A kinase-independent function of CDK6 links the cell cycle to tumor angiogenesis

Abstract

In contrast to its close homolog CDK4, the cell cycle kinase CDK6 is expressed at high levels in lymphoid malignancies. In a model for p185BCR-ABL+ B-acute lymphoid leukemia, we show that CDK6 is part of a transcription complex that induces the expression of the tumor suppressor p16INK4a and the pro-angiogenic factor VEGF-A. This function is independent of CDK6's kinase activity. High CDK6 expression thus suppresses proliferation by upregulating p16INK4a, providing an internal safeguard. However, in the absence of p16INK4a, CDK6 can exert its full tumor-promoting function by enhancing proliferation and stimulating angiogenesis. The finding that CDK6 connects cell-cycle progression to angiogenesis confirms CDK6's central role in hematopoietic malignancies and could underlie the selection pressure to upregulate CDK6 and silence p16INK4a.

Graphical abstract

Figure 1

Enforced Expression of CDK6 in p185^BCR-ABL-Transformed Cells Unmasks Its Tumor-Suppressing Activity (A) ³[H]-thymidine incorporation of Cdk6^+/+ (expressing a pMSCV-puro-based retrovirus) and Cdk6^+/++Cdk6 (expressing a pMSCV-Cdk6-puro-based retrovirus) p185^BCR-ABL-transformed cells (n = 4; ^∗∗p = 0.009). (B) Colony-forming assays were performed by seeding a defined number of p185^BCR-ABL-transformed Cdk6^+/+ or Cdk6^+/++Cdk6 cells in growth factor-free methylcellulose. Left side: representative set of pictures. Right side: number of colonies per dish after incubation for 5 days (n = 4/genotype; ^∗p = 0.023). (C) Kaplan-Meier plot of Nu/Nu mice injected subcutaneously with Cdk6^+/+ or Cdk6^+/++Cdk6 cells (n = 4 cell lines/genotype; n = 2 mice/cell line; mean survival: 18 [Cdk6^+/+] versus 26 [Cdk6^+/++Cdk6] days; ^∗∗p = 0.002). (D) Kaplan-Meier plot of Rag2^−/− mice transplanted intravenously with Cdk6^+/+ or Cdk6^+/++Cdk6 cells (n = 3 cell lines/genotype; n = 6 mice/genotype; mean survival: 18 [Cdk6^+/+] and 29 [Cdk6^+/++Cdk6] days; ^∗p = 0.014). (E and F) Immunohistochemical stainings for the proliferation marker Ki-67 of Cdk6^+/+ (n = 4) and Cdk6^+/++Cdk6 (n = 9) tumors were quantified with HistoQuest software. A representative set of pictures is given (E). Original magnification 20×. Bar graphs depict percentage of Ki-67-positive tumor cells (F; p = 0.029 [^∗]). Error bars indicate the mean ± SEM. See also Figure S1.

Figure 2

CDK6 Acts as a Transcriptional Regulator of p16^INK4a (A) Immunoblot for CDK6 and p16^INK4a of three individually derived Cdk6^+/+ cell lines with (+Cdk6) or without (w/o vector, empty vector) enforced CDK6 expression. (B) p16^INK4a mRNA levels of Cdk6^+/+ and Cdk6^+/++Cdk6 cells were analyzed with qPCR (n = 6; ^∗p = 0.03). (C) Immunoblot for CDK6 and p16^INK4a of Cdk6^−/− cells expressing a doxycycline-inducible tet-on Cdk6 vector (Cdk6^−/−+tet-on Cdk6). Lane 1: Cdk6^−/− cells. Lanes 2–5: Cdk6^−/−+tet-on Cdk6 cells 48 hr after treatment with doxycycline (0, 0.3, 1, 3 μM). Lane 6: Cdk6^+/+ cells. Lane 7: Cdk6^+/++Cdk6 cells. (D) ChIP assays were performed using Cdk6^+/++Cdk6 and Cdk6^−/− cells. Protein-DNA complexes were immunoprecipitated using an anti-CDK6-antibody and analyzed by PCR for the presence of p16^INK4a and as specificity control of ARF promoter sequence (region depicted in Figure 2F as open rectangles). (E) Promoter ChIP assays were performed using Cdk6^+/+, Cdk6^+/++Cdk6, and Cdk6^−/− cells. Protein-DNA complexes were immunoprecipitated using antibodies specific for the indicated histone modification. ChIP DNA was analyzed with qPCR for the presence of a p16^INK4a promoter sequence (region 1 in Figure 2F). The relative enrichment to a control region is shown (Tbp promoter for H3K9ac, H3K4me2, and H3K4me3; Neurog1 promoter for H3K27me3). The mean and SEM of two independent experiments is shown (H3K4me3: Cdk6^+/+ versus Cdk6^−/−, ^∗p < 0.05; H3K27me3: Cdk6^+/+ versus Cdk6^+/++Cdk6,^∗∗p < 0.01; Cdk6^+/++Cdk6 versus Cdk6^−/−, ^∗p < 0.05). (F) H3K36me3 ChIP assays were performed using Cdk6^+/+ and Cdk6^+/++Cdk6 cells. Protein-DNA complexes were immunoprecipitated using an antibody specific for H3K36me3. ChIP DNA was analyzed with qPCR for the presence of p16^INK4a sequences 1–4, which are depicted as filled black rectangles in the lower panel (middle of the amplicon relative to the TSS [arrow symbol marked +1] of p16^INK4a: 1, −169 base pairs; 2, +538 base pairs; 3, +2,278 base pairs; 4, +4,312 base pairs). The relative enrichment to a Gapdh gene body region is shown. The mean and SEM of two independent experiments is shown (sequence 3: Cdk6^+/+ versus Cdk6^+/++Cdk6,^∗p < 0.05; sequence 4: Cdk6^+/+ versus Cdk6^+/++Cdk6,^∗∗p < 0.01). (G) Immunoblot for CDK6 and p16^INK4a of Cdk6^+/+, Cdk6^+/++Cdk6, and Cdk6^+/++Cdk6K43M cells. See also Figure S2.

Figure 3

Inverse Relation between CDK6 and p16^INK4a Expression in Human Lymphomas (A) [³H]thymidine incorporation of Cdk6^+/+, INK4a/ARF^−/−, INK4a/ARF^−/−+Cdk6, and INK4a/ARF^−/−+Cdk6R31C p185^BCR-ABL-transformed cells (n = 3; Cdk6^+/+ versus INK4a/ARF^−/−, ^∗∗p = 0.002; INK4a/ARF^−/−+Cdk6, ^∗∗p = 0.001; INK4a/ARF^−/−+Cdk6R31C,^∗∗p = 0.005). (B) Kaplan-Meier plot of Rag2^−/− mice intravenously transplanted with Cdk6^+/+, INK4a/ARF^−/−, INK4a/ARF^−/−+Cdk6, and INK4a/ARF^−/−+Cdk6R31C cells (n = 3 cell lines/genotype; mean survival: 18 [Cdk6^+/+], 15 [INK4a/ARF^−/−], 15 [INK4a/ARF^−/−+Cdk6], and 15.5 [INK4a/ARF^−/−+Cdk6R31C] days; ^∗p = 0.02). (C) Immunohistochemical stainings of a B cell lymphoma tissue array including 16 lymphoma samples and two control lymph nodes for CDK6 and p16^INK4a. Representative examples including different types of B cell lymphoma, a diffuse large B cell lymphoma (patient A), two follicular lymphomas (patients B and C), and one control lymph node (left) are depicted. Original magnification 20×. (D and E) The expression of CD30, CDK6, and p16^INK4a was analyzed for 17 NPM-ALK-positive (E) and 11 NPM-ALK-negative (D) lymphoma cases with immunohistochemistry. Representative cases for NPM-ALK-positive (patients D and E) and NPM-ALK-negative (patients F and G) cases are depicted. Original magnification 20×. Error bars indicate the mean ± SEM. See also Figure S3.

Figure 4

CDK6 Regulates NPM-ALK-Induced Disease Progression (A) Cdk6^+/+, Cdk6^+/−, and Cdk6^−/− mice crossed with NPM-ALK transgenic (tg) mice developed a T cell lymphoma after several weeks (n ≥ 8; mean survival 120 [Cdk6^+/+], 143.5 [Cdk6^+/−], and 212 [Cdk6^−/−] days; ^∗∗p = 0.001). (B) [³H]thymidine incorporation of NPM-ALK-transformed Cdk6^+/+ and Cdk6^−/− cells (n ≥ 3; ^∗p = 0.009). (C) Immunoblot for CDK6 and p16^INK4a of NPM-ALK-transformed Cdk6^+/+ and Cdk6^−/− cells. (D) Methylation-specific PCR of a part of the p16^INK4a promoter CpG island in NPM-ALK-transformed murine cells. The visible PCR product indicates the presence of methylated alleles. wt BM, bone marrow of a healthy mouse; pos. contr., control for methylated samples; neg. contr., control for unmethylated samples. (E) Immunoblot for CDK6 of untransformed human T cells (three individually derived samples), human T-lymphoid leukemic cell lines (Sudhl1, CCRF, PEER), untransformed murine T cells (two individually derived samples), murine Cdk6^+/+ NPM-ALK-transformed cells, untransformed murine B cells, untransformed human B cells, and human B-lymphoid leukemic cell lines (SUP-B15). (F) Immunoblot for CDK6 of untransformed human B cells (three individually derived samples), human B-lymphoid leukemic cell lines (SUP-B15, RL-7, JVM2), untransformed murine B cells (two individually derived samples), murine Cdk6^+/+ p185^BCR-ABL-transformed cells (three individually derived cell lines), and murine Cdk6^+/++Cdk6 p185^BCR-ABL-transformed cells (three individually derived cell lines). Error bars indicate the mean ± SEM. See also Figure S4.

Figure 5

CDK6 Regulates Tumor Angiogenesis (A–C) Cdk6^+/+, Cdk6^−/−, Cdk6^−/−+Cdk6, Cdk6^−/−+Cdk6R31C, or Cdk6^−/−+Cdk6K43M p185^BCR-ABL-transformed cells were injected subcutaneously (sc) into Nu/Nu mice (n = 3 cell lines/genotype; n ≥ 6 tumors/genotype). (A) Immunofluorescence staining for CD31 (red) was performed to analyze blood vessel formation in sc tumors. Original magnification 20×. Representative cases of each genotype are depicted. (B) Quantitative assessment (HistoQuest) of the blood vessels of the subcutaneous tumors (n ≥ 4 tumors of three independent cell lines; Cdk6^−/− versus: Cdk6^+/+, ^∗∗p = 0.001; Cdk6^−/−+Cdk6, ^∗∗∗p < 0.0001; Cdk6^−/−+Cdk6R31C, ^∗∗∗p = 0.0002; Cdk6^−/−+Cdk6 K43M, ^∗∗∗p < 0.0001). (C) Tumor weight was detected after 8 days of injection (Cdk6^−/− versus: Cdk6^+/+, ^∗∗p = 0.003; Cdk6^−/−+Cdk6R31C, ^∗∗∗p = 0.0001). (D and E) Murine endothelial cell (mEC) spheroids were cultured in methylcellulose with 20% supernatant derived from indicated cells for 24 hr. (D) Quantitative analysis of the relative sprout length was measured with ImageJ software (n ≥ 3; Cdk6^−/− versus: Cdk6^+/+, ^∗p = 0.042; Cdk6^−/−+Cdk6, ^∗∗p = 0.002; Cdk6^−/−+Cdk6R31C, ^∗∗p = 0.009; Cdk6^−/−+Cdk6K43M, ^∗p = 0.038). (E) One representative set of pictures is given. (F and G) A monolayer wounding assay was performed to analyze migration of mECs incubated with supernatant derived from indicated cells. After 2 and 24 hr, pictures were taken and mEC migration quantified (% open area after 24 hr) with TScratch Software. (F) n ≥ 5; Cdk6^−/− versus: Cdk6^+/+, ^∗∗∗p = 0.0007; Cdk6^−/−+Cdk6, ^∗∗p = 0.0014; Cdk6^−/−+Cdk6R31C, ^∗∗p = 0.0092; Cdk6^−/−+Cdk6K43M, ^∗∗∗p = 0.0009). (G) One representative set of pictures is given. Error bars indicate the mean ± SEM. See also Figure S5.

Figure 6

CDK6 Regulates Transcription of the Pro-Angiogenic Factor Vegf-A (A) Ex vivo VEGF-A protein levels (pg/ml) of the subcutaneous tumors (see Figures 5A–5C) were analyzed with an ELISA experiment (n = 3). (B) Relative Vegf-A mRNA levels of indicated cells were analyzed by qPCR. The fold change compared to Cdk6^−/−Vegf-A mRNA level is shown (n ≥ 4; Cdk6^−/− versus: Cdk6^−/−+Cdk6, ^∗∗∗p < 0.0001; Cdk6^−/−+Cdk6R31C, ^∗p = 0.03; Cdk6^−/−+Cdk6K43M, ^∗∗p = 0.004). (C and D) ChIP assays were performed on (C) Cdk6^−/− and Cdk6^−/− cells expressing an HA-tagged CDK6 using different amounts of an anti-HA-antibody as well as on (D) Cdk6^−/− and Cdk6^+/++Cdk6 cells using an anti-CDK6-antibody. PCR was performed to detect the Vegf-A promoter sequence. (E) Dose-response curve of Cdk6^+/+ p185^BCR-ABL-transformed cells treated 24 hr with the CDK6/4 inhibitor PD0332991 (n = 3). (F) Vegf-A mRNA levels of Cdk6^+/+ p185^BCR-ABL-transformed cells (n = 3) treated 24 hr with 0, 30, 100, 300, and 1000 nM PD0332991 were analyzed by qPCR. The fold change compared to untreated Vegf-A mRNA levels is shown. (G) p16^INK4a mRNA levels of Cdk6^+/+ p185^BCR-ABL–transformed cells (n = 3) treated 24 hr with 0, 30, 100, 300, and 1,000 nM PD0332991 were analyzed with qPCR. The fold change compared to untreated p16^INK4a mRNA levels is shown. Error bars indicate the mean ± SEM. See also Figure S6.

Figure 7

Transcriptional Interaction Partners of CDK6 (A) An anti-STAT3 co-immunoprecipitation (co-IP) was performed with Stat3^Δ/Δ, Cdk6^−/−, Cdk6^+/+, and Cdk6^+/++Cdk6 cell extracts and immunoblotted for STAT3 and CDK6. (B) An anti-c-JUN co-IP was performed with c-Jun^Δ/Δ, Cdk6^−/−, Cdk6^+/+, and Cdk6^+/++Cdk6 cell extracts and immunoblotted for c-JUN and CDK6. (C and D) Vegf-A mRNA levels of (C) c-Jun^Δ/Δ versus c-Jun^Δ/Δ+Cdk6 and (D) Stat3^Δ/Δ versus Stat3^Δ/Δ+Cdk6 p185^BCR-ABL-transformed cells were analyzed with qPCR (n ≥ 3; Stat3^Δ/Δversus Stat3^Δ/Δ+Cdk6: ^∗p = 0.025). (E and F) p16^INK4a mRNA levels of (E) c-Jun^Δ/Δ versus c-Jun^Δ/Δ+Cdk6 and (F) Stat3^Δ/Δ versus Stat3^Δ/Δ+Cdk6 p185^BCR-ABL-transformed cells were analyzed with qPCR (n ≥ 3; c-Jun^Δ/Δ versus c-Jun^Δ/Δ+Cdk6: ^∗∗p = 0.008). (G) A potential interaction between CDK6 and c-JUN or STAT3 was analyzed in Cdk6^+/++Cdk6 cells with ChIP-Re-ChIP experiments at the promoter regions of Vegf-A and p16^INK4a. Antibodies used for ChIP (1st AB) and Re-ChIP (2nd AB) are shown on the right. (H–K) Vegf-A (H and I) and p16^INK4a (J and K) mRNA levels of Cyclin D1/2/3^−/− MEFs versus cyclin D1/2/3^−/− MEFs enforced expressing CDK6 as well as cyclin D1/2/3^+/+ MEFs versus cyclin D1/2/3^+/+ MEFs enforced expressing CDK6 were analyzed with qPCR. Error bars indicate the mean ± SEM. See also Figure S7.

Figure 8

Schematic Representation of the Kinase-Dependent and the Noncanonical Kinase-Independent Functions of CDK6