Dual parameter analysis of myeloma cells by flow cytometry. DNA content of cells containing monotypic cytoplasmic immunoglobulin

Dual-parameter flow cytometric analysis of monotypic cytoplasmic immunoglobulin (CIg) and DNA content on 15 myeloma marrows allowed S-phase determination of the CIg(+) tumor separately from the CIg(-) hematopoietic cell pool. The median percentage of the CIg(+) cells in S-phase was 2% compared with 5% for the CIg(-) cells. The median survival of patients with more than 2%, and those with 2% or less CIg(+) S-phase cells was 2 months and more than 13 months, respectively, from the time of study. Ploidy analysis identified four patterns of plasma cell DNA content: entirely hyperdiploid, entirely diploid, combined diploid and tetraploid stemlines, and tumors containing diploid and aneuploid CIg(+) cells. A monotypic CIg(+) double stemline myeloma was distinguished from two aneuploid tumors containing admixed normal, diploid polyclonal plasma cells. This technique provides an improved and expedient means for determining the proliferating fractions of myeloma cells and enhances recognition of double stemline tumors and clonal evolution in myeloma.     

Karyotyping and DNA flow cytometry of mature residual teratoma after intensive chemotherapy of disseminated nonseminomatous germ cell tumor of the testis: a report of two cases

Karyotyping and DNA flow cytometry was performed on mature residual teratoma cells following intensive chemotherapy of disseminated nonseminomatous germ cell tumor of the testis to study its biology. We report herein a successful method for short-term tissue culture and karyotyping of retroperitoneal residual mature teratoma in two cases. In vitro morphology confirmed that the cultured cells were nonembryonal carcinoma cells. Both mature residual teratomas were highly aneuploid and possessed the i(12p) marker characteristic of testicular germ cell tumors. A clone in the retroperitoneal residual lesion of one of the patients showed a DNA-index different from the primary tumor and might represent a clone unmasked by chemotherapy. In view of these data, which are in agreement with recent reports on secondary non-germ cell malignancies arising in mature residual teratoma, aggressive surgery of mature residual lesions seems justified.     

Multiparametric deoxyribonucleic acid and cell cycle analysis of breast carcinomas by flow cytometry. Clinicopathologic correlations.

Using intact ethanol-fixed cytokeratin monoclonal (CAM 5.2) and propidium iodide dual-stained cells, we have performed two-color multiparametric flow cytometric (FCM) DNA analysis and S-phase fraction (SPF) determination on 165 mechanically dissociated breast carcinomas. Sixty-seven patients were axillary node positive, 33 patients node negative; 59 had biopsy only and in 8, FCM was performed on tissue from metastatic lesions. Overall, 62% of the tumors contained aneuploid cell populations. Abnormal cellular DNA content (aneuploidy) was significantly correlated with high nuclear grade (p less than 0.001), lack of estrogen receptors (p less than 0.001), presence of vascular invasion (p less than 0.04), high histologic grade (p less than 0.04), and tumor size (p less than 0.03) but not with patient age (p greater than 0.07) or axillary node status (p greater than 0.50). SPF values derived from ungated histograms had a positively skewed frequency distribution (range 2 to 30%, N = 152) with an overall median of 11% (diploid, 8.9%; aneuploid, 15.7%). Higher SPF values were significantly correlated with aneuploidy (p less than 0.001), presence of necrosis (p less than 0.001), lack of estrogen receptor (p less than 0.0001), high nuclear grade (p less than 0.001), vascular invasion (p less than 0.003), tumor size (p less than 0.006), and high histologic grade (p less than .004) but not the presence of lymph node metastases (p greater than 0.56). Mean SPF values were significantly higher when calculated from cytokeratin gated DNA histograms (14.1% versus 11.5%, p less than 0.001), probably due to exclusion of contaminating stromal/inflammatory cells; and significantly lower when calculated from debris subtracted histograms (7.8% versus 11.4%). Cytokeratin gated and debris subtracted SPF values both had a greater degree of correlation than ungated values with clinicopathologic factors of known prognostic significance.(ABSTRACT TRUNCATED AT 250 WORDS)     

Expression of intermediate filament proteins in normal and diseased thyroid glands.

A total of 67 samples from normal and pathological thyroid glands were stained (as formalin fixed paraffin sections) with a panel of monoclonal antibodies directed against intermediate filament proteins. The study confirmed previous reports of cytokeratin and vimentin coexpression in primary thyroid carcinomas, but coexpression was also detected in normal thyroid and in a range of benign conditions including follicular adenomas, Hashimoto's thyroiditis, and diffuse hyperplasia (thyrotoxicosis). Prekeratin expression was found (using antibodies recognising higher molecular weight cytokeratins) predominantly in areas of squamous change, independent of the underlying thyroid pathology. This study does not therefore support previous findings that prekeratin expression provides a reliable means of distinguishing follicular pattern papillary carcinoma from follicular carcinoma with its poorer prognosis or that it helps distinguish benign from malignant papillary lesions. No evidence of desmin or neurofilament expression was seen, and in particular, neurofilaments could not be detected in any of the cases of medullary carcinoma studied.     

Distribution of intermediate filament proteins in normal and diseased human glomeruli.

The distribution of intermediate filament proteins (vimentin, desmin, and cytokeratin) was studied by means of immunofluorescence in the normal human and rat glomerulus and in pathologic human glomeruli. Antifibronectin antibodies were used as mesangial markers. In normal human glomeruli, vimentin antibodies stained endothelial cells, podocytes, and mesangial cells; desmin antibodies, surprisingly, stained podocytes. In normal rat glomeruli, the pattern of vimentin staining was the same as in humans, but desmin antibodies stained both mesangial cells and podocytes. In human and rat glomeruli cytokeratin staining was confined to segments of Bowman's capsule. In human pathologic glomeruli, vimentin and desmin antibodies stained the structures that were positive in normal glomeruli, giving a characteristic pattern for each pathologic condition examined. These results are compatible with the mesenchymal origin of podocytes and mesangial cells and suggest that both cells have smooth muscle-like phenotypic features. Mesangial cells may have slightly different differentiation paths in humans and rats, leading to a distinct expression of intermediate filament proteins.     

The use of flow cytometry as a tool for monitoring filament formation of fungi.

Flow cytometry (FC) has the ability to discriminate a variety of cell parameters including cell size and complexity, and fluorescence intensity. As yeast cells or fungal spores germinate they undergo a morphological transformation from round oval shaped cells to elongate filamentous forms. To date, monitoring these events has been performed using microscopic examination. Microscopic examination is a labor intensive process that examines a very small percentage of the total cell population. We have developed a method using FC that is rapid, simple to perform, and reproducible. The major advantages of FC include analysis of a larger number of cells, increased objectivity due to nonselective measurements of all cells in the population studied, and the computer related data analysis capability of the flow cytometer.     

Expression of intermediate filament proteins in fetal and adult human kidney: modulations of intermediate filament patterns during development and in damaged tissue.

The expression of intermediate filament proteins, particularly individual cytokeratins (CKs), vimentin, and glial filament protein, was immunohistochemically investigated using frozen sections and Carnoy-fixed, paraffin-embedded tissue from normal fetal and adult human kidneys as well as from pathologically altered kidneys. In fetal kidneys, the co-expression of CKs and vimentin was detected in the visceral and parietal epithelium of the glomerulus, the proximal tubules, the thin loops of Henle, and the collecting ducts. In contrast, in the tubules of normal adult kidneys, the presence of vimentin and CKs was nearly always mutually exclusive. While CKs 8 and 18 were present in all tubular epithelia, CKs 19 and 7 each exhibited a distinctive distribution pattern, there being a striking alteration between positive and negative segments and, not infrequently, intratubular heterogeneities. In certain segments, particular cell types (e.g., "plica cells," intercalated cells) could thus be recognized. In tubular epithelia altered by various injurious conditions, novel or enhanced expression of vimentin, CK 19 and CK 7, and, less frequently, CK 17 and glial filament protein was noted in certain segments. The increase in intermediate filament protein expression in altered (particularly proximal) tubules appeared to parallel the reduction in the degree of differentiation. Vimentin was never detected in distal tubules. The present results reveal a considerable similarity between the intermediate filament patterns in non-neoplastic proximal tubules of fetal and damaged kidney tissue and those in clear-cell and chromophilic renal cell carcinomas. They also serve to illustrate that the analysis of both fetal development and reactive cell changes may significantly contribute to our understanding of differentiation phenomena in malignant tumors.     

Myofibroblasts from diverse pathologic settings are heterogeneous in their content of actin isoforms and intermediate filament proteins

We examined by immunofluorescence the distribution of vimentin, desmin, alpha-smooth muscle actin and alpha-sarcomeric actin in normal human soft tissues and in pathologic tissues containing myofibroblasts, including normally healing granulation tissue, hypertrophic scar, and fibromatosis. The pattern of actin isoforms was also documented biochemically by two-dimensional gel electrophoresis. Fibroblastic and/or myofibroblastic cells in each setting always expressed vimentin and never alpha-sarcomeric actin. Moreover, these cells showed an heterogeneous cytoskeletal composition which defined four phenotypes: (a) cells expressing only vimentin; (b) cells expressing vimentin, alpha-smooth muscle actin and desmin; (c) cells expressing vimentin and alpha-smooth muscle actin; and (d) cells expressing vimentin and desmin. Given this, two groups of lesions are distinguished: the first contains only vimentin cells and consists of normally healing granulation tissue, eschars and normally healed scars; the second contains vimentin cells admixed with variable proportions of vimentin, alpha-smooth muscle actin and desmin, vimentin and alpha-smooth muscle actin, and vimentin and desmin cells and consists of hypertrophic scars and fibromatoses. Immunogold electron microscopy showed that alpha-smooth muscle actin was present in a proportion of cells with ultrastructural features of myofibroblasts. Our findings suggest that contrary to myofibroblasts of normally healing granulation tissue and normally healed scars, myofibroblasts of pathologic conditions characterized by chronic retraction express always immunochemical features indicative of smooth muscle differentiation.     

Complexity of expression of intermediate filament proteins, including glial filament protein, in endometrial and ovarian adenocarcinomas.

The expression patterns of intermediate filament proteins of primary and metastatic endometrial (n = 18) and ovarian (n = 24) adenocarcinomas were analyzed by immunocytochemistry using a panel of specific antibodies and by gel electrophoresis of cytoskeletal preparations, followed by immunoblotting. All cells of all endometrial adenocarcinomas studied contained the "simple epithelial"-type cytokeratins (CKs) 8, 18, and (mostly) 19, with variable numbers of cells also positive for CK 7 and vimentin. In addition, most of these tumors contained individual cells or groups of cells that were positive for the stratification-related CKs 4, 5, 6, 13, 14, and 17. The latter CKs were often associated with squamous cell foci, but were also found in some single (nonsquamous) tumor cells, indicative of early stages of squamous cell differentiation. Ovarian carcinomas of various histologic types and grades contained predominantly CKs 7, 8, 18, and 19. Serous, endometrioid, and anaplastic tumors, but not mucinous and clear cell tumors, also contained minor amounts of stratification-related CKs in variable combinations, mostly including CK 4. In all tumor types except mucinous tumors, vimentin was consistently detected in variable proportions of tumor cells which, however, were rather low in anaplastic carcinomas. Surprisingly, glial filament protein was detected in a minor proportion (< or = 20%) of tumor cells in seven of 14 serous and endometrioid ovarian carcinomas and in three of 18 endometrial carcinomas. These different intermediate filament expression patterns of müllerian duct-type carcinomas, only partly related to the morphologic appearance of the specific type of tumor, might reflect the multipotentiality of differentiation of müllerian duct-derived epithelia. Cytoskeletal features of potential diagnostic value, especially in metastatic carcinomas, are discussed.     

Application of BrdU/Hoechst-ethidium bromide two parameter flow cytometry in studying synchronous and non-synchronous mouse cells.

BrdU/Hoechst-EB bivariate flow cytometry has a wide application in the study of factors controlling cell cycle for asynchronous cells such as embryonic stem cells (ES), and for synchronous cells such as stimulated resting B cells (Bo). The technique allows one to calculate the average cell cycle duration time. ES cells are found to cycle every 8-10 h, and most B cells are 11-12 h, but there is a small subset of B cells with a cycle time of only 6-7 h. Using this technique, we also study the roles of different T lymphocytes on B cell activation when B cells are stimulated with anti-IgM antibodies (commonly used, anti-mu). Exposure to anti-mu recruits small B cells into the cell cycle, but arrests them in the G1 phase of the second cycle. Interleukin (IL) 4 is a costimulator of anti-mu. In addition, IL-4 is an S-phase progression factor. Contrary to that seen when B cells are stimulated by other mitogens, very few cells are in the G2 compartments after anti-mu plus IL-4 stimulation. This phenomenon is reminiscent of embryonic cells. Our findings provide strong evidence to propose that there are two restriction points for B cell activation: at the transition from G0 to G1 and at the transition from G1 to S phase.     

Heterogeneity and co-expression of intermediate filament proteins in adenoid cystic carcinoma of salivary glands.

Among a series of 76 adenoid cystic carcinomas (ACC), 51 cases with cibriform or trabecular patterns were selected for an immunohistochemical study. These tumors were composed of three types of cells: basaloid, myoepithelial and ductal luminal acinous or tubular cells with a variable amount of each cell type from one case to another. Monoclonal antibodies (MoAb) against keratins (PKK1, KL1, K8.12, K8.13, K4.62 anti-cytokeratins antibodies) and against vimentin were used. Tubular cells were characteristically marked by anti-cytokeratins MoAb, but with a great heterogeneity of keratin distribution. In myoepithelial cells, keratin was absent or slightly positive and vimentin was present, with co-expression of the two types of filaments in some cells. Like myoepithelial cells, basaloid cells were positive to anti-vimentin antibody and negative or slightly positive to anti-keratin antibodies, with sometimes a co-expression of vimentin and keratin filaments in the same cell. Histogenesis of adenoid cystic carcinomas was discussed. Progenitor intercalated duct reserve cells may change into ductal luminal and myoepithelial tumor cells. Otherwise, basaloid cells, arising also from intercalated duct reserve cells, are able to acquire some secretory organites.     

Immunohistochemical localization of intermediate filament and S-100 proteins in several non-endocrine cells of the human pituitary gland

The presence and distribution of glial fibrillary acidic protein, vimentin, neurofilament protein, cytokeratins No. 8 (52 Kd), No. 18 (45 Kd) and No. 19 (40 Kd) and S-100 protein in pituicytes, folliculo-stellate cells, the epithelium of the Rathke's cysts and squamous cell nests of the pars tuberalis were investigated immunohistochemically by the peroxidase-antiperoxidase (PAP) method in eleven normal human pituitary glands. An identical immunostaining pattern was expressed by both folliculo-stellate cells and pituicytes. In both cell types the immunostaining for glial fibrillary acidic protein (GFAP), S-100 protein and vimentin was strongly positive. These results indicate the probable glial origin of the folliculo-stellate cell, and enlarge the group of glial cell types expressing vimentin. The co-expression of cytokeratins No. 8 and 19, both characteristic for simple epithelia, and S-100 protein was evident in the epithelial cells lining the Rathke's cysts and the squamous cell nests of the pars tuberalis. Furthermore, some epithelial cells of the Rathke's cysts co-expressed cytokeratins, S-100 protein and GFAP, a fact seldom reported and only in relation to rare neoplasms. The cytokeratin No. 18, characteristic for glandular epithelia, was not clearly demonstrated. Finally, the neurofilament protein was detected only in axons of the neurohypophysis; no immunopositive cells could be found throughout the adenohypophysis. Similarities in the antigenic patterns of these cell populations and the possible relation with their origin and nature are discussed.     

Evaluation of multiparameter flow cytometry for the detection of breast cancer tumor cells in blood samples

We comparatively evaluated different cytokeratin (CK) reagents analyzed by flow cytometry (FCM) for the identification of the best combination of DNA/CK staining for detecting minimal numbers of breast cancer cells in peripheral blood (PB). In 59 primary breast cancer tumors, we comparatively analyzed the reactivity for up to 6 different anti-CK reagents using multiparameter FCM: anti-CK7, anti-CK20, anti-pan-CK, anti-CK8/CK18, anti-CK8, and anti-CK18. Afterward, dilutional experiments of Michigan Cancer Foundation (MCF)7 breast cancer cells in PB were performed, and the sensitivity of a DNA/CK18 staining was evaluated. Our results showed that anti-CK18 reagents were those providing the brightest and more sensitive staining for primary breast cancer tumor cells by FCM. Dilutional experiments of MCF cells in PB showed that the DNA/anti-CK18 double staining was highly specific for the identification of epithelial cells; its sensitivity ranged between 10(-6) and 10(-7) (detection of 1 tumor cell among 10(6) to 10(7) nucleated blood cells). Combined assessment of DNA cell contents and reactivity for CK18 by FCM is a sensitive method for the specific identification of breast cancer cells in PB.     

Cytokeratin-specific monoclonal antibodies are reactive with tumours of smooth muscle derivation. An immunocytochemical and biochemical study using antibodies to intermediate filament cytoskeletal proteins

Formalin fixed and paraffin wax embedded tissue from 13 leiomyosarcomas, three leiomyoblastomas, five leiomyomas and fresh tissue from one leiomyosarcoma and one leiomyoma was studied. Tumours were stained by the avidin-biotin-peroxidase technique with antibodies to desmin, vimentin, cytokeratins and S-100 protein. Neoplastic cells in all the cases were positive for desmin and vimentin but were negative for S-100; antibody CAM 5.2 gave strong reactivity with tumour cells of 12 leiomyosarcomas and all leiomyomas, but failed to stain the leiomyoblastomas; LP34 weakly stained two leiomyosarcomas and one leiomyoblastoma. Immunoblot studies using whole cell extracts of one leiomyosarcoma and one leiomyoma respectively, showed 55-58 kd and 35-38 kd bands reactive with monoclonal antibody CAM 5.2. However, no cytokeratin components were demonstrated when cytoskeletal extracts of the same tissue samples were immunoblotted with CAM 5.2. The implications of these findings for the diagnostic histopathologist are discussed.     

Cytogenetic analysis of the mature and immature teratoma components of a metastatic testicular nonseminomatous germ cell tumor.

A case is described in which the mature and immature teratoma components of metastases of the same testicular nonseminomatous germ cell tumor were karyotyped. The highly similar karyotypes of both components suggest that the phenotypic difference is predominantly epigenetically determined.     

Antibodies to intermediate filament proteins as molecular markers in clinical tumor pathology. Differentiation of carcinomas by their reaction with different cytokeratin antibodies

Antibodies to human and bovine epidermal prekeratin and antibodies to mouse liver cytokeratin component D (Mr 49 000) have been applied in indirect immunofluorescence microscopy on sections of human tumors of mammary gland and liver. In non-neoplastic mammary gland all epithelial cells were stained with these antibodies. In pre-invasive and invasive ductal and lobular carcinomas a cell population was observed which was not significantly stained with antibodies to epidermal prekeratin but did strongly react with antibodies to liver cytokeratin D. In the liver, the antibodies to epidermal prekeratin as well as those directed against liver cytokeratin D strongly decorated bile duct epithelia. In contrast, significant staining of the hepatocytes was only achieved with antibodies to liver cytokeratin D. This different staining reaction was maintained in liver tumors of hepatocellular and cholangiocellular origin. Antibodies to vimentin stained mesenchymal cells and tumors of mesenchymal derivation but reacted not significantly with any of the epithelial and carcinoma cells examined. The difference is of practical importance for the discrimination between anaplastic carcinomas and sarcomas of unknown origin. Cytokeratin could also be detected by antibody staining using the peroxidase-antiperoxidase (PAP) technique in formaldehyde-fixed and paraffin-embedded material of skin, gastrointestinal, respiratory, urinary and genital tract as well as various glands, liver and kidney. Examples of positive reactions were shown in a squamous cell carcinoma, a basalioma and a pleomorphic adenoma of the parotis. It is concluded that the immunohistochemical analysis of intermediate filament proteins has diagnostic potential in clinical pathology and may help to elucidate histogenesis and differentiation of tumors and possibly also prognosis of tumor growth. It is further suggested to use antibodies recognizing different subsets of proteins of the cytokeratin family in order to distinguish between different types of carcinomas.     

Intermediate filament proteins and actin isoforms as markers for soft tissue tumor differentiation and origin. II. Rhabdomyosarcomas.

A series of 15 rhabdomyosarcomas was examined by light microscopy, transmission electron microscopy, two-dimensional gel electrophoresis (2D-GE) and indirect immunofluorescence, the latter using monoclonal or affinity-purified polyclonal antibodies to desmin, vimentin, alpha-smooth muscle and alpha-sarcomeric (alpha-sr) actins. By light microscopy, the authors diagnosed 1 botrioid, 1 alveolar, and 7 embryonal rhabdomyosarcomas, 4 pleomorphic spindle cell sarcomas, and 2 spindle cell sarcomas, one nondistinct, the other with a hemangiopericytomatous pattern. By transmission electron microscopy, 13 neoplasms disclosed rhabdomyoblastic differentiation; the remaining 2, myogenic differentiation. By immunofluorescence microscopy, all neoplasms expressed vimentin and alpha-sr actin, 12 expressed, in addition, desmin, and 1 expressed alpha-smooth muscle actin. Among the 11 neoplasms studied by means of 2D-GE, 7 demonstrated an alpha-actin spot, while 4 showed only beta and gamma spots. One tumor disclosed, in addition to alpha, beta, and gamma spots, a spot with a molecular weight corresponding to actin, but more acidic than alpha-actins. This study demonstrates that alpha-sr actin antibody represents a valuable marker for the diagnosis of rhabdomyosarcoma, because it was present in all neoplasms, including the one negative for desmin. This antibody further allowed the recognition of pleomorphic variants and morphologically atypical forms of rhabdomyosarcomas. The presence of alpha-smooth muscle actin in 1 case of rhabdomyosarcoma suggests that this actin isoform may be expressed during skeletal muscle differentiation.     

流式细胞术检测调节性T细胞方法的建立

目的建立流式细胞术（FCM）检测外周血CD4＋CD25＋调节性T细胞（Tregs）的方法，并观察紫杉醇联合卡铂治疗对晚期肺癌和乳腺癌患者外周血CD4＋CD25＋Tregs数量的影响。 方法19例晚期肺癌和10例乳腺癌患者均给予紫杉醇联合卡铂方案化疗，于化疗前1d和化疗后第7天采集患者外周血，分别加入鼠抗人CD4-FITC（异硫氰酸荧光素）/CD8-PE（藻红蛋白）/CD3-PerCP（多甲藻叶绿素蛋白）、CD25-FITC/CD127-PE/CD4-PerCP、CD3-FITC/CD（16＋56）-PE/CD45-PerCP单抗，并以分别加入同型鼠抗人IgG1-FITC、IgG1-PE、IgG1-PerCP抗体作为阴性对照。采用流式细胞术（FCM）检测化疗前后外周血CD3＋、CD4＋、CD8＋T细胞和CD4＋CD25＋Tregs、NK细胞所占比例并进行数据分析。实验重复3次。 结果与化疗前比较，晚期肺癌和乳腺癌患者化疗后外周血CD4＋CD25＋Tregs的比例均明显降低（6.82%±3.11%vs.5.48%±2.13%，P=0.045；6.38%±1.84%vs.3.88%±1.69%，P=0.007）；晚期肺癌患者化疗后外周血CD4＋T细胞的比例升高（48.84%±16.44%vs.56.35%±14.50%，P=0.006），CD8＋T细胞的比例降低（51.18%±16.44%vs.43.65%±14.50%，P=0.006），CD4＋/CD8＋T细胞比值升高（1.12±0.60vs.1.57±0.88，P=0.008），而CD3＋T细胞和NK细胞的比例均无明显变化；乳腺癌患者化疗后外周血CD3＋、CD4＋、CD8＋T细胞、NK细胞的比例和CD4＋/CD8＋T细胞比值均无明显变化。 结论成功建立了FCM检测CD4＋CD25＋Tregs的方法，联合应用CD4、CD25、CD127检测CD4＋CD25＋Tregs简便可行、重复性好，检测结果可靠、准确，比较适用于临床检验。紫杉醇联合卡铂能够降低晚期肺癌和乳腺癌患者外周血CD4＋CD25＋Tregs的数量。     

Enzyme-induced aggregation and disaggregation of tumor cells via the cell surface glycocalyx in association with deoxyribonucleic acid

Serine proteases cause aggregation of the rat ascites tumor cell lines AH-130, AH-109A and YS in vitro, and the tumor cell aggregates are dissolved by treatment with DNase I. We previously demonstrated that these events played a critical role in the augmentation or reduction of experimental blood-borne metastasis of these cell lines. In the present study, the ultrastructural features of this protease-dependent aggregation were analysed. Transmission and scanning electron microscopy revealed that after the protease treatment each tumor cell was surrounded by a thin membranous (sleeve-like) structure. This sleeve-like structure was stained with ruthenium red to an intensity similar to the cell surface of the control. Adjacent cells became attached to each other with microvilli via this fine structure. Immuno-electron microscopy revealed DNA antigen as dense patches on the sleeve-like structure or as faint and diffuse deposits on the outer surface of the cells by indirect immunoperoxidase staining using an anti-DNA monoclonal antibody. Both the sleeve-like structure and immunopositive deposits disappeared after treatment with DNase I. Neither cell viability nor the normal ultrastructure of their organelles was influenced by the enzyme treatment. These results indicate that serine protease-induced tumor cell aggregation is due to cellular contact via the sleeve-like structure, which probably originates from the cell surface glycocalyx in association with DNA molecules of unknown origin.