ZBTB46 coordinates angiogenesis and immunity to control tumor outcome

Abstract

Tumor angiogenesis and immunity show an inverse correlation in cancer progression and outcome¹. Here, we report that ZBTB46, a repressive transcription factor and a widely accepted marker for classical dendritic cells (DCs)^2,3, controls both tumor angiogenesis and immunity. Zbtb46 was downregulated in both DCs and endothelial cells by tumor-derived factors to facilitate robust tumor growth. Zbtb46 downregulation led to a hallmark pro-tumor microenvironment (TME), including dysfunctional vasculature and immunosuppressive conditions. Analysis of human cancer data revealed a similar association of low ZBTB46 expression with an immunosuppressive TME and a worse prognosis. In contrast, enforced Zbtb46 expression led to TME changes to restrict tumor growth. Mechanistically, Zbtb46-deficient endothelial cells were highly angiogenic, and Zbtb46-deficient bone marrow progenitors upregulated Cebpb and diverted the DC program to immunosuppressive myeloid lineage output, potentially explaining the myeloid lineage skewing phenomenon in cancer⁴. Conversely, enforced Zbtb46 expression normalized tumor vessels and, by suppressing Cebpb, skewed bone marrow precursors toward immunostimulatory myeloid lineage output, leading to an immune-hot TME. Remarkably, Zbtb46 mRNA treatment synergized with anti-PD1 immunotherapy to improve tumor management in preclinical models. These findings identify ZBTB46 as a critical factor for angiogenesis and for myeloid lineage skewing in cancer and suggest that maintaining its expression could have therapeutic benefits.

Cooperation between tumor cells and the TME is a defining characteristic of cancer⁵. Targeting this interaction has gained much attention^6,7, with strategies like immune checkpoint blockade (ICB) immunotherapy becoming prominent. ICB has shown remarkable efficacy in some patients by promoting long-term tumor control and complete regression⁸. Unfortunately, most patients still do not respond to ICB therapies⁹. One confounding factor that limits the effectiveness of ICB is the abnormal tumor vessels featuring impaired perfusion and excessive leakage, leading to a hypoxic and immunosuppressive microenvironment^6,10. Vascular normalization supports more antitumor immunity and improves response to ICB therapy. Identifying TME programs that can enhance the outcomes of ICB holds tremendous promise as adjuvants in immunotherapies.

ZBTB46 belongs to the BTB-ZF family of transcriptional repressors and is considered a classical DC marker. Although classical DCs, endothelial cells (ECs) and fibroblast/mesenchymal cells constitutively express Zbtb46, its deficiency does not lead to any discernible DC-related or EC-related abnormalities (Extended Data Fig. 1a–c)^2,3,11 In homeostasis, ZBTB46 keeps DCs and ECs in a quiescent state^12,13; however, its function has not been addressed in pathological conditions. We investigated the role of ZBTB46 in solid tumors using mouse cancer models. Zbtb46 expression was downregulated in the tumor-associated ECs and DCs in the wild-type (WT) mice bearing 1956 sarcoma, Lewis lung carcinoma (LLC) and PyMT-BO1 (orthotopic) and MMTV-PyMT (genetic) breast cancers (Fig. 1a and Extended Data Fig. 1d–h). Zbtb46 was also downregulated in the bone marrow (BM) of the tumor-bearing mice (Fig. 1a). Zbtb46^gfp/gfp (Zbtb46 knockout (KO) or ZKO) mice exhibited more robust tumor growth (Fig. 1b–e), showing pro-tumor TME features: enhanced angiogenesis, lower cytotoxic T lymphocyte (CTL)/regulatory T (T_reg) cell ratio, decreased conventional type 1 dendritic cells (cDC1; including XCR1⁺F4/80⁺ myeloid cells) and increased fibroblast activation (Fig. 1f,g and Extended Data Fig. 1i). Remarkably, individuals with cancer who had lower ZBTB46 expression had reduced antitumor immune components, and reduced ZBTB46 expression was associated with a worse prognosis in multiple cancer types (Fig. 1h and Extended Data Fig. 1j,k). Together, these data suggest an inhibitory role for ZBTB46 in tumor growth.

Fig. 1 |. ZBTB46 is a negative regulator of tumor progression.

a, Zbtb46 expression in tumor ECs (TECs; n = 4 for day D8 and n = 6 for D11, 15 and 19), tumor DCs (TDCs; n = 6), isolated at indicated days, and bone marrow (T.BM, n = 3) cells from 1956 tumor-bearing mice. Lung ECs (LECs; n = 8), splenic DCs (SDCs; n = 5) and BM (H.BM, n = 3) from healthy mice are shown as controls. b–e, Tumor growth in 1956 sarcoma (n = 20 for WT and 17 for ZKO) (b), PyMT-BO1 breast cancer (n = 12 for WT and 8 for ZKO) (c), LLC carcinoma (n = 10) (d) and MMTV-PyMT breast cancer (n = 10) (e) in WT and ZKO mice. f, Representative images with quantification of tumoral CD31⁺ vessel density in WT (n = 6 for 1956, 5 for PyMT-BO1, 10 for LLC and 11 for MMTV-PyMT) and ZKO (n = 12 for 1956, 10 for PyMT-BO1, 10 for LLC and 7 for MMTV-PyMT) mice. g, Measurements of CTL/T_reg, M1/M2, CD11c⁺MHCII⁺XCR1⁺cDC1-like cells (including XCR1⁺F4/80⁺ myeloid cells) and fibroblast activation protein (FAP⁺) in the TME of 1956 sarcoma and PyMTBO1 breast tumor-bearing mice at endpoints described in b and c (n = 4). h, Presence of CD8⁺ T cells and classical DCs in high-versus-low ZBTB46-expressing tumors in individuals from The Cancer Genome Atlas (TCGA) database analyzed with XCell. i, Tumor growth of 1956 sarcoma and PyMT-BO1 breast cancer in stromal ZKO and hematopoietic ZKO mice (n = 5). j, Tumor growth of 1956 sarcoma and PyMT-BO1 breast cancer in WT (n = 7 for 1956 and n = 3 for PyMT-BO1), VEC-cre ZKO (n = 7 for 1956 and n = 3 for PyMT-BO1), CD11c-cre ZKO (n = 4) and ZKO mice (n = 6). Data are the mean ± s.d. P values were determined using one-way analysis of variance (ANOVA) with Dunnett’s test for a (left and middle), two-tailed Student’s t-test for a (right)–g, i and j, and unpaired t-tests with Holm–Sidak multiple-comparison test for h. Scale bars, 100 μm (f). NS, not significant.

We examined the relative contribution of EC-specific and DC-specific Zbtb46 expression in the tumor growth by using constitutive endothelial (VEC-Cre), tamoxifen-inducible endothelial (iVEC-Cre) or hematopoietic (CD11c-Cre or VAV-Cre)-specific deletion of Zbtb46. We also generated BM chimeras for the same purpose (Extended Data Fig. 2a). In all the systems, ZBTB46 was required in both the endothelial and hematopoietic cells for optimal tumor control (Fig. 1i,j and Extended Data Fig. 2b–e). The most robust tumor growth was observed in the global ZKO mice, suggesting a collaborative role for the EC-specific and DC-specific Zbtb46 expression in suppressing tumor growth.

ZKO mice supported more angiogenesis in a Matrigel plug assay, supporting an EC-intrinsic role of ZBTB46 in regulating angiogenesis (Fig. 2a). Notably, enhanced tumor angiogenesis was mainly associated with the endothelial, not hematopoietic, deletion of Zbtb46 (Extended Data Fig. 2f,g). Zbtb46 knockdown (ZKD) mouse cardiac endothelial cells (MCECs) showed enhanced sprouting, tube formation and proliferation, while Zbtb46 overexpression led to reduced sprouting, tube formation and proliferation, similarly to a previous report¹³ (Fig. 2b,c and Extended Data Fig. 2h). Bulk RNA sequencing (RNA-seq) revealed that while the parental MCECs were enriched in migration and angiogenic pathways, Zbtb46-overexpressing MCECs were enriched in hypoxic-stress response and immune-supportive pathways (Fig. 2d).

Fig. 2 |. ZBTB46 normalizes tumor vasculature.

a, Representative images and quantification of CD31⁺ vessels in the Matrigel plug implanted in WT and ZKO mice (n = 5). b,c, Representative images and quantifications from the (b) fibrin gel sprouting (n = 8) and (c) Matrigel tube formation (n = 6) assays. d, Gene Ontology (GO) enrichment analysis of bulk RNA-seq data from WT and Zbtb46-overexpressed (OE) MCECs. e–g, Tumor growth of 1956 sarcoma (e), PyMT-BO1 breast cancer (f) and LLC carcinoma (g) in WT mice with empty vector (EV) or Zbtb46 (ZOE) lentiviral overexpression construct (intratumor) treatment. h, Representative images with quantification of tumoral CD31⁺ vessel density in WT mice with EV (n = 4 for 1956, 6 for LLC and 4 for PyMT-BO1) or Zbtb46 (WT + ZOE, n = 10 for 1956, 6 for LLC and 12 for PyMT-BO1) lentiviral overexpression construct (intratumor) treatment. i, Representative images and quantification for vascular leakage and intratumoral hypoxia as measured by the fluorescein isothiocyanate (FITC)–dextran 70-kD spread and the relative abundance of Hypoxyprobe-1 binding, respectively, in the 1956 sarcoma tumor tissue of WT mice with EV (n = 9 for vascular leakage and 10 for hypoxia) or Zbtb46 (WT + ZOE, n = 9 for vascular leakage and 7 for hypoxia) or of ZKO mice with EV (n = 6 for vascular leakage and 9 for hypoxia) lentiviral overexpression construct (intratumor) treatment. j, Schematic and analysis of relative migration of CD3⁺CD8⁺ CTL cells through WT (n = 8) and Zbtb46-overexpressed (ZOE, n = 7) MCEC barrier in a leukocyte trans-endothelial migration assay. Created with BioRender.com. k, Schematics and tumor growth of 1956 sarcoma cells co-transplanted with WT ECs (TC + WTECs; n = 6) or ZOE ECs (TC + ZOEECs; n = 8). Created with BioRender.com. Data are the mean ± s.d. P values were determined using a two-tailed Student’s t-test for a, e–h, j and k, one-way ANOVA with Dunnett’s test for b, c and i, and Fisher’s exact test with false discovery rate correction for d. Scale bars, 100 μm (a–c, h and i).

Intratumoral lentiviral delivery of Zbtb46 led to sustained expression in the tumor ECs, tumor DCs and tumor cells and reduced tumor mass (Fig. 2e–g and Extended Data Fig. 2i). Importantly, Zbtb46-overexpressing tumor lines had no growth difference compared with the parental lines, suggesting a TME-centric role for ZBTB46 in suppressing tumor growth (Extended Data Fig. 2j). While tumor vasculature was more dysfunctional in the ZKO mice, intratumoral lenti-Zbtb46 expression reduced angiogenesis (Fig. 2h) and led to vascular normalization with increased pericyte coverage, enhanced perfusion, alleviated intratumoral hypoxia and reduced leakage (Fig. 2i and Extended Data Fig. 2k,l). Normalized tumor vessels are known to support more antitumor immune and stromal TME with enhanced cytotoxic T cell infiltration and effectiveness^14–16, which could partly explain the decreased antitumor immune components observed in the ZKO mice and the opposite following the enforced Zbtb46 expression (Figs. 1g and 3c). In support of this, Zbtb46-overexpressing MCECs allowed more CD8⁺ T cell trans-endothelial migration (Fig. 2j). Remarkably, EC-specific Zbtb46 overexpression restricted the tumor progression in a tumor and EC co-transplantation system, supporting the EC-intrinsic role of ZBTB46 in tumor suppression (Fig. 2k and Extended Data Fig. 2m).

Fig. 3 |. ZBTB46 promotes antitumor immunity.

a, 1969 regressive sarcoma growth in WT, VEC-cre, VAV-cre and global ZKO mice. b, Analysis of GMP, pre-cDC, cDC, CD3 and Gr1⁺ cells in the BM, spleen and PB of the 1956 sarcoma-bearing mice (n = 3). c, Analysis of the tumor immune microenvironment of 1956 sarcoma-bearing (n = 4) and PyMT-BO1 tumor-bearing (n = 5) WT mice with EV or Zbtb46 (ZOE) lentiviral overexpression construct (intratumor) treatment. d, Gr1⁺ and MHCII⁺ cell generation from WT or ZKO mouse BM cells with EV-mCherry (WT and ZKO) or Zbtb46-mCherry (ZOE) lentiviral overexpression (n = 3). Created with BioRender.com. e, Gr1⁺ and MHCII⁺ cell generation from WT or ZKO mouse KSL cells sorted from BM (n = 4). f, Analysis of donor-derived TAMs and DCs after 5 days of intratumoral transfer of enriched monocytes from WT donor mouse BM (CD45.1) into 1956 sarcoma-bearing WT recipient mice (CD45.2) with EV-mCherry (WT + EV, n = 6) or Zbtb46-mCherry (WT + ZOE, n = 5) lentiviral overexpression. Created with BioRender.com. g, ChIP–qPCR analysis of ZBTB46 recruitment to a potential binding site in the Cebpb promoter in BM-derived precursors cells (n = 3). h, Analysis of Gr1⁺ and MHCII⁺ cell generation from BM cells of ZKO mice with EV-mCherry lentiviral overexpression alone (ZKO, n = 4) or with Cebpb short hairpin RNA (shRNA) constructs (ZKO + CKD, n = 3) and of WT mice with Zbtb46-mCherry lentiviral overexpression alone (ZOE, n = 4) or with Cebpb overexpression (ZOE + COE, n = 3). i, ChIP–qPCR analysis for CEBPB enrichment at CEBPB peaks in BM-derived precursors from WT or ZKO mice (n = 3). j, Tumor progression after 5 days of intratumoral transfer of enriched monocytes from healthy WT BM into 1956 sarcoma-bearing WT mice with EV or Zbtb46 (ZOE) lentiviral overexpression (n = 11). Data are the mean ± s.d. P values were determined using a two-tailed Student’s t-test for b, c and e–j and one-way ANOVA with Dunnett’s test for d.

Zbtb46 deficiency does not compromise classical DC function in a steady state^2,3,12. To determine if ZBTB46 was needed for DC in the context of the tumor, we challenged the global, EC-specific and hematopoietic-specific ZKO mice with 1969 sarcoma tumors, whose rejection in immune-competent mice depends on DC function¹⁷. Strikingly, all the ZKO mice rejected the 1969 tumor (Fig. 3a and Extended Data Fig. 3a), indicating no direct involvement of ZBTB46 in classical DC function. The hematopoietic system in ZKO mice was comparable to the WT mice during homeostasis in both the progenitor and committed cell types in lymphoid organs and peripheral blood (PB), except with a slight increase in the Gr1⁺ cells in PB (Extended Data Fig. 3b). However, upon tumor challenge, ZKO mice displayed more pronounced myeloid-biased pro-tumor immune characteristics as evidenced by the increased granulocyte–monocyte progenitors (GMPs) and reduced pre-cDCs in both BM and spleen, as well as decreased T cells and cDCs and increased Gr1⁺ and myeloid-derived suppressor cells in PB (Fig. 3b and Extended Data Fig. 3c). In contrast, intratumoral lenti-Zbtb46 expression led to a more antitumor immune microenvironment: increased CTL/T_reg cell ratio, enhanced cDC1 (including XCR1⁺F4/80⁺ myeloid cells) and natural killer (NK) cell population and reduced fibroblast activation (Fig. 3c).

To further understand the impact of Zbtb46 on the immune outcome, we assessed the lineage output of ZKO BM cells in culture². ZKO BM cells generated fewer major histocompatibility complex (MHC) class II-positive (MHCII⁺) DCs but increased Gr1⁺ myeloid and cells with a myeloid-derived suppressor cell phenotype than the WT BM cells (Fig. 3d and Extended Data Fig. 3d). Notably, ZKO KSLs (cKit⁺Sca1⁺Lin⁻; containing hematopoietic stem and progenitor cells, HSPCs), common myeloid progenitors (CMPs) and GMPs produced more Gr1⁺ cells and fewer MHCII⁺ DCs than the WT cells, suggesting that downregulation of Zbtb46 expression in HSPCs leads to myeloid lineage skewing (Fig. 3e and Extended Data Fig. 3e). Conversely, enforced Zbtb46 expression in either the total BM cells or the sorted progenitors reversed this trend to generate more DCs and fewer myeloid cells (Fig. 3d and Extended Data Fig. 3f), like a previous observation². Importantly, enforced Zbtb46 expression in BM-derived progenitors, mostly containing monocytes and lineage-negative cells, generated more DCs and fewer macrophages when transferred into an established tumor, a system that closely reflects the tumor-infiltrating monocyte differentiation¹⁸ (Fig. 3f).

We identified critical myeloid genes, including Cebpb, which encodes a crucial transcription factor for emergency granulopoiesis¹⁹ and monocyte/macrophage gene regulation²⁰, as potential direct targets of ZBTB46 from previously published chromatin immunoprecipitation followed by sequencing (ChIP–seq)¹² and microarray² data. ChIP coupled to quantitative PCR (ChIP–qPCR) confirmed that ZBTB46 binds the Cebpb promoter in BM-derived progenitors (Fig. 3g and Extended Data Fig. 3g). Cebpb expression was upregulated in the ZKO BM cells but was downregulated after Zbtb46 overexpression in WT BM cells (Extended Data Fig. 3h). Moreover, overexpression or knockdown of Cebpb partially reversed the overexpression or knockdown effect of Zbtb46, respectively, on myeloid lineage output from BM cells (Fig. 3h and Extended Data Fig. 3i). These observations and the fact that the consensus DNA recognition site for both ZBTB46 and CEBPB is highly similar^12,21 (Extended Data Fig. 3j) raise the possibility that CEBPB can upregulate myeloid genes, normally repressed by ZBTB46, in the absence of ZBTB46. Indeed, CEBPB binding to its transcriptional targets, such as Cebpb itself and Csf3r, was enhanced in ZKO BM cells (Fig. 3i and Extended Data Fig. 3k). A reporter system consisting of Cebpb transcriptional response element (TRE; containing tandem repeats of consensus CEBP-DNA recognition motifs) further demonstrated that ZBTB46 can inhibit the CEBPB-induced activation of TRE (Extended Data Fig. 3l,m). This interplay between ZBTB46 and CEBPB is functionally important as BM cells from tumor-challenged ZKO mice had higher macrophage-related gene expression and lower DC-related gene expression; Zbtb46 overexpression reversed this pattern (Extended Data Fig. 3n), suggesting that repression of Cebpb is at least partly responsible for the enforced Zbtb46-mediated immunostimulatory TME. Like the EC-specific overexpression outcome, intratumoral injection of Zbtb46-overexpressed BM-derived progenitors partially restricted tumor growth (Fig. 3j), demonstrating the collaboration of the cell-type-specific roles of ZBTB46 in tumor suppression.

To better understand the TME changes following lenti-Zbtb46 injection, we performed single-cell RNA-seq of the non-tumor components from the PyMT-BO1 tumors (Fig. 4a and Methods). Unsupervised analysis of 36,757 cells resulted in distinct clusters of monocyte/macrophages, neutrophils, fibroblasts, ECs, DCs, T and NK cells (Fig. 4a,b and Supplementary Fig. 1a,b). Within the lymphoid compartment, four CD8⁺ T clusters were identified: T_naive(CD8T_s1), T_effector(CD8T_ s2), T_exhaust(CD8T_s3) and proliferative-exhausted T_p-ex(CD8T_s4). Although frequencies of the exhausted clusters (s3 and s4) were similar, T_effector(CD8T_s2) was greatly increased and T_naive(CD8T_s1) was reduced by lenti-Zbtb46 expression (Fig. 4c, Extended Data Fig. 4a–c and Supplementary Fig. 2a,b). Intriguingly, the T_exhaust(CD8T_s3) cluster was associated with angiogenesis pathways, suggesting that the terminally exhausted CD8⁺ T cells could contribute to immunosuppression by promoting angiogenesis (Extended Data Fig. 4d). Lenti-Zbtb46 expression increased the frequency of the CD4⁺ T cluster while downregulating the genes vital for the immunosuppressive functions of the T_reg cells (for example, Cxcr3 and Lef1)^22,23 (Extended Data Fig. 4d–f). Lenti-Zbtb46 expression also increased the frequency of activated T cells comprising both CD4⁺ and CD8⁺ T cells (Extended Data Fig. 4d,e). Of the five NK clusters, lenti-Zbtb46 expression increased NK_s1 (activate) and decreased NK_s2 (inhibit) clusters (Fig. 4d and Extended Data Fig. 4a,b,g–i).

Fig. 4 |. ZBTB46 remodels the TME into an antitumor immunostimulatory milieu.

a, Schematics and tumor growth of PyMT-BO1 breast cancer in WT mice with either control (EV) or Zbtb46 (ZOE) lentiviral overexpression construct (intratumor) treatment (n = 5). Two-tailed Student’s t-test at endpoint. On day 16, the non-tumor live TME cells were sorted from the pooled tumor masses and sent for single-cell RNA-seq (Methods). Created with BioRender.com. b, t-distributed stochastic embedding (t-SNE) plot of intratumoral cells from the control and treatment group merged. c, Violin plots displaying expression levels of select genes in CD8T_s1–s4 clusters along with the percentages of cells in the clusters across different treatments. d, Violin plots displaying expression levels of select genes in NK_s1–s5 clusters along with the percentages of cells in the NK_s1 and NK_s2 clusters across different treatments. e, t-SNE plots highlighting Cx3cr1- and Nos2-expressing cells in the macrophage/monocyte population. f, Violin plots displaying expression levels of select genes in Mac_s1–s6 clusters along with the percentage of cells in the Mac_s1 cluster across different treatments. g, Violin plots displaying expression levels of select genes in DC_s1–s3 clusters. h, Percentages of cells in Strm_s1, Strm_s2 and Strm_s6 clusters of stromal cell populations in the TME across different treatments. i, Summary of TME remodeling after intratumoral Zbtb46 lentiviral treatment.

Among the myeloid cells, the macrophage/monocyte population constituted the largest cluster and exhibited remarkable remodeling following lenti-Zbtb46 administration, as evident by the changes in the expression patterns of Cx3cr1 and Nos2; macrophages/monocytes in the treatment was associated with inflammatory pathways (Fig. 4e and Extended Data Fig. 5a,b). Particularly, Mac_s1(M1-like) and Mac_s5 (IFNg-responsive-M1-like) were increased, while Mac_s3 (Arg⁺ activated tumor-associated macrophages (TAMs)) and Mac_s6 (proliferating M2-like) were reduced by lenti-Zbtb46 expression (Fig. 4f, Extended Data Fig. 5c–f and Supplementary Fig. 3a). Intriguingly, in the M2-like clusters (s2 and s4), expressions of pro-inflammatory markers were increased, and anti-inflammatory markers were reduced in the lenti-Zbtb46 treatment (Extended Data Fig. 5g). Next, among the DC clusters, DC_s2 (migratory DC-like) and DC_s3 (monocyte DC-like) remained similar, and DC_s4 (plasmacytoid dendritic cell (pDC)-like) was reduced by lenti-Zbtb46 expression. Unexpectedly, the DC_s1 (cDC1-like) frequency was decreased in the lenti-Zbtb46 treatment group (Extended Data Fig. 6a–c and Supplementary Fig. 4a). However, the Mac_s2 cluster, particularly in the lenti-Zbtb46-administered tumors, contained cells that were largely positive for Itgax, H2-Ab1 and Xcr1 (which were the fluorescence-activated cell sorting (FACS)-based markers for the cDC1), and was enriched in DC-associated pathways (Extended Data Figs. 5f and 6d), explaining the overall increase of the XCR1⁺ cDC1-like myeloid cells in the TME after lenti-Zbtb46 treatment. The expression of DC maturation and activation genes²⁴ were also increased in the DC_s1–s3 clusters by lenti-Zbtb46 expression (Fig. 4g and Extended Data Fig. 6e). Lastly, of the three neutrophil clusters, Neu_s1 and Neu_s3 (with tumorigenic activation and NETosis) were reduced and Neu_s2 (with antigen-presenting capabilities) was increased by lenti-Zbtb46 expression (Extended Data Fig. 7a–d). The N_s2/(N_s1 + N_s3) ratio was increased in the treatment; intratumoral neutrophils in the lenti-Zbtb46 treatment were enriched for antitumor immune pathways (Extended Data Fig. 7e,f). Intriguingly, we also observed a small mast cell cluster, which was reduced by lenti-Zbtb46 expression (Extended Data Fig. 7g,h).

In the stromal compartment (Extended Data Fig. 8a,b), the endothelial (Strm_s1) cluster was reduced by lenti-Zbtb46 expression, reflecting inhibition of angiogenesis by ZBTB46ZBTB46 (Fig. 4h). Strm_s1 in the treatment was enriched with extracellular matrix (ECM) modification, cell cycle checkpoints and antigen processing and presentation pathways and depleted for growth factor, NOTCH4, and interleukin (IL)-6-induced JAK–STAT3 signaling (Extended Data Fig. 8c). Among the five fibroblast clusters, Strm_s2 (low-angiogenic) was increased, while Strm_s6 (myofibroblastic cancer-associated fibroblast (CAF)) was reduced by lenti-Zbtb46 expression (Fig. 4h and Extended Data Fig. 8d,e). Importantly, lenti-Zbtb46 expression increased the expression of co-stimulatory molecules such as Cd80 and decreased the genes implicated in facilitating metastasis such as Olfml3 (ref. 25) across the clusters (Extended Data Fig. 8f).

Analysis of the hashtags (tagging either transduced cells that were identified as mCherry-reporter-positive or non-transduced cells identified as mCherry-reporter-negative) revealed that the lymphoid compartment (T cells and NK cells) mostly comprised reporter-negative non-transduced cells, whereas the stromal population (ECs and fibroblasts) were mostly reporter-positive transduced cells; the myeloid population (monocytes/macrophages, DCs and neutrophils) comprised both (Supplementary Fig. 1c,d). Collectively, lenti-Zbtb46 expression, both directly and indirectly, remodeled the TME to have more effector CD8⁺ T cells, M1-like pro-inflammatory macrophages, XCR1⁺ cDC1-like myeloid cells, mature and activated NK cells, and low-angiogenic fibroblasts while reducing M2-like anti-inflammatory macrophages, pro-tumor neutrophils and fibroblasts, and angiogenic ECs (Fig. 4i).

Since lenti-Zbtb46 expression led to an immune-hot TME, which is a prerequisite for an effective ICB therapy^1,8, we assessed whether Zbtb46 administration could synergize with anti-PD1 immunotherapy. Systemic administration of the p5RHH peptide-based Zbtb46 mRNA nanoparticle^16,26,27 was effective in sustaining Zbtb46 expression in tumor DCs and tumor ECs and resulted in the restriction of tumor growth (Extended Data Fig. 9a–f). Although tumor cells also acquired Zbtb46 expression (Extended Data Fig. 9b,e), tumor cells overexpressing Zbtb46 had no growth difference compared with the parental lines (Extended Data Fig. 2j), denoting TME remodeling by lenti-Zbtb46 as the driver for the tumor control. The tumor restriction was associated with an immunostimulatory TME (Extended Data Fig. 9c,f). Remarkably, Zbtb46 nanoparticles promoted anti-PD1 response in both anti-PD1-responsive (1956 sarcoma) and anti-PD1-refractory (PyMT-BO1) tumor¹⁶ models, generating long-term complete remission of tumor in many of the treated animals (Fig. 5a–d). An extended Zbtb46-nanoparticle treatment produced complete remission even in the anti-PD1-refractory PyMT-BO1 tumor-bearing mice (Fig. 5e). The addition of an anti-VEGFR2 (DC101) to Zbtb46 nanoparticles and anti-PD1 combination enhanced the response, particularly in the anti-PD1-refractory PyMT-BO1 model, hinting at the collaborative nature of ZBTB46 and vascular endothelial growth factor (VEGF) pathways in the tumor control (Fig. 5a–d). The tumor-eliminated mice spontaneously rejected a secondary challenge, indicating the development of an immunological memory (Fig. 5f). Intriguingly, co-transplanting tumors with Zbtb46-overexpressed ECs also improved the anti-PD1 treatment (Extended Data Fig. 9g), reinforcing the importance of the EC-specific Zbtb46 expression in tumor suppression. Post-treatment histological and FACS analysis of the TME, lymphoid organs, PB and a non-tumor organ revealed that the immunostimulatory and vascular normalization impacts of the Zbtb46 nanoparticles were primarily restricted to the TME (Extended Data Fig. 9h–j and Supplementary Fig. 5a,b).

Fig. 5 |. Therapeutic maintenance of ZBTB46 improves cancer immunotherapy outcome.

a, Schematics of WT tumor-bearing mouse treatment with Zbtb46 mRNA nanoparticles, anti-PD1 and DC101. Created with BioRender.com. b, Tumor growth of 1956 sarcoma and PyMT-BO1 breast cancer in WT mice with control (EGFP mRNA nanoparticles + IgG, n = 8 for 1956 and 6 for PyMT-BO1), Zbtb46 mRNA nanoparticles (ZmR, n = 4 for 1956 and 8 for PyMT-BO1), anti-PD1 (AP, n = 8), DC101 (DC, n = 8 for 1956 and 7 for PyMT-BO1), DC + AP (n = 8), ZmR + AP (n = 9 1956 and 10 for PyMT-BO1) and ZmR + AP + DC (n = 10) treatment at 12 days after treatment initiation. c,d, Tumor growth kinetics in 1956 sarcoma (c) and PyMT-BO1 breast cancer (d) in WT mice with control (EGFP mRNA nanoparticles + IgG), ZmR, anti-PD1 and DC101 treatment. e, Tumor growth kinetics in PyMT-BO1 breast cancer in WT mice with either an extended regimen of ZmR and anti-PD1 (AP) combination treatment (n = 15) or control (EGFP mRNA nanoparticles + IgG, n = 5). f, Schematic and tumor growth in the treatment-responded 1956 sarcoma tumor-eliminated mice (n = 13) challenged with secondary 1956 sarcoma transplantation. The tumor-naive mice (n = 5) were age matched to the tumor-eliminated mice. Tumor injections were administered contralateral to the original tumor site and were kept similar for both the tumor-naive and tumor-eliminated mice. Created with BioRender.com. Data are the mean ± s.d. P values were determined using one-way ANOVA with Dunnett’s test for b. CR, complete remission. i.p., intraperitoneal; i.v., intravenous.

Finally, we tested tumor factors to regulate Zbtb46 expression. Tumor-conditioned media downregulated Zbtb46 expression in vitro (Extended Data Fig. 10a). We tested a few likely candidates^18,28 and found that the reactive oxygen species, prostaglandin E2 (PGE2), retinoic acid (RA) and VEGF downregulated Zbtb46 expression in both ECs and DCs (Extended Data Fig. 10a,b); RA and PGE2 also skewed BM cells output toward more Gr1⁺ and less MHCII⁺ cells in vitro (Extended Data Fig. 10c). Pharmacological targeting of the PGE2, RA and reactive oxygen species pathways using US Food and Drug Administration-approved and commercially available inhibitors modestly prevented tumor-mediated Zbtb46 suppression both in tumor ECs and tumor DCs and in the total BM cells, while partially restricting tumor growth (Extended Data Fig. 10d–i), further implying the translatability of regulating Zbtb46 expression in the tumor.

In summary, our study uncovers a tumor-suppressive role of ZBTB46 in the vascular and hematopoietic system. Although thought to be restricted to cDCs, our study shows that ZBTB46 also controls EC functions in tumors. One striking finding was the potential suppressive role of ZBTB46 in tumor fibroblasts. Zbtb46 is normally, although at low levels, expressed in fibroblast/mesenchymal cells in homeostatic tissues. In ZKO mice, fibroblast activation protein-positive fibroblasts were increased, while Zbtb46 overexpression led to an expansion of low-angiogenic fibroblasts, warranting further studies of ZBTB46 in tumor-associated fibroblasts. With a recent study showing that ZBTB46 also controls the function of group 3 innate lymphoid cells in maintaining intestinal homeostasis²⁹, Zbtb46’s role seems to expand across many different cell types. Our data suggest that ZBTB46-mediated tumor vessel normalization leads to improved antitumor immunity, further supporting the inverse relationship between tumor angiogenesis and tumor immunity^30–34. Additionally, our study sheds light on the mechanisms underlying myeloid lineage skewing in cancer, implicating Zbtb46 downregulation and subsequent de-repression of Cebpb as contributing factors. This insight into the molecular pathways involved in myeloid cell generation in cancer may have implications for therapeutic strategies aimed at modulating myeloid cell populations within the TME. The synergistic effect of Zbtb46 mRNA nanoparticles with anti-PD1 treatment in controlling tumor growth, inducing long-term remissions and promoting immunological memory highlights the therapeutic potential of modulating Zbtb46 expression. Overall, our findings provide a proof of concept for further exploring Zbtb46 expression as an effective therapeutic target and underscores its importance in the context of cancer immunotherapy.

Online content

Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41590–024-01936–4.

Methods

TCGA dataset analysis for immune microenvironment and survival analysis

We used RNA-seq data from 18 non-hematological TCGA tumor types. Cancer types profiled include: adrenocortical carcinoma (ACC), bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), colon adenocarcinoma (COAD), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), cholangiocarcinoma (CHOL), head and neck squamous cell carcinoma (HNSC), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), liver hepatocellular carcinoma (LIHC), Lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), Pancreatic adenocarcinoma (PAAD), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), skin cutaneous melanoma (SKCM) and stomach adenocarcinoma (STAD). RNA-seq data were downloaded from the Genomic Data Commons (GDC) pan-cancer portal (https://gdc.cancer.gov/about-data/publications/pancanatlas/)³⁵. Data were processed using the Firehose pipeline with upper-quantile normalization. The primary tumor sample was favored for individuals with more than one RNA-seq sample. xCell, a gene signatures-based enrichment approach, was used to delineate the enrichment of 64 immune and stromal cell types as described previously³⁶. Briefly, the xCell R package was used to generate raw enrichment scores, transform into linear scale, and apply a spillover compensation to derive corrected enrichment scores. The distribution of enrichment scores for individuals with high (>50%) and low (<50%) levels of ZBTB46 expression were compared. The Mann–Whitney test was used to calculate statistical significance with an alpha value of 0.05.

For survival analysis, the Pathology section of The Human Protein Atlas was accessed through its website (https://www.proteinatlas.org/) and searched for ZBTB46 gene expression in different tumor types along with the clinical outcome^37,38. Briefly, TCGA project of GDC collects and analyzes multiple human cancer samples. RNA-seq data from 17 cancer types representing 21 cancer subtypes with a corresponding major cancer type in the Human Pathology Atlas were included to allow for comparisons between the protein staining data from the Human Protein Atlas and RNA-seq from TCGA data. The TCGA RNA-seq data were mapped using the Ensembl gene ID available from TCGA, and the number fragments per kilobase of exon per million reads (FPKMs) for each gene were subsequently used for quantification of expression with a detection threshold of 1 FPKM. Based on the FPKM value of each gene, individuals were classified into two expression groups, and the correlation between expression level and survival was examined. The prognosis of each group of individuals was examined by Kaplan–Meier survival estimators, and the survival outcomes of the two groups were compared by log-rank tests.

Animals

C57BL/6 mice were used as WT mice in this study. Zbtb46^gfp/gfp (ZKO) mice were a gift from K. M. Murphy at Washington University in St. Louis. We also obtained Zbtb46^fl/fl mice (B6(Cg)-Zbtb46tm2Mnz/J, strain no. 026850) and crossed them with VEC-Cre (B6.FVB-Tg(Cdh5-cre)7Mlia/J, strain no. 006137), CD11c-Cre (B6.Cg-Tg(Itgax-cre)1–1Reiz/J, strain no. 008068) and VAV-Cre (B6.Cg-Commd10Tg(Vav1-icre)A2Kio/J, strain no. 008610) mice (The Jackson Laboratory,) to generate VEC-Cre; Zbtb46^fl/fl conditional KO mice (mostly EC-specific ZKO or VEC ZKO), CD11c-Cre; Zbtb46^fl/fl conditional KO mice (DC-specific ZKO or CD11c ZKO) and VAV-Cre; Zbtb46^fl/fl conditional KO mice (hematopoietic ZKO or VAV ZKO). Moreover, we generated tamoxifen-inducible VEC-creERT2 Zbtb46 conditional KO (EC-specific inducible ZKO or iVEC-cre ZKO) mice by crossing the Tg(Cdh5-cre/ERT2)1Rha mice (model no 13073, Taconic Biosciences) with the Zbtb46^fl/fl mice. For cre induction, tamoxifen (T5648, MilliporeSigma; 10 mg ml⁻¹ in corn oil and ethanol) was injected to the postnatal day (P)18–20 pups for three consecutive injections. A fourth booster dose of 1 mg per mouse was injected a month later. Deletion efficiency and tumor studies were performed 1 week after the fourth dose of Tamoxifen. MMTV-PyMT mice were a gift from M. Egeblad, Cold Spring Harbor Laboratory, and crossed with Zbtb46^gfp/gfp mice to generate ZKO in the presence of MMTV-PyMT transgene (MMTV-PyMT Zbtb46^gfp/gfp). All the mice used in the study were backcrossed into C57BL6 for at least ten generations. Littermate subjects were used as a control with the different KO mice. Both male and female mice were used equally in any given experiment except in experiments utilizing both the genetic and orthotopic breast tumor models, where only female mice were used. The ages of the experimental animals were between 10 and 12 weeks. All test animals were kept in the Washington University animal facility at an ambient temperature of 22°C ± 5 °C and a humidity of 50–70%. A 12-h day–night cycle was maintained to avoid fluctuations in the circadian rhythm.

Study approval

Animal husbandry, generation, handling and experimentation were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee of Washington University School of Medicine in St. Louis.

BM chimeric mice generation

As described previously³⁹, WT recipient mice (CD45.1) were lethally irradiated with 950-rad irradiation. Donor BM from the control (CD45.2) and ZKO (CD45.2) mice was transplanted into the recipient mice retro-orbitally 24 h after irradiation. Flow cytometric analysis of the PB after 5 months of transplantation confirmed the successful BM reconstitution and generation of ‘hematopoietic ZKO BM chimeric’ mice. Alternatively, lethally irradiated WT (CD45.2) and ZKO (CD45.2) recipient mice received BM transplantation from WT (CD45.1) donors and generated ‘stromal ZKO BM chimeric’ mice.

Arterial blood pressure, heart rate and pressure-diameter measurement in mice

As described previously⁴⁰, ZKO and WT littermates were secured under 1.5% isoflurane anesthesia. A Millar pressure catheter (SPR-671, Millar) was introduced to the ascending aorta, and heart rate, arterial systolic, diastolic and mean blood pressures were recorded using the PowerLab data acquisition system (ADInstruments). The ascending aorta and left common carotid artery of the mice were dissected and mounted on metal cannulae in a pressure myograph (Danish Myo Technology). Intravascular pressure was increased from 0 to 175 mm Hg in increments of 25 mm Hg, and the vessel diameter was recorded at each pressure point. The average of three measurements at each pressure was reported.

Mammary tumorigenesis

MMTV-PyMT transgenic mice were utilized to generate a spontaneous model of breast cancer, where MMTV-LTR drives the expression of mouse mammary gland-specific polyomavirus middle T-antigen⁴¹. Palpable tumors in the mammary gland of the MMTV-PyMT Zbtb46^gfp/gfp (ZKO) mice were measured every week until 21 weeks of age to track the development and progression of tumorigenesis. Tumor volume was calculated by the equation: volume = (largest diameter) × (smallest diameter)² × 0.5.

Tumor transplantation studies

In total, 1 ml of growth factor-reduced Matrigel (354248; Corning) was mixed with 1 ml of cultured GFP-expressing LLC tumor cell²⁶ suspension (2 × 10⁶ per ml in PBS); 100 μl of the mixture was subcutaneously injected into the back of the mice. 1956 and 1969 sarcoma cells^42,43 were subcutaneously injected as 1 × 10⁶ cells in 150 μl PBS + Matrigel solution (1:1) per mouse to the flank of the mice. PyMT-BO1 cells⁴⁴ were orthotopically injected as 1 × 10⁵ cells in 50 μl PBS + Matrigel solution (1:1) per mouse to the mammary fat pad of the mice, as described previously⁴⁵. Palpable tumors started to develop 4–5 days after transplantation, and tumor growth was measured until the end of the study. For co-transplantation assay, a 1:2 ratio of 1956 sarcoma tumor cells and ECs was mixed in a 1:1 solution of PBS and Matrigel, and a total of 2 × 10⁶ cells in a 200 μl final volume per mouse was injected into the back of the mice. For overexpression studies, relevant lentiviral particles were intratumorally injected as 15 μl per injection for a viral content of 2 × 10⁶ transducing units per injection, as many times as indicated in the relevant figures. For in vivo treatment studies, rat IgG2aκ anti-mouse PD1 antibody (P372, clone: RMP1–14, Leinco Technologies) was injected intraperitoneally at a dosage of 200 μg per day, and DC101 (anti-VEGFR2) was injected intraperitoneally at a dose of 40 mg per kg of body weight (BE0060, Bio X Cell). Rat IgG2a isotype control at an equivalent dose was used as control.

Preparation of Zbtb46 mRNA-p5RHH peptide nanoparticles

For the mRNA nanoparticle treatment study, the Zbtb46 ORF sequence

‘ATGAACAACCGAAAGGAAGATATGGAAATCACTTCTCACTACCGGCATCTGCTTCGAGAGCTCAATGAGCAGAGGCAGCACGGAGTCCTCT-GTGATGCGTGCGTCGTGGTGGAGGGCAAGGTCTTCAAGGCACATAA-GAACGTCTTGCTTGGGAGCAGCCGCTACTTTAAGACGCTCTACTGC-CAGGTACAGAAGACATCTGACCAGGCCACCGTCACTCACTTGGACATT-GTTACAGCCCAGGGCTTCAAGGCCATTATTGACTTCATGTACTCCGCC-CATCTGGCTCTCACTAGTAGGAATGTCATCGAGGTGATGTCAGCTGC-CAGCTTCCTACAGATGACTGACATTGTGCAGGCCTGCCATGATTTCATCAAGGCTGCACTGGACATCAGCATAAAGTCAGATGCCTCCGAT-GAACTCTCAGAATTTGAGATTGGCACCCCAGCCAGCAACAGTACAGAG-GCGTTGATCTCAGCTGTGATGGCTGGAAGGAGTATCTCCCCATGGTTG-GCTCGGAGAACAAGTCCTGCCAATTCTTCTGGAGACTCTGCCATTGC-CAGCTGTCATGAAGGAGGAAGCAGCTATGGGAAGGAGGACCAGGAAC-CCAAAGCTGATGGCCCTGATGACGTTTCTTCACAGTCTTTGTGGCCTG-GAGATGTAGGCTATGGGTCTCTGCGCATCAAGGAAGAACAGATTTCAC-CATCACATTATGGAGGGAGTGAGCTTCCATCTTCCAAGGACACTGCAATACAGAATTCTTTATCAGAACAGGGTTCTGGGGATGGCTGGCAGCC-CACAGGCCGGAGGAAGAATCGGAAAAACAAAGAGACTGTCCGACACATCACCCAGCAGGTGGAGGAGGACAGCCAGGCTGGCTCTCCAG-TACCTTCATTCCTACCCACATCGGGATGGCCTTTCAGCAGCCGAGA-CTCAAATGTAGACCTGACGGTCACTGAGGCCAGCAGCTTGGACAGC-CGAGGCGAGAGAGCAGAGCTCTATGCTCACATCGATGAGGGCCTAC-TAGGAGGAGAAACCAGCTACTTGGGCCCACCCCTCACCCCAGAGAA-GGAAGAAGCACTACACCAGGCTACTGCAGTGGCCAATCTTCGTGCTG-CACTCATGAGTAAGAACAGTCTGCTGTCACTCAAGGCTGACGTGCTCG-GTGATGATGGCTCACTTCTGTTCGAGTACCTGCCCAAAGGTGCCCACTCACTGTCTCGTAAGTGCAAGTTCTGGTGTGTCACTGTGTCTTCCTTTGGTTTAAGCACCTCAGTTCAGCCCTTCAGACCCTGGAGTCACTGA’ was made into a modified mRNA transcript with complete substitution of pseudo-U in RNase-free water from TriLink Biotechnologies. CleanCap EGFP mRNA (L-7201–100) was used as mRNA control. In total, 8 μl (1 μg per μl) of the mRNA solution was mixed with 5 μl of 20 mM p5RHH peptide solution and 187 μl of 1× HBSS (Gibco) to prepare the nanoparticle complex and immediately injected into the mouse through the tail vein^{16,26,27,46–49}.

Flow cytometric analysis

BM was isolated from the experimental animals by flushing the tibia and femur. Spleen was collected and meshed into a single-cell suspension using the back side of a sterile 5 ml syringe plunger. PB was collected by intracardiac puncture from the euthanized animals immediately and processed for making a single-cell suspension following standard procedure. Lungs were minced and processed by standard enzymatic digestion consisting of collagenase I (LS004194, Worthington). Tumor-draining lymph nodes were isolated, disrupted and digested with collagenase IV (LS004186, Worthington). Tumor tissues were harvested, minced into fine pieces and dissociated into single-cell suspensions with an enzymatic digestion buffer consisting of collagenase II (for subcutaneous tumors; LS004176, Worthington) or collagenase III (for breast tumors; LS004182, Worthington), along with Dispase II (D4693, MilliporeSigma) and DNase I (LS002139, Worthington). Next, the cell suspensions were incubated with LIVE/DEAD Fixable Blue Dead Cell Stain Kit (L34961) along with different panels of fluorophore-conjugated surface staining antibodies. For subsequent intracellular staining, cell suspensions were fixed and permeabilized using either a Foxp3/transcription factor staining buffer set (00–5523-00, Thermo Fisher Scientific) or an intracellular fixation and permeabilization buffer set (88–8824-00, Thermo Fisher Scientific) and subsequently stained with intracellular antibodies. Samples were analyzed using either a BD LSRFortessa X-20 (BD Biosciences) or a BD FACSymphony A3 (BD Biosciences); data were collected with FACSDiva software (BD Biosciences) and later processed with FlowJo software (BD Biosciences). CD45⁻CD31⁺ ECs and CD45⁺CD11c⁺MHCII⁺ cDCs were FACS sorted using a BD FACSAria II (BD Biosciences). Sorted cells were purity tested by secondary flow cytometry and later processed for downstream applications including total RNA isolation using an RNeasy mini kit (74104, Qiagen) following the manufacturer’s instructions. Markers used for different cell lineages were:

Lineage (Lin): CD3⁺Ter119⁺B220⁺Gr1⁺

MkP (Megakaryocyte Progenitors): Lin⁻Sca1⁻cKit⁺CD41⁺CD150⁺

GMP (Granulocyte-Macrophage Progenitors): Lin⁻Sca1⁻cKit⁺CD

41⁻CD150⁻CD16/32⁺

Pre-cDC: Lin⁻Sca1⁻IL7R⁻MHCII⁻CD16/32⁻CD11c⁺cKit⁻CD135⁺

CMP (Common Myeloid Progenitor): Lin⁻Sca1⁻IL7R⁻MHCII⁻CD16

/32⁻CD11c⁻CD41⁻CD135⁺cKit^hi

CDP (Common DC Progenitor): Lin⁻Sca1⁻IL7R⁻MHCII⁻CD16/32⁻C

D11c⁻CD41⁻CD135⁺cKit^int cDC: CD45⁺CD11c⁺MHCII⁺ cDC1: CD45⁺CD11c⁺MHCII⁺Xcr1⁺ CTL (Cytotoxic T Lymphocytes): CD45⁺CD3⁺CD8⁺

T_reg: CD45⁺CD3⁺CD4⁺CD25⁺Foxp3⁺

M1 (M1 Macrophage): CD45⁺CD11b⁺F4/80⁺iNOS⁺

M2 (M2 Macrophage): CD45⁺CD11b⁺F4/80⁺CD206⁺CX3CR1⁺

B: CD45⁺B220⁺

NK (Natural Killer cells): CD45⁺NK1.1⁺

EC: CD31⁺CD45⁻

FACS analysis was done with the following antibodies: anti-mouse CD3 APC/Cy7 (100221, BioLegend), anti-mouse CD4 PE (100408, BioLegend), anti-mouse CD8a PE/Cy5 (100710, BioLegend), anti-mouse CD11b BUV737 (612800, BD Biosciences), anti-mouse CD11c BUV496 (750483, BD Biosciences), anti-mouse CD25 BV650 (102038, BioLegend), anti-mouse CD41 PE/Dazzle 594 (133935, BioLegend), anti-mouse CD45 BV605 (103155, BioLegend), anti-mouse CD49b BUV563 (741280, BD Biosciences), anti-mouse CD86 BUV661 (741502, BD Biosciences), anti-mouse CD127 PE/Cy5 (15–1271-81, Thermo Fisher), anti-mouse CD172a BUV805 (741997, BD Biosciences), anti-mouse CD206 APC Alexa Fluor 700 (141734, BioLegend), anti-mouse Ly6C BV510 (128033, BioLegend), anti-mouse F4/80 PerCp/Cy5.5 (123127, BioLegend), anti-mouse B220 BUV395 (563793, BD Biosciences), anti-mouse MHCII (I-A/I-E) BV711 (107643, BioLegend), anti-mouse CX3CR1 PE/Dazzle 594 (149014, BioLegend), anti-mouse XCR1 APC (148205, BioLegend), anti-mouse CD117 APC/Cy7 (10–1172-82, Thermo Fisher), anti-mouse B220 PE/Cy7 (103222, BioLegend), anti-mouse CD3e PE/Cy7 (100320, BioLegend), anti-mouse Ter119 PE/Cy7 (116222, BioLegend), anti-mouse Gr1 PE/Cy7 (108416, BioLegend), anti-mouse Sca1 PerCp/Cy5.5 (108124, BioLegend), anti-mouse CD150 BV785 (115937, BioLegend), anti-mouse CD48 BUV395 (740236, BD Biosciences), anti-mouse CD135 BV421 (562898, BioLegend), anti-mouse CD16/32 PE (553145, BioLegend) and anti-mouse CD115 BV605 (135517, BioLegend). Antibodies were used at a dilution of 1:200.

Immunofluorescence studies

Harvested tumors were thinly sliced and fixed in 10% buffered formalin (16004–112, VWR), immersed in 30% (wt/vol) sucrose solution for 48 h to cryo-protect the tissue, frozen in NEG-50 frozen section medium (6502, Thermo Fisher Scientific) using liquid nitrogen and 2-methyl butane system, and sectioned at a thickness of 8 μm using a Leica Cryostat microtome (CM1850). Afterward, tissue sections were blocked using freshly made blocking buffer (3% essentially IgG-free BSA (A9085), 0.3% Triton X-100 (X100) and Fc blocker (101301)). Next, sections were incubated with different primary antibodies for 16 h at 4 °C, followed by visualization with appropriate secondary antibodies. Finally, the sections were counterstained for nuclei with DAPI, cured with ProLong Diamond Antifade mountant and sealed with nail polish for preservation. To detect intratumoral hypoxia in mice, Hypoxyprobe-1 solution (HP6–100Kit) was intraperitoneally administered 90 min before death as 100 mg per kilogram of body weight. FITC-conjugated anti-pimonidazole mouse IgG1 monoclonal antibody was used to detect the extent of hypoxia. For the vascular leakage and perfusion experiments, FITC-conjugated 70-kD dextran (60 mg per kilogram of body weight; 46945) and FITC-conjugated lectin (8 mg per kilogram of body weight; L9381), respectively, were injected through the tail vein 15 min before death. Hamster anti-mouse CD31 (clone: 2H8, MA3105) was used as the primary antibody. AF568 goat anti-hamster (A-21112) was utilized as the secondary antibody. The sections were examined using the Olympus Fluoview 1200 confocal microscope system and minimally processed with Imaris (Bitplane) software (version 9.1). At least five pictures from every section were processed using ImageJ software (National Institutes of Health (NIH)) for quantification purposes.

Cell lines

MCECs were purchased from CELLutions Biosystems (CLU510). 1956 and 1969 sarcoma cell lines were obtained from R. D. Schreiber at Washington University in St. Louis. Zbtb46-overexpressing MCEC and 1956 cell lines were generated by transducing parental MCEC and 1956 cells with Zbtb46-overexpressing lentiviral particles followed by blasticidin selection (A1113903, Thermo Fisher Scientific). ZKD MCEC and 1956 cell lines were generated by transducing the parental MCECs and 1956 cells with lentiviral particles from a set of five shRNA clones for Zbtb46 followed by puromycin selection (A1113802, Thermo Fisher Scientific). LLC-GFP cells (American Type Culture Collection (ATCC) CRL-1642) were a gift from A. S. Krupnick, University of Virginia. PyMT-BO1 cells were obtained from K. N. Weilbaecher at Washington University in St. Louis. HEK293T cells (ATCC, CRL-3216) were purchased from the ATCC. All cell lines tested negative for mycoplasma contamination.

Cell culture

LLC-GFP, PyMT-BO1 and HEK293T cells were cultured in DMEM high-glucose (11965092, Thermo Fisher Scientific) growth medium supplemented with 10% (vol/vol) FBS (12103C, MilliporeSigma), 100 units per ml of penicillin–streptomycin (15140122, Thermo Fisher Scientific). 1956 and 1969 sarcoma cells were cultured in RPMI 1640 growth medium supplemented with 10% (vol/vol) FBS, 100 units per ml of penicillin–streptomycin, 1% (vol/vol) l-glutamine (200 mM; BW17–605E, Thermo Fisher Scientific), 1% (vol/vol) sodium pyruvate (100 mM; BW13–115E, Thermo Fisher Scientific), 0.5% (vol/vol) sodium bicarbonate (7.5% wt/vol stock; BW17–613E, Thermo Fisher Scientific) and 0.1% (vol/vol) 2-mercaptoethanol (M-6250, MilliporeSigma). All MCEC lines were maintained in M199 growth medium (11150067, Gibco), supplemented with 20% (vol/vol) FBS, 100 units per ml of penicillin–streptomycin and 10 mM HEPES (15630106, Thermo Fisher Scientific). In BM cell-related experiments, BM cells were cultured in Iscove’s Modified Dulbecco’s media supplemented with 10% (vol/vol) FBS (12103C, MilliporeSigma), 100 units per ml of penicillin–streptomycin (15140122, Thermo Fisher Scientific).

Lentiviral shRNA and overexpression particle production

pLKpuro lentiviral mouse Zbtb46 shRNA clones TRCN0000125839 (NM_028125.1–2420s1c1), TRCN0000125840 (NM_028125.1–364s1c1), TRCN0000125841 (NM_028125.1–1358s1c1), TRCN0000125842 (NM_028125.1–321s1c1) and TRCN0000125843 (NM_028125.1–1231s1c1) and mouse Cebpb shRNA clones TRCN0000231407 (NM_009883), TRCN0000231408 (NM_009883), TRCN0000231410 (NM_009883), TRCN0000231411 (NM_009883) and TRCN0000231409 (NM_009883) were purchased from MilliporeSigma. HEK293T cells were transfected with the mentioned shRNA clones or pCSII-EF1-Zbtb46-IRES2-Bsr or pCSII-EF1-Cebpb-IRES2-Bsr constructs along with pCAG-HIVgp and pCMV-VSV-G-RSV-Rev (with a ratio of 4:3:1) by using a calcium phosphate method. Sixteen hours after transfection, the medium was changed, and cells were grown for an additional 48 h. Subsequently, the supernatant was harvested and concentrated by Lenti-X-Concentrator (631232, Clontech). The virus titer was determined using the Lenti-X p24 Rapid Titer Kit (632200, Clontech).

BM cell differentiation assay

Total BM cells were harvested by flushing out the tibia and femur of the experimental animals. Cells were transfected with different lentiviral particles by spin infection at 800g for 30 min at 32 °C in the presence of 2 μg ml⁻¹ polybrene. Cells were later cultured in complete IMDM media with 10 ng ml⁻¹ of Flt3L (250–31L, PeproTech), stem cell factor (250–03, PeproTech), GM-CSF (315–03, PeproTech) and G-CSF (250–05, PeproTech). Cells were harvested and analyzed by FACS after 4 days of culture.

KSL (Lin⁻cKit⁺Sca1⁺), CMP (Lin⁻cKit⁺CD34⁺CD16/CD32^int) and GMP (Lin⁻cKit⁺CD34⁺CD16/CD32^hi) cell differentiation assay

Total BM cells were harvested by flushing out the tibia and femur of the experimental animals. Cells were then stained with PE/Cy7-conjugated anti-Gr-1 (RB6–8C5), CD11b (M1/70), B220 (RA3 6B2), Ter119 (TER119) and CD3 (145–2C11), in combination with APC/Cy7 c-Kit (2B8), PerCp/Cy5.5 Sca1 (E13–161.7), PE CD34 (RAM34) and BV421 CD16/CD32 antibodies for 40 min on ice. KSL (Lin⁻cKit⁺Sca1⁺), CMP (Lin⁻cKit⁺CD34⁺C D16/32^int) and GMP (Lin⁻cKit⁺CD34⁺CD16/32^hi) cells were sorted on a FACS Aria II (BD Biosciences) sorter using 85-μm nozzles directly into a round-bottom 96-well plate at a density of 1,000 cells per well. Culture media consisted of StemSpan serum-free base medium (StemCell Technologies), 10% FBS, penicillin (50 U ml⁻¹) and streptomycin (50 U ml⁻¹), stem cell factor (25 ng ml⁻¹, PeproTech), FLT3L (20 ng ml⁻¹, PeproTech), IL-3 (1% supernatant), murine thrombopoietin (20 ng ml⁻¹, PeproTech), IL-6 (10 ng ml⁻¹, PeproTech), IL-11 (10 ng ml⁻¹, PeproTech), M-CSF (10 ng ml⁻¹, PeproTech), G-CSF (10 ng ml⁻¹, PeproTech) and GM-CSF (10 ng ml⁻¹, PeproTech). Cells were harvested and analyzed by FACS after 2 to 6 days of culture as indicated in the relevant figures.

BM monocyte enrichment

Tibias and femurs of the experimental animals were harvested and flushed, and later monocytes were enriched from the total BM cells using the monocyte isolation kit (BM) from Miltenyi Biotec (130–100-629) through a negative selection process following the manufacturer’s instruction. This enriched population contained mostly monocytes and precursor lineage-negative cells.

ChIP–qPCR assay

As described previously⁵⁰, enriched monocyte and lineage-negative cells from BM were cross-linked with 1% formaldehyde and lysed with cell membrane lysis buffer, followed by further incubation with nuclear membrane lysis buffer. For the Zbtb46 overexpression system, before proceeding to lysate preparation, after the isolation of the enriched population, Zbtb46-HA overexpression lentiviral particles were used to transduce the cells by spin infection at 800g for 30 min at 32 °C in the presence of 2 μg ml⁻¹ polybrene and then incubated at 37 °C for 24 h in complete IMDM media. The lysate was incubated with either anti-CEBPb (23431–1-AP, Proteintech) or anti-HA (ab9110, Abcam) followed by protein A/G Sepharose beads (sc-2002, Santa Cruz Biotechnology). The beads were washed to isolate immunoprecipitated DNA fragments that were later subjected to qPCR. For the anti-ZBTB46 experiment, Cebpb peak-1 enrichment (DNA location Chr20: 50189053–50189756; corresponding mouse region: Chr2: 167687660–167688195)⁵¹ was assessed by qPCR using the primer set forward sequence ‘CCCCAGCTCAGCAGATAACA’ and reverse sequence ‘AGGCTTCTCAGGTGATTGCG’. For the anti-CEBPb experiment, Cebpb peak-1 enrichment (DNA location Chr2: 167687679–167688023)⁵² was assessed by qPCR with the primer set forward sequence ‘AAGGGCACAGGGAGATGTCA’ and reverse sequence ‘GGTGTTGCTCAACCTTCGGT’; Csf3r peak-6 enrichment (DNA location Chr4: 126017669–126017995)⁵² was assessed by qPCR with primer set forward sequence ‘GACAACGCTGGCACTTTTGTA’ and reverse sequence ‘TGTGCAAGCAGGTCATTGTG’.

Intratumoral transfer of enriched BM monocytes

Enriched monocytes and precursor lineage-negative cells from CD45.1 WT donor mice were transduced with either EV-mCherry or Zbtb46-mCherry lentiviral particles by spin infection at 800g for 30 min at 32 °C in the presence of 2 μg ml⁻¹ polybrene. Later, as described previously⁵³, the cells were resuspended in PBS as 1 × 10⁵ cells per 50 μl and injected intratumorally into the day-10 post-transplant 1956 sarcoma tumor-bearing CD45.2 WT recipient mice. Tumor volumes were measured periodically and harvested at day 15 after transplantation for flow cytometric analysis to track donor-derived (CD45.1⁺) transduced (mCherry⁺) cell polarization into either macrophages or DCs.

Tube formation assay

Overnight serum-starved ECs were plated on Matrigel (96992, Corning)-coated 24-well plates (3 × 10⁴ cells per well) and incubated at 37 °C for 6 h before imaging with a Leica DFC 310 FX microscope system as described before²⁶. The angiogenesis analyzer module of ImageJ software (NIH) was used to quantify the total tube length, number of loops and number of branches.

Fibrin gel bead sprouting angiogenesis assay

As described previously⁵⁴, parental and modified ECs were coated on Cytodex 3 microcarrier beads (17–0485-01, GE Healthcare). Coated beads were then resuspended in Fibrinogen solution (F8630, MilliporeSigma), supplemented with aprotinin (A1153, Millipore-Sigma). Next, 500 μl of this bead suspension was transferred to each well of a 24-well plate containing 20 μl of thrombin (50 unit per ml in DPBS; T3399, MilliporeSigma). After the gel solidified, NIH/3T3 mouse fibroblast cells were added to the wells. The wells were monitored for the sprouting using a bright-field microscope, and images were taken using a Leica DFC 310 FX microscope system. ImageJ (NIH) was used to assess the number of sprouts per bead and average sprout lengths.

Matrigel plug assay

As described before⁵⁵, 500 μl of overnight-thawed Matrigel (96992, Corning) was subcutaneously injected to the flanks of the experimental mice. On day 14 after injection, the mice were euthanized and Matrigel plugs harvested. Harvested plugs were fixed in 10% buffered formalin (16004–112, VWR), immersed in 30% (wt/vol) sucrose solution for 48 h to cryo-protect the tissue, frozen in NEG-50 frozen section medium (6502, Thermo Fisher Scientific) using liquid nitrogen and 2-methyl butane system, and sectioned at a thickness of 8 μm using a Leica Cryostat microtome (CM1850, Germany). Afterward, plug sections were processed for immunofluorescence imaging following standard procedure as described above.

T cells trans-endothelial migration assay

As described before¹⁶, T cell migration potential across an endothelial barrier in vitro was assessed using a QCM leukocyte trans-endothelial migration colorimetric assay kit (ECM557, MilliporeSigma) following the manufacturer’s instruction. T cells used in the study were isolated from the spleens harvested from 1956 sarcoma subcutaneous tumor-bearing WT mice with APC CD45 (103112, clone: 30-F11), PE CD3 (100206, clone: 17A2), BV650 CD4 (100469, clone: GK1.5) and PerCP/Cy5.5 CD8a (100734, clone: 53–6.7). Cytotoxic T cells were sorted using a BD FACSAria II (BD Biosciences) as the CD45⁺CD3⁺CD4⁻CD8⁺ population. Endothelial monolayers were treated with either tumor necrosis factor (TNF; 20 ng ml⁻¹) for 12 h and washed with PBS before placing the harvested T cells on the endothelial monolayer. The relative abundance of migrated T cells was calculated by measuring the absorption of the samples at 450 nm following the WST-1 reagent staining.

Real-time RT–qPCR

cDNA was prepared with qScript cDNA SuperMix (101414–106, VWR) according to the manufacturer’s protocol. Gene expression was measured by real-time RT–qPCR using these primer sets: Zbtb46 (forward ‘ATCACTTCTCACTACCGGCAT’ and reverse ‘AAGACGTTCTTATGTGCCTTGAA’), Cebpb (forward ‘CGCCTTATAAACCTCCCGCT’ and reverse ‘TGGCCACTTCCATGGGTCTA’), Runx3 (forward ‘CAGGTTCAACGACCTTCGATT’ and reverse ‘GTGGTAGGTAGCCACTTGGG’), Irf4 (forward ‘TCCGACAGTGGTTGATCGAC’ and reverse ‘CCT-CACGATTGTAGTCCTGCTT’), Ifi44 (forward ‘AACTGACTGCTCGCAATAATGT’ and reverse ‘GTAACACAGCAATGCCTCTTGT’), Itgax (forward ‘CTGGATAGCCTTTCTTCTGCTG’ and reverse ‘GCACACTGTGTCCGAACTCA’), Ccr7 (forward ‘TGTACGAGTCGGTGTGCTTC’ and reverse ‘GGTAGGTATCCGTCATGGTCTTG’), Lifr (forward ‘TACGTCGGCAGACTCGATATT’ and reverse ‘TGGGCGTATCTCTCTCTCCTT’), Mertk (forward ‘CAGGGCCTTTACCAGGGAGA’ and reverse ‘TGTGT-GCTGGATGTGATCTTC’), Mafb (forward ‘TTCGACCTTCTCAAGTTCGACG’ and reverse ‘TCGAGATGGGTCTTCGGTTCA’), Vegfa (forward ‘CTGCCGTCCGATTGAGACC’ and reverse ‘CCCCTCCTTGTACCACTGTC’) and Csf3r (forward ‘CTGATCTTCTTGCTACTCCCCA’ and reverse ‘GGTGTAGTTCAAGTGAGGCAG’),

Bulk RNA-seq

RNA-seq data from WT and Zbtb46-overexpressing MCECs were preprocessed and analyzed for differentially expressed genes using edgeR⁵⁶. Genes with fewer than five counts in at least three samples were filtered out. Count data were normalized to account for varying library sizes. Normalized data were fit to a negative binomial generalized log-linear model, and differentially expressed genes were extracted. In-built functions for GO analysis were used to identify GO terms enriched in WT and Zbtb46-overexpression MCECs.

Single-cell RNA-seq of TME

Single-cell RNA-seq analyses of the non-tumor components of the PyMT-BO1-bearing mice on day 16 after either EV control or Zbtb46-overexpressing lentiviral intratumoral treatments on days 9, 12 and 15 after tumor transplantation were performed. Briefly, on day 16, tumor masses from a total of five mice were pulled in each group. Following enzymatic digestion, the cells were FACS sorted as the live non-tumor (GFP−) population. For multiplexing, cells were split into separate tubes as either mCherry⁺ or mCherry⁻ cells (mCherry is a fluorescence reporter included in both the EV and Zbtb46-overexpressing vector), marked with distinctive hashtags (Hashtag-B0301 for mCherry⁺ and Hashtag-B0302 for mCherry⁻ cells), and pulled back to make two samples: control and treatment. The cells were then processed for single-cell library generation, using the 10x Genomics platform, and sequenced with the Illumina NovaSeq 6000. Unsupervised k-nearest-neighbors graph-based community detection cell clustering analysis of 36,757 cells resulted in distinct clusters of monocyte/macrophages, neutrophils, fibroblasts, ECs, DCs, T cells and NK cells. We computationally separated different major cell types and reanalyzed them at higher granularity, as described below.

Single-cell RNA-seq processing and analysis

Sequencing outputs were aligned to the mm10 reference genome and feature-counted using Cell Ranger 7.2.0. Filtered Cell Ranger outputs were then processed and analyzed using R (version 4.3.0) and Seurat (v5.0)⁵⁷. Preliminary quality control of each sample was performed using miQC (v1.10.0)⁵⁸ and DoubletFinder (v2.0.4)⁵⁹ to remove low-quality cells and doublets. Low-quality cells were identified using miQC with a posterior cutoff of 0.75, and doublet predictions were made using DoubletFinder with the top 30 principal components (PCs), pN of 0.25 and pK of 0.09, assuming a 7.5% doublet formation rate. Transcript levels were log-normalized using a scaling factor of 10,000, and variable features were identified using the mean.vars. plot method. All variable features except T cell antigen receptor chain A and chain B genes were scaled and used for principal component analysis. Samples were then integrated using Harmony (v1.2.0)⁶⁰. Uniform manifold approximation and projection and t-SNE embeddings were calculated using the top 30 PCs after Harmony correction. Unsupervised clustering was performed using FindNeighbors and FindClusters functions with the top 30 Harmony embeddings and default parameters. Preliminary annotation was performed using a clustering resolution of 0.6, and the dataset was divided into lymphoid, myeloid and stromal compartments for analysis and annotation at higher resolutions⁶¹. Additional doublets were removed based on the gene expression signature of clusters within each compartment. The processed and annotated subsets were merged to create the final whole dataset. The average expression levels of genes of interest for clusters from the whole dataset and the subsets were calculated using the AggregateExpression function from Seurat.

Differential expression analysis

Differentially expressed genes were calculated using the function FindAllMarkers in Seurat by comparing clusters within the same cell type but different subtypes. Default parameters were used.

Differential expression analysis between conditions within a cluster was performed for clusters of interest using the FindMarkers function with the MAST method, log₂ fold-change threshold of 0 and min. pct = 0. Genes with adjusted P values below 0.05 were considered as differentially expressed genes across conditions.

Differentially expressed genes were selected from markers and aggregated by condition and cluster identities. Average expression levels are then visualized using the Heatmap function from the ComplexHeatmap package (v2.18.0) with hierarchically clustered rows.

Gene-set enrichment analysis

Functional analysis of differentially expressed genes was performed using fgsea with provided genes ranked by log₂ fold-change⁶². For gene-set enrichment analysis of cross-cluster differentially expressed genes, all genes reported by FindAllMarkers were used as input. For gene-set enrichment analysis of cross-condition differentially expressed genes, genes with adjusted P values below 0.05 were used, except for ECs that had a limited number of genes with adjusted P values below 0.05 due to their small cluster size.

Analysis of antigen-presenting cell-like neutrophils

The marker spreadsheet was downloaded from supplementary data of ref. 63. Markers for neutrophil-derived antigen-presenting cells nAPC.1 and nAPC.2 were imported. Genes with log fold-change > 0 and expression below 80% of cells outside the cluster were sorted by the area under the curve method, and the top 100 genes were used to create corresponding nAPC signature gene sets, where nAPC.1 has 75 genes selected for gene-set creation and nAPC.2 has 100 genes in the gene set. Cross-cluster differentially expressed genes of neutrophils were used as input to fgsea.

Similarly, neutrophil markers generated from differential expression analysis with adjusted P value > 0.05 and expression in over 80% of outside-of-cluster cells were removed. The remaining genes are grouped by cluster identity and tested against the two nAPC gene sets.

Visualization

t-SNE plots were made using Seurat Dimplot and FeaturePlot functions. Gene expression heat maps were made using the Heatmap function from the ComplexHeatmap package (v2.18.0).

Clusters in different cell compartments

Lymphoid compartment.

CD4 T: CD4⁺ T cell cluster

T_act (activated T cells): Both CD4⁺ and CD8⁺ T cells; associated with T cell antigen receptor signaling and showed more of an inflammatory characteristic (for example, association with the TNF signaling via NF-kB pathways).

CD8T_s1 (T_naive): Featured by genes associated with naive T cells (for example, Sell and Cd7)^64,65.

CD8T_s2 (T_effector): Featured by genes associated with T cell activation-exhaustion (for example, Havcr2 (TIM-3), Lag3, Pdcd1, Isg15, Prf1 and Gzmb)^66,67.

CD8T_s3 (T_exhaust): Featured by genes associated with T cell activation-exhaustion (for example, Havcr2 (TIM-3), Lag3, Pdcd1, Isg15, Prf1 and Gzmb)^66,67 and terminal exhaustion (for example, Mt1)^68–70.

CD8T_s4 (T_p-ex) (proliferative-exhausted): Featured by exhaustion-related genes with high proliferative activities⁷¹.

NK_s1 (activate): Characterized by a more mature and active NK signature (for example, more cells expressing Tbx21, Lef1, Ccl4 and Ccl5). Also, the presence of more of Klrc2-expressing cells (an activating receptor for NK cells^72,73) and less of Klrg1-expressing cells (an inhibitory receptor for NK cells^74,75).

NK_s2 (inhibit): Featuring a more immature NK signature (for example, lack of cells expressing Tbx21, Lef1, Ccl4 and Ccl5). Additionally, the presence of less of Klrc2-expressing cells (an activating receptor for NK cells^72,73) and more of Klrg1-expressing cells (an inhibitory receptor for NK cells^74,75).

NK_s3: Featured by a somewhat similar gene expression profile as NK_s1 (activate) but with a stronger association with the cytokine signaling and inflammatory pathways.

NK_s4: Preferential expression of Cd69 and Isg15 (refs. 76,77) and was marked by JAK–STAT signaling pathways.

NK_s5: A proliferating NK cluster.

Myeloid compartment.

Mac_s1 (M1-like): Largest of the clusters. Cells in this cluster were characterized by the lack of alternatively activated and maturation markers such as Mrc1, Arg1, Trem2, Mertk and Cx3cr1 and having more expression of the classically activated and pro-inflammatory markers such as Nos2, Chil3, Ccr2 and Il1b^78–80. Enriched with inflammatory pathways such as TLR, NOD1/NOD2 and IL-1 signaling pathways.

Mac_s2 (M2-like): Featured by high expression patterns of the alternatively activated and maturation markers such as Mrc1, Mertk, Trem2 and Cx3cr1 and the lack of expression of Nos2.

Mac_s3 (Arg⁺ activated TAMs): Characterized by expression of the alternatively activated and maturation markers Mrc1 and Arg1. Enriched with TNF, IL-2–STAT5, IL-10 and TGF-β signaling pathways and associated with pro-tumor epithelial-to-mesenchymal-transition (EMT), degradation of the ECM and hypoxia pathways.

Mac_s4 (CD206^hi M2-like): Featured by very high expression of Mrc1, along with a high expression pattern of the alternatively activated and maturation markers such as Mertk, Trem2 and Cx3cr1 and the lack of expression of Nos2.

Mac_s5 (IFN-γ-responsive M1-like): Characterized by the expression of pro-inflammatory markers such as Nos2, Chil3, Ccr2, and Il1b, along with the interferon signaling-related genes such as Isg15, Ifit1, Ifit3, Ifitm3 and Rsad2. Enriched for interferon signaling pathways and other antitumor inflammatory and immunomodulatory pathways.

Mac_s6 (proliferating M2-like): Consists of anti-inflammatory macrophages positive for Mrc1, Mertk, Trem2 and Cx3cr1 that featured high proliferative activities.

DC_s1 (cDC1-like): Showed characteristics of cDC1 cells as they expressed canonical markers such as Zbtb46, Batf3, Irf8, Id2, Clec9a, Cadm1 and Xcr1 (ref. 81).

DC_s2 (migDC-like): Contained cells that shared characteristics of migratory DCs (for example, high expression of Socs2 and Cacnb3)⁸².

DC_s3 (moDC-like): Cells were positive for Irf4, Flf4, Cd14, Cd209a, Itgam, Mrc1 and Sirpa but were negative for Batf3 and Xcr1, indicating that they were possibly monocyte-derived DCs^81,83,84. Associated with pathways of the classical monocytes and were enriched for interferon response pathways.

DC_s4 (pDC-like): Shared characteristics of pDCs; for example, the cells in this cluster were positive for Ifr4, Irf7, Irf8 and Siglech, while lacking Batf3, Zbtb46 and Id2 (ref. 81).

Neu_s1: Featured by tumorigenic activation and NETosis (for example, high expression of s100a8 and enrichment of EMT, PI3K and hypoxia pathways)^85,86.

Neu_s2: Featured by preferential expression of pro-inflammatory molecules and enrichment of peroxisome and Hedgehog signaling pathways. Enriched with the characteristic pathways of DCs and gene signatures of the antigen-presenting neutrophils⁸⁷.

Neu_s3: Featured by characteristics of tumorigenic activation and NETosis (for example, high expression of s100a8 and enrichment of EMT, PI3K and hypoxia pathways)^85,86.

Stromal compartment.

Strm_s1 (endothelial): Featured by expressions of Myct1, Cdh5 and Pecam1.

Strm_s2 (low-angiogenic CAF): CAF cluster. Enrichment of pathways related to bile acid synthesis and metabolism, along with a reduction of pathways associated with EMT, angiogenesis and cell cycle.

Strm_s3: Characterized as a metabolically active inflammatory CAF cluster. Enriched for pathways related to glycolysis and mitochondrial ATP synthesis.

Strm_s4: Inflammatory CAFs. Characterized by more expression of genes associated with inflammation and potential metastatic pathways.

Strm_s5 (apCAF): Featured by the expression of genes associated with antigen-presenting CAFs such as H2-Ab1, Cd74, Lgals1 and Saa3 (ref. 88).

Strm_s6 (myCAF): Characterized by the features of myofibroblastic CAFs, such as the expression of Acta2, Rgs5 and Myh11 (refs. 89–91).

Statistical analysis

GraphPad Prism 10 software was used for performing statistical analysis and generating graphs/plots. Data are presented as the mean and s.d. for all the measurements. Statistical significance was determined by two-tailed unpaired Student’s t-test (for two groups) and one-way ANOVA with Dunnett’s or Tukey’s multiple-comparison test, as appropriate (for more than two groups). Non-parametric tests were used for non-log-transformed gene expression data from the TCGA database. P < 0.05 was considered statistically significant.

Graphical presentations

Schematics in Figs. 2–5, Extended Data Figs. 3 and 10 and Supplementary Fig. 5 were prepared with BioRender.com.

Reporting summary

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

Extended Data

Extended Data Fig. 1 |. Zbtb46 is a tumor suppressor and its expression is downregulated in tumor.

a, Zbtb46 expression in different cell population in mouse Trachea and Heart analyzed by single cell RNA sequencing. Data acquired from the Tabula Muris portal: https://tabula-muris.ds.czbiohub.org/. b-c, Analysis of (b) vital cardiovascular parameters and (c) pressure-diameter measurements in the Zbtb46 KO (n = 4) and wild-type (n = 11) mice. d, Zbtb46 expression in ECs and DCs isolated from the mammary gland (MEC and MDC) of the healthy and from the PyMT-BO1 tumor-bearing (TEC and TDC) wild-type mice (n = 4 for D19-TEC and -TDC, 5 for SDC, and 6 for the remaining groups). e-f, Zbtb46 expression in ECs from Lung (LEC) and mammary gland (MEC) of the healthy wild-type mice and from the tumors (TEC) of the (e) LLC carcinoma (n = 3) and (f) MMTV-PyMT (n = 4 for MEC and 3 for TEC) bearing mice. g-h, Analysis of ZBTB46 + (GFP+) (g) ECs and (h) DCs from the lungs and spleen of the healthy and tumors of the 1956 sarcoma-bearing Zbtb46^gfp/+ mice (n = 4). i, Measurements of B, CD8T, Treg, and XCR1⁺ cDC1-like myeloid cells in the tumor-draining lymph nodes of 1956 sarcoma-bearing mice (n = 3). j, Overall survival of patients with indicated cancer separated by ZBTB46 expression as high (>50th percentile) and low (<50th percentile). Survival data were derived from publicly available clinical records of TCGA patients. k, Activated DCs, M1 macrophages, B cells, and CD8 + T effector memory cells in high vs. low ZBTB46 expressing tumors in patients from the TCGA database analyzed with the CIBERSORT algorithm. Data are mean±SD. P-values were determined using two-tailed Student’s t-test for b, e-h, and i, One-way ANOVA with Tukey’s multiple comparison test for c or with Dunnett’s test for d, Log ranked test for j, and unpaired t-tests with Holm-Šídák multiple comparison test for k. ns=not significant.

Extended Data Fig. 2 |. Both endothelial and hematopoietic Zbtb46 expression contribute to the suppression of tumor progression and ZBTB46 maintains endothelial cells quiescent in the tumor context.

a, Stromal and hematopoietic Zbtb46 KO BM chimera generation scheme and FACS analysis of CD45.1 (for stromal KO) or CD45.2 (for hematopoietic KO) repopulation in PB (n = 4). b, 1956 sarcoma and PyMT-BO1 progression in wild-type (n = 7 for 1956 and 3 for PyMT-BO1), VEC-cre Zbtb46 KO (n = 4 for 1956 and 3 for PyMT-BO1), VAV-cre Zbtb46 KO (n = 7 for 1956 and 5 for PyMT-BO1), and Zbtb46 KO (n = 6 for 1956 and 3 for PyMT-BO1) mice. c, PyMT-BO1 growth in wild-type, VEC-cre Zbtb46 KO, CD11c-cre Zbtb46 KO, and Zbtb46 KO mice (n = 3). d, Zbtb46 expression in Lung-ECs from wild-type and Tamoxifen⁻inducible VEC-cre Zbtb46 CKO (iVEC-cre ZKO) mice (n = 3). e,1956 sarcoma growth in wild-type (n = 6), iVEC-cre ZKO (n = 6), and Zbtb46 KO (n = 5) mice. f-g, Representative images with quantification of tumoral CD31+ vessels in (f) wild-type (n = 10), VEC-cre Zbtb46 KO (n = 6), VAV-cre Zbtb46 KO (n = 7), CD11c-Cre Zbtb46 CKO (n = 4), Zbtb46 KO (n = 11), and (g) iVEC-cre ZKO mice (n = 10). h, Proliferation of the cultured parental, Zbtb46 knockdown (ZKD), and Zbtb46 overexpressing (ZOE) MCEC cells (pooled from 2 biological replicates, n = 3/replicates). i, Zbtb46 expression in tumor-ECs, tumor-DCs, and GFP+ tumors from 1956 sarcoma (n = 9) and PyMT-BO1 (n = 9 for tumor-EC and tumor-DC and 4 for tumor cells) -bearing wild-type mice with empty vector (EV) or Zbtb46 (ZOE) lentiviral overexpression construct intra-tumor treatment. j, 1956 sarcoma and PyMT-BO1 progression with empty vector (1956-EV and BO1EV) or Zbtb46 (1956-ZOE and BO1-ZOE) lentiviral overexpression or Zbtb46 shRNA construct expression (1956-ZKD and BO1-ZKD) in wild-type mice (n = 6). k-l, Representative images and quantification for (k) vascular perfusion (n = 7) and (l) pericyte coverage (n = 5) as measured by the FITC-lectin binding to vessels and NG2+ vascular area, respectively, in the 1956 sarcoma tissue of intra-tumor EV or ZOE lentivirus treated wild-type and Zbtb46 KO mice. m, Representative images with quantification of tumoral CD31+ vessels in wildtype mice bearing 1956 sarcoma co-transplanted with parental ECs (TC + WTEC) or Zbtb46-overexpressed ECs (TC + ZOEEC) (n = 6). Data are mean±SD. P-values were determined using One-way ANOVA with Dunnett’s test for a-c, e, f, h, and k and two-tailed Student’s t-test for d, g, i, j, l, and m. ns=not significant. Scale bars=100μm (f, g, and k-m).

Extended Data Fig. 3 |. Zbtb46 KO hematopoietic system is similar to wild-type in homeostasis but more immunosuppressive in the tumor and ZBTB46 sustains DC lineage generation while suppressing myeloid lineages in the tumor context.

a, 1969 regressive sarcoma growth in wild-type (n = 5), VEC-cre Zbtb46 KO (n = 5), CD11c-cre Zbtb46 KO (n = 6), and Zbtb46 KO (n = 4) mice. b, Hemavet and FACS analysis of PB, BM, and spleen in the wild-type and Zbtb46 KO mice in healthy condition (n = 3). c, MDSCs (CD11b + Gr1+) in the PB of the tumor-bearing mice in wild-type and Zbtb46 KO mice (n = 3). d, MDSC generation from wild-type and Zbtb46 KO mice BM (n = 3). Created with BioRender.com. e, Gr1+ and MHCII+ cell generation from BM-sorted CMPs and GMPs of wild-type or Zbtb46 KO mice (n = 4). f, Gr1+ and MHCII+ cell generation from wild-type BM-sorted KSL, CMPs, and GMPs with empty vector-mCherry (WT) or Zbtb46-mCherry (ZOE) overexpression (n = 3). g, Genomic snapshots depicting the ZBTB46 binding regions at the indicated genomic loci. h, Cebpb expression in BM cells of wild-type or Zbtb46 KO mice with empty vector (EV) or Zbtb46 (ZOE) lentiviral overexpression (n = 3). i, Zbtb46 and Cebpb expression in BM cells of wild-type or Zbtb46 KO mice with empty vector (EV), or Zbtb46 (ZOE), or Cebpb shRNA constructs (CKD), or Cebpb (COE) lentiviral overexpression (n = 3). j, Consensus motif for ZBTB46 and CEBPB ChIP sequences. k, Genomic snapshots depicting the CEBPB binding regions at the indicated genomic loci. l, Schematics of reporter lentivector core expression cassette for CEBP signaling pathway. Created with BioRender.com. m, Chemiluminescence measurement of SEAP activity in the reporter-only (control), or reporter with Cebpb overexpressed (COE), or reporter with Cebpb and Zbtb46 overexpressed (COE + ZOE) assay system (n = 4). n, Analysis of a few DC and macrophage signature genes in BM cells of 1956 sarcoma-bearing wild-type and Zbtb46 KO mice with empty vector (EV) or Zbtb46 (ZOE) lentiviral overexpression (n = 3). Data are mean±SD. P-values were determined using two-tailed Student’s t-test for b-f and One-way ANOVA with Dunnett’s test for h, i, m, and n. ns=not significant.

Extended Data Fig. 4 |. Lenti-Zbtb46 treatment remodels tumor infiltrating lymphocytes.

a, tSNE plot from merged scRNAseq data of exclusively intratumoral lymphocytes. b, tSNE plots of lymphocytes showing select marker-gene expression. c, Violin plots displaying expression levels of select genes in CD8T cell clusters. d, Percentages of cells in CD4 and T_act clusters across different treatment. e, Violin plots displaying expression levels of select genes in CD4T cell cluster between control and treatment group. f, Heatmap of GSEA identifying pathway enrichment by T cell clusters. g, Violin plots displaying expression levels of select genes in NK cell clusters. h, Percentages of cells in NK_s3–5 clusters across different treatment. i, Heatmap of GSEA identifying pathway enrichment by NK cell clusters.

Extended Data Fig. 5 |. Lenti-Zbtb46 treatment remodels intratumoral macrophage/monocytes.

a, Violin plots displaying Cx3cr1 and Nos2 expression levels in the macrophage/monocyte population. b, Heatmap of GSEA identifying pathway enrichment in the macrophage/monocyte population by treatment. c, tSNE plot from merged scRNAseq data of exclusively intratumoral macrophages/monocytes. d, Violin plots displaying expression levels of select genes in macrophage/monocyte clusters. e Percentages of cells in Mac_s2-s6 clusters across treatment. f, Heatmap of GSEA identifying pathway enrichment by macrophages/monocytes clusters. g, Violin plots displaying expression levels of select genes in Mac_s2 and Mac_s4 clusters between control and treatment group.

Extended Data Fig. 6 |. Lenti-Zbtb46 treatment remodels intratumoral dendritic cell population.

a, tSNE plot from merged scRNAseq data of exclusively intratumoral dendritic cells (DC). b, Heatmap of GSEA identifying pathway enrichment by DC clusters. Heatmap displaying normalized expression of select genes in each DC cluster. c, Percentages of cells in DC_s1-s4 clusters across different treatment. d, tSNE plots highlighting H2-Ab1, Itgax, and Xcr1 expressing cells in the macrophage/monocyte population. Heatmap of GSEA identifying pathway enrichment by DC clusters. e, Violin plots displaying expression levels of select genes in DC_s1-s3 clusters between control and treatment group.

Extended Data Fig. 7 |. Lenti-Zbtb46 treatment remodels intratumoral neutrophil cell population.

a, tSNE plot from merged scRNAseq data of exclusively intratumoral neutrophils. b, Percentages of cells in Neu_s1-s3 clusters across different treatment. c, Heatmap displaying normalized expression of select genes in each neutrophil cluster. d, Heatmap of GSEA identifying pathway enrichment by neutrophil clusters. e, Ratio of the frequencies of neutrophil clusters N_s2 over (N_s1+N_s3). f, Heatmap of GSEA identifying pathway enrichment in the total intratumoral neutrophil population by Zbtb46 treatment. g, Percentage of mast cells in the tumor microenvironment across different treatment. h, Heatmap of GSEA identifying pathway enrichment in the mast cell population by treatment.

Extended Data Fig. 8 |. Lenti-Zbtb46 treatment reshapes the landscape of intratumoral stromal cell population.

a, tSNE plot from merged scRNAseq data of exclusively intratumoral stromal cells. b, Heatmap displaying normalized expression of select genes in each stromal cluster. c, Heatmap of GSEA identifying pathway enrichment in the endothelial cell population (Strm_c1 cluster) by treatment. d, Heatmap of GSEA identifying pathway enrichment by remaining stromal clusters. e, Percentages of cells in different stromal clusters across treatment. f, Violin plots displaying expression levels of select genes in Strm_s2-s5 (fibroblasts) cell clusters between control and treatment groups.

Extended Data Fig. 9 |. Systemic Zbtb46 mRNA nanoparticle treatment restricts tumor growth and enhances outcomes of anti-PD1 immunotherapy.

a-f, Tumor growth kinetic (a, d; n = 7 for control and 8 for ZmR in a; n = 8/group in d), Zbtb46 expression (b, e; n = 3 for control and 4 for ZmR), and immune microenvironment (c, f; n = 3 for control and 4 for ZmR) of 1956 sarcoma (a-c) and PyMT-BO1 breast cancer (d-f) in wild-type mice with control (EGFP mRNA nanoparticle) and Zbtb46 mRNA nanoparticle (ZmR) treatment. g, Growth of 1956 sarcoma cells co-transplanted with parental endothelial cells (WTEC) or Zbtb46 overexpressed endothelial cells (ZOEEC) with IgG or anti-PD1 treatment (n = 6 for WTEC+IgG, 7 for WTEC+Anti-PD1, 6 for ZOEEC+IgG, and 8 for ZOEEC+Anti-PD1). CR= Complete Remission. h-j, Representative images and quantification for (h) tumoral CD31+ vessels (n = 15 for control, 10 for Anti-PD1, 11 for ZmR, and 12 for Anti-PD1+ZmR), (i) pericyte coverage as measured by the NG2+ vascular area (n = 5 for control and 4 for ZmR), and (j) Gr1+ cells (n = 3), respectively, in 1956 sarcoma of wild-type mice with Zbtb46 mRNA nanoparticle (ZmR) or control (EGFP mRNA nanoparticle + IgG) treatment. Data are mean±SD. P-values were determined using two-tailed Student’s t-test for a-f, i, and j and One-way ANOVA with Dunnett’s test for g and h. ns=not significant.

Extended Data Fig. 10 |. Tumor-derived factors suppress Zbtb46 expression.

a-b, Zbtb46 expression in (a) MCEC cells (n = 5 (left); n = 3 for PBS and H₂O₂ and 4 for the rest) and (b) BM-derived DC (BMDC) (n = 10 for PBS and PGE2 and 12 for VEGF and RA) with indicated treatments for 24 hours. c, Gr1+ and MHCII+ cell generation from wild-type BM cells with Control (PBS), or RA, or PGE2 as indicated (n = 3). Created with BioRender.com. d, Schematics depicting treatment plan for tumor-bearing mice. Created with BioRender.com. e-i, Tumor growth kinetic (e, h; n = 6 for PBS, 7 for BMS493, and 8 for NS398 and NAC-APO in e; n = 6 for PBS and BMS493 and 8 for NS398 and NAC-APO in h), Zbtb46 expression (f, i; n = 3 for tumor-EC and tumor-DC and 5 for total-BM in f; n = 3 in i), and FACS analysis of BM (g, n = 5) of 1956 sarcoma (e-g) and PyMT-BO1 breast cancer (h-i) in wild-type mice with BMS493 (inverse pan⁻retinoic acid receptor agonist), NS398 (selective cyclooxygenase-2 inhibitor), and NAC + APO (ROS scavengers) treatment. Data are mean±SD. P-values were determined using two-tailed Student’s t-test for a(left), e, f, h, and i and One-way ANOVA with Dunnett’s test for a(right)-c, f(right), and g. ns=not significant.

Supplementary Material

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41590–024-01936–4.

Data availability

Sequencing data are available at the Gene Expression Omnibus under accessions GSE226087 (bulk RNA-seq) and GSE264124 (scRNA-seq). Human bulk RNA tumor datasets are available from the TCGA database. Source data are provided with this paper.