Aspirin prevents metastasis by limiting platelet TXA2 suppression of T cell immunity

Main

Despite advances in primary cancer treatment, many patients treated for early-stage cancers develop metastatic recurrence months to years later owing to the eventual growth of disseminated micrometastases⁴. Micrometastases are vulnerable to immune attack, as they are deprived of the highly immunosuppressive microenvironment found within established tumours³. This creates an opportunity for anti-metastatic therapies that utilize the immune system to prevent recurrence in patients with early-stage cancer at risk of metastasis.

Aspirin is an irreversible inhibitor of COX enzymes⁵. Meta-analyses of large randomized controlled trials have shown that daily aspirin treatment is associated with reduction in metastasis at multiple sites in individuals with cancer⁶ (hazard ratio (HR) 0.64, 95% confidence interval (CI) 0.48–0.84). Moreover, low-dose (75–300 mg) aspirin treatment is associated with a reduction in the rate of cancer death in individuals without metastasis at the time of cancer diagnosis^6,7,8 (HR 0.49, 95% CI 0.30–0.79). In colorectal cancer, the association of aspirin use with improved survival appears to be restricted to tumours that express high levels of human leukocyte antigen (HLA) class I, suggesting that its effect has an immune component⁹. COX-1 (also known as prostaglandin G/H synthase 1) is constitutively expressed in most tissues, including in platelets, where it is required for TXA₂ production, whereas COX-2 is predominantly induced during inflammation^5,10,11. Aspirin has a short half-life (around 20 min), such that only frequent high doses of aspirin can achieve sustained pharmacological inhibition of COX-1 and COX-2 in nucleated cells. By contrast, daily low-dose aspirin primarily targets platelet COX-1, and consequently the production of TXA₂, since anucleated platelets cannot resynthesize their COX-1 pool, which becomes irreversibly inhibited^5,10. Collectively, these results suggest a relationship between T cell immunity, suppression of platelet TXA₂ and the anti-metastatic activity of aspirin, but the direct relation between these has not been established.

In this study, we show that platelet TXA₂ suppresses immunity to cancer metastasis by activating a T cell-intrinsic immunosuppressive pathway that is dependent on the guanine exchange factor ARHGEF1. Consequently, restricting the availability of TXA₂ using aspirin, selective COX-1 inhibitors or platelet-specific deletion of COX-1 reduced the rate of metastasis in a manner that was dependent on T cell-intrinsic expression of ARHGEF1 and signalling by TXA₂ in vivo. These findings reveal a novel immunosuppressive pathway that limits T cell immunity to cancer metastasis, providing a mechanistic basis for the anti-metastatic activity of aspirin and paving the way for the development of more effective anti-metastatic immunotherapies.

ARHGEF1 suppresses T cell immunity to metastasis

We previously performed a large in vivo genetic screen to identify host regulators of cancer metastasis¹². This screen identified 15 genes whose disruption in host tissues reduced the frequency of lung metastases after intravenous injection of syngeneic B16 melanoma cells, including the gene encoding the 115-kDa RHO guanine exchange factor ARHGEF1^13,14. Confirming the results of our initial screen using littermate controls, we found a reduced number of metastases in the lungs of Arhgef1-deficient mice compared with wild-type controls after intravenous injection of syngeneic B16 melanoma cells (Fig. 1a,b). This corresponded to a reduction in the frequency of mCherry-labelled tumour cells in the lungs of Arhgef1-deficient mice upon intravenous injection of B78ChOva melanoma cells (Extended Data Fig. 1a). We similarly observed a decrease in the rate of lung metastasis after intravenous injection of syngeneic LL/2 Lewis lung carcinoma cells into Arhgef1-deficient mice (Fig. 1c and Extended Data Fig. 1b). The reduction in the rate of metastasis upon ARHGEF1 loss was not restricted to the lungs, since we also observed reduced numbers of liver metastases upon intrasplenic implantation of B16 cells in Arhgef1-deficient mice (Fig. 1d and Extended Data Fig. 1c).

**Fig. 1: ARHGEF1 suppresses T cell immunity to cancer metastasis.**

Mice bearing the MMTV-PyMT germline mutation develop primary breast cancers that spontaneously metastasize to the lungs¹⁵. Although we observed no major difference in the growth of primary mammary tumours in Arhgef1-deficient mice compared with wild-type controls when crossed onto the MMTV-PyMT background (Extended Data Fig. 1d), we noted a reduction in the frequency of metastatic nodules in the lungs of Arhgef1-deficient mice, when mice with similarly sized primary mammary tumours were analysed (Fig. 1e). Similar to the lack of substantial effect on the growth of primary MMTV-PyMT breast tumours, ARHGEF1 deficiency had minimal effect on the growth of subcutaneously implanted syngeneic MC38 colorectal adenocarcinoma tumours (Extended Data Fig. 1e). These findings showed that loss of ARHGEF1 expression in host tissues reduces the development of cancer metastases at multiple metastatic sites.

ARHGEF1 is predominantly expressed in cells of the haematopoietic lineage^13,14,16. We did not observe major differences in the frequency of mature haematopoietic lineages in tumour-bearing lungs of Arhgef1-deficient mice compared with wild-type controls (Extended Data Fig. 2). However, we found that loss of ARHGEF1 in haematopoietic cells was sufficient to confer the observed anti-metastatic effect, since lethally irradiated wild-type mice reconstituted with bone marrow haematopoietic cells from Arhgef1-deficient mice exhibited a reduced frequency of B16 lung metastases compared with mice reconstituted with wild-type haematopoietic cells (Fig. 1f). Metastatic colonization of the lungs of Arhgef1-deficient mice was associated with increased expression of genes associated with immune activation and cytotoxic function, including Ccl4, Nkg7, Gzma, Gzmb, Cxcl5 and Klrg1, compared with wild-type controls (Fig. 1g and Supplementary Table 1). These findings led us to hypothesize that ARHGEF1 exerts an immunosuppressive effect in cells of the haematopoietic lineage.

To determine the haematopoietic cell types in which ARHGEF1 exerts its immunosuppressive function, we generated a floxed Arhgef1 mouse allele (hereafter Arhgef1^fl), enabling conditional deletion of ARHGEF1 in specific cellular lineages upon Cre-mediated recombination (Fig. 1h). We crossed Arhgef1^fl/fl mice to Ncr1^cre, Lyz2^cre and Cd4^cre mice, enabling loss-of-function analysis of ARHGEF1 in haematopoietic cells of the natural killer (NK) cell, macrophage and T cell lineages, respectively^17,18,19. Whereas NK cell- and macrophage-specific ablation of ARHGEF1 did not result in reduced frequency of lung metastases after intravenous administration of B16 melanoma cells, we found that T cell-specific ablation of ARHGEF1 in Arhgef1^fl/fl Cd4^cre mice (hereafter Arhgef1-cKO mice) resulted in markedly reduced lung metastasis compared with Arhgef1^+/+ Cd4^cre (hereafter cWT) controls (Fig. 1i–l and Extended Data Fig. 3a), recapitulating the resistance to metastasis observed in mice lacking ARHGEF1 in all tissues.

Loss of ARHGEF1 enhances T cell function

Cytokine polyfunctionality is a hallmark of effective anti-viral and anti-tumour immune responses^20,21. To better understand how ARHGEF1-deficient T cells control pulmonary metastasis, we analysed lung-infiltrating T cells using flow cytometry. Although a similar number of CD4⁺ and CD8⁺ T cells were found in tumour-bearing lungs of cWT and Arhgef1-cKO mice (Extended Data Fig. 3b), intracellular cytokine staining of T cells from tumour-bearing lungs revealed markedly increased frequencies of polyfunctional cells that co-express two or more cytokines among IFNγ, IL-2 and TNF in Arhgef1-cKO mice compared with cWT mice (Fig. 2a–c).

**Fig. 2: Loss of ARHGEF1 promotes CD8⁺ T cells with enhanced cytokine polyfunctionality.**

T cell exhaustion is associated with reduced cytokine polyfunctionality and anti-tumour activity^22,23,24. Analysing the phenotype of T cells in the lungs by high-dimensional flow cytometry, we found that metastatic colonization induces populations of CD4⁺ and CD8⁺ T cells that express high levels of the exhaustion marker PD-1, including both terminally exhausted TIM3⁺TIGIT⁺TOX⁺CD39⁺TCF1^low (cluster 10) cells and progenitor exhausted SLAMF6⁺TIM3⁻TIGIT⁻TCF1^hi (cluster 14) cells, whose frequency was reduced in Arhgef1-cKO mice compared with cWT mice (Fig. 2d–h and Extended Data Figs. 3c,d and 4a). Of note, we observed increased cytokine polyfunctionality and reduced expression of the exhaustion markers PD-1 and TOX among tumour antigen-specific CD8⁺ T cells responding to ovalbumin (OVA) expressed by B78ChOva metastases in lungs of Arhgef1-cKO mice (Fig. 2i,j and Extended Data Fig. 4b).

Corresponding to augmented adaptive immune responses to metastasis, Arhgef1 deletion reduced the late accumulation of mCherry-labelled cancer cells in the lungs at days 11 and 17 after tumour injection, but did not affect their early accumulation and growth at days 1 and 7 after tumour injection (Extended Data Fig. 4c). PD-1 expression is induced upon sustained exposure to tumour antigen²⁵. We found that the frequency of PD-1⁺ cells among antigen-experienced CD8⁺ T cells did not correlate with the number of metastases when either wild-type or Arhgef1-deficient mice were analysed, suggesting that reduced PD-1 expression was unlikely to be the consequence of decreased tumour burden in the lungs of Arhgef1-deficient mice (Extended Data Fig. 4d). Moreover, the function of ARHGEF1 was not restricted to tumour-specific T cell responses, since OVA antigen-specific CD8⁺ T cells responding to infection with Listeria monocytogenes expressing OVA exhibited a less terminally differentiated phenotype with increased levels of the memory marker CD127 in Arhgef1-cKO compared with cWT mice²⁶ (Extended Data Fig. 5a–c). Collectively, these experiments show that ARHGEF1 functions intrinsically in T cells to limit effector functions and anti-metastatic immunity in vivo.

TXA₂ suppresses T cells via ARHGEF1

We sought to define upstream receptors and ligands that drive the immunosuppressive function of ARHGEF1 in T cells so as to reveal extracellular components of the pathway that might be amenable to therapeutic targeting. G-protein-coupled receptors (GPCRs) sense a variety of extracellular signals and control diverse cellular responses. ARHGEF1 is activated by a subset of GPCRs coupled to the G-protein subunits Gα12 and Gα13, and in turn activates the GTPase RHOA to drive intracellular signal transduction^27,28,29. To identify candidate GPCRs that act upstream of ARHGEF1 in T cells, we performed a bioinformatic analysis to identify Gα12/13-coupled GPCRs expressed by T cells. Previous work has biochemically quantified ligand-induced interactions between 148 GPCRs and known G-protein subunits³⁰. Using these data, we generated a list of Gα12/13-coupled GPCRs and examined their expression at an mRNA level in naive and activated CD8⁺ T cells. This revealed 18 Gα12/13-coupled GPCRs that were expressed at significant mRNA levels by naive or activated CD8⁺ T cells (Fig. 3a).

**Fig. 3: TXA₂ suppresses activation and proliferation of T cells via ARHGEF1.**

To determine which among the identified Gα12/13-coupled GPCRs act upstream of ARHGEF1 to suppress T cell activation, we performed a screen to test whether their known ligands or agonists suppress T cells in an ARHGEF1-dependent manner in vitro. The majority of ligands had little effect on T cell activation, or affected T cell activation in a manner that was independent of ARHGEF1. However, T cell activation was suppressed in the presence of a stable analogue of TXA₂, U46619, which is a ligand of the Gα12/13-coupled TXA₂ receptor (TP (also known as TXA₂-R)), in a manner that was almost completely dependent on ARHGEF1 (Fig. 3b and Extended Data Fig. 6). Confirming the results of the screen, we found that U46619 (hereafter TXA₂ analogue) potently suppressed the activation and proliferation of fluorescence-activated cell sorting (FACS)-purified wild-type naive CD8⁺ T cells upon stimulation, whereas it had a minimal effect on Arhgef1-deficient cells (Fig. 3c,d and Extended Data Fig. 7a,b). Moreover, TXA₂ analogue treatment limited the expansion of wild-type but not Arhgef1-deficient CD8⁺ T cells in a manner that was dependent on the activity of its receptor TP (encoded by Tbxa2r), since the effect of TXA₂ was reversed upon genetic or pharmacological disruption of TP in Arhgef1-proficient cells (Fig. 3e–g and Extended Data Fig 8a–d). Notably, TXA₂ receptor signalling suppressed T cell proliferation and activation in the absence of any other cell type, suggesting that TXA₂ can suppress T cells independently of effects on T cell–dendritic cell interactions³¹. We did not observe differences in apoptosis upon treatment of T cells with the TXA₂ analogue (Extended Data Fig. 7c). These findings suggest that ARHGEF1 has a critical role in transducing TXA₂ signalling in T cells, limiting T cell activation and proliferation in response to T cell receptor (TCR) signalling.

TXA₂ limits TCR-driven kinase signalling

To better define the effect of TXA₂-driven ARHGEF1 activity on the global T cell activation programme, we performed whole-transcriptome RNA sequencing (RNA-seq) of naive FACS-sorted wild-type and Arhgef1-deficient CD8⁺ T cells undergoing stimulation with anti-CD3ε and anti-CD28 antibodies in the presence of TXA₂ analogue. Consistent with phenotypic analyses, we observed that treatment of cells with TXA₂ analogue suppressed the expression of genes involved in T cell activation and effector differentiation, including Gzma, Fasl, Ccl5, Ccr5 and Prdm1 in an ARHGEF1-dependent manner (Fig. 4a and Supplementary Table 2). Accordingly, we observed highly significant enrichment in the expression of genes upregulated in memory versus naive CD8⁺ T cells in TXA₂ analogue-treated Arhgef1-deficient cells compared with wild-type cells³² (Fig. 4b, Extended Data Fig. 9 and Supplementary Tables 3–5). Together, these findings show that TXA₂ signalling acts via ARHGEF1 to regulate the global TCR-driven transcriptional programme in T cells.

**Fig. 4: Thromboxane signalling suppresses TCR-driven T cell and kinase pathway activation via TP, ARHGEF1 and RHOA.**

ARHGEF1 activates RHOA to control a variety of processes including cellular differentiation, motility, proliferation and kinase signalling^33,34,35. Kinase signalling has a key role in TCR-driven T cell activation, proliferation and differentiation³⁶. We found that treatment of T cells with TXA₂ analogue profoundly inhibited TCR stimulation-driven phosphorylation of S6 and ERK in a manner that was almost entirely dependent on ARHGEF1 (Fig. 4c and Extended Data Fig. 10a,b), suggesting that ARHGEF1 restricts TCR-driven PI3K and MAPK signalling in response to TXA₂ signalling. We also observed increased phosphorylation of S6 and ERK following in vitro stimulation of Arhgef1-deficient cells with phorbol 12-myristate 13-acetate (PMA) plus ionomycin (Extended Data Fig. 10c), suggesting effects on signal transduction downstream of proximal TCR signal transduction³⁷. Active RHOA pull-down assays using Rhotekin beads showed that stimulation of wild-type but not Arhgef1-deficient T cells with TXA₂ analogue increased the abundance of GTP-bound active RHOA, demonstrating that ARHGEF1 is critical to TXA₂-mediated RHOA activation (Fig. 4d). Consistent with a critical role of RHOA activation in the suppressive function of ARHGEF1, complementation of Arhgef1-deficient CD8⁺ T cells with a constitutively active mutant RHOA-Q63L (RHOA^CA)³⁸ was sufficient to reverse hyper-phosphorylation of S6 and ERK following anti-CD3 crosslinking (Fig. 4e). Moreover, genetic disruption of Rhoa using CRISPR-based mutagenesis in primary Cas9-expressing CD8⁺ T cells resulted in increased stimulation-driven S6 and ERK phosphorylation (Extended Data Fig. 10d,e).

Activation of PTEN by the RHOA effector ROCK enables RHOA to negatively regulate the AKT–S6 pathway, providing a means for crosstalk between small G-protein and kinase signalling^33,35. ROCK recruits PTEN to the cell cortex upon activation³³. Confocal immunofluorescence microscopy of wild-type and Arhgef1-deficient T cells in vitro showed that T cells respond rapidly to TXA₂ sensing by recruiting PTEN to the cell cortex in a manner dependent on both the TXA₂ receptor TP and ROCK, since pre-treatment of cells with either the ROCK inhibitor Y-27632 or the TP inhibitor SQ 29548 prevented TXA₂-mediated PTEN recruitment to the cell cortex (Fig. 4f and Extended Data Fig. 11a). Biochemical analysis of cell lysates showed that TXA₂-driven suppression of AKT phosphorylation is dependent on the activity of TP, ARHGEF1, ROCK and PTEN, whereas TXA₂-driven suppression of ERK phosphorylation is dependent on TP and ARHGEF1, but not ROCK or PTEN (Fig. 4g and Extended Data Fig. 11b). The PI3K–AKT pathway has a critical role in T cell activation, differentiation and cytokine production^39,40,41. Pharmacological inhibition of AKT using the well-characterized allosteric AKT inhibitor, AKT inhibitor VIII, reduced excessive S6 phosphorylation and cytokine production following stimulation of Arhgef1-deficient CD8⁺ T cells (Fig. 4h,i). This suggests that inhibition of kinase pathway activation by ARHGEF1 is critical to its suppressive function in T cells. Collectively, these findings show that ARHGEF1 and downstream activation of RHOA are required for suppression of TCR-driven kinase pathway phosphorylation and T cell activation upon TXA₂ signalling in vitro.

Aspirin releases T cells from suppression by TXA₂

The biosynthesis of TXA₂ is blocked by inhibitors of COX enzymes, including aspirin. Meta-analyses of large randomized controlled trials have shown that aspirin use is associated with protection against metastatic recurrence at multiple sites after primary cancer diagnosis in humans^6,7. Our observation that TXA₂ limits T cell activation in an ARHGEF1-dependent manner in vitro led us to hypothesize that aspirin exerts an anti-metastatic effect by releasing T cells from TXA₂-driven suppression mediated by ARHGEF1 in vivo.

Supplementation of the drinking water of mice with the TXA₂ analogue U46619 increased the frequency of pulmonary metastases following intravenous administration of B16 cells (Fig. 5a). Treatment of mice with aspirin decreased the abundance of TXA₂ as determined by measurement of its stable metabolite TXB₂ in the serum of mice (Fig. 5b). Of note, aspirin treatment reduced metastasis frequency in control mice but not in mice bearing a T cell-specific conditional deletion of ARHGEF1 (Fig. 5c), suggesting that its activity is immune-dependent. Moreover, aspirin treatment in vivo resulted in increased capacity for TCR-driven S6 phosphorylation among antigen-experienced CD4⁺ and CD8⁺ T cells from tumour-bearing lungs of wild-type mice, to increased levels observed in Arhgef1-deficient mice treated with either aspirin or vehicle control (Fig. 5d and Extended Data Fig. 12a,b). Notably, supplementation of aspirin-treated wild-type mice with TXA₂ analogue U46619 reversed the anti-metastatic activity of aspirin, but neither aspirin nor TXA₂ analogue affected metastasis frequency in mice carrying a T cell-specific conditional deletion of ARHGEF1 (Fig. 5e), showing that the anti-metastatic activity of aspirin is dependent on its ability to reduce TXA₂ abundance, releasing T cells from TXA₂-driven suppression by ARHGEF1. The importance of TXA₂ receptor signalling to the anti-metastatic activity of aspirin was further supported by bone marrow reconstitution experiments that showed that the anti-metastatic activity of aspirin is dependent on TP expression by cells of the haematopoietic lineage (Fig. 5f). Additionally, the anti-metastatic activity of aspirin was dependent on adaptive immune cells, since we observed no difference in metastasis frequency when Rag2-deficient mice (which lack all T cells and B cells, resulting in NK cell expansion and a reduced background rate of metastasis)¹² were treated with aspirin (Extended Data Fig. 12c).

**Fig. 5: Aspirin promotes anti-metastatic immunity by releasing T cells from ARHGEF1-dependent suppression by TXA₂.**

Platelet COX-1 suppresses anti-metastatic immunity

Aspirin can inhibit both COX-1 and COX-2^5,10. We sought to determine their relative contribution to the observed effect of aspirin on metastasis. We found that selective COX-1 inhibition was able to recapitulate the effect of aspirin, reducing metastasis frequency in a manner that was dependent on T cell-restricted expression of ARHGEF1 (Fig. 5g). Reduction in metastasis frequency upon COX-1 inhibition was associated with a decrease in the abundance of the stable TXA₂ catabolite TXB₂ in the serum of mice treated with COX-1 inhibitors. By contrast, COX-2 inhibition did not result in a decrease in either metastasis frequency or the systemic abundance of TXB₂ (Fig. 5h). Moreover, inhibition of P2Y₁₂, which is involved in alternative ADP-driven platelet activation, did not affect the rate of pulmonary metastasis (Extended Data Fig. 12d).

We next tested whether platelets are the primary source of TXA₂ responsible for promoting metastasis. Platelet transwell co-culture experiments revealed contact-independent suppression of T cell activation by platelets in a manner that was partially dependent on T cell expression of ARHGEF1 (Fig. 5i and Extended Data Fig. 13a–c). Moreover, antibody-mediated platelet depletion resulted in a profound reduction in metastasis frequency in mice that were inoculated intravenously with B16 cells, in which context ARHGEF1 proficiency was unable to promote metastasis (Extended Data Fig. 13d,e). Platelets produce TXA₂ in a manner that is dependent on the expression of COX-1. Given that COX-1 inhibition resulted in an anti-metastatic effect dependent on expression of ARHGEF1 in T cells, we tested whether the rate of metastasis is affected in Pf4^cre Ptgs1^fl/fl mice, which harbour a platelet and megakaryocyte-specific conditional deletion of COX-1^42,43 (Fig. 5j). Of note, platelet and megakaryocyte-specific conditional deletion of COX-1 resulted in a marked reduction in metastasis, accompanied by decreased urinary abundance of the TXA₂ catabolite TXM, but not of other prostanoid metabolites that we evaluated. To formally test the role of TXA₂ in ARHGEF1-dependent suppression of anti-metastatic immunity in vivo, we crossed mice with a genetic deficiency in the thromboxane synthase gene Tbxas1⁴⁴ with Arhgef1-deficient mice. This allowed us to examine the effect of absence of systemic TXA₂ on metastasis frequency in the presence or absence of ARHGEF1. Among ARHGEF1-proficient mice, genetic absence of thromboxane synthase resulted in a reduction in metastasis frequency, whereas in mice lacking ARHGEF1, loss of thromboxane synthase did not affect the rate of metastasis (Fig. 5k). Collectively, these data show that a critical anti-metastatic function of aspirin and COX-1 inhibitors is to release T cells from suppression by TXA₂ produced in a platelet COX-1-dependent manner.

Discussion

Here we show that platelet-derived TXA₂ functions as a potent immunoregulatory molecule that suppresses T cell immunity to cancer metastasis by inducing the immunosuppressive function of ARHGEF1. Consequently, COX-1 inhibition, including using aspirin, enhances anti-metastatic immunity by releasing T cells from TXA₂–ARHGEF1-mediated suppression (Extended Data Fig. 14). These findings add to our mechanistic understanding of the observed anti-metastatic effects of aspirin^6,7,8,9.

Although aspirin provides a potentially attractive opportunity for anti-metastatic therapy given its low cost, more selective targeting of the TXA₂–ARHGEF1 pathway could enable enhanced anti-metastatic activity and/or reduced bleeding risk and gastric toxicity. Understanding of the immunostimulatory effect of aspirin raises the possibility that aspirin may be used to synergize with other adjuvant immunotherapies. These findings build on previous data that show that in patients with colon cancer, the association of aspirin use with improved survival appears to be restricted to cancers with high HLA class I expression⁹, and suggest that additional immune biomarkers may help to stratify patients who are most likely to benefit from the anti-metastatic activity of aspirin. Given contradictory evidence relating to aspirin efficacy in distinct groups of patients^6,8,45, our findings underscore the need for detailed biomarker identification studies in the context of prospective randomized controlled trials to definitively establish the cancer types and patient populations in which aspirin has the greatest efficacy.

We observed that genetic disruption of ARHGEF1 reduced the frequency of exhausted T cells at the metastatic site and increased the frequency of memory-precursor effector cells (MPECs) in pathogen-specific CD8⁺ T cell responses in vivo. How the observed modulation of kinase pathways by TXA₂–ARHGEF1 relates to phenotypic changes in exhaustion and MPEC differentiation is unclear and remains an important topic for future investigation. The findings also suggest that the TXA₂–ARHGEF1 pathway may have broader roles in immune regulation, including in control of T cell memory and exhaustion during chronic viral infection^24,25,46,47. We observed mild effects of ARHGEF1 deficiency on T cells cultured ex vivo in the absence of TXA₂, which were more pronounced in short-term assays directly ex vivo than in longer-term culture experiments. Whether ARHGEF1 is mildly basally active in the absence of TXA₂ signalling, or tonically active after exposure to homeostatic levels of TXA₂ or a stable metabolite in vivo are interesting questions for future study.

Our findings build on a broader understanding of the importance of prostanoids in cancer development and immune evasion. In particular, COX-2 has well-established and multifaceted roles in carcinogenesis, and prostaglandin E₂ production by tumour cells has a significant role in immune evasion by established tumours⁴⁸. Recent findings also suggest that platelet TXA₂ can act directly on cancer cells to promote COX-2-dependent carcinogenesis⁴³, although it remains to be determined whether the ability of TXA₂ to stimulate prostaglandin E₂ production adds to the immunomodulatory potential of TXA₂ in the context of metastasis by COX-2-expressing tumours.

This work establishes TXA₂ as a regulator of T cell immunity, with implications for cancer prevention and therapy. The identification of this pathway provides mechanistic insights into the anti-metastatic effects of aspirin, a potential basis for its more targeted use, and targets for development of new therapeutic strategies for preventing metastatic disease.

Methods

Mice and reagents

All animal experiments were conducted in accordance with UK Home Office guidelines and were approved by the University of Cambridge Animal Welfare and Ethics Review Board. Littermate controls or age- and sex-matched mice were used in experiments as indicated. All mice, unless otherwise stated, were housed at the University of Cambridge University Biomedical Services Gurdon Facility or the Babraham Institute Biological Support Unit. Wild-type C57BL/6 mice were obtained from Charles River. Ptprc^a (CD45.1) congenic, OT-1 TCR^tg, Rag2^–/– and MMTV-PyMT (B6.FVB-Tg(MMTV-PyVT)634Mul/LellJ) mice were obtained from the Jackson Laboratory. Arhgef1-KO (Arhgef1^Tm1a) mice have previously been described¹². Tbxas1-KO (Tbxas1^tm1Swl) and Tbxa2R-KO (Tbxa2r^tm1Cof) mice, as previously described^44,49, were provided by S.-W. Lin. Arhgef1^fl/fl mice were generated by crossing Arhgef1^Tm1a mice with FlpO-deleter mice⁵⁰, and subsequently crossed with Cre-expressing strains, Ncr1^cre (Ncr1^{tm1.1(icre)Viv}), Lyz2^cre (Lyz2^tm1(cre)Ifo) and Cd4^cre (Tg(Cd4-cre)1Cwi) mice to generate conditional-knockout mice, respectively^17,18,19. Mice were genotyped by Transnetyx. Pf4^cre Ptgs1^fl/fl mice have been described previously⁴² and were housed at the animal facility of the “G. d’Annunzio” University of Chieti-Pescara and the animal experiments were performed under the European Communities Council (EEC) Directive of 22 September 2010 (2010/63/EU) and the National Ethical Committee (authorization no. 434/2024-PR).

Tumour metastasis model

Mouse metastatic melanoma cell lines B16-F10 (purchased from Kerafast) and B78ChOva-mCherry (provided by M. F. Krummel)⁵¹ were passaged in DMEM (Thermo Fisher Scientific) supplemented with 10% FBS and antibiotics. For lung metastasis model: mice were injected intravenously with 5 × 10⁵ B16-F10 cells or B78ChOva-mCherry cells in 150 µl Dulbecco’s PBS (DPBS) and lungs were dissected between day 14 and 17 for metastasis enumeration and subsequent analysis. LL/2 murine Lewis lung carcinoma cells were passaged in DMEM supplemented with 10% FBS and antibiotics. Mice were injected with 1 × 10⁶ LL/2 cells in 150 μl DPBS and lungs were dissected on day 17. Lungs were fixed in 4% formalin before being embedded in paraffin. Sections were taken from the centre of the lungs and stained with H&E. Slide images were taken using a Pannoramic digital slide scanner (3DHistech) and analysed using QuPath Software. Tumour burden was calculated as a percentage of total tissue area for each sample. For the liver metastasis model: mice were anaesthetized with isoflurane and a small incision was made in the left flank to expose the spleen. 3.5 × 10⁵ B16-F10 cells suspended in 50 μl DPBS were injected into the spleen. The wound was closed with sutures and skin staples. Mice were euthanized and livers were dissected on day 11 for metastasis enumeration and subsequent analyses. Aspirin (Aspégic, Sanofi Aventis and Sigma) was resuspended in drinking water at 600 mg l⁻¹. TXA₂ analogue U46619 (Cayman) was diluted in DMSO and delivered in drinking water at 50 μg kg⁻¹. All the drinking water contained 1% sucrose and was replaced 3–4 times per week. For platelet depletion, mice were administered R300 antibody (Emfret) intraperitoneally at 0.25 mg kg⁻¹ every 2–3 days from 1 day before intravenous injection of tumour cells. For inhibition studies, COX-1 inhibitor SC-560 (SelleckChem and Abcam), COX-2 inhibitor Celecoxib (SelleckChem), P2Y12 inhibitor Ticagrelor (SelleckChem) or vehicle control were administered at 30 mg kg⁻¹ via daily oral gavage 5 days before intravenous injection of tumour cells and continuing throughout the study.

Spontaneous germline MMTV-PyMT cancer metastasis model

Mice were palpated for mammary tumours once per week. Mammary tumours were assessed by taking length and width measurements with digital callipers three times a week after the first palpable mammary tumour was detected. When the total mammary tumour area exceeded 2.25 cm², mice were euthanized and lungs were collected and fixed in 4% formalin before being embedded in paraffin. Sections were taken from the centre of the lungs and stained with H&E. Slide images were taken using a Pannoramic digital slide scanner (3DHistech) and analysed using QuPath Software. Tumour burden was calculated as a percentage of total tissue area for each sample.

MC38 subcutaneous tumour model

MC38 mouse colorectal cancer cells were passaged in DMEM supplemented with 10% FBS and antibiotics. Mice were injected subcutaneously into one flank with 2.5 × 10⁵ cells in 100 µl PBS. Tumour growth was assessed by taking length and width measurements with digital calipers 3 times a week starting from day 7 after injection.

L. monocytogenes infection and kinetic analysis

For experiments assessing the response of CD8⁺ T cells to bacterial infections, bacteria were grown in BHI medium to an OD₆₀₀ of 0.1 before each experiment. The mice were infected with a sub-lethal dose of 5 × 10⁶ colony-forming units attenuated (ΔactA) L. monocytogenes expressing OVA by intravenous administration⁵². Blood samples were collected via the tail vein at serial time points following infection. CD8⁺ T cell responses were detected and their phenotypes were analysed by flow cytometry.

Bone marrow reconstitution experiments

For bone marrow reconstitution experiments, C57BL/6 mice were administered 1,000 Gy total-body γ-radiation from a ¹³⁷Cs source before intravenous injection of bone marrow cells from single-cell bone marrow preparations from 8- to 12-week-old mice. Mice were administered neomycin for two weeks following irradiation and reconstitution to limit infection risk, and were used in flow cytometry and cancer metastasis experiments at two to three months after reconstitution.

Flow cytometry

Single-cell suspensions from lymphoid tissues were prepared by mechanical dissociation through 40-µm cell strainers (BD Biosciences). Lungs were minced in media containing 20 µg ml⁻¹ DNase I (Roche) and 1 mg ml⁻¹ collagenase (Sigma Aldrich) and incubated with agitation at 37 °C for 40 min before also being dissociated through 40-µm cell strainers. Erythrocytes were lysed using ice-cold ACK Lysing Buffer (Gibco) for 5 min. Cells requiring intracellular staining of cytokines were stimulated prior to flow cytometry analysis using PMA, ionomycin, brefeldin A and monensin for 4 h in complete RPMI 1640 (Thermo Fisher Scientific). Viable cells were discriminated by first staining alone with Zombie UV fixable viability dye (Biolegend) or eFluor 780 fixable viability dye (eBioscience) in PBS, according to the manufacturer’s instructions. Cells were then incubated with specific surface antibodies on ice for 40 min in FACS buffer, in the presence of 2.4G2 monoclonal antibodies to block FcγR binding. Cell surface phosphatidylserine was labelled using the eBioscience Annexin V Apoptosis Detection Set (Thermo Fisher Scientific) according to the manufacturer’s protocol. For intracellular staining, the eBioscience Foxp3/Transcription Factor Staining Buffer Set (Thermo Fisher Scientific) or BD Cytofix/Cytoperm Fixation/Permeabilization Kit was used in accordance with the manufacturer’s instructions followed by intracellular staining with fluorochrome-conjugated antibodies for 40 min. Samples were acquired using Cytek Aurora (Cytek), BD LSR Fortessa (BD Biosciences) or Beckman CytoFLEX (Beckman Coulter) cytometers with their respective software: SpectroFlo (v3.3.0), BD FACSDiva (v8.0.1), and CytExpert (v2.5.0.77). Data were analysed using FlowJo v10.10.0 software (TreeStar LLC).

Antibodies used for flow cytometry are as follows: anti-CD103-Pacific Blue (2E 7, BioLegend, 121418, 1/200), anti-CD127-BV650 (A7R34, BioLegend, 135043, 1/200), anti-CD127-PE-Cy7-A7R34 (BioLegend, 135014, 1/200), anti-CD16/CD32 (93, BioLegend, 101302, 1/200), anti-CD25-PE-Cy7 (PC61, Invitrogen, 25-0251-82, 1/200), anti-CD25-BUV395 (PC61, BD, 564022, 1/200), anti-CD39-AF647 (Duha59, BioLegend, 143808, 1/400), anti-CD4-AF700 (RM4-5, BioLegend, 100536, 1/200), anti-CD4-BV650 (RM4-5, BioLegend, 100546, 1/400), anti-CD44-BV510 (IM7, BioLegend, 103044, 1/200), anti-CD44-BV786 (IM7, BD, 563736, 1/400), anti-CD44-PerCP-Cy5.5 (IM7, Invitrogen, 45-0441-82, 1/400), anti-CD44-APC (IM7, Invitrogen, 17-0441-83, 1/400), anti-CD45.2-ef506 (104, Invitrogen, 69-0454-82, 1/100), anti-CD45.2-FITC (104, BioLegend, 109805, 1/200), anti-CD61-PE (2C9.G3, Invitrogen, 12-0611-82, 1/200), anti-CD62L-BUV737 (MEL-14, BD, 612833, 1/400), anti-CD62L-APC (MEL-14, BioLegend, 104412, 1/400), anti-CD69-PECy5 (H1.2F3, BioLegend, 15-0691-82, 1/300), anti-CD69-PE-Dazzle (H1.2F3, BioLegend, 104536, 1/200), anti-CD8α-BUV395 (53-6.7, BD Horizon, 563786, 1/200), anti-CD8α-BUV805 (53-6.7, BD, 612898, 1/200), anti-CD8α-BV510 (53-6.7, BioLegend, 100752, 1/200), anti-CD8α-FITC, (53-6.7, Invitrogen, 11-0081-86, 1/200), anti-CD90.1-FITC (OX-7, BioLegend, 202503,1/100), anti-CD90.1-PerCP (OX-7, BD, 557266, 1/100), anti-Foxp3-APC (FJK-16S, eBioscience, 17-5773-82, 1/200), anti-IFNγ-BUV737 (XMG1.2, BD, 612769, 1/400), anti-IFNγ-FITC (XMG1.2, BioLegend, 505806, 1/200), anti-IL-2-PE (JES6-5H4, BioLegend, 503808, 1/200), anti-Ki67-PerCP-ef710 (SolA15, Invitrogen, 46-5698-80, 1/200), anti-KLRG1-APC (2F1/KLRG1, BioLegend, 138412, 1/200), anti-KLRG1-BV605(2F1/KLRG1, BioLegend, 138419, 1/200), anti-Ly108-APC (330-AJ, BioLegend, 134610, 1/200), anti-Ly6G-FITC (RB6-8C5, eBioscience, 11-5931-85, 1/400), anti-PD-1-APCef780 (J43, Invitrogen, 47-9985-82, 1/200), anti-PD-1-PE-Cy7 (RMP1-30, BioLegend, 109110, 1/200), anti-pErk T202/Y204-AF488, (197G2, Cell Signalling, 13214S, 1/100), anti-pErk T202/Y204-AF647 (197G2, Cell Signalling, 13148S, 1/100), anti-pS6 S235/6-PE (D57.2.2E, Cell Signalling, 5316S, 1/150), anti-ST2-PerCP-ef710 (RMST2-2, eBioscience, 46-9335-82, 1/200), anti-TCF-1-AF488 (C63D9, Cell Signalling, 6444S, 1/200), anti-TCRβ-BV570, (H57-597, BioLegend, 109231, 1/200), anti-TCRβ-FITC (H57-597, BioLegend, 109206, 1/200), anti-TCRβ-PerCP-Cy5.5 (H57-597, BioLegend, 109228, 1/200), anti-TIGIT-PE-Dazzle (1G9, BioLegend, 142110, 1/100), anti-TIGIT-PE (GIGD7, Invitrogen, 12-9501-82, 1/100), anti-TIM3-BV421 (RMT3-23, BioLegend, 119723, 1/100), anti-TIM3-BV785 (RMT3-23, BioLegend, 119725, 1/100), anti-TNF-APC (MP6-XT22, BioLegend, 506308, 1/200), anti-TNF-BV650 (MP6-XT22, BioLegend, 506333, 1/200), anti-TOX-PE (REA473, Miltenyi, 130-120-716, 1/200) and TER119-FITC (TER119, Invitrogen, MA5-17822, 1/200)

Computational analysis of flow cytometry data

Data were processed as previously described^53,54. In brief, Flow Cytometry Standard (FCS) 3.0 files were first imported in FlowJo version 10.10.0 to eliminate dead cells by manual gating, and select CD45⁺ leukocytes, subjected to biexponential transformation, then exported for computational analysis by a custom-made script making use of PhenoGraph (K value set at 30). Here, we modified the Linux-community and the core.py script to fix the seed to “123456” (run in Python version 3.7.3). Data were then converted in comma-separated value (CSV) files and merged into a single file by using the pandas package. The obtained data, exported as new CSV file (one for each cluster), were further imported in FlowJo and analysed to define the percentage of cells positive for each protein as well as their median fluorescent intensity. Data were finally metaclustered using the gplots R package. UMAP was performed with the UMAP Python package.

Fluorescence-activated cell sorting

Pre-enrichment of CD8⁺ T cells from single-cell suspensions was done using the MagniSort Mouse CD8⁺ T cell Enrichment Kit (Thermo Fisher Scientific) according to the manufacturer’s protocol. Markers required for cell sorting were stained using flow cytometry cell surface antibodies, diluted 1/100, while cell suspensions were being labelled with the Enrichment Antibody Cocktail from the kit. Cells were filtered again and resuspended in RPMI 1640 containing DAPI for live/dead discrimination before sorting. Cell sorting was performed using a BD Fusion or Aria instrument (BD Biosciences). Cells were sorted into solutions of RPMI 1640 supplemented with 25% Charcoal-Stripped FBS (Gibco) before being prepared for experiments as described below.

Primary T cell culture

Naive CD8⁺ T cells were isolated by FACS from the spleens and lymph nodes of wild-type and Arhgef1-KO mice and stimulated with immobilized Ultra-LEAF purified anti-mouse CD3ε (BioLegend) and Ultra-LEAF purified anti-mouse CD28 (37.51; BioLegend) in culture medium containing rhIL-2 (5 ng ml⁻¹; Peprotech) for 4–5 days. In some cases, cells were labelled with 2.5 μM CTV proliferation dye (Thermo Fisher) before stimulation, according to the manufacturer’s instructions.

Gα12/13 GPCR ligand stimulation assays

Naive CD8⁺ T cells were purified as described above by FACS from the spleens and lymph nodes of wild-type and Arhgef1-KO mice. Purified naive CD8⁺ T cells (0.8 × 10⁵) labelled with 2.5 μM CTV were activated by plate-bound anti-CD3ε and soluble anti-CD28 (5 µg ml⁻¹ each; BioLegend) in lipid-free medium containing 10% charcoal-stripped FBS for 5 days in the presence of rhIL-2 (5 ng ml⁻¹) and Gα12/13 GPCR ligands as follows: CXCL12 (1 μg ml⁻¹; R&D), Thrombin (10 units ml⁻¹; Sigma Aldrich), PAR1 peptide (10 μM; Abcam), PAR2 peptide (10 μM; Abcam), S1P (5 μM; Acros Organics), 20:4 LysoPI (1 μM; Avanti Polar Lipids), 18:1 LysoPS (1 μM; Avanti Polar Lipids), LPA (1 μM; Avanti Polar Lipids), 9-HODE (1 μM; Cayman), histamine dihydrochloride (5 mM; Sigma Aldrich), prostaglandin E₂ (0.5 μM; Cayman) and U46619 (5 μM; Calbiochem). At the end of the cell culture, cells were collected, and cell proliferation and activation were measured by flow cytometry. For the inhibition study, cells were treated with the TP inhibitor SQ 29548 (Cayman) at the concentration of 10 μM, beginning 1 h before stimulation and continuing until the end of the cell culture.

Acute T cell stimulation and analysis of protein phosphorylation

CD8⁺ T cells from wild-type and Arhgef1-KO OT-1 TCR^tg mice were stimulated in vitro with 10 nM OVA_257–264 peptide in the presence of rhIL-2 for 2 days, followed by a 3-day expansion with rhIL-2. Before TCR crosslinking, cells were serum-starved overnight in RPMI 1640 containing 0.1% fatty acid-free BSA (Sigma Aldrich). Cells were then incubated with soluble anti-CD3ε antibody (Thermo Fisher Scientific) at 4 °C for 20 min. After washing, CD3 molecules bound with anti-CD3ε antibodies were crosslinked using goat anti-Armenian hamster IgG (H+L) (Jackson ImmunoResearch) in a lipid-free medium containing 5 μM U46619 or DMSO control for 5 min. For western blots, following crosslinking, cells were immediately lysed in Pierce RIPA buffer with cOmplete Mini protease and PhosSTOP phosphatase inhibitor cocktails (Roche). Cells were then denatured in 2× Laemmli buffer (Bio-Rad) with 2-mercaptoethanol and boiled for 10 min at 98 °C before gel loading. Western blotting on PVDF membrane was performed using TGX reagents (Bio-Rad Laboratories) and protocols. Following transfer, blots were blocked with 5% BSA then incubated with primary antibodies against pAkt-Ser 473 (193H12, Cell Signaling), pan-Akt (9272, Cell Signaling), pS6-Ser 235/236 (D57.2.2E, Cell Signaling), S6 (5G10, Cell Signaling), pMEK1/2-Ser 217/221 (9121, Cell Signaling), MEK1/2 (9122, Cell Signaling), pERK1/2-Thr202/Tyr204 (D13.14.4E, Cell Signaling), ERK1/2 (3A7, Cell Signaling), ARHGEF1 (D25D2, Cell Signaling), β-actin (AC74, Sigma Aldrich) and GAPDH (1E6D9, Proteintech) with appropriate horseradish peroxidase-conjugated secondary antibodies (Bio-Rad). Horseradish peroxidase-conjugated secondary antibodies blots were developed using chemiluminescence (Thermo Fisher) and gel images were captured in the darkroom using photographic films. For inhibition studies, cells were pre-treated for 1 h with the following inhibitors: AKT inhibitor VIII (1 μM, Calbiochem), TP inhibitor SQ 29548 (10 μM, Cayman), ROCK inhibitors Y-27632 (30 μM, Tocris) and GSK269962A (10 μM, ApexBio) or PTEN inhibitor bpV(pic) (2.5 μM, Sigma Aldrich) before TCR crosslinking.

Ex vivo phosflow assay

Single-cell suspensions from the spleens and lymph nodes and lungs from mice of indicated genotype and treatments were prepared as described above except for using lipid-free medium throughout the procedure. Two-to-three million cells were incubated with specific surface antibodies and soluble anti-CD3ε antibody (Thermo Fisher Scientific) at 4 °C for 30 min. After washing, CD3 molecules bound with anti-CD3ε antibodies were crosslinked using goat anti-Armenian hamster IgG (H+L) (Jackson ImmunoResearch) in a lipid-free medium containing 5 μM U46619 or vehicle control for 5 min. Following crosslinking, the cells were immediately fixed with ice-cold 2% formaldehyde at 4 °C for 20 min. After fixation, the cells were permeabilized with ice-cold 90% methanol at 4 °C for 20 min. Phosphorylation of S6 and ERK was measured by flow cytometry using anti-pERK T202/Y204 AF488- or AF647-conjugated antibodies (197G2) and anti-pS6 S235/6-PE or -AF647 (D57.2.2E) antibodies.

Assays for quantification of RHOA activation

The activity of RHOA was measured using the RHOA Pull-Down Activation Assay Biochem Kit (Cytoskeleton). Following overnight serum starvation, 10 × 10⁶ wild-type and Arhgef1-KO OT-1 TCR^tg CD8⁺ T cells were stimulated with 5 μM U46619 for 5 min at 37 °C. Cells were immediately spun down, washed and lysed in the lysis buffer with the phosphatase and protease inhibitor cocktails at 4 °C. Fifty micrograms of Rhotekin-RBD beads (Cytoskeleton) were added to the lysates and rotated for 1 h at 4 °C. The samples were spun down, washed three times, denatured in 2× Laemmli buffer with 2-mercaptoethanol and boiled for 10 min at 98 °C before gel loading. The primary antibody against RHOA (Abcam) was used for western blotting.

Plasmids, cloning, retroviral transduction and CRISPR–Cas9 mutagenesis

Platinum-E ecotropic packaging cells (Cell Biolabs) were plated one day before transfections on poly-d-lysine-coated 10-cm plates (Corning) at a concentration of 6 × 10⁶ cells per plate. Packaging cells were transfected with 10 μg of retroviral plasmid DNA encoding MSCV-IRES-Thy1.1 (pMIT) or pMIT-RHOA^CA along with 6 μg pCL-Eco plasmid DNA using Transit293 (Sigma Aldrich) in OptiMEM (Thermo Fisher Scientific) for 8 h in antibiotic-free medium. Medium was replaced 8 h after transfection and cells were incubated for a further 48 h. Retroviral supernatants were used to spinfect Cas9-expressing OT-1 TCR^tg CD8⁺ T cells following stimulation with 10 nM OVA_257–264 peptide for 24 h. In brief, CD8⁺ T cells were collected, suspended in viral supernatant and spun at 2,000g for 2 h at 32 °C in 24-well plates in the presence of 8 μg ml⁻¹ polybrene (Sigma Aldrich) with slow acceleration and deceleration. Cells were cultured for 3–4 further days in rhIL-2-containing media prior to acute restimulation and phospho-flow assays.

For CRISPR–Cas9 mutagenesis, three different pairs of sgRNAs for Rhoa were designed using the CRISPick tool (Broad Institute), subcloned into optimized retroviral MSCV-sgRNA-puro-Thy1.1 vectors and retrovirally transduced to Cas9-expressing CD8⁺ T cells as described above. Deletion of Tbxa2r in naive T cells was as previously described⁵⁵. In brief, three different sgRNAs targeting Tbxa2r: 5′-CCAGAGAAGCTCATGACAGG-3′, 5′-UUAGGAGCCAUGUGGCCCAA-3′ and 5′-CGAGGUGCCAUUGGGCCACA-3′ or their negative control: scrambled sgRNA#1 (Synthego) were incubated with Alt-R S.p.HiFi Cas9 Nuclease V3 Cas9 (Integrated DNA Technologies) to form sgRNA–Cas9 ribonuclear protein complexes before electroporating into naive MACS-sorted CD8⁺ T cells labelled with CTV using a Lonza 4D-Nucleofector system (DS137). Cells were rested overnight before activation in culture for 3 days by plate-bound anti-CD3ε and soluble anti-CD28 (2 µg ml⁻¹ each; BioLegend) and rhIL-2 (5 ng ml⁻¹) in lipid-free medium containing 5 μM U46619 or vehicle control.

RNA-seq analysis

Lung tissues were immediately immersed in the RNAlater Stabilization solution (Thermo Fisher Scientific) for storage at −80 °C. The whole tissue was homogenized in 1 ml RLT Plus lysis buffer (Qiagen) by using a Qiagen TissueLyser II homogenizer and RNA was isolated by using the RNeasy Plus Mini kit (Qiagen) according to the manufacturer’s instructions. Naive CD8⁺ T cells (CD62L⁺CD44⁻CD8⁺) were purified by total CD8⁺ T cells enrichment followed by FACS, as described above and activated by plate-bound anti-CD3 and soluble anti-CD28 (5 µg ml⁻¹ each) in lipid-free medium containing 10% Charcoal-Stripped FBS for 5 days in the presence of recombinant rhIL-2 (5 ng ml⁻¹) and U46619 (1 μM). Cells were then pelleted and resuspended in 40 µl RNAlater Stabilization solution for storage at −80 °C. Processing of cell samples was performed using the QIAshredder Kit (Qiagen) and RNA was then extracted using the RNeasy Plus Mini Plus Kit (Qiagen) according to the manufacturer’s protocol. RNA libraries were prepared by Novogene using the mRNA library preparation kit (poly A enrichment) and sequencing was performed by Novogene on NovaSeq PE150. The FastQ files were then subjected to quality control using FastQC and then alignment to the NCBIM37 Mus musculus genome annotation using the STAR workflow. Differential gene expression analysis was performed on all expressed genes (>20 detected reads) using DESeq2⁵⁶, and differentially expressed genes were further analysed and visualized using R. Expression heat maps were generated with the R package heatmap.

Histopathological analysis of tissues

Lungs were fixed in 4% formalin before being embedded in paraffin. Sections were taken from the centre of the lungs and stained with H&E, using routine histology methods. Slides were genotype-blinded and independently scored for pathological features.

Quantification of urinary prostanoid metabolites

Urine samples from mice that received intravenous injections of B16-F10 cells were collected using metabolic cages. Systemic biosynthesis of prostaglandin E₂, prostaglandin D₂, prostaglandin I₂ and TXA₂ was evaluated by quantifying their major urinary enzymatic metabolites: PGEM, PGDM, PGIM and TXM, respectively, using liquid chromatography–tandem mass spectrometry as previously described^42,57,58. Urine samples (0.2 ml aliquots) were added with internal standards: tetranor PGEM-d₆ (final concentration of 10 ng ml⁻¹, Cayman), tetranor PGDM-d₆ (final concentration of 10 ng ml⁻¹, Cayman), 2,3-dinor-6-keto-PGF_1α-d₉ (sodium salt) (final concentration of 10 ng ml⁻¹, Cayman) and 2,3-dinor-TXB₂-d₉ (final concentration of 10 ng ml⁻¹, Cayman). Samples were incubated at room temperature for 15 min, followed by addition of formic acid (5 μl). After 15 min of incubation, methoxyamine HCl (1 g ml⁻¹, 0.1 ml) (Sigma Aldrich) was added. Following 30 min of incubation at room temperature, urine samples were diluted to 1 ml with water adjusted to pH 3 with HCl and extracted with Number Strata-X 33u polymeric reversed phase (30 mg ml⁻¹, Phenomenex) that had been preconditioned with 1 ml of acetonitrile and 1 ml of water^58,59. The Number Strata-X 33u polymeric reversed phase column loaded with samples was washed with 1 ml of water (5% acetonitrile) and then eluted with 1 ml of 5% acetonitrile in ethyl acetate. The eluate was evaporated, and the dried residue was resuspended in 100 μl mobile phase (10% acetonitrile in water), and 30 μl was injected into an ACQUITY UPLC I-Class/Xevo TQS micro-IVD System (Waters) equipped with an electrospray ionization source (ESI Z-Spray), under negative ionization conditions, as previously described⁵⁸.

Platelet and T cell transwell co-cultures

Platelets were isolated as previously described⁶⁰. In brief, blood was collected via cardiac puncture using syringes containing 100 μl of acid citrate dextrose (Sigma Aldrich) from B16-F10 bearing mice. Samples were immediately mixed with modified Tyrode’s buffer to prevent premature activation of platelets. Blood was centrifuged at 300g for 5 min and the uppermost platelet-rich plasma (PRP) layer was collected. PRP was subsequently centrifuged at 200g for 8 min to increase purity. Prostacyclin I₂ (Sigma Aldrich) was added at a final concentration of 1 μg ml⁻¹ to limit premature platelet activation and the sample was immediately centrifuged for 11 min at 1,000g. The cells were resuspended in modified Tyrode’s buffer. Platelet count and purity were determined by flow cytometry using the following antibodies: Armenian hamster anti-mouse/rat CD61-PE antibody (2C9.G2, BioLegend), rat anti-mouse Ter119-FITC antibody (TER119, eBioscience) and anti-mouse CD45.2-PerCp-Cy5.5 antibody (Ly5.2). Platelet purity was greater than 90%.

Platelets and naive FACS-sorted CD8⁺ T cells labelled with 2.5 μM CTV were co-cultured with a ratio of 30:1 in 24-well transwell plates (0.4 μm pore size, Corning). Naive CD8⁺ T cells on the bottom of the transwell were activated by plate-bound anti-CD3ε and soluble anti-CD28 (5 µg ml⁻¹ each) in a lipid-free medium in the presence of rhIL-2 (5 ng ml⁻¹) as described above. After 5 days, supernatants were carefully collected and TXB₂ levels were measured using a Thromboxane B₂ ELISA Kit (Cayman) according to the manufacturer’s instructions. T cells were collected and processed for flow cytometry as previously described.

Confocal immunofluorescence microscopy

CD8⁺ T cells from wild-type and Arhgef1-KO OT-1 TCRtg mice were generated in vitro for 5 days as described above. 5 × 10⁵ serum-starved cells were seeded on 13-mm coverslips (VWR) pre-coated with 10 μg ml⁻¹ Ultra-LEAF™ purified anti-mouse CD3ε in a 24-well plate. Cells were stimulated with 10 μM U-46619 or DMSO control for 10 min at 37 °C. After fixation and washing with 15 mM glycine and PBS. Coverslips were blocked with Fc blocking anti-FcR antibody (clone 2.4G2, BD). Cells were then incubated with the rabbit anti-mouse PTEN (138G6, Cell Signalling) antibody, followed by the donkey anti-rabbit IgG-Alexa Fluor 647 (A31573, Thermo-Fischer) antibody and Coralite 594-Phalloidin (Proteintech). Coverslips were mounted onto VWR SuperFrost Microscopy Slides (Appleton Woods MS527) using ProLong Diamond Antifade Mountant with DAPI (Thermo-Fischer P36962). For inhibition experiments, cells were pre-treated with 10 μM TP inhibitor SQ 29548, ROCK inhibitors 30 μM Y-27632 and 10 μM GSK269962A or DMSO for 1 h before seeding.

Images were captured on a Leica TCS SP8 inverted confocal microscope using the Leica Application Suite X (LAS X) software (v1.4.6.28433) with a 63× oil-immersion objective lens, in three channels, F-actin (yellow) and PTEN (red) plus a DAPI nuclear counterstain (blue) to locate the cells. z-stacks were obtained to capture the entirety of cells in each region. Three to five distinct locations were imaged for each replicate of each treatment condition. Images analysed were maximum intensity projections of four consecutive z-slices selected from the centre of a cross-volume image stack. An image analysis pipeline was constructed using Cellprofiler 4⁶¹. In brief, cell masks were generated from the F-actin image using as a seed a nuclei mask segmented from the DAPI image. These cell masks could be further compartmentalized into membrane and cytoplasmic regions by using the F-actin image. Ratiometric measurements of mean fluorescence intensity of PTEN on the membrane versus intracellular were used to report its re-localization to the membrane.

Single-sample gene set enrichment analysis

Every expressed gene detected in the RNA-seq experiment in Fig. 4a (14,025 genes; Supplementary Table 3) was subject to single-sample Gene Set Enrichment Analysis (ssGSEA) analysis⁶². ssGSEA was used to test the relative enrichment of each gene set comprising the MsigDB C7 Immunologic signature gene sets in the total expressed gene transcriptional profile of each replicate sample. ssGSEA first calculates for each sample the differential expression of each expressed gene between that sample and the remainder of the samples. The list of expressed genes in each sample is then ranked by their level of differential expression compared to the other samples. The non-random distribution (enrichment) of genes within each C7 gene set for each sample is then calculated, yielding a sample enrichment score for each gene set per sample. This is achieved by analysing the distribution of the genes in each C7 gene set in the list of expressed genes rank ordered by their differential expression for each sample. The sample enrichment score for each differentially enriched gene set (false discovery rate (FDR) < 0.2, |FC| > 1.5) is represented in the heat map provided and the identity of each gene set in the heat map is given in Supplementary Table 4. Each gene within each of the gene sets included in the analysis can be looked up using the following link: https://www.gsea-msigdb.org/gsea/msigdb/human/genesets.jsp?collection=C7.

Cell line authentication

Validated B16-F10 melanoma cells were obtained from ATCC. Plat-E cells were obtained from Cell Biolabs. Validated MC38 colorectal adenocarcinoma cells were obtained from Kerafast. All lines were validated as mycoplasma-free by suppliers and expanded at low passage frequency before cryopreservation.

Statistical testing

Data were analysed using unpaired two-tailed Student’s t tests, Mann–Whitney U-test, ordinary one-way ANOVA or two-way ANOVA test where stated. Most experiments did not require blinding since objective quantitative assays, such as flow cytometry, were used. For tumour experiments littermate controls or age- and sex-matched mice of different genotypes were randomized and the operator blinded to genotype before injection and again before counting of metastatic nodules or assessment of histology images to allow for objective assessment. Experimental sample sizes were chosen using power calculations, preliminary experiments, or were based on previous experience of variability in similar experiments. Samples which had undergone technical failure during processing were excluded from subsequent analysis.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.