Hamamatsu). After H&E staining, pathologic evaluation of the tissue sections was performed by Dr. Suzanne Ledet at the Ascension Seton Medical Center where regions of definite and probable endometrial stroma, endometrial glands, and hemosiderin were identified within the endometriosis lesions. Eutopic endometrial tissue was also confirmed by pathology. Tissue sections were prepared using a lysis buffer protocol we have detailed in literature ( Scheme S1 ) ( 31 ). The use of the buffer directly on tissue sections facilitates hemolysis and erythrocyte removal in order to selectively reduce ion suppression from blood proteins including hemoglobin and enhance detection and molecular coverage of other endogenous tissue proteins. An HTX M5 sprayer (HTX Technologies) was used to apply super-DHB matrix (40 mg/mL, 90/10/0.1 acetonitrile/water/trifluoroacetic acid) over 12 passes at a flow rate of 0.1 mL/min, nozzle velocity of 1200 mm/min, a nozzle height of 40 mm, track spacing of 2 mm, and nozzle temperature of 75 °C. Rehydration in a sealed chamber containing 50% acetic acid vapor for 5 min at 37 °C was utilized. MALDI imaging data was acquired on a Bruker RapifleX MALDI TOF/TOF mass spectrometer (Bruker Daltonics) in linear positive-ion mode from m/z 2000 to 24000. Mass calibration was performed using Protein Calibration Standard I (Bruker Daltonics). MALDI images were collected at a spatial resolution of 50 μm using full pixel ablation and 2000 laser shots per pixel. Mass resolution was optimized for m/z 12000. Mass spectral information recorded on FlexImaging 5.1 (Bruker Daltonics) was imported into SCiLS Lab MVS 2025a Core ( http://scils.de/ ; Bremen, Germany) after acquisition for preprocessing and visualization. A baseline subtraction (convolution algorithm), normalization (root mean squared algorithm), and weak spatial denoising were applied to all data acquired prior to generating average representative mass spectra over measurement regions. Putative protein identification was performed based on comparison of measured m/z to data from MSiMass List ( http://maldi-msi.org/ ). Statistical Analysis of Intact Proteins: For comparison of peak areas between two groups, a t-test assuming normal distribution was used with a significance threshold p < 0.05. For N-glycan imaging experiments, after sectioning and drying in a desiccator, tissue sections were washed in ethanol (70%, 100%, 30 s each), Carnoy’s fluid (2 min), ethanol (100%, 30 s), water (30 s), and ethanol (100%, 30 s). PNGase F (0.1 μg/μL in water) was applied using a HTX M5 sprayer (HTX Technologies) over a series of 15 passes, flowrate of 0.025 mL/min, nozzle velocity of 1200 mm/min, a nozzle height of 40 mm, track spacing of 3 mm in a crisscross pattern, and nozzle temperature of 45°C to cleave N-linked oligosaccharides from proteins. The samples were incubated in a 37°C oven for 2 hours in a prewarmed sealed polystyrene petri dish (100 mm diameter x 15 mm depth) lined with a disposable cloth wipe (WypAll) saturated with water. A heating pad held at a temperature of 40°C was placed on top of the petri dish within the oven during the incubation to prevent condensation and delocalization of analytes. After enzymatic digestion, CHCA (10 mg/mL, 70% acetonitrile, 0.1% trifluoroacetic acid) was applied using a HTX M5 sprayer (HTX Technologies) over a series of 4 passes, flowrate of 0.12 mL/min, nozzle velocity of 1200 mm/min, a nozzle height of 40 mm, track spacing of 3 mm in a HH spray pattern, and nozzle temperature of 75°C. MALDI imaging data were acquired on a Bruker timsTOF fleX MALDI mass spectrometer (Bruker Daltonics) from m/z 600 to 3500. Mass calibration for N-glycan peptide was performed using Red Phosphorous (Sigma-Aldrich). MALDI images were collected at a spatial resolution of 50 μm using 1500 laser shots per pixel. Tune parameters were optimized for sensitivity of N-glycans as follows: Funnel 1 RF of 450 Vpp; Funnel 2 RF of 500 Vpp; Multipole RF of 500 Vpp; Collision energy of 10 eV; Collision RF of 2500 Vpp; Transfer Time of 120 μs; and Pre Pulse Storage of 11 μs. Mass spectral information recorded on FlexImaging 5.1 (Bruker Daltonics) was imported into SCiLS Lab MVS 2025a Core ( http://scils.de/ ; Bremen, Germany) after acquisition for preprocessing and visualization. All imaging data was visualized using the root mean squared algorithm and weak spatial denoising was applied prior to generating average representative mass spectra over measurement regions and further data processing. An N-glycan peak list was created by exporting the images as .imzML files, uploading to METASPACE ( https://metaspace2020.eu/ ), and compiling a list of all annotations with a FDR of 10% generated over all samples using the NGlycDB-v1 database (3667 entries). Relevant annotation and search setting parameters used in METASPACE are as follows: positive polarity, no chemical modifications, sodium adducts, mass tolerance = 10 ppm. The images were exported as .imzML files using only the N-glycan peak list created in METASPACE to reduce data size. Mass spectra from regions of endometrial glands and stroma or eutopic endometrium were extracted using MSiReader. Statistical Analysis of N-Glycans: Established statistical approaches to analyze multivariate molecular data obtained from MS imaging experiments such as Significance Analysis of MicroArrays (SAM) was employed using the “samr” package in the R programming language (version 4.3.1) to determine significantly different N-glycans comparing endometriosis lesions and eutopic endometrium (FDR = 0.01). Putative N-glycan structures were assigned based on comparison to literature reports ( 32 – 37 ). Importantly, as N-glycan identification was performed by matching m/z values to a composition, multiple N-glycan structural isomers are possible for a single m/z value. Putative structures shown in Table S2 represent one example of a commonly reported candidate structure detected from mammalian tissues using MALDI imaging. For information on instrument resolving power and N-glycan identification, please see Supporting Information: Supplemental Information on Glycan Identification . To perform tryptic peptide imaging after N-glycan MALDI-MS imaging, the MALDI matrix was removed by washing the slide in ethanol (70%, 100%, 30 s each), Carnoy’s fluid (2 min), ethanol (100%, 30 s), water (30 s), and ethanol (100%, 30 s). MS-grade trypsin (0.05 mg/mL in 80% 100 mM ammonium bicarbonate pH 8, 10% 100 mM acetic acid, 10% acetonitrile) was applied using a HTX M5 sprayer (HTX Technologies) over 12 passes, flowrate of 0.01 mL/min, nozzle velocity of 750 mm/min, a nozzle height of 40 mm, track spacing of 3 mm in a HH spray pattern, and nozzle temperature of 30°C. Samples were incubated in a sealed polystyrene petri dish (100 mm diameter x 15 mm depth) containing 1 mL of water at 37°C for 4 hours. CHCA was applied using an HTX M5 sprayer. MALDI imaging data were acquired on a Bruker timsTOF fleX MALDI mass spectrometer (Bruker Daltonics) from m/z 600 to 4000. Mass calibration for peptide imaging was performed using Red Phosphorous (Sigma-Aldrich). MALDI images were collected at a spatial resolution of 50 μm using 2000 laser shots per pixel. Tune parameters were optimized for sensitivity of peptides as follows: Funnel 1 RF of 450 Vpp; Funnel 2 RF of 500 Vpp; Multipole RF of 600 Vpp; Collision energy of 10 eV; Collision RF of 2500 Vpp; Transfer Time of 110 μs; and Pre Pulse Storage of 22 μs. Mass spectral information recorded on FlexImaging 5.1 (Bruker Daltonics) was similarly imported into SCiLS Lab MVS 2025a Core ( http://scils.de/ ; Bremen, Germany) after acquisition for preprocessing and visualization. All imaging data was visualized using the root mean squared algorithm and weak spatial denoising was applied. Statistical Analysis of Tryptic Peptides: A Pearson’s correlation analysis was performed in SCiLS lab MVS 2025a Core using the top 1070 peaks selected using the sliding window feature to determine m/z values co-localized to regions of endometriosis lesions. Tryptic peptides were extracted directly from serial sections after enzymatic digestion with trypsin and submitted to the University of Texas at Austin CBRS Biological Mass Spectrometry Facility (RRID:SCR_021728) for protein identification by LC-MS/MS using the Thermo Ultimate 3000 RSLCnano UPLC coupled to the Orbitrap Fusion. Prior to HPLC separation, the peptides were desalted using Millipore U-C18 ZipTip Pipette Tips following the manufacturer’s protocol. A 2 cm long x 75 µm I.D. C18 trap column was followed by a 75 µm I.D. x 25 cm long analytical column packed with C18 3 µm material (Thermo Acclaim PepMap 100) running a gradient from 5–35% B over 60 min. MS data is collected in the orbitrap with resolution set to 120,000, and 3 sec cycle time MS/MS are acquired in HCD ion trap mode. Raw LC-MS/MS data files were analyzed using Proteome Discoverer (version 2.5.0.400) and Sequest HT for database searching and peptide identification ( 38 , 39 ). The number of allowed missed cleavages for each peptide was set to a maximum of 2. No fixed modifications were performed or selected. N-terminal protein acetylation and methionine oxidation were selected as variable modifications. The peptide mass tolerance for precursor ions was 10 ppm and mass tolerance for fragment ions was 0.6 Da. MS/MS spectra were also searched against a contaminant database reported by Hao et al. and possible contaminants were identified as appropriate ( 38 ). Percolator was used to validate mass spectra based on q-value, which represents the minimal FDR at which identifications are deemed significant (Maximum Delta Cn = 0.05, Maximum Rank = 0, Target FDR (strict) = 0.01, Target FDR (relaxed) = 0.05).( 39 ) In the Percolator algorithm, the FDR was calculated using a target-decoy strategy to estimate the number of false positive protein identifications ( 39 ). The LC-MS/MS data was manually correlated with the MALDI MS imaging data based on comparison of the theoretical m/z values for detected peptides in the LC-MS/MS experiments with the detected m/z values of peptides in the MALDI imaging experiments. Specifically, m/z values for SAM features significantly higher in endometriosis were manually correlated to the detected peptides in the pooled LC-MS/MS endometriosis samples, while m/z values for SAM features significantly higher in endometrium were correlated to peptides detected in the pooled LC-MS/MS endometrium sample. Identifications were assigned within ±0.02 Da mass error. Note that if multiple identifications were equally likely based on mass error, all possible identifications were noted. In the case of multiple protein isoforms or variants identified for the same tryptic peptide, the most general term was reported. Note that in this study no specific post-translational modifications were studied. Further, no quantification measurements were performed, and LC-MS/MS data was only used to identify detected peptides/proteins of interest in the MS imaging data.

In this study, we employed MALDI-MS imaging assays to spatially characterize the distribution of proteins, tryptic peptides, and N-glycans within endometriosis tissues and directly correlate the molecular data to histological features indicative of disease state. As histologic features of interest within endometriosis tissues such as glands and stroma can be observed as small as 50 μm, the spatial resolution of MALDI-MS imaging enabled acquisition of molecular information specific to the tissue microenvironment of endometriosis lesions, which are particularly heterogeneous. MS imaging thus helps overcome challenges related to tissue homogenization steps utilized in traditional LC-MS protocols that results in mixing of lesions with surrounding tissue and a potential decrease in specificity ( 44 ). Capitalizing on these capabilities, statistical analysis employed on the molecular data allowed identification of changes in the abundance of specific molecular species within eutopic endometrium and ectopic endometriosis lesions in tissues from patients with confirmed cases of endometriosis. The results obtained through intact protein analysis suggest that endometriosis may present increased abundance of neutrophil defensin proteins in tissues endometriosis lesions ( Fig. 1 ). Various literature studies have provided evidence that the development of endometriosis lesions involve dysfunction of the immune system ( 45 ). In particular, immune cells such as neutrophils, macrophages, and natural killer cells, among others, have been suggested to play a role in the formation of endometriosis lesions ( 45 ). The immune cells secrete proteins such as cytokines and neutrophil defensins which are a part of the innate immune system in the local endometriosis tissue microenvironment ( 46 , 47 ). Neutrophil defensins belong to a family of antimicrobial peptides and thus display cytotoxic activity against a wide range of microorganisms such as bacteria and fungi. These proteins have also been reported to play a role in regulating inflammatory and immune response in tissues ( 46 , 48 ). Interestingly, our results are in agreement with a previous study by Milewski et al. that quantified levels of neutrophil defensin proteins in the peritoneal fluid of endometriosis patients in comparison to control women ( 48 ). Neutrophil defensins 1–3 were observed to be significantly increased in the peritoneal fluid of endometriosis patients in comparison to control women and were also observed to be correlated with disease severity. Remarkably, the same trend in increased abundances of neutrophil defensins 1 and 2 in endometriosis tissues were also observed in our study. Overall, our results contribute to the growing evidence that there may be an association between neutrophil defensin proteins and endometriosis pathogenesis. Future research steps with focus on longitudinal analyses are thus motivated towards elucidating if the elevated levels of these proteins are a causative factor in endometriosis development or a consequence of the disease. Since the initiation of endometriosis cannot be studied in patient cohorts due to definitive diagnosis requiring symptom presentation and confirmatory laparoscopic surgery after the disease has already developed, experimental models, such as animal studies, are necessary to elucidate events associated with disease initiation. Further corroboration with proteomic data in tissue samples obtained from patient cohorts comparing protein abundance across different stages of endometriosis and its correlation with disease severity, could further clarify the relationship between disease causation and progression. Overall, the distinction of this relationship is critical for understanding the mechanisms underlying development and progression of endometriosis and for identifying potential therapeutic targets and treatment strategies. We also performed MALDI MS imaging of N-glycans comparing eutopic endometrium and endometriosis lesions to evaluate if alterations in N-glycosylation machinery may be involved in endometriosis pathogenesis. Previous studies using MALDI-MS imaging of N-glycans have shown that changes in glycosylation are hallmarks of disease state and that cancer cells frequently display aberrant alterations in glycan structure and composition in comparison to normal cells ( 30 , 34 , 37 , 49 ). As endometriosis shares some similarities with neoplastic processes ( 50 , 51 ), we aimed to evaluate if alterations in N-glycosylation patterns in endometriosis could be observed. Our results showed that the majority of N-glycans selected as significantly higher in endometriosis tissues were complex and/or fucosylated N-glycans. Furthermore, several of the N-glycans significantly higher in endometriosis tissues had increased levels of branching from biantennary to tri- and tetraantennary structures ( Fig. 3B ). Interestingly, previous studies have shown that complex glycans are commonly observed in nontumor stromal regions, although also detected in tumor regions depending on the tumor type ( 30 , 52 ). Still, increased levels of branching of complex biantennary N-glycans to tri- and tetraantennary structures is typically associated with tumorigenic processes ( 30 ). In our study, the N-glycans with tri- and tetra-antennary structures detected at significantly higher abundances in endometriosis lesions suggest that increased levels of N-glycan branching may be associated with disease state. In addition, our data provides evidence that increased levels of fucosylation may be associated with endometriosis lesions. Fucose is an essential sugar that is involved in various glycosylation reactions which play a critical role in many cell-cell interaction and signaling processes ( 53 ). Tumor cells in particular are known to modify their cell surface proteins through increased levels of fucosylation to escape immune recognition which may contribute to decreased adhesion and uncontrolled growth ( 54 ). Our results are in agreement with a study by Vincente-Muñoz et al. which observed increased plasma concentrations of fucose in endometriosis patients in comparison to control women ( 51 ). In the study, the authors hypothesized that increased levels of fucose may be released into the blood circulation of endometriosis patients associated with endometriosis lesion progression. These results are corroborated in our study, providing further evidence that dysregulation in cell surface proteins occurs in tissue endometriosis lesions. Lastly, we investigated alterations in proteomic profiles of endometriosis tissues by performing on-tissue enzymatic digest followed by MALDI MS imaging of tryptic peptides. Among the top co-localized features detected within the tissue microenvironment endometriosis lesions, a tryptic peptide corresponding to extracellular matrix (ECM) protein collagen alpha-1 (XIV) ( m/z 929.55) was observed to be highly localized to regions of endometriosis. Several studies have suggested that mechanisms for the spread of endometriosis may involve cell-cell and cell-ECM interactions between endometriosis cells and the peritoneal lining ( 55 , 56 ). In particular, collagen alpha-1(XIV) is a protein encoded by the COL14A1 gene and plays an adhesive role in binding together collagen types, as well as various components of the ECM components ( 57 , 58 ). The spread of endometriosis in the peritoneal cavity has been hypothesized to result from attachment of endometriotic cells to the surface of the peritoneum after they are released from the peritoneal fluid from other pre-existing lesions ( 55 ). After initial attachment to the peritoneum, endometriotic cells are thought to rapidly divide and invade the ECM which may potentially cause the formation of new lesions within the peritoneal cavity ( 55 ). As such, several ECM proteins of the peritoneum, such as collagen, have been investigated as potential binding targets for endometriotic cells. For example, in a study by Adachi et al. endometriotic cyst stromal cells were observed to have higher adhesive properties to ECM proteins collagen type I and collagen type IV, in comparison to normal eutopic endometrial stromal cells, thus providing evidence that cell-ECM adhesion may play an important role in endometriosis pathogenesis ( 55 ). Overall, the spatially-resolved molecular data we obtained of endometriosis lesions contributes to the growing evidence that suggests ECM proteins may be involved in the development and progression of endometriosis lesions within the peritoneal cavity.

In our study, several molecular changes within the endometriosis tissue microenvironment were identified and characterized. Although endometrial glands and stroma within endometriosis lesions are histologically similar to eutopic endometrium, molecular alterations at both the protein and N-glycan level were observed comparing the two tissue types using statistical analyses. In this study, an increased abundance of neutrophil defensin proteins detected in regions of endometriosis lesions in intact protein imaging experiments suggest the role of immune and inflammatory processes involved in endometriosis. We also observed differences in the N-glycan molecular profiles comparing eutopic endometrium and endometriosis lesions such as increased levels of N-glycan fucosylation and branching in regions of endometriosis lesions. These results suggest that dysregulation in cell-cell interactions and signaling processes such as through modification of cell surface proteins may be associated with endometriosis development. Molecular imaging of tryptic peptides revealed ECM proteins such as collagen localized within the endometriosis lesion tissue microenvironment, providing evidence that ECM proteins and cell-ECM adhesion interactions may play an important role in endometriosis pathogenesis. Overall, the results of our study provide an improved understanding of endometriosis disease mechanisms, which could aid in identifying potential diagnostic markers and improving treatment options for patients.

Endometriosis is gynecological condition in which endometrial-like cells abnormally grow outside the uterine cavity. The disease affects ~10% of women in reproductive age, with even higher incidence of ~30–50% among women presenting symptoms of pelvic pain and infertility ( 1 – 4 ). Despite its high prevalence and the significant impairment that endometriosis causes to patients’ quality of life, the biological mechanisms of endometriosis are poorly understood, and the disease is often misdiagnosed due to the lack of pre-operative diagnostic methods ( 5 ). Additionally, diagnosis and treatment of endometriosis can be complex as patients often present a range of clinical phenotypes, symptoms, and disease presentation ( 6 ). While many patients present severe and chronic abdominal pain and subfertility, for others the disease is asymptomatic ( 7 ). In terms of disease presentation, endometriosis lesion phenotypes are highly heterogeneous, presenting as superficial peritoneal endometriosis, ovarian endometriomas, deep infiltrating endometriosis, among others ( 6 , 8 ). Along with the heterogeneous nature of the disease, endometriosis patients often suffer from other comorbidities including autoimmune disorders, endocrine disorders, and increased risk of gynecological cancers ( 9 , 10 ), which contributes to the complexity in diagnosis. Currently, there are no proven pre-operative diagnostic markers of endometriosis ( 6 , 11 ), and exploratory laparoscopic surgery followed by pathology remains as the main diagnostic and treatment option for patients ( 12 ). Several studies have found that dysregulation of cellular and biochemical pathways is a significant factor in endometriosis, resulting in inherent differences between the endometrial tissue within the uterus of endometriosis patients compared to those without the condition. In addition, studies have shown that although histologically similar, molecular differences exist between ectopic endometrial tissue found outside the uterine cavity compared to the eutopic endometrial tissue in endometriosis patients ( 13 – 16 ). Thus, molecular characterization of endometrial and endometriosis tissues could help advance the understanding of endometriosis pathogenesis, leading to improved diagnosis and treatment for patients ( 17 ). A few studies have used mass spectrometry (MS) proteomics and lipidomics approaches to evaluate alterations in metabolites, lipids, and proteins from biospecimens obtained from endometriosis patients ( 18 – 21 ). Li et al. applied liquid chromatography (LC) MS to analyze lipid extracts of eutopic endometrial biopsies from patients with endometriosis and patients with unrelated infertility, resulting in the identification of five glycerophospholipids that allowed identification of endometriosis with 90.5% sensitivity and 75.0% specificity ( 22 ). In a study by Mear et al. , 543 differentially expressed proteins were identified in eutopic endometrium tissues of endometriosis patients when compared to endometrium tissues from women without endometriosis using LC-MS/MS. The focal adhesion and PI3K/AKT signaling pathways involved in regulation of fundamental cellular functions, such as cell proliferation, cell growth, and apoptosis were enriched in endometriosis tissues based on the upregulated proteins ( 18 ). More recent studies have leveraged direct MS and MS imaging techniques to identify molecules specific to endometriosis tissues ( 20 , 23 – 25 ). Adamyan et al. utilized tissue spray MS to directly analyze ectopic and eutopic endometrial tissues from endometriosis patients, revealing significant differences in 15 lipid species. Feider et al. utilized desorption electrospray ionization (DESI) MS imaging to explore alterations in metabolites and lipids between eutopic endometrial tissues and ectopic endometriosis tissue lesions obtained from patients. Differences in the relative abundances and spatial distributions of small metabolites, free fatty acids, and lipids, including phosphatidylserine 36:1 were detected within regions of endometriosis lesions ( 23 ). MALDI-MS imaging has been used to investigate molecular alterations associated with several gynecologic malignancies including endometrial cancers ( 26 – 29 ). In a study by Mittal et al. on endometrial cancer tissues, proteomic alterations of tryptic peptides such as higher intensity of annexin A2 peptide in regions of endometrial tumors were observed when compared to normal adjacent tissue ( 27 ). Alterations in N-glycosylation patterns such as higher abundance of oligomannose in regions of cancer tissue were also detected ( 29 ). This study showcases the increased interest in glycomics studies to investigate surface glycans and their role in cell signaling, cell-cell interaction, adhesion, growth, and metastasis in gynecologic diseases ( 30 ). Yet, characterization of proteins and glycans in endometriosis tissues remains poorly explored. Here, we investigate protein and N-glycan molecular signatures in eutopic endometrium tissue and ectopic endometriosis lesions from patient tissues using MALDI-MS imaging. Intact protein imaging revealed that neutrophil defensin proteins 1 and 2 were detected at higher abundances in regions of endometriosis lesions, suggesting that dysregulation in immune and inflammatory processes may be associated with endometriosis. Higher relative abundances of fucosylated N-glycans and N-glycans with increased levels of branching from biantennary to tri- and tetraantennary structures were also detected in endometriosis lesions, suggesting that cell-cell interactions, adhesion, and uncontrolled growth may be involved in endometriosis development. Lastly, we found a high degree of spatial localization of extracellular matrix proteins, such as collagen, to regions of endometriosis lesions, which may be related to the development of endometriosis lesions within the peritoneal cavity. Overall, this study reveals protein and N-glycan molecular alterations within endometriosis lesions tissue microenvironment that may provide new insights into disease development and progression.