Cyclin C promotes development and progression of B-cell acute lymphoblastic leukemia by counteracting p53-mediated stress responses

Abstract

Despite major therapeutic advances in the treatment of acute lymphoblastic leukemia (ALL), resistances and long-term toxicities still pose significant challenges. Cyclins and their associated cyclin-dependent kinases are one focus of cancer research when looking for targeted therapies. We discovered cyclin C to be a key factor for B-cell ALL (B-ALL) development and maintenance. While cyclin C is not essential for normal hematopoiesis, CcncΔ/Δ BCR::ABL1+ B-ALL cells fail to elicit leukemia in mice. RNA sequencing experiments revealed a p53 pathway deregulation in CcncΔ/Δ BCR::ABL1+ cells resulting in the inability of the leukemic cells to adequately respond to stress. A genome-wide CRISPR/Cas9 loss-of-function screen supplemented with additional knock-outs unveiled a dependency of human B-lymphoid cell lines on CCNC. High cyclin C levels in B-cell precursor (BCP) ALL patients were associated with poor event-free survival and increased risk of early disease recurrence after remission. Our findings highlight cyclin C as a potential therapeutic target for B-ALL, particularly to enhance cancer cell sensitivity to stress and chemotherapy.

Figure 1.

Malignant B-lymphoblastic cells depend on cyclin C. (A) CRISPR dependencies for CCNC, CDK8 and CDK19 in a panel of human cancer cell lines obtained from the Broad Institute Cancer Dependency Map (DepMap). Dots correspond to individual cell lines (N=1,095), y-axis represents CRISPR dependency score (DepMap Public 23Q2+Score, Chronos). Negative gene effect scores imply dependency of a cell line on a given gene as gene knock-out results in impaired cell line growth and/or death. Scores are normalized so that all non-essential genes have a median score of 0 and all common essential genes have a median score of -1. Cell lines with probability of dependency >0.5 are considered dependent. (B) Representation of the top 8 significantly enriched lineages (P<0.0005, t test) in the Depmap CRISPR dependency screen for CCNC. The number of cell lines included in each lineage subset is denoted in parentheses. Ranking was based on effect size.

Figure 2.

Cyclin C is not essential for normal hematopoiesis. (A) (Left) Breeding scheme for the generation of Ccnc^fl/fl VavCre^-/-(Ccnc^fl/fl) and Ccnc^fl/fl VavCre^+/- (Ccnc^fl/fl VavCre) mice. (Right) Cyclin C immunoblot analysis of bone marrow (BM) and spleen from Ccnc^fl/fl and Ccnc^fl/fl VavCre mice (N=3 per genotype). HSC70 served as loading control. (B) Lin-Sca-1⁺c-kit⁺ (LSK) frequencies in BM of Ccnc^fl/fl (N=12) and Ccnc^fl/fl VavCre (N=11) mice, one representative out of 2 independent experiments is shown. (C) Relative fold change of BM subpopulations from Ccnc^fl/fl VavCre normalized to the mean value from Ccnc^fl/fl mice (N=9-12). (D) (Left) Representative flow cytometry plots depicting the gating scheme for early B-cell development populations according to expression of the markers CD43, B220, BP-1, CD19, IgM, IgD and (right) summary of frequencies in the BM of Ccnc^fl/fl (N=12) and Ccnc^fl/fl VavCre (N=10) mice normalized to mean values from Ccnc^fl/fl mice. Details on flow cytometric analyses are provided in the Online Supplementary Methods. Bar graphs represent mean ± Standard Deviation. Box and whiskers plot center values represent median, the box 25^th to 75^th percentiles, and whiskers minimum to maximum. (B-D) Levels of significance were calculated using Mann-Whitney U test. HSC: hematopoietic stem cells; MPP: multipotent progenitor population; CLP: common lymphoid progenitors.

Figure 3.

Cyclin C plays a pivotal role in the oncogenic transformation and immortalization of BCR::ABL1^p185+ B-cell acute lymphoblastic leukemia. Bone marrow (BM) cells from Ccnc^fl/fl and Ccnc^fl/fl VavCre mice were isolated and either (A) infected with a retrovirus encoding BCR::ABL1^p185 prior to plating in growth-factor free methylcellulose or (B) directly plated in methylcellulose containing interleukin-7 (IL-7). Pictures show individual colonies for each genotype. The number of colonies (colony forming units [CFU]) were counted and are depicted normalized to mean values from Ccnc^fl/fl mice. (A) N=12 per genotype, pooled from 4 independent experiments, performed in technical duplicates. (B) N=3 per genotype, performed in technical duplicates. One representative result from 2 experimental set-ups with different concentrations of plated BM cells is shown. (C) Statistics on the percentage of immortalized monoclonal cell lines proliferating in FCS-supplemented medium after picking individual colonies from a BCR::ABL1^p185-induced colony formation assay (N=14-30 picked colonies / genotype). Error bars represent Confidence Intervals (CI) calculated using the Wilson Score interval (95% CI). (D-F) BM cells from Ccnc^fl/fl and Ccnc^fl/fl VavCre mice were isolated, transformed with a pMSCV-Bcr-Abl1-p185-IRES-eGFP-based retrovirus, and cultured in liquid medium to monitor outgrowth. (D) Bar graphs summarizing the result of Annexin/7-AAD stainings performed six weeks after transformation of BM of Ccnc^fl/fl and Ccnc^fl/fl VavCre mice with the BCR::ABL1^p185oncogene. Frequencies of living (Annexin^- 7-AAD^-), early apoptotic (Annexin⁺ 7-AAD^-), and late apoptotic/necrotic (Annexin⁺ 7-AAD⁺) fractions from N=6-8 biological replicates / genotype are shown. (E) Growth curves of stable Ccnc^fl/fl and Ccnc^Δ/Δ BCR::ABL1^p185+ cell lines 12 weeks after initial transformation (N=4-5 cell lines/genotype, performed in technical duplicates). One representative result out of 3 independent experiments is depicted. (F) Flow cytometric analysis after PI cell cycle staining of Ccnc^fl/fl and Ccnc^Δ/Δ BCR::ABL1^p185+ cell lines (N=3 stable, independently established cell lines / genotype) in standard medium supplemented with 10% FCS and 24 hours after FCS removal (0% FCS). Experiments were performed in technical duplicates 12-18 weeks after initial transformation. (A, B, D, E, F) Graphs represent mean ± Standard Deviation. Levels of significance were calculated using (A, B, D) Mann-Whitney U-test, (C) Fisher’s exact test, or (E) unpaired t test on log-transformed counts from day 7 post seeding. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

Figure 4.

Cyclin C deficiency impairs in vivo leukemia establishment. (A) Scheme depicting experimental setup of data shown in (B) and (C). 2,500 Ccnc^fl/fl or Ccnc^Δ/Δ BCR::ABL1p¹⁸⁵⁺cells were injected intravenously (i.v.) into NSG mice (N=9-10/genotype, 3 independent cell lines per genotype were injected). (B) Representative pictures of spleens on day 26 post injection. (C) Representative pictures of blood smears on day 26 post injection after Hemacolor Rapid staining. Blasts are indicated with turquoise arrows. (D) Scheme depicting experimental setup for data shown in (E) and (F). (E) Intravenous injection of Ccnc^fl/fl and Ccnc^Δ/Δ BCR::ABL1^p185+ cells into NSG mice (N=9-10/genotype, 3 independent cell lines per genotype were injected, data pooled from 2 independent experiments). Survival of recipient mice was monitored for up to a maximum of 224-281 days in case of absent disease symptoms. Median survival of mice receiving Ccnc^fl/fl injections was 23 days; survival of the only diseased mouse in the Ccnc^Δ/Δ cohort (high-lighted turquoise) was 54 days. Black triangles indicate time points at which mice were eliminated without appearance of disease symptoms. Level of significance was calculated using log rank (Mantel-Cox) test. (F) Flow cytometric analysis of bone marrow (BM) infiltration in mice receiving Ccnc^fl/fl and Ccnc^Δ/Δ BCR::ABL1^p185+ injections as depicted in (D) and (E). Graph shows mean ± Standard Deviation. Level of significance was calculated using Mann-Whitney U test. ****P<0.0001.

Figure 5.

Cyclin C represses p53 responses in transformed cells. (A) Scheme depicting experimental setup for RNA sequencing (RNA-Seq) analyzed in (B) and (C). Bone marrow (BM) from Ccnc^fl/fl and Ccnc^fl/fl VavCre mice was isolated and infected with retrovirus encoding BCR::ABL1^p185 to generate stable cell lines (“in vitro” samples) which were intravenously (i.v.) injected into NSG mice and retrieved from bone marrow of recipient mice ten days later (“ex vivo” samples). (B) Venn Diagram showing the number of differentially expressed genes (Benjamini-Hochberg adjusted P value <0.1) in Ccnc^Δ/Δ versus Ccnc^fl/fl BCR::ABL1^p185+ cells in vitro (blue) and ex vivo (orange). Intersecting area shows the overlap between the in vitro and ex vivo datasets. (C) Significantly upregulated hallmark gene sets (normalized enrichment score >1, false discovery rate <0.2, P<0.05) from Gene Set Enrichment Analysis (GSEA) of ex vivo derived Ccnc^Δ/Δ versus Ccnc^fl/fl BCR::ABL1^p185+ cells which were not significantly enriched in cyclin C-deficient BCR::ABL1^p185+ cell lines in vitro. (D) In vitro proliferation of Ccnc^fl/fl and Ccnc^Δ/Δ BCR::ABL1^p185+ cells in standard culture conditions: medium supplemented with 10% fetal calf serum (FCS) (top), after reducing FCS to 1% (bottom left) and in reduced cell density (bottom right) (N=4-5 cell lines / genotype). (E) Immunoblot analysis of stable Ccnc^fl/fl and Ccnc^Δ/Δ BCR::ABL1^p185+ cell lines (N=3 per genotype) in standard culture medium (10% FCS) and after reducing FCS to 1% for 24 hours (hr) and 96 hr. Levels of cyclin C, cyclin D2, PLK2, NDRG1, GADD45a and 14-3-3σ (Sfn) were assessed on 3 separate immunoblots with the same lysates. HSC70 served as loading control. EMT: epithelial-mesenchymal transition; ROS: reactive oxygen species.

Figure 6.

Disruption of functional p53 signaling restores the leukemogenicity of cyclin C-deficient BCR::ABL1^p185+ cells. (A) In vitro proliferation of Ccnc^fl/fl and Ccnc^Δ/Δ BCR::ABL1^p185+ cells six days after direct delivery of a ribonucleoprotein (RNP) complex consisting of Cas9 enzyme and guide RNA targeting Gadd45a, Sfn or Plk2. A non-targeting control (sgNTC) and mock treated cells (-) served as controls. Growth curves were performed in technical duplicates in standard culture conditions: medium supplemented with 10% fetal calf serum (FCS) (top), with lowered FCS reduced to 1% (bottom left) or reduced cell density (bottom right). Graphs show mean ± Standard Deviation. Levels of significance were determined by one-way ANOVA followed by Dunnett’s test comparing log-transformed counts from day 6 post seeding with sgNTC as control group for Ccnc^fl/fl and Ccnc^Δ/Δ BCR::ABL1^p185+ cells, respectively. (B) Immunoblot analysis showing levels of cyclin C and p53 in spleen infiltrating BCR::ABL1^p185+ cells of diseased mice from experiment shown in Figure 4E. HSC70 served as loading control. (C) Kaplan-Meier plot of NSG mice after intravenous injection of Ccnc^fl/fl and Ccnc^Δ/Δ BCR::ABL1^p185+ cells carrying spontaneous mutations in the DNA binding domain of p53 (p53^mut) (N=4 per genotype, median survival 16 vs. 18 days). (D) Stable Ccnc^fl/fl and Ccnc^Δ/Δ BCR::ABL1^p185+ cell lines were infected with retrovirus encoding dominant negative p53 (dn p53), intravenously injected into NSG mice and survival was monitored: 2 independent cell lines per genotype were used, N=5 mice were injected with Ccnc^fl/fl + dn p53 and N=5 received Ccnc^Δ/Δ + dn p53 BCR::ABL1^p185+ cells; median survival 20 vs. 22 days. Levels of significance were determined using (C and D) log rank (Mantel-Cox) test . *P<0.05, **P<0.01, ***P<0.001.

Figure 7.

Cyclin C as a potential new target for B-cell acute lymphoblastic leukemia treatment. (A) Gene expression profile of CCNC in control (bone marrow [BM] mononuclear cells, N=56) versus primary B-cell precursor acute lymphoblastic leukemia (BCP-ALL) samples (N=362), presented as violin plot. Sample size per subtype: BCR::ABL1: N=97; BCR::ABL1-like: N=51; ETV6::RUNX1: N=6; KMT2A rear: N=56; ZNF384 rear: N=20; low hypodiploid: N=45; high hyperdiploid: N=17; near haploid: N=2; TCF3::PBX1: N=6; DUX4 rear: N=22; PAX5 P80R: N=14; PAX5alt: N=10: BCL2/MYC: N=4; CDX2/UBTF: N=7; NUTM1 rear: N=2; ZEB2/CEBP: N=3. (B) Analysis of 12 human B-lymphoblastic leukemia / lymphoma cell lines showing p53 status and dependency probabilities for CCNC from a genome-wide CRISPR/Cas9 knock-out screen (DepMap Public 23Q2+Score, Chronos). Cell lines with dependency probabilities >0.5 are considered dependent. (C) Immunoblot showing cyclin C levels in bulk cell culture after CRISPR/Cas9 mediated targeting of cyclin C in NALM-6 cells. Three different guide RNA targeting cyclin C (sgCCNC) were used, guide RNA targeting HPRT1 served as control (Ctrl). HSC70 was used as loading control. (D) Bulk cell lines depicted in (E) were single cell sorted using a BD FACSAria III cell sorter and outgrowth of single cell clones was monitored. One representative result is shown; a similar result was obtained using limiting dilution to generate monoclonal cell lines. (E) Probability of event-free survival (EFS) in pediatric BCP-ALL patients with high versus low CCNC expression. The high / low CCNC cut-off was based on median expression among BCP-ALL samples in cohort. Death in induction, death, relapse, non-response, and secondary malignancy were counted as events for EFS. (F) Proportion of pediatric BCP-ALL patients who reached remission after induction treatment, but for which an event (relapse / death) was reported prior to 36 months. Error bars represent 95% Confidence Interval calculated using the adjusted Wald method. Levels of significance were determined using (E) log rank (Mantel-Cox) test and (F) Fisher’s exact test. ****P<0.0001.