Loss of endothelial EMCN drives tumor lung metastasis through the premetastatic niche

EMCN^flox/flox mice and Tek-CreERT2 mice were purchased from Beijing VIEWSOLID Biotechnology Co., Ltd., and raised in a temperature-controlled facility with a 12-h light/dark cycle in a specific pathogen-free environment in a separate ventilation cage. Mice had ad libitum access to food and water. All animal experiments were approved by the Animal Experiments Committee of the Chinese Academy of Medical Sciences (IACUC: GR21003). EMCN^flox/flox mice were crossed with Tek-CreERT2 mice to generate endothelial cell-specific EMCN knockout mice (EMCN^ecko). All female or male mice aged 8–10 weeks (weight 23–25 g) were randomly divided into groups for follow-up experiments. All genotyping was confirmed by PCR. Gene deletion by Cre recombinase was achieved by intraperitoneal (i.p.) injection of tamoxifen (75 mg/kg body weight) (Sigma‒Aldrich, CAS# 10540-29-1) every day for 5 days, starting seven days before tumor cell injection. EMCN knockout efficiency was determined by Western blot.

LLC (LL/2) murine lung carcinoma cells, B16-F10 melanoma cells and PUMC-HUVEC-T1 (later HUVEC, SV40T transformed immortalized human umbilical vein endothelial cells) were purchased from the National Infrastructure of Cell Line Resource and maintained in DMEM (Gibco) supplemented with 10% FBS (Gibco). The human salivary gland adenoid cystic carcinoma cell line SACC-LM (highly invasive) was obtained from the Peking University Hospital of Stomatology. The cells were cultured in RPMI 1640 supplemented with 10% fetal bovine serum (FBS) and a 1% penicillin‒streptomycin solution (Invitrogen). Cell line authentication was performed by short tandem repeat (STR) assessment. All experiments were performed with mycoplasma-free LLC-luciferase clones produced by continuous puromycin (800 μg/ml) screening of LLC cells overexpressing the lentiviral-driven luciferase gene. For EMCN deletion, three human shRNA sequences were cloned into a plasmid vector, and lentivirus was packaged by Shanghai Jikai GENE Biological Company. HUVECs were infected with lentivirus (MOI = 20). HUVEC cell lines with stable EMCN knockdown were screened by puromycin. Negative control and EMCN knockdown were termed HUVEC/shNC and HUVEC/shEMCN, respectively. HUVEC proliferation was analyzed by CCK8 kits (Dojindo, C0038). For angiogenesis, the concentrated and reduced growth factor matrix gel (Corning) was placed into a 24-well plate, and the plate was incubated at 37 °C for 30 min. HUVECs/shNC and HUVECs/shEMCN (1 × 10⁵ cells) suspended in serum-containing media were added to 24-well plates with solidified Matrigel. Images were obtained with a contrast microscope and analyzed using the Fiji plug-in for ImageJ software.

HUVECs were infected with lentivirus (MOI = 20). Cells with stable NICD knockdown were selected in puromycin-supplemented medium. NICD knockout efficiency and EMCN expression were detected by Western blotting. Condition medium (CM) was prepared by seeding 1 × 10⁶ LLC, SACC-LM and B16-F10 cells in common culture medium supplemented with 10% FBS for 48 h. The harvested CM was then centrifuged at 250g for 10 min to remove cells and debris. Then, HUVECs were washed with PBS and incubated further for 48 h in different tumor cell-conditioned media.

To deplete Ly6G⁺ neutrophils, mice were intraperitoneally injected with 7.5 mg/kg anti-Ly6G⁺ (clone 1A8; BioXCell) once every three days. Neutrophil depletion was confirmed by immunofluorescence. For visualization of the luciferase reporter gene that is expressed by the LLC cells, d-Luciferin (PerkinElmer cat. #122796) was intraperitoneally injected (150 mg/kg), and the lungs were dissected 10 min after injection. Luciferase-positive regions were imaged using IVIS Lumina II (Caliper Life Sciences).

Tumor-bearing WT mice and EMCN^ecko mice were injected intravenously with 100 µl of a mixture containing 2 mg/ml rhodamine-conjugated dextran (70 kDa) in PBS. After 30 min, the mice were euthanized, and cardiac perfusion was performed with 10 ml PBS. Lung tissue was fixed with 4% paraformaldehyde (PFA) at 4 °C for 24 h; washed in PBS for 5 min; passed through sucrose solutions of 10%, 20% and 30% for 24 h; and embedded in OCT. Frozen blocks were cut into 10-µm cryosections. Images were obtained using a Leica confocal microscope. The fluorescence intensity of 70 kDa rhodamine-dextran was quantified by using ImageJ software.

Lung and tumor tissues were fixed in 10% formaldehyde solution in PBS at room temperature followed by routine dehydration, paraffin embedding, and tissue sectioning. Paraffin sections (4 µm) were stained with hematoxylin–eosin to observe the structure of the main organs and lung metastasis. For immunofluorescence staining, in brief, paraffin sections were dewaxed and placed into water. The sections were placed into sodium citrate buffer, heated in a microwave oven for antigen repair, restored to room temperature, and blocked with goat serum at room temperature for 1 h. Sections were incubated with the primary antibodies anti-Ly6G⁺ (ab238132, 1:200), anti-Arg2⁺ (ab264071, 1:200) and anti-NOS2 (ab115819, 1:200) at 4 ℃ overnight. The slices were washed thrice with PBS for 3 min each time followed by incubation with Alexa Fluor 488- or 594-conjugated secondary antibodies (1:5000 dilutions) for 1 h. DAPI was used to stain the nucleus for 15 min, and samples were subject to imaging with a Leica fluorescence microscope. A routine protocol was performed for immunohistochemistry. The primary antibody dilutions were S100A8/A9 (ab22506, 1:1000), anti-CD31 (D8V9E, 1:500), and anti-Ki67 (ab16667, 1:1000).

Total RNA from cultured cells (HUVECs/shNCs and HUVECs/shEMCNs) and mouse lung tissues was isolated with TRIzol reagent (Invitrogen) as instructed. cDNA was synthesized from 2 μg of total RNA with random primers using a Thermo kit, and the concentration was measured by Colibri. mRNA expression was assessed based on the threshold cycle (Ct), and relative expression levels were calculated as 2^−ΔΔct after normalization to GAPDH expression. The primers used for quantitative real-time PCR are listed in Additional file 6: Table S1.

Western blot analysis was implemented using a standard protocol. The primary antibodies used for Western blot analysis included anti-endomucin (sc-65495, 1:500), anti-endomucin (ab96315, 1:1000), anti-claudin-5 (ab131259, 1:1000), anti-MMP-9 (ab283575, 1:1000), anti-ZO-1 (ab216880, 1:1000), and anti-TGF-β (ab215715, 1:1000). The secondary antibodies included goat anti-mouse (ZB-2055, 1:10000), goat anti-rat (ZB-2040, 1:10000) and goat anti-rabbit (ZB-2306, 1:10000). Anti-β-actin (ab8226, 1:2000) was used as a loading control.

Total RNA was extracted using TRIzol reagent according to the manufacturer’s protocol. RNA purity and quantification were evaluated using a NanoDrop 2000 spectrophotometer (Thermo Scientific, USA). RNA integrity was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Then, the libraries were constructed using the TruSeq Stranded mRNA LT Sample Prep Kit (Illumina, San Diego, CA, USA) according to the manufacturer’s instructions. Transcriptome sequencing and analysis were performed by OE Biotech Co., Ltd. (Shanghai, China).

The libraries were sequenced on an Illumina HiSeq X Ten platform, and 150-bp paired-end reads were generated. Transcriptome sequencing and analysis were performed by OE Biotech (Shanghai, China). Raw data (raw reads) were processed using Trimmomatic [16]. The reads containing poly-N and the low-quality reads were removed to obtain clean reads. Then, the clean reads were mapped to the reference genome using HISAT2 [17]. The FPKM [18] value of each gene was calculated using cufflinks [19], and the read counts of each gene were obtained by htseq-count [20]. DEGs were identified using the DESeq R package functions estimate SizeFactors and nbinomTest. A P value < 0.05 and fold change > 2 or fold change < 0.5 were set as the thresholds for significant differential expression. Hierarchical cluster analysis of DEGs was performed to explore gene expression patterns. GO enrichment and KEGG [21] pathway enrichment analyses of DEGs were performed using the R package based on the hypergeometric distribution. The sequencing coverage and quality statistics for each sample are summarized in Additional file 7: Table S2.

TCGA database provides high-throughput analysis of different tumors, including data on mRNA expression. Combining bioinformatics analysis with patient clinical information lays a foundation for improving cancer prevention and discovering new targets for treatment. In this study, we downloaded RNA-seq data of lung adenocarcinoma from TCGA database. mRNA data from 594 samples, including 59 normal samples and 535 lung cancer samples, and mRNA data from 1222 samples, including 112 normal breast samples and 1110 breast cancer samples, were used in this analysis. Using the R software package, the downloaded data were normalized and differentially analyzed to obtain differentially expressed mRNAs. We used the R package Survival to examine the prognostic potential of EMCN expression levels in cancers. Survival-relevant mRNAs with a log-rank P value < 0.05 were considered significant. The relevant data provided by TCGA are publicly available.

All graphs and statistical analyses were completed using GraphPad Prism software v8.0. Significant differences were evaluated using an independent sample T test, and multiple treatment groups were compared within individual experiments by ANOVA. TCGA data were downloaded from the cBioPortal website, and the log-rank test was used for comparison of survival outcomes with the Kaplan‒Meier method. Values of p < 0.05 were considered significant. All values are presented as the mean ± SD.

To identify the key genes and pathways involved in tumor metastasis, we compared the transcriptomic profiles of 55 normal and 538 lung cancer samples from TCGA database. In total, 3719 genes were significantly upregulated, and 1811 genes were significantly downregulated (Fig. 1A and B). Similar transcriptomic analysis of 112 normal 1110 breast cancer samples revealed upregulation of 7393 and downregulation of 3199 genes (Additional file 1: Fig S1A and S1B). Given that tumor cell interactions with endothelial cells play a critical role in cancer metastasis [22], we first focused on the expression analysis of these genes. We found that EMCN is specifically expressed in microvascular endothelial cells and showed remarkable downregulation in the majority of both metastatic lung and breast tumors. We hypothesized that EMCN downregulation may be important for tumor metastasis and recurrence (Fig. 1C and Additional file 1: Fig. S1C). To further investigate the clinical significance of EMCN expression in metastatic lung and breast cancer patients, we analyzed EMCN expression in patients with and without metastasis. The results showed that EMCN expression was significantly reduced in metastatic tumors (Fig. 1D and Additional file 1: Fig. S1D) and was strongly reduced in tumor recurrence (Fig. 1E and Additional file 1: Fig. S1E). These analyses revealed an important role of EMCN in the progression of tumor metastasis.

We used an inducible knockout model using Tek-creERT2 mice to drive endothelial-specific deletion of EMCN (hereafter referred to as EMCN^ecko). Endothelial EMCN inactivation was confirmed at the gene level by PCR and at the protein level by Western blot (Additional file 2: Fig. S2A and S2B). EMCN^ecko mice appeared phenotypically normal; however, histological examination revealed an increase in red blood cells in the lungs of these animals compared with the wild-type (WT) mice (Additional file 2: Fig. S2C). To determine whether EMCN mediates tumor growth, syngeneic LLC cells were injected subcutaneously into the flanks of WT and EMCN^ecko mice. Neoplasms were formed within 2 weeks and displayed no distinct impairment in growth rate or tumor angiogenesis between EMCN^ecko and WT mice, which was also evidenced by Ki67 and CD31 staining, respectively (Fig. 2A–D). Consequently, EMCN expression in ECs does not affect tumor growth in vivo.

Cancer cells penetrate blood vessels through intravasation and extravasation, and both processes require the ability to break through endothelial cells [23]. Tumor cell metastasis typically occurs in small capillaries [24], and EMCN is a glycoprotein expressed in endothelial cells of these capillaries. We hypothesized that EMCN depletion in tumors may obliterate the integrity of the endothelial barrier and lead to an increase in cell infiltration. To determine the effect of EMCN deficiency on tumor metastasis in vivo (Fig. 2E), EMCN^ecko mice were injected with LLC cells through the tail vein. The pathology results showed significantly higher metastasis in EMCN^ecko mice (Fig. 2F), suggesting that EMCN depletion caused an increase in metastasis. Furthermore, EMCN knockout mice showed significantly reduced survival compared to WT mice when LLC cells were injected into the venous circulation (Fig. 2G).

To construct a mouse model that is more consistent with the metastasis of clinical tumor patients, we resected subcutaneous tumors 2 weeks after implantation to simulate surgical resection in tumor patients and allowed time for disseminated tumor cells to form detectable metastases. The effect of EMCN knockout on lung metastasis was observed after resection of LLC primary tumors (Fig. 2H). EMCN^ecko mice exhibited a significant increase in metastasis upon primary tumor resection compared with WT mice (Fig. 2I and Fig. S2D). We used a second animal model to confirm the same results, indicating that the occurrence of lung metastasis is not cell specific (Additional file 2: Fig. S2E). These results demonstrated that (i) loss of EMCN in the vascular endothelium results in markedly greater spontaneous metastases to the lung and (ii) the establishment of EMCN^ecko model mice with greater spontaneous metastases.

To further understand the mechanism of endothelial EMCN in tumor metastasis, we constructed a stable EMCN knockdown cell line in vitro. The results showed that EMCN knockdown significantly inhibited endothelial cell proliferation (Additional file 3: Fig. S3A and S3B). Next, we evaluated the gene expression profiles of EMCN (HUVEC/shEMCN) knockdown and control endothelial cells (HUVEC/Con313). Differentially expressed genes were selected as those showing either upregulation or downregulation by > twofold in HUVECs/shEMCN compared with HUVECs/Con313. KEGG analysis of differentially expressed genes showed that EMCN knockdown in HUVECs significantly affected cell junctions (Fig. 3A). Patricia A. D’Amore et al. reported that EMCN controls angiogenesis by regulating the activation of VEGFR2 in vitro [13]. We further confirmed whether EMCN controls endothelial cell junctions in vitro. HUVECs/Con313 and HUVECs/shEMCN were seeded on Matrigel to facilitate capillary tube formation. Strikingly, EMCN deficiency impeded HUVEC angiogenesis, which is consistent with previous studies [13] (Fig. 3B). In the in vivo study, we found that EMCN did not affect angiogenesis but did affect tumor metastasis. By comparing EMCN expression levels in naive, premetastatic and metastatic lung tissue, we found that EMCN expression levels were decreased significantly in tumor metastasis (Additional file 3: Fig. S3C). Thus, we hypothesized that endothelial EMCN deficiency changes vascular barrier properties, such as vascular permeability. Thus, we established a pulmonary microvascular permeability model by intravenous injection of rhodamine-dextran (70 kDa). After 30 min, the permeability of high-molecular-weight rhodamine in the lung parenchyma was analyzed by fluorescence microscopy. The results revealed slight leakage in the lungs of EMCN^ecko mice compared with WT mice (Additional file 3: Fig. S3D). Interestingly, EMCN^ecko tumor-bearing mice exhibited significantly increased lung vascular permeability compared with WT mice, as indicated by the fluorescence signal intensity observed in frozen lung sections (Fig. 3C, D). The intercellular junction of endothelial cells composed of adhesive junctions and tight junctions is the key factor in maintaining vascular integrity. ZO-1 and Claudin5 are important adhesion molecules that maintain endothelial cell‒cell contact [25‐27]. To explore how EMCN regulates vascular permeability in vivo, we further detected changes in ZO-1 and Claudin5 in the lungs of WT and EMCN^ecko tumor-bearing mice. The results showed that EMCN deficiency caused a significant reduction in ZO-1 and Claudin5 expression (Fig. 3E). In brief, these results suggested that EMCN depletion affected the expression of ZO-1 and Claudin5 and the integrity of the vascular barrier of pulmonary capillaries, causing an increase in vascular permeability.

Next, we attempted to clarify the steps of the metastatic cascade affected by EMCN defects. Based on our results indicating (i) no difference in primary tumor growth between WT and EMCN^ecko mice (Fig. 2B) and that (ii) EMCN deficiency increased lung metastasis (Fig. 2F and I) and pulmonary vascular permeability (Fig. 3D), we hypothesized that EMCN deletion might contribute to enhanced colonization of metastatic cells after changes in the host microenvironment, which relies on the formation of a lung metastatic niche. Vascular permeability is considered to be one of the important characteristics of the premetastatic niche [28] that plays an important role in tumor metastasis [29‐31]. Transcriptome sequencing and protein detection were used to evaluate whether EMCN defects in endothelial cells could affect the formation of pulmonary premetastatic niches by affecting vascular permeability. MMP9, TGF-β and S100A8/A9 have been demonstrated to increase lung inflammation or immunosuppressive factors and play a key role in the initiation of the premetastatic niche [28, 32]. Next, transcriptome analysis of lung tissue was performed in EMCN^ecko mice, WT mice and tumor-bearing WT mice. We focused on the changes in premetastatic niche-related genes. Differentially expressed genes were identified as those upregulated or downregulated by > 1.5-fold in the groups of tumor-bearing WT mice vs. WT mice and EMCN^ecko vs. WT mice. We found that EMCN deficiency led to changes in the premetastatic niche in the lung, which was similar to the premetastatic niche induced by LLC tumors (Fig. 4A). Furthermore, EMCN deficiency was associated with changes affecting the host microenvironment, resulting in an increase in genes related to the premetastatic niche (TGFB, S100A8/A9, MMPs, etc.). The key factors in the premetastatic niche were detected, and the expression of S100A8/A9, MMP9 and TGF-β was increased in the lungs of EMCN^ecko mice. The expression levels of MMP9, TGF-β and S100A8/A9 were confirmed by Western blot and IHC, respectively (Fig. 4B–D). In brief, these results indicated that EMCN deficiency leads to the formation of a lung metastatic niche.

It has been demonstrated that overexpression of EMCN can prevent neutrophil-endothelial cell interactions in vitro and inhibit the infiltration of CD45⁺ and NIMPR14⁺ (neutrophil marker) cells, but there were no changes in F4/80⁺ (monocyte/macrophage marker) cells [12]. An increasing number of studies have confirmed that neutrophils play an important role in the formation of a premetastatic niche in tumor lung metastasis [33, 34]. We analyzed the premetastatic niche lungs of WT and EMCN^ecko tumor-bearing mice by RNA-seq. The results showed that EMCN deletion affected the inflammatory response and neutrophil chemotaxis (Fig. 4E). The increase in cytokines was measured by qRT‒PCR (Fig. 4F). Therefore, we hypothesized that EMCN deficiency may lead to the aggregation of neutrophils in the lung, promote the formation of a premetastatic niche, and thereby promote growth and metastasis. We next detected neutrophil infiltration in the premetastatic niche and metastatic lung of WT and EMCN^ecko mice by immunofluorescence staining and found that Ly6G⁺ and S100A9⁺ (neutrophil marker) cells were significantly recruited in the lungs of EMCN^ecko tumor-bearing mice compared with WT tumor-bearing mice (Fig. 4G, Additional file 3: Fig. S3E). The above results led to the conclusion that EMCN deletion can cause neutrophil infiltration in vivo.

Next, we evaluated whether the increased neutrophils caused by EMCN deficiency are functional, revealing an important mechanism of lung metastasis. Neutrophils in vivo were depleted with Ly6G⁺ antibody before and after the tail vein injection of LLC cells into EMCN^ecko mice. Through the formation of lung metastasis foci and in vivo imaging, we found that neutrophil depletion reduced lung metastasis in vivo (Fig. 5A and B). In brief, these results demonstrated that Ly6G⁺ neutrophils play an important role in lung metastasis caused by EMCN deficiency.

A deeper understanding of the metastasis process is needed to develop new therapies for cancer patients. We assessed whether the increased TGF-β was functionally important in LLC lung metastasis of EMCN^ecko mice. It has been reported that the polarization of N2 neutrophils promotes tumor metastasis of breast and gastric cancers [35, 36]. TGF-β has been shown to play an important role in inducing N1 to N2 polarization [37]. Through bioinformatics analysis of TCGA dataset and experimental verification, we found a negative correlation between EMCN and TGF-β (Figs. 4C and 5C). Therefore, we hypothesized that EMCN deletion led to an increase in TGF-β in the microenvironment, promoted the polarization of N1 neutrophils into N2 neutrophils, and thereby promoted lung metastasis. Therefore, specific blockade of TGF-β-induced neutrophil polarization is a meaningful therapeutic strategy to treat lung metastasis.

To further assess the anti-TGF-β suppressive effect of neutrophil polarization, we injected LLC cells subcutaneously into WT and EMCN^ecko mice. EMCN^ecko mice were treated with or without anti-TGF-β antibody and examined for N1-/N2-associated markers. Consistent with the above-described results (Fig. 2B), EMCN deletion did not affect growth but promoted lung metastasis of primary tumors in the WT and EMCN^ecko groups (Fig. 5G and H). However, compared with EMCN^ecko mice, anti-TGF-β antibody did not affect tumor growth but inhibited lung metastasis caused by EMCN deletion (Fig. 5E–H and Additional file 3: Fig S3F). We also examined the expression of relevant N1- and N2-neutrophil markers in lung metastatic tumors by immunofluorescence analysis and found that EMCN^ecko mice could induce N2 neutrophils (Arg2⁺) to infiltrate metastatic tumors. However, anti-TGF-β antibody significantly inhibited N2 neutrophils, and N1 neutrophil infiltration was primarily observed (Fig. 5I and Additional file 3: Fig S3G). Taken together, these results strongly suggest that the anti-TGF-β antibody is a promising therapeutic target that can be used to prevent or treat lung metastasis caused by neutrophil polarization.

Previous studies have identified the key regulatory relationship of the Notch signaling pathway in the regulation of EMCN expression in hepatic ischemia/reperfusion injury. Pharmaceutical Notch blockade dramatically upregulated EMCN and prevented trans-endothelial migration of neutrophils in vitro. Furthermore, in RBPj (integrated transcription factor of typical Notch signaling) knockout mice, the Notch signal was inactivated, but the expression of EMCN was upregulated [15]. Through bioinformatics analysis of TCGA data, an obvious negative correlation was noted between EMCN and Notch1 (Fig. 6A). To further investigate whether Notch regulates EMCN expression, EMCN protein expression was detected by inhibiting NICD (a key protein of the Notch signaling pathway). The results showed that NICD inhibition significantly increased the expression level of EMCN in HUVECs (Fig. 6B and Additional file 4: Fig. S4A). After we cocultured HUVECs with different tumor cell-conditioned media, the results showed that tumor cells significantly activated NICD in HUVECs, which was consistent with previous studies [9]. Interestingly, NICD activation significantly inhibited the expression of EMCN on HUVECs (Fig. 6C and Additional file 4: Fig. S4B). Accordingly, we used the Notch pathway inhibitor DAPT, a γ-secretase inhibitor, to detect its effect on tumor growth and metastasis using a syngeneic mouse model. To determine whether DAPT can inhibit tumor growth and metastasis, we treated tumor-bearing mice with DAPT or vehicle. The results showed that DAPT slightly inhibited tumor growth and metastasis compared with the vehicle group (Fig. 6D–F). Immunohistochemistry results for Ki67 staining confirmed these findings (Additional file 4: Fig. S4C and 4D). Western blot analysis showed that EMCN expression was significantly upregulated in the lungs of DAPT-treated mice compared with that in the vehicle-treated mice (Additional file 4: Fig. S4E). These results confirmed that DAPT could inhibit tumor growth and elicit a suppressive effect on lung metastases by upregulating EMCN.

TGF-β weakens the intrinsic antitumor potential of immune cells in the tumor microenvironment by increasing the inhibitory effect on key immune cells of innate and adaptive immunity. Therefore, the antitumor response of myeloid cells and lymphocytes is hypothesized to be enhanced by inhibiting TGF-β [38]. It is worth mentioning that due to the limited clinical efficacy of TGF-β inhibitor monotherapy, a large number of combined therapies are being assessed, such as combined cytotoxic drugs, radiotherapy, and immune checkpoint inhibitors [39‐42]. In light of the data described above, DAPT has a good therapeutic effect on tumor growth and metastasis in syngeneic mice, whereas the use of DAPT as a Notch signaling pathway inhibitor in combination with anti-TGF-β antibody has not been reported. We again used the mouse tumor metastasis model of syngeneic subcutaneous tumor grafting and surgical resection. Once tumors are palpable, the animals were assigned to different treatment systems as shown in Fig. 6D. Combined DAPT and anti-TGF-β administration did not significantly affect mouse body weight or damage to important organs compared with control mice, indicating that the regimen with a low dose of DAPT and anti-TGF-β is relatively safe (Additional file 5: Fig. S5A and B). The growth of subcutaneous tumors was significantly inhibited (Fig. 6D and E). Immunohistochemical analysis of Ki67 revealed differences in proliferation (Additional file 4: Fig. S4C). Tumor metastasis was found in the lung metastasis foci and HE-stained sections (Fig. 6F, Additional file 4: Fig. S4D). Notably, we found that the combination of DAPT and anti-TGF-β antibody dramatically reduced the area of the metastatic nodules and slightly inhibited tumor growth compared to the DAPT group (Fig. 6F). We used a second animal model to confirm the same results (Additional file 5: Fig. S5C and D). The survival rate of mice was significantly improved (Additional file 5: Fig. S5E). Taken together, we demonstrated the important role of DAPT combined with anti-TGF-β antibody in tumor growth and metastasis, especially the significant inhibition of tumor metastasis and improved survival using a syngeneic mouse model.

To further validate our experimental results, we examined the clinical significance of Notch1 and EMCN in lung cancer by analyzing TCGA dataset. We evaluated the relationship between Notch1 and EMCN expression levels and patient survival. The results showed that patients with high EMCN expression exhibited a higher survival rate, whereas patients with high Notch1 expression exhibited a significantly lower survival rate (Additional file 5: Fig. S5F).

In conclusion, we demonstrate a new antimetastatic effect of EMCN on lung metastasis. EMCN deletion caused a remarkable increase in metastasis by affecting the formation of the premetastatic niche. TGF-β in the premetastatic niche promoted the polarization of neutrophils and exacerbated lung metastasis. Pharmacological inhibition of Notch improved EMCN expression, inhibited tumor metastasis, and showed additive effects when combined with anti-TGF-β antibody therapy. Taken together, we demonstrate a possible new therapeutic strategy involving Notch inhibitors and anti-TGF-β antibodies in clinical tumor patients (Fig. 7).