Abstract Estrogens regulate eosinophilia in asthma and other inflammatory diseases. Further, peripheral eosinophilia and tumor-associated tissue eosinophilia (TATE) predicts a better response to immune checkpoint blockade (ICB) in breast cancer. However, how and if estrogens affect eosinophil biology in tumors and how this influences ICB efficacy has not been determined. Here, we report that estrogens decrease the number of peripheral eosinophils and TATE, and this contributes to increased tumor growth in validated murine models of breast cancer and melanoma. Moreover, estrogen signaling in healthy female mice also suppressed peripheral eosinophil prevalence by decreasing the proliferation and survival of maturing eosinophils. Inhibiting estrogen receptor (ER) signaling decreased tumor growth in an eosinophil-dependent manner. Further, the efficacy of ICBs was increased when administered in combination with anti-estrogens. These findings highlight the importance of ER signaling as a regulator of eosinophil biology and TATE and highlight the potential near-term clinical application of ER modulators to increase ICB efficacy in multiple tumor types. __________________________________________________________________ Anti-estrogen therapy increases the efficacy of immunotherapy by promoting the antitumor activity of eosinophils. INTRODUCTION Breast cancer is the most commonly diagnosed cancer in women in the United States with 287,850 new cases and 43,250 deaths from this disease reported in 2022 ([60]1). Recent advances in the treatment of breast cancer have had a marked effect on overall outcomes although most of the benefit is seen in patients with early/localized breast cancer ([61]2). There remains a substantial need for new approaches that can be used to treat patients whose tumors progress while on standard of care and/or to prevent and treat patients whose disease has a high risk of recurrence. Independent of the breast cancer subtype, the treatment regimens for most patients will involve surgical resection of the primary tumor (where indicated), regional radiation therapy, and systemic treatments (chemotherapy and/or targeted therapies) ([62]3). For estrogen receptor (ER)–positive breast cancer, endocrine therapies that target the production of estrogens (aromatase inhibitors or GnRH agonists) and/or target ER directly [e.g., the selective ER modulator (SERM), tamoxifen, or the selective estrogen receptor degraders (SERDs) fulvestrant (fulv) or elacestrant], are the cornerstone of interventions in patients with localized or metastatic disease ([63]3). CDK4/6 inhibitors used in combination with endocrine therapies are now considered standard of care for high-risk, early-stage and advanced ER-positive breast cancer ([64]4). Trastuzumab, an antibody that targets the receptor tyrosine kinase HER2 (ERBB2) has revolutionized the treatment of patients whose tumors harbor an ERBB2 gene amplification or in which the HER2 protein is overexpressed ([65]3). The ability of trastuzumab to inhibit the kinase activity of HER2 is an important contributor to its efficacy although underappreciated is the fact that this drug also facilitates antibody-dependent cell-mediated cytotoxicity and is thus an immunotherapy ([66]5). Beyond classical chemotherapy the treatment options, at least with respect to targeted therapies, for triple-negative breast cancer (TNBC) have been limited. Encouraging, however, is the recent approval of pembrolizumab (in combination with chemotherapy) for patients with locally recurrent unresectable or metastatic TNBC whose tumors express PD-L1 (combined positive score, ≥10) based on results from the phase 3 KEYNOTE-355 trial ([67]6). Supported by the results of the KEYNOTE-522 trial, pembrolizumab was later approved for all patients with high-risk early-stage TNBC ([68]7). The observation of benefit in these trials has encouraged investigation as to how to increase the number of patients who can be treated successfully with contemporary immunotherapies and has also begged a reexamination of the impact of existing medicines on immune cell repertoire and immune cell function in cells with a view to expanding immunotherapies to all breast cancer subtypes and to other cancer types. This puts into context the results of studies from our laboratory and others, which have demonstrated that inhibition of ER signaling in immune cells, notably myeloid cells, increases the efficacy of immunotherapies in disease relevant animal models of different cancers ([69]8, [70]9). These data suggest the possibility that endocrine therapies could be extended to all breast cancer subtypes and to other cancers that are classically defined as ER-negative in the anticipation that remodeling of the tumor immune microenvironment may increase the efficacy of immunotherapies. There is emerging evidence that tumor immune cell repertoire influences outcome/response to therapy in patients with breast cancer. Bioinformatic analysis of genomic data from a large cohort of ER^+ve and ER^−ve breast tumors indicated that increased numbers of eosinophils and monocytes were significantly associated with improved outcome especially in patients with ER^+ve breast cancer ([71]10). Moreover, it has been determined that increased tumor eosinophilia is associated with better response to immunotherapy in patients with TNBC ([72]11). Eosinophils are bone marrow–derived granulocytes with immunoregulatory functions in cancer ([73]12, [74]13). The presence of eosinophils in tumors [termed tumor-associated tissue eosinophilia (TATE)] has been correlated with better prognosis in colorectal, prostate, esophageal, and gastric cancers. In breast cancer, while the role of TATE is not well explored, increased blood eosinophil count is associated with improved cancer specific survival in patients and with better response to immunotherapy in patients with TNBC ([75]11, [76]13, [77]14). Estrogens regulate eosinophil accumulation in peripheral tissues but the role of this hormone in regulating TATE is unknown ([78]13). Here, we explore the hypothesis that estrogen signaling modulates eosinophil biology in tumors and that appropriate targeting of this signaling axis will enhance the anti-tumor activity of eosinophils. Using validated animal models of breast cancer, we have defined the roles of estrogens/ER in eosinophil biology in tumors and have determined that inhibiting ER signaling results in increased eosinophil number/activity in tumors and results in increased immune checkpoint blockade (ICB) efficacy. RESULTS Endocrine therapies increase the number of peripheral eosinophils in patients with breast cancer We performed a retrospective analysis of patient data to explore how ovarian suppression and endocrine therapies (GnRH agonists, and aromatase inhibitors and/or tamoxifen) affect peripheral immune cell repertoire in premenopausal women with ER-positive breast cancer ([79]Fig. 1A). These studies were focused on premenopausal women as they experience a very acute and sustained inhibition of ER signaling upon commencement of endocrine therapy (ET). Using preset inclusion and exclusion criteria (see Supplementary Methods), we identified a relevant cohort of patients with breast cancer and compared each patient’s prechemo, pre-ET “baseline” complete blood count (CBC) to blood cell counts collected 1 to 6 months after the initiation of ET (“early ET” or “T1”) and to blood cell counts collected 6 to 24 months after ET initiation (“late ET” or “T2”). Of the cell types examined, the most robust (and sustained) increase occurred in blood eosinophils at T1 and T2 compared to baseline ([80]Fig. 1, B and C). A very small increase in the number of basophils was also observed in patients after treatment with ET. To control for the effects of chemotherapy and surgery on eosinophil counts, we performed three subanalyses within this patient cohort. First, we compared each patient’s postchemotherapy CBC (collected at least 1 month after the completion of chemotherapy but before start of ET) to their “T2” CBC collected 6 to 24 months after the initiation of ET. The sample size (N value) in this subanalysis was much smaller than in the main study because most patients started ET concomitant with (or very shortly after) chemotherapy. Despite the greatly reduced sample size, we still found a statistically significant increase in eosinophil counts at T2 compared to postchemotherapy, suggesting that ET, not chemotherapy, is driving the increase in the number of eosinophils noted ([81]Fig. 1D). Second, there was no statistically significant change in eosinophil counts postchemotherapy compared to the prechemotherapy baseline ([82]Fig. 1E); however, a significant increase in eosinophils is observed at T2 compared to either pre- or postchemotherapy making it unlikely that chemotherapy alone is driving the differences noted in eosinophil numbers. These results were also extended to other immune cells (fig. S1I). Third, we identified 25 patients who underwent surgery before other treatments and compared their presurgery CBC to a postsurgery CBC collected at least 1 month after surgery but before any additional treatments. This subanalysis showed no independent effect of surgery on eosinophil count, suggesting that surgery is not an important confounding variable ([83]Fig. 1F). Considering these findings, and what is known of the importance of eosinophils in tumor biology in humans ([84]12), we proceeded to explore how estrogens regulate eosinophil number and activity and how this affects breast tumor pathobiology. Fig. 1. ET/Ovarian suppression increases the prevalence of blood eosinophils in patients with breast cancer. [85]Fig. 1. [86]Open in a new tab (A) Schematic depiction of the study design that was used to evaluate peripheral blood immune cell counts in patients with breast cancer that received endocrine therapy (ET). (B) Changes in the prevalence of various blood immune cells obtained from complete differential blood counts in patients taken at baseline (CBC collected before chemotherapy but no more than 3 months before diagnosis) and an early timepoint (T1) after the initiation of ET (n = 70). (C) Changes in the prevalence of various blood immune cells obtained from complete differential blood counts in patients taken at baseline and a late timepoint (T2) after the initiation of ET (n = 69). (D) Changes in the prevalence of blood eosinophils in patients taken at postchemo baseline compared to ET treatment at T2 (n = 41). (E) Changes in the prevalence of blood eosinophils in patients collected at the prechemo baseline compared to the postchemo timepoint (n = 41). (F) Changes in the prevalence of blood eosinophils in patients comparing pre- versus postbreast tumor resection surgery (n = 25). Data are shown as the mean ± SD. Significance was calculated using two-way ANOVA with repeated measures followed by Bonferroni correction of individual immune cell values normalized to variance. This was followed by paired Student’s t test for comparing baseline and treatment for individual immune cells. *P ≤ 0.05, **P ≤ 0.005, and ***P ≤ 0.0001. Estrogens decrease the number of eosinophils in tumors and in the blood in mice and this results in increased tumor growth Cancer cell–intrinsic estrogen signaling drives the proliferation and metastasis of ER^+ breast cancers, and endocrine therapies that target this signaling axis are among the most effective treatments for patients with this disease subtype ([87]15). However, estrogen signaling can also modulate the functionality of immune cells and other stromal cells that are present in the tumor microenvironment ([88]13). Recent studies from our group and others have highlighted the roles of estrogen signaling in tumor-associated macrophages and myeloid-derived suppressor cells, on tumor biology ([89]8, [90]9, [91]16, [92]17). These findings, and our observation that ET is associated with increased numbers of peripheral eosinophils in patients with breast cancer, encouraged us to explore the extent to which estrogens regulate eosinophil biology in validated animal models of breast cancer. To this end, we evaluated the impact of ER signaling on immune cell repertoire and activity in tumors derived from A7C11 and E0771 breast cancer cells propagated in syngeneic hosts. These specific ER-unresponsive cancer cells were chosen for this study as their biology is not directly influenced by estrogens in vitro (fig. S1, A and B) affording us the opportunity to assess the cancer cell–extrinsic effects of estrogens, SERMs and SERDs on tumor pathology and tumor immunobiology. Specifically, cancer cells were injected into the mammary fat pad (thoracic #3) of ovariectomized C57BL/6 mice (to accomplish estrogen deprivation) or ovariectomized mice supplemented with 17β-estradiol (E2, estrogen replete). It was determined that E2 promoted tumor growth in both the A7C11 and E0771 models when compared to placebo ([93]Fig. 2, A and C). The growth of these tumors was unaffected by E2 when propagated in the mammary fat pads of immune compromised nonobese diabetic severe combined immunodeficiency disease γ (NSG) mice ([94]Fig. 2, B and D). Notably, the stimulatory effect of estrogens on uterine wet weights, a measure of water imbibition and hypertrophy in response to E2, were similar in both C57BL/6 and NSG mice (fig. S1F). These data suggest that an intact immune system is required for the tumor-promoting effects of E2 in these models. Immune profiling of tumors was performed and a decrease in the frequencies of tumor-associated eosinophils (CD45^+SSC^highSiglecF^+CD11b^+Ly6G^lowMHCII^low) ([95]Fig. 2, E and F, and fig. S1C) was observed. An E2-dependent increase in the M2/M1 macrophage ratio was also observed (fig. S1, D and E) as we have reported previously in murine models of melanoma and lung cancer ([96]8). In these published studies, it was demonstrated that depletion of ERα expression in macrophages prevented estrogen-mediated M2 polarization, indirectly increasing T cell activity, resulting in reduced tumor growth ([97]8). However, in the breast tumor models (A7C11 and E0771), we have evaluated in this study depletion of ERα in macrophages was without effect (fig. S1, G and H), suggesting that the action of E2 on tumor pathobiology in the selected models is likely to be mediated by other immune cells in the tumor microenvironment. Whereas macrophages are a/the primary target of ER modulators in animal models of melanoma (B16F10 and BPD6) ([98]8), it was observed that E2 also decreased the number of intratumoral eosinophils in these models ([99]Fig. 2, G and H). Eosinophils reside primarily in the gut and have been shown to play a role in suppressing tumor growth in patients with colorectal cancer (CRC) ([100]18). We demonstrated that E2 decreased eosinophil number in an organoid model of CRC (AKPST tumor model) ([101]Fig. 2I). Further, we used immunohistochemical staining of A7C11 tumors to identify the spatial location of eosinophils in the tumor and observed that the invasion front (periphery) of the tumors have higher eosinophil presence as compared to the center. The presence of eosinophils was also lower in the tumors from E2-treated mice ([102]Fig. 1K and fig. S1J). The decreased tumor eosinophil number reflected a similar decrease in blood eosinophil number in tumor-bearing mice receiving E2 treatment ([103]Fig. 2J). Together, it appears that E2 decreases TATE in well-validated models of breast cancer and in models of other solid cancers and that this activity correlates with increased tumor growth; a finding of importance given the established antitumor activities of eosinophils ([104]19). Fig. 2. Estrogens decrease tumor and blood eosinophil numbers and increase tumor growth. [105]Fig. 2. [106]Open in a new tab (A) The growth of orthotopic A7C11 tumors (2 × 10^4 cells) in ovariectomized female C57BL/6 mice. Following ovariectomy, mice were supplemented with either placebo or E2 (2.72 mcg/ml in drinking water) throughout the study (n = 9). (B) The growth of orthotopic A7C11 tumors (2 × 10^4 cells) in ovariectomized, immune-compromised NSG mice, supplemented with placebo or E2 (n = 10). (C) E0771 (2 × 10^5 cells) tumor growth in syngeneic C57BL/6 mice ovariectomized and supplemented with placebo or E2 (n = 10). (D) E0771 (2 × 10^5 cells) tumor volume in immune-compromised NSG mice ovariectomized and supplemented with placebo or E2 (n = 10). (E to I) Tumor-infiltrated eosinophils (CD45^+Ly6G^lowSSC^hiMHCII^−CD11b^+SiglecF^+) in A7C11 (E), E0771 (F), B16F10 (G), BPD6 (H), and AKPS (I) tumor models. (J) Quantification of blood eosinophil counts in tumor-bearing mice in A7C11 and BPD6 tumor models (n = 6). (K) Representative immunohistochemistry images (20×) showing siglecF (red)–expressing cells (eosinophils) in the periphery and center of A7C11 tumor. Data are shown as the mean ± SD. Results in (A) to (H) and (J) are representative of two independent experiments. (A to D) Two-way ANOVA followed by Tukey’s multiple comparison test; (E to J) Unpaired t test. *P ≤ 0.05, **P ≤ 0.005, and ***P ≤ 0.0001. Eosinophils have antitumorigenic properties in murine models of breast cancer In a study of patients with TNBC who were eligible for immunotherapy, it has been reported that increased numbers of eosinophils in the blood predicted a favorable response to ICB ([107]11). To assess whether tumor eosinophil prevalence is also prognostic of outcome independent of treatment with immunotherapies, we assessed the expression of an eosinophil signature ([108]20) in the Molecular Taxonomy of Breast Cancer (Metabric) dataset ([109]21, [110]22) in all patients with breast cancer. This revealed that higher activity of an intratumoral eosinophil gene signature was associated with a survival advantage in patients with breast cancer ([111]Fig. 3A). Similar advantages in survival were also observed in patients whose tumors expressed higher eosinophil activity measured by the eosinophil peroxidase gene, epx, whose encoded protein is only expressed in eosinophils (fig. S3A). These data encouraged us to explore potential cause-and-effect relationships between estrogen deprivation, increased TATE, and decreased growth of tumors in syngeneic cancer models. To this end, we performed an eosinophil depletion study using an anti-SiglecF antibody ([112]Fig. 3B). Using this approach, we accomplished a quantitative depletion of eosinophils in both the blood and within tumors in mice ([113]Fig. 3, C and D). Depletion of eosinophils abrogated the protective effects of estrogen deprivation (as seen in ovariectomized/placebo mice) noted in the murine A7C11 tumor model ([114]Fig. 3, E and F). Since siglecF is also expressed in a subpopulation of neutrophils ([115]23), we further confirmed the specific involvement of eosinophils using the ΔdblGATA1 mouse model that lack eosinophils ([116]Fig. 3G) ([117]24). While estrogen deprivation reduced tumor growth compared to estrogen-treated mice in the littermate control ΔdblGATA1^−/+ and WT mice, the protective effects of estrogen deprivation is lost in the ΔdblGATA1^−/− mice ([118]Fig. 3, H and I). Together, these data highlight how estrogens, through their actions in eosinophils, affect primary tumor growth in established models of breast cancer. Fig. 3. Eosinophils have antitumorigenic properties in murine models of breast cancer. [119]Fig. 3. [120]Open in a new tab (A) Survival analysis using an eosinophil gene signature ([121]20) in patients with breast cancer (Metabric dataset). The patients were stratified into low and high expressing groups (of the eosinophil gene signature) according to the median. The P value was calculated using a log-rank test in R (n = 1992). (B) Schematic of the experimental plan for the eosinophil depletion study. (C and D) Quantification of blood and tumor eosinophils confirming depletion of eosinophils with anti-SiglecF antibody given once every 72 hours (n = 10). (E and F) A7C11 tumor growth and final day tumor volume in ovariectomized C57BL/6J mice treated with placebo or E2 and given either IgG or anti-SiglecF antibody (1 mg/kg dose) once every 72 hours (n = 10). (G) Quantification of A7C11 tumor eosinophils in ΔdblGATA1 mice ovariectomized and supplemented with placebo or E2. Eosinophils are absent in the ΔdblGATA1^−/− (n = 9) mice as compared to littermate controls ΔdblGATA1^−/+ mice (n = 7). (H and I) Tumor growth and final day tumor weight in ΔdblGATA1^−/− and littermate control ΔdblGATA1^−/+ mice ovariectomized and supplemented with placebo or E2 for 7 days before orthotopic injection of A7C11 (2 × 10^4 cells) (n = 7 to 9). Data are shown as the mean ± SD. (C, D, F, H, and I) One-way ANOVA followed by Tukey’s test. (E and H) Two-way ANOVA followed by Tukey’s test. *P ≤ 0.05, **P ≤ 0.005, and ***P ≤ 0.0001. E2 regulates both the biogenesis and the activity of eosinophils Eosinophils mature in the bone marrow before entering peripheral organs ([122]12). In this study, it was observed that both blood ([123]Fig. 2J) and bone marrow eosinophil numbers were decreased in tumor-bearing and in healthy mice upon E2 treatment, suggesting that this hormone regulates fundamental aspects of the biology of these immune cells ([124]Fig. 4, A and B). We thus explored the impact of E2 on the biogenesis and differentiation of bone marrow–derived eosinophils (BMEos) in vitro. The standard BMEos differentiation protocol was modified to enable a specific assessment of the role of E2 ([125]Fig. 4C). Aliquots of cells were taken on days 8, 11, and 14 during the differentiation assay to assess cell viability and the expression of markers that read on differentiation. BMEos are defined as CD45^+SiglecF^+SSC^hi and their frequency (among the live cells in culture) increased from ~16% on day 8 to ~90% on day 14 in presence or absence of E2 treatment (fig. S2B). However, E2 treatment decreases the total number of live progenitors and fully matured eosinophils throughout the time course of the differentiation ([126]Fig. 4D). It was determined that E2 treatment decreased both the viability of BMEos (fig. S2A), as well as the proliferation of live BMEos as measured by Ki67 staining ([127]Fig. 4E). The expression of interleukin-5Ra (IL-5Ra) and CCR3 receptors on BMEos was not affected by E2 (not shown). The results of this analysis suggested that E2 treatment reduced both the viability and proliferative potential of maturing eosinophils without affecting maturation per se, likely explaining the reduction in the numbers of eosinophils in circulation and within tumors that we have observed in E2-treated mice. In vitro, we observe that greater than 72 hours of E2 treatment is required for significant changes in eosinophil viability and proliferation. Fig. 4. E2 regulates the biogenesis and antitumor activity of eosinophils. [128]Fig. 4. [129]Open in a new tab (A) Blood eosinophils in healthy C57BL/6J mice ovariectomized and supplemented with placebo or E2 (n = 6). (B) Bone marrow eosinophils in A7C11 tumor-bearing and healthy C57BL/6J mice ovariectomized and supplemented with placebo or E2 (n = 6). (C) Schematic representation of experimental plan used to generate BMEos in vitro. (D and E) Quantification of live BMEos and Ki67^+ BMEos in the experiment at various timepoints (n = 3). (F) Schematic representation of the experimental plan used to study BMEos cytotoxicity activity in vitro. (G and H) Quantification of live, pre-apoptotic, apoptotic, and necrotic A7C11 cells after 5 hours of coculture with BMEos differentiated in presence or absence of E2 (n = 4). (I) Volcano plot of differentially expressed genes in BMEos (n = 3). (J and K) Negative enrichment plots for hallmark Myc target V2 and hallmark G[2]M checkpoint pathways in E2 versus Veh treatment comparison (n = 3). (L) Normalized counts from RNAseq for mRNAs that encode proteins in the cytotoxic granules of eosinophils; Epx, eosinophil peroxidase; Ear1, eosinophil-associated RNAse1; Prg2, major basic protein (n = 3). (M) qPCR analysis of BMEos (day 11) for the expression of Epx, Ear1, Prg2, Myc, Cdc6, and E2f1 genes (n = 3). (N) Quantification of BMEos viability (live/dead staining of eosinophils) and proliferation (Ki67^+ BMEos) at day 11 with the addition of an ER degrader fulv (100 nM) in addition to E2 (n = 4). Data are shown as the mean ± SD. Results in (A), (B), (D), (E), (H), (M), and (N) are representative of three independent experiments. (A and B) unpaired t test; (D, E, H, and M) two-way ANOVA followed by Tukey’s multiple comparison test; (N) one-way ANOVA followed by Tukey’s multiple comparison test. *P ≤ 0.05, **P ≤ 0.005, and ***P ≤ 0.0001. In addition to impacting biogenesis, we also assessed whether E2 influences eosinophil cytotoxicity. To this end, an equal number of live BMEos on day 14, differentiated in the presence or absence of E2, were cocultured with A7C11 cancer cells (1:15 ratio) for 5 hours. The number of live, pre-apoptotic, apoptotic, and necrotic cancer cells in the coculture were quantified using annexin V and Sytox staining. It was observed that the cytotoxic capability of E2-treated BMEos was reduced compared to vehicle-treated cells resulting in a significantly reduced number of necrotic cancer cells ([130]Fig. 4, F to H). To elucidate the mechanisms by which estrogens affect eosinophil biology, we performed RNA sequencing (RNAseq) analysis of BMEos differentiated in the presence of E2 or vehicle control. This approach to study ER-target gene expression in BMEos was validated by confirming the increased expression of Pgr and Tgm2 with E2 treatment, two canonical ER/E2 target genes ([131]25, [132]26) ([133]Fig. 4I and table S1). Pathway analysis indicated that E2 treatment down-regulated the expression of genes that are associated with proliferation (Myc) cell cycle (G[2]M checkpoint) and with eosinophil cytotoxicity [eosinophil peroxidase (Epx), eosinophil-associated ribonuclease (Ear1), and eosinophil major basic protein (Prg2)] ([134]Fig. 4, J to L, and fig. S2C). The expression of these genes was confirmed in a separate experiment using quantitative polymerase chain reaction (qPCR; [135]Fig. 4M). Paradoxically, E2 treatment was also found to regulate pathways associated with increased innate immune cell activation and inflammatory responses (fig. S2C). This was unexpected given our observation that BMEos differentiated in the presence of E2 have reduced cytotoxic capabilities. Thus, it appears that the effects of estrogens on different aspects of ER regulated biology are not the same, an observation that highlights a potential clinical utility of SERMs or SERDs that can differentially regulate these biologies. The negative effects of E2 on eosinophil viability and proliferation observed in vitro were reversed using fulvestrant a clinically important SERD ([136]Fig. 4N). In addition to reduced eosinophil biogenesis, adoptive transfer of equal numbers of CD45.1 BMEos into CD45.2 tumor-bearing mice indicated that E2 treatment is associated with decreased eosinophil recruitment to tumors (fig. S2, D to G). Recruitment of adoptively transferred eosinophils to other organs such as spleen was minimal and not affected by E2, indicating that E2 affects TATE specifically (fig. S2H). The decreased TATE observed may relate to changes in the expression of CCL11, CCL24 and CCL5 in the tumors; cytokines that have been shown to be important for eosinophil recruitment (fig. S2, I to K). Some of the changes observed in the intratumoral abundance of cytokines upon treatment with E2 might be a result of decreased TATE as eosinophils secrete many of these cytokines ([137]27). It is likely, however, that E2 action in macrophages and other immune cells may also influence the composition of the cytokine milieu ([138]28). Notwithstanding the complexities of estrogen signaling in the tumor microenvironment, some of the biology of this hormone results from its ability to decrease eosinophil production in the bone marrow ([139]Fig. 4B) and inhibit their activity in tumors. To assess the potential clinical significance of estrogen signaling in eosinophils, we generated two gene signatures from the E2 (a) up- and (b) down-regulated genes from our RNAseq analysis. The top 30 up- or down-regulated differentially expressed genes based on fold change with a significant adjusted P value (<0.01) were used to generate mouse E2 up- and down-regulated gene signatures. The human corollary of the mouse signature was used for subsequent studies (table S1). The signature derived from the E2 down-regulated genes (and not up-regulated) appears to have prognostic value in two different breast cancer datasets analyzed. These include the Metabric and the Sweden Cancerome Analysis Network–Breast datasets. Specifically, higher activity (increased expression of the E2 down-regulated genes) of this signature predicts better survival in patients with breast cancers when analyzed as a group and more specifically in patients with ER^+ breast tumors (fig. S2, L and M), suggesting that down-regulation/inhibition of ER signaling in eosinophils may be beneficial. Note that none of the top 30 down-regulated genes included in this analysis are part of the hallmark G[2]M checkpoint or hallmark Myc target V2 pathways, indicating that this signature is not just a reflection of decreased proliferation and viability associated with E2 treatment on eosinophils (table S1). However, we do note the limitation of using these signatures since some of the genes in the signatures we derived are expressed in cells other than eosinophils. Together, these results suggest that estrogens directly suppress eosinophil proliferation, survival, and cytotoxicity by affecting the expression of genes associated with these functions and that inhibition of ER signaling in eosinophils will likely increase the antitumorigenic effects of eosinophils. Inhibiting ERα signaling in eosinophils promotes their antitumorigenic activity The results of our in vitro studies revealed favorable activities of an ER-degrader (SERD) on eosinophil survival and proliferation ([140]Fig. 4N). Similarly, it has been demonstrated in animal models of airway inflammation that the actions of estrogens on tissue eosinophilia are mediated by ERα ([141]13). Thus, we explored the extent to which eosinophil-intrinsic estrogen signaling, mediated by ERα, regulates eosinophil function in tumors. To this end, we generated mice that lack ERα expression in mature eosinophils. Specifically, we crossed EpxCre mice ([142]29) with Esr1^flox/flox mice to generate EpxCre^+Esr1^f/f (EosERKO) mice and EpxCre^+Esr1^w/w (EosERWT) control mice ([143]Fig. 5A). Mature eosinophils derived from the bone marrow of EosERKO mice have substantially reduced ERα expression compared to those from the control mice ([144]Fig. 5B). The expression of the EpxCre driver in mice is highest in mature eosinophils with some expression noted in eosinophil progenitors. EPX is not expressed in common myeloid progenitors or in granulocyte-macrophage progenitors (GMPs) from which eosinophils arise ([145]29). Therefore, using the mouse model created, we were able to probe the specific role(s) of ERα signaling on the biology of mature eosinophils. The growth of the syngeneic A7C11 and E0771 cell–derived breast tumors were significantly impaired in EosERKO mice when compared to EosERWT control mice, in placebo-treated, estrogen-deprived mice in both the breast cancer models ([146]Fig. 5, C to F). These effects were not restricted to models of breast cancer as we observed that the growth of BPD6 cell–derived melanoma tumors was also compromised in EosERKO mice compared to EosERWT control mice (fig. S3, C and D). The impact of estrogens on tumor growth in the EosERKO mice was also attenuated when compared to control mice. Depletion of ERα in mature eosinophils did not reverse the negative effects of estrogens on the accumulation of eosinophils in tumors ([147]Fig. 5G and fig. S3, B and E). This is perhaps expected as ERα is only depleted in mature eosinophils and not in progenitor cells in EosERKO mice. However, depletion of ERα in mature eosinophils reversed the inhibitory effect of E2 on the expression of tumor EPX, a marker of eosinophil activity ([148]Fig. 5H). Thus, it appears that estrogens, acting on ERα, regulate eosinophil activity but that the effects of this hormone on TATE per se occur in an indirect manner or are mediated by the receptor expressed in progenitors in the bone marrow where we could not accomplish a genetic knockout. Despite the decrease in TATE observed with E2 in EosERKO mice, the expression of tumor eosinophil peroxidase, a marker of eosinophil activity, was not suppressed by E2 treatment in EosERKO mice ([149]Fig. 5H). These results agree with our in vitro cytotoxicity assays and suggest that estrogens, acting through ERα, suppress the antitumorigenic properties of mature eosinophils. Fig. 5. Inhibiting ERα signaling in eosinophils promotes their antitumorigenic activity. [150]Fig. 5. [151]Open in a new tab (A) Schematic showing breeding strategy to generate EosERWT and EosERKO mice. (B) Western blot of ERα expression in eosinophils derived from the BM of either EosERWT or EosERKO mice (n = 3). (C and D) Tumor growth and final day tumor volume in EosERKO and EosERWT mice ovariectomized and supplemented with placebo or E2 for 7 days before orthotopic injection of A7C11 (2 × 10^4 cells) (n = 9). (E and F) Tumor growth and final day tumor volume in EosERKO and EosERWT mice ovariectomized and supplemented with placebo or E2 for 7 days before orthotopic injection of E0771 (2 × 10^5 cells) (n = 9). (G and H) Quantification of tumor eosinophils (n = 7) and tumor eosinophil peroxidase levels (n = 3) in the A7C11 tumors from EosERWT and EosERKO mice. Data are shown as the mean ± SD. Results in (C) and (D) are representative of two independent experiments. (C and E) Two-way ANOVA followed by Tukey’s multiple comparison test. (G and H) One-way ANOVA followed by Tukey’s multiple comparison test. *P ≤ 0.05, **P ≤ 0.005, and ***P ≤ 0.0001. It has been established that eosinophils increase the number of intratumoral cytotoxic CD8^+ T cells to inhibit tumor growth ([152]30). In large part this has been ascribed to increases in cytokines that facilitate T cell recruitment to tumors ([153]30). We observed an increase in the number of effector CD8^+ T cells [CD69^+CD8^+ and interferon-γ^+ (IFN-γ^+) CD8^+ cells] in placebo-treated EosERKO mice. This change was accompanied by a decrease in protumorigenic M2 macrophages in E2-treated EosERKO mice (fig. S3, F to H). Pharmacological inhibition of ER signaling increases TATE to suppress tumor growth Having established that estrogens suppress TATE to increase tumor growth, we next evaluated the effects of systemically inhibiting estrogen signaling, using two different classes of clinically relevant ER modulators. To this end, we undertook a comparative analysis of the activity of fulvestrant, a clinically approved injectable SERD and lasofoxifene (laso), an oral SERM that is in late-stage clinical development as a treatment for metastatic breast cancer ([154]31, [155]32). Using the A7C11 model, we demonstrated that laso inhibited E2-induced tumor growth. Fulv also decreased tumor growth but in general its effects were less robust and not sustained (fig. S4A). The lack of efficacy of fulv may relate to its mechanism of action (an ER degrader) or to its poor pharmacokinetic properties ([156]33). However, we did confirm adequate drug exposure in these studies by demonstrating that fulv (and laso) inhibited E2 stimulated increases in uterine wet weight (fig. S4, B and D). Considering its superior efficacy, its excellent pharmaceutical properties, and that it has been shown to be well tolerated in multiple clinical studies [as a treatment for osteoporosis and the climacteric symptoms associated with menopause ([157]34)], we undertook a more comprehensive evaluation of laso as a regulator of TATE/tumor growth. The efficacy of laso given alone or in combination with E2 was evaluated in two breast cancer models (A7C11 and E0771). In both models, laso reduced tumor growth below that observed in placebo-treated mice and reversed E2 induced tumor growth ([158]Fig. 6, A and B). Lasofoxifene cotreatment also reversed the inhibitory effects of E2 on eosinophil numbers in the A7C11 model with a trend in this direction noted in studies performed in the E0771 model ([159]Fig. 6C and fig. S4C). The E2-dependent suppression of tumor eosinophil activity, as measured by tumor EPX levels, was rescued by laso treatment ([160]Fig. 6D). We next evaluated the extent to which the protective effect of laso on tumor growth was dependent on tumor-associated eosinophils. This was accomplished by evaluating the efficacy of laso in tumor models in which eosinophils had been depleted using an anti-siglecF antibody ([161]Fig. 6E). Quantification of blood and tumor eosinophils confirmed that anti-SiglecF antibody treatment efficiently depleted eosinophils in mice (fig. S4, E and F). We noted that depletion of eosinophils completely abrogated the antitumor activity of laso ([162]Fig. 6, F and G). Fig. 6. Pharmacological inhibition of ER signaling increases TATE to suppress tumor growth. [163]Fig. 6. [164]Open in a new tab (A) A7C11 tumor volume (after injecting 2 × 10^4 cells) in ovariectomized C57BL/6 mice supplemented with E2 or placebo and given subcutaneous laso (10 mg/kg per day) or vehicle daily starting from day 2 after tumor injection. Significant increase in tumor growth with E2 treatment was reversed with the addition of laso (n = 10). (B) E0771 tumor volume (after injecting 2 × 10^5 cells) in ovariectomized C57BL/6 mice supplemented with E2 or placebo and given subcutaneous laso (10 mg/kg per day) or vehicle daily starting from day 2 after tumor injection. Significant increase in tumor growth with E2 treatment was reversed with the addition of laso (n = 10). (C) Quantification of tumor eosinophils. The decrease in eosinophils with estrogen treatment was reversed by lasofoxifene treatment. An increase in tumor eosinophils when laso was given alone was also noted (n = 8). (D) Quantification of tumor eosinophil peroxidase (EPX) levels in A7C11 tumors, indicating rescue of eosinophil activity in the tumors upon lasofoxifene treatment (n = 5). (E) Schematic of experimental methods for eosinophil depletion study and laso treatment. (F and G) A7C11 tumor growth and final day tumor volume in ovariectomized C57BL/6 mice supplemented with placebo or E2 and given either intraperitoneal IgG or anti-SiglecF antibody (1 mg/kg per dose) once every 72 hours and subcutaneous laso (10 mg/kg per day) or vehicle given daily from day 2 after tumor injections (n = 8). Data are shown as the mean ± SD. Results in (A) and (C) are representative of two independent experiments. (A, B, and F) Two-way ANOVA followed by Tukey’s multiple comparison test; (C, D, and G) One-way ANOVA followed by Tukey’s multiple comparison test. *P ≤ 0.05, **P ≤ 0.005, and ***P ≤ 0.0001. Lasofoxifene enhances the antitumor efficacy of immune checkpoint inhibitors Increased TATE has been shown to enhance the efficacy of ICB in patients with TNBC ([165]11). Thus, we explored the extent to which manipulating eosinophil function using anti-estrogens influenced ICB efficacy in E2-treated A7C11 cell–derived tumors, which express PD-L1 (fig. S5A). Notwithstanding the positive benefits of laso in these models, the potential near-term clinical utility of an approved ER modulator, like fulv, encouraged us to extend our studies and compare the SERM (laso) with the SERD (fulv). In this model, ICB treatment [anti–PD1 (10 mg/kg) and anti-CTLA (4 to 5 mg/kg)] alone demonstrated modest antitumor activity ([166]Fig. 7, A to C). With respect to tumor growth, it was determined that both laso and fulv increased ICB efficacy ([167]Fig. 7, A to C). The ability of both fulv and laso to inhibit E2-dependent increases in uterine weights was used to confirm adequate drug exposure in these animals (fig. S5B). As expected, E2 decreased TATE in this model, which was reversed upon treatment with both laso and fulv but not ICB given alone ([168]Fig. 7D). It was also observed that E2 decreased the number of CD8^+ T cells and that this was reversed by treatment with either laso or ICB or fulv. Of particular importance was the observation that the number of CD8^+ T cells was substantially increased by cotreatment with ICB and laso or by cotreatment with ICB and fulv when compared to no drug treatment control ([169]Fig. 7E). While E2-induced tumor growth was modestly suppressed by ICB or fulv when given alone, we observed a robust suppression of tumor growth when anti-estrogens were combined with ICB. We also observed a robust increase in intratumoral CD8 T cells, an observation that correlates with better response to ICB treatment. Thus, it appears that fulv and laso, both clinically relevant drugs, have similar activity with respect to their ability to increase ICB efficacy. Thus, differences in properties of these drugs (i.e., tolerability and toxicity) are likely to be key differentiators of these drugs when used in combination with existing ICBs. Fig. 7. Lasofoxifene enhances the antitumor efficacy of immune checkpoint inhibitors. [170]Fig. 7. [171]Open in a new tab (A and B) Individual and average A7C11 tumor growth curves in C57BL/6 mice that were ovariectomized and given placebo or E2 treatment were further treated with subcutaneous laso (10/mg/kg per day) or subcutaneous vehicle treatment starting from day 2 after tumor cell injection, intraperitoneal ICB (anti-CTLA, 4 to 5 mg/kg; anti-PD1, 10 mg/kg) or IgG given once every 72 hours starting from day 2, intramuscular fulv (25 mg/kg) or vehicle given once weekly starting from day 2 after tumor injections or the combinations as indicated (n = 8). (C) Final day tumor volume (A7C11) in treatment groups as indicated (n = 8). (D and E) Quantification of tumor eosinophils (D) and CD8 T cells (E) (n = 8). Data are expressed as the mean ± SD. Results in (A) to (E) are representative of two independent experiments. Two-way ANOVA followed by Tukey’s multiple comparison test. *P ≤ 0.05, **P ≤ 0.005, and ***P ≤ 0.0001. DISCUSSION It has been reported that the increased number of eosinophils within tumors, (i.e., TATE) is associated with favorable outcomes and a better response to therapy across multiple cancers ([172]11, [173]30, [174]35). Notable are data in patients with CRC in which elevated tumor eosinophilia is strongly associated with decreased risk for cancer death and with better prognosis ([175]36, [176]37). Likewise, increased TATE predicts a favorable prognosis/reduced cancer-associated death in patients with esophageal, gastric, oral, melanoma, and liver cancers ([177]12, [178]13). Less studied is the impact of TATE on the pathobiology of breast cancers although there are studies that suggest that increased numbers of peripheral eosinophils are associated with better survival and/or response to chemotherapy ([179]38). A recent study reported that in patients with TNBC, an increase in both systemic and intratumoral eosinophils correlated with better response to ICB treatment, a finding that was corroborated in studies performed in animals ([180]11). In this regard, our observation that patients diagnosed with breast cancer that receive ET have increased prevalence of blood eosinophils indicates a positive activity of endocrine therapies that heretofore has not been appreciated. While the observational data suggesting that TATE is correlated with good prognosis in many different cancers are strong ([181]12), little has been done to establish cause-and-effect relationships and even less has been done to develop approaches to increase TATE in tumors. Using mouse models of breast cancer and melanoma, we have demonstrated that depletion of eosinophils results in increased tumor growth and that inhibition of ER signaling in eosinophils decreases tumor growth. In this study, and in other studies reported by our group and others, it has been shown that cancer cell extrinsic actions of estrogens promote the growth of ER-negative tumors or tumors in which ER is not engaged in the regulation of proliferative responses ([182]8). Whereas the growth-promoting effects of estrogens can be attributed to cancer cell–intrinsic actions of ER in some tumor models, we have now established that ER directly regulates eosinophil biology and that this affects tumor pathology. The specific importance of eosinophils as mediators of estrogen action was confirmed by showing that specific genetic depletion of ER within eosinophils resulted in increased tumor growth. However, we were limited in these studies to using the Epx-cre model where ER depletion is accomplished only in mature eosinophils or eosinophil committed progenitors that express Epx gene. What remains to be determined is whether the noted effects of estrogens in regulating eosinophil biogenesis are due to direct effects of estrogens on ER in hematopoietic progenitor cells such as GMPs and multipotent progenitor or if it occurs in an indirect manner. The direct cytotoxic actions of eosinophils toward cancer cells likely explain some of their antitumor activity ([183]18). However, recent studies have demonstrated that these cells are multifunctional and can regulate other immune cells. Eosinophils were shown to secrete cytokines (such as CXCL9/10), which facilitate the recruitment of CD8^+ T cells into tumors ([184]11, [185]30). While others have indicated that eosinophils can modulate the CD8^+ T cell population in the tumors ([186]30), we demonstrated that depletion of eosinophils with concomitant treatment with E2 did not significantly affect intratumoral CD8^+ T cell number in the breast tumor models. However, we observed that suppressing ERα signaling in eosinophils enhanced the infiltration of activated CD8^+ T cells into tumors. Therefore, we cannot completely rule out the involvement of T cells in the antitumor effects of eosinophils in these models. In addition to T cells, activation of bone marrow eosinophils was shown to be required for survival and retention of plasma cells postimmunization ([187]39, [188]40). Eosinophils also regulate humoral immunity via their impact on B cell homeostasis and proliferation upon activation ([189]41). Likewise, cytokines such as IFN-γ, IL-5, and eotaxins secreted by other cells such as T-cells, macrophages and fibroblast can influence eosinophil recruitment and activation ([190]28). Whereas increased TATE is generally positive in tumors the antitumor activity of these cells is likely counterbalanced somewhat by their ability to produce IL-13 and IL-4, which facilitate M2 polarization of tumor-associated macrophages ([191]13, [192]42). We and others have established that estrogens increase M2 polarization of macrophages and that in some models (i.e., melanoma) this is due to direct actions of estrogens on ER in myeloid cells. In our study, we have also found that E2 increases the number of M2 -polarized macrophages in tumors and that this was significantly attenuated in mice that lack eosinophil-intrinsic ERα signaling. Inhibiting estrogen signaling in eosinophils also increased tumor eosinophil peroxidase levels indicating increased eosinophil activity. Further studies are required to distinguish the direct versus indirect (via eosinophils) influence of estrogens on various immune cells in regulating tumor immunity. Regardless, we conclude that inhibiting ERα signaling in eosinophils enhances their antitumor activity by increasing eosinophil cytotoxicity, as well as indirectly affecting CD8^+ T cell infiltration/activity and macrophage polarization. Eosinophils mature within the bone marrow and then passage through the blood stream to accumulate in peripheral organs. While these cells have a very short half-life in the blood, within tissues, they can survive for extended periods where they exert physiological or pathological functions ([193]43). The decrease in TATE, we observed in murine models of breast, melanoma, and CRC cancers in estrogen-treated animals was mirrored by a decreased number of eosinophils in the blood and within the bone marrow. Unexpectedly, we also observed that E2 decreased peripheral eosinophil counts in healthy ovariectomized mice, indicating that, absent the presence of a tumor, this hormone regulates the biogenesis of eosinophils in the bone marrow. While interesting, the biological significance of this of this observation remains to be determined. It is known that eosinophil counts are suppressed during pregnancy to protect fetus from excessive inflammation. Blood eosinophil counts spike postpartum and this increase correlates with lower E2 levels ([194]44). In studies of eosinophil production/differentiation from bone marrow progenitors in vitro, we observed that E2 decreases the number of mature eosinophils by decreasing both cell proliferation and survival. Further, transcriptomic analysis of in vitro–differentiated BMEos revealed that E2 suppressed the expression of genes that are associated with the G[2]M checkpoint and with Myc signaling. E2 also promoted the expression of the progesterone receptor (Pgr) although the physiological consequence of this activity remains to be determined. The observation in vitro that differentiation, proliferation, and thus the total number of maturing BMEos can be enhanced by inhibiting ER signaling using fulv further confirms the role of E2-ER signaling in regulating the biology of BMEos. The mechanisms by which ER/E2 affects eosinopoiesis remain to be determined. In addition to decreasing the absolute numbers of eosinophils, our data also suggest that E2 decreases their recruitment to tumors. We observed that E2 decreased the intratumoral expression of cytokines (i.e., CCL5, CCL11, and CCL22) in mice that others have shown to be important for eosinophil recruitment to tumors and thus negatively affect their survival ([195]28). While the source of some of these cytokines could be eosinophils themselves, there are other cells such as fibroblasts and macrophages in the TME that can influence eosinophil recruitment ([196]13, [197]45, [198]46). Regardless, our observation is in line with that of others, which demonstrated that administration of E2 to ovariectomized mice suppresses peritoneal eosinophilia in a classic model of acute inflammation ([199]47). Together, we believe that E2 decreases the biogenesis and maturation of eosinophils in the bone marrow, decreases their migration to tumor, and decreases their cancer cell directed cytotoxic activities ([200]Fig. 8). These observations in murine models are further corroborated by the clinical observation that ET is associated with an increase in peripheral eosinophil numbers in young women with breast cancer ([201]Fig. 1). Fig. 8. Model describing how modulating ER influences eosinophil biology. [202]Fig. 8. [203]Open in a new tab E2 suppresses peripheral eosinophil prevalence and TATE in tumor models. E2 also decreases eosinophil cytotoxicity by suppressing the expression of cytotoxic granular contents. These deleterious effects of estrogens can be countered using anti-estrogens such as laso and fulv. Targeting the E2/ERα signaling axis is an effective therapeutic strategy for ER^+ breast cancer at all stages ([204]15). In late-stage disease, where patients have progressed on one or more aromatase inhibitors, the SERD fulv is currently the standard of care but will likely be replaced by elacestrant for those patients whose tumors express specific ESR1 mutations ([205]48–[206]50). Studies from our group and others have also demonstrated the utility of the SERM laso as an approach to inhibit the growth of tumors that express ERα mutations ([207]51, [208]52). The antitumor activity of laso in therapy-resistant breast cancer that expresses the most common ERα mutations was shown in murine models of breast cancer ([209]51). These findings have led to clinical studies to evaluate laso in patients whose tumors express ESR1 mutations ([210]31, [211]53). Our studies suggest that laso could have more widespread clinical utility. Notably, we found that laso is superior to fulv in suppressing E2-induced tumor growth in breast tumor models where the cancer cells are not directly responsive to E2. Laso enhanced TATE, an activity we demonstrated that was required for the antitumor efficacy of this drug. Further, laso increased the efficacy of anti-PD1 and anti-CTLA4 in the tumor models examined. Laso was superior to fulv in reversing E2-suppressed TATE when evaluated in ICB combinations. As yet, we do not know whether this represents a fundamental difference between SERMs and SERDs or whether it relates to a specific pharmacological property of fulv. Regardless, these findings establish the importance of ER signaling in eosinophils on tumor pathobiology and highlight the potential utility of combining ER modulators (SERMs/SERDs) with ICBs in patients with breast cancer and other cancers where the presence of TATE is beneficial. MATERIALS AND METHODS Mice C57BL/6 and ΔdblGATA1 female mice were purchased from Jackson Laboratories (Bar Harbor, ME). Age-matched mice (7 to 12 weeks old) were used for all the studies. EpxCre mice were obtained from Mayo Clinic ([212]29) and were bred to Esr1^f/f (a gift from K. Korach, NIEHS) mice to generate Esr1^f/fEpxCre and littermate controls EpxCre and Esr1^f/f mice. ER^flox/flox/LysMCre mice were generated as previously described ([213]8). Mice were housed in a secure animal facility on a 12-hour light:dark cycle at temperature around 25°C and 70% humidity. Throughout the study, the mice have access to ad libitum food and water. NSG (NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ) mice were purchased from the Division of Laboratory Animal Resources (Duke University). The NSG animals were fed with GL3 diet and were kept in pathogen-free conditions. All animal experiments were performed according to guidelines from Duke Institutional Animal Care and Use Committee (IACUC) under the approved animal use protocol number A2442312. Tumor models and cells Mouse breast cancer line E0771 was acquired from the lab of M. Dewhirst (Duke University) and A7C11 cells were provided by J. Conejo-Garcia (Duke University). E0771 cells were maintained in RPMI-1640 media (Sigma-Aldrich) supplemented with 8% fetal bovine serum (FBS), 0.1 mM nonessential amino acid (NEAA), and 1 mM NaPy and A7C11 were maintained in RPMI-1640 media (Sigma-Aldrich) supplemented with 2-mercaptoethanol (Gibco 21985-023; 1:1000), 8% FBS, 0.1 mM NEAA, and 1 mM sodium pyruvate (NaPy). B16F10 were purchased from American Type Culture Collection (Manassas, VA). BPD6 were donated by B. Hanks, Duke University. The cells were split three times/week at a 1:10 ratio when confluent and were kept in 37°C incubator at 5% CO[2]. For breast tumor models, A7C11 (2 × 10^4 cells) and E0771 (2 × 10^5 cells) cells were subcutaneously injected into the right upper third thoracic mammary fat pad of mice. Tumors were measured thrice weekly with the help of an electronic caliper. Tumor volume was calculated by the formula volume = length × (width × width)/2. For tumor growth rate studies, mice were euthanized wither before or when the tumors reach 2000 mm^3, as specified by Duke IACUC. For colon cancer studies, ApcD/D;KrasG12D;Trp53D/D;Smad4D/D;tdTomato (AKPST) mouse colon cancer organoids were obtained as a gift from Ö. Yilmaz (Massachusetts Institute of Technology). AKPST organoids were orthotopically engrafted into the colonic submucosa of recipient syngeneic C57BL/6 mice via colonoscopy-guided injection (700 organoids/injection), as previously described ([214]54, [215]55). Tumors were monitored with optical colonoscopy and mice were euthanized after 1 month. Ovariectomy and estrogen supplementation Seven-week-old C57BL/6 female mice were subjected to ovariectomy 8 days before tumor inoculation. Using an inhalation chamber (2% isoflurane), mice were first anesthetized and subsequently maintained in half the dose of isoflurane (1%) via nose cone throughout the surgical process. Before surgery, mice were administered with carprofen (5 mg/kg) subcutaneously. The area on the back of the mice below the ribs was shaved with an electronic razor, and the skin was sterilized by rubbing with betadine and alcohol three times alternating. This was followed by a horizontal incision through the skin above the ovary fat pad and then a vertical incision through the abdominal muscle wall. The ovary was externalized and removed using cauterizing scissors. The fat pad was replaced, muscle walls were realigned and sutured (one to two stitches), and one drop of bupivacaine (0.25%) was added on top of the incision site. The skin was realigned, and a wound clip is placed on the incision site. This was repeated for the other ovary. Subsequently, the mice were removed from anesthesia and kept in a clean cage and monitored until it regained consciousness. The mice were monitored for recovery for 10 days. Drug treatments The treatments with E2 were started on day 2 after ovariectomy. E2 was dissolved in drinking water at 2.72 mcg/ml, and the placebo group received vehicle (ethyl alcohol) at acidified water (2 ml/liter). Lasofoxifene (10 mg/kg) was dissolved in 10% dimethyl sulfoxide (DMSO)/40% polyethylene glycol (molecular weight, 400)/50% sterile water and injected subcutaneously once daily from day 2 after tumor injections until the end of the study. Fulvestrant was dissolved in corn oil and given once a week intramuscularly at 25 mg/kg. The ICB treatments were a combination of anti-PD1 at 10 mg/kg (catalog no. BE0146; Bio X Cell, Lebanon, NH) and anti-CTLA4 at 5 mg/kg (catalog no. BE0164, Bio X Cell, Lebanon, NH) given intraperitoneally once every 3 days starting at day 2 after tumor injections. Immunoglobulin G (IgG; catalog no. BE0089; Bio X Cell, Lebanon, NH) was used as the control for ICB treatments. Eosinophil depletion studies For the purposes of eosinophil depletion, C57BL/6 mice were injected with mouse SiglecF antibody (clone no. 238047; R&D catalog no. MAB17061) at 1 mg/kg per dose or monoclonal rat IgG[2A] isotype control (clone no. 54447; R&D catalog no. MAB006) reconstituted at 0.05 mg/ml in sterile phosphate-buffered saline (PBS). Antibody injections were performed 24 hours before tumor injections and once every 72 hours after the first dose. The efficacy of eosinophil depletion was analyzed at the end of the experiment by collecting cardiac blood and tumor-infiltrating immune cells and performing flow cytometry for eosinophil subpopulation. In vitro bone marrow eosinophil differentiation Bone marrow–derived cells were aseptically collected from 8- to 10-week-old female C57BL/6 mice by crushing the leg bones in PBS containing 1% FBS and 2 mM EDTA. The solution was filtered through a 40-μm strainer to remove bone pieces. Cells were subjected to ACK buffer to lyse the red blood cells. After centrifugation, the cells were washed once in PBS containing 0.1% bovine serum albumin (BSA). The bone marrow cells were cultured at 10^6/ml in media containing RPMI 1640 (Invitrogen) with 20% charcoal-stripped FBS, penicillin (100 IU/ml), and streptomycin (10 μg/ml; Invitrogen), 2 mM glutamine (Invitrogen), 25 mM Hepes, 0.1 mM NEAA, 1 mM sodium pyruvate (Gibco), and 50 μM β-mercaptoethanol (Gibco 21985-023). Cells were supplemented with either E2 (1nM) or DMSO throughout the differentiation process and with stem cell factor (SCF; 100 ng/ml; PeproTech catalog no. 250-03), and FLT3-Ligand (FLT3-L; 100 ng/ml; PeproTech catalog no. 250-31L) from day 0 to day 4. On day 4, the media containing SCF and FLT3-L was replaced with media containing recombinant mouse interleukin-5 (rmIL-5; 10 ng/ml; PeproTech catalog no. 215-15) only. On day 8, the cells were moved to new flasks and maintained in fresh media supplemented with rmIL-5. Every other day, from this point forward, the media was replaced with fresh media containing rmIL-5, and the concentration of the cells was adjusted each time to 10^6/ml. Cells were enumerated at day 0 and on days indicated thereafter using a hemocytometer. qPCR of eosinophils RNA was isolated using Aurum total RNA mini kit (catalog no. 7326820, Bio-Rad, Hercules, CA) followed by cDNA synthesis using iScript cDNA synthesis kit (catalog no. 170-7691, Bio-Rad, Hercules, CA) as per the manufacturer’s instructions. Quantitative amplification was performed with Sybr Green (catalog no. 1725124, Bio-Rad, Hercules, CA) using CFX-384 Real Time PCR detection system. The 36B4 gene expression was used as an internal control, and the expression levels of each target gene were normalized to 36B4, then to vehicle (DMSO)–treated controls. The corresponding primers were as follows: mEar1: Fw, TAGGGGCCTTAGCCACACTC; and Rv, CTGCTATGCAGTCTCGAAGGA. mPrg2: Fw, GTCTCAGGTCAGGATGTGACA; and Rv, GCGGACTGGATTCCGAAGTT. mEpx: Fw, TAGGGGCCTTAGCCACACTC; and Rv, CTGCTATGCAGTCTCGAAGGA. m36B4: Fw, AGATTCGGGATATGCTGTTGGC; and Rv, TCGGGTCCTAGACCAGTGTTC. mMyc: Fw, ATGCCCCTCAACGTGAACTTC; and Rv, CGCAACATAGGATGGAGAGCA. mCdc6: Fw, CCGTGTGTGGACGTAAAACTT; and Rv, GGGGAGTGTTGCACAGGTT. mE2f1: Fw, CAGAACCTATGGCTAGGGAGT; and Rv, GATCCAGCCTCCGTTTCACC. Proliferation assays A7C11 or E0771 cells were plated in RPMI media (without phenol red) supplemented with 8% charcoal-stripped FBS (with further addition of 2-mercaptoethanol for A7C11). Cells were plated at a concentration of 1000 cells per well on a 96-well plate and incubated for 2 days in 200 μl of media. After 2 days, 50 μl of media was removed and replenished with 50 μl of fresh media containing 4× concentration of vehicle (DMSO), E2, or E2 + fulvestrant, thus maintaining the original concentration. Cells were collected every 24 hours by discarding the media from the plates and then freezing plates at −80°C. Frozen plates were thawed at room temperature, and 100 μl of water was added to each plate for 1 hour at 37°C to mediate cell lysis. DNA content in each well was determined by the addition of 25 μl of Hoechst dye (Hoechst 33258 catalog no. H3570 Thermo Fisher Scientific, Waltham, MA) in 10 ml of TNE buffer for 45 min at room temperature, and the fluorescence was read at excitation of 346 nm and emission at 460 nm using a microplate reader. Flow cytometry staining Single-cell suspensions (10^6 cells in 50 μl) were incubated with live/dead fixable dead cell stain in PBS for 10 min at 4°C. Cells were spun down at 2000 RPM and were incubated with anti-CD16/32 (1:100; catalog no. 14-0161-85, Thermo Fisher Scientific, Waltham, MA) in flow buffer (made by dissolving 10 g of BSA in 1 liter of PBS) for 15 mins. Following this, cells were stained with antibody cocktails in BV buffer (catalog no. 566349, Thermo Fisher Scientific, Waltham, MA) for 30 min at 4°C. For intracellular staining, cells were fixed and permeabilized using transcription factor staining kit buffer (catalog no. 00-5523-00, Thermo Fisher Scientific, Waltham, MA), followed by intracellular staining with the desired antibodies for 30 min at 4°C. Multicolor flow cytometry was performed using the BD Fortessa 16 color analyzer. Results were analyzed using FlowJo_V10 software (FlowJo, LLC). The antibodies used are as follows: live/dead staining dye (Invitrogen [216]L34964, 1:200), BV650 anti-mouse CD45 (30-F11, BioLegend 103139, 1:800), APC anti-mouse SiglecF (S17007L, BioLegend 155508, 1:100), PerCpCy5.5 anti-mouse CD3 (17A2, BD 560572, 1:50), AF700 anti-mouse CD11b (M1/70, BioLegend 101222, 1:100), BV711 anti-mouse MHC II (M5/114, BD 563414, 1:100), PE anti-mouse CD193 (J073E5, BioLegend 144506, 1:100), PE-Cy7 anti-mouse CD125 (DIH37, BioLegend 153408, 1:100), BUV395 anti-mouse Ki67 (B56, BD 564071, 1:100), AF647 anti human/mouse granzyme B (GB11, BioLegend 515406, 1:50), AF700 anti mouse/human CD44 (IM7, BioLegend 103026, 1:100), APC-Cy7 anti-mouse CD69 (H1.2F3, BioLegend 104526, 1:100), BV650 anti-mouse CD8 (53-6.7, BioLegend 100742, 1:100), BV785 anti-mouse CD4 (RM4-5, BioLegend 100552, 1:100), BV711 anti-mouse IFN-g (XMG1.2, BioLegend 505835, 1:50), PE anti-mouse FOXP3 (FJK-16s, eBioscience 12-5573-82, 1:100), BV510 anti-mouse PD1 (29F1A12, BioLegend 135241, 1:100), APC anti-mouse CD25 (PC61.5, eBioscience 17-0251-82, 1:100), PE anti-mouse CD11b (M1/70, BioLegend 101208, 1:50), PE-TR anti-mouse CD19 (6D5, Thermo Fisher Scientific, RM7717, 1:100), AF488 anti-mouse CD206 (C06C2, BioLegend 141710, 1:100), BUV496 anti-mouse CD24 (M1/69, BD 564664, 1:100), PerCPCy5.5 anti-mouse CD64 (X54-5/7.1, BioLegend 139308, 1:100), PerCPCy5.5 anti-mouse PD-L1 (10F.9G2, BioLegend 124333, 1:100), APC anti-mouse F4/80 (BM8, BioLegend 123116, 1:100), APC-Cy7 anti-mouse CD11c (HL3, BD 561241, 1:50), PE-CY7 anti-mouse B220 (RA3-6B2, BioLegend 103222, 1:100), PE-CY7 anti-mouse MHCII (M5/114-15.2, eBioscience 25-5321-82, 1: 600), BV711 anti-mouse Ly6 C (HK1.4, BioLegend 128037, 1:100), and BV786 anti-mouse Ly6G (1A8, BioLegend 127645, 1:100). Immunoblotting Cells were washed three times with 2 ml of ice-cold PBS and lysed with 0.15 ml of phospho-RIPA lysis buffer [tris-Cl (pH 7.5), 50 mM; NaCl, 150 mM; NP-40, 1%; sodium deoxycholate, 0.5%; SDS, 0.05%; EDTA, 5 mM; sodium fluoride, 50 mM; sodium pyrophosphate, 15 mM; β-glycerophosphate, 10 mM; and sodium orthovanadate, 1 mM] with protease inhibitor cocktail (Millipore-Sigma, P-8340). Equal amounts of protein per sample/lane were denatured and resolved by SDS–polyacrylamide gel electrophoresis. Proteins were transferred to immunoblot polyvinylidene difluoride membranes (Bio-Rad catalog no. 1620177). Primary antibodies used were anti–ER-α (1:1000, clone no. 6F11 Leica in 5% milk/PBST) and anti–β-actin (Cell Signaling, cat no. 8457, dilution 1:20,000). The secondary antibodies used were horseradish peroxidase (HRP)–conjugated goat anti-mouse IgG (Bio-Rad catalog no. 1706516; dilution 1:10,000) and HRP-conjugated goat anti-rabbit (catalog no. 1706515; Bio-Rad; dilution 1:15000). Bulk RNAseq Day 11 BMEos differentiated in the presence of vehicle or E2 were flow sorted for live cells, followed by RNA isolation using the RNeasy micro kit from Qiagen. Following RNA QC, library preparation was performed using stranded KAPA hyperPrep kit followed by high-throughput sequencing to obtain 50-bp paired-end reads using Illumina NovaSeq 6000. RNAseq reads were trimmed by Trim Galore ([217]www.bioinformatics.babraham.ac.uk/projects/trim_galore/) and trimmed reads were aligned to mouse genome (GRCm38) using STAR (version 2.7.7a) ([218]56). Quantification of gene expression was obtained by Subread featureCounts (version 2.0.1) ([219]57). Differential expression analysis using gene counts was performed by DESeq2 ([220]58). Data wrangling were performed and plots were generated using tidyverse ([221]www.tidyverse.org/), Enhanced Volcano ([222]https://github.com/kevinblighe/EnhancedVolcano), and base R packages. Pathway enrichment analysis was performed using ClusterProfiler ([223]59) and gene set enrichment analyses ([224]60). EPX ELISA Tumor eosinophil peroxidase (EPX) levels were quantified using mouse the EPX enzyme-linked immunosorbent assay (ELISA) kit from Biomatik corporation (catalog no. 501502404/[225]EKN45030 96 Tests), following manufacturer’s protocol. An equal amount of protein isolated from tumor homogenates (prepared according to manufacturer’s recommendations) was used along with protein standards provided by the kit in a sandwich enzyme immunoassay to quantify EPX levels. Optical density was measured at 450 nm using a TECAN microplate reader. Immunohistochemistry Tumors were fixed in 10% formalin overnight and then transferred to 30% sucrose before being embedded and frozen in optimal cutting temperature. Tumor tissue sections were cut at a thickness of 10 μm using a cryostat at −20°C. Tissue sections were washed in 1× tris-buffered saline (TBS) twice and then blocked using immunofluorescence blocking buffer (Cell Signaling, 12411) in a humidified chamber for 1 hour at room temperature. Tumor sections were stained for Siglec-F expression (Invitrogen, catalog no. 14-1702-82) or for rat IgG2a κ isotype control (Invitrogen, catalog no. 14-4321-81) at a 1:25 dilution overnight in a humidified chamber at 4°C. Sections were washed three times in 1× TBST and then stained for anti-rat IgG Alexa Fluor 568 (Invitrogen, catalog no. A11077) at 1:300 dilution for 1 hour at room temperature in a humidified chamber the following day. Sections were then again washed three times in 1× TBST, and then immersed in 1× TBS twice for 5 min. Sections were then counterstained with 4′,6-diamidino-2-phenylindole (Sigma-Aldrich, catalog no. D9542; 1:1000 dilution in 1× PBS) for 7 min in a humidified chamber at room temperature. Sections were washed three times in 1× TBST, and then three times in 1× TBS. All sections were imaged using a Leica SP5 inverted confocal microscope. Measurement of tumor cytokine levels Protein from tumor homogenates were subjected to cytokine ELISA array (Ray Biotech) analysis for 96 different cytokines. The membranes were probed according to the manufacturer’s protocol and quantified by scanning densitometry. Analysis of human correlates Bulk RNAseq was performed on murine BMEos differentiated in the presence of E2 or placebo. The top 30 genes that were up-regulated in E2 group and the top 30 genes that were down-regulated with E2 based on fold change and P value were identified. The human analogs of these genes were used to create an “E2-up-regulated” gene signature and an “E2 down-regulated” gene signature. Survival analyses were performed using the “survival package” analysis with R. Patient populations were partitioned using median expression values and compared using the log-rank test. Retrospective chart review study The SlicerDicer tool, a feature of the Epic Electronic Medical Record, was used to identify a group of 135 patients within the Duke University Health System (approved protocol no. pro00114047) who satisfy the following criteria. Inclusion criteria are as follows: females with a breast cancer diagnosis between October 2018 and October 2023, age younger than 45 at time of diagnosis, at least two CBCs with differential within 12 months of diagnosis, and either (A) concurrent treatment with a GnRH agonist and an aromatase inhibitor or (B) concurrent treatment with a GnRH agonist and tamoxifen. Exclusion criteria are as follows: diagnosis of distant metastases (stage IV disease), diagnosis of menopause or menopause symptoms before breast cancer diagnosis, previous breast cancer diagnoses, and any history of treatment with CDK4/6 inhibitors. For each patient, three CBCs with differential were identified: a “baseline” CBC collected before chemotherapy but no more than 3 months before diagnosis, an “early ET” CBC collected 1 to 6 months after the end of chemotherapy and the initiation of both GnRH agonist and AI/tamoxifen, and a “late ET” CBC collected 6 to 24 months after the same timepoint referenced above. Six measurements were recorded from each CBC: total white blood cell count, neutrophil count, lymphocyte count, monocyte count, eosinophil count, and basophil count. Fifty-two of the original 135 patients were excluded from the study due to incomplete information in the chart or notes, indicating noncompliance with ET. Study data were securely recorded and managed using Research Electronic Data Capture (REDCap) tools hosted at Duke University ([226]61, [227]62). Statistics Description of samples (mice, cell number, and biological replicates) is provided in the figure legends. Data were plotted and analyzed using Graph Pad Prism 8.0 software. Statistical significance was calculated by either two-tailed Student’s t test, one-way analysis of variance (ANOVA), or two-way ANOVA as indicated in the legends. For both one-way and two-way ANOVA, posttest analysis was performed using Tukey’s multiple correction. Acknowledgments