CDK12 and PAK2 as novel therapeutic targets for human gastric cancer

Abstract

Background: Gastric cancer remains the second leading cause of cancer-related death, and the third in mortality due to lack of effective therapeutic targets for late stage cancer patients. This study aims to identify potential druggable target biomarkers as potential therapeutic options for patients with gastric cancer. Methods: Immunohistochemistry of human gastric tumor tissues was conducted to determine the expression level of cyclin-dependent kinase 12 (CDK12). Multiple in vitro and in vivo assays such as RNAi, mass spectrometry, computer docking models, kinase assays, cell xenograft NU/NU mouse models (CDXs) and patient-derived xenograft NOD/SCID mouse models (PDXs) were conducted to study the function and molecular interaction of CDK12 with p21 activated kinase 2 (PAK2), as well as to find CDK12 inhibitors as potential treatment options for human gastric cancer. Results: Here we identified that CDK12 is a driver gene in human gastric cancer growth. Mechanistically, CDK12 directly binds to and phosphorylates PAK2 at T134/T169 to activate MAPK signaling pathway. We further identified FDA approved clinical drug procaterol can serve as an effective CDK12 inhibitor, leading to dramatic restriction of cancer cell proliferation and tumor growth in human gastric cancer cells and PDXs. Conclusions: Our data highlight the potential of CDK12/PAK2 as therapeutic targets for patients with gastric cancer, and we propose procaterol treatment as a novel therapeutic strategy for human gastric cancer.

Keywords: CDK12; PAK2; gastric cancer; phosphorylation; procaterol.

Figure 1

CDK12 is a potential therapeutic target for human gastric cancer. A. The expression of CDK12 was examined by immunohistochemistry (IHC) staining using a gastric tumor array. The upper panels show quantification of all samples and paired samples (n=58). The lower panels show expression of CDK12 in patients with different age, gender, clinical stages and tumor grades. The representative photographs of each group of paired samples are shown. 100 × magnification. B. The expression of CDK12 in human gastric cancer cell lines: HCG27, AGS, SNU-1, KATOIII, and NCI-N87 were evaluated by western blotting (WB). C. The level of CDK12 knock-down by lentiviral transduction (shMock, shCDK12#2 and shCDK12#5) was measured by western blotting in SNU-1 and KATOIII cell lines. D. Cell viability was measured after CDK12 knock-down by MTT (72 h after seeding cells) assay. E. Colony number was checked after CDK12 knocking down by anchorage-independent colony formation assay, representative photographs are shown. F. Protein levels were detected by WB after overexpression of CDK12 or control. G. Cell proliferation was measured by colony formation assay after overexpression of CDK12 or control, representative images are shown. Data represent means ±SD. P<0.05: *, p<0.01: **, p<0.001: ***. Significance determined by two-tailed Student's t test.

Figure 2

Knock-down of CDK12 induces cell cycle arrest and delays tumor progression in vivo. A. Cell cycle was detected by flow cytometry after CDK12 knock-down in SNU-1 and KATOIII cells 24 h after seeding the cells. B. Longitudinal tumor volume, tumor weight, final tumor size and percentage of mice with tumors less than 1000 mm³ at the endpoint were measured after the inoculation of shMock, shCDK12#2 and shCDK12#5 SNU-1 cells into NU/NU mice. Tumor size was measured every 3-7 days, and the volume was calculated by V = (length) × (width) × (height). The tumors in each group are shown. C. Longitudinal tumor volume and tumor weight were measured after injecting lentivirus (shMock, shCDK12#2 and shCDK12#5) directly into patient-derived xenografts (PDX) tumors at ~ 200 mm³. Two different cases (upper panels: LSG50, lower panels: LSG45) were tested. Tumor weight was measured at the endpoints. The tumors in each case are shown. Data represent means ±SD. P<0.05: *, p<0.01: **, p<0.001: ***. Significance determined by two-tailed Student's t test.

Figure 3

CDK12 knocking down inhibits tumor growth by blocking MAPK signaling pathway. A. Representative images and histogram of H&E and CDK12, Ki67, phospho-MEK, phospho-ERK IHC staining after formalin-fixed paraffin-embedded SNU-1 xenografts and PDXs tumors. Data represent means ±SD. P<0.05: *, p<0.01: **, p<0.001: ***. Significance determined by two-tailed Student's t test.

Figure 4

CDK12 interacts with PAK2 and positively correlate with each other. A. SNU-1 cell lysates and recombinant CDK12 and PAK2 proteins were immunoprecipitated with IgG or anti-CDK12/anti-PAK2 antibodies. Immunoprecipitated samples were subjected to western blotting with the indicated antibodies to detect the interaction of CDK12 and PAK2 ex vivo and in vitro. B. The domain architecture of human PAK2 (right) contains an N-terminal Cdc42/Rac interactive binding motif (CRIB) and kinase domain (aa249 to aa499. PAK2 can be cleaved into PAK2-p27 (aa2 to aa212) and PAK2-p34 (aa213 to aa524). (Left) HEK293T cells were transfected with MYC-tagged CDK12 and HA-tagged WT or N-terminally truncated PAK2 including p27, p34, and kinase domain. Immunoprecipitation was performed using anti-HA affinity gel followed by western blots using anti-MYC antibody. The PAK2-p27 binding to CDK12 is essential for this protein-protein interaction. C. The interaction between CDK12 (blue) and PAK2 (red) was predicted using a computational docking model. D. Representative images show that CDK12 (green) and PAK2 (red) are positively co-localized in nucleus (DAPI: blue) and cytoplasm shown by laser scanning confocal microscope through an immunofluorescence assay. E. The expression of PAK2 was examined by IHC staining with a gastric tumor array. The left panels show quantification of all samples and paired samples (n=72). The right panels show representative photographs of each group of paired samples. 100 × magnification. F. The IOD value statistics of CDK12 and PAK2 in human gastric tumor/adjacent tissues with IHC is shown in scatter diagram. CDK12 is positively correlated with PAK2 in human gastric tumors. The representative images are shown to the right from the same patient tissue. The R value calculated by Pearson formula. Data represent means ±SD. P<0.05: *, p<0.01: **, p<0.001: ***. Significance determined by two-tailed Student's t test.

Figure 5

PAK2 act as an oncogene in human gastric cancer. A. The expression level of PAK2 in human gastric cancer cell lines was evaluated by western blotting. B. The level of PAK2 after knock-down by lentiviral transduction (shMock, shPAK2#2 and shPAK2#3) was measured by western blotting in SNU-1 and KATOIII cells. C. Cell viability after PAK2 knock-down was assessed by MTT (72 h after seeding cells) assay. D. Colony number of PAK2 knocking down cells was measured by anchorage-independent colony formation assay. The representative images are shown. E. Longitudinal tumor volume, tumor weight, final tumor size and percentage of mice with tumors less than 1000 mm³ at the endpoint were measured from shMock, shPAK2#2 and shPAK2#3 treated tumors of SNU-1. Tumor size was measured every 5-7 days, and the volume was calculated by V= (length) × (width) × (height). F. The effect of PAK2 knock-down in tumors was examined by WB. The representative images are shown. Data represent means ±SD. P<0.05: *, p<0.01: **, p<0.001: ***. Significance determined by two-tailed Student's t test.

Figure 6

CDK12 phosphorylates PAK2 at threonine 134/169 and activates MAPK signaling pathway. A. Kinase assay of recombinant GST-tagged CDK12 with PAK2 in vitro. CDK12 (200 ng) and inactive PAK2 (400 ng) were mixed with ATP in kinase buffer at 30 ℃ for 30 min. Phospho-Threonine and GST were detected by western blots. B. Kinase assay of mutated PAK2 proteins (WT, T134A, T169A, 2A-T134A/T169A) with CDK12. Phospho-Threonine and GST were detected by western blots. C. Protein level of PAK2 was measured by western blotting after GFP-tagged lentiviral transfection of different mutation type (Vector, WT, 2A) of PAK2 in HGC27 cells. D. Cell viability of different mutation type (Vector, WT, 2A) PAK2 transfected cells was measured by MTT assay (72 h) and representative fluorescence of GFP are shown. E. Cell proliferation of different mutation type (Vector, WT, 2A) PAK2 transfected cells was determined by crystal violet foci formation assay and representative images are shown. F. Co-localization of CDK12 (orange), PAK2 (Vector, WT, 2A, red) and GFP (green) were measured by immunofluorescence after stable overexpression of PAK2 in HGC27. CDK12 and PAK2-WT are evenly distributed throughout the nucleus (DAPI: blue) and cytoplasm, while PAK2-V and PAK2-2A with CDK12 were mainly in nucleus. Representative images are shown. G. MAPK signaling pathway analysis by western blotting. Phospho-MEK and Phospho-ERK was measured by western blots in HEK293T cells with empty vector, PAK2-WT, or PAK2-2A transfected. Representative results are shown. H. Longitudinal tumor volume, tumor weight, final tumor size and percentage of mice with tumors less than 1000 mm³ at the endpoint were measured from empty vector, CDK12-WT, and CDK12-2A-overexpressed tumors of HGC27 in a xenograft model. Tumor size was measured every 5-7 days, the volume was calculated by V = (length) × (width) × (height). Tumors in each group are shown. Data represent means ±SD. P<0.05: *, p<0.01: **, p<0.001: ***. Significance determined by two-tailed Student's t test.

Figure 7

Procaterol is a potent CDK12 inhibitor. A. Colony formation of gastric cancer cells with vehicle control and procaterol (0.5 μM) treatment. Representative images are shown. B. Cell viability in different types of human cancer (gastric cancer, colon cancer, esophageal cancer, and lung cancer) cell lines with vehicle control and procaterol (1 μM) treatment, the representative result of 72 h are shown. C. The binding of procaterol to CDK12 in SNU-1 cell lysates was determined using sepharose 4B and procaterol-conjugated sepharose 4B beads. D. The interaction between procaterol and CDK12 was predicted by a computational docking model. (Left) The representative images show that procaterol binds with CDK12 at kinase responsible site (ASP877) and nucleotide binding site (MET816). (Right) Ligand interaction diagram of procaterol bind with CDK12. E. The effect of procaterol on cell cycle progression of gastric cancer cells. Cells were treated with 0.5, 1 or 2 μM of procaterol and then incubated for 48 h for cell cycle analysis. F. The effect of procaterol on apoptosis in gastric cancer cells. Cells were treated with 0.5, 1 or 2 μM of procaterol and then incubated for 72 h for the Annexin-V staining assay. Representative images are shown. Data represent means ±SD. P<0.05: *, p<0.01: **, p<0.001: ***. Significance determined by two-tailed Student's t test.

Figure 8

Procaterol inhibits gastric tumor growth in PDXs. A. Patient information of different PDX cases. Gender, age, pathological grade, clinical phases, smoking history, alcohol history and basic diseases are shown. B. Effects procaterol (75 μg/kg/day) administered by gavage. Tumor volumes of four different cases of PDX models were recorded. Tumors in each case are shown. Data represent means ±SD. P<0.05: *, p<0.01: **, p<0.001: ***. Significance determined by two-tailed Student's t test.

Figure 9

Procaterol inhibits gastric tumor growth via MAPK signaling pathway verified in PDXs tumors. A. H&E staining and IHC staining of Ki67, p-MEK, p-ERK in PDX tumor tissues. Representative images are shown. 100 × magnification. B. Proposed schematic diagram of CDK12 in whole study. CDK12 binds with and phosphorylates PAK2 at Threonine134/169 to activate MAPK signaling pathway such as p-MEK and p-ERK. This leads to the acceleration of cell proliferation and tumor growth in gastric tumor with high level of CDK12 (left). CDK12 inhibition by procaterol suppresses PAK2 phosphorylation and MAPK signaling (right). Data represent means ±SD. P<0.05: *, p<0.01: **, p<0.001: ***. Significance determined by two-tailed Student's t test.