Aerobic glycolysis in cancer associated fibroblasts ...

This article was downloaded by: [111.47.206.15] On: 26 August 2015, At: 21:21 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5 Howick Place, London, SW1P 1WG

Cell Cycle Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/kccy20

The reverse Warburg effect: Aerobic glycolysis in cancer associated fibroblasts and the tumor stroma Stephanos Pavlides, Diana Whitaker-Menezes, Remedios Castello-Cros, Neal Flomenberg, Agnieszka K. Witkiewicz, Philippe G. Frank, Mathew C. Casimiro, Chenguang Wang, Paolo Fortina, Sankar Addya, Richard G. Pestell, Ubaldo E. Martinez-Outschoorn, Federica Sotgia & Michael P. Lisanti Published online: 01 Dec 2009.

To cite this article: Stephanos Pavlides, Diana Whitaker-Menezes, Remedios Castello-Cros, Neal Flomenberg, Agnieszka K. Witkiewicz, Philippe G. Frank, Mathew C. Casimiro, Chenguang Wang, Paolo Fortina, Sankar Addya, Richard G. Pestell, Ubaldo E. Martinez-Outschoorn, Federica Sotgia & Michael P. Lisanti (2009) The reverse Warburg effect: Aerobic glycolysis in cancer associated fibroblasts and the tumor stroma, Cell Cycle, 8:23, 3984-4001, DOI: 10.4161/cc.8.23.10238 To link to this article: http://dx.doi.org/10.4161/cc.8.23.10238

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Versions of published Taylor & Francis and Routledge Open articles and Taylor & Francis and Routledge Open Select articles posted to institutional or subject repositories or any other third-party website are without warranty from Taylor & Francis of any kind, either expressed or implied, including, but not limited to, warranties of merchantability, fitness for a particular purpose, or non-infringement. Any opinions and views expressed in this article are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor & Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions It is essential that you check the license status of any given Open and Open Select article to confirm conditions of access and use.

Report

Cell Cycle 8:23, 3984-4001; December 1, 2009; © 2009 Landes Bioscience

The reverse Warburg effect Aerobic glycolysis in cancer associated fibroblasts and the tumor stroma Stephanos Pavlides,1,2,† Diana Whitaker-Menezes,1,2,† Remedios Castello-Cros,1,2 Neal Flomenberg,2,3 Agnieszka K. Witkiewicz,2,4 Philippe G. Frank,1,2 Mathew C. Casimiro,1,2 Chenguang Wang,1,2 Paolo Fortina,1,2 Sankar Addya,1,2 Richard G. Pestell,1,2 Ubaldo E. Martinez-Outschoorn,2,3 Federica Sotgia1,2,* and Michael P. Lisanti1-3,* 3

1 Departments of Stem Cell Biology & Regenerative Medicine, and Cancer Biology; 2The Jefferson Stem Cell Biology and Regenerative Medicine Center; Department of Medical Oncology; and 4Department of Pathology, Anatomy & Cell Biology; Kimmel Cancer Center; Thomas Jefferson University; Philadelphia, PA USA

These authors contributed equally to this work.

†

Downloaded by [111.47.206.15] at 21:21 26 August 2015

Keywords: caveolin-1, tumor stroma, myo-fibroblast, cancer-associated fibroblast, aerobic glycolysis, M2-isoform of pyruvate kinase, lactate dehydrogenase, Warburg effect

Here, we propose a new model for understanding the Warburg effect in tumor metabolism. Our hypothesis is that epithelial cancer cells induce the Warburg effect (aerobic glycolysis) in neighboring stromal fibroblasts. These cancer-associated fibroblasts, then undergo myo-fibroblastic differentiation, and secrete lactate and pyruvate (energy metabolites resulting from aerobic glycolysis). Epithelial cancer cells could then take up these energy-rich metabolites and use them in the mitochondrial TCA cycle, thereby promoting efficient energy production (ATP generation via oxidative phosphorylation), resulting in a higher proliferative capacity. In this alternative model of tumorigenesis, the epithelial cancer cells instruct the normal stroma to transform into a wound-healing stroma, providing the necessary energy-rich micro-environment for facilitating tumor growth and angiogenesis. In essence, the fibroblastic tumor stroma would directly feed the epithelial cancer cells, in a type of host-parasite relationship. We have termed this new idea the “Reverse Warburg Effect.” In this scenario, the epithelial tumor cells “corrupt” the normal stroma, turning it into a factory for the production of energyrich metabolites. This alternative model is still consistent with Warburg’s original observation that tumors show a metabolic shift towards aerobic glycolysis. In support of this idea, unbiased proteomic analysis and transcriptional profiling of a new model of cancer-associated fibroblasts [caveolin-1 (Cav-1) deficient stromal cells], shows the upregulation of both (1) myo-fibroblast markers and (2) glycolytic enzymes, under normoxic conditions. We validated the expression of these proteins in the fibroblastic stroma of human breast cancer tissues that lack stromal Cav-1. Importantly, a loss of stromal Cav-1 in human breast cancers is associated with tumor recurrence, metastasis, and poor clinical outcome. Thus, an absence of stromal Cav-1 may be a biomarker for the “Reverse Warburg Effect,” explaining its powerful predictive value.

Activated myo-fibroblasts are critical for normal wound healing and are generated via the TGFb mediated differentiation of normal fibroblasts.1 However, they have also been implicated in a number of human diseases related to fibrosis, such as systemic sclerosis, interstitial pulmonary fibrosis, as well as renal fibrosis following ischemia. Fibrosis and collagen deposition are also known risk factors for malignancy, and it is, therefore, not surprising that cancer-associated fibroblasts share many characteristics with activated myo-fibroblasts.1 In fact, cancer has been referred to by many as a “wound that does not heal.”1 As such, the tumor micro-environment has been postulated to play a key role in tumor initiation, progression, and metastasis. In this regard, cancer-associated fibroblasts may be thought of as activated myofibroblasts that cannot regress to the unactivated state. They are stuck in the “on position.”

Thus, one way to create a model of cancer-associated fibroblasts might be to cause constitutive activation of TGF signaling. To test this hypothesis directly, we examined the phenotypic behavior of fibroblasts derived from mice that specifically lack a known inhibitor of TGFb signaling, namely caveolin-1 (Cav-1). Cav-1 is known to function as an inhibitor of the TGFb type I receptor kinase.2 In accordance with this hypothesis, Cav-1 (-/-) null mice are prone to the development of fibrotic disease, and Cav-1 (-/-) null skin fibroblasts share characteristics with scleroderma fibroblasts,3,4 which also exhibit a constitutive myofibroblastic phenotype. Thus, a loss of Cav-1 expression may be sufficient to induce a constitutive myo-fibroblastic phenotype. In accordance with this hypothesis, we previously demonstrated that Cav-1 expression is dramatically downregulated in human breast cancer-associated fibroblasts, relative to matched

*Correspondence to: Federica Sotgia and Michael P. Lisanti; Email: [email protected] and [email protected] Submitted: 10/02/09; Accepted: 10/05/09 Previously published online: www.landesbioscience.com/journals/cc/article/10238 3984

Cell Cycle

Volume 8 Issue 23

Report


Report

normal mammary fibroblasts isolated from the same patients.5 Thus, we postulated that a loss of Cav-1 in fibroblasts is a new marker of the cancer-associated fibroblast phenotype. To test this hypothesis more directly and to establish a cause-effect relationship, we studied the phenotype of Cav-1 (-/-) null mammary stromal fibroblasts in culture.6 Specifically, we showed that genetic ablation of Cav-1 appeared to be sufficient to confer the cancer-associated fibroblast phenotype. Cav-1 (-/-) mammary stromal fibroblasts behaved as myo-fibroblasts, exhibiting: (1) contraction/retraction; (2) the upregulation of musclerelated genes; and (3) the upregulation of TGFb ligands, related factors, and responsive genes (TGFb-2/3, procollagen genes, interleukin-11, and CTGF).6 Thus, a loss of Cav-1 in fibroblasts appears to drive the onset of the myo-fibroblastic differentiation program. Finally, unbiased genome-wide expression profiling of Cav-1 (-/-) mammary stromal fibroblasts showed significant transcriptional overlap with human breast cancerassociated fibroblasts.6 Thus, Cav-1 (-/-) mammary fibroblasts provide the first cell culture model for breast cancer-associated fibroblasts.6 Since a loss of Cav-1 is associated with and can confer the cancer-associated fibroblast phenotype, we speculated that it may have clinical relevance as a biomarker.7 To test this hypothesis directly, we immuno-stained a well-annotated breast cancer tumor micro-array containing 160 consecutive patients.8 Our results indicated that an absence of stromal Cav-1 was specifically associated with a high rate of tumor recurrence, metastasis and tamoxifen-resistance, resulting in poor clinical outcome.8 An absence of stromal Cav-1 had predictive value that was independent of epithelial marker status, indicating that it was an effective biomarker for all the major subclasses of breast cancer, including ER+, PR+, HER2 +, and triple-negative patients (ER- /PR- /HER2-).8 An absence of stromal Cav-1 expression was most predictive in lymph-node positive patients, where it showed an 11.5-fold stratification of recurrence-free survival at 5 years (Cav-1(+), 80% survival versus Cav-1(-), 7% survival).8 Similar results were also obtained with DCIS patients, indicating that a loss of stromal Cav-1 may be linked both to tumor initiation and progression, as well as metastasis.9 DCIS patients lacking stromal Cav-1 had an overall recurrence rate of 100% and 80% of these patients underwent progression to invasive breast cancer.9 In prostate cancer patients, an absence of stromal Cav-1 was specifically associated with advanced prostate cancer and metastatic disease,10 indicating that a loss of stromal Cav-1 may have predictive value in multiple forms of human cancer. The above results related to breast cancer metastasis and survival have also now been independently validated in a second unrelated patient cohort.11 Based on the above mechanistic and clinical data, it is now clear that Cav-1 (-/-) null mice are a new valuable resource for studying how myo-fibroblasts/cancer-associated fibroblasts contribute to tumor initiation, progression, and metastasis. For example, Cav-1 (-/-) stromal cells could be used to test new therapeutic strategies that target the tumor micro-environment. Here, we have used stromal cells derived from Cav-1 (-/-) null mice as an engine to drive new stromal biomarker discovery. In order to identify which proteins are selectively upregulated by

www.landesbioscience.com

an absence of Cav-1, we subjected Cav-1 (-/-) stromal cells to extensive unbiased proteomic analysis and genome-wide transcriptional profiling. Results from this screening approach were then validated by the immuno-staining of human breast cancer samples that lack stromal Cav-1 expression. Using a proteomic approach, we have now identified >25 candidate stromal biomarkers that are upregulated by a loss of Cav-1 in stromal cells. Interestingly, these proteins include five myofibroblast markers, three signaling molecules, one oncogene, eight metabolic and glycolytic enzymes, as well as three extra-cellular matrix proteins—known to be associated with fibrosis and tumorigenesis. The eight glycolytic enzymes include the M2-isoform of pyruvate kinase and lactate dehydrogenase (Ldha), which are key regulators known to mediate the Warburg effect.12,13 Two markers of oxidative stress were also upregulated, suggesting the over-production of reactive oxygen species (ROS) in Cav-1 deficient stromal cells. Based on these findings, we predict that a loss of stromal Cav-1 may be a novel biomarker for the Warburg effect (aerobic glycolysis) in the tumor stromal compartment. We have termed this new idea the “Reverse Warburg Effect” because it suggests that aerobic glycolysis may take place in the “fibroblastic” tumor stromal compartment, rather than in the epithelial cancer cells themselves. Interestingly, previous metabolic studies with skin myo-fibroblasts have clearly shown that they perform “aerobic glycolysis,” with increased glucose utilization and lactate production/secretion,14 suggesting that the “Reverse Warburg Effect” may be a general feature of both myo-fibroblasts and cancer-associated fibroblasts (CAFs). Results Proteomic analysis of Cav-1 (-/-) null stromal cell lysates: Identification of new potential tumor stromal biomarkers. Since a loss of Cav-1 protein expression in the breast tumor stroma, in both DCIS and breast cancer patients, is predictive of poor clinical outcome,7-9 we hypothesized that molecules that are upregulated in the absence of stromal Cav-1 may be novel biomarkers. To identify potential new biomarkers, we subjected Cav-1 (-/-) stromal cell lysates to extensive unbiased proteomic analysis. For this purpose, we used primary cultures of bone-marrow derived stromal cells (BMSCs) from WT and Cav-1 (-/-) null mice. We chose to use BMSCs, as CAFs are thought to originate from mesenchymal stem cells of the bone marrow.15-17 Two-dimensional separation of WT and Cav-1 (-/-) stromal cell lysates yielded at least 60 protein spots which were differentially expressed. We used mass spec/protein micro-sequencing to determine the molecular identity of 22 protein spots that were specifically upregulated. See Supplemental Figure 1 for representative 2-D gel analysis. Interestingly, five of the protein spots that were upregulated we identified as known markers of the myo-fibroblast and/or cancer-associated fibroblast (CAF) phenotype (vimentin, calponin2, tropomyosin, gelsolin and prolyl 4-hydroxylase alpha) by mass spectrometry analysis (Table 1).

Cell Cycle

3985


Figure 1. Genetic ablation of Cav-1 in stromal cells upregulates eight metabolic enzymes associated with the glycolytic pathway. Enzymes and metabolites that are part of glycolysis, the pentose phosphate pathway, fatty acid synthesis, triglyceride synthesis, lactose synthesis and the TCA cycle are shown. Note that the protein spots identified by proteomics analysis/mass spec (listed in Table 1) are eight enzymes associated with the glycolytic pathway (highlighted in pink). This list includes pyruvate kinase, a key enzyme which is sufficient to mediate the Warburg effect, driving aerobic glycolysis in tumors. This diagram was modified (with permission) from Beddek et al. Proteomics 2008; 8:1502–15, an article on the proteomics of the lactating mammary gland, and is based on the KEGG pathway database (www.genome.jp/kegg/pathway.html).

3986

Cell Cycle

Volume 8 Issue 23

Table 1. Proteomic analysis: identification of cellular proteins upregulated in Cav-1 (-/-) bone marrow stromal cells Fold change (KO/WT)

Accession number

Protein spot number

2.21

gi|148676699

7

Myo-fibroblast associated proteins gelsolin calponin 2

1.83

gi|6680952

44

tropomyosin 2; beta

1.86

gi|123227997

40

vimentin

1.82

gi|31982755

30

vimentin

1.75

gi|2078001

22

vimentin

1.7

gi|31982755

33

prolyl 4-hydroxylase alpha(I)-subunit (P4HA1)

1.7

gi|836898

11 38

Signaling molecules annexin A1

2.0

gi|124517663

annexin A2

1.86

gi|6996913

39

Rho, GDP dissociation inhibitor (GDI) beta

1.67

gi|33563236

50

M2-type pyruvate kinase

2.78

gi|1405933

15


Glycolytic and metabolic enzymes phosphoglycerate kinase 1

2.41

gi|70778976

31

lactate dehydrogenase A (Ldha)

2.11

gi|13529599

43

fructose-bisphosphate aldolase A

1.87

gi|6671539

32

glycerol 3-phosphate dehydrogenase 2

1.83

gi|224922803

12

enolase 1 (Eno1)

1.77

gi|34784434

24

triosephosphate isomerase 1

1.7

gi|6678413

57


1.65

gi|6678413

58

phosphoglycerate mutase 1

1.65

gi|10179944

54

1.94

gi|13124192

41

peroxiredoxin 1

1.88

gi|6754976

60

catalase

1.74

gi|442441

17

Oncogenes elongation factor 1-delta; EF-1-delta Anti-oxidants associated with oxidative stress

All peptide sequences used for protein identification correspond to the Mus musculus protein product, ruling out contamination by serum proteins.

Also, several signaling molecules were overexpressed, such as the annexins (A1 and A2), as well as RhoGDI. One of the proteins we identified is a known oncogene, Elongation factor 1-delta (EF-1-delta) (Table 1). Overexpression of EF-1-delta is sufficient to drive cell transformation and tumorigenesis.18,19 Perhaps, most importantly, a loss of Cav-1 resulted in the upregulation of eight glycolytic/metabolic enzymes, including the M2-isoform of pyruvate kinase (Table 1). The M2-isoform of pyruvate kinase is generated by gene splicing and is known to be sufficient to confer the “Warburg Effect”, i.e., aerobic glycolysis, which is thought to be a characteristic of tumor cells, stem cells, and cancer stem cells.12,13 The M1 isoform is the “adult” isoform, while the M2 isoform is the corresponding “embryonic or developmental” isoform, both generated by alternate splicing from the same gene. The M2-isoform of pyruvate kinase can also act a nuclear co-factor to stimulate the transcriptional effects of Oct4, an iPS transcription factor that confers pluripotency in ES cells.20 Figure 1 shows that all 8 of these enzymes sequentially map to the glycolytic pathway, which should result in the over-production of the metabolites pyruvate and lactate, which could then be secreted into the medium to “feed” adjacent tumor cells. Thus,


for the first time, we have now identified that the Warburg Effect can originate in the tumor stroma. Two anti-oxidant markers of oxidative stress (peroxiredoxin 1 and catalase) were also upregulated, suggesting the over-production of reactive oxygen species (ROS) in Cav-1 deficient stromal cells, under normoxic conditions. Proteomic analysis of Cav-1 (-/-) null stromal cell “conditioned media.” To identify secreted proteins that are upregulated in Cav-1 (-/-) stromal cells, we also subjected “conditioned media” from bone-marrow derived stromal cells (BMSCs) to proteomic analysis. Two-dimensional separation of WT and Cav-1 (-/-) stromal cell “conditioned media” yielded greater than 60 protein spots which were differentially expressed. We used mass spec/ protein micro-sequencing to determine the molecular identity of 8 protein spots that were specifically upregulated. Our results are summarized in Table 2. See Supplemental Figure 2 for representative 2-D gel analysis. Three of the protein spots that were upregulated we identified as extracellular matrix proteins [collagen I (Col1a1 and Col1a2)] and SPARC), that are known to be associated with fibrosis, the myofibroblast phenotype, and the tumor-associated stroma (Table 2).

Cell Cycle

3987

Table 2. Proteomic analysis: Identification of secreted proteins upregulated in Cav-1 (-/-) stromal cell “conditioned media” Fold change (KO/WT)

Accession number

Protein spot number

Extracellular matrix proteins Col1a1 protein

3.92

gi|13096810

23

Col1a2 protein

2.88

gi|15214623

25

SPARC (secreted acidic cysteine rich glycoprotein)

2.59

gi|148701546

14

albumin

5.77

gi|163310765

53

Albumin and alpha-fetoprotein: liver-specific secreted proteins alpha-fetoprotein

4.39

gi|191765

43

alpha-fetoprotein

4.24

gi|191765

17

alpha-fetoprotein

2.42

gi|191765

4

3.0

gi|74151816

58

Other unnamed protein product (Glutaredoxin (GRX) family, SH3BGR (SH3 domain binding glutamic acid-rich protein) subfamily)


All peptide sequences used for protein identification correspond to the Mus musculus protein product, ruling out contamination by serum proteins.

Interestingly, two liver-specific secreted proteins were also upregulated, such as albumin and alpha-fetoprotein (Table 2). This is consistent with the idea that Cav-1 (-/-) null mesenchymal stem cells may have an increased tendency towards liver-specific differentiation.21 Using in situ hybridization, alpha-fetoprotein expression has been previously localized selectively to myo-fibroblasts within the tumor stromal compartment in human breast cancers, but not to the corresponding epithelial cancer cells.22 Validation of stromal marker expression using human breast cancer tissues. We have now identified 19 candidate cell-based biomarkers whose levels are upregulated in response to a loss of Cav-1 in stromal cells (Table 1). These include five myofibroblast markers (such as vimentin), one oncogene (EF-1-delta), three signaling molecules (annexins A1, A2 and RhoGDI), eight glycolytic enzymes (including M2-pyruvate kinase) and two antioxidant molecules related to oxidative stress (peroxiredoxin 1 and catalase). Similarly, we identified six secreted proteins that are upregulated in Cav-1 negative stromal cells (Table 2). For example, Cav-1 negative stromal cells showed the overexpression of both collagen I (Col1a1 and Col1a2) and SPARC. Liver-specific secretory proteins were also overexpressed: albumin and alpha-fetoprotein. To validate the potential clinical relevance of these candidate stromal biomarkers for human breast cancer, we immunostained sections from human breast cancer tumor tissues that were pre-selected based on a lack of stromal expression of Cav-1. Figure 2 illustrates that stromal Cav-1 is absent or greatly dimished in these samples (with the exception of the vasculature), and that vimentin is greatly increased. Using this immuno-histochemical approach, we also validated the tumor stromal expression of other myo-fibroblast markers (calponin and P4HA1), the annexins (A1 and A2), an oncogene (EF-1-delta), two key glycolytic enzymes (M2-pyruvate kinase and lactate dehyrogenase), and an extracellular matrix protein (SPARC) (Fig. 3–7). Importantly, M2-pyruvate kinase, the key glycolytic enzyme involved in conferring the “Warburg Effect” was abundantly

3988

expressed in the tumor stromal compartment, using isoform-specific antibody probes that specifically recognize the M2-isoform. Furthermore, our proteomic studies using Cav-1 (-/-) null stromal cells detected a peptide specific for the murine M2-isoform (EAEAAIYHLQLFEELRR), but not the corresponding peptide expected from the M1-isoform. Since annexin A2 and SPARC expression are associated with tenascin C overexpression in the extracellular matrix,23-25 we also examined the expression of tenascin C in breast cancer tissues lacking Cav-1 expression. Interestingly, Figure 8 shows the robust expression of tenascin C in the tumor stroma, as predicted. Validation of stromal marker expression using transcriptional gene profiling. In order to independently assess the expression profiles of the gene products that are differentially expressed in WT and Cav-1 (-/-) stromal cells, we also performed gene expression studies using transcriptome analysis. Briefly, mRNA species isolated from WT and Cav-1 (-/-) stromal cells were subjected to transcriptional profiling using the Affymetrix GeneChip® Mouse Exon 1.0 ST array. Our results are summarized in Tables 3–5. Table 3 shows that many of the cellular proteins that we identified as upregulated in Cav-1 (-/-) stromal cells by proteomics were also transcriptionally upregulated by gene profiling. These include myo-fibroblast markers (Gsn, Cnn2, Tpm2, Vim, P4ha1), signaling molecules (Anxa2, Arhgdib), glycolytic enzymes (Pkm2, Pgk1, Ldhal6b, Aldoa, Gpd2, Eno1, Tpi1, Pgam1), oncogenes (Eef1d) and antioxidants (Cat). Also, additional glycolytic enzymes that were transcriptionally upregulated are included in this analysis. Similarly, Table 4 shows that the secreted proteins that we identified as upregulated in Cav-1 (-/-) stromal cells by proteomics were also transcriptionally upregulated, such as extracellular matrix proteins (Col1a1, Col1a2, Spock1), and liver-specific proteins (Alb, Afp). We also examined the expression of other functionally related gene products. Table 5 shows that TGFb receptor signaling molecules, glucose and lactate transporters, cancer-associated fibroblast markers, muscle-related genes, and

Cell Cycle

Volume 8 Issue 23


Figure 2. Vimentin is highly overexpressed in the stroma of human breast cancers that lack stromal Cav-1 expression. We selected a number of human breast cancer cases that lack stromal Cav-1 expression for validating the stromal expression of new biomarkers. The upper panel shows a representative example of a loss of Cav-1 stromal staining; however, note that the endothelial vasculature still remains Cav-1 positive (see arrows), as we have previously documented. Note that sections from the same tumor show the overexpression of vimentin in the tumor stromal compartment. These results directly demonstrate the feasibility and success of our proteomics analysis. Original magnification, 40X.

complement regulatory proteins, were all upregulated in Cav-1 (-/-) stromal cells. The upregulation of TGFb receptor related signaling molecules, cancer-associated fibroblast markers, and muscle-specific genes is consistent with the onset of a myo-fibroblastic phenotype. Furthermore, the upregulation of glucose transporters, glycolytic enzymes, and lactate transporters is consistent with a shift towards aerobic glycolysis, i.e., the “Warburg Effect.” Finally, the overexpression of complement regulatory genes is in accordance with the idea that the Cav-1 negative tumor stroma would provide a micro-environment like a “wound that does not heal.”


Figure 3. Validating the tumor stromal expression of myo-fibroblast markers, calponin and prolyl 4-hydroxylase, in breast cancers that lack stromal Cav-1 expression. Paraffin-embedded tissue sections from human breast cancer samples were immuno-stained with antibodies directed against calponin and prolyl 4-hydroxylase, alpha I subunit (P4HA1). Slides were counterstained with hematoxylin. Note that breast cancer tumor sections show the overexpression of calponin and P4HA1 in the tumor stromal compartment. Original magnification, 40X.

These studies provide critical independent validation for our results obtained by proteomic analysis of Cav-1 (-/-) stromal cells. Discussion The “Warburg Hypothesis” was first formulated by Otto H. Warburg in the early 1920s.26 He hypothesized that tumor metabolism is different from normal metabolism, and relies on glycolysis for the production of energy in the form of ATP, despite the presence of oxygen. Thus, aerobic glycolysis has come to be known as the “Warburg Effect,” and was originally attributed to mitochondrial mal-functioning.26 This is thought to result from

Cell Cycle

3989


Figure 4. Annexins (A1 and A2) are overexpressed in the stroma of human breast cancers that lack stromal Cav-1 expression. Paraffin-embedded tissue sections from human breast cancer samples were immuno-stained with antibodies directed against annexins (A1 and A2). Slides were counterstained with hematoxylin. Note that breast cancer tumor sections show the overexpression of annexins (A1 and A2) in the tumor stromal compartment. Also, note that tumor cell “nests” surrounded by annexin-positive stroma were observed (upper). For annexin A2, virtually identical results were obtained with antibodies directed against both the heavy and light chains (not shown). Upper panels, original magnification, 20X; Lower panels, original magnification, 40X.

the adaptation of cancer cells to a hypoxic micro-environment. A subset of proteins overexpressed in breast cancer (such as Cyclin D1) have also been shown to inhibit mitochondrial metabolism, thereby driving epithelial cell aerobic glycolysis in vitro and in vivo.27,28 Importantly, aerobic glycolysis results in the production of two metabolic end-products, pyruvate and lactate, which can then be secreted by cancer cells. Secreted lactate and pyruvate can be taken up by adjacent cancer cells and provides a feedforward mechanism for tumor growth, as these metabolites can then enter into the TCA cycle in cancer cells which are using oxidative metabolism. Lactate dehyrogenase (LDH) is essential for this process, as it is a bi-directional enzyme that coverts lactate to pyruvate and visa-versa. So LDH converts lactate to pyruvate, which enters the TCA cycle. Bi-directional transport of lactate and pyruvate (into and out of cancer cells) is accomplished by

3990

a family of mono-carboxylate transporters (such as MCT1 and MCT4).29 This allows cancer cells undergoing aerobic glycolysis to “feed” adjacent cancer cells and other cells which are undergoing oxidative metabolism. Until recently, the “Warburg Effect” was thought to be confined to cancer cells. However, a recent paper by Vincent et al. 2008 directly demonstrates that myo-fibroblasts from skin conduct aerobic glycolysis much like cancer cells.14 As compared with normal fibroblasts, myo-fibroblasts consumed more glucose and produced more secreted lactate, behaving exactly like cancer cells.14 Furthermore, pre-treatment with quercetin—a well-known lactate transport inhibitor—is sufficient to prevent the conversion of normal fibroblasts to myo-fibroblasts.30-34 Interestingly, the idea that myo-fibroblasts exhibit the “Warburg Effect” has never been considered in the context of tumorigenesis or Warburg’s original

Cell Cycle

Volume 8 Issue 23


Figure 5. The M2-isoform of pyruvate kinase (M2-PK) is overexpressed in the stroma of human breast cancers that lack stromal Cav-1 expression. Paraffin-embedded tissue sections from human breast cancer samples were immuno-stained with antibodies directed against the M2-isoform of pyruvate kinase (PK-M2). Slides were counterstained with hematoxylin. Note that breast cancer tumor sections show the overexpression of PK-M2 in the tumor stromal compartment (see arrow). Virtually identical results were obtained with two different monoclonal antibodies directed against PK-M2. Upper panel, original magnification, 40X; Middle and Lower panels, original magnification, 60X.

hypothesis. Thus, aerobic glycolysis may be a key feature of the myo-fibroblast phenotype. This has important implications for tumorigenesis, if we consider the striking similarities between myo-fibroblasts and cancer-associated fibroblasts. In direct support of this notion, we show here that Cav-1 (-/-) deficient stromal cells exhibit the characteristics of myo-fibroblasts, as they overexpress five myo-fibroblast marker proteins.


Moreover, these Cav-1 (-/-) deficient stromal cells also overexpress 8 glycolytic enzymes, including the M2-isoform of pyruvate kinase and lactate dehydrogenase. Transcriptome analysis provided independent validation for these proteomic studies, and showed the overexpression of cancer-associated fibroblast markers, muscle-related genes, glucose and lactate transporters, glycolytic enzymes, and TGFb related signaling molecules, in Cav-1 (-/-) deficient stromal cells. Thus, our current results with Cav-1 (-/-) deficient stromal cells provide genetic support for the hypothesis that aerobic glycolysis is a key feature of the myofibroblast phenotype. Cav-1 (-/-) deficient stromal cells also functionally behave like human cancer-associated fibroblasts.6 By unbiased genome-wide expression profiling, we have previously shown significant overlap between the transcriptomes of Cav-1 (-/-) mammary stromal fibroblasts and human breast cancer-associated fibroblasts.6 Furthermore, Cav-1 (-/-) mammary stromal fibroblasts overexpress a number of growth factors associated with cancer-associated fibroblasts (PDGF, VEGF, TGFb, IL-6, among others), and can stimulate, in a paracrine-fashion, the onset of an EMT in normal mammary epithelial cells.6 Finally, an absence of stromal Cav-1 expression in human breast cancers is associated with metastasis and recurrence, resulting in poor-clinical outcome.8,9 With this in mind, we used human breast cancer tissues which lack the stromal expression of Cav-1 to validate our results from the proteomic analysis of Cav-1 (-/-) deficient stromal cells. Here, we show that eight of these proteins are indeed expressed in the stromal or extracellular matrix compartment of human breast cancer tissues. Importantly, two of these proteins are key Warburg-related glycolytic enzymes (the M2-isoform of pyruvate kinase and lactate dehydrogenase) that are prominently expressed in cancer-associated fibroblasts seen within the tumor stroma, but not within the adjacent cancer cells. Although these observations are consistent with the idea that aerobic glycolysis is a key feature of tumor metabolism, they directly suggest that this effect can also be confined to the tumor stromal compartment. As an absence of stromal Cav-1 is powerful predictive biomarker, these results suggest that the presence of aerobic glycolysis in the tumor stroma may have dire consequences for patient survival. Thus, based on these data, we would like to propose a new model for understanding the Warburg effect in tumor metabolism. Our hypothesis is that epithelial cancer cells induce the Warburg effect (aerobic glycolysis) in neighboring stromal fibroblasts. These cancer-associated fibroblasts, then undergo myo-

Cell Cycle

3991


fibroblastic differentiation, and secrete lactate and pyruvate (energy metabolites resulting from aerobic glycolysis). Epithelial cancer cells could then take up these energy-rich metabolites and use them in the mitochondrial TCA cycle, thereby promoting efficient energy production (ATP generation via oxidative phosphorylation), resulting in a higher proliferative capacity. This new model is illustrated schematically in Figure 9. In this alterative model of tumorigenesis, the epithelial cancer cells instruct the normal stroma to transform into a wound-healing stroma, providing the necessary energy-rich micro-environment for facilitating tumor growth and angiogenesis. In essence, the tumor stroma would directly feed the epithelial cancer cells, in a type of host-parasite relationship. We have termed this new idea the “Reverse Warburg Effect.” In this scenario, the tumor cells “corrupt” the normal stroma, turning it into a factory for the production of energy-rich metabolites. This alternative hypothesis is still consistent with Warburg’s original observation that tumors show a metabolic shift towards aerobic glycolysis. Thus, we should consider that the Warburg effect may be a stromal phenomenon. Consistent with this new stromal hypothesis, lactate by itself is sufficient to promote aerobic glycolysis, fibrosis and angiogenesis.35-37 So, secreted lactate and pyruvate derived from cancer-associated fibroblasts may also be used by endothelial progenitor cells, pericytes, and endothelial cells to drive angiogenesis. Similarly, a high tumor content of lactate is a powerful predictive biomarker for recurrence, metastasis and poor clinical outcome.38-41 Finally, it has been recognized as early as 1985 that TGFb treatment is sufficient to induce aerobic glycolysis and lactate production/secretion in fibroblasts.42 TGFb is also sufficient to induce the fibroblast-to-myofibroblast transition. This hormonal connection via TGFb also provides key evidence linking myofibroblasts with aerobic glycolysis and lactate production. A simple prediction of the “Reverse Warburg Effect” is that tumors with an increased percentage of stroma would have a worse prognosis, because they would be expected to have increased lactate production/secretion. In fact, when colon cancer patients are stratified based on tumor stromal content,43,44 patients with high tumor stromal content show dramatically increased tumor progression and poor survival, without the need for any biomarker analysis. The “Reverse Warburg Effect” may have important new implications for novel therapies targeting the tumor stromal microenvironment. If the “Reverse Warburg Effect” is correct, then lactate transport inhibitors would be a promising therapeutic strategy, as they would metabolically uncouple the stromal fibroblasts from the epithelial cancer cells. Lactate transport inhibitors would be predicted to kill cancer-associated fibroblasts, as they would prevent the secretion of lactate, leading towards intracellular acidification. Similarly, lactate transport inhibitors would starve epithelial tumor cells by eliminating their extracellular source of lactate and pyruvate. Non-metabolizable derivatives of glucose, such as 2-deoxy-glucose, could also be used to therapeutically target the fibroblastic tumor stroma. These ideas undoubtedly deserve further study. Further mechanistic experiments will be necessary to determine exactly how a loss of stromal Cav-1 coordinately induces

3992

Figure 6. Lactate dehydrogenase, a key glycolytic enzyme, is highly expressed in the breast cancer tumor stroma. Paraffin-embedded tissue sections from human breast cancer samples were immuno-stained with antibodies directed against lactate dehydrogenase (LDH). Slides were counterstained with hematoxylin. Note that breast cancer tumor sections show the overexpression of LDH in the tumor stromal compartment (see boxed area). Upper panel, original magnification, 40X; Lower panel, boxed area shown at higher magnification. Arrow points at LDHpositive stromal fibroblasts.

five myo-fibroblast markers and eight glycolytic enzymes. Interestingly, an informatics analysis of our proteomics results reveals that many of these protein products (including myofibroblast markers and glycolytic enzymes) are normally upregulated by hypoxia and/or are targets of the HIF genes.45 Thus, stromal induction of HIF-responsive genes could possibly explain our current findings. Notably, the stromal expression of HIF2alpha is associated with progression and poor clinical outcome in human colon cancer patients.46,47 Interestingly, the over-production of reactive oxygen species (ROS) is sufficient to induce HIF through its stabilization under normoxic conditions.48,49 Thus, a loss of stromal Cav-1 may somehow induce oxidative stress. In accordance with this hypothesis, Cav-1 (-/-) deficient stromal cells also show the upregulation of two anti-oxidant proteins (peroxiredoxin 1 and catalase) normally associated with ROS-production and oxidative stress.

Cell Cycle

Volume 8 Issue 23


Figure 8. Expression of tenascin C in the extracellular matrix in breast cancer tumor tissue lacking stromal Cav-1. Paraffin-embedded tissue sections from human breast cancer samples were immuno-stained with antibodies directed against tenascin C. Slides were counterstained with hematoxylin. Note that breast cancer tumor sections show the overexpression of tenascin C in the extracellular matrix. Note the presence of tumor cell nests outlined by tenascin C. Original magnification, 20X.

Figure 7. Validating the tumor stromal expression of an oncogene (EF1-delta) and an extracellular matrix protein (SPARC). Paraffin-embedded tissue sections from human breast cancer samples were immunostained with antibodies directed against EEF1D (eukaryotic translation elongation factor 1 delta) and SPARC (Secreted Protein Acidic and Rich in Cysteine). Slides were counterstained with hematoxylin. Note that breast cancer tumor sections show the overexpression of EEF1D in the tumor stromal compartment and SPARC in the extracellular matrix. Original magnification, 40X.

Previous unrelated studies have shown the Cav-1 expression is essential for liver regeneration. Thus, Cav-1 (-/-) null mice subjected to partial hepatectomy have extremely low survival rates.5052 This defect can be rescued by dietary supplementation of Cav-1 (-/-) deficient mice with glucose, but not fatty acids, as an energy source.50,51 As such, these functional/physiological observations are also consistent with the idea that Cav-1 (-/-) mice are more metabolically-dependent on glucose and aerobic glycolysis for routine energy production. The annexins are a family of membrane-targeted calciumbinding proteins. Interestingly, we show here that annexins A1 and A2 are both upregulated in Cav-1 (-/-) null stromal cells and


are present in the tumor stromal compartment of human breast cancer tissue sections. This is consistent with previous studies showing that annexins A1 and A2 are localized to myo-fibroblasts within human breast cancers.53 In accordance with our overall hypothesis, both annexins A1 and A2 have been shown to be upregulated in response to hypoxia and/or are HIF target genes.54,55 Importantly, annexin A1 may also be a therapeutic target as radio-labeled antibodies directed against annexin A1 are sufficient to block tumor growth and prolong survival in an animal model, using a rat mammary adenocarcinoma cell line.56 Interestingly, RhoGDI was also increased nearly twofold in Cav-1 (-/-) stromal cells. Originally, RhoGDI was thought to function as a negative regulator of the Rho family of small GTPases (Rho/Rac/Cdc42). However, it appears that RhoGDI is required for proper Rac-activation and targeting.57 Thus, RhoGDI is now considered to be a positive regulator of down-stream targets, such as NADPH oxidase.57,58 As such, an increase in RhoGDI expression would be expected to stimulate NADPH oxidase, leading to the increased production of reactive oxygen species (ROS) and oxidative stress. This may also explain why we see the concomitant upregulation of two antioxidant proteins (peroxiredoxin 1 and catalase) in Cav-1 (-/-) stromal cells. Materials and Methods Materials. Antibodies for immuno-staining were obtained from commercial sources: anti-annexin A1 (#610066, BD Biosciences); anti-annexin A2 [#610070 (LC, light chain) and #610068 (HC, heavy chain), BD Biosciences]; anti-calponin 1/2/3

Cell Cycle

3993

Table 3. Transcriptome analysis: validation of cellular proteins upregulated in Cav-1 (-/-) bone marrow stromal cells, continued Fold change (KO/WT)

Accession number

p-value

NM_146120

0.02

Myo-fibroblast associated proteins Gsn

gelsolin

2.05

Cnn2

calponin 2

1.95

NM_007725

0.006

Cnn1

calponin 1

1.37

NM_009922

0.09

Tpm3

tropomyosin 3

2.30

NM_022314

0.02

Tpm2

tropomyosin 2

1.71

BC014809

0.04

Tpm4

tropomyosin 4

1.60

NM_001001491

0.05

Vim

vimentin

1.58

ENSMUST00000028062

0.09

P4htm

prolyl 4-hydroxylase, transmembrane (ER)

1.70

NM_028944

0.03

P4ha1

prolyl 4-hydroxylase, alpha polypeptide I

1.35

NM_011030

0.1

P4hb

prolyl 4-hydroxylase, beta polypeptide

1.42

NM_011032

0.02

1.70

NM_013472

0.02 0.03

Signaling molecules


Anxa6

annexin A6

Anxa3

annexin A3

1.65

NM_013470

Anxa2

annexin A2

1.62

NM_007585

0.1

Anxa8

annexin A8

1.60

NM_013473

0.04

Anxa7

annexin A7

1.50

NM_009674

0.03

Anxa11

annexin A11

1.38

NM_013469

0.01

Arhgdib

Rho GDP dissociation inhibitor (GDI) beta

1.49

NM_007486

0.03

Arhgdig

Rho GDP dissociation inhibitor (GDI) gamma

1.38

NM_008113

0.01

Glycolytic and related metabolic enzymes Pkm2

pyruvate kinase, muscle

1.50

NM_011099

0.01

Pgk1

phosphoglycerate kinase 1

2.05

NM_008828

0.01

Pgk2

phosphoglycerate kinase 2

1.56

NM_031190

0.1

Ldhal6b

lactate dehydrogenase A-like 6B

2.04

NM_175349

0.007

Ldhb

lactate dehydrogenase B

2.44

NM_008492

0.02

Ldhc

lactate dehydrogenase C

1.48

NM_013580

0.04

Ldhd

lactate dehydrogenase D

1.53

NM_027570

0.04

Aldoa

aldolase A, fructose-bisphosphate

1.84

NM_007438

0.05

Aldob

aldolase B, fructose-bisphosphate

1.43

NM_144903

0.03

Aldoc

aldolase C, fructose-bisphosphate

2.82

NM_009657

0.01

Gpd2

glycerol-3-phosphate dehydrogenase 2

2.50

NM_001145820

0.005

Gpd1l

glycerol-3-phosphate dehydrogenase 1-like

1.94

NM_175380

0.02

Eno1

enolase 1, (alpha)

2.11

NM_023119

0.0002

Eno2

enolase 2 (gamma, neuronal)

1.31

NM_013509

0.07

Tpi1


2.02

NM_009415

0.006

Pgam1

phosphoglycerate mutase 1 (brain)

1.93

NM_023418

0.02

Pgam2

phosphoglycerate mutase 2 (muscle)

1.67

NM_018870

0.1

Pgam5

phosphoglycerate mutase family member 5

1.63

NM_028273

0.05

Hk1

hexokinase 1

1.87

NM_001146100

0.004

Hk2

hexokinase 2

2.40

NM_013820

0.04

Gck

glucokinase (hexokinase 4)

1.48

NM_010292

0.002

Gpi1

glucose phosphate isomerase 1

1.81

NM_008155

0.01

Pgm1

phosphoglucomutase 1

1.25

NM_025700

0.04

Pgm2


1.76

NM_028132

0.02

Pgm3


2.06

NM_028352

0.03

Pfkl

phosphofructokinase, liver

1.85

NM_008826

0.04

Gene products that were also identified by proteomics analysis are shown in bold; Other related genes and family members are also shown. 3994

Cell Cycle

Volume 8 Issue 23

Table 3. Transcriptome analysis: validation of cellular proteins upregulated in Cav-1 (-/-) bone marrow stromal cells, continued Pfkm

phosphofructokinase, muscle

1.94

NM_021514

0.03

Pfkfb3

6-phosphofructo-2-kinase/fructose-2,6-biphosphatase3

1.36

NM_133232

0.02

Gapdh

glyceraldehyde-3-phosphate dehydrogenase

1.58

NM_008084

0.04

Eef1d

eukaryotic translation elongation factor 1 delta

1.61

NM_029663

0.03

Oncogenes Anti-oxidants associated with oxidative stress Prdx4

peroxiredoxin 4

1.74

NM_016764

0.095

Prdx5

peroxiredoxin 5

1.64

NM_012021

0.1

Prdx2

peroxiredoxin 2

1.44

NM_011563

0.08

Cat

catalase

1.34

NM_009804

0.1

Gene products that were also identified by proteomics analysis are shown in bold; Other related genes and family members are also shown.


Table 4. Transcriptome analysis: validation of secreted proteins upregulated in Cav-1 (-/-) stromal cell “conditioned media” Fold change (KO/WT)

Accession number

p-value

collagen, type I, alpha1

1.59

NM_007742

0.05

Extracellular matrix proteins Col1a1 Col1a2

collagen, type I, alpha2

1.69

NM_007743

0.03

Spock1

sparc/osteonectin, proteoglycan (testican)1

1.93

NM_009262

0.0005

Spock3

sparc/osteonectin, proteoglycan (testican)3

1.43

NM_023689

0.02

Sparcl1

SPARC-like 1 (hevin)

1.61

NM_010097

0.007

Smoc1

SPARC related modular calcium binding 1

2.24

NM_001146217

0.02

Albumin and alpha-fetoprotein: liver-specific secreted proteins Alb

albumin

1.93

NM_009654

0.09

Afp

alpha-fetoprotein

1.49

NM_007423

0.06

Sh3bgrl3

SH3 domain binding glutamic acid-rich protein like 3

1.78

NM_080559

0.003

Other Gene products that were also identified by proteomics analysis are shown in bold; Other related genes and family members are also shown.

(#sc-28545, FL-297, Santa Cruz Biotech); anti-caveolin-1 (#sc894, N-20, Santa Cruz Biotech); anti-EF-1-delta (#ab13962, EEF1D, Abcam); anti-lactate dehydrogenase (#ab47010, Abcam), anti-prolyl 4-hydroxylase, alpha subunit (#12658-1-AP, P4HA1, Proteintech Group); anti-pyruvate kinase M2 (#S-1, clone DF-4, ScheBo Biotech; and ab55602, Abcam); anti-SPARC (ab14174, Abcam); tenascin C (NCL-TENAS-C, Novocastra/Leica Microsystems); and anti-vimentin (#M0725, clone V9, Dako; and ab8545, Abcam). Animal studies. All animals were housed and maintained in a pathogen-free environment/barrier facility at the Kimmel Cancer Center at Thomas Jefferson University under National Institutes of Health (NIH) guidelines. Mice were kept on a 12-hour light/ dark cycle with ad libitum access to chow and water. Cav-1 (-/-) deficient mice were generated, as we previously described.6,59 All wild-type and Cav-1 knockout (KO) mice used in this study were in the FVB/N genetic background. For most of the studies, 2.5-month-old virgin mice were used, unless stated otherwise. Animal protocols used for this study were pre-approved by the institutional animal care and use committee.


Isolation and culture of bone-marrow derived stromal cells (BMSCs). Bone marrow cells were collected from 10-week-old wild-type and Cav-1 (-/-) mice by flushing the hind leg femurs and tibias. Bone marrow cells were washed twice, and plated in 10 cm tissue culture dishes with Minimum Essential Media alpha (alpha-MEM; A10490-01, Gibco-Invitrogen), containing 10% FBS. Once the culture reached 80–90% confluency, cells were trypsinized and replated. Bone-Marrow Derived Stromal Cells (BMSCs) were used for experiments at passage 3. BMSCs prepared in this fashion are also referred to in the literature as mesenchymal stem cells (MSCs). Preparation of “conditioned media” from bone-marrow derived stromal cells. WT and Cav-1 (-/-) stromal cells were cultured in normal medium until they reached confluence. Then, the cells were washed 3 times with PBS and cultured in low-serum α-MEM (supplemented with 0.1% FBS). After 48 hours, tissue culture supernatants were collected, filtered through a 0.45 μm pore filter, and concentrated by Centriprep YM-10 (Millipore), according to the manufacturer’s instructions. Concentrated samples were stored at -80°C until proteomic analysis.

Cell Cycle

3995

Table 5. Transcriptome analysis of Cav-1 (-/-) stromal cells: TGFb signaling, glucose and lactate transporters, cancer-associated fibroblast markers and complement regulatory proteins, continued. Fold change (KO/WT)

Accession number

p-value


TGFbeta receptor signaling, ligands and target genes Ctgf

connective tissue growth factor

2.16

NM_010217

0.02

Tgfbi

transforming growth factor, beta-induced, 68 kDa

2.83

NM_009369

0.002

Tgfbr2

transforming growth factor, beta receptor II (70/80 kDa)

1.82

NM_009371

0.01

Tgfbr3

transforming growth factor, beta receptor III

1.70

NM_011578

0.03

Tgfb1

transforming growth factor, beta1

1.67

NM_011577

0.01

Tgfb1i1

transforming growth factor beta1 induced transcript 1

1.43

NM_009365

0.04 0.03

Smad3

SMAD family member 3

1.75

NM_016769

Smad6


1.61

NM_008542

0.03

Smad9


1.62

NM_019483

0.02

Smad5


1.48

NM_008541

0.04

Twist2

twist homolog 2 (Drosophila)

1.39

NM_007855

0.002

Snai2

snail homolog 2 (Drosophila); Slug

1.30

NM_011415

0.03

Loxl1

lysyl oxidase-like 1

1.80

NM_010729

0.05

Loxl3


1.87

NM_013586

0.05


1.75

NM_053083

0.007

Loxl4

Family of facilitated glucose transporters (GLUTs) Slc2a6

solute carrier family 2, GLUT6

2.03

NM_172659

0.01

Slc2a5


1.87

NM_019741

0.008

Slc2a3


1.77

NM_011401

0.02

Slc2a8


1.30

NM_019488

0.01

Family of mono-carboxylate transporters (MCT1/4) Slc16a1

solute carrier family 16, member 1 (MCT1)

1.62

NM_009196

0.04

Slc16a3

solute carrier family 16, member 3 (MCT4)

1.53

NM_030696

0.03

chemokine (C-X-C motif) ligand 9

7.99

NM_008599

0.04

Mme

CD10 membrane metallo-endopeptidase

6.48

NM_008604

0.01

Cxcl12

chemokine (C-X-C motif) ligand 12 (SDF1)

5.14

NM_021704

0.002

Cancer-associated fibroblast markers Cxcl9

Cxcl11

chemokine (C-X-C motif) ligand 11

3.97

NM_019494

0.01

Ly6e

lymphocyte antigen 6 complex, locus E; stem cell antigen-2

3.78

NM_008529

0.009

Ccl5

chemokine (C-C motif) ligand 5 (RANTES)

3.37

NM_013653

0.03

Hgf

hepatocyte growth factor (hepapoietin A; scatter factor)

3.09

NM_010427

0.04

Met

met proto-oncogene (Hgf receptor)

1.36

NM_008591

0.04

Vegfa

vascular endothelial growth factor A

2.56

NM_001025250

0.04

Cdh11

cadherin 11, type 2, OB-cadherin (osteoblast)

2.37

NM_009866

0.004

Pdpn

podoplanin

2.10

NM_010329

0.01

Pdgfrl

platelet-derived growth factor receptor-like

2.02

NM_026840

0.05

Pdgfrb

platelet-derived growth factor receptor, beta polypeptide

1.78

NM_001146268

0.03

Pdgfra

platelet-derived growth factor receptor, alpha polypeptide

1.57

NM_011058

0.05

Pdgfa

platelet derived growth factor A

1.32

NM_008808

0.008

Cd34

Cd34 molecule

1.92

NM_001111059

0.003

Smtn

smoothelin

1.78

NM_001159284

0.02

Igf2

insulin-like growth factor 2 (somatomedin A)

1.76

NM_010514

0.0004

Cd248

Cd248 molecule, endosialin, Tem1

1.53

NM_054042

0.04

Pecam1

platelet/endothelial cell adhesion molecule; Cd31

1.47

NM_008816

0.04

Retnla

resistin like alpha; Fizz1

1.34

NM_020509

0.03

3996

Cell Cycle

Volume 8 Issue 23

Table 5. Transcriptome analysis of Cav-1 (-/-) stromal cells: TGFb signaling, glucose and lactate transporters, cancer-associated fibroblast markers and complement regulatory proteins, continued. Cav1

caveolin-1; caveolae-associated protein

0.048

NM_007616

0.0001


Muscle-related genes Acta1

actin, alpha1, skeletal muscle

1.65

NM_009606

0.006

Actc1

actin, alpha, cardiac muscle 1

1.68

NM_009608

0.006

Actg1

actin, gamma1

2.68

NM_009609

0.004

Actl6a

actin-like 6A

1.80

NM_019673

0.002

Actl6b

actin-like 6B

1.63

NM_031404

0.01

Actn2

actinin, alpha2

2.50

NM_033268

0.03

Actn3

actinin, alpha3

1.45

NM_013456

0.04

Actn4

actinin, alpha4

1.41

NM_021895

0.03

Actr1a

ARP1 actin-related protein 1 homolog A, centractin (yeast)

1.70

NM_016860

0.03

Actr6

ARP6 actin-related protein 6 homolog (yeast)

1.90

NM_025914

0.02

Myo10

myosin X

2.15

NM_019472

0.02

Myo15

myosin XV

2.08

NM_010862

0.02

Myo18a

myosin XVIIIA

2.44

NM_011586

0.01

Myo1b

myosin IB

1.74

NM_010863

0.03

Myo1c

myosin IC

1.66

NM_001080775

0.02

Myo1e

myosin IE

1.47

NM_181072

0.03

Myo1g

myosin IG

1.49

NM_178440

0.0004

Myo1h

myosin IH

1.71

ENSMUST00000031566

0.04

Myo5b

myosin VB

1.45

NM_201600

0.02

Myo5c

myosin VC

1.29

NM_001081322

0.03

Myo6

myosin VI

1.99

NM_001039546

0.04

Myo7a

myosin VIIA

1.94

NM_008663

0.006

Myo7b

myosin VIIB

1.64

NM_032394

0.03

Myo9b

myosin IXB

1.93

NM_001142322

0.01

Myod1

myogenic differentiation 1

1.52

NM_010866

0.01

Myom1

myomesin 1, 185 kDa

1.75

NM_010867

0.02

Myom2

myomesin (M-protein) 2, 165 kDa

1.59

NM_008664

0.02

Myh10

myosin, heavy chain 10, non-muscle

1.41

NM_175260

0.004

Myh11

myosin, heavy chain 11, smooth muscle

3.12

NM_013607

0.04

Myh14

myosin, heavy chain 14

1.88

AY363100

0.04

Myh6

myosin, heavy chain 6, cardiac muscle, alpha

2.34

NM_010856

0.04

Myh7

myosin, heavy chain 7, cardiac muscle, beta

1.96

NM_080728

0.05

Myh9

myosin, heavy chain 9, non-muscle

3.14

NM_022410

0.008

Myl10

myosin, light chain 10, regulatory

1.75

NM_021611

0.03

Myl2

myosin, light chain 2, regulatory, cardiac, slow

1.89

NM_010861

0.03

Mylk

myosin light chain kinase

1.34

NM_139300

0.001

Mylk3

myosin light chain kinase 3

1.48

NM_175441

0.02

Mylpf

myosin light chain, phosphorylatable, fast skeletal muscle

1.27

NM_016754

0.04

Complement regulatory proteins Cfb

complement factor B

7.80

NM_008198

0.01

C3

complement component 3

4.35

NM_009778

0.004

C1r

complement component 1, r subcomponent

2.04

NM_023143

0.02

C1ql1

complement component 1, q subcomponent-like 1

1.87

NM_011795

0.01

C1ql2


1.40

NM_207233

0.03

C1ql3


1.85

AB044560

0.004


Cell Cycle

3997


Table 5. Transcriptome analysis of Cav-1 (-/-) stromal cells: TGFb signaling, glucose and lactate transporters, cancer-associated fibroblast markers and complement regulatory proteins, continued. C1qtnf2

C1q and tumor necrosis factor related protein 2

1.66

NM_026979

0.02

C1qtnf5

C1q and tumor necrosis factor related protein 5

1.99

NM_145613

0.005 0.02

C2


1.65

NM_013484

C4b

complement component 4B

1.49

NM_009780

0.01

C8a

complement component 8, alpha polypeptide

1.56

NM_146148

0.03

C8b

complement component 8, beta polypeptide

1.27

NM_133882

0.04

C8g

complement component 8, gamma polypeptide

1.35

NM_027062

0.03

C9


1.42

NM_013485

0.003

Proteomic analysis. 2-D DIGE (two-dimensional difference gel electrophoresis) 60 and mass spectrometry protein identification were run by Applied Biomics (Hayward, CA). Image scans were carried out immediately following SDS-PAGE using Typhoon TRIO (Amersham BioSciences) following the protocols provided. The scanned images were then analyzed by Image QuantTL software (GE-Healthcare), and then subjected to in-gel analysis and cross-gel analysis using DeCyder software version 6.5 (GE-Healthcare). The ratio of protein differential expression was obtained from in-gel DeCyder software analysis. The selected spots were picked by an Ettan Spot Picker (GE-Healthcare) following the DeCyder software analysis and spot picking design. The selected protein spots were subjected to in-gel trypsin digestion, peptides extraction, desalting and followed by MALDI-TOF/TOF (Applied Biosystems) analysis to determine the protein identity. Immuno-histochemical analysis. Immunohistochemical staining was performed essentially as we previously described.8,9 Briefly, 5-μm sections from paraffin-embedded breast cancer tissue were de-paraffinized and rehydrated by passage through a graded series of ethanol. Antigen retrieval was performed by heating the slides in 10 mM sodium citrate buffer, pH 6.0, for 10 minutes using a pressure cooker. Endogenous peroxidase activity was quenched with 3% H2O2 for 10 minutes. Then, slides were washed with phosphate-buffered saline (PBS) and blocked with 10% goat serum in PBS for 1 hour at room temperature. Samples were incubated with the primary antibodies diluted in 1% BSA in PBS overnight at 4°C. After washing in PBS (three times, 5 minutes each), slides were stained with the LSAB2 system kit (Dako Cytomation, Glostrup, Denmark), according to the manufacturer’s recommendations. Briefly, samples were incubated with biotinylated linker antibodies for 30 minutes, washed in PBS (three times, 5 minutes each), and then incubated with a streptavidin-horseradish peroxidase-conjugated solution for 30 minutes. After washing, samples were incubated with the diaminobenzidine reagent until color production developed. Finally, the slides were rinsed with tap water and counter-stained with hematoxylin, dehydrated, and mounted with coverslips. Importantly, critical negative controls were performed in parallel for all of the immunohistochemical studies. Genome-wide transcriptional profiling. Total RNA was isolated from WT and Cav-1 (-/-) stomal cells using RNAeasy mini columns (Qiagen). RNA was prepared from three wildtype and three Cav-1 (-/-) stromal cell isolates. DNase-treated

3998

RNA was ethanol precipitated and quantified on a NanoDrop ND-1000 spectrophotometer, followed by RNA quality assessment by analysis on an Agilent 2100 Bioanalyzer (Agilent Inc., Palo Alto, CA). RNA amplification and labeling was performed by the WT-Ovation Pico RNA amplification system (NuGen Technologies, Inc.,). Briefly, 500 pg to 50 ng of total RNA was reverse transcribed using a chimeric cDNA/mRNA primer, and a second complementary cDNA strand was synthesized. Purified cDNA was then amplified with ribo-SPIA enzyme and SPIA DNA/RNA primers (NuGEN Technologies, Inc.). Amplified DNA was purified with the QIAquick PCR purification kit (Qiagen). Sense transcript cDNA (ST-cDNA) was generated from 3 μg amplified cDNA using WT-Ovation Exon module (NuGen Technologies, Inc.). Purified ST-cDNA was assed for yield using the Nanodrop Spectrophotometer (NanoDrop Technologies, Inc.). 2.5 μg ST-cDNAs were fragmented and chemically labeled with biotin to generate biotinylated ST-cDNA using FL-Ovation cDNA biotin module V2 (NuGen Technologies, Inc.,). Each Affymetrix GeneChip® Mouse Exon 1.0 ST array (Affymetrix, Santa Clara, CA) was hybridized with fragmented and biotinlabeled target (2.5 μg) in 110 μl of hybridization cocktail. Target denaturation was performed a 99°C for 2 min. and then 45°C for 5 min., followed by hybridization for 18 hrs. Arrays then were washed and stained using Genechip Fluidic Station 450, and hybridization signals were amplified using antibody amplification with goat IgG (Sigma-Aldrich) and anti-streptavidin biotinylated antibody (Vector Laboratories, Burlingame, CA). Chips were scanned on an Affymetrix Gene Chip Scanner 3000, using Command Console Software. Background correction and normalization were done using the Robust Multichip Average (RMA) with Genespring V 10.0 software (Agilent, Palo Alto, CA). The Robust Multichip Average (RMA) signal was computed for exon and gene-level probeset summaries by performing the RMA-sketch analysis for CORE probesets in Affymetrix Expression Console Version 1.1 (http://www.affymetrix.com). Log2 RMA expression values were exported as a text file and additional calculations were performed in Matlab 2009a (The MathWorks, Natick, MA; www.mathworks.com) and Microsoft Excel (Microsoft, Redmond, WA; www.microsoft.com). The difference between average expression in knockout samples and the average expression in wild type samples was computed as well as the fold-change between the two sample groups. Results from one exon for each gene analyzed were selected for inclusion in Tables 3–5. A p-value of