Myeloid activation clears ascites and reveals IL27-dependent regression of metastatic ovarian cancer

Ovarian cancer (OvCa) is the most lethal gynecological cancer in the United States and the fifth leading cause of cancer-related deaths in women due in large part to metastases (Siegel et al., 2023). Unlike other cancers that metastasize through circulation, OvCa cells mostly disseminate directly into the peritoneal cavity and preferentially seed in the omentum prior to metastasizing to other peritoneal organs (Ma, 2020). Because it is difficult to detect at early stages, OvCa is often diagnosed at later stages with metastatic lesions present throughout the peritoneal cavity (stages III and IV) (Lengyel, 2010). Despite an initial positive response to chemotherapy, most patients relapse and ultimately present with intraperitoneal chemoresistant disease and poor prognosis (Colombo et al., 2017). Meanwhile, recent breakthroughs in immune checkpoint therapies have led to little improvement in OvCa prognosis (Monk et al., 2021; Pujade-Lauraine et al., 2021). Therefore, there is an urgent need to develop alternative therapies for end-stage metastatic OvCa.

Multiple mechanisms contribute to therapy resistance in metastatic OvCa. First, the presence of disseminated cancer cells in the peritoneal fluid following initial treatment may promote therapy resistance in relapsed patients (Shield et al., 2009). A growing body of evidence suggests that these disseminating cells, which are present in malignant ascites and cannot be surgically resected, exhibit cancer stem cell characteristics that render them highly invasive and broadly resistant to therapy (Latifi et al., 2012; Liao et al., 2014; Shepherd and Dick, 2022; Shield et al., 2009; Ward Rashidi et al., 2019). Targeting these cells remains a significant challenge to overcoming therapy resistance and relapse.

The second mechanism of therapy resistance is likely driven by the highly immunosuppressive microenvironment in the peritoneal cavity (Almeida-Nunes et al., 2022). Its immunosuppressive nature is predominantly supported by myeloid cells, namely macrophages (MΦs) and neutrophils, which are the most abundant cell type found in OvCa tumors and malignant ascites (Charoentong et al., 2017; Izar et al., 2020; Lee et al., 2019; Raghavan et al., 2019; Rickard et al., 2021). The peritoneal cavity consists of at least three anatomical compartments: the peritoneal fluid, the omentum, and the peritoneal membrane. Recent studies have revealed that tissue-resident MΦs in all three compartments can suppress immune responses and promote OvCa progression (Casanova-Acebes et al., 2020; Etzerodt et al., 2020; Long et al., 2021; Miyamoto et al., 2023; Zhang et al., 2021a). Of note, omental MΦs are known to contribute to the formation of the pre-metastatic niche (Etzerodt et al., 2020; Krishnan et al., 2020) and neutrophils recruited to the omentum during early OvCa progression have also been reported to promote metastasis into the omentum (Lee et al., 2019). Therefore, alternative immunotherapy approaches directly targeting myeloid cells in the peritoneal cavity may be necessary to overcome disease progression and resistance.

Nearly a century ago, it was discovered that the administration of dead pathogens (specifically Coley’s toxin) could stimulate an antitumor response in some patients, likely via activating myeloid cells that promoted cancer killing (Wiemann and Starnes, 1994). Here, we sought to utilize a similar strategy to treat metastatic OvCa by administering β-glucan alongside IFNγ to activate myeloid cells in the peritoneal cavity. β-glucan is a polysaccharide derived from yeast cell walls and a known activator of myeloid cells. It has been shown to inhibit tumor progression in several non-OvCa tumor models (Bradner et al., 1958; Cheung et al., 2002; Hong et al., 2004; Kalafati et al., 2020; Woeste et al., 2023). Whether it can inhibit metastatic OvCa is not known. IFNγ is an immunogenic cytokine that is essential for innate and adaptive immunity. It was originally identified as an MΦ activation factor (Celada et al., 1984; Schreiber et al., 1982; Svedersky et al., 1984) and is crucial for reprogramming tumor-associated MΦs in multiple tumor models (Alspach et al., 2019; Sun et al., 2021). Importantly, IFNγ alone failed in the most recent clinical trial to improve the prognosis of patients with metastatic OvCa, indicating additional activation signals are necessary (Miller et al., 2009). Both β-glucan and IFNγ are currently in separate clinical trials to investigate efficacy in treating multiple cancers (NCT05159778 and NCT04628338). Importantly, emerging evidence suggests that therapies combining IFNγ and pathogen-derived molecules (e.g., β-glucan or lipopolysaccharide [LPS]) can reverse the immunosuppressive microenvironments in a few clinically relevant, orthotopic cancer models (Sun et al., 2021; Wattenberg et al., 2023). However, whether or how these therapies could treat metastatic OvCa remains poorly understood.

Here, we report that the intraperitoneal administration of β-glucan and IFNγ (BI) successfully induced the robust regression of metastatic OvCa tumors and controlled cancer metastasis. In the peritoneal fluid, we identified two complementary mechanisms through which β-glucan eliminates cancer cells: one that requires macrophage-mediated clotting in the peritoneal fluid and another that requires Dectin-1-Syk–dependent NETosis in the omentum. In solid metastases, we found that both agents for BI treatment are required for tumor regression. BI induced anti-tumor immunity in omentum tumors in part via MΦ-derived IL27. In vitro, BI directly induced IL27 secretion in MΦs, which consequently activates CD8⁺ T cells. Moreover, we found that higher IL27 expression predicted better overall survival in patients with metastatic OvCa. Overall, this study proposes a new promising therapeutic approach for treating metastatic OvCa and reveals novel mechanisms of tumor control in the peritoneal fluid and omentum tumors.

As β-glucan is a well-known activator of innate immunity and has been reported to control other tumor types in vivo (Bradner et al., 1958; Cheung et al., 2002; Hong et al., 2004; Kalafati et al., 2020; Woeste et al., 2023), we first evaluated whether β-glucan alone could effectively treat tumors in the commonly utilized ID8 mouse model of OvCa. Luciferase- and GFP-tagged ID8 cells were seeded intraperitoneally (i.p.) to simulate metastatic OvCa and mice were treated once every 2 wk with β-glucan (Fig. S1 A). Mouse tumor burden was significantly decreased in β-glucan treated mice (Fig. 1 A). Additionally, malignant ascites (the accumulation of tumor cell– and red blood cell–containing fluid in the peritoneal cavity—a hallmark of metastatic disease) was completely inhibited (Fig. 1 B) and ID8 cells were undetectable in the peritoneal lavage following β-glucan treatment (Fig. 1 C). These data indicate that β-glucan is sufficient to control ID8 tumor progression in vivo.

Although widely utilized for the past 20 years, it is now appreciated that ID8 cells do not harbor any common patient-relevant mutations or somatic copy-number alterations seen in human OvCa (Iyer et al., 2021; Roby et al., 2000; Walton et al., 2016). Therefore, we next wanted to test β-glucan efficacy in another syngeneic model of OvCa that utilizes a recently characterized cell line containing OvCa patient-relevant genetic alterations, KPCA (KRAS^G12VTrp53^R172HCcne1^{OE (OE)}Akt2^OE, Fig. 1 D) (Iyer et al., 2021). KRAS, TP53, CCNE1, and AKT2 are mutant alleles observed in high-grade serous ovarian tumors with frequencies of 12%, 96%, 19%, and 6%, respectively (Iyer et al., 2021; Zhang et al., 2021b). Interestingly, the amplification of KRAS and CCNE1 was recently identified as a marker for chemoresistant OvCa (Smith et al., 2023) and these tumors were indeed reported to be resistant to chemotherapy (Iyer et al., 2021). Therefore, utilizing this model is valuable for developing effective therapies against therapy-resistant OvCa in patients, an unmet clinical need.

Mice bearing KPCA tumors have a median survival of only 35 days, which is a nearly threefold decrease compared to the ID8 model which has a median survival of 114 days (Iyer et al., 2021; Roby et al., 2000). To account for this, treatment was initiated 1 wk following KPCA injection, and mice were treated with two doses of β-glucan on days 7 and 14 (Fig. S1 B). We confirmed the presence of KPCA cells (GFP⁺luciferase⁺) in the peritoneal fluid (Fig. 1 E and Fig. S1 C) as well as in the omentum at the time of treatment initiation (Fig. 1 F; and Fig. S1, D and E), which models metastatic OvCa (stage III or IV) in patients (Prat, 2014). For reasons that are not known, KPCA cells lose their luminescence in the peritoneal fluid. Therefore, a bioluminescent signal is indicative of solid tumor burden only and fluid tumor burden was exclusively analyzed by flow cytometric enumeration of GFP⁺ cells. In contrast to ID8 tumors, β-glucan could not reduce the KPCA solid tumor burden (Fig. 1 G). However, β-glucan once again was able to inhibit the accumulation of malignant ascites (Fig. 1 H and Fig. S1 F) and significantly reduced KPCA cell number in the peritoneal lavage (Fig. 1 I). Taken together these data suggest that β-glucan is sufficient to control fluid tumor burden independent of tumor type. However, β-glucan was unable to control the metastatic growth of KPCA tumors. The ability of β-glucan to control solid tumor progression of ID8 tumors but not KPCA further highlights the critical role of tumor mutation status on therapy response and emphasizes the importance of utilizing preclinical models which better model human cancer.

Given that β-glucan significantly reduced the presence of cancer cells in the ascites of tumor-bearing mice, we next sought to determine the mechanism by which β-glucan eliminates cancer cells from the peritoneal fluid.

One distinguishing characteristic of peritoneal resident macrophages (PRMΦs) is their ability to rapidly aggregate around foreign particles or pathogens, entrapping them in clot-like structures and facilitating their clearance from the peritoneal fluid (Barth et al., 1995). This process is known as the MΦ disappearance reaction (MDR) and is critical to control infection in the peritoneal cavity (Vega-Pérez et al., 2021; Zhang et al., 2019). To test whether MDR could be responsible for the disappearance of cancer cells from the peritoneal fluid, we administered GFP-labeled OvCa cells concurrently with β-glucan and analyzed peritoneal lavage 5 h later (Fig. S2 A). We first confirmed MDR in our model by observing the disappearance of PRMΦs (F4/80^hiCD11b^hiICAM2^hi) in the peritoneal fluid following β-glucan administration (Fig. 2 A). We next looked for the presence of GFP⁺ cancer cells in the peritoneal fluid to see if they also disappeared. Indeed, the presence of both ID8 cells (Fig. S2 B) and KPCA cells (Fig. 2 B and Fig. S2 C) was significantly reduced in β-glucan–treated mice. Because cancer cell disappearance does not appear to be dependent on cell type, KPCA cells were used for the remaining experiments. Similar to what has been previously reported in infection models, clot-like structures also formed in our cancer model following β-glucan treatment. These structures could be visualized as ∼1–2 mm GFP⁺ clots freely floating in the peritoneal cavity (Fig. 2 B and Fig. S2 D) and flow cytometric analysis of these structures confirmed the presence of CD45⁻GFP⁺ KPCA cells within (Fig. 2 D), thus suggesting that MDR can indeed capture cancer cells in the peritoneal fluid. To confirm this, we examined the peritoneal lavage of mice treated with clodronate-loaded liposomes (CLL), which deplete PRMΦs (Fig. S2 E) (van Rooijen and Hendrikx, 2010). Indeed, KPCA disappearance was partially impaired in CLL-treated mice (Fig. S2 F). Moreover, concurrent administration of β-glucan with heparin, an anticoagulant which inhibits MDR by disrupting PRMΦ clotting (Zhang et al., 2019), also partially impaired KPCA disappearance in the peritoneal lavage (Fig. 2 E). Finally, OvCa cells captured within these clots were more apoptotic than untreated cancer cells freely floating in the peritoneal fluid as determined by TUNEL staining analyzed by flow cytometry (Fig. S2 G). Taken together, these data confirm that intraperitoneal administration of β-glucan can capture OvCa cells floating in the peritoneal fluid into clot-like structures by activating MDR, leading to sequestration of OvCa cells out of the peritoneal fluid.

We next wanted to investigate the molecular mechanism underpinning β-glucan–mediated cancer removal from the peritoneal fluid. Dectin-1, a C-type lectin receptor expressed by myeloid cells, recognizes β-glucan and signals via downstream spleen tyrosine kinase (Syk) (Brown et al., 2002). To determine the contribution of Dectin-1-Syk signaling in cancer cell disappearance, we first analyzed the peritoneal lavage of mice with Syk-deficient myeloid cells (Syk^MyeΔ). Indeed, the disappearance of cancer cells from the fluid was impaired in Syk^MyeΔ mice as compared to their littermate controls (Syk^WT, Fig. 2 F). Surprisingly, impaired OvCa cell disappearance occurred independent of MDR as PRMΦs were still undetectable in the peritoneal lavage of Syk^MyeΔ mice after β-glucan treatment (Fig. 2 G). Moreover, neither Syk deficiency nor MDR inhibition completely reversed OvCa cell removal from the peritoneal fluid (Fig. 2, E and F), thus implying the existence of two independent cancer sequestration mechanisms: one which involves MDR and another which requires Syk signaling in myeloid cells.

The omentum is an immune-rich adipose tissue which is critical for clearing peritoneal contaminants and coordinating protective immune responses during peritonitis (Català et al., 2022; Meza-Perez and Randall, 2017). Notably, zymosan (a type of β-glucan) has been reported to be rapidly sequestered into the omentum following i.p. injection (Jackson-Jones et al., 2020). Therefore, to further elucidate the MDR-independent, Syk-dependent mechanism of cancer clearance, we examined the omentum in β-glucan–treated mice. To test whether the omentum can capture OvCa cells following intraperitoneal β-glucan administration, we imaged the omentum in situ 5 h following KCPA cell and β-glucan injection. Mouse body cavities were opened, and the omentum was gently stretched across the liver to ensure low background fluorescence and clear imaging. Green fluorescing OvCa cells were clearly visualized in the omentum following β-glucan treatment (Fig. 2 H), indicating that the omentum can indeed sequester OvCa cells following β-glucan administration. Moreover, OvCa cell disappearance from the peritoneal lavage was partially but significantly reversed in mice whose omentum was surgically removed by omentectomy (OMX, Fig. 2 I). Like what was observed in Syk^MyeΔ mice, reversal of OvCa cell disappearance occurred in OMX mice independent of MDR (Fig. S2 H), thus supporting the notion that sequestration by the omentum and MDR likely occurs independently of one another. To confirm this, we administered heparin in OMX mice where both MDR and omentum capture were inhibited, and observed cancer cell disappearance from the peritoneal fluid was completely reversed in these mice (Fig. 2 I). This supports the existence of two complementary pathways in two independent structures (clots formed in the peritoneal fluid and the omentum) that are required for total OvCa cell sequestration by β-glucan.

Given that cancer cell disappearance in Syk^MyeΔ mice phenocopied what was observed in OMX mice, we next wanted to test whether Dectin-1-Syk signaling could drive cancer cell sequestration in the omentum. Indeed, fewer cancer cells were observed in the omentum of Syk^MyeΔ mice as seen by imaging and flow cytometry (Fig. 2 J and Fig. S2 I). Moreover, like OMX mice, heparin administration in Syk^MyeΔ and constitutive Dectin-1 knockout mice (Dectin-1 KO) once again completely reversed OvCa cell disappearance in the peritoneal lavage (Fig. 2, K and L). Moreover, depletion of MΦs using CLL in Syk^MyeΔ mice also completely reversed OvCa cell disappearance (Fig. S2 J), further confirming that Dectin-1-Syk signaling drives OvCa cell capture in the omentum independent of MDR.

Finally, we sought to identify the mechanism through which OvCa cells were trapped within the omentum. A recent report demonstrated that the capture of zymosan in the omentum is mediated by neutrophil recruitment and activation of neutrophil extracellular traps (NETs) (Jackson-Jones et al., 2020). Additionally, Syk signaling has been reported to be a master regulator of NETosis in response to β-glucan (Nanì et al., 2015; Negoro et al., 2020; Zhu et al., 2023). Given that Syk signaling in myeloid cells is crucial for OvCa cell capture in the omentum, we posited that OvCa cells could be captured via Syk-dependent NETosis in the omentum. Whole-mount confocal imaging of the omentum 5 h after β-glucan and KPCA administration demonstrated that NETs, marked by citrullinated histone H3 (cHH3), were indeed induced by β-glucan and localized with neutrophils and OvCa cells in the omentum (Fig. 2 M). Moreover, this signal decreased in Syk^MyeΔ mice (Fig. 2 M), confirming that Syk-mediated NETosis may drive OvCa cell capture by the omentum. Finally, we found that inhibiting NETs directly by genetically deleting peptidyl arginine deiminase type IV in mice (PAD4 KO) (Li et al., 2010) reversed OvCa cell capture by the omentum (Fig. 2 N) and that adding heparin in PAD4 KO mice once again fully reversed OvCa cell disappearance from the peritoneal lavage (Fig. 2 O). Thus, β-glucan induces the capture of OvCa cells by the omentum via Dectin-1-Syk–mediated NETosis.

Taken together, we identified two nonredundant yet complementary pathways that mediate sequestration of disseminating OvCa cells from the peritoneal fluid by β-glucan: (1) an MDR-mediated intraperitoneal clotting mechanism in the peritoneal fluid and (2) a Dectin-1-Syk–dependent NETosis mechanism in the omentum. A simplified illustration of these pathways can be found in Fig. S2 K.

Although β-glucan treatment alone effectively controlled fluid tumor burden, it was unable to control metastatic growth of KPCA tumors in vivo (Fig. 1 H). Emerging evidence suggests that therapies combining IFNγ and pathogen-derived molecules can be utilized against cancers and have been tested in a few clinically relevant, orthotopic cancer models (Sun et al., 2021; Wattenberg et al., 2023). In light of this, we adopted a similar approach and posited that β-glucan in combination with IFNγ could potentially treat KPCA tumors.

To test the efficacy of BI against KPCA tumors, 1 × 10⁶ luciferase-tagged KPCA cells were seeded i.p. and treated as shown in Fig. 3 A. BI significantly reduced KPCA tumor burden as visualized by bioluminescence imaging, flow cytometry, and omentum weight (Fig. 3 B; and Fig. S3, A and B), but single-agent β-glucan or IFNγ did not. Reduction of KPCA tumor burden was completely impaired in IFNγ receptor KO mice (IFNγR KO, Fig. 3 C) and BI treatment did not reduce KPCA number in vitro (Fig. S3 C), indicating that the therapy is not directly cytotoxic to tumors but requires cancer-extrinsic IFNγR-mediated host immune responses. MΦ activation can promote tumor killing by activating T cells (Sun et al., 2021). To test whether T cells are required for the antitumor activity of BI, we depleted T cells in vivo using monoclonal antibodies against CD4 and CD8 (αCD4+αCD8). Indeed, mice lacking CD4⁺ and CD8⁺ T cells are unable to control tumor burden following BI (Fig. 3 D). Taken together, these data indicate that BI activates host immunity to robustly control metastatic KPCA tumors in vivo.

We next sought to determine whether BI could drive tumor regression. To this end, we tracked tumor growth longitudinally following treatment via bioluminescence imaging (Fig. S3 D). Notably, there was no difference in tumor burden across all treatment groups on days 8, 10, or 14; however, there was a significant decrease in tumor burden in only BI-treated mice on day 21 (Fig. 3, E and F; and Fig. S3 E). These data indicate that BI may induce direct tumor killing and disease regression, although it does not rule out the possibility that BI may reduce tumor growth. Compartmental imaging of the omentum and the non-omentum body cavity (Fig. S1 D) revealed that the omentum had the greatest tumor burden in untreated mice. Compartmental imaging of BI-treated mice revealed that both β-glucan and IFNγ were required to control tumors in the omentum (Fig. 3 G) as well as throughout the rest of the cavity (Fig. 3 H). Moreover, ascites accumulation was once again eradicated in both β-glucan– and BI-treated mice (Fig. S3, F and G). To test the toxicity of BI, we performed a complete clinical chemistry panel with IDEXX and analyzed multiple organ damage markers in the serum of mice on day 21. Of the 14 parameters measured, only a change in phosphorous levels reached statistical significance (P = 0.0020, Fig. S3 H) being decreased in the serum of BI-treated mice as compared to PBS-treated. However, AST, ALT, alkaline phosphatase, and creatine kinase concentrations trended up while bilirubin and blood urea nitrogen levels trended down in BI-treated mice as compared to PBS-treated mice (Fig. S3 H). Finally, there was no difference in body weight between PBS- and BI-treated mice (Fig. S3 I). These data suggest that two rounds of BI treatment may induce some insignificant amount of toxicity in mice; however, a more thorough toxicity study is warranted prior to testing in humans.

Taken together, these data demonstrate that BI activates host immunity to robustly kill KPCA tumors in metastatic sites in the peritoneal cavity in vivo.

BI can signal through receptors expressed by MΦs (Dectin-1 and IFNγR, respectively). To test whether MΦs are required for BI-induced antitumor immunity, we depleted MΦs using CLL during the course of BI treatment and found that mice treated with CLL had a higher tumor burden than BI+PBS-treated mice (Fig. 4 A), suggesting that MΦs are required for the optimal tumor control induced by BI.

Next, we sought to investigate how MΦs facilitated antitumor immunity. We focused on omentum tumors because they were the largest tumor burden and very few tumor nodules were present in any other peritoneal compartment or the fluid following BI (Fig. 3, G and H). Flow cytometric analysis of omentum tumors revealed a reduction in pro-tumor Arginase-1⁺ and Tim-4⁺ MΦs (Fig. S4 A), which have been reported to promote tumor progression by suppressing T cell functions (Noy and Pollard, 2014; Xia et al., 2020). To determine whether there is a unique subset of monocytes/MΦs which may promote antitumor activity, we performed scRNA-seq on omentum tumors and identified eight clusters of monocytes/MΦs (Fig. 4, B and C). Using the Immunological Genome Project dataset as a reference, Clusters 1 and 5 were identified as monocytes; all other clusters were identified as MΦs. Notably, Cluster 2 frequency was selectively enriched in BI-treated mice, while Cluster 1 was reduced (Fig. 4 D). Cluster 1 monocytes selectively expressed inflammation-related genes such as Il1b and Vcan, whereas Cluster 2 MΦs were enriched with genes related to antigen presentation (H2-Aa, H2-Eb1), immune activation (Ccl8, Vcam1), and complement activation (C1qa, C1qb, C1qc) (Fig. 4 C).

To better understand the origin and development of Cluster 2 MΦs we performed Slingshot trajectory analysis of our single-cell RNA sequencing (scRNA-seq) dataset. Setting Cluster 5 monocytes as the origin, we found three distinct differentiation pathways (Fig. 4 E). All trajectories passed from Cluster 5 through Clusters 1 and 3 before diverging and terminating in Clusters 2, 7, and 8. These data suggest that Cluster 2 MΦs developed from monocytes through a unique pathway. To support this hypothesis, we cultured monocytes isolated from bone marrow and stimulated them with BI during their in vitro differentiation into MΦs. MΦs that differentiated in the presence of BI upregulated multiple markers of Cluster 2 MΦs, such as Vcam1, H2-eb1, and Ccl8, but not C1qa (Fig. 4 F). Additionally, we investigated whether BI could systematically augment immune progenitors and monocytes in vivo. To this end, we analyzed bone marrow cells by flow cytometry 1 wk after BI treatment and found that Lin-Sca1+cKit+ progenitor cells, long-term hematopoietic stem cells, multipotent progenitor cells, and Ly6C^hi monocytes were all increased in number compared to PBS-treated mice (Fig. S4 B). Taken together, these data suggest that Cluster 2 MΦs develop from bone marrow monocytes but require additional signals in the tumor microenvironment to mature fully.

To understand what signals Cluster 2 MΦs may upregulate to activate T cells to drive an antitumor response, we chose to focus on MΦ-derived cytokines, which are key approaches MΦs use to activate T cells. Ingenuity pathway analysis (IPA) specifically focusing on cytokine regulators in Cluster 1 and 2 confirmed the pro-inflammatory phenotype of Cluster 1 monocytes as inflammatory cytokine pathways were highly enriched, such as IL1β and IL6 (Fig. 4 G). In contrast, both subunits of the IL27 cytokine heterodimer (IL27 and EBI3) were significantly enriched in Cluster 2 (Fig. 4 G), and Il27 and Ebi3 co-expression was predominately detected in Cluster 2 MΦs (Fig. 4 H). Moreover, monocyte-derived MΦs in the presence of BI expressed higher levels of Il27 when compared to untreated MΦs (Fig. 4 I). IL27 is an unconventional IL12-family cytokine that has both pro- and antitumor properties depending on tumor types (Fabbi et al., 2017; Yoshida and Hunter, 2015), while IL12 is the classical MΦ-derived antitumor cytokine (Chan et al., 1991; Kaczanowska et al., 2021). Surprisingly, IL12 was almost undetectable in any of our MΦ subpopulations (Fig. S4 C).

Lastly, to determine whether IL27 signaling is relevant in OvCa patients, we analyzed a recently published scRNA-seq dataset of tumors and ascites from patients with metastatic OvCa (Vázquez-García et al., 2022). In these tumors, IL27 and EBI3 were exclusively co-expressed in only monocytes/MΦs (Fig. S4, D and E), specifically in the “M2.CXCL10” subset (Fig. 4, J and K). To our surprise, we did not find significant expression of IL12 in these tumor samples (Fig. S4 F), consistent with the result from our mouse model. Finally, utilizing public datasets, we analyzed the correlation between overall patient survival and IL27 expression. Indeed, higher co-expression of IL27 and EBI3 significantly correlated with improved survival in patients with stage III/IV metastatic OvCa (Fig. 4 L, P = 0.0059). In summary, these results suggest that BI treatment causes regression of metastatic OvCa by promoting differentiation of monocytes into IL27⁺ antitumor MΦs.

To test whether IL27 contributes directly to the antitumor response of BI, we neutralized IL27 using a monoclonal antibody against IL27p28 in the presence of BI in vivo. Indeed, IL27 neutralization significantly impaired the antitumor activity in both omental and mesenteric metastases as compared to IgG controls (Fig. 5, A and B). Given the requirement of T cells for BI efficacy (Fig. 3 D), we next evaluated changes in T cells following BI treatment and examined the potential role of IL27 underlying these responses. T cells from PBS- and BI-treated omentum tumors were restimulated ex vivo and analyzed by flow cytometry. T cells secrete cytokines such as IFNγ and TNF upon activation to drive anti-tumor immune responses. In BI-treated tumors, although CD8⁺ T cell frequencies did not change (Fig. S5 A), IFNγ⁺ and TNF⁺ CD8⁺ T cells were enriched (Fig. S5 B) and IFNγ and TNF mean fluorescent intensity (MFI) was increased (Fig. 5 C). Interestingly, in CD4⁺ T cells, only the frequency of TNF⁺ was increased, and no other significant changes were observed (Fig. S5 C). Therefore, BI treatment is immunostimulatory and its efficacy likely relies on the activation of CD8⁺ cytotoxic T cells.

To test whether MΦ-derived IL27 contributes to the immunostimulatory effect seen in BI treatment, we established an in vitro system using MΦs and CD8⁺ T cells. First, to test whether the IL27 can be directly secreted from MΦs stimulated by BI, we treated bone marrow–derived MΦs (BMDMs) with BI for 48 h and tested the supernatant for secreted IL27 using an IL27/EBI3 heterodimer-specific ELISA. Interestingly, while β-glucan alone can generate a small amount of IL27, both β-glucan and IFNγ are required for robust IL27 secretion (Fig. 5 D). It has been previously reported that the p28 subunit of IL27 (also known as IL30 when IL27p28 forms a monomer) can be generated in MΦs stimulated with IFNγ or LPS (another pathogen-associated molecular pattern [PAMP] molecule) as detected by an IL27p28 ELISA (not specific for IL27 heterodimer). In this context, a single agent is sufficient to induce IL27p28 production, but combining the two agents synergizes to produce a maximal amount of IL27p28 (Liu et al., 2007). Notably, we see a similar pattern in IL27p28 secretion in BMDMs treated with β-glucan or IFNγ (Fig. S5 D). Therefore, secretion of the IL27 heterodimer is specific to BI treatment despite IL27p28 generation induced by single agents, suggesting that both agents may be necessary for the transcription of EBI3, but how this occurs remains unknown. Surprisingly, IL27 did not require Dectin-1-Syk signaling as BMDMs from Dectin-1 KO and Syk^Mye∆ mice still secreted IL27 following BI treatment (Fig. 5 E). On the other hand, IFNγR was crucial for IL27 production as loss of this receptor completely ablated IL27 secretion in BMDMs (Fig. 5 E). Therefore, we show that secretion of the IL27 heterodimer is a specifically regulated event which requires IFNγ receptor signaling but not Dectin-1 or Syk. Given the role IL27 plays in BI efficacy (Fig. 5 A), the specific generation of IL27 by BI may offer one explanation as to why single agent β-glucan or IFNγ treatment was not sufficient in controlling OvCa tumor burden (Fig. 3 B).

To test whether MΦ-derived IL27 could drive the immunostimulatory phenotype in CD8⁺ T cells, naïve OT-I CD8⁺ T cells were co-cultured with PBS- or BI-pretreated BMDMs, OVA peptide, and dendritic cells (DCs) in the presence of control or an IL27p28 neutralization antibody. Indeed, co-culturing T cells with BI-pretreated BMDMs increased the frequency of IFNγ⁺, TNF⁺, and IFNγ⁺TNF⁺ CD8⁺ T cells (Fig. S5, E and F) and increased IFNγ and TNF MFI (Fig. 5 F) in an IL27-dependent manner. BI-pretreated MΦs also increased Granzyme B expression (a cytotoxic effector molecule in CD8⁺ T cells) but did so independently of IL27 (Fig. S5 G). Taken together, these data suggest that BI treatment directly stimulates IL27 expression in MΦs and that MΦ-derived IL27 promotes the anti-tumor activity of BI treatment via activating cytotoxic CD8⁺ T cells.

To test whether βI can be combined with platinum-based chemotherapy (standard-of-care for OvCa patients) to treat metastatic OvCa, we treated the homologous recombination (HR)–proficient chemoresistant KPCA and HR-deficient chemo-sensitive BPPNM (p53^−/−R172HBrca1^−/−Pten^−/−Nf1^−/−Myc^OE) tumors (Iyer et al., 2021) with BI once a week for 2 wk with or without carboplatin and monitored their overall survival. As expected in the chemoresistant KPCA model, carboplatin alone did not yield any survival advantage compared to PBS controls, but BI significantly extended the overall survival and led to the survival of 20% of mice at the end of the study. In addition, combining BI with carboplatin did not lead to a survival advantage compared to BI alone. On the other hand, in the HR-deficient chemo-sensitive BPPNM model, carboplatin alone significantly extended the overall survival, while BI modestly prolonged the survival (Fig. 6 B). In sharp contrast to the KPCA model, combining BI with carboplatin led to 80% of mice alive at the end of the study in mice carrying BPPNM tumors (Fig. 6 B). These results suggest that combining BI with platinum-based chemotherapy may yield a significant therapeutic advantage compared to chemotherapy alone in HR-deficient OvCa, while BI may be a treatment option for chemoresistant HR-proficient OvCa.

T-cell-based immunotherapy has achieved significant advancements in the treatment of various cancers; however, it has yet to yield effective therapeutic outcomes for metastatic OvCa. In this study, we present an alternative immune therapy which harnesses myeloid cells against aggressive metastatic OvCa. In summary, we report here that β-glucan, a PAMP molecule, in combination with the immunogenic cytokine IFNγ coordinates a robust antitumor response in a patient-relevant murine model of metastatic OvCa. Tumor burden regressed in multiple metastatic compartments including the peritoneal fluid, omentum, and other parts of the peritoneal cavity. Surprisingly, β-glucan alone was sufficient to control fluid tumor burden despite not controlling solid tumor progression. Mechanistically, we demonstrated that sequestration of tumor cells out of the peritoneal fluid requires intraperitoneal clotting accompanied by MDR and Dectin-1-Syk–dependent NETosis in the omentum (Fig. S2 K). Systemic tumor regression following BI was facilitated through tumor-extrinsic signaling, requiring both T cells and IFNγR signaling from the host. ScRNA-seq analysis of omentum tumors revealed a selective enrichment of monocyte-derived IL27⁺ MΦs following BI. Neutralizing IL27 in vivo significantly impaired BI-induced antitumor immunity, demonstrating that IL27 indeed drives antitumor immunity likely through MΦs. In agreement with this notion, MΦs treated directly with BI in vitro can secrete IL27 and activate CD8⁺ T cells in an IL27-dependent manner. Moreover, in patients with metastatic OvCa, IL27/EBI3 co-expression predicted better overall patient survival. Finally, BI significantly extended the overall survival of mice with clinically relevant metastatic OvCa and dramatically enhanced the efficacy of platinum-based chemotherapy in vivo in a clinically relevant, HR-deficient metastatic OvCa model. Taken together, these data suggest the therapeutic potential of BI in treating metastatic OvCa in patients.

In contrast to the hematological route of metastasis seen in other cancers, OvCa cells readily disseminate directly into the peritoneal fluid as single cells or multicellular aggregates prior to seeding in secondary peritoneal metastatic sites. This route of dissemination is uniquely challenging as these fluid-bound spheroids are functionally dormant, rendering them resistant to anoikis (a form of apoptosis following cell detachment) and proliferation-targeting chemotherapy (Shepherd and Dick, 2022). At the same time, highly invasive and therapy-resistant OvCa stem cells have been reported to be enriched in malignant ascites in patients (Raghavan et al., 2019; Ward Rashidi et al., 2019). One theory of relapse suggests that cancer cells left behind in the fluid following surgical debulking and therapy give rise to chemoresistant patient relapse (Liao et al., 2014; Shepherd and Dick, 2022). Therefore, numerous studies have attempted to elucidate cancer intrinsic vulnerabilities of disseminating cancer cells in the peritoneal fluid in an effort to improve therapy response (Buensuceso et al., 2020; Haagsma et al., 2023; Latifi et al., 2012). To our knowledge, this study is the first of its kind to demonstrate a cancer extrinsic approach for targeting these cells. Here we demonstrate that peritoneal immune responses which canonically target pathogens can be exploited to target disseminating cancer cells in the peritoneal fluid, seemingly independently of cell types (Fig. 2 B and Fig. S2 B). Moreover, not only did intraperitoneal β-glucan administration sequester cancer cells out of the peritoneal fluid, it also induced acute cancer killing in peritoneal clots (Fig. S2 G). Interestingly, BI enhanced cancer killing in these clots as well (Fig. S5, H and I). The consequence of clot biology in killing cancer cells remains an open question and warrants further study. Meanwhile, although the acute fate of cancer cells trapped by the omentum was not investigated in this study, the consequence of rapid sequestration of cancer cells into the omentum still has the potential to revolutionize OvCa treatment. Despite efforts in improving therapeutics against OvCa, one of the greatest predictors of prognosis in stage III and IV patients is still the maximal removal of tumors during cytoreduction surgery (Bristow et al., 2023). Because most OvCa patients will undergo omentectomy as part of treatment, preoperative intraperitoneal administration of β-glucan could potentially improve surgical outcomes by trapping fluid cancer cells in the organ prior to its removal. However, further research is needed to test the efficacy and safety of such an approach. Still, this study shows for the first time that exploring cancer extrinsic mechanisms of cancer targeting in the peritoneal fluid holds promise to address a critical need.

The pleiotropic cytokine IL27 has been reported to contribute to tumor immunity in a context-dependent manner (Fabbi et al., 2017). It was initially reported as an IL12-like cytokine secreted by antigen-presenting cells that drive T-cell activation (Pflanz et al., 2002) and antitumor immunity (Liu et al., 2022; Patidar et al., 2022). More recently, accumulating evidence demonstrates the role of IL27 in promoting expansion of regulatory T cells (Do et al., 2017; Hall et al., 2012), expression of T cell checkpoint receptors (Carbotti et al., 2015; Hirahara et al., 2012), and survival of tumor cells (Jia et al., 2016), revealing its context-dependent role in modulating tumor immunity. This starkly contrasts with IL12, which is universally considered to be antitumor. IL12 is strongly induced by PAMP/IFNγ stimulation in vitro and is a hallmark of antitumor MΦs. Recent in vivo work has demonstrated that co-stimulation of MΦs by MPLA (a PAMP molecule, acting as a TLR4 agonist) and IFNγ in breast cancer and OvCa induces IL12 production by MΦs and promotes antitumor T cell activity (Sun et al., 2021). However, the serious toxicity of IL12 in humans prevented its clinical usage (Leonard et al., 1997). To this end, IL27 could be a promising cytokine that stimulates local rather than systemic antitumor immunity. Interestingly, in published human and our own mouse scRNA-seq datasets of OvCa, expression of IL12 was not highly induced (Fig. S4, C and F), suggesting that different tumor microenvironments can induce differential antitumor cytokine-secretion in MΦs. These suggest that IL27 is a promising antitumor cytokine in treating metastatic OvCa. Whether IL27 is a viable therapeutic target in OvCa requires future investigations. Finally, as to the regulation of IL27 expression and secretion, we found that both agents in BI treatment were required to stimulate IL27 heterodimer secretion in vitro. Meanwhile, either β-glucan or IFNγ on their own is sufficient to drive the secretion of IL27p28 (the monomer form is also known as IL30), although BI is still necessary to produce the highest response (Fig. S5 D). This is consistent with previous reports which demonstrated that LPS (another PAMP molecule) induces IL27p28 expression and secretion, and this effect was enhanced by the addition of IFNγ (Liu et al., 2007). Therefore, understanding the genetic regulation of EBI3, the other subunit of IL27, or the secretion of the IL27 heterodimer following BI treatment may ultimately dictate IL27 regulation in MΦs. Moreover, future studies will answer how IL27 is induced in vivo and definitively demonstrate its cellular source and target.

To our surprise, we identified two β-glucan–driven mechanisms which did not require canonical Dectin-1-Syk signaling: (1) MDR and MDR-mediated cancer cell capture (Fig. 2, G, K, and L) and (2) IL27 stimulation in MΦs (Fig. 5 E). This is surprising because Dectin-1 is considered to be the primary receptor for β-glucan signaling in MΦs (Brown et al., 2002; Goodridge et al., 2011). To our knowledge, while Syk-independent β-glucan signaling has been identified (Gringhuis et al., 2009; Herre et al., 2004), how β-glucan signals in MΦs independent of Dectin-1 is not well understood. β-glucan can bind to complement and therefore may signal through complement receptor 3 in addition to Dectin-1. However, this interaction has been predominately studied in neutrophils and its contribution to phagocytosis of β-glucan particles in MΦs (like what is used for this study) is insignificant (Goodridge et al., 2009). Therefore, identifying the mechanism through which β-glucan drives MΦ clotting and IL27 is ongoing.

The ability of the innate immune system to generate memory responses to secondary infectious or inflammatory responses is called immune training (Netea et al., 2016). Immune training can be induced by β-glucan and was initially thought to be only involved in infection and inflammation (Netea et al., 2011; Quintin et al., 2012). Recently, it has been appreciated that β-glucan-mediated immune training could be one approach for the treatment of several cancers, inducing robust antitumor activity and protecting against relapse in preclinical models (Ding et al., 2023; Geller et al., 2022; Kalafati et al., 2020; Woeste et al., 2023). Although we did not set out to comprehensively analyze trained immunity in this study, we did observe an increase in progenitor cell numbers in the bone marrow of mice 1 wk after BI treatment (Fig. S4 B), which may hint at a role of trained immunity for BI efficacy. Conversely, however, we also found that T cells were indispensable for BI-induced antitumor activity against metastatic OvCa. This finding opposes one key hallmark of trained immunity; that is its activity occurs independently of T cells (Kalafati et al., 2020; Priem et al., 2020). Therefore, it is unclear from this study whether immune training contributes to BI antitumor activity. However, future studies are warranted to evaluate this remaining question comprehensively.

In summary, we have established an alternative immune therapy which combines β-glucan and IFNγ to drive regression of metastatic OvCa in vivo. While more work is to be done on elucidating specific antitumor mechanisms in MΦs and other cell types following β-glucan+IFNγ and its impact on hematopoiesis and immune training, our data suggest (1) a novel concept of targeting tumor cells floating in the fluid by “relocating” them to adherence to an immune-rich environment for better killing and (2) a novel role of IL27 in promoting antitumor immunity in metastatic OvCa. We acknowledge that the antitumor role of IL27⁺ MΦs is only partially responsible for tumor elimination. Mechanisms mediated by other cell types will be investigated in future studies. Moreover, we propose to further investigate BI because of its tremendous therapeutic potential, which could improve the lives and survival of metastatic OvCa patients. Finally, while most immunotherapeutic approaches focus on targeting adaptive immunity, our data highlight the importance of harnessing innate immunity in developing robust anti-cancer immunotherapies.

The original ID8 cell line, derived from spontaneous in vitro malignant transformation of C57BL6 mouse ovarian surface epithelial cells (Roby et al., 2000), was modified to express GFP and firefly luciferase. KPCA, and BPPNM cells, which were recently generated and characterized (Iyer et al., 2021), were generously gifted to us by Drs. Robert Weinberg and David Pepin at the Whitehead Institute, Cambridge, MA, USA, and the Mass General Research Institute, Boston, MA, USA, respectively. KPCA cells were previously modified to express GFP and firefly luciferase. We modified BPPNM cells to express firefly luciferase. All cell lines were cultured in Dulbecco’s modified Eagle’s medium (10-017-CV; DMEM; Corning) supplemented with 4% FBS (16140-071; Gibco), 1% penicillin/streptomycin, 1X Insulin-Transferrin-Selenium (cat# 41400-045; Gibco), and 2 ng/ml mouse epidermal growth factor at 37°C supplied with 5% CO₂. Cells were passaged no more than five times prior to injection into mouse peritoneal cavities.

C57BL/6J (WT, #000664), B6.129S7-Ifngr1^tm1Agt/J (IFNγRKO, #003288), B6.129S6-Clec7atm1Gdb/J (Dectin-1 KO, # 012337) and B6.Cg-Padi4^tm1.1Kmow/J (PAD4 KO, #030315) mice were purchased from Jackson Laboratories. Lyz2^Cre/+;Syk^fl/fl (Syk^Mye∆) and their Cre negative littermate controls (Syk^WT) mice were generated by crossing Lyz2^Cre mice (#004781; Jackson) with Syk^fl/fl mice, which were generously gifted to us by Dr. John Lukens from the University of Virginia, Charlottesville, VA, USA. All mice were housed in individual microisolator cages in a rack system with filtered air in Wistar’s mouse barrier facility and provided with shelter and enrichment to reduce stress. Reducing stress in mice is critical as stress has been reported to impede anti-cancer immunotherapy (Yang et al., 2019). Anecdotally, we observed similar effects in our model. Mice were kept on a 12-h light-dark cycle and had access to food and water ad libitum. All animal procedures were performed in accordance with the Wistar Institutional Animal Care and Use Committee under protocol 201536. Genotyping was performed utilizing Transnetyx automated genotyping services.

Cells were harvested with trypsin-EDTA (corning), washed in PBS, and injected i.p. into mice. 3 × 10⁶ ID8 cells were injected in 100 μl PBS into 8–10-wk-old female WT mice and allowed to seed for 2 wk prior to β-glucan treatment. Mice were treated with 500 µg sonicated whole β-glucan particles (tlrl-wgp; Invivogen) in PBS i.p. or PBS vehicle control once every other week for 2 wk for a total of two injections (Fig. S1 A). 2 wk after the final dose of β-glucan, tumor burden was assessed by IVIS Spectrum Imaging (PerkinElmer). For KPCA tumors, 1 × 10⁶ KCPA cells were injected i.p. into WT or IFNγR KO in 200 μl of a 1:1 matrigel:PBS mix (Matrigel Matrix Basement Membrane, 354234; Corning). Tumors grew for 1 wk prior to treatment. Mice were treated with 500 µg β-glucan, 20 µg recombinant mouse IFNγ (315-05; Peprotech), βI, or PBS vehicle control once a week for 2 wk (Fig. S1 B). 1 wk after the final treatment, mice were imaged by IVIS. Importantly, β-glucan was sonicated intermittently on high for 15 min immediately prior to injection to ensure thorough disruption of β-glucan aggregates. Recombinant mouse IFNγ was gently reconstituted in molecular grade, sterile H₂O and diluted in PBS to its working concentration. To preserve IFNγ activity, solutions were handled gently to reduce the presence of bubbles and never vortexed. For macrophage depletion studies, mice were treated with 100 μl of CLL (C-005; Liposoma) 1 wk prior to cancer capture studies. For tumor studies, 100 μl of CLL was injected i.p. 5, 9, 14, and 19 days after tumor seeding. On day 14, CLL was administered 4 h before BI treatment. For T cell depletion studies, αCD4 (C2838; Leinco Tech) and αCD8 (C2850; Leinco Tech) monoclonal antibodies were injected 150 μg each i.p. in WT KPCA tumor-bearing mice 3 days following cancer seeding and then once a week for 2 wk for a total of three injections. For the IL27 neutralization studies, 200 µg αIL27p28 monoclonal antibody (BE0326; InVivoMAb) (Marillier et al., 2014) was injected 2 days prior to βI treatment, at the same time as treatment, and then two times a week for 2 wk following treatment for a total of six injections. For treatment of experiments with carboplatin, 10–30 mg/kg once a week of carboplatin was administered with or without βI, starting at day 7 for 2 wk. The dose and timing of β-glucan and IFNγ were the same as mentioned above. On day 21, mice were imaged by IVIS and further monitored for survival analysis.

In vivo and ex vivo bioluminescence imaging was performed on an IVIS 50 (Living Image 4.3.1; PerkinElmer), with exposures of 1 s to 1 min, binning 2–8, field of view 12.5 cm, f/stop 1, and open filter. For in vivo imaging, D-Luciferin (150 mg/kg in PBS; Gold Biotechnology) was injected into the mice i.p. and imaged 10 min later. Mice were maintained under general anesthesia by continuous inhalation of 2–1.5% isoflurane in 60% oxygen. For ex vivo imaging, mice were injected with D-Luciferin and euthanized after 5 min. Mice peritoneal cavities were exposed, and the omentum was excised. Mouse carcasses (“non-omentum”) and omentum were placed in the machine and imaged as shown in Fig. S2 D. The total photon flux (photons/s) was measured from regions of interest using the Living Image 2.6 program.

Mesentery metastasis score was calculated based on the following criteria. Whole disseminated mesenteric tumors were counted and each mouse score was determined as follows:

• 0: no tumor was detected,

• 1: number of tumor nodules is <10,

• 2: number of nodules is 10–30,

• 3: number of nodules is over 30.

2 × 10⁶ cancer cells in 100 μl of PBS were injected i.p. into 8–12-wk male and female WT, Syk^Mye∆, Syk^WT, Dectin-1 KO, or PAD4 KO mice immediately followed by 500 µg sonicated β-glucan in 100 μl PBS. 5 h later, mice were euthanized, peritoneal lavage was taken, and the omentum were imaged as described below (Fig. S2 A). Clots were harvested 24 h after β-glucan injection and digested prior to analysis by FACS as described below. Peritoneal lavage and clots were analyzed by FACS for the presence of GFP⁺CD45⁻ cancer cells and F4/80^hiICAM2^hiCD11b^high PRMΦs. Omentum was imaged to identify the presence of GFP+ cancer cells. To reduce background fluorescence, the omentum was gently stretched over the liver and imaged. For mice treated with heparin, 100 U/mouse of heparin was given in the same syringe as β-glucan. In the MΦ depletion model, mice were treated with 100 μl of CLL 1 wk prior to cancer cell capture assays. Omentectomized mice were allowed to recover for 4 wk prior to entering this study. Data from combined experiments are presented as fold change. Fold change was calculated as n ÷ avg of the contol where n = experimental value.

Extended-release buprenorphine (3.25 mg/kg) was given subcutaneously preoperatively for pain management. Surgical removal of the omentum was accomplished under general anesthesia by continuous inhalation of 2–3% isoflurane in 60% oxygen using a veterinary vaporizer. Aseptic techniques were performed to maintain sterility in the surgical field. 6–8-wk-old male and female mice were used. Mice were placed on a heating pad in a supine position. A midline incision was made in the region of the stomach and the greater omentum was carefully exposed. The omentum is a thin, elongated adipose tissue that is located under the stomach and between the spleen and pancreas. The omentum was removed via electrocautery to avoid bleeding and the midline incision was closed with absorbable sutures in two layers (first the peritoneal wall was closed and then the skin). Mice were immediately placed in a clean heated cage and monitored until awake. A liquid recovery diet was provided, and mice were monitored daily for 7 days for signs of infection. Removal of both the entire greater and lesser omentum results in malperfusion of the stomach and spleen and thus was not feasible. Mice were allowed to recover for 4 wk before any further experimental procedures were performed.

Peritoneal lavage was collected by flushing the peritoneal cavity with 6 ml FACS buffer (PBS with 2 mM EDTA and 0.1% BSA). Mice were gently massaged to ensure optimal collection of peritoneal cells. Peritoneal clots were collected 24 h after β-glucan injection and omentum tumors were collected at the conclusion of each study. Clots and tumors were digested in the same way using a cocktail of 1 mg/ml collagenase IV and 100 µg/ml DNase I, in 1–4 ml DMEM with 10% FBS depending on tissue size. The tissues were minced into small pieces prior to digestion at 37°C for 30 min shaking at 800 rpm. Samples were then passed through a 70 µm cell strainer to collect single-cell suspension to be analyzed by flow cytometry. If necessary, single-cell suspensions were treated with 1X red blood cell lysis buffer (555899; BD Bioscience) on ice for 10 min.

Single-cell suspensions were collected as described above prior to staining with primary conjugated antibodies at their indicated dilutions (Table S1). Surface stain antibodies were incubated with cells on ice for 30 min and washed with 1 ml FACS buffer (PBS with 2 mM EDTA and 0.1% BSA). Intracellular staining was carried out using True-Nuclear Transcription Factor Buffer Set (424401; Biolegend) according to manufacturer instructions. Briefly, cells were fixed for 45 min–1 h at room temperature in the dark. Notably, KPCA cells lose GFP signal following fixation. Following fixation, cells were washed 1× in permeabilization buffer and then incubated in permeabilization buffer with intracellular stain antibodies overnight at 4°C. Counting beads (424902; Biolegend) were added to each sample to determine cell numbers. Samples were analyzed on a BD FACSymphony A3 Cell Analyzer using FlowJo software. For IFNγ and TNF staining in T cells, single cells from the omentum tumor were incubated for 4 h with Cell Activation Cocktail (with Brefeldin A) (catalog no. 423303; Biolegend) prior to staining.

To analyze bone marrow progenitors, 2–5 × 10⁶ bone marrow cells were stained as single cells suspension in FACS buffer for 100 min at 4°C as follows. Cells were incubated with Aqua Live/Dead (BV510) and anti-CD16/32 BV711. After 10 min, the anti-CD34 PE was added. After 30 min, the remaining antibodies cocktail was added (anti-LY6G BUV563, B220 FITC, CD90.2 FITC, NK1.1 BUV661, Sca1 PerCpCy5.5, CD117 [cKit] BV786, CD135 BV4221, CD150 PECy5, CD48 AF700, LY6C BV605, CD81 PECy7, CD115 PE/Dazzle, CD11B BUV805, CD106 BUV737). Cells were washed and resuspended in 500 μl of FACS Buffer and acquired with a BD FACSymphony A5 Cell Analyzer and analyzed using FlowJo software.

5 h after the injection of β-glucan and KPCA cells, the omentum was excised and fixed in 4% paraformaldehyde overnight at 4°C. On the next day, the tissues were rinsed with 3x PBS, blocked with 3% BSA and 1% Triton in PBS at room temperature for 60 min, and incubated with primary antibodies at their optimized dilutions (Table S1)—α-cHH3 and α-S100A9 overnight at 4°C. The tissues were then washed with 3x PBS before being incubated with respective secondary antibodies at room temperature for 1 h. After washing, the tissues were stored in PBS. Whole-mount confocal images were collected using a Leica SP8 microscope.

In situ omentum images were also taken using a Leica M205 FA fluorescence stereo microscope. Briefly, shortly following euthanization, the mice peritoneal cavities were exposed, and the omentum was located. Because several organs autofluoresce, such as the intestines (Fig. S2 D), the omentum was gently stretched over the liver, which does not autofluoresce, prior to imaging to minimize background.

The femur and tibia were harvested from 8-wk-old C57BL6 mouse. Two ends of the bones were cut and the bone marrow was flushed with a 26 g needle filled with cold sterile 1X PBS through a 40 μm cell strainer. 10 ml total PBS was used for all four bones. The bone marrow was centrifuged at 500 g for 5 min at 4°C and then the pellet was resuspended in 500 μl of FACS buffer.

Monocytes were isolated using EasySep Mouse Monocyte Isolation Kit (#19861; Stem Cell Technologies) and plated at a density of 200,000 cells per well of a 24-well plate (4 × 10⁵ cells/ml). Monocytes were differentiated in the presence of 20 ng/ml of CSF1 in RPMI with 10% FBS and 1x penicillin/streptomycin. After 24 h in culture, 10 μg/ml β-glucan and 33 ng/ml IFNγ were added for an additional 48 h. Total RNA was isolated from cultured cells using TRIzolTM reagent (Invitrogen), according to the manufacturer’s protocols. Glycoblue (Invitrogen) was added as a co-precipitant when handling <10⁶ cells.

cDNA was synthesized using a High-Capacity RNA-to-cDNA Kit (#4387406; Applied Biosystems), according to the manufacturer’s protocols. qRT-PCR was performed using SYBR Green PCR Master Mix (#4344463; Applied Biosystems) on a QuantStudio 3 Real-Time PCR Instrument (Applied Biosystems). The following primers were used for qRT-PCR:

Vcam1 forward primer 5′-AGT​TGG​GGA​TTC​GGT​TGT​TCT-3′; Vcam1 reverse primer 5′-CCC​CTC​ATT​CCT​TAC​CAC​CC-3′; C1qa forward primer 5′-AAA​GGC​AAT​CCA​GGC​AAT​ATC​A-3′; C1qa reverse primer 5′-TGG​TTC​TGG​TAT​GGA​CTC​TCC-3′; H2-eb1 forward primer 5′-GCG​GAG​AGT​TGA​GCC​TAC​G-3′; H2-eb1 reverse primer 5′-CCA​GGA​GGT​TGT​GGT​GTT​CC-3′; Ccl8 forward primer 5′-TCT​ACG​CAG​TGC​TTC​TTT​GCC-3′; Ccl8 reverse primer 5′-AAG​GGG​GAT​CTT​CAG​CTT​TAG​TA-3′; Il27 forward primer 5′-CTG​TTG​CTG​CTA​CCC​TTG​CTT-3′; Il27 reverse primer 5′-CAC​TCC​TGG​CAA​TCG​AGA-3′; Gapdh forward primer 5′-AGG​TCG​GTG​TGA​ACG​GAT​TTG-3′; Gapdh reverse primer 5′-TGT​AGA​CCA​TGT​AGT​TGA​GGT​CA-3′.

All data were normalized to Gapdh quantified in parallel amplification reactions.

Prior to sequencing, immune cells (minus B cells, CD45⁺CD19⁻) were sorted from omentum tumor samples (three mice/group) and pooled. Samples were uniquely barcoded using TotalSeq-B mouse hashtag antibodies (BioLegend), as per the manufacturer’s directions, to allow for sample multiplexing for the 10x Genomics Chromium Controller single-cell platform (10x Genomics). Specifically, 1–2 million cells of each sample were first blocked with TruStain FcX PLUS anti-mouse CD16/32 antibody and then incubated with 0.5 μg of various anti-mouse hashtag antibodies carrying unique cell barcodes. One 10x G chip lane was loaded with a pool of four uniquely barcoded samples and single-cell droplets were generated using the Chromium Next GEM single cell 3′ kit v3.1 (10x Genomics). cDNA synthesis and amplification, library preparation, and indexing were done using the 10x Genomics Library Preparation kit (10x Genomics), according to the manufacturer’s instructions. Overall library size was determined using the Agilent Bioanalyzer 2100 and the High Sensitivity DNA assay and libraries were quantitated using KAPA real-time PCR. One library consisting of a total of four samples was pooled and sequenced on the NextSeq 2000 (Illumina) using a P3 100 cycle kit (Illumina), paired end run with the following run parameters: 28 base pair × 8 base pair (index) × 90 base pair.

Pre-processing of the scRNA-seq data was performed using Cell Ranger Suite (pipeline v7.0.0, https://support.10xgenomics.com) with refdata-gex-mm10-2020-A transcriptome as a reference to map reads on the mouse genome (mm10) using STAR (Dobin et al., 2013). Cells with over 5% mitochondrial content were filtered out as were those with <200 genes with reads to remove cells with low quality and/or cells that are likely dying. The remaining 13534 cells were used for downstream analysis. Batch effect was not observed and hence not corrected for. Seurat v4 (Hao et al., 2021) was used for cell clustering, marker identification, and visualization. The R package SingleR (Aran et al., 2019) was used to determine the initial cell types of the clusters using the MouseRNASeq dataset as a reference for cell-specific gene signatures and then verified using known cell-type markers unique to clusters. The MΦ/monocyte clusters were subset and reclustered to identify subclusters of interest. The R package Slingshot (Street et al., 2018) was used for trajectory analysis of the MΦ/monocyte subclusters. Differential expression between samples in specific clusters was performed using the Wilcoxon Rank Sum Test. Statistically significant differentially expressed genes were used as inputs for enrichment analysis using Qiagen IPA. Raw data have been deposited onto Gene Expression Omnibus (GSE278380).

We obtained a scRNA-seq dataset deposited by Vázquez-García et al. from the National Center for Biotechnology Information Gene Expression Omnibus (GSE180661) (Vázquez-García et al., 2022). The dataset consisted of quality-filtered matrices of 929,686 cells. The Seurat package v4.3.0 (RRID:SCR_016341) in R software v4.3.1 (RRID:SCR_001905) and the scanpy package v1f.9.3 (RRID:SCR_018139) in Python 3.10 (RRID:SCR_008394) were used for downstream processing. Dimensionality reduction was performed using principal component analysis and uniform manifold approximation and projection (UMAP), the same protocol as in the original article. We used the cell type annotation assigned in the original article. The R package scCustomize v1.1.1 and the scanpy function score_genes were used to generate joint plots.

The KM Plotter Online tool (https://kmplot.com/analysis/) (Győrffy, 2024) was used to evaluate the relationship between high gene expression and clinical outcomes in patients with late-stage OvCa (stages III and IV). This open-access The Cancer Genome Atlas–based database contains bulk RNA sequencing datasets from 1,268 late-stage OvCa patients, which allowed us to investigate the correlation between overall survival and enriched genes identified by our scRNA-seq analysis in patients. To analyze the correlation between IL27/EBI3 co-expression and overall survival, we utilized the multiple-gene analysis and assigned equal weights to IL27 and EBI3.

The femur and tibia were harvested from 8-wk-old C57BL6, Syk^Mye∆, Dectin-1 KO, or IFNγR-KO mouse. Two ends of the bones were cut, and the bone marrow was flushed with a 26 g needle filled with cold sterile 1X PBS through a 40-μm cell strainer. 10 ml total PBS was used for all four bones. BMDMs were differentiated from total bone marrow cells by growing them in the presence of 20 ng/ml of CSF1 in RPMI with 10% FBS and 1x penicillin/streptomycin. Media was supplemented with 0.5 the volume of initial media+CSF1 and differentiation was complete by day 6. After differentiation, the media was changed and BMDMs were stimulated with 10 μg/ml β-glucan and 33 ng/ml IFNγ for 24–48-h. IL27 heterodimer and IL27p28 ELISAs were performed according to their manufacturer protocols (438707; Biolegend, and M2728; R&D Biotechne, respectively).

For T cell activation assay, we treated BMDM with 10 μg/ml β-glucan and 33 ng/ml IFNγ for 24 h. OT-I T cells (CD8⁺) were isolated from spleens of OT-I mice using EasySep Mouse CD8⁺ T Cell Isolation Kit (19853A; Stem Cell Technologies). DCs were isolated from spleens of WT mice using EasySep Mouse Pan-DC Enrichment Kit II (19863; Stem Cell Technologies). Co-cultures were set up in four replicates with BMDM primed as above, OT-I T cells, and DCs in the presence of OVA257-264 peptide (AS-60193-1; Ana Spec, Inc.) at 0.5 ng/ml for 3 days. DCs were added at a 1:10 ratio to T cells.

For staining surface markers, cells were incubated with a fluorescent-conjugated antibody cocktail for staining at 4°C for 30 min. For intracellular staining, cells were stimulated with 20 ng/ml of PMA, 1 μg/ml of ionomycin, 3 μg/ml of brefeldin A, and 2 μM of monensin and incubated at 37°C for 4 h. Cells were incubated with an antibody cocktail for surface markers at 4°C for 30 min and then fixed using a True nuclear transcription factor buffer set (Cat # 424401; Biolegend) for 20 min at 4°C. Cell pellets were incubated with an antibody cocktail for intracellular markers prepared in permeabilization buffer and incubated at 4°C for 30 min. All the antibodies were used at 1:400 dilution. Samples were then washed and resuspended in 1x PBS and acquired on BD FACS Symphony flow cytometer. Data were analyzed using FlowJo v10 (Treestar, Inc.).

Statistical analyses were performed in Prism (GraphPad Software, Inc.). Statistical tests used and other relevant details are noted in the figure legends. Statistical analysis was performed using the Student’s t test for unpaired samples or one-way ANOVA with a post-hoc Tukey’s multiple comparisons test. Results were considered significant at P < 0.05. Results display all replicated experiments and are presented as mean ± SEM.

Graphical abstract and Fig. S2 K were generated using https://Biorender.com.

Fig. S1 illustrates in vivo treatment regimens, staging of tumor pathology at time of treatment and clearance of ascites following β-glucan treatment. Fig. S2 provides additional support for β-glucan–mediated cancer capture including ID8 capture and clodronate-loaded liposome macrophage depletion experiments. Fig. S2 also provides an illustration describing the cancer capture mechanism discovered in this study. Fig. S3 includes additional tumor burden evaluations following BI treatment as well as an IDEXX analysis of BI toxicity. Fig. S4 analyzes MΦ subpopulations and progenitor cells following BI treatment. Fig. S4 also includes immune cell population clustering of human scRNA-seq data and expression of IL12A and IL12B in myeloid subclusters from human OvCa tumors. Fig. S5 shows expression of TNF and IFNg in CD4⁺ and CD8⁺ T cells following BI treatment with and without IL27 neutralization. Additionally, Fig. S5 shows TUNEL staining quantification in clots from BI-treated mice. Finally, Table S1 lists information for all antibodies used in this study.

The human sequencing data in Fig. 4 and Fig. S4 are openly available in the Gene Expression Omnibus (GSE278380). Remaining data are available from the corresponding author upon reasonable request. All the data needed to evaluate the conclusions of the paper are present in the paper or the online supplemental material.

We thank Wistar core facilities (imaging, histology, flow cytometry, animal facility, and genomics). We thank Drs. Chris Hunter and Li-Fan Lu for discussions on IL27 and Dr. Maureen Murphy for proofreading the manuscript.

This study was supported by National Institutes of Health (NIH) Career Enhancement Program through Hopkins-Penn-Wistar Ovarian Cancer Specialized Ovarian Cancer Research (P50CA228991; N. Zhang), Department of Defense Ovarian Cancer Research Program (HT94252410206; N. Zhang), W.W. Smith Charitable Trust (C2205; N. Zhang), National Institute of Allergy and Infectious Diseases (NIAID) (K99AI151198; N. Zhang), National Institute of Neurological Disorders and Stroke (1R01NS131912; F. Veglia), National Cancer Institute (NCI) (1R37CA280869; R. Shinde), NCI (1R21CA259240; R. Shinde), NIAID (5R21AI175738; L. Huang), Cancer Center Support Grant (P30CA010815; N. Zhang), NIH Wistar Training Program in Basic Cancer Research (T32 CA009171; B. Murphy), and Japan Society for the Promotion of Science (202360517; T. Miyamoto).

Author contributions: B. Murphy: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing - original draft, Writing - review & editing, T. Miyamoto: Conceptualization, Investigation, Methodology, Validation, Writing - original draft, Writing - review & editing, B.S. Manning: Formal analysis, Investigation, Visualization, G. Mirji: Formal analysis, Investigation, Visualization, Writing - review & editing, A. Ugolini: Investigation, T. Kannan: Data curation, Formal analysis, Visualization, K. Hamada: Data curation, Formal analysis, Writing - review & editing, Y.P. Zhu: Validation, D.T. Claiborne: Data curation, Resources, Supervision, Writing - review & editing, L. Huang: Writing - review & editing, R. Zhang: Resources, Supervision, Writing - review & editing, Y. Nefedova: Resources, Writing - review & editing, A. Kossenkov: Data curation, Formal analysis, Software, Validation, Visualization, Writing - review & editing, F. Veglia: Investigation, Writing - review & editing, R. Shinde: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Supervision, Validation, N. Zhang: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Validation, Visualization, Writing - original draft, Writing - review & editing.