USP22 maintains gastric cancer stem cell stemness and promotes gastric cancer progression by stabilizing BMI1 protein

Abstract

Increased ubiquitin-specific protease 22 (USP22) has been associated with poor prognosis in several cancers including gastric cancer. However, the role of USP22 in gastric tumorigenesis is still unclear. Gastric cancer stem cells have been identified and shown to correlate with gastric cancer initiation and metastasis. In this study, we found that silencing of USP22 inhibited proliferation of gastric cancer cells and suppressed the cancer stem cell spheroid formation in serum-free culture. Furthermore, cancer stem cell markers, such as CD133, SOX2, OCT4 and NANOG were down-regulated. Additionally, knockdown of USP22 inhibited gastric cancer xenografts growth. Our analysis of TCGA database indicated that BMI1 overexpression may predict gastric cancer patient survival, and TAT-BMI1 proteins reversed the USP22 knockdown-mediated decreased in cancer stem cell properties, and elevated the expression of stemness-associated genes. Furthermore, we found that overexpression of USP22 stabilized the BMI1 protein in gastric cancer cells. Taken together, our study demonstrates that USP22 is indispensable for gastric cancer stem cell self-renewal through stabilization of BMI1. These results may provide novel approaches to the theranostics of gastric cancer in the near future.

Keywords: BMI1; USP22; deubiquitinase; gastric cancer; gastric cancer stem cell.

Figure 1. Knockdown of USP22 inhibits GC cell proliferation

MGC-803 cells were infected with lentiviral vectors expressing shCtrl (Control) or shUSP22-1# and 2# (shRNAs for USP22 knockdown). (A) RT-qPCR validation of USP22 knockdown efficiency. (B) Western blot analysis of USP22 expression. GAPDH was used as a loading control. (C) Cell proliferation was measured using wst-1 assays. (D) Effect of USP22 depletion on cell colony formation. After 10 d of infection, the cells were fixed and stained with crystal violet solution. The histogram shows the colony number. (E) MGC-803 and SGC-7901 cells were infected with shCtrl or shUSP22 lentiviral vector. After 72 h of infection, the scratch wound-healing assay was performed to study the effect of USP22 on cell migration. The results are shown in (F). (G) PI/Hoechst staining was conducted to evaluate the effect of USP22 on cell death. (H) The histograms representing the cell death percentage in (G). The data were from three independent experiments. Statistical comparisons between groups were conducted by unpaired Student's t-test. Bar graph shows the mean± SEM. Statistical significance: **P<0.01, ***P<0.001, compared with control. P<0.05 was considered to be significant.

Figure 2. Inhibitory effect of USP22-silencing on gastric CSC formation

(A) Cultured gastric CSCs derived from GC cell lines MGC-803 and SGC-7901 cells in serum-free culture. Scale bar=100 μm. (B) RT-qPCR analysis of the gastric CSC markers in MGC-803 cells and the MGC-803 derived stem cells (SCs). (C-D) Western blot analysis of the expression of gastric CSC markers in SCs and control. α-tubulin was chosen as endogenous control. (E) The effect of USP22 depletion on gastric CSC formation in MGC-803 and SGC-7901 cells in serum-free culture. (F) Histograms show the stem cell spheroid formation and the sizes of the spheres. (G) The stem cell spheroids in (E) (F) were passaged 2 times, and the percentage of spheroid formation and the sizes of the spheres were calculated. (H) RT-qPCR analysis of the expression of gastric CSC markers in control (shCtrl) and USP22 knockdown (shUSP22) cells. Data are presented as mean±SEM. Statistical comparisons between groups were conducted by unpaired Student's t-test. Statistical significance: *P<0.05, **P<0.01, ***P<0.001, compared with the control. P<0.05 was considered to be significant.

Figure 3. USP22 silencing suppresses tumor growth in GC xenografts in vivo

(A) Male 4-week-old BALB/c mice were subcutaneously inoculated into two hind flanks with stably expressed GFP-tagged shCtrl or shUSP22 cells. (B) Tumor volumes were calculated after 8 d every other day by measuring the length and width of tumor until 30 d and plotted. (C) Tumor growth progression was monitored by in vivo imaging of the xenografts at 30 d after inoculation. (D) Representative photos of tumors 30 d after subcutaneous xenografting (n=4). Xenografts were weighed as shown in (E). (F) H&E staining of the frozen sections of xenografts. Scale bar=100 μm. (G) Immunostaining of the frozen sections with KI67 antibody. Arrowheads indicate the KI67 positive cells. Scale bar=100 μm. (H) The relative KI67-positive cells were calculated, and statistical results are shown in the histogram. (I) RT-qPCR was performed to detect the mRNA expression of USP22, BMI1 and gastric CSC markers. Data are presented as mean± SEM. Statistical comparisons between groups were conducted by unpaired Student's t-test. Statistical significance: *P<0.05, **P<0.01, ***P<0.001, compared with the control. P<0.05 was considered to be significant.

Figure 4. GC patient survival plots of ‘death-from-cancer’ genes

(A) A schematic representing the prediction model for multiple cancers from death-from-cancer signatures. (B) GC patient survival plots of death-from-cancer genes were analysed in 127 samples from TCGA database. The GC patients were divided into two groups according to gene expression. Blue curves indicate the upper mean group, and red curves indicate the lower mean group.

Figure 5. USP22 silencing inhibits gastric CS formation by regulating BMI1 protein levels

(A) Immunofluorescence staining to analyse the co-localization of USP22 and BMI1 in MGC-803 cells. Scale bar=10 μm. (B) Effect of USP22 and BMI1 silencing on colony formation in MGC-803 cells and SGC-7901 cells. Colonies were calculated and statistical results are shown in histograms. (C) Effect of knockdown of USP22 or BMI1 on GC stem cell spheroids formation using MGC-803 and SGC-7901 cells in serum-free culture. Histograms show the percentage of spheres. (D) Western blot was performed to analyse the effect of USP22 and BMI1 depletion on proliferation-related and gastric SC markers using respective antibodies. GAPDH was chosen as a loading control. (E) Schematic of purified TAT-BMI1 protein containing a His tag, a TAT transduction domain and a HA tag (left). Validation of the transportation of BMI1 into MGC-803 cells (right). Vehicle (PBS) or TAT-BMI1 (0.05 μM) was added to cultured MGC-803 (F) and SGC-7901 (G) cells stably expressing the shCtrl, shUSP22 or shBMI1 in vitro sphere-forming assays. The number of gastric CSC spheres was counted and plotted relative to initial cell number (10³ cells) per well in a low-attachment 24-well plate. Data are presented as mean± SEM. Statistical comparisons between groups were conducted by unpaired Student's t-test. Statistical significance: *P<0.05, **P<0.01, ***P<0.001, compared with control. P<0.05 was considered to be significant.

Figure 6. USP22 is associated with BMI1 protein stability

(A) Stable USP22-silenced (shUSP22) or control (shCtrl) MGC-803 cells were treated with CHX (20 μM) for 0, 15, 30, 45, 60 and 75 min, and endogenous BMI1 proteins were examined by Western blot. GAPDH was used as an endogenous control. (B) The densitometry curves of BMI1 normalized to GAPDH were plotted against the indicated time points to determine its half-lives according to (A). (C) Western blot was conducted to verify the effect of knockdown of USP22 on PcG protein EZH2 level and H3K27me3. GAPDH was used as a loading control. (D) A proposed working model of the function of USP22 in regulating gastric tumorigenesis.

Figure 7. Overexpression of USP22 correlates with BMI1 in GC tissue samples and is association with poor prognosis

(A) RT-qPCR analysis of mRNA expression of USP22 and BMI1 in 10 pairs of GC (T) and adjacent non-cancerous tissues (T). (B) Western blot to detect USP22 and BMI1 expression in GC clinical specimens. GAPDH was used as a loading control. (C) Representative immunohistochemistry analysis from the human protein atlas (HPA) database. USP22 and BMI1 levels were plotted according to TNM staging. (D) Immunofluorescence staining of GC tissue sections to analyse the expression and co-localization of USP22 and BMI1 in clinical specimens. Arrowheads indicate the USP22 positive/BMI1 positive cells. Scale bar=50 μm. (E) Kaplan-Meier survival plots shows USP22 and overall survival rate of 1,065 patients from the GC database (www.Kmplot.com).