A, Table of the clonally-derived tumor cell cultures. Single tumor cells from sorted populations were isolated and clonally expanded in vitro. No differences in clonal expansion and pleiomorphic differentiation could be detected when compared to parental cultures. B, Phase contrast micrographs of tumor cells obtained after clonal expansion. Glandular organization similar to the parental cultures was observed. a, Tu-7 (passage 3), clone D4 (X10) - 3 month culture. b, Tu-7 (passage 3), clone G2 (X10) - 3 month culture. c, Tu-7 (passage 4), clone C3 (X4) - 2.5 month culture. d, Tu-12 (passage 1) clone F7 (X10) - 1 month culture. e, Tu-12 (passage 1) clone F7 (X4) - 1 month culture. f, Tu-12 (passage 1) clone G12 (X4) - 1 month culture. C, Lineage commitment of clonally derived tumor cells. Tu-12 clones F7, E9 and G12 were stained for HLA-ABC (FITC, green) and Muc-1 or Muc-2 (Alexa-594, red). The nuclear counterstain was Hoechst 33342 (blue). Heterogeneous expression of Muc-2 and Muc-1 indicate the presence of both goblet and enterocyte lineages. D, Illustrative G-banded karyotypes of Tu-12 and Tu-21, parents and clones. Three of twenty metaphase cells analyzed are presented here for each tumor cell culture. The results revealed hypodiploid to near-tetraploid (Tu-12) and near-hexaploid (Tu-21) karyotypes with several clonal and nonclonal numerical and structural chromosomal aberrations in all three tumor cell cultures. Chromosome number ranged from 40 to 94 (Tu-12) and from 38 to 135 (Tu-21). *, abnormal chromosomes; red, chromosomal aberrations common to all three cultures; blue, chromosomal aberrations common only in one or two cultures. The variability in the karyotypes of these tumor cells suggests the presence of chromosomal instability.

A, Flow cytometric contour plots (5% probability) demonstrating the expression of cancer stem cell markers on one representative culture of tumor tissues from metastatatic colon cancer (Tu-18). The dot plots depict two-color staining of live tumor cells from Tu-18 patient on a logarithmic scale. Quadrant markers were positioned to include in the lower left quadrant greater than 98% of control unstained live human tumor cells (not shown). The numbers are the percentage of live cells present in each quadrant after staining. B, Flow cytometric dot plots indicating the gating strategy and result of the CD133 positive and negative cell isolation. The dot plots depict two-color staining of live tumor cells from Tu-27 on a logarithmic scale. First plot, propidium iodide/forward scatter indicates the gate for live cells (49.4%), second plot indicates the gate for human tumor cells (84.1%, HLA⁺) and the third plot is the staining for CD133 (and CD49f). Quadrant markers were positioned to include in the lower left quadrant greater than 98% of control unstained live human tumor cells (not shown). The numbers are the percentage of live cells present in each quadrant after staining. The re-analyses of the isolated CD133⁺ and CD133⁻ subpopulations are presented. C, Table of tumorigenic activity of CD133⁺ and CD133⁻ human colon cancer cells. For Tu-12 and Tu-21, clones derived from a single cell were expanded in vitro and transplanted in mice to generate xenograft tumors described in the result section. Tumor cells isolated from the xenograft or from the culture (Tu-27) were sorted for CD133⁺ and CD133⁻ cell populations as described. Cell populations with identical numbers of CD133⁺ and CD133⁻ cells were transplanted to the left (CD133⁻) and right (CD133⁺) flanks of immunodeficient mice. Tumors were collected six to twelve weeks after transplantation.

A, CEA secretion in vitro in long-term cultures and in vivo in Rag2/γc^−/− mouse sera after tumor cell transplantation. Parental and clonal cultures (derived from one single tumor cell) expressed high levels of CEA in vitro. The transplanted animals with engrafted tumor cells were identified by their CEA secretion. Tu-7, Tu-12 and Tu-14 were expanded in vitro before transplantation. Tu-18, Tu-19 and Tu-22 were tumor cells transplanted in mice without previous in vitro expansion. P, passage. B, Xenograft tumors in mice injected subcutaneously with metastatic colon cancer derived from three patients. Frozen sections of Tu-12 (expanded in vitro, 8 weeks after transplantation), Tu-18 (not expanded, 6 weeks after transplantation), and Tu-22 (not expanded, 8 weeks after transplantation) were stained with anti-HLA-ABC (Alexa-488, green) and anti-Muc-2 (Alexa-594, red). Nuclear counterstaining was done with Hoechst 33342 (blue). The observed heterogeneous expression of Muc-2 demonstrates the presence of goblet cells in the moderately to well-differentiated adenocarcinomas. HLA negative cells present around the xenograft represent the murine stromal cells. C, Multilineage commitment in clonally derived tumor cells. Serial sections of Tu-12 clone F7 and non-serial sections of Tu-21 E12 were stained for HLA-ABC (Alexa-488, green) and Muc-1, Muc-2, chromogranin A (CgA) or villin (Alexa-594, red). Nuclear counterstaining was done with Hoechst 33342 (blue). Tumor generated from one cancer cell generated both columnar cells (Muc-1⁺ and villin⁺ cells) as well as goblet cells (Muc-2⁺ cells). Rare CgA+ enteroendocrin cells were detected in Tu-21 but not in Tu-12. D, Goblet cell maturation in xenograft tumors of Tu-12 (clone F7). Left panel, transmission electron micrograph of xenograft section showing the presence in tumor tissues of mucous-containing goblet cells (G) surrounded by columnar cells with apical microvilli and tight junctions. Right upper panel, Muc-2 staining (red) and Hoechst 33342 (blue) showing mature goblet cells. Not every aberrant crypt contains Muc-2 positive goblet cells. Right lower panel, HLA-ABC (green) and Muc-2 (red) staining with Hoechst 33342 (blue) showing both goblet and columnar cells present in xenograft tumors.

A, tumor cells derived from different patients formed similar organotypic structures, with glandular organization indicating maturation potential. Photomicrographs of metastatic colon tumors Tu-12 and Tu-10 presenting pleiomorphic maturation. For Tu-11, photomicrographs correspond to sequential representation of tumor colony growth and maturation. Tumor cells exhibit an undifferentiated phenotype and differentiation occurs gradually with the formation of organized epithelial structures resembling abnormal crypts. B, Immunohistochemical analyses of Tu-11, Tu-12 and Tu-25 in culture. Immunostaining with anti-CK19 (Alexa-594, red) differentiated human tumor cells from stroma cells. Both Muc-1 (Alexa-488, green) and Muc-2 (Alexa-594, red) positive human cells are found in rare populations of tumor cells. The nuclear counterstain was Hoechst 33342. C, Flow cytometric dot plot profiles of metastatic colon tumor cells isolated from different patients and cultured under identical conditions. Density plots using pseudo-color for Tu-18 (passage 1), Tu-21 (passage 0) and Tu-22 (passage 0). The dot plots depict two-color staining of live tumor cells from three patients on a logarithmic scale. Quadrant markers were positioned to include in the lower left quadrant greater than 98% of control unstained live human tumor cells (not shown). The numbers are the percentage of live cells present in each quadrant after staining. The summary table of Panel C represents flow cytometric analyses of tumor cells. Tumor samples expanded in culture revealed a common flow cytometric profile suggesting the expansion of a similar population of cancer cells. Flow cytometric analyses of multiple passages were performed for each tumor and summarized in this table, Tu-7 (passage 2, 4), Tu-10 (passage 0, 1), Tu-11 (passage 0, 1), Tu-12 (passage 0, 1, 3, 4, 6, 7, 8), Tu-14 (passage 0, 1, 2), Tu-18 (passage 0, 1, 2), Tu-21 (passage 0, 1), Tu-22 (passage 0, 1, 3), Tu-25 (passage 0, 1, 2) Tu-27 (passage 2, 4) and Tu-28 (passage 2). P, positive when all of the tumor cells were positive for the surface marker (seven markers in all of the tumor samples), H, heterogeneous populations of tumor cells with some cells positive and other cells negative among the tumor cell populations (three markers in most of the tumor samples). N, all tumor cells negative for the surface marker (nine markers in most of the tumor samples). Five markers, including CD133 were inconsistent among the tumor samples studied. ND, not determined. Passage 0 corresponds to the first in vitro expansion. D, Summary of colony forming frequency after limiting dilution analyses (LDA) on parental (non-clonally derived) and clonally derived tumor cells. Cultured tumor cells were sorted into 96-well plates at limiting dilutions using positive cell surface markers present on human cells. Cultures were stopped after 2 to 4 weeks and colonies visualized. Regression analyses were performed to determine the frequency of tumor colony forming cells. When n≥3 experiments, standard deviation was calculated using the “nonbiased” method.