Targeting PTPRZ inhibits stem cell-like properties and tumorigenicity in glioblastoma cells

Abstract

The R5 subfamily of receptor-type protein tyrosine phosphatases (RPTPs) comprises PTPRZ and PTPRG. A recent study on primary human glioblastomas suggested a close association between PTPRZ1 (human PTPRZ) expression and cancer stemness. However, the functional roles of PTPRZ activity in glioma stem cells have remained unclear. In the present study, we found that sphere-forming cells from the rat C6 and human U251 glioblastoma cell lines showed high expression levels of PTPRZ-B, the short receptor isoform of PTPRZ. Stable PTPRZ knockdown altered the expression levels of stem cell transcription factors such as SOX2, OLIG2, and POU3F2 and decreased the sphere-forming abilities of these cells. Suppressive effects on the cancer stem-like properties of the cells were also observed following the knockdown of PTPRG. Here, we identified NAZ2329, a cell-permeable small molecule that allosterically inhibits both PTPRZ and PTPRG. NAZ2329 reduced the expression of SOX2 in C6 and U251 cells and abrogated the sphere-forming abilities of these cells. Tumor growth in the C6 xenograft mouse model was significantly slower with the co-treatment of NAZ2329 with temozolomide, an alkylating agent, than with the individual treatments. These results indicate that pharmacological inhibition of R5 RPTPs is a promising strategy for the treatment of malignant gliomas.

Figure 1

Decreased cell sphere formation of stem cells in Ptprz-knockdown glioblastoma cells. (A) Sphere formation assay. C6 and U251 cells (parent), Ptprz-knockdown C6 cells (RZ-KD#2), and PTPRZ1-knockdown U251 cells (RZ1-KD#5U) were cultured in serum-free cancer stem cell medium for 7 days (CSC) or DMEM supplemented with 10% FBS for 3 days (Normal). Scale bars, 100 µm. Images are representative of five independent cultures. The plot shows the number of cell spheres as the mean value. **P < 0.01 (Student’s t-test). (B) PTPRZ expression. Cells cultured as in A were analyzed by Western blotting with anti-PTPRZ-S (for the detection of rat PTPRZ) and anti-RPTPβ (for human PTPRZ1). The blot is representative of five independent cultures. The plot shows the arbitrary densitometric units of the staining intensity of PTPRZ-B relative to the parental cells cultured in normal serum-containing medium. *P < 0.05, **P < 0.01 (Student’s t-test). Full-length blots and gels are presented in Supplementary Fig. S6.

Figure 2

Altered expression of core transcription factors in Ptprz-knockdown glioblastoma cells. Western blots using antibodies against SOX2, OLIG2, POU3F2, and SALL2. Parental C6 and RZ-KD#2, and parental U251 and RZ1-KD#5U cells were cultured in CSC medium as in Fig. 1A. SALL2 proteins were not detected in C6 or U251 cells. Sample loading was verified by immunostaining with GAPDH. Images are representative of five independent cultures. The plots show the relative densitometric units of the staining intensity in the PTPRZ-knockdown cells and the staining intensity in the parental cells. *P < 0.05; **P < 0.01 (Student’s t-test). Full-length blots and gels are presented in Supplementary Fig. S6.

Figure 3

Decreased tumor growth in the Ptprz-knockdown glioblastoma cell xenografts. Subcutaneous xenograft tumor model. Nude mice were implanted subcutaneously with parental and RZ-KD#2 cells (5 × 10⁶ cells). The results were obtained from five independent cell preparations for the C6 group and two independent cell preparations for the RZ-KD#2 cell group. Images are representative of six animals per group 30 days after the cell injection, in which the tumor rims were surrounded by red dotted lines. Tumor size was measured until it reached the humane endpoint for sacrifice (>3,000 mm³ or 50 days after the cell injection): the C6 group reached the predetermined humane endpoint at 30 days after cell injections. The plot shows tumor growth for each animal. One mouse (a gray square symbol) was sacrificed because of tumor necrosis and bleeding at 34 days. The days required to reach a tumor volume of 3,000 mm³ were significantly different between the parental C6 and RZ-KD#2 groups (P < 0.01; Mann-Whitney U-test). The average tumor sizes at 30 days were significantly different between the two groups (P < 0.01, Student’s t-test).

Figure 4

NAZ2329, an allosteric inhibitor of R5 RPTP subfamily members. (A) Structure of NAZ2329. (B) Effect of preincubation of NAZ2329 on its inhibitory activity against PTPRZ. Recombinant human PTPRZ1 enzyme (the whole intracellular region) was premixed with NAZ2329 for the indicated times and followed by addition of a fluorogenic DiFMUP (non-specific PTP substrate, 6,8-difluoro-4-methylumbiliferyl phosphate) to measure PTP inhibitory activity. (C) Time dependence. The PTPRZ1 enzyme was added to the inhibitor–substrate mixture, and the inhibitory activity was determined at each time point relative to the vehicle control. (D) Reversibility of inhibition. PTPRZ1 enzymes were preincubated with NAZ2329 at the indicated concentrations for 30 min. The mixture was then 20-fold diluted, and recovered PTP activity was expressed as the relative activity compared with the vehicle-treated enzyme. (E) Lineweaver-Burk plot analysis. PTPRZ1 enzymes were preincubated with NAZ2329 for 60 min, and the PTP activity was measured using DiFMUP. (F) Concentration-inhibition curves of NAZ2329 for representative PTP members, including human PTPRZ and PTPRG (another R5 RPTP subfamily member), human PTPRS (R2A subfamily), human PTPRM (R2B subfamily), mouse PTPRB (R3 subfamily), human PTPRA (R4 subfamily), human PTPN1 (non-transmembrane PTP, NT1 subfamily), and human PTPN6 (NT2 subfamily), were generated. Assays were performed using DiFMUP. IC₅₀ values obtained are shown in the inset. Data represent the averages of two or three separate experiments.

Figure 5

Structural basis for PTPRZ inhibition of NAZ2329. (A) An enlarged view of the X-ray structure of human PTPRZ1-D1 complexed with NAZ2329 (PDB ID: 5H08) (left), X-ray structure of the apo (open) form of PTPRZ1-D1 (PDB ID: 5AWX, ref. , middle) and a computer-modeled structure of the closed form of PTPRZ1-D1 complexed with a competitive inhibitor, SCB4380 (ref. , right), are shown. PTPRZ1-D1 is shown as a gray surface representation, in which Cys1933 at the active site and Val-1911 at the bottom of the allosteric pocket are indicated in red and yellow, respectively. Inhibitors are colored according to the atomic species in a stick figure representation as follows: oxygen (red), nitrogen (blue), sulfur (yellow), and carbon (green). The bottom figures show the conformation of the WPD loop (orange) as cartoon representations, in which Asp1901 and Cys1933 are indicated by orange and red spheres, respectively. Side chains of Arg1939 at the active site, Met1902 in the WPD loop, and Val-1911 at the bottom of the allosteric binding area are also shown in the stick representation. (B) LigPlot representation of the NAZ2329 interaction with PTPRZ1-D1. The predicted hydrogen bonds (broken black lines) and residues (red spoked arcs) involved in the hydrophobic interaction are shown. (C) Predicted steric hindrance toward NAZ2329 binding after substitution of Val-1911 with a bulky Phe residue. The wild-type (upper) and replaced (lower) structures of PTPRZ1-D1 are shown in stick and transparent-sphere models. The structure was built by direct replacement of Val-1911 by Phe using the mutagenesis tool in the PyMOL software. (D) Inhibitor sensitivity to point substitution mutants of PTPs. Inhibition by 10 µM NAZ2329 is presented as a percentage of the DMSO control for each enzyme (mean ± S.E. of three separate experiments). **P < 0.01, significantly different from the wild-type enzyme (Student’s t-test).

Figure 6

Effects of NAZ2329 treatment on oligodendrocyte differentiation. Mouse oligodendrocyte-lineage OL1 cells were cultured in differentiation media containing the indicated concentrations of NAZ2329. After 10 days, cells were fixed with formalin and stained with anti-NG2 proteoglycan (oligodendrocyte precursor cells, red) and anti-MBP (oligodendrocyte, green) antibodies, in conjunction with the DAPI-labeling of nuclei (blue). Scale bars, 100 µm. The plot shows the ratio of MBP-positive cells to NG2-positive cells, in which each dot corresponds to an independent cell culture (n = 4 each). *P < 0.05; **P < 0.01, significantly different from vehicle-treated cells (one-way ANOVA with Bonferroni post hoc tests).

Figure 7

Cellular effects of NAZ2329 on the malignant phenotypes of C6 cells. (A) Phosphorylation of paxillin at Tyr-118. C6 cells were incubated with NAZ2329 for the indicated periods. Immunoprecipitated paxillin was analyzed by Western blotting using anti-pY118-paxillin and anti-paxillin. Blots are representative of five independent cultures. The plot shows the intensity of pY118 staining relative to the paxillin level, normalized to the vehicle control in each experiment. **P < 0.01, significantly different from the vehicle by one-way ANOVA with Bonferroni post hoc tests). (B) Cell proliferation assay. C6 cells were incubated for 48 h with NAZ2329 in normal medium containing 2% FBS. The plot shows the percentage increase in the cell number. *P < 0.05, **P < 0.01 (one-way ANOVA with Bonferroni post hoc tests). (C) Boyden chamber assay. Cells were allowed to migrate for 3 h. DAPI-stained nuclei are shown before (total) and after (migrated) the removal of cells remaining in the top chamber. Scale bars, 100 µm. The plot shows the migrated cell number normalized to the vehicle. *P < 0.05, **P < 0.01 significantly different from the vehicle (one-way ANOVA with Bonferroni post hoc tests). (D,E) Sphere formation (D) and SOX2 expression (E). C6 cells were cultured in CSC medium for 7 days with indicated concentrations of NAZ2329. Images are representative of five independent cultures. Scale bars, 100 µm. The plots show the sphere number per well (D) and staining intensity of SOX2, normalized to the vehicle (E). *P < 0.05, **P < 0.01 significantly different from the vehicle (one-way ANOVA with Bonferroni post hoc tests). (F) Self-renewal of C6 spheres. C6 spheres were initially developed in CSC medium for 7 days, followed by an incubation with NAZ2329 in CSC medium for 5 days. Images are representative of five independent cultures. Scale bars, 100 µm. The plot shows the sphere number per well. *P < 0.05, **P < 0.01 (one-way ANOVA with Bonferroni post hoc tests). Full-length blots and gels are presented in Supplementary Fig. S7.

Figure 8

Inhibitory effects of NAZ2329 on PTPRZ activity in C6 cells. (A,B) Protein expression of PTPRZ-1B and paxillin (A, left), the overall tyrosine (Tyr)-phosphorylation pattern of cellular proteins (A, right), and the Tyr-phosphorylation levels of paxillin at Tyr-118 (B). Sample loading was verified by immunostaining with GAPDH. RZ-KD#2 cells transfected with an expression plasmid for wild-type (WT) PTPRZ1-B, the V1911F (VF) or C1933S (CS) mutants of human PTPRZ1-B, or empty plasmid (Moc) were treated with 25 µM NAZ2329 or vehicle for 60 min, and the cell extracts were subjected to analyses by Western blotting. Blots are representatives of four independent experiments. Paxillin proteins were immunoprecipitated, and their Tyr-phosphorylation levels at Tyr-118 were analyzed as in Fig. 7A: pY118 levels were normalized to the vehicle-treated mock cells in each experiment. (C to E) Cell proliferation (C, 48 h), cell migration (D, 3 h), and sphere formation (E) in RZ-KD#2 cells transfected with the indicated construct were analyzed as in Fig. 7B,C and D, respectively. Scale bars, 100 µm. Statistical analyses of the data were performed as follows: *P < 0.05; **P < 0.01, significantly different from the mock-transfected cells in each treatment group (vehicle or NAZ2329) (one-way ANOVA with Bonferroni post hoc tests). ^# P < 0.05; ^## P < 0.01, significantly different between the vehicle and NAZ2329 treatment (Student’s t-test in each transfectant). The blots and images are representative of four independent experiments, and their full-length blots (A and B) and representative microscopic images (D) are presented in Supplementary Fig. S8.

Figure 9

Effects of Ptprg knockdown on C6 stem cell-like properties. (A) Quantitative RT-PCR analyses. C6 cells were electroporated with short interfering RNA (siRNA) for rat Ptprz, Ptprg, or Ptprz plus Ptprg. After a 48-h culture, RNA was extracted from cells and subjected to quantitative RT-PCR. The plots show the mRNA expression of Ptprz, Ptprg, and Sox2 normalized to Gapdh expression. (B) Sphere formation assay. Cells transfected with the indicated siRNAs were cultured for 7 days in CSC medium. Scale bars, 100 µm. Images are representative of six independent cultures. The plot shows sphere number per well. *P < 0.05, **P < 0.01, significantly different from the scramble control (Student’s t-test).

Figure 10

Antitumor effects of NAZ2329 on the C6 mouse xenograft model. (A) Nude mice were subcutaneously implanted with C6 cells (5 × 10⁶ cells), and tumor size was monitored until the criteria were met (>150 mm³). At this time, mice were randomly divided into 4 treatment groups. DMSO as the vehicle control (n = 7), NAZ2329 (45 µmol (22.5 mg)/kg body weight, n = 9), temozolomide (TMZ, 50 mg/kg, n = 9), and the combination of NAZ2329 (45 µmol/kg) with TMZ (50 mg/kg) (n = 9) were administered intraperitoneally twice per week until the humane endpoint (>3,000 mm³ tumor size or 40 days after the treatment). Tumor growth in each animal and the number of days required to reach a tumor volume of 3,000 mm³ are shown in the graph. Five mice were sacrificed due to tumor necrosis (shown by gray symbols in the graph), and two mice unexpectedly died suddenly during the observation period (black symbols in the graph). (B) A Kaplan-Meier analysis of the four treatment groups shown in A. Significant differences were observed in the days until the tumor volume reached 3,000 mm³ between the vehicle vs NAZ2329/TMZ (P < 0.05), NAZ2329 vs NAZ2329/TMZ (P < 0.05) and TMZ vs NAZ2329/TMZ (P < 0.05) (Kruskal-Wallis test followed by Steel-Dwass test).

Figure 11

CSC differentiation effects of the pharmacological inhibition by R5 RPTPs. Because cancer stem-like cells (CSCs) contribute to therapeutic resistance and tumor recurrence in malignant tumors including glioblastoma, a more effective therapy should target chemoresistant CSCs and proliferating and invading cancer cells. PTPRZ is strongly expressed in malignant gliomas^, , and the expression levels of PTPRZ transcripts are closely associated with cancer stemness in primary human glioblastomas. The short receptor form, PTPRZ-B, is the major PTPRZ isoform in rat C6 and human U251 glioblastoma cells (Fig. 1D), the expression of which is significantly stronger in sphere-forming cells than normal cultured cells. The cell-permeable R5 RPTP inhibitor, NAZ2329, suppresses the stem cell-like properties of CSCs and the strong proliferation and migration characteristics of adherent glioblastoma cells (non-CSCs). The synergistic antitumor effects of NAZ2329 with temozolomide (TMZ) on the C6 glioblastoma xenograft, which was previously reported to be relatively resistant to TMZ (ref. 34), indicates that the combined use of R5 RPTP inhibitors and TMZ will be an effective treatment for malignant gliomas.