Epithelial cells from normal human endometrium express a tumor-associated glycoprotein (TAG-72) epitope in vitro Kevin G. Osteen, PhD,.· b Ted L. Anderson, PhD,.· bJoel T. Hargrove, MD: George A. Hill, MD: and Fred Gorstein, MDb Nashville, Tennessee Monoclonal antibody 872.3 identifies a tumor-associated glycoprotein (TAG-72) epitope derived from a human breast carcinoma metastasis. Recently, expression of this epitope was noted in normal endometrium during the secretory, but not proliferative, menstrual interval. In light of known hormonal control of normal endometrial growth and differentiation, we investigated in vitro expression of TAG-72 epitope in purified endometrial epithelium cultured under serum-free conditions on Matrigel biomatrix. Cells from secretory endometrium exhibited homogeneous tumor-associated glycoprotein 72 epitope expression. Unexpectedly, epithelium from the proliferative interval developed expression after 5 to 6 days of culture. Epithelial cells from both intervals maintained expression over 12 days of culture without exogenous estradiol and progesterone. Spontaneous, uniform expression of tumor-associated glycoprotein 72 epitope by normal endometrial epithelial cells in vitro is in marked contrast to the cyclic, heterogeneous expression observed in vivo. Such expression also differs from published in vitro observations of cancer cell lines that express this epitope. (AM J OSSTET GVNECOL 1990;163:479-84.)
Key words: Endometrium, epithelium, matrix, tissue culture, differentiation, tumor epitope Monoclonal antibody B 72.3 is directed against a high molecular weight membrane-associated determinant extracted from a breast tumor metastasis. I Shown to be immunoreactive with carcinomas of the breast, ovary, colon, and uterus,2.5 this antibody was originally reported to be nonreactive with normal adult epithelium from a variety of tissues. 3 • 4 Subsequently, however, monoclonal antibody B72.3 was reported to recognize normal human endometrial epithelium, but only during the secretory menstrual interval.' The antigen recognized by monoclonal antibody B 72.3 has been designated tumor-associated glycoprotein 72 (TAG-72), which has not been well characterized. A partial purification of TAG-72 from a xenograph of a colon cancer cell line (LS-174T) in athymic mice found the epitope to be a neuraminidase-sensitive mucin-like molecule. 6 Recent studies indicate that monoclonal antibody B72.3 reacts with the core reFrom the Department of Obstetrics and Gynecologya and the Department of Pathology: Vanderbilt University School of Medicine. Supported in part by Biomedical Research Support Grant No. RR-05424, American Cancer Society Institutional Research Grant No. IN-25-28, and National Institutes of Health Grant No. HD05797-18 to the Center for Reproductive Biology Research. Presented in part at the Seventy-second Annual Meeting of the Federation of the American Societies for Experimental Biology, Las Vegas, Nevada, May 1-5,1988. Received for publication December 5, 1989; revised March 27, 1990; accepted April 24, 1990. Reprint requests,' Dr. Kevin G. Osteen, Department of Obstetrics and Gynecology, Vanderbilt University School of Medicine, Nashville, TN
37232. 6/1/21798
gion of O-linked sialosyl-2~6o:-N-acetylgalactosaminyl (sialosyl-Tn) epitope. 7 Although the degree ofTAG-72 expression in tumor cell xenografts approximates that observed in tumor masses in situ, this level of expression is not exhibited by tissue cultures of similar tumors. S Maintenance of TAG-72 expression by LS-174T cells in vitro is poor, and appears to depend on the three-dimensional architecture of cells in the culture system; LS-174T cells grown in agar gels expressed a lO-fold higher level of TAG-72 compared with cells grown on plastic. s Our attention has focused on the stage-specific recognition of normal human endometrium by monoclonal antibody B72.3, which promises to be a useful differentiation marker in this tissue and may provide unique opportunities to examine the regulation of TAG-72 epitope expression. To this end, methods have been developed to isolate, with high purity and viability, glandular epithelium from normal human endometrium. 9 When cultured appropriately on biomatrix, these cells were found to establish and maintain an in vivo-like architecture and polarity.'o In the present study we examined the pattern of B72.3-reactive epitope expression in vitro by normal human endometrial epithelium obtained during proliferative versus secretory intervals of the menstrual cycle, and we compared such expression with that observed in vivo.
Material and methods Sources of tissue. Endometrial biopsy samples were obtained from normally cycling women for indications
479
480 Osteen et al.
not associated with uterine dysfunction during the midproliferative (cycle days 5 to 10; n = 7) and mid-to-Iate secretory (cycle days 21 to 26; n = 9) intervals. All samples were obtained and used with prior approval of the Vanderbilt University Committee for the Protection of Human Subjects. Specimens were collected in a 1: 1 mixture of phenol red-free Dulbecco's modified Eagle'S medium and Ham's nutrient mixture F-12 with 15 mol/L hepes, antibiotics, and 10% charcoal-stripped fetal bovine serum (DME/F-12/FBS) for transport to the laboratory. A portion of each sample was placed in 4% buffered formalin for routine histologic evaluation and for immunohistochemical studies as described later. Purification of epithelial cells. Endometrial biopsy tissue was dissected into small pieces (l to 2 mm') and washed by centrifugation (400 g) in fresh medium to remove debris and most blood cells. Epithelial cell isolation was accomplished by a procedure involving selective enzymatic digestion, filtration, unit gravity sedimentation, and selective plating, as previously described." Briefly, tissues were incubated with DME/F12 containing 0.5% collagenase, 0.02% deoxyribonuclease, and 2% chicken serum for 1 hour at 37° C in a shaking water bath. Cell clumps were dispersed by aspiration through a siliconized Pasteur pipette after 30 minutes of digestion. After initial tissue digestion, most stromal cells were separated from large epithelial clumps by filtration through two nylon membrane filters (105 and 88 fLm, respectively). Retained epithelial clumps were washed in Hank's balanced salt solution (pH 7.4) by centrifugation (400 g) before further enzymatic digestion in Hank's balanced salt solution containing 0.5% collagenase, 0.1 % hyaluronidase, 0.1 % pronase, 0.02% deoxyribonuclease, and 2% chicken serum. Subsequent to the second enzymatic digestion clumps of epithelial cells were collected on an 88 fLm nylon mesh filter (allowing contaminating stromal cells to pass), washed twice, and resuspended in 2 ml DME/F-12/FBS. This suspension was layered over 10 ml of DME/F-12/FBS in a 15 ml conical tube for a 30-minute unit gravity sedimentation at 37° C. This step was repeated with only the bottom 2 ml of cells from the first sedimentation. Final purification of epithelial cells was achieved by allowing rapid selective adherence of the few remaining stromal cells to 25 cm2 tissue culture flasks. Epithelial clumps were suspended in DME/F-12/FBS and plated for 30 minutes at 37° C in a humidified environment with 5% carbon dioxide. The nonattached epithelial cells were collected and the procedure was repeated once. Epithelial cell viability was ascertained in a 0.04% solution of trypan blue in phosphate-buffered saline solution (pH 7.2), and an estimate of cell numbers
August 1990 Am J Obstet Gynecol
within each cell preparation was determined with a hemocytometer. Epithelial cell cultures. For routine assessment of culture purity, epithelial cells were cultured for 2 to 3 days as a monolayer on Thermanox tissue culture disks (Miles Laboratories, Naperville, Ill.) placed in four-well culture plates. These cells were evaluated morphologically and with an immunohistochemical probe for the epithelial cell-specific intermediate filament, cytokeratin, as previously described. 9 Otherwise, purified epithelial cells were cultured in Millicell-CM tissue culture plate inserts (Millipore Products Division, Bedford, Mass.) coated with a layer of Matrigel biomatrix (Collaborative Research, Boston, Mass.). Between 1 to 5 X 105 cells/insert were seeded and allowed to attach in DME/F-12/FBS for 48 to 72 hours. Subsequently, cultures were maintained up to 12 days in serum-free DME/F-12 medium supplemented with insulin, transferrin, selenium, bovine serum albumin, and linoleic acid (ITS+; Collaborative Research, Boston). On termination, inserts containing cells cultured on biomatrix were washed in ice-cold phosphate-buffered saline solution and fixed in 4% buffered formalin. Subsequently, the support membranes with attached epithelial cells were carefully removed from inserts with a scalpel and embedded in paraffin for localization of cellular TAG-72 epitope by monoclonal antibody B72.3 immunoreactivity. Immunohistochemistry. The cytokeratin monoclonal antibody PKKI, directed against a 40 to 60 kd cytoskeletal fraction from pig kidney epithelial cells was purchased from Labsystems (Morton Grove, Ill.). Monoclonal antibody B72.3 used for this study was generated in the laboratory of Dr. ] effrey Schlom (National Cancer Institute Section on Tumor Immunology, Bethesda, Md.).1-5 Immunolocalizations of TAG-72 epitope or cytokeratin were accomplished with the avidinbiotin complex technique,l1 with streptavidin to lower background staining. As a negative control, an identical isotype antibody (monoclonal antibody MOPC-21; Litton Bionetics, Charleston, S.C.) was used under the same conditions. Sections from paraffin-embedded biopsy specimens or cells grown on biomatrix were pretreated with heat-inactivated horse serum, washed in phosphatebuffered saline solution, and exposed to primary antibody (1: 100 in phosphate-buffered saline solution containing 1% horse serum) for 1 hour at room temperature. Specimens then were washed and incubated with biotinylated secondary antibody (horse antimouse immunoglobulin G) for 1 hour. After further washing, tissues were exposed to streptavidin-biotin-peroxidase complex for 1 hour and reactivity was visualized with diaminobenzidine. All specimens were counterstained with Gill's hematoxylin and examined by light micros-
Volume 163 Number 2
copy. Immunoreactivity was considered positive when a distinct, brown, granular diaminobenzidine precipitate was evident.
Results Immunoreactivity of endometrial biopsy specimens to monoclonal antibody B72.3. No sample (n = 7) of intact endometrium obtained during the proliferative interval of the menstrual cycle intially showed monoclonal antibody B72.3 immunoreactivity in any region of the tissue (Fig. 1, A). Conversely, all (n = 9) biopsy specimens obtained during the secretory menstrual interval showed immunoreactivity to monoclonal antibody B72.3 in glandular epithelium and, to a lesser extent, luminal surface epithelial cells (Fig. 1, B). Considerable heterogeneity of TAG-72 epitope expression was noted between luminally situated glands (strongest reactivity) and basal glands (no reactivity). Further- . more, some cellular heterogeneity was observed among adjacent luminal glands such that the number of immunoreactive cells among the luminal glands ranged from approximately 10% to >30%. Immunoreactive material was often noted in the lumen of reactive glands. Isolation and assessment of endometrial epithelium. The yield of purified epithelium from the portion of biopsy material available for study did not vary appreciably in relation to either patient age or menstrual interval and consistently ranged between 10 to 15 X 106 cells. Cell viability after purification procedures was routinely >90%. Thus a minimum of 10 culture replicates could be established from each biopsy sample. Epithelial cells cultured on Thermanox tissue culture disks assumed a flattened, whorled configuration after 3 to 4 days, and the purity of these cell populations was assessed by cytoplasmic immunoreactivity to anticytokeratin antibody. Consistent with previous observations; contamination of epithelial cell preparations by stromal or endothelial cells was
Key words: Endometrium, epithelium, matrix, tissue culture, differentiation, tumor epitope Monoclonal antibody B 72.3 is directed against a high molecular weight membrane-associated determinant extracted from a breast tumor metastasis. I Shown to be immunoreactive with carcinomas of the breast, ovary, colon, and uterus,2.5 this antibody was originally reported to be nonreactive with normal adult epithelium from a variety of tissues. 3 • 4 Subsequently, however, monoclonal antibody B72.3 was reported to recognize normal human endometrial epithelium, but only during the secretory menstrual interval.' The antigen recognized by monoclonal antibody B 72.3 has been designated tumor-associated glycoprotein 72 (TAG-72), which has not been well characterized. A partial purification of TAG-72 from a xenograph of a colon cancer cell line (LS-174T) in athymic mice found the epitope to be a neuraminidase-sensitive mucin-like molecule. 6 Recent studies indicate that monoclonal antibody B72.3 reacts with the core reFrom the Department of Obstetrics and Gynecologya and the Department of Pathology: Vanderbilt University School of Medicine. Supported in part by Biomedical Research Support Grant No. RR-05424, American Cancer Society Institutional Research Grant No. IN-25-28, and National Institutes of Health Grant No. HD05797-18 to the Center for Reproductive Biology Research. Presented in part at the Seventy-second Annual Meeting of the Federation of the American Societies for Experimental Biology, Las Vegas, Nevada, May 1-5,1988. Received for publication December 5, 1989; revised March 27, 1990; accepted April 24, 1990. Reprint requests,' Dr. Kevin G. Osteen, Department of Obstetrics and Gynecology, Vanderbilt University School of Medicine, Nashville, TN
37232. 6/1/21798
gion of O-linked sialosyl-2~6o:-N-acetylgalactosaminyl (sialosyl-Tn) epitope. 7 Although the degree ofTAG-72 expression in tumor cell xenografts approximates that observed in tumor masses in situ, this level of expression is not exhibited by tissue cultures of similar tumors. S Maintenance of TAG-72 expression by LS-174T cells in vitro is poor, and appears to depend on the three-dimensional architecture of cells in the culture system; LS-174T cells grown in agar gels expressed a lO-fold higher level of TAG-72 compared with cells grown on plastic. s Our attention has focused on the stage-specific recognition of normal human endometrium by monoclonal antibody B72.3, which promises to be a useful differentiation marker in this tissue and may provide unique opportunities to examine the regulation of TAG-72 epitope expression. To this end, methods have been developed to isolate, with high purity and viability, glandular epithelium from normal human endometrium. 9 When cultured appropriately on biomatrix, these cells were found to establish and maintain an in vivo-like architecture and polarity.'o In the present study we examined the pattern of B72.3-reactive epitope expression in vitro by normal human endometrial epithelium obtained during proliferative versus secretory intervals of the menstrual cycle, and we compared such expression with that observed in vivo.
Material and methods Sources of tissue. Endometrial biopsy samples were obtained from normally cycling women for indications
479
480 Osteen et al.
not associated with uterine dysfunction during the midproliferative (cycle days 5 to 10; n = 7) and mid-to-Iate secretory (cycle days 21 to 26; n = 9) intervals. All samples were obtained and used with prior approval of the Vanderbilt University Committee for the Protection of Human Subjects. Specimens were collected in a 1: 1 mixture of phenol red-free Dulbecco's modified Eagle'S medium and Ham's nutrient mixture F-12 with 15 mol/L hepes, antibiotics, and 10% charcoal-stripped fetal bovine serum (DME/F-12/FBS) for transport to the laboratory. A portion of each sample was placed in 4% buffered formalin for routine histologic evaluation and for immunohistochemical studies as described later. Purification of epithelial cells. Endometrial biopsy tissue was dissected into small pieces (l to 2 mm') and washed by centrifugation (400 g) in fresh medium to remove debris and most blood cells. Epithelial cell isolation was accomplished by a procedure involving selective enzymatic digestion, filtration, unit gravity sedimentation, and selective plating, as previously described." Briefly, tissues were incubated with DME/F12 containing 0.5% collagenase, 0.02% deoxyribonuclease, and 2% chicken serum for 1 hour at 37° C in a shaking water bath. Cell clumps were dispersed by aspiration through a siliconized Pasteur pipette after 30 minutes of digestion. After initial tissue digestion, most stromal cells were separated from large epithelial clumps by filtration through two nylon membrane filters (105 and 88 fLm, respectively). Retained epithelial clumps were washed in Hank's balanced salt solution (pH 7.4) by centrifugation (400 g) before further enzymatic digestion in Hank's balanced salt solution containing 0.5% collagenase, 0.1 % hyaluronidase, 0.1 % pronase, 0.02% deoxyribonuclease, and 2% chicken serum. Subsequent to the second enzymatic digestion clumps of epithelial cells were collected on an 88 fLm nylon mesh filter (allowing contaminating stromal cells to pass), washed twice, and resuspended in 2 ml DME/F-12/FBS. This suspension was layered over 10 ml of DME/F-12/FBS in a 15 ml conical tube for a 30-minute unit gravity sedimentation at 37° C. This step was repeated with only the bottom 2 ml of cells from the first sedimentation. Final purification of epithelial cells was achieved by allowing rapid selective adherence of the few remaining stromal cells to 25 cm2 tissue culture flasks. Epithelial clumps were suspended in DME/F-12/FBS and plated for 30 minutes at 37° C in a humidified environment with 5% carbon dioxide. The nonattached epithelial cells were collected and the procedure was repeated once. Epithelial cell viability was ascertained in a 0.04% solution of trypan blue in phosphate-buffered saline solution (pH 7.2), and an estimate of cell numbers
August 1990 Am J Obstet Gynecol
within each cell preparation was determined with a hemocytometer. Epithelial cell cultures. For routine assessment of culture purity, epithelial cells were cultured for 2 to 3 days as a monolayer on Thermanox tissue culture disks (Miles Laboratories, Naperville, Ill.) placed in four-well culture plates. These cells were evaluated morphologically and with an immunohistochemical probe for the epithelial cell-specific intermediate filament, cytokeratin, as previously described. 9 Otherwise, purified epithelial cells were cultured in Millicell-CM tissue culture plate inserts (Millipore Products Division, Bedford, Mass.) coated with a layer of Matrigel biomatrix (Collaborative Research, Boston, Mass.). Between 1 to 5 X 105 cells/insert were seeded and allowed to attach in DME/F-12/FBS for 48 to 72 hours. Subsequently, cultures were maintained up to 12 days in serum-free DME/F-12 medium supplemented with insulin, transferrin, selenium, bovine serum albumin, and linoleic acid (ITS+; Collaborative Research, Boston). On termination, inserts containing cells cultured on biomatrix were washed in ice-cold phosphate-buffered saline solution and fixed in 4% buffered formalin. Subsequently, the support membranes with attached epithelial cells were carefully removed from inserts with a scalpel and embedded in paraffin for localization of cellular TAG-72 epitope by monoclonal antibody B72.3 immunoreactivity. Immunohistochemistry. The cytokeratin monoclonal antibody PKKI, directed against a 40 to 60 kd cytoskeletal fraction from pig kidney epithelial cells was purchased from Labsystems (Morton Grove, Ill.). Monoclonal antibody B72.3 used for this study was generated in the laboratory of Dr. ] effrey Schlom (National Cancer Institute Section on Tumor Immunology, Bethesda, Md.).1-5 Immunolocalizations of TAG-72 epitope or cytokeratin were accomplished with the avidinbiotin complex technique,l1 with streptavidin to lower background staining. As a negative control, an identical isotype antibody (monoclonal antibody MOPC-21; Litton Bionetics, Charleston, S.C.) was used under the same conditions. Sections from paraffin-embedded biopsy specimens or cells grown on biomatrix were pretreated with heat-inactivated horse serum, washed in phosphatebuffered saline solution, and exposed to primary antibody (1: 100 in phosphate-buffered saline solution containing 1% horse serum) for 1 hour at room temperature. Specimens then were washed and incubated with biotinylated secondary antibody (horse antimouse immunoglobulin G) for 1 hour. After further washing, tissues were exposed to streptavidin-biotin-peroxidase complex for 1 hour and reactivity was visualized with diaminobenzidine. All specimens were counterstained with Gill's hematoxylin and examined by light micros-
Volume 163 Number 2
copy. Immunoreactivity was considered positive when a distinct, brown, granular diaminobenzidine precipitate was evident.
Results Immunoreactivity of endometrial biopsy specimens to monoclonal antibody B72.3. No sample (n = 7) of intact endometrium obtained during the proliferative interval of the menstrual cycle intially showed monoclonal antibody B72.3 immunoreactivity in any region of the tissue (Fig. 1, A). Conversely, all (n = 9) biopsy specimens obtained during the secretory menstrual interval showed immunoreactivity to monoclonal antibody B72.3 in glandular epithelium and, to a lesser extent, luminal surface epithelial cells (Fig. 1, B). Considerable heterogeneity of TAG-72 epitope expression was noted between luminally situated glands (strongest reactivity) and basal glands (no reactivity). Further- . more, some cellular heterogeneity was observed among adjacent luminal glands such that the number of immunoreactive cells among the luminal glands ranged from approximately 10% to >30%. Immunoreactive material was often noted in the lumen of reactive glands. Isolation and assessment of endometrial epithelium. The yield of purified epithelium from the portion of biopsy material available for study did not vary appreciably in relation to either patient age or menstrual interval and consistently ranged between 10 to 15 X 106 cells. Cell viability after purification procedures was routinely >90%. Thus a minimum of 10 culture replicates could be established from each biopsy sample. Epithelial cells cultured on Thermanox tissue culture disks assumed a flattened, whorled configuration after 3 to 4 days, and the purity of these cell populations was assessed by cytoplasmic immunoreactivity to anticytokeratin antibody. Consistent with previous observations; contamination of epithelial cell preparations by stromal or endothelial cells was
