Is there a role for mammary stem cells in inflammatory breast carcinoma?

A review of evidence from cell line, animal model, and human tissue sample experiments

Authors

  • Steven Van Laere MSc, PhD,

    Corresponding author
    1. Translational Cancer Research Group, Laboratory of Pathology, University of Antwerp/University Hospital Antwerp, Edegem, Belgium
    2. Oncology Center, General Hospital Sint-Augustinus, Wilrijk, BelgiumThe articles in this supplement were presented at the First International Inflammatory Breast Cancer Conference, Houston, Texas, December 5-7, 2008
    • Wetenschappelijk Labo—Oncologisch Centrum, AZ Sint-Augustinus, Oosterveldlaan 24, 2610 Wilrijk, Belgium
    Search for more papers by this author
    • Fax: (011) 32 (0)3 443 30 36

  • Ridha Limame MSc,

    1. Translational Cancer Research Group, Laboratory of Pathology, University of Antwerp/University Hospital Antwerp, Edegem, Belgium
    2. Oncology Center, General Hospital Sint-Augustinus, Wilrijk, BelgiumThe articles in this supplement were presented at the First International Inflammatory Breast Cancer Conference, Houston, Texas, December 5-7, 2008
    Search for more papers by this author
  • Eric A. Van Marck MD, PhD,

    1. Translational Cancer Research Group, Laboratory of Pathology, University of Antwerp/University Hospital Antwerp, Edegem, Belgium
    2. Oncology Center, General Hospital Sint-Augustinus, Wilrijk, BelgiumThe articles in this supplement were presented at the First International Inflammatory Breast Cancer Conference, Houston, Texas, December 5-7, 2008
    Search for more papers by this author
  • Peter B. Vermeulen MD, PhD,

    1. Translational Cancer Research Group, Laboratory of Pathology, University of Antwerp/University Hospital Antwerp, Edegem, Belgium
    2. Oncology Center, General Hospital Sint-Augustinus, Wilrijk, BelgiumThe articles in this supplement were presented at the First International Inflammatory Breast Cancer Conference, Houston, Texas, December 5-7, 2008
    Search for more papers by this author
  • Luc Y. Dirix MD, PhD

    1. Translational Cancer Research Group, Laboratory of Pathology, University of Antwerp/University Hospital Antwerp, Edegem, Belgium
    2. Oncology Center, General Hospital Sint-Augustinus, Wilrijk, BelgiumThe articles in this supplement were presented at the First International Inflammatory Breast Cancer Conference, Houston, Texas, December 5-7, 2008
    Search for more papers by this author

Abstract

Stem cells are pluripotent cells, with a large replicative potential, which perform normal physiological functions such as tissue renewal and damage repair. However, because of their long lifespan and high replicative potential, stem cells are ideal targets to accumulate multiple mutations. Therefore, they can be regarded as being responsible for the initiation of tumor formation. In the past, numerous studies have shown that the presence of an elaborate stem cell compartment within a tumor is associated with aggressive tumor cell behavior, frequent formation of metastases, resistance to therapy, and poor patient survival. From this perspective, tumors from patients with inflammatory breast cancer (IBC), an aggressive breast cancer subtype with a dismal clinical course, are most likely to be associated with stem cell biology. To date, this hypothesis is corroborated by evidence resulting from in vitro and in vivo experiments. Both gene and microRNA expression profiles highlighted several stem cell-specific signal transduction pathways that are hyperactivated in IBC. Also, these stem cell-specific signal transduction pathways seem to converge in the activation of nuclear factor-kappa B, a molecular hallmark of IBC, and induction of epithelial-to-mesenchymal transition. Recently, the latter mechanism was identified as a prerequisite for the induction of stem cell characteristics in breast cancer cells. Cancer 2010;116(11 suppl):2794–805. © 2010 American Cancer Society.

Stem Cells and the Cancer Stem Cell Hypothesis

Stem cells are cells with a capacity to self-renew and to generate daughter cells that can differentiate down several lineages to form all of the cell types that are found in mature tissue. Stem cells might go through an asymmetric cell division, producing a novel stem cell to maintain the stem cell compartment and a cell that differentiates, via additional cell divisions, into terminally differentiated cells. The cells that form the intermediates between stem cells and terminally differentiated cells are usually referred to as progenitor cells, transit cells, or transit amplifying cells. In addition, a stem cell might divide symmetrically, resulting in 2 identical daughter cells. These 2 daughter cells can either be 2 stem cells (expansion of the stem cell compartment), 2 transit/progenitor cells (depletion of stem cell compartment), or again 1 stem cell and 1 transit/progenitor cell. Physiologically, stem cells perform different functions in normal human tissue, among others tissue renewal (eg, bone marrow, small intestine, skin) or damage repair (eg, satellite cells of muscle).1 The latter suggests that stem cells are present in all tissues of the human body.

Because of their large replicative potential and their long lifespan, stem cells can accumulate several mutations over time, which makes them excellent candidates for the cells of origin of cancer. Also, mutations might cause restricted progenitors or differentiated cells to acquire properties of cancer stem cells, such as increased self-renewal potential. Either way, the cancer stem cells would be distinguished from other cancer cells by their tumorigenic potential, their ability to generate additional cancer stem cells (self-renewal), and their ability to generate phenotypically diverse nontumorigenic cancer cells with more limited proliferative potential (pluripotency).2 From the perspective of tumorigenesis, the latter 2 properties are particularly interesting, as self-renewal pathways are thought to be the engines that drive neoplastic proliferation, and pluripotency provides an explanation for the heterogeneous cell population that is found in a tumor.

Stem Cells in Breast Cancer

The existence of a stem cell population in the adult human mammary gland is suggested by the finding that the terminal ductal lobular units (TDLUs) are monoclonal in origin. The TDLUs are structures in the human breast that consist of alveoli that are clustered around a distinct duct and ductal side branches. It was observed that the pattern of X chromosome-linked gene inactivation is the same for all cells composing the TDLUs.3 The clonal origin of individual TDLUs supports a model of adult stem cells that are distributed throughout the mammary gland.1 The further characterization of mammary (cancer) stem cells was aided by the development of an in vitro propagation system for stem cells in suspension culture.4 In analogy with the in vitro propagation system for neuronal stem cells, which form neurospheres in suspension, mammary stem cells form mammospheres in suspension culture. These mammospheres are spherical cell structures that are able to survive and proliferate in the absence of attachment to an exogenous substratum. As is the case for neurospheres, the mammospheres are highly enriched in bi- or multipotent cells, as demonstrated by the ability of single cells isolated from mammospheres to generate multilineage colonies when cultured under conditions promoting differentiation. In addition, it was demonstrated that single cells obtained from mammospheres are capable of generating secondary mammospheres, suggesting that these single cells are capable of self-renewal.5 The development of the in vitro propagation system for mammary (cancer) stem cells has aided in their further molecular characterization by means of transcriptional profiling.4

The contribution of mammary stem cells to breast carcinogenesis was first suggested by Al-Hajj et al.6 Uncultured specimens of human breast cancer cells from 9 patients were separated into fractions that expressed different surface molecules, and then injected in immunodeficient mice. As few as 200 CD44+/CD24 cancer cells were able to consistently form tumors, whereas injection of thousands of cancer cells that had other phenotypes failed to form tumors. These CD44+/CD24 cells were able to generate new cancer stem cells as well as cells with a distinct phenotype and lower tumorigenic potential.6 To date, several other markers besides dual-labeling with CD44 and CD24 have been used to characterize putative stem cell populations in breast cancer, including ALDH1,7 CD133,8 and Hoechst-dye exclusion.9, 10 Applying these markers to human tissue samples, it was demonstrated that the presence of cancer stem cells in breast tumors is particularly associated with specific subtypes like basal-like (triple negative) breast tumors,11, 12 ErbB2+ breast tumors,13 and breast tumors with mutations in the BRCA1 gene.8, 14 Similarly, it has also been shown that the presence of stem cells in breast tumors is associated with the occurrence of metastases and poor prognosis15-18 and resistance to therapy,19, 20 and that breast cancer stem cells harbor tumor-initiating capacity.21

Molecular Repertoire of the Breast Cancer Stem Cell

In the past, several studies have focused on the molecular characteristics of breast cancer stem cells. The main focus of these studies was on signal transduction pathways associated with embryonic development, such as Notch, Wnt, Hedgehog, tumor growth factor (TGF)β, and bone morphogenetic protein (BMP) signaling. These signaling pathways have repeatedly been associated with tumor progression, from the induction of a hyperproliferative phenotype to tumor cell dissemination and metastatic spread.22 That tumor cells use pathways critical to the developing embryo is not surprising, as many of the normal developmental programs include processes that are also seen during tumor progression to a metastatic phenotype, including epithelial-to-mesenchymal transition, tissue-specific morphogenesis, cellular motility, and invasion. The Notch, Wnt, Hedgehog, TGFβ, and BMP signaling pathways have also been associated with tumor progression in breast cancer, the evidence for which is briefly reviewed in the following paragraphs.

Notch signaling pathway

A role for Notch signaling in breast carcinogenesis was suggested by the observation that Notch1 and Notch4 are targets for insertion and rearrangement by the mouse mammary tumor virus, and that these mutations promote epithelial mammary tumorigenesis.23 This suggestion was corroborated by a study of Stylianou et al showing that activation of the Notch signaling pathway leads to transformation of normal breast epithelial cells.24 To date, Notch signaling has been associated with several clinicopathological parameters25-28 and poor overall survival.29, 30 A possible mechanistic link between Notch signaling and breast tumorigenesis was provided by Dontu and colleagues, who have demonstrated that Notch signaling plays a critical role in the biology of mammary stem/progenitor cells by affecting self-renewal and lineage-specific differentiation.31 Also, Farnie and colleagues have demonstrated that the mammosphere-forming efficiency of specimens from ductal carcinoma in situ is reduced when Notch signaling is inhibited.32

Wnt signaling pathway

With regard to Wnt signaling, Li et al have shown that overexpression of Wnt ligands in stroma of transgenic mice or activation of β-catenin in murine mammary epithelium leads to an increased number of mammary stem cells.33 In a preceding study, the same group showed that mammary tumors induced by components of the Wnt pathway contain heterogeneous cell types and express early developmental markers, in contrast to tumors induced by other signaling elements. These results suggest that mammary stem cells may be the targets for oncogenesis by Wnt signaling elements.34 The identification of overexpressed members of the Wnt/Frizzled pathway within the transcriptional profile of mammosphere-derived cells further strengthens this hypothesis.5 Lindvall and colleagues have demonstrated that the Wnt signaling receptor LRP5 is required for mammary ductal stem cell activity and Wnt1-induced tumorigenesis.35

Hedgehog signaling pathway

The association between Hedgehog signaling and breast cancer was suggested by Moraes and colleagues, who demonstrated that constitutive activation of Smoothened, the receptor for Hedgehog ligands, in mammary glands of transgenic mice leads to increased proliferation, altered differentiation, and ductal dysplasia.36 In human breast cancer, Xuan and colleagues have demonstrated that all Hedgehog signaling molecules are overexpressed at the protein level in invasive ductal breast carcinoma compared with normal breast epithelium. Also, associations between the expression of Hedgehog signaling molecules and progesterone receptor expression, cell proliferation, lymph node metastasis, and tumor stage were described.37 Liu and colleagues demonstrated a role for Hedgehog signaling in human breast cancer stem cells by showing that human breast cancer stem cells characterized by the CD44+/CD24−/lowLin phenotype have activated Hedgehog signaling.38

TGFβ signaling pathway

The association between TGFβ signaling and breast cancer has already been extensively demonstrated. Introduction of dominant negative TGFβ receptors into metastatic breast cancer cells has been shown to inhibit epithelial-to-mesenchymal transition, motility, invasiveness, survival, and the formation of metastases, thereby indicating the potential roles of TGFβ in breast carcinogenesis.39 Consistent with this view, Grau et al have shown that breast cancer patients with high plasma levels of TGFβ have poor overall survival.40 These observations can be explained by the effect of TGFβ signaling on the generation of cancer stem cells by the process of epithelial-to-mesenchymal transition secondary to RAS/MAPK activation.41, 42 Shipitsin and colleagues have demonstrated that inhibition of TGFβ signaling in CD44+ mammary cancer stem cells resulted in a more epithelial, less differentiated phenotype.43 Interestingly, Hedgehog signaling, which is also activated in CD44+/CD24 breast cancer cells, induces TGFβ signaling,44 and thereby induces epithelial-to-mesenchymal transition by upregulating ZEB1 and ZEB2 through nuclear factor-kappa B (NF-κB).

BMP signaling pathway

The role of the BMP pathway has been established in ER-positive breast carcinomas, for which the expression of BMPR1B is a major hallmark of tumor progression and dedifferentiation. For ER-positive breast carcinomas, the expression of BMPR1B was associated with high tumor grade, high tumor proliferation, cytogenetic instability, and poor prognosis.45 In addition to the receptors, several BMP ligands (specifically BMP4 and BMP7) have also been associated with breast cancer progression, including the induction of cell motility, hormone independent growth, and angiogenesis.46-49 More recently, Montesano has shown that BMP4 has a negative effect on the in vitro formation of alveolar-like spherical cysts from mammary epithelial cells and induces scattering of individual cells into the surrounding collagen matrix.50 In addition, it was shown that BMP4 potentiates the mitogenic activity of multiple growth factors and hence may contribute to mammary gland morphogenesis as well as to breast cancer progression.51 Although the studies by Montesano suggest that BMPs are implicated in the biology of mammary stem cells, direct evidence has not been published so far.

Evidence for the Involvement of Breast Cancer Stem Cells in Human Inflammatory Breast Cancer

Inflammatory breast cancer (IBC) is an aggressive subtype of locally advanced breast cancer characterized by explosive local growth and the frequent formation of metastases. At diagnosis, virtually all patients have lymph node metastases, and 1 of 3 of the patients have metastases in distant organs. A pathological hallmark of IBC is the frequent presence of tumor emboli, which exhibit a mammosphere-like, anchorage-independent growth within the lumen of dermal and parenchymal lymph vessels. Histologically, IBC is a poorly differentiated tumor and demonstrates a specific and diffuse growth pattern characterized by the absence of a well-defined tumor mass and the presence of skip zones, which are constituted of tumor-free areas between small nodules of tumor cells. From the molecular perspective, IBC is characterized by a frequent absence of ER protein expression, amplification of ErbB2 and/or overexpression of epidermal growth factor receptor, and an elevated p53 mutation rate. These molecular characteristics agree with the predominance of the basal-like and ErbB2+ molecular breast cancer subtypes, as defined by Perou and colleagues,52 in IBC.53, 54 Also, recent data suggest that the majority of the ER+ IBC samples can be classified within the luminal B molecular subtype (unpublished data). The latter subtype is thought to be associated with resistance against endocrine therapy,55 corroborating the view that ER+ IBC samples are resistant to endocrine therapy. Altogether, the histological and biological features of IBC in combination with its poor-prognosis profile and the frequent occurrence of metastases suggest that cancer stem cells play an important role in IBC.

Recently, Xiao et al have shown that lymphovascular emboli of the MARY-X model, a human xenograft model of IBC, demonstrate a stem cell-like phenotype.56 Spheroids generated from tumor cells from MARY-X expressed embryonal stem cell markers (Stellar, Rex1, Nestin) and transcription factors that are associated with self-renewal and developmental potential (Oct4, Sox2, Nanog). Most importantly, the spheroids generated from tumor cells from MARY-X express a stem cell phenotype characterized by the expression of known stem cell markers (eg, CD44+/CD24, ALDH1, and CD133). As few as 100 cells derived from single-cell clonogenic expansion were tumorigenic with recapitulation of the IBC phenotype. Prototype stem cell signaling pathways such as Notch were active in MARY-X. Interestingly, the stem cell phenotype of the lymphovascular emboli of MARY-X was also observed in the lymphovascular emboli of human IBC cases, corroborating the hypothesis that lymphovascular emboli in human IBC resemble mammosphere-like structures. In addition to these findings, it was demonstrated that a recently established IBC cell line (MDA-IBC-1) exhibits a stem cell-like phenotype.

A stem cell expression profile of IBC

In 2007, Liu et al published the Invasiveness Gene Signature. This signature consists of 186 genes with a differential gene expression profile between CD44+/CD24 tumorigenic breast cancer cells and normal breast epithelium. It was demonstrated that breast tumors with a gene expression profile that is positively correlated with the Invasiveness Gene Signature demonstrate a poor prognosis profile compared with breast tumors with a gene expression profile that is negatively correlated with the Invasiveness Gene Signature.15 By using unsupervised hierarchical clustering, we demonstrated that IBC and non-IBC demonstrate a different gene expression profile with respect to the Invasiveness Gene Signature. Two sample clusters were identified, 1 enriched in IBC samples (67% IBC samples vs 33% non-IBC samples) and a second enriched in non-IBC samples (8% IBC samples vs 92% non-IBC samples) (Fig. 1A). Then, we analyzed whether the gene expression profiles of IBC or non-IBC samples were positively correlated with the Invasiveness Gene Signature. A centroid-mediated classification algorithm was applied, and demonstrated that the gene expression profiles of IBC samples were significantly more often positively correlated with the Invasiveness Gene Signature compared with the gene expression profiles of non-IBC samples (Table 1).57 In addition to the Invasiveness Gene Signature, other gene expression signatures associated with mammary stem cells have been published.5, 11, 15, 30, 43, 58 These mammary gene expression signatures have been applied to IBC and non-IBC gene expression data using a similar centroid-mediated classification algorithm as applied for the Invasiveness Gene Signature. Results are displayed in Table 1. Overall, these mammary stem cell gene signatures demonstrate that, on average, 74% of the IBC samples and 44% of the non-IBC samples are predicted to have an elaborate stem cell compartment. In addition, when comparing the list of genes that are differentially expressed between IBC and non-IBC (false discovery rate <, 0.01), with the mammary stem cell signatures it was observed that the IBC-specific gene set shows a significant overlap with most of the mammary stem cell signatures, except for leukemia viral BMI1 onogene (BMI1)-regulated gene sets and gene sets associated with nonmammary stem cells (Fig. 1B). This suggests that genes specifically associated with mammary stem cells are also differentially expressed between IBC and non-IBC and vice versa, again suggesting that stem cells are indeed involved in the biology of IBC.

Figure 1.

(A) Unsupervised hierarchical clustering analysis using the Invasiveness Gene Signature is shown. Data are represented in matrix format, with rows representing genes and columns representing samples. The gene expression data are color-coded, with red denoting overexpression and blue denoting underexpression. Color saturation is associated with the gene expression level. Two sample clusters are clearly visible, 1 enriched in inflammatory breast cancer (IBC) samples (Left) and 1 enriched in non-IBC (non-IBC) samples (Right). The colored bar underneath the dendrogram represents the sample grouping, with blue denoting non-IBC and red denoting IBC. (B) This heat map demonstrates the associations between different mammary stem cell signatures and a list of genes with differential expression between IBC and non-IBC. The color saturation, ranging from white to dark blue, provides a measure for overlap between different gene expression signatures, with dark blue denoting a complete overlap and white denoting no overlap. Unsupervised hierarchical clustering identifies 3 sets of mammary stem cell signatures, which is demonstrated by the dendrograms in the x axis and y axis. The set of differentially expressed genes between IBC and non-IBC (second from the top) is associated with the majority of mammary stem cell signatures, except for the BMI-regulated genes set (second bifurcation) and the gene sets associated with nonmammary stem cells (first bifurcation). The colored bar underneath the bar denotes the P value of a gene set enrichment analysis comparing the mammary stem cell signatures with the list of differentially expressed genes between IBC and non-IBC. The colors range from orange (highly significant) to purple (not significant). It can be observed that the majority of the gene sets clustered with the set of IBC-specific genes are also significantly associated with the IBC gene set using gene set enrichment analysis.

Table 1. Association Between Mammary Stem Cell Signatures and Breast Cancer Phenotypes
SignatureIBCnIBCPBasal-likeErbB2+Luminal ALuminal BNormal-likePStudy
  • IBC indicates inflammatory breast cancer; nIBC, non-IBC; +, positive; −, negative; BMI1, leukemia viral BMI1 oncogene; CNS, central nervous system; PNS, peripheral nervous system; SC, stem cells; HSC, hematopoietic SCs; ESC, embryonal SCs; NSC, neuronal SCs; NA, not applicable.

  • a

    Genes with 2-fold differential expression.

  • b

    Genes with no expression in 1 condition.

ALDH1+ vs. ALDH174%28%.00172%92%10%89%0%<.001Charafe-Jauffret 200958
CD44+ vs CD24+74%50%.10057%92%35%45%69%.018Shipitsin 200743
Mammospheres vs differentiated (2-fold)a79%43%.01272%75%25%45%77%.011Dontu & Wicha 20055
Mammospheres vs differentiated (on/off)b84%38%<.001100%66%20%45%69%<.001Dontu & Wicha 20055
CD44+/CD24 vs non-CD44+/CD2474%40%.02672%83%45%78%0%<.001Shipitsin 200743
CD44+ vs CD4458%21%.00728%58%10%56%31%.024Honeth 200811
Invasiveness gene signature84%19%<.00172%92%5%66%8%<.001Liu 200715
Notch/survivin signature74%34%.005100%75%10%11%69%<.001Reedijk 200530
BMI-regulated signature52%40%.41529%58%20%33%8%.065Glinsky 200577
Overlap (BMI, CNS, and PNS neurospheres)63%12%<.00157%58%10%66%8%.004Glinsky 200577
Overlap (BMI, PNS neurospheres)58%79%.12672%0%0%0%0%<.001Glinsky 200577
Overlap (BMI and CNS Neurospheres)74%100%.00157%50%40%89%8%.002Glinsky 200577
Overlap (mammospheres, neuronal SC)89%69%.114100%92%60%66%74%.149Dontu & Wicha 20055
Overlap (mammospheres, embryonical SC)89%48%.00285%83%30%56%85%.002Dontu & Wicha 20055
Overlap (mammospheres, hematopoietic SC)84%36%<.001100%75%15%44%62%<.001Dontu & Wicha 20055
Overlap (mammospheres, HSC, ESC, and NSC)68%55%.403100%58%50%78%8%<.001Dontu & Wicha 20055
Average (range)74% (52-89%)44% (12-100%).02273% (29-100%)69% (0-92%)24% (0-60%)54% (0-89%)36% (0-85%).017NA

The predominance of basal-like and ErbB2+ breast cancer subtypes in IBC, or more generally, the frequent loss of ER protein expression in IBC, which is a hallmark of dedifferentiation, might explain the observed link between IBC and the mammary stem cell signatures. Several recently published studies indicate an increase in breast cancer stem cells within the basal-like and ErbB2+ breast cancer subtypes.11, 13, 28 Also, samples belonging to the luminal B breast cancer subtype, which is the predominant molecular breast cancer subtype in the ER+ subset of IBC samples (unpublished data), are thought to have elevated mammary stem cell compartments.59 These observations are confirmed when comparing the mammary stem cell signature predictions with the subdivision of the same samples according to the molecular subtypes (Table 1). Basal-like, ErbB2+, and to some extent luminal B samples are predicted to have higher fractions of samples with an elaborate stem cell compartment (basal-like: 74%; ErbB2+: 69%; luminal B: 54%). Conversely, luminal A and normal-like samples are predicted to have lower fractions of samples with an elaborate mammary stem cell compartment (luminal A: 24%; normal-like: 36%). Hence, the predicted elevated stem cell compartment in IBC according to the mammary gene signatures can be a cause or consequence of the predominance of specific molecular breast cancer subtypes.

Stem cell-associated signaling in IBC

The Invasiveness Gene Signature represents 2 major signal transduction pathways: the Iκ, NF-κB pathway and the RAS/MAPK pathway,15 indicating that these pathways play an important role in the molecular biology of mammary cancer stem cells. This finding was confirmed in several studies. Zhou and colleagues demonstrated that inhibitors of the NF-κB signal transduction pathway have a preferential effect on mammary stem cells.60 Cao and colleagues demonstrated that activated NF-κB signaling is required for self-renewal of ERBB2-transformed mammary tumor-initiating cells.61 With respect to the RAS/MAPK pathway, Morel and colleagues have shown that cancer stem cells can be derived from human mammary epithelial cells after the activation of the RAS/MAPK pathway and subsequent epithelial-to-mesenchymal transition.41 In recent studies, we were able to demonstrate that IBC samples are characterized by more frequent activation of the NF-κB and the RAS/MAPK pathway (Fig. 2).62, 63 Lerebours and colleagues confirmed the frequent activation of the NF-κB pathway in IBC samples. They demonstrated a significant up-regulation of 58% of NF-κB target genes in IBC by quantitative reverse transcriptase polymerase chain reaction (RT-PCR).64 Interestingly, we observed an association between NF-κB activation in IBC and activation of the RAS/MAPK pathway, suggesting that RAS/MAPK activation is responsible for NF-κB activation in IBC. Also, it has been demonstrated that activation of NF-κB is responsible for epithelial-to-mesenchymal transition,65 the very same process responsible for the generation of mammary cancer stem cells from more differentiated epithelial cells downstream of RAS/MAPK activation.41 Remarkably, NF-κB activation in IBC is also associated with loss of ER expression, a hallmark of dedifferentiation. Altogether, current data suggest that the RAS/MAPK/NF-κB pathway in IBC is potentially responsible for epithelial-to-mesenchymal transition, which is responsible for dedifferentiation (loss of ER expression) and the generation of mammary cancer stem cells.

Figure 2.

Gene plots for the (A) nuclear factor-kappa B (NF-κB) and (B) RAS/MAPK pathways are shown. In these plots, each gene is represented by a color-coded bar, with a red bar denoting overexpression of the associated gene in inflammatory breast cancer (IBC) and a green bar denoting overexpression of the associated gene in non-IBC (nIBC). The height of a bar corresponds to the standardized gene expression level (Z score; [x − μ]/σ, with x being the raw unstandardized gene expression level for a given gene, and μ and σ being the mean and standard deviation, respectively, of a reference population for the same gene estimated by random class label permutations). When considering a standardized gene expression of 2 (horizontal black line) as the threshold for significance (P<.05), it can be observed that most genes in both signal transduction pathways are differentially expressed between IBC and non-IBC, with elevated expression levels of the majority of these genes in the IBC samples. Globally, the NFκB and RAS/MAPK pathways are significantly associated with the difference between IBC and non-IBC (global test, P = .004 and P = .002, respectively).

To assess the contribution of the developmental signal transduction pathways to the tumor biology of IBC, we performed an analysis on the IBC and non-IBC gene expression data.54 By using the Gene Ontology database, all Gene Ontology identifiers related to BMP, TGFβ, Wnt, Hedgehog, and Notch signaling as well as their associated gene lists were retrieved. Each of these gene lists was mapped onto the lists of informative genes, and all gene lists consisting of at least 2 genes were analyzed for global expression differences between IBC and non-IBC using the global test. This resulted in 8 significant Gene Ontology identifiers representing Hedgehog, BMP, and Notch signaling pathways as well as TGFβ receptor activity. For Wnt signaling, no differences were observed. For each of these significant Gene Ontology identifiers, pathway scores were calculated. The pathway score is a numeric value that represents the extent of activation or inhibition of each signal transduction pathway in each sample. By using univariate analysis, it was demonstrated that IBC samples are characterized by increased TGFβ and Hedgehog signaling, whereas non-IBC samples are characterized by increased BMP signaling (Fig. 3A). In multivariate analysis, it was demonstrated that Hedgehog and BMP signaling are independent predictors of the IBC/non-IBC phenotype in addition to progesterone receptor status and tumor grade. This demonstrates that with respect to the developmental signal transduction pathways, IBC is specifically characterized by a hyperactivation of the Hedgehog pathway and a hypoactivation of the BMP pathway. Interestingly, Hedgehog signaling induces epithelial-to-mesenchymal transition by the up-regulation of JAG2, a Notch ligand, to induce SNAI1 expression via the Notch pathway and by induction of TGFβ1 to induce ZEB1 and ZEB2 via the TGFβ receptor and NF-κB.44 The conjunction of the Hedgehog and TGFβ signaling pathways to induce epithelial-to-mesenchymal transition explains why Hedgehog and TGFβ signaling are not both independent predictors of the IBC phenotype in multivariate analysis.

Figure 3.

Box plots display (A) differential activation of bone morphogenetic protein (BMP) signaling, tumor growth factor-β (TGFβ) signaling, and Hedgehog signaling pathways and (B) differential expression of CDKN1B, FOXO3, and AKT2 between inflammatory breast cancer (IBC) (orange) and non-IBC (nIBC, blue). Hedgehog signaling and TGFβ signaling are more active in IBC compared with non-IBC, whereas BMP signaling is more active in non-IBC compared with IBC. In addition, between estrogen receptor-positive (ER+) IBC and ER+ non-IBC specimens, we observed a significant difference in gene expression for CDKN1B, FOXO3, and AKT2. AKT2 is up-regulated in IBC, whereas CDKN1B and FOXO3 are down-regulated in IBC, which is consistent with the finding by Belguise et al.66, 69 In addition, these data suggest that tumor cells from ER+ IBC samples have a more dedifferentiated phenotype compared with tumor cells from ER+ non-IBC samples, because the AKT2/FOXO3/CDKN1B axis is involved in the maintenance of the epithelial cell phenotype. TGFβ indicates TGFβ receptor.

Another interesting observation was made recently by Belguise and colleagues.66 They identified a novel PKCθ-Akt pathway that leads to deregulation of ER synthesis and activation of NF-κB family members. Also, these authors demonstrated that PKCθ activated AKT, which results in inactivation of FOXO3, leading to decreased synthesis of ER and CDKN1B (p27Kip1). These data agree with our findings regarding the inverse association between NF-κB activation and ER expression.63 Moreover, Gonzalez-Angulo and colleagues have reported loss of CDKN1B expression in >80% of human IBC samples.67 Also, within the same report it was demonstrated that CDKN1B expression was associated with resistance against therapy and survival. In breast cancer in general, CDKN1B expression is associated with resistance against endocrine therapy.68 In addition, preliminary data comparing ER+ IBC samples to ER+ non-IBC samples (matching for ER status) indeed suggest that FOXO3 and CDKN1B are repressed in IBC, whereas AKT2 is overexpressed (Fig. 3B). Interestingly, the FOXO3/ER/CDKN1B axis normally maintains an epithelial cell phenotype, and inactivation of this axis leads, through epithelial-to-mesenchymal transition, to a less differentiated and a more stem cell-like phenotype.69

MicroRNA expression profile of IBC

MicroRNAs are single-stranded noncoding RNAs that play a critical role in regulating gene expression by binding at a target sequence in the 3′ untranslated region of mRNA molecules.70 Previous gene expression profiling results have suggested that at least part of the IBC-specific gene expression pattern is associated with expression of specific microRNAs such as miR-221/-222 and miR-18/-106B/-20.57 Studies with various types of stem cells indicate an intricate network of microRNAs regulating key stem cell-specific signal transduction pathways.71 Hedgehog signaling induces TGFβ, which down-regulates miR-141, -200a, -200b, -200c, -205, and -429, which in turn results in the up-regulation of ZEB1 and ZEB2, resulting in the induction of epithelial-to-mesenchymal transition.44, 72 Also, Yu and colleagues have demonstrated a reduced expression of the Let-7 microRNA family in breast tumor-initiating cells. Infecting breast tumor-initiating cells with Let-7 reduced the mammosphere-forming capacity by targeting, among others, the RAS/MAPK signaling pathway.73 Interestingly, down-regulation of the Let-7 family has been described in mammospheres from an IBC cell line (MDA-IBC-1).

By using quantitative RT-PCR, 384 microRNAs were profiled in a set of 21 IBC samples and 50 non-IBC samples and screened for differential expression between IBC and non-IBC. Among the down-regulated microRNAs in IBC, we identified members of the Let-7-family, miR-200a, -205, and -335. As discussed above, the Let-7 microRNAs are implicated in the maintenance of the stem cell population in breast cancer by targeting the RAS/MAPK pathway,73 miR-200a and -205 are involved in the induction of epithelial-to-mesenchymal transition secondary to Hedgehog and TGFβ signaling, and miR-335 is associated with metastasis-suppressive characteristics, in part because of the regulation of SOX4, a molecule belonging to the SOX-family of DNA-binding proteins that have been associated with stem cell biology.74 Finally, our previous results with regard to overexpression of miR-221 and -222 in IBC are confirmed.54 Both microRNAs are associated with resistance against endocrine therapy in breast cancer by targeting CDKN1B,75, 76 1 of the molecules necessary to maintain an epithelial cell phenotype according to Belguise and colleagues.69 Altogether, the microRNA data globally confirm our previous results with respect to CDKN1B repression and RAS/MAPK, Hedgehog, and TGFβ signaling. Also, these data add considerable extra evidence to associate stem cell biology with IBC.

Conclusions

The finding that IBC is an aggressive tumor with elevated metastatic potential and a dismal clinical course, associated with resistance to therapy, lends credit to the idea that mammary stem cells are implicated in the biology of IBC. From histological (eg, presence of lymphovascular tumor cell emboli) and biological (eg, gene and microRNA expression profiles) perspectives, this hypothesis is further corroborated. Interestingly, the process of epithelial-to-mesenchymal transition appears to take center stage, suggesting that the stem cell characteristics of IBC tumor cells are perhaps a consequence of dedifferentiation instead of IBC tumor cells originating from true mammary stem cells. However, the evidence so far is only circumstantial, and should be validated using alternative techniques that allow for the direct quantification and comparison of the numbers of mammary (cancer) stem cells in IBC and non-IBC. Notwithstanding, the evidence from human tissue samples, in combination with evidence based on cell lines (MDA-IBC-1) and animal models (MARY-X), provides clear writing on the wall that the involvement of stem cells in IBC is a worthwhile topic for further research.

CONFLICT OF INTEREST DISCLOSURES

This supplement was sponsored by the Houston Affiliate of Susan G. Komen for the Cure, the National Cancer Institute, and the State of Texas Rare and Aggressive Breast Cancer Research Program. The First International Inflammatory Breast Cancer Conference was supported in part by GlaxoSmithKline, Pfizer, Eli Lilly and Company, and Cardinal Health.