Human luminal A type of breast cancer MCF-7 and ZR-75-1 cells, human TNBC MDA-MB-231 and MDA-MB-453 cells, human mammary epithelial cells (HMEpC), and mouse TNBC 4T1 cells of BALB/cfC3H strain were purchased from the cell bank of Chinese Academy of Sciences (Shanghai, China). MCF-7 and HMEpC cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (Gibco BRL, Grand Island, NY, USA), while MDA-MB-231, MDA-MB-453, and ZR-75-1 cells were cultured in DMEM/Nutrient Mixture F-12 medium (Thermo Fisher Scientific, Shanghai, China). 4T1 cells were cultured in RPMI-1640 medium (Thermo Fisher Scientific, Shanghai, China), supplemented with 10% FBS (fetal bovine serum) in a humidified atmosphere of 5% CO₂. The cell lines had been authenticated to be negative for mycoplasma infection with a short tandem repeat analysis using a PCR-based Universal Mycoplasma Detection kit (American Type Culture Collection, Manassas, VA, USA). The detailed information for antibodies, reagents, and kits is shown in Table S1 in Supplementary Material .

The methods have been previously described (6, 24). In brief, an NRP-1 shRNA pSuppressorNeo vector targeting NRP-1 gene sequence (GGACAGAGACTGCAAGTAT) (corresponding to nucleotides 519-537 of human NRP-1 [RefSeq transcript: NM_003873.7] and nucleotides 721-739 of mouse NRP [RefSeq transcript: NM_008737.2]) and a scrambled shRNA vector (Sc-shRNA) were constructed. MDA-MB-231 and 4T1 cells were seeded in 10-cm dishes, grown to approximately 67% confluence, and then transfected with 4 μg of each vector by using Lipofectamine2000. Cells were detached by trypsinization 48 h after transfection and seeded in the selection medium containing geneticin (500 μg/ml). Stable transfectants were selected after 3-4 weeks of culture, and from MDA-MB-231 and 4T1 cells, four stably transfected cell lines, namely MDA-MB-231-NRP^low, MDA-MB-231-Sc, 4T1-NRP^low, and 4T1-Sc cells were generated, respectively.

The full-length of the 3’-untranslated region (UTR) of human NRP1 mRNA (NCBI Gene ID: 8829) was cloned into a pMIR-REPORT luciferase reporter vector (Ambion) to generate an SV40 promoter-driving luciferase reporter vector, which was used as a wild-type vector for miR-124-3p because it contained the putative miR-124-3p binding sequence (CUAUGUCCUCUCAAGUGCCUUUUUG). This binding sequence was mutated to “CUAUGUCCUCUCAACACGGAAUUUG” to generate a mutated vector. Luciferase reporter transfection and assay were performed as described previously (7). Briefly, the above vectors and an empty vector without 3’-UTR of NRP-1 were transfected into cells. Luciferase activities in cells were measured by using a luciferase assay kit (Promega, Madison, WI), and miRNA function was expressed as the percentage of the luciferase activity of the reporter vector with 3’-UTR of NRP-1 over that of the empty vector.

Three sets of animal experiments were carried out to examine the role of NRP-1 and/or miR-124-3p on the growth of TNBC tumors and liver metastasis. Female 6-8-week nude BALB/c-nu/nu mice and BALB/c mice were purchased from the SLAC Laboratory Animal Co., Ltd. (Shanghai, China). Animals were housed at the Animal Research Center, the First Affiliated Hospital of Harbin Medical University. The protocols have been approved by the Institute Animal Ethics Committee with a permit (No. SYXK20020009), which complies with the Experimental Animal Regulations by the National Science and Technology Commission, China.

All the other analyses and assays used in the study have been reported previously (6, 23, 25, 26), and are described in detail in the Supplementary Material .

GraphPad Prism 8.02 (GraphPad Software, San Diego, CA, USA) was employed for statistical analyses. Data are expressed as mean values ± standard deviation. Multiple comparisons were made with a one-way analysis of variance (ANOVA) followed by a Tukey post-hoc test. Comparisons between two groups were made by a t-test. The relationship between the expression of NRP-1 and miR-124-3p was analyzed by using Pearson’s correlation coefficient. P < 0.05 was considered statistically significant.

The order of cell lines with the highest to the lowest expression of NRP-1 was MDA-MB-231, MDA-MB-453, 4T1, MCF-7, ZR-75-1 and HMEpC cells; and TNBC cells (MDA-MB-231, MDA-MB-453 and 4T1) expressed higher levels of NRP-1 than non-TNBC cells (MCF-7 and ZR-75-1) ( Figures 1A, B ). Compared with parental cells, MDA-MB-231-NRP^low cells expressed significantly lower, while MDA-MB-231-Sc cells expressed similar levels of NRP-1 and were used as controls ( Figures 1C, D ). Similarly, 4T1-NRP^low cells expressed significantly lower, while 4T1-Sc cells expressed similar levels of NRP-1, compared with 4T1 cells ( Figures 1E, F ).

MDA-MB-231-NRP^low cells had significantly lower viability, but MDA-MB-231-Sc cells had similar viability, compared with MDA-MB-231 cells ( Figure 2A ). The depletion of NRP-1 significantly downregulated the expression of cyclin E and cyclin-dependent kinase 2 (CDK2), but not had little effect on cyclin D1, in agreement with the previous studies (6, 7). For further exploring the upstream factors that regulate cyclin E and CDK2, we found that NRP-1 depletion significantly upregulated the expression of p27 but not p21 ( Figure 2B ). Since cyclin E and CDK2 are key regulators for cell cycle, we examined cell cycle distribution by using flow cytometry, which showed that 60.21% of MDA-MB-231-NRP^low cells were arrested at the G0/G1 phase, which was significantly higher than that in MDA-MB-231-Sc cells (40.73%) ( Figures 2C, D ). The BrdU incorporation assay showed that depletion of NRP-1 significantly inhibited cell proliferation as 31.6% of MDA-MB-231-NRP^low cells were BrdU-positive, which was significantly lower than that of MDA-MB-231-Sc cells (78.2%) ( Figures 2E, F ).

We next screened potential miRNAs that have putative binding sites with the 3’UTR of human NRP-1 gene by using multiple miRNA prediction tools including miRanda (https://omictools.com/miranda-tool), TargetScan (http://www.targetscan.org/), miRWalk (http://mirwalk.umm.uni-heidelberg.de/), mirdb (http://mirdb.org/), and miRTarBase (http://mirtarbase.mbc.nctu.edu.tw/), which identified 12 miRNA candidates ( Figure S1 in Supplementary Material ). The full-length of 3’UTR of human NRP1 mRNA (NCBI Gene ID: 8829) was inserted into an SV40 promoter-driving luciferase reporter vector, which was co-transfected into MDA-MB-231 cells with each miRNA mimics or negative control oligonucleotides ( Table S2 in Supplementary Material ). Among the 12 miRNAs, miR-124-3p displayed the strongest ability to inhibit the luciferase activity ( Figure 3A ). A further investigation showed a highly conserved miR-124-3p binding site on the 3’UTR of NRP1 in all the available species ( Figure 3B ) and thus miR-124-3p was selected for further investigation. A mutated luciferase reporter vector was constructed by inserting a mutated 3’-UTR of NRP-1 gene ( Figure 3C ). The inhibitory effect of miR-124-3p on the luciferase activity was shown to be in a dose-dependent manner ( Figure 3D ). MiR-124-3p mimics or antagomiR-124-3p significantly altered the luciferase activity in cells co-transfected with the wild-type vector but not the mutated vector ( Figure 3E ). The results indicate that the binding site on NRP-13’-UTR is essential for miR-124-3p to display its regulatory effect in MDA-MB-231 cells.

NRP-1 depletion had shown no effect on miR-124-3p expression in MDA-MB-231 cells ( Figure 4A ) and mouse 4T1 cells ( Figure 4B ). Human miR-124-3p (MIMAT0000422) and mouse miR-124-3p (MIMAT0000134) share the same gene sequence (http://www.mirbase.org). MiR-124-3p mimics transfection resulted in the downregulation of NRP-1, while antagomiR-124-3p transfection led to an increase in NRP-1 expression, compared with negative control, in both human MDA-MB-231 cells ( Figure 4C ) and mouse 4T1 cells ( Figure 4D ). The results are in agreement with a highly conserved miR-124-3p binding site on the 3’UTR of NRP1 in all the species ( Figure 3B ).

By applying qRT-PCR analyses, we detected the expression of miR-124-3p in available cell lines and found that MDA-MB-231 cells expressed the lowest level of miR-124-3p ( Figure S2A in Supplementary Material ). By using a Pearson test, we analyzed the correlation of miR-124-3p expression levels with the expression levels of NRP-1, and a negative correlation was observed ( Figures S2B, C in Supplementary Material ).

Transfection of miR-124-3p mimics inhibited the proliferation of MDA-MB-231-Sc cells, while antagomiR-124-3p could partially abolish the reduced proliferation of MDA-MB-231-NRP^low cells by NRP-1 depletion ( Figure 5A and Figure S3 in Supplementary Material ). Similar results were obtained in mouse 4T1 cells ( Figure S4 in Supplementary Material ). In mechanistic exploration, the role of miR-124-3p on cell proliferation was supported by its regulatory effects on the expression of CDK2, cyclin E and p27 through downregulating NRP-1 expression ( Figure 5B ). Specifically, miR-124-3p mimics downregulated the expression of cyclin E and CDK2, and upregulated the expression of p27, while antagomiR-124-3p showed an opposite effect ( Figure 5B ). Since antagomiR-124-3p inhibits the negative regulatory function of miR-124-3p, which binds to the 3’-UTR of NRP-1 mRNA, while NRP-1 shRNA targets the coding sequence (CDS) of NRP-1 mRNA, two regulatory mechanisms on NRP-1 expression exist in MDA-MB-231-NRP^low cells transfected with antagomiR-124-3p. In addition, NRP-1 shRNA and antagomiR-124-3p may also transfect different cells because they are unable to transfect 100% of cells. Therefore, antagomiR-124-3p only partially abolished the effects of NRP-1 depletion ( Figure 5B ).

NRP-1 has been shown to play a role in promoting the migration of cells from other cancer types (7, 23). Here we confirmed that NRP-1 depletion significantly inhibited the migrating ability of MDA-MB-231 cells by Transwell ( Figures 6A, B ) and scratch ( Figures 6C, D ) assays. It is well known that breast cancer cells gain their ability to migrate and invade through the epithelial-to-mesenchymal transition (EMT) (27). We showed here that NRP-1 depletion led to significantly downregulation of N-cadherin, matrix metalloproteinase (MMP)-2 and MMP-9, and upregulation of E-cadherin ( Figure 6E ). The results of gelatin zymography assays also showed that MDA-MB-231-NRP^low cells had lower activities of MMP-2 and MMP-9 ( Figure 6F ).

Furthermore, miR-124-3p mimics reduced the number of migrated cells, while antagomiR-124-3p partially abolished the effect of NRP-1 depletion on cell migration ( Figure 6G ); and similar results were obtained with mouse 4T1 cells ( Figure S4 in Supplementary Material ). The above effects of miR-124-3p were in accordance with the changes in E-cadherin and N-cadherin expression ( Figure 6H ) and the results of Matrigel invasion assay ( Figure S5 in Supplementary Material ).

Compared with control MDA-MB-231-Sc tumors, MDA-MB-231-NRP^low tumors grew significantly faster ( Figure 7A ), and were significantly larger (1035.4±101.6 cm³) and heavier (1254.3±158.5 mg) than control tumors (478.2±88.5 cm³ and 524.6±109.7 mg) at the end of experiments ( Figure 7B ). In agreement with NRP-1 expression in cultured cells ( Figure 1D and Figure 2B ), MDA-MB-231-NRP^low tumors had a weaker expression of NRP-1 as examined by immunohistochemistry ( Figure 7C ). MDA-MB-231-NRP^low tumors also had weaker cell proliferation detected by immunohistochemistry with an Ab against Ki-67 ( Figures 7C, D ).

The effects of NRP-1 depletion on liver metastasis of TNBC cells were examined by adopting mouse 4T1 cells, which can spontaneously metastasize from the primary tumors to distant sites including livers (28). Primary 4T1-NRP^low tumors were significantly smaller than control 4T1-Sc tumors ( Figure S6 in Supplementary Material ). 4T1-NRP^low cells injected into the mice had formed significantly fewer metastatic nodules (Number: 45.6± 8.5) in livers (with an average number of 45.6± 8.5) than 4T1-Sc cells (Number: 102.7 ±18.6) ( Figures 7E, F ).

Tumors treated with miR-124-3p mimics were significantly smaller (611.6±74.8 mm³) than vehicle-injected tumors (1082.4 ± 120.5 mm³), which is not significantly different from negative control-treated tumors (994.7±113.4 mm³), 20 days after treatment commencement ( Figure 8A ). The downregulation of NRP-1 and reduced expression of Ki-67 in miR-124-3p mimics-treated tumors harvested 3 days after gene injection were confirmed by immunohistochemistry ( Figures 8B–D ). Western blot analysis of tumor homogenates also showed downregulation of NRP-1, cyclin E, CDK2 and N-cadherin, and upregulation of p27 and E-cadherin in miR-124-3p mimics-treated tumors ( Figure 8E ).

It has been reported that NRP-1 can co-activate the TGF-β pathway (29), which plays a key role in the progression of TNBC (30). We confirmed the binding of TGF-β and NRP-1 in MDA-MB-231 cells as detected by co-immunoprecipitation assay ( Figure S7 in Supplementary Material ), in agreement with published studies (8, 31). We also showed that the interaction of NRP-1 and TGF-β ligand-induced the activation of the TGF-β pathway in MDA-MB-231 cells. Recombinant human TGF-β protein, a stimulator of the TGF-β pathway, and galunisertib, which is a specific inhibitor of TGF-β receptor (TGF-βR) and has been tested to treat breast cancer in clinical trials (32), or their combination were added to cultured cells. TGF-β protein and/or galunisertib had little effect on the expression of NRP-1 or TGF-βRI ( Figure 9A ). However, recombinant TGF-β protein increased, while galunisertib reduced, the expression of p-TGF-βRI. The activation of the TGF-β signaling pathway by TGF-β protein, or the inhibition by galunisertib or NRP-1 depletion, had no effect on the expression of Smad2/3 or Samd4. However, TGF-β protein incubation induced an increase in p-Smad2/3, Snail and N-cadherin expressions, and reduced the expression of E-cadherin; while inhibition of TGF-β signaling by galunisertib or NRP-1 depletion reduced the expression of p-Smad2/3, Snail and N-cadherin, and increased the expression of E-cadherin, and also abolished the effects of TGF-β protein in part ( Figure 9A ).