The immune landscape of uterine fibroids as determined by mass cytometry.

Portions of uterine Fib and paired Myo were collected from patients (n=10) who were not on immune-altering medications or hormonal treatments for at least 3 months prior to surgery at Harbor-UCLA Medical Center. Informed consent was obtained from all patients prior to surgery and the protocol was approved by the IRB at The Lundquist Institute (18CR-31752–01R). The mean age of the patients was 47.4 ± 2.5 years and none were post-menopausal, had prior embolization, or had any autoimmune diseases. Patient demographics and indications for surgery are listed in Table 1 . The Fib samples used ranged in size from 3~5 cm in diameter and were intramural (FIGO stage 3 and 4). Myometrium samples were obtained distant from the fibroid. Pathologic evaluation confirmed non-degenerating fibroids and myometrium. Some specimens had adenomyosis as indicated in Table 1 . Seven of the 10 tumors analyzed were confirmed to have the MED12 mutation in exon 2 as determined by Sanger sequencing. Immediately following surgery, tissue samples were collected and digested, for 20h in collagenase and DMEM and strained through a 40 μm filter. The media was then centrifuged at 800 × g for 5 min in order to separate cells from the supernatant. The cells were washed with 1 mL PBS and then counted and aliquoted for further analysis. After cell isolation, an aliquot of 3 × 10^6 of both Fib and myometrial cells were used for staining. The cells were allowed to incubate with cisplatin and Fc block before a panel of 24 extracellular metal-conjugated antibodies was added ( Supplemental Table 1 ). Cells were then fixed, permeabilized and stained with a mixture of 5 intracellular metal-conjugated antibodies. All 29 antibodies, 24 extracellular and 5 intracellular, were provided by the Flow Cytometry Core Lab at UCLA. The antibodies were validated and optimal concentration was determined by the UCLA Flow Cytometry Core facility. Iridium was then added and the cells were allowed to incubate overnight before being washed with PBS and Milli-Q water. Samples were then analyzed on the HELIOS-II machine (Flow Cytometry Core Lab at UCLA). Data visualization was manually performed using the OMIQ software (Dotmatics, Boston, Massachusetts, USA). All data pre-processing, including de-barcoding and bead normalization, was performed at the Flow Cytometry Core Lab at UCLA. Bead normalization was performed by re-suspending the cells in ddH2O and EQ Four Element Calibration Beads (Fluidigm, San Francisco, California, USA). Data was then normalized using OMIQ software. Mass cytometry data was gated following the method of Iyer et al.[ 12 ], which selects for both live cell populations and singlets. CD45+ cells were then gated to allow for identification of only immune cells. Gating was performed manually for each immune cell population and was kept consistent across paired samples. This allowed obtaining the most consistent data, as the distribution of Fib and Myo cells was different across patients but remained the same within the same patient. Files were then separated and concatenated depending on whether they were Fib or Myo cells. UMAPs were then generated in the OMIQ software. Paired Fib and Myo (n=5) were fixed in 4% paraformaldehyde in PBS, and then transferred to a PBS solution containing 30% sucrose. Tissue samples were embedded in paraffin and 5um sections were cut and allowed to adhere to poly-L-lysine coated slides. Slides were rehydrated by a sequential ethanol wash and incubated in HistoVT One (Nacalai Tesque, Kyoto, Japan) in a microwave for 20 minutes for antigen retrieval. Tissue sections were blocked with PBS-5% with primary antibody mouse anti-CD117 (1:200 dilution, Bio-Rad, Hercules, California, USA) or mouse anti-Tryptase (1:200 dilution, Zeta Corporation, Sierra Madre, California, USA). Tissue sections were then washed 3 times with 1x PBS and incubated with secondary antibody goat anti-mouse (1:500 dilution, Novus, Centennial, Colorado, USA)for 1 hour. Antigens were visualized using Impact DAB Perxoidase (Vector Laboratories, Newark, California, USA) and counterstained with hematoxylin and eosin. Immunofluorescent sections were incubated with Alexa Fluor 594 anti-CD3 (1:200 dilution, BioLegend, San Diego, California, USA), PerCp/Cyanine 5.5 anti-CD4 (1:200 dilution, BioLegendSan Diego, California, USA), FITC anti-CD56 (1:200 dilution, BioLegendSan Diego, California, USA), AlexaFluor 594 anti-iNOS (1:200 dilution, Novus, Centennial, Colorado, USA), or FITC anti-CD68 (1:200 dilution, BioLegendSan Diego, California, USA). Antibodies for immunofluorescence were diluted in Signal Enhancer Hikari (Nacalai Tesque, Kyoto, Japan). Immunofluorescent slides were incubated overnight at 4°C in the dark, then washed 3 times with 1x PBS. Sections were stained with DAPI (1:1000 dilution) and cover slipped using ProLong Gold Antifade Mountant (Invitrogen, Waltham, Massachusetts, USA) and placed in the dark. For both IHC and IFC, sections of tonsil tissues were stained with the same antibodies used for analysis as a positive control. For the IHC samples, a tissue section was incubated with antibody diluent instead of the primary antibody which served as a negative control. For the IFC study, a negative control consisted of incubation of section with a secondary antibody conjugated to either FITC, AF-594, or PerCp-Cy5.5 eliminating the primary antibody in order to assess background fluorescence. Five images were captured at 20x magnification for both Myo and Fib tissue sections (n=5) using the Leica Photonic Microscope (Leica Microsystems, Deerfield, Illinois). Images were chosen at random focusing on capturing unique quadrants of the section (i.e. top-left, bottom-right). Both IHC and IFC images were processed and analyzed using the HALO V.3.4.2986 Image Analysis Platform (Indica Labs, Albuquerque, New Mexico). IHC images were processed using the Multiplex IHC HALO module and fluorescent images were processed using the Highplex FL module. Co-localization of fluorophores was calculated by the program. Total protein from six pairs of matched Fib and Myo was isolated and concentrations were normalized. Samples were analyzed by the UCLA Immune Assessment Core where a Luminex Human 38-plex cytokine/chemokine assay was performed. Quality control was performed by the UCLA Immune Assessment Core prior to analysis. An ‘a priori’ sample size determination was made based upon previously defined molecular endpoints [ 7 ]. Based upon a minimal detectable difference of 40% in the mean between experimental groups and 35% expected standard deviation, a sample size of 10 would be needed in order to achieve a power = 80% and α= 0.05. All statistical data presented throughout the text was processed by the software PRISM v.10.0.3 (Graph-Pad, San Diego, CA) and is presented as mean ± SEM. Data in Figures 1 – 4 was not normally distributed; therefore, a non-parametric test was used for the analysis. Pairwise comparisons were conducted using the Wilcoxon matched-pairs signed rank test. Statistical significance was established at P<0.05.

The first objective was to visualize the cell populations using UMAP [ 13 ] in order to identify distinct immune characteristics of each tissue compartment. Immune cells were gated and then UMAPs were generated for Myo ( Figure 1A ) and Fib ( Figure 1B ). All clusters shown were generated based upon the gating done previously. Each color on the plot represents a different immune cell. These plots showed clear differences in the composition of dendritic cells (cDCs), T cells, NK cells, and macrophages in Fib as compared with Myo. Statistical analysis indicated a significant increase in the number of total macrophages ( Figure 1F ) and their M2 subtype ( Figure 1H ) in Fibs (p=0.0020 and p=0.0273 respectively). There was no significant change in the number of M1 macrophages ( Figure 1G ) or the M1:M2 ratio, however, the increase in M1 macrophages in fib nearly reached statistical significance (p=0.053). There was a significant increase in the number of conventional dendritic cells, or cDCs (p=0.0098)( Figure 1J ), and NK cells (p=0.0156) in Fibs as compared with Myo ( Figure 1I ). Both CD3 ( Figure 1C ) and CD4 ( Figure 1D ) T cell numbers were significantly lower in Fibs as compared to Myo (p=0.0039 and p=0.0273 respectively). Mass cytometry did not show a significant difference in total mast cell number ( Figure 1K ), Tregs, or CD8 T cells ( Figure 1E ) number between Myo and Fib. Figures for Tregs, M1:M2 ratio, and other non-significant immune cells are shown in Supplementary Figure 1 . There were no significant effects of race/ethnicity, MED12 mutation status of fib, presence of adenomyosis or menstrual cycle phase on immune cell populations although the sample size was limited to detect the impact of these variables. In order to validate the CytoF data immunofluorescent staining was performed for several of the cell types we found to be significantly different by mass cytometry. In Figures 2A – C , DAPI was used for nuclear staining and each image was taken at 20x magnification with scale bars denoting 90μm. Macrophage staining is shown in Figure 2A ; CD68 (a pan-macrophage marker) [ 14 ] fluoresced green, iNOS (a M1 macrophage marker)[ 15 ] fluoresced cyan, and CD206 (a M2 macrophage marker) [ 16 ], fluoresced red. Positive staining for general macrophages is represented by the white arrow, fluorescing solely green; M2 macrophage cell staining is indicated by the yellow arrow, fluorescing orange; M1 macrophage cell staining is indicated by the green arrow. Image analysis indicated there was a significantly higher number of total macrophages and M2 macrophages in Fibs as compared with Myo (p<0.05) with no significant difference in M1 macrophage number ( Figure 2A ). Figure 2B , shows staining for NK cells; Antibodies for CD3 (red) and CD56 (green) antibodies were to identify NK and NKT cells [ 17 ]. Positive staining for NK cells is indicated by the white arrow, and shows solely green CD56 fluorescence, while positive staining for NKT cells, indicated by the yellow arrow, represents both CD3 and CD56 overlap. Image analysis using the HALO software ( Figure 2B ), indicated significantly (p<0.05) higher number of NK cells in Fib as compared with Myo, with no significant differences in NKT cells. The number of CD3 was significantly lower (p<0.05) in Fib. Figure 2C shows IFC staining for CD3 (red), CD4 helper (cyan), and CD8 cytotoxic (green) T cells. Positive staining for CD4 T cells is indicated by the green arrow and positive CD8 staining by the yellow arrow. Image analysis revealed a significant decrease (p<0.05) in the number of both CD3 and CD4 T cells in Fibs when compared to Myo with no significant change in the number of CD8 T cells ( Figure 2C ). In order to determine any differences in mast cell activation , immunohistological staining was performed on paired Myo and Fibs for both CD117, a marker for resting mast cells, and Tryptase, a marker for activated mast cells. The results are shown in Figure 3 , with positive staining being denoted by the red arrow pointing to a brown spot and nuclear staining denoted by the blue spots. There were no significant differences in resting mast cell number between Fibs and Myo, which supports the findings from mass cytometry ( Figure 1I ). However, Fibs had significantly more tryptase-positive mast cells as compared to Myo (p=0.015). Differences in cytokine expression was assessed through a Luminex cytokine assay of (n=5) pairs of matched Fib and Myo protein isolates. Analysis revealed significant differences (p<0.05) in the cytokine profile between Fib and Myo, which is shown in Figure 4 . Fibs had significantly higher levels of IFNA2, IL-1α, and PDGF-AA and significantly lower levels of IL-1RA and M-CSF when compared to Myo. There was no significant difference in IL-4, IL-5, IL-6, IL-8, IL-10, IL-12 (p40), IL-15, IL-17E, IL-18, IL-22, MCPI, CXCL9, MIP-1b, PDGF-AB, TNF-B, or VEGF-A.

In conclusion, the results presented show significant differences in the immune cell populations of Fibroid and its matched Myometrium as determined by CytoF and IFC/IHC analysis which was also associated with altered cytokine/chemokine profile. Our data indicated a significant increase in the number of total and M2 macrophages, NK cells, and dendritic cells with a significant decrease in the number of CD3/CD4 T cells in Fibs. IHC also confirmed the lack of differences in the number of total resting mast cells in Fibs, but a significant increase in tryptase-positive mast cells which is indicative of their activation. This study provides a foundation for future studies exploring the potential role of immune cells in Fib pathogenesis and targeting them for novel therapies.

The results of this study indicate marked differences in the immune landscape of fibroids as compared with adjacent myometrium. Mass cytometry using 29 different antibodies representing different immune cell markers indicated significant differences in the number of naive and M2 macrophages, NK cells, cDCs, with all being expanded in Fibs as compared with Myo. In the case of T cells, Fibs showed lower number of CD4/CD8 cells as compared with matched Myo. CytoF results were confirmed by immunofluorescent staining, which indicated a significant increase (p<0.05) in the number of total macrophages, M2 macrophages, and NK cells and a significant decrease in the number of CD3/CD4 T cells in Fibs as compared with Myo. IHC revealed a significant increase in the number of tryptase-positive mast cells indicative of activation of mast cells, but not resting mast cell number in Fibs confirming the CytoF results. The multiplex cytokine assay also revealed distinct differences in the cytokine profiles of Fib and Myo, with increased levels of IFNA2, Il-1α, and PDGF-AA and decreased levels of IL-1RA in Fib as compared with Myo. Of the immune cell population analyzed macrophages were found to be the most significantly altered immune cells in Fib, with tumors having significantly more macrophages when compared to matched Myo. It is known that inflammatory process is key in the development of fibrosis [ 18 ] and that macrophages are the key inflammatory mediators in the fibrotic response [ 19 ] and other smooth muscle cancers [ 20 ]. In other fibrotic tissues macrophages have also been found to be key regulators of CD4 T cells, CD3 T cells, and NK cells [ 21 ], which we also found to be significantly altered in Fibs. Little is known about the cytokines and chemokines produced by macrophages in Fib however, it has been noted that Fib have increased MCP-1 [ 22 ] and GM-CSF [ 23 ] levels, which are often associated with macrophage recruitment [ 24 , 25 ]. Furthermore, our multiplex cytokine assay found significant upregulation of IL-1α, a cytokine associated with macrophage-induced angiogenesis [ 26 ], inflammation [ 27 ] and differentiation [ 28 ], and significant downregulation of IL-1RA, a IL-1α receptor antagonist [ 29 ], in Fibs. M2 polarized macrophages were also significantly higher in Fibs. Despite being labeled as anti-inflammatory, M2 macrophages have been found to play a vital role in the development of fibrosis through their interaction with the ECM [ 30 , 31 ]. M2 macrophages are integral in the transformation of stromal tissue during the wound healing process [ 32 ], however, this pathway is often found to be dysregulated in the development of fibrotic tumors [ 33 ]. In fact, PDGF-AA, a cytokine released by M2 macrophages [ 34 ] that is integral in the wound healing process [ 35 , 36 ], was upregulated in Fibs. Additionally, TGF-B [ 37 , 38 ] activin-a [ 39 ], and TNF-a [ 40 ], which all influence ECM composition [ 41 ], are upregulated in Fibs. TGF-B has also been found to facilitate a positive feedback loop that drives further M2 macrophage polarization [ 42 ]. Despite these findings, there was no significant change in the M1:M2 ratio between Fib and Myo which is typically found in other tumors [ 43 , 44 ]. Also, M-CSF, a cytokine that drives both general and M2 macrophage differentiation [ 45 , 46 ], was significantly decreased in Fibs which would be contrary to expectation. Our results indicated that both CD3 and CD4 T cell numbers were significantly lower in Fib. This decrease in T cell number may provide a mechanism for immune evasion by the Fib as in many malignancies [ 47 ]. Evading immune surveillance is a key hallmark of a growing tumor and is necessary for its development [ 48 ]. In fact, in previous fib studies, CD4 T cell count was linked to prognostic outcomes [ 49 ]. In recent studies, we reported marked dysregulation of the tryptophan catabolic pathway with a significant upregulation of TDO2 and kynurenine [ 50 , 51 ]. The metabolites of the tryptophan catabolism are known to be immunosuppressive [ 52 , 53 ] and could account for immune evasion in Fibs. The ability and degree of activation of T cells is largely dependent on its interactions with the innate immune system. As previously stated, macrophages are key regulators of T cells in other fibrotic tissues [ 21 ] and possibly contribute to an immunosuppressive environment as in other cancers through their interactions with CD4 cells [ 54 ]. Dendritic cells, which were also upregulated in Fibs, also alter the immune microenvironment in cervical cancer through its release of IDO1 [ 55 ], an enzyme that catalyzes tryptophan degradation [ 56 ], and also has been found to be inhibit CD8 and Treg cell activity [ 55 ]. Furthermore, the inhibition of DC maturation in cervical cancer results in the expansion of CD4 T cells [ 57 ]. CD4 T cells are key regulators of the cytokines and chemokines present in the immune environment [ 58 ]. Surprisingly, our cytokine profile found that IFNA2, a cytokine often associated with T cell activation and decreased tumor growth [ 59 , 60 ], was significantly increased in Fibs. In a report comparing the peripheral immune cells between healthy donors and patients with Fibs, patients with Fibs had a significantly higher number of CD4 T cells [ 7 ] in contrast to our findings which are based on the tumor. Previous studies indicated that the features of circulating peripheral T cells do not necessarily reflect the T cells found in tissues [ 61 ]. No significant difference was found in the number of T regulatory (Treg) cells in Fib and Myo tissue. Previous studies have shown the important role Treg cell infiltration plays in the modulation of both tumor development [ 62 , 63 ] and inflammation [ 64 ]. T regulatory cells play an important role in maintaining homeostasis of the immune environment [ 65 ], partly through the inhibition of cytotoxic CD8 T cells [ 66 ]. CD8 T cells were also not significantly different between Fibs and Myo, which was confirmed by both CytoF analysis and IFC, lending further support that T regulatory cells may not be acting as an immunosuppressant in this environment. The balance between Treg and Th17 T cells which our Cyto F panel did not cover has also been noted to be highly important in the maintenance of inflammation [ 67 , 68 ]. In this study, resting mast cell numbers were not altered in fib however activated mast cell numbers were increased similar to previous studies reporting an increased number of tryptase and leptin-positive mast cells [ 4 , 69 ]. Resting mast cells are key inflammatory cells that promote tumor angiogenesis and invasion in sarcomas [ 70 ] and other smooth muscle tumors [ 71 ], and, when activated, can release proteases that degrade the ECM [ 72 ]. Additionally, many of the chemokines released by mast cells are known to regulate the recruitment and activation of various immune cells [ 73 – 76 ]. It has also been suggested that mast cells may function as antigen-presenting cells for CD4 T cells [ 77 ]. This type of interaction could account for our findings showing activation of mast cells and decreased T cell number in Fibs. Tranilast, an anti-allergy drug used in Japan and South Korea that is known to target mast cells [ 78 ], has had positive therapeutic effects both in vivo and in vitro in Fib [ 79 , 80 , 81 ], lending evidence to the possible significance of mast cells in Fib pathogenesis.

Fibroids (Fib) are benign tumors of the smooth muscle cells of the uterus affecting up to approximately 70% of women in their lifetime. These tumors may cause heavy bleeding, pain and infertility. These symptoms can eventually lead to a full or partial hysterectomy [ 1 ]. It is known that immune evasion is imperative for a tumor to grow [ 2 ], however, little is known about the immune response to a growing fibroid. Further understanding of this immune response can provide insight into possible therapeutic options targeting immune cells. A hallmark of Fib development is excessive inflammation [ 3 ], which suggests dysregulation of inflammatory immune cells. Previous studies have found that Fib growth are associated with an increase in the number of activated mast cells [ 4 ], which could possibly be a contributing factor to increased inflammation. Fibs as compared with healthy myometrium (Myo) have increased macrophage invasion and increased MCP-1 expression which is a chemoattractant for macrophages [ 5 ]. Macrophages, and their many different phenotypes, are integral to both the inflammatory and tissue-repair pathways [ 6 ]. Previous studies have also found dysregulation among T cell populations in Fibs. There was significantly higher number of Treg, CD4, and CD8 T cells in peripheral blood of patients with Fib when compared to healthy controls [ 7 ]. Differences in NK cell populations in the peripheral blood was also reported, with patients with Fibs having reduced NK cell counts when compared to healthy controls [ 8 ]. Patients with uterine Fibs also have significantly more gene polymorphisms in cytokines associated with Th1/Th2 T cell populations [ 8 ]. Polymorphisms associated with Fib risk were also noted in the gene PCDC6 [ 9 ], which facilitates programmed cell death of T cells. While the findings reported in these studies are significant, they also come with limitations. Immune cells were obtained from the peripheral blood in many of these studies, rather than the tissue, which may not be representative of the immune environment within the tumor. Moreover, there are no studies to date comparing immune cell populations in Fib tissue with its matched Myo from the same patient, which would provide much greater insight into the key differences between tumor and healthy tissue. In order to accomplish this, cytometry by time of flight, or CytoF, was employed to characterize the immune cell populations in the Fib with its matched Myo. CytoF is an application of mass cytometry which uses antibodies conjugated with a metal isotope in order to measure the abundance of both intracellular and extracellular cell markers [ 10 ]. There are many advantages to using mass cytometry when compared to the more conventional flow cytometry, such as the ability to analyze many more antibodies at once, little to no background, and much less overlap between antibodies [ 11 ].

Supplemental Figure 1. Pairwise statistical analysis comparing immune cell populations obtained from mass cytometry. (A-H) represent immune cell populations that did not reach significance. Note that in (E) the y-axis is listed as M1 divided by M2 rather than a percentage of CD45. Supplemental Table 1. Table representing the list of antibodies used for CytoF analysis in Figure 1 . The column designated Label represents the metal isotope each antibody was conjugated to.