Regulation of alternative splicing of CD45 by antagonistic effects of SR protein splicing factors.

CD45 is a transmembrane glycoprotein possessing tyrosine phosphatase activity, which is involved in cell signaling. CD45 is expressed on the surface of most leukocytes and can be alternatively spliced by the inclusion or skipping of three variable exons (4, 5, and 6 or A, B, and C) to produce up to eight isoforms. In T cells, the splicing pattern of CD45 isoforms changes after activation; naive cells express high m.w. isoforms of CD45 which predominantly express exon A (CD45RA), whereas activated cells lose expression of exon A to form low m.w. isoforms of CD45 including CD45RO. Little is known about the specific factors controlling the switch in CD45 splicing which occurs on activation. In this study, we examined the influence of the SR family of splicing factors, which, like CD45, are expressed in tissue-specific patterns and have been shown to modulate the alternative splicing of a variety of transcripts. We show that specific SR proteins have antagonistic effects on CD45 splicing, leading either to exon inclusion or skipping. Furthermore, we were able to demonstrate specific changes in the SR protein expression pattern during T cell activation.

Alternative splicing of CD45 pre-mRNA is uniquely obedient to conditions in lymphoid cells.

The leucocyte common antigen (LCA or CD45) consists of various isoforms generated by alternative splicing of variable exons 4, 5 and 6 (or A, B and C). To follow splicing behaviour in different cell types we developed a human CD45 mini-gene and analysed its expression in transfected cell lines and transgenic mouse tissues. In Cos-1, HeLa and 3T3 cells we found distinct expression patterns which could only be modulated slightly by protein synthesis inhibitors but not by variation in culture conditions like pH, serum concentration and cell density, or by stimulation with phorbol ester (TPA). In all non-lymphoid transgenic tissues the default splicing pattern (CD45R0) was found, while the expression profile in lymphoid cells, where all eight isoforms are present, mimics that of the endogenous mouse LCA gene products. Next, to examine the factors involved in alternative exon use we analysed the expression pattern of members of the family of SR proteins, well known splicing regulators with arginine/serine-rich (R/S) domains. Cell lines expressed variable levels of SRp75, SRp30 and SRp20 and constant amounts of SRp40. Mouse tissues expressed large amounts of SRp75, SRp55 and SRp40, additional expression of SRp30s and SRp20 was restricted to lymphoid tissues. Therefore, SRp30 and SRp20 may contribute to forming the appropriate cellular conditions for alternative use of CD45 exons 4-6 in the haematopoietic compartment.

Reduced expression of protein tyrosine phosphatase gamma in lung and ovarian tumors.

Based on LOH studies protein tyrosine phosphatasegamma (PTPgamma) has been suggested as a candidate tumor suppressor gene involved in the oncogenesis of lung and renal cancers. In order to assess the involvement of PTPgamma in tumor development we developed a PTPgamma-specific monoclonal antibody (gammaTL1) (IgM isotype) by immunization with a synthetic peptide of 15 amino acids corresponding to the amino acid sequence nos. 1423-1438 just outside the phosphatase domain-II. In line with the fact that the antibody was raised to an intracellular domain of the PTPgamma molecule the antibody labeled the cell membrane of fixed cells but did not stain the outside of the cell membrane in the immunofluorescence assay. The Mab gammaTL1 recognized a full-length baculovirus recombinant PTPgamma protein of 185 kDa, in addition to putative cleavage products of 120 kDa, 114/110 kDa and 80 kDa, on Western blots of lysates of PTPgamma-gene transfected Sf9 insect cells but not of tumor cell lysates. Based on immunoperoxidase and immunofluorescence assays on cryostat sections, however, PTPgamma was expressed in more than 90% of both normal, human tissue samples and in the (non-) tumor cells of carcinoma samples. However, PTPgamma was not found in 28% of the overall lung tumor samples, i.e. in 50% of the lung adenocarcinoma samples, while the expression was weak and heterogeneous in 71% of squamous lung cell carcinomas. PTPgamma was not suppressed in the normal cells between the lung carcinoma cells. The presence of PTPgamma, assayed by immunofluorescence in lung tumor cell lines (H69, H128, H82, C3) was confirmed by RT-PCR assay. Interestingly, the 90% expression score of PTPgamma protein in normal ovarian tissue samples was reduced dramatically to 44 and 38% in both the non-tumorous and tumorous cells, respectively, in ovarian tumor samples. PTPgamma was absent in the HT29 human colon carcinoma cell line both by immunofluorescence and RT-PCR assay. In summary, we have developed a PTPgamma-specific monoclonal antibody, that demonstrated that the expression of PTPgamma is severely reduced (>50%) in lung tumors and ovarian tumors.

Cell surface GPI-anchoring of CD45 isoforms.

We have designed a new cell surface expression plasmid to study the structural and membrane-topological requirements for functioning of different isoforms of CD45, a leucocyte specific member of the protein tyrosine phosphatase (PTPase) family of proteins. Use of this vector in cell transfection experiments enabled us to produce multiple CD45 isoforms (ABC, B, Null), with their extracellular segment intact, and the entire membrane spanning and intracellular C-terminal domain replaced by a GPI-membrane-anchor and VSV-tag. Our strategy facilitated the identification and analysis of chimeric proteins and selection of cell clones from low transfection efficiency experiments. We demonstrate here that simple expression of GPI-anchored CD45 isoforms on transfected Cos-1 cells does not facilitate binding to CD22+ lymphoid cells. This suggests that not only the mere presence of CD45 extracellular domains but also their assembly into higher order structures at the cell surface, is necessary in order to engage in the recognition and/or signalling processes normally used by B- and T-cells.

Differentiation margins of ovarian tumor pathology: first incidences of epithelial ovarian tumors monitored by marker antibodies.

The first incidence of ovarian tumors in The Netherlands was analyzed during the PALGA data. The first incidences of benign epithelial ovarian tumors reach a plateau, at a level of 60 to 65 cases per 100,000 women beyond the age of 40 years. The borderline malignant epithelial ovarian tumors account for 10 per 100,000 women aged 30 to 85, while the ovarian carcinomas reach a plateau level of 25 to 35 per 100,000 women after the age of 50. Despite the long lag period (+/- 10 years) between benign and malignant ovarian tumors, the question of whether or not all epithelial ovarian cancers develop via an intermediate step of cystadenomas is still unanswered. Therefore, we examined whether the expression pattern of intermediate filaments and tumor antigens in normal, benign, and malignant ovarian tissues might contribute to this question. The following changes in expression pattern were observed: generally, all tumor cells retained the keratin profile of the corresponding original cell type. However, in a limited number of tumor samples ectopic keratin types, such as nos. 4, 10, 13, and 14, became expressed additionally. Most epithelial ovarian tumors and mesothelial cells coexpressed vimentin. The panepithelial marker BW495/36 clearly distinguished between negatively stained normal ovarian surface mesothelium and the positively stained (inclusion) cystic epithelium. TAG-72 as well as OV-632 marked a subsequent tumor stage by discriminating between negative serous adenomas and positive serious carcinomas. TAG-72, however, stained both mucinous adenomas and carcinomas. The ovarian tumor markers (OC125, OV-TL 33, OV-TL16, MOv18), all showed an increasing expression level in the sequence order from normal cells to benign and malignant ovarian tumors. Both our epidemiological and our immunohistochemical data have shown that in the Dutch population there is a lag period of at least 10 years between the plateau levels of benign and malignant ovarian tumors. The early transformation of mesothelial cells to benign and/or malignant tumors is clearly marked by a switch on of the BW495/36 marker. Although no general transformation or progression marker from adenomas to carcinomas emerged from this study, TAG-72 might be considered a (progression) marker between the subgroup of benign and malignant serous ovarian tumors.

Design of immunoliposomes directed against human ovarian carcinoma.

Factors (protein/lipid ratio, pH of incubation medium, incubation time, anchor molecule density in the bilayer) affecting the covalent binding of anti-ovarian carcinoma Fab' to liposomes containing the anchor molecule MPB-PE (N-(4-(p-maleimidophenyl)butyryl)phosphatidylethanolamine) were explored. Standard experimental conditions were chosen and information on the relevant physicochemical parameters of the liposome dispersions was collected (mean particle diameter, size distribution, charge). The reproducibility of standard immunoliposomes prepared in subsequent batches in terms of Fab' binding, particle size and charge was established. In addition, preservation of immunoreactivity, no marker loss, and no aggregation/fusion was found for the standard immunoliposomes over a period of at least 3 weeks at 4 degrees C. In vitro up to 35,000 immunoliposomes were estimated to bind per human ovarian carcinoma cell. Internalization of the immunoliposomes could not be demonstrated. Electron micrographs showed binding of specific immunoliposomes to human ovarian carcinoma cells growing intraperitoneally in athymic nude mice.

The applicability of a keratin 7 monoclonal antibody in routinely Papanicolaou-stained cytologic specimens for the differential diagnosis of carcinomas.

The monoclonal antibody OV-TL 12/30, which detects keratin 7, was tested for its usefulness in cytologic diagnosis by reincubating previously Papanicolaou-stained slides. For this purpose malignant effusions of 73 patients with histologically confirmed cancers of the colon, ovary, mesothelium, breast, lung, esophagus, pancreas, urinary bladder, stomach, kidney, and prostate were used. All malignant cells from ovarian adenocarcinomas were positive, whereas malignant cells from colonic adenocarcinomas and malignant mesotheliomas were negative. Adenocarcinomas of gastric, renal, pancreatic, esophageal, and mammary origin demonstrated variable staining. Transitional cell carcinomas were positive, whereas squamous and small cell lung carcinomas were negative. Because OV-TL 12/30 does not react with normal and atypical mesothelial cells in these preparations, this reagent is a valuable tool in distinguishing benign mesothelial cells and adenocarcinoma cells. The authors' results demonstrate that this antibody is an excellent tool in the differential diagnosis of malignant cells in effusions and can be used in routinely stained cytologic specimens to determine primary tumor localization. In addition to its ability to distinguish between ovarian and colonic adenocarcinomas, its negativity in mesotheliomas may prove helpful in several diagnostic considerations.

Immunocytochemical detection of ovarian carcinoma cells in serous effusions.

Monoclonal antibodies OC 125, OV632, OV-TL 3, MOv18 and OV-TL 23, directed against distinct ovarian carcinoma-associated antigens, were examined for their value in cytopathologic diagnosis. Their sensitivity and specificity in staining ovarian carcinoma cells in serous effusions we determined using the indirect immunoperoxidase technique. Smears prepared from 140 serous effusions (73 benign, 67 malignant) were immunostained with the five antibodies. OC 125 and MOv18 reacted positively with 96% and 81% of the smears of effusions from ovarian carcinoma patients, respectively, while OV-TL 3, OV632 and OV-TL 23 stained a lower percentage of the samples (73%, 65% and 62%, respectively). In discriminating ovarian carcinoma cells from benign (mesothelial or inflammatory) cells in serous effusions, MOv18 demonstrated the highest specificity (100%) since none of the 73 samples from benign effusions were stained upon incubation with this antibody. OC 125 cannot be used for this purpose due to its reactivity with mesothelial cells in benign samples. Staining cytologic preparations of malignant effusions from cancer patients with carcinomas not originating in the ovary revealed that OV632 and OV-TL 23 may be useful adjuncts to determine the origin of the carcinoma cells found in serous effusions. The reactivity of these antibodies was highly selective for ovarian carcinoma cells, staining only 6% and 0% of the samples from the non-ovarian carcinoma samples, respectively. It is concluded that MOv18 is the most suitable antibody for distinguishing ovarian carcinoma cells from mesothelial cells in serous effusions, while OV632 and OV-TL 23 especially may help to assess whether carcinoma cells found in effusions originate in the ovary.

OV-TL 12/30 (keratin 7 antibody) is a marker of glandular differentiation in lung cancer.

The immunoreactivity of OV-TL 12/30, a monoclonal antibody to keratin 7 was investigated on paraffin-embedded human lung cancer tissues of 61 patients. A modified AEC-immunoperoxidase method with pepsin pre-digestion was used. In normal lung tissue keratin 7 was found in bronchial and bronchiolar epithelium, pneumocytes and compound glands. Squamous metaplasia of the bronchial tree was negative. All 24 squamous cell carcinomas were negative irrespective of grade of differentiation. All differentiation grades of 20 adenocarcinomas including bronchioalveolar carcinomas were positive. Since six large cell anaplastic carcinomas did not react with keratin 7 antibody these tumours are considered to be of squamous cell rather than adenocarcinomatous origin. Small cell anaplastic carcinomas were negative in 10 of 11 cases. Our study demonstrates that this keratin 7 antibody is useful in differentiating between squamous cell carcinoma and adenocarcinoma of the lung and it may be particularly useful in making the correct diagnosis in small lung biopsy specimens.

Changes in expression of differentiation markers between normal ovarian cells and derived tumors.

The marker profile of 18 samples of normal human ovarian tissues and 138 samples of their derived tumors was established using 51 monoclonal antibodies directed against intermediate filaments, ovarian carcinoma-specific antigens, general tumor-associated antigens and MHC-I/II antigens. Our data show that vimentin and keratins 7, 8, 18, and 19 were found in both epithelial and some nonepithelial ovarian tumors. Several tumor samples contained additional keratins 4, 10, 13, and 14, as well as desmin. BW 495/36 and to a lesser extent HMFG-2 were usually found in all ovarian tumors that contained simple epithelial keratins, except the absence of HMFG-2 in gonadal tumors as well as in dysgerminomas. In contrast to the keratin antibodies, these two panepithelial antibodies were negative in normal mesothelial cells and granulosa cells of the ovarian follicles. In general, the marker TAG-72 appeared useful for its discrimination between positively stained mucinous adenomas, the ovarian carcinomas as well as germ cell tumors, and the negatively stained gonadal tumors, serous adenomas, and cystomas. OV632 appeared useful in the distinction between negatively stained serous adenomas and positively stained serous carcinomas. In contrast, the monoclonal antibodies OC 125, OV-TL 3, OV-TL 16, and MOv 18 can be considered as pan-ovarian carcinoma markers, however without the discriminative capability as seen for OV632. These ovarian carcinoma-associated antigens were hardly found expressed in gonadal and germ cell tumors, except in the group of endodermal sinus tumors. HLA-I was found to be expressed in almost all nucleated cells, although loss of HLA-I expression was seen in areas of tumor cells. HLA-DR was negative in normal ovarian tissue, but heterogeneous expression was noticed in most of the epithelial tumors.

The expression of cytoskeletal proteins during the differentiation of rat granulosa cells.

This report examines the relationship between aromatase activity and progesterone production and the expression of actin and vinculin in rat granulosa cells, exposed to insulin, follicle stimulating hormone (FSH) and human chorionic gonadotrophin (HCG). Granulosa cells of pre-antral follicles from juvenile rats treated with diethylstilbestrol (DES) were cultured on collagen A-coated plastic coverslips in serum-free medium. At a moderate or low level of steroidogenesis (FSH alone), the expression of vinculin was diminished while vinculin plaques disappeared completely. At a high level of steroidogenesis (both FSH and insulin), actin in stress fibre form was also decreased considerably. Under conditions of progesterone production (pretreatment with FSH and subsequent incubation with HCG), a concomitant increase of actin in soluble form was found. It is concluded from these studies that the higher the steroidogenesis level of the granulosa cells, the lower the organization of the microfilaments vinculin and actin, which are regulated independently of each other.

Specific and nonspecific immunoassays to detect HAMA after administration of indium-111-labeled OV-TL 3 F(ab')2 monoclonal antibody to patients with ovarian cancer.

The development of human anti-mouse antibodies (HAMA) may cause problems in radioimmunotargeting studies, but may also improve survival of patients. To identify the presence of HAMA in blood samples from patients intravenously injected with 1 mg of 111In-labeled OV-TL3-F(ab')2, we developed three specific OV-TL 3-based HAMA assays and tested these along with two commercially available nonspecific HAMA assays (Sorin and Immunomedics). The specific assays were positive for HAMA with 10 postinjection serum samples from 7 patients. Eight of the 10 samples were also HAMA positive with one or both nonspecific HAMA assays. Conflicting results were observed with half the number of samples. The two nonspecific assays also reacted positively with another 11 serum samples from 5 patients including their preinjection samples. Despite some contradictory results, the nonspecific HAMA assays identify both pre-existent and Mab-induced HAMA, whereas the specific OV-TL3-based HAMA assays identify specific immune-responses occurring after the OV-TL 3 injection.

In vivo targeting of OV-TL 3 immunoliposomes to ascitic ovarian carcinoma cells (OVCAR-3) in athymic nude mice.

Specific binding of immunoliposomes to target tumor cells was investigated in a xenograft model (athymic nude mice) of i.p. growing human ovarian carcinoma (OVCAR-3). For the first time, quantitative evidence is presented that attachment of a tumor-specific antibody (OV-TL 3) dramatically enhances the association of liposomes with i.p. growing OVCAR-3 cells. The OV-TL 3-mediated binding of liposomes to the OVCAR-3 cells was rapid; 30 min after i.p. injection approximately 70% of the injected dose of OV-TL 3 immunoliposomes was associated with the OVCAR-3 cells while for unconjugated liposomes a value of only approximately 3% was obtained. At 2 h after injection, a maximal binding level of 84% was achieved in case of the OV-TL 3 immunoliposomes whereas the binding level of unconjugated liposomes was still about 3%. Twenty-four h after injection about 83% of the injected dose OV-TL 3 immunoliposomes still was associated with the OVCAR-3 cells, compared to about 10% of the injected dose of unconjugated liposomes. Accordingly, unconjugated liposomes disappeared from the peritoneal cavity much faster than the OV-TL 3 immunoliposomes. By comparison with immunoliposomes bearing irrelevant antibody, the specificity of the binding of the OV-TL 3 immunoliposomes to the OVCAR-3 cells was demonstrated. In addition, it was observed that the sustained high OV-TL 3 immunoliposome levels found in the peritoneal cavity are the result of both reduced particle clearance from the peritoneal cavity and the tenacious binding of the immunoliposomes to the tumor cells. Finally, data are presented showing that the degree of binding of OV-TL 3 immunoliposomes to OVCAR-3 cells in vitro and in vivo correlates positively with the antibody (Fab') density on the liposomes.

Biodistribution of iodine-125 and indium-111 labeled OV-TL 3 intact antibodies and F(ab')2 fragments in tumor-bearing athymic mice.

The monoclonal antibody OV-TL 3, directed against an ovarian carcinoma-associated antigenic determinant, was tested as a vehicle for radioimmunolocalization of ovarian carcinomas in athymic mice bearing NIH:OVCAR-3 xenografts. The biodistribution of intact. OV-TL 3 was compared with the distribution of OC 125. Tumor uptake with OV-TL 3 was significantly higher than with OC 125, and almost 7 times higher than with a non-specific control antibody (OV-TL 19). Administration of a mixture of intact OV-TL 3 and OC 125 did not improve tumor uptake in comparison with OV-TL 3 alone. Subsequently, intact OV-TL 3 and its F(ab')2 fragments were labeled with either 111In or 125I. The highest tumor uptake was obtained with 111In-labeled intact OV-TL 3 (14.7% ID/g, 48 hr p.i.). For both antibody forms uptake of 111In in liver, spleen and kidneys was very high. Furthermore, 111In cleared more slowly from most tissues than 125I. As a result, tumor/tissue ratios with 111In-labeled OV-TL 3 were lower than with 125I-labeled OV-TL 3. The highest tumor/tissue ratios (6.9 to 53) were obtained with 125I-labeled OV-TL 3 F(ab')2 fragments, 48 hr post injection. 111In-labeled OV-TL 3 F(ab')2 has already been shown to be a clinically useful label for the detection of ovarian cancer. The results of our comparative animal study suggest that these clinical results may even be improved by using 123I-labeled OV-TL 3 F(ab')2.

High CA-125 concentrations in peritoneal fluid of normal cyclic women with various infertility-related factors as demonstrated with two-step immunoradiometric assay.

OBJECTIVE: To determine CA-125 concentrations and total amounts in peritoneal fluid (PF) of women with various infertility-related factors throughout the menstrual cycle. DESIGN: Peritoneal fluid was obtained at laparoscopy. CA-125 was determined using the assessed two-step immunoradiometric assay (IRMA) which, in contrast to the one-step IRMA, gives valid results. SETTING: University Hospital Nijmegen, Nijmegen, The Netherlands. PATIENTS: One hundred six infertile women with a regular and ovulatory cycle were included. INTERVENTIONS: None. MAIN OUTCOME MEASURE(S): The mean PF CA-125 concentration and total amount were significantly lower during the luteal phase as compared with other phases of the menstrual cycle. No correlation was found with the presence or absence of endometriosis, adhesions, a male and/or cervical mucus infertility factor, and with patent or closed fallopian tubes. RESULTS: Peritoneal fluid CA-125 concentrations varied from 630 to 12,000 arbitrary units/mL (mean +/- SD = 3,437 +/- 2,286). Total PF CA-125 amounts (concentration x PF volume) varied from 1,760 to 13,300 arbitrary units (mean +/- SD = 30,219 +/- 26,841). CONCLUSIONS: CA-125 secretion into the abdominal cavity varies during the menstrual cycle. Retrograde menstruation is not the main source of CA-125 in PF.

Immunohistochemical demonstration of keratin 7 in routinely fixed paraffin-embedded human tissues.

The immunoreactivity of OV-TL 12/30, a monoclonal anti-keratin 7 antibody (Mab), was investigated on frozen as well as paraffin-embedded human tissues. Its reactivity patterns were compared with another well-characterized monoclonal antibody to keratin 7 (RCK 105), and with broadly cross-reacting monoclonal (OV-TL 12/5) as well as polyclonal (pKer) keratin antisera. In frozen sections of normal and malignant human tissues both keratin 7 Mabs gave similar staining patterns. The immunoreactivity for OV-TL 12/30 and the polyclonal antibody (pKer) in tissue sections fixed in 4 per cent formalin or Bouin solution, was completely restored when pretreated with 0.1 per cent pronase, 0.1 per cent trypsin in phosphate-buffered saline (PBS) or with 0.5 per cent pepsin in 0.01 N HCl. Except for loss of immunoreactivity on human normal stomach surface epithelium and glandular mucous cells, Mab OV-TL 12/30 reacted strongly positive with essentially all those formalin- or Bouin-fixed paraffin-embedded tissues that had been shown to stain in non-fixed, frozen sections. In addition to the good correlation in human tissues, a complete correlation between the reactivity on frozen and paraffin-embedded human carcinomas (n = 86) was found as well. While both RCK 105 (anti-keratin 7) and OV-TL 12/5 (anti-keratin 5, 7, 14, 19) did not stain on paraffin-embedded sections, the polyclonal control antiserum (pKer) lost immunoreactivity in some cell types (e.g. mucous cells in compound glands, hepatocytes, pancreatic acinar cells, and proximal and distal convoluted tubules of the kidney). Our study shows that the keratin 7 Mab OV-TL 12/30 is an excellent marker for tumour histopathology since it is reactive in paraffin-embedded formalin-fixed human tissues.

Immunocytochemical markerprofile of endometriotic epithelial, endometrial epithelial, and mesothelial cells: a comparative study.

A comparative immunocytochemical study was performed on endometriotic epithelial versus endometrial epithelial and normal mesothelial cells in order to obtain further evidence for either the endometrial implantation or mesothelial metaplasia theory. The three cell types could not be distinguished by keratin subtyping, using monoclonal antibodies (MAbs) to keratins 5, 7, 8, 14, 18, and 19. The epithelial markers HMFG-2 and BW 495/36, and a newly developed MAb NEND-3 (against endometrial cells) discriminated between generally negatively reacting mesothelial cells and positively reacting endometrial and endometriotic epithelial cells. The MAb NEND-3 appeared to be specific for endometrial (and endometriotic) epithelial cells since no reactivity with other epithelial cell types was found. Typing with MAbs against ovarian carcinoma related antigens (OV-TL 3, OV-TL 10 and OC 125) did not permit sufficient distinction. The marker profile of cultured endometrial, endometriotic and mesothelial cells confirmed the immunocytochemical findings on frozen sections. Although the data are consistent with the endometrial implantation theory, mesothelial metaplasia can not be excluded with regard to the histogenesis of endometriosis since metaplastic mesothelium may express different antigen markers.

Retrograde seeding of endometrial epithelial cells by uterine-tubal flushing.

OBJECTIVE: To estimate the amount of endometrial epithelial cells in peritoneal fluid (PF) after uterine-tubal flushing (40 mL) throughout the menstrual cycle. DESIGN: We cultured the cell pellet of flush medium present in the peritoneal cavity. SETTING: University Hospital Nijmegen, The Netherlands. PATIENTS: Ninety-two women with various infertility-related factors. INCLUSION CRITERIA: (1) ovulatory cycle, (2) patent tubes, and (3) no adhesions. INTERVENTIONS: None MAIN OUTCOME MEASURE(S): The number of developing epithelial cell colonies were counted after 7 days. We started to record the amount of flush medium recovered during the study. RESULTS: The amount of flush medium recovered was positively correlated with the presence of endometriosis (P = 0.017). Endometrial epithelial cells were identified in 85 flush media (92%). The number of epithelial cell colonies varied from 0 to 100 or more and was higher when flushing was performed during the early follicular phase (P less than 0.01). High estradiol-17 beta and progesterone levels in culture medium did not change the number of developing cell colonies. Methylene blue significantly reduced the number of cell colonies (P = 0.002). CONCLUSIONS: Uterine-tubal flushing results in varying numbers of endometrial epithelial cells in PF. Methylene blue adversely affects the growth potential of these cells.