Phospholipid scramblases TMEM16F and Xkr8 mediate distinct features of phosphatidylserine (PS) externalization and immune suppression to promote tumor growth

Abstract

The phospholipid scramblases Xkr8 and TMEM16F externalize phosphatidylserine (PS) by distinct mechanisms. Xkr8 is activated by caspase-mediated proteolytic cleavage and, in synergy with the inactivation of P4-ATPase flippases, results in the irreversible externalization of PS on apoptotic cells and an "eat-me" signal for efferocytosis. In contrast, TMEM16F is a calcium-activated scramblase that reversibly externalizes PS on viable cells via the transient increase in intracellular calcium in live cells. The tumor microenvironment (TME) is abundant with exposed PS, resulting from prolonged oncogenic and metabolic stresses and high apoptotic indexes of tumors. Such chronic PS externalization in the TME has been linked to host immune evasion from interactions of PS with inhibitory PS receptors, such as TAM and TIM family receptors. Here, in an effort to better understand the contributions of apoptotic vs live cell PS-externalization to tumorigenesis and immune evasion, we employed an EO771 orthotopic breast cancer model and genetically ablated Xkr8 and TMEM16F using CRISPR/Cas9. While neither the knockout of Xkr8 nor TMEM16F showed defects in cell intrinsic properties related to proliferation, tumor-sphere formation, and growth factor signaling, both knockouts suppressed tumorigenicity in immune-competent mice, but not in NOD/SCID or RAG-knockout immune-deficient strains. Mechanistically, Xkr8-KO tumors suppressed macrophage-mediated efferocytosis, and TMEM16F-KO suppressed ER stress/calcium-induced PS externalization. Our data support an emerging idea in immune-oncology that constitutive PS externalization, mediated by scramblase dysregulation on tumor cells, supports immune evasion in the tumor microenvironment. This links apoptosis/efferocytosis and oncogenic stress involving calcium dysregulation, contributing to PS-mediated immune escape and cancer progression.

Fig. 1. Ablation of Xkr8 on EO771 tumor cells blocks PS externalization during apoptosis without affecting intrinsic oncogenic properties.

A q-RT-PCR analysis of Xkr isoform expression (Xkr4, Xkr8, Xkr9) in EO771 tumor cells. Xkr8 is the most highly expressed isoform among the Xkr family (Ordinary One-way ANOVA, n = 4, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. B Validation of Xkr8 knockout (KO) in EO771 tumor cells using western blotting with a custom polyclonal anti-Xkr8 antibody (Abclonal). Complete ablation of Xkr8 is observed in KO clones. C Real-time measurement of PS externalization in Xkr8 KO and wild-type (WT) EO771 cells upon apoptosis induction with 100 nM Staurosporine using fluorescently labeled Annexin-V, as measured by Incucyte. D Cell proliferation assay comparing Xkr8 KO and WT EO771 tumor cells as measured by bright field phase object confluence (%) by Incucyte. Tumor-sphere formation assay in Matrigel comparing Xkr8 KO and WT EO771 tumor cells was also measured by Incucyte. E Quantification of tumor-sphere area μM²/Image and F tumor sphere counts were compared for Xkr8 KO and WT EO771 tumor-spheres.

Fig. 2. Xkr8 knockout reduces tumor growth in an orthotopic model of breast cancer.

A Experimental timeline for orthotopic injection of EO771 wild-type (WT) or Xkr8 knockout (KO) tumor cells into the mammary fat pads of C57BL/6 wild-type (WT) mice. Tumor volumes were measured periodically to assess growth. Tumor weights and spleen weights were measured at endpoint. B Relative tumor volumes of EO771 WT and Xkr8 KO tumors at various time points (Mean values ± SD, Two-way ANOVA, Mixed effects model, n = 15). C Tumor weights of EO771 WT and Xkr8 KO tumors at the time of sacrifice point (Mean values ± SD, One-way ANOVA, n = 15). D Spleen weights of EO771 WT and Xkr8 KO tumor-bearing mice point (Mean values ± SD, One-way ANOVA, n = 15). E Relative tumor volumes in immune-deficient NSG mice, injected with EO771 WT or Xkr8 KO cells (Mean values ± SD, Two-way ANOVA, Mixed effects model, n = 9). F Tumor weights in NSG mice injected with EO771 WT or Xkr8 KO cells (Mean values ± SD, One-way ANOVA, n = 9). G Relative tumor volumes (Mean values ± SD, Two-way ANOVA, Mixed effects model, n = 10) and H tumor weights (Mean values ± SD, One-way ANOVA, n = 9) in Rag1 knockout (KO) mice, which lack mature T and B cells. (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Fig. 3. TMEM16F knockout prevents PS externalization in response to calcium stress without altering intrinsic oncogenic properties.

A q-RT-PCR analysis showing the relative expression of TMEM16 isoforms (16C, 16D, 16F, 16G, and 16J) in EO771 cells (Mean Relative gene expression values ± SD, Ordinary One-way ANOVA, n = 3, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). B Western blot validation of TMEM16F knockout (KO) in EO771 cells. The absence of TMEM16F is confirmed in the KO clones using a custom antibody (courtesy of Lily Jan, UCSF). C Fluorescent microscopic analysis of PS externalization in response to calcium influx using A23187 (calcium ionophore) and Annexin V-FITC staining. EO771 wild-type (WT) cells show robust PS externalization, while TMEM16F KO cells exhibit significantly reduced PS externalization. D Fold change in MFI for Annexin V-PE staining by flow cytometry of PS externalization in EO771 WT and TMEM16F KO cells upon treatment with A23187. E Cell proliferation assay comparing EO771 WT and TMEM16F KO cells using Incucyte, as measured by bright field phase object confluence % by Incucyte. Tumor-sphere formation assay in Matrigel comparing TMEM16F KO and WT EO771 tumor cells was also measured by Incucyte. F Quantification of tumor-sphere area μM²/Image and G tumor sphere counts were compared for TMEM16F KO and WT EO771 tumor-spheres.

Fig. 4. TMEM16F knockout reduces tumor growth in syngeneic orthotopic models of breast and colon cancer.

A Relative tumor volume % from C57BL/6 WT mice injected with EO771 WT or TMEM16F KO cells into mammary fat pads (Mean values ± SD, Two-way ANOVA, Mixed effects model, n = 15). B Tumor weight (g) measurements from the same C57BL/6 WT mice showing reduced tumor weight in TMEM16F KO tumors compared to WT tumors (Mean values ± SD, One-way ANOVA, n = 15). C Spleen weight (g) measurements from C57BL/6 WT mice injected with EO771 WT or TMEM16F KO cells (Mean values ± SD, One-way ANOVA, n = 15). D Relative tumor volume % (Mean values ± SD, Two-way ANOVA, Mixed effects model, n = 10), E tumor weight (g) (Mean values ± SD, One-way ANOVA, n = 10) and F spleen weight (g) (Mean values ± SD, One-way ANOVA) measurements from an independent clone: 2 of EO771 TMEM16F KO cells, showing consistent reduction in tumor growth compared to WT tumors. F Kaplan–Meier survival curve showing improved survival in mice bearing EO771 TMEM16F KO tumors compared to those with WT tumors (n = 10). G Relative tumor volume (Mean values ± SD, Two-way ANOVA, Mixed effects model, n = 10). H tumor weight (Mean values ± SD, One-way ANOVA, n = 10), and I spleen weight measurements from MC38 colon cancer model in C57BL/6 WT mice (Mean values ± SD, One-way ANOVA, n = 10). J Relative tumor volume (Mean values ± SD, Two-way ANOVA, n = 9) and K tumor weight (g) (Mean values ± SD, One-way ANOVA, n = 9) measurements in NSG mice injected with EO771 WT or TMEM16F KO cells, showing no significant difference between the two groups. L Relative tumor volume (Mean values ± SD, Two-way ANOVA, Mixed effects model, n = 10)and M tumor weight measurements in Rag1 KO mice injected with EO771 WT or TMEM16F KO cells (Mean values ± SD, One-way ANOVA, n = 15) (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Fig. 5. Apoptotic tumor cells induce efferocytosis by Xkr8.

A Schematic showing the workflow for efferocytosis assays, wherein apoptotic EO771 cells (WT, TMEM16F KO and Xkr8 KO) were labeled with pHrodo red, fed to bone marrow-derived macrophages, and after 3 hours, % of uptake was measured by flow cytometry. B Rate of efferocytosis as shown by % of pHrodo red + macrophage cells (Mean ± SD, One-way ANOVA, n = 6 *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). C PS externalization was disrupted in Xkr8 KO apoptotic cells, but not in WT and TMEM16F KO apoptotic cells (D), providing evidence that Xkr8-mediated PS externalization is essential for efficient recognition and phagocytosis by BMDMs.

Fig. 6. Intracellular calcium regulation and PS externalization in tumor cells.

A Intracellular calcium levels were upregulated in both WT and TMEM16F KO EO771 cells upon treatment with calcium ionophore, as indicated by the increased calmodulin-GFP fluorescence. B PS externalization as measured by relative MFI of Annexin V staining in calcium ionophore-treated WT and TMEM16F KO cells (Mean values ± SD, One-way ANOVA, n = 3 *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). C Flow plots showing intracellular calcium levels as shown by Fluo-AM staining. D PS externalization as measured by relative MFI of Annexin V staining in WT and TMEM16F KO cells exposed to glucose deprivation for 24 h (n = 2). E Live cell imaging with Incucyte shows calcium upregulation and PS externalization (Annexin V) for up to 24 h with glucose deprivation. F Gating strategy shows that PS externalization was measured on the live cells that stained negative for active caspase 3, as well as the viability stain. G PS externalization on cells treated with low dose MG132, a proteasome inhibitor, as shown by relative MFI of annexin V staining (Mean values ± SD, One-way ANOVA, n = 3 *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). H Flow plots showing intracellular calcium levels upon MG132 staining as shown by Fluo-AM staining. I Live cell imaging with Incucyte showed calcium upregulation and PS externalization (Annexin V) for up to 24 hours MG132 treatment.

Fig. 7. TMEM16F and XKr8 KO affect the TME through distinct pathways.

A Schematic of the Nanostring Gene Expression Analysis used to assess the tumor immune microenvironment (TME) in tumors grown in C57BL6 WT mice at day 15 after injection. B Clustered heat map showing the top 50 differentially regulated genes in Xkr8 KO and TMEM16F KO tumors compared to EO771 WT tumors is shown (n = 12, genes for p-values < 0.05 are shown and were calculated using ANOVA). WT-1,2,3,4, T-1,2,3,4 and X-1,2,3,4 correspond to EO771 WT, TMEM16F KO and Xkr8 KO tumors, respectively. Gene expression analysis showed downregulation of IL-10 (C), a cytokine released after efferocytosis, and CTLA-4 (D), an immune checkpoint involved in immune suppression in Xkr8 KO tumors. In contrast, HMGB1, a damage-associated molecular pattern (DAMP) released during immunogenic cell death, was upregulated in Xkr8 KO tumors (E). F–I TMEM16F KO tumors exhibited downregulation of MFGE8 (F), a protein involved in PS binding and efferocytosis, and MMP9 (G), which is involved in matrix remodeling and cancer progression. Caspase-1 (H) and CSF-1 (I) were upregulated in TMEM16F KO tumors, indicating increased cell death and macrophage infiltration. J Both Xkr8 KO and TMEM16F KO tumors showed downregulation of EGFR expression, suggesting an impact on EGFR-expressing cell populations (Mean log₂ normalized expression values ± SD, Two-way ANOVA, Mixed effects model, n = 12 *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Fig. 8. PS externalization on tumor cells by scramblases Xkr8 and TMEM16F mediates pro-tumorigenic effects by distinct pathways.

A The working model shows that PS externalization by scramblases Xkr8 and TMEM16F mediates pro-tumorigenic effects by unique pathways, wherein Xkr8 upregulates PS exposure on apoptotic cells, leading to increased efferocytosis and secretion of immune-suppressive cytokines, whereas TMEM16F is regulated by calcium stress in live cells, leading to MFGE8-mediated matrix remodeling. B Temporal activation of the scramblases can be explained as such: TMEM16F is activated chronically in live cells undergoing chronic ER stress and upregulated intracellular calcium, whereas Xkr8 activation is a late signal induced by caspase-mediated cell death.