A New Model for Inflammatory Breast Carcinoma

Institution: University of California, Los Angeles
Investigator(s): Sanford Barsky, M.D. -
Award Cycle: 1999 (Cycle V) Grant #: 5JB-0104 Award: $200,000
Award Type: IDEAS II
Research Priorities
Biology of the Breast Cell>Pathogenesis: understanding the disease

Initial Award Abstract (1999)
Inflammatory breast cancer is an unusual type of the disease that is characterized by a very high degree of invasion into the blood and lymphatic vessels. We have established the first human model of this form of the disease, a transplantable inflammatory carcinoma xenograft model called MARY-X. This is a human tumor obtained from a patient with inflammatory breast carcinoma, which grows in mice similar to humans. This allows us to study the detailed molecular mechanisms that regulate inflammatory carcinoma. Our specific interest is the invasion of cancer cells into the blood and lymphatic vessels. MARY-X is extremely aggressive in this invasion process, and it even turns the skin of the mouse bright red just like inflammatory cancer does in humans.

We compared both MARY-X and MARY (the corresponding human cell line) with non-inflammatory breast carcinoma cell lines and xenografts for molecules including (1) angiogenic factors (molecules which cause blood vessel formation), (2) proteolytic enzymes (molecules which help cancer cells spread), and (3) adhesion molecules (molecules which help cancer cells stick to one another and to blood vessels) to determine differences that could account for the inflammatory phenotype. We found that angiogenic factors and proteases do not account for these differences. However, we noted a marked overexpression of MUC1 (a glycoprotein mucin found in the cell membrane of a variety of tumor types) and surprisingly E-cadherin (a cell-cell adhesion molecule). E-cadherin has been shown to be lost in most breast cancers rather than overexpressed, so our observation was quite surprising.

Based on these initial studies, we will pursue two aims designed to advance our knowledge of the mechanism of the inflammatory phenotype. First, we plan functional studies designed to examine whether MUC1/E-cadherin expression actually causes or contributes to the inflammatory phenotype. We will alter the expression of these proteins using both gene transfection and neutralizing antibodies and determine the effect on lymphovascular invasion in mice. Secondly, we plan a molecular comparison of inflammatory carcinoma vs. other non-inflammatory carcinomas using a technique of ‘differential display’ to identify those genes that are uniquely expressed or not expressed in inflammatory carcinoma.

Our goal is to gain further understanding the characteristics of inflammatory breast carcinoma and what sets this deadly disease of women into motion.

Final Report (2002)
Note: This grant was extended 1-yr to complete the aims and budget expenditures.

Introduction: Inflammatory breast cancer (IBC) is a poorly understood, little studied form of breast cancer which is very aggressive and particularly devastating in disadvantaged minority women. IBC is characterized by florid tumor emboli within lymphovascular spaces, a phenotype which distinguishes it from other forms of breast cancer.

Topic Addressed: Using a novel human-SCID (i.e., immunodefient mice) model of IBC, we have conceptually analyzed this phenotype in three parts.

Progress toward Specific Aims: First, the tumor cell embolus (IBC spheroid) forms on the basis of an intact and overexpressed E-cadherin / alpha-, beta-catenin axis which mediates tumor cell tumor cell adhesion analogous to the embryonic blastocyst and accounts for both the compactness of the embolus and its complete dissolution with anti-E-cadherin antibodies, absent calcium, or E-cadherin dominant-negative mutant approaches. Dissolution of the tumor cell embolus by any of these approaches induces apoptosis via an anoikis pathway. The compactness of the embolus results in its resistance to chemotherapy/radiation therapy and its efficiency at metastasis formation and therefore therapeutic strategies which disadhere it are highly desirable. Secondly, the tumor cell embolus (IBC spheroid), in contrast, fails to bind the surrounding vascular endothelial cells because of complete absence of sialyl-Lewis X/A carbohydrate ligand-binding epitopes on its overexpressed MUCl which are necessary for binding endothelial cell E-selectin. This natural tumor cell-endothelial cell aversion of the tumor cell embolus (IBC spheroid) further contributes to the compactness of the IBC spheroid and its passivity in metastasis dissemination. Experiments with purified glycosyltransferases which add sialyl-Lewis X/A to MUC1 on the IBC spheroids, produce strong electrostatic repulsions which disrupt the E-cadherin homodimers and cause disadherence. Finally, The tumor cell embolus finds itself within the vascular lumen in the first place because it stimulates a vascular channel to form around it rather than intravasating into a pre-existing lymphatic or capillary. The enveloping vascular channel does not form from angiogenesis, but rather froms vasculogenesis, as evidenced by experiments where tumor cell emboli (IBC spheroids) are admixed with murine embryonal fibroblasts labeled with green fluorescent protein (GFP) and injected into SCID mice. Tumor emboli are observed within lymphovascular spaces where the endothelial cells express vasculogenesis markers as well as endothelial markers. These endothelial cells also express GFP, evidence that they must have formed from the injected GFP-labeled murine embryonal fibroblasts.

Further details on our results can be referenced in the publications listed below.

Future Directions and Impact: The molecular basis of IBC provides a mechanism by which IBC bypasses the traditional steps of intravasation, dissemination and extravasation in its metastasic pathway and affords opportunities for therapeutic intervention.

Symposium Abstract (2005)
Lymphovascular invasion (LVI) is a rate-limiting step in the spread of breast cancer through the body (metastasis) and consists of clumps of tumor cells, called emboli, within a lymphatic or blood vessel channel. The mechanisms behind LVI remain unknown and the goals of this study were to better understand this process. We decided first to use our previously developed human model of inflammatory breast cancer (MARY-X) which exhibited florid LVI in the mouse to address these mechanisms. We then investigated whether these same mechanisms occurred in humans with breast cancer. In the mouse, we conducted single and dual-color fluorescent imaging studies to gain insights into the dynamics of the process. In the human studies, we identified patients who had received an allogeneic bone marrow transplant from a male donor for leukemia and who years later developed an unrelated breast cancer. In the mouse experiments, unlabelled MARY-X tumoral spheroids, co-injected with either GFP- or RPF-murine embryonal stem cells (labeled with green or red tags) gave rise to tumoral nodules which exhibited intense fluorescence. Examination of these tumoral nodules revealed florid LVI in which either GFP or RFP immunoreactivity was present within the tumoral stroma as well as the endothelial lining cells (confirmed by PECAM-1, a marker of blood vessels) which surrounded the unlabelled tumoral emboli. Adjacent murine angiogenesis could be grossly visualized as dark vascular structures which coursed over a fluorescent tumor nodule. Unlabelled MARY-X spheroids, co-injected with RFP-murine stem cells, into GFP-transgenic nude mice, gave rise to nodules exhibiting a central zone of red fluorescence followed temporally by a zone of circumferential green fluorescence and then an evolving intermediate zone of yellow fluorescence. These gross findings were confirmed by microscopic evidence of encircling stem cell vasculogenesis, adjacent angiogenesis and bridging anastomoses between the two vessel types. In a second set of experiments designed to examine whether the bone marrow was a source of these stem cells, marrow from male GFP-transgenic nude mice were transplanted into lethally - irradiated female nudes. Mice exhibiting marrow engraftment were co-injected with unlabelled MARY-X spheroids, and RFP-murine stem cells as before. GFP, RFP and hybrid fluorescence was present within the tumoral stroma as well as the endothelial lining cells. These results were confirmed in humans: examination of their breast cancer tissues with X and Y chromosome-specific probes revealed that a fraction of their stromal and endothelial cells were of male origin and hence derived from the bone marrow. These studies were confirmed by chromosome specific sequence PCR studies performed on the stroma and endothelial cells which were microdissected (removed) from their tissues.

Our animal imaging and human studies together indicate that the LVI story is really the story of the “ship in the bottle.” The evidence of dual bone marrow- and locally-derived stem cell vasculogenesis with secondary angiogenic anastomoses suggests that the bottle is built around the ship and built from both local as well as systemic building blocks. This new paradigm suggests the need for new strategies of delivering therapeutic agents that block LVI and subsequent metastasis.

Symposium Abstract (2007)
Inflammatory breast cancer (IBC) is a type of human breast cancer with high metastasis due to the presence of florid lymphovascular tumor emboli. We have recapitulated this phenotype in a human xenograft mouse model of IBC (MARY-X). In this xenograft, the tumor emboli are mediated by an overexpressed E-cadherin / alpha and beta-catenin axis and grow in vitro as spheroids. These spheroids are very similar to the embryonal blastocyst which also overexpresses E-cadherin. When the spheroids were disrupted with anti-E-cadherin, the disadhered cells underwent apoptosis. When undisrupted, these spheroids and corresponding emboli resist the apoptosis-inducing effects of chemotherapy / radiotherapy. Because of the resemblance of these spheroids to the embryonal blastocyst, the source of embryonal stem (ES) cells, we reasoned that the spheroids might not only contain stem cells but express stem cell-associated signaling which might be exploited therapeutically. We indeed found known stem cell markers (Stellar, H19, Rex-1, Nestin) to be highly expressed within the MARY-X spheroids. In addition, RT-PCR analyses of MARY-X revealed the expression of OCT4, SOX2, and Nanog, transcriptional determinants essential for the pluripotency and self-renewal of stem cells. Most importantly, CD133, a surface marker for human leukemic / gliomic stem cells, was also strongly expressed by 60% of the cells of the MARY-X spheroids and the lymphovascular emboli of human IBC. Only this CD133+ subpopulation expressed both the stem cell transcriptional determinants mentioned previously as well as the capacity for self-renewal and tumorigenicity. Tumorigenicity capacity could be observed within spheroids with as few as 50 cells or within single cells if grown in and co-injected with Matrigel. These stem cells also exhibited stem cell-associated signaling. Both increased expression of notch receptors and increased notch signaling via notch cleavage and downstream increased expression of target genes, HES-1, HES-5, were detected. In addition, gamma-secretase inhibitors, inhibitors which effectively blocked notch cleavage and downstream signaling of target genes, caused marked apoptosis of the CD133+ IBC cells in vitro, even when they were cytoprotected within their embolic niche. This effect was also observed in vivo with marked shrinkage of the size of tumor xenograft nodules with gamma-secretase inhibitor treatment. This anti-tumor effect was independent of the E-cadherin / alpha and beta-catenin axis which remained intact. Our results suggest that activation of notch signaling pathways are essential for the maintenance of viable tumor emboli and CD133+ "stem cells", in particular, and that inhibiting these notch pathways with gamma-secretase inhibitors may achieve selective targeting of stem cells within the lymphovascular emboli and ultimately a novel and effective therapy for IBC.

Symposium Abstract (2010)

Background: The existence of stem cells in human cancers has been inferred by clonality experiments and marker studies in vitro and clinical observations in vivo concerning tumor recurrences and emerging drug resistance. We have studied the existence of stem cells in a human model of inflammatory breast cancer, termed MARY-X, a model which forms spheroids which are similar to normal tissue stem cell-derived mammospheres. Comparing MARY-X with common non-IBC breast carcinoma and normal cell lines, we found specific embryonal stem cell markers within the MARY-X spheroids. RT-PCR analyses of MARY-X also revealed the expression of transcriptional determinants essential for the pluripotency and self-renewal of human embryonal stem cells. Since MARY-X spheroids, when injected into mice, form tumors with florid lymphovascular emboli, we wondered whether tumor emboli from actual human cancers also recapitulated a mammosphere stem cell phenotype.

Design: We carried out laser capture microdissection of tumor emboli in 200 cases of human breast cancer including cases of both infiltrating ductal as well as lobular cancer. These cases included all of the common molecular classes of breast cancer including luminal A, luminal B, luminal C, Her-2/neu positive and triple negative including the basal subtype. The common denominator which was studied in all of these cases was the lymphovascular embolus.

Result: By both RT-PCR and IHC, lymphovascular emboli exhibited five-ten fold higher stem cell markers including Stellar, H19, Rex-1, Nestin, CD133 and Aldehyde Dehydrogenase 1 (ALDH1) as well as stem cell transcriptional determinants including OCT4, SOX2, and Nanog than the non-embolic tumoral areas. In addition, stem cell signaling pathways specifically involved in self-renewal and pluripotency including Bmi-1, Hedgehog and Notch 3 were activated selectively within the lymphovascular tumor emboli. These observations held true irrespective of the molecular class of breast cancer from which the embolus was derived and irrespective of the adhesion status (presence or absence of E-cadherin) of the embolus.

Conclusion: Our findings indicate that the lymphovascular embolus is not simply a cellular fragment that detaches from the main tumor but rather represents a selection for a stem cell phenotype. This finding may explain the increased resistance of lymphovascular tumor emboli to chemotherapy and the decreased disease-free survival and poorer prognosis exhibited by patients with significant lymphovascular emboli.

A novel human xenograft model of inflammatory breast cancer
Periodical:Cancer Research
Index Medicus: Cancer Res
Authors: Alpaugh ML, Tomlinson JS, Shao ZM, Barsky SH
Yr: 1999 Vol: 59 Nbr: 20 Abs: Pg:5079-84

An intact overexpressed E-cadherin/alpha,beta-catenin axis characterizes the lymphovascular emboli of inflammatory breast carcinoma
Periodical:Cancer Research
Index Medicus: Cancer Res
Authors: Tomlinson JS, Alpaugh ML, Barsky SH
Yr: 2001 Vol: 61 Nbr: 13 Abs: Pg:5231-41

Cooperative role of E-cadherin and sialyl-Lewis X/A-deficient MUC1 in the passive dissemination of tumor emboli in inflammatory breast carcinoma
Index Medicus: Oncogene
Authors: Alpaugh ML, Tomlinson JS, Kasraeian S, Barsky SH
Yr: 2002 Vol: 21 Nbr: 22 Abs: Pg:3631-43

Reversible model of spheroid formation allows for high efficiency of gene delivery ex vivo and accurate gene assessment in vivo
Periodical:Human Gene Therapy
Index Medicus: Hum Gene Ther
Authors: Alpaugh ML, Barsky SH
Yr: 2002 Vol: 13 Nbr: 10 Abs: Pg:1245-58

Relationship of sialyl-Lewis(x/a) underexpression and E-cadherin overexpression in the lymphovascular embolus of inflammatory breast carcinoma
Periodical:American Journal of Pathology
Index Medicus: Am J Pathol
Authors: Alpaugh ML, Tomlinson JS, Ye Y, Barsky SH
Yr: 2002 Vol: 161 Nbr: 2 Abs: Pg:619-28