Three-Dimensional Modeling of Breast Cancer Progression

Institution: Lawrence Berkeley National Laboratory
Investigator(s): Carlos Ortiz de Solorzano, Ph.D. -
Award Cycle: 2002 (Cycle VIII) Grant #: 8WB-0150 Award: $329,901
Award Type: STEP Award
Research Priorities
Biology of the Breast Cell>Pathogenesis: understanding the disease



Initial Award Abstract (2002)
The mammary gland is a highly heterogeneous organ. The activity and function of its cells vary between different parts of the gland and as a function of the particular developmental stage. The responsiveness of the cells to external signals, transmitted through hormones and growth factors, is at the very core of this functional diversity.

Breast cancer, a disease of the mammary epithelial cells, is also heterogeneous in its cellular composition. It is our working hypothesis that both the position within the mammary gland and the developmental status play an important role in determining which cells become tumorgenic. Therefore we hypothesize that the timing and locus of breast cancer initiation is determined by the hormone responsiveness of the cells being targeted by the carcinogenic event, and so it is the pattern of growth of the disease. Therefore, we want know which parts of the gland and which developmental stages are more likely to develop the disease. This would put us in a better position to ask whether the cells in those parts/stages display a particular response to the external hormonal and growth signaling.

We will take a quantitative look at the evolving three-dimensional organization and cellular composition (based on hormone responsiveness) of the mammary gland before and after the onset of breast cancer. This involves using a computerized microscopy system to image and reconstruct at multiples level of resolution the mammary gland of engineered (i.e., transgenic) mice. These transgenic mice overexpress an oncogene (c-neu, the mouse variant of the human erbB2) and as a consequence develop long latency mammary tumors which mimic their human counterparts. This fact makes them a perfect model of cancer initiation and progression. We will stage these animals and periodically extract and study the morphology and characteristics of their mammary glands. At the lowest resolution, we will study where in the mammary gland the tumors appear and whether tumor onset is preceded by impaired or abnormal mammary gland development. At the highest level of resolution, we will plot and quantify the cell-by-cell presence and co-localization of key proteins involved in breast cancer, such as the progesterone receptor, estrogen receptor, and the c-neu oncogene.

The end result will be comparable to producing a progressive "atlas" of breast cancer that combines tissue and microscopic morphology along with the distribution of critical breast cancer markers to visualize the disease. We expect that this model of cancer progression will help understanding the pathogenesis and progression of erbB2 positive human cancers, which are normally associated with a very poor prognosis. This will eventually help targeting or repairing the molecular mechanisms that are responsible for the origin and growth of these tumors.


Final Report (2004)
Program note: a discussion with photos comparing mouse and human mammary development and anatomical structures is found on a UC Davis site.

The overall goal that we set out to achieve with this project was to develop a 3-dimensional microscopic model the progression of breast cancer using mice. Our main goal was to determine the location and hormonal responsiveness of the pre- and post-neoplastic areas where cancer appears, and then to follow the spread of the disease within the mammary gland. Our model for this study is a genetically modified mice strain that over-expresses c-neu (i.e., the mouse homolog of the human Her-2 oncogene). These mice are known to develop early pre-neoplastic (benign) lesions that evolve into malignant tumors with a relatively long latency (i.e., less than 2 years, in almost 100% of the animals). Since there are no in vivo (whole animal) imaging methods than can provide morphological information at the level of resolution required for this study, we proposed to examine 24 of these female animals (3 animals every 6 weeks during 48 weeks), extract their mammary glands, and use an existing 3-dimensional computer-aided microscopy system to look for pre-neoplastic and neoplastic lesions. These regions would be located in the context of the entire gland, and we planned to examine the hormonal responsiveness of individual cells within and in the neighborhood of those breast lesions. To determine whether the glands have a predisposition to developing these lesions, due to an abnormal response to hormones, we compared both the morphology and the hormone responsiveness of glands from animals before tumor onset.

At the end of the funding period, we have partially reconstructed 10 glands of wild type animals and 11 glands of c-neu transgenic mice sampled in 6-week intervals. To be able to handle the vast amount of tissue processing required, we have made important software developments aimed at automating the acquisition, registration of entire tissue sections and we have automated the detection of histological structures (mammary ducts and tumor) from stained sections. Comparison between the wild-type (normal) and c-neu transgenic glands shows retarded ductal elongation and abnormal end-bud formation in the transgenic animals. The abnormal end buds give rise to ducts with excessive lateral branching, although not in all the animals, since some have normal ductal structures and some show both normal and abnormal structures. Next, we tried to unravel the hormonal status of the transgenic glands, based on their expression of the estrogen receptor (ER) and progesterone receptor (PR), and compared these to of wild-type animals. Previous studies have shown that PR is usually co-expressed with ER-? in non-proliferating mammary epithelial cells, and it has been suggested that estrogen signaling though ER-? induces PR expression. Her-2 and ER-?, however, are mutually antagonistic, resulting in over-expression of only one receptor at a time. Our data is consistent with this observation, showing that when Her-2 is over-expressed, an inhibition of ER-? results, when comparing the ER abundance in transgenics with those of same-age wild-type animals. Interestingly, the transgenic mice showed a dramatic increase of PR expression in mammary terminal end bud regions compared to the wild- type, with no appreciable change of ductal expression. The excessive lateral branching and precocious (i.e., happening earlier in development) lobuloalveolar morphogenesis associated with the c-neu phenotype is consistent with this finding, providing further evidence that PR is responsible for these two morphological events. According to current models of normal receptor expression, we had expected that the down-regulation of ER-? to result in lower levels of PR. Our results show higher levels of PR in the TEBs, suggesting that an alternate aberrant pathway, Her-2 up-regulation of PR, could then induce the phenotypic abnormalities of the transgenic ductal tree.

This work will continue, and we eventually hope to shed light on the pre-neoplastic events in both mammary morphology and hormone receptor expression that predisposes an individual to breast cancer.


Symposium Abstract (2003)
Introduction: The aim of this study is to determine if there is a particular morphological environment and/ or developmental stage that favors both initiation and progression of breast cancer, and if that environment can be associated with a pattern of hormone and growth factor receptor status.

Methods: We use wild-type and transgenic c-neu over-expressing mice that develop long latency tumors with a high penetrance (24-48 months). This strain is a good model for human erbB2-positive breast cancers, which are usually associated with a poor prognosis. Three nulliparous animals of each type will be sacrificed every 6 weeks for 48 weeks. We combine H&E staining and immunohistochemis-try to quantify, cell-by-cell, the levels of expression of the receptors at consecutive time points before and after cancer onset. Among all receptors, we chose those for estrogen (ERa), progesterone (PR) and erbB2 because of their clinical relevance as markers for breast cancer prognosis. The distribution of receptors in the wild type animals will be used as a control for the study done on the c-neu model. We use an existing computerized three-dimensional microscopy system and automated image analysis to virtually reconstruct the glands from serial tissue sections integrating the receptors status information.

Preliminary Results: The first collection of the glands started on the 7/9/02 and will end around the 7/7/ 03. We began the reconstruction of the first glands using sections stained for histology (H&E). We set up the protocol for immunostaining of ERa, PR and erbB2 and we are now working on the analysis of the distribution of PR in the first gland. Our preliminary results allowed us to measure the feasibility of the project and to develop and adapt the three-dimensional system to this purpose.

Discussion and perspectives: From our results we will first compare the mammary gland three-dimensional structure and patterns of receptor expression of wild type versus c-neu mice, at periodic stages in the development of the animals, before and after tumor onset. Based on that analysis, we will determine a) if abnormal morphology is an indicator of neoplasia, and identify the particular locus and timing of cancer initiation and b) if there is a correlation between the locus and time of cancer initiation and a particular pattern of receptor status. We will create a model of cancer initiation and progression that combines receptor status, and both micro and macroscopic histology. This will have an important impact in understanding carcinogenesis and neo-plastic progression in breast cancer.


Symposium Abstract (2005)
Our aims were to study the role of hormone receptors in normal mammary gland development and neoplasia. To this end we have compared the morphology and cell-by-cell hormone receptor status of the mammary gland of wild type mice versus those of genetically modified (transgenic) mice known to develop early pre-neoplastic (benign) lesions that evolve into malignant tumors with a relatively long latency (less than 2 years, in almost 100% of the animals).

Standard imaging methods cannot provide morphological and molecular information at the level of resolution required for this study.. Thus, we sampled female animals (6 weeks during 48 weeks), extracted and fully sectioned their mammary glands, and used a three-dimensional computer-aided microscope to reconstruct the morphology of the glands to analyze the receptor status at the cellular level. Specifically, we have reconstructed 10 glands of wild-type animals and 11 glands of Her2 transgenic mice sampled in six week intervals. To be able to handle the vast amount of tissue processing required, we have made important software developments aimed at automating the acquisition and registration of entire tissue sections. We have automated the detection of histological structures (mammary ducts and tumors) from H&E stained sections.

Comparison between the wild-type and transgenic glands showed a retarded ductal elongation and abnormal end-bud formation in the transgenic animals. The abnormal end buds give rise to ducts with excessive lateral branching, although not in all the animals, since some presents normal ductal structures and some present both normal and abnormal structures. Previous studies have shown that PR is usually co-expressed with ER-alpha in non-proliferating mammary epithelial cells, and it has been suggested that estrogen signaling though ER-alpha induces PR expression. HER2 and ER-alpha however, are mutually antagonistic, resulting in overexpression of only one receptor at a time, with the subsequent downregulation of the other. Our data is consistent with this observation, showing that when HER2 is overexpressed, an inhibition of ER-alpha results, when comparing the ER numbers with those of same-age wild type animals. Interestingly, the transgenic mice showed a dramatic increase of PR expression in the terminal end buds compared to the wild type, with no appreciable change of ductal expression.

The excessive lateral branching and precocious lobuloalveolar morphogenesis associated with the c-neu phenotype is consistent with this finding, providing further evidence that PR is responsible for these two morphological events. According to current models of normal receptor expression, we had expected that the downregulation of ER-alpha to result in lower levels of PR. Our results show higher levels of PR in the terminal end buds, suggesting that an alternate aberrant pathway, HER2 upregulation of PR, could then induce the phenotypic abnormalities of the transgenic mammary ductal tree.

Automatic segmentation of histological structures in mammary gland tissue sections.
Periodical:Journal of Biomedical Optics
Index Medicus: J Biomed Optics
Authors: Fernandez-Gonzalez R, Deschamps T, Idica A, Malladi R, Ortiz de Solorzano C.
Yr: 2004 Vol: 9 Nbr: 3 Abs: Pg:444-53