Volume 63, Issue 1, Pages 102-113 (July 2015) Targeting cyclin dependent kinase 5 in hepatocellular carcinoma – A novel therapeutic approach  Sandra M. Ehrlich, Johanna Liebl, Maximilian A. Ardelt, Thorsten Lehr, Enrico N. De Toni, Doris Mayr, Lydia Brandl, Thomas Kirchner, Stefan Zahler, Alexander L. Gerbes, Angelika M. Vollmar  Journal of Hepatology  Volume 63, Issue 1, Pages 102-113 (July 2015) DOI: 10.1016/j.jhep.2015.01.031 Copyright © 2015 European Association for the Study of the Liver Terms and Conditions

Fig. 1 Expression and activity of Cdk5 in HCC. (A) Immunostainings of HCC and corresponding healthy liver tissue of a human tissue microarray show staining for Cdk5 (red, upper panels) and p35 (red, lower panels) together with hematoxylin (blue, nuclei). Quantitative evaluations of Cdk5 and p35 mean immunoreactivities are displayed in the respective bar graphs. Each: Rank Sum-test, *p<0.001. (B) Immunoblots from patient samples of HCC and normal (N) liver tissue probed with antibodies for Cdk5 and p35 are shown. Quantitative evaluations are displayed. (C) mRNA expression of Cdk5 (upper panel) and p35 (lower panel) from patient samples of HCC and normal (N) liver tissue are shown. (D) Immunoblots show Cdk5 expression and Cdk5 Tyr15 phosphorylation in primary human hepatocytes and HCC/hepatoblastoma cell lines HepG2, HuH7, Hep3B, HuH6, and HepT1. (E) Cdk5 kinase activity is indicated by phosphorylated (p-35) histone 1 (p-H1). The immunoblot in the left panel shows Cdk5 activity in HepG2 and HuH7 cells. The bar graph displays quantification of Histone H1 phosphorylation from HepG2 and HuH7 cells (Rank Sum-test, *p<0.05, n=4). The immunoblot in the right panel shows endogenous Cdk5 activity and Cdk5 activity after overexpression of Cdk5/p35 in HuH7 cell lysates with/without addition of Roscovitine. Precipitation with the IgG isotype antibody serves as negative control, n=3. (This figure appears in colour on the web.) Journal of Hepatology 2015 63, 102-113DOI: (10.1016/j.jhep.2015.01.031) Copyright © 2015 European Association for the Study of the Liver Terms and Conditions

Fig. 2 Genetic downregulation and pharmacological inhibition of Cdk5 decreases HCC cell proliferation and clonogenic survival. (A) Proliferation of nt and Cdk5 shRNA HuH7 cells after synchronization by aphidicolin is shown. Corresponding doubling time is shown. One Way ANOVA, Holm-Sidak *p<0.05, n=3. (B) Proliferation of HuH7 cells treated with Roscovitine at indicated concentrations for 72h is shown. EC50=14.2μM. (C) Proliferation of HepG2 cells transfected with nt or Cdk5 siRNA is shown. Corresponding doubling time is shown. t test, *p<0.05, n=3. (D) Proliferation of HepG2 cells treated with Roscovitine at indicated concentrations is shown EC50=10.9μM. (E) Proliferation of Hep3B cells transfected with nt or Cdk5 siRNA is shown. Corresponding doubling time is shown. t test *p<0.05, n=3. (F) Clonogenic growth of HepG2 cells treated with roscovitine is shown. One Way ANOVA on Ranks, Dunn’s *p<0.05, n=3. (G) Clonogenic growth of HuH7 cells that were either left untreated (con) or treated with roscovitine for 24h and freshly seeded for 7days is shown. One Way ANOVA on Ranks, Dunn’s *p<0.05, n=3. (H) Left panel: Clonogenic growth of nt or Cdk5 siRNA transfected HuH7 cells is shown. t test *p<0.001, n=4. Right panel: Clonogenic growth of nt or Cdk5 shRNA HuH7 cells is shown. One Way ANOVA, Holm-Sidak *p<0.05, n=3. (I) Proliferation of HuH6 cells transfected with empty vector (EV) or Cdk5/p35 is shown. Corresponding doubling time is shown, t test *p<0.05, n=3. (J) Proliferation of HepT1 cells transfected with EV or Cdk5/p35. Corresponding doubling time is shown, t test *p<0.05, n=3. (K) Clonogenic growth of HuH6 cells transfected with EV or Cdk5/p35 is shown, paired t test *p<0.05, n=3. (L) Clonogenic growth of HepT1 cells transfected with EV or Cdk5/p35, paired t test *p<0.05, n=3. (This figure appears in colour on the web.) Journal of Hepatology 2015 63, 102-113DOI: (10.1016/j.jhep.2015.01.031) Copyright © 2015 European Association for the Study of the Liver Terms and Conditions

Fig. 3 Inhibition or distinct knockdown of Cdk5 reduces HCC growth in vivo. (A) Tumors of nt shRNA HuH7 and Cdk5 shRNA HuH7 cells grown in SCID mice are shown (n=11). (B) Tumor volume over the time is shown, t test *p<0.05; n=11. Modeling of tumor growth was performed using an exponential tumor growth model and shows a significantly reduced tumor growth rate of 7.1% of Cdk5 shRNA HuH7 tumors compared to nt shRNA HuH7 tumors (p<0.01). (C) Tumor weight after dissection is shown, t test *p<0.05. (D) Tumors of HuH7 cells grown in SCID mice that were treated with roscovitine or solvent are shown (n=8). (E) Tumor volume over the time is shown. Modeling of tumor growth was performed using an exponential tumor growth model and shows a significantly reduced tumor growth rate of 10.4% by roscovitine treatment compared to solvent treatment (*p<0.001). (F) Tumor weight after dissection is shown. (G and H) Immunostaining of indicated tumors for Ki-67 (red) and hematoxylin (nuclei, blue) is shown. Quantifications of percentages of proliferating cells in the tumors are shown. (G) t test, *p<0.05; n=11. (H) t test, *p<0.05; n=7. (This figure appears in colour on the web.) Journal of Hepatology 2015 63, 102-113DOI: (10.1016/j.jhep.2015.01.031) Copyright © 2015 European Association for the Study of the Liver Terms and Conditions

Fig. 4 Cdk5 is activated and localized in the nucleus of proliferating cells. (A) Immunoblots of HuH7 cells treated with FCS for the indicated times and probed with antibodies for phosphorylated Cdk5 (Tyr15), Cdk5, and actin is shown. (B) Radioactive kinase assays of HuH7 cells treated with FCS for the indicated times are shown. Phosphorylated (p-32) histone 1 (p-H1) indicates Cdk5 activity. (C) Immunoblots from nt and Cdk5 shRNA HuH7 cells treated with FCS as indicated and probed with antibodies for phosphorylated Akt, Akt, phosphorylated ERK and ERK are shown. (D) Immunostainings for Cdk5 (green), Ki67 (red) and Hoechst33342 (blue) are shown. Arrowheads indicate nuclear Cdk5 localization in Ki67 positive cells. Arrows indicate absence of nuclear Cdk5 in Ki67 negative cells. (E) Immunostainings for Cdk5 (green), p27 (red) and Hoechst33342 (blue) are shown. Arrowheads indicate localization of Cdk5 and p27 in the nucleus. Arrows indicate absence of Cdk5 and p27 in the nucleus. (F) Immunoblots of nuclear and cytoplasmic fractions of nt or p27 siRNA transfected HuH7 cells probed with antibodies for p27 and Cdk5 are shown. CREB and β-tubulin are shown as fractionation controls. Quantitative evaluation is shown. One Way ANOVA, Holm-Sidak *p<0.05, n=3. (G) Immunoblots of nuclear and cytoplasmic fractions of HuH7 cells probed with antibodies for phosphorylated Cdk5 and Cdk5 are shown. (H) Radioactive kinase assays of nuclear and cytoplasmic fractions from HuH7 cells are shown. Phosphorylated (p-32) histone 1 (p-H1) indicates Cdk5 activity. (This figure appears in colour on the web.) Journal of Hepatology 2015 63, 102-113DOI: (10.1016/j.jhep.2015.01.031) Copyright © 2015 European Association for the Study of the Liver Terms and Conditions

Fig. 5 Cdk5 is active during G2 cell cycle phase and gets activated by treatment with DNA damaging agents. (A) Cell cycle progression of HuH7 cells after synchronization by aphidicolin (10μM, 24h) and subsequent release is shown. (B) Nuclear and cytosolic fractions from HuH7 cells that were synchronized with aphidicolin and released for the indicated times were immunoblotted and probed with antibodies for phosphorylated Cdk5, Cdk5, and p27. Immunoblots for CREB and tubulin indicate fractionation and equal loading. (C) Radioactive kinase assays of nuclear and cytoplasmic fractions from HuH7 cells 0h, 12h, and 24h after synchronization are shown. Phosphorylated (p-32) histone 1 (p-H1) indicates Cdk5 activity. (D) Immunoblots of HuH7 cells treated for 24h with Sn38 (5ng/ml) or ABT-888 (100μM) probed with antibodies for phosphorylated Cdk5, Cdk5 and actin are shown. (E) Radioactive kinase assay of nuclear fractions from HuH7 cells treated with Sn38 (5ng/ml) or ABT-888 (100μM) for 24h are shown. Phosphorylated (p-32) histone 1 (p-H1) indicates Cdk5 activity. (F) Immunoblots from nt and Cdk5 shRNA HuH7 cells treated with Sn38 (5ng/ml) and probed with antibodies for p-H2A.X and actin are shown. (G) Immunostaining for p-H2A.X (red) and Hoechst33342 (blue, nuclei) from nt and Cdk5 shRNA HuH7 cells are shown. (H) Immunoblots from nt and Cdk5 shRNA HuH7 cells with Sn38 (5ng/ml) treatment and removal for the indicated times probed with antibodies for p-H2A.X are shown. (This figure appears in colour on the web.) Journal of Hepatology 2015 63, 102-113DOI: (10.1016/j.jhep.2015.01.031) Copyright © 2015 European Association for the Study of the Liver Terms and Conditions

Fig. 6 Cdk5 is important for DNA damage checkpoint activation. (A) Cell cycle analysis from nt (blue) and Cdk5 (red) shRNA HuH7 cells after treatment with Sn38 (5ng/ml), ABT-888 (100μM) or nocodazol (50ng/ml) is shown. Bar graphs display respective quantitative evaluations. (B) Cell cycle analysis from nt (blue) and Cdk5 (red) shRNA HuH7 cells after synchronization and treatment with Sn38 (5ng/ml) is shown. Bar graphs display respective quantitative evaluations. (This figure appears in colour on the web.) Journal of Hepatology 2015 63, 102-113DOI: (10.1016/j.jhep.2015.01.031) Copyright © 2015 European Association for the Study of the Liver Terms and Conditions

Fig. 7 Cdk5 is involved in ATM phosphorylation and thereby regulates DNA damage response. (A and B) Immunoblots from nt and Cdk5 HuH7 cells treated with (A) Sn38 (5ng/ml) or (B) ABT-888 (100μM) and probed with antibodies for phosphorylated ATM (p-ATM), p-Chk2, p-BRCA1, p-cdc2, p-ATR, and p-Chk1 are shown. (C) Immunoblots from nt and Cdk5 HuH7 cells treated with Sn38 (5ng/ml) and probed with antibody for p-ATM (Ser794). Journal of Hepatology 2015 63, 102-113DOI: (10.1016/j.jhep.2015.01.031) Copyright © 2015 European Association for the Study of the Liver Terms and Conditions

Fig. 8 Cdk5 knockdown sensitizes HCC cells for treatment with DNA damaging agents and sorafenib. (A) Proliferation of nt and Cdk5 shRNA HuH7 cells that were synchronized by aphidicolin and treated with Sn38 (1ng/ml), ABT-888 (10μM), etoposid (0.1μM), and doxorubicin (5nM) for the indicated times is shown. Representative curves and evaluation of the corresponding doubling time is displayed. Sn38 and doxorubicin: One Way ANOVA, Holm-Sidak *p<0.05, n=3; ABT-888 and etoposid: One Way ANOVA on Ranks, Dunn’s *p<0.05, n=3. (B) Colony formation of nt and Cdk5 shRNA HuH7 cells treated as indicated for 24h is shown. Sn38 and etoposid: One Way ANOVA, Holm-Sidak *p<0.05, n=3; ABT-888 and doxorubicin: One Way ANOVA on Ranks, Dunn’s *p<0.05, n=3. (C) Apoptosis rate of nt and Cdk5 shRNA HuH7 cells after Sn38 (5ng/ml) treatment (48h) and subsequent removal (7 d). One Way ANOVA, Holm-Sidak *p<0.05, n=3. (D) Tumors of HuH7 cells grown in SCID mice that were treated with solvent (control, con), roscovitine (rosco), irinotecan (irino), or a combination of roscovitine and irinotecan (rosco+irino) are shown (control: n=8, roscovitine: n=7, irinotecan: n=7, combination: n=8). Tumor volume over the time after end of treatment is shown. Modeling of tumor growth was performed using an exponential tumor growth model and demonstrated a significantly reduced tumor growth rate by combined treatment with roscovitine and irinotecan (p<0.05). (This figure appears in colour on the web.) Journal of Hepatology 2015 63, 102-113DOI: (10.1016/j.jhep.2015.01.031) Copyright © 2015 European Association for the Study of the Liver Terms and Conditions