Jerry A. Peterson

Characterization of cell surface antigens of human mammary epithelial cells with monoclonal antibodies prepared against human milk fat globule

Hybridomas have been prepared that secrete monoclonal antibodies against three different surface antigens of normal human mammary epithelial cells by fusion of mouse myeloma cells with spleen cells from mice and rats immunized with delipidated human milk fat globules. Using a novel method for molecular weight determination, the three different monoclonal antibodies, BLMRL-HMFG-Mc3, BLMRL-HMFG-McR2, and BLMRLHMFG-Mc5, were found to identify molecules with apparent molecular weights of 46,000, 70,000, and 400,000 daltons, respectively. The latter is a mucin-like glycoprotein with a high sugar content and has not previously been described as a component of the human milk fat globule or of human mammary epithelial cell membranes. Single-cell quantitation of binding of monoclonal BLMRL-HMFG-Mc5 to three breast tumor cell lines using a Microscope Spectrum Analyzer and indirect immunofluorescence revealed a heterogeneous expression. Further, using a competitive radioimmunoassay, it was found that breast tumor cell lines differed by at least 10-fold in the 400,000-molecular-weight antigen content. None of the three antigens are detectable on several nonbreast cell lines, including normal breast fibroblasts.

CHARACTERIZATION OF CELLS CULTURED FROM EARLY LACTATION MILKS

SummaryTwo major types of cells can be cultured from early lactation human milks: a colony-forming epithelial cell and an adherent nondividing cell referred to as a foam cell The epithelial cells show a positive reaction with a specific antiserum reactive against membrane components of the milk fat globule, whereas the foam cells do not. The nondividing foam cells are phagocytic and can be killed by silica particles; they produce lysozyme, are resistant to trypsinization, and have Fc receptors. These properties, together with the lack of reaction with antiserum to the milk fat globule membrane, suggest that the foam cells are not terminally differential epithelial cells, but tissue macrophages.

Quantitation of human mammary epithelial antigens in cells cultured from normal and cancerous breast tissues

SummaryA sensitive radioimmunoassay technique was developed to quantitatite the level of human breast celltype specific antigens on cells from normal breast and from various established cell lines of breast and nonbreast origins. Polyacrylamide gel electrophoresis revealed four major proteinaceous components (150,000; 75,000; 60,000; and 48,000) in human milk fat globule membranes that were used to immunize rabbits in order to elicit antimammary epithelial cell antibody. Antisera obtained were rendered specific by abosorptions and were able to recognize three specific mammary epithelial components of the breast epithelial cell. Human mammary epithelial (HME) antigen expression was highest (1290 ng/106 cells) in normal breast epithelial cells from primary cultures of normal breasts. Lower levels (range: 955 to 330 ng/106 cells) were found in breast epithelial cells from cell lines established from cancerous breast tissue. Cells of nonbreast origins as well as fibroblasts from breast gave much lower values (less than 30 ng/106 cells). On treatment, with trypsin, of two breast epithelial cell lines (MDA-MB-157 and MCF-7) 80 to 85% of their HME antigen expression was lost, suggesting that a majority of these breast antigens reside on the cell surface.

The widespread nature of phenotypic variability in hepatomas and cell lines, in the form of a geometric series☆

The phenomenon of geometric phenotypic variability is described and its widespread occurrence is established by a new analysis of data from a literature survey of quantitative variation in 39 different enzymes and other cell products in hepatomas and cell lines. The range of variation from hepatoma to hepatoma or from cell line to cell line was between 3- and 700-fold, depending on the particular cell product. By collating together and normalizing the data for the enzymes and other cell products surveyed, it was demonstrated in a statistically valid manner that the quantitative variation for most, if not all, of the enzymes and serum albumin was not random, but followed a geometric series, the consecutive terms of which differed by a factor of square root 2. In addition, examples are presented to show that quantitative inheritance in normal tissues also occurs along this geometric series.

Characterization of differentiation antigens of the mouse mammary epithelial cell (MME antigens) carried on the mouse milk fat globule

The characterization of mouse mammary epithelial (MME) antigens from the mouse milk fat globule (MMFG) is described. Heterologous antisera against fat-free MMFG material identifies by indirect immunofluorescence techniques, cell-type-specific MME antigens on the surface of mouse mammary epithelial cells and MMFGs and not on cells of other tissues. Mammary epithelial cells of all mouse strains tested were positive. However, the intensity of staining differed. Specificity of anti-MME for MME antigens is demonstrated by competition and cross-absorption experiments. The MMFG is composed of 5 main proteinaceous components, 3 of which contain carbohydrate and which are shown to contain the specific MME antigens.

Analysis of variability in albumin content of sister hepatoma cells and a model for geometric phenotypic variability (quantitative shift model)

A model (quantitative shift model) is presented that can account for the phenomenon termed geometric phenotypic variability in which quantitative variation occurs along a √2-fold geometric series. The model is based on data for variability in albumin content in hepatoma cells and has three basic assumptions: (i) The albumin genes on each chromosome are active and under independent quantitative regulation, (ii) The rate of albumin gene transcription per cell is the sum of the rates of transcription of the genes of each chromosome. (iii) The mechanism that controls the rate of transcription is highly variable so that at each cell cycle there is a high probability (P=0.1–0.3) that on the newly synthesized chromatid the rate of transcription of the albumin gene will be different from that of the old chromatid. However, if the rate is different, it will usually be either half or twice the levels of the old chromatid. A computer program was developed based on the model that can generate quantitative diversity in single cells that mimics the pattern of variability in albumin content between sister hepatoma cells, among cells in clonal colonies, and in cell populations where the periodicities in single-cell distributions are compared by Fourier transform analysis. It was determined that the rate of phenotypic variability is indirectly proportional to the magnitude of the quantitative shift in albumin content and that it can be as high as 0.3 per cell per generation. Also, the √2 factor in geometric phenotypic variability appears to be an average of a family of values and not the actual value of the smallest quantal shift.

The removal of cell surface material by enzymes used to dissociate mammary-gland cells.

SummaryTreatment of mouse mammary epithelial cells (MMEC) with various enzymes used for dispersing and transferring cells results in extensive digestion of materials on the cell surfaces. MMEC biosynthetially labeled with [3H]fucose, [14C]fucose and [3H]amino acids or with125I by the lactoperoxidase method were exposed to either collagenase plus hyaluronidase, followed by pronase, or to trypsin in concentrations and conditions currently used for cell dispersion. Whereas the latter enzyme preparation solubilized 76% of the trichloroacetic acid precipitable radioactive fucose and 96% of the protein-bound125I, collagenase plus hyaluronidase treatment released lesser amounts of each label. Subsequent treatment of the cells with pronase removed additional surface-labeled materials, but the total amounts released were still less than when the trypsin preparation alone was employed. Released cell surface materials were analyzed by gel chromatography. Some of the peaks obtained also were examined by polyacrylamide gel electrophoresis. The labeled materials that remained attached to the MMEC after enzymatic treatment were investigated by these two methods as well. We could show that collagenase plus hyaluronidase solubilized three main glycoprotein components from the cell surface. In addition, we could show that the extensive cell surface damage caused by these two enzyme preparations was due to the high proteolytic activity present in these preparations as judged by their ability to hydrolyze rabbit gamma globulin labeled with125I. Even though their membranes were extensively damaged by the enzyme treatments, the dispersed cells could be cultured successfully in vitro and could incorporated fucose into their surfaces in a manner similar to that by intact tissue. Through the use of gel-filtration (cochromatography of [14C]fucose and [3H]fucose cell surface materials), we could demonstrate the identity of cell surface glycoproteins synthesized by cultured cells and by intact tissue.

Human mammary epithelial antigens (HME-Ags) in the circulation of nude mice implanted with a breast tumor and non-breast tumors

Human mammary epithelial antigens (HME‐Ags) obtained from the membrane of the human milk fat globule (HMFG) were tested for their possible role as breast tumor markers. Specific antisera raised against HME‐Ags were used to monitor plasma concentrations of these antigens in nude mice implanted with a human breast tumor. The level of plasma HME‐Ags, determined by radioimmunoassay, was significantly higher in animals transplanted with a human breast tumor (mean ± standard error; 687 ± 184 ng/ml) than those with other types of human tumors (colon carcinoma: 50 ± 29; lung carcinoma: 82 ± 78; medulloblastoma: <30; and Wilson melanoma: <30) and healthy control animals (49 ± 22). Removal of the breast tumor resulted in a significant drop of HME‐Ags level to “background” values, suggesting that animals with the breast tumor did release into the circulation HME‐Ags which could be possibly used as breast‐tumor markers in breast tumor diagnosis.

Quantitative phenotypic variation in single normal and malignant cells from liver and breast occurs along a geometric series

Single cell variability in albumin content of normal rat hepatocytes and hepatoma cells and in a specific breast surface antigen content in normal and malignant human mammary epithelial cells was studied by a quantitative immunoperoxidase method. The range of variability was nearly 10-fold for both normal and malignant cells. This wide range of single cell heterogeneity was generated very rapidly in clonal colonies of less than 30 hepatoma cells. By a grid test for periodicity that we devised, and also by Fourier transform analysis, the distribution of albumin content in single hepatocytes and hepatoma cells, and of a breast surface antigen content in normal and malignant breast cells, was shown to be discontinuous, where the cells distributed with a periodicity that fit a geometric series, of which consecutive values differed by a factor of √2. The fit to this geometric series was best for normal cells. Significant quantal shifts occurred at a high rate (>10−2 per cell per generation) and to higher and lower levels with equal probability. These results demonstrate that geometric phenotypic variability is a normal phenomenon and should be taken into account when studying gene regulation and population dynamics.

Glycosyl transferases in mouse and human milk fat globule membranes

SummaryMembranes isolated from mouse and human milk fat globules were found to contain the enzymes responsible for the synthesis of dolichol monophosphate mannose and dolichol monophosphate glucose as well as those involved in the transference of the glycosyl residues from the two dolichol derivatives to dolichol diphosphate oligosaccharides. The levels of most of the enzymes were comparable to those found in mouse mammary gland microsomes. The presence of enzymes involved in protein glycosylation via dolichol derivatives in the milk fat globule membrane provides evidence in favor of an outward flow of membrane components from the rough endoplasmic reticulum, where these enzymes are active in vivo, towards the cell surface.