Abstract

   Immune checkpoint blockade (ICB) therapies for solid tumors often fail
   due to resistance, necessitating new strategies. While efforts target
   IFNγ signaling or antigen presentation, other immune evasion mechanisms
   are unclear. Here, we identify Monoacylglycerol O-Acyltransferase 1
   (Mogat1) as a critical modulator of tumor immune evasion using an in
   vivo transcriptomic screen in progressing tumors. We find that tumors
   exploit Mogat1 to sequester fatty acids into triglycerides, a metabolic
   adaptation that fuels growth and fosters an immunosuppressive
   microenvironment, enabling immune escape. Genetic inhibition of Mogat1
   suppresses tumor growth by promoting T-cell infiltration and enhancing
   their tumor-killing ability. Importantly, Mogat1 loss sensitizes tumors
   to PD-1 blockade, overcoming resistance and suggesting reduced reliance
   on conventional antigen presentation. Our findings reveal a lipid
   metabolism-centered immune evasion mechanism and highlight Mogat1 as a
   potential target to improve cancer immunotherapy.

   Subject terms: Breast cancer, Immunoediting
     __________________________________________________________________

   Despite initial clinical success of immune checkpoint blockade,
   unfortunately not all patients respond to these therapies due to immune
   evasion mechanisms. Here, the authors discover that Monoacylglycerol
   OAcyltransferase 1 (Mogat1) regulation of lipid homeostasis modulates
   tumor immune evasion in breast cancer.

Introduction

   Immune checkpoint blockade (ICB) therapies have achieved remarkable
   clinical success, yet a majority of patients either fail to respond
   initially or develop acquired resistance^[36]1–[37]5. This resistance
   is often attributed to tumor-intrinsic alterations, such as mutations
   disrupting interferon-γ (IFNγ) signaling or antigen presentation
   pathways^[38]1–[39]5. Critically, effective strategies to overcome
   acquired ICB resistance are lacking, underscoring the urgent need for
   alternative approaches that can reignite anti-tumor immunity within the
   tumor microenvironment (TME).

   While genome-wide genetic perturbation screens (e.g., CRISPR-Cas9)
   provide valuable insights into gene function^[40]4–[41]7, particularly
   in vitro or under simplified in vivo conditions, they may not fully
   capture the adaptive changes occurring during tumor progression within
   a complex, evolving TME. Such screens identify effects following
   introduced modifications. Recognizing these limitations, we pursued a
   distinct approach: an in vivo selection strategy using longitudinal
   transcriptomic analysis to identify genes whose expression is naturally
   modulated by immune pressure during tumor development.

   To decipher the dynamic interplay between tumor cells and the immune
   system during cancer progression, we first characterized the evolving
   immune landscape in an immunocompetent mouse model of MMTV-PyMT breast
   cancer. Building on this model, we employed an unbiased in vivo
   selection strategy coupled with transcriptomic profiling across
   different stages of tumor development. This approach, designed to
   capture genes naturally enriched under immune pressure, led us to
   identify a previously uncharacterized role for Monoacylglycerol
   O-acyltransferase 1 (Mogat1) in tumor immune evasion. Mogat1 is a key
   enzyme in triglyceride (TAG) synthesis, catalyzing the formation of
   diacylglycerol (DAG) from monoacylglycerol (MAG)^[42]8. Its expression
   is typically restricted to metabolically active tissues like the
   intestine, liver, and adipose tissue^[43]9,[44]10, where it influences
   systemic metabolism; for instance, hepatic Mogat1 inhibition improves
   glucose homeostasis^[45]11. Despite its known roles in normal
   physiology and hepatic steatosis, the function of Mogat1 in cancer
   biology, particularly within the TME, has remained unexplored. Here, we
   reveal an unexpected role for Mogat1 in driving metabolic adaptations
   during tumor progression. We demonstrate that tumor cells leverage
   Mogat1 to sequester excess fatty acids into TAG stored in lipid
   droplets, effectively rewiring metabolic pathways to simultaneously
   fuel growth and establish an immunosuppressive niche. This study
   unveils a lipid metabolism-centric mechanism of immune evasion mediated
   by Mogat1 and identifies it as a promising therapeutic target to
   enhance cancer immunotherapy.

Results

Immune infiltration dynamics during breast cancer progression

   A robust anti-tumor immune response is critical in suppressing tumor
   development. However, tumor cells often evolve mechanisms to evade
   immune surveillance, enabling their survival and expansion. Uncovering
   the genetic drivers of this immune escape is essential for developing
   effective immunotherapies.

   As an alternative to direct genetic manipulation approaches, we
   employed an unbiased in vivo selection strategy coupled with
   transcriptomic profiling to identify genes whose expression is
   naturally modulated during tumor progression under immune pressure. By
   analyzing the transcriptomes of tumor cells isolated at various stages
   of development within an immunocompetent host, we aimed to capture
   dynamic gene expression changes reflecting the intricate tumor-immune
   interactions across different phases of progression. We reasoned that
   leveraging the natural selective pressures of the host immune system in
   situ would allow us to identify clinically relevant adaptive changes
   potentially missed by in vitro or simplified in vivo perturbation
   screens. Specifically, instead of introducing exogenous genetic
   modifications, we injected GFP-expressing MMTV-PyMT breast cancer cells
   into the mammary fat pads of immunocompetent FVB mice, allowing the
   tumors to evolve naturally. We hypothesized that subsequent
   transcriptomic analysis of tumor cells harvested at distinct time
   points would reveal genes upregulated or downregulated through natural
   selection, thereby identifying factors essential for tumor growth and
   immune evasion in this physiologically relevant context.

   Following tumor induction, we monitored tumor growth and conducted
   detailed immune profiling weekly (Fig. [46]1a). This approach allowed
   us to track the kinetics of immune cell responses within the tumor
   microenvironment as tumors progressed from early to late stages
   (Fig. [47]1b). As expected, flow cytometry analysis revealed a
   significant decline in both the number and percentage of
   tumor-infiltrating CD45^+ immune cells as tumors advanced (Fig. [48]1c,
   [49]e), a finding confirmed by immunofluorescence analysis of CD45^+
   cells in tumor sections (Fig. [50]1f, [51]h). This progressive decline
   in immune cell infiltration led to a global reshaping of the tumor
   immune compartment. We further validated this observation using the
   MMTV-PyMT spontaneous breast cancer genetically engineered mouse model
   (GEMM), which also demonstrated a progressive decrease in CD45^+ immune
   cell infiltration with tumor progression (Supplementary Fig. [52]1a, b
   and Supplementary Fig. [53]1e, f). This convergence of findings across
   both models confirms that our transplantation model effectively mirrors
   key aspects of immune modulation observed during spontaneous tumor
   development.

Fig. 1. Immune infiltration dynamics during breast cancer progression.

   [54]Fig. 1
   [55]Open in a new tab

   a Schematic diagram of the orthotopic breast cancer model utilizing
   MMTV-PyMT tumor cells inoculated into fat pad of immunocompetent FVB
   mice. Created in BioRender. [56]https://BioRender.com/penf9v4. b Tumor
   growth kinetics of MMTV-PyMT tumors in FVB mice (n = 5 mice per group).
   c–e Flow cytometry analysis of total immune cell infiltration in
   MMTV-PyMT tumors at different time points. c Percentage of CD45^+
   immune cells among live cells within the tumor. d Total number of
   CD45^+ immune cells isolated from per gram tumors. e Representative
   flow cytometry plots CD45^+ immune cells gated on live cells. (n = 5
   mice per group). f Representative immunofluorescence images of
   MMTV-PyMT tumor sections at indicated time points, stained for CD45
   (green) to visualize total immune cells. Scale bar: 100 μm. g
   Quantification of CD45^+ immune cell density (cells/mm^2) in MMTV-PyMT
   tumors. (Data represent the mean value from n = 5 biologically
   independent mice per group. For each mouse, 3 tumor sections were
   analyzed, and the average value was used for plotting and statistical
   analysis.). h Quantification of the distance between CD45^+ cells and
   the tumor margin in MMTV-PyMT tumors, assessed from immunofluorescence
   images. (Data represent the mean value from n = 5 biologically
   independent mice per group.). i, j Flow cytometry analysis of T cell
   and NK cells within the tumor-infiltrating lymphocyte population at
   different time points. i Representative flow cytometry plots gated on
   lymphoid cells (gated CD45^+ CD11B^- cells). j Quantification of T
   cells (CD3^+ NK1.1^-) as a percentage of lymphoid cells. The gating
   strategy for these panels is further detailed in Supplementary
   Fig [57]2a. (n = 5 mice per group). Data c–j are representative of two
   independent experiments from 5 mice per group per experiment. Data are
   presented as mean ± SD (d, e, g, h, j) and mean ± SEM (b). Statistical
   significance was calculated using one-way ANOVA (d, e, g, h, j). Source
   data are provided as a [58]Source Data file.

   To further dissect the evolving immune landscape within TME, we
   characterized the dynamic changes in immune cell subsets. We observed a
   marked shift in the balance between lymphoid and myeloid lineages
   within the CD45^+ compartment (Supplementary Fig. [59]2a, b).
   Specifically, the proportion of lymphoid cells decreased significantly,
   while myeloid cells became progressively enriched as tumors expanded
   (Supplementary Fig. [60]2a, b). Within the TME, infiltrating
   myeloid-derived suppressor cells (MDSCs) and tumor-associated
   macrophages (TAMs) play important regulatory roles in immunosuppression
   of tumor hosts to promote tumor progression and modulate the function
   of tumor-infiltrating lymphocytes^[61]12,[62]13. Among myeloid cells,
   MDSCs exhibited the most dramatic increase over time (Supplementary
   Fig. [63]2c, d), consistent with their established role in creating a
   pro-tumorigenic immunosuppressive environment^[64]14,[65]15, while TAMs
   proportions remained relatively stable (Supplementary Fig. [66]2c, d),
   as previously reported in established tumors^[67]16.

   While the percentage of total T cells progressively declined within the
   lymphoid compartment (Fig. [68]1i, j), the composition of the T cell
   compartment shifted. Although CD8^+ T cells as a percentage of total T
   cells showed a moderate increase, peaking in the middle and late stages
   of tumor growth (Supplementary Fig. [69]2e, f), the overall decline in
   total T cells suggests this did not represent a substantial increase in
   absolute CD8^+ T cell numbers. In contrast, the proportion of CD4^+ T
   cells steadily decreased throughout tumor progression (Supplementary
   Fig. [70]2e, f). Concurrently, we observed a marked expansion of both
   NK cell populations, increasing by ~5-fold during tumor development
   (Fig. [71]1i, Supplementary Fig. [72]2g). These immune shifts mirrored
   findings in the MMTV-PyMT GEMM model, where the proportion of CD3^+ T
   cells within CD45^+ immune cells progressively declined (Supplementary
   Fig. [73]1c, d), and CD8^+ T cell infiltration into tumor tissue
   diminished (Supplementary Fig. [74]1e, [75]g). Similarly, NK cell
   proportions gradually increased with tumor progression in the MMTV-PyMT
   model (Supplementary Fig. [76]2g), further underscoring the conserved
   nature of these immune dynamics. Collectively, these results
   demonstrate that the progressive decline in immune infiltration and the
   resulting remodeling of the tumor immune compartment recapitulate key
   features of immune responses observed in spontaneous breast cancer
   model.

   Importantly, RNA sequencing of intratumoral CD8^+ T cells isolated at
   different stages of tumor progression revealed a progressive
   upregulation of inhibitory receptors including Pd1, Lag3, and Tigit
   (Supplementary Fig. [77]2h), consistent with previously described
   patterns of T cell exhaustion^[78]17–[79]19. Notably, this upregulation
   of immune checkpoint molecules occurred simultaneously with increased
   expression of cytotoxic effector molecules like Granzyme B, IFNγ, and
   TNF-α, reflecting the complex dynamics of T cell responses within the
   tumor microenvironment^[80]19,[81]20. This pattern suggests a state of
   T cell exhaustion that develops progressively during tumor
   growth^[82]20,[83]21, characterized by the paradoxical co-expression of
   both inhibitory and effector programs^[84]22. Together, these results
   highlight the dynamic and evolving nature of the tumor immune
   microenvironment during breast cancer progression^[85]23. While our
   MMTV-PyMT transplantation model differs experimentally from spontaneous
   tumor development, this detailed characterization confirms that it
   effectively recapitulates key immunological features, including
   evolving immune-tumor cell interactions and the establishment of an
   immunosuppressive environment characteristic of breast cancer
   progression^[86]24,[87]25, thus validating its suitability for our
   subsequent investigations into the molecular drivers of these
   processes.

Mogat1 expression increases along breast tumor advancement and correlates
with lipid accumulation

   Tumor expansion necessitates acquiring immune evasion mechanisms that
   enable cancer cells to circumvent immune surveillance. To unveil the
   molecular underpinnings of tumor immune evasion and growth using the
   strategy outlined above, we performed RNA sequencing on GFP-expressing
   MMTV-PyMT tumor cells isolated at early (week 1–2), middle (week 3),
   and late (week 4) stages of tumor development in our model
   (Fig. [88]1a). Principal component analysis (PCA) and clustering
   analysis revealed distinct transcriptional profiles for each stage,
   highlighting the dynamic evolution of tumor cells during progression
   (Fig. [89]2a).

Fig. 2. Mogat1 expression increases along breast tumor advancement and
correlates with lipid accumulation.

   [90]Fig. 2
   [91]Open in a new tab

   a Principal component analysis (PCA) of MMTV-PyMT-GFP tumor cells
   sorted from tumor tissue at week 2, 3, and 4, based on variable gene
   expression. b Venn diagrams illustrating the overlap of upregulated
   genes in MMTV-PyMT tumors at week 3 versus week 2 and week 4 versus
   week 3. c Heatmap depicting differentially expressed genes in MMTV-PyMT
   tumors at week 2, 3, and 4, as determined by RNA sequencing. Each row
   represents a gene, and each column represents a sample. d Volcano plots
   show casing differential gene expression in MMTV-PyMT tumor cells at
   week 4 versus week 3. Red dots indicate upregulated genes (log2 fold
   change [FC] > 1, false discovery rate [FDR] < 0.05), and green dots
   indicate downregulated genes (log2FC < −1, FDR < 0.05). e Log2FC
   expression values of upregulated genes associated with enriched Gene
   Ontology (GO) terms in MMTV-PyMT tumors at week 4 versus week 3. f
   RNA-seq analysis of Mogat1 expression in MMTV-PyMT tumors at weeks 2–4
   post-implantation. (n = 3 mice in week 2, n = 4 mice in week 3 and week
   4). The median value (center line), lower quartile and upper quartile
   (box edges) and maximum and minimum value whiskers are indicated in the
   boxplot. g qPCR validation of Mogat1 expression in tumor cells sorting
   from MMTV-PyMT tumors at week 1–4 post-tumor induction, normalized to
   Gapdh and relative to week 1. (n = 3 biological tumor samples per
   group). h qPCR validation of Mogat1 expression in MMTV-PyMT GEMM tumors
   at week 8–11. (n = 5 biological tumor samples per group). i Relative
   Mogat1 expression by qPCR in human breast tumors and tumor-adjacent
   normal breast. (n = 30 breast tumors and n = 30 tumor-adjacent normal
   breast were analyzed. Statistics were by two-tailed paired student’s
   t-test.). j, k Representative IHC images and quantification of Mogat1
   and Ki67 expression in the indicated human breast tumor tissues and
   tumor-adjacent normal breast tissues. (n = 30 breast tumors and n = 30
   tumor-adjacent normal breast were analyzed. Statistics were by
   two-tailed unpaired student’s t-test.) Data are representative of two
   independent experiments (n = 3-4 for RNA-seq). Statistical significance
   was calculated using one-way ANOVA (f, g, h). Source data are provided
   as a [92]Source Data file.

   Intriguingly, our analysis identified 34 genes that were consistently
   upregulated throughout tumor development from early to late stages
   (Fig. [93]2b, [94]c). Among these 34 genes, Mogat1, encoding
   monoacylglycerol O-acyltransferase 1, a key enzyme in triglyceride
   synthesis^[95]8, emerged as the prominently upregulated gene in
   late-stage tumors (Fig. [96]2c, d and Supplementary
   Fig. [97]3a).Differential gene expression analysis revealed a
   significant upregulation of genes associated with energy metabolism,
   particularly lipid metabolism, as tumors progressed. Compared to
   early-stage tumors, middle-stage tumors exhibited significant
   enrichment of pathways related to negative regulation of lipid
   transport, PI3K signaling, and cellular metabolic processes
   (Supplementary Fig. [98]3b). This metabolic shift persisted in
   late-stage tumors, which showed further enrichment of lipid and fatty
   acid metabolism pathways (Fig. [99]2e). Collectively, these
   transcriptomic shifts suggest that adaptations in lipid metabolism,
   potentially involving Mogat1, may play an important role during tumor
   development.

   We sought to define Mogat1’s unique role by comparing its expression
   with other triglyceride synthesis genes in the monoacylglycerol
   (Mogat2, Dgat1, Dgat2; 26) and glycerol-3-phosphate (Gpat, Agpat1-5;
   27) pathways (Supplementary Fig. [100]3c, d). Notably, while most other
   lipid metabolism synthases, including Agpat3-5, exhibited unchanged or
   decreased expression, Mogat1 was consistently upregulated throughout
   tumor progression (Fig. [101]2f and Supplementary Fig. [102]3c, d), an
   observation corroborated by qPCR (Fig. [103]2g). In the MMTV-PyMT GEMM
   model, we observed higher mogat1 expression in tumor tissues compared
   to normal mammary glands, with expression further increasing during
   tumor progression (Fig. [104]2h), while its levels remained stable in
   normal mammary glands across different ages (Supplementary
   Fig. [105]3e). This pattern highlights the specific and sustained
   upregulation of Mogat1 during tumor development. To investigate whether
   this translated into changes in lipid storage, we examined lipid
   droplet content in tumor tissues and normal breast tissue. Oil Red O
   staining revealed a progressive increase in lipid droplets with tumor
   development (Supplementary Fig. [106]3f, g), while they remained stable
   in normal mammary tissue (Supplementary Fig. [107]3h, i). The
   observation of increased Mogat1 expression and lipid droplet
   accumulation occurring concurrently with the previously noted decline
   in immune cell infiltration (Fig. [108]1) raises the hypothesis that
   Mogat1-mediated alterations in lipid metabolism might contribute to
   both tumor growth and the establishment of an altered immune
   environment.

   Clinical validation in 30 paired breast cancer specimens further
   revealed significantly elevated Mogat1 RNA levels in tumors versus
   adjacent normal tissue (Fig. [109]2i), paralleled by heightened Mogat1
   protein expression and Ki67^+ proliferation in tumor cells
   (Fig. [110]2j, k). Pan-cancer analysis of TCGA datasets confirmed
   Mogat1 upregulation across multiple malignancies (Supplementary
   Fig. [111]4a), with high expression correlating with poorer overall
   survival in breast cancer (BRCA, Luminal A BRCA), thymic cancer (THYM),
   glioma (LGG), and HPV+ head and neck squamous carcinoma (HPV-HNSC)
   (Supplementary Fig. [112]4b). Notably, Mogat1 expression was highest in
   triple-negative breast cancer (TNBC) compared to luminal or HER2^+
   subtypes (Supplementary Fig. [113]4c), aligning with its aggressive
   clinical course. These findings establish Mogat1 as a potentially
   conserved factor associated with tumor progression and poorer prognosis
   in both murine models and diverse human cancers. The consistent
   correlation between stage-specific Mogat1 upregulation, lipid pathway
   enrichment, increased lipid storage in vivo, and adverse clinical
   outcomes strongly suggests that alterations in lipid metabolism
   involving Mogat1 represent an important facet of tumor adaptation,
   warranting functional investigation into its specific roles in driving
   tumor growth and influencing the immune response, which we explore in
   subsequent sections.

Mogat1 inhibition suppresses lipid droplet formation

   Given Mogat1’s localization to the endoplasmic reticulum (ER), the
   primary site of lipid droplet formation (Fig. [114]3b and Supplementary
   Fig. [115]5a), we investigated its role in lipid metabolism within
   cancer cells, focusing on lipid droplet accumulation and diacylglycerol
   (DAG) production, the precursor of triacylglycerols (TG) and
   phospholipids (Fig. [116]3a,^[117]26,[118]27). To functionally assess
   Mogat1, we used short hairpin RNAs (shRNAs) to achieve robust knockdown
   in a panel of human (MDA-MB231) and mouse (MMTV-PyMT, 4T1, B16F10)
   cancer cell lines. This efficient Mogat1 suppression enabled systematic
   dissection of its lipid metabolic functions (Fig. [119]3c and
   Supplementary Fig. [120]5b). We first performed colony formation assays
   to assess Mogat1’s impact on long-term tumor cell survival. Consistent
   with this, Mogat1 knockdown significantly suppressed colony formation
   across all models(Supplementary Fig. [121]5c), while restoring Mogat1
   expression in depleted cells reversed the colony formation defects
   (Supplementary Fig. [122]5c). Our findings suggest a crucial role for
   Mogat1 in supporting the sustained growth and survival required for
   colony establishment, likely through its regulation of lipid
   metabolism.

Fig. 3. Mogat1 inhibition suppresses lipid droplet formation.

   [123]Fig. 3
   [124]Open in a new tab

   a Schematic of the glycerol phosphate pathway and the monoacylglycerol
   pathway for triacylglycerol (TAG) synthesis. Key enzymes are indicated:
   GPAT (glycerol-3-phosphate acyltransferase), AGPAT
   (1-acylglycerol-3-phosphate acyltransferase), PAP (phosphatidic acid
   phosphatase), DGAT (diacylglycerol acyltransferase), MGAT
   (monoacylglycerol acyltransferase). b Confocal microscopy analysis of
   Mogat1 localization in MDA-MB-231 cells overexpressing Mogat1 tagged
   with neogreen (Mogat1-neogreen). Cells were stained for endoplasmic
   reticulum (ER) markers Calnexin and Calreticulin. Scale bar: 10 μm. c
   Western Blot of Mogat1 knockdown efficiency and Mogat1 overexpression
   in MMTV-PyMT and MDA-MB231 tumor cells. Quantification of d
   diacylglycerol (DAG) and e triacylglycerol (TG) content in MMTV-PyMT
   and MDA-MB231 tumor cells. (n = 4 biological cell cultures in data d,
   n = 3 biological cell cultures in data e). f Analysis of lipid droplet
   content in MMTV-PyMT and MDA-MB231 tumor cells. Representative images
   of Oil Red O staining and quantification of Oil Red O staining
   intensity. Scale bar: 20 μm. (n = 6 biological cell cultures). g
   Representative confocal microscopy images of lipid droplets in
   MMTV-PyMT and MDA-MB231 tumor cells following Mogat1 knockdown and
   mogat1 overexpression. Lipid droplets are visualized with BODIPY
   493/503 (green). Scale bar: 20 μm. (n = 6 biological cell cultures).
   h–j Confocal microscopy analysis of lipid droplet marker Perilipin 2
   (PLIN2) in h MDA-MB231 and i MMTV-PyMT cells. j Quantification of lipid
   droplet marker Perilipin 2. Scale bar: 20 μm. (n = 6 biological cell
   cultures). k, l Flow cytometry analysis of BODIPY 493/503 staining (a
   measure of neutral lipids) in MMTV-PyMT and MDA-MB231 tumor cells. k
   Representative histograms. l Mean fluorescence intensity (MFI) values.
   (n = 4 biological cell cultures). m, n Assessment of fatty acid uptake
   in MMTV-PyMT and MDA-MB231 tumor cells with or without Mogat1 knockdown
   and mogat1 overexpression using BODIPY FL C16. m Representative
   histograms. n Quantification of BODIPY FL C16 uptake. (n = 4 biological
   cell cultures). Data in (d, e, f, g, j, l, n) are presented as the
   mean ± SD; Statistical significance was calculated using one-way ANOVA
   (d, e, f, g, j, l, n). ns, not significant. Representative immunoblots
   and gels shown in (c) were repeated at least two times independently
   with similar results. Source data are provided as a [125]Source Data
   file.

   We next examined the effect of Mogat1 on lipid synthesis. Mogat1
   knockdown significantly reduced DAG and TG levels in all tested tumor
   cells (Fig. [126]3d, e and Supplementary Fig. [127]5d, e). This
   correlated with a substantial reduction in lipid droplet accumulation,
   confirmed by Oil Red O (Fig. [128]3f and Supplementary Fig. [129]5g),
   BODIPY 493/503 (Fig. [130]3g and Supplementary Fig. [131]5h). Perilipin
   2 staining further revealed impaired lipid droplet assembly upon Mogat1
   knockdown (Fig. [132]3h–j). Conversely, restoration of Mogat1 rescued
   DAG and TG levels, lipid accumulation, and lipid droplet assembly.
   Notably, Mogat1 inhibition did not alter expression of
   glycerol-3-phosphate pathway enzymes (GPAT and AGPAT, Supplementary
   Fig. [133]5f), mirroring RNA-seq findings from progressive tumor stages
   (Fig. [134]2 and Supplementary Fig. [135]3c, d). Together, these data
   establish Mogat1 as a non-redundant regulator of tumor lipid droplet
   biogenesis and metabolic plasticity.

   Interestingly, Mogat1 knockdown did not affect fatty acid uptake or the
   expression of fatty acid translocase protein (Cd36, Cd68, Ld1r, Scarb1,
   Lrp1, and Slc27a1-4), as demonstrated by BODIPY FL C16 staining
   (Fig. [136]3m, n and Supplementary Fig. [137]5j) and qPCR analysis
   (Supplementary Fig. [138]5k). These findings indicate that Mogat1’s
   primary role is in lipid synthesis rather than uptake of exogenous
   lipids. Collectively, these findings demonstrate that Mogat1 promotes
   tumor cell growth by enhancing intracellular lipid synthesis.

Mogat1 inhibition effectively suppresses tumor growth in vivo

   To validate the in vivo relevance of Mogat1 in tumor progression, we
   established mouse allograft models using multiple cancer cell lines.
   Consistently, MMTV-PyMT cells with Mogat1 knockdown exhibited
   significantly slower tumor growth compared to controls, as evidenced by
   reduced tumor volume and weight (Fig. [139]4a–c and Supplementary
   Fig. [140]6a). This effect was replicated in 4T1 breast cancer
   allograft (Fig. [141]4d–f and Supplementary Fig. [142]6b) and further
   corroborated using B16F10 melanoma cells (Fig. [143]4g–i and
   Supplementary Fig. [144]6c), demonstrating the consistency of Mogat1’s
   tumor-promoting role across different cancer types.

Fig. 4. Mogat1 inhibition suppresses tumor growth in vivo.

   [145]Fig. 4
   [146]Open in a new tab

   a–c Growth curves of MMTV-PyMT tumors in immunocompetent FVB mice
   following inoculation with cells transduced with control shRNA (shCTRL,
   black), shMogat1_1 (blue), or shMogat1_2 (red). b Quantification of
   tumor volume and c weight at day 36 in the MMTV-PyMT model. (n = 8 mice
   in shCTRL and shMogat1_1 group, n = 7 mice in shMogat1_2 group). d–f
   Growth curves of 4T1 tumors in immunocompetent Balb/c mice following
   inoculation with cells transduced with shCTRL (black), shMogat1_1
   (blue), or shMogat1_2 (red). e Quantification of tumor volume and f
   weight at day 25 in the 4T1 model. (n = 8 mice in shCTRL and shMogat1_1
   group, n = 7 mice in shMogat1_2 group). g–i Growth curves of B16/F10
   tumors in immunocompetent C57BL/6 mice following inoculation with cells
   transduced with shCTRL (black), shMogat1_1 (blue), or shMogat1_2 (red).
   h Quantification of tumor volume and i weight at day 17 in the B16/F10
   model. (n = 6 mice per group). j, k Oil Red O staining of neutral
   lipids in tumor tissues from MMTV-PyMT mice with control shRNA (shCTRL)
   or two independent Mogat1 shRNAs (shMogat1_1, shMogat1_2). Scale bar:
   100 μm. (n = 6 mice per group). Data are representative of two
   independent experiments from 6 to 8 mice per group per experiment. Data
   are presented as mean ± SD (b, c, e, f, h, i, k) and mean ± SEM (a, d,
   g). Statistical analysis was conducted by two-way ANOVA followed by
   Tukey’s multiple comparison test (a, d, g). Statistical significance
   was calculated using one-way ANOVA (b, c, e, f, h, i, k). ns, not
   significant. Source data are provided as a [147]Source Data file.

   Our initial screening revealed a striking correlation between Mogat1
   expression and enhanced lipid droplet formation in breast cancer cells
   (Supplementary Fig. [148]3f, g). To establish a causal relationship, we
   investigated the effects of Mogat1 inhibition in both breast and
   melanoma tumors. Notably, Mogat1 knockdown resulted in a dramatic
   reduction of lipid droplets (Fig. [149]4j, k and Supplementary
   Fig. [150]6d, e), coinciding with significant tumor regression
   (Fig. [151]4a–i). Because triglyceride synthesis and subsequent lipid
   droplet formation can occur via two primary pathways: the
   monoacylglycerol pathway and the glycerol-3-phosphate
   pathway^[152]27,[153]28. The specific and substantial reduction in
   lipid droplets following Mogat1 knockdown underscores its critical and
   non-redundant role in this process. Notably, the inability of other
   enzymes in these pathways to compensate for Mogat1 depletion highlights
   its unique function in triglyceride synthesis and lipid droplet
   formation.

   This in vivo tumor data not only confirms the specific role of Mogat1
   in breast cancer progression but also reveals its broader implications
   in cancer biology. The consistent effects observed across different
   cancer types suggest that Mogat1-mediated lipid metabolism may
   represent a conserved mechanism in tumor development and maintenance.
   This reduction in tumor size also inversely correlated with enhanced
   immune cell infiltration (see more below). Collectively, these findings
   validate the robustness of our in vivo selection and transcriptomic
   profiling strategy and highlight Mogat1-driven lipid metabolism as a
   critical fuel source for tumor expansion.

   To elucidate the molecular mechanisms underlying Mogat1’s effects, we
   performed RNA-seq on MMTV-PyMT cells isolated from allograft
   (Supplementary Fig. [154]6f, g). Hierarchical clustering clearly
   distinguished Mogat1 knockdown from control tumors (Supplementary
   Fig. [155]6f). Differential expression analysis revealed 401
   significantly upregulated and 279 downregulated genes in
   Mogat1-depleted tumors (LogFC > 1 or <−1, P < 0.05; Supplementary
   Fig. [156]6f). Functional enrichment analysis of upregulated genes
   highlighted several key processes affected by Mogat1 knockdown,
   including negative regulation of cell proliferation, positive
   regulation of programmed cell death, metabolic process regulation
   (Supplementary Fig. [157]6g). Conversely, downregulated genes were
   enriched in pathways related to lipid export, cell growth, calcium ion
   transport regulation, and cell communication (Supplementary
   Fig. [158]6g). These findings align with our in vitro observations and
   provide mechanistic insights into Mogat1’s role in tumor progression.
   Collectively, these results demonstrate that Mogat1 knockdown
   significantly impairs tumor growth in vivo across multiple cancer
   models. The transcriptomic changes induced by Mogat1 depletion suggest
   a multifaceted role in regulating cancer cell proliferation, survival,
   and metabolism, underscoring its crucial role in tumor progression
   through changes of lipid droplet formation.

Mogat1 inhibition disrupts lipid homeostasis and energy metabolism in breast
cancer

   To investigate the metabolic consequences of Mogat1 inhibition in
   breast cancer cells, we performed non-targeted metabolomics profiling
   on different breast cancer subtypes GFP-labeled 4T1 tumor cells (TNBC)
   (Fig. [159]5) and MMTV-PyMT tumor cells (luminal) (Supplementary
   Fig. [160]7a) isolated from allograft tumors. Principal component
   analysis of the metabolomics data revealed distinct metabolic profiles
   between Mogat1 knockdown and control 4T1 tumor groups (Fig. [161]5a),
   with 488 metabolites downregulated and 327 metabolites upregulated upon
   Mogat1 knockdown (Fig. [162]5b, c).While multiple metabolite changes
   were also found in MMTV-PYMT tumors (Supplementary Fig. [163]7a, b).

Fig. 5. Mogat1 inhibition reprograms breast cancer cell metabolism by
suppressing TCA cycle activity.

   [164]Fig. 5
   [165]Open in a new tab

   a Orthogonal partial least squares discriminant analysis (OPLS-DA)
   score plot comparing the metabolic profiles of shCTRL and shMogat1 4T1
   tumor cells (n = 3 biological replicates per group). b Volcano plot
   depicting differentially abundant metabolites between shCTRL and
   shMogat1 4T1 tumor cells. Metabolites with a variable importance in
   projection (VIP) score > 1 from the OPLS-DA analysis and a P < 0.05
   from two-tailed unpaired Student’s t-test are highlighted. c Heatmap
   illustrating the relative abundance of differentially expressed
   metabolites (VIP > 1, P < 0.05) identified by non-targeted mass
   spectrometry. d Schematic diagram depicting metabolic flux from glucose
   and glutamine into the tricarboxylic acid (TCA) cycle and glycolysis
   pathways. e–g Targeted mass spectrometry analysis of TCA cycle
   metabolites in shCTRL and shMogat1 tumor cells. f 4T1 cells. g
   MMTV-PyMT cells. h B16/F10 cells. (n = 4 biological cell cultures per
   group). h Heatmap displaying the enrichment of KEGG metabolic pathways
   in shCTRL and shMogat1 4T1 tumor cells, based on gene expression data.
   Data are representative of two independent experiments. Data in (e, f,
   g) are presented as the mean ± SD. Statistical significance was
   determined using a two-tailed unpaired Student’s t-test. Source data
   are provided as a [166]Source Data file.

   As expected, Mogat1 inhibition led to a significant decrease in neutral
   lipids, particularly triglycerides and diacylglycerols (Fig. [167]5c).
   We also observed substantial changes in membrane phospholipids,
   including phosphatidylcholine (PC) and phosphatidylethanolamine (PE),
   consistent with Mogat1’s role in DAG synthesis and subsequent lipid
   metabolism. Intriguingly, these lipid alterations were accompanied by
   marked changes in central carbon metabolism. Levels of glutamate, a key
   intermediate in glutamine metabolism and the TCA cycle, were
   significantly altered in Mogat1 knockdown cells, with a notable
   decrease in malate, suggesting impaired TCA cycle activity
   (Fig. [168]5c, d). Targeted metabolomics analysis confirmed significant
   downregulation of multiple TCA cycle metabolites upon Mogat1 knockdown
   in breast cancer (Fig. [169]5e–g). Further analysis revealed that
   Mogat1 knockdown impacted a broad spectrum of metabolic pathways,
   including glycerophospholipid metabolism, carbon metabolism, fatty acid
   metabolism, nucleotide metabolism, and sphingolipid signaling
   (Fig. [170]5h, Supplementary Fig. [171]7c).

   These findings demonstrate that Mogat1 inhibition not only disrupts
   lipid droplet formation but also substantially reprograms cellular
   metabolism in breast cancer cells, including suppressing the TCA cycle
   and altering phospholipid and fatty acid metabolism. These significant
   alterations in intrinsic tumor cell metabolism suggest a disruption of
   core bioenergetic and biosynthetic pathways necessary for rapid tumor
   expansion. While these cell-autonomous metabolic effects likely
   contribute to impairing tumor fitness, their relative contribution
   versus the impact on the immune microenvironment in mediating the
   overall tumor suppression observed in vivo required further
   investigation, as explored in subsequent sections.

Mogat1 loss inflames tumor microenvironment

   Given the critical role of the tumor microenvironment (TME) in shaping
   anti-tumor immunity, we next investigated the impact of tumor with
   Mogat1 depletion on immune cell infiltration. T-Distributed Stochastic
   Neighbor Embedding (tSNE) analysis of immune profiling data by FACS
   revealed a significant increase in immune cell infiltration into
   Mogat1-depleted tumors compared to controls, with a concomitant
   decrease in the proportion of tumor cells (Fig. [172]6a, d,
   Supplementary Fig. [173]8a–c).

Fig. 6. Mogat1 inhibition promotes an inflamed tumor microenvironment.

   [174]Fig. 6
   [175]Open in a new tab

   a–c t-distributed stochastic neighbor embedding (t-SNE) plots
   visualizing single-cell flow cytometry data from shCTRL and shMogat1
   MMTV-PyMT tumors. a Total immune and non-immune cell populations. b All
   cell populations. c Subsets of CD4^+ and CD8^+ T cells based on PD1
   expression. The comprehensive flow cytometry gating strategy for these
   panels is presented in Supplementary Fig. [176]8a, b. d–g
   Quantification of immune cell populations in shCTRL and shMogat1
   MMTV-PyMT tumors by flow cytometry. These quantitative data (d) for
   total immune cells (CD45^+). e for total CD3^+T cells and T cell sub
   populations. f for B, NK and DC cells. g for tumor-associated
   macrophages (TAMs, CD45^+CD11b^+F4/80^+) and tumor-associated
   neutrophils (TANs, CD45^+CD11b^+Ly6G^+), were all derived using the
   gating strategies presented in Supplementary Fig. [177]8a, b. n = 5
   mice per groups. h Representative immunofluorescence images of shCTRL
   and shMogat1 MMTV-PyMT tumor sections stained for CD8 (red) to
   visualize cytotoxic T cell infiltration. Scale bar: 100 μm. i, j FACS
   analysis of the IFNγ, Granzyme B (i) and exhaustion marker TOX, CTLA-4
   and LAG3 expressing CD8 T cells infiltrating MMTV-PyMT tumors. n = 5
   mice per groups. In vitro cytotoxicity assay assessing the
   susceptibility of shCTRL or shMogat1 (k) MMTV-PyMT-OVA and (n)
   B16/F10-OVA tumor cells to CD8^+ T cell-mediated killing.
   Quantification of tumor cell killing at various E:T ratios. (n = 4
   biological cell cultures per group). IFNγ secretion by CD8^+ T cells
   following co-culture with shCTRL or shMogat1 l MMTV-PyMT-OVA and o
   B16-OVA cells, as measured by ELISA. (n = 4 biological cell cultures
   per group). Flow cytometry analysis of the frequency of Granzyme
   B-positive (GraB^+) CD8^+ T cells after co-culture with shCTRL or
   shMogat1 (m) MMTV-PyMT-OVA and (p) B16-OVA cells. (n = 4 biological
   cell cultures per group in data m, n = 3 biological cell cultures per
   group in data p). q, r Schematic representation of a transwell assay
   involving in vitro activated OT1 CD8^+ T cells combined with shCTRL or
   shMogat1 MMTV-PyMT-OVA tumor cells media. r CD8^+ T Cell migration and
   PI^+ dead cells were analyzed by transwell assay (n = 4 biological cell
   cultures in shCTRL group, n = 3 biological cell cultures in shMogat1
   group). q is created in BioRender. [178]https://BioRender.com/k35b440
   (n = 4 biological cell cultures in shCTRL group, n = 3 biological cell
   cultures in shMogat1 group). Data (a–j) are representative of two
   independent experiments from 5 mice per group per experiment. Data
   (k–r) are representative of two independent experiments. Data are
   presented as mean ± SD. Statistical significance was calculated using
   unpaired Student’s t tests. ns, not significant. Source data are
   provided as a [179]Source Data file.

   Specifically, we observed significant increases in the infiltration of
   T cells, B cells, NK cells, and dendritic cells (DCs) (Fig. [180]6b–f).
   Mogat1 knockdown tumors had significantly decreased proportions of
   tumor-associated macrophages(TAMs), while tumor-associated neutrophils
   are no change (Fig. [181]6b, g), suggesting a shift towards a more
   immunogenic TME. Notably, Mogat1 knockdown led to a substantial
   increase in the proportion of tumor-infiltrating T cells (Fig. [182]6b,
   c, e), tumor tissue analysis using specific anti-CD8 antibody confirmed
   the increased abundance of intra-tumoral CD8^+ cytotoxic T cells in
   Mogat1-knockdown tumors (Fig. [183]6h). These suggest that Mogat1
   knockdown creates a more permissive tumor microenvironment for T-cell
   entry, which can be crucial for mounting an effective immune response
   against the tumor. Interestingly, the proportion of PD-1^+ CD8^+ T
   cells was markedly increased (Fig. [184]6c, e, right). PD-1 expression
   on CD8^+ T cells typically indicates an exhausted state due to chronic
   antigen exposure^[185]29–[186]32. In contrast, the proportion of PD-1^+
   CD4^+ T cells did not change, indicating that Mogat1 inhibition
   specifically affects CD8^+ T cell activation without altering the
   activation state of CD4^+ T cells (Fig. [187]6e right). Intriguingly,
   we observed a marked increase in PD-1^- CD4^+ and PD-1^- CD8^+ T cells
   within Mogat1-knockdown tumors (Fig. [188]6c, e), suggesting that a
   portion of the T cell population remains in a non-exhausted,
   potentially more functional state, which could contribute to a more
   robust anti-tumor immune response^[189]33–[190]35.

   To directly assess the functional state of tumor-infiltrating CD8^+ T
   cells in Mogat1-knockdown tumors, we analyzed expression of key
   effector and exhaustion markers. Flow cytometric analysis revealed that
   Mogat1 knockdown significantly increased the expression of the
   cytotoxic cytokines IFNγ and Granzyme B in tumor-infiltrating CD8^+ T
   cells (Fig. [191]6i). Furthermore, these cells exhibited reduced
   expression of the exhaustion markers TOX, CTLA-4, and LAG3
   (Fig. [192]6j), indicating that Mogat1 depletion enhances CD8^+ T cell
   cytotoxicity while mitigating exhaustion.

   Importantly, RNA sequencing of CD8^+ T cells isolated from
   Mogat1-depleted tumors revealed a distinct transcriptional profile
   compared to controls (Supplementary Fig. [193]8d). Notably, T cells
   from Mogat1-knockdown tumors exhibited upregulated pathways associated
   with proliferation, differentiation, motility, and cell adhesion,
   suggesting a more active and functional state^[194]29–[195]31.
   Conversely, pathways related to T cell death, stress response, hypoxia
   response, and lipid metabolism were downregulated, suggesting a shift
   away from a dysfunctional and exhausted phenotype (Supplementary
   Fig. [196]8d). Next, we examine whether enhanced T cell infiltration
   translates into direct increased tumor cell killing, we performed in
   vitro co-culture assays using primary T cells and either wild-type or
   Mogat1-knockdown MMTV-PyMT-OVA breast cancer and B16-OVA melanoma
   cells. Mogat1 knockdown did not affect tumor cell proliferation
   (Supplementary Fig. [197]8E). Importantly, Mogat1 depletion
   significantly increased tumor cell susceptibility to T cell-mediated
   lysis at all tested effector-to-target ratios (Fig. [198]6k, n, and
   Supplementary Fig. [199]8f, g). Crucially, control wells containing
   tumor cells incubated without T cells showed minimal background lysis,
   confirming that the observed tumor cell killing was indeed specifically
   mediated by the co-cultured T cells (Fig. [200]6k, n). This enhanced
   susceptibility to T cell lysis was accompanied by increased T cell
   cytotoxic function, as evidenced by significantly higher secretion of
   IFNγ and Granzyme B (Fig. [201]6l, m, o, p). Furthermore, chemotaxis
   assays demonstrated that supernatants from Mogat1-depleted tumor cells
   enhanced the migration and reduced cell death of activated OT1 CD8^+ T
   cells (Fig. [202]6q, r), suggesting that Mogat1 knockdown promotes
   CD8^+ T cell chemotaxis and infiltration into the tumor
   microenvironment via secreted factors. To test whether such secreted
   factors also modulate T cell effector function, we cultured activated
   CD8^+ T cells with conditioned media (CM) harvested from shCTRL or
   shMogat1 tumor cells. Indeed, CD8^+ T cells incubated with CM from
   Mogat1-depleted cells displayed significantly increased intracellular
   expression of both IFNγ and Granzyme B compared to those cultured with
   control CM (Supplementary Fig. [203]9a, b). This provides direct
   evidence that soluble factors released by Mogat1-deficient tumor cells
   contribute to the enhancement of CD8^+ T cell cytotoxic potential.
   Together, these in vitro findings demonstrate that Mogat1 depletion
   renders tumor cells more susceptible to T cell killing while
   simultaneously enhancing T cell effector functions, partly through the
   modulation of secreted factors, suggesting a multifaceted role in
   shaping anti-tumor immunity within the TME.

Mogat1 inhibition enhances T cell-mediated anti-tumor immunity in vivo

   The enhanced T cell infiltration and intrinsic cytotoxic capacity
   prompted us to determine whether T cells are directly responsible for
   restricting tumor growth upon Mogat1 depletion in vivo. To address
   this, we first utilized an orthotopic MMTV-PyMT breast cancer model. T
   cell depletion via anti-CD4/8 antibody injection during tumor
   progression (Fig. [204]7a, b) revealed that Mogat1 knockdown-mediated
   tumor growth suppression was abrogated in T cell-deficient mice
   (Fig. [205]7c). Tumor growth rate, volume, and weight in
   Mogat1-knockdown, T cell-depleted animals were comparable to controls
   (Fig. [206]7d, e). These findings indicate that the anti-tumor immune
   response elicited by Mogat1 loss is primarily T cell-dependent.

Fig. 7. Mogat1 loss in tumor cells stimulate anti-tumor immunity by T cells
in vivo.

   [207]Fig. 7
   [208]Open in a new tab

   a Schema of CD4/8 antibody depletion in orthotopic MMTV-PyMT breast
   tumor-bearing immunocompetent FVB mice. b Frequency of CD8^+ and CD4^+
   T cells within tumor after inject IgG or CD4/8 antibody with flow
   cytometry in day 26. See Supplementary Fig. [209]8a, b for gating
   strategy. c Tumor growth of FVB immunocompetent mice inoculated with
   shCTRL and shMogat1 MMTV-PyMT tumor cells versus T cell depletion.
   (n = 6-8 mice per group). d, e Tumor volumes and tumor weight from (c)
   at day 26. (n = 8 mice in anti-IgG group; In anti-CD4/8 group, n = 7
   mice in shCTRL and shMogat1_2 group, n = 6 mice in shMogat1_1 group). f
   Schematic representation of MDA-MB231 orthotopic tumor model in severe
   immunodeficient NOG mice. g Tumor growth of severe immunodeficient NOG
   mice inoculated with MDA-MB231 tumor cells. h Tumor volumes and tumor
   weight at day 34. (n = 7 mice in shCTRL and shMogat1 group, n = 6 mice
   in shMogat1+ OE group). i Schematic representation of MDA-MB231
   orthotopic tumor model in MHCI/II-DKO NOG mice tail vein injected human
   PBMC at day 0 and day 16. (n = 6–7 mice per group). j Tumor growth of
   MHCI/II-DKO NOG mice inoculated with MDA-MB231 tumor cells. k Tumor
   volumes and tumor weight at day 36. (n = 6 mice in shCTRL and shMogat1
   group, n = 7 mice in shMogat1+ OE group). l The proportion of
   tumor-infiltrating human CD8^+ T cells was detected by flow cytometry
   in MHCI/II-DKO NOG mice at day 36. Gated mCD45^+hCD45^+ cells. (n = 5
   mice per group). The flow cytometry gating strategy is presented in
   Supplementary Fig. [210]9c, d. m Schema of B16/F10 melanoma animal
   model in Rag1^-/- mice. (n = 5 mice per group). n Growth curves of
   B16/F10 tumors in immunodeficient Rag1^-/- mice. o Quantification of
   tumor volume at day 15 in the B16/F10 model. (n = 5 mice per group).
   Data are representative of two independent experiments. Data are
   presented as mean ± SD (d, e, h, k, l, o) and mean ± SEM (c, g, j, n).
   Statistical analysis was conducted by two-way ANOVA followed by Tukey’s
   multiple comparison test (c, g, j, n). Statistical significance was
   calculated using unpaired Student’s t tests (o) and one-way ANOVA (d,
   e, h, k, l). ns, not significant. Figure (a, f, i, m) are created in
   BioRender. [211]https://BioRender.com/yduci19. Source data are provided
   as a [212]Source Data file.

   Next, we verify our findings in human cancer model by employing the
   human breast cancer cell line MDA-MB-231 in a severely immunodeficient
   NOG mouse model (Fig. [213]7f). Consistent with previous results,
   Mogat1 knockdown did not significantly affect the growth rate, tumor
   volume, or tumor weight of MDA-MB-231 xenografts compared to controls
   (Fig. [214]7g, h). To explore the anti-tumor immune responses induced
   by Mogat1 knockdown, we orthotopically implanted MDA-MB-231 cells and
   subsequently transferred human peripheral blood mononuclear cells
   (PBMCs) into NOG mice deficient in both MHC class I and II molecules
   (Fig. [215]7i). By day 16, human immune cells were successfully
   detected in the mice’s peripheral blood, prompting a second PBMC
   injection (Supplementary Fig. [216]9c). Tumor growth analysis revealed
   that PBMC infusion significantly suppressed tumor growth in the Mogat1
   knockdown group compared to the shCTRL group, whereas Mogat1
   overexpression rescued tumor growth, as demonstrated by both tumor
   volume and weight measurements (Fig. [217]7j, k). Flow cytometry
   indicated that intra-tumoral human immune cells were predominantly CD8⁺
   T cells (Supplementary Fig. [218]9d). Notably, the proportion of CD8⁺ T
   cells was significantly elevated in Mogat1-knockdown tumors compared to
   the shCTRL group, and this increase was abrogated by Mogat1
   overexpression (Fig. [219]7l).

   Finally, to validate our findings in an alternative tumor model, we
   employed the B16/F10 melanoma system in Rag1-/- mice, which are
   deficient in T and B cells (Fig. [220]7m). Consistent with observations
   in immunocompetent mice (Fig. [221]4c), Mogat1 knockdown did not induce
   tumor shrinkage in Rag1-/- mice (Fig. [222]7n, o). These collective
   findings across multiple models, including T-cell depletion in
   immunocompetent mice (Fig. [223]7c–e), lack of efficacy in
   immunodeficient NOG and Rag1-/- mice (Fig. [224]7g, h, n, o), and
   PBMC-dependent suppression in a humanized system (Fig. [225]7j–l),
   demonstrate that the significant tumor suppression mediated by Mogat1
   knockdown in vivo is critically dependent on T cell-mediated adaptive
   immunity. While Mogat1 inhibition clearly impacts intrinsic tumor cell
   metabolism (Figs. [226]3, [227]5), these in vivo results highlight the
   predominant role of extrinsic, immune-mediated effects in controlling
   tumor growth in an immune-competent setting. Collectively, our data
   show that Mogat1 depletion enhances T cell infiltration and cytotoxic
   capacity (Figs. [228]6, [229]7), suggesting Mogat1 inhibition operates
   through a mechanism involving substantial immune modulation,
   potentially complementing existing immunotherapies.

Synergistic enhancement of anti-tumor immunity by Mogat1 knockdown and PD-1
blockade

   While essential for preventing autoimmunity by dampening excessive T
   cell activation, the PD-1 pathway is often exploited by tumors to evade
   immune destruction^[230]36,[231]37. Tumor-infiltrating T cells,
   chronically exposed to tumor antigens, upregulate PD-1, effectively
   hindering their cytotoxic potential. Intriguingly, Mogat1 depletion not
   only led to a substantial influx of T cells into the tumor
   microenvironment, but also significantly increased the proportion of
   intratumoral CD8^+ T cells expressing PD-1. This enrichment of PD-1^+
   CD8^+ T cells within the tumor, potentially rendered more susceptible
   to PD-1 checkpoint inhibition, prompted us to investigate whether
   combining Mogat1 inhibition with PD-1 blockade could synergistically
   enhance anti-tumor immunity.

   To investigate the potential synergy between Mogat1 inhibition and PD-1
   blockade, we constructed breast cancer and melanoma tumor models by
   injecting MMTV-PyMT or B16/F10 tumor cells, respectively. Mice were
   then treated with PD-1 antibodies according to a specified timeline
   (Fig. [232]8a, b). The tumor growth curves for breast cancer
   (Fig. [233]8c) demonstrated that mice injected with Mogat1 knockdown
   tumor cells and treated with PD-1 antibodies exhibited the slowest
   tumor growth. This pattern was consistently observed in the melanoma
   tumor model (Fig. [234]8d). Importantly, the combination of Mogat1
   knockdown and PD-1 immunotherapy was more effective at suppressing
   tumor growth than either treatment alone, as confirmed by the survival
   curves of the mice (Fig. [235]8e, f). These findings collectively
   suggest that Mogat1 inhibition not only reshapes the TME to favor
   anti-tumor immunity but also enhances the efficacy of immune checkpoint
   blockade. Targeting Mogat1 thus holds promise as a therapeutic strategy
   for cancer treatment, particularly in combination with immunotherapies
   (Fig. [236]8g).

Fig. 8. Mogat1 inhibition enhances Anti-PD-1 therapy response in breast
cancer and melanoma.

   [237]Fig. 8
   [238]Open in a new tab

   a, b Schematic representation of the experimental design for evaluating
   the effects of Mogat1 knockdown and anti-PD-1 therapy in mouse models.
   a MMTV-PyMT tumor cells (shCTRL or shMogat1) were implanted into FVB
   mice. b B16/F10 melanoma cells (shCTRL or shMogat1) were implanted into
   C57BL/6 mice. Mice received intraperitoneal injections of either IgG
   control antibody or anti-PD-1 antibody at the indicated time points.
   Created in BioRender. [239]https://BioRender.com/9gwjt01. Tumor growth
   curves of c FVB mice in MMTV-PyMT breast cancer model and d C57BL/6
   mice in B16/F10 melanoma model (n = 7 per group). Arrows indicate
   anti-PD-1 injection days. Kaplan-Meier survival curves of (e) FVB mice
   in MMTV-PyMT breast cancer model and (f) C57BL/6 mice in B16/F10
   melanoma model (n = 7 per group). Arrows indicate anti-PD-1 injection
   days. g Graphical summary of Mogat1 drives metabolic adaptations to
   evade immune surveillance. Created in BioRender.
   [240]https://BioRender.com/fhksjom. Data are representative of two
   independent experiments. Error bars in (c, d) represent standard error
   of the mean (SEM). Statistical significance for tumor growth curves (c,
   d) was determined using two-way ANOVA. Survival curves (e, f) were
   analyzed using the log-rank test. *P < 0.05, **P < 0.01, ***P < 0.001,
   ****P < 0.0001. ns, not significant. Figure (a, b, g) are created by
   BioRender. Source data are provided as a [241]Source Data file.

Discussion

   Tumors orchestrate complex strategies to evade immune surveillance,
   with metabolic reprogramming emerging as a central mechanism that both
   supports malignant growth and shapes the immune
   landscape^[242]38–[243]41. This metabolic adaptation is now recognized
   as a key driver of immune resistance, allowing tumors to manipulate the
   immune microenvironment to their advantage^[244]42. While genetic
   perturbation screens like CRISPR-Cas9 have transformed our
   understanding of gene function^[245]4–[246]7, these approaches may not
   fully capture the dynamic processes occurring during tumor progression
   within a complex TME^[247]43. Complementary strategies are needed.
   Here, we utilized an in vivo selection strategy coupled with
   transcriptomic profiling, monitoring natural gene expression changes as
   tumors evolve under immune pressure. This unbiased approach identified
   the metabolic enzyme Mogat1 as a critical mediator of immune evasion,
   demonstrating the power of such methods to uncover clinically relevant
   immune escape mechanisms operating across different stages of cancer
   progression^[248]44.

   While Mogat1’s role in lipid metabolism is established^[249]45,[250]46,
   we reveal a function in tumor immune evasion, driven by its critical
   role in metabolic reprogramming during tumor progression. This
   metabolic reprogramming, a hallmark of cancer, is increasingly
   recognized as a key factor in immune evasion^[251]47. Our multi-stage
   transcriptomic analysis identified Mogat1 as central to this
   adaptation, with its persistent upregulation and lipid pathway
   enrichment suggesting tumors co-opt lipid synthesis for energy and
   stress resilience, consistent with the broader understanding of cancer
   cell metabolic flexibility^[252]48. We show that increased Mogat1
   expression during breast cancer progression enhances lipid storage,
   promoting immune escape by depleting resources and inhibiting T cell
   infiltration, a mechanism akin to metabolic checkpoints that suppress
   anti-tumor immunity^[253]49. Critically, the anti-tumor effects of
   Mogat1 inhibition depend on an intact adaptive immune system, revealing
   a link between lipid metabolism and immune evasion, a connection also
   observed in other metabolic enzymes and immune pathways^[254]50. This
   discovery provides a clear mechanism for Mogat1’s role in tumor
   progression and opens immunotherapeutic avenues, where metabolic
   targets are increasingly being investigated^[255]51. Indeed, Mogat1
   inhibition suppresses tumor growth while simultaneously enhancing
   anti-tumor immunity, increasing T cell infiltration and augmenting
   their cytotoxic function, evidenced by heightened IFN-γ and Granzyme B
   secretion and increased tumor cell lysis. Furthermore, our findings
   demonstrating that secreted factors from Mogat1-depleted tumor cells
   enhance T cell migration and function highlight a functional
   interconnection between the intrinsic metabolic state and extrinsic
   immune modulation, mediated at least in part by soluble factors. While
   the precise secreted factors or cell-surface interactions downstream of
   Mogat1 that mediate these immune effects require further detailed
   investigation, representing an important avenue for future studies, our
   current data robustly establish the critical role of Mogat1 in
   regulating the tumor-immune interface.

   Furthermore, our findings reveal a promising therapeutic synergy
   between Mogat1 inhibition and PD-1 blockade. This occurs despite a
   complex increase in both PD-1^+ and PD-1- CD8^+ T cells upon Mogat1
   depletion. Crucially, the elevated PD-1^+ T cell fraction displayed
   heightened effector function and reduced co-expression of canonical
   exhaustion markers (TOX, CTLA-4, LAG3). This indicates a functionally
   potent, activated state amenable to PD-1 blockade^[256]52,[257]53,
   rather than terminal exhaustion. Therefore, the observed synergy likely
   stems from PD-1 blockade acting upon an expanded pool of these highly
   activated T cells fostered within the improved immune microenvironment
   created by Mogat1 deficiency. These results strongly support Mogat1
   inhibition as a potential strategy to enhance or restore sensitivity to
   ICB therapies.

Methods

Experimental model and subject details

Cell culture

   MDA-MB231, MMTV-PyMT, 4T1, B16/F10, MMTV-PyMT-OVA, B16/F10-OVA, and
   293 T cells were devoid of mycoplasma and cultured in Dulbecco’s
   Modified Eagle Medium (Gibco, Cat#C11995500BT), which was supplemented
   with 1% penicillin-streptomycin (Gibco, Cat#10378016) and 10% fetal
   bovine serum (Gibco, Cat#10500064). All cell lines were cultured in a
   humidified 5% CO[2] incubator at 37 °C and passaged at 80% confluence
   for a maximum of 20 passages.

Mice strains

   For this research, C57BL/6 mice were supplied by the Shanghai
   Laboratory Animal Center, Chinese Academy of Sciences (Shanghai,
   China). Various female mouse strains, including FVB/N, BALB/c, NOG
   (NOD.Cg-Prkdc-scid Il2rgtm1sug/JicTac), and NOG-MHCI/II-2 KO
   (NOD.Cg-B2Mem1Tac-Prkdc-scid H2-Ab1tm1Doi Il2rgtm1sug/JicTac) aged 6–8
   weeks, were procured from Beijing Vital River Laboratory Animal
   Technology Company (Beijing, China). The breast cancer model, female
   MMTV-PyMT transgenic mice (FVB/N-Tg(MMTV-PyVT)634Mul/J, JAX stock
   #002374), were acquired from The Jackson Laboratory (Bar Harbor, Maine,
   USA). All animals were housed in a specific pathogen-free (SPF)
   environment maintained at approximately 22 °C, provided with ad libitum
   access to food and water, and subjected to a 12-h light/dark cycle.
   Experimental procedures involving mice were performed under the ethical
   guidelines and regulations of China, with all protocols specifically
   approved by the Institutional Animal Care and Use Committee (IACUC) of
   Westlake University (permission numbers: 19-022-XZJ).

Human samples collection

   Paired tumor and adjacent normal breast tissue samples were collected
   from 30 female patients undergoing surgery for breast cancer at The
   Affiliated Hangzhou First People’s Hospital. The study protocol
   (20240628XZJ001) was approved by the Ethics Committee of The Affiliated
   Hangzhou First People’s Hospital, and all procedures were performed in
   accordance with relevant guidelines and regulations. All clinical
   samples were collected and analyzed with the patients’ informed
   consent, in accordance with the approved protocol: 20240628XZJ001. In
   addition, all procedures involving the handling and use of human
   genetic resources complied with the regulations of the China Ministry
   of Science and Technology, and all necessary approvals were obtained.
   The median age of the patient cohort was 53.5 years (Interquartile
   Range: 47–61 years). To ensure participant de-identification,
   individual patient ages are presented as ranges in Supplementary
   Data [258]3. Sex/gender of participants was determined based on
   clinical records (assigned sex at birth). As this study focused on
   breast cancer in a cohort comprising exclusively female participants,
   sex/gender was not a variable in the analysis.

Method details

shRNA-mediated gene knockdown

   Short-hairpin RNA (shRNA) sequences targeting Mogat1 were designed
   (Supplementary Data [259]1) and cloned into pLL3.7-GFP/PURO and
   inducible pTRIPZ-TurboRFP lentiviral vectors to construct shRNA
   plasmids. Lentivirus was produced by co-transfecting HEK293T cells with
   two lentiviral helper plasmids (pVSV-G and pMD8.9). The lentivirus was
   collected from the 48-h supernatant by passing it through a 0.45 µm
   filter. The collected lentivirus was either used directly to infect
   tumor cells with the addition of 8 µg/mL polybrene (Sigma-Aldrich,
   Cat#H9268) or frozen at −80 °C for later use. To eliminate negative
   interference, tumor cells infected with shRNA-pLL3.7-GFP/PURO
   lentivirus were sorted based on GFP fluorescence using a cytometry
   sorter (Sony MA900) or selected using puromycin. Meanwhile, tumor cells
   infected with shRNA-pTRIPZ-TurboRFP lentivirus were induced for RFP
   expression with 1 µg/mL doxycycline for 1 week and purified using a
   cytometry sorter (Sony MA900).

Cell colony formation assay

   Tumor cells were cultured until they reached 70% confluence under
   optimal growth conditions. The cells were then digested into
   single-cell suspensions and resuspended in fresh medium. After
   counting, the cells were appropriately diluted and seeded into 6-well
   or 12-well plates at densities of 1000 or 500 cells per well,
   respectively. The cells were cultured in a humidified incubator with 5%
   CO[2] at 37 °C for 7 days, with fresh medium supplemented on day 3
   without removing the old medium. Following the incubation period, cell
   colonies were stained with a crystal violet solution (0.5% w/v crystal
   violet powder, 80% v/v H2O, and 20% v/v methanol; MACKLIN,
   Cat#548-62-9) and imaged using a Zeiss AxioZoom V16 microscope (Zeiss,
   Oberkochen, Germany). Colony images were analyzed using ImageJ
   software.

Mouse tumor models

   For the MMTV-PyMT and 4T1 orthotopic breast tumor models, 5 × 10^5
   reconstituted tumor cells were mixed with 50% Matrigel (Corning,
   Cat#354234) and injected into the left fat pad of FVB and BALB/c mice,
   respectively. For the B16/F10 melanoma model, 2 × 10^5 B16/F10 cells
   were mixed with 50% Matrigel and subcutaneously injected into the right
   flank on day 0. For the MDA-MB231 orthotopic breast tumor model,
   1 × 10^6 reconstituted tumor cells were mixed with 50% Matrigel and
   injected into the left fat pad of NOG mice. For human PBMC treatment,
   1 × 10^7 human PBMC cells were injected via the tail vein into
   MHCI/II-DKO-NOG mice on day 0 and day 16. On day 1, 1 × 10^6 MDA-MB231
   tumor cells were injected into the left fat pad of the mice. For
   anti-PD1 antibody treatments, mice were injected intraperitoneally with
   200 µg of anti-PD1 antibody (BioXCell, clone RMP1-14, Cat#BE0146) on
   indicated days after tumor injection, with Rat IgG2a isotype antibody
   (BioXCell, clone 2A3, Cat#BE0089) used as a control. For T cell
   depletion experiment, FVB mice were injected intraperitoneally with
   200 µg of anti-CD4 antibody (BioXCell, clone RMP1-14, Cat# BE0003) and
   on anti-CD8a antibody (BioXCell, clone RMP1-14, Cat# BE0061) indicated
   days after tumor injection, with Rat IgG2a isotype antibody (BioXCell,
   clone 2A3, Cat#BE0089) used as a control. Tumor size was measured every
   3–4 days post-challenge. Measurements were taken manually by assessing
   the longest dimension (length) and the longest perpendicular dimension
   (width). Tumor volume was calculated using the formula: V = L × W^2/2.
   Mice were euthanized via CO[2] inhalation when tumors reached 2000 mm³
   or upon ulceration/bleeding. All procedures complied with the local
   institutional animal ethics board (permission number: 19-022-XZJ). The
   maximal tumor diameter permitted by the ethics committee was no more
   than 2 cm for single tumor and the maximal tumor size did not exceed
   the limit. Humane euthanasia was performed upon reaching predefined
   endpoints: tumor diameter exceeding 2 cm, body weight loss >20%
   relative to baseline, ulceration severity >Grade 3, or observable
   mobility deficits.

Mouse tumor dissection and dissociation

   At the conclusion of the animal experiments, tumor tissue was dissected
   and finely chopped into 1 mm3 pieces using a razor blade. Concurrently,
   a tissue digestion solution was prepared by supplementing DMEM/F12
   medium (Gibco, Cat#C11330500BT) with 2 mg/mL collagenase type IV
   (Gibco, Cat#9001-12-1), 1 mg/mL insulin (Gibco, Cat#12-585-014),
   20 U/mL DNase I (Sigma-Aldrich, Cat#9003-98-9), and 5% fetal bovine
   serum. The tumor tissue pieces were resuspended in the digestion
   solution and gently agitated for 45 min at 37 °C. Following digestion,
   the tumor tissue was collected via centrifugation, and residual red
   blood cells were removed using a blood cell lysis buffer (Beyotime
   Biotechnology, Cat#C3702). The digested tissue was then washed with
   phosphate-buffered saline (PBS) containing 2% fetal bovine serum and
   filtered through a 70 µm mesh to eliminate tissue clumps. Finally,
   single cells from the dissociated tumor tissue were counted and stored
   in an appropriate medium for subsequent experiments.

Staining and flow cytometry

   Single-cell suspensions derived from tumors were stained with the
   eFluor™ 780 Fixable Viability Dye (Invitrogen, Cat#65-0865-14) at 4 °C
   for 20 min, followed by blocking with TruStain FcX™ PLUS (anti-mouse
   CD16/32 antibody, BioLegend, Cat#156604) at 4 °C for an additional
   20 min. Subsequently, the cells were stained with a panel of cell
   surface marker antibodies, including: PE-Cy7 anti-mouse CD45 (1:200,
   BioLegend, Cat#S18009F), eFluor™ 450 anti-mouse CD3 (1:200, Invitrogen,
   Cat#48-0032-82), PerCP/Cy5.5 anti-mouse CD8a (1:200, BD Biosciences,
   Cat#551162), PE anti-mouse CD8a (1:200, BioLegend, Cat#100708), BV510
   anti-mouse CD4 (1:200, BioLegend, Cat#100559), BUV395 anti-mouse CD19
   (1:200, BD Biosciences, Cat#563557), BV605 anti-mouse CD11b (1:200,
   BioLegend, Cat#101257), AF700 anti-mouse Ly6G (1:200, BioLegend,
   Cat#127622), BV421 anti-mouse F4/80 (1:200, BioLegend, Cat#123132),
   BV785 anti-mouse CD279 (PD-1) (1:200, BioLegend, Cat#135225), APC
   anti-mouse NK1.1 (1:200, BD Biosciences, Cat#550627), PE anti-human
   CD45 (1:200, BioLegend, Cat#982322), FITC anti-human CD3 (1:200,
   BioLegend, Cat#300452), APC anti-human CD8 (1:200, BioLegend,
   Cat#300911), BV510 anti-human CD4 (1:200, BioLegend, Cat#300545). After
   surface marker staining, the intracellular proteins were stained, like
   BUV615 anti-mouse IFNγ (1:100, BD Biosciences, Cat#366-7311-82) and
   PE-eflour610 anti-mouse Granzyme B (1:100, BD Biosciences,
   Cat#61-8898-80). Concurrently, cells were stained for 10 min at 37 °C
   with BODIPY™ 493/503 and BODIPY™ FL C16. After three rounds of washing,
   the cells were analyzed using the Cytek™ Aurora (Cytek Biosciences,
   Fremont, CA) and CytoFLEX™ (Beckman Coulter, Fullerton, CA, USA)
   instruments. Data were analyzed using FlowJo™ software (v10.4.2, FlowJo
   LLC). Cell counts were determined by normalizing the number of cells to
   the recorded bead count, divided by the volume of the tumor aliquot and
   the mass of the tumor. The analysis gate was established based on
   isotype controls.

Immunofluorescence assays

   For frozen tumor section, tissues were fixed in 4% paraformaldehyde
   (PFA) (BBI Life Sciences, Cat#E672002-0500) overnight at 4 °C and
   subsequently immersed in 30% sucrose overnight for infiltration. The
   fixed tissues were then embedded in O.C.T. compound (Tissue Tek®, Miles
   Scientific, Illinois) and frozen at −80 °C prior to sectioning. For
   cells grown on microscope slide, fixation was performed using 4% PFA at
   37 °C for 15 min. After washing with phosphate-buffered saline (PBS),
   the samples were permeabilized with PBS containing 0.1% Triton X-100
   (Sigma-Aldrich, Cat#V90050) for 5 min. To prevent nonspecific binding,
   the samples were treated with a 5% donkey serum blocking solution at
   room temperature for 1 h. Subsequently, the samples were incubated with
   primary antibodies overnight at 4 °C, including: Rat Anti-mouse CD8A
   (1:100, Novusbio, Cat#NBP-49045SS), Rat anti-mouse CD45(1:200,
   Invitrogen, Cat#14-0451-85), Goat anti-Calnexin (1:300, Abcam,
   Cat#ab219644), Rabbit Anti-Calreticulin (1:200, CST, Cat#12238)
   overnight at 4 °C. Then samples were stained with
   fluorophore-conjugated secondary antibodies: Donkey Anti-Rat IgG H&L
   conjugated 647 (Jackson Immunoreaserch, Cat#712-605-153), Donkey
   Anti-Rat IgG H&L conjugated 594 (Jackson Immunoreaserch,
   Cat#712-585-153), Donkey-anti-Rabbit AF568 (Abcam, Cat#ab175693,),
   Donkey-anti-Rabbit AF647 (Thermo-Invitrogen, Cat#A142730),
   Donkey-anti-Goat AF647 (Thermo-Invitrogen, Cat#A21447) for 1 h at room
   temperature. Finally, slides were stained with 1ug/mL^1x Hoechst
   (ThermoFisher, Cat#62249) for 10 min and mounted with Antifade Mounting
   Medium (Beyotime, Cat#P0126). Images were taken using a Olympus FV3000
   Confocal (Germany) fluorescence microscope and analyzed by Image J.

In vitro T cell coculture assay

   CD8^+ T cells were isolated from splenocytes of OVA-specific T cell
   receptor transgenic OT-1 mice using the EasySep™ Mouse CD8^+ T Cell
   Isolation Kit (StemCell Technologies, Cat#19853). The isolated CD8^+ T
   cells were resuspended in RPMI 1640 medium supplemented with 10% fetal
   bovine serum (FBS), 1% penicillin-streptomycin (P/S), 50 µM
   β-mercaptoethanol (Thermo Fisher Scientific, Cat#21985023), and
   10 ng/mL recombinant interleukin-2 (IL-2) (PeproTech, Cat#200-02-10).
   The T cells were then activated with CD3/CD28 antibodies (Bio X Cell,
   Cat#BE0001, BE00015) for 48 h and plated at a density of 2 × 10^5 cells
   per well in a 96-well plate. Simultaneously, 1 × 10^4 MMTV-PyMT-OVA or
   B16-OVA shCTRL and shMogat1 tumor cells were seeded into a 96-well
   plate and allowed to adhere for 12 h. Preactivated OT-1 CD8^+ T cells
   were added to the tumor cells at various effector-to-target (E: T)
   ratios and co-cultured for 24 h. After co-culture, the cells were
   collected, and the percentage of apoptosis in tumor cells was analyzed
   using propidium iodide (PI) staining and flow cytometry (Beckman
   CytoFLEX, Fullerton, CA, USA).

CD8^+ T cell transwell assay

   CD8^+ T cells were seeded into transwell inserts equipped with 5 µm
   transwell filters to perform a transwell migration assay. OT-1 CD8^+ T
   cells were diluted to a concentration of 1 × 10^6 cells/mL in T cell
   culture medium. A 200 µL aliquot of the T cell suspension was added to
   the upper chamber, while tumor cell supernatants derived from shCTRL or
   shMogat1 MMTV-PyMT-OVA cells cultured for 48 h were added to the lower
   chamber. The plates were incubated at 37 °C for 8 h. After incubation,
   the upper chamber was carefully removed, and cells in the lower chamber
   were collected, counted, and stained with propidium iodide (PI) for
   analysis by flow cytometry.

Determination of DAG and TAG content

   Diacylglycerol (DAG) and triacylglycerol (TAG) concentrations were
   measured using a DAG ELISA kit (Beijing Bioroyee, Cat# RE4257) and a
   TAG assay kit (Beijing Solarbio, Cat# BC0625), respectively, following
   the manufacturer’s instructions. Briefly, tumor cells were resuspended
   in fresh lysis buffer for DAG extraction or in triglyceride assay
   buffer for TAG extraction, followed by ultrasonication until the
   solution became clear. DAG and TAG concentrations were determined from
   the supernatant of each sample after generating a standard curve using
   a SpectraMax ABS PLUS microplate reader (Molecular Devices, Shanghai,
   China). The concentrations were normalized to either the total protein
   concentration or the tumor cell number.

Co-culture of CD8^+ T cells with tumor cell-conditioned media

   Activated OT-1 CD8^+ T cells were prepared by stimulating with CD3/CD28
   antibodies in complete T cell medium for 48 h. Subsequently, 1 × 10^6
   activated T cells were seeded per well in 500 μL medium within a
   12-well plate. Tumor cell-conditioned media was generated by collecting
   supernatants from 48 h cultures of shCTRL or shMogat1 MMTV-PyMT-OVA
   cells. Equal volumes (500 μL) of tumor-conditioned media were added to
   T cell cultures and co-incubated at 37 °C for 12 h. To facilitate
   intracellular cytokine detection, GolgiPlug™ Protein Transport
   Inhibitor (BD Biosciences, 559029) was added at 1 μL/mL 4–6 h prior to
   cell harvesting. Cytokine production (IFN-γ and Granzyme B) was
   subsequently analyzed by flow cytometry using standard intracellular
   staining protocols.

RNA extraction and RT-qPCR

   All reagents, buffers, and containers used for RNA work were either
   RNase-free grade or treated with 0.1% v/v DEPC (Sigma-Aldrich, Cat#
   D5758) to eliminate RNase contamination in the experiments described in
   this section and other relevant sections. For total RNA extraction,
   cultured cells were washed and lysed using TRIzol reagent (Life
   Technologies, Cat# 15596018), following the manufacturer’s
   instructions. Subsequently, 1 µg of RNA was reverse-transcribed into
   cDNA using the HiScript II Q RT SuperMix Kit (Vazyme, Cat# R223-01)
   according to the manufacturer’s protocol. HiEff qPCR SYBR Green Master
   Mix (YEASEN, Cat# 11201ES) and specific primers (Supplementary
   Data [260]1) were used for real-time quantitative PCR (RT-qPCR), which
   was performed on a Real-time Thermal Cycler (Analytik Jena, Jena,
   Germany). Finally, the RT-qPCR data were normalized to GAPDH expression
   levels.

RNAseq analysis

   For different cell types, cells were purified using a flow cytometry
   sorter (Sony MA900) based on surface markers and fluorescence, followed
   by total cellular RNA extraction. RNA quantification was performed
   using Qubit (Invitrogen, Carlsbad, USA), and RNA integrity was assessed
   using an Agilent Bioanalyzer (Agilent, Santa Clara, USA). Subsequently,
   RNA sequencing (RNA-seq) was conducted on the Illumina NovaSeq 6000
   platform at Shanghai Genenergy Biomedical Technology Co., Ltd.
   (Shanghai, China). Raw RNA-seq data were processed using FastQC for
   quality control, Trim Galore for trimming, Hisat2 for mapping, and
   Subread for counting. Differentially expressed genes were identified
   using the limma software, with thresholds of log2(FC) > 1 and P < 0.05.
   Additionally, datasets of tumor cells and CD8^+ T cells sorted from
   tumor tissue have been uploaded to the Gene Expression Omnibus (GEO,
   accession number [261]GSE274902).

Non-targeted mass spectrometry-based metabolomics

Sample preparation

   Approximately 5 × 10^6 MMTV-PyMT or 4T1 tumor cells per biological
   replicate (n = 3 biological replicates per group: shCTRL vs. shMogat1)
   were collected. Cells were washed twice with ice-cold PBS. Metabolite
   extraction was performed by adding 1 ml of pre-chilled extraction
   solution (acetonitrile: methanol: water, 2:2:1, v/v/v) containing a
   mixture of isotopically labeled internal standards (a commercial
   internal standard mix was used for quality control, Shanghai Baiqu
   Biomedical Technology Co., Ltd., Shanghai, China). Samples were
   subjected to three rapid freeze-thaw cycles using liquid nitrogen and a
   37 °C water bath, with vortexing for 30 s after each thaw. Protein
   precipitate and cell debris were removed by centrifugation at
   13,000 × g for 15 min at 4 °C after a 10-min ice water bath. The
   resulting supernatant was transferred to a new tube, directly used for
   LC-MS analysis. Pooled quality control (QC) samples were prepared by
   mixing equal aliquots from each experimental sample and were injected
   periodically throughout the analytical run.

LC-MS analysis

   Metabolomic profiling was performed using a Thermo Vanquish UHPLC
   system coupled to a Thermo Q Exactive HF hybrid quadrupole-Orbitrap
   mass spectrometer (Service Provider: Shanghai Baiqu Biomedical
   Technology Co., Ltd., Shanghai, China). Chromatographic separation was
   achieved on a Waters ACQUITY UPLC BEH Amide column (2.1 mm × 50 mm,
   1.7 μm) maintained at 40 °C. The mobile phase consisted of A: [water
   with 25 mmol/L ammonia and 25 mM ammonium acetate] and B: [acetonitrile
   with 0.1% formic acid]. The LC gradient was as follows: [0-0.25 min,
   95% B; 0.25-3.5 min, 95–65% B; 3.5–4 min, 65–40% B; 4–4.5 min, 40% B;
   4.5–4.55 min, 40–95% B; 4.55–6 min, 95%B]. The flow rate was
   0.5 mL/min, and the injection volume was 2 µL. Mass spectrometry was
   performed in both positive and negative ion modes using a Heated
   Electrospray Ionization (HESI) source. Key MS parameters were: spray
   voltage, 3.8 kV (positive), −3.4 kV (negative); capillary temperature,
   320 °C; sheath gas flow rate, 50 Arb; auxiliary gas flow rate, 15 Arb.
   Full scan MS1 data were acquired over a mass range of m/z (70–1200) at
   a resolution of 60,000. The data were acquired in full scan and
   data-dependent MS/MS (ddMS2) mode.

Data processing and metabolite identification

   Raw data files (.raw) were converted to (.mzXML) format using
   ProteoWizard msConvert (version 3.0.20287). Peak picking, retention
   time alignment, and feature detection were performed using XCMS package
   in R (V3.6.0). Metabolite annotation was performed by matching
   precursor m/z values (mass tolerance ± 25 ppm), retention times
   (tolerance ± 0.5 min), and MS/MS fragmentation patterns against an
   internal MS2 database (BiotreeDB, V3.0) and public database (HMDB,
   mona, NIST) with an MS/MS spectral match score threshold of 0.7.

Statistical analysis and pathway enrichment

   The result was normalized to internal standards. Multivariate
   statistical analysis, including supervised Orthogonal Partial Least
   Squares Discriminant Analysis (OPLS-DA), was performed using SIMCA-P
   (16.0.2, Umetrics) to identify differentially abundant metabolites
   between shCTRL and shMogat1 groups. Metabolites with a Variable
   Importance in Projection (VIP) score >1 from the OPLS-DA model and a
   statistically significant difference (two-tailed Student’s t-test,
   P < 0.05, with FDR correction if applied) were considered differential.
   Pathway enrichment analysis of these differential metabolites was
   performed using KEGG pathway mapping tools and R (ggplot2, V3.3.5).

   The datasets have been uploaded to the China National GeneBank Database
   (CNGBdb, #CNP0006104,
   [262]https://db.cngb.org/search/project/CNP0006104/).

Survival curve analysis using TIMER2.0

   We utilized TIMER2.0, available at
   [263]http://timer.comp-genomics.org/, to analyze patient survival
   outcomes, applying a split expression threshold of 20%. Using the
   “Gene_Outcome” function within the “Cancer Exploration” module of the
   TIMER2.0 platform, we examined the impact of the gene MOGAT1 on the
   survival curves of various tumors.

Statistics and reproducibility

   All quantitative data from controlled laboratory experiments are
   presented as mean ± standard error of the mean (SEM) or mean ± standard
   deviation (SD), as specifically indicated in the respective figure
   legends. The number of biological replicates (e.g., individual mice or
   independent cell culture preparations) and the number of independent
   experimental repetitions are detailed in the corresponding figure
   legends for each experiment. Key experiments were independently
   repeated at least two to three times to ensure the reproducibility of
   the findings, unless otherwise specified. Sample sizes for animal
   experiments (typically n = 5–8 mice per group, as indicated in figure
   legends) were chosen based on prior experience with these in vivo
   models and common standards in the field, aiming to provide sufficient
   power to detect biologically meaningful differences while adhering to
   ethical animal use principles. Formal statistical power calculations
   were not performed prior to conducting the experiments. No data were
   excluded from the analyses. For animal studies, mice were assigned to
   experimental groups to balance for initial body age and tumor volume at
   the start of treatment where applicable Investigators performing tumor
   measurements and downstream sample processing were blinded to group
   allocation where practicable.

   Statistical analyses were performed using GraphPad Prism (Version 8.0;
   GraphPad Software, La Jolla, CA, USA), Comparisons between two groups
   were assessed using a two-tailed unpaired Student’s t-test, or a paired
   Student’s t-test for paired samples (e.g., human tumor vs. adjacent
   normal tissue, as specified in figure legends). Comparisons among three
   or more groups were performed using one-way or two-way analysis of
   variance (ANOVA) followed by an appropriate post-hoc test for multiple
   comparisons (e.g., Tukey’s, Dunnett’s, or Sidak’s test, as specified in
   figure legends). Tumor growth curves were analyzed using two-way ANOVA
   with repeated measures. Survival data were analyzed using the log-rank
   (Mantel-Cox) test. Correlations were assessed using Pearson’s or
   Spearman’s correlation coefficient where appropriate. Statistical
   significance was defined as P < 0.05. Specific P values are indicated
   in the figures or legends, with significance levels denoted as follows:
   *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not
   significant. Exact P-values are provided for key findings where
   appropriate.

   All experiments were repeated at least two to three times, unless
   otherwise specified. Data are presented as mean ± Standard Error of the
   Mean (SEM) or Standard Deviation (SD). Graphs and additional
   statistical analyses were generated using GraphPad Prism Version 8.
   Comparisons between two groups were assessed using either unpaired or
   paired Student’s t-tests. We employed F-test for comparing variances
   and two-tailed, unpaired Student’s t-test to determine the statistical
   discrepancy for two groups. Log-rank test was used for patient survival
   analysis. Two-way ANOVA followed by Tukey’s multiple comparison test
   was used for all tumor growth studies. The specific tests used, as well
   as the number of animals and experimental replicates, are detailed in
   the corresponding figure legends. Statistical significance was defined
   as follows: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Reporting summary

   Further information on research design is available in the [264]Nature
   Portfolio Reporting Summary linked to this article.

Supplementary information

   [265]Supplementary Information^ (14.1MB, pdf)
   [266]41467_2025_62134_MOESM2_ESM.pdf^ (55.9KB, pdf)

   Description of Additional Supplementary Information
   [267]Supplementary Data 1^ (12.3KB, xlsx)
   [268]Supplementary Data 2^ (16.1KB, xlsx)
   [269]Supplementary Data 3^ (10.6KB, xlsx)
   [270]Reporting Summary^ (103.8KB, pdf)
   [271]Transparent Peer Review file^ (20.3MB, pdf)

Source data

   [272]Source Data^ (386.9KB, xlsx)

Acknowledgements