FAK inhibition in ovarian cancer releases omega-3 fatty acids to program CXCL13-producing anti-tumor resident peritoneal macrophages

SUMMARY

High-grade serous ovarian cancer (HGSOC) is a lethal malignancy characterized by therapy resistance. Focal adhesion kinase (FAK) is highly expressed in HGSOC, yet its impact on tumor-immune communication remains incompletely defined. Using three syngeneic ovarian cancer models, we show that FAK inhibition (FAKi) increased macrophage CXCL13 expression and promoted peritoneal B cell infiltration. Combining FAKi with low-dose pegylated doxorubicin and anti-T cell immunoreceptor with Ig and ITIM domains (TIGIT) checkpoint blockade suppressed orthotopic ovarian tumor growth, extended survival, and induced tertiary lymphoid structures. Macrophage lineage factor GATA6 inactivation reduced CXCL13 expression, enhanced FAK-knockout tumor growth, and limited ascites B cell accumulation. Mechanistically, FAKi-treated or FAK-deficient tumor cells release exosomes enriched in omega-3 fatty acids that stimulated macrophage CXCL13 production. Exposure of macrophages to tumor-derived omega-3 lipids or eicosapentaenoic acid induced anti-tumor reprogramming and CXCL13 expression. Together, these findings reveal a tumor lipid-macrophage signaling axis activated by FAKi that supports B cell recruitment and anti-TIGIT immunotherapy.

In brief

FAK tyrosine kinase drives ovarian cancer tumor progression in part via effects on the tumor microenvironment. Chen et al. show that ovarian tumor FAK inhibition triggers release of omega-3 fatty acid-containing exosomes, impacting GATA6⁺ peritoneal macrophage anti-tumor reprogramming, CXCL13 cytokine production, and anti-TIGIT immunotherapy.

Graphical Abstract

INTRODUCTION

High-grade serous ovarian cancer (HGSOC) is the most common and aggressive type of ovarian cancer.¹ Surgery followed by platinum and paclitaxel chemotherapy is the standard of care, but many patients develop chemotherapy resistance and succumb to disease within 5 years.² Platinum-resistant recurrent disease is difficult to treat, and topoisomerase inhibitors such as topotecan or pegylated liposomal doxorubicin (PLD) may be used as recurrent therapies.³ Several immune checkpoint immunotherapies (ICIs) have been tested for HGSOC, but tumor-intrinsic properties support the generation of a strong immunosuppressive tumor microenvironment (TME), which limits therapy efficacy.⁴ Despite this, patients with HGSOC tumors containing elevated numbers of T cells, tumor-infiltrative B cells, and dendritic cells exhibit improved survival.⁵ Thus, a greater understanding of tumor-intrinsic signals driving immunosuppression may identify new combinatorial immunotherapeutic approaches.

Tumor-associated macrophages (TAMs) are the most abundant immune cells in HGSOC.⁶ TAMs possess important regulatory immune functions via phagocytosis of cellular debris, lipid accumulation, and cytokine secretion.⁷ Despite traditional macrophage roles being anti-tumoral and cytotoxic, TAMs also display considerable pro-tumor plasticity.^8,9 Analysis of TAM function in HGSOC requires consideration of distinct populations of either tissue-resident large peritoneal macrophages (LPMs) expressing lineage-determining markers such as the GATA6 transcription factor or recruited small peritoneal macrophages (SPMs), derived from differentiated bone marrow-derived myeloid cells.¹⁰ While cytotoxic chemotherapies transiently decrease total TAM populations, surviving TAMs may either support or inhibit immune surveillance.¹¹ Alterations in protein marker expression can indicate pro-tumor and anti-inflammatory or anti-tumor and pro-inflammatory TAM reprogramming.⁷ Although GATA6⁺ LPMs are considered immune guardians of the peritoneal cavity,^10,12 the role of GATA6⁺ macrophage inactivation in murine tumor models remains unclear.¹³

In addition to important roles for TAMs and T cells in peritoneal immune surveillance, increased B cell recruitment^14,15 and the formation of tertiary lymphoid structures (TLSs) are prognostic factors for better survival in copy-number-driven tumors such as HGSOC.^5,16,17 Structurally, TLSs are an ectopic collection of non-encapsulated B, T, and dendritic cells occurring at inflammation sites, including those in autoimmune disease and cancer.¹⁸ In general, patient tumors with TLSs show greater response to ICIs.¹⁹ Mechanistically, the dialogue of cytokine signaling impacting B cells and T cells in the TLS facilitates both T and B cell activation.¹⁶ Thus, the generation of B cell recruitment signals within an HGSOC immunosuppressive TME appears critical for adaptive T cell responses.²⁰ Cytokines released by tumor and stromal cells also play a significant role in the formation, maturation, and location of TLSs within tumors.²¹ In mice, CXCL13 knockout (KO) limits B cells in the peritoneal and pleural cavities²² and is important in lymph node development,²³ and exogenous CXCL13 addition with ICI antibody administration can limit ovarian tumor growth in a CD8⁺ T cell-dependent manner.²⁴ Notably, CXCL13 is part of a cytokine signature promoting TLS formation.²¹

The molecular mechanisms generating an HGSOC immunosuppressive TME are driven by genomic alterations.^25,26 HGSOC tumors contain a mutated p53 tumor suppressor, which is observed together with several recurring regional chromosomal gains and losses.²⁷ Chromosomal gains at 8q24.2-8q24.3 encompass the focal adhesion kinase (FAK) tyrosine kinase gene (PTK2). FAK gains are prognostic for decreased patient survival²⁸ and are prevalent in >75% of HGSOC tumors.²⁹ PTK2 gains are associated with elevated FAK mRNA and protein levels with enhanced FAK tyrosine phosphorylation.²⁸ In HGSOC patient tumors, elevated FAK mRNA levels are associated with expression of the CD112 and CD155 ligands for the TIGIT (T cell immunoreceptor with Ig and ITIM domains) immune checkpoint receptor.³⁰ Markers of FAK activation co-stain with CD155 in patient tumor biopsies, and inhibition of FAK activity reduces CD155 expression in vitro and in vivo.³⁰ We have developed an aggressive, chemotherapy-resistant C57BL/6 syngeneic ascites-producing ovarian tumor model termed KMF (KRas, Myc, FAK amplified) that contains DNA gains and losses in gene regions comparable to human HGSOC.^28,30 In the KMF model, genetic or small-molecule FAK inhibitor (FAKi; VS4718) regulated the CD155 ligand for the TIGIT receptor, and FAKi plus anti-TIGIT inhibition extended survival and elevated TME-associated CXCL13 levels in vivo.³⁰ However, the FAKi-stimulated cellular CXCL13 source and mediating signals remain unknown.

Herein, we use ifebemtinib, a FAKi, in clinical testing.³¹ We find that FAKi or genetic tumor FAK inactivation increased GATA6⁺ LPM expression of CXCL13 in vivo, associated with elevated B cell peritoneal infiltration in three independent syngeneic ovarian tumor models. Macrophage-specific GATA6 inactivation prevented FAK-KO-induced CXCL13 expression and reduced ascites B cells, leading to an increase in FAK-KO tumor growth compared to wild-type (WT)-GATA6-expressing mice. FAKi combined with low-dose PLD (chemotherapy) and anti-TIGIT (immunotherapy) significantly extended survival with ascites immune infiltration and omental tumor-associated TLS formation. The FAKi CXCL13-inducing signal present in tumor conditioned medium (CM) was 56°C heat stable and associated with elevated omega-3 fatty acid levels in exosomes released from FAK-KO and FAKi-treated WT KMF tumor cells. Notably, FAKi and the omega-3 fatty acid eicosapentaenoic acid (EPA) pushed murine LPMs toward an anti-tumor phenotype and stimulated human macrophages isolated from HGSOC patient ascites to express CXCL13. Together, we show that ovarian tumor FAK inhibition releases omega-3 fatty acids, triggering mouse GATA6 LPMs to produce CXCL13, which was associated with ascites B cell recruitment, and the strengthening of an anti-TIGIT immunotherapy response.

RESULTS

Ifebemtinib FAKi chemotherapy combinations in ovarian cancer

As pre-clinical studies have shown that FAKi combinations with paclitaxel or cisplatin can provide ovarian tumor control,^28,32,33 and as platinum with paclitaxel chemotherapy is the standard of care for HGSOC, we evaluated the effects of oral ifebemtinib (FAKi) and intraperitoneal (i.p.) cisplatin plus (+) paclitaxel (CPT) with or without FAKi (Figure 1A) on KMF ascites-associated tumor growth and peritoneal immune cell infiltration by lavage collection followed by flow cytometry (Figure S1; STAR Methods). Single-agent FAKi reduced tumor cell number by >90%, with CPT and FAKi + CPT exhibiting maximal >95% tumor inhibition at day 25 (Figure 1B). FAKi also slightly increased peritoneal B cell infiltration in tumor-bearing mice compared to the control (p = 0.042). However, whereas the combination of FAKi + CPT exhibited strong tumor control and increased T cells in the TME, this was not accompanied by elevated B cell, dendritic cell, or TAM infiltration within the peritoneal microenvironment (Figure 1B).

Figure 1. Ifebemtinib FAK inhibitor with pegylated doxorubicin combine to prevent ovarian tumor growth with enhanced B cell infiltration.

(A) Experimental schematic of KMF ovarian tumor ascites model with oral FAK inhibitor (FAKi) and intraperitoneal (i.p.) cisplatin + paclitaxel (CPT) chemotherapy.

(B) Flow cytometry of tumor and immune cells collected by peritoneal lavage at day 25, representing control (vehicle [Veh], blue), FAKi (red), CPT (green), and FAKi + CPT (purple) treatments. Cell populations were defined as tumor (CD45⁻), T cell (CD45⁺ TCRβ⁺), B cell (CD45⁺ B220⁺), dendritic cell (CD45⁺ CD11c⁺), and TAM (CD45⁺ CD11b⁺ F4/80⁺). Points are from individual mice (n = 9).

(C) Schematic of KMF tumor model treated with oral FAKi and i.p. pegylated doxorubicin (PLD).

(D) Flow cytometry of tumor and immune cells collected at day 25, representing control (blue), FAKi (red), PLD (green), and FAKi + PLD (purple) treatments. Points are from individual mice (n = 9).

(E and F) Quantitation of omental B cells (E) or omental T cells (F) from the KMF tumor experiment shown in (D). Points are from at least 8 sections from 3 mice per experimental group.

(G) Representative omental KMF tumor implant staining of B cells (purple), T cells (yellow), and cell nuclei (Hoechst, blue). Scale: 100 μm.

(B, D, F, and G) Values are means ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; one-way ANOVA with Tukey’s multiple comparisons test). (B) t test as noted (*p = 0.042).

To test other FAKi chemotherapy combinations on the TME, KMF tumor-bearing mice were treated with oral FAKi and/or low levels (0.15 mg/kg) of i.p. PLD (Figure 1C). The combination of ifebemtinib FAKi and PLD is being tested in a randomized phase 2 trial for platinum-resistant ovarian cancer (ClinicalTrials.gov: NCT06014528). As expected, FAKi reduced the tumor cell number by >90%, with PLD and PLD + FAKi exhibiting >95% reduction in peritoneal tumor cell number at day 25 (Figure 1D). Notably, FAKi, PLD, and FAKi + PLD individually and together significantly enhanced B and dendritic cell presence in peritoneal infiltrates of tumor-bearing mice (Figure 1D). In general, seeding KMF tumor cells in the peritoneal space allows for anchorage-independent tumorsphere growth with no primary tumor site. However, KMF cells are invasive within omental tissue, but control KMF tumor-bearing mice contain few omental B and T cells (Figures 1E–1G). Notably, increased B cell staining nearby omental tumor implants was detected in FAKi-treated mice (Figure 1E), and the combination of FAKi + PLD significantly increased omental T cell detection (Figures 1F and 1G). Taken together, while FAKi + CPT suppressed tumor growth and increased T cell infiltration, FAKi + PLD also controlled tumor growth and promoted B and dendritic cell accumulation, indicating that distinct FAKi immune-modulating effects may depend on the chemotherapeutic partner.

Transgenic FAK-KO ovarian tumor model

FAK inactivation prior to mouse adulthood can result in developmental lethality.³⁴ To study the effects of FAK KO in a transgenic model of ovarian cancer, FAK^fl/fl mice were crossed with mice expressing the SV40 T antigen (TAg⁺) under the control of the Müllerian inhibiting substance type II receptor³⁵ (Figure S2). Spontaneous ovarian tumors were isolated, and cells were expanded and genotyped as mouse ovarian carcinoma (MOVCAR) cells. Adenoviral (Ad) Cre recombinase was used to inactivate FAK expression ex vivo, and growth of FAK^fl/fl MOVCAR and Ad-Cre-treated MOVCAR FAK^−/− cells in culture was equivalent. However, intrabursal injection of FAK^fl/fl cells into syngeneic mice³⁵ yielded large and invasive tumors compared to FAK^−/− cells that remained as small clusters within the intrabursal ovary space. MOVCAR FAK^−/− tumors exhibited significantly elevated CD45⁺ immune cell staining compared to FAK^fl/fl tumors. In ascites at 12 weeks, decreased FAK^−/− tumor cell number was associated with significantly increased CD45⁺ immune cells compared to FAK^fl/fl tumor-bearing mice. Decreased FAK^−/− tumor burden was also associated with increased B cell infiltration and elevated CXCL9 and CXCL13 cytokine levels in peripheral blood. These results support the notion that tumor FAK loss may elicit signals that impact the immune TME (Figure S2).

Ifebemtinib FAKi and PLD increase macrophage CXCL13 expression

Our previously published studies showed that VS4718 FAKi increased Cxcl13 mRNA in non-tumor cells of the KMF TME.³⁰ To determine the FAKi-stimulated cellular source of CXCL13, flow cytometry and single-cell RNA sequencing (scRNA-seq) were performed on lavage samples from KMF tumor-bearing mice treated with ifebemtinib FAKi, PLD, or FAKi + PLD at day 25 (Figure 2A). In FAKi-treated mice, stimulated CXCL13 expression was highest in CD45⁺ CD11b⁺ F4/80^high TAMs compared to CD45⁺ CD11b⁺ F4/80^low TAMs or CD45⁺ CD11b⁻ F4/80⁻ myeloid cells (Figures 2B and 2C). Interestingly, PLD treatment also increased CD45⁺ CD11b⁺ F4/80^high TAM CXCL13 expression (Figure 2C).

Figure 2. Flow cytometry and single-cell sequencing of KMF peritoneal TME reveal FAKi- and PLD-induced myeloid CXCL13 expression.

(A) Schematic of KMF ovarian tumor model treated with oral FAKi and intraperitoneal PLD.

(B) Representative flow cytometry of FAKi-treated KMF tumor-bearing mice of CD45⁺ CD11b⁺ F4/80^high LPMs, CD45⁺ CD11b⁺ F4/80^low small peritoneal macrophages (SPMs), and CD45⁺ CD11b⁻ F4/80⁻ immune cells.

(C) Flow cytometry of CD45⁺ CD11b⁺ F4/80⁺ CXCL13⁺ TAMs from vehicle-, FAKi−, PLD−, and PLD + FAKi-treated mice bearing KMF tumors isolated at day 25. Values are the means ± SEM (*p < 0.05 and ***p < 0.001; one-way ANOVA with Tukey’s multiple comparisons test; n = 10).

(D) UMAP dimensionality reduction plot of single-cell RNA sequencing (scRNA-seq) performed on peritoneal tumor and immune cells collected at day 25. Shown are composite cell identifications denoted by color (legend at right) for all experimental groups.

(E) UMAP plot (single color) denoting CXCL13 mRNA-expressing cells in vehicle, FAKi, PLD, and FAK + PLD experimental groups at day 25. Scale is shown at left.

(F) Relative CXCL13 mRNA expression in B, myeloid, or T/NK cells in KMF tumor-bearing mice as determined by scRNA-seq and separated by experimental groups.

(G) UMAP plot of CD45⁺ myeloid cells as determined by scRNA-seq showing changes in cell subpopulations in vehicle, FAKi, PLD, and FAK + PLD experimental groups at day 25.

(H) UMAP plot (single color) of myeloid cells expressing Cxcl13, Gata6, and Timd4 mRNA in FAKi-treated experimental group at day 25. Scale is shown at right.

scRNA-seq analyses on lavage samples confirmed that FAKi, PLD, and FAKi + PLD increased TME B cells compared to vehicle control, as represented by uniform manifold approximation and projection (UMAP) dimensional reduction plots (Figure 2D). Cxcl13 mRNA expression was elevated in myeloid cells by FAKi, PLD, and FAKi + PLD treatment of tumor-bearing mice (Figure 2E; Table S1). Strong induction of Cxcl13 mRNA expression was detected in myeloid but not B or T/natural killer (NK) cell populations by FAKi, PLD, and FAKi + PLD (Figure 2F). UMAP sub-clustering of myeloid cells revealed population-level changes in mRNA expression upon FAKi, PLD, or PLD + FAKi treatment of KMF tumor-bearing mice (Figure 2G). Notably, FAKi-induced Cxcl13 expression partially overlapped with Gata6 and Timd4 mRNAs (Figure 2H), which are markers of LPMs.³⁶ Moreover, TCGA analyses revealed that HGSOC tumor mRNAs associated with CXCL13 include CXCL9 chemokine, TIGIT, and other T cell markers (Table S2). Taken together, our results support the notion that FAKi and PLD increase CXCL13 in TME TAMs with Gata6⁺ or Timd4⁺ LPMs as the putative cellular source.

FAKi with PLD strengthens anti-TIGIT immunotherapy and extends survival

Co-expression of CXCL13 and TIGIT occurs in HGSOC (Table S2, Spearman’s correlation: 0.767). As TIGIT functions as an immune checkpoint inhibitory receptor with elevated expression in exhausted T cells,³⁷ and active FAK controls tumor-associated CD155 ligand expression for TIGIT,³⁰ we tested whether including an anti-TIGIT inhibitory antibody as part of a FAKi + PLD chemo- and immunotherapy combination over 24 days impacted B and T cell TME recruitment (Figure 3A). Anti-TIGIT alone had negligible effects on KMF tumor cell number or changes in immune cell infiltration compared to IgG control (Figure 3B). Anti-TIGIT + PLD potently reduced tumor cell number and resulted in a TME that was enriched in B cells but not CD8⁺ T cells (Figure 3B). TIGIT + FAKi reduced tumor cell number at day 25, dramatically reduced CD155 TIGIT ligand on KMF cells, and increased B cells in the TME compared to the control. Interestingly, TIGIT + FAKi increased dendritic cell TME infiltration compared to anti-TIGIT alone (Figure 3B). TIGIT + FAKi + PLD resulted in tumor control, tumor CD155 reduction, elevated B cells, increased CD8⁺ T cells, and greater dendritic cell TME recruitment compared to single or double therapies (Figure 3B).

Figure 3. Combination FAKi and PLD chemotherapy strengthen anti-TIGIT immunotherapy and promote extended survival with TLS formation.

(A) Schematic of KMF ovarian tumor model testing combinations of FAKi, PLD, and anti-TIGIT. Treatments were stopped at day 24, and in separate experiments, cells were harvested and analyzed at day 25 (B) or mice were evaluated for survival (C).

(B) Flow cytometry of vehicle (V; blue), TIGIT (T; red), TIGIT + PLD (T+D; green), TIGIT + FAKi (T+F; purple), and TIGIT + PLD + FAKi (T+D+F; clear) experimental groups (n = 9). Cell populations were defined as tumor (CD45⁻ and CD155⁺), B cells (CD45⁺ B220⁺), CD8 T cells (CD45⁺ TCRβ⁺ CD8⁺), and dendritic cells (CD45⁺ CD11c⁺). Points are from individual mice (n = 9).

(C) KMF tumor-bearing mouse survival. Experimental groups (n = 10) match those in (B) with the addition of PLD only and PLD + FAKi.

(D) Quantitation of infiltrated B and T cell clusters within omentum at day 25. Points are from non-overlapping tumor sections, and immune cell values are presented as the percentage of the total area.

(E) Serial fixed sections from omental tumor implants stained by H&E and multiplex immunohistochemistry (IHC) with antibodies to B cells (B220, green), T cells (CD3, yellow), and CXCL13 (magenta). Scale bar: 100 μm.

(B, C, and E) Values are the means ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; one-way ANOVA with Tukey’s multiple comparisons test).

In independent experiments, KMF tumor-bearing mice were treated with the same 24-day protocol, treatment was stopped, and extended survival was measured (Figures 3A and 3C). All control mice reached a humane endpoint by day 35. Anti-TIGIT alone did not extend survival. PLD-only mice exhibited tumor regrowth that limited survival to 40–45 days, and PLD + FAKi-treated mice exhibited survival up to 50 days (Figure 3C). Significantly greater survival (>60 days) occurred in TIGIT + PLD + FAKi-treated mice (Figure 3C). In a second independent experiment, dual combinations TIGIT + FAKi and TIGIT + PLD yielded equivalent survival beyond 50 days (Figure 3C). However, TIGIT + PLD + FAKi treatment extended survival significantly longer than either dual drug combination (Figure 3C). Our results support future combinatorial testing of ICIs such as anti-TIGIT with FAKi + PLD chemotherapies.

TLS formation with FAKi, PLD, and anti-TIGIT treatments

The molecular mechanisms supporting TLS formation remain under investigation.¹⁹ As immune cell populations collected and analyzed in the peritoneal TME can differ from solid tumors, T and B cell staining of omental KMF tumor implants was performed (Figures 1G and S3). Control tumors had few immune cells, PLD alone increased T cell infiltration, FAKi treatment resulted in B cell-enriched sites within tumors, and PLD + FAKi promoted tumor-associated T and B cell small clusters (Figure 1G). Increased T and B cell infiltration of solid tumors with PLD + FAKi treatment correlated with increased survival (Figure 3C).

Although anti-TIGIT alone did not impact survival or T cell infiltration in the KMF model, the combination of TIGIT + PLD + FAKi resulted in high levels of omental immune cell infiltration and B-T cell clusters, many of which were mature TLSs formed by experimental day 25 (Figures 3D and 3E). Despite strong TIGIT + PLD anti-tumor effects by day 25 (Figure 3B), this combination increased omental T cell abundance (Figure S3) but not organized TLSs. Importantly, mouse survival beyond 60 days was associated with TLS formation upon TIGIT + PLD + FAKi treatment. Additionally, omental tumor staining detected F4/80⁺ CXCL13⁺ TAMs near tumor-associated clusters of B and T cells in tumor-bearing TIGIT + PLD + FAKi-treated mice (Figure 3E). Taken together, our results show that FAKi and PLD chemotherapy strengthen anti-TIGIT immunotherapy with enhanced B cell recruitment, survival, and TLS formation.

Tumor FAK expression and activity control macrophage CXCL13 expression and B cell TME recruitment

To better define the role of tumor-intrinsic FAK signals impacting the immune TME, we employed clonal KMF FAK-KO cells generated by CRISPR and made comparisons to KMF FAK-KO cells stably reconstituted with FAK-WT or a kinase-inactive FAK-K454R point mutant. FAK-WT and FAK-R454 proteins were equally expressed (Figure S3A), with FAK-WT cells producing significantly greater tumor burden than FAK-KO and FAK-R454 cells within 25 days (Figures S3B and S3C). scRNA-seq was performed on tumor and immune cells obtained by peritoneal lavage at day 25, and a notable UMAP difference between groups was the increased representation of B cells in the FAK-KO and FAK-R454 TME compared to the FAK-WT TME (Figure 4A). Interestingly, this pattern of B cell presence paralleled myeloid-associated CXCL13 expression in the FAK-KO and FAK-R454 but not the FAK-WT TME (Figure 4B). To extend these results, we used CRISPR to inactivate FAK or the related PYK2 kinase expression in murine HGS2 fallopian-tube-derived tumor cells from p53^−/− PTEN^−/− BRCA2^−/− mice (Figure S3D). Loss of FAK, but not PYK2, inhibited HGS2 syngeneic tumor growth (Figure S3E), and flow cytometry analyses showed that loss of FAK, but not PYK2, prevented HGS2 CD155 TIGIT ligand expression in vivo (Figure S3F). Analysis of CD45⁺ F4/80⁺ CD11b⁺ TAMs showed significantly increased CXCL13⁺ TAMs from FAK^−/− compared to PYK2^−/− and parental HGS2 tumor-bearing mice (Figure S3G). Taken together, three independent ovarian syngeneic tumor models (KMF, MOVCAR, and HGS2) link tumor FAK inactivation with decreased tumor burden, increased TAM CXCL13 production, and enhanced B cell TME infiltration.

Figure 4. Tumor FAK inhibition releases a heat-stable macrophage polarization and CXCL13-inducing factor into conditioned media.

(A) UMAP plot of scRNA-seq performed on peritoneal cells from KMF FAK-KO, FAK-WT, and FAK R454 tumor-bearing mice at day 25. Shown are composite cell identifications denoted by color (legend at right).

(B) UMAP plot (single color) denoting Cxcl13 mRNA-expressing cells in FAK-KO, FAK-WT, and FAK R454 tumor-bearing mice at day 25. Scale is shown at right.

(C) Schematic of transwell KMF tumor cell (bottom chamber) with primary C57BL/6 peritoneal macrophage (top chamber) co-culture signaling assay.

(D) Flow cytometry of peritoneal CD45⁺ CD11b⁺ F4/80⁺ macrophages for changes in GATA6 and CXCl13 expression in response to control (CTRL), FAK-KO, or KMF FAK-WT cell conditioned medium (CM).

(E) Quantification of CD11b⁺ F4/80⁺ CXCL13⁺ PMJ2-R macrophages by flow cytometry with the indicated heat-treated KMF cell CM incubations.

(F) Quantification of CD11b⁺ F4/80⁺ GATA6⁺ CXCL13⁺ PMJ2-R macrophages by flow cytometry with the indicated heat-treated HGS2 cell CM incubations.

(G) RNA-seq heatmap of elevated (red) or decreased (blue) peritoneal macrophage mRNAs after incubation with FAK-KO or FAK-WT cell CM.

(H) Volcano plot of differentially expressed macrophage mRNAs upon incubation with FAK-KO or FAK-WT CM.

(D–F) Values are the means ± SEM (*p < 0.05, ***p < 0.001, and ****p < 0.0001; one-way ANOVA with Tukey’s multiple comparisons test, n = 3).

Tumor FAKi releases a heat-stable CXCL13-inducing and macrophage anti-tumor reprogramming factor

To determine if FAK-KO-induced signals promoting TAM CXCL13 expression in vivo were mediated by cell-cell contact or via an indirect soluble mediator, KMF FAK-KO or FAK-WT cells were incubated with unstimulated C57BL/6 myeloid macrophages isolated by peritoneal lavage. FAK-KO but not FAK-WT co-culture stimulated macrophage (CD45⁺ CD11b⁺ F4/80⁺) CXCL13 expression, and this was reproduced in cell-separated transwells (Figure 4C). Interestingly, both FAK-KO and FAK-WT cells in transwells stimulated increased macrophage GATA6⁺ expression, but only FAK-KO cells boosted CXCL13 and increased double-positive GATA6⁺ CXCL13⁺ macrophages compared to signals generated by FAK-WT cells (Figure 4D). The mediator of macrophage CXCL13 induction was a soluble factor present in serum-free CMs of FAK-KO and FAK-R454 but not FAK-WT cells (Figure 4E). Importantly, heat treatment at 56°C for 30 min to denature tumor-released proteins and cytokines did not alter KMF FAK-KO or FAK-R454 CM-enhanced macrophage CXCL13 expression (Figure 4F). This tumor-associated heat-stable factor was also generated by pre-treating KMF FAK-WT or HGS2 cells with FAKi prior to CM collection, indicating that FAKi was “stimulating” the release of this reprogramming factor (Figures 4E, S5A, and S5B).

To determine if increased CXCL13 protein expression was associated with mRNA changes, primary C57BL/6 peritoneal macrophages were incubated with heat-treated KMF FAK-KO or FAK-WT CM and analyzed by bulk RNA-seq (Table S3), with selected log2 changes shown as a heatmap (Figure 4G). Importantly, differential expression analyses revealed that FAK-KO CM induced elevated levels of Cxcl13, Adgre1, Fgfr1, Itga6, Vsig4, Itgb1, and Lrg1 mRNAs (Figure 4H). Notably, these mRNAs were also part of the KMF tumor-bearing Cxcl13 myeloid population identified by scRNA-seq (Table S1). In total, 843 macrophage mRNAs were downregulated and 231 upregulated (log₂ fold change > 1) when comparing FAK-WT versus FAK-KO CM effects (Figure 4H). Anti-tumor macrophage markers (Gas6, CD80, CD68, and Ccl3) were upregulated in the presence of FAK-KO CM, while pro-tumor macrophage markers (Retnla, Chil3, Tgfb3, Col1a1, Col6a1, and Col12a1) were elevated in FAK-WT CM-treated macrophages. Fold changes in Cxcl13, Gas6, Col1A1, Col6a1, and Col12a1 were verified by real-time qPCR (Figure S6). In total, our results support the notion that soluble heat-stable signals released from FAK-inhibited tumor cells increase CXCL13 and push peritoneal macrophages toward an anti-tumor phenotype.

Macrophage GATA6 KO prevents FAK-KO-induced CXCL13 expression and augments FAK-KO tumor growth

As scRNA-seq showed that FAKi-associated macrophage Cxcl13 mRNA expression overlapped with Gata6 in myeloid-specific analyses (Figure 2H), we used LysM Cre macrophage-specific inactivation of GATA6^fl/fl mice to test the impact of GATA6 loss on KMF FAK-KO CM-stimulated CXCL13 expression (Figure 5A). LysM Cre⁺ GATA6^fl/fl mice exhibit a selective loss of resident LPMs but not macrophages in other organs.³⁶ Real-time qPCR confirmed GATA6 mRNA reduction from total peritoneal lavage-isolated macrophages from LysM Cre⁺ GATA6^fl/fl (GATA6-KO) compared to LysM Cre⁻ GATA6^fl/fl (GATA6-WT) mice (Figure 5B). FAK-KO but not FAK-WT CM stimulated CXCL13 expression from GATA6-WT but not GATA6-KO (CD45⁺ CD11b⁺ F4/80⁺) macrophages (Figures 5C and 5D). These results show the importance of GATA6 LPMs in stimulated CXCL13 expression.

Figure 5. Macrophage GATA6 knockout prevents FAK-KO-induced CXCL13 expression with augmented FAK-KO syngeneic tumor growth and fewer ascites B cells.

(A) Schematic of peritoneal macrophage isolation from LysM Cre⁻ GATA6^fl/fl YFP⁺ (GATA6-WT) and LysM Cre⁺ GATA6^fl/fl YFP⁺ (GATA6-KO) mice and stimulation with heat-treated FAK-KO or FAK-WT CM. Created in BioRender (https://BioRender.com/mpnfcap).

(B) Real time qPCR of Gata6 exons 2–3 from total peritoneal macrophages isolated from unstimulated GATA6-WT and GATA6-KO mice. Values are the means ± SEM (***p < 0.001 by unpaired t test).

(C) Representative flow cytometry of induced CXCL13 expression in CD45⁺ CD11b⁺ F4/80⁺ peritoneal GATA6-WT but not GATA6-KO macrophages by FAK-KO CM.

(D) Quantitation of FAK-KO CM-stimulated CXCL13 expression in GATA6-WT macrophages in vitro.

(E) 3 million KMF FAK-KO or FAK-WT cells were i.p. injected into GATA6-KO and GATA6-WT mice and monitored for tumor growth by IVIS (n = 5 per group).

(F) Quantitation of CD45⁺ CD11b⁺ CXCL13 expression from tumor experiment in (E).

(G) 5 million KMF FAK KO cells were i.p. injected in GATA6-KO and C57BL/6 mice and monitored for tumor growth by IVIS (n = 5 or 6 mice per group).

(H) Ascites CD45⁻ KMF FAK-KO quantitation from (G).

(I) Ascites TAM (CD45⁺ CD11b⁺ F4/80⁺) quantitation.

(J) Ascites B cell (CD45⁺ CD19⁺) quantitation.

(K) Ascites B1 cell subset (CD45⁺CD19⁺B220^lowCD43⁺) quantitation.

(L) Ascites B2 cell subset (CD45⁺CD19⁺B220^highCD43⁻) quantitation.

(D–F) Values are the means ± SEM (**p < 0.01 and ****p < 0.0001 by ANOVA with Tukey post hoc test). (G–L) Values are the means ± SEM (**p < 0.01, ***p < 0.001, and ****p < 0.0001 by unpaired t test).

To determine the effect of GATA6 loss on syngeneic tumor growth, KMF FAK-KO and FAK-WT tumor cells were i.p. injected into GATA6-WT and GATA6-KO mice (Figure 5E). As expected, KMF FAK-WT cells generated greater tumor burden than FAK-KO cells in both GATA6-WT and GATA6-KO mice. Whereas FAK-WT KMF cells grew equivalently in GATA6-KO and GATA6-WT mice, FAK-KO KMF cells exhibited enhanced growth in GATA6-KO mice (Figure 5E). Notably, GATA6-KO TAMs have low F4/80 surface expression, and it was the CD45⁺ CD11b⁺ F4/80^high MHC class II^low TAM population that produced CXCL13 in mice with KMF FAK-KO compared to FAK-WT tumors (Figures 5F and S7). These results support the notion that FAK-KO tumor growth may be limited by GATA6 macrophage production of cytokines such as CXCL13. To explore the role of GATA6 macrophage function further, KMF FAK-KO tumor cells were seeded in GATA6-KO and C57BL/6 mice, which revealed significantly enhanced tumor burden in GATA6-KO mice within 30 days (Figures 5G and 5H). Notably, ascites-associated TAMs and B cells were significantly reduced in GATA6-KO mice (Figures 5I and 5J). Surprisingly, the ratio of the ascites B1 subset was increased, whereas the ascites B2 cell subset was decreased, in GATA6-KO compared to C57BL/6 mice with FAK-KO tumors (Figures 5K and 5L). As B cells found in TLSs exhibit similar phenotypic and functional characteristics of a B2-type cell phenotype,³⁸ increased FAK-KO tumor growth in GATA6-KO mice is consistent with less B cell immune restraint.

CXCL13 induction by omega-3 fatty acid-enriched exosomes from FAK-KO CM

Building on the fact that the CXCL13-inducing factor in CM was heat stable, we analyzed intact cells (Figure 6A) and methanol-extracted CMs from KMF FAK-KO and FAK-WT cells by lipid mass spectrometry (Figure 6B; Tables S4 and S5). Several differences were detected, and α-linoleic acid (αLA), an omega-6 fatty acid, and EPA, an omega-3 fatty acid, were elevated in FAK-KO compared to FAK-WT CM (Figure 6B). EPA is a dietary-acquired omega-3 fatty acid that is heat stable, found in salmon and fish oil supplements, and known for its potential heart health benefits.³⁹ To determine if these lipids could impact C57BL/6 LPMs in vitro, purified αLA or EPA were independently added to FAK-WT CM and assayed for CXCL13-promoting activity (Figure 6C). EPA but not αLA addition potently induced CXCL13 expression in CD45⁺ CD11b⁺ F4/80⁺ macrophages. Using human THP-1 macrophages, EPA and the omega-3 fatty acid docosahexaenoic acid (DHA) significantly increased CXCL13 mRNA levels within 18 h (Figure 6D). These results suggest that omega-3 fatty acids may be signaling lipids in FAK-inhibited CM.

Figure 6. Exosome-associated omega-3 fatty acids from FAK-KO tumor cells stimulate macrophage CXCL13 expression.

(A) Heatmap of cellular lipid differences in FAK-KO and FAK-WT KMF cells as identified by mass spectrometry. C, control is serum-free cell media.

(B) Quantification of CM lipids from FAK-KO (blue circles) and FAK-WT (black circles) cells. Values (n = 3) are the fold change compared to the control. α-linolenic acid (αLA) and eicosapentaenoic acid (EPA) in FAK-KO media are highlighted in red.

(C) Quantitation of CXCL13 in CD45⁺ CD11b⁺ F4/80⁺ peritoneal macrophages upon stimulation with FAK-WT CM, FAK-WT CM with 0.01% purified αLA, with added EPA, and compared to FAK-KO CM.

(D) EPA and docosahexaenoic acid (DHA) but not αLA stimulate CXCL13 mRNA expression in THP-1 human macrophages.

(E) Schematic summary of exosome purification from serum-free FAK-KO and FAK-WT CMs by repeated ultracentrifugation. Created in BioRender (https://BioRender.com/qu75alu).

(F) Nanoparticle tracking profile of purified FAK-KO and FAK-WT exosomes.

(G) Percentage of CD45⁺ CD11b⁺ F4/80⁺ CXCL13⁺ murine peritoneal macrophages by flow cytometry after control (PBS) or increasing volumes of purified FAK-KO or FAK-WT exosomes.

(H) Heatmap differences of mass spectrometry lipidomic analyses of purified FAK-KO or FAK-WT exosomes. DHA and EPA are highlighted.

(C, D, and G) Values are the means ± SEM (**p < 0.01, ***p < 0.001, and ****p < 0.0001; one-way ANOVA with Tukey’s multiple comparisons test, n = 3).

In aqueous solutions, lipids can form structures such as micelles and bilayers. Exosomes are secreted lipid-bilayer-coated vesicles ranging in size from 30 to 150 nm that play important roles in cell-to-cell signaling within the TME.⁴⁰ Exosomes from FAK-KO and FAK-WT CMs were purified by repeated centrifugation steps, suspended in PBS, and analyzed by nanoparticle tracking (Figure 6E). Starting from the same number of cells, purified FAK-WT and FAK-KO exosomes had approximately the same final concentration, with particle size ranging from 50 to 300 nm (Figure 6F). Notably, FAK-KO but not FAK-WT exosomes induced CD45⁺ CD11b⁺ F4/80⁺ macrophages to make CXCL13 (Figure 6G) and contained elevated levels of DHA and EPA omega-3 fatty acids (Figure 6H; Table S6). Together, these results show that tumor FAK inactivation releases exosomes with an elevated level of omega-3 fatty acids that can stimulate peritoneal macrophages to make CXCL13.

CXCL13 induction by FAKi or EPA using cells from human ovarian tumor ascites

Abdominal paracentesis is used to relieve patients with ovarian cancer of malignant ascites, which is a mix of tumor and immune cells that can be evaluated ex vivo (Figure 7A; Table S6). Human macrophages are identified by different surface markers/anti-bodies (CD14 and CD68) compared to mice, and human macrophages can be polarized toward anti-tumor (CD86⁺) or pro-tumor (CD163⁺) phenotypes (Figure 7B). Incubation of total ascites cells with FAKi or EPA significantly increased (CD45⁺ CD14⁺ CD68⁺ CD86⁺) TAM markers and decreased (CD45⁺ CD14⁺ CD68⁺ CD163⁺) surface TAM markers (Figures 7C and 7D). The increase in (CD45⁺ CD14⁺ CD68⁺) TAMs was paralleled by increased CXCL13 expression in these cells (Figure 7E). Using gradient-purified TAMs from patient ascites, EPA increased CXCL13 mRNA (Figure 7F). EPA-stimulated CXCL13 was produced by CD45⁺ CD14⁺ CD68⁺ but not CD45⁺ CD14⁻ CD68⁻ TAMs (Figure 7G), with EPA and DHA omega-3 fatty acids showing equivalent CXCL13-inducing activities (Figure 7H). Taken together, our results show that FAKi effects on tumors and EPA effects on TAMs promote an anti-tumor murine and human peritoneal macrophage phenotype with increased CXCL13 expression.

Figure 7. Human ascites macrophages produce CXCL13 after FAKi-tumor or omega-3 fatty acid stimulation.

(A) Schematic of ovarian cancer patient ascites collection by paracentesis, total cell or gradient-purified macrophage isolation, and evaluation of effects of FAKi or EPA addition after 48 h. Created in BioRender (https://BioRender.com/c5fsm81).

(B) Simplified model of surface protein markers that define human macrophages (CD14⁺ and CD68⁺) and either anti-tumor (CD86⁺) or pro-tumor (CD163⁺) polarization. Created in BioRender (https://BioRender.com/qqyyx92).

(C) Flow cytometry quantitation of CD45⁺ CD14⁺ CD68⁺ CD86⁺ macrophages upon FAKi (1 μM) or purified EPA (50 μM) addition.

(D) Quantitation of CD45⁺ CD14⁺ CD68⁺ CD163⁺ macrophages as in (C).

(E) Quantitation of CD45⁺ CD14⁺ CD68⁺ CXCL13⁺ macrophages as in (C).

(F) Quantitative real-time RT-PCR of CXCL13 mRNA levels in EPA-stimulated purified patient macrophages.

(G) Representative flow cytometry gating of EPA-stimulated CD45⁺ CD14⁺ CD68⁺ macrophage CXCL13 expression.

(H) Percentage of CD45⁺ CD14⁺ CD68⁺ CXCL13-expressing macrophages upon EPA or DHA stimulation.

(G) (C–F and H) Values are the means ± SEM (**p < 0.01, ***p < 0.001, and ****p < 0.0001; one-way ANOVA with Tukey’s multiple comparisons test, n = 3). CTRL represents no addition to base media.

DISCUSSION

The small-molecule FAKi defactinib, in combination with the RAF-MEK inhibitor avutometinib, has received accelerated FDA approval for recurrent low-grade serous ovarian cancer.⁴¹ However, HGSOC is a molecularly different tumor with p53 mutations and FAK copy-number gains without RAS-MAPK pathway mutational activation. Moreover, pre-clinical tumor models show that small-molecule FAKi can impact both tumor and stromal cells of the TME.^42,43 Increased complexity associated with interpretations of FAKi action is due in part to different tumor types and the use of pre-clinical FAKi compounds that inhibit both FAK and the related PYK2 kinase.⁴³

Herein, we used ifebemtinib, a FAK but limited PYK2 inhibitor, that is being tested in combination with PLD in a phase 2 trial for recurrent platinum-resistant HGSOC (ClinicalTrials.gov: NCT06014528). Using multiple syngeneic ovarian tumor models, we revealed that tumor-intrinsic FAK loss, genetic FAK kinase inactivation, or oral ifebemtinib FAKi treatment of mice promoted B cell infiltration into the peritoneal TME. FAK-inhibited tumor cells released omega-3 fatty acid-enriched exosomes as a lipid-associated signal to resident peritoneal macrophages to make CXCL13, also known as B cell recruitment factor. Macrophage GATA6 lineage determination factor inactivation prevented CXCL13 expression in vivo, enhanced FAK-KO tumor growth, and reduced ascites B cell infiltration—mechanisms that are consistent with less TME immune restraint. Overall, our studies with mice and human tumor ascites revealed that ovarian tumor FAKi educates macrophages to express CXCL13—highlighting a potential therapeutic pathway linking FAKi, omega-3 fatty acid-containing exosomes, and macrophage reprogramming toward an anti-tumor phenotype.

Surprisingly, we found that FAKi immune-modulating effects are impacted by the chemotherapeutic partner. Whereas the combination of FAKi with cisplatin and paclitaxel provided strong tumor inhibition, beneficial effects on the TME were limited. However, combinations of FAKi and low levels of PLD exhibited tumor control and facilitated recruitment of B cells and dendritic cells into the peritoneal tumor TME. Interestingly, the addition of an anti-TIGIT ICI antibody did not inhibit tumor growth or alter ovarian TME by itself, but adding anti-TIGIT to FAKi + PLD chemotherapy resulted in maximal ovarian tumor inhibition; increased B cell, CD8⁺ T cell, and dendritic cell infiltration and TLS formation; and significantly enhanced survival. Our results support the hypothesis that low-level i.p. PLD chemotherapy serves as a localized danger signal; tumor FAKi alters exosome lipids, thereby impacting macrophage chemokine signaling; and checkpoint blockade (TIGIT) supports adaptive immune effector function.

CXCL13 plays a critical role in orchestrating T and B cell interactions essential for maturation of the immune response.^44,45 CXCL13 is recognized as a positive prognostic indicator in HGSOC,²⁴ and in this context, it was expected that the CXCL13 source would be follicular T cells.⁴⁶ Surprisingly, flow cytometry and single-cell sequencing show that GATA6⁺ tissue-resident LPMs are a FAKi-stimulated CXCL13 source in mice. This differs from other studies, which noted that ovarian tumor cells can produce CXCL13 upon CDK4/6 inhibition⁴⁷ and that FAKi alters cytokine expression from other tumors.⁴⁸ Changes in cytokine expression upon FAK inhibition were conserved between KMF and HGS2 ovarian tumor models, and increased levels of CXCL9 and CXCL13 could be detected in the peripheral blood of mice with MOVCAR FAK^−/− compared to FAK^+/+ tumors.

Tumor cells can uptake omega-3 fatty acids from the environment.⁴⁹ This is especially true for HGSOC, which typically spreads to the omentum, an adipocyte-rich tissue that contributes to circulating fatty acid levels.⁵⁰ Omega-3 fatty acids, such as DHA and EPA, are best known for supporting cardiovascular health, but EPA also possesses anti-tumor activity and is in clinical trials for colon cancer.⁵¹ Our studies add to the growing literature on the positive role of omega-3 fatty acids in supporting an immune-favorable TME.⁵² Overall, our studies define a tumor-to-macrophage signaling linkage via exosome lipids that reprogram macrophages and support immunotherapy.

Limitations of the study

Our analyses of HGSOC patient tumor samples collected by paracentesis were limited in sample number but revealed a potential tumor-to-macrophage signaling response that may be independent of prior chemotherapy treatments or a tumor genetic-limiting response. In addition, as GATA6 expression in human macrophages differs from that in mice,⁵³ our parallel findings with human ovarian TAMs in ascites support the notion that tumor-FAKi and EPA-macrophage signaling is a conserved link-age. Our results also complement earlier studies showing that GATA6 macrophages regulate antibody production through peritoneal B cells⁵⁴ and that CXCL13 is important for B cell homing, antibody production, and body cavity immunity.²² Accordingly, several unknowns exist in this signaling linkage, starting with the molecular mechanisms of FAKi-initiated release of omega-3-enriched exosomes, the potential lipid receptors on macrophages, and the transcriptional signals in mouse and human TAMs that increase CXCL13 mRNA. Future studies will address the role of B cells and CXCL13 in FAKi-associated phenotypes through depletion strategies to better determine their role in tertiary lymphoid formation or blood biomarkers for monitoring therapy effects. Moreover, our syngeneic mouse ovarian tumor models will provide insights into how FAKi impacts tumor bioactive lipid exosome content, signaling to GATA6⁺ macrophages, and regulation of B cell infiltration, with knowledge gained on how to re-normalize the ovarian TME.

RESOURCE AVAILABILITY

Lead contact

Requests for further information and materials should be directed to the lead contact, David D. Schlaepfer (dschlaepfer@health.ucsd.edu).

Materials availability

All materials used in this study are available from the lead contact with a completed materials transfer agreement.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice

All experimental mouse procedures were reviewed and approved by The University of California San Diego Institutional Animal Care and Use Committee (Protocol S07331). Female C57BL/6 mice (8–10 weeks old) were purchased from Charles River Laboratories. LysMcre GATA6^fl/fl YFP⁺ mice (GATA6 WT or KO) were kindly provided by Gwendalyn Randolph (Washington University). FAK^fl/fl mice were used as described.⁶² Transgenic mice expressing the transforming region of SV40 T antigen (TAg) under control of the Mullerian inhibitory substance type II receptor (MISIIR) gene promoter were from Denise Connolly (Fox Chase Cancer Center). MISIIR-TAg⁺ and MISIIR-TAg^Low mice were developed as described.^35,57 Female NSG (NOD scid gamma) mice (NOD.CgPrkdc^scid Il2rg^tm1Wjl/SzJ) were purchased from Jackson Labs. HGS2 fallopian tube p53^−/− PTEN^−/− BRCA2^−/− tumor cells (syngeneic with C57Bl6 mice) were from Francis Balkwill (Queen Mary University of London).⁶³ Primers used for mouse genotyping at listed (Key Resources).

Cells and culture

Murine KMF ovarian tumor cells were isolated and cultured as previously described.^28,55 KMF FAK knockout (FAK-KO) was created by CRISPR, isolated as a clone (KT13), and exome sequenced as described.²⁸ KT13 FAK-KO cells were stably reconstituted with wildtype (FAK-WT) or catalytically inactive (FAK-R454) FAK as GFP fusion proteins.²⁸ KMF cells were cultured in DMEM medium with 10% Fetal Bovine Serum (FBS) and 1% penicillin/streptomycin. The p53 tumor suppressor is weakly expressed in KMF cells, and sequencing revealed that the p53 transcript is altered.⁶² HGS2 cells were grown in Advanced DMEM/F-12 medium supplemented with 4% FBS, 1X insulin-transferrin-selenium, 100 ng/mL hydrocortisone, 20 ng/mL EGF, 1X penicillin-streptomycin and L-glutamine. THP-1 cells were cultured in RPMI 1640 medium plus 10% FBS and 1% penicillin/streptomycin. PMJ2-R cells were cultured in modified DMEM medium (ATCC, 30–2002) containing 4 mM L-glutamine, 4500 mg/L glucose, 1 mM sodium pyruvate, 1500 mg/L sodium bicarbonate and 10% FBS. All cell lines were maintained in a humidified incubator at 37°C with 5% CO2. Mouse ovarian carcinoma (MOVCAR) cells were isolated from ascites of tumor-bearing MISIIR-TAg⁺; FAK^fl/fl mice and expanded by growth in non-adherent conditions following methods as described.⁶⁴ For 2D growth, MOVCARs were seeded (3 × 10⁵ cells per well) in tissue culture-treated six-well plates (Costar). After 5 days, cells were phase-contrast imaged (Olympus CKX41), collected by trypsinization, and enumerated (ViCell XR, Beckman). For 3D anchorage independent growth, cells were seeded at 10,000 cells/ml in poly-HEMA-coated 24-well plates (Costar), and after 7 days, collected by centrifugation and enumerated (ViCell XR).

HGS2 FAK and Pyk2 knockout cell lines were generated using CRISPR/Cas9-mediated gene editing using plasmid designed and created by VectorBuilder Inc. Lentiviral plasmids to target murine FAK (Ptk2) were pRP[2CRISPR]-Puro-Cas9_D10A-U6>mPTK2[gRNA3] and pRP[2CRISPR]-Puro-Cas9_D10A-U6>mPTK2[gRNA4] with gRNA protospacer sequences GACTCACCT GGGTACTGGCA and ATACTCGTTCCATTGCACCA, respectively. Lentiviral plasmids to target murine PYK2 (Ptk2b) pRP[2CRISPR]-Puro-Cas9_D10A-U6>mPTK2b[gRNA1] and pRP[2CRISPR]-Puro-Cas9_D10A-U6>mPTK2b[gRNA2] with gRNA protospacer sequences GCCCATAGCATTCAGCCAGC and GCTGCACCCACAGATGACCG, respectively. Transfected cells were selected by puromycin treatment (1.5 μg/mL, 36 h), and single cell clones were isolated by flow cytometry sorting (FACSAria, BD Biosciences). The loss of FAK and PYK2 protein expression was verified by immunoblotting. Cell lines were authenticated by DNA sequencing, cell-specific assays, and verification of knockdown by protein immunoblotting. Routine mycoplasma testing, performed prior to stock cryopreservation and after 10 cell passages, was negative.

Human samples

Newly diagnosed or recurrent ovarian cancer patients (9 females) with ascites were consented and materials from paracentesis were collected by the Department of Pathology, University of California School of Medicine, Moores Cancer Center, USA under Institutional Review Board approved protocol (IRB 110805). All available patient samples were evaluated by experienced pathologists for confirmation of histological type and biomarkers as indicated (Table S6). Peritoneal malignant ascites was centrifuged at 300 × g for 10 min at 4°C. Cell pellets were treated with red blood cell lysis buffer (BioLegend, 420302) for 5 min at room temperature, followed by the addition of PBS to terminate the reaction. Intact cells were pelleted by centrifugation and resuspended in growth medium containing RPMI 1640 supplemented with 10% heat-inactivated FBS, 2 mM L-glutamine, and 1% penicillin-streptomycin.

Cell suspensions were washed by repeated centrifugation and filtered through a 70 μm cell strainer. Total cell counts were obtained using a ViCell XR automated hemocytometer (Beckman Coulter). Collected cells were either frozen, directly used for experiments, analyzed by flow cytometry, or purified by density gradient separation. Crude patient ascites cell suspensions, which contain both tumor and immune cells, were seeded in 6-well poly-HEMA-coated plates (Corning, 3471). Cells were cultured for 48 h, and 2 × 10⁶ cells per well were treated with either ifebemtinib FAK inhibitor (IN10018, 1 μM), eicosapentaenoic acid (EPA, 50 μM), or docosahexaenoic acid (DHA, 50 μM) for 48 h, and analyzed by flow cytometry.

Purification of human tumor associated macrophages

To isolate macrophages from patient ascites, a discontinuous Percoll gradient was prepared as previously described.⁶⁵ Briefly, 10 mL of 70% Percoll, 15 mL of 45% Percoll, 20 mL of 25% Percoll (containing the cell suspension), and 5 mL of PBS were layered sequentially into a 50 mL conical tube and centrifuged at 800 × g for 30 min at room temperature without braking. Macrophage-enriched cells were collected from the 25–45% Percoll interface. Fractions were washed with growth medium and centrifuged at 800 × g for 7 min. To further enrich for macrophages, cells from the 25–45% Percoll interface were plated in tissue culture dishes and incubated for 2 h at 37°C. Non-adherent cells were gently removed by washing twice with warm PBS, leaving a population enriched for adherent macrophages.

Sex as a biological variable

These studies focused on ovarian cancer, a disease that biologically affects females only. Accordingly, all in-vitro and in-vivo experiments involved female-derived cell lines or animal models where applicable. The TCGA-OV (The Cancer Genome Atlas Ovarian Cancer Collection) is a de-identified and publicly available patient dataset https://www.cancer.gov/ccg/research/genome-sequencing/tcga.

METHOD DETAILS

Immunoblotting

Cells in culture or peritoneal cells from tumor-bearing mice were collected by lavage, washed with cold PBS, whole cell protein lysates were made by RIPA Lysis Buffer (Pierce) addition, and lysates were clarified by centrifugation (16,000 x g, 5 min). Complete Mini ETDA-free Protease inhibitor cocktail (Millipore Sigma) and PhoSTOP phosphatase inhibitor cocktail (Millipore Sigma) were added Lysis Buffer prior to use. Total protein levels were determined using a bicinchoninic acid assay (Pierce), 25 μg of protein were resolved on Mini-Protean TGX precast gels (4–15% Tris/Glycine gel, BioRad) and transferred to polyvinylidene difluoride membranes using a TransBlot Turbo (Bio-Rad). Immunoreactive protein bands were detected using HRP-conjugated anti-mouse or anti-rabbit antibodies with Clarity Western ECL (enhanced chemiluminescence) substrate reagent and visualized using a ChemiDoc Imaging System (Bio-Rad).

Tumor growth and mouse survival

All animal experiments were performed in accordance with The Association for Assessment and Accreditation for Laboratory Animal Care guidelines and approved by the UCSD Institutional Animal Care and Use Committee (S07331). For tumor implantation studies, cells were transduced with pUltra-Chili-Luciferase (Addgene plasmid #48688) enabling bicistronic expression of dTomato and luciferase and sorted by flow cytometry for equivalent dTomato and GFP expression prior to mouse implantation. For the KMF experiments, unless otherwise indicated, 5 × 10⁶ pUltra-Chili-luciferase-labeled tumor cells were suspended in a 1:1 mixture of culture medium (without FBS) and Matrigel (Corning, 354262), and 200 μL were administered by intraperitoneal (IP) injection into 8- to 10-week-old female C57BL/6 mice (Charles River). Mice were randomized into experimental groups on day 5 after tumor cell injection and treated daily with ifebemtinib FAK inhibitor (IN10018, 25 mg/kg, InxMed) by oral gavage. Control group received saline injections. Additional groups were treated with chemotherapeutics including cisplatin (2 mg/kg) plus paclitaxel (10 mg/kg) (CPT), pegylated liposomal doxorubicin (Doxil; 0.15 mg/kg) weekly, or in combinations with the FAK inhibitor. Low endotoxin Anti-TIGIT monoclonal (1B4-mAb; 200 μg) or isotype control (Invivomab, mouse IgG1; Bio X Cell, BE0083) antibody were administered via IP injection at specified time points. Tumor growth was monitored by bioluminescent imaging (IVIS, PerkinElmer).

Tumor-bearing mice were sacrificed at indicated endpoints. Peritoneal tumor and immune cells were harvested by peritoneal lavage using 5 mL of ice-cold buffer (PBS supplemented with 2 mM EDTA and 2% BSA). Cells were processed with red blood cell lysis buffer (BioLegend, 420302), filtered through a 70 μm cell strainer, enumerated, assessed for viability (ViCell XR, Beckman Coulter), and analyzed by flow cytometry and single-cell RNA sequencing. Tumor cell numbers were calculated by multiplying the percentage of CD45⁻ cells by the total number of peritoneal cells collected. Murine omental and pancreatic tissues were harvested, processed through alcohol and formalin, and paraffin-embedded for histochemical staining. For survival analysis, mice were monitored daily after therapy cessation on day 24 for ascites tumor burden (appearance and body mass increase) and interference of normal behavior (feeding, grooming, lethargy) prior humane euthanasia as defined IACUC guidelines. Results presented are from one of at least two independent experiments.

MOVCAR orthotopic injection into the ovary-bursa space was performed as described.³⁵ Briefly, MOVCAR or HGS2 cells were mixed with growth factor reduced Matrigel (Corning), at a concentration of 0.5 × 10⁶ cells per 7 μL and injected into the surgically identified right ovary using a Hamilton syringe, 29.5-gauge needle, and dissecting microscope. Incisions were closed with surgical stables and mice were evaluated for health daily. High resolution ultrasound imaging (Vevo 2100, Visual Sonics) was used to monitor MOVCAR tumor growth in syngeneic TAg⁺ or NSG mice.

Flow cytometry

Cells were resuspended in flow staining buffer (PBS supplemented with 2 mM EDTA and 2% BSA) and incubated with anti-mouse CD16/32 (BD Biosciences, 553142) or Human TruStain FcX (BioLegend, 422302) Fc-blocking reagent for 15 min. For murine samples, surface staining was performed at 4°C for 30 min using the following antibodies: LIVE/DEAD Fixable Aqua Dead Cell Stain (Thermo Fisher Scientific, L34957), CD45 (BD Biosciences, 560510), B220 (BioLegend, 103255), CD19 (BioLegend, 152408), CD43 (BioLegend, 143217), TCRβ (eBioscience, 47-5961-82), CD4 (BioLegend, 100509), CD8 (BD Biosciences, 552877), MHC Class II (I-A/I-E; Invitrogen, 50-112-8850), CD11b (BioLegend, 101233), F4/80 (eBioscience, 45-4801-82), CD11c (BioLegend, 117334), CD206 (BioLegend, 141732), and CD155 (BioLegend, 131510). For intracellular staining, cells were fixed and permeabilized using the eBioscience Foxp3/Transcription Factor Staining Buffer Set (eBioscience, 00-5523-00), followed by staining with CXCL13 (eBioscience, 17-7981-82) and GATA-6 (Cell Signaling, 26452). Human peritoneal ascites samples were stained with antibodies against CD45 (BioLegend, 304041), CD14 (BioLegend, 301803), CD86 (BioLegend, 305441) and CD163 (BioLegend, 333613) followed by intracellular staining with CD68 (BioLegend, 333827) and CXCL13/BLC/BCA-1 (R&D Systems, IC8012A-100). Flow cytometry was performed using BD LSR Fortessa and Fortessa X-20 cytometers. Data were analyzed using FlowJo software (version 10).

Serum cytokine quantification

Tumor-bearing mice were exsanguinated by trans cardiac puncture and blood was allowed to clot at for 30 min at RT and centrifuged at 1,000 × g for 10 min to isolate serum. Cytokines and chemokines were quantified using the LEGENDplex Custom Mouse Panel 750 (BioLegend) according to the manufacturer’s instructions.

Multiplex immunofluorescence staining

Murine omental and pancreatic tissues from tumor-bearing mice were harvested, fixed in 10% neutral-buffered formalin, and paraffin-embedded. Tissue sections (5 μm) were baked at 60°C for 1 h, followed by deparaffinization in xylene (3 × 5 min) and rehydration through successive alcohols (2 × 100%, 2 × 95%, 2 × 70%) into distilled water. Antigen retrieval was performed using citrate-based Antigen Unmasking Solution (Vector Laboratories, H-3300) at 95°C for 30 min. Staining was performed using the Intellipath Automated IHC Stainer (Biocare Medical). Endogenous peroxidase activity was quenched using BLOXALL Blocking Solution (Vector Laboratories, SP-6000) for 10 min, followed by two washes in TBST (Tris-buffered saline with 0.1% Tween 20). Non-specific binding was inhibited with Blocker BLOTTO buffer (Thermo, 37530) for 10 min.

For multiplex staining with anti-CD3 and anti-B220, sections were incubated with anti-CD3 primary antibody (Abcam, ab16669; 1:500 dilution) for 1 h, followed by two TBST washes. Detection was performed using HRP-conjugated anti-rabbit polymer (Cell IDx, 2RH-50) for 30 min, and developed with Opal 570 fluorophore (Akoya Biosciences, FP1488001KT) for 10 min. Slides were washed twice in TBST before proceeding to the second round of antigen retrieval. The second antigen retrieval was performed using the same citrate-based buffer (Vector, H-3300) at 95°C for 30 min. After a 5-min Bloxall block and two TBST washes, sections were incubated with anti-B220 antibody (BD Biosciences, 553086; 1:200 dilution) for 1 h, followed by two TBST washes. Detection was achieved using HRP-conjugated anti-rat polymer (Cell IDx, 2AH-50) for 30 min and developed with Opal 690 fluorophore (Akoya Biosciences, FP1497001KT) for 10 min.

For multiplex staining with anti-B220, anti-CD3, anti-F4/80 and anti-CXCL13, sequential rounds of antigen retrieval and antibody staining were performed as described above. The following primary antibodies and detection reagents were used: anti-B220 (BD Biosciences, 553086, 1:200), anti-rat HRP Polymer (Cell IDX, 2AH-50) and Opal 520 (Akoya, FP1487001KT); anti-CD3 (Abcam, ab16669, 1:500), Anti-rabbit HRP Polymer (Cell IDX, 2RH-50), Opal 570 (Akoya, FP1488001KT); anti-F4/80 (Bio-Rad, MCA497BB, 1:50), anti-rat HRP Polymer (Cell IDX, 2AH-50), Opal 620 (Akoya, FP1495001KT); anti-CXCL13 (R&D, AF470, 1:400), anti-goat HRP Polymer (Cell IDX, 2GH-50), Opal 690 (Akoya, FP1497001KT). Each antibody was incubated for 1 h. Between each staining cycle, slides were washed twice in TBST. All antigen retrieval steps were performed in citrate-based buffer (Vector, H-3300) at 95°C for 30 min. Each cycle included peroxidase blocking (5 min) prior to primary antibody incubation. After the final staining cycle, sections were washed twice in TBST and twice in distilled water, followed by nuclear counterstaining with DAPI (1μg/mL, 15 min). Slides were mounted with VECTASHIELD Vibrance Antifade Mounting Medium (Vector Laboratories, H-1700-10) and cover slipped. Whole-slide images were acquired using the PhenoImager Fusion imaging system (Akoya Biosciences).

For MOVCAR tumor staining, FITC-conjugated anti-CD45 antibodies (Invitrogen) at 1 mg/mL in 5% BSA and PBS were incubated for 2 h with FITC-conjugated IgG2b isotype antibodies (Invitrogen) as a negative control. Cell nuclei were visualized by incubation with Heochst 33342 (Invitrogen). Images were sequentially captured at an inverted microscope (IX81; Olympus) at 40X using Hamamatsu ORCA-AG camera, pseudo-colored, overlaid, and merged using Photoshop software. Percent CD45⁺ staining expressed as percent of nuclei using ImageJ. For MOVCAR analysis of spontaneous lung metastasis.

TLS identification

Tertiary lymphoid structures (TLSs) in the omentum of ovarian tumor-bearing mice were identified by H&E staining and multiplex immunofluorescence on consecutive tissue sections. TLSs were characterized by the clustered presence of T cells (CD3⁺) and B cells (B220⁺), and their numbers were quantified within defined omental areas. Quantitative image analysis was performed using QuPath (version 0.5.1). Regions of interest (ROIs) (minimum of 3 and maximum of 6 ROIs per tissue section) were manually annotated using the software’s annotation tool. Each antibody and DAPI nuclear stain were pseudo colored, and signal intensities were equalized across different slides by thresholding. Cells were segmented within ROIs based on their nuclear DAPI expression. Single marker measurement classifier was created for each antibody and quantification of positive detection in each ROI was exported. Data are presented as percentage of positive detections to all cells in each ROI, mean of the group +− SEM from at least two individual mice experimental group.

Murine peritoneal macrophage isolation

Female C57BL/6 mice (8–12 weeks old) or LysMcre GATA6^fl/fl YFP⁺ mice (GATA6 WT or KO) were euthanized in accordance with Institutional Animal Care and Use (IACUC) guidelines. Macrophages were isolated by injecting 5 mL of cold sterile PBS into the peritoneal cavity, the abdomen was gently massaged for 30 s to dislodge resident cells, and peritoneal fluid was withdrawn and transferred into sterile 50-mL tubes on ice. Peritoneal fluids from 3 to 5 mice were pooled to ensure sufficient cell yield. Typically, 2–3 × 10⁶ peritoneal cells per mouse was recovered. Cells were pelleted by centrifugation at 400 × g for 10 min at 4°C, the supernatant was discarded, and the cell pellet was resuspended in RPMI-1640 medium supplemented with 10% FBS and 1% penicillin-streptomycin. Cells were seeded onto tissue culture plates and incubated at 37°C (5% CO2) for 2 h to allow macrophage adherence. Non-adherent cells were removed by washing 3x with warm PBS and the remaining adherent cells were >90% macrophages. These murine peritoneal macrophages cells were subsequently used for downstream applications.

Conditioned medium collection

Ovarian cancer cell lines (KMF or HGS2) were cultured in their respective complete growth media under standard conditions (37°C, 5% CO₂). Upon reaching ~70% confluency, cells were washed once with PBS, and the medium was replaced with serum-free medium. After 48 h, the conditioned medium (CM) was collected, centrifuged at 300 × g for 5 min to remove cell debris, and filtered through a 0.22-μm filter (Millipore, SE1M179M6) to ensure sterility. The filtered CM was heat-inactivated at 56°C for 30 min in 1 mL aliquots, cooled on ice, and stored at −20°C until further use.

Transwell assay

Primary peritoneal macrophages were isolated and seeded into the upper chamber of transwell inserts (8-μm pore size) with KMF FAK-WT or FAK-KO tumor cells in the lower chamber. After 72 h of co-culture, macrophages were harvested and analyzed by flow cytometry. Macrophages were identified as CD45⁺CD11b⁺F4/80⁺ and assessed for expression of GATA6, CXCL13, or double-positive expression by flow cytometry.

Peritoneal macrophage RNA sequencing

Peritoneal macrophages (PMs) were isolated and cultured in growth medium (RPMI-1640 supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin). After 24 h, 1:1 volume of tumor cell CM and fresh PM growth medium were added. After 72 h, PMs were harvested, and total RNA was extracted using the PureLink RNA Purification Kit (Invitrogen, 12183018A) following the manufacturer’s protocol. RNA library preparation and sequencing was performed by Novogene Inc. (Beijing, China). RNA library preparation was performed using Abclonal mRNA-seq Lib Prep Kit for Illumina (RK20302) and samples were paired-end 150-bp sequenced using NovaSeq X Plus Series (PE150) platform generating >50 million reads per sample. Clean reads (>95%) were aligned to the reference genome using HISAT2. Novogene data pipeline used ClusterProfiler software for enrichment analysis, including GO Enrichment, DO Enrichment, KEGG and Reactome database enrichment to analyze and visualize functional profiles of genomic coordinates, genes and gene clusters. Novogene performed differential expression analysis of two conditions/groups by using the DESeq2-R software. Clustering and grouping analyses used transcripts with FPKM values > 1 and an adjusted p value <0.05. Each dataset was subject to Gene Set Enrichment Analysis (GSEA) and The Molecular Signatures Database (MSigDB) analysis.

Further differential expression analysis and visualization were performed using Python tools: Expression data were imported from spreadsheets using Pandas, and DEG were filtered based on log₂ fold change (log₂FC) and p-values. Thresholds were set at log₂FC > 0.5 and p < 0.05. Volcano plots were generated using matplotlib, with selected genes of interest highlighted and annotated to enhance biological interpretation. AdjustText was used to improve label clarity where necessary. A heatmap comparing WT and KO samples was created using Matplotlib with Seaborn, with custom color gradient via Matplotlib.colors.

Single-cell RNA sequencing

Tumor and immune cells were isolated by peritoneal lavage at experimental day 24 and subjected to single cell 3’ RNA sequencing using the 10x Genomics Chromium Next GEM Single Cell 3’ HT v3.1 platform. Cell barcoding, reverse transcription, cDNA amplification, and library construction were performed according to the manufacturer’s protocol. Sequencing was carried out on an Illumina NovaSeq 6000 platform, generating approximately 1 billion reads per sample. Demultiplexed FASTQ files were processed using Cell Ranger (v6.1.2, 10x Genomics) for unique molecular identifier (UMI) quantification and alignment to the mm10 mouse reference genome.

Downstream analysis was conducted in R (v4.1.2) using Seurat (v4.3.0) software. Cells were filtered based on the following quality control criteria: 250–10,000 detected genes per cell, 1,000–100,000 UMIs, <20% mitochondrial gene content, and <40% ribosomal gene content. Putative doublets were removed using DoubletFinder (v2.0.3). Data normalization and variance stabilization were performed using SCTransform with regularized negative binomial regression to enhance biological signal. Dimensionality reduction was performed using principal component analysis (PCA), and an integrated shared nearest neighbor (SNN) graph was constructed for clustering. Clusters were identified using the modularity optimization-based SNN clustering algorithm (40 principal components, resolution = 0.4). Differential gene expression analysis across clusters was conducted using the FindAllMarkers function in Seurat, applying the default Wilcoxon rank-sum test. Initial cell type annotation was performed using the ScType method, followed by manual curation based on canonical marker expression.

Raw fastq files were aligned to the prebuilt mouse reference genome (refdata-gex-mm10-2020-A) and quantified using CellRanger v7.1.0. Seurat v5.3.0 was used for downstream analysis. All samples were quality filtered to remove cells with less than 250 features and greater than 10,000 features, overall counts <40,000, percent mitochondrial reads <20% and percent ribosomal reads <40%. Data were transformed using SCTransform with percent mitochondrial reads and percent ribosomal reads as covariates. Samples were integrated and clusters were identified with UMAP visualization at a resolution of 0.8. Doublets were filtered out using DoubletFinder. Cell types were defined by marker genes and informed by predictions from SingleR using the mouse immune cell reference (ImmGenData). Genes of interest were overlaid on the UMAP visualizations with FeaturePlot. Violin plots were made for genes of interest with the VlnPlot function.

Quantitative RT-PCR

Total RNA was extracted using the PureLink RNA Mini Kit (Invitrogen, 12183018A) according to the manufacturer’s protocol. cDNA was synthesized using the iScript Reverse Transcription Supermix for RT-qPCR (Bio-Rad, 1708841). Quantitative PCR was performed using iTaq Universal SYBR Green Supermix (Bio-Rad, 1725121) with gene-specific primers and cDNA templates on a CFX Opus 96 Real-Time PCR Systems (Bio-Rad). Gene expression levels were normalized to GAPDH as a housekeeping gene control. Primer sequences used are listed in key resources table.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Alexa Fluor® 700 Rat Anti-Mouse CD45	BD Biosciences	Cat# 560510; RRID AB_1645208
APC anti-mouse CD155 (PVR) Antibody	BioLegend	Cat# 131510; RRID: AB_10645507
LIVE/DEAD™ Fixable Aqua Dead Cell Stain Kit	Thermo Fischer	Cat# L34957
Brilliant Violet 711™ anti-mouse/human CD45R/B220 Antibody	BioLegend	Cat# 103255; RRID: AB_2563491
PE anti-mouse CD19 Antibody	BioLegend	Cat# 152408; RRID: AB_2629817
PE/Dazzle™ 594 anti-mouse CD43 Antibody	BioLegend	Cat# 143217; RRID: AB_2800664
TCR beta Monoclonal Antibody (H57-597), APC-eFluor™ 780	eBioscience	Cat# 47-5961-82; RRID: AB_1272173
FITC anti-mouse CD4 Antibody	BioLegend	Cat# 100509; RRID: AB_312712
BD Pharmingen™ PE-Cy™7 Rat Anti-Mouse CD8a	BD Biosciences	Cat# 552877; RRID: AB_394506
Brilliant Violet 605™ anti-mouse CD11c Antibody	BioLegend	Cat# 117334; RRID: AB_2562415
Brilliant Violet 570™ anti-mouse/human CD11b Antibody	BioLegend	Cat# 101233; RRID: AB_10896949
F4/80 Monoclonal Antibody (BM8), PerCP-Cyanine5.5	eBioscience	Cat# 45-4801-82; RRID: AB_914345
CXCL13 Monoclonal Antibody (DS8CX13), APC	eBioscience	Cat# 17-7981-82; RRID: AB_2762702
GATA-6 (D61E4) XP® Rabbit mAb (PE Conjugate)	Cell Signaling	Cat# 26452; RRID: AB_2798924
MHC Class II (I-A/I-E) Monoclonal Antibody (M5/114.15.2)	eBioscience	Cat# 50-112-8850
PE/Dazzle™ 594 anti-mouse CD206 (MMR) Antibody	BioLegend	Cat# 141732; RRID: AB_2565932
Brilliant Violet 605™ anti-human CD45 Antibody	BioLegend	Cat# 304041; RRID: AB_2562105
FITC anti-human CD14 Antibody	BioLegend	Cat# 301803; RRID: AB_314185
Brilliant Violet 421™ anti-human CD68 Antibody	BioLegend	Cat# 333827; RRID: AB_2800881
Brilliant Violet 785™ anti-human CD86 Antibody	BioLegend	Cat# 305441; RRID: AB_2616793
PE/Cyanine7 anti-human CD163 Antibody	BioLegend	Cat# 333613; RRID: AB_2562640
Human CXCL13/BLC/BCA-1 APC-conjugated Antibody	R&D Systems	Cat# IC8012A-100
Human TruStain FcX™ (Fc Receptor Blocking Solution)	BioLegend	Cat# 422302; RRID: AB_2818986
BD Pharmingen™ Purified Rat Anti-Mouse CD16/CD32 (Mouse BD Fc Block™)	BD Biosciences	Cat# 553142; RRID: AB_394657
Anti-TIGIT [1B4] VivopureX	Absolute Antibody	Cat# Ab01258-1.1-VXB
InVivoMAb mouse IgG1 isotype control	Bio X Cell	Cat# BE0083; RRID: AB_1107784
Anti-FAK, clone 4.47	Millipore Sigma	Cat# 05-537; RRID: AB_2173817
Phospho-FAK (Tyr397) Monoclonal Antibody (141-9)	Thermo Fischer	Cat# 44-625G; RRID: AB_1500096
PYK2 (5E2) Mouse mAb	Cell Signaling	Cat# 3480; RRID: AB_2174093
E-Cadherin (24E10) Rabbit mAb	Cell Signaling	Cat# 3195; RRID: AB_2291471
N-Cadherin (13A9) Mouse mAb	Cell Signaling	Cat# 14215; RRID: AB_2798427
Beta Actin Monoclonal antibody	Proteintech	Cat# 60008-1-Ig; RRID: AB_2289225
β-Tubulin (9F3) Rabbit mAb	Cell Signaling	Cat# 2128; RRID: AB_823664
Anti-CD3 epsilon antibody [SP7]	Abcam	Cat# ab16669; RRID: AB_443425
BD Pharmingen™ Biotin Rat Anti-Mouse CD45R/B220	BD Biosciences	Cat# 553086; RRID: AB_394616
Mouse CXCL13/BLC/BCA-1 Antibody	R&D Systems	Cat# AF470; RRID: AB_355378
Rat anti Mouse F4/80:Biotin	Bio-Rad	Cat# MCA497BB; RRID: AB_323893
UltraPolymer Goat anti-Rabbit IgG (H&L) – HRP	Cell IDx	Cat# 2AH-015
UltraPolymer Goat anti-Rat IgG (H&L) – HRP	Cell IDx	Cat# 2AH-050
UltraPolymer Donkey anti-Goat IgG (H&L) – HRP	Cell IDx	Cat# 2GH-050
VIMPCS - murine plasma mimetic medium	Wisent Bioproducts	Cat# 319-268-cL
IgG2b isotype antibodies	Invitrogen	Cat# 14-4031-82
Antigen Unmasking Solution	Vector Laboratories	Cat# H-3300
BLOXALL Blocking Solution	Vector Laboratories	Cat# SP-6000
BLOTTO blocking buffer	Thermo Fischer	Cat# 37530
HRP-conjugated anti-rabbit polymer	Cell IDx	Cat# 2RH-50
Opal 570 fluorophore	Akoya Biosciences	Cat# FP1488001KT
citrate-based buffer	Vector	Cat# H-3300
HRP-conjugated anti-rat polymer	Cell IDx	Cat# 2AH-50
Opal 690 fluorophore	Akoya Biosciences	Cat# FP1497001KT
Opal 520 fluorophore	Akoya Biosciences	Cat# FP1487001KT
Opal 620 fluorophore	Akoya Biosciences	Cat# FP1495001KT
VECTASHIELD	Vector Laboratories	Cat# H-1700-10
Growth factor reduced Matrigel	Corning	Cat# 354262
Bouin’s solution	Sigma	Cat# HT10132
Red blood cell lysis buffer	BioLegend	Cat# 420302
Bacterial and virus strains
Ad-Cre-GFP	Vector Biolabs	Cat# 1700
Biological samples
Human ascites from paracentesis	This paper	N/A
Chemicals, peptides, and recombinant proteins
RPMI 1640	Thermo Fischer	Cat# 11875093
DMEM	Thermo Fischer	Cat# 11995065
Modified DMEM	ATCC	Cat# 30-2002
Advanced DMEM/F-12	Thermo Fischer	Cat# 12634010
Fetal bovine serum	R&D Systems	Cat# S12450
Insulin/transferrin/selenium	Invitrogen	Cat# 51300
hydrocortisone	Sigma	Cat# H0135
Murine Epidermal Growth Factor (EGF)	Sigma	Cat# E4127
RBC Lysis Buffer (10X)	BioLegend	Cat# 420302
RIPA Lysis and Extraction Buffer	Thermo	Cat# 89900
Mouse IL-4 Recombinant Protein	PeproTech	Cat3 214-14
Lipopolysaccharides (LPS)	Sigma-Aldrich	Cat# L6529
Mouse M-CSF Recombinant Protein	PeproTech	Cat3 315-02
Eicosapentaenoic Acid (EPA)	MCE	Cat# HY-B0660
Linolenic Acid (LA)	Thermo Fischer	Cat# 21504
Docosahexaenoic Acid (DHA)	Cayman Chemical	Cat# 90310
PROTAC Degrader FC11	Tocris Bioscience	Cat# 7306
DOXOrubicin HCI Liposome Injection	Dr. Reddy’s	NDC 43598-0283-35
Paclitaxel	Pfizer	NDC 61703-0342-09
Cisplatin	West-Ward	NDC 0143-9504-01
Foxp3/Transcription Factor Staining Buffer Set	eBioscience	Cat# 00-5523-00
Antigen Unmasking Solution, Citrate-Based	Vector Laboratories	Cat# H-3300-250
BLOXALL® Endogenous Blocking Solution	Vector Laboratories	Cat# SP-6000-100
Blocker™ BLOTTO in TBS	Thermo Fisher	Cat# 37530
Opal 570 Reagent Pack	Akoya Biosciences	Cat# FP1488001KT
Opal 690 Reagent Pack	Akoya Biosciences	Cat# FP1497001KT
Opal 520 Reagent Pack	Akoya Biosciences	Cat# FP1487001KT
Opal 620 Reagent Pack	Akoya Biosciences	Cat# FP1495001KT
VECTASHIELD Vibrance® Antifade Mounting Medium	Vector Laboratories	Cat# H-1700-10
Ifebemtinib (IN10018) FAK Inhibitor	InxMed	clinical compound
IVISbrite D-Luciferin Potassium Salt Bioluminescent Substrate	Revvity	Cat# 122799
Corning® Matrigel® Matrix High Concentration (HC), Phenol Red Free, LDEV-free	Corning	Cat# 354262
Heochst 33342	Thermo Fischer	Cat# 62249
Mini ETDA-free Protease inhibitor cocktail	Sigma	Cat#11836170001
PhoSTOP™ phosphatase inhibitor cocktail	Millipore	Cat# 4906845001
Clarity Western ECL	BioRad	Cat# 1705060S
Critical commercial assays
eBioscience™ Foxp3/Transcription Factor Staining Buffer Set	eBioscience	Cat# 00-5523-00
LEGENDplex Custom Mouse Panel 750	BioLegend	Cat# 900001850
PureLink RNA Mini Kit	Invitrogen	Cat# 12183018A
iScript Reverse Transcription Supermix for RT-qPCR kit	Bio-Rad	Cat# 1708841
iTaq Universal SYBR Green Supermix	Bio-Rad	Cat# 1725121
Abclonal mRNA-seq Lib Prep Kit	Illumina	Cat# RK20302
Transwell chambers (8 μm)	Costar	N/A
0.22-μm filter	Millipore	Cat# SE1M179M6
Pierce™ BCA Protein Assay Kits	Thermo Fischer	Cat# 23225
Mini-PROTEAN® TGX™ Precast Gels	BioRad	Cat# 4561033
Corning™ Costar™poly-HEMA-coated 24-well plates	Thermo Fischer	Cat# 09-761-146
Deposited data
Mouse bulk RNA sequencing of macrophage	This paper	GEO: GSE320157
Mouse ovarian cancer single cell RNA sequencing	This paper	GEO: GSE313445
Experimental models: Cell lines
KMF (ID8-IP)	Ward et al.55	Schlaepfer Lab
THP-1	ATCC	Cat# TIB-202
HGS2	Ximbio	Cat# 160538
PMJ2-R	ATCC	Cat# CRL-2458
293T	ATCC	Cat# CRL-3216
MOVCAR	This paper	N/A
Experimental models: Organisms/strains
Mouse: C57BL/6	Charles River Laboratories	C57BL/6 Mouse
Mouse: FAK fl/fl	Shen et al.56	N/A
Mouse: TAg	Connolly et al.57	N/A
Mouse: LysMcre GATA6fl/fl YFP+ mice	Gautier et al.36	N/A
Oligonucleotides
Mouse GATA6 Forward CAGCAGGACCCTTCGAAAC Reverse CCATTCATCTTGCTGTAGAGACC	This paper	N/A
Human CXCL13 (BCA1) Forward TCTCTGCTTCTCATGCTGCT Reverse TCAAGCTTGTGTAATAGACCTCCA	This paper	N/A
Human GAPDH Forward GTCTCCTCTGACTTCAACAGCG Reverse ACCACCCTGTTGCTGTAGCCAA	This paper	N/A
Mouse GAPDH Forward CATCACTGCCACCCAGAAGACTG Reverse ATGCCAGTGAGCTTCCCGTTCAG	This paper	N/A
Mouse CXCL13 Forward CTCCAGGCCACGGTATTCTG Reverse CCAGGGGGCGTAACTTGAAT	This paper	N/A
Mouse GAS6 Forward ACAGGCTCAACTACACCCGAACAT Reverse TGACGGGTGCAGAAATCACCGATA	This paper	N/A
Mouse Col1a1 Forward GGGGCAAGACAGTCATCGAA Reverse GAGGGAACCAGATTGGGGTG	This paper	N/A
Mouse Col6a1 Forward AGGGCTACAAGGAACCATGC Reverse GGTATGTGTGGTCTGTGGCA	This paper	N/A
Mouse Col12a1 Forward CCGTACAATGGGCAAGGCTA Reverse TGCCGCGAGATTTCCATACA	This paper	N/A
MISIIR Forward CAGCCAGAATGTGCTCATTCG Reverse GCTCAGTATCTCCCACAGTAG	This paper	N/A
FAK fl/fl Forward TAAGAGTCTAATCCACCACAGCA Reverse TCAGTCATCAGTTCTGCTCCTTA	This paper	N/A
GATA6 fl/fl Forward GTGGTTGTAAGGCGGTTTGT Reverse ACGCGAGCTCCAGAAAAAGT	This paper	N/A
LysMcre_mutant Forward CTTGGGCTGCCAGAATTTCTC Reverse CCCAGA AATGCCAGATTACG	This paper	N/A
LysMcre_WT Forward CTTGGGCTGCCAGAATTTCTC Reverse TTACAGTCGGCCAGGCTGAC	This paper	N/A
YFP_mutant Forward AAAGTCGCTCTGAGTTGTTAT Reverse AAGACCGCGAAGAGTTTGTC	This paper	N/A
YFP_WT Forward AAAGTCGCTCTGAGTTGTTAT Reverse GGAGCGGGAGAAATGGATATG	This paper	N/A
Recombinant DNA
pUltra-Chili-luciferase	Addgene	Cat# 48688
Software and algorithms
FlowJo 10	Becton-Dickinson	https://www.flowjo.com/
Prism 10	Graphpad	https://www.graphpad.com/
QuPath-0.5.0-x64	Bankhead et al.58	https://qupath.readthedocs.io/en/latest/index.html
NovoMagic	Novogene	https://cssamerica.novogene.com/pub/novoMagic
Python	Python Software Foundation	https://www.python.org
ImageJ	Schneider et al.59	https://imagej.nih.gov/ij/
RawConverter	N/A	https://doi.org/10.1021/acs.analchem.5b02721
Maven	N/A	https://lsi.princeton.edu/research/faculty-publications/lc-ms-data-processing-maven-metabolomic-analysis-and-visualization
LipidSearch 5.0	thermofisher.com	Cat# OPTON-30880
R (v4.1.2)	N/A	https://www.r-project.org/
Seurat (v4.3.0)	N/A	https://satijalab.org/seurat/
DoubletFinder (v2.0.3)	N/A	https://github.com/chris-mcginnis-ucsf/DoubletFinder
SCTransform	N/A	https://satijalab.org/seurat/articles/sctransform_vignette.html
CellRanger v7.1.0	N/A	10xgenomics.com/software
FeaturePlot	N/A	https://satijalab.org/seurat/reference/featureplot
SingleR	N/A	https://www.bioconductor.org/packages/release/bioc/html/SingleR.html
Adobe Photoshop 2025	Adobe.com	https://www.adobe.com/products/photoshop.html
DESeq2-R	Love et al.60	https://www.bioconductor.org/packages/devel/bioc/vignettes/DESeq2/inst/doc/DESeq2.html
ClusterProfiler	Yu et al.61	https://bioconductor.org/packages/release/bioc/html/clusterProfiler.html
Canvas X Draw (v7.1 build 7112)	canvasgfx.com	https://www.canvasgfx.com/products/canvas-x-draw
Other
M10 (GRCm38.p4) mouse reference genome	N/A	https://www.gencodegenes.org/mouse/release_M10.html
Mouse immune cell reference (ImmGenData)	N/A	https://immunology.hms.harvard.edu/resources/immgen

Lipidomics

150 μL of cold 5:3:2 MeOH:ACN:H2O (v/v/v) solution was added to the dried cell pellet derived from 5 × 10⁵ FAK-WT and FAK-KO KMF cells cultured in a murine plasma mimetic medium termed VIMPCS with 1X with physiological supplements.⁶⁶ Samples were vortexed for 30 min at 4°C, then centrifuged for 10 min at 18k RCF. Using 10 μl injection volumes, the supernatants were analyzed by ultra-high-pressure-liquid chromatography coupled to mass spectrometry (UHPLC-MS - Vanquish and Exploris, Thermo). Metabolites were resolved across a 1.7 μm, 2.1 × 150 mm Kinetex C18 column using a 5-min gradient previously described.⁶⁷ 500 μL of cold 5:3:2 MeOH:ACN:H2O (v/v/v) solution was used to reconstitute 40 μl dried media extract. Cold MeOH was added in a 1:25 ratio to the stored media. Samples were then vortexed vigorously for 30 min at 4°C, then centrifuged for 10 min at 18k RCF. Using 10 μL injection volumes, the supernatants were analyzed by ultra-high-pressure-liquid chromatography coupled to mass spectrometry (UHPLC-MS —Vanquish and Exploris, Thermo). Metabolites were resolved across a 1.7 μm, 2.1 × 150 mm Kinetex C18 column using a 5-min gradient previously described.⁶⁷ Using 10 μL injection volumes, non-polar lipids were resolved using UHPLC coupled to ddMS2 using a 5-min gradient method as previously described.⁶⁸

Following data acquisition, .raw files were converted to mzXML using RawConverter software. Metabolites were then assigned based on intact mass, ¹³C isotope pattern and retention times in conjunction with the KEGG database and an in-house standard library. Peaks were integrated using Maven (Princeton University). Quality control was assessed as using technical replicates run at beginning, end, and middle of each sequence as previously described. Lipidomics data were analyzed using LipidSearch 5.0 (Thermo Scientific), which provides lipid identification by intact mass, isotopic pattern, and fragmentation pattern to determine lipid class and acyl chain composition.

Exosome purification and analysis

KMF FAK-WT or FAK-KO cancer cells were cultured in six 15-cm culture dishes in DMEM supplemented with 10% FBS under standard conditions (37°C, 5% CO₂). Upon reaching 70% confluency, cells were washed twice with 10 mL sterile PBS to remove residual FBS-derived vesicles. Cells were then incubated in serum-free medium for 48 h. For exosome isolation, conditioned media was transferred to 50-mL conical tubes and centrifuged at 500 × g for 15 min at 4°C to remove cells. The resulting supernatants were centrifuged at 10,000 rpm for 20 min at 4°C to remove cell debris and large vesicles. Supernatants were transferred into SW28 tubes and exosomes were pelleted by ultracentrifugation at 110,000 × g for 70 min at 4°C (Beckman Coulter, SW28 rotor). Supernatants were discarded carefully. Pellets were resuspended in sterile PBS and combined into a single SW28 tube. A second ultracentrifugation was performed at 110,000 × g for 70 min at 4°C to wash the vesicles. Following centrifugation, supernatants were discarded, and the final exosome pellets (from 6 plates) were resuspended in 200 μL of PBS. The size distribution and concentration of isolated exosomes were determined using Nanoparticle Tracking Analyzer (NTA) on a ZetaView instrument (Particle Metrix) according to the manufacturer’s instructions.

QUANTIFICATION AND STATISTICAL ANALYSIS

Unless indicated, results presented are from at least two independent experiments. Statistical analyses were performed in Prism v10 (GraphPad Software). For experimental groups of three or more, statistical significance was calculated based on one-way ANOVA with Tukey’s multiple comparison test. Unpaired t test was used to determine statistical difference between the means from two different samples. p values < 0.05 were considered significant.

Supplementary Material

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2026.117009.

Highlights.

• FAKi, pegylated doxorubicin, and anti-TIGIT enhance survival and tertiary lymphoid formation

• Tumor FAK inhibition released omega-3 fatty acid exosomes, enhancing macrophage CXCL13

• Macrophage GATA6 loss enhanced FAK-knockout tumor growth and reduced peritoneal B cells

• Omega-3 fatty acids stimulate human ovarian tumor-associated macrophage CXCL13

Data and code availability

• RNA-seq FASTQ files have been deposited to Gene Expression Omnibus under the accession number GEO: GSE313445.

• All original code has been deposited at https://github.com/UCSD-Fisch-Lab/KMF_FAK_single_cell and is publicly available at the time of publication (https://doi.org/10.5281/zenodo.18203756).

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.