Fig 1.
Tumor cells with a SOX2+ (or NANOG+) HIF-1α+ RNApII-S2P-/low phenotype are found in human glioblastoma tissues.
a: Hematoxylin-eosin (H-E) staining and single-color immunostaining for SOX2, NANOG, HIF-1α, and RNApII-S2P (serine 2 phosphorylation of the C-terminal domain of RNA polymerase II) in tumor tissue from a representative case of glioblastoma. Suppression of RNApII-S2P immunoreactivity was noted around the necrotic area (N). b: Triple immunofluorescent staining for SOX2, HIF-1α, and RNApII-S2P. SOX2+ HIF-1α+ RNApII-S2P-/low cells (arrows) were found around the necrotic area (N). c: Triple immunofluorescent staining for NANOG, HIF-1α, and RNApII-S2P. NANOG+ HIF-1α+ RNApII-S2P-/low cells (arrows) were found around the necrotic area (N). Insets at lower left of the panels b and c show higher magnification of the boxed areas in the merged images. N, necrotic area; V, blood vessels; DAPI, 4′,6-diamidino-2-phenylindole (nuclear stain); DIC, differential interference contrast image. Scale bars, 25 μm. d: Summary of the staining methods used in b and c. Sections were incubated with goat anti-SOX2 (b) or anti-NANOG (c) antibody and then with Alexa Fluor 546-conjugated donkey anti-goat IgG secondary antibody. Next, rabbit anti-RNApII-S2P antibody and then Alexa Fluor 635-conjugated goat anti-rabbit IgG secondary antibody were applied. The sections were reacted with mouse anti-HIF-1α antibody and then with biotinylated donkey anti-mouse IgG secondary antibody and Alexa Fluor 488-conjugated streptavidin.
Fig 2.
Detection of SOX2+ (or NANOG+) HIF-1α+ RNApII-S2P-/low tumor cells by multi-color chromogenic immunostaining in glioblastoma tissues reveals that these cells are associated with large ischemic necroses.
a-e: Serial sections of glioblastoma tissue around a large ischemic necrosis. H-E staining (a, c), triple staining for SOX2/HIF-1α/RNApII-S2P (b, d) or NANOG/HIF-1α/RNApII-S2P (e) are shown (SOX2 or NANOG, red; HIF-1α, blue; RNApII-S2P, brown). Boxes in a and b indicate the magnified areas in c and d, respectively. A curved line in c represents the border between a large ischemic necrosis and viable tumor tissue. Arrows in d and e mark SOX2+ HIF-1α+ RNApII-S2P-/low cells and NANOG+ HIF-1α+ RNApII-S2P-/low cells, respectively. Insets in d and e show higher magnifications of these cells (arrows). NL, large ischemic necrosis; V, blood vessels. Scale bars, 50 μm. f: Correlation between the relative area of necrosis and the frequency of SOX2+ HIF-1α+ RNApII-S2P-/low cells in 21 glioblastoma cases. Line indicates the regression line (y = 0.188x + 1.46, Pearson’s correlation coefficient r = 0.64). g: Summary of the staining methods used in b and d–f. Sections were incubated with goat anti-SOX2 or anti-NANOG antibody and then with alkaline phosphatase (AP)-conjugated donkey anti-goat IgG secondary antibody, and color was developed with Vulcan Fast Red. After denaturing, mouse anti-HIF-1α antibody and rabbit anti-RNApII-S2P antibody were applied, and then the sections were reacted with MACH 2 Double Stain 1 (a secondary antibody cocktail of AP-conjugated anti-mouse IgG and horseradish peroxidase-conjugated anti-rabbit IgG antibodies). Color was developed with Perma Blue/AP for HIF-1α and diaminobenzidine for RNApII-S2P.
Fig 3.
SOX2+ (or NANOG+) HIF-1α+ RNApII-S2P-/low tumor cells are not found in zones around small pseudopalisading necroses or in areas showing no necrotic change.
These cells are found in glioblastomas (WHO grade IV), but not in diffuse astrocytomas (grade II) and anaplastic astrocytomas (grade III). a, b: Serial sections of glioblastoma tissue containing small pseudopalisading necroses. c, d: Serial sections of glioblastoma tissue showing no necrotic changes. H-E staining (a, c) and triple immunostaining for SOX2/HIF-1α/RNApII-S2P (b, d) are shown. Although SOX2+ and/or HIF-1α+ tumor cells were found, they were RNApII-S2P+. Therefore, SOX2+ HIF-1α+ RNApII-S2P-/low cells were not observed. NS, small pseudopalisading necrosis; V, blood vessel. Scale bars, 50 μm. e: Triple immunostaining for SOX2/HIF-1α/RNApII-S2P in astrocytomas of WHO grade II and III. No SOX2+ HIF-1α+ RNApII-S2P-/low cells were observed. Scale bars, 25 μm. f, g: Frequency of SOX2+ HIF-1α+ RNApII-S2P-/low cells (f) and NANOG+ HIF-1α+ RNApII-S2P-/low cells (g) in cases of astrocytic tumors of WHO grade II–IV. *, P < 0.05; **, P < 0.01 (grade IV vs. combined group of grades II and III).
Fig 4.
Triple immunofluorescent staining for SOX2, HIF-1α, and RNApII-S2P in zones around small pseudopalisading necroses or in areas showing no necrotic changes.
a: Glioblastoma tissue containing small pseudopalisading necrosis. Although SOX2+ and/or HIF-1α+ tumor cells were found, they were RNApII-S2P+. Therefore, SOX2+ HIF-1α+ RNApII-S2P-/low cells were not observed. b: Glioblastoma tissue showing no necrotic changes. SOX2+ HIF-1α+ RNApII-S2P-/low cells were not found. Insets at lower left of the panels a and b show higher magnification of the boxed areas in the merged images. Ns, small pseudopalisading necrosis; V, blood vessels; DIC, differential interference contrast image. Scale bars, 25 μm.
Fig 5.
The localization of SOX2+ HIF-1α+ RNApII-S2P-/low tumor cells is related to their distances from necrotic areas and blood vessels.
a, b: Serial sections of glioblastoma tissue around a large ischemic necrosis. Double immunostaining for HIF-1α/RNApII-S2P (a) and triple immunostaining for SOX2/HIF-1α/RNApII-S2P (b) are shown (SOX2, red; HIF-1α, blue; RNApII-S2P, brown). Arrows in b indicate SOX2+ HIF-1α+ RNApII-S2P-/low tumor cells. NL, large ischemic necrosis; V, blood vessels. Scale bars, 50 μm. c: Relationship between distances from nearest blood vessels and from necrotic areas in SOX2+ HIF-1α+ RNApII-S2P-/low cells. The data from 135 SOX2+ HIF-1α+ RNApII-S2P-/low cells found in 12 cases of glioblastoma are plotted. The line indicates the regression line (y = 0.205x + 17.4, r = 0.34). The gray region represents y ≤ x.
Fig 6.
Significance of emergence of NANOG+ HIF-1α+ RNApII-S2P-/low cells in spheroid cultures of T98G glioblastoma cells.
a: Scheme of the experiments. Spheroids were generated under normoxic conditions (20% O2), and then divided into 2 groups (normoxic and hypoxic). After normoxic or hypoxic (5% O2) culture for 9 or 24 h, histological analysis and sphere formation assay under normoxic conditions were performed. Data for the normoxic and hypoxic groups sampled after the same culture time were compared. b–j: Histological analysis of the spheroids cultured under normoxic or hypoxic conditions. Since the spheroids of normoxic group cultured for 9 and 24 h showed similar results, only the spheroids cultured for 24 h are shown. H-E staining (b–d), chromogenic double immunostaining for HIF-1α/RNApII-S2P (e–g), and triple immunostaining for NANOG/HIF-1α/RNApII-S2P (h–j) are presented (NANOG, red; HIF-1α, blue; RNApII-S2P, brown). NANOG+ HIF-1α+ RNApII-S2P-/low cells (purple cells) were detected in spheroids cultured in hypoxia for 24 h (j, arrows). The inset in j shows a higher magnification of these cells. These data are representative of at least 3 independent experiments. Scale bars, 50 μm. k: Sphere-forming efficiency of the cells derived from the spheroids of normoxic or hypoxic groups. Values are relative ones in which the results for the normoxic spheroids sampled after the same culture time were set to 1. The results are expressed as the mean ± SD of at least 7 independent experiments. N.S., not significant; *, P < 0.05.
Fig 7.
A schematic model illustrating the main findings of this study.
The SOX2+ (or NANOG+) HIF-1α+ RNApII-S2P-/low cells identified in this study represent hypoxic quiescent stem-like tumor cells, and the emergence of these cells is associated with an increased sphere-forming activity. Thus, the most common locations of these cells, namely, the neighborhood of large ischemic necroses in glioblastoma tissues (“peri-necrotic niche”) and the zone of intermediate depth from the surface of the spheroids cultured under 5% O2, indicate a microenvironment supporting tumorigenicity. Since a gradient of O2 concentration is present within the glioblastoma tissues and spheroids, this microenvironment is formed under an appropriate level of hypoxia.