58
Introduction
Growth of cancer cells is subjected to multiple signals including autocrine signals from cancer cells and paracrine signals from the surrounding cancer cells as well as various cell types in the local environment (also known as microenvironment).
Accumulating evidence suggests that microenvironment plays a vital role in the initiation and progression of many cancers (Liotta and Kohn, 2001). In breast cancer, cancer-associated fibroblasts (CAFs) and tumor-cancer-associated macrophages (TAMs) promote breast cancer progression and metastasis by secreting growth factors and chemokines.
For example, HGF and SDF-1 secreted from CAFs stimulate breast cancer cell growth through their receptors c-Met and CXCR4, respectively (Orimo et al., 2005; Rong et al., 1992; Tyan et al., 2011). CCL18 secreted from TAMs promotes the metastasis of breast cancer cells (Chen et al., 2011). Deciphering the underlying molecular mechanisms of crosstalk among heterotypic cells is of great interest because new targets for improving diagnosis or novel therapeutic strategies could be identified.
Adipocyte is an abundant cell type in human mammary gland. During breast cancer development, invasion of tumor cells through the basement membrane results in a close interaction between cancer cells and adipocytes (Schaffler et al., 2007). Several clinical observations demonstrate that local invasion of the adipose is correlated with poor prognosis (Kimijima et al., 2000; Yamaguchi et al., 2008). In addition, larger breast size at a young age is also associated with a higher incidence of premenopausal breast cancer (Kusano et al., 2006), suggesting that local adiposity in breast may have a significant impact on breast cancer progression.
Cytokines secreted from adipocytes (also known as adipokines (Vona-Davis and
59
Rose, 2007)) play a role in cancer progression. For instance, leptin and interleukin-6 secreted from adipocytes, promote breast cancer growth and invasion (Dirat et al., 2011;
Surmacz, 2007). The expression of metalloproteinase (MMP)-11/stromelysin-3 in adipocytes is induced by breast cancer cells at the proximity of the invasive front, indicative of a role in the extracellular matrix (ECM) remodeling during breast cancer development (Andarawewa et al., 2005). However, to our knowledge systematic studies to explore the crosstalk between adipocytes and breast cancer cells have not been performed.
We applied co-culture system using breast cancer cells and adipocytes isolated from normal parts of breast cancer mastectomy specimens (called patient-derived mammary gland adipocytes, PMGAs) to identify receptor involved in the crosstalk between cancer cells and adipocytes. Through microarray analyses, monocarboxylate transporter 2 (MCT2) was identified as a new player in PMGAs-mediated promotion of tumorigenic activity. Furthermore, adipocytes secreted β-hydroxybutyrate, an intrinsic HDAC inhibitor, increased histone H3K9 acetylation and induced expression of several tumor-promoting genes including IL1β and LCN2 to enhance tumorigenicity of breast cancer cells expressing MCT2. Consistently, elevated expressions of MCT2 as well as β-hydroxybutyrate-induced genes, IL1β and LCN2, were significantly correlated with poor prognosis in breast cancer patients in two independent cohorts. These findings reveal a mechanism underlying the crosstalk between tumor and its microenvironment, and provide new insights into the connection between adipocytes and breast cancer malignancy.
60
Materials and Methods Ethics statement
Human samples were obtained from National Taiwan University Hospital (NTUH) and Wan Fang Hospital, Taipei Medical University. The samples were encoded to protect patient confidentiality and used under protocols approved by the Institutional Review Board of Human Subjects Research Ethics Committee of National Taiwan University, Taipei, Taiwan (IRB no. 200902001R) and Wan Fang Hospital, Taipei Medical University (IRB no. WFH-IRB-99049). Clinical information was obtained from pathology reports, and the characteristics of these cases are given in Table III-1 and III-2.
Patients with at least 5 years follow-up were included in this study. The animal studies were approved by the Institutional Animal Care and Use Committee of the Academia Sinica, Taipei, Taiwan (Protocol # 14-05-708).
Cell lines
Human breast cancer cell lines MCF7, MB-157, MB-231, MDA-MB-361, MDA-MB-468, and SKBR3 were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10 % fetal bovine serum (FBS), 2 mM L-glutamine, 1 mM non-essential amino acid (NEAA), 1 mM sodium pyruvate and antibiotics/antimycotics in a humidified 37℃ incubator supplemented with 5 % CO2.
Statistical methods
Except for the clinical correlation and quantification for specific immunoblots, all data were presented as means ± SD, and Student’s t test was used to compare control and
61
treatment groups. * indicated statistical significance with P < 0.05. The following analyses were performed using MedCalc statistics software. The association between MCT2 gene expression and survival in breast cancer patients was evaluated using univariate Cox proportional-hazards regression analysis. Multivariate Cox regression analysis was used to adjust the association between survival and MCT2 gene expression level for varying clinical parameters including age, tumor size, lymph-node status, tumor grade and estrogen receptor expression. Hazard ratio was evaluated using the method of Grambsch and Therneau. No violation of the proportional assumption was detected. The optimal cut-off value of MCT2 gene expression level for 5-year survival was determined using ROC (receiver operating characteristic) analysis. Kaplan-Meier method was plotted and log-rank test was used to evaluate the statistical significance between patients with high and low MCT2 expression in survival.
Primary patient-derived mammary gland adipocytes (PMGAs) isolation
Human breast adipose tissue samples were collected from breast cancer mastectomy specimens in NTUH. Adipose tissue pieces were minced and immediately incubated with digestion buffer (Collagenase I (250 U/mL, Sigma) dissolved in PBS containing 2 % bovine serum albumin (BSA, Sigma)) as previously described (van Harmelen et al., 2005). For the incubation, the tubes are closed tightly shacked in a 37°C incubator for 2 hr. Stromal vascular fraction (SVF) was separated by centrifugation (5 min, room temperature, 300g). Primary PMGAs were cultured in DMEM/F12 medium supplemented with 10 % fetal bovine serum (FBS) and antibiotics/antimycotics.
62
Soft agar colony formation assay
Primary PMGAs (10~25 µl) mixed with 0.5 % agar/complete DMEM/F12 growth medium were added into one well of 12-well plate to form a base layer. After the agar was solidified, 1500 breast cancer cells suspended in 0.35 % agar/1 % FBS DMEM medium were seeded on top of bottom agar. Cells were maintained in a humidified 37℃ incubator for 14 days and the colonies were fixed with ethanol containing 0.05 % crystal violet for quantification. For supplement experiments, 50 µl serum-free medium containing various dose of β-hydroxybutyrate, lactate and pyruvate (Sigma) were added every 4 days.
Xenograft assay in NOD/SCID/γnull mice
For tumorigenicity assay, 106 breast cancer cells mixed with PBS or 50 µl primary PMGAs were mixed with equal volume of Matrigel (BD bioscience), and then co-injected into NOD/SCID/γnull mice subcutaneously. Tumor volumes were evaluated every 4 days after initial detection. Student’s t-test was used to test the significant differences. The systemic administration of β-hydroxybutyrate (500 mg/kg/mouse) was performed by daily intra-peritoneal (i.p.) injection, and PBS was injected as control.
Immunoblotting
Immunoblot analysis was performed after 10 % or 15 % SDS-PAGE, with overnight incubation of 1:200 dilution of anti-MCT2 (sc-50322, Santa Cruz), or 1:1000 dilution of anti-ENPP1 (GTX103447, Genetex), anti-acetyl-Histone H3 (Lys9) (#9649, Cell Signaling Technology), anti-Histone H3 (#4499, Cell Signaling Technology)
63
antibodies followed by a 1:10000 dilution of horseradish peroxidase-conjugated anti-mouse or anti-rabbit secondary antibody (GeneTex). Signals were detected using Immobilon Western Chemiluminescent HRP Substrate (Millipore). Protein concentration was determined by the Bradford assay (Bio-Rad) before loading and verified by α-tubulin level using a 1:10000 dilution of anti-α-tubulin antibody (GTX72360, GeneTex). The intensity of western blot bands was quantified using Image J software (NIH).
Immunohistochemistry
Formalin-fixed paraffin embedded primary tumor tissue sections were stained by using the Ventana automated immunostainer BenchMark LT (Ventana Medical Systems).
Homemade mouse anti-MCT2 monoclonal antibody (1 µg/µl) was diluted to 1:200, and incubated at room temperature for 2 hours. Signals were detected using the OptiView DAB IHC Detection Kit (Ventana Medical Systems). All slides were counterstained with hematoxylin, and the images were taken using an Aperio Digital Pathology System.
Samples were identified as MCT2 positive if more than 5 % of the tumor cells were positive for membrane staining.
Co-culture and conditioned medium collection
Conditioned medium from PMGAs or SVF cells were harvested using a Trans-well culture system (0.4-mm pore size; Millipore). 50-100 µl PMGAs or 105 SVF cells were seeded in the top or bottom chamber of the Transwell system (12 well format) in the culture medium of PMGAs/SVF cells. Conditioned medium (CM) was harvested by overnight cultivation in serum-free medium. The fractionated CM was collected using
64
ultrafiltration with different molecular weight cut-off filters (Amicon®, Millipore). For co-culture of PMGAs and breast cancer cells, PMGAs and breast cancer cells were seeded into top and bottom chamber of Transwell (12-well format) in the culture medium of PMGAs and breast cancer cells, respectively.
Plasmids and shRNA
The cDNA of ARMCX1, ENPP1 and MCT2 were amplified from MDA-MB-231 cells and cloned into pLVX-IRES-Neo mammalian expression vector (Clontech) using restriction enzymes and primers listed in Table III-3. The lentiviral shRNA expression vectors of pLKO.1-shLacZ, shABCG2 (mixture of TRCN59798-59802), shAREG (TRCN117993, 117995, and 117996), shARMCX1 (TRCN133686, 138265, 160564, 160630-160631, and 161844), shATP2B1 (TRCN43068-43071), shENPP1 (TRCN2537 and 2540), shFRAS1 (TRCN73564, 73566 and 73567), shFRMD5 (TRCN128657 and 131096), shGPC6 (TRCN123094-123098), shGPR126 (TRCN11562, 273811 and 273865), shIL6R (TRCN372671, 372728, and 378748), shOSBPL8 (TRCN281229), shRAB8B (TRCN379718, 380132, 380248, 380390 and 382238), shRFTN1 (TRCN130134, 130387, and 130642), shRHOF (TRCN48050-48051), shMCT2 (TRCN38508, 286881 and 294347), shTHBD (TRCN53923-53927 and 174207) were purchased from the National RNAi Core Facility (Taipei, Taiwan). For lentivirus production, HEK-293T cells were cotransfected with shRNA containing lentiviral vector, envelope plasmid pMD.G and packaging plasmid pCMVΔR8.91. Virus containing supernatant was harvested at 24 and 48 h post transfection. 157, MDA-MB-231 and MCF7 cells were infected with lentivirus and then selected with 2 µg/ml
65
puromycin.
RNA isolation, reverse-transcription, (real-time) PCR assay, and gene expression using microarray analysis
Total RNA from cell culture and tumor tissue was isolated using TRIzol reagent (Invitrogen) and reverse-transcribed with Transcriptor First Strand cDNA Synthesis kit (Roche) for gene expression analysis according to instructions from the manufacturers.
Quantitative real-time RT–PCR was performed using KAPA SYBR FAST qPCR kit (KAPA Biosystems) for gene expression according to the manufacturer’s instruction and analyzed on a StepOnePlus Real-Time PCR system (Applied Biosystems). GAPDH mRNA was used as an internal control for mRNA expression. Expression levels were calculated according to the relative ΔCt method, and the specificity of each primer pairs were determined by dissociation curve analysis. All primers are listed in Table III-4.
Affymetrix U133 Plus 2.0 human oligonucleotide microarrays (Phalanx Biotech Group) were used to detect gene expression.
Measurement of β-hydroxybutyrate, lactate and pyruvate
The measurements of β-hydroxybutyrate, lactate and pyruvate levels in conditioned medium were performed using ELISA colorimetric assay kits (BioVision) according to instructions from the manufacturer. The levels of β-hydroxybutyrate, lactate and pyruvate were normalized with protein concentration.
66
Results
Patient-derived mammary gland adipocytes (PMGAs) isolated from mammary glands enhance malignancy of distinct subtypes of breast cancer cells
To address whether PMGAs from mammary glands have a role in promoting tumor malignancy, a panel of breast cancer cell lines including estrogen receptor (ER) positive (MCF7 and MB-361) and ER negative (MB-231, MB-157, MDA-MB-468 and SK-BR3) were used for soft agar colony formation assay with or without PMGAs co-culture in the bottom agar layer, respectively. In this co-culture system, PMGAs and breast cancer cells were separated, but crosstalk between them through diffused soluble factors could occur. The human primary PMGAs were isolated from normal parts of the breast cancer mastectomy specimens according to protocols referred in (van Harmelen et al., 2005) (Fig. III-1A and 1B). As shown in Fig. III-1C, PMGAs promoted colony formation of several breast cancer cell lines, but not MDA-MB-468 and SK-BR3 cells. There were no correlations with the ERα status of breast cancer cells. This activity was not conferred by patient-specific PMGAs, because similar results were also observed using two independent batches of PMGAs isolated from different breast cancer patients. The colony number increased when greater numbers of PMGAs were present in the co-culture system (Fig. III-2A and 2B). Next, we used a xenograft tumor model to examine the tumorigenicity of breast cancer cells co-cultured with PMGAs. To avoid potential interference by adipocytes in the mouse fat pad, we co-implanted human PMGAs and breast cancer cells subcutaneously. As shown in Fig. III-3A and 3B, the breast cancer cells mixed with PMGAs grew faster and also had greater tumor weights than the control groups. However, there was little or no difference in
67
MDA-MB-468 cells either in vitro or in vivo (Fig. III-1C and 3C). Taking together these data suggested that promotion of breast cancer malignancy by PMGAs was selective depending on breast cancer cell lines and the effects are likely mediated by soluble factors secreted from the PMGAs.
Identification of membrane proteins in cancer cells involved in the promotion of tumorigenicity by PMGAs
To explore potential candidates in tumor cells that are involved in PMGAs-mediated tumorigenic enhancement, we compared the mRNA expression profiles among MDA-MB-231, MDA-MB-361, MDA-MB-157 and MDA-MB-468 cells using human cDNA microarrays (Fig. III-4A). The genes up-regulated two folds or greater in MDA-MB-231, MDA-MB-361, and MDA-MB-157 when compared to MDA-MB-468, were selected as candidate genes. Among the 253 genes up-regulated in MDA-MB-231, MDA-MB-361, and MDA-MB-157, 16 genes encoded membrane or membrane-associated proteins (Fig.
III-4B). Q-PCR analysis confirmed that the expression profiles of these genes were similar to that of the results from cDNA microarray (Fig. III-5). Next, the expression of these 16 candidate genes in the MDA-MB-157 cells were depleted by pooled RNAi knockdown to evaluate their roles in PMGAs-mediated tumorigenicity enhancement (Fig. III-6). Depletion of ARMCX1, ENPP1, FRMD5, GPC6, GPR126, RFTN1, and MCT2, respectively, abrogated the colony increment induced by co-culture with PMGAs (Fig. III-4C). However, only ARMCX1, ENPP1 and MCT2 showed a better correlation between phenotype and expression levels as high expressions of these three genes were detected in MCF7, MDA-MB-361, MDA-MB-231 and MDA-MB-157 cells, to which
68
PMGAs had a promoting activity, while low expressions were found in MDA-MB-468 and SKBR3 cells, to which PMGAs co-culture had no significant effects (Fig. III-4C and 4D).
To further verify that ARMCX1, ENPP1, and MCT2 contributed to PMGAs-mediated effects, we depleted ARMCX1, ENPP1 and MCT2 expression in two additional cell lines MCF7 and MDA-MB-231. Similar results were also observed (Fig.
III-7A). Furthermore, we ectopically expressed these genes in MDA-MB-468 and SK-BR3 cells, two breast cancer cells whose colony growth was not enhanced by PMGAs, only overexpression of MCT2, not ARMCX1 nor ENPP1 enhanced colony formation when co-cultured with PMGAs (Fig. III-7B). In addition, subcutaneous co-injection of MCT2-depleted MDA-MB-231 cells or MCT2-overexpressing MDA-MB-468 cells with PMGAs in NOD/SCID/γnull mice reduced or accelerated tumor growth compared to the control groups, respectively (Fig. III-7C and 7D). Taken together, these results derived from loss- and gain-of-function analyses suggest that MCT2 in breast cancer cells mediates the tumorigenic effects when co-cultured with PMGAs.
β-hydroxybutyrate secreted from PMGAs enhances breast cancer malignancy Based on the design of our in vitro co-culture system, the effects from PMGAs are most likely mediated by secreted factors. Adipocytes are known to produce many secretory factors, known as adipokines (Vona-Davis and Rose, 2007), however, a systematic analysis of factors secreted from PMGAs was required. We collected PMGAs-conditioned medium and fractioned secreted factors into <10 kD, 10-50 kD and
>50 kD fractions via selected filters and centrifugation. Only the <10 kD fractionated
69
conditioned medium stimulated colony formation when compared to cell culture medium (not PMGAs’ conditioned medium) in the soft agar colony formation assay using MDA-MB-231 and MCF7 cells (Fig. III-8A). However, treatment with <10 kD fractionated conditioned medium did not enhance colony formation of MDA-MB-468, consistent with our previous observation that MDA-MB-468 is a non-responsive cell line in the PMGAs co-culture system. Furthermore, depletion or ectopic expression of MCT2 in MDA-MB-231, MCF7 and MDA-MB-468 cells, respectively, either abrogated or increased the response to <10 kD fractionated conditioned medium treatment (Fig. III-8B), suggesting MCT2 plays an important role in mediating the effects of the conditioned medium.
MCT2 is a monocarboxylate transporter, which has been shown to transport monocarboxylic acid, such as pyruvate, lactate and β-hydroxybutyrate from extracellular space into cells (Broer et al., 1999; Lin et al., 1998). Based on the observation that only
<10 kD fractionated conditioned medium promoted colony formation in a MCT2-dependent manner and the biological function of MCT2 in metabolites transport, we hypothesized that these small molecules pyruvate, lactate and β-hydroxybutyrate might play roles in PMGAs-mediated tumorigenicity.
To evaluate the effect of pyruvate, lactate and β-hydroxybutyrate on tumorigenicity, we first determined the levels of secreted pyruvate, lactate and β-hydroxybutyrate from PMGAs and corresponding stromal vascular fraction (SVF) cells, the two major cell types present in the human breast adipose tissue. Although pyruvate, lactate and β-hydroxybutyrate were detected in conditioned medium harvested from both PMGAs and SVF cells, the levels of β-hydroxybutyrate were higher in the PMGAs’ conditioned
70
medium compared to corresponding SVF cells’ (Fig. III-8C). To directly test the potential roles of pyruvate, lactate and β-hydroxybutyrate in tumorigenesis, these small molecule supplements were added to knockdown MDA-MB-231 and MCT2-overexpressing MDA-MB-468 cells, respectively, in colony formation assay.
Interestingly, supplement with β-hydroxybutyrate and lactate could either effectively or partially enhance colony formation in a MCT2 in a dose dependent manner, but supplement of pyruvate in addition to the basal level of 1 mM pyruvate in culture medium did not affect colony formation (Fig. III-9A and 9B). Since deprivation of pyruvate dramatically abolished the colony formation regardless of the status of MCT2 expression, it was less likely pyruvate played a key role in PMGAs-mediated colony promotion (Fig. III-10). Consistently, administration of β-hydroxybutyrate via daily intra-peritoneal (i.p.) injection in xenograft models promoted tumor growth of MDA-MB-231 and MDA-MB-157 cells, but not their MCT2-depleted counterparts (Fig. III-9C). These results support the notion that β-hydroxybutyrate promotes breast cancer tumorigenisis though MCT2.
β-hydroxybutyrate induces changes in gene expression through epigenetic modifications
β-hydroxybutyrate has been identified as an endogenous inhibitor of class I histone deacetylase (HDAC) recently (Shimazu et al., 2013). Treatment with β-hydroxybutyrate increased the global histone acetylation, such as histone H3 Lys 9 and Lys14 (H3K9 and H3K14) acetylation, which in turn results in gene activation. Therefore, it is likely that MCT2 plays some roles in β-hydroxybutyrate-mediated global histone acetylation.
71
Indeed, upon treatment with β-hydroxybutyrate, the level of H3K9 acetylation was enhanced in a dose- and time-dependent manner in the presence of MCT2 in MDA-MB-231, MCF7 as well as MDA-MB-157 cells (Fig. III-11A and 11B). Consistently, co-culture with PMGAs also increased the acetylation of H3K9 (Fig. III-11C), implicating that the tumorigenic effect by β-hydroxybutyrate is likely through epigenetic modifications of histone followed by up-regulation of expression of tumor-promoting genes.
To identify β-hydroxybutyrate-induced genes that affect tumor cells phenotypes, MDA-MB-231 cells cultured in the presence of β-hydroxybutyrate were subjected to microarray analyses. A total of 34 genes were found to be significantly upregulated upon β-hydroxybutyrate treatment (Fig. III-12A). By performing gene ontology analysis of each gene using the Biological Process (BP) GO term, these genes were classified into six groups accordingly (Fig. III-12B). Intriguingly, most genes on the list were highly expressed in breast cancers compared to normal mammary tissue (Data was adapted from Oncomine database, https://www.oncomine.org/resource/login.html) (Table. III-5).
The genes belonging to “Cell growth and proliferation” and “Apoptosis related pathway”
were selected for further validation by Q-PCR analysis. IL1β and LCN2 were significantly increased upon β-hydroxybutyrate treatment (Fig. III-12C). Moreover, co-culture with PMGAs also induced the expression of IL1β and LCN2 (Fig. III-12D).
Taken together, these results suggest that tumorigenesis promotion by β-hydroxybutyrate is likely resulted from up-regulation of tumor-promoting genes through epigenetic modification of chromatin.
72
Elevated MCT2 levels correlate with poor prognosis in breast cancer patients
To further affirm the significance of MCT2 in breast cancer tumorigenesis, RNA
To further affirm the significance of MCT2 in breast cancer tumorigenesis, RNA