SDS lysis buffer was used to harvest cells. For western blot, protein samples were boiled and separated on 8-12% SDS-PAGE gels. After blocking with 2% nonfat milk in Tris-buffered saline containing 1% Tween-20 for 1 hour, the membranes were incubated with specific primary antibodies overnight at 4°C. Antibody-bound proteins were detected by chemiluminescence (BIORAD, USA). The β-actin-HRP (Abways Technology, #AB2001), E-Cadherin (CST, #3195), N-Cadherin (CST, #13116), Vimentin (CST, #5741), PCNA (ABmart, #P30108s), Slug (CST, #9585), Cleaved caspase-3 (CST, #9661), Caspase-3 (CST, #9662), Anti-rabbit IgG-HRP (CST, #70742), Anti-mouse IgG-HRP (CST, #7076).

The human breast cancer cell lines LM2 and HEK293T cells were grown in Dulbecco’s modified Eagles’ medium (Hyclone, Beijing) supplemented with 10% fetal bovine serum (FBS, Gibco BRL, Gaithersburg, MD). The luciferase-expressing mouse mammary tumor cells 4T1 were cultured in RPMI-1640 Medium supplemented with 10% FBS. The mouse mammary tumor cells Py8119 were cultured in DMEM/F12(1:1) Medium containing Hydrocortisone (0.5μg/ml), mEGF (20ng/ml), Insulin(10μg/ml) and 10% FBS. All cells applied in this study were cultured at 37°C in a humidified 5% CO2 atmosphere.

2 × 10⁵ Py8119 cells were plated in each well of 12-well plates. On the next day, the culture medium was changed to medium with different concentrations BCAA, and the cells were cultured for 48 hours. Cells were collected and 10μL cell suspension were taken, where 10μL of a 0.4% (m/v) previously prepared trypan blue solution was added. After stirring, the mixture was deposited on the cell counting plate. The cell number and viability were measured with automatic cell counter (Countstar). Independent triplicates were performed by analyzing three wells per group in each experiment.

Mouse mammary tumor cells Py8119 (1×10⁵) were suspended with Matrigel (3:1) in a total volume of 75μl and were orthotopically transplanted directly into the inguinal mammary fat pads of wild type or Pp2cm knockout female mice at the age of 6-7 weeks. The Pp2cm knockout mice were constructed by our laboratory (35). 3-4 weeks after injection, mice were euthanized and tumors were dissected and examined for tumor size and immunohistochemical staining. Tumor volume was calculated using the following formula: tumor volume (cubic millimeters (mm³)) = 0.5× (length × width²).

For the lung metastasis assay, the collected lungs were fixed with formalin and submitted for hematoxylin-eosin staining (H&E). The fixed lung tissue was sent to Google, a biological company, for subsequent paraffin embedding, sectioning, and staining experiments.

Human breast cancer cells LM2 (1×10⁵) were suspended with Matrigel (3:1) in a total volume of 75μl and orthotopically transplanted into the inguinal mammary fat pads of NOD/SCID female mice at the age of 6 weeks. After injection, mice were feed with normal BCAA diet or high BCAA diet for 8 weeks. The high BCAA diet (Research Diets, Inc, A12030801) was 1.5 times higher than the normal BCAA diet (Research Diets, Inc, A11072001) (36). Eight weeks after injection, mice were euthanized and tumors were dissected and examined for tumor size and immunohistochemical staining. Tumor volume was measured weekly and calculated as before described.

The 5-6-week-old BALB/C mice were obtained from SLAC Laboratory Animals Company Limited, Shanghai, China. The purchased mice were transferred to Animal facility for 3 days and pretreated with a low BCAA diet (LBCAA, Research Diets, Inc, A12030802) or a normal BCAA diet (NBCAA, Research Diets, Inc, A11072001) for 2 weeks (36). BALB/C mice (n = 10, 11 per group) were injected in their tail veins with 5×10⁵ Luciferase-expressing 4T1 cells. After injection, BCAA diets were continued to the end of experiment. To investigate tumor metastasis, 12 days after injection, each mouse was intraperitoneally injected 200 mg/kg D-Luciferin. Bioluminescence was analyzed by IVIS system.

Tumor Dissociation Kit (Miltenyi Biotec, #130-096-730) was used for single cell suspensions from mouse tumor tissue. The single cell suspension of spleen was obtained by grinding, and white blood cells were obtained by 5% Ficoll density gradient centrifugation. Red Blood Cell Lysis Solution (10×) (Miltenyi Biotec, #130-094-183) was used to remove the erythrocytes. For cell-surface analysis, cells were stained with Fixable Viability Stain 780 (BD Horizon, #565388), anti-mCD45(BD Pharmingen, #561087), anti-mCD3e (BD Pharmingen, #557984), anti-mCD4 (BD Pharmingen, #561831), anti-mCD8a (BD Pharmingen, #561109), anti-CD11b (BD Pharmingen, #561098), anti-mF4/80 (BD Pharmingen, #743280), anti-mCD86 (BD Pharmingen, #564198) in recommended antibody concentrations and incubated at 4°C for 30 min. For the NK intracellular IFN-γ (BD Pharmingen, #562333) and granzymeB (Miltenyi Biotec, #130-101-356) cytokine staining, cells were fixed and permeabilized after stimulation with Leuko Act Cktl with GolgiPlug (BD Pharmingen, #550583) for 6 h. For the CD206 (BD Pharmingen, #565250) and NK1.1 (BD Pharmingen, #563220) staining, cells were also fixed and permeabilized. Cytofix/Cytoperm Soln Kit (BD Pharmingen, #554714) was used to fix and permeabilize the cells. Flow cytometry data was analyzed using FlowJo version 10.

2×10⁵ LM2 cells were cultured in a six-well plate (5×10⁵ cells were cultured in a 6cm plate) until 90-100% confluence and then carefully scratched with a 10μl/200μl pipette tip. After washing three times with 1×PBS to remove detached cells, the culture medium was changed to BCAA gradient medium. The BCAA gradient medium was formulated with BCAA-free DMEM and BCAA chemicals. Images in 3 different wound fields were captured at respective time points (0h and 48h) to evaluate the migration of cells.

Chambers (#3422, 8µm pore, Corning, NY, USA) without or with matrigel (#356234, BD Biosciences, CA, USA) were used to investigate migration and invasion ability of cells, respectively. Systems without Matrigel were used to measure the migration ability of cells. BCAA concentration gradient medium was used in the experiment. 2×10⁴ cells suspended in 100μL serum-free medium were seeded onto the upper chamber of 24-well plates, and 700μL of medium with 10% FBS was added to the lower chamber. 22 hours later, the medium was removed from the upper chamber. The cells on the membrane were fixed with 4% paraformaldehyde for 30 minutes, and the cells were stained with 0.1% of crystal violet (Sangon Biotech) overnight at 37°C. Then non-invading cells on the upper side of the chamber were gently removed thoroughly with a clean wet cotton swab and rinsed in clean water and dried overnight in oven at 37°C. Finally, the stained cells were counted by microscopy. Results represent the average number of cells in three fields per membrane and the experiment was repeated three times independently. As for cell invasion assays, the matrigel was diluted according to the manufacturer’s recommendations and added onto the chambers before seeding cells, then performed in the same manner as cell migration assays.

All animal experiments were performed in accordance with relevant institutional and national guidelines and regulations of Shanghai Medical Experimental Animal Care Commission.

The prognostic significance of the mRNA expression of BCAA catabolism genes in TNBC was evaluated using the Kaplan-Meier plotter (www.kmplot.com), an online database including gene expression data and clinical data. With the purpose to assess prognostic value of a specific gene, the patient samples were divided into two cohorts according to the median expression of the gene (high vs. low expression). The relapse-free survival (RFS) of TNBC patients was analyzed by using a Kaplan-Meier survival plot. Log rank p-value and hazard ratio (HR) with 95% confidence intervals were calculated. The Affymetrix ID of each gene in this study: BCAT2 (ID203576 at), BCKDK (ID202030 at), PPM1K (ID244011 at).

The individual gene expression level of BCAT2, BCKDK and PPM1K was analyzed by Oncomine. The mRNA levels of cancer vs. normal patient datasets were compared. 1.25fold change, p-value=0.05, and top 10% gene rank were selected as threshold.

Statistical analysis was performed using GraphPad Prism 7.0 Software (GraphPad, San Diego, CA). The results were expressed as mean ± SD. 1-way ANOVA or a two-tailed Student’s t test was performed to analyze the statistically significance. P-values < 0.05 was considered as significant. *p < 0.05; **p < 0 01; ***p < 0.001.

To understand the role of BCAA metabolism in breast cancer in human, we analyzed the BCAA metabolic gene expression in human breast tumors using Oncomine database. Among the genes involved in the initial 2 shared steps, the mRNA expression of BCAT2 and BCKDK was significantly higher, whereas the expression of PPM1K was lower, in the invasive carcinoma than that in normal breast tissue ( Figures 1A–C ). The prognostic value of BCAA metabolic gene mRNA expression in breast cancer was assessed according to relapse-free survival (RFS) using Kaplan–Meier plotter. The Kaplan-Meier plotter revealed that the higher level of BCAT2 was correlated with preferable RFS ( Figure 1D ). Increased expression of BCKDK was correlated with poor outcome in breast cancer patients ( Figure 1E ). Breast cancer patients with up-regulated PPM1K demonstrated better RFS ( Figure 1F ). These results suggest that BCAA metabolic reprogramming is strongly associated with outcome in breast cancer patients.

We investigated the impacts of elevated BCAA on breast tumor progression in Pp2cm-dificient mice in which BCAA accumulated due to the impaired BCAA catabolism ( Figure 2A ) (35). As expected, the plasma levels of BCAA were elevated in Pp2cm-dificient mice, compared with those in the control wildtype mice ( Figure 2B ). We constructed an orthotopic allograft model using Py8119, a kind of murine TNBC cell line (37), in Pp2cm-dificient and control mice. Unexpectedly, the tumor growth was significantly suppressed in Pp2cm knockout mice ( Figure 2C ).

To explore the underling mechanism, we assessed the cell proliferation, cell death, and angiogenesis in tumor tissues in Pp2cm-deficient and control mice. Immunohistochemistry staining showed no significant differences in the expression of proliferation markers Ki67 ( Figure 2D ) and proliferating cell nuclear antigen (PCNA) ( Figure 2G ) and angiogenesis marker CD31 ( Figure 2F ). Unexpectedly, the apoptosis marker cleaved caspase-3 was significantly up-regulated in the tumor in Pp2cm-deficient mice, compared with those in the control wildtype mice ( Figures 2E, G ).

We assessed the direct impacts of high level BCAA on breast cancer cell growth and death in vitro. The concentration of BCAA in Dulbecco’s Modified Eagle Medium (DMEM) is 800 μM, higher than their fasting plasma levels (~50-200 μM) under physiological conditions. Py8119 cells were cultured with different concentrations of BCAA in medium. High level of BCAA did not suppress breast cancer cell growth ( Figure 2H ) or induce cell death ( Figure 2I ). Therefore, the increased cell death in tumors in Pp2cm-deficient mice was unlikely resulted from the direct effects of BCAA on cancer cells. Together, these results suggested high BCAA suppressed TNBC tumor growth by promoting tumor cell death via an indirect mechanism.

Immune system plays an important anti-tumor role in the development of cancer. We then analyzed the immune responses to TNBC tumor in Pp2cm-deficient mice. The populations of macrophage, NK cells, CD4⁺ T cells, CD8⁺ T cells, as well as the population of NK cells expressing interferon γ (IFN-γ) showed no significant changes in the spleens of Pp2cm-deficient mice, compared with those in control mice ( Figures 3A–E ). Among the tumor-infiltrating immune cells, no difference was detected for the populations of macrophage, type-I macrophage, type-II macrophage, CD4⁺ T cells, and CD8⁺ T cells ( Figures 3F–L ). However, the populations of NK cells and NK cells expressing IFN-γ were significantly increased in the tumors in Pp2cm knockout mice ( Figures 3I, J ). These results suggested that high BCAA enhanced the activity of tumor-infiltrating NK cells which could kill cancer cells and inhibit tumor development.

We next investigated whether increasing dietary BCAA intake could affect breast cancer progression in an orthotopic xenograft model. The human breast cancer cell LM2 was injected into the inguinal mammary fat pad of female non-obese diabetic/severe combined immunodeficiency (NOD/SCID) mice. Mice were fed with different BCAA diets for 8 weeks ( Figure 4A ). The plasma concentrations of BCAA in the high-BCAA diet (HBCAA)-fed mice were significantly higher than those in the normal-BCAA diet (NBCAA)-fed group ( Figure 4B ). Importantly, the tumor mass in the HBCAA-fed mice was significantly smaller than that in the NBCAA-fed group ( Figure 4C ). HBCAA significantly inhibited the rate of tumor growth ( Figure 4D ). In agreement with the orthotopic allograft model ( Figures 2E, G ), significantly up-regulated apoptosis marker in the orthotopic xenograft tissues of HBCAA group was observed, compared with that of NBCAA group ( Figures 4E, F ).

In NOD/SCID mice, the adaptive immune system is deficient, accompanied with normal macrophage and low NK cell activity. Macrophage population showed no difference in tumor ( Figure 4G ) or spleen ( Figure 4H ) between HBCAA and NBCAA groups. However, HBCAA feeding increased the population of NK cells ( Figure 4I ) and NK cells expressing granzyme B ( Figure 4J ) in spleen. The populations of tumor-infiltrating NK cells ( Figure 4K ) and NK cells expressing granzyme B ( Figure 4L ) were not significantly changed.

The vast majority of breast cancer deaths are due to metastasis but not the primary tumor (38). To gain insight into the impacts of elevated BCAA on breast cancer metastasis, the tumor metastases in lung tissues was analyzed in the Pp2cm knockout and wildtype mice bearing tumor under inguinal mammary fat pads. Interestingly, the lung metastases in Pp2cm knockout mice were less than those in the control group ( Figures 5A–C ).

In addition to the genetic mouse model in which BCAA was elevated, we also analyzed the impacts of altering BCAA dietary intake on breast cancer metastasis. The tail-vein-metastasis model was established by injecting luciferase-expressing 4T1 cells in wildtype mice, and low BCAA diet significantly promoted lung colonization of breast cancer cells in this model ( Figures 5D–F ). Overall, these results provide evidence that high BCAA inhibits lung metastasis of breast cancer.

To determine whether BCAA directly affected breast cancer cell migration, the transwell assay was conducted with different concentrations of BCAA in culture medium. The results showed that the migration of LM2 was inhibited when BCAA level increased ( Figures 6A, B ). Wound healing assay also showed high BCAA in medium inhibited the migration ability of breast tumor cells ( Figures 6C, D ). In addition, the invasion ability of LM2 was also suppressed by high level of BCAA ( Figures 6E, F ). PP2cm expression is low in LM2 cell. It remains unclear whether inactivation of PP2Cm makes the same impacts on breast cancer cell migration and invasion.

The epithelial-to-mesenchymal transition (EMT) is a biological process strongly associated with tumor progression, invasion, and metastasis (39, 40). As the concentration of BCAA increased in the culture medium, the expression of N-Cadherin, a key mesenchymal marker, dramatically decreased in LM2 cells ( Figure 6G ). The expression of N-Cadherin was also decreased in tumor tissues of HBCAA-fed mice in the orthotopic xenograft model ( Figure 4F ). On the other hand, the expression of E-Cadherin, Slug, and Vimentin showed no difference between different BCAA groups ( Figure 6G ). Furthermore, N-Cadherin expression was down-regulated in the breast tumor tissues in Pp2cm knockout mice compared with that in control mice, but the expression of E-Cadherin and Vimentin was not changed ( Figure 6H ).