Vascular abnormalization confers tumor resistance to antiangiogenic therapy via downregulating angiostatic factors

Abstract

Background

Angiogenesis is a hallmark of cancer. However, the efficacy of antiangiogenic therapy is hindered by drug resistance. The underlying mechanisms are complex and multifactorial, and the practical strategies to overcome resistance remain limited.

Methods

By establishing single cell clones (SCCs) from EO771 breast tumor cell line, we classified SCCs as resistant and sensitive to Anlotinib, a new antiangiogenic agent. Resistant and sensitive SCCs were then mixed in different ways to uncover novel mechanisms of antiangiogenic resistance by integrating orthotopic tumor models, RNA sequencing (RNA-seq), immunohistochemistry, blocking antibodies, and recombinant proteins.

Results

Increasing the diversity and quantity of sensitive SCCs overcame the influences of resistant SCCs and improved Anlotinib efficacy. Sensitive SCC tumors exhibited a more normalized vasculature and increased intratumoral CD8⁺ T cells compared to resistant SCC tumors. Moreover, elevated vessel perfusion was significantly correlated with better Anlotinib efficacy. RNA-seq data revealed upregulation of dual-function genes (e.g., Ifng, Cxcl9, and Vcam1) linking angiostasis and immune activation. Strikingly, the addition of an angiostatic factor (e.g., recombinant murine IFN-γ or Endostar) converted resistant breast tumors to sensitive to Anlotinib treatment and induced vascular normalization, whereas neutralizing IFNγ, a key angiostatic factor in sensitive SCC tumors, impeded their responses to Anlotinib.

Conclusions

The downregulation of angiostatic factors promotes breast tumor vascular abnormalization and resistance to antiangiogenic therapy. Therefore, integrating angiostatic factors with pro-angiogenic blockers may overcome resistance, offering a translatable approach to improve antiangiogenic outcomes in breast cancer.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13058-026-02229-4.

Introduction

Tumor angiogenesis is a hallmark of cancer, which facilitates tumor progression and metastasis [1, 2]. Angiogenesis is the formation of new blood vessels from the existing vasculature. Blood vessel formation is regulated by the interplay between pro-angiogenic factors, e.g., vascular endothelial growth factors (VEGFs), platelet-derived growth factors (PDGFs), fibroblast growth factors (FGFs), placental growth factor (PGF), and angiostatic factors, e.g., interferon γ (IFN-γ), C-X-C motif chemokine ligand 9 (CXCL9), tissue inhibitor of metalloproteinase 3 (TIMP3), thrombospondins (THBSs) [2–5]. Within the tumor parenchyma, the relentless proliferation of tumor cells creates a hypoxic tumor microenvironment (TME), which induces the secretion of numerous pro-angiogenic factors, including VEGFs, FGFs, and PGF, resulting in an imbalance between pro-angiogenic and angiostatic factors to promote tumor angiogenesis [3, 6]. Thereby, Bevacizumab, the first antibody to block VEGF/VEGFR signaling, has been approved to treat various solid cancers since 2004. However, the survival benefit is in the order of weeks and drug resistance develops rapidly [7, 8].

Since tumors usually upregulate multiple pro-angiogenic factors, the compensation of other pro-angiogenic factors could be a major mechanism of tumor resistance to Bevacizumab [9, 10]. Therefore, multi-targeted tyrosine kinase inhibitors (TKIs), such as Anlotinib, Apatinib, and Regorafenib, have been developed to simultaneously inhibit various pro-angiogenic signaling pathways [2, 3]. Unfortunately, the survival benefits of TKIs are still marginal [11–13]. Although several resistant mechanisms of antiangiogenic therapy have been reported, including the activation of compensatory pro-angiogenic signaling [14, 15], the recruitment of stromal and bone marrow-derived cells that promote angiogenesis [16, 17], and hypoxia-induced metabolic reprogramming [18], they are complex and multifactorial.

Drug resistance is one of the major challenges facing all kinds of cancer treatment modalities, and tumor cell heterogeneity (TCH) is one of the key resistance mechanisms [19, 20]. TCH is that a tumor is usually composed of different tumor cell clones with diverse genetic, epigenetic, and/or phenotypic characteristics [21–23]. TCH has been successfully used to elucidate the mechanisms of tumor resistance to the therapies targeting oncogenes, e.g., epidermal growth factor receptor (EGFR) and B-Raf proto-oncogene (BRAF) [24–26].

Here, we adapted TCH to investigate the resistance mechanisms of antiangiogenic therapy. Our data showed that resistant single cell clone (SCC) tumors exhibited a more abnormal tumor vasculature with fewer intratumoral CD8⁺ T cells compared to sensitive SCC tumors. Moreover, resistant SCC tumors upregulated the expression of pro-angiogenic factors, while downregulating the expression of angiostatic factors, compared to sensitive SCC tumors. Interestingly, increasing the diversity of sensitive SCCs promoted vascular normalization and reversed tumor resistance to Anlotinib treatment, which was primarily mediated by elevated angiostatic factors. These results suggest that vascular abnormalization mediates tumor resistance to antiangiogenic agents primarily via the imbalance between pro-angiogenic factors and angiostatic factors. Thus, pro-angiogenic blockers combined with angiostatic factors could overcome tumor resistance to antiangiogenic therapy.

Materials and methods

Animals

All mice in this study were maintained in a specific pathogen-free (SPF) animal facility at Soochow University. Female C57BL/6 mice (6–8 weeks old) were used (Shanghai Laboratory Animal Center; SLAC, China). All animal experiments in this study were conducted in accordance with the Animal Care and Use Regulations of China and were approved by the Institutional Animal Care and Use Committee of Soochow University.

Cell lines and reagents

Two murine mammary carcinoma cell lines, EO771 (obtained from CH3 Biosystems, Amherst, NY, USA) and AT3 (kindly provided by Dr. Changshun Shao, Soochow University), were used in this study. SCCs of EO771 were generated by limiting dilution in 96-well plates, ensuring each well contained no more than one single cell. Cells were maintained under continuous culture until stable SCC lines were established. All tumor cells were cultured at 37 °C in a humidified 5% CO₂ incubator. The culture medium consisted of DMEM (Gibco, USA) with 10% fetal bovine serum (FBS; Hyclone, USA) and 1% penicillin–streptomycin (Biosharp, China).

Anlotinib (provided by Jiangsu ChiaTai Tianqing Pharmaceutical Co. Ltd., Nanjing, China) was prepared as a 10 mg/ml stock solution in double-distilled water (ddH₂O). Endostar is a recombinant endostatin approved by the National Medical Products Administration (NMPA). Anti-CD8a monoclonal antibody (Clone 53-6.72), the corresponding IgG2a isotype control (clone 2A3), and an anti-mouse IFN-γ monoclonal antibody (Clone XMG1.2) were purchased from Bio X Cell (New Hampshire, USA). Recombinant murine IFN-γ (catalog 315-05) was obtained from PeproTech (Cranbury, NJ, USA).

Tumor growth and treatments

To establish orthotopic breast cancer models, EO771, SCCs, SCC mixtures (2 × 10⁵ cells), or AT3 (5 × 10⁵ cells) were inoculated into the mammary fat pads of mice. When tumors reached approximately 4 × 5 mm, mice were randomly assigned to receive control or Anlotinib treatment by daily oral gavage (EO771/SCC/SCC mixtures: 2 mg/kg; AT3: 4 mg/kg). Endostar (10 mg/kg) was administered daily by intraperitoneal injection. In some experiments, anti-mouse IFN-γ antibody (250 µg per mouse) was administered every 3 days via intraperitoneal injection, while in separate experiments, recombinant murine IFN-γ (3 µg per mouse) was administered via intratumoral injection at the same interval. Depletion of CD8⁺ T cells in vivo was performed as previously described [27]. Briefly, mice were intraperitoneally administered with an anti-CD8a monoclonal antibody or isotype-matched IgG2a (200 µg per mouse) on days 7, 9, and 14 following SCC tumor cell inoculation. Tumor growth was monitored every 3 days, and tumor volume was calculated as follows: (long axis × short axis² × π)/6. Tumors were surgically removed and weighed at the experimental endpoint.

Cell proliferation assays

M9, S2, Q5, M8, Q1, Q6, and EO771 cells (2 × 10⁴ cells) were plated into 96-well plates (n = 4 replicates per condition). Cell viability was determined by CCK-8 assay (Beyotime, Shanghai, China) at 24, 48, and 72 h, with absorbance measured at 450 nm.

Tumor vessel analysis

Tumor vessel perfusion and tumor vessel staining were evaluated based on established protocols [27, 28]. Functional vessels were labeled by intravenous administration of Ho33342 (10 mg/kg in PBS; Sigma-Aldrich, USA), followed by systemic perfusion with PBS (Biosharp, China) after 5 min of circulation. Excised tumors were subsequently fixed (4% paraformaldehyde, 2–3 h), cryoprotected (30% sucrose, overnight) and embedded in optimal cutting temperature (O.C.T.) compound. Tumor Sect. (20 μm) were incubated with an anti-CD31 primary antibody (Sigma-Aldrich, USA) overnight at 4 °C, followed by an Alexa Fluor 647-conjugated secondary antibody (Jackson ImmunoResearch, USA) for 2 h at room temperature. Nuclear counterstaining was achieved with Sytox Green (Molecular Probes, USA). All procedures were performed in the dark in humidified chambers. Confocal images were acquired on an Olympus FV3000 confocal microscope and quantified using Image-Pro Plus and ImageJ software.

Flow cytometric analysis

Flow cytometric analysis was conducted according to previously published methods [27]. Briefly, tumor tissues were enzymatically dissociated at 37 °C for 45 min using a mixture of hyaluronidase, collagenase type IA, and DNase I in DMEM. After filtration through 70-µm strainers, single-cell suspensions were resuspended in cold flow buffer (PBS with 1% bovine serum albumin and 0.1% NaN₃) and blocked with a rat anti-mouse CD16/CD32 antibody (BD Pharmingen, USA). Cells were subsequently stained with different fluorochrome-conjugated anti-mouse antibodies: CD45-BV421, CD11b-BV510, CD8a-FITC, CD4-PE, CD25-APC, F4/80-FITC, Gr1-APC-Cy7, CD206-APC, CD11c-PE-Cy7 (all from BioLegend, USA). Nonviable cells were excluded using 7AAD (eBioscience, USA). All samples were run on a Gallios flow cytometer (Beckman, USA) and analyzed using Kaluza software.

RNA sequencing

Transcriptome profiling was performed on SCC cell lines and 4–6 mm tumors of different SCCs by high-throughput RNA sequencing. Library construction and RNA sequencing were performed by LC-Bio Technology Co., Ltd. on an Illumina Novaseq 6000 platform. Gene expression levels were estimated and normalized using the fragments per kilobase of transcript per million mapped reads (FPKM). P < 0.05 was used to select genes with significant changes in expression. All sequencing data were processed and analyzed with OmicStudio tools.

Quantitative real-time PCR (qPCR)

Total RNA was extracted from mixed SCC tumors using a FreeZol Reagent (Vazyme, R711-02, China). The extracted RNA was then converted into complementary DNA (cDNA) in a reverse transcription reaction employing HiScript III RT SuperMix (Vazyme, R323-02, China). qPCR was conducted with ChamQ SYBR Color qPCR Master Mix (Vazyme, Q421-02, China) on a LightCycler 480 system (Roche, Switzerland). The fold changes of gene expression were calculated by the comparative threshold cycle method, with β-actin serving as the normalization control.

Statistical analysis

All quantitative data were analyzed using GraphPad Prism 10. We employed unpaired two-tailed Student’s t-tests for comparisons between two groups and one-way ANOVA for comparisons across multiple groups. Two-way ANOVA was used when two categorical independent variables were involved. Correlations were assessed by Pearson correlation analysis and both correlation coefficients (r) and P values were provided. All data are presented as the mean ± standard error of the mean (SEM) and P < 0.05 was considered statistically significant. mTGI = (mean tumor volume of control group - mean tumor volume of Anlotinib treatment group)/mean tumor volume of control group×100.

Results

SCC tumors resistant to Anlotinib contain fewer CD8⁺ T cells and possess a more abnormal tumor vasculature

To adapt TCH to investigate the resistance mechanisms of antiangiogenic therapy, we established a library of SCCs from EO771 breast tumor cell line using limited dilution. SCCs or parental EO771 tumor cells were then orthotopically inoculated into female mice. When tumors reached 4–5 mm in diameter, mice were treated with 2 mg/kg of Anlotinib daily for 9 days according to a previous publication (Fig. 1A) [27]. SCC tumors displayed various sensitivities to Anlotinib (Fig. 1B, C). To accurately determine the sensitivity of different SCC tumors to Anlotinib, mean tumor growth inhibition (mTGI) of tumor volume was calculated. Since parental EO771 showed an intermediate response to Anlotinib with an mTGI of 48.15% (Supplementary Fig. S1A), SCC tumors with mTGI ≥ 48% were classified as sensitive SCCs, including M9, S2, and Q5, while SCC tumors with mTGI < 48% were classified as resistant SCCs, including M8, Q1, and Q6 (Fig. 1D).

Fig. 1.

The SCC tumors exhibited heterogeneous responses to Anlotinib.

SCC breast tumor cells (M9, S2, Q5, M8, Q1, Q6) (2 × 10⁵ cells) were orthotopically inoculated into the mammary fat pad of female C57BL/6 mice. When tumors reached 4–5 mm in diameter, mice were randomly grouped and administered with ddH₂O or Anlotinib (2 mg/kg) daily by oral gavage. Tumor size was recorded every 3 days. The tumor volume was estimated by the formula = [(long axis) × (short axis)² × π/6]. Tumor weight was measured at the end of the experiments. A Experimental design to establish SCCs from the EO771 breast tumor cell line. B and C Tumor growth curves and tumor weight of SCC tumors. Significance was determined by two-way ANOVA and unpaired two-tailed Student’s t-tests. All data are presented as means ± SEM. NS, no significant difference; **P < 0.01; ***P < 0.001. D mTGI of Anlotinib in different SCC tumors. mTGI = (mean tumor volume of control group - mean tumor volume of Anlotinib treatment group)/mean tumor volume of control group×100. The tumor volume on Day 9 after vehicle control or Anlotinib treatments was used. Data are from one experiment representative of two independent experiments with similar results (n = 10–14 tumors per group)

To investigate the mechanisms underlying the heterogeneous responses of SCC tumors to Anlotinib, we first evaluated the proliferation rates of different SCCs and EO771 in vitro and in vivo and there were no significant differences among the parental EO771, sensitive SCCs, and resistant SCCs (Supplementary Fig. S1B, C).

Next, we characterized the features of tumor vascular in EO771 and SCC tumors. We assessed vessel density by CD31 staining and vascular perfusion by intravenous injection of Hoechst 33,342 (Ho33342) dye 5 min prior to tumor tissue harvest to label functional blood vessels and perfused area. By comparing the baseline vascular density and perfusion among parental EO771, sensitive SCCs, and resistant SCCs, the vascular features of parental EO771 were more similar to those of resistant SCCs, whereas sensitive SCCs exhibited markedly enhanced vessel perfusion (Supplementary Fig. S1D). The data show that parental EO771 and resistant SCCs possess abnormal tumor vasculature, while sensitive SCCs possess relatively normalized tumor vasculature.

Following Anlotinib treatment, tumor vessel area was proportionally reduced in all SCC tumors (Fig. 2A–C). However, Anlotinib treatments significantly increased the ratios of perfused tumor vessels in sensitive SCC tumors, which was not observed in resistant SCC tumors (Fig. 2D). Moreover, the individual TGI values were positively and significantly correlated with the ratios of perfused tumor vessels after Anlotinib treatment (Fig. 2E).

Fig. 2.

Improved tumor vessel perfusion was positively associated with the responses of SCC tumors to Anlotinib.

SCC tumor-bearing mice were prepared and treated as described in Fig. 1. Mice were intravenously injected with 200 µg/mouse Ho33342 five minutes prior to tumor tissue harvested. A, B The representative images of Ho33342 perfusion (blue) and CD31-staining (red). Scale bar = 100 μm. C, D The CD31⁺ areas and Ho33342⁺ areas were analyzed by ImageJ. Ho33342⁺ area reflects the perfused function of tumor vessels and CD31⁺ area represents tumor vessel density. Ho33342⁺ area/CD31⁺ area shows the ratios of perfused tumor vessels. Unit is pixel. The significance was determined by two-way ANOVA. The data are presented as means ± SEM. NS, no significant difference; **P < 0.05; ***P < 0.001. E The association between the values of individual TGI and the ratios of perfused tumor vessels was analyzed by Pearson Correlation Analysis (r: 0.75, ***P < 0.001). Individual TGI = (mean tumor volume of the control group – tumor volume of each tumor in the Anlotinib treatment group) / mean tumor volume of the control group×100. Data are from one experiment representative of two independent experiments with similar results (n = 5–6 tumors per group)

We further analyzed the proportions of intratumoral immune cells in SCC tumors because they usually produce various angiogenesis-related factors [25–28]. Sensitive SCC tumors had higher proportions of CD8⁺ T cells and lower proportions of Tregs, and Anlotinib treatment increased the proportions of CD4⁺ and CD8⁺ T cells compared to resistant SCC tumors (Supplementary Figs. S2 and S3). Moreover, the proportions of myeloid cell populations, such as tumor-associated macrophages (TAMs), neutrophils, and monocytes, showed no consistent differences between sensitive and resistant SCC tumors (Supplementary Fig. S4).

Moreover, the depletion of CD8⁺ T cells abrogated the anti-tumor effects of Anlotinib in sensitive tumors (M9, S2). In contrast, CD8⁺ T cell depletion had no impact on tumor growth in resistant SCC tumors (M8, Q6), which remained unresponsive to Anlotinib treatment (Supplementary Fig. S5). The data suggest the crucial role of CD8⁺ T cells in the response of sensitive SCCs to Anlotinib.

Taken together, the data suggest that resistant SCC tumors contain fewer CD8⁺ T cells and possess a more abnormal tumor vasculature. This is consistent with our previous report that the accumulation and activation of CD8⁺ T cells promotes tumor vascular normalization [28, 29].

Increasing the diversity and quantity of sensitive SCCs improves Anlotinib efficacy

Next, SCCs were mixed in different ways to mimic tumors with different sensitivities to Anlotinib (Fig. 3A and Supplementary Fig. S6A). Tumors derived from mixing resistant (R) SCCs in equal proportions, either 2R tumors (M8 + Q6, M8 + Q1) or 3R tumors (M8 + Q1 + Q6), were still resistant to Anlotinib treatment, and their mTGI values were comparable to those of resistant SCC tumors (Fig. 3 and Supplementary Fig. S6). Similarly, tumors derived from mixing sensitive (S) SCCs in equal proportions, either 2S tumors (M9 + S2, S2 + Q5) or 3S tumors (M9 + S2 + Q5), were still sensitive to Anlotinib treatment (Fig. 3 and Supplementary Fig. S6).

Fig. 3.

Increasing the diversity and quantity of sensitive SCCs improved Anlotinib efficacy.

Female C57BL/6 mice were orthotopically inoculated with different SCC mixtures 2 × 10⁵ cells). When tumors reached 4–5 mm in diameter, mice were randomly grouped and administered with ddH₂O or Anlotinib (2 mg/kg) daily by oral gavage. Tumor size was recorded and analyzed as described in Fig. 1. A The schematic of different SCC mixtures to establish tumors with differential tumor cell heterogeneity. B, C Tumor growth curves of 3R, 1R1S, 2R2S, 3R3S (1:1), 3S, 3R3S (3:1) and 3R3S (1:3). D mTGI of Anlotinib in differentially mixed SCC tumors. mTGI was calculated as described in Fig. 1. Significance was determined by two-way ANOVA. Data are presented as mean ± SEM. NS, no significant difference; **P< 0.01; *** P< 0.001. Data are from one experiment representative of three independent experiments withsimilar results (in B, n = 8–12 tumors per group)

The resistant and sensitive SCCs were further mixed in equal or unequal proportions. The mTGI values of Anlotinib in 1R1S tumors M8 + M9, Q1 + S2, and Q6 + S2 were 34.79%, 28.99%, and 29.72%, respectively, which were similar to those of resistant SCCs (Fig. 3B, D and Supplementary Fig. S6C, D). The mTGI values in 2R2S tumors (M8 + Q1 + S2 + M9, M8 + Q6 + S2 + M9) and 3R3S (1:1) tumors were 46.89%, 52.12%, and 55.59%, respectively, which were higher than those of resistant SCCs and gradually closer to those of sensitive SCCs (Fig. 3B–D). Moreover, changing the ratios between resistant and sensitive SCCs, the mTGI values in 3R3S (3:1) and 3R3S (1:3) tumors were 31.28% and 60.64%, respectively, which were similar to those of resistant and sensitive SCCs, respectively (Fig. 3C, D). Together, the data show that increasing the diversity and quantity of sensitive SCCs improves Anlotinib efficacy.

Increasing the diversity and quantity of sensitive SCCs promotes tumor vascular normalization with elevated intratumoral CD8⁺ T cells

Previous data suggest the association between Anlotinib sensitivity, vessel perfusion, and intratumoral CD8⁺ T cells in SCC tumors. Therefore, we also analyzed the vasculature in the mixed tumors. In the vehicle control group, the tumor vessel area gradually decreased from 3R, 1R1S, 2R2S, 3R3S (1:1), 3R3S (1:3) to 3S mixed tumors, while the levels of tumor vessel perfusion were significantly higher in the 2R2S, 3R3S (1:1), 3R3S (1:3), and 3S mixed tumors than in the 3R and 1R1S mixed tumors (Fig. 4A, B). The data suggest that increasing the diversity of sensitive SCCs promotes vascular normalization in mixed tumors. Moreover, Anlotinib treatment continuously increased the ratios of perfused tumor vessels from 3R, 1R1S, 2R2S, 3R3S (1:1), 3R3S (1:3) to 3S mixed tumors, with significant differences in the mixed tumors of 2R2S, 3R3S (1:1), 3R3S (1:3), 3S and 3R, 1R1S compared to their vehicle controls (Fig. 4A–C). In addition, the proportions of intratumoral CD8⁺ T cells were continuously increased from 3R, 1R1S, 2R2S, 3R3S (1:1), 2S to 3S mixed tumors (Supplementary Fig. S7). These findings indicate that increasing the diversity of sensitive SCCs facilitates Anlotinib to induce vascular normalization with elevated intratumoral CD8⁺ T cells in the mixed tumors.

Fig. 4.

Increasing the diversity and quantity of sensitive SCCs promoted tumor vascular normalization.

Mixed SCC tumor-bearing mice were prepared, and Anlotinib treatments (2 mg/kg/daily) were conducted as described in Fig. 3. Tumor tissues were harvested on Days 9 after vehicle control or Anlotinib treatments. Five minutes prior to tumor harvest, mice were intravenously injected with 200 μg/mouse Ho33342. A The representative images of Ho33342 perfusion (blue) and CD31-staining (red) of tumors derived from different SCC mixtures. Scale bar = 100 μm. B, C The CD31⁺ area and Ho33342⁺ area. The ratios of perfused tumor vessels were calculated by Ho33342⁺ area/CD31⁺ area. Unit is pixel. The significance was determined by two-way ANOVA. The data are presented as means ± SEM. NS, no significant difference; **P < 0.01; ***P < 0.001. Data are from one experiment representative of two independent experiments with similar results (n = 6–8 tumors per group)

Sensitive and resistant SCC tumors express higher levels of angiostatic and pro-angiogenic genes, respectively

The above data show that resistant SCC tumors have a more abnormal tumor vasculature, and that increasing the diversity of sensitive SCCs can reverse tumor resistance to Anlotinib and facilitate tumor vascular normalization. To unveil the underlying mechanisms, resistant and sensitive SCC tumor tissues were collected for RNA sequencing (RNA-seq).

Several studies have defined angiogenesis-related genes associated with good or poor prognosis, named as GPAGs and PPAGs, respectively [30, 31]. PPAGs generally promote tumor vascular abnormalization, while GPAGs may induce tumor vascular normalization. Gene Set Variation Analysis (GSVA) also revealed that the PPAG-related pathways were predominantly enriched in the resistant SCC tumors (Fig. 5A and Supplementary Fig. S8A). In contrast, the GPAG-related pathways were predominantly enriched in the sensitive SCC tumors (Fig. 5B and Supplementary Fig. S8B). Moreover, Gene Set Enrichment Analysis (GSEA) revealed the enrichment of pro-angiogenic and hypoxic signaling pathways in the resistant SCC tumors, including genes related to growth factors and their receptors (e.g., Fgf23, Flt1, Pgf, Fgf7, Ptgs1, Ereg, Hbegf, Pdgfra), cytokines (e.g., Cxcl1, Cxcl2, Cxcl3, Cxcl5, Il6), extracellular matrix (e.g., Bmp7, Mmp2, Mmp19, Col2a1), and hypoxia (e.g., Hif1a, Hif3a) (Fig. 5C, E and Supplementary Fig. S8C, D). Conversely, GSEA revealed the enrichment of angiostatic signaling in the sensitive SCC tumors, including Cxcl9, Ifng, Ifngr2, Ifnar1, Thbs3, Thbs4, Cxcr3, Dll1, Icam1, Chga, and Timp3 (Fig. 5D, F and Supplementary Fig. S8D).

Fig. 5.

Resistant and sensitive SCC tumors expressed higher levels of pro-angiogenic and angiostatic genes, respectively.

Female C57BL/6 mice were orthotopically inoculated with 2 × 10⁵ sensitive SCCs (M9, S2, Q5) and resistant SCCs (M8, Q1, Q6). When tumors reached 4–6 mm in diameter, tumor tissues were harvested for RNA sequencing (n = 3 tumors per SCC). A, B: GO terms that associated with GPAGs and PPAGs. C, D: Pathways associated with pro-angiogenic and angiostatic factors analyzed by GSEA. E, F: Expression of selected genes of pro-angiogenic and angiostatic signalings

Given that recent studies suggest the crosstalk between tumor vessels and immune cells [28–30], we also analyzed immune-related pathways in SCC tumors. The RNA-seq data analysis revealed that T cell receptor (TCR) signaling, antigen processing and presentation, and immune activation pathways were enriched in sensitive SCC tumors compared to resistant SCC tumors (Supplementary Fig. S9A–C). Moreover, Venn diagram of angiostatic signaling and immune activation signaling showed that Ifng, Cxcl9, and Vcam1 were three commonly upregulated genes in sensitive SCC tumors, compared with resistant SCC tumors (Supplementary Fig. S9D). The three genes are critical mediators of the recruitment and activation of CD8⁺ T cells. This is consistent with the data in Supplementary Fig. S3 showing that sensitive tumors contain higher levels of CD8⁺ T cells. However, the analysis of RNA-seq of SCC cells revealed that angiostatic signaling, rather than immune-related pathways, was enriched in sensitive SCC cells, compared to resistant SCC cells (Supplementary Fig. S10).

Collectively, the above data show that resistant SCC tumors upregulate the expression of pro-angiogenic factors, while sensitive SCC tumors upregulate the expression of genes with dual angiostasis and immune activation, such as Ifng, Cxcl9, and Vcam1. However, sensitive SCC cells upregulated angiostasis-related genes, but not immune activation-related genes. In line with the profiles of angiogenic-related gene expression, the tumor vasculature was more abnormal in resistant SCC tumors than in sensitive SCC tumors. These results indicate that tumor cell-intrinsic angiostatic program dictates the characteristics of the tumor vasculature and the sensitivity of Anlotinib at least partially via regulating the recruitment and activation of CD8⁺ T cells.

Increasing the diversity of sensitive SCCs elevates angiostatic gene transcription while decreasing pro-angiogenic gene transcription in the mixed tumors

Next, we analyzed the baseline levels of angiogenesis-related genes in the mixed tumors with differential diversity of SCCs prior to Anlotinib treatment. From the 3R, 1R1S, 2R2S, 3R3S (1:1), 3R3S (1:3) to 3S mixed tumors, the levels of angiostatic genes, e.g., Ifng, Cxcl9, Tnfa, Icam1, Chga, Cxcr3, Dll1, Notch1, were continuously increased, while the levels of pro-angiogenic factors, e.g., Pgf, Vegfa, Flt1, Hif1a, Ang2 and Tgfbr3, were gradually decreased (Fig. 6A, B). We also examined angiogenic-related genes in four representative mixed SCC tumors (3R, 1R1S, 3R3S (1:1), and 3S) after vehicle control or Anlotinib treatments. In the control group, the transcription of most angiostatic genes was increased while the transcription of most pro-angiogenic genes was decreased with increasing the diversity of sensitive SCCs (Supplementary Fig. S11). Anlotinib treatments upregulated angiostatic genes, such as Ifng, Cxcl9, and Cxcr3 in the sensitive mixed tumors (3R3S (1:1), 3S), while suppressed pro-angiogenic gene transcription primarily in resistant mixed tumors (3R, 1R1S) (Supplementary Fig. S11). Taken together, the data show that increasing the diversity of sensitive SCCs elevates the transcription of angiostatic genes while decreasing the transcription of pro-angiogenic genes in the mixed tumors. This finding is consistent with the data in Fig. 4, showing that increasing the diversity of sensitive SCCs in the mixed tumors promotes tumor vascular normalization.

Fig. 6.

Increasing the diversity and quantity of sensitive SCCs upregulated angiostatic genes while downregulating pro-angiogenic genes in the mixed SCC tumors.

Different mixed tumors (3R, 1R1S, 2R2S, 3R3S (1:1), 3R3S (1:3) and 3S) were prepared and harvested at a size of 4–5 mm. Changes in gene transcription were determined by qPCR. A Transcription of Ifng, Tnfa, Cxcl9, Icam1, Chga, Cxcr3, Dll1, and Notch1 was analyzed by qPCR. B Transcription of Pgf, Vegfa, Flt1, Ang2, Hif1a, and Tgfbr3 was analyzed by qPCR. The significance was determined by one-way ANOVA (n = 6–8 tumors per group). The data are presented as means ± SEM. NS, no significant difference; *P < 0.05; **P < 0.01; ***P < 0.001

The addition of an angiostatic factor converts tumors from resistant to sensitive to Anlotinib therapy and facilitates tumor vascular normalization

To test whether elevated angiostatic factors could sensitize tumors to Anlotinib, we first investigated the roles of IFN-γ in Anlotinib therapy because IFN-γ serves as both an immune-activating and angiostatic factor (Fig. 5F and Supplementary Figure S9D). Blockade of IFN-γ impeded the responses of sensitive tumors (3R3S (1:1), 3S) to Anlotinib therapy, without changing the efficacy of Anlotinib in resistant tumors (3R, 1R1S) (Supplementary Fig. S12). Conversely, intratumoral injection of recombinant murine IFN-γ improved the Anlotinib efficacy in resistant SCC tumors (Q6), whereas it had no significant effect on sensitive SCC tumors (S2) (Supplementary Fig. S13A). Mechanistically, the combination of recombinant IFN-γ and Anlotinib reduced vessel area, enhanced vessel perfusion, and increased the ratios of perfused vessels compare to Anlotinib or recombinant IFN-γ alone in Q6 tumors, while only a modest improvement in the ratios of perfused vessels was seen in S2 tumors (Supplementary Fig. S13B). The data show that IFN-γ contributes to the efficacy of Anlotinib.

We also added Endostar to Anlotinib therapy, because Endostar is a recombinant endostatin and has been approved by the National Medical Products Administration (NMPA) in China for the treatment of non-small-cell lung cancer [32]. Indeed, the combination of Endostar and Anlotinib significantly inhibited the growth of resistant tumors (3R and 1R1S), while Anlotinib or Endostar alone showed marginal tumor growth inhibition (Fig. 7A and B). However, in the sensitive 3S and 3R3S (1:1) tumors, the addition of Endostar did not improve Anlotinib efficacy (Fig. 7A, B). We further analyzed the impacts of Endostar and Anlotinib combination therapy on the tumor vasculature. The combination therapy had no significant effect on tumor blood vessels in sensitive tumors (3S) compared to Anlotinib alone (Fig. 7C, D). However, in resistant tumors (1R1S), the combination therapy further reduced tumor vessel density and increased vessel perfusion and the ratios of perfused vessels compared to Anlotinib or Endostar alone (Fig. 7C, D). The data suggest that the addition of an angiostatic factor can reverse the resistance to Anlotinib therapy and facilitate tumor vascular normalization.

Fig. 7.

The addition of Endostar, an angiostatic factor, converted tumors from resistant to sensitive to Anlotinib therapy and facilitated tumor vascular normalization.

Female C57BL/6 mice were orthotopically inoculated with 2 × 10⁵ SCC mixtures (3S, 3R3S (1:1), 3R and 1R1S (S2 + Q6)). When tumors reached 4–5 mm in diameter, mice were randomly assigned to four groups and then treated with vehicle control, Anlotinib (2 mg/kg), Endostar (10 mg/kg) or Anlotinib + Endostar. A, B Tumor growth curves of 3R, 1R1S, 3R3S (1:1) and 3S tumors, and the mTGI values of Anlotinib and Anlotinib + Endostar in the mixed tumors. C The representative images of Ho33342 perfusion (blue) and CD31-staining (red) of different mixed tumors. Scale bar = 100 μm. D The CD31⁺ area and Ho33342⁺ area. The ratios of perfused tumor vessels were calculated by Ho33342⁺ area/CD31⁺ area. Significance was determined by two-way ANOVA (A) and one-way ANOVA (D) (n = 6–8 tumors per group). Data are presented as mean ± SEM. NS, no significant difference; *P < 0.05; **P < 0.01; ***P < 0.001

Moreover, we conducted long-term Anlotinib and Endostar treatment in representative SCC and mixed SCC tumor models to recapitulate the antiangiogenic regimens in the clinic. Prolonged combination therapy resulted in durable and stable tumor growth suppression in resistant tumors (Q6, S2 + Q6) compared to Anlotinib alone, while did not improve Anlotinib efficacy in sensitive tumors (S2, 3S) (Supplementary Fig. S14). These results suggest that the addition of an angiostatic factor not only reverses resistance but also stabilizes long-term responsiveness to Anlotinib.

Finally, we validated the above findings in two parental breast tumor models. Both parental EO771 and AT3 tumors exhibited moderate sensitivity to Anlotinib (mTGI = 46.23% and 47.47%, respectively). Notably, Endostar enhanced the anti-tumor efficacy of Anlotinib in both tumor models (mTGI = 66.05% and 80.95%, respectively) (Supplementary Fig. S15). Taken together, the data show that tumor resistance to Anlotinib is due to lower levels of angiostatic factors, and that the integration of angiostatic factors with pro-angiogenic blockers improve antiangiogenic efficacy.

Discussion

Although antiangiogenic therapy has been applied in the clinic for approximately twenty years, drug resistance remains a major challenge. Here, we adapted TCH, a well-known mechanism that mediates tumor resistance to therapy [19, 24], to explore the mechanisms of antiangiogenic resistance. We found that sensitive SCC tumors contained a more normalized vasculature and higher levels of CD8⁺ T cells compared to resistant SCC tumors. Moreover, the levels of vessel perfusion were positively correlated with the sensitivity of SCC tumors to Anlotinib. Mechanistically, RNA-seq analysis showed that sensitive SCC tumors upregulated genes with dual angiostasis and immune activation, especially Ifng, Cxcl9, and Vcam1. In sensitive SCC tumor cells, however, only angiostatic genes were significantly enriched compared to resistant ones. Taken together, these data suggest that cancer-cell-intrinsic angiostasis dysregulation shapes the characteristics of the tumor vasculature and antiangiogenic sensitivity via modulating the recruitment and activation of CD8⁺ T cells at least partially. In line with this novel mechanism, neutralizing IFN-γ hindered the responses of sensitive tumors to Anlotinib therapy, while the addition of IFN-γ or Endostar, converted tumors from resistant to sensitive to Anlotinib and induced tumor vascular normalization, underscoring the indispensable role of angiostatic signaling in Anlotinib sensitivity. Therefore, the integration of angiostatic factors with simultaneously blocking pro-angiogenic factors could overcome tumor resistance to conventional antiangiogenic agents.

In the clinic, there is still no feasible strategy to overcome antiangiogenic resistance [3, 10]. Current approaches to combat drug resistance of antiangiogenic therapy are usually to suppress multiple pro-angiogenic signaling pathways concomitantly [10, 33, 34]. However, these types of treatments usually create a hypoxic TME which triggers the secretion of more pro-angiogenic factors, resulting in rapid drug resistance [35–37]. By investigating the impacts of TCH on antiangiogenic therapy, we found that increasing the diversity of sensitive SCCs in the mixed tumors overcame the effects of resistant SCCs and improved the efficacy of Anlotinib treatment, which was associated with increased expression of angiostatic factors. Since tumors usually persistently produce pro-angiogenic factors with few angiostatic factors, the combination of angiostatic factors with pro-angiogenic blockers would be more feasible to restore the imbalance between pro-angiogenic and angiostatic factors and therefore overcome antiangiogenic resistance.

TCH is a major resistance mechanism of cancer therapies [19]. In this study, we proposed that TCH could mediate antiangiogenic resistance via inducing tumor vascular abnormalization. In the TME, multiple components could regulate tumor angiogenesis. First, the myeloid lineage plays dynamic and diverse roles in tumor angiogenesis [38–41]. Although our data showed that the overall proportions of monocytes, neutrophils and TAMs were comparable between sensitive and resistant SCC tumors, the subtypes of myeloid cells and their function status may contribute to antiangiogenic resistance. For example, the accumulation of Ly6C^lo monocytes or Ly6G⁺ neutrophils within tumors may hinder anti-VEGFR2 therapy via the secretion of CX3CL1 [42]. Targeting myeloid subsets, such as CSF1R⁺ or CD244⁺ monocytes/macrophages, could improve antitumor effects in combination with antiangiogenic therapy or immunotherapy [41, 43, 44]. Second, lymphoid effectors may impede tumor angiogenesis to overcome antiangiogenic resistance. The baseline levels of intratumoral CD8⁺ T cells in sensitive SCC tumors were higher than that of resistant SCC tumors and Anlotinib treatments further elevated them. Furthermore, the depletion of CD8⁺ T cells partially abolished the antitumor efficacy of Anlotinib in the sensitive SCC tumors. Consistently, Ifng, Cxcl9, and Vcam1 were three commonly upregulated genes between angiostatic signaling and immune activation signaling in sensitive SCC tumors, compared with resistant SCC tumors. The three genes are critical mediators of the recruitment and activation of CD8⁺ T cells. Third, tumor cells can produce diverse factors to modulate the tumor vasculature. The analysis of RNA-seq of SCC tumor tissues showed that immune activation and angiostatic signaling were enriched in sensitive SCC tumors compared to resistant SCC tumors. However, the analysis of RNA-seq of SCC cell lines revealed that angiostatic signaling, rather than immunostimulatory-related pathways, was enriched in sensitive SCC cells, compared to resistant SCC cells. Given the facts that different SCC tumors have distinct tumor vasculature and immune microenvironment, and that different SCCs were derived from a same tumor cell line and were inoculated in the same inbred mice, we concluded that the distinct characteristics of tumor vasculature and immune microenvironment are dictated by the intrinsic properties of SCC tumor cells. In addition, our previous studies have shown that the accumulation and activation of CD8⁺ T cells promotes tumor vascular normalization and improves immunotherapy [28, 29]. Therefore, these results suggest that tumor cell-intrinsic angiostatic program dictates the characteristics of the tumor vasculature and the sensitivity of Anlotinib at least partially via modulating the recruitment and activation of CD8⁺ T cells.

It is well-known that higher tumor cell heterogeneity is associated with poor prognosis in oncogene targeted therapy [24, 45]. However, our study showed that increasing the diversity of sensitive SCCs could reverse the effects of resistant SCCs and improve Anlotinib efficacy. RNA-seq revealed that sensitive SCCs expressed heterogeneous angiostatic genes. For example, except for IFN-γ signaling pathway-related genes, M9 highly expressed Cxcr3 and Dll1, S2 highly expressed Vcam1 and Thbs3, and Q5 highly expressed Notch1. Conversely, 3S mixed tumors highly expressed Ifng, Cxcl9, Ifngr2, Cxcr3, Dll1, Vcam1, Thbs3, and Notch1 simultaneously. The expression levels of these angiostatic genes in 3R3S (1:1) and 3R3S (1:3) mixed tumors were closer to those in 3S mixed tumors. The activation of DLL1/Notch signaling can induce tumor vascular normalization [46], and CXCL9 also can directly inhibit angiogenesis by binding to CXCR3 on endothelial cells [46, 47]. This could be a reason that anti-IFN-γ cannot completely convert sensitive mixed tumors to resistant to Anlotinib. Thus, the diversity and quantity of sensitive SCCs could increase the diversity and quantity of angiostatic factors that facilitate Anlotinib to induce tumor vascular normalization and improve therapeutic outcomes.

Due to the high heterogeneity of tumors in the clinic, biomarkers are a powerful approach to stratify patients for precision cancer treatments and to predict treatment responses. For example, in oncogene targeted therapy, molecular profiling is routinely used to guide the selection of targets for patients with a range of cancer types [48–50]. However, the lack of appropriate predictive biomarkers limits the clinical application of antiangiogenic therapy. Our study identified SCCs with sensitive or resistant to Anlotinib. However, the levels of Anlotinib targets were not determinants of Anlotinib sensitivity. Interestingly, the efficacy of Anlotinib was positively correlated with the levels of vessel perfusion in SCC tumors, suggesting that tumor vessel perfusion may dictate the efficacy of Anlotinib treatment. Our recent reports show that tumor vessel perfusion can predict tumor response to immune checkpoint blockade therapy [28, 51]. Thus, radiological approaches, such as contrast-enhanced CT and MRI, to monitor and evaluate tumor vessel perfusion may guide the development of precision antiangiogenic therapy.

Our study also has certain limitations. The mixed-tumor models were composed of up to six SCCs, representing a restricted level of intratumoral diversity. It remains unclear whether increasing clonal diversity beyond this (e.g., 4R4S or 8R8S) would further enhance sensitivity to Anlotinib or instead reach a plateau. Future work needs to further dissect the effects of TCH on the acquired resistant of antiangiogenic therapy.

Conclusion

In summary, our study employed tumor cell heterogeneity to investigate the mechanisms of antiangiogenic resistance. Sensitive SCC tumors contained a more normalized vasculature with higher levels of intratumoral CD8⁺ T cells compared to resistant SCC tumors. Moreover, the levels of vessel perfusion were positively correlated with Anlotinib efficacy in SCC tumors. Mechanistically, sensitive SCC tumors upregulated genes with dual angiostasis and immune activation, e.g. Ifng, Cxcl9, and Vcam1. Notably, the addition of angiostatic factor converted tumors from resistant to sensitive to Anlotinib treatment, while blocking angiostatic factors impeded the responses of sensitive tumors to Anlotinib. Taken together, the data suggest that vascular abnormalization confers tumor resistance to antiangiogenic therapy via downregulating angiostatic factors, partially through suppressing the recruitment and activation of CD8⁺ T cells. Therefore, our findings may offer a feasible strategy to overcome antiangiogenic resistance by integrating angiostatic factors with pro-angiogenic blockers.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

X.M., H.Z., and Y.H. designed the research; X.M., X.C., L.Q., M.G., N.Z., P.F., Z.Q., and T.L. performed experiments and analyzed the data; X.M., X.C., H.Z., and Y.H. interpreted the results and wrote the paper with inputs from all the other authors. All authors reviewed the manuscript.

Funding

This work was supported by grants from the National Natural Science Foundation of China (82150106 to YH, 82473305 to YH, 81972877 to YH), Provincial-level Talent Program for National Center of Technology Innovation for Biopharmaceuticals (NCTIB2024JS0101 to YH), Key Program for Basic Research in Jiangsu Province (BK20243008), Interdisciplinary Basic Frontier Innovation Program of Suzhou Medical College of Soochow University (YXY2303028 to YH), the Collaborative Innovation Center of Hematology, and the Priority Academic Program Development of Jiangsu Higher Education Institutions.

Data availability

All the data that support the findings of this study are included in the article and/or Supplementary Information.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

The animal study was reviewed and approved by the Institutional LaboratoryAnimal Care and Use Committee of Soochow University